FULL-LENGTH ORIGINAL RESEARCH

Purinergic control of hippocampal circuit hyperexcitability in Dravet syndrome  Feng Gu, Anupam Hazra, Ahmad Aulakh, and Jok ubas Ziburkus Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

SUMMARY

Dr. Feng Gu is a  mentor to Dr. Ziburkus (University of Houston) and a postdoctoral fellow at Stanford University.

Objective: Severe myoclonic epilepsy in infancy (SMEI) or Dravet syndrome is one of the most devastating childhood epilepsies. Children with SMEI have febrile and afebrile seizures (FS and aFS), ataxia, and social and cognitive dysfunctions. SMEI is pharmacologically intractable and can be fatal in 10–20% of patients. It remains to be elucidated how channelopathies that cause SMEI impact synaptic activities in key neural circuits, and there is an ongoing critical need for alternative methods of controlling seizures in SMEI. Using the SCN1A gene knock-in mouse model of SMEI (mSMEI), we studied hippocampal cell and circuit excitability, particularly during hyperthermia, and tested whether an adenosine A1 receptor (A1R) agonist can reliably control hippocampal circuit hyperexcitability. Methods: Using a combination of electrophysiology (extracellular and whole-cell voltage clamp) and fast voltage-sensitive dye imaging (VSDI), we quantified synaptic excitation and inhibition, spatiotemporal characteristics of neural circuit activity, and hyperthermia-induced febrile seizure-like events (FSLEs) in juvenile mouse hippocampal slices. We used hyperthermia to elicit FSLEs in hippocampal slices, while making use of adenosine A1R agonist N6-cyclopentyladenosine (CPA) to control abnormally widespread neural activity and FSLEs. Results: We discovered a significant excitation/inhibition (E/I) imbalance in mSMEI hippocampi, in which inhibition was decreased and excitation increased. This imbalance was associated with an increased spatial extent of evoked neural circuit activation and a lowered FSLE threshold. We found that a low concentration (50 nM) of CPA blocked FSLEs and reduced the spatial extent of abnormal neural activity spread while preserving basal levels of excitatory synaptic transmission. Significance: Our study reveals significant hippocampal synapse and circuit dysfunctions in mSMEI and demonstrates that the A1R agonist CPA can reliably control hippocampal hyperexcitability and FSLEs in vitro. These findings may warrant further investigations of purinergic agonists as part of the development of new therapeutic approaches for Dravet syndrome. KEY WORDS: SCN1A gene, Seizure, Hyperthermia, Dravet syndrome, Imaging, Adenosine A1 receptor.

Severe myoclonic epilepsy in infancy (SMEI) or Dravet syndrome is a deleterious form of childhood epilepsy with onset in the first year of life, usually beginning with febrile Accepted October 21, 2013; Early View publication January 13, 2014. Department of Biology and Biochemistry, University of Houston, Houston, Texas, U.S.A. 1  Address correspondence to Jokubas Ziburkus, Department of Biology and Biochemistry, University of Houston, Houston, TX 77204, U.S.A. E-mail: [email protected] Wiley Periodicals, Inc. © 2014 International League Against Epilepsy

seizures (FS).1 FS often culminate in status epilepticus, and patients with SMEI may endure a number of neurological complications.1–7 Genetic studies show that 70–80% of the SMEI phenotype can be accounted for by mutations in the SCN1A gene. Using transgenic mice carrying mutations of SCN1A, a gene that codes for voltage-gated sodium channel protein (Nav1.1), several recent studies have shown that the animals develop epilepsy phenotypes that recapitulate many features of the human SMEI disorder, including a decreased threshold for febrile seizures and early death in homozygous

245

246 F. Gu et al. mutants.8–12 It has also been demonstrated that this and similar SCN1A mutations specifically affect neocortical12 and hippocampal13 inhibitory interneurons, causing them to fail to reliably generate action potentials. Despite these recent advances in the understanding of the pathophysiology of SMEI, effective treatments for it still remain a great challenge,10 and SMEI patients are clinically refractory with a 10–20% mortality rate.14,15 To improve our understanding of epileptogenesis in Dravet syndrome, further studies must determine how SCN1A mutations and hyperthermia impact synaptic and circuit physiology, ultimately leading to improved treatments. Recent advances in fast functional imaging, including voltage-sensitive dye imaging (VSDI), provide a way to simultaneously measure changes in excitability of neuronal populations across wide spatial areas, enabling identification of hyperexcitable circuits. VSDI signals are linearly correlated with relative changes in postsynaptic neuronal membrane potentials17–18 and can be used reliably to visualize evoked19,20 or spontaneous epileptiform activity.21 In chronic epilepsy models,19 VSDI reveals circuit hyperexcitability, synonymous with significantly wider area of evoked neural activation. We applied this approach here to study pathophysiology of neural circuits in SMEI. To characterize the synaptic and neural circuit activity in mouse model of SMEI (mSMEI), we focused on hippocampal circuits. The hippocampus is highly susceptible to seizures, may be involved in the generation of SMEI seizures,22 and is a key structure for learning and memory. Using electrophysiology and fast VSDI in hippocampal slices, we found that the knock-in SCN1A gene mutation leads to a significant excitation/inhibition (E/I) imbalance. This shift results in abnormally large hippocampal circuit activation and lowered threshold for hyperthermia-induced febrile seizure-like events (FSLEs). To curtail neural circuit hyperexcitability in mSMEI tissue, we used adenosine A1 receptor (A1R) agonist N6-cyclopentyladenosine (CPA). A1R agonists have been shown to be anticonvulsant in several experimental epilepsy models23–26; however, the efficacy of A1R agonists on hyperexcitable SMEI circuits has not been tested. Herein, we report that the A1R agonist CPA efficiently curtails aberrant SMEI circuit activity and hyperthermia-induced hyperexcitability without blocking excitatory synaptic transmission.

Methods Animals The model of SMEI used in these studies is caused by a knock-in nonsense substitution (CgG to TgA in exon 21) made within a loop between segments 5 and 6 in mice with C57BL/6/129 background (87.5%/12.5%, respectively). All experiments were performed in accordance with animal protocols approved by the International Animal Care and Use Committee (IACUC) of the University of Houston. KnockEpilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

in mice were generously provided by Drs. K. Yamakawa and I. Ogiwara12 (RIKEN, Japan). Heterozygous (HET) and wild-type (WT) littermate mice were used. The genotyping was performed by Transnetyx. Seizures in this mSMEI have been previously reported to occur starting at P18.12 Spontaneous generalized Racine scale 6 seizures27 in heterozygous animals in the colony studied here were observed starting at P14. Slice preparation Mice (P16–22) were anesthetized with isoflurane, decapitated, and the brains immediately removed. Transverse dorsal isolated hippocampal sections (350 lm thickness) were cut in cold dissection solution (in mM: 2.6 KCl, 1.23 NaH2PO4, 24 NaHCO3, 0.1 CaCl2, 2 MgCl2, 205 sucrose, and 10 glucose) using a Vibratome and were incubated for a half hour in normal artificial cerebrospinal fluid (ACSF; pH 7.3, 30°C) containing (in mM): 130 NaCl, 1.2 MgSO4, 3.5 KCl, 1.2 CaCl2, 10 glucose, 2.5 NaH2PO4, and 24 NaHCO3, aerated with 95% O2-5% CO2. After the incubation, slices were stained with 3-{4-[2-(6-dibutylamino)-2-naphthyl]trans-ethenylpyridinium} propane sulfonate (Di-4ANNEPS) (final dye concentration is 0.05 mg/ml) and left to recover for an additional hour at 30°C.21 The slices were transferred to a submersion recording chamber (Warner Instr., Hamden, CT, U.S.A.) and continuously perfused (2 ml/min, at 32°C) with the oxygenated ACSF. Optical imaging and electrical recordings We used a combination of in vitro electrophysiology (extracellular field potential recording and whole-cell patch clamp recordings) and fast VSDI. All electrical recordings were performed using MCC 700 amplifiers (Axon Instruments). Electrical data were acquired at 4 kHz, digitized at 10 kHz using a Digidata DAC board and pClamp software (Molecular Devices, Sunnyvale, CA, U.S.A.). Optical data were sampled at a 250 Hz rate with MiCam 02 (192 9 128 pixels, Sci Media, Costa Mesa, CA, U.S.A). For electrical whole-cell voltage-clamp recordings, we used borosilicate glass micropipettes (4–7 MΩ) containing (in mM): 116 Cesium gluconate, 6 KCl, 0.5 EGTA, 20 HEPES, 10 phosphocreatine, 0.3 NaGTP, 2 NaCl, 4 MgATP and 0.3% Neurobiotin (pH 7.25, 295 mOsm) and 5 mM QX-314 (fast voltage-gated conductance blocker). The calculated chloride reversal potential was 70 mV at 30°C. Spontaneous inhibitory postsynaptic currents (sIPSCs) and spontaneous excitatory postsynaptic currents (sEPSCs) were recorded in CA1 pyramidal cells. sIPSCs were recorded at 80 mV in the presence of APV (100 lM) and CNQX (40 lM). sEPSCs were recorded at 70 mV in the presence of inhibitory synaptic transmission blocker picrotoxin (50 lM). Both were recorded at 30°C. The sEPSCs and sIPSCs were detected using Mini Analysis (Synaptosoft, Fort Lee, NJ, U.S.A.). The detection threshold was set as twice the root mean square (RMS) of baseline noise, and the

247 Adenosine Control of Hyperexcitability on c-aminobutyric acid (GABA) receptors.28 The differences in responses from stained and unstained slices were not statistically significant and were pooled together for the final analysis. Schaffer collaterals were stimulated at 40 Hz frequency using 10 pulse (0.2 ms) trains (gamma frequency range) repeated every 15 s with stimulus intensity set at the half maximal fEPSP amplitude. fEPSP and VSDI optical data trials were synchronized and averaged. Half maximal value was determined for each experiment. The average stimulation intensity was 458.3 lA in WT and 508.3 lA in HET. Action potentials (APs) in whole-cell patch clamped cells were induced by depolarizing 500 msec-long square wave current injections, incrementally increased by 30 pA for each step. The average values of the amplitude and full width at half-maximum amplitude of APs were calculated using 20 APs per cell using Matlab software (Mathworks, Natick, MA, U.S.A.). Threshold of AP was calculated separately using ramp current injections. The steepest change in the slope of membrane potential corresponded to the start of AP.

sEPSCs and sIPSCs detected by the software were visually checked to avoid potential errors. Average RMS values were statistically similar for both IPSC and EPSC between HET and WT tissue. For IPSC calculations, we studied activity in three 20 s segments (20 s apart) for each cell. Because EPSCs have lower frequency, we used 2 min long segments for their analysis. Between 200 and 400 EPSCs and IPSCs were analyzed per each cell. Cumulative distribution of the amplitudes and interevent interval (frequency) were plotted in Prism software (GraphPad, La Jolla, CA, U.S.A.). To characterize excitatory neural circuit activity, stimulating electrodes (concentric bipolar metal, 200 lm in diameter (FHC, Bowdoin, ME, U.S.A.) were placed on the Shaffer collaterals. Field excitatory postsynaptic potential (fEPSP) recordings were performed in hippocampal slices concurrently with the VSDI (Fig. 1). Extracellular recording electrodes (1–2 MΩ, 0.9% saline) were placed in the CA1 radiatum layer. To monitor VSDI signals, the slices were illuminated with halogen (150 W; excitation 522–550 nm; emission – 580 nm; dichroic – 565 nm). Input-output (I-O) characteristics of fEPSPs were calculated using the same intensities of stimulation and incrementally (100 lA) raising them until the maximal responses or, occasionally, population spikes were obtained. I-O calculations were performed in stained and unstained slices to rule out the possible modulation of Di-4ANEPPS A

Hyperthermia-induced FSLEs FSLEs were induced using hyperthermia in a submersion bath chamber by raising the temperature of ACSF with reduced magnesium (0.6 mM) gradually by 1°C/min from

B

C

D

E

Figure 1. Increased excitation in the CA1 of mSMEI. (A) evoked fEPSP amplitude measurements. Incremental increases in the stimulation currents of Schaffer collaterals elicited significantly larger responses in the HET hippocampi area CA1 at a wide range of stimulus amplitudes tested (N = 8 WT, green and N = 7 HET, green; p < 0.05, unpaired point-by-point t-test). Inset shows representative WT and HET half maximal fEPSP traces. Scale bar: 5 msec, 0.2 mV. Regression analysis of these data showed that the slope of the regression lines between WT and HET were also statistically significant (p = 0.0329). (B, C) Electrical traces of sEPSCs from the pyramidal cells recorded in HET and WT tissue. Downward deflections in the electrical traces are spontaneous inward excitatory currents or sEPSCs. (D) Cumulative distribution plot of the interevent interval duration or frequency of occurrence of sEPSCs (N = 12 WT, green; N = 16 HET, red; p = 0.9877, K–S test). (E) Cumulative distribution plot of the sEPSC amplitudes. (N = 12 WT, green; N = 16 HET, red; p < 0.0001, K–S test). sEPSC amplitude was increased in the HET tissue, as indicated by the rightward shift in the cumulative distribution plots. **Statistical significance. Mean values and unpaired t-test results: sEPSC IEI: 896.70  107.70 msec (HET) versus 1087.00  193.00 msec (WT), p = 0.3684. sEPSC amplitude: 20.89  1.67 pA (HET) versus 15.21  1.37 pA (WT), p = 0.0186. IEI, interevent interval. Epilepsia ILAE Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

248 F. Gu et al. 32°C up to 42.5°C. Using this protocol, typically 2–4 FSLEs were elicited in each slice. FSLEs were defined by extracellular synchronized bursts, followed by a depolarizing shift and the typical ictal-like depolarization and ringing recorded extracellularly. FSLEs could not be induced with just 0.6 mM Mg2+ in the ACSF at 32–36°C or during hyperthermia (37–42.5°C) in ACSF containing 0.6 mM Mg2+ and A1R agonist CPA. Whole-cell recordings of the pyramidal cell activity were also performed before and during FSLEs. Passive and active membrane properties in WT and HET slices were quantified at 32°C. Drugs and application To modulate hippocampal hyperexcitability and FSLEs, we used A1R agonist CPA dissolved in dimethyl sulfoxide (DMSO) and added to the bath ACSF solution. CPA took effect within 2 min of its application. We tested 10 and 1 lM and 500, 100, 50, and 10 nM concentrations of CPA. Because CPA acts through G-protein–linked mechanisms, the washout of CPA was not possible in the current experiments. In a subset of control experiments (n = 5 slices), we confirmed that the vehicle DMSO on its own does not affect the size of fEPSP responses (final DMSO dilution in the ACSF was 0.0005%). A1R blocker DPCPX concentration was 200 nM. All drugs and dyes were obtained from Sigma-Aldrich (St. Louis, MO, U.S.A.). Data and statistical analysis Optical and electrical data were analyzed using BrainVision (SciMedia), Mini Analysis (Synaptosoft), and pClamp (Molecular Devices) softwares. To increase the signal-tonoise ratio, data analysis in individual slices and during pharmacologic manipulations was performed on the averages of 15 files (electrical and optical). Standard electrophysiologic analysis techniques were used to analyze the recorded fEPSP, EPSC, and IPSC characteristics and action potential (AP) characteristics. For optical analysis, 1,060 msec of data were fitted with an approximate Gaussian curve,29 and full width at half maximal (FWHM) values over the distance of 950 micrometers from the stimulating electrode toward subiculum was calculated using BrainVision (SciMedia). Here, FWHM quantifies distance of neuronal signal spatial spread. FWHM calculations along the orthodromic neural activity propagation trajectory included 15 frames (60 ms) before and 194 frames (376 ms) after the 40 Hz train stimulation, which has a duration of 225 msec and spanned over the evoked signal as shown in Figure 4. All results are reported as grouped averages with standard error of the mean. Results from WT and HET or treated versus untreated groups were compared using unpaired and paired t-tests, respectively. A p < 0.05 was regarded as a statistically significant value. Nonparametric KolmogorovSmirnov (K-S) test was used to compare the cumulative EPSC and IPSC probability distributions. Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

Results Increased synaptic excitation in mSMEI To study the impact of SCN1A mutation on the synaptic excitation/inhibition balance in the hippocampus, we performed extracellular and whole-cell patch clamp recordings in the CA1 area. fEPSPs were evoked by Schaffer collateral stimulation in the normal ACSF solution (Fig. 1). fEPSP amplitudes were measured in WT and HET knock-in tissue using the same stimulation intensities. Stimulation-response (or input-output) measurements showed that lower amplitude electrical stimulation over a broad range of stimulus amplitudes evoked higher amplitude fEPSPs in the HET tissue than in the WT (Fig. 1A). To further elucidate how the SCN1A mutation affects synaptic excitation, we performed whole-cell patch clamp recordings of sEPSCs in the hippocampal CA1 pyramidal cells (Fig. 1B,C). Analysis of the distributions of the frequency and amplitude of sEPSCs (Fig. 1D,E) showed that the amplitude increased but that the frequency of sEPSCs was not significantly different in HET hippocampi as compared to WT. Therefore, both the evoked and spontaneous excitatory responses are increased in amplitude in the hippocampal tissues of animals with SCN1A mutation. To test whether increase in sEPSC amplitude was associated with increased action potential (AP) events or excitability in the pyramidal cells, we tested and found that neither the resting membrane potentials (WT: 56.65  0.91, N = 10, HET: 55.86  2.03, N = 8, p = 0.6970), nor AP generation threshold ( 46.53  1.09, N = 8, 46.06  0.89, N = 8, p = 0.7443), AP amplitude (WT: 66.58  2.74, N = 10, HET: 70.84  4.22, N = 8, p = 0.3928), and AP width (WT: 1.319  0.053, N = 10, HET: 1.461  0.058, N = 8, p = 0.0927; unpaired t-tests for all) were different in WT as compared to the HET cells. Impaired synaptic inhibition in mSMEI To further study the synaptic E/I balance in this model, analysis of spontaneous inhibitory postsynaptic currents (sIPSCs) was performed (Fig. 2A,B). Cumulative distributions of sIPSC frequency and amplitude were calculated and statistically compared (Fig. 2C,D). sIPSCs were significantly less frequent and smaller in amplitude in the CA1 pyramidal cells from the HET tissue. Changes in membrane input resistances of the pyramidal cells could not account for the differences in sEPSC/ sIPSC amplitudes (WT: 190.4  11.49 MΩ, N = 14 cells and HET: 183.6  16.78 MΩ, N = 11 cells, p = 0.7327). Aberrant hippocampal activity spread in mSMEI To further analyze how this knock-in mutation affects CA1 circuit activity, we performed electrical fEPSP train stimulations and simultaneously visualized the extent of the neural activity in the hippocampal circuits using VSDI

249 Adenosine Control of Hyperexcitability A

B

C

D

Figure 2. Decreased inhibition in mSMEI. (A, B) Whole-cell voltage-clamp traces from CA1 pyramidal cells obtained in the presence of glutamatergic transmission blockers CNQX (40 lM) and APV (100 lM). Arrows indicate a portion of expanded traces on the right. (C, D) Cumulative distribution of sIPSC interval (C) and amplitudes (D). Both frequency (C) and amplitudes (D) of the sIPSCs were significantly reduced in the HET pyramidal cells (frequency: N = 10 HET, red; N = 8 WT, green; p < 0.0001, K–S test; amplitude: N = 10 HET, N = 8 WT, p < 0.0001, K–S test). Mean values and unpaired t-test results: sIPSC IEI: 201.70  29.88 msec (HET) versus 97.30  25.79 msec (WT), p = 0.0206. sIPSC amplitude: 12.93  1.38 pA (HET) versus 17.65  1.37 pA (WT), p = 0.0295. IEI, interevent interval. Epilepsia ILAE

(Fig. 3). Fast optical acquisition allowed us to visualize the 40 Hz trains of stimulation. When Schaffer collaterals were stimulated at 40 Hz, intensity of stimulation of half maximal response led to abnormally large and sustained amplitude of fEPSP responses (Fig. 3A–C) and activated neural activity in a wider area in HET tissue as compared to WT tissue (Fig. 3D,E). We calculated the difference in the spatial extent of the evoked neuronal responses from the site of the stimulating electrode using FWHM value of the peak normalized fluorescence measure (Fig. 3F–H).21 Evoked neural activity in the HET tissue was detected further distances from the stimulation site, consistent with the E/I imbalance. It is also important to note that the 40 Hz trains often produced antidromic hippocampal circuit activation in the HET hippocampi (pulses 6 and 7, Fig. 3D). Coupled with E/I imbalance, this further suggests a significant disruption in the proper spatiotemporal hippocampal circuit activation pattern. Lowered febrile seizure-like event threshold in mSMEI In SMEI, neural circuit activity is highly sensitive to increases in temperature. To test whether hyperthermia would reveal differences in FSLE temperature threshold and incidence in mSMEI, we increased temperature 1°C/ min up to 42.5°C. We were able to reliably induce FSLEs (Fig. 4A,B) and found that hippocampal slices from the

HET animals were highly susceptible to prolonged FSLEs. Temperature and reduced magnesium were sufficient to reliably induce FSLEs (Fig. 4). In warmed ACSF, shorter bursts preceded prolonged and sustained ictal-like activities. Ictal-like activity was preceded by prolonged subthreshold depolarizations, gradually turning into bursts of action potentials. FSLEs terminated when the action potential bursting subsided into the subthreshold depolarizations in hyperthermia (Fig. 4B). Interictal bursting in the pyramidal cells was infrequent (asterisks in Fig. 4A). A series of studies with Mg2+[1.2 mM]o failed to produce reliable FSLEs (n = 5 HET and n = 5 WT slices), which indicates that glutamate activation of the N-methyl-D-aspartate (NMDA) receptor and decrease in the surface charge screening are likely important mechanisms contributing to ictal activity formation in hyperthermia. Hippocampal slices from HET animals showed significantly lower temperature threshold for induction of FSLE. FSLEs in the HET tissue were significantly higher in incidence and longer in duration in comparison to those recorded in the WT hippocampi (Fig. 5C–E). Adenosine A1 agonist controls hippocampal circuit hyperexcitability Adenosine, a purine and the core of ATP, has recently piqued clinical interest due to its role as an endogenous Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

250 F. Gu et al. A

B

C

D

F

E

G

H

Figure 3. Increased propagation of neural activity in mSMEI hippocampal circuits. (A) Average fEPSP amplitudes during the 40 Hz 10 pulse stimulation in WT (red) and HET (green) slices. Stimulation was produced at half maximal responses. HET tissue showed larger half maximal responses that persisted throughout the stimulation train. Regression analysis showed that the intercepts of the WT and HET regression lines were significantly different (p < 0.0001). (B, C) Individual representative examples of the evoked fEPSP responses in WT (b) and HET (c) slices. Scale bars: 0.2 mV, 25 msec. (D) Photomicrographs in the upper and lower panels depict transverse hippocampal slices overlaid with the normalized average (15 trials) VSD signals in HET (upper) and WT (lower) tissues. Thick black line is the stimulating electrode (200 lm tip) and the site of Schaffer collateral stimulation in CA1 area. Each frame corresponds to the peak of the signal produced by 10 stimulation pulses (P1–P10). Forty Hertz train stimulation in the wild type (WT) tissue evoked a typically small and concise neuronal activity map. The same intensity stimulation in the SMEI hippocampi activated neural activity detectable in wider area. Note antidromic signal activation in the CA3 area of HET hippocampus. Scale bar = 250 lm. (E) Optical traces of 40 Hz stimulation in WT (B) and HET (D) tissue from a representative pixel (*symbol and arrow) in CA1 region. (F) Example of neural activity spread calculation using FWHM. To analyze the average distance of the spatial spread of signal, we used a 950-lm–long line (indicated by the black arrow) that extended over the approximate center region of the evoked signal. Scale bar = 200 lm. (G) Average propagating neural activity (distance, y axis) over time (time, x axis) is shown for 265 frames, which included time periods before, during, and after the 40 Hz stimulation. (Circled letter S indicates location of the stimulating electrode). (H) FWHM measurements of the optical signal show that in HET tissue the signal evoked by the stimulation was detected further distances from stimulation site toward the subiculum region as compared with the WT responses (HET: 640.5  75.08 lm, N = 8, red; WT: 443.0  33.80 lm, N = 9, green; p = 0.0247, unpaired t-test). Epilepsia ILAE

Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

251 Adenosine Control of Hyperexcitability A

B

C

D

E

Figure 4. Decreased FSLE threshold in mSMEI. (A) Repetitive seizures evoked by hyperthermia simultaneously recorded using extracellular and patch clamp electrodes. Patch clamp was performed in the pyramidal CA1 cell in HET tissue. Red gradient arrow illustrates increase in ACSF temperature. Temperature was leveled when the first FSLE began (39°C). (B) Example of the enlarged FSLE from another pyramidal cell in HET tissue. Top trace - whole cell and bottom trace - extracellular recording traces. Inset illustrates further expanded electrical traces. (C) Incidence of FSLEs was significantly higher in the HET hippocampal tissue (HET = 74.1%, N = 27; WT = 33.3%, N = 24). (D) FSLEs in HET tissue had a significantly lower temperature threshold during hyperthermia (HET = 38.74  0.36°C, N = 20, red; WT = 40.61  0.48°C, N = 8, green; p = 0.0076, unpaired t-test). (E) FSLEs that formed in HET tissue were also significantly longer in duration (HET = 39.88  3.96 s, N = 25; WT = 26.54  4.12 s, N = 12; p = 0.0445, unpaired t-test). Epilepsia ILAE

anticonvulsant.30,31 The majority of adenosine’s neuroprotective and anticonvulsant effects are mediated by A1R, a G-protein–coupled receptor. Its activation reduces presynaptic calcium influx into synaptic terminals, increases potassium currents, and inhibits the release of glutamate.32 To determine the effects of adenosine agonist on WT and HET neural circuit activity, we studied the effects of A1R agonist CPA on the evoked neural activity propagation at 32°C as well as on FSLEs evoked by hyperthermia. Our initial experiments using 1 and 10 lM concentrations of CPA33 showed that at these concentrations, CPA completely eliminated fEPSPs in WT and HET tissues. We then used CPA concentrations of 500, 100, 50, and 10 nM, and found that 50 nM concentration was sufficient to signifi-

cantly reduce synaptic hyperexcitability in HET and WT tissues without blocking synaptic transmission (Fig. 5A–D). Data presented here were obtained using a 50 nM concentration of CPA. In a few instances, when the recording electrode was placed in the CA1 pyramidal layer, low amplitude stimulation resulted in population spike responses in the HET tissue. In those instances, 50 nM concentration of CPA effectively reduced the exaggerated spike into fEPSP (Fig. 5A). Because A1Rs are metabotropic, to confirm that CPA is acting through A1R, we used A1R antagonist DPCPX (Fig. 5B). Although CPA reliably reduced the evoked responses in the control ACSF, in the presence of A1R antagonist DPCPX (200 nM), the A1R agonist CPA Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

252 F. Gu et al. A

C

B

D

E

F

G

Figure 5. A1R agonist reliably controls synaptic, circuit, and hyperthermia-induced hyperexcitability in mSMEI. (A) Electrical traces of the evoked potentials recorded extracellularly. In this representative example, very low stimulation intensity (300 lA) evoked population spike (bottom trace, recording electrode placed in the pyramidale layer). Addition of 50 nM CPA reduced this spike into a fEPSP response. (B) Pharmacologic manipulation of fEPSP response amplitudes with A1R agonist CPA and antagonist DPCPX. Example traces of the responses recorded in the CA1 during the 40 Hz train stimulation in: (1) regular ACSF; (2) ACSF with 50 nM CPA; (3) ACSF with CPA and DPCPX. DPCPX prevented CPA from decreasing fEPSP amplitude. (C) Average fEPSP measurements in HET (N = 6) tissue before and after addition of CPA during 10 pulse 40 Hz stimulation pulses. Fifty nanomolar CPA significantly reduced fEPSP trains in the HET hippocampus (paired t-test; p < 0.0001). (D, E) Example of evoked neural activity maps (dF/Fmax) in HET tissue at P19. Fifty nanomolar of CPA reduced the abnormally wide circuit excitation. Scale bar (white line) – 200 lm. (F) FWHM measurements of the spatial extent of the detected neural activity showed that CPA significantly confined the evoked activity in the presence of CPA (HET: 591  106 lm, green; HET + CPA: 491  106 lm, red; p = 0.0321; N = 7; paired t-test). (G) Example of extracellular recording trace, showing that addition of CPA after the first FSLE blocks formation of subsequent FSLEs (N = 10 HET slices). Epilepsia ILAE

was, as expected, ineffective at reducing evoked fEPSPs (N = 5 slices). To determine if reduction in fEPSP size was associated with the control of spatial extent of neural activity, we again used electrophysiology and VSDI. Spatiotemporal measurements showed that CPA reduced the extent of the evoked neural activity (Fig. 5B–E). Unsurprisingly, when CPA was applied following the formation of the FSLEs, subsequent FSLEs were efficiently blocked. Inclusion of CPA in the ACSF prior to hyperthermia failed to evoke FSLEs in HET tissues (N = 7). This suggests that A1R can significantly modulate hyperthermia-induced hyperexcitability and the spread of aberrant neural activity in developing epileptic circuits. Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

Discussion Synaptic impairments caused by SCN1A mutation SCN1A gene mutations impair functions of sodium channel protein Nav1.1, resulting in decreased sodium currents13 and action potential firing in the parvalbumin-positive inhibitory interneurons.12 This impairment is specific to neocortical and hippocampal interneurons and has not been reported to affect pyramidal cell AP generation. Our analysis of pyramidal cell passive and active membrane properties also supports this notion. In the present study, E/I imbalance in the SCN1A mutant during the third postnatal week is due to both loss of synaptic inhibition and increased excitation. Loss of inhibition is

253 Adenosine Control of Hyperexcitability consistent with the Nav1.1 location in the axon initial segment of the inhibitory cells. Increase in the spontaneous and evoked excitation levels in mSMEI tissue suggest an additional, compounding problem. Because passive and active membrane properties of the pyramidal cells were similar in HET and WT tissues, this indicates that the increase in excitation in CA cells could be due to potentiated synapses or increased postsynaptic glutamatergic signaling. Other recent studies of the knock-out mSMEI model also suggest that the hippocampus is hyperexcitable.22,34 Different from our results showing significant changes in frequency and amplitudes of sIPSC (decrease) and sEPSC (increase) reported here, Han et al. found that only the frequencies of sIPSCs and sEPSCs were altered. These discrepancies with current results could be due to a different model and age of animals studied, distinct experimental settings, and methods of analysis. Nonetheless, to develop a more comprehensive understanding of SMEI, future studies will need to determine how this mutation affects synaptic E/I balance, plasma membrane characteristics, and firing properties in the variety of the inhibitory cell subtypes. Several factors, including developmental E/I imbalance, increased excitation (pre- or postsynaptic), and inhibitory failure caused by the SCN1A mutation, possibly contribute to the formation of early life febrile seizures and the associated sequel of cognitive and social dysfunctions in SMEI. Fast functional imaging of E/I imbalance in neural circuits Microcircuit and larger scale imaging modalities act as important tools for understanding the interactions between various brain regions and the cells within them. VSDI provides a way to quantify characteristics of neural activity propagation and to identify regions that drive epileptiform activity. In epileptiform tissue from chronic epilepsy models, VSDI studies show that evoked activity maps are spatially increased, and often, neural activity does not follow normal synaptic connectivity techniques.19 Our previous work in the 4-aminopyridine model using VSDI showed that increases in synchrony even during shorter duration spontaneous interictal-like bursts are also associated with the wider area of burst propagation in the hippocampus.21 At present, the use of a combination of electrophysiology and VSDI to study synaptic and circuit activity has for the first time allowed visualization of neural circuit activity in the transgenic model of pediatric epilepsy. We show that the previously reported loss of inhibition results in circuitwide dysfunction, reflected by an increase in the evoked signal in the HET tissues. Evoked responses in HET tissue were also more likely to exhibit antidromic activation, suggesting that loss of functional inhibition would disrupt anatomic specificity in hippocampal SMEI circuit activation. Therefore, in the near future, it is important to eluci-

date whether the other pathways connecting hippocampal and entorhinal cortices are also affected by the SCN1A mutation and to determine how individual excitatory and inhibitory cell activity21,35 dynamically creates seizures in SMEI. Conceivably, hyperexcitability in the CA1 could be compensated for by the surrounding hippocampal regions, for example, via decreased excitation or increased inhibition along the perforant or the mossy fiber pathways of the hippocampus. Febrile seizure-like events in vitro Although it has been reported that hyperthermia does not induce seizures in Scn1a+/ mice at P17–18,36 our preliminary results show that, in our strain, hyperthermia (38–42°C) can induce seizures both in heterozygous mice and, with a much lower frequency, in WT mice at P16–22 (data not shown), as it has been reported in Lewis rats.37,38 In vitro, models of febrile seizures using hyperthermia present a unique opportunity to mimic some of the conditions and isolate some of the cellular mechanisms that could potentially lead to FS in vivo. Although in most instances, the convulsant chemicals or the electrical stimulations in combination with heating are used to evoke FS in vitro,39 at least one recent report indicates that FSLEs in WT tissue can be induced with just hyperthermia40 and that, in a different experimental setup, hyperthermia in slices induces spreading depression.22,41 The difference in experimental outcomes is likely due to the differences in age of the animals, the structures studied, tissue preparation, and even the type of recording chamber used. FSLEs recorded during hyperthermia here suggest that mSMEI tissue can be used to study FSLEs at P16–22. Furthermore, our results show that there is a significant difference in the incidence and temperature threshold of FSLEs, with mSMEI tissue demonstrating a significantly lower threshold. These “febrile seizures in a dish” may become a useful model for testing actions of novel compounds on neural activity during hyperthermia and FSLEs. Several lines of evidence suggest that, on a cellular level, GABAergic inhibition begins to fail during hyperthermia.42,43 The most recent report suggests that in the juvenile hippocampus (P13–16), inhibitory interneurons are highly sensitive to temperature increases.40 Our preliminary studies of the isolated IPSCs also suggest that IPSCs are more sensitive to temperature increases compared to EPSCs (data not shown). Sensitivity of inhibition to hyperthermia presents a compounding problem in SCN1A mice, which already have lowered levels of inhibition at ambient temperatures (30–32°C) due to the mutation in the interneuron-specific Nav1.1 channel. Conversely, hyperthermia was reported to increase postsynaptic excitatory signaling.44 This effect suggests that an SCN1A mutation would make hippocampal circuits highly susceptible to hyperexcitability during hyperthermia. FSLEs and even seizures in Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

254 F. Gu et al. vivo may occur when the E/I imbalance during hyperthermia reaches a critical state in which, however transiently, excitation is potentiated and inhibition is nearly diminished. On the other hand, it is likely that the mechanisms that lead to FSLEs evoked by hyperthermia in lowered magnesium solution may be distinct from those that lead to FS in vivo. Modulating network hyperexcitability with A1R agonist SMEI remains one of the most pharmacoresistant forms of epilepsy.10 Currently, the most common treatment for SMEI involves the use of GABA modulators to enhance inhibition. Valproate is commonly prescribed to prevent the recurrence of febrile seizures, and benzodiazepines are used for long-lasting seizures, but both are often insufficient.10 Some other drugs, like lamotrigine, carbamazepine, and phenobarbital have also been previously tested for their efficacy against SMEI seizures, but none are known to work reliably.10,45 Increasing GABA synthesis should work well in the networks that contain functionally intact inhibitory cells. Unfortunately, in many forms of epilepsy, including the SMEI models, the inhibitory neurons are affected and may lose their ability to fire action potentials and release GABA. Therefore, novel approaches to treating SMEI and new insights about how different neuromodulators affect neural network activity are most needed. In the present study, we tested purinergic agonist CPA. A1R activation reduces presynaptic depolarization via the activation of delayed rectifying potassium channels46 and blockade of voltage-gated calcium channels.47 Compared to the agents that are typically used to treat SMEI by boosting GABA, use of A1R agonists or small molecule inhibitors downstream of A1R like adenosine kinase48 presents an opportunity to reduce excitation. Several studies show that A1R agonists have anticonvulsant properties in spontaneous electrographic, kindling, kainate, and seizures induced with combination of hyperthermia and an A1R antagonist.23–25,49,50 Furthermore, researchers have demonstrated that a ketogenic diet acting through A1Rs produces anticonvulsant effects in SCN1A mutants.25,51 Most recent experimental evidence also suggests that A1R activation during seizures can likewise prevent depolarizing GABA actions.52 Experiments with the A1R antagonist presented here showed that DPCPX can actually enhance fEPSP response compared to baseline. This implies that endogenous adenosine release may be curtailing the evoked fEPSPs and hyperexcitability. Therefore, the known ontogeny of adenosine dynamics and synaptic E/I imbalance in mSMEI presents unique opportunities to intervene in the epileptogenic process early on, during the critical period of robust synaptic plasticity with the goal of modulating A1R or its downstream cellular targets to rebalance fragile epileptogenic circuits and prevent formation of SMEI. Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487

Acknowledgments This research was supported by the Epilepsy Foundation of America  Dravet Syndrome Foundation (J.Z.),  EFA (EFA) Research Grant (J.Z.), Pre-Doctoral Fellowship award (A.H.), and by the Biology of Behavior Institute at the University of Houston Graduate (F.G.) and Undergraduate Summer Fellowships (A.A.). We thank Drs. Kazuhiro Yamakawa and Ikuo Ogiwara (RIKEN, Japan) for their generous gift of the knock-in mice used in this study. We thank Drs. Chris Dulla, Jeffrey L. Noebels, and Jason Eriksen for their invaluable suggestions and comments on the manuscript.

Disclosure None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

References 1. Dravet C, Bureau M, Oguni H, et al. Severe myoclonic epilepsy in infancy: Dravet syndrome. Adv Neurol 2005;95:71–102. 2. Dravet C. The core Dravet syndrome phenotype. Epilepsia 2011;52 (Suppl. 2):3–9. 3. Guerrini R, Falchi M. Dravet syndrome and SCN1A gene mutation related-epilepsies: cognitive impairment and its determinants. Dev Med Child Neurol 2011;53(Suppl. 2):11–15. 4. Marini C, Scheffer IE, Nabbout R, et al. The genetics of Dravet syndrome. Epilepsia 2011;52(Suppl. 2):24–29. 5. Mulley JC, Scheffer IE, Petrou S, et al. SCN1A mutations and epilepsy. Hum Mutat 2005;25:535–542. 6. Ragona F. Cognitive development in children with Dravet syndrome. Epilepsia 2011;52(Suppl. 2):39–43. 7. Scheffer IE, Zhang YH, Jansen FE, et al. Dravet syndrome or genetic (generalized) epilepsy with febrile seizures plus? Brain Dev 2009;31:394–400. 8. Oakley JC, Kalume F, Catterall WA. Insights into pathophysiology and therapy from a mouse model of Dravet syndrome. Epilepsia 2011;52 (Suppl. 2):59–61. 9. Catterall WA, Kalume F, Oakley JC. NaV1.1 channels and epilepsy. J Physiol 2010;588:1849–1859. 10. Chiron C. Current therapeutic procedures in Dravet syndrome. Dev Med Child Neurol 2011;53(Suppl. 2):16–18. 11. Escayg A, Goldin AL. Sodium channel SCN1A and epilepsy: mutations and mechanisms. Epilepsia 2010;51:1650–1658. 12. Ogiwara I, Miyamoto H, Morita N, et al. Nav1.1 localizes to axons of parvalbumin-positive inhibitory interneurons: a circuit basis for epileptic seizures in mice carrying an Scn1a gene mutation. J Neurosci 2007;27:5903–5914. 13. Yu FH, Mantegazza M, Westenbroek RE, et al. Reduced sodium current in GABAergic interneurons in a mouse model of severe myoclonic epilepsy in infancy. Nat Neurosci 2006;9:1142– 1149. 14. Genton P, Velizarova R, Dravet C. Dravet syndrome: the long-term outcome. Epilepsia 2011;52(Suppl. 2):44–49. 15. Sakauchi M, Oguni H, Kato I, et al. Mortality in Dravet syndrome: search for risk factors in Japanese patients. Epilepsia 2011;52(Suppl. 2):50–54. 16. Zecevic D, Djurisic M, Cohen LB, et al. Imaging nervous system activity with voltage-sensitive dyes. Curr Protoc Neurosci 2003; Chapter 6:Unit 6.17. 17. Grinvald A, Hildesheim R. VSDI: a new era in functional imaging of cortical dynamics. Nat Rev Neurosci 2004;5:874–885. 18. Chemla S, Chavane F. Voltage-sensitive dye imaging: technique review and models. J Physiol Paris 2010;104:40–50. 19. Coulter DA, Yue C, Ang CW, et al. Hippocampal microcircuit dynamics probed using optical imaging approaches. J Physiol 2011;589:1893–1903.

255 Adenosine Control of Hyperexcitability 20. Bandyopadhyay S, Hablitz JJ. Dopaminergic modulation of local network activity in rat prefrontal cortex. J Neurophysiol 2007;97:4120–4128. 21. Hazra A, Rosenbaum R, Bodmann B, et al. Beta-adrenergic modulation of spontaneous spatiotemporal activity patterns and synchrony in hyperexcitable hippocampal circuits. J Neurophysiol 2012;108:658–671. 22. Liautard C, Scalmani P, Carriero G, et al. Hippocampal hyperexcitability and specific epileptiform activity in a mouse model of Dravet syndrome. Epilepsia 2013;54:1251–1261. 23. Guttinger M, Padrun V, Pralong WF, et al. Seizure suppression and lack of adenosine A1 receptor desensitization after focal long-term delivery of adenosine by encapsulated myoblasts. Exp Neurol 2005;193:53–64. 24. Boison D, Scheurer L, Tseng JL, et al. Seizure suppression in kindled rats by intraventricular grafting of an adenosine releasing synthetic polymer. Exp Neurol 1999;160:164–174. 25. Masino SA, Li T, Theofilas P, et al. A ketogenic diet suppresses seizures in mice through adenosine A(1) receptors. J Clin Invest 2011;121:2679–2683. 26. Fukuda M, Suzuki Y, Hino H, et al. Adenosine A1 receptor blockage mediates theophylline-associated seizures. Epilepsia 2010;51:483– 487. 27. Luttjohann A, Fabene PF, van Luijtelaar G. A revised Racine’s scale for PTZ-induced seizures in rats. Physiol Behav 2009;98: 579–586. 28. Mennerick S, Chisari M, Shu HJ, et al. Diverse voltage-sensitive dyes modulate GABAA receptor function. J Neurosci 2010;30: 2871–2879. 29. Petersen CC, Grinvald A, Sakmann B. Spatiotemporal dynamics of sensory responses in layer 2/3 of rat barrel cortex measured in vivo by voltage-sensitive dye imaging combined with whole-cell voltage recordings and neuron reconstructions. J Neurosci 2003;23:1298– 1309. 30. Boison D. Adenosine and seizure termination: endogenous mechanisms. Epilepsy Curr 2013;13:35–37. 31. Gouder N, Fritschy JM, Boison D. Seizure suppression by adenosine A1 receptor activation in a mouse model of pharmacoresistant epilepsy. Epilepsia 2003;44:877–885. 32. Dunwiddie TV, Masino SA. The role and regulation of adenosine in the central nervous system. Annu Rev Neurosci 2001;24:31–55. 33. Migita H, Kominami K, Higashida M, et al. Activation of adenosine A1 receptor-induced neural stem cell proliferation via MEK/ERK and Akt signaling pathways. J Neurosci Res 2008;86: 2820–2828. 34. Han S, Tai C, Westenbroek RE, et al. Autistic-like behaviour in Scn1a+/ mice and rescue by enhanced GABA-mediated neurotransmission. Nature 2012;489:385–390. 35. Ziburkus J, Cressman JR, Barreto E, et al. Interneuron and pyramidal cell interplay during in vitro seizure-like events. J Neurophysiol 2006;95:3948–3954.

36. Oakley JC, Cho AR, Cheah CS, et al. Synergistic GABA-enhancing therapy against seizures in a mouse model of Dravet syndrome. J Pharmacol Exp Ther 2013;345:215–224. 37. Fukuda M, Morimoto T, Suzuki Y, et al. Interleukin-6 attenuates hyperthermia-induced seizures in developing rats. Brain Dev 2007;29:644–648. 38. Fukuda M, Suzuki Y, Ishizaki Y, et al. Interleukin-1beta enhances susceptibility to hyperthermia-induced seizures in developing rats. Seizure 2009;18:211–214. 39. Tancredi V, D’Arcangelo G, Zona C, et al. Induction of epileptiform activity by temperature elevation in hippocampal slices from young rats: an in vitro model for febrile seizures? Epilepsia 1992;33:228– 234. 40. Kim JA, Connors BW. High temperatures alter physiological properties of pyramidal cells and inhibitory interneurons in hippocampus. Front Cell Neurosci 2012;6:27. 41. Wu J, Fisher RS. Hyperthermic spreading depressions in the immature rat hippocampal slice. J Neurophysiol 2000;84:1355–1360. 42. Qu L, Liu X, Wu C, et al. Hyperthermia decreases GABAergic synaptic transmission in hippocampal neurons of immature rats. Neurobiol Dis 2007;27:320–327. 43. Qu L, Leung LS. Mechanisms of hyperthermia-induced depression of GABAergic synaptic transmission in the immature rat hippocampus. J Neurochem 2008;106:2158–2169. 44. Schiff SJ, Somjen GG. The effects of temperature on synaptic transmission in hippocampal tissue slices. Brain Res 1985;345:279– 284. 45. Oakley JC, Kalume F, Yu FH, et al. Temperature- and age-dependent seizures in a mouse model of severe myoclonic epilepsy in infancy. Proc Natl Acad Sci USA 2009;106:3994–3999. 46. Takigawa T, Alzheimer C. Phasic and tonic attenuation of EPSPs by inward rectifier K+ channels in rat hippocampal pyramidal cells. J Physiol 2002;539:67–75. 47. Manita S, Kawamura Y, Sato K, et al. Adenosine A(1)-receptormediated tonic inhibition of glutamate release at rat hippocampal CA3CA1 synapses is primarily due to inhibition of N-type Ca(2+) channels. Eur J Pharmacol 2004;499:265–274. 48. Boison D. The adenosine kinase hypothesis of epileptogenesis. Prog Neurobiol 2008;84:249–262. 49. Boison D. Adenosine dysfunction and adenosine kinase in epileptogenesis. Open Neurosci J 2010;4:93–101. 50. Masino SA, Kawamura M Jr, Ruskin DN, et al. Purines and neuronal excitability: links to the ketogenic diet. Epilepsy Res 2011;100:229– 238. 51. Dutton SB, Sawyer NT, Kalume F, et al. Protective effect of the ketogenic diet in Scn1a mutant mice. Epilepsia 2011;52:2050–2056. 52. Ilie A, Raimondo JV, Akerman CJ. Adenosine release during seizures attenuates GABAA receptor-mediated depolarization. J Neurosci 2012;32:5321–5332.

Epilepsia, 55(2):245–255, 2014 doi: 10.1111/epi.12487