The Perils of Generalizing about GABA in Seizure Generalization

Depolarized GABAergic Signaling in Subicular Microcircuits Mediates Generalized Seizure in Temporal Lobe Epilepsy.

Wang Y, Xu C, Xu Z, Ji C, Liang J, Wang Y, Chen B, Wu X, Gao F, Wang S, Guo Y, Li X, Luo J, Duan S, Chen Z Neuron 2017;95:92–105.

Secondary generalized seizure (sGS) is a major source of disability in temporal lobe epilepsy (TLE) with unclear cellular/circuit mechanisms. Here we found that clinical TLE patients with sGS showed reduced volume specifically in the subiculum compared with those without sGS. Further, using optogenetics and extracellular electrophysiological recording in mouse models, we found that photoactivation of subicular GABAergic neurons retarded sGS acquisition by inhibiting the firing of pyramidal neurons. Once sGS had been stably acquired, photoactivation of GABAergic neurons aggravated sGS expression via depolarized GABAergic signaling. Subicular parvalbumin, but not somatostatin subtype GABAergic, neurons were easily depolarized in sGS expression. Finally, photostimulation of subicular pyramidal neurons genetically targeted with proton pump Arch, rather than chloride pump NpHR3.0, alleviated sGS expression. These results demonstrated that depolarized GABAergic signaling in subicular microcircuit mediates sGS in TLE. This may be of therapeutic interest in understanding the pathological neuronal circuitry underlying sGS.

Commentary

A key distinction in classifying various epilepsies is whether the seizure onsets are focal and limited to one brain area, or generalized and occur bilaterally across several brain areas. In addition, seizures can begin as focal events but spread throughout the brain via secondary generalization. For example, although temporal lobe epilepsy (TLE) is the most common form of partial epilepsy (1), many patients with TLE experience secondarily generalized seizures. This study focused on the local cellular and circuit mechanisms that may drive seizure generalization in TLE, focusing on the subiculum, the first region that is an output target of the hippocampus, and which has thus been postulated as a gate in the generalization of seizures arising in the hippocampus. Intriguingly, the results point to a critical role for subicular GABAergic neurons in focal seizure generalization, but with distinctly different effects on seizure generalization in nonepileptic versus chronically epileptic animals.

The authors first used high-resolution MRI imaging to measure the volume of the subiculum and various hippocampal subregions in patients with TLE. Although the entorhinal cortex, CA1, CA3, and dentate gyrus were all smaller in patients with TLE compared with controls, those patients who experienced secondarily generalized seizures also showed smaller volume of the subiculum. This clinical finding thus provided a rationale to hone in on the subiculum in mechanistic investigations using mouse models. First, the authors used a mouse that expresses the optogenetic protein channelrhodopsin (ChR2) in GABAergic neurons, to enable selective activation of GABAergic neurons at certain time points by delivery of blue light to the subiculum. In the first experiment, the authors delivered electrical kindling stimulation to the hippocampus of nonepileptic mice, and observed that subicular GABAergic neuron activation shortly after the kindling stimulation reduced the seizure severity and increased the degree of kindling stimulation required to drive development of generalized seizures. Furthermore, activation of the inhibitory optogenetic proton pump archaerhodopsin (Arch) in subicular GABAergic neurons had the opposite effect, and decreased the amount of stimulation needed to drive generalized seizure acquisition. These results are consistent with a model of GABAergic interneurons providing inhibitory input to subicular principal cells, and tamping down their activity to prevent further spread of seizure activity emanating from the hippocampus.

Interestingly, however, the effect of optogenetic GABAergic neuron activation on seizure generalization was reversed once the mice were fully kindled (i.e., the seizures elicited by the electrical stimulation regularly generalized). In this state, optogenetic activation of the subicular GABAergic neurons had a worsening effect and increased the duration of the secondarily generalized seizures, whereas Arch-mediated inactivation of the GABAergic neurons reduced seizure generalization. Similarly, chemogenetic activation of subicular GABAergic neurons, using hM3Dq, a G_q-coupled “designer receptor exclusively activated by designer drugs” (DREADD) (2), also suppressed seizure generalization in response to acute intra-hippocampal injection of the convulsant kainic acid (KA), but boosted secondarily generalized seizure duration in mice that developed chronic epilepsy weeks after the intrahippocampal KA injection. Together, these results indicate that the effect of GABAergic neuron activation on subicular principal cell activity switches from inhibition to excitation as kindling or epileptogenesis progresses.

Although GABA mediates the vast majority of synaptic inhibition in the brain (3), the effects of GABA on a postsynaptic cell depend on a combination of the type of GABA receptor expressed (e.g., ionotropic GABA_A vs metabotropic GABA_B) and the concentrations of chloride and anions both inside and outside the postsynaptic cell. Bicuculline, a GABA_A receptor antagonist, blocked the ability of optogenetic GABAergic neuron activation to suppress seizure generalization in the mice undergoing kindling, indicating that GABA_A receptors are primarily involved. Whether GABA_A receptor activation leads to a depolarizing or hyperpolarizing response in the postsynaptic cell is tuned by the relationship of the reversal potential for chloride (E_Cl) to the resting membrane potential (V_rest). Most mature neurons maintain low levels of intracellular chloride, which sets the E_Cl more hyperpolarized than V_rest, thus leading to hyperpolarizing responses upon opening of GABA_A receptors. Immature neurons and certain populations of mature neurons, however, express high levels of the Na-K-2Cl co-transporter (NKCC1), which pumps chloride ions into the cell and elevates the level of intracellular chloride, setting the E_Cl at a more depolarized value. If the E_Cl is more depolarized than V_rest, then the response to GABA_A receptor activation can be depolarizing, and if sufficiently strong, can drive action potential firing in the postsynaptic cell (4, 5).

The authors postulated that as kindling progresses and focal seizure generalization is stably acquired, the degree of NKCC1 expression in the subiculum changes. Indeed, subicular protein levels of NKCC1 increased, whereas expression of the K-Cl co-transporter (KCC2), which extrudes chloride from neurons, was decreased. Furthermore, when kindled animals were injected in the subiculum with an shRNA targeting NKCC1, thus reducing NKCC1 expression, the ability of optogenetic GABAergic neuron activation to attenuate seizure generalization was restored. Therefore, NKCC1-mediated changes in intracellular chloride levels of the postsynaptic subicular principal cells likely drive the differential response to GABAergic neuron activation in nonepileptic versus epileptic mice. In this regard, Arch-mediated suppression of principal cell activity was effective at reducing seizure generalization in epileptic mice, but use of another inhibitory opsin, halorhodopsin (NpHR3.0), was not. This difference likely results from the different mechanisms of hyperpolarizing action of these two opsins; Arch extrudes protons from the cell, whereas NpHR3.0 drives chloride accumulation. If the intracellular level of chloride is already sufficiently elevated, NpHR3.0 activation would not have a further hyperpolarizing effect. Indeed, the ability of NpHR3.0 in the subiculum to suppress seizure generalization could be restored by treatment with the NKCC1 antagonist bumetanide, which would restore intracellular chloride levels to a point at which GABA_A receptor activation would once again be hyperpolarizing.

Together, these results suggest that once kindling is complete or chronic epilepsy is established, subicular principal cells persistently accumulate intracellular chloride to a high enough degree that: 1) GABA_A receptor activation becomes predominantly depolarizing; and 2) the seizure-suppressing efficacy of NpHR3.0 is reduced. This finding contrasts with previous results demonstrating that NpHR3.0 activation in CA1 pyramidal cells can abort spontaneous hippocampal seizure activity in epileptic mice (6). Therefore, although hippocampal, entorhinal cortical, and other cells may also accumulate chloride with seizure activity (7, 8), the degree may not be sufficient to impair NpHR3.0-mediated seizure suppression. However, these dynamic changes in intracellular chloride levels, and switches in the response to GABA_A receptor activation, have profound implications for the use of anticonvulsants and antiepileptic drugs that act via potentiation of GABA_A receptor-mediated signaling, such as benzodiazepines and phenobarbital.

In summary, the work in the highlighted study supports a rationale for continued focus on the subiculum as a target in stopping the secondary generalization of seizures. Furthermore, these findings should be taken as a reminder of the perils of thinking of GABA and GABA_A receptors as unidirectional and simply “inhibitory.”