The Autoimmune Concept of Multiple Sclerosis

Multiple sclerosis (MS) is a neurodegenerative disease with unresolved pathogenesis. Up to now, an autoimmune attack against myelin autoantigens is considered to be the primary and main event in the pathogenesis of MS (Hemmer et al. 2003; Pender and Greer 2007). Molecular mimicry has been considered to be the potential trigger for central nervous system (CNS) autoimmunity (Chastain and Miller 2012). The anti-inflammatory therapeutic approach is in line with these assumptions (Hohlfeld and Wekerle 2004; Steinman 2007). It is presumed that the immune therapy is able to prevent secondary neurodegeneration. Therefore, an early onset of immune-directed therapy has been recommended (Coyle and Hartung 2002; Hartung et al. 2005).

Paradigm Shift

Recently, the autoimmune concept for MS pathogenesis has been challenged from different directions: (i) Despite multiple attempts to substantiate a role of myelin autoantigens in MS, no unequivocal proof has been supplied for this idea. In fact, T-cells that specifically recognize peptides of myelin proteins have been shown to be equally present in healthy subjects and in MS patients (Pette et al. 1990; Holmøy 2007). Evidence is also lacking for the hypothesis that a defective immune regulation would play a role in MS (Goverman 2009). (ii) Several lines of evidence suggest that inflammation in MS could be a secondary phenomenon. No immune cells have been found by histopathology in developing MS lesions at the onset of an MS relapse (Barnett and Prineas 2004). The latter finding has been limited to few cases that had been available for pathological analysis. It has been discussed, however, that the MS plaque may be subjected to some kind of metamorphosis (Kornhuber 2006) with inflammation as a secondary event: 3 months prior to the onset of contrast enhancement, subtle but significant alterations take place at the plaque site in the brain tissue as detected by magnetization transfer imaging (Filippi et al. 1998). As contrast enhancement is taken as a correlate of inflammation (Smith et al. 1993), it seems to be likely, that non-inflammatory events occur within the MS plaque before immune cells come into play. In contrast, inflammation parameters in the blood display quite a different time course. They rise within a few weeks before the acute attack (Beck et al. 1988). This rise could be due to various types of systemic infections, which are well known to contribute to an acute attack (Correale et al. 2006). Lesion metamorphosis would also be in line with different types of plaques that have been detected by histopathology (Lucchinetti et al. 2000). Thus, different types of MS need not necessarily be taken into consideration to explain this phenomenon. In fact, the postulation of diverse separate subtypes of MS disagrees with the results of epidemiologic studies which suggest that all known clinical subtypes are part of one and the same disease process (Confavreux and Vukusic 2006a). A primary neurodegenerative process would also fit the finding that 20 % of the axons in the cerebral white matter have been lost already when MS is diagnosed. N-acetyl-aspartic acid has been shown to be significantly reduced in the cerebral white matter of MS patients with early disease compared to control subjects by magnetic resonance spectroscopy (De Stefano et al. 2002; Filippi et al. 2003). This diffuse axonal loss can hardly be attributed to the few MS lesions that are commonly present on magnetic resonance images in this early state of the disease (De Stefano et al. 2002; Filippi et al. 2003). In addition, axonal loss and demyelination have been demonstrated to progressively decrease irrespective of an ongoing immunomodulatory therapy (Anlar et al. 2003; Parry et al. 2003).

Immune Therapy and Progression of the Disease

Excess mortality of patients with MS as compared with the general population has declined significantly since 1950 (Llorca et al. 2005, Brønnum-Hansen et al. 2006). An influence of immune therapy on this phenomenon has not yet been unequivocally established, and there may be further factors of influence such as improvements of economic status and general health care including symptomatic MS treatment. Provided that inflammation is the driving force of the disease, disability would follow the inflammatory activity. Such correlations are weak at best, and not given when the disease progresses steadily (Confavreux et al. 2003). Moreover, the inflammatory activity in MS does not exert a detectable influence on the ongoing process of cerebral atrophy (Cheriyan et al. 2012). It is common knowledge that MS patients with a progressive course from onset do not benefit from any immunomodulatory therapy. Actually, the impact of immune modulation in MS segregates into a moderate effect on relapses and a lacking or at most weak effect on disease progression (Kornhuber et al. 2005). Accordingly, it has been shown, that a longer lasting therapy with interferon beta has no influence on disability, neither in patients with a relapsing-remitting type of disease (Confavreux and Vukusic 2006b; Shirani et al. 2012), nor in patients with secondary progression (Confavreux and Vukusic 2006b; La Mantia et al. 2012). This finding has been attributed by some authors to the relatively short study periods or to imprecise clinical scales (Hohlfeld 2012). However, if MS is considered to be a primary neurodegenerative disease, the failure of immune therapy is consistent.

Alternative Hypothesis: HERV/Superantigen Pathogenesis

In our eyes, the hypothesis of MS as a primary autoimmune disease is hardly tenable any more. Therefore, alternative concepts have to be developed that better fit the real findings in MS patients that have been gathered over the last decades. Actually, simple and attractive views in agreement with the dualism of degeneration and inflammation in MS are already available: A clear and reproducible association of human endogenous retroviruses (HERVs) in MS-patients has been shown previously (for review see e.g., Perron and Lang 2010; Nexø et al. 2011). In 1989, Perron and coworkers detected retroviral activity in an MS patient (Perron et al. 1989). Consequently, the concept of an MS-associated retrovirus (MSRV) has been developed (Perron et al. 1997). Today, of the more than 30 HERV-families, few have been considered to be associated with MS (Perron and Lang 2010; Christensen 2010; Nexø et al. 2011; Tai et al. 2008).

At least 8 % of the human genome is composed of endogenous retroviral sequences. These sequences were integrated into the human genome in the course of the evolution and are now transmitted from generation to generation like other genes. Some of these sequences are involved in normal physiological functions (Dupressoir et al. 2012). HERVs have been found to be associated with different diseases, e.g., endogenous psychoses, psoriasis, diabetes mellitus type 1, rheumatoid arthritis and diverse malignant tumors (Dolei 2006; Balada et al. 2009). The majority of HERVs integrated in our genome is not competent to replicate and most HERV sequences are presumably silent. Thus, harmful properties of HERVs possibly depend on the circumstances and the organ in which they are expressed. In comparison to healthy individuals, MS patients have been shown to display genetic differences in the promoter region of HERV-Fc1 (Nexø et al. 2011). Such differences might contribute to the differential expression of HERVs in MS patients as compared with healthy subjects (Nexø et al. 2011). Other factors have been reported to influence HERV expression (Table 1). For example, estradiol primes human breast cancer cells for subsequent progesterone-induced HERV-K expression (Ono et al. 1987). Similarily, it was shown that HERV expression in keratinocytes increases under the influence of UV light (Hohenadl et al. 1999). Due to the impact of endogenous and exogenous factors in MS, it would be worth to investigate further such factors (sexual hormones, light exposure) in the context of HERV expression. Environmental parameters could also play a role in regard to the distribution of MS prevalence in different geographical regions.

Table 1 Activators of HERV expression

It is well known that the gammaherpesvirus Epstein Barr virus (EBV) plays a role in MS. Actually, MS has been shown to become manifest within a few years after EBV-infection (Levin et al. 2010). When HERVs are thought to be involved in the pathogenesis of MS, there should be some link between EBV and HERV activation. Indeed, HERVs can be transactivated by herpes viruses including Herpes simplex virus 1, cytomegalovirus and EBV (Sutkowski et al. 2001; Nellåker et al. 2006; Assinger et al. 2013). The potent transactivating activity of HSV has been attributed to the HSV alpha gene product immediate early protein 0 (ICP0) (Everett 1984; Gelman and Silverstein 1985; O’Hare and Hayward 1985).

Furthermore, it was shown that ICP0 of HSV-1 transactivates the LTR-directed transcription of the human endogenous retrovirus K (Kwun et al. 2002). In addition to molecular mimicry and epitope spreading triggered by an excessive and prolonged immune stimulation, transactivation of HERVs by EBV and subsequent superantigenic stimulation of T cells might be involved in autoimmune phenomena (Dreyfus 2011). Transactivation by HSV-1 of the retrovirus of the LM7 cell line that was later termed MSRV has been shown as early as 1993 (Perron et al. 1993).

T-Cell Superantigens

If HERVs would play a causative role in MS, all MS features should fit into this scenario. First of all, a cellular inflammation takes place in MS plaques. Actually, gene products encoded by endogenous retrovirus sequences have been shown to induce clonal deletion of lymphocytes in a V-beta specific manner resembling that known for superantigens (SAgs) (MacDonald et al. 1988). It has been shown previously that the envelope protein of MSRV leads to polyclonal expansion of Vbeta16 T-lymphocytes in vitro (Perron et al. 2001). Furthermore, SCID-mice engrafted with human peripheral blood mononuclear cells intraperitoneally developed T-lymphocyte dependent brain hemorrhage albeit without encephalitis after i.p. injection of MSRV-virions (Firouzi et al. 2003). Of course, superantigens have been tested in the context of experimental autoimmune encephalomyelitis (EAE). However, controversial results have been presented with both, augmentation or attenuation of EAE (Rott et al. 1992; Perron et al. 2013). To test, if SAgs could induce inflammation similar to that in MS within central nervous tissue directly and not by the detour of EAE, we have previously used bacterial exotoxins (Kornhuber et al. 2002; Emmer et al. 2008, 2010). In healthy rodents without prior immune stimuli, SEA present locally within the brain does not induce a major inflammatory response (Kornhuber et al. 2002). The latter can be markedly amplified, however, after i.v.-injection of spleen cells that had been non-specifically activated by Concanavalin A (Kornhuber et al. 2002). Similarly, acute attacks in MS could be triggered by non-specific immune stimuli i.e., stress or infectious diseases. Interestingly, CD8-positive T-cells dominate the SEA-driven perivascular round cell reaction (Emmer et al. 2010). This type of lymphocyte has been found to predominate also in inflammatory MS lesions (Friese and Fugger 2009). The oligoclonal expansion of T-cells in the cerebral tissue of MS patients (Junker et al. 2007) appears to accord to a superantigenic stimulus. Furthermore, inflammation is detectable in non-myelinated areas such as the retina. While this finding is well in accord with a superantigenic stimulus, it would not be expected when a myelin auto-antigen is involved in the MS pathogenesis. Again, no separate disease entity can be proposed based on whether the retina is involved or not in MS patients. Thus, it has been shown, that MS patients with or without retinal involvement do not show any differences (Schmidt et al. 2001).

B-Cell Superantigens

Beside cellular inflammation, a humoral immune reaction is an essential feature of MS: oligoclonal bands (OCB) can be detected in over 90 % of MS-patients by isoelectric focusing in the cerebrospinal fluid. It has been reported, that only 1 % of these OCB contain antibodies that are directed against CNS antigens (Kaiser et al. 1997; Owens et al. 2009). In contrast, the spectrum of antigen specificities that is present in OCB is so diverse, that the term “nonsense antibodies” has been used for this phenomenon (Mattson et al. 1980). If SAgs play a role in the pathogenesis of MS, evidence should be presented that OCB may be induced by them. In fact, preliminary results support the notion, that B-cell SAgs can induce OCB in in vitro cultures of peripheral blood leukocytes from human blood donors (Emmer et al. 2011). Thereby, the envelope glycoprotein 120 (gp120) of human immunodeficiency virus (HIV) was capable to activate blood leukocytes to form antibodies against unrelated antigens including Measles virus, Varicella-zoster virus and Rubella virus (Emmer et al. 2011).

Antigen-Presentation

As far as we can see, the effects of HERV-encoded envelope proteins have not been studied in the context of cerebral antigen presenting cells (APCs) such as microglial cells or astrocytes or eventually also monocytes or dendritic cells (DC) that may enter the CNS during states of inflammation (D’Agostino et al. 2012). Nevertheless, interactions of conventional SAgs such as e.g., Staphyloccal enterotoxins or toxic shock syndrome toxin-1 (TSST-1) with different APCs have been reported.

For cerebral APCs only few studies are available so far. Treatment of primary microglia with purified Staphylococcal enterotoxin B (SEB) has been shown to augment the TNF-α production in response to the TLR2 ligand Pam3Cys (Vidlak et al. 2011). SEB and TSST-1 have been reported to bind with high affinity to MHC class II antigen expressing astrocytes (Hassan-Zahraee et al. 2000). Contrasting results have, however, been reported by Rott et al. (1993).

Toxic shock syndrome toxin-1 (TSST-1) has been shown to potently induce IL-1 in human monocytes (Ikejima et al. 1984; Parsonnet et al. 1986). SEA, SEB, and TSST-1 were shown to activate and induce IL6 in peritoneal macrophages from lipopolysaccharide-responsive C3HeB/FeJ mice (Fleming et al. 1991). Furthermore, SEB and TSST-1 have been demonstrated to induce IL-1 and TNF secretion in human monocytes and monocytic cell lines (Trede et al. 1991). Eventually, the activation of monocytes by TSST-1 is mediated by tyrosine phosphorylation (Scholl et al. 1992). The maturation process that has been induced by systemic administration of SEB was shown to be accompanied by upregulation of CD40, CD80 and CD86 expression in splenic interdigitating dendritic cells (IDCs) but not in other APCs such as macrophages and B cells (Yoon et al. 2001). Beside the known binding of SAgs by MHC class II molecules on the surface of APCs, SAgs have been demonstrated to be taken up in DCs by transportation and trafficking (Ganem et al. 2013). It has been speculated that this SAg uptake might increase the local SAg concentration and thus enhance their presentation on the cell surface, e.g., to lymphocytes (Ganem et al. 2013). Furthermore, SAgs up-regulate monocyte surface toll-like receptor (TLR) 2 and TLR 4 expression through MHC class II signaling (Hopkins et al. 2005, 2008). The SAg SEC1 has been suggested to play a role in the differentiation of bovine peripheral blood mononuclear cells into DC (Seo et al. 2009).

Beside stimulatory effects on APCs, SAgs have been shown to lead to APC apoptosis. Thus, SEB was shown to selectively increase the number of apoptotic CD80(−) monocytes, presumably via CD95 dependend pathway (Takahashi et al. 2001). Furthermore, a significant depletion of Langerhans cells has been reported to be induced by SEA or exfoliative toxin but not by TSST-1 (Pickard et al. 1994).

From the above cited reports it becomes apparent that the interaction of SAgs locally expressed in the brain such as HERV-encoded envelope proteins might lead to the activation of cerebral APCs. In fact such activated APCs such as microglial cells are typically seen in the context of widespread demyelination of cerebral tissue of MS patients (Lassmann 2013). By way of contrast, microglia has been found to be not or at most mildly activated during the course of EAE (Vainchtein et al. 2014). Furthermore, the SAg–APC interaction might lead to the liberation of cytokines such as IL1, IL6, and TNF-α. This in turn could lead to reactions of the cerebral endothelial cells and to enhanced recruitment of immune effector cells, i.e., to the initiation of an inflammatory response.

Superantigens and HLA-Association

In MS patients, certain HLA-types have been shown to be significantly more common than in the general population (Ramagopalan et al. 2009). This HLA-association is complex and does hardly allow any conclusion with respect to the cause of the disease. Like other antigens, T-cell-SAgs bind to HLA-molecules which present them to the T-cell receptor in a Vβ-dependent manner. If a role for HERV-encoded SAgs is assumed within the MS pathogenesis, then differences in the HLA-association should be present between patients with a relapsing-remitting type of course and those who show slow progression from the disease onset. Actually, such differences have been repeatedly found (Madigand et al. 1982; Van Lambalgen et al. 1986; Francis et al. 1987; Olerup et al. 1989). Furthermore, an association between the HLA-type and the disease course could be expected especially in those patients who actually show signs of inflammation in form of relapses, and who respond to immune modulating therapy. Contradictory results have been reported regarding this issue. Nevertheless, as a reproducible finding that may hint in this direction, the manifestation age of MS depends on the presence of the HLA DRB1*15:01 allele (Masterman et al. 2000; Sawcer et al. 2011). If the HLA DRB1*15:01 allele is absent, MS becomes manifest at a higher age. If this relation holds true, it could be that the different types of MS course form a continuum in which patients with a primary type of course from onset would experience their first relapse at an advanced age. In this case, it would be difficult to distinguish with certainty a relapse from the ongoing disease process.

HERV and Neurodegeneration

How could primary degeneration in the CNS agree with a HERV-pathogenesis? Oligodendrocyte apoptosis has been detected in MS plaques in an early plaque stage with immune reactivity against caspase 3 (Prineas and Parratt 2012). Apoptosis of oligodendrocytes has been described in the context of diverse viral diseases, e.g., in progressive multi-focal leukoencephalopathy in JC-virus infection (Merabova et al. 2008). Furthermore, axonal loss and demyelinisation is a feature of HIV-encephalopathy (Bell 1998). Certain proteins of HIV may induce apoptosis markers like caspase 3 or TUNEL (Hauser et al. 2009). Overexpression of the HERV-W env has been reported to be associated with elevated levels of small conductance Ca(2+)-activated K(+) channel protein 3 (SK3) in human neuroblastoma cells. This finding has been discussed in the context of neuronal excitotoxicity (Li et al. 2013). These findings suggest that other retroviruses, eventually including HERVs might be capable to induce degeneration of oligodendrocytes or axons, too. Indeed, cytotoxic effects of HERV products on oligodendrocytes have been described (Antony et al. 2004; Kremer et al. 2013). In addition to a direct cytopathic effect of HERV, SAg-activated T cells might be able to destroy bystander cells. Under certain circumstances, strong activation of T cells can induce non-HLA restricted cytotoxic activity against innocent bystander cells (Dick et al. 1993; Staege et al. 1996, 1998, 2000, 2003). Such bystander lysis could contribute to secondary neurodegeneration.

Despite the fact that many researches stick to EAE as their favorite model for MS, the concept of MS as an autoimmune disease driven by anti-myelin autoantigenic mechanisms is hardly tenable any more. Alternatively, a key role in MS pathogenesis might be played by HERVs that are not safely silenced. In fact, such HERVs have been shown to become locally activated in the CNS of MS patients. In this context, it is not far to speculate that HERV-encoded envelope proteins act as SAgs and thereby cause a cerebral inflammatory reaction as has been established for bacterial SAgs already. Notably, the super-stimulatory effects would be expected to concern both, T-cells (inflammatory plaques) and B-cells (oligoclonal bands). Which consequences for MS therapy evolve, if a HERV/SAg pathogenesis is taken into account? (i) If MS is caused by HERVs, immune suppression or modulation can hardly be expected to influence slow disease progression. (ii) A substantial attenuation of neurodegeneration would require effective suppression of HERV-activation. (iii) It is readily explicable that immune therapy can reduce the rate and severity of relapses, while disease progression is hardly influenced. (iv) The extensive and expensive use of immune modulating therapies from the disease onset without taking the disease dynamics and progression characteristics into account seems to be questionable (Fig. 1). (v) The development of ever new immune modulators does not warrant the necessary progress in the sense to avoid disability on the long term. By way of contrast, to develop mechanisms of silencing HERV expression poses an attractive base for a future therapy of this disabilitating disease. If in turn efficacy of such a therapy could be proven in progressive forms of MS, this would strongly support a HERV driven pathogenesis.

Fig. 1
figure 1

A model for HERV/superantigen mediated neurodegeneration and neuroinflammation. HERVs constitute an integral part of our genome. Under normal conditions, expression of HERVs is switched off epigenetically (0). Triggered by diverse factors (see Table 1), reactivation of HERVs expression can be induced (1). The majority of HERV-encoded RNAs contain point mutations and deletions that hinder synthesis of HERV-encoded proteins. Such RNAs might be involved in dysregulation of RNA metabolism (2). HERV-encoded RNAs with intact open reading frames can be translated into proteins (3). In rare cases, complete virions can be formed (4) which might have direct cytopathic effect. In addition, single proteins can act as antigens or superantigens for T cell stimulation (5). Such superantigens can lead to depletion of V beta families but also to activation of proliferation of oligoclonal T cells (6). Activated T cells secrete cytokines which can lead to immune dysregulation (7). In addition, activated T cells might be able to interact and kill other cells including the superantigen expressing cells (8). However, bystander cell killing can be independent from the expression of the superantigen which was responsible for the induction of the immune response (9)