Inflammatory and neurodegenerative responses in the central nervous system (CNS) are strongly affected by infections that occur in the periphery [1]. This is clinically illustrated by the fact that infections are an important risk factor for the development of relapses in multiple sclerosis (MS) [2, 3]. In addition, infections may enhance the severity and/or duration of clinical exacerbations. Evidence from experimental models also demonstrates that systemic administration of bacterial compounds such as LPS or staphylococcal enterotoxin induces clinical relapses in mice recovered from experimental autoimmune encephalomyelitis (EAE) [4, 5]. Infections may cause activation of autoreactive lymphocytes in the periphery in a non-specific manner [5, 6]. Alternatively, bacterial compounds when released into the bloodstream may increase blood–brain barrier permeability [7], subsequently activating microglia, which secrete proinflammatory cytokines promoting neurodegeneration [8]. Recent studies further demonstrate that manipulation of the normal gut microbiota by antibiotics, germ-free conditions and administration of Bacteroides fragilis polysaccharides affects EAE incidence and severity [9–11]. It is conceivable that both exogenous (LPS) and endogenous phosphorylated compounds such as adenosine triphosphate (ATP) at least partly mediate these effects. Together, these findings suggest that patients with MS may benefit from an early control of infections and neutralization of microbial compounds of normal gut flora.

A promising strategy for the neutralization of bacterial endotoxin and pro-inflammatory extracellular nucleotides is treatment with alkaline phosphatase (AP), an enzyme ubiquitously expressed in mammalian tissues and present in body fluids. AP hydrolyzes the diphosphoryl lipid A moiety of LPS, generating the non-toxic monophosphoryl lipid A [12]. Endogenous AP plays a role in the defense against Gram-negative bacteria [13] and is pivotal for normal gut homeostasis [14, 15]. AP has beneficial effects in several animal models of inflammatory diseases, including sepsis, inflammatory bowel disease and colitis [16–18].

In addition to detoxification of exogenous compounds, such as LPS and bacterial CpG [19], AP also deactivates endogenous molecules such as ATP, which serves as an immunological danger signal when present at high concentrations (>100 μM) in the extracellular space [20]. ATP, produced by bacteria and released in large concentrations from damaged cells, is sensed by purinergic P2 receptors [21]. In the intestinal lamina propria, ATP is critical for the differentiation of Th17 cells [22].

CD39 and CD73 are two other enzymes that are involved in ATP metabolism, thus having overlapping functions with AP. CD39, like AP, mediates the conversion of ATP via ADP to AMP. CD73 and AP both convert AMP into adenosine. The two enzymes are expressed by regulatory T cells (Treg) and are crucial for the immunosuppressive function of Treg by decreasing local ATP concentrations and increasing the immunosuppressive adenosine [23]. In addition, CD73 expression and adenosine signalling is pivotal for leukocyte entry into the CNS of mice with EAE [24].

Because infections often precede MS relapses and given that AP detoxifies endogenous and exogenous innate activating signals, we hypothesize that AP has a beneficial role in limiting neuroinflammation in MS. In this study we therefore aimed to: 1) determine the prophylactic and therapeutic potential of AP in EAE, a mouse model of MS; 2) determine plasma AP levels in MS patients in relation to endotoxin exposure; and 3) assess the expression and cellular sources of AP in relation to CD39 and CD73 in MS brain tissue.

MS exacerbations are often associated with preceding infections [2, 3]. Moreover, recent experimental studies demonstrate that exposure to LPS or the commensal microbiota is necessary for the development of neuroinflammatory disease in mice with myelin-reactive T cells [5, 11]. These findings strongly suggest that microbes play a crucial role in the initiation and propagation of inflammatory responses that mediate CNS pathology. This may occur by activation of autoreactive T cells in the periphery [5] or through activation of microglia that initiate inflammatory responses in the brain parenchyma [31]. In the latter scenario, systemic inflammation leads to a phenotypic switch of microglia, from anti-inflammatory to proinflammatory [8, 31]. Importantly, preactive MS lesions, characterized by activated microglia, do not always develop into demyelinating lesions [32] and it is tempting to speculate that systemic infections could act as a driving force of the development of demyelinating lesions. Irrespective of the mechanism, it can be argued that MS patients may benefit from neutralization of microbial compounds and endogenous danger signals as an attempt to prevent subsequent propagation of inflammatory responses in the CNS. In this study we hypothesized a protective role for AP through its potent detoxifying activity of LPS and endogenous danger signals, such as ATP. We demonstrate that pre-symptomatic AP administration to animals in which autoimmunity to MOG peptide was induced, resulted in the reduction of clinical signs of EAE.

There are several mechanisms that could explain how AP treatment resulted in a reduction of clinical severity in EAE (Figure 6). First, AP may neutralize the LPS present in the adjuvant and needed for the priming of pathogenic MOG-reactive T cells. In experimental colitis induced with dextran sodium sulfate, the LPS-neutralizing activity of AP is regarded as the principal mode of action [17]. Second, AP may detoxify endogenous substrates, such as ATP, and hence interfere with CD4⁺ T cell activation, as ATP was recently shown to provide autocrine signals for activated T cells [33]. Our finding that splenocytes of AP-treated mice proliferated less efficiently to MOG-peptide is in support of both mechanisms. The overall proliferation capacity of splenocytes was not affected by in vivo AP treatment; however, it must be noted that the proliferation to PHA was measured after three days and not at the optimum of one day.

AP treatment may also act to limit inflammatory responses in the brain. ATP has been shown to mediate the microglial response to local brain injury [34] and has excitotoxic effects on oligodendrocytes [35]. We did not find a beneficial role for AP treatment (via the i.p. route) during the acute or chronic phase of disease. However, it is still possible that local administration of AP can interfere with microglial activation and oligodendrocyte death. The lack of clinical improvement after AP administration during the acute or chronic phase of disease may also be explained by the fact that the plasma residence time of the applied bovine AP is relatively short (in the order of minutes to a maximum of two hours [16]). Thus, ongoing inflammation is only in-part and transiently modulated. In contrast, the neutralization of endotoxin and ATP by the short-acting AP may have more impact at the onset of inflammation, as observed in experimental models of inflammatory bowel disease [18].

The efficacy of AP has been studied in several human inflammatory diseases and conditions. AP treatment was safe [36, 37] and was associated with short-term clinical improvement and reduction of C-reactive protein in ulcerative colitis [38]. AP treatment also resulted in more prominent recovery of creatinine clearance in patients with sepsis-induced acute kidney injury [37]. The safety and efficacy of short-term AP treatment is currently under investigation for acute rheumatoid arthritis (ClinicalTrials.gov Identifier: NCT01416493). Animal models in which AP was shown to be effective were mostly acute conditions, including septic shock, acute myocardial infarction and peritonitis (summarized in Table 3). However, beneficial effects have also been reported in chronic colitis [17, 39].

We also determined the expression of AP and its related enzymes CD39 and CD73 in MS brain tissue. The observed AP expression pattern was largely similar to that in rodents and primates as reported previously [40, 41]. In contrast to Alzheimer’s disease [42], we did not find increased AP expression in MS brain compared to controls. It is possible that enzyme histochemistry as used in this study is less sensitive in detecting differences than enzyme assays of whole tissue homogenates. CD39 was predominantly expressed on blood vessel endothelium, as was AP, and on microglia, which is consistent with an earlier study [40]. In white matter lesions of MS patients, CD39 expression was also found on microglia. CD39 may play a role in the microglial response to inflammation and tissue damage as it converts extracellular toxic ATP, which is released upon tissue damage, to ADP and AMP. Not all HLA-DR⁺ cells co-expressed CD39, especially those that surrounded capillaries and some cells in the center of the lesion, suggesting that CD39 is differentially expressed by (perivascular) macrophages and microglia. In the periphery, CD39 is also expressed by Treg cells [43]. CD39⁺ Treg cells were recently shown to suppress pathogenic Th17 cells and were reduced in peripheral blood of MS patients [44].

In contrast to AP and CD39, human cerebral endothelium did not express CD73. Instead, we found CD73 expression at the border of the Virchow-Robin space, which may reflect CD73⁺ astrocyte endfeet at the glia limitans [45]. CD73 was also evident as punctate staining in HLA class II⁺ cells, most likely representing macrophages or microglia that have phagocytosed myelin, which is known to contain CD73 [40]. Collectively, the expression patterns of AP, CD39 and CD73 in MS lesions further underscore the importance of extracellular nucleotides, which were previously reported to drive leukocyte entry into the brain [24], mediate the suppressive action of Treg cells [43], and signal microglia to respond to tissue damage [46].

The production of AP in vivo is increased upon LPS challenge [47, 48], and, also, IgM and IgG antibodies to the core domain of endotoxin are modulated by exposure to LPS [49]. To determine the need for AP supplementation in MS, we measured AP activity and determined endotoxin exposure by direct and indirect methods. We found no significant differences in AP activity and isoenzyme frequencies in MS patients compared to controls, although AP activity and AP isoenzyme distribution were slightly altered in PP-MS patients. Another study also addressed AP activity in RR-MS patients and reported comparable AP activity levels in MS versus healthy controls [50]. To our knowledge there are no other reports on AP activity in PP-MS patients. Parallel to an increase in AP levels in PP-MS, we detected a significant decrease of Endocab IgM levels in PP-MS patients compared to RR-MS patients and healthy controls, a finding that may be explained by differences in age as we found a significant negative correlation between age and Endocab IgM levels. In other human disease conditions, decreased Endocab IgM levels have been interpreted as consumption of antibodies caused by systemic release of gut endotoxin, for example, after cardiac surgery [49]. Similarly, low base-line levels of Endocab IgM and IgG are associated with increased mortality and prolonged hospitalization after surgery [51] and with the development of systemic inflammatory response syndrome (SIRS) in children with organ failure that occurred post-operatively or after head injury [52]. In contrast, elevated levels of IgG Endocab are found in Crohn’s disease, perhaps reflecting chronic exposure to endotoxin due to impaired gut barrier function [53]. Since the Endocab IgG levels in PP-MS patients were comparable to controls, it is unlikely that PP-MS patients have been chronically exposed to endotoxin, at least to the rough-type of LPS used in the Endocab assay.

In conclusion, using the EAE model we found that administration of AP during the priming phase was effective in reducing clinical severity and proliferation of T cells in response to MOG35-55. We also demonstrated a strong expression of the AP-like enzyme CD39 in MS lesions, as a possible reflection of the microglial response to inflammation and tissue damage. Although the number of patients that were used to make comparisons and correlations in our study is modest and confirmation of the findings is required in an independent cohort, our results do suggest that MS patients may have sufficient circulating AP levels and neutralizing antibodies to inactivate and/or clear endotoxin. Nevertheless, upon microbial exposure or gut barrier dysfunction, when endotoxin is released into the systemic circulation, peripheral leukocytes could still be activated in a non-specific manner driving neuroinflammatory disease. The interrelationships between infections, microbial translocation of gut microbiota compounds and MS disease activity, therefore, warrants further investigation.