Immune Protection of the Brain—Efficient and Delicate

The brain is in general considered the body’s noblest organ. Its parenchyma is densely packed with electrically active neurons and glial cells that constantly undergo spatial rearrangements, but there is little connective tissue. The brain is vulnerable; it requires a particular degree of protection. Quite obviously, the brain is protected against external physical forces by a robust capsule formed by skull bones and meninges. Less obvious and by no means trivial is how the body protects its brain from endogenous danger, for example microbial infection. In fact, there has been a longstanding debate concerning the existence of immune reactivity within the central nervous system (CNS). In 1986, my colleagues and I reviewed the traditional concept of neuroimmunology—that the CNS was an immunoprivileged site, excluded from immune surveillance, and inaccessible to immunocompetent migratory cells [1] The evidence supporting the concept of immunologic privilege rested on three structural peculiarities that distinguish the CNS from other tissues. First, there is a specialized endothelial blood-brain barrier (BBB) that secludes the CNS parenchyma from the circulating blood and prevents most blood components from entering the tissue. Second, the CNS lacks fully organized drainage via lymphatic vessels. These are of crucial importance for the transport of antigen-presenting (dendritic) cells to travel from peripheral tissues to immune organs, where they trigger a full immune response. Third, major histocompatibility complex (MHC) determinants are conspicuously absent in the mature CNS. Without MHC proteins expressed on cell membranes, there is no presentation of antigenic peptide to specific T lymphocytes. Perhaps of even more importance, “professional” antigen-presenting cells, the cells specialized for the transport of peripheral antigen to immune organs and for activation of antigen-inexperienced naive T cells, are missing in the CNS A wealth of recent information indicates that the BBB, lack of lymphatic drainage, and the lack of MHC antigens by no means reflect an absolute absence of immune reactivity within the CNS. It is now clear that these features are characteristic of the resting CNS and that in “emergencies” the CNS parenchyma can support a local immune response. The CNS can be stimulated to create a microenvironment that allows the unfolding of productive immune responses. Hence immune reactivity within the CNS is not constitutive, as in most other tissues, but is inducible, becoming available on temporary demand Why did the CNS develop this peculiar variant of immune surveillance? I speculate that immune surveillance in the CNS has adapted to the particular nature and the needs of its target organ. It has been shaped in order to provide immune protection to the CNS and, at the same time, to reduce to the absolute minimum inflammatory bystander damage and other unwanted effects on neuronal function Experimental autoimmune encephalomyelitis (EAE) has been enormously useful as a paradigm in which to study autoimmune disease in the CNS. Studies that have used EAE models have helped to elucidate the basic features of physiologic immune reactivity in the CNS. EAE is caused by an autoimmune attack directed by T cells, which are specifically programmed to recognize protein components in the CNS. The best-studied targets of EAE are proteins of myelin or myelin-forming oligodendrocytes. In addition, astrocytic and even neuronal proteins may qualify as encephalitogenic autoantigens [2] Encephalitogenic T cells are located in the natural healthy immune repertoire. Most are of CD4 lineage and recognize their target antigenic peptide in context of MHC class II proteins, although recently, encephalitogenic CD8 T cells (restricted by class I proteins) have been described [3]. Like all T lymphocytes, the encephalitogenic T cells are formed in the thymus, but for reasons that are incompletely understood, they are not cleared from the repertoire, although a number of encephalitogenic target autoantigens are present in the thymus [4] In most persons, the brain-specific T cell clones are kept in a life-long, solid state of quiescence. Spontaneous attacks against the CNS are rarely seen. Encephalitogenic T cells unfold their autoaggressive pathogenic potential only upon maximal activation. Microbial infections are held responsible for accidental activation of autoaggressive T cells. Molecular mimicry (structural cross-reaction between a microbial with a myelin-derived peptide) is one possible activation mechanism. There may also be stimulation by microbial superantigens (which bind to class-specific determinant of the T cell receptor, rather than to its specific antigen-binding site, and thus activate classes of T cells defined by the type of T cell receptor used). Third, microbial infections may use mechanisms of innate immunity to create local milieus that support “spontaneous” activation of (autoimmune) T cells. Finally, some T cells express more than one specific antigen receptor. At least in theory, one receptor type could be specific for (myelin) autoantigen and the other for microbial antigen. Exposure to the microbial antigen could activate the T cell, which by virtue of its myelin-specific alternative receptor could attack the CNS The autoimmune CNS attack by pathogenic T cells thus rests on two properties. Activation is required to unfold the T cell’s pathogenic potential, while the correct T cell receptor specificity is needed to target the T cells into the right CNS target tissue [5] Activated CNS-specific T cells thus are programmed to travel to the CNS; they reach their target like magic bullets. Numerous investigators have used this competence to study T cell functions in EAE and, in a more general sense, to investigate T cell migration into and within the CNS as required for immune surveillance. These studies depend on unequivocal markers for the identification of the T cells upon in vivo transfer. Radioactive and fluorescent dyes were traditionally used as labels, but their dilution over time by membrane permeation or during cell division posed severe problems Recently, studies in my laboratory led to a highly efficient self-replenishing fluorescent marker. The method relies on retroviral introduction of the gene encoding the green fluorescent protein (GFP) into the genome of autoimmune T cells. The genetically modified T cells keep all of their immunologic functions. In particular, they are as encephalitogenic as their unmodified “wild type” counterparts, but they produce the fluorescent marker GFP throughout their life, which is also true for all their cellular progeny. Furthermore, the intensity of GFP fluorescence is correlated with the T cell’s state of activation. Highly activated GFP T cells produce large quantities of GFP and therefore are brighter than resting T cells [6]. Autoimmune GFP-engineered T cells have become useful probes to trace the migratory pathways of encephalitogenic cells before and during EAE. In addition, this approach allows for characterization of the functional changes that T cells undergo on their way to the CNS and within the developing EAE lesion Results of early studies that used myelin basic protein–specific T cell lines indicated that freshly activated T cells mediate EAE in a highly predictable manner. Of importance, clinical and histologic changes never develop immediately following transfer of the pathogenic T cells. Regardless of the number of pathogenic cells transferred, there is a minimal interval of about 3 days separating intravenous (iv) transfer and onset of EAE. It has been known for some time that a small number of activated T cells go through the naive BBB within a few hours after iv transfer [1, 7]. The T cells of the first immigration wave seem to endure in CNS parenchyma until onset of disease, but their function is not definitely known. It appears that the activated T cells of the first wave release activation-dependent cytokines (e.g., interferon [IFN]-γ and tumor necrosis factor [TNF]-α), which prime brain cells (glia, endothelium, and possibly neurons) to express immunologically relevant proteins. The early immigrant T cells would thus prime the CNS tissue to set the scene for the large scale (auto) immune reaction that flares up during onset of clinical disease, 3 days after T cell transfer (figure 1) Photomicrograph of experimental autoimmune encephalomyelitis infiltrate shows formation of T cell infiltrates. In the first wave (0–24 h after infection), freshly activated cells total perhaps >10,000; by the second wave (60–80 h after infection), there are millions of postactivated migratory cells For technical reasons, little is known about the fate of the majority of encephalitogenic T cells before EAE onset. Encephalitogenic T cell lines are regular T cells, indistinguishable from other T cells by membrane markers or other characteristics. Once introduced into a syngeneic recipient animal, the transferred T cells are embedded in the host’s immune system. By conventional techniques, it has been nearly impossible to distinguish a transferred encephalitogenic T cell from a host T cell of undetermined function The recent availability of GFP-engineered encephalitogenic T cells has facilitated tracing and characterization of pathogenic T cells during the prodromal period and during subsequent clinical disease. The work led to surprising observations [8]. Contrary to (naive) expectation, the transferred activated GFP T cells did not settle randomly in the diverse lymphoid tissues but took a strictly ordered itinerary. The first tissues to accumulate the autoimmune T cells (and control GFP T cells specific for ovalbumin) were the parathymic lymph nodes. This small group of lymph nodes is located at both sides of the thymus and drains the pleural area. The GFP T cells stay there about a day, enter the blood, and travel onward to the spleen. Predictably, after about 60–70 h in the recipient’s body, following an unknown signal, masses of the T cells leave the spleen. Brain-specific T cells head to the CNS while their ovalbumin-specific counterparts settle primarily in the peripheral lymph nodes The migratory activity of GFP T cells is accommodated by profound changes in their gene expression pattern. As mentioned, encephalitogenic (and other organ-specific autoreactive) T cells must be in a state of high activity in order to be efficiently autoaggressive. Activated T cells up-regulate the expression of a particular gene. Prominent examples are CD25 (the α chain of the interleukin-2 receptor) and the OX-40 membrane protein. Both markers, but also proinflammatory cytokines such as IFN-γ and TNF-α, are high in activated GFP T cells at the time of iv injection. Within hours after transfer, however, the activation markers are decreased. Concomitant with this down-regulation, another class of genes, especially receptors for several chemokines, is up-regulated. Down-regulation of activation markers and up-regulation of chemokine receptors are seen on protein and transcription levels. In remarkable distinction, MHC class II proteins, which are missing on the membrane of resting rat T cells, are also enhanced after transfer of GFP T cells, but there is no concomitant increase of class II gene transcription. Thus, the class II determinants seem to be acquired by T cells from neighboring cells Encephalitogenic T cells assume a novel functional phenotype after transfer into a recipient animal. This migratory phenotype expresses new chemokine receptors, which seem to be required for guiding the cells through the immune system and eventually to the CNS target tissue. At the same time, repression of activation-dependent genes decreases the actual cytotoxic potential of the effector T cells during recirculation As mentioned, about 3 days after transfer, a large number of T cells leave the spleen. They reach the CNS via the blood circulation and rapidly pass through the BBB. This migration coincides or briefly precedes onset of clinical EAE. The number of GFP-labeled encephalitogenic T cells that enter the CNS is unexpectedly high. In early stages of EAE (4 days after transfer), more than 90% of all CD4 T cells recovered from infiltrates are GFP T cells. This proportion decreases over time and is less than 10% by day 8 After CNS entry, the effector T cells undergo rapid reactivation with renewed induction of activation markers CD25 and OX-40 and of proinflammatory cytokines. T cell activation seems to be the consequence of autoantigen presentation because T cell receptor expression on the surfaces is reduced as the consequence of the formation of the immunologic synapse between T cell and antigen-presenting cell How is an EAE lesion terminated? In classical monophasic models of EAE, the CNS infiltrates resolve within a few days after formation [9]. To date, emigration of encephalitogenic T cells has never been shown with certainty; however, there is much evidence of apoptotic cell death of inflammatory cells [10–12]. Studies of GFP T cells show that the effector cells undergo rapid apoptosis within the fresh lesion. In areas with neurodegeneration, T cell apoptosis often occurs close to neurons and some dying T cells are taken up by the neighboring nerve cell [13] (figure 2). Nonspecifically recruited T cells and monocytes/macrophages follow with some delay Apoptotic encephalitogenic T cell engulfed by motor neuron of facial nerve after ipsilateral axotomy. Green fluorescent protein–engineered myelin basic protein–specific T cell (arrow) 4 days after intravenous injection I have stressed that the intact CNS tissues constitute a milieu unfavorable for immune responses. MHC antigens, cell adhesion molecules, soluble mediators, and other factors required to support immune reactions are largely missing. On the other hand, a remarkably broad spectrum of neuropathologic changes result in the induction of “immune molecules” within the brain [14]. These include infections, tumors, and, most intriguing, neurodegenerative processes. How could such diverse lesions converge to create an “immune friendly” milieu in the CNS? Experimental studies show that immune reactivity in the CNS is under the strict control of positive and negative elements. Functionally intact neurons suppress induction of immune molecules in their surrounding glial cells (and in themselves), thus exerting suppressive function. Yet, strong proinflammatory signals (high doses of IFN-γ or TNF-α) can overcome neuronal suppression of immune gene expression, allowing local production of MHC proteins, cytokines, and cell adhesion molecules Experimental models have impressively documented the regulatory effects of neuronal function and of proinflammatory mediators [15]. As shown by induction of EAE, an especially strong inflammatory stimulus can overcome the natural antiinflammatory disposition of CNS tissue. Conversely, experimental induction of localized neuronal degeneration (e.g., by remote axotomy) creates a circumscript milieu that attracts encephalitogenic T cells, presumably by a relaxation of neuronal immune repression due to postaxotomy degeneration (figure 3) Preferential infiltration of neurodegenerative central nervous system lesions by encephalitogenic green fluorescent protein–engineered T lymphocytes. Bars = 10 μm One useful paradigm in which to study in more detail the regulation of immune molecules within the CNS tissue relies on CNS tissue explant cultures (e.g., from the neonatal rat hippocampus) [16]. Under favorable conditions, these tissue fragments partly regenerate in vitro. In particular, neurons situated inside the explants recover to form functionally intact neuronal networks. Immunocytochemical analysis of hippocampus explants shows that, as in the intact brain, individual glial or neuronal cell components fail to produce MHC class II molecules. Abundant amounts of class II protein are, however, induced by treatment with tetrodotoxin plus IFN-γ. Tetrodotoxin, a blocker of sodium channels, paralyzes neuronal function and thus neutralizes neuronal suppression of immune gene expression. IFN-γ then is required to provide a positive proinflammatory signal to induce MHC class II The molecular mechanisms of neuronal gene suppression are not clarified in detail; however, neurotrophins contribute at least part of these suppressive signals, in particular the classical nerve growth factor, NGF. NGF is secreted by (hippocampal) neurons dependent on neuronal activity [16, 17]. In cultures treated with tetrodotoxin, there is a deficit of NGF production that results in increased inducibility of class II molecules. This effect is neutralized by the addition of exogenous NGF. NGF thus qualifies as a natural immunosuppressant in the CNS, a function that has been verified by NGF prevention of EAE [18, 19] Electric activity also controls expression of MHC genes in the neurons themselves. Electrically active cultured neurons do not express MHC determinants and are quite resistant to their induction by IFN-γ. After tetrodotoxin paralysis, however, IFN-γ treatment leads to induction of MHC on neuronal cell bodies and on their processes [20]. The neuronal class I proteins are functionally mature (figure 4). They present endogenous (e.g., viral) antigens in a way that is recognizable by specific CD8 T cells. These killer T cells destroy the class I–induced neurons, first cutting their neurites [20, 21] and then driving the entire cell into apoptosis [20, 22] Major histocompatibility complex class I induction in neurons. Interferon-γ is a prerequisite for a positive signal and TTX for a negative signal. Bar = 10 μm The experimental evidence presented supports a concept of tightly regulated immune reactivity within the CNS. Immune reactions are required to identify and remove potentially dangerous structures (microbes, tumor cells) from the CNS, but they must be limited to the essential minimum in order to avoid unwanted side effects. Exuberant uncontrolled immune responses could lead to severe disturbances of neuronal function. Soluble inflammatory mediators (cytokines, chemokines) modify neuronal function, as reflected by disturbed sleep, feeding, and temperature responses. In the extreme case, inflammation may result in cytotoxic “bystander” killing of irreplaceable neurons [23] I propose that fine tuning of the CNS immune response is achieved by several sets of regulatory strategies that act on the effector T cells and on the interacting CNS cells. CNS immune reactivity is controlled both in spatial and temporal terms. Spatial control of CNS immune responses is possible thanks to the regulation of the CNS immune milieu. Under regular conditions, expression of immune molecules is firmly repressed by neuronal mechanisms. These genes can be induced more easily when neuronal function is reduced as proven in (localized) neuronal degeneration, where degenerative areas express MHC antigens, cell adhesion molecules, and cytokines required to and support circulating T and other inflammatory cells. activated T cells, which from fresh (auto) immune responses and that have the to pass through the intact BBB in small would provide proinflammatory to areas and thus contribute to the of an immune milieu. The number of these activated T cells seems to be small to profound effects in the CNS. They become by their migration to lesion areas and their local A specific immune response would develop with the of large of T cells that the on their way to the CNS these cells do not high of proinflammatory mediators and much bystander effects on cells. after at the CNS lesion site, the migratory T cells recognize their specific antigen and reach maximal of activation within They act against their antigen under strictly controlled spatial control of the CNS immune response is of the CNS by specific effector T cells is a rapid and highly not of these effector cells are within the same lesion within a few This concept that are relevant for mechanisms and of CNS disease. must which features of CNS immune reactivity can be used to modify responses in autoimmune disease, in The experimental models be of I to the of the and work is the for this