Galactosylsphingosine (psychosine)-induced demyelination is attenuated by sphingosine 1-phosphate signalling

Globoid cell leukodystrophy (Krabbe disease) is a rare autosomal recessive neurodegenerative disorder affecting 1:100,000 live births in the United States (Wenger et al., 1997). This lysosomal disorder typically has an early onset, is rapidly progressing and is invariably fatal in infants. The vast majority (85–90%) of cases are of the infantile form, with the juvenile and adult-onset forms being considered extremely rare (Wenger et al., 1997). The hallmark symptoms of the infantile form include irritability, hypersensitivity, psychomotor arrest and hypertonia. This is followed by rapid mental and motor deterioration, seizures and optic atrophy. Death usually ensues within the first two years of life and there is currently no cure (Davenport et al., 2011). Krabbe disease is caused by a mutation in the lysosomal enzyme galactosylceramidase (GALC) (Suzuki, 2003). This GALC deficiency results in the accumulation of a toxic lipid metabolite psychosine (galactosylsphingosine) and, to a lesser extent, β-galactosylceramide (Giri et al., 2002). Pathological features of Krabbe disease include profound demyelination and almost complete loss of oligodendrocytes in the white matter, reactive astrocytosis and infiltration of numerous multinucleated macrophages termed ‘globoid cells’. These globoid cells accumulate around blood vessels and in the regions of demyelination and are a unique feature of Krabbe disease (Suzuki, 2003). Progressive accumulation of psychosine in the brains of Krabbe disease patients is thought to be the critical pathogenic mechanism of this illness (Davenport et al., 2011). In some cases the levels of psychosine rise more than 100-fold, from sub-nanomolar concentrations to those in the micromolar range (Svennerholm et al., 1980). In Krabbe disease, the high levels of psychosine escape from lysosomes and dying cells forming aggregates (Orfi et al., 1997). Several reports have demonstrated that psychosine causes direct cellular cytotoxicity by mechanisms that include mitochondrial dysfunction (Haq et al., 2003), caspase activation, alteration of lipid rafts and modulation of the protein kinase C (PKC), Jun N-terminal kinase (JNK) and NFκB signalling pathways (Davenport et al., 2011; Haq et al., 2003; Yamada et al., 1996). Inflammation is also now accepted to play an important role in the pathogenesis of Krabbe disease (LeVine and Brown, 1997). Inflammatory molecules, such as AMP-activated protein kinase (AMPK), prostaglandin D, inducible nitric oxide synthase (iNOS) and pro-inflammatory cytokines have all been implicated in Krabbe disease as well as in the twi/twi mouse animal model of this disease (Giri et al., 2008). Taken together, this increasing evidence now suggests that the loss of oligodendrocytes and widespread demyelination seen in Krabbe disease is due to apoptotic processes as well as aberrant inflammatory response (Giri et al., 2008,, 2006; Haq et al., 2003; Tohyama et al., 2001).

The GALC enzyme, mutated in Krabbe disease, is involved in the complex pathway of sphingolipid metabolism, which includes bioactive lipids such as ceramide, sphingosine and sphingosine 1-phosphate (S1P), all of them particularly important in regulating neural cell function. In particular, the family of S1P receptors (S1PRs) are G-protein coupled and expressed in a range of cells, including those of the immune, cardiovascular and central nervous systems (Dev et al., 2008; Fyrst and Saba, 2010). These receptors are targets for the drug pFTY720 (fingolimod), which is the first oral therapy for relapsing remitting multiple sclerosis (Kappos et al., 2010). The proposed mechanism of action for pFTY720 is reported as being dependent on internalisation of S1PRs in T cells, which limits their S1P-mediated egress from lymph nodes and thus attenuates the inflammatory response in the brains of multiple sclerosis patients (Adachi and Chiba, 2008). Importantly, a number of studies have now demonstrated that compounds such as pFTY720 can also regulate neuronal and glial cell function (Balatoni et al., 2007; Choi et al., 2011; Fischer et al., 2011; Osinde et al., 2007). Indeed, we and others have shown that S1PRs regulate a number of intracellular signalling pathways in astrocytes and promote astrocyte migration (Mullershausen et al., 2007,, 2009). In addition, modulation of S1PRs also promotes oligodendrocyte differentiation and survival (Dev et al., 2008; Miron et al., 2008b). S1PRs have also been shown to limit events of demyelination and promote remyelination, which are probably mediated by the dampening of pro-inflammatory cytokine levels (Miron et al., 2010; Sheridan and Dev, 2012). Overall, therefore, S1PRs represent an important drug target that can be exploited for use in neuroinflammatory, demyelinating and neurodegenerative diseases, as documented by a growing body of literature (Asle-Rousta et al., 2013; Deogracias et al., 2012). Here we investigate whether regulation of S1P signalling alters psychosine-induced astrocyte dysfunction, pro-inflammatory cytokine release and demyelination.

The modulation of psychosine on astrocyte cell function is less well studied in Krabbe disease than the effect of psychosine on oligodendrocytes and myelination. We therefore first investigated the effects of this toxin on human astrocyte survival and also demonstrated the protective effects of pFTY720. Cultured human astrocytes were serum starved for 4 h and then pre-treated with pFTY720 for 1 h before treatment with psychosine, at the timepoints and concentrations indicated (Fig. 1A). Psychosine reduced human astrocyte numbers in a concentration-dependent manner: psychosine treatment for 4 h significantly reduced astrocyte cell survival (10 µM, 55.4±3.8%; 15 µM, 44.3±2.7%; and 20 µM, 39.8±1.8%, compared with control; mean±s.e.m.). Importantly, pre-treatment with 1 µM pFTY720 significantly attenuated the psychosine-induced cell death (reduced by 9.8±4.1%, 21.1±2.5% and 18.4±2.2%, respectively) (Fig. 1B). The psychosine (10 µM)-induced decrease in survival of human astrocytes was also observed to be time dependent (2 h, 79.2±3.2%; 4 h, 63.5±2.4%; 6 h, 37.1±3.3%) and again significantly attenuated in the presence of pFTY720 (reduced by 17.4±4.6% at 4 h and 23.1±4.3% at 6 h) (Fig. 1C). These effects of pFTY720 were also concentration dependent: 1 µM pFTY720 significantly increased cell survival by 34.3±3.2%, compared with 10 µM psychosine treatment alone (Fig. 1D).

During the course of our experiments demonstrating psychosine-induced astrocyte cell death and reversal by pFTY720 (Fig. 1A–D), we noted these effects were also dependent on the density of cultured astrocytes. To quantify these observations, human astrocytes were seeded at densities ranging from 1.2×10⁵ to 6×10⁵. The cells were then pre-treated with 1 μM pFTY720 followed by 10 μM psychosine treatment for 2 h and imaged (Fig. 2A). Human astrocytes seeded at low densities of 1.2×10⁵ and 2.4×10⁵ were most sensitive to psychosine insult, with 81.5±6.2% and 68.5±5.1% cell death occurring after 2 h treatment. Astrocytes seeded at a density of 3.9×10⁵ displayed 39.4±5.3% cell death for psychosine treatment alone, which was significantly attenuated to 20.6±3.6% in the presence of pFTY720 (Fig. 2B). At the two highest densities, 4.8×10⁵ and 6×10⁵, 10 μM psychosine treatment for 2 h did not induce overt astrocyte toxicity, with cell densities of 94.4±3.7% and 99.9±5.4%, respectively (Fig. 2C).

Increasing evidence now suggests the involvement of apoptosis as a mechanism underlying oligodendrocyte cell death seen in Krabbe disease, a disorder that involves mitochondrial cytochrome c release, alterations in electron transport and loss of mitochondrial membrane potential (ΔΨm) (Haq et al., 2003). Here, the ΔΨm was measured using the membrane-permeant dye tetraethylbenzimidazolylcarbocyanine iodide (JC-1), which exhibits potential-dependent accumulation in mitochondria. At high ΔΨm this dye forms aggregates yielding an emission at 590 nm (red) whereas at low ΔΨm JC1 is primarily in monomeric form yielding an emission at 530 nm (green) (Fig. 3A). Cultured astrocytes were serum starved and pre-treated for 1 h with pFTY720 (1 µM) before treatment with psychosine (5 µM, 10 µM, 15 µM, 20 µM). Cells were then loaded with 1 μM JC-1 and after 30 min the emission spectra were measured. Psychosine treatment of mouse astrocytes decreased the aggregate:monomer ratio of JC-1, indicating loss of ΔΨm compared with control (Fig. 3B). Notably, treatment with pFTY720 attenuated the psychosine-induced decrease in aggregate:monomer ratio of JC-1, returning the ΔΨm close to control levels. These findings were supported by use of the MTT colorimetric cell viability assay (based on the reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) into formazan crystals by metabolically active cells), which showed that psychosine caused a significant concentration-dependent reduction of rat astrocyte viability (10 µM, 73.3±2.7%; 15 µM, 63.5±3.6%; and 20 µM, 49.9±1.7%, compared with control; mean±s.e.m.). In agreement, pFTY720 significantly attenuated this psychosine-induced cell death by 14.4±1.7% and 17.3±2.5% at 10 µM and 20 µM psychosine concentrations, respectively (Fig. 3C).

The expression of several pro-inflammatory cytokines and chemokines has been observed in the twi/twi mouse model (LeVine and Brown, 1997). Psychosine has also been seen in vitro to markedly potentiate the LPS-induced production of pro-inflammatory cytokines in primary rat astrocyte cultures (Giri et al., 2002). Here, we investigated the effect of psychosine on the levels of IL6, TNFα and IL1β in mouse astrocytes in the presence and absence of LPS. Firstly, LPS (1 ng/ml to 10 µg/ml) was shown to induce a concentration-dependent increase in the levels of IL6 (Fig. 4A), TNFα (Fig. 4B) and IL1β (data not shown): 100 ng/ml LPS was selected as the optimal concentration for use in further experiments. Next, cultured mouse astrocytes were serum starved, pre-treated with pFTY720 (1 µM for 1 h) and then treated with LPS (100 ng/ml) and/or psychosine (1 µM and 10 µM) for 3 h, 6 h and 12 h. No discernible release of IL6, TNFα or IL1β was observed after 3 h treatment with LPS and/or psychosine (data not shown). In contrast, the treatment of astrocytes for 6 h with LPS induced at least a 2-fold increase in the levels of IL6 (5.1±0.9 pg/ml vs 31.5±4.3 pg/ml; mean±s.e.m.) (Fig. 4C), TNFα (19.4±3.4 pg/ml vs 48.3±2.1 pg/ml) (Fig. 4D) and IL1β (1.1±0.5 pg/ml vs 14.3±5.5 pg/ml) (data not shown), compared with control. Moreover, psychosine (10 µM), while having little effect alone, significantly augmented the LPS-induced production of IL6 (31.5±4.3 pg/ml vs 44.8±4.2 pg/ml) (Fig. 4C) and TNFα (48.3±2.1 pg/ml vs 59.1±2.8 pg/ml) (Fig. 4D) (unpaired Student's t-test P<0.05). Importantly, pFTY720 treatment significantly attenuated the increased levels of IL6 (44.8±4.2 pg/ml vs 25.6±4.9 pg/ml) (Fig. 4C) and TNFα (59.1±2.8 pg/ml vs 42.4±4.4 pg/ml) (Fig. 4D) induced by LPS or psychosine (10 µM). Similar findings were observed at 12 h: pFTY720 attenuated the LPS and/or psychosine (10 µM) mediated increase in levels of IL6 (132.0±3.4 pg/ml vs 56.4±21.3 pg/ml) (Fig. 4E) and TNFα (183.8±88.9 pg/ml vs 78.8±25.3 pg/ml) (Fig. 4F). Collectively, these findings suggest that psychosine may enhance the LPS-induced levels of pro-inflammatory cytokines in astrocytes as previously reported (Giri et al., 2002), and that pFTY720 attenuates these effects.

Rapid and complete loss of myelin and the myelin-forming oligodendrocytes is one of the main pathological features of Krabbe disease (Davenport et al., 2011). pFTY720 promotes remyelination as well as limiting the demyelination induced by the bioactive lipid lysolecithin (lysophosphatidylcholine, LPC) (Miron et al., 2010; Sheridan and Dev, 2012). With this in mind, we first determined whether psychosine induced demyelination in cerebellar slices and secondly examined whether pFTY720 attenuated this psychosine-induced demyelination. Organotypic cerebellar slices were exposed to LPC (0.5 mg/ml) or psychosine (100 nM, 1 µM, 20 μM) in the presence or absence of pFTY720 (0.1 nM, 1 nM) for 18 h and treated for a further 30 h with pFTY720 (0.1 nM, 1 nM) or control medium (Fig. 5A). In agreement with our previous studies (Pritchard et al., 2014; Sheridan and Dev, 2012), pFTY720 attenuated LPC-induced demyelination compared with control, as observed by expression of myelin oligodendrocyte glycoprotein (MOG) (Fig. 5B) and myelin basic protein (MBP) (Fig. 5C–D). Importantly, the exposure of the slice cultures to psychosine also decreased the expression of MOG (Fig. 5B, Fig. S1), MBP (33.4±3.8 vs 12.4±2.5, 1 µM psychosine; mean±s.e.m.) (Fig. 5C–D) and myelin proteolipid protein (PLP) (Fig. S2), as well as decreasing the expression of neurofilament H (NFH) (29.9±5.4 vs 13.4±4.6, 1 µM psychosine). pFTY720 (0.1 nM, 1 nM) prevented the psychosine-induced decrease in expression of MOG (Fig. 5B, Fig. S1), MBP (12.43±2.5 vs 31.6±4.9), PLP (Fig. S2) and NFH (13.4±4.6 vs 40.7±10.3) (Fig. 5C-D). Taken together, therefore, these results demonstrate that pFTY720 reverses psychosine-induced demyelination and neuronal toxicity in cerebellar slice cultures.

The involvement of pro-inflammatory cytokines in the pathogenesis of demyelination has been previously investigated (di Penta et al., 2013), and a model in which S1PR activation may reduce demyelination via a mechanism involving attenuation of cytokine/chemokine release has been proposed (Sheridan and Dev, 2012). Hence we investigated whether psychosine treatment would induce the release of the pro-inflammatory cytokines IL6, TNFα and IL1β from organotypic cerebellar slices. As with the above protocol, organotypic cerebellar slices were exposed to LPC (0.5 mg/ml) or psychosine (100 nM) in the presence or absence of pFTY720 (0.1 nM) for 18 h and treated for a further 30 h with pFTY720 (0.1 nM). After the 30 h incubation the medium was collected and analysed by ELISA. LPC induced at least a 4-fold increase in levels of IL6 (136.2±61.5 pg/ml vs 1543.2±89.4 pg/ml; mean±s.e.m.), TNFα (15.0±4.6 pg/ml vs 61.9±12.3 pg/ml) and IL1β (0.6±0.6 pg/ml vs 178.8±32.7 pg/ml), compared with controls (Fig. 6A–C). pFTY720 treatment attenuated the LPC-induced release of IL-6 (1543.2±89.4 pg/ml vs 1273.1±22.1 pg/ml) (Fig. 6A), TNFα (61.94±12.3 pg/ml vs 42.32±11.2 pg/ml) (Fig. 6B) and IL1β (178.8±32.7 pg/ml vs 14.0±14.1 pg/ml) (Fig. 6C). Interestingly, psychosine (100 nM) treatment did not induce the release of IL6, TNFα or IL1β, in agreement with our data showing the treatment of mouse astrocytes with psychosine alone had little effect on the release of IL6 (Fig. 4C and E), TNFα (Fig. 4B and F) and IL1β (data not shown). Furthermore, we did not observe significant effects of psychosine on ionized calcium binding adaptor molecule 1 (Iba1) (microglia) staining, nor did BV2 microglia cells treated with psychosine show enhanced levels of IL6 cytokine release (Fig. S3). Taken together, these results suggest that, in organotypic cerebellar slices, psychosine-induced demyelination occurs via a mechanism that is probably independent from the release of pro-inflammatory cytokines. Moreover, pFTY720 inhibits this type of demyelination.

Accumulation of psychosine, propagation of pro-inflammatory cytokines, demyelination and the widespread loss of oligodendrocytes are all hallmarks of the Krabbe disease brain (Suzuki, 1998; Wenger et al., 2001). To date, little is known about the role of astrocytes in Krabbe disease, where the majority of studies have mainly been focused on the role oligodendrocytes in this illness. Altered astrocytic function has gained recognition as a major contributing factor to a growing number of neurological disorders (Claycomb et al., 2013) and the belief that astrocytic dysfunction significantly contributes to the development of inflammation in the CNS has gained traction in the past number of years (Sharma et al., 2010). In addition, the immunomodulatory functions of astrocytes are now being shown to actively participate in the pathogenesis of a number of demyelinating disorders (Sharma et al., 2010). It has also been reported that astrocytic processes may closely surround demyelinating fibres and that astrocytes release factors such leukaemia inhibitory factor (LIF) to promote myelination. Remarkably, when oligodendrocytes from twi/twi mice are transplanted into the shiverer mouse model of demyelination, these oligodendrocytes are capable of myelinating the shiverer axons (Kondo et al., 2005). This indicates that demyelination in Krabbe disease may not be solely attributed to oligodendrocyte dysfunction and that given correct environmental support these twi/twi oligodendrocytes may function normally. Thus, astrocytic reactivity in Krabbe disease may not represent a secondary response to demyelination, but may possibly be a primary response to psychosine and in turn astrocytes may significantly contribute to the pathogenesis of Krabbe disease (Claycomb et al., 2013).

In our current study, we obtained data that demonstrated a direct effect of psychosine on astrocyte cell death and attenuation by pFTY720. We investigated the effect of the cytotoxic lipid metabolite psychosine on cultures of human astrocytes. Psychosine caused a time- and concentration-dependent decrease in astrocyte cell numbers, as previously reported (Giri et al., 2002; Sugama et al., 1990). In agreement with the current literature (Davenport et al., 2011; Haq et al., 2003; Jatana et al., 2002), our JC-1 experiments show that psychosine induces mitochondrial dysfunction. Hence, we suggest, the astrocytic cell death induced by psychosine is likely to occur via an apoptotic process. Importantly, pFTY720 attenuated psychosine-induced cell death as well as restoring psychosine-induced mitochondrial dysfunction. Moreover, we found that although psychosine itself did not induce increased levels of pro-inflammatory cytokines in mouse astrocytes, it did enhance LPS-mediated release of IL6, TNFα and IL1β, and these effects were again reduced by pFTY720. We also report here, for the first time, that direct application of psychosine to organotypic slice cultures induces demyelination in a manner that does not include enhanced pro-inflammatory cytokine release. These data corroborate the idea that psychosine in the brains of Krabbe disease patients may directly induce demyelination. In these set of experiments, pFTY720 attenuated LPC-induced demyelination, which was shown to include enhanced levels of pro-inflammatory cytokines, as we have reported before (Sheridan and Dev, 2012). Of most interest, pFTY720 also reduced the demyelination caused by psychosine, in a manner that did not include enhanced levels of IL6, TNFα and IL1β. Overall, these studies suggest that S1PRs may regulate myelination state in both inflammatory and non-inflammatory paradigms.

Inflammatory processes have been implicated in the pathogenesis of Krabbe disease and the expression of pro-inflammatory cytokines have been reported in the twi/twi mouse brain (Claycomb et al., 2013; LeVine and Brown, 1997). However, the mechanism governing psychosine-mediated cell toxicity and the direct role of pro-inflammatory cytokines in the degeneration of astrocytes and/or oligodendrocytes is still not fully understood. Here we found that psychosine treatment itself did not alter pro-inflammatory cytokine levels, and therefore such an effect on cytokines is unlikely to explain the decrease in astrocyte cell numbers that we observed after psychosine treatment. Instead, psychosine may induce astrocyte cell death by altering mitochondrial function and electron transport, as determined by increased JC1 levels in the cytosol and decreased NAD(P)H oxidase function measured by MTT, respectively. In agreement with this idea, previous studies have demonstrated that psychosine alters mitochondrial function and electron transfer, probably via a mechanism involving changes in the lipid environment of the membrane (Cooper et al., 1993; Tapasi et al., 1989). Moreover, studies have also reported that pFTY720 can stabilise mitochondrial function, supporting our findings that pFTY720 rescues mitochondrial dysfunction induced by psychosine. Interestingly, in this current study, although psychosine alone had no effect on pro-inflammatory cytokine levels, it augmented the LPS-induced release of IL6 from mouse astrocytes, with a similar trend for the levels of TNFα and IL1β. This finding is comparable to those reported previously where psychosine potentiated LPS-induced production of TNFα, IL6, IL1β and NO in primary rat astrocytes, which in turn was suggested to induce oligodendrocyte cell death (Giri et al., 2002,, 2006). These enhanced levels of cytokines were reduced by pFTY720, in agreement with previous studies from our and other groups demonstrating that S1PRs play a role in regulating the levels of cytokines in a number of immune and glial cells (Choi et al., 2011; Sheridan and Dev, 2012; Wang et al., 2007; Zhang et al., 2008). Thus, in astrocytes, it appears that psychosine directly modulates mitochondrial function to induce cell death, while in parallel enhancing cytokine levels under conditions of LPS-induced inflammation, and that pFTY720 can attenuate these effects of psychosine.

Profound demyelination and almost complete loss of oligodendrocytes are two of the major pathological features of Krabbe disease. The hypothesis that supraphysiological levels of psychosine kill oligodendrocytes and result in widespread demyelination is now widely accepted. The mechanisms by which psychosine induces demyelination remains unclear at present. Increasing evidence however now suggests that the widespread demyelination and loss of oligodendrocytes seen in Krabbe disease and induced by psychosine is due to apoptotic processes, probably via caspase dependent pathways (Zaka and Wenger, 2004; Tohyama et al., 2001; Giri et al., 2006; Giri et al., 2008; Haq et al., 2003). Another proposed mechanism by which psychosine induces toxicity involves the preferential accumulation of the molecule in lipid rafts, associated with regional cholesterol increases and inhibition of PKC activity (White et al., 2011; Davenport et al., 2011; Yamada et al., 1996; Hannun and Bell, 1987). Psychosine treatment of oligodendrocytes has also been shown to induce the generation of LPC and arachidonic acid, and inhibition of secreted phospholipase A2 (sPLA2) attenuates psychosine-induced increases in both LPC and arachidonic acid as well as attenuating psychosine-induced cell death (Giri et al., 2006). Furthermore, psychosine accumulation has been reported to induce phosphorylation of neurofilament proteins, resulting in reduced radial growth of axons in the twi/twi mouse model and to induce axonal defects and cell death in isolated neuronal cultures (Cantuti-Castelvetri et al., 2012; Castelvetri et al., 2011). S1PRs are known to play many roles in the regulation of differentiation, cell survival and apoptosis of oligodendrocytes, astrocytes and microglia (Dev et al., 2008; Miron et al., 2008b). In addition, previous studies have also shown that S1PRs are expressed on Schwann cells and that treatment with compounds such as pFTY720 can promote Schwann cell survival and regulate peripheral nerve myelination (Kim et al., 2011). Moreover studies have also shown that compounds such as pFTY720 can significantly improve motor function recovery in animal models of spinal cord injury (Lee et al., 2009). Therefore, we used organotypic slice cultures as a more complex cellular model to investigate the effect of psychosine on demyelination. Here, for the first time, we showed that psychosine can directly induce a concentration-dependent demyelination, as expressed by a decrease in the levels of MOG, PLP and MBP, and that pFTY720 can attenuate these effects. These findings are in agreement with our and other previous reports demonstrating that pFTY720 rescues myelination (Coelho et al., 2007; Jung et al., 2007; Mattes et al., 2010; Miron et al., 2008a,,b; Sheridan and Dev, 2012). Interestingly, unlike LPC, psychosine treatment alone did not induce the release of IL6, TNFα or IL1β from cerebellar slices, which was in agreement with our astrocyte data. These findings suggest that psychosine induces demyelination independently from the release of pro-inflammatory cytokines and most importantly that pFTY720 can inhibit demyelination independently from the regulation of pro-inflammatory cytokines. Taking these current findings and previous studies that have demonstrated pFTY720 can promote neuronal and oligodendrocyte survival (Dev et al., 2008; Miron et al., 2008b) into account, we suggest that pFTY720 may rescue the toxic effects induced by psychosine by having multimodal effects on both glial and neuronal cells. However, we acknowledge the need for further studies, including: (1) ultrastructural EM analysis of remyelination to support our immunostaining data, which is not a direct measurement of enhanced myelination per se, (2) testing the effects of pFTY720 on myelination in twi/twi cerebellar slice cultures to further test the hypothesis of psychosine-induced demyelination and (3) most importantly, testing the effects of pFTY720 on disease progression in twi/twi mice. Although these studies remain outstanding and while recent clinical data shows a lack of efficacy for pFTY720 in progressive forms of multiple sclerosis, the cellular data presented in the current study goes some way in suggesting that S1PRs may be useful targets in demyelinating illnesses such as Krabbe disease.

Human, rat and mouse astrocytes were cultured as we have described before (Healy et al., 2013; Rutkowska et al., 2015). BV2 microglia were cultured in DMEM supplemented with 2% FBS (Labtech) and 1% penicillin/streptomycin (Sigma) at 37°C and 5% CO₂. When 80% confluent cells were split into 24-well plates to be treated. In all cases, before treatment cells were serum starved for 3–4 h by incubating in serum-free DMEM/F12 (Fisher) at 37°C and 5% CO₂. Specific treatment details are indicated in the figure legends. Organotypic cerebellar slice cultures were prepared exactly as we have described previously (Pritchard et al., 2014; Sheridan and Dev, 2012). In brief, tissue isolated from postnatal day 10 (P10) C57BL/6 mice and 400 µm parasagittal slices of cerebellum were grown on cell culture inserts (5–6 slices each) (Millicell PICMORG50 Millipore). Slices were cultured using an interface method with 1 ml of medium per 35 mm well. For the first 3 days in vitro (DIV), slices were grown in serum-based medium (50% Opti-Mem (Invitrogen), 25% Hanks’ buffered salt solution (HBSS) (Gibco), 25% heat-inactivated horse serum and supplemented with 2 mM Glutamax, 28 mM d-glucose, 100 U/ml penicillin/streptomycin (Sigma) and 25 mM HEPES (Sigma), at 35.5°C and 5% CO₂. After 3 DIV, slices were transferred to serum-free medium (98% Neurobasal-A and 2% B-27 (Invitrogen), supplemented with 2 mM Glutamax, 28 mM d-glucose, 100 U/ml penicillin/streptomycin and 25 mM HEPES). Demyelination was induced at 12 DIV and examined at 14 DIV. All tissue was isolated in accordance with EU guidelines and protocols approved by the Trinity College Dublin ethics committee.

For cytokine analysis supernatants from cell culture were removed and examined for their cytokine content using ELISA kits for IL6 (DY406), TNFα (DY410) and IL-1β (DY401) according to the manufacturer's instructions (R&D Systems), and exactly as we have described before (Pritchard et al., 2014). For western blotting, cerebellar slices were scraped from the culture membrane and suspended in PTxE buffer (PBS, 1% Triton-x, 1 mM EDTA) using mechanical homogenisation and sonication. Samples were denatured and electrophoresis carried out on 10% SDS-polyacrylamide gels exactly as we have previously reported (Pritchard et al., 2014). Primary antibodies used were: anti-MOG (1:2000; Millipore: MAB5680) and anti-tubulinβ (1:5000; Millipore MAB3408). Secondary antibody used was: HRP-conjugated mouse (1:10,000; Sigma: A8924).

JC-1 (tetraethylbenzimidazolylcarbocyanine iodide) exhibits potential-dependent accumulation in mitochondria. Measurements of mitochondrial membrane potential (ΔΨm) using membrane-permeant dyes such as JC-1 are therefore widely used in apoptosis studies to monitor mitochondrial health. Mouse astrocytes were plated into a black 96-well plate in 100 μl DMEM/F12 and incubated for 72 h at 37°C and 5% CO₂. On the day of the assay, the cells were incubated for 3 h in serum-free DMEM/F12 at 37°C and 5% CO₂ before receiving a pre-treatment of pFTY720 (1 µM) for 1 h. Cells were then treated with psychosine (5 µM, 10 µM, 15 µM, 20 µM) for 2 h. All wells were then loaded with 1 μM JC-1 for 30 min. After this incubation the plate was centrifuged for 5 min at 400 g at room temperature. The supernatant was aspirated and the wells washed with 200 μl of PBS. This wash step was repeated and the emission spectra at 535 nm and 590 nm was measured (SpectraMAX Gemini XS). A colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay was used to assess cell viability as per manufacturer's instructions (Invitrogen, M6494), similar to our previous studies (Mullershausen et al., 2009; Rutkowska et al., 2015).

All images of human astrocytes were taken using a CKX41 Olympus inverted microscope (Mason Technologies) at 10× magnification. ImageJ software was used to calculate the cell density of the surviving cells by measuring the percentage of empty space compared with their own controls. For organotypic cerebellar slice cultures, immunostaining was performed as we have described previously (Sheridan and Dev, 2012). Primary antibodies used were: anti-MBP (1:1000; Abcam: ab40390), anti-NFH (1:1000; Millipore: MAB5539), anti-PLP (1:500; Millipore: MAB388), anti-MOG (1:1000; Millipore: MAB5680) and anti-Iba1 (1:1000; Wako: 019-1974). Secondary antibodies used were: anti-chicken 633 (1:1000; Invitrogen Alexa: A21103), anti-rabbit 488 (1:2000; Invitrogen Alexa: A27034) and anti-mouse Dylight 549 (1:2000; Jackson ImmunoResearch: 715-505-020). Confocal images were captured using a LSM 510 Meta microscope at 10× or 20× magnification. These resulting images were analysed using ImageJ software. A total of 5–6 slices were used per condition and the fluorescence of each cerebellar slice was captured using 5–6 independent regions of interest (ROI). The ROI were selected randomly to cover the whole slice and the mean fluorescence was calculated using a total of 25–36 independent ROI observations for each independent experiment.

All statistical analysis was performed using Prism 5 GraphPad Software package. A one-way ANOVA with Newman–Keuls post-hoc test was used to compare groups and an unpaired Student’s t-test was used to compare two sets of data with each other. Individual statistical tests are described in text and figure legends. The significance levels (or alpha levels) were set at P<0.05, P<0.01 and P<0.001.

We are grateful to Dr Anis Mir (Novartis Pharma, Basel, Switzerland) for supply of pFTY720 and to Prof. Veronica Campbell (School of Medicine, Trinity College Dublin) for supply of BV2 microglia.