Patients, neuropathology and genetic testing

Brain specimens from 37 patients operated for drug-resistant epilepsy (aged from 3 months to 16 years) at the Rothschild Foundation Hospital in Paris, France, between 2016 and 2020 were investigated for molecular and cellular studies. The cohort consisted of cases with FCDII or hemimegalencephaly (HME) (n = 24) and epilepsy surgical cases used as controls with a neuropathological diagnosis of FCDI (n = 5) or mild malformation of cortical development (mMCD; n = 8). Non-essential brain tissues for neuropathological diagnostic purposes were attributed to research by the neurosurgeon in agreement with the neuropathologist. All specimens were immediately frozen in liquid nitrogen except four FCDII brain specimens that were fixed immediately after surgery for electron microscopy. Post hoc genetic analysis on frozen bulk tissue was performed by targeted deep sequencing using a gene panel consisting of approximately 50 genes involved in mTOR signaling and focal cortical malformations, as previously published8. In addition, two FCDII samples (patient ID 8 and patient ID 9) were provided by the Sainte-Anne and Lariboisière Hospitals and were kept oxygenated for ex vivo MEA recordings after surgery. Neuropathological diagnosis was made according to the classification of the Diagnostic Methods Commission of the International League Against Epilepsy55,56.

MEA recordings on human cortical slices

Cortical specimens collected in the surgery room were immediately transported to the laboratory within 15 min in ice-cold (0–4 °C) oxygenated solution (O2/CO2 95/5%), containing (in mM): N-methyl-d-glucamine 93, KCl 2.5, NaH2PO4 1.2, NaHCO3 30, HEPES 20, D-glucose 20, ascorbic acid 5, sodium pyruvate 3, MgSO4 10 and CaCl2 0.5 (300–310 mOsm, pH 7.4). Transverse 400-μm-thick cortical slices were prepared in the same solution using a vibratome (HM650V, Microm). They were maintained at 37 °C in an interface chamber containing artificial cerebrospinal fluid (ACSF) composed of (in mM): D-glucose 10, KCl 3.5, NaHCO3 26, NaH2PO4 1.25, NaCl 126, CaCl2 1.6 and MgCl2 1.2 (290 mOsm), equilibrated with 5% CO2 in 95% O2. MEA recordings were performed using an MEA2100 station (MultiChannelSystems) equipped with a 120-microelectrode array chamber (10 × 12 layout, 30 µm TiN electrodes spaced 1,000 µm vertical and 1,500 µm horizontal). Slices were maintained in the recording chamber using a homemade platinum/nylon harp and perfused with pre-warmed (37 °C) oxygenated ACSF at a rate of 6 ml min−1. Slices were imaged using a video microscope table (MEA-VMT1, MultiChannelSystems) to register the location of electrodes with respect to the slice. Extracellular signals were acquired at a sampling rate of 10 kHz and filtered with a low-pass Bessel filter (order 2; 40 Hz) or high-pass Bessel filter (order 2; 100 Hz) using Multi Channel Experimenter (MultiChannelSystems, version 2.20). Analyses were performed offline using homemade software (MATLAB, version R2018b). The semi-automated IILD and MUA detection was performed according to a standard procedure previously reported57. In brief, for IILD detection, the signal was denoised, filtered in the 1–40-Hz range, squared and then normalized over the entire recording. IILD detection was then semi-automatic, using a user-defined threshold. For MUA, a similar procedure was used on signal high-pass filtered above 250 Hz. SigmaPlot 13.1 (SPSS) was used for statistical analysis and graph generation.

Immunofluorescence on human acute cortical slices

Immediately after MEA recordings, 400-µm-thick slices were fixed in 4% paraformaldehyde (PFA) for 12 h before transfer in PBS. Slices were blocked in PBS + 2% Triton X-100 + 2% BSA + 10% normal donkey serum for 48 h on an orbital shaker at room temperature. Slices were then incubated in PBS + 1% Triton X-100 + 1% normal donkey serum with appropriate primary antibody for 48 h on an orbital shaker at 4 °C. After 3 × 30-min rinsing in PBS, slices were incubated in PBS + 0.5% Triton X-100 with appropriate secondary antibody and 0.1 mg ml−1 DAPI for 24 h on an orbital shaker at 4 °C. Slices were then rinsed in PB 0.1 M for 24 h on an orbital shaker at room temperature. Finally, slices were subjected for 24 h on an orbital shaker at room temperature to tissue clarification in RapiClear (SUNJIN Lab, RapiClear 1.49), mounted on homemade microscope slide mounts and imaged using a Nikon A1R HD25 confocal microscope with a ×10 glycerol objective in resonant scan mode. z-stacks per 0.16-mm2 field of acquisition were acquired with a 10-µm step over the 400-µm thickness at the three color-coded dashed areas in the tissue. Between 10 and 170 DNs were counted per area.

Orientation of cortical slices for correlation between MEA and immunostainings

After MEA recordings, 400-µm-thick slices were individually placed in 12-well plates, fixed in 4% PFA and immunostained in the same well to preserve tissue integrity. Superimposition of images of slices taken during MEA recordings and after immunostaining was performed based on selective anatomical markers visually identified from the images. Finally, high-resolution confocal imaging was performed in the selected outlined area.

SaβGal colorimetric assay and immunohistochemistryHuman tissue

SaβGal colorimetric assay was performed following the manufacturer’s protocol (Cell Signaling Technology, 9860) with an incubation period of 12 h. All flash-frozen human brain slices (cases and controls) were processed together in the exact same experimental conditions, such as pH, temperature, humidity and incubation time—critical parameters that may influence SaβGal reactivity. Cortical sections were then fixed for 15 min in 4% PFA before washing in PB 0.1 M and immediate subsequent immunohistochemistry. For immunohistochemistry, samples were incubated in 1% H2O2 for 5 min, rinsed three times in PBS–Tween 20 0.05% and blocked in PBS–Tween 20 0.05% + 10% normal goat serum + 0.2% Triton X-100 for 30 min. Slides were then incubated in appropriate primary antibody diluted in blocking solution overnight at 4 °C. The next day, samples were rinsed three times in PBS–Tween 20 0.05% and incubated in appropriate HRP-coupled secondary antibody diluted in blocking solution for 1 h at room temperature. Slides were then rinsed three times in PBS–Tween 0.05% before incubation in amplification ABC vector kit (Vector Laboratories, PK6100). Samples were then rinsed three times in PBS–Tween 0.05% and dipped in ddH2O, and revelation was performed using a DAB kit (Vector Laboratories, SK4100) before standard counterstaining with hematoxylin and eosin. Samples were finally dehydrated in ethanol solutions before mounting in xylene and automatic scanning using a NanoXoomer slide scanner (Hamamatsu). Images were visualized using NDP.view 2 software (Hamamatsu). The following antibodies were used: p53 (1:300, DAKO, M7001, mouse), p16 (1:200, Abcam, 108349, rabbit), pS6 S240/244 (1:1,000, Cell Signaling Technology, 5364, rabbit), SMI311R (1:400, BioLegend, 837801, mouse), VIM (1:400, DAKO, M0725, mouse), p21 (1:200, Abcam, ab188224, rabbit), Hmgb1 (1:100, Cell Signaling Technology, 6893, rabbit), LaminB1 (1:100, Cell Signaling Technology, 17416, rabbit), NeuN (1:500, Merck, MAB377, mouse), Olig2 (1:200, Abcam, ab109186, rabbit | 1:100, Merck, MABN50, mouse) and Gfap (1:300, LifeTechnologies, MA515086, mouse).

Mouse tissue

Mouse brain slices were processed as human brain slices only with a shorter incubation period of 6 h. Quantifications of SaβGal+ cells were performed semi-automatically on Fiji software (ImageJ2, version 2.9.0/1.53t). In brief, RGB images were used to set a detection threshold for blue pixels on WT brain tissues. The threshold was then applied to images of vehicle-treated or DQ-treated animals and individually saved to ROI Manager. Automatic counting of objects greater than 25 px2 was performed and used as a proxy of the number of SaβGal+ cells in the different considered areas. For immunostainings, mounted slices were permeabilized for 1 h in PBS + 0.2% Triton X-100 + 5% BSA and then incubated in PBS + 1% BSA with primary antibody against pS6 S240/244 (1:1,000, Cell Signaling Technology, 5364, rabbit) overnight at 4 °C. Slides were then rinsed in PBS and incubated in PBS with secondary antibody anti-rabbit Alexa Fluor 555 (Thermo Fisher Scientific, A27039, goat) for 1 h at room temperature, before washing and 0.1 mg ml−1 DAPI incubation for 10 s. Quantification of pS6+ cells was performed semi-automatically on Fiji software. Eight-bit images were used to set a user-based fluorescence detection threshold, converted into binary files and subjected to watershed. Automatic counting of objects greater than 25 px2 was performed to quantify the density of pS6+ cells (that is, the number of objects per field of view).

Quantifications

For SAβGal colorimetric assay and DAB co-stainings on human tissue, between n = 42 and n = 310 SAβGal+ cells were manually counted on a 1-cm2 region of interest (ROI) per sample on n = 5 FCDII samples. For SAβGal colorimetric assay and DAB co-stainings on mouse tissue, between n = 359 and n = 405 SAβGal+ cells were manually counted on two 0.5-mm2 somatosensory cortical column ROIs (one rostral and one caudal) per animal on n = 3 animals (n = 6 samples in total). For fluorescent stainings on human tissue, cellular density (pS6+NeuN+ DNs and pS6+NeuN− BCs) was semi-automatically quantified using the Fiji Analyze Particle module on 2.5-mm2 ROIs per sample. For fluorescent stainings on mouse tissue, cellular density and mean fluorescence intensity (pS6+ electroporated neurons) were semi-automatically quantified using the Fiji Analyze Particle module on 1-mm2 ROIs. For SAβGal colorimetric assay quantification, cellular density was semi-automatically quantified using the Fiji Analyze Particle module on 0.5-mm2 ROIs on n = 4 animals.

Electron microscopy on FCDII samples

Samples (n = 6) were immediately fixed in the operating room after surgical removal in 2% glutaraldehyde + 2% PFA + 2 mM CaCl2 in 0.1 M sodium cacodylate buffer, pH 7.4, for 1 h at room temperature. Tissues were post-fixed with 1% osmium tetroxide in water for 1 h at room temperature, rinsed three times with water and contrasted ‘en bloc’ for 1 h at room temperature with 2% aqueous uranyl acetate. Small pieces (1 mm3) of gray matter were dissected and progressively dehydrated in 50%, 70%, 80%, 90% and 100% ethanol solution (10 min each). Final dehydration was performed twice in 100% acetone for 20 min. Infiltration with an epoxy resin (EMbed 812) was performed in two steps: one night at 4 °C in a 1:1 mixture of Epon and acetone in an airtight container and twice for 1 h at room temperature in freshly prepared resin. Finally, samples were placed in molds with fresh resin. Polymerization was performed at 56 °C for 48 h in a dry oven. Blocks were cut with a Leica UC7 ultramicrotome. Semi-thin sections (0.5 μm thick) were stained with 1% toluidine blue in 1% borax, allowing identification of DNs and BCs. Ultra-thin sections (70 nm thick) were contrasted with Reynold’s lead citrate and observed with a Hitachi HT7700 electron microscope operating at 70 kV. Pictures were taken with an AMT41B camera.

Laser capture microdissection and ddPCR

Laser capture microdissection was performed in two FCDII/HME tissues with pathogenic variants in MTOR and PIK3CA (ID 3 and ID 13) using a Leica LMD7000 system on 20-µm frozen brain sections mounted on PEN-membrane slides after SAβGal colorimetric assay and immunohistochemistry against pS6. Pools of n = 200 double-positive SAβGal+pS6+ enlarged cells (soma diameter >25 µm) were microdissected and collected in AdhesiveCap 500 Opaque tubes (Zeiss) for DNA extraction. ddPCR was performed as previously described58 using specific probes to detect variants MTOR:p.T1977K and PIK3CA:p.R1047H.

Mouse models

Two mouse models were used in this study; both males and females were used unless otherwise specified; and all animals were used at adult age (from postnatal day (P) 28). All mice were kept and bred under controlled conditions with a 12-h/12-h light/dark cycle, 45–65% humidity, a temperature of 22 °C as well as food and water ad libitum. All efforts were made to minimize the suffering and number of animals used in this study.

Depdc5cKO strains (on a C57BL6/J background, Janvier Labs) were previously reported22. Depdc5flox/flox mice were generated by flanking exons 1–3 with loxP sites by genOway, a fee-for-service external company. Mice were crossed with Synapsin1-Cre mice (B6.Cg-Tg(Syn1-cre)671Jxm/J, no. 003966, The Jackson Laboratory) to obtain Depdc5flox/flox;Syn-Cre+/− animals named Depdc5cKO and Depdc5flox/flox;Syn-Cre−/− animals named Depdc5WT.

For the MtorS2215F model, MtorS2215F mice were generated by IUE at E14.5 in Swiss/CD1 embryos (Janvier Labs) as in ref. 59. DNA solution contained 0.5 mg ml−1 pCAG-EGFP and 2.5 mg ml−1 pCAGIG-mTOR (p.S2215F) (kindly provided by Alfonso Represaʼs team at INMED, Marseille), a plasmid encoding the recurrent MTOR: p.S2215F variant found in patients with FCDII. At birth, pups were selected based on GFP fluorescence in the head visualized under a microscope.

Brain lysate preparation

Animals were killed by beheading. Whole brains of Depdc5cKO mice (n = 5) and control littermates (n = 5) were dissected out of the skull and immediately placed in a 2-ml microcentrifuge tube dropped in liquid nitrogen. For MtorS2215F models, GFP+ cortical regions were scooped out of the brain, placed in a 2-ml microcentrifuge tube and dropped in liquid nitrogen. Lysates were prepared by transferring half of the brain samples in tubes containing FastPrep homogenizer beads with 200 µl of ice-cold lysis buffer (Cell Signaling Technology, 9803) complemented with anti-phosphatase and anti-protease. Homogenization was performed using a FastPrep homogenizer. Homogenate was transferred without beads to a new 1.5-ml microcentrifuge tube and centrifuged at 10,000g for 10 min at 4 °C, and supernatant was collected as final solution and stored at −80 °C.

Western blotting

Total protein concentrations were quantified using a BCA Protein Assay Kit on an automated plate reader (SpectraMax, Molecular Devices). Then, 50 µg of proteins per sample was separated on 4–12% Bis-Tris gel and transferred to a nitrocellulose membrane. After Ponceau coloration to visualize protein bands, membranes were cut according to the targeted protein molecular weights, blocked for 1 h in PBS–Tween 20 0.05% + 5% BSA before incubation in appropriate primary antibody diluted in blocking buffer overnight at 4 °C. Membranes were then incubated in appropriate HRP-coupled secondary antibody for 2 h. Membranes were then rinsed three times and incubated for 5 min in an ECL kit, and revelation was performed on autoradiographic films. Densitometry on Fiji was performed to quantify the expression of targeted proteins normalized to actin. The following primary antibodies were used: Depdc5 (1:250, Abcam, ab185565, rabbit); p53 (1:300, DAKO, M7001, mouse); p19 (1:2, CNIO, rat); actin (1:1,000, Merck, A2066, rabbit); pS6 S240/244 (1:2,000, Cell Signaling Technology, 5364, rabbit); and total ribosomal protein S6 (1:1,000, Cell Signaling Technology, 2317S, mouse). Secondary HRP antibodies from Cell Signaling Technology were used at 1:2,000 (anti-mouse, 7076; anti-rabbit, 7074; and anti-rat, 7077). Biological replicates (3–5 mice for each genotype or condition) were used.

Multiplex immunoassays

The assay was performed following the manufacturer’s protocol (Meso Scale Discovery (MSD), V-Plex Mouse Cytokine 29-Plex Kit). Brain lysates (250 µg) and neuronal conditioned medium (50 µl) were used per well. Every measure was done in duplicate. n = 3 brain lysates per genotype per age were used (with two technical replicates each time), and n = 6 conditioned medium per genotype were used. Plates were read on an MSD QuickPlex.

Mouse surgery and intracranial electrode implantation

MtorS2215F mice (n = 16, groups 1, 2 and 3) dedicated for video EEG experiments were selected based on visual confirmation of handling-induced behavioral seizures before intracranial electrode implantation. Mice aged 2 months were administered 0.1 mg kg−1 buprenorphine 30 min before being anesthetized with 2% isoflurane and placed in a stereotaxic frame. As previously described59, enamel-coated stainless steel electrodes were implanted on the left and right primary motor cortex (M1: AP, 2.2 mm; MD, 2.2 mm), on the right and left lateral parietal association cortex (LPta: AP, −1.8 mm; MD, −1.2 mm) and a common reference in the cerebellum. All coordinates were derived and adjusted from the Paxinos and Watson mice brain atlas60.

Video EEG recordings and analyses

MtorS2215F mice (n = 12) were placed under freely moving conditions and connected to an A.D.C. amplifier (BRAINBOX EEG-1166), part of an EEG video acquisition system (DeltaMed, Natus). EEG signals were acquired at 2,048 Hz and band-pass filtered between 0.5 Hz and 70 Hz. The video was synchronized to the electrophysiological signal and recorded at 25 frames per second. MtorS2215F animals were recorded from 3–6 months of age for 72 continuous hours weekly. Analysis of EEG recordings was manually performed by an experimenter blinded to the drug administration protocol. Fast Fourier transform (FFT) analyses were achieved using the Gabor function running in MATLAB (MathWorks, version R2015b) and used to semi-automatically identify ictal events that were then confirmed visually based on pattern criteria used for human patients with epilepsy. Morlet wavelets were computed in the 1–50-Hz frequency.

Senolytics administration protocol

The typical scheme of drug administration is oral gavage of vehicle (DMSO) or senolytic DQ agents for four consecutive days, followed by 2 days of intermission and five additional days as previously described32. Stock solutions were prepared in DMSO 1 day before protocol initiation, aliquoted accordingly and stored at −20 °C as follows: 100 mg ml−1 dasatinib (Merck, CDS023389) and 60 mg ml−1 quercetin (Merck, PHR1488). The final solution was prepared immediately before gavage at a concentration of 12 mg kg−1 dasatinib and 50 mg kg−1 quercetin (or equivalent volume of DMSO for preparation of vehicle) with 30% PEG300 and 5% Tween 20 in ddH2O. Only one animal died during the protocol between phases of vehicle and DQ administration in one experimental design, likely due to failure in proper gavage procedure (outlined in red on Fig. 5d).

Animal well-being monitoring

The mouse grimace scale, as outlined by the National Center for the Replacement Refinement and Reduction of Animals in Research61, was used for assessing pain in mice as a proxy of well-being after DQ gavage. In brief, a score of 0 (no presence), 1 (moderate presence) and 2 (obvious presence) is given for the observation of five typical facial expressions used as proxies: orbital tightening, nose bulge, cheek bulge, ear position and whisker change. A score of 0 was assigned to each of the five features for both the DQ-treated and the vehicle-treated group of mice, before, during and at the end of the treatment. In parallel, animal well-being was assessed by monitoring body weight every day before the oral gavage procedure.

PTZ administration and seizure severity assessment

We used the PTZ-induced experimental model of epilepsy62. PTZ (Merck, P6500) was dissolved in 0.9% saline at 20 mg ml−1. Swiss/CD1 mice (Janvier Labs) received oral administration of the vehicle or DQ for 9 days (as described above) and received, on the 9th day, a single dose of 40 mg ml−1 PTZ. Seizures had a rapid onset (average <60 s), spontaneously resolved, were never lethal and were graded by an observer blinded to the experimental condition based on the following modified Racine scale: score 1, sudden arrest; score 2, head twitches and myoclonic jerks; score 3, repetitive forelimb clonus; score 4, generalized tonic-clonic convulsions with loss of righting reflex; and score 5, lethal seizure.

Statistics

All scatter dot plots with bars are presented as mean ± s.e.m. All statistical analyses were performed in GraphPad Prism 8 (GraphPad Software; Prism 10 for macOS, version 10.1.1 (270)). All statistical analyses were two-tailed. Exact P values are provided when possible. Only P values lower than 0.05 were considered statistically significant.

Study approval

For human data, the study protocol received approval by the ethics committee of CPP Île-de-France II (no. ID-RCB/EUDRACT-2015-A00671-48) and the INSERM Ethics Review Committee (no. 22-879) by the INSERM Institutional Review Board (IRB00003888, IORG0003254 and FWA00005831). This study is registered on ClinicalTrials.gov (NCT02890641). Informed and written consent was obtained from all participants (or their parents on their behalf).

For mouse data, the study protocols received approval from the French Ministry of Research (APAFIS 26557, 37296 and 40207). All efforts were made to minimize the suffering and number of animals used in this study.

Statistics and reproducibility

No statistical methods were used to predetermine sample size. For experiments with quantitative measurements, sufficient numbers of samples were used to derive non-parametric statistical tests. No data were excluded from the analyses. Experimenters were blinded to allocation during experiments and outcome assessment regarding the genotype or the drug (DQ versus vehicle). For in vivo experiments, attribution to experimental groups was randomized. Colorimetric assays and immunostainings on human frozen tissue and mouse tissue were repeated at least three times. Western blots were repeated at least two times. MEA recordings and histology stainings in Fig. 1 could be performed only once on each slice (four slices in patient 8 and two slices in patient 9). Electron microscopy imaging was repeated at least three times (Extended Data Fig. 6). Statistical tests used are non-parametric two-tailed Mann–Whitney tests because no a priori exists regarding the directionality of variation, and no distribution to meet required assumption of normality is required.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.