Grafted hPSC-derived GABA-ergic interneurons regulate seizures and specific cognitive function in temporal lobe epilepsy

Abstract

Interneuron loss/dysfunction contributes to spontaneous recurrent seizures (SRS) in chronic temporal lobe epilepsy (TLE), and interneuron grafting into the epileptic hippocampus reduces SRS and improves cognitive function. This study investigated whether graft-derived gamma-aminobutyric acid positive (GABA-ergic) interneurons directly regulate SRS and cognitive function in a rat model of chronic TLE. Human pluripotent stem cell-derived medial ganglionic eminence-like GABA-ergic progenitors, engineered to express hM4D(Gi), a designer receptor exclusively activated by designer drugs (DREADDs) through CRISPR/Cas9 technology, were grafted into hippocampi of chronically epileptic rats to facilitate the subsequent silencing of graft-derived interneurons. Such grafting substantially reduced SRS and improved hippocampus-dependent cognitive function. Remarkably, silencing of graft-derived interneurons with a designer drug increased SRS and induced location memory impairment but did not affect pattern separation function. Deactivation of DREADDs restored both SRS control and object location memory function. Thus, transplanted GABA-ergic interneurons could directly regulate SRS and specific cognitive functions in TLE.

Fig. 1. The overall experimental design and the sequence of experiments performed in the study.

Status epilepticus (SE) was induced in two-month-old male F344 rats through graded kainic acid injections. Two months later, animals exhibiting chronic temporal lobe epilepsy (TLE) typified by spontaneous recurrent seizures (SRS) were chosen for bilateral grafting of human medial ganglionic eminence (hMGE)-like cells. The hMGE cells were generated from hM4Di-receptor (a designer receptor activated only by designer drugs, DREADDs) expressing human embryonic stem cell (hESC)-line through a directed differentiation protocol. In the fourth month after grafting, rats were implanted with electrodes, and 2 weeks later, continuous video-encephalographic (video-EEG) recordings were taken before (days 1–5), during (days 6–8), and 2 days after (days 11–14) clozapine-N-oxide (CNO) treatment. Seven days after completing video-EEG recordings, the animals were probed with the object location tests (OLTs), with and without CNO administration, to assess hippocampus-dependent cognitive function. Seven days after completing the second OLT, the animals were investigated with the pattern separation tests (PSTs), with and without CNO administration, to examine their pattern separation ability. Following behavioral tests, the animals were perfused, and fixed brain tissues were processed to analyze the differentiation and integration of graft-derived cells through dual and triple immunofluorescence methods and confocal imaging.

Fig. 2. Evaluation of the effects of grafting human medial ganglionic eminence (hMGE) progenitor cells expressing the Gi-protein-coupled receptor hM4Di into the hippocampus of chronically epileptic rats (CERs) on spontaneous recurrent seizure (SRS) activity.

Quantification in the 4th month after grafting via continuous video-EEG recordings revealed that compared to the group of CERs receiving no grafts, the group of CERs receiving hMGE cell grafts displayed greatly decreased frequencies of all SRS (a) and stage V SRS (b). The grafted animals also spent much less time in seizure activity (c). d–f illustrate electroencephalographic (EEG) traces during the pre-clozapine-N-oxide (CNO), CNO, and post-CNO periods. Values in bar charts are presented as mean ± S.E.M. ****p < 0.0001 (unpaired, two-tailed Student’s t test).

Fig. 3. Effects of silencing graft-derived neurons with clozapine-N-oxide (CNO) on spontaneous recurrent seizure (SRS) activity in chronically epileptic rats.

The bar charts a–c compare all SRS and stage V SRS frequencies and times spent in SRS activity (% of recorded time) during pre-CNO, CNO, and post-CNO periods. The bar charts d–f compare all SRS and stage V SRS frequencies and times spent in SRS activity during the pre-CNO (days 1–5), CNO (days 6–8), and post-CNO (days 11–14) periods. The bar chart g compares the average electroencephalographic (EEG) power (i.e., spectral density) recorded in interictal periods during pre-CNO, CNO, and post-CNO phases. Values in bar charts are presented as mean ± S.E.M. *p < 0.05; **p < 0.01; NS, non-significant (one-way ANOVA with Newman–Keuls multiple comparisons test).

Fig. 4. Effects of clozapine-N-oxide (CNO) on spontaneous recurrent seizure (SRS) activity in chronically epileptic rats receiving no grafts.

The bar charts a–c compare all SRS and stage V SRS frequencies and times spent in SRS activity (% of recorded time) during the pre-CNO, CNO, and post-CNO periods in CERs receiving no grafts. Values in bar charts are presented as mean ± S.E.M. NS, non-significant (one-way ANOVA with Newman–Keuls multiple comparisons test).

Fig. 5. Effects of grafting human medial ganglionic eminence (hMGE) progenitor cells expressing the Gi-protein-coupled receptor hM4Di into the hippocampus of chronically epileptic rats (CERs) on cognitive function.

a depicts the various trials involved in an object location test (OLT). The bar charts in b–e compare percentages of time spent with the object in the familiar place (OIFP) and the object in the novel place (OINP) in naive control rats (b), chronically epileptic rats (CERs; c), and CERs with hMGE grafts before and during the clozapine-N-oxide (CNO) treatment (d, e). The bar chart in f compares the time spent with the OINP across the four groups with ANOVA. Object location memory was impaired in CERs with no grafts and CERs with grafts when graft-derived interneurons were silenced. g shows the various trials involved in a pattern separation test (PST). The bar charts in h–k compare percentages of time spent with the familiar object on pattern 2 (FO on P2) and the novel object on pattern 2 (NO on P2) in naive control rats (h), CERs (i), and CERs with hMGE grafts before and during the clozapine-N-oxide CNO treatment (j, k). The bar chart in l compares the time spent with the NO on P2 across the four groups with ANOVA. Note that pattern separation ability was impaired in CERs with no grafts. However, CERs with grafts displayed pattern separation ability even when the graft-derived interneurons were silenced with CNO. Values in bar charts are presented as mean ± S.E.M. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, NS, non-significant (b–e, h–k, unpaired, two-tailed Student’s t test; f, l, one-way ANOVA with Newman–Keuls multiple comparisons test).

Fig. 6. Evaluation of Gi-protein-coupled receptor hM4Di expression (with mCherry reporter) in human nuclear antigen-positive (HNA+) cells, neuron-specific nuclear antigen-positive (NeuN+) neurons, and gamma-aminobutyric acid-positive (GABA+) interneurons derived from human medial ganglionic eminence progenitor cell grafts in the hippocampus of chronically epileptic rats.

Note that mCherry is displayed in virtually all HNA+ graft-derived cells (a–c), NeuN+ neurons (d–f), and GABA-ergic interneurons (j–l). g–i demonstrate that a vast majority (mean ± S.E.M, 80.8 ± 1.1%) of HNA+ graft-derived cells differentiated into GABA-ergic interneurons. Scale bars: a–l, 20 µm.

Fig. 7. Graft-derived interneurons expressed hM4Di and formed putative synapses with host neurons.

Gi-protein-coupled receptor hM4Di expression (with mCherry reporter) in parvalbumin (PV) and neuropeptide Y (NPY) expressing interneurons derived from human medial ganglionic eminence progenitor cell grafts in the hippocampus of chronically epileptic rats (a–f), and putative synapse formation between graft-derived axons and host neurons (g–p). Note that mCherry is apparent in PV and NPY+ interneurons derived from graft-derived cells (a–f). g, l illustrate putative synapse formation between graft-derived presynaptic boutons (green colored structures expressing human synaptophysin (hSyn) and the host postsynaptic density protein 95 (PSD95, red particles) elements on microtubule-associated protein-2 (MAP-2) positive dendrites (blue) in the host CA1 stratum radiatum (g) and the dentate gyrus molecular layer (l). h, m are magnified views of boxed regions in g, l, respectively. i–k, n–p illustrate MAP-2, hSyn, and PSD95 elements in red, green, and blue channels. Scale bars: a–f, 20 µm; g, l, 5 µm; i–k, n–p, 0.5 µm.