Yet another role for the cystic fibrosis transmembrane conductance regulator.

In 1983, Paul Quinton showed that the chloride permeability of sweat duct epithelium was essentially abolished in cystic fibrosis (CF) (1). Shortly after, cyclic adenosine monophosphate (cAMP)–dependent transport of chloride across airway epithelia was also shown to be greatly reduced in CF, and the defect was localized to the apical membrane (2, 3). Therefore, it was to be anticipated that, within a couple of years of cloning the CF gene, its protein product (the cystic fibrosis transmembrane conductance regulator [CFTR]) was shown to be a cAMP-activated chloride channel (4). The situation is now more complex. First, CFTR itself has been proposed to transport water, urea, formate, adenosine triphosphate (ATP), glutathione, and organic anions in addition to halide ions (5–9). Second, CFTR has been shown to influence a large number of cell functions in addition to transepithelial transport of chloride ( see Table 1). Not all of the effects listed in Table 1 are generally accepted, and in some cases the initial report has not been confirmed by subsequent studies. For instance, the defective acidification hypothesis was highly appealing in that it could explain the pleiotropic effects of CF (10). According to the hypothesis, failure of CFTR in intracellular vesicles led to decreased acidification of their contents. This would clearly affect the posttranslational processing and trafficking of many proteins. But several later studies failed to show differences in intravesicular pH in CF (11). The key questions to the many proposed actions of CFTR are ( 1 ) whether the methods are adequate and ( 2 ) do the results have any relevance to the function of affected human epithelia in vivo. Specific methodologic problems include random variation among cell lines, use of nonpolarized cells to draw conclusions about epithelial function, overexpression of CFTR, inadequacies in the transgenic mouse model of CF, and use of nonspecific pharmacologic agents. Many studies have compared CF and non-CF cell lines, most of which are poor models of native epithelium because they do not form tight junctions and polarize. Realizing this, many workers supplement their results on cell lines with data from primary cell cultures. These can very closely resemble the native epithelium in both their structure and function (12), and CF and non-CF cultures can be closely matched for cell number, structure, transepithelial resistance, and other parameters. However, the primary cultures used are generally very poorly characterized. At minimum, investigators should check whether their primary cultures are polarized, but even this simple assay (for transepithelial resistance) is rarely performed. Transfection of cell lines and expression of exogenous CFTR is also commonly used. The most sophisticated application of this approach involves the “recovery” of CF cell lines by expression of wild-type CFTR. The problem, however, with expression experiments is that the expressed CFTR is usually present at levels far greater than found in native epithelia. Excess of any ion channel in the membrane might be expected to have multiple effects on function. It is also possible that CFTR trafficking pathways may become saturated and then CFTR trafficked to unusual locations. In retinal pigment epithelium, for instance, CFTR is normally trafficked to the apical membrane, but exogenously expressed CFTR appears in the basolateral (S. Miller, personal communication). Similarly, in rat Fischer thyroid cells, the relative amounts of exogenous CFTR passing to the apical and basolateral membranes depend on the level of expression (13). Furthermore, if trafficking pathways are overwhelmed with CFTR, then trafficking of proteins other than CFTR may be affected. To control for the effects of overexpression, it is usual to transfect control cells with an irrelevant reporter gene such as lacZ. Yet this is not an entirely adequate control because the reporter gene is usually not an integral membrane protein and uses trafficking pathways different from CFTR. The CF mouse is a powerful investigational tool (14), but it must be remembered that mouse airways and their epithelia differ greatly from those of humans. Mouse airway epithelium has comparatively large numbers of Caactivated chloride channels (15) and lacks mucous glands and goblet cells (16). In fact, ion transport by the tracheal epithelium of CF transgenic mice is indistinguishable from ( Received in original form October 5, 1999 )