Acidic Submucosal Gland pH and Elevated Protein Concentration Produce Abnormal Cystic Fibrosis Mucus

SUMMARY

In response to respiratory insults, airway submucosal glands secrete copious mucus strands to increase mucociliary clearance and protect the lung. However, in cystic fibrosis, stimulating submucosal glands has the opposite effect, disrupting mucociliary transport. In CF pigs, loss of CFTR anion channels produced submucosal gland mucus that was abnormally acidic with an increased protein concentration. To test whether these variables alter mucus, we produced a microfluidic model of submucosal glands using mucus vesicles from banana slugs. Acidic pH and increased protein concentration decreased mucus gel volume and increased mucus strand elasticity and tensile strength. However, once mucus strands were formed, changing pH or protein concentration largely failed to alter the biophysical properties. Likewise, raising pH or apical perfusion did not improve clearance of mucus strands from cystic fibrosis airways. These findings reveal mechanisms responsible for impaired mucociliary transport in cystic fibrosis and have important implications for potential treatments.

Graphical Abstract

eTOC Blurb

In cystic fibrosis, airway submucosal glands produce mucus that disrupts mucociliary transport. Using pigs, which have lungs like humans, and a microfluidic model of glands, Xie et al. show that acidic pH and increased protein concentration increase the elasticity and tensile strength of mucus strands, thereby preventing normal mucociliary transport.

INTRODUCTION

Cystic fibrosis (CF) is a common autosomal recessive disease caused by mutations in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR) anion channel (Quinton, 1990; Riordan et al., 1989). Loss of CFTR function impairs respiratory defenses, including mucociliary transport (MCT), which protects airways by trapping potential pathogens and particulate material in mucus and then propelling the mucus up the airways through the action of cilia beating (Fahy and Dickey, 2010; Robinson and Bye, 2002; Wanner et al., 1996; Wine and Joo, 2004). Defective MCT is a primary CF host defense defect that contributes to the mucus obstruction, bacterial infection, and inflammation that produce most CF mortality and morbidity (Ermund et al., 2018; Hoegger et al., 2014; Stoltz et al., 2015; Wine et al., 2018).

In cartilaginous airways of humans, pigs, and other large mammals, submucosal glands produce much of the mucus that facilitates removal of material from the lung (Widdicombe and Wine, 2015). Mucus is produced and stored in mucus vesicles within the gland’s mucous cells. Exocytosis of the mucus vesicles releases mucus into the lumen of the submucosal gland acinus where it very rapidly expands in liquid secreted by serous cells and forms a gel. When faced with a challenge from inhaled or aspirated material or pathogens, cholinergic stimulation induces submucosal glands to secrete abundant amounts of mucus (Ballard and Spadafora, 2007; Wine and Joo, 2004). Previous studies using pigs, which have submucosal glands like those of humans (Joo et al., 2010; Rogers et al., 2008a), revealed that after traveling up the submucosal gland duct, mucus emerges onto the airway surface in the form of strands (Ermund et al., 2017; Fischer et al., 2019; Oppenheimer, 1981; Ostedgaard et al., 2017; Simel et al., 1984; Trillo-Muyo et al., 2018). Beating cilia on surface epithelial cells pull mucus strands toward the larynx. The strands then break and sweep across the surface capturing particulate material and pathogens and clearing them from the lung. In CF airways, mucus strands sometimes remain attached to the submucosal gland duct and thereby impair MCT (Ermund et al., 2018; Hoegger et al., 2014; Ostedgaard et al., 2017; Wine et al., 2018).

Several factors have limited the ability of previous studies to answer questions about how submucosal glands determine the biophysical properties of mucus in non-CF and CF airways. It is difficult to collect mucus from healthy human airways, although more can be collected from diseased airways. But in both cases, the material is a mixture of mucus from submucosal glands, surface epithelia, and distal airways (beyond 10 generations in humans). In addition, mucus collected from humans often contains proteases, products of neutrophils and macrophages, inflammatory proteins, DNA, and phagocytic and epithelial cells. In contrast to humans, mice have only a few submucosal glands near the larynx, and they do not develop lung disease like that in humans with CF (Grubb and Boucher, 1999; Guilbault et al., 2007). Although human and porcine submucosal gland cells have been cultured as epithelia on planar permeable supports (Widdicombe et al., 2012; Yamaya et al., 1991), and the electrolyte transport properties of isolated serous cells have been assessed (Lee and Foskett, 2010a, b), these cells lack the structure of submucosal glands.

Therefore, we studied newborn pigs, which have airways like those of humans, including abundant submucosal glands. Pigs lacking CFTR (CFTR^−/−) exhibit host defense defects in MCT and antimicrobial protein killing of bacteria, and they develop CF airway disease that is similar to that in humans (Ermund et al., 2018; Hoegger et al., 2014; Joo et al., 2010; Ostedgaard et al., 2011; Pezzulo et al., 2012; Rogers et al., 2008a; Rogers et al., 2008b; Stoltz et al., 2010). We found that newborn CF porcine submucosal glands produce mucus that is abnormally acidic and has an increased protein concentration. To learn how pH and mucus protein concentration determines the properties of mucus strands, we turned to mucus produced by banana slugs (Ariolimax columbianus) (Deyrup-Olsen et al., 1983). Importantly for our studies, intact mucus vesicles (also called granules) can be collected from slugs, so that we could control variables at the moment that mucus is released from vesicles to form a mucus gel. Finally, we asked whether the viscoelastic properties of mucus could be changed after formation of the gel; the answer has important implications for understanding mucus hydrogels and for potential therapies.

RESULTS

Mucus from newborn porcine CF submucosal glands has an abnormally acidic pH, a reduced volume, and an increased protein concentration.

To access mucus produced by submucosal glands, we removed tracheas from newborn pigs, opened them along the ventral surface, and mounted them on a 3D-printed device (Fig. 1A). The apical side was cleaned with phosphate-buffered saline, dried with a stream of air, and covered with water-saturated mineral oil. The basolateral side was perfused with Krebs buffer with HCO₃⁻/CO₂ (pH 7.4) at 37°C. We added methacholine basolaterally to stimulate submucosal gland secretion and found that mucus droplets formed rapidly above submucosal gland ducts (Fig. 1B). We measured their volume and found that CF submucosal glands secreted less liquid than non-CF (Fig. 1C), consistent with earlier reports in human and porcine airways (Joo et al., 2010; Joo et al., 2006).

Figure 1. CF submucosal glands secrete abnormally acidic and concentrated mucus.

(A) Trachea from newborn pigs was opened along the ventral surface and mounted on a 3D printed device. Its apical side was cleaned, air dried, and covered with mineral oil. The basolateral side was perfused with Krebs buffer with HCO⁻₃/CO₂ (pH 7.4) and methacholine (2.5 mg/mL). A nanoliter pipet was used to collect mucus for analysis.

(B) Image of mucus droplets secreted from individual submucosal glands 30 min after beginning perfusion with methacholine. Scale bar 200 μm.

(C-F) Volume, pH, protein concentration, and protein amount of mucus secreted in 30 min after methacholine addition. Submucosal glands were from non-CF and CF pigs. N=10 for CF; N=8 for non-CF.

(G,H) Effect of perfusing the basolateral surface of non-CF airways with buffer containing bumetanide (10 μM) and HCO₃⁻-free HEPES-buffered saline at pH 7.4 on pH (G) and volume of submucosal gland mucus (H). N=7.

(I-K) Effect of perfusing the basolateral surface of non-CF airways with buffer containing bumetanide (10 μM) and HCO₃⁻-free HEPES-buffered saline at pH 7.4 on protein concentration (I), volume (J), and protein amount (K) of mucus secreted from submucosal glands in 30 min after methacholine addition. N=6.

Each data point or set of data points and lines indicates a different pig, and data points represent average of 3 measurements per pig in panels D, E, F, G, I and K; and 50 measurements per pig with exclusion of mucus doublets in panel C, H and J. Bars indicate mean±SEM. Data were analyzed using un-paired t-tests (C-F), one-way ANOVA and Tukey post hoc test (G, H), or paired t-tests (I-K). * P < 0.05; **P < 0.01; ***P < 0.001.

We collected the mucus droplets and analyzed their pH with the fluorophore BCECF (Fig. S1). Mucus from non-CF submucosal glands had a pH of 7.46±0.07, a value similar to the pH of the basolateral solution (pH=7.4) (Fig. 1D). In contrast, CF mucus was more acidic (pH 7.05±0.10), consistent with loss of HCO₃⁻ secretion in the continued presence of H⁺ secretion (Shah et al., 2016). A previous report of human submucosal gland secretions had suggested that pH was similar in CF and non-CF (Jayaraman et al., 2001). However, when the same group expanded the study in a subsequent report, they found that submucosal glands produced more acidic secretions in CF (pH 6.6±0.1) than non-CF (pH 7.2±0.1) airways (Song et al., 2006).

We also measured the protein concentration of the collected submucosal gland mucus (Fig. S1). Compared to non-CF pigs, mucus from CF pigs had a higher protein concentration (Fig. 1E). The amount of protein produced by single glands, calculated from the volume and protein concentration, showed a trend toward a reduction in CF (Fig. 1F).

CFTR anion channels conduct both HCO₃⁻ and Cl⁻ (Chen et al., 2010b; Linsdell et al., 1997; Poulsen et al., 1994; Smith and Welsh, 1992). To assess the contributions of HCO₃⁻ secretion, Cl⁻ secretion, or both to submucosal gland mucus pH and volume, we perfused the basolateral surface with three additional solutions: a) HCO₃⁻/CO₂-free Krebs solution (HEPES buffer, pH 7.4) to prevent HCO₃⁻ secretion, b) Krebs solution containing bumetanide to inhibit basolateral Cl^– entry and thereby Cl⁻ secretion (Ballard and Spadafora, 2007; Trout et al., 1998), and c) HCO₃⁻/CO₂-free Krebs containing bumetanide to inhibit both HCO₃⁻ and Cl⁻ secretion. Preventing HCO₃⁻ secretion decreased the pH of submucosal gland mucus, whereas inhibiting Cl⁻ secretion had little effect on pH (Fig. 1G). Conversely, inhibiting Cl⁻ secretion decreased the volume of mucus produced, whereas preventing HCO₃⁻ secretion had a trend toward volume reduction that was not statistically significant (Fig. 1H). When both HCO₃⁻ and Cl⁻ secretion were inhibited, the volume and pH of secretions fell (Fig. 1G,1H). Preventing anion secretion with HCO₃⁻/CO₂-free Krebs plus bumetanide also increased the protein concentration of submucosal gland mucus (Fig. 1I). The increased protein concentration together with a reduced volume (Fig. 1J) decreased the total amount of secreted protein (Fig. 1K).

These data indicate that in the absence of CFTR, cholinergic stimulation produces submucosal gland mucus that is more acidic and has a higher protein concentration than that in non-CF airways. HCO₃⁻ plays the major role in buffering secreted protons, and Cl⁻ plays a major role in volume secretion. These findings correspond with our earlier observation that inhibiting both HCO₃⁻ and Cl⁻ secretion reproduces much of the CF mucus phenotype in pig airways (Hoegger et al., 2014; Ostedgaard et al., 2017). Together, they suggest the hypothesis that both pH and protein concentration influence the biophysical properties of mucus.

At the time when mucus gels are formed, pH, Ca²⁺, and protein concentration determine the gel’s volume, elasticity, and viscosity.

Submucosal gland mucous cells tightly pack mucus into vesicles containing an acidic pH and a high Ca²⁺ concentration, [Ca²⁺] (Ambort et al., 2012; Raynal et al., 2003; Ridley et al., 2014). When exocytosis releases mucus into the gland lumen, it expands in an ionic environment determined in large part by electrolyte transport by submucosal gland serous cells (Basbaum et al., 1990; Lee and Foskett, 2010a, b).

Learning how the ionic environment affects the properties of mucus gels, requires control of the chemical composition to which mucus is exposed at the moment of its release. However, it is problematic to control the chemical composition in the lumen of submucosal gland, ensure a solution with a large enough volume to maintain constant conditions, elicit mucus release from vesicles at a defined time, and assess the small volume of mucus produced by single mucus vesicles from submucosal gland cells. Therefore, we turned to banana slugs (Ariolimax columbianus) for a model of mucus release and unfolding (Deyrup-Olsen et al., 1983).

The mucus producing tissue in A. columbianus resembles that of mucous cells in porcine submucosal glands, and lectin staining and scanning electron microscopy revealed a similar appearance of pig and slug mucus (Fig. 2A–2C). Slug mucus vesicles were elliptical with a normally distributed volume of 0.12±0.08 pL (mean±SD) (Fig. S2).

Figure 2. Mucosal structures and mucus of banana slugs share features with pig airway.

(A) Transmission electron micrographs show a mucous cell with mucus vesicles from a slug’s epithelial surface and a pig’s submucosal gland. Scale bars: 1 μm.

(B) Mucus strands stained with wheat germ agglutinin lectin. Scale bars: 100 μm.

(C) Scanning electron microscopic images of mucus. Scale bars: 500 nm.

To release mucus, we disrupted the vesicle membrane with sonication, nystatin, or both, and measured volume by confocal imaging of fluorescent nanospheres attached to their surface. With vesicle rupture in a pH 7, Ca²⁺-free solution, mucus expanded to a volume 122±31-fold the original vesicle volume. Expansion was rapid with a half time of 0.4 sec (Fig. S3, Video S1), consistent with previous reports (Tanaka and Fillmore, 1979).

We tested the hypothesis that pH affects mucus properties during formation of the mucus gel by varying pH over a large range (pH 5 to 9) and measuring the volume of mucus produced by individual vesicles (Fig. 3A). We also tested the effect of Ca²⁺. Mucus vesicles contain a high [Ca²⁺], and [Ca²⁺] has been proposed to alter mucus properties in CF (Chen et al., 2010a; Tang et al., 2016; Verdugo, 1990). An increased [Ca²⁺] was also found in sputum produced by people with CF (Smith et al., 2014). Therefore, we also compared formation of mucus gels in the presence and absence of Ca²⁺ (10 mM Ca²⁺ or 10 mM EGTA with nominal Ca²⁺). Both in the presence and absence of Ca²⁺, mucus gels were smaller when produced in an acidic solution compared to an alkaline pH (Fig. 3A). Moreover, at constant pH values, Ca²⁺ reduced the size of mucus gels.

Figure 3. Acidic pH, Ca²⁺, and increased protein concentration reduce the volume of mucus gels and increase elasticity.

(A) The effect of pH and Ca²⁺ on the volume of individual mucus gels. N=23, 19, 26, 25 slugs for each group. Data for each slug are the average of measurements of at least 5 mucus gels.

(B) The effect of pH and Ca²⁺ on elastic modulus (G’) and viscous modulus (G”) of slug mucus gels. N=10.

(C) Relationship between mucus gel protein concentration and elastic modulus (G’) and viscous modulus (G”) at pH 5 with 10 mM Ca²⁺ and at pH 9 without Ca²⁺. N=6. Lines are simple linear regression.

Data points indicate data from individual animals. Bars indicate mean±SEM. Data were analyzed using one-way ANOVA and Tukey post hoc test (A,B), or F test for linear regression (C). *P < 0.05; **P < 0.01; ***P < 0.001.

To further assess the properties of the mucus, we used oscillatory rheometry to measure the elastic modulus (G’), which indicates a gel’s tendency to recover its original shape following stress-induced deformation, and the viscous modulus (G”), which indicates a gel’s resistance to flow (Fig. S4). Under all conditions studied, the elastic modulus was ~10-fold greater than the viscous modulus, indicating that elasticity dominates the behavior of the mucus gel (Fig. 3B) (Nielsen et al., 2004). Forming mucus gels at pH 5 doubled elasticity compared to pH 9 (Fig. 3B). There was also a trend for Ca²⁺ to increase elasticity. The effects of pH and Ca²⁺ on viscosity were not statistically significant.

To test the effect of protein concentration, [protein], we ruptured varying amounts of mucus vesicles in a defined volume. Of note, [protein] represents the total protein concentration in mucus, rather than the concentration of mucin. As [protein] increased, elasticity increased (Fig. 3C). Moreover, the slope of the relationship was much steeper when mucus was formed in a pH 5 solution with Ca²⁺ compared to a pH 9, Ca²⁺-free solution. The [protein], pH, and Ca²⁺ had similar effects on viscosity (Fig. 3C).

After formation, mucus gels resist modifications in volume and elasticity in response to variations in pH, Ca²⁺ and solution volume.

Once mucus strands emerge onto the airway surface, they are exposed to airway surface liquid, which may have pH and Ca²⁺ concentrations that differ from those in the submucosal gland where mucus gels are formed. In addition, inhaled solutions and therapeutics may alter the composition of airway surface liquid. Therefore, we asked if changing pH, Ca²⁺, and/or protein concentration would alter the biophysical properties of mucus gels after they had formed (Fig. 4A).

Figure 4. After formation, mucus gels resist modification by changes in pH, Ca²⁺, and protein concentration.

(A) Example of surface rendering of an individual mucus vesicle and resulting mucus gel after rapture and formation, and then after modification of experimental conditions, measurements including elastic modulus (G’) and viscous modulus (G”).

(B) Effect of altering pH and/or [Ca²⁺] after mucus formation on elastic and viscous moduli. N=5–6.

(C) Effect of altering protein concentration through evaporation after mucus formation on elastic and viscous moduli. N=5–6.

Each set of data points and lines indicates data from individual slugs. Bars indicate mean±SEM. Data were analyzed using paired t-test (B), or one-way ANOVA and Tukey post hoc test (C). Differences were not statistically significant (B). *P < 0.05.

We found that switching from extremes of pH and Ca²⁺ at gel formation to the opposite extreme afterwards, failed to alter elasticity or viscosity (Fig. 4B). Likewise, forming a gel and then reducing the volume to increase the protein concentration by evaporating liquid did not increase elasticity or viscosity (Fig. 4C). Thus, once gels were formed, their elasticity and viscosity largely resisted changes in response to altered pH, Ca²⁺, and protein concentration.

An artificial submucosal gland produces mucus strands.

Previous studies suggested that the distal portion of respiratory submucosal glands comprise serous cells, which produce much of the liquid, and proximal to that lie mucous cells, which release mucus (Meyrick et al., 1969). After formation in the gland acinus, the mucus gel moves through the gland duct to the epithelial surface where it emerges in the form of mucus strands. Mucus strands grow in length, break, and then cilia sweep them up the airways as they capture particulate material (Ermund et al., 2017; Fischer et al., 2019; Hoegger et al., 2014; Ostedgaard et al., 2017). To test the hypothesis that the liquid composition at the time of mucus gel formation influences subsequent mucus strand behavior, we produced a microfluidic device to mimic a submucosal gland.

Inspired by the organization between mucous cells and serous cells in submucosal glands, we designed a microfluidic device that featured two up-stream inlets to inject slug mucus vesicles and a solution with varying pH and Ca²⁺ concentration (Inlets 1 and 2) (Fig. 5A). To duplicate the rapid release of mucus by exocytosis in submucosal glands, we placed a micro-sized interdigital transducer beneath the channel to quickly rupture vesicles by acoustic streaming (Lighthill, 1978) (Fig. 5A, Video S2). To mimic emergence of mucus strands from gland ducts, the released mucus was anchored on a vertical pillar in a flowing stream of liquid (Inlet 3). We visualized mucus strands by incorporating fluorescent nanospheres (20 nm, red), which bound to the mucus (Fig. 5B). Stretching and breaking of mucus strands was quantitatively assessed by labeling with fluorescent microspheres (1 μm, green). When we progressively increased the flow rate from inlet 3, mucus strands stretched, eventually broke, and then the proximal portion snapped back (Fig. 5B). This behavior resembled that of mucus strands emerging from submucosal gland ducts in pig airways (Hoegger et al., 2014). To quantify the stretch of mucus strands, we calculated strain by the displacement between the anchored pillar and attached microspheres (Fig. S5). To quantify strand breakage, we measured the percentage of a strand that remained attached to the pillar in the flowing stream of liquid.

Figure 5. Ca²⁺ and acidic pH decrease strain and breakage in a microfluidic model of submucosal gland.

(A) Schematic showing microfluidic model of submucosal gland. Inset shows an interdigital transducer that generates ultrasound to rupture mucus vesicles. Scale bar 100 μm.

(B) Example of a mucus strand. After mucus release from vesicles, the strand hangs on the pillar in the flowing liquid stream from inlet 3. In left panel, mucus strand is visualized with red nanospheres (20 nm). In three right panels, green microspheres (1 μm) reveal the displacement and breakage of mucus strands as flow rate progressively increases. Arrowheads indicate individual green microspheres on a single mucus strand. Scale bar 100 μm.

(C) Left panel shows flow switching between 10 and 50 μL/min (17 and 83 mm/s, respectively) and right panel shows resulting strain. Solution was pH 7 with Ca²⁺. In response to the periodic changes in flow rate, mucus strands stretched and relaxed back to the initial length with trivial residual strain. This test was conducted on mucus strands formed at all the pH and Ca²⁺ levels, and the results were similar for all conditions.

(D) Protocol for increasing flow rate from 10 to 1000 μL/min (from 17 mm/s to 1.7 m/s, respectively).

(E) Strain at the flow rate of 250 μL/min. In most cases strands failed to form in pH 9 solution without Ca²⁺. Data for strands from each slug represent an average of 5 measurements of strain. N=6.

(F) Percentage of mucus strands remaining at the flow rate of 1000 μL/min (1.7 m/s). N=6. In panels E and F, data points are from different slugs. Bars indicate mean±SEM. Data were analyzed using one-way ANOVA and Tukey post hoc test. *P < 0.05; **P < 0.01; ***P < 0.001.

The pH and Ca²⁺ at the time of mucus strand formation determine its strain and breakage.

To determine whether mucus strands exhibit primarily elastic or viscous behavior, we applied alternating low (10 μL/min, ~17 mm/sec) and high (50 μL/min, ~83 mm/sec) flow rates through Inlet 3 to stretch and relax mucus strands (Fig. 5C). Of note, the flow velocity we applied is larger than the maximum velocity at cilia tips (~1 mm/sec) (Salathe, 2007). During five cycles with a 20 sec period, the mucus strand returned to the original state when flow slowed, resisting permanent deformation (Fig. 5C, Video S3). These results indicate that the mucus strands exhibit predominantly elastic behavior, a result consistent with our assay of mucus gels with classical rheometry.

Elastic behavior led us to investigate strain and breaking of mucus strands that formed at different pH and Ca²⁺. To gradually stretch and break mucus strands, we increased flow from 10 to 1000 μL/min (1.7 meters/sec) in 50 μL/min steps at 10 sec intervals (Fig. 5D). The solution in all inlets was the same. Flow rates <250 μL/min (~417 mm/sec) were defined as stretch-dominant regions because we observed little breaking at those flows. Reducing the pH (Inlets 2 and 3) decreased strain, indicating a greater resistance to deformation (Fig. 5E, S6A). The presence of Ca²⁺ also reduced strain. As flow rate increased, strands broke and were swept away (Video S3). As pH of the solutions fell, fewer strands broke (Fig. 5F, S6B). Ca²⁺-containing solution also reduced strand breakage. Reduced strain and breakage indicate mucus strands have a greater elasticity and tensile strength when formed at a low pH and in the presence of Ca²⁺.

Once strands are formed, changing pH and [Ca²⁺] has little impact on strain and breakage.

To test the hypothesis that changing the solution after mucus gels had formed would change their properties, we modified the microfluidic chamber design so that we could stop flow from the formation condition (Inlet 3) and switch it to a modification condition (Inlet 4) (Fig. 6A). During the switch, the flow rate was constant and remained laminar.

Figure 6. After mucus strands formed, changing the pH and [Ca²⁺] had minimal effects on their strain and breakage.

(A) Schematic of microfluidic device to alter chemical conditions after mucus strand formation. Mucus was formed and attached to the pillar at a specified pH and [Ca²⁺] (identical solution flowing through inputs 1, 2, and 3). After mucus was formed, flow through inputs 1 and 2 was then stopped. After 2 min of mucus strand formation, flow was switched from input 3 to input 4 to vary the solution composition. The switch and laminar flow are shown with colored solution from inlets 3 and 4. Flow rate through inlet 4 was then progressively increased as shown in Fig. 5D to measure strain and strand breakage. Mucus strands were formed at pH 5 with 10 mM Ca²⁺. Scale bar 200 μm.

(B,C) Mucus strand strain and the percentage of strands remaining after switching flow to solutions with the indicated pH and Ca²⁺. N=6. In panels B and C, each point indicates data from individual slugs. Data for each slug represent the average of 5 measurements of strain. Bars indicate mean±SEM. Differences were not statistically significant by one-way ANOVA with a Tukey post hoc test.

We formed mucus strands at pH 5 with Ca²⁺. We then switch to a pH 9 solution with or without Ca²⁺ and then measured strain and breakage. Increasing pH and removing Ca²⁺ failed to increase strain (Fig. 6B, S6C). Likewise, breakage was largely unaltered, although there was a trend for increased breakage at pH 9 without Ca²⁺ when the flow rate reached values of 1000 μL/min (1.7 meters/sec) (Fig. 6C, S6D). Thus, once mucus strands had formed, their biophysical properties were largely fixed; post-formation adjustment of pH and Ca²⁺ had minimal impact on strain or breakage.

Raising pH and/or reducing [Ca²⁺] does not break mucus strands on CF airways.

Based on our finding that 1) porcine CF submucosal gland mucus is abnormally acidic and concentrated, 2) acidic pH, Ca²⁺, and increased protein concentration increase elasticity and tensile strength of A. columbianus mucus, and 3) once the mucus gel has formed it resists modification of its biophysical properties, we predicted that raising apical pH and/or reducing apical Ca²⁺ would fail to increase breakage of mucus strands emerging from CF submucosal glands.

To test this prediction, we opened porcine tracheas ventrally and mounted them in a chamber that allowed separate control of solutions bathing the apical and basolateral surfaces (Fig. 7A). We stimulated submucosal gland secretion with methacholine added basolaterally and labeled mucus by including 20 nm fluorescent nanospheres in the apical HCO₃⁻/CO₂ Krebs solution. Inspection of the airway surface revealed mucus strands being swept to the edge of the chamber propelled by cilia, consistent with earlier reports (Ermund et al., 2017; Fischer et al., 2019; Hoegger et al., 2014). To quantify mucus strands that remained attached on the airway surface, we obtained panoramic images of the airway surface with confocal microscopy and used an averaging technique that allowed us to preferentially observe stationary strands (Hoegger et al., 2014). We counted attached strands using a grid-based analysis strategy (Fig. S7). We then changed the apical solution to test the effect of solution composition on strand behavior.

Figure 7. Modification of apical pH and [Ca²⁺] had minimal effects on mucus strand clearance in excised pig trachea.

(A) A 1 cm length of pig trachea was opened ventrally and mounted on a dual chamber device that allowed separate perfusion of apical and basolateral surfaces. Prior to beginning measurements, the basolateral solution was perfused for 30 min with Krebs Ringer solution that contained methacholine (12.8 μM). The apical surface was perfused with pH 7.3 HCO₃⁻/CO₂ Krebs solution that contained fluorescent nanospheres (20 nm). During this preincubation period, mucus strands formed on the apical surface. Images of the surface were obtained at times 0, 5, and 10 min. The apical perfusate was then switched to a) the same solution, b) solution at pH 8.3, c) Ca²⁺-free solution, d) pH 8.3 and Ca²⁺-free solution, or e) solution containing 10 mM DTT. Images of the surface were then obtained at times 15, 20, and 25 min.

(B) Examples of mucus strands on the surface of non-CF airway, CF airway, and non-CF airway basolaterally perfused with HCO₃⁻/CO₂-free HEPES buffered solution containing 10 μM bumetanide. Images were obtained at 10 min. Scale bar 2 mm. A grid was placed over each image and the percentage of grid sections containing a mucus strand was determined (Fig. S7). The percentage of grid sections containing a mucus strand on non-CF airways was 7.7±1.6% (n=6), on CF airways was 33±3% (n=27), and on non-CF airways basolaterally perfused with HCO₃⁻-free solution with bumetanide was 38±4% (n=33).

(C,D) Decrease in mucus strands covering the airway. Data are the decrease in percentage of grids that contained a mucus strand at time 25 min vs. 10 min. In panel C, data are non-CF and CF airways with HCO₃⁻/CO₂ solution on the basolateral surface. In panel D, data are non-CF epithelia perfused basolaterally with HCO₃⁻/CO₂-free HEPES buffered solution containing 10 μM bumetanide. See also Fig. S7. N=5–7. In panels C and D, each data point is from a different pig. Bars indicate mean±SEM. Data were analyzed using one-way ANOVA and Tukey post hoc test. *P < 0.05; **P < 0.01.

The surface of non-CF airways had very few attached mucus strands (8±2% of grids, n=6) (Fig. 7B,S7B). In contrast, the surface of CF airways treated in the same way had numerous stationary mucus strands that were attached to submucosal gland ducts (33±3% of grids, n=27) (Fig. S7C–S7G). These results are consistent with earlier data (Hoegger et al., 2014). Perfusing the apical surface of CF airways with a solution at pH 8.3, a Ca²⁺-free solution, or both pH 8.3 and Ca²⁺-free caused little breakage and clearance of mucus strands (Fig. 7C,S7C–S7G). As a positive control, we perfused the apical surface with 10 mM dithiothreitol (DTT), which reduces disulfide bonds between mucin molecules. DTT broke the mucus strands and they were swept from the surface.

We also examined non-CF airways and removed basolateral HCO₃⁻/CO₂ to prevent HCO₃⁻ secretion and added bumetanide to prevent Cl⁻ secretion. These interventions reduce the pH and increase the protein concentration of submucosal gland mucus, reproducing CF-like conditions. With disrupted Cl⁻ and HCO₃⁻ secretion, mucus strands failed to break after emerging from submucosal gland ducts, and they accumulated on the apical surface (38±4% of grids, n=33) (Fig. 7B,S7H–S7L). Like we found with CF mucus strands, raising apical pH to 8.3, reducing apical [Ca²⁺] to approximately zero, or both failed to break mucus strands (S7D,S7H-S7L). However, 10 mM DTT broke the mucus strands and they were cleared from the surface.

DISCUSSION

Mucus cells secrete mucus into the lumen of submucosal glands located in cartilaginous airways. Under basal conditions, a small amount of mucus flows up submucosal gland ducts and is released onto the airway surface (Fischer et al., 2019; Joo et al., 2002). When challenged by particulates, toxic conditions, infection or inflammation, neuronal and humoral signals induce submucosal glands to secrete profuse amounts of mucus (Widdicombe and Wine, 2015). Thus, they serve an important protective function in clearing material from the lung. Earlier studies revealed substantial CFTR expression and function in submucosal gland serous cells suggesting that loss of CFTR would disrupt gland function and thereby contribute to CF lung disease (Engelhardt et al., 1992; Lee and Foskett, 2010a). Our findings further emphasize the importance of submucosal glands in the pathogenesis of CF lung disease, highlighting two defects in submucosal glands caused by loss of CFTR: decreased HCO₃⁻ secretion produces an abnormally acidic pH, and decreased Cl⁻ secretion decreases liquid secretion thereby increasing protein concentration.

Induced release of mucus from isolated A. columbianus mucus vesicles allowed us to control the solution composition and volume at the moment the mucus gel was formed, and we found that an acidic pH, increased protein concentration, and Ca²⁺ reduced the volume and increased the elasticity and viscosity of the mucus gel. Elasticity was the dominate property. Use of a microfluidic model of submucosal glands allowed us to generate mucus strands and further revealed the importance of an acidic pH and Ca²⁺. However, after the gel was formed, the same factors that had influenced elasticity, tensile strength and volume at the moment of formation, then had little effect on the biophysical properties. These results plus our subsequent studies of mucus on porcine airways explain the finding that mucus strands secreted from CF submucosal glands disrupt MCT, an important host defense. This defect plus reduced antibacterial activity due to the decreased pH (Abou Alaiwa et al., 2014; Pezzulo et al., 2012) generate an environment for bacterial seeding and initiate a cascade of infection, inflammation, airway remodeling, and respiratory failure. Thus, loss of CFTR subverts the normally protective effect of stimulating submucosal gland secretion.

Previous studies have suggested that pH, Ca²⁺, and protein concentration can influence the formation of mucus gels. For example, relative to non-CF, the reduced pH of CF airway surface liquid was necessary and sufficient to increase the viscosity of airway surface liquid on cultured human and porcine CF airway epithelia and collected from newborn CF pigs (Tang et al., 2016). Gastric mucin forms a gel at low pH values, and dynamic light scattering suggested that acidic pH induces an extended conformation that exposes and facilitates interactions between hydrophobic sites (Cao et al., 1999). Electrostatic and hydrogen bond interactions may modulate gastric mucus biophysical properties in response to acidic pH (Bansil and Turner, 2006; Bhaskar et al., 1991; Bromberg and Barr, 2000; Hughes et al., 2019). Previous studies also indicate that Ca²⁺ may contribute intra- and inter-molecular bridges, thereby strengthening mucin structure. For example, increased [Ca²⁺] and decreased pH reduced swelling of cervical mucus (Espinosa et al., 2002). Previous studies of salivary mucus showed that Ca²⁺ allows interactions between mucin molecules (Raynal et al., 2003). Studies of MUC5B and recombinant N-terminal fragments revealed a Ca²⁺-binding site in the D3-domain of MUC5B and found reversible interactions at low pH and high Ca²⁺ (Ridley et al., 2014). An increase concentration of mucus can also increase viscosity and elasticity (Hill et al., 2014).

Once mucus strands and gels were formed, their volume, elasticity, viscosity, and tensile strength largely resisted changes in response to conditions that determined those properties at the time of formation - pH, Ca²⁺, and protein concentration. Stabilization of mucus strands likely involves multiple interacting factors. End-to-end linking of mucin molecules by disulfide bonds is key for forming long chains, and preventing those linkages with reducing agents disrupted the long strands (see also (Fischer et al., 2019)). The entanglement of the sugar side chains and protein may further contribute to the biophysical properties of mucus. Other factors that stabilize mucus strands likely include a complex mix of hydrophobic interactions, electrostatic forces, hydrogen bonds, and entanglement of the glycan-rich protein chain.

Advantages and limitations of this study

This work has a number of advantages. a) We studied porcine airways, which closely resemble those of human airways, including the submucosal glands that have similar numbers, size, and development (Rogers et al., 2008a), and CF pigs replicate the features of human CF (Ermund et al., 2018; Ostedgaard et al., 2011; Rogers et al., 2008b; Stoltz et al., 2010). b) Studying mucus vesicles from banana slugs allowed precise control of liquid volume and composition at the moment mucus is released from vesicles. Development of a microfluidic model to study slug mucus allowed us to test mucus behavior under defined conditions that resemble those of submucosal glands. c) The analysis of native unprocessed mucus avoids the secondary effects of purification and processing of mucus. d) Studying mucus from slugs and mucus immediately after release from airways of newborn pigs allowed us to evaluate mucus strands, i.e., the shape of mucus emerging onto the airway surface, while eliminating potential confounding factors associated with inflammation, infection, and proteases.

Studying mucus from slugs and mucus immediately after release from airways of newborn pigs allowed us to evaluate mucus strands, i.e., the shape of mucus emerging onto the airway surface, while eliminating potential confounding factors associated with inflammation, infection, and proteases.

This study also has limitations. a) Slug mucins and mucus differ from those of humans and pigs. Mucins are the main structural component of mucus (Fahy and Dickey, 2010; Lillehoj et al., 2013; Thornton et al., 2008; Voynow and Rubin, 2009), and earlier analyses of evolutionary relationships revealed substantial predicted amino acid sequence similarity across species including humans and mollusks (Aplysia californica) (Lang et al., 2016). Mucins are also heavily glycosylated proteins, and slugs and pigs will have different patterns of glycosylation, which could affect biophysical properties. However, the behaviors we studied yielded concordant and complimentary results. b) For studies in pigs, we focused on mucus produced by submucosal glands, because they produce much of the mucus in proximal airways, and their secretions disrupt MCT in CF (Ermund et al., 2018; Hoegger et al., 2014; Ostedgaard et al., 2017). However, goblet cells and secretory cells in surface epithelia also produce mucus. We expect that changes in pH, and protein concentration will produce similar effects on the biophysical properties of that mucus. c) Additional studies beyond our work with newborn pigs would be informative. For example, it would be important to learn how increased age, airway wall remodeling, and disease affect mucus biophysical properties. It would be interesting to learn how modulators of mutant CFTR affect submucosal gland mucus. It would also be valuable to repeat these studies with human tissue. d) We studied mucus, which is a mixture of many proteins (Fahy and Dickey, 2010; Voynow and Rubin, 2009). This has the advantage of in vivo relevance, and it avoids purification and separation that could change properties. Although mucins are the main structural proteins of mucus, the contribution of other proteins remains uncertain.

Implications for CF

The development of new medicines that increase the function of mutant CFTR has remarkably improved lung function for people with CF (Accurso et al., 2010; Clancy et al., 2019; Middleton et al., 2019). However, approximately 10 percent of people with CF have CFTR mutations that are not targeted by current medicines or adverse effects of the medicines prevent their use. Moreover, current medications do not completely normalize lung function, or eliminate inflammation and infection. Thus, additional treatments will be of value. Our results have implications for therapeutic strategies for two general approaches, modification of airway mucus and modification of cell function.

Finding that mucus strands have altered biophysical properties suggests that pharmacologically changing those properties might be of value. For example, agents that disrupt disulfide bonds might allow abnormally anchored mucus to break free and carry bacteria and particulates up the airways. That would be consistent with our findings with DTT. Another example might be hypertonic saline (7% NaCl), which is currently administered to people with CF to reduce the frequency of respiratory exacerbations (Donaldson et al., 2006; Elkins et al., 2006). The increased NaCl concentration might alter electrostatic interactions that are important for the integrity of mucus strands. The same may be true of agents that interfere with glycan:glycan interactions (Smith et al., 2017). Based on our observations, increasing airway surface liquid pH or reducing the Ca²⁺ concentration might not be particularly effective at altering submucosal gland mucus strands. Consistent with that suggestion, giving an aerosol of the Ca²⁺ chelator EDTA for 3 months to people with CF did not improve pulmonary function (Brown et al., 1985). Of course, these considerations do not exclude the possibility that such interventions might affect mucus produced by surface epithelial cells. In addition, interventions such as increasing airway surface liquid pH might have other beneficial effects such as enhancing activity of antimicrobial proteins (Pezzulo et al., 2012).

The contribution of submucosal glands to CF lung disease raises the question of whether it will be important for new therapeutics to improve function of submucosal gland epithelia? Current effective medicines that target mutant CFTR do not answer this question because they are delivered systemically and thus improve CFTR function throughout the body, including in submucosal gland cells. However, some potential new therapeutic strategies could be delivered through the airway lumen and target airway surface epithelia but might not reach submucosal gland epithelia. Examples are gene transfer (Boyd et al., 2020; Yan et al., 2019), gene editing (Hodges and Conlon, 2019; Mention et al., 2019), activation of Ca²⁺-activated Cl⁻ channels (Danahay et al., 2020), and delivering small molecule anion channels (Muraglia et al., 2019). It is possible that such interventions might improve some but not all host defense defects; future studies will be required to understand their benefit for airway disease.

STAR METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Michael J. Welsh (michael-welsh@uiowa.edu).

Materials Availability

This study did not generate unique reagents.

Data and Code Availability

This study did not generate datasets.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Sus scrofa

We previously reported the generation of CFTR^−/− pigs (Rogers et al., 2008b). CFTR^−/− (CF) and CFTR^+/+ (non-CF) pigs were the products of CFTR^+/− matings. Pigs were obtained from Exemplar Genetics and studied 8 to 15 hours after birth. All experiments used an approximately equal distribution of male and female newborn pigs. Animals were maintained in accordance with University of Iowa Animal Care and Use Committee guidelines. Anesthesia was with ketamine (20 mg/kg, I.M.) (Phoenix Pharmaceutical, Inc., USA) and xylazine (2 mg/kg, I.M.) (Lloyd, USA). Euthanasia was with intravenous Euthasol (Virbac, France). For ex vivo studies, a tracheal segment between the opening of the right cranial lobe and the larynx was cut with a surgical blade between cartilage rings and removed from piglets. The tracheal segments were immediately placed in the Krebs-Ringer saline and stored at 4 °C for 1–12 hr before experiments. The University of Iowa Animal Care and Use Committee (IACUC) approved all animal studies.

Ariolimax columbianus

Large terrestrial banana slugs native to the Pacific northwest region of North America, were purchased from Niles Biological, Inc. They were housed and raised in the laboratory at 13 °C.

METHOD DETAILS

Solutions

For most ex vivo studies on pig trachea, we used a HCO₃⁻/CO₂ buffered Krebs-Ringer saline containing (in mM): 118.9 NaCl, 25 NaHCO₃, 10 dextrose, 2.4 K₂HPO₄, 0.6 KH₂PO₄, 1.2 CaCl₂, 1.2 MgCl₂, pH=7.4 with 5% CO₂. For some experiments, we used HCO₃⁻-free HEPES buffered saline containing (in mM): 135 NaCl, 5 HEPES, 10 dextrose, 2.4 K₂HPO₄, 0.6 KH₂PO₄, 1.2 CaCl₂, 1.2 MgCl₂, pH =7.3. In some preparations, Cl transport was inhibited with 10 μM bumetanide, which inhibits NKCC-mediated entry of Cl⁻ across the basolateral membrane.

For experiments with slug mucus, we designed a series of solutions with different levels of pH and Ca²⁺ concentrations. They include pH 5 Ca²⁺ (50 mM HEPES, 10 mM CaCl₂), pH 7 Ca²⁺ (50 mM HEPES, 10 mM CaCl₂), pH 9 Ca²⁺ (50 mM HEPES, 10 mM CaCl₂), and pH 5 Ca²⁺-free (50 mM HEPES, 10 mM EGTA), pH 7 Ca²⁺-free (50 mM HEPES, 10 mM EGTA), and pH 9 Ca²⁺-free (50 mM HEPES, 10 mM EGTA). Immediately before use, the pH of solutions was adjusted with either 1N HCl or 3N NaOH and the osmolarity was adjusted to 210 mOSM with 1N NaCl.

Measuring the volume, pH and protein concentration of mucus secreted from SMGs

The apical surface of a trachea segment (length of ~1 cm) was cleaned with cotton tips, and air-dried. The trachea was pinned on a 3D printed device and further mounted on a perfusion chamber (PDMI-2, Harvard Apparatus, USA). The apical surface of trachea was covered with approximately 50 μL water-saturated mineral oil (Sigma Aldrich, USA). The basolateral side of trachea was perfused with Krebs-Ringer with 5% CO₂ at pH=7.4, or HEPES buffer at pH=7.4 through the built-in channels in the 3D printed device. In some experiments, methacholine (2.5 mg/mL), and bumetanide (10 μM) were added in the basolateral solutions. Perfusion flow rate was maintained at 0.1 mL/min with syringe pumps (neMESYS Syringe Pumps, Cetoni GmbH, Germany). The system was maintained at 37 °C. Mucus secretion from submucosal glands occurred immediately after the start of basolateral perfusion. Mucus production lasted for 30 min prior to collection of mucus and analysis of volume, pH, and protein concentration.

To measure the volume of mucus secreted from individual SMGs, a photograph of mucus droplets on trachea surface was taken by a digital camera (DFC 7000T, Leica, Germany) mounted on a surgery microscope (M205 FA, Leica, Germany). Diameters of 50 mucus droplets were measured with ImageJ software, the volume of each droplet was calculated by:

$$V=\frac{1}{6}\pi d^3,$$

where d and V are the diameter and volume of each mucus droplet, respectively.

To measure the pH of mucus, 2 μL BCECF acid (Thermo Fisher Scientific, USA) with a concentration of 0.25 mg/mL in ethanol was picked up with a glass capillary of inner diameter 0.5 mm (Wiretrol 5–000-1001, Drummond, USA). A uniform layer of BCECF was coated on the inner wall of the glass capillary by air-drying the ethanol solution. The coated glass capillaries were kept in the dark environment before experiments. Mucus droplets were collected with a nano-injector (Nanoject 2, Drummond, USA) mounted on a micro-manipulator (ROE-200, Sutter Instrument Company, USA). Mucus and mineral oil were then transferred to the BCECF-coated capillaries and sealed with tube sealant (Critoseal, Leica, Germany). Mucus and oil were separated in the capillary with a hematocrit centrifuge (CT-3400, Adams Readacrit, USA). After 5 min centrifugation, mucus was layered under mineral oil due to their density differences (Fig. S1A). The pH of mucus in the capillary was assessed by ratiometric measuring BCECF fluorescence at excitation wavelengths of 440 and 490 nm and an emission wavelength of 535 nm (Fig. S1B). Fluorescent images were captured with an inverted microscope (Eclipse TE200, Nikon, Japan) and an EMCCD camera (Evolve 512, Photometrics, USA) and analyzed with ImageJ software. A standard curve with PBS buffer at pH 6, 6.5, 7, 7.5, and 8 in capillary was used to calibrate pH of mucus (Fig. S1C). Three technical replicates for each individual trachea were averaged.

To measure the protein concentration of mucus, the mucus droplets were collected, transferred to a non-coated glass capillary, sealed with tube sealant, and centrifuged to isolate from mineral oil (Fig. S1D). The volume of mucus collected in the capillary was calculated based on its length occupied in capillary. Mucus, together with mineral and sealant, were placed in a 96-well plate with 150 μL CBQCA Protein Quantitation Kit (Thermo Fisher Scientific, USA) (Fig. S1E). After incubating for 1h, the protein concentration in each well was accessed by fluorescent measurement at excitation of 490 nm and emission of 535 nm. Because transferring mucus from capillaries to 96-well plates diluted the protein, the original protein concentration was calculated by the volume of solution in the well and mucus in capillaries. Bovine serum albumin (BSA) solution with varying concentrations was used as a standard (Fig. S1F). Negative control experiments were conducted with pure mineral oil and tube sealant. Three technical replicates on each trachea were averaged.

Measuring mucus strand clearance at different apical/basolateral conditions with a dual chamber assay

To separate the apical and basolateral side of the tracheas, a dual chamber was developed. A trachea segment (length of ~1 cm) removed from newborn pigs was opened with a longitudinal cut on the ventral aspect and inserted between the two halves of the dual chamber and different buffers or solutions were perfused into the top and bottom chambers according to the aim of the experiment.

Mucus strands were imaged with fluorescent nanospheres. Yellow-green fluorescent carboxylate-modified nanospheres (20 nm diameter, FluoSpheres, Molecular Probes, USA) were sonicated in saline and added at a final dilution of 1:10,000. The top chamber was filled with 200 uL fluorescent nanosphere-containing solution and the bottom chamber was filled with 1 mL solution containing 1.28×10⁻⁵ mol/L methacholine. Fluorescent nanospheres were visualized using 10x dry objective on an upright Nikon A1R confocal microscope. Experiments were conducted in a temperature-controlled environment (32–35 °C).

Time-averaged panoramic imaging was conducted to quantify stationary mucus strands. To visualize the entire tracheal surface in the chamber, the confocal microscope was synchronized to a stage controller (MS-2000–500, Applied Scientific Instrumentation, USA) that moved the microscope stage to different positions on the tracheal preparations and panoramic images were generated. We used 3 fields × 3 fields to visualize the entire tracheal surface. To preferentially image stationary mucus, we applied a time-averaging technique. For each microscopic field, a series of 32 images were acquired over the course of 4 sec. The average pixel fluorescence was computed for the 32 images and a single time-averaged microscopic field was generated. With this time-averaging technique, the signal from stationary fluorescent entities was similar to that in a single image. However, the signal of fluorescent entities in motion was markedly diminished in the time-averaged microscopic field. This time averaging technique combined with panoramic image acquisition allowed preferential visualization of stationary fluorescent entities on entire tracheal segments. Imaging was started 30 min after the start of methacholine (MCH) stimulation (t=0). Perfusion (flow rate 100 μL/min) was performed using a syringe pump from t=5 min until the end of the experiment. Multiple panoramic images were recorded during a 25 min imaging period.

To quantify mucus strands, we created a scoring system. Images at specific time points (0, 5, 10, 15, 20, and 25 min) were printed with an overlying grid consisting of 1 cm × 1 cm squares (Fig. S7). The grid covered the surface of the airways but excluded the edges to avoid mucus attached to and accumulated at the edges of the chamber; such mucus was readily distinguishable and was not scored. Squares with readily discernable mucus strands were counted. Two individuals scored each grid. Scorers were blinded to conditions and times. The data are presented as % of squares with strands.

Measuring expansion of slug mucus from granules

Mucus vesicles were collected by flushing the slug body with 1 mL slug Ringer solution (10 mM HEPES, 50 mM NaCl, pH 7, osmolarity 210 mOsm) into an Eppendorf tube. The collected mucus vesicles were diluted 100 times with experimental solutions into a 35 mm² glass-bottom dish. We added FM4–64, which stains the mucus vesicle membranes, and green fluorescence nanospheres (1:1000 v/v, F8787, Invitrogen, USA), which allowed visualization of individual mucus expansion upon rupture of the mucus vesicle membrane and release of the mucus (Video S1). Mucus expansion was observed and imaged with a Z-stack confocal image (with interval 1 micron) (LSM 880, Zeiss, Germany) 30 min after vesicle rupture.

To test modification of the volume of mucus gels, we perfused individual mucus gels with 50 mL solution with varied pH (5, 7, or 9), and/or [Ca²⁺] (10 mM, or ~0 by adding 10 mM EGTA) and recorded the volumes of individual mucus gels 30 min after perfusion.

Imaris software was used to construct and render the surface of the 3D structure of the expanded mucus from a series of Z-stack microscope images. After rendering the surface of the expanded mucus, the surface volume was calculated as the mucus expansion volume. The investigator that performed the experiments and analyzed the data was blinded to conditions.

Measuring the elastic and viscous moduli of slug mucus gel

Slug were euthanized, the slug trunk was shocked with 5 V to stimulate mucus secretion, and mucus vesicles were collected with slug Ringer solution. To investigate effects of pH and [Ca²⁺] on the elastic and viscous moduli of mucus, 500 μL of vesicle solution was introduced to 500 μL HEPES buffer with varied pH (5, 7, or 9) and [Ca²⁺] (10 mM, or ~0 by adding 10 mM EGTA) to form the working vesicle solutions. Vesicles were ruptured by addition of 10% (v/v) of nystatin stock (10 mg/mL) introduced into the mucus vesicle solutions or sonicating the samples for 10 min to form a homogenous mucus gel. Gels were maintained for 1 hr at room temperature before measurement.

To investigate the impact of protein concentration on elastic and viscous moduli at mucus formation, 10, 100, 500, 900 μL mucus vesicle solution was diluted into HEPES buffer with pH 5 Ca²⁺ (10 mM) and pH 9 EGTA solutions to a final volume of 1 mL. After mucus vesicles are ruptured by ultrasound, the elastic and viscous modulus are measured as above. The mucus protein concentration was determined by BCA assay (Pierce™ BCA Protein Assay Kit, ThermoFisher Scientific, USA). 10 μL mucus vesicles were immediately transferred and mixed with working reagent following the protocol. After 30 min incubation at 37 °C, the absorbance was measured at 562 nm with a microplate spectrophotometer.

To investigate effects of pH and Ca²⁺ after mucus formation, we formed two mucus gels under the same conditions. Then, we introduced solutions with different pH and/or Ca²⁺; as a control, the same volume of solution without a change in pH or [Ca²⁺] was introduced. Mucus gels were maintained for 30 min at room temperature before measuring elastic and viscous moduli with a rheometer.

To investigate effects of protein concentration, [protein], after mucus formation, we formed three mucus gels with relative protein concentration of 1x, 1.8x (volume of 1 mL) and of 1x gel (volume of 1.8 mL) in pH 5 with 10 mM Ca²⁺ solution. Then the gel with relative [protein] of 1x and volume of 1.8ml was evaporated to 1 mL of volume and reached its relative [protein] to 1.8x. Their elastic and viscous moduli were measured with a rheometer.

The elastic and viscous moduli were measured by cone and plate oscillation rheometer (Haake Rheostress 1 Rheometer; Thermo Scientific, USA). The 35 mm/4° cone and 35 mm measurement plate were used to measure 1 mL mucus gels that were formed in different pH and Ca²⁺ solutions. The elastic modulus and viscose modulus were measured in consistent stress mode with shear stress value τ=1 changing with ω from 0.5 to 105 rad/s (frequencies from 0.079 to 16.7 Hz) (Fig. S4). To best estimate the elasticity and viscosity of the gel (Lai et al., 2009), we reported all the elasticity and/or viscosity as their linear viscoelastic moduli, which are G’ (f=0.206 Hz, ω=1.29) and G” (f=0.206 Hz, ω=1.29) respectively.

A microfluidic engineered SMG model to measure stretch and breakage of mucus strands

The microfluidic model of SMG includes: a substrate to generate ultrasound as a surface acoustic wave (i.e. SAW substrate) (Ozcelik et al., 2018); a PDMS-based microfluidic channel; and the bonding of the PDMS microchannel to the substrate. To fabricate the SAW substrate, chrome-plated gold (Cr/Au, 5 nm/ 50 nm) was deposited on a photoresist-patterned (SPR-3012, MicroChem, USA) 128° Y-cut lithium niobate (LiNbO3) wafer (500 μm thick), double-side polished (PWLN-431232, Precision micro-optics, USA), followed by a lift-off protocol to form the pair of interdigital transducers. The PDMS based microfluidic channel was fabricated with a standard soft-lithography and mold replica protocol (Xia and Whitesides, 1998) using photoresist (SPR-955, MicroChem, USA). Holes were drilled in the PDMS microfluidic channel for inlets and outlets with a 1.0-mm punch (Harris Uni-Core, USA). To bond the PDMS microfluidic channel with a LiNbO₃-based SAW substrate, both were treated with oxygen plasma in a plasma cleaner (Harrick Plasma, USA) for 3 min. The PDMS microchannel was aligned and bonded to the SAW substrate between the transducers. The whole device was cured at 65 °C for at least three days before experiments, to avoid leakage between channel and SAW substrate. Experiments were conducted on the stage of an inverted microscope (Eclipse TE200, Nikon, USA). To eliminate the virtual image from the double-side polished LiNbO3 substrate, a polarizer was placed in the light path and adjusted to a requisite angle. Sample solutions were prepared in 1 and 3 mL plastic syringes (Becton, Dickinson and Company, USA), and they were injected through polyethylene tubing (Becton, Dickinson and Company, USA) into the microchannel with syringe pumps (neMESYS, Cetoni GmbH, Germany). RF functions were generated (E4422B, Agilent, USA) at desired frequencies and were amplified (100A250A, Amplifier Research, USA) to generate SAW. In experiments, a frequency of 140.5 MHz was used. An EM-CCD camera (Evolve 512, Photometrics, USA) was positioned facing the XY plane of the device. To record the stretch and break of mucus strands, red fluorescent nanospheres (20 nm diameter, EX580/EM605, Carboxylate, Life technology, USA) was used to show the outline of mucus strand; green fluorescent microspheres (1 μm diameter, EX480/EM520, Carboxylate, Bangs Laboratory, USA) were used to trace the stretch of mucus strands.

Strain of mucus strands was defined as the ratio of deformation to the initial length of a mucus strand under a certain flow rate. In experiments, green fluorescent microspheres (1 μm diameter) were used to label mucus strand (Fig. S5). After displacement of microspheres was recorded, strain from multiple microspheres was averaged as the strain of mucus strand, which was calculated by:

$${\mathit{\text{strain}}}_{Q2}=\frac{1}{n}\sum _{i=1}^n\left(\frac{y_{i,Q2}-y_{i,Q1}}{y_{i,Q1}-y_p}\right)$$

where Q2 is the flow rate to stretch mucus strand, Q1 is the initial flow rate; n is the number of microspheres attached to a mucus strand; in each experiment, positions of 5 microspheres were calculated to obtain an average strain (n=5). The strain was calculated at flow rates ranging from 10–250 μL/min, because we realized once beyond this flow range, the mucus strand begins to break. Therefore, calculated strain in this flow range characterized the elastic properties of mucus strand.

Breakage of mucus strands was represented by the percentage of a mucus strand that remained at defined flow rates. As an example, at flow rate Q3, the break of a mucus strand was calculated by:

$$\mathit{\text{Microspheres remaining }}\%=\frac{i}{n}\times 100\%$$

Breakage was assessed from the lowest flow rate (10 μL/min) to the highest flow rate (1000 μL/min).

Scanning electron microscopy

Samples of pig trachea and slug foot were fixed for 2 h in perfluorocarbon (FC-72, 3M, USA) containing 2% osmium tetroxide, then dehydrated in three changes of 100% ethanol over 3 h. Samples were then transitioned to hexamethyldisilizane and air-dried overnight. Tissues were mounted on aluminum stubs, sputter coated with 80:20 gold:palladium, and imaged with a scanning electron microscopy (Hitachi S-4800 FE-SEM, Hitachi High Technologies America, Inc., USA).

Transmission electron microscopy

Samples of piglet trachea and slug foot were rinsed with PBS and fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate. Then, samples were post-fixed in 2% osmium tetroxide, en bloc-stained with 2.5% uranyl acetate, dehydrated through a graded ethanol series, infiltrated with Eponate 12 (Ted Pella), and polymerized at 60 °C for 24 h. Sections with thickness of 70 nm were cut with a ultramicrotome (Leica EM UC6, Leica Microsystems, Germany) and counterstained with 5% uranyl acetate and Reynold’s lead citrate before examination on a transmission electron microscope (JEM1230, JEOL Ltd., Japan). Images were captured with a CCD camera (2k × 2k, Gatan Inc., USA).

Lectin stain of mucus in vitro from slugs and pigs

The slug mucus was collected after the slug crawled onto a glass slide (Superfrost Plus Slides, Thermo Fisher Scientific, USA). The pig mucus was collected from trachea lumen after in vivo methacholine treatment, and transferred to a glass slide. The lectins, WGA-rhodamine (Vector Laboratories, USA) and Jacalin-FITC (Vector Laboratories, USA) at 1:1000 dilution, were introduced to stain the mucus for 1 hour. The mucus samples were rinsed by PBS before coverslipped. Images ware captured by a confocal microscopy (Fluoview FV3000, Olympus Life Science, USA).

QUANTIFICATION AND STATISTICAL ANALYSIS

Most data are represented with mean±SEM unless otherwise stated. P value was determined using one-way ANOVA, paired t-test, or un-paired t-test as appropriate unless otherwise stated. Statistical analyses were performed in GraphPad Prism.

Supplementary Material

2

Video S1. Mucus gel expansion from a single mucus vesicle, Related to Figure 3.

3

Video S2. Acoustic effect in a microfluidic channel, Related to Figure 5.

4

Video S3. Stretch and breakage of a mucus strand in a microfluidic submucosal gland, Related to Figure 5.

KEY RESOURCE TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, Peptides, and Recombinant Proteins
FM4–64 Dye (N-(3-Triethylammoniuumpropyl)-4-(6-(4-(Diethylamino) Phenyl) Hexatrienyl) Pyridinium Dirbromide)	Thermo Fisher	Cat# T13320
FluoSpheres™ carboxylate-modified microspheres: Yellow/Green (505/515); 20nm; 2% solid	Thermo Fisher	Cat# F8787
WGA: Rhodamine labeled Wheat Germ Agglutinin (WGA)	Vector Laboratories	Cat# RL-1022
Jacalin: Fluorescein labeled Jacalin	Vector Laboratories	Cat# FL-1151U
Methacholine chloride	Sigma	CAS# 62–51-1
Forskolin	Sigma	CAS# 66575–29-9
3-Isobutyl-1-methylxanthine (IBMX)	Sigma	CAS# 28822–58-4
Bumetanide	Sigma	CAS# 28395–03-1
Critical Commercial Assays
BCA assay: PierceTM BCA Protein Assay Kit	Thermo Fisher	Cat# 23225
CBQCA Protein Quantitation Kit	Thermo Fisher Scientific	C6667
Experimental Models: Organisms/Strains
Wildtype pigs	Exemplar Genetics
CFTR−/− pigs	Exemplar Genetics
Banana slug: Ariolimax Columbianus	Niles Biological, Inc.	SKU: SLUGB E
Software and Algorithms
Inkscape	https://inkscape.org/	Version 0.92.4
ImageJ	https://imagej.nih.gov/ij/	Version 1.8.0 172
GraphPad Prism 8	https://www.graphpad.com/	Version 8.1.2
NIS-Element Ar	Nikon Instruments Inc.

Highlights.

• Cystic fibrosis submucosal glands produce abnormally acidic and concentrated mucus

• A microfluidic model using banana slug mucus produced strands of mucus

• Acidic, concentrated mucus formed strands with high elasticity and tensile strength

• After mucus strands formed, their biophysical properties resisted further change