Allosteric modulation of glycine receptors

Abstract

Inhibitory (or strychnine sensitive) glycine receptors (GlyRs) are anion-selective transmitter-gated ion channels of the cys-loop superfamily, which includes among others also the inhibitory γ-aminobutyric acid receptors (GABA_A receptors). While GABA mediates fast inhibitory neurotransmission throughout the CNS, the action of glycine as a fast inhibitory neurotransmitter is more restricted. This probably explains why GABA_A receptors constitute a group of extremely successful drug targets in the treatment of a wide variety of CNS diseases, including anxiety, sleep disorders and epilepsy, while drugs specifically targeting GlyRs are virtually lacking. However, the spatially more restricted distribution of glycinergic inhibition may be advantageous in situations when a more localized enhancement of inhibition is sought. Inhibitory GlyRs are particularly relevant for the control of excitability in the mammalian spinal cord, brain stem and a few selected brain areas, such as the cerebellum and the retina. At these sites, GlyRs regulate important physiological functions, including respiratory rhythms, motor control, muscle tone and sensory as well as pain processing. In the hippocampus, RNA-edited high affinity extrasynaptic GlyRs may contribute to the pathology of temporal lobe epilepsy. Although specific modulators have not yet been identified, GlyRs still possess sites for allosteric modulation by a number of structurally diverse molecules, including alcohols, neurosteroids, cannabinoids, tropeines, general anaesthetics, certain neurotransmitters and cations. This review summarizes the present knowledge about this modulation and the molecular bases of the interactions involved.

Introduction

Strychnine-sensitive glycine receptors (GlyRs) are pentameric anion channels. Five GlyR subunits have been cloned from mammalian tissue and designated α1–α4 and β (Lynch, 2004). Each GlyR subunit contains an amino-terminal extracellular domain, four transmembrane domains (TM) and a large intracellular loop between TM3 and TM4, which configure the ligand binding region, the ion channel pore and sites for intracellular modulation respectively (Lynch, 2004; Sine and Engel, 2006; Baenziger and Corringer, 2011).

In the adult, most GlyRs are composed of α1 and β subunits probably in a 2(α1)/3β stoichiometry (Grudzinska et al., 2005). This GlyR subtype is the main mediator of glycinergic inhibition in the adult CNS. Many of these GlyRs colocalize with the postsynaptic scaffolding protein gephyrin (Todd et al., 1995; Waldvogel et al., 2010). Early in development most GlyRs are α2 homomers, which become replaced around postnatal day 14 by α1β heteromers in most CNS areas (Malosio et al., 1991; Lynch, 2004). In a few selected areas, such as the retina, α2 persists however into adulthood. Like α1 subunits, α3 subunits are mainly found in the adult but their expression is spatially much more restricted. Immunohistochemistry and quantitative RT-PCR studies in mice have shown that α3-GlyR subunits are predominantly expressed in the superficial laminae of the dorsal horn (Harvey et al., 2004; Anderson et al., 2009) and in the respiratory network of the brainstem (Manzke et al., 2010). RNA edited α2- and α3-GlyRs may serve a peculiar function as extrasynaptic high affinity GlyRs in the hippocampus (Meier et al., 2005; Legendre et al., 2009). The gene encoding for the GlyR α4 subunit is a pseudogene in humans due to the presence of a premature stop codon (Simon et al., 2004).

The subunit composition of the ion channel complex and the arrangement of the different subunits within this complex determine its pharmacological profile. According to recent data, the glycine binding site of α/β heteromeric GlyRs is jointly formed by α and β subunits (Grudzinska et al., 2005), while β subunits interact with gephyrin and thereby mediate synaptic clustering of GlyRs (Pfeiffer et al., 1982; Schmitt et al., 1987; Kim et al., 2006). Although very few established drugs primarily act through inhibitory GlyRs (Laube et al., 2002), a number of endogenous messenger molecules and some drugs do modulate GlyRs function (Figure 1). Such compounds include different cations, in particular zinc, cannabinoids, neuroactive steroids, tropeines, alcohols, avermectins, butyrolactones and general anaesthetics. Most of these molecules do not directly interact with the glycine binding sites but rather bind to allosteric sites within the GlyR complex (Figure 2). It is at present not yet established to what extent these interactions contribute to physiology or to drug actions in vivo, but their existence clearly establishes a possibility for specific pharmacological intervention. The analysis of the molecular bases of the interaction of these compounds with GlyRs should hence foster the development of specific GlyR modulators.

Figure 1.

Allosteric modulation of GlyR subtypes. The scheme summarizes the interactions between several groups of allosteric modulators with different GlyR subtypes. Examples of some representative chemical structures for each group of compounds are shown.

Figure 2.

Molecular sites for positive allosteric modulators of GlyR function. Critical residues are shown for several allosteric modulators that exert positive effects on GlyR function. The amino acid positions described were identified in functional experiments performed on the α1-GlyR mutants. The molecular sites involved in the effects elicited by neuroactive steroids, some cannabinoid ligands (i.e. endocannabinoids and synthetic cannabinoids) and glutamate receptor ligands remain to be defined. Data are from Mascia et al. (1996b), Mihic et al. (1997), Laube et al. (2000), Lynch et al. (1998), Maksay et al. (2009), Lynagh and Lynch (2010), Xiong et al. (2011) and Yevenes et al. (2008).

Possible indications of GlyR as a therapeutic target

Early knowledge about possible roles of glycinergic neurotransmission in physiological functions or in diseases has mainly been obtained through pharmacological blockade of GlyR with the rodent poison strychnine (for a review see Callister and Graham, 2010). These early studies have identified a critical role of GlyRs in the control of muscle tone. Severe cramps of the skeletal musculature are the leading symptom of strychnine poisoning. Apart from these motor symptoms altered sensory perception such as an increased sensitivity to acoustic or tactile stimuli has also frequently been observed. Further evidence for the involvement of glycinergic inhibition in sensory processing has come from studies in several strains of GlyR mutant mice (spasmodic, oscillator and spastic) which carry mutations in the GlyR α1 subunit (spasmodic, oscillator) or in the β subunit (spastic) (Buckwalter et al., 1994; Mülhardt et al., 1994; Ryan et al., 1994). These mice do not only exhibit increased muscle tone but also show a strong hyperekplexic phenotype, very much reminiscent of human startle disease (Koch et al., 1996). In fact, mutations in GlyR subunit genes are frequently found in human patients suffering from hyperekplexia/startle disease (Rees et al., 2001).

There is also evidence that part of the spinal component of inflammatory hyperalgesia (i.e. an increased sensitivity to painful stimuli as a consequence of peripheral inflammation) comes from diminished glycinergic inhibition caused by the phosphorylation and inhibition of α3-GlyRs (Ahmadi et al., 2002; Harvey et al., 2004; Reinold et al., 2005). In the spinal cord, these GlyRs are largely confined to the superficial dorsal horn, the main termination area of nociceptive afferent nerve fibres. This result fits nicely to early reports showing that exaggerated nociceptive responses can be triggered by intrathecal injection of strychnine in rats (Beyer et al., 1985; Yaksh, 1989). Very recent evidence indicates that α3-GlyRs also serve an important function in brainstem respiratory control where their dephosphorylation through serotonin 5-HT1A receptor activation antagonizes opioid-induced respiratory depression (Manzke et al., 2010).

Although glycinergic innervation is largely confined to the spinal cord, brainstem and cerebellum, GlyRs are widely expressed also in the forebrain, where they might become activated by ambient glycine. The affinity to glycine of un-edited receptors is normally too low for such an activation. However, high affinity receptors can be generated through cytidine deamination of GlyR transcripts (RNA editing) (Meier et al., 2005). This RNA editing gives rise to novel isoforms of α2 and α3-GlyRs carrying a proline to leucine point mutation (α2[P192L] and α3[P185L]) (Meier et al., 2005; Legendre et al., 2009). Recent evidence suggests that such high affinity extrasynaptic GlyRs contribute to pathological changes in temporal lobe epilepsy through the silencing of hippocampal neurons (Eichler et al., 2008).

In the disease states discussed above, GlyR function is affected through inherited mutations, RNA editing or posttranslational modifications such as phosphorylation. Patients suffering from diseases caused by diminished inhibition would probably benefit most from facilitated glycinergic inhibition, for example, through positive allosteric GlyR modulators, while in temporal lobe epilepsy an inhibition of GlyRs might be desirable. The following sections address the mechanisms of and the molecular sites for a positive allosteric modulation.

Cannabinoid ligands

A series of reports published starting in 2005 focused on a possible role of endocannabinoids and structurally or functionally related molecules as GlyR modulators. Endocannabinoids are endogenous activators of G-protein coupled cannabinoid receptors (CB1 and CB2 receptors) (Piomelli, 2003). N-arachidonoyl ethanol amide (AEA, also known as anandamide) was the first endocannabinoid discovered, followed by 2-arachidonyl-glycerol (2-AG). Both are lipid signalling molecules, structurally related to arachidonic acid. Although many lines of evidence indicate that G-protein coupled cannabinoid receptors are the primary targets of 2-AG and AEA, several studies showed that they interact with additional targets including several ion channels (Oz, 2006).

A direct modulation of GlyR by 2-AG and AEA was first reported in hippocampal neurons where both endocannabinoids reduced the amplitude of glycinergic membrane currents and altered their rise time, desensitization and deactivation kinetics in a concentration-dependent manner (Lozovaya et al., 2005). This modulation was insensitive to CB1 receptor antagonists (SR141716A) and remained intact when the recorded cell was perfused with the ubiquitous G-protein inhibitor GDP-β-S (Lozovaya et al., 2005). Direct modulation of GlyR by AEA has also been found in oocytes expressing recombinant α1-GlyRs (Hejazi et al., 2006). Neither SR141716A nor the cannabinoid reuptake inhibitor AM404 prevented the potentiating actions of AEA, again indicating that modulation occurred independent of CB1 receptors.

Following the identification of the two endocannabinoids 2-AG and AEA, additional, structurally related endogenous molecules were discovered (Huang et al., 2001). Several of these, such as N-arachidonoyl glycine (NA-Gly) and N-arachidonoyl serine (NA-Ser), bind CB1 receptors only very weakly, but still modulate GlyRs or other ion channels (Guo et al., 2008; Yang et al., 2008; Barbara et al., 2009). Other molecules with agonistic activity at CB1 or CB2 receptors but structurally unrelated to endocannabinoids also modulate GlyRs. Among these are some ingredients of the Cannabis sativa plant (Δ⁹-tetrahydrocannabinol [Δ⁹-THC], cannabidiol) and several synthetic CB₁ and/or CB₂ receptor ligands (HU-210, WIN 55,212-2), which either potentiate or inhibit GlyR currents, sometimes in a subunit-specific manner (compare Table 1). Although a consistent picture has yet to emerge, these data suggest that different molecular determinants exist in the target protein for CB receptor activation and GlyR modulation.

Table 1.

Cannabinoid ligand effects on native and recombinant GlyRs

Compound	Native GlyRs(EC50 or concentration range examined)	Comments	Reference
AEA	↓ (0.1–1 µM) Neonatal rat hippocampal pyramidal neurons1	100 µM glycine (≈EC50)	Lozovaya et al., 2005
2-AG	↓ (1 µM) Neonatal rat hippocampal pyramidal neurons1
WIN 55,212-2	Little effect on amplitude, but τdes and τon decreased (0.1–10 µM)
AEA	↑ (230 nM) Acutely dissociated VTA neurons	5 µM glycine	Hejazi et al., 2006
Δ9-THC	↑ (115 nM) Acutely dissociated VTA neurons
Δ9-THC	↑ (0.03–1 µM) Cultured spinal neurons	10 µM glycine (≈EC2)	Xiong et al., 2011
Meth-AEA	No effect on the amplitude or kinetics of glycinergic mIPSCs from spinal cord dorsal horn slices	5 µM meth-AEA	Anderson et al., 2009

	Recombinant GlyRs (EC50 or concentration range examined)
	α1	α1/β	α2	α3
Δ9-THC	↑ (86 nM)	↑(73 nM)	–	–	3–5 µM glycine	Hejazi et al., 2006
Δ9-THC	↑ (0.03–50 µM)2	–	↑ (0.03–50 µM)2	↑ (0.03–50 µM)2	≈EC2 (subunit-dependent)	Xiong et al., 2011
AEA	↑ (320 nM)	↑ (320 nM)	–	–	3–5 µM glycine	Hejazi et al., 2006
AEA	↑ (38 nM)	↑ (75 nM)	No effect	No effect	≈EC10 (subunit-dependent)	Yang et al., 2008
HU-210	↑ (270 nM)	↑	↓ (90 nM)	↓ (50 nM)
WIN 55,212-2	No effect	No effect	↓ (0.22 µM)	↓ (86 nM)
N-arachidonyl-glycine	Complex action3	Complex action3	↓(3.03 µM)	↓ (1.32 µM)
Ajulemic acid	↑ (9.7 µM)	↑ (12.4 µM)	–	–	10 µM glycine	Ahrens et al., 2009b
Cannabidiol	↑ (12.3 µM)4	↑ (18.1 µM)4	–	–	10 µM glycine	Ahrens et al., 2009a

¹Qualitatively similar effects were obtained in cerebellar Purkinje neurons.

²EC₅₀ values were not reported. The sensitivity to Δ⁹-THC was α1 = α3 > α2.

³Complex actions with initial potentiation and subsequent inhibition.

⁴Direct activation was observed at 10- to 20-fold higher concentrations.

2-AG, 2-arachidonyl-glycerol; AEA, N-arachidonoyl ethanol amide; GlyR, glycine receptor; Δ⁹-THC; Δ⁹-tetrahydrocannabinol.

The studies discussed above consistently found that most of the cannabinoid related compounds did not directly activate GlyRs but in most cases caused a leftward shift of the glycine concentration response curve. Another important aspect is that native GABA_A receptors (in rat ventral tegmental area neurons) and recombinant α2β3γ2 GABA_A receptors expressed in Xenopus laevis oocytes were not modulated by AEA (Hejazi et al., 2006).

First analyses of possible molecular sites for these allosteric effects were performed by Hejazi and coworkers, who found that the sensitivity to AEA was similar in homomeric α1 and heteromeric α1β-GlyRs indicating that α subunits were sufficient for this modulation (Hejazi et al., 2006). They next investigated the influence of a serine to glutamine amino acid exchange in α1 at position 267 (S267Q), which was previously shown to abolish the potentiation of GlyRs by ethanol and general anaesthetics (Mihic et al., 1997). No change in the potentiating action of AEA or Δ⁹-THC was found in this mutant. However, mutation of the serine 267 into an isoleucine (I), which also abolishes GlyR potentiation by ethanol (Mihic et al., 1997), prevented potentiation by three molecules structurally-related to Δ⁹-THC (cannabidiol, HU210 and ajulemic acid) (Foadi et al., 2010) suggesting a role of TM2 residues for the actions of these cannabinoids ligands. More recently, Xiong and coworkers demonstrated that potentiation of α1- and α3-GlyRs by Δ⁹-THC involves a TM3 serine residue (S296 on α1 or S307 on α3-GlyRs), which likely contributes to a direct interaction of Δ⁹-THC via hydrogen bonds (Xiong et al., 2011, see also Figure 2).

Over the last several years, convincing evidence has accumulated for a direct modulatory action of cannabinoid-related compounds on recombinant GlyRs. Data supporting a significant contribution of these effects to the in vivo actions of (endo-)cannabinoids were however lacking until recently. The report by Xiong et al. (2011) provides the first evidence in support of an in vivo relevance showing that mice lacking α3-GlyRs exhibit a pronounced reduction in Δ⁹-THC-induced analgesia. An important piece of information which is still missing in the puzzle is data demonstrating a direct amplification or prolongation of glycinergic synaptic currents by (endo-)cannabinoids.

A second issue which is particularly relevant, when cannabinoid-related molecules are considered as lead structures for the development of GlyR modulators, is their lack of specificity. Almost all of these molecules also interfere with the function of other ion channels (Oz, 2006, see also Table 2) and many of them also exhibit activity at CB1 or CB2 receptors. Again, the report by Xiong et al. (2011) provides new insights. Introduction of slight chemical modifications to the Δ⁹-THC molecule significantly decreased affinity to CB1 receptors while fully retaining activity at GlyRs. Although comprehensive analyses of the molecular determinants are definitely still needed, the recent studies indicate that some of the cannabinoid-related molecules discussed above may constitute interesting lead compounds for the development of GlyR modulators.

Table 2.

GlyR positive allosteric modulators and their additional targets

Group	Representative ligands	GlyR (relative potency1)	Additional targets	References
Volatile anaesthetics	Isoflurane, enflurane	↑↑	GABAA-R, voltage-gated Ca2+ channels, NMDA-R, 2P-domain K+ channels	Yamakura et al., 2001; Franks, 2008
Intravenous anaesthetics	Propofol	↑↑	GABAA-R, L-type Ca2+ channels, 11β-hydroxylase	Yamakura et al., 2001; Rudolph and Antkowiak, 2004; Franks, 2008
n-alcohols	Ethanol	↑	GABAA-R, NMDA-R, GIRK channels	Aguayo et al., 2002; Harris et al., 2008
Avermectins	Ivermectin	↑↑2	nAch-Rs, GABAA-R, P2X4-R	Krusek and Zemkova, 1994; Krause et al., 1998; Adelsberger et al., 2000; Silberberg et al., 2007; Jelinkova et al., 2008
Tropeines	Tropisetron	↑↑↑/↓↓3	5HT3-R	Thompson and Lummis, 2007
Cannabinoid ligands	Anandamide, THC	↑↑↑/↓↓	CB-R, TRPV1, nAch-Rs/5HT3-R	Piomelli, 2003; Oz, 2006
Bivalent cations	Zn2+	↑↑/↓↓3	GABAA-R, NMDA-R, TrkB-R	Smart et al., 2004; Mony et al., 2009; Sensi et al., 2009
Glutamatergic ligands	AP5, NMDA	↑↑	NMDA-R	Dingledine et al., 1999

Potentiation or inhibition ranges: ↑↑↑ or ↓↓↓, nM; ↑↑ or ↓↓, µM; ↑ or ↓, mM.

¹Mainly from functional studies using electrophysiology (see text).

²Direct activator.

³Biphasic modulation, inhibition at >10–50 µM.

Ethanol

Evidence from biochemical and electrophysiological experiments consistently indicates that GlyR currents are potentiated by ethanol at concentrations reached in humans after moderate ethanol intake. This potentiation originates from a decrease in the glycine EC₅₀ without a change in maximal currents (Aguayo et al., 1996; Mihic, 1999). Potentiation of GlyRs by ethanol is apparently not restricted to certain CNS areas but occurs in neurons throughout many parts of the CNS, including spinal cord, hippocampus, hypoglossal nucleus and ventral tegmental area (Aguayo et al., 1996; Jiang and Ye, 2003; Eggers and Berger, 2004). Experiments performed in motoneurons from brainstem and spinal cord slices have shown that ethanol increases the amplitude of glycinergic postsynaptic currents suggesting that modulation of synaptic GlyRs by ethanol could potentially explain some of the alterations caused by ethanol in motor control and respiratory rhythms (Gibson and Berger, 2000; Ziskind-Conhaim et al., 2003; Eggers and Berger, 2004).

Homomeric α1-GlyRs are more sensitive to ethanol than α2-GlyRs especially at concentrations below 100 mM (Mascia et al., 1996b; Perkins et al., 2008; Yevenes et al., 2010). This differential sensitivity correlates well with data from neuronal preparations, in which neonatal GlyRs (mostly α2-GlyRs) were less sensitive to ethanol than mature (α1) GlyRs (Tapia and Aguayo, 1998; Eggers et al., 2000; Sebe et al., 2003).

Most of the knowledge about the molecular mechanisms underlying the modulation of GlyRs by ethanol originally came from the electrophysiological analysis of a set of chimeric GlyR α1/GABA_Aρ1 receptors (Mihic et al., 1997). This seminal report identified residues in TM2 (S267) and TM3 domains (A288) which abolish the ethanol sensitivity of α1-GlyRs. Subsequent experiments combining site-specific mutagenesis, molecular modelling and covalent binding of alcohol analogues to cysteine mutants consistently determined that TM2 and TM3 residues jointly shape a water-filled cavity serving as an ethanol-binding pocket (Ye et al., 1998; Mascia et al., 2000). Other studies showed that the extracellular loop 2 and TM1 residues also play a role in the alcohol modulation of GlyRs, although it is not clear if they shape additional binding pockets (Crawford et al., 2008; Lobo et al., 2008). These studies clearly demonstrate that both ethanol binding pockets and regulatory elements for the ethanol actions are within the TM domains of GlyRs. However, ethanol sensitivity can also be effectively controlled by intracellular signalling, possibly suggesting that part of the ethanol actions occur indirectly through other ethanol-sensitive proteins. For instance, the ethanol-induced potentiation of recombinant and native GlyRs is attenuated by protein kinase C inhibitors (Mascia et al., 1998; Jiang and Ye, 2003) and by ct-GRK2, a specific G-protein βγ sequester peptide (Yevenes et al., 2008). The importance of the Gβγ signalling for the alcohol effects on GlyRs also has been demonstrated recently using Gβγ-insensitive α1-GlyRs, in which the mutation of two intracellular residues (KK385-386) attenuated ethanol effects without altering potentiation induced by general anaesthetics (Yevenes et al., 2008). Additionally, a recent report also showed that the differential ethanol sensitivity of α1- and α2-GlyRs can be better explained by a selective Gβγ modulation rather than by specific TM ethanol-binding pockets, which are conserved between these isoforms (Yevenes et al., 2010). It is still a matter of debate whether direct or indirect actions are more relevant, but conceivably both the direct binding of ethanol to the receptor (reviewed in Harris et al., 2008) and the indirect modulation of signalling components by ethanol (reviewed in Morrow et al., 2004) could be equally important and act cooperatively to elicit the final effects on GlyRs.

The physiological importance of the molecular sites for the ethanol actions in vivo has been investigated through genetic approaches in mice carrying the ethanol-insensitive S267Q mutation in the α1-GlyR gene. Transgenic expression of S267Q mutated GlyR in mice decreased ethanol sensitivity in behavioural assays without inducing apparent behavioural changes in the absence of alcohol (Findlay et al., 2002). Although these results support the importance of this GlyR site for alcohol actions in vivo, they should be interpreted cautiously as a subsequent publication of the same group investigating S267Q point-mutated (‘knock-in’) mice has yielded different results. Mice homozygous for the S267 point mutation exhibited spontaneous seizures and died 3 weeks after birth. Heterozygous mice survived but still displayed a severe increase in the acoustic startle responses (Findlay et al., 2003). In vitro experiments demonstrated that the S267Q mutation in α1-GlyR significantly reduced the glycine-evoked chloride uptake in spinal cord synaptoneurosomes from heterozygous knock-in mice and dramatically disrupted receptor function at the single-channel level (Findlay et al., 2003).

General anaesthetics

Many studies on recombinant GlyRs consistently demonstrate that volatile anaesthetics, such as isoflurane, enflurane, halothane and sevoflurane potentiate homomeric α1-GlyR currents at anaesthetic concentrations (Downie et al., 1996; Mascia et al., 1996a; Krasowski and Harrison, 1999; Yamakura et al., 2001). This potentiation is not specific for α1-GlyRs, as homomeric α2-GlyRs are also sensitive to isoflurane (Harrison et al., 1993), while α3-GlyRs remain to be investigated. The anaesthetics studied were unable to activate GlyRs by themselves (Harris et al., 1995; Downie et al., 1996; Krasowski and Harrison, 1999), but rather caused a leftward shift of the glycine concentration–response curve. These effects have been reproduced in native receptors. Isoflurane potentiated the glycine-activated currents in rat medullary neurons (Downie et al., 1996), and prolonged the decay kinetics and increased the frequency of mIPSCs in rat trigeminal nucleus and spinal motoneurons (Yamashita et al., 2001; Cheng and Kendig, 2002). Because glycinergic inhibition is largely confined to the hindbrain and spinal cord, it is unlikely that the loss of consciousness by volatile anaesthetics is caused via an interaction with GlyRs. However, immobility is an action of volatile anaesthetics, which is much more likely related to interactions with GlyR (Sonner et al., 2003; Rudolph and Antkowiak, 2004). In line with this idea, isoflurane, enflurane and sevoflurane indeed significantly reduced spontaneous action potential firing in neurons recorded in organotypic slice cultures of the rat ventral horn (Grasshoff and Antkowiak, 2004; 2006;). An interaction with GlyRs might also be relevant for sensory processing at the spinal dorsal horn level. Extracellular recordings from wide-dynamic range neurons in intact rats have shown that halothane induced depression in the responses to thermal and mechanical noxious stimuli and that this depression was partially reversed by strychnine at doses which had no per se effect on wide-dynamic range neuron firing (Yamauchi et al., 2002). In line with these studies, in vivo experiments performed in rats demonstrated that spinal GlyRs are important, although not the only, mediators of the isoflurane-induced immobility (Zhang et al., 2003).

Intravenous anaesthetics may also have some effect on GlyR function, but these actions are more controversial. Propofol displayed significant modulatory activity at GlyR, but the degree of this modulation was much less than that of volatile anaesthetics, in particular at clinically relevant concentrations (Pistis et al., 1997; Krasowski and Harrison, 1999). Homomeric α1-GlyRs, α1/β heteromers and homomeric α2-GlyRs are similarly sensitive to propofol (Mascia et al., 1996b; Pistis et al., 1997), while at least α1-GlyRs appeared to be insensitive to etomidate (Mascia et al., 1996a; Pistis et al., 1997). Despite the lack of information on the sensitivity of other GlyR subunits or subunit combinations, available evidence suggests that GlyRs are unlikely to play a major role in the in vivo effects of the intravenous anaesthetics (but see Nguyen et al., 2009). This is also supported by reports, which showed that propofol-induced immobility was produced exclusively via spinal GABA_A receptors (Sonner et al., 2003; Grasshoff and Antkowiak, 2004).

The molecular determinants of GlyR modulation by anaesthetics have been worked out hand-in-hand with those of GABA_A receptors. In fact, pioneering studies performed in the late 1990s found sites for volatile anaesthetics within TM domains of GlyRs through the analysis of chimeric receptors between the enflurane-sensitive α1-GlyRs and enflurane-insensitive ρ1-GABA_A receptors (Mihic et al., 1997). This and other studies have consistently shown that specific residues within TM2 and TM3 domains of α1-GlyRs potentially shape an intra-subunit cavity which serves as a general anaesthetic binding pocket and which also acts as an acceptor for ethanol and other n-alcohols with longer carbon chains (reviewed in Krasowski and Harrison, 1999; Lobo and Harris, 2005). Unfortunately, the knowledge regarding the molecular sites for anaesthetics on GlyRs has not yet been translated into genetic mouse models. This would be necessary in order to address the role of GlyRs on the general anaesthetic actions in vivo.

Glutamate

A recent publication (Liu et al., 2010) provides strong evidence that glutamate, the principal fast excitatory neurotransmitter in the CNS, can act as a positive allosteric GlyR modulator. This potentiation was seen in spinal neurons in culture and in slices as well as in HEK293 cells transiently expressing GlyRs. Potentiation of GlyR currents manifested in increased single channel open probability and occurred not only by glutamate but also by NMDA, AP5, kainate, quisqualate, aspartate and kynurenic acid, while CNQX and NBQX inhibited GlyR currents indicating that the pharmacology of this modulation did not match with that of any known glutamate receptor. Experiments performed in isolated membrane patches suggest a direct binding of glutamate to the GlyR channel complex. Homomeric α1-GlyR currents were doubled by glutamate, while potentiation of α1/β heteromeric channels was in the rage of about 40–60%, possibly suggesting that the putative binding site resides on the GlyR α1 subunit. Potentiation of glycinergic inhibition by glutamate may provide an extremely fast feedback mechanism for the maintenance of balanced synaptic excitation and inhibition. Although the relevant binding sites have not yet been determined and because the findings certainly require independent verification, this previously unknown modulation may point to an additional possibility for therapeutic intervention with GlyRs.

Ivermectin

Avermectins are a family of macrocyclic lactones derived from the bacterium Streptomyces avermitilis and commonly used as antiparasitic and insecticide agents. They act mainly through an allosteric modulation or a direct activation of glutamate-gated chloride channels (GluCls) expressed by nematodes and insects (Wolstenholme and Rogers, 2005). Ivermectin is one member of this group whose activity on invertebrate GluCls has been characterized in detail in recombinant systems (Arena et al., 1992; Cully et al., 1994; 1996; Kane et al., 2000). Notably, ivermectin also modulates cationic and anionic ligand-gated ion channels including GlyRs in vertebrate (see Table 2). Early electrophysiological studies showed that ivermectin influences recombinant homomeric α1 and heteromeric α1β-GlyRs at submicromolar concentrations (Shan et al., 2001). More recent studies revealed that different amino acid substitutions of the TM3 residue A288 differentially affected ivermectin's action on α1-GlyR currents. The mutation A288G increased the ivermectin sensitivity to the nanomolar range, whereas the A288F substitution completely abolished its agonistic actions (Lynagh and Lynch, 2010; Figure 2). Equivalent mutations in the corresponding residue in a nematode GluCl ion channel showed a similar pattern of effects suggesting that avermectins affect vertebrate GlyRs and nematode GluCl ion channels through similar molecular sites (Lynagh and Lynch, 2010). This is further supported by the observation that homologous residues act as binding sites for ethanol and general anaesthetics on GABA_A and GlyRs (reviewed in Krasowski and Harrison, 1999; Yamakura et al., 2001; Lobo and Harris, 2005). The direct activation of GlyRs by ivermectin makes this compound an interesting template to design GlyR ligands with agonistic activity. However, ivermectin is not specific for GlyRs (see Table 2) and furthermore inhibits GlyRs in some preparations (Dawson et al., 2000). A better understanding of the mechanisms underlying the effects of ivermectin on GlyRs will be necessary to design new analogues with improved selectivity.

Neuroactive steroids

Endogenous neurosteroids are cholesterol metabolites produced locally in the CNS. They induce fast changes in neuronal excitability through a direct interaction with ion channels. Best established is the facilitating action on GABA_A receptors by 5α-pregnan-3α-ol-20-one (3α,5α-tetrahydroprogesterone; 3α,5α-THPROG; allopregnanolone) and 5β-pregnan-3α-ol-20-one (3α,5β-THPROG; pregnanolone), collectively called 3α-reduced neurosteroids. These neurosteroids neither directly activate nor potentiate GlyRs (Pistis et al., 1997; Lambert et al., 2001; Weir et al., 2004). 3α,5β-THPROG in fact causes a small but significant inhibition of spinal GlyR currents (Wu et al., 1997; Fodor et al., 2006) and 3β-sulphates of pregnenolone have been shown to inhibit recombinant GlyRs expressed in oocytes (Maksay et al., 2001).

By contrast, synthetic neurosteroids such as minaxolone, Org20599 and alphaxalone significantly enhance homomeric α1-GlyR currents in recombinant systems (Weir et al., 2004; Ahrens et al., 2008). Another recent report has shown that two synthetic pregnanolone analogues potentiate homomeric α3-GlyR currents in a voltage-dependent fashion (Jin et al., 2009). Although facilitation occurred in these studies with EC₅₀ values approximately 10-fold higher than those required for GABA_A receptors and with generally lower efficacies (Weir et al., 2004), experiments on spinal dorsal horn neurons have shown that low micromolar concentrations of minaxolone prolong the decay time kinetics of glycinergic mIPSCs in lamina II neurons (Mitchell et al., 2007). At 10 µM, minaxolone additionally increased the amplitude, but not the frequency of glycinergic mIPSCs. Tonic glycinergic currents found in the same dorsal horn neurons were insensitive to minaxolone (Mitchell et al., 2007). The molecular sites involved in the modulation elicited by synthetic steroids on different GlyR subtypes are still not investigated in depth.

Tropeines

Tropeines were originally identified as potent 5-HT₃ receptor antagonists. Tropisetron, also known as ICS-205,930, is one of the best-known compounds of this group. It is used mainly as an anti-emetic following chemotherapy due to its ability to target 5-HT₃ receptors involved in vomiting reflexes. An increasing body of evidence has shown that tropeines also allosterically modulate GlyRs of different subunit composition. Pioneering electrophysiological recordings in cultured spinal neurons have revealed that two tropeines, MDL-72222 and tropisetron, were able to potentiate GlyR chloride currents at nanomolar concentrations (Chesnoy-Marchais, 1996). In contrast, higher micromolar concentrations caused inhibition. Subsequent studies determined that potentiation only occurred in the presence of GlyR agonists, depended on the agonist concentration, and was also present in outside-out patches (Chesnoy-Marchais, 1996; Supplisson and Chesnoy-Marchais, 2000; Yang et al., 2007). The potentiation elicited by tropisetron remained unaltered in the presence of zinc, ethanol or propofol, suggesting different binding sites and mechanisms (Chesnoy-Marchais, 1999). Interestingly, tropisetron also displayed subunit-specificity. Studies in recombinant GlyRs showed that tropisetron potentiated homomeric α1 but inhibited homomeric α2-GlyRs. Furthermore the expression of β subunits significantly increased the potentiation sensitivity of α1 and switched α2-GlyR inhibition to potentiation. These results suggest that the tropeine potentiating site lies within the α-α or α-β interface (Supplisson and Chesnoy-Marchais, 2000). Other studies also found that α2-GlyR was more effectively inhibited by tropisetron than α1-GlyR, but in contrast, they did not find any potentiation even in the presence of β subunits (Maksay et al., 1999). Despite these differences, the electrophysiological data correlated well with binding studies in recombinant and native membrane preparations. For example, several tropeines have been shown to inhibit ³[H]strychnine binding to GlyRs with high nanomolar affinity. In addition, they increase the glycine potency to displace ³[H]strychnine, suggesting direct effects on glycine binding sites (Maksay, 1998; Maksay et al., 2004). In general terms, structure-activity analysis suggests that the tropeine ring itself, the tropeine nitrogen, an aromatic ring and a carbonyl group are necessary for binding and functional potentiation (Maksay, 1998; Chesnoy-Marchais et al., 2000; Maksay et al., 2004). The tropeine ring, on the other hand, appears to be a primary requirement for functional inhibition (Yang et al., 2007; Maksay et al., 2009).

Recently, several studies addressed the location of the tropeine binding sites on GlyRs. In agreement with studies performed in 5-HT₃ receptors (Yan and White, 2005; Joshi et al., 2006), tropeines appear to bind to cavities within the extracellular domain located close to the ligand binding sites. Using recombinant GlyRs, Yang et al. (2007) showed that mutations to N102 in the α1, but not in the β subunit (N125), abolished tropisetron inhibition without affecting the potentiation. Subsequent work performed with a structurally related tropeine (3α-(3′-methoxy-benzoyloxy) nortropane, MBN) determined that other amino acid substitutions close to the agonist-binding domain of α1-GlyRs also alter the MBN inhibition or potentiation of GlyRs (see Figure 2, Maksay et al., 2009). In addition, homology models and molecular docking simulations also suggest that the biphasic modulation elicited by tropeines on GlyRs is likely to involve different docking modes in adjacent binding sites within the agonist-binding region (Maksay et al., 2009).

The high affinity binding and the remarkable sensitivity of GlyRs to tropeines makes this group of compounds one of the most promising candidates for the development of specific drugs targeting GlyRs. Despite the existence of some interesting differences between the chemical determinants required for tropeine binding to GlyRs and 5-HT₃ receptors, most tropeines still bind and modulate 5-HT₃ receptors with high affinity (Maksay et al., 2004). In addition, the biphasic nature of tropeine-GlyR modulation and the significant overlap between the requirements for potentiation and inhibition is also an important impediment to their use as enhances of GlyR function. A better understanding of the mechanisms underlying the potentiation of GlyR subtypes by tropeines will hopefully lead to new tropeine derivatives lacking glycinergic inhibition and 5-HT₃ receptor binding.

Zinc

The interaction of GlyRs with the cation zinc is probably at present the best characterized form of allosteric modulation of GlyR. Previous research has not only consistently established the molecular sites involved, but work in point-mutated mice has also firmly established a physiological role of this modulation. Zinc modulates GlyRs in a biphasic manner. Potentiation dominates at low (<10 µM) concentrations while inhibition occurs at higher concentrations (>10 µM) (Bloomenthal et al., 1994; Doi et al., 1999; Laube et al., 2000). This bidirectional modulation involves different molecular sites. Potentiation is due to an increase in the affinity of GlyRs to glycine, while inhibition occurs through reduced efficacy. Amino acids involved in the potentiation by zinc are D80, E192, E194 (Lynch et al., 1998; Laube et al., 2000), while inhibition involves H107, H109, T112 and T133 (all positions refer to α1-GlyR) (Harvey et al., 1999; Laube et al., 2000; Miller et al., 2005). The different GlyR isoforms differ in their susceptibility to modulation by glycine. Zinc inhibits α2-GlyR and α3-GlyR to a lesser degree than α1-GlyR. This difference is apparently due to the substitution of amino acid H107 in α1-GlyR by an asparagine residue in the corresponding positions in α2- and α3-GlyR. Generation of a point-mutated mouse carrying a D80A substitution, which largely ablates the potentiating effects of zinc without changing glycine sensitivity, expression level or receptor trafficking to the synapse, revealed a physiological function of this modulatory site (and of zinc itself) in spinal cord neuronal circuits (Hirzel et al., 2006). Homozygous D80A point-mutated mice exhibit a progressive hyperekplexia-like phenotype starting about at day P12, when α1-GlyRs replace embryonic α2-GlyRs. Whether these zinc modulatory sites are suitable for therapeutic targeting is however at present not known.

Conclusions

Pharmacological modulation of glycinergic inhibition could represent a novel therapeutic strategy against a variety of diseases involving altered synaptic inhibition primarily in the spinal cord and brain stem but possibly also at supraspinal sites. Several endogenous molecules including neurotransmitters and neuromodulators, and exogenous substances such as anaesthetics and alcohols have been identified that modulate GlyR function. As most pathologies linked to GlyR dysfunction involve diminished GlyR activity, positive allosteric modulation appears desirable in the majority of cases. Most currently available GlyR modulators are rather promiscuous and by no means specific for GlyRs (compare Table 2). These compounds are therefore not suitable for a therapeutic approach targeted specifically towards GlyRs. However, for several of them, direct modulation through allosteric sites is either firmly established or very likely. The existence of putative distinct sites for allosteric modulation on GlyRs, however, indicates future possibilities for a specific modulation of GlyR subtypes by novel synthetic ligands. This perhaps optimistic view is supported by the recent report which showed that an unbiased high throughput screening approach led to the identification of several highly specific GlyR modulator peptides (Tipps et al., 2010). A comprehensive mapping of the molecular sites and mechanism involved will certainly facilitate the identification and development of small molecules specifically targeting GlyRs. In the absence of high-resolution structures for GlyRs (and hence also of structural data of these receptors with bound allosteric modulators or agonists), our knowledge is restricted to what can be inferred from functional studies using recombinant mutant receptors. Alternatively, advances might also come from the analysis of structurally related channels. High-resolution X-ray structures have recently been obtained from bacterial pentameric ligand-gated ion channel (reviewed recently by Baenziger and Corringer, 2011). For the bacterial proton-activated ion channel G. violaceus ligand-gated ion channel crystal structures have even been obtained with a general anaesthetics bound (Nury et al., 2011). It is to be hoped that such data will foster future structure-function studies on GlyRs, for example, through improved homology modelling and molecular dynamic simulations. Through these and other new approaches, the discovery and development of new synthetic drugs targeting GlyRs with improved specificity and efficacy appears to be not too far fetched.

Glossary

Abbreviations

2-AG

2-arachidonyl-glycerol

3α,5α-THPROG

3α,5α-tetrahydroprogesterone, allopregnanolone

5β-pregnan-3α-ol-20-one

3α,5β-THPROG, pregnanolone

AEA

N-arachidonoyl ethanol amide

CB receptor

cannabinoid receptor

GlyR

glycine receptor

NA-glycine

N-arachidonoyl glycine

TM

transmembrane

Conflicts of interest

The authors state no conflict of interests.