Theta waves, neural spikes and seizures can propagate by ephaptic coupling in vivo

Abstract

Electric field coupling has been shown to be responsible for non-synaptic neural activity propagation in hippocampal slices and cortical slices. Epileptiform and slow-wave sleep activity can propagate by electric field coupling without using synaptic connections at speeds of ~0.1 m/s in vitro. However, the characteristics of the events that can propagate using electric field coupling through a volume conductor in vivo have not been studied. Thus, we tested the hypothesis that various types of neural signals such as interictal spikes, theta waves and seizures could propagate in vivo across a transection in the hippocampus. We induced epileptiform activity in 4 rats under anesthesia by injecting 4-aminopyridine in the temporal region of the hippocampus, four recording electrodes were inserted along the longitudinal axis of the hippocampus. A transection was made between the electrodes to study the propagation of the neural activity. Although 54% of the interictal spikes could propagate through the cut, only those spikes with a high amplitude and short duration had a high probability to do so. 70% of seizure events could propagate through the cut but parameters distinguishing between propagating and non-propagating seizure events could not be identified. Theta activity was also observed to propagate at a mean speed of 0.16 ± 0.12 m/s in the characteristic range of propagation using electric field coupling through the transection. The electric field volume conduction mechanism was confirmed by showing that propagation was blocked by placing a dielectric layer within the cut. The speed of propagation was not affected by the transection thereby providing further evidence that various types of neural signals including activity in the theta range can propagate by electric field coupling in-vivo.

1. Introduction

Electric field coupling has been shown to be responsible for the propagation of neural activity in hippocampal tissue in vitro without involving any of the common modes of propagation such as synaptic transmission, gap junctions, or diffusion (Zhang et al., 2014). Endogenous or applied electric fields are involved in modulating the extracellular voltages and have a causal effect on cognitive functions such as memory formation (Anastassiou et al., 2011; Francis et al., 2003; Fröhlich and McCormick, 2010) and cortical column synchronization (Rebollo et al., 2021). The effect of local field potential on several cognitive processes in animals and humans has been reported in several studies (Harvey et al., 2009; Kreiman et al., 2006, 2000; Quiroga et al., 2005). Although it is known that epileptiform activity can be generated without synaptic transmission, (Cruikshank et al., 2004; Haglund and Hochman, 2005; Radman et al., 2007) the mechanisms of non-synaptic propagation of neural activity by electric field coupling in vivo has yet to be analyzed.

While epileptiform and slow-wave activity have been found to propagate at ~0.1 m/s using electric field coupling in the hippocampus in recent studies, (Chiang et al., 2019; Shivacharan et al., 2019; Zhang et al., 2016) the speed of propagation for another propagation mechanism such as diffusion is significantly different with speeds between 0.0004 and 0.008 m/s. Axon conduction speed is between 0.3 and 0.5 m/s (Francis et al., 2003; Jensen and Jensen, 2008; Kibler et al., 2012; Lian et al., 2001; Meeks and Mennerick, 2007; Miles et al., 1988). However previous studies have shown non-synaptic waves can propagate through a complete cut in vitro in the hippocampus (Chiang et al., 2019; Shivacharan et al., 2019) and more recently in the cortex (Rebollo et al., 2021) thereby eliminating axons as a possible mechanism of propagation. Although the speed of propagation seems to be a characteristic feature of electric field coupling propagation in vitro, parameters of the activity such as frequency, amplitude and duration could affect the propagation. This study focuses on the role of electric field effects in vivo by analyzing the propagation of neural activity through a transection of the hippocampus in anesthetized rats. 4-aminopyridine (4-AP), an epileptogenic agent was used to induce the neural activity such as interictal spikes and seizure events (Kibler and Durand, 2011). One of the parameters that is common to several types of epileptiform activity is the speed of propagation at ~0.1 m/s (Zhang and Durand, 2013) which remains within a narrow range irrespective of the model used. In addition to epileptiform activity propagating at ~0.1 m/s, theta oscillations are also reported to propagate at speeds of 0.08–0.107 m/s in the hippocampus of rats (Lubenov and Siapas, 2009).

The study focuses on the characteristics of spikes, seizure events and theta wave propagation in the hippocampus along the longitudinal axis through a transection of the neural tissue by electric field coupling.

2. Materials and methods

2.1. Animals

Four adult male and female Sprague-Dawley rats (150-350 g) were used for each of the in vivo acute experiments in the study. The animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee of Case Western Reserve University. The animals were anesthetized using 1–3% isoflurane for the entire duration of the experiment with vitals being monitored. A stereotactic apparatus was used to secure the head of the animal for the surgery (Kopf Instruments). The skull of the animal was exposed after a small incision rostrocaudally, followed by hydrogen peroxide and electrocautery before drilling five burr holes for electrode insertion in the hippocampus. A rat brain atlas was used to determine the appropriate coordinates for insertion along the longitudinal axis (Herman and Watson, 1987), two electrodes in the Temporal side, Temporal electrode (T), Electrode 1 (Elec 1) (bregma −4.6 mm, lateral 5.1 mm, depth 7.0 mm), Medial Temporal (MT) electrode Elec 2 (bregma −5.3 mm, lateral 5.4 mm, depth 4.6 mm), and two in the septal side, Medial Septal (MS) electrode, Elec 3 (bregma −2.5 mm, lateral 2.0 mm, depth 3.0 mm), and Septal electrode (S), Elec 4 (bregma −2.0 mm, lateral 2.0 mm, depth 3.4 mm). The reference electrode was a stainless-steel electrode inserted near the posterior bone sutures on the skull. A dental cement (TEETS Denture Material; Co-Oral-Ite Dental Manufacturing Company) was used to secure the electrodes in place after insertion. The site of injection for the epileptogenic agent was bregma −4.1 mm, lateral 4.6 mm, depth 8.0 mm, a 5-μL micro syringe (Hamilton, Reno, NV) was also attached to the stereotactic apparatus. Finally, the coordinates for the transection/cut made with the help of a stereotactic knife (M120 Retractable Wire Knife, Kopf Instruments) was bregma −4 mm, lateral −2.5 mm, depth − 2-4.5 mm, between the Elec 2 and Elec 3, the temporal and septal side (Fig. 1.B). For the electric field blocking experiments, only the Elec 1 and Elec 4 were used; a glass slide was cut to obtain the glass piece required of dimensions 150 μm in width and 5 mm in breadth (Fig. 5.A).

Fig. 1.

Induced interictal spikes propagate across a transection in the hippocampus. (A) Hippocampus anatomical structure - sagittal view. (B) Enlarged hippocampus with the four Temporal and Septal electrodes colour coded to the signals acquired. Gray patch in the hippocampus denotes the transverse transection/cut about 400 μm wide. (C) Interictal spikes propagating through the cut from the Temporal electrode (Elec 1 - Brown), Medial Temporal electrode (Elec 2 - Violet) to the Medial Septal electrode (Elec 3 - Pink) ending in Septal electrode (Elec 4 - Orange). (D) Example of a propagating spike across all electrodes. (E) Non-Propagating spikes that stop in the Medial Temporal electrode (Elec 2 - Violet) and do not cross the cut to the Medial Septal and Septal electrodes. (F) Example of a non-propagating spike in the Temporal electrodes not observed in the Septal electrodes.

Fig. 5.

Theta wave propagation across the transection in the hippocampus. (A) Enlarged hippocampus with the four Temporal and Septal electrodes colour coded to the signals acquired. Gray patch in the hippocampus denotes the transverse transection/cut about 400 μm wide (B) Power Spectrum Density of the Raw signals from all four electrodes as acquired from the animal depicting peaks in the Theta range of 4–12 Hz (n = 4 animals) Power Spectrum Density of the Theta signals digitally filtered, showing clear peaks around 6 Hz. (C) The Theta waves as filtered from the raw signals, Temporal electrode (Brown), Medial Temporal Electrode (Violet), Medial Septal Electrode (Pink) and Septal Electrode (Orange) that propagated across the transection/cut. (D)No significant difference in peak theta frequency in the PSD from all four electrodes (n = 4 animals, N = 32 events, z = 0.00, −1.00, p > 0.05).

2.2. Data acquisition

The neural signals were continuously recorded from the electrodes inserted and sampled at 20 kHz (Power Lab DAQ; AD Instruments) amplified by a gain of 100 and band pass filtered at 0.1 Hz – 5 kHz (Model 1700, A-M systems). The neural activity from the hippocampus was analyzed using MATLAB (Mathworks) to extract the interictal spikes, seizure events, and theta waves. The seizure events were continuous spikes with frequency content more than 5 Hz for a duration of >5 s (Couturier and Durand, 2018; Nissinen et al., 2000). The amplitude and duration of these seizure events were measured using the Root Mean Square (RMS) value with the help of an envelope method. The interictal spikes were paroxysmal events induced before and after the seizure events and were present constantly during the entire course of the experiment. The amplitude and duration of the spikes were calculated using the peak-to-peak value and full width at half height respectively. The theta periods were observed occasionally between these events; a band pass filter at 4–12 Hz was used to extract these theta waves. Cross-correlation and Phase Difference were used to calculate the speed of propagation of the acquired activity.

2.3. Experimental protocol

The epileptogenic agent 4-aminopyridine (4-AP) of concentration of 30 mmol·L–1 was administered in increments of 0.5 μL every 20 min until seizure events were triggered to establish a baseline activity of continuous seizure events for 30 min. The knife was used to make the transection/cut between the Medial Temporal, Elec 2, and Medial Septal, Elec 3 electrodes. The transection/cut was made at an angle of about 45 degrees to the sagittal plane cutting the longitudinal axis of the hippocampus in the transverse direction. Following the cut, the seizure events recurred entirely within 30 min and did not require any more injections of 4-AP. The entire experiment was about two hours of continuous recording of neuronal activity. For the electric field blocking experiment, only two changes were made, only the Temporal electrode, Elec 1, and the Septal electrode, Elec 4 were used, and the glass piece was also attached to the stereotactic apparatus to be inserted in the same coordinates as the transection/cut was made.

2.4. Statistical analysis

All the statistical analyses were performed using IBM SPSS Statistics 20. The data had a non-normal distribution, and thus non-parametric tests were performed to determine statistically significant differences between the propagating and non-propagating activity. Wilcoxon signed-rank tests were employed to test for significance between any two groups, significance level set at 0.05. A z-score analysis of the speed of propagation values was used to differentiate propagating and non-propagating theta waves. All results are represented as their mean values ± the standard deviation values.

2.5. Signal processing

Each neural activity category was analyzed separately using MATLAB to study their other characteristics. The interictal spikes were extracted from the entire recording, each individual spike was categorized based on being observed in all four electrodes, as propagating spikes and as non-propagating spikes otherwise, observed only in Temporal and Medial Temporal Electrodes, Elec 1 and 2. Further, the propagating spikes had a constant delay as they traveled between electrodes, obtained using cross-correlation of the spikes (Zhang et al., 2014) and was processed in MATLAB. The distance between the electrodes was calculated using the coordinates used at the time of insertion during the surgery, Elec 1–2 as 2.5 mm, Elec 2–3 as 4.6 mm, and Elec 3–4 as 0.64 mm. Thus, the speed of propagation is calculated by dividing the distance with the delay values. For amplitude, the peak-to-peak value is used, and full width at half height for the duration of the spikes. The seizure events were analyzed by first taking the RMS values, (Singh and Pachori, 2017) The envelope was computed using the Root Mean Square (RMS) of the magnitude of voltage signals. The upper envelope was calculated using MATLAB with 0.5 s window length. There is a consistent incline and drop of the envelope indicating the begging of the end of the seizure event as the frequency content becomes less than 5 Hz. Cross-correlation analysis was used to determine the delay at each of the four electrodes and therefore the speed. The seizure wavefront events were clearly distinguishable from the baseline and hence onset and offset were based on the frequency of the ictal spikes. The interictal spikes were classified in two categories: “propagating” and “non-propagating” based on the following criterion: spikes with a similar delay (constant speed) observed at all four electrodes were classified as “propagating”. All other spikes were classified as “non-propagating”. In addition to cross-correlation analysis, for theta waves, a phase difference method was also used to determine the speed of propagation. The time delay ($\Delta \mathrm{t}$) between two theta waves was calculated using the equation,

$$\Delta \mathrm{t}=\Delta \phi ∕2\pi \mathrm{f}$$ (1)

dividing the phase difference ($\Delta \phi$) determined from the vector dot product (Cappaert et al., 2009) by the fundamental frequency (f) observed. The distance between the electrodes was used to calculate the speed of propagation. Cross-correlation was also used to compare the seizure events before and after the dielectric layer was inserted to block the propagation of the activity.

3. Results

3.1. Propagating interictal spikes have higher amplitude and narrower width than non-propagating spikes

We first analyzed the propagation of 4-AP induced spiking activity recorded at four electrodes sites along the hippocampal axis. Although seizure-like activity was also observed, only spikes without underlying seizures were included in this section. All spikes propagated (100%) from the induction site in the temporal lobe towards the septal area of the hippocampus at a speed of 0.12 ± 0.04 m/s (n = 400 spikes, 4 rats) before the transection. A transection was then made between electrode 2 and 3 and spontaneous seizure activity interlaced with interictal spikes was observed from the four electrodes (Fig. 1.A, B). Some spikes were observed to propagate to the other side of the transection at similar speed of 0.087 ± 0.05 m/s and a delay of 69.5 ± 32.3 ms (see example of propagating spikes in Fig. 1.C and D). Analysis of 400 spikes showed that only 218 spikes propagated (54%) while 195 spikes did not cross the transection. Some spikes (46%) were only observed in the Temporal electrodes, Elec 1–2, and were categorized as non-propagating spikes (Fig. 1.E and F).

Given the stable speed of propagation before the transection, we tested the hypothesis that propagating spikes would arrive on the other side of the transection with an actual delay similar to an expected delay based on speed measured before the transection. The results show that there is no significant difference between the expected delay and actual delay observed after the transection in all the three electrodes (Table 1). Further, the speed of propagation measured after the transection at each electrode was in same constant range of 0.02–0.25 m/s. This shows that the spikes crossed the transection and propagated to the septal electrodes, Elec 3, and Elec 4. The speed is also within the range of characteristic speeds for non-synaptic signal propagation by electric field coupling (Zhang et al., 2016). The amplitude of the spikes decreased once they crossed the transection, the average amplitude dropped to 0.35 ± 0.2 mV at Elec 3 from the previous observed value of 1.2 ± 0.54 mV at Elec 2, similar to before the transection where the amplitude also decreased from 0.98 ± 0.29 mV (Elec 2) to 0.48 ± 0.26 mV (Elec 3) (Table 1).

Table 1.

Comparison of delay and amplitude before and following the transection.

Before the transection N = 400		After the transection N = 218						Before the transection	After the transection
Speed Between	Actual speed in m/s	Location	Expected delay in ms	Actual delay in ms	Speed Between	Actual speed in m/s	Location	Amplitude in mV	Amplitude in mV
Elec 1–2	0.17 ± 0.15	At Elec 2	69. ±54.2	56.58 ± 32.1	Elec 1–2	0.07 ± 0.07	At Elec 2	0.98 ± 0.29*	1.2 ± 0.54*
Elec 2–3	0.11 ± 0.08	At Elec 3	76.7 ± 54.7	69.53 ± 32.3	Elec 2–3	0.09 ± 0.05	At Elec 3	0.48 ± 0.26*	0.35 ± 0.2*
Elec 3–4	0.09 ± 0.12	At Elec 4	6.38 ± 5.3	5.61 ± 3.13	Elec 3–4	0.11 ± 0.2	At Elec 4	0.43 ± 0.22*	0.32 ± 0.14*

Tabulated values are means across the four animals. Each animal measurement is the mean of several observations for that animal. The speed observed between the electrodes was consistently in the range 0.02–0.25 m/s (Speed before the transection). Hence, the expected delay was calculated based on speed before the transection and compared to the actual observed delay after the transection. No significant difference between the expected delay and the actual delay after the transection (highlighted bold section). The propagating spikes thus arrived at the electrodes after the transection with similar constant speed in the same range (Speed after the transection). This supports our hypothesis that the spikes propagate through the transection. The last two columns show the amplitude of the spikes

^*p < 0.005 there was a significant difference between the amplitudes before and after the transection in all three electrodes (Last two columns).

Various characteristics of the spikes such as amplitude, width, power, rising time, falling time, and speed of propagation from one electrode to another were analyzed to identify the characteristics of the propagating and the non-propagating spikes. Only two of the five parameters used to classify propagating vs non-propagating spikes were significantly different: amplitude, and width. The amplitude was calculated using the peak-to-peak voltage and width was defined as to the full width at half-height of the individual spike (Fig. 2.A, B). The amplitude of the propagating spikes was 1.48 ± 0.5 mV which is significantly higher (n = 413, Z = −10.156, p < 0.001, 4 rats) than that of the non-propagating spike mean amplitude, 0.85 ± 0.35 mV (Fig. 2.D). The mean value of the width of the propagating spikes (84 ± 20 ms) was significantly lower than the non-propagating spike width mean of 160 ± 32 ms (Fig. 2.E), (n = 413, Z = −11.989, p < 0.001, 4 rats). Although there was an overlap between the two groups, the data show that spikes that could propagate through the transection have large amplitudes and narrow widths compared to the non-propagating ones. (Fig. 2.C). A distribution of the amplitude values also supports the significant difference between the propagating amplitudes (1–3 mV) and the non-propagating amplitudes (0.4–1.6 mV) (Fig. 2.F). Similarly, the distribution of duration of propagating spikes (30–130 ms) and the duration of non-propagating spikes (60–230 ms) shows the significant difference between the two groups (Fig. 2.G).

Fig. 2.

Propagating interictal spikes have higher amplitude and narrower width. (A) Individual propagating spike shown with the definition of the amplitude and width measurements, example of high amplitude and narrow width (B) Individual non-propagating spike with smaller amplitude and wider width (C) Scatter plot of propagating (Green) and non-propagating (Red) spikes, width in seconds in the X-axis versus amplitude in mV in the Y-axis, the position of the cluster of green dots indicate that the large narrow spikes propagate over the small wider ones though there is an overlap of the two groups. (D) Scatter plot of propagating and non-propagating spike amplitudes with propagating mean of 1.48 ± 0.5 mV significantly higher than the mean amplitude of the non-propagating spikes 0.85 ± 0.35 mV for n = 4 animals (Z = −10.156, p < 0.001). (E) Scatter plot of widths with the mean width of the propagating spikes 84 ± 20 ms was found to be significantly lower than that of the non-propagating spikes 160 ± 32 ms (Z = −11.989, p < 0.001). (F) Histogram shows the range of propagating amplitude to be between 1 and 3 mV wider than that of non-propagating amplitude at 0.4–1.6 mV. (G) Histogram shows the range of propagating spike duration to be between 30 and 130 smaller than that of non-propagating spike duration at 60–230 ms.

3.2. Seizure events propagate through the transection/cut in the hippocampus

Since some of the interictal activity could cross the gap, we next asked if the seizures events could also propagate. Seizure events, defined as spiking activity with frequency content more than 5 Hz for a duration of >5 s, spontaneously developed following the 4-AP injections and were observed to propagate along the temporal-septal axis (Fig. 3.A). Similar events could also be observed 10 min after making the transection/cut propagating in the same direction. We first classified these events into “propagating” seizure events (Fig. 3.B, C) and “non-propagating” seizure events (Fig. 3.D, E). Propagating events were identified as such if they were observed arriving at electrode 1,2,3 and 4 with an average delay of 29.8 ± 14.1 ms corresponding to a speed of ~0.1 m/s using cross-correlation of the signal. The speed of propagation was calculated (Fig. 4.C), and no significant difference in the speed between the four electrodes was found (Z1 = −0.503, Z2 = −1.51, p > 0.05). The speed from the temporal to the medial-temporal electrode was 0.12 ± 0.06 m/s, from medial-temporal to medial-septal electrode was 0.15 ± 0.09 m/s and from medial-septal to septal electrode was 0.1 ± 0.05 m/s.

Fig. 3.

Seizure events propagate across a transection in the hippocampus. (A) Enlarged hippocampus with the four Temporal and Septal electrodes colour coded to the signals acquired. Gray patch in the hippocampus denotes the transverse transection/cut about 400 μm wide (B)(C) Post experimental histological figures showing the complete knife-cut transections (D) Example of seizure events that propagate through the cut from the Temporal electrode (Brown), Medial Temporal electrode (Violet) to the Medial Septal electrode (Pink) ending in Septal electrode (Orange) using electric field coupling with a delay (E) Visible delay between the propagating signals indicating propagation from the Temporal to the Septal electrodes (F) Non-Propagating seizure events that stop in the Medial-Temporal electrode and do not cross the cut to the Medial Septal and Septal electrodes (G) Enlarged seizure spikes that do not propagate to the Septal electrodes.

Fig. 4.

Speed and characteristics of propagating seizure events. (A) Enlarged hippocampus with the four Temporal and Septal electrodes colour coded to the signals acquired. Gray patch in the hippocampus denotes the transverse transection/cut about 400 μm wide (B) Example of envelope calculated using RMS value of the seizure event, Temporal (Brown) and Medial Temporal Electrode (Violet) signals from n = 4 animals (C) The mean speed from the Temporal to the Medial Temporal electrode was 0.12 ± 0.06 m/s, from Medial Temporal to Medial Septal electrode was 0.15 ± 0.09 m/s and from Medial Septal to Septal electrode was 0.1 ± 0.05 m/s. No significant difference was observed in the speed of propagation between electrodes as shown (z = −0.503, p > 0.05, z = −1.511, p > 0.05). (D) Comparison of mean amplitude of the propagating seizure envelope in the Medial Temporal electrode at 0.47 ± 0.26 mV with the mean amplitude of the non-propagating seizure envelope at 0.56 ± 0.12 mV. No significant difference was observed (n = 4 animals, N = 27 events, Z = −0.35, p > 0.05). (E) Comparison of mean duration of the propagating and non-propagating envelope at 16.35 ± 2.4 s and 17.2 ± 6.1 s. No significant difference between them (n = 4 animals, N = 27 events, Z = −0.56, p > 0.05).

Further analysis showed that about 70% of the seizure events could propagate through the transection/cut and reach the septal electrodes. The envelope of the seizure events was obtained using the RMS analysis to study the characteristics of propagating events such as amplitude, duration, and speed of the seizure events (Fig. 4.B). The mean amplitude of the propagating seizure envelope in the medial-temporal electrode was 0.47 ± 0.26 mV which was not significantly different (n = 4 animals, N = 27 events, Z = −0.35, p > 0.05, 4 rats) from 0.56 ± 0.12 mV (Fig. 4.D). Similarly, the mean duration of the propagating and non-propagating envelope was 16.35 ± 2.4 s and 17.2 ± 6.1 s respectively with no significant difference between them (n = 4 animals, N = 27 events, Z = −0.56, p > 0.05, 4 rats) (Fig. 4.E.

3.3. Theta waves propagate through the transection/cut in the hippocampus

We then analyzed the low amplitude signals with no seizure events nor interictal spikes. Frequency spectrum analysis revealed high power density in the frequency range of 1–20 Hz close to the theta waves range (Fig. 5.B). The time-frequency analysis of the signals from the recording electrodes showed signals in the theta bands in all four electrodes. These time segments were further processed using a digital band pass filter of 4–12 Hz to detect the theta wave signals (Fig. 5.C). Some of these theta waves signals were observed in all four electrodes with a delay within a narrow range (24.9 ± 15.8 ms) indicating propagation of theta waves travelling to each electrode along the longitudinal axis (Fig. 6.B). Waves that were observed only in the temporal electrodes were categorized as the non-propagating waves, while 88% of the waves propagated to the septal electrodes and were categorized as the propagating waves. The speed of the theta waves was calculated using two methods, cross-correlation of extracted theta segments about 5 s in duration and phase difference calculation using the vector dot product (Cappaert et al., 2009). In case of the propagating Theta waves, there was no significant difference between the values obtained from cross-correlation (CC) and phase difference (PD) (n = 4 animals, N = 32 events, z = −1.962, z = −1.874, z = −1.31, p > 0.05). And there was no significant difference among the speeds of propagation from Elec 1 to Elec 2 (CC - 0.18 ± 0.1 m/s, PD - 0.13 ± 0.06 m/s), from Elec 2 to Elec 3 (CC - 0.16 ± 0.12 m/s, PD - 0.1 ± 0.04 m/s), from Elec 3 to Elec 4 (CC - 0.17 ± 0.17 m/s, PD - 0.16 ± 0.12 m/s), (Fig. 6.C). The speed values for non-propagating theta waves were also determined. A z-score analysis was used to compare the speed values of non-propagating and propagating theta waves (Fig. 6.D). The threshold for the waves to be classified as non-propagating was set at ±2.5, (Sakai et al., 2015) the speed of the propagating theta waves had a score of about −0.5. No other distinguishing feature was observed between the propagating and non-propagating theta waves. Thus, theta waves in the presence of 4-AP could propagate non-synaptically across the cut as indicated by recordings from the four recording electrodes. Taken together these results indicate both interictal spikes and seizure activity can propagate through a complete transection but that not all events can cross the transection. Similarly, a significant portion of the theta events analyzed could also propagate through the transection. The speed of all three types of activity fell within a narrow range (0.05–0.35 m /s) previously reported in vitro as characteristic of electric coupling.

Fig. 6.

Speed of theta wave propagation across the transection. (A) Enlarged hippocampus with the four Temporal and Septal electrodes colour coded to the signals acquired. Gray patch in the hippocampus denotes the transverse transection/cut about 400 μm wide. (B) Visible delay between the signals indicating propagation from the Temporal to the Septal electrodes. (C) Example of delay calculation using cross-correlation analysis of two signals. (D) Histogram of Z-scores calculated from speed values, showing a clear threshold at ±2.5 for the theta waves to classify them into propagating (green) and non-propagating waves (red). (E) Speed of propagation between the electrodes calculated using Cross-correlation (CC - Purple) and Phase Difference (PD - Peach). Speed from Temporal electrode to the Medial Temporal electrode was CC - 0.18 ± 0.1 m/s, PD - 0.13 ± 0.06 m/s, from the Medial Temporal electrode to the Medial Septal electrode was CC - 0.16 ± 0.12 m/s, PD - 0.1 ± 0.04 m/s, from the Medial Septal electrode to the Septal electrode was CC - 0.17 ± 0.17 m/s, PD - 0.16 ± 0.12 m/s. No significant difference was found between the values obtained from Cross-correlation and Phase Difference (n = 4 animals, N = 32 events, z = −1.962, z = −1.874, z = −1.31, p > 0.05).

3.4. Seizure event propagation is blocked by a dielectric layer inserted within the cut in the hippocampus

The non-synaptic mechanism mediating the propagation through the cut is thought to be ephaptic or electric field coupling since axons have been cut and diffusion is too slow to account for the speed. Electric fields (E) and current density J (J = σE) can propagate through volume conductors but not through dielectric material. To further test the hypothesis of ephaptic coupling as a mechanism of propagation, we designed an experiment for measuring propagation through the transection with either conductive solution (Cerebrospinal Fluid) or dielectric material by inserting a thin piece of glass (150um, 5 × 10 mm) inside the cut. Before the glass insertion, the seizure events could propagate to the Septal electrode, Elec 4 from the Temporal electrode, Elec 1 (Fig. 7.B). The glass piece was placed perpendicularly to the longitudinal axis of the hippocampus similar to the cut orientation (Fig. 7.A). The signals were recorded before glass insertion, following glass insertion, and following glass removal. The seizure events propagation was blocked immediately by glass insertion (Fig. 7.C). During the baseline recording period, 100% of events could propagate (N = 19 events, n = 4 animals). Following the insertion of the glass, complete block was observed (0% seizure event propagation). Following the removal of the glass layer 79 ± 8% seizure events could propagate after a 30 min period. (Fig. 7. D, E). The distribution of the cross-correlation coefficient for the events observed from both the electrodes before, during, and after the glass insertion is shown in Fig. 7.F. To show the blocking of the seizure events by the glass piece, we used cross-correlation analysis between signals obtained from electrodes just before the transection and after the transection and classified those events into propagating (r² > 0.3) and non-propagating ones (r² < 0.3). With the glass in place no event could propagate and r² remained below 0.2. However, the correlation for events before the insertion and following the removal of the glass layer was significantly higher (r² = 0.8), the speeds of propagation of the seizure were determined to be 0.34 ± 0.34 m/s and 0.35 ± 0.21 m/s respectively. No significant difference was observed between the propagation speeds of the events before glass insertion and after its removal (n = 4 animals, Z = −1.48, p > 0.05).

Fig. 7.

Seizure event propagation is blocked by a dielectric layer inserted within the transection. (A) Enlarged hippocampus with the four Temporal and Septal electrodes colour coded to the signals acquired, with a representation of the glass piece inserted between the Medial Temporal Electrode and the Medial Septal electrode to block the electric field coupling. (B) Experiment picture with glass inserted in the hippocampus of a rat (C) Cryosection of the brain post experiment showing the electrode positions and glass piece position after removal (D) Example of a seizure event with the RMS envelope before the glass was inserted. (E) Example of a seizure event with the RMS envelope after the glass was inserted (F) Example of a seizure event with the RMS envelope after the glass was removed, hence transecting the hippocampus transversely across its longitudinal axis similar to the transection/cut from the previous experiments. (G) Comparison of the percentage of seizure events propagation, 100% propagation in the baseline condition without any intervention (dark green), then complete block shown by the 0% seizure event propagation after the glass was inserted, and 79 ± 8% seizure event remission after the glass piece was removed (light green) in 4 animals (H) Histogram of the cross-correlation values obtained in all three conditions, after the glass was inserted, the value dropped to below 0.2 (red) from above 0.7 (dark green) value before the glass insertion. Light green bar above 0.8 shows the cross-correlation value of the seizure events that recurred after the glass was removed.

4. Discussion

We report that three types of neural activity can propagate across a complete transection of the rodent hippocampus in vivo. The propagation of interictal spikes, seizure events, and theta waves in the longitudinal direction of the hippocampus was based on the assumption that electrical activity can propagate through a volume conductor with a negligible propagation delay (Plonsey, 1999). This study focuses on understanding the mechanisms and characteristics that contribute to the in vivo propagation of various types of neural activity across the transection.

Interictal spikes have been observed in relation to epileptogenesis, (Staley and Dudek, 2006) and some previous studies have proposed that these spikes can inhibit seizure events (De Curtis and Avanzini, 2001) thereby attesting their significance in understanding the mechanism of epilepsy (Staley et al., 2011). In the present study interictal spikes induced by the injection of 4-AP could propagate across the transection with a speed of about 0.1 m/s similar to spike propagation in human patients (Schevon et al., 2012). Previous studies have reported observations of interictal spike propagation in the cortex and hippocampus, mapped their synaptic pathways, and established the reproducibility in animals and humans from both in vitro and in vivo experiments (Conrad et al., 2020; Sabolek et al., 2012; Stoop and Pralong, 2000; Ulbert et al., 2004). While several studies indicate the importance of quantifying different characteristics of the spikes such as amplitude and duration to discern the mechanism responsible for their generation, (Deeba et al., 2019; Frost Jr et al., 1986; Varma et al., 1997) the mechanism of propagation is assumed to be axonal conduction and synaptic transmission. We report here that some of neural events can propagate by volume conduction, and we have identified the characteristics of some of those events. Sharp interictal spikes with high-amplitude and narrow-width are more likely to propagate across the transection, compared to spikes with small amplitudes and long duration. The high amplitude of the spikes generates a corresponding large electric field thought to be a key factor for maintaining propagation of activity by electric field (Qiu et al., 2015). The significance of narrow width is not clear but the data in Fig. 2 indicate that most of the high amplitude spikes also have narrow width. The propagation of narrow width spikes through a transection could be clinically significant since spikes preceding seizures also have a narrow width (Chauvière et al., 2012). Wider spikes of long duration did not propagate to the septal electrodes across the physical transection. The mechanism for the lack of propagation of low amplitude longer spike is not known but could be a special property of electric field coupling whereby low amplitude, long-lasting electric field events could generate inactivation of NMDA receptors and prevent excitation of local neurons. The speed of the propagating interictal spikes was in the range of 0.05–0.2 m/s similar to the speed reported in the hippocampus (Stoop and Pralong, 2000).

Epileptic seizure events induced in the experiments also propagated longitudinally along the hippocampus similar to the interictal spikes, these events were studied using their ictal wave front. Although some of the parameters were similar to those observed in human studies (Schevon et al., 2012; Smith et al., 2016) we did not find any criteria of the seizure events that would allow us to classify the events into either propagating or non-propagating through a transection. This result suggests that the non-propagating events require intact axons and synaptic transmission. Although propagation of seizure activity by axonal transmission is a well-established phenomenon, (Milton, 2003) our results indicate that there is another mechanism involving ephaptic transmission that can underlie propagation in addition to axonal propagation. Using ephaptic transmission, the ictal wave front propagated across the transection with a constant speed similar to that of the interictal spikes, while other studies have shown that the wave front can propagate at slower speeds (Weiss et al., 2013) as the focus is moving and generating fast moving spikes. This discrepancy can be explained by the fact that in our preparation, the seizure focus was fixed at the site of injection and did not propagate. Seizure propagation and suppression by low frequency stimulation in the contralateral hippocampus has been studied in vitro and in vivo by our group (Couturier and Durand, 2018; Toprani and Durand, 2013). A special intact hippocampal preparation in vitro showed propagation of induced seizure activity using the commissural tracts to the contralateral hippocampi. Cortical seizures propagated to the hippocampus at varied speeds in vivo and were suppressed by low frequency stimulation of the corpus collosum. The current study however focuses on the short-range propagation by ephaptic coupling across a transection in the singular hippocampus and will need further studies to determine the long-range propagation to the contralateral hippocampus. Another study from our lab (Couturier and Durand, 2018) does show that focal injections of 4-AP induced seizure front generalize and propagates from cortex to the hippocampus in acute in vivo experiments and reverse propagation to the cortex was also observed. The time delay between the seizure events was recorded to be a wide range of values from milliseconds to seconds that can be contributed to different mechanisms including ephaptic coupling and axonal propagation.

Experiments in the present study were performed in the CA1 region of the hippocampus known for its ability to generate theta wave activity. Studies support the prominence of regular and high amplitude theta waves in the CA1 region (Buzsáki, 2002; Manns et al., 2007). Similar theta power is preserved in the same layer along the longitudinal axis of the hippocampus (Bullock et al., 1990). The importance of theta oscillations in different cognitive functions have been studied in hippocampal slices, freely moving animals, and in human patients. They play a vital role in spatial memory, navigation, (Buzsáki and Moser, 2013; Ekstrom et al., 2005; Kahana et al., 1999) and their connection to interictal spikes in epilepsy has also been well documented (Colom et al., 2006; Fu et al., 2018). The relatively quieter periods without seizures and spikes were processed using power spectral density analysis as 5 s segments. Our analysis of the raw data showed 90% of the theta segments analyzed had sharp peaks in the slow theta range. We categorize the analysis of our data of theta waves as slow-theta (4–6 Hz). We observed the propagation of theta waves along the longitudinal axis of the hippocampus at speeds similar to that of the spikes and to theta wave from other preparations (Lubenov and Siapas, 2009; Zhang et al., 2018) suggesting a common mechanism. The fact that these speeds match those observed in the intact hippocampus in vivo (Anastassiou et al., 2010) and can travel through a complete transection of the hippocampus suggest that these waves may not require axonal transmission for propagation.

The speed of propagation of the different neural signals observed remained consistently in the range of 0.05–0.35 m/s in line with literature. Previous studies have reported the speed of epileptiform propagation by ephaptic coupling at ~0.1 m/s (Zhang and Durand, 2013) in the absence of chemical or electrical synaptic transmission. This speed is characteristic of non-synaptic propagation by electric field effect (ephaptic coupling) since other modes of propagation in the neural tissue such as axonal conduction or diffusion are significantly different from this reported speed.

Electric field effects in the extracellular space plays a significant role in synchronization of activity in densely packed tissue such as the hippocampus (Dudek et al., 1998; Jefferys, 1995). Buzsáki et al., 1991 speculated that non-synaptic effect of ephaptic interactions could explain the propagation of interictal spikes in the longitudinal hippocampus at speeds (0.2–3 m/s) by a mechanism different from conduction by hippocampal fibers (~0.5 m/s). Both endogenous and applied weak electric fields can affect neural activity (Anastassiou et al., 2010; Qiu et al., 2015; Radman et al., 2007). Weiss et al. (Weiss and Faber, 2010) and Wei et al. (Wei et al., 2014) also postulated that the functional significance of electric field effects in physiological conditions. A recent study indicated the ephaptic coupling is responsible for the synchronization of neural activity within cortical columns (Rebollo et al., 2021). These reports in conjunction with the results discussed indicate the prevalence of electric field coupling and can explain the non-synaptic propagation of the various types of neural activity across a transection.

Studies from our laboratory have also shown that epileptiform activity can propagate in vitro across a transection (Shivacharan et al., 2019). The mechanism of the propagation was attributed to electric field coupling (ephaptic effect) confirmed by placing a dielectric material across the cut (Shivacharan et al., 2019). Therefore, the hypothesis that electric field coupling was also involved in the in vivo experiments reported above with a similar experiment whereby a dielectric material (glass slide) is placed inside the transection separating the two sides of the hippocampus. Similar to the results observed in vitro, 100% of seizure events were blocked when the glass slide was inserted, and propagation resumed upon removal of the slide providing further evidence for ephaptic coupling as the mechanism for propagation of these events.

It is highly unlikely that propagation of the activity through a transection can be explained by the presence of other axonal pathways around the cut or uncut axons within the transaction. The first argument against that hypothesis is that the propagation delay was not affected by the transection ruling out a longer pathway. The second argument is that the size of the glass side place within the cut was at least 3.3 times the width of the hippocampus (1–1.5 mm) (Duffell and Soames, 1998) cross-section effectively cutting possible other non-hippocampal pathways near the recording electrodes. However, the physical obstruction caused by the glass barrier in the tissue also hinders diffusion and does not rule out the possibility that diffusion is involved in the synchronization of neural activity across a cut as previously reported. However propagation of neural activity by diffusion has been directly tested and the speed has been found to be several orders of magnitude too slow to explain the reported results (Lian et al., 2001). Computer simulations of potassium diffusion as a mechanism of propagation are consistent with this result (Chizhov and Sanin, 2020; Durand et al., 2010) leaving electric coupling as the only explanation for the propagation through a transection.

Hence, ephaptic coupling is responsible for the non-synaptic propagation of the epileptiform activity in vivo in this study. This finding is clinically significant as it could explain the failure of resections as a long-term treatment option for epilepsy, where the epileptic region is isolated by making several transections. Studies have shown that Multiple Subpial Transections (MST) has a particularly low success rate in adults and children (Blount et al., 2004; Ntsambi-Eba et al., 2013). Long-term seizure outcome study comparing different surgical interventions for epilepsy shows that only 16% of the patients remained free of all seizures after MST (Téllez-Zenteno et al., 2005). The result from the study supports the conclusion that severing the synaptic connections does not completely prevent the propagation of the epileptiform activity.

In summary, this study focuses on the specific parameters that facilitate propagation of neural activity by ephaptic coupling. The similar speed of propagation observed in vivo supports ephaptic coupling as the propagation mechanism. The results presented also show that propagation is blocked by the presence of a dielectric barrier inside the cut. Further studies on characteristics of ephaptic coupling could lead to therapeutic solutions for neurological conditions of epilepsy by controlling the electric fields in/near the focus to prevent the propagation of the epileptiform activity.

Abbreviations:

4-AP

4-aminopyridine

CA

Cornu Ammonis

CC

Cross Correlation

Elec

Electrode

MS

Medial Septal

MST

Multiple Subpial Transections

MT

Medial Temporal

PD

Phase Difference

PSD

Power Spectrum Density

RMS

Root Mean Square

S

Septal

T

Temporal

Data availability

The data that support the findings of this study are available from the corresponding author, upon reasonable request.