Assembly, Maturation, and Turnover of KATP Channel Subunits*

ATP-sensitive K+, or KATP, channels are comprised of KIR6.x and sulfonylurea receptor (SUR) subunits that assemble as octamers, (KIR/SUR)4. The assembly pathway is unknown. Pulse-labeling studies show that when KIR6.2 is expressed individually, its turnover is biphasic; ∼60% is lost with t½ ∼36 min. The remainder converts to a long-lived species (t½ ∼26 h) with an estimated half-time of 1.2 h. Expressed alone, SUR1 has a long half-life, ∼25.5 h. When KIR6.2 and SUR1 are co-expressed, they associate rapidly and the fast degradation of KIR6.2 is eliminated. Based on changes in the glycosylation state of SUR1, the half-time for the maturation of KATP channels, including completion of assembly, transit to the Golgi, and glycosylation, is ∼2.2 h. Estimation of the turnover rates of mature, fully glycosylated SUR1 associated with KIR6.2 and of KIR6.2 associated with Myc-tagged SUR1 gave similar values for the half-life of KATP channels, a mean value of ∼7.3 h. KATP channel subunits in INS-1 β-cells displayed qualitatively similar kinetics. The results imply the octameric channels are stable. Two mutations, KIR6.2 W91R and SUR1 ΔF1388, identified in patients with the severe form of familial hyperinsulinism, profoundly alter the rate of KIR6.2 and SUR1 turnover, respectively. Both mutant subunits associate with their respective partners but dissociate freely and degrade rapidly. The data support models of channel formation in which KIR6.2-SUR1 heteromers assemble functional channels and are inconsistent with models where SUR1 can only assemble with KIR6.2 tetramers.

The rules that govern the maturation and assembly of oligomeric membrane proteins, particularly ion channels, are not well understood. Glycosylation, endoplasmic reticulum (ER) 1 retention and exit mechanisms, and timing are implicated as important factors restricting trafficking to fully assembled complexes (1)(2)(3)(4)(5)(6)(7). Multisubunit ion channels provide excellent models to study these issues. ATP-sensitive K ϩ , or K ATP , channels are large, octameric (K IR 6.x-SUR) 4 complexes (2,8,9) composed of two disparate subunits, potassium inward rectifiers K IR 6.1 (KCNJ8) or K IR 6.2 (KCNJ11), that form the pore of the channel and sulfonylurea receptors SUR1 (ABCC8) or SUR2 (ABCC9), ATP-binding cassette (ABC) proteins that activate the pore and regulate its gating. Glycosylation takes place at two residues within SUR (7,10) and is required for efficient surface expression, implying quality control via the glycoprotein-ER-associated degradation (GERAD) system (11). In addition, dibasic ER retention motifs, specifically an -RKRmotif, have been identified in both K IR and SUR subunits. Their removal or mutation results in loss of the requirement of the partner subunit for correct trafficking (3,12). The retention mechanism is poorly understood. Recent reports (13,14) indicate that the dibasic motif can interact with COPI proteins, resulting in retention (reviewed in Ref. 15). Binding with COP1 complexes can be antagonized by 14 -3-3 isoforms to facilitate ER exit, and the interaction between dimeric 14 -3-3 and the -RKR-motif in K IR 6.2 has been reported to be enhanced by assembly of K IR subunits (14).
Reduced or loss-of-function mutations in K IR 6.2 and SUR1 are known to account for ϳ50% of the mild and severe cases of hyperinsulinemic hypoglycemia (16,17). Nonsense and splice-site mutations that truncate SUR1 result in loss of functional channels at the cell surface in patient ␤-cells and retention of truncated SUR1 in the ER (4,18). Missense mutations have been identified that result in loss of stimulation by MgADP (19 -22), and other mutations are reported to affect trafficking to the cell surface as a consequence of improper folding or possibly by altering interactions with the ER retention motif(s) (23)(24)(25). One general suggestion has been that retention in the ER provides additional time for completion of channel assembly. K IR 6.2 subunits lacking the -RKR-motif can assemble channels in the absence of a SUR, and these will reach the surface indicating SUR is not obligatory for assembly of the channel pore. The assembly pathway is unknown, and at what stage SUR interacts with K IR is a matter of conjecture. Similarly, the lifetimes of the individual subunits, of mutant subunits, and of the channel complex have not been reported.
We have used 35 S[Met/Cys] pulse-labeling methods to determine the lifetimes of SUR1 and K IR 6.2 alone, to examine the effect of two severe hyperinsulinemic mutations on subunit lifetime, and to follow the maturation of channels by monitoring changes in the glycosylation state of SUR1. The results show that when SUR1 is expressed in the absence of the inward rectifier it is long-lived in contrast, for example, to CFTR, which has a half-life of ϳ30 min in the ER (26,27). When K IR 6.2 is expressed in the absence of SUR1, a fraction (ϳ60%) degrades rapidly, but the remainder is converted to a long-lived species. When K IR 6.2 and SUR1 are co-expressed, complexes form quickly and the rapid degradation of K IR is eliminated. In the complex, SUR1 undergoes a time-dependent change in its glycosylation state as the channel transits to the cell surface. Two loss-of-function mutations, ⌬F1388 in SUR1 and W91R in K IR 6.2, both drastically decrease subunit lifetime irrespective of whether they are co-assembled with their partner subunit.

EXPERIMENTAL PROCEDURES
Molecular Biology and Plasmid Construction-K IR 6.2 and SUR1 proteins were modified by addition of extracellular Myc tags. The details of construction of the hamster pECE SUR1, pECE SUR1Myc, and human pECE K IR 6.2 have been described (2,4,28).
Cell Culture and Transient Transfection-COS m6 cells were cultured in Dulbecco's modified Eagle's medium, 4.5 g/liter glucose, supplemented with 10% fetal bovine serum. Approximately 7 ϫ 10 6 cells were transfected with 8 g of the pECE SUR1 and/or 1 g of the pECE K IR 6.2 plasmids by electroporation (BioRad Laboratories, Inc.) following the manufacturer's directions (950 F, 0.220 V in 0.4-mm cuvettes in RPMI medium supplemented with 10% fetal bovine serum and 1.25% Me 2 SO). The average efficiency of transfection was 7-10%, estimated by co-transfection with a green fluorescent protein-marker plasmid. There was no significant difference in transfection efficiency when multiple plasmids were used. Pulse-chase experiments were carried out 18 h after transfection.
Pulse-Chase and Immunoprecipitation Protocol-Cells were transfected and seeded in 6-well plates in duplicate at 70 -80% confluence. The following day cells were incubated for 2 h in L-methionine/L-cysteine-free media (Invitrogen) at 37°C with 5% CO 2 . Subsequently, cells were labeled for 60 min with a mixture of [ 35 S]methionine/cysteine (200 Ci/ml; EaseTag TM mix, PerkinElmer Life Sciences)and then incubated with pre-warmed Dulbecco's modified Eagle's medium supplemented with 5 mM unlabeled L-methionine/L-cysteine. At the indicated times, wells were washed twice with cold phosphate-buffered saline, collected, and then lysed on a rotating wheel for 6 h at 20°C in 1 ml of 1% digitonin (Sigma) in phosphate-buffered saline plus cysteine and serine protease inhibitors (Roche Applied Science). Lysates were clarified at 9000 ϫ g for 15 min at 4°C. The supernatants were pre-absorbed for 1 h with protein-G plus and then incubated with anti-Myc or anti-K IR 6.2 (N-18) antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 20°C followed by a 3-h incubation with protein-G-agarose (Santa Cruz Biotechnology). The agarose beads were washed five times with modified lysis buffer containing 0.1% digitonin. A modified 2ϫ loading sample buffer (2.5% SDS, 0.1 M dithiothreitol, 0.06 M Tris base, 20% glycerol, 0.008 M EDTA, pH 6.8) was added to the beads. Proteins were separated on 7.5% or 10% SDS-polyacrylamide gels (30). Gels were stained, fixed in 50% methanol, 10% acetic acid, then treated with Fluorenhance (Amersham Biosciences) for 30 min, dried, and exposed to either x-ray film (Hyperfilm; (Eastman Kodak Co.) at 80°C for 16 -24 h or put on an intensifying screen for quantification with a Storm TM PhosphorImager (Amersham Biosciences).
Note on Sample Preparation-The mobility of SUR1 on SDS-polyacrylamide gels is dependent on its glycosylation state; the immature or "core" glycosylated form displays an apparent molecular mass of ϳ140 kDa and the mature, fully glycosylated form a mass of ϳ150 -170 kDa (1,2). Because maturation of membrane glycoproteins occurs in the medial Golgi apparatus (reviewed in Ref. 31), the mobility difference is a marker for trafficking of SUR1 (4). During the course of these experiments we observed that inclusion of the usual heating step (95°C, 5 min) in the preparation of samples for SDS-PAGE resulted in aggregation. Fig. 1, left panel shows the effect of heating solubilized immunoprecipitates containing SUR1Myc. The right panel shows a similar effect for fully glycosylated, Myc-tagged SUR1 present when the SUR and K IR subunits are co-expressed. Additionally, the lower right panel confirms an earlier report on the reduced recovery of K IR 6.2 following heating (32). The appearance of the heated, Myc-tagged SUR1 samples is not affected by treatment with endoglycosidases (data not shown). The samples in Fig. 1 were identified by photolabeling with 125 I[azido] glibenclamide (2), but equivalent results were obtained using [ 35 S]Met/ Cys-labeled subunits. To avoid aggregation, samples were not subjected to heating before SDS-PAGE.
Data Analysis-The relative intensities of bands were determined using phosphorimaging (Molecular Dynamics). The data are plotted as intensity values or as the fraction remaining obtained by dividing by the zero time value. We have assumed homogenous compartments and that degradation of a given molecule is a random event. Single exponential functions were fit to the data, with the exception of K IR 6.2 expressed in the absence of SUR where two exponentials were used. The results are expressed as the half-life for a given species. The half-life values can be converted to the mean residence times for a molecule in a compartment by dividing by 0.693. We used a simple two-compartment model in Modelmaker 4 (Cherwell Scientific, Ltd., Oxford, UK) to simulate the conversion of monomeric K IR 6.2 to the tetramer and estimate the half-time for the assembly. All other fitting was done using the non-linear routines in Origin Pro (OriginLabs Corp., Northampton, MA). The results are given as mean Ϯ S.D.

Free K ATP Channel Subunits Turn Over with Different Kinet-
ics-SUR1 is a member of the ABC superfamily, specifically the ABCC subfamily. CFTR, another ABCC protein, whose turnover and trafficking has been examined in some detail, degrades rapidly in the ER, a t1 ⁄2 ϳ30 min (26,27), with only 20 -40% estimated to leave the ER. The mature, fully glycosylated CFTR at the cell surface is longer-lived, reported t1 ⁄2 values of 7.5-16 h (26,27). P-glycoprotein (MDR1), another ABC protein, has reported half-lives of Ͼ24 h, whereas mutants have been described with t1 ⁄2 values of ϳ3 h (33-35). 35 S[Met/Cys] pulse-chase experiments with COS m6 cells transfected with Myc-tagged SUR1, in the absence of K IR 6.2, show there is no conversion to the mature species and that the immature receptor turns over slowly in the ER ( Fig. 2A). A single exponential decay fits the data points reasonably well, giving an estimated half-life of ϳ25.5 Ϯ 4.4 h (n ϭ 6) for SUR1 subunits in the ER in the absence of the inward rectifier.
The turnover of K IR 6.2 expressed alone is markedly different from SUR1. Turnover is biphasic with ϳ60% of the 35 S[Met/ Cys]-labeled subunits being rapidly degraded with a half-life ϳ35.7 Ϯ 11.8 min. The remaining 40% turns over slowly with a half-life ϳ26.1 Ϯ 8.2 h (n ϭ 3). K IR 6.2 can assemble functional tetrameric pores in the absence of SUR that will reach the cell surface if their ER retention sequences are altered or deleted (36 -38). Thus we propose that the biphasic decay of the K IR 6.2 reflects a difference in the rates of degradation of unassembled subunits versus assembled pores. We used this two-state assumption and the estimated degradation rates to model the turnover assembly process. The estimated half-life for conversion of a K IR 6.2 monomer into a stable species, assumed to be a tetramer, is ϳ1.2 h. The solid line in Fig. 2B was calculated using these parameters.
Maturation of K ATP Channels-The 35 S[Met/Cys] pulsechase protocol followed by immunoprecipitation with either anti-K IR 6.2 or anti-Myc antibodies was used to assess the as-FIG. 1. Heating in SDS alters the mobility of SUR1. COS m6 cells were transfected with Myc-tagged SUR1 with or without K IR 6.2, photolabeled with [ 125 I]azido glibenclamide, and subjected to immunoprecipitation as described. The immunoprecipitates were solubilized in a modified SDS sample buffer and either heated (95°C, 5 min) or held at room temperature. Fig. 1, left panel shows the effect of heating solubilized immunoprecipitates containing Myc-tagged SUR1. The right panel shows a similar effect for fully glycosylated SUR1 present when Myc-tagged SUR and K IR subunits are co-expressed. Additionally, the lower right panel confirms an earlier report on the reduced recovery of K IR 6.2 following heating (32). The appearance of the heated Myc-tagged SUR1 samples is not affected by treatment with endoglycosidases. These samples were identified by photolabeling with [ 125 I]azido glibenclamide, but equivalent results were obtained using [ 35 S]Met/Cys-labeled subunits and wild type SUR1. The addition of 1 M unlabeled glibenclamide (glib) eliminates photolabeling.

FIG. 2. Turnover of individual K ATP channel subunits.
A, in the absence of K IR 6.2, SUR1 is long-lived. COS m6 cells transfected with Myc-tagged SUR1 were pulse-labeled with [ 35 S]methionine/cysteine for 60 min and then processed as described under "Experimental Procedures" using anti-Myc antibodies. A single exponential function was fit to the data, given as the fraction remaining. The t1 ⁄2 ϭ 25.5 Ϯ 4.4 h (n ϭ 6). B, the same protocol, with anti-K IR 6.2 antibodies, was used to analyze the turnover of K IR subunits in the absence of SUR. The results show that turnover is biphasic with ϳ60% of the subunits being rapidly lost. A double exponential function was fit to the data. The t1 ⁄2 values were 35.7 Ϯ 11.8 min and 26.1 Ϯ 8. sembly and maturation of K ATP channels in COS m6 cells co-transfected with both SUR1 and K IR 6.2. In principle, using two antibodies should provide information on the assembly of SUR1 with K IR 6.2 complexes and vice versa. The results using the anti-K IR antibodies are shown in Fig. 3A. Co-expression with SUR1 markedly affects the turnover of K IR 6.2. The rapid decay, evident in Fig. 2B when K IR 6.2 is expressed alone, is missing. The amount of K IR 6.2 in the immunoprecipitates increases over the first 60 min and then decays with a half-life of ϳ8.5 Ϯ 4 h, providing one measure of channel lifetime. The result implies SUR1 must rapidly associate with and stabilize the early assembly intermediates (Fig. 3A). This interpretation is supported by the observation that K IR 6.2 and SUR1 coimmunoprecipitate at the earliest time point, increasing in amount for ϳ2 h (compare first bands in Fig. 3, A and B). In addition, K IR 6.2-SUR1 complexes were detectable after brief (10 min) pulse-labeling periods (data not shown). As shown in Fig. 3B, there is essentially no detectable fully glycosylated SUR1 present at the earliest times, but the mature receptor becomes evident within 2 h, peaking at ϳ10 h. Because the immunoprecipitation was carried out with anti-K IR 6.2 antibodies, SUR1 must be associated with K IR 6.2. The appearance of the mature receptor as the core species disappears is consistent with a precursor-product relationship. The estimated half-life for conversion to mature SUR1 is ϳ2.2 Ϯ 0.14 h, based on fitting a single exponential (Fig. 3B, dotted line) to the disappearance of the immature form of the receptor beginning with the 2-h time point. The estimated half-life of the channel complex, based on fitting a single exponential to the mature SUR1 data, beginning with the 10-h time point, is ϳ5.9 Ϯ 1 h. This is in reasonable agreement with the ϳ8.5 Ϯ 4 h estimate for K IR 6.2 because the K IR number will include contributions from any subunits not assembled with SUR1.
The results for co-immunoprecipitation of Myc-tagged SUR1 and K IR 6.2 with anti-Myc antibodies are shown in Fig. 4. Consistent with the anti-K IR 6.2 immunoprecipitation result (Fig.  3A), K IR 6.2 does not turn over rapidly when associated with SUR1. There is a progressive increase in the amount of K IR 6.2 precipitating with SUR1, with a peak at ϳ5 h, consistent with slow sequential assembly. The estimated channel half-life, based on the lifetime for K IR 6.2 associated with SUR1, is 7.8 Ϯ 1.1 h. The precursor-product relation between immature and mature SUR1 is apparent with nearly the same timing (halflife for conversion of core-to-complex SUR1 is ϳ2.6 Ϯ 0.6 h). The estimated channel half-life, based on fitting a single exponential to the mature SUR1 data, is 6.9 Ϯ 0.2 h. The ratio of immature SUR1-to-K IR 6.2 at the end of the 35 S[Met/Cys] pulse (t ϭ 0) is higher than that observed for immunoprecipitation with anti-K IR 6.2 (9.5 versus 2.9, respectively), implying SUR1 is in excess under these transfection conditions. This idea is supported by comparison of the later time points in the SUR1 intensity data in Figs. 3B and 4B. The maturation of SUR1, marked by conversion to the mature form, is essentially complete in less than 20 h for SUR1 associated with K IR 6.2 (Fig.  3B), whereas the immature form remains readily detectable after 30 h in the anti-Myc immunoprecipitates (Fig. 4B). The result is consistent with there being non-K IR 6.2-associated SUR1 in the ER that turns over slowly as shown in Fig. 1A.
Increased Rates of Turnover for Two K ATP Channel Mutants-A number of mutations in either SUR1 or K IR 6.2 are known to cause familial hyperinsulinism (10,39,40); several have been shown to affect trafficking, reportedly as a consequence of improper folding. We have used the 35 S[Met/Cys] pulse-chase immunoprecipitation protocol to examine the turnover of two mutant subunits, SUR1 ⌬F1388 and K IR 6.2 W91R, expressed alone or with their respective partner. The ⌬F1388 mutation is in the second nucleotide binding domain of SUR1 and has been identified as a common cause of familial hyperinsulinism in Ashkenazi populations (41). The W91R mutation substitutes an arginine for the second in a pair of tryptophan residues near the top of K IR 6.2 (10). In KcsA, this tryptophan pair is in the pore helix and is thought to contribute significantly to the stability of the tetramer (42). Both mutations are associated with severe forms of hyperinsulinism. Fig. 5 compares the turnover of K IR 6.2 W91R versus wild type K IR 6.2 subunits. K IR 6.2 W91R is degraded rapidly in either the presence or absence of SUR1. The conversion to a long-lived species is missing in the mutant. Co-immunoprecipitation experiments show that K IR 6.2 W91R subunits do associate with SUR1, but this association does not appear to slow degradation significantly. The estimated half-lives for the K IR 6.2 W91R subunits in the absence and presence of SUR1 are 26.4 Ϯ 2.4 min (n ϭ 2) versus 34.0 Ϯ 2.6 min (n ϭ 2). These values are not significantly different from those determined for the fastest component observed when wild type K IR 6.2 is expressed in the absence of SUR1 (35.7 Ϯ 11.8 min; Fig. 2B). The results are consistent with the hypothesis that the W91R subunits are unable to assemble into stable tetramers and turn over rapidly as free monomers. To support this idea, we expressed Myc-tagged K IR 6.2 with either K IR 6.2 W91R or wild type K IR 6.2 and then determined the levels of W91R versus wild type K IR 6.2 in anti-Myc immunoprecipitates (Fig. 6). The Myc tag does not substantially alter the properties of K IR 6.2-SUR1 channels but has a greater molecular mass and serves to identify the "carrier" subunits in the immunoprecipitates. K IR 6.2 W91R is degraded more rapidly than wild type K IR 6.2, being nearly undetectable after 5 h, whereas the turnover of wild type K IR 6.2 parallels that of the Myc-tagged K IR 6.2 subunits (Fig. 6).
The turnover of SUR1 ⌬F1388 is dramatically faster than the wild type receptor (Fig. 7). The half-lives for SUR1 ⌬F1388 versus wild type SUR1, expressed alone, are 3 Ϯ 0.4 (n ϭ 2) versus 25.5 Ϯ 4.4 (n ϭ 3) hours, respectively. The SUR1 ⌬F1388 subunits are able to assemble with K IR 6.2 (Fig. 7B), but this association does not affect turnover of the mutant receptor significantly. We see no mature, fully glycosylated mutant receptors, consistent with their not leaving the ER. The estimated half-life for the K IR 6.2-SUR1 ⌬F1388 complex is 2.1 Ϯ 0.6 h, not significantly different from the mutant receptor alone. The association with SUR1 ⌬F1388 does significantly slow the degradation of K IR 6.2 (Fig. 7B). We assume there is a constantly changing mixture of SUR1-K IR 6.2 complexes in this experiment but have fit a single exponential function, t1 ⁄2 ϭ 5.6 Ϯ 2.3 h, to the data for purposes of comparison. It is noteworthy that both immunoprecipitations show enrichment of K IR 6.2 subunits versus SUR1 ⌬F1388 at later times, consistent with dissociation of ⌬F1388 from the complexes and subsequent degradation.

Turnover of K ATP Channel Subunits in the INS-1 ␤-cell line-
The kinetics of turnover and maturation of K ATP channel subunits in INS-1 ␤-cells, determined by immunoprecipitation with anti-K IR antibodies, were similar to what was observed in transfected COS m6 cells (Fig. 8). The level of expression of K ATP channel subunits is lower than in transfected cells, but mature SUR1 was detectable between two and three hours after a 1-h [ 35 S]methionine/cysteine pulse. Based on the disappearance of immature SUR1, the half-life for maturation was 2.2 Ϯ 0.5 h. The average half-life of fully assembled channels, the mean of the values determined for K IR 6.2 (6.2 Ϯ 2.0 h) and mature SUR1 (6.8 Ϯ 2.4 h), is 6.5 Ϯ 2.0 h (n ϭ 3).

DISCUSSION
Pulse-chase methods were employed to determine the turnover rates of K ATP channel subunits expressed alone or together. Expressed individually, both subunits have long-lived species, half lives near 24 h. This stability is consistent with the idea that these channels assemble slowly. The difference between the extended half-life of SUR1 and the rapid turnover of the related ABCC protein, CFTR, is noteworthy (reviewed in Refs. 34 and 43). Whereas SUR1 is long-lived in the ER, the behavior of unaccompanied K IR 6.2 is more closely akin to the turnover observed for CFTR, with ϳ60% of the K IR 6.2 subunits being degraded before they can convert to a stable long-lived species. The means of stabilization are different; CFTR eventually folds correctly and leaves the ER, whereas based on the fact that SUR1 is not obligatory for formation of the pore and K IR 6.2 subunits can assemble functional homomeric channels, we suggest the short-and long-lived species are K IR monomers and tetramers, respectively, and have interpreted the results using this assumption. A simulation, based on this two-state model, using the measured turnover rates, indicates the half-time for assembly of tetramers from K IR 6.2 monomers is ϳ1.2 h.
The formation of K IR 6.2-SUR1 heteromers is rapid. We have not attempted to determine their rate of association, but complexes are detectable after a 10-min labeling pulse and their formation stabilizes K IR monomers against rapid degradation.
Following completion of assembly, channels transit to the Golgi and SURs convert from the core to the mature, fully glycosylated form in a precursor-product fashion. We have assumed that octameric K ATP channels, once formed, do not dissociate readily and have estimated channel half-life by determining the rates of loss of K IR 6.2 or SUR1 from immunoprecipitates obtained using either anti-K IR 6.2 or anti-Myc antibodies, respectively. The assumption of channel stability is supported by the similar half-life values determined for the accompanying partner subunit. The four values range from 5.9 -8.5 h, with a mean value of 7.3 Ϯ 1.6 h. The pattern of maturation and turnover of K IR 6.2 and SUR1 in INS-1 ␤-cells is similar with an estimated channel half-life of 6.5 Ϯ 2.2 h, implying the data obtained with reconstituted channels reflects subunit processing in ␤-cells.
The precursor-product relation between the immature and mature receptors was used to estimate the timing of channel maturation. Subunit synthesis and initial glycosylation of SUR1 take place concurrently in the ER, whereas mature glycosylation takes place in the medial Golgi. We observe core glycosylated SUR1-K IR 6.2 complexes at the earliest times and can follow their conversion to the fully glycosylated species. Based on the loss of core glycosylated SUR1 from the complex (Figs. 3B and 4B), the estimated half-time for maturation of the K IR 6.2-SUR1 complexes is ϳ2.2 h. What maturation entails is not well understood but must involve subunit folding, assembly into a complete channel, transit through the Golgi, addition of sialic acid residues in the Golgi, and transit to the cell surface.
FIG. 5. K IR 6.2 W91R turns over rapidly. COS m6 cells expressing K IR 6.2 W91R subunits, in the presence or absence of Myc-tagged SUR1, were pulse-labeled and then processed as described under "Experimental Procedures" using either anti-K IR 6.2 or anti-Myc antibodies. Single exponential functions were fit to the data. The t1 ⁄2 values, with (filled squares) or without (open squares) SUR1, are 34.0 Ϯ 2.6 (solid line) and 26.4 Ϯ 2.4 (dashed line) minutes (n ϭ 2), respectively. The dotted line illustrates the turnover of wild type K IR 6.2 subunits expressed in the absence of SUR (taken from Fig. 2A). The gel illustrates K IR 6.2 W91R turnover in the absence of SUR1, but the results were equivalent when Myc-tagged SUR1 was present and anti-Myc antibodies were used. The times shown are 0, 0.5, 1, 2, 3, 5, 10, 21, 25, and 30 h. Values are mean Ϯ S.D.
FIG. 6. W91R subunits are eliminated from K IR 6.2 complexes. COS m6 cells were co-transfected with equimolar amounts of either the W91R or a wild type K IR 6.2 plasmid plus the Myc-tagged K IR 6.2 plasmid, pulse-labeled, and then processed as described under "Experimental Procedures" using anti-Myc antibodies. W91R subunits co-assemble with Myc-tagged K IR 6.2 but dissociate and are progressively lost from the complexes, whereas the wild type and Myc-tagged K IR s have indistinguishable rates of turnover. Note that the pattern of turnover of Myc-tagged K IR 6.2 differs somewhat from that of K IR 6.2, suggesting the Myc-tagged subunit may convert more slowly to the stable species. Myc-tagged K IR 6.2, open circles and squares; wild type K IR 6.2, filled circles; K IR 6.2 W91R, filled squares.
FIG. 7. SUR1 ⌬F1388 turns over rapidly. COS m6 cells were transfected with Myc-tagged SUR1 ⌬F1388 with and without K IR 6.2, pulse-labeled, and then processed as described under "Experimental Procedures" using either anti-Myc antibodies (A) or anti-K IR 6.2 antibodies (B). A, the turnover of SUR1 ⌬F1388 is essentially the same in the presence or absence of K IR 6.2, 2.1 Ϯ 0.6 versus 3 Ϯ 0.4 h, open versus filled circles (n ϭ 3), respectively. The dotted line illustrates the turnover of Myc-tagged SUR1 in the absence of K IR 6.2 (from Fig. 2A). Only the immature form of ⌬F1388 subunits is present, indicating failure to transit to the Golgi apparatus. The top gels illustrate the turnover of both K IR 6.2 and SUR1 ⌬F1388. Note that although the Myc-tagged ⌬F1388 is barely detectable, K IR 6.2 is enriched. The ⌬F1388 bands are intentionally overexposed. The times shown are 0, 0.5, 1, 2, 3, 5, 10, 21, 25, and 30 h. B, assembly with SUR1 ⌬F1388 slows the turnover of K IR 6.2 subunits associated with SUR1 ⌬F1388. A single exponential function, t1 ⁄2 ϭ 5.6 Ϯ 2.3 h (n ϭ 2), was fit to the data, although multiple populations are presumably present. The dotted line illustrates the turnover of wild type K IR 6.2 subunits expressed in the absence of SUR (from Fig. 2B). The lower gels illustrate subunit turnover using anti-K IR 6.2 immunoprecipitation. The times shown are 0, 0.5, 1, 2, 3, 5, 10, 21, and 25 h. K IR 6.2 is enriched at the later time points. Values are mean Ϯ S.D. The 2.2-h figure is an estimate of the average time between synthesis and the final addition of sialic acid residues. Estimates for the transit time of monomeric proteins between ER and Golgi vary widely, but most are shorter that 2.2 h. GLUT1 and GLUT4 are reported to take 5 versus 20 min, respectively, in 3T3-L1 adipocytes (44). The mean residence time of a green fluorescent protein-tagged mutant of the vesicular-stomatitis virus protein in both the ER and Golgi is estimated to be ϳ40 min (45,46). If the residence time(s) for complete K IR 6.2-SUR1 channels in the ER and Golgi compartments are similar, the estimated maturation time of 2.2 h is consistent with the idea that although K IR 6.2 and SUR1 subunits rapidly associate, the rate-limiting step is the slow assembly of these heteromeric complexes into complete octameric channels. This idea is supported by the slow formation of homomeric K IR 6.2 pores (t1 ⁄2 ϳ1.2 h), by the slow maturation of K ATP channel subunits in INS-1 ␤-cells, and by the relatively slow maturation of other K ϩ channels, e.g. 80% conversion of immature Shaker subunits to the mature form in 1.5 h (47), independently of the presence of the Kv␤2 subunit, and a t1 ⁄2 ϳ3 h for conversion of core glycosylated Kv1.4 to the fully glycosylated form (48).
Subunit Mutations and Degradation-Multiple mutations in both SUR1 and K IR 6.2 have been identified in patients with congenital hyperinsulinism (10,39,40). Truncations of the C terminus of SUR1 have been reported to affect trafficking to the cell surface (4). A number of point mutations have been identified that exhibit channel activity in isolated patches and display normal inhibition by ATP but fail to be stimulated by MgADP (19,20,49,50). The ⌬F1388 mutation in SUR1 was one of the first mutations to be identified and is a frequent mutation in Ashkenazi populations (41). This is a severe mutation, and no channel activity was detectable when SUR1 ⌬F1388 subunits were expressed with wild type K IR 6.2, although binding/labeling studies showed the mutant receptor was expressed (41). Interestingly, SUR1 ⌬F1388 was later shown to associate with K IR 6.2, and a limited number of channels can be made to traffick to cell surface if the -RKR-motif is changed to -AAA- (24). Analysis of the K IR 6.2-SUR1 ⌬F1388 channels in which the -RKR-motif has been substituted with -AAA-showed they have a defective stimulatory response to MgADP. We show that the SUR1 ⌬F1388 subunits have a significantly shorter residence time in the ER than wild type SUR1, 3 Ϯ 0.4 versus 25.5 Ϯ 4.4 h, consistent with the idea that they are defective and are targeted for degradation. The defect must be subtle because the ⌬F1388 subunits are able to associate with K IR 6.2 subunits and protect them against rapid degradation. Although we fit a single exponential to the K IR decay data, there are almost certainly multiple populations present because we are able to precipitate K IR 6.2 free of SUR1 ⌬F1388 at the later time points (Fig. 7, Ͼ10 h).
The W91R mutation was identified by our group in a patient of Palestinian origin with severe hypoglycemia but has  9. Summary. The turnover results are summarized within the framework of a model for K ATP channel formation. We hypothesize that K IR monomers assemble slowly into tetramers. In the absence of SUR, K IR monomers are degraded rapidly, whereas tetramers are long-lived. When both subunits are expressed, core glycosylated SUR* forms complexes rapidly with K IR 6.2 as implied by the larger arrow. SUR* associated with K IR 6.2 is converted slowly to complex glycosylated SUR1**. This maturation process (identified by the dashed arrows), an estimated half-time of 2.2 h, involves multiple steps including completion of channel assembly, transit through the Golgi complex, and the addition of sialic acid residues. The estimated half-life for the complete, fully glycosylated channel is ϳ7.3 h, the mean value of the turnover rates for both subunits in complete channels. not been studied in detail (10). Co-expression of K IR 6.2 W91R with wild type SUR1 failed to produce active channels at the cell surface, and we did not detect mature receptors. Trp-91 is the second of a pair of tryptophan residues near the top of K IR 6.2. In KcsA, an equivalent tryptophan pair is positioned with other aromatic amino acids to stabilize the selectivity filter via hydrophobic interactions (42). Based on the KcsA structure, the W91R substitution is expected to disrupt stabilization by burying four charged arginine residues within this hydrophobic layer. Our results show the W91R subunits can associate with SUR1 and will associate with other K IR 6.2 subunits. This association appears to be transient, and like SUR1 ⌬F1388, the mutant subunits are lost from the channel complex.
Summary Model-The results are summarized in Fig. 9. In the absence of SUR1, K IR 6.2 is either degraded (ϳ60%) or, we propose, assembles relatively slowly into long-lived tetramers. This situation is not physiological, and these pores presumably remain in the ER, retained by their -RKR-retention motifs (3). The lack of K ATP channels in Sur1 knock-out mice supports this conclusion (51). When both subunits are present, K IR 6.2-SUR1 complexes form rapidly (indicated by the thick arrow) and are detectable using short labeling times. K IR 6.2 subunits are stable when complexed with SUR1. The slow assembly of K IR 6.2 monomers combined with the disappearance of the rapid degradation phase when SUR1 is present argues against the idea that SUR1 can only assemble with K IR 6.2 tetramers. By following the glycosylation state of SUR (symbolized by the asterisk) subunits associated with K IR , we estimate a maturation time for channels consisting of completion of assembly, transit to the Golgi, and full glycosylation (double asterisk) of ϳ2.2 h (dashed arrows). It is not clear whether association with SUR increases the rate of channel formation. We assume the glycosylation reaction(s) are fast; although the transit times of monomeric proteins from the ER to the Golgi vary widely, they are generally shorter than one hour, whereas other K ϩ channels that require assembly have comparable times. We have not attempted to estimate the time of transit from the Golgi to the cell surface using this data, but for other membrane proteins this step is relatively rapid. Once assembly is complete, the channel subunits do not appear to dissociate as judged by their comparable turnover rates, which are consistent with a half-life on the cell surface of about 7.3 h.
A striking feature of both of the hyperinsulinemic mutations examined here is that although they are able to form complexes with their partners the complexes are short-lived, implying the subunits dissociate and degrade. This produces the interesting situation, for example, where K IR 6.2-SUR1 ⌬F1388 complexes assemble transiently, protect the K IR 6.2 monomer against degradation, and then dissociate and turn over, whereas the K IR 6.2 tetramers live on. This idea is supported by the lack of SUR1 ⌬F1388 in the anti-K IR 6.2 immunoprecipitates after Ͼ10 h (Fig. 7). A similar situation occurs when W91R subunits associate with wild type K IR s. We expect that the rapid turnover and instability of the K IR -SUR complexes would significantly reduce the likelihood that complete octameric channels assemble and thus are able to exit the ER. This provides a plausible explanation for the failure of SUR1 ⌬F1388 to reach the cell surface when expressed with wild type K IR 6.2 (24,41) and suggests that the substitution of the -RKR-retention motif with -AAA-in SUR1 ⌬F1388 may improve surface expression primarily by allowing incomplete channels to exit the ER.
The SUR1 ⌬F1388 and K IR W91R subunits are representative of a class of defective channel assembly mutations. These mutations are recessive; patients homozygous for either mutation lack functional K IR 6.2-SUR1 K ATP channels at the ␤-cell surface. Heterozygous carriers would be expected to have a reduced complement of normal channels because the defective subunits would be weeded out by virtue of their increased tendency to dissociate from channels and be degraded.