Positive GABAA receptor modulating steroids and their antagonists: Implications for clinical treatments

GABA is the main inhibitory neurotransmitter in the brain and GABAergic transmission has been shown to be of importance for regulation of mood, memory and food intake. The progesterone metabolite allopregnanolone (Allo) is a positive GABAA receptor modulating steroid with potent effects. In humans, disorders such as premenstrual dysphoric disorder (PMDD), hepatic encephalopathy and polycystic ovarian syndrome are associated with elevated Allo levels and increased negative mood, disturbed memory and increased food intake in some individuals. This is surprising because Allo shares many properties with benzodiazepines and is mainly considered to be anxiolytic and anti‐depressant. However, it is well established that, in certain individuals, GABAA receptor activating compounds could have paradoxical effects and thus be anxiogenic in low physiological plasma concentrations but anxiolytic at high levels. We have demonstrated that isoallopregnanolone (Isoallo), the 3β‐OH sibling of Allo, functions as a GABAA receptor modulating steroid antagonist (GAMSA) but without any effects of its own on GABAA receptors. The antagonistic effect is noted in most GABAA subtypes investigated in vitro to date. In vivo, Isoallo can inhibit Allo‐induced anaesthesia in rats, as well as sedation or saccadic eye velocity in humans. Isoallo treatment has been studied in women with PMDD. In a first phase II study, Isoallo (Sepranolone; Asarina Pharma) injections significantly ameliorated negative mood in women with PMDD compared with placebo. Several GAMSAs for oral administration have also been developed. The GAMSA, UC1011, can inhibit Allo induced memory disturbances in rats and an oral GAMSA, GR3027, has been shown to restore learning and motor coordination in rats with hepatic encephalopathy. In humans, vigilance, cognition and pathological electroencephalogram were improved in patients with hepatic encephalopathy on treatment with GR3027. In conclusion GAMSAs are a new possible treatment for disorders and symptoms caused by hyperactivity in the GABAA system.


| INTRODUC TI ON
It is common knowledge that altered GABA-ergic function is associated with neurological and psychiatric disorders. Neuroactive steroids, mainly metabolites of stress and sex steroids (eg, deoxycorticosterone and progesterone) are known to be potent positive modulators of the GABA A receptor. However, the existence of negative GABA A receptor modulators and antagonists to positive steroid GABA A receptor modulators is perhaps less well known. Extensive research has shown that positive GABA A receptor modulators such as the neuroactive steroid allopregnanolone (Allo) have pronounced effects in the brain. Some of these effects are beneficial, and allopregnanolone is thus promoted as a therapy for neuropsychiatric disorders, 1 although some behavioural and neurological effects are aversive and disliked by the patients. Below, we discuss the clinical disorders and symptoms that are related to increased neuroactive steroid concentrations and/or increased sensitivity to neuroactive steroids. We will also discuss some experimental animal studies and clinical human studies of the effects of isoallopregnanolone (Isoallo) and an oral compound (GR3027) that are GABA A receptor modulating steroid antagonists (GAMSAs), opening up the possibility that the negative effects of GABA A receptor modulating steroids can be counteracted.

| CLINIC AL S ITUATI ON S WHERE THERE IS A TEMP OR AL REL ATI ON S HIP B E T WEEN THE PRE S EN CE OF NEUROAC TIVE S TEROIDS AND SYMP TOMS
There are several situations where production of neuroactive steroids and plasma concentrations are naturally fluctuating and through which it is possible to detect relationships between symptoms and variations in steroid blood concentrations. For example, both Allo and Isoallo are produced by the corpus luteum of the ovary or secreted from the adrenal cortex during stress and Allo is also synthesised directly in the brain. 2,3 The existence of such relationships is discussed below, although there are limitations to the interpretation since neuroactive steroids produced within the brain can reach effective levels locally without any corresponding variation in peripheral blood. However, we know that an increase in blood concentrations is mirrored in the brain and influences brain activity. 4 Examples of situations with fluctuating levels of neuroactive steroids are during the menstrual cycle, pregnancy and the post-partum period in women, as well as during acute and chronic stress. There are also specific disorders associated with an increased brain concentration of neuroactive steroids (eg, hepatic encephalopathy). 5

| Mood disorders
In premenstrual dysphoric disorder/premenstrual syndrome (PMDD/PMS), the symptoms arrive close to the allopregnanolone rise after ovulation at the beginning of the luteal phase. Symptom severity continues to rise until the end of the menstrual cycle and the start of the full flow of menstrual bleeding ( Figure 1). 6 In the beginning of the next menstrual cycle, the symptoms decline rapidly and usually disappear within 3-4 days of the onset of menstrual bleeding. During the postmenstrual phase, there is a period of wellbeing, closely related to oestrogen production, ending with the oestradiol peak 6 ( Figure 1). This pattern suggests that a symptomprovoking factor is indeed produced by the corpus luteum of the ovary. This is further supported by the finding that, in anovulatory cycles, either spontaneous or induced, Allo is not produced and no symptom cyclicity occurs. tions of progesterone and Allo during the luteal phase. The consensus today is that the plasma concentration does not predict the PMDD diagnosis but, in sensitive patients, the levels appear to have importance. [9][10][11] The sensitivity of the GABA A receptor for Allo has been indicated as pathogenesis in PMDD. This is interesting because patients with PMDD have different sensitivity to Allo compared to controls 12 and serotonin reuptake inhibitors normalise neurosteroid sensitivity in patients with PMDD. 13 In a previous investigation of major depressive disorder, both Allo and Isoallo were measured in 11 patients. 14 The results showed a decrease in Allo concentrations and an increase in Isoallo concentrations in comparison with levels at recovery from major depression. 14 However, in patients with panic disorder, the same group found an increased Allo level and a decreased Isoallo level compared to controls. 15 Ten patients with panic disorder were also investigated during a sodium lactate induced panic attack and Allo and their Isoallo levels were compared with those of a matched control group. A decrease in Allo concentration and an increase in Isoallo concentration were noted during the attack compared to the control group. 16 The same group also investigated panic induction with high doses of cholecystokinin tetrapeptide in healthy volunteers. All volunteers displayed a pronounced panic reaction, although there was no change in either Allo or Isoallo plasma levels. 17 Postpartum depression has been subject to neurosteroid investigations. It is well known that there is a major decrease in

| Menstrual migraine
In women with migraine, approximately 50% report an association between the occurrence of migraine attacks and menstrual cycle phase. Compared to the expected number of migraine attacks, women with menstrual migraine (MM) suffer from significantly more frequent attacks during the late luteal and peri-menstrual phase than during the follicular phase. 21,22 It is hypothesised that this increased incidence of attacks may be related to the withdrawal of the effects of Allo on the GABA A receptor. However, the data do not support the existence of a comorbidity between PMDD and menstrual migraine. In PMDD, a relief in symptoms during menstruation is always observed, whereas MM has its peak symptoms during menstruation. 23

| Catamenial epilepsy
It has long been recognised that seizure frequency can vary with the menstrual cycle in women with epilepsy and with the oestrus cycle in animal models. 24 where there is a seizure-exacerbation during the menstrual phase in a subset of women. 24,[26][27][28] In addition, a more recent prospective multicentre study of women with partial epilepsy showed a twofold greater seizure occurrence on the first day of menstrual flow compared to the mid luteal days. 29 Overall, 39% of the women fulfilled criteria for catamenial epilepsy. 30 In ovulatory cycles, the predominant pattern was an increase of seizures during menstruation closely following the rapid decrease in plasma levels of Allo. The exacerbation of seizure frequency during menstruation could thus be a withdrawal effect similar to that seen when an antiepileptic drug is suddenly discontinued. In experimental studies, progesterone increases the electroshock seizure threshold in animals, protects against metrazol-induced seizures and decreases the frequency of interictal spikes from an epileptic focus. 31,32 Moreover, luteal phase serum concentrations of progesterone reduced inter-ictal spike frequency in four out of seven women with partial epilepsy, and the effect was dependent on the plasma progesterone binding. In this experimental study, women with the highest plasma binding rate did not respond. 33 The antiepileptic effect of progesterone is suggested to be mediated by its metabolite Allo. 34

| Body movement disorders
Postural control could vary during the menstrual cycle, measured as the displacement area in a balance testing situation. Women with PMS had a significantly greater displacement area during the luteal phase compared to women without cyclical changes in one such study. 35 The incidence of sports-related injuries is one clinical aspect that seems to be related to the luteal phase of the menstrual cycle. 36 In other body movement disorders (eg, essential tremor), extra synaptic GABA A receptors have been implicated as regulators of the oscillatory activity. In patients with essential tremor, physical or mental stress exacerbates the tremor; for example, in acute stress situations, such as when giving an oral presentation. 37 During stress, the concentration of Allo in the brain and plasma increases to a concentration that has a physiological impact on GABA A receptor activity. 2,38 The results from a neuroimaging study indicated that there is an increased sensitivity to GABAergic stimulus in the cerebellarthalamocortical pathway generating the tremor. 39 Tourette's syndrome is a disorder presenting as different kinds of motor and phonic manifestations called tics, with the symptoms increasing during stress. 40 The influence of neuroactive steroids on the frequency of tics has been studied in both humans and animal models. Inhibition of the 5α-reductase enzyme reduces the number of tics.
5α-Reductase plays a central role in the enzymatic pathway from progesterone to Allo. In addition, Allo administration increases tic frequency. 41

| Cognition and dementia
Here, we discuss the effects of acute and chronic stress on learning and memory in humans and rodents, as well as the effects in rodent dementia models and the risk of developing dementia in humans.
Allo and tetrahydrodeoxycorticosterone (THDOC) are both positive GABA A receptor modulating steroids and they increase during stress. 2,38 An acute short exposure of Allo, such as in an acute stress situation, is known to decrease neural activity in the hippocampus of rats. 42 Allo inhibits learning in rats tested in the Morris water maze (MWM) [43][44][45] and inhibits episodic memory in humans. 46 Episodic memory is a type of memory that is disturbed early in patients diagnosed with Alzheimer's disease (AD). 47 Allo also inhibits longterm potentiation and cholinergic action in rat hippocampus. 48,49 Functional magnetic resonance imaging tudies in women show that progesterone and Allo impair memory by reducing the activity in brain regions related to memory formation and retrieval. 50 Other In wild-type mice, chronic Allo treatment at low-stress levels, for 5 months, produces a permanent deterioration in learning and memory especially in females. 73 In this experimental study, MWM  61 The increase in dementia frequency was not considered to be the result of ischaemic events but rather to a biological/hormonal factor, probably MPA itself. 74 Oestrogen alone did not increase the dementia risk in the same study. 75  can induce anaesthesia when given to rats. 76

| Food consumption and weight increase
Obesity is one of the major causes of poor health and the most important factor in development of obesity is the consumption of more calories than needed. 77 Still, the mechanism behind the many endogenous driving forces that contribute to over-eating is not well understood. However, GABAergic transmission has been shown to be of importance for regulation of food intake. 78,79 In humans, situations with elevated Allo levels are associated with increased food intake, eating disorders, overweight and obesity. 80 The importance of GABA A receptors in the regulation of food intake was investigated in an animal study where the GABA A receptor antagonist bicuculline inhibited food intake in hungry, food-deprived rats. 81 The brain areas of particular interest in this perspective are the hypothalamic arcuate nucleus and paraventricular nucleus. 78 Studies of benzodiazepine-induced food intake indicate that the most abundant α-subunits of the GABA A receptor in the arcuate nucleus and paraventricular nucleus are α3 and α2. 82 Allo in concentrations as low as 2 nm significantly increased the spontaneous postsynaptic current in single dissociated cells in both the arcuate nucleus and paraventricular nucleus. 83 Animal studies have shown that Allo increased meal size in a dose-dependent manner 84 and that calorie dense food was preferred compared to less palatable food. 85 Moreover, Allo compared to placebo increased body weight significantly in rats after five treatment days, and weight gain was highly correlated with increased food intake. 86 In humans, there is a correlation between serum Allo concentration and uncontrolled eating in obese women with polycystic ovarian syndrome. 87 Overweight and obese women have a different sensitivity to Allo compared to normal weight women, indicating that obesity changes the sensitivity to Allo. 88 Excessive weight-gain during pregnancy is related to high Allo levels compared to women with low weight-gain. 89 In Prader-Willis syndrome, hyperphagia and obesity are key symptoms caused by a total lack of satiety sensations. 90 These patients have a deletion in chromosome 15 that contains genes coding for one GABA A receptor subtype. 90 Prader-Willis syndrome patients have a compensatory increased expression of α4 (12-fold), γ2 (5fold), and α1 and α3 (>1.5-fold), as well as a higher density of GABA A receptors with the subunits α4, βx and δ visualised in neuroimaging studies. 91,92 Also, the α4, βx, δ receptor subtype is known to be hypersensitive to Allo. 93

| Rational for treating positive GABA A receptor modulating steroid dependent disorders with GABA A modulating steroid antagonist (GAMSA)
A plausible hypothesis is that the symptoms and disorders discussed above are related to increased Allo concentrations in the brain or withdrawal effects after exposure to high Allo levels, with both mechanisms mediated via the GABA A receptor. In addition, the sensitivity to Allo exposure varies between individuals and may be increased by the up-regulation of more sensitive GABA A receptor subtypes. In this context, a modulation of the specific Allo effect on the GABA A receptor by a neurosteroid with antagonistic properties (ie, a GAMSA) would be beneficial, decreasing the negative symptoms.

| In vitro studies
It is well known that the progesterone metabolites Allo (3αhydroxy-5α-pregnan-20-one) and its isomer pregnanolone (Preg, 3α-hydroxy-5β-pregnan-20-one) enhance GABA action via their own binding sites on the GABA A receptor. 94 Their 3β-isomers, Isoallo We have used patch-clamp recordings of spontaneous inhibitory postsynaptic currents (sIPSCs) to study different 3β-steroids in dissociated neuronal cells from rat hypothalamus. We tested the 3β-steroids together with Allo or singly without Allo for their ability to modulate the spontaneous GABA-mediated chloride flux.
The Allo-enhanced GABA response was reduced by the tested 3βsteroids. Figure 2  The other three steroids tested, including Isoallo, had little or no effect on GABA induced receptor opening. These results suggest that some 3β-steroids may be used as therapeutics and act by counteracting the negative effects of Allo. 103,104

| In vivo animal studies
The interaction of Isoallo with Allo induced anaesthesia was studied in male rats using burst suppression of the electroencephalog- As shown above, Allo could impair learning and memory in animals. 43 In another study, a new GAMSA (3β-20β-dihydroxy-5αpregnane) was evaluated as a blocker of Allo-induced effects on memory and learning of rats in vivo and of chloride ion uptake into hippocampal microsacs in vitro. The time taken by the rats to find the hidden platform in the MWM was significantly reduced in the group injected with Allo + 3β-20β-dihydroxy-5α-pregnane compared to the Allo-injected group. The rats injected with 3β-20β-dihydroxy-5α-pregnane located the platform as fast as the placebo treated controls. In line with these findings, chloride ion uptake to GABA-ergic neurones increased with Allo + GABA but was hindered by 3β-20βdihydroxy-5α-pregnane. 107

| Human studies
As noted above, Allo exerts an anesthetic and sedative/hypnotic effect through potentiation of GABA at the GABA A receptor. 105,111 The In the first clinical trial, Isoallo was used in a s.c. preparation called Sepranolone (Asarina Pharma). The disorder treated was PMDD. Sepranolone was given as a luteal phase treatment in a randomised, parallel-group, placebo-controlled study design. The results obtained showed that Sepranolone reduced PMDD symptoms significantly better than placebo based on the per protocol analysis.
The effect was better in the population treated as intended compared to those starting and ending the treatment too early, which suggests that Sepranolone needs to be present in appropriate serum levels during the late luteal phase to exert the best possible effect.
It was concluded that the results constituted an initial indication that Sepranolone could reduce symptom severity and impairment in PMDD more efficiently than placebo. 113 Golexanolone (GR3027) (Umecrine Cognition) is a novel oral GAMSA developed specifically for the treatment of cognitive and vigilance impairment as a result of allosteric over-activation of GABA A receptors. From animal studies, the compound GR3027 has been shown to counteract the effects of Allo and has therefore been considered as a possible future pharmaceutical product for humans suffering from Allo-related over-activation of GABA A receptors. The GR3027-mediated antagonistic effect on the Allo-mediated reduction of SEV and increased self-rated sedation have been studied in a double-blind, placebo-controlled, cross-over study. GR3027 inhibited the Allo-induced decrease in SEV and Allo-induced sedation.
Thus, the oral compound GR3027 could mitigate the inhibition of brain function induced by Allo at well tolerated doses. 114 In a recent study, the effects of Golexanolone on the electro-

| CON CLUS IONS
Hyperactivity or an increased GABAergic tone as a result of either exposure to positive allosteric GABA A receptor modulating steroids, or increased sensitivity to Allo, is related to several symptoms and disorders. Treatment with GABA A modulating steroid antagonists appears to mitigate the severity of the symptoms and disorders.

ACK N OWLED G EM ENTS
Some of the studies included are supported by an EU-Grant, H2020 project number 721802, to Umecrine AB via the consortium SYNDEGEN with a PhD position for RD, Umecrine AB, Umecrine Cognition AB, Asarina Pharma AB Umeå University foundations and Swedish research council project number 4X-11198.

CO N FLI C T O F I NTE R E S T S
TB has shares in Umecrine AB, Umecrine Cognition AB and Asarina Pharma AB. RD is employed by Umecrine AB. F I G U R E 3 Change in the same subjects in saccadic eye velocity (SEV), self-rated sedation, Allo (AP) and Isoallo (ISO) serum levels following i.v. administration of Allo alone (white squares; n = 12), Allo:Isoallo low (light grey squares; n = 12) or Allo:Isoallo high (dark grey squares; n = 11), respectively. Data are given as the mean ± SEM. Both doses of AP + ISO had a significantly smaller reduction in SEV compared to AP alone. Allo:Isoallo low, but not Allo:Isoallo high, had a significant reduction in sedation compared to Allo alone. Reprinted with permission from Bengtsson et al 99

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/jne.13013.