Ripples and fast ripples in epileptogenesis: a characterization of hippocampal high-frequency oscillations after status epilepticus
High-frequency oscillations (HFOs) are spontaneous, transient, and fast oscillations (80- 500 Hz) observed in the cortical regions of mammals. Recorded mainly during slow-wave sleep and quiet waking, physiological HFOs (a.k.a. ripples) participate in sensory perception and memory consolidation. Conversely, pathological HFOs (a.k.a. fast ripples) occur at a higher frequency than ripples and are commonly recorded in seizure-prone structures of epileptic subjects. Despite the strong association between fast ripples and seizure onset zones, it is unclear whether they emerge from the degradation of the ripple generation network or if they develop de novo in the adult brain after cell death and synaptic reorganization triggered by a brain insult.
Here, we used electrophysiological recordings and an animal model of status epilepticus to study the expression of HFOs associated with hippocampal epileptogenesis in the mouse. Animals were chronically implanted with bilateral hippocampal and cortical electrodes and a guide cannula targeting the right dorsal hippocampus. After baseline recordings (for at least two days, minimum six h/day), animals received a single intrahippocampal administration of pilocarpine (700 µg/µl). As we have previously shown, a sustained status epilepticus (SE) evolves after that.
Initially, our analysis focused on detecting, extracting, and classifying HFOs recorded in the hippocampal electrodes. First, features such as the central HFOs frequency, duration, power, and rate were obtained in a 5-minute slow-wave sleep episode using the RIPPLELAB, a Matlab-based toolbox for semi-automatic extraction of HFOs. Then, we merged the extracted features over five time-points: before SE (baseline period), one day after the SE (SE+1), two days after the SE (SE+2), the first week after the SE (SE+7), and between 10 and 30 days after SE (SE+30).
We identified 1,334 true-positive ripples from 6 animals. Ripple rate monotonically decreased after the SE (F(4,24)= 2.98, p=0.039, factor time, one-way ANOVA), dropping from 16.2±7.2 events/min to 7.8±4.3 events/min in the baseline period and SE+1, respectively. We recorded the lowest ripple rate at the chronic period (SE+30) (6.9±6.2 events/min). Ripple central frequency also decreased after SE (F(4,1329)= 29.9, p<0.01, factor time, one-way ANOVA) although the most substantial reduction was observed in SE+2 (in comparison to baseline: unpaired mean difference= -9.9 Hz, 95%CI: [-11.8 and -8.0]; p<0.01, two-sided permutation t-test). We observed no statistical differences in ripple duration regardless of the time after SE. The average ripple duration was 39.8±12.8 msec. Finally, SE significantly reduced ripple power over time F(4,1329)= 10.8, p<0.01, factor time, one-way ANOVA) and reached the lowest power one day after SE (in comparison to baseline: unpaired power mean difference = -0.07 mV²/Hz, 95%CI: [-0.09 and -0.05]; p<0.01, two-sided permutation t-test). Spearman correlation analysis revealed a positive association between ripple frequency and power at SE+1 (r=0.30, p<0.01) and SE+7 (r=0.71, p<0.01) considering all detected events. In this line, looking at each animal at each time point, ripple rate and power showed a positive correlation (r=0.49, p<0.01) as well as power and frequency (r=0.38, p<0.05). Surprisingly, latency to the first seizure, maximum Racine scale during SE, and length of SE did not correlate with any ripple feature. A positive trend was seen with body weight and rate of events between days, however, no significance was achieved (Spearman correlation, r=0.31, p=0.12).
We also analyzed the appearance of fast ripples in SE. In this case, we focused our analysis on transient events with frequencies above 200 Hz. Surprisingly, in all animals, fast ripples appeared in the first 10 minutes of the acute, pilocarpine-induced seizures (n=409; 68.1±34.0 events/animal within 30 min of SE). Fast ripples usually co-occurred with the ictal spike, and they were followed by slower oscillations (circa 60-100 Hz). We observed phase reversal between this electrode and adjacent ones. The fast ripple mean frequency was 283.8±28.6 Hz, the mean power was 0.5±0.1 mV²/Hz, and their mean duration was 31.7±13.7 msec. As SE evolved, the rate of those events decreased, and they were completely abolished by diazepam (10mg/kg, i.p.). All animals presenting fast ripples during acute SE also showed fast ripples in the following days. The rate of fast ripples during 5-min slow-wave sleep was higher on the first day (SE+1) and decreased in the following days (events per min: 14.0±6.8, 15.0±2.5, and 3.8±3.6 in SE+1, SE+2 and SE+30 periods, respectively; F(4,19)=5.4, p<0.01, one-way ANOVA). As expected, fast ripple mean frequency was higher than ripples (mean difference=+156 Hz, 95%CI: 153 and 158, p<0.05, unpaired t-test). Fast ripples also showed stronger power (mean difference=+0.119 mV²/Hz, 95%CI: 0.098 and 0.138, p<0.05, unpaired t-test), and shorter duration (mean difference=-8.12 msec, 95%CI: -9.55 and -9.58, p<0.05, unpaired t-test) than physiological ripples.
To our knowledge, this work is the first study to describe the temporal changes of ripples features during epileptogenesis extensively. Our results demonstrate that the ripple alterations are partially explained by SE severity. Also, we showed that pathological HFOs (fast ripples) occurred in the first minutes of the SE. This observation suggests that fast ripples do not need cell death or synaptic reorganization to be expressed but rather reflect abnormal balance between excitation and inhibition usually observed in acute and induced seizures. Future analysis will demonstrate if the reduction in ripple rate and power can explain the poor performance in memory and learning tasks observed in animals with spontaneous seizures.