Glycine Receptors in Spinal Nociceptive Control—An Update

Abstract

Diminished inhibitory control of spinal nociception is one of the major culprits of chronic pain states. Restoring proper synaptic inhibition is a well-established rational therapeutic approach explored by several pharmaceutical companies. A particular challenge arises from the need for site-specific intervention to avoid deleterious side effects such as sedation, addiction, or impaired motor control, which would arise from wide-range facilitation of inhibition. Specific targeting of glycinergic inhibition, which dominates in the spinal cord and parts of the hindbrain, may help reduce these side effects. Selective targeting of the α3 subtype of glycine receptors (GlyRs), which is highly enriched in the superficial layers of the spinal dorsal horn, a key site of nociceptive processing, may help to further narrow down pharmacological intervention on the nociceptive system and increase tolerability. This review provides an update on the physiological properties and functions of α3 subtype GlyRs and on the present state of related drug discovery programs.

1. Introduction

Inhibitory neurotransmission plays a crucial role in the maintenance of a physiologically meaningful state of pain sensitivity and helps separating innocuous and noxious signal relay. Following injury and inflammation, reduced inhibition contributes to the phenomena of hyperalgesia (an increased sensitivity to input from nociceptive fibers) and allodynia, which describes a painful sensation elicited by input from non-nociceptive fibers. While both phenomena may help protect injured tissue from further damage and may foster its healing, under unfortunate conditions they outlast the healing process and may then severely compromise the quality of life of affected patients. The spinal dorsal horn is a key site for endogenous pain control and maladaptive plasticity which underlies many chronic pain conditions. Various processes triggered by peripheral inflammation or nerve damage compromise synaptic inhibition at this site and in corresponding brainstem areas. Among these processes are alterations in the excitatory drive to inhibitory dorsal horn neurons [1], a compromised electrochemical gradient of chloride ions [2,3], and altered responsiveness of inhibitory neurotransmitter receptors [4]. While the first two mechanisms are triggered by peripheral nerve damage and affect both GABAergic and glycinergic inhibition, peripheral inflammation has a specific impact on the function of dorsal horn GlyRs. General aspects of glycinergic neurotransmission are illustrated in Figure 1.

2. GlyRs in Inflammatory Hyperalgesia and Allodynia

Cyclooxygenase-2 (COX-2) derived prostaglandin E2 (PGE₂) is a pivotal mediator of inflammation and inflammatory hyperalgesia. In response to peripheral inflammatory insults, PGE₂ is not only produced in the periphery at the site of inflammation, but also in the CNS, especially in the spinal dorsal horn, where it contributes to the phenomenon of central sensitization. A series of reports published between 2002 and 2005 established the critical role of α3 GlyRs in inflammation and PGE₂-mediated central sensitization. PGE₂, but not other prostaglandins including PGD₂, PGI₂, or PGF_2α, reduces glycinergic synaptic transmission in superficial dorsal horn neurons through a postsynaptic mechanism involving the activation of EP2 receptors, production of cAMP, and subsequent activation of protein kinase A (PKA) [5]. This inhibitory effect was lost in mice that lack a specific GlyR subtype defined by the inclusion of the α3 subunit in the pentameric receptor complex (GlyR α3) [6]. The expression of this subunit in the spinal cord is largely confined to layer II of the dorsal horn, which is also known as the substantia gelatinosa. This site constitutes the termination area of most nociceptive fibers arriving in the CNS from the periphery. A systematic in silico screen of the GlyR α3 protein sequence revealed a strong consensus site for PKA-dependent phosphorylation in the large intracellular loop between transmembrane segments 3 and 4. This consensus site comprises the amino acid sequence arginine-glutamate-serine-arginine (RESR), in which the serine at position 346 constitutes the actual site of phosphorylation. The critical role of this site for PGE₂ triggered GlyR α3 phosphorylation was supported by results obtained in HEK293 cells. Heterologous expression of EP2 and α3 GlyRs reconstituted the inhibition of glycinergic membrane currents by PGE₂. Replacing the serine residue at position 346 by an alanine, an amino acid of similar size but lacking the OH group required for phosphorylation, prevented inhibition by PGE₂ [6].

The availability of mice lacking α3 GlyRs allowed assessing the contribution of α3 GlyRs and their regulation by PKA to baseline nociception and different forms of hyperalgesia. GlyRα3 deficient mice behaved normally in tests of baseline nociception (noxious heat or punctate mechanical stimulation with von Frey filaments). This lack of a pronociceptive phenotype may be linked to unaltered baseline glycinergic neurotransmission in GlyR α3 deficient mice and may hint at a compensatory up-regulation of other GlyR subunits. Pronounced alterations in the development of hyperalgesia became apparent when the mice were challenged with peripheral inflammation. In wild-type mice, subcutaneous injection of complete Freund’s adjuvant or with the yeast extract zymosan A induces thermal and mechanical hyperalgesia, which lasted for several days to weeks depending on the amount injected. GlyR α3 deficient mice showed strongly reduced thermal and mechanical hyperalgesia especially during the later phases of inflammation. A virtually identical phenotype was observed in a side-by-side comparison of GlyR α3 deficient mice and mice lacking prostaglandin EP2 receptors [7].

Although the reconstitution experiments in HEK293 cells supported the critical role of a direct PKA-dependent phosphorylation of α3 GlyRs, unequivocal proof for its relevance to in vivo hyperalgesia was still lacking. The recent generation of a genetically engineered S346A point mutated mouse line allowed filling this gap (Figure 2). This mouse carries a serine to alanine amino acid exchange at position 346 (S346A mutation) in the PKA consensus sequence of the GlyR α3 subunit [8]. Electrophysiological recordings from substantia gelatinosa neurons in spinal cord slices of these mice demonstrated not only that the point-mutated α3GlyRs were resistant to inhibition by PGE₂ but also confirmed the critical contribution of phosphorylation at this site to inflammatory hyperalgesia. Mice carrying the S346A point mutation were resistant to the hyperalgesic effects of intrathecally injected PGE₂ and developed much less hyperalgesia after injection of zymosan A into one hindpaw.

Besides inflammation, neuropathy is another major source of chronic pain and hyperalgesia. A contribution of COX-2 or PGE₂ to neuropathic pain has been proposed [9] but is still rather controversial. GlyR α3 deficient mice have previously been tested in the constriction injury model of neuropathic pain [10]. Both wild-type mice and GlyR α3 deficient mice developed prolonged thermal and mechanical hyperalgesia following the constriction injury of the sciatic nerve, suggesting that the phosphorylation of GlyR α3 subunits is dispensable for the development of hyperalgesia following peripheral nerve injury. This finding was recently confirmed in experiments with S346A point mutated mice, which also developed unaltered hyperalgesia following peripheral nerve injury [8]. It should be stressed that the normal development of neuropathic hyperalgesia in the GlyR α3 deficient or S346A point mutated mice does not exclude that α3 GlyRs still control neuropathic hyperalgesia; it only means that these receptors are not phosphorylated or inhibited in the course of neuropathy. In fact, two relatively recently developed GlyR modulators, AM-1488 [11] and 5-desoxy-THC/DH-CBD [12], reduce neuropathic pain in rodents (for a more detailed discussion of the action of these molecules see Section 5).

Additional insights have also been gained into the molecular mechanisms that link phosphorylation at S346 to decreased glycinergic currents [13]. In HEK293 cells cotransfected with cDNAs encoding for GlyRα3 and EP2 receptors, application PGE₂ led to a decrease in glycinergic membrane currents with no change in their inactivation kinetics or plasma membrane expression. In single channel recording experiments, activation of PKA progressively reduced single current amplitudes to about 66%, which is close to the inhibition of glycinergic synaptic currents observed in mouse spinal cord slices [5,8,14]. Introducing the phospho-mimicking serine to glutamate (S→E) mutation reduced single channel conductance to a similar degree with no effect on single channel open probability. Although previous work had suggested that S346 phosphorylation elicits structural changes in the α3 glycine-binding site [15], the introduction of the S346E mutation had no effect on the potency (EC₅₀) of GlyR currents. Substituting S346 with phospho-deficient alanine left the single channel amplitude and open probability almost unaffected.

3. Circuit Aspects of Glycinergic Control of Spinal Nociception

To fully understand of the role of α3 GlyRs in spinal control of nociception, precise knowledge of the neural circuits controlled by these receptors is essential. On a very gross scale, spinal hyperalgesia may be viewed as an imbalance of excitatory nociceptive input and local inhibitory control. In reality, the situation is likely much more complex. Figure 3 depicts some of the polysynaptic pathways of the spinal dorsal horn which become functional in different pathological pain states.

The simplest mechanism by which GlyRs may contribute to spinal nociceptive control might be the direct inhibition of nociceptive output from the spinal cord. This output occurs mainly via projection neurons located in lamina I of the dorsal horn, which relay nociceptive signals to the brainstem. These neurons receive direct excitatory drive from nociceptors and are controlled by GABAergic and glycinergic input [18]. The glycinergic input to lamina I neurons is reduced in rats with inflamed paws [19]. Altered heat hyperalgesia in the α3 GlyR-deficient mice [6] may reflect changes in the glycinergic control of superficial dorsal horn neurons, as heat stimuli are primarily processed in this area.

However, compared to neurons of the deeper dorsal horn, projection neurons of lamina I receive relatively little spontaneous inhibitory input [18], potentially suggesting that GlyRs might primarily be relevant to sensory processing in the more complex circuits of the deeper dorsal horn. Glycinergic neuron somata are in fact more prevalent in the deeper dorsal horn layers (laminae III and deeper) than in lamina I and II [20]. Accordingly, postsynaptic glycine responses are also larger and more prevalent in the deep than in the superficial dorsal horn [21,22]. It should however be noted that even in lamina II, the glycinergic component of inhibitory postsynaptic currents still outweighs the GABAergic component [23]. This anatomical gradient likely bears functional implications, as the deep dorsal horn receives mainly tactile (Aβ fiber) input while the nociceptive (C and Aδ fiber) input dominates in the superficial dorsal horn. Impaired segregation of signal relay in the superficial versus deep dorsal horn is thought to underlie allodynia, the painful sensation evoked by tactile stimuli. It is believed to result from the abnormal activation of lamina I neurons by Aβ fiber input. Lamina projection neurons normally become activated only in response to noxious input. After blockade of GABA_A receptors and GlyRs, these neurons become excitable also by input from Aβ fibers through a polysynaptic pathway [24].

The first neuron type identified in this pathway were protein kinase Cγ (PKCγ) expressing excitatory interneurons [25], which are located at the border between lamina II and lamina III and hence at the interface of innocuous tactile and noxious input [26]. Subsequent work has identified several additional elements of this circuit (for a recent review on allodynia circuits of the dorsal horn, see also [16]). Very recent work suggests that several different pathways exist, which are differentially recruited in inflammatory of neuropathic pain states [17]. For several of these neuron types, the presence of GlyRs (as well as GABA_A receptors) on their surface has been directly demonstrated [25,27,28]. These GlyRs become activated in a feed-forward mechanism initiated by input from non-nociceptive tactile fibers [29]. It is believed that excitatory interneurons of these allodynia circuits are normally under strong inhibitory control, leading to the gating (closure) of the polysynaptic connection under normal conditions [30]. Consistent with this model, ablation of dorsal horn glycinergic neurons induces behavioral signs of allodynia and spontaneous discomfort in mice [23]. Similar phenotypes have also been observed after ablation or silencing of genetically defined subsets of glycinergic neurons, such as inhibitory parvalbumin and dynorphin neurons [31,32]. Despite these new insights, the specific location of α3 GlyRs on particular types of dorsal horn neurons and the subtypes of inhibitory interneurons that target α3 GlyRs are still unknown.

Inhibitory dorsal horn neurons do not only provide classical postsynaptic inhibition to intrinsic dorsal horn neurons, but also target axon terminals of primary sensory afferent nerve fibers. These terminals express GABA_A receptors but no GlyRs. Accordingly, glycine does not contribute to so-called primary afferent depolarization or presynaptic inhibition of primary afferent input. However, GlyRs reside on presynaptic terminals of central neurons, where they increase transmitter release. Such presynaptic GlyRs have first been found in neurons of the auditory brainstem, where their activation increases glycine release [33]. A similar action has later been reported for glycinergic input onto commissural neurons of the spinal dorsal horn [34]. The subunit composition of these presynaptic GlyRs is unknown. However, unlike postsynaptic GlyRs, presynaptic GlyRs may be homomeric receptors (i.e., lack β subunits) as presynaptic GlyRs are not clustered by gephyrin (see also Figure 1). Whether α3 subunits contribute to presynaptic GlyRs in the dorsal horn is unknown but experiments in hypoglossal motoneurons have shown that forskolin (cAMP)-induced facilitation of transmitter release was reduced in mice lacking α3 GlyRs [35].

4. Genetic Evidence of Glycinergic Pain Control in Humans

Direct evidence to support the presence of glycinergic pain control in humans is difficult to obtain given the present lack of compounds suitable for clinical testing in humans. A recent study in human patients suffering from inherited hyperekplexia provides however supporting evidence [36]. Most hyperekplexia patients carry homozygous (or compound heterozygous) loss-of-function mutations in GlyR genes (mainly α1 and β) or in the glycine transporter GlyT2, whose dysfunction leads to impaired loading of glycinergic terminals with glycine. The main symptom in these patients is an exaggerated startle response upon exposure to sudden sensory stimuli such as loud and abrupt noise or unexpected touch. Using a quantitative sensory testing battery, the recent study revealed in addition decreased pain thresholds in hyperekplexia patients.

While mutations in GLRA1 and GLRB as well as in SLC6A5 (encoding for GlyT2) are well-established causes of hyperekplexia, GLRA3 has not been linked to any human disease yet. This may suggest that pharmacological targeting of α3 GlyRs may be relatively safe, but it also means that human genetic evidence supporting a specific role of α3 GlyRs is lacking. On a positive note, the PKA consensus site in the large intracellular loop of α3 GlyRs is conserved in humans. It is also noteworthy that the phenotype described in α3 GlyR deficient mice (reduced inflammatory pain) might not be easily detectable in humans. Furthermore, a screening of the human genetic variation database (www.ncbi.nlm.nih.gov/variation/) for loss of function (non-sense) variants revealed few hits. All of which occurred with frequencies too low to allow systematic clinical trials in such persons (for a discussion on this topic, see [8]).

5. Synthetic Glycine Receptor Modulators with Potential Analgesic Effects

The rather selective expression of α3 GlyRs at a site critical for nociceptive processing and the phenotype of GlyR α3 deficient mice have sparked considerable interest in these receptors as targets for novel analgesics. Unlike the closely related GABA_A receptors, GlyRs have remained therapeutic orphans. Nevertheless, several drug companies (e.g., AMGEN [11,37,38] and Neusentis [39]) and academic groups [40,41,42] have recently reported the synthesis or identification of small molecule potentiators of GlyRs. In addition, AMGEN reported a series of pan (α1 and α3) and selective (α1 or α3) GlyR antibodies with functional (agonistic or antagonistic) activity profiles suggesting that selectivity within this receptor family can be achieved [43]. Figure 4 provides an overview over different small molecule GlyR activators and potentiators and their binding sites in the channel complex.

Recent drug screening efforts at AMGEN have led to the discovery of novel glycine receptor potentiators [11,37,38]. Of particular interest, AM-1488, a highly selective positive allosteric modulator of α1 and α3 GlyRs acts through a high affinity binding (in the sub-micromolar range) to a newly discovered site in the large extracellular domain [11]. AM-1488 (20 mg/kg; p.o.) has been tested in a mouse spared nerve injury (SNI) model of neuropathic pain, where it was as effective as the reference compound pregabalin (30 mg/kg; p.o.), while its enantiomer was inactive at the same dose [11,37].

Other evidence supporting potential therapeutic benefit of GlyR potentiators comes from cannabinoid derivatives. Δ⁹-tetrahydrocannabinol (THC) modulates GlyR function, in addition to activating G protein coupled CB₁ and CB₂ receptors. Chemical modification of THC has led to the derivatives 5-desoxy-THC (identical to DH-CBD), 1-desoxy-THC, and di-desoxy-THC. Further, 5-desoxy-THC/DH-CBD and di-desoxy-THC are devoid of activity at CB₁ and CB₂ receptors but still bind to GlyRs. In addition, 5-desoxy-THC/DH-CBD acts as a positive allosteric modulator and di-desoxy-THC as its competitive antagonist of GlyRs [40]. In mouse inflammatory pain models, 5-desoxy-THC/DH-CBD exerted antihyperalgesic effects that were absent in α3 GlyR deficient mice but retained in CB₁ and CB₂ deficient mice [12].

Propofol (2,6-diisopropylphenol) is an intravenous anesthetic that targets primarily GABA_A receptors. At higher concentrations, it also acts as a GlyR modulator. The propofol derivative 2,6-di-tert-butylphenol (2,6-DTBP) is devoid of activity at the major GABA_A receptor subtypes [46] but retains activity at GlyRs [47] and possesses antihyperalgesic activity in neuropathic mice at high doses [14]. Critical for the interaction of propofol derivatives with GlyRs is a phenylalanine residue (F388 in the α3 subunit) in the large intracellular loop and close to the S346 phosphorylation site [48]. Interestingly, potentiation of synaptic GlyRs by 2,6-DTBP depends on the phosphorylation status of the GlyR α3 subunit. In mouse spinal cord slices, 2,6-DTBP prolonged the kinetics of glycinergic IPSCs only when α3 GlyRs were phosphorylated (“primed”) with PGE₂ or when it was tested in spinal cord slices prepared from animals with inflamed paws [14].

Molecules with positive allosteric activity at GlyRs have also been isolated from a natural compound library generated from Australian and Antarctic marine invertebrates and algae [42,49,50]. These compounds have highly complex chemical structures hampering their artificial synthesis. Probably for this reason, in vivo activities have not been tested yet. Interestingly, one of the compounds specifically modulates α3 GlyRs with no activity at α1 GlyRs [49], suggesting that the development of α3 selective compounds is in principle feasible.

6. Other Compounds with Modulatory Actions at Glycine Receptors

GlyRs are in addition modulated by a number of endogenous molecules or synthetic compounds that either lack drug-like properties or activate primarily targets different GlyRs. Although such compounds are unlikely to be used therapeutically, they have led to the discovery of sites for allosteric modulation of different GlyRs. As such, they may provide starting points for drug discovery programs. Below, we provide a relatively short summary of this work. A more comprehensive coverage of this topic can be found elsewhere [51,52,53,54].

Endocannabinoids are lipid signaling molecules that primarily activate the G protein coupled cannabinoid CB₁ and CB₂ receptors. Arachidonoyl ethanolamide (AEA) and 2-arachidonoyl glycerol (2-AG) are pivotal endogenous activators of these receptors. Related arachidonoyl conjugates potentiate GlyR function with partially different effects on the different GlyR subtypes [55,56]. While neutral compounds such as AEA, N-arachidonoyl-serotonin, and N-arachidonoyl-dopamine potentiate α1, α2, and α3 GlyRs, the acidic compounds N-arachidonoyl-glycine, N-arachidonoyl-serine, N-arachidonoyl-L-alanine, arachidonic acid, and N-arachidonoyl-GABA potentiate α1 but inhibit α2 and α3 GlyRs [55]. Building on these differential effects has allowed the identification of relevant sites in the GlyR protein through the generation and analysis of chimeric GlyR constructs. These experiments revealed a relatively complex scenario with one relevant amino acid located in the extracellular domain (alanine at position 52 in α1), two amino acids in transmembrane segment 2 (glycine 254 in α1 and alanine 265 in α3), and in the large intracellular loop (lysine 385 in α1). Other work provides a detailed analysis of effects of the acyl chain length and numbers and sites of double bonds within the acyl chain on the modulation of GlyRs with different subunit combinations [57].

Tropeines are antiemetic drugs that act as antagonists at ionotropic serotonin (5-HT3) receptors. Two tropeines, MDL-72222 and tropisetron, also potentiate GlyR activity at nanomolar concentrations [58,59,60]. They bind to an interface between two α or one α and a β subunit in the extracellular domain. The effects of tropeines have been mainly explored on α1 GlyRs, and less in α3 GlyRs. A recent report explored the modulation of homopentameric α3 GlyRs by tropeines [44]. Tropisetron did not potentiate α3 GlyRs, but rather caused concentration-dependent inhibition in the low micromolar range. In silico docking confirmed that tropeines may bind to the extracellular domain of α3 GlyRs. Since tropisetron also displayed excusive inhibitory effects on α2 GlyRs [59], these data suggest a subunit-specific effect on α1 GlyRs.

Zonisamide is an antiepileptic drug, which has recently been reported to facilitate activation of recombinant and native α1, α2, and α3 GlyRs at therapeutic concentrations [61]. Zonisamide has many potential therapeutic targets including voltage gated Na⁺ channels and T-type Ca²⁺ channels and it is unclear which target is responsible for its antiepileptic effect. Some evidence suggests that zonisamide possesses antihyperalgesic activity in different preclinical models [62]. This is, however, a common action also of anticonvulsant drugs including the gabapentinoids, carbamazepine, and phenytoin and it is hence unclear whether potential analgesia might originate from its modulatory activity at GlyRs. More remarkable is the absence of the potential side effects of GlyR modulation discussed above (muscle relaxation, respiratory dysfunction, addiction) [63].

Lastly, glycine receptor activity can also be modulated in a bidirectional fashion by zinc, where sub-micromolar concentrations (20 nM–1 µM) lead to potentiation and micromolar concentration (20–50 µM) to inhibition [64], reviewed in [65]). The biological presence of zinc documented in synaptic vesicles in the brain, suggest indeed that it bears unique neurotransmission modulation roles within the central nervous system [66,67,68,69]; reviewed in [65]). Thus, adding zinc chelators (such as compatible with the screening platforms selected) is highly recommended in future HTS-type screens to reduce the rate of false-positive hits moving-forward (see [70] for a comprehensive list of zinc chelators readily available).

7. Is Specificity for the α3 Subtype Preferred or Required?

This question cannot be answered yet [52]. GlyRs do not only control nociception but several other physiological functions. Best known among them is the control of motoneuron activity and hence muscle tone. In addition, GlyRs are abundant in the pre-Bötzinger complex, which controls respiration, in the ventral tegmental area (VTA) and the nucleus accumbens (NAc), which form the brain’s reward system, the retina and the auditory system. While pharmacological modulation of inhibitory transmission appears to have little impact on visual or auditory perception, effects on motoneuron activity, respiratory control, and addiction are areas of potential concern. Motoneurons virtually lack α3 GlyRs, suggesting that α3 selective modulators should be devoid of undesired muscle relaxation. In contrast, the pre-Bötzinger complex, the VTA, and the NAc contain α1 and α3 GlyRs [71,72,73,74,75,76]. Selectivity for the α3 subtype may protect from some of these potential undesired drug effects but it may also reduce efficacy as most GlyRs even in the substantia gelatinosa contain both α1 and α3 subunits [77]. While several lines of evidence strongly support that GlyR potentiators will reduce pathological hyperalgesia, it is currently uncertain whether subtype specific modulators should be preferred in drug discovery programs. With the recent progress in identifying selective GlyRs small and large molecules, it is only a question of time before highly selective and potent drug candidates can allow the field to test whether selectivity is preferred or required.