A. Tinker, Room F2, 4th Floor, Centre for Clinical Pharmacology, Department of Medicine, University College London, The Rayne Institute, 5 University Street, London WC1E 6JJ, UK, Fax: + 44 20 76912838, Tel.: + 44 20 76796192, E-mail: email@example.com
ATP-sensitive K+ channels are an octameric assembly of two proteins, a sulfonylurea receptor (SUR1) and an ion conducting subunit (Kir 6.0). We have examined the role of the C-terminus of SUR1 by expressing a series of truncation mutants together with Kir6.2 stably in HEK293 cells. Biochemical analyses using coimmunoprecipitation indicate that SUR1 deletion mutants and Kir6.2 assemble and that a SUR1 deletion mutant binds glibenclamide with high affinity. Electrophysiological recordings indicate that ATP sensitivity is normal but the response of the mutant channel complexes to tolbutamide, MgADP and diazoxide is disturbed. Quantitative immunofluorescence and cell surface biotinylation supports the idea that there is little disturbance in the efficiency of trafficking. Our data show that deletions of the C-terminal most cytoplasmic domain of SUR1, can result in functional channels at the plasma membrane in mammalian cells that have an abnormal response to physiological and pharmacological agents.
ATP-sensitive potassium channels (KATP) are present in the plasma membrane of a number of tissues and are also present in endomembranes such as mitochondria. They have been proposed to be involved in a number of physiological and pathophysiological processes and form a link between cellular metabolism and membrane excitability. For example, in the pancreas, KATP regulates insulin release and in vascular smooth muscle, it is regulated by vasodilators and influences blood flow in certain vascular beds. KATP is an octameric protein complex composed of two subunit types namely a pore forming subunit (Kir6.1, Kir6.2), a member of the inwardly rectifying family of K+ channel, and the sulfonylurea receptor subunit, a member of the ATP binding cassette family of proteins (SUR1, SUR2A, SUR2B). The assembly of a particular pore forming subunit with a particular SUR generates currents with a characteristic single-channel conductance, nucleotide regulation and pharmacology [1–4]. Kir subunits have a cytoplasmic N and C terminus with two transmembrane domains and a pore forming H5 loop [5,6]. SUR has multiple transmembrane domains with two large intracytoplasmic loops, the first and second nucleotide binding domains (NBD1 and NBD2), which contain consensus sequences for the hydrolysis of nucleotides (Walker A and B motifs) [7,8].
The trafficking of the KATP channel complex has been the subject of some investigation. It was initially observed that coexpression of the two proteins was necessary to generate significant plasmalemmal currents . A series of studies have supported the idea that the Kir6.2 and SUR1 subunits have small peptide motifs (RKR) that either prevent the export of the protein from the ER and/or retrieve it from the Golgi . The simultaneous masking of these two signals by interaction of the two proteins allows the channel complex to proceed through the biosynthetic pathway to the membrane. It has also been suggested that the most distal part of the C-terminus of SUR1 contains an anterograde signal that allows export from the ER . Potentially, this has clinical ramifications as a number of mutations in persistent hyperinsulinaemic hypoglycaemia of infancy  cluster in this domain of the protein. Persistent hyperinsulinaemic hypoglycaemia of infancy is an hereditary disease characterized by inappropriately high levels of insulin release and hypoglycaemia in children at birth. It is hypothesized that deletion of the most distal part of the C-terminus (only seven amino acids) leads to the removal of a forward trafficking signal and retention in the ER. A phenylalanine (1574) and a leucine (1566) were established as being particularly important. However a splice variant of SUR1 has been described and cloned from a hypothalamic cDNA library that removes exon 33 and results in a frameshift and premature termination of the C-terminus preceding the Walker A and B motifs . Two other deletions were constructed at the beginning and end of the C-terminus in exon 33. All these were functional upon expression and the truncation of 253 amino acids did not affect the magnitude of macroscopic currents. In a related vein, a recent report has shown a SUR1-MRP1 C-terminal chimaera is able to traffic to the plasma membrane when coexpressed with Kir6.2 . There are several potential explanations for these different results. Firstly, there are differences in the expression system used (Cos cells in  vs. Xenopus laevis oocytes in  and ) and secondly the details of the deletions vary. To try to resolve these differences we have constructed a series of deletions in the second nucleotide binding domain, expressed and studied their biochemical and functional behaviour in a human kidney cell line (HEK293 cells).
Materials and methods
Standard subcloning techniques were used throughout. A SUR1 mutant with a myc epitope was used as previously described . In our initial studies we constructed a 146 amino acid deletion of SUR1 with the myc epitope. The pcDNA3 vector was digested with NotI/ApaI. Sense and antisense oligonucleotides corresponding to an artificial gene fragment encoding the myc epitope and stop codon together with compatible overhangs were annealed and ligated into pcDNA3. A NotI fragment of SUR1 (from SUR1 in pcDNA3) was then subcloned into this construct. Subsequently a range of SUR1 deletions, tagged with the 10 amino acids constituting the myc epitope, were constructed from SUR1 in pBluescript (SK–). A two stage PCR strategy was used to generate a deletion cassette with a 5′-SfiI site in the clone and the myc epitope followed by a stop codon and an AvrII site. The native fragment was replaced by this cassette and resulting product subcloned into the XhoI/XbaI sites of pcDNA3 using a SalI/SpeI digest. Kir6.2 and Kir6.2mycHis6 was expressed in pcDNA3.1/Zeo (Invitrogen). Kir6.2mycHis6 was generated from a previous study . The sequence of all mutants was confirmed by DNA sequencing using the dRhodamine Terminator cycle sequencing kit (Applied Biosciences) and an automatic sequencer (ABI 377, Perkin-Elmer).
Cell culture and transfection
HEK293 cells were cultured, transfected and stable cell lines expressing Kir6.2, Kir6.2mycHis6 and SUR1-myc and deletion mutants generated as previously described . A number of new monoclonal lines have been generated in this study and they are detailed as appropriate in the text. For selection with G418 we used 727 µg·mL−1, for Zeocin 364 µg·mL−1 and in combination 727 µg·mL−1 G418 and 364 µg·mL−1 Zeocin.
Antiserum production in rabbits
A peptide corresponding to amino acid residues 942–955 of hamster SUR1 (ETVMERKASEPSQGC, final cysteine added for coupling purposes) was synthesized and linked to keyhole limpet haemocyanin before injection into rabbits using standard protocols (Regal Group Ltd, Great Bookham, Surrey, UK). Bleeds were assayed for activity using an antibody capture assay  and bleeds showing reactivity were affinity purified. Kir6.2 antisera to the C-terminus of the channel and myc hybridoma cells were used as previously described [15,17].
Gel electrophoresis, radioligand binding, immunoprecipitation and immunofluorescence
SDS/PAGE, Western blotting, radioligand binding, immunoprecipitation and immunofluorescence staining were carried out as previously described [15,17] with some modifications for colocalization experiments. In colocalization experiments slides were incubated with the first primary antibody [a 1 : 500 dilution of anti-(Kir6.2 C-terminus) Ig] and the appropriate secondary antibody (a 1 : 300 dilution of a rhodamine-linked secondary goat anti-rabbit Ig) for 1 h each as previously described . After washing, the second primary antibody was applied overnight (a 1 : 500 dilution of anti-myc Ig) and the appropriate secondary antibody (a 1 : 300 dilution of a fluorescein-linked secondary goat anti-mouse Ig) was then applied the following day for 1 h. The antibody reactive to the Kir6.2 C-terminus was raised to the peptide sequence DALTLASSGPLRKRSC and has been characterized previously . The anti-myc Ig used was purified on Protein A Sepharose CL-4B (Amersham-Pharmacia biotech) from mouse 9E10 hybridoma cell line culture supernatant under high salt conditions using a standard method (p311; ). All fluorophore conjugated secondary antibodies were purchased from Molecular Probes Inc. Gel densitometry was performed using scion image as detailed in the help section in the program.
For the imaging work, slides were viewed and analysed using a computer based image analysis system (openlab 3.1, Improvision) coupled to a Zeiss Axiovert 100M microscope set up for epifluorescence equipped with an appropriate filter set that allowed imaging of both fluorescein and rhodamine fluorophores (XF66-1 multiband filter, Omega optical). Images used for quantitative analysis and colocalization experiments represented a focal plane taken through the middle of the cell. Scattered and out-of-focus fluorescent signal was removed from the image by deconvolution. The image was deconvolved from a z-stack of 31 images with 0.2 µm spacing using the openlab software (volume deconvolution module). The quantitative assay was performed on the deconvolved images using the openlab advanced measurements module. The theory behind the quantitative assay is given in Fig. 6A. In colocalization experiments, deconvolved images of the cell at each excitatory wavelength were obtained and merged using the openlab software.
Procedure for surface biotinylation
Stable cell lines were cultured in 100 mm2 tissue culture dishes as previously described  until 75–95% confluent. Each dish was washed three times with ice-cold phosphate buffered saline (NaCl/Pi– 10 mm phosphate buffer, 2.7 mm KCl, 137 mm NaCl, pH 7.4, prepared from tablets supplied from Sigma, Poole, UK) before incubation with 3 mL 0.5 mg·mL−1 EZ-Link Sulfo-NHS-LC-Biotin (Pierce, Rockford, Illinois, USA) in NaCl/Pi at 4 °C for 30 min. Dishes were washed twice with ice-cold NaCl/Pi before incubation with NaCl/Pi + 100 mm glycine at 4 °C for 20 min to quench any remaining unreacted biotin reagent. Cells were then lysed by scraping into 250 µL 1% (w/v) SDS in Tris buffered saline (Tris/NaCl – 50 mm TrisHCl, 150 mm NaCl, 5 mm KCl, pH 7.4) containing 1 mm phenylmethylsulfonyl fluoride and a protease inhibitor cocktail (Roche Complete EDTA-free, Roche Diagnostics Limited, Lewes, Sussex, UK). The lysates were subsequently incubated at 65 °C for 10 min before dilution with 1 mL 1% (v/v) Triton in Tris/NaCl containing 1 mm phenylmethanesulfonyl fluoride and the protease inhibitor cocktail. Lysates were then incubated on ice for 1 h before being briefly sonicated (2 s using an MSE Soniprep 150 probe sonicator at half power). Sonicated lysates were centrifuged at 20 000 g for 30 min to pellet any insoluble material and biotinylated proteins were isolated from the supernatant as described below. A 50-µL sample was removed at this point and used for subsequent SDS/PAGE and Western blotting analysis.
Isolation of surface biotinylated proteins
The supernatant obtained from the surface biotinylation procedure was incubated with 150 µL of a pre-equilibrated 1 : 1 slurry of Ultra-link Immobilized NeutrAvidin biotin binding protein (Pierce) overnight at 4 °C with gentle rotation. The slurry was pre-equilibrated with binding buffer [1% (v/v) Triton, 0.20% (w/v) SDS in Tris/NaCl]. After the incubation period, the binding resin was pelleted by centrifugation (20 000 g for 2 min at 4 °C) followed by five washes with 1 mL binding buffer. Bound protein was recovered by incubation with 80 µL 6 × Laemmli gel loading buffer (350 mm TrisHCl pH 6.8, 10.28 (w/v) SDS, 36% (v/v) glycerol, 0.012% (w/v) bromophenol blue, 200 mm dithiothreitol) at 100 °C for 3 min. Eluted proteins were subsequently analysed by SDS/PAGE followed by Western blotting.
Whole-cell and inside-out patch clamp recordings were performed using an Axopatch 200B amplifier (Axon Instruments) and were digitized using a Digidata 1200 interface before capture to a computer hard disk. Whole-cell current signals were filtered at 1 kHz and sampled at 2 kHz, and analysed using pclamp6 software (Axon Instruments). Unitary single-channel currents were filtered at 2 kHz and sampled at 5 kHz, and analysed to determine either mean NPo (Number of channels × Open probability) during a 30 second sweep (using pClamp6), or mean current (using Satori, Intracel Ltd). Patch pipettes were pulled using a PP-830 pipette puller, and fire-polished using a MF-830 microforge (both Narishige). Pipettes had resistances of 1.5–3 MΩ for whole-cell recording and 6–9 MΩ for single-channel recordings. The capacitance of pipettes was reduced by coating pipettes with a parafilm/mineral oil suspension and compensated for by using the amplifier. Series resistance during whole cell recording was compensated to at least 70% using amplifier circuitry. The pipette/bath solution contained in mm; 107 KCl, 1.2 MgCl2, 1 CaCl2, 10 EGTA, 5 Hepes (with 33 mm KOH to pH 7.2) and the bath/pipette solution; 140 KCl, 2.6 CaCl2, 1.2 MgCl2, 5 Hepes (pH 7.4) for whole cell/single-channel work. Whole-cell pipette solutions were supplemented with ATP and ADP, and pH was re-adjusted to 7.2 (see figure legends for nucleotide concentrations used in each stable line).
Radioligand binding experiments were fitted by the binding isotherm y = Bmax·x/(x + Kd) where y is the bound specific radioligand (pmol·mg protein−1) and x is the radioligand concentration. For dose–response curves with inhibition of current by tolbutamide the data from individual experiments were expressed and fitted to: % inhibition of maximal current = a + (b − a)/(1 + (x/Ki)h) where b was fixed at zero and a represents the limiting inhibition (expressed as percentage inhibition of maximal current), h the hill coefficient and Ki the EC50 for inhibition. Mean Ki was calculated by fitting individual experiments with the curves and calculating mean parameters as indicated. Statistical analysis was carried out using one-way anova with an appropriate posthoc test or Students t-test as appropriate (origin v6.0 and prism v3.0). Statistical significance is as indicated in the legend and text. Data are presented as mean ± SEM.
To facilitate biochemical studies, we generated a polyclonal rabbit antisera raised to a short peptide sequence in the first nucleotide binding domain of SUR1 (see Materials and methods). Figure 1A and B show the characterization of the specificity of the antisera for immunoblotting and immunofluorescence, respectively. It is apparent that the antisera recognizes bands of the correct molecular mass in a SUR1 + Kir6.2 stable line (a band at ≈ 150 kDa and another at ≈ 170 kDa) but not in stable lines expressing SUR2A + Kir6.2, SUR2B + Kir6.1 and wildtype nontransfected HEK293 cells (Fig. 1A). The antisera also recognizes a number of other proteins of lower molecular mass endogenous to HEK293 cells however, this does not influence the nature of our conclusions below. When assayed using immunofluorescence, the antisera reacted with a stable line expressing SUR1 + Kir6.2 but not one expressing SUR2B + Kir6.1 (Fig. 1B). The signal was competed with by the immunogenic peptide (Fig. 1B) and incubation with the secondary alone led to no signal (not shown).
Generation of SUR1 deletions and stable cell lines
Using standard molecular cloning methods (see Materials and methods) we generated a series of SUR1 deletions tagged with the 10 amino acid myc epitope (Fig. 2A). In our initial studies we first examined a 146 amino acid deletion (SUR1del146myc) and subsequently engineered a series of these. We generated a polyclonal stable cell line in HEK293 cells stably expressing SUR1del146myc by selection with G418 and stable monoclonal cell lines expressing Kir6.2 + SUR1myc (line A), Kir6.2 + SUR1del101myc (line B), Kir6.2 + SUR1del145myc (line C1) and Kir6.2mycHis6 + SUR1del146myc (line C2), Kir6.2 + SUR1del196myc (line D) and Kir6.2 + SUR1del249myc (line E) with G418 and Zeocin as previously described [15,18]. Lines were screened using biochemical and subsequently electrophysiological methods. Line C1 was labile with multiple passages often losing current. Line C2 was more stable and could be passaged for long periods without loss of current. As a result most electrophysiological studies were performed on this line.
Kir6.2 interacts with SUR1 and the SUR1 deletions
We used a coimmunoprecipitation strategy to examine the interaction of Kir6.2 with SUR1-myc and SUR1-myc deletions. Immunoprecipitation of SUR1-myc and deletion constructs was performed using a mouse monoclonal antibody to myc to avoid problems with discrimination between the rabbit immunoglobulin heavy chain and Kir6.2 that might occur if the precipitating antibody were rabbit in origin. Immunoprecipitation of SUR1 was confirmed by probing with the rabbit polyclonal antisera to SUR1. Figure 2B shows the coimmunoprecipitation of Kir6.2 with SUR1-myc and each of the deletions. The upper blots show the level of protein expression in the selected lines prior to immunoprecipitation. Kir6.2 is detected using a rabbit polyclonal antisera raised to the C-terminus of the protein (see Materials and methods and ). Controls for the immunoprecipitation protocol have previously been shown . We further showed the close association and localization of Kir6.2 and SUR1-myc by costaining cells with the myc mouse monoclonal antibody and the rabbit anti-(Kir6.2 C-terminus) Ig. In addition, we also examined the colocalization of Kir6.2 with the most profound deletion SUR1del249myc. In Fig. 2C it is apparent that the two signals colocalize in both line A and line E. Thus the ability of Kir6.2 to assemble with SUR1 is not grossly affected by deletion of the second nucleotide binding domain.
Functional properties of the deletion mutants coexpressed with Kir6.2
Electrophysiological techniques revealed that all the deletion mutants were able to express significant currents at the plasma membrane of HEK293 cells. In Fig. 3A and B representative traces are shown from whole cell patch clamp recordings from line A (Ai and Aii) as compared to the most profound deletion in line E (Bi and Bii). After rupture of the membrane to gain access to the intracellular contents, whole-cell currents increased to a steady-state value. The magnitude of this was dependent on the pipette ATP concentration. Currents in line A were substantially inhibited by 100 µm tolbutamide, whereas those in line E were only partially inhibited (see below). Both currents were completely inhibited at hyperpolarized potentials by 10 mm Ba2+ (Fig. 3Aii and Bii). All the SUR1 deletions in the monoclonal stable lines gave rise to currents significantly in excess of those from wildtype cells . However the magnitude of these varied with clonal isolates and some lines possessed smaller currents than others. With 1.2 mm ATP in the patch pipette, the current density at −50 mV was: in line A = 586 ± 117 pA/pF (n = 14); in line B = 9.5 ± 2.1 pA/pF (n = 11); in line D = 51.2 ± 14.1 pA/pF (n = 7) and in line E = 492 ± 136 pA/pF (n = 11). With 0.6 mm ATP in the patch pipette, the current density at −50 mV in line C2 was 313 ± 44 (n = 8) pA/pF. This electrophysiological behaviour was reflected in variations in the expression of the relevant components as determined by immunoblotting. Figure 3C shows representative recordings of single-channels in the inside-out configuration. It shows the inhibition of channels in line A and E by ATP. The deletion of the C-terminus of SUR1 does not substantially affect the response to this nucleotide (Fig. 3Ci and ii) and quantitation is shown in Fig. 5B.
However other electrophysiological and pharmacological parameters were disturbed. The interaction of the channel complex with the sulfonylurea class of drug was examined next. We first performed radioligand binding with tritiated glibenclamide to the polyclonal SUR1del146myc cell line. The Kd for drug binding is not significantly changed: for SUR1del146myc Kd = 2.02 ± 0.35 nm (n = 6) (a representative example is shown in Fig. 4A) and for SUR1-myc as previously published Kd = 1.25 ± 0.17 nm (n = 4) (not significant, P = 0.12 unpaired t-test after logarithmic transformation). Next, we examined the electrophysiological effect of tolbutamide using the whole cell configuration of the patch clamp (Fig. 4B.C). Concentrations of tolbutamide (100 µm) known to act selectively by binding to SUR1 and not to have significant pore blocking effects were used. The lines containing the SUR1 deletions had currents that were reduced but not to the same extent compared to control. Figure 4B summarizes the percentage inhibition of current in these lines. Thus progressive deletion of the C-terminus of SUR1 affects the efficacy of drug response. To further characterize this, dose–response curves were constructed for two of these and it was found that the Ki was not significantly affected once the differences in limiting value were taken into account. The Ki for tolbutamide inhibition of the current in line A was 8.1 ± 3.3 µm and h = 1.45 ± 0.1 and a = 87.3 ± 3.9 (n = 10), compared with 18.7 ± 7.5 µm, h = 1.63 ± 0.3 and a = 51.3 ± 4.0 for line C2 (n = 6) and 6.2 ± 2.8 µm, h = 1.43 ± 0.33 and a = 33.0 ± 4.2 for line E (n = 8) (for KiP = 0.09 after log transformation for all comparisons, one way anova with Bonferronni correction). Figure 4C shows the mean pooled data at each point fitted with the best-fit curve (to the means) using nonlinear regression. The resultant parameters are very similar but not identical to those determined after analysis of each individual experiment. The latter however, allows statistical analysis of the relevant data.
The response to bath applied diazoxide was determined in the whole-cell configuration. The application of 300 µm diazoxide induced large whole-cell currents in line A but led to little increase in the other lines (Fig. 4D). Recent studies have shown an interrelationship between diazoxide stimulation and MgADP stimulation [19,20]. Thus we examined bath applied MgADP in inside-out patches on line A and line E. In the former line there was pronounced stimulation whilst in the latter there was a much more modest stimulation (Fig. 5A and B). We also performed analogous experiments with line C2 and observed very little stimulation with MgADP (not shown).
In summary, the deletion of the C-terminus of SUR1 results in functional channels that cannot be stimulated by diazoxide and MgADP. The affinity of sulfonylurea interaction is not affected, however, the net resultant inhibitory effect is decreased as the C-terminus is progressively deleted.
Membrane trafficking studied using immunofluorescence
The presence of significant membrane currents is an indication that trafficking is not significantly impaired. However the possibility of differences in the relative level of expression of the different mutants may confound such a picture. For example, it is possible for there to be significant membrane currents but for a significant fraction of protein to be retained intracellularly. A number of elegant and imaginative methods have been developed to examine these problems [10,21]; however, they suffer from the complication that calibration for relative expression must be obtained independently. We took a different approach. The fraction of immunofluorescent signal at or close to the plasma membrane was calculated in a number of cells. First a mask was defined round the edge of the cell and then a further mask was defined (performed by shrinking the original mask by 0.4 µm in each direction), which excludes the membrane and membrane associated area but includes the intracellular contents. The total fluoresence in each situation is calculated from mean pixel intensity multiplied by the number of pixels. The ‘membrane’ associated fluoresence is calculated by subtracting the two and the ‘membrane’ signal is normalized to the total fluorescence. Potentially this controls internally within each cell for differences in expression. The principle of the assay is shown in Fig. 6A.
The sensitivity of such an assay was tested in detecting the membrane translocation of SUR1 and Kir6.2 when expressed independently and in the situation where they are coexpressed. To facilitate these studies we generated stable lines expressing Kir6.2 with Zeocin selection and SUR1 with G418 (see Materials and methods). The fraction of immunofluorescence associated with the membrane was compared between SUR1 and SUR1 + Kir6.2 stable lines probed with the anti-SUR1 Ig and between Kir6.2 and Kir6.2 + SUR1 stable lines probed with the anti-(Kir6.2 C-terminus) Ig. Representative images and the quantitative data are shown in Fig. 6B–D. In both cases there was a measurable and significant increase in membrane associated immunofluorescence in keeping with the visual appearance of the images (see Discussion). The fraction retained upon single expression of a subunit (SUR1 or Kir6.2) and the increase upon coexpression was approximately the same when detected by either antibody.
We then used this approach to examine the trafficking of Kir6.2 in lines A and line E. Representative images and the quantitative data are shown in Fig. 7A and B. We were unable to detect a significant difference in trafficking of Kir6.2 when coexpressed with full length SUR1-myc and the most profound deletion, SUR1del249myc. In other words even with the most profound deletion there is no gross alteration in the efficiency of trafficking as assessed using immunofluorescence.
As our assay will not distinguish between protein located at the membrane or just below it, we used a further experimental approach to examine this question. We labelled membrane proteins with biotin using a cell impermeable reagent (see Materials and methods) and purified the resulting products using an avidin derivative complexed to a solid support. A sample of the cellular lysate prior to purification and samples after purification are then subjected to Western blotting with antibodies to SUR1 and Kir6.2. Figure 8A shows that only small amounts of SUR1 are labelled when expressed in a stable cell line alone. However, coexpression of SUR1-myc with Kir6.2 in line A results in significant labelling. Furthermore, coexpression of Kir6.2 with the most profound deletion, SUR1del249myc, in line E results in comparable signal. We also noted that Kir6.2 was purified under such conditions (Fig. 8B) and due to the presence generally of a single distinct band was better suited to quantification. Thus assuming a high efficiency for purification it is possible to estimate the fraction of cell surface protein by performing gel densitometry (see Materials and methods and figure legend). In line A 2.55 ± 1.2% and in line E 1.93 ± 0.38% of Kir6.2 is surface biotinylated (n = 4, not a significant difference). Finally, our conclusion is further supported by the presence of substantial and comparable currents in line A and line E.
In this study we report a comprehensive study in a mammalian cell line of the effects of deletion of the SUR1 C-terminus on assembly, trafficking and function of the KATP channel complex. The data reported here show that the cytoplasmic C-terminus of SUR1 has a key role in channel function but is not absolutely required for complex assembly or trafficking. Our data support aspects of previous studies but also extend those observations [11,13,14].
The cytoplasmic C-terminus of SUR1 containing NBD2 is mutated in persistent hyperinsulinaemic hypoglycaemia of infancy. The disease is autosomal recessive and is characterized by profound neonatal hypoglycaemia and inappropriate insulin secretion from the pancreas. A number of point mutations cluster in this region of SUR1; the resulting frameshift mutations lead to premature truncation of the protein . A number of studies have examined various aspects of the role of the C-terminus, including NBD2, in KATP channel function. Sharma et al.  performed limited deletions of the C-terminus and identified a forward trafficking signal that increased plasma membrane currents. The removal of as few as seven amino acids led to loss of channels at the plasma membrane based on the reduction of sulfonylurea inhibited Rb+ flux, a chemiluminescent trafficking assay and the absence of an apparent higher molecular mass glycosylated species of SUR1. In contrast, Sakura et al.  observed pronounced expression of plasma membrane currents that was not quantitatively different from wildtype. In addition, Schwappach et al.  showed that a SUR1-MRP1 C-terminal chimaera (in which the C-terminus of SUR1 was replaced by MRP1) when coexpressed with Kir6.2 could reach the plasma membrane. One potential explanation for this is that the trafficking in these two systems is quite different. Indeed it is now well established that the commonest of the mutations (ΔF508) in the CFTR protein (cystic fibrosis transmembrane conductance regulator) causing cystic fibrosis forms a functional chloride conductance in Xenopus laevis oocytes but not in mammalian cells. The explanation is that the trafficking is temperature dependent: it occurs at 23 but not 37 °C [22,23]. Furthermore, there are a number of rarer mutations in cystic fibrosis that affect mainly the C-terminus of CFTR and lead to premature truncation of the protein in an analogous fashion to those occurring in SUR1 in persistent hyperinsulinaemic hypoglycaemia of infancy. In a series of elegant studies, Lukacs and colleagues have shown that trafficking through the secretory pathway is not disturbed and maturation occurs normally [24,25]. However C-terminally deleted CFTR is degraded more quickly than wild-type protein via a novel transport mechanism that occurs from the plasma membrane to the proteosome [24,25]. Our data in a human cell line support the idea that the deletion of the C-terminus of SUR1 does not give channel complexes predisposed to temperature dependent trafficking. We also examined whether there was a subtle relative trafficking defect by using a quantitative immunofluorescence assay and cell surface biotinylation and were unable to demonstrate major disturbances in the trafficking. However the assays may not be sensitive enough to pick up a decrease in forward trafficking from ER to Golgi or increase in degradation rate with C-terminal deletion as occurs with CFTR. It is also conceivable, however, that these mutants may traffic differently in different cell lines and that there may be trafficking determinants in the region defined by the largest and smallest deletions. It is worth pointing out that the immunofluoresence assay will not distinguish between protein at the membrane or in submembranous vesicles. As might be expected given the ability of the SUR1 deletion mutants to form functional channels at the membrane, there was also no gross disturbance of interaction between the subunits as assessed by coimmunoprecipitation.
However, our data do reveal quite profound disturbances in the function of the mutant channel complex. The ATP sensitivity is not significantly disturbed and this is supported by more detailed studies in Xenopus laevis oocytes . More interestingly however, as observed by Sakura et al., profound deletion of the C-terminus of SUR1 generates channels that are insensitive to MgADP and diazoxide. In addition we demonstrate here that only relative minor deletions also impair the ability of diazoxide to activate the channels. In particular intact Walker A and B motifs are necessary for diazoxide activity. The current models of the interrelationship of channel opener binding and action, ATP binding and hydrolysis and MgADP binding and stimulation are complex [19,20,26,27,27–34]. Much of this data is compatible with the idea that the hydrolysis of MgATP at NBD2 results in an MgADP bound conformation that leads to channel activation. Openers stabilize the latter state and this effect is modulated by nucleotide binding at NBD1. Our data emphasize the central role that a fully intact NBD2 of SUR1 plays in stimulation by MgADP and diazoxide and support the above hypothesis.
Sulfonylurea action is more subtly altered. The Kd for tritiated glibenclamide binding is not significantly changed. However tolbutamide, at doses known to interact predominantly with the sulfonylurea receptor , causes only a partial reduction in current but with an identical Ki. Tolbutamide has the same affinity for interaction but the efficacy is decreased. Our data differ from those of Sakura et al.  who found no difference for high affinity tolbutamide block when measuring whole-cell currents in two-electrode voltage clamp recordings of macroscopic currents in Xenopus laevis oocytes (they did see some reduction in effect in inside-out patches). However, they used metabolic poisoning to elicit these currents and this may account for the difference. In addition, we demonstrate that this phenomenon occurs with even relatively modest deletions.
What do the observations say about the mechanism of sulfonylurea block? Recent studies [36,37] indicate that the binding pocket is located within the final group of transmembrane domains between NBD1 and NBD2. It is interesting that the coexpression of two SUR hemi molecules is necessary for the full reconstitution of the sulfonylurea binding site . Our data are compatible with the idea that the binding is unaltered but occupation of its site results in less pronounced effects and a change in efficacy dependent on NBD2. It emphasizes that the occupation of the sulfonylurea binding site need not necessarily translate itself into channel closure . Secondly it suggests that a significant proportion of the sulfonylurea effect may be directly related to the antagonism of MgADP induced openings dependent on an intact Walker A and B motif in NBD2. Speculatively, it supports an emerging picture in which NBD2 and the final group of transmembrane domains interact functionally to modulate the pore forming subunit. Finally, these studies demonstrate as in our previous work  that assembly and trafficking may be well preserved but functional coupling between SUR1 and Kir6.2 is critically determined by a number of interdependent factors.
This work was supported by BBSRC, British Heart Foundation, Diabetes UK and the Wellcome Trust. We are grateful to Professor S. Seino for providing Kir6.2 cDNAs and Professor J. Bryan for providing SUR1 cDNA.