Enhanced hippocampal type II theta activity AND altered theta architecture in mice lacking the Cav3.2 T-type voltage-gated calcium channel

T-type Ca2+ channels are assumed to contribute to hippocampal theta oscillations. We used implantable video-EEG radiotelemetry and qPCR to unravel the role of Cav3.2 Ca2+ channels in hippocampal theta genesis. Frequency analysis of spontaneous long-term recordings in controls and Cav3.2−/− mice revealed robust increase in relative power in the theta (4–8 Hz) and theta-alpha (4–12 Hz) ranges, which was most prominent during the inactive stages of the dark cycles. Urethane injection experiments also showed enhanced type II theta activity and altered theta architecture following Cav3.2 ablation. Next, gene candidates from hippocampal transcriptome analysis of control and Cav3.2−/− mice were evaluated using qPCR. Dynein light chain Tctex-Type 1 (Dynlt1b) was significantly reduced in Cav3.2−/− mice. Furthermore, a significant reduction of GABA A receptor δ subunits and GABA B1 receptor subunits was observed in the septohippocampal GABAergic system. Our results demonstrate that ablation of Cav3.2 significantly alters type II theta activity and theta architecture. Transcriptional changes in synaptic transporter proteins and GABA receptors might be functionally linked to the electrophysiological phenotype.


Temperature profile in Ca v 3.2 +/+ and Ca v 3.2 −/− mice.
Besides biopotentials and activity, the TA10ETA-F20 transmitter is capable of recording body temperature. The latter were averaged for the DC and LC for both genotypes. As expected, temperature values in Ca v 3.2 +/+ and Ca v 3.2 −/− mice mimicked those results obtained for activity ( Figs. 2A,B, 3A,B). As mice are nocturnal animals, both genotypes exhibited significant temperature increase during the DC1 compared to the LC1 of baseline recording R1 (Ca v 3.2 +/+ : 35.12 ± 0.13 °C (DC1) vs. 34.55 ± 0.11 °C (LC1), p < 0.0001; Ca v 3.2 −/− : 34.70 ± 0.24 °C (DC1) vs. 34.04 ± 0.25 °C (LC1), p = 0.0006) (Fig. 3A). No significant differences were observed within the DC1 and LC1 between both genotypes. These findings again mirror the results obtained from the activity study (Fig. 2), pointing out that no significant alterations in activity architecture exists between Ca v 3.2 +/+ and Ca v 3.2 −/− mice. Temperature analysis of R2 confirmed results from R1 with a significant increase in DC2 compared to LC2 in both genotypes (Ca v 3. FFT based frequency analysis from spontaneous EEG long-term recordings in Ca v 3.2 +/+ and Ca v 3.2 −/− . In both Ca v 3.2 +/+ and Ca v 3.2 deficient mice, two 24 h long-term (baseline) EEG recordings (R1, R2) from the CA1 region were carried out at day 10 and day 17 post implantation of the TA10ETA-F20 transmitter (Fig. 1A). This regime guarantees a recovery period of 10 days, sufficient for the animals to regain standard physiological parameters, i.a., in CNS electrophysiology and circadian patterns 48 . Subsequently, an FFT based EEG frequency analysis was performed distinguishing between the LC and DC as well as the non-active (inactive) state (NAS) and active state (AS). EEG power values are presented as relative values (%).
To confirm these results, a second urethane injection was carried out at day 25 post transmitter implantation. Again, significant increases in relative power were monitored for θ 1 , θ 2 and α frequency bands (θ 1 , 24.426 ± 3.129% (Ca v 3. As for both 24 h long-term EEG recordings, the urethane studies (U1, U2) clearly confirm an increase in θ 2 and α activity in the CA1 hippocampal area in Ca v 3.2 deficient mice.
In order to get a closer insight into the hippocampal theta/alpha architecture of Ca v 3.2 +/+ and Ca v 3.2 −/− mice, we analyzed power spectrum density (PSD) plots for theta/alpha peak frequencies. Representative PSD plots for both genotypes from the baseline and post-urethane state are depicted in Fig. 9A. Notably, the peak frequency was increased in Ca v 3. These findings indicate that there is not only an increase in theta/alpha activity in Ca v 3.2 −/− mice but also a shift in theta peak frequency and thus in global theta architecture.
Transcriptional alterations in the hippocampus of Ca v 3.2 deficient mice. To gain insight into the mechanisms of theta/alpha augmentation in Ca v 3.2 −/− mice, we performed qPCR analysis of gene candidates (Table 1). These genes were previously detected in a transcriptome analysis from the hippocampi of Ca v 3.2 +/+ and Ca v 3.2 −/− mice 49 . Importantly, Ca v 3.2 −/− mice exhibited a significant decrease in transcript levels for dynein light chain Tctex-Type 1 (Dynlt1b) by a fold change (FC) of − 5.208 (p = 0.0002) (Fig. 10B, Table 2).
As the dynein related transportome complex is involved in the transport of GABA receptors, we expanded our qPCR study including the GABA A receptor delta subunit (Gabrd), GABA A receptor gamma subunit (Gabrg2), GABA B1 receptor subunit (Gabbr1) and the GABA B2 receptor subunit (Gabbr2). In Ca v 3.2 −/− mice, a significant decrease was observed for Gabrd (FC, − 1.385; p = 0.015; Fig. 10H, Table 2) and Gabbr1 (FC, − 1.105, p = 0.010, Fig. 10J, Table 2). These findings correlate with the decrease in dynein light chain Tctex-Type 1 transcripts suggesting an overall reduction in the GABA receptor transportome complex and synaptic/extrasynaptic GABA receptor density in the hippocampus, particularly in the hippocampal interneurons.
To exclude that other T-type VGCCs contribute to the theta/alpha phenotype in Ca v 3.2 deficient mice, we also checked for compensatory alterations in Ca v 3.1 and Ca v 3.3 Ca 2+ channel transcripts. Notably, no changes were observed between both genotypes, which further stresses the idea that the theta/alpha alterations in transgenic mice are due to Ca v 3.2 ablation itself (Fig. 10, Table 2).

Discussion
VGCCs play a key role in the generation of theta oscillations in dendrites of hippocampal pyramidal cells 12 , associated with various movement related behaviors 50 , learning tasks and memory processing 51,52 . The septohippocampal circuitry involved in theta genesis acquires innervation from various brain regions to code motor and sensory information processing 6,50,53 , and can trigger the regulation of theta/alpha waves in relation to specific behavioral conditions. Recently, Gangadharan et al. reported that ablation of Ca v 3.1 VGCCs results in increased theta activity probably based on tonic inhibition of hippocampal GABAergic interneurons via septal GABAergic interneurons. The latter was hypothesized to disinhibit hippocampal pyramidal neurons and to cause increased theta activity 43 .
Although Ca v 3.1 is prominently expressed in the septohippocampal system, the expression of Ca v 3.2 clearly predominates 34,44 . Therefore, we investigated the role of Ca v 3.2 in the generation and architecture of theta/alpha activity in Ca v 3.2 deficient mice. Previous studies had suggested a complex phenotype upon Ca v 3.2 ablation including, i.a., impaired memory formation and elevated anxiety 46 . Decreased memory function in Ca v 3.2 −/− mice was originally described using two hippocampal recognition settings, i.e., the novel object recognition (NOR) and spatial object recognition (SOR) testing. The Ca v 3.2 −/− mice did not exhibit preference for the novel or the relocated object compared to wild type animals. Importantly, this altered response was not due to an impairment   46 . Given these findings and the fact that Ca v 3.2 VGCC expression outnumbers Ca v 3.1 expression in the septohippocampal system 44 , we analyzed the role of Ca v 3.2 in theta genesis and theta architecture relevant for memory formation. Using implantable EEG radiotelemetry from the hippocampal CA1 region and frequency analysis, we elaborated that Ca v 3.2 Ca 2+ channels substantially contribute to atropine-sensitive type II theta oscillations and modulate theta architecture. Thus, this is the first direct functional link between Ca v 3.2 VGCCs and rodent theta oscillation in vivo. Importantly, the significant increase in θ 2 and α relative EEG power was observed during the NAS of the LC as well as the DC of R1 and R2 and also U1 and U2. The NAS is characterized, i.a., by alert immobility, a physiological state known to exhibit hippocampal type II theta activity. Consequently, theta alterations in Ca v 3.2 −/− mice are likely to be related to atropine-sensitive type II theta. These findings are further confirmed by our urethane injections studies. Pharmacodynamically, urethane serves as a multi-target drug with both agonistic and antagonistic effects on various ligand-and voltage-gated ion channels. Whereas muscarinic and nicotinic AChRs, GABA A receptors, and glycine receptors are stimulated upon urethane injection, NMDA and AMPA receptors are inhibited 54,55 . Urethane is known to induce type II theta activity and Ca v 3.2 −/− mice www.nature.com/scientificreports/ again revealed an increase in the relative EEG power in the theta/alpha band and the theta peak frequencies in this pharmacological setting. Notably, motor activity can have an important impact on theta I/theta II distribution. It is thus important to stress that both Ca v 3.2 +/+ and Ca v 3.2 −/− mice display characteristic circadian activity profiles. No differences in activity were observed between both genotypes indicating that alterations in the hippocampal θ 2 and α band are not related to alterations in locomotion. It should be noted that besides the consistent changes in the θ 2 and α frequency bands in R1, R2, U1 and U2, inconsistent alterations were observed for α and σ bands during the LC of the AS in R1 and R2 and for θ 1 in U1 and U2. No changes were detected for the AS of the DC in R1 and R2. Anxiety related behavior is another aspect that might influence hippocampal type II theta activity associated with alert immobility. Anxiety analysis in Ca v 3.2 −/− mice using the light/dark conflict test/context 56 and spontaneous exploratory behavior analysis via open field test and the elevated-plus maze (EPM) suggested increased anxiety in Ca v 3.2 −/− mice not associated with repetitive and compulsive behaviors 46 . Importantly, these results contrast with a previous study from Choi et al. pointing out a lack of anxiety-related behavior in Ca v 3.2 −/− mice using the light/dark conflict test 57 . This apparent discrepancy might be due to the genetic background of the Ca v 3.2 −/− mice 58 and the behavioral procedure used in the two studies 43,56,57,59-61 . Also, increased anxiety does not necessarily coincide with an increase in theta activity. Ablation of the septal PLCβ 4 isoform for example caused attenuated type II theta rhythm but increased anxiety 14,19,62 . Thus, θ 2 and α alterations in Ca v 3.2 deficient mice do not seem to be attributable to potential changes in anxiety levels.
Next, we investigated the molecular mechanisms underlying the theta/alpha related changes in Ca v 3.2 −/− mice. In general, VGCCs are crucial for LTP, learning and memory functions [63][64][65][66] . Disruption of T-type Ca 2+ channel activity was shown to severely change the induction and maintenance of LTP in the hippocampus, visual cortex and cerebellum 45,67,68 . Furthermore, T-type Ca 2+ channels interfere with the neurotransmitter release machinery and modulate synaptic transmission [69][70][71] . Recently, Gangadharan et al. found that Ca v 3.1 −/− mice exhibit increased type II theta activity. This increase was related to a shift in the firing pattern of septal GABAergic interneurons from the burst mode to the tonic mode. LVA T-type Ca 2+ channels are known to mediate low-threshold Ca 2+ spikes and burst activity [72][73][74] . Thus, ablation of Ca v 3.1 resulted in tonic inhibition of hippocampal GABAergic interneurons via projecting septal GABAergic interneurons. Subsequent perisomatic disinhibition of hippocampal pyramidal neurons was supposed to enhance theta activity in Ca v 3.1 −/− mice 43,75,76 . Given the fact, that Ca v 3.2 expression outnumbers the expression of Ca v 3.1 in the septohippocampal system 44 , we hypothesized that Ca v 3.2 ablation causes a similar sequence of septal GABAergic tonic inhibition and disinhibition of pyramidal cells as observed in Ca v 3.1 −/− mice. To confirm this mechanism of action in Ca v 3.2 −/− mice, we had a closer look at the functional aspects of the septohippocampal system once again. We first investigated the outcomes of our previous transcriptome analysis from the hippocampus of Ca v 3.2 +/+ and Ca v 3.2 −/− mice 49 . qPCR analysis revealed a significant reduction in dynein light chain Tctex-Type 1 (Dynlt1b) in Ca v 3.2 −/− mice. The dynein light chain is part of a GABA receptor transportome complex mediating the translocation of GABA receptors to the subsynaptic or extrasynaptic membrane [77][78][79] . This was a first indication that the GABAergic system is indeed altered in the septohippocampal system of Ca v 3.2 deficient mice. Therefore, we next analyzed transcript levels of GABA A and GABA B receptors. In Ca v 3.2 −/− mice, transcript levels for the GABA A receptor δ subunit and the GABA B 1 receptor subunit were significantly reduced. These findings strongly support our GABA hypothesis of enhanced θ 2 /α activity in Ca v 3.2 −/− mice, as GABA A receptor-mediated inhibition within the CNS occurs by fast synaptic transmission and sustained tonic inhibition 80,81 . As observed in dentate gyrus granule cells and thalamic neurons, extrasynaptically located GABA A receptors that contain, e.g., δ subunits, mediate tonic current that is relevant for neuronal/interneuronal excitability in response to ambient GABA concentrations [82][83][84] . On the other hand, GABA B1-subunit containing receptors can be detected within dendritic spines and mediate slow postsynaptic inhibition 85,86 . www.nature.com/scientificreports/ Importantly, we have no indication from our own microarray analysis or qPCR studies that there are compensatory transcriptional alterations of the other T-type Ca 2+ channels, i.e., Ca v 3.1 and Ca v 3.3, in the hippocampus of Ca v 3.2 −/− mice. Thus, the changes observed seem to be solely attributable to Ca v 3.2 ablation itself.
In summary, our qPCR findings might support the hypothesis that both postsynaptic and extrasynaptic GABA receptors are decreased upon tonic inhibition of hippocampal interneurons and that diminished plasma membrane density is due to an impaired dynein/GABA receptor containing transportome complex. Additional synaptic transporter studies and patch-clamp recordings in hippocampal slices are necessary to directly prove a potential septohippocampal disinhibition in Ca v 3.2 −/− mice. It should also be noted that hippocampal frequency characteristics can be age and gender specific [87][88][89] . We decided to use males in our study to avoid any potential interference with the estrous cycle in females. Hierarchical fights as a potential source of variability Our study is the first one to prove that Ca v 3.2 ablation results in increased atropine sensitive type II theta activity and altered theta architecture in the CA1 region. We hypothesize that tonic inhibition of hippocampal GABAergic interneurons and subsequent disinhibition of pyramidal cells due to Ca v 3.2 ablation might result in compensatory changes in the GABAergic system. These imply both the downregulation of the dynein containing GABA receptor transporter/trafficking complex and GABA A and B receptor complexes themselves. Notably, compensatory changes in other neuronal cell types and circuitries affecting the septohippocampal network cannot be excluded in the gobal Ca v 3.2 knockout used in our study. Recently, Dinamarca et al. 90 have shown that GABA B receptors (GBR) form complexes with amyloid precursor protein (APP). This GBR/APP complex is supposed to stabilize APP at the surface membrane and to reduce proteolysis from APP to Aβ. Impaired GABA receptor trafficking and GBR expression in Ca v 3.2 −/− mice might therefore alter APP stability in these animals. Future studies will be necessary to unravel the potential functional interdependence between T-type VGCCs, the GABAergic system and APP and its relevance in the aetiopathogenesis of Alzheimer's disease.

Study animals. In this study, Ca v 3.2 +/− embryos (kindly provided by Kevin Campbell via MMRCC-Mutant
Mouse Resource & Research Centers) were re-derived with C57BL/6J mice. All genotypes were obtained using random intra-strain mating. In total, eight Ca v 3.2 +/+ mice (all ♂, mean age: 124 ± 1 days) and eight Ca v 3.2 −/− animals (all ♂, mean age: 129 ± 4 days) were analyzed electroencephalographically. Experimental animals were housed in clear Macrolon cages type II in groups of 3-4 with ad libitum access to drinking water and standard food pellets. Mice were maintained under controlled environmental conditions using the ventilated cabinet Model 9AV125P (Tecniplast, Germany) and the UniProtect cabinet (Bioscape, Germany) with the following settings: ambient temperature 21 ± 2 °C, relative humidity 50-60%, and conventional 12 h/12 h light/dark cycle starting at 5:00 a.m.
All animal experiments were carried out in accordance with the guidelines of the German council on animal care and experimental protocols were approved by the local institutional and national committee on animal care ( Pre-surgical management of experimental animals and transmitter implantation. For presurgical preparation of experimental animals including selection of mouse lines, age and gender, anesthesia, temperature support, pain management, etc. please refer to our detailed descriptions 91,92 . Further details on the transmitter implantation are provided in Refs. 91-93 . Intrahippocampal electrode placement for electrohippocampal recordings. For intracerebral, deep EEG recordings from the hippocampal CA1 region, the differential electrode of the TA10ETA-F20 transmitter (Data Science International, DSI, USA), technical specifications: weight 3.9 g, volume 1.9 cc, input voltage range ± 2.5 mV, channel bandwidth (B) 1-200 Hz, nominal sampling rate (f) 1000 Hz (f = 5 B), temperature operating range 34-41 °C, warranted battery life 4 months, on-off mechanism magnetically actuated) was positioned www.nature.com/scientificreports/ at the following stereotaxic coordinates: (+)-lead, caudal − 2 mm, lateral of bregma 1.5 mm (right hemisphere), and dorsoventral (depth) 1.5 mm. The epidural reference electrode was positioned on the surface of the cerebellar cortex at the following stereotaxic coordinates: (−)-lead, bregma − 6 mm and lateral of bregma 1 mm (right hemisphere). For intracerebral recordings, the sensing lead of the transmitter was mechanically clipped to the deep electrode [91][92][93] . Notably, the deep tungsten electrodes (FHC, USA) are encapsuled with epoxylite with an impedance of 50-100 kΩ (measured at 1000 Hz) and a shank diameter of 250 μm. Epidural and intracerebral electrodes were fixed using glass ionomer cement (Kent Dental, Kent Express Ltd., UK) and the scalp was closed using over-and-over sutures (Ethilon, 6-0). Due to the body surface/body volume ratio, mice are highly susceptible to hypothermia. Thus, supplemental warmth was given to the animals during the entire period of anesthesia/ surgical procedure and the first two days post implantation using a heating pad. A detailed description of the stereotaxic EEG electrode placement and transmitter implantation was previously given by Weiergräber and colleagues 91,92,94 . For peri-and post-operative pain management, carprofen (5 mg/kg, Rimadyl, Parke-Davis/ Pfizer, Germany) was injected subcutaneously. Mice were given 10 days to fully recover after surgery. This recovery period was determined by the finding that no alterations in basic physiological/behavioral parameters such as water and food uptake, locomotion, surface and body core temperature, etc. could be detected between radiotransmitter-implanted, non-implanted, and sham-operated mice 10 days post surgery 48 .

Confirmation of EEG electrode placement.
To confirm that electrodes were positioned in the exact CA1 target area, brains were extirpated post mortem and fixed in 4% formaldehyde solution. Afterwards, brains were cut to 60 μm slices using a Vibroslice Tissue Cutter EMS 5000-MZ (Campden Instruments Limited, UK). Brain slices were stained with hematoxylin/eosin to visualize the branch canal ( Supplementary Fig. 6). Mice in which EEG electrodes were not placed correctly in the defined target region were removed from the subsequent analysis.
Radiotelemetric EEG data acquisition. In each experimental animal, the first 24 h baseline recording (R1) from the CA1 hippocampal region (electrohippocampogram) was obtained at day 10 post surgery. This recovery period is based on the observation that 10 days post surgery no differences in physiological parameters between transmitter implanted, non-implanted, and sham-operated animals could be detected 48,95 .
A second 24 h long-term baseline recording (R2) was conducted at day 17 post implantation to check whether potential alterations in relative EEG frequency range power are robust over time or whether there are developmental changes [96][97][98] (Fig. 1A).
In addition, two EEG recordings were performed following urethane injection (U1, U2) with 800 mg/kg i.p. (Sigma, Germany, freshly dissolved in 0.9% NaCl) at day 18 and 25 after implantation, respectively. CA1 intrahippocampal EEG data were acquired using the Dataquest ART 4.2 software (Data Sciences International, DSI, USA). Note that EEG data were sampled at a nominal rate of 1000 Hz with no a priori filter cutoffs.
Based on the Shannon-Nyquist theorem and limit, EEG frequency analysis was carried out up to 500 Hz (upper gamma range) 99 .
Besides biopotentials (such as EEG), the TA10ETA-F20 transmitter also provides temperature and activity data. As the transmitter was placed in a subcutaneous pouch on the back of the experimental animal in our setting, the recorded subcutaneous temperature values do not represent body core values. However, subcutaneous temperature data were shown to correlate with body core temperature under environmentally controlled conditions and can thus be compared within and between the individual genotypes [91][92][93]100,101 . Further note, that activity data are provided by the telemetry system in relative values (relative activity). These relative data represent activity in the horizontal plane and integrate trip distance, velocity and acceleration. Our EEG-activity correlation is based on a binary system with activity = 0 for the inactive state and activity > 0 for the active state. For details see also 91 .
For FFT based analysis, the duration of the individual EEG epochs was determined as 2 s 91-93 . Mean relative EEG power (%) of the individual frequency ranges was calculated for the individual circadian stages, i.e., two dark cycles (DC1, DC2, 12 h each) and two light cycles (LC1, LC2, 12 h each), and 6 h post urethane 1 and 2 injection phases (U1, U2). Potential EEG artefacts were identified by both manual inspection of the EEG and the automated artefact detection tool of Neuroscore and were eliminated for EEG relative power analysis 91,92,94 .
Relative activity counts and temperature data were also analyzed for baseline (R1, R2) and post urethane recordings (U1, U2) mentioned above. Importantly, activity data (active state, i.e., activity units > 0, or inactive state, i.e., activity units = 0) during the conventional 12 h/12 h light/dark cycle (starting at 5:00 a.m.) were correlated with the relative EEG power of the individual frequency bands from the hippocampal CA1 deflection.
Data were statistically analyzed and displayed as mean ± SEM. Statistics for FFT based frequency analysis were performed using multiple Student's t-test, corrected for multiple comparison using the Holm-Sidak approach www.nature.com/scientificreports/ (*p < 0.05; **p < 0.01; ***p < 0.001). Statistics and graphical representations were conducted using GraphPad Prism 6 for Windows (Graphpad Software, Inc., USA). In a second approach, qPCR analysis of selected gene candidates of the GABAergic system was performed including GABA A receptor delta subunit (Gabrd), GABA A receptor gamma 2 subunit (Gabrg2), GABA B 1 receptor subunit (Gabbr1) and GABA B 2 receptor subunit (Gabbr2). These subunits were selected for the following reasons: In mammals, sequences of six α, three β, three γ, one δ, three ρ, one ε, one π and one θ GABA A receptor subunits have been described [105][106][107][108] . A majority of GABA A receptor subtypes contains α, β and γ subunits with a stoichiometry of 2α.2β.1γ 105,109 . Notably, most GABA A receptors containing the γ2 subunit tend to form clusters at the postsynaptic membrane, whereas GABA A receptors incorporating the δ subunit seem to be exclusively localized extrasynaptically [110][111][112][113] . We have thus checked for the GABA A δ and γ subunits in our qPCR study. Concerning GABA B receptors, we analyzed both GABA B1 and GABA B2 subunits as they regulate both pre-and postsynaptic activity [114][115][116][117][118] .

Quantitative real time PCR (qPCR).
Finally, Ca v 3.1 (Cacna1g) and Ca v 3.3 (Cacna1i) were analyzed to check for potential alterations in other LVA T-type Ca 2+ channel transcript levels. Forward and reverse primer sequences of gene candidates are displayed in Table 1.
Total RNA was extracted from the hippocampus of male Ca v 3.2 +/+ animals (mean age: 19.32 ± 0.44 weeks, n = 8) and male Ca v 3.2 −/− mice (mean age: 20.43 ± 0.41 weeks, n = 8). Additionally, the hippocampus of a female Ca v 3.2 +/+ mouse aged 24.14 weeks was dissected, which served as a calibrator in our study. The calibrator contains RNA from the genes of interest and the housekeeping genes and is used on every PCR plate for every gene tested.
First, hippocampal tissue was dissected in RNAprotect Tissue Reagent (Qiagen, Germany) and snap-frozen in liquid nitrogen. Total hippocampal RNA was extracted using RNeasy Lipid Tissue Mini Kit (Qiagen, Germany) including DNA degradation (additional DNase digestion step). Quality and quantity of the extracted RNA was evaluated using Nanodrop (Nanodrop 1000, Thermo Fisher Scientific, Germany). To obtain a 50 µl cDNA volume, 1 µg of total RNA from each animal was reversely transcribed in a two-step RT-PCR approach using both anchored-oligo (dT) 18 and hexamer primers (Transcriptor First Strand cDNA Synthesis Kit, Roche, Switzerland). Gene candidates were tested in triplicates in each animal using 2 µl cDNA as a template. In addition, a triplicate of calibrator cDNA was carried out for normalization of potential inter-run variations. Duplicates of two negative controls, i.e., no template controls and no reverse transcriptase controls were performed to exclude false positive results. Note, that mice used for qPCR analysis did not undergo transmitter implantation and EEG recordings.
qPCR was conducted in a Light Cycler 480 System (Roche, Switzerland) using the following protocol: 95 °C (10 min, pre-incubation step); 95 °C (10 s, denaturation step); 60 °C (20 s, annealing step); 72 °C (30 s, extension step), 35 cycles. This protocol was applied to all tested primer pairs (Table 1). SYBR Green 1 Master (Roche, Switzerland) was used for signal detection and the specificity of amplification was evaluated by melting curve analysis.
The CP values received from the Light Cycler 480 Software (Roche, Switzerland) were exported to qBase + software (Biogazelle, Belgium) and analyzed based on a delta-Cq quantification model with qPCR efficiency correction, reference gene normalization considering the reference target stability of the selected housekeeping genes (HPRT, β-actin) and inter-run calibration 119 . The results were characterized as Calibrated Normalized Relative Quantity (CNRQ) and statistically analyzed using the Mann-Whitney test.

Data availability
Relative EEG power data of this study are available at Mendeley data (doi: 10 www.nature.com/scientificreports/