Potassium (see Chapter 7: Potassium Channels), water (see Chapter 8: Water Channels), and glutamate (see Chapter 9: Glutamate Metabolism) form a trio of critical neuron–astrocyte synaptic systems that might be called the “trio of synaptic gliostasis” (Fig. 14.1). It is interesting to note that altering the function of one of these systems often alters another, for example, inwardly rectifying potassium channel K_ir4.1^−/− mice have impaired glutamate uptake [45] and aquaporin-4 (AQP4)^−/− mice have impaired K⁺ uptake [46]. Therefore, there is functional coordination among water, potassium, and glutamate uptake. However, this may not require a direct intermolecular interaction. For example, there is no difference in expression of K_ir4.1 protein [46] or K_ir4.1 immunoreactivity [47] in AQP4^−/− mice nor AQP4 immunoreactivity in K_ir4.1^−/− mice [47]. In addition, no alterations were observed in membrane potential, barium-sensitive K_ir4.1 K⁺ current, or current–voltage curves in AQP4^−/− retinal Müller cells [48] or brain astrocytes [49]. Lack of alteration of K_ir channels in AQP4^−/− mice suggests the interesting possibility that the slowed [K⁺]_o decay may be a secondary effect of slowed water extrusion (“deswelling”) following stimulation, and this has been modeled carefully [50] but not directly demonstrated. Similarly, slowed “deswelling” could also relate to slowed seizure termination in AQP4^−/− mice [46].

In epilepsy, changes in the synaptic gliostasis of all three have been described. For potassium, K_ir4.1-mediated K⁺ uptake is diminished in epilepsy [41,42,51–54]. For water, downregulation and redistribution of AQP4 has been described in both animal and human epilepsies [55–57]. For glutamate, both glutamate transporters (downregulation of glutamate transporter-1 (GLT1) [58]) and glutamate metabolism (downregulation of GS [59]) have been found to be altered. We have recently found dramatic reduction in expression of GLT1, which is responsible for greater than 90% of glutamate uptake at the synapse, during the latent phase of epileptogenesis in the dorsal hippocampus in the intrahippocampal kainic acid mouse model (Hubbard and Binder, unpublished observations).

All of these changes would be expected to be proepileptogenic as described in chapters “Potassium Channels,” “Water Channels,” and “Glutamate Metabolism.” A goal for future AED development, then, would be to develop selective activators or regulators of K_ir4.1, AQP4, GLT1, and/or GS. If successful, such drugs would be potentially restorative of synaptic gliostasis. Currently, there are no drugs that selectively activate K_ir4.1 to our knowledge [60]. For AQP4, there are no selective activators despite great efforts in this area [61].

Modulation of glutamate uptake by astrocytes offers potential for decreasing excessive excitability associated with the development of epilepsy. Global deletion of GLT1 leads to intractable seizures and early postnatal lethality [62], which was confirmed recently with conditional deletion of GLT1 in astrocytes [63]. Conversely, overexpression of GLT1 in transgenic mice attenuated epileptogenesis and reduced chronic seizure frequency in a pilocarpine-induced model of epilepsy [64]. Until 2005, no pharmacological intervention was able to modulate GLT1 protein expression. Rothstein et al. [65] showed that ceftriaxone, a β-lactam antibiotic, is a potent stimulator of GLT1 transcription and glutamate uptake, acting via the nuclear factor-κB (NF-κB) signaling pathway [66], although these results are controversial [67,68]. Ceftriaxone has been shown to reduce extracellular glutamate levels [69] and have antiseizure effects [70–72]. The β-carboline alkaloid harmine can also increase GLT1 gene expression and glutamate uptake activity in vitro [73]. The β-lactamase inhibitors clavulanic acid and tazobactam decreased seizure-like activity in an invertebrate model [74], although their effects on GLT1 have not been explored. Recently, however, the effectiveness of clavulanic acid as an antiepileptic therapeutic has been debated [75,76]. Nevertheless, the approach to upregulate GLT1 expression with these related β-lactam drugs remains to be explored fully in standard animal models of epileptogenesis.

Other strategies to upregulate GLT1 include: (1) adenosine A₁ and A_2A receptors as well as equilibrative nucleoside transporter-1 (ENT1) can be targeted to modulate GLT1-mediated glutamate transport and release [77–84] (see Chapter 10: Adenosine Metabolism for details); (2) a new small molecule activator of GLT1 translation developed at Ohio State [64,85]; and (3) delivery of cells [86] or viruses [87] to express GLT1 locally at sites of pathology. Key questions that arise regarding GLT1 restoration as a therapeutic strategy include: (1) selectivity for epilepsy; (2) anatomic specificity to sites of “epileptic” GLT1 dysregulation; and (3) would GLT1 upregulation cause side effects similar to other AEDs such as sedation and cognitive problems?

Subcellular targeting and compartmentalization of K_ir4.1, AQP4, and GLT1 is another interesting opportunity for identifying new modulatory targets. Both K_ir4.1 and AQP4 demonstrate perivascular and perisynaptic targeting, and for AQP4 it is clear that the perivascular and perisynaptic pools are lost even at time points when total AQP4 protein may be elevated [56,57]. Recent studies have also shown subcellular compartmentalization of GLT1. Activity-dependent mobility of GLT1 transporters into and out of the synaptic region has recently been described [88], identifying another site and source of regulation of the efficacy of perisynaptic glutamate uptake. Biochemical studies have recently identified that GLT1 targeting to plasma membrane versus intracellular sites is regulated by sumoylation [89] and S-nitrosylation [90]. Thus, if GLT1 transporters are not only downregulated but also intracellularly dysregulated, restoration of perisynaptic GLT1 levels through intracellular targeting could be another therapeutic strategy. Further understanding of the activity regulation and intracellular (perisynaptic and perivascular) targeting and mobility of K_ir4.1, AQP4, and GLT1 will be of critical importance to establish feasibility. In this way, restoring “synaptic gliostasis” may not be as simple as upregulating mRNA or protein levels of K_ir4.1, AQP4, or GLT1, but in addition effectively manipulating plasma membrane versus intracellular fractions and perisynaptic versus extrasynaptic fractions. Furthermore, the physiological significance of homo- versus heterodimeric K_ir4.1 (K_ir4.1/K_ir5.1) [60] and that of different AQP4 isoforms [91] and assembly into orthogonal arrays of particles (OAPs [92,93]) remain to be explored in the context of epilepsy.

It has been observed for some time from studies both in vitro and in vivo that hyperosmolarity protects against seizures, whereas hypoosmolarity promotes generalized seizures [94–98]. The consequences of cell swelling and reduction of the extracellular space (ECS) include increased extracellular resistance [96,98], magnified effect of local extracellular ion and transmitter accumulation [99,100], and enhanced neuronal synchrony and excitability [95,96].

Based on the considerations above and those described in chapters “Potassium Channels,” “Water Channels,” and “Glutamate Metabolism,” a specific role for astrocyte swelling in increasing neuronal excitability is a compelling area for future investigation (Fig. 14.2) [101]. Astrocyte susceptibility to swelling during pathological states has been suspected for quite some time, dating back to early electron microscopy (EM) studies which noted that astrocytes “are very susceptible to edema and are among the first elements of the nervous system to swell in poorly fixed preparations” [102]. Recent evidence based on real-time volume measurements using two-photon microscopy indicates that astrocytes are much more prone to swelling than neurons [103,104]. Astrocyte susceptibility to volume changes has been attributed to selective expression of AQP4 [103,105–107]. During the buildup of excitability leading up to seizure (ictal) discharges, it is speculated that elevated K⁺ released from neurons during synaptic transmission is taken up into astrocytes along with water, causing astrocyte swelling and progressive reduction of the ECS. There is strong evidence in cultured astrocytes in vitro that astrocyte swelling leads to opening of volume-regulated anion channels (VRACs) to produce a regulatory volume decrease. Release of water through astrocytic VRAC is accompanied by substantial amounts of glutamate [108–111]. Among the first targets encountered by astrocytically released glutamate are extrasynaptic N-methyl D-aspartate receptors (NMDARs), leading to slow inward currents (SICs), and potentially interictal and ictal (seizure-like) discharges. In support of this possibility, SIC-like currents have been evoked in CA1 pyramidal neurons by cell volume changes alone in acute hippocampal slices in vitro [112]. Recent evidence indicates a very strong relationship between slight changes in osmolarity, astrocyte swelling, and the generation of NMDAR-dependent SICs [113].

The possibility for involvement of astrocyte swelling in seizure generation is even more compelling when taking into account the specific changes taking place in astrocytes during epileptogenesis. Loss and redistribution of AQP4 and K_ir4.1 away from astrocytic endfeet is expected to exacerbate astrocyte swelling by increasing water influx into astrocyte processes at synapses while decreasing efflux via endfeet into the cerebrovasculature [114]. This could prolong seizures by slowing the recovery of astrocytic volume, a possibility supported by the increased seizure duration observed in AQP4^−/− mice [46]. Upregulated expression of group I metabotropic glutamate receptors (mGluRs) in reactive astrocytes could further enhance modulation of AQP4 [115] and exacerbate swelling and swelling-evoked release of glutamate by astrocytes. Increased release of adenosine triphosphate (ATP) due to elevated secretion of inflammatory molecules and perhaps also by ectopic expression of the NMDAR subunit NR2B by astrocytes could significantly potentiate release of glutamate from VRACs as observed in vitro [116,117]. In addition, reduced expression of GS [59] would elevate cytoplasmic concentrations of glutamate in astrocytes, providing more glutamate to be released when VRACs open. Furthermore, extrasynaptic NR2B subunit NMDAR expression increases [118], providing additional targets for astrocytically released glutamate. All of these changes would be expected to exacerbate swelling, astrocytic release of glutamate, and stimulation of extrasynaptic NMDARs, contributing to the development of epilepsy and spontaneously recurring seizures. Although the therapeutic potential of inhibition of astrocytic VRAC has not yet been tested in epilepsy, the astrocyte-specific VRAC inhibitor DCPIB (4-(2-butyl-6,7-dichloro-2-cyclopentyl-indan-1-on-5-yl)oxobutyric acid) exhibits powerful neuroprotective effects in a rat model of ischemia [119]. Selective inhibition of astrocyte swelling, astrocyte glutamate release through VRAC, extrasynaptic NMDARs, and EphB2/ephrinB2-mediated NMDAR potentiation offer exciting avenues for the development of new strategies for the treatment of epilepsy.

A challenge facing future studies exploring specific astrocytic mechanisms of seizure generation and development of epilepsy is dissecting the relative contributions of astrocyte Ca²⁺ versus astrocyte swelling. This is difficult given that astrocyte Ca²⁺ elevations resulting from receptor activation will occur alongside glutamate and potassium uptake, water influx, and cell volume changes. This issue is complicated further by a report suggesting that ATP-induced astrocyte Ca²⁺ elevations in situ activate VRAC [120]. Furthermore, the changes taking place in reactive astrocytes during hippocampal sclerosis would be expected to affect both processes. However, tools are available to begin to differentiate between astrocyte Ca²⁺ sources and astrocyte swelling in the generation of epileptiform activity. Especially intriguing are the IP₃R2 knockout mice, in which astrocyte Ca²⁺ elevations are abolished [121]. Neuronal SICs can be readily generated in hippocampal slices from IP₃R2 knockout mice [112] suggesting that SICs occur independent of astrocyte Ca²⁺ elevations. It will be especially interesting in future studies using IP₃R2 KO mice to determine not only the extent to which astrocyte IP₃-mediated Ca²⁺ elevations play a role in the changes taking place during epileptogenesis, but also on the generation of epileptiform activity in vitro and in vivo. These mice might also be used to examine alternate Ca²⁺ sources in astrocytes that may become available over the course of epileptogenesis, such as astrocytic expression of NR2B NMDARs.

Adenosine levels are elevated during seizure activity [122,123]. Adenosine exerts a powerful inhibitory effect on excitatory synaptic transmission primarily through its interaction with presynaptic A₁ adenosine receptors (A₁Rs) to suppress neurotransmitter release [124–126]. Therefore, the cycle of adenosine release and breakdown is especially important in cases of excessive excitability including epilepsy. Once released from neurons and astrocytes, ATP is rapidly converted into adenosine monophosphate (AMP) and then into adenosine by extracellular nucleotidases [127]. The reuptake of adenosine occurs through equilibrative nucleoside transporters [128], and phosphorylation by the astrocyte-specific enzyme ADK breaks down adenosine and therefore clears excess adenosine from the ECS. Minor changes in ADK activity affect the active cycle between adenosine, AMP, ADK, and 5′-nucleotidase and lead to major changes in extracellular adenosine levels [129]. Therefore, alterations in ADK are especially relevant to the generation of seizures. Increased levels of ADK are associated with seizures whereas decreased levels may lead to seizure suppression [130]. Increased ADK expression has been linked to seizure activity in both human tissue and experimental models of epilepsy [37,131–133] (see Chapter 10: Adenosine Metabolism for details). Seizure induction in experimental epilepsy was found to decrease extracellular adenosine concentrations through the upregulation of ADK [134]. In the kainic acid model of temporal lobe epilepsy (TLE) in mice, profound astrogliosis and increased ADK activity was observed [133]. This coincides with the findings of Aronica et al. [37], who demonstrated prolonged increases in ADK—for at least 3–4 months—in the rat hippocampus and cortex after induction of status epilepticus (SE). This increase was also detected in the hippocampus and temporal cortex of TLE patients.

Collectively, the above findings support the ADK hypothesis of epileptogenesis [131,132], including the dysregulation of ADK and its contribution to the epileptogenic cascade. Adenosine, adenosine receptor agonists, and ADK inhibitors have well-established anticonvulsant efficacy [135–138]. Intracranial injection of adenosine prevents seizures in rats [139]. In addition, the use of transgenic mice revealed that reduced forebrain ADK protects against epileptogenesis [140]. Other studies involving adenosine augmentation therapies (AAT) include a silk protein-based release system for adenosine [141] and the local release of adenosine from grafted cells [142], both of which resulted in seizure suppression. Focal adenosine delivery, such as slow-release polymers, cellular implants, gene therapy, or pump systems, has been suggested as a new pharmacological tool to treat refractory epilepsy with minimal side effects [130].

Particularly exciting is the recent finding that even a transient adenosine augmentation may have longer-lasting epigenetic effects that are antiepileptogenic [143,144]. Probably the most effective treatment would be a brain-permeant peripherally administered small molecule inhibitor of ADK. This would hopefully obviate systemic side effects seen with direct adenosine delivery. If effective in triggering long-lasting antiepileptogenesis, such a drug would ideally need to be given only during an isolated therapeutic window just after an epileptogenic stimulus. This would potentially minimize long-range side effects while maintaining efficacy.

Impaired glutamate uptake and neurotoxic release of glutamate from growing gliomas have been observed in vitro [145]. It is thought that growing glial tumors actively kill surrounding neurons through the release of excessive quantities of glutamate, which may also contribute to the seizures frequently seen in conjunction with glioma [145,146]. Recent studies by Harald Sontheimer’s laboratory have identified specific pathological alterations in glioma mouse models and human tissues that correlate with tumor-associated epilepsy [147]. They recently identified that a hypothesized glutamate release pathway, cystine/glutamate transporter (SXC), is active in a subset of gliomas [148]. SLC7A11/xCT, the catalytic subunit of SXC, demonstrated elevated expression in about 50% of patient tumors. Compared with tumors lacking this transporter, SLC7A11-positive tumors were associated with faster growth, peritumoral glutamate excitotoxicity, seizures, and worse survival. In a translational pilot study, use of the FDA-approved SXC inhibitor sulfasalazine in nine patients with biopsy-proven SXC expression led to inhibition of glutamate release from the tumor in vivo as assessed by magnetic resonance (MR) spectroscopy [148]. This exciting study demonstrates that phenotyping tumors for glial-associated transport molecules will lead to selective pharmacological targeting to prevent or ameliorate tumor-associated epilepsy, and gets at the pathological mechanism of glutamate release from tumor cells rather than standard AED approaches of globally suppressing synaptic transmission. Similarly, “disconnecting” invading astrocytoma cells by targeting tumor “microtubes” which connect distant membrane protrusions by GJs is a more rational approach to limit glioma invasion than standard chemo- and radiotherapeutic “cell kill” approaches which cause collateral damage on the normal brain and body [149,150].

Until recently, the literature in the field of GJs and epilepsy was confusing and contradictory, relying on studies that demonstrated up- and downregulation of astrocytic and/or neuronal connexins and positive or negative effects of GJ inhibitors in epilepsy models [151]. It is important to recognize that nearly all pharmacological GJ inhibitor drugs (eg, carbenoxolone, halothane) are “dirty” drugs with other effects, thus interpretation of the available studies is limited.

Functional coupling analysis, obtained by patch-clamping astrocytes and filling the astrocyte syncytium with dyes to quantitatively measure coupling, is clearly the gold standard and has led to recent seminal findings from Christian Steinhäuser’s laboratory (see Chapter 11: Gap Junctions for details). In this study [36], the gap junctional connectivity of astrocytes from 119 specimens from patients with mesial temporal lobe epilepsy (MTLE) with (n =75) and without (n =44) sclerosis were examined. They found that in MTLE specimens with typical hippocampal sclerosis, there is a complete absence of typical “classical” astrocytes and an absence of astrocyte gap junctional coupling. In contrast, coupled astrocytes were abundant in nonsclerotic hippocampus. In the intracortical kainic acid (ICKA) model of TLE, mice exhibited decreased astrocytic coupling 4–5 days post injection and completely lacked coupling 3 and 6 months after SE in the sclerotic hippocampus [36]. In the nonsclerotic hippocampus, however, coupling remained intact. Interestingly, decreased astrocyte coupling preceded apoptotic neuronal death and the onset of spontaneous seizures. Decreased GJ coupling also impaired K⁺ clearance 4 hours post injection. The authors found that proinflammatory cytokines induced the uncoupling of hippocampal astrocytes in vivo [36], which agreed with similar in vitro findings that proinflammatory cytokines have an inhibitory effect on astrocytic GJ coupling [152]. To test the hypothesis that inflammation may contribute to the pathophysiology of seizures, lipopolysaccharide was injected into mice. Five days post injection, animals exhibited reduced GJ coupling and reduced connexin 43 (Cx43) protein levels, but unchanged connexin 30 (Cx30) protein amounts. The uncoupling effect of lipopolysaccharide could be fully prevented with the anti-inflammatory and AED levetiracetam [36]. In addition, uncoupling was prevented in Toll-like receptor 4 (TLR4) knockout mice. Taken together, these important data suggest that inflammation may contribute to rapid uncoupling of astrocytes and the uncoupling of astrocytes may be involved in epileptogenesis.

Of importance for the future will be to investigate how general a mechanism uncoupling is for other forms of epilepsy. In the SE models (kainic acid, pilocarpine), there is cell death and sclerosis. Do epilepsy models without significant cell death also demonstrate uncoupling of astrocytes? What is the timing of astrocyte uncoupling and loss of K⁺ homeostasis compared with the onset of epilepsy in these models? Further work in this area will be critical to establish the mechanisms and thresholds for astrocyte uncoupling in a variety of models of epilepsy. Restoration of gap junctional coupling in astrocytes, perhaps via modulation of the TLR4 pathway, represents another novel therapeutic strategy [153].

Post-traumatic epilepsy (PTE) and post-stroke epilepsy (PSE) are good examples of injury-induced secondary epileptogenesis in which the inciting event is clear (see Chapter 5: Neuropathology of Human Epilepsy for details). Thus, these syndromes are in need of antiepileptogenic therapies delivered after the insult that would prevent subsequent development of epilepsy. No such therapy is currently available, and all previous antiepileptogenic trials for PTE have been ineffective [17,154–161]. Interestingly, restoration of GLT1 expression after lateral fluid percussion injury (FPI) by the β-lactam antibiotic ceftriaxone decreased gliosis and reduced cumulative post-traumatic seizure duration in rats, providing proof-of-principle for the idea of restoration of glutamate homeostasis as an antiepileptogenic strategy against PTE [162]. A recent genetics and biomarker cohort study of 256 patients with moderate to severe TBI found evidence linking interleukin-1β (IL-1β) to PTE risk, and provides rationale for testing targeted IL-1β therapies as prophylaxis against PTE [163].

One thing that PTE and PSE have in common is breakdown of the blood–brain barrier (BBB) at the time of the initial event (see Chapter 12: Blood–Brain Barrier Disruption for details). Here, studies of the gliovascular junction and BBB disruption-induced epileptogenesis may be therapeutically relevant [164]. The structure and roles of the gliovascular junction [165] and the role of astrocytes in BBB permeability [166] and control of microcirculation [167–170] have only recently been appreciated. Local pathological alterations in the gliovascular junction could perturb blood flow, K⁺, and H₂O regulation and constitute an important mechanism in the generation of hyperexcitability.

Transient opening of the BBB is sufficient for focal epileptogenesis [171]. Extravasated albumin can be taken up by astrocytes which activates the transforming growth factor-β (TGF-β) pathway leading to focal epileptogenesis. Kaufer and colleagues [172] have worked out a mechanism by which albumin induces excitatory synaptogenesis through astrocyte TGF-β/ALK5 signaling. This mechanism provides an astrocytic basis for BBB disruption-induced epileptogenesis and suggests antiepileptogenic therapeutic approaches (TGF-β inhibition). Indeed, losartan, a TGF-β inhibitor and FDA-approved antihypertensive medication, has been found to exert antiepileptogenic effects in this model [173,174].

Another event that can occur with BBB disruption is infiltration of leukocytes into the brain. Migration of leukocytes, which are more abundant in human cortical CNS tissue of patients with epilepsy than in control tissue [175], is controlled by chemokines in physiological and pathological conditions. Leukocyte–endothelium interactions and subsequent recruitment of leukocytes in brain parenchyma represent key components of the epileptogenic cascade [176]. In a mouse model of epilepsy, Fabene et al. [175] found that seizures induced elevated expression of vascular cell adhesion molecules and enhanced leukocyte rolling and arrest in brain vessels mediated by the leukocyte mucin P-selectin glycoprotein ligand-1 (PSGL-1) and leukocyte integrins α4β1 and αLβ2. Moreover, the blockage of leukocyte–vascular adhesion attenuated BBB leakage, suggesting a pathogenic link between leukocyte–vascular interactions, BBB damage, and seizure generation [175]. Recent studies have demonstrated that lack of perforin, a downstream factor of natural killer (NK) and cytotoxic T cells, reduces BBB damage and mortality in the rat pilocarpine model of epilepsy [177]. Via astrocyte–lymphocyte interactions and expression of various adhesion molecules, astrocytes are well positioned to serve as “gatekeepers” for immune cell infiltration into the CNS [178]. Thus, adhesion molecules themselves may serve as antiepileptogenic targets to be delivered in the appropriate therapeutic time window [178].

Neuroinflammation is an integrated response of all CNS cell types, including microglia, macroglia, neurons, and infiltrating leukocytes to an initial injury (see Chapter 13: Inflammation for details). Innate immunity, or the early inflammatory response triggered by an insult, occurs in the brain after systemic infection. This phenomenon can eventually progress to an adaptive immune response in which the immune system can recognize and remember specific pathogens; this is mediated by activated lymphocytes recruited from the blood [179,180]. When a local inflammatory reaction is triggered in the brain following an injury, both microglia and astrocytes become activated and release a number of proinflammatory cytokines such as IL-1β, tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6). These proinflammatory mediators can alter the properties of both glial autocrine actions and glial–neuron paracrine signaling.

Specific inflammatory pathways are chronically activated during epileptogenesis in both microglia and astrocytes [181,182]. Along with astrogliosis, microglial activation has been shown within the sclerotic hippocampus of patients with TLE [181–185] and those exhibiting focal cortical dysplasia (FCD) [186,187]. Reactive astrocytes of lesioned areas in the hippocampus of MTLE patients overexpress NF-κB-p65, a transcription factor that can activate the transcription of numerous proinflammatory genes [181]. Activated microglia and astrocytes chronically express IL-1β in the hippocampus of patients with TLE [182], which coincides with an increase in interleukin-1 receptor type 1 (IL-1R1) activation in the rat forebrain during SE [188]. The infiltration of leukocytes into sclerotic tissue was detected in both human specimens and animal models using immunohistochemistry [185]. Microglia activation and proliferation is prevalent in resected human epileptic tissue [189].

Both innate and adaptive immunity are activated in various forms of epilepsy. The complement pathway, an inflammatory response cascade that is part of innate and adaptive immunity, is overexpressed in reactive astrocytes, microglia, and macrophages in human TLE [190,191], tuberous sclerosis complex (TSC) [192,193], and FCD [187]. Alterations in BBB permeability were also associated with inflammation in TSC-associated lesions [193]. A combination of proinflammatory cytokines and the components of the complement cascade may contribute to the spread of the inflammatory response and increased network excitability in the sclerotic hippocampus of patients with TLE, in lesions of patients with TSC, and in the dysplastic tissue of FCD.

Several studies have provided evidence in support of anti-inflammatory modulation for the treatment of epilepsy. A recent study used minocycline, a known inhibitor of inflammation, to determine whether innate immunity plays a causal role in mediating the long-term epileptogenic effects of early-life seizures [194]. Mice were induced with SE at postnatal day 25, which caused an increase in microglial activation. Mice induced with a second SE 2 weeks later responded with greater microgliosis and shorter latency to seizure expression. Minocycline abolished the acute seizure-induced microglial activation and decreased seizure susceptibility, suggesting that anti-inflammatory therapy after SE may be useful in blocking the epileptogenic process and mitigating the long-term damaging effects of early-life seizures. In a different study, treatment with aspirin, a nonselective cyclooxygenase inhibitor, reduced both the frequency and duration of spontaneous and recurrent seizures following pilocarpine-induced SE in rats [195]. Moreover, aberrant migration of granule cells, mossy fiber sprouting, and hippocampal neuronal cell loss were attenuated by aspirin.

The idea to block inflammation as a treatment for epilepsy is attractive but raises at least two problems. First, several immune agents and processes are triggered in response to an initial insult, and depression of all immune signaling would also depress the endogenous anti-inflammatory agents. Second, inflammation may contribute to the repair process that protects against major neuronal circuit changes that promote the emergence of spontaneous seizures [196]. A more sensible approach may be to target a single inflammatory cascade, such as regulation of the balance between brain IL-1β and IL-1RA [197,198]. The exogenous application of IL-1β prolongs seizures in an IL-1R1-mediated manner, and intrahippocampal application of recombinant IL-1RA inhibits motor and electroencephalographic seizures induced by bicuculline in mice [198]. Inhibition of IL-1β production using selective inhibitors of interleukin-converting enzyme (ICE/caspase-1) or caspase-1 gene deletion have been shown to block seizure-induced production of IL-1β in the hippocampus of rats. Reduction in ICE/caspase-1 activity resulted in a significant decrease in seizure onset and duration [199]. A group of recent studies has indicated the importance of high-mobility group box-1 (HMGB1) release from neurons and glia and its interaction with TLR4, a key receptor of innate immunity [200]. It appears that the HMGB1–TLR4 axis is active in human epileptic tissue [200,201], providing a new target for anti-inflammatory drug therapy [202,203].

Last but not least, epilepsy is much more than just seizures; nearly all epilepsy syndromes are associated with cognitive and behavioral comorbidities. Should these comorbidities be viewed now in the context of astrocyte dysregulation? Dysregulation at the level of individual gliosynaptic units and K⁺/glutamate/water dyshomeostasis, and also more broadly in tissue-wide disruption of the astrocyte syncytium [36] and astrocyte domain organization [35] could conceivably lead to widespread effects on cognition, memory, and behavior.

Let us consider two examples of astrocyte-based neuromodulation gone awry that could contribute to epilepsy comorbidities. First, Boison and Aronica [204] have advanced the “adenosine hypothesis of comorbidities.” Astrocyte overexpression of ADK induces tissue-wide deficiency in the “homeostatic tone” of adenosine. Boison and Aronica adduce evidence from patient samples demonstrating ADK overexpression not only in epilepsy but also in Alzheimer’s disease, Parkinson’s disease, and amyotrophic lateral sclerosis (ALS or Lou Gehrig’s disease). A transgenic “comorbidity model” with overexpression of ADK [205] and adenosine deficiency [206] is found to cause not only seizures [140,207], but altered dopaminergic function, impairment of attention, and cognitive and sleep regulation deficits [204,208,209]. In this conception, dysregulation of astrocyte adenosine metabolism produces widespread effects not only confined to seizures and epilepsy but also to cognitive comorbidities. Thus, normalization of adenosine by augmentation therapy might effectively treat not only manifestations of the “primary” condition (eg, seizures) but also comorbidities via the same mechanism [204].

A second example is that of AQP4 dysregulation. As described earlier and in Chapter 8: Water Channels in detail, AQP4 exhibits downregulation and/or altered subcellular distribution in epilepsy [55–57]. What effect does deletion of the perisynaptic pool of AQP4 [99,210] have on synaptic plasticity and cognitive function? Interestingly, AQP4^−/− mice were found to have profound and selective deficits in synaptic plasticity and memory. Specifically, AQP4^−/− mice exhibited a selective defect in hippocampal long-term potentiation (LTP) and long-term depression (LTD) without a change in basal synaptic transmission or short-term plasticity [211]. The impairment in LTP in AQP4^−/− mice was specific for the type of LTP that depends on the neurotrophin brain-derived neurotrophic factor (BDNF [212]), which is induced by stimulation at theta rhythm [theta-burst stimulation (TBS)-LTP], but there was no impairment in a form of LTP that is BDNF-independent, induced by high-frequency stimulation (HFS-LTP). LTD was also impaired in AQP4^−/− mice, which was rescued by a scavenger of BDNF or blockade of Trk receptors. AQP4^−/− mice also exhibited a cognitive defect in location-specific object memory but not Morris water maze or contextual fear conditioning. These results suggest that AQP4 channels in astrocytes may play an unanticipated role in neurotrophin-dependent plasticity and influence behavior [211]. Similar results have subsequently been obtained by other research groups [213–215]. Based on these results, downregulation of AQP4 may not only lead to increased neural excitability due to abnormalities of water and potassium homeostasis but may also lead directly to abnormalities in synaptic plasticity (both LTP and LTD). This provides a potential explanation for the way that astrocytic changes in epilepsy may contribute not only to seizures but also to cognitive comorbidities [216]. Cognitive impairment is very important because patients with TLE have many alterations in cognitive function and in particular hippocampal-dependent tasks such as spatial memory [217–220]. In addition, many other forms of plasticity are operative in the epileptic brain such as potentiation of synapses, reorganization of neuronal circuitry, and alteration in postnatal neurogenesis [221–223]. Thus, AQP4 deficiency or dysregulation could cause deficits in synaptic function and memory [224]. Restoration of AQP4 homeostasis could conceivably mitigate or reverse these deficits.