Hippocampal P2X7 and A2A purinoceptors mediate cognitive impairment caused by long-lasting epileptic seizures

Abstract

Rationale: Cognitive impairment and depression are salient comorbidities of mesial temporal lobe epilepsy; it is still unclear whether this frequently drug resistant disease is a cause or consequence of hippocampal damage and its interplay with long-lasting seizure activity (status epilepticus; SE). Thus, a major therapeutic advance in this field is badly needed. Methods: We modeled enduring behavioral and electroencephalographic (EEG) seizures in mice by the intraperitoneal injection of kainic acid (KA), and measured the dynamics of the intracellular Ca²⁺ signals in the hippocampal CA1 area by fiber photometry. Learning and memory were controlled by the Morris Water-Maze and Novel Object Recognition tests on whole animals and by the induction of long-term potentiation in CA1 pyramidal neurons in brain slices. Depressive-like reactions were evaluated by the Tail Suspension, Forced Swim, and Sucrose Preference tests. Results: The intraperitoneal injection of the blood-brain permeable, highly selective, P2X7 and A2A receptor (R) antagonists, JNJ-47965567, and KW6002/SCH58261, respectively, counteracted the effects of KA-induced SE both on seizure activity and the increase of Ca²⁺ signals (as a measure of changes in the intracellular Ca²⁺ concentration) in neurons and astrocytes of the hippocampal CA1 area. In addition, these drugs also prevented the impairment of the hippocampus-dependent spatial and non-spatial learning abilities by KA-SE. The knockdown of P2X7Rs in CA1 astrocytes, but not neurons prevented the cognitive deterioration, suggesting that the release of astrocytic signaling molecules onto neighboring neurons might be the cause of this effect. In accordance with our observations, in hippocampal slices prepared from mice which underwent KA-SE, a selective sensitivity increases to the prototypic P2X7R agonist dibenzoyl-ATP (Bz-ATP) manifested in CA1 neurons. This sensitivity increase appeared to be due to a postsynaptic interference between P2X7Rs and the release of excitatory neurotransmitters during SE. In spite of a P2X7 and A2AR-mediated increase of Ca²⁺ signaling in the medial prefrontal cortex, no similar change was noted after KA-SE in depressive-like reactions or the open-field behavior. Conclusions: SE induced the release of ATP and adenosine from the hippocampus and in consequence decreased the cognitive abilities of mice. The pharmacological blockade of P2X7 and A2ARs prevented the SE-induced seizure activity and cognitive deterioration, but not depressive-like behavior.

Keywords: A2A receptors; ATP; P2X7 receptors; adenosine; hippocampus; pharmacological antagonists; status epilepticus.

Figure 1

Seizure stages on the Racine scale in mice during the first 1 h after the injection of kainic acid (30 mg/kg) or its application together with a preceding injection of JNJ-47965567 (30 mg/kg), KW6002, or SCH58261 (3 mg/kg each), all via the i.p. route. While the curve of the seizure stage was displaced to the right by JNJ-47965567, it was unchanged by KW6002 and SCH58261.The Mann-Whitney test was used for statistical evaluation; the levels of significance reached were *P < 0.05, **P < 0.01 (A, 30-40 min, U = 3.500, P = 0.0202; B, 40-50 min, U = 6, P = 0.0204; D, 30-40 min, U = 15, P = 0.2692; G, 40-50 min, U = 17, P = 04371). In addition, we also calculated the onset until SE (stages 4-5) and the duration of SE, which were in all cases modified by JNJ-47965567, KW6002, and SCH58261. The statistical evaluation with the unpaired t-test yielded the following results: *P < 0.05, **P < 0.01, *P < 0.001 (B, t = 7.475; C, t = 3.093; E, t = 6.483; F, t = 5.409; H, t = 3.277; I, t = 5.734 (n = 7 each).

Figure 2

Effect of kainic acid (KA) injection alone or in combination with the P2X7R antagonist JNJ-47965567 on the intracellular concentration of Ca²⁺ in CA1 neurons of the mouse hippocampus. Two weeks before KA application (30 mg/kg, i.p.), the virus complex rAAV-hSyn-GCaMP6f-EGFP generating specifically in neurons a genetic Ca²⁺ indicator was microinjected into the hippocampus. JNJ-47965567 (30 mg/kg, i.p.) was injected 1 h before KA. The Ca²⁺ changes were measured with fiber photometry and expressed in ΔF/F₀; gross electrical activity of the brain was measured with EEG telemetry and presented in µV amplitude signals (left panels). The ΔF/F₀ fluorescence curves are shown as mean ± S.E.M. of the recordings occurring during the seizure stages 4 and 5. KA was dissolved in saline, while JNJ-47965567 was dissolved in 30% SBE-β-CD + 70% saline (the solvent of JNJ-4438079 is indicated in the Figure panels as saline for the sake of simplicity). All details of the procedures used are described in the Methods Section. The heatmaps of the Ca²⁺ and EEG recordings are shown in the middle panels, while the right panels show the area under the curve (AUC) of the ΔF/F₀ changes and the total power of the EEG as n-time changes from the baseline. The levels of statistical significance reached are marked with *P < 0.05, **P < 0.01, and ***P < 0.001. The unpaired t-test was used for evaluation in the case of normal distribution of data (A, t = 10.00; C, t = 2,685; D, t = 6,517; E, t = 11.69; F, t = 8.031; G, t = 1.189; H, t = 8.577). The Mann-Whitney test was used for evaluation in panel B because of non-normal distribution of the data (U = 0). The number of experiments was 5 in each panel.

Figure 3

Effect of kainic acid (KA) injection alone or in combination with the P2X7R antagonist JNJ-4438079 on Ca²⁺ signals in CA1 astrocytes of the mouse hippocampus. Two weeks before KA application (30 mg/kg, i.p.), the virus complex rAAV-GfaABC1D-GCaMP6f-EGFP generating specifically in astrocytes a genetic Ca²⁺ indicator was microinjected into the hippocampus. JNJ-47965567 (30 mg/kg, i.p.) was injected 1 h before KA. The changes in Ca²⁺ signaling was measured with fiber photometry and expressed in ΔF/F₀; gross electrical activity of the brain was measured with EEG telemetry and presented in µV amplitude signals (left panels). The ΔF/F₀ fluorescence curves are presented as mean ± S.E.M. of the recordings occurring during the seizure stages 4 and 5. KA was dissolved in saline, while JNJ-47965567 was dissolved in 30% SBE-β-CD + 70% saline (the solvent of JNJ-4438079 is indicated in the Figure panels as saline for the sake of simplicity). All details of the procedures used are described in the Methods Section. EEG recordings are shown in the middle panels, while the right panels show the area under the curve (AUC) of the ΔF/F₀ changes and the total power of the EEG as n-time changes from the baseline. The levels of statistical significance reached are marked with *P < 0.05, **P < 0.01, and ***P < 0.001. The unpaired t-test was used for evaluation throughout (A, t = 7.809; B, t = 20.26; C, t = 0.9751; D, t = 17.52; E, t = 9.273; F, t = 11.20; G, t = 11.20; H, t = 7.902). The number of experiments was 5 in each panel.

Figure 4

Effects of kainic acid (KA; 30 mg/kg, i.p.) injection on Ca²⁺ signals in neurons and astrocytes in the CA1 area of the hippocampus and in the medial prefrontal cortex (mPFC). In some of the experiments, the n-time change of the EEG total power was also measured. The increase of Ca²⁺ signals was prevented by the P2X7R antagonist JNJ-47965567 (30 mg/kg, i.p.), as well as by the A2AR antagonists KW6002 and SCH58261 (3 mg/kg, i.p., each). The time-course of the experiments and further methodological details are described in the Legends to Figures 1 and 2. In contrast to these Figures, the effect of KA is presented as AUC minus the solvent-induced effect. (A) The increase of Ca²⁺ signals by KA was similar in neurons and astrocytes of the hippocampal CA1 region and the mPFC each, but this increase was much smaller in the latter than in the former. (B) JNJ-47965567 greatly inhibited the KA-induced increases in the Ca²⁺ concentration both in neurons and astrocytes of the CA1 area of the hippocampus. Similarly, JNJ-47965567 markedly inhibited the EEG activity. (C) JNJ-47965567 greatly inhibited the KA-induced increases in the Ca²⁺ concentration both in neurons and astrocytes of the mPFC. JNJ-47965567 again markedly inhibited the EEG activity. (D) KW6002 and SCH58261 also strongly inhibited the Ca²⁺ concentration increase in CA1 neurons and astrocytes. The levels of statistical significance reached are marked with *P < 0.05, **P < 0.01, and ***P < 0.001. The unpaired t-test was used for evaluation in A-C (a, left, t = 9.270; A, right, t = 6.528; B, left, t = 8024, B, middle, t = 5.893; B, right, t = 22.91; C, left, t = 10.23, C, right, t = 19.11). The number of experiments was 5 for CA1 neurons and astrocytes, as well as 6 for mPFC neurons and astrocytes. The number of experiments was 10 for the EEG measurements. The two-way ANOVA was used for evaluation in D, left (Row F_4,8 = 3.883, P = 0.0486; Column F_2,8 = 47.11, P < 0.0001; KA vs. KW+KA, P < 0.0001; KA vs. SCH+KA, P < 0.0001) and D, right (Row F_4,8 = 0.5372, P = 0.7131; Column F_2,8 = 45.18, P < 0.001; KA vs. KW+KA, P < 0.0004; KA vs. SCH+KA, P < 0.0001).

Figure 5

Effect of KA injection (30 mg/kg, i.p.), 2 weeks after microinjection into the hippocampal CA1 area rAAV-hSyn-ATP1.0 to initiate the synthesis of a neuron-specific ATP-responsive sensor or its control virus, rAAV-hSyn-ATP1.0mut; in other experiments rAVV-hSynAdo1.0med was microinjected into the CA1 area to induce the synthesis of an adenosine (ADO)-responsive sensor or its control virus, rAVV-hSyn-Ado1.0mut. (A, left) The ΔF/F₀ fluorescence ratio obtained was a measure of the ATP release, 1 h after KA injection, in the CA1 area around neurons. The heatmap qualitatively characterizes the time-dependent release of ATP (A, middle), while the area under the curve (AUC) gives a quantitative measure of the released ATP (A, right). (B) Analogous presentation of ATP release 24 h after KA injection. Analogous presentation of adenosine release around CA1 neurons, 1 h (C) and 24 h (D) after KA injection. The composition of the panels was the same as that in A, B. The levels of statistical significance reached are marked with *P < 0.05, **P < 0.01, and ***P < 0.001, n.s. P > 0.05. The unpaired t-test was used for evaluation throughout (A, t = 10.00; B, t = 0.2969; C, t = 3.306; D, t = 1.106). The number of experiments was 5 in each panel. The release of ATP was much larger than the release of adenosine (AUC A vs. AUC C; t = 4.406, P = 0.0023).

Figure 6

Effects of kainic acid (KA; 30 mg/kg, i.p.) on hippocampus-dependent spatial memory of mice in the Morris Water Maze test, when applied alone or in combination with JNJ-47965567 (30 mg/kg, i.p.), KW6002, or SCH58261 (3 mg/kg, i.p., each). Four training days were followed by a probe trial on the 5^th day, as described in the Methods Section. The tracks covered by the mice during the training days are shown in A, D, G. After the last training day measurement, JNJ-47965567 or solvent injection was followed after a 1 h-interval by KA or its solvent application (A). Then, the next day the probe trial measurement was carried out. The results of the path length and escape latencies on each day are shown in B. The number of platform crossings and duration of stays in the target quadrant are presented in C. KA was dissolved in saline, while JNJ-47965567 was dissolved in 30% SBE-β-CD + 70% saline. With an analogous application protocol, KW6002 (D-F) or its solvent, or SCH58261 or its solvent (G-I; 15% DMSO + 85% saline, in each case) were injected i.p. Con, control; Solv, solvent. Two-way ANOVA was used in this (C) and also in the next sets of experiments (F, I) for statistical evaluation; the levels of significance reached are marked with *P < 0.05, **P < 0.01, ***P < 0.001, n.s. P > 0.05 (C below left, Row F_7,28 = 0.4175, P = 0.8832; Column F_4-28 = 2.305, P = 0.0292; C below right, Row F_7-28 = 0.5680, P = 0.7754, Column F_4,28 = 2.953, P = 0.0374; F below left, Row F_7-28 = 1.027, P = 0.4345, Column F_4-28 = 3.309, P = 0.0243; F below right, Row F_7-28 = 0.9130, P = 0.5110, Column F_4-28 = 3.957, P = 0.0114; I below left, Row F_7-28 = 0.9806, P = 0.4647, Column F_4-28 = 4.995, P = 0.0036; I below right, Row F_7-28 = 2.1185, P = 0.0668, Column F_4-28 = 6.348, P = 0.0009). The number of experiments was 8 in each column.

Figure 7

Probe trial data of the Morris Water Maze test in control mice or mice injected with kainic acid (KA; 30 mg/kg, i.p.) after previous micro-injection into their lateral ventricles of neuron-specific or astrocyte-specific shRNAs directed against P2X7Rs. KA was dissolved in saline and was injected 24 h before the collection of data on the day of the probe trial. A group of mice was injected with P2X7R shRNA and another group with control shRNA lacking mP2X7, which is responsible for the knockdown of the receptor. The results show that the neuron-specific abrogation of P2X7Rs does not change the cognitive achievements of mice in the MWM test, whereas the astrocyte-specific abrogation of this receptor inhibits the cognitive achievements. C, control; KD, knockdown; MWM, Morris Water Maze. Effect of neuron-specific shRNA on the number of platform crossings (A) and the duration of stays in target quadrants (B) (mean ± SEM of 8-10 mice). Kruskal-Wallis ANOVA on ranks (A) or one-way ANOVA (B) was used for statistical evaluation; the levels of significance reached are marked with *P < 0.05, **P < 0.01 (A, Groups = 6, Kruskal-Wallis statistics = 29.27, P < 0.0001; Dunn's test, KD-shRNA vs. KD-shRNA+KA, P = 0.0011; B, F_5,49 = 6.568, P < 0.0001; Dunn's-test, C-shRNA vs. C-shRNA+KA, P = 0069, KD-shRNA vs. KD-shRNA+KA, P = 0.0110). In addition, pairs of columns were compared with each other by the unpaired t-test or the Mann-Whitney test, as adequate and the levels of significance are marked with ^§§P < 0.01, ^§§§P < 0.001. A (C-Saline vs. KA, t = 4.234; C-shRNA vs. C-shRNA+KA, t = 2.982; KD-shRNA vs. KD-shRNA+KA, t = 5.393) and B (C-Saline vs. KA, t = 3.414). Effect of astrocyte-specific shRNA on the number of platform crossings (C) and the duration of stays in target quadrants (D) (mean ± SEM of 10-15 mice). Kruskal-Wallis ANOVA on ranks (C, D) was used for statistical evaluation; the levels of significance reached are marked with *P < 0.05, n.s. P > 0.05 (C, Groups = 6, Kruskal-Wallis statistics = 21.62, P = 0.0002; Dunn's test, C-Saline vs. KA, P = 0.0222; D, Groups = 6, Kruskal-Wallis statistics = 15.25, P = 0.0093). In addition, pairs of columns were compared with each other by the unpaired t-test or the Mann-Whitney test, as adequate and the levels of significance are marked with ^§§P < 0.01, ^§§§P < 0.001, n.s. P > 0.05 (C, C-shRNA vs. C-shRNA+KA, Mann-Whitney U = 4.5; D, C-shRNA vs. C-shRNA+KA, Mann-Whitney U = 17).

Figure 8

Effects of kainic acid (KA; 30 mg/kg, i.p.) on hippocampal-dependent non-spatial memory of mice in the Novel Object Recognition test, when applied alone or in combination with JNJ-47965567 (30 mg/kg, i.p.), KW6002, or SCH58261 (3 mg/kg, i.p., each) (A-C). On the first 2 days mice were trained to recognize an object as described in the Methods Section. Then, KA was injected, and on the following day the time spent with the novel object and the preference index for the correctly recognized objects were determined. Further details are described in the Methods Section. KA was dissolved in saline, while JNJ-47965567 was dissolved in 30% SBE-β-CD + 70% saline. With an analogous application protocol, JNJ-47965567 or its solvent (A), KW6002 or its solvent (B; 15% DMSO+85% saline), or SCH58261 or its solvent (C; 15% DMSO + 85% saline) were injected i.p. Con, control; Solv, solvent. Two-way ANOVA was used in all experiments for statistical evaluation; the levels of significance reached are marked with *P < 0.05, **P < 0.01, ***P < 0.001, n.s. P > 0.05 (A left, Row F_7,28 = 1.257, P = 0.3066; Column F_4-28 = 14.02, P < 0.0001; A right, Row F_7,28 = 1.876, P = 0.1118; Column F_4-28 = 8.668, P < 0.0001; B left, Row F_7,28 = 1.286, P = 0.2930; Column F_4-28 = 19.37, P < 0.0001; B right, Row F_7,28 = 0.7618, P = 0.6234; Column F_4-28 = 6.513, P < 0.0008; C left, Row F_7,28 = 1.729, P = 0.1426; Column F_4-28 = 12.52, P < 0.0001; C right, Row F_7,28 = 0.9046, P = 0.5170; Column F_4-28 = 8.596, P < 0.0001). The number of experiments was 8 in each column. Effects of kainic acid (KA; 30 mg/kg, i.p.) and its solvent in the open field test (D, E). The tracks of mice, when injected with KA or its solvent (D). Effects of KA and its solvent on the total running distance, the time spent in the center, the time spent in the border and the rearing frequency (E). None of these parameters changed in comparison with their solvent treated counterparts. The unpaired t-test was used to check statistically significant differences (E left, t = 0.4654; E left middle, t = 0.5673; E right middle, t = 0.7624; E right, t = 1.521). The number of experiments was 8 in each column. The acute depressive-like reactions in the tail suspension test (TST), forced swim test (FST) and sucrose consumption test (SCT) also did not differ in comparison between solvent- and KA-induced effects. The unpaired t-test was used to check statistically significant differences ( F left, t = 0.7031; F middle, t = 0.6666; F right, t = 0.8979). The number of experiments was 10 (TST) and 8 (FST, SCT).

Figure 9

Effects of kainic acid (KA; 30 mg/kg, i.p.) on NMDA and Bz-ATP-induced current amplitudes in hippocampal CA1 neurons and astrocytes, 1 hour and 1 day after KA application; exclusion of a presynaptic modulation of glutamate release from CA1 neurons induced by double-pulse stimulation, with endogenously released ATP. Representative patch-clamp recordings of current responses to NMDA (100 µM) and Bz-ATP (1000, 3000 µM) from astrocytes and neurons without and with the preceding injection of KA (A, B, D, E). The mean ± S.E.M. current amplitudes for astrocytes (C) and neurons (F) are plotted in the respective panels. Further details are described in the Methods Section. One-way ANOVA was used for statistical evaluation; the levels of significance reached are marked with ^§§§P < 0.001, n.s., P > 0.05 (C, F_5,48 = 6.816, P < 0.0001; differences from control Bz-ATP 1000, P = 0.2301 and P > 0.9999, respectively; F, F_5,42 = 12.06, P < 0.0001, differences from control Bz-ATP 3000, P = 0.0007 and P = 0.0001, respectively). The differences between the pairs of white and red columns were calculated by the t-test; the levels of significance reached are marked with *P < 0.05, **P < 0.01, ***P < 0.001 (C, control, t = 2.293, P = 0.0357; KA 1 h, t = 1.959, P = 0.0677; KA 1 d, t = 4.923, P = 0.0002; F, control, t = 0.2017, P = 1.340; KA 1h, t = 3.238, P = 0.0060; KA 1 d, t = 0.4075, P = 0.0011. Paired-pulse potentiation (PPP) in CA1 hippocampal neurons of saline- and KA (30 mg/kg, i.p.)-injected mice, in standard aCSF (G, H) and gabazine (10 µM)-containing aCSF (I, J). Original recordings of electrically-evoked (e)EPSCs under both conditions (G, I). The stimulation artifacts were retouched from the recordings. eEPSCs were evoked by two stimuli (7 mA strength, 100 µs duration) with an inter-pulse interval of 50 ms, every 20 s. A438079 (10 µM) did not alter the P2/P1 ratio in the mean ± S.E.M. of 10 cells (H) in the absence of gabazine, although in a minority of cells there was a difference (G; n = 2) between preparations taken from saline- and KA-treated mice. There was no such variability (I, J) when gabazine blocked the GABA_AR component of the eEPSC (n = 8). The statistical evaluation of data showed that the PPP in aCSF in comparison with that measured in aCSF + A438079, did not differ from each other in a statistically significant manner (H, right panel; t-test, t = 1.413). The same occurred in the presence of gabazine (J, right panel; t = 0.158).

Figure 10

Kainic acid (30 mg/kg., i.p.) injection to mice blocks long-term potentiation (LTP) caused in hippocampal CA1 neurons by stimulation of the Schaffer collaterals. Field excitatory potentials (fEPSPs) were evoked in brain slice preparations by 100 µs duration square wave pulses every 20 s (stimulation strength, 7 mA). LTP was induced by high frequency stimulation (100 Hz for 1 s, repeated two times at an interval of 20 s). Experimental arrangement and original tracings in preparations taken from saline- and KA-treated mice (A). Electrophysiological measurement were made 24 h after injecting KA. The LTP was determined in the last 10 min of recording as a % over baseline value (B-F). (B) fEPSP slopes recorded for 30 min before and 60 min after high frequency stimulation in saline- and KA-injected animals. (C) Abolition of the effect of KA, when it was injected together with JNJ-47965567 (30 mg/kg, i.p.), on the fEPSP slope. (D) Abolition of the effect of KA injection when JNJ-47965567 (1 µM) was superfused onto brain slices. (E, F) Abolition of the effect of KA, when it was injected together with KW6002 or SCH58261 (3 mg/kg, i.p.) of mice before preparing brain slices. The unpaired t-test was used to calculate the levels of statistical significance in case of normal distribution; *P < 0.05, n.s. P > 0.05 (B, t = 2.205; C, t = 0.1701; D, t = 0.1099; F, t = 0.8476). The Man-Whitney test was used to calculate statistical significance in case of non-normal distribution; n.s. P > 0.05 (E, U = 23). The number of experiments varied between 6 and 9.