GABAergic disinhibition and impaired KCC2 cotransporter activity underlie tumor-associated epilepsy

Abstract

Seizures frequently accompany gliomas and often escalate to peritumoral epilepsy. Previous work revealed the importance of tumor-derived excitatory glutamate (Glu) release mediated by the cystine-glutamate transporter (SXC) in epileptogenesis. We now show a novel contribution of GABAergic disinhibition to disease pathophysiology. In a validated mouse glioma model, we found that peritumoral parvalbumin-positive GABAergic inhibitory interneurons are significantly reduced, corresponding with deficits in spontaneous and evoked inhibitory neurotransmission. Most remaining peritumoral neurons exhibit elevated intracellular Cl(-) concentration ([Cl(-) ]i ) and consequently depolarizing, excitatory gamma-aminobutyric acid (GABA) responses. In these neurons, the plasmalemmal expression of KCC2, which establishes the low [Cl(-) ]i required for GABAA R-mediated inhibition, is significantly decreased. Interestingly, reductions in inhibition are independent of Glu release, but the presence of both decreased inhibition and decreased SXC expression is required for epileptogenesis. We suggest GABAergic disinhibition renders peritumoral neuronal networks hyper-excitable and susceptible to seizures triggered by excitatory stimuli, and propose KCC2 as a therapeutic target.

Keywords: GABA; KCC2; glioma; peritumoral epilepsy.

Figure 1

xCT expression in several patient-derived xenograph tissue samples and seizure activity from GBM22-implanted tumor animals (A). Patient tissue samples show tumors with either high or low xCT expression. (B) cresyl violet staining of a GBM22 tumor slice, showing a large tumor mass in the cortex. scale bar: 200 μm. (right) DIC image of a distinctive mass of GBM22 tumor cells in a cortical slice. Scale bar: 5 μm (C) EEG recordings from a sham and GBM22 tumor animal (D), showing a pronounced spontaneous seizure event lasting over 30 s.

Figure 2

Peritumoral neurons display hyperexcitability. (A) Representative recordings in ACSF of sPSCs in sham (top) and spontaneous epileptiform event observed in tumor-bearing slice (bottom). Inset below tumor trace shows two spontaneously occurring epileptiform events in tumor-bearing slice on an expanded time scale. Spontaneous epileptiform events were never observed in recordings from sham slices. (B) Representative recordings from sham (left) and tumor (right) slices, showing the latency to (10 μM) Bic-induced epileptiform activity was significantly shorter in tumor-bearing slices. Inset below sham and tumor traces shows inward currents on an expanded time scale. (C) Summary of the percent of slices showing epileptiform activity in sham and tumor-bearing slices. (D) Summary of the latency to Bic-induced epileptiform events in sham and tumor-bearing slices.

Figure 3

Decreased inhibition in peritumoral cortex. (A) Cortical PV-ir shown at low magnification (top) of a sham control and tumor-bearing slice. Scale bar, 50 μm. (Bottom) Higher magnification showing round shape individual PV-positive neurons from sham and longer flattened PV-neurons from peritumoral cortex. Scale bar, 10 μm. (B) Quantification of PV cell loss in layer 2/3 of the peritumoral cortex. Cells were counted ~100 μm from tumor edge. (C) Sample recordings of sIPSCs and mIPSCs (F) from sham (left) and tumor-associated cells (right) recorded from layer II/III pyramidal cells. Lower traces show individual events on an expanded timescale. (D) Summary of the frequency of sIPSC and mIPSC (G) and amplitude of sIPSC (E) and mIPSC (H). (I) Pooled data of the mean threshold stimulus intensity required to evoke IPSCs in sham and peritumoral neurons. (J) Stimulus response curve of the peak amplitude plotted versus stimulation intensities from neurons in sham and tumor-bearing slices. Representative IPSC traces evoked in sham (top) and tumor-bearing slices (bottom) in response to increasing stimulation intensities.

Figure 4

Depolarized GABA response and decreased KCC2 expression in tumor-bearing cortex. Sample recordings of membrane potential responses to 10 μM GABA from sham (A) and (B) tumor-bearing slices. (C) Scatter plot showing the mean E_GABA values measured in identified layer II/III pyramidal cells from sham and tumor-bearing slices. The E_GABA values for tumor-bearing slices were more depolarized compared to shams. (D) Summary of E_Cl values from neurons in sham and tumor-bearing slices. (E) Western blot showing a decrease in KCC2 expression in the peritumoral (PT) cortex compared to contralateral (C) cortex. (F) Quantification of KCC2 protein expression, c refers to control cortex in F and G. (G) Western blot of NKCC1 expression in peritumoral and contralateral cortex. (H) Quantification of NKCC1 showing no significant change in the peritumoral cortex.

Figure 5

Alteration in KCC2 expression in PV- and NeuN-positive peritumoral cells. (A,B) Confocal images from sham slices labeled for KCC2 (green), PV (red) and merged images (right). Merged images show colocalization of KCC2 on PV-positive cells at the periphery of the soma from a sham slice. (C,D) same as A and B, but in a tumor-bearing slice showing a decrease in the expression of KCC2 on the periphery of cell body. (I and J) Confocal images from sham slices labeled for KCC2 (green), NeuN (red). Merged image show colocalization of KCC2 on NeuN-positive cells at the periphery of the soma from a sham slice. (E, M) Summary data of the relative fluorescent intensity of KCC2 staining in the soma of PV-positive and NeuN-positive cells from sham and tumor-bearing animals, respectively. (G, O) Percent of the cell volume colocalization of PV and KCC2 (G) and NeuN and KCC2 (O) in tumor-bearing and sham slices. Line scanning across cell soma of a PV- positive cell showing KCC2 expression in a sham (F) and a peritumoral cell (H). Line scan of the relative fluorescence intensity of NeuN and KCC2 expression across a sham (N) and peritumoral neuron (P). The yellow line indicates the scanned line. Scale bars: 20 μm (A, C, I, and K) and 5 μm (B, D, J and L). The yellow line indicates the scanned line.

Figure 6

Low SXC expressing GBM14-implanted tumors present with depolarized GABA responses and hyperpolarized E_GABA. (A) Sample EEG recording from a GBM14 tumor-implanted mouse, which have a low expression of SXC, showing interictal spikes. These tumor-bearing animals do not display long-lasting seizure events. (B) Scatter plot of E_GABA values measured in peritumoral neurons from GBM14-implanted animals. (C) Sample recording of sPSCs from a GBM22 peritumoral neuron showing epileptiform events in the presence of Bic (10 μM). Application of SAS (250 μM) blocked the epileptiform events after 25 mins. (D) Summary of the frequency and duration (E) of epileptiform events in the presence of Bic and following SAS application.

Figure 7

Decreasing KCC2 expression in glioma correlates with reduced patient survival. The Rembrandt database was used to perform a Kaplan-Meier survival analysis of glioma patients with decrease KCC2 expression. Down-regulation was defined as a 2-fold or greater deviation from intermediate expression. Among glioma patients (n = 413) there was a significant difference in survival between patients with decreased KCC2 gene expression (n = 297) compared to patients with no change in KCC2 expression (n = 116) (log-rank test; P = 1.0^e-5).