| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Medicine (L.A.-B.) and Cell Biology (J.B.), Baylor College of Medicine, Houston, Texas 77030
| Abstract |
|---|
|
|
|---|
C subunits with SUR restores
normal KATP channel activity | I. Introduction |
|---|
|
|
|---|
-subunits of
the amiloride-sensitive, non-voltage-gated epithelial Na+
channels, ENac, activate channel activity resulting in Liddles
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
Bartters 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 KATP 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 KATP 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 KATP channel, namely, the high-affinity sulfonylurea receptor, SUR1, and the K+ inward rectifier, KIR6.2, 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 KATP channels composed of the high-affinity sulfonylurea receptor, SUR1 (OMIM 600509), and KIR6.2, 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 KATP 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 KIR6.x isoforms have been cloned, KIR6.1 (uKATP-1) (25) and KIR6.2 (BIR) (26). Interestingly, the SUR and KIR genes are paired on human chromosomes, with SUR1 and KIR6.2 adjacent on chromosome 11, and SUR2 and KIR6.1 near each other on chromosome 12. When the low-affinity receptors are reconstituted with KIR6.1 or KIR6.2, they produce KATP channels with distinctive pharmacologies which, along with their tissue distribution, suggest they make up the cardiac-skeletal and vascular smooth muscle KATP channels (23, 24, 27, 28, 29).
| II. How Are KATP Channels Defined? |
|---|
|
|
|---|
|
-helices that form the pore, or of amino acids
within the gate. Silent intervals or gaps delineate the bursts. The
application of ATP, without Mg2+, reduces the time spent in
the open state (the mean open time) by prolonging the lifetime of the
gaps and by shortening the duration of the bursts (38, 39). MgATP also
inhibits channel activity. The IC50 for inhibition of the
ß-cell channel, SUR1/KIR6.2, by MgATP is reported to be
higher than for ATP4- (40), while the reverse has been
reported for the cardiac KATP channel,
SUR2A/KIR6.2 (41). The IC50 values for
inhibition by ATP have generally been reported to be higher for the
cardiac than for the ß-cell channel, although the variation of the
IC50 values, as well as the Hill coefficients, is quite
large for both (42). Our own values for the human
SUR1/KIR6.2 (ß-cell) and SUR2A/KIR6.2
(ventricular myocyte) channels are in the low micromolar range,
5
and 20 µM, respectively, with Hill coefficients near 1
with no dependence on Mg2+ (28). Whether the reported
variations in ATP sensitivity are due to tissue- specific
regulatory factors vs. experimental problems has not been
resolved. In the absence of Mg2+, ADP is also inhibitory.
Under physiological conditions with Mg2+ present, ADP
stimulates ATP-inhibited channels, providing one route for regulation
(43, 44).
|
|
| III. How Do KATP Channels Affect the Membrane Potential of Pancreatic ß-Cells? |
|---|
|
|
|---|
|
As illustrated in Fig. 3
, VM in the ß-cell is regulated
by glucose metabolism. When the external glucose concentration is less
than 3 mM, VM 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 KATP 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 KATP 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 ICl,islet, which is activated by cAMP, glibenclamide (110 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. ICl,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 (ICRAN). 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 ICRAN (60, 100, 101) or ICRAC for Ca2+ release-activated currents (102, 103, 104, 105). The use of the term ICRAC in this context appears to be somewhat unfortunate and this channel, which has a conductance in the 2040 picosiemens (pS) range, should not be confused with the low-conductance Ca2+ release-activated Ca2+ channel, ICRAC, described previously (106, 107, 108).
Dukes et al. (60) have reviewed the evidence for the role of ICRAN 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 ICRAN 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 ICRAN, 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 KATP channels. Coupling the control of ICRAN to emptying of internal Ca2+ stores is able to explain the effects of compounds such as maitotoxin (109, 110, 111), which activates ICRAN, 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 ICRAN remains to be elucidated.
6. Potassium channels. In addition to KATP
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
(KIR3.1, 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 KATP channels play in bursting? There is general agreement that KATP 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 KATP channels, in effect mimicking the signal from glucose metabolism, thus depolarizing the ß-cell membrane. Titration of KATP 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 KATP channels, thus shifting VM more toward EK. The question of how KATP 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 KATP 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 KATP 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 KATP channels were blocked by sulfonylureas. These results indicate KATP channels are not essential for bursting to proceed, but that residual channel activity may play a role in normal repolarization. Finally, if KATP 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 KATP channel-dependent and independent mechanisms for controlling insulin secretion (see for example Refs. 121, 122, 123).
The physiological ligand(s) or mechanism(s) controlling
KATP 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
KATP channels (30, 124). However, the IC50 for
inhibition of channel activity was determined to be in the 1050
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, KATP channel activity was strongly inhibited.
Subsequent studies showed that MgADP could open KATP
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 KIR6.2 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 |
|---|
|
|
|---|
|
1. Cloning of KIR6.x cDNAs. Using a fragment of KIR3.1 (GIRK1) cDNA as a probe, Inagaki et al. (144) cloned KIR6.1 (ruKATP-1) from a rat library. The tissue distribution of KIR6.1 was broad; hence the designation "u" KATP-1 for ubiquitous. Expression of KIR6.1 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 KIR6.1 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 KIR6.1 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 KIR retention signal. This report further added to the confusion surrounding the cloning and identification of KATP channels since KIR1.1 (1), KIR3.4 (147), and KIR6.1 (144) were all reported initially to be ATP-sensitive potassium channels.
KIR6.1 was not detected in ß-cell lines, and Inagaki et al. (144) used the KIR6.1 cDNA to clone the cDNA for KIR6.2 (originally designated BIR for ß-cell inward rectifier). Subsequent work showed that KIR6.2 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 KIR6.2 gene was found to contain no introns and is therefore relatively easy to obtain using the PCR.
KIR6.1 and KIR6.2 have significant blocks of
identical sequence. Seventy percent, 276 of the 390 residues in human
KIR6.2, are identical to those of KIR6.1 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. KIR6.1 has several
intriguing features, including a triple repeat at its C-terminal end,
S(M/I)RRNN.
|
TC-6 pancreatic
-cell line (134, 156), heart (157, 158), and from
brain (148, 149, 159, 160). Some effort was made to solubilize and
purify these proteins, and Bernardi et al. (161) reported
the purification of a 150-kDa high-affinity sulfonylurea
receptor from brain.
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, KD
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
KATP 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 KD. As described below, the 38- to 40-kDa
species is KIR6.2.
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 150170 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, KD
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 KATP 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 KATP 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 KIR6.2,
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 KATP channel activity. The link between sulfonylurea receptors and KATP 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 KATP 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 KIR6.2 channels, without receptors, or whether high passage HIT cells have lost KIR6.2 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 KIR6.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 SURs 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
p21ras-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 TMDNBF14 TMDNBF2). 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 TMDNBF16 TMDNBF2, emphasizes the existence within the MRP subfamily proteins of an multidrug resistance protein-like core consisting of 6 TMDNBF16 TMDNBF2. 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 KDs 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 KATP 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 |
|---|
|
|
|---|
|
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 KIR6.1 and SUR1 would form
KATP channels, but these were not characterized. These
authors also argued for greater promiscuity and pairing of the receptor
with KIR1.1 (ROMK1) and KIR3.4
(rcKATP-1/CIR/GIRK4). Clement et al. (146) have confirmed
that KIR6.1 and SUR1 produce metabolically activated
KATP channels, while coexpressing SUR1, and
KIR1.1 (ROMK1) or KIR3.4 (CIR, rcKATP) failed
to show association or KATP 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 KIR2.1 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
KATP 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/KIR6.1 channel have all been with intact cells, since
observations on excised patches have been hampered by rapid channel
rundown (145).
| VI. KATP Channel Structure |
|---|
|
|
|---|
A. KIR6.2 forms the pore of a KATP 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 KIR6.2
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
KIR6.2, like those of other inward rectifiers, were part of
the permeation pathway. The discovery that truncated KIR6.2
subunits, missing 1635 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 KATP
channels would resemble that of the other members of the
KIR channel family.
B. SUR1 and KIR6.x are physically associated
The requirement for coexpression of both SUR and KIR
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 KIR6.2 became clear from
[125I]azidoglibenclamide labeling studies on COS cells
expressing the cloned subunits (146). In these experiments SUR1
photolabeled independently of KIR6.2, but
KIR6.2 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
KIR6.2 must be in close proximity, since the lifetime of
the photo-activated species is brief.
C. Coexpression with KIR6.2 affects the maturation of
SUR1
The finding that coexpression of SUR1 and KIR6.2
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-KIR 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
KIR6.2 also generated mature complex glycosylated receptors
(146). Chromatography on wheat germ agglutinin was used to show that
KIR6.2 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
KIR6.2 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
KIR6.2 could traffic independently to the plasma membrane,
but the result indicated SUR1 did not move efficiently through the
medial Golgi in the absence of KIR. The later observation
that C-terminally truncated KIR6.2 subunits could generate
K+ channels raised the question of whether KIR
subunits could traffic independently to the plasma membrane (51).
D. Complex glycosylated SUR1 and KIR6.2 assemble a
large multimer
The finding that maturation of SUR1 was coupled with expression of
KIR6.2 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 KIR6.2
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 KIR is both necessary
and sufficient to make KATP channels
The simplest model for the assembly of SUR1 with
KIR6.2 was a 1:1 association. Therefore, cDNAs encoding
fusion proteins with defined stoichiometries were engineered to
determine whether a 1:1 SUR/KIR stoichiometry was both
sufficient and necessary for the formation of functional
KATP channels (146). Expression of
SUR1
KIR6.2, a fusion of the N terminus of
KIR6.2 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 SUR1
KIR6.2
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
SUR1
KIR6.2, like the wild-type subunits, produced a
950-kDa multimer. The results showed that a 1:1 stoichiometry was
sufficient for formation of KATP channels.
Expression of a "triple" fusion construct,
SUR1
(KIR6.2)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
(KIR6.2)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 KATP channels
and, together with the mass estimates, argued for a tetrameric
organization similar to other members of the KIR channel
family.
F. Other KIR channels are tetramers
Several reports (208, 209) indicate that the KIR
channels, like the KV channels (138, 139, 210, 211, 212), are
tetrameric. Glowatzki et al. (208) coexpressed weak
(KIR1.1, ROMK1) and strong (KIR4.1, 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 KIR2.1 channel (IRK1) by fusing
subunits to form multimers, a strategy that had been used earlier by
Liman et al. (211) on KV1.1 (RCK1) channels. The
trimeric and tetrameric constructs, (KIR2.1)3
and (KIR2.1)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 KATP channels
is (SUR1/KIR6.2)4
A strategy based on the formation of heterologous channels was
used to show that functional KATP channels are tetramers
(146). cDNAs were engineered to express triple-fusion proteins,
SUR1
(KIR6.2)2 and
SUR1
(KIR6.2N160D)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