Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Integration of the CA2 region in the hippocampal network during epileptogenesis

Integration of the CA2 region in the hippocampal network during epileptogenesis INTRODUCTIONA striking characteristic of pyramidal cells of the cornu ammonis 2 (CA2) region is their resistance toward hyperactivity‐induced cell death in mesial temporal lobe epilepsy (MTLE; Steve et al., 2014). MTLE is the most common form of focal epilepsies in adult patients and characterized by seizures which involve the hippocampus and surrounding temporal brain structures and which are often resistant to pharmacological treatment. Furthermore, MTLE is frequently associated with hippocampal sclerosis including substantial loss of principal cells and inhibitory interneurons and accompanying gliosis (Blümcke et al., 2013; Engel, 2001; Thom, 2014).Instead of a single “antiepileptic” mechanism, the resistance of CA2 pyramidal cells in MTLE is most likely based on a combination of their unique characteristics, which might include the regulation of plasticity through perineuronal nets (Carstens et al., 2016, 2021; Hayani et al., 2018), specific Ca2+‐buffering properties (Lee et al., 2010; Leranth & Ribak, 1991; Simons et al., 2009), upregulation of neuroprotective factors like brain‐derived neurotrophic factor (Tulke et al., 2019) and integration into a strong inhibitory network (Botcher et al., 2014). However, despite their resistance, the physiological identity of CA2 pyramidal cells in MTLE is altered (Williamson & Spencer, 1994), including the generation of spontaneous bursts in acute hippocampal slices from human specimens (Knowles et al., 1987; Reyes‐Garcia et al., 2018; Wittner et al., 2009). In addition, a recent study in the pilocarpine mouse model for MTLE has found increased excitability in and decreased inhibition onto CA2 pyramidal cells (Whitebirch et al., 2022).The integration of CA2 pyramidal cells into the epileptic network in and beyond the sclerotic hippocampus is, however, still far from being understood. We and others have shown that mossy fibers sprout to the CA2 pyramidal cell layer to build aberrant synapses onto principal cell somata in the intrahippocampal kainate (KA) mouse model for MTLE (Häussler et al., 2016), the rat pilocarpine model (Althaus et al., 2016) and in human tissue (Freiman et al., 2021; Houser et al., 1990; Wittner et al., 2009). In the pilocarpine model, mossy fibers provide stronger input to CA2 pyramidal cells in comparison to control, although they do not sprout (Whitebirch et al., 2022), whereas CA2 pyramidal cells themselves provide increased excitatory input to all regions of the cornu ammonis in slices. Indeed, CA2 contributes to epileptic activity (EA) in vivo (Tulke et al., 2019) and its blockade using chemogenetic tools resulted in reduced seizure occurrence (Whitebirch et al., 2022). All together, these data point to a central role of CA2 in the epileptic network.It remains unclear whether the altered input via the mossy fibers translates into dentate gyrus (DG) engaging CA2 during EA in vivo and whether it changes the coupling of both regions during non‐pathological activity patterns like theta oscillations. Furthermore, whereas DG granule cells are supposed to lack direct connections to the contralateral hemisphere (Hjorth‐Simonsen & Laurberg, 1977) and cannot directly convey EA there, CA2 pyramidal cells connect bilaterally (Cui et al., 2013) and therefore might contribute to propagation of EA beyond the ipsilateral hippocampus.To address these questions, we used the intrahippocampal KA mouse model for MTLE and corroborated the local origin of EA and theta oscillations in ipsilateral CA2 by silicon probe recordings followed by current source density (CSD) analysis. Viral tracing confirmed the preservation of the projection from ipsilateral CA2 to the contralateral hippocampus in MTLE. By assessing the time course of EA in ipsi‐ and contralateral DG and CA2, as well as co‐occurrence of EA in those regions, we found a strong coupling of the DG with ipsi‐ and to a lesser extent contralateral CA2 during EA. Furthermore, theta rhythm in CA2 was coherent with the slowed‐down oscillation in the DG (Kilias et al., 2018) already early after status epilepticus (SE). In situ hybridization revealed the loss of glutamic acid decarboxylase 67 (Gad67) mRNA‐expressing interneurons directly after the initial SE in ipsi‐ but not contralateral CA2, which might contribute to the early alterations in excitability. We conclude that CA2 is an active part of the epileptic network that may act as one gate for propagation of EA beyond the sclerotic hippocampus.METHODSSurgical procedures and LFP recordingsAnimalsAll experiments were performed in adult (>8 weeks, 22‐30 g) male C57Bl/6 wild‐type mice and transgenic B6(SJL)‐Tg(Amigo2‐icre/ERT2)1Ehs (Amigo2‐icreERT2) mice (Alexander et al., 2018) or their negative littermates (all bred at CEMT Freiburg). We assume that the inducible Amigo2‐icreERT2 allele does not play a role for the characteristics of KA‐induced MTLE, as this has recently been shown for the constitutive Amigo2‐cre line (Lisgaras & Scharfman, 2022). Mice were housed under standard conditions (12 h‐light/dark cycle, room temperature [RT] 22 ± 1°C, food and water ad libitum). Experiments were carried out following the guidelines of the European Community's Council Directive of 22 September 2010 (2010/63/EU) and were approved by the regional council (Regierungspräsidium Freiburg). In total, 54 mice were used of which a subset was already analyzed for different aspects in a previous study (Tulke et al., 2019).Intrahippocampal injectionsSurgery was performed under deep anesthesia (ketamine hydrochloride 100 mg/kg, xylazine 5 mg/kg, atropine 0.1 mg/kg body weight, i.p.) with analgesic treatment (buprenorphine hydrochloride 0.1 mg/kg, carprofen 4 mg/kg, s.c.). Stereotaxic injections of 50 nl kainic acid (KA, 20 mM in 0.9% saline, Tocris, Cat 0222, RRID:SCR_003689) or 0.9% sterile saline into the right dorsal hippocampus were performed at coordinates anterioposterior [AP] = −2.0 mm, mediolateral [ML] = −1.4 mm, and dorsoventral [DV] = −1.8 mm from bregma using a stainless steel cannula (Harry Rieck Edelstahl GmbH) connected via polyethylene tubing to a 0.5 μl Hamilton syringe (7000.5 N, Hamilton Bonaduz) and a micropump (Carnegie Medicine). In KA‐injected mice, the occurrence of behavioral SE, characterized by head nodding, mild convulsions, rotations and immobility was verified as previously described (Häussler et al., 2016; Tulke et al., 2019). Postsurgical analgesia was performed with burprenorphin and carprofen for 2 days.To trace the projections from ipsilateral CA2 to the contralateral hippocampus Amigo2‐icreERT2 mice were injected with an adeno‐associated virus (AAV, serotype 1) that delivered phSyn1(S)‐FLEX‐tdTomato‐T2A‐SypEGFP‐WPRE (Addgene plasmid #51509, Penn Vector Core or Viral Vector Facility ETH Zürich, RRID:Addgene_51509; Oh et al., 2014) into CA2 at coordinates AP = −2.0 mm, ML = −2.6 mm from bregma, DV = −1.0 mm from the cortex surface. When AAV and KA injection were combined, AAV was injected 2 days prior to KA. The AAV (250 nl) was injected using a glass capillary and a programmable nanoliter injector (Nanoject III, Drummond Scientific Company) during 10 min. The capillary was left in place for another 3–5 min to avoid reflux. To induce expression of the AAV‐delivered construct, Amigo2‐icreERT2 mice were injected (i.p.) with tamoxifen in sunflower oil (100 mg/kg body weight; Sigma‐Aldrich, T5648) once per day for five consecutive days starting from 5 or 6 days after AAV injection onward (timelines for experiments in Supplementary Figure 1).Electrode implantations and in vivo recordingsElectrode implantation was either performed in the same surgery or 14 days later (anesthesia and analgesia see above). Wire electrodes (platinum‐iridum, teflon‐insulated, ∅125 μm, World Precision Instruments, PTT0502) were implanted into the DG (AP = −2.0 mm, ML = ±1.4 mm, DV = −1.9 mm) and CA2 (AP = −2.0 mm, ML = ±2.6 mm, DV = −1.4 mm) of one or both hemispheres. Jewelers' screws above the frontal cortex served as ground and reference. Electrodes were soldered to a connector, which was permanently fixed to the skull with cyanoacrylate and dental cement. Post hoc Nissl staining (section thickness 50 μm for wires) was used to verify the location of wire tips in the molecular layer or granule cell layer of the DG and in stratum oriens (so), stratum pyramidale (sp), or stratum lucidum/stratum radiatum (sl/sr) of CA2. Note that mossy fiber sprouting in CA2 prevents a clear discrimination of sr and sl borders in CA2 (Häussler et al., 2016). Only data from electrodes located unambiguously in CA2 and DG were used for analysis, which results in different numbers of electrodes in the analyses that depended on proper placement of individual electrodes or on pairs of electrodes (Supplementary Table 1). Custom‐made single shaft silicon probes (16 iridium‐oxide [IrOx] electrodes of 35 μm diameter, spacing 45 μm; Herwik et al., 2009, 2011; Lanzilotto et al., 2016) were chronically implanted into CA2 (AP = −2.0 mm, ML = −2.6 mm, probe tip DV = −1.7 mm from the cortex surface). Probe positions were reconstructed on Nissl‐stained sections (section thickness 35 μm), on 4′,6‐diamidino‐2‐phenylindole (DAPI)‐stained sections or with Purkinje cell protein 4 (PCP4) immunocytochemistry (see Section 2.2.1; Supplementary Figure 2). Individual electrode positions of silicon probes were reconstructed post hoc based on histological traces of the probe track and the amplitude profile of action potentials to delineate sp (Supplementary Figure 2).Recordings were performed 1–2 days (≤48 h), 7, 14, and 21 days after KA injection for mice implanted with wires and 21/22 days after KA for mice implanted with probes (Supplementary Figure 1). Freely moving mice were connected to one or two miniature preamplifiers (MPA8i, Multichannel Systems). LFPs were filtered, amplified (gain: wires and probes 500‐fold, filter: wires 1 Hz–5 kHz, probes: 1 Hz–1 kHz; PGA32, Multichannel Systems) and digitized (sampling rate wires: 10 kHz, probes for LFP: 2 kHz, Power1401 mk2/ADC16, Spike2 software; CED, RRID: SCR_000903).Histological analysisPerfusion, slice preparation, immunocytochemistry, and in situ hybridizationMice were transcardially perfused under deep anesthesia with 0.9% saline followed by paraformaldehyde (4% in 0.1 M phosphate buffer [PB], pH 7.4) >14 days after implantation or at 2, 7, 14, and 21 days after KA/NaCl injection for Gad67 fluorescence in situ hybridization (FISH). Brains were dissected and postfixed in the same fixative overnight (4°C), cut on a vibratome in the frontal or horizontal plane (50 μm thickness), the next day and collected in PB for analysis of viral tracing. For FISH, brains were cryoprotected in 30% sucrose, frozen in isopentane and stored at −80°C. Cryostat sections (50 μm) were prepared in the frontal plane and collected in PB or in 2x saline sodium citrate.For immunocytochemistry, slices were preincubated with 0.25% Triton X‐100 (Sigma‐Aldrich) and 10% normal goat serum in 0.1 M PB (30 min, free‐floating), followed by incubation for 4 h at RT+ overnight at 4°C with polyclonal rabbit anti‐PCP4 (1:400; Sigma‐Aldrich GmbH, HPA005792, RRID:AB_1855086) or polyclonal rabbit anti‐green fluorescent protein (GFP, 1:1000; Abcam, ab6556, RRID:AB_305564) primary antibodies with 0.1% Triton and 1% normal goat serum. A secondary goat anti‐rabbit antibody conjugated to Cy2 (1:400; Jackson ImmunoResearch, 111‐225‐144, RRID:AB_2338021) was applied for 2–3 h at RT. Counterstaining was performed with DAPI (1:10,000, Roche, Cat 236276). After washing, sections were mounted on microscope slides and coverslipped with Immu‐Mount (Thermo Shandon Ltd).The expression of Gad67 mRNA was investigated by FISH with a digoxigenin (DIG)‐labeled cRNA probe generated by in vitro transcription from appropriate plasmids (Kulik et al., 2003). Sections were hybridized overnight at 45°C, followed by detection with an anti‐DIG antibody and tyramide signal amplification as described (Marx et al., 2013; Tulke et al., 2019). FISH was always followed by PCP4 immunohistochemistry for precise location of CA2 as described above, but for double‐staining Triton X‐100 was omitted.Microscopy and quantificationSections were imaged with an epifluorescence microscope (Axio Imager 2, Zen software, Zeiss, RRID:SCR_018876, RRID:SCR_013672) with a 10‐fold objective (Plan Apochromat, NA 0.45), appropriate filter settings and a digital camera (AxioCam 506). All images that were compared were taken with identical exposure times. To analyze the DV extent of AAV labeling of somata (tdTomato) and axonal projections (synaptophysin‐bound EGFP), images of horizontal sections throughout the DV extent of the hippocampus were taken and assigned to the planes of a horizontal brain atlas (Franklin & Paxinos, 2007). All planes were checked for tdTomato‐positive somata and EGFP‐expressing axons and the location of the most dorsally and most ventrally located tdTomato‐positive somata and the most dorsally and most ventrally located EGFP‐expressing axons in the ipsi‐ and contralateral hippocampus, respectively, were visually determined by two independent observers. To measure density profiles across the layers of ipsi‐ and contralateral CA2 in the horizontal plane we selected the section closest to the KA injection site (W‐like shape of the granule cell layer). Images were converted to 16‐bit in ImageJ (version 1.53a, National Institute of Health) and intensity profiles along a straight line (length 330 μm, drawn from so to sr/sl) were measured at three equidistant positions and averaged. All intensity values are given relative to the maximum intensity in ipsilateral CA2 of the respective slice. For measurement of intensity profiles in CA3 and CA1 we proceeded likewise, only with different line lengths (350 μm for CA3, 450 μm for CA1). Values are again given relative to the maximum in ipsilateral CA2.For quantification of Gad67 mRNA‐expressing cells in CA2, we selected dorsal slices surrounding the KA injection site (two to four slice/mouse). The location of CA2 was determined in the PCP4 immunostaining, subsequently three regions of interest (ROIs) were placed in the FISH images framing so, sp and stratum lacunosum moleculare (slm)/sr/sl of CA2. All Gad67+ cells within every ROI were counted manually and average values per mouse are given as cells/mm2. For all analyses, blinding to treatment or time point was not possible due to the salient progressive changes after KA injection (cell loss, granule cell, and CA2 dispersion).Data analysisDetection of epileptiform activityFirst, we controlled for sufficient data quality: As mentioned above, only recordings with proper electrode placement in the DG or CA2 were used. In addition, noisy or artifact‐rich recordings, either due to electrode/soldering quality or mouse behavior were discarded. Finally, not all mice were recorded at all days of the series. This results in different numbers of mice for each recording site and recording day (Supplementary Table 1).Data analyses of preselected data were done in Python 3.7.3 (Python Software Foundation, RRID:SCR_008394) and Matlab 2022a (Mathworks, RRID:SCR_001622). Individual epileptiform discharges in the LFP were detected with a custom‐made algorithm in Python as previously described (Heining et al., 2019; code accessible at Zenodo: https://doi.org/10.5281/zenodo.4110614). In brief, LFP data were resampled at 500 Hz, artifacts were manually removed and discharges were detected based on a combination of frequency composition and amplitude threshold. Discharges occurring within 2.5 s were grouped into one burst. Bursts were classified according to number of discharges, median interdischarge interval, and coefficient of variation using Ward's hierarchical clustering into mild, moderate, and severe events. Detection was performed for all channels individually (DG ipsi/contra, CA2 ipsi/contra).We defined the coupling between two hippocampal regions as the overlap of EA in time. Hereby, we focused our analysis on severe EA in a seed region (DGi, CA2i) and quantified the temporal overlap with co‐occurring EA of any class detected in a target region (CA2i, DGi, CA2c). The overlap is presented as a fraction of the total severe EA time in the seed region. Due to low sample size, in particular at 21d, we did not include DGc in this analysis.Analysis of theta periodsTheta properties were analyzed as previously described (Kilias et al., 2018). In brief, LFP periods (≥9 s) dominated by theta activity and free of EA, slow wave sleep, large irregular activity associated with resting, or movement artifacts were identified manually a all available recording channels. LFP recordings were resampled at 1 kHz. Autocorrelation and power spectral density (Thomson's multitaper power spectral density estimate, 5 Slepian tapers) were calculated on sliding windows (4 s shifted in 0.25 s steps) for the selected LFP periods. For each window, we detected the spectral power maximum and its corresponding frequency within theta band (5–12 Hz). We smoothed the resulting power–frequency distributions (median filtering, 0.75 Hz × 1.5 × 10−3 mV2/Hz). Theta peak frequency was defined as the frequency at the maximum of the theta frequency probability distribution after a second smoothing step (Savitzky–Golay, first order, 1.25 Hz window). This analysis was done for all animals (nanimals = 12) and all recording sites (nrec sites = 35).Recording sessions in which the autocorrelation and the power spectra lacked a pronounced peak in the theta band and where no peak in the power–frequency distributions could be detected, were excluded from further analysis. This was particularly relevant for recordings performed ≤48 h after SE (see Figure 4b).Coherence was averaged across sliding windows (4 s with 3 s overlap) of all sessions for 2–250 Hz at 0.25 Hz resolution. Coherence peaks in theta band were detected as local maxima in the Sawitzky‐Golay‐filtered coherence between 5 and 12 Hz after averaging across animals.CSD estimationWe performed CSD analysis on data recorded 21/22 days after KA injection in animals implanted with a single shaft 16‐channel silicon probe covering all layers of CA2 (nKA = 3). To identify the sink‐source profile of EA in CA2 we detected negative peaks that exceeded an amplitude of 0.9 mV in a 50 Hz low‐pass filtered LFP recorded in slm. Data were extracted in a window of −150 to +500 ms around identified peaks, averaged across a subset of 340 randomly selected windows per animal. CSDs were estimated from these averages. Subsequently, CSDs were cropped to seven channels covering slm to sp. In one animal, we bootstrapped one channel in the middle of sl/sr to account for the structural variability. The cropped individual CSDs were averaged across all three mice to obtain a final CSD profile.To identify the sink‐source profile of theta activity in CA2 we filtered the LFPs (two mice, one mouse was excluded because movement artifacts impeded theta analysis) for the theta band (band‐pass, second order, butterworth) and z‐scored to its mean and standard deviation (SD) across all selected theta periods per session. In the z‐scored theta periods, we detected all theta cycles with peaks that exceeded a threshold of 0.5 × SD and with both adjacent peaks above threshold. We cut 1 s windows around the peak times of these cycles and averaged across 3900 randomly selected windows and CSDs were calculated for these averages. We cropped and averaged the CSDs as described above to obtain an interindividual mean. We computed the spectrum of the CSD traces originating from sr (Thomson's multitaper power spectral density estimate, 3 Slepian tapers) and identified the frequency of maximal power.To circumvent effects of slight differences in gains and noise of the two independent preamplifiers, where one MPA8I amplified signals from all even and the other from all odd electrodes of the silicon probe, we calculated separate CSDs for even and odd electrodes, with 90‐μm electrode spacing, respectively, and these were subsequently merged and smoothed (2D median filter, 2 electrodes × 15 ms).StatisticsStatistical analysis was performed with MATLAB for electrophysiological data and GraphPad Prism 9 (GraphPad Software LLC, RRID:SCR_002798) for histology. Individual values, mean ± SD are presented. Results from the EA analysis were statistically evaluated by a two‐ or three‐way analysis of variance (ANOVA) with time point, structure (DG/CA2) and hemisphere if applicable as factors followed by a Tukey's multiple comparison test. Changes in theta frequency were compared across time by a repeated measure three‐way ANOVA and compared to a previously published distribution of theta frequency in healthy and KA‐injected mice with a Kolmogorov–Smirnov test. Theta coherence peaks were compared to predominant theta frequency by using a one‐sample t test with the mean frequency of both tested electrodes as the hypothesized mean of the underlying distribution.For statistical evaluation of density of Gad67+ cells, a two‐way ANOVA was performed for factors treatment/time point and hemisphere followed by a Tukey's multiple comparison test for treatment/time point and a Šidák posttest for pairwise comparison between hemispheres for each time point.RESULTSEA occurs locally in ipsilateral CA2To determine the role of CA2 under epileptic conditions, we implanted wire electrodes into the DG and CA2 of the ipsi‐ and contralateral hippocampus of KA‐injected mice and measured EA as well as periods dominated by theta activity but free from any ictal or interictal events (Figure 1). Positioning of electrodes in the molecular layer or granule cell layer of the DG and in sl/sr, sp or so of CA2 was verified in post hoc Nissl staining (Figure 1a). In a subset of mice, the placement of electrodes or silicon probes in CA2 was further confirmed by PCP4 immunostaining (Figure 1a). Episodes of EA, mostly occurring during quiet wakefulness, alternated with theta episodes during sleep or locomotion (Figure 1b, 7 days after KA). We applied an automated detection algorithm with subsequent classification of EA into three different categories—mild, moderate, or severe—based on the number of discharges, interdischarge intervals and coefficient of variation for each channel independently (Figure 1b; Heining et al., 2019). Theta episodes were identified manually by cross‐validating all available channels per session and animal (Figure 1b).1FIGUREEpileptic and theta activity in CA2 have a source local to CA2. (a, left) Implantation scheme for bilateral electrodes in the dorsal dentate gyrus (DG) and CA2 with color‐coded severity of kainate (KA)‐induced hippocampal reorganization (darker red ≙ stronger damage). (a, middle) Nissl‐stained hippocampal sections show the tips of the wire electrodes (white arrows). (a, right) Section of the ipsilateral hippocampus of a mouse implanted with a silicon probe immunostained for Purkinje cell protein 4 (PCP4, green) with 4′,6‐diamidino‐2‐phenylindole (DAPI) counterstaining (magenta). The probe shank crossed the pyramidal layer of CA2. (b, top) Local field potential (LFP) traces of ipsi‐ and contralateral DG and CA2 at 7 days after KA. Manually identified theta periods (green bars) alternated with automatically detected epileptic activity (EA) of different severity (violet: severe, lilac: moderate, pink: mild EA). (b, bottom left) Zoom into a period of EA co‐occurring at all four positions with discharges of large amplitude in ipsilateral DG and CA2 and more sparse epileptic spikes in the contralateral hippocampus (DGc and CA2c). (b, bottom right) Theta oscillations could be observed simultaneously at all four positions. Broadband activity (black trace) is shown overlayed with its theta‐filtered version (colored trace). (c) DAPI staining with reconstructed track of a 16‐channel silicon probe implanted in ipsilateral CA2 with the recording sites covering all layers. (d, top) Average amplitude profile from outer stratum lacunosum moleculare (slm, brown) to stratum pyramidale (sp, yellow) of CA2 for large epileptic spikes (mean of 340 spikes per mouse averaged across three mice). (d, bottom) Average current source density (CSD) profile based on traces in (d, top) revealed local sink‐source alternation across distal and proximal apical dendritic areas and across time. Inset shows the CSD time course at an individual electrode (*) in sr. The sink in sr is delayed by 8.5 ms to that in slm and reaches its maximum after 44 ms. CSDs were computed individually for each mouse (n = 3 mice), cropped to seven channels covering the area between slm and sp and averaged across mice. (e) Corresponding average amplitude profile and CSD profile of theta oscillations in CA2 (n = 2 mice, 3900 theta peaks per mouse). Scale bars in (a) left 2 mm, middle 100 μm, right 500 μm, in (c) 100 μm. so, stratum oriens; sp, stratum pyramidale; sr, stratum radiatum; slm, stratum lacunosum moleculare; ml, molecular layer; gcl, granule cell layer; pol, polymorphic layerEpileptic spikes of large amplitudes occurred in the ipsilateral and the contralateral DG, as previously described (Häussler et al., 2012; Janz et al., 2018). In ipsilateral CA2, EA occurred mostly concomitant with the ipsilateral DG whereas contralateral CA2 usually showed shorter EA with low amplitudes (Figure 1b, EA example). Accordingly, automatically detected EA was usually classified into less severe categories in the contralateral hippocampus.Theta oscillations were found bilaterally in the DG and CA2 (Figure 1b, θ example), but with lower power in the ipsilateral DG, as described previously (Kilias et al., 2018). To determine whether the LFPs of epileptic and theta activity recorded in CA2 were generated locally in CA2 or a volume‐conducted signal, we implanted 16‐channel silicon probes sampling all layers of CA2 (n = 3 mice, 21/22 days after KA, Figure 1c). We recorded spatio‐temporally resolved activity profiles (Figure 1d,e) and estimated CSD profiles of epileptic spikes (Figure 1d) and theta oscillations (Figure 1e).For large epileptic spikes detected in slm (Figure 1d, 21 days after KA, top trace, amplitude profile of average across n = 3 mice, 340 epileptic spikes per mouse, layers color‐coded), the averaged CSD profile revealed a strong dipole at the onset of the epileptic spike consisting of a profound current sink in slm at the position of distal dendrites opposed by a source in sl/sr. This was followed after ~9 ms by a strong sink in sl/sr which reached its maximum after ~44 ms and propagated toward sp, and a subsequent long‐lasting source of ~200 ms duration spanning all layers of CA2 (Figure 1d).During theta oscillations, an alternating sink‐source pattern was visible with a dipole which inverted its polarity spatially at the border of slm and sl/sr and temporally with a frequency of ~6.6 Hz (see Figure 4c, right) (Figure 1e, n = 2 mice). The strongest sink was visible in the proximal half of sl/sr, approximately the region where mossy fibers and CA2 recurrent connections terminate.Both patterns are in favor or a local origin of activity in CA2 instead of volume‐conducted potentials from neighboring regions. Together, our data show that CA2 in chronically epileptic mice (i.e., at a time point when CA2 pyramidal cell dispersion and mossy fiber sprouting have developed), actively contributes to EA and theta oscillations in the hippocampus.Preservation of the connectivity between ipsilateral CA2 and the contralateral hippocampus under epileptic conditionsWe hypothesized that CA2 with its interhemispheric connectivity might serve as a gateway for EA propagation. A prerequisite for this is the preservation of the projection of ipsilateral CA2 to the contralateral hippocampus under epileptic conditions. To test this, we injected an AAV carrying a floxed copy of tdTomato under the control of the synapsin promoter and EGFP bound to the vesicle protein synaptophysin into the CA2 region of Amigo2‐icreERT2 mice prior to KA injection and analyzed the fluorescently labeled projections in frontal and horizontal sections (Figure 2a). In control mice that received only the AAV injection (n = 1 mouse frontal, n = 2 mice horizontal), we detected tdTomato labeling mainly in CA2 somata and dendrites and occasionally in larger axons (Figure 2b,c,e). Synaptophysin‐bound EGFP labeling was strongest in axons and synapses, but also visible in somata of pyramidal cells. Co‐staining with a CA2‐specific marker (PCP4, data not shown) confirmed the specificity of the viral transduction. Within the hippocampus, projections to ipsilateral CA3, CA2, and CA1 and to the contralateral side, in particular to CA2 and CA1 were strongly labeled (Figure 2b frontal plane, Figure 2c,e horizontal plane). At 21 days after KA injection, labeling was overall similar, but the density of tdTomato‐labeled cells was lower, mainly due to dispersion of the CA2 pyramidal cell layer, as shown previously (Häussler et al., 2016) and some loss of CA2 pyramidal cells (Figure 2b,d,f; n = 1 mouse frontal, n = 5 mice horizontal). The projection from the dispersed CA2 to the adjacent CA1 region was strongly reduced in sections close to the injection site reflecting the loss of CA1 neurons in the sclerotic hippocampus at this position (Figure 2b, Supplementary Figure 3). Nevertheless, the projection to contralateral CA2, but also to contralateral CA3 and CA1 was preserved (Figure 2b–f, Supplementary Figure 3). We determined the DV extent of tdTomato‐labeled somata in the horizontal plane of ipsilateral CA2 as well as the most dorsal and most ventral position at which EGFP‐positive CA2 axons and synapses were visible in the ipsi‐ and contralateral hippocampus, respectively, for all mice. We did not find any salient differences between control and KA‐injected hippocampi (Figure 2g). Similarly, the layer‐resolved (so to sr) intensity profiles of EGFP‐expression in contralateral CA2 relative to the intensity maximum on the ipsilateral side (to account for the altered density of transduced cells) did not differ between KA and control (Figure 2h). In summary, our viral tracing experiments show that the projections of ipsilateral CA2 to the contralateral hippocampus persist after KA injection and offer a direct pathway for activity propagation from the ipsi‐ to the contralateral hippocampus.2FIGUREPreservation of CA2‐CA2 connectivity in epilepsy. (a) Scheme depicting the site of tracer injection and slicing angles. (b) Tracing with a cre‐dependent adeno‐associated virus (AAV) in the CA2‐specific Amigo2‐icreERT2 mouse line; frontal sections at the level of the injection site. Upper row: ipsi‐ and contralateral hippocampus of a control mouse. Lower row: epileptic mouse 21 days after kainate (KA) injection. Virally delivered tdTomato (red) is expressed in somata and dendrites of ipsilateral CA2, the axonal projections (but also local somata and dendrites) express synaptophysin‐bound EGFP (green). Note the dispersion of the GCL and cell loss in CA3 and CA1 (and the corresponding loss of CA2 projections to these regions) on the ipsilateral site after KA injection. The projection from ipsilateral to contralateral CA2 appears weaker but preserved in epileptic mice (arrows). (c) Representative control section in horizontal plane at the dorso‐ventral (DV) level of the injection site (injection in right hippocampus), tdTomato expression in somata and processes (red), synaptophysin‐bound EGFP (green) with 4′,6‐diamidino‐2‐phenylindole (DAPI) counterstaining (blue). (d) Same as in c but without DAPI. Small images: Ipsilateral CA2 at injection site and contralateral CA2 at corresponding level. Dotted line illustrates the extent along which optical density measurement (shown in h) has been performed. (e) Horizontal section of a KA‐injected brain 21 days after KA. The dispersion of the GCL is visible. (f) Same as in e but without DAPI. The projection of ipsi‐ to contralateral CA2 is preserved. Inset shows cells at the AAV injection site—note that this is taken from a section slightly more anterior than the large image to show location of maximal labeling of somata. (g) Qualitative analysis of DV extent of labeled cells in horizontal sections (for all sequential sections cut in horizontal plane; DV position according to Franklin and Paxinos “Atlas of the mouse brain”) from control mice (n = 2 mice) and KA‐injected Amigo2‐icreERT2 mice (n = 5 mice). The DV extent is depicted, along which labeled axons/synapses within the ipsilateral hippocampus (axons ipsi) and to the contralateral hippocampus (axons contra) were visible. (h) Mean relative intensity profiles measured from stratum oriens (so) to stratum radiatum (sr) are given for ipsi‐ and contralateral CA2 in control and KA‐injected mice for sections around the DV level of the KA/AAV injection site. Scale bars (b, c) 500 μm, small images in (e, f) 100 μm. GCL, granule cell layer, SO, stratum oriens; SP, stratum pyramidale; SR, stratum radiatumTime course of EA and coupling between DG and CA2 in both hemispheresGiven the active participation of ipsilateral dorsal CA2 in EA and its preserved contralateral projections, we next explored a potential role of CA2 as a gate for propagation of EA in the epileptic network. In particular cell loss in CA3 and CA1, the main up‐ and downstream regions of CA2, and the progressive development of mossy fiber sprouting onto CA2 somata during the first 2 weeks after KA (Figure 3a, schematic drawing; Häussler et al., 2016) raised the possibility of an enhanced coupling between CA2 and its remaining pre‐ and postsynaptic partners during EA.3FIGUREBilateral coupling of epileptic activity in the dentate gyrus and CA2. (a, top) Schematic drawing of both hippocampi before kainate (KA) injection (injection site indicated) and 2–3 weeks after KA injection when hippocampal sclerosis (cell loss mainly in CA3 and CA1 and astrogliosis), and the dispersion of dentate granule cells and CA2 pyramidal cells has taken place in the ipsilateral hippocampus. The contralateral hippocampus is spared from any major changes. (a, bottom) Time spent in epileptic activity (EA) as a fraction of total recording time for ipsi‐ and contralateral DG and CA2 measured over 21 days after the initial status epilepticus (SE). (b) For each time point and recording site, EA was classified into mild (b, left); moderate (b, middle); and severe (b, right) events and the fraction of time in each class relative to total EA time is given. (c, left) Likelihood of measuring EA of any type in ipsi‐ and contralateral CA2 while severe EA was detected in ipsilateral DG. (c, right) Likelihood of measuring EA of any type in ipsilateral DG or contralateral CA2 while severe EA was detected in ipsilateral CA2. Mean ± standard deviation (filled circles) and values for individual mice (small dots) are given for all analyses. DGi, ipsilateral dentate gyrus; DGc, contralateral dentate gyrus; SE, status epilepticusFirst, we compared the mean fraction of time spent in EA (all severity classes included) which was higher in the ipsilateral hippocampus than in the contralateral hippocampus at all time points (Figure 3a, n = 10 mice, time‐independent mean fraction ± SD DGi: 0.26 ± 0.13, CA2i: 0.31 ± 0.12, DGc: 0.19 ± 0.14, CA2c: 0.14 ± 0.11; difference in ratio between hemispheres in post hoc Δ = 0.10, p = .001). Interestingly, we did not find any major differences between the time points for the individual recording sites (main effect time p = .43), indicating that there was no or only a very short silent period after SE but instead EA occurred from early on.To determine whether EA changes in severity during the first weeks after KA injection, we next regarded the classification of EA into mild, moderate, and severe events, as described above, and determined the ratio of time spent in each class relative to total EA for all time points (Figure 3b). We found that mild events made up for a large part of total EA time early after KA injection (≤48 h), but diminished thereafter at all recording sites (Figure 3b; three‐way ANOVA for time, structure and hemisphere, post hoc: ≤48 h vs. 7 days p = 2.3 × 10−4, ≤48 h vs. 14 days p = .0014, ≤48 h vs. 21 days p = .61). However, in contralateral CA2, mild events made up for the largest fraction of EA time, which was significantly higher than in the ipsilateral hippocampus at all time points (DGi vs. CA2c p = .0012, CA2i vs. CA2c p = 2.7 × 10−5). For all other positions, the mild event rate did not differ. The share spent in moderate events was constant over time and comparable for all recording sites. In contrast, severe events were rare at ≤48 h at all recording sites and strongly increased until the 7 days time point, remained at this level at 14 days, but slightly dropped at 21 days (Figure 3b; post hoc time: ≤48 h vs. 7 days p = 2.2 × 10−5, ≤48 h vs. 14 days p = 2.3 × 10−4, ≤48 h vs. 21 days p = .74, 7 days vs. 21 p = .04). The fraction of severe events was always higher in the ipsilateral compared to the contralateral hippocampus, but did not differ between ipsilateral DG and CA2 (post hoc: ipsi vs. contra p = .0022). The high degree of similarity between EA classes in ipsilateral DGi and ipsilateral CA2 supports the idea of a strong EA coupling between these regions. The propagation to contralateral CA2, however, seems to be much less prominent and might depend on large events. To analyze this relationship, we inspected severe EA in the ipsilateral DG and tested for co‐occurring EA (independent of its class) in ipsi‐ or contralateral CA2. As hypothesized, the rate of coupling between DGi and CA2i was high (Figure 3c; mean fraction ± SD: 62 ± 19%) and despite ongoing mossy fiber sprouting, did not increase over time (two‐way ANOVA main effect time, p = .97). The coupling of EA between ipsilateral DG and contralateral CA2 was much weaker (34 ± 25%, difference of DGi to CA2i vs. CA2c Δ = 30.26%, p = 2.9 × 10−4) but also constant over time.Interestingly, when testing in opposite direction taking severe EA in ipsilateral CA2 as a basis, the chance to observe simultaneous EA in the ipsilateral DG was initially lower (34 ± 14%) but increased during the first 2 weeks to 66 ± 7%, a level comparable to the inverse coupling from the ipsilateral DG to ipsilateral CA2 (Figure 3c). The coupling from ipsi‐ to contralateral CA2 was overall lower (DGi vs. CA2c Δ = 24.63%, two‐way ANOVA post hoc: p = 9.5 × 10−5), but also increased over the time course of 3 weeks by ~23% (main effect time 0.004, post hoc ≤48 h vs. 21 days Δ = 27.34% structures pooled, p = .019).In summary, we found stronger coupling of EA within the ipsilateral hippocampus which was stable over time when taking the DG as basis. When taking ipsilateral CA2 as basis, both, coupling to the DG and to contralateral CA2 increased over time, suggesting a dynamic reorganization of EA sources in CA2.Theta frequency reduction and coherence in ipsi‐ and contralateral CA2Given the temporal reorganization of EA sources in CA2 and the rather weak coupling of EA with the contralateral side, we asked next how CA2 is embedded into the hippocampal network during theta oscillations. We have previously shown that despite chronic MTLE theta oscillations are prominent in the DG and medial entorhinal cortex, but occur at lower prevalence compared to controls (Kilias et al., 2018).When identifying theta periods of 9 s or longer, we found that the fraction of theta periods relative to total recording duration did not significantly change over time after KA injection (Figure 4a, repeated measure ANOVA, time p = .64). However, when we calculated the autocorrelation and spectra of theta periods, there was no clear peak in the theta range in several recording sessions (Figure 4b top). In particular at ≤48 h after SE, a large fraction of recording sites in the ipsilateral hippocampus did not show clear oscillatory theta activity, whereas there was an identifiable peak on the contralateral side (Supplementary Table 1; three‐way ANOVA for time point, targeted structure and hemisphere of recordings without identifiable theta peak, post hoc ≤48 h ipsilateral was different to all other time points and sites: ipsi/contra at 7 days p = .027/.014, 14 days p = .012/.013, 21 days p = .013/.0064). The remaining recording sessions were characterized by prominent secondary maxima in the autocorrelation at 83–200 ms lag (Figure 4b, middle) and a spectral power peak in the range of 5–12 Hz (Figure 4b, bottom). Only sessions fulfilling these criteria were used for further analysis.4FIGURETheta oscillations in the DG‐CA2 network of both hemispheres. (a) The average (mean across animals) proportional time covered by all theta periods (≥9 s) at different time points after kainate (KA) does not significantly change over time. (b, top row) Fractions of recordings without a prominent secondary maximum in the autocorrelation or a spectral power peak corresponding to theta range. For each channel, the fraction is given relative to the total available recordings at this particular site and time point. The remaining data sets showed prominent peaks in the autocorrelation at 83–200 ms lag (b, middle) and spectral peaks in the range of 5–12 Hz (b, bottom) for all time points. (c) Dominant theta frequency for ipsi‐ and contralateral DG and CA2 for all time points. Theta frequency is stable across time, comparable over structures and not different from previously published data from KA‐injected chronically epileptic animals (KAREF, Kilias et al., 2018; mean ± standard deviation (SD) of the earlier data set shown as green dashed line and shaded area, respectively) but to data from NaCl‐injected animals (CtrREF, Kilias et al., 2018; mean ± SD in gray). (d) Coherence for electrode pairs of DGi, CA2i, and CA2c for each time point (mean ± SD as line and shadow in corresponding colors). Coherence peaks were found in the theta range for all three possible combinations of recording sites and at all time points. The number of data sets for each combination and time point is indicated in the plot.Next, we determined the dominant frequency within the theta band. We have shown earlier that theta oscillations, irrespective of the underlying behavior, are by ~1 Hz lower in the DG and medial entorhinal cortex under epileptic conditions (Kilias et al., 2018). Since this did not depend on local reorganization of the DG and since the hippocampus is a system of coupled oscillators (Vertes, 2005), we expected a proportional reduction in theta frequency in CA2 and indeed found a predominant frequency of 6.43 ± 0.35 Hz. This frequency range was not significantly different from our previous dataset obtained in the DG of chronically epileptic KA‐injected mice (Kilias et al., 2018; Figure 4c, all recordings of DG and CA2 pooled, one‐sample Kolmogorov–Smirnov test, p = .06). It was, however, substantially lower than the theta oscillations found in equivalent NaCl‐injected controls (Kilias et al., 2018: 7.54 ± 0.9 Hz, p = 1.75 × 10−14). Interestingly, the reduction in theta frequency was not only invariant across hippocampal regions, but also stable over time (repeated measure ANOVA, time p = .44). Thus, the reduced frequency is not a consequence of the progressive reorganization of the hippocampal formation but an immediate effect of SE‐associated processes.Our CSD analysis revealed an alternation of strong sinks and sources within sr of CA2, the target zone of the mossy fibers. Given the progressive mossy fiber sprouting during the first weeks after SE, we asked whether the coherence in theta band between DG and CA2 changes over time. We found a coherence peak between the ipsilateral DG and ipsi‐ and contralateral CA2 at all time points tested in the first 3 weeks after SE (Figure 4d). The previously described absence of theta oscillations in ~67% of the ipsilateral recordings at ≤48 h but an identifiable oscillation on the contralateral electrodes implies a loss of coherence to contralateral theta (Figure 4b). Nevertheless, when theta oscillations were present in DG and CA2, their coherence peak frequency was not significantly different from the mean of their individual frequencies (Figure 4d). At 7and 21 days, coherent oscillations were slightly faster than the observed individual theta oscillations at the individual recording sites (7 days, mean theta of DGi and CA2i—Δf = 0.4 Hz, p = .0002; 21 days DGi—CA2i Δf = 0.38 Hz, p = .04; 21 days CA2i—CA2c Δf = 0.58 Hz, p = .0052, one‐sample t test with the mean frequency of the individual electrodes as hypothesized mean), but never large enough to compensate for the reduction in frequency relative to controls seen before (Δf = 1.04 Hz, Kilias et al., 2018).In summary, we found that CA2 is fully integrated in the hippocampal theta oscillation network and that the overarching reduction of theta frequency happens early after SE.Loss of GABAergic neurons in ipsilateral CA2 onlyFinally, we asked whether local changes in inhibition could contribute to the strong transmission probability of EA from ipsilateral DG to CA2 early after SE, although mossy fiber sprouting becomes prominent only at 14 days after KA (Häussler et al., 2016). Furthermore, the increasing transmission from ipsi‐ to contralateral CA2 might be promoted by a progressive interneuron reduction in contralateral CA2. To test this, we analyzed whether the density of Gad67 mRNA‐expressing interneurons changes with time after KA injection. After locating the position of CA2 in PCP4 immunostaining (Figure 5a,b), we quantified the density of GABAergic cells in all layers of ipsi‐ and contralateral CA2 in a FISH for Gad67 mRNA. In NaCl‐injected controls, the ipsi‐ and contralateral hippocampus did not differ indicating that all effects were KA‐specific (Figure 5c,i–k). In contralateral CA2 the density of Gad67 mRNA‐expressing cells was comparable to control in all layers of CA2 at all time points (2, 7, 14, 21 days after KA; Figure 5g,i–k). In ipsilateral CA2, the density was strongly reduced compared to the contralateral hippocampus and to control in so, sp, and slm/sr/sl already at 2 days after KA and persistent for all time points (n = 4–6 mice per group; so: two‐way ANOVA for treatment/time point p < .001, hemisphere p < .0001, post hoc ipsi: NaCl vs. 2—14 days p < .0001, NaCl vs. 21 days p = .0065; post hoc for hemispheres: 2—14 days p < .0001; sp: two‐way ANOVA for treatment/time point p < .001, hemisphere p < .0001, post hoc ipsi: NaCl vs. 2—21 days p < .0001; post hoc for hemispheres: 2—21 days p < .0001; slm/sr/sl: two‐way ANOVA for treatment/time point p < .001, hemisphere p < .0001, post hoc ipsi: NaCl vs. 2—21 days p < .0001; post hoc for hemispheres: 2—14 days p < .0001, 21 days p = .0003). Interestingly, there was a tendency for higher interneuron density at 21 days after KA compared to all earlier time points after KA.5FIGUREGABAergic interneurons in CA2. (a, b) Representative images of Purkinje cell protein 4 (PCP4) immunostaining in sections of the dorsal hippocampus to localize CA2. (a) NaCl‐injected control. Granule cells and CA2 pyramidal cells are PCP4‐positive. (b) Ipsilateral hippocampus, 21 days after KA injection. PCP4‐positive granule cells and CA2 pyramidal cells are dispersed (arrows). (c–h) Representative images of a fluorescence in situ hybridization (FISH) for glutamic acid decarboxylase 67 (Gad67) mRNA displays GABAergic interneurons. (c) NaCl‐injected control, same section as in (a). GABAergic interneurons are abundant throughout the slice and densely placed in CA2. (d) Ipsilateral hippocampus, 2 days after KA. A strong loss of GABAergic interneurons is visible, with some preserved interneurons in CA3/CA2. (e) Ipsilateral hippocampus, 7 days after KA. Similar to 2 days. (f) Ipsilateral hippocampus, 14 days after KA. A progressive dispersion of the GCL is visible, some GABAergic cells are preserved in CA3/CA2. (g) Contralateral hippocampus, 21 days after KA. The contralateral hippocampus is comparable to control at all time points. Occasionally, some interneuron loss occurs in distal CA1. (h) Ipsilateral hippocampus, 21 days after KA, same slice as in (b). In some mice, more Gad67 mRNA‐expressing cells are visible in CA3/CA2 than at earlier time points. (i–k) Layer‐resolved quantification of the density of Gad67 mRNA‐expressing cells in the dorsal CA2 region (close to the injection site) in NaCl‐injected controls and for all time points after KA injection (in cells/mm2). Mean ± standard deviation and individual values per animal (means of 2–4 sections per mouse) are given. Values for the animals chosen for display in (a–h) are indicated with a green circle. (i) Stratum oriens (so). Two‐way analysis of variance (ANOVA): treatment/time point p < .001, hemisphere p < .0001. (j) Stratum pyramidale (sp). Two‐way ANOVA for treatment/time point p < .001, hemisphere p < .0001. (k) Strata lacunosum moleculare, radiatum, and lucidum (slm/sr/sl). Two‐way ANOVA: treatment/time point p < .001, hemisphere p < .0001. Comparison for treatment/time point was made with a Tukey's multiple comparison test: **p < .01, ***p < .001; pairwise comparison was made with a Šidák's test: ###p < .001. Scale bar for (a–h) 100 μm. GCL, granule cell layer; CA2, cornu ammonis 2; slm/sr/sl, stratum lacunosum moleculare/stratum radiatum/stratum lucidum; sp, stratum pyramidale; so, stratum oriensTogether these data indicate the loss of local Gad67 mRNA expression exclusively in ipsilateral CA2 immediately after KA injection with an eventual small recovery in the chronic stage.DISCUSSIONThe degree of preservation of CA2 has traditionally been used among other criteria for classification of hippocampal sclerosis in MTLE (Blümcke et al., 2013; Thom, 2014; Wyler et al., 1992). It is, however, less well known how CA2 is integrated into the epileptic network. In the present study we have used the intrahippocampal KA mouse model for MTLE to shine light on structural and functional determinants of the integration of CA2 in the epileptic network of both hippocampi. We show that EA is generated locally in CA2 and CSD analysis of EA in CA2 revealed two spatially and temporally segregated sinks in distal and proximal dendrites of CA2 pyramidal cells. The projections from the ipsilateral CA2 region to the contralateral hippocampus were preserved despite hippocampal sclerosis; however, in the contralateral hippocampus, the fraction of time spent in EA and the severity of events was substantially lower than in the ipsilateral hippocampus. In accordance with this hemispheric difference, our time‐course analysis of Gad67‐positive cell density showed a massive reduction of cells already at 2 days after KA only in ipsilateral CA2. When investigating the coupling of EA, we found that the ipsilateral coupling was always stronger than with contralateral CA2. On the ipsilateral side, severe events in the DG were reliably accompanied by EA in CA2 but in the converse direction, the EA overlap increased over time.Analysis of theta episodes revealed that CA2 actively participates in theta rhythm and shows the same frequency reduction observed in other hippocampal areas.Changes in CA2 connectivity and structure under epileptic conditionsEarlier we have shown that the length of the CA2 region was increased at 14 days after KA injection (but not at 7 days after KA) and remained constant afterward, indicating a transient phase of dispersion around 14 days (Häussler et al., 2016). Similarly, sprouted mossy fiber terminals which target CA2 pyramidal cell somata can be observed from ~14 days onward. When analyzing the expression of Bdnf mRNA, a marker for structural and functional plasticity mechanisms under epileptic conditions (reviewed in Binder et al., 2001), we found the upregulation at 2 and 14 days after KA injection but not at 7 and 21 days (Tulke et al., 2019). Together these data point to an early, most likely SE‐induced and a later, transient phase of high structural plasticity in CA2.To further analyze this, we here performed a time‐resolved analysis of the density of Gad67 mRNA‐expressing neurons in CA2 and found at strong reduction already 2 days after KA injection similar to what we have observed in the hilus and entire hippocampus (Marx et al., 2013), which indicates that SE has a devastating effect on interneurons also locally in CA2. Interestingly, there seems to be a partial recovery of Gad67 mRNA‐expressing neurons which was restricted to CA2 and distal CA3 at 21 days. There was, thus, no complete loss of GABAergic interneurons in CA2 after KA injection, but instead some interneurons might have only temporally lost the expression of the GABA synthesizing enzyme and therefore probably also the ability for GABAergic transmission. These data match earlier observations using magnetic resonance spectroscopy where the recovery of the GABA concentration in the hippocampus has been shown at >16 days after KA injection after a transient loss of GABA in the hippocampus (Hamelin et al., 2021; Janz, Schwaderlapp, et al., 2017). Although there is the very unlikely option that the loss of GABAergic interneurons is followed by migration of interneurons from other areas, it is much more plausible that some interneurons undergo recovery processes and restart the transcription of Gad67 mRNA and accordingly the production of GABA. This would be in agreement with the concept of dormant interneurons (Sloviter, 1991) or transient reduction of expression of markers for inhibitory neurons in humans (Wittner et al., 2001). It remains to be determined whether our observations on the mRNA level are accompanied by changes in the synaptic input to CA2 pyramidal cells. Together with our previous studies, our current results support the idea of an early phase of severe damage, followed by a second phase of structural alterations inducing aberrant connectivity between 7 and 14 days and eventually a third phase of partial recovery with the aim to reach the “least epileptic state possible.” Interestingly, this also matches the time course of severe EA described here and the time course of epileptic spike rate and burst length in an earlier study (Janz et al., 2018).Concerning the outbound connectivity of CA2 pyramidal cells, we found a reduction of projections from ipsilateral CA2 to ipsilateral CA3/CA1 in agreement with cell loss in these areas, but the preservation of projections to contralateral CA3, CA2, and CA1 at 21 days after KA injection. Importantly, the DV extent of ipsi‐ and contralateral hippocampal projections was comparable to control, indicating that despite the local changes, CA2 pyramidal cells remained part of a network beyond the ipsilateral hippocampus. Certainly, our current study does not provide a comprehensive analysis of all CA2 projections under epileptic conditions since those to the subiculum (own unpublished observations), to ventral CA1 (Meira et al., 2018), to the septum (Leroy et al., 2018) and other targets (Cui et al., 2013) which might also play an important role for EA propagation were not analyzed here.Active role of CA2 in epilepsyThere are strong indications that CA2 plays an active role in MTLE since increased intrinsic excitability as well as increased excitatory input to and output from CA2 have been shown in the pilocarpine mouse model (Whitebirch et al., 2022). Similarly, in hippocampal slices from human specimens resected during curative MTLE surgery CA2 generated spontaneous interictal activity (Knowles et al., 1987; Reyes‐Garcia et al., 2018; Wittner et al., 2009). Our layer‐resolved silicon probe recordings are in line with these observations showing local EA in CA2. We observed a conspicuous pattern of sink‐source‐distribution in epileptic spikes in the CSD analysis: two successive sinks in slm and sl/sr point toward distinct inputs arriving in sequence. We hypothesize that the sink in slm reflects input from the medial entorhinal cortex which arrives in the DG and in CA2 nearly simultaneously and activates both. Under normal conditions, entorhinal input to CA2 is strong and highly efficient in inducing action potentials (Chevaleyre & Siegelbaum, 2010; Srinivas et al., 2017; Sun et al., 2014) and entorhinal input nearly simultaneously induced field responses in the DG and CA2 but not in CA3 or CA1 (Bartesaghi & Gessi, 2004). Anterograde tracing from the medial entorhinal cortex confirmed the preservation of these synaptic inputs to CA2 in the intrahippocampal KA model (Janz, Savanthrapadian, et al., 2017). We can only speculate on their efficiency under epileptic conditions, but sprouting of entorhinal projections, formation of novel synapses and strengthening of a subset thereof in the DG of the intrahippocampal KA model (Janz, Savanthrapadian, et al., 2017) are in favor of preservation or even strengthening of projections to CA2, too.The simultaneous activation of granule cells might lead to subsequent propagation of DG activity via the mossy fibers to CA2 which is reflected in a current sink in sl/sr in CA2 following the sink in slm with some delay. Under normal conditions mossy fiber transmission onto CA2 is weak and mostly activates feed‐forward inhibition (Kohara et al., 2014; Sun et al., 2017). However, considering the reduction of GABAergic interneurons shown here and the sprouting of mossy fibers in CA2 (Häussler et al., 2016), a strengthening of this input, similar to the pilocarpine model (Whitebirch et al., 2022), is conceivable. A contribution of input from CA3 to this second peak is unlikely due to the strong loss of CA3 pyramidal cells, but it cannot be excluded, in particular with respect to input from more temporal or contralateral CA3 where tissue preservation is better. Similarly, we cannot rule out a component from the recurrent CA2 connections (Okamoto & Ikegaya, 2019). The test of our hypotheses would require simultaneous recordings in the DG and CA2 with placement of probes parallel to the cells' orientation for allow for more detailed analysis of sequential inputs.We were wondering why severe EA in the DG had a high likelihood to occur together with EA in CA2 from early on, whereas the opposite coupling increased over time. Taking into account that we implanted silicon probes only in late stages to avoid strong tissue reorganization, we can only hypothesize on underlying mechanisms and recordings of the whole time course with silicon probes will be necessary in the future. The reliable transmission from DG to CA2 from early on is in agreement with the direct connectivity of the two regions via the mossy fibers together with the early loss of GABAergic interneurons—and with that most likely also the loss of feed‐forward inhibition which suppresses this transmission under healthy conditions (Kohara et al., 2014). The time course of the increase in coupling in the opposite direction—from CA2 to DG—matches the time course of ongoing sprouting processes, including mossy fiber sprouting (Häussler et al., 2016) and sprouting of perforant path synapses onto the DG (Janz, Savanthrapadian, et al., 2017)—and possibly onto CA2, too. In addition, CA2 projections which lose their target cells in CA1 might also sprout to the remaining ipsilateral targets in a similar time frame (own unpublished observations). Together, this may develop a network of high interconnectivity and recurrence, which favors reciprocal synchronization.Concerning the role of CA2 as a possible gate for propagation of EA, we found that EA in contralateral CA2 is less severe at all time points. This indicates that the preservation of GABAergic cells in contralateral CA2 is efficient in dampening EA there. Yet, more severe events from the ipsilateral side might have a higher probability to propagate to the contralateral side because it has been shown that repeated stimulation of the inputs to CA2 results in long‐term depression of inhibitory neurons in the CA2 region (Nasrallah et al., 2017). This could explain the increase of propagation from ipsi‐ to contralateral CA2 between 2 and 7 days, when the incidence of more severe events increases on the ipsilateral side but the later increase. Although we did not observe an increase in the density of CA2‐CA2 projections under epileptic conditions—which might be due to a slight reduction of CA2 pyramidal cells under epileptic conditions and thus a lower density of virus‐expressing cells in ipsilateral CA2, to different virus expression properties or to altered tamoxifen effects under epileptic conditions—we cannot exclude that sprouting of projections from ipsi‐ to contralateral CA2 might take place due to the loss of ipsilateral targets. It remains to be determined whether the balance of ipsilateral projections to contralateral CA2 in changed in favor of excitation. In addition, our data do not rule out the propagation via other pathways, for example, via the few remaining CA3 cells in the dorsal hippocampus, via ventral CA3, via mossy cells from the ventral hippocampus, where they are partially preserved (Volz et al., 2011) or other polysynaptic pathways.Non‐EA in the epileptic networkIn the earlier work, we have shown a reduction in theta frequency by more than 1 Hz in chronic epilepsy (>16 days after KA) in the DG of both hemispheres and in the medial entorhinal cortex (Kilias et al., 2018). Here, we have included the CA2 region of both hippocampi and expanded our analysis to earlier time points to see whether it is a network‐wide phenomenon and to determine whether structural reorganization within the hippocampal formation might contribute to the slowdown of the rhythm. The frequency reduction at all positions was already visible at ≤48 h and persisted without any changes in theta peak frequency throughout the experiment, ruling out sprouting as a potential cause. Cell death in the hippocampal formation, in particular loss of interneurons as the predominant target of theta‐rhythmic inputs, has nearly completely developed at this time point (Bouilleret et al., 1999; Franz et al., 2022; Marx et al., 2013; Suzuki et al., 1995). As shown here for CA2, the loss of interneurons is predominantly a focal phenomenon sparing the contralateral hippocampus, making an early damage of extrahippocampal theta pacemakers, like the medial septum during/after SE a more plausible explanation. The early reduction is in agreement with a theta frequency reduction at 4 days after pilocarpine injection in rats which is further supporting its independence from local hippocampal sclerosis because cell loss in this systemic model is uniform across hemispheres and restricted to the hilus and only parts of CA1 and CA3 (Chauvière et al., 2009).Interestingly, it was not possible to detect distinct theta periods in the ipsilateral hippocampus of two‐thirds of the mice at ≤48 h after KA injection, but theta recovered in the following. We assume that rather SE‐induced short‐term effects on intracellular properties (LeDuigou et al., 2005) might account for this temporary loss of theta rhythmicity than long‐lasting changes in cell resonance characteristics (Marcelin et al., 2009).Despite a clear theta peak at all recording sites at 14 days after KA, we found that the coherence peak in theta range was less pronounced than at all other time points, in particular for coupling of ipsilateral DG and CA2. Interestingly, this coincides with the time point of strongest synaptic reorganization and Bdnf‐associated plasticity in CA2 (Häussler et al., 2016; Tulke et al., 2019). It is therefore conceivable that projections from theta pacemakers also undergo intermittent plasticity processes due to loss of their hippocampal targets. Indeed, neurons from the medial septum or the supramammillary nucleus (SUM) strongly project onto GABAergic interneurons in CA2 and induce transient disinhibition (GABAergic septal projections) or feed‐forward inhibition (glutamatergic projections from SUM; Robert et al., 2021) of CA2 pyramidal cells. The loss of GABAergic target cells might induce sprouting and synapse formation onto CA2 pyramidal cells which might strongly disturb the balanced system of inputs that guarantees theta oscillations under normal conditions.Although there is no direct evidence for behavioral or mnemonic impairment associated with a slowed theta rhythm under epileptic conditions (Shuman et al., 2017), theta stimulation has been reported to have beneficial effects on spatial learning (Lee et al., 2017). Thus, given that CA2 firing is strongly coupled to theta rhythm (Oliva et al., 2016), social cognition deficits observed in patients (Bora & Meletti, 2016) and TLE models (Mikulecká et al., 2019) could relate to the observed rhythmopathy in CA2.In summary, our data show that CA2 is an active part of the epileptic network that might contribute to the propagation of EA to brain areas beyond the sclerotic hippocampus. Furthermore, we propose that early SE‐induced cell death plays an important role for epileptic and theta activity not only in CA2 but also in the entire hippocampal formation and is followed by a phase of compensatory sprouting, reflected in changes of reciprocal interaction of DG and CA2.ACKNOWLEDGMENTSThe authors are grateful to Prof Dr Carola A. Haas (Medical Center—University of Freiburg) for providing valuable input, lab equipment and materials. The Amigo2‐icreERT2 mouse line was a kind gift from Dr Serena Dudek and Dr Georgia Alexander (NIEHS, Durham, NC). The authors thank Tobias Holzhammer for the assembly of silicon probes with polyimide‐based ribbon cables. The work was funded by the German Research Foundation (DFG grant HA7597) and as a part of the Cluster of Excellence “BrainLinks‐BrainTools” within the framework of the German Excellence Initiative (grant number EXC1086 to U.H. and P.R.), by BrainLinks‐BrainTools, which is funded by the Federal Ministry of Economics, Science and Arts of Baden‐Württemberg within the sustainability program for projects of the excellence initiative II, and a grant from the Research Commission of the Medical Faculty, University of Freiburg. NB received a PhD scholarship from the Center for Basics in Neuromodulation, Freiburg. Open Access funding enabled and organized by Projekt DEAL.DATA AVAILABILITY STATEMENTThe data that support the findings of this study are available from the corresponding author upon request.REFERENCESAlexander, G. M., Brown, L. Y., Farris, S., Lustberg, D., Pantazis, C., Gloss, B., Plummer, N. W., Jensen, P., & Dudek, S. M. (2018). CA2 neuronal activity controls hippocampal low gamma and ripple oscillations. eLife, 7, e38052. https://doi.org/10.7554/eLife.38052Althaus, A. L., Zhang, H., & Parent, J. M. (2016). Axonal plasticity of age‐defined dentate granule cells in a rat model of mesial temporal lobe epilepsy. Neurobiology of Disease, 86, 187–196. https://doi.org/10.1016/j.nbd.2015.11.024Bartesaghi, R., & Gessi, T. (2004). Parallel activation of field CA2 and dentate gyrus by synaptically elicited perforant path volleys. Hippocampus, 14, 948–963. https://doi.org/10.1002/hipo.20011Binder, D. K., Croll, S. D., Gall, C. M., & Scharfman, H. E. (2001). BDNF and epilepsy: Too much of a good thing? Trends in Neurosciences, 24, 47–53. https://doi.org/10.1016/S0166-2236(00)01682-9Blümcke, I., Thom, M., Aronica, E., Armstrong, D. D., Bartolomei, F., Bernasconi, A., Bernasconi, N., Bien, C. G., Cendes, F., Coras, R., Cross, J. H., Jacques, T. S., Kahane, P., Mathern, G. W., Miyata, H., Moshé, S. L., Oz, B., Özkara, Ç., Perucca, E., … Spreafico, R. (2013). International consensus classification of hippocampal sclerosis in temporal lobe epilepsy: A task force report from the ILAE Commission on Diagnostic Methods. Epilepsia, 54, 1315–1329. https://doi.org/10.1111/epi.12220Bora, E., & Meletti, S. (2016). Social cognition in temporal lobe epilepsy: A systematic review and meta‐analysis. Epilepsy & Behavior, 60, 50–57. https://doi.org/10.1016/j.yebeh.2016.04.024Botcher, N. A., Falck, J. E., Thomson, A. M., & Mercer, A. (2014). Distribution of interneurons in the CA2 region of the rat hippocampus. Frontiers in Neuroanatomy, 8, 104. https://doi.org/10.3389/fnana.2014.00104Bouilleret, V., Ridoux, V., Depaulis, A., Marescaux, C., Nehlig, A., & Le Gal La Salle, G. (1999). Recurrent seizures and hippocampal sclerosis following intrahippocampal kainate injection in adult mice: Electroencephalography, histopathology and synaptic reorganization similar to mesial temporal lobe epilepsy. Neuroscience, 89, 717–729. https://doi.org/10.1016/S0306-4522(98)00401-1Carstens, K. E., Lustberg, D. J., Shaughnessy, E. K., McCann, K. E., Alexander, G. M., & Dudek, S. M. (2021). Perineuronal net degradation rescues CA2 plasticity in a mouse model of Rett syndrome. The Journal of Clinical Investigation, 131(16), e137221. https://doi.org/10.1172/JCI137221Carstens, K. E., Phillips, M. L., Pozzo‐Miller, L., Weinberg, R. J., & Dudek, S. M. (2016). Perineuronal nets suppress plasticity of excitatory synapses on CA2 pyramidal neurons. The Journal of Neuroscience, 36, 6312–6320. https://doi.org/10.1523/JNEUROSCI.0245-16.2016Chauvière, L., Rafrafi, N., Thinus‐Blanc, C., Bartolomei, F., Esclapez, M., & Bernard, C. (2009). Early deficits in spatial memory and theta rhythm in experimental temporal lobe epilepsy. The Journal of Neuroscience, 29, 5402–5410. https://doi.org/10.1523/JNEUROSCI.4699-08.2009Chevaleyre, V., & Siegelbaum, S. A. (2010). Strong CA2 pyramidal neuron synapses define a powerful disynaptic cortico‐hippocampal loop. Neuron, 66, 560–572. https://doi.org/10.1016/j.neuron.2010.04.013Cui, Z., Gerfen, C. R., & Young, W. S. (2013). Hypothalamic and other connections with dorsal CA2 area of the mouse hippocampus. The Journal of Comparative Neurology, 521, 1844–1866. https://doi.org/10.1002/cne.23263Engel, J. (2001). A proposed diagnostic scheme for people with epileptic seizures and with epilepsy: Report of the ILAE Task Force on classification and terminology. Epilepsia, 42, 796–803. https://doi.org/10.1046/j.1528-1157.2001.10401.xFranklin, K. B. J., & Paxinos, G. (2007). The mouse brain in stereotaxic coordinates (3rd ed.). Academic Press.Franz, J., Barheier, N., Tulke, S., Haas, C. A., & Häussler, U. (2022). Differential vulnerability of neuronal subpopulations of the subiculum in a mouse model for mesial temporal lobe epilepsy. Biorxiv:2022.06.02.494518. https://doi.org/10.1101/2022.06.02.494518Freiman, T. M., Häussler, U., Zentner, J., Doostkam, S., Beck, J., Scheiwe, C., Brandt, A., Haas, C. A., & Puhahn‐Schmeiser, B. (2021). Mossy fiber sprouting into the hippocampal region CA2 in patients with temporal lobe epilepsy. Hippocampus, 31, 580–592. https://doi.org/10.1002/hipo.23323Hamelin, S., Stupar, V., Mazière, L., Guo, J., Labriji, W., Liu, C., Bretagnolle, L., Parrot, S., Barbier, E. L., Depaulis, A., & Fauvelle, F. (2021). In vivo γ‐aminobutyric acid increase as a biomarker of the epileptogenic zone: An unbiased metabolomics approach. Epilepsia, 62(1), 163–175. https://doi.org/10.1111/epi.16768Häussler, U., Bielefeld, L., Froriep, U. P., Wolfart, J., & Haas, C. A. (2012). Septotemporal position in the hippocampal formation determines epileptic and neurogenic activity in temporal lobe epilepsy. Cerebral Cortex, 22, 26–36. https://doi.org/10.1093/cercor/bhr054Häussler, U., Rinas, K., Kilias, A., Egert, U., & Haas, C. A. (2016). Mossy fiber sprouting and pyramidal cell dispersion in the hippocampal CA2 region in a mouse model of temporal lobe epilepsy. Hippocampus, 26, 577–588. https://doi.org/10.1002/hipo.22543Hayani, H., Song, I., & Dityatev, A. (2018). Increased excitability and reduced excitatory synaptic input into fast‐spiking CA2 interneurons after enzymatic attenuation of extracellular matrix. Frontiers in Cellular Neuroscience, 12, 149. https://doi.org/10.3389/fncel.2018.00149Heining, K., Kilias, A., Janz, P., Häussler, U., Kumar, A., Haas, C. A., & Egert, U. (2019). Bursts with high and low load of epileptiform spikes show context‐dependent correlations in epileptic mice. eNeuro, 6(5), ENEURO.0299–ENEU18.2019. https://doi.org/10.1523/ENEURO.0299-18.2019Herwik, S., Kisban, S., Aarts, A. A. A., Seidl, K., Girardeau, G., Benchenane, K., Zugaro, M. B., Wiener, S. I., Paul, O., Neves, H. P., & Ruther, P. (2009). Fabrication technology for silicon‐based microprobe arrays used in acute and sub‐chronic neural recording. Journal of Micromechanics and Microengineering, 19, 074008. https://doi.org/10.1088/0960-1317/19/7/074008Herwik, S., Paul, O., & Ruther, P. (2011). Ultrathin silicon chips of arbitrary shape by etching before grinding. Journal of Microelectromechanical Systems, 20, 791–793. https://doi.org/10.1109/JMEMS.2011.2148159Hjorth‐Simonsen, A., & Laurberg, S. (1977). Commissural connections of the dentate area in the rat. The Journal of Comparative Neurology, 174(4), 591–606. https://doi.org/10.1002/cne.901740404Houser, C. R., Miyashiro, J. E., Swartz, B. E., Walsh, G. O., Rich, J. R., & Delgado‐Escueta, A. V. (1990). Altered patterns of dynorphin immunoreactivity suggest mossy fiber reorganization in human hippocampal epilepsy. The Journal of Neuroscience, 10, 267–282. https://doi.org/10.1523/JNEUROSCI.10-01-00267.1990Janz, P., Hauser, P., Heining, K., Nestel, S., Kirsch, M., Egert, U., & Haas, C. A. (2018). Position‐ and time‐dependent arc expression links neuronal activity to synaptic plasticity during epileptogenesis. Frontiers in Cellular Neuroscience, 12, 244. https://doi.org/10.3389/fncel.2018.00244Janz, P., Savanthrapadian, S., Häussler, U., Kilias, A., Nestel, S., Kretz, O., Kirsch, M., Bartos, M., Egert, U., & Haas, C. A. (2017). Synaptic remodeling of entorhinal input contributes to an aberrant hippocampal network in temporal lobe epilepsy. Cerebral Cortex, 27, 2348–2364. https://doi.org/10.1093/cercor/bhw093Janz, P., Schwaderlapp, N., Heining, K., Häussler, U., Korvink, J. G., von Elverfeldt, D., Hennig, J., Egert, U., LeVan, P., & Haas, C. A. (2017). Early tissue damage and microstructural reorganization predict disease severity in experimental epilepsy. eLife, 6, 25742. https://doi.org/10.7554/eLife.25742Kilias, A., Häussler, U., Heining, K., Froriep, U. P., Haas, C. A., & Egert, U. (2018). Theta frequency decreases throughout the hippocampal formation in a focal epilepsy model. Hippocampus, 28, 375–391. https://doi.org/10.1002/hipo.22838Knowles, W. D., Traub, R. D., & Strowbridge, B. W. (1987). The initiation and spread of epileptiform bursts in the in vitro hippocampal slice. Neuroscience, 21, 441–455. https://doi.org/10.1016/0306-4522(87)90134-5Kohara, K., Pignatelli, M., Rivest, A. J., Jung, H.‐Y., Kitamura, T., Suh, J., Frank, D., Kajikawa, K., Mise, N., Obata, Y., Wickersham, I. R., & Tonegawa, S. (2014). Cell type‐specific genetic and optogenetic tools reveal hippocampal CA2 circuits. Nature Neuroscience, 17, 269–279. https://doi.org/10.1038/nn.3614Kulik, A., Vida, I., Luján, R., Haas, C. A., López‐Bendito, G., Shigemoto, R., & Frotscher, M. (2003). Subcellular localization of metabotropic GABA(B) receptor subunits GABA(B1a/b) and GABA(B2) in the rat hippocampus. The Journal of Neuroscience, 23, 11026–11035. https://doi.org/10.1523/JNEUROSCI.23-35-11026.2003Lanzilotto, M., Livi, A., Maranesi, M., Gerbella, M., Barz, F., Ruther, P., Fogassi, L., Rizzolatti, G., & Bonini, L. (2016). Extending the cortical grasping network: Pre‐supplementary motor neuron activity during vision and grasping of objects. Cerebral Cortex, 26, 4435–4449. https://doi.org/10.1093/cercor/bhw315LeDuigou, C., Wittner, L., Danglot, L., & Miles, R. (2005). Effects of focal injection of kainic acid into the mouse hippocampus in vitro and ex vivo. Journal of Physiology (London), 569, 833–847. https://doi.org/10.1113/jphysiol.2005.094599Lee, D. J., Izadi, A., Melnik, M., Seidl, S., Echeverri, A., Shahlaie, K., & Gurkoff, G. G. (2017). Stimulation of the medial septum improves performance in spatial learning following pilocarpine‐induced status epilepticus. Epilepsy Research, 130, 53–63. https://doi.org./10.1016/j.eplepsyres.2017.01.005Lee, S. E., Simons, S. B., Heldt, S. A., Zhao, M., Schroeder, J. P., Vellano, C. P., Cowan, D. P., Ramineni, S., Yates, C. K., Feng, Y., Smith, Y., Sweatt, J. D., Weinshenker, D., Ressler, K. J., Dudek, S. M., & Hepler, J. R. (2010). RGS14 is a natural suppressor of both synaptic plasticity in CA2 neurons and hippocampal‐based learning and memory. PNAS, 107, 16994–16998. https://doi.org/10.1073/pnas.1005362107Leranth, C., & Ribak, C. E. (1991). Calcium‐binding proteins are concentrated in the CA2 field of the monkey hippocampus: A possible key to this region's resistance to epileptic damage. Experimental Brain Research, 85, 129–136. https://doi.org/10.1007/BF00229993Leroy, F., Park, J., Asok, A., Brann, D. H., Meira, T., Boyle, L. M., Buss, E. W., Kandel, E. R., & Siegelbaum, S. A. (2018). A circuit from hippocampal CA2 to lateral septum disinhibits social aggression. Nature, 564, 213–218. https://doi.org/10.1038/s41586-018-0772-0Lisgaras, C. P., & Scharfman, H. E. (2022). Robust chronic convulsive seizures, high frequency oscillations, and human seizure onset patterns in an intrahippocampal kainic acid model in mice. Neurobiology of Disease, 166, 105637. https://doi.org/10.1016/j.nbd.2022.105637Marcelin, B., Chauvière, L., Becker, A., Migliore, M., Esclapez, M., & Bernard, C. (2009). h channel‐dependent deficit of theta oscillation resonance and phase shift in temporal lobe epilepsy. Neurobiology of Disease, 33, 436–447. https://doi.org/10.1016/j.nbd.2008.11.019Marx, M., Haas, C. A., & Häussler, U. (2013). Differential vulnerability of interneurons in the epileptic hippocampus. Frontiers in Cellular Neuroscience, 7, 167. https://doi.org/10.3389/fncel.2013.00167Meira, T., Leroy, F., Buss, E. W., Oliva, A., Park, J., & Siegelbaum, S. A. (2018). A hippocampal circuit linking dorsal CA2 to ventral CA1 critical for social memory dynamics. Nature Communications, 9, 4163. https://doi.org/10.1038/s41467-018-06501-wMikulecká, A., Druga, R., Stuchlík, A., Mareš, P., & Kubová, H. (2019). Comorbidities of early‐onset temporal epilepsy: Cognitive, social, emotional, and morphologic dimensions. Experimental Neurology, 320, 113005. https://doi.org/10.1016/j.expneurol.2019.113005Nasrallah, K., Piskorowski, R. A., & Chevaleyre, V. (2017). Bi‐directional interplay between proximal and distal inputs to CA2 pyramidal neurons. Neurobiology of Learning and Memory, 138, 173–181. https://doi.org/10.1016/j.nlm.2016.06.024Oh, S. W., Harris, J. A., Ng, L., Winslow, B., Cain, N., Mihalas, S., Wang, Q., Lau, C., Kuan, L., Henry, A. M., Mortrud, M. T., Ouellette, B., Nguyen, T. N., Sorensen, S. A., Slaughterbeck, C. R., Wakeman, W., Li, Y., Feng, D., Ho, A., … Zeng, H. (2014). A mesoscale connectome of the mouse brain. Nature, 508, 207–214. https://doi.org/10.1038/nature13186Okamoto, K., & Ikegaya, Y. (2019). Recurrent connections between CA2 pyramidal cells. Hippocampus, 29, 305–312. https://doi.org/10.1002/hipo.23064Oliva, A., Fernández‐Ruiz, A., Buzsáki, G., & Berényi, A. (2016). Spatial coding and physiological properties of hippocampal neurons in the cornu ammonis subregion. Hippocampus, 26, 1593–1607. https://doi.org/10.1002/hipo.22659Reyes‐Garcia, S. Z., Scorza, C. A., Araújo, N. S., Ortiz‐Villatoro, N. N., Jardim, A. P., Centeno, R., Yacubian, E. M. T., Faber, J., & Cavalheiro, E. A. (2018). Different patterns of epileptiform‐like activity are generated in the sclerotic hippocampus from patients with drug‐resistant temporal lobe epilepsy. Scientific Reports, 8, 7116. https://doi.org/10.1038/s41598-018-25378-9Robert, V., Therreau, L., Chevaleyre, V., Lepicard, E., Viollet, C., Cognet, J., Huang, A. J., Boehringer, R., Polygalov, D., McHugh, T. J., & Piskorovski, R. A. (2021). Local circuit allowing hypothalamic control of hippocampal area CA2 activity and consequences for CA1. eLife, 10, e63352. https://doi.org/10.7554/eLife.63352Shuman, T., Amendolara, B., & Golshani, P. (2017). Theta rhythmopathy as a cause of cognitive disability in TLE. Epilepsy Currents, 17, 107–111. https://doi.org./10.5698/1535-7511.17.2.107Simons, S. B., Escobedo, Y., Yasuda, R., & Dudek, S. M. (2009). Regional differences in hippocampal calcium handling provide a cellular mechanism for limiting plasticity. Proceedings of the National Academy of Sciences of the United States of America, 106, 14080–14084. https://doi.org/10.1073/pnas.0904775106Sloviter, R. S. (1991). Permanently altered hippocampal structure, excitability, and inhibition after experimental status epilepticus in the rat: The “dormant basket cell” hypothesis and its possible relevance to temporal lobe epilepsy. Hippocampus, 1, 41–66. https://doi.org/10.1002/hipo.450010106Srinivas, K. V., Buss, E. W., Sun, Q., Santoro, B., Takahashi, H., Nicholson, D. A., & Siegelbaum, S. A. (2017). The dendrites of CA2 and CA1 pyramidal neurons differentially regulate information flow in the cortico‐hippocampal circuit. The Journal of Neuroscience, 37, 3276–3293. https://doi.org/10.1523/JNEUROSCI.2219-16.2017Steve, T. A., Jirsch, J. D., & Gross, D. W. (2014). Quantification of subfield pathology in hippocampal sclerosis: A systematic review and meta‐analysis. Epilepsy Research, 108, 1279–1285. https://doi.org/10.1016/j.eplepsyres.2014.07.003Sun, Q., Sotayo, A., Cazzulino, A. S., Snyder, A. M., Denny, C. A., & Siegelbaum, S. A. (2017). Proximodistal heterogeneity of hippocampal CA3 pyramidal neuron intrinsic properties, connectivity, and reactivation during memory recall. Neuron, 95, 656–672.e3. https://doi.org/10.1016/j.neuron.2017.07.012Sun, Q., Srinivas, K. V., Sotayo, A., & Siegelbaum, S. A. (2014). Dendritic Na+ spikes enable cortical input to drive action potential output from hippocampal CA2 pyramidal neurons. eLife, 3, 04551. https://doi.org/10.7554/eLife.04551Suzuki, F., Junier, M. P., Guilhem, D., Sørensen, J. C., & Onteniente, B. (1995). Morphogenetic effect of kainate on adult hippocampal neurons associated with a prolonged expression of brain‐derived neurotrophic factor. Neuroscience, 64, 665–674. https://doi.org/10.1016/0306-4522(94)00463-FThom, M. (2014). Hippocampal sclerosis in epilepsy: A neuropathology review. Neuropathology and Applied Neurobiology, 40, 520–543. https://doi.org/10.1111/nan.12150Tulke, S., Haas, C. A., & Häussler, U. (2019). Expression of brain‐derived neurotrophic factor and structural plasticity in the dentate gyrus and CA2 region correlate with epileptiform activity. Epilepsia, 60, 1234–1247. https://doi.org/10.1111/epi.15540Vertes, R. P. (2005). Hippocampal theta rhythm: A tag for short‐term memory. Hippocampus, 15, 923–935. https://doi.org/10.1002/hipo.20118Volz, F., Bock, H. H., Gierthmuelen, M., Zentner, J., Haas, C. A., & Freiman, T. M. (2011). Stereologic estimation of hippocampal GluR2/3‐ and calretinin‐immunoreactive hilar neurons (presumptive mossy cells) in two mouse models of temporal lobe epilepsy. Epilepsia, 52(9), 1579–1589. https://doi.org/10.1111/j.1528-1167.2011.03086.xWhitebirch, A. C., LaFrancois, J. J., Jain, S., Leary, P., Santoro, B., Siegelbaum, S. A., & Scharfman, H. E. (2022). Enhanced excitability of the hippocampal CA2 region and its contribution to seizure activity in a mouse model of temporal lobe epilepsy. Neuron, 110(19), 3121–3138.e8. https://doi.org/10.1016/j.neuron.2022.07.020Williamson, A., & Spencer, D. D. (1994). Electrophysiological characterization of CA2 pyramidal cells from epileptic humans. Hippocampus, 4, 226–237. https://doi.org/10.1002/hipo.450040213Wittner, L., Huberfeld, G., Clémenceau, S., Eross, L., Dezamis, E., Entz, L., Ulbert, I., Baulac, M., Freund, T. F., Maglóczky, Z., & Miles, R. (2009). The epileptic human hippocampal cornu ammonis 2 region generates spontaneous interictal‐like activity in vitro. Brain, 132, 3032–3046. https://doi.org/10.1155/2017/7154295Wittner, L., Maglóczky, Z., Borhegyi, Z., Halász, P., Tóth, S., Eross, L., Szabó, Z., & Freund, T. F. (2001). Preservation of perisomatic inhibitory input of granule cells in the epileptic human dentate gyrus. Neuroscience, 108, 587–600. https://doi.org/10.1016/S0306-4522(01)00446-8Wyler, A. R., Curtis, D. F., Schweitzer, J. B., & Berry, A. D. (1992). A grading system for mesial temporal pathology (hippocampal sclerosis) from anterior temporal lobectomy. Journal of Epilepsy, 5, 220–225. https://doi.org/10.1016/S0896-6974(05)80120-3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Hippocampus Wiley

Integration of the CA2 region in the hippocampal network during epileptogenesis

Loading next page...
 
/lp/wiley/integration-of-the-ca2-region-in-the-hippocampal-network-during-aAVr8UFy8Q
Publisher
Wiley
Copyright
© 2022 Wiley Periodicals LLC.
ISSN
1050-9631
eISSN
1098-1063
DOI
10.1002/hipo.23479
Publisher site
See Article on Publisher Site

Abstract

INTRODUCTIONA striking characteristic of pyramidal cells of the cornu ammonis 2 (CA2) region is their resistance toward hyperactivity‐induced cell death in mesial temporal lobe epilepsy (MTLE; Steve et al., 2014). MTLE is the most common form of focal epilepsies in adult patients and characterized by seizures which involve the hippocampus and surrounding temporal brain structures and which are often resistant to pharmacological treatment. Furthermore, MTLE is frequently associated with hippocampal sclerosis including substantial loss of principal cells and inhibitory interneurons and accompanying gliosis (Blümcke et al., 2013; Engel, 2001; Thom, 2014).Instead of a single “antiepileptic” mechanism, the resistance of CA2 pyramidal cells in MTLE is most likely based on a combination of their unique characteristics, which might include the regulation of plasticity through perineuronal nets (Carstens et al., 2016, 2021; Hayani et al., 2018), specific Ca2+‐buffering properties (Lee et al., 2010; Leranth & Ribak, 1991; Simons et al., 2009), upregulation of neuroprotective factors like brain‐derived neurotrophic factor (Tulke et al., 2019) and integration into a strong inhibitory network (Botcher et al., 2014). However, despite their resistance, the physiological identity of CA2 pyramidal cells in MTLE is altered (Williamson & Spencer, 1994), including the generation of spontaneous bursts in acute hippocampal slices from human specimens (Knowles et al., 1987; Reyes‐Garcia et al., 2018; Wittner et al., 2009). In addition, a recent study in the pilocarpine mouse model for MTLE has found increased excitability in and decreased inhibition onto CA2 pyramidal cells (Whitebirch et al., 2022).The integration of CA2 pyramidal cells into the epileptic network in and beyond the sclerotic hippocampus is, however, still far from being understood. We and others have shown that mossy fibers sprout to the CA2 pyramidal cell layer to build aberrant synapses onto principal cell somata in the intrahippocampal kainate (KA) mouse model for MTLE (Häussler et al., 2016), the rat pilocarpine model (Althaus et al., 2016) and in human tissue (Freiman et al., 2021; Houser et al., 1990; Wittner et al., 2009). In the pilocarpine model, mossy fibers provide stronger input to CA2 pyramidal cells in comparison to control, although they do not sprout (Whitebirch et al., 2022), whereas CA2 pyramidal cells themselves provide increased excitatory input to all regions of the cornu ammonis in slices. Indeed, CA2 contributes to epileptic activity (EA) in vivo (Tulke et al., 2019) and its blockade using chemogenetic tools resulted in reduced seizure occurrence (Whitebirch et al., 2022). All together, these data point to a central role of CA2 in the epileptic network.It remains unclear whether the altered input via the mossy fibers translates into dentate gyrus (DG) engaging CA2 during EA in vivo and whether it changes the coupling of both regions during non‐pathological activity patterns like theta oscillations. Furthermore, whereas DG granule cells are supposed to lack direct connections to the contralateral hemisphere (Hjorth‐Simonsen & Laurberg, 1977) and cannot directly convey EA there, CA2 pyramidal cells connect bilaterally (Cui et al., 2013) and therefore might contribute to propagation of EA beyond the ipsilateral hippocampus.To address these questions, we used the intrahippocampal KA mouse model for MTLE and corroborated the local origin of EA and theta oscillations in ipsilateral CA2 by silicon probe recordings followed by current source density (CSD) analysis. Viral tracing confirmed the preservation of the projection from ipsilateral CA2 to the contralateral hippocampus in MTLE. By assessing the time course of EA in ipsi‐ and contralateral DG and CA2, as well as co‐occurrence of EA in those regions, we found a strong coupling of the DG with ipsi‐ and to a lesser extent contralateral CA2 during EA. Furthermore, theta rhythm in CA2 was coherent with the slowed‐down oscillation in the DG (Kilias et al., 2018) already early after status epilepticus (SE). In situ hybridization revealed the loss of glutamic acid decarboxylase 67 (Gad67) mRNA‐expressing interneurons directly after the initial SE in ipsi‐ but not contralateral CA2, which might contribute to the early alterations in excitability. We conclude that CA2 is an active part of the epileptic network that may act as one gate for propagation of EA beyond the sclerotic hippocampus.METHODSSurgical procedures and LFP recordingsAnimalsAll experiments were performed in adult (>8 weeks, 22‐30 g) male C57Bl/6 wild‐type mice and transgenic B6(SJL)‐Tg(Amigo2‐icre/ERT2)1Ehs (Amigo2‐icreERT2) mice (Alexander et al., 2018) or their negative littermates (all bred at CEMT Freiburg). We assume that the inducible Amigo2‐icreERT2 allele does not play a role for the characteristics of KA‐induced MTLE, as this has recently been shown for the constitutive Amigo2‐cre line (Lisgaras & Scharfman, 2022). Mice were housed under standard conditions (12 h‐light/dark cycle, room temperature [RT] 22 ± 1°C, food and water ad libitum). Experiments were carried out following the guidelines of the European Community's Council Directive of 22 September 2010 (2010/63/EU) and were approved by the regional council (Regierungspräsidium Freiburg). In total, 54 mice were used of which a subset was already analyzed for different aspects in a previous study (Tulke et al., 2019).Intrahippocampal injectionsSurgery was performed under deep anesthesia (ketamine hydrochloride 100 mg/kg, xylazine 5 mg/kg, atropine 0.1 mg/kg body weight, i.p.) with analgesic treatment (buprenorphine hydrochloride 0.1 mg/kg, carprofen 4 mg/kg, s.c.). Stereotaxic injections of 50 nl kainic acid (KA, 20 mM in 0.9% saline, Tocris, Cat 0222, RRID:SCR_003689) or 0.9% sterile saline into the right dorsal hippocampus were performed at coordinates anterioposterior [AP] = −2.0 mm, mediolateral [ML] = −1.4 mm, and dorsoventral [DV] = −1.8 mm from bregma using a stainless steel cannula (Harry Rieck Edelstahl GmbH) connected via polyethylene tubing to a 0.5 μl Hamilton syringe (7000.5 N, Hamilton Bonaduz) and a micropump (Carnegie Medicine). In KA‐injected mice, the occurrence of behavioral SE, characterized by head nodding, mild convulsions, rotations and immobility was verified as previously described (Häussler et al., 2016; Tulke et al., 2019). Postsurgical analgesia was performed with burprenorphin and carprofen for 2 days.To trace the projections from ipsilateral CA2 to the contralateral hippocampus Amigo2‐icreERT2 mice were injected with an adeno‐associated virus (AAV, serotype 1) that delivered phSyn1(S)‐FLEX‐tdTomato‐T2A‐SypEGFP‐WPRE (Addgene plasmid #51509, Penn Vector Core or Viral Vector Facility ETH Zürich, RRID:Addgene_51509; Oh et al., 2014) into CA2 at coordinates AP = −2.0 mm, ML = −2.6 mm from bregma, DV = −1.0 mm from the cortex surface. When AAV and KA injection were combined, AAV was injected 2 days prior to KA. The AAV (250 nl) was injected using a glass capillary and a programmable nanoliter injector (Nanoject III, Drummond Scientific Company) during 10 min. The capillary was left in place for another 3–5 min to avoid reflux. To induce expression of the AAV‐delivered construct, Amigo2‐icreERT2 mice were injected (i.p.) with tamoxifen in sunflower oil (100 mg/kg body weight; Sigma‐Aldrich, T5648) once per day for five consecutive days starting from 5 or 6 days after AAV injection onward (timelines for experiments in Supplementary Figure 1).Electrode implantations and in vivo recordingsElectrode implantation was either performed in the same surgery or 14 days later (anesthesia and analgesia see above). Wire electrodes (platinum‐iridum, teflon‐insulated, ∅125 μm, World Precision Instruments, PTT0502) were implanted into the DG (AP = −2.0 mm, ML = ±1.4 mm, DV = −1.9 mm) and CA2 (AP = −2.0 mm, ML = ±2.6 mm, DV = −1.4 mm) of one or both hemispheres. Jewelers' screws above the frontal cortex served as ground and reference. Electrodes were soldered to a connector, which was permanently fixed to the skull with cyanoacrylate and dental cement. Post hoc Nissl staining (section thickness 50 μm for wires) was used to verify the location of wire tips in the molecular layer or granule cell layer of the DG and in stratum oriens (so), stratum pyramidale (sp), or stratum lucidum/stratum radiatum (sl/sr) of CA2. Note that mossy fiber sprouting in CA2 prevents a clear discrimination of sr and sl borders in CA2 (Häussler et al., 2016). Only data from electrodes located unambiguously in CA2 and DG were used for analysis, which results in different numbers of electrodes in the analyses that depended on proper placement of individual electrodes or on pairs of electrodes (Supplementary Table 1). Custom‐made single shaft silicon probes (16 iridium‐oxide [IrOx] electrodes of 35 μm diameter, spacing 45 μm; Herwik et al., 2009, 2011; Lanzilotto et al., 2016) were chronically implanted into CA2 (AP = −2.0 mm, ML = −2.6 mm, probe tip DV = −1.7 mm from the cortex surface). Probe positions were reconstructed on Nissl‐stained sections (section thickness 35 μm), on 4′,6‐diamidino‐2‐phenylindole (DAPI)‐stained sections or with Purkinje cell protein 4 (PCP4) immunocytochemistry (see Section 2.2.1; Supplementary Figure 2). Individual electrode positions of silicon probes were reconstructed post hoc based on histological traces of the probe track and the amplitude profile of action potentials to delineate sp (Supplementary Figure 2).Recordings were performed 1–2 days (≤48 h), 7, 14, and 21 days after KA injection for mice implanted with wires and 21/22 days after KA for mice implanted with probes (Supplementary Figure 1). Freely moving mice were connected to one or two miniature preamplifiers (MPA8i, Multichannel Systems). LFPs were filtered, amplified (gain: wires and probes 500‐fold, filter: wires 1 Hz–5 kHz, probes: 1 Hz–1 kHz; PGA32, Multichannel Systems) and digitized (sampling rate wires: 10 kHz, probes for LFP: 2 kHz, Power1401 mk2/ADC16, Spike2 software; CED, RRID: SCR_000903).Histological analysisPerfusion, slice preparation, immunocytochemistry, and in situ hybridizationMice were transcardially perfused under deep anesthesia with 0.9% saline followed by paraformaldehyde (4% in 0.1 M phosphate buffer [PB], pH 7.4) >14 days after implantation or at 2, 7, 14, and 21 days after KA/NaCl injection for Gad67 fluorescence in situ hybridization (FISH). Brains were dissected and postfixed in the same fixative overnight (4°C), cut on a vibratome in the frontal or horizontal plane (50 μm thickness), the next day and collected in PB for analysis of viral tracing. For FISH, brains were cryoprotected in 30% sucrose, frozen in isopentane and stored at −80°C. Cryostat sections (50 μm) were prepared in the frontal plane and collected in PB or in 2x saline sodium citrate.For immunocytochemistry, slices were preincubated with 0.25% Triton X‐100 (Sigma‐Aldrich) and 10% normal goat serum in 0.1 M PB (30 min, free‐floating), followed by incubation for 4 h at RT+ overnight at 4°C with polyclonal rabbit anti‐PCP4 (1:400; Sigma‐Aldrich GmbH, HPA005792, RRID:AB_1855086) or polyclonal rabbit anti‐green fluorescent protein (GFP, 1:1000; Abcam, ab6556, RRID:AB_305564) primary antibodies with 0.1% Triton and 1% normal goat serum. A secondary goat anti‐rabbit antibody conjugated to Cy2 (1:400; Jackson ImmunoResearch, 111‐225‐144, RRID:AB_2338021) was applied for 2–3 h at RT. Counterstaining was performed with DAPI (1:10,000, Roche, Cat 236276). After washing, sections were mounted on microscope slides and coverslipped with Immu‐Mount (Thermo Shandon Ltd).The expression of Gad67 mRNA was investigated by FISH with a digoxigenin (DIG)‐labeled cRNA probe generated by in vitro transcription from appropriate plasmids (Kulik et al., 2003). Sections were hybridized overnight at 45°C, followed by detection with an anti‐DIG antibody and tyramide signal amplification as described (Marx et al., 2013; Tulke et al., 2019). FISH was always followed by PCP4 immunohistochemistry for precise location of CA2 as described above, but for double‐staining Triton X‐100 was omitted.Microscopy and quantificationSections were imaged with an epifluorescence microscope (Axio Imager 2, Zen software, Zeiss, RRID:SCR_018876, RRID:SCR_013672) with a 10‐fold objective (Plan Apochromat, NA 0.45), appropriate filter settings and a digital camera (AxioCam 506). All images that were compared were taken with identical exposure times. To analyze the DV extent of AAV labeling of somata (tdTomato) and axonal projections (synaptophysin‐bound EGFP), images of horizontal sections throughout the DV extent of the hippocampus were taken and assigned to the planes of a horizontal brain atlas (Franklin & Paxinos, 2007). All planes were checked for tdTomato‐positive somata and EGFP‐expressing axons and the location of the most dorsally and most ventrally located tdTomato‐positive somata and the most dorsally and most ventrally located EGFP‐expressing axons in the ipsi‐ and contralateral hippocampus, respectively, were visually determined by two independent observers. To measure density profiles across the layers of ipsi‐ and contralateral CA2 in the horizontal plane we selected the section closest to the KA injection site (W‐like shape of the granule cell layer). Images were converted to 16‐bit in ImageJ (version 1.53a, National Institute of Health) and intensity profiles along a straight line (length 330 μm, drawn from so to sr/sl) were measured at three equidistant positions and averaged. All intensity values are given relative to the maximum intensity in ipsilateral CA2 of the respective slice. For measurement of intensity profiles in CA3 and CA1 we proceeded likewise, only with different line lengths (350 μm for CA3, 450 μm for CA1). Values are again given relative to the maximum in ipsilateral CA2.For quantification of Gad67 mRNA‐expressing cells in CA2, we selected dorsal slices surrounding the KA injection site (two to four slice/mouse). The location of CA2 was determined in the PCP4 immunostaining, subsequently three regions of interest (ROIs) were placed in the FISH images framing so, sp and stratum lacunosum moleculare (slm)/sr/sl of CA2. All Gad67+ cells within every ROI were counted manually and average values per mouse are given as cells/mm2. For all analyses, blinding to treatment or time point was not possible due to the salient progressive changes after KA injection (cell loss, granule cell, and CA2 dispersion).Data analysisDetection of epileptiform activityFirst, we controlled for sufficient data quality: As mentioned above, only recordings with proper electrode placement in the DG or CA2 were used. In addition, noisy or artifact‐rich recordings, either due to electrode/soldering quality or mouse behavior were discarded. Finally, not all mice were recorded at all days of the series. This results in different numbers of mice for each recording site and recording day (Supplementary Table 1).Data analyses of preselected data were done in Python 3.7.3 (Python Software Foundation, RRID:SCR_008394) and Matlab 2022a (Mathworks, RRID:SCR_001622). Individual epileptiform discharges in the LFP were detected with a custom‐made algorithm in Python as previously described (Heining et al., 2019; code accessible at Zenodo: https://doi.org/10.5281/zenodo.4110614). In brief, LFP data were resampled at 500 Hz, artifacts were manually removed and discharges were detected based on a combination of frequency composition and amplitude threshold. Discharges occurring within 2.5 s were grouped into one burst. Bursts were classified according to number of discharges, median interdischarge interval, and coefficient of variation using Ward's hierarchical clustering into mild, moderate, and severe events. Detection was performed for all channels individually (DG ipsi/contra, CA2 ipsi/contra).We defined the coupling between two hippocampal regions as the overlap of EA in time. Hereby, we focused our analysis on severe EA in a seed region (DGi, CA2i) and quantified the temporal overlap with co‐occurring EA of any class detected in a target region (CA2i, DGi, CA2c). The overlap is presented as a fraction of the total severe EA time in the seed region. Due to low sample size, in particular at 21d, we did not include DGc in this analysis.Analysis of theta periodsTheta properties were analyzed as previously described (Kilias et al., 2018). In brief, LFP periods (≥9 s) dominated by theta activity and free of EA, slow wave sleep, large irregular activity associated with resting, or movement artifacts were identified manually a all available recording channels. LFP recordings were resampled at 1 kHz. Autocorrelation and power spectral density (Thomson's multitaper power spectral density estimate, 5 Slepian tapers) were calculated on sliding windows (4 s shifted in 0.25 s steps) for the selected LFP periods. For each window, we detected the spectral power maximum and its corresponding frequency within theta band (5–12 Hz). We smoothed the resulting power–frequency distributions (median filtering, 0.75 Hz × 1.5 × 10−3 mV2/Hz). Theta peak frequency was defined as the frequency at the maximum of the theta frequency probability distribution after a second smoothing step (Savitzky–Golay, first order, 1.25 Hz window). This analysis was done for all animals (nanimals = 12) and all recording sites (nrec sites = 35).Recording sessions in which the autocorrelation and the power spectra lacked a pronounced peak in the theta band and where no peak in the power–frequency distributions could be detected, were excluded from further analysis. This was particularly relevant for recordings performed ≤48 h after SE (see Figure 4b).Coherence was averaged across sliding windows (4 s with 3 s overlap) of all sessions for 2–250 Hz at 0.25 Hz resolution. Coherence peaks in theta band were detected as local maxima in the Sawitzky‐Golay‐filtered coherence between 5 and 12 Hz after averaging across animals.CSD estimationWe performed CSD analysis on data recorded 21/22 days after KA injection in animals implanted with a single shaft 16‐channel silicon probe covering all layers of CA2 (nKA = 3). To identify the sink‐source profile of EA in CA2 we detected negative peaks that exceeded an amplitude of 0.9 mV in a 50 Hz low‐pass filtered LFP recorded in slm. Data were extracted in a window of −150 to +500 ms around identified peaks, averaged across a subset of 340 randomly selected windows per animal. CSDs were estimated from these averages. Subsequently, CSDs were cropped to seven channels covering slm to sp. In one animal, we bootstrapped one channel in the middle of sl/sr to account for the structural variability. The cropped individual CSDs were averaged across all three mice to obtain a final CSD profile.To identify the sink‐source profile of theta activity in CA2 we filtered the LFPs (two mice, one mouse was excluded because movement artifacts impeded theta analysis) for the theta band (band‐pass, second order, butterworth) and z‐scored to its mean and standard deviation (SD) across all selected theta periods per session. In the z‐scored theta periods, we detected all theta cycles with peaks that exceeded a threshold of 0.5 × SD and with both adjacent peaks above threshold. We cut 1 s windows around the peak times of these cycles and averaged across 3900 randomly selected windows and CSDs were calculated for these averages. We cropped and averaged the CSDs as described above to obtain an interindividual mean. We computed the spectrum of the CSD traces originating from sr (Thomson's multitaper power spectral density estimate, 3 Slepian tapers) and identified the frequency of maximal power.To circumvent effects of slight differences in gains and noise of the two independent preamplifiers, where one MPA8I amplified signals from all even and the other from all odd electrodes of the silicon probe, we calculated separate CSDs for even and odd electrodes, with 90‐μm electrode spacing, respectively, and these were subsequently merged and smoothed (2D median filter, 2 electrodes × 15 ms).StatisticsStatistical analysis was performed with MATLAB for electrophysiological data and GraphPad Prism 9 (GraphPad Software LLC, RRID:SCR_002798) for histology. Individual values, mean ± SD are presented. Results from the EA analysis were statistically evaluated by a two‐ or three‐way analysis of variance (ANOVA) with time point, structure (DG/CA2) and hemisphere if applicable as factors followed by a Tukey's multiple comparison test. Changes in theta frequency were compared across time by a repeated measure three‐way ANOVA and compared to a previously published distribution of theta frequency in healthy and KA‐injected mice with a Kolmogorov–Smirnov test. Theta coherence peaks were compared to predominant theta frequency by using a one‐sample t test with the mean frequency of both tested electrodes as the hypothesized mean of the underlying distribution.For statistical evaluation of density of Gad67+ cells, a two‐way ANOVA was performed for factors treatment/time point and hemisphere followed by a Tukey's multiple comparison test for treatment/time point and a Šidák posttest for pairwise comparison between hemispheres for each time point.RESULTSEA occurs locally in ipsilateral CA2To determine the role of CA2 under epileptic conditions, we implanted wire electrodes into the DG and CA2 of the ipsi‐ and contralateral hippocampus of KA‐injected mice and measured EA as well as periods dominated by theta activity but free from any ictal or interictal events (Figure 1). Positioning of electrodes in the molecular layer or granule cell layer of the DG and in sl/sr, sp or so of CA2 was verified in post hoc Nissl staining (Figure 1a). In a subset of mice, the placement of electrodes or silicon probes in CA2 was further confirmed by PCP4 immunostaining (Figure 1a). Episodes of EA, mostly occurring during quiet wakefulness, alternated with theta episodes during sleep or locomotion (Figure 1b, 7 days after KA). We applied an automated detection algorithm with subsequent classification of EA into three different categories—mild, moderate, or severe—based on the number of discharges, interdischarge intervals and coefficient of variation for each channel independently (Figure 1b; Heining et al., 2019). Theta episodes were identified manually by cross‐validating all available channels per session and animal (Figure 1b).1FIGUREEpileptic and theta activity in CA2 have a source local to CA2. (a, left) Implantation scheme for bilateral electrodes in the dorsal dentate gyrus (DG) and CA2 with color‐coded severity of kainate (KA)‐induced hippocampal reorganization (darker red ≙ stronger damage). (a, middle) Nissl‐stained hippocampal sections show the tips of the wire electrodes (white arrows). (a, right) Section of the ipsilateral hippocampus of a mouse implanted with a silicon probe immunostained for Purkinje cell protein 4 (PCP4, green) with 4′,6‐diamidino‐2‐phenylindole (DAPI) counterstaining (magenta). The probe shank crossed the pyramidal layer of CA2. (b, top) Local field potential (LFP) traces of ipsi‐ and contralateral DG and CA2 at 7 days after KA. Manually identified theta periods (green bars) alternated with automatically detected epileptic activity (EA) of different severity (violet: severe, lilac: moderate, pink: mild EA). (b, bottom left) Zoom into a period of EA co‐occurring at all four positions with discharges of large amplitude in ipsilateral DG and CA2 and more sparse epileptic spikes in the contralateral hippocampus (DGc and CA2c). (b, bottom right) Theta oscillations could be observed simultaneously at all four positions. Broadband activity (black trace) is shown overlayed with its theta‐filtered version (colored trace). (c) DAPI staining with reconstructed track of a 16‐channel silicon probe implanted in ipsilateral CA2 with the recording sites covering all layers. (d, top) Average amplitude profile from outer stratum lacunosum moleculare (slm, brown) to stratum pyramidale (sp, yellow) of CA2 for large epileptic spikes (mean of 340 spikes per mouse averaged across three mice). (d, bottom) Average current source density (CSD) profile based on traces in (d, top) revealed local sink‐source alternation across distal and proximal apical dendritic areas and across time. Inset shows the CSD time course at an individual electrode (*) in sr. The sink in sr is delayed by 8.5 ms to that in slm and reaches its maximum after 44 ms. CSDs were computed individually for each mouse (n = 3 mice), cropped to seven channels covering the area between slm and sp and averaged across mice. (e) Corresponding average amplitude profile and CSD profile of theta oscillations in CA2 (n = 2 mice, 3900 theta peaks per mouse). Scale bars in (a) left 2 mm, middle 100 μm, right 500 μm, in (c) 100 μm. so, stratum oriens; sp, stratum pyramidale; sr, stratum radiatum; slm, stratum lacunosum moleculare; ml, molecular layer; gcl, granule cell layer; pol, polymorphic layerEpileptic spikes of large amplitudes occurred in the ipsilateral and the contralateral DG, as previously described (Häussler et al., 2012; Janz et al., 2018). In ipsilateral CA2, EA occurred mostly concomitant with the ipsilateral DG whereas contralateral CA2 usually showed shorter EA with low amplitudes (Figure 1b, EA example). Accordingly, automatically detected EA was usually classified into less severe categories in the contralateral hippocampus.Theta oscillations were found bilaterally in the DG and CA2 (Figure 1b, θ example), but with lower power in the ipsilateral DG, as described previously (Kilias et al., 2018). To determine whether the LFPs of epileptic and theta activity recorded in CA2 were generated locally in CA2 or a volume‐conducted signal, we implanted 16‐channel silicon probes sampling all layers of CA2 (n = 3 mice, 21/22 days after KA, Figure 1c). We recorded spatio‐temporally resolved activity profiles (Figure 1d,e) and estimated CSD profiles of epileptic spikes (Figure 1d) and theta oscillations (Figure 1e).For large epileptic spikes detected in slm (Figure 1d, 21 days after KA, top trace, amplitude profile of average across n = 3 mice, 340 epileptic spikes per mouse, layers color‐coded), the averaged CSD profile revealed a strong dipole at the onset of the epileptic spike consisting of a profound current sink in slm at the position of distal dendrites opposed by a source in sl/sr. This was followed after ~9 ms by a strong sink in sl/sr which reached its maximum after ~44 ms and propagated toward sp, and a subsequent long‐lasting source of ~200 ms duration spanning all layers of CA2 (Figure 1d).During theta oscillations, an alternating sink‐source pattern was visible with a dipole which inverted its polarity spatially at the border of slm and sl/sr and temporally with a frequency of ~6.6 Hz (see Figure 4c, right) (Figure 1e, n = 2 mice). The strongest sink was visible in the proximal half of sl/sr, approximately the region where mossy fibers and CA2 recurrent connections terminate.Both patterns are in favor or a local origin of activity in CA2 instead of volume‐conducted potentials from neighboring regions. Together, our data show that CA2 in chronically epileptic mice (i.e., at a time point when CA2 pyramidal cell dispersion and mossy fiber sprouting have developed), actively contributes to EA and theta oscillations in the hippocampus.Preservation of the connectivity between ipsilateral CA2 and the contralateral hippocampus under epileptic conditionsWe hypothesized that CA2 with its interhemispheric connectivity might serve as a gateway for EA propagation. A prerequisite for this is the preservation of the projection of ipsilateral CA2 to the contralateral hippocampus under epileptic conditions. To test this, we injected an AAV carrying a floxed copy of tdTomato under the control of the synapsin promoter and EGFP bound to the vesicle protein synaptophysin into the CA2 region of Amigo2‐icreERT2 mice prior to KA injection and analyzed the fluorescently labeled projections in frontal and horizontal sections (Figure 2a). In control mice that received only the AAV injection (n = 1 mouse frontal, n = 2 mice horizontal), we detected tdTomato labeling mainly in CA2 somata and dendrites and occasionally in larger axons (Figure 2b,c,e). Synaptophysin‐bound EGFP labeling was strongest in axons and synapses, but also visible in somata of pyramidal cells. Co‐staining with a CA2‐specific marker (PCP4, data not shown) confirmed the specificity of the viral transduction. Within the hippocampus, projections to ipsilateral CA3, CA2, and CA1 and to the contralateral side, in particular to CA2 and CA1 were strongly labeled (Figure 2b frontal plane, Figure 2c,e horizontal plane). At 21 days after KA injection, labeling was overall similar, but the density of tdTomato‐labeled cells was lower, mainly due to dispersion of the CA2 pyramidal cell layer, as shown previously (Häussler et al., 2016) and some loss of CA2 pyramidal cells (Figure 2b,d,f; n = 1 mouse frontal, n = 5 mice horizontal). The projection from the dispersed CA2 to the adjacent CA1 region was strongly reduced in sections close to the injection site reflecting the loss of CA1 neurons in the sclerotic hippocampus at this position (Figure 2b, Supplementary Figure 3). Nevertheless, the projection to contralateral CA2, but also to contralateral CA3 and CA1 was preserved (Figure 2b–f, Supplementary Figure 3). We determined the DV extent of tdTomato‐labeled somata in the horizontal plane of ipsilateral CA2 as well as the most dorsal and most ventral position at which EGFP‐positive CA2 axons and synapses were visible in the ipsi‐ and contralateral hippocampus, respectively, for all mice. We did not find any salient differences between control and KA‐injected hippocampi (Figure 2g). Similarly, the layer‐resolved (so to sr) intensity profiles of EGFP‐expression in contralateral CA2 relative to the intensity maximum on the ipsilateral side (to account for the altered density of transduced cells) did not differ between KA and control (Figure 2h). In summary, our viral tracing experiments show that the projections of ipsilateral CA2 to the contralateral hippocampus persist after KA injection and offer a direct pathway for activity propagation from the ipsi‐ to the contralateral hippocampus.2FIGUREPreservation of CA2‐CA2 connectivity in epilepsy. (a) Scheme depicting the site of tracer injection and slicing angles. (b) Tracing with a cre‐dependent adeno‐associated virus (AAV) in the CA2‐specific Amigo2‐icreERT2 mouse line; frontal sections at the level of the injection site. Upper row: ipsi‐ and contralateral hippocampus of a control mouse. Lower row: epileptic mouse 21 days after kainate (KA) injection. Virally delivered tdTomato (red) is expressed in somata and dendrites of ipsilateral CA2, the axonal projections (but also local somata and dendrites) express synaptophysin‐bound EGFP (green). Note the dispersion of the GCL and cell loss in CA3 and CA1 (and the corresponding loss of CA2 projections to these regions) on the ipsilateral site after KA injection. The projection from ipsilateral to contralateral CA2 appears weaker but preserved in epileptic mice (arrows). (c) Representative control section in horizontal plane at the dorso‐ventral (DV) level of the injection site (injection in right hippocampus), tdTomato expression in somata and processes (red), synaptophysin‐bound EGFP (green) with 4′,6‐diamidino‐2‐phenylindole (DAPI) counterstaining (blue). (d) Same as in c but without DAPI. Small images: Ipsilateral CA2 at injection site and contralateral CA2 at corresponding level. Dotted line illustrates the extent along which optical density measurement (shown in h) has been performed. (e) Horizontal section of a KA‐injected brain 21 days after KA. The dispersion of the GCL is visible. (f) Same as in e but without DAPI. The projection of ipsi‐ to contralateral CA2 is preserved. Inset shows cells at the AAV injection site—note that this is taken from a section slightly more anterior than the large image to show location of maximal labeling of somata. (g) Qualitative analysis of DV extent of labeled cells in horizontal sections (for all sequential sections cut in horizontal plane; DV position according to Franklin and Paxinos “Atlas of the mouse brain”) from control mice (n = 2 mice) and KA‐injected Amigo2‐icreERT2 mice (n = 5 mice). The DV extent is depicted, along which labeled axons/synapses within the ipsilateral hippocampus (axons ipsi) and to the contralateral hippocampus (axons contra) were visible. (h) Mean relative intensity profiles measured from stratum oriens (so) to stratum radiatum (sr) are given for ipsi‐ and contralateral CA2 in control and KA‐injected mice for sections around the DV level of the KA/AAV injection site. Scale bars (b, c) 500 μm, small images in (e, f) 100 μm. GCL, granule cell layer, SO, stratum oriens; SP, stratum pyramidale; SR, stratum radiatumTime course of EA and coupling between DG and CA2 in both hemispheresGiven the active participation of ipsilateral dorsal CA2 in EA and its preserved contralateral projections, we next explored a potential role of CA2 as a gate for propagation of EA in the epileptic network. In particular cell loss in CA3 and CA1, the main up‐ and downstream regions of CA2, and the progressive development of mossy fiber sprouting onto CA2 somata during the first 2 weeks after KA (Figure 3a, schematic drawing; Häussler et al., 2016) raised the possibility of an enhanced coupling between CA2 and its remaining pre‐ and postsynaptic partners during EA.3FIGUREBilateral coupling of epileptic activity in the dentate gyrus and CA2. (a, top) Schematic drawing of both hippocampi before kainate (KA) injection (injection site indicated) and 2–3 weeks after KA injection when hippocampal sclerosis (cell loss mainly in CA3 and CA1 and astrogliosis), and the dispersion of dentate granule cells and CA2 pyramidal cells has taken place in the ipsilateral hippocampus. The contralateral hippocampus is spared from any major changes. (a, bottom) Time spent in epileptic activity (EA) as a fraction of total recording time for ipsi‐ and contralateral DG and CA2 measured over 21 days after the initial status epilepticus (SE). (b) For each time point and recording site, EA was classified into mild (b, left); moderate (b, middle); and severe (b, right) events and the fraction of time in each class relative to total EA time is given. (c, left) Likelihood of measuring EA of any type in ipsi‐ and contralateral CA2 while severe EA was detected in ipsilateral DG. (c, right) Likelihood of measuring EA of any type in ipsilateral DG or contralateral CA2 while severe EA was detected in ipsilateral CA2. Mean ± standard deviation (filled circles) and values for individual mice (small dots) are given for all analyses. DGi, ipsilateral dentate gyrus; DGc, contralateral dentate gyrus; SE, status epilepticusFirst, we compared the mean fraction of time spent in EA (all severity classes included) which was higher in the ipsilateral hippocampus than in the contralateral hippocampus at all time points (Figure 3a, n = 10 mice, time‐independent mean fraction ± SD DGi: 0.26 ± 0.13, CA2i: 0.31 ± 0.12, DGc: 0.19 ± 0.14, CA2c: 0.14 ± 0.11; difference in ratio between hemispheres in post hoc Δ = 0.10, p = .001). Interestingly, we did not find any major differences between the time points for the individual recording sites (main effect time p = .43), indicating that there was no or only a very short silent period after SE but instead EA occurred from early on.To determine whether EA changes in severity during the first weeks after KA injection, we next regarded the classification of EA into mild, moderate, and severe events, as described above, and determined the ratio of time spent in each class relative to total EA for all time points (Figure 3b). We found that mild events made up for a large part of total EA time early after KA injection (≤48 h), but diminished thereafter at all recording sites (Figure 3b; three‐way ANOVA for time, structure and hemisphere, post hoc: ≤48 h vs. 7 days p = 2.3 × 10−4, ≤48 h vs. 14 days p = .0014, ≤48 h vs. 21 days p = .61). However, in contralateral CA2, mild events made up for the largest fraction of EA time, which was significantly higher than in the ipsilateral hippocampus at all time points (DGi vs. CA2c p = .0012, CA2i vs. CA2c p = 2.7 × 10−5). For all other positions, the mild event rate did not differ. The share spent in moderate events was constant over time and comparable for all recording sites. In contrast, severe events were rare at ≤48 h at all recording sites and strongly increased until the 7 days time point, remained at this level at 14 days, but slightly dropped at 21 days (Figure 3b; post hoc time: ≤48 h vs. 7 days p = 2.2 × 10−5, ≤48 h vs. 14 days p = 2.3 × 10−4, ≤48 h vs. 21 days p = .74, 7 days vs. 21 p = .04). The fraction of severe events was always higher in the ipsilateral compared to the contralateral hippocampus, but did not differ between ipsilateral DG and CA2 (post hoc: ipsi vs. contra p = .0022). The high degree of similarity between EA classes in ipsilateral DGi and ipsilateral CA2 supports the idea of a strong EA coupling between these regions. The propagation to contralateral CA2, however, seems to be much less prominent and might depend on large events. To analyze this relationship, we inspected severe EA in the ipsilateral DG and tested for co‐occurring EA (independent of its class) in ipsi‐ or contralateral CA2. As hypothesized, the rate of coupling between DGi and CA2i was high (Figure 3c; mean fraction ± SD: 62 ± 19%) and despite ongoing mossy fiber sprouting, did not increase over time (two‐way ANOVA main effect time, p = .97). The coupling of EA between ipsilateral DG and contralateral CA2 was much weaker (34 ± 25%, difference of DGi to CA2i vs. CA2c Δ = 30.26%, p = 2.9 × 10−4) but also constant over time.Interestingly, when testing in opposite direction taking severe EA in ipsilateral CA2 as a basis, the chance to observe simultaneous EA in the ipsilateral DG was initially lower (34 ± 14%) but increased during the first 2 weeks to 66 ± 7%, a level comparable to the inverse coupling from the ipsilateral DG to ipsilateral CA2 (Figure 3c). The coupling from ipsi‐ to contralateral CA2 was overall lower (DGi vs. CA2c Δ = 24.63%, two‐way ANOVA post hoc: p = 9.5 × 10−5), but also increased over the time course of 3 weeks by ~23% (main effect time 0.004, post hoc ≤48 h vs. 21 days Δ = 27.34% structures pooled, p = .019).In summary, we found stronger coupling of EA within the ipsilateral hippocampus which was stable over time when taking the DG as basis. When taking ipsilateral CA2 as basis, both, coupling to the DG and to contralateral CA2 increased over time, suggesting a dynamic reorganization of EA sources in CA2.Theta frequency reduction and coherence in ipsi‐ and contralateral CA2Given the temporal reorganization of EA sources in CA2 and the rather weak coupling of EA with the contralateral side, we asked next how CA2 is embedded into the hippocampal network during theta oscillations. We have previously shown that despite chronic MTLE theta oscillations are prominent in the DG and medial entorhinal cortex, but occur at lower prevalence compared to controls (Kilias et al., 2018).When identifying theta periods of 9 s or longer, we found that the fraction of theta periods relative to total recording duration did not significantly change over time after KA injection (Figure 4a, repeated measure ANOVA, time p = .64). However, when we calculated the autocorrelation and spectra of theta periods, there was no clear peak in the theta range in several recording sessions (Figure 4b top). In particular at ≤48 h after SE, a large fraction of recording sites in the ipsilateral hippocampus did not show clear oscillatory theta activity, whereas there was an identifiable peak on the contralateral side (Supplementary Table 1; three‐way ANOVA for time point, targeted structure and hemisphere of recordings without identifiable theta peak, post hoc ≤48 h ipsilateral was different to all other time points and sites: ipsi/contra at 7 days p = .027/.014, 14 days p = .012/.013, 21 days p = .013/.0064). The remaining recording sessions were characterized by prominent secondary maxima in the autocorrelation at 83–200 ms lag (Figure 4b, middle) and a spectral power peak in the range of 5–12 Hz (Figure 4b, bottom). Only sessions fulfilling these criteria were used for further analysis.4FIGURETheta oscillations in the DG‐CA2 network of both hemispheres. (a) The average (mean across animals) proportional time covered by all theta periods (≥9 s) at different time points after kainate (KA) does not significantly change over time. (b, top row) Fractions of recordings without a prominent secondary maximum in the autocorrelation or a spectral power peak corresponding to theta range. For each channel, the fraction is given relative to the total available recordings at this particular site and time point. The remaining data sets showed prominent peaks in the autocorrelation at 83–200 ms lag (b, middle) and spectral peaks in the range of 5–12 Hz (b, bottom) for all time points. (c) Dominant theta frequency for ipsi‐ and contralateral DG and CA2 for all time points. Theta frequency is stable across time, comparable over structures and not different from previously published data from KA‐injected chronically epileptic animals (KAREF, Kilias et al., 2018; mean ± standard deviation (SD) of the earlier data set shown as green dashed line and shaded area, respectively) but to data from NaCl‐injected animals (CtrREF, Kilias et al., 2018; mean ± SD in gray). (d) Coherence for electrode pairs of DGi, CA2i, and CA2c for each time point (mean ± SD as line and shadow in corresponding colors). Coherence peaks were found in the theta range for all three possible combinations of recording sites and at all time points. The number of data sets for each combination and time point is indicated in the plot.Next, we determined the dominant frequency within the theta band. We have shown earlier that theta oscillations, irrespective of the underlying behavior, are by ~1 Hz lower in the DG and medial entorhinal cortex under epileptic conditions (Kilias et al., 2018). Since this did not depend on local reorganization of the DG and since the hippocampus is a system of coupled oscillators (Vertes, 2005), we expected a proportional reduction in theta frequency in CA2 and indeed found a predominant frequency of 6.43 ± 0.35 Hz. This frequency range was not significantly different from our previous dataset obtained in the DG of chronically epileptic KA‐injected mice (Kilias et al., 2018; Figure 4c, all recordings of DG and CA2 pooled, one‐sample Kolmogorov–Smirnov test, p = .06). It was, however, substantially lower than the theta oscillations found in equivalent NaCl‐injected controls (Kilias et al., 2018: 7.54 ± 0.9 Hz, p = 1.75 × 10−14). Interestingly, the reduction in theta frequency was not only invariant across hippocampal regions, but also stable over time (repeated measure ANOVA, time p = .44). Thus, the reduced frequency is not a consequence of the progressive reorganization of the hippocampal formation but an immediate effect of SE‐associated processes.Our CSD analysis revealed an alternation of strong sinks and sources within sr of CA2, the target zone of the mossy fibers. Given the progressive mossy fiber sprouting during the first weeks after SE, we asked whether the coherence in theta band between DG and CA2 changes over time. We found a coherence peak between the ipsilateral DG and ipsi‐ and contralateral CA2 at all time points tested in the first 3 weeks after SE (Figure 4d). The previously described absence of theta oscillations in ~67% of the ipsilateral recordings at ≤48 h but an identifiable oscillation on the contralateral electrodes implies a loss of coherence to contralateral theta (Figure 4b). Nevertheless, when theta oscillations were present in DG and CA2, their coherence peak frequency was not significantly different from the mean of their individual frequencies (Figure 4d). At 7and 21 days, coherent oscillations were slightly faster than the observed individual theta oscillations at the individual recording sites (7 days, mean theta of DGi and CA2i—Δf = 0.4 Hz, p = .0002; 21 days DGi—CA2i Δf = 0.38 Hz, p = .04; 21 days CA2i—CA2c Δf = 0.58 Hz, p = .0052, one‐sample t test with the mean frequency of the individual electrodes as hypothesized mean), but never large enough to compensate for the reduction in frequency relative to controls seen before (Δf = 1.04 Hz, Kilias et al., 2018).In summary, we found that CA2 is fully integrated in the hippocampal theta oscillation network and that the overarching reduction of theta frequency happens early after SE.Loss of GABAergic neurons in ipsilateral CA2 onlyFinally, we asked whether local changes in inhibition could contribute to the strong transmission probability of EA from ipsilateral DG to CA2 early after SE, although mossy fiber sprouting becomes prominent only at 14 days after KA (Häussler et al., 2016). Furthermore, the increasing transmission from ipsi‐ to contralateral CA2 might be promoted by a progressive interneuron reduction in contralateral CA2. To test this, we analyzed whether the density of Gad67 mRNA‐expressing interneurons changes with time after KA injection. After locating the position of CA2 in PCP4 immunostaining (Figure 5a,b), we quantified the density of GABAergic cells in all layers of ipsi‐ and contralateral CA2 in a FISH for Gad67 mRNA. In NaCl‐injected controls, the ipsi‐ and contralateral hippocampus did not differ indicating that all effects were KA‐specific (Figure 5c,i–k). In contralateral CA2 the density of Gad67 mRNA‐expressing cells was comparable to control in all layers of CA2 at all time points (2, 7, 14, 21 days after KA; Figure 5g,i–k). In ipsilateral CA2, the density was strongly reduced compared to the contralateral hippocampus and to control in so, sp, and slm/sr/sl already at 2 days after KA and persistent for all time points (n = 4–6 mice per group; so: two‐way ANOVA for treatment/time point p < .001, hemisphere p < .0001, post hoc ipsi: NaCl vs. 2—14 days p < .0001, NaCl vs. 21 days p = .0065; post hoc for hemispheres: 2—14 days p < .0001; sp: two‐way ANOVA for treatment/time point p < .001, hemisphere p < .0001, post hoc ipsi: NaCl vs. 2—21 days p < .0001; post hoc for hemispheres: 2—21 days p < .0001; slm/sr/sl: two‐way ANOVA for treatment/time point p < .001, hemisphere p < .0001, post hoc ipsi: NaCl vs. 2—21 days p < .0001; post hoc for hemispheres: 2—14 days p < .0001, 21 days p = .0003). Interestingly, there was a tendency for higher interneuron density at 21 days after KA compared to all earlier time points after KA.5FIGUREGABAergic interneurons in CA2. (a, b) Representative images of Purkinje cell protein 4 (PCP4) immunostaining in sections of the dorsal hippocampus to localize CA2. (a) NaCl‐injected control. Granule cells and CA2 pyramidal cells are PCP4‐positive. (b) Ipsilateral hippocampus, 21 days after KA injection. PCP4‐positive granule cells and CA2 pyramidal cells are dispersed (arrows). (c–h) Representative images of a fluorescence in situ hybridization (FISH) for glutamic acid decarboxylase 67 (Gad67) mRNA displays GABAergic interneurons. (c) NaCl‐injected control, same section as in (a). GABAergic interneurons are abundant throughout the slice and densely placed in CA2. (d) Ipsilateral hippocampus, 2 days after KA. A strong loss of GABAergic interneurons is visible, with some preserved interneurons in CA3/CA2. (e) Ipsilateral hippocampus, 7 days after KA. Similar to 2 days. (f) Ipsilateral hippocampus, 14 days after KA. A progressive dispersion of the GCL is visible, some GABAergic cells are preserved in CA3/CA2. (g) Contralateral hippocampus, 21 days after KA. The contralateral hippocampus is comparable to control at all time points. Occasionally, some interneuron loss occurs in distal CA1. (h) Ipsilateral hippocampus, 21 days after KA, same slice as in (b). In some mice, more Gad67 mRNA‐expressing cells are visible in CA3/CA2 than at earlier time points. (i–k) Layer‐resolved quantification of the density of Gad67 mRNA‐expressing cells in the dorsal CA2 region (close to the injection site) in NaCl‐injected controls and for all time points after KA injection (in cells/mm2). Mean ± standard deviation and individual values per animal (means of 2–4 sections per mouse) are given. Values for the animals chosen for display in (a–h) are indicated with a green circle. (i) Stratum oriens (so). Two‐way analysis of variance (ANOVA): treatment/time point p < .001, hemisphere p < .0001. (j) Stratum pyramidale (sp). Two‐way ANOVA for treatment/time point p < .001, hemisphere p < .0001. (k) Strata lacunosum moleculare, radiatum, and lucidum (slm/sr/sl). Two‐way ANOVA: treatment/time point p < .001, hemisphere p < .0001. Comparison for treatment/time point was made with a Tukey's multiple comparison test: **p < .01, ***p < .001; pairwise comparison was made with a Šidák's test: ###p < .001. Scale bar for (a–h) 100 μm. GCL, granule cell layer; CA2, cornu ammonis 2; slm/sr/sl, stratum lacunosum moleculare/stratum radiatum/stratum lucidum; sp, stratum pyramidale; so, stratum oriensTogether these data indicate the loss of local Gad67 mRNA expression exclusively in ipsilateral CA2 immediately after KA injection with an eventual small recovery in the chronic stage.DISCUSSIONThe degree of preservation of CA2 has traditionally been used among other criteria for classification of hippocampal sclerosis in MTLE (Blümcke et al., 2013; Thom, 2014; Wyler et al., 1992). It is, however, less well known how CA2 is integrated into the epileptic network. In the present study we have used the intrahippocampal KA mouse model for MTLE to shine light on structural and functional determinants of the integration of CA2 in the epileptic network of both hippocampi. We show that EA is generated locally in CA2 and CSD analysis of EA in CA2 revealed two spatially and temporally segregated sinks in distal and proximal dendrites of CA2 pyramidal cells. The projections from the ipsilateral CA2 region to the contralateral hippocampus were preserved despite hippocampal sclerosis; however, in the contralateral hippocampus, the fraction of time spent in EA and the severity of events was substantially lower than in the ipsilateral hippocampus. In accordance with this hemispheric difference, our time‐course analysis of Gad67‐positive cell density showed a massive reduction of cells already at 2 days after KA only in ipsilateral CA2. When investigating the coupling of EA, we found that the ipsilateral coupling was always stronger than with contralateral CA2. On the ipsilateral side, severe events in the DG were reliably accompanied by EA in CA2 but in the converse direction, the EA overlap increased over time.Analysis of theta episodes revealed that CA2 actively participates in theta rhythm and shows the same frequency reduction observed in other hippocampal areas.Changes in CA2 connectivity and structure under epileptic conditionsEarlier we have shown that the length of the CA2 region was increased at 14 days after KA injection (but not at 7 days after KA) and remained constant afterward, indicating a transient phase of dispersion around 14 days (Häussler et al., 2016). Similarly, sprouted mossy fiber terminals which target CA2 pyramidal cell somata can be observed from ~14 days onward. When analyzing the expression of Bdnf mRNA, a marker for structural and functional plasticity mechanisms under epileptic conditions (reviewed in Binder et al., 2001), we found the upregulation at 2 and 14 days after KA injection but not at 7 and 21 days (Tulke et al., 2019). Together these data point to an early, most likely SE‐induced and a later, transient phase of high structural plasticity in CA2.To further analyze this, we here performed a time‐resolved analysis of the density of Gad67 mRNA‐expressing neurons in CA2 and found at strong reduction already 2 days after KA injection similar to what we have observed in the hilus and entire hippocampus (Marx et al., 2013), which indicates that SE has a devastating effect on interneurons also locally in CA2. Interestingly, there seems to be a partial recovery of Gad67 mRNA‐expressing neurons which was restricted to CA2 and distal CA3 at 21 days. There was, thus, no complete loss of GABAergic interneurons in CA2 after KA injection, but instead some interneurons might have only temporally lost the expression of the GABA synthesizing enzyme and therefore probably also the ability for GABAergic transmission. These data match earlier observations using magnetic resonance spectroscopy where the recovery of the GABA concentration in the hippocampus has been shown at >16 days after KA injection after a transient loss of GABA in the hippocampus (Hamelin et al., 2021; Janz, Schwaderlapp, et al., 2017). Although there is the very unlikely option that the loss of GABAergic interneurons is followed by migration of interneurons from other areas, it is much more plausible that some interneurons undergo recovery processes and restart the transcription of Gad67 mRNA and accordingly the production of GABA. This would be in agreement with the concept of dormant interneurons (Sloviter, 1991) or transient reduction of expression of markers for inhibitory neurons in humans (Wittner et al., 2001). It remains to be determined whether our observations on the mRNA level are accompanied by changes in the synaptic input to CA2 pyramidal cells. Together with our previous studies, our current results support the idea of an early phase of severe damage, followed by a second phase of structural alterations inducing aberrant connectivity between 7 and 14 days and eventually a third phase of partial recovery with the aim to reach the “least epileptic state possible.” Interestingly, this also matches the time course of severe EA described here and the time course of epileptic spike rate and burst length in an earlier study (Janz et al., 2018).Concerning the outbound connectivity of CA2 pyramidal cells, we found a reduction of projections from ipsilateral CA2 to ipsilateral CA3/CA1 in agreement with cell loss in these areas, but the preservation of projections to contralateral CA3, CA2, and CA1 at 21 days after KA injection. Importantly, the DV extent of ipsi‐ and contralateral hippocampal projections was comparable to control, indicating that despite the local changes, CA2 pyramidal cells remained part of a network beyond the ipsilateral hippocampus. Certainly, our current study does not provide a comprehensive analysis of all CA2 projections under epileptic conditions since those to the subiculum (own unpublished observations), to ventral CA1 (Meira et al., 2018), to the septum (Leroy et al., 2018) and other targets (Cui et al., 2013) which might also play an important role for EA propagation were not analyzed here.Active role of CA2 in epilepsyThere are strong indications that CA2 plays an active role in MTLE since increased intrinsic excitability as well as increased excitatory input to and output from CA2 have been shown in the pilocarpine mouse model (Whitebirch et al., 2022). Similarly, in hippocampal slices from human specimens resected during curative MTLE surgery CA2 generated spontaneous interictal activity (Knowles et al., 1987; Reyes‐Garcia et al., 2018; Wittner et al., 2009). Our layer‐resolved silicon probe recordings are in line with these observations showing local EA in CA2. We observed a conspicuous pattern of sink‐source‐distribution in epileptic spikes in the CSD analysis: two successive sinks in slm and sl/sr point toward distinct inputs arriving in sequence. We hypothesize that the sink in slm reflects input from the medial entorhinal cortex which arrives in the DG and in CA2 nearly simultaneously and activates both. Under normal conditions, entorhinal input to CA2 is strong and highly efficient in inducing action potentials (Chevaleyre & Siegelbaum, 2010; Srinivas et al., 2017; Sun et al., 2014) and entorhinal input nearly simultaneously induced field responses in the DG and CA2 but not in CA3 or CA1 (Bartesaghi & Gessi, 2004). Anterograde tracing from the medial entorhinal cortex confirmed the preservation of these synaptic inputs to CA2 in the intrahippocampal KA model (Janz, Savanthrapadian, et al., 2017). We can only speculate on their efficiency under epileptic conditions, but sprouting of entorhinal projections, formation of novel synapses and strengthening of a subset thereof in the DG of the intrahippocampal KA model (Janz, Savanthrapadian, et al., 2017) are in favor of preservation or even strengthening of projections to CA2, too.The simultaneous activation of granule cells might lead to subsequent propagation of DG activity via the mossy fibers to CA2 which is reflected in a current sink in sl/sr in CA2 following the sink in slm with some delay. Under normal conditions mossy fiber transmission onto CA2 is weak and mostly activates feed‐forward inhibition (Kohara et al., 2014; Sun et al., 2017). However, considering the reduction of GABAergic interneurons shown here and the sprouting of mossy fibers in CA2 (Häussler et al., 2016), a strengthening of this input, similar to the pilocarpine model (Whitebirch et al., 2022), is conceivable. A contribution of input from CA3 to this second peak is unlikely due to the strong loss of CA3 pyramidal cells, but it cannot be excluded, in particular with respect to input from more temporal or contralateral CA3 where tissue preservation is better. Similarly, we cannot rule out a component from the recurrent CA2 connections (Okamoto & Ikegaya, 2019). The test of our hypotheses would require simultaneous recordings in the DG and CA2 with placement of probes parallel to the cells' orientation for allow for more detailed analysis of sequential inputs.We were wondering why severe EA in the DG had a high likelihood to occur together with EA in CA2 from early on, whereas the opposite coupling increased over time. Taking into account that we implanted silicon probes only in late stages to avoid strong tissue reorganization, we can only hypothesize on underlying mechanisms and recordings of the whole time course with silicon probes will be necessary in the future. The reliable transmission from DG to CA2 from early on is in agreement with the direct connectivity of the two regions via the mossy fibers together with the early loss of GABAergic interneurons—and with that most likely also the loss of feed‐forward inhibition which suppresses this transmission under healthy conditions (Kohara et al., 2014). The time course of the increase in coupling in the opposite direction—from CA2 to DG—matches the time course of ongoing sprouting processes, including mossy fiber sprouting (Häussler et al., 2016) and sprouting of perforant path synapses onto the DG (Janz, Savanthrapadian, et al., 2017)—and possibly onto CA2, too. In addition, CA2 projections which lose their target cells in CA1 might also sprout to the remaining ipsilateral targets in a similar time frame (own unpublished observations). Together, this may develop a network of high interconnectivity and recurrence, which favors reciprocal synchronization.Concerning the role of CA2 as a possible gate for propagation of EA, we found that EA in contralateral CA2 is less severe at all time points. This indicates that the preservation of GABAergic cells in contralateral CA2 is efficient in dampening EA there. Yet, more severe events from the ipsilateral side might have a higher probability to propagate to the contralateral side because it has been shown that repeated stimulation of the inputs to CA2 results in long‐term depression of inhibitory neurons in the CA2 region (Nasrallah et al., 2017). This could explain the increase of propagation from ipsi‐ to contralateral CA2 between 2 and 7 days, when the incidence of more severe events increases on the ipsilateral side but the later increase. Although we did not observe an increase in the density of CA2‐CA2 projections under epileptic conditions—which might be due to a slight reduction of CA2 pyramidal cells under epileptic conditions and thus a lower density of virus‐expressing cells in ipsilateral CA2, to different virus expression properties or to altered tamoxifen effects under epileptic conditions—we cannot exclude that sprouting of projections from ipsi‐ to contralateral CA2 might take place due to the loss of ipsilateral targets. It remains to be determined whether the balance of ipsilateral projections to contralateral CA2 in changed in favor of excitation. In addition, our data do not rule out the propagation via other pathways, for example, via the few remaining CA3 cells in the dorsal hippocampus, via ventral CA3, via mossy cells from the ventral hippocampus, where they are partially preserved (Volz et al., 2011) or other polysynaptic pathways.Non‐EA in the epileptic networkIn the earlier work, we have shown a reduction in theta frequency by more than 1 Hz in chronic epilepsy (>16 days after KA) in the DG of both hemispheres and in the medial entorhinal cortex (Kilias et al., 2018). Here, we have included the CA2 region of both hippocampi and expanded our analysis to earlier time points to see whether it is a network‐wide phenomenon and to determine whether structural reorganization within the hippocampal formation might contribute to the slowdown of the rhythm. The frequency reduction at all positions was already visible at ≤48 h and persisted without any changes in theta peak frequency throughout the experiment, ruling out sprouting as a potential cause. Cell death in the hippocampal formation, in particular loss of interneurons as the predominant target of theta‐rhythmic inputs, has nearly completely developed at this time point (Bouilleret et al., 1999; Franz et al., 2022; Marx et al., 2013; Suzuki et al., 1995). As shown here for CA2, the loss of interneurons is predominantly a focal phenomenon sparing the contralateral hippocampus, making an early damage of extrahippocampal theta pacemakers, like the medial septum during/after SE a more plausible explanation. The early reduction is in agreement with a theta frequency reduction at 4 days after pilocarpine injection in rats which is further supporting its independence from local hippocampal sclerosis because cell loss in this systemic model is uniform across hemispheres and restricted to the hilus and only parts of CA1 and CA3 (Chauvière et al., 2009).Interestingly, it was not possible to detect distinct theta periods in the ipsilateral hippocampus of two‐thirds of the mice at ≤48 h after KA injection, but theta recovered in the following. We assume that rather SE‐induced short‐term effects on intracellular properties (LeDuigou et al., 2005) might account for this temporary loss of theta rhythmicity than long‐lasting changes in cell resonance characteristics (Marcelin et al., 2009).Despite a clear theta peak at all recording sites at 14 days after KA, we found that the coherence peak in theta range was less pronounced than at all other time points, in particular for coupling of ipsilateral DG and CA2. Interestingly, this coincides with the time point of strongest synaptic reorganization and Bdnf‐associated plasticity in CA2 (Häussler et al., 2016; Tulke et al., 2019). It is therefore conceivable that projections from theta pacemakers also undergo intermittent plasticity processes due to loss of their hippocampal targets. Indeed, neurons from the medial septum or the supramammillary nucleus (SUM) strongly project onto GABAergic interneurons in CA2 and induce transient disinhibition (GABAergic septal projections) or feed‐forward inhibition (glutamatergic projections from SUM; Robert et al., 2021) of CA2 pyramidal cells. The loss of GABAergic target cells might induce sprouting and synapse formation onto CA2 pyramidal cells which might strongly disturb the balanced system of inputs that guarantees theta oscillations under normal conditions.Although there is no direct evidence for behavioral or mnemonic impairment associated with a slowed theta rhythm under epileptic conditions (Shuman et al., 2017), theta stimulation has been reported to have beneficial effects on spatial learning (Lee et al., 2017). Thus, given that CA2 firing is strongly coupled to theta rhythm (Oliva et al., 2016), social cognition deficits observed in patients (Bora & Meletti, 2016) and TLE models (Mikulecká et al., 2019) could relate to the observed rhythmopathy in CA2.In summary, our data show that CA2 is an active part of the epileptic network that might contribute to the propagation of EA to brain areas beyond the sclerotic hippocampus. Furthermore, we propose that early SE‐induced cell death plays an important role for epileptic and theta activity not only in CA2 but also in the entire hippocampal formation and is followed by a phase of compensatory sprouting, reflected in changes of reciprocal interaction of DG and CA2.ACKNOWLEDGMENTSThe authors are grateful to Prof Dr Carola A. Haas (Medical Center—University of Freiburg) for providing valuable input, lab equipment and materials. The Amigo2‐icreERT2 mouse line was a kind gift from Dr Serena Dudek and Dr Georgia Alexander (NIEHS, Durham, NC). The authors thank Tobias Holzhammer for the assembly of silicon probes with polyimide‐based ribbon cables. The work was funded by the German Research Foundation (DFG grant HA7597) and as a part of the Cluster of Excellence “BrainLinks‐BrainTools” within the framework of the German Excellence Initiative (grant number EXC1086 to U.H. and P.R.), by BrainLinks‐BrainTools, which is funded by the Federal Ministry of Economics, Science and Arts of Baden‐Württemberg within the sustainability program for projects of the excellence initiative II, and a grant from the Research Commission of the Medical Faculty, University of Freiburg. NB received a PhD scholarship from the Center for Basics in Neuromodulation, Freiburg. Open Access funding enabled and organized by Projekt DEAL.DATA AVAILABILITY STATEMENTThe data that support the findings of this study are available from the corresponding author upon request.REFERENCESAlexander, G. M., Brown, L. Y., Farris, S., Lustberg, D., Pantazis, C., Gloss, B., Plummer, N. W., Jensen, P., & Dudek, S. M. (2018). CA2 neuronal activity controls hippocampal low gamma and ripple oscillations. eLife, 7, e38052. https://doi.org/10.7554/eLife.38052Althaus, A. L., Zhang, H., & Parent, J. M. (2016). Axonal plasticity of age‐defined dentate granule cells in a rat model of mesial temporal lobe epilepsy. Neurobiology of Disease, 86, 187–196. https://doi.org/10.1016/j.nbd.2015.11.024Bartesaghi, R., & Gessi, T. (2004). Parallel activation of field CA2 and dentate gyrus by synaptically elicited perforant path volleys. Hippocampus, 14, 948–963. https://doi.org/10.1002/hipo.20011Binder, D. K., Croll, S. D., Gall, C. M., & Scharfman, H. E. (2001). BDNF and epilepsy: Too much of a good thing? Trends in Neurosciences, 24, 47–53. https://doi.org/10.1016/S0166-2236(00)01682-9Blümcke, I., Thom, M., Aronica, E., Armstrong, D. D., Bartolomei, F., Bernasconi, A., Bernasconi, N., Bien, C. G., Cendes, F., Coras, R., Cross, J. H., Jacques, T. S., Kahane, P., Mathern, G. W., Miyata, H., Moshé, S. L., Oz, B., Özkara, Ç., Perucca, E., … Spreafico, R. (2013). International consensus classification of hippocampal sclerosis in temporal lobe epilepsy: A task force report from the ILAE Commission on Diagnostic Methods. Epilepsia, 54, 1315–1329. https://doi.org/10.1111/epi.12220Bora, E., & Meletti, S. (2016). Social cognition in temporal lobe epilepsy: A systematic review and meta‐analysis. Epilepsy & Behavior, 60, 50–57. https://doi.org/10.1016/j.yebeh.2016.04.024Botcher, N. A., Falck, J. E., Thomson, A. M., & Mercer, A. (2014). Distribution of interneurons in the CA2 region of the rat hippocampus. Frontiers in Neuroanatomy, 8, 104. https://doi.org/10.3389/fnana.2014.00104Bouilleret, V., Ridoux, V., Depaulis, A., Marescaux, C., Nehlig, A., & Le Gal La Salle, G. (1999). Recurrent seizures and hippocampal sclerosis following intrahippocampal kainate injection in adult mice: Electroencephalography, histopathology and synaptic reorganization similar to mesial temporal lobe epilepsy. Neuroscience, 89, 717–729. https://doi.org/10.1016/S0306-4522(98)00401-1Carstens, K. E., Lustberg, D. J., Shaughnessy, E. K., McCann, K. E., Alexander, G. M., & Dudek, S. M. (2021). Perineuronal net degradation rescues CA2 plasticity in a mouse model of Rett syndrome. The Journal of Clinical Investigation, 131(16), e137221. https://doi.org/10.1172/JCI137221Carstens, K. E., Phillips, M. L., Pozzo‐Miller, L., Weinberg, R. J., & Dudek, S. M. (2016). Perineuronal nets suppress plasticity of excitatory synapses on CA2 pyramidal neurons. The Journal of Neuroscience, 36, 6312–6320. https://doi.org/10.1523/JNEUROSCI.0245-16.2016Chauvière, L., Rafrafi, N., Thinus‐Blanc, C., Bartolomei, F., Esclapez, M., & Bernard, C. (2009). Early deficits in spatial memory and theta rhythm in experimental temporal lobe epilepsy. The Journal of Neuroscience, 29, 5402–5410. https://doi.org/10.1523/JNEUROSCI.4699-08.2009Chevaleyre, V., & Siegelbaum, S. A. (2010). Strong CA2 pyramidal neuron synapses define a powerful disynaptic cortico‐hippocampal loop. Neuron, 66, 560–572. https://doi.org/10.1016/j.neuron.2010.04.013Cui, Z., Gerfen, C. R., & Young, W. S. (2013). Hypothalamic and other connections with dorsal CA2 area of the mouse hippocampus. The Journal of Comparative Neurology, 521, 1844–1866. https://doi.org/10.1002/cne.23263Engel, J. (2001). A proposed diagnostic scheme for people with epileptic seizures and with epilepsy: Report of the ILAE Task Force on classification and terminology. Epilepsia, 42, 796–803. https://doi.org/10.1046/j.1528-1157.2001.10401.xFranklin, K. B. J., & Paxinos, G. (2007). The mouse brain in stereotaxic coordinates (3rd ed.). Academic Press.Franz, J., Barheier, N., Tulke, S., Haas, C. A., & Häussler, U. (2022). Differential vulnerability of neuronal subpopulations of the subiculum in a mouse model for mesial temporal lobe epilepsy. Biorxiv:2022.06.02.494518. https://doi.org/10.1101/2022.06.02.494518Freiman, T. M., Häussler, U., Zentner, J., Doostkam, S., Beck, J., Scheiwe, C., Brandt, A., Haas, C. A., & Puhahn‐Schmeiser, B. (2021). Mossy fiber sprouting into the hippocampal region CA2 in patients with temporal lobe epilepsy. Hippocampus, 31, 580–592. https://doi.org/10.1002/hipo.23323Hamelin, S., Stupar, V., Mazière, L., Guo, J., Labriji, W., Liu, C., Bretagnolle, L., Parrot, S., Barbier, E. L., Depaulis, A., & Fauvelle, F. (2021). In vivo γ‐aminobutyric acid increase as a biomarker of the epileptogenic zone: An unbiased metabolomics approach. Epilepsia, 62(1), 163–175. https://doi.org/10.1111/epi.16768Häussler, U., Bielefeld, L., Froriep, U. P., Wolfart, J., & Haas, C. A. (2012). Septotemporal position in the hippocampal formation determines epileptic and neurogenic activity in temporal lobe epilepsy. Cerebral Cortex, 22, 26–36. https://doi.org/10.1093/cercor/bhr054Häussler, U., Rinas, K., Kilias, A., Egert, U., & Haas, C. A. (2016). Mossy fiber sprouting and pyramidal cell dispersion in the hippocampal CA2 region in a mouse model of temporal lobe epilepsy. Hippocampus, 26, 577–588. https://doi.org/10.1002/hipo.22543Hayani, H., Song, I., & Dityatev, A. (2018). Increased excitability and reduced excitatory synaptic input into fast‐spiking CA2 interneurons after enzymatic attenuation of extracellular matrix. Frontiers in Cellular Neuroscience, 12, 149. https://doi.org/10.3389/fncel.2018.00149Heining, K., Kilias, A., Janz, P., Häussler, U., Kumar, A., Haas, C. A., & Egert, U. (2019). Bursts with high and low load of epileptiform spikes show context‐dependent correlations in epileptic mice. eNeuro, 6(5), ENEURO.0299–ENEU18.2019. https://doi.org/10.1523/ENEURO.0299-18.2019Herwik, S., Kisban, S., Aarts, A. A. A., Seidl, K., Girardeau, G., Benchenane, K., Zugaro, M. B., Wiener, S. I., Paul, O., Neves, H. P., & Ruther, P. (2009). Fabrication technology for silicon‐based microprobe arrays used in acute and sub‐chronic neural recording. Journal of Micromechanics and Microengineering, 19, 074008. https://doi.org/10.1088/0960-1317/19/7/074008Herwik, S., Paul, O., & Ruther, P. (2011). Ultrathin silicon chips of arbitrary shape by etching before grinding. Journal of Microelectromechanical Systems, 20, 791–793. https://doi.org/10.1109/JMEMS.2011.2148159Hjorth‐Simonsen, A., & Laurberg, S. (1977). Commissural connections of the dentate area in the rat. The Journal of Comparative Neurology, 174(4), 591–606. https://doi.org/10.1002/cne.901740404Houser, C. R., Miyashiro, J. E., Swartz, B. E., Walsh, G. O., Rich, J. R., & Delgado‐Escueta, A. V. (1990). Altered patterns of dynorphin immunoreactivity suggest mossy fiber reorganization in human hippocampal epilepsy. The Journal of Neuroscience, 10, 267–282. https://doi.org/10.1523/JNEUROSCI.10-01-00267.1990Janz, P., Hauser, P., Heining, K., Nestel, S., Kirsch, M., Egert, U., & Haas, C. A. (2018). Position‐ and time‐dependent arc expression links neuronal activity to synaptic plasticity during epileptogenesis. Frontiers in Cellular Neuroscience, 12, 244. https://doi.org/10.3389/fncel.2018.00244Janz, P., Savanthrapadian, S., Häussler, U., Kilias, A., Nestel, S., Kretz, O., Kirsch, M., Bartos, M., Egert, U., & Haas, C. A. (2017). Synaptic remodeling of entorhinal input contributes to an aberrant hippocampal network in temporal lobe epilepsy. Cerebral Cortex, 27, 2348–2364. https://doi.org/10.1093/cercor/bhw093Janz, P., Schwaderlapp, N., Heining, K., Häussler, U., Korvink, J. G., von Elverfeldt, D., Hennig, J., Egert, U., LeVan, P., & Haas, C. A. (2017). Early tissue damage and microstructural reorganization predict disease severity in experimental epilepsy. eLife, 6, 25742. https://doi.org/10.7554/eLife.25742Kilias, A., Häussler, U., Heining, K., Froriep, U. P., Haas, C. A., & Egert, U. (2018). Theta frequency decreases throughout the hippocampal formation in a focal epilepsy model. Hippocampus, 28, 375–391. https://doi.org/10.1002/hipo.22838Knowles, W. D., Traub, R. D., & Strowbridge, B. W. (1987). The initiation and spread of epileptiform bursts in the in vitro hippocampal slice. Neuroscience, 21, 441–455. https://doi.org/10.1016/0306-4522(87)90134-5Kohara, K., Pignatelli, M., Rivest, A. J., Jung, H.‐Y., Kitamura, T., Suh, J., Frank, D., Kajikawa, K., Mise, N., Obata, Y., Wickersham, I. R., & Tonegawa, S. (2014). Cell type‐specific genetic and optogenetic tools reveal hippocampal CA2 circuits. Nature Neuroscience, 17, 269–279. https://doi.org/10.1038/nn.3614Kulik, A., Vida, I., Luján, R., Haas, C. A., López‐Bendito, G., Shigemoto, R., & Frotscher, M. (2003). Subcellular localization of metabotropic GABA(B) receptor subunits GABA(B1a/b) and GABA(B2) in the rat hippocampus. The Journal of Neuroscience, 23, 11026–11035. https://doi.org/10.1523/JNEUROSCI.23-35-11026.2003Lanzilotto, M., Livi, A., Maranesi, M., Gerbella, M., Barz, F., Ruther, P., Fogassi, L., Rizzolatti, G., & Bonini, L. (2016). Extending the cortical grasping network: Pre‐supplementary motor neuron activity during vision and grasping of objects. Cerebral Cortex, 26, 4435–4449. https://doi.org/10.1093/cercor/bhw315LeDuigou, C., Wittner, L., Danglot, L., & Miles, R. (2005). Effects of focal injection of kainic acid into the mouse hippocampus in vitro and ex vivo. Journal of Physiology (London), 569, 833–847. https://doi.org/10.1113/jphysiol.2005.094599Lee, D. J., Izadi, A., Melnik, M., Seidl, S., Echeverri, A., Shahlaie, K., & Gurkoff, G. G. (2017). Stimulation of the medial septum improves performance in spatial learning following pilocarpine‐induced status epilepticus. Epilepsy Research, 130, 53–63. https://doi.org./10.1016/j.eplepsyres.2017.01.005Lee, S. E., Simons, S. B., Heldt, S. A., Zhao, M., Schroeder, J. P., Vellano, C. P., Cowan, D. P., Ramineni, S., Yates, C. K., Feng, Y., Smith, Y., Sweatt, J. D., Weinshenker, D., Ressler, K. J., Dudek, S. M., & Hepler, J. R. (2010). RGS14 is a natural suppressor of both synaptic plasticity in CA2 neurons and hippocampal‐based learning and memory. PNAS, 107, 16994–16998. https://doi.org/10.1073/pnas.1005362107Leranth, C., & Ribak, C. E. (1991). Calcium‐binding proteins are concentrated in the CA2 field of the monkey hippocampus: A possible key to this region's resistance to epileptic damage. Experimental Brain Research, 85, 129–136. https://doi.org/10.1007/BF00229993Leroy, F., Park, J., Asok, A., Brann, D. H., Meira, T., Boyle, L. M., Buss, E. W., Kandel, E. R., & Siegelbaum, S. A. (2018). A circuit from hippocampal CA2 to lateral septum disinhibits social aggression. Nature, 564, 213–218. https://doi.org/10.1038/s41586-018-0772-0Lisgaras, C. P., & Scharfman, H. E. (2022). Robust chronic convulsive seizures, high frequency oscillations, and human seizure onset patterns in an intrahippocampal kainic acid model in mice. Neurobiology of Disease, 166, 105637. https://doi.org/10.1016/j.nbd.2022.105637Marcelin, B., Chauvière, L., Becker, A., Migliore, M., Esclapez, M., & Bernard, C. (2009). h channel‐dependent deficit of theta oscillation resonance and phase shift in temporal lobe epilepsy. Neurobiology of Disease, 33, 436–447. https://doi.org/10.1016/j.nbd.2008.11.019Marx, M., Haas, C. A., & Häussler, U. (2013). Differential vulnerability of interneurons in the epileptic hippocampus. Frontiers in Cellular Neuroscience, 7, 167. https://doi.org/10.3389/fncel.2013.00167Meira, T., Leroy, F., Buss, E. W., Oliva, A., Park, J., & Siegelbaum, S. A. (2018). A hippocampal circuit linking dorsal CA2 to ventral CA1 critical for social memory dynamics. Nature Communications, 9, 4163. https://doi.org/10.1038/s41467-018-06501-wMikulecká, A., Druga, R., Stuchlík, A., Mareš, P., & Kubová, H. (2019). Comorbidities of early‐onset temporal epilepsy: Cognitive, social, emotional, and morphologic dimensions. Experimental Neurology, 320, 113005. https://doi.org/10.1016/j.expneurol.2019.113005Nasrallah, K., Piskorowski, R. A., & Chevaleyre, V. (2017). Bi‐directional interplay between proximal and distal inputs to CA2 pyramidal neurons. Neurobiology of Learning and Memory, 138, 173–181. https://doi.org/10.1016/j.nlm.2016.06.024Oh, S. W., Harris, J. A., Ng, L., Winslow, B., Cain, N., Mihalas, S., Wang, Q., Lau, C., Kuan, L., Henry, A. M., Mortrud, M. T., Ouellette, B., Nguyen, T. N., Sorensen, S. A., Slaughterbeck, C. R., Wakeman, W., Li, Y., Feng, D., Ho, A., … Zeng, H. (2014). A mesoscale connectome of the mouse brain. Nature, 508, 207–214. https://doi.org/10.1038/nature13186Okamoto, K., & Ikegaya, Y. (2019). Recurrent connections between CA2 pyramidal cells. Hippocampus, 29, 305–312. https://doi.org/10.1002/hipo.23064Oliva, A., Fernández‐Ruiz, A., Buzsáki, G., & Berényi, A. (2016). Spatial coding and physiological properties of hippocampal neurons in the cornu ammonis subregion. Hippocampus, 26, 1593–1607. https://doi.org/10.1002/hipo.22659Reyes‐Garcia, S. Z., Scorza, C. A., Araújo, N. S., Ortiz‐Villatoro, N. N., Jardim, A. P., Centeno, R., Yacubian, E. M. T., Faber, J., & Cavalheiro, E. A. (2018). Different patterns of epileptiform‐like activity are generated in the sclerotic hippocampus from patients with drug‐resistant temporal lobe epilepsy. Scientific Reports, 8, 7116. https://doi.org/10.1038/s41598-018-25378-9Robert, V., Therreau, L., Chevaleyre, V., Lepicard, E., Viollet, C., Cognet, J., Huang, A. J., Boehringer, R., Polygalov, D., McHugh, T. J., & Piskorovski, R. A. (2021). Local circuit allowing hypothalamic control of hippocampal area CA2 activity and consequences for CA1. eLife, 10, e63352. https://doi.org/10.7554/eLife.63352Shuman, T., Amendolara, B., & Golshani, P. (2017). Theta rhythmopathy as a cause of cognitive disability in TLE. Epilepsy Currents, 17, 107–111. https://doi.org./10.5698/1535-7511.17.2.107Simons, S. B., Escobedo, Y., Yasuda, R., & Dudek, S. M. (2009). Regional differences in hippocampal calcium handling provide a cellular mechanism for limiting plasticity. Proceedings of the National Academy of Sciences of the United States of America, 106, 14080–14084. https://doi.org/10.1073/pnas.0904775106Sloviter, R. S. (1991). Permanently altered hippocampal structure, excitability, and inhibition after experimental status epilepticus in the rat: The “dormant basket cell” hypothesis and its possible relevance to temporal lobe epilepsy. Hippocampus, 1, 41–66. https://doi.org/10.1002/hipo.450010106Srinivas, K. V., Buss, E. W., Sun, Q., Santoro, B., Takahashi, H., Nicholson, D. A., & Siegelbaum, S. A. (2017). The dendrites of CA2 and CA1 pyramidal neurons differentially regulate information flow in the cortico‐hippocampal circuit. The Journal of Neuroscience, 37, 3276–3293. https://doi.org/10.1523/JNEUROSCI.2219-16.2017Steve, T. A., Jirsch, J. D., & Gross, D. W. (2014). Quantification of subfield pathology in hippocampal sclerosis: A systematic review and meta‐analysis. Epilepsy Research, 108, 1279–1285. https://doi.org/10.1016/j.eplepsyres.2014.07.003Sun, Q., Sotayo, A., Cazzulino, A. S., Snyder, A. M., Denny, C. A., & Siegelbaum, S. A. (2017). Proximodistal heterogeneity of hippocampal CA3 pyramidal neuron intrinsic properties, connectivity, and reactivation during memory recall. Neuron, 95, 656–672.e3. https://doi.org/10.1016/j.neuron.2017.07.012Sun, Q., Srinivas, K. V., Sotayo, A., & Siegelbaum, S. A. (2014). Dendritic Na+ spikes enable cortical input to drive action potential output from hippocampal CA2 pyramidal neurons. eLife, 3, 04551. https://doi.org/10.7554/eLife.04551Suzuki, F., Junier, M. P., Guilhem, D., Sørensen, J. C., & Onteniente, B. (1995). Morphogenetic effect of kainate on adult hippocampal neurons associated with a prolonged expression of brain‐derived neurotrophic factor. Neuroscience, 64, 665–674. https://doi.org/10.1016/0306-4522(94)00463-FThom, M. (2014). Hippocampal sclerosis in epilepsy: A neuropathology review. Neuropathology and Applied Neurobiology, 40, 520–543. https://doi.org/10.1111/nan.12150Tulke, S., Haas, C. A., & Häussler, U. (2019). Expression of brain‐derived neurotrophic factor and structural plasticity in the dentate gyrus and CA2 region correlate with epileptiform activity. Epilepsia, 60, 1234–1247. https://doi.org/10.1111/epi.15540Vertes, R. P. (2005). Hippocampal theta rhythm: A tag for short‐term memory. Hippocampus, 15, 923–935. https://doi.org/10.1002/hipo.20118Volz, F., Bock, H. H., Gierthmuelen, M., Zentner, J., Haas, C. A., & Freiman, T. M. (2011). Stereologic estimation of hippocampal GluR2/3‐ and calretinin‐immunoreactive hilar neurons (presumptive mossy cells) in two mouse models of temporal lobe epilepsy. Epilepsia, 52(9), 1579–1589. https://doi.org/10.1111/j.1528-1167.2011.03086.xWhitebirch, A. C., LaFrancois, J. J., Jain, S., Leary, P., Santoro, B., Siegelbaum, S. A., & Scharfman, H. E. (2022). Enhanced excitability of the hippocampal CA2 region and its contribution to seizure activity in a mouse model of temporal lobe epilepsy. Neuron, 110(19), 3121–3138.e8. https://doi.org/10.1016/j.neuron.2022.07.020Williamson, A., & Spencer, D. D. (1994). Electrophysiological characterization of CA2 pyramidal cells from epileptic humans. Hippocampus, 4, 226–237. https://doi.org/10.1002/hipo.450040213Wittner, L., Huberfeld, G., Clémenceau, S., Eross, L., Dezamis, E., Entz, L., Ulbert, I., Baulac, M., Freund, T. F., Maglóczky, Z., & Miles, R. (2009). The epileptic human hippocampal cornu ammonis 2 region generates spontaneous interictal‐like activity in vitro. Brain, 132, 3032–3046. https://doi.org/10.1155/2017/7154295Wittner, L., Maglóczky, Z., Borhegyi, Z., Halász, P., Tóth, S., Eross, L., Szabó, Z., & Freund, T. F. (2001). Preservation of perisomatic inhibitory input of granule cells in the epileptic human dentate gyrus. Neuroscience, 108, 587–600. https://doi.org/10.1016/S0306-4522(01)00446-8Wyler, A. R., Curtis, D. F., Schweitzer, J. B., & Berry, A. D. (1992). A grading system for mesial temporal pathology (hippocampal sclerosis) from anterior temporal lobectomy. Journal of Epilepsy, 5, 220–225. https://doi.org/10.1016/S0896-6974(05)80120-3

Journal

HippocampusWiley

Published: Nov 24, 2022

Keywords: Amigo2; dentate gyrus; interneurons; mesial temporal lobe epilepsy; mossy fiber sprouting; seizures; theta oscillations

References