Loss of KCC2 in GABAergic Neurons Causes Seizures and an Imbalance of Cortical Interneurons

Abstract

K-Cl transporter KCC2 is an important regulator of neuronal development and neuronal function at maturity. Through its canonical transporter role, KCC2 maintains inhibitory responses mediated by γ-aminobutyric acid (GABA) type A receptors. During development, late onset of KCC2 transporter activity defines the period when depolarizing GABAergic signals promote a wealth of developmental processes. In addition to its transporter function, KCC2 directly interacts with a number of proteins to regulate dendritic spine formation, cell survival, synaptic plasticity, neuronal excitability, and other processes. Either overexpression or loss of KCC2 can lead to abnormal circuit formation, seizures, or even perinatal death. GABA has been reported to be especially important for driving migration and development of cortical interneurons (IN), and we hypothesized that properly timed onset of KCC2 expression is vital to this process. To test this hypothesis, we created a mouse with conditional knockout of KCC2 in Dlx5-lineage neurons (Dlx5 KCC2 cKO), which targets INs and other post-mitotic GABAergic neurons in the forebrain starting during embryonic development. While KCC2 was first expressed in the INs of layer 5 cortex, perinatal IN migrations and laminar localization appeared to be unaffected by the loss of KCC2. Nonetheless, the mice had early seizures, failure to thrive, and premature death in the second and third weeks of life. At this age, we found an underlying change in IN distribution, including an excess number of somatostatin neurons in layer 5 and a decrease in parvalbumin-expressing neurons in layer 2/3 and layer 6. Our research suggests that while KCC2 expression may not be entirely necessary for early IN migration, loss of KCC2 causes an imbalance in cortical interneuron subtypes, seizures, and early death. More work will be needed to define the specific cellular basis for these findings, including whether they are due to abnormal circuit formation versus the sequela of defective IN inhibition.

Keywords: KCC2; development; excitatory GABA; excitatory/inhibitory balance; interneuron; parvalbumin interneuron; seizure; somatostatin interneuron.

FIGURE 1

Precocious expression of KCC2 in layer 5 INs. (A) KCC2 immunoreactivity in P0 and P13 WT cortex (orange heatmap images) and its overlay with Dlx5:cre tdTomato reporter, which marks cortical interneurons (KCC2 red, Dlx5 reporter green). (B) Magnified areas of P0 cortex indicated by a white rectangle in (A) and matching images of barrel field cortex at P13. (C) Confocal images from layer 5 showing plasmalemmal KCC2 expression in two WT cells expressing a Dlx5 GFP reporter.

FIGURE 2

Loss of KCC2 expression in INs of Dlx5 KCC2 cKO. (A) KCC2 expression in WT (top row) and Dlx5 KCC2 cKO (bottom row) cortical hemisections at P0, P13, and P19. Note lack of KCC2 immunoreactivity in somatosensory cortex (SSC) of Dlx5 KCC2 cKO at P0, but not at later ages. Loss of immunoreactivity persists in striatum (Str), hippocampal hilus (hyl), and reticular nucleus of thalamus (nRT). Regions of interest are outlined by a dashed white line. (B) Close-up images showing loss of plasmalemmal KCC2 immunoreactivity in INs marked by cre tdTomato reporter (KCC2 green, INs red). Arrows point out several examples of KCC2-expressing INs in WT.

FIGURE 3

Interneuronal loss of KCC2 does not alter perinatal IN migration. (A) Distribution of INs, marked by tdTomato cre reporter, in P0 Dlx5 KCC2 cKO and age-matched WT. Top graph compares cortical layers of somatosensory cortex, and bottom compares medial and lateral cortex. Center image shows exemplar distribution of INs in P0 WT cortex with labels of layers/regions used for comparison. (B) Distribution of E13.5-labeled EdU+ INs in P0 Dlx5 KCC2 cKO and age-matched WT. EdU was injected into G13.5 pregnant dams, and distribution of EdU+ INs (shown in center images) was compared between Dlx5 KCC2 cKO and WT in cortical layers of somatosensory cortex (top graph), and between lateral and medial cortex (bottom graph). For both (A,B), no significant differences found by one-way ANOVA with Sidak’s multiple comparisons. Comparisons were also made between individual layers of medial and lateral cortex with no differences found (data not shown). For (A,B), n = 13 WT and 5 Dlx5 KCC2 cKO mice. Abbreviations: (MZ), marginal zone; (L2-4), layers 2-4; (L5), layer 5; (L6), layer 6; (layers 2-6), L2-6; (SP), subplate; (WM), white matter.

FIGURE 4

Dlx5 KCC2 cKO mice exhibit failure to thrive and seizures. (A–C) Graphs showing that Dlx5 KCC2 cKO pups are born at expected 25% Mendelian ratio (A), but have decreased weight and length (B), and die in the postnatal period (C). One-way ANOVA with Sidak’s multiple comparisons: *P < 0.05, ****P < 0.0001. For body length, paired comparisons against an average of matched littermate lengths were done using a ratio paired T-test, ****P < 0.0001. (D) Mild hypoglycemia in Dlx5 KCC2 cKO pups. T-test: ***P < 0.001. (E) Susceptibility of Dlx5 KCC2 cKO pups to faster onset of flurothyl-induced generalized tonic clonic seizures (GTC) at P4–P5 and P10. Kruskal–Wallis non-parametric ANOVA: **P < 0.01. The number of mice examined is indicated by n values in each panel. For bar graphs, these are listed at the bottom of each bar.

FIGURE 5

Increase in Layer 5 INs in P12–14 Dlx5 KCC2 cKO Cortex. Bar graph shows densities of INs, labeled by cre reporter, by different layers of barrel field cortex in P12–14 WT, heterozygote, and Dlx5 KCC2 cKO. Images show examples of IN reporter expression in WT and Dlx5 KCC2 cKO cortex. One-way ANOVA with Sidak’s multiple comparisons: **P < 0.01. n = 10 WT, 5 Dlx5 KCC2 cKO mice.

FIGURE 6

Normal sIPSC frequency, faster decay in pyramidal neurons in KO. (A) Sample recordings of spontaneous IPSCs in layer 5 pyramidal neurons in barrel cortex of P12–14 WT and Dlx5 KCC2 cKO. (B) Amplitude, area, decay (mean ± SEM), and inter-event interval (IEI, box and whisker plot showing median, 25/75 percentile, and maximum/minimum) of spontaneous IPSCs in layer 5 pyramidal neurons in barrel cortex of P12–14 WT and Dlx5 KCC2 cKO. T-test: *P < 0.05. Non-parametric Mann–Whitney test was used for IEI only. n values at the bottom of each bar/box and whisker plot indicate the number of cells recorded, taken from ≥5 mice, for each group.

FIGURE 7

Altered distribution of somatostatin and parvalbumin INs in maturing cortex of Dlx5 KCC2 cKO. (A) Images showing complete co-localization of SST (left, green) and PV (right, green) immunopositive neurons with Ai14 cre reporter used to label INs (red) in P13 barrel field cortex. (B) Average density with SEM of SST+ cells in different layers of P12–14 somatosensory cortex with exemplary images of SST immunolabeling. Yellow dashed line demarcates layer 5. n = 14 WT, 12 Dlx5 KCC2 cKO mice. One-way ANOVA with Sidak’s multiple comparisons: **P < 0.01. (C) Paired comparisons of density of PV+ cells in layers 2–4 (L2–4), 5 (L5), and 6 (L6) in P12–14 somatosensory cortex of Dlx5 KCC2 cKO and control littermates, with exemplary images of PV immunolabeling. Box and whisker plots in each graph plot percent difference change in each littermate pair, n = 12 littermate mouse pairs. Wilcoxon matched-pairs signed rank test, **P < 0.01, *P < 0.05. (D) Survival plot of Nkx2.1 KCC2 knockout, conditional only to PV and SST INs. n = 46 WT, 15 heterozygote, 26 Nkx2.1 KCC2 knockout mice.

FIGURE 8

Selective loss of KCC2 in parvalbumin neurons does not cause seizure susceptibility nor failure to thrive. (A) PV INs labeled by Ai14 cre reporter in hemisections of P64 PV KCC2 cKO and sibling control. (B) Body weight of PV KCC2 cKO. One-way ANOVA with Sidak’s multiple comparisons: ****P < 0.0001. Note normal PV KCC2 cKO body weight until adulthood, when it is slightly diminished. (C) Postnatal survival is unaffected in PV KCC2 cKO. (D) Seizure susceptibility (right) measured by time to first myoclonic jerk and generalized tonic clonic seizure (left) and seizure mortality (right) resulting from flurothyl seizure induction in PV KCC2 cKO mice. Note that PV KCC2 cKO do not exhibit faster onset of induced seizures, but are highly likely to die from them. Chi squared with Yates’ correction: *P < 0.05. The number of mice examined is indicated by n values in each panel. For bar graphs, these are listed at the bottom of each bar.