Epilepsia, JO(2): 127-137, 1999 Lippincott Williams & Wilkina, Inc., Philadelphia 0 International League Against Epilepsy

Laboratory Research

Hippocampal and Entorhinal Cortex High-Frequency Oscillations (100-500 Hz) in Human Epileptic Brain and in

Kainic Acid-Treated Rats with Chronic Seizures

*tAnatol Bragin, *?§Jerome Engel, Jr., *?$Charles L. Wilson, $§Itzhak Fried, and *$$Gary W. Mathern

*Reed Neurological Research Center, 7Department of Neurology, $Division of Neurosurgery, and $Brain Research Institute, UCLA School of Medicine, Los Angeles, California, U.S.A.

Summary: Purpose: Properties of oscillations with frequen- cies >I00 Hz were studied in kainic acid (-)-treated rats and compared with those recorded in normal and kindled rats as well as in patients with epilepsy to determine differences as- sociated with epilepsy.

Methods: Prolonged in vivo wideband recordings of electri- cal activity were made in hippocampus and entorhinal cortex (EC) of (a) normal rats, (b) kindled rats, (c) rats having chronic recurrent spontaneous seizures after intrahippocampal KA in- jections, and (d) patients with epilepsy undergoing depth elec- trode evaluation in preparation for surgical treatment.

Results: Intermittent oscillatory activity ranging from 100 to 200 Hz in frequency and 50-150 ms in duration was recorded in CAI and EC of all three animal groups, and in epileptic human hippocampus and EC. This activity had the same char- acteristics i n all groups, resembled previously observed “ripples” described by Buzsaki et al., and appeared to repre- sent field potentials of inhibitory postsynaptic potentials (IPSPs) on principal cells. Unexpectedly, higher frequency in-

termittent oscillatory activity ranging from 200 to 500 Hz and 10-100 ms in duration was encountered only in KA-treated rats and patients with epilepsy. These oscillations, termed fast ripples (FRs), were found only adjacent to the epileptogenic lesion in hippocampus, EC, and dentate gyrus, and appeared to represent field potential population spikes. Their local origin was indicated by correspondence with the negative phase of burst discharges of putative pyramidal cells.

Conclusions: The persistence of normal-appearing ripples in epileptic brain support the view that inhibitory processes are preserved. FRs appear to be field potentials reflecting hyper- synchronous bursting of excitatory neurons and provide an opportunity to study the role of this pathophysiologic phenom- enon in epilepsy and seizure initiation. Furthermore, if FR ac- tivity is unique to brain areas capable of generating spontane- ous seizures, its identification could be a powerful functional indicator of the epileptic region in patients evaluated for sur- gical treatment. Key Words: Kainic acid-Temporal lobe epi- lepsy-Ripples-Oscillations.

Ripples are 120-200-Hz field oscillations described in hippocampal (1-3) and parahippocampal areas (4) of normal rats. They are produced by inhibitory postsynap- tic potentials (IPSPs) occurring during high-frequency bursts of inhibitory interneurons, which converge on principal neurons (1,3), and they reflect a state of net- work synchronization that predisposes to normal hyper-

synchronous events found in the rat hippocampus, such as sharp waves (SPWs) (5) .

Epileptogenesis in human mesial temporal lobe epi- lepsy and several chronic experimental models, such as those produced by kindling and kainic acid (KA) admin- istration, is associated with neuronal loss and synaptic reorganization in the hippocampus with resultant strengthening of certain interneuronal connections and increased propensity for hypersynchronization (6-17).

Given this abnormal synaptic reorganization, the pur- pose Of this investigation was to determine first, whether ripple activity could be recorded in kindled or KA- treated rats, and second, whether it was present in the

human hippocampus. We hypothesized

Accepted August 12, 1998. Address correspondence and reprint requests to Dr. .I. Engel, Jr. at

Reed Neurological Research Center, Department of Neurology, 710 Westwood Plaza, UCLA School of Medicine, Los Angeles, CA 90095- 1769, U.S.A.

Permanent address of Dr. Bragin: Institute of Experimental and Theoretical Biophysics, Puschino, Russia. that if ripple activity were present, epilepsy-related hy-

127

128 A. BRAGIN ET AL.

persynchronization might be reflected as an alteration in the characteristics of the ripples recorded in vivo from epileptogenic brain areas in kindled and KA-treated rats, as well as in the epileptic human hippocampus.

METHODS

Animal studies

Control and kindled rats Adult male (250-350 g) Sprague-Dawley rats (n = 3)

were anesthetized with a mixture of ketamine (100 mg/ kg), xylazine (5.2 mg/kg), and acepromazine (1.0 mg/ kg). Pairs of tungsten wires (60 p m in diameter) with 0.5-mm vertical tip separation were placed in the right angular bundle or bilaterally to stimulate the perforant path (PP) afferents to the hippocampus (AP = -7.0 mm from bregma, L = 3.5 mm from midline, and V = 3.0 mm).

Movable microelectrodes (tungsten, 60 pm in diam- eter, with 200-300 pm horizontal space and 500 pm vertical space separation) were implanted bilaterally above the dorsal hippocampi (AP = -3.0, L = 2.6), in the posterior hippocampi (AP = -5.2, L = 4.0) and in entorhinal cortex (EC) (AP = -8.0, L = 4.2) (18). After recovery (2-3 days), electrophysiologic recording from control rats began. The tips of the movable electrodes were gradually lowered into the hippocampus or EC dur- ing PP stimulation, so that evoked field potentials could be used to guide the positioning of the microelectrodes. Two stainless steel watch screws driven into the bone above the cerebellum served as indifferent and ground electrodes.

Experiments on the control rats were carried out over a period of 2-3 days. The locations of the recording electrodes in the dentate gyrus (DG), CAI area of hip- pocampus, and EC were verified on the basis of the shape of the evoked potentials elicited by PP stimulation. The response of those areas usually consisted of a posi- tive population excitatory postsynaptic potentials (pop- EPSPs) with a superimposed negative population spike (popSP), of 3- to 8-ms latency for stratum granulosum of DG, a response with a popSP latency of 10-15 ms for stratum pyramidale of CA1, and 2- to 5-ms latency popSP without a popEPSP in EC. Measurements were made continuously through all stages of sleep and wake- fulness.

After completion of these control studies, PP kindling was begun in the same group of rats. Stimulation con- sisted of a 10-s train of 0.2-ms square pulses at 5-7 Hz and 300-800 pA, which was above the popSP threshold in the DG. The stimulation trains were administered once every 24 h for 15 days, usually evoking afterdischarges (ADS) 20-60 s long. If, on any day, the stimulation train did not evoke an AD, stimulation was repeated again after 30-60 min with increased intensity of current until

an AD occurred. The stimulation current never exceeded 1 .O mA. To exclude the short-term influence of electrical stimulation, the rate and frequency of ripples were mea- sured before each kindling stimulation. Usually this was done every morning between 9 and 11 a.m.

Kainic acid-treated rats Adult male (250-350 g) Sprague-Dawley rats were

given atropine (0.04 mg, i.m.), anesthetized with chloral hydrate (400 mg/kg, i.p.), and unilaterally injected with KA (0.4 mg/0.2 ml normal saline) in the right posterior hippocampus (5.6 mm posterior, 4 mm lateral, and 7 mm deep to bregma). Beginning 3-4 months after injection, rats were observed during repeated 16- to 24-h video monitoring periods for 1-2 weeks every 1-2 months to detect spontaneous behavioral seizures. Five rats that showed recurrent spontaneous seizures were chosen for experiments.

A week before electrode implantation, each rat was again monitored for spontaneous seizures. After comple- tion of preliminary video monitoring, rats were anesthe- tized with a mixture of ketamine (1 00 mg/kg), xylazine (5.2 mg/kg), and acepromazine (1 .O mg/kg). The mov- able recording and stimulating electrodes were implanted bilaterally at the same coordinates as in control and kindled rats. The same electrophysiologic criteria were used to identify the position of the recording electrodes. In two rats, tungsten microelectrodes were implanted bi- laterally in EC (AP = -8.0; L = 5.0: V = 5-6), dorsal (AP = -3.5: L = 2.0; V = 3.0-4.5), and posterior hippocampi (AP = -6.0; L = 5.0; V = 5.5-6.0), and piriform cortex (AP = +2.0: L = 5.0; V = 9.0-9.5) and fixed on the scalp with dental cement.

Data acquisition Five four-channel MOSFET input operational ampli-

fiers, mounted in the cable connector, served to eliminate cable movement artifacts (19). Physiologic data were re- corded wideband (0.1 Hz to 1 kHz for field potentials or 1.0 Hz to 5.0 kHz for units and field potentials) and sampled to 10 kHz/channel (16 channels) with 12-bit precision on a Pentium PC using RC-Electronics soft- ware. The data were stored on JAZ cartridges or DAT. EEG was continuously recorded by using a Biological Monitoring Systems Inc. (BMSI-4000; Nicolet, Madi- son, WI, U.S.A.) telemetry system with video monitoring 16-24 h a day.

Histologic procedures At the end of the electrophysiologic experiments, the

rats were deeply anesthetized and perfused with glutar- aldehyde before Nissl and neo-Timm’ s staining to verify electrode placements and evaluate mossy fiber reorgani- zation in the dentate gyrus (1 3,20). Estimation of inten- sity of mossy fiber sprouting was done by using the graduated score suggested by Tauck and Nadler (21) and

Epilrpsiu. V d . 40, No. 2, 1999

FAST RIPPLE ACTIVITY IN EPILEPSY 129

also by measurement of the gray value in outer and inner molecular layers in the dentate gyrus of ipsilateral and contralateral hippocampi (22).

Human studies Patients (n = 9) requiring depth electrode investiga-

tion to verify the location of a temporal lobe epilepto- genic region before surgical resection (23) gave their informed consent for participation in these studies under the approval of the UCLA Human Subjects Protection Committee. Electrodes, stereotactically implanted under general anesthesia by using computerized magnetic reso- nance imaging (MRI) guidance, consisted of flexible polyurethane tubing with seven platinum contacts, through which a bundle of nine platinum-iridium mi- crowires (40 km in diameter) was inserted (24). Each clinical electrode containing microelectrodes was se- cured within a screw guide placed at the calculated co- ordinate in the temporal bone. The seven clinical con- tacts were led to one connector for seizure monitoring, and the nine microwires were led to a separate connector for experimental monitoring. Typically five to seven depth electrodes were inserted into each hemisphere with four to five directed to mesial temporal structures, in- cluding EC, amygdala, anterior, middle, and posterior hippocampus, or parahippocampal gyrus. The area of EC targeted was located between the amygdala and the an- terior hippocampus, medial and inferior to the hippocam- pus. The middle hippocampal electrodes were located at the level of the hippocampus lateral and inferior to the lateral geniculate nucleus, and the posterior hippocampus and parahippocampal gyms electrodes were placed -1 .O cm posterior to the middle hippocampus [see Behnke et al. (24) for in situ MRI scans of electrode sites]. The patients were observed postoperatively for 48 h in the neurosurgery intensive care unit before transfer to the neurology telemetry unit. Exact electrode positions were verified several days later with an MRI scan of the pa- tient’s brain with the electrodes in place. Seizures and interictal spike activity were recorded by using Biologi- cal Monitoring Systems Inc. BMSI-400 and Telefactor (Philadelphia, PA, U.S.A.) monitoring and seizure- detection systems. Local field potentials and unit activity were recorded wideband (0.1 Hz to 1 kHz for field po- tentials or 1.0 Hz to 5.0 kHz for units and field poten- tials) and sampled at 10 kHz/channel (16 channels) with 12-bit precision, simultaneously on a Pentium PC using RC-Electronics software.

For those patients in whom an epileptic focus was localized, anteromesial temporal lobe resection was per- formed (25,26). Hippocampal specimens were collected and stained by using the neo-Timm’s method (20). Es- timation of mossy fiber sprouting was made as described in the animal histologic procedure section.

Data analysis Analysis of results was carried out off-line. The re-

corded data were digitally filtered at 120 dB/octave to select the frequency of interest: for unit activity (250 Hz-5.0 kHz), and various EEG bands: 50-200 Hz for ripples and 200-800 Hz for fast ripples. Putative single units with a signal-to-noise ratio of a 3 times background were detected by amplitude discrimination. They were considered single if their autocorrelograms showed a re- fractory period: an absence of spikes within > 1 ms from the zero point. Unit activity that did not show a refractory period was considered multiunit activity. No more than two units were taken for analysis from each microelec- trode. Field potential and unit activity cross-correlations were carried out by using RC Electronics software. The frequency and amplitude of sharp waves, interictal spikes, ripples, and fast ripples were assessed by aver- aging 10 of each during each daily recording session over 3-4 weeks of experiments. Ripples and fast ripples were filtered before averaging. From these daily mea- sures, the overall mean amplitude, duration, and standard deviation of each waveform as recorded for the entire duration of the study was calculated by using the statis- tics software “GB-STAT” (Dynamic Microsystems, Inc.). Statistical significance of differences in the fre- quency and duration of ripples and fast ripples were de- termined with paired t tests.

RESULTS

Animal Studies Data were analyzed from three normal rats, the same

three rats after kindling, from five rats with spontaneous seizures 6-12 months after KA injection into the right hippocampus and from nine patients with epilepsy. The results are summarized in Tables 1 and 2.

Control rats In normal rats during immobility and slow-wave sleep,

ripple activity was recorded from stratum pyramidale of the CAI area and from the middle layers of EC. Ripple amplitude varied between 100 and 300 FV (mean, 210 FV), and the rate of occurrence of ripple oscillations ranged from 0.1 to 36/min (mean, 6/min). Ripple dura- tion ranged between 50 and 150 ms, and the mean fre- quency of ripples in normal rats was 148 f 22 Hz, within the frequency band of 130-180 Hz (Table 1; Here and below “fn” indicates standard deviation, SD). Ripples could occur simultaneously in both hemispheres or inde- pendently in either hemisphere. Recording simulta- neously from two electrodes with tips placed in stratum pyramidale and stratum radiatum, 56 f 13% of ripples occurred during SPWs of negative polarity recorded in stratum radiatum and tended to occur in the middle of these transients (Fig. 1A). SPW amplitude ranged from 0.5 to 2.5 mV (mean, 1.1 mV).

Epilepsia, Vol. 40, No. 2, 1999

130 A. BRAGIN ET AL.

TABLE 1. Comparison of characteristics of ripples and fast ripples

Control rats Kindled rats KA rats Human (n = 3) (n = 3) (n = 5) (n = 6)

Ripples Behavioral state

Location Mean frequency (Hz)

Mean duration (ms)

Mean rate (per min)

Total number of ripples for each mean

Behavioral state

Location Mean frequency (Hz)

Mean duration (ms)

Mean rate (per min)

Total number of FRs

(range)

(range)

(range)

FR

(range)

(range)

(range)

for each mean

Sleep CA1, EC

bilateral 148

(1 30-200) 80

(50-1 50) 6

(0.1-36)

210

NA

Sleep CA1, EC

bilateral 164

(1 50-200) 85

(30- 160) 10

(3-60)

415

NA

~~

Sleep CAI, EC

bilateral 161

( 120-200) 73

(30- 140) 8

(0.1-30)

1,110

Sleep CAI,DG,

EC, unilateral 357"

(250-500) 2Sh

(1 0-65) 1.6

(1-5)

1,300

~~

Sleep Hip, EC

bilateral 108

60 (50-120)

5 (0.1-20)

180

Sleep Hip, EC unilateral 364"

(250-540) 23'

( 1040) 2.1

(0.5-6)

150

(80-160)

CAI, CAI area of hippocampus; DG, dentate gyrus; EC, entorhinal cortex; FR, fast ripples; KA, kainic acid. 6 p < 0.001. a p < 0.0001.

In the hilus of DG, 40- to 100-Hz gamma activity and dentate spikes were recorded as described in previously published studies (27), but no ripples were observed in this region.

Kindled rats At the end of the 15-day kindling period we observed

that ADS were usually accompanied by stage 3 or stage 4 behavioral seizures [criteria of Racine (28)]. The rela- tion of ripples to sleep state, ripple amplitude, rate of occurrence, duration, and mean frequency in kindled rats were not significantly different from those in naive rats (Table 1).

However, kindling did cause an increase in SPW am- plitude recorded in the stratum radiatum of the CAI area, as previously reported (29,30). In kindled animals, the amplitude of SPWs was higher than that found in naive rats, and SPWs with amplitudes >3.0 mV were consid- ered to be interictal spikes (IISs), because in the earlier

TABLE 2. Correlation of neurons with ripples and FRs

Mean No. of frequency

neurons (impk)

1 20 2 12 3 1.3 4 3.0 5 0.5

~

SWS Paradoxical Correlation bursts sleep with ripples

+ Theta Yes + Theta Yes + No theta No + No theta No + No theta No

~ ~~

Correlation with FRs

No No Yes Yes Yes

FRs, fast ripples; SWS, slow-wave sleep.

Epilepsia, Vol. 40, No. 2, 1999

studies, SPWs with amplitudes >2.5 mV were rarely found in control rats. In other brain areas, any sharp events with an amplitude double the mean amplitude of the EEG were determined to be 11%. 11% began to occur after the fifth to eighth kindling stimulus and were ob- served in all recorded areas including CA1, DG, and EC on both sides.

When recording simultaneously both from pyramidal

A B

100ms I SPW IIS

FIG. 1. Local field potentials and unit activity in the dorsal CAI area of the rat's hippocampus: simultaneous recording with three microelectrodes located in the upper (pl), lower (p2) parts of the pyramidal layer, and in the stratum radiatum (r). The scheme on the left represents the presumed position of the electrodes. The location of the electrodes was determined on the basis of the shape of the averaged evoked potentials to perforant path stimu- lation [see Bragin et al. (27) for details]. A: Normal rat. Ripple (dashed box) and unit discharges of the presumed interneurons coincide with the sharp wave in CAI area. B: Kindled rat. With the same positions of electrodes, the ripple (dashed box) and unit discharges now precede the interictal spike after completion of the kindling procedure.

FAST RIPPLE ACTIVITY IN EPILEPSY 131

and radiatum layers after completion of kindling, 82 ~ t_

3% of ripples preceded IIS, as shown in Fig. 1B. In the EC, such a relation between ripples and IIS was not obvious. Three neurons having action potentials with spontaneous firing rates >10 Hz were found to discharge at high frequency on the negative wave of the ripples (Fig. IB).

The amplitude and frequency of gamma oscillations in the hilus of the DG also did not change significantly after completion of kindling, and no ripples occurred in this area.

Kuinic acid-treated rats In rats that displayed chronic recurrent seizures 4-9

months after unilateral intrahippocampal injection of KA, we found 120-200 Hz (mean, 161 Hz) ripples bi- laterally in the CA 1 area of dorsal and ventral hippocam- pi and in the EC. The parameters of these ripples did not differ significantly from those in the control and kindled rats (Table 1). 11% were recorded in all KA rats. They were found in both the lesioned hippocampus and con- tralateral hippocampus, as well as in recorded areas out- side of the hippocampus. When electrical activity was recorded simultaneously from the pyramidal and radia- tum layers of CA1 area, ripple activity did not have a clear relation with IISs.

In addition to this normal-appearing ripple activity, a striking new pattern that we termed fast ripple (FR) ac- tivity was found in these rats. This high-frequency os- cillatory activity ranged in frequency from 250 to 500 Hz (mean, 357 Hz) and, like ripples, was observed primarily during immobility and slow-wave sleep. Unlike ripples and IIS, histologic verification of electrode positions showed that FRs were recorded only in areas adjacent to the lesion in CAI (Fig. 2A), and in DG and EC ipsilateral to the lesioned hippocampus (Fig. 3A). Their amplitude varied between 200 FV and 1.5 mV (mean, 720 pV). FRs could occur alone or could be superimposed on 11%.

A

FIG. 2. A: Spontaneous fast ripples (FRs) in the rat’s right posterior CA1 (RpCA1) and right entorhinal cortex (REC) 4 months after KA injection in the right posterior hippocampus. B: The unit activity of presumed pyra- midal neurons (bottom) discharge in a phase-locked fashion with the negative wave of the FR. Three su- perimposed events are shown. This neuron revealed rare (fewer than one per 10 s) burst discharges with 1.5- to 2.0-ms interspike intervals. The FRs were re- corded with 100- to 800-Hz bandpass filter, unit activity was recorded with 0.5- to 5.0-kHz bandpass filter. LEC and LpCAl correspond to left entorhinal cortex and left posterior CA1 areas.

However, we found no clear relation between FRs and ripples or FRs and 11%.

FRs were not observed in the contralateral hemisphere at symmetric points of the CA1 area, DG and EC, nor in any ipsilateral or contralateral areas of the dorsal hippo- campus distant from the lesion. FRs with frequencies similar to those occurring spontaneously could be evoked in ipsilateral DG by orthodromic stimulation of PP or in ipsilateral EC by antidsomic stimulation of PP on the side of the epileptogenic KA lesion (Fig. 3B and C). These responses showed phase reversal when record- ing from electrodes that crossed the granule layer of the DG.

Because previous studies established the relation be- tween unit activity and ripples in normal and kindled rats, as shown in Fig. 1A and B (see refs. 1, 3, 30), we quantitatively analyzed data only for the KA animals. Among 32 neurons recorded in the CA1 area of hippo- campus in KA rats, we found two that showed theta modulation during exploratory activity or paradoxic sleep. The mean frequency of discharge of one neuron was 121s and the other was 201s. These two neurons showed correlation of their discharges with normal ripple activity similar to that shown in Fig. 1B for kindled rats, but not with FRs. Their action potentials occurred on the negative wave of the ripples.

All other recorded neurons displayed low firing rates (0.5-6.0/s), and irregularly discharged single or complex spikes. Three of these neurons fired bursts synchronously with FRs, but not with ripples (Table 2). Among these neurons, two revealed a complex spike discharge with amplitude decrement, whereas the third fired in bursts with no visible spike amplitude decrement. The action potentials of the neurons recorded during FRs also oc- curred on the negative wave of the FRs (Fig. 2B).

The confirmed location of recording and stimulating electrodes in target areas were in agreement with elec-

B

REC

LpCAi

LEC 40 msec

Epilepsia, Vol. 40, No. 2, 1999

132 A. BRAGIN ET AL.

A

RdDG __w_c_c *

LpDG

LEC 20rnsec

FIG. 3. Fast ripples (FRs) in the chronic epileptic rat brain. A: Spontaneous FR recorded in the rat‘s dentate gyrus (DG) and en- torhinal cortex (EC) 8 months after unilateral intrahippocampal kainic acid (KA) injection. The record- ings were performed in the right dorsal hippocampus and at sym- metric points in the right and lefl EC and ventral hippocampi. No- tice that 300-Hz FR occurs in the right ventral DG and ipsilateral EC. B: FR in response to perfo- rant path electrical stimulation. The same rat and the same posi- tion of the recording sites from the

I 106RpDG top to the bottom: DG of the dorsal

C

----.- hippocampus (RdDG), dentate gyrus of the right posterior hippo- campus (RpDG), right EC (REC), 2 0 c

dentate gyrus of the left posterior hippocampus (LpDG), left EC (LEC). Single electrical pulses (average of five) evoke population EPSPs with superimposed population spikes followed by oscillation at -300 Hz in the right ventral DG. In the REC, the evoked response consists of a negative-positive wave with a superimposed 250- to 300-Hz small-amplitude oscillation. There is no high-frequency oscillatory activity in the DG of the dorsal hippocampus, although amplitude of the evoked response is higher than in RpDG, and recordings are from symmetric points of the contralateral hemisphere.

trophysiologic results. Timm’s stain showed prominent sprouting of mossy fibers into the inner molecular layer of the ipsilateral dorsal and ventral dentate gyms in all rats receiving KA injection. Mild sprouting also was found in the contralateral dorsal and ventral dentate gy- N S .

Human studies Spontaneous ripples were observed in six of nine pa-

tients with temporal lobe epilepsy. They occurred bilat- erally in EC in all six cases (Fig. 4A), and in the hippo- campus in four cases. Their frequency varied from 80 to 160 Hz (mean, 120 -+ 57), and other parameters were similar to ripples in the rat (see Table 1). They were only observed during slow-wave sleep or quiet rest with eyes closed. Neurons that fired in bursts during spontaneous activity discharged predominantly on the negative phase of ripples (Fig. 4B). This was determined not only on the basis of visual observation, but also by averaging 50 sweeps of ripple activity and carrying out a cross- correlation with unit activity recorded from the same electrode tips.

In addition, spontaneous FR field oscillations (300- 500 Hz) were observed in EC and hippocampus in five of the six patients showing ripples. In four patients, these occurred only in areas identified as epileptogenic zones on the basis of presurgical evaluation (Fig. 5). In the remaining patient, FRs occurred only contralateral to the epileptogenic zone, although ripples were recorded bi- laterally. No clear relation between FRs and either ripples or IISs could be demonstrated. In areas of EC where FRs appeared spontaneously, they also could be evoked in the superficial layers by electrical stimulation

of the deep layers (not shown). Unlike human IISs, which are often synchronous over a large area, the field- potential distribution of FRs was quite localized.

Three of six patients exhibiting FRs and one of three patients without these oscillations underwent anterome- sial temporal lobe resection. In these four cases, histo- logic analysis showed hippocampal sclerosis and sprout- ing of mossy fibers in the DG.

DISCUSSION

There are two major findings in this study. First, ripple activity similar to that previously described in the normal rat (1) was found bilaterally in EC and hippocampus of control, kindled, and KA-treated rats, as well as in the hippocampus and EC of patients with epilepsy. Second, high-frequency ripple activity, not previously described, was recorded during the interictal state from areas adja- cent to lesions in the hippocampus of KA-treated rats with recurrent spontaneous seizures, and in the epilepto- genic zones of patients with epilepsy. Whereas ripples are well known in rats (1,2,4), their existence in the human brain was not demonstrated until now, nor has the oscillatory activity we describe as FRs in this report been previously observed or recognized as a spontaneous in- terictal event in the epileptogenic hippocampus and EC of rats and patients with spontaneous epileptic seizures.

High-frequency EEG activity has been described in patients with epilepsy in several reports (31-37), but it occurred chiefly during the ictal period and was of lower frequency than the oscillations reported here. The brief, intermittent interictal events observed in our experiments

Epilepsia, Vol. 40, No. 2, 1999

FAST RIPPLE ACTIVITY IN EPILEPSY 133

FIG. 4. A: Ripple activity in the human entorhinal cortex (EC). Recordings were carried out with 40-pm platinum-iridium microwires implanted within the mesial temporal lobe at the tips of clinical EEG electrodes. Ripples simultaneously oc- curred in the right and the left ECs, but not in the left and right hippocampi, although 80- to 100-ms sharp waves occurred there at the same time. B: Cross-correlation be- tween ripples and unit activity. Ba, an ex- ample of the burst of action potentials of a presumed interneuron. Bb, average of ripples (n = 50) filtered off-line with the bandpass filter 50-500 Hz. The same ripple peak was used for averaging, and as a zero point for generation of the cross- correllograms. Bc and Bd are cross- correllograms of unit discharges of two presumed interneurons (un. 1 and un. 2) selected by different levels of the window discriminator during recordings from the same microelectrode.

A

LEC

LmHip 4 are clearly different from the more continuous and ictal- related gamma activity noted in previous studies.

Ripples The ripple activity we observed corresponds well to

normal ripples reported in the literature (1-4). Our find- ings that ripples correlate with the firing of presumed interneurons agrees with the suggestion of Buzshki et al. (1) that ripple frequency depends on the interspike inter- val of bursts of interneuron action potentials (3).

Ripples recorded in kindled and KA rats were the same as those in control rats with respect to pattern, rate

A B ' a . I

I

- 20 ms

FIG. 5. A: Ripples and fast ripples (FRs) recorded in a patient with bundles of microwires implanted in the hippocampus and entorhinal cortex (EC). B: Expanded part of the dashed rect- angle. Ripples (125 Hz) preceded an interictal spike. Both the ripples (arrow) and the interictal spike first occurred in the record- ing point REC2 and spread to other brain areas. However, FR (double-headed arrow) occurs only in REC2. REC and LEC, right and left EC; LmHip, left middle hippocampus.

B a un.1

20Op"l

301 w b - 1

-32

of occurrence, behavioral state, location, and bilaterality. However, in contrast to control rats in which ripples tended to occur in the middle of a sharp wave, ripples in kindled rats preceded interictal spikes, appearing to drive the 11s discharge. The amplitude of the 11s field potential is so large, however, it is difficult to determine whether the ripple overlaps the peak of the spike. This suggests that the ripple oscillation has a limited field-potential distribution, whereas the much large amplitude spike component of the 11s represents the contribution of a greater population of synchronously discharging cells. In addition, the occurrence of ripples with these EEG tran- sients increased from 56% for normal SPW in control rats to 82% for 11s in kindled rats. The relation between ripples and 11s was not so clearly defined in KA-treated rats, and therefore the possible role of ripples in the generation of 11% requires additional investigation. However, the presence of ripples with normal character- istics in these two models of epilepsy supports the view that inhibitory mechanisms are preserved (8,11,14).

In the epileptic human temporal lobe, the mean fre- quency of ripples was somewhat lower than that in rats, consistent with findings that some other types of human EEG activity are also slower than in rat (38,39), but there were numerous similarities in ripple activity between hu- mans and rats. As in rats, human ripples occurred during slow-wave sleep, or quiet rest with eyes closed, and were not observed during rapid eye movement (REM) sleep or the awake state. In addition, we recorded neurons that discharged with bursts of action potentials and correlated with the negative phase of ripples. These results suggest that ripple activity in humans has a mechanism of gen- eration similar to that in rats, in which interneuron firing also has been shown to occur on the negative phase of ripple activity (3) . Our results also imply preservation of

Epilepsia, Vol. 40, No. 2, 1999

I34 A. BRAGIN ET AL.

functional inhibitory mechanisms in the human epileptic hippocampus and EC during the interictal state.

High-frequency ripples We found no field oscillations <200 Hz in normal or

kindled rats in the areas recorded in this study. However, in KA-treated rats with chronic recurrent spontaneous seizures, FR field oscillations were present in the CA1 and EC areas adjacent to the lesion. FRs occurred during the same behavioral states as ripples, but showed no consistent temporal relation to ripples or IISs. In addi- tion, FRs were observed in the DG adjacent to the lesion, although in normal rats, the frequency of DG oscillatory activity is never >60-100 Hz (27,40). The amplitude of FRs recorded in this study was as high as 1.5 mV, sub- stantially higher than the previously reported maximal ripple amplitude (3,4), but the significance of this differ- ence awaits detailed comparisons of the depth profiles of ripples and FRs. It is important to note that FRs were found only in areas of structural damage adjacent to the epileptogenic KA lesion and not in distant areas of ipsi- lateral hippocampus, or in symmetric points of the con- tralateral hemisphere, both of which also showed mossy fiber sprouting. Given that kindled animals did not show FRs, it remains to be determined whether the association of FRs and spontaneous seizures in areas near cell loss indicates that FRs reflect epileptogenic mechanisms, or whether the FR is merely a correlate of cell damage independent of epilepsy. Evidence that FRs are involved in the transition to spontaneous seizures, however, would tend to favor a role in epileptogenesis (Bragin et al., unpublished data).

FRs in the human epileptic brain occurred during the same behavioral states as ripples and FRs in rats, and had a similar anatomic distribution (i.e., within hippocampus and EC). Human FRs showed remarkable similarities to KA rat FRs in frequency, duration, and rate. Although they clearly occurred in hippocampus and EC, it was not possible to determine their precise location in the human hippocampus because of the spatial limitation of MRI localization of electrode tips. Finally, they were similar to KA rat FRs in being found in the vicinity of the lesion or epileptogenic region, with the exception of one pa- tient, who seemed FRs contralaterally. It is not uncom- mon, however, for patients with temporal lobe epilepsy to have occasional seizures originating contralaterally as well as in the area of primary seizure onset (40,41). Therefore, it is possible that the FR is limited only to brain tissue capable of generating spontaneous epileptic seizures.

Putative mechanisms of FR generation On the basis of the extracellular single unit and field-

potential recordings obtained in our study, two hypoth- eses for FR generation were considered. First, similar to the mechanism that was previously described for ripples

in the CA1 area of rats (1,3), FRs may reflect IPSPs on the somata of principal cells generated during simulta- neous depolarization of basket and pyramidal cells by the bursting activity of CA3 neurons. Although we found that cells that resembled the characteristic firing pattern of interneurons (action-potential frequencies of 2 15 Hz and theta modulation during paradoxic sleep or explor- atory activity; 42) occasionally fired during ripples, none of these putative interneurons fired with a frequency as high as FRs, nor did they fire in synchrony with FRs. The absence of any correlation of the firing of interneuron- like cells with FRs argues against the first hypothesis.

The second hypothesis for FR generation is that they reflect the rhythmic field potentials of population spikes generated by the hypersynchronous bursting of groups of principal neurons, and indeed, we were able to record from bursting neurons that showed a clear correlation with FRs. In our experiments, the three cells that fired in synchrony with FRs had baseline firing frequencies 4 Hz and displayed complex spike patterns with decremen- tal amplitudes, which, on the basis of accepted criteria (1,3,42,43), suggests that they were pyramidal cells not interneurons. Although the number of units showing such correlation with field oscillatory activity was small, the observation of interneuron-like activity correlating only with ripples, and pyramidal cell-like activity cor- relating with FR, supports the hypothesis that FR field oscillations reflect synchronous burst firing of principal neurons. Of the 32 cells recorded, only 9% fired in syn- chrony with FRs. Based on the low number of neurons (7%) that have previously been found to change their firing rate during subclinical epileptiform discharges in the human (44), the small number of cells correlating with FR activity is not surprising. The generation of population spikes by action-potential discharge from hip- pocampal neurons has been accepted since the work of Andersen et al. (45), who formulated the “population spike hypothesis” as follows: ‘‘The population spike is the potential produced at the site of the recording elec- trode by the summation of the individual action poten- tials of many neighboring, synchronously discharging cells (p. 209).”

Extracellular recordings from cells considered to be pyramidal cells in the normal rat hippocampus com- monly show a complex spike pattern of discharge during immobility and slow-wave sleep (46), which consists of bursts of three to seven action potentials with 1.5- to 6.0-ms interspike intervals (46,47). Intracellularly, these Na’ spikes are superimposed on depolarizing potentials, mediated by a voltage-dependent inward Ca2+ current, which is important for spike burst generation (47,48). A similar bursting pattern of discharge has been shown for neurons within the epileptic penicillin focus in the neo- cortex and hippocampus of animals (49-51) and in hu- mans with epilepsy (52,53). The mechanisms by which

Epilepsiu. Vol. 40, N o . 2, I999

FAST RIPPLE ACTIVITY IN EPILEPSY 135

these bursting neurons become synchronized is a key question to be answered.

There are few data from in vivo chronic models of epilepsy, on mechanisms of abnormal synchronization of neuronal discharges, but a number of studies used acute models and in vitro hippocampal slice studies of acute or chronic epilepsy. Blockade of y-aminobutyric acid (GABA)ergic inhibition causes synchronization of action potentials, presumably through recurrent or associational excitatory input (1 1,50,55). Hypersynchronous popula- tion spike field potential bursts with frequencies of 200- 300 Hz were described in dentate gyrus and CA1 area of the hippocampus in penicillin models of epilepsy (SO). They also were observed in rats after systemic KA in- jection, either spontaneously (56) or in response to single-pulse stimulation (56-59). Buckmaster and Dudek (60) described high-frequency bursts of population spikes in the dentate gyrus of rats showing dramatic losses of parvalbumin neurons after KA injection. It has been assumed in all of these experiments that such bursts primarily reflect discharges of principal cells.

Many factors may contribute to the synchronous burst- ing of neurons in limbic structures of KA-treated rats and in humans with temporal lobe epilepsy. Mossy fiber sprouting and other synaptic reorganization in rat and human (6,10,17,21,56,61-66), formation of gap junc- tions between local clusters of neurons [(57), also see refs. 11 and 671 and development of volume-conducted electrical currents (68) all have been postulated to con- tribute to hypersynchrony in epilepsy and may underlie FRs. Our preliminary results, however, do not support a principal role of mossy fiber sprouting in the generation of FRs. Whereas all rats with KA injection showed mossy fiber sprouting and FR activity near the lesion site, mossy fiber sprouting also occurred in ipsilateral dorsal DG and in some contralateral sites where no FR activity was observed. In human, all four patients for whom tissue was available for histologic analysis showed mossy fiber sprouting, whereas only three showed FR activity. More detailed correlation of quan- tified Timm’s reaction densities and evidence of sprout- ing in other areas of hippocampus with FR field activity is the subject of future study. Further experiments with intracellular recordings followed by morphologic identi- fication of recorded neurons with dye injection are re- quired to determine the membrane events inducing FRs.

In vivo studies of epilepsy with extracellular record- ings noted variable correlation between single unit ac- tivity and epileptiform field potentials. In the acute peni- cillin model, 90% of neurons were reported to fire syn- chronous with interictal EEG spikes (48). In contrast, human epileptic hippocampal recordings show as few as 10% of local cells may participate in seizure activity (44). The paucity of participating neurons may explain why single unit studies yielded conflicting data on syn-

chronous firing in the human epileptogenic region (69- 71). In the epileptic human neocortex or hippocampal formation, it was shown that, from a population of single units, only -20-25% show bursting patterns (53,72). In addition, cross-correlations between hippocampal cells of patients with temporal lobe epilepsy show that only -25% of these tend to fire in synchrony with one another (69). Thus the number of cells that show synchronous firing and simultaneous burst firing is only -556%. This means that a large total population of neurons must be recorded to determine the existence of presumably ab- normal synchronously bursting cells. If robust FR field potentials are generated by small populations of hyper- synchronously bursting cells, recording of FRs may pro- vide a means of studying the role of this pathophysi- ologic phenomenon in epileptogenesis and seizure initia- tion without recording large population of single units from large groups of subjects. The discovery of high- frequency field oscillations associated with areas of sei- zure generation also may provide a more powerful means of delineating the epileptogenic region for surgical re- section.

CONCLUSIONS

Ripples reflect the firing of inhibitory intemeurons: their unaltered presence in epileptic animal models and patients with epilepsy indicates that these inhibitory mechanisms are preserved in epilepsy.

FRs appear to represent excitatory synaptic activity generated by abnormal synchronous pyramidal cell burst firing. Such synchronous, high-frequency discharges may exist only in brain tissue capable of generating spontaneous seizures. Further studies are needed to de- termine the contribution of these events to seizure gen- esis and perhaps to the development of epilepsy itself (e.g., as a repetitive source of endogenous stimulation that eventually results in kindling).

Furthermore, if the FR is unique to tissue capable of generating spontaneous seizures, it may become a useful marker for identification of the epileptogenic region in patients who are surgical candidates.

Acknowledgment: This work was supported by NIH grants NS-33310, NS-02808, and KO8 NS 01603. We thank Dr. Gyorgy Buzshki for helpful discussions of the results; Elizabeth Vizentin for research assistance; and Joyel Almajano, Tony Fields, and Eric Behnke for essential technical support. Maria Melendez kindly assisted in manuscript preparation.

REFERENCES

1. Buzsaki G, Horvath Z, Urioste R, Hetke J, Wise K. High- frequency network oscillation in the hippocampus. Science 1992;

2. O’Keefe J, Nadel L. The hippocampus as a cognitive map. Oxford:

3. Ylinen A, Bragin A, Nadazdy Z, et al. Sharp wave-associated

256:1025-7.

Clarendon Press, 1978.

Epilepsia, Vol. 40, No. 2, 1999

136 A. BRAGIN ET AL.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

high-frequency oscillation (200 Hz) in the intact hippocampus: network and intracellular mechanisms. J Neurosci 1995; 15:30-46. Chrobak JJ, Buzsiki G. High-frequency oscillations in the output networks of the hippocampal-entorhinal axis of the freely behaving rat. J Neurosci 1996:16:3056-66. Buzs6ki G. Hippocampal sharp waves: their origin and signifi- cance. Brain Res 1986;398:242-52. Ben-Ari Y, Tremblay E, Ottersen OP, Meldrum BS. The role of epileptic activity in hippocampal and remote cerebral lesions in- duced by kainic acid. Brain Res 1980;191:79-97. Ben-Ari Y. Limbic seizure and brain damage produced by kainic acid: mechanisms and relevance to human temporal lobe epilepsy. Neuroscience 1985;14:375-403. Sloviter RS. On the relationship between neuropathology and pathophysiology in the epileptic hippocampus of humans and ex- perimental animals. Hippocampus 1994;4:25&3. Corsellis JA, Bruton CJ. Neuropathology of status epilepticus in humans. Adv Neurol 1983;34: 129-39. Babb TL. Research of the anatomy and pathology of epileptic tissue. In: Liiders H, ed. Epilepsy surgery. New York Raven Press, 199 1 :7 19-27. Schwartzkroin PA. General introduction. In: Schwartzkroin PA, ed. Epilepsy: models, mechanisms, and concepts. Cambridge: Cambridge University Press, 1993:l-18. Wasterlain CG, Fujikawa DG, Penix L, Sankar R. Pathophysi- ological mechanisms of brain damage from status epilepticus. Epi- lepsia 1993;34:S37-53. Mathem GW, Babb TL, Pretorius JK, Melendez M, LCvesque MF. The pathophysiologic relationships between lesion pathology, in- tracranial ictal EEG onsets, and hippocampal neuron losses in tem- poral lobe epilepsy. Epilepsy Res 1995:2: 133-47. Engel J Jr. Excitation and inhibition in epilepsy. Can J Neurol Sci 1996;23: 167-73. Fujikawa DG. The temporal evolution of neuronal damage from pilocarpine-induced status epilepticus. Brain Res 1996;24: 1 1-22. Wuarin J-P, Dudek FE. Electrographic seizures and new current excitatory circuits in the dentate gyres of hippocampal slices from kainate-treated rats. J Neurosci 1996;16:4438-48. Sutula T. Reactive changes in epilepsy: cell death and axon sprout- ing induced by kindling. Epilepsy Res 1991; 10:62-70. Paxinos G, Watson C. The rat brain in stereotaxic coordinates. 3rd ed. San Diego: Academic Press, 1997. Buzsiki G, Bickford RG, Ryan LJ, et al. Multisite recordings of brain field potentials and unit activity in freely moving rats. J Neurosci Methods 1989;28:209-17. Babb TL, Kupfer WR, Pretorius JK, Crandal PH, LCvesque MF. Synaptic reorganization by mossy fibers in human epileptic fascia dentata. Neuroscience 1991;42:35 1-63. Tauck D, Nadler J. Evidence of functional mossy fiber sprouting in hippocampal formation of kainic acid-treated rats. J Neurosci 1985;5:101&22. Mathem GW, Bertram EH, Babb TL, et al. In contrast to kindled seizures, the frequency of spontaneous epilepsy in the limbic status model correlates with greater aberrant fascia dentata excitatory and inhibitory axon sprouting, and increased staining for N-methyl+- aspartate, AMPA and GABA, receptors. Neuroscience 1997;77: 1003-19. Engel J Jr. Current concepts: surgery for seizures. N Engl I Med 1996;334:647-52. Behnke EJ, Wilson CL, Fried I, Maidment NT, Engel J Jr. An intracranial probe for in vivo single neuron recording and micro- dialysis in the human brain. J Neurosci Methods (in press). Spencer DD, Spencer SS, Matson RH, et al. Access to the posterior medial temporal lobe structures in the surgical treatment of tem- poral lobe epilepsy. Neurosurgery 1984;15:667-71. Doyle WK, Spencer DD. Anterior temporal lobe resections. In: Engel J Jr, Pedley TA, eds. Epilepsy: a comprehensive textbook. Vol 2. New York: Lippincott-Raven, 1998: 1807-18. Bragin AG, Jando G, Nadazdy Z, Hetke J, Wise K, Buzsiki G. Gamma (40-100 Hz) oscillation in the hippocampus of the behav- ing rat. J Neurosci 1995;15:47-60. Racine RJ. Modification of seizure activity by electrical stimula-

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

40. 41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

tion. 11. Motor seizure. Electroencephalogr Clin Neurophysiol

Buzsiki G, Ponomareff LG, Bayardo F, Ruiz R, Gage FH. Neu- ronal activity in the subcortically denervated hippocampus: a chronic model for epilepsy. Neuroscience 1989:28:527-38. Buzs&i G, Hsu M, Slamka C, Gage FH, Horvath Z. Emergence and propagation of interictal spikes in the subcortically denervated hippocampus. Hippocampus 1991; 1: 163-80. Huang C-M, White LE Jr. High-frequency components in epilep- tiform EEG. J Neurosci Methods 1989;30: 197-201. Bany E, Sussman N, O’Connor MJ, Hamer R. Presurgical elec- trographic patterns and outcome from anterior temporal lobec- tomy. Arch Neurol 1992;49:21-7. So N, Gloor P, Quesney FJ, Jones-Gotman M. Depth electrode investigations in patients with epileptiform abnormalities. Arm Neurol 1989;25:423-31. Allen PJ, Fish DR, Smith SJM. Very high-frequency rhythmic activity during SEEG suppression in frontal lobe epilepsy. Elec- troencephalogr Clin Neurophysiol 1992;82: 155-9. Fisher WRS, Webber RP, Lesser S, Arroyo S, Uematsu S. High- frequency EEG activity at the start of seizures. J Clin Neurophysiol 1992;9:441-8. Quirk JA, Kendall B, Kingsley DPE, Boyd SG, Pitt MC. EEG features of cortical displasia in children. Neuropediatrics 1993;24:

Raymond AA, Fish DR. EEG features of focal malformations of cortical development. J Clin Neurophysiol 1996;13:495-506. Marescaux C, Micheletti G, Vergness M, Depaulis A, Rumbach L, Warter JM. A model of chronic spontaneous petit mal-like seizures in the rat: comparison with pentylenetetrazol-induced seizures. Epilepsia 1984;25:3263 1. Vadazs C, Carpi D, Jando G, et al. Genetic threshold hypothesis of neocortical spike-and-wave discharges in the rats: an animal model of petit ma1 epilepsy. Am J Med Genet 1995;60:55-63. Buzsiki G, Leung L, Vanderwolf CH. Cellular basis of hippocam- pal EEG in the behaving rat. Bruin Res Rev 1983;6:139-71. Engel J Jr. Seizure and epilepsy. Philadelphia: FA Davis, 1989. Engel J Jr. Introduction to temporal lobe epilepsy. Epilepsy Res 1996;26: 141-50. Fox SE, Rank J. Electrophysiological characteristics of hippocam- pal complex-spike cells and theta cells. Exp Brain Res 1981;41: 299-313. Olmstead CE, Best PJ, Mays LE. Neural activity in the dorsal hippocampus during paradoxical sleep, slow wave sleep and wak- ing. Brain Res 1973;60:381-91. Babb TL, Wilson CL, Isokawa-Akesson M. Firing pattern of hu- man limbic neurons during stereoencephalography (SEEG) and clinical temporal lobe seizures. Electroencephalogr Clin Neuro-

Andersen P, Bliss TVP, Skrede KK. Unit analysis of hippocampal population spikes. Exp. Brain Res 1971;13:209-21. Ranck JB Jr. Studies on single neurons in dorsal hippocampal formation and septum in unrestrained rats. I. Behavioral correlates and firing repertoires. Exp Neurol 1973;41:461-531. Kandel ER, Spencer WA. Electrophysiology of hippocampal neu- rons. 11. After-potentials and repetitive firing. J Neurophysiol

Wong RKS, Prince DA. Afterpotentials generation in hippocampal pyramidal cells. J Neurophysiol 198 1 ;45:86-97. Matsumoto H, Ajmone-Marsan C. Cortical cellular phenomena in experimental epilepsy: ictal manifestations. Exp Neurol I964;9: 305-26. Dichter M, Spencer WA. Penicilline-induced interictal discharges from the cat hippocampus. 11. Mechanisms underlying origin and restriction. J Neurophysiol 1969;32:663-87. Calvin WH, Ojemann GA, Ward AA. Human cortical neurons in epileptic foci: comparison of inter-ictal firing patterns to those in “epileptic” neurons in animals. Electroencephalogr Clin Neuro- physiol 1973;34337-51. Ward AA Jr. The epileptic neuron: chronic foci in animals and man. In: Jasper HH, Ward AA, Pope A, eds. Basic mechanisms oj the epilepsies. Boston: Little Brown, 1969:63-288. Wyler AR, Ojemann G, Ward AA Jr. Neurons in human epileptic

1972;32:28 1-94.

193-9.

physiol 1987;00467-82.

1961 ;24:243-59.

Epilqsia, Vol. 40, No. 2, 1999

FAST RIPPLE ACTIVITY IN EPILEPSY 137

cortex: correlation between unit and EEG activity. Ann Neurol 1982;l 1:301-8.

54. Rayport M. Single neuron studies in human epilepsy. In: Somjen GG, ed. Neurophysiology studied in man. Amsterdam: Excerpta Medica, 1972: 100-9.

55. Prince DA. Neurophysiology of epilepsy. Annu Rev Neurmwi 1978;1:395415.

56. Franck JE, Kunkel DD, Baskin DG, Schwartzkroin PA. Inhibition in kainate-lesioned hyperexcitable hippocampi: physiologic, auto- radiographic, and immunocytochemical observations. J Neurosci 1988;8:1991-2002.

57. MacVicar B, Dudek FE. Electronic coupling between granule cells of rat dentate gyrus: physiological and anatomical evidence. J Neu- rophysiol 1982;47:579-92.

58. Nakajima S, Franck JE, Bilkey D, Schwartzkroin PA. Local circuit synaptic interactions between CAI pyramidal cells and interneu- rons in the kainate-lesioned hyperexcitable hippocampus. Hippo- campus 199 I ; I :67-78.

59. Schweitzer JS, Patrylo PR, Dudek FE. Prolonged field bursts in the dentate gyrus: dependence on low calcium, high potassium, and nonsynaptic mechanisms. J Neurophysiol 1992;68:2016-25.

60. Buckmaster PS, Dudek FE. Neuron loss, granule cell axon reor- ganization, and functional changes in the dentate gyrus of epileptic kainate-treated rats. J Comp Neurol 1997;385:385404.

61. Davenport CJ, Brown WJ, Babb TL. Sprouting of GABAergic and mossy fibers axons in the dentate gyrus following intrahippocam- pal kainate in the rats. Exp Neurol 1990;109:180-90.

62. Cronin EP, Dudek FE. Chronic seizures and collateral sprouting of dentate mossy fibers after kainic acid treatment in rats. Brain Res 1988;474: 1814 .

63. Mathern GW, Cifuentes F, Leite JP, Pretorius JK, Babh TL. Hip-

pocampal EEG excitability and chronic spontaneous seizures are associated with aberrant synaptic reorganization in the rat intra- hippocampal kainate model. Electroencephulogr Clin Neuro- physiol 1993;87:326-39.

64. Perez Y, Morin F, Bealieu C, Lacaille J-C. Axonal sprouting of CA1 pyramidal cells in hyperexcitable hippocampal slices of kain- ate-treated rats. Eur J New-nsci 1996;8:73648.

65. de Lanerolle N, Kim JM, Robbins RJ, Spencer DD. Hippocampal interneurons loss and plasticity in human temporal lobe epilepsy. Brain Res 1989;495:387-95.

66. Houser CR. Morphological changes in the dentate gyrus in buman temporal lobe epilepsy. In: Ribak CE, Gall CM, Mody I, eds. The dentate g y m s and its role in seizures (Epilepsy Res Suppl 7 ) . Amsterdam: Elsevier, 1992:223-34.

67. Dichter MA, Ayala GF. Cellular mechanisms of epilepsy: a status report. Science 1987;237: 157-237.

68. Dudek FE, Snow RW, Taylor CP. Role of electrical interactions in synchronization of epileptiform bursts. Adv Neurol 1986;44:593- 617.

69. Colder BW, Wilson CL, Frysinger RC, Harper RM, Engel J Jr. Interspike intervals during interictal periods in human temporal lobe epilepsy. Bruin Res 1996;719:96-103.

70. Colder BW, Wilson CL, Frysinger RC, Chao LC, Harper RM, Engel J Jr. Neuronal synchrony in relation to burst discharge in epileptic human temporal lobes. J Neurophysiol 1996;75:2496- 508.

71. Isokawa-Akesson M, Wilson CL, Babb TL. lnhibition in synchro- nously firing hippocampal neurons. Epilepsy Res 1989;3:23647.

72. Colder BW, Frysinger RC, Wilson CL, Harper RM, Engel J Jr. Decreased neuronal burst discharge near site of seizure onset in epileptic human temporal lobes. Epilepsia 1996;37: 113-21.

Epilepsia, Vol. 40, No. 2, 1999