Divergent paths to seizure-like events

Abstract

Much debate exists about how the brain transitions into an epileptic seizure. One source of confusion is that there are likely to be critical differences between experimental seizure models. To address this, we have compared the evolving activity patterns in two widely used in vitro models of epileptic discharges. Brain slices from young adult mice were prepared in the same way and bathed either in 0 Mg²⁺ or 100 µmol/L 4AP artificial cerebrospinal fluid. We have found that while local field potential recordings of epileptiform discharges in the two models appear broadly similar, patch-clamp analysis reveals an important difference in the relative degree of glutamatergic involvement. 4AP affects parvalbumin-expressing interneurons more than other cortical populations, destabilizing their resting state and inducing spontaneous bursting behavior. Consequently, the most prominent pattern of transient discharge ("interictal event") in this model is almost purely GABAergic, although the transition to seizure-like events (SLEs) involves pyramidal recruitment. In contrast, interictal discharges in 0 Mg²⁺ are only maintained by a very large glutamatergic component that also involves transient discharges of the interneurons. Seizure-like events in 0 Mg²⁺ have significantly higher power in the high gamma frequency band (60-120Hz) than these events do in 4AP, and are greatly delayed in onset by diazepam, unlike 4AP events. We, therefore, conclude that the 0 Mg²⁺ and 4AP models display fundamentally different levels of glutamatergic drive, demonstrating how ostensibly similar pathological discharges can arise from different sources. We contend that similar interpretative issues will also be relevant to clinical practice.

Keywords: Epilepsy; ictal events; interictal events; interneurons.

Figure 1

Similarities between evolving epileptiform activity, in 0 Mg²⁺ and 4AP ACSF. (Ai) Example trace from 0 Mg²⁺‐induced epileptiform activity. (Aii) Equivalent recording in 4AP. (B) The 4AP model leads to SLE significantly faster than the 0 Mg²⁺ model (Mann–Whitney test, **P = 0.0017). (C) There is no difference between the duration of seizures between the 4AP and 0 Mg²⁺ models (Mann–Whitney test, P = 0.1733). (D) There is a significant difference in the number of SLE between the 4AP and 0 Mg²⁺ models (Mann–Whitney test, ***P = 0.0002). (E) The 4AP model enters LRD significantly earlier than the 0 Mg²⁺ model (Mann–Whitney test, **P = 0.005).

Figure 2

Different levels of glutamatergic drive in early interictal‐like events, in the two models. (Ai) Example patch‐clamp recording of typical early interictal‐like events onto a layer 5 pyramidal cell induced by washout of Mg²⁺ ions (Cell is being held at −30 mV). (Aii) Pie chart demonstrating the proportion of putative IPSCs (33.3%, represented with an upward blue arrow), EPSCs (42.3%, represented with a downward pink arrow), and composite events (events containing both a putative IPSC and an EPSC; 24.4%, represented with both an upward blue arrow and a downward pink arrow) of early interictal‐like postsynaptic currents onto pyramidal cells after washout of Mg²⁺ ions (4002 events were analyzed from 10 slices). (Bi) Equivalent recording in 4AP as shown in Ai (Cell is being held at −30 mV). (Bii) Pie chart demonstrating the proportion of putative IPSCs (87.0%, represented with an upward blue arrow), EPSCs (7.8%, represented with a downward pink arrow), and composite events (5.2%, represented with both an upward blue arrow and a downward pink arrow) of early interictal‐like postsynaptic currents onto pyramidal cells after addition of 4AP (9992 events were analyzed from nine slices, Proportion data from 4AP is significantly different from the 0 Mg²⁺ paradigm, χ² test, P < 0.0001).

Figure 3

Different levels of glutamatergic drive during preictal activity, in the two models. (Ai) Example patch‐clamp recording of typical SLE induced by washout of Mg²⁺ ions (Cell is being held at −30 mV). (Aii) Pie chart demonstrating the proportion of putative IPSCs (42.8%, represented with an upward blue arrow), EPSCs (27.0%, represented with a downward pink arrow), and composite events (events containing both a putative IPSC and an EPSC; 30.2%, represented with both an upward blue arrow and a downward pink arrow) of preictal postsynaptic currents onto pyramidal cells after washout of Mg²⁺ (152 events were analyzed from 10 slices). (Bi) Equivalent recordings in 4AP as shown in Ai (Cell is being held at −30 mV). (Bii) Pie chart demonstrating the proportion of putative IPSCs (88.0%, represented with an upward blue arrow), EPSCs (7.0%, represented with a downward pink arrow), and composite events (5.0%, represented with both an upward blue arrow and a downward pink arrow) of preictal postsynaptic currents onto pyramidal cells after addition of 4AP (105 events were analyzed from 9 slices, Proportion data from 4AP is significantly different from the 0 Mg²⁺ paradigm, χ² test, P < 0.0001).

Figure 4

4AP interictal‐like activity is sustained in the presence of glutamatergic blockers but not in the 0 Mg²⁺ paradigm. (A) Representative LFP trace of a 0 Mg²⁺ recording before and after addition of Glu‐X (NBQX and AP5) and Glu‐X plus Gabazine (N = 3). (B) Equivalent recordings in 4AP (N = 3).

Figure 5

4AP preferentially induces increased action potential firing in the PV cell population compared to pyramidal and SST cells. (A) Spontaneous activity: representative traces of a pyramidal cell, a PV‐interneuron, and a SST‐interneuron following wash in of 100 µmol/L 4AP with Glutamate (NBQX and AP5) and GABA (Gabazine and CGP‐55845) receptor blockers (N = 6). (B) Spontaneous action potential firing in the PV cells is significantly higher than in the pyramidal or SST cells (Kruskal–Wallis Test with post hoc test, **P = 0.008 for pyramidal vs PV, *p = 0.042 for SST vs PV). (C) Evoked activity: representative traces showing responses of (Ci) a pyramidal cell, (Cii) a PV interneuron, and (Ciii) a SST interneuron to 3s long hyperpolarisation (−100pA) and depolarization (100pA) current pulses, in the presence of 4AP (GABA and glutamatergic neurotransmission blocked). Scale bars: horizontal, 500 ms; vertical, 10 mV. (D) 4AP increased the number of action potentials fired by PV interneurons (Wilcoxon matched‐pairs rank test, P = 0.008; N = 9) in response to 100pA current injection, but not by pyramidal cells (paired t‐test, P = 0.769; N = 6) or SST interneurons (Wilcoxon matched‐pairs rank test, P = 0.125; N = 6).

Figure 6

4AP affects the firing properties in both PV‐ and SST‐expressing interneurons, and also pyramidal cells. (Ai) 4AP significantly increased the action potential half‐width of pyramidal cells (paired t‐test, P = 0.033; N = 6). Inset: representative pyramidal cell action potential trace before (black) and after (green) 4AP. Scale bars: horizontal, 1 ms; vertical, 20 mV. (Aii) 4AP significantly increased the action potential half‐width of PV interneurons (paired t‐test, P < 0.0001; N = 9). Inset: representative PV interneuron action potential trace before (black) and after (orange) 4AP. Scale bars: horizontal, 1ms; vertical, 20 mV. (Aiii) 4AP significantly increased the action potential half‐width of SST interneurons (paired t‐test, P = 0.002; N = 6). Inset: representative SST interneuron action potential trace before (black) and after (blue) 4AP. Scale bars: horizontal, 1ms; vertical, 20 mV. (Bi) 4AP significantly reduced the action potential threshold of pyramidal cells (paired t‐test, P = 0.023; N = 6). Inset: representative phase plots of pyramidal cell action potential trace before (black) and after (green) 4AP. (Bii) 4AP significantly reduced the action potential threshold of PV interneuron (paired t‐test, P = 0.021; N = 9). Inset: representative phase plots of PV interneuron action potential trace before (black) and after (orange) 4AP. (Biii) 4AP had no effect on the action potential threshold of SST interneuron (paired t‐test, P = 0.159; N = 6). Inset: representative phase plots of SST interneuron action potential trace before (black) and after (blue) 4AP.

Figure 7

PV firing during interictal events is more sustained in 4AP than in 0 Mg²⁺ events. (A) Example paired recordings of pyramidal cells (“Pyr”, recorded in whole cell (“wc”) voltage‐clamp mode at −30 mV) and PV interneurons (recorded in cell‐attached (“ca”) mode) in 0 Mg²⁺. Note the intense bursts of PV firing coincident with upwards deflections of the pyramidal trace, consistent with the PV activity being the source of the GABAergic synaptic barrages. (B) Equivalent example paired recordings in 100 µmol/L 4AP. (C) PV interneurons fire significantly more action potentials per interictal event in 4AP than they do in 0 Mg²⁺ (79 events from six slices analyzed from the 0 Mg²⁺ paradigm while 57 events from four slices were analyzed for the 4AP model; Mann–Whitney test, **P = 0.0095). (D) There is no difference in the PV cell firing rates during interictal events between the models. (E) Duration of interictal events are longer in the 4AP model than in 0 Mg²⁺ model (Mann–Whitney test, **P = 0.0095).

Figure 8

0 Mg²⁺ and 4AP SLEs differ in higher frequency power. 0 Mg²⁺ seizures display more power in the higher frequency range then 4AP seizures (mean summed power 60–120Hz (normalized traces), 0 Mg²⁺ = 7.08e‐7 ± 8.99e‐7; 4‐AP = 1.78e‐7 ± 4.21e‐7; P = 0.0155; 22 and 24 brain slices, respectively). CI, confidence interval.

Figure 9

0 Mg²⁺ model and the 4AP model differ in their sensitivity to diazepam. (A) Representative LFP trace demonstrating the effect of diazepam on 0 Mg²⁺ induced epileptiform activity. (B) Equivalent recording in 4AP. (C) Diazepam significantly delays seizure‐like events in the 0 Mg²⁺ model while it has no effect on 4AP induced seizures (ANOVA with post hoc test, *P = 0.001). (D) Diazepam significantly increases the duration of the early interictal period in the 0 Mg²⁺ model while having no effect in 4AP (ANOVA with post hoc test, *P = 0.001). (E) Diazepam significantly increases the number of early interictal‐like events in the 0 Mg²⁺ model while this remains unchanged in the 4AP (ANOVA with post hoc test, *P = 0.001). (F) Diazepam significantly increases the rate of early interictal‐like events in the 0 Mg²⁺ model while having no effect on the 4AP induced early interictal‐like events (ANOVA with post hoc test, *P = 0.001).