PHYSIOLOGICAL REVIEWS   Vol. 79 No. 1 January 1999, pp. S109-S144
Copyright ©1999 by the American Physiological Society

Pharmacology of CFTR Chloride Channel Activity

B. D. SCHULTZ, A. K. SINGH, D. C. DEVOR, AND R. J. BRIDGES

University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

I. INTRODUCTION
II. CHANNEL BLOCKERS
    A. Sulfonylureas and Diarylsulfonylureas
    B. Disulfonic Stilbenes
    C. Arylaminobenzoates
III. CHANNEL OPENERS
    A. Xanthines
    B. Phosphatase Inhibitors
    C. Isoflavones and Flavones (Genistein)
    D. Benzimidazolones
    E. Benzoxazoles (Chlorzoxazone)
    F. Psoralens
IV. HUMAN AIRWAY EPITHELIAL STUDIES
REFERENCES

	ABSTRACT

Top References

Schultz, B. D., A. K. Singh, D. C. Devor, and R. J. Bridges. Pharmacology of CFTR Chloride Channel Activity. Physiol. Rev. 79, Suppl.: S109-S144, 1999.  The pharmacology of cystic fibrosis transmembrane conductance regulator (CFTR) is at an early stage of development. Here we attempt to review the status of those compounds that modulate the Cl channel activity of CFTR. Three classes of compounds, the sulfonylureas, the disulfonic stilbenes, and the arylaminobenzoates, have been shown to directly interact with CFTR to cause channel blockade. Kinetic analysis has revealed the sulfonylureas and arylaminobenzoates interact with the open state of CFTR to cause blockade. Suggestive evidence indicates the disulfonic stilbenes act by a similar mechanism but only from the intracellular side of CFTR. Site-directed mutagenesis studies indicate the involvement of specific amino acid residues in the proposed transmembrane segment 6 for disulfonic stilbene blockade and segments 6 and 12 for arylaminobenzoate blockade. Unfortunately, these compounds (sulfonylureas, disulfonic stilbenes, arylaminobenzoate) also act at a number of other cellular sites that can indirectly alter the activity of CFTR or the transepithelial secretion of Cl. The nonspecificity of these compounds has complicated the interpretation of results from cellular-based experiments. Compounds that increase the activity of CFTR include the alkylxanthines, phosphodiesterase inhibitors, phosphatase inhibitors, isoflavones and flavones, benzimidazolones, and psoralens. Channel activation can arise from the stimulation of the cAMP signal transduction cascade, the inhibition of inactivating enzymes (phosphodiesterases, phosphatases), as well as the direct binding to CFTR. However, in contrast to the compounds that block CFTR, a detailed understanding of how the above compounds increase the activity of CFTR has not yet emerged.

	I. INTRODUCTION

Top Next References

Experimentally, there is a considerable need for specific high-affinity ligands that could be used to probe the structure and function of the cystic fibrosis transmembrane conductance regulator (CFTR). Insights gained from such experimentation would allow us to better understand CFTR's role in cell biology, physiology, and ultimately the underlying mechanisms leading to clinical pathology. There is little doubt that CFTR is the Cl channel that mediates cAMP-dependent Cl secretion in various epithelia. The availability of potent and specific CFTR modulators would greatly aid our understanding of how the different domains of CFTR interact to form a protein kinase A- and ATP-regulated Cl channel; the mechanisms of anion conduction; how CFTR interacts with other ion channels; the role CFTR plays in intracellular compartments to regulate pH, glycoprotein sulfation, and vesicle fusion; as well as how mutations in CFTR cause cystic fibrosis (CF). Equally important will be the therapeutic impact of CFTR modulators in the treatment of respiratory disorders including CF, chronic obstructive pulmonary disease, asthma, bronchitis, emphysema, pneumonia as well as secretory diarrhea, polycystic kidney disease and reproductive dysfunctions (congenital bilateral absence of the vas deferens, Ref. 270; testicular and sperm abnormalities, Ref. 88). Compared with cation channels, nature has not been generous in providing Cl channel modulators. Here we attempt to review the status of those compounds thought to interact with or modulate the Cl channel activity of CFTR. The presentation is divided into two main sections discussing those compounds that block CFTR and those that open or increase the activity of CFTR.

The channel activity of CFTR can be altered at multiple levels (Fig. 1). Because CFTR is activated by the cAMP signal transduction pathway, alterations in the binding of an agonist to its receptor, the G protein-mediated activation of adenylyl cyclase, the hydrolysis of cAMP by phosphodiesterase (PDE), and the activity of protein kinases or of protein phosphatases can influence the activity of CFTR. The opening and closing of CFTR is also dependent on ATP and ADP (sects. III and IV) and affected by the cellular redox potential (373). Thus changes in cellular metabolism can influence the activity of CFTR. Therefore, a compound may act at one or more of these sites of action as well as bind directly to CFTR to alter channel gating. Studies using compounds to inhibit or increase the activity of CFTR must consider each of these potential sites of action when attempting to interpret results from cellular-based assays of CFTR channel activity such as ¹²⁵I efflux, 6-methoxy-N-(3-sulfopropyl)quinoline (SPQ) fluorescence, or whole cell membrane patch recordings. The transepithelial secretion of Cl also depends on three basolateral membrane proteins: the Na⁺-K⁺-ATPase to maintain a Na⁺ gradient, the Na⁺-K⁺-2Cl cotransporter for Cl entry into the cell, and K⁺ channels to recycle K⁺ and maintain the necessary membrane potential to drive Cl exit across apical membrane Cl channels (24, 154). At least three biophysically and pharmacologically distinct types of K⁺ channels are thought to contribute to the basolateral membrane K⁺ conductance: a cAMP-activated K⁺ channel, a Ca²⁺-activated K⁺ channel, and a maxi K⁺ channel. In addition to CFTR, two other Cl channels, an outwardly rectifying Cl channel (ORCC) and a Ca²⁺-activated Cl channel, are reported to contribute to the apical membrane Cl conductance. Other Cl conductances such as members of the Cl channel gene family (ClC) and voltage- and osmolyte-sensitive anion conductance have not historically received as much attention, and thus their relative contribution to transepithelial ion movement remains to be defined (193, 399). The activities of each of these basolateral and apical membrane transporters and channels are tightly coordinated in the regulated secretion of Cl. Thus, in addition to effects on the signal transduction pathways, compounds may have inhibitory or stimulatory effects on one or more of these transporters and channels and thereby alter the secretion of Cl. In general, many of the compounds that have been used to alter the channel activity of CFTR are not specific for CFTR. We attempt, in this review, to present the pharmacology of these compounds, specifically, the documented actions of these compounds at sites other than CFTR. Our intent is to provide a basis upon which one may formulate an informed interpretation of the sometimes confusing array of published observations using these compounds as well as to assist in the design of future studies with these compounds.

View larger version (30K): [in this window] [in a new window]   FIG. 1.   Schematic diagram of a Cl secretory epithelial cell showing selected transporters and channels that participate in transepithelial Cl secretion. Included are selected second messenger pathways (cAMP cascade and Ca2+ cascades) reported to affect ion transport mechanisms. Note that each step in these pathways and transport proteins themselves are susceptible to pharmacological modulation as discussed in text. Four types of Cl channels are indicated in apical membrane: ClORCC, an outwardly rectifying Cl channel; ClVOL, a volume-regulated Cl channel that may also be present in basolateral membrane; ClCa, a Ca2+-activated Cl channel, and ClCFTR, a cAMP/protein kinase A (PKA)-activated Cl channel. Molecular identities of ClORCC, ClVOL, and ClCa are not known, nor is relative contribution of various Cl channels in response to different secretory agonists. In addition to Na+-K+-2Cl cotransport and Na+-K+-ATPase, 3 different K+ channels are shown in basolateral membrane: KBK, a large-conductance K+ channel; KcAMP, a cAMP-activated K+ channel; and KCa, a Ca2+-activated K+ channel. Evidence suggests KCa corresponds to recently cloned HIK-1 K+ channel, whereas molecular identities of KBK an KcAMP remain unknown. In addition to binding to basolateral membrane receptors, agonists may also bind to apical membrane receptors (not shown) to stimulate Cl secretion. Agonists that elevate cAMP can do so by activating one of several isoforms of adenylate cyclase, some of which are regulated by intracellular Ca2+, and in turn intracellular Ca2+ can be regulated by cAMP levels. Thus, in addition to various isoforms of PKA, protein kinase C, protein phosphatases (PPase), and phosphodiesterases (PDE), the two signal transduction cascades can interact at the second messenger level as well as the transport protein, ion channel level to regulate secretion of Cl.

The pharmacology of CFTR is driven in large measure by the desire to develop agents that will be useful in the treatment of CF. Since the discovery of the gene coding for CFTR (201, 302, 306), over 700 mutations that cause CF have been reported to the CF Genetic Analysis Consortium (13; accessible electronically at http://www.genet.sickkids.on.ca). Mutations that cause one or more problems in the transcription, translation, protein processing, or channel activity of CFTR have been described (394, 395, 423, 424, 442). The end result of the mutations in CFTR is an alteration in fluid and electrolyte transport in the epithelia of the sweat glands, pancreas, intestines, reproductive organs, and airways. One approach to the pharmacological treatment of CF is to restore normal function to the mutant CFTR protein. Although the debate continues on how many functions CFTR serves, most investigators do agree the restoration of CFTR Cl channel activity in the apical membrane of the affected epithelia is a desired goal. Unfortunately, the vast majority of CF-causing mutations result in the absence or reduced expression of apical membrane CFTR protein. Thus, for most mutations, one must develop a pharmacological means of facilitating the transcription, translation, or protein processing of CFTR to deliver a mature protein to the apical membrane. If, in addition, the mutation causes the channel to be dysfunctional, then the channel activity must also be pharmacologically modified to achieve normal function. The deletion of phenylalanine at position 508 (F508) in the first nucleotide binding fold (NBF) is the single, most frequent CF-causing mutation, being present on ~67% of mutant allels (193), although the relative proportion of specific mutations varies depending on the population of inference (307). The F508 mutation causes both protein processing and channel activity to be dysfunctional (85, 89, 103, 159, 178). Therefore, the treatment of F508 CF patients will require the correction of both the protein processing and channel activity defects of the F508 CFTR protein.

Although two entirely different classes of compounds may be required to restore normal function to F508 CFTR, specific high-affinity compounds that interact with CFTR to modulate channel activity may serve both purposes. This notion is illustrated by recent exciting studies on the multidrug resistance (MDR) protein P-glycoprotein (230). P-glycoprotein, like CFTR, is a member of the ATP binding cassette (ABC) superfamily of transport proteins. Several classes of compounds are known to be transported by or act as inhibitors of P-glycoprotein. Loo and Clarke (230) have recently shown that these same compounds can assist in the protein processing of mutant P-glycoprotein that would have otherwise not reached the plasma membrane. These results lend great promise to the hoped for discovery of CFTR-specific compounds that will improve the delivery of F508 CFTR to the plasma membrane. So far, the only compounds known to bind to CFTR are those that alter channel activity. Thus the discovery and the development of compounds that can restore normal protein processing to some mutant forms of CFTR may result from the further development of specific and potent CFTR channel modulators.

The certainty that specific high-affinity CFTR channel modulators can be developed is supported by the pharmacology of another member of the ABC transporter superfamily, the sulfonylurea receptor (SUR). The sulfonylurea receptor is found in pancreatic -cells (2) and in muscle cells (185), where it associates with Kir6.2 to form an ATP-sensitive K⁺ channel complex, K_ATP (184). In this complex, SUR provides for both nucleotide and high-affinity sulfonylurea sensitivity (146). The release of insulin from -cells can be modulated by sulfonylureas, several of which bind to SUR with high affinity [e.g., dissociation constant (K_D) for glibenclamide <1 nM]. The development of specific high-affinity compounds like glibenclamide for SUR required 27 years after the discovery of hypoglycemic drugs and the synthesis of at least 12,000 derivatives by numerous pharmaceutical companies (14, 41, 252). A similar investment in effort may be required to obtain CFTR channel modulators of high potency and specificity. Certain to be of importance will be a detailed understanding of the kinetics and mechanism of action of the candidate compounds affecting CFTR channel activity. The intent of this review is to present our current understanding of the mechanisms of action of those compounds now known to affect CFTR channel activity. Although there are only a few classes of compounds and fewer still to which a mechanism can be ascribed, the list is growing, as is our understanding of their mechanisms of action. Thus, although still in its infancy, the pharmacology of CFTR Cl channels is progressing and is certain to yield significant future benefit.

	II. CHANNEL BLOCKERS

Top Previous Next References

Sulfonamide compounds were first developed and clinically employed for their bacteriostatic activities. When patients were being treated for typhoid with 2254 RP in 1942, it was noted that they became hypoglycemic (231, 422). In 1956, the sulfonylureas, carbutamide and tolbutamide (Fig. 2), were identified as nonbacteriostatic, clinically useful hypoglycemic agents. Subsequently, in excess of 12,000 sulfonylurea compounds were synthesized and tested for their ability to treat diabetes mellitus (14). This quest for a higher affinity (e.g., <1 nM) hypoglycemic agent resulted in the discovery of glibenclamide (Fig. 2) some 13 years later (252) and diazoxide, a sulfonamide with opposite effects, i.e., a hyperglycemic agent which proved useful for the treatment of insulinomas (14, 41). It was determined that the effects of sulfonylureas on pancreatic insulin secretion are mediated by the inhibition of a K⁺ channel that is physiologically stimulated by ADP and inhibited by ATP (K_ATP; Refs. 320, 321, 370). In 1995, 43 years after identifying hypoglycemic drugs and 26 years after identifying glibenclamide as a high-affinity modulator of insulin secretion, SUR was cloned and discovered to be a member of the ABC transporter superfamily of proteins (2, 287). More recently, it was shown that the SUR regulates the activity of a 38-kDa protein (Kir6.2) that copurifies with it and together function as the K_ATP channel (184, 383).

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Structures of compounds that have been used to pharmacologically block selected components of Cl secretion across epithelium.

Numerous sulfonylurea-based structures and associated synthetic protocols have been published during their development as antibacterial and antidiabetic drugs (69, 123). Extensive structure-activity relationships (SAR) for effects on insulin secretion and modulation of ATP-dependent K⁺ channels from a variety of laboratories indicate that the most potent sulfonylurea-based modulators include a cyclohexyl group linked to the urea portion of the sulfonylurea backbone and a para-substituted phenyl group linked to the sulfonyl portion (e.g., glibenclamide, Fig. 2; Ref. 219). Perhaps surprisingly, meglitinide, the benzoic acid derivative of glibenclamide, retains antidiabetic effects, indicating that the sulfonylurea moiety is not required, per se, for regulation of SUR/Kir6.2 (219). Furthermore, the para-aromatic acid or aliphatic amine present in numerous high-affinity modulators of SUR is not absolutely required on the sulfonyl phenyl, since tolbutamide (Fig. 2), which includes only a para-methyl substituent, is also an effective ligand, albeit at much higher concentrations. Rather, any combination of these moieties, when appropriately linked to a sulfonylurea backbone, provides for the highest potencies (219, 310). More recently, researchers at Lilly Laboratories have developed a subclass of sulfonylureas, the diarylsulfonylureas (Fig. 2), as oncolytic agents that do not universally have antidiabetic effects (170, 171, 260, 385). Although the SAR for oncolytic activity is seemingly extensive (171, 260), the oncolytic mechanism of action remains undefined (150, 203, 285, 286, 311). Antidiabetic sulfonylurea derivatives are largely available from commercial sources. Oncolytic diarylsulfonylureas are not commercially available, although synthetic protocols have been published (171, 260).

The realization that SUR is, like CFTR, a member of the ABC transporter family is one of many parallels that have been drawn between the SUR/Kir6.2 complex and CFTR. For example, both SUR/Kir6.2 and CFTR are reciprocally regulated by ATP and ADP. Intracellular ATP decreases and ADP increases the open probability of SUR/Kir6.2 (106, 120), whereas the reverse is true for CFTR (12, 333, 410, 432). In combination with these nucleotides, Mg²⁺ also modulates the gating of both CFTR and SUR/Kir6.2 (14, 152, 323). Sheppard and Welsh (347) first reported that antidiabetic sulfonylureas, compounds used to pharmacologically characterize SUR/Kir6.2, reduced whole cell Cl currents in NIH 3T3 cells expressing CFTR. Blocker concentrations in excess of 1,000-fold greater than those reported to inhibit SUR/Kir6.2 were required to similarly affect CFTR, although the rank order of potency remained unchanged (glibenclamide > tolbutamide). Unfortunately, compounds that open SUR/Kir6.2 (diazoxide, minoxidil; Refs. 14, 131) caused a reduction in CFTR-dependent Cl conductance (347). Thus CFTR does not have an activator site comparable to the diazoxide-responsive site on SUR/Kir6.2, although the rank order of potency for glibenclamide and tolbutamide suggests that the sulfonylurea-sensitive site is weakly conserved.

Both glibenclamide and tolbutamide have been shown to reduce the open probability of CFTR in a structure-dependent and concentration-dependent manner by direct interaction with the channel as recorded in cell-free membrane patches from CFTR-expressing C127 cells, Madin-Darby canine kidney (MDCK) cells, HEK 293 cells, and L cells (324, 331, 346, 409). Inhibition of CFTR channel currents in excised membrane patches of HT-29 cells, T84 cells (293), and Xenopus oocytes (250) has also been reported. In each case, the effective concentration of glibenclamide was similar with the IC₅₀ or K_D in the range of 2-30 µM. One discrepancy, however, is that reversal of effect with washout was rapid in mammalian expression systems (293, 324, 409) but failed to occur after an extended period in Xenopus oocytes (250). The lack of reversibility in Xenopus oocytes might be explained by channel rundown in the presence of glibenclamide or by other known effects of glibenclamide (see below), since Devor et al. (90) reported that inhibition by 300 µM glibenclamide was immediately reversible in shark rectal gland cells.

Kinetic analysis, modeling, and simulation of results employing glibenclamide and tolbutamide showed that the compounds reversibly interact exclusively with the open state of CFTR channels to interrupt Cl conduction (324, 346, 409). The blocking efficiency of glibenclamide is altered by changes in pH, which changes the proportion of glibenclamide in the anionic form, voltage, and extracellular Cl concentration (346). The simplest kinetic models to account for nucleotide- and kinase-dependent regulation of CFTR employ a four-state model of channel gating (64, 158, 333, 410, 432).

Thus the interaction of a sulfonylurea with the open state sequesters CFTR in an activated, albeit nonconducting, state and, in this linear scheme, reduces the likelihood of channel inactivation. The greater potency of glibenclamide compared with tolbutamide is chiefly attributable to a reduced off rate constant (k_off) (<100 vs. 1,210 s¹; Refs. 324, 346, 409), indicating that the glibenclamide-CFTR interaction is substantially more stable than the tolbutamide-CFTR interaction. From a theoretical and clinical perspective, these conclusions have significant ramifications. First, in the resting state, there is no expected effect of sulfonylureas on CFTR. Second, if activation is slowed or inactivation is accelerated by particular mutations as has been suggested (178, 322, 333, 428), then the effect of the sulfonylureas would be to prolong the duration which CFTR remains in an activated state before inactivation. All open-channel modulators would be expected to have similar effects on the state distribution, with the most potent compounds exhibiting the longest duration of interaction (i.e., the lowest k_off). Most effective in the symptomatic treatment of CF would be compounds that similarly decrease the inactivation rate by interacting with the open state but maintain some ionic conduction. Additionally, any such compound must, unlike glibenclamide or tolbutamide, be shown to have a high degree of selectivity for CFTR (see below).

On the basis of the knowledge that sulfonylureas block CFTR by a direct interaction, it has become widely accepted to employ glibenclamide-dependent inhibition of anion transport to indicate that CFTR participates in a particular physiological system of interest [e.g., mouse intestinal crypt secretion, Ref. 400; guinea pig ventricular myocyte Cl currents, Refs. 388, 389; rat nephron terminal collecting duct, Ref. 175; human kidney cyst epithelial cells, Ref. 155; mIMCD-K2 cells, Ref. 403; shark rectal gland cells, Ref. 90; NS004-, NS1619-, 1-EBIO-, and psoralen-stimulated secretion in T84 cells, Refs. 93, 94; protein kinase A (PKA)-stimulated Cl flux across rat nephron cortical brush-border membranes, Ref. 34; rat fetal lung epithelium, Ref. 377; and cAMP-stimulated ¹²⁵I efflux from MDCK cells, Ref. 261]. Alternatively, the lack of effect of glibenclamide has been used as an indicator that CFTR does not mediate the response in some systems (e.g., rat choroid plexus, Ref. 202; mouse mandibular salivary gland, Ref. 212; bovine pancreatic duct cells, Ref. 5; ATP-stimulated ion transport in rabbit tracheal epithelium, Ref. 189; and ATP-stimulated ion transport in epithelial cells from Mongolian gerbil middle ear, Ref. 127).

It must be emphasized that glibenclamide-dependent inhibition should not be used as a sole indicator of CFTR participation in a response. Because glibenclamide has a K_D for the SUR of <10 nM and at micromolar concentrations has been shown to have multiple effects which include the inhibition of a variety of K⁺ channels (32, 110), the inhibition of numerous enzymes (62) including PKA (275), and the inhibition of other Cl channels (255, 293, 324, 438), caution must be exercised in interpreting effects in intact tissues. Both nonunity Hill coefficients for concentration-dependent inhibition and the inability to "wash out" the inhibitory effect of glibenclamide might reflect concerted effects on intracellular enzymes including PKA rather than an ongoing bimolecular interaction with CFTR (90, 250, 347, 389). Likewise, a lack of effect or modest effect of glibenclamide can be misinterpreted to indicate that CFTR is not present. Ionic, pH, or voltage conditions can be setup such that either CFTR or glibenclamide is not in an optimal or permissive physicochemical confirmation for inhibition to occur (346). Additionally, one must be especially careful in interpreting experiments employing <30 µM glibenclamide, since, at these concentrations, a significant block of CFTR would not be expected in any situation. Although glibenclamide can effectively block CFTR and appears to be securely entrenched in CFTR literature, it is obvious that the field would benefit from a more potent and selective inhibitor of CFTR.

Evidence from our laboratory indicates that diarylsulfonylureas provide greater selectivity and slightly higher potency than glibenclamide for the inhibition of CFTR (328, 329, 331). The mechanism of action is identical to that presented above for glibenclamide and tolbutamide, interaction only with the open state of CFTR channels to interrupt Cl conduction. In contrast to glibenclamide, the diarylsulfonylureas have (by definition) substituted phenyl moieties linked to both the sulfonyl and urea of the sulfonylurea backbone (Fig. 2). The binding site of CFTR is like the SUR in that the pharmacophore does not require an aromatic acid or aliphatic amine linked to the sulfonyl phenyl group for functional interaction, e.g., tolbutamide. A p-methyl substituent on the sulfonyl phenyl appears to be adequate for interaction with the binding site (329, 409), although compounds including either benzofuran (LY-295501; Fig. 2) or indanyl (LY-186641; Fig. 2) groups linked to the sulfonyl have been shown to be effective (328). Potency of CFTR inhibition varies over more than three orders of magnitude depending on the urea-phenyl substituents. Electron-withdrawing substituents (e.g., in Fig. 2, R' = Cl, NO₂, or CF₃) at the para- and/or meta-positions of the urea phenyl have been shown to provide for higher affinity interactions than electron-donating or electroneutral substituents (e.g., R' = OH or H) (329). Both LY-186641 (p-chloro) and LY-295501 (m,p-dichloro) have been shown to block CFTR, but not to affect alternative Cl channels when expressed in Xenopus oocytes [GABA_C (84, 348) and rabbit gastric ClC (CIC2G; Schultz, unpublished observation and Ref. 236)], not to compete with glibenclamide for binding to SUR (328), and not to affect the SUR/Kir6.2, in vivo, as assessed by the failure to affect blood glucose levels in rats (171). That diarylsulfonylureas might be safely administered to humans is evidenced by the fact that both LY-186641 and LY-295501 have been employed in clinical trials (196, 271, 420; S. Beardslee, personal communication). These attributes (a partially defined mechanism of action on CFTR, a unique SAR, and clinical safety of closely related compounds) make the diarylsulfonylureas attractive as lead compounds in the development of CFTR Cl channel modulators.

The mechanisms that communicate sulfonylurea or diarylsulfonylurea binding to changes in CFTR channel gating remain to be determined. Although the CFTR channel protein is available and much is known about its nucleotide- and kinase-dependent regulation, pharmaceutical compounds (and more importantly, a set of compounds known to interact with a common binding site) that regulate its activity are not widely available. We have demonstrated specific binding of [¹²⁵I]iodoglibenclamide to CFTR (354) and have initial data to show that diarylsulfonylureas interact with this binding site in a structure-dependent way. Clearly, additional studies will provide insights to define the binding site and mechanism of action. Numerous parallels have been drawn between the pharmacology of CFTR, the SUR, and other ABC proteins. Like CFTR, neither the sulfonylurea binding site on the SUR protein nor the mechanistic relationship between sulfonylurea binding and changes in K⁺ channel activity is known. Mayorga-Ward et al. (241) have suggested that the glibenclamide binding site on Necturus intestine K⁺ channels resides on or near the cytoplasmic face of the channel. Aguilar-Bryan et al. (2) reported that covalent modification of SUR with sulfonylurea ligands occurred within the amino-terminal third of the protein but that further identification of the binding site had not been completed. Subsequently, a second SUR (SUR2) has been cloned which, when coexpressed with Kir6.2, exhibits pharmacological characteristics of cardiac or skeletal muscle K_ATP (185). Because the affinity of this isoform for glibenclamide is reduced (compared with SUR1) and because SUR are reported to have a single sulfonylurea binding site (146), sequence comparison or chimeric expression will likely provide indications of the binding site local. Continuing research on SUR/Kir6.2 and CFTR with this class of compounds will certainly produce SAR that will enlighten us regarding their mechanistic basis and direct the synthesis of more potent, selective, and thus more useful compounds as we attempt to understand and modulate the many functions of CFTR.

Maddy (234) synthesized SITS (Fig. 2) as a general fluorescent, impermeant covalent modifier of membrane amino groups. Subsequently, SITS and other stilbene disulfonates (25) were recognized as novel inhibitors of band 3-mediated Cl / HCO₃ anion exchange in human red blood cells (206). Much of the work done in understanding the reversible nature of inhibition by SITS revealed that the blockade of anion transport was not the result of intrinsic chemical modification, but rather of an "explicit" interaction between the disulfonic stilbene probe and an apparent binding site (58). Since then, several purely noncovalent reversibly acting (e.g., 4,4'-dinitrostilbene-2,2'-disulfonic acid, DNDS; Fig. 2) and covalent irreversibly acting (e.g., DIDS; Fig. 2) disulfonic stilbene derivatives have been designed and synthesized (58-60). These agents have played a critical role in the definitive identification of the anion exchanger protein band 3 (58-60, 222, 441).

The disulfonic stilbenes are among the most potent inhibitors of band 3- or AE1 (anion exchanger)-mediated (6) anion exchange. The inhibitory constants (IC₅₀) of various disulfonic stilbene derivatives for AE1 span a concentration range of over four orders of magnitude (0.8-500 µM, Ref. 57). Disulfonic stilbenes also inhibit the recently cloned AE2 and AE3 anion exchangers found in a variety of tissues (7, 87, 213, 216, 226). The backbone of this class of inhibitor consists of a trans-2,2'-disulfonic stilbene structure, with varying 4,4'-substituents (26, 58). A quantitative SAR of disulfonic stilbene derivatives revealed that there is a positive correlation with the Hamett constant (, a measure of the capacity to exchange electrons), and the Hansch constant (, a measure of hydrophobicity) of the 4,4'-substituents. The reversible type of inhibition has been studied dynamically using various techniques such as spectrophotometry (111, 126), fluorimetry (58, 283, 412), and NMR (117, 130), or by utilizing radioactively labeled forms of disulfonic stilbenes (58-60, 222, 351). On the other hand, the irreversible type of binding with membrane components has been studied by radioactively labeled forms of disulfonic stilbenes or by Western blotting (57). Despite the advantages of using disulfonic stilbene derivatives with noncovalently reactive 4,4'-substitutions (e.g., DNDS), most studies have used the covalently reactive derivatives DIDS or SITS. The quality of commercially available DIDS and SITS is not uniform, because some companies provide them as free acids and others as sodium salts with varying degrees of cis/trans-isomerization and different ratios of NCS and NH₂ groups. Because of the hygroscopic properties and tendency or ease of intermolecular reactions of the NCS group with NH₂ groups, especially with sodium salts of the disulfonic stilbenes, polymerization can occur yielding DIDS-DIDS, DIDS-4,4'-diaminostilbene-2,2'-disulfonic acid (DADS) (Fig. 2), and DIDS-DADS-DIDS polymers. As suggested by Racker (294) "DIDS which often `dids' more than advertised" should be used with caution.

The disulfonic stilbenes, most often DIDS, have been shown to block a wide variety of Cl channels expressed in numerous cell types (1, 17-19, 38, 42, 43, 47, 71, 72, 96, 101, 136, 138, 160, 190, 191, 200, 204, 205, 224, 225, 255, 258, 259, 267, 279, 289, 304, 308, 315-317, 341, 343, 344, 400, 401, 411, 418, 434). Remarkably, DIDS does not block transepithelial Cl secretion across rat colonic mucosa, trachea, or T84 monolayers (49, 340). Extracellular DIDS also fails to block CFTR-mediated whole cell Cl currents (83, 99, 139, 140, 265, 300, 309, 336, 371, 372, 384, 403, 431). Indeed, CFTR is one of the very few Cl channels that is not blocked by extracellular DIDS. The disulfonic stilbenes do block CFTR from the intracellular side (see below) but, because the sulfonate groups of the disulfonic stilbenes are present as fully ionized divalent anions at physiological pH (57), they do not penetrate the plasma membrane. The disulfonic stilbenes, preferably DNDS, could be used diagnostically to implicate CFTR-mediated Cl secretion. However, because one anticipates a negative result from this type of experiment, additional positive controls are needed before one can conclude Cl secretion is mediated by CFTR.

Although the disulfonic stilbenes do not appear to directly block CFTR from the extracellular side, they may by indirect means influence CFTR-mediated Cl secretion. DIDS has been reported to increase short-circuit current and serosal-to-mucosal Cl fluxes across stripped rabbit colonic mucosa (364). Basolateral addition of DIDS or SITS at a concentration between 10 and 200 µM has been reported to produce a transient increase in short-circuit current, which was followed by a gradual inhibition across T84 monolayers (48). This transient stimulation with DIDS was also observed for whole cell Cl currents in isolated T84 cells grown on permeable supports as well as the human airway cell line Calu-3 (360) and was attributed to the elevation of cytosolic Ca²⁺ levels. Although the exact mechanism by which DIDS increases Ca²⁺ levels is not known, the fact remains that the disulfonic stilbenes are capable of elevating cytosolic free Ca²⁺ in both confluent as well as nonconfluent T84 epithelial cells (48). The mechanistic basis for the gradual decrease in short-circuit current in T84 cells by the disulfonic stilbenes is also unknown but might result from alteration in intracellular pH due to the inhibition of anion exchange. Disulfonic stilbenes have also been shown to effect various epithelial K⁺ channels such as I_SK (55, 342, 407), K_ATP (128), and the Ca²⁺-activated K⁺ channel in smooth muscle cells (167).

A disulfonic stilbene-sensitive epithelial Cl channel that has received considerable attention by CF investigators is the ORCC. DNDS reversibly blocked the ORCC incorporated into planar lipid bilayers with a K_D of 2-3 µM when applied to the extracellular side, and DIDS caused an irreversible inhibition (49, 353). Structure-activity studies with the disulfonic stilbenes and a structurally similar class of compounds, the sulfonated calixarenes, have led to the development of a highly potent blocker (K_D = 0.6 nM) of ORCC, TS-TM-calix[4]arene (5,11,17,23tetrasulfonato-25,26,27,28-tetramethoxy-calix[4]arene; Fig. 2; Ref. 358). Before the discovery of CFTR, ORCC was implicated as the Cl channel whose PKA regulation was defective in CF (see Ref. 333a). Although it is now recognized that CFTR is the defective Cl channel in CF, it has been suggested that CFTR regulates ORCC and that both channels may contribute to transepithelial Cl secretion (179, 335, 336). Schwiebert et al. (335) made use of the differential sensitivity of CFTR and ORCC to DIDS blockade in short-circuit current studies on primary cultures of rat tracheal epithelial monolayers stimulated with cAMP-dependent agonists or ATP. DIDS (1 mM) and TS-TM-calix[4]arene (1 µM) inhibited 31 and 21%, respectively, of the short-circuit current stimulated by 8-bromo-cAMP (8-BrcAMP). Addition of hexokinase to remove any extracellular ATP caused a similar inhibition in the 8-BrcAMP-stimulated current. In contrast, DIDS and TS-TM-calix[4]arene inhibited nearly all, 95 and 70%, respectively, of the short-circuit current stimulated by ATP. Consistent with these results, Schweibert and co-workers (179, 335) proposed a model whereby extracellular ATP, released in response to cAMP stimulation, binds to a purinergic receptor that in turn activates ORCC. The inhibition in short-circuit current by DIDS and TS-TM calix[4]arene was interpreted as the blockade of ORCC and the remaining unblocked portion of the short-circuit current attributed to CFTR. Thus both ORCC and CFTR were suggested to contribute to transepithelial Cl secretion. There are a few caveats that warrant consideration with this interpretation. Most importantly, DIDS is a known antagonist of purinergic receptors (53, 54, 56, 104, 105, 116, 249, 257, 366, 430, 443). The inhibition constant (K_i) for DIDS blockade of various purinergic receptors ranges between 1.6 and 300 µM. Therefore, it is not clear if the inhibitory effects of 1 mM DIDS in the studies reported by Schwiebert and co-workers (335, 336) are due to the blockade of ORCC or an antagonism at the ATP receptor. In the studies by Hwang et al. (179), mucosal DIDS blocked the stimulation in short-circuit current by mucosal ATP but did not block stimulation by serosal ATP. These results support the notion that DIDS may be acting at a purinergic receptor, since mucosal DIDS will not have access to a basolateral membrane purinergic receptor but will have access to an apical membrane purinergic receptor. On the basis of the very close structural similarity between TS-TM-calix[4]arene and the disulfonic stilbenes, the inhibitory effects of the calixarene may also be due to ATP receptor antagonism. Second, we (357) and others have failed to see an inhibitory effect of the disulfonic stilbenes or TS-TM-calix[4]arene on transepithelial Cl secretion in a variety of secretory epithelia including primary cultures of rat tracheal epithelial cells stimulated with a number of different agonists. Thus, contrary to the conclusions reached by Schwiebert et al. (336) and Hwang et al. (179), we suggest that the inhibitory effects of DIDS in their studies may have an alternative explanation and that ORCC does not contribute to transepithelial Cl secretion. Glibenclamide and diphenylamine-2-carboxylate (DPC) were also used in these studies with the intention of discriminating between CFTR and ORCC. Unfortunately, glibenclamide and DPC also block ORCC (293, 324, 353) and may interfere with the cAMP signal transduction cascade (166, 214, 426), thus complicating the interpretation of the results obtained with these compounds. These concerns extend to any studies using disulfonic stilbenes (e.g., DIDS, DNDS), sulfonylureas (e.g., glibenclamide), or arylaminobenzoates [e.g., DPC, 5-nitro-2(3-phenylpropylamino)benzoate (NPPB)] on cellular-based macroscopic measurements of epithelial Cl currents.

Linsdell and Hanrahan (227) have shown that DNDS and DIDS cause a voltage-dependent block of CFTR-mediated Cl currents when applied to the cytoplasmic side of excised inside-out membrane patches from baby hamster kidney cells expressing wild-type or R347D CFTR. Extracellular DNDS or DIDS did not block the CFTR-mediated Cl currents. Inhibition from the intracellular side by DNDS displayed a voltage-dependent K_D of 62 µM at 100 mV, 111 µM at 50 mV, 465 µM at +50 mV as would be expected for block of the channel pore by a negatively charged molecule acting from the intracellular side. Fitting these data to the Woodhull equation (433) gave a K_D of 236 µM at 0 mV. Substitution of the positively charged arginine at position 347 to a negatively charged asparate significantly reduced the affinity of block of DNDS by eightfold and DIDS by threefold. Tabcharani et al. (382) had previously shown that the R347D mutation reduces the single-channel conductance, eliminates channel blockade by SCN, and abolishes anomalous mole fraction behavior seen in Cl-SCN mixtures. Tabcharani et al. (382) have suggested that R347 contributes to an important anion-binding site close to the cytoplasmic end of the channel pore. Linsdell and Hanrahan (228) suggest that CFTR channel pore may contain a relatively large inner vestibule accessible from the intracellular side to large blocking anions such as DNDS, gluconate, and glutamate and that arginine-347 may be involved in anion binding within this region of the pore. Further structure-activity studies with additional disulfonic stilbenes and perhaps calixarene derivatives together with additional mutational analysis will be useful in defining the CFTR structure at this site as well as the development of higher affinity blockers of CFTR.

The arylaminobenzoate DPC (Fig. 2) was developed by Di Stefano et al. (100) as a blocker of the basolateral membrane Cl conductance in the thick ascending limb of the loop of Henle (TAL) and in the apical membrane Cl conductance of shark rectal gland tubules (RGT). The DPC had an IC₅₀ of 26 µM when added to the basolateral side of the rabbit cortical and mouse medullary portion of the TAL (cTAL and mTAL, respectively), both of which are NaCl-reabsorptive epithelia (163, 164, 417). When added to the apical side in the RGT, a NaCl-secreting epithelia, DPC showed similar inhibition of the apical membrane Cl conductance (143, 144, 157). The inhibition by DPC on both the renal and rectal gland epithelia was shown to be rapid and reversible. Di Stefano et al. (100) have shown the presence of a DPC (100 µM)-sensitive Cl channel in excised inside-out patches from the apical membrane of RGT and the basolateral membrane of rabbit cTAL; however, the concentration dependence of DPC block was not evaluated. An SAR of 219 arylaminobenzoates led to the discovery of more potent blockers such as NPPB (Fig. 2), which inhibited the basolateral membrane Cl conductance of cTAL with an IC₅₀ of 80 nM (416). To our knowledge, the single-channel identity of a Cl channel with an IC₅₀ of 26 µM for DPC or 80 nM for NPPB has not been shown in the cTAL or any other epithelia.

Since the discovery of the arylaminobenzoates, high concentrations of DPC and NPPB have been widely used in several macroscopic assays as inhibitors of Cl transport in numerous epithelia (57). These include rabbit, canine, and sheep tracheal epithelia (189, 374), luminal membrane of the rectal gland of dogfish (145), cultured human fetal alveolar epithelial cells (244), primary cultures of human and rabbit distal colonic crypt cells (312, 313), human and rat epididymal epithelia (71, 73), mouse inner medullary collecting duct (mIMCD-K2) (403), cultured A6 renal epithelial cells (50, 67, 211, 266, 268, 269, 350), equine sweat gland epithelium (209), human jejunum and colon (52), frog skin epithelium (51), epithelial cells (intestine 407) (215), retinal pigment epithelial cells (174, 398), cultured human airway epithelial cells (242, 336), human pancreatic duct cells (411), rabbit colonic epithelia (142), mouse muscle cells (142, 414), and cultured human biliary cells (305). In all of these studies, the concentrations used were considerably higher than required to inhibit the Cl conductive pathway across cTAL and RGT, although the inhibition of Cl transport was considered to result from the blockade of Cl channels.

With the use of several microscopic assays such as planar lipid bilayer and excised membrane patches, DPC and NPPB have been shown to inhibit endogenous and heterologously expressed Cl channels in several epithelial cells. These include the ORCC from the human carcinoma cell line HT-29 (102, 161), rat colonic enterocytes (353), and cultured human respiratory cells (217), a Ca²⁺-dependent Cl channel from sheep airway epithelium (8, 9), volume-sensitive outwardly rectifying Cl channels from various epithelia (1, 17, 71, 72, 138, 215, 242), and CFTR in various epithelia (see below). Tilmann et al. (387) have shown that NPPB inhibits an ORCC from HT-29 cell line with a K_i of 0.9 µM when added to the cytosolic side and a K_i of 0.1 µM when added to the outer membrane side of the channel. The reason for this difference in the K_i values was attributed to the fact that the NPPB interaction site is likely accessible only from the extracytosolic side of the channel, so cytosolic addition precludes NPPB reaching its interaction site. However, apart from being anions, with pK values of 3-5, the arylaminobenzoates are lipophilic. At a pH of 7.4, ~50-90% are distributed into the lipid phase, e.g., DPC has a partition coefficient of 0.58 (water/CH₂Cl₂) (100), so by the addition of these compounds on one side of the cell membrane, one cannot rule out their penetration across the cell membrane. Indeed, we have demonstrated that both DPC and NPPB showed similar inhibition from the extracellular and intracellular sides of the ORCC incorporated into planar lipid bilayer with K_i values of 600 and 25 µM, respectively (353).

The concentrations at which the arylaminobenzoates DPC and NPPB have been used in different studies do raise a question about their specificity to inhibit one type of Cl channel as compared with other channels or intracellular processes. The arylaminobenzoates share some structural similarity with loop diuretics. In the first paper describing the SAR of this class of compounds, it was shown that some of the arylaminobenzoates also had an affinity for the Na⁺-K⁺-2Cl cotransporter (416). For example, DPC and NPPB had IC₅₀ values of 100 and 30 µM, respectively, for inhibiting the cotransporter. Interestingly, the loop diuretics furosemide and bumetanide have recently been shown to block CFTR (296, 408). Furosemide and bumetanide have K_i values of 40 and 5 µM, respectively, for the inhibition of CFTR (408). Arylaminobenzoates have also been shown to block nonselective cation channels (75, 132, 133, 290, 301), L-type Ca²⁺ channels (415), volume-sensitive basolateral K⁺ channels in HT-29/B6 cells (180), an inwardly rectifying Ca²⁺-dependent K⁺ channel in turtle colon (301), and a Ca²⁺- and cAMP-activated low-conductance K⁺ channel in the basolateral membrane of human colonic crypt cells (229). It is especially important to consider the inhibition of basolateral membrane K⁺ channels and the basolateral membrane cotransporter by these compounds when studying Cl secretion. In a typical Cl secretory epithelia, the apical membrane Cl conductance and the basolateral membrane K⁺ conductance are tightly coupled to maintain a sustained level of Cl secretion. Inhibition of either apical membrane Cl channels, basolateral membrane K⁺ channels, or the basolateral membrane cotransporter will inhibit Cl secretion. Hence, where investigators have intended to used DPC as an inhibitor of apical membrane CFTR to differentiate between cAMP-activated Cl secretion via CFTR and other Cl channels, there could well have been an inhibition of the basolateral membrane K⁺ conductances or cotransporter that is the actual cause of the inhibition of Cl secretion.

Diphenylamine-2-carboxylate, NPPB, and some other arylaminobenzoates have been shown to have considerable inhibitory effects on intracellular adenylyl cyclase in the Cl secretory human colonic cell lines HT-29/B6 and T84 (166, 214, 426). Kreusel et al. (214) have shown that 1 mM DPC inhibited >85% of forskolin (10 µM)-activated cAMP production in HT-29/B6 cells. Diphenylamine-2-carboxylate was unable to inhibit Cl secretion activated by dibutyryl cAMP. In a similar study with the T84 cell line, 500 µM DPC or NPPB inhibited forskolin-stimulated cAMP production by 28 and 56%, respectively (166). Thus it is possible that the mechanism of action of DPC in this and other cellular-based studies results from the inhibition of cAMP production rather than a direct inhibition on Cl channels. Diphenylamine-2-carboxylate has also been reported to inhibit prostaglandin D₂ synthesis from arachidonic acid in primary cultures of canine tracheal epithelium (375). The likely site of action of DPC in this tissue is the inhibition of cyclooxygenase, the enzyme responsible for prostaglandin synthesis. Prostaglandins are important modulators of electrolyte transport in a number of epithelia (33, 233, 237, 262, 292, 295, 365, 391, 413). Thus the arylaminobenzoates appear to inhibit both the cAMP and eicosanoids signal transduction pathways. The arylaminobenzoates have also been shown to have profound effects on intracellular pH in LLC-PK₁ cells (50) and could thereby modulate a number of intracellular processes and channel activities. The interpretation of arylaminobenzoate inhibition of macroscopic Cl secretion is, at best, difficult because of their nonselectivity for Cl channels coupled with their inhibition of basolateral membrane K⁺ channels and the Na⁺-K⁺-2Cl cotransporter. As with the disulfonic stilbenes, the use (abuse) of the arylaminobenzoates as Cl channel blockers has become entrenched in the literature. Clearly, more specific reagents are needed to study the contribution of CFTR in anion secretion by intact epithelia.

Caution aside, DPC and NPPB have been shown to inhibit CFTR (11, 61, 83, 86, 139, 140, 243-245, 291, 300, 309, 336, 376, 381, 384, 403), and in a series of elegant studies by McCarty et al. (243) and McDonough et al. (245), DPC was successfully used to probe the conduction pathway of CFTR. In their initial studies, McCarty et al. (243) demonstrated that DPC and flufenamic acid (FFA; 3'-trifluoromethyldiphenylamine-2-carboxylic acid; Fig. 2) blocked CFTR-mediated whole cell and single-channel Cl currents in Xenopus oocytes. Blockade was voltage dependent; currents at positive potentials were not affected, but currents at negative potentials were blocked. Both DPC and FFA blocked single-channel openings in an excised patch when applied directly to the cytoplasmic side of the channel. As expected from the whole cell data, DPC and FFA blockade of CFTR in excised patches was also voltage dependent. The onset of blockade by extracellularly applied DPC or FFA was biphasic, with an early rapid phase and a later phase that developed over several minutes. The slow development of the blockade was attributed to the permeation of the blocker into the cell to an intracellular binding site on CFTR. Blockade by DPC and FFA was fully reversible. With the use of the voltage dependence of the blockade and the Woodhull equation (433), the K_D for DPC at 0 mV was 912 µM and at 100 mV was 237 µM. Similarly, the K_D for FFA at 0 mV was 1.22 mM and at 100 mV was 289 µM. The apparent electrical distance sensed by both the blockers was 41% as measured from the inside of the membrane. The studies of McDonough et al. (245) extended these findings to demonstrate that the interaction of DPC with CFTR was consistent with an open-channel mechanism of blockade. Furthermore, site-directed mutagenesis of residues in the putative transmembrane segments (TM) 6 and 12 significantly altered DPC blockade of the channel. Most notably, mutation of serine-341 to an alanine caused a fivefold increase in the K_D at 100 mV (wild type, 276 µM vs. S341A, 1,251 µM). This result is of special interest, since the predicted position of residue 341 lies 40% through the proposed TM6. Although S1141 in TM12 is predicted to have a similar position as S341 in TM6, mutation of serine-1141 to an alanine did not show an appreciable effect on DPC binding. Furthermore, when the methionine and threonine residues immediately adjacent to S1141 were changed to isoleucine and phenylalanine to match the residues immediately adjacent to S341, DPC bound with an affinity close to that of the wild-type channel (S341A-M1140I-T1142F). This showed that the DPC-binding site on TM6 could be transferred to TM12. Mutation of threonine residue 1134 to a phenylalanine caused a threefold improvement in the affinity for DPC (T1134F, 74 µM). Like residue 341, residue 1134 lies 40% through the proposed TM12. These results strongly support the notion that both TM6 and TM12 contribute to forming the pore. They speculated that the carboxy group of DPC interacted with S341 on TM6 and the phenyl ring with T1134 on TM12. The studies of McCarty et al. (243), and McDonough et al. (245) with DPC, Linsdell and Hanrahan (227) with DNDS, and Schultz and co-workers (324, 329), Sheppard and Robinson (346), and Venglarik et al. (409) with sulfonylureas provide excellent illustrations of how one can judiciously use these small ligands to probe the structure and kinetics of CFTR channel activity.

	III. CHANNEL OPENERS

Top Previous Next References

Paleolithic man is credited with the discovery of the central nervous system (CNS) stimulatory effects of drinks made from alkylxanthine (caffeine, theophylline, theobromine; Fig. 3) containing plants (coffee, tea, cocoa) (303). Alkylxanthines are now known, in addition to their CNS effects, to relax smooth muscle, most notably bronchial smooth muscle, stimulate cardiac muscle, and act on the kidney to cause diuresis. The treatment of asthmatic patients with strong coffee was initiated more than 100 years ago (314). Recent clinical studies suggest CF patients may also benefit from the use of bronchodilators such as the alkylxanthines (82). The cellular sites of action of the alkylxanthines include the mobilization of intracellular Ca²⁺ (115), the inhibition of phophodiesterases (28), the inhibition of phosphatases (81, 118), and the blockade of adenosine receptors (192). Adenosine receptor blockade is thought to be the relevant site of action at dietarily achieved alkylxanthine concentrations (50 µM). Ten- to 20-fold higher concentrations of caffeine or theophylline are required to increase intracellular Ca²⁺ or inhibit PDE. The inhibition of PDE to elevate cAMP levels has been the often intended reason for using alkylxanthines, especially 3-isobutly-1-methyxanthine (IBMX; Fig. 3), in epithelial transport studies. However, Becq and co-workers (29, 30) have also made use of the phosphatase inhibitory effects of IBMX to study the phosphorylation-dependent regulation of CFTR.

View larger version (20K): [in this window] [in a new window]   FIG. 3.   Structures of compounds that have been used to stimulate Cl secretion by directly or indirectly modulating activity of cystic fibrosis transmembrane conductance regulator.

The cAMP-mediated activation of Cl secretion has been recognized for more than 30 years (24, 154) and the failure of CF epithelia to respond to cAMP for nearly 20 years. Studies before the discovery of CFTR demonstrated the impaired responsiveness of CF epithelia lies distal to the formation of cAMP and the activation of PKA. Soon after its discovery, CFTR was shown to be a PKA-activated ATP-dependent Cl channel as reviewed in detail in References 129a and 333a. Salient to CFTR activation by alklyxanthines, McPherson et al. (251) were the first to demonstrate that IBMX partially restored amylase and mucin secretion in submandibular acinar cells from CF patients. Subsequently, Drumm et al. (103) were the first to show that CFTR constructs with naturally occurring mutations in the first NBF could be activated by high concentrations of IBMX. In their studies, Xenopus oocytes were injected with wild-type or mutant CFTR (e.g., F508, G551D) mRNA, and Cl current was measured in response to a stimulation cocktail including 10 µM forskolin, 200 µM 8-(4-chlorophenylthio)-cAMP, and various concentrations of IBMX. Mutant CFTR were less sensitive than wild-type CFTR to IBMX. For example, oocytes expressing wild-type CFTR were nearly completely stimulated with 1 mM IBMX, but those expressing G551D or F508 CFTR required 5 mM IBMX to achieve complete stimulation. The reduction in sensitivity of the mutant CFTR was also found to be correlated with the severity of the CF in patients carrying the corresponding mutations.

The differential sensitivity of CFTR bearing mutations in NBF1 or NBF2 to IBMX stimulation was further evaluated by Smit et al. (362). Mutation of a conserved glycine (G551 and G1349) in the putative linker domains of either NBF reduced the sensitivity to IBMX, and mutations of this site in both NBF produced an additive effect. In contrast, substitutions in the Walker A and B motifs produced strikingly different effects in NBF1 and NBF2. Substitutions for the conserved lysine (K464, Walker A) or asparate (D572, Walker B) in NBF1 resulted in a marked decrease in sensitivity to IBMX, whereas the same changes in NBF2 (K1250, Walker A; D1370, Walker B) produced an increase in sensitivity. Smit et al. (361) went on to show that the missense mutation G480C associated with CFTR protein mislocalization was equally sensitive to IBMX stimulation when compared with wild-type CFTR. Wilkinson et al. (428) undertook a quantitative analysis of the rates of activation and inactivation of these same mutants in response to stimulation and removal of 10 µM forskolin and 5 mM IBMX. Consistent with the steady-state current measurements of their previous studies, substitutions at G551 in NBF1 or G1349 in NBF2 reduced the rate of activation and increased the rate of inactivation. Substitutions at K464 or K1250 also reduce the rate of activation but had opposite effects on the rate of inactivation; K464 substitution increased the rate of inactivation while K1250 substitution decreased the rate of inactivation. Substitutions of D572 decrease the rate of activation and increased the rate of inactivation. In contrast, D1370 substitutions did not alter the activation rate but markedly slowed the inactivation rate. Thus D1370 mutants remained active long after the removal of the stimulation cocktail. These elegant macroscopic measurements demonstrated that mutations in either NBF1 or NBF2 can influence the activation and inactivation of CFTR and suggest the nature or the exact consequences of nucleotide binding differ for the two domains (428), observations that have since been demonstrated in numerous patch-clamp studies.

An important outcome of the above studies in oocytes was the demonstration that mutant forms of CFTR could indeed be activated and function as Cl channels. These results thus lent great support to the suggestion by McPherson et al. (251) that alkylxanthines could be useful in the treatment of CF. This notion received further support from the studies of Yang et al. (440), who extended the observations in oocytes to murine fibroblast cells (L cells) expressing mutant forms of CFTR. With the use of the halide-sensitive fluorophore SPQ assay, 4 mM IBMX was found to activate F508 and G551D CFTR-mediated anion efflux. Activation by IBMX of the F508 CFTR expressing cells was observed in cells maintained at 37°C, indicating some F508 CFTR protein had reached the plasma membrane. Similar stimulatory effects of IBMX were also reported by Haws et al. (159) using mouse mammary epithelial cells (C127 cells) expressing F508 CFTR. Haws et al. (159) also demonstrated, with patch-clamp studies, that F508 CFTR was present in the plasma membrane albeit at a lower channel density than wild-type CFTR-expressing cells and with a lower open probability (P_o) of 0.11 compared with wild-type CFTR (P_o = 0.33).

Beavo (28) has provided a recent review of the cyclic nucleotide PDE. An estimated 25 PDE isoforms are thought to exist. The pharmaceutical industry has attempted to capitalize on the tissue-specific expression of the various PDE isoforms in the development of antiasthmatic, antithrombotic, antihypertensive, cardiotonic, and antidepressant agents. There are, as a result of these drug development efforts, a number of isoform-specific PDE inhibitors that are in clinical use. Kelly et al. (198) set out to determine which of the specific classes of PDE were involved in the activation of CFTR in epithelial cells. Milrinone and amrinone (Fig. 3), class III PDE inhibitors, were found to stimulate ¹²⁵I efflux in Calu-3 and 16HBE human airway epithelial cells, whereas the class IV PDE inhibitor rolipram and the class V PDE inhibitor dipyridamole were much less effective. Stimulation of ¹²⁵I efflux by milrinone and amrinone did not require the inclusion of an adenylyl cyclase activator (e.g., forskolin), nor did stimulation correlate with cAMP levels. However, stimulation by milrinone and amrinone was inhibited by the cell-permeant PKA inhibitor N-(2-[methylamino]ethyl)-5-isoquinolinesulfonamide (H-8) and the cAMP antagonist R_p-cAMP. These results are consistent with a cAMP/PKA-mediated activation of CFTR that results from a compartmentalized pool of cAMP. These studies were extended to show that ³⁶Cl efflux could be stimulated by milrinone plus isoproterenol in the transformed nasal polyp cells (CF-T43) homozygous for F508-CFTR (197). The CFTR antisense oligonucleotides prevented the increase in ³⁶Cl efflux in response to milrinone and isoproterenol. These results again demonstrate that some F508 CFTR is functionally expressed in the plasma membrane and that it can be activated by milrinone and isoproterenol. Kelly et al. (199) have since shown the efficacy of forskolin and milrinone to hyperpolarize the nasal epithelium indicative of a Cl secretory response in F508 CFTR-expressing mice. Whereas the combination of forskolin and milrinone was ineffective in altering the nasal potential difference in the CFTR (/) mice, the F508 CFTR-expressing mice displayed a change in potential difference of ~50% of that observed in mice expressing at least one wild-type CFTR allele. As in the in vitro studies with F508 CFTR-expressing cells, both an adenylyl cyclase agonist and milrinone were required to cause a response in vivo.

Collectively, these studies on human salivary acinar cells, Xenopus oocytes, human airway cells, murine fibroblasts, and murine nasal epithelia suggest the superactivation of the cAMP-PKA regulatory pathway may activate certain mutant forms of CFTR and thus be of therapeutic utility in the treatment of CF. However, there are a few caveats that bear consideration. In addition to IBMX inhibition of PDE, one must also consider the inhibition of phosphatases as a possible site of action as demonstrated by Becq and co-workers (29, 30) (see below). If phosphatase inhibition or a combination of phosphatase inhibition and PDE inhibition is the mechanism, then one must modify the interpretation of these results, and consequently, a different strategy will be required to optimize this therapeutic approach. The studies with milrinone were performed at a single high concentration of 100 µM. The K_i of milrinone is 0.3 µM for inhibition of class III PDE (28). Milrinone concentration-response studies as well as the biochemical demonstration of the presence of class III PDE in the apical membrane of airway epithelial cells are needed before one can conclude the stimulatory effects of milrinone are mediated by inhibition of class III PDE. In addition, the assumed activation of the cAMP-PKA regulatory pathway by IBMX or milrinone is expected to lead to the phosphorylation of the mutant CFTR proteins. Protein phosphorylation studies are needed to demonstrate that IBMX and milrinone do alter the phosphorylation status of CFTR under the experimental conditions used in these studies. Positive results from such biochemical studies would lend great support to the conclusions reached by these investigators. These studies must also be reconciled with the observations of Grubb et al. (148), who evaluated the combined use of adenylyl cyclase activators (forskolin or isoproterenol) and IBMX on normal and CF airway epithelia in vitro and in vivo. High concentrations of IBMX (5 mM) did not augment forskolin-stimulated Cl secretion but instead inhibited Cl secretion in primary cultures from normal patients. Neither forskolin nor forskolin plus IBMX had any effect in cells from CF patients with the F508 mutation even in the presence of a favorable Cl gradient. Nasal potential difference measurements failed to detect an additive effect of IBMX with isoproterenol in either normal or CF subjects. Grubb et al. (148) concluded that the combination of adenylyl cyclase activators and IBMX is not effective in initiating Cl secretion in CF epithelia. These authors also note that agents that raise cell cAMP in CF airways may further increase the abnormally high basal rate of sodium absorption. Thus Cl secretagogues that elevate cAMP levels may be counterproductive in the treatment of CF airways (148).

The studies of Becq et al. (29) suggest an alternative explanation for the effects of high concentrations of IBMX. These investigators observed that channel inactivation (rundown) that is often observed upon patch excision could be slowed by theophylline, IBMX, and levamisole (Fig. 3) in wild-type CFTR-expressing CF pancreatic cell line CFPAC-PLJ-CFTR-6. These same substances reduced the apical membrane-associated alkaline phosphatase activity by 70-75%. A polyclonal antialkaline phosphatase antibody that detected and reduced apical membrane alkaline phosphatase activated quiescent CFTR Cl channels. Subsequently, Becq et al. (30) showed in CFTR expressing Chinese hamster ovary (CHO) and human airway epithelial cells that the alkaline phosphatase inhibitors bromotetramisole, IBMX, theophylline, and vanadate (Fig. 3; each at 1 mM) slowed the rundown of CFTR channel activity in excised membrane patches. These same substances also reduced the dephosphorylation of CFTR protein in isolated membranes. 3-Isobutyl-1-methylxanthine was the most effective at slowing channel rundown. Caffeine (Fig. 3) and dipyridamole, two PDE inhibitors that only poorly inhibit phosphatases, were ineffective in slowing rundown as was okadaic acid (10 µM), a type 1 and 2A protein phosphatase inhibitor. ()-p-Bromotetramisole and IBMX also activated wild-type and the disease-causing mutant CFTR, R117H, G551D, and F508 in cell-attached patches. We have also observed that IBMX slows the rundown of F508-CFTR in excised membrane patches from L cells (322, 326). These results in excised membrane patches and isolated membrane vesicles cannot be explained by the inhibition of PDE but rather suggest the inhibition of membrane-associated phosphatases, perhaps alkaline phosphatase, as the site and mechanism of action. However, on a cautionary note, one must consider the concentrations of the substances used in these studies. ()-p-Bromotetramisole, the most potent alkaline phosphatase inhibitor used by Becq et al. (29) at 1 mM, has a K_i of 1 µM for alkaline phosphatase (254). In the enzymatic studies of Farely et al. (118), IBMX and theophylline had K_i values of 600 and 200 µM, respectively, for alkaline phosphatase inhibition. The patch-clamp studies of Becq et al. (30) evaluated these compounds at 1 mM and suggest IBMX was the more potent compound at slowing CFTR channel rundown. Therefore, concentration-response studies that show a EC₅₀ on CFTR channel activity consistent with the K_i for alkaline phosphatase inhibition would aid in delineating the site and mechanism of action of these substances.

The direct binding of the alkylxanthines to CFTR also warrants some consideration when attempting to understand their mechanism of action on CFTR channel gating. We and others (322, 428) have observed that IBMX causes a decrease in the single-channel amplitude of CFTR and thus appears to act as an open-channel blocker. Thus IBMX will by mass action hold CFTR in the open (albeit partially blocked) state. If only closed CFTR channels can be dephosphorylated, IBMX, by stabilizing the open state, will slow dephosphorylation and channel inactivation (rundown). In support of