GABAkines – Advances in the Discovery, Development, and Commercialization of Positive Allosteric Modulators of GABA_A Receptors¹

A. History.

Although GABAkines have existed for years, it was not until the 1970s that GABA was implicated in their pharmacological actions. Traditional medicine utilized a host of hypnotic and sedative agents (e.g., ethanol, opioids, cannabinoids) with a transition to the use of paraldehyde, chloral hydrate, and bromides at the end of the 19^th century. Barbiturates were initially synthesized by Adolf von Baeyer in 1864 and were introduced to the market as sedatives, hypnotics and antiepileptics in the early 20^th century (López-Muñoz et al., 2005). Their use was rapidly expanded by the introduction of approximately 50 barbiturates, producing diverse clinical profiles but narrow therapeutic windows which created the potential for overdose, abuse, dependence and for exacerbation of seizures upon withdrawal (Isbell, 1950). The mechanism of action of barbiturates is not completely understood where modulation of multiple voltage-gated and ligand-gated ion channels has been proposed (Löscher and Rogawski, 2012). The main modulatory effect is through modulation of GABA_ARs with potentiation at low concentrations and direct activation at higher concentrations, pharmacological properties that decrease the safety of barbiturate GABAkines relative to the benzodiazepine GABAkines. Inter-subunit binding sites in the transmembrane domain of the human GABA_AR were identified (Chiara et al., 2013) but precise binding domains and their structural impact upon binding remain to be fully characterized (Scott and Aricescu, 2019). The medical use of barbiturates declined in the 1950s after the introduction of meprobamate and the 1,4-benzodiazepines.

From the era of meprobamate and the benzodiazepines, there have been many new GABAkines discovered and some of these advanced into clinical development. The primary therapeutic areas of interest for these compounds were anxiety and epilepsy. However, other areas of interest were explored as well.

B. Anxiolytic Activity and Sedation.

One issue with the benzodiazepine anxiolytics that is key to understanding their therapeutic value as well as an aspect of their pharmacology that impedes therapeutic utility is dose-dependent sedation. While sedation is sometimes desired, as in sleep induction, it is a dose-limiting side-effect for the majority of therapeutic applications.

From the standpoint of both efficacy and side-effect liabilities, improvements in standard-of-care medicines for the symptomatic relief of anxiety are needed. Some antidepressant drugs are used for the treatment of anxiety (Bandelow, 2020). Although these compounds are generally not sedating, they bear other side effects such as treatment-mergent weight gain and sexual dysfunction (Atmaca, 2020; Gill et al., 2020). Moreover, their anxiolytic actions, like their effects on depression (Katz et al., 2004), require weeks of daily dosing (Cvjetkovic-Bosnjak et al., 2015). Due to such limitations, benzodiazepine anxiolytics continue to be used for the acute treatment of anxiety. Drugs like alprazolam are widely prescribed but come with the burden of sedation and motor-impairment, as well as abuse and dependence potential (Străulea and Chiriţă, 2009; Reissig et al., 2015; Duke et al., 2020). Moreover, drugs like alprazolam can facilitate the suppression of respiration that has led to emergency room visits and deaths (Jann et al., 2014; Hedegaard et al., 2018; Witkin et al., 2019a).

Rational drug discovery efforts directed at creating improved GABAkines came from basic pharmacological data along with the discovery of the benzodiazepine receptor (Möhler and Okada, 1977; Squires and Brastrup, 1977) and its role in potentiation of GABA currents by GABAkines (Choi et al., 1977) (see historical overviews by (Schallek et al., 1979; Tallman et al., 1980; Haefely, 1989)). This discovery enabled establishment of binding assays (Williamson et al., 1978), with promising ligands evaluated in animal models for efficacy and reduced unwanted side effects (reviewed in Skolnick, 2012).

Multiple compounds progressed into clinical trials due to their favorable preclinical profile including bretazenil, abecarnil, alpidem, and ocinaplon (Fig. 4). Alpidem was approved as an anxiolytic with relatively little sedation (Musch et al., 1988) but was later withdrawn due to a high occurrence of hepatitis (Baty et al., 1994). Ocinaplon demonstrated anxiolytic effects without sedation (Lippa et al., 2005; Czobor et al, 2010), but was later withdrawn from further development due to concern over incidents of liver enzyme elevation.

Initial pursuit of the ideal GABAkine had been directed toward the creation of a non-sedating anxiolytic (see Skolnick, 2012 for review). The potential for the discovery process to proceed on rational grounds was first given hope by Lippa and colleagues. Based upon the ability of some compounds to produce anxiolytic-like effects without sedation in animal models, it was hypothesized that multiple benzodiazepine receptors exist and that some mediate anxiolytic effects and others sedation (Klepner et al., 1979; Lippa et al., 1981; Lippa et al., 1982). The non-sedating anxiolytic-like compound, CL218,872 (Fig. 5) was the first to jump start this idea.

The advent of molecular biology enabled further refinement to the search for anxio-selective drugs. The concept of benzodiazepine type 1 and type 2 receptors (Lippa et al., 1978, 1981, 1982; Klepner et al., 1979) was integrated into the current understanding of the structure and function of the GABA_AR. Basic and applied research in this area focused on the specific α-subunit comprising the GABA_AR assembly since modifications of this subunit produced major changes in pharmacological activity. Genetic, pharmacological, and behavioral evidence was used to suggest that α1-subtype-containing GABA_ARs preferentially mediate the sedative, amnestic, ataxic effects, and dependence production of compounds (Rudolph et al., 1999; McKernan et al., 2000; Wafford, 2005; Licata et al., 2009; Ator et al., 2010; Tan et al., 2010), whereas α2- and α3-subtypes mediate anxiolytic effects (Löw et al., 2000; Rudolph and Möhler, 2014), anticonvulsant (Rivas et al., 2009), and pain therapeutics (Dias et al., 2005; Lewter, 2019). The α5-subtype has been implicated in memory function (Collinson et al., 2002; Dawson et al., 2006). Based primarily on the data associating α1-containing GABA_ARs receptors with sedation, discovery efforts were directed at the identification of GABAkines with preference for α2- and α3- relative to α1-containing GABA_ARs (Rudolph and Knoflach, 2011) (Fig. 5).

The α1 hypothesis has been challenged by both animal and human pharmacological data. For example, ocinaplon (Fig. 4), produced no sedation at the dose that produced significant anxiolytic effects, despite having higher efficacy at potentiating currents evoked in α1 GABA_ARs than α−2, −3 and −5-containing GABA_ARs (Lippa et al., 2005). And, as reviewed later, the α1-sparing, subtype-selective GABAkines showed more sedation than predicted from their pharmacological profiles in the preclinical studies. There continues to be a great deal unknown about the structural determinants of specific GABA_ARs that correspond to specific efficacy and side-effects (Sieghart and Savić, 2018; Skolnick, 2012). This uncertainty is exacerbated by the complexity of behavioral endpoints, specific GABA_AR localization, and pharmacology (see also Maramai et al., 2020 for an overview).

One of the first such molecules investigated was L–838,417 (Fig. 5), a partial agonist at α2,3- and α5-containing GABA_ARs and a negative allosteric modulator at α1-containing receptors. L-838,417, produced anxiolytic-like effects in the elevated plus maze but did not impair motor activity (McKernan et al., 2000; Carling et al., 2005). Another three compounds progressed into clinical studies: two analogs of L–838,417 (TPA-023 and MRK-409) and a structurally unrelated compound, TPA-023B (Fig. 5) (Atack, 2011). All three compounds were partial agonists at α2/3 subtypes with no substantial efficacy at α1-containing GABA_ARs in vitro (Atack et al., 2006) and they were all efficacious in animal models of anxiety without observed sedation (Atack, 2009). Two other α1-sparing, subtype-selective GABAkines have also been brought forward: NS11821 from Neurosearch (structure not disclosed) and AZD7325 from Astra Zeneca (Fig. 5).

Despite preclinical hopes, clinical data on these compounds presented a more complex picture. All of these compounds produced more sedation, dizziness, drowsiness, and motor incoordination than was hoped (de Haas et al., 2007, 2008, 2012; Atack, 2009; Atack et al., 2010; Fujita et al., 1999). NS11821and AZD7325 also exhibited more dizziness, somnolence, and sedation in humans than hoped but were otherwise well-tolerated (Zuiker et al., 2016) (Chen et al., 2014; Jucaite et al., 2017). For example, in the study by (Zuiker et al., 2016), NS11821 is a partial GABA_AR agonist with relatively dominant α2,3 and α5 subtype efficacy but negligible α1 agonism. The first-in-human study was performed in healthy male subjects using a single-dose, parallel, double blind, placebo-controlled, randomized, dose-escalation study design with six cohorts (N=48) enrolled. The eight subjects of each cohort received NS11821 (10 mg, 30 mg, 75 mg, 150 mg, 300 mg or 600 mg) or placebo in a 6:2 ratio. At low dose levels, NS11821 had relatively low exposure, probably due to poor solubility. Saccadic peak velocity decreased in a dose-related manner while limited impairments were seen on body sway and the visual analogue scale for alertness. The most common adverse events were somnolence and dizziness, which were more prominent with the higher doses. Although no positive control was used in this study, the results were compared post hoc with a dataset for lorazepam (2 mg). The maximum saccadic peak velocity effects seemed comparable to the typical effects of lorazepam, whereas the other central nervous system effects were smaller.

None of these compounds has yet progressed to become a medicine. There are many reasons for a drug not achieving the difficult milestone of becoming an approved therapeutic and the present review will not address this complex topic. For subtype selectivity discussions, the commentary by Sieghart and Savić (2018) and Skolnick (2012) is recommended.

Nonetheless, these compounds continue to be highly utilized research tools. For example, L838,417 was also utilized in modeling the pharmacophore for driving potency and efficacy of α_1,2,3,5βγ₂ GABA_A receptor interactions where the α₁-Gly²⁰¹/α₃-Glu²²⁵ appears to be pivotal (Söderhielm et al., 2018).

In an elegant series of experiments, Lorenzo et al. (2020) used L838,417 as a tool to help dissect the α2-GABA_AR basis for regulation of neuropathic pain. The low abuse potential of this compound was shown by its lack of self-administration in Rhesus monkeys (Huskinson et al., 2019; Berro et al., 2021); triazolam and lorazepam were self-administered (Berro et al., 2021). As with chronic pain, itch continues to require improved therapies. TPA 023B was utilized as one of the tools to help define α2- and α3-containing GABA_ARs in the regulation of itch (Ralvenius et al., 2018). Zeilhofer and colleagues also reported that TPA-023B suppresses the affective component (tonic) of pain (Neumann et al., 2021). Taken together, these findings lead to increased mechanistic understanding of diseases and symptoms that are in search of better medicines.

Even though benzodiazepines, as a class, act at all γ subunit containing GABA_ARs (α1,2, 3, and 5), some compounds produce less sedation than others. One such compound is the 1,5 benzodiazepine clobazam (Wildin et al., 1990; Sankar, 2012), whose milder sedative liability could have contributed to its approval as an add-on therapy for Lennox-Gastaut syndrome (Ng et al., 2011). A small proof of concept clinical trial also reported reduction of capsaicin-induced hyperalgesia with clobazam (Besson et al., 2015). The activity of clobazam might be due, at least in part, to its active metabolite, N-desmethyl-clobazam. N-desmethyl-clobazam exhibits functional selectivity for α2/3/5-containing GABA_ARs (Ralvenius et al., 2016). N-desmethyl-clobazam produced significant analgesia in rodent models without sedation (Ralvenius et al., 2016). A patent for the clinical use of N-desmethyl-clobazam for chronic pain has been filed (Ralvenius et al., 2016). A Phase-1 clinical trial with N-desmethyl-clobazam showed no sedation, but there was also no significant efficacy in reversing hyperalgesia (Matthey et al., 2020). The compound has been registered for a Phase-2 clinical trial for treatment of peripheral neuropathic pain (Besson, 2020).

C. Epilepsy.

Many first-generation antiepileptic drugs have primary actions as GABAkines. These include valproic acid (Depakene), clorazepate dipotassium (Tranxene), clonazepam (Klonopin), diazepam (Valium), phenobarbital (Luminal), and primidone (Mysoline) (Gitto et al., 2010), some of which are still in use for the treatment of epilepsy (Gitto et al., 2010). For example, valproate is a GABAkine with use in generalized seizures and status-epilepticus (Rahman and Nguyen, 2020; Liampas et al., 2021). GABA-related side effects of dizziness and somnolence are induced and for which discontinuation of medication can occur (Rahman and Nguyen, 2020).

Despite the continued use of some older generation GABAkines, improvements are needed in both efficacy and safety (Witkin et al., 2021). It is estimated that in about 70% of patients, existing drugs are not fully efficacious in controlling seizures (Marson et al., 2007; Sinha and Siddiqui, 2011; Banerjee et al., 2014). Despite being maintained on multiple antiepileptic medicines, many patients continue to have seizures (Błaszczyk et al., 2018). For these reasons, some patients elect to have invasive therapeutic procedures such as surgical resection or disconnection (Adelson, 2001; Hwang and Kim, 2019). Pharmacoresistance to anticonvulsant therapy continues to be one of the key obstacles to the treatment of epilepsy (Franco et al., 2014). A recent review highlights the GABAkines that are being considered for focal epilepsy (Janković et al., 2021).

D. Other Therapeutic Indications.

Since GABA is a pervasive and primary neuro-inhibitor, GABA influences other biological functions in addition to anxiety and seizure activity. As such, GABAkines have been considered as potential therapeutic modulators for diverse disorders including schizophrenia, pain, depression, anxiety, cognition, and disorders of cardiovascular function. A recent review is highly informative (Maramai et al., 2020). This section will provide a very brief summary of the current focus of GABA_ARs that have α4, α5, or α6 subunit composition. The therapeutic potential of compounds acting at these receptors is less defined than those acting receptors containing α1-, α2, and α3 subunits.

Early work on GABAkines that specifically amplified GABA signaling through α5-containing GABA_ARs identified potential therapeutic value in the areas of schizophrenia and airway muscle relaxation (c.f., (Clayton et al., 2015)) (see reviews by Jacob, 2019; Mohamad and Has, 2019). Current work outlining a rationale for the potential therapeutic value of these selective GABAkines has recently been published in the diverse areas of schizophrenia, pain, depression, anxiety, cognition, and cardiovascular function (c.f., (Batinić et al., 2017; Donegan et al., 2019; Hernández-Reyes et al., 2019; Prevot et al., 2019; Bojić et al., 2021; Davenport et al., 2021; Fee et al., 2021; Franco-Enzástiga et al., 2021); however, see (Xue et al., 2017)).

α4-and α6-containing GABA_ARs are not sensitive to typical GABA modulators (diazepam-insensitive) and their focus is beyond the scope of this review. α4β3δ assemblies are considered extrasynaptic and as such have key relevance to some of the newer advancing neuroactive steroid GABAkines. α4-containing GABA_ARs also appear relevant for the control of breathing (Yocum et al., 2016).

The prospect of a therapeutic impact of α6-containing GABA_ARs has also not been neglected for multiple areas including pain and Tourette syndrome (c.f., (Huang et al., 1999; Chiou et al., 2018; Vasović et al., 2019; Tzeng et al., 2020; Cadeddu et al., 2021)).

A. Neuroactive steroids.

Three of the principal neuroactive steroids in clinical development are shown in Fig. 6. These GABAkines have potential for treating a host of neurological and psychiatric disorders (Gasior et al., 1999; Miziak et al., 2020), and are being developed primarily for epilepsy and depression. Indeed, a role for neuroactive steroids in epilepsy has been postulated since the publication of early preclinical data (Belelli et al., 2019; Gasior et al., 1999; Reddy and Rogawski, 2012; Miziak et al., 2020).

B. PF‐06372865 (darigabat, formerly CVL-865)

C. KRM-II-81

KRM-II-81 was first disclosed as a non-sedating anxiolytic-like compound in 2016 from the laboratory of James M. Cook (Poe et al., 2016) (Fig. 7). KRM-II-81 is an imidazodiazepine that preferentially activates α2/3-containing GABA_ARs (Table 2). Preclinical data support the proposition that KRM-II-81 demonstrates reduced sedation, motor-impairing effects, tolerance development, and abuse liability compared to 1,4-benzodiazepine GABAkines. KRM-II-81 has also shown efficacy in animal models that are used to detect compounds that are anxiolytic, antidepressant, as well as those used in chronic and neuropathic pain states. KRM-II-81 has most extensively been studied as an antiepileptic where it has shown greater potency and efficacy than diazepam in several animal models and greater efficacy in animal models of pharmaco-resistant epilepsy. In addition, KRM-II-81 has demonstrated in situ activity in human translational studies using brain tissue from treatment-resistant epileptic patients. No clinical data have been reported with KRM-II-81. It is currently under development by EndeavorRx, a business unit of RespireRx Pharmaceuticals Inc.

KRM-II-81 was first synthesized from HZ-166 (Poe et al., 2016). HZ-166 (Fig. 7) was identified as an α2/3-preferring GABAkine that displayed reduced propensity for sedation and motor impairment compared to diazepam (Cook et al., 2009), and was active in models that detect anxiolytic drugs (Fischer et al., 2010), anticonvulsant compounds (Rivas et al., 2009), and antinociceptive compounds (Di Lio et al., 2011). HZ-166 contains a metabolically-labile ester function that reduces its overall bioavailability (Poe et al., 2016); KRM-II-81 was rationally designed to increase oral bioavailability and retain selectivity for α2/3-containing GABA_ARs by replacing the ethyl ester with an oxazole bioisostere. KRM-II-81 displayed good bioavailability after i.p. and oral administration in rats with exposure of plasma and brain (Poe et al., 2016). Both oral and intraperitoneal dosing in rats produced detectable unbound plasma levels of KRM-II-81 from 0.25–12h post dosing (Witkin et al., 2019a).

Electrophysiological experiments have established that KRM-II-81 potentiates α2- and α3-containing GABA_ARs with little amplification of α1-containing GABA_ARs (Table 2). In contrast, diazepam also amplifies responses of GABA in recombinant cells containing α1- and α5-containing GABA_ARs (Table 2). Compared to PF-06372865 (α2/3/5-selective), KRM-II-81 is 9000 times less potent at α1 and 11 times less potent at α5. Thus, in addition to low efficacy at α1- and α 5-containing GABA_ARs, KRM-II-81 requires higher concentrations to potentiate these ‘off-target’ GABA_ARs. When studied for its activity against a host of other receptor proteins, KRM-II-81 did not display any off-target liabilities (Poe et al., 2016).

Preclinical data have been reported in a broad range of seizure models using rodents and human epileptic brain tissue. KRM-II-81 displayed wide-ranging anticonvulsant activity against acute chemoconvulsant challenges (Witkin et al., 2018; Knutson et al., 2020) and electrically-induced seizures (Witkin et al., 2018; Knutson et al., 2020). KRM-II-81 also blocked seizure sensitization (kindling) (Witkin et al., 2018; Knutson et al., 2020). Importantly, like the neuroactive steroids, but unlike diazepam, KRM-II-81 prevented the development of the seizure-kindling process (Knutson et al., 2020). That α2-containing GABAARs are sufficient for anticonvulsant activity is evidenced by the finding that blockade of α1-containing GABA_ARs by β-CCT did not nullify the anticonvulsant effects of diazepam at doses that were sufficient to block its motor-impairing effects (Witkin et al., 2018).

Traumatic brain injury engenders long-term negative health outcomes including posttraumatic epilepsy that is associated with affective, neurocognitive, and psychosocial disruption of life (Semple et al., 2019) and for which there are no effective treatments (Temkin, 2009; Wat et al., 2019). Effects of KRM-II-81 on inhibiting cortical network activity was studied using an in vivo two-photon imaging technique in which neuronal activity in cortical layer II/III pyramidal neurons was assessed in mice with traumatic brain injury. Dramatic increases in fluorescence, measuring the intensity of spiking activity of individual neurons and fraction of active neurons, was observed 3 months after brain injury. Both the mean integrated fluorescence and fraction of active neurons were markedly decreased after KRM-II-81 (Witkin et al., 2020). The possibility that KRM-II-81 might also positively reduce the probability of post-traumatic epilepsy is raised by these findings.

In animal models of pharmaco-resistant epilepsy, KRM-II-81 is also active and, in many cases, better than standard of care antiepileptics (Witkin et al., 2020). The mechanisms for pharmaco-resistance are likely multifaceted; nonetheless, tolerance development has been suggested to play a major role (Löscher and Schmidt, 2006). KRM-II-81 did not produce tolerance, over five days of dosing, to the suppression of convulsions in mice induced by pentylenetetrazol (Witkin et al., 2018). In a kainate-induced chronic epilepsy model, KRM-II-81 given daily for five days did not exhibit decreases in anticonvulsant efficacy (Witkin et al., 2020). Similar findings were reported by Rivas et al. (2009) with HZ-166. Longer term dosing studies are required to determine whether KRM-II-81 has a reduced propensity to produce tolerance as suggested by the data in pain models (described below) and as predicted from point mutation studies implicating α2-containing GABA_ARs as a target for the relief of tolerance development (Ralvenius et al., 2015).

To further establish translatability of the anticonvulsant effects of KRM-II-81 to patients, microelectrode recordings were made from freshly transected cortical tissue slices from the brains of juvenile epileptic patients that were refractory to antiepileptic medications. KRM-II-81 decreased cortical firing rates across the neural network (Witkin et al., 2018).

Pain pathways rely heavily on GABA_AR-driven inhibitory neurotransmission (Hammond and Drower, 1984; Dirig and Yaksh, 1995; Enna and McCarson, 2006; Zeilhofer et al., 2015; Etlin et al., 2016). A nice illustration was presented over 25 years ago where inflammatory pain induced by formalin was reduced by intrathecal injection of the direct-acting agonist muscimol. That these effects of muscimol were due to stimulation of GABA_ARs was confirmed through the use of the GABA_AR antagonist bicuculline (Dirig and Yaksh, 1995). More recent studies have shown that intrathecal muscimol can also reduce neuropathic pain endangered by spinal cord injury in rats (Hosseini et al., 2014).

Although GABA is known to be an integral biological mediator of pain, GABAkines are generally not used to control pain (Chou et al., 2017) although prescriptions for pain patients occurs (Wright, 2020). The sedative properties of benzodiazepines have been proposed to interfere with the attainment of the doses needed to produce analgesic benefit (McKernan et al., 2000; Knabl et al., 2009; Munro et al., 2009; Atack, 2010; Ralvenius et al., 2015; Rudolph et al., 1999). A potential role for α1-containing GABA_ARs in the mitigation of the antinociceptive effects of diazepam has been demonstrated in point mutation studies (Knabl et al., 2008, 2009; Ralvenius et al., 2015; Paul et al., 2014; Tudeau et al., 2020). That α2-containing GABA_ARs are responsible for this emergent anti-nociceptive activity was provided by the finding that α2-selective GABAkines, HZ-166 (di Lio et al., 2017), L-838417 (Knabl et al., 2008; Nickolls et al., 2011; Hofmann et al., 2012; Lorenzo et al., 2020), NS11294 (Nunro et al., 2008, 2009), NS16085 (de Lucas et al., 2015), and TPA023B (Nickolls et al., 2011) are analgesic in rodent models.

KRM-II-81 and structural analogs have been reported to be active in animal models of acute and inflammatory pain (Fisher et al., 2017; Lewter et al., 2017; Moerke et al., 2019; Witkin et al., 2019) and chronic neuropathic pain induced by L5/6 nerve ligation (Witkin et al., 2019) or by the chemotherapeutic agent paclitaxil (Biggerstaff et al., 2020). The potential for this mechanism to serve to reduce opioid burden (the amount of opioid needed to control pain) was reported by Rahman et al. (2021). When dosed together with morphine, the KRM-II-81 analog, MP-III-024, engendered synergistic pain reduction.

Tolerance to the antinociceptive effects of pain medications creates serious health risks both by placing the patient in undue pain and increased risk of medication dependence and lethality (Kalant et al., 1971; Huxtable et al., 2011). No tolerance was observed with KRM-II-81 against acute inflammatory pain after 11 days of dosing; midazolam showed tolerance development (Lewter, 2019). Similar findings were observed in a neuropathic pain preparation with chronic constrictive nerve injury (Lewter, 2019). Over 22 consecutive days of dosing, neither KRM-II-81 or MP-III-080 produced tolerance to their antinociceptive effect against chonic neuropathic pain in mice (Biggerstaff et al., 2020).

KRM-II-81 and analogs produced anxiolytic-like effects in rodents at doses that did not produce sedation or motor-impairment. This was reported in a marble-burying assay in mice (Poe et al., 2016; Knutson et al., 2020) and in a Vogel-conflict study in rats (Poe et al., 2016; Witkin et al., 2017). KRM-II-81 and two structural analogs also produced antidepressant-like effects in the mouse forced swim test although the efficacy was not as great as that of imipramine (Methuku et al., 2018).

On multiple measures of sedation and motor impairment, KRM-II-81 was less impacting than 1,4-benzodiazepines like diazepam (Poe et al., 2016; Lewter et al., 2017; Witkin et al., 2018). Given the reduced motoric impact of KRM-II-81, the structural basis of this reduced side-effect profile was interrogated with structural docking studies where alprazolam was shown to bind more strongly to the α1β3γ2L GABA_AR compared to KRM-II-81 (Witkin et al., 2020). Respiratory depression is another potential medical liability of GABAkines when given with respiratory depressing drugs like opioids. In multiple measures of respiratory function, KRM-II-81 was either inactive or less potent than alprazolam (Witkin et al., 2019a). GABAkines can impair memory and cognitive function (Crowe and Stranks, 2018; Engin et al., 2018). Indeed, some benzodiazepines like midazolam are used as pre-anesthetic sedatives where they convey the useful side effect of surgical memory loss (Kim et al., 2015; Patel and Kurdi, 2015). In one study evaluating potential cognitive impact, midazolam reduced spontaneous alternation in a T-maze from 70 to 33% whereas KRM-II-81 did not significantly reduce this measure of cognition (61%) (Lewter, 2019).

1,4-Benzodiazepine anxiolytics can be abused (Griffiths and Wolf, 1990; Woods et al., 1992) and are legally scheduled as controlled substances for this reason. The U.S. FDA updated its boxed warning for this class of compounds in September 2020 over safety issues associated with their combined use with opioids (Hirschtritt et al., 2021). In two models of abuse liability, intracranial self-administration (Moerke et al., 2019) and drug discrimination (Lewter, 2019), diazepam or midazolam were detected but not KRM-II-81).

D. Early-Stage Compounds

Medicinal chemistry efforts to identify improved structures for GABA_AR ligands is ongoing. This research enterprise has brought new compounds into use as tools to identify biological bases for disease and has led to increased understanding of the structural and molecular underpinnings of key drug-protein interactions. Importantly, these newer compounds stand as potential novel compounds from which the next breakthrough GABAkines will come. An excellent review of newly-patented compounds is available (Crocetti and Guerrini, 2020).

Reports have emerged on a new compound that is a selective GABAkine of α3-containing GABAARs. A study from Rowlett’s lab (Meng et al., 2020) showed that YT-III-31 (8-ethynyl-N-methyl-6-phenyl-4H-benzo[f]imidazo[1,5-a][1,4]diazepine-3-carboxamide) did not increase punished responding of Rhesus monkeys, a finding that is opposite to that of the anxiolytic-like effects found with traditional benzodiazepine anxiolytics and with the α2/3-GABAkine, KRM-II-81 outlined earlier in this review. Instead, YT-III-31 had unique sedative-like effects. These findings are consistent with the idea that α2 but not α3-containing GABA_ARs (for which YT-III-31 is preferring) are pivotal for driving anxiolytic-like effects. Another study from Rowlett’s lab showed that in Rhesus monkeys discriminating ethanol, only triazolam (non-selective), an α5-preferring GABAkine, and an α2/3/5-preferring GABAkine fully reproduced ethanol like discriminative stimulus effects. In monkeys self-administering ethanol orally, an α2/3-(HZ-166) and an α3-preferring (YT-III-31) GABAkine facilitated ethanol but not sucrose drinking (Berro et al., 2021). A study in rats from by the Savić lab showed that YT-III-31 was able to produce anxiolytic-like efficacy in the open field and elevated plus-maze assays at low, non-sedating doses (Batinić et al., 2018). They also did work to help identify the in vivo activity of this new compound. They used brain exposures to estimate in vivo potentiation of specific GABAAR configurations: YT-III-31 at doses that were anxiolytic-like in rats did not potentiate α1βγ2 GABAARs. Their estimation also revealed only a modest selectivity of YT-III-31 for α3γ2 over α2γ2 and α5γ2 sites.

Medicinal chemistry efforts have also brought forth novel GABAkines acting at GABA_BRs (Lobina et al., 2021). Novel compounds for the GABA_AR have also been recently disclosed with both agonist and allosteric modulatory properties (Olander et al., 2018), α5-impacting GABAkines acting at extrasynaptic GABA_ARs (Etherington et al., 2017), a new chemical series of 1,2,3-triazolo-benzodiazepine anticonvulsant compounds (Shafie et al., 2020), new series of 11-dialkylaminomethyl-2,3,4,5-tetrahydrodiazepino[1,2-a]benzimidazoles with potent anxiolytic properties (Maltsev et al., 2021), new analogs of 4,6-diphenylpyrimidin-2-ol with in vivo activity (Khoramjouy et al., 2021), 5-(2-aryloxy-4-nitrophenyl)-4H-1,2,4-triazoles and 5-(2-aryloxy-3-pyridyl)-4H-1,2,4-triazoles, possessing C-3 thio or alkylthio substituents with potent anticonvulsant activity (Navidpour et al., 2021), a new series of anticonvulsants based upon the 3-{2-[1-acetyl-5-(substitutedphenyl)-4,5-dihydropyrazol-3-yl]hydrazinylidene}-1,3-dihydro-2H-indol-2-ones platform (Kerzare et al., 2021), and a novel series of 1,3-dihydro-2H-1,4-benzodiazepin-2-one azomethines and 1,3-dihydro-2H-1,4-benzodiazepin-2-one benzamides was disclosed for which the lead compound was more efficacious than diazepam under their assay conditions and with lead-like properties (Nilkanth et al., 2020).

The design and synthesis of three different series of 1,5-benzodiazepines substituted at the 2 and 4 position were reported with a lead compound having diazepam-like high potency and drug-like properties (Verma et al., 2020). A new series of 4-phenyl-6H-imidazo[1,5-a]thieno[3,2-f][1,4]diazepine-7-carboxylate esters were synthesized by Di Capua et al. (2020) and shown to have in vivo activity in a range of benzodiazepine-detecting assays systems; some compounds were active without notable ancillary side-effects. Pandey et al (2020) reported new analogs of HZ-166. These compounds used 2,4-disubstituted oxazoles and oxazolines as bioisosteric replacement of the ester function with the goal of improving oral bioavailability. Two new structural analogs, LKG-I-70 and KPP-III-5, were shown to be devoid of motor-impairing effects when given up to 100 mg/kg. MIDD0301 (Yocum et al., 2019) was reported by Forkuo et al. (2018) to selectively amplify α_1–3,5β₃γ₂ GABAARs. The compound was orally active with a long half-life. Their data suggest that MIDD0301 represents a new candidate compound that relaxes airway smooth muscle, reduces lung inflammation and mitigates airway hyper-responsiveness in a mouse model of asthma.

Newer methods are also being introduced to help identify molecular substrates of compound binding to and functional activation of GABA_ARs as well as the identification of novel drug therapies. For example, in a recent study by Crocetti and colleagues (2021) 8-methoxypyrazolo[1,5-a]quinazolines were examined with proximity frequencies analyses (quantification of the frequencies that a compound intercepts two or more amino acids in the process of binding). Their work led to the elucidation of a combination of amino acids αVAL203- γTHR142 and αTYR 160- γTYR 58 that could predict GABAkine function. Sansolone et al. (2019) described new photochemical methods to interrogate individual GABA_ARs, a technology that should have important consequences for the design of improved GABAkines. Structural studies on a newer series of pyrazoloquinolinones gave some insights into the structural dynamics of the pharmacophore that might help guide new compound discovery (Iorio et al., 2020). A new potential drug target for diazepam-refractory status epilepticus was announced - proinflammatory cytokine high mobility group box-1 (HMGB1) (Zhao et al., 2020). Along with this, Burman et al. (2019) reported on other potential mechanisms of diazepam resistance. Novel compounds have been used to help identify new binding domains for neuroactive steroids (Yu et al., 2019; Jayakar et al., 2020). Another auxiliary protein associated with GABA_ARs was also recently disclosed; Shisa7 regulates GABA_AR trafficking, function, and pharmacology relevant to the control of GABAkine activity (Han et al., 2020).

A recent review of the Translocator Protein 18 kDa (TSPO), long-postulated to be a binding site for benzodiazepine anxiolytics is now available (Barresi et al., 2021). Alpidem was used as a prototype to create novel structures interacting with the TSPO binding site (Gudasheva et al., 2020). The dipeptide, GD-102 (N-phenylpropionyl-l-tryptophanyl-l-leucine amide), was shown to produce potent anxiolytic-like effects in vivo that were blocked by a TSPO binding site antagonist.

Taken as a whole, it is clear that the search for improved GABAkine medications continue.