Combined agonists act synergistically to increase mucociliary clearance in a cystic fibrosis airway model

Mucus clearance, a primary innate defense mechanism of airways, is defective in patients with cystic fibrosis (CF) and CF animals. In previous work, the combination of a low dose of the cholinergic agonist, carbachol with forskolin or a β adrenergic agonist, isoproterenol synergistically increased mucociliary clearance velocity (MCCV) in ferret tracheas. Importantly, the present study shows that synergistic MCCV can also be produced in CF ferrets, with increases ~ 55% of WT. Synergistic MCCV was also produced in pigs. The combined agonists increased MCCV by increasing surface fluid via multiple mechanisms: increased fluid secretion from submucosal glands, increased anion secretion across surface epithelia and decreased Na+ absorption. To avoid bronchoconstriction, the cAMP agonist was applied 30 min before carbachol. This approach to increasing mucus clearance warrants testing for safety and efficacy in humans as a potential therapeutic for muco-obstructive diseases.

Combined agonists did not induce airway smooth muscle contraction or airway narrowing. The cAMP and Ca 2+ -elevating agonists that increase MCCV also affect airway smooth muscle. Used alone they have opposite effects: Ca 2+ -elevating agonists contract muscles whereas cAMP-elevating agonists relax them. For therapeutic use, the potential for producing unwanted bronchoconstriction with the combined agonists is a safety concern. To determine which effect predominates, we measured airway smooth muscle responses to carbachol ± 10 µM forskolin or formoterol using two different methods: measuring muscle tension and lumen area. Tension of ferret trachealis muscle bundles was measured to increasing carbachol concentrations ± 10 µM forskolin. Forskolin abolished tension increases to 0.3 and 0.6 µM carbachol and greatly reduced responses to higher doses of carbachol ( Fig. 2A,B). Lumen area, imaged in thin sliced piglet or ferret tracheal rings, displayed a sustained 20-40% reduction with exposure to 0.3 µM carbachol, but when carbachol was preceded by either forskolin or formoterol, it induced only a transient decrease of 5% or less (Fig. 2C-F). Importantly, this same protective effect was observed in CF ferrets (Fig. 2F).
The velocity of mucus clearance reflects the transportability of mucus and ciliary beat frequency. Transportability is in turn largely determined by hydration/concentration 17 and pH (or bicarbonate content) of the mucus 18,19 . A major source of upper airway fluid is submucosal glands, and the agonists we used to stimulate MCCV also stimulate submucosal gland secretion [20][21][22][23][24][25][26][27][28][29] . ASL depth and composition are also modified by surface epithelia that secrete and absorb electrolytes/fluid. Indeed, this is the principle means of controlling ASL in airways that lack submucosal glands 30 . In prior work by us and others 26,[31][32][33] evidence was found for cholinergic inhibition of Na + absorption, which would tend to increase the fluidity and transportability of mucus. Finally, CBF is increased by elevating either Ca 2+34 or cAMP 35 . The following experiments sought evidence to support or challenge a possible contribution of each of these mechanisms to synergistic increases in MCCV. www.nature.com/scientificreports/ Synergistic glandular mucus secretion in WT pigs, WT ferrets, and CF ferrets. We hypothesize that synergistic increases in MCCV rely partly on increased mucus secretion from submucosal glands. This hypothesis arises from evidence that combinations of [Ca 2+ ] i -elevating and [cAMP] i -elevating agonists produce synergistically elevated rates of mucus from submucosal glands of humans 22 , pigs 24 , and ferrets 21 . However, in those experiments different specific concentrations of agonists were used. To determine if the same protocols used here produced synergistic increases in secretion from submucosal glands, we measured mucus secretion rates of individual tracheal glands in WT pigs and ferrets and in CF ferrets via time-lapse optical imaging 28 while stimulating them with the same concentrations of agonists and durations of exposure used for MCCV studies.
All secretion rates are reported as nanoliters/min/gland.   Fig. 3A). The basal rate was significantly increased by each agonist alone and was further increased by their combination in either order. Rates to the combined agonists were significantly larger than the arithmetic sum of their individual responses: additive sum = 1.26 ± 0.19, 7 pigs versus combined agonists = 2.86 ± 0.25 (2.3-fold larger, P < 0.01, 8 pigs). Data for individual pigs is shown in Fig. 3B for forskolin first and in Fig. 3C for carbachol first. (See also Supplementary Movie 1).
WT Ferrets gave similar results ( Fig. 3D-F). In ferret tracheal glands, the average unstimulated secretion rate was ~ zero (0.003 ± 0.001, 67 glands, 7 ferrets, Fig. 3D). It was significantly increased by forskolin (0.26 ± 0.07, 37 glands, 7 ferrets P < 0.05) and by carbachol (0.94 ± 0.28, 30 glands, 7 ferrets P < 0.05). The secretion rates to the combined agonists were significantly larger than the arithmetical sum of their individual responses (Fig. 3D): overall arithmetic sum = 1.27 ± 0.23 versus 2.46 ± 0.39 for the combined agonists (1.9-fold larger, P < 0.05, 55-67 glands, 5-7 ferrets. Data for individual WT control ferrets is shown in Fig. 3E for forskolin first and in Fig. 3F for carbachol first. Importantly, CF ferrets (CFTR KO/KO ) also showed synergistic gland secretion in spite of having no response to forskolin alone. We were able to test only two CF ferrets ( Fig. 3G-I). One CF ferret was stimulated first with 10 µM forskolin, the other with 0.3 µM carbachol, and both with the synergy paradigm. Unstimulated secretion rates were ~ zero, as in WT ferrets. Forskolin alone failed to stimulate secretion as expected (0.01, 7 glands), carbachol alone increased the average secretion rate to 0.45 ± 0.16, and the combined agonists increased average rates to 1.23 ± 0.35, 7 glands, and 1.31 ± 0.19, 7 glands. When agonists were combined synergy was seen in both orders of addition. The averaged secretion rate across both ferrets to the combined agonists was 1.27 ± 0.15, which is 2.8 times the arithmetic sum of the two agonists used alone, and about half of the response of WT ferrets of 2.46 ± 0.39 16 .
To summarize this section, the rates of mucus secretion across both species and including CF ferrets is increased to values beyond the additive sum of the agents used alone, providing circumstantial evidence that gland secretion rates contribute to MCCV in our system. Combined agonists stimulate epithelial surface anion secretion and inhibit Na + absorption. The surface epithelia also modify ASL. Figure 4A is a cartoon of the main ion flows controlling ASL depth: anion-mediated fluid secretion increases, and Na + -mediated fluid absorption decreases ASL depth. We hypothesize that the combined agonists increase ASL depth and thus MCCV by stimulating secretion and inhibiting absorption (see also Fig. 6). Figure 4B shows our best example of an I sc trace from pig tracheal mucosa stimulated with forskolin followed by carbachol. Forskolin caused a sustained I sc increase with no measurable change in conductance. When 0.3 µM carbachol was then added, it induced a transient I sc increase followed by slow decreases in I sc and conductance, with conductance reduced to 84% of the pre-and immediate post-forskolin value after ~ 30 min. The ENaC inhibitor benzamil (Bz) did not cause further changes in I sc or conductance, suggesting that carbachol completely inhibited ENaC-dependent Na + absorption. At this point the epithelium is www.nature.com/scientificreports/ secreting anions, indicated by steep drops in I sc and conductance produced by the two anion channel inhibitors, BPO-27 and niflumic acid, with no counterbalancing absorption, so ASL depth is predicted to increase (dotted gold line in Fig. 6A) unless MCCV increases. Our evidence shows that MCCV does increase. Figure 4C-F shows summary plots of ΔI sc as a function of time and stimulation. Each panel shows responses to 10 µM forskolin or 0.3 µM carbachol for the first 30 min and then the combined agonists for the next 30 min for WT pigs (Fig. 4C,D) and WT ferrets (Fig. 4E,F). Forskolin increased ∆I sc as expected for both species, but when carbachol was added the ∆I sc diminished slowly (Fig. 4C,E). Our interpretation of I sc in the forskolin + carbachol condition is that forskolin mainly increased I sc by stimulating anion secretion while carbachol largely decreased I sc by inhibiting Na + absorption. Inhibiting Na + absorption would increase net fluid accumulation on www.nature.com/scientificreports/ the surface. When carbachol is added first, the ΔI sc decreased directly or after a transient increase (Fig. 4D,F). In both cases the subsequent ∆I sc increase to carbachol + forskolin is smaller than to forskolin alone, because of their opposite effects on I sc but additive effects on ASL depth. . We tested to see if CBF might display synergistic increases to the combined agonists. CBF (in Hz) of unstimulated human nasal cells in KRB (Krebs buffer solution) was 6.79 ± 1.69 at 25 °C and 10.46 ± 0.95 at 37 °C (4 subjects, P = 0.01) (see "Methods" section). As shown in Fig. 5, neither agonist increased CBF significantly, but when combined, their additive effects produced a 27.2% increase to 13.31 ± 0.77 Hz. This was a significant increase compared to unstimulated CBF (n = 4, P < 0.05), but not to the arithmetic sum of ∆CBF to the two agonists: combined agonists: 2.85 ± 0.76, and arithmetic sum: 2.19 ± 0.66 www.nature.com/scientificreports/ (n = 4, P = 0.47). Thus, while increases in CBF will contribute to increases in MCCV, they are unlikely to account for the synergistic increase in MCCV seen with the combined agonists (see "Discussion" section).

Discussion
Main findings. We have six main findings. (1) Combined agonists produced synergistic increase of MCCV in CF ferrets to 19.95 ± 4.12 mm/min, which is ~ 55% of MCCV in WT ferrets tested in similar conditions. (2) Little or no airway narrowing was produced by the combined agonists, even in CF ferrets. (3) Pigs also showed synergistic increases in MCCV, so the effect is not species-specific. As for mechanisms, we found: (4) synergistic increases in glandular mucus secretion in pigs, ferrets, and CF ferrets; (5) increased anion secretion and decreased Na + absorption by surface epithelia; and (6) increased CBF, but only with an additive effect. The magnitude of synergistic MCCV increases were multiple-fold higher than to either agonist alone or to their summed responses and were close to maximal values reported in vivo. In anesthetized ferrets, basal MCCV in vivo was 18.2 ± 1.0 mm/min, and was increased to 32.0 ± 3.8 mm/min with maximal anticholinesterase treatment 36 . In anesthetized pigs, averaged basal MCCV in vivo was ~ 7 mm/min, and averaged maximal MCCV was ~ 12 mm/ min 37 . If synergy also occurs in vivo, it should help mobilize mucus in certain obstructive airway diseases. The most significant result was the synergistic increase in MCCV of CF ferrets. This is intriguing because in CF ferret tracheas forskolin alone did not increase MCCV (Fig. 1A) or stimulate gland mucus secretion (Fig. 3H). Also, using a different synergy paradigm human submucosal gland secretion was lost in airways from subjects with CF 22 . Therefore, the combined agonists used in the present study must be activating a CFTR-independent anion secretion pathway that is refractory to forskolin alone (see below).
Strategies to increase mucus clearance. Strategies to increase mucus clearance are mainstays of cystic fibrosis treatment but are only modestly effective 6,[8][9][10] . Pulmozyme (recombinant human DNase), hypertonic www.nature.com/scientificreports/ saline, and mannitol all improve mucus clearance in CF, while inhalation of bicarbonate or tromethamine improved CF sputum rheology 38 . Long before Pulmozyme or hypertonic saline treatments, numerous studies documented that β-adrenergic (cAMP) agonists increased MCC 13,39 . Indeed β-adrenergic agonists, considered as bronchodilators, are now used ubiquitously for treating obstructive diseases. However, the doses needed to stimulate increased MCC are higher than those that reliably produce bronchodilation 39 , and so it is not clear to what extent the doses presently used are increasing MCC. Unlike β-adrenergic agents, cholinergic (Ca 2+ ) agents cause bronchial constriction, which is the basis for the methacholine challenge test 40 , although increased mucus transport in humans by cholinergic stimulation has been reported 41 . Cholinergic agents also stimulate mucus secretion, and it is widely held that mucus over-production contributes to muco-obstructive disease 42,43 . Thus, it is not surprising that no one has previously advocated a therapeutic use for inhaling an agent that stimulates mucus secretion and causes bronchoconstriction. Indeed, anti-cholinergic agents are used as treatments for COPD, with modest effectiveness apparently resulting primarily from increased bronchodilation 44 . Thus, our finding that a combination of forskolin (or a β-adrenergic, formoterol) and a low-dose cholinergic markedly increased MCCV was unexpected.
Our hypothesis is that the combined agonists increase MCCV mainly because they increase ASL volume via three processes: synergistic increases in gland mucus secretion, increased fluid secretion and decreased absorption by surface epithelia (Fig. 6). The combined agonists produced only modest, additive increases in CBF measured in Krebs solution. It is possible that larger increases in CBF depend on increases in ASL volume, which occurred in the MCCV experiments but not in the CBF experiments. CBF increases have been observed using micro optical coherence tomography (μOCT) to visualize transport in intact tracheas 45,46 . Importantly, all of this occurs in the absence of airway narrowing.
The concept that MCCV will be faster if ASL depth is increased is supported by studies of patients with pseudohypoaldosteronism (PHA), where loss of function mutations in ENaC subunits eliminate Na + absorption from the airway surface, which more than doubles the volume of ASL and causes a fourfold increase in 0-20 min clearance rates of inhaled tracer from the lungs 47 . Previously, we demonstrated that agonist-induced MCCV in ferrets was ~ doubled when ENaC was inhibited 16 . In those experiments 16 , stimulation with either forskolin or www.nature.com/scientificreports/ carbachol in the presence of ENaC inhibition increased MCCV to values similar to those seen with the combined agonists, providing additional evidence that synergistic MCCV results, in part, from ENaC inhibition. The idea that increased ASL provides faster clearance also underlies the logic of using β-agonists 39 and hypertonic saline 8,9 to increase clearance. It is also supported by studies of ex vivo pig tracheas, where stimulating secretion increased MCCV, blocking secretion slowed MCCV, and blocking absorption increased MCCV of tracheas after secretion had been blocked 20 .

Potential molecular and cellular mechanisms. Molecular and cellular mechanisms responsible for
synergistic MCCV by β-adrenergic and cholinergic agonists were not addressed in this study, but given our evidence that inhibition of ENaC contributes, prior works on molecular mechanisms of ENaC inhibition are relevant. A common theme is the role of elevated [Ca 2+ ] i [48][49][50] , which can be achieved with a wide range of agonists, including ATP, UTP, histamine, thapsigargin, and bradykinin 51 . Cholinergic agonists increase [Ca 2+ ] i ; other mechanisms include increasing extracellular antiproteases 27,52 ; and other ENaC inhibitors 25,53 by stimulating secretions from airway glands and surface epithelia.
Because we observed synergistic increases of MCCV and glandular secretion in CF ferrets, mechanisms that bypass CFTR must be involved. Intracellular crosstalk between cAMP and Ca 2+ signaling pathways via inositol 1,4,5-triphosphate receptor-binding protein release with IP 3 (IRBIT) has been shown to mediate synergy in salivary gland and pancreatic ducts 54 . Synergistic secretion by lacrimal glands in response to cAMP and cholinergic agonists was partly due to inhibition of p44/p42 mitogen-activated protein kinase (MAPK) by the cAMP agonist 55 . A previous study 50 demonstrated that synergistic fluid secretion by cAMP + Ca 2+ agonists could result from Ca 2+ release by a cAMP-dependent Ca 2+ release mechanism in addition to a Ca 2+ agonist in isolated serous cells from human nasal and WT & CFTR -/pig tracheal glands. Another study in HEK 293 cells 49 , however, has shown that a cAMP agonist, such as parathyroid hormone or isoproterenol, did not increase [Ca 2+ ] i by itself, but when combined with carbachol, a cAMP agonist potentiated carbachol-induced Ca 2+ release by unmasking a discrete Ca 2+ pool in ER. Discrepancies in previous reports might arise in part from using different cell or organ preparations and in part from using different measurement parameters, e.g., [Ca 2+ ] i versus [HCO 3 ] i ([pH] i ). Our earlier studies 22,24 have shown that there are CFTR-dependent and -independent paths in synergistic glandular mucus secretions, depending on the doses of β-adrenergic and cholinergic agonists.
Potential therapeutic relevance for muco-obstructive disease. Procedures to enhance mucociliary clearance are needed for people with muco-obstructive airway disease, including substantial numbers of people with CF 11,56 . Because β-agonists and methacholine are used routinely (the latter to test for hyperactive airways), little should stand in the way of testing them in combination except that it seems counterintuitive. Our ex vivo data show this combination is effective in speeding mucus clearance without inducing airway narrowing, even in CF animals (Fig. 2), which have airway muscles with increased sensitivity to cholinergic agonists 57 . However, as we found with WT ferrets, the combined agonists minimizes/prevents airway narrowing induced by carbachol in WT pigs and in CF ferret airways. Our results are consistent with an earlier study where greatly reduced bronchoconstriction was observed when a β-adrenergic agonist was administered prior to methacholine in CF children 58 .
It remains to be seen if this combination is safe in individuals with hyperactive airways. If results do warrant further testing in people with CF, it will be important to start early with healthier airways, because the trend observed with β-agonist improvement of MCC was that healthier airways showed more benefit than diseased airways 39 .

Materials and methods
Airway tissue procurement. CF ferret tissues. Seven transgenic CF ferret tracheas (five CFTR G551D/G551D , one CFTR ∆F508/∆F508 , one CFTR G551D/KO ) were used for MCC assays. These ferrets were raised on the CFTR modulator VX770. Dosing was stopped at least 3 weeks prior to euthanasia; no residual drug effect is expected or was observed (zero response to forskolin). Two CFTR KO/KO ferret tracheas were used for the tracheal, single gland mucus secretion rate assay. Two CF ferret tracheas (one CFTR ∆F508/∆F508 and one CFTR G551D/∆F508 ) ferret trachea were used for tracheal smooth muscle contraction assay. All isolated CF ferret tracheal trims (2-3 cm in length) were placed in DMEM culture medium immediately after euthanasia and shipped from the University of Iowa via overnight priority express. Human tissues. Human nasal mucosal tissues were obtained from nasal biopsies during endoscopic sinus surgeries at Yonsei University Hospital. All methods using human tissues were carried out in accordance with relevant guidelines and regulations of Yonsei University, Seoul, South Korea. All experimental protocols were Ciliary beat frequency measurement. Ciliary beat frequency was measured using human nasal mucosa in the lab where ferret and pig tracheal mucosa was not readily accessible. Human nasal mucosa from endoscopic nasal biopsies was further dissected under a microscope and placed in a chamber controlled for temperature and pH control. Perfused Krebs bicarbonate buffer was maintained at 37 °C and pH 7.4. Cilia were visualized with a Zeiss microscope equipped with 40 × or 60 × objectives (Munich, Germany) using differential interference contrast (DIC) optics. Images were viewed live and were captured automatically at 2,000 fps with a high frame-rate digital camera (optiMOS and NIS-Elements microscope imaging software (Nikon, Japan)) and converted to TIFF images. Images were obtained for 10 s at each condition and experiments were performed in the following sequence: (1) unstimulated CBF at room temperature; (2) unstimulated at 37 °C; (3) CBF at 37 °C with 0.3 μM carbachol; (4) wash for 10 min; (5) CBF with 10 μM forskolin; and (6) CBF with 0.3 µM carbachol added to the forskolin. Each condition was maintained for at least 10 min. Note that this paradigm differs from the synergy paradigms used for measuring MCCV and gland mucus secretion rates in that the exposure to agonists was ≥ 10 min instead of 30 min, and it omits the condition in which forskolin was added in addition to carbachol. All recordings at each condition were made at three different areas of the epithelia, and the analyzed CBF was averaged for each experiment. To analyze captured images and calculate CBF, an in-house coding with a MATLAB software (MA, USA) was used.
Airway smooth muscle contraction measurement. Two methods were used to measure tracheal smooth muscle contraction. One is designed to measure airway narrowing using thin sliced tracheal rings. Piglet or ferret tracheal ring preparations of ~ 2 mm were submerged and securely pinned on a Sylgard-lined Petri dish filled with KRB solution at 37 °C and pH 7.4. Digital images of tracheal ring contractions in response to agonists for 1-10 min intervals were recorded with a Nikon digital camera and the inner lumen surface area of the tracheal ring was calculated using ImageJ (NIH, MD/USA). The other method is using a force transducer. One end of an isolated ferret trachealis muscle bundle was secured in a Sylgard-lined Petri dish filled with KRB solution and the other end was attached by 26-gauge wire to a previously calibrated strain gauge (series 400A force transducer system, Cambridge Technology, MA/USA). Tension responses to increasing carbachol doses ± 10 µM forskolin were obtained and displayed with PowerLab Chart4 software (ADInstruments, CO/USA).