GHB for cataplexy: Possible mode of action

Introduction

The sleep disorder narcolepsy is caused by the loss of orexinergic (hypocretinergic) neurones in the lateral hypothalamus (Sakurai, 2013), probably due to an autoimmune process (Mahlios et al., 2013). The prevalence of narcolepsy has been estimated to be between 0.03–0.16% in the general population (Nishino, 2007); in the USA one out of 2000 people may suffer from the disorder (Mignot, 1998). It has complex symptomatology that reflects the intrusion of both slow wave sleep (excessive daytime sleepiness, sleep attacks) and rapid eye-movement sleep (cataplexy, hypnagogic hallucinations, sleep paralysis) into wakefulness (Ohno and Sakurai, 2008). Cataplexy is an especially troublesome symptom: it consists of the sudden loss of muscle tone in response to strong emotions. Narcolepsy, however, can also occur without cataplexy (Andlauer et al., 2012; Oka et al., 2006). Narcolepsy patients with cataplexy have reduced orexin concentrations in the cerebrospinal fluid (CSF) (Ohno and Sakurai, 2008; Scammell and Winrow, 2011). On the other hand, only a subgroup of patients without cataplexy have reduced CSF orexin levels (Oka et al., 2006); it has been proposed that these patients may develop cataplexy at a later stage (Andlauer et al., 2012).

The symptoms of ‘idiopathic’ narcolepsy and cataplexy may be mimicked by brain lesions affecting the neuronal circuitry involved (‘symptomatic’ or ‘secondary’ narcolepsy and cataplexy) (Nishino and Kanbayashi, 2005). It has been shown that hypothalamic lesions can lead to narcolepsy, with or without cataplexy, and some of the patients affected have reduced CSF orexin levels (Kanbayashi et al., 2011; Nishino and Kanbayashi, 2005; Scammell et al., 2001). However, brainstem lesions, not involving the hypothalamus and sparing the orexin system, can also lead to narcolepsy (Mathis et al., 2007) or isolated cataplexy (D’Cruz et al., 1994; Reynolds and Roy, 2011). These clinical reports indicate that the consequences of orexin deficiency can be mimicked by lesions of structures downstream from them, and illustrate the importance of the brainstem in the regulation of sleep, arousal and muscle tone (Burgess and Scammell, 2012; Luppi et al., 2011). Furthermore, they are consistent with recent experimental evidence of the role of the pontine locus coeruleus (LC) in mediating the effects of hypothalamic orexin neurones on arousal (Carter et al., 2013).

While excessive daytime sleepiness can be treated with psychostimulants (amphetamine, modafinil), cataplexy does not respond to these drugs. Two other classes of drug, antidepressants (especially selective serotonin reuptake inhibitors, SSRIs) and sodium oxybate (γ-hyroxybutyric acid (GHB)) have shown effectiveness in reducing the frequency of cataplectic attacks (Houghton et al., 2004; Zeitzer et al., 2006).

The clinical effectiveness of sodium oxybate in narcolepsy is well established (Alshaikh et al., 2012; Boscolo-Berto et al., 2012; Mayer, 2012; Robinson and Keating, 2007). It is effective in alleviating all three core symptoms (excessive daytime sleepiness, fragmented night-time sleep and cataplexy) of narcolepsy. It is absorbed fast and has a short elimination half-life (Schep et al., 2012), necessitating twice-nightly administration (Robinson and Keating, 2007). While its effects on nocturnal sleep are observable following acute administration (Plazzi et al. 2014), a reduction in the frequency of cataplectic attacks becomes detectable only after several weeks (Huang and Guilleminault, 2009; US Xyrem Multicenter Study Group, 2004). This suggests that the anticataplectic effect may reflect some adaptive changes. In the clinically recommended dosage range, sodium oxybate is well tolerated (Alshaikh et al., 2012; Lecendreux et al., 2012), and has a very low abuse potential (Wang et al., 2009).

It is not known how GHB may exert its therapeutic effect in cataplexy.

GHB

GHB is an analogue of γ-aminobutyric acid (GABA), the natural inhibitory neurotransmitter; for clinical use it is marketed as ‘sodium oxybate’. GHB exerts sedative effects by interacting with specific inhibitory GHB and GABA_B receptors (Crunelli et al., 2006). The interaction of GHB with GABA_B receptors is well established. By stimulating these receptors, GHB inhibits central cholinergic and serotonergic neurones (Kohlmeier et al., 2013), and inhibits dopamine release (Madden and Johnson, 1998), leading to catalepsy (Koek and France, 2008). It is controversial to what extent the two receptors mediate the anticataplectic effect of GHB. A clinical comparison of sodium oxybate and baclofen, a GABA_B receptor agonist (Bowery, 2006), in adolescent patients with narcolepsy/cataplexy, has shown that only GHB was effective in reducing cataplexy and improve daytime sleepiness, while baclofen was ineffective (Huang and Guilleminault, 2009). On the other hand, in a recent study, baclofen proved to be more effective than GHB in alleviating murine cataplexy (Black et al., 2014).

Recently it has been proposed that the GHB receptor is in fact a subtype of the GABA_A receptor (Absalom et al., 2012; Bay et al., 2014; Enna, 2012). However, the importance of the interaction of GHB with GABA_A receptors has been questioned (Connelly et al., 2013a). GABA_A receptors, and probably also GABA_B receptors, occur not only subsynaptically in close association with the GABAergic nerve terminal, but also at some distance from it (‘extrasynaptic receptors’), where they mediate a more diffuse slow (‘tonic’) inhibitory effect (Brickley and Mody, 2012).

LC and control of muscle tone

The sudden loss of muscle tone in cataplexy is due, at least partly, to the switching off of the activity of the LC, a major central noradrenergic nucleus. The LC sends a direct excitatory output to spinal motoneurones (Szabadi, 2013), and plays an important role in the maintenance of muscle tone (McGregor and Siegel, 2010). During attacks of cataplexy there is cessation of LC activity (Wu et al., 1999), leading to total atonia (Peever, 2011). It is likely that emotions trigger cataplexy via the activation of the amygdala (Dauvilliers et al., 2014), a limbic nucleus known to play a central role in processing emotions (Phelps and LeDoux, 2005). The central nucleus of the amygdala projects to the LC (Szabadi, 2013), mediating an excitatory influence on LC neurones (Bouret et al, 2003). Furthermore, emotional memories evoke parallel activation of the amygdala and the LC (Sterpenich et al., 2006). The involvement of the amygdala in cataplexy is highlighted by the observation that its experimental lesion leads to the attenuation of cataplexy (Burgess et al., 2013). The LC also receives a strong excitatory orexinergic input from the lateral hypothalamus (Szabadi, 2013); this input becomes deficient in narcolepsy (Ohno and Sakurai, 2008). The loss of the orexinergic input to the LC may be responsible for the paradoxical effect of emotional stimuli in narcolepsy: LC neurones, rather than getting facilitated, become deactivated (see below).

Autoregulation of LC neurones

LC activity is regulated by inhibitory α₂-adrenoceptors (autoreceptors) on the noradrenergic neurones. Stimulation of inhibitory somatodendritic α₂-adrenoceptors by noradrenaline, released from somatodendritic sites and recurrent axon collaterals, dampens the firing of the neurone, providing a mechanism for autoregulation (Andrade and Aghajanian, 1984; Huang et al., 2012). As the LC neurone is stimulated, firing rate increases as a function of stimulus strength (frequency, number of pulses in a train), and when firing rate reaches 15–20 Hz, the autoinhibition mechanism ‘kicks in’ (Huang et al., 2012). Further stimulation leads to inhibition of the neurone, the degree of which is related to the strength of stimulation (Marzo et al., 2014). At very high frequency stimulation the LC may completely ‘switch off’. This may explain the observation that while the activation of the LC by optogenetic photostimulation at moderate frequencies leads to increases in arousal and locomotion, high frequency stimulation causes behavioural arrest (Carter et al., 2010). The relationship between stimulus strength and firing rate can be described by an inverted U curve (Figure 1). At lower levels of stimulus strength the firing rate increases linearly with stimulus strength. However, after autoinhibition has been triggered, further stimulation results in a progressive decrease in firing rate. The curve can be defined by the following parameters: maximum firing rate attained (‘critical firing rate’); stimulus strength above which further stimulation leads to inhibition (‘reversal threshold’); stimulus strength at which firing ceases (‘cessation threshold’).

OPEN IN VIEWER

Figure 1.

Hypothetical relationship between input stimulus strength and firing rate of a locus coeruleus (LC) neurone. Top curve: normal situation; bottom curves: cataplexy (solid line: untreated; broken line: treated). Abscissa: stimulus strength (arbitrary units); letters indicate reversal thresholds (a: normal situation; b: cataplexy; c: cataplexy treated with γ-hydroxybutyric acid (GHB) or selective serotonin reuptake inhibitor (SSRI)); numbers indicate cessation thresholds (1: normal situation; 2: cataplexy; 3: cataplexy treated with GHB or SSRI). Ordinate: firing rate (Hz); 1 and 2: baseline firing rate (1: normal situation and cataplexy; 2: cataplexy treated with GHB or SSRI); 3 and 4: critical firing rate (3: normal situation; 4: cataplexy, untreated and treated). See text for details. ES: emotional stimulus.

Proliferation of autoreceptors in narcolepsy

It has been reported that α₂-adrenoceptors on LC neurones proliferate in narcolepsy (Fruhstorfer, 1989). This may reflect ‘heterologous denervation supersensitivity’ (Tassin et al., 1987), related to the loss of the orexinergic input to the LC (Figure 2). The increase in autoreceptor number could trigger the auto-inhibition mechanism at a lower firing rate, resulting in the lowering of the critical firing rate, and the reversal and cessation thresholds. Thus an emotional stimulus, that normally would cause little change in the firing of LC neurones, may switch them off completely (Figure 1), leading to cataplexy. Interestingly, it has been reported that α₂-adrenoceptor antagonists, such as yohimbine, suppress cataplexy, whereas some α₂-adrenoceptor agonists aggravate it (Nishino et al., 1990), consistent with the involvement of the auto-inhibition mechanism in the generation of cataplexy.

OPEN IN VIEWER

Figure 2.

Control of muscle tone by the locus coeruleus (LC) in the normal situation and cataplexy, both before and after treatment. Nuclei: LC, amygdala (A), lateral hypothalamic (LH) area; motoneurones (MNs) in: interneurones, raphe nucleus (R). Connections are indicated by arrows. All connections are excitatory, with the exception of two inhibitory inputs to LC (recurrent noradrenergic collaterals and input from γ-aminobutyric acid (GABA)-ergic interneurons). Neurotransmitters: noradrenaline (NA); corticotrophin releasing factor (CRF); orexin (Ox); GABA; 5-hydroxytryptamine (serotonin (5-HT)). Receptors: α₁ and α₂: adrenoceptors. Cataplexy: the proliferation of somatodendritic inhibitory α₂-adrenoceptors, resulting from the deficiency of the orexinergic input, leads to enhanced autoinhibition of LC neurones: this leads to weakened noradrenergic stimulation of spinal motoneurones, resulting in loss of muscle tone. Treatment of cataplexy: (1) γ-hydroxybutyric acid (GHB) enhances GABAergic inhibition of the LC by stimulating subsynaptic and extrasynaptic GABA receptors on LC neurones; (2) antidepressants (SSRIs) increase the GABAergic inhibition of the LC by potentiating the serotonergic facilitation of GABAergic interneurones. See text for details.

Some antidepressants, such as mirtazapine (Anttila and Leinonen, 2001) and trazodone (Cusack et al., 1994), although they also interact with a number of other neuroreceptors, act as antagonists of α₂-adrenoceptors, suggesting that they may possess anticataplectic effects. According to one case report, trazodone was effective in treating narcolepsy/cataplexy (Sandyck, 1985). The relationship between the antagonism of α₂-adrenoceptors and the activation of the LC is highlighted by the effect of yohimbine as a sexual enhancer, an effect attributed to LC activation (Corazza et al., 2014).

Interestingly, it has been reported that not only α₂-adrenoceptors, but also muscarinic M2 cholinoceptors proliferate in the brains of narcoleptic dogs, and this change has been implicated in the generation of cataplexy (Nishino and Mignot, 1997). However, no change in central muscarinic receptors could be detected in the brains of narcoleptic patients, using positron emission tomography (PET) imaging (Sudo et al., 1998). Furthermore, pharmacological attempts, using selective cholinoceptor agonists and antagonists, to demonstrate the involvement of a cholinergic mechanism in human narcolepsy/cataplexy, have been unsuccessful (Nishino and Mignot, 1997).

GABAergic regulation of LC activity and the action of GHB

The LC also receives an inhibitory input from GABAergic interneurons (Gobert et al., 2000) (Figure 2). Both GABA_A and GABA_B receptors are present in the LC (Olpe et al., 1988). Some of the GABA_A receptors may be extrasynaptic (Delgado-Lezama et al., 2013); their activity may be potentiated by the stimulation of GABA_B receptors (Connelly et al., 2013b). The GABAergic interneurons may tonically dampen LC activity by stimulating extrasynaptic GABA receptors. GHB may enhance the GABAergic tonic inhibition of LC neurones via its agonistic action at GABA_A and GABA_B receptors, leading to lowering of the baseline activity of LC neurones. Indeed, it has been reported that sustained treatment of rats with GHB lowers the spontaneous firing rate of LC neurones (Szabo et al., 2004).

As illustrated in Figure 1, lowering of the baseline firing rate of LC neurones would change the relationship between the strength of the input stimulus and the increase in firing rate evoked by it, reflected in increases in the reversal and cessation thresholds. Thus an emotional stimulus reaching the LC via the amygdala might not lead to complete inhibition of the firing of LC neurones, and the cataplectic attack may be averted. Interestingly, lowering of the baseline firing rate of LC neurones by GHB is not expected to affect the autoinhibition mechanism per se, indicated by the lack of change in critical firing rate. On the other hand, the anticataplectic effect of the α₂-adrenoceptor antagonist yohimbine (Nishino et al., 1990) is likely to be mediated via the curbing of autoinhibition.

It may appear paradoxical that a reduction in LC activity can have both cataplectic (as provoked by the emotional stimulus) and anti-cataplectic (as brought about by GHB treatment) effects. However, the emotional stimulus may suddenly switch off LC activity completely (‘phasic effect’; see Bouret et al., 2003; Wu et al., 1999), whereas GHB treatment is likely to cause a more moderate ‘tonic’ dampening of baseline neuronal activity (Szabo et al., 2004), allowing fundamental LC functions, such as the maintenance of muscle tone, to continue.

It may be important to administer GHB in a sustained fashion orally to obtain its anticataplectic effect (Robinson and Keating, 2007), since it has been reported that acute intravenous administration of GHB to narcoleptic patients may provoke cataplexy (Price et al., 1981). Significantly, provocation of cataplectic attacks is not a complication of the treatment of cataplexy with sodium oxybate (European Medicines Agency, 2014). It is likely that while sustained administration of GHB in moderate doses tones down the firing of LC neurones (Szabo et al., 2004), acute intravenous administration may lead to complete cessation of LC activity, as seen in narcoleptic attacks evoked by strong emotions (Wu et al., 1999).

As GHB is an orthosteric GABA_A and GABA_B receptor agonist, its action is independent of allosteric sites on the receptor. This is important since GABA_A receptors on LC neurones are insensitive to allosteric modulation by benzodiazepines (Chen et al., 1999), although they contain the γ2 subunit required for benzodiazepine sensitivity, and can bind benzodiazepines (Hellsten et al., 2010). Therefore benzodiazepines would not be good candidates for the treatment of cataplexy. On the other hand, it would be of interest to see whether gaboxadol, a selective orthosteric extrasynaptic GABA_A receptor agonist (Brickley and Mody, 2012), has any therapeutic efficacy in cataplexy. Unfortunately, this remains only a theoretical possibility, since the clinical development of gaboxadol, a promising candidate for the treatment of insomnia, has been halted. However, there are a number of other orthosteric GABA_A receptor agonists in development (Brickley and Mody, 2012).

The GABAergic interneurones are facilitated by a serotonergic input from the dorsal raphe nucleus (Szabadi, 2013) interacting with excitatory 5HT_2C receptors. Stimulation of these receptors, via the activation of the GABAergic interneurones, leads to tonic inhibition of LC activity (Gobert et al., 2000). The facilitation of the GABAergic inhibition of the LC by serotonergic neurones may be enhanced by the blockade of serotonin uptake by antidepressants, especially SSRIs: this may underlie the anticataplectic efficacy of these drugs (Houghton et al., 2004; Zeitzer et al., 2006). Although GHB may attenuate the indirect GABAergic inhibition of the LC by inhibiting serotonergic neurones (Kohlmeier et al., 2013), this effect is likely to be superseded by GHB’s direct action at GABA receptors on and in the vicinity of the LC.

Conclusion

In conclusion, the loss of the orexinergic input to the LC in narcolepsy may result in sensitization of the auto-inhibition mechanism, leading to lowering of the stimulus intensity threshold required to switch off LC activity. GHB, and indirectly SSRIs, by tonically inhibiting LC activity, may increase the threshold and thus prevent the cessation of LC firing evoked by an emotional stimulus.