Correction for 1/fn to evaluate group differences across different gamma bands. A, The typical 1/f trend of the power spectrum complicates the examination of a range of gamma activity from 30 to 150 Hz. Representative spectrum from LFP at the SLM from a control and an epileptic rat is shown to exemplify the issue. The codes of the rats are given in parentheses. The factor n better fits the 1/f power law change for different frequency bands. Numbers at fitting lines identify the n factor for different bands in the examples shown (4–2400, 30–210, 150–500, and 500–2400 Hz). Note the fitting consistency between bands when no biological activity is recorded (dead condition for the epileptic rat; rightmost plot). B, Group differences of the n scale factor for different frequency bands. Note the similarities between control and epileptic rats, except for the 150–500 Hz high-frequency band. *p < 0.05. C, Mean group power spectrum (±95% confidence interval) from SLM channel of all control rats (n = 10) and epileptic rats (n = 8). Note the occurrence of some spectral trends across gamma bands but poor difference between groups due to 1/f scaling. D, Same data as in C after power law detrending between 30 and 210 Hz. Note the differences between groups.
Different types of gamma activity recorded in the dorsal hippocampus of epileptic rats. A, Representative example of an epileptic rat recorded with 16-channel silicon probes during object exploration. Left, Coronal section immunostained against NeuN shows the probe track. Scale bar, 200 µm. Right, Representative LFP signals. Note gamma activity along the descending phase of some theta cycles at the SP (trace: 30–60 Hz, black arrowheads). Mid-gamma activity (60–90 Hz) is typically recorded at the positive theta peak of some cycles (gray arrowheads). SR, Stratum radiatum; ML, molecular layer of the dentate gyrus. B, Similar laminar phase shift of LFP and CSD theta cycles with respect to SLM (mean ± 95% confidence interval; n = 10 control, n = 8 epileptic). C, Spatial expression of gamma activity across hippocampal layers in a representative control and epileptic rat (top plot). Note the stereotyped profiles characterized by stronger activity at the hilus and distinct gamma bands at different layers. To avoid the hilar saturating effect, a grand average of three control and three epileptic rats recorded at comparable locations is shown at bottom along the SP and SR layers. D, Relative power after 1/f detrending from LFP signals recorded at the CA1 SP (mean ± 95% confidence interval). Data are grouped according to proximal (close to CA3) and distal [close to subiculum (sub)] recording locations (see scheme at right). The three gamma bands are indicated by different colors: gamma-s (γs; 30–60 Hz; light purple); gamma-m (γm; 60–90 Hz; purple); and gamma-f (γf; 90–140 Hz; dark purple). Right, Nissl staining section showing proximal and distal locations along with the probe track (arrowheads). Scale bar, 500 µm. E, Group differences on the three gammas for LFP recorded at SP, SR, and SLM from all rats. Data (mean ± SD) are presented as proximal (n = 5 control rats; n = 4 epileptic rats) and distal (n = 5 control rats; n = 4 epileptic rats) for SR and SLM layers. Note strong statistical differences accumulated at the fast gamma band. *p < 0.05; **p < 0.005
Theta-phase modulation of gamma activity. A, Representative example of theta–gamma modulation recorded from the control rat shown in Figure 2C. Theta-phase modulation of LFP gamma activity is shown at left for SP, SR, and SLM channels (black-yellow plots). CSD signals help to account for local gamma generators (blue-red plot at right). Theta-phase modulation of CSD gamma activity is shown below the CSD plot for SR and SLM channels. All signals are aligned by the theta peak recorded at SP (vertical discontinuous line). Note the clear theta-phase segregation of the slow and mid-gamma bands from MUA recorded at the SP. Gamma-s (γs) CSD activity (30–60 Hz) is better isolated at the SR during the descending phase of theta cycles in association with the corresponding SR theta sink (vertical arrows). Gamma-m (γm) CSD activity (60-90 Hz) is isolated at the SLM during the SP theta peak, in association with the corresponding SLM sink. SR and SLM sinks are indicated by gray arrowheads in the CSD plot. B, Same as in A for the epileptic rat shown in Figure 2C. Note the poor slow activity compared with the control rat. C, Phase difference between the SR and SLM CSD signals in all control (top plot, n = 10) and epileptic rats (bottom plot; n = 8). Note the poor modulation of SR slow gamma in epileptic rats (only data from six phases are shown). D, Group differences of the SR slow and SLM mid-gamma CSD generators were found only at proximal locations. The p values refer to group effects in a two-way ANOVA for groups and four phases. E, Differences of the gamma modulation index between groups were found only in the slow 30–60 Hz band at proximal locations both for LFP and CSD signals. *p < 0.05; **p < 0.005. F, Correlation between the LFP slow gamma modulation index at SR and the theta power of the SR CSD signal for both groups together (black) and within the epileptic group (blue).
Relationship between slow gamma (30–60 Hz) and performance in the what-where-when episodic-like memory task. A, Schematic representation of the task. The scheme shows object configurations in the sample and test phases (3 min each). The retention interval is 50 min. B, Group data (mean ± SD) of the total exploratory time during the test phase for control rats (n = 6), epileptic rats tested in periods free of epileptiform activities (n = 6), and epileptic rats exhibiting seizures and/or interictal spikes within 2 h of the behavioral task (n = 4; epileptic Sz/IS group), as judged from simultaneous multisite silicon probe recordings. Note the reduced exploratory behavior of rats experiencing epileptiform activities compared with their Sz/IS free epileptic mates. *p < 0.05. C, Data from three rats evaluated in the what-where-when task during periods with and without epileptiform events. Note the different exploratory behavior in the very same animals performing the task with or without signs of epileptiform activities in the dorsal hippocampus. D, Group data (mean ± SD) for the discrimination ratio of object A1 (black) and B2 (green in A), carrying most of the temporal and spatial memory of the exploratory episode. Note epileptic rats discriminate at chance level (0.25). Right plot shows group data for the spatial and temporal memory indices, where and when, respectively. *p < 0.05; **p < 0.005; ***p < 0.0001. Data from six control rats and six epileptic rats free from epileptiform activities. E, Correlation between the slow gamma CSD modulation index and the discrimination index for object B2 (A, green). Significant Pearson correlation was found for both groups (black) and the epileptic group alone (blue). F, Correlation between the slow gamma CSD modulation index and a spatial memory index (where) for both groups (black), but not for either group alone.
Tetrode recordings of single cell activity. A, Eight independent tetrodes were advanced into the dorsal hippocampus to target CA1 cells at the stratum pyramidale of normal and epileptic rats. Sections immunostained against NeuN were used to identify proximal and distal recording locations. Immunostaining against the CA2-specific protein PCP4 was used to validate proximal CA1 locations. Blue is bisbenzimide used to stain cell nuclei. B, Typical spike waveform and autocorrelogram of a putative pyramidal cell (red) and an interneuron (blue) recorded from an epileptic rat (KWGAT6). C, Units were classified according to several criteria, including spike width data. A group of units remained unclassified (gray). The number of units is given in the figure. Data from three control and three epileptic rats. D, Firing rate data from putative pyramidal cells and interneurons recorded at proximal and distal locations in control and epileptic rats. *p < 0.05; **p < 0.001.
Disruption of interneuronal firing patterns at the proximal CA1 of epileptic rats. A, Representative tetrode recording from a control rat (only one channel is shown). Spikes from individually sorted pyramidal cells and one interneuron are shown in different colors. Large black ticks mark the position of theta peaks at SP. B, Same as in A for an epileptic rat. C, Proximodistal group differences (mean ± SD) of slow gamma modulation index as obtained from tetrodes (eight proximal and eight distal from three control rats, and six proximal and six distal from three epileptic rats) and silicon probes (four proximal and five distal from nine control rats; and three proximal and three distal from six epileptic rats). Different degree of spike contamination in tetrode and silicon probe recordings likely prevents a quantitative comparison between these datasets. D, Theta phase-locked firing (top row) of the interneuron and pyramidal cell shown in A. Firing phase modulation (8.7º bin size) was quantified using the theta-phase vector length. Bottom histograms represent the unit firing autocorrelograms (1 ms bin size). Insets show the power spectrum of the unit autocorrelogram, from which a theta autocorrelation index was estimated. This interneuron was subclassified as a putative PV basket cell (Table 1, unit 0405RMN_OF_T1R-01_part1_1). E, Same as in D for the interneuron and pyramidal cell shown in the epileptic rat in B. The putative interneuron could not be quantitatively subclassified, but it resembled an OLM interneuron (Table 1, unit 2911KWGAT6_OF_1_btp_01_T5R15-03_3). F, Group data (mean ± SD) for the theta-phase vector length (top plot) and the theta autocorrelation index (bottom) for all units. Note differences between groups concentrated at the autocorrelation index for proximal interneurons. *p < 0.05. G, Computer simulation of theta phase-locked firing with different degrees of firing rate (mean ± SD from 100 simulations). Decreasing the global firing rate had a stronger impact in the autocorrelation index than in the phase vector length, indicating different sensitivity of these measures. Arrows point to simulations shown at left. H, I, Data from simulations indicated by arrows in G. Both the theta-phase histograms (top row) and the autocorrelograms (bottom) are shown. Note poor rhythmicity in the autocorrelogram but not apparent change in theta-phase firing preference when the global firing rate decreases to <20%.
Rhythmic firing of proximal parvalbumin basket cells is specifically disrupted in epileptic rats. A, Putative interneurons were subclassified according to criteria previously confirmed in identified types, including PV basket cells (PV-BC) and OLM interneurons. Firing dynamics at run/stop transitions was one of these criteria. An example is shown for two tetrodes from a control rat. Note poor firing modulation at run/stop transition of the interneuron in tetrode 8 (OLM; Table 1, unit 0505RMN_TrD_1_T8_1) vs strong modulation of the interneuron in tetrode 1 (PV-BC; unit 0505RMN_TrD_1_T1-03_1). The second unit shown at each tetrode was a putative pyramidal cell. Bottom traces show LFP from one channel from each tetrode. B, Firing of the units shown before during SPW ripples enlarged from A (thick line). Note different participation of interneurons classified as OLM and PV-BC in the perievent histograms at right. One of the ripple events is expanded at the bottom. C, Theta phase-locked firing of the units shown before. D, Same as in A for a putative PV basket cell in an epileptic rat showing modulation by run/stop transition (Table 1, unit 1612KAW90_OF_T5R16-01.mat_2). E, Same as in B. Histograms show SPW-triggered firing of the PV-BC shown at left and that of the putative OLM cell shown in G. Note that large-amplitude, high-frequency oscillations typical of the epileptic hippocampus that saturate the voltage scale shown here. F, Theta-phase firing histogram of the putative PV-BC shown before. G, Theta-phase firing histogram of a putative OLM interneuron from rat KWGAT6 (unit 2911KWGAT6_OF_1_btp-01_T4-01_1). H, Neurochemically identified PV basket cell recorded intracellularly from an epileptic rat (arrow; scale bar, 50 µm). Note PV-positive synaptic boutons around pyramidal cell somata shown at right (arrowheads, two 2-µm-thick optical section; scale bar, 15 µm) and negative immunoreactivity to somatostatin (SOM) at bottom (scale bar, 50 µm). I, Activity of the PV basket cell shown in H during theta (left) and SPW ripples (right) validate classification criteria in the context of TLE. J, Theta-phase vector length and autocorrelation index for theta and gamma from all proximal putative PV basket cells (n = 4 control rats, n = 4 epileptic rats) confirm preferential disruption of their rhythmicity in TLE. *p < 0.05; **p < 0.001. See also Table 1.