Airway Surface Liquid

In this issue of The Journal, Jayaraman and colleagues 2001 describe a novel technique to measure the osmolality of airway surface liquid (ASL) using fluorophore-encapsulated liposomes. As in a recent paper reporting measurements of ASL depth, salt concentration, and pH (Jayaraman et al. 2001), this group has once again exploited fluorescence microscopy in an innovative fashion to avoid the shortcomings of the traditional methods of harvesting ASL and analyzing its composition. As reported, ASL osmolality in the mouse was found to be 330 ± 36 mOsM (comparable to that of plasma), and no difference was found between wild-type and CF knockout mice. In addition to providing the beginning of a clear picture of ASL composition under normal conditions, these studies may have important implications for the pathogenesis of cystic fibrosis (CF) lung disease.

The use of fluorescent techniques to infer the composition of ASL is a tremendous leap forward in technology, avoiding for the first time the pitfalls associated with techniques based on harvesting ASL or on in situ measurements. The mainstay of ASL studies over the years has been the application of filter paper to the surface of the epithelium (Boucher et al. 1981; Joris et al. 1993; Knowles et al. 1997). This technique depends on the capillarity generated by the paper fibers, which act as a myriad of high energy capillaries in parallel. The filter paper technique is quite efficient at collecting ASL. Indeed, it may be too effective, generating a sufficiently high driving force to result in the movement of macromolecules from the submucosal surface to the airway lumen (Erjefält and Persson 1990). Measurements of ASL harvested with this technique have consistently shown [K⁺] values substantially above plasma, suggesting the possibility of epithelial damage. Indeed, in our hands, ASL harvested using polyethylene capillaries (Cowley et al. 1997, Cowley et al. 2000), which generate much less force than filter paper, consistently yield [K⁺] lower than those found with filter paper. Nevertheless, the polyethylene approach is likely also to generate artefacts because it involves direct contact of the sampling probe with the epithelium. In addition, because of the hydrophobicity of polyethylene, this technique may bias the collection to samples of ASL with a surface tension below the critical surface tension of polyethylene (Landry et al. 2000). Other approaches, such as salt-sensitive electrodes (Caldwell et al. 2000) and electron probe X-ray crystallography (Baconnais et al. 1998), similarly suffer from potential artefacts related to distortion of the epithelium or the ASL itself.

The elegant but simple system of osmotically sensitive liposomes reported in the present study seems to be well suited to measurements of ASL osmolality. The fluorophore used is stable, and the liposomes are sufficiently small so as not to disturb the ASL while being able to distribute homogeneously though it. The authors have demonstrated that the method is sufficiently sensitive to detect the effects of evaporation or addition of salt on ASL osmolality. The results appear to be both reproducible and consistent in cell culture and in vivo models. A key advantage over previous techniques based on harvesting ASL, salt-sensitive electrodes, and studies of rapidly frozen airways, is the ability to conveniently follow changes in ASL osmolality over time. For the first time, the rate at which intact epithelium responds to stresses like changes in osmolality can be measured directly, thereby providing important insights into the dynamic properties of the airway epithelium. Jayaraman and co-workers (2001) have exploited this capability to estimate the permeability of epithelial cells to water, yielding an independent estimate of epithelial water permeability that is in general agreement with previous results (Matsui et al. 2000). The ability to track the time course of epithelial responses to stimuli is an important strength of this approach, and we look forward to its application in a variety of disease models in which epithelial regulation of ASL composition and volume may be important.

The results of greatest interest in this study are those carried out in vivo. The authors have taken advantage of the small size of mice to extend fluorescence microscopy to the airway epithelium of intact animals, which is an important advance. Nevertheless, results from these studies must be interpreted cautiously. The fluorescent techniques developed in the Verkman lab have been described as a noninvasive approach to measure ASL composition in mice. Certainly this approach perturbs the epithelium less than techniques involving direct contact with the epithelial surface; however, it is probably better characterized as minimally invasive rather than noninvasive. The current method involves a midline incision in the neck, isolation of the trachea, as well as the oral insertion of a tracheal microcatheter for instillation of liposomes. One possible consequence of these manipulations is the triggering of an epithelial leak from the submucosa, leading to the introduction of plasmalike material into the ASL. This possibility was effectively excluded by experiments in which a small window in the trachea was created and no discernible effects on the results were detected. Similarly, the authors report that no leak was detected after injection of FITC-labeled macromolecules, which has been used previously to detect leak in the airways of allergen-challenged animals (Erjefält and Persson 1990). Another source of error in these studies is the possible contribution of glandular secretion. In their previous report, Jayaraman et al. 2000 described detecting evidence of gland secretions, measured as changes in ASL depth, in fewer mice treated with atropine than in untreated mice (10 vs. 30%). Although this is reassuring, it does not exclude the possible contribution of gland secretion at rates below the detection threshold of the ASL depth measurements, or that may have occurred before the administration of the liposomes. Also, the administration of atropine most likely does not block all gland secretion. However, the stability of the ASL depth measurements over time does make the possibility of the ASL composition being altered by bulk flow from glands less likely. Despite these caveats, the results reported by Jayaraman et al. 2001 are an important advance and raise interesting questions about current models of ASL regulation.

Although there has been some interest in the composition and regulation of ASL in other diseases (Wong et al. 1999; Anderson and Daviskas 2000), the driving force behind the study of ASL has been the attempt at understanding the pathogenesis of CF lung disease. Although more than a decade has elapsed since the discovery of CFTR (Riordan et al. 1989), the precise mechanisms by which its absence leads to chronic infection, bronchiectasis, and respiratory failure remain unclear. CFTR is a member of the ATP binding cassette family of proteins, and exhibits at least two of the properties characteristic of this family. CFTR is an anion channel that can switch between conducting and nonconducting conformations using the energy of ATP hydrolysis. In its conducting configuration, CFTR promotes the movement of anions both into and out of cells. CFTR can also regulate other membrane proteins, particularly the epithelial Na⁺ channel (ENaC), a process that requires the phosphorylation of CFTR (Stutts et al. 1995). Despite the functional characterization of CFTR, there is no consensus regarding how loss of function leads to lung disease. In recent years, two hypotheses have come to dominate the debate, each makes specific predictions concerning the regulation and composition of ASL. The osmolality data presented by Jayaraman and colleagues directly addresses this debate, appearing to rule out one hypothesis and raising questions about the other.

The first hypothesis, termed the “isosmotic volume hypothesis” (Matsui et al. 2000) is based on the notion that the respiratory epithelium, which is absorptive under normal circumstances, becomes hyper-reabsorptive in CF (Knowles et al. 1983; Matsui et al. 1998). It is proposed that this hyper-reabsorption leads to a reduction in volume of ASL on the epithelial surface, with a consequent increase in ASL viscosity and a reduction in the effectiveness of mucociliary clearance. This model emphasizes the importance of CFTR as a regulatory molecule, specifically as an inhibitor of ENaC (Stutts et al. 1995). Absence of CFTR is thought to permit excessive ENaC activity leading to excessive absorption of Na⁺ and hyper-reabsorption of ASL. The ultimate consequence is a decrease in ASL depth and an increase in its viscosity, preventing cilia from effectively clearing foreign particles (Matsui et al. 1998), thereby predisposing to infection and the eventual development of chronic lung disease.

The alternate hypothesis emphasizes the role of ASL composition in the passive defense against bacterial infection. Smith and colleagues 1996 observed in cultured cells, that ASL possesses antibiotic properties that are present only when the salt concentration of ASL is well below that of plasma. In their hands, the absence of CFTR leads to elevation of ASL salt concentrations, thereby inhibiting the antibacterial properties of ASL. Several compounds present in ASL, including defensins (Singh et al. 1998) and cathelicidins (Bals et al. 1998), have been proposed to account for the salt-sensitive antibiotic actions of ASL. This theory is supported by several studies in animals (Cowley et al. 1997, Cowley et al. 2000; Baconnais et al. 1998) and in humans (Joris et al. 1993; Knowles et al. 1997), reporting ASL salt concentrations in the airway below those of plasma. Furthermore, in at least one study (Joris et al. 1993) the [Na⁺] of CF ASL was significantly higher than that of normal subjects (121 ± 3.5 vs. 82 ± 6 mM). Conceptually, the high salt hypothesis emphasizes the role of CFTR as an apical chloride channel. In contrast to the volume hypothesis, it is proposed that Cl⁻ moves from the airway lumen into the airway epithelial cell such that under normal conditions, an excess of Na⁺ and Cl⁻ is reabsorbed leaving behind hypotonic ASL, or at least ASL with a salinity lower than plasma. In CF, the absence of CFTR leads to elevation of the ASL [Na⁺] because Na⁺ absorption is decreased due to poor epithelial Cl⁻ permeability in a manner analogous to the sweat duct epithelium (Joris et al. 1993). The higher salt concentration in CF ASL would block its antibiotic function, thereby predisposing CF patients to chronic infection in the airways.

Recently, the high salt hypothesis has been challenged on several levels. Despite the consistent finding of low salt concentrations in ASL under normal circumstances, the finding of higher [Na⁺] in the ASL of CF patients (Joris et al. 1993) has not been replicated in vivo in patients (Knowles et al. 1997) or in mice (Cowley et al. 1997; Jayaraman et al. 2001). The most extensive study of human ASL composition to date found both salt concentrations and osmolality to be similar in CF and normal subjects (Knowles et al. 1997). This hypothesis has been criticized on theoretical grounds as well. For the salt concentrations of ASL to be low, reabsorption of water isotonically with salt must be prevented somehow. This could happen if the airway epithelium was relatively impermeable to water. Measurements of water permeability in the present paper (Jayaraman et al. 2001) as well as other recent reports (Matsui et al. 2000) indicate that the airway epithelium is quite permeable to water, making this mechanism unlikely. Alternatively, some force could hold water in the lumen against the osmotic gradient created by hypertonic reabsorption of NaCl (Zabner et al. 1998). It has been proposed that the capillarity of the cilia, or perhaps surface tension at the tips of the cilia, could act to limit water absorption. The presence of a mucus layer in ASL prevents the establishment of the needed air–liquid interface, making this an unlikely mechanism (Matsui et al. 1998). It is also possible that unmeasured osmotic particles allow ASL to be isotonic with plasma despite a low [NaCl] (Zabner et al. 1998). No evidence of these missing osmotically active molecules so far has been presented, and direct measurements of ASL osmolality in the past (Willumsen et al. 1994; Folkesson et al. 1996; Matsui et al. 2000) and in the present study (Jayaraman et al. 2001) have suggested that Na⁺, Cl⁻, and K⁺ are likely sufficient to account for ASL osmolality.

Although the above suggests that the high salt theory is on the ropes, does this mean that the low volume hypothesis is now confirmed? The work of Jayaraman and colleagues raises challenges for the low volume hypothesis as well. Although the new finding of near isotonic osmolality of ASL in both CF knockout mice and controls, as well as the previous observation that ASL [Na⁺], [Cl⁻], and pH are similar in normal and CF (Jayaraman et al. 2001) support the low volume hypothesis, the failure to detect a difference in ASL thickness argues against it. The central prediction of the low volume hypothesis is that ASL depth is abnormally low, resulting in disordered ciliary function (Matsui et al. 1998). Although fluorescence techniques are capable of detecting decreases in ASL depth caused by evaporation or the addition of salt (Jayaraman et al. 2001), no decreases in ASL depth were found in CF knockout mice in vivo. Furthermore, ASL thickness in cultured epithelial systems was found by Jayaraman to be substantially less than in vivo, casting doubt on the physiological relevance of the cell culture data that has been extensively used to support both the low volume and high salt hypotheses. The fluorescently measured ASL depth reported here is in good agreement with previous measurements using other techniques (Matsui et al. 1998; Wu et al. 1998) as well as with theoretical predictions (Widdicombe 1997).

If the findings of Jayaraman are inconsistent with the two most popular hypotheses to explain CF lung disease, what alternatives are suggested by these data? First, it is important to stress that the in vivo fluorescence studies so far have been limited to mice. It remains possible that conditions in human airways are different, although the results using bronchial segments (Jayaraman et al. 2001) do not support this notion. It is also possible that the physiology of the respiratory epithelium is different in the lung periphery, where glands are less prominent and the epithelial structure is different (Ranga and Kleinerman 1978). No doubt these arguments will be made by those defending both the low volume and the high salt theories. Nevertheless, the present findings are compelling and call out for alternative hypotheses.

One option is to revisit the importance of CFTR in airway secretion, rather than absorption (Wine 1999), and consider the idea that the submucosal glands are central to CF pathogenesis. In the postnatal lung, CFTR expression is highest in the serous cells of the submucosal glands (Engelhardt et al. 1992). Submucosal gland serous cells are a potentially important source of antibiotic compounds and CFTR is critical for glandular secretion (Trout et al. 1998). There is also evidence that secretions of submucosal glands contribute to ASL volume and composition. For example, in a recent study of the ferret tracheal model (Wang et al. 2001), the presence of submucosal glands was associated with more abundant ASL volume and a higher [Cl⁻], which is consistent with the high expression of CFTR in the submucosal glands. Absence of CFTR is likely to decrease the volume of glandular secretions and might well affect their viscosity (Inglis et al. 1997). Although the results by Jayaraman et al. 2001 and others (Knowles et al. 1997) suggest that gland dysfunction in CF has little impact on basal ASL volume and composition, they do not exclude the possibility that, when stimulated, the glands are an important part of host defense. Abnormal submucosal gland secretions in CF could predispose to infection through a variety of mechanisms including high viscosity, which may only be evident under conditions when the glands are stimulated. Intriguingly, the Verkman group recently presented preliminary fluorescence data regarding murine ASL viscosity measured in vivo (Jayaraman et al. 2000), suggesting that more information about the viscosity of gland secretions in CF may soon be forthcoming. The advent of fluorescence microscopy techniques for the investigation of ASL composition and regulation may take us full circle, back to the classical model of CF, a disease of abnormal glandular secretions.