A total of thirty-seven adult male Sprague-Dawley rats (Harlan, San Pietro al Natisone, Italy), ranging 260–270 g of body weight, and 66 4-week-old male CD-1 mice (Charles River, Calco, Italy) were used in this study. All animals were housed in a specific pathogen-free facility under controlled environment with ad libitum access to water and food.

EP-80317 was obtained through conventional solid phase synthesis, dissolved in a physiologic solution and administered through intraperitoneal (i.p.) injection (330 μg/kg). GW9662 and dimethyl sulfoxide (DMSO) were purchased from Sigma–Aldrich Chemicals (St. Louis, MO, United States). GW9662 was dissolved in DMSO and diluted with saline and then administered (2 mg/kg, i.p.) (Rani et al., 2016).

All experiments were in compliance with the European Directive 2010/63/EU and carried out according to the national guidelines on animal experimental research of the Italian Ministry of Health (DM 126/2011 – B and DM 92/2013). The University of Modena and Reggio Emilia Ethics Committee approved the study protocol. All efforts were made to refine procedures to improve the welfare and to reduce the number of animals that were used.

For the pilocarpine model, we subdivided rats into four groups for behavior and video electrocorticographic (ECoG) recordings: (i) 10 control rats received saline (1 ml/kg, i.p.), (ii) 10 rats received EP-80317 alone, (iii) seven rats received GW9662 alone, and (iv) 10 rats received GW9662 10 min prior to EP-80317.

For the repeated 6-Hz corneal stimulation model, we considered a number of mice (n = 39) for behavior and video-ECoG recordings, and the others (n = 27) for immunohistochemical and immunofluorescence analysis. In particular, for behavior and video-ECoG recordings, we subdivided mice into the following groups: (i) nine control mice received saline (1 ml/kg, i.p.), (ii) 12 mice received EP-80317 alone (330 μg/kg, i.p.), (iii) nine mice received GW9662 alone (2 mg/kg, i.p.), and (iv) nine mice received GW9662 10 min prior to EP-80317, both at the already indicated doses. For immunohistochemical analyses, four control mice were used to determine basal levels of the investigated marker and were neither treated nor stimulated. Still, they were handled and exposed to the same procedure as the others. Twelve were saline-treated mice, while the other 11 mice received an i.p. EP-80317 injection. Moreover, these animals were used for qualitative immunofluorescence experiments but, in this case, we used only three mice per treatment group.

Pilocarpine (Sigma–Aldrich, Milan, Italy) was injected i.p. (380 mg/kg) to induce SE. The pilocarpine injection was preceded by methylscopolamine (1 mg/kg, i.p.; Sigma–Aldrich) to prevent the peripheral effects of cholinergic stimulation (Curia et al., 2008). EP-80317, GW9662 alone, or saline were injected i.p. 20 min after methylscopolamine and 10 min before pilocarpine. In the group of rats treated with GW9662 and then with EP-80317, the first drug was administered 10 min prior to the second one. In all rats experiencing SE, diazepam (20 mg/kg, i.p.; Hospira Italia, Naples, Italy) was injected 10 min after the SE onset to guarantee survival. This procedure stops convulsive seizures, leaving non-convulsive SE unaltered for hours (Gualtieri et al., 2012).

Drug-induced responses in rats were observed directly and graded by blind to treatment expert raters. Seizures were graded according to a modification of the Racine’s scale (Racine, 1972). In particular, we considered: (i) non-convulsive seizures, ranked as stage 1 to stage 3; (ii) convulsive seizures, ranked as stage 4 to stage 5; (iii) SE (stage 6), considered as the stage in which rats either did not recover normal behavior (i.e., exploration, grooming, or motor reaction to stimuli) between one seizure and the other, or in which they displayed continuous shaking for more than 5 min (Lowenstein, 1999). Rats were sacrificed 72 h after SE to assess the absence of cerebral lesions caused by electrode implant, and the presence of injuries developed during the SE.

Mice were stimulated once and allowed to recover for 72 h before being stimulated up to four sessions. Corneal stimulation was performed as previously described (Giordano et al., 2015, 2016). Briefly, a topical eye anesthetic (0.4% oxybuprocaine hydrochloride eye drops, Novesin, Novartis, Switzerland) was applied 10 min before stimulation. All mice received the injection of EP-80317, GW9662, or saline 10 min before each session of corneal stimulation. When GW9662 preceded EP-80317, it was administered 20 min before each session. Stimulation (fixed current intensity of 32 mA, pulse width of 0.2 ms, duration of 3 s, frequency of 6 Hz) was delivered via corneal electrodes connected to a stimulator (ECT Unit 5780; Ugo Basile, Comerio, Italy).

Seizure severity was ranked according to the following score: (i) stunned posture and eye blinking; (ii) head nodding, Straub tail and repetitive rhythmic movements (stereotypies) such as chewing; (iii) unilateral or alternating forelimb clonus; (iv) generalized tonic-clonic convulsions without loss of posture and rearings; (v) generalized tonic-clonic convulsions with loss of posture. Seizure scores were first recorded through direct observation, then reanalyzed on video recordings to quantify the duration of behavioral changes by an investigator unaware of the stimulation session. Recovery from seizures was defined as the reappearance of a normal exploratory behavior.

For electrode implantation, anesthesia was induced with volatile isoflurane (4% induction and 1–2% maintenance) in rats, whereas mice were anesthetized with ketamine (150 μg/g, i.p.) + xylazine (10 μg/g, i.p.). After deep anesthesia was reached (assessed by deep breath, loss of tail and eye reflexes), the skin was shaved, disinfected with povidone-iodine 10% (Betadine^® skin solution; Meda Pharma, Milano, Italy), cut and opened to expose the skull. Guiding holes were drilled and epidural electrodes (stainless steel Ø = 1 mm; PlasticsOne, Roanoke, VA, United States) were implanted in frontal (bregma 0 mm, 3.5 mm lateral from midline in rats and bregma 0 mm, 3 mm lateral from midline in mice) and occipital cortices (bregma -6.5 mm, 3.5 mm lateral from midline in rats and bregma -3.5 mm, 3 mm lateral from midline in mice) of both hemispheres. One electrode was implanted below the lambda on the midline in all animals and used as a reference. Electrodes were connected through steel wire to terminal gold pins (Bilaney Consultant GmbH, Düsseldorf, Germany) inserted in a plastic pedestal (PlasticsOne) cemented on heads. At the end of the surgery, gel containing 2.5 g lidocaine chloride, 0.5 g neomycin sulfate and 0.025 g fluocinolone acetonide (Neuflan^® gel; Molteni Farmaceutici, Scandicci, FI, Italy) was applied to reduce acute pain and risk of infection. All animals were monitored until complete recovery from anesthesia, and were housed in single cages without grids or environmental enrichments to reduce risk of headset loss.

For brain activity recording, animals were placed in cages with paper filter cover that allowed cable connection between headset and preamplifiers. Electrical brain activity was digitally filtered (0.3 Hz high-pass, 500 Hz low-pass), acquired at 1 kHz per channel, and stored on a personal computer after the mathematical subtraction of traces of recording electrodes from trace of reference electrode, using a PowerLab8/30 amplifier connected to 4 BioAmp preamplifiers (ADInstruments; Dunedin, Otago, New Zealand). Videos were digitally captured through a camera connected to the computer and synchronized to the ECoG traces through LabChart 7 PRO internal trigger. Four days after electrode implantation, all animals were connected to the recording system and received the sequence of treatments described above. To facilitate handling and pharmacological manipulations, recordings were stopped, animals were temporarily disconnected while being injected and reconnected soon after. In particular, rats were recorded for 20 min after methylscopolamine administration, 10 min after EP-80317, GW9662 or saline injection, 10 min after SE onset and at least 2 h after diazepam administration. Rats that did not experience SE were recorded for at least 90 min after pilocarpine injection, until signs of normal behavior (exploring, grooming) reappeared. Mice were recorded after EP-80317, GW9662 or saline injection and 6-Hz corneal stimulation, until signs of normal behavior reappeared.

Electrocorticographic traces were digitally filtered offline (band-pass: high 50 Hz, low 1 Hz) and manually analyzed using LabChart 7 PRO software (AD Instruments) by expert raters.

In rats, we measured the duration of each electrographic seizure, characterized as epileptiform ECoG patterns with trains of 150–250 ms long spikes with amplitudes at least twice as the previous 2 s baseline. Seizures occurring within 5 s of each other were defined as one epileptic event.

In mice, we quantified the ECoG signal of each seizure performing a power spectral analysis, that described the distribution of signal power over frequency (Dressler, 2004). In particular, the power spectrum was obtained by fast Fourier transformation of the ECoG waveforms (LabChart 7 PRO, ADInstruments) and used to obtain the mean power spectra for the experimental groups. The maximum value of power spectra (peak value) was considered for each mouse. Peak values of power, recorded from frontal and occipital electrodes, were evaluated to investigate whether modifications of ictal activity had occurred from the first to the fourth stimulation.

Control mice and mice receiving up to one or three sessions of 6-Hz corneal stimulation and treated with saline or EP-80317 were used for tissue analysis. Mice deeply anesthetized with isoflurane were transcardially perfused with phosphate buffered saline (PBS, pH 7.4) followed by Zamboni’s fixative (pH 6.9), 14–17 h after the seizure. Brains were post-fixed at 4°C in the same fixative for 24 h, cryoprotected in 15 and 30% sucrose solutions (Vinet et al., 2016) and stored at -80°C until used. Horizontal sections of 50 μm were cut using a freezing sliding microtome (Leica SM2000 R; Leica, Nussloch, Germany). For immunostaining, sections were washed in Tris-buffered saline (TBS) and incubated in 3% H₂O₂ in TBS (30 min) to quench endogenous peroxidase activity. Following another washing step, sections were blocked 1 h in TBS containing 2% bovine serum albumin, 0.3% Triton X-100 (Tx) and 5% normal goat serum. Sections were then placed at 4°C with a polyclonal rabbit anti-PPARγ (Ab209350; Abcam, Cambridge, United Kingdom, dilution 1:2000, 48 h). After washing, sections were incubated for 1 h with a biotinylated anti-rabbit secondary antibody (Vector Laboratories, Burlingame, CA, United States; 1:200), and later with the avidin-biotin-peroxidase complex (Elite ABC Kit; Vector Laboratories, Burlingame, CA, United States). The immunostaining was performed in 0.05% 3,3-diaminobenzidine tetrahydrochloride for 5 min (DAB, Sigma–Aldrich, Milan, Italy) and developed by adding 0.03% H₂O₂. Finally, sections were washed again in TBS, mounted on gelatin-coated slides and cover slipped with Eukitt (Eukitt^®, O. Kindler GmbH & Co., Freiburg, Germany).

Immunostained sections related to bregma from -8.04 mm to -5.04 mm for hippocampal CA1 and CA3 regions, and hilus of the dentate gyrus (DH) were analyzed using an Axioskop microscope (Carl Zeiss Vision GmbH, Munchen, Germany) equipped with a 10X objective. Images were digitally captured by a Sony CCD-IRIS B–W video camera, along the ventrodorsal direction of the brain (approximately 6–7 serial horizontal sections separated by 0.5 mm). A mouse brain atlas-C57BL/6J horizontal was used to assess brain sections. The image analysis was carried out using the KS300 software (Carl Zeiss Vision GmbH), as previously described (Biagini et al., 2005; Curia et al., 2013; Giordano et al., 2015, 2016). PPARγ immunoreactivity was measured as field area values, corresponding to the addition of areas of the specific profiles obtained after discrimination from background staining. Background values in each section were obtained from areas devoid of specific immunostaining, such as the angular bundle. All measurements were taken bilaterally and the final values represent the left–right average.

We performed a PPARγ/67-kDa glutamate decarboxylase (GAD67) qualitative double immunofluorescence to assess PPARγ expression in interneurons. PPARγ expression was further evaluated in different interneuron subsets by PPARγ/parvalbumin (PV), PPARγ/somatostatin-28 (SOM) or PPARγ/vasoactive intestinal peptide (VIP) qualitative double immunofluorescences. For PPARγ/GAD67 immunofluorescence, sections were washed in TBS at room temperature and unmasking in sodium citrate at 98°C and permeabilized for 1 h in TBS/0.02% Triton X-100 containing 5% normal goat serum. Concerning PPARγ/PV, PPARγ/SOM and PPARγ/VIP immunofluorescences, sections were washed in TBS at room temperature and permeabilized for 1 h in TBS/0.1% Triton X-100 containing 5% normal goat serum. Then, sections were incubated for 48 h with primary antibodies: rabbit anti-PPARγ (Ab209350; Abcam, Cambridge, United Kingdom, dilution 1:50), mouse anti-GAD67 (MAB5406; Chemicon International, Billerica, MA, United States, dilution 1:200), mouse anti-PPARγ (E-8, sc-7273; Santa Cruz Biotechnology, Santa Cruz, CA, United States, dilution 1/50), rabbit anti-PV (Ab11427; Abcam, Cambridge, MA, United States, dilution 1:2000), rabbit anti-SOM (20089; Immunostar, Hudson, WI, United States, dilution 1/1000) and rabbit anti-VIP (20077; Immunostar, Hudson, WI, United States, dilution 1/500). Following a washing step in TBS, sections were incubated for 3 h at room temperature with secondary antibodies Alexa Fluor 488-labeled goat anti-mouse antibody (A-11001; Invitrogen, Carlsbad, CA, United States, dilution 1:500) and Alexa Fluor 594-labeled goat anti-rabbit (A-11012; Invitrogen, Carlsbad, CA, United States, dilution 1:500). After rinsing sections in TBS, brain sections were placed on gelatinized glass slides, dried, and mounted with Mowiol after incubation with DAPI. Images were acquired using a Leica TCS SP2 (Leica Laser Technik, Heidelberg, Germany) confocal microscope. All images were taken using a 40X magnification.

Chi-Square test (χ²) was used to compare the percentage (%) of rats experiencing SE in the various treatment groups. Pairwise comparisons with Fisher’s exact test (using α correction) were used to establish differences among groups. Additionally, the time interval (min) required to develop seizures and SE in the various groups was analyzed by Kruskal–Wallis test. Dunn’s test was used for multiple comparisons. One-way analysis of variance (ANOVA) was used to compare electrographic seizure duration (sec) in each treatment groups.

The χ² test was also used to compare mice experiencing seizures with loss of posture in the various treatment groups. Again, pairwise comparisons with Fisher’s exact test (using α correction) were used to establish differences among groups. Electrocorticographic recordings, analyzed as peak values of power spectra, were compared using two-way ANOVA, considering treatments as the between-factor and sessions as the within-factor. The Holm-Šídák test was used for multiple comparisons. All statistical analyses were performed using Sigmaplot 11 (Systat Software, San Jose, CA, United States). Unless otherwise indicated, results are shown as mean ± standard error of the mean (SEM); p-values lower than 0.05 were considered as statistically significant.

Response to pilocarpine was assessed by expert raters who directly annotated the motor responses observed during seizure development according to the modified Racine’s scale (Racine, 1972; Lucchi et al., 2013), and by offline analysis of video-EcoG recordings. Four (n = 1 saline group, n = 2 EP-80317 group, and n = 1 GW9662 group) animals died during continuous tonic-clonic seizures and were excluded from the analysis. Consistently with previous findings (Biagini et al., 2011), pilocarpine induced SE in 100% of saline-treated rats. Instead, only 50% of rats treated with EP-80317 developed SE (p < 0.05 vs. saline-treated rats, Fisher’s exact test). Additionally, we also found that the percentages of rats developing SE after pilocarpine administration in groups treated with GW9662 or GW9662+EP-80317 (respectively, 83% and 80%) were not significantly different from that observed in saline-treated rats (Table 1).

We calculated the time interval required to develop the first stage 1-3 and stage 4-5 seizure after pilocarpine administration, as median and interquartile range (IRQ) values. Moreover, we calculated the time interval required to develop SE (Table 2). No significant differences were found when comparing the latency periods of the first stage 1-3 seizure in the various treatment groups. Instead, the latency periods for developing the first stage 4-5 seizure were significantly different in GW9662-treated or GW9662+EP-80317-treated rats compared to, respectively, saline-treated rats (p < 0.05 for both groups, Dunn’s test) and the EP-80317 group of treatment (p < 0.05). In addition, the latency periods for developing SE after the first stage 4-5 seizure were significantly shorter in GW9662-treated and GW9662+EP-80317-treated rats compared to saline-treated rats (p < 0.05).

We measured the average seizure duration in traces obtained from ECoG recordings (Figure 1) performed in the various groups of treatment. No differences were found for animals treated, respectively, with saline (65.32 ± 8.75 s) or GW9662 (67.15 ± 10.95 s). Although the EP-80317 group presented a lower seizure duration (45.75 ± 9.24 s), this difference did not reach a statistically significant level. Mean seizure duration was also lower (49.91 ± 12.64 s) in GW9662+EP-80317 but, again, not enough to be statistically relevant.

Experimenters first visually monitored seizures induced through 6-Hz corneal stimulation and their pharmacological modulations. The first seizure duration was significantly shorter in mice treated with EP-80317 compared to control mice (p < 0.05, Holm-Šídák test), thus confirming the anticonvulsant properties of this molecule (Biagini et al., 2011; Giordano et al., 2016). No differences were noticed by comparing controls to mice treated with GW9662 alone or prior to EP-80317 (Figure 2A). Similarly seizure severity, defined as the percentage of mice displaying loss of posture during seizures, was also less pronounced in the presence of EP-80317 (p < 0.05, EP-80317 vs. saline in the first session of stimulation; Fisher’s exact test). No differences were observed by comparing the other groups with saline-treated mice (Figure 2B).

As expected (Giordano et al., 2015), seizures significantly shortened by repeating 6-Hz corneal stimulation (p < 0.01, sessions three and four vs. session one in saline-treated mice; Holm-Šídák test). Consequently, no differences between mice treated with saline and EP-80317 were observed from the second session onward (Figure 2A). Moreover, the severity of seizures progressed in all groups, including mice treated with EP-80317 (Figure 2B), which confirmed the appearance of resistance to the effects of EP-80317 (Giordano et al., 2016). Consistently, in the fourth session of 6-Hz corneal stimulation no differences were present within the various treatment groups.

Seizures induced by 6-Hz corneal stimulation and their pharmacological modulations were further monitored through video ECoG recordings. In particular, we compared power spectra of seizures recorded from frontal (not shown) and occipital electrodes (Figure 3A). As previously reported (Giordano et al., 2015), power peaks did not change for ictal events recorded through frontal cortex electrodes, in all groups, whereas in the occipital recording we noticed the presence of an epileptogenic process. In particular, we found a significant increase in power peaks recorded from occipital electrodes in the fourth session, both in saline-treated and GW9662-treated mice (p < 0.01 for saline-treated mice; p < 0.05 for GW9662-treated mice, session four vs. session one; Holm-Šídák test; Figure 3B). Conversely, power peaks in mice treated with EP-80317 did not change from the first to the fourth session of 6-Hz corneal stimulation, suggesting an antiepileptogenic effect. This effect was not completely prevented by administering GW9662 prior to EP-80317, although a trend was present (p = 0.09). Interestingly, significantly lower values were found in the fourth session of mice treated with EP-80317 compared to saline-treated mice (p < 0.05; Figure 3B).

We evaluated the effects of repeated exposure to 6-Hz corneal stimulation on PPARγ immunoreactivity in the stratum pyramidalis of CA1 (Figure 4), in both stratum pyramidalis and stratum radiatum/lacunosum-moleculare of CA3 (Figure 5), and finally in the DH (Figure 6). First, we quantified the changes occurring after the first and third session of seizure induction in pyramidal cells of the hippocampus and in mossy cells of the DH. PPARγ levels were scantily detectable in all sampled regions of unstimulated control mice. PPARγ immunoreactivity was not significantly changed in regions of interest of saline-treated mice. Notably, in animals treated with EP-80317 we observed a prominent increase in PPARγ levels after the first 6-Hz corneal stimulation, reaching a statistically significant level in all the considered regions (p < 0.05 vs. controls, in CA1; p < 0.001 vs. controls and saline-treated mice, in CA3 and DH; Holm-Šídák test; Figures 4–6). After the third seizure induction, PPARγ immunoreactivity returned to basal levels in all hippocampal regions.

To evaluate the response in interneurons, we analyzed the changes occurring in the stratum radiatum/lacunosum-moleculare of the CA3 (Table 3). Again, PPARγ levels were barely detectable in unstimulated control mice and saline-treated mice, but a significant increase in PPARγ levels was found in mice treated with EP-80317 after the first 6-Hz corneal stimulation (p < 0.01 vs. controls; p < 0.05 vs. saline-treated mice, Holm-Šídák test; Table 3). Consistently, PPARγ levels returned to basal levels after the third seizure induction.

To confirm that PPARγ was expressed in interneurons, we performed a double immunofluorescence experiment with PPARγ and GAD67 antibodies. We found that not all interneurons coexpressed GAD67 and PPARγ immunoreactivity (Figure 7, top panels). Thus, to assess the subset of interneurons expressing PPARγ immunoreactivity, we evaluated PPARγ/PV, PPARγ/SOM and PPARγ/VIP colabeling. PPARγ-positive cells were frequently observed within PV interneurons (Figure 7, in which arrows indicate colabeling), and less frequently identified within SOM interneurons. In contrast, PPARγ never colocalized with VIP in interneurons (Figure 7, in which arrowheads indicate lack of colabeling).