The initiation site for triggering epileptiform activity was investigated via optical imaging using voltage-sensitive dyes in the neocortical slice perfused with artificial cerebral spinal fluid containing nominally zero magnesium. The neocortical slices (400-μm thick) were harvested from Sprague-Dawley rats (P21–28). Optical imaging was made by using a high speed photodiode array. Spontaneous epileptiform activity emerged 20–40 min after the preparation was perfused with zero-magnesium solution. There was a good correspondence between electrical and optical signals (n = 46), although the details of the two recordings were somewhat different. The initiation sites were measured optically in 11 preparations. Among them, four were found to be located in superficial layers, two were found in middle layers, and five were found in deep layers. Repeated recordings revealed that these initiation sites were relatively stable; shifting of the initiation site was not observed. Therefore spontaneous epileptiform activity could be initiated in various cortical layers, from layer I to layer VI. The activation started from a small area <0.04 mm3 and spread smoothly from the initiation site to adjacent cortical areas, suggesting that the initiation site is very confined to one of the cortical layers. The initiation sites were distributed randomly in various cortical areas, and no higher probability was found in a special cortical region. Electrical stimulation delivered via a glass microelectrode filled with 2 M NaCl (2–5 MΩ) could reliably trigger epileptiform activity that had the same characteristics as the spontaneous activity. The cortical neurons activated directly by the stimulation were around the electrode’s tip and estimated to be within a 50-μm area, suggesting that only a few neurons were needed to form an initiation site. Because the timing for stimulation was arbitrary and the evoked events were initiated independent of discharges of neurons in any other layers, it is likely that the initiation site for epileptiform activity in various cortical layers is independent of the control of layer V pyramidal neurons. Together these finding suggest that the epileptiform focus is confined and can be formed in several (probably all) neocortical layers and in many cortical areas. The initiating neurons may be of different types because neuronal types in various cortical layers are different.
In a number of epilepsy models, epileptiform activity can be initiated from cortical structures (McNamara 1994; Traub et al. 1994), indicating that the initiation of epileptiform activity can be intrinsic to the cortex. Studying the process of epilepsy initiation has been attractive not only because the understanding of its mechanism would benefit more efficacious prevention and treatment of the disease but also because the study of neuronal interactions during this process would reveal important features of the functional neuronal organization in the cortex. Pyramidal neurons in layer V of the neocortex have been thought to be responsible for initiation because pyramidal neurons have the highest excitability in the neocortex (Connors 1984), they have little detectable inhibition (Chagnac-Amitai and Connors 1989b), and layer V alone is sufficient for generating epileptiform activity in a zero-Mg2+ epilepsy model (Silva et al. 1991). However, it recently has been reported that isolated superficial layers also can initiate epileptiform activity and may even dominate the initiation process in the intact cortex perfused with 10–20 μM bicuculline methiodide (Albowitz and Kuhnt 1995). Field potential recordings also suggested that epileptiform activity could start in layer II/III of the cortex (de Curtis et al. 1994). Because intrinsic firing properties of cortical neurons may underlie the initiation of epileptiform activity (Miles and Wong 1983;Prince 1967, 1969; Traub and Wong 1982), cortical neurons with repetitive firing properties might be capable of initiating epileptiform activity. This implies that various neurons from different layers may potentially become initiation cells for epileptiform activity if they possess repetitive and rhythmic firing properties. In fact, cortical neurons in both superficial and deep layers can be induced to fire spontaneously (Flint and Connors 1996), and therefore these neurons potentially should be able to initiate epileptiform activity. The location of initiation cells for triggering epileptiform activity in different cortical layers indicates the involvement of different neuron types in the formation of the epileptiform focus. To clarify in which layer the initiating cells are located, it is necessary to simultaneously monitor neuronal activity in all cortical layers during epileptiform activity.
Examination of the spatial and temporal characteristics of epileptiform activity in cortical slice preparations using conventional electrophysiological techniques requires an electrode array. The number of electrodes in such an array has been limited to just a few (Chagnac-Amitai and Connors 1989b) because of practical difficulties. Current-source density analysis has been attempted in a study of the origin of epileptiform activity, but the location of initiation site could not be determined accurately (de Curtis et al. 1994). The timing of epileptiform activity onset from the different regions is a good indicator for where the epileptiform activity starts but only can be compared when these regions are simultaneously measured. Because this transient signal tends to be in the millisecond domain, such measurements need high temporal resolution.
Optical recording has been applied to achieve the measurement of rapid changes in transmembrane potential, such as action potentials. The optical signals from neuronal processes stained with voltage-sensitive dyes are linear with membrane potential (Ross et al. 1977). If multiple neurons are recorded optically by a single detector, the optical signal represents the average of the change in transmembrane potential (Tsau et al. 1996). When multiple regions are simultaneously recorded by a photodiode array, the timing of activities in these regions can be compared and the initiation site for epileptiform activity thus can be determined. In this study, we applied high speed (1 frame per millisecond) optical imaging using voltage-sensitive dyes to measure epileptiform activity over a region of the cortex ∼4.5 × 1.5 mm2 in an attempt to directly record the initiation process. We have reported that dominant foci for initiating epileptiform activity emerge when the preparation is perfused with zero-Mg2+ artificial cerebral spinal fluid (ACSF) or normal ACSF containing bicuculline (20–50 μM) (Tsau et al. 1998). In this report, we image these stationary foci in neocortical tissue. We ask in which cortical layer these foci are located. We also use electrical stimulation via a microelectrode to trigger epileptiform activity in various cortical layers to visualize the initiation process.
The experiments were carried out on neocortical slices from Sprague-Dawley rats (P21–28) of either sex. The animals were decapitated under CO2-induced narcosis, following the guidelines set by the National Institutes of Health. The brains were removed and placed in normal ACSF consisting of (in mM) 132NaCl, 3 KCl, 2 CaCl2, 2 MgSO4, 1.25 NaH2PO4, 26 NaHCO3, and 10 dextrose bubbled with 95% O2-5% CO2 (pH = 7.4), and chilled in an ice bath. The preparations were 400-μm-thick coronal sections from the temporal cortex of Bregma −2 to −5 mm and were incubated for ≥1 h in normal ACSF at room temperature before staining. After staining the slice was submerged in a recording chamber under a Leitz Ortholux II microscope and perfused with (nominally) zero-Mg2+ ACSF at 28–31°C for electrical and optical recordings (Fig. 1). Because optical recordings are sensitive to mechanical vibrations, several steps were taken to reduce them: an air table (Newport) was used to reduce the floor vibration; the lamp house was sealed to eliminate the illumination fluctuation due to air flow around the light bulb; the perfusion was paused during the optical recording trials; and a cover slip was placed on top of the recording chamber to eliminate the fluctuations at the air-water interface.
Recording electrodes were made from borosilicate glass containing a capillary fiber (0.75 mm ID, 1.0 mm OD). Extracellular glass electrodes were filled with 1 M NaCl and had a resistance ranging from 2 to 5 MΩ. In most experiments, the recording electrodes were placed in layer II of the cortex. The tips of electrodes were inserted ∼100 μm into the slice. Recordings were made with the use of two preamplifiers (IE201, Warner Instrument, Hamden, CT). In the experiments where we recorded evoked epileptiform activity, one electrode was used as a stimulating electrode. Stimuli consisted of rectangular current pulses 2–5 μA (10–30 V) in amplitude and 100–200 μs in duration. The data from electrical recordings were stored on video tapes through a digital recorder (CRC VR-100) and then transferred to a PC for further analysis.
The neocortical slices were stained with the voltage-sensitive dye RH479 (first synthesized by R. Hildesheim and A. Grinvald and kindly provided by Dr. L. Loew, University of Connecticut, Farmington, CT, as JPW 1131) solution (0.05 mg/ml in ACSF) for 1–2 h. Transmitted light was used for optical recording. Light from a Tungsten filament lamp (12 V, 100 W) was passed through a 705 ± 60 nm filter (mean ± SD) and then to the preparation. The image was projected onto a 124 element photodiode array (Centronics, Newbury Park, CA) through a ×2.8 objective lens (Fig. 1). A low-power objective lens was used to maximize the recording area. Each pixel received light from a 0.3 × 0.3 mm2 area of the objective plane. The optical field was adjusted so that the epileptiform activity initiation site and propagation would be best recorded optically because only part of the cortical slice could be included in the recording field. Each photodiode signal was amplified in parallel by a set of 128 parallel amplifiers (purchased for the Department of Physiology, Yale University, New Haven, CT). The first stage of amplification was performed with a current to voltage converter circuit using a feedback resistor of 5 MΩ. The second stage amplified the voltage by an additional factor of 1,000 or 2,000, with high-pass filtering with a 100-ms time constant and four-pole Bessel analogue low-pass filtering of 300-Hz corner frequency. The signals from the 124-element photodiode array and two electrodes then were multiplexed and digitized at 1,000 frames per second with a DAP3200e/214 12-bit data-acquisition board installed in a Pentium PC and 2 MSXB expansion boards (Microstar Laboratories, Bellevue, WA) controlled by a program written in BASIC. Data were acquired into the buffer memory on the data-acquisition board and saved to a hard drive. Optical recordings were made without averaging. Four to 20 trials were recorded from each preparation. To capture spontaneous epileptiform activity, each trial had a duration of 8–16 s. The signal-to-noise ratio was >20, and the optical signal size (ΔI/I, change in intensity divided by resting light intensity) for epileptiform activity was ∼0.1–0.5%. Data analysis was performed with a Pentium PC running the Interactive Data Language (IDL, Research Systems) program NeuroPlex (OptImaging, LLC, Fairfield, CT). The optical signals were filtered digitally with 300-Hz low-pass (Butterworth filter) and 1.2-Hz high-pass (RC filter) and then normalized for the movie display. The contour mode displayed the areas of equal signal height as the same color. The spaces between diodes are interpolated; the display is not in discrete pixels. Variable display mode was used where the scale for each diode is independent and the timing information is displayed more accurately.
Spontaneous epileptiform activity initiation sites revealed by optical recordings
Spontaneous epileptiform activity emerged 20–40 min after the preparation was perfused with zero-Mg2+ ACSF. Epileptiform events occurred in episodic bursts, each burst starting with a large initial spike and followed by 7–10 Hz oscillations lasting 5–30 s (Flint and Connors 1996; Silva et al. 1991; also see Fig. 2). We have validated the optical recordings by comparing simultaneously recorded electrical and optical signals throughout all experiments (n = 46). These results indicate a very good correspondence between electrical and optical signals (Fig. 2). Both recordings show a large initial spike followed by prolonged 7- to 10-Hz oscillations. The timing of the initial spike and some peaks of the oscillation waves was very similar in the two traces. However, details of the two recordings were not the same probably because electrical recordings can be affected by current from distant areas, whereas optical recordings strictly reflect the neuronal membrane potential changes in the area the image of which is projected to the recording photodiode (Ross et al. 1977; Salzberg et al. 1973). For example, the electrical recording is very sensitive to a stimulus current delivered at a substantial distance, whereas the optical signal is not affected at all. Therefore it was anticipated that the components of field potential recording would differ somewhat from those of the optical recording. In this study, we were interested mainly in the timing and location of the initial spike in different cortical laminae to understand the initiation process of epileptiform activity. Because the timing of the initial spike in optical recordings correlated well with that of the local field potential recordings, we can use the photodiode array to locate the starting focus of the initial spike. In the following figures, we will only describe the characteristics of the initial spike of each epileptiform event. The secondary oscillations have different characteristics and have been described elsewhere (Wu et al. 1999).
Epileptiform activity could be recorded from the entire preparation when the preparation was perfused with zero-Mg2+ ACSF (n = 40), even in a small section (2 mm wide horizontally) of a cortical slice with all cortical layers intact. This was true of neocortical slices from randomly chosen regions including frontal, temporal, and occipital areas. This result indicates that initiation sites for generating epileptiform activity can be formed spontaneously regardless of the region of cortex. The initiation site in each stained cortical slice was identified optically and moved to an appropriate place in the imaging field. Because the resolution of optical imaging data is relatively poor compared with the cortical architecture due to the limited number of pixels, more than one cortical layer often was projected to a single detector, and distinctions between layers could not always be made based on the imaging data. For example, the image of six cortical layers was projected to five photodiodes in Fig.3, and so some photodiodes detect activity from more than one cortical layer. Nevertheless, it is clearly shown by traces from the photodiodes and by the consecutive image display in Fig. 3 that the epileptiform event started from a middle layer and propagated to superficial and deep layers as well as horizontally across the cortical slice. The onset of the epileptiform event started earliest in the trace recorded from the middle layer and was delayed (by ∼10 ms) in the adjacent detectors. The further the physical location was from the initiating site, the more delayed the onset (Fig. 3, top). This also is seen in the consecutive image display where a red spot representing activation of cortical neurons appeared in a middle layer and then expanded to the superficial and deep layers as well as in lateral directions (Fig. 3,bottom). These results indicated that epileptiform activity was initiated by an epileptiform focus located in layer III or IV in the parietal cortex. The location of such a focus was quite stable; repeated optical recordings (n = 10) from the same preparation revealed the same initiation site and propagation profile. Figure 4 is a second example from the same preparation. This is consistent with the finding that there often exists a dominant epileptiform focus (Tsau et al. 1998).
In some preparations, the initiation sites were found to be located in deep layers (probably layer V or VI; Fig.5, top). The epileptiform events started from deep layers and spread out to the adjacent areas. In other preparations, the initiation sites were found to be in superficial layers (probably layer I or II; Fig. 5, bottom). Optical recordings were repeated 4–10 times on each preparation. These initiation sites were stable, and shifting of the initiation site was not observed. Therefore it seemed that spontaneous epileptiform activity could be initiated in any of the cortical layers. All of the imaging data on the epileptiform focus showed that the activation started from a small confined area <0.3 × 0.3 × 0.4 mm, or 0.04 mm3, and spread smoothly from the initiation site to adjacent cortical areas.
The initiation site across the cortical layers varied from preparation to preparation. These sites were scattered over the slice and were observed in neocortical slices from occipital, parietal, temporal, and perirhinal areas in different preparations (n = 11). Among the 11 preparations, 5 had initiation sites in deep layers, 2 in middle layers, and 4 in superficial layers (Fig.6). These initiation sites seemed to be distributed randomly and not to have higher susceptibility in any particular cortical region. Although evidence has been presented that the temporal cortex may be more likely to originate epilepsy (e.g.,Wardas et al. 1990), our data suggest that there is no preference for a particular cortical area or an exclusive cortical layer in establishment of an initiation site for epileptiform activity when the cortex is hyperexcitable and capable of conducting traveling epileptiform events.
Continuous recordings of spontaneous firing of cortical neurons
Our optical imaging data have shown the dynamic process of epileptiform activity initiation. In all the experiments, no consistent firing of cortical neurons was optically observed before an epileptiform event was initiated. No consistent firing was observed optically even at the initiation site. Because each photodetector covers 0.04 mm3 of cortical tissue, significant firing of a single neuron may be masked by light noise from many other cortical cells. Another limitation of optical imaging is that each trial of recording is limited to a few seconds due to dye bleaching. We used a microelectrode in many locations in the slices to detect constant firing of neurons. This allowed us to record spikes from the neurons close to the recording electrode’s tip. In most experiments (68 of 72), no continuous firing was observed. Occasionally (4 experiments), continuous firing from a single neuron was observed before as well as during epileptiform events. However, none of the activity was correlated to the initiation of epileptiform events. These data indicate that spikes in a single cortical neuron may not be able to initiate epileptiform activity. Although the recording period can be as long as several hours, the major limitation of the electrical recording is that it is very difficult to accurately position the electrode at the initiation site of an epileptiform event.
Electrically evoked epileptiform events
We showed in the preceding text that spontaneous epileptiform events can be initiated in several cortical layers; it seems that cortical neurons in several layers have this capability. However, it is possible that an initiation site that appears in middle or superficial layers is actually initiated by layer V pyramidal neurons the discharges of which trigger the activation of an initiation site located in one of the other cortical layers. Optical recordings might not be able to detect a small initiating discharge from layer V pyramidal neurons. This implies that even though we have seen the activity starting from the superficial or middle layers, it actually might be triggered by undetectable activity in the deep layers, and neurons in superficial layers might not be capable of initiating epileptiform activity themselves. To test this possibility, a microelectrode was used to measure the ability of an electrical stimulus to initiate epileptiform events in different layers. A single electrical pulse (10 V, 0.1 ms) was delivered via an electrode (its tip diameter was <1 μm, usually ∼0.1 μm) with impedance of 2–5 MΩ, placed in one of the cortical layers. Assuming that the electrical impedance of fluid and brain tissue is much lower than that of the microelectrode and can be ignored, the stimulation currentI = V/R would be 2–5 μA at the electrode’s tip. The current density would be 2,500 times smaller at a distance of 50 μm from the tip on the assumption of current spread from a point source in a volume conductor. Therefore only the neurons close to the tip would be activated directly by the stimulus, and the number of neurons should be very limited. We found that epileptiform events were evoked reliably in all cortical layers by the electrode stimulation. The evoked epileptiform events had the same characteristics as those of spontaneous events and propagated vertically and horizontally. The initiation site depended only on the position of the stimulation electrode’s tip (e.g., Fig.7). The images, made at 2-ms intervals, show clearly that activation first appeared at the electrode tip with no delay and then spread smoothly to the neighboring areas. No jumping of activation from the electrode tip to deep layers or other cortical areas was observed in any of the experiments. Various areas of the cortex including frontal, occipital, parietal, temporal, perirhinal, and hippocampal cortex were tested; all the cortical structures were found to be capable of generating evoked epileptiform events (n = 36). Because the timing for stimulation was arbitrarily set, the evoked epileptiform events must have started independently of any spontaneous pyramidal neuron discharges.
Our optical imaging data directly revealed spontaneous initiation sites. These initiation sites were found to be stable in every preparation. In different preparations, the initiation sites were in different cortical layers where different types of neurons are located. Continuous spontaneous activity correlated with epileptiform events was not observed in either the focus or other regions between epileptiform events. We also used extracellular microelectrodes to monitor single unit activity during as well as between epileptiform events and did not observe consistently firing neurons in the neocortical slices. Therefore it is reasonable to assume that most neurons, except those responsible for initiation, are not continuously firing between epileptiform events. We previously have hypothesized that a dominant initiation focus may be organized by a dynamic process in a group of local neurons (Tsau et al. 1998); our current results suggest that this kind of process is not laminar-specific, and it can happen in various cortical layers.
Optical signals reflect the membrane potential change in all membranes stained with voltage-sensitive dyes. Because neurons contribute to the optical signal regardless of their physical size, a change in membrane potential of virtually all the cortical neurons could contribute to the optical signals. As we report in this study, the initiation site can be located in several cortical layers and so would be composed of neurons of different types because each cortical layer has distinctive cell types. Different types of neurons in various layers may become initiation neurons and trigger epileptiform activity.
Previous studies showed that elimination of layer V abolishes spontaneous activity in cortical slices (Silva et al. 1991). To explain our data of visualizing initiation foci in superficial layers, one may assume that an epileptiform event may be triggered by an initiation site located in layer V pyramidal neurons, but the activities in layer V are too small to be detected by our optical recording technique. We used the microstimulus experiment to reduce this possibility. When epileptiform events were evoked by electrical stimulation, the number of neurons activated directly by stimulation presumably was limited. In this experiment, we always saw the activity start from the simulation point instead of from layer V. This suggests that an epileptiform event starts at the layers where the stimulation electrode is placed. The current density of stimulation is high around the microelectrode’s tip, and the effective area is probably smaller than a distance of 50 μm. This volume of cortical tissue can contain only 100 neurons with a 10-μm diameter. It is therefore likely that the neurons directly activated by the stimulation electrode also start the epileptiform activity. However, our data do not rule out the possibility that layer V pyramidal neurons provide background activity for the neurons in other layers; certain patterns of the spontaneous activities in layer V may trigger other layers to start all-or-none epileptiform events. It is difficult for our current experimental setting to demonstrate that pyramidal neurons provide background activities. However, Demir et al. (1999) suggested similar activities, a “preepileptiform activity” that can trigger an all-or-none epileptiform event in other places. Cortical neurons in both superficial and deep layers in the neocortex can have repetitive discharge properties (Flint and Connors 1996). Neurons in isolated superficial layers of entorhinal cortex also can fire repetitively (Dickson and Alonso 1997). It long has been believed that neurons with repetitive discharge properties are capable of initiating epileptiform activity (Miles and Wong 1983; Prince 1967, 1969; Traub and Wong 1982). Our data further suggest that neurons in various layers may be capable of initiating epileptiform activity and that the initiation process may be less reliant on one specific type of neuron.
The initiation site revealed by our optical imaging data was composed of a confined volume of <0.04 mm3, indicating that the initiation of epileptiform events is a localized process and may be from local neuronal clusters with potentiated connections (Tsau et al. 1998). As this is the smallest volume a photodiode could detect in our experiments, the real epileptiform focus could be smaller. We hypothesize that a neuron pool with repetitive discharge properties in such a confined epileptiform focus may synchronously activate its adjacent neurons and thus initiate epileptiform activity. This is consistent with the stimulation experiment where a single-pulse stimulus generates synchronized activation of cortical neurons around the microelectrode tip and results in an all-or-none epileptiform event.
We thank Drs. L. Cohen and N. Hershkowitz for helpful comments, and A. Schaefer for technical assistance with the experiments and manuscript.
This work was supported by National Institute of Neurological Disorders and Stroke Grant NS-31425, Department of Defense Grant 17-93-V3018, and a grant from Epilepsy Foundation of America.
Address for reprint requests: J.-Y. Wu, Institute for Cognitive and Computational Sciences, WP 24A Research Building, Georgetown University Medical Center, 3970 Reservoir Rd., N.W., Washington, DC 20007. E-mail:
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 1999 The American Physiological Society