Contribution of Pannexin1 to Experimental Autoimmune Encephalomyelitis

Pannexin1 (Panx1) is a plasma membrane channel permeable to relatively large molecules, such as ATP. In the central nervous system (CNS) Panx1 is found in neurons and glia and in the immune system in macrophages and T-cells. We tested the hypothesis that Panx1-mediated ATP release contributes to expression of Experimental Autoimmune Encephalomyelitis (EAE), an animal model for multiple sclerosis, using wild-type (WT) and Panx1 knockout (KO) mice. Panx1 KO mice displayed a delayed onset of clinical signs of EAE and decreased mortality compared to WT mice, but developed as severe symptoms as the surviving WT mice. Spinal cord inflammatory lesions were also reduced in Panx1 KO EAE mice during acute disease. Additionally, pharmacologic inhibition of Panx1 channels with mefloquine (MFQ) reduced severity of acute and chronic EAE when administered before or after onset of clinical signs. ATP release and YoPro uptake were significantly increased in WT mice with EAE as compared to WT non-EAE and reduced in tissues of EAE Panx1 KO mice. Interestingly, we found that the P2X7 receptor was upregulated in the chronic phase of EAE in both WT and Panx1 KO spinal cords. Such increase in receptor expression is likely to counterbalance the decrease in ATP release recorded from Panx1 KO mice and thus contribute to the development of EAE symptoms in these mice. The present study shows that a Panx1 dependent mechanism (ATP release and/or inflammasome activation) contributes to disease progression, and that inhibition of Panx1 using pharmacology or gene disruption delays and attenuates clinical signs of EAE.


Introduction
In multiple sclerosis (MS) and in the animal model experimental autoimmune encephalomyelitis (EAE), the acute disease state is associated with T cell extravasation into the CNS, elevated levels of macrophage/monocyte derived cytokines such as interleukin-(IL)-1b, and macrophage mediated myelin phagocytosis, whereas the chronic disease phase is associated with ongoing cellular losses [1,2].
ATP is involved in diverse signaling platforms and is well known for inducing excitotoxic cell death within the nervous system. In myelinated tissue, ATP initiates an excitotoxic cascade that culminates in apoptosis of oligodendrocytes. Pharmacologic inhibition of the ATP-sensitive ionotropic P2X 7 receptor (P2X 7 R) reduces demyelination, axonal damage, and oligodendrocyte cell death in white matter ischemia and in EAE [3,4]. The non-lytic mechanism of ATP release in EAE has yet to be defined.
Under whole-cell patch clamp recordings, P2X 7 R activation by ATP is marked by an initial small conductance followed by a protracted larger conductance permissive to molecules up to ,1 kDa that is due to recruitment of Pannexin1 (Panx1) channels [5][6][7]. Because of their association, P2X 7 R activation can trigger cellular ATP release by opening Panx1 channels, and thereby the P2X 7 R-Panx1 complex can function as an ATP-sensitive ATP release unit. Independently of P2X 7 R, Panx1 channels can also be activated by voltage, mechanical stretch, and extracellular K + , with high [K + ] out activating Panx1 channels in neurons and astrocytes [8][9][10]. Panx1 channels can be blocked by gap junction channel blockers but at much lower concentrations [11], including mefloquine (MFQ), a quinine derivative that reversibly blocks Panx1 channels at the nM range [12].
Expression of Panx1 in neurons and astrocytes indicates that these channels could be involved in neurodegeneration. Indeed, Panx1-mediated ATP release was documented for cultured spinal cord astrocytes and oligodendrocytes in response to inflammatory mediators and to oxygen/glucose deprivation [3,13]. In vivo, Panx1 has been shown to contribute to neuronal cell death in retinal ischemia [14] and in enteric colitis [15].
In the present study, we tested the hypothesis that inhibition or deletion of Panx1 improves neurologic and pathologic signs of EAE. We show that enhanced Panx1 activity and ATP release is characteristic of the EAE spinal cord. Moreover, blockade of Panx1 channels with MFQ or genetic deletion of Panx1 confers some degree of protection against clinical and pathological signs of EAE, which occurs in part by reducing the infiltration of inflammatory cells during the acute phase and by attenuating the amount of ATP release from diseased CNS tissue. Nevertheless, during the chronic phase of the disease, when no improvement of clinical signs was conferred by deleting Panx1, the upregulation of P2X 7 R found in EAE spinal cords most likely overcomes the decreased ATP release seen in Panx1 KO.

Animals
Male adult Lewis rats and female C57Bl/6 mice were obtained from IFFA CREDO and Charles River, respectively. Panx1 deficient mice were obtained from UCDavis KOMP (allele: Panx1 tm1a(KOMP)Wtsi ) as heterozygous (HT) mice on C57Bl/6 background and bred to homozygosity (Panx1 KO and Panx1 WT) and maintained in an SPF animal facility at Albert Einstein College of Medicine. Only female mice were used in this study.

Induction of Experimental Autoimmune Encephalomyelitis (EAE)
Acute EAE was induced in Lewis rats with inoculum containing 100 mg of guinea pig myelin basic protein (MBP; Sigma) in water, emulsified in equal volumes of Incomplete Freund's Adjuvant (Sigma), supplemented with 500 mg of heat-inactivated M. tuberculosis H37Ra (DIFCO Laboratories). Chronic EAE was induced in 8-10 week old female mice by subcutaneous immunization with 300 mg of myelin oligodendrocyte glycoprotein MOG

Mefloquine Treatment
Daily i.p. injections of the Panx1 channel blocker mefloquine (MFQ; Bioblocks-QU024-1) were started 14 days post-immunization of female mice and clinical signs followed for 38 days. In some experiments, rats and mice received daily i.p. injections of MFQ starting at one week post-immunization and followed thereafter for one or two weeks.

Axon Conduction Latency
Conduction latency of axons of the corticospinal tract of EAE mice treated and untreated with MFQ was measured as previously described [4]. Briefly, conduction latency was measured in anesthetized mice (tribromoethanol: 240 mg/kg, i.p.; Sigma-Aldrich) using a silver chloride recording electrode placed in the vertebral canal at the L2 level and a stimulatory (0-90 V, 0.2 Hz, 50 ms) bipolar electrode placed on the surface of the primary motor cortex (1 mm anterior from Bregma, 2 mm from midline).

Dye Uptake
Mouse spinal cords were removed by insufflation with cold PBS, cut into 330 mm slices using a tissue chopper and immediately incubated in air-bubbled ACSF (145 mM NaCl, 2.5 mM KCl, 3.1 mM CaCl 2 , 1.3 mM MgCl 2 , 10 mM glucose, 10 mM HEPES, pH 7.4) for 30-40 min, at 34uC. Uptake of the dye YoPro was performed as previously described [9]. For that, cervical and thoracic spinal cord slices from non-EAE and EAE mice were bathed for 1 h in ACSF containing 5 mM YoPro-1 (Molecular Probes; Invitrogen), washed 3610 min with ACSF, and fixed in 4% paraformaldehyde overnight. YoPro fluorescence was measured using an epifluorescence microscope equipped with a 4x objective, 488/512 nm excitation/emission filter sets and MetaFluor software version 7.1 (Molecular Devices). YoPro fluorescence was measured from 15-25 slices of cervical/thoracic spinal cord from each mouse. Fluorescence intensity values obtained from regions of interested placed on images obtained from the gray matter tissues of EAE mice were normalized to non-EAE controls and expressed as mean 6 s.e.m.

Quantification of Cell Death
Cell death was assessed by measuring propidium iodide (PI) fluorescence from spinal cord slices incubated for 10 min in ACSF containing PI (10 mM) prior to the addition of YoPro. Spinal cords slices were processed as described above and PI fluorescence intensity was measured from regions of interest placed in the gray matter, using 594 nm excitation filter set.

ATP Assays
ATP released into the ACSF bathing lumbar and sacral spinal cord slices was measured with a Promega Luciferin/Luciferase assay kit and a Turner luminometer. Fifty microliters of ACSF were collected after 30 min incubation and the amount of ATP present in the ACSF normalized to total amount of protein.

Protein Measurements
Spinal cord slices were sonicated in lysis buffer (150 mM NaCl, 10 mM Tris-base, 1% Triton-X, pH 7.4) containing protease inhibitor cocktail (complete, EDTA-free; Roche). After spin down supernatants were used to measure protein content using the BCA assay (Thermo Scientific).

Splenocyte Cultures
Spleens from Panx1 WT and KO mice were homogenized with a 1 ml syringe plunger, passed through a 40 mm filter, centrifuged 8 min at 250 g, and re-suspended in red blood cell lysis buffer (155 mM NH 4 Cl, 10 mM KHCO 3 , 0.1 mM EDTA, pH 7.2) for 5 min with occasional shaking. Cells were centrifuged and resuspended in Dulbecco's Modified Eagle Medium (DMEM; Gibco) supplemented with 10% fetal bovine serum (FBS; Gibco), 1% penicillin/streptomycin (P/S; Gibco), and MEM non-essential amino acids (Gibco). Cells were plated at a density of 6610 7 per 35 mm dish. The next day, non-adherent cells were removed by shaking. Adherent cells were maintained in culture for 2 weeks before use, at which time we found by immunocytochemistry that about 97% cells expressed the macrophage marker CD11b.

IL-1b ELISA Assay
Panx1 WT and KO splenic macrophages were incubated in serum-free DMEM, 1% P/S, MEM non-essential amino acids overnight, and then treated with 1 mg/ml lipopolysaccharide (LPS; Sigma-Aldrich) in media overnight. Cells were then washed in phosphate buffered saline (Dulbecco's PBS) and exposed for 20 min to 5 mM ATP (Sigma). Supernatants were collected at the end of the 20 min ATP stimulation, centrifuged for 10 sec at 10,000 rpm to remove cellular debris, and the clarified supernatant was used for mouse IL-1b ELISA according to manufacturer instructions (PeproTech). The amount of IL-1b in the supernatant was normalized to the total cellular protein levels obtained from cells harvested in 100 ml lysis buffer (1% TritonX-100, 150 mM NaCl, 10 mM Tris-base, Roche EDTA-free complete protease inhibitor; pH 7.4) using the BCA assay. Data are reported as mean 6 SEM of triplicate measurements of IL-1b. Three independent experiments were performed.

Histopathology
Non-EAE and EAE mice were anesthetized with isoflurane, perfused with 20 ml PBS and 20 ml 4% paraformaldehyde, and spinal cords dissected. In some experiments, spinal cords were removed by insufflation with cold PBS. Paraffin-embedded sections from thoracic, lumbar and sacral spinal cord were stained with hematoxylin and eosin (H & E). Lesions were quantified in the ventral funiculus. Areas of parenchymal inflammation were manually outlined, measured using NIH ImageJ software [16] and presented as a percentage of ventral funiculus white matter area.

Western blotting
Mice were anesthetized with isoflurane, decapitated, and spinal cords removed by insufflation with cold PBS. Lumbar spinal cord was sonicated in lysis buffer containing protease inhibitors. Each sample (30 mg) was electrophoresed on 4-20% gradient sodium dodecyl sulfate-polyacrylamide minigels (BioRad) and transferred to nitrocellulose membranes. Membranes were blocked with 2% milk and incubated for 2 hours with antibodies to P2X 7 R (Alomone Lab., 1:1000) or GAPDH (Fitzgerald, 1:10,000). HRPconjugated secondary antibodies were used at 1:3000 (Santa Cruz). Membranes were exposed to enhanced chemilluminescent substrate (Millipore) and detected with X-ray films. Densitometric analysis of bands was performed with ImageJ.

qRT-PCR
Cerebellar tissues of 3 naïve and 3 EAE wild-type mice were used to quantify the levels of Panx1 transcripts. Tissues were minced and homogenized with a Bullet Blender (Next Advance Inc.) and total RNA was extracted using the plus mini kit (Qiagen) according to the manufacturer's protocol. Complementary DNA was synthesized from 1 mg/10 ml of RNA, using a Superscript VILO cDNA Synthesis Kit (Invitrogen). Primers used were Pannexin1 (F: AGCCAGAGAGTGGAGTTCAAAGA; R: CAT-TAGCAGGACGGATTCAGAA) and 18S ribosomal RNA (F: CACGGCCGGTACAGTGAAAC; R: AGAGGAGCGAGC-GACCAAA). The ribosomal 18S was used as a house-keeping gene for normalization. Real-time qRT-PCR was performed using SYBR Green PCR Master Mix with 7300 Fast Real-Time PCR system (Applied Biosystems). Reaction mixtures were denatured at 95uC for 10 min, followed by 40 PCR cycles. Each cycle consisted of the following three steps: 94uC for 15 sec, 57uC for 15 sec, and 72uC for 1 min. Each sample was normalized against internal control (18S ribosomal RNA); the relative values for target abundance was extrapolated from standard curves generated from the reference standard.

Statistical Analyses
Statistical comparisons were made using GraphPad Prism 5.0 software. Clinical signs of disease were compared between groups using unpaired student's t-test or one-way ANOVA followed by Newman-Keuls post hoc analysis, as indicated in the text. P,0.05 was considered significant.

Blockade of Panx1 Channels Delays Onset and Ameliorates EAE Signs
The contribution of Panx1 channels to EAE was first evaluated using the transient EAE model in rats immunized with MBP that were either untreated or treated daily with MFQ (1.0 and 5.0 mg/ kg; starting 7 days post-immunization: 7 dpi). As shown in Fig. 1A, a significant improvement in neurological scores was measured in rats that received the highest concentration of the Panx1 channel blocker (P = 0.022, paired t-test; N = 4 animals per group). When administered at 5 mg/kg, MFQ delayed the disease onset and reduced the EAE symptoms compared to those of untreated rats ( Fig. 1B; P,0.05; N = 13-14 rats). Similarly to what was recorded in rats, daily injections of MFQ (5.0 mg/kg; starting 7 dpi) in mice immunized with MOG also delayed onset and reduced EAE symptoms compared to MFQ-untreated EAE mice ( Fig. 1C; P,0.05; N = 19-20 mice).
In order to evaluate whether the beneficial effect of MFQ was related to an attenuation of peripheral and/or CNS inflammatory responses, we injected MFQ (5 mg/kg) in mice after the onset of the disease. Fourteen days post-immunization, a time point in which the majority of mice had a score 4 (hindlimb hemiparalysis), MFQ was injected daily and scoring performed for 24 more days. As indicated in Fig. 1D, a slight but significant improvement of symptoms was recorded in female mice that received the Panx1 channel blocker, with MFQ-treated mice displaying on average at day 38 post-immunization paralyzed tail and the untreated ones hindlimb weakness.
Further evidence for the beneficial effect of MFQ was obtained by measuring the conduction latency in spinal tracts of EAE mice treated and untreated with MFQ (5 mg/kg). Untreated EAE mice displayed a conduction latency of 5.260.24 ms (N = 16 mice) while EAE mice that were daily treated with MFQ had a significantly shorter conduction latency (4.560.08 ms; N = 17 mice; P,0.05; Fig. 1E).
Thus, together these data indicate a beneficial effect of MFQ and suggest that Panx1 channels contributed to the disease, both by delaying the onset and attenuating EAE symptoms.

Delayed Onset and Reduced Mortality in Panx1 KO Mice with EAE
To more directly evaluate the contribution of Panx1 channels to EAE, we immunized Panx1 KO female mice and compared the clinical score with those obtained from Panx1 WT mice. We performed three independent experiments in which mice were followed for 34 days and conducted another additional experiment which was terminated at 13 dpi to harvest tissue. The clinical course of EAE in Panx1 WT and Panx1 KO female mice obtained from four independent experiments is summarized in Figure 2A To verify whether the differences in disease outcome seen during the chronic phase of EAE in Panx1 WT mice treated with the Panx1 channel blocker (Fig. 1D) and those of Panx1 KO mice ( Fig. 2A) were related to pleiotropic action of MFQ, we compared Our data showing a significant delay in EAE onset in the Panx1 deficient mice suggested that during the initial phase of the disease infiltration of inflammatory cells could be attenuated in these mice. To verify this possibility, histological analysis of inflammatory lesions was performed using H & E stained paraffin sections of spinal cords, focusing on the ventral funiculus white matter, a common site of EAE lesions. Consistent with the hypothesis, at the initial phase of EAE (12-13 dpi), the calculated percent area of ventral funiculus white matter occupied by inflammatory cells in Panx1 WT mice was more extensive (9.662.4%, N = 5 mice) than that measured for Panx1 KO mice (3.960.9%, N = 5 mice; p = 0.04, t-test) (Fig. 2B). In contrast, at the later phase (35-36 dpi) of disease when no significant clinical difference between the two genotypes was measured, the extent of lesions in the spinal cords of Panx1 WT and Panx1 KO mice was similar (Fig. 2C). No inflammation was observed in non-EAE mice. Thus, these data indicate that lack of Panx1 restricts the extent of inflammation during the initial phase of EAE disease and does not contribute to EAE progression but reduces the incidence of animal death.

Reduced IL-1b Release from Panx1 KO Splenic Macrophages
IL-1b released from monocyte-lineage cells contributes to the development of EAE lesions. To evaluate the possibility that the delayed onset of EAE seen in Panx1 KO mice was related to attenuated cytokine production by macrophages, we measured the levels of IL-1b released from activated cultures of splenic macrophages. For that, CD11b-positive adherent splenocytes derived from Panx1 WT and Panx1 KO mice were treated with 1 mg/ml LPS overnight to up-regulate inflammasome components, followed by 5 mM ATP stimulation for 20 minutes and supernatants assayed for IL-1b by ELISA. As shown in Figure 3, LPS activated and ATP stimulated Panx1 WT splenic macrophages released significantly higher IL-1b amounts (49.4167.1 pg/mg protein) than did activated Panx1 KO cells (8.062.9 pg/mg protein). Thus, these results suggest that deficient cytokine production by Panx1 KO splenic macrophages might be related to the reduced area of inflammation in white matter spinal cords which likely delayed the onset of EAE seen in these mice.

Reduced ATP Release from Spinal Cords of EAE Panx1 KO Mice
Activation of Panx1 channels can contribute to elevated levels of extracellular ATP, which has excitotoxic effects [3]. To evaluate whether Panx1 channel activity was altered during EAE, we used the dye-uptake assay. For that, we measured the fluorescence intensity of the dye YoPro in the grey matter of acutely isolated slices of cervical spinal cord from Panx1 WT control mice and from mice displaying mild and severe EAE clinical scores. As shown in Figure 4A, YoPro uptake was significantly increased in Panx1 WT mice with EAE as compared to Panx1 WT controls. A dose-dependent increase in dye uptake with disease severity was also recorded. Spinal cord slices from Panx1 WT mice with mild clinical signs (hindlimb weakness) displayed 1.1860.03 fold higher YoPro uptake compared with control Panx1 WT non-EAE mice (P,0.001). Spinal cords of Panx1 WT mice with severe clinical signs (hindlimb paralysis) displayed significantly higher uptake of YoPro than slices from Panx1 WT mice with mild EAE (p,0.001), and 1.5260.04 fold higher than control Panx1 WT non-EAE. No significant dye uptake was recorded from spinal cords of Panx1 KO mice with or without EAE (Fig. 4A).
The extent to which increased YoPro fluorescence measured in spinal cords of EAE mice was related to cell damage/death was evaluated using the propidium iodide (PI) uptake assay, which at low concentrations yields high fluorescent signals only in damaged cells. For that, spinal cord slices from EAE and non-EAE Panx1 WT and Panx1 KO mice were bathed in ACSF containing 10 mM PI, and fluorescence intensity measured. Slices from Panx1 WT and Panx1 KO animals sensitized for EAE exhibited a similar 1.1260.02 fold increase in PI fluorescence intensity, which was significantly higher compared to their non-sensitized controls (P,0.0001; Fig. 4B). Thus, these data suggest that cell damage is likely to account for about 12% of the YoPro uptake measured in spinal cords of EAE mice, such that Panx1-mediated dye-uptake would be expected to correspond to 40% increase in YoPro fluorescence intensity in cases of severe EAE symptoms.
Based on our findings of increased Panx1 activity in EAE, we then quantified the levels of extracellular ATP in spinal cord of these mice. For that, we measured ATP concentration in aliquots of ACSF bathing the spinal cord slices from control Panx1 WT mice and from Panx1 WT mice sensitized for EAE. We found significantly higher levels of ATP released from slices of EAE (0.3460.04 pmol/mg protein, N = 5 animals; P,0.001) compared to non-EAE (0.1460.02 pmol/mg protein, N = 10 animals) tissues (Fig. 4C). As expected, spinal cord slices from Panx1 WT mice with EAE released more ATP than those from Panx1 KO mice with EAE (WT: 0.3460.04 pmol/mg protein, N = 5 mice; KO: 0.1760.01 pmol/mg protein, N = 8 mice; P,0.001). Spinal cord slices from Panx1 KO mice with EAE displayed a significantly higher level of extracellular ATP (0.1760.01 pmol/mg protein, N = 5 mice; P,0.05) than non-EAE Panx1 KO mice (0.1160.01 pmol/mg protein, N = 8 mice; Fig. 4C), indicating that about 0.06 pmol ATP/mg protein is released via a Panx1independent mechanism, most likely due to cell damage.
Quantitative real time PCR (qPCR) performed on RNA extracted from cerebellum of Panx1 WT mice with EAE (35 dpi; hindlimb paralysis) indicated that Panx1 transcript was 2.8 fold higher than in non-EAE Panx1 WT mice (Fig. 4D).
Together, the dye-uptake and ATP release data indicate that in EAE there is increased Panx1 activity which results in elevated extracellular levels of ATP in spinal cord slices of EAE mice. This increased activity is likely resultant from increased Panx1 expression levels in EAE.

Increased Expression of P2X 7 Receptors in EAE
In view of the results obtained showing that Panx1 KO mice developed as severe EAE as Panx1 WT mice despite decreased extracellular ATP levels, we evaluated whether altered expression levels of the ATP sensitive P2X 7 receptors could account for these results. Indeed we found that these receptors where expressed at higher levels in lumbar spinal cords from EAE mice of both genotypes, compared to non-EAE mice (Fig. 5). Thus, it is likely that upregulation of P2X 7 R provides a mechanism that could lead to the development of EAE despite reduced ATP levels, as seen the Panx1 KO mice.

Discussion
Although the initial event in MS is thought to be autoimmune demyelination, axonal damage and tissue atrophy are frequently observed in chronic MS and correlate well with neurological impairment [17]. Likewise, in the chronic phase of murine EAE, a CD4+T cell inflammatory response is frequently accompanied by damage to axons and loss of neurons and oligodendrocytes [18][19][20]. This neurodegenerative phase may be mediated by abundant free radicals, proinflammatory cytokines such as IL-1b and TNFa, glutamate and ATP. In this study, we present evidence that Panx1 channels contribute to EAE pathology, in part by releasing ATP from diseased tissue. A recent report also indicates that mice with global deletion of Panx1, or with neuronal-targeted conditional deletion of Panx1, are protected from excitotoxic cell death in retina ischemia reperfusion, consistent with a role for Panx1 in direct neurotoxicity [14].
In the present study, we found that female Panx1 KO mice exhibited a significant 2.5-day delay in the onset of clinical signs, whereas at 33 dpi the Panx1 KO mice were as sick as the surviving s.e.m. values of Panx1 transcript (normalized to ribosomal 18S) measured from RNA samples extracted from cerebellum of naïve and EAE Panx1 WT mice. Samples are from 3 naïve and 3 EAE mice. ***P,0.001 and *P,0.05 (ANOVA followed by Newman-Keuls multiple comparison test); P value in part D was obtained using unpaired t-test analysis. doi:10.1371/journal.pone.0066657.g004 Panx1 WT mice. We also recorded decreased number of deaths among Panx1 KO than Panx1 WT mice with EAE. Consistent with their less severe clinical scores in the acute phase, histopathologic analysis of EAE spinal cords at 12-13 dpi showed that inflammatory lesions were decreased in the Panx1 KO. Parenchymal lesions were still present in the Panx1 KO EAE spinal cord, but were less frequent and tended to be smaller in size than those found in Panx1 WT EAE lesions, suggesting that Panx1 is not strictly required for leukocyte blood brain barrier transmigration and lesion initiation, but does participate in lesion expansion.
We also showed that MFQ, a Panx1 channel blocker, delayed and provided protection against clinical signs of EAE in rats and mice. It remains unclear why Panx1 WT mice treated with MFQ, but not Panx1 KO mice, showed decreased disease severity in the chronic phase of EAE. One possibility is related to the hypomorphic phenotype of Panx1 KO mice which display about 30% Panx1 mRNA of that of WT mice [21]. Another possibility may be due to pleiotropic effects of the drug. For instance, mefloquine has been shown to inhibit p-glycoprotein transporter [22] and the adenosine A2A receptor [23], two important players in EAE [24,25]. However, given that MFQ did not alter the outcome of EAE when administered to Panx1 KO mice, these possibilities seem unlikely. Nonetheless, we show by pharmacologic inhibition and genetic deletion that in both conditions Panx1 contributes to disease expression, and in the latter case, also increases mortality of mice with EAE.
In this study we also examined whether Panx1 channels were activated in EAE by measuring the amount of ATP released and YoPro uptake in acute spinal cord slices from mice sensitized for EAE and healthy non-EAE control mice. In control conditions we found that these channels are likely quiescent since equivalent amounts of ATP were detected in the ACSF bathing spinal cord slices of healthy Panx1 WT and Panx1 KO mice. Similarly, no significant differences in dye-uptake were detected in these tissues. In the EAE setting, however, we found that Panx1 channels are active and contribute to ATP release from CNS tissues given that Panx1 KO EAE spinal cords released significantly less ATP than did Panx1 WT EAE spinal cords. It is also likely that the increased Panx1 mRNA expression levels detected in chronic EAE tissues of Panx1 WT is related to the increased levels of extracellular ATP detected in these tissues. At present, it is difficult to evaluate by western blot whether Panx1 expression is altered and/or whether there are post-translational changes during EAE that affect Panx1 activity. Recent study indicates that different Panx1 antibodies yield different bands of sizes in different tissues, including Panx1 knockout mice [26]. Our findings are similar to those previously reported for retinal tissues, where, under normoxic conditions, Panx1 WT and Panx1 KO tissues and cellular permeability to dye were found to be similar, changing only after 30 min of oxygenglucose deprivation [14].
There are several possible ways by which Panx1 channels might contribute to EAE. One mechanism is by inducing ATPdependent excitotoxicity, due to the extrusion of ATP through open Panx1 channels. High extracellular ATP levels activating P2X 7 receptors, either alone or in a physical association with Panx1 in the membrane, would result in membrane permeabilization, influx of Ca 2+ , and cell death. Indeed, previous work has shown that exogenous application of ATP leads to demyelination and oligodendrocyte losses, and that such effects could be prevented by pharmacologic inhibition of the ATP-sensitive P2X 7 R [4]. Our experiments using ex-vivo spinal cords indicate that Panx1 is a predominant mechanism for ATP release in EAE, but we cannot distinguish the cellular source. Panx1 protein and transcript are expressed in neurons, astrocytes, oligodendrocytes, macrophages, and T cells [3,8,9,14,[27][28][29][30]. Because our ATP experiments were conducted on spinal cords during the chronic phase, when the T cell burden is low (Fig. 2), this suggests that at least in the latter phase of the disease, Panx1-mediated ATP release does not come from infiltrating lymphocytes. The fact that P2X 7 R was found to be upregulated in chronic EAE spinal cords, suggest the presence of a mechanism by which disease progresses despite reduced ATP release. This possibility could explain the lack of clinical score improvement in Panx1 KO mice compared to those of surviving WT mice; nevertheless, our results suggest a certain degree of protection in the absence of Panx1 given that number of EAE related death in the null genotype was reduced compared to that of Panx1 WT mice.
The contribution of Panx1 to EAE can also be due to increased levels of adenosine, which is rapidly generated from ATP by CD73. Increased adenosine levels are observed after cellular stress, neuroinflammation and brain injury (for reviews see, [31][32][33]). However, it is not clear whether adenosine has protective or deleterious roles in multiple sclerosis models. Mice that lack CD73 expression have been shown to be protected from EAE [34], and blockade of adenosine receptors with caffeine, a broad spectrum adenosine receptor antagonist have been shown to attenuate EAE symptoms [35]. However, reports in which the deletion of specific adenosine receptors was evaluated indicate a dependence of disease severity on receptor subtypes. Thus, A1 and A2A adenosine receptor activation is protective in EAE models ( [24,36] but see [34]) whereas the opposite effect is observed with A2B receptors [37].
Another way by which Panx1 may contribute to EAE involves inflammasome activation. In neurons and astrocytes, Panx1 and P2X 7 receptors co-immunoprecipitate with each other and with components of the inflammasome [8]. In this modality, ATP released through Panx1 activates P2X 7 R triggering the recruitment of caspase-1 to the inflammasome, which then cleaves pro-IL1b into its mature form [30,38]. IL-1b is found at high levels in EAE and MS lesions [39], primarily derived from macrophages and microglia, but also can be produced in neurons and astrocytes. IL-1b is a potent activator of astrocytes [40], and is important for inducing IL-17 production from CD4 and gamma-delta T-cells [41][42][43]. Disruption of the Panx1-P2X 7 R complex, in the absence of Panx1 signaling, with consequent prevention of inflammasome activation and IL-1b processing represents an attractive explanation for the delayed onset of clinical and pathological signs of EAE observed in our Panx1 KO mice and in rats treated with MFQ. Such possibility is consistent with our results showing that IL-1b secretion is impaired in ATP-stimulated splenic macrophages from Panx1 KO mice. Similar findings were also reported for macrophages treated with Panx1 siRNA [6], and for neurons and astrocytes treated with the Panx1 inhibitor probenecid [8]. Moreover, retinas from global and neuronal-targeted Panx1 KO were also shown to fail to activate the inflammasome and the processing of IL-1b triggered by ischemic events [14].
Several lines of evidence also point to a role for P2X 7 R itself in white matter damage. P2X 7 R expression is elevated in MS white matter [4,44]. Similarly, P2X 7 R is elevated in neurons and astrocytes during acute EAE and after symptom recovery [45,46]. A gain-of function P2X 7 R polymorphism that increases Ca 2+ permeability occurs at higher frequency in MS patients than in the general population [47]. The contribution of P2X 7 R signaling to neuroinflammation has also been directly tested in EAE. Genetic deletion of P2X 7 R alters severity of and susceptibility to EAE, although differing effects were found in independently derived P2X 7 R KO mice [48,49]. The P2X 7 R inhibitors brilliant blue G (BBG) and oxidized-ATP (oATP) attenuate EAE signs, even when administered after onset of disease [4]. Interestingly, it has been recently shown that both BBG and oATP are potent inhibitors of Pannexin1 channels [50,51].
It should be kept in mind, however, that mechanisms other than the ones highlighted here are likely involved in EAE. Also possible is that responses to the absence of Panx1 vary between the immune system and the CNS. Thus, the lack of Panx1 may attenuate the immune response during the initial stages of EAE (as supported by data displayed in Fig. 3), while ongoing CNS damage in the chronic EAE phase may rely, in addition to Panx1, in other yet unknown mechanisms. Micro-array analysis would certainly provide an unbiased approach to unveil such mechanisms.
In summary, we have shown that heightened ATP release is characteristic of the EAE spinal cord, and that most of the ATP release can be accounted for by Panx1 channel activity. Panx1 is shown to contribute to both early and late stages of EAE, but the disease-inducing effect of Panx1 seems most profound during the acute phase. Genetic ablation or blockade of Panx1 delays the onset of clinical signs of EAE. During the chronic phase, wild-type mice had high incidence of mortality, while Panx1 KO mice were protected from this mortality, although surviving mice expressed clinical signs of disease. Our data suggest that a Panx1-mediated pathway (ATP release and/or inflammasome activation) contributes to neuroinflammatory damage in the diseased spinal cord, and open the possibility that modulation of Panx1 could be a novel therapeutic target in multiple sclerosis.