Reactive astrocyte-driven epileptogenesis is induced by microglia initially activated following status epilepticus

Extensive activation of glial cells during a latent period has been well documented in various animal models of epilepsy. However, it remains unclear whether activated glial cells contribute to epileptogenesis, i.e., the chronically persistent process leading to epilepsy. Particularly, it is not clear whether interglial communication between different types of glial cells contributes to epileptogenesis, because past literature has mainly focused on one type of glial cell. Here, we show that temporally distinct activation profiles of microglia and astrocytes collaboratively contributed to epileptogenesis in a drug-induced status epilepticus model. We found that reactive microglia appeared first, followed by reactive astrocytes and increased susceptibility to seizures. Reactive astrocytes exhibited larger Ca2+ signals mediated by IP3R2, whereas deletion of this type of Ca2+ signaling reduced seizure susceptibility after status epilepticus. Immediate, but not late, pharmacological inhibition of microglial activation prevented subsequent reactive astrocytes, aberrant astrocyte Ca2+ signaling, and the enhanced seizure susceptibility. These findings indicate that the sequential activation of glial cells constituted a cause of epileptogenesis after status epilepticus. Thus, our findings suggest that the therapeutic target to prevent epilepsy after status epilepticus should be shifted from microglia (early phase) to astrocytes (late phase).


Introduction 44
Epileptogenesis; i.e., the process leading to epilepsy, is a common sequel of 45 brain insults such as brain injury, cerebrovascular disease, or status epilepticus 46 (SE) [1,2] Such brain insults are typically followed by a latent period, during  previous studies showed that astrocyte calcium activity could contribute to 83 excitotoxic neuronal death through glutamate release following SE [24,25]. 84 However, the functional changes including Ca 2+ signaling of reactive astrocytes 85 after SE and their causal roles in epileptogenesis remain largely uncertain. 86 To evaluate the role of inter-glial communication between different types of 87 glial cells in the process of epileptogenesis, we assessed the spatiotemporal 88 dynamics of glial activation following SE. Using cell-type specific manipulation, 89 we show that relative alterations of both microglia and astrocytes play causal 90 roles in epileptogenesis. Moreover, reactive glia are temporally distinct and 91 collaboratively contribute to epileptogenesis. Reactive microglia appear first and 92 induce reactive astrocytes in the hippocampus after SE. These reactive 93 astrocytes present larger IP 3 R2-mediated Ca 2+ signals, which are essential for 94 induction of the increased seizure susceptibility after SE. We clearly 95 demonstrate that inhibition of microglial activation reduces astrogliosis, aberrant 96 astrocytic Ca 2+ signaling, and seizure susceptibility. We therefore conclude that 8 97 the sequential activation of glial cells; i.e., the initial activation of microglia 98 followed by astrocytic activation, is a cause of epileptogenesis after SE.

102
Astrocytic activation follows microglial activation after SE 103 To determine the contributions of glial cells to epileptogenesis, we used the 104 pilocarpine model of epilepsy in mice, a model known to be highly isomorphic 105 with human temporal lobe epilepsy [26,27]. Repeated low doses of pilocarpine 106 (100 mg kg −1 ) were injected intraperitoneally (i.p.) until the onset of SE (Fig 1A). 107 This ramping protocol has been shown to reduce mortality after SE [28,29]. To 108 investigate how glial cell activation affects the epileptogenic process, we first 109 examined the spatiotemporal pattern of microglial and astrocytic activation in the 110 hippocampus following SE. We initially assessed microglial and astrocytic 111 activation with immunohistochemistry using cell-type-specific activation markers 112 at 1, 3, 7, and 28 days after SE (Fig 1B and 1D). The area of Iba1-positive 113 microglia was significantly increased in CA1 from 1 to 7 days after SE, which 114 was followed by an increase in the area of GFAP-positive astrocytes in CA1 from 9 115 7 to 28 days after SE (Fig 1C and 1E).  To examine whether the first SE increased seizure susceptibility, the second 136 SE was induced 4 weeks after the first SE. A lower dose of pilocarpine was 137 required for the induction of the second SE in mice with prior exposure to 138 pilocarpine-induced SE at 8 weeks of age (PP) compared to those without such 139 exposure (SP) (Fig 1F). In addition, a lower dose of pilocarpine was required for 140 the induction of the second SE compared to the first SE ( Fig 1G). These data    showing Ca 2+ signal amplitudes (dF/F) (K) and frequency (M) (n = 57, 32, 85 170 cells/2, 2, 3 mice, ***P < 0.001, unpaired t-test). Cumulative probability plots 171 showing Ca 2+ signal amplitudes (dF/F) (L) and frequency (N) (P < 0.001,  (2-APB; 100 μM). 2-APB also significantly reduced the amplitude of astrocytic 188 Ca 2+ signals after SE (Fig 2C, 2H  in the dose of pilocarpine required for the induction of the first SE were observed 201 between IP 3 R2KO and WT mice (Fig 1F and 2O). These data indicated that 202 IP 3 R2-mediated Ca 2+ signaling in astrocytes does not alter the acute responses 203 to pilocarpine.

204
In IP 3 R2KO mice, the area of Iba1-positive microglia was significantly 14 205 increased in CA1 at 1 day after SE, suggesting that microglial activation after SE 206 was comparable in IP 3 R2KO and WT mice (S1 Fig). However, there was no 207 significant change in the dose of pilocarpine required for the induction of the 208 second SE in SP compared with PP mice (Fig 2O). There was no significant 209 change in the dose of pilocarpine required for the induction of the first and 210 second SE in IP 3 R2KO mice ( Fig 2P). These results suggested that 211 IP 3 R2-mediated astrocytic Ca 2+ hyperactivity is essential for the induction of the 212 increased seizure susceptibility after SE. tested, we found that Tnf and Il1b mRNAs were also significantly upregulated in 224 the isolated hippocampal microglia at 1 day after SE (Fig 3A).   To clarify whether microglial activation is required for astrogliosis, we  after the first SE (Fig 3L and 3N). In addition, Aif1 and Tnf mRNA levels were 271 significantly decreased at 1 day after SE with PLX5622 treatment compared with 272 those in the control diet group (Fig 3P). Similarly, the increased area of 273 GFAP-positive astrocytes in CA1 from 7 to 28 days after SE in control diet 274 (AIN-76A) mice was prevented in PLX5622 treated mice (Fig 3M and 3O). To 275 identify the optimal timing of microglial inhibition to prevent astrogliosis, we 276 applied PLX5622 from 3 weeks after SE (Fig 4A). This later PLX5622 treatment  (Fig 4C and 4E). These findings showed that the initial reactive microglia are 280 required to induce morphological activation of astrocytes after SE.  We then investigated whether microglial activation is required for astrocytic Ca 2+ 296 hyperactivity after SE. We also used a pharmacological approach to inhibit the 297 early microglial activation after SE. Microglia inhibition with minocycline reduced 298 the larger and frequent Ca 2+ signals of astrocytes (S1 Movie) (Fig 5A, 5B   prevented the increased seizure susceptibility (Fig 6A and 6B). No difference 333 was observed between control diet and PLX5622-treated mice in the dose of 334 pilocarpine required for the induction of the first SE (Fig 6C), indicating that PLX5622-treated mice (Fig 6E and 6F). In contrast, a lower dose of pilocarpine 343 was required for the induction of the second SE in later PLX5622 treatment 344 mice, similar to that in control diet mice (Fig 6G, 6H, 6I and 6J). These data 345 suggested that the inhibition of initial microglial activation rescues the increased 346 seizure susceptibility.  Here, we demonstrate that SE induces sequential activation of glial cells; i.e., the 373 initial activation of microglia, followed by astrocytic activation, which is essential potential. Here, we found that inhibiting microglia at the acute phase (0 to 7 days 398 after SE) but not the late phase (21 to 28 days after SE) reduced susceptibility to 399 the second SE, suggesting that activated microglia trigger the epileptogenic 400 process including astrocytic activation, but do not exert a direct proconvulsive 401 effect on the later phase after SE.

402
In the present study, we demonstrate that astrocytic activation develops  Notably, we found that genetic deletion of IP 3 R2 is sufficient to rescue the 414 increased seizure susceptibility and reduce astrogliosis. Our study thus 415 suggests that IP 3 R2-mediated Ca 2+ signaling in reactive astrocytes plays a 416 proconvulsive role in the epileptic brain and can contribute to epileptogenesis.  In this study, we also demonstrate that pro-inflammatory cytokines of  In summary, our findings identify a sequence of glial activation in the 481 hippocampus that contributes to the epileptogenic process. In this process, 482 microglial activation is identified as a crucial event to induce reactive astrocytes.

483
In turn, astrocytic Ca 2+ activation mediated by IP 3 R2 was essential for the  All studies used male C57BL/6J mice (SLC Japan, Shizuoka, Japan). IP 3 R2KO 494 mice on a C57BL/6 background were available from a previous study [32]; their 495 generation and maintenance have been previously described in detail.

496
Glast-CreERT2::flx-GCaMP3 mice on a C57BL/6 background were also 497 available from a previous study [30,31]; their generation and maintenance have 498 been previously described in detail. In the present study, we performed 499 immunohistochemistry and confirmed that GCaMP3 was co-localized with 500 GFAP, an astrocyte marker, but not with Iba1 or NeuN (S2 Fig and S1 Table).

503
Mice were housed on a 12 h light (6 am)/dark (6 pm) cycle with ad libitum 504 access to water and rodent chow. The animals were allowed to adapt to 505 laboratory conditions for at least 1 week before starting the experiments. All     and their amplitude (dF/F) and duration (full width at half maximum) using the 672 Originlab "peak analysis" function.