SEARCH

SEARCH BY CITATION

Keywords:

  • Temporal lobe epilepsy;
  • Neurosteroids;
  • GABAA receptors;
  • Tonic current

Summary

  1. Top of page
  2. Summary
  3. Hormones in the Brain
  4. Interneurons, GABAA Receptors and Tonic Inhibition
  5. Neurosteroids and Tonic Inhibition
  6. Alterations of Neurosteroid-Sensitive GABAA Receptors in an Animal Model of TLE
  7. Disclosure
  8. References

Epilepsies consist of a spectrum of neurologic disorders typically characterized by unpredictable and dysfunctional network behaviors in the central nervous system (CNS), which lead to discrete episodes of large bouts of uncontrolled neuronal synchrony that interfere with the normal functioning of the brain. Temporal lobe epilepsy (TLE) is accompanied by changes in interneuronal innervation and modifications in different γ-aminobutyric acid (GABA)A receptor subunits. Hormones play an important role in modulating the overall excitability of neurons, and at the same time hormonal pathways are frequently modified during epilepsy. This review focuses on TLE-correlated modifications of GABAergic transmission, and in particular on the implications of some of our own findings related to GABAARs containing the δ subunits (δ-GABAARs). These are extra- or perisynaptic GABAARs that mediate tonic inhibition, a major component of the inhibitory mechanism in the brain. The most potent endogenous modulators of δ-GABAARs are neurosteroids, which act as positive allosteric modulators. Plasticity of δ-GABAARs during TLE consists of down-regulation of the subunit in the dentate gyrus granule cells (DGGCs), while being up-regulated in interneurons. Surprisingly, the level of tonic inhibition in DGGCs remains unchanged, consistent with the idea that it becomes mediated by GABAARs containing other subunits. In parallel, tonic inhibition in a TLE model ceases to be sensitive to neurosteroid potentiation. In contrast, as predicted by the anatomic plasticity, interneuronal tonic current is increased, and remains sensitive to neurosteroids. These findings have important pharmacologic implications. Where neurosteroids normally have sedative and anticonvulsant effects, bimodal and cell-type specific modulations in their natural targets might weaken the inhibitory control on the dentate gate, under circumstances of altered neurosteroids levels (stress, ovarian cycle, or the postpartum period).


Hormones in the Brain

  1. Top of page
  2. Summary
  3. Hormones in the Brain
  4. Interneurons, GABAA Receptors and Tonic Inhibition
  5. Neurosteroids and Tonic Inhibition
  6. Alterations of Neurosteroid-Sensitive GABAA Receptors in an Animal Model of TLE
  7. Disclosure
  8. References

Network excitability may be viewed as the summation of the activity of discrete neuronal networks resulting from a balanced reciprocal control between inhibitory and excitatory components. Shifting the weight of one or the other neuronal population will determine the degree of excitability of a given brain area. This modulation can be achieved by regulating the expression and kinetics of ligand-gated or voltage-gated ion channels, by shifting the concentration gradient of the corresponding permeable ions, or by regulating neurotransmitter release probability and duration of action. Of the many possible ways of modulating network excitability, hormonal modulation stands out as a long-range and efficient mechanism for the task.

Hormones are classified into different chemical classes, and they act on diverse families of receptors. They serve as messengers for long-range, fast-acting, and long-lasting communication between organs and, as tuning molecules, they contribute to metabolic needs and functionality of the organism. Their ability to dynamically modulate neuronal outputs ultimately leads to their efficacy in influencing brain activity, behavior, and mood. Given these premises, it is not surprising how in pathologic states such as epilepsy, perturbation of the homeostatic hormonal effects on the brain can bring about an aggravation of the underlying disease, and reciprocally, physiologic hormone secretion can be disrupted by dysfunctional neuronal activity. Biochemically, hormones can be classified into lipid-derived (of which cholesterol-based steroids are the most abundant class), protein and peptide-derived, and catecholamines (adrenaline, noradrenaline, and dopamine) (Koeppen & Stanton, 2010).

Hormonal receptors can be found in different cellular compartments, depending on the nature of the ligand and the specific effect they trigger. For example, receptors for estrogen and progesterone (ER-α and ER-β and PR, respectively) are located in the cytoplasm or in the nucleus, in a strategic position for gene expression regulation. In this case, receptor activation normally leads to slow and long-lasting effects. Hormonal receptors can also be present in the plasma membrane, as is the case of the metabotropic receptors for dopamine and endocannabinoids, or the G-protein–coupled estrogen receptors (GPER1) that mediate nongenomic, fast effects of estrogen (Olde & Leeb-Lundberg, 2009). Their activation leads to cascades involving second messengers that can both act in a fast and transitory way by modulating the responsiveness of ion channels, or create long-lasting effects by enhancing or dampening the activity of transcription factors (Gómez-Ruiz et al., 2007; Mackie, 2008; Wegener & Koch, 2009; Beaulieu & Gainetdinov, 2011; Girault, 2012). Lastly, some hormones can bind directly to specific binding sites on ligand-gated and voltage-gated ion channels, resulting in fast and transitory effects on network excitability. Synthesis of neuroactive hormones can take place both peripherally (gonads, adrenal gland), and in the central nervous system (CNS). Basal levels of steroid hormone production (testosterone, estrogens) are always present in both the female and male brain, but in some cases, as for example following brain injury, their synthesis is tremendously increased (Mirzatoni et al., 2010). Locally synthetized steroids can act in a cell-autonomous way, or in a paracrine fashion on neighboring cells.

Changes to several hormonal pathways have been implicated in the epileptogenic process and in the triggering of ictal periods, both in human and animal models of TLE. Herein we report a few examples of hormonal pathways that are directly involved in excitability modifications during epilepsy. Estradiol is an ovarian hormone, which in rats has been found to increase hippocampal seizure susceptibility while lowering their severity. The mechanism of action seems to be mediated by estrogen receptor-α (ER-α), and involves decrease in γ-aminobutyric acid (GABA) release accompanied by an augmented release of the anticonvulsant molecule neuropeptide Y (NPY) (Ledoux et al., 2009). There is evidence for a switch in the expression of ER-α and ER-β from principal cells to reactive astrocytes in the epileptic hippocampus of rats (Sakuma et al., 2009) and for a selective increase in ER-α expression in human hippocampi under certain antiepileptic drug (AED) treatments (Killer et al., 2009).

Another hormonal pathway that is profoundly altered in epilepsy is that of corticotrophin-releasing hormone (CRH). Numerous independent studies have reported increase in CRH synthesis in different brain areas, including the hippocampus, in epileptic humans and mice (Wang et al., 2001; Wu et al., 2012). Although it was long known that CRH can act as proconvulsant in the developing brain (Baram & Schultz, 1991), its role together with the hypothalamic pituitary adrenal (HPA) axis during epilepsy is currently under extensive investigation (Sarkar et al., 2011).

Neuregulin-1 (NRG-1) has recently been implicated in the regulation of excitability of a subclass of interneurons and seems to play a protective role in epilepsy (Li et al., 2012; Tan et al., 2012). Briefly NRG-1 is a peptide that acts as neurotrophic factor by activating the receptor tyrosine-protein kinase ErbB4. It increases inhibition by increasing the excitability of fast-spiking basket cells, and the downstream molecular pathway has been shown to be downregulated in human chronic epilepsy (Li et al., 2012). Further studies are needed to understand to what extent this hormonal pathway can be a potential target for treatment. Of interest, it seems that NRG-1 secretion is under progesterone control, via cytoplasmic progesterone receptor activation (Lacroix-Fralish et al., 2007). Progesterone is a well-established anticonvulsant and sedative (Majewska et al., 1986; Söderpalm et al., 2004), and these effects are mostly mediated by its neuroactive form allopregnanolone through an increase in GABAA receptor–mediated inhibition (Belelli & Lambert, 2005). Nonetheless, NRG-1 is a good candidate as an additional molecular pathway through which progesterone can decrease brain excitation.

Ongoing investigation of hormonal molecular pathways, and of their modulation during health and disease, constitutes an important basis for development of targeted pharmacologic approaches in order to lower frequency and magnitude of seizure activity in epileptic patients. This review will focus on 3α-hydroxy ring A–reduced pregnane steroids and their role during states of altered neuronal excitability such as epilepsy. Our main findings pertain to neurosteroid regulation of the inhibitory component of neuronal networks in health and disease. Neurosteroids are brain-derived metabolites of gonadal and adrenal gland steroids and are synthetized in both neurons and glia by the enzyme 5-α-reductase. In particular, our main topics will be the progesterone-derived allopregnanolone (ALLO) and the cortisol-derived tetrahydrodeoxycorticosterone (THDOC), which at physiologic concentrations preferentially act as positive allosteric modulators of extrasynaptic δ-containing GABAA receptors, and through this biochemical mechanism, fulfill their sedative, anticonvulsant, and anxiolytic effects (Belelli & Lambert, 2005).

Interneurons, GABAA Receptors and Tonic Inhibition

  1. Top of page
  2. Summary
  3. Hormones in the Brain
  4. Interneurons, GABAA Receptors and Tonic Inhibition
  5. Neurosteroids and Tonic Inhibition
  6. Alterations of Neurosteroid-Sensitive GABAA Receptors in an Animal Model of TLE
  7. Disclosure
  8. References

In the mammalian brain, the GABAergic system is in charge of the inhibitory control of neuronal output, and it plays a pivotal role in orchestrating synchronicity of local networks and functional coupling of different brain regions (Mann et al., 2005; Montgomery & Buzsáki, 2007; Buzsáki & Wang, 2012). Its fundamental components are highly specialized and heterogeneous families of neurons that synthesize and release GABA (Freund & Buzsaki, 1996; Klausberger & Somogyi, 2008), and an equally complex and diverse system of receptors that bind GABA (Mody & Pearce, 2004). Epilepsy comprises many syndromes, each of which is characterized by pathologic modifications of neuronal excitability and network synchrony (Engel, 2006; Engel et al., 2009). Hence it is not surprising that a number of brain region–specific changes in both GABA-releasing cells and GABA-binding receptors have been implicated in TLE, although the direct functional consequences of each modification are difficult to predict. Equally problematic is a clear-cut separation between compensatory and aggravating alterations (Mody, 1998; Zhang et al., 2007; Mann & Mody, 2008; Avoli & de Curtis, 2011). Understanding the many facets of GABAergic plasticity during epileptogenesis and during the chronic phase of TLE will be a fundamental step toward the development of specific drugs for the control and prevention of seizures. In order to investigate the functional consequences of GABAergic plasticity in TLE, in the following section we review anatomy, physiology, and the functional role of the GABAergic system in the healthy CNS.

Among GABAergic neurons are the interneurons that coordinate the activity of local networks (Klausberger & Somogyi, 2008; Isaacson & Scanziani, 2011), but some GABAergic neurons can project long range, and synchronize far apart brain regions (Alonso & Köhler, 1982; Jinno et al., 2007; Melzer et al., 2012). Interneurons can be classified by their electrophysiologic properties, by certain exclusive proteins they express, by the neuronal compartments they innervate, or by areas they project to (Freund & Buzsaki, 1996). A powerful restraint on principal cell excitability is held by perisomatic inhibition. This type of GABAergic innervation is established by a subclass of interneurons which contacts on principal cells are localized on the soma and the axon initial segment, in a particularly advantageous position for the control of neuronal output (Freund & Katona, 2007). Both human and mouse-model TLE exhibit heterogeneous modifications of perisomatic GABAergic innervation of principal cells (Maglóczky, 2010; Wyeth et al., 2010; Li et al., 2012; Tan et al., 2012), suggesting the importance of this type of inhibitory innervation in maintaining adequate levels of neuronal excitability.

On the other side of the synaptic cleft, and its vicinity, are GABAARs. These receptors are heteropentameric ligand-gated Cl/inline image permeable channels, which belong to the Cys-loop receptor family, and are usually made of three different proteins, chosen from an array of several subunits (α1-6, β1-3, γ1-3, δ, ε, θ, π, and ρ1-2) (Laurie et al., 1992; Wisden et al., 1992; Sieghart & Sperk, 2002; Sun et al., 2004). The subunit composition will determine the pharmacologic characteristics of the receptor. For example, γ-subunit–containing receptors mediate different benzodiazepine effects depending on what α subunit they are paired with (Rudolph et al., 1999). In addition, certain α subunits (α4 and α6) are naturally insensitive to benzodiazepines. Moreover, the presence of a δ subunit in the GABAAR complex confers a high degree of neurosteroid-dependent modulation to the receptor (Belelli & Herd, 2003; Stell et al., 2003; Belelli & Lambert, 2005). Molecular and functional alterations of this subunit on different families of neurons during TLE are the main focus of this review. The varied pharmacologic properties of GABAARs allow for the development of highly specific drugs for differential application in a wide range of clinical situations (epilepsy, depression, anxiety, amnesia, insomnia).

Anatomic localization of the different receptors is another variable that correlates with subunit composition. The same neuron (excitatory or inhibitory) can express assorted combinations of subunits in separate subcellular compartments, and some subunits are typical of different brain areas (Lüddens & Wisden, 1991; Hevers & Lüddens, 1998; Pirker et al., 2000). For example, γ and δ subunits are mutually exclusive (Shivers et al., 1989; Araujo et al., 1998), with δ subunits generally found outside the synapses (Nusser et al., 1998b; Wei et al., 2003). In addition, δ subunits classically pair up with α6 in the cerebellum, whereas in the forebrain, they pair up with α4 in principal cells and α1 in interneurons (Glykys et al., 2007). Likewise, various subunits can be differentially expressed during development and in the adult brain (Laurie et al., 1992).

In addition to inhibition mediated by postsynaptic GABAARs, there is increasing evidence for the existence of axonal GABAARs (Trigo et al., 2008). Their physiologic role is controversial: activation of presynaptic GABAARs has been shown to both facilitate and inhibit transmitter release from the synapse, depending on the brain area, the developmental stage of the animal, and the neuron type. Moreover, presynaptic GABAAR agonism might function in a bimodal way, acting as excitatory at states of low receptor activation and inhibitory at higher degrees of activation (Jang et al., 2005). Although little is known of the subunit composition and the pharmacologic properties of presynaptic GABAARs on different kinds of cells, anatomic accessibility of mossy fiber boutons to patch clamp experiments has allowed for extensive functional investigation of presynaptic GABAARs on these terminals (Alle & Geiger, 2007; Ruiz et al., 2010). Although anatomic evidence of axonal δ-GABAARs presence is missing, the GABAARs of mossy fiber boutons pharmacologically behave as δ-GABAARs and their activation seems to facilitate synaptic transmission (Ruiz et al., 2010).

Lastly, subunit composition is responsible for the electrophysiologic behaviors of the GABAARs (Farrant & Nusser, 2005). Upon agonist binding, an ionic conductance ensues, which correlates with the synaptic or peri/extrasynaptic localization of the receptor, and the nature of the conductance depends on the Cl reversal potential. In addition, the receiving pyramidal cell seems capable of controlling local Cl concentration depending on what type of interneuron synapses onto it, thanks to differential postsynaptic expression of ClC-2, a hyperpolarization-activated Cl channel (Armstrong & Soltesz, 2012). ClC-2 is also found in interneurons, suggesting that this mechanism that is actively controlled by the postsynaptic cell might also involve interneuron-to-interneuron synapses. Synaptic GABAARs generate rapid postsynaptic currents known as inhibitory postsynaptic currents (IPSCs) and thus mediate phasic inhibition (Macdonald et al., 1989; Mody et al., 1994; Haas & Macdonald, 1999). Nonsynaptic GABAARs are localized either far or in the proximity of GABAergic synapses and generate a more recently discovered, persistent “tonic” conductance and are implicated in the management and maintenance of tonic inhibition (Mody, 2001; Semyanov et al., 2004; Farrant & Nusser, 2005). Tonic inhibition accounts for three fourths of the inhibitory charge received by the postsynaptic cell (Mody & Pearce, 2004), and pathologic modifications that directly interfere with its functionality during epilepsy will greatly influence the overall gain of neuronal networks (Semyanov et al., 2004). Functional relevance of tonic inhibition has recently been demonstrated in vivo (Duguid et al., 2012), as modulation of tonic inhibition in cerebellar granule cells (CGCs) has been shown to be a regulator of sensory information processing. In particular, basal levels of tonic inhibition seem to be optimal for seamless and accurate sensory transmission: CGCs spontaneous activity is kept low, while they are still readily responsive to sensory stimuli. Blockade or enhancement of CGCs tonic inhibition shifts the network from this optimal zone where signal-to-noise ratio is at a maximum, toward areas where increase in spontaneous activity or decrease of the power of sensory-evoked responses, respectively, decrease the efficiency of sensory transmission.

As it is true for interneurons, GABAARs can also be greatly altered in TLE, both in number and in subunit composition (Houser & Esclapez, 1996; Nusser et al., 1998a; Loup et al., 2000; Houser & Esclapez, 2003; Peng et al., 2004). In addition, some genetic epilepsies result from mutations that affect specific GABAAR subunits (Macdonald et al., 2010). These receptor modifications alter the response of target cells to an already modified inhibitory network, most likely pushing baseline activity toward more excitable and synchronized states.

Depending on the neuron type, tonic inhibition can be mediated solely by δ-containing GABAARs (δ-GABAARs), α5-containing GABAARs, or a combination of both (Glykys et al., 2008). Interest in δ-GABAARs increased once it was discovered that they are uniquely sensitive to low concentrations of neurosteroids, which act on them as potent positive allosteric modulators, thus increasing tonic inhibition in a fast and sustained way. The three main functional characteristics that distinguish phasic inhibition–mediating and tonic nhibition–mediating receptors are their differential affinity for GABA, desensitization rate, and agonist efficacy.

Typical of extrasynaptic GABAARs is their high affinity for GABA (half maximal effective concentration [EC50] 1 μm), which allows them to respond to very low GABA concentrations, and makes them an ideal sensor for ambient GABA, which oscillates in the near micromolar range (Nyitrai et al., 2006). In the DGGCs of the hippocampus, δ-GABAARs are localized in proximity to GABAergic synapses, where they can readily detect GABA spilled over from nearby boutons (Wei et al., 2003). In other cases, δ-GABAARs are found extrasynaptically scattered across the dendritic tree, as in CGCs (Nusser et al., 1998b), indicating the existence of ambient GABA fields larger than those found around inhibitory synapses. Cerebellar astrocytes were recently identified as a source for ambient GABA, and upon activation they can tonically release GABA through nonspecific anion channels (Best-1) (Lee et al., 2010). In addition, GABA can be released in the extracellular compartment by a particular type of cortical interneuron, the neurogliaform cell (NGFC), the release of which produces effects on δ-GABAARs independently of synaptic transmission (Oláh et al., 2009). The axonal cloud of these cells can cover an area with a diameter of 200 μm, rendering this type of communication particularly fit to level and synchronize the amount of excitation of a network over broad distances. GABA spillover and volume transmission are the sources of extracellular GABA. However, the net ambient GABA concentration detected by extrasynaptic GABAARs results from the summation of GABA release and GABA reuptake, which relies on an efficient class of transporters, localized both on neurons and glia (GAT-1 and GAT-3, respectively) (Walker & Semyanov, 2008). Dysfunction of the reuptake machinery will alter the level of tonic inhibition in receiving neurons, as shown in neocortex layer II–III pyramidal cells poststroke (Clarkson et al., 2010). After depolarization, GABA transporters may act in reverse and release GABA into the extracellular space. Of interest, there is evidence of decreased ambient GABA in epilepsy (Minamoto et al., 1992), whereas evidence of alteration in transporter expression and function during epilepsy is controversial. Altered distribution was reported in human in human postsurgical TLE hippocampi (Lee et al., 2006), where overall decrease in transporter expression was reported in several mouse model studies (During et al., 1995; Bouilleret et al., 2000; Silva et al., 2002). Nonetheless, the mechanism of action of some AEDs like tiagabine is mediated by selective blockade of GABA uptake (Madsen et al., 2010), or as in the case of vigabatrin, by a reduced GABA catabolism. In both cases, ambient GABA levels would be increased, and tonic currents would become potentiated. The mechanism of action of these AEDs, which are already in use in the treatment of chronic epilepsy, suggests that increasing the tonic current may be a good strategy for the development of novel drugs.

The second characteristic that makes δ-GABAARs capable of sustaining a current that is always on in the continuous presence of GABA is their relatively modest desensitization to the agonist (Haas & Macdonald, 1999). Desensitization is a common property of ligand-gated ion channels, and describes a state of prolonged closures during which the agonist is still bound to the channel (Jones & Westbrook, 1996). In other words, rapidly desensitizing channels are designed to respond to fast, transitory, and large changes in agonist concentration, whereas nondesensitizing channels are capable of being functional for the entire time they are exposed to the agonist. Ambient GABA oscillations are slow and small compared to changes in synaptic GABA, which means that δ-GABAARs need to be able to respond to extracellular GABA concentrations that vary little in the short run. δ-GABAARs in the thalamus and cerebellum have been shown to be unable to respond to GABA spilled over from the synapse, and this was mainly ascribed to their desensitization in the presence of ambient GABA. This would suggest that discrete populations of δ-GABAARs, simultaneously activated, contribute to the generation of tonic conductance. Once a population becomes desensitized, tonic inhibition would be mediated by another population of extrasynaptic GABAARs (Bright et al., 2011).

Thirdly, despite their high affinity for the agonist, GABA efficacy on δ-GABAARs is relatively low compared to other GABAARs (Meera et al., 2009). This results from a low-level coupling between channel gating and agonist binding. This property offers extensive modulatory potential of δ-GABAAR-mediated tonic currents. In other words, their high affinity and low efficacy make these receptors very good listeners but very poor translators of extracellular GABA concentrations. Increasing their efficacy might be the major mechanism by which tonic inhibition is controlled in the CNS. Endogenously, neurosteroids promote tonic inhibition by increasing the efficacy of GABA on δ-GABAARs, in response to physiologic homeostatic needs. Exogenously, this constitutes a pharmacologic mechanism that can be used to clinical advantage. The sedative effects of steroid metabolites have been described decades ago (Atkinson et al., 1965), and the molecular mechanisms involving the GABAergic system through which these effects are mediated have been thoroughly investigated (Majewska et al., 1986; Stell et al., 2003; Belelli & Lambert, 2005; Herd et al., 2007). Neurosteroids are the most potent endogenous modulators of GABA-mediated transmission, and at low concentrations (nanomolar range) they enhance the GABA efficacy of δ-GABAARs. Their selectivity is likely to be ascribable to their kinetic effect rather than to a preferential interaction with δ-GABAARs: the already high efficacy of GABA on non-δ-GABAARs makes neurosteroid potentiation on them less likely.

Following GABAAR gating, the effect of GABA transmission on the excitability of a network depends on the Cl reversal potential of the postsynaptic cell. In particular, the balanced expression of two Cl transporters will determine if the cell will be depolarized or hyperpolarized by the opening of its GABAARs. NKCC1 and KCC2 are two symporters that, respectively, use the Na+ or K+ gradient generated by the Na-K pump to move Cl across the plasma membrane. In particular, NKCC1 activity increases intracellular Cl, whereas KCC2 decreases it. The developmental GABA switch from excitatory to inhibitory neurotransmitter is caused by molecular changes implemented in mature neurons, which increase KCC2 and lower NKCC1 expression (Ben-Ari, 2002). KCC2 expression has been shown to be impaired in the rat hippocampus following pilocarpine-induced status epilepticus and in the latent period, and chronically in sclerotic hippocampi of patients with TLE (Huberfeld et al., 2007; Pathak et al., 2007; Barmashenko et al., 2011). A reversed Cl gradient will further reduce the capability of the GABAergic network to maintain control of fleeting bursts of excitation. It follows that key structures to keep excitation in check, as for example the dentate gate, which is normally under stringent inhibitory control, start losing reliability in the way they process and pass on information.

Lastly, it is important to remember that a depolarizing GABA current will not necessarily excite a neuron. After all, inhibition can be achieved not only by hyperpolarization, but also by clamping the cell membrane at a certain potential, still hyperpolarized to the threshold of powerful voltage-gated conductances (Na+ or Ca2+). This phenomenon de facto dampens the neuron firing probability in what’s known as shunting inhibition (Staley & Mody, 1992; Mitchell & Silver, 2003). The effectiveness of shunting inhibition is proportional to GABA conductance, and at least in some interneurons seems to come into play only at higher ambient GABA concentrations (Song et al., 2011). This means that certain cells naturally maintain a membrane potential so close to GABA reversal potential that they will respond with either an increase or a decrease in their gain depending on the extracellular GABA concentration: at low concentration, the depolarizing effect will prevail, whereas it will be overcome by shunting inhibition at higher concentrations. In conclusion, the final network effect of GABAergic transmission rests upon a dynamic and delicate equilibrium between GABA release, GABAARs modulation, GABA transporters, and Cl gradient.

In the epileptic brain, this equilibrium is modified (Bernard et al., 1999; Cohen et al., 2002; Engel et al., 2009; Zhang et al., 2012), and although most of the time inhibition is capable of finely regulating excitation, changes in the many modulators of the GABAergic system, which would fall under normal homeostatic fluctuations in the healthy brain, may hence facilitate the triggering of epileptic discharges typical of clinical seizures.

Neurosteroids and Tonic Inhibition

  1. Top of page
  2. Summary
  3. Hormones in the Brain
  4. Interneurons, GABAA Receptors and Tonic Inhibition
  5. Neurosteroids and Tonic Inhibition
  6. Alterations of Neurosteroid-Sensitive GABAA Receptors in an Animal Model of TLE
  7. Disclosure
  8. References

Neurosteroids are synthetized in the cytoplasm of neurons and glia starting from steroid hormones released by the gonads and the adrenal gland, or produced locally in the cytoplasm of both neurons and glia (Belelli & Lambert, 2005). Plasma concentrations of progesterone and cortisol can greatly oscillate from virtually no detectable levels to hundreds of nm, depending on the physiologic state of the organism. In mammals during the ovarian cycle, progesterone levels vary between 2 nm in the follicular phase and 15 nm in the luteal phase. Physiologic stress response, both acute and chronic, modifies plasma levels of cortisol from a few nM to tens of nm (Reddy & Rogawski, 2002; Porcu et al., 2003). During the last third of pregnancy, circulating levels of progesterone are at their highest, around 100 nm (Paul & Purdy, 1992). Modifications in neurosteroid levels are also typical of physiologic aging, when circulating steroid levels are reduced (Schumacher et al., 2003). In general, neurosteroids are readily synthetized in the brain, and their concentration correlates to that of their precursors: hence, oscillations in steroid levels are faithfully followed by similar oscillations in neurosteroids (Paul & Purdy, 1992). Local steroidogenesis can take place both in neurons and glia, and the rate-limiting enzyme is the cholesterol-translocator transporter protein TSPO localized on the mitochondrial membrane (Rupprecht et al., 2010). Neurosteroids synthetized in neurons or glia could potentially serve different purposes and overall have different effects on network excitability by acting in a cell-autonomous way as opposed to paracrine action influencing neurons of specific astrocytic microdomains (Fig. 1).

image

Figure 1.   Schematic representation of two pathways for neurosteroid synthesis and possible differential neurosteroid action on GABAARs. Brain-derived precursors are synthetized from cholesterol both in neurons and glia. The rate-limiting step is the cholesterol transport into the mitochondrial matrix by the transporter protein TSPO (formerly known as peripheral benzodiazepine receptor, PBR). Cholesterol is converted to pregnenolone by the enzyme P450scc, located on the inner mitochondrial membrane. Pregnenolone is the precursor for ALLO and THDOC, which are synthetized in the smooth endoplasmic reticulum by the enzyme 5-α-reductase. Neurosteroidogenesis can also use plasma-derived precursors. Progesterone and cortisol/corticosterone synthesized in the ovaries and adrenal cortex, respectively, can diffuse into neurons and glia, where they are converted into ALLO and THDOC. The highly lipophilic neurosteroids accumulate in the plasma membrane and thus can modulate GABAARs. Plasma membrane concentrations of neurosteroids might be sufficient to modulate both δ-GABAARs and non-δ-GABAARs in a cell-autonomous fashion. However, low concentrations of neurosteroids diffusing through the extracellular space from neighboring neurons and glia may only act on δ-GABAARs, in a cell nonautonomous manner.

Download figure to PowerPoint

Up to 100 nm, neurosteroids preferentially if not uniquely potentiate δ-GABAAR-mediated tonic current (Stell et al., 2003), by increasing GABA efficacy on these receptors. The functional effect is a fast and strong increase in tonic inhibition on the receiving cell. The amount of tonic current augmentation is neurosteroid concentration–dependent, and in DGGCs at 10 and 100 nm it ranges from 100% to >300% increase compared to control (Stell et al., 2003). Given the fact that δ-GABAARs distribution is not homogenous throughout the brain, and that certain neurons express a weighted combination of δ-GABAARs and α5-GABAARs to sustain their tonic inhibition, neurosteroid effects will be stronger on those cells and in those brain areas, which preferentially utilize δ-GABAARs for their tonic GABA conductance. In the hippocampus, δ-GABAARs are heavily expressed on the dendrites of DGGCs, and on certain interneurons, whereas hippocampal pyramidal cells, exclusively (CA3) or preferentially (CA1) express α5-GABAARs (Glykys et al., 2007). The hippocampal dentate gyrus has been extensively investigated in epilepsy, and malfunction of its inhibitory component has been postulated to play an important role in the initiation or propagation of seizures (Heinemann et al., 1992). This function of the dentate gyrus has been termed as a filter or gate, and this gate is under inhibitory control. Dentate excitability is naturally very low compared to other areas of the hippocampus: excitatory input onto the dentate from the entorhinal cortex through the perforant path excites both granule cells and local interneurons (feed-forward inhibition), an activity fundamental for the maintenance of the gate function (Coulter & Carlson, 2007). Typically in TLE, GABAAR expression in both DGGCs and dentate interneurons changes, and this may contribute to a compromised efficiency of the gate.

At higher concentrations, neurosteroids become agonists on δ-GABAARs, thus triggering channel gating independently of GABA presence (Callachan et al., 1987). Lastly, exogenous drugs, such as ethanol and certain antidepressants, can alter neurosteroid levels. For instance, there is increasing evidence that ethanol effects on GABAARs might be mediated, at least in part, by neurosteroids, through an NMDA receptor–dependent increase in their synthesis (Sanna, 2004; Tokuda et al., 2011). Fluoxetine seems to normalize neurosteroid levels in major depression, posttraumatic stress syndrome (PTSD), and after social isolation, conditions characterized by reduced neurosteroid synthesis (Uzunova et al., 1998; Pinna et al., 2003; Pinna, 2004). This led to hypothesize that neurosteroids contribute to the antidepressant actions of this drug (Pinna et al., 2006).

The molecular mechanism of action of the allosteric modulation of GABAARs by neurosteroids most likely involves direct interaction of neurosteroids with the receptor. Two sites on the α subunit have been identified as necessary for the potentiating and activating effects of neurosteroids (Hosie et al., 2006), although there is a debate on whether these sites physically bind neurosteroids (Li et al., 2009). Recently a new site on β3 subunit has been described to directly bind a neurosteroid analog, although its functional role is yet to be investigated (Chen et al., 2012). There is increasing evidence that neurosteroids might gain access to a membranous GABAAR site of action through lateral diffusion into the plasma membrane. The interaction between neurosteroids and receptor would happen in the lipid phase, and would be dependent on the amount of neurosteroids that are accumulated into the lipid bilayer. This has several implications. First, the real neurosteroid concentration around the receptor might be hundreds of times larger than that calculated in the aqueous solution. Second, neurosteroid range of action will be limited to the synthetizing cell and immediate neighboring cells (autocrine and paracrine action). Third, this solubility-dependent, low-affinity and high-concentration interaction needs to be taken into account for the development of drugs that aim to mimic neurosteroid effect on GABAARs (Chisari et al., 2010).

The possibility of mimicking neurosteroid effects in pathologic conditions characterized by excitation–inhibition dysfunctions has allowed for the development of drugs that can selectively and directly enhance δ-GABAAR–mediated tonic inhibition. Clinical application of such pharmacologic agents has received increasing interest, as animal studies have shown intriguing potential. For example, direct enhancement of δ-GABAARs with gaboxadol (4,5,6,7-tetrahydroisoxazolo(5,4-c)pyridin-3-ol, or THIP) has been shown to lessen negative symptoms in a mouse model of postpartum depression (Maguire & Mody, 2008). At 500 nm, THIP acts as a δ-GABAAR-selective agonist (Brown et al., 2002) and was previously proposed as a sleep-promoting drug (Wafford & Ebert, 2006), only to fail phase III clinical trials because of its side effects (Brickley & Mody, 2012). Safer δ-GABAAR-specific agonists are currently under investigation (Wafford et al., 2009).

Ganaxolone is a synthetic neurosteroid currently under investigation as an AED for catamenial epilepsy, given the typical exacerbation pattern of this kind of epilepsy, which correlates with ovarian-cycle–linked steroid oscillations in the plasma of affected women. As it is true for other AED indications, not all kinds of epilepsies will benefit from drugs that increase tonic inhibition. In fact the effect of neurosteroid-like drugs on epileptic patients will depend on the anatomic and functional alterations typical of their specific syndrome. The clearest example is that of absence seizures. Of interest, THIP was used to trigger absence-like seizures in rats (Fariello & Golden, 1987), and in a case report, exogenous administration of progesterone increased seizure frequency in an absence epilepsy patient (Grünewald et al., 1992). These effects are most likely mediated by further potentiation of an already pathologically increased tonic inhibition on thalamic relay neurons described in animal models of absence seizures (Cope et al., 2009). In these cases, drugs with diametrically opposite effects might be of use. Unfortunately, to date no specific δ-GABAAR antagonists have been developed.

Lastly, ethanol is another molecule that can, at intoxicating concentrations (20–30 mm), increase δ-mediated tonic current in both interneurons and DGGCs (Glykys et al., 2007). Whether the effect is mediated by a direct interaction with the receptor or through the neurosteroid system via stimulation of neurosteroid synthesis is a topic of current controversy and fervent investigation.

Alterations of Neurosteroid-Sensitive GABAA Receptors in an Animal Model of TLE

  1. Top of page
  2. Summary
  3. Hormones in the Brain
  4. Interneurons, GABAA Receptors and Tonic Inhibition
  5. Neurosteroids and Tonic Inhibition
  6. Alterations of Neurosteroid-Sensitive GABAA Receptors in an Animal Model of TLE
  7. Disclosure
  8. References

GABAAR plasticity in human TLE and animal models of the disease is complex and encompasses neuron-type and brain region–specific modifications in the expression of different subunits and will affect both types of inhibition, phasic and tonic (Schwarzer et al., 1997; Fritschy et al., 1999; Loup et al., 2000; Scimemi et al., 2005; Goodkin et al., 2008). For the extrasynaptic GABAARs, changes in expression and modulation have been reported in different animal models, and consistently show decreased expression of δ-GABAARs in the molecular layer in the DG starting 4 days after the initial epileptogenic insult, and progressively continuing until it stabilizes at constant low levels at about 12 weeks (Schwarzer et al., 1997; Peng et al., 2004; Zhang et al., 2007). In addition, the γ2 subunit begins to be ectopically expressed perisynaptically, where it most likely pairs up with α4 subunits, which also show an increased level of expression. Of interest, tonic current in these cells remains at control levels, and is largely mediated by α5-GABAARs, and possibly by ectopic γ2-GABAARs, in a molecular shift that will have important functional and pharmacologic consequences (Mtchedlishvili & Kapur, 2006; Zhang et al., 2007; Rajasekaran et al., 2010).

First of all, a neuron that changes its array of extrasynaptic receptors will respond to GABA very differently. Although there are no in vivo studies on how different GABA concentrations influence the activation of various GABAARs, traditionally δ-GABAARs are considered as having high affinity for GABA, whereas α5-GABAARs seem to need concentrations 10 times higher to be activated. This implies that possibly a tonic current that shifts from being δ-GABAAR mediated to being mostly α5-GABAAR mediated will require higher ambient GABA levels in order to be as effective. This consideration, in addition to evidence of decreased overall GABA content in epileptic brains (Minamoto et al., 1992), corroborates the idea of a multilevel GABAergic inadequacy for the control of DGGC excitability.

Second, pharmacologic properties of α5-GABAARs and δ-GABAARs are very different. The low GABA efficacy on δ-GABAARs, makes these receptors the preferential site of action of neurosteroids. In addition, δ-GABAARs are naturally insensitive to benzodiazepines, because these drugs need γ subunit in order to bind to GABAARs. Moreover, α5-GABAARs are typically insensitive to zolpidem. In our own findings, diazepam-mediated increase in DGGC tonic current was unchanged after pilocarpine-induced SE, and there is little to no anatomic evidence of α5-GABAAR up-regulation in the DG molecular layer, and on the contrary, α5 might even be slightly down-regulated (Zhang et al., 2007). Maintenance of DGGC tonic current at control levels in epileptic animals most likely relies on GABAARs other than α5-GABAARs, namely perisynaptic α4γ2-GABAARs or receptors containing only α4 and β subunits. Pharmacology of the tonic current mediated by these receptors will also be different, given the low sensitivity of α4-GABAARs to diazepam (Wisden et al., 1991; Benke et al., 1997). Perhaps the most important pharmacologic consequence of extrasynaptic GABAAR plasticity in TLE is the loss of a major fraction of δ-GABAARs. Functionally, DGGCs of epileptic animals do not respond to neurosteroid potentiation of tonic current, and in parallel, the ability of neurosteroids to lower dentate excitability turns into a net effect of increasing excitability (Zhang et al., 2007).

Before reaching final conclusions on δ-GABAAR plasticity in TLE and its functional and practical pharmacologic consequences, we also need to consider δ-GABAAR expression on interneurons. In the dentate, in molecular layer interneurons, the δ subunit is expressed paired with α1, rather than α4 subunits (Glykys et al., 2007), and the δ-GABAARs mediate a neurosteroid sensitive tonic conductance. Of interest, in the same mouse model of TLE, δ expression on interneurons appears to be increased, contrary to what happens in granule cells (Zhang et al., 2007). Tonic current is also increased in molecular layer interneurons cells, and neurosteroid sensitivity is maintained (Wei W, Mody I, unpublished data). Increasing tonic inhibition onto the inhibitory cells might have a dual effect. On one hand, slightly decreasing GABAergic drive onto DGGCs might help reduce their level of synchronization, since some interneurons are very efficient at synchronizing large populations of neurons (Cobb et al., 1995). More so if we keep in mind that a typical anatomic alteration in TLE is the formation of functional aberrant recurrent connections within DGGCs (Tauck & Nadler, 1985), which have been shown to facilitate the onset of DGGCs synchronization. On the other hand, some of the δ-GABAAR-labeled interneurons in the hippocampus might be neurogliaform cells (NGFCs) and ivy cells (Armstrong et al., 2011, 2012). NGFCs abundantly express δ-GABAARs (Oláh et al., 2009), and since they largely contribute to ambient GABA levels, decrease of their excitability would bring a net decrease in ambient GABA. Moreover, inhibiting the inhibitors can by itself lead to significant alterations in dentate excitability and bring basal excitability levels dangerously close to seizure threshold.

Most of the time the epileptic brain is capable of maintaining adequate and functional levels of excitation and inhibition, but compared to the healthy brain, it is a precarious equilibrium. Even small deviations can cause the inhibitory control to lose its grasp on the dentate gate. Typically TLE is sensitive to triggers, and there is much literature on the efficiency of stress as a trigger of seizures (Joëls, 2009). Catamenial epilepsy is a type of epilepsy during which seizures cluster around a specific period of the ovarian cycle (Herzog et al., 1997). In both these cases, neurosteroid fluctuations in the brain will be fast and large, and although they will fail to decrease DGGCs excitability, they will be able to inhibit interneurons by potentiating their already higher tonic current, possibly shifting the balance sufficiently to facilitate an epileptic event. Lastly, the role of ethanol as a precipitant of seizures might be played through its preferential δ-GABAARs enhancement at low concentrations (Hauser et al., 1988; Ng et al., 1988).

However, during epilepsy a neurosteroid basal tone is necessary, because blocking neurosteroid synthesis with finasteride greatly increases seizure frequency, whereas it has no apparent effect on control mice (Lawrence et al., 2010). Finasteride is a 5-α-reductase blocker commonly used in the treatment of certain prostatic cancers, and at lower doses, as a remedy for androgenic alopecia (Mella et al., 2010; Nacusi & Tindall, 2011). In one case report, finasteride increased, whereas progesterone decreased, seizure frequency in a patient with catamenial epilepsy (Herzog & Frye, 2003).

One way to reconcile these results is to think of neurosteroid modulation of neuronal excitability as a heterogeneous phenomenon: compartmentalization of neurosteroid synthesis in neurons and astrocytes will dissociate their effects on different types of neurons and populations of receptors. Therefore, the fact that in vitro DGGCs do not show neurosteroid sensitivity could mean that part of the recorded tonic current was already potentiated by neurosteroid locally synthetized in the slice. If neurosteroid levels were sufficiently elevated to reach a ceiling effect on the few remaining δ-GABAARs, then addition of neurosteroid will not cause a further increase in the tonic conductance. Therefore, in TLE, basal levels of neurosteroid are important to keep inhibitory–excitatory equilibrium, but at the same time the epileptic brain seems to be more sensitive to neurosteroid changes. Physiologic conditions of modified neurosteroid levels such as the ovarian cycle, pregnancy, and stress are accompanied by plastic homeostatic changes in δ-GABAARs, but we do not yet know if similar changes are taking place in the epileptic hippocampus (Maguire et al., 2005; Maguire & Mody, 2007, 2008; Maguire et al., 2009). Nonetheless, in the face of the GABAAR alterations in the dentate gyrus, rapid neurosteroid increases will likely favor more excitable states.

In conclusion, different hormonal pathways have the ability to regulate neuronal excitation. Neurosteroids are a potent class of hormones that enhance inhibition by positively modulating the efficacy of GABA at δ-GABAARs. Plasticity of neurosteroid-sensitive GABAARs has been consistently found in different models of TLE, and although neurosteroid effects on individual cell excitability have been tested, predicting the net effect of abrupt neurosteroid changes on the overall gain of a specific network in the epileptic brain is not straightforward. Stress and the ovarian cycle modify neurosteroid concentration in the brain and are also known seizure triggers in epileptic patients. Overall, neurosteroids appear capable of producing bimodal effects on excitability in the epileptic brain. Although basal neurosteroid synthesis seems to be necessary for control of seizure frequency, the epileptic brain is also more sensitive to neurosteroid changes, and inhibition is more likely to lose control over the dentate gate after substantial neurosteroid increase, since interneuronal sensitivity to neurosteroid is maintained while excitatory cells lose their ability to respond to neurosteroids. This bimodal behavior has to be taken into account when designing AEDs that act by potentiating the function of δ-GABAARs, especially in patients exposed to abrupt changes in endogenous neurosteroids.

Disclosure

  1. Top of page
  2. Summary
  3. Hormones in the Brain
  4. Interneurons, GABAA Receptors and Tonic Inhibition
  5. Neurosteroids and Tonic Inhibition
  6. Alterations of Neurosteroid-Sensitive GABAA Receptors in an Animal Model of TLE
  7. Disclosure
  8. References

None of the authors has any conflict of interest to disclose.

We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

References

  1. Top of page
  2. Summary
  3. Hormones in the Brain
  4. Interneurons, GABAA Receptors and Tonic Inhibition
  5. Neurosteroids and Tonic Inhibition
  6. Alterations of Neurosteroid-Sensitive GABAA Receptors in an Animal Model of TLE
  7. Disclosure
  8. References