cystic fibrosis transmembrane conductance regulator
Na+ bicarbonate cotransporter
Ca2+-activated chloride channel
Cervical mucus thinning and release during the female reproductive cycle is thought to rely mainly on fluid secretion. However, we now find that mucus released from the murine reproductive tract critically depends upon concurrent bicarbonate (HCO3−) secretion. Prostaglandin E2 (PGE2)- and carbachol-stimulated mucus release was severely inhibited in the absence of serosal HCO3−, HCO3− transport, or functional cystic fibrosis transmembrane conductance regulator (CFTR). In contrast to mucus release, PGE2- and carbachol-stimulated fluid secretion was not dependent on bicarbonate or on CFTR, but was completely blocked by niflumic acid. We found stimulated mucus release was severely impaired in the cystic fibrosis ΔF508 reproductive tract, even though stimulated fluid secretion was preserved. Thus, CFTR mutations and/or poor bicarbonate secretion may be associated with reduced female fertility associated with abnormal mucus and specifically, may account for the increased viscosity and lack of cyclical changes in cervical mucus long noted in women with cystic fibrosis.
Changes in cervical mucus properties and bicarbonate (HCO3−) secretion within the female reproductive tract are critically linked to fertility. A HCO3−-rich alkaline pH environment is crucial for optimal sperm motility (Muschat, 1926) and capacitation (Wang et al. 2003; Chan et al. 2006, 2009). Both HCO3− and mucus change dramatically throughout the menstrual cycle, with concentrations of HCO3− varying from 35 mm at the follicular phase to at least 90 mm at ovulation (Maas et al. 1977), which corresponds to the minimum in mucus viscosity (Blair et al. 1941).
Bicarbonate douching is reported to improve cervical mucus viscoelasticity and improve sperm penetration (Ansari et al. 1980; Everhardt et al. 1990), and this suggests that HCO3− itself may have an important influence on the properties of mucus. The abnormalities of generalized thick mucus and defects in HCO3− secretion in the genetic disease cystic fibrosis (CF) also suggest that HCO3− may be a critical determinant of mucus properties. More specifically, CF is characterized by pathologies resulting from mucus obstructions in almost all affected organs including the pancreas (Farber et al. 1943; Zuelzer & Newton, 1949), small intestine (Eggermont, 1996), hepatobiliary tree (Bhaskar et al. 1998), small airways (Burgel et al. 2007) and salivary and most other exocrine glands (Gugler et al. 1967; Oppenheimer & Esterly, 1975). The sweat gland, which secretes virtually no mucus, is an exception (Johansen et al. 1968; Quinton, 1999). Along with a loss of Cl− conductance (Quinton, 1983), mutations of the cystic fibrosis transmembrane conductance regulator (CFTR) Cl− channel protein also impede HCO3− transport, the extent of which appears to correlate with the severity of the CF phenotype (Kopelman et al. 1989; Choi et al. 2001; Quinton, 2001; Reddy & Quinton, 2003).
Mucus pathology and reproductive physiology intersect in CF women where cervical mucus thinning is absent (Kopito et al. 1973a), fertility is reduced (Hilman et al. 1996; Edenborough et al. 2000) and cervical mucus plugs may develop (Oppenheimer et al. 1970). HCO3− transport is characteristically impaired in CF affected organs (Kopelman et al. 1988; Smith & Welsh, 1992; Seidler et al. 1997; Clarke & Harline, 1998; Pratha et al. 2000; Choi et al. 2001; Quinton, 2001; Ishiguro et al. 2009); consequently, it seems highly possible that the thick and tenacious cervical mucus of women with CF involves impaired HCO3− secretion. We therefore investigated the effect of altering bicarbonate and fluid secretion on stimulated mucus release in the reproductive tracts of wild type (WT) and homozygous ΔF508 CF female mice ex vivo. We describe a new role for HCO3− in mucus release, where it seems likely to be critical for normal mucin expansion and transport.
The University of California San Diego institutional animal care and use committee approved all procedures used in this study. All our studies comply with the policies and regulations of The Journal of Physiology (Drummond, 2009).
WT adult C57BL/6 mice were either purchased from Harlan Laboratories Inc. or taken from our own breeding colony and maintained on standard laboratory chow. The mice were allowed free access to food and water until surgery. The ΔF508 mice were obtained from Case Western Reserve University and were generated by targeted replacement of the WT exon 10 allele with the ΔF508 mutant allele (Zeiher et al. 1995). To increase survival, the ΔF508 mice were maintained on an osmotic laxative containing electrolytes and polyethylene glycol 3350 (GoLYTELY; Braintree Laboratories, Inc., Braintree, MA, USA) administered ad libitum in the drinking water (Clarke et al. 1996) and a liquid diet of Peptamen Af (Nestlé HealthCare Nutrition, Minnetonka, MN, USA). Mice were used at the oestrous stage of the reproductive cycle, which was determined by the presence of round nucleated epithelial cells, cornified cells and leukocytes obtained from a wet vaginal smear.
The mice were anaesthetized with ketamine (100 mg kg−1) and xylazine (10 mg kg−1) administered subcutaneously. Once the hindlimb flexor withdrawal reflex ceased, the reproductive tract was excised intact, and the animals were killed immediately by cervical dislocation.
All chemicals and drugs used were purchased from Sigma-Aldrich (St Louis, MO, USA) with the exception of GlyH-101 (Muanprasat et al. 2004) and MalH-1 (Sonawane et al. 2006), which were gifts from N. D. Sonawane and A. S. Verkman (University of California, San Francisco) and human MUC5B antibodies, which were a gift from J. K. Sheehan and M. Kesimer (University of North Carolina, Chapel Hill). MUC5B-N-terminal and C-terminal antibodies were polyclonal (anti-rabbit), while MUC5B-central domain antibody was monoclonal (anti-mouse). The human MUC5B antibodies appear to react with mouse Muc5b.
All drugs were dissolved in dimethyl sulphoxide (DMSO) with the exception of prostaglandin E2 (PGE2), which was dissolved in ethyl alcohol, and carbachol, which was dissolved in water. PGE2 was used at 10−6 m (Fong & Chan, 1998) and carbachol was used at 10−5 m for a maximal secretory response (Hammarstrom, 1980).
The basolateral phosphate-buffered Ringer solution contained (in mm): 150 Na+, 5 K+, 1 Ca2+, 1 Mg2+, 150 Cl−, 2.5 PO4x− and 10 d-glucose. The luminal perfusion solution was a glucose free Ringer solution of the same composition except that mannitol replaced glucose. The HCO3−-buffered Ringer solution contained (in mm): 150 Na+, 5 K+, 1 Ca2+, 1 Mg2+, 125 Cl−, 25 HCO3−, 2.5 PO4x− and 10 d-glucose. For fluid secretion measurements a luminal sodium-free solution was used with the following composition: 150 N-methyl-d-glucamine (NMDG), 5 K+, 1 Ca2+, 1 Mg2+, 125 Cl−, 2.5 PO4x− and 10 d-glucose. All solutions were adjusted to pH 7.4 and equilibrated with either 100% O2 (phosphate-buffered Ringer solution) or 95% O2–5% CO2 (HCO3−-buffered Ringer solution) and maintained at 36 ± 1°C during all experiments.
Tissue preparation and mucus collection
The excised reproductive tract was placed immediately in chilled Ringer solution containing indomethacin (10−5 m) to minimize endogenous prostaglandin release during fine dissection and manoeuvres. The reproductive tracts were mounted vertically with the uterine horn oriented toward the bottom in a custom-built constant temperature chamber. The perfusion chamber was a smaller version of that previously described (Garcia et al. 2009), adapted by using a 3 ml plastic syringe as the perfusion chamber sealed within a 20 ml plastic syringe with rubber gaskets at each end to form a water jacket (Fig. 1). The top of one uterine horn was ligated to the fire polished end of a glass capillary (0.8 mm bore Ø× 8 cm length) and the lumen of the whole tract flushed with glucose free Ringer solution. The end of the other horn was ligated closed. The vaginal end of the reproductive tract was then attached to the polished end of another glass capillary (1.2 mm Ø× 3.5 cm length). The lower glass capillary (uterine horn end) was connected to a variable speed peristaltic fluid pump by Silastic (Dow-Corning) tubing, and the lumen perfused from bottom to top with glucose-free, phosphate-buffered Ringer solution at a rate of approximately 0.4 ml min−1. The upper capillary was connected to Silastic tubing elevated to maintain a slight hydrostatic pressure to ensure that the lumen remained patent and directed into a sample collection tube. After mounting, the reproductive tract was bathed in phosphate-buffered Ringer solution for 10 min and then switched as needed to a bath of HCO3−-buffered Ringer solution for the remainder of the experiment. The lumen was perfused continuously with glucose/HCO3− free Ringer solution. Samples were collected at 5 min intervals during a pre-stimulation period of 25 min as well as for the duration of the experiment after stimulation.
The amount of mucus released was analysed using peroxidase conjugated wheat germ agglutinin-lectin (Triticum vulgare) (lectin) binding of samples loaded onto a polyvinylidene fluoride membrane using a dot blot apparatus. In brief, 200 μl of perfusate samples along with known mucus standards were loaded in triplicate onto the membrane. The membrane was incubated in 3% albumin from bovine serum in Tris-buffered saline (TBS) to block non-specific binding for 2 h, followed by a 2 h incubation in lectin (5 μg ml−1 in TBS) at room temperature. For the MUC5B antibody assay, the membrane was incubated in the primary antibody overnight at 4°C (1 in 500 dilution), followed by a 1 h incubation in the secondary antibody (1 in 1000 dilution) at room temperature. The membrane was then washed for 30 min in TBS and incubated in 3,3′-diaminobenzidine enhanced liquid substrate for 5 min. For the periodic acid-Schiff (PAS) assay, the membrane was incubated for 10 min in 0.5% periodic acid, followed by 30 min incubation in Schiff's reagent. The intensity of the lectin, PAS and antibody binding was quantified using the density analysis function of Adobe Photoshop.
The mean amount of mucus released after stimulation was normalized as a percentage increase over the basal mucus release (the average value of mucus released 10 min prior to stimulation taken as 100%). See Fig. 2A for a representative trace of mucus release before normalization.
Histology and immunostaining
The horns from freshly isolated uteri were separated for use as matched pairs. Each horn was stimulated with PGE2 and carbachol for 20 min, one horn in the presence of HCO3− and the other in its absence. The tissue was fixed immediately in methanol–Carnoy's fixative (Johansson et al. 2008). Paraffin-embedded sections were dewaxed and hydrated. Antigens were retrieved using 10 mm sodium citrate, (pH 6). Sections of 5 μm were stained with PAS/hematoxylin or with an antibody to the central domain of MUC5B labelled with a TRITC-conjugated anti-mouse secondary antibody (sigma). DAPI (Roche) was used to counterstain nuclei. Micrographs were obtained with a spinning disc confocal (BD Carv II) microscope (Zeiss, Axiovert 200).
Fluid secretion measurement
Fluid secreted from the reproductive tract was measured by cannulating a uterine horn of the excised tissue with a fire polished glass capillary (0.28 mm Ø× 3 cm length). The lumen was gently flushed using sodium-free Ringer solution to diminish sodium-dependent fluid absorption. The contralateral horn and the vaginal canal were ligated closed to form a cannulated sac isolated from the external solution. After equilibrating, fluid secreted into the lumen displaced fluid into the capillary cannula. To avoid changes in volume due to spontaneous smooth muscle contractions, nifedipine (10−6 m) was applied prior to mounting in the basolateral solution to inhibit muscle activity for the duration of the measurement. The rate of fluid secretion was measured by monitoring the displacement of the air/liquid meniscus in the cannulating capillary under a low power microscope. The rate of secretion was determined from the volume of fluid, calculated from the diameter of the capillary multiplied by the displacement of the meniscus in the capillary, divided by the collection interval in minutes.
Statistical analysis of the data was obtained using repeated measures two-way ANOVA with a Bonferroni post hoc test for analysis of mucus and fluid released under different conditions over a period of time. Data are presented as means ± standard error of the mean, n is the number of experiments (one animal per experiment), and P < 0.05 was taken as significantly different.
Stimulating mucus release
Both the cholinergic agonist carbachol and the prostaglandin PGE2 stimulate cervical mucus release (Hammarstrom, 1980; Duchens et al. 1993). Initial experiments established that both PGE2 (10−6 m) (maximal mucus release after 5 min = 230 ± 38%; n= 5) and carbachol (10−5 m) (289 ± 116%; n= 5) stimulated an acute mucus release within 5 min of applying the agonists (Fig. 2B and C). The agonists appeared to act synergistically when added together (800 ± 239%; n= 6) (Fig. 2B and C). Since maximal mucus release occurred when both agonists were added simultaneously, tissues were stimulated with both carbachol and PGE2 in all subsequent protocols. After the initial peak, the rate of mucus release returned to modestly above basal, but continuously decreased to pre-stimulus levels (Fig. 2A and B).
To confirm that the lectin used for assays was binding predominantly to mucins secreted in the reproductive tract, we compared the binding of lectin, PAS and three MUC5B antibodies in the samples collected after stimulation with PGE2 and Carbachol. In assays of the same samples, the amount of lectin binding correlated well with both the intensity of PAS staining and the amount of MUC5B antibody binding (Fig. 2D). We took these results as confirmatory evidence that lectin binding gives a good measure of mucus released in this preparation. Since the assay is efficient and economical, we used lectin binding to determine the amounts of mucus released in the samples of all collected perfusates.
Mucus release requires HCO3− and CFTR
Addition of PGE2 and carbachol stimulated mucus release into the perfusate by about 8-fold over baseline (Fig. 2A). In the absence of HCO3− in the bathing medium (Fig. 3A) or in the presence of 4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid (DIDS; 10−4 m) (Fig. 3B) to inhibit the Na+/HCO3− cotransporter (NBC) (Romero et al. 1997), stimulated mucus release did not increase significantly above basal rates (104 ± 15%, n= 4 and 108 ± 30%, n= 4, respectively).
While CFTR is best known for its role in Cl− transport in fluid secretion and salt absorption, it is also well known to support HCO3− secretion (Clarke & Harline, 1998; Choi et al. 2001; Welsh & Smith, 2001; Ishiguro et al. 2009). To determine if CFTR might be required for HCO3− secreted during mucus release, we applied GlyH-101 (2 × 10−5 m) and MalH-1 (10−5 m) to inhibit CFTR (Sonawane et al. 2006). The inhibitors were combined and added 15 min prior to stimulation to both the basolateral and the luminal solutions. Since MalH-1 is cell impermeant and unlikely to access the recessed gland lumens, it seems most likely that most of the effect would be due to the permeant GlyH-101 form of the CFTR inhibitor. Nonetheless, for efficiency, we added MalH-1 to the luminal bath to inhibit accessible surfaces and achieve maximal effect.
Stimulated mucus release was inhibited (99 ± 29%; n= 4; P < 0.001) (Fig. 3C) suggesting CFTR is involved in mucus release. However, as the inhibitors may affect other Cl− channels (Muanprasat et al. 2004; Caputo et al. 2008), we tested mucus release in homozygous CF ΔF508 mice where mutant CFTR is dysfunctional. Stimulated mucus release was also significantly depressed even though the tissues were bathed in HCO3− replete Ringer solution (106 ± 50%; n= 4; P < 0.001) (Fig. 3C and D).
Does mucus release require fluid secretion?
Mucus release is often, if not always, accompanied by fluid secretion (Nadel et al. 1979; Joo et al. 2001). Thus, simultaneous addition of carbachol and PGE2 may stimulate mucus release that is dependent on fluid secretion. To show that fluid secretion and mucus release occur concurrently, we monitored fluid secretion from native, cannulated reproductive tracts. Spontaneous fluid secretion was measured for 25 min prior to addition of PGE2 and carbachol and ranged between 0.30 and 1.15 μl min−1 g−1, 5 min prior to stimulation. PGE2 (4.5 ± 1.9 μl min−1 g−1; n= 7) and carbachol (12.2 ± 4.9 μl min−1 g−1; n= 5) each stimulated fluid secretion but with a maximal effect when added simultaneously (17.9 ± 3.6 μl min−1 g−1; n= 10) (Fig. 4A). The stimulated fluid secretory response coincided closely with that of mucus released as measured in other tissues (compare Fig. 4A to Fig. 2).
To determine whether mucus release was dependent on fluid secretion, we measured stimulated mucus release in the presence of bumetanide, which inhibits fluid secretion by virtue of its selective inhibition of the Na+–K+–2Cl− cotransporter (NKCC). Bumetanide (10−4 m) added to the bath solution almost completely blocked fluid secretion (2.1 ± 1.0 μl min−1 g−1; n= 4; P < 0.001) (Fig. 4B and C) and also prevented stimulated mucus release into the perfusate (95 ± 31%; n= 4) (Fig. 4D).
Is fluid secretion HCO3− dependent?
Since blocking fluid secretion clearly impaired mucus release (Fig. 4D), and since HCO3− may enhance fluid transport (Cremaschi et al. 1979; Heintze et al. 1981), we sought to determine whether the HCO3−-dependent mucus release observed here might be due to a reduced HCO3−-dependent fluid secretion. However, we found that in contrast to mucus release, PGE2 and carbachol-stimulated fluid secretion was not significantly different in the presence or absence of HCO3− (17.5 ± 6.0 μl min−1 g−1; n= 4; P= 0.84) (Fig. 5A and C) or in the presence of basolateral DIDS (12.1 ± 0.86 μl min−1 g−1; n= 4; P= 0.27) (Fig. 5B and C).
Fluid secretion does not require CFTR
We also found that mucus release was severely impaired both in tissues from homozygous ΔF508 mice, and in tissues from WT mice whose CFTR function was inhibited. Since some forms of fluid secretion depend on CFTR (Sato & Sato, 1984; Ianowski et al. 2008), these results raised the question of whether the impaired mucus release is due to a loss of CFTR-dependent fluid secretion. We found that fluid secretion stimulated in the presence of GlyH-101 and MalH-1 was not significantly reduced (21.7 ± 5.1 μl min−1 g−1; n= 6; P= 0.68) (Fig. 6A and C). Corroborating these results, PGE2 and carbachol-stimulated fluid secretion in the ΔF508 mouse reproductive tract was not significantly different from fluid secretion in WT tissue either (16.3 ± 4.3 μl min−1 g−1; n= 4; P= 0.69) (Fig. 6A and C). On the other hand, niflumic acid (10−4 m) effectively inhibited fluid secretion (0.24 ± 0.12 μl min−1 g−1; n= 4; P < 0.001) (Fig. 6B and C). Although the simplest interpretation of this niflumic acid effect is that fluid secretion is mediated through a Ca2+-activated Cl− channel (CaCC) (White & Aylwin, 1990), niflumic acid also affects other Cl− transporters such as Cl−/HCO3− exchangers (Cousin & Motais, 1979) and CFTR (Scott-Ward et al. 2004). However the lack of effect of GlyH-101, which is reported to at least partially block CaCC at high concentrations (Caputo et al. 2008), raises the possibility that an alternative Cl− channel may support fluid secretion under these conditions.
Mucus is retained in the cervico-uterine glands in the absence of HCO3−
Tissue was prepared for histological analysis in an attempt to determine the deposition of mucus after stimulation. PAS and MUC5B antibody micrographs of PGE2 and carbachol stimulated tissue show that in the presence of HCO3−, uterine glands appear to retain only relatively small amounts of mucus (Fig. 7A and B), whereas in the absence of HCO3−, the uterine glands were largely filled with a PAS positive substance that also correlated with MUC5B labelling (Fig. 7C and D).
Our working hypothesis is that HCO3− is critical for mucus release in the female reproductive tract. A role of HCO3− in the release and expansion of mucins into the lumens of exocrine organs has recently been proposed (Quinton, 2008; Garcia et al. 2009). The present study demonstrates that HCO3− is critical for mucus release in the female reproductive tract, where HCO3− and mucus normally change concurrently, with a peak in HCO3− secretion corresponding with a decrease in mucus viscosity (Blandau et al. 1958; Kopito et al. 1973b; Maas et al. 1977). The results presented here in addition to those observed in the intestine (Garcia et al. 2009) support the notion that the hypothesis may hold for gel forming mucin release in general, and not necessarily be a tissue specific phenomenon.
Implications for fertility
Cervical mucus volume in the female reproductive tract increases dramatically during the menstrual cycle with maximal amounts seen at ovulation (Long & Evans, 1922). An increase in the water content is thought to be the principal factor contributing to cyclical changes in cervical mucus properties (Katz et al. 1997; Kopito et al. 1973b). Biochemical changes in cervical mucins, such as dramatic changes in O-glycosylation at ovulation may also contribute to the changes in the physical properties of cervical mucins, as well as promote sperm penetration due to low sialic acid content (Andersch-Bjorkman et al. 2007).
Bicarbonate has also been shown to be critical for improving mucus viscoelastic properties (Everhardt et al. 1990), and our results here demonstrate its importance for normal mucin release. Specifically, a loss of bicarbonate secretion may underlie an absence of cyclical cervical mucus thinning in CF women at ovulation. In addition to a loss of sperm capacitation in the absence of HCO3− (Wang et al. 2003), mucus that retains increased viscosity throughout the reproductive cycle is likely to be a significant impediment to sperm mobility (Steinberg, 1955; Kopito et al. 1973a; Sher & Katz, 1976; Scott et al. 1977).
Although the factors involved in infertility in CF women are often multifactorial, (Johannesson et al. 1998; Timmreck et al. 2003; Wang et al. 2003; Hodges et al. 2008), abnormal mucus has long been associated with infertility (Scott et al. 1977; Daunter & Khoo, 1984; Punnonen et al. 1984). Therefore the effects of poor HCO3− secretion on mucus release may be an unappreciated factor in low fertility associated with abnormal mucus in general and specifically in women with CF (Oppenheimer et al. 1970). In view of the fact that Young's syndrome and conditions of congenital bilateral absence of the vas deferens (CBAVD) are associated with mutations in the CFTR gene (Chillon et al. 1995), it is tempting to consider that some cases of reduced fertility in females might be associated with putative mild mutations in this gene with a consequence of abnormal cervico-uterine mucus release due to inadequate HCO3− secretion. An example of this might be the case of two infertile sisters with significantly abnormal cervical mucus who were found to be compound heterozygote carriers of the cystic fibrosis ΔF508 and R117H/7T mutations (Schoyer et al. 2008).
Lectin-binding assays mucus
Due to the complex biophysical and biochemical nature of mucus, routine quantification can require extensive sample preparation and time consuming techniques. Lectin binding assays have been developed as an alternative efficient method to assay for these large molecules as a general class (Rhodes & Ching, 1993). Lectins are carbohydrate-binding proteins that bind selectively and non-covalently to a specific sugar sequence in oligosaccharides and glycoconjugates. Analysis of the O-linked carbohydrate chains in purified cervical secretions collected midcycle revealed high levels of galactose, N-acetylglucosamine, fucose and neuraminic acid (Yurewicz & Moghissi, 1981). We therefore used wheat germ agglutinin lectin in this study, which is specific for N-acetylglucosamine and neuraminic acid (Gravel, 2002). We demonstrated that lectin binding correlates well with the binding of three different MUC5B antibodies and to PAS stain, a well established assay for polysaccharides (Kirkeby et al. 1992) (Fig. 2D). The binding of other lectins such as peanut agglutinin lectin and Erythrinal cristagalli lectin to cervical secretions is also known to correlate with binding of MUC5B in cervical mucus (Argueso et al. 2002). Therefore, although not a direct measure of mucus secretion, the use of the lectin-binding assay to semi-quantitatively determine the amount of mucus released from the reproductive tract seems well justified.
Independent impacts of HCO3− and fluid secretion
Neither the changes in viscoelasticity of mucus in the normal menstrual cycle nor the cause of the thick aggregated mucus associated with CF are well explained. Both have been commonly attributed to mucus hydration, i.e. water content. The concept of dehydrated mucus due to hyperabsorption through the epithelial sodium channel (ENaC) (Boucher, 2007) has been widely accepted for CF airways; however, hyperabsorption does not seem to explain mucus aggregation in organs where there are no known mechanisms of fluid absorption, such as the pancreas (Novak & Hansen, 2002).
The data presented here show that stimulated mucus release in the female reproductive tract is highly dependent on HCO3− secretion. In the CF ΔF508 reproductive tract, stimulated mucus release was severely impaired in the presence of basolateral HCO3−, suggesting transport of HCO3− into the lumen and consequently mucus release depends on CFTR. It is feasible that addition of HCO3− to the luminal side would rescue mucus release; however, the complex morphology of the reproductive tract would likely prevent HCO3− from reaching the site of mucus release by diffusion. In addition, the secretory flow out of the gland would likely prevent HCO3− from interacting with mucins as they are secreted, and in a similar manner to that in the small intestine (Garcia et al. 2009) in sufficient to rescue mucus release.
Although fluid secretion is clearly important for mucus release, fluid secretion in the CF ΔF508 reproductive tract was preserved, whereas mucus release was severely impaired. Moreover, removing HCO3− did not significantly impair fluid secretion in WT mice, indicating that the reduction in stimulated mucus release is not due to impaired HCO3− dependent fluid secretion. These results support the notion that the loss of CFTR-dependent HCO3− secretion per se is the principal factor contributing to impaired mucus release rather than to the simple loss of cAMP mediated CFTR-dependent fluid secretion or to the loss of HCO3−-dependent fluid secretion. It is noteworthy that these results are not tissue specific and that similar results were recently reported in the small intestine (Garcia et al. 2009). HCO3− in the reproductive tract appears to thin mucus (Ansari et al. 1980; Everhardt et al. 1990) and in CF the loss of HCO3− secretion appears to coincide with abnormally thick mucus in affected organs. Together, these observations implicate a crucial role for HCO3− in releasing gel forming mucins in general, but how?
Proposed impact of bicarbonate on mucus
Mucins, the essential components of mucus, are extremely large, negatively charged linear polymers of glycoprotein molecules that are packaged in a highly condensed state in intracellular mucus granules. As proposed by Verdugo et al. (1987), in order for these enormous molecules with their dense population of fixed negative sites to be stored in a highly condensed form, the repulsive forces between the anionic charges must be ‘shielded’ electrostatically by cations, namely high concentrations of intragranular Ca2+ and H+. However, once released from the cell, the shielding cations must be extracted to allow the repulsive electronegative forces to expand the polymer rapidly (decondensation) (Verdugo et al. 1987) into an extracellular network of ‘tangled strings’ (Lee et al. 1977). In the process the molecular volume increases by as much as 1000-fold in less than a couple of seconds (Verdugo, 1990). Recently, we proposed that the presence of extracellular HCO3− (and CO32−), which readily complexes with Ca2+ and protons, appears to be critical for extracting the ‘cationic shields’ from mucin anions upon exocytosis of the granule (Fig. 8) (Quinton, 2008; Garcia et al. 2009). Once unshielded, the repulsive electrostatic forces of exposed anionic sites not only expand the mucin molecules, but also result in a very low coefficient of friction between the expanded gel mucin molecules and the intrinsic mucins of the apical membrane of epithelial cells (Pettersson & Dedinaite, 2008). We surmise that this loose association between mucins is required to efficiently promote their movement from within the crypts of the cervico-uterine epithelium, the lumens of the mucus glands, and into the cervico-uterine tract.
HCO3− seems most likely to be involved extracellularly, where exocytosed mucins would be immediately exposed to high concentrations of HCO3− as opposed to an intracellular role in processes that lead up to exocytosis. De Lisle suggested that the effect of HCO3− may arise intracellularly with HCO3− entering goblet cells from neighbouring cells through gap junctions and then into mucin granules (De Lisle, 2009) or alternatively by HCO3− uptake into mucus cells upon stimulation. A gap junction route of entry would require that HCO3− be directed passively through adjacent epithelial cells through gap junctions into the mucus cell. Such a mechanism would seem to limit the amount of HCO3− that could be rapidly delivered for mucus expansion and would seem to at least require a large accumulation of HCO3− in the adjacent cells and/or a significant depolarization of the mucus cell relative to the adjacent cells. However, bestrophin Cl− channels are highly permeable to HCO3− (Qu & Hartzell, 2008), and have recently been shown to be expressed in goblet cells (Yu et al. 2010), which potentially could provide entry to HCO3−. On the other hand, fluid-secreting epithelia can actively transport large quantities of HCO3− into luminal spaces into which mucins are released (Trout et al. 1998; Joo et al. 2001; Simpson et al. 2005; Kreindler et al. 2009). In this case, the amount of HCO3− delivery would be limited only by the number and secretory capacity of these cells forming the epithelial compartment, where it is conceivable that the bicarbonate concentration could be increased several fold above interstitial fluid levels. We note that goblet cells are generally single cells surrounded by putative fluid transporting epithelial cells.
Recent work on the mechanism by which MUC5B mucin is released from granules of salivary cells (Kesimer et al. 2009) suggests the process of mucin unravelling may occur in more than one phase, with the initial phase involving the removal of Ca2+ from the negatively charged oligosaccharides of the mucin core that permits initial expansion followed by a second phase that requires freeing the terminal, poorly glycosylated protein regions of the mucins from nodes. The protein regions appear to be more tightly bound by covalent bonds and/or possibly by strong Ca2+ binding. HCO3− (CO32−) could be crucial to this phase as well to (1) sequester tightly bound Ca2+ from protein binding sites in the binding nodes, (2) simply help initial expansion of the packed oligosaccharide cores sufficiently to allow access to the nodal binding sites for enzymatic cleavage that permits the mucin molecule to further unravel, (3) maintain an optimal pH for enzymatic cleavage of intramolecular bonds, or (4) serve as a possible cofactor in the process (Fig. 8) (Quinton, 2009).
Why is mucus release depressed in the absence of HCO3−?
Mucus release in the presence of HCO3− was 8-fold higher than in its absence. Since the reproductive tract was perfused continuously, why was mucus not released into the perfusate even if it remained poorly unexpanded in the absence of HCO3−? It is unlikely that the tissue was unresponsive under these conditions since under these same conditions (no HCO3−), a robust fluid secretory response was present, demonstrating that the tissue was both viable and responsive. Histology of the reproductive tract stimulated in the presence and absence of HCO3− demonstrated that without HCO3− mucus is stuck in the cervico-uterine glands, apparently unavailable for ready transport through the lumens of the tissue (Fig. 7). Thus, the luminal perfusate contained significantly less mucus when the tissue was deprived of secreted HCO3−. But why is the mucus trapped? Without HCO3−, it is possible that Ca2+ and H+ cations are not efficiently extracted and remain associated with mucin molecules. They may increase intra- as well as intermolecular interactions (due to divalent Ca2+ bridges, hydrogen bonds and electrostatic interactions). If these interactions involve molecules of the intrinsic mucin layer, which are tethered to the membrane lining the lumen of glands (Johansson et al. 2008), the complete release of the gel-forming mucins should be suppressed and their transport into the luminal perfusate impeded. Interestingly, CFTR is highly expressed in the uterine glands at the oestrous stage of the cycle (Chan et al. 2002), which is when maximum mucus release occurs. Our data demonstrate mucus release depends on CFTR, and in the absence of HCO3− mucus remains stuck in the glands; strongly suggesting a critical role for CFTR mediated HCO3− transport in mucus ‘transportability’.
The data presented here do not prove that a lack of bicarbonate prevents unravelling of mucin molecules, but the results reported here are consistent with results in the small intestine (Garcia et al. 2009), so that whatever the role of HCO3− in normal mucin release, it appears to be fundamental to the process and not tissue specific.
HCO3− is required for normal release of gel-forming mucins to form transportable mucus in the female reproductive tract. Poor HCO3− secretion seems likely to be a component in cases of low fertility as well as in reduced fertility in women with cystic fibrosis specifically. This novel role for HCO3− in mucus release should be relevant in other mucus producing organs as well, where gel forming mucins are crucial to numerous physiological functions.
P.Q. conceived and designed the study; R.M. performed the research at the University of California, San Diego and analysed the data; R.M. and P.Q. wrote and edited the manuscript. Experiments were performed at the University of California, San Diego.
This study was supported by grants from Cystic Fibrosis Research Inc., Nancy Olmsted Trust, Cystic Fibrosis Foundation and NIH USPHS R01-HL084042. We thank A. Verkman and N. Sonawane for the kind gift of the drugs GlyH-101 and MalH-1; J. Sheehan and M. Kesimer for the kind gift of the MUC5B antibodies; M. Drumm and A. Wilson for the CF ΔF508 mice; UCSD Moores cancer centre histology core for preparation of slides and G. Flores and K. Taylor for technical assistance.