Development of a Novel Benzodiazepine to Delineate Peripheral GABA-A Signaling Mechanisms in Visceral Pain Syndromes

Abstract

Background & Aims

Visceral pain is a cardinal symptom of many disorders affecting the gut and other abdominal organs. Modulators of gamma-aminobutyric acid (GABA) such as benzodiazepines may attenuate such pain but the specific contribution of peripheral GABA-A receptors (GABRAs) remains unclear as current agonists have prominent central effects.

Methods

Using medicinal chemistry optimization of the benzodiazepine scaffold, we developed a novel and potent positive allosteric modulator (PAM), LI-633, with no significant central nervous system (CNS) penetration.

Results

The locomotor activity of rats placed in an open field was unchanged with LI-633 at doses up to 30 mg/kg, confirming its lack of a CNS effect. LI-633 produced robust potentiation of GABA-induced inward current, with EC50 values ranging from 8 nM (α5β2γ2) to 128 nM (α3β2γ2). In vitro electrophysiological studies confirmed its ability to reduce excitability of human dorsal root ganglion (DRG) neurons by GABA. LI-633 potentiated muscimol-induced GABAergic currents in rat DRG neurons in a dose-dependent manner, with an EC₅₀ of 70.4 nM. In vivo, LI-633 significantly attenuated visceral hypersensitivity and pain behavior in a rat model of irritable bowel syndrome (IBS) and functional dyspepsia (FD), indicating the presence of physiologically relevant concentrations of GABA in the colon and stomach. In the IBS model, administration of the drug also resulted in decreased excitability of colon-specific DRG neurons and significantly reduced the colonic afferent response to balloon distention as measured by recordings of neural activity in dorsal ganglia rootlets.

Conclusions

These findings highlight the potential of targeting peripheral GABRAs for pain management in disorders associated with visceral hypersensitivity.

Summary.

LI-633, a novel peripherally restricted GABA-A positive allosteric modulator, reduced excitability of human dorsal root ganglion neurons and decreased pain behaviors in rodent models of post-IBD IBS and functional dyspepsia, supporting peripheral GABA-A receptors as therapeutic targets for these indications.

What You Need to Know.

Background

Chronic abdominal pain is a major unmet need in gastrointestinal disorders. Benzodiazepines can relieve pain but are limited by sedation and dependence due to central nervous system effects.

Impact

LI-633 is a novel peripherally restricted benzodiazepine that reduces visceral hypersensitivity in preclinical models of irritable bowel syndrome and functional dyspepsia, correlating with decreased excitability in human and rat nociceptive neurons.

Future Directions

These findings establish peripheral gamma-aminobutyric acid type A receptors as viable analgesic targets for visceral pain. Peripheral-acting gamma-aminobutyric acid type A-positive allosteric modulators like LI-633 could offer non-sedating, non-opioid relief for chronic gastrointestinal pain syndromes.

Chronic pain involves complex interactions between the peripheral and central nervous systems (CNS).¹ In the periphery, tissue damage or inflammation can lead to sensitization of nociceptors, making them more responsive to stimuli. This peripheral sensitization can trigger changes in the CNS, particularly in the spinal cord and brain, leading to central sensitization. It has been difficult to assess the relative contributions of these 2 nervous systems to chronic pain, as almost all analgesics cross the blood brain barrier and therefore can modulate the function of both. However, such insight has important therapeutic implications—such as avoiding off-target adverse effects (peripheral or central) or providing more effective relief by targeting both.²

This is particularly relevant in the development of drugs for disorders characterized by chronic visceral pain, such as irritable bowel syndrome (IBS) and functional dyspepsia (FD), which together affect more than 10% of the population in the United States and worldwide.³ Uniquely, gastrointestinal sensory afferents originate and/or terminate in 2 large, complex nervous systems (the enteric nervous system [ENS] and the CNS, respectively) with large overlaps in both neurotransmitters and their receptors, including, among others, serotonin, dopamine, and gamma-aminobutyric acid (GABA).⁴ GABA is the endogenous agonist for 2 distinct families of receptors, designated GABA-A and GABA-B. GABA-A receptors (GABRAs) comprise a large group of ligand-gated channels that allow the passage of chloride ions when open. These heteropentameric channels are generally composed of 2 alpha subunits, 2 beta subunits, and a single gamma subunit.⁵ Of the 6 alpha receptors, only α₁, α₂, α₃, and α₅ (GABRA1, 2, and 5) are benzodiazepine-sensitive. GABA is widely distributed in the CNS, where it exerts a hyperpolarizing, inhibitory effect contributing to the well-known sedating effects of drugs such as benzodiazepines and other positive allosteric modulators (PAMs) of the GABRAs. In the periphery, GABA-A signaling can be either hyperpolarizing or depolarizing, depending on the local chloride gradient across the membrane of the target cell.⁶ Although its function in the gastrointestinal tract has yet to be fully worked out, the ENS has the highest concentration of GABA among all peripheral organs, and there is intriguing evidence for a role for GABA signaling both as part of IBS pathophysiology and as a potential therapeutic target.^7,8

However, the site of action of both the clinical and preclinical drugs are difficult to assess because the modulators used (benzodiazepines and muscimol, respectively) cross the blood brain barrier. Although the potential utility of GABA-A PAMs for the treatment of chronic pain is supported by a large body of literature that shows that these compounds exhibit anti-allodynic and anti-hyperalgesic effects,^9,10 some of these effects are likely to also be mediated centrally by α₂, and to a lesser extent, α₃-containing GABRAs within the spinal cord.¹¹ Nevertheless, recent evidence supports a prominent role for peripheral GABRAs in mediating analgesia that appears to be distinct from their central effects. GABRAs and exogenous delivery of GABA, or GABA agonists to the dorsal root ganglion (DRG) has been shown to effectively attenuate nociceptive behavior in models of inflammatory and neuropathic pain.^12,13 With respect to visceral sensitization, muscimol has been shown to inhibit attenuate colonic afferent activity in isolated colon-pelvic nerve preparations and behavioral pain responses to noxious colonic distention in normal mice.¹⁴ Further, both diazepam and GABA attenuate visceral hypersensitivity in mice with acute colitis.¹⁵

We therefore hypothesized that enhancement of GABA-A signaling only within peripheral primary sensory neurons (ie, without engagement of central receptors) could be an effective strategy to treat visceral pain. Herein, we report the characterization of a promising lead GABA-A PAM, LI-633, that is pharmacologically similar to classical benzodiazepines, but with highly restricted access to the CNS. This agent reduces excitability in human sensory neurons in vitro and is effective in preclinical models of aspects of visceral pain. The development of this compound has the potential to identify more precisely the respective contributions of the gut and the brain to the pathogenesis of chronic visceral symptoms such as pain as well as spur the development of novel therapies for the same.

Results

Lead Discovery Characterization

The lead compound LI-633 (Figure 1A) emerged from medicinal chemistry optimization of the benzodiazepine scaffold for GABA-A-positive allosteric modulation potency, efficacy, pharmacokinetics, and minimal brain penetration. Examination of the cryo-EM structure of the GABA-A receptor with bound diazepam indicates that the lactam moiety of diazepam is solvent-exposed.¹⁶ In several other prototypical benzodiazepines such as midazolam and rilmazolam, this lactam is replaced by a fused 5-membered heterocycle. Recognizing that this solvent-exposed portion of the molecule would tolerate many substituents, we sought to introduce polar or charged functional groups at this position. The carboxylic-acid substituted imidazole present in LI-633 was ultimately found to have an acceptable balance potency and peripheral restriction and was advanced into further characterization. In the ³H-flunitrazepam displacement assay, LI-633 was found to be a potent binder of GABRA (Figure 1B), with a K_i of 21 nM. This value is comparable to the affinity of diazepam, which has a K_i of 7.1 nM.

Figure 1.

(A) Structure of LI-633; (B) displacement of^3H-flunitrazepam from rat brain tissue by LI-633; (C) time-course of plasma concentrations of LI-633 in Sprague-Dawley rats following single oral doses (1, 3, 10 mg/kg); and (D) locomotor activity of rats following administration of LI-633; (a) locomotor activity measured in 5-minute bins over 120 minutes; (b) cumulative locomotor activity during the 120-minute observation period. n = 11, 10, 10, 10 for vehicle, 3 mg/kg, 10 mg/kg, and 30 mg/kg LI-633 groups, respectively. Data are presented as mean ± SEM.

To determine its functional effect across various GABRA subunits, and to reveal any selectivity for particular subtypes, LI-633 was tested by manual patch clamp electrophysiology in cell lines expressing α₁β₂γ₂, α₂β₂γ₂, α₃β₂γ₂, and α₅β₂γ₂, receptors. As shown in Table 1, LI-633 produced robust potentiation of GABA-induced inward current, with EC₅₀ values ranging from 8 nM (α₅β₂γ₂) to 128 nM (α₃β₂γ₂).

Table 1.

Activity of LI-633 Against Different GABA_A Receptor Subtypes

GABAA receptor	EC50, μM	Emax, %a
α1ß2γ2	0.093	236
α2ß2γ2	0.009	222
α3ß2γ2	0.128	208
α5ß2γ2	0.008	79.4

GABA, gamma-aminobutyric acid.

^aE_max is normalized to the current elicited by 2 μM GABA; an E_max of 100% thus represents a doubling of the response to 2 μM GABA.

LI-633’s specificity was tested using a panel of 87 diverse targets, including enzymes, G protein-coupled receptors (GPCRs), and transporters in rat brain preparations. As expected, LI-633 resulted in displacement of the benzodiazepine antagonist flumazenil and benzodiazepine flunitrazepam. No other significant interactions were observed in this assay, including GABA-B receptors (data not shown).

The in vitro absorption, distribution, metabolism, and excretion (ADME) profile of LI-633 is shown in Table 2. Apparent permeability was assessed in MDR1-MDCK cell monolayers and was found to be low. The efflux ratio in MDR1-MDCK cells was 2.8, suggesting modest efflux by the P-gp transporter. LI-633 was unchanged after incubation with human or rat hepatocytes or liver microsomes. In a panel of 5 CYP450 enzymes, LI-633 displayed no inhibition up to a concentration of 50 uM. Moderate plasma protein binding was observed in human (10.9% unbound) and rat (18.6% unbound).

Table 2.

In Vitro ADME (Absorption, Distribution, Metabolism, and Excretion) properties of LI-633

Kinetic solubility at pH 7.4 (μg/ml)	125
MDR1-MDCK Papp A-B (nm/s)	3.1
MDR1-MDCK Efflux Ratio	2.8
R,H hepatocytes T1/2 (min)	> 120
R,H microsomes T1/2 (min)	> 120
Rat PPB, % unbound	18.6
Human PPB, % unbound	7.5
Rat BTB, % unbound	13.1
CYP450 IC50, μM	> 50

NOTE. MDR1-MDCK: Madin-Darby canine kidney cells transfected with the human MDR1 gene encoding P-glycoprotein (P-gp, a membrane transport protein that actively pumps many drugs and xenobiotics out of cells).

NOTE. Papp A-B: apparent permeability from apical to basolateral side.

NOTE. Efflux ratio: Ratios ≥ 2 are commonly interpreted as evidence of active efflux and that the compound is likely an MDR1/P-gp substrate; high ratios (eg, > 3–5) are associated with poor CNS penetration and/or limited oral absorption depending on the context.

BTB, brain tissue binding; CYP450, cytochrome P450 enzymes; H, human; PPB, plasma protein binding; R, rat; T1/2, Half-life of disappearance.

Pharmacokinetic Profile

The plasma levels of LI-633 were measured after administering oral doses of 1, 3, or 10 mg/kg in Sprague-Dawley rats (Figure 1C). LI-633 was rapidly absorbed, reaching a maximum plasma concentration less than 1 hour after administration.

Additionally, the concentration of LI-633 in the small and large intestine, DRG, and sciatic nerve (SN) was measured at 3 time points following an oral dose of 10 mg/kg. The LI-633 concentrations were modest in DRG and SN (the ratio of DRG/plasma and SN/plasma were 0.367 and 0.283, respectively, 1 hour after administration while not detectable after 4 and 12 hours). However, a large amount of LI-633 was retained in the small and large intestines. The ratio of small intestinal to plasma concentration was 49, 58.1, and 41.7 at 1, 4, and 12 hours after administration, respectively, whereas the ratio of colon to plasma was 0.916, 124, and 374 at 1, 4, and 12 hours, respectively, after administration.

The potential for CNS penetration of LI-633 was assessed following a single 100 mg/kg oral dose. The total brain/plasma ratio at 1-hour post-dose was 0.021 (K_p), and the unbound brain/plasma ratio was 0.015 (K_pu,u). The cerebrospinal fluid (CSF)/plasma ratio 1-hour post-dose was determined to be 0.007 (K_p). These results demonstrate that LI-633 does not readily cross the blood-brain barrier, even after a high oral dose.

Open-field Locomotor Activity

Following single oral doses of 3, 10, or 30 mg/kg, the locomotor activity of rats placed in an open field was unchanged compared with vehicle-treated animals, as shown in Figure 1D, indicating that LI-633 does not produce measurable sedation at these doses.

Electrophysiological Effects of LI-633 on Rat DRG Neurons

Next, we evaluated the ability of LI-633 to potentiate muscimol-induced currents in rat DRG neurons. In a concentration-response experiment, the GABA-A selective agonist muscimol was found to induce GABRA activation with an EC₅₀ of 5.7 uM (Figure 2A, top), consistent with previous reports.^17,18 Representative current traces showing the potentiation of muscimol current by LI-633 are shown in Figure 2A (middle). In the presence of an EC₃₀ concentration of muscimol (3 uM), LI-633 potentiated GABAergic currents in a dose-dependent manner, with an EC₅₀ of 70.4 nM and an E_max value of approximately 100% (Figure 2A, bottom). Thus, LI-633 acts as a potent GABA-A PAM in native rodent tissue, in agreement with in vitro EC₅₀ results from heterologous expression systems (Table 1).

Figure 2.

LI-633 potentiates muscimol-induced inhibition of current in rat and human DRG neurons. (A) Top: GABAergic currents elicited by muscimol using whole cell manual patch clamp recordings were made on rat DRG neurons. (a) Representative currents shown with 4-second agonist exposure. Maximal current activation was demonstrated with 100 μM (no further increase with 300 μM). Currents showed expected activation and inactivation characteristics. Activation was more rapid with higher concentrations and desensitization increased as with higher concentrations of muscimol. (b) Mean I/Imax data (±SD) with fitted concentration response curve. Middle: Representative recording from a rat DRG neuron, showing potentiation of muscimol-induced current by LI-633. Bottom: Concentration-dependence of potentiation of currents by LI-633 in rat DRG neurons. (B) Top: Data from 2 groups of human DRG neurons in culture. Group A consisted of neurons to whom GABA was applied twice after a 5-minute washout period with no apparent desensitization of evoked calcium fluorescence. In group B, muscimol (M) was applied initially, followed by muscimol and bicuculline (B), with a robust attenuation of the calcium signal. Bottom: Time course of calcium signal in response to muscimol and its inhibition by bicuculline in an illustrative experiment.

Effects of LI-633 on Excitability of Human DRG Neurons

To test feasibility, we first examined the effects of high concentrations of muscimol, a GABA-A receptor agonist on human DRG neurons. A robust calcium signal was obtained, which was reproduced after washout and a second application of GABA (n = 13). In a separate group of neurons (n = 18), bicuculline blocked the effects of the second application, suggesting at high concentrations at least, the excitatory effects of GABA on human DRG neurons were mediated by GABRAs (Figure 2B).

We next tested the effects of GABA itself on stimulated human neuronal excitability. Electrical field stimulation (EFS) is an established technique for studying neuronal excitability, and has been adapted for use with human DRG neurons.¹⁹ Using stimulation with voltage of 1500 to 2000 mV, among the sub-population of neurons that were GABA-responsive, excitability (as measured by calcium fluorescence using Fluo-8) was decreased by GABA in a concentration-dependent manner (Figure 3; n = 12). As positive control, tetrodotoxin (TTX), a Na channel blocker, completely diminished the EFS response in the DRGs. In another set of experiments (n = 14), the effect of GABA (8 uM) was significantly potentiated by in a dose dependent manner by LI-633 (30 and 300 nM).

Figure 3.

GABA-induced reduction in EFS-induced excitability and its potentiation by LI-633 in human DRG neurons.Left: GABA dose-dependently decreased excitability of human DRG neurons in response to EFS. Also shown is the complete suppression of the response to EFS by TTX. Right: LI-633 potentiates the effect of submaximal dose of GABA on human DRG neuronal excitability in response to EFS. Data are normalized to the average of all the baseline responses to EFS. Individual values are shown along with means ± SEM; comparisons indicate q values.

Visceromotor Response to Noxious Colorectal Distension in a Rat Model of IBS

The effects of LI-633 on hyperalgesia were tested in a well-established and accepted rat model of IBS-like pain.²⁰ To assess acute drug treatment effects, IBS and control rats (n = 7–8 per group) were treated with 0, 1, 3, 10, or 30 mg/kg of LI-633 (5 mL/kg, PO). Another group of IBS rats was treated with buprenorphine (0.5 mg/kg, SC), an opioid analgesic, as a positive control. One hour after the first dose, colorectal pain sensitivity was assessed by measuring the VMR response to CRD at 20, 40, 60, and 80 mmHg pressures. The VMR responses to CRD were measured by EMG. In each treatment group, the EMG responses for 4 pressures were calculated into area under the curve (AUC) to present the data for each treatment. As shown in Figure 4, left, IBS rats showed a significant increase in VMR response to CRD. Treatment with LI-633 dose-dependently reduced the hyperalgesia in the IBS rats. Buprenorphine completely blocked the VMR response to CRD in IBS rats, significantly below that of the saline-vehicle group

Figure 4.

Effects of acute and repeated treatments with different doses of LI-633 on visceral hyperalgesia in a rat IBS model. Nocifensive responses were assessed by the VMR response to CRD using 20, 40, 60, and 80 mmHg pressures. Left: VMR responses 1 hour after oral LI-633 using different doses (1, 3, 10, 30 mg/kg). Right: VMR responses 1 day (day 6) after oral LI-633 using different doses (1, 3, 10, 30 mg/kg) daily for 5 days. Data are expressed as an AUC of EMG response to 4 pressures in each treatment group. Individual values (n = 7–8 per group) are shown along with means ± SEM; comparisons indicate q values.

To determine the persistent effects of repeated treatment of LI-633, we continued LI-633 at the same dose for 5 days, and VMR responses to CRD were tested on the day after the last dose. As shown in Figure 4 (right), a day after stopping drug administration, the buprenorphine-treated rats rebounded to the same level as the IBS-vehicle group. However, the effect of LI-633 seemed to persist for 24 hours after the last dose, with the highest doses (10 and 30 mg/kg) still showing statistically significant reductions in behavioral responses compared with the IBS-vehicle group by post-hoc testing.

DR Responses to LI-633 in a Rat Model of IBS

To determine whether the effects of LI-633 on reduction of visceral hyperalgesia in IBS model is peripheral to the spinal cord, we measured the activity of DR single nerve fibers in response to CRD at several intervals over a 2-hour period of time. Colorectal-innervated single fibers in the dorsal root nerve were identified by their response to CRD. The nerve activity in response to CRD was recorded before and after systemic administration of vehicle (n = 3) or LI-633 (5 mg/kg, iv; n = 5) in IBS rat model. When normalized to their baseline, the dorsal root nerve responses were significant attenuated in rats that received LI-633 for both 40 mmHg (2-way ANOVA for drug effect: F_(1,6) = 6.89; P = .039) and 80 mmHg (F_(1,6) = 10.5; P = .018), with post-hoc analyses showing differences at several time points (Figure 5). These results indicate that LI-633-induced reduction of hyperalgesia in IBS rats is associated with attenuation of peripheral nociceptive signals.

Figure 5.

Effects of LI-633 on dorsal root single fiber responses to CRD. CRD was performed at 40 mmHg (left) and 80 mmHg (right) in the rat model of IBS and the response to LI-633 (5 mg/kg, iv; n = 5) or vehicle (n = 3). Data are presented as single fiber activity (means ± SEM) in response of CRD normalized to baseline (before treatment with vehicle or LI-633). ∗q < .05; ∗∗q < .01.

Effects of LI-633 on Excitability of Colonic Sensory Neurons in the Rat IBS Model

To further establish the peripheral site of action of LI-633, we performed whole-voltage patch clamping on colon-specific spinal sensory neurons isolated from the DRG of IBS and control rats treated with 1 and 10 mg/kg of LI-633 (See Methods). The results are shown in Figure 6 (n = 19–22 per group). Resting membrane potential was not affected by either sensitization or drug treatment. Rheobase was reduced in the IBS model but not affected by LI-633 treatment. Membrane resistance trended towards an increase in the IBS model (although not significant by correction for multiple comparisons) but was significantly decreased by both doses of LI-633. Excitability as measured by the number of spikes induced by 2×-rheobase current was significantly higher in IBS rats compared with controls, but this was suppressed to control levels by both doses of LI-633.

Figure 6.

Effects of LI-633 on colonic sensory neuronal electrophysiology. After 5 days of treatment with vehicle or LI-633 (1 and 10 mg/kg, PO), colonic sensory neurons labeled with DiI (see Methods) were dissociated from lumbosacral DRGs and studied by patch clamping for changes in resting membrane potential, rheobase (minimum amount of current to elicit an action potential spike), membrane resistance, and number of spikes elicited after current stimulation with 2× rheobase. Individual values (n = 19–22 per group) are shown along with means ± SEM; comparisons indicate q values.

VMR to Noxious GD in a Rat Model of FD

To further test whether LI-633 also reduces hyperalgesia induced by other disorders, we tested the effect of acute treatment with LI-633 in a rat model of FD.²¹ Sensitized and control rats were treated with LI-633 (10 mg/kg, PO). One hour after administration, visceral hypersensitivity of the rats was assessed by the VMR responses to GD at pressures of 20, 40, 60, and 80 mmHg. Three-way ANOVA revealed significant (P < .001) effects of all 3 parameters: CRD pressure (F_{(3, 40)} = 44.42), sensitization (F_{(1, 40)} = 41.65), and treatment (F_{(1, 36)} = 22.83). VMR responses were significantly increased in FD rats compared with vehicle-treated controls an effect that was significantly attenuated by treatment with LI-633, as shown in Figure 7.

Figure 7.

Effect of LI-633 in a rat model of functional dyspepsia. Acute treatments of LI-633 significantly reduced hyperalgesia in a rat model of FD as assessed by VMR response to GD with 20, 40, 60, and 80 mmHg pressures. Data are presented as mean ± SEM. ∗q < .05; ∗∗q < .01 compared with corresponding CRD pressure for control-vehicle group; ^#q < .01 compared with corresponding CRD pressure for FD-vehicle group.

Discussion

The gut-brain axis involves intricate bidirectional communication through multiple pathways, including neural, endocrine, immune, and metabolic routes. This complexity makes it difficult to isolate and study individual components or mechanisms. In particular, when dealing with disorders affecting this axis, it becomes challenging to separate the central vs peripheral effects of drugs that modulate neurotransmitters, given that several of these are present in both nervous systems. As an example, the role of GABA, although abundantly expressed in the gut along with its receptors, has received less attention in part because current pharmacological agents can reach therapeutic concentrations in both the CNS and peripheral nervous systems. The Gold laboratory was among the first to suggest that the use of peripherally restricted ligands could provide analgesia while sparing the CNS adverse effects associated with GABA-A activation.²² An important step towards understanding the pathogenic role and therapeutic potential of GABAergic signaling in the periphery therefore is the development of pharmacological tools that can clearly separate the peripheral from the central effects. If peripheral GABA signaling is also shown to have a prominent and independent effect on pain, this will provide a robust foundation for developing drugs that can avoid off-target adverse effects in the CNS.²

Towards this end, here we have developed a peripherally restricted GABA-A PAM, LI-633. This compound binds specifically to the GABA-A receptor, with nanomolar affinity to its expected benzodiazepine site, in the classical flunitrazepam displacement binding assay. Further functional electrophysiological characterization using multiple overexpressing cell lines confirmed that LI-633 is devoid of direct GABA agonist activity at all concentrations tested but showed potent potentiation of GABA-evoked current, consistent with being a PAM of GABA-A. Unsurprisingly, the “benzodiazepine-insensitive” α₄ and α₆ containing receptors²³ showed no potentiation in the presence of LI-633 (data not shown). Across the various receptor subunit combinations that were examined, LI-633 demonstrated little to no selectivity to individual GABA-A receptor subunits, similar to most benzodiazepines. The maximum current amplitude evoked in the presence of LI-633 is comparable to that produced by diazepam across all subtypes tested, which is frequently used as a reference point for comparison. Additional screening against a diverse panel of pharmacological targets revealed no major off-target effects by LI-633.

In vitro ADME profiling revealed that LI-633 has good metabolic stability in both rat and human microsomes, relatively high unbound fraction in plasma, and no measurable inhibition of 5 cytochrome P450 enzymes, suggesting broad utility for in vivo testing. In MDR1-MDCK transfected cells, LI-633 showed low passive permeability, as well as moderate efflux by the transporter P-gp. It has been suggested that small-molecule drugs possessing this combination of properties have a lower likelihood of crossing the blood-brain barrier.²⁴ Indeed, the unbound brain to plasma ratio (K_puu) after even a very high oral dose of 100 mg/kg shows that only ∼2% of circulating drug reaches the CNS. Furthermore, LI-633 did not depress locomotor activity following doses up to 30 mg/kg, suggesting that CNS effects would not confound further efficacy experiments.

The direct ability of LI-633 to modulate GABAergic signaling in the periphery was evaluated using native, labeled, and isolated rat DRG sensory neurons, which express a diverse array of GABA-A subunits. The EC₅₀ obtained in rat DRGs (Figure 2A) was in broad agreement with the data obtained by electrophysiology in cell lines. However, an over-reliance on rodent models has been implicated in the poor success rate in translating new drug candidates to the clinic,²⁵ and this concern prompted us to examine the effects of LI-633 in an ex vivo preparation of human DRG neurons. Although others had shown that human DRG neurons express GABRA and other subunits, and are responsive to GABA and its potentiation by diazepam,^26,27 Zhang et al were the first to show that GABA induced both large inactivating low affinity currents and smaller persistent high-affinity currents in all human DRG neurons and that these were blocked by GABA-A antagonists.²² Differences were also noted between human and rat DRGs—most notably, the Cl^- equilibrium potential of −41 mV in human DRG was more hyperpolarized (by about 20 mV) than that observed in rats. Suppression of calcium transients (evoked high potassium) by GABA has also been previously shown in about 30% of cutaneous DRG neurons in culture from naïve rats.²⁸ The results of our experiments build on these studies by demonstrating that both muscimol and GABA were capable of reducing DRG neuron excitability when stimulated electrically. Likewise, LI-633 potentiated the effect of submaximal GABA in a dose-dependent manner. This series of experiments confirm previous work that functional GABRA receptors are present on human DRG neurons, but also that by activating these receptors, their action potential firing is reduced.

Our results are also in good agreement with the findings of Du et al,¹² who showed that GABA application to rat DRG neurons generally decreased the number of action potentials fired under current-clamp conditions. To our knowledge, this is the first demonstration that the excitability of human DRG neurons is reduced by GABA-A activation, and this in keeping with emerging knowledge of GABRA expression in human sensory neurons. A recent comprehensive classification of human DRG neurons based on transcriptomes of cells and nuclei from 126 donors sampled across cervical, thoracic, and lumbar DRGs, showed that GABRA2 is the dominant α-subunit transcript and appears in 2 distinct neuronal classes: non-peptidergic C-fiber nociceptors (C-NP.MRGPRX1/GFRA2) and proprioceptive A-fibers (A-Propr.EPHA3), with the strongest expression in the former. GABRA5 is primarily confined to another nociceptor population (C-NP.MRGPRX1/GFRA2), with no significant enrichment elsewhere. Other α-subunits (GABRA1, 3, 4, 6) were not significantly enriched in any neuronal subtype in this atlas.²⁹ Enhancement of GABA-A signaling in DRG neurons is therefore expected to inhibit 2 major classes of human nociceptors.

By contrast to the CNS, GABA causes depolarization of peripheral neurons, including those in the DRG, because of higher intracellular chloride concentrations.²² Such primary afferent depolarization (PAD) may result in decreased excitability by inactivating voltage-gated sodium channels, activating potassium channels, shunting/increased conductance, and/or decreasing neurotransmitter release, as others have suggested, although these mechanisms remain to be proven definitively.^22,28,30 Such a mechanism has also been postulated to underlie the congenital insensitivity to pain experienced by carriers of sodium channel Na_v1.9 gain-of-function mutations.³¹ In these individuals, persistent activation of a sodium channel leads to a large depolarization of resting membrane potential, which, in turn, inactivates a significant population of other sodium channels, leading to the failure of action potentials to propagate. Interestingly, we show in this study that LI-633 reduces the membrane resistance of colonic sensory neurons, supporting the depolarization shunting theory to explain reduced excitability of peripheral sensory neurons.

Proof of concept studies were then carried out to demonstrate the translation of these in vitro effects to in vivo analgesic efficacy. GABA-A PAMs are not direct agonists but augment the effect of GABA and require physiologically relevant concentrations of GABA to have an effect. Experimentally, they therefore also allow us to identify which organs rely on endogenous GABA signaling for modulation of sensory signals. Disorders of the colon such as IBS are an obvious choice because of the relatively high local concentrations of GABA in the tissue (particularly in the myenteric plexus) in both animals and humans.^7,32 Several benzodiazepines have been shown to be effective in relieving so-called functional gastrointestinal symptoms including pain,^33,34 and 1 benzodiazepine- (in combination with an antispasmodic) containing formulation (chlordiazepoxide/clidinium) is approved by the United States Food and Drug Administration for the treatment of IBS, a condition characterized by abdominal pain and altered bowel movements.³⁵ Interestingly, humans with IBS also have reduced levels of glutamic acid decarboxylase (GAD), the enzyme that is responsible for GABA production along with reduced GABA levels in their colons, indicating that GABA-augmenting therapy may potentially also be disease-modifying.³⁶

The Gold laboratory first showed the importance of endogenous colonic GABA in modulating colonic afferent tone primarily via GABA-A signaling. GABA-A antagonists increased afferent excitability and nocifensive responses to colonic distention, whereas agonists had the opposite effect.¹⁴ Further, they showed that activation of low affinity GABRAs (presumably possessing γ subunits) by diazepam and high-dose muscimol attenuates visceral hypersensitivity in mice, whereas activation of high-affinity receptors (presumably with δ subunits) increases it.¹⁵ However, acute inflammation is not a feature of IBS. We therefore used a model that reliably produces hypersensitivity to CRD, even in the absence of overt tissue injury or inflammation.^20,37 Sensitized animals showed significantly reduced behavioral responses to CRD after acute treatment with LI-633 at oral doses as low as 1 mg/kg. From the pharmacokinetic data presented in Figure 1C, this dose is expected to produce unbound plasma concentrations of 52 nM at 1 hour post-dose. Thus, a plasma concentration at, or slightly below, the rat DRG EC₅₀ is sufficient for analgesic efficacy in this model. Examination of the buprenorphine-treated group shows a nearly complete loss of sensitivity to mechanical stimulus, indicating an expected generalized “numbing” effect. By contrast, LI-633 reduces the response only to the level of unsensitized animals. This suggests that LI-633 may reverse the hypersensitivity associated with IBS, while preserving normal sensory function.

A similar effect on visceral hypersensitivity was also seen in rats modeled to have FD, which is equally, if not more, prevalent than IBS in the general population and for which there is no approved drug available in the United States. Although we could only find one study that reported the expression of GABA in gastric tissue (both in the mucosa and muscularis propria),³⁸ the fact that LI-633 can attenuate hypersensitivity to gastric distention in a rat model of FD strongly suggests a role for endogenous GABA in gastric sensory nerve modulation and attests to the use of a peripherally restricted GABA-A PAM to identify such mechanisms.

This study has established that a peripheral mechanism is important in enhancing signals to the brain in a model of visceral hypersensitivity. We cannot conclusively rule out an additional role at the DRG cell body, which is suggested by studies showing that direct application of GABA agonists at this site attenuates neuropathic hypersensitivity.^12,13 Expression of GABA and molecules involved in GABA synthesis and trafficking have been shown using real-time polymerase chain reaction (RT-PCR) and immunostaining,^12,39 and GABA release by DRG cell bodies has been demonstrated.¹² However, this is in contrast to single-cell sequencing studies of DRG neurons that have not reported significant expression of these molecules.^29,39 Regardless, the role of endogenous GABA-A signaling at the DRG soma (as assessed by the application of antagonists) remains unsettled, with one study reporting animals as showing spontaneous nocifensive behavior,¹² while another showed no effect on thermal sensitivity/withdrawal behavior.¹³ On the other hand, Loeza-Alcocer et al were able to demonstrate attenuation of colonic afferent responses distal (ie, peripheral) to the DRG in an ex vivo preparation.¹⁴ LI-633 suppressed the excitability of DR nerve fibers as well as decreased the excitability of colon-specific sensory neuronal bodies, establishing a peripheral mechanism of action for GABA-A as had been suggested in previous studies.

The finding that LI-633 exerts an antinociceptive effect after repeat dosing, even after a 24-hour washout period, can perhaps be explained by the sustained, high concentrations of drug that were observed within the intestinal tissue. A compound with low solubility and very low passive permeability is expected to create a “reservoir” of drug that is slowly absorbed over many hours, and it is likely that therapeutic concentrations are maintained within the gut even after plasma drug levels have returned to baseline. However, the effects on colonic sensory neuronal excitability are more difficult to explain; isolation and culture of these neurons would ensure that there is no exposure to LI-633 at the time of the patch clamping. Although we have no evidence to support this, a possible mechanism may be neuroplastic changes in response to several days of treatment with LI-633 that may result in sustained loss of excitability. In this context, neuroplasticity of CNS neurons has been observed with benzodiazepines in both early and late periods of life.40, 41

Although this study focused on the effects of peripheral GABA-A signaling on nociception, GABA likely plays an important role in gastrointestinal motility as well with well-established mechanisms and often opposing effects in both the CNS and ENS.^8,42 With the use of tools such as LI-633, it is anticipated that these actions can now be further clarified.

Materials and Methods

Rat Pharmacokinetic Experiments

Experiments were carried out at Pharmaron (Beijing, China) using male Sprague-Dawley rats.

Nonspecific GABA Binding Assay

Tritiated flunitrazepam displacement from rat brain tissue was performed as described by Speth et al⁴³ at CEREP, Celle l'Evescault, France.

Manual Patch-clamp in Recombinant Cell Lines

Experiments were carried out by B’Sys GmbH (Witterswil, Switzerland). Chinese hamster ovary (CHO) cells stably expressing human GABA-A α₁ß₂γ₂ receptors and leukocyte tyrosine kinase (LTK) cells stably expressing human GABA-A α₂ß₂γ₂, α₃ß₂γ₂, or α₅ß₂γ₂ receptors were used for all experiments.

Off-target Binding Panel

LI-633 was tested against a panel of 87 enzymes, receptors, and ion channels (Eurofins Panlabs). LI-633 was tested in duplicate at a concentration of 10 uM.

Rat DRG Cell Electrophysiology

Experiments were carried out at Metrion Biosciences (Cambridge, United Kingdom). DRG neurons were isolated from P8-P12 Sprague-Dawley rats, were prepared and maintained in culture for up to 4 days on glass coverslips. Whole-cell manual patch clamp recordings were obtained from cell bodies.

Human DRG Neuron Excitability

Experiments were carried out at AnaBios (San Diego, California) using tissue obtained by legal consent from organ donors in the United States. For this study, human DRG neurons were obtained from a deceased 44-year-old (for the muscimol studies) and a 25-year-old (for the GABA and LI-633 studies): both were Hispanic females. DRGs were transferred into a dissection vessel containing a cold (4°C) proprietary dissection solution. DRGs were maintained completely submerged in dissection solution and dissected appropriately. Each DRG was enzymatically dissociated as per AnaBios’ proprietary methodologies and cultured for 3 days prior to experiments with the test article. Responsiveness to GABA was defined as a greater than 25% decrease in response to the test article in the setting of electrical stimulation with 12 of 22 neurons (∼54%) tested meeting this criterion.

Open-field Locomotor Activity

Eight-week-old male rats were ordered from Charles River and allowed to acclimate to our animal facility for 1 week. Rats were then handled daily (∼1 minute) for 7 consecutive days prior to behavioral testing. On the test day, the rats were transported to the behavioral testing room, weighed (290–330 g), and left to acclimate for 1 hour. After the 1-hour waiting period, the rats were dosed (by mouth [PO]; volume: 5 mL/kg) with either vehicle (5% dimethyl sulfoxide [DMSO] in phosphate-buffered saline [PBS]; pH = 8) or a single dose of LI-633 (3, 10, or 30 mg/kg) and then placed into a novel, acrylic open-field arena (445 mm length × 445 mm width) for 2 hours. The sessions were recorded, and locomotor activity was analyzed using Topscan (CleverSys, Inc) automated behavioral analysis software.

Neonatal IBS Model and Pseudoaffective Responses to Noxious Colorectal Distention

The IBS rat model was generated as we have previously described.²⁰ Briefly, male Sprague Dawley rat pups at postnatal day 6 were purchased from Harlan Laboratories with dams (previous studies have shown that GABA-A has equipotent effects on hyperalgesia in males and female rodents).¹³ At postnatal day 10 to 12, pups received a colorectal infusion of 0.2 ml of 0.5% acetic acid (AA) and were allowed to grow up until 8 to 12 weeks of age. For acute treatment, adult IBS or control rats were treated with LI-633 (1, 3, 10, and 30 mg/kg, 5 mL/kg, PO) or vehicle (5% DMSO in 0.1 M Na-phosphate buffer [pH 8]). As a positive control, another group of IBS rats was treated with buprenorphine (0.5 mg/kg, subcutaneously [SC]). One hour later, hyperalgesia was assessed by visceromotor reflex (VMR) responses (measured by electromyographic recordings of the external oblique abdominal muscle) to colorectal distention (CRD) at 20, 40, 60, and 80 mmHg pressures. For repeated treatment, control and IBS rats were treated with LI-633 (1, 3, 10, and 30 mg/kg, 5 mL/kg, PO) or vehicle (5% DMSO in 0.1 M Na-Phosphate buffer [pH 8]). As control, another group of IBS rats was treated with buprenorphine (0.5 mg/kg, subcutaneously) once a day for 5 days. The VMR response to CRD was then measured on day 6, 24 hours after the last administration.

Dorsal Rootlet Single Nerve Unit Recording

To determine the effect of LI-633 on peripheral sensory signal to the CNS, we tested the activity of dorsal rootlet (DR) single fibers using single-unit afferent recordings as previously described by us.⁴⁴ Briefly, after anesthesia, the T13 to S1 vertebral column was mounted on a stereotaxic frame and fixed by clamping. The spinal cord (L1–6) was exposed by laminectomy. In an oil pool, DR were then also exposed by carefully cutting and deflecting the dura mater using a fine forceps. To record afferent fiber activity, the lumbar 5 DR was first cut centrally (as close to the cord entry as possible) and freed from the spinal cord. The nerve activity was recorded at the distal ends using a silver hook electrode. Single units that innervate the colon were identified by a consistent spike rate in response to CRD. Signals were amplified with Iso-DAM8A Bio-amplifier (WPI) and analyzed using SPIK 2 software program (Cambridge Electronic Design). A 20-second baseline was recorded, followed by CRD for 20 seconds. Two pressures, 40 mmHg and 80 mmHg, were used for CRD. The single unit nerve activity was measured before and 10, 30, 60, 90, and 120 minutes after administration of LI-633 (5 mg/kg, intravenously [iv]). The data were normalized to the baseline signal activity obtained before administration of the drug.

Ex Vivo Measure of Excitability on Colonic Sensory Neurons in IBS Rats

To label DRG neurons that innervated the distal colon, a retrograde dye, 1,1’-dioleyl-3,3,3’,3-tetramethylindocarbocyanine methanesulfonate [DiI] (10 mg/mL methanol) was injected into the distal colon wall (1μL/site × 10 sites) of rats, previously sensitized with saline or AA as neonates as described by us.⁴⁵ Two weeks later, rats were treated with vehicle or LI-633 (1 or 10 mg/kg, 5 mL/kg, PO) once a day for 5 days. On the day after the last treatment, rats were administrated with 1 dose of the treatment and were sacrificed 1 hour later. DRGs (L4–S3) were harvested, dissociated, and plated, and whole-cell voltage patch-clamp recordings of the cultured DRG neurons were conducted in DiI-positive neurons as previously described by us.⁴⁶

Effects of LI-633 Treatment Visceral Hypersensitivity in a Rat Model of FD

A rat model of FD was generated using transient neonatal gastric irritation previously described by us.^21,44 Ten-day-old male rats received 0.2 mL 0.1% iodoacetamide (IA) in 2% sucrose daily by oral gavages for 6 days; controls received 2% sucrose. Adult FD and control rats (∼ 12 weeks old) were administered with LI-633 (10 mg/kg, PO) or vehicle (5% DMSO in 0.1 M Na-phosphate buffer, pH 8). One hour after the treatment, gastric hyperalgesia of the rats was examined by the VMR responses to gastric distention (GD), as measured by electromyographic recordings of the acromotrapezius muscle.

Statistical Analysis

Data were analyzed by Student t-test, 1-, 2-, or 3-way analysis of variance (ANOVA) using PRISM (Graphpad Software) unless otherwise indicated. If a significant difference was detected, post hoc testing using 2-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli for correction for multiple comparisons was used to evaluate differences between individual groups and statistical significance reported by q values as appropriate. Data are expressed as mean ± standard error of the mean (SEM), unless otherwise indicated.

Conclusion

In summary, the present data demonstrate that LI-633 is a potent and selective PAM of GABRAs, with robust reductions in the excitability of both rodent and human sensory neurons and antinociceptive activities in a rat IBS and FD model. Taken together with previous studies and human DRG data, these results support the potential clinical benefit of drugs acting on GABRA receptors to enhance GABA signaling within peripheral sensory afferents in a variety of painful visceral disorders.