Wistar male rats (2 months old, 260–300 g; from the Central Animal Laboratory of Ribeirao Preto Medical School) were housed at room temperature (25 ± 2°C), 12:12 light/dark cycle (lights on at 7 a.m.), and free access to food and water. All procedures were performed in agreement with the ethical principles of animal experimentation adopted by the Brazilian College of Animal Experimentation (COBEA) and approved by the Ethics Committee on Animal Experiments (CETEA) of Ribeirao Preto Medical School (protocol #051/2010).

Rats were anesthetized with ketamine hydrochloride 10% (0.5 ml/kg i.p. + 0.75 ml/kg i.m.; Agener, Brazil) and xylazine (0.25 ml/kg i.p. + 0.4 ml/kg i.m.; Dopaser/Calier, Spain) and body temperature was maintained constant with a heating pad (37 ± 0.5°C). The level of anesthesia was constantly checked by tail pinch reflex and, if necessary, reinforced with 10% of the initial dose. Following standard coordinates (AP = -5.6 mm, LL = ± 4.5 mm, H = -3.0 mm; Paxinos and Watson, 1997), stereotaxic surgery was performed to implant one cannula (12 mm 23G stainless steel needles; BD, Brazil) in the left dorsal hippocampus and two monopolar recording electrodes (13 mm tungsten wire coated with Teflon 60 μm diameter; AM system Inc., USA), one in the right and one in the left dorsal hippocampal hilus. A silver wire welded to a micro-screw (placed in the suture of the parietals and occipital bones) served as reference electrode. All electrodes and the cannula were fixed in the skull using dental cement. Animals were allowed to recover for 7 days before administration of pilocarpine, CBD, or vehicle.

Status epilepticus was induced by 1 μL pilocarpine hydrochloride i.h. (2.4 mg/μL diluted in 0.9 % saline; Sigma-Aldrich, USA), injected with a microsyringe (10 μL Hamilton syringe; Sigma, USA) connected to a polyethylene tube (30 cm, PE10) and gingival needle (curved 13 mm, 30G; Injex, Brazil), at 1 μL/min flow. SE was kept for 2 h, and seizure was aborted with sodium thiopental i.p. (30 mg/kg; Abbott, Brazil). CBD (99.9% pure; STI Pharmaceuticals, Brentwood, UK) was prepared 10 mg/kg diluted in 98% saline (0.15 M NaCl) and 2% Tween 80 (Sigma-Aldrich, UK) and injected i.p. in a volume of 1 mL/kg, 1 h before SE induction (groups SE+CBDp and SE+CBDt). Additionally, CBD was administered every 12 h up to euthanasia in the SE+CBDt group. Rats from all groups were euthanized at 1 or 3 days post SE.

Control groups were rats that received either one injection of vehicle i.p. (98% saline + 2% Tween 80; VH group), one injection of CBD (10 mg/kg i.p. in vehicle; CBD group), or one injection of the vehicle before SE induction (SE group). All SE animals were analyzed for SE onset latency and severity. Classification of seizure severity was according to the modified Racine’s scale: (1) mouth and facial movements, (2) head nodding, (3) forelimb clonus, (4) rearing, (5) rearing and falling (Racine, 1972).

The dose of CBD was chosen based on anticonvulsant effects and pharmacokinetic studies (Karler et al., 1979; Turkanis et al., 1979; Siemens et al., 1980; Consroe et al., 1982; Jones et al., 2009). Gas chromatography has used to evaluate the CBD plasma levels during the experiments and ensure CBD was present during our experiment. Blood samples were collected from rat’s tails 90 min after CBD injection for CBD, SE+CBDp, and SE+CBDt groups at day 0 (i.e., SE induction day) to assess acute CBD presence, and at days 1 and 3 for SE+CBDp and SE+CBDt. Gas chromatography evaluation followed previously established protocols for CBD detection (Karschner et al., 2010; Andrews and Paterson, 2012). CBD was detectable in the plasma at day 0 (Supplementary Figure S1). One and three days after administration, CBD plasma concentration was reduced or absent in SE+CBDp rats, whereas increasing levels of CBD were detected in the plasma of SE+CBDt rats (Supplementary Figure S1). SE+CBDt rats also had a higher variability in CBD plasma levels than SE+CBDp rats. Thus, CBD was detectable in almost all rats from the SE+CBD groups within the temporal window of our experiment.

Spontaneous LFPs were recorded 7 days after the animal surgery. Experiments were performed in a homemade wooden box placed inside a Faraday cage, which allowed free movement of the rats (Supplementary Figure S2). Analogic signal were amplified, filtered (×100, 0.5–500 Hz, P55 pre-amplifier, Grass Technologies), and digitized (sampling rate of 400 Hz) utilizing an analogic-digital converter (Powerlab/16s; ADInstruments, Inc., Australia), connected to a computer, and recorded on LabChart 7.2 software (ADInstruments, Australia). Concomitant video recording was performed with a digital camera (webcam) positioned inside the wooden box (video-EEG).

LFP recording was carried out during 20 min (baseline recording or after administration CBD) followed by and additional 2-h recording after SE induction. Oscillatory activities of SE groups (SE, SE+CBDp, and SE+CBDt) were monitored 1 h per day, at day one and day three after SE. Signal artifacts (i.e., events with amplitude 2.5 times superior to the standard deviation of the segment) were removed from LFP, and data was analyzed using custom-made MATLAB scripts (The Mathwork, Natick, MA, USA). Power spectra densities (PDs) were calculated using mtspectrumc scripts from Chronux package¹. Chronux parameters used were: fpass = [0 100], tapers = [3 5], err = [0 0.05], trialave = 1, pad = 0 (Lopes-Aguiar et al., 2013). In addition, relative power spectra (normalized to the PSD sum across frequencies) in delta (0.5–4 Hz) and theta (4–10 Hz) were calculated. The relative (normalized) power band estimates the spectral density at delta and theta in relation the entire spectrogram (Moshel et al., 2013). Spectral energy from 58 to 62 Hz was excluded to avoid noise contamination. Power spectra were compared among groups to evaluate the effect of CBD treatment on oscillatory activity.

Open field test was performed to evaluate the post ictal locomotion of the rats. An activity monitor apparatus (Insight, Brazil) was used to analyze total distance explored, the number of vertical activities (the number of times an animal rises in its hind limbs), time explored on the borders and center of the arena (Supplementary Figure S3). The testing protocol consisted of placing the animal in the center of the box (habituation) for 3 min and then test for 8 min, according to previously tested protocols (Wolf et al., 2016). After each test, open field was cleaned with acetic acid (0.1% v/v). All data were automatically collected via software (Insight, Brazil).

One or three days after SE, rats were anesthetized with sodium pentobarbital (80 mg/kg i.p.) and perfused intracardially with 100 mL phosphate buffered saline (PBS, 0.1 M pH 7.4), followed by 250 mL 4% (w/v) paraformaldehyde in PBS. Brains were then removed from skull, post-fixed by immersion in 4% paraformaldehyde-PBS for 4 h, dehydrated and embedded in paraffin. Eight-micrometer-thick coronal sections (-3.3 to -3.6 mm from bregma) were mounted on slides and submitted to Fluoro-Jade B (FJB) histochemistry and NeuN immunohistochemistry for evaluation of neurons in degeneration and neuronal density, respectively.

Fluoro-Jade B histochemistry followed published protocol (Schmued et al., 1997; Do Val-da Silva et al., 2016). Brains were deparaffinized, hydrated, and immersed in the following sequence: NaOH 1% (w/v) solution in 80% (v/v) ethanol (5 min); 70% (v/v) ethanol (2 min); distilled water (2 min); potassium permanganate 0.06% (w/v) (15 min); two baths of distilled water (1 min); 0.01% Fluoro-Jade B (Histo-Chem Inc., USA) in 0.01% acetic acid (30 min); and three baths of distilled water (1 min). Sections were dehydrated and mounted with Krystalon (Harleco/EMD Millipore, USA).

Adjacent coronal sections were submitted to immunohistochemistry for the detection of NeuN epitope with an anti-mouse antibody (code MAB377, Chemicon/EMD Millipore, USA), following previously published protocols (Peixoto-Santos et al., 2012; Kandratavicius et al., 2014b). Briefly, sections were submitted to endogenous peroxidase block, microwave antigenic retrieval with 50 mM Tris-HCl pH 9.6, and overnight NeuN antibody solution in skim milk blocking buffer at 1:1,000 dilution. Primary antibody detection was done with rabbit anti-mouse IgG (E0354, Dako, Denmark) in blocking buffer (1:200), avidin-biotin-peroxidase kit (Elite ABC kit, Vector, USA) amplification, and diaminobenzidine (DAB, Pierce/ThermoFisher, USA) as a chromogen. Sections were dehydrated and mounted with Krystalon (Harleco, USA).

Micrographs from hippocampal subfields were collected with AxioVision software and MR AxioCam 15MP coupled to an Axioscope A1 microscope (Zeiss, Germany). FJB positive (FJB+) cells were counted in 30 squares of 2,500 μm² in the contralateral and ipsilateral hilus, CA3, and CA1 subfields. NeuN positive cells were counted in 12 squares of 2,500 μm² in granule cell layer, 6 squares of 2,500 μm² in the hilus, and 10 squares of 2,500 μm2 in CA3 and CA1 subfields, and neuron density was estimated with Abercrombie correction, as described elsewhere (Do Val-da Silva et al., 2016).

Student’s t-test or ANOVA followed by the Sidak post hoc test were used for parametric variables, whereas non-parametric variables were evaluated with Mann–Whitney’s. For analysis over time between different groups, we used two-way ANOVA for repeated measures mixed model (RM ANOVA) followed by a Sidak post hoc test. In the results section, the mention of Sidak post hoc test was omitted, but this post hoc test was applied to all ANOVA tests and is mentioned in the figure captions. All statistical procedure was performed using SPSS 17. Statistical significance was considered at p < 0.05.

As expected, VH and CBD groups showed no behavioral change. A higher number of SE+CBD rats did not developed SE when compared to SE group (35.1% vs. 15.9%; χ², p = 0.025). The SE+CBD groups, when compared to SE group, had increased latency to SE (44 ± 3 min vs. 20 ± 2 min; t-test, p < 0.001; Figure 1A) and lower SE severity (Racine 3 vs. Racine 4; Mann–Whitney, p < 0.001; Figure 1B). Moreover, none of the SE+CBD rats presented Racine 5 seizures or died, whereas 36.1% of rats from SE group had Racine 5 seizures and one rat died during SE.

Locomotory activity was assessed to evaluate post-ictal changes (i.e., post ictal lethargy) in rats submitted to pilocarpine-induced SE. Compared to controls (VH and CBD) and SE+CBDp, SE rats had less vertical exploratory behavior 1 day after SE (ANOVA; p ≤ 0.02; Figures 2A–F), whereas at day 3 SE had lower vertical exploratory activity than VH group (ANOVA, p = 0.008; Figure 2A). There was no difference between groups regarding total exploratory behavior or time spent on the arena borders vs. arena center (data not shown). Therefore, CBD treatment modulated the behavioral features of pilocarpine-induced SE and normalized the post-SE locomotory activity in the SE+CBDp group shortly after SE.

The LFP recording of animals during SE showed increased time to contralateral epileptiform afterdischarges onset in SE+CBD groups compared to SE group (8.09 ± 2.42 min vs. 2.10 ± 0.52 min; t-test, p = 0.04; Figure 3A). There was no difference regarding afterdischarge onset ipsilateral to the pilocarpine microinjection (7.63 ± 2.52 min vs. 2.36 ± 0.46 min; t-test, p = 0.06). A qualitative evaluation of the raw LFP recording as well as the whole frequencies from 0.5–25 Hz shows differences in electrographic activity between the SE and the SE+CBD groups (Figure 3B). Amongst the frequencies in the 0.5–25 Hz interval, increased hippocampal delta (0.5–4 Hz) and theta (4–10 Hz) frequencies are hallmarks of ictal electrographic activity, as mentioned above. Thus, ictal powers of these low-frequency oscillations are relevant to understand the neuroprotective actions of CBD during seizures and the post-ictal electrographic changes in the pilocarpine-induced SE rat model. SE+CBD groups had lower relative power in theta frequency contralateral to pilocarpine microinjection, when compared to SE group, at 40–80 min (0.12 ± 0.013 vs. 0.22 ± 0.02; RM ANOVA, p = 0.01) and in the full block of time (0.11 ± 0.009 vs. 0.17 ± 0.01; RM ANOVA, p = 0.016; Figure 3C). Contralateral delta band oscillation changed over time (RM ANOVA, p < 0.001), with a trend toward difference between SE+CBD and SE groups at 0–20 min (0.36 ± 0.06 vs. 0.54 ± 0.04) but no difference in the full block of time (Figure 3D). There was no difference in relative powers in delta and theta frequencies in the hippocampus ipsilateral to pilocarpine injection, when comparing the SE+CBD and SE groups (RM ANOVA; data not shown).

The evaluation of spectral power 1 and 3 days after SE showed continuous drop in relative power of delta frequencies of the SE+CBD groups, with more prominent drop in the SE+CBDp group (compare Figure 3D full block of time with Figure 4A), whereas the relative power in theta frequency of SE group dropped over time while the theta frequencies of SE+CBDp and SE+CBDt groups increased (compare Figure 3C and Figure 4B). At day one, there was no difference between SE, SE+CBDp, and SE+CBDt regarding contralateral relative power in delta (Figure 4A) or theta frequencies (Figure 4B). At 3 days post-SE, SE+CBDp had lower contralateral relative power in delta band compared to SE group (0.34 ± 0.03 vs. 0.47 ± 0.02; RM ANOVA, p = 0.02; Figure 4A). There was no difference between SE+CBDp, SE+CBDt, and SE in the theta frequency on day three (Figure 4B). There was also no difference in the ipsilateral delta and theta frequencies between the groups at 1 or 3 days (RM ANOVA, data now shown). CBD and VH groups had no difference in baseline or in delta and theta frequencies (data not shown). In summary, CBD increased afterdischarge latency and prevented some of the theta and delta changes seen in the pilocarpine-induced SE.

Acutely after SE, neurons in the hippocampus often enter a process of degeneration that leads to neuron loss over time. In our study, neurodegeneration was measured by presence and number of Fluoro-Jade B positive (FJB+) cells. No FJB+ cells were seen in the control groups VH and CBD, 1 or 3 days after vehicle or CBD injection. SE group had a high number of FJB+ cells in the contralateral and ipsilateral hilus, CA3, and CA1 at 1 (Figure 5) or 3 days (Figure 6) post SE induction. CBD injection significantly reduced SE-induced neurodegeneration, as can be visually seen in the comparison of SE+CBDp (B,E) and SE+CBDt (C,F) with SE hilus (A,D) of Figures 5, 6. In agreement with the visual evaluation, SE+CBDp and SE+CBDt had significantly lower FJB+ cells in the hilus (ANOVA, p ≤ 0.001) and CA3 (ANOVA, p ≤ 0.03) of the ipsilateral and contralateral hippocampus, compared to SE group, 1 day after SE (Figures 5G,H). Three days after pilocarpine-induced SE (Figures 6G,H), SE+CBDp and SE+CBDt groups had lower FJB+ cells only in the contralateral hilus (ANOVA, p ≤ 0.014), compared to SE group. All SE groups (SE, SE+CBDp, and SE+CBDt) had FJB+ cells in ipsilateral or contralateral CA1 subfield at 1 and 3 days after SE, with no statistical difference between groups. Thus, CBD reduced the neurodegeneration common to the pilocarpine-induced SE rat model.

To evaluate neuron loss following SE, we analyzed neuron density, measured as NeuN positive cell density, at 1 and 3 days post-SE. One day after SE induction, SE group had lower neuron density than the controls (VH and CBD) in the contralateral (Figure 7A) and ipsilateral (Figure 7B) hilus, CA3, and CA1 (ANOVA, p < 0.001), as well as in the ipsilateral granule cell layer (ANOVA, p < 0.001). CBD injection prevented SE-induced neuron loss, as SE+CBDp, and SE+CBDt groups had higher neuron density than SE group in contralateral (Figure 7A) and ipsilateral (Figure 7B) hilus, CA3, and CA1 (ANOVA, p < 0.001), and in ipsilateral granule cell layer (ANOVA, p < 0.001; Figure 7B). Moreover, SE+CBDp and SE+CBDt groups had similar neuron density than the control groups in almost all hippocampal subfields evaluated. Only in contralateral CA3 and CA1, SE+CBDp group had higher neuron density than CBD and SE+CBDt (ANOVA, p < 0.006), and than VH, CBD, and SE+CBDt (ANOVA, p < 0.03; Figure 7A), respectively. There was no difference between groups in contralateral granule cell layer.

The 3 days post-SE histological evaluation showed lower neuron density in the SE group, compared to the controls (VH and CBD), in contralateral granule cell layer and in contralateral and ipsilateral, hilus, CA3, and CA1 (ANOVA, p < 0.01; Figures 8A,B, 9). SE had also lower neuron density than CBD in the ipsilateral granule cell layer (ANOVA, p = 0.042). SE+CBDp, and SE+CBDt groups had higher neuron density than SE group in the contralateral and ipsilateral hilus, CA3, and CA1 (ANOVA, p ≤ 0.04). In the contralateral hippocampus (Figure 8A), SE+CBDt had lower neuron density than CBD in the granule cell layer and CA1 (ANOVA, p ≤ 0.037), and than SE+CBDp in the granule cell layer (ANOVA, p = 0.048). In ipsilateral granule cell layer (Figure 8B), SE had lower neuron density than CBD and SE+CBDp (ANOVA, p ≤ 0.042) SE+CBDt group had lower neuron density than SE+CBDp (ANOVA, p < 0.04). In ipsilateral CA3 (Figure 8B), SE+CBDp also had higher neuron density than CBD and SE+CBDt (ANOVA, p ≤ 0.038). SE+CBDt had lower neuron density than VH, CBD, and SE+CBDp in ipsilateral CA1 (ANOVA, p ≤ 0.04; Figure 8B). Our findings indicated that CBD administration, both before (SE+CBDp) and before and after (SE+CBDt) pilocarpine-induced SE protected hippocampal sector from neuron loss up to 3 days after the SE.