The Immune Pathogenesis of Experimental Autoimmune Encephalomyelitis: Lessons Learned for Multiple Sclerosis?

The Preimmunization Neuroantigen-Specific T Cell Repertoire

E xperimental autoimmune encephalomyelits (EAE) does not develop spontaneously in wild-type animals. Neuroantigen-specific T cells are present in such animals, but they coexist with the endogenous antigen in an apparent state of stable tolerance before the immunization. What do we know about the immune biology of these precursor cells that, after immunization, can give rise to the effector T cells that mediate EAE?

As T cells mature in the thymus, their T cell receptors (TCRs) are generated by random rearrangements. Therefore, in addition to the TCRs that can recognize the yet to be encountered universe of foreign antigens, inevitably also TCRs are generated that have specificity for self-, including neuroantigens. The classic mechanism that accounts for the silencing of such potentially autoaggressive T cells is negative selection, also called central tolerance. Immature T cells that encounter (auto)antigens in the thymus undergo apoptosis. Special mechanisms have evolved to ensure that even tissue-specific autoantigens are presented in the thymus and induce central tolerance (Cohn 2009; Gardner and others 2009). Comparisons of the myelin basic protein (MBP)-specific T cell repertoires in wild-type mice (in which MBP is an autoantigen) and congenic MBP-deficient shiverer mice (in which MBP is a foreign antigen) provided the first evidence for negative selection to a neuroantigen (Targoni and Lehmann 1998). Immunization of shiverer mice with MBP triggered a high affinity T cell response that was directed against a single immunodominant peptide, MBP 79-87. In wild-type mice, MBP and this peptide were immunogenic only when injected at a 1,000-fold higher dose, and the T cell clones isolated from shiverer mice responded to a 10,000-fold lower concentration of MBP 79-87. The T cells selected in the presence of MBP in wild-type mice, therefore, had up to 10,000-fold lower functional avidity (affinity) for MBP than the T cells that were selected in the absence of MBP in the shiverer mice. These data showed that the mature MBP-specific T cells, which circulate in the immune periphery of wild-type mice, have escaped negative selection in the thymus due to their low affinity for the neuroantigen, and that they neglect the autoantigen in the immune periphery [including the central nervous system (CNS)] due to their low affinity for MBP.

Are such preimmunization neuroantigen-specific T cells naïve precursor cells—similar to foreign antigen-specific T cells that have not yet encountered their antigen—or have they been preprimed by the endogenous autoantigen? Spontaneous priming is seen in mice that are transgenic for neuroantigen-specific TCRs (Lafaille and others 1994; Waldner and others 2000; Pöllinger and others 2009). At a first glance this finding might suggest that also in normal mice the neuroantigen-specific precursor T cells are not naïve, but the endogenous autoantigen has caused them to differentiate into memory/effector cells. This conclusion might not be warranted, however, since spontaneous priming of T cells is also seen in TCR-transgenic mice that bear TCRs specific for foreign antigens, such as ovalbumin (Paul V. Lehmann, unpublished observation). In the latter mice the antigen itself cannot contribute to the T cell activation. Therefore, the spontaneous T cell priming that is observed in mice with transgenic TCRs might also result from a unique biology in these highly artificial animals including decreased numbers or complete lack of regulatory T cells, which results in uncontrolled T cell activation (Furtado and others 2001). In unimmunized wild-type mice one cannot detect preprimed neuroantigen-specific memory T cells (Forsthuber and others 1996; Yip and others 1999). Memory T cells secrete cytokines such as interferon (IFN)-γ, interleukin (IL)-2, IL-4, IL-5, or IL-17. Naïve T cells do not produce these cytokines. When lymph nodes or spleens of wild-type mice are tested for neuro- (or foreign) antigen reactivity before immunization, one cannot detect antigen-reactive T cells that produce any of the above cytokines even when ultra-sensitive ELISPOT assays are used that permit the robust detection of as few as one in a million cytokine-producing cells (Forsthuber and others 1996; Yip and others 1999; Kuerten and others 2010). Also, the outcome of immunizations suggests that in wild-type animals the neuroantigen-specific T cells are naïve before immunization. Unlike memory T cells that are more or less committed to a cytokine effector lineage, naïve T cells are uncommitted, and can readily be induced to differentiate into the T_H1, T_H2, and T_H17 cytokine effector lineages (Forsthuber and others 1996; Yip and others 1999; Batoulis and others 2010; Murphy and Stockinger 2010). In naïve (not preimmunized) wild-type mice one can induce highly polarized neuroantigen-specific T_H1, T_H2, or T_H17 immunity with the same ease as can be done with foreign antigens, also suggesting that the neuroantigen-specific T cells had been naïve and uncommitted at the time point of the immunization (Lichtenegger and others 2007; Kuerten and others 2010).

When foreign antigens are injected into naïve mice with different adjuvants, it is the adjuvant itself that dictates the class of T cell response that emerges. Immunizations with alum or incomplete Freund's adjuvant (IFA) trigger highly polarized T_H2 immunity (Forsthuber and others 1996; Heeger and others 2000; Yip and others 1999), injections with CpG adjuvants generate unipolar T_H1 immunity (in the absence of T_H17) and immunizations with complete Freund's adjuvant (CFA) trigger mixed T_H1 and T_H17 immunity in the absence of T_H2 engagement (Tigno-Aranjuez and others 2009b). These polarizing effects of the adjuvants tend to override intrinsic properties of antigens, genetic predisposition of the host, or the effects of antigen dose (Yip and others 1999). The adjuvants selectively engage the respective microenvironments that cause naïve T cells to differentiate into the different cytokine effector lineages: IL-12 induces the differentiation of IFN-γ-producing T_H1 cells; IL-23, IL-6, and TGF-β promote the generation of IL-17-producing T_H17 cells; and IL-4 causes the maturation of IL-4-producing cells T_H2 cells (Murphy and Stockinger 2010). The same rules that apply for the adjuvant-guided differentiation of foreign antigen-reactive T cells upon primary immunization (when verifiably naïve T cells are induced to differentiate into T_H1, T_H2, or T_H17 cytokine effector lineages) were also found to hold true when neuroantigens were injected in different adjuvants (Yip and others 1999; Heeger and others 2000; Lichtenegger and others 2007; Tigno-Aranjuez and others 2009a; Kuerten and others 2010). Therefore, the fact that neuroantigen-specific precursor T cells can readily differentiate into all known effector cell lineages suggests that these cells have a naïve phenotype before immunization.

The ability to engage different T cell lineages by defined immunization protocols also permits the evaluation of when the induction of an autoimmune response to a neuroantigen results in autoimmune pathology, and when it is tolerated without pathogenic consequences to the host—or even results in protection from subsequent autoimmune disease.

The Cytokine Signature of the Effector T Cells That Mediate EAE

In mice, the minimum requirement for the induction of EAE is the injection of neuroantigen in CFA, which triggers mixed T_H1 and T_H17 immunity in the absence of a T_H2 component (Hofstetter and others 2005b; Kuerten and others 2010). It should be noted, however, that IL-3 produced by antigen-specific T cells induces bystander IL-4 production in spleen cells that can be misinterpreted as an antigen-specific T_H2 component (Karulin and others 2002). In all EAE models studied, the cytokine signature of the neuroantigen-specific CD4⁺ T cells induced by immunization with CFA was IL-2⁺, IL-3⁺, IFN-γ⁺, IL-17⁺, and IL-4⁻, IL-5⁻ (Hofstetter and others 2005b; Kuerten and others 2006; Kuerten and others 2010) (Table 1). In several EAE models pertussis toxin (PTX) needs to be injected in addition to trigger clinical disease. PTX is a T_H1/T_H17 adjuvant of its own right (Fedele and others 2010). In addition, PTX supports the autoimmune pathogenesis by permeabilizing the blood–brain barrier (BBB), and by facilitating autoantigen recognition in the CNS, activating the tissue-resident antigen-presenting cells (APCs) in the target organ (see below and Hofstetter and others 2002; Hofstetter and others 2007).

Immunization with neuroantigen in CpG adjuvant triggers a T_H1 T cell response of similar magnitude as seen after injection with CFA, measured by the frequencies of IFN-γ- and IL-2-producing CD4⁺ T cells. Also, the CpG-induced response does not entail a T_H2 component as evidenced by the lack of IL-4- and IL-5-producing T cells (Tigno-Aranjuez and others 2009a) (Table 1). Here too, however, mind the bystander production of IL-4. Unlike the CFA-triggered response, after CpG immunization no IL-17-producing CD4⁺ T cells were detectable. The T cells induced with CpG did not cause any clinical or histological manifestations of EAE and could not mediate delayed-type hypersensitivity (DTH) after the classic subcutaneous challenge (Tigno-Aranjuez and others 2009a). The latter finding is also crucial since IL-17 production by T cells has been closely linked to the ability of the T cells to mediate DTH (Nakae and others 2002; Umemura and others 2007; Tigno-Aranjuez and others 2009a), and because EAE essentially results from a DTH-type reaction in the CNS. Injection of an IL-17 neutralizing antibody prevents the development of CFA/neuroantigen-induced EAE (Hofstetter and others 2005a). IL-23 knock-out mice that can generate T_H1, but not T_H17 cells are resistant to EAE (Cua and others 2003). In contrast, IFN-γ knock-out and IFN-γ receptor knock-out mice show unaltered or even increased EAE susceptibility (Ferber and others 1996; Krakowski and Owens 1996; Willenborg and others 1996). Collectively, these findings show that induction of T_H1 immunity alone is not sufficient for eliciting encephalitogenic effector T cells, whereas the induction of IL-17-producing neuroantigen-specific T cells seems to be a requirement for EAE development.

Injection of the same neuroantigen in IFA or alum induces type 2 cytokines in addition to IL-2, in the absence of EAE and DTH (Forsthuber and others 1996; Yip and others 1999; Heeger and others 2000; Lichtenegger and others 2007; McKee and others 2009) (Table 1). After such injections mice typically even become resistant to subsequent attempts to induce EAE (Lichtenegger and others 2007). When neuroantigens are injected with IFA and PTX, EAE can result due to the prevalence of the intrinsic adjuvant effect of PTX (Hofstetter and Forsthuber 2010).

One major lesson these types of immunizations can teach us is that the induction of an autoimmune T cell response in itself is not necessarily hazardous to the host. Neuroantigen-specific T cells can become primed, for example, due to the infection with a cross-reactive microorganism, or due to tissue damage, yet this will not lead to CNS pathology unless a specific type of effector cell becomes engaged. Notably, in all of the above studies the clonal sizes (frequencies) of neuroantigen-specific T cells triggered by the CpG, IFA, or alum immunization were not smaller than those elicited by CFA immunization; only their cytokine signature (effector class) differed. The second major lesson learned is that the induction of T_H17 cells that also mediate DTH seems to be a prerequisite for the encephalitogenicity of the neuroantigen-specific T cells.

EAE Results from T Cell-Mediated DTH

DTH is one of the basic, stereotypic response patterns of T cell-mediated immunity (Kobayashi and others 2001). It has been first noted in the context of the immune response to Mycobacterium tuberculosis (MTB) and has also been termed type IV reaction/allergy. The first requirement for DTH induction is the priming of a special effector class of antigen-specific T cells, apparently T_H17 (Fig. 1). In the case of MTB infection, MTB itself is the adjuvant that elicits MTB-specific T_H17 cells. In the case of immunizations with CFA, it is the inactivated mycrobacteria that create the micromilieu for the T_H17 differentiation of the T cells that are specific for the antigen mixed into the adjuvant. In this way, T_H17 memory cells are generated that leave the site of their differentiation, the draining lymph node, and start recirculating in the entire organism with the mission to detect that same antigen. In the context of MTB infection that antigen encounter could be anywhere the microorganism has spread, including the lung, gut, and brain. Antigen recognition by the T_H17 cell leads to their activation and the induction of cytokine synthesis at the site of antigen re-encounter. The cytokines that the antigen-specific T cells secrete activate tissue-resident macrophages, and recruit immature monocytes/macrophages from the blood steam, which also get activated (Fig. 1). IL-17 is one of the key cytokines involved. Other mediators that have been generally identified in the process of DTH are the macrophage/monocyte chemoattractant protein-1 (MCP-1), IL-2, IL-3, IL-8, macrophage migration inhibitory factor (MIF), granulocyte/macrophage colony-stimulating factor, and tumor necrosis factor (TNF)-β (Kobayashi and others 2001). Notably, while the T cell is essential to initiate the DTH reaction, it results from a cascade of cellular reactions that involve positive and negative feedback loops, including other cell types, mostly macrophages (Fig. 1). In EAE main attention has been drawn to IL-17, but based on this general knowledge it is likely that many other T cell- and macrophage-derived cytokines are also critical.

OPEN IN VIEWER

FIG. 1.

Antigen-driven differentiation of the effector CD4 cells that mediate EAE. In the sensitization phase, naïve neuroantigen-specific CD4 cells encounter their antigen—either the neuroantigen itself after subcutaneous immunization, or a cross-reactive protein expressed by an infectious agent. This encounter happens in secondary lymphoid tissues and results in the proliferation and differentiation of the T cell. The CD4 cells differentiate into IL-17-producing Th17 cells if a specific microenvironment is induced by the adjuvant or the infection that entails TGF-β, IL-1β, IL-6, IL-21, and IL-23—nonpathogenic CD4 cells emerge in other microenvironments (see Table 1). After 5 to 10 days, the newly generated effector cells leave the lymph nodes, and disseminate into the entire organism, a few of them also reaching the CNS where they encounter the endogenous neuroantigen. These T cells initiate in the CNS the effector phase by recruiting and activating macrophages. The activated macrophages and their secretory products are the primary mediators of the local inflammation causing the CNS injury. The basic mechanism involved corresponds to a delayed hypersensitivity reaction. DTH is a multicellular, complex event, in which IL-17 possibly plays a central role in disease initiation, but is one of many contributing factors of downstream effects. APC, antigen presenting cell; CNS, central nervous system; DTH, delayed-type hypersensitivity; EAE, experimental autoimmune encephalomyelitis; GM-CSF, granulocyte/monocyte colony-stimulating factor; IL, interleukin; MCP-1, macrophage/monocyte chemoattractant protein-1; MHC, major histocompatibility complex; MIF, macrophage migration inhibitory factor; MIP, macrophage inflammatory protein; NO, nitric oxide; PGE, prostaglandine; TCR, T cell receptor; TNF, tumor necrosis factor; TNFR, tumor necrosis factor receptor.

After a brief initial phase that is dominated by polymorph nuclear cells (neutrophils), 24–48 h after the antigen encounter by the T cell, a characteristic dense mononuclear cell infiltrate develops at the site, consisting of activated macrophages. These macrophages are the actual effector cells of the DTH reaction. In the case of tuberculosis they form a defense wall to encapsulate the infectious agent that cannot be degraded. With prolonged disease duration granulomas develop. With time, the granulatomatous tissue can even reorganize to constitute a quasi lymphoid organ that behaves similar to lymph nodes in as much that it becomes a preferred destination for T cell trafficking, while also supporting T and B cell priming and differentiation. In this way, a tertiary lymphoid tissue is generated in the target organ of the T cell attack that can contribute to the amplification and maintenance of the localized immune response. As will be outlined below, most of these basic elements of DTH can also be seen in EAE.

Entry and Activation of T Cells in the CNS

After the subcutaneous injection of neuroantigen in CFA, the neuroantigen-specific CD4⁺ T cells proliferate and undergo antigen-driven cytokine differentiation in the draining lymph nodes, from which they emerge as T_H1 and T_H17 cells within 5 days (Targoni and others 2001). While their numbers continue to increase in the draining lymph nodes until day 10 (after which their numbers rapidly decline), already as early as day 5 after immunization the primed neuroantigen-specific T cells start disseminating in the entire organism, by day 14 reaching high frequencies even at irrelevant sites such as the peritoneal cavity (Targoni and others 2001). As part of the random dissemination, a minor number of the neuroantigen-specific T cells also seeds into the CNS. Such recently activated T cells have the ability to transmigrate through the BBB—a property that they share with any activated T cell, irrespective of its antigen specificity. The difference between neuroantigen- and third-party antigen-specific T cells resides in the fact that the latter return into circulation after a short dwelling in the perivascular space of the CNS, whereas the neuroantigen-specific T cells stay in that space if they recognize antigen (Hickey and others 1991). This antigen encounter in the perivascular space of the CNS stimulates the neuroantigen-specific T cells to secrete the very cytokine(s) they have been programmed to express during the antigen-driven differentiation in the draining lymph nodes, including IL-17: T_H17 cells will initiate a local DTH reaction that eventually matures into a perivascular lesion.

The question arises why a low affinity neuroantigen-specific memory T cell that apparently neglected the endogenous neuroantigen before the immunization becomes stimulated by it after the immunization. During the immunization itself an artificially high dose of antigen is injected resulting in the priming of such cells, but how can the same T cell neglect the autoantigen as a naïve cell and be stimulated by it after the immunization? While (unlike for B cells) the antigen receptor on T cells (TCR) does not undergo affinity maturation after immunization, activated T cells upregulate a number of accessory and costimulatory molecules that increase their antigen responsiveness. Subsequently, neuroantigen expression levels that were at a sub-stimulatory level for the naïve T cell become stimulatory to the memory T cell (Lehmann and others 1998; Zhu and others 2010). Notably, even the activation threshold of the memory cell is not fixed, but continues to increase with the time elapsed since the last antigen encounter (Hesse and others 2001). This type of fluctuation of the T cell activation threshold could contribute to the subsequent effector function of the T cells causing remissions and relapses of the disease.

The activation threshold of the T cell is also affected by the activation state of the APC. Neuroantigen recognition in the CNS by neuroantigen-specific memory T cells therefore also critically depends on the activation state of the APCs in the CNS itself. Most EAE protocols in mice require the administration of PTX. In EAE models on the C57BL/6 background immunizations with CFA alone induce a pathogenic T cell type in the periphery (eg, spleen) that can mediate disease upon adoptive transfer, but will not result in histological or clinical EAE unless PTX is also administered. In the absence of PTX, such T cells can even promote wound healing in the CNS (Hofstetter and others 2003). Pertussis causes massive induction of chemokines and accessory molecules (Hofstetter and others 2002) and apparently this pre-excitation of the APC in the target organ constitutes a third signal that can largely contribute to the susceptibility of the target organ to a T cell attack (Darabi and others 2004). The SJL/PLP model seems to contradict the notion that APC activation in the CNS is a prerequisite for antigen-recognition by neuroantigen-specific T cells because here, immunization with peptide in CFA alone suffices to trigger the disease. However, CFA itself induces a cytokine storm in the host that also affects the CNS. When combined with a highly elevated T cell response that PLP peptide induces, CFA alone seems to suffice to provide sufficient activation of the APCs in the CNS so that they can stimulate the PLP-specific T cells. Also, the fact that adoptive transfer of neuroantigen-specific T cells can cause EAE without administration of PTX does not contradict the third signal hypothesis. Adoptive transfers for induction of passive EAE require that the neuroantigen-specific T cells have been preactivated within a week before the injection, and such recently stimulated T cells produce massive amounts of cytokines and chemokines that can activate APCs in the CNS. The overall picture that arises is that in addition to the quality and magnitude of neuroantigen-specific T cell immunity, also the responsiveness of the target organ itself plays a critical role in defining whether disease develops and what severity and course it takes.

The Postimmunization Peripheral T Cell Repertoire

Before we consider the further course of the inflammatory response induced by the neuroantigen-specific T cells in the CNS, let us take a look at these T cells in the immune periphery, that is, outside of the brain. Even at the peak of EAE the vast majority (over 99%) of the neuroantigen-specific T cells that have been primed by the immunization continue to be in the immune periphery, distributed between spleen, blood, peritoneal cavity, and other organs; only few can be detected in the draining lymph nodes (Targoni and others 2001; Hofstetter and others 2005c). These peripheral T cells have the same affinity (antigen dose responsiveness) for the neuroantigen as those that migrated to the CNS (Targoni and others 2001; Hofstetter and others 2005c). Therefore, there is no evidence that among the T cell clones that have been engaged by immunization, those that have a relatively higher affinity for the neuroantigen are retained in the CNS (constituting the actual effector cells), whereas only the low affinity end of the spectrum would continue to recirculate in the immune periphery and the blood (being irrelevant for the pathogenesis of the disease). Also, the cytokine signature of these T cells in the CNS is precisely reflected in the immune periphery and blood (Targoni and others 2001; Hofstetter and others 2005c). This notion is important because most of the studies of the autoantigen-specific T cells are conducted by convenience or necessity on cells isolated from the immune periphery (spleen for mice, and the blood for humans): the neuroantigen-specific T cells detected in the periphery seem from all what one can tell to be representative for those present in the target organ, the brain. Confirming this notion, the T cells that are isolated from the periphery at the peak of the disease, or in the chronic stages of the disease, can adoptively transfer EAE proving that they contain the encephalitogenic clones. In models, in which the disease is monophasic, that is, mice recover after a single episode of EAE, the neuroantigen-specific memory T cells continue to persist in largely unchanged frequencies and with unchanged affinities or cytokine signatures. These postrecovery T cells retain their encephalitogenic potential since they can adoptively transfer EAE. It looks like that this peripheral pool of preprimed potentially pathogenic effector cells can again become ignorant of the self-antigen, but can be reactivated by stimuli such as activation by superantigen (Menezes and others 2007).

A question of immunodiagnostic relevance is whether the frequencies of neuroantigen-specific T_H17 (and/or T_H1) cells in the blood permit predictions on the concurrent or subsequent disease activity. In mice this was tested by longitudinally measuring the T cell reactivity to the neuroantigen in the blood of individual animals after the immunization while following the clinical disease course. In MOG peptide-immunized C57BL/6 mice, there was a highly significant correlation between the magnitude of the antigen-specific T cell response before disease onset, and the subsequent development of EAE severity. This relationship was not seen, however, in PLP peptide 139-151-immunized SJL mice. Also, no clear correlation was seen between the clonal sizes in the blood preceding the relapse, and the severity of the relapse (Kuerten and others 2010). These results indicate that while neuroantigen-specific T cells are an absolute requirement for the elicitation of EAE, the extent of CNS inflammation and neuropathology is not solely dictated by the magnitude of the effector T cell pool. Since these T cells operate via the engagement of a DTH reaction, the cell types of the innate immune system that mediate DTH, that is, primarily macrophages, will also play an important role in defining the severity of the inflammation. Moreover, the susceptibility of the CNS tissue to the inflammatory damage also defines the magnitude of pathology (Stefanie Kuerten, manuscript in preparation).

The number of neuroantigen-specific T cells that get primed through immunization in the periphery is relatively low and can be estimated to be around 20,000 to 30,000 cells per mouse (Targoni and others 2001). Of these cells only a few percent can be detected in the CNS at any given time point (Targoni and others 2001). Thus, less than 1,000 of the peripherally primed neuroantigen-specific T cells reside within the CNS at any given time of EAE. As the initial T cells trespass the BBB more or less randomly, those that recognize autoantigen start generating an inflammatory focus. They recruit macrophages, and other T cells to the site among which a few will also be neuroantigen-specific and gradually, perivascular infiltrates develop. The vast majority of T cells in this infiltrate are not specific for the neuroantigen (Steinman 1996). When considering the fate of the neuroantigen-specific T cells, it is important to distinguish whether they migrate into the perivascular space and the CNS parenchyma proper (Bauer and others 1998; Lehmann 1998). The perivascular mesenchyma is conductive to immune responses—it is the site where most of the recruited T cells and macrophages accumulate forming the perivascular cuffs. These perivascular cuffs represent tertiary lymphoid tissues, in which T cells and other immune cells not only survive, but proliferate and differentiate, to which they are recruited and that they can leave. From here T cells can also enter the parenchyma; however, once having migrated there, the T cells are prone to undergo apoptotic death (Bauer and others 1998). T cell migration to the CNS parenchyma is a 1-way path. With time, as T cells die in the parenchyma, the peripheral pool of the neuroantigen-specific T cells gets drained and eventually the first wave of effector cells (that was primed by the peripheral immunization) exhausts (Targoni and others 2001; Kuerten and others 2010).

Determinant/Epitope Spreading

The perivascular cuffs that develop during EAE increasingly gain functions of tertiary lymphoid tissues. In addition to memory cells, also naïve T cells get recruited to these sites, including naïve neuroantigen-specific T cells that are unrelated in antigen-specificity to the immunizing neuroantigen (eg, PLP-reactive T cells after eliciting EAE by immunizing with MBP). These naïve T cells will encounter the endogenous neuroantigen (in this example PLP) presented on professional APCs in a tertiary lymphoid tissue, and will be stimulated to engage in an immune response. Thus, after immunization with a single peptide of a neuroantigen, primed T cell reactivities can be detected to unrelated peptides of the same or different neuroantigens providing evidence for an amplificatory reaction, called determinant/epitope spreading (Lehmann and others 1992, 1993; McRae and others 1995). The priming of these second-wave T cells has been shown to occur in the CNS itself rather than in the lymph nodes that drain the CNS (Targoni and others 2001; Tompkins and others 2002).

Attention has been drawn to lymphoid organization within the CNS of multiple sclerosis (MS) patients (Serafini and others 2004; Magliozzi and others 2007). The development of tertiary lymphoid tissues can be particularly well studied in a novel EAE model, the MBP-PLP fusion protein (MP4)-induced disease of the C57BL/6 mouse (Kuerten and others 2006). This model displays the formation of high endothelial venules, the expression of chemokines that attract lymphocytes to home, and the development of B cell follicles—all providing evidence for a lymphoid organization of the CNS that can serve as a fertile local ground for the maintenance and amplification of the immune attack on the brain (Stefanie Kuerten, manuscript in preparation).

The Autoimmune Response Against Neuroantigens Occurs in Waves

After immunization with a neuroantigenic peptide, the frequency of the T cells recognizing this peptide rapidly increases to peak values within 10 days, but subsequently the numbers of these first wave effector cells gradually decrease to reach barely detectable levels after 3–4 weeks as they die in the CNS parenchyma (Targoni and others 2001; Hofstetter and others 2005c). At the same time, a second wave of effector cells emerges trough determinant spreading that is specific for unrelated neuroantigens, and subsequently also these T cells exhaust as they head for apoptosis induction in the CNS parenchyma (Targoni and others 2001). A third wave of T cell response might follow, etc. Overall, a dynamic autoimmune repertoire evolves with shifting antigen fine-specificity (Tuohy and others 1999). The T effector cell population therefore is dynamic and can cause considerable interindividual variability in the course of the immune attack on the CNS.

Attenuation of the DTH Reaction

The neuroantigen-specific T cells are the drivers of EAE. They target an abundant antigen in the CNS, which cannot be cleared. Thus, one might expect that these T cells will continue to expand as they are boosted by the endogenous autoantigen destroying the target organ leading to and causing lethal disease. While in human MS this outcome can be seen, in mice, acute lethal EAE is the exception. It is observed only in few EAE models, and even in those it affects only a fraction of the mice. Typically, the animals either recover from an initial episode of EAE and do not undergo subsequent relapses, or even if EAE continues, the relapses occurs after a period of recovery. Thus, the initial T cell attack on the CNS is typically self-limiting. The attenuation of the autoimmune attack on the CNS does not result from the exhaustion of the effector cell pool—during the recovery the numbers of neuroantigen-specific effector T cells present in the spleen and blood continues to be high and these T cells that are present in the periphery retain their encephalitogenic potential upon adoptive transfer (Targoni and others 2001; Hofstetter and others 2005c; Menezes and others 2007). It is not known why the T cell attack on the CNS comes to a spontaneous halt, but apparently the self-limiting nature of the disease needs to be sought within the CNS itself. It is possible that the T cells in the CNS undergo spontaneous attenuation. Since T_H17 cells are the mediators of the disease, and since recently it has become clear that T_H17 cells can convert into inducible T regulatory cells (iTregs) (reviewed in Zhou and others 2009), it is possible that the neuroantigen-specific T cells switch from a pathogenic phenotype to a protective one, a hypothesis that has not yet been tested in EAE. Another possibility is that after an initial excitability by T cells, the macrophages in the inflamed brain tissue become refractory to the T cell-mediated stimulation: while the T cells continue to secrete pro-inflammatory cytokines, the macrophages do not respond to them anymore. The first observation of such an attenuation reaction was lipopolysaccharide (LPS)-resistance (Fujihara and others 2003): injection of bacterial LPS causes full-blown macrophage activation with TNF release leading to a lethal vascular shock reaction. If, however, the animals were injected with a sublethal dose of LPS first, they tolerated a many fold higher of the otherwise lethal dose of LPS without developing any symptoms. In this scenario, the macrophages have become resistant to LPS stimulation and do not produce TNF after the renewed exposure.

The development of resistance of the innate immune system to T cell-mediated inflammation has been first described in the context of graft-versus-host (GvH) disease (Lehmann and others 1991). The injection of alloreactive CD4⁺ T cell clones can cause lethal GvH disease resulting from vascular leakage. Preinjecting a sublethal dose of the clone, however, induces complete resistance to the subsequent injection of an otherwise many fold lethal number of the clone (Lehmann and others 1991). This GvH-resistance results from the failure of the innate immune system to respond to the recurrent stimulation by the CD4⁺ T cells.

In tumor immunology recent attention has been drawn to an immune suppressive microenvironment created by tumors that prevent the T cell attack on the tumor (Singer and others 2011). It is conceivable that also normal tissues can protect themselves in a similar manner against an ongoing T cell attack and build a protective environment. Therefore, if we want to understand EAE pathogenesis, in addition to the T cells that drive the inflammation, we also need to consider the responsiveness of the innate immune system and that of the target organ proper, both of which underlie regulatory mechanisms of their own.

Reversible and Irreversible CNS Damage During EAE

During the DTH reaction, T cells and macrophages secrete lymphokines, chemokines, prostaglandins, oxygen radicals, and many other agents of inflammation. Jointly, they create a microenvironment that in the context of an immune response against an infectious agent helps keeping the infection localized. These cytokines inhibit local protein synthesis, thereby suppressing the replication of infectious agents in infected cells (but also suppressing regular functions of noninfected cells in the vicinity). In the context of an autoimmune T cell response to neuroantigen, the DTH that develops in the CNS has similar consequences, which can be summarized as what general pathology describes as functio laesa: the cells in the inflamed tissue minimize their metabolism and service functions for the organism. In EAE, oligodendrocytes stop producing myelin. These cells survive, but show reduced functions and demyelination occurs. The naked/uncoated axons stop transmitting signals normally and neurological deficits manifest themselves leading to the characteristic symptoms. Demyelination initially is reversible and tends to be transient. Remyelination can occur as soon as the inflammation ceases, which leads to the disappearance of the clinical disease symptoms. However severe and chronic inflammation will lead to permanent destruction of CNS tissue, with scar formation and axonal damage/transection (Trapp and others 1998; Trapp 2004).

Antibody- and CD8⁺ T Cell-Mediated Autoimmune Pathology

Classic EAE models rely on immunizations with protein antigens. Due to general rules of antigen processing and presentation, immunizations with proteins inherently induce CD4⁺ T cells. Encephalitogenic peptides of neuroantigens have also been defined through immunizations with the proteins, and thus were identified as major histocompatibility complex class II restricted determinants that are recognized by CD4⁺ T cells. While it has become apparently clear that CD4⁺ T cells are sufficient to induce EAE, it is not clear whether they can account for the full spectrum of disease manifestations that are seen in MS. Immunizations with neuroantigens or their peptides will inevitably result in the induction of an antibody response too. In most EAE models, however, these antibodies do not seem to contribute to the disease (Bourquin and others 2003; Oliver and others 2003). Studies of MP4- (MBP-PLP fusion protein)-induced EAE, in contrast, suggested that an antibody response against extracellular PLP domains was required for the development of full-blown EAE, and in particular for demyelination to occur (Kuerten and others 2011). In MS there is ample evidence for B cell involvement, ranging from the disease-specific oligoclonal immunoglobulin bands in the cerebrospinal fluid to B cell infiltration and follicle formation in the brain (reviewed in Franciotta and others 2008; Cross and Waubant 2011). Myelin antigen-specific antibodies on their own, however, do not mediate pathology (Kuerten and others 2011). Thus, CD4⁺ T cells are required for inducing EAE, but the engagement of an additional autoantibody response against neuroantigens can modify the disease. Like the CD4⁺ T cell response to neuroantigens, also the antibody response underlies the antigen-spreading reaction (Robinson and others 2003), possibly contributing to the variability of clinical outcome.

Presently, it is unclear what role CD8⁺ T cells play in autoimmune CNS pathology. While for the aforementioned experimental reasons the field has been biased toward CD4⁺ T cell-mediated EAE, evidence has emerged that also CD8⁺ T cells can contribute (Ji and Goverman 2007). Also in MP4-induced EAE, in which the immunization elicits CD4⁺ T cells only due to the injection of protein antigen, possibly through determinant spreading, CD8⁺ T cells have been shown to contribute to the development of the disease (Kuerten and others 2006).

Multiple Faces of EAE and MS

MS is one of the most difficult to diagnose neurological diseases because its clinical manifestations are extremely variable and the disease course also shows unpredictable individual patterns. We are far from understanding the complexities that underlie this variability, but certain patterns clearly emerge. First, it has become clear that different genetic backgrounds will lead to different manifestations of an autoimmune T cell attack on the CNS (Hoppenbrouwers and Hintzen 2010). It is also clear that the actual neuroantigen targeted by the T cell attack (elicited by immunization) will result in a differential targeting of anatomical regions of the CNS (Kuerten and others 2007). Differences in lesion localization are a typical feature of MS, termed dissemination in space, and are likely to cause heterogeneity in clinical symptoms. There is evidence for the prevalence of either T cell-mediated (dominated by the DTH-type mononuclear cell infiltration) or more antibody-biased variants of MS in humans with complement depositions in the target organ prevailing (Lucchinetti and others 2000). In addition to these rather defined parameters, there are dynamic elements of the inflammatory cascade that can result in interindividual variations of disease progression. Among these are the extent of antigen spreading, the prevalence of neuroantigens to which the spreading occurs, and the rate at which compensatory reactions of the innate immune system surface to counterregulate the damage of the target organ. The pathology of EAE and MS might be exemplified by the movie Groundhog Day, in which minute changes in events result in fundamentally different outcomes even though the actors and the play are otherwise stereotypically the same.