An amino acid triplet in the NH2 terminus of rat ROMK1 determines interaction with SUR2B.

ATP-regulated (K(ATP)) channels are formed by an inward rectifier pore-forming subunit (Kir) and a sulfonylurea (glibenclamide)-binding protein, a member of the ATP binding cassette family (sulfonylurea receptor (SUR) or cystic fibrosis transmembrane conductance regulator). The latter is required to confer glibenclamide sensitivity to K(ATP) channels. In the mammalian kidney ROMK1-3 are components of K(ATP) channels that mediate K(+) secretion into urine. ROMK1 and ROMK3 splice variants share the core polypeptide of ROMK2 but also have distinct NH(2)-terminal extensions of 19 and 26 amino acids, respectively. The SUR2B is also expressed in rat kidney tubules and may combine with Kir.1 to form renal K(ATP) channels. Our previous studies showed that co-expression of ROMK2, but not ROMK1 or ROMK3, with rat SUR2B in oocytes generated glibenclamide-sensitive K(+) currents. These data suggest that the NH(2)-terminal extensions in both ROMK1 and ROMK3 block ROMK-SUR2B interaction. Seven amino acids in the NH(2)-terminal extensions of ROMK1 and ROMK3 are identical (amino acids 13-19 in ROMK1 and 20-26 in ROMK3) and may determine ROMK-SUR2B interaction. We constructed a series of hemagglutinin-tagged ROMK1 NH(2)-terminal deletion and substitution mutants and examined glibenclamide-sensitive K(+) currents in oocytes when co-expressed with SUR2B. These studies identified an amino acid triplet "IRA" within the conserved segment in the NH(2) terminus of ROMK1 and ROMK3 that blocks the ability of SUR2B to confer glibenclamide sensitivity to the expressed K(+) currents. The position of this triplet in the ROMK1 NH(2)-terminal extension is also important for the ROMK-SUR2B interactions. In vitro co-translation and immunoprecipitation studies with hemagglutinin-tagged ROMK mutants and SUR2B indicted that direct interaction between these two proteins is required for glibenclamide sensitivity of induced K(+) currents in oocytes. These results suggest that the IRA triplet in the NH(2)-terminal extensions of both ROMK1 and ROMK3 plays a key role in subunit assembly of the renal secretary K(ATP) channel.

ATP-regulated inwardly rectifying K ϩ (K ATP ) 1 channels are widely expressed in excitable cells (1)(2)(3) and kidney (4,5) where they couple cell metabolism to electrical activity or K ϩ secretion. These channels are formed by octameric complexes of two types of subunits: a pore-forming subunit, the inwardly rectifying K ϩ channel (Kir6.1, Kir6.2, or Kir1.1), and a regulatory subunit, the sulfonylurea receptor (SUR) or the cystic fibrosis transmembrane conductance regulator (CFTR), members of the ATP binding cassette transporter protein family (3, 6 -9). Sensitivity of K ATP channel K ϩ currents to sulfonylurea drugs (e.g. glibenclamide) requires the associated SUR or CFTR subunit (9 -11). Three isoforms of SUR have been cloned and are important for the regulation of physiological and pharmacological functions of K ATP channels (2,10,12,13). SUR1 associates with Kir6.2 to form the neuronal/pancreatic ␤-cell-type K ATP channel; SUR2A and SUR2B, splice variants of a single gene, form either the cardiac-type (SUR2A/Kir6.2) or the vascular smooth muscle-type (SUR2B/Kir6.1 or Kir6.2) K ATP channels.
Renal K ATP channels have been identified in the apical membranes of the thick ascending limb of the loop of Henle (TAL) and principal cells of the cortical collecting duct where they mediate K ϩ secretion (4,5). These K ATP channels appear to be formed by association of ROMK (Kir1.1) (5,14) and CFTR (9,11) and/or SUR2B (15). Three rat ROMK splice variants (ROMK1-3) are expressed in rat kidney and are identical except for distinct NH 2 -terminal extensions of 19 and 26 amino acids in ROMK1 and ROMK3, respectively (16 -18). We recently cloned SUR2B from rat kidney and showed that it was expressed in TAL and cortical collecting duct, similar to the localization of ROMK2 (15). Co-expression of ROMK2 with SUR2B in Xenopus laevis oocytes generated glibenclamidesensitive K ϩ currents, indicating that SUR2B could be involved in forming and regulating renal K ATP channels (15). Surprisingly neither ROMK1 nor ROMK3 formed glibenclamide-sensitive K ϩ channels when co-expressed with SUR2B (15), suggesting that the unique NH 2 -terminal extensions of ROMK1 and ROMK3 inhibited interactions with SUR2B. In the present study we evaluated the role of the unique NH 2 -terminal amino acids of ROMK1 in determining interactions with SUR2B using electrophysiology and co-immunoprecipitation methods.

MATERIALS AND METHODS
Construction of HA-tagged Mutant ROMK1 and SUR2B-The hemagglutinin (HA) tag was introduced into the 3Ј-end of the ROMK cDNA just before the stop codon by polymerase chain reaction using ROMK1/pSPORT1 and ROMK2/pSPORT1 as templates with primers * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  The polymerase chain reaction products were digested with MscI and NheI and subcloned into MscI-and NheI-digested wild-type ROMK to generate the HA-tagged ROMK, ROMK-HA (19). A series of ROMK1 NH 2 -terminal truncation mutants and other mutants were designed and prepared by polymerase chain reaction methods. A schematic representation of ROMK1, ROMK1 truncations, and ROMK2 are shown in Figs. 1 and 2. Each construct sequence begins with the same "Kozak" sequence (20) and the translation initiation codon (GCCACCATG) found in ROMK1. R1-N⌬5-HA indicates that the first 5 amino acids in the NH 2 -terminal extension of ROMK1-HA were removed. Constructs R1-N⌬10-HA, R1-N⌬12-HA, R1-N⌬15-HA, R1-N⌬16-HA, and R1-N⌬17-HA represent ROMK1-HA mutants where the first 10, 12, 15, 16, or 17 amino acids in the NH 2 terminus were deleted. Other mutant constructs were R1-N⌬(13-19)-HA (amino acids 13-19 deleted from the NH 2 terminus of ROMK1-HA), R1-NS-HA (amino acids 13-19 moved to the start of the NH 2 terminus of ROMK1-HA), R1-N(13-15)V-HA (amino acids 13-15 replaced with valine residues), and R1-N(16 -18)V-HA (amino acids 16 -18 replaced with valine residues). R1-N13V-HA, R1-N14V-HA, and R1-N15V-HA are ROMK1-HA mutants where amino acids 13-15 were individually replaced by valine. All ROMK constructs were expressed in pSPORT1. SUR2B cDNA, cloned from rat kidneys, was subcloned into the pGEMHE vector. The sequences of all constructs were conformed using the cycle sequencing method (PerkinElmer Life Sciences).
Two-electrode Voltage Clamp-Ba 2ϩ -sensitive K ϩ currents in oocytes injected with cRNA from each of the rat ROMK1-HA, ROMK2-HA, and HA-tagged ROMK1 mutants with rat SUR2B (5 ng of ROMK1-HA, ROMK2-HA, or HA-tagged ROMK1 mutants, 50 ng of SUR2B; molar ratio ϭ 1 ROMK1-HA, ROMK2-HA, or HA-tagged ROMK1 mutant:3 SUR2B) were examined by two-electrode voltage-clamping as described previously (16,17). Recordings were performed 96 h after cRNA coinjection at 22 Ϯ 2°C by two-electrode voltage clamp at a holding potential of Ϫ100 mV (Axoclamp 2A, Axon Instruments, Inc.). Currents were measured from Ϫ160 to ϩ40 mV for 50 ms in 20-mV steps. The composition of the bath solution was as follows: 96 mM NaCl, 1 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM HEPES (pH 7.4) with 5 mM Ba 2ϩ or 0.2 mM glibenclamide. The average resting potential was Ϫ100 mV (Ϫ70 to Ϫ104 mV) at an external K ϩ concentration of 1 mM. As indicated in our previous study (9), glibenclamide sensitivity is best observed with 1 mM external K ϩ . Glibenclamide was dissolved in Me 2 SO (Sigma) and diluted from a 1000ϫ stock solution. An equal amount of Me 2 SO was added to the control bath solution.
In Vitro Translation of ROMK1, ROMK1 Mutants, or ROMK2 with SUR2B-ROMK1-HA, ROMK2-HA, or HA-tagged ROMK1 mutants with or without SUR2B wild-type cDNAs were translated in vitro using the TNT ® -coupled reticulocyte lysate system and [ 35 S]methionine in the presence of canine pancreatic microsomal membranes according to the instructions of the manufacturer (Promega). Reaction mixtures includ-ing cDNA of each ROMK-HA or HA-tagged ROMK1 mutants with or without SUR2B wild-type were incubated at 30°C for 90 min after the addition of [ 35 S] methionine. Protein products were resolved by 8% SDS-polyacrylamide gel electrophoresis, and the [ 35 S] methioninelabeled proteins were visualized by autoradiography (15).
Immunoprecipitation-Immunoprecipitaton was performed using an anti-HA monoclonal antibody (clone 12CA5, Roche Molecular Biochemicals) as described previously (15). The in vitro-translated reaction mixture was washed twice with ice-cold phosphate-buffered saline buffer and lysed in a buffer containing 20 mM Tris/HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40 (v/v) at 4°C for 20 min followed by centrifugation at 10,000 ϫ g for 15 min to obtain supernatants. Proteins were immunoprecipitated by incubation of supernatants with anti-HA monoclonal antibody for 4 h followed by overnight incubation with protein A-Sepharose at 4°C. The proteins were then recovered by incubation of protein A with 2ϫ Laemmli SDS sample buffer at 55°C for 15 min and separated by an 8% SDS-polyacrylamide gel. The [ 35 S]methionine-labeled proteins were visualized by autoradiography.
FIG. 3. Glibenclamide sensitivity of ROMK1, ROMK1 truncation mutants, and ROMK2 co-expressed with SUR2B. ROMK1 (or ROMK2) wild type or ROMK1 mutant construct cRNAs (5 ng) were co-injected with SUR2B cRNA (50 ng). Whole cell currents were monitored in X. laevis oocytes by two-electrode voltage clamp with 1 mM bath K ϩ , and all currents could be abolished by 5 mM Ba 2ϩ . The oocytes were voltage-clamped at Ϫ100 mV and pulsed in steps of 20 mV for 50 ms over the range of Ϫ160 to ϩ40 mV.
In Vitro Co-translation and Co-immunoprecipitation of ROMK1 Truncation Mutants and SUR2B-We previously showed that the functional and biochemical interactions between ROMK2 or ROMK1 and SUR2B are directly correlated (15). To assess whether the functional interaction of the HAtagged ROMK1 truncation mutants with SUR2B also indicates direct interaction between the channel subunits, the proteins were translated in vitro and immunoprecipitated with anti-HA antibody. In vitro translation of ROMK1-HA, ROMK2-HA, and the HA-tagged ROMK1 truncation mutants in the presence of canine pancreatic microsomal membranes showed the expected molecular mass bands between 42 and 45 kDa. Fig. 4 shows that co-translation of ROMK1-HA, ROMK2-HA, or the HAtagged ROMK1 truncation mutants with SUR2B produced two distinct bands in each lane: the 174-kDa band representing SUR2B and a band between 43 and 45 kDa representing the HA-tagged ROMK protein. To determine whether SUR2B coprecipitated with the ROMK proteins, we used anti-HA antibody to immunoprecipitate ROMK1-HA, ROMK2-HA, or the HA-tagged ROMK1 truncation mutants and then resolved proteins by SDS-polyacrylamide gel electrophoresis. The SUR2B protein did not co-immunoprecipitate with ROMK1-HA or the ROMK1-HA truncation mutants R1-N⌬5-HA, R1-N⌬10-HA, and R1-N⌬12-HA (Fig. 5). In contrast, immunoprecipitation of R1-N⌬16-HA, R1-N⌬17-HA, or ROMK2-HA brought down the SUR2B protein (Fig. 5). When R1-N⌬15-HA was co-translated with SUR2B and the proteins were immunoprecipitated with anti-HA antibody, a mixed result was obtained: sometimes the anti-HA precipitated a complex of R1-N⌬15-HA and SUR2B proteins, but other times it just precipitated the R1-N⌬15-HA protein (data not shown). This is consistent with the variable glibenclamide sensitivity seen with this mutant construct. Thus, our biochemical data are consistent with the electrophysiological results: glibenclamide sensitivity occurs only when ROMK and SUR2B directly associate, and amino acids in the extended NH 2 terminus of ROMK1 inhibit its association with SUR2B. Specifically, these results indicate that amino acids 13-15 in the NH 2 terminus of ROMK1 inhibit the interaction between ROMK1 and SUR2B.
In Vitro Co-translation and Co-immunoprecipitation of Individual ROMK1 Mutants with SUR2B- Fig. 7 shows the results of the in vitro co-translation of R1-N⌬(13-19)-HA or R1-NS-HA and SUR2B performed in the presence of canine pancreatic microsomal membranes. Anti-HA antibody co-immunoprecipitated each of these ROMK1 mutants together with SUR2B. When R1-N(13-15)V-HA or R1-N(16 -18)V-HA and SUR2B were co-translated, SUR2B was co-immunoprecipitated with R1-N(13-15)V-HA but not with R1-N(16 -18)V-HA (Fig. 8). These results are consistent with the glibenclamide sensitivity of the K ϩ currents observed when these mutants were coexpressed with SUR2B. Moreover, they also confirm that the amino acid triplet (amino acids 13-15, IRA) in ROMK1 plays a key role determining the assembly of ROMK1 with SUR2B, and thereby conferring glibenclamide sensitivity to the K ATP channel. DISCUSSION The present study defines the molecular site on ROMK1 that inhibits its interaction with SUR2B to form a sulfonylureasensitive K ϩ channel. Our previous study demonstrated that ROMK2, but not ROMK1 or ROMK3, formed a glibenclamidesensitive K ϩ channel when co-expressed with SUR2B (15). ROMK1 and ROMK3 are alternatively spliced forms of the renal ATP-regulated K ϩ channel that contain NH 2 -terminal extensions of 19 and 26 amino acid residues, respectively. Compared with ROMK2, ROMK1 and ROMK3 share a 7-amino acid segment just preceding the ROMK2 initiator methionine (Fig.  1) and we had proposed that this segment could be involved in blocking the interaction of ROMK and SUR2B (15). In the present study we confirmed that ROMK2, but not ROMK1, formed a glibenclamide-sensitive K ϩ channel when co-expressed with SUR2B, and we have now identified three amino acids (in ROMK1 amino acids 15-17, IRA) that inhibit the interaction with SUR2B and thus render the ROMK1 channel insensitive to glibenclamide. This amino acid triplet is contained within the 7-amino acid region, 13-19, which is identical to amino acids 20 -26 in ROMK3 and could also account for the lack of ROMK3 interaction with SUR2B.
Since glibenclamide sensitivity was only observed when ROMK co-immunoprecipitated with SUR2B, sulfonylurea sensitivity requires direct interaction between these two proteins. While we do not know the stoichiometry of ROMK-SUR2B to form glibenclamide-sensitive K ϩ channels in oocytes, previous studies of SUR1 and the ATP-sensitive K ϩ channel Kir6.2 demonstrated a one-to-one stoichiometry in a 4:4 hetero-octamer (3). The NH 2 terminus of Kir6.2 has been suggested to provide a site for coupling to the sulfonylurea receptor (21). Future studies will be required to investigate what regions and amino acid residues in ROMK2 play a similar role in the interaction with SUR2B. If a similar region in ROMK2 is required for interacting with SUR2B, then it is possible that the inhibitory "IRA" triplet in ROMK1 interferes with this NH 2 -terminal SUR2B binding site by either binding to this The upper 174-kDa band represents SUR2B protein, and the ϳ45-kDa lower bands are HA-tagged ROMK. Above each lane is a representation of the specific construct. R2, see Fig. 2; aa, amino acids. region or altering its structure. Interestingly the relative position within the NH 2 terminus of this amino acid triplet appears to be critical since transposing it to the beginning of the NH 2 terminus allows ROMK1 interaction with SUR2B.
It is generally accepted that ROMK forms the small conductance ATP-regulated and glibenclamide-sensitive K ϩ channel in distal nephron segments of the mammalian kidney mediating K ϩ secretion (5,14). When ROMK is expressed alone in X. laevis oocytes it is not sensitive to glibenclamide (9,11,15), indicating that it must be co-associated with a sulfonylurea receptor protein in native tissue. In this regard, ROMK2 has been shown to form glibenclamide-sensitive K ϩ currents when co-expressed with the CFTR (9,11) and with the sulfonylurea receptor proteins SUR1 (22) and SUR2B (Ref. 15 and Fig. 3, B and C). CFTR, like the sulfonylurea receptor, also is a member of the ATP binding cassette transporter gene family and has been shown to impart glibenclamide sensitivity to other channels (23). SUR1 is not expressed in kidney epithelia, but both SUR2B (15,24) and CFTR transcripts and protein (25) are found in several distal nephron segments that also express ROMK isoforms (17). Thus, both CFTR and SUR2B may contribute to forming ATP-regulated, small conductance K ϩ secretory channels in the kidney.
While ROMK2-SUR2B and ROMK2-CFTR can form glibenclamide-sensitive K ϩ currents, CFTR, but not SUR2B, can also impart sulfonylurea sensitivity to K ϩ currents when co-expressed with ROMK1. ROMK2 is expressed in all distal nephron segments from the TAL to the cortical collecting duct, while ROMK1 is found only in the collecting duct and ROMK3 in TAL cells. The medullary TAL expresses ROMK2, ROMK3, and CFTR, while the cortical TAL expresses ROMK2, ROMK3, CFTR, and SUR2B. The cortical collecting duct expresses ROMK1, ROMK2, CFTR, and SUR2B. Thus TAL and collecting duct cells may express different combinations of ROMK-and sulfonylurea-interacting proteins to give rise to the ATP-regulated, glibenclamide-sensitive K ϩ channels found in apical membranes of these nephron segments.
In conclusion, our results demonstrate that the specific location of amino acid triplet IRA (residues 13-15) in the NH 2terminal extension of ROMK1 (and likely ROMK3) plays a key role in the inhibition of subunit assembly of the renal secretary K ATP channel ROMK with SUR2B and can account for the lack of glibenclamide sensitivity when SUR2B is co-expressed with ROMK1 or ROMK3.