Von Rüden, 2018-Neuroscience_HSP70 genetic modulation.pdf

INTRODUCTION

Excessive and persistent inflammatory signaling involving Toll-like receptor (TLR)-mediated glia activation is considered as a key factor contributing to the development and manifestation of a permanent hyperexcitability during the course of epileptogenesis (Devinsky et al., 2013;Vezzani et al., 2013). Considering a facilitating impact of inflammatory mediators on excitatory neurotransmission, inflammatory signaling can also serve as one factor triggering ictogenesis following epilepsy onset (Vezzani, 2014).

High-mobility group box 1 (HMGB1) has been intensely studied as a TLR ligand with enhanced release and overexpression following an epileptogenic insult and in the epileptic brain (Dey et al., 2016). A series of studies has addressed the functional role of HMGB1 in the context of epileptogenesis and ictogenesis reporting relevant effects of targeting strategies in different rodent epilepsy models (Maroso et al., 2010;Chen et al., 2015;Yang et al., 2017). Whereas a role of HMGB1-triggered TLR signaling has thus been confirmed, it is still an open question whether further TLR ligands might contribute to the excessive inflammatory signaling and its functional consequences during epileptogenesis and in the epileptic brain. Using a comprehensive proteome analysis, we have recently demonstrated a prominent regulation of inducible heat shock protein 70 (hsp70) in the hippocampus and parahippocampal cortex during the early post-insult phase and the latency phase in a rat post-status epilepticus model of epileptogenesis (Walker et al., 2016). This result raised the question whether HSPA1A overexpression might affect hyperexcitability based on its interaction with TLRs.

Whereas it is well accepted that hsp70 serves as an endogenous ligand and modulator of TLR 2 and 4 (Vabulas et al., 2002;Calderwood et al., 2007), the exact quality of the receptor interaction is still not completely understood. Contrasting findings led subgroups of researchers either to suggest that hsp70 activates different cell types through innate immune receptors including TLRs causing enhanced release of pro-inflammatory cytokines or that hsp70 acts as a negative regulator of inflammatory responses (Asea et al., 2000(Asea et al., , 2002Aneja et al., 2006;Singleton and Wischmeyer, 2006;Ferat-Osorio et al., 2014). Thus, it remains difficult to predict whether increasing hsp70 expression exerts beneficial effects with anticonvulsant and preventive effects or detrimental effects with proconvulsant and disease-promoting effects. So far, limited data have been reported from acute seizure models and a kindling paradigm with chemical induction of single or repeated seizures (Ammon-Treiber et al., 2007;Ekimova et al., 2010). The results of these studies suggested protective effects of HSPA1A. However, the models used in these studies are biased by the use of chemoconvulsants, which directly interact with glutamatergic or GABAergic neurotransmission and by global hsp70 overexpression. These interactions might have affected the outcome thereby limiting general conclusions. In this context, it needs to be considered in particular that hsp70 might serve as a modulator of GABAergic neurotransmission rendering a direct interaction with the GABA antagonist pentylenetetrazol (PTZ) rather likely (Hsu et al., 2000;Huang et al., 2001;Lo¨scher, 2011).

In the present study, we have addressed the hypothesis that hsp70 can modulate excitability resulting in an impact on seizure thresholds and progression of seizures in an electrical kindling paradigm. Mice overexpressing human HSPA1A have been used for the experiments taking into account that hsp70 can auto-regulate its expression with a negative feedback (Morimoto, 1993;Plumier et al., 1995). Thus, overexpression of HSPA1A would rather suppress hsp70 induction. Plumier et al. (1995) have previously demonstrated that overexpression of human HSPA1A does not interfere with constitutive expression and cell stress-associated upregulation of murine Hspa1a. Therefore, we have chosen the transgenic (TG) mouse line C57BL/10.D2-Tg NSE(VSV/Hsp70)5769OGLE to study the consequences of HSPA1A overexpression in the amygdala kindling model of temporal lobe epilepsy. This mouse line constitutively overexpresses hsp70 in neuronal cells. Therefore, cell-specific consequences can be drawn.

EXPERIMENTAL PROCEDURES

Animals C57BL/10.D2-TgNSE(VSV/Hsp70)5769OGLE, in the following HSPA1A-TG mice (Carsillo et al., 2006(Carsillo et al., , 2009Kim et al., 2013a;Kim et al., 2013b), homozygous breeding pairs were obtained from the Department of Veterinary Biosciences at the Ohio State University.

HSPA1A-TG mice overexpress the human HSPA1A gene under the control of the neuron-specific enolase promoter (Carsillo et al., 2006(Carsillo et al., , 2009. The resultant protein is . For the experiments, we used adult male mice (34 HSPA1A-TG mice and 33 wild type (WT) mice) in an age range between 10 and 16 weeks. WT (C57BL/10) breeding pairs were purchased from Envigo (Horst, Netherlands). Mice were bred under controlled environmental conditions (temperature: 20-24°C; humidity: 45-65%; 12 h dark/light cycle). Experimental animals received food (Ssniff R/M Haltung, ssniff Spezialdia¨ten GmbH, Soest, Germany) and tap water ad libitum. During the experiments, we kept the animals individually in type II Makrolon cages with selected poplar wood bedding (LignocelÒ Select, Altromin Spezialfutter GmbH & Co. KG, Lage, Germany). For enrichment red polycarbonate houses and nesting material (Nestlets TM , Ancare, Bellmore, NY, USA) were provided. We made all efforts to minimize pain or discomfort and to reduce the number of the animals used in the study. The Government of Upper Bavaria approved the study (permit number: 55.2-1-54-2532-58-15) and the experimental procedures were carried out in accordance with the European Communities Council Directive of 22 September 2010 (2010/63/EU), the German Animal Welfare Act and comply with the ARRIVE guidelines.

Genotyping HSPA1A-TG mice were genotyped by polymerase chain reaction. The genotype of each animal was assessed twice using genomic DNA isolated from an ear punch (before experiments) or from a tail biopsy (after experiments). For extracting the DNA and detecting the inserted intragenomic sequence a commercial Kit (Phire Tissue Direct PCR Master Mix, ThermoFisher Scientific) was utilized using sense 5 0 TTGGAGGCACTTCTGACTT GCA3 0 and antisense 5 0 TTCGCGTCAAACACGGTGTT3 0 primers (Carsillo et al., 2009). We performed all procedures according to the manufacturer's specifications. Gene expression was quantified using a TProfessional Basic Thermocycler (Biometra GmbH, Go¨ttingen, Germany).

Electrode implantation

For electrical kindling and electroencephalography (EEG) recording, mice underwent stereological electrode implantation. For analgesia we administered 1 mg/kg meloxicam subcutaneously (MetacamÒ 5 mg/ml; Boehringer Ingelheim Vetmedica GmbH, Ingelheim/ Rhein, Germany) 30 min prior and 24 h after the electrode implantation. Animals were anesthetized by intraperitoneal injection of chloral hydrate (600 mg/kg, 40 mg/ml solved in saline, Merck KGaA, Darmstadt, Germany). We stabilized the mice in a stereotaxic frame (TSE Systems GmbH, Bad Homburg, Germany) and bupivacaine (Bupivacaine 0.5% with epinephrine 0.0005% JenapharmÒ; Mibe GmbH, Brehna, Germany) was administered subcutaneously for local anesthesia. We drilled four holes into the exposed skull bone, facilitating the implantation and fixation of the Teflon-isolated stainless steel electrode (diameter 280 mm) into the right amygdala. The stereotaxic coordinates in millimeters relative to bregma were determined in previous test implantations: rostro-caudal À1.4; lateral +3.4; dorso-ventral À5.1 for HSPA1-TG mice and rostro-caudal À1.4; lateral +3.1; dorso-ventral À5.0 for WT mice.

Kindling procedure

Animals were randomly distributed to experimental and control groups (www.randomizer.org). The person who performed the experiments was unaware of the genotype of the animals. We treated control animals exactly in the same way as the experimental mice, except for electrical stimulations. The bodyweight of all animals has been scaled before and after the kindling process. For acclimatization to the experimental conditions, we transferred the animals to the laboratory 30 min before the start of the experiments. Electrical stimulations were performed in a video monitored glass aquarium (40 Â 35 Â 40 cm). The initial and the post-kindling threshold were determined between 6.30 a.m.-2.30 p.m. and daily electrical stimulations were performed between 8.30 a.m.-10.30 a.m., except for the last electrical stimulation being scheduled 24 h before the euthanasia and perfusion.

After a recovery period of two weeks, we assessed the animals initial afterdischarge threshold by electrical stimulation using an HSE Type 215E12 SV1 Stimulator (Hugo Sachs Elektronik, Harvard Apparatus, March-Hugstetten, Germany). An ascending stair-step procedure with an initial current of 8 mA and an increase by 20% of the previous current in intervals of 1 min was used as previously described (Pekcec et al., 2007). The lowest current which triggered characteristic electroencephalographic seizure activity (spikes: amplitude at least twice as high as basal electroencephalogram, lasting at least five seconds) was defined as afterdischarge threshold. Animals were kindled with current above the initial afterdischarge threshold (700 mA, 1 ms, monophasic square-wave pulses, 50 Hz for 1 s) until the WT mice exhibited three consecutive generalized seizures (in total nine stimulations). We decided to use supra-threshold stimulation with 700 mA based on the spread of the initial afterdischarge threshold. Our group has already shown in former studies that there is no difference between suprathreshold stimulation and individual stimulation strength (von Ru¨den et al., 2015). For the determination of the seizure severity a slightly modified Racine scale (Racine, 1972) was used: score 1: immobility, eye closure, ear twitching, twitching of vibrissae, sniffing, facial clonus; score 2: head nodding associated with more severe facial clonus; score 3: clonus of one forelimb; score 4: bilateral forelimb clonus with or without rearing; score 5: bilateral fore-and hind limb clonus and falling with or without rearing (von Ru¨den et al., 2015). The experimenter measured the behavioral seizure duration with a digital timer and the EEG was recorded with LabChart 7 (ADInstruments Ltd, Hastings, United Kingdom). The sum of all electrographic afterdischarge durations during the kindling process was defined as cumulative afterdischarge duration. Following the kindling procedure, we determined the post-kindling afterdischarge threshold using the stepwise procedure as already described before. Cumulative seizure parameters and the number of stimulations to reach the first generalized seizure were calculated with values of the daily kindling stimulations.

In total, seven animals (four WT mice, three HSPA1A-TG mice) were excluded from the experiment: two animals died after surgery, four animals had to be euthanized with respect to humane end points or loss of the stimulation electrode and the last one was excluded because of spontaneous seizure activity one day after surgery.

Tissue preparation and immunohistochemistry

Twenty-four hours after the last kindling stimulation and six hours after post-kindling threshold determination, animals were deeply anesthetized by an intraperitoneal injection of 600 mg/kg pentobarbital (NarcorenÒ, Merial GmbH, Halbergmoos, Germany) and were transcardially perfused with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). We removed the brains from the skull and post-fixed them in 4% paraformaldehyde (pH 7.4) for 24 h before being cryoprotected in a solution of 30% sucrose in 0.01 M PBS. Coronal sections (40 lm) were cut using a freezing microtome (Model 1205 Frigomobil, Reichert-Jung, Heidelberg-Nußloch, Germany). Brain sections were transferred to a cryoprotectant solution (1:1 glycerol and 0.1 M phosphate buffer, pH 7.4) and stored at -80°C. We processed serial sections, taken at 200 lm intervals, for immunohistochemistry. For the verification of the electrode localization, one entire series was stained with Thionine acetate (9937 -Carl Roth, Germany).

Immunohistochemical staining of the ionized calcium binding adaptor molecule 1 (Iba1), a microglia/ macrophage-specific calcium-binding protein, was performed with free-floating tissue sections. Following three washing steps, we conducted heat-induced antigen retrieval at 80°C for 30 min using sodium citrate buffer (10 mM, pH 6.0). Subsequently, the tissue sections cooled down on ice for 10 min. After three additional washing steps, sections were incubated in 0.45% TBS-buffered H 2 O 2 for one hour and were rinsed three times. Blocking of unspecific binding sites was performed by incubating the sections for one hour at room temperature in a blocking solution containing 10% donkey serum (Jackson, West Grove, USA) and 2% bovine serum albumin (Sigma-Aldrich Cat# A9647, Germany). Sections were incubated with a polyclonal rabbit anti-Iba1 primary antibody (Wako Cat# 019-19471, RRID:AB_2665520, Lot# SAJ2266 1:10000, Japan) overnight at 4°C. Following washing in TBS, sections were placed in biotin-labeled donkey-anti-rabbit secondary antibody (Jackson ImmunoResearch Labs Cat# 711-065-152, RRID:AB_2340593, Lot# 122849, USA) for one hour at room temperature. Subsequent to three washing steps, sections were incubated in horseradish peroxidase-labeled streptavidin (Jackson ImmunoResearch Labs Cat# 016-030-084, RRID:AB_ 2337238, USA) for one hour at room temperature. We conducted the nickel-enhanced diaminobenzidine reaction (0.05% 3,3 0 -diaminobenzidine (CN75, Carl Roth GmbH & Co. KG, Karlsruhe, Germany), 0.01% nickel ammonium sulfate (A1827, Sigma Aldrich Chemie GmbH, Munich, Germany) and 0.01% H 2 O 2 one minute after rinsing with TBS. Following final washing steps, sections were mounted and air-dried overnight and subsequently, we used EntellanÒ (107960, Merck, Darmstadt, Germany) to coverslip microscope slides. We processed negative controls in parallel omitting the incubation with the primary antibody.

In addition, we analyzed the expression profile of a second microglia marker CD11b. CD11b (= integrin alpha M) associates with CD18 to form the type 3 complement receptor (CR3). Although it is constitutively expressed on resting and activated microglia, its up-regulation in the brain in response to an insult is considered to be indicative of activation of microglia. Immunohistochemical staining was performed on free floating tissue sections. Endogenous peroxidase was quenched by incubating the sections with 3% H 2 O 2 for 15 min and heat induced epitope retrieval was achieved with sodium citrate buffer (10 mM, pH 6.0) in a water bath at 80°C for 30 min. In between, sections were washed with TBST. We blocked unspecific binding sites with 5% goat serum in antibody diluent (2% goat serum, 0.1% Triton X in TBS). Afterward, we incubated the sections with a monoclonal anti-CD11b primary antibody (1:500, Abcam Cat# ab133357, RRID:AB_2650514, Abcam, Cambridge, UK) overnight at 4°C. Sections were incubated for one hour at room temperature with a biotinylated goat anti-rabbit antibody (1:1500, Vector Laboratories Cat# BA-1000, RRID:AB_2313606, USA) for one hour at room temperature and with peroxidaselabeled streptavidin (1:1400, Vector Laboratories Cat# PK-4000, RRID:AB_2336818, USA) for one hour at room temperature. In between sections were washed with TBST. Subsequently, the diaminobenzidine reaction was performed for one minute and sections were washed in TBS and in bidistilled water, mounted, airdried and coverslipped with EntellanÒ (Merck, Darmstadt, Germany). Negative controls omitting the primary antibody were prepared in parallel.

Evaluation of microglia activation

Microglia activation of Iba1-positive microglia cells was assessed with the semi-automatic image analysis method to analyze microglial morphology as described by Hovens et al. (2014) in both brain hemispheres in the cornu ammonis regions 1 and 3 (CA1 and CA3) and in the hilus of the hippocampal formation. An operator unaware of the animals' genotype and treatment condition evaluated five brain sections (À1.22, À1.7, À2.3, À2.8, À3.4 caudal to bregma). Therefore, images were captured at 400x magnification with an Olympus BH2 microscope with, a single chip charge-coupled device (CCD) color camera (Axiocam; Zeiss, Go¨ttingen, Germany), and an AMD Athlon TM 64 Processor based computer with an image capture interface card (Axiocam MR Interface Rev.A; Zeiss, Go¨ttingen, Germany). We evaluated two visual fields (1388 Â 1040 pixels) in the CA1 and CA3 and one visual field (1388 Â 1040 pixels) in the hilus by ImageJ software (ImageJ 1.51 version, https://imagej. nih.gov/ij).

Iba-1 staining has a specific signal distribution, with the cell body presenting higher intensity compared to the rest of the cell. Using the threshold mask, the software filters out most of the dendritic processes and the use of a size filter ensures to exclude all highstaining pixel clusters that are smaller than 150 pixels and belong to dendritic processes.

According to Hovens et al. (2014), the total cell area, referring to the total Iba1 expression, was measured using the threshold function with the default mask, a variation of the IsoData algorithm (Ridler and Calvard, 1978), which was automatically provided by the program. To measure the total cell body area the threshold was lowered by 40 points and a size filter (150-infinity pixels) was applied to exclude all pixel clusters that are smaller than a small cell body (radius >35 lm) as described by Kozlowski and Weimer (2012). Both, the total cell area and the cell body area of all microglia in the area of interest were measured with ImageJ analyze particle function.

The obtained values were used to estimate the microglial activation as follows:

Microglia activationð%Þ ¼ total cell area=cell body area

We assessed microglia activation of CD11b-positive microglia in a randomized sequence (www.randomizer. org) using a semi-automatic image analysis method in both brain hemispheres in the same brain regions as above. The investigator was unaware of the group allocation. Images were captured at 200Â magnification with the same microscope and camera as mentioned above. We analyzed two visual fields in the CA1 region, one visual field in the two rostral bregma regions and two visual fields in the three caudal bregma regions of the CA3 region, and one visual field in the hilus with ImageJ software (ImageJ 1.51 version, https://imagej. nih.gov/ij). One visual field comprised 1388 by 1040 pixels equal 586 by 439 mm (2.37 pixel/mm conversion scale).

For assessment of the cell dimension range of CD11b-positive cells, a total of 60 cells were processed with the contour tool and measured for their total area. The cell size ranged from 20.11 to 90.797 mm 2 . Based on these measurements, we chose a size filter of 20 mm 2 referring to the smallest captured cell size. Images were converted into 8-bit format. The threshold mask (triangle method (Zack et al., 1977) added by two points for each image) excludes most of the dendritic processes and the additional size filter excludes all high-staining pixel clusters that are smaller than 20 mm 2 belonging to dendritic processes or background staining.

The area of interest was measured and the number of CD11b-positive cells was assessed using the analyze particle tool. We summed up the counted particles and the area of interest per analyzed Bregma area and region and expressed the ratio as cell density:

Statistics

We performed the statistical analysis with GraphPad Prism 5.04 for Windows (GraphPad Prism Software, San Diego, USA). Group differences at threshold stimulation were determined by using the unpaired Student's t-test or Mann-Whitney test. We analyzed seizure severity, seizure duration and afterdischarge duration during the course of kindling as well as bodyweight development by two-way ANOVA for repeated measurements and compared individual stimulation days as well as selected group comparisons of the bodyweight by unpaired Student's t-test. The response to stimulation at the first day of the kindling phase with daily stimulations was analyzed with a Fisher's exact test (seizure score !1: yes/no). We used the paired Student's t-tests to compare pre-and postkindling parameters for individual groups. The evaluation of activated microglia (Iba1 and CD11b) was performed using a two-way ANOVA and the Bonferroni post hoc correction for multiple testing. We considered a p value <0.05 statistically significant and expressed all descriptive statistics as mean ± SEM.

RESULTS

Enhanced initial seizure susceptibility in HSPA1A-TG mice

We evaluated the pre-kindling seizure susceptibility in WT and HSPA1A-TG mice in order to obtain information about basal seizure thresholds and spread of seizure activity.

The analysis revealed an increased susceptibility for electrically triggered ictogenesis with seizure thresholds in HSPA1A-TG mice reaching 43% lower levels as compared to WT animals (WT 200.29 ± 29.64 mA;

HSPA1A-TG 113.17 ± 25.10 mA, see Fig. 1A). Seizure parameters at threshold stimulation including seizure severity as well as duration of behavioral and electrographic seizure activity did not differ significantly between groups (see Fig. 1A-E). In addition, we observed seizures with a severity score of 1 in ten WT and seven HSPA1A-TG mice and a severity score of !1 in seven WT and eleven HSPA1A-TG mice, indicating a comparable seizure spread (Fisher's exact test p = 0.3175). Mean seizure severity amounted to 1.59 ± 0.21 in WT mice and to 2.28 ± 0.31 in HSPA1A-TG mice (see Fig. 1D). The duration of afterdischarges reached 10.76 ± 0.75 s in WT mice and 16.94 ± 3.07 s in HSPA1A-TG mice (Fig. 1C) and the duration of motor seizures amounted to 13.82 ± 4.53 s in WT mice to 18.0 ± 11.27 s in HSPA1A-TG mice (Fig. 1E).

Accelerated kindling progression in HSPA1A-TG mice

In contrast to WT mice, HSPA1A-TG mice exhibited generalized seizure activity early during the kindling paradigm. We observed seizures with a severity score of four and higher in a significantly larger number of animals at day one of the phase with daily kindling stimulations (Fisher's exact test p < 0.0001). The mean seizure severity reached significantly higher levels at days one to six of the stimulation phase (F (1.264) = 46.90; p < 0.0001; see Fig. 2A). Along with an increased seizure severity score, behavioral and electrographic seizure duration proved to be prolonged as a consequence of HSPA1A overexpression (seizure severity: F (1.264) = 46.90; p < 0.0001; seizure duration F (1.264) = 9.48; p < 0.004; afterdischarge duration F (1.264) = 6.12; p < 0.02). Differences to WT animals were evident at stimulation days one to three as well as day five for motor seizure activity (see Fig. 2B) and at days one to three for electrographic afterdischarges (see Fig. 2C).

Toward the end of the 9-day stimulation phase, seizure parameters reached comparable levels in both groups of animals (see Fig. 2A-C).

We further confirmed the difference in the kindling acquisition rate by demonstrating that the number of stimulations necessary for induction of the first generalized seizure was significantly lower in HSPA1A-TG mice (WT n = 3.24 ± 0.33, HSPA1A-TG n = 1.11 ± 0.08, see Fig. 2D). Moreover, we calculated the cumulative duration of all electrographic seizure events during the kindling paradigm until the first generalized seizure was observed in individual animals. In HSPA1A-TG mice the respective cumulative afterdischarge duration until the first generalized seizure was reduced by 93.46% in comparison with WT mice (WT 34.00 ± 5.47 s, HSPA1A-TG 2.22 ± 1.53 s, see Fig. 2E). In addition, we determined an increased total cumulative afterdischarge duration in HSPA1A-TG mice as compared with WT mice (p = 0.0186, Fig. 2F).

Comparison of post-kindling thresholds between

HSPA1A-TG and WT mice

Following the 9-day stimulation phase, post-kindling thresholds were determined along with an assessment of seizure parameters at threshold stimulation and the relative change of individual afterdischarge threshold. Threshold analysis in WT mice revealed the typical kindling-associated decline in seizure thresholds (p = 0.0021) indicating that seizure susceptibility has significantly increased during the kindling paradigm (see Fig. 1A). In HSPA1A-TG mice, in which low seizure thresholds have been evident before kindling, no difference was observed when comparing pre-and postkindling seizure thresholds (p = 0.4585, see Fig. 1A). Thus, kindling in HSPA1A-TG mice did not result in a further increase in seizure susceptibility. When the relative change of individual afterdischarge thresholds was analyzed a significant group difference became evident (p = 0.039). Whereas WT mice show a decrease by 23%, HSPA1A-TG mice exhibit an increase of individual afterdischarge thresholds by 78% (see Fig. 1B).

Regardless of the genotype, the duration of afterdischarges at post-kindling threshold stimulations exceeded that at pre-kindling threshold stimulation with a 93% increase in WT mice and an 89% increase in TG mice (WT p < 0.0001 and HSPA1A-TG mice p = 0.0003, see Fig. 1C). Thereby, significantly enhanced seizure severity and prolonged duration were evident in HSPA1A-TG mice when directly comparing with WT data (seizure severity p = 0.0188; seizure duration p < 0.0001; Fig. 1F, G).

Impact of kindling on microglia activation in WT and

HSPA1A-TG mice

We analyzed microglia activation in the CA1, CA3 and hilus sub-region of the hippocampal formation by Iba1 and CD11b immunohistochemistry.

For all of these sub-regions data proved to be comparable for both hemispheres and homogeneous along the rostro-caudal part of the hippocampus.

Activation of microglia assessed based on analysis of

Bodyweight development is affected by HSPA1A overexpression in kindled and non-kindled mice

Considering the fact that short heat stress can affect appetite probably mediated via induction of heat shock proteins (Pearce et al., 2014), we carefully controlled the development of bodyweight in all groups. No significant group differences in bodyweight were evident when animals were entering the study (see Table 1). At the end of the experiment, the bodyweight of non-kindled animals WT and HSPA1A-TG mice increased by 5% and 2%, respectively (see Table 1). Kindled animals also exhibited an increase in bodyweight during the course of the kindling paradigm reaching 4% in WT and 2% in TG mice (see Table 1). The bodyweight of kindled WT and HSPA1A-TG differed at the end of the experiment with a significant impact of the genotype (F (1.33) = 5.35, p = 0.03 and Table 1), with WT mice weighing more.

DISCUSSION

As a consequence of the overexpression of human HSPA1A, mice exhibited an increased seizure Representative histological images of the hilus of a non-kindled electrode implanted WT control mouse, of a kindled WT mouse, of a non-kindled electrode implanted HSPA1A-TG control mouse, and of a kindled HSPA1A-TG mouse. Note the different morphology of resting microglia cells with a small cell body and elongated cell processes and activated microglia cells with a swollen cell body and retracted cell processes. All data are given as mean ± SEM, Two-way ANOVA. Control: non-kindled electrode implanted controls, Kindling: kindled electrode implanted mice, WT: WT mice, TG: HSPA1A-TG mice, group size n = 6, Scale bar = 25 lm (upper four pannels) and 10 mm (lower two pannels).

susceptibility already evident during the first stimulation. Moreover, we observed an early development of generalized seizures and an accelerated kindling process in HSPA1A-TG mice.

In support of our hypothesis, determination of the initial afterdischarge threshold, revealed an impact of HSPA1A-TG. The findings indicate a facilitated ictogenesis in animals, which have not been previously exposed to any epileptogenic trigger. The fact that the severity and duration of induced seizures during this first pre-kindling threshold determination did not differ between WT and TG mice suggests that HSPA1A overexpression promotes seizure generation, whereas seizure spread and endogenous seizure termination remain unaffected in non-kindled animals.

These findings are of interest considering that neuronal damage following brain insults can result in enhanced release of damage-or danger-associated molecular patterns (DAMPS) including hsp70 (Matin et al., 2015;Thundyil and Lim, 2015). According to our findings, increased concentrations of hsp70 may then contribute to early seizures, which represent a frequent event following brain insults such as traumatic brain injury (Alexander et al., 2017;Aquino et al., 2017;Wang et al., 2017). Respective effects of hsp70 may be related to its role as a modulator of TLR 2 and 4 (Akira and Hemmi, 2003; Matin et al., 2015). It has been demonstrated that an interaction of hsp70 with these receptors can contribute to an activated state of microglial cells. However, in the present study, we could not confirm an impact of hsp70 on microglia activation analyzed based on Iba1 and CD11b staining and morphology of immunopositive cells. Thus, more subtle molecular alterations in the microglial functional state might have contributed to the impact of HSPA1A overexpression on kindling. Hsp70mediated TLR signaling can trigger the synthesis of inducible nitric oxide synthase (iNOS) and of matrix metalloproteinases. These events may also affect seizure thresholds. Excess amounts of iNOS have already been suggested as a contributor to ictogenesis (Murashima Yoshiya et al., 2002). Extracellular matrix and synaptic remodeling mediated by increased matrix metalloproteinase activity may promote enhanced neuronal excitability and ictogenesis (Pitka¨nen et al., 2014;Bronisz and Kurkowska-Jastrze z bska, 2016). iNOS, matrix metalloproteinases 2 and 9 are the most important inflammatory associated proteins in this family of proteins. However, in previous studies using a comprehensive proteome analysis (Walker et al., 2016;Keck et al., 2018), we have analyzed proteins associated with microglia activation, cell stress, extracellular matrix and angiogenesis at three different time points in a rat model of epileptogenesis. In this data set, none of the abovementioned proteins proved to be regulated. Moreover, in another proteome analysis in a mouse model of temporal lobe epilepsy (Bitsika et al., 2016) regulation of proteins associated with microglia activation were confirmed, but again iNOS, matrix metalloproteinases 2 and 9 were not detected. Thus, we did not expect relevant alterations in the kindling paradigm, which is characterized by a milder neuropathology. Therefore, we decided to not further analyze the expression of these proteins in the present study. The impact of excessive TLR signaling has previously been demonstrated by studies targeting receptor function (Maroso et al., 2010;Iori et al., 2017). For instance, injection of TLR 4 antagonists or of antagonists of its ligand HMGB1 results in anticonvulsant effects in a mouse model with spontaneous seizures (Maroso et al., 2010). On the other hand, administration of HMGB1 or TLR 4-activating lipopolysaccharide exerted pro-convulsant effects (Maroso et al., 2010;Vezzani, 2014). Taking these findings into account, enhanced TLR 4-mediated hilar microglia activation with its molecular and functional consequences might have contributed to higher seizure susceptibility in HSPA1A-TG mice.

Testing of TG mice in the kindling paradigm also provides valuable information about the development of a persistent hyperexcitable state. During this process, the progression of the development of seizure severity and duration is of particular interest. HSPA1A-TG mice exhibited significantly increased severity scores and an increased seizure duration from their first stimulation (i.e., subsequent to initial threshold determination) and thereafter. Thus, the history of a single focal seizure tremendously triggered the response to electrical stimulation in these mice, so that generalized seizures were already observed with the first supra-threshold stimulation during kindling acquisition. The data demonstrate that seizure spread, generalization, and duration can also be enhanced as a consequence of HSPA1A overexpression resulting in a very rapid kindling phenomenon. Evidence for an enhanced kindling rate received further confirmation based on the analysis of the cumulative afterdischarge until the first generalized seizure, which proved to be reduced in HSPA1A-TG mice. This finding argues against a mere pro-convulsant role of HSPA1A overexpression and supports an additional impact on kindling progression. Thus, our data provide first evidence that hsp70 may not only promote ictogenesis, but may also exert proepileptogenic effects. Respective conclusions require further confirmation by testing in chronic models with development of spontaneous seizures following an initial brain insult.

Again, an enhanced TLR 4-mediated inflammatory response might play a functional role for the consequences of the genetic modulation. Various experiments demonstrated that anti-inflammatory treatment can delay or block progressive development of kindled seizures or the development of late spontaneous seizures in post-status epilepticus models (Ravizza et al., 2011;Vezzani et al., 2013;Terrone et al., 2016). Thus, it seems likely that excessive hsp70 expression in TG mice triggered TLR 4-mediated microglia activation and its molecular consequences, which in turn accelerated the generation of a hyperexcitable kindled network. However, we were not able to detect a difference in the level of activated microglia in the hippocampus of WT and HSPA1A-TG mice. In this context, it needs to be considered that we have focused the analysis on the morphology of Iba-1 expressing microglia cells. As mentioned above we cannot exclude more subtle molecular differences in the cellular functional state.

Despite the slower kindling rate, WT mice exhibited comparable responses to stimulation during the final phase of the kindling process. However, a difference to TG mice was still evident when seizure parameters were analyzed during the post-kindling threshold determination. At this time point, both, seizure severity and duration in TG mice exceeded that in WT mice. In contrast, thresholds proved to be in a comparable range in both groups. The lack of a kindling-associated decrease in thresholds in TG mice might be related to the fact that these animals already exhibited very low thresholds prior to kindling. However, an analysis of kindling-associated changes even pointed to an increase in individual thresholds, which might have further contributed to post-kindling thresholds that were comparable to those of WT mice.

Differences in post-kindling seizure parameters demonstrate that HSPA1A overexpression still exerts relevant effects when a hyperexcitable neuronal network has been generated. In particular, the analysis of seizure parameters at threshold stimulation indicated that enhanced hsp70 release or synthesis may also promote severity and duration of seizures following epilepsy manifestation. Again, this assumption requires further experimental support from models with spontaneous seizure development.

The findings of the present study seem to be in apparent contrast with one previous study that explored the consequence of HSPA1A overexpression in mice. In this study Ammon-Treiber et al. (2007) described a protective effect of the genetic modulation in a chemical kindling paradigm with repeated PTZ administration. Considering an increased sensitivity to PTZ, TG mice received lower PTZ doses than WT mice for kindling induction. In response to these doses, TG mice exhibited a delayed kindling progression, reduced mortality as well as lower seizure severity in a re-exposure to PTZ seven days following completion of the kindling phase (Ammon-Treiber et al., 2007). The data from the PTZ dose-finding experiment in naı¨ve mice support a proconvulsant effect of HSPA1A overexpression, and are thus, consistent with our findings of a lowered seizure threshold prior to kindling. The discrepancies in the impact of hsp70 on the course of kindling might be related to different factors. First of all, the adjusted PTZ dosing in WT and TG mice might have caused a bias toward a slower kindling in TG mice in the study by Ammon-Treiber et al. (2007). As already mentioned in the introduction, the fact that hsp70 has been suggested to act as a modulator of GABAergic neurotransmission might have influenced the response to the GABA A receptor antagonist PTZ (Hsu et al., 2000;Huang et al., 2001;Lo¨scher, 2011). Finally, the consequences of a widespread HSPA1 overexpression under control of the beta-actin promoter, which mediates expression in the brain and the periphery, have been analyzed in the earlier study (Ammon-Treiber et al., 2007). Under these conditions, hsp70 overexpression is affected in all different cell types of the brain (e.g. neurons, astrocytes, microglia, oligodendroglia, and endothelial cells). Therefore, conclusions about the mechanistic cell dependent action of hsp70 cannot be made. Moreover, peripheral consequences of the genetic modulation might have exerted an additional effect for instance based on a cross-talk between peripheral and central inflammatory signaling. In contrast, we have assessed the specific consequences of neuronal overexpression. Thus, hsp70 expression levels of responding cells is unaffected and does not interfere with neuronal hsp70. Furthermore, the decision for the HSPA1A-TG mouse line with neuronal hsp70 overexpression was taken, as enhanced release and synthesis of the DAMP in the clinical situations of interest stems from neuronal damage.

In this context, we would like to point out that it would have been of interest to analyze hsp70 expression levels in this kindling study. For instance, respective data would have allowed testing for a putative correlation with in vivo seizure data. However, we faced technical problems and despite intense efforts did not succeed in establishing a method for expression analysis of mouse hsp70.

CONCLUSION

The findings demonstrate that hsp70 can exert proconvulsant effects promoting ictogenesis in animals not previously exposed to epileptogenic triggers. The pronounced impact on the response to subsequent stimulations with an accelerated kindling phenomenon gives first evidence that excessive hsp70 concentration may also contribute to epileptogenesis. Thus, strategies antagonizing hsp70 or inhibiting its expression might be of interest for prevention of seizures and epilepsy. However, conclusions about a putative pro-epileptogenic effect of hsp70 require further investigations in models with development of spontaneous recurrent seizures.