The ATP-Sensitive K⁺ Channel ABCC8 S1369A Type 2 Diabetes Risk Variant Increases MgATPase Activity

Cultured tsA201 cells were transfected with the KCNJ11 and ABCC8 clones using the calcium phosphate precipitation technique (32). Transfected cells were identified using fluorescent optics in combination with coexpression of a green fluorescent protein plasmid (Life Technologies, Gaithersburg, MD). Macroscopic K_ATP channel recordings were then performed 48–72 h after transfection. The excised inside-out patch-clamp technique was used to measure macroscopic recombinant K_ATP channel currents in transfected tsA201 cells as described in detail previously (32). Experiments were performed at room temperature. The bath solution for recording in the absence of Mg²⁺ ions contained no MgCl₂, and 2 mmol/L EGTA were added. Beryllium fluoride (BeF₂) was prepared as a 33% stock solution dissolved in 50 mmol/L KF to produce the γ-phosphate analogs BeF_X (BeF₃⁻ and BeF₄²⁻) (29). For preparing bath solutions to be used for experiments in the presence of BeFx, 50 mmol/L KCl were replaced with 50 mmol/L KF.

BeF₂, MgATP, and Na⁺ salts of guanosine triphosphate (GTP) and guanosine diphosphate (GDP) were obtained from Sigma-Aldrich (Oakville, ON, Canada). ATP, GTP, and GDP were prepared as 100 mmol/L stocks in double deionized H₂O immediately before use.

The human K_ATP channel Kir6.2 and SUR1 subunit clones were provided by Dr. J. Bryan (Pacific Northwest Diabetes Research Institute, Seattle, WA). The E23K/S1369A variants and the Q1372A mutation were introduced into the KCNJ11 and ABCC8 cDNAs, respectively, using site-directed mutagenesis (QuikChange; Stratagene, La Jolla, CA) and confirmed by sequence analysis.

The homology model of the SUR1 NBD1 and NBD2 dimer (30), based on the prokaryotic ATP binding cassette protein crystal structure of MJ0796 (Protein Data Bank accession number 1F30) was provided by Dr. C.G. Nichols (Washington University, St. Louis, MO) and used to predict the location of key regions and residues in the NBD1/2 dimer using PyMOL software (Fig. 1).

A 681 base pair–long fragment of NBD2 (aa 1301–1528), containing the S1369 or A1369 variant, was subcloned into the pGEX-4T-1 expression vector. The recombinant plasmids were sequenced and then transformed into BL21 (DE3) cells for protein expression. The GST-NBD2 fusion proteins were detected with a monoclonal anti–GST tag antibody (1:5,000 dilution, Santa Cruz Biotechnology, Inc., Santa Cruz, CA). After purification, proteins (50 µg) were incubated at 40°C for 30 min followed by 1-h incubation at 4°C to allow homogeneous dimerization of the NBD2 monomers. MgATPase activities of NBD2 dimers were determined at 37°C by monitoring the rate of production of the fluorophore resorufin, which was coupled stoichiometrically to the formation of ADP (ADP Quest; DiscoverX, Fremont, CA). ADP-free ATP (ATP-Gold, DiscoverX) was used as substrate. Resorufin formation was measured continuously in a fluorescence microplate reader (Gemini XPS; Molecular Devices, Sunnyvale, CA) with excitation and emission wavelengths of 560 and 591 nm, respectively, and with an emission filter at 590 nm. Initial rates (v) were determined in quadruplicate at several ATP concentrations from the slopes of early pseudo-linear portions of fluorescence (relative fluorescence unit) versus time curves (SoftMax Pro software; Molecular Devices). Velocity data were fitted with the Michaelis-Menten equation to obtain K_M and V_max values.

Macroscopic K_ATP channel currents were normalized and expressed as changes in test current relative to control current. Macroscopic current analysis was performed using pClamp 10.0 (Axon Instruments, Foster City, CA) and Origin 6.0 software. Statistical significance was assessed using the unpaired Student t test or one-way ANOVA with a Bonferroni post hoc test. P < 0.05 was considered statistically significant. Data are expressed as the mean ± SEM.

Our recent data indicated that K_ATP channels containing the ABCC8 A1369 variant are less sensitive to inhibition by MgATP, whereas the intrinsic open probability at zero ATP remained unchanged (26). There are three possible explanations for these observations indicated as follows: 1) K_ATP channels containing the A1369 variant are more sensitive to activation by MgADP; 2) the Kir6.2-mediated ATP inhibition is reduced; or 3) the enzymatic MgATPase activity of the channel is increased, leading to elevated local levels of MgADP that relieve ATP inhibition. The first two possibilities are unlikely because we previously demonstrated that the MgADP stimulatory effects are unaltered in the K23/A1369 channel haplotype, and the presence of the KCNJ11 (Kir6.2) E23 or K23 variants do not change the IC₅₀ for ATP inhibition via the Kir6.2 subunit (26). Therefore, we tested the notion that the observed decrease in MgATP sensitivity in K_ATP channels containing the ABCC8 A1369 variant is caused by an increase in the intrinsic MgATPase activity of the K_ATP channel complex.

Analysis of the NBD1/2 dimer homology model (30) places residue 1369 within a β-sheet/hairpin loop that is ∼10 residues proximal to the primary MgATPase activity site 2 in the dimer (Fig. 1). Because MgATP can activate the K_ATP channel complex after being converted to MgADP through intrinsic MgATPase activity, we explored whether the presence of the A1369 variant affects K_ATP channel activity by altering the MgATPase enzymatic activity of SUR1. It is well established that the nucleotide GTP can be used as a substrate for SUR1 ATPase activity but is a very weak inhibitor of channel activity via the Kir6.2 subunit (32,33). However, GDP (similar to ADP) generated by SUR1 MgATPase activity has a stimulatory effect on K_ATP channels. By monitoring changes in K_ATP channel activity, GTP can be used to quantify MgATPase activity. To evaluate the effects of the K23 and A1369 T2D risk variants on MgATPase activity, we measured GTP-elicited current responses in recombinant human K_ATP channels containing the E23/S1369 nonrisk haplotype or the K23/A1369 T2D risk haplotype using the conventional excised inside-out patch technique. In the presence of 0.1 mmol/L ATP, GTP (1 mmol/L) stimulated K_ATP channel activity to a significantly greater extent in channels containing the K23/A1369 T2D risk haplotype compared with channels containing the nonrisk E23/S1369 haplotype (3.24 ± 0.05- vs. 1.83 ± 0.14-fold increase, respectively, P < 0.0001) (Fig. 2A, B, and E). To test whether this observed enhancement of GTP stimulation was due to the presence of either the K23 or A1369 variant alone, we expressed quasi-heterologous K_ATP channels containing E23 and A1369 and observed that GTP caused a similar enhancement of channel activity to the K23/A1369 T2D risk variant (2.99 ± 0.14-fold for E23/A1369 vs. 1.83 ± 0.14-fold increase for E23/S1369, P < 0.0001) (Fig. 2C and E). These results indicate that it is the A1369 variant (and not K23) that is responsible for the enhanced GTP stimulatory effect, suggesting that the K_ATP channel complex’s intrinsic MgATPase activity is enhanced by the presence of the A1369 variant (Fig. 2F). Homology modeling of NBD2 predicts that the S1369 residue forms a hydrogen bond with Q1372 and that the A1369 variant lacks the ability to form this bond (Fig. 1). Accordingly, we introduced the Q1372A mutation in the presence of S1369 (E23/S1369-Q1372A) to test whether the GTP stimulatory effect is similarly enhanced. The Q1372A mutation significantly enhanced the stimulatory effect of GTP compared with K_ATP channels containing E23/S1369 and Q1372 (2.42 ± 0.15- vs. 1.83 ± 0.14-fold, P < 0.0001) (Fig. 2D and E).

To provide further evidence for the A1369 variant altering MgATPase activity, we performed similar experiments in the absence of Mg²⁺ ions or in the presence of the MgATPase inhibitor BeF₂. Mg²⁺ ions are a necessary cofactor required for MgATPase activity, and removal of Mg²⁺ ions from the intracellular solution should prevent GTP-mediated stimulation of K_ATP channels if MgATPase activity is responsible. In the absence of Mg²⁺ ions, no GTP stimulatory effect was observed in K_ATP channels containing the E23/S1369 nonrisk or K23/A1369 T2D risk haplotype (1.06 ± 0.01- vs. 1.06 ± 0.02-fold increase, respectively) (Fig. 3). In the absence of Mg²⁺ ions, the IC₅₀ for ATP inhibition was not significantly different for the E23/S1369 nonrisk or K23/A1369 T2D risk haplotype (7.7 ± 2.2 vs. 7.9 ± 1.3 μmol/L) (Fig. 3D).

BeF₂ (BeF₃⁻ and BeF₄²⁻, abbreviated in text as BeF_X) is a potent inhibitor of many ATPases, including K_ATP channels, locking the ATPase in a prehydrolytic state by stable trapping of the MgADP-BeF_X complex in the catalytic sites (28,34–37). We observed that the stimulatory effect of GTP (1 mmol/L) is lost in the presence of 1 mmol/L BeF_X in K_ATP channels containing the E23/S1369 or K23/A1369 haplotype (Fig. 4A–C). If the observed GTP stimulatory effect is due to GTP hydrolysis via MgATPase activity, then this effect should be lost with the nonhydrolyzable GTP analog GMP-PNP. Indeed, GMP-PNP did not enhance currents recorded from K_ATP channels containing the E23/S1369 or K23/A1369 haplotype (Fig. 5). These results provide further evidence for the observed stimulatory effects of GTP being mediated by the intrinsic MgATPase activity of the K_ATP channel complex.

Although we previously demonstrated that the ADP sensitivity is similar in K_ATP channels containing the E23/S1369 or K23/A1369 haplotype (26), the possibility remains that the GDP-mediated stimulation may be different in these two K_ATP channel haplotypes. Accordingly, we measured the current responses of the two K_ATP channel haplotypes to 0.1 or 1 mmol/L GDP while the channels were partially inhibited by 0.1 mmol/L ATP. We observed no significant difference in the ability of GDP to stimulate K_ATP channels containing the E23/S1369 or K23/A1369 haplotype, indicating that both variants are equally sensitive to GDP (Fig. 6) (1.63 ± 0.05 vs. 1.79 ± 0.06 at 0.1 mmol/L GDP and 3.37 ± 0.2 vs. 3.13 ± 0.15 at 1 mmol/L GDP for the E23/S1369 and K23/A1369 haplotypes, respectively). These results suggest that it is not GDP sensitivity that is altered in the K23/A1369 haplotype but rather that the elevated generation of GDP from increased MgATPase activity is responsible for the observed differences in the GTP stimulatory effect (Fig. 2).

Candidate gene and genome-wide association studies indicate that T2D risk is only associated with the homozygous state, in other words, K_ATP channels containing only the K23 and A1369 variants. Therefore, we sought to determine whether K_ATP channel subunits containing a mixture of all the E23, K23, S1369, and A1369 variants would display an enhanced GTP stimulatory effect. To achieve this, we transfected tsA201 cells with equimolar amounts of the individual plasmids containing all four variants to generate the expression of quasi-heterozygous K_ATP channels. No significant difference was observed between the stimulatory effects of GTP (1 mmol/L) in K_ATP channels containing the E23/S1369 nonrisk haplotype and K_ATP channels composed of subunits containing the E23, K23, S169, and A1369 variants (E+K/S+A) (Fig. 7A–C).

The electrophysiological data presented above provide good evidence for enhanced MgATPase activity in K_ATP channels containing the A1369 variant. However, we sought further confirmation of this concept using a direct assay of MgATPase activity in GST-NBD2 fusion proteins containing the S1369 or A1369 variants. A significant increase in MgATP binding and MgATPase activity was observed in NBD2 containing the A1369 variant compared with S1369 (K_M = 82 ± 1.7 vs. 130 ± 16 µmol/L; V_max = 694 ± 37 vs. 365 ± 16 pmol ADP/min/mg, P < 0.05, n = 3) for A1369 and S1369, respectively (Fig. 7D). To provide further evidence that ATP hydrolysis is required, ATP was replaced with the nonhydrolyzable analog 5′-adenylylimidodiphosphate (AMP-PNP) in MgATPase assays using the S1369 GST-NBD2 fusion protein. As expected, the MgATPase activity of the GST-NBD2 fusion protein was abolished with the replacement of ATP with AMP-PNP (Fig. 7E). The ATP inhibitor BeF_x used in this study inhibited the MgATPase activity of the GST-NBD2 S1369 fusion protein with an IC₅₀ of 40 μmol/L (data not shown).

Previous studies investigating the precise molecular mechanism(s) that underpin the documented association of the K_ATP channel KCNJ11 K23 variant have yielded variable results (21–25). This may reflect that in these earlier studies, the K23 variant was investigated in recombinant K_ATP channels composed of subunits from several different species without controlling for A1369. Our recent work was the first to document the properties and pharmacology of human K_ATP channels that contain the K23 and A1369 T2D risk haplotype and indicated that it is the A1369 variant, but not K23, that alters the MgATP-sensing properties of the K_ATP channel (26). The underlying molecular mechanism for the reduced MgATP inhibition observed in human K_ATP channels containing the common K23/A1369 T2D risk haplotype is of obvious importance to understand the development of T2D in humans carrying two copies of this haplotype. Insights into this molecular mechanism can be gleaned from previous studies on rare heterozygous ABCC8 mutations (R1380L and R1380C) that cause neonatal diabetes via increases in MgATPase activity (29). Homology modeling places R1380 within the predicted major MgATPase catalytic site 2 (Fig. 1), increasing MgATPase activity and rendering the K_ATP channel markedly less sensitive to MgATP inhibition compared with free ATP inhibition, leading to severely impaired insulin secretion in affected patients (29). Results from the current study now demonstrate that the A1369 variant also increases MgATPase activity as measured in intact K_ATP channels using electrophysiological techniques (Figs. 2–5) and in a direct assay of NBD2 MgATPase activity (Fig. 7D and E). Although the underlying mechanism of increased MgATPase activity may be the same for rare monogenic mutations at residue 1380 and for the A1369 variant (as demonstrated in this study), the clinical phenotype is strikingly different: overt neonatal diabetes versus T2D risk. This may be explained by the difference in the magnitude of channel activation at physiological millimolar ATP concentrations. Monogenic mutations at residue 1380 result in large increases in K_ATP channel activity at millimolar MgATP (29), effectively clamping β-cells in an electrically quiescent state preventing insulin release. Indeed, inhibition of overactive K_ATP channels with sulfonylurea therapy has been shown to be an effective therapy in many cases of neonatal diabetes resulting from K_ATP channel activation mutations (38). In contrast, the A1369 variant may subtly increase the K_ATP channel’s intrinsic MgATPase activity, leading only to modest increases in K_ATP channel activity at millimolar ATP that would oppose, but not prevent, insulin secretion and thus increase the risk for T2D without precipitating overt neonatal diabetes. This latter concept is supported by our previous observation that the MgATP sensitivity of K_ATP channels containing the A1369 variant is slightly reduced, leading only to small but measurable increases in single K_ATP channel activity at millimolar ATP (26). From a molecular perspective, analysis of the SUR1 NBD1/2 dimer homology model (30) places the 1369 residue adjacent to a hairpin loop in the β-sheet that forms the backbone of the NBD2 portion of the major MgATPase catalytic site 2 (Fig. 1). Therefore, substitution of Ser to an Ala at residue 1369 may alter the secondary or tertiary structure of this region, resulting in an enhanced MgATPase activity of the NBD1/2 dimer. Indeed, our results provide strong evidence for this concept because the increased GTP stimulatory effect in channels containing the A1369 variant (Fig. 2) is not due to altered GDP sensitivity (Fig. 5) but is lost in the absence of Mg²⁺ ions in the presence of the ATPase inhibitor BeF_X or with the nonhydrolyzable GTP analog GMP-PNP (Figs. 3–5). Furthermore, an assay of MgATPase activity of GST-NBD2 fusion proteins containing the S1369 or A1369 variant also confirmed that the A1369 directly increases MgATPase activity (Fig. 7D and E). It should be noted that a previous study reported increased GDP sensitivity in K_ATP channels containing the K23 variant, although K_ATP channel subunit clones from several species were used, and the importance of the A1369 variant was unknown at the time and thus not controlled for (23).

Although it is not possible to determine conclusively the exact structural consequences of the S1369A substitution from analysis of homology models alone, it is nevertheless possible to speculate on the possible mechanism(s). The side chain of Ser at residue 1369 is modeled to point toward the hairpin loop’s interior with the terminal hydroxyl group hydrogen bonding with the NH₂-terminal group on the side chain of Q1372 (Fig. 1). This hydrogen bond would likely constrain the structure of the hairpin loop and the β-sheet proximal to the major MgATPase activity site 2 (Fig. 1). Substitution of the Ser with an Ala at position 1369 (A1369) removes only the side-chain terminal hydroxyl group that would prevent any hydrogen bond formation with Q1372. This would effectively reduce the structural constraint on the hairpin loop, allowing a greater movement in this region resulting in the observed increase in MgATPase activity. This concept is supported further by our data showing that introduction of the Q1372A mutation into channels containing the S1369 variant enhanced the GTP stimulatory effect (Fig. 2D and E). We therefore propose that the A1369 T2D risk variant causes subtle changes in the conformation of the dimer structure proximal to the NBD2 region that forms part of the MgATPase catalytic site 2.

The results obtained in this study further confirm that it is the A1369 variant and not K23 that alters the observed GTP stimulatory effect via increased MgATPase activity because the enhanced GTP stimulatory effect was similar in K_ATP channels containing E23 or K23 (Fig. 2C and E). These results are consistent with our previous findings indicating that the A1369 variant, and not the K23 variant, is responsible for the observed reduction in MgATP sensitivity (26). Our results further highlight that future mechanistic studies on the K23 variant should include the A1369 variant (27).

Previous candidate gene and genome-wide association studies have shown that the association of the KCNJ11 K23 and ABCC8 A1369 variants with T2D is present only in the homozygous state (37,39). These clinical findings suggest that all eight subunits in the K_ATP channel hetero-octameric complex must contain these variants to increase T2D susceptibility. Our results support this notion because the enhanced GTP stimulatory effect afforded by the A1369 variant was not preserved in K_ATP channel subunits containing a combination of the E23, K23, S1369, and A1369 variants (Fig. 6).

From a physiological perspective, K_ATP channels encoded by the KCNJ11 and ABCC8 genes are expressed in several different tissues. The physiological consequence of the K23/A1369 T2D risk haplotype regarding enhanced MgATPase activity would be to suppress first-phase K_ATP channel–dependent insulin secretion. Although earlier studies on homozygous carriers of the K23 variant did not observe any reduction in insulin secretion (16,40), a recent clinical study demonstrated an impaired insulin secretory response in humans homozygous for the K23 variant (25). In addition, K_ATP channels are involved in the appropriate α-cell glucagon secretory response (41), and the K23/A1369 T2D risk haplotype may also deleteriously affect glucagon secretion. Subtle alterations in CNS K_ATP channel activity may also affect the regulation of whole-body glucose homeostasis via alterations in the excitability of hypothalamic neurons. Despite no overt clinical phenotype, subjects carrying two copies of the K23/A1369 T2D risk haplotype may be at an increased risk of developing diabetes in combination with other genetic variations or environmental factors, such as diet and lifestyle (42).

In summary, our results provide evidence that the ABCC8 A1369 variant plays a major role in determining the properties of K_ATP channels containing the K23/A1369 T2D risk haplotype and describe a plausible molecular mechanism by which ∼10% of the general population homozygous for this haplotype may be at increased risk for developing T2D.

D.S. was supported by an Alberta Innovates-Health Solutions trainee award. A.H. was supported by a grant from the Canadian Institutes of Health Research. P.E.L. was supported by a grant from the Canadian Institutes of Health Research, received salary support as an Alberta Innovates-Health Solutions Senior Scholar, and holds the Dr. Charles A. Allard Chair in Diabetes Research.

M.F. and M.R. performed experiments, analyzed data, and contributed to manuscript writing and discussion. C.C. constructed mutations and contributed to discussion. D.S. performed experiments, analyzed data, and contributed to manuscript writing and discussion. A.H. designed and optimized the direct ATPase activity experiments and contributed to data analysis. P.E.L. designed experiments, wrote the manuscript, and is the guarantor of the article.

No potential conflicts of interest relevant to this article were reported.

Parts of this study were presented in abstract form at the 70th Scientific Sessions of the American Diabetes Association, Orlando, Florida, 25–29 June 2010.

The authors thank Dr. Joanne Lemieux (University of Alberta) for constructive discussions on structural biology.