Obstructive sleep apnea (OSA), characterized by a cessation of breathing produced by a narrowed or collapsed upper airway, represents a significant public health risk [1, 2]. Current OSA prevalence estimates indicate that 14 % of American men and 5 % of American women suffer from OSA, and that those estimates are rising [1]. More importantly, there are strong associations between OSA and other diseases, such as type 2 diabetes, hypertension, stroke, and coronary heart disease [3–5]. The “gold standard” of OSA treatment is to splint open the upper airway via continuous positive air pressure; however, treatment tolerance is low and patients do not adhere to treatment despite obvious health benefits [2]. Safe and effective pharmacological treatments for OSA remain to be identified, and such efforts have been limited by incomplete knowledge of the central and peripheral neural mechanisms controlling respiration during sleep [2, 6]. Recently, activating the inhibitory G_i/o-associated cannabinoid (CB) receptors have been proposed as novel pharmacological intervention to treat OSA [7, 8].

The endocannabinoid system, associated with cannabinoid type 1 (CB₁) and cannabinoid type 2 (CB₂) receptors located on peripheral nerves and central neurons, can be targeted therapeutically to modify disease states [9, 10]. Dronabinol, a synthetic version of Δ9-THC, is a FDA-approved CB₁ and CB₂ receptor agonist used to suppress chemotherapy-induced nausea and stimulate appetite in AIDS patients [11]. Dronabinol, when administered to patients with OSA [12], or to rats chronically-instrumented to measure respiration during sleep [13], decreased apneas. The mechanism of dronabinol’s effect in decreasing apnea propensity appeared to be, in part, due to the activation of both CB₁ and CB₂ receptors located on nodose ganglia of the vagus nerves [14, 15], which transmit vital information from the lungs to the brainstem, contributing to reflex responses regulating: tidal volume, respiratory frequency, augmented breaths and bronchoconstriction [16, 17]. In a well-established model of vagally-mediated reflex apnea [18], dronabinol injected into nodose ganglia attenuated apneas [15]. Antagonism of CB₁, CB₂, or both reversed dronabinol’s attenuation of apneas [14]. More importantly, dronabinol also increased phasic upper airway activity via activation of CB receptors at the nodose ganglia [14, 15]. Though recent evidence confirms the role the vagus nerve plays in apnea propensity [14, 15, 19–21], less is known about the role that central CB receptors have in apnea induction or suppression. Dronabinol is highly lipophilic and readily crosses the blood-brain barrier into the central nervous system [22], where CB receptors are widely distributed [23, 24] [10], including brain areas vital to respiratory control [23, 25–28]. Complicating the issue further is that activating central CB receptors can inhibit evoked release of excitatory or inhibitory neurotransmitters [29], thereby inhibiting or disinhibiting neuronal activity [9, 30]. It is unknown if central versus peripheral administration of a CB agonist would have similar or dissimilar effects on respiration. Therefore, it is important to understand dronabinol’s global effects on the central nervous system without activating CB receptors in the peripheral nervous system.

Here, we hypothesized that global central administration via intracerebroventricular (ICV) injection of dronabinol would attenuate reflex apneas and increase upper airway activity.

Reflex apneas induced via IV infusion of 5-HT were conducted in rats before and after ICV injections of various concentrations of dronabinol (Fig. 1). Also, breath duration, coefficient of breath durations, and phasic and tonic genioglossus activity were quantified before and after ICV injections of various concentrations of dronabinol (Figs. 2, 3, and 4, respectively).Fig. 1

Apnea duration quantified from acute 5-HT-induced apnea experiments before (baseline; N = 30) and after ICV injections of various concentrations of dronabinol (100, 10, 1 or 0.1 μg; N = 6 for each dose) or vehicle (DMSO; N = 6). ICV injections of dronabinol at any concentration did not significantly (p = 0.19) attenuate reflex apneas. Data (mean ± SEM) were analyzed using mixed model analysis with a repeated/fixed measure (ICV treatment)

At baseline (before ICV injections), IV 5-HT infusion produced reflex apneas lasting 7.29 ± 0.75 seconds (N = 30). ICV injections of 100 μg (11.80 ± 2.66 seconds, N = 6), 10 μg (7.73 ± 0.85 seconds, N = 6), 1 μg (7.48 ± 0.87 seconds, N = 6), or 0.1 μg (9.67 ± 1.46 seconds, N = 6) of dronabinol, or injection of vehicle (DMSO; 7.79 ± 0.93 seconds, N = 6) did not significantly (F_{5, 35.6} = 1.90, p = 0.12 for main effect of “treatment”) alter apnea durations compared to baseline (Fig. 1).

Breath durations were averaged from 5 breaths prior to 5-HT-induced apneas. Average breath duration before ICV injections was 0.89 ± 0.03 seconds (N = 30), and was not significantly (F_{5, 32.7} = 1.89, p = 0.12) altered by 100 μg (0.84 ± 0.06 seconds, N = 6), 10 μg (0.84 ± 0.08 seconds, N = 6), 1 μg (0.75 ± 0.03 seconds, N = 6), or 0.1 μg (0.86 ± 0.06 seconds, N = 6) ICV injection of dronabinol, or ICV injection of vehicle (DMSO; 0.99 ± 0.11 seconds, N = 6; Fig. 2). Coefficient of variation of 30 breath durations before (pre) and after (post) 5-HT infusion was quantified as a measurement of respiratory instability (Fig. 3) [35]. There was no ICV treatment main effect (F_{5, 11.0} = 1.03, p = 0.45) or interaction between ICV treatment and time (F_{5, 11.0} = 1.01, p = 0.46). There was a main effect of time (F_{1, 14.5} = 87.7, p < 0.01); breathing was more unstable following 5-HT infusion compared to before infusion (p < 0.01; Fig. 3).

Genioglossus activity was measured at the start of inspiration (Phasic EMGgg) and at the end of expiration (tonic EMGgg) prior to reflex apneas (Fig. 4). ICV Injections of 100 μg (0.89 ± 0.28 a.u., N = 6), 10 μg (1.17 ± 0.41 a.u., N = 6), 1 μg (0.59 ± 0.12 a.u., N = 6), or 0.1 μg (0.93 ± 0.24 a.u., N = 6) of dronabinol, or injection of vehicle (DMSO; 0.63 ± 0.11 a.u., N = 6) did not significantly (F_{4, 9.50} = 0.94, p = 0.48 for main effect of “treatment”) alter phasic EMGgg (Fig. 4a). Similarly, ICV injections of 100 μg (0.97 ± 0.13 a.u., N = 6), 10 μg (1.46 ± 0.20 a.u., N = 6), 1 μg (1.33 ± 0.45 a.u., N = 6), or 0.1 μg (1.23 ± 0.29 a.u., N = 6) of dronabinol, or injection of vehicle (DMSO; 1.35 ± 0.46 a.u., N = 6) did not significantly (F_{4, 7.96} = 1.22, p = 0.37 for main effect of “treatment”) alter tonic EMGgg (Fig. 4b).Fig. 4

Phasic (a) and tonic (b) genioglossus electrogram amplitude (arbitrary units, a.u.; dotted line corresponds to baseline EMGgg) quantified after ICV injections of various concentrations of dronabinol (100, 10, 1 or 0.1 μg; N = 6 for each dose) or vehicle (DMSO; N = 6). There were no significant differences in phasic (p = 0.48) or tonic (p = 0.37) EMGgg in the treatment groups. Data (mean ± SEM) were analyzed using mixed model analysis with a repeated/fixed measure (ICV treatment)

With the increasing prevalence [1], comorbidity with other diseases [3–5], and a lack of tolerable and effective treatment options [2, 6], OSA is a significant health problem. An impediment to effective treatments of OSA is insufficient knowledge of peripheral and central neural mechanisms of respiratory control, especially during sleep. Though we have shown previously that modulation of vagal afferents via activation of CB receptors located on nodose ganglion neurons can attenuate reflex apneas [14, 15], it was uncertain what role central CB receptors play in respiratory pattern control. Here we show that ICV injection of dronabinol, a non-specific agonist of CB₁ and CB₂ receptors, does not attenuate peripherally-induced reflex apneas.

Vagal afferent neurons, which relay important information about respiratory drive and upper airway muscle tone [16, 17, 36], provide input to the excitatory or inhibitory neurons located in nucleus of the solitary tract (NTS) [37, 38]. The NTS projects to the respiratory centers of the brainstem, including the hypoglossal nucleus (XII) and the rostral ventrolateral medulla (RVLM) [36, 39]. The NTS contains a variety of neuronal populations that fire at distinct points in the expiratory-inspiratory phases of breathing [40]. Reflex apneas are induced peripherally via the vagus nerve by pharmacologically activating excitatory receptors located on nodose ganglia, or by mechanically activating stretch receptors located in the lung [16, 41]. Reflex apneas are also induced “downstream” by activation of glutamatergic receptors in the NTS [42–45]. Conversely, vagally-induced apnea can be reversed by microinjection of a GABA agonist [46], or a glutamatergic NMDA antagonist [45], into the NTS. Vagal afferents also synapse in regions of the NTS that modulate sympathetic activity, with activation of vagal afferents not only inducing apnea, but decreasing heart rate and blood pressure [41]. This trifecta of responses, known as the Bezold-Jarisch reflex, is modulated by different regions of the NTS, and these regions, when activated or inhibited, induce variable responses in blood pressure, heart rate, and breathing [45–47]. The NTS contains both CB₁ and CB₂ receptors that, when activated, inhibit or promote NTS activity, and attenuate other vagally-mediated behaviors [26, 27, 48–51]. CB₁ activation decreases glutamate release in the NTS [51], thus CB₁ activation would have the effect of suppressing apneas. CB₁ activation also disinhibits second order NTS neurons by pre-synaptically decreasing release of GABA [51–53], thus CB₁ activation would have the opposite effect of potentiating apneas. It might be plausible that the dual effects of CB₁ activation occurred, with the consequence of neither suppressing nor potentiating apneas.

Moreover, the XII, which modulates the phasic and tonic activation of the genioglossus muscle [36], also contains CB₁ receptors [23, 25, 28, 54, 55], of which the physiologic role is unclear. CB₁ activation in the XII is known to disinhibit XII by preventing release of glycine, an inhibitory neurotransmitter [25, 54]. In fact, a CB agonist microinjected in the XII activated the genioglossus in awake, but not sleeping, rats [55]. A recent report showed cardiorespiratory anomalies, including unstable breathing and apneas, in CB₁ knockout mice [56]. We hypothesized that the activation of CB receptors located at crucial respiratory centers in the brainstem would modulate reflex apneas and increase genioglossus activity; however, we saw no effect of a centrally-administered CB agonist, dronabinol.

The present work focused on global activation of CB receptors located in the brain, which mimicked patients undergoing dronabinol treatment [12], and did not elucidate any specific local effects of the respiratory centers of the brainstem. Due to the variability of Bezold-Jarisch reflex responses to NTS activation or inhibition, and the location of CB receptors on both excitatory and inhibitory neurons of the NTS, it is possible that non-specific and global activation of central CB receptors via ICV injection of dronabinol led to multiple and potentially opposing responses, masking any specific local effects. Therefore no physiologic response was observed. For example, Padley et al. induced apnea by microinjecting a CB₁ agonist centrally into the RVLM, which sends projections to inhibitory neurons located in the respiratory centers of the brainstem [39]. Also, injection of CBs peripherally in the nodose ganglia increased genioglossus activity, presumably through disinhibition of parasympathetic input into respiratory centers of the brain [15]. Carley et al. showed respiratory stability with intraperitoneal injections of dronabinol, which would involve both peripheral and central CB receptors [13]. Moreover, low and high doses of CBs are known to have biphasic effects [31, 57]; however, we saw no changes in apnea response at low or high doses of dronabinol. Since CB receptors are widely distributed in the brain, on both excitatory and inhibitory neurons, and are activated differentially by different concentrations of CBs, specific microinjections in the NTS or XII will need to be completed to see if CBs have any effect on breathing and genioglossus activity in rats.

In conclusion, we show that ICV injections of dronabinol, a non-specific CB agonist, had no effect on 5-HT-induced reflex apnea, and had no effect on genioglossus activity. These results suggest that central CB receptors have a minimal or no effect on breathing when activated globally; however, there might be specific local effects of CBs due to the diverse population of neuronal inputs and outputs of the NTS. Previous work showing stabilization of breathing from CBs might be derived from activation of CB receptors located on peripheral nerves [13–15], suggesting that pharmacotherapies targeting only peripheral CB receptors for OSA treatment might be sufficient. Future work will concentrate on elucidating specific local effects of CBs on breathing in the NTS and XII.