This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Aguilar-Bryan, L. Articles by Bryan, J. Search for Related Content PubMed PubMed Citation Articles by Aguilar-Bryan, L. Articles by Bryan, J.

Molecular Biology of Adenosine Triphosphate-Sensitive Potassium Channels¹

Departments of Medicine (L.A.-B.) and Cell Biology (J.B.), Baylor College of Medicine, Houston, Texas 77030

	Abstract

Top Abstract I. Introduction II. How Are KATP... III. How Do KATP... IV. KATP Channel Subunits V. Reconstitution of KATP... VI. KATP Channel Structure VII. Regulation of KATP... VIII. Human SUR1 and... IX. KATP Channels and... X. Linking PHHI to... XI. Other Issues XII. KATP and NIDDM XIII. The Leptin Connection XIV. Summary and Conclusions References

 

I. Introduction

II. How Are K_ATP Channels Defined?

III. How Do K_ATP Channels Affect the Membrane Potential of Pancreatic ß-Cells?

IV. K_ATP Channel Subunits

A. The K_IR family of inwardly rectifying K+ channels

B. Sulfonylurea receptors

V. Reconstitution of K_ATP Channel Activity from SUR1 and K_IR6.2

A. The question of "promiscuous coupling" of SUR1 with other inward rectifiers

VI. K_ATP Channel Structure

A. K_IR6.2 forms the pore of a K_ATP channel

B. SUR1 and K_IR6.x are physically associated

C. Coexpression with K_IR6.2 affects the maturation of SUR1

D. Complex glycosylated SUR1 and K_IR6.2 assemble a large multimer

E. A 1:1 stoichiometry of SUR to K_IR is both necessary and sufficient to make K_ATP channels

F. Other K_IR channels are tetramers

G. The stoichiometry of active ß-cell K_ATP channels is (SUR1/K_IR6.2)₄

VII. Regulation of K_ATP Channel Activity

A. How do ATP and ADP exert their effects on K_ATP channels?

B. Where are the nucleotide binding sites located?

C. C-terminally truncated K_IR6.2 channels show abnormal kinetics

D. Coexpression of K_IR6.2C subunits with SUR restores normal K_ATP channel activity

E. Why are K_IR6.2 channels silent?

F. The N terminus of K_IR6.2 limits burst duration

G. Where do the openers bind and how do they work?

H. Do SURs have adenosine triphosphatase (ATPase) activity?

I. Do SURs have transport activity?

J. Is there an endogenous substrate?

VIII. Human SUR1 and K_IR6.2 Genes

IX. K_ATP Channels and Persistent Hyperinsulinemic Hypoglycemia of Infancy (PHHI)

A. HI-GK

B. HI-GlnDH

C. HI-"unknown"

D. HI-K_IR6.2

E. HI-SUR1

X. Linking PHHI to Defects in K_ATP Channel Activity

A. ß-Cells from newborns diagnosed with "sporadic" PHHI lack K_ATP channel activity

B. PHHI ß-cells with the SUR1 exon 35 mutation lack K_ATP channel activity

C. Why is there a lack of dominant negative mutations?

D. Development of mouse models

XI. Other Issues

A. Nesidioblastosis does not cause PHHI

B. "Diffuse" vs. "focal" forms of PHHI

XII. K_ATP and Non-Insulin-Dependent Diabetes Mellitus (NIDDM)

A. ß-Cell type K_ATP channels in the brain

XIII. The Leptin Connection

XIV. Summary and Conclusions

	I. Introduction

Top Abstract I. Introduction II. How Are KATP... III. How Do KATP... IV. KATP Channel Subunits V. Reconstitution of KATP... VI. KATP Channel Structure VII. Regulation of KATP... VIII. Human SUR1 and... IX. KATP Channels and... X. Linking PHHI to... XI. Other Issues XII. KATP and NIDDM XIII. The Leptin Connection XIV. Summary and Conclusions References

 
ION CHANNELS are present in the plasma membrane and intracellular organelles of all cells, where they coordinate such diverse functions as neurotransmission, contraction, secretion, and control of cell volume. Over the past decade it has become increasingly clear that mutations in the genes that encode the subunits of ion channels can result in pathological states in a wide variety of tissues, including those not generally thought of as "excitable." The number of ion channels that give rise to "channelopathies" is increasing. The list currently spans all of the major channel types, including the following nonexhaustive survey. Perhaps the most prevalent ion channelopathy is cystic fibrosis (Online Inheritance in Man, OMIM 219700), which is due to reduced chloride ion transport resulting from mutations in the cystic fibrosis transconductance regulator, CFTR, a member of the large ATP-binding cassette (ABC) superfamily of proteins. A dominant form of myotonia congenita (OMIM 160800), Thomsen’s disease, is associated with mutations in CLCN1, a voltage-gated chloride channel. Mutations in the gene encoding another member of this family, CLCN5, give rise to hypercalciuric nephrolithiasis (Dent’s disease; OMIM 300009), an X-linked recessive disorder. Mutations in the genes encoding the ß- and -subunits of the amiloride-sensitive, non-voltage-gated epithelial Na⁺ channels, ENac, activate channel activity resulting in Liddle’s syndrome (OMIM 177200), which is associated with hypertension and hypokalemia. Various episodic disorders have been shown to arise from mutations of voltage-dependent ion channels in nerve and muscle. Periodic paralysis I, or hypokalemic periodic paralysis (OMIM 170400), results from mutations in the gene encoding the 1-subunit of the muscle dihydropyridine-sensitive calcium channel, while periodic paralysis II, hyperkalemic periodic paralysis (OMIM 170500), is associated with mutations in the gene encoding a voltage-gated Na⁺ channel subunit. Mutations and CAG repeat expansions in the gene encoding a second Ca²⁺ channel 1-subunit isoform, termed isoform 4, have been identified in familial hemiplegic migraine 1 (OMIM 141500), episodic ataxia type 2 (OMIM 108500), and spinocerebellar ataxia 6 (OMIM 183086). Mutations in the genes that encode potassium channels, perhaps the most diverse class of ion channels, give rise to a number of disorders; for example, variants of the long QT syndrome, LQT1 (OMIM 192500) and LQT2 (OMIM 152427) are caused by mutations in a voltage-gated K⁺ channel, KVLQT1, and in HERG (human ether-a-go-go related gene), a Ca²⁺-modulated K⁺ channel, respectively. Mutations in the KCNJ1 gene encoding ROMK1, the first discovered potassium inward rectifier (1) give rise to the antenatal variant of Bartter’s syndrome with renal tubular hypokalemic alkalosis (OMIM 601678). Finally, mutations in the gene encoding a water channel, aquaporin 2, cause the renal form of diabetes insipidus (OMIM 125800). Ion channelopathies have been reviewed recently (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17). The reader is referred to the extensive bibliographies cited in the Online Mendelian Inheritance in Man database (http://www3.ncbi.nlm.nih.gov/Omim/) for references to each of the channelopathies cited above and for current information on others.

Our objective is to review the molecular biology of ATP-sensitive K⁺ channels, or K_ATP channels, which are present in pancreatic ß-cells in the islets of Langerhans where they play a key role in stimulus-secretion coupling by providing a link between changes in metabolism and membrane electrical activity. Evidence suggests this channel is regulated by changes in adenine nucleotide levels, an increase in the ATP/ADP ratio, resulting from changes in glucose metabolism, and is the target for a class of drugs called sulfonylureas used in the treatment of non-insulin-dependent diabetes mellitus. Closure of K_ATP channels as a result of increased glucose metabolism or by sulfonylureas leads to the release of insulin. Nearly 50 mutations in either of the two subunits that make up the ß-cell K_ATP channel, namely, the high-affinity sulfonylurea receptor, SUR1, and the K⁺ inward rectifier, K_IR6.2, are responsible for a channelopathy, the recessive genetic form of persistent hyperinsulinemic hypoglycemia of infancy (PHHI), characterized by the uncoupling of glucose metabolism from ß-cell electrical activity (see Refs. 18, 19, 20, 21 for reviews). Investigation of this recessive form of PHHI has led to the identification of at least two other causes of dominant forms of persistent neonatal hypoglycemia and should provide insight into the regulation of the insulin-secretory pathway in both normal and in diabetic individuals.

The objective of this article is to review developments in the molecular biology of ATP-sensitive potassium channels. We will focus mainly on recent studies of the molecular biology of the "classic" ß-cell type K_ATP channels composed of the high-affinity sulfonylurea receptor, SUR1 (OMIM 600509), and K_IR6.2, a potassium inward rectifier subunit (OMIM 600937), but will draw comparison with the striated and smooth muscle type channels where appropriate. Over the last several years, the subunits of the ß-cell K_ATP channel have been cloned, expressed, and reconstituted into functional nucleotide-sensitive channels and have been used to investigate their overall structure and regulation by nucleotides, potassium channel openers, and sulfonylureas. The availability of cDNAs for SUR1 (22) allowed the isolation of the low-affinity sulfonylurea receptors, SUR2A and SUR2B (23, 24). In humans, the mRNAs encoding the SUR2A and SUR2B receptors originate from a single SUR2 gene by differential splicing of the last exon (19). Two K_IR6.x isoforms have been cloned, K_IR6.1 (uKATP-1) (25) and K_IR6.2 (BIR) (26). Interestingly, the SUR and K_IR genes are paired on human chromosomes, with SUR1 and K_IR6.2 adjacent on chromosome 11, and SUR2 and K_IR6.1 near each other on chromosome 12. When the low-affinity receptors are reconstituted with K_IR6.1 or K_IR6.2, they produce K_ATP channels with distinctive pharmacologies which, along with their tissue distribution, suggest they make up the cardiac-skeletal and vascular smooth muscle K_ATP channels (23, 24, 27, 28, 29).

	II. How Are KATP Channels Defined?

Top Abstract I. Introduction II. How Are KATP... III. How Do KATP... IV. KATP Channel Subunits V. Reconstitution of KATP... VI. KATP Channel Structure VII. Regulation of KATP... VIII. Human SUR1 and... IX. KATP Channels and... X. Linking PHHI to... XI. Other Issues XII. KATP and NIDDM XIII. The Leptin Connection XIV. Summary and Conclusions References

 
The application of ATP to the intracellular face of an excised patch was initially used to identify K_ATP channels. These observations were made by Cook and Hales (30) in the United States, by Noma (31) in Japan, and by Trube and Hescheler (32) in Germany. The first reference to ATP-sensitive K⁺ channels that we can uncover is an abstract by Trube and Hescheler (33) outlining the properties of the cardiac K_ATP channel from isolated patches of cardiac cell membranes. Historically, a large number of studies defined the electrophysiological properties and pharmacology of these channels, including their single-channel current properties, pattern of regulation by nucleotides, and response to channel openers and blockers. It is now possible to begin to define these channels in terms of their molecular composition. K_ATP channels are heteromultimers of two types of subunits, inward rectifiers, K_IR6.x, and sulfonylurea receptors, SURs, members of the ABC superfamily. The available evidence, discussed in more detail below, indicates functional channels are assembled as tetramers (SUR/K_IR)₄. Table 1 relates the known subunits to channel types identified by tissue type. The list is provisional and is not meant to be exhaustive, as it is not known whether all the subunit isoforms have been identified. This classification is based largely on pharmacological criteria, and there remains a strong need for tissue and cellular localization work to confirm this classification and to look for overlapping expression of subunit types.

View larger version (32K): [in this window] [in a new window]   Figure 1. Electrical activity of human SUR1/KIR6.2 ATP-sensitive K+ channels. A, Recordings from excised membrane patches of COSm6 cells transfected with human SUR1 and human KIR6.2 illustrate channel activity during perfusion of the cytoplasmic face of the patch with the indicated reagents. Panels B and C are traces plotted at shorter time intervals chosen to show open and closed states of a single channel within a burst of activity. D, A current vs. voltage plot to show the inward rectification (reduced current at more positive potentials) of KATP channels and the effect of an added polyamine. The single-channel conductance is approximately 75 pS. E, The ATP inhibition curve is the average of five experiments, and the bars are the ±SE. The activity in the absence of ATP was taken as 100%; the IC50 was 14.7 µM with a Hill coefficient of 1.1.

View larger version (20K): [in this window] [in a new window]   Figure 2. Chemical structures of compounds that activate and block KATP channels.

	III. How Do KATP Channels Affect the Membrane Potential of Pancreatic ß-Cells?

Top Abstract I. Introduction II. How Are KATP... III. How Do KATP... IV. KATP Channel Subunits V. Reconstitution of KATP... VI. KATP Channel Structure VII. Regulation of KATP... VIII. Human SUR1 and... IX. KATP Channels and... X. Linking PHHI to... XI. Other Issues XII. KATP and NIDDM XIII. The Leptin Connection XIV. Summary and Conclusions References

 
The function of K_ATP channels is best understood in pancreatic ß-cells, the membrane potential of which is responsive to external glucose concentration. ß-Cells show a remarkably complex electrical bursting behavior (see Fig. 3) in response to an increase in glucose level. It is generally agreed that the oscillatory electrical activity leads to oscillatory changes in [Ca²⁺]_i, which drive pulsatile insulin release in isolated islets (see for example Ref. 52). Various suggestions on how K_ATP channels may participate in this response have been put forward; these range from maintaining the resting membrane potential when glucose is low (53) to playing a major role in control of bursting when glucose is high (54, 55) or terminating individual bursts (56). The origin and control of the electrical bursting activity in pancreatic ß-cells remain controversial, and a detailed discussion of this area is beyond the scope of this review. The reader is directed to recent articles for detailed discussions (57, 58, 59, 60, 61, 62, 63). Our intention is to provide a framework for understanding the molecular biology and structure of K_ATP channels, how they affect membrane potential and thus how they can affect bursting and insulin release, and finally, how their absence leads to familial hyperinsulinism.

View larger version (16K): [in this window] [in a new window]   Figure 3. Illustration of electrical activity and [Ca2+]i in a pancreatic ß-cell. Islets, loaded with fura-2, were stimulated with 12 mM glucose, and simultaneous recordings were made from the same ß-cell using a microelectrode and a microfluorometer (modified from Fig. 1 in Ref. 60). The earliest change, marked phase 0, is a gradual depolarization of the membrane and a decrease in [Ca2+]i. An abrupt initiation of a sustained period of depolarization with superimposed Ca2+-dependent action potentials or spikes marks phase 1. [Ca2+]i rises abruptly during phase 1, the result of Ca2+ mobilization from intracellular stores and from Ca2+ influx through voltage-dependent Ca2+ channels. Phase 1 can be correlated temporally with first-phase insulin release. Phase 2 is marked by the initiation of steady-state oscillations in both membrane potential and [Ca2+]i, commonly called "bursting." The frequency of bursting is dependent on the glucose level (see for example Ref. 88). F340/F380 is the fluorescence ratio for fura-2 used as a measure of [Ca2+]i. [Adapted with permission from I. D. Dukes et al.: Curr Opin Endocrinol Diab 4:262–271, 1997 (60 ). © Lippincott Williams and Wilkins.]

As illustrated in Fig. 3, V_M in the ß-cell is regulated by glucose metabolism. When the external glucose concentration is less than 3 mM, V_M is near -65 to -70 mV. Increasing the external glucose causes gradual depolarization to a new steady-state level. Further increases cause the membrane to depolarize to a threshold potential, near -50 mV, where electrical activity, due in part to Ca²⁺ currents through voltage-dependent Ca²⁺ channels, is initiated. Several types of pumps and ion channels have been identified in ß-cells, which could or have been suggested to contribute to this observed electrical bursting behavior, including the following.

1. Na⁺/K⁺ ATPase. The asymmetric distribution of sodium and potassium ions is maintained by the ß-cell Na⁺/K⁺ ATPase, which has been identified and subjected to limited study (65, 66). The activity of this enzyme is generally considered insensitive to K_ATP channel blockers and openers, although high concentrations of glibenclamide have been reported to be inhibitory (67). Inhibition of the Na⁺/K⁺ ATPase with ouabain results in membrane depolarization and release of insulin. In a recent report, Ding et al. (56) used manipulations of the activity of the Na⁺/K⁺ ATPase to change the ATP/ADP ratio and affect K_ATP channel activity and suggest that a fall in [ATP]_i (and presumably an increase in [ADP]_i), terminates a burst of electrical activity.

2. Na⁺ channels. Early studies on the effects of veratridine and tetrodotoxin suggested that Na⁺ channels were functionally important for insulin release from pancreatic islets (68, 69), although Donatsch et al. (70) found tetrodotoxin did not affect insulin release from mouse islets. Several reports have described a transient, voltage-dependent, inward Na⁺ current that is blocked by tetrodotoxin (71, 72). The canine channel shows steep activation and inactivation between -50 and -40 mV (73, 74). Philipson et al. (75) identified Na⁺ channel subunit cDNAs in canine, human, and rodent islets, and in hamster and mouse insulinoma cell lines, which are most closely related to the rat brain III 1 isoform of Na⁺ channel subunits. This conductance could play a role in the initial depolarization, but its role is usually discounted in rodent ß-cells because the voltage dependence of inactivation of the rodent channel suggests it will be largely inactive (71, 76). Na⁺ channels may contribute to the action potential in canine and human ß-cells (73, 74), but their role is generally believed to be minimal.

3. Chloride channels. Several reports have suggested chloride ions can affect insulin release and ß-cell electrical activity (77, 78). The mechanism(s) of these effects are not known, and it is unclear whether or not chloride ions are passively distributed across the ß-cell membrane. Kinard and Satin (79) have identified an ATP-sensitive chloride channel in ß-cells, termed I_Cl,islet, which is activated by cAMP, glibenclamide (1–10 µM), and cell swelling. Decreasing [ATP]_i reduces the amplitude of the current; thus, Kinard and Satin suggest this channel may be under metabolic control. I_Cl,islet mediates a large inward current that would depolarize the ß-cell membrane. A second, Ca²⁺-dependent Cl^- conductance has been described recently in the ßTC-3 ß-cell line (80), but its physiological significance is uncertain.

4. Ca²⁺ channels. The importance of Ca²⁺ in insulin release has been reviewed by many others (81, 82, 83, 84, 85, 86, 87, 88). Insulin release is attenuated in Ca²⁺-deficient media and by the action of Ca²⁺ channel blockers. The early reports describing Ca²⁺ currents in ß-cells (89, 90, 91) were followed by others identifying at least two types of Ca²⁺ channels in ß-cells, which could be distinguished by their kinetics and pharmacology (71, 72, 92, 93, 94, 95, 96). The larger conductance channel has the properties of a fast deactivating, dihydropyridine-sensitive L-type Ca²⁺ channel with an activation threshold near -30 mV. The smaller conductance is similar to the T-type Ca²⁺ channel, is slowly deactivated, and has a lower activation threshold, near -50 mV. The T-type channel is not sensitive to dihydropyridines, but is inhibited by NiCl₂, a selective inhibitor of other T-type calcium channels (96). It has been suggested that the slow T-type channel could be responsible for the slow Ca²⁺ waves, while the L-type channels contribute the spikes. The evidence for and against this idea has been reviewed by Satin and Smolen (58). The possible involvement of T-type channels in control of insulin secretion has become more accessible with the cloning of these channels (97, 98) and the finding of T-type channel mRNAs in the pancreas. In addition to the voltage-dependent Ca²⁺ channels, a glucose-activated calcium channel has been reported in ß-cells (99).

5. Ca²⁺-release-activated nonselective cation channels (I_CRAN). Several lines of evidence indicate the importance of a nonselective cation channel, which is activated by depletion of internal Ca²⁺ stores, in the regulation of ß-cell electrical bursting. This current has been termed I_CRAN (60, 100, 101) or I_CRAC for Ca²⁺ release-activated currents (102, 103, 104, 105). The use of the term I_CRAC in this context appears to be somewhat unfortunate and this channel, which has a conductance in the 20–40 picosiemens (pS) range, should not be confused with the low-conductance Ca²⁺ release-activated Ca²⁺ channel, I_CRAC, described previously (106, 107, 108).

Dukes et al. (60) have reviewed the evidence for the role of I_CRAN in bursting. This model underscores the importance of Ca²⁺ mobilization from the endoplasmic reticulum through activation of the inositol triphosphate (IP₃) receptor by IP₃ which is generated by membrane depolarization. Emptying of internal Ca²⁺ stores signals I_CRAN to open. Under physiological conditions Na⁺ is the primary current carrier; thus, as described above, the increased Na⁺ permeability depolarizes the ß-cell causing L-type Ca²⁺ channels to open, which provides Ca²⁺ to help refill internal calcium stores. Replenishment reduces the signal to and activity of I_CRAN, allowing the cell to repolarize and continue cycling. Since the extent of repolarization will be determined by the K⁺ permeability and will determine whether bursting continues, this is one potential control point for K_ATP channels. Coupling the control of I_CRAN to emptying of internal Ca²⁺ stores is able to explain the effects of compounds such as maitotoxin (109, 110, 111), which activates I_CRAN, and thapsigargin, an inhibitor of the endoplasmic reticulum Ca²⁺ pump [sarcoplasmic or endoplasmic reticulum Ca-ATPase (SERCA)] (112, 113), on ß-cell membrane potential and insulin release (105, 114, 115). How the filling level of the internal Ca²⁺ store is determined and the nature of the chemical signal that passes to I_CRAN remains to be elucidated.

6. Potassium channels. In addition to K_ATP channels, other K⁺ channels have been identified in pancreatic ß-cells. This area has been reviewed recently by Dukes and Philipson (87) and need not be repeated beyond indicating that delayed rectifier channels (Kv1.x), Ca²⁺-activated K⁺ channels (maxi-K), an -adrenoreceptor-activated K⁺ channel, and G protein-gated K⁺ channels (K_IR3.1, 3.2 and 3.4) have all been identified in ß-cells either by electrical recording or molecular biological methods. The Ca²⁺-activated K⁺ channel was proposed to play a role in bursting, but inhibitors of this channel have little effect on electrical activity.

What role do K_ATP channels play in bursting? There is general agreement that K_ATP channels are the predominant conductance in a ß-cell in low glucose, and that the initial depolarization, which starts a train of bursts, results from an increase in glucose metabolism, which reduces their activity. The action of sulfonylureas is readily understood within this context. Sulfonylureas close K_ATP channels, in effect mimicking the signal from glucose metabolism, thus depolarizing the ß-cell membrane. Titration of K_ATP channel activity with tolbutamide will lead to bursting at intermediate glucose concentrations (116, 117). The action of a KCO is also readily understood. The application of diazoxide inhibits insulin release by hyperpolarizing ß-cells by opening more K_ATP channels, thus shifting V_M more toward E_K. The question of how K_ATP channels participate in the actual dynamics of bursting is controversial. Using whole-cell recording, Smith et al. (118) were unable to see changes in conductance in K_ATP channels during bursting, indicating they are not involved in the dynamics of bursting. Larsson et al. (119) have reported the reverse: that the conductance of K_ATP channels oscillates during slow bursting in single mouse ß-cells and in small clusters of ß-cells. Rosario et al. (120) have demonstrated bursting in ß-cells, incubated with high extracellular Ca²⁺, when K_ATP channels were blocked by sulfonylureas. These results indicate K_ATP channels are not essential for bursting to proceed, but that residual channel activity may play a role in normal repolarization. Finally, if K_ATP channels are stimulated by diazoxide during an episode of bursting, the increase in K⁺ permeability hyperpolarizes the ß-cell membrane and thereby terminates bursting. These results have led to the idea that there are both K_ATP channel-dependent and independent mechanisms for controlling insulin secretion (see for example Refs. 121, 122, 123).

The physiological ligand(s) or mechanism(s) controlling K_ATP channels remains somewhat controversial. In excised, inside-out patches, the application of ATP results in channels closing (Fig. 1), which led to the early suggestion that changes in the intracellular concentration of ATP might regulate ß-cell K_ATP channels (30, 124). However, the IC₅₀ for inhibition of channel activity was determined to be in the 10–50 µM range, well below the millimolar concentrations estimated for intracellular ATP levels (125, 126, 127); when millimolar concentrations of ATP were applied to the intracellular face of excised patches, K_ATP channel activity was strongly inhibited. Subsequent studies showed that MgADP could open K_ATP channels inhibited by ATP, implying that the ADP/ATP ratio was the important variable (43, 44, 128, 129, 130). As described below, studies on channels reconstituted from a mutant SUR1 identified in a PHHI patient and wild-type K_IR6.2 are consistent with this idea and suggest further that fluctuations in the intracellular ADP concentration that occur as a result of glucose metabolism (126, 127, 131) are a critical regulator in ß-cells.

	IV. KATP Channel Subunits

Top Abstract I. Introduction II. How Are KATP... III. How Do KATP... IV. KATP Channel Subunits V. Reconstitution of KATP... VI. KATP Channel Structure VII. Regulation of KATP... VIII. Human SUR1 and... IX. KATP Channels and... X. Linking PHHI to... XI. Other Issues XII. KATP and NIDDM XIII. The Leptin Connection XIV. Summary and Conclusions References

 
A. The K_IR family of inwardly rectifying K⁺ channels
The observations (30, 44, 124, 132, 133, 134) that ATP-sensitive K⁺ channels are both potassium selective and inwardly rectifying suggested they would have some relationship to the inwardly rectifying potassium channel superfamily, the definition of which began with the cloning of three distinct inwardly rectifying K⁺ channels: K_IR1.1 (ROMK1), a weak inward rectifier found in rat kidney (1), K_IR2.1 (IRK1), a strong inward rectifier cloned from a macrophage cell line (135), and K_IR3.1 (KGA, GIRK1), a G protein-gated, strong inward rectifier isolated from rat heart (136, 137). The descriptive terms "weak" and "strong" in this context are indicators of the amount of current the channels can pass in the outward direction when the membrane potential is greater than E_K; weak rectifiers conduct more current. These proteins are all smaller than the voltage-gated K⁺ channels and, based on hydropathy plots, were predicted to have two transmembrane domains (TMDs) or helices, termed M1 and M2. These segments flanked a sequence with a high degree of similarity to the P (pore) or H5 loop first identified in the voltage-gated K⁺ channel (K_V) family (138, 139) and shown to be part of the potassium selectivity "filter." Insight into how the pore is probably formed has been provided by the determination of the structure of the pore of the KcsA channel (34, 35) [1BL8 in the Protein Data Bank (140, 141)] which is depicted in Fig. 4. Although the KcsA channel is a pH-regulated, bacterial K⁺ channel, the general similarity between the pore regions of the Kv channels, the inward rectifiers, and KcsA, all of which share an architecture with two transmembrane helices flanking a GYG or GFG sequence responsible for the K⁺ selectivity, indicates their structures will be similar. As expected from earlier studies, the M2, or inner pore helix, actually forms the permeation pathway with the signature GYG sequence positioned to act as a selectivity filter. The structure and studies on the dynamics of KcsA suggest the actual gate is formed by amino acids at the cytoplasmic ends of the M2 helices. Lateral or twisting motions of these helices could control access or gating of internal K⁺ ions into the water- filled vestibule immediately below the selectivity filter. The availability of a general structural model, even one missing the cytoplasmic domains that presumably regulate motion of the M2 helices, should allow real progress toward understanding gating mechanisms.

View larger version (64K): [in this window] [in a new window]   Figure 4. Three-dimensional structure of the Streptomyces lividans K+ channel. The top view looks down into the conduction pathway through the selectivity filter assumed to be composed of the backbone carbonyl oxygen atoms of the gly-phe-gly motif found in other potassium channels. The side view conveys the overall structure, including the vestibule below the selectivity filter and the convergence of the four M2 helices, which form the conduction path at the cytoplasmic face of the channel, to form a gate which can open and close (35 ). The figure was generated using the coordinates, 1BL8, obtained from the Protein Data Bank (140 141 ) using RasMol (373 ). [Derived from D. A. Doyle et al.: Science 280:69–77, 1998 (34 ).]

1. Cloning of K_IR6.x cDNAs. Using a fragment of K_IR3.1 (GIRK1) cDNA as a probe, Inagaki et al. (144) cloned K_IR6.1 (ruK_ATP-1) from a rat library. The tissue distribution of K_IR6.1 was broad; hence the designation "u" K_ATP-1 for ubiquitous. Expression of K_IR6.1 in HEK293 cells suggested it formed a K⁺ channel whose activity was blocked by 1 mM ATP, and activated by diazoxide, but was not inhibited by sulfonylureas. Subsequent expression of K_IR6.1 in other cell types has failed to generate novel K⁺ channel activity with similar properties (145, 146), and it is unclear whether the original observation was due to coupling of K_IR6.1 to an unidentified SUR-like protein in HEK293 cells (21), to mischaracterization of an endogenous channel, or to having a sufficiently high level of expression to overcome a K_IR retention signal. This report further added to the confusion surrounding the cloning and identification of K_ATP channels since K_IR1.1 (1), K_IR3.4 (147), and K_IR6.1 (144) were all reported initially to be ATP-sensitive potassium channels.

K_IR6.1 was not detected in ß-cell lines, and Inagaki et al. (144) used the K_IR6.1 cDNA to clone the cDNA for K_IR6.2 (originally designated BIR for ß-cell inward rectifier). Subsequent work showed that K_IR6.2 mRNA was relatively abundant in pancreatic islets and ß-cell lines and was expressed in brain, heart, and skeletal muscle (see Ref. 21 for a review). The K_IR6.2 gene was found to contain no introns and is therefore relatively easy to obtain using the PCR.

K_IR6.1 and K_IR6.2 have significant blocks of identical sequence. Seventy percent, 276 of the 390 residues in human K_IR6.2, are identical to those of K_IR6.1 after conservative alignment (Fig. 5). We have marked the proposed transmembrane helices and the P loop by comparison with the KcsA structure and have marked the approximate positions of the PHHI mutations described below. K_IR6.1 has several intriguing features, including a triple repeat at its C-terminal end, S(M/I)RRNN.

View larger version (59K): [in this window] [in a new window]   Figure 5. Schematic representation of an inward rectifier potassium channel. The diagram is based on the Streptomyces lividans K+ channel described by Doyle et al. (34 ) and on data described in the text. M1 and M2 identify the two TMDs; the region determining the K+ selectivity contains the gly-phe-gly motif. The asterisks mark the positions of mutations in KIR6.2 that have been identified in patients with PHHI. The dotted circles identify segments of the KATP channel referred to in the text. The alignment compares the sequences of human KIR6.1 with KIR6.2. Identical residues are shaded. The M1 and M2 segments are boxed. Note the repeated sequence in the C terminus of KIR6.1 marked by the bar.

A significant advance in the biochemistry of these receptors came with the realization that [³H]glibenclamide could function as a photoaffinity probe and would label a polypeptide with an apparent molecular mass of 140 kDa in rat ß-cell tumor membranes (151). A ¹²⁵I-labeled derivative of glibenclamide, [¹²⁵I]iodoglibenclamide (152) (Fig. 2), identified a similar protein in membranes isolated from HIT T15 cells and had the advantage of higher specific activity, which made studies of the receptor and its purification feasible. Nelson et al. (154) investigated the photolabeling of membrane proteins by this reagent and showed that the 140-kDa protein accounted for most of its high-affinity binding activity, K_D 5 nM, in HIT T15 cell membranes. Rajan et al. (134) and Ronner et al. (156) used [¹²⁵I]iodoglibenclamide and [³H]gli-benclamide, respectively, to identify the high-affinity receptor in TC-6 cells, a glucagon-secreting cell line. Both groups characterized the sulfonylurea-sensitive K_ATP channels in these cells by rubidium efflux and single-channel recording and concluded they were equivalent to the ß-cell channel.

Schwanstecher et al. (162) synthesized a 4-azidosalicyloyl analog of glibenclamide, ¹²⁵I-iodo-azidoglibenclamide (Fig. 2), and demonstrated that this derivative would photolabel a 38- to 40-kDa protein in addition to SUR1 (163). Competition binding experiments indicated that both the novel 38-to 40-kDa species and SUR1 were labeled with ¹²⁵I-iodo-azidoglibenclamide with the same apparent K_D. As described below, the 38- to 40-kDa species is K_IR6.2.

2. SUR1 is differentially glycosylated. Labeling of TC-6 cells with [¹²⁵I]iodoglibenclamide showed two labeled species with estimated molecular masses of 140 kDa and 150–170 kDa, the latter being a diffuse band (134). The higher molecular mass receptor accounted for approximately half of the high-affinity labeling in TC-6 cells. Competition binding studies demonstrated that the two receptors had the same affinity for iodoglibenclamide, K_D 3.5 nM. Rajan et al. (134) suggested the 150- to 170-kDa receptor resulted from differential glycosylation, but at the time the possibility of an additional gene product could not be eliminated. Nelson et al. (164) demonstrated that the 140 and 150- to 170-kDa receptors were present in several other cell types. In RINm5F and TC-6 cells the two glycosylated species are present in about equal amounts, while in HIT cells the predominant species was the 140-kDa receptor, with the 150- to 170-kDa species accounting for less than 10% of the high-affinity binding activity. Ozanne et al. (165) used [³H]glibenclamide to label receptors in rat pancreatic islets, insulinomas, and CRI-G1 cells and were able to identify both forms of the receptor. The 150- to 170-kDa species was the major form in islets and insulinomas, while the 140-kDa species was predominant in CRI-G1 cells.

Biochemical studies indicate the mass difference between the two forms of the receptor is the result of differential glycosylation (164). Photolabeling of membranes isolated from RINm5F cells grown in tunicamycin, which blocks N-linked glycosylation, produced a single photolabeled band with an estimated mass of 137 kDa, distinct from the 140-kDa species. A 137-kDa species was also generated by digestion with endoglycosidases. The 140 and 150- to 170-kDa receptors could be separated using concanavalin A and wheat germ agglutinin, the 140 kDa receptors binding to concanavalin A-agarose, while the 150- to 170-kDa receptors bound to wheat germ agglutinin-agarose. The results indicate that the 140-kDa receptors are "core" glycosylated with mannose-containing glycosyl groups that bind to concanavalin A, while the 150- to 170-kDa receptors are "complex" glycosylated with terminal sialic acid residues that bind to wheat germ agglutinin (164). Since core glycosylation takes place in the endoplasmic reticulum after protein synthesis, while trimming of the mannose chains and addition of sialic acid groups takes place in the medial Golgi, the data indicate that the 150- to 170-kDa species is the "mature" form of the receptor and is either in, or in transit to, the plasma membrane. This is consistent with the results of Ozanne et al. (165) showing that the "immature" 140-kDa receptors in insulinomas were found in internal membranes, whereas the 150- to 170-kDa receptors were in granules and plasma membranes. The results raise questions about how glycosylation of SUR1 is regulated, whether the core glycosylated receptors on internal membranes might have a function apart from K_ATP channels, and whether the subunits can traffic independently to the plasma membrane. For example, Eliasson et al. (166) have suggested that sulfonylureas can stimulate insulin secretion independently of their blockage of K_ATP channels, perhaps acting through SUR1 in granule membranes.

The available sequence and site-directed mutagenesis has confirmed there are two N-glycosylation sites in SUR1 at residues 10 and 1050. The N10Q or N1050Q mutations partially eliminate glycosylation, while the double mutant is not glycosylated. All three of the glycosylation mutants generate channels when expressed with K_IR6.2, although the number of channels, estimated by ⁸⁶Rb⁺ efflux, is reduced (J. P. Clement IV and J. Bryan, unpublished data). Interestingly, the glycosylation patterns of SUR1 appear to differ between endocrine and neuronal tissues as well as - and ß-cell lines (164). The SUR1 in brain, for example, appears to be only the highly glycosylated form (162).

3. Linking the sulfonylurea receptor with K_ATP channel activity. The link between sulfonylurea receptors and K_ATP channels was the subject of some early controversy. The simple view that receptor and channel might be a single entity (167) was questioned by Khan et al. (168), who reported the dissociation of K_ATP channel activity and sulfonylurea receptors in a rat insulin-secreting clonal cell line, CRI-D11. Aguilar-Bryan et al. (169) produced evidence that receptor and channel activity were lost in parallel during serial passage of HIT cells at approximately the same passage number at which glucose sensitivity for insulin release was lost (170). These studies have not been reexamined using the molecular probes now available, and it is unclear, for example, whether the CRI-D11 cells actually have active K_IR6.2 channels, without receptors, or whether high passage HIT cells have lost K_IR6.2 in addition to SUR1.

4. Receptor purification. Hamster SUR1, prelabeled with [¹²⁵I]iodoglibenclamide, was purified from HIT cell membranes using chromatographic and electrophoretic methods (22, 164). The general strategy was to prelabel the receptor with [¹²⁵I]glibenclamide, then purify the iodinated protein. The end point was obtaining sufficient pure receptor for N-terminal peptide sequencing. The purification has been described in detail (164, 171) and will not be discussed further. The availability of cloned cDNAs has allowed introduction of specific markers, for example hexa-histidine tags, to facilitate purification of expressed recombinant receptors (146, 172) and K_IR6.x subunits.

5. Cloning strategy. The N-terminal protein sequence was used to design oligonucleotide primers for amplification of a small segment of the SUR1 cDNA by the PCR (22, 164). The resulting cDNA fragment was used to screen hamster, rat, mouse, and human pancreatic cDNA libraries to obtain full-length clones that encode homologous proteins with molecular masses of approximately 176 kDa (1582 amino acids for the rodent receptors and 1581 or 1582 amino acids for the human receptor). The cDNA and gene sequences for SUR1 and SUR2 are readily available through the National Library of Medicine (http://www.ncbi.nlm.nih.gov/Web/Search/index.html).

6. SUR1 is a member of the ATP-binding cassette superfamily. The rodent and human SUR1 sequences are quite similar with absolute identities of >90%. Blast searches (173) of the Genbank database show similarities between SUR’s and a large number of ABC proteins including multidrug resistance proteins (174), multidrug resistance associated proteins (MRPs) (175), CFTRs (176), canalicular multispecific organic anion transporter (177, 178), and the cadmium-binding protein YCF1 (179). Dendrograms and the overall topology of SURs indicate they can be placed within the MRP subfamily of ABC proteins (180) with two potential nucleotide-binding folds (NBFs), multiple TMDs, and a very hydrophobic segment at their N termini. The greatest sequence similarities are in the NBFs that have the Walker A (-GlyXXGlyXGlyLysSer/Thr-, where X is any amino acid) and B (-YYYYAsp-, where Y is a hydrophobic amino acid) consensus motifs (181) and a conserved -LeuSerGlyGlyGln- sequence in the segment linking the two Walker motifs. NBF1 matches this motif exactly, while NBF2 has a somewhat degenerate sequence, -PheSerGlnGlyGln- (SUR1), -PheSerValGlyGln- (SUR2). Manavalan et al. (182) have noted the similarity of the NBF sequence between CFTR and G proteins. X-ray crystal structures for the three major classes of GTPases, the small p21^ras-like proteins, the heterotrimeric G proteins, and EF-Tu, indicate that the Walker A motif interacts with Mg²⁺ and with the oxygen atoms of the - and ß- phosphates (see Ref. 183 for a review). The Gln in the -LeuSerGlyGlyGln- linker sequence has been proposed to act as a general base during nucleotide hydrolysis, but this has been difficult to establish (184). The lack of a structure for the NBFs of the ABC cassette proteins has hindered progress. The recent preliminary report on the determination of the structure of an NBF from the ribose transporter by Armstrong et al. (185) should allow major advances in this area. A major question with regard to the sulfonylurea receptors is whether they can hydrolyze ATP, and, if so, how this hydrolysis is coupled to the regulation of channel activity. The purification and ATPase activity of the closely related MRP1 protein have been described (186). Interestingly, the ATPase activity of this ABC protein is stimulated several fold by low concentrations of nucleoside diphosphates (187).

7. Proposed membrane topology for SUR1. Aguilar-Bryan et al. (22) predicted a topology for SUR1 based on plots of hydrophobicity and hydrophobic moments using the analysis of Eisenberg and colleagues (188). The constraints on the proposed topology were that the N terminus, glycosylated at N10 (22), was extracellular and the two NBFs were intracellular. The resulting model consisted of nine TMDs followed by a nucleotide-binding fold, NBF1, a second transmembrane region with four spanning helices followed by a second nucleotide-binding fold, NBF2 (i.e., 9 TMD–NBF1–4 TMD–NBF2). This topology resembled the model originally put forward for MRP1 (175). Tusnády et al. (180) have revised the model of the MRP subfamily using multisequence alignments of recently cloned ABC proteins. Their model, 11 TMD–NBF1–6 TMD–NBF2, emphasizes the existence within the MRP subfamily proteins of an multidrug resistance protein-like core consisting of 6 TMD–NBF1–6 TMD–NBF2. This core is preceded, in the MRP subfamily, by a highly hydrophobic N-terminal extension with 5 or more TMDs. Work is needed to establish the topology experimentally and to determine what function the additional TMDs might have. We have discussed the topology of the receptors in more detail elsewhere (19).

8. Expression of SUR1 cDNAs. Transfection of SUR1 cDNAs into COSm6 cells lacking endogenous sulfonylurea receptors led to the expression of proteins that could be photolabeled with [¹²⁵I]glibenclamide and had an apparent mobility on SDS polyacrylamide gels equivalent to the native receptor (22). The discrepancy in mass between 176 kDa predicted from the cDNA sequence and 140 kDa estimated by gel electrophoresis is artifactual. Purification of receptors tagged with hexa-histidine at either the N- or C termini yield the same size protein. Membranes isolated from COS cells expressing the recombinant hamster and rat receptors contained high-affinity [¹²⁵I]glibenclamide-binding activity with K_Ds of 10 and 2 nM, respectively (22). These values were in good agreement with a value of 5 nM determined for the native hamster receptor in isolated HIT cell membranes (19, 154, 169). Efforts to show that COS cells expressing SUR1 alone had novel K⁺ channel activity were unsuccessful, suggesting other components were required to form a K_ATP channel. A puzzling result, from these early studies, was that only the 140-kDa core glycosylated protein was expressed, while the 150- to 170-kDa receptors were not detected.

	V. Reconstitution of KATP Channel Activity from SUR1 and KIR6.2

Top Abstract I. Introduction II. How Are KATP... III. How Do KATP... IV. KATP Channel Subunits V. Reconstitution of KATP... VI. KATP Channel Structure VII. Regulation of KATP... VIII. Human SUR1 and... IX. KATP Channels and... X. Linking PHHI to... XI. Other Issues XII. KATP and NIDDM XIII. The Leptin Connection XIV. Summary and Conclusions References

 
Coexpression of SUR1 and K_IR6.2 in COS cells and in Xenopus oocytes produced a novel K⁺ conductance that was characterized by ⁸⁶Rb⁺ efflux and single-channel recording (26). Metabolic poisoning with 2-deoxyglucose and oligomycin revealed a novel ⁸⁶Rb⁺ efflux pathway not detected in cells transfected with ß-galactosidase, or with either SUR1 or K_IR6.2 alone. Efflux was blocked by tolbutamide and gli-benclamide and was activated by diazoxide, but was poorly activated by cromakalim or pinacidil (23). Single-channel recordings characterized a moderate, inwardly rectifying potassium-selective channel with a conductance of about 65 pS in 140 mM KCl at -60 mV (26). The extent of inward rectification was dependent upon the presence of Mg²⁺ or polyamines on the intracellular side of the channel (19, 146). Channel activity was half-maximally inhibited by ATP at approximately 10 µM (26). Later measurements indicated the recombinant channels were activated by MgADP in the presence of ATP (189, 190). Table 2 provides a comparison of the properties of recombinant and native K_ATP channels. The recombinant channels appear to mimic all of the basic electrophysiological and pharmacological properties of native ß-cell ATP-sensitive potassium channels. These results have been confirmed by studies in oocytes (145) and HEK cells (191). Additional work will be required to determine whether and by what means the recombinant K_ATP channels are coupled to the receptors for the various peptides, such as galanin, reported to affect their activity (Ref. 192 but see Ref. 193), somatostatin (194), GLP-1 (195), CGRP (196), and leptin (197, 198, 199, 200, 201).

A. The question of "promiscuous coupling" of SUR1 with other inward rectifiers
The full range of inward rectifiers that SUR1 will partner with has not been established. Ämmälä et al. (202) showed that K_IR6.1 and SUR1 would form K_ATP channels, but these were not characterized. These authors also argued for greater promiscuity and pairing of the receptor with K_IR1.1 (ROMK1) and K_IR3.4 (rcKATP-1/CIR/GIRK4). Clement et al. (146) have confirmed that K_IR6.1 and SUR1 produce metabolically activated K_ATP channels, while coexpressing SUR1, and K_IR1.1 (ROMK1) or K_IR3.4 (CIR, rcKATP) failed to show association or K_ATP channel activity. Using Xenopus oocytes, Gribble et al. (145) were unable to reproduce the original observations on coupling of SUR1 to ROMK1 and showed that K_IR2.1 also failed to couple to SUR1. We infer that, although all the potential inward rectifier candidate subunits have not been screened, the degree of promiscuity involved in K_ATP channel formation will not be large. Wellman et al. (203) have reached a similar conclusion using coronary arterial myocytes. Interestingly, the reports on the SUR1/K_IR6.1 channel have all been with intact cells, since observations on excised patches have been hampered by rapid channel rundown (145).

	VI. KATP Channel Structure

Top Abstract I. Introduction II. How Are KATP... III. How Do KATP... IV. KATP Channel Subunits V. Reconstitution of KATP... VI. KATP Channel Structure VII. Regulation of KATP... VIII. Human SUR1 and... IX. KATP Channels and... X. Linking PHHI to... XI. Other Issues XII. KATP and NIDDM XIII. The Leptin Connection XIV. Summary and Conclusions References

 
The reconstitution experiments demonstrated that both SUR and K_IR subunits were required to form K_ATP channels, but raised a number of questions about the interactions, stoichiometry, and functional role(s) of each subunit within a channel and about the trafficking and activity of individual subunits. It was not at all clear, for example, why K_IR6.2 was silent in the absence of SUR. Some of these questions have been addressed, and it is becoming clear that essentially all of the kinetic, pharmacological, and regulatory properties of K_ATP channels including their sensitivity to inhibitory ATP and trafficking to the plasma membrane are highly dependent on interactions between SUR and K_IR subunits.

A. K_IR6.2 forms the pore of a K_ATP channel
An early question was which subunit formed the pore or permeation pathway of the channel and whether a part of SUR was involved. Two lines of evidence have been developed that show that K_IR6.2 subunits are sufficient to form the permeation pathway. Clement et al. (146) and Shyng et al. (204) showed that substitution of an aspartate for the asparagine at position 160, N160D, produced strongly rectifying channels; whereas asparagine, in the wild-type channel produced weaker rectification. The result did not eliminate the possibility that SUR1 could contribute directly to the permeation pathway, but did indicate that the M2 segments of K_IR6.2, like those of other inward rectifiers, were part of the permeation pathway. The discovery that truncated K_IR6.2 subunits, missing 16–35 amino acids from their C-terminal ends, can assemble a K⁺ channel provided a direct demonstration that SUR1 was not required to form the pore (51). These findings suggested that the underlying architecture of the pore of K_ATP channels would resemble that of the other members of the K_IR channel family.

B. SUR1 and K_IR6.x are physically associated
The requirement for coexpression of both SUR and K_IR to make an ATP-sensitive K⁺ channel implied these subunits were associated, but other alternatives were possible. Al-Awqati (205), for example, proposed that SUR might regulate a physically separate potassium channel by releasing ATP, as CFTR has been proposed to regulate the outwardly rectifying chloride channel (206, 207). The first direct evidence that SUR1 was associated with a second subunit came from the work of Schwanstecher, Panten, and colleagues (162, 163) as mentioned above. These authors developed an azido derivative of glibenclamide, [¹²⁵I]azidoglibenclamide (Fig. 2), and demonstrated that in addition to labeling SUR1 in brain membranes, an additional 38-kDa protein displayed high-affinity labeling. The identification of the 38-kDa component as K_IR6.2 became clear from [¹²⁵I]azidoglibenclamide labeling studies on COS cells expressing the cloned subunits (146). In these experiments SUR1 photolabeled independently of K_IR6.2, but K_IR6.2 did not bind glibenclamide directly and was only labeled when coexpressed with SUR1. These findings implied that ¹²⁵I-iodo-azidoglibenclamide was bound to SUR1 and that K_IR6.2 must be in close proximity, since the lifetime of the photo-activated species is brief.

C. Coexpression with K_IR6.2 affects the maturation of SUR1
The finding that coexpression of SUR1 and K_IR6.2 affected the maturation or glycosylation state of the receptor provided additional support for subunit association, provided a means to demonstrate complex formation directly, and was the first indication that SUR-K_IR assembly was needed for SUR to traffick to the plasma membrane. Expression of SUR1 alone generated the immature 140-kDa core glycosylated receptor, while coexpression with K_IR6.2 also generated mature complex glycosylated receptors (146). Chromatography on wheat germ agglutinin was used to show that K_IR6.2 was preferentially associated with mature SUR1 and was assembled into a stable multimer that could be isolated after solubilization in digitonin (146). These findings suggested a plausible assembly pathway in which the immature receptor associates with K_IR6.2 in the endoplasmic reticulum and maturation to the complex glycosylated form occurs in the medial Golgi as the assembled channel moves to the plasma membrane. These results agree with the observations of Ozanne et al. (165) that the 150-kDa species of SUR1 is in plasma membranes. It was not clear whether K_IR6.2 could traffic independently to the plasma membrane, but the result indicated SUR1 did not move efficiently through the medial Golgi in the absence of K_IR. The later observation that C-terminally truncated K_IR6.2 subunits could generate K⁺ channels raised the question of whether K_IR subunits could traffic independently to the plasma membrane (51).

D. Complex glycosylated SUR1 and K_IR6.2 assemble a large multimer
The finding that maturation of SUR1 was coupled with expression of K_IR6.2 and that these subunits assembled a stable complex argued that the multimer might be the channel itself. The mass of the stable multimer, estimated by sucrose gradient sedimentation, was approximately 950 kDa, consistent with a large channel composed of four complex glycosylated SUR1 subunits plus four K_IR6.2 subunits (an expected protein mass of 885 kDa = 4 x 176 kDa + 4 x 45 kDa, plus an unspecified carbohydrate mass) (146).

E. A 1:1 stoichiometry of SUR to K_IR is both necessary and sufficient to make K_ATP channels
The simplest model for the assembly of SUR1 with K_IR6.2 was a 1:1 association. Therefore, cDNAs encoding fusion proteins with defined stoichiometries were engineered to determine whether a 1:1 SUR/K_IR stoichiometry was both sufficient and necessary for the formation of functional K_ATP channels (146). Expression of SUR1K_IR6.2, a fusion of the N terminus of K_IR6.2 with the C terminus of SUR1 through a 6-glycine linker produced potassium-selective channels that were activated by metabolic poisoning and inhibited by sulfonylureas. Single-channel recording showed that the SUR1K_IR6.2 channels were weak inward rectifiers, which had the same conductance as unfused channels, although the sensitivity to both ATP and glibenclamide were reduced. Sucrose gradient analysis showed that SUR1K_IR6.2, like the wild-type subunits, produced a 950-kDa multimer. The results showed that a 1:1 stoichiometry was sufficient for formation of K_ATP channels.

Expression of a "triple" fusion construct, SUR1(K_IR6.2)₂, with a defined 1:2 stoichiometry, did not generate ATP-sensitive potassium currents. However, the triple fusion could be rescued by coexpression with monomeric SUR1, implying that the additional receptor(s) reestablished the required 1:1 stoichiometry. The SUR1 + SUR1(K_IR6.2)₂ channels had the same conductance properties as unfused channels, but were less sensitive to both ATP and sulfonylureas. The results showed that a 1:1 stoichiometry was necessary for assembly of K_ATP channels and, together with the mass estimates, argued for a tetrameric organization similar to other members of the K_IR channel family.

F. Other K_IR channels are tetramers
Several reports (208, 209) indicate that the K_IR channels, like the K_V channels (138, 139, 210, 211, 212), are tetrameric. Glowatzki et al. (208) coexpressed weak (K_IR1.1, ROMK1) and strong (K_IR4.1, BIR10) inward rectifiers and observed intermediate rectification properties expected from the assembly of heterologous channels. The differential sensitivity of these channels to voltage-dependent block by spermine was used to study the distribution of channel types assembled in cotransfected cells. Five types were identified that approximated the binomial distribution expected for a pore composed of four subunits. The polyamine binding site, in the pore, is therefore assumed to be assembled from amino acids contributed by four M2 segments. Yang et al. (209) studied the subunit stoichiometry of the strongly rectifying K_IR2.1 channel (IRK1) by fusing subunits to form multimers, a strategy that had been used earlier by Liman et al. (211) on K_V1.1 (RCK1) channels. The trimeric and tetrameric constructs, (K_IR2.1)₃ and (K_IR2.1)₄, both produced channels. Coexpression of the strongly rectifying wild-type trimers with a weakly rectifying mutant monomer produced a heterologous channel with intermediate rectification and weaker affinity for spermidine, while coexpression of the mutant monomer with the wild-type tetramer did not affect rectification or polyamine binding. The results were consistent with a tetrameric channel.

G. The stoichiometry of active ß-cell K_ATP channels is (SUR1/K_IR6.2)₄
A strategy based on the formation of heterologous channels was used to show that functional K_ATP channels are tetramers (146). cDNAs were engineered to express triple-fusion proteins, SUR1(K_IR6.2)₂ and SUR1(K_IR6.2_N160D)₂, which generated moderate or strongly rectifying channels, respectively, when rescued by SUR1 monomers. When these constructs were coexpressed, three classes of channels could be identified: the parental weak and strong rectifiers plus a third species with intermediate rectification. The presence of a speci