Preseizure increased gamma electroencephalographic activity has no effect on extracellular potassium or calcium

Journal of Neuroscience Research 85:906–918 (2007)

Preseizure Increased Gamma Electroencephalographic Activity Has No Effect on Extracellular Potassium or Calcium Marita Broberg,1* Kenneth J. Pope,2 Michael Nilsson,3 Angus Wallace,2 Jodie Wilson,4 and John O. Willoughby1 1

Center for Neuroscience and Department of Medicine, Flinders University, Adelaide, South Australia, Australia 2 School of Informatics and Engineering, Flinders University, Adelaide, South Australia, Australia 3 Institute of Neuroscience and Physiology, Go¨teborg University, Go¨teborg, Sweden 4 Biomedical Engineering, Flinders Medical Centre, Bedford Park, South Australia, Australia

Extracellular ion concentrations change during seizures in seizure models. [K+]o increases and [Ca2+]o decreases, resulting from population discharges, enhanced neuronal excitability, though not obviously before seizure onset. In acute pharmacological epilepsy models, there are striking increases in preictal high-frequency (gamma) electroencephalographic (EEG) activity. It is not known whether enhanced gamma EEG results in ionic changes, because gamma and ions have not been measured simultaneously. In this study, unanesthetized, paralyzed rats were given intravenous injections of kainic acid or picrotoxin to induce EEG discharges. Changes in EEG, [K+]o, and [Ca2+]o in cortex and hippocampus were recorded. Kainic acid caused small [K+]o fluctuations, without a temporal relationship of these with increased gamma EEG or with onset of discharges. Gamma EEG increases after picrotoxin also failed to affect [K+]o and [Ca2+]o. Picrotoxin-induced electrical discharges led to [K+]o rises of >9 mM and [Ca2+]o falls of 0.1–0.2 mM. Kainic acidinduced discharges generated only moderate (2–3 mM) rises in [K+]o and no changes in [Ca2+]o. In both models, there were large potassium rises (15–80 mM) and calcium falls (>0.5 mM), suggesting spreading depressions. Small [K+]o fluctuations after kainic acid are consistent with disruption in potassium homeostasis, possibly because of depolarization of astrocytes. To reveal possible latent [K+]o or [Ca2+]o changes, we injected fluorocitrate intracortically to impair astrocytic function, before administering picrotoxin. Even fluorocitrate did not cause gamma-related ion changes but did cause low-magnitude, transient, potassium increases and slower potassium homeostasis during discharges, minor changes consistent with involvement of both astrocytes and neurons in [K+]o regulation. VC 2007 Wiley-Liss, Inc. Key words: ions; homeostasis; seizure; astrocytes

Tight regulation of extracellular potassium ([K+]o) and calcium [Ca2+]o is important for normal neural function. Neuronal firing is known to increase [K+]o levels and ' 2007 Wiley-Liss, Inc.

to decrease [Ca2+]o, and both of these changes lead to increased neuronal firing (Pan and Stringer, 1997; Lerche et al., 2001; Bikson et al., 2002), postulated to enhance excitability during seizure development (Schwartzkroin et al., 1998; Steinlein and Noebels, 2000). Astrocytes are important in this step as regulators of potassium (Sykova, 1983), through passive KCl uptake (Walz and Wuttke, 1999), the Na/K-ATPase pump (Walz and Wuttke, 1999), and redistribution of [K+]o (Futamachi and Pedley, 1976; Amzica et al., 2002). The convulsant kainic acid (Lothman and Collins, 1981; Lothman et al., 1981) is a glutamate agonist activating glutamate receptors of the AMPA and kainate type that are distributed throughout the central nervous system (Wisden and Seeburg, 1993). Receptors responding to kainic acid are present on both astrocytes (Sontheimer et al., 1988) and neurons. After systemic administration, there is a delay of 30–45 min between kainic acid injection and behavioral seizures (Willoughby et al., 1997), during which time there are both marked increases in the power of high-frequency electroencephalographic (EEG) rhythms (around 40 Hz, gamma rhythms) and nonconvulsive EEG discharges (Olsson et al., 2006; Medvedev et al., 2000). It is not known whether possible changes in potassium or calcium caused by preictal gamma EEG activity contributed to seizures caused by kainic acid.

Contract grant sponsor: Swedish Foundation for International Cooperation in Research and Higher Education (STINT). *Correspondence to: Marita Broberg, Department of Medicine, Flinders University, P.O. Box 2100, Adelaide, South Australia 5001, Australia. E-mail: marita.broberg@flinders.edu.au Received 23 August 2006; Revised 12 October 2006; Accepted 31 October 2006 Published online 22 January 2007 in Wiley InterScience (www. interscience.wiley.com). DOI: 10.1002/jnr.21162

Ion and Gamma Changes Preseizure

The convulsant picrotoxin blocks the chloride channel linked to g-aminobutyric acid type A (GABAA) receptors that are widely distributed postsynaptically and extrasynaptically (Mody and Pearce, 2004) as well as on astrocytes (Bureau et al., 1995). Picrotoxin leads to electrical discharges and seizures within 10 min of its systemic administration, during which time there are short-lived spindles of EEG activity associated with myoclonic jerks as well as small increases in gamma EEG activity (Willoughby et al., 1995; Olsson et al., 2006). Blockage of the GABA-chloride channel impairs hyperpolarization of neurons during GABA-mediated inhibition, thus diminishing inhibition. Many studies examining the role of potassium and calcium in seizure onset have used picrotoxin or similar agents to induce seizures. Many of these experiments were undertaken in in vitro preparations in which highfrequency EEG activity might not occur or was not examined. Although studies conducted in anesthetized intact animals do not point to persistent preictal ionic changes (Futamachi et al., 1974; Fisher et al., 1976; Lux and Heinemann, 1978; Somjen and Giacchino, 1985; Lux et al., 1986), it is not known whether these preparations are capable of exhibiting increases in preictal gamma EEG activity. In awake animal models of epilepsy, when increased gamma EEG activity has been recorded (Leung et al., 2005), no recordings of potassium were made. Fluorocitrate administered in low concentrations or small doses within the central nervous system is known to lead to reversible, selective dysfunction of astrocytic metabolism by selective uptake (Hassel et al., 1992) and interruption of the tricarboxylic acid cycle (Paulsen et al., 1987). Given the importance of astrocytes in potassium regulation, fluorocitrate administration would be expected to challenge ionic regulation further in pharmacologic models, as has already been demonstrated in an electrical seizure model (Xiong and Stringer, 1999). Lian and Stringer (2004b) have also shown increased sensitivity to kainate following fluorocitrate administration. In this study, we tested the hypothesis that intensified gamma rhythms (analyzed between 30 and 200 Hz) elevate [K+]o or decrease [Ca2+]o and thus increase neuronal excitability, triggering electrical discharges. We also used fluorocitrate in some experiments to test further the role of astrocytes in this process. We were unable to demonstrate persistent preictal changes in ion concentration, but we did observe the well-described changes during discharges and with spreading depressions (SDs), both of which were affected by fluorocitrate. MATERIALS AND METHODS Ion-Selective Microelectrodes Double-barrelled ion-selective microelectrodes (ISMs) were manufactured by using a modification of the technique described by Oehme and colleagues (Oehme and Simon, 1976; Ammann, 1986). Thick-walled, borosilicate theta glass capillaries (2.0 mm OD; Clark Electromedical Instruments) were pulled using a Flaming/Brown micropipette puller (model P97; Sutter Instrument Co., Novato, CA). Silanization of the ionJournal of Neuroscience Research DOI 10.1002/jnr

907

selective barrel was achieved by dipping the tip of the pulled electrodes into tributylchlorosilane (3%; Fluka, Heidelberg, Germany) in 1-chloronaphthalene (Aldrich, Milwaukee, WI) after protection of the reference barrel with water. The silanization was fixed by heating of the electrodes (1808C, 1 hr). For K+-selective microelectrodes, the tip (inner diameter 2–3 lm) of the ion-selective barrel was filled to 100–300 lm with a lowimpedance membrane cocktail based on the neutral K+-selective ion carrier valinomycin (Fluka Cocktail A 60031; Wuhrmann et al., 1979). The ion-selective barrel was then back-filled with 150 mM KCl solution, giving a tip resistance between 70 and 90 MO. For Ca2+-selective microelectrodes, the tip (inner diameter 2–3 lm) of the ion-selective barrel was filled to 100– 300 lm with a calcium-ionophore (Fluka Cocktail A 21048). The ion-selective barrel was then back-filled with 150 mM CaCl2 solution, giving a tip resistance between 70 and 90 MO. The reference barrel was filled with 150 mM NaCl solution (tip resistance 2–20 MO). Calibration of each electrode in saline (2 mM to 50 mM KCl or 0.19 to 25 mM CaCl2, respectively) was performed before and after experiments, giving slopes of 50–58 mV per tenfold increase in K+ concentration or 24–29 mV per tenfold increase in Ca2+ concentration, respectively, in accordance with the Nikolsky equation. On some occasions, a correction for drifting of the Ca2+-selective electrodes had to be performed. Drifting of Ca2+ electrodes was assumed to be consistent throughout the experiment (as control measurements implied), and corrections were made by comparing pre- and postexperiment calibration values (slopes being within limits given).

Animal Preparation All animal experiments were approved by the Animal Welfare Committee of Flinders University. Male inbred adult Sprague Dawley rats (n ¼ 84, 260–400 g), known not to express spike wave discharges spontaneously (Willoughby and Mackenzie, 1992), were obtained from the animal house of the Flinders Medical Centre and were anesthetized with pentobarbitone (ip, 62.5 mg/kg; Boehringer, Ingelheim, Germany). A right atrial catheter was inserted via the jugular vein exteriorized at the back of the neck before the animal was fixed in a stereotaxic device. Holes were drilled through the skull and stainless-steel screws for EEG recording were implanted on the surface of frontal cortex area 1 and/or hindlimb cortex. For hippocampal recordings, fine wire (teflon-coated, 80% platinum–20% iridium 25.4-lm wire; Leico Industries Inc., New York, NY) electrodes were stereotaxically implanted in hippocampus (Bregma –5.3; lateral 4.2; dura –2.6). An indifferent electrode was placed anteriorly over the frontal sinus, and an earth electrode in the occipital bone overlying cerebellum. A window through the skull, centred about 4 mm away from the closest implanted EEG electrode, covered by a steel cylinder (4.5 mm in diameter) and sealed with bone wax, was also prepared in some animals. The EEG electrodes, connectors, catheter, and steel cylinder were embedded in dental cement. Buprenorphine (0.1 mg/kg, im; Reckitt & Colman) was administered, and the animals were allowed to recover postsurgically for 1–3 days or for 1 month after hippocampal electrode implantation.

908

Broberg et al.

Experimental Procedures On the day of experiments, the rats were paralyzed (methylscopolamine, 50 lg/kg iv, Sigma plus suxamethonium 5 mg/ kg iv; Astra Zeneca), strapped into a head-shaped mould, and artificially respired throughout the experiment, with a facemask covering the nose. Paralysis itself, in the absence of surgery, is nonnoxious in humans. Methylscopolamine, rather than scopolamine, was used to avoid central nervous system muscarinic antagonist effects (Ferreira et al., 2003). Animals were reparalyzed by injection of more methylscopolamine and suxamethonium at the first sign of whisker movement or, at the latest, 40 min after previous administration. This procedure was witnessed and approved by the Animal Welfare Committee of Flinders University. The bone wax was removed and xylocaine (1%) administered to the dura before a hole in the dura was opened to gain access to the cortical surface. The ISM was introduced (0.6–1.3 mm) into the cerebral frontal cortex or farther in to hippocampus (3.2–3.4 mm). We have previously found intracerebral insertion of microelectrodes in unparalyzed, unanesthetized animals to be nonnoxious. While we introduced the K+ electrodes into the brain, rapid [K+]o increases (presumed cell ruptures) occurred, returning to basal level within seconds. No similar disturbance was seen for the Ca2+ electrodes, but a concentration (activity) shift on entering the brain could be detected. In some animals, an EEG signal was taken from the reference barrel of the ISM before filters were applied. The ISM was allowed to stabilize for a few minutes before convulsants or fluorocitrate were administered. Epileptiform activity was induced by systemic kainic acid or picrotoxin administration. Kainic acid (Sigma, St. Louis, MO) 10 mg/kg or picrotoxin (Sigma) 2.1 mg/kg was dissolved in 10% dimethylsulfoxide (Ajax Chemicals, Sydney, Australia) in 0.9% saline and infused iv over 30 sec. In two animals not receiving injections, the ISM was left for 20 min, during which time no [K+]o changes or abnormal EEG activity was observed. In two animals, EEG activity during paralysis for 1.5 hr was recorded without any ISM present. To cause local astrocytic dysfunction, animals received intracortical injections of sodium fluorocitrate. Fluorocitrate (0.8 nmol in 0.125 ll saline, prepared as described by Paulsen et al., 1987) was pressure injected in the right frontal hindlimb cortex (1.2 mm below pia) via a glass micropipette. We have previously shown this to inhibit astrocytic metabolism selectively (Willoughby et al., 2003). Control animals were injected with 0.8 nmol sodium citrate in 0.125 ll saline. The ISM was then introduced (0.6–1.3 mm) into the cerebral cortex close to (0.1–0.6 mm) or at a distance from (2.4–3.0 mm) the injection site. ISM and EEG recordings were performed for 1.5 hr after fluorocitrate/citrate injection, before picrotoxin (iv, 2.1 mg/kg; Sigma) was given, to reveal any effects of astrocytic dysfunction during picrotoxin-induced epileptiform activity. A small proportion of animals exhibited discharges after fluorocitrate alone (Willoughby et al., 2003), and these were excluded from further study.

Instruments, Sydney, Australia). Half-amplitude cutoff frequencies were at 0.1 and 1,000 Hz. For ion measurements, differential recording between the ion-selective and the reference barrel in the ISM using a high-input impedance amplifier was measured and digitized with the same hardware. A low-pass filter at 10 Hz was applied. EEG data were analyzed by using fast Fourier transformation with 1- or 2-Hz resolution and half-overlapping epochs of 0.5 or 1 sec to generate spectra and spectrograms of EEG power relative to a 1-min segment before drug administration. For spectral comparisons, segments of 10–60 sec were taken late in the recordings, when no discharges, spindles, or postseizure depressions were visible. The EEG spectrograms provide a second-by-second display of changes in strength or power of different frequencies (1–200 Hz), which are contained within the EEG signals but are not evident on visual inspection. The EEG changes in the range up to 20-fold above pretreatment EEG power are represented by a gray-scale stripe from white to black. Successive adjacent 1-sec stripes then provide an image of EEG changes for the entire experiment. These spectrograms together with EEG recordings were analyzed both at second and at minute levels and used to define the time point of onset of spikes, spindles, and electrical discharges. In the kainic acid model, discharges were defined as high-voltage, sharp waves (greater than twice normal EEG oscillations) at low frequencies (