Dual role of K_ATP channel C-terminal motif in membrane targeting and metabolic regulation

Kir6.2 Associates with Ankyrin-B.

To define molecular determinants of K_ATP channel membrane organization, we assessed the Kir6.2 and SUR cytoplasmic domains (Fig. 1A) for potential targeting motifs. Interestingly, sequence analysis of Kir6.2 revealed striking similarity between a C-terminal cytoplasmic sequence and a motif recently established to target voltage-gated channels (via ankyrin proteins) to specialized domains of excitable neurons and myocytes (6–10) (Fig. 1 B and C). Further analysis of this motif revealed high sequence conservation between Kir6.2 orthologs (Fig. 1C).

We evaluated potential interaction of Kir6.2 and ankyrin-B (AnkB) by co-immunoprecipitation. AnkB Ig co-immunoprecipitates Kir6.2 from detergent-soluble lysates generated from pancreas and heart (Fig. 1D and Fig. S1A). We observed no interaction between Kir6.2 and control Ig (Fig. 1D and Fig. S1A). Moreover, AnkB was unable to co-immunoprecipitate the related Kir6.1 gene product from heart (Fig. S1B). We examined a potential direct AnkB/Kir6.2 interaction using purified proteins. In vitro binding assays using radiolabeled full-length Kir6.2 revealed a direct interaction with the AnkB membrane-binding domain (MBD) (Fig. 1E), with robust AnkB MBD/Kir6.2 binding in salt concentrations of 500 mM (Fig. S2). Both the AnkB spectrin-binding domain (SBD) and C terminal domain (CTD) displayed minimal binding activities for radiolabeled Kir6.2 similar to GST alone (Fig. 1E). Together, our findings indicate that AnkB and Kir6.2 are protein partners in vivo and that this direct interaction is mediated by the AnkB MBD.

AnkB Associates with Kir6.2 C Terminus.

Kir6.2 contains a cytosolic N terminus, two transmembrane domains, a P-loop, and a cytosolic C terminus (Fig. 1F). We generated Kir6.2 mutants lacking either the N- or C-terminal cytoplasmic domains (Fig. 1F) to establish the structural requirements on Kir6.2 for AnkB association. Consistent with the role of a putative AnkB-binding motif (ABM) in the Kir6.2 C terminus, Kir6.2 lacking the C terminus (Kir6.2ΔCT) is devoid of AnkB-binding activity (Fig. 1F Right). In contrast, Kir6.2ΔNT binds AnkB MBD (Fig. 1F Center), similar to full-length Kir6.2 (Fig. 1F). These findings establish the Kir6.2 C-terminal domain as necessary for AnkB association.

A Kir6.2 C-Terminal Motif Is Required for AnkB Association.

We evaluated the role of the Kir6.2 C-terminal motif for ankyrin-binding using in vitro binding assays and two additional Kir6.2 mutants (Fig. 1G). The first mutant was engineered with a premature stop codon following the putative binding motif (Kir6.2 Δ331; Fig. 1G Right), while the second mutant was engineered with a premature stop codon preceding the putative binding motif (Kir6.2 Δ315; Fig. 1G Center). In vitro binding with immobilized GST-AnkB MBD revealed that only Kir6.2 Δ331 displayed binding activity for AnkB (Fig. 1G Right). Binding of Kir6.2 Δ315 for GST-AnkB was similar to GST alone (Fig. 1G Center). These data identify an eight-residue motif in Kir6.2 C terminus that is necessary for interaction with AnkB. In fact, Kir6.2 residues 313–325 are necessary and sufficient for the ankyrin interaction, as a biotinylated peptide mimicking the putative Kir6.2 ABM associates with AnkB (Fig. 1H). In agreement with co-immunoprecipitation data, the analogous, but nonidentical motif found in Kir6.1 lacked AnkB-binding activity (Fig. 1H)

AnkB^+/− Mice Display Abnormal Kir6.2 Expression and Localization.

We evaluated a role for AnkB in Kir6.2 membrane expression in vivo using mice with reduced AnkB expression (AnkB^+/− mice). AnkB is highly expressed in the pancreatic β cell (Fig. 2A), with minimal levels observed in other islet cell types or exocrine acinar cells (11). AnkB^+/− mice are viable and express 50% reduction in islet AnkB levels (Fig. 2B, ≈50% decrease in pancreas by immunoblot, Fig. 2C). We observed a striking reduction in expression levels of Kir6.2 in AnkB^+/− pancreas by immunoblot and immunostaining (Fig. 2 F, I, and J). In contrast, we observed no difference in expression of Ca_v1.2, SUR1, ankyrin-G, or NHERF1 (Fig. 2 D, E, G, and H and Fig. S3). Moreover, the levels of two proteins affected by loss of AnkB in heart, Na/Ca exchanger and Na/K ATPase (12), were unchanged in AnkB^+/− mouse pancreas (Fig. S3). Real-time PCR revealed no difference in the levels of mRNA encoding Kir6.2 (KCNJ11), suggesting that abnormal expression of Kir6.2 in AnkB^+/− pancreas is likely due to posttranscriptional events.

AnkB Is Required for Kir6.2 Membrane Expression.

As AnkB^−/− mice die shortly after birth, we created a primary cell culture system lacking AnkB to directly test the role of AnkB for Kir6.2 membrane targeting. Primary fibroblasts were isolated from wild-type (WT) and AnkB^−/− postnatal day 1 mice. Absence of AnkB in primary null cells was tested by genotyping, immunostaining, and immunoblot (Fig. 2 K, L, N, O, cells also lack Kir6.2, and Fig. 2M) GFP-tagged Kir6.2 (but not GFP) was localized to the membrane of transfected WT fibroblasts (Fig. 2 P and Q, cells cotransfected with SUR1). In contrast, GFP-Kir6.2 was not targeted to the membrane of AnkB^−/− fibroblasts, but instead was localized to a perinuclear region of AnkB^−/− cells (Fig. 2R).

Finally, we definitively tested the role of AnkB for Kir6.2 membrane expression using inside-out patch clamp measurements of I_KATP in WT, AnkB^+/−, and AnkB^−/− fibroblasts transfected with Kir6.2 and SUR1 (Fig. 3). In agreement with qualitative staining experiments, quantitative functional analyses of GFP-Kir6.2 membrane expression revealed a marked reduction of K_ATP channel current in AnkB^+/− and AnkB^−/− primary cells compared with WT fibroblasts (measured at 0 mM ATP to stimulate opening of all membrane associated K_ATP channels; Fig. 3, A, B, and E). As expected, I_KATP channel currents were blocked by 1 mM ATP (Fig. 3 C–E). Finally, in both immunostaining and functional electrophysiology experiments, defective Kir6.2 membrane targeting phenotypes in AnkB^−/− cells were corrected by re-expression of AnkB cDNA (Fig. S2 and Fig. 3 A–E). Together, these findings support a requirement of AnkB for efficient Kir6.2 membrane localization.

Human Diabetes Disease Mutation Abolishes AnkB Interaction.

We previously identified a human mutation in the ankyrin-binding motif of cardiac Na_v1.5 that causes arrhythmias in the Brugada syndrome (8) (Fig. 4A). At the cellular level, Brugada syndrome is characterized by decreased Na_v1.5 membrane activity. Consistent with this phenotype, the Na_v1.5 E1053K mutant lacks ankyrin-binding activity (Fig. S4) and is not targeted to the membrane of primary myocytes (resulting in decreased I_Na) (8). Permanent neonatal diabetes mellitus (PNDM) is characterized by the development of diabetes within the first 6 months of birth and life-long insulin dependence (13). In 2004, Vaxillaire identified a Kir6.2 mutation in a 10 year 10 month old insulin-dependent female with PNDM (presented with hyperglycemia at three days old) (14). Strikingly, the disease-causing mutation (E322K) is located in the AnkB-binding motif (ABM) at a homologous position to the Na_v1.5 Brugada syndrome disease variant E1053K (8) (Fig. 4A). We engineered the human PNDM mutation (E322K) in human Kir6.2 and examined AnkB-binding. Additional lysine and alanine substitutions were generated in this and adjacent negatively charged residues E321 and D323. All charge reversal mutants (E321K, E322K, and D323K) eliminated Kir6.2 AnkB-binding activity (Fig. 4B). Consistent with these findings, a peptide against the Kir6.2 AnkB-binding motif engineered with the human E322K mutation displayed minimal binding activity for AnkB (Fig. 4 C and D). In contrast, neutral alanine substitutions were without effect on AnkB-binding (Fig. 4B). Together, these findings identify a key acidic stretch in the ankyrin-binding motif as critical for AnkB/Kir6.2 interaction and demonstrate that the human Kir6.2 E322K disease mutation disrupts the Kir6.2/AnkB association.

Kir6.2 Disease Mutant Lacking Ankyrin-Binding Activity Is Not Efficiently Targeted.

Based on the recognized role of ankyrins for membrane protein targeting (15), we further evaluated the membrane localization of the Kir6.2 human diabetes mutation E322K. In contrast to the WT channel, Kir6.2 E322K (and E321K, D323K) was not efficiently targeted to the membrane of HEK293 cells (Fig. 4 E–I) or primary mouse fibroblasts (both cell types express AnkB; Fig. 4 J and K). Specifically, these mutant subunits were primarily clustered in a perinuclear pattern (Fig. 4 G–K). These results demonstrate that the Kir6.2 C-terminal ABM is essential for efficient Kir6.2 membrane targeting, such that disruption of this motif will lead to loss of membrane Kir6.2, an effective Kir6.2 loss-of-function.

Kir6.2 Mutant Lacking AnkB-Binding Activity Displays Decreased ATP Sensitivity.

Reduced K_ATP channel trafficking due to the E322K mutation should result in a loss of cellular K_ATP channel activity, unless the residual membrane channels display strongly augmented K_ATP channel opening probability. In fact, the prediction that E322K mutant cells show loss-of-function due to defective membrane insertion is in apparent contrast to the observed mild gain-of-function PNDM phenotype of the human Kir6.2 E322K mutation (14). Based on this discrepancy between the cellular trafficking (Kir6.2 loss-of-function) and mild human (gain-of-function) E322K phenotypes, we analyzed multiple cell lines to identify a mammalian cell system suitable for studying the biophysical gating properties of Kir6.2 trafficking-defective E322K-GFP mutants at the plasma membrane. COSm6 cells lack Kir6.2 but retain AnkB expression (Fig. S5) and have enhanced activity of an AnkB-independent K_ATP channel trafficking pathway compared with all other primary and cultured cells examined (see Figs. 2–4). Still, while Kir6.2 E322K trafficking remains significantly impaired in COSm6 cells [≈20% as assessed by qualitative immunostaining (Fig. S6) and quantitative functional ⁸⁶Rb⁺ efflux experiments performed in “metabolic inhibition” conditions (oligomycin and 2-deoxyglucose) designed to lower cellular [ATP]:[ADP] ratios and fully activate membrane K_ATP channels (16) (Fig. S6B)], sufficient Kir6.2 E322K-GFP is targeted to the plasma membrane to permit assessment of channel function. Inside-out macropatch recordings of COSm6 cells expressing Kir6.2 E322K revealed a marked difference in ATP sensitivity compared with WT channels (Fig. 5 A–C), consistent with previous findings in nonmammalian cells (17). The E322K channel is significantly less sensitive to inhibition by ATP (K_1/2 for ATP inhibition is shifted from 15 μM in WT to 62 μM in the mutant, Fig. 5C) compared to WT channels. In contrast, charge neutralization of E322 (Kir6.2 E322A, retains ankyrin-binding activity; Fig. 4B) has no significant effect on Kir6.2 ATP sensitivity or targeting (N.S.; Fig. 5C, n = 21 WT, n = 13 E322A). Consistent with decreased ATP sensitivity (channel gain-of-function), COSm6 cells expressing Kir6.2 E322K displayed increased ⁸⁶Rb⁺ efflux under basal conditions (simulates physiologic ATP conditions; Fig. S6C). These data are consistent with E322K as a Kir6.2 gating gain-of-function mutation and support dual roles for the Kir6.2 C-terminal ABM in both targeting to the membrane and regulation of channel activity by ATP.

Ankyrin/Kir6.2 Binding Regulates I_KATP.

We next tested whether the Kir6.2 C-terminal ABM directly affects K_ATP channel gating using a peptide designed against the Kir6.2 ABM motif to competitively inhibit the ankyrin/Kir6.2 interaction. In agreement with a role of AnkB in regulating K_ATP channel biophysical activity, we observed increased K_ATP channel currents when the ABM peptide was applied to the cytoplasmic side of the channel in excised macropatches from COSm6 cells (Fig. S7 A and C). This change in activity was not observed for Kir6.2 E322K channels under identical conditions (Fig. S7 B and C). Additionally, a control peptide lacking ankyrin-binding activity (contains the E322K mutation) had no effect on WT or mutant Kir6.2 E322K channel activity (Fig. S7 A–C).

Defects in Kir6.2 E322K Cause Dual Cellular Phenotypes Resulting in Human Disease.

Interestingly, the E322K mutation results in two counteracting I_KATP phenotypes: decreased membrane localization and reduced ATP sensitivity. A severe hypoinsulinemic/hyperglycemic phenotype can result from even a mild decrease in ATP sensitivity (18). Our data demonstrate a significant reduction in ATP sensitivity (≈50 μM shift in K_1/2 of ATP binding); however, the E322K phenotype presents with hyperglycemia that is much less severe than in individuals expressing other gain-of-function mutations in Kir6.2 [fasting blood glucose of 342 mg/dL (19 mM) compared with values ranging from 403 mg/dL (22.4 mM) to 1260 mg/dL (70 mM) for other PNDM mutation carriers] (14). Mathematical modeling of our experimental data (Fig. S8 and SI Appendix) support the idea that reduced I_KATP conductance due to decreased membrane channel expression may protect the β cell from complete elimination of action potential firing (Fig. S8 E–G) that would otherwise result if the enhanced opening probability of E322K mutant channels was unopposed by reduced membrane expression. Therefore, our data suggests that the dual molecular phenotype of defects in ATP sensitivity plus abnormal trafficking (i.e., fewer channels with gain-of-function properties) results in the ultimate PNDM clinical phenotype of E322K probands. These findings are consistent with recent findings of other neonatal diabetes mutations that show loss of channel density combined with reduced ATP sensitivity (19). While the mechanism(s) underlying dual phenotypes of other neonatal diabetes mutations are still unclear, our results strongly support dysfunction in ankyrin-based pathways as the mechanism underlying the Kir6.2 E322K human disease phenotype.

Ankyrin Forms a Ternary Complex with Kir6.2 and SUR1.

Interaction of Kir6 and SUR subunits is critical for normal K_ATP channel regulation (20). In fact, mutations or genetic disruption of either Kir6 or SUR subunits resulting in a loss of subunit association, produce K_ATP channel dysfunction (21, 22). We therefore tested whether AnkB-binding affected association of Kir6.2 with SUR1. HEK293 cells (express AnkB, lack Kir6.2, and SUR1) were transfected with combinations of Kir6.2 or Kir6.2 E322K plus or minus SUR1, and the cell lysates were immunoprecipitated using Kir6.2 Ig (Fig. S9A). As expected, Kir6.2 Ig co-immunoprecipitated AnkB from cells expressing Kir6.2 alone (Fig. S9A). Likewise, Kir6.2 Ig co-immunoprecipitated SUR1 and AnkB from cells expressing both Kir6.2 with SUR1, demonstrating that Kir6.2 associates with both AnkB and SUR1 proteins. Finally, while lacking ankyrin-binding activity, Kir6.2 E322K maintained its association with SUR1, as demonstrated by the ability of Kir6.2 Ig to co-immunoprecipitate SUR1 in cells that express Kir6.2 E322K and SUR1(Fig. S9A). We performed similar experiments using AnkB Ig (Fig. S9B). Consistent with our prior data, AnkB associated with Kir6.2, but not Kir6.2 E322K(Fig. S9B). Additionally, we did not detect interaction of AnkB with SUR1 in cells expressing only SUR1(Fig. S9B). However, co-expression of SUR1 with Kir6.2, but not Kir6.2 E322K, resulted in ability of AnkB to co-immunoprecipitate the SUR1 subunit (Fig. S9B). Finally, we tested the potential of an ankyrin/Kir6.2/SUR1 ternary complex in vivo. Kir6.2 Ig co-immunoprecipitated both AnkB and SUR1 from mouse pancreas (Fig. S9 C and D). Similarly, AnkB Ig co-immunoprecipitated both Kir6.2 and SUR1 (Fig. S9 E and F). Our combined data identify a ternary protein complex required for normal Kir6.2 targeting and regulation in excitable cells. Moreover, our data strongly suggest that AnkB and SUR1 display unique binding sites on the Kir6.2 subunit. In support of these findings, other groups have reported SUR binding regions on the Kir6 N-terminal and first transmembrane domains (23).

Animals.

All mice evaluated were age-matched male littermates and were housed and fed under identical conditions.

Immunoblots and Co-Immunoprecipitations.

See SI Methods for details.

Binding Experiments.

See SI Methods for details.

Neonatal Fibroblast Isolation and Immunofluorescence.

See SI Methods for details.

Transfection.

See SI Methods for details.

Patch Clamp Electrophysiology.

See SI Methods for details.