Ex vivo model of epilepsy in organotypic slices—a new tool for drug screening

Pregnant female Sprague-Dawley rats were acquired from Charles River (Barcelona, Spain). The Portuguese law and European Union guidelines (2010/63/EU) were respected in all procedures regarding the protection of animals for scientific purposes. All efforts were made to minimize animal suffering and to use the minimum number of animals. This study was approved by the “iMM’s Institutional Animal Welfare Body–ORBEA-iMM and the National competent authority–DGAV (Direção Geral de Alimentação e Veterinária).”

Organotypic slice cultures were prepared from 6- to 7-day-old Sprague-Dawley rats, according to the interface culture method [45]. After decapitation, brains were removed and placed in cold Gey’s balanced salt solution (GBSS, Biological Industries, Kibbutz Beit Haemek, Israel) with 25 mM d-glucose (Sigma, St. Louis, MO, USA), under sterile conditions. The hippocampus, together with the entorhinal cortex (EC) and perirhinal cortex (PC), (Fig. 1a), was dissected out and sliced transversely at 350 μm using a McIlwain tissue chopper. Five slices were placed onto porous (0.4 μm) insert membranes (PICM 03050, Millipore, Bedford, MA), which were transferred to six-well culture trays (Corning Costar, Corning, NY). Each well contained 1 ml of culture medium composed of 50% Opti-MEM, 25% Hanks’ balanced salt solution (HBSS), 25% heat-inactivated horse serum (HS) (all from Invitrogen, Paisley, UK), 25 mM d-glucose, penicillin (100 units/ml), and streptomycin (100 μg/ml) (Sigma). Slices were maintained at 37 °C with 5% CO₂ and 95% atmospheric air for the following 2 weeks.

Slices were randomly divided in two groups, which undertook different culture conditions (Fig. 1). Those designated control slices (CTL, Fig. 1b) were kept in a serum-based (25% HS) Opti-MEM medium, with medium renewal twice a week, and did not develop epileptiform activity. Those slices which spontaneously develop epileptiform activity (EL, Fig. 1c) were changed at 3 days in vitro (DIV) to a chemically defined serum-free based medium, Neurobasal A (Invitrogen), supplemented with 2% B27 (Invitrogen), l-glutamine (1 mM) (Invitrogen), penicillin (100 units/ml), and streptomycin (100 μg/ml), and decreasing HS concentrations (15%, 10% and 5%), until a serum-free condition was reached at 9 DIV. In EL slices, medium was renewed every second day. Our protocol is different from other reports in the sense that slices are gradually deprived of serum.

Field potential recordings were performed in an interface-type chamber as previously described by others [20]. At 14 DIV, slices were transferred to the interface recording chamber, with a humidified (5% CO₂/95% O₂) atmosphere at 36 °C, and with their growth medium continuously superfused and recirculating at a rate of 2 ml/min. Each slice was visually inspected, ensuring slice integrity and organization. After an equilibration period of 20 min, spontaneous field potential recordings were performed with an extracellular microelectrode (4 M NaCl, 2–4 MΩ resistance) positioned in the CA3 pyramidal cell layer over 30–40 min. Recordings were obtained with Axoclamp 2B amplifier, digitized (Axon Instruments, Foster City, CA), and analyzed by the Clampex software version 10.2 (Molecular Devices, Sunnyvale, CA, USA). All recordings were band-pass filtered (eight-pole Bessel filter at 60 Hz and Gaussian filter at 600 Hz).

Slices were transferred to an Axioskop 2FS upright microscope (Zeiss, Jena, Germany) equipped with a differential interference contrast-infrared (DIC-IR) CCD video camera (VX44, Till Photonics, Gräfelfing, Germany) and fixed with a grid in a recording chamber continuously superfused by a gravitational superfusion system at 2–3 ml/min with artificial cerebrospinal fluid (aCSF) containing (in mM) NaCl 124, KCl 3, NaH₂PO₄ 1.25, NaHCO₃ 26, MgCl₂ 1, CaCl₂ 2, Glucose 10, pH 7.4 (gassed with 95% O₂, 5% CO₂), at room temperature. Patch pipettes (4–9 MΩ) were pulled from borosilicate glass capillaries (1.5 mm outer diameter, 0.86 mm inner diameter, Harvard Apparatus, Holliston, MA, USA) with PC-10 Puller (Narishige Group, London, UK) and were filled with an internal solution containing (in mM) K-gluconate 125, KCl 11, CaCl₂ 0.1, MgCl₂ 2, EGTA 1, HEPES 10, MgATP 2, NaGTP 0.3 and phosphocreatine 10, pH 7.3 adjusted with KOH (1 M), 280–290 mOsm. Recordings were performed in current-clamp mode using an EPC-7 electrical amplifier (List Biologic). Signals were low-pass filtered using a 3- and 10-kHz three-pole Bessel filter of an EPC-7 amplifier, digitized at 10 kHz using a Digidata 1322A board, and registered by the Clampex software version 10.2 (Molecular Devices). Resting membrane potential (RMP) was measured immediately after establishing whole-cell configuration, and action potential firing was systematically evoked in current-clamp mode by injecting current pulses of − 50 to + 300 pA, in 10 or 50 pA increments for 1000 ms from an initial holding potential of − 70 mV. The threshold for action potential (AP) generation was determined as the difference between the resting membrane potential and the membrane potential at which phase plot slope reached 10 mV/ms [46].

Samples (35 μg total protein/well) and protein molecular weight marker (NZYColour Protein Marker II, NZYTech, Lisbon, Portugal) were boiled at 95 °C for 10 min, electrophoresed on a 12% SDS-PAGE, and electrotransferred to PVDF membranes (Merck-Millipore, Feltham, UK). Following blocking, membranes were probed with the primary antibodies and with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (1:10000, Santa Cruz Biotechnology, Heidelberg, Germany) as previously described [47]. Immunoreactions were visualized with ECL Western Blotting Detection System (GE Healthcare, Buckinghamshire, UK). The integrated intensity of each band was calculated using computer-assisted densitometry analysis with ImageJ software 1.44b. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the loading control.

Images were prepared for printing in Image Lab software 5.2.1 (software available in ChemiDoc XRS+ system, Bio-Rad). For each protein evaluated, the chemiluminescence image was merged with the colorimetric image of the molecular weight marker.

The primary antibodies used were mouse monoclonal antibody anti-αII-spectrin (1:500) and rabbit polyclonal antibody anti-Caspase-3 (1:1000) from Santa Cruz Biotechnology; rabbit polyclonal antibody anti-GFAP (1:5000) from Sigma; and goat polyclonal antibody anti-Iba1 (1:1000), rabbit polyclonal antibody anti-NLRP3 (1:300), and mouse monoclonal antibody anti-GAPDH (1:5000), all from Abcam, Cambridge, UK.

Hippocampi were isolated from ten slices, and RNA was extracted according to QIAGEN RNeasy Mini Kit (QIAGEN, Hilden, Germany). The collected tissue was dissociated with a 25G needle in the presence of QIAzol lysis reagent. RNA concentration was determined using Nanodrop 1000 (ND-1000 Spectrophotometer, Thermo Fisher Scientific, Waltham, MA, USA).

In vitro reverse transcription was performed from 2 μg of total RNA (in 20 μl) and carried out with SuperScript II Reverse Transcriptase (EC 2.7.7.49, Invitrogen, Carlsband, CA, USA) in a thermoclycler (MyCycle, Bio-Rad), according to the manufacturer’s recommendations (SuperScript First Strand Synthesis Systems for RT-PCR, Invitrogen). cDNA amplification was carried out in a Rotor-Gene 6000 real-time rotary analyzer thermocycler (Corbett Life Science, Hilden, Germany) as former described [47]. Briefly, the reactions took place in the presence of SYBR Green Master Mix (Applied Biosystems, Foster City, CA, USA) and 0.2 μM of each gene-specific primers. RT-qPCR parameters included an initial denaturation step for 2 min at 94 °C and 50 cycles with 30 s at 94 °C, 90 s at 60 °C, and 60 s at 72 °C. Reaction specificity was evaluated in all assays by a melting curve (Additional file 1: Figure S1).

The threshold cycle (Ct) and the melting curves required for the relative quantification [48] were acquired with Rotor-Gene 6000 Software 1.7 (Corbett Life Science). GAPDH was used as the reference internal standard.

The primers used for the proinflammatory cytokines (see Table 1) were designed using the OligoAnalyzer 3.1, provided by Integrated DNA Technologies (Coralville, IA, USA) and acquired to Invitrogen. All mRNA sequences from Rattus norvegicus were obtained from the GenBank sequence database of the National Centre for Biotechnology Information (http://​www.​ncbi.​nlm.​nih.​gov/​nucleotide/​). Accession numbers are indicated in Table 1.

Cell death was assessed by monitoring the cellular uptake of the fluorescent dye propidium iodide (PI, 3,8-diamino-5-(3-(diethylmethylamino)propyl)-6-phenyl phenanthridinium diiodide, Sigma). PI is a polar compound, which only enters cells with damaged cell membranes and interacts with DNA emitting red fluorescence (630 nm; absorbance 493 nm). It is not toxic to cells and was proven to be a feasible marker of neuronal cell death in organotypic slice cultures [49].

PI imaging was carried out at 14 DIV. CTL and EL slices were incubated with 2 μM sterile PI, diluted in culture medium for 4 h before imaging. Cellular uptake of PI was recorded by fluorescence microscopy on a wide field fluorescence microscope (Axiovert 200, Zeiss, Germany) using a rhodamine filter. Fluorescence photomicrographs were acquired, under identical lighting, with an EC Plan-NeoFluar 5x objective (Zeiss) with a numerical aperture of 0.16. For assessment of PI uptake, the regions of interest (DG, CA1 and CA3) were delineated and the fluorescence intensity was quantified using the software ImageJ 1.44b (NIH). The intensity value of each analyzed region was obtained by correction with a fluorescence background image. PI uptake by each hippocampal region was expressed in arbitrary units of fluorescence intensity.

Slices were fixed, at 14 DIV, for 1 h with 4% paraformaldehyde (PFA, Sigma) diluted in PBS at room temperature (RT), followed by an incubation in increasing concentrations of a sucrose (Sigma) solution (10 and 20% in PBS) at RT. Slices were kept in a 30% sucrose solution at 4 °C until further use.

Slices were cut out of the insert and put in slides. Each slice was surrounded with DAKO pen (Dako, Glostrup, Denmark) to protect staining areas from drying out and from mixing with each other. Following PBS washes, slices were incubated for 3 h at RT in blocking solution containing 10% HS, 10% bovine serum albumin (BSA, Sigma), and 1% Triton X-100 (Sigma) in PBS, which ensure simultaneous permeabilization and blocking of the tissue. Subsequently, slices were incubated with the primary antibodies for 24 h at 4 °C. Slices were rinsed with PBST (PBS containing 0.1% Tween-20), and the fluorophores coupled-secondary antibodies were applied to the slices for 4 h at RT. The nuclei were stained with Hoechst 33342 (20 μg/ml, Invitrogen) for 40 min at RT. The slices were mounted in Mowiol (Sigma).

For each marker, images were acquired under identical lighting between conditions, with a frame size of 1024 × 1024 pixels on an inverted confocal laser scanning microscope (Zeiss LSM 710, Zeiss) equipped with a Plan-Apochromat 20x objective (Zeiss) with a numerical aperture of 0.80. Hoechst 33342 fluorescence was detected with a 405-nm diode laser (30 mW nominal output). Alexa Fluor 488 fluorescence was detected using the 488-nm line of an Argon laser (25 mW nominal output), and Alexa Fluor 568 was detected using a 561-nm DPSS laser (15 mW nominal output). Images were prepared for printing using Illustrator software from Adobe Systems (San Jose, CA, USA).

Doublecortin (DCX) fluorescence intensity was quantified in the DG of CTL and EL slices, using the software ImageJ 1.44b. The final intensity value was obtained by correction with a fluorescence background image. Fluorescence intensity values are presented as a percentage of CTL slices.

The microglia cell body diameter in CTL and EL slices stained with Iba1 was measured, using the software ImageJ 1.44b, in 50–70 microglia cells per each hippocampal area. Whenever the cell body was oval, the cell body diameter was considered the largest distance passing through the center of the ellipse.

The primary antibodies used were mouse monoclonal antibody anti-GFAP (1:1000, Sigma), goat polyclonal antibody anti-Iba1 (1:1000, Abcam), and mouse anti-doublecortin (1:500, Santa Cruz Biotechnology). The secondary antibodies were donkey anti-mouse-Alexa Fluor 488, donkey anti-goat-Alexa Fluor 488, and donkey anti-mouse-Alexa Fluor 568 (1:500, Invitrogen).

Hippocampi were dissected from five slices and homogenized, as for western blot, in ice-cold RIPA buffer containing a mixture of protease inhibitors (Sigma). The lysates were centrifuged, and the supernatants, containing the cytosolic and membrane fraction, were collected to measure IL-1β, TNF-α, and IL-6 in the tissue. Total protein was quantified using the Bio-Rad DC Protein Assay Kit (Bio-Rad). IL-1β, TNF-α, and IL-6 protein levels were quantified by ELISA, according to the manufacturers’ suggested protocol (R&D Systems, Abingdon, UK), using selective antibodies. Absorbance was read at 405 nm. The detection limit was < 3 pg/ml.

All statistical analysis was performed using GraphPad Prism software (San Diego, CA, USA). Most comparisons between CTL and EL slices were made using an unpaired t test. Regarding PI uptake experiments and microglia cell body diameter measurements, statistical analysis was performed with one-way ANOVA, followed by Bonferroni’s comparison test. The number of independent cultures or cells (n) used in each assay is indicated in the legend of each figure. In all figures, data are presented as mean ± SEM. Values of p < 0.05 were considered to account for statistically significant differences.

As depicted in Fig. 1a, the organotypic slices used throughout this work contain the hippocampal complex, namely the dentate gyrus (DG) and the cornu ammonis areas (CA1 and CA3) of the hippocampus, as well as the entorhinal (EC) and perirhinal (PC) cortex. Indeed, clinical research and animal models of epilepsy have suggested that these cortical areas play an important role in seizure generation [50].

Spontaneous activity of the slices was assessed at 14 DIV. CTL and EL slices were transferred to the interface chamber, and field potentials were recorded (40 min) under a constant superfusion with growth medium. Figure 2 shows representative field potential recordings of spontaneous activity obtained in CA3 area of CTL (Fig. 2a–CTL) and EL (Fig. 2a–EL) organotypic slices at 14 DIV. CTL slices depict a spontaneous typical physiological activity, while EL slices exhibit spontaneous interictal-like discharges, which resemble in vivo epilepsy, in accordance with what was previously described by others [42].

To evaluate the resting membrane potential of neurons, CTL and EL slices were transferred to the chamber of an upright microscope and RMP values of CA3 pyramidal cells were measured immediately upon establishing whole-cell configuration. We found no statistically significant differences between RMP measured in neurons from CTL slices when compared with neurons from EL slices (CTL − 62.57 ± 1.59 mV and EL − 59.25 ± 1.89 mV, p > 0.05, t test) (Fig. 2b).

Action potentials were then evoked under current-clamp mode by injecting current pulses (− 50 to + 300 pA, in 10 or 50 pA increments for 1000 ms) from an initial holding potential of − 70 mV. The threshold for AP generation was measured as the beginning of the upward rise of the AP when dV/dT exceeded 10 mV/ms. The delta between the RMP of each neuron and the threshold potential (Vt), dRMP-Vt, at which they first fired an AP was not significantly different (p > 0.05, t test) between CTL (dRMP-Vt 21.63 ± 1.63 mV) and EL (dRMP-Vt 21.40 ± 1.81 mV) slices (Fig. 2c, d).

Over the last decade, changes in hippocampal neurogenesis have emerged as a hallmark of temporal lobe epilepsy (TLE) [14].

Doublecortin (DCX) is a microtubule-associated phosphoprotein, which is widely used as a marker for immature neurons derived from newly generated neural precursor cells and has been validated to assess changes in the level of neurogenesis [11].

Thus, an immunofluorescence protocol was performed to evaluate the DCX staining in the DG of CTL and EL slices at 14 DIV.

As can be observed in Additional file 2: Figure S2, DCX staining was increased in slices that display epileptiform activity. Indeed, DCX fluorescence intensity was found to be significantly higher (*p < 0.05, t test) in the DG of EL slices, thus suggesting increased neurogenesis.

Neuronal death is considered an important feature of epilepsy. To evaluate if neuronal death was increased in EL slices, we quantified αII-spectrin cleavage and caspase-3 activation by a western blot assay, carried out with protein extracts obtained from the hippocampal region of the slices. Calpain-mediated degradation of αII-spectrin results in the formation of two unique and highly stable SBDPs with a molecular weight of 145 and 150 kDa (SBDP145 and SBDP150), which can be further cleaved by caspase-3 yielding shorter fragments. The presence of the calpain-cleaved fragments is usually indicative of necrotic and excitotoxic neuronal death. On the other hand, caspase-3-mediated αII-spectrin cleavage leads to the formation of a fragment with 150 kDa (SBDP150), which is further degraded by caspase-3, producing the apoptotic-specific SBDP120, with 120 kDa [51].

Thus, an antibody which recognizes the full length αII-spectrin, as well as the calpain- and caspase-3-signature SBDPs, was used. As can be observed in Fig. 3a, this antibody identified two bands: a high molecular weight band, which corresponds to full length αII-spectrin (250 kDa), and a lower one for SBDPs. The expression of SBDP145 and SBDP150 was not possible to visualize separately (Fig. 3a). However, the apoptotic-specific SBDP120, which results from the caspase-3 cleavage of αII-spectrin, was absent from both conditions, indicating that the lower band recognized by αII-spectrin antibody corresponds exclusively to the calpain-signature SBDPs.

CTL and EL slices have a similar expression of full length αII-spectrin (Fig. 3c, p > 0.05, t test). The results also indicate that the ratio SBDP145/150 to αII-spectrin was significantly increased (Fig. 3d, *p < 0.05, t test) in EL slices, pointing to a higher αII-spectrin cleavage by calpain in these slices. Furthermore, the band for active caspase-3 (17 kDa) is absent from the immunoblot (Fig. 3b) and no differences are found in pro-caspase 3 expression (Fig. 3e, p > 0.05, t test). These results corroborate the sole detection of SBDP145/150 originated by calpain-mediated cleavage of αII-spectrin and point to an increased necrosis in EL slices.

To evaluate neuronal death across the different regions of the hippocampus, the PI uptake assay was carried out in CTL and EL slices at 14 DIV.

The representative fluorescence photomicrographs of both groups of slices (Fig. 4a) suggested that PI incorporation was increased in EL slices. This observation was corroborated by the quantification of PI uptake in each hippocampal region, expressed in arbitrary units of fluorescence intensity (Fig. 4b). PI uptake in granular and pyramidal neurons of EL slices was significantly higher (*p < 0.05, one-way ANOVA) than that detected in the same regions of CTL slices. Also, in both groups of slices, CA1 pyramidal neurons show a significantly higher PI uptake than granular neurons (^#p < 0.05, one-way ANOVA). Specifically, for DG, CTL 86.71 ± 8.016 vs EL 121.3 ± 10.51; for CA3, CTL 105.4 ± 6.920 vs EL 135.5 ± 9.846; and for CA1, CTL 130.1 ± 10.67 vs EL 169.8 ± 11.83.

Activation of astrocytes and microglia are considered a molecular hallmark of epilepsy, and recent findings point to a role of glial cells in the pathogenesis of this disorder [52, 53].

To investigate if gliosis was associated with epileptiform activity, two experimental approaches were used: a western blot assay to evaluate the overall expression of GFAP and Iba1 and an immunohistochemistry assay to evaluate morphological changes in astrocytes and microglia.

The western blot was carried out with 14 DIV hippocampal lysates obtained from CTL and EL slices. The immunoblots (Fig. 5a, c) showed an increased expression of GFAP and Iba1 in EL slices. Indeed, the densitometry analysis (Fig. 5b) attested a significant increase in GFAP expression in EL slices (*p < 0.05, t test). Likewise, the increased Iba1 expression in EL slices, evident in the immunoblot (Fig. 5c), was supported by the densitometry analysis (Fig. 5d), which indicated that Iba1 expression was significantly higher (***p < 0.001, t test) in EL slices, when compared to CTL ones.

An immunofluorescence analysis was also performed in CTL and EL slices to assess the morphological features of astrocytes (Fig. 6) and microglia (Fig. 7) in the main hippocampal areas. At 14 DIV, slices were fixed with PFA and handled as described before.

In the DG and CA regions of CTL slices (Fig. 6a–A1, A2, A3), most GFAP-positive astrocytes have thin processes and display preservation of individual domains, which can be observed in the magnified panels, Fig. 6a–A1a, A2a, and A3a. However, some hypertrophic cell bodies can also be observed suggestive of a mildly activated state. This basal astrogliosis in organotypic slices has long been reported by others [54]. However, in EL slices (Fig. 6b), a moderate/reactive astrogliosis can be noticed in all regions of the hippocampus, which include the presence of tick and highly superimposed astrocytic processes, resulting in the disruption of individual domains. These features are visible in the magnified panels, Fig. 6b–B1a, B2a, and B3a.

Figure 7 shows the occurrence of microglia cells, identified by Iba1 expression, in all hippocampal regions of CTL and EL slices. In CTL slices (Fig. 7a), most microglia have a ramified morphology, characteristic of a resting state, as can be noticed in Fig. 7a–A1a, A2a, and A3a. In EL slices (Fig. 7b), microglia are more abundant in all regions of the hippocampus, which agrees with the western blot results described above. Furthermore, most microglia have larger cell bodies and have changed their morphology from ramified to bushy-like cells with fewer processes, as observable in Fig. 7b–B1a, B2a, and B3a. In comparison with CTL slices, microglia in EL slices have a significantly increased cell body diameter in all regions of the hippocampus (***p < 0.001, one-way ANOVA), as attested by the quantification depicted in Fig. 7c. Within each condition, no statistical difference was found between hippocampal areas (p > 0.05, one-way ANOVA). The mean microglia cell body diameter for CTL slices (DG 11.6 μm ± 0.32; CA3 11.9 μm ± 0.30; CA1 11.5 μm ± 0.31) is higher than 7.5 μm reported by others [55] in resting microglia. This points to a mixed population of resting and reactive microglia in CTL slices, which is in accordance with some basal microglia activation known to occur in organotypic slices [54]. However, in EL slices, the mean microglia cell body diameter increases approximately twofold in relation to CTL ones (DG 18.9 μm ± 0.82; CA3 17.6 μm ± 0.64; CA1 19.8 μm ± 0.63), supporting the qualitative morphological analysis.

Thus, the described morphological alterations depicted by astrocytes and microglia corroborate an increased astrogliosis and microgliosis in EL slices.

Recently, inflammatory-related events have been intensely discussed in the context of epilepsy and were proposed as potential therapeutic targets for new antiepileptic drugs [56, 57].

After showing the occurrence of gliosis in EL slices, it was pertinent to evaluate alterations in the expression of proinflammatory cytokines, namely IL-1β, TNF-α, and IL-6, which are expressed at very low levels in the normal brain, but undergo a fast increase in expression after the induction of seizures [58, 59].

Thus, the expression of these proinflammatory cytokines was evaluated at the mRNA and protein level in CTL and EL slices at 14 DIV.

The transcript expression was evaluated by RT-qPCR, and the results are depicted in Fig. 8a. As can be observed, the mRNA expression of IL-1β, TNF-α, and IL-6 is significantly upregulated in EL slices (*p < 0.05, t test).

The protein concentration of these cytokines in the tissue, Fig. 8b, was calculated by an ELISA assay. At the time point evaluated, protein levels of IL-1β and IL-6 were found to be increased in EL slices (*p < 0.05, t test), but TNF-α change at the protein level did not reach statistical significance (p > 0.05, t test).

Recently, studies indicate a correlation between enhanced NLRP3 inflammasome expression and several neurological-related disorders [30]. NLRP3 activation leads to the processing of pro-IL-1β and pro-IL-18 by caspase-1, with the consequent increased production of the mature cytokines. Since IL-1β was upregulated in EL slices, it was relevant to evaluate if NLRP3 expression was altered in these slices.

To evaluate the differential expression of NLRP3 inflammasome between CTL and EL slices, a western blot was carried out with 14 DIV hippocampal lysates. The immunoblot (Fig. 9a) shows that NLRP3 expression was increased in EL slices. This result was confirmed by the densitometric analysis (Fig. 9b), which indicated a statistical significant increase (*p < 0.05, t test) in NLRP3 expression in EL slices.