Increased expression of the ATP‐gated P2X7 receptor reduces responsiveness to anti‐convulsants during status epilepticus in mice

Refractory status epilepticus is a clinical emergency associated with high mortality and morbidity. Increasing evidence suggests neuroinflammation contributes to the development of drug‐refractoriness during status epilepticus. Here, we have determined the contribution of the ATP‐gated P2X7 receptor, previously linked to inflammation and increased hyperexcitability, to drug‐refractory status epilepticus and its therapeutic potential.

including an increased risk of developing epilepsy and cognitive decline (Holmes, 2015). The most common causes of status epilepticus are inadequate doses of anticonvulsant drugs in patients with epilepsy, stroke, traumatic brain injury (TBI), toxic-metabolic encephalopathies and neurological and systemic infections (Mayer et al., 2002).
Frontline pharmacotherapy for status epilepticus is primarily via treatment with benzodiazepines, such as lorazepam or diazepam, followed by anti-seizure drugs (ASDs), such as phenytoin or phenobarbital (Crawshaw & Cock, 2020). Treatments fail, however, in 31-43% of cases, resulting in refractory status epilepticus (RSE). This is defined as status epilepticus which fails to respond to ≥2 ASDs including at least one non-benzodiazepine, with the time spent in status epilepticus increasing the probability of developing drug unresponsiveness. Critically, RSE is linked with a three-fold increase in mortality compared to normal status epilepticus and considerable morbidity (Betjemann & Lowenstein, 2015). The causes and mechanisms of RSE development are incompletely understood, but molecular changes thought to contribute include the internalization of GABA receptors and increased surface expression of excitatory glutamate (NMDA and AMPA) receptors (Betjemann & Lowenstein, 2015). Consequently, there is a pressing need to identify new drug targets that act either independently of GABAergic inhibitory signalling or that can What is already known • P2X7 receptor expression is increased in the brain after status epilepticus.
• Drugs targeting the P2X7 receptor modulate seizure severity during status epilepticus.

What this study adds
• Elevated P2X7 receptor expression leads to resistance to anticonvulsant drugs.
• ATP-mediated extracellular signalling may contribute to drug refractoriness during status epilepticus.

What is the clinical significance
• The P2X7 receptor contributes to unresponsiveness to anticonvulsant treatments during status epilepticus.
• Blockade of the P2X7 receptor may represent a novel adjunctive therapy for drug-refractory status epilepticus. be given as adjunctive treatment to enhance the effects of current anti-seizure medication.
A role for inflammatory mediators in neuronal function and excitability is well established . Cytokines, such as IL-1β or TNF-α, modulate the function of GABA receptors and potentiate excitatory NMDA receptor synaptic transmission, alter voltagegated ion channel function and expression, and regulate presynaptic exocytosis of both excitatory and inhibitory neurotransmitters (Huang et al., 2010;Rossi et al., 2012;Vezzani & Viviani, 2015). Importantly, several drugs targeting inflammatory signalling pathways have anticonvulsive effects during status epilepticus . An emerging target is the ionotropic, ATP-gated P2X7 receptor (Beamer et al., 2021). This ion channel functions as a gatekeeper of inflammation by driving the assembly of the NLRP3 inflammasome and subsequent release of IL-1β (Di Virgilio et al., 2017). The P2X7 receptor has some unique characteristics that make it particularly attractive: (a) In comparison to other P2X receptors, it has a low sensitivity to ATP, suggesting that it is mainly activated under pathological conditions when high volumes of ATP are released, as during a seizure (Beamer et al., 2019). This could spare patients from unwanted side effects during treatment based on P2X7 receptor antagonists. (b) expression of P2X7 receptors increases in the brain following status epilepticus and (c) blockade of the P2X7 receptor with drugs modulates seizure severity and alters the resulting neuropathology in several experimental models of status epilepticus (Beamer et al., 2021;Engel et al., 2012;Jimenez-Pacheco et al., 2013;Jimenez-Pacheco et al., 2016;Kim & Kang, 2011). Whether increased expression or function of the P2X7 receptor contributes to drug-refractoriness during status epilepticus, however, has not been explored to date.
Here, we used a model of focal-onset status epilepticus in mice to determine whether P2X7 receptors contributed to the development of the refractory form of status epilepticus, RSE. Expression of P2X7 receptors was studied by using P2X7 receptor reporter mice and highly specific P2X7 receptor nanobodies (Danquah et al., 2016;Kaczmarek-Hajek et al., 2018). P2X7 receptor function was investigated using a combination of pharmacological and genetic mouse models. We show that drug refractoriness in status epilepticus is, at least in part, mediated via P2X7 receptor-driven inflammation and that inhibition of the receptor can restore responses to current antiseizure medication. All mice used in our experiments were 8-12 weeks old, with a weight range between 25 and 30 g. The following strains and sexes were used: male C57Bl/6 OlaHsd wild-type (WT) mice, obtained from the Biomedical Research Facility at RCSI; heterozygous FVB/N-Tg (RP24-114E20-P2X7/StrepHisEGFP)Ani (line 17) BAC transgenic mice that express the enhanced green fluorescent protein (EGFP) immediately upstream of a P2rx7 BAC clone over-express P2X7 receptors and are referred to in this paper as P2X7-OE mice (Kaczmarek-Hajek et al., 2018) and the respective WT litter mates (to reduce breeding, we used male and female mice); male C56BL/6N-P2rx7 tm1d(EUKOMM)wtsi P2X7R knock-out (KO, P2X7 À/À ) mice, obtained from A. Nicke, LMU Munich; and male mice C57BL/6-Nlrp3 tm1Vmd KO for NLRP3 (Nlrp3 À/À ), C57BL/6-Pycardt m1Vmd KO for ASC (Pycard À/À ), B6N.129S2-Casp1 tm1Flv/J KO for caspase-1 and caspase-11 (Casp1/11 À/À ), obtained from I. Couillin, University of Orleans.
Convulsive status epilepticus in rodents has been used extensively to model seizures and to identify potential novel treatments (Loscher, 2017). Mice are a valuable experimental model of status epilepticus for several reasons including the fact that they share major aspects of brain circuitry with humans, such as the organization and function of the hippocampus. Mice are also large enough to enable multi-channel EEG recordings to score seizures and undergo stereotyped behavioural responses during seizures that have human correlates (e.g., tonic-clonic components). Another reason for using mice is the availability of numerous transgenic lines to study effects of a specific target gene, using P2X7 receptor KO or overexpression (OE).
The exact numbers of mice per experimental group are provided in the respective figure legends. The sample size was calculated using G*Power 3.1.9.4 software with inputs based on data recorded in similar previously performed experiments to determine suitable sample sizes necessary for detecting differences. Inputs for power analyses in these studies were taken from raw data reported in Engel et al. (2012) and Jimenez-Mateos et al. (2012). Student's t test: confidence intervals were set at 0.95, giving α = 0.05. Power was set at 1 À β = 0.8.
The difference between groups (μ1 À μ2) was 35, while standard deviation (σ) was 21. Using these parameters, a group size of 14 was derived.

| Mouse model of status epilepticus
Status epilepticus was induced as described previously (Alves et al., 2019). Briefly, during stereotaxic procedures, mice were anaesthetised using isoflurane (5% induction, 1-2% maintenance) and maintained normothermic (body temperature was maintained between 36 C and 37 C) by means of a feedback-controlled heat blanket, controlled by a rectal probe (Harvard Apparatus Ltd, Edenbridge, Kent, UK). The depth of the anaesthesia was frequently tested by checking the plantar nociception or corneal reflex. Additionally, to minimize pain during and post-surgery, mice were treated with buprenorphine (0.05 mgÁkg À1 ) and EMLA cream (Aspen Pharma, Maidenhead, UK) which was applied to head wounds and ear bars.
Once fully anaesthetised, mice were placed in a stereotaxic frame and a midline scalp incision was performed to expose the skull. A guide cannula (coordinates from Bregma; AP = À0.94 mm, L = À2.85 mm) and three electrodes (Bilaney Consultants, Sevenoaks, UK), one on top of each hippocampus and with the reference on top of the frontal cortex, were fixed in place with dental cement. An Xltek recording system (Optima Medical, Guildford, UK) was used to record electroencephalogram (EEG). Following a recovery period of approximately 1 h post-surgery, status epilepticus was induced via a microinjection of 0.2-μg (FvB/NJ background) or 0.3 μg (C57/Bl6 background) of kainic acid (KA) in 0.2 μl phosphate-buffered saline (PBS) into the right basolateral amygdala into awake, hand-restrained mice. Vehicle-injected control animals received 0.2 μl of PBS. The anticonvulsants lorazepam (6 mgÁkg À1 ; Mouri et al., 2008), phenytoin (10 mgÁkg À1 ; Loscher, 2007), carbamazepine (40 mgÁkg À1 ; Twele et al., 2016) or midazolam (8 mgÁkg À1 ; Diviney et al., 2015) were injected i.p., 40 min following intra-amygdala KA at doses previously shown to reduce seizures. EEG was recorded for 10 min before intra-amygdala KA (baseline), during the time of intra-amygdala KA until treatment with anticonvulsant (status epilepticus) and for 60 min post-anticonvulsant treatment (post-status epilepticus). Post-status epilepticus, mice were evaluated using a scoring sheet for scoring endpoints in rodents, approved by the HPRA, and the mouse Grimace scale. Mice were humanely killed via cervical dislocation by a trained individual, if not otherwise indicated.

| EEG and behavioural analysis
EEG data were uploaded onto Labchart7 software (AD Instruments), as described earlier (Alves et al., 2019). EEG total power (μV 2 ) is a function of EEG amplitude over time and was analysed by integrating frequency bands from 0 to 100 Hz. Power spectral density heat maps were generated within LabChart7 (spectralview), with the frequency domain filtered from 0 to 40 Hz and the amplitude domain filtered from 0 to 50 mV. Data are presented as n-fold to baseline recordings prior to intra-amygdala KA injection, if not indicated otherwise.
Behavioural seizures were scored according to a modified Racine Scale, as reported previously (Jimenez-Mateos et al., 2012): score 1, immobility and freezing; score 2, forelimb and or tail extension, rigid posture; score 3, repetitive movements, head bobbing; score 4, rearing and falling; score 5, continuous rearing and falling; score 6, severe tonic-clonic seizures. Mice were scored every 5 min for 40 min after KA injection. The highest score attained during each 5-min period was recorded by an observer, who was blinded to treatment.

| Plasma and brain concentration of AFC-5128
To ensure sufficiently high brain levels of the P2X7 receptor antagonist AFC-5128, mice were treated i.p. with 50 mgÁkg À1 AFC-5128, as before (Fischer et al., 2016). Plasma was prepared from blood collected into tubes containing 15 μl of 0.5-M EDTA (pH 7.4) via puncture of the saphenous vein 30 min following i.p. injection of 50 mgÁkg À1 AFC-5128. Plasma was prepared by centrifuging the tubes at 1300 Â g, for 10 min at 4 C. Blood (terminal bleed) and brains were collected 60 min following 50 mgÁkg À1 AFC-5128 i.p.
injection. Blood was immediately frozen down and brains removed following perfusion with ice cold PBS and immediately put on dry ice.
Blood, plasma and brain samples from mice treated with AFC-5128 were analysed at the Lead Discovery Center (Dortmund, Germany).
Briefly, AFC-5128 was extracted from plasma and homogenized brain by protein precipitation using acetonitrile. Resulting filtrates were analysed by liquid chromatography tandem-mass spectrometry (LC-MS/MS) using a Prominence UFLC system (Shimadzu, Duisburg, Germany) coupled to a QTrap 5500 instrument (ABSciex, Darmstadt, Germany). The compound was separated on a Zorbax Eclipse Plus C18 column (Agilent Technologies, Santa Clara, CA, USA) with an acetonitrile/water gradient containing 0.1% formic acid as solvent.
Plasma, blood and brain concentrations were calculated by means of a standard curve (Fischer et al., 2016) (Figure S1).

| RNA extraction and qPCR
RNA extraction was performed using the Trizol method, as described before (Alves et al., 2019). Quantity and quality of RNA were measured using a Nanodrop Spectrophotometer (Thermo Scientific, Rockford, IL, USA). Samples with a 260/280 ratio between 1.8 and 2.0 were considered acceptable; 500 ng of total RNA was used to produce complementary DNA (cDNA) by reverse transcription using SuperScript III reverse transcriptase enzyme (Invitrogen, Waltham, MA, USA) primed with 50 pmol of random hexamers (Sigma). Quantitative real-time polymerase chain reaction (qPCR) was performed using the QuantiTech SYBR Green kit (Qiagen Ltd, Hilden, Germany) and the LightCycler 1.5 (Roche Diagnostics, GmbH, Mannheim, Germany). Each reaction tube contained 2 μl cDNA sample, 10 μl

| Cytokine measurement in brain tissue
Levels of IL-1β in the hippocampus were measured using the DuoSet ELISA kits from R&D Systems (Abingdon, UK) following the manufacturer's instructions (mouse IL-1β/IL-1F2, Cat #: DY401-05). In a 96-well ELISA plate, the detection antibody was incubated overnight at room temperature. Then, 100 μl of the samples (50 ng) and standard curve (IL-1β: from 15.6-1000 pgÁml À1 ) were added to the wells and incubated for 2 h at room temperature, followed by incubation with 100 μl of streptavidin-HRP complex. A colour reaction, caused by the addition of a substrate solution (100 μl) and terminated via stopping solution (50 μl), was quantified at 450 and 570 nm using a microplate reader. The cytokine concentration was obtained following the manufacturer's recommendations; 570 nm values were subtracted from the 450 nm values. The log10 of the standard curve values were plotted, and a line of best fit was generated. The amount of cytokines was extrapolated using standard curve and average of calculated triplicate samples. Cytokine concentration was then expressed per mg of total protein in tissue.

| Fluoro-Jade B
Status epilepticus-induced neurodegeneration was analysed as before using Fluoro-Jade B (FjB) staining (Alves et al., 2019). Seventy-two hours post-status epilepticus, mice were transcardially perfused with 15 ml of PBS and brains were removed and flash frozen in 2-methylbutane (Sigma-Aldrich). Coronal tissue sections (12 μm thick) at the medial level of the hippocampus (Bregma AP = À1.94 mm) were cut directly onto a glass slide using a CM1900 cryostat (Leica, Wetzlar, Germany) and stored at À80 C. Tissue was fixed in 4% paraformaldehyde (PFA), rehydrated in ethanol, and then transferred to a 0.006% potassium permanganate solution followed by incubation with 0.001% FjB (Chemicon Europe Ltd, Chandlers Ford, UK). DPX mounting solution was used to mount the sections. Using an epifluorescence microscope, FjB-positive cells including all hippocampal subfields (dentate gyrus [DG], CA1 and CA3) were counted, without knowledge of the treatment, under a 40x lens in two adjacent sections and the average determined for each animal.
Sections were mounted using FluorSave (Merck Millipore), and two images from each hippocampal subfield were obtained using a 40Â lens in the Nikon 2000s epifluorescence microscope. The number of cells was then counted, without knowledge of the treatment, and the results presented as the average count from two images.

| Immunofluorescence
Mice were transcardially perfused with PBS (5 ml) and 4% PFA (10 ml) and brains removed. Following an additional 24 h long post-fixation in 4% PFA at 4 C, brains were transferred to PBS and immersed into 4%

| Three-dimensional morphological analysis of microglia
To analyse morphological changes of microglia, we carried out immunofluorescence staining, as described in the previous section. Images were reconstructed with ImageJ to determine an average greyscale threshold value. Images were subsequently rendered in 3D using FluoRender Version 2.25.0. Three cells from each subfield were selected at random by a reviewer, blind to groups. Cells were isolated, with background noise removed as much as possible using the "diffusion paintbrush" option. Once removed, volumetric analysis was performed, using a minimum greyscale threshold value obtained from FIJI so as to only perform analysis on "real" signal, produced by antibody binding. Cell process length was then measured on the same software, using the multipoint measurement tool, beginning from the centre of the soma (located using DAPI) to the most extreme points of the cell process. Only primary processes were analysed, which means processes that extended directly from the soma, as opposed to secondary or tertiary processes branching of the primary cell process. Average process length was calculated as the mean length of all primary processes extending from the cell body of each individual cell.

| Ex vivo electrophysiology
Status epilepticus was induced in WT and P2X7R-OE mice, as described above (Section 2.2) and allowed to continue for 40 min.
After this time, mice were killed by cervical dislocation for ex vivo brain slice preparation (mice were not treated with lorazepam and slices recorded without knowledge of genotype). Brains were quickly dissected and submerged in oxygenated ice-cold sucrose artificial cerebrospinal fluid (ACSF; composition: in mM: 205 sucrose, 10 glucose, 26 NaHCO 3 , 1.2 NaH 2 PO 4 .H 2 O, 2.5 KCl, 5 MgCl 2 , 0.1 CaCl 2 ); 400-μm horizontal slices were prepared using a vibratome (Campden 7000 smz II, Campden Instruments, Loughborough, UK), with bath temperature held at $1 C. Only slices from the hemisphere ipsilateral to KA injection were retained and stored at room temperature in a submerged-style holding chamber, filled with oxygenated ACSF (in mM: 125 NaCl, 10 glucose, 26 NaHCO 3 , 1.25 NaH 2 PO 4 .H 2 O, 3 KCl, 2 CaCl 2 , 1 MgCl 2 ). For recording, slices were transferred to a membrane chamber which was perfused with oxygenated ACSF, heated to 34 C, at a rate of 16 mlÁmin À1 (Morris et al., 2016). Slices Data are presented as means ± SEM. One-way ANOVA parametric statistics with post hoc Fisher's protected least significant difference test was used to determine statistical differences between three or more groups. Unpaired Student's t test was used for two-group comparisons. Two-way ANOVA was used for repeated measures between groups where a series of measurements have been taken from the same mouse at different time-points. Statistical analysis was undertaken only for studies where each group size was at least N = 5.
Other data where group sizes were below N = 5, are indicated as preliminary data. Significance was accepted at *P < 0.05.

| P2X7 receptor overexpression causes broad resistance to anticonvulsants during status epilepticus
To test whether P2X7 receptor overexpression contributes to broad drug-unresponsiveness in status epilepticus, we tested three additional commonly used ASDs for status epilepticus in our model. These were the positive allosteric GABA receptor modulator midazolam and the anticonvulsive drugs, phenytoin and carbamazepine. For the latter, the exact mechanism of action is not known but appears to involve sodium channel blocking.
As before, no differences could be observed in seizure severity between WT and P2X7R-OE mice during the time from intraamygdala KA until the administration of anticonvulsants 40 min later (Figure 3a-c). In contrast and as observed for lorazepam, all  and proliferation has been described (Monif et al., 2009). Moreover, microglia are among the first cell types to respond to injury, and this is mediated via extracellular ATP (Davalos et al., 2005). We therefore hypothesized that the effects of P2X7 receptor overexpression during status epilepticus are mediated, at least in part, via its proinflammatory function in microglia and focused on this cell type.

| P2X7 receptor overexpression-induced unresponsiveness to anticonvulsants is a consequence of increased inflammation
Double-immunostaining of brain sections from WT mice collected 60 min post-intra-amygdala KA injection, a time-point when sensitivity to lorazepam is reduced in this model , showed a more intense staining of P2X7 receptors on Iba-1-positive microglia in the hippocampus including all three hippocampal subfields. Microglia analysed at the same time-point showed also an increased immunoreactivity to Iba-1 and larger cell bodies, indicating a more activated state (Figures 4a and S3). These findings are consistent with status epilepticus driving an early up-regulation of P2X7 receptors in microglia that might influence their inflammatory state and suggest that P2X7 receptor-driven unresponsiveness to anticonvulsants is  Kobayashi et al., 2013). No difference in response to lorazepam treatment was seen between P2X7R-OE and WT mice, if mice were pretreated with minocycline ( Figure 4d). This is consistent with the hypothesis that a reduced response to anticonvulsants in P2X7R-OE mice is partly mediated by increased inflammation. Notably, we found no gross differences in microglia morphology between WT and P2X7R-OE mice under naïve conditions with microglia from both genotypes showing similar volumes of both soma and whole cell bodies including all microglial processes, thus displaying the typical morphology of resting microglia (Figure 4e). In contrast, microglia from P2X7R-OE mice obtained 40 min post-KA injection showed a slightly smaller cell volume and significantly less primary processes than those from WT mice (Figure 4e), consistent with a more activated, amoeboid and pro-inflammatory phenotype.
Finally, to investigate whether there is an effect of P2X7 receptor overexpression on brain network activity during status epilepticus (40 min post-KA treatment), we analysed electrophysiological responses in hippocampal slices from WT and P2X7R-OE mice subjected to status epilepticus (Figure 5a). This revealed that slices from P2X7R-OE mice displayed an increased population synaptic potential in hippocampal CA1, indicating greater synaptic activation within the hippocampal network, which may contribute to the diminished response to anticonvulsants observed in these animals (Figure 5b,c).

| P2X7 receptor overexpression effects during status epilepticus are independent of NLRP3 inflammasome activation
As overexpression of the P2X7 receptor impaired responses to anticonvulsants during status epilepticus, we reasoned that loss of these receptors may improve their efficiency. As shown for P2X7R-OE mice, P2X7 À/À mice showed a normal expression of genes involved in neurotransmission including Slc6a1, GRM5 and Gria2 (Figure 6a).
As reported previously, using the same P2X7 À/À strain (Conte et al., 2020), P2X7 À/À mice and WT mice showed similar seizure severity during status epilepticus (Figure 6b,c). However, in line with the P2X7 receptor contributing to anticonvulsant unresponsiveness during status epilepticus, P2X7 À/À mice showed a better response to F I G U R E 5 P2X7R-OE mice exhibit more network excitability in the hippocampus after status epilepticus (post-SE). (a) Diagram showing electrode placement. A stimulating electrode was placed in the Schaffer collateral pathway, and two recording electrodes placed in CA1 strata pyramidale (for population spike) and radiatum (for population postsynaptic potential). (b) Representative traces taken from P2X7R-OE and WT mice, after 40 min of status epilepticus. (c) Overall data (means ± SEM) show that population spike amplitude was unchanged between P2X7R-OE and WT mice. Population synaptic potential (PSP) slope was significantly higher in P2X7R-OE mice after 40 min status epilepticus. All data are from N = 4 mice per genotype and two to three slices per mouse. *P < 0.05, significant effect of treatment; repeated measures two-way ANOVA lorazepam when compared with WT mice (Figure 6b,c). ELISA assays using hippocampal tissue extracted 40 min post-intra-amygdala KA showed reduced IL-1β levels in P2X7 À/À mice, when compared with WT mice (Figure 6d), suggesting altered inflammatory pathways in P2X7 À/À mice.
The P2X7 receptor has been described as one of the main activators of the NLRP3 inflammasome leading to the release of IL-1β (Pelegrin, 2021). If effects of P2X7 receptors on responsiveness to anticonvulsants during status epilepticus were mediated via the P2X7 receptor-NLRP3 axis, then responses of mice deficient in components F I G U R E 6 P2X7R effects during status epilepticus are independent of the P2X7R-NLRP3 axis. (a) Graph showing absence of P2rx7 transcripts in the hippocampus of P2X7R KO mice (P2X7 À/À ), compared with WT mice, in naive conditions. Data are shown as individual values with means ± SEM; N = 5 per group. *P < 0.05, significantly different from WT; unpaired Student's t test. No difference in hippocampal mRNA levels of Grm5, slc6a1 and Gria2 in P2X7 À/À mice, compared with WT mice, under naïve conditions (N = 5 per group). (b, c) No difference in EEG total power between WT and P2X7 À/À mice from the time of intraamygdala KA until lorazepam administration 40 min later. Reduced EEG total power in P2X7 À/À mice compared with WT mice, during a 60-min recording period post-status epilepticus (post-SE). Data are shown as individual values with means ± SEM; N = 8 per group. *P < 0.05, significantly different from WT; unpaired Student's t test. (d) Graph showing a reduction in the level of the cytokine IL-1β in P2X7 À/À mice 40 min following intra-amygdala KA treatment. Data are shown as individual values with means ± SEM; N = 4 (WT) and 7 (P2X7 À/À ). *P < 0.05, significantly different from WT; unpaired Student's t test. (e) No difference was observed in the EEG total power during and after status epilepticus, between WT and ASC À/À , NLRP3 À/À and CASP1 À/À mice. N = 9 (WT), 7 (ASC À/À ), 9 (NLRP3 À/À ) and 7 (CASP1 À/À ) of the NLRP3 signalling pathway should mimic the improved response of P2X7 À/À mice during status epilepticus. To test this hypothesis, caspase-1/11, NLRP3 or the inflammasome adaptor protein apoptosis-associated speck-like protein containing a CARD (ASC) KO mice were subjected to intra-amygdala KA and seizure severity, compared with WT mice. Unexpectedly, no significant differences were observed between WT mice and the different KO strains either during status epilepticus or post-lorazepam administration (Figure 6e), suggesting that resistance to anticonvulsants is NLRP3-independent.
Of note, although not statistically different, Casp 1/11 À/À mice showed consistently increased seizure severity when compared with WT mice. Moreover, no significant difference in hippocampal neurodegeneration between genotypes could be observed 72 h poststatus epilepticus ( Figure S4).
Taken together, our results suggest that the effects of the P2X7 receptor on anti-convulsant drug responses during status epilepticus involve microglia activation but are independent of NLRP3 inflammasome activation and that NLRP3 inflammasome inhibition has no effect on seizure severity or neuropathology in the intraamygdala KA mouse model.

| P2X7 receptor antagonists improve responses to anticonvulsants during status epilepticus in mice pretreated with LPS
We next sought to explore whether P2X7 receptor antagonism could be effective in a more clinically-relevant model of elevated inflammatory tone at the time of status epilepticus, compared to our genetic model. To induce inflammation, mice were pretreated i.p. with the bacterial component LPS (Beutler & Poltorak, 2000), prior to intraamygdala KA (Figure 7a). Preliminary data shows that treatment with LPS led to increases in Iba-1 and P2X7 receptor protein expression in the hippocampus, 72 h post-LPS treatment ( Figure 7b). Next, we sought to establish whether LPS-treated mice, similar to P2X7R-OE mice, were also less responsive to anticonvulsants during status epilepticus and, if so, whether deletion or blocking of P2X7 receptors could reverse LPS-induced drug-unresponsiveness. EEG analysis revealed that, whereas vehicle-treated WT mice showed seizure severity during status epilepticus, similar to that in LPS-treated WT mice, they responded significantly better to anticonvulsant treatment than LPSpretreated mice (Figure 7c). Further supporting a P2X7 receptormediated drug unresponsiveness, LPS-pretreated P2X7 À/À mice responded $50% better to the anticonvulsant lorazepam adminis-  Figure S5).
In summary, we conclude that P2X7 receptor expression was increased under inflammatory conditions in the brain, leading to a lower response to anticonvulsants during status epilepticus. Inhibition of P2X7 receptors, therefore, represents a promising novel treatment strategy for status epilepticus where underlying neuroinflammation is one of the pathological hallmarks. showing that the administration of ATP and the ATP analogue 2 0 ,3 0 -O-(4-benzoyl-benzoyl)ATP (BzATP) into the brain increases seizure severity during status epilepticus Sebastian-Serrano et al., 2016). Moreover, P2X7 receptors appear to interfere with drug responses at later time-points, when refractoriness to anticonvulsants becomes evident . P2X7 receptors are activated by high extracellular concentrations of ATP (EC 50 ≥ 100 μM) (Surprenant et al., 1996), most likely reached under ongoing pathological conditions, such as prolonged seizures and sustained inflammation (Vezzani et al., 2011). Also, its high activation threshold (0.3-to 0.5-mM ATP) decreases during inflammatory conditions (0.05-to 0.1-mM ATP) (Di Virgilio et al., 2017). Several studies have shown only moderate or no anticonvulsant effects of P2X7 receptor antagonism during acute seizures such as in the maximal electroshock, the pentylenetetrazol and the 6 Hz psychomotor test (Fischer et al., 2016;Nieoczym et al., 2017). This is in line with the need for an ongoing pathological process and ATP-enriched extracellular milieu for activation of P2X7 receptors in seizures and suggests that treatments based on these receptors might be more effective once pathological processes, such as inflammation, have been initiated.

| DISCUSSION
Our findings that P2X7 À/À mice responded better to treatment with lorazepam and that targeting of P2X7 receptors improves LPS-induced drug-refractoriness makes a strong case for the use of antagonists of these receptors, as adjuvant therapy in RSE. In line with this, the combination of P2X7 receptor antagonists with the ASD carbamazepine, increased the seizure threshold in the maximal electroshock seizure test (Fischer et al., 2016) and intra-amygdala KA induced-status epilepticus   (Amhaoul et al., 2016) and frequency (Jimenez-Pacheco et al., 2016) of seizures in epileptic mice and rats. Of note, mice, in which epilepsy was induced via intra-amygdala KA, were shown to be resistant to low doses of carbamazepine (Welzel et al., 2020).
How P2X7 receptors promote drug-refractoriness during status epilepticus remains to be established. Inflammatory processes induced by activated P2X7 receptors represent, however, a likely explanation.
Expression of P2X7 receptors has repeatedly been shown to be increased in the brain post-status epilepticus Jimenez-Pacheco et al., 2013) with the majority of studies suggesting a dominant up-regulation in microglia (Dona et al., 2009;Kaczmarek-Hajek et al., 2018). By using highly specific P2X7 receptor nanobodies (Danquah et al., 2016), we have shown that P2X7 receptor expression is increased in microglia as soon as 60 min post-KA injection, when mice show resistance to lorazepam  Our results suggest that P2X7 receptor-induced inflammatory responses are mainly mediated by microglial activation. We cannot, however, exclude the contribution of other cell types. While we did not find expression of P2X7 receptors in astrocytes here, these have also been shown to be activated by these receptors (Khan et al., 2019). Blockade of P2X7 receptors has further been shown to prevent astrocyte death following status epilepticus (Kim et al., 2009).
Of note, while microglia had previously been ascribed a mainly proconvulsant role, recent data also suggests a protective function of microglia with the depletion of microglia leading to a decreased seizure threshold (Badimon et al., 2020). This suggests that we must identify and target specific pro-convulsant pathways in microglia, rather than inhibiting global microglia function.
Which pathways are activated by P2X7 receptors in microglia during status epilepticus requires further investigation. An obvious explanation would be that pro-convulsive effects are mediated via increasing IL-1β levels. IL-1β reduced the anticonvulsant action of the benzodiazepine midazolam in primary murine cortical neuron cultures (Clarkson et al., 2017) and IL-1 receptor blockers such as anakinra protected the brain from damage induced by status epilepticus (Noe et al., 2013). An unexpected finding was, therefore, that genetic ablation of different components of the NLRP3 inflammasome pathway led a seizure phenotype, similar to that in WT mice. This is in sharp contrast to previous studies, which reported that suppression of the NLRP3 inflammasome-caspase-1 axis provided neuroprotection and resulted in a milder epileptic phenotype (Meng et al., 2014;Shen et al., 2018;Vezzani et al., 2010). Differences in the specific animal models, the use of antagonists versus genetic models, different timepoints of treatment and other factors may account for these opposing results. We demonstrated that, while antagonists of P2X7 receptors facilitated the effect of anticonvulsants, either genetic ablation or pharmacological inhibition of components of the NLRP3-caspase-1 signalling cascade had either no effect or even increased seizure severity. This not only suggests that blocking P2X7 receptors offers a better therapeutic strategy to treat status epilepticus compared with blockers of the NLRP3 pathway but also implies that P2X7 receptor signalling that causes drug unresponsiveness is independent of NLRP3 activation. Activation of the P2X7 receptors has been shown to activate several other cellular pathways, including those involving ERK, Akt and NF-κB, that might contribute to inflammation and are activated following seizures or during epilepsy (Beamer et al., 2016;Vezzani et al., 2013). In addition, cathepsins, proteases which have been shown to contribute to IL-1β maturation, have also been linked to signalling by P2X7 receptors (Lopez-Castejon et al., 2010). Nevertheless, independent of the specific mechanism, our data strongly suggest that the drug resistance induced by P2X7 receptor activation was mediated by microglia.
One limitation of our study is that we have not tested if mice remain resistant to treatment if treated with a combination of anticonvulsants. Also, anticonvulsants have only been used at a single dose. This needs to be addressed in future studies. Another limitation of our study is that we have used only one animal model.
While the intra-amygdala KA model closely mimics the human condition, in terms of the neurodegeneration and lack of response to ASDs , our experiments should be replicated in status epilepticus models independent of KA. Finally, while P2X7 receptor-driven inflammation seems to be the most likely explanation for the effects of these receptors on responses to anticonvulsants, it is important to keep in mind that P2X7 receptors have been involved in numerous other pathological and also physiological mechanisms, including blood-brain barrier disruption, changes in neurotransmitter levels, synaptic reorganization, and neurogenesis (Sperlagh & Illes, 2014 (Thijs et al., 2019).