Animals

Experiments were conducted on adult (2–3.5 months of age) wild-type C57BL/6J male mice (Janvier Labs). Mice were housed at the local animal facility, under 12 h light/12 h dark cycle with access to food and water ad libitum, and in individually ventilated cages. Three cohorts of animals were used: cohort 1, n = 6; cohort 2, n = 10; cohort 3, n = 15 (total n = 31). A maximum of 5 mice was kept in each cage, mice were tagged with an earmark number and randomly assigned to experimental groups. After viral injections, 3 mice died (total n = 28; PSAM n = 21; GFP-only/Ctrl n = 7), and during status epilepticus induction 10 mice died (total n = 18; PSAM n = 14; GFP-only/Ctrl n = 4), leaving a total of 18 mice video-EEG monitored. EEG signal had insufficient quality for analysis in 5 animals (PSAM n = 3; GFP-only/Ctrl n = 2), and 4 animals did not develop SRS nor ESs, leaving a total of 9 animals used for analysis (total n = 9; PSAM n = 7; GFP-only/Ctrl n = 2).

The experimental procedures performed were approved by the Malmö/Lund Animal Research Ethics Board, ethical permit number 2998/2020-m, and conducted in agreement with the Swedish Animal Welfare Agency regulations and the EU Directive 2010/63/EU for animal experiments.

Organotypic slices were prepared using Sprague-Dawley rats at Instituto de Medicina Molecular João Lobo Antunes (iMM). All experiments were conducted according to European Union Guidelines (2012/707/EU) and to the Portuguese legislative action (DL 113/2013) for the protection of animals used for scientific purposes. The methods described were approved by “iMM’s Institutional Animal Welfare Body (ORBEA-iMM, Lisboa, Portugal) and authorized by the Portuguese authority for Animal Welfare (Direção Geral de Alimentação e Veterinária—DGAV).

Virus production and injection

AAVs were produced as previously described [74]. In brief, AAV8-CaMKIIα-PSAM4-GlyR-IRES-eGFP (was a gift from Scott Sternson; Addgene plasmid # 119744; http://n2t.net/addgene:119744; RRID: Addgene_119744) and AAV8-CaMKIIα-GFP genomes were packaged separately into AAV8 via PEI transfection. HEK293T cells were double transfected with AAV8-CaMKIIα-PSAM4-GlyR-IRES-eGFP and pDP8.ape (PlasmidFactory GmbH & Co. KG; # PF478). AAVs were harvested 72 h post-transfection using polyethylene glycol 8000 (PEG8000) precipitation and chloroform extraction, followed by PBS exchange in concentration columns. Purified AAVs were titered using RT-qPCR with standard curves and primers specific for the ITRs. AAVs were stored in glass vials at 4 °C until use. Titers used were around 8 × 1013 gc/ml, and if needed were normalized using PBS.

Intrahippocampal injections were performed via stereotaxic surgery. In brief, mice were anesthetized with isoflurane/O2 mixture 5% and placed in the stereotaxic frame, thereafter anesthesia was maintained at 1.5%. After bupivacaine injection (<0.5 ml, Marcaine, AstraZeneca) a middle incision was made on top of the skull. Small holes were drilled at the injection sites and bilateral injections were targeted to the following coordinates (in mm, from Bregma): position 1: AP −2.2, ML ± 1.7, DV −1.9; position 2: AP −3.3, ML ± 3.0, DV −3.7 and −2.7. A total volume of 0.4 µl was injected at a speed of 0.1 μl/min with an additional 3 min allowed for diffusion before careful retraction of the needle and moving to the next coordinate. A total of 1.2 µl was injected in each hemisphere using a glass capillary and injection pump (Nanoliter 2010, World Precision Instruments). The wound was sutured with resorbing thread (Vicryl, Ethicon).

Kainic acid-induction of status epilepticus in adult mice and electrode implantation

IHKA was chosen as a well-established mouse model of chronic TLE with HS. This model is used by the NIH/NINDS Epilepsy Therapy Screening Program to test the efficacy of new antiepileptic treatments since it mimics drug-resistant seizures as described in TLE patients [75].

A minimum of two weeks after virus injection, mice were again anesthetized and placed in the stereotaxic frame as described above. Unilateral KA injections were performed at the right dorsal hippocampus following coordinates (in mm, from Bregma): AP −2.0, ML + 1.6, DV −1.9. A total of 45 nl of KA solution (20 mM i.e., total dose 0.9 nM KA; Abcam ab120100) was injected at a rate of 25 nl/min followed by 3 min of waiting time before retraction of the glass capillary.

During the same surgery, the electrode (E363/2, P1 Technologies) and, after blunt dissection creating a skin pocket over the right dorsal thoracic region, a telemetry transmitter (MT10B, KAHA Sciences) for EEG recording were implanted. In brief, an insulated stainless-steel electrode (203 µm diameter) was inserted at the same coordinates as for KA injection. The positive lead of the transmitter was connected before insertion by clamping to the electrode connector, and sealed with conductive paint (Bare Conductive). The negative (reference) lead was similarly connected to a skull screw (AgnTho’s) in the occipital or parietal bone. Connections were fixed on the skull and insulated with first cyanoacrylate glue (Super Glue, Loctite) followed by dental acrylic cement (AgnTho’s) and the opening of the skin pocket containing the transmitter was sutured with resorbing thread.

In our study, the IHKA mouse model deviated from the previously described model in the literature [75,76,77,78], complicating the assessment of the therapeutic outcomes. Typically, the majority of IHKA-treated mice exhibit numerous focal nonconvulsive electrographic seizures, with behavioral seizures being infrequent. However, in the current study, 4 out of 9 IHKA mice developed electrographic seizures exclusively, while all displayed frequent behavioral seizures. This divergence from prior studies could be attributed to the concurrent IHKA injections and electrode implantation during the same surgical session, potentially causing a greater spread of the kainic acid (KA) within the hippocampus, including along the electrode track. This likely resulted in more extensive tissue damage and disruption of the BBB, leading to a more severe epileptic phenotype. Future experimental designs should take into account the potential increased KA spread and BBB disruption when combining procedures. Additionally, the experimental design might need to address the issue of neuronal loss following SE induction post-viral vector injection, which could reduce the effectiveness of the uPSEM817 treatment.

Intrahippocampal EEG recordings and drug treatment

The EEG was recorded using a telemetry system (MT110 tBase, Kaha Sciences) connected to an ADC (PowerLab, AD Instruments) with LabChart Pro software (AD Instruments) on a Windows PC. The sampling rate was 1 kHz. The video was registered using two HD ethernet cameras (Axis Communications), combined in the open-source Open Broadcaster Software (OBS Studio), and synchronized and recorded using LabChart. After the surgery, the animals were video-EEG monitored for status epilepticus for at least 24 h before being returned to stables. Two weeks later, continuous video-EEG monitoring commenced during which the different treatment conditions were administered according to the schedule (Supplementary Fig. 2C): uPSEM817 (PSAM4 ligand), saline (vehicle and negative control treatment), and PHB (ASM, positive control treatment).

All treatments were administered at the same time of the day (10:30 a.m.) to avoid possible circadian influence. uPSEM817 diluted in saline at the concentration of 0.05 mg/ml was administered i.p. at the dose 0.03 mg/kg [44]. As a negative control treatment, the corresponding volume of saline was administered i.p. As a positive control for uPSEM817, we injected i.p. PHB at the dose of 40 mg/kg [51]. After the end of the in vivo experiments, the animals were either perfused or used for in vitro electrophysiological experiments.

EEG data analysis

The in vivo data were analyzed using LabChart software and Matlab R2019b. LabChart was used to manually review the recordings and label treatment administration and large behavioral seizures (LBSs) [79]. The behavioral severity was quantified using the Racine scale [80] with added sixth class representing wild running seizure type. For loading the signals and labels from the LabChart *.adicht files into the Matlab environment we wrote our scripts and utilized ADInstruments (LabChart) SDK [81].

For the detection of ESs and IEDs, we used a semi-automatic Matlab-based Electrographic seizure analyzer (ESA) available at https://github.com/KM-Lab/Electrographic-Seizure-Analyzer [79]. The thresholds for the detection of spikes and artifacts were set individually for each animal using the plotting function of ESA to maximize the number of true detections while not allowing false detections. We used the default settings of the ESA except for the ES detection which we required to have at least 8 spikes in each 4 s. The minimum seizure length was 4 s and ESs closer than 4 s were glued together. For the ES detection, we used negative spikes. In the animals in which ESs were present, we considered only positive spikes to be IEDs. In the animals which did not present ESs, we have chosen the polarity of spikes which was more frequent to be considered as IEDs. We considered only narrow spikes as IEDs (Fig. 4A).

For the ESs we analyzed the following five parameters: ES rate, mean ES duration, mean number of spikes constituting the ES, and the mean spike amplitude within the ES. For the IEDs, we determine their rate and mean amplitude. In the calculation of the ES rate, we excluded the time spent in behavioral seizures or the artifact-contaminated periods. To determine the IED rate, we also excluded the time spent in the ESs.

The effect of uPSEM817 was analyzed from 20 min to 240 min after i.p. injection, to avoid any possible injection effect contamination and due to previous data in PSEM pharmacokinetics [44, 64]. For PHB analysis data from 10 min to 70 min after i.p. injection was considered following the previous reasoning [51]. Data are represented as the median ± interquartile range of the median fold change of the subjects after normalization to saline injection. Importantly, for the GFP-only group, no animals displayed ESs, and for IED analysis #32 corresponds to a repeated recording in animal #31.

In vitro electrophysiology

Whole-cell patch-clamp recordings were performed in acute hippocampal slices. First, mice, after being removed from the video-EEG monitoring, were briefly anesthetized with isoflurane and decapitated. Brains were transferred to an ice-cold sucrose-based cutting solution containing (in mM): 75 sucrose, 67 NaCl, 26 NaHCO3, 25 D-glucose, 2.5 KCl, 1.25 NaH2PO4, 0.5 CaCl2, 7 MgCl2 (all from Sigma Aldrich), equilibrated with carbogen (95% O2/5% CO2), with pH 7.4 and osmolarity ~305–310 mOsm. Hemispheres of the brains were separated and cut on a vibratome (VT1200S, Leica Microsystems) into 400 μm thick slices. The right hemisphere was cut sagittal to better visualize the dorsal hippocampus where the electrode was implanted, and the left hemisphere was cut quasi-horizontal [82]. Slices were incubated in this cutting solution for 30 min at 34 °C, and subsequently transferred to aCSF containing (in mM): 119 NaCl, 26 NaHCO3, 25 D-glucose, 2.5 KCl, 1.25 NaH2PO4, 2.5 CaCl2 and 1.3 MgCl2 (pH 7.4, osmolarity 305–310 mOsm). Slices were kept in aCSF at room temperature (RT) until recordings were performed. The individual cells in the slices were visualized for whole-cell patch-clamp recordings using infrared differential interference contrast video microscopy (BX51WI; Olympus). Recordings were performed from GFP+ cells (identified under fluorescent 470 nm light) at 32 °C using a glass pipette filled with a solution containing (in mM): 140 K-Gluconate, 4 NaCl, 10 KOH-HEPES, 0.2 KOH-EGTA, 2 Mg-ATP, and 0.3 Na3GTP (~300 mOsm, pH 7.2; all from Sigma-Aldrich). The average pipette tip resistance was between 2.5 and 6 MΩ. Pipette capacitance was corrected online before GΩ seal formation while fast capacitive currents were compensated for during cell-attached configuration. All recordings were done using a HEKA EPC9 amplifier (HEKA Elektronik) and sampled at 10 kHz with a 3 kHz Bessel anti-aliasing filter and using PatchMaster software for data acquisition. For each recording, baseline parameters were measured immediately after accessing the cell and again 3–5 min later, just before initiating drug wash-in. The measurements taken just prior to the application were used for analysis. uPSEM817 was then introduced at a final concentration of 3 nM, dissolved in aCSF, using a perfusion pump. The effects of uPSEM817 were assessed 30–40 min after the start of the wash-in. Finally, 30–40 min after beginning the wash-out period, during which regular aCSF without the drug was perfused, measurements for the wash-out parameters were taken.

Passive membrane properties of GFP+ cells and spontaneous synaptic activity: After the formation of a GΩ seal, the patch was ruptured giving direct access to the intracellular compartment. RMP was determined in current-clamp mode at 0 pA immediately after establishing the whole-cell configuration. Series resistance (Rs) and input resistance (Ri) were determined in voltage clamp at −70 mV from a 5 mV negative voltage pulse applied through the patch pipette and monitored throughout the experiment. A series of square current steps of 500 ms duration from −40 pA to 200 pA in 10 pA steps, were applied at a membrane potential of approximately −70 mV with holding current as needed, to determine the cells’ ability to generate AP. AP characteristics were assessed by administration of a depolarizing ramp current over 1 s, from a holding potential of −70 mV, starting with a 0–25 pA ramp and up to a 0–500 pA ramp in various cells. Spontaneous postsynaptic potentials were recorded at 0 pA holding current.

Preparation of rhinal cortex-hippocampus organotypic slices

Rhinal cortex-hippocampus organotypic slices were prepared from 6–7 days-old Sprague-Dawley rats, according to previous reports [47, 48]. No sex determination was conducted on the pups. Each experiment utilized the entire litter, which included both male and female pups. Rats were euthanized by decapitation and the brain removed, under sterile conditions, to a 60 mm plate with ice-cold Gey’s balanced salt solution (GBSS) (Biological Industries, Israel) supplemented with 25 mM D-glucose (Sigma, USA). The hippocampus, entorhinal cortex, and perirhinal cortex were separated and sliced transversely (350 µm thick) using a McIlwain tissue chopper. Four slices were transferred to porous (0.4 µm) insert membranes (PICM 03050, Millipore, USA) placed in wells of six-well culture trays (Corning Costar, Corning, USA) containing 1.1 mL of culture medium composed of 50% Opti-MEM, 25% Hanks’ balanced salt solution (HBSS), 25% heat-inactivated horse serum (HS), 30 μg/mL gentamycin (Thermo Fisher, USA), and 25 mM D-glucose (Sigma). Slices were maintained at 37 °C with 5% CO2 and 95% O2 for the following 2 weeks. From 3 days in vitro (DIV) on, slices were changed to supplemented Neurobasal A (NBA) medium (NBA, 2% B-27, 1 mM L-glutamine, gentamycin 30 μg/mL, all from Thermo Fisher) and subjected to decreasing HS concentrations (15%, 10% and 5%) until a serum-free medium was reached. Culture medium was changed every other day.

At 3 DIV, 2 µl of either AAV8-CAMKIIα-eGFP or AAV8-CAMKIIα-uPSAM4-GlyR-IRES-eGFP viral vector, diluted 1:10 in NBA, was carefully placed on the top of the hippocampus. Just before recording, each slice was visually inspected, to ensure slice integrity, and a fluorescence image was acquired in a Zeiss Axiovert 200 fluorescence microscope, equipped with an AxioCam MRm, using the AxioVision imaging software (Zeiss, Germany).

Spontaneous activity recordings from organotypic slices

Under a gradual deprivation of serum rhinal cortex-hippocampus organotypic slices depict spontaneous epileptiform discharges [47, 48]. At 14 DIV individual slices were transferred to an interface-type chamber with a humidified (5% CO2/95% O2) atmosphere at 37 °C, and with the NBA medium continuously recirculating at a rate of 2 mL/min. After a stabilization period of 5 min, the baseline epileptiform activity was recorded for 20 min. The superfused medium was then changed to NBA supplemented with 6 nM uPSEM817. After an equilibration period of 15 min, to ensure a complete renewal of medium bathing the slice, a 20 min recording was acquired (Recording scheme in Fig. 4A). Spontaneous field potentials were recorded using a glass micropipette electrode (2–4 MΩ) filled with artificial cerebrospinal fluid (aCSF) containing (in mM): 124 NaCl, 3 KCl, 1.2 NaH2PO4, 25 NaHCO3, 10 glucose, 2 CaCl2 and 1 MgSO4 (pH 7.4), and positioned in the CA3 pyramidal cell layer. Recordings were obtained with an Axoclamp 2B amplifier (Axon Instruments, USA), and digitized with the WinLTP software (WinLTP Ltd., UK) [83].

In this study, ictal-like discharges were defined as continuous discharges lasting more than 10 s (bursts) or with a minimum frequency of 2 Hz, and the end of a burst was defined when the inter-spike interval was longer than 2 s [84]. Continuous activity that did not fit within these parameters was not considered burst activity. The pCLAMP Software Version 10.7 (Molecular Devices Corporation, California, USA) was used to automatically detect the events during a recording. All recordings were band-pass filtered (eight-pole Bessel filter at 60 Hz and Gaussian filter at 600 Hz).

The baseline used in pCLAMP to detect the events was specific to each recording and was established right above the noise of oscillations. The number of bursts per slice, the number of events within a burst, and the duration of each burst, as well as the frequency of events and the average positive peak amplitude (amplitude between the baseline and the peak of the spike) in a burst, were automatically evaluated in an in-house program developed in C++ language.

Immunofluorescence

Hippocampal slices were fixed after acute recordings overnight at 4 °C with 4% PFA and changed to KPBS after. Then, slices were washed thoroughly three times with KPBS, incubated for 1 h at RT in permeabilization solution (0.02% BSA  + 1% Triton X‐100 in PBS) and 2 h at RT in blocking solution (5% normal serum +1% BSA + 0.2% Triton X‐100 in PBS). Primary antibodies (Chicken anti-GFP, 1:400, Abcam ab13970) were diluted in blocking solution and incubated for 48 h at 4 °C. Then, slices were incubated again in blocking solution for 2 h at RT. Secondary antibodies (AlexaFluor Plus 488 Goat anti-chicken, 1:1000, ThermoFisher) were applied in blocking solution for 48 h at 4 °C. Nuclei were counterstained with Hoechst 33342 (1:1000) diluted in the last rinsing with PBS for 20 min before mounting with PVA-DABCO mounting media. Images were acquired by confocal microscopy (Nikon Confocal A1RHD microscope).

For mossy fiber sprouting and dentate GC layer dispersion, slices were subcutted in the microtome to 30 μm thickness. Then, sections were washed thoroughly three times with KPBS and then pre-incubated for 1 h in blocking solution containing 10% normal serum (of the species-specific to the secondary antibody) in KPBS containing 0.25% Triton-X-100, for 1 h at RT. Primary antibodies diluted in the blocking solution were incubated overnight at 4 °C (Rabbit anti-ZnT3 [Synaptic systems 197003, 1:500], and mouse anti-NeuN [Millipore MAB377, 1:400]). Following primary antibody incubation, sections were washed three times with KPBS and incubated again in blocking solution for 1 h at RT. Then, sections were incubated with secondary antibodies for 2 h at RT (AlexaFluor Plus 555 goat anti-rabbit or mouse, 1:1000, ThermoFisher). Nuclei were counterstained with Hoechst 33342 (1:1000) diluted in the last rinsing with PBS before mounting with PVA-DABCO mounting media. Images were acquired by epifluorescence microscopy (Olympus BX61 and Leica DMi8).

Statistical analyses

Whole-cell patch-clamp recordings were analyzed offline with Igor Pro (Wavemetrics). The cell was excluded from analysis if Rs changed during the recording by more than 20%. On-ramp recordings, AP amplitude was measured from threshold to peak and the minimum current needed for the first AP was measured. On step recordings, AP amplitude and threshold were measured as before, and the amplitude of the afterhyperpolarization (AHP) was measured as the difference between the AHP peak and the AP threshold. For the 500 pA steps, the steady-state amplitude was measured as the difference between the holding potential, −70 mV, and the potential at the plateau of the pulse response.

For fluorescence quantification, an image of the right dorsal DG, at the level of the electrode placement was digitally acquired using the 10X objective (epifluorescence microscopy, Olympus BX61). All images were taken with the same exposure time and ISO and treated equally. At least three different hippocampal sections were used for the analysis of each animal. Subsequently mean fluorescence intensity was measured using Fiji/ImageJ software (NIH, Annapolis, MD, USA). DG area was outlined and the mean gray value was measured within the selected area. Moreover, the maximum gray value was also measured for each area.

Statistical analysis of the data was performed using Prism 7 (GraphPad). Data were represented as paired, and medians ± interquartile range was used for group representation. Wilcoxon rank test was used for comparison of medians with Bonferroni’s post hoc test for multiple comparisons of medians. Statistical analysis between the activity parameters evaluated in the baseline recording of each organotypic slice and the ones assessed in the recording under uPSEM817 superfusion was achieved by a paired Wilcoxon test. The level of significance for the tests was set at p < 0.05. All data is presented in the figures as medians ± interquartile range.