Address correspondence and reprint requests to Dr. A. Bragin at David Geffen School of Medicine at UCLA, 710 Westwood Plaza, Los Angeles, CA 90095, U.S.A. E-mail: email@example.com
Summary: High-frequency oscillations (HFOs) have been described in normal and epileptic brains of animals and humans. These oscillations reflect a short-term integration within neuronal networks and have important functional consequences for normal and pathological processes. We performed a comparative voltage depth profile analysis of normal and pathological HFOs after intrahippocampal kainic acid injection. Sixteen channel recording probes, with 100–200 μm separation between the tips of microelectrodes, were implanted along the CA1—dentate gyrus axis in the anterior hippocampus of adult rats. Guide cannulae were implanted in the CA3 area. After a week of baseline recording kainic acid (KA) (0.2μg/0.2μl) was injected into the CA3 area. Electrical activity continued to be record for the next 3–4 weeks after KA induced status epilepticus. Voltage depth profiles and power spectral analysis of HFOs were performed off-line using DataPac software. Ripple oscillations (80–200 Hz) in the CA1 area and gamma activity (40–80 Hz) in the dentate gyrus remained after status epilepticus. In the group of rats that later developed seizures a new pattern consisting of bursts of population spikes (BPS) occurred. The maximum of amplitude for BPS generated in CA1 was in the pyramidal layer and for those generated in the dentate gyrus was in the granular layer. BPS appeared 2–3 days after status epilepticus and remained for the rest of the experiments. The frequencies of intraburst spikes varied between 80 Hz and 600 Hz. With increasing distance from the area of the burst generation, this activity took on the appearance of HFOs. The occurrence of spontaneous BPS appear to be a primary electrophysiological consequence of status epilepticus when progressive epileptogenesis occurs with maximum of amplitude in the cellular layer. In areas outside of the generator of the BPS, this activity looks more like pathological high-frequency oscillations (pHFO), which were observed in earlier experiments.
The processes leading to recurrent spontaneous seizures after an initial precipitating event remain unclear. Hundreds of genes are up- or down-regulated after status epilepticus, resulting in changes in the properties of different receptors, channels, and transporters. Localization of specific patterns of electrical activity that could be markers of ongoing epileptogenesis and thereby predict recurrent seizure occurrence is one of the initial steps in identifying the molecular mechanisms of progressive epileptogenesis.
In our earlier study (Bragin et al., 2004), we showed the appearance of high-frequency oscillations (HFOs) (80–500 Hz) in the hippocampus ipsilateral to prior unilateral intrahippocampal kainic acid (KA) injection. There was a strong positive correlation between the occurrence and persistence of fast ripples (250–500 Hz) as well as pathological ripple frequency oscillations (80–200 Hz) in the dentate gyrus and the development of recurrent spontaneous seizures.
In normal brain during immobility and slow wave sleep, ripples occur in the CA1 region of hippocampus (Buzsaki et al., 1992) and not in the dentate gyrus. We do not know, at this time, whether the same neuronal network generates normal ripples as well as pathological high-frequency oscillations (pHFO) in both ripple frequency and fast ripple frequency, or whether they are generated by two or three different networks. To further study this question, we performed an analysis of voltage depth profiles of HFO at multiple recording sites after status epilepticus induced by unilateral intrahippocampal KA injection.
Microelectrode implantation and recording
Experiments were performed on eight adult Wistar rats (200–250 g) with implanted 16-channel recording probes across the right CA1-DG areas with coordinates AP =−3.5; L = 2.0; V = 4.5 for the deepest microelectrode. Probes consisted of sixteen 20-μm tungsten wires glued together, and cut so the distance between recording sites within each probe was 200 μm. Ground and indifferent electrodes were placed in the skull above the cerebellum. Guide cannulae for KA injection were implanted above unilateral area CA3 with coordinates AP =−3.5; L = 5.0; V = 3.0.
After 1 week of recovery rats were taken for electrophysiological experiments, where electrical activity was recorded during exploratory activity, slow wave sleep and REM sleep. The amplified physiological data were recorded wide-band (0.1–4.0 kHz) and sampled at 10 kHz/channel with 12-bit precision on a Pentium PC using DataPac data acquisition software (Mission Viejo, CA, U.S.A.).
Kainic acid injection
After 3 days of recording of baseline electrical activity KA was injected through implanted cannula into the CA3 area of the right hippocampus adjacent to the recording site (0.2μg/0.2μl). Beginning the next day, electrographic activity was recorded with video monitoring for seizures every day for 8 h/day for at least 30 days.
Analysis of electrophysiological data was carried out off-line on a Pentium computer, using Datapac (Run Technologies, Mission Viejo, CA, U.S.A.) software. Location of recording sites within different layers of CA1 and dentate gyrus was determined on the basis of the shape of evoked potentials to perforant path stimulation and phase reversal of sharp waves (Lomo, 1971; Buzsaki et al., 1992) (see Fig. 1A). Electrical activity was separated by filtering low-pass activity (less than 100 Hz) and HFOs (80–600 Hz). The peak of interictal events was detected by Datapac software in the channel with maximum amplitude and an averaging procedure was performed using this channel as a reference. During power spectral analysis, spectrograms before and after status epilepticus were compared only during EEG patterns indicative of slow wave sleep. Voltage depth profile analysis was performed on the averages for at least 10 events.
After completion of electrophysiological experiments rats were deeply anesthetized (Nembutal, 50 mg/kg) and perfused with 2.5% paraformaldehyde. Then brains were removed and kept in the same solution overnight. On the next day brains were sectioned (60 μm) on a vibratom. Location of recording and stimulating electrodes were verified on the Nissl stained sections.
Local field potential activity in CA1-DG areas before status epilepticus
REM sleep and exploratory activity, theta rhythm occurred in the hippocampus with maximum amplitude near the hippocampal fissure. During slow wave sleep and immobility sharp waves (SPWs) occurred in the CA1 area of hippocampus with maximum negative amplitude in the str. radiatum. They coincided with ripple oscillations in the pyramidal layer (Fig. 1A, double-headed arrow) (Buzsaki et al., 1992; Ylinen et al., 1995). In the dentate gyrus during slow wave sleep dentate spikes occurred, and showed phase reversal in the str. moleculare of the dentate gyrus. Gamma waves in the frequency range 40–80 Hz usually were superimposed on the dentate spikes (Fig. 1A, diamonds) (Bragin et al., 1995a).
Local field potential activity in CA1-DG areas after status epilepticus
On the 2nd or 3rd days after KA injection EEG interictal spikes appeared in the CA1-dentate gyrus areas (Fig. 2B). Some of these interictal spikes (Fig. 2B, insets) were accompanied by HFOs in the frequency range 80–600 Hz (Fig. 2C, D). The amplitude of these oscillations varied from 0.2 mV to 2.5 mV both in the stratum pyramidale of CA1 area and in the dentate gyrus. Ripple oscillations continued to be present in str. pyramidale of the CA1 area however their rate of occurrence was significantly higher after status epilepticus than before (54 ± 19 ± SD compared to 26 ± 21/min ±SD, respectively). Before status epilepticus CA1 ripples were accompanied by sharp waves that were negative in str. radiatum (Fig. 1A), after status epilepticus ripple oscillations were accompanied by positive sharp waves (e.g., Fig. 4A). Power spectral analysis did not demonstrate significant differences in the frequency of ripple oscillations in CA1 pyramidal layer before and after status epilepticus (not shown).
In addition to ripple oscillations another pattern of electrical activity was observed in CA1. This pattern consisted of 2–3 population spikes or bursts of up to 12–15 population spikes (Fig. 3, left part). The frequency of population spikes within bursts varied between 80 Hz and 600 Hz. These bursts of population spikes (BPS) could occur locally in the CA1 area in association with positive sharp waves in stratum radiatum or coincide with ripple or fast ripple oscillations in the dentate gyrus (Fig. 4A). These BPS usually occurred intermixed with ripple oscillations during the same time period.
In the dentate gyrus oscillations in the frequency range of 100–600 Hz appeared only after status epilepticus. They occurred at the rate of 18 ± 11/min (±SD) and were superimposed either on negative or positive waves (Fig. 1B, C). Negative waves had maximum amplitude in the hilus and phase reversed around the CA1 pyramidal layer (Fig. 1B). Positive waves also had maximum amplitude in the hilus with decrements both toward the CA1 area and the thalamus (Fig. 1C) without phase reversal. Depending on the location of the recording electrode these oscillations could look like either waves, or spikes (Fig. 3, right part).
The amplitude of population spikes sharply decreased and the shape dramatically changed in the recording sites located outside of cellular layers (Fig. 3C). Activity recorded from apical dendrites in 300 μm from the cellular layer outlined by dashed box. The bottom record is band-pass (80–600 Hz) filtered and amplified 10 times.
Relationship between pathological HFOs in the CA1 region and dentate gyrus
Sixty to 70% of HFOs that occurred in the dentate gyrus after status epilepticus were accompanied by ripple oscillations in the CA1 area. Because of the high variability in the shape of these oscillations, it was not possible to determine which of these two areas was leading. Fig. 4B illustrates voltage depth profiles of three HFOs. Two of them, with peak outlined by letters “a” and “b” have maximum amplitude within the CA1 pyramidal layer and one with peak outlined by letter “c” within the hilus. Oscillation “a” has phase reversal just below the pyramidal layer, while “b” has phase reversal around the hippocampal fissure. Oscillation “c” has phase reversal even deeper, within the molecular layer. The difference between the voltage depth profiles of deep parts of CA1 oscillations “a” and “b” could indicate an interaction between CA1 oscillations “b” and dentate oscillations “c,” rather than indicate two truly distinctive neuronal generators.
As was shown in the present experiments, ripple oscillations remain in the CA1 area after status epilepticus. Whereas in normal rats they are accompanied by negative sharp waves (Buzsaki et al., 1992), after status epilepticus they are associated with positive sharp waves. Anatomical data (Ramon y Cajal, 1938) indicate that the CA3a area sends most of its Schaffer collaterals to the CA1 stratum oriens, while the CA3b area sends Schaffer collaterals to the CA1 stratum radiatum. Sharp waves that are negative in the stratum radiatum of CA1 indicate that their source is the CA3b area of hippocampus. Sharp waves that are positive in the stratum radiatum indicate that their source is the CA3a area of hippocampus. We suggest that KA injection preferentially kills CA3b and CA3c neurons and that the change in sharp wave polarity reflects activity of surviving cells in the remaining CA3a area.
Another new pattern was found to occur in the CA1 area after status epilepticus. As a ripple oscillations, it had maximum amplitude in the CA pyramidal layer. Normal ripple oscillations reflect IPSPs on the somata of CA1 pyramidal layer (Ylinen et al., 1995), however it consisted of BPS. The frequencies of intraburst spikes varied between 80 Hz and 600 Hz.
In the dentate gyrus, gamma oscillations recorded in normal rats remained after status epilepticus. In addition a new pattern occurred, which as in the CA1 area, consisted of BPS but with maximum amplitude in the hilus. Intraburst spike frequency also varied between 80 Hz and 600 Hz. Some population spikes were superimposed on positive waves similar to the population spikes evoked by perforant path electrical stimulation (Lomo, 1971). Others, however, were superimposed on negative waves. In both cases the maximum amplitudes of waves and spikes were in the hilus of the dentate gyrus. It is possible that positive waves and spikes indicate that the electrographic event is generated in the recorded areas while negative waves indicate that the recording electrode is located some distance from the area generating electrical activity.
Differences in voltage depth profiles of oscillations with maximum amplitude in the dentate gyrus and those with maximum amplitude in stratum pyramidale of the CA1 area indicate that oscillations recorded in these two areas are independently generated and are not the result of volume conduction from one of these areas. The shape of the spike bursts changed dramatically with increasing distance from the point of maximum appearance (see Fig. 3) creating an oscillating pattern resembling HFOs recorded in majority cases with single microelectrodes.
At the generation site, pHFOs look like BPS. In areas more remote from the generator, pHFOs look more like oscillations due to the change of their shape while propagating along dendrites and probably low-pass filtering effect of intervening brain tissue.
Finding that pHFos reflect BPS is in line with the hypothesis (Bragin et al., 2000) that during a latent period a network of pathologically interconnected neuron clusters (PIN clusters) is formed as a substrate for fast propagation of epileptiform activity across the brain. Long-term potentiation evoked as a result of BPS discharges generated by PIN clusters may be a mechanism for formation of the network of PIN clusters during epileptogenesis.
Acknowledgments: This work was supported by National Institutes of Health Grants NS-33310 and NS-02808. We thank Joyel Almajano, Josephine Ruidera, and Tony Fields for essential assistance in the experiments.