Pyramidal cells accumulate chloride at seizure onset

Abstract

Seizures are thought to originate from a failure of inhibition to quell hyperactive neural circuits, but the nature of this failure remains unknown. Here we combine high-speed two-photon imaging with electrophysiological recordings to directly evaluate the interaction between populations of interneurons and principal cells during the onset of seizure-like activity in mouse hippocampal slices. Both calcium imaging and dual patch clamp recordings reveal that in vitro seizure-like events (SLEs) are preceded by pre-ictal bursts of activity in which interneurons predominate. Corresponding changes in intracellular chloride concentration were observed in pyramidal cells using the chloride indicator Clomeleon. These changes were measurable at SLE onset and became very large during the SLE. Pharmacological manipulation of GABAergic transmission, either by blocking GABA(A) receptors or by hyperpolarizing the GABA(A) reversal potential, converted SLEs to short interictal-like bursts. Together, our results support a model in which pre-ictal GABA(A) receptor-mediated chloride influx shifts E(GABA) to produce a positive feedback loop that contributes to the initiation of seizure activity.

Figure 1. Targeted Path Scanning of interneurons and principle cells

A) Using two-photon TPS, a laser path (yellow dotted line) is selected that includes cells stained with Indo-1 only (principal cells, green) and cells that express GFP and are stained with Indo-1 (interneurons, pink). B) 50 µM 4-AP ACSF produces SLEs that have a characteristic pattern (mean calcium: black, LFP: blue) beginning with a pre-ictal burst, followed by the tonic phase of the SLE, and finishing with clonic discharges. C) The delay between the pre-ictal burst and the onset of the tonic phase of the SLE varied within a range of 0–5s. D) A sample of 7 interneuron (red) and 7 principal cell (blue) calcium traces from a single pre-ictal burst (at t=0) in a single slice suggests that calcium transients are larger in interneurons. E) Indeed pooling calcium data across all recorded pre-ictal bursts (n=22 SLEs, 533 principal cells, 115 interneurons) reveals a significant shift in the balance of interneuron vs. principal cell calcium concentration (p<0.01, green asterisks indicate time points at which I-E ratio is significantly different from baseline).

Figure 2. Interneurons fire at higher rates than principal cells at ictogenesis

Using both whole-cell and loose-patch clamp recordings (A and B, respectively), we observed (insets) uncorrelated interneuron (red) and principal cell (blue) firing before SLE onset and precise I-before-E firing immediately following the interneuron dominated pre-ictal burst (PB). C) Instantaneous spike rates calculated at the time of the pre-ictal burst (t=0 is time of pre-ictal burst onset, shaded regions include mean ± SEM) further validate that the elevated interneuron calcium levels observed in Figure 1 are a result of an increase in interneuron firing rate.

Figure 3. Intracellular chloride is elevated during ictogenesis

A) At SLE onset there was a sharp decrease in clomeleon ratio (YFP/CFP), corresponding to an increase in intracellular chloride. B) Blocking GABA_A with 10uM GABAzine eliminated SLEs and the corresponding high-amplitude chloride transients, leaving interictal-like discharges at a rate of ~0.1Hz. C,D) Acute slices treated with 4-AP and zero-Mg²⁺ ACSF produced similar results to those observed in A,B, indicating that the effect is not specific to chronically epileptic tissue. E) Aligning chloride measurements at the time of the pre-ictal bursts reveals a pre-ictal burst-evoked increase in intracellular chloride. The mean chloride concentration at the time of SLE onset (indicated by * for each of the 8 SLEs) was 6.27±1.23mM above baseline. A,B, and insets share a scale bar. C and D also share a scalebar.

Figure 4. GABA becomes depolarizing during seizure, is necessary for ictogenesis

A) During SLE, chloride increased by 22.95±0.73mM, corresponding to a calculated 26±3mV depolarization of the GABA_A receptor reversal potential. B) GABA_A blockade eliminated SLEs, leaving burst-evoked chloride peaks of 2.97±0.07mM. C) Under control conditions, organotypic slices produced bouts of electrical activity of short duration, corresponding to interictal bursting and of long duration, corresponding to SLEs, while D) GABAzine eliminated prolonged SLEs, leaving behind only short duration events. Log-spaced bins (but not axes) were used for these histograms to highlight the mixture of many short-duration events and occasional long-duration events.

Figure 5. Ictal pH changes

Organotypic slices prepared from wild-type mice were stained with the ratiometric pH indicator SNARF-1 AM. Spontaneous SLEs were apparent (red bar) in a synchronously recorded field potential (green). Both the polarity (alkali ictal transient) and low amplitude (0.080 ± 0.025, mean±SEM, n=3 slices) of observed pH changes (blue) were insufficient to account for changes observed in intracellular chloride using Clomeleon.

Figure 6. Bicarbonate current contributes to ictogenesis

A) In 4-AP-treated acute slices from GIN mice, pharmacologically hyperpolarizing the GABA_A reversal potential by blocking intracellular bicarbonate production with acetazolamide (ACTZ) eliminates SLEs, leaving only interictal bursts. Upon washout of ACTZ, SLEs return. B) ACTZ reversibly eliminates SLEs in 4-AP (n = 11, ** indicate that the number of SLEs is significantly different from zero, p<0.05).