Phospholipase A2 is involved in galactosylsphingosine-induced astrocyte toxicity, neuronal damage and demyelination

Krabbe disease is a fatal rare inherited lipid storage disorder affecting 1:100,000 births. This illness is caused by mutations in the galc gene encoding for the enzyme galactosylceramidase (GALC). Dysfunction of GALC has been linked to the toxic build-up of the galactolipid, galactosylsphingosine (psychosine), which induces cell death of oligodendrocytes. Previous studies show that phospholipase A2 (PLA2) may play a role in psychosine induce cell death. Here, we demonstrate that non-selective inhibition of cPLA2/sPLA2 and selective inhibition of cPLA2, but not sPLA2, also attenuates psychosine-induced cell death of human astrocytes. This study shows that extracellular calcium is required for psychosine induced cell death, but intracellular calcium release, reactive oxygen species or release of soluble factors are not involved. These findings suggest a cell autonomous effect, at least in human astrocytes. Supporting a role for PLA2 in psychosine-induced cell death of oligodendrocytes and astrocytes, the results show inhibition of PLA2 attenuates psychosine-induced decrease in the expression of astrocyte marker vimentin as well as myelin basic protein (MBP), myelin oligodendrocyte glycoprotein (MOG) and the neuronal marker SMI-32 in organotypic slice cultures. These findings provide further mechanistic details of psychosine-induced death of glia and suggest a role for PLA2 in the process. This work also supports the proposal that novel drugs for Krabbe disease may require testing on astrocytes as well as oligodendrocytes for more holistic prediction of pre-clinical and clinical efficacy.


Introduction
Krabbe disease is a rare inherited lipid storage disorder resulting in oligodendrocyte cell death and subsequent loss of myelin. Krabbe disease is associated with mutations in the galc gene encoding for the enzyme galactosylceramidase (GALC) [1] resulting in the build-up of one particular galactolipid, called psychosine, which is suspected to be responsible for the pathology of this illness [2]. Most studies on Krabbe disease have focused on the toxicity of psychosine on oligodendrocytes. Several pathways have been associated with psychosine-induced (DMSO, Sigma, D8418); and EDTA (Sigma E5132) which was prepared as 100mM stock solution in sterile water. Concentrations of these drugs were used as suggested by pervious literature [9,33,34].

MTT assay
Astrocytes were plated in 96-well plates and cultured for 24 hours until 70% confluent. The cells were starved with serum free media (DMEM/F12) for three hours, treated with or without compounds, after which the media was removed and replaced by 100μL of fresh, pre-warmed media supplemented with 10μL of 12mM MTT (Invitrogen, M6494). The cells were incubated with MTT for 4 hours at 37˚C. After 4 hours of incubation, 75μL of media was removed and 50μL of DMSO per well was added. The cells were incubated for 10 min at 37˚C. The plate was mixed and the absorbance read at 540nm.

Experimental design and statistical analysis
Psychosine was first tested on human astrocytes culture in order to obtain time and concentration curves. For all experiments using astrocytes culture, cells have been starved using serum free media for 3 hours, pre-treated with the specific inhibitor for 1 hour and then treated with psychosine. We also tested our inhibitors in a more complex model using organotypic cerebellar slices culture. Slices were used after 12 days of culture, then treated with psychosine and the inhibitor for 18 hours. The media was changed but treatment with the inhibitor was performed for another 30 hours. Slices were then fixed, stained and visualized using a confocal microscope. All statistical analysis was performed using GraphPad Prism 5 Software package (Graph-PAD Software for Science, San Diego, CA, U.S.A). Unpaired Student's t-test was used to compare two sets of data from two independent samples, whereas one-way ANOVA was used to compare three or more sets of data from unrelated samples. Newman-Keuls multiple comparison post-hoc test was used after an analysis of variance (ANOVA) for the comparisons of different pairs of mean. Keuls multiple comparison post-hoc, is based on the studentized range distribution, which test all differences among pairs of means. Means were ranked from smallest to largest, and then, the smallest mean was compared to the largest mean. If the test was not significant, then no pairwise tests were significant and no further testing was performed. The significance levels (or alpha levels) were set at p<0.05, p<0.01 and p<0.001.

Psychosine-induced changes in human astrocyte morphology and is partially reversed by DEDA
To confirm the protective effect of DEDA on psychosine-induced toxicity we examined the effect of psychosine on the type III intermediate filament astrocyte marker vimentin, which is involved in cytoskeleton formation in astrocytes. Treatment of human astrocytes with psychosine (10μM) for 3 hours or 24 hours induced a reduction of vimentin expression in astrocytes, particularly in the cell processes, suggesting a deregulation of cellular cytoskeleton (Ctrl: 1.03 ± 0.14; 24h: 0.67 ± 0.15) (Fig 2). This change likely reflected the rounding of astrocytes before detachment induced by psychosine. Importantly, these changes in vimentin localization induced by psychosine were reduced by 1 hour pre-treatment with DEDA (5μM) (3h: 1,40 ± 0,15 vs 1.03 ± 0,14; 24h: 1,40 ± 0,14 vs 0.67 ± 0.15) (Fig 2). Overall, these results support the idea that altered cytoskeleton structure may precede psychosine induced cell death and that DEDA attenuates these effects, suggesting the involvement of PLA2.
Psychosine induced astrocyte cell death does not involve production of ROS or the release of a toxic soluble factors Psychosine induces reactive oxygen species (ROS) in the oligodendrocyte MO3.13 cell line [33]. We thus investigated if pre-treatment with a specific ROS inhibitor (N-acetylcysteine, NAC) (1mM) 1 hour before psychosine (10μM) had protective effects on astrocyte viability, but found no obvious rescue (3h: 39.3 ± 1.8% vs 36.1 ± 1.2%; 20h: 48.9 ± 5.9% vs 51.6 ± 5.7%) Human astrocytes were pre-treated with (A) the specific sPLA2 inhibitor Varespladib (10μM for 1h) (n = 4) or (B) the specific cPLA2 inhibitor BDDPAA (1μM for 1h) (n = 4-5) before treatment with psychosine (10μM for 3h or 20h). Cell viability was quantified by an MTT assay. Data is presented as mean + s.e.m. The main graph shows the ratio between control vs psychosine and drug vs drug +psychosine and smaller graphs show comparisons to vehicle control and psychosine treatment.
https://doi.org/10.1371/journal.pone.0187217.g003 ( Fig 5A). To further examine the mechanism by which psychosine induces astrocyte cell death, we investigated if human astrocytes treated with psychosine may release toxic mediators into the media that can induce cell death in a cell non-autonomous manner. Human astrocytes treated with psychosine (10μM) for one hour were washed free from psychosine and fresh media was added. After five-hour incubation, the conditioned media ("psy-media") was collected, boiled at 90˚C for 5 min, and/or directly incubated with naive human astrocytes for 20 hours. While psychosine induced significant human astrocyte cell death compared to control (37.8 ± 2.8% vs 87.3 ± 3.0%), no cell death was induced by the psy-media (98.7 ± 1.9 vs 87.3 ± 3.0%) or by the boiled psy-media ( Fig 5B). These results suggest that psychosine induced astrocyte cell death does not involve production of ROS or the release of a toxic soluble factors, and that psychosine toxicity occurs in a cell autonomous manner, at least in isolated astrocyte cell cultures. Inhibition on PLA2 attenuates psychosine-induced decrease in the astrocyte marker vimentin In our previous studies we have shown that psychosine induces demyelination in organotypic slice cultures [17,18]. Thus, to further these previous studies and examine whether our isolated astrocyte culture data held true in a more complex cellular system, we next examined the effects of psychosine on expression of the astrocyte marker vimentin, using these brain slice Human astrocytes were pre-treated with a specific ROS inhibitor N-acetylcysteine (NAC) (1mM for 1h) (n = 3-4) before treatment with psychosine (10μM for 3h or 20h). The main graph shows the ratio between control vs psychosine and drug vs drug+psychosine and smaller graphs show comparisons to vehicle control and psychosine treatment. (B) Human astrocytes were serum starved for 3h before treated with psychosine (10μM for 1h). After washing three times with serum free media to wash out the added psychosine, fresh complete media was incubated for 5h with the treated astrocytes and this conditioned media ("psy-media") was then added to naive human astrocytes and incubated for 20h. One-way ANOVA and Newman-Keuls Multiple Comparison post-test, compared to control (*p<0.05, **p<0.01, ***p<0.005). The number of experiments is indicated in the parenthesis (n = 4-28). Cell viability was quantified by an MTT assay. Data is presented as mean + s.e.m. cultures. For the first time, we show that psychosine reduces the expression of vimentin in cerebellar slices (74.38 ± 7.13% vs 100%) (Fig 6B and 6C). This data is in agreement with our current astrocyte culture studies data and furthers our previous findings, suggesting psychosine reduces cell viability of astrocytes. Importantly, in these experiments, DEDA attenuated the decrease in expression of vimentin caused by psychosine (83.50 ± 16.45% vs 74.38 ± 7.13%) (Fig 6B and 6C). We have also reported psychosine does not regulate the expression levels of the microglia marker Iba1 (ionized calcium binding adapter molecule 1), although it should be noted Iba1 is not a specific marker of altered microglia reactivity. Here we confirmed that psychosine does not alter the expression of Iba1 (92.13 ± 10.43% vs 100%) and that DEDA had no effects on the expression of this marker either (102.1 ± 23.27% vs 100%) (Fig 6B and 6C).

Psychosine-induced demyelination is reversed by the inhibition of PLA2
A recent study has identified a pathogenic phenotype of reactive astrocytes found in a number of neurodegenerative and neuroinflammatory disorders. These neurotoxic cells can directly induce neuronal and oligodendroglial death in addition to loss of their homeostatic function [37]. We have also shown previously that psychosine induced astrocyte cell death in dissociated cell cultures and demyelination in organotypic slice cultures can be rescued by the S1PR drugs pFTY720 and BAF312, both of which are efficacious in multiple sclerosis [17,18]. Here we aimed to confirm psychosine-induced demyelination in organotypic cerebellar slices and to examine the role of PLA2 in this complex brain slice model system. Organotypic cerebellar slices were exposed to psychosine (100 nM) in the presence or absence of DEDA (100 nM) for 18 h and treated for a further 30 h with DEDA (100 nM) or control medium (Fig 7A). In agreement with our previous studies [17,18], psychosine induced a modest decrease in the levels of neurofilament H (NFH) (86.38 ± 25.82% vs 100%), with significant demyelination compared with control, as observed by expression of myelin basic protein (MBP) (67.88 ± 10.82% vs 100%) and myelin oligodendrocyte glycoprotein (MOG) (70.17 ± 17.19% vs 100%) (Fig 7B). Importantly, this psychosine-induced demyelination was significantly attenuated by DEDA (100 nM) (MBP: 102.3 ± 15.44% vs 67.88 ± 10.82%; MOG: 108.0 ± 30.52% vs 70.17 ± 17.19%) (Fig 7B). These results further support the idea that PLA2 plays a role in psychosine-induced cell toxicity and may have a multimodal effect in both astrocytes and oligodendrocytes.

Axonal damage induced by psychosine is reversed by the inhibition of PLA2
Neurofilament H is a structural cytoskeletal protein specific to neurons [38] comprising phosphorylated and non-phosphorylated epitopes of NFH [39]. It has been previously shown that non-phosphorylated NFH epitopes, such as SMI-32, are mainly expressed by purkinje and neuronal cell bodies under physiological conditions, with minimal expression in the axonal tracts [40,41]. More interestingly, an increase in the expression of SMI-32 in the axonal tracts has been associated with disturbances in axonal transport, demyelination and neuronal damage in multiple sclerosis [42,43]. In our previous studies we have utilised SMI-32 as a marker for neuronal damage and have shown an increase in the expression of SMI-32 in the white matter tracts under demyelinating GOX-CAT treatment in mouse cerebellar slice culture [30]. In agreement with that, and to further our previous findings [17][18], here we report an increase of SMI-32 in the white matter tracts under psychosine treatment (Fig 8A). Importantly, we find that inhibition of sPLA2 by DEDA attenuated this increment in SMI-32 (193% ± 8.81% vs Psy: 593.3% ± 81.6%, ### p<0.0001). Notably, DEDA did not induce the expression of SMI-32 in the axonal tracts by itself compared to control (136.2% ± 11.47% vs 100%) ( Fig  8B).

Summary of findings
Recent studies in our laboratory have demonstrated that psychosine induces cell toxicity in astrocytes [17,18]. In the current study, we investigated whether PLA2, calcium and ROS PLA2 and krabbe disease pathways are involved in this process. Results demonstrate that PLA2 inhibitors rescue the decrease of human astrocyte cell viability induced by psychosine treatment. We show that the cytosolic isoform of PLA2 (cPLA2), rather than the secretory form (sPLA2), is involved in psychosine-induced toxicity in astrocytes. We also find that extracellular calcium is required for psychosine-induced toxicity, whereas limiting intracellular calcium has little effect. Our experiments show that ROS generation does not play a significant role in psychosine-mediated astrocyte cell death. In addition, the data show that media obtained from astrocytes treated with psychosine does not cause further astrocyte cell death, suggesting that psychosine induces toxicity in a cell-autonomous manner. Using a more complex cellular system, we find that psychosine inhibits expression of vimentin in organotypic slice cultures and these effects are attenuated by the PLA2 inhibitor DEDA, corroborating our dissociated astrocyte cell culture findings. We find also that PLA2 attenuates psychosine-induced decrease in the expression of myelin basic protein (MOG) and myelin oligodendrocyte glycoprotein (MOG) in organotypic slice cultures. Lastly, we demonstrate that axonal damage, as measured by the expression of SMI-32 in the axonal tracts, induced by psychosine was attenuated under inhibition of PLA2. Taken together, our results suggest that psychosine modifies astrocyte as well as oligodendrocyte function, likely leading to subsequent axonal damage and that the PLA2 signalling pathway plays a role in this process.

The cPLA2 isoform plays a role in psychosine-induced toxicity in astrocytes
The PLA2 inhibitor DEDA has a protective effect on psychosine-induced cell death in oligodendrocytes [9]. Activation of PLA2 leads to the generation of arachidonic acid and metabolites, which are involved in processes such as inflammation, apoptosis, cell growth and differentiation [44,45]. Arachidonic acid leads to the activation of p38 MAPK and transcription factors such as PLA2 and krabbe disease AP1 and NFκB [46]. The metabolism of arachidonic acid also gives rise to ROS, which may induce oxidative stress [47]. The arachidonic acid (and/or lysophosphatidylcholine, LPC) signalling pathway(s) likely result in the loss of oligodendrocytes, where inhibitors of PLA2, such as DEDA, can rescue psychosine-induced cell death [9]. Previous studies have demonstrated that astrocytes release arachidonic acid (with phospholipase A2 involvement) under ROS conditions [48,49]. We note however, no previous research, to our knowledge, has investigated psychosine-induced levels of arachidonic acid or LPC in astrocytes. A caveat in our current study is we also did not investigate if psychosine changes levels of LPC and/or AA. Here, our data suggested that DEDA itself promoted moderately the number of astrocytes and more importantly protected these cells from psychosine toxicity. Given that DEDA may inhibit both sPLA2 and cPLA2, and also 5-lipoxygenase and formation of leukotrienes, we also used specific PLA2 inhibitors. The low molecular weight (13-15 kDa) sPLA2 requires high concentrations of Ca2+ (mM range) for activity [10,11], whereas cPLA2 is larger (85 kDa) and requires a lower concentration of Ca2+ (μM range) [50]. Here, we used Varespladib, a sPLA2 inhibitor with an IC50 of 9.6nM [35] and BDDPAA, a cPLA2 inhibitor, with an IC50 of 1.8nM [51]. Our results showed that the cPLA2 inhibitor BDDPAA, but not the sPLA2 specific inhibitor Varespladib, attenuated psychosine-induced toxicity, at least in astrocytes. These results are in contrast to previous studies showing that sPLA2 plays a role in psychosine-induced toxicity in oligodendrocytes [9] and highlights possible mechanistic differences between cell types. It should be noted that we also observed that BDDPAA treatment by itself modestly increased the number of human astrocytes. Arachidonic acid has been shown to act as a second messenger and regulate apoptosis [44,52]. Inhibiting cPLA2 activity may lead to a decrease in arachidonic acid and a loss of cell ability to produce lipid mediators [53,54]. This hypothesis could explain why inhibiting cPLA2 using BDDPAA has a modest increase in cell number.

Extracellular calcium signalling is required for psychosine-induced toxicity in human astrocytes
Previous studies in oligodendrocytes have shown that psychosine leads to a rapid calcium peak after 5-8 minutes treatment and followed by a slower second peak that precedes cell death [33]. It has been demonstrated that the chelation of extracellular calcium extends oligodendrocyte cell survival and reduces mitochondrial ROS production [33]. Here, we used EDTA to chelate extracellular calcium and the ryanodine antagonist Dandrolene to inhibit release of calcium from intracellular stores. The data showed that EDTA, but not Dandrolene, significantly rescued psychosine-induced cell death of human astrocytes. This protective effect was observed to be more pronounced than the PLA2 inhibitors and lasted both after short and long periods of psychosine treatment. Our data suggest that, similar to oligodendrocytes, psychosine might cause a calcium deregulation, which also plays a role in astrocyte cell death.

ROS is not thought to be involved in psychosine-induced toxicity in astrocytes
In addition to deregulating PLA2 and/or calcium signalling, psychosine might induce the production of intra-mitochondrial ROS production [33]. This ROS production might be associated with changes in calcium signalling, in particular, intra-mitochondrial calcium signalling [55]. With regard to PLA2, the over-production of arachidonic acid from PLA2 enzymatic activity might also lead to ROS production [56]. We used the ROS inhibitor N-acetylcysteine (NAC), which limits ROS generation via inhibiting c-Jun N-terminal kinase, p38 MAP kinase, redox-sensitive activating protein-1 and NFκB [57,58]. In contrast with oligodendrocytes, our data show that inhibiting ROS production did not affect human astrocyte cell survival when treated with psychosine. Oligodendrocytes are particularly sensitive to oxidative stress [59], where the production of ROS is damaging more so to oligodendrocytes, compared to astrocytes or microglia [60]. Therefore, in agreement with our findings, ROS inhibitors may have protective effects on psychosine-mediated oligodendrocyte cell death, while not so in astrocytes.
Psychosine mediates astrocyte cell death in a cell-autonomous manner Others and we, have shown that psychosine does not alter the levels of cytokines in astrocytes [5,17,18]. However, psychosine can enhance the production of cytokines such as TNFα, IL8 or MCP1 in addition to LPS in peripheral blood mononuclear cell (PBMC) form patients [3], and stimulates the cytokine-mediated production of nitric oxide (NO) in C6 glial cell line [5]. Based on these findings, we investigated whether psychosine-induced toxicity in astrocytes is associated with the release of toxic mediators into the media and/or if psychosine-induced toxicity occurs in a cell-autonomous manner. To investigate this hypothesis, healthy human astrocytes were incubated with media from astrocytes that had been pre-treated with psychosine. Our data showed that this conditioned media did not cause cell toxicity when exposed to healthy human astrocytes. This data suggests that psychosine-mediated astrocytes cell death does not involve the release of mediators into the extracellular space and therefore occurs in a cell-autonomous manner.

Psychosine deregulates astrocytes and oligodendrocytes in concert
Many studies have shown that psychosine causes demyelination and that the dysfunction of oligodendrocytes is associated with Krabbe disease. To support the idea that psychosine also deregulates astrocytes we examined the effect of this toxin on markers of all three glia cell types in a more intact cell culture model using organotypic slice cultures. In agreement with our previous studies [17,18], psychosine reduced the expression of MBP and MOG suggesting demyelination, while showing no effect on the microglia marker Iba1. For the first time, here we show that psychosine reduced expression of the astrocyte marker vimentin. Importantly, DEDA attenuated the psychosine-induced decrease in expression of MBP and MOG as well as vimentin supporting the idea that astrocyte function as well as oligodendrocyte myelination state is altered in Krabbe disease. Furthermore, we note that the increased expression of SMI-32 in the axonal tracts-a marker of axonal damage and demyelination-induced by psychosine is attenuated by DEDA treatment.

Conclusion
This study was initially designed to investigate whether the signalling pathways involved in psychosine-mediated cell death of oligodendrocytes are shared in astrocytes, with a focus toward the role of PLA2. For the first time, we show here that PLA2 and calcium signalling is involved in psychosine-induced toxicity in astrocytes, similar to oligodendrocytes. We confirm that astrocytes are responsive to psychosine treatment in a cell autonomous manner and during psychosine-induced demyelination. These findings provide new mechanistic details psychosine-mediated cell death of glial cells and suggest that targeting both oligodendrocytes and astrocytes for therapeutic intervention may be preferential. We conclude that testing PLA2 inhibitors in the twitcher mouse model may be highly desirable and positive outcomes may pave a way toward a clinical trial for such drugs in Krabbe disease.