cAMP- and Ca2+-independent activation of cystic fibrosis transmembrane conductance regulator channels by phenylimidazothiazole drugs.

Patch-clamp, iodide efflux, and biochemical techniques were used to evaluate the ability of phenylimidazothiazoles to open normal and mutated cystic fibrosis transmembrane conductance regulator (CFTR) chloride channels and to investigate the mechanism of activation. As reported previously for bromotetramisole, levamisole activated wild-type CFTR channels stably expressed in Chinese hamster ovary cells in the absence of other secretagogues and without elevating intracellular cAMP or calcium. The protein kinase A (PKA) inhibitor N - (2-(p-bromocinnamylamino)ethyl)-5-isoquinolinesul-fonamide abolished activation by forskolin but only partially inhibited stimulation by levamisole, suggesting the involvement of other kinases. CFTR channels bearing mutations at multiple phosphorylation sites, in the membrane domains, and in the first nucleotide binding domain (including the disease-causing mutations G551D and ΔF508) all responded to phenylimidazothiazoles. Moreover, levamisole and bromotetramisole increased the activity of wild-type and mutant channels already exposed to PKA + MgATP, consistent with the inhibition of a constitutive, membrane-associated phosphatase activity. We conclude that phenylimidazothiazole drugs can open normal and mutated CFTR channels by stabilization of phosphoforms of CFTR that are produced by basal activity of PKA and alternative protein kinases. If similar stimulation is observed in humans in vivo, phenylimidazothiazoles may be useful in the development of pharmacological therapies for cystic fibrosis.

Patch-clamp, iodide efflux, and biochemical techniques were used to evaluate the ability of phenylimidazothiazoles to open normal and mutated cystic fibrosis transmembrane conductance regulator (CFTR) chloride channels and to investigate the mechanism of activation. As reported previously for bromotetramisole, levamisole activated wild-type CFTR channels stably expressed in Chinese hamster ovary cells in the absence of other secretagogues and without elevating intracellular cAMP or calcium. The protein kinase A (PKA) inhibitor N -(2-(p-bromocinnamylamino)ethyl)-5-isoquinolinesulfonamide abolished activation by forskolin but only partially inhibited stimulation by levamisole, suggesting the involvement of other kinases. CFTR channels bearing mutations at multiple phosphorylation sites, in the membrane domains, and in the first nucleotide binding domain (including the disease-causing mutations G551D and ⌬F508) all responded to phenylimidazothiazoles. Moreover, levamisole and bromotetramisole increased the activity of wild-type and mutant channels already exposed to PKA ؉ MgATP, consistent with the inhibition of a constitutive, membrane-associated phosphatase activity. We conclude that phenylimidazothiazole drugs can open normal and mutated CFTR channels by stabilization of phosphoforms of CFTR that are produced by basal activity of PKA and alternative protein kinases. If similar stimulation is observed in humans in vivo, phenylimidazothiazoles may be useful in the development of pharmacological therapies for cystic fibrosis.
Cystic fibrosis is a disease that mainly affects epithelia and is caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) 1 gene (1)(2)(3). CFTR is a chloride channel with characteristics that are now well documented in epithelia (4 -7) and various heterologous expression systems (8 -10), although it may also have other functions. Cystic fibrosis epithelia from the airways, pancreatic ducts, intestine, and sweat glands have abnormal chloride transport that, depending on the particular mutation, may be caused by a defect in the targeting of CFTR to the plasma membrane or to channel dysfunction at the cell surface (11).
CFTR has two membrane domains, each comprised of six transmembrane spanning regions, two nucleotide-binding folds (NBF 1 and 2), and a large cytoplasmic regulatory (R) domain. Channel activity is controlled by cAMP-dependent phosphorylation of the R domain (12)(13)(14)(15) and nucleotide interactions at the nucleotide-binding folds (16,17). Protein kinase A (PKA) can phosphorylate CFTR at multiple sites, and some are more strongly phosphorylated in vivo than others; however, the precise functional role of each site has not yet been characterized. Alteration of all 10 dibasic consensus sequences and five additional monobasic sequences reduces but does not abolish activity (15,58). Moreover, there is some phosphorylation of CFTR even in unstimulated cells, presumably catalyzed by basal activity of PKA and other kinases (12)(13)(14)(15). CFTR is kept inactive under resting conditions by a membrane-associated phosphatase activity in transfected Chinese hamster ovary cells (CHO) (8), human pancreatic duct cells (7), and in human airway cell lines expressing endogenous or heterologous CFTR (18,19). It is clear, however, that other protein phosphatases play a role in cardiac (20) and sweat duct cells (21). Phosphatase inhibitors have been suggested as potential therapies for cystic fibrosis (18); however, the phosphatase that regulates CFTR in airway epithelia and CHO cells remains to be identified. It does not appear to be PP1, PP2A, or PP2B (8,18); therefore PP2C or an unidentified protein phosphatase with functional similarities to alkaline phosphatase are the most likely candidates. The role of PP2C, which was defined based on its magnesium dependence, has been difficult to establish definitively because there are no specific inhibitors of this enzyme. PP2C, alkaline phosphatase (23) and the CFTR channel itself (22) are all magnesium-dependent.
We recently obtained evidence that CFTR is associated with a membrane-bound phosphatase activity that controls spontaneous deactivation (rundown) of channel activity in excised patches (8,18) and the level of CFTR phosphorylation (18). Several phosphodiesterase and/or alkaline phosphatase inhibitors were found to slow rundown and inhibit CFTR protein dephosphorylation in isolated membranes. The present study was designed to further characterize the effects of bromotetramisole and to examine its congener levamisole, which is already used to treat other human disorders (33)(34)(35)(36) and therefore could be brought to clinical trials more readily.
Single-channel Patch-Clamp Recording-CHO cells were plated at low density on glass coverslips and cultured at 37°C in 5% CO 2 for 1-4 days before use. The methods used for patch-clamp recordings are similar to those previously described (8,18). Single-channel currents were recorded from both cell-attached and excised patches. Data were continuously recorded using an Axopatch 200A amplifier (Axon Instruments, Inc., Foster City, CA) and video tape recorder. Recordings were low pass filtered at 150 Hz (8-pole Bessel) and digitized at 1 kHz for analysis by a laboratory computer and custom software. Current amplitudes were determined using a semi-automated procedure in which unitary currents were measured at each potential from amplitude histograms computed for short segments of record, and values were displayed as current-voltage (I/V) relationships at the end of the run. Reversal potentials and conductances data were obtained from the I/V curves by least square regression analysis. Voltage was referenced to the external surface of the membrane patch. In all illustrations, outward currents (flowing into the patch electrode) are displayed in the upward direction.
Iodide Efflux-CFTR(Ϫ) or CFTR(ϩ) CHO cells were loaded for 1 h at room temperature with the following buffer (in mM): 136 NaI, 3 KNO 3 , 2 Ca(NO 3 ) 2 , 11 glucose, and 20 HEPES, pH 7.4. Extracellular NaI was removed by thoroughly washing with efflux buffer (136 mM NaNO 3 replacing the same amount of NaI). Cells were equilibrated for 1 min in a final 1-ml aliquot. The first four aliquots were used to establish a stable baseline efflux, and then agonist was added (at time 0) and the amount of iodide in each aliquot was determined using an iodide-specific electrode. Alternatively, a radioisotopic method for measuring iodide was used to measure iodide efflux and gave equivalent results. After removing the culture medium, CFTR(Ϫ) or CFTR(ϩ) CHO cells were washed twice with 2 ml of modified Earle's salt solution (solution B) containing (in mM): 137 NaCl, 5.36 KCl, 0.4 Na 2 HPO 4 , 0.8 MgCl 2 , 1.8 CaCl 2 , 5.5 glucose, and 20 HEPES, pH 7.4. Cells were loaded in solution B containing 1 M KI (0.1 Ci of 125 I-Na/ml, Amersham Corp.) for 30 min. The first two aliquots were used to establish a stable baseline efflux, and then the buffer was replaced with fresh solution containing agonist for subsequent aliquots. The medium was recovered at the end of the incubation, cells were solubilized in 1 N NaOH, and radioactivity was determined using a ␥ counter (Compu Gamma, LKB). Efflux curves were constructed by plotting the percent of cellular content accumulated in the medium versus time.
Cyclic-AMP Determination-Cells were grown to confluence, washed twice with Ca 2ϩ -and Mg 2ϩ -free phosphate-buffered saline and detached from plates with phosphate-buffered saline containing 0.1% EDTA. They were spun down at 800 rpm for 10 min at 4°C and resuspended in KRH buffer (140 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 6 mM glucose, 25 mM HEPES-NaOH, pH 7.4). Samples were taken for cyclic AMP determination by using a cyclic AMP enzyme immunoassay kit (Sigma).
Fluorescence Measurement of Intracellular Ca 2ϩ Concentration-Cells were loaded with fura 2 acetoxymethyl ester (Molecular Probes, Eugene, OR) as described previously (24). Briefly, individual coverslips were mounted in the bottom of a temperature-controlled, laminar flowthrough chamber mounted on a Nikon inverted microscope equipped for epifluorescence (Diaphot, Nikon, Tokyo, Japan). Excitation light was from a 75 W mercury-xenon arc lamp. The emitted fluorescence was deflected to the eyepieces or to an intensified charge-coupled device video camera (Model 2468, Hamamatsu Photonics, Hamamatsu City, Japan). Analyses were performed using a computer-based image analysis system (Fluor-1; Universal Imaging, West Chester, PA) and are presented as ratios of the fluorescence intensities at 520 nm measured during excitation at 340 or 380 nm.
Solutions and Chemicals-The catalytic subunit of PKA was obtained from the laboratory of M. P. Walsh and was described previously (8). The protein kinase inhibitor N-(2-(p-bromocinnamylamino)ethyl)-5-isoquinolinesulfonamide (H89) and protein phosphatase inhibitors calyculin A and okadaic acid were from RBI (Natick, MA). Levamisole and bromotetramisole were purchased from Aldrich (Metuchen, NJ), forskolin was from Calbiochem (San Diego, CA), and other chemicals were from Sigma.
Statistical Analysis-The data are presented as the means Ϯ S.E. Statistical significance was assessed at the 95% confidence level using the Student's t test.

RESULTS
The Phenylimidazothiazole Drugs Levamisole and Bromotetramisole Activate CFTR Channels-We showed previously that bromotetramisole can activate single CFTR channels on cultured cells; however, it was not clear if such stimulation would generate a significant macroscopic flux. Thus we compared 125 I effluxes from CFTR(Ϫ) and CFTR(ϩ) CHO cells during stimulation by bromotetramisole or forskolin. Fig. 1A shows that 125 I efflux from CFTR-expressing CHO cells was stimulated by 5 M forskolin, consistent with previous studies of these and other CFTR-expressing cells (9). Bromotetramisole (2 mM) evoked a similar iodide efflux from CFTR(ϩ) but not from CFTR(Ϫ) cells (Fig. 1B). Importantly, bromotetramisole (2 mM) stimulated 125 I efflux from CFTR(ϩ) CHO cells in the absence of forskolin (Fig. 1, B and D), and the response was comparable with the one evoked by forskolin. Many cells have calciumactivated chloride conductances that could potentially mediate 125 I effluxes. Therefore, we examined the effect of raising intracellular Ca 2ϩ using a calcium ionophore. The ionophore A23187 (10 M) failed to modulate iodide effluxes from CFTR(Ϫ) or CFTR(ϩ) CHO cells, suggesting that elevating calcium is not sufficient to activate this anion permeability (compare Figs. 1, C and D).
Single channel activity was studied in cell-attached patches from CFTR(ϩ) and CFTR(Ϫ) CHO cells with or without phenylimidazothiazole drugs in the bath (Fig. 2). Spontaneous openings of CFTR channels did not occur on resting CFTR(ϩ)cells (n ϭ 15) but were consistently observed following addition of 600 M levamisole to the bath (Fig. 2, A and B). The maximum number of channels activated by levamisole (N) averaged 6.5 Ϯ 1.04 (n ϭ 6 patches), as estimated from the largest number of simultaneous openings during long recordings. Channel open probability (P o ) averaged 0.35 Ϯ 0.01 (n ϭ 6) during stimulation by levamisole alone. The current-voltage relationship of levamisole-stimulated CFTR channels and the magnitude and time course of the levamisole-induced iodide efflux were similar to those obtained during forskolin stimulation (Fig. 2, C and D) and are consistent with previous reports using CHO cells (8) and other preparations (4 -10). On CFTR(ϩ) CHO cells, CFTR channel activity was observed in 86% of experiments with 1 mM levamisole (20 out of 23 patches; Fig. 2E) and in 66% of experiments in the presence of 1 mM bromotetramisole (14 out of 21 patches; data not shown). Importantly, forskolin and other cAMP agonists were not present during these studies of levamisole and bromotetramisole effects. Both drugs failed to activate channels on CFTR(Ϫ) CHO cells (Fig. 2E).
Stimulation of CFTR channels by levamisole was concentration-dependent (Fig. 3). Open probability was 0.03 Ϯ 0.01 with 60 M levamisole (n ϭ 3); however, this may be overestimated because the number of channels tends to be underestimated when P o is low because all the channels in the patch rarely open simultaneously. By contrast, the number of active CFTR channels in cell-attached patches was 14 Ϯ 1 (n ϭ 4), and mean P o was 0.45 Ϯ 0.2 (n ϭ 4) during exposure to 5 mM levamisole. The concentration of levamisole required to elicit 50% of the maximal response (EC 50 ) was 450 M. Maximal stimulation of the P o was obtained with millimolar levels of levamisole (Fig. 3, B and C). This potency is considerably lower than reported previously for inhibition of alkaline phosphatase by levamisole (11 M) and bromotetramisole (1.2 M; Ref. 25), indicating low membrane permeability or the phosphatase regulating CFTR may be distinct from known alkaline phosphatases despite having an epitope recognized by an antibody to alkaline phosphatase (7). Two lines of evidence suggest that activation by phenylimidazothiasole drugs was relatively specific for CFTR. First, we did not observe activation of any other ion channels by levamisole or bromotetramisole in the 38 cell-attached patches or many more that were recorded in the inside-out configuration. Second, bath addition of the dextro isomer D-bromotetramisole (1 mM, n ϭ 8), which does not inhibit alkaline phosphatase and is often used as a control for nonspecific actions (26), had no effect on CFTR channels in the cell-attached configuration. The protein phosphatase 1 and 2A inhibitors calyculin A (10 M, n ϭ 5) and okadaic acid (10 M, n ϭ 10) also failed to activate CFTR channels in cell-attached patches, as expected from their inability to slow rundown of channel activity in excised patches (18).
Mechanism of Activation of CFTR Channels by Phenylimidazothiazole Drugs-We examined the effects of phenylimidazothiazoles on signaling pathways by measuring intracellular cAMP and free Ca 2ϩ concentrations in CHO cells during maximal stimulation by levamisole. cAMP levels increased from 6.4 Ϯ 0.1 pmol/well in resting CHO cells (n ϭ 3, Fig. 4A) to 66.6 Ϯ 10 pmol/well during exposure to 15 M forskolin (n ϭ 3, Fig.  5A). By contrast, incubation with levamisole (1 mM) or bromotetramisole (1 mM) had no effect on cAMP levels (8.6 Ϯ 0.5 pmol/well (n ϭ 3, Fig. 5A) and 5.3 Ϯ 0.8 pmol/well (n ϭ 3), respectively). Thus phenylimidazothiazoles do not increase anion permeability by raising intracellular cAMP. To examine the possibility that calcium mobilization mediates chloride channel activation by these drugs, we determined cytosolic free Ca 2ϩ concentrations before and during exposure to levamisole or the calcium ionophore ionomycin using digital fluorescence imaging. CHO cells were superfused with 1 or 5 mM levamisole for 8 min, allowed to recover for 10 min, and then exposed to 100 M ionomycin for another 8 min. As shown in Fig. 5B, levamisole had no effect on the fluorescence ratio, whereas ionomycin caused a more than 3-fold increase. Similar results were obtained using levamisole (1 mM, data not shown). These data showing that phenylimidazothiazole drugs do not affect the cytosolic free Ca 2ϩ concentration are consistent with those in Fig. 2 (C and D), which demonstrated that the Ca 2ϩ ionophore A23187 does not stimulate iodide efflux.
To investigate if basal PKA activity in unstimulated CHO cells mediates the activation by phosphatase inhibitors, CFTR channels were recorded on CFTR(ϩ) cells after incubating them for 10 -15 min with the PKA inhibitor H89 (22). Levamisole (1 mM) still activated CFTR channels in the presence of high concentrations of H89 (50 M) in five out of eight experiments (mean number of channels; n ϭ 1.6 Ϯ 1.3). H89 effectively inhibited the PKA-dependent pathway under these conditions because forskolin stimulation was completely blocked (n ϭ 5). Nevertheless, the number of active CFTR channels per patch was decreased by 75% compared with control conditions (i.e., levamisole without H89). Partial activation by levamisole in the presence of H89 suggests that other kinases in addition to PKA contribute to CFTR activation during exposure to phosphatase inhibitors.
Rich et al. (27) showed that a mutant CFTR containing eight serine-to-aspartate substitutions generated channels that could open without PKA-dependent phosphorylation. We studied a similar mutant in which the eight R domain serines had been replaced with glutamates (8SE). 8SE channel activity was occasionally observed on unstimulated cells (Fig. 5, A and B), although open probability and number of active channels per patch were similar with (P o ϭ 0.11 Ϯ 0.1, n ϭ 8; n ϭ 1.1 Ϯ 0.8, n ϭ 8 patches) or without exposure to 15 M forskolin (P o ϭ 0.12 Ϯ 0.15, n ϭ 7; n ϭ 1.4 Ϯ 0.9, n ϭ 7 patches). This ineffectiveness of forskolin contrasted with levamisole and bromotetramisole (1 mM), which increased the P o and the number of active 8SE-CFTR channels by 3.5-fold (Fig. 5, A and B). Fig.  5C shows a representative experiment in which 1 mM bromotetramisole was added to an unstimulated cell stably expressing 8SE-CFTR. The low level of constitutive activity before the addition of the drug was increased significantly by adding bromotetramisole (middle and lower traces in Fig. 5C). Similar activation by levamisole was obtained using cell-attached patches, which contained mutants that lacked all 10 dibasic PKA consensus sequences due to the substitution of alanine for serine (n ϭ 1.3 Ϯ 0.57, n ϭ 5; Ref. 15). These data are compatible with phenylimidazothiazole drugs acting on CFTR, in part, through a PKA-independent mechanism. Functional Response of CFTR Channels with Disease-causing Mutations-Three CFTR mutations associated with severe (⌬F508 and G551D) or mild (R117H) forms of cystic fibrosis were studied after stable transfection into CHO cells. We examined the activity of the mutated channels in cell-attached patches after incubation with phenylimidazothiazole drugs in the absence of forskolin. For each mutant, we estimated the number of active CFTR channels, open probability (P o ), unitary conductance, and frequency of activation among different experiments. All three mutated channels were activated by levamisole or bromotetramisole in the cell-attached configuration. The results are summarized in Table I. The basic biophysical properties of ⌬F508 and R117H versions of CFTR were similar to those reported previously (11,28,29). R117H had a slightly lower unitary conductance and significantly reduced P o (Table I). Single ⌬F508 channels were detected after treatment with levamisole (1 mM) or forskolin (15 M) if cells were maintained at 23°C prior to patch-clamp experiments (n ϭ 5) but not if they were maintained at 37°C. The P o of ⌬F508 CFTR channels was reduced by ϳ60%; however, unitary conductance was the same as wild-type CFTR (Table I).
We also examined a CFTR mutation (R347D) that is at a residue where disease-causing mutations have been identified (R347P and R347H). These mutations all reduce single channel conductance and abolish multi-ion pore behavior (29,30). Spontaneous activity of R347D channels was never observed on resting cells (Table I) but was elicited by exposure to 15 M forskolin (80% of patches; Fig. 6A) or l mM levamisole (60% of patches; Fig. 6, B and C). The unitary conductance of R347D channels determined after stimulation by forskolin was 2.9 Ϯ 0.2 pS (n ϭ 8) and by levamisole was 2.8 Ϯ 0.3 pS (n ϭ 6; Fig.  6D).
Bromotetramisole increases the incorporation of radiolabeled PO 4 into CFTR when isolated membranes are incubated in the presence of PKA and [ 32 P]ATP and is therefore believed to potentiate CFTR activity by enhancing its phosphorylation (18). Evidence that the phosphatase regulating CFTR is constitutively active was obtained by exposing excised patches to phenylimidazothiazoles in the presence of 180 nM PKA and 1 mM ATP. Levamisole further increased channel activity despite the presence of high levels of PKA and ATP. Similar results were obtained after comparing bromotetramisole (P o ϭ 0.51 Ϯ 0.06, n ϭ 7.5 Ϯ 0.72 channels/patch, n ϭ 7 patches) with stimulation with PKA ϩ MgATP alone (P o ϭ 0.25 Ϯ 0.04, n ϭ 2.87 Ϯ 0.04, n ϭ 12 patches). By contrast, okadaic acid (10 M, n ϭ 3), an inhibitor of type 1 and 2A protein phosphatases, had no effect on CFTR activity under these conditions (P o ϭ 0.26 Ϯ 0.05; n ϭ 3.2 Ϯ 0.9). Tonic inhibition of mutant channels by phosphatases was also relieved by phenylimidazothiazoles. G551D channel activity in the presence of PKA and ATP was lower than wild-type CFTR (P o ϭ 0.17 Ϯ 0.06, n ϭ 2.6 Ϯ 0.57, n ϭ 3 patches) but was nevertheless increased 2-fold by addition of bromotetramisole in the presence of PKA (P o ϭ 0.36 Ϯ 0.08, n ϭ 6.2 Ϯ 0.56, n ϭ 7 patches). Although these values for P o may be overestimated because they are based on the estimated number of channels without locking all the channels open using AMP-PNP, the increase in open probability was consistent and was also observed with the mutants R117H and R347D (data not shown).
Bromotetramisole and levamisole activated channels on unstimulated cells, and 25% of the response was insensitive to the PKA inhibitor H89. To investigate whether the residual stimulation results from direct stimulation and is independent of phosphorylation, we examined the effects of phenylimidothia- zoles on excised patches in the nominal absence of kinase activity (Fig. 7). CFTR channels did not respond to the addition of levamisole (n ϭ 5) or bromotetramisole (n ϭ 4) within 10 min, although subsequent addition of PKA (180 nM) to the same excised patches produced vigorous activation. This indicates that levamisole and bromotetramisole do not directly stimulate CFTR in the absence of phosphorylation.

DISCUSSION
The alkaline phosphatase inhibitor bromotetramisole can activate CFTR channels on cultured airway epithelial and CHO cells (18). This paper investigates the mechanism of this stimulation and extends the study to a related molecule, levamisole. Stimulation by levamisole is potentially significant because this drug is already used to treat other human disorders (33) and therefore might be brought more readily to clinical trials.
The present results show that some phosphorylation is required for stimulation of CFTR by phenylimidazothiazole drugs to occur. The data are consistent with inhibition of a membrane-associated phosphatase activity by these drugs, which enables phosphorylation to accumulate. Approximately 75% of the phenylimidazothiazole response is blocked by the PKA inhibitor H89. The H89-insensitive component of the levamisole response presumably reflects accumulation of phosphorylation due to other kinases, which are no longer opposed by the phosphatase.
Levamisole Activates CFTR Channels in Cultured Cells-Both levamisole and its brominated analogue bromotetramisole are stereospecific inhibitors of alkaline phosphatases (e.g., EC 3.1.3.1; Fig. 8; Ref. 31). Levamisole is about 10-fold less potent than bromotetramisole but has been in veterinary and clinical use for many years as a treatment for parasitic worms (32), colorectal carcinoma (33), and upper respiratory diseases (34 -36). Two isomers of bromotetramisole are available commercially that differ in the position of the asymmetric carbon (see asterisk in Fig. 8). The L form of bromotetramisole inhibits alkaline phosphatases from the liver, bone, and kidney whereas D-bromotetramisole does not. Significantly, bromotetramisole inhibits alkaline phosphatase from lung but is not effective against intestinal alkaline phosphatase (see Table II).
Our results can be summarized as follows: (i) levamisole and bromotetramisole promote the opening of cell-attached CFTR channels in the absence of forskolin, (ii) elevation of intracellular cAMP and Ca 2ϩ are not required for this stimulation, (iii) at least four mutant CFTRs (G551D, R117H, R347D, and ⌬F508) can also be activated on-cell by these drugs in the absence of forskolin, (iv) both phenylimidazothiazoles further enhance CFTR channel activity when excised patches were exposed to high levels of PKA, indicating that their target is associated with the membrane, and (v) activation by both drugs requires some kinase activity.
Among the substances that have been reported to open CFTR channels, phenylimidazothiazole drugs were examined because

TABLE I Effect of phenylimidazothiazole drugs in cell-attached patches from CHO cells on various parameters of channels
Experiments were performed using cell-attached patches, in the absence (control) or presence of 1 mM levamisole in the bath. The frequency represents the ratio between the number of patches containing activated CFTR channels and the number of patches tested. For each successful experiment, the number of activated CFTR channels, the corresponding open probability (P o ) and conductance (g) are given. Data are expressed as the mean Ϯ S.E. Note that no activity was observed in control conditions.  (26,31,37). Alkaline phosphatase is a major membrane-associated phosphoprotein phosphatase in human and rat liver (37,38). Immunoprecipitated CFTR protein can be dephosphorylated by the addition of exogenous purified alkaline phosphatase (18). CFTR channel activity is inhibited ϳ90% when alkaline phosphatase is added to excised patches from various cell preparations (7,9,18), and phenylimidazothiazoles further enhanced CFTR activity in patches that are exposed to PKA (Fig. 7).
Although the precise mechanism of action is uncertain, it is likely that phenylimidazothiazole drugs inhibit a phosphatase activity, thereby stabilizing a phosphointermediate form of CFTR (37,39). Several families of compounds have been shown to modulate the activity of chloride channels. The xanthine derivatives 3-isobutyl-1-methylxanthine (40), 8-cyclopentyl-1,3-dipropylxanthine (41), and the epoxygenase inhibitor ketoconazole (42) have been proposed as therapeutics. The mechanisms of stimulation by 8-cyclopentyl-1,3-dipropylxanthine and ketoconazole are not understood; however, 3-isobutyl-1-methylxanthine appears to act through several mechanisms, one involving inhibition of the membrane associated phosphatase activity (18). There is evidence that the tyrosine kinase inhibitor genistein also activates CFTR by inhibiting a phosphatase (43).
Levamisole was chosen for this study because of its similarity to bromotetramisole and because it is already used in humans as mentioned above. There are anecdotal reports that levamisole reduces the frequency, duration, and severity of upper respiratory tract infections in children (34 -36), but whether this is due to its ability to activate CFTR channels is uncertain. Regardless, stimulation of the immune system could potentially be a useful side effect in the treatment of CF.
Specificity of Levamisole Action-The spectrum of alkaline phosphatases inhibited by levamisole is relatively narrow. Among the three major types of alkaline phosphatase, intestinal, placental, and liver-bone-kidney, levamisole is highly specific for the liver-bone-kidney phosphatases but is relatively ineffective against the intestinal and placental forms. Table II summarizes the relative activities of this drug on various enzymes. Importantly, several ATPases, 5Ј-nucleotidases, and phosphodiesterases are not inhibited by levamisole nor are acid phosphatase, type 1 or 2A protein phosphatases, or various tyrosine phosphatases. Although circumstantial, this agrees with the growing body of evidence that alkaline phosphatase is indeed a phosphoprotein phosphatase (37,38,43). The dextroisomer of bromotetramisole ((ϩ)-p-bromotetramisole), which does not inhibit alkaline phosphatases, also failed to activate  CFTR channels in cell-attached patches and did not slow the rundown of CFTR channels in excised patches. The phosphatases regulating CFTR apparently vary according to cell type. In some cells, the PP1 and PP2A inhibitors okadaic acid and calyculin A have profound effects on CFTR activity. PP2A is a relatively abundant cytosolic protein phosphatase. It can dephosphorylate immunoprecipitated CFTR protein (14,18) and may regulate it under particular conditions or at a particular stage during the life of the protein. Protein phosphatase 2C remains a good candidate for the CFTR phosphatase in CHO and airway cells. Like alkaline phosphatase, it requires magnesium, but no data are available concerning its sensitivity to phenylimidazothiazoles. Drugs such as levamisole clearly can activate CFTR channels on cultured cells, but their effectiveness in vivo remains to be demonstrated. Bromotetramisole (1 mM) and levamisole (1 mM) had no effect on short circuit current across trachea, jejunum, or caecum freshly isolated from normal and transgenic G551D mice. 2 Interestingly, forskolin did cause significant stimulation of the short circuit current, probably due to cAMPinduced calcium mobilization and stimulation of Ca 2ϩ -activated chlorine conductance (45). It is not known if levamisole or bromotetramisole inhibit the rundown of CFTR channel activity in patches excised from mouse airway cells. If CFTR contributes little to chlorine secretion across the epithelium of mouse airways, it may be necessary to test levamisole using some other in vivo preparation, for example, by examining its effects on human nasal potentials. An intestinal phenotype does occur in the G551D mouse; however, it is unclear whether the same phosphatases regulate CFTR in airway and intestinal cells and the lack of effect might be explained by the known insensitivity of intestinal alkaline phosphatase to bromotetramisole.
Sensitivity to phosphatase inhibitors may depend on the level of CFTR expression. Phosphatase activity may be sufficient to keep endogenous CFTR channels closed in vivo but not in cells expressing heterologous CFTR at high levels. The role of this safety factor in determining CFTR responsiveness to phosphatase inhibitors should be explored.
Phenylimidazothiazoles are well tolerated in vivo and in vitro. Oral administration of very high doses of levamisole causes few if any side effects according to short term clinical records of children with upper respiratory infections (34 -36). Extensive toxicity studies of oral levamisole have been carried out in various animal species (32,46). In preliminary toxicological tests, cultured cells grown for up to 5 days in medium containing 1 mM levamisole appeared normal and had normal cellular protein content. 3 Further laboratory experiments to identify more potent congeners combined with limited clinical studies should indicate whether phenylimidazothiazole drugs will be beneficial in the treatment of cystic fibrosis patients.