Identification of the cystic fibrosis transmembrane conductance regulator domains that are important for interactions with ROMK2.

In addition to functioning as a cAMP-activated chloride channel, the cystic fibrosis transmembrane conductance regulator (CFTR) plays an important role in conferring regulatory properties on other ion channels. It is known, with respect to CFTR regulation of ROMK2 (renally derived K(ATP) channel), that the first transmembrane domain and the first nucleotide binding fold domain (NBF1) of CFTR are necessary for this interaction to occur. It has been shown that under conditions that promote phosphorylation, the ROMK2-CFTR interaction is attenuated. To elucidate the complex nature of this interaction, CFTR constructs were co-expressed with ROMK2 in Xenopus oocytes, and two microelectrode voltage clamp experiments were performed. Although the second half of CFTR can act as a functional chloride channel, our results suggest that it does not confer glibenclamide sensitivity on ROMK2, as does the first half of CFTR. The attenuation of the ROMK2-CFTR interaction under conditions that promote phosphorylation is dependent on at least the presence of the R domain of CFTR. We conclude that transmembrane domain 1, NBF1, and the R domain are the CFTR domains involved in the ROMK2-CFTR interaction and that NBF2 and transmembrane domain 2 are not essential. Lastly, the R domain of CFTR is necessary for the attenuation of the ROMK2-CFTR interaction under conditions that promote phosphorylation.

In addition to functioning as a cAMP-activated chloride channel, the cystic fibrosis transmembrane conductance regulator (CFTR) plays an important role in conferring regulatory properties on other ion channels. It is known, with respect to CFTR regulation of ROMK2 (renally derived K ATP channel), that the first transmembrane domain and the first nucleotide binding fold domain (NBF1) of CFTR are necessary for this interaction to occur. It has been shown that under conditions that promote phosphorylation, the ROMK2-CFTR interaction is attenuated. To elucidate the complex nature of this interaction, CFTR constructs were co-expressed with ROMK2 in Xenopus oocytes, and two microelectrode voltage clamp experiments were performed. Although the second half of CFTR can act as a functional chloride channel, our results suggest that it does not confer glibenclamide sensitivity on ROMK2, as does the first half of CFTR. The attenuation of the ROMK2-CFTR interaction under conditions that promote phosphorylation is dependent on at least the presence of the R domain of CFTR. We conclude that transmembrane domain 1, NBF1, and the R domain are the CFTR domains involved in the ROMK2-CFTR interaction and that NBF2 and transmembrane domain 2 are not essential. Lastly, the R domain of CFTR is necessary for the attenuation of the ROMK2-CFTR interaction under conditions that promote phosphorylation.
The CFTR 1 gene encodes for a multifunctional transmembrane protein that is both a cAMP-activated chloride channel and a regulator of other ion channels (1,2). It is a member of the ATP binding cassette superfamily, and like other members of this group it is comprised of transmembrane spanning domains and nucleotide binding folds (NBFs); however, it has a regulatory domain (R domain), which is unique to CFTR (3). The domains of CFTR involved in its regulation of other ion channels may be functionally distinct from those involved in its Cl Ϫ channel function.
CFTR expression has been linked to protein kinase A regulation of the outwardly rectifying chloride channel (ORCC), the epithelial sodium channel (ENaC), and a K ATP channel, ROMK2 (4,5,6). The first nucleotide binding domain of CFTR and the R domain are important for it's regulation of ORCC and ENaC (7,8). For CFTR to regulate ROMK2 channels, in vitro, an intact NBF1 is required (9). Although many groups are investigating the role of NBF1 and NBF2 with regard to channel gating, their role with regard to regulation of other ion channels has yet to be elucidated. For instance, it is not known if NBF2 can have the same effect as NBF1 with respect to regulation of ROMK2 channels.
Many of the disease-producing mutations are located in the NBF domains of CFTR, predominantly within NBF1. These mutations are often associated with abnormal CFTR channel activity, as well as abnormal kinase regulation of ORCC and ENaC. For example, the protein kinase activation of ORCCs depends on CFTR expression and is affected by mutations in NBF1, as is the cAMP-dependent protein kinase-dependent inhibition of ENaC (4,10). The majority of cAMP-dependent protein kinase phosphorylation sites are located in the R domain of CFTR (11). It is not known what role the phosphorylation-rich R domain plays in these processes. With respect to the ROMK2-CFTR interaction, it has been shown that this interaction is attenuated under conditions that promote phosphorylation. It is not known if the R domain of CFTR is important in this phosphorylation-dependent attenuation (6).
To examine further the structure-function relationships of the NBFs and the R domain, with respect to the ability of CFTR to act as a regulator of ROMK2 channels, CFTR constructs (see Fig. 1) and ROMK2 were co-expressed in Xenopus oocytes, and two-microelectrode voltage clamp techniques were applied. First, to determine if the R domain of CFTR is involved in the phosphorylation process that attenuates ROMK2/CFTR coupling, we co-expressed ROMK2 with CFTR constructs containing truncations before and after the R domain. Next, we substituted SUR1 for CFTR in our ROMK2/CFTR co-expression experiments to learn more about the dephosphorylation/phosphorylationdependent ROMK2-CFTR interaction. SUR1 is an ABC transporter that does not contain an R domain and is known to regulate another K ATP channel (12). Lastly, we investigated if CFTR constructs containing NBF2 alone can confer glibenclamide on ROMK2.

Preparation of Oocytes for Voltage Clamp Experiments-Stage V-VI
Xenopus laevis oocytes were isolated and injected with RNA as described previously (6). RNA concentrations were adjusted such that a 1:4 molar ratio was preserved between ROMK2 and the ABC transporter (50 nl of total RNA was injected). Experiments were performed on days 1-6 after injection.
The experimental protocol consisted of a 10-min equilibration in control solution, impalement, a 5-min control period, a 15-min exposure to 500 M glibenclamide (a sulfonylurea) in standard bath solution followed by a 3-min exposure to 2 mM barium (Ba 2ϩ ) before a control solution wash (5 min) (Fig. 2). All experiments were done under two conditions (i) basal and (ii) after stimulation with 100 M forskolin (FSK), 1 mM 3-isobutyl-1-methylxanthene (IBMX)(Sigma), a mixture used to maximize cAMP-dependent phosphorylation processes (13,14). FSK/IBMX was added to the preincubation, control, and glibenclamidecontaining solutions. Glibenclamide was diluted to the required concentration from a 100 mM stock solution dissolved in an ethanol/dimethyl sulfoxide (Me 2 SO) mixture 2:1 v/v ratio. Stock solutions of FSK (20 mM) and IBMX (200 mM) were prepared with Me 2 SO as the vehicle.
To address the possibility that the current observed, under conditions that promote phosphorylation, was stimulated Cl Ϫ current and not K ϩ current, a subset of experiments were subjected to an alternate protocol. It contained an additional 3-min barium pulse after the initial control period. These experiments were performed on oocytes expressing ROMK2/CFTR-K593X and ROMK2/CFTR-D835X and showed that the current was Ba 2ϩ -sensitive K ϩ current and not stimulated Cl Ϫ current.
For oocytes injected with ROMK2 cRNA, only cells expressing Ն500 nA whole-cell current were selected. Experiments were discounted if a stable baseline current was not obtained (Ͼ10% variation in control current). Data were compared using a paired Student's t test within a single experiment or with an analysis of variance (ANOVA) when comparing multiple conditions and/or groups. A p value Ͻ0.05 was considered significant.
CFTR Constructs and Method of Mutagenesis-Cells were injected with ROMK2 alone, ROMK2/CFTR-WT, ROMK2/SUR, ROMK2/CFTR-D835X (a CFTR construct with an intact nucleotide binding fold and a functional R domain), ROMK2/CFTR-K593X (a CFTR construct with an intact nucleotide binding fold but no R domain), or ROMK2/ RT2N2CFTR (a CFTR construct containing the R domain, the second nucleotide binding fold domain, the second transmembrane domain, and the first 159 bases of CFTR-WT so as to include the Kozak methionine for translation initiation) ( Fig. 1). Site-directed mutagenesis was performed as described by Kunkel et al. (15). Mutations were created in the CFTR clone pBQ4.7 by standard oligonucleotide-directed mutagenesis of single-stranded DNA using the Muta-Gene Phagemid in vitro mutagenesis kit as described previously (16,17). The oligonucleotides used for mutagenesis were CFTR-K593X, 5Ј-CTGTTAACTGATGGCT-AGCAAACTAGG-3Ј and CFTR-D835X, 5ЈCACGAAAAGTGTCACTGG-CCCCTCAGGCAAACTTCGATATATTACTGTCCACAAGAGCTTAAT-TTTGTGC-3Ј. The RT2N2CFTR mutation was constructed as outlined by Devidas et al. (18). The mutations were confirmed by DNA sequencing. To prepare cRNA, the plasmid was linearized with appropriate restriction enzymes and transcribed in vitro using T7 RNA polymerase (mMessage mMachine, Ambion) as described previously (9,16). CFTR-D835X and CFTR-K593X are expressed at membrane level, in varying expression systems (9,19). In addition it has been demonstrated that the second half of CFTR forms a functional chloride channel (18). These experiments were repeated, and confirmed RT2N2CFTR is expressed at membrane level with 0.278 Ϯ 0.041 A of current (V HOLD ϭ Ϫ40 mV using a two-microelectrode voltage clamp) (p Ͻ 0.05 when compared with an uninjected oocyte) (Fig. 2).

RESULTS
Treatment with FSK/IBMX, a Mixture That Increases Intracellular cAMP, Attenuates the ROMK2-CFTR Interaction-To assess how conditions that promote phosphorylation affect the ROMK2-CFTR interaction, ROMK2 and CFTR-WT were coexpressed in Xenopus oocytes and exposed to 100 M/1 mM FSK/IBMX. Under these conditions the majority of observed current was Ba 2ϩ -sensitive K ϩ current. This current was insensitive to glibenclamide (9 Ϯ 2.9% inhibition, n ϭ 9) (Fig. 3). This finding was similar to the glibenclamide insensitivity observed when ROMK2 was expressed alone (17 Ϯ 4.3% inhibition under basal conditions, n ϭ 11, 26 Ϯ 6% inhibition with FSK/IBMX, n ϭ 6) ( Table I, Fig. 4). These data support the previous observation made in excised patches that cAMP-dependent protein kinase phosphorylation can uncouple the ROMK2-CFTR interaction. This finding can be contrasted to the enhanced glibenclamide sensitivity of the resultant K ϩ current when ROMK2 is co-expressed with CFTR-WT in the absence of FSK/IBMX (55 Ϯ 5.7% inhibition, n ϭ 19) (p Ͻ 0.0001) ( Table I, Fig. 4). These data suggest that phosphorylation of CFTR, ROMK, or both uncouples this interaction and led us to investigate which domains of CFTR participate in this process.

The Presence of an R Domain Is Required for Attenuation of the ROMK2-CFTR Interaction under Conditions That Promote
Phosphorylation-To investigate, under conditions that maximize intracellular cAMP, which domain of CFTR is important for the attenuated ROMK2-CFTR interaction, CFTR constructs with mutations involving the R domain were chosen to be co-expressed with ROMK2. When ROMK2 was co-expressed with a CFTR construct containing transmembrane domain 1 and NBF1 without an R domain (CFTR-K593X) and exposed to FSK/IBMX, the resultant Ba 2ϩ -sensitive K ϩ current was in-  (20). Each of these ABC transporters was co-expressed with ROMK2 in Xenopus oocytes. hibited by glibenclamide (66 Ϯ 4.3% inhibition, n ϭ 18) ( Table  I, Figs. 2-4). Without cAMP stimulation glibenclamide inhibited 46 Ϯ 4.5% (n ϭ 6) of K ϩ current when ROMK2 and CFTR-K593X were co-expressed, which is similar to previous findings. These data suggest that CFTR constructs without the R domain or NBF2 can influence ROMK2 currents. In addition the data suggest there is no attenuation of the interaction by cAMP stimulation when the R domain is absent. Furthermore, after cAMP stimulation there is a significant enhancement of glibenclamide effect on ROMK2/CFTR-K593X currents (Table I).
However, when ROMK2 and CFTR-D835X (a CFTR construct with an intact nucleotide binding fold and a functional R domain) were co-expressed and stimulated with the cAMPenhancing mixture, the resultant K ϩ current demonstrated an attenuated sulfonylurea response (26 Ϯ 5% inhibition, n ϭ 16) (Fig. 4). In the absence of FSK/IBMX there was a 48 Ϯ 10% inhibition of K ϩ current when ROMK2 and CFTR-D835X were co-expressed (n ϭ 9) (Table I, Figs. 2-4). Taken together, these data suggest that phosphorylation uncouples the ROMK2-CFTR interaction in the presence of an R domain.
To further assess the effect of an R domain on this interaction, SUR1 (pancreatic ␤ cell sulfonylurea receptor), an ABC transporter that does not contain the R domain (20,21), was substituted for CFTR in the ROMK2-ABC transporter interaction. SUR1 can substitute for CFTR in imparting sulfonylurea sensitivity on ROMK2 potassium current with 55 Ϯ 9% inhibition, (n ϭ 10) (Table I, Fig. 4), in agreement with previous findings (22). After stimulation with FSK/IBMX the ROMK2-SUR interaction retained sulfonylurea sensitivity, that is, 55 Ϯ 6.5% of the Ba ϩ -sensitive K ϩ current was inhibited by glibenclamide (n ϭ 7). This suggests that the ROMK2/SUR proteins remain functionally interactive under conditions that increase cAMP-dependent phosphorylation unlike the ROMK/CFTR-WT and ROMK2/CFTR-D835X proteins. However, in contrast to the ROMK2/CFTR-K593X currents, glibenclamide inhibition is not enhanced after cAMP stimulation.
Can the Second Half of CFTR Substitute for the First Half in Conferring Channel Regulation upon ROMK2?-To assess if the second half of the CFTR protein can confer properties of channel regulation on ROMK2, a CFTR construct comprised of the second half of the molecule was substituted for CFTR-WT in the ROMK2-CFTR interaction. When RT2N2CFTR was coexpressed with ROMK2, the resultant Ba 2ϩ -sensitive K ϩ current remained glibenclamide insensitive (19 Ϯ 3.1% inhibition, n ϭ 10) (Table I, Fig. 5). This suggests that co-expression of the RT2N2CFTR construct with ROMK2 does not alter the sulfonylurea sensitivity of the K ϩ current. Conditions that promote phosphorylation did not alter the glibenclamide insensitive  A, B, and E) in Xenopus oocytes, using two-microelectrode voltage clamp techniques. Shown are outward currents elicited at V HOLD Ϫ60 mV plotted against time. After an initial control equilibration period (5 min), the oocyte was perfused with control solution containing 0.5 mM glibenclamide (Glib). After 14 min of glibenclamide exposure the oocyte was then perfused with 2 mM barium (to determine total K ϩ current), and this was followed with a wash off period with control solution. An identical protocol was followed for each condition with 100 M forskolin and 1 mM 3-isobutyl-1-methylxanthine added to all solutions, including preincubation. In a subset of experiments (C and D), to determine that the majority of current after FSK/IBMX application was still K ϩ current and not Cl Ϫ current, generated from activated CFTR, a barium pulse is applied and subsequently washed off with control solution before the application of glibenclamide.

FIG. 3. A representative family of whole cell currents from oocytes injected with either ROMK2/CFTR-D835X or ROMK2/ CFTR-K593X, and traces under basal conditions (in the absence of FSK/IBMX) and after stimulation with FSK/IBMX (100 M/1 mM) are compared.
Oocytes were held at Ϫ60 mV and thereafter pulsed for 20 ms from Ϫ100 to 40 mV in 20-mV increments, and currents elicited are shown for control, after glibenclamide and during barium application for each condition. It demonstrates that in ROMK2/ CFTR-K593X-injected oocytes, glibenclamide inhibition of whole cell K ϩ current remains prominent, despite application of FSK/IBMX, but that in ROMK2/CFTR-D835X-injected oocytes the FSK/IBMX mixture attenuates this inhibitory response.
(29 Ϯ 5.3% inhibition, n ϭ 6). Therefore, although the second half of CFTR can act as a functional chloride channel (18), these data indicate that it does not act as a conductance regulator of ROMK2 K ϩ channels. DISCUSSION Both the first and second halves of the CFTR protein can act as functional chloride channels (18,23); however, they do not have the same function when it comes to conferring properties of channel regulation on ROMK2 channels. The first nucleotide binding fold along with the first transmembrane domain is necessary for the ROMK2-CFTR interaction to occur (9); however, the attenuation of the ROMK2-CFTR interaction, under conditions of phosphorylation, only occurs in the presence of an R domain. It suggests that the R domain is not essential for this coupling to take place; however, it is necessary for the regulation of this interaction. The data presented also suggest that NBF2 is not involved in the ROMK2-CFTR interaction.
It is unknown whether the ability of CFTR to act as a channel regulator is because of a direct protein-protein interaction or another process. Studies on the CFTR regulation of ORCC suggest that the NBF1 of CFTR is an essential domain TABLE I Sensitivity of average ROMK2 K ϩ currents to glibenclamide when expressed alone and when co-expressed with CFTR, SUR, and CFTR constructs p values were obtained from ANOVA, and differences between groups were isolated using both the Bonferroni test and the Student-Newman-Kuels test. In addition to all pairwise comparisons, we analyzed the data via ANOVA comparing all coexpression groups against oocyte expressing ROMK2 alone (using it as the reference or control group).  FIG. 5. Effect of glibenclamide on Ba 2؉ -sensitive currents obtained from Xenopus oocytes using two-microelectrode voltage clamp techniques. Summary of data obtained for ROMK2, ROMK2/ RT2N2CFTR, ROMK2/CFTR-WT, and ROMK2/CFTR-D835X as indicated on the x axis. Represented on the y axis is the percentage of total barium-sensitive K ϩ current inhibited by glibenclamide. Experiments were all performed under basal conditions. It demonstrates that the second half of the CFTR molecule (RT2N2CFTR) does not confer properties of sulfonylurea sensitivity on ROMK2 and essentially behaves very like ROMK2 alone. In contrast, the first half of the CFTR molecule (CFTR-D835X), when co-expressed with ROMK2, behaves in a similar pattern to ROMK2/CFTR-WT and assumes sensitivity to sulfonylureas. for this interaction (7). Investigators postulate that there is direct protein-protein interaction involving NBF1 and ORCC, but no direct evidence has been provided. Evidence of a direct interaction between CFTR and other channel proteins is provided by the analysis of CFTR domains and recombinant ENaC subunits in yeast two hybrid assays. These assays demonstrate that NBF1 and the R domain of CFTR interact directly with the C-terminal tail of ␣-recombinant ENaC (8).
CFTR may be coupled directly with ROMK2 in the cell membrane, providing the missing domain to the cloned K ϩ channel and thus restoring the sulfonylurea sensitivity seen in native tissue (25). Such a coupling process would be similar to the subunit interactions proposed for SUR and K ir 6.2 where SUR couples with K ir 6.2 and forms the native ␤ cell K ATP channel (26).
Recently, it has been suggested that additional proteins may be involved in the interactions of CFTR with other ion channels such as cytoskeletal proteins or PDZ-containing proteins (24). If these proteins are involved, it would suggest that the interaction is not because of a direct coupling of CFTR with other ion channel pores but secondary to a complicated cascade of events.
The uncoupling of this interaction by phosphorylation may result from a conformational change in the entire ROMK2/ CFTR heteromultimeric structure such that the glibenclamide binding site is altered. Previous studies have shown that when phosphorylated the R domain of CFTR interacts with NBF1 and alters CFTR channel properties (27). Such a process may also alter the ability of CFTR to act as a channel regulator.
Lastly, the human multidrug resistance P-glycoprotein, a multifunctional ABC transporter, is postulated to have two functionally distinct states with respect to its ability to act as a channel regulator and its transport activities (28). It has been shown that active transport of chloride is not necessary for the ROMK2-CFTR interaction to occur, as this interaction occurs without CFTR activation (6). It may be that CFTR is also demonstrating bifunctional properties: a phosphorylated state to act primarily as a Cl Ϫ channel and a basal state to act purely as a channel regulator. This is supported by the ability of CFTR to regulate ROMK2, ORCC, and ENaC independent of its Cl Ϫ channel activity. When assessing cystic fibrosis phenotype/genotype, it has been suggested that the severity of pulmonary disease in cystic fibrosis may be primarily associated with the regulatory properties of CFTR (29); thus understanding the regulatory properties of CFTR is crucial to potential future interventions for cystic fibrosis.