Pannexin1 Mediates Early-Life Seizure-Induced Social Behavior Deficits

Abstract

There is a high co-morbidity between childhood epilepsy and autism spectrum disorder (ASD), with age of seizure onset being a critical determinant of behavioral outcomes. The interplay between these comorbidities has been investigated in animal models with results showing that the induction of seizures at early post-natal ages leads to learning and memory deficits and to autistic-like behavior in adulthood. Modifications of the excitation/inhibition (glutamate/GABA, ATP/adenosine) balance that follows early-life seizures (ELS) are thought to be the physiological events that underlie neuropsychiatric and neurodevelopmental disorders. Although alterations in purinergic/adenosinergic signaling have been implicated in seizures and ASD, it is unknown whether the ATP release channels, Pannexin1 (Panx1), contribute to ELS-induced behavior changes. To tackle this question, we used the ELS-kainic acid model in transgenic mice with global and cell type specific deletion of Panx1 to evaluate whether these channels were involved in behavioral deficits that occur later in life. Our studies show that ELS results in Panx1 dependent social behavior deficits and also in poor performance in a spatial memory test that does not involve Panx1. These findings provide support for a link between ELS and adult behavioral deficits. Moreover, we identify neuronal and not astrocyte Panx1 as a potential target to specifically limit astrogliosis and social behavioral deficits resultant from early-life seizures.

Keywords: astrocytes; ion channel; purinergic signaling; sociability; spatial memory; status epilepticus.

Figure 1.

Outcomes of KA injection in P21 mice. (A) Number of Panx1^f/f (FF), global Panx1 knockout (KO), NFH-cre:Panx1^f/f (NFH), and GFAP-cre:Panx1^f/f (GFAP) mice that developed status epilepticus (SE; black bars), that did not developed status (NO SE, dark gray bars), and that died (DEAD, light gray bars) after i.p. injection of KA. Inset in (A) shows the normalization (fraction of mice) of the outcomes per genotype. (B) Violin plot showing the median and quartile values of the time that Panx1^f/f (FF; n = 53), global Panx1 knockout (KO; n = 50), NFH-Cre:Panx1^f/f (NFH; n = 18), and GFAP-Cre:Panx1^f/f (GFAP; n = 22) mice spent in SE. ***p = 0.001, ns = not significant (Kruskal-Wallis followed by Dunn’s multiple comparison test).

Figure 2.

Panx1 and early life seizure do not affect locomotor activity and anxiety-like behavior. Violin plots showing the median values of the (A) total distance travelled, (B) velocity, (C) number of ambulatory episodes, and (D) time spent in the outer zone of the open field arena obtained for 2 months old Panx1^f/f (FF), global Panx1 knockout (KO), NFH-Cre:Panx1^f/f (NFH), and GFAP-Cre:Panx1^f/f (GFAP) mice that were injected with saline (Sal) and kainic acid (KA) at age P21. KA groups correspond to mice that developed SE. **p = 0.001, ns = not significant (Kruskal-Wallis followed by Dunn’s test). Number of mice: FF_sal (30), FF_KA (18), KO_sal (15), KO_KA (10), NFH_sal (14), NFH_KA (15), GFAP_sal (22), GFAP_KA (14).

Figure 3.

Panx1 dependence of social behavior deficits induced by early life seizures. (A) Means ± sem values of the time that 2 months old Panx1^f/f (FF), global Panx1 knockout (KO), NFH-Cre:Panx1^f/f (NFH), and GFAP-Cre:Panx1^f/f (GFAP) mice, injected with saline (Sal) and kainic acid (KA) at age P21, spent with an object (O) and with a conspecific (A) measured during the three-chamber test. ****p < 0.0001, ***p = 0.0001, **p = 0.007, ns = not significant (two ways ANOVA followed by šidák’s multi-comparison test). (B) Means ± sem values of the sociability (percent time spent with conspecific to the total time spent with object and conspecific) obtained from 17 FF_sal, 11 FF_KA, 12 KO_sal, 10 KO_KA, 10 NFH_sal, 8 NFH_KA, 13 GFAP_sal, 11 GFAP_KA mice. ****p < 0.0001, ***p = 0.0003 (one sample t-test).

Figure 4.

Stereotypic behavior is not altered by Panx1 and early-life seizures. Means ± sem values of the (A) fraction of buried marbles and (B) number of stereotypic counts measured from 2 months old Panx1^f/f (FF), global Panx1 knockout (KO), NFH-Cre:Panx1^f/f (NFH), and GFAP-Cre:Panx1^f/f (GFAP) mice, injected with saline (Sal) and kainic acid (KA) at age P21. In part A, values were obtained from 30 FF_sal, 17 FF_KA, 15 KO_sal, 13 KO_KA, 10 NFH_sal, 15 NFH_KA, 16 GFAP_sal, 9 GFAP_KA mice. In part B, values were obtained from 19 FF_sal, 16FF_KA, 15 KO_sal, 14 KO_KA, 14 NFH_sal, 14 NFH_KA, 22 GFAP_sal, 9 GFAP_KA mice. ns = not significant (two ways ANOVA followed by Šidák’s multiple comparison test).

Figure 5.

Long-term reference memory impairment. (A1–D1) Mean ± sem of the (A1) fraction of reference memory errors (RME), (B1) correct choices (correct arms), (C1) working memory incorrect errors (WM-I), and (D1) working memory correct errors (WM-C) obtained during the training sessions. (A2 - D2) Violin plot showing the median values of RME (a), correct arms (B), WM-C (D), and WM-I (C) obtained during the test phase for 2 months old mice that were saline (Sal) and kainic acid (KA) injected at age P21. Panx1^f/f (FF_sal, n = 15; FF_KA, n = 21) and global Panx1 knockout (KO_sal, n = 21; KO_KA, n = 19) mice. ns: not significant, *p < 0.03, **p = 0.007, ***p = 0.0005 (Kruskal-Wallis followed by Dunn’s multiple comparison tests).

Figure 6.

Increased GFAP expression levels following ELS is abrogated in mice lacking neuronal Panx1. (A) Mean ± sem of GFAP expression measured from whole brain homogenates of 2-months old control (saline-injected at P21: Sal) Panx1^f/f (FF), global Panx1 knockout (KO), GFAP-Cre:Panx1^f/f (GFAP), and NFH-Cre:Panx1^f/f (NFH) mice. *p = 0.0172, ***p = 0.0007, (one-way ANOVA followed by Dunnett’s multiple comparison tests, n = 4 mice per genotype). (B) Mean ± sem of the fractional changes in GFAP expression levels measured from whole brains of 2-months old mice that were injected at P21 with kainic acid (KA) relative to saline-injected (Sal) mice. ns = not significant, *p = 0.03, **p = 0.006, (one-way ANOVA followed by Šidák’s multi-comparison test; n = 4 mice per group). Top panel: Representative Western blots showing GFAP and β-tubulin expression levels in whole brains of control (Sal) and ELS (KA) mice.