Chemogenetic stimulation of the hypoglossal neurons improves upper airway patency

Obstructive sleep apnea (OSA) is characterized by recurrent upper airway obstruction during sleep. OSA leads to high cardiovascular morbidity and mortality. The pathogenesis of OSA has been linked to a defect in neuromuscular control of the pharynx. There is no effective pharmacotherapy for OSA. The objective of this study was to determine whether upper airway patency can be improved using chemogenetic approach by deploying designer receptors exclusively activated by designer drug (DREADD) in the hypoglossal motorneurons. DREADD (rAAV5-hSyn-hM3(Gq)-mCherry) and control virus (rAAV5-hSyn-EGFP) were stereotactically administered to the hypoglossal nucleus of C57BL/6J mice. In 6–8 weeks genioglossus EMG and dynamic MRI of the upper airway were performed before and after administration of the DREADD ligand clozapine-N-oxide (CNO) or vehicle (saline). In DREADD-treated mice, CNO activated the genioglossus muscle and markedly dilated the pharynx, whereas saline had no effect. Control virus treated mice showed no effect of CNO. Our results suggest that chemogenetic approach can be considered as a treatment option for OSA and other motorneuron disorders.

after CNO and saline treatments. CNO induced a striking 3.12 fold increase in tonic GG activity, which was observed within 15 min of CNO administration (Fig. 2) and lasted the entire 6 hr experiment in all mice. CNO also induced a 1.23 fold increase in phasic GG activity, but the response varied between mice. In contrast, saline treatment had no effect in the same animals. The specificity of the CNO effect was tested in six mice infected with the control virus and six additional mice, which were not infected. In these animals CNO had no effect on EMG GG (see Fig. 2 for control virus data; uninfected mice not shown).
Nine out of thirteen DREADD-infected mice were examined in a dynamic MR imaging protocol. The pharynx was imaged in the mid-sagittal and multiple axial planes throughout respiratory cycle, both before and after injection of CNO (n = 6) or saline (n = 3) (Fig. 3). Both sagittal and axial dynamic images demonstrated that CNO dilated the pharynx throughout the respiratory cycle. The oropharynx closed at baseline (mice are obligate  , moving average (∫EMG GG ) and respiratory effort recorded at baseline (left) and after CNO administration (right). Note the robust increase in both phasic and tonic EMG activity after CNO. (B) EMG response to CNO or saline in the same DREADD treated animals (n = 13; 15 minutes after injections) and EMG response to CNO in mice treated with control virus (n = 6, 15 minutes after injection) normalized to peak phasic EMG at baseline. a.u. arbitrary units. *, p < 0.001. nasal breathers) was opened by CNO (Fig. 3A,B; Suppl. Fig. 2). At the rim of the soft palate, 4 mm caudal to the hard-soft palate junction, CNO increased the pharyngeal cross-sectional area independent of respiratory phase (p < 0.05), from 2.08 ± 0.29 mm 2 to 3.45 ± 0.87 mm 2 during inspiration and from 1.88 ± 0.45 mm 2 to 3.32 ± 0.93 mm 2 during expiration (Figs 3C and 4). In contrast, saline injections had no effect on upper airway patency. Three control virus treated mice (n = 3) were examined in the same MRI protocol. CNO had no effect on the upper airway dimensions throughout the respiratory cycle in these animals. For example, the pharyngeal cross-sectional area 4 mm caudal to the hard-soft palate junction during expiration was 2.13 ± 0.99 mm 2 and 2.09 ± 0.92 mm 2 before and after CNO, respectively.

Discussion
To our knowledge, this is the first study examining the effect of chemogenetic stimulation on the upper airway musculature. We report that DREADD-mediated excitation of hypoglossal motorneurons (1) dramatically increased both tonic and peak phasic GG muscle activity and (2) markedly improved upper airway patency throughout the respiratory cycle. Our data may open a new line of research for pharmacotherapy of OSA.
Considerable research effort has been dedicated to the development of pharmacological agents for OSA over the last several decades 8,12 . In patients with an anatomic predisposition to OSA due to adiposity or facial anatomy, dilator muscles maintain pharyngeal patency during wakefulness. Sleep leads to a decrease in muscle tone of upper airway dilators, including the GG muscle of the tongue, especially in REM sleep, resulting in OSA 13 . OSA patients also have blunted neuromuscular responses to the upper airway obstruction, which further contribute to nocturnal collapse of the upper airway 14,15 . Adrenergic and serotoninergic mechanisms stimulate hypoglossal motorneurons, but serotonin and noradrenaline reuptake inhibitors have not been effective [16][17][18] . More recently the inward rectifying potassium 2.4 channel (Kir2.4) has been identified as a novel drug target for OSA, but molecules modulating this channel have not been discovered 19,20 . The failure of therapeutics to relieve upper airway obstruction prompted us to examine whether artificially engineered receptors can recruit hypoglossal motorneurons. DREADDs are G-protein coupled human cholinergic receptors which have been chemogenetically engineered to recognize CNO but no naturally occurring mammalian ligands 11 . The excitatory hM3 (Gq) DREADD has been used to modulate neuronal function in multiple studies of feeding behavior 21,22 , energy expenditure 23,24 , memory 25 , sleep 26,27 and social behavior 28 , but to our knowledge DREADD has not been previously used to activate motorneurons. In the present study, we successfully deployed DREADDs in the hypoglossal nucleus. Specific activation of DREADDs with CNO increased both respiratory related phasic and tonic activity of GG, the main tongue protrudor muscle. EMG GG activation was observed for 6 hours suggesting a lasting effect. An array of protrudors, retractors and intrinsic muscles are involved in the dynamic control of tongue position and pharyngeal patency, and protrudor activation does not always result in the airway opening [29][30][31] . Despite complex tongue neuromotor control mechanisms, it is particularly significant that CNO produced substantial oropharyngeal dilation, suggesting that chemogenetics can provide a viable approach for treating upper airway obstruction.
Our study had several limitations. First, both EMG GG and MRI were performed in anesthetized rather than sleeping mice. Additional studies will be required to demonstrate that DREADDs can relieve airflow obstruction during natural sleep. Second, we used non-Cre dependent DREADDs, which were expressed non-selectively across cells and motorneuron pools 11 . Upper airway patency could be further improved by stimulating specific lingual muscles 30 with Cre-dependent DREADDs activated by retrograde viral vectors containing Cre-recombinase 32 . Third, neurophysiological recordings in brain slices to test the CNO effect on the excitability of hypoglossal neurons have not been performed.
Notwithstanding these limitations, our findings imply that DREADDs administered to the hypoglossal nucleus can dilate the upper airway, and suggest that gene therapy with DREADDs may provide novel treatment for OSA. Our findings also suggest potential utility of DREADDs to treat a broad spectrum of motorneuron diseases.  Histology. Mice were sacrificed by isoflurane overdose and rapidly perfused with ice-cold 4% paraformaldehyde in phosphate buffered saline (PBS). The brains were carefully removed, postfixed in 4% paraformaldehyde for 24 h at 4 °C and cryoprotected in 20% sucrose in PBS overnight at 4 °C. The next morning, brains were frozen on dry ice and stored in antifreeze solution at − 20 °C until further use. The medulla was cut into 30-μ m-thick coronal sections on a sliding microtome (Thermo Scientific HM 560; Waltham, MA, USA). The sections were performed via the entire medulla, mounted on glass slides sealed with antifade medium (Vector; Burlingame, CA, USA). Localization of DREADDs in the medulla was confirmed by visualization of mCherry protein expression using Texas Red Filter, whereas localization of the control virus was confirmed by visualization of EGFP expression using FITC filter (Zeiss Axio D.1 microscope Waltham, MA, USA).

Methods
Electromyography of the Genioglossus Muscle (EMG GG ). EMG GG recordings were performed 6-8 wks after DREADD or control virus administration. EMG GG was acquired and analyzed as previously described 34 . In brief, mice were anesthetized with isoflurane (2-3%) initially; thereafter, isoflurane was held at 1-2% to maintain a respiratory rate at 1 Hz (0.9-1.1 Hz). The two Teflon-insulated wire hook electrodes (stainless steel, Teflon-coated, full hard, 0.005-in. bare, 0.008-in. coated; A-M Systems, Carlsborg, WA) were inserted in the genioglossus muscle toward the base of the tongue. The bared ends (0.5 mm) were passed through 27G insulin needle and folded at the bevel in hook fashion. The needle directed the intramuscular placement of the hooks and the wires were sutured to the neck musculature to maintain placement.
The EMG GG signal was amplified, band-pass filtered from 30 to 1,000 Hz (alternating-current preamplifier; model P511K, Grass Instruments), and digitized at a sampling rate of 1,000 Hz (LabChart Pro 7). The EMG GG was rectified, and a 100 ms time constant was applied to compute the moving average (LabChart Pro 7). Respiratory effort was monitored with a sensing bladder wrapped around the mouse as previously described 35 .
DREADD infected mice (n = 13) were treated with clozapine-N-oxide (CNO, 1 mg/kg in saline i.p.) and vehicle (saline) four days apart. Control virus infected mice (n = 6) were treated with CNO only. EMG GG was recorded at baseline for at least 1 h before treatment, followed by continuous EMG GG recording for additional 6 hs after intervention. Both treatments were performed only if mice appear healthy and demonstrated normal weight gain, grooming behavior, and normal food and water intake. For quantitative analysis, the tonic (expiratory) and peak phasic (inspiratory) components were measured for 10 randomly selected breaths 15 minutes after CNO or saline injections. Tonic and phasic EMG GG measurements were normalized and expressed as a percent of average peak phasic moving average at baseline.
Another subset of mice (n = 6) was treated with CNO (1 mg/kg in saline i.p.), which was not preceded by viral vector infection (negative control) with EMG GG recorded at baseline and after CNO treatment as described above.
Mice were anesthetized by inhalation of 2.5% isoflurane in oxygen in an induction box, followed by maintenance at 1-2% isoflurane as needed to maintain stable respiration at 1 Hz. A stream of warm air was used to maintain body temperature between 36-38 °C during loading of the mouse into the scanner and while scanning. The mice were loaded into a cradle containing a conical anesthesia mask with the head fixed in place with ear bars. Respiration was monitored using a pneumatic sensor placed between the cradle and one side of the mouse's abdomen while rectal temperature was measured with a fiber optic sensor (SA Instruments, Stony Brook, NY). The mice were inserted rear feet first and prone into a Bruker Biospec Avance 7 Tesla MRI scanner equipped with a 120 mm actively-shielded gradient/shim coil and 35 mm linear birdcage transmit/receive resonator (Bruker Biospin, Ettlingen, Germany). Guided by initial pilot scans, a single midline sagittal image was acquired to measure the neck angle and to identify the junction between the hard palate and soft palate, which was used as a landmark for defining subsequent axial scans. This image was acquired with a fat-suppressed fast spin echo (RARE) pulse sequence triggered at end-expiration with repetition time TR = 1 breath (769-1000 ms), echo time TE = 8 ms, echo train length ETL = 8, two signal averages, slice thickness 1 mm, field-of-view FOV = 3 × 3 cm (head-foot × anterior-posterior) and matrix size MTX = 128 × 128, resulting in a voxel size of 234 μ m × 234 μ m. The neck angle was defined as the angle between the ventral margins of the brain and spinal cord. The position of the mice was adjusted outside the magnet, as needed, to achieve a neck angle between 110° and 130°. B0 field map-based shimming was performed to second order over a 1.8 × 2.3 × 5.9 mm voxel centered at the base of the tongue prior to collection of dynamic images. Localized 1H NMR spectroscopy of this voxel was performed with a respiratory-triggered PRESS sequence and yielded a water peak linewidth (full width at half maximum) of 35-45 Hz. Using the same midline sagittal geometry as for the above fast spin echo scan, a dynamic gradient echo scan was performed using a FLASH pulse sequence with parameters TR = 15 ms, TE = 2.5 ms, flip angle FA = 15°, 40 dynamic frames and four signal averages. MR images covered 600 ms of the respiratory cycle, from mid-inspiration to end-expiration. Typical scan time for each single-slice dynamic experiment was 7-10 minutes, depending upon the actual respiration rate. Additional dynamic scans were then performed for an axial slice through the junction of the hard and soft palates and for similar parallel slices centered 1, 2, 3 and 4 mm caudal to this landmark. Dynamic scans of these axial slices were acquired with the same parameters as for the midline sagittal slice, except FOV = 2 × 2 cm (anterior-posterior × left-right) was used, resulting in a voxel size of 156 μ m × 156 μ m.
After acquiring a complete set of sagittal and axial dynamic scans, the cradle containing the DREADD-infected mice (n = 9) were removed from the magnet, carefully lifted off the cradle and an i.p. injection of CNO (1 mg/kg in saline, n = 6) or saline (n = 3) was performed. The mice were carefully repositioned to ensure a reproducible neck angle, the cradle was reinserted into the magnet and the above imaging protocol was repeated to identify tonic and phasic changes in upper airway geometry in response to CNO. Control virus -infected mice (n = 3) were examined in the same manner, except that only CNO injections were performed. Data were transferred to ImageJ (NIH, Bethesda, MD) for manual delineation and cross-sectional area measurement of the pharyngeal airways in each dynamic frame for each axial slice. All cross-sectional area measurements were performed in a blinded fashion by a single observer (H.P.). Care was taken to match pre-CNO and post-CNO axial slices by identifying landmarks such as the tympanic bullae in each image. Analytic Methods. Mixed-effect multivariable linear regression models were developed to examine whether mouse EMG GG amplitude and pharyngeal area in the axial plane significantly changed after treatment while accounting for between-mice variations. Specifically, EMG GG and pharyngeal area averaged over a phase of respiration (inspiration or expiration) were modeled as functions of treatment (CNO vs. saline) and time point (baseline and after treatment). Separate analyses were performed on the tonic and phasic components of the EMG GG . Pharyngeal area analysis was stratified by respiratory phase. Analyses were performed with XTMIXED (STATA 12, Statacorp LP, College Station, TX) and R with LME package (www.R-project.org).