Are Astrocytes Central Players in the Pathophysiology of Multiple Sclerosis?

Multiple sclerosis (MS) is a chronic inflammatory demyelinating and neurodegenerative disease of the central nervous system (CNS) probably triggered by an environmental factor in genetically susceptible individuals. The initiation, maintenance, and suppression of immune responses are complex events, requiring the coordination of signals among different cell types. An interaction between antimyelin T cells and glial cells plays a pivotal role in the initiation and development of the lesions. A proposed dichotomy has emerged between microglia and astrocytes in terms of their respective roles in the formation of lesions in MS. A more complete characterization of the roles played by these cell types in CNS immune responses might lead to better therapeutic strategies to combat the disease. Many investigators favor microglia over astrocytes as the main "villain" in MS, whereas others believe that astrocytes play a more prominent role in the pathophysiology of the disease. We present the hypothesis that astrocytes are centrally involved in the proximate initiation sequence that ultimately culminates in the development of CNS immune responses in MS.

A rigid selection process in the thymus eliminates most self-reactive T cells. However, antimyelin T cells, among many other autoreactive cell clones, appear to escape this negative selection process.^{1 - 2} This explains why T cells reactive to myelin antigens, including myelin basic protein, myelin proteolipid protein, and myelin oligodendrocyte glycoprotein, are easily found in the peripheral blood of healthy individuals.^{3 - 5} These observations indicate that additional factors are required before autoreactive T cells can become pathogenic. Activation of autoreactive T cells is a critical event in the induction of CNS disease. Once activated, T cells can enter the normal CNS parenchyma as part of their normal "program" of seeking antigen.^{1 ,6} T cells can be activated in the periphery through, for example, presentation of viral and bacterial antigens or peptides that have sufficient sequence or molecular shape similarity corresponding to an immunodominant self-peptide (molecular mimicry). It appears that peptides from a wide variety of common human pathogens may be involved in the activation of antimyelin T cells in the peripheral blood.⁷

Activated antimyelin T cells that have infiltrated the CNS do not necessarily produce injury unless they recognize their specific antigen, at which time they may become reactivated.^{1 ,6 ,8} The inflammatory cascade and the resultant lesion formation in MS are contingent on the ability of activated antimyelin T cells to recognize their specific antigen in the context of class II major histocompatibility complex (MHC) molecules expressed on the membrane of antigen-presenting cells. Immigrant activated T cells in the CNS parenchyma have the capability to release proinflammatory cytokines, such as interferon γ, which induce the expression of class II MHC molecules on antigen-presenting cells.^{1 ,6 ,8} The formation of the trimolecular complex (consisting of T-cell receptor, antigenic peptide, and MHC molecule) results in either an enhanced immune response against the bound antigen or anergy, depending on the type of signaling that results from interactions with surface costimulatory molecules and their ligands.^{1 - 2 ,6} The B7 family of costimulatory molecules expressed on antigen-presenting cells is capable of providing a second signal to T cells at least via 2 receptors, CD28 and CTLA-4. Costimulation provided by B7 molecules to its ligand CD28 is important in the initiation of the autoimmune response,⁹ which involves a cascade of inflammatory events that lead to injury to myelin, oligodendrocytes, and axons. This inflammatory cascade has been extensively described in other reviews.^{1 ,8 ,10} The interaction between B7 and CTLA-4 plays a critical role in down-regulating the immune response.

One of the most controversial and debated issues in the MS literature is whether microglia or astrocytes represent the principal CNS antigen-presenting cells. Under nonpathologic conditions, human microglia can constitutively express MHC class II antigens.^{11 - 13} In addition, microglial MHC class II and costimulatory molecules are readily up-regulated in the context of many forms of CNS pathologic conditions,^{14 - 15} including ischemic brain and spinal cord lesions. It is therefore difficult to understand why patients with ischemic stroke do not develop inflammatory demyelinating lesions despite the presence of activated T cells against myelin proteins in their circulation.¹⁶

Initial reports^{17 - 21} showing that astrocytes in active MS lesions express class II MHC molecules have been contradicted by a series of observations that failed to confirm these findings.^{13 ,22 - 23} A recent observation corroborates the initial reports that some scattered astrocytes in active MS lesions, especially at the plaque edges, express MHC class II molecules.²⁴ In contrast, in focal ischemic neuronal damage, which is also associated with the invasion of inflammatory cells that release proinflammatory cytokines, and in spinal cord of patients with amyotrophic lateral sclerosis, all astrocytes were found to be MHC class II negative.²⁴

There is no doubt that microglia play a crucial role in the pathophysiology of MS, but they may not necessarily be the cells that exclusively initiate aberrant CNS immune reactions. In some functional assays, carefully isolated microglia were found to be poor stimulators of T-cell proliferation.^{25 - 27} It has even been suggested that microglia may act to protect the CNS parenchyma from the unwanted actions of patrolling lymphocytes. Thus, rather than encouraging their activation and proliferation, microglia may actually induce their death through apoptotic mechanisms.^{26 ,28} Alternately, investigators who favor microglia over astrocytes as the prevailing CNS antigen-presenting cell in MS will argue that the astrocytes predominantly play a role in limiting CNS inflammation by inducing T-cell suppression and apoptosis.^{29 - 30} Astrocytes are also a potential source of mediators, such as transforming growth factor β, that inhibit MHC class II antigens and intercellular adhesion molecule 1 expression in microglia.³⁰

Many investigators consider astrocytes to be poor antigen-presenting cells, since they do not constitutively express class II MHC molecules and lack costimulatory molecules. However, it appears that astrocytes may act as antigen-presenting cells under certain conditions. Some investigators found a lack of antigen-specific T-cell activation by cytokine-treated astrocytes,²⁹ whereas others found activation of primed CD4⁺ T cells^{31 - 32} and activation of naive CD4⁺ T cells by antigen-pulsed, interferon γ–treated astrocytes.³³ In some cultures of astrocytes, interferon γ induces not only class II MHC expression but also the expression of adhesion molecules (intercellular adhesion molecule 1 and vascular cell adhesion molecule 1) and costimulatory signals (B7-1 and B7-2), in addition to the antigen-processing machinery.^{33 - 39} So far, this has been more convincingly demonstrated in rodent than in human astrocytes, and the few studies^{11 ,40} in MS plaques failed to detect B7-1 and B7-2 expression on astrocytes. This finding deserves additional research, because MHC class II molecules on astrocytes in MS plaques were also detected on a few scattered reactive astrocytes in MS plaques, only by some investigators and not by others. The conflicting data in the literature regarding the ability of astrocytes to express class II MHC and costimulatory molecules suggest that astrocytes may have limited potential to act as facultative immunocompetent CNS antigen-presenting cells under certain conditions.

It has been shown that reactivation of primed T cells can be induced by class II MHC interactions alone without costimulatory signals.^{9 ,41 - 42} Myelin basic protein–reactive T cells in MS patients are less dependent on CD28 costimulation than in healthy controls, suggesting that these T cells were previously primed in vivo.⁹

Astrocytes are subject to regulatory influences that serve to adaptively prohibit the initiation of CNS immune responses. Such influences are likely to be a constituent of the regulatory apparatus that constitutes CNS immunoprivilege. Astrocytes express a wide range of neurotransmitter receptors, and we are only beginning to understand their roles.^{43 - 44} Of particular interest is that some neurotransmitters, including norepinephrine, glutamate, and vasoactive intestinal polypeptide, are involved in suppressing the induction of MHC class II and adhesion and costimulatory molecules on the astrocyte membrane.^{45 - 48} This may explain the absence of MHC class II and costimulatory molecules expression on astrocytes in vivo under normal and many pathologic conditions, including animals with experimental allergic encephalitis.⁴⁹

It has been shown that norepinephrine dose dependently inhibits interferon γ–induced MHC class II antigen expression on astrocytes in vitro through β₂-adrenergic signal transduction mechanisms.^{45 - 46} This effect could be attenuated by the β-adrenergic receptor antagonist propranolol but not by the β₁-selective antagonist atenolol or the α-adrenergic antagonist phentolamine. The direct β₂-adrenergic receptor agonist isoproterenol produced the same effect as norepinephrine. Taken together, all these results provided compelling evidence that β₂-adrenergic receptor signaling was responsible for norepinephrine-mediated inhibition of MHC class II expression on astrocytes (Figure 1). It was also recognized that β₂-adrenergic receptor mechanisms were mediated through the activation of adenylate cyclase with the consequent escalation of intracellular cyclic adenosine monophosphate (cAMP). Direct administration of the cAMP analogue dibutyryl-cAMP to astrocyte cultures could recapitulate the inhibitory effects of norepinephrine on interferon γ–induced MHC class II expression.⁴⁵ Extending these observations, the phosphodiesterase inhibitor dipyridamole, which inhibits the breakdown of cAMP in astrocytes, produced a similar inhibitory effect.⁴⁵

How does β₂-adrenergic receptor activation inhibit interferon γ–induced MHC class II expression? Current evidence indicates that interferon γ induces MHC class II expression in astrocytes at the transcriptional level by activating the type IV class II transactivator promotor (CTIIA).^{50 - 51} Norepinephrine may act through the inducible cAMP early repressor, which suppresses the activity of the cAMP response binding element protein that cooperates with CTIIA-mediated MHC class II expression.^{52 - 53}

Norepinephrine does not cross the blood-brain barrier. The norepinephrine acting at astrocytic β₂-adrenergic receptors may, at least partly, be derived from the noradrenergic innervation of the CNS microvasculature, which also contacts astrocytes.^{54 - 56} The cells of origin in this system are contained in the locus ceruleus within the dorsorostral pontine tegmentum. A pulsatile release of norepinephrine through this diffuse ascending adrenergic projection system is likely important for the normal regulation of the CNS immunoprivileged state. Pulsatile treatment of astrocytes with norepinephrine produced a greater degree of inhibition of MHC class II expression compared with single-dose exposure.⁴⁵ However, it was found that on increasing the concentration of interferon γ, there was less capability for norepinephrine to down-regulate this response. Thus, astrocytes may become facultative antigen-presenting cells when proinflammatory signals such as interferon γ exceed the ability of inhibitory signals such as norepinephrine to down-regulate astrocyte-mediated immune responses and effector mechanisms.

It was recently discovered that β₂-adrenergic receptors were absent on astrocytes in MS brain tissue.^{57 - 58} This observation was not limited to areas of active or inactive plaque lesions but was also observed in the normal-appearing white matter.⁵⁷ In contrast to astrocytes, neurons continued to show a normal distribution of β₂-adrenergic receptors.^{57 - 58} The loss of these receptors likely facilitates the deviation of astrocytes to function as facultative immunocompetent CNS antigen-presenting cells (Figure 1).

Repeated exposure with activated T cells and their proinflammatory cytokines, without the counterbalance of appropriate astrocyte MHC class II down-regulation, would then trigger unregulated inflammatory responses, leading to focal areas of myelin, oligodendrocyte, and axonal damage. This would occur only from time to time in limited areas of the CNS when the proinflammatory signals exceed the ability of inhibitory circuitry to down-regulate the immune responses and effector mechanisms by astrocytes.

The absence of astrocytic β₂-adrenergic receptor expression in normal-appearing MS white matter without any evidence of inflammation and the findings that β₂-adrenergic receptors on astrocytes normally up-regulate in other types of CNS injury^{57 ,59 - 60} militate against the hypothesis that the loss of these receptors in MS is simply an epiphenomenon. Interestingly, an animal model of inflammatory demyelination that involves the loss of β₂-adrenergic receptors further corroborates this contention. Canine distemper (CD) virus primarily infects astrocytes and causes a demyelinating disease in dogs that closely resembles MS.⁶¹ In dogs with CD encephalitis, β₂-adrenergic receptors were normally present on neurons but were lost on astrocytes in both acute and chronic demyelinated lesions and also in normal-appearing white matter.⁶² Astrocytic β₂-adrenergic receptors were preserved in other forms of encephalitis.⁶² Similar to MS, several astrocytes at the boundary of the demyelinated lesions in CD encephalitis expressed MHC class II antigens. The CD virus has been shown to reduce the density of β₂-adrenergic receptors in rat glioma cells.⁶³ Perhaps a similar infection by one or more viruses, in genetically predisposed individuals, might be the cause of the loss of β₂-adrenergic receptors on astrocytes in MS.

It is likely that microglia and astrocytes participate in a complex coordinated interplay of interactions that culminate in transitions from proinflammatory to immunoregulatory phases of the disease process in MS. By virtue of their positional relationships to the cerebrovascular endothelium, via end feet, astrocytes may provide the sensing apparatus that represents a scaffolding across the blood-brain barrier interface that could provide for the sampling of antigens and their subsequent presentation to immune cells. Such processes influence lymphocyte recruitment and redistribution of these cells into the CNS through the elaboration of cytokines and chemokines derived from astrocytes and may constitute the earliest events in the development of inflammatory demyelination. It has recently been suggested that astrocytes may undergo a phenotypic switch from a cell that exerts protective reparative and maintenance functions for the oligodendrocyte-axonal-synaptic apparatus to one that can mediate inflammatory events, compromise the integrity of the conduction properties of axons, and ultimately lead to gliosis and neurodegeneration.⁶⁴ Astrocytes can be activated and observed to produce nitric oxide synthase and NADPH diaphorase activity with the consequent induction of proinflammatory cytokines, such as interferon γ, interleukins 1 and 6, and tumor necrosis factor α.^{48 ,65 - 69}

We suggest that the initial inflammatory phase of MS may occur consequent to astrocyte activation. This hypothesis is provocative, and we clearly need more research to obtain a better view on the respective roles played by microglial and astroglial cells in the pathophysiology of MS. However, the observation that astrocytes in MS selectively lose their β₂-adrenergic receptors suggests that it represents a specific regulatory defect that may be germane to understanding the underlying disease mechanism. It may predispose MS patients to waves of focal inflammation because it leads to an inappropriate counterbalance of inhibitory regulation. Recurrent episodes of inflammation may lead to determinant spreading and a broader diversity of immune responses against CNS antigens. With the uncovering of a broader complement of sequestered epitopes in proximity to activated astrocytes, microglia, and macrophages, a greater expansion of immune responses ensues, producing a perpetual cascade of autoimmune attacks on the CNS.

Further work is required to investigate both the cause and significance of the acquired loss of β₂-adrenergic receptors on astrocytes in MS. Infectious and genetic factors must be explored. If the β₂-adrenergic receptor defect can be elucidated, it should be possible to restore the regulatory mechanisms of astrocytes and promote a phenotypic switch back to a more protective and adaptive functional state of these cells.

Corresponding author: Jacques De Keyser, MD, Department of Neurology, Academisch Ziekenhuis Groningen, Hanzeplein 1, 9713 GZ Groningen, the Netherlands (e-mail: j.h.a.de.keyser@neuro.azg.nl).

Accepted for publication June 21, 2002.

This study was supported by Nederlandse Organisatie voor Wetenschappelijk Onderzoek, Hersenstichting Nederland, and Serono Benelux, The Hague, the Netherlands, and the Yellow Rose Foundation and the Hawn Foundation, Dallas, Tex.