The role of spinal cord extrasynaptic α5GABAA receptors in chronic pain

Abstract Chronic pain is an incapacitating condition that affects a large population worldwide. Until now, there is no drug treatment to relieve it. The impairment of GABAergic inhibition mediated by GABAA receptors (GABAAR) is considered a relevant factor in mediating chronic pain. Even though both synaptic and extrasynaptic GABAA inhibition are present in neurons that process nociceptive information, the latter is not considered relevant as a target for the development of pain treatments. In particular, the extrasynaptic α5GABAARs are expressed in laminae I‐II of the spinal cord neurons, sensory neurons, and motoneurons. In this review, we discuss evidence showing that blockade of the extrasynaptic α5GABAARs reduces mechanical allodynia in various models of chronic pain and restores the associated loss of rate‐dependent depression of the Hoffmann reflex. Furthermore, in healthy animals, extrasynaptic α5GABAAR blockade induces both allodynia and hyperalgesia. These results indicate that this receptor may have an antinociceptive and pronociceptive role in healthy and chronic pain‐affected animals, respectively. We propose a hypothesis to explain the relevant role of the extrasynaptic α5GABAARs in the processing of nociceptive information. The data discussed here strongly suggest that this receptor could be a valid pharmacological target to treat chronic pain states.

population of GABA A Rs is present in somatic, dendritic, and axonal membranes, they are not in opposition to presynaptic terminals. Extrasynaptic receptors mediate an alternative form of inhibition that modulates the neurons' excitability by a persistent increase in conductance and tonic shunt (Brickley et al., 1996;Farrant & Nusser, 2005;Kullmann et al., 2005). The transmitter release from nociceptive and non-nociceptive primary afferents is under presynaptic GABAergic control through the axo-axonic synapses (Rudomin & Schmidt, 1999). According to the gate theory, this presynaptic inhibition regulates the excitatory input onto projection neurons preventing its activation (Melzack & Wall, 1965).
Based on its pharmacological properties, GABA A Rs can be divided into sensitive and nonsensitive to benzodiazepines. This feature is related to the subunit composition of the receptor because the action of these drugs is mediated by a specific binding site located in the α 1,2,3,5 and γ subunits. GABA A Rs containing α 4 or α 6 subunits do not bind benzodiazepines and are associated with the δ subunit (Rudolph & Knoflach, 2011;Zeilhofer et al., 2012). The side effects of benzodiazepines, such as sedation and addiction after long-term use, are considered important issues in developing new GABA A Rs subtypeselective compounds to overcome the limitations of classical benzodiazepines .

CHRONIC PAIN
The GABAergic inhibition, mediated by GABA A Rs, in the spinal cord is so relevant that its blockade with bicuculline, an antagonist of GABA A Rs, produces allodynia and hyperalgesia in healthy rodents (Roberts et al., 1986). Consequently, based on this fact, the development of chronic and neuropathic pain is considered a result of a loss of GABA A inhibition in the spinal cord (Bonin & De Konick, 2013;Munro et al., 2011;Zeilhofer et al., 2012). Many mechanisms have been described to explain the sensitization at the spinal dorsal horn produced by the loss of synaptic GABA A inhibition. One of these studies shows a substantial increase in polysynaptic input onto lamina II neurons in the presence of bicuculline (Baba et al., 2003). Another interesting study was performed by recording second-order lamina I neurons, which express neurokinin 1 receptors and receive sensory excitatory synaptic inputs exclusively from C and Aδ nociceptors. In the presence of GABA A R and GlyR antagonists, polysynaptic inputs appeared in response to Aβ fiber activation (Torsney & MacDermott, 2006). Similar anomalous synaptic inputs were recorded from neurons of the substantia gelatinosa in transverse spinal cord slices from animals with chronic pain (Baba et al., 1999). The presence of this polysynaptic excitatory input has been proposed to underlie allodynia activated in vivo after the intrathecal application of bicuculline or | 3 of 10 DELGADO-LEZAMA Et AL.
strychnine . These studies indicate the relevance of GABA A synaptic inhibition as a pharmacological target for reversing pathological pain states. Taking into consideration, all these evidence by transgenic mice with mutated α subunits, a positive allosteric modulators (PAM) battery of α 2,3,5 GABA A R with less sedating benzodiazepines action has been developed to restore the loss of synaptic GABA A inhibition to relieve pain (Bonin & De Koninck, 2013;Knabl et al., 2008;Munro et al., 2011;Zeilhofer et al., 2012). In parallel, the use of point-mutated mice has shown that the ablation of the α 2 GABA A R has a strong antihyperalgesic effect with reduced side effects in different pain models . In addition, the pharmacological targeting of this receptor could prevent the development of tolerance.
All these studies indicate that the different subtypes of benzodiazepine-sensitive GABA A Rs contribute to spinal antihyperalgesia with the rank order α 2 >α 3 >α 5 >α 1 . The antihyperalgesic action of benzodiazepines has been tested in three pain models, that is, zymosan A, chronic constriction injury of sciatic nerve and formalin test . In addition, it has been reported that α 5 GABA A Rs activation, via the administration of PAMs, does not induce analgesia in inflammatory and neuropathic pain models (Munro et al., 2011;Zeilhofer et al., 2012).

OF α 5 GABA A R IN DIABETIC NEUROPATHY
Investigating the mechanisms underlying hyperalgesia developed by the formalin test in diabetic rats, it was found a paradoxical reduction of glutamate, the increase of GABA, and downregulation in the expression of the K + -Cl − co-transporter 2 (KCC2) in the laminae I-II of the dorsal horn spinal cord (Jolivalt et al., 2008;Morgado et al., 2008). Unexpectedly bicuculline reduced formalin-evoked flinching and alleviated tactile allodynia (Jolivalt et al., 2008), suggesting that reduced KCC2 expression in parallel with the increased GABA release contribute to the allodynia and hyperalgesia in diabetes (Jolivalt et al., 2008). The relevant role of GABA A Rs in diabetic neuropathy is highlighted by the evidence that in dorsal horn neurons, even when subjected to intense glycinergic inhibition (Yoshimura & Nishi, 1995), the blockade of GABA A Rs is enough to reduce allodynia and produce it in healthy animals, respectively. Interestingly, some of the molecular and cellular alterations produced in the spinal cord in diabetic neuropathy are quite similar to that reported in nerve injury models (Coull et al., 2003). In this case, KCC2 is also downregulated by the binding of BDNF to TrK1 receptors producing the depolarization of the E Cl− to a level high enough to generate action potentials when GABA A Rs are activated (Coull et al., 2003(Coull et al., , 2005. Therefore, KCC2 downregulation in diabetic neuropathy might also produce a switch from GABA A inhibition to the excitation of laminae I-II neurons involved in the nociceptive processing information as in peripheral nerve injury. Unexpectedly, but in accordance with Jolivalt et al. (2008), our results have shown that the blockade of the α 5 GABA A Rs is enough to reverse mechanical allodynia in streptozotocin (STZ)-induced diabetic rats (Hernández-Reyes et al., 2019), suggesting that this receptor contribute significantly to pain in diabetic neuropathy.

AND PRONOCICEPTIVE ROLE OF α 5 GABA A R IN CHRONIC INFLAMMATION AND NEUROPATHIC PAIN
It is well known that the tonic current mediated by extrasynaptic GABA A Rs has great relevance in physiological and pathological events both at the level of individual neurons and in neural networks (Farrant & Nusser, 2005). Even though the α 5 GABA A Rs are expressed in laminae I-II neurons of the network that processes nociception, their role in chronic and neuropathic pain is unknown. Only one study carried out in Gabra 5 −/− mice shows that these receptors mediate a tonic current in those neurons. This work also concludes that α 5 GABA A Rs do not modulate acute nociception but play a pronociceptive role at the late inflammation stages ( (Castro et al., 2011), and motoneurons (Canto-Bustos et al., 2017) where they are activated by the endogenous GABA reducing its excitability tonically. It is well known that the presynaptic inhibition of low threshold primary afferents, associated with its depolarization (PAD), is mediated by the activation of synaptic α 2 GABA A Rs (Witschi et al., 2011) and depresses the monosynaptic reflex (MSR) (Rudomin & Schmidt, 1999). Interestingly, L-655,708 facilitated the MSR without affecting the PAD, suggesting that besides synaptic α 2 GABA A Rs also extrasynaptic α 5 GABA A Rs are involved in motor control (Canto-Bustos et al., 2017;Loeza-Alcocer et al., 2013). The role of α 5 GABA A Rs in motor control has been evidenced by the Hoffmann reflex (HR). The property of HR, known as rate-dependent depression (RDD), has been proposed as a biomarker to detect the presence of neuropathic pain (Marshall et al., 2017). As shown in rats in humans, the RDD is also impaired in diabetic neuropathy (Hernández-Reyes et al., 2019;Lee-Kubli & Calcutt, 2014;Marshall et al., 2017). Interestingly, bicuculline restores the loss of RDD, suggesting that the disruption of GABAergic inhibition mediated by GABA A Rs is involved in the impairment of this property (Jolivalt et al., 2008;Lee-Kubli & Calcutt, 2014). We have shown that the intrathecal application of L-655,708 abolishes RDD in healthy animals and reestablished it in neuropathic diabetic rats in parallel with the reduction of tactile allodynia (Figure 1, Hernández-Reyes et al., 2019). This result demonstrates that the impairment of RDD is mediated by extrasynaptic α 5 GABA A Rs. There is evidence that formalin-or capsaicin-induced secondary hyperalgesia is associated with the activation of antidromic action potentials in sensory neurons produced by the facilitated PAD (inflammatory conditions). This phenomenon, known as dorsal root reflex (DRR), propagates retrogradely to the peripheral terminal, where it evokes the release of inflammatory neuropeptides (substance P and CGRP) (Willis, 1999), contributing to hyperalgesia (Cervero et al., 2003). In this context, instead of activating GABA A Rs, it was proposed to induce analgesia by blocking them. In line with this, the α 5 -selective negative allosteric modulator (NAM) α5IA-II was evaluated in the formalin and carrageenan tests. In both cases, it reverses mechanical hypersensitivity and weight-bearing deficits. In contrast, it did not affect nociception in a neuropathic pain model (Munro et al., 2011). Unexpectedly, in the turtle, we showed that L-655,708 depresses the DRRs without affecting PAD, suggesting that extrasynaptic α 5 GABA A Rs might be tonically depolarizing the primary afferents to reach the threshold to activate the DRRs (Loeza-Alcocer et al., 2013). In agreement with this result, the peripheral and intrathecal pretreatment or post-treatment with L-655,708 prevents and reverses the long-lasting allodynia and hyperalgesia in the formalin-induced test and restores the loss RDD of the HR (Bravo-Hernández et al., 2016).

| HOW SELECTIVE IS L -655,708?
The pharmacological approach to investigate the function of α 5 GABA A Rs has the disadvantage of the differential selectivity of L-655,708 for the GABA A Rs. It displays more selectivity for the α 5 subunit-containing receptors than for those containing α 1 , α 2 , α 3 , and α 6 subunits (Quirk et al., 1996). To circumvent this issue, we have used the excitability Walltest in healthy rats. For this, the L1 vertebrae segment was removed to expose the lumbar enlargement, and the antidromic compound action potential (test cAP) recorded in the tibial nerve was evoked by the spinal cord electrical stimulation and was conditioned by the electrical stimulation of the sural and peroneal nerves. The test and the conditioned responses indicate the phasic and the tonic excitability of primary afferent fibers, respectively. Interestingly, the phasic excitability of primary afferent fibers mediated by the activation of synaptic α 2/3 GABA A Rs was not affected by L-655,708, while the tonic excitability mediated by extrasynaptic α 5 GABA A Rs was increased (Figure 2; Hernández-Reyes et al., 2019). Therefore, this result, together with the action of the α 5 GABA A R siRNA in healthy and reserpine-induced fibromyalgia (De la Luz-Cuellar et al., 2019), confirms the antinociceptive and pronociceptive role of α 5 GABA A Rs in healthy and chronic pain conditions, respectively.

RELEVANT IN HEALTH AND PAIN?
α 5 GABA A Rs are expressed in laminae I and II neurons, where they mediate a tonic inhibitory current in healthy animals (Perez-Sanchez et al., 2017). KCC2 is downregulated in the dorsal horn of rodents with nerve injury neuropathy (Coull et al., 2003) and diabetic neuropathy (Jolivalt et al., 2008). Therefore, the action of GABA A R might be switching from inhibition into excitation, which has been shown in nerve injury (Coull et al., 2003). Given that the action of extrasynaptic GABA A Rs is about six times more intense than the synaptic receptors (Ataka & Gu, 2006), we have hypothesized that the extrasynaptic spinal α 5 GABA A Rs might be tonically hyperpolarizing the projection neurons in the health condition, while in chronic pain they might be causing a tonic depolarization of the cell membrane. In the first case, they prevent the excitatory synaptic inputs of the low threshold afferent fibers from activating the projection neurons of laminae I-II (Baba et al., 2003;Torsney & MacDermott, 2006), and in the second condition, the gate opens, allowing the fibers mentioned above to activate the neurons (Baba et al., 1999; Figure 3).
Although we have shown that α 5 GABA A Rs have a pronociceptive role in several models of chronic pain, there remains an unanswered question: after blocking these receptors, what | 5 of 10 DELGADO-LEZAMA Et AL.
inhibits the projection neurons, preventing them from being stimulated by the activation of the low-threshold afferent fibers? One possibility is the participation of the GABA B receptors (GABA B Rs). The expression of GABA B Rs in dorsal horn neurons and primary afferent fibers in the spinal circuitry processing nociceptive information is well documented (Malcangio 2018;Schuler et al., 2001;Towers et al., 2000). Indeed, GABA B Rs are tonically active inhibiting transmitter release (Peshori et al., 1988). At postsynaptic neurons, they control neuronal excitability. Moreover, activation of these receptors with baclofen inhibits the plateau properties, wind-up, and post-discharge induced by dorsal root stimulation in dorsal horn neurons even in the presence of tetrodotoxin (Russo et al., 1998) or by noxious mechanical stimulation in vivo (Reali et al., 2011). Lee-Kubli et al. (2021) confirmed that the GABA B Rs in conjunction with the GABA A Rs contribute to the inhibitory circuitry involved in the modulation of RDD of the HR and tactile allodynia in diabetic rats (Lee-Kubli et al., 2021).
Interestingly, phaclofen, an antagonist of the GABA B Rs, but not bicuculline, impaired the RDD in 4-week diabetic rats, suggesting that in these animals the normal RDD is mediated by functional GABA B Rs. It is well known that in 8-week diabetic rats, bicuculline and L-655708 restored the impaired RDD and reversed the mechanical allodynia (Hernández-Reyes et al., 2019;Jolivalt et al., 2008). However, unexpectedly, the administration of phaclofen in these rats reverted the restoration of RDD by bicuculline. In addition, the intrathecal application of baclofen alleviated mechanical allodynia in diabetic rats. These results suggested to the authors that GABA B Rs, together with GABA A Rs, are mediating the GABAergic inhibition in the dorsal horn of the spinal cord F I G U R E 1 Rate-dependent depression of the Hoffmann reflex (HR). (a) First and second HR evoked at 0.1 (black trace) and 5 Hz in control (blue trace) condition and after the intrathecal administration of L-655,708 (red trace) in naïve rats. (b) AUC (area under the curve) of rectified HR2 normalized with respect to the mean area of the first HR at 0.1 Hz (black circles), 5 Hz (blue circles), and 5 Hz in the presence of L-655,708 (red circles). *p = 0.0001, significantly different with respect to the mean value of HR obtained at 0.1 and 5 Hz (red circle), by 2-sample Student´s t-test. (c) First and second HR evoked at 5 Hz before (blue trace) and after 2 h (red trace) of intrathecal administration of L-655,708 in STZ-diabetic rats. (d) AUC of rectified HR2 normalized with respect to the first HR's mean area at 5 Hz before (blue circles) and after the intrathecal administration of L-655,708 (red circles). *p = 0.0001, significantly different with respect to the mean values of HR obtained at 5 Hz (blue circle) and 0.1 Hz (black circle) in STZdiabetic rats, by two-sample Student t-test. Data are the mean ± SD from six animals where nociceptive information is processed. Consequently, if GABA B Rs are closing the gate to the excitatory input of low-threshold afferent fibers on projection neurons, it would be interesting to know whether intrathecal administration of a GABA B receptor antagonist restores mechanical allodynia in diabetic rats after having removed it by blocking α 5 GAB-A A Rs receptors.

RECEPTORS REGULATION
Previous pharmacological studies have reported sex differences in α 5 GABA A Rs function. PAMs of α 5 GABA A Rs reduce stress-induced behaviors in females but not in male rodents (Piantadosi et al., 2016). More recently, our group reported a sex-dependent effect of α 5 GABA A Rs activation in chronic pain (De la Luz-Cuellar et al., 2019;Franco-Enzástiga et al., 2021). We found that α 5 GABA A Rs mRNA and protein changes in DRG and spinal cord are modulated in a sex-dependent way. Nerve injury increases the expression of α 5 GABA A R in female DRG and spinal cord in female rodents, but not in males . These changes are activated by DNA methylation in the CpG island of Gabra5 in males but not females. In contrast, the expression and function of spinal α 5 GABA A Rs in females is associated with the presence of 17β-estradiol .

| CONCLUSIONS
Multiple and irreversible plastic changes involving nonneuronal and neuronal cells in the spinal cord circuits  processing nociceptive information convey by sensory neurons underlie chronic and neuropathic pain. Between the complexity of adaptations in the function of the nervous system, the loss of inhibition mediated by synaptic GABA A Rs has been identified as a target for developing drugs to relieve pain. Reestablishing GABAergic inhibition without knowing where and how the inhibition is lost represents a very complex issue. The reversion of allodynia by blocking the α 5 GABA A Rs in pain induced by diabetes, nerve injury, fibromyalgia, and chronic inflammation indicates that the spinal inhibitory dysfunction might be related to a downregulation of KCC2 transporter affecting the homeostasis of Cl − in projection neurons, where the tonic current through this receptor might be switched from inhibition to excitation. Therefore, blocking α 5 GABA A Rs may be a feasible strategy to treat chronic pain.