The ATP-sensitive potassium (KATP) channel controls insulin secretion by coupling glucose metabolism to excitability of the pancreatic β-cell membrane. The channel comprises four subunits each of Kir6.2 and the sulphonylurea receptor (SUR1), encoded by KCNJ11 and ABCC8, respectively. Mutations in these genes that result in reduced activity or expression of KATP channels lead to enhanced β-cell excitability, insulin hypersecretion and hypoglycaemia, and in humans lead to the clinical condition congenital hyperinsulinism (CHI). Here we have investigated the molecular basis of the focal form of CHI caused by one such mutation in Kir6.2, E282K. The study led to the discovery that Kir6.2 contains a di-acidic ER exit signal, 280DLE282, which promotes concentration of the channel into COPII-enriched ER exit sites prior to ER export via a process that requires Sar1-GTPase. The E282K mutation abrogates the exit signal, and thereby prevents the ER export and surface expression of the channel. When co-expressed, the mutant subunit was able to associate with the wild-type Kir6.2 and form functional channels. Thus unlike most mutations, the E282K mutation does not cause protein mis-folding. Since in focal CHI, maternal chromosome containing the KATP channel genes is lost, β-cells of the patient would lack wild-type Kir6.2 to rescue the mutant Kir6.2 subunit expressed from the paternal chromosome. The resultant absence of functional KATP channels leads to insulin hypersecretion. Taken together, we conclude that surface expression of KATP channels is critically dependent on the Sar1-GTPase-dependent ER exit mechanism and abrogation of the di-acidic ER exit signal leads to CHI.
ATP-sensitive potassium (KATP) channels are inhibited by ATP and activated by Mg-ADP, a property that endows them with the ability to regulate insulin secretion in response to changes in blood glucose (1,2). When blood glucose levels rise, uptake and metabolism of glucose by the pancreatic β-cell increases. This results in an increase in intracellular [ATP]/[ADP] ratio, leading to closure of KATP channels, depolarization of the plasma membrane, opening of depolarization stimulated calcium channels and flux of Ca2+ into the cell. Rise in intracellular Ca2+ triggers insulin secretion. Thus KATP channels have the unique ability to translate changes in cell metabolism into changes in membrane potential and thereby regulate insulin secretion. Genetic mutations that alter their nucleotide sensitivity lead to disease states: Mutations that increase the sensitivity to [ATP]/[ADP] (loss of function) lead to congenital hyperinsulinism (CHI) (OMIM 600937) characterized by unregulated insulin secretion and severe hypoglycaemia (1,3–5). Conversely, mutations that reduce the sensitivity to [ATP]/[ADP] (gain of function) lead to neonatal diabetes mellitus (NDM) (OMIM 606176), where insulin secretion is seriously compromised with severe hyperglycaemia (6,7). Some mutations cause the disease by affecting the plasma membrane density of the channel by interfering with normal trafficking of the channel (8,9).
Structurally, the KATP channel is an octameric complex composed of four subunits each of Kir6.2, a member of the inwardly rectifying potassium channel, and the sulphonylurea receptor (SUR1), a member of the ABC protein superfamily (1–4,10). Kir6.2 subunits form a tetramer through which K+ ions permeate and also comprise inhibitory binding sites for ATP. SUR1 subunits are regulatory subunits; they surround the Kir6.2 tetramer and contain binding sites for the stimulatory Mg-nucleotides (2,11). Cell biological studies showed that neither subunit can exit the ER when expressed alone due to the presence of dibasic ER retention/retrieval signals, RKR (arginine–lysine–arginine), present on both the subunits (12). The details of how KATP channels escape the ER are not entirely clear, but assembly of the subunits into an octameric complex is thought to mask the retention signals enabling the channel to escape the ER. A recent study has indicated that 14-3-3 proteins play a role in masking the ER retention signals (13).
Kir6.2 is encoded by KCNJ11 and SUR1 by ABCC8. Genetic mutations leading to diseases of abnormal insulin secretion, viz. CHI and NDM, have been reported in both the genes (1). Mutations that cause severe functional defects lead to DEND (developmental Delay, Epilepsy and Neonatal Diabetes) syndrome (1). Although these diseases are rare, recent large scale genome-wide association studies have revealed a strong link between polymorphic mutations in KCNJ11 (e.g. E23K) and the more commonly occurring type 2 diabetes (OMIM 600937.0014) (14). The mechanism by which polymorphic mutations predispose individuals to type 2 diabetes, however, is unclear.
Studies of disease causing genetic mutations can be very rewarding in terms of providing novel insights into the molecular mechanisms underlying a disease. They also have the potential to reveal previously unrecognized mechanisms regulating the expression, function and cellular trafficking of proteins. In this paper, we studied the G844A heterozygous paternal mutation in the KCNJ11 gene of a Swedish patient with the focal form of CHI (15). Focal CHI is characterized by the loss of maternal chromosome 11p15, an area containing the KCNJJ11, ABCC8 and imprinted tumour suppressor genes. This unique combination reduces the genotype to hemi-zygosity for the mutant allele in the affected β-cells and leads to β-cell hyperplasia. Resection of the focal hyperinsulinaemic area of the pancreas resulted in the clinical cure of the patient (15).
The G844A mutation leads to the substitution of lysine (K) with glutamate (E) at position 282 of Kir6.2. Using recombinant systems, we demonstrate that this mutation causes ER retention, resulting in the complete absence of KATP channels from the cell surface, but by a process that is mechanistically different from all other mutations reported in the literature (16,17); unlike other mutations, the E282K mutation does not affect protein folding. Detailed investigations led to the new finding that KATP channels contain a di-acidic ER exit signal, 280DXE282, on Kir6.2; that the channels use this signal to recognize the COPII coat machinery and exit the ER via COPII vesicles; and that the mutation blocks ER exit by disabling the exit signal. Co-expression of the wild-type subunit rescued the mutant subunit from ER retention and resulted in functional expression. These results explain the phenotype of the patient as well as response of the patient to resection of the focal area of the pancreas (see Discussion).
The E282K mutation in Kir6.2 prevents ER exit and cell surface expression of KATP channels
To study the functional consequence of the E282K mutation, we introduced the mutation into Kir6.2 and expressed in Xenopus oocytes along with its obligatory subunit, SUR1 (12). Combined application of the metabolic poison, azide, which depletes ATP from the cell and diazoxide, a pharmacological activator of KATP channel, failed to evoke K+ currents (Fig. 1A). Under the same conditions, the wild-type KATP channel subunits gave rise to robust currents that are sensitive to glibenclamide, a specific inhibitor of KATP channels. When sections of these oocytes were immunostained for Kir6.2, fluorescence was apparent at the cell surface for the wild-type, but not for the mutant, channels (Fig. 1B). Expression of the mutant channel was seen only in the intracellular regions of the cell. Thus, the lack of currents in oocytes expressing the mutant channels appears to be due to impaired membrane trafficking.
To investigate the mechanism underlying the impaired surface expression, we used Kir6.2 tagged with the extracellular haemagglutinin A (HA) epitope (HA-Kir6.2) (18,19). HEK293 cells were co-transfected with the tagged wild-type or E282K mutant Kir6.2 subunit plus SUR1 and stained for the surface (red) and intracellular channel protein (green). As with the oocytes, there was distinct surface staining for the wild-type, but not for the mutant channel, despite robust intracellular expression (Fig. 1C). The intracellular fluorescence (green) showed almost complete co-localization (yellow) with the ER resident protein, calreticulin (red) (Fig. 1D), indicating ER retention of the mutant subunit. This is supported by the biochemical evidence (Fig. 1E): c-myc epitope-tagged SUR1 (myc-SUR1), when co-expressed with the wild-type Kir6.2 matures into a heavy glycosylated form (top ∼175 kDa band in Fig. 1E); however, such glycosylation was absent when myc-SUR1 was co-expressed with the mutant subunit, or expressed alone. Given that the assembly of Kir6.2 with SUR1 is a prerequisite for the ER exit (12), the data could suggest that the E282K mutation prevents interaction of Kir6.2 with SUR1. However, Figure 1F shows that, just like the wild-type subunit, mutant Kir6.2 was able to associate with SUR1: anti-HA antibodies were able to co-immunoprecipitate myc-SUR1 from the lysates of cells expressing myc-SUR1 plus either the wild-type or mutant (E282K) Kir6.2, but not myc-SUR1 alone. W91R is another CHI mutation which has been shown previously to cause ER retention of the channel due to mis-folding, but does not prevent association of Kir6.2 with SUR1 (20) and was used as a control in this experiment (Fig. 1F).
Kir6.2 contains a functional di-acidic ER-exit code which is obligatory for Sar1 GTPase dependent ER exit of KATP channels
E282 is preceded by an aspartate (D) at −2 position in the primary sequence of Kir6.2 (Fig. 2A). Thus, the 280DLE282 sequence in Kir6.2 conforms to the di-acidic ‘(D/E)X(D/E)’ ER exit motif (X = any amino acid) described in other membrane proteins (21,22). To test whether the DLE sequence serves as an ER exit code, we studied the effect of mutating this motif. Mutation of either of the conserved residues, but not the middle non-conserved residue, prevented surface expression of the channel (Fig. 2B and C), supporting the idea that the 280DLE282 sequence is a potential ER exit signal.
Membrane proteins bearing the ER exit signals are concentrated at the COPII-enriched ER exit sites (ERES) (22–25). This involves recognition of exit signals on the cargo protein by the COPII-coat machinery and packaging of cargo into COPII vesicles at the ERES (23). The COPII-coat machinery consists of the small GTPase Sar1 and the heteromeric protein complexes Sec23–Sec24 and Sec31–Sec13. Exchange of GTP for GDP by Sar1 GTPase initiates the assembly of cargo into COPII vesicles, and the GTP hydrolysis triggers the release of vesicles from the ERES (25–27). Thus, COPII vesicle-mediated ER exit of cargo can be prevented with dominant negative mutants of Sar1 such as Sar1T39N, which is restricted in its ability to exchange GDP for GTP, and Sar1H79G, which is incapable of hydrolysing GTP (27,28). When tested, both the Sar1 dominant negatives prevented the surface expression of KATP channels (Fig. 2D and E) by blocking the ER exit (Fig. 2F). Similar results were obtained with the insulin secreting rat β-cell line, INS1e (See Supplementary Material, Fig. S1), indicating the physiological relevance of the mechanism. Taken together, the data in Figure 2 indicate that KATP channels require the ‘DLE’ ER exit signal and Sar1 GTPase for ER exit.
The E282K mutation prevents entry of KATP channels into ERES
To confirm the idea that the ER exit signal promotes concentration of newly made KATP channels into the COPII-enriched ERES, we took advantage of a temperature-sensitive mutant of the vesicular stomatitis virus glycoprotein tagged with the green fluorescent protein (VSVG-ts045-GFP) and selective temperature block of ER exit (28,29). This protein uses the DXE motif to enter ERES. At the non-permissive temperature (39.5°C), VSVG-ts045-GFP is mis-folded and retained in the ER. Upon shift to the permissive temperature (32°C or lower), it assumes correct folding and exits the ER, but this exit step can be arrested by switching from 39.5 to 10°C, allowing functional visualization of ERES (29). Figure 3A shows control data: at 37°C, robust fluorescence of VSVG-ts045-GFP is seen at both the cell surface and in intracellular compartments. However, when pre-incubated at 39.5°C, and then shifted to 10°C, fluorescence was concentrated in distinct puncta, representing ERES. This effect could be replicated at 37°C by inhibiting the ER exit by co-expression of Sar1H79G. Using both approaches, we demonstrate co-localization of KATP channel with the co-expressed VSVG-ts045-GFP in the punctate ERES (Fig. 3B, top panels and 3C). In contrast, no such co-localization was apparent with the E282K mutant channels (Fig. 3B, lower panels). These data indicate that entry of KATP channels into ERES is dependent on an intact ER exit signal on Kir6.2 and that the E282K mutation prevents entry of the channels into ERES.
Kir6.2 and SUR1 subunits can enter ERES and the ER-Golgi intermediate compartment independently
Although previous studies reported that Kir6.2 and SUR1 subunits co-assemble prior to ER exit (12), there are no data to indicate that co-assembly of the subunits is essential for entry into the ERES. We therefore examined if Kir6.2 and SUR1 are capable of entering ERES independently. Inhibition of Sar1 function with Sar1H79G (Fig. 4A) resulted in the accumulation of both Kir6.2 and SUR1 in the VSVG-ts045-GFP-labelled ERES. The results were unexpected (see Discussion), but suggest that association of the two subunits is not a prerequisite for entry of Kir6.2 and SUR1 into ERES. Since proteins entering ERES are expected to exit the ER and enter ERGIC (ER-Golgi intermediate compartment), we tested if the KATP subunits expressed alone would colocalize with ERGIC-53, a marker for the ERGIC (30). Figure 4B shows that this indeed was the case; both Kir6.2 and SUR1 colocalized with ERGIC-53, indicating that the individual subunits are able to escape the ER. The data also suggest that SUR1 likely has an ER exit signal, the nature of which remains to be determined; it is worth noting that the ER exit of CFTR, a relative of SUR1, has been shown to occur via COPII vesicles (27).
Kir6.2 interacts with the COPII machinery with its C-terminal domain
To gain an insight into the molecular basis of interaction between the COPII coat machinery and the DLE motif on Kir6.2, we have examined the available structural information. In the crystal structure, the binding site for the DXE peptide was found on Sec24 of Sec23/24-Sar1 complex of COPII at an estimated distance of 20 Å from the membrane interface (24,31) (See Supplementary Material, Fig. S2). In a 3D structural model of Kir6.2, the 280DLE282 peptide sequence is solvent exposed (Fig. 5) and is located at ∼25 Å from the putative membrane interface. Thus, the ER exit signal on Kir6.2 appears to be placed favourably for the concave-shaped Sec23/24-Sar1 complex to bind and bend the membrane to form the COPII vesicle (Supplementary Material, Fig. S2). Although data presented above (Fig. 4) indicated that when expressed individually both Kir6.2 and SUR1 can enter ERES, and therefore could exit the ER, under natural conditions, when both subunits are co-expressed, co-assembly of the two subunits could occur before entering ERES. Previous studies indeed reported that Kir6.2 assembles with SUR1 prior to ER exit (12). However, if complete octomeric assembly occurs in the ER, SUR1 subunits surrounding the central Kir6.2 tetramer could present stearic hindrance for interaction with Sec24. Thus, it could be argued that the signal on Kir6.2 is incapable of binding the COPII machinery and the observed effects of mutating the signal could be indirect.
To address this, we used a fusion protein where we replaced the entire C-terminal domain of CD4 with that of Kir6.2. In this construct, CD4-Kir6.2CT, we hoped that the distance of the DLE signal relative to the intracellular membrane face would be preserved (Fig. 5A) to allow COPII vesicle formation. The rationale is that CD4, a type 1 membrane protein, which lacks ER exit signals, would traffic to the plasma membrane in a COPII independent manner, but the addition of C-terminus of Kir6.2 would confer the ability to become COPII dependent. As predicted, the surface expression of CD4 was found to be COPII independent, as demonstrated by the inability of Sar1H79G to prevent its surface expression (Fig. 5B). In contrast, CD4-Kir6.2CT showed normal surface expression, but failed to reach the cell surface in the presence of Sar1H79G co-expression; instead, the fusion protein was retained at ERES labelled with VSVG-ts045-GFP (Fig. 5C), just like the wild-type KATP channel. Furthermore, introduction of the E282K mutation into the C-terminus of CD4-Kir6.2CT prevented its surface expression (Fig. 6A), and entry of the fusion protein into ERES, as judged by the absence of co-localization with VSVG-ts045-GFP in Sar1H79G co-transfected cells (Fig. 6B; bottom panel). Taken together, these data indicate that direct interaction of DLE sequence in Kir6.2 with the COPII coat machinery is responsible for the concentration of KATP channels into ERES.
We next asked if a membrane-permeable peptide containing the ER exit signal sequence on Kir6.2 would interfere with the ER exit of the channel by competing for the cargo binding site on the COPII coat machinery. For this, we used a synthetic peptide containing the tat sequence fused to residues 277HHQDLEIIV286 of Kir6.2. The tat sequence derived from the HIV viral protein facilitates cellular uptake of peptides and proteins to which it is fused (32). Figure 7 shows that the fusion peptide has no effect on the expression of the wild-type KATP channel, but completely blocks membrane trafficking. The control tat peptide as well as the fusion peptide containing the E282K mutation had no effect on membrane trafficking. These results further support our argument that DLE sequence of Kir6.2 interacts with the COPII machinery.
The wild-type subunit can rescue the surface expression and function of the E282K mutant subunit
We next asked if the mutant subunit could associate with the wild-type subunit and reach the cell surface. For this, we have co-expressed the wild-type Kir6.2 lacking the HA epitope with HA-Kir6.2E282K, plus SUR1, and stained with anti-HA antibodies. Figure 8A shows surface anti-HA staining, which must be owing to heterooligomers comprising the wild-type and mutant subunits. The result suggests that not all four ER exit signals are required for the ER exit of the channel. Furthermore, the E282K mutation, unlike all other mutations reported, does not cause ER retention through protein mis-folding. In contrast, heterooligomers containing the W91R CHI mutation, which is outside of the ER exit signal and known to cause mis-folding (20), did not show surface expression.
We next asked whether the E282K mutant subunits can form functional channels with the wild-type subunit. To address this, we injected Xenopus oocytes with the mutant and wild-type Kir6.2 cRNA in a 1:1 ratio and measured glibenclamide sensitive K+ currents. The composition of the oligomers predicted to form in the co-injected cells is shown schematically in Figure 8B, where the percentage of each oligomer is calculated based on the assumptions that the mutation does not bias the association and the assembly follows a binomial distribution. Accordingly, 6.25% channels are expected to be homomeric mutant channels that would be retained in the ER. The remaining 93.75% channels will have between 1 and 4 wild-type subunits, and depending on the number of ER exit signals required, all or a proportion of these oligomers will exit the ER. Figure 8C shows that mean current amplitude from co-injected oocytes is ∼60% less than that from cells injected with the wild-type Kir6.2 cRNA only. Quantitative analysis of the surface expression (Fig. 8C) revealed a corresponding reduction in the surface expression of the channels. These data suggest that more than one ER exit signal per oligomer may be required for ER exit.
It could be argued (albeit unlikely given the data in Fig. 1F) that the wild-type subunits do not associate with the mutant subunits in the oocytes and the currents seen are due to wild-type channels (resulting from the 50% of the injected cRNA) expressed at the cell surface. To address this, we made a tandem dimeric construct between the wild-type and mutant subunits. This dimer, like the wild-type–wild-type tandem dimer, gave rise to characteristic KATP currents (Fig. 8D). These data also suggest that two ER exit signals are enough to promote surface expression of the channel because the tandem dimer will contain only two ER exit signals per channel complex.
Our investigation into how the E282K genetic mutation in Kir6.2 causes CHI led to revelation of a previously unrecognized ER exit signal that proved to be crucial for membrane trafficking of KATP channels, and that abrogation of the signal underlies the disease.
Electrophysiological experiments showed that Kir6.2 containing the E282K mutation does not form functional channels (Fig. 1A). Confocal imaging and biochemical studies indicated that the lack of functional expression is due to impaired trafficking caused by the retention of the channel in the ER (Fig. 1B–E). However, unlike other mutations reported in KATP channels (8) and other membrane proteins (16), including CFTR (17), the mutation does not cause protein mis-folding. This is evident from the fact that the mutant subunit, Kir6.2E282K, was able to assemble with the wild-type Kir6.2 and SUR1 (Fig. 1F) and give rise to functional channels at the cell surface (Fig. 8). Thus, the mechanism of ER retention by the E282K mutation is mechanistically different from the previously reported mutations in that mis-folding is not the cause of ER retention.
Mutation of residues surrounding the E282 residue revealed that the 280DLE282 sequence in Kir6.2 may function as an ER exit code, as with several other membrane proteins (22). Proteins bearing such signals are recognized by the COPII coat machinery in a reaction that involves Sar1-GTPase and packaged into COPII vesicles that appear at ERES (22–27). Such concentrative step ensures efficient delivery of proteins to the cell surface. Several lines of evidence indicated that KATP channels use the 280DLE282 sequence as a signal to assemble into COPII vesicles and exit the ER. First, mutation of the key residues of the DLE sequence prevented surface expression (Fig. 2B and C), Second disruption of the Sar1-GTPase function with dominant negative constructs prevented delivery of wild-type KATP channels to the cell surface (Fig. 2D and E). Third, when hydrolysis of Sar1-bound GTP (required for release of COPII vesicles from the ER) was prevented using the Sar1H79G construct, the wild-type channel accumulated in ERES; in contrast, channels containing the E282K mutation failed to enter ERES (Fig. 3). Fourth, surface expression of a reporter protein, CD4, which lacks ER exit signals, was Sar1-independent, but substitution of its C-terminal domain with that of Kir6.2 conferred dependency on Sar1-GTPase; this is evident from the finding that the CD4-Kir6.2CT fusion protein, unlike CD4, accumulates in ERES when Sar1H79G was co-expressed (Fig. 5). Furthermore, E282K mutation prevented surface expression of CD4-Kir6.2CT (Fig. 6). Finally, a membrane permeable tat-fusion peptide containing the ER exit signal of wild-type Kir6.2 (HHQDLEIIV), but not the mutant form (HHQDLKIIV) prevented ER exit of the channel (Fig. 7). Taken together, we conclude that KATP channels contain a functional di-acidic ER exit signal which promotes concentration of the channel into COPII enriched ER exit sites prior to secretion. In the absence of such a concentrative step, delivery of the channel to the cell surface is almost completely prevented, indicating that the ER exit signal is indispensable for cell surface expression of the channel.
Zerangue et al. (12) reported that both Kir6.2 and SUR1 contain the ‘RKR’ ER localization signals which restrict surface expression to fully assembled channels, and retain unassembled subunits and partial complexes in the ER. The authors proposed that Kir6.2 subunits form a tetramer first, exposing the RKR motifs, but subsequent binding of SUR1 masks the signals on the tetramer and SUR1 itself, allowing the trafficking of fully assembled octomeric complexes to the cell surface. In the light of this report, our finding that both Kir6.2 and SUR1 could be recruited into COPII-enriched ERES (Fig. 3) and escape to ERGIC (Fig. 4) even when expressed independently was somewhat unexpected, but this result sheds new light on our understanding of the assembly and ER exit of KATP channels. The result suggests that interaction between Kir6.2 and SUR1 is not required for recruitment into COPII vesicles. Furthermore, the ‘RKR’ ER localization signals do not seem to prevent independent entry of the channel subunits into COPII-enriched ERES and ERGIC. Taken together with the previous report (12), it seems that the channel subunits can exit the ER as individual subunits as well as co-assembled complexes. Whether the fully assembled octomeric channel complex can enter ERES is unclear, but this would seem unlikely for two reasons: first, in the fully assembled state (33), the ER exit signals on Kir6.2 are likely be masked by SUR1 preventing its recognition by the COPII machinery. Second, the fully assembled complex is ∼900 kDa and has a diameter of 18 nm (33); one would expect the large size to present mechanistic problems for the COPII coat machinery to package the complex into COPII vesicles, which appear to be 60–70 nm in diameter (26,34). Individual subunits and partial complexes of the channel, on the other hand, could be more easily accommodated into COPII vesicles. The fact the E282K mutant subunit is retained in the ER (does not enter ERES; Fig. 3B), but the wild-type subunit can rescue it from ER retention suggests that Kir6.2 subunits exit the ER as a multimer, most likely as a tetramer.
The finding that both Kir6.2 and SUR1 can exit the ER and enter ERGIC (Fig. 4) suggests that complete assembly of the channel likely occurs in the ERGIC or the cis-Golgi compartment. It has been suggested that 14-3-3, a protein capable of probing the assembly status of the channel, plays a role in the assembly of KATP channels in a post-ER compartment (13). Any unassembled subunits, bearing exposed ‘RKR’ signals could be recognized by the COPI machinery, and retrieved to the ER via COPI vesicles (13,35). It therefore seems plausible that the ‘RKR’ ER localization signals in the KATP channel subunits may serve as retrieval signals, rather than retention signals.
The patient bearing the E282K mutation has the focal form of CHI, with the mutation found on the paternal chromosome. In focal CHI, the affected β-cells express the mutant subunits from the paternal chromosome, but no wild-type subunits to rescue the surface expression of the mutant subunits, because of the loss of maternal chromosome 11p15, an area containing the KCNJJ11 gene (15). The consequent loss of functional channels in the foci of the islets would explain the phenotype of the patient. Following the resection of the focal hyperinsulineamic area, however, the patient was clinically cured (15); this could be explained by the fact that the non-focal region of the pancreas would express both the wild-type and E282K mutant subunits that are expected to give rise to heterozygous functional channels capable of regulating insulin secretion.
In summary, studies of a CHI causing KCNJ11 mutation led to the discovery that Kir6.2 contains a di-acidic ER exit signal that is mandatory for the ER exit of KATP channels. The signal promotes concentration of the channels at the COPII-enriched ERES, a step that is crucial for the delivery of the channel to the cell surface. Importantly, the finding that abrogation of this signal leads to CHI underpins the physiological importance of the signal. Although disease causing mutations have been reported in genes encoding the components of the COPII machinery, including Sar1 (36) and Sec23 (37,38), this study provides the first example of a mutation in a cargo protein causing disease in humans. Besides revealing the molecular basis of the focal form of CHI detected in the patient, our results also provide insights into fundamental mechanisms underlying the assembly and ER exit of KATP channels by raising the possibility that complete assembly of the channel complex could occur in a post-ER compartment.
MATERIALS AND METHODS
The HEK293-MSRII cell line was provided by Glaxo-Smith Kline, Stevenage, UK. tsA-201 were from ECACC (Ref. 96121229). HA-tagged Kir6.2 in pCDNA3 and myc-tagged SUR1 in pCDNA6 were used for the expression in mammalian cell lines as described previously (19). Kir6.2-FLAG and SUR1 sequences, both in the pKS-globin vector, were used for preparation of cRNA for expression in Xenopus occytes (9). Wild-type and mutants of hamster Sar1a (acc. AAB30321.1) in pCDNA3.1 were kindly provided by WE Balch, The Scripps Research Institute, La Jolla, CA, USA versus (27). VS VG-ts045-GFP was a gift from Kai Simons, Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany. Human CD4 (M12807) and CD4 containing the substitution of the C-terminal domain (residues 178–364) of Kir6.2 (CD4-Kir6.2-CT), both in pCDNA3, are as described (19). Rat anti-HA antibodies (clone 3F10) were purchased from Roche Diagnostics, GmbH. Mouse anti-myc antibodies were from Cell Signaling Technology, Danvers, MA, USA. AlexaFluor488-conjugated anti-rat secondary antibodies were obtained from Invitrogen Ltd, Paisley, UK. Anti-rat FITC- and Cy3-conjugated secondary antibodies were purchased from Jackson ImmunoResearch (West Grove, PA). Horse radish peroxidase (HRP)-conjugated goat anti-rat secondary antibodies were from Sigma (St Louis, MO, USA). Mouse anti-CD4 monoclonal antibodies (Q4120) were a gift from Dr P. Beverley (Edward Jenner Vaccine Research, Compton, UK). Protein G-Sepharose was from Upstate, Buckingham, UK. All tissue culture media and antibiotics were from Invitrogen Ltd, Paisley, UK. Lumigen PS-Atto substrate was from Lumigen Inc, Southfield, MI, USA. All other chemicals were from Sigma (St Louis, MO, USA). Tat-fusion peptides were custom-made by Peptide 2.0, Chantilly, VA, USA.
Molecular biology and cell transfections
Point mutations were introduced by the QuikChange mutagenesis method (Stratagene). Tandem dimeric Kir6.2 constructs were produced by inserting an EcoRI site prior to the stop codon of Kir6.2 protomer-1 and engineering an EcoRI site prior to the ATG start codon of Kir6.2 protomer-2. Protomer-1 was then ligated to protomer-2 at the EcoRI site using T4 DNA ligase to generate the Kir6.2WT-Kir6.2WT tandem dimer. The E282K mutation was introduced into promoter-2 to generate the Kir6.2WT-Kir6.2E282K tandem dimer. The tandem dimers were cloned into pKS-globin. cRNA was prepared from the pKS-globin vector-cDNA constructs for oocyte expression as described previously (9). HEK293-MSRII cells were cultured in DMEM/F12 supplemented with 10% foetal calf serum, 100 U ml−1 penicillin, 100 µg ml−1 streptomycin and 400 µg ml−1 G418. Cells were grown on poly-D-lysine coated coverslips (immunocytochemistry) or 24 well plates (chemiluminescence) and transiently co-transfected with pCDNA6-SUR1 and pCDNA3-HA-Kir6.2 (3:1 ratio) plus other clones as desired using Fugene6™ (Roche Diagnostics, GmbH). For immunoprecipitation, tsA201 cells (grown in DMEM/10% foetal calf serum) were grown in 100 mm dishes and transfected using the standard calcium phosphate method.
Expression and electrophysiological analysis of KATP channels was performed as described previously (9). Xenopus oocytes were co-injected with 5–20 ng of cRNA encoding Kir6.2 (WT, mutant or tandem dimer) and 3-fold excess of SUR1 cRNA and cultured for 48–72 hours before measuring K+ currents. Currents were recorded in Ringer's solution containing 90 mm KCl, 1 mm MgCl2, 1.8 mm CaCl2 and 5 mm HEPES (pH 7.4) using two-electrode voltage clamp. Currents were measured at −120 mV during a ramp protocol (−150 to +70 mV) using a holding potential of −30 mV. A mixture of 200 µM diazoxide and 3 mm sodium azide was used to maximally activate currents and 1 µM glibenclamide to inhibit the currents. Measurements were made from at least three occytes. Representative time-course data are presented. For Figure 5B, oocytes were injected with cRNA for the wild-type, mutant or a mixture of wild-type and mutant channel. Glibenclamide sensitive currents are presented as mean and standard errors of mean (SEM). Paired Student t-test was applied to determine the significance (P) values.
Transfected cells grown on coverslips were fixed in 2% PFA in phosphate buffered saline (PBS). Non-specific binding was reduced by pre-incubating in 5% goat serum in PBS (blocking buffer) and diluting all antibodies in blocking buffer. Surface channels were labelled by incubation for 1 h in primary rat anti-HA antibodies (1:400 dilution) followed by secondary Cy3-conjugated donkey anti-rat IgG antibodies (1:800 dilution) and washing in PBS between the steps. Where detection of intracellular channels was also desired, the same cells were permeabilized in 50% acetone/methanol solution (−20°C) at 4°C for 10 min before treating with anti-HA antibodies and AlexaFluor488-conjugated donkey anti-rat IgG (1:400 dilution). Calreticulin was probed using monoclonal mouse anti-calreticulin antibody (1:400 dilution; Stressgen), in conjunction with Cy3 conjugated anti-mouse antibody (1:800 dilution). CD4 or CD4-Kir6.2-CT fusion proteins were stained using the anti-CD4 antibodies to an extracellular epitope. Cells were viewed under a Zeiss 510-META laser scanning confocal microscope under an oil-immersion ×63 objective (NA = 1.40). AlexaFluor488 (494 nm excitation: 519 nm emission) was excited using an argon laser fitted with a 488 nm filter and Cy3 (550 nm excitation: 570 nm emission) was excited using a helium/neon laser fitted with a 543 nm filter. Immunostaining of Xenopus oocyte sections was performed as described previously (9).
Quantification of cell surface density of KATP channels
Density of HA-tagged KATP channels at the cell surface was estimated by the chemiluminescence method as described previously (19). Briefly, cells transfected in 24 well plates were fixed, labelled with rat anti-HA antibody (1:500 dilution) for 1 h at room temperature, and after washing, incubated with the HRP-conjugated goat anti-rat IgG (1:500 dilution). Cells were washed, solubilized overnight at 4°C in 400 µl of 2% sodium deoxycholate in PBS supplemented with Benzonase (5 units) and 1X Protease inhibitors (Roche). Duplicate aliquots (50 µl) of the lysate were treated with 50 µl of Lumigen PS-Atto substrate in a 96 well plate (Nunc white) for 30 s, and the luminescence measured using PolarStar Optima (BMG Labteck, GmbH). Protein content was measured on a 40 µl aliquot using the Bicinchoninic acid (BCA) method. Luminescence values were normalized to the cell protein content. This gives a measure of the surface density of the channels. In a parallel set of experiments, cells were permeabilized after fixing to determine the total expression of the channel. Surface expression values were divided by total expression values. The resultant values were normalized to the wild-type. Data were obtained from three separate transfection experiments, each measured in duplicate and expressed as mean ± SEM. Statistical significance of differences was determined using One-Way ANOVA and Bonferroni test; P < 0.05 was accepted as significant. Chemiluminescence-based assay of surface expression on oocytes was performed as described previously (12).
Immunoprecipitation and western blotting
tsA201 cells were transfected in 100 mm dishes with the desired constructs and lysed in RIPA buffer (50 mm Tris–HCl, pH 7.4, 150 mm NaCl, 1 mm EDTA, 1% Nonidet P40, 0.25% Na-deoxycholate). Lysates were subjected to immunoprecipitation using Protein G-Sepharose preadsorbed with anti-HA antibodies. Following three to four washes with the RIPA buffer, the beads were suspended in SDS sample buffer (1.25% SDS, 0.06 m Tris–HCl, pH 6.8, 10% glycerol, 8 mm EDTA) and incubated at 37°C for 30 min prior to SDS–PAGE on an 8% gel. For western blotting, proteins were transferred to nitrocellulose membrane (Trans-Blot; Bio-Rad). Blots were probed with anti-myc antibodies and the bands detected by chemiluminescence using Lumigen PS-atto.
The work was supported by the Medical Research Council, UK. S.K. is funded by the Overseas Associateship Programme, Department of Biotechnology, Government of India.
R.K. and A.J.S. received PhD studentships from Overseas Research Studentships Awards Scheme and the University of Leeds.
Conflict of Interest statement. None declared.