Molecular basis of Kir6.2 mutations associated with neonatal diabetes or neonatal diabetes plus neurological features

Peter Proks*†, Jennifer F. Antcliff*†, Jon Lippiat*, Anna L. Gloyn‡§, Andrew T. Hattersley‡, and Frances M. Ashcroft*¶

*University Laboratory of , Oxford University, Oxford OX1 3PT, United Kingdom; and ‡Institute of Biomedical and Clinical , Peninsula Medical School, Exeter EX2 5DW, United Kingdom

Edited by Charles F. Stevens, The Salk Institute for Biological Studies, La Jolla, CA, and approved November 1, 2004 (received for review July 2, 2004) Inwardly rectifying potassium channels (Kir channels) control cell opmental delay, muscle weakness, and epilepsy (4–6). We refer membrane K؉ fluxes and electrical signaling in diverse cell types. to this as severe disease. Of the 27 patients with Kir6.2 mutations Heterozygous mutations in the human Kir6.2 gene (KCNJ11), the studied to date, Ϸ30% had neurological problems. pore-forming subunit of the ATP-sensitive (KATP) channel, cause Fig. 1B maps some of the mutations associated with PNDM permanent neonatal diabetes mellitus (PNDM). For some muta- onto a homology model of Kir6.2 (7). Two mutations in the same tions, PNDM is accompanied by marked developmental delay, residue (R201H and R201C), which lies in the putative ATP- muscle weakness, and epilepsy (severe disease). To determine the binding site (7, 8), cause mild disease. In contrast, the mutations molecular basis of these different phenotypes, we expressed Q52R, V59M, and V59G, which are associated with severe wild-type or mutant (R201C, Q52R, or V59G) Kir6.2͞sulfonylurea disease, lie some distance from the ATP-binding site, within the receptor 1 channels in Xenopus oocytes. All mutations increased N-terminal region of the protein; moreover, V59 lies within the resting whole-cell KATP currents by reducing channel inhibition by ‘‘slide helix,’’ a domain postulated to be involved in the opening ATP, but, in the simulated heterozygous state, mutations causing and closing (gating) of Kir channels (7). PNDM alone (R201C) produced smaller KATP currents and less Previous work has shown that mutation of R201 to histidine

change in ATP sensitivity than mutations associated with severe PHYSIOLOGY decreases the ATP sensitivity of the KATP channel (4). However, disease (Q52R and V59G). This finding suggests that increased KATP all patients identified to date are heterozygotes, and no obvious currents hyperpolarize pancreatic beta cells and impair insulin loss of ATP sensitivity was observed when the heterozygous state secretion, whereas larger KATP currents are required to influence was simulated by coinjection of R201H and wild-type Kir6.2 (4). extrapancreatic cell function. We found that mutations causing The cause of PNDM was therefore unclear. Furthermore, the PNDM alone impair ATP sensitivity directly (at the binding site), reason why some mutations are associated with muscle weak- whereas those associated with severe disease act indirectly by ness, developmental delay, and epilepsy has not been addressed. biasing the channel conformation toward the open state. The In this paper, we compare the functional effects of mutations effect of the mutation on ATP sensitivity in the heterozygous state associated with severe disease (Q52R and V59G) with those that reflects the different contributions of a single subunit in the Kir6.2 cause mild disease (R201C and R201H). tetramer to ATP inhibition and to the energy of the open state. Our results also show that mutations in the slide helix of Kir6.2 (V59G) Materials and Methods influence the channel kinetics, providing evidence that this domain Mutagenesis and RNA Preparation. Human Kir6.2 (GenBank ac- is involved in Kir channel gating, and suggest that the efficacy of cession no. NM000525) and rat SUR1 (GenBank accession no. sulfonylurea therapy in PNDM may vary with genotype. L40624) were used in this study. Because human Kir6.2 contains ͉ ͉ ͉ two common polymorphisms, E23K and I377V, we used the potassium channel KATP channel KCNJ11 ATP-binding site most common allele at these residues (i.e., E at position 23 and I at position 377). Site-directed mutagenesis of Kir6.2 was TP-sensitive potassium channels (KATP channels) control performed with the QuikChange XL system (Stratagene). Wild- Aelectrical signaling in diverse cell types by coupling cellular type and mutant mRNAs were expressed in Xenopus laevis metabolism to potassium movement across cell membranes. In oocytes as described in ref. 9. Currents were recorded 1–3 days pancreatic beta cells, KATP channels link changes in blood after injection with 0.1 ng of wild-type or mutant Kir6.2 mRNA glucose concentration to insulin secretion (1); in brain, they and Ϸ2 ng of SUR1 mRNA (giving a 1:20 ratio). For each batch contribute to glucose sensing and seizure protection, in cardiac of oocytes, all mutations were injected to enable direct compar- muscle they protect against ischemic stress; and in skeletal ison of their effects. muscle, they influence tone (1). KATP channels comprise four pore-forming inwardly rectifying potassium channel (Kir chan- Two-Electrode Voltage-Clamp Studies. Whole-cell currents were nel) 6.2 subunits and four regulatory sulfonylurea receptor recorded from intact oocytes in response to voltage steps of Ϯ20 (SUR) subunits (2). The SUR1 isoform is found in beta cells. mV from a holding potential of Ϫ10 mV, and they were filtered ATP binding to Kir6.2 closes the channel, whereas interaction of at 1 kHz and digitized at 4 kHz. Oocytes were constantly Mg-nucleotides with SUR1 opens it (3). Consequently, increased metabolism leads to channel closure, membrane depolarization, and electrical activity and, conversely, metabolic inhibition This paper was submitted directly (Track II) to the PNAS office. opens KATP channels and suppresses electrical activity (see Abbreviations: Kir channel, inwardly rectifying potassium channel; SUR1, sulfonylurea Fig. 1A). receptor 1; KATP channel; ATP-sensitive potassium channel; PNDM, permanent neonatal Mutations in the human Kir6.2 gene (KCNJ11) were recently diabetes mellitus. identified that severely impair insulin secretion and lead to †P.P. and J.F.A. contributed equally to this work. permanent neonatal diabetes mellitus (PNDM) (4–6). One class §Present address: Oxford Centre for Diabetes, Endocrinology and Metabolism, Churchill of mutations produces only PNDM, and we refer to this as Hospital, Oxford OX3 7LJ, United Kingdom. ‘‘mild’’ disease. The other is associated with a more severe ¶To whom correspondence should be addressed. E-mail: [email protected]. phenotype in which PNDM is accompanied by marked devel- © 2004 by The National Academy of of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0404756101 PNAS ͉ December 14, 2004 ͉ vol. 101 ͉ no. 50 ͉ 17539–17544 Downloaded by guest on September 25, 2021 perfused at 20–24°C with solution containing 90 mM KCl, 1 mM The values of IC50,i and PO,i were estimated from the following MgCl2, 1.8 mM CaCl2, and 5 mM Hepes, pH 7.4 with KOH. equations (for further details see Supporting Text, which is Metabolic inhibition was produced by 3 mM Na-azide. Tolbut- published as supporting information on the PNAS web site): amide (0.5 mM) was tested on all oocytes to confirm that the ϭ ͑͞ ϩ ␭i⅐ ͒ PO,i PMAX 1 AS [4] observed current flowed through KATP channels. and Patch-Clamp Studies. Macroscopic currents were recorded from ϭ ⅐ ͑͞␭i⅐ ͒ giant excised inside-out patches by using the patch-clamp tech- IC50,i IC50,WT PO,WT PO,i , [5] nique. Currents were filtered at 0.15 kHz and digitized at 0.5 kHz. The pipette solution contained 140 mM KCl, 1.2 mM where PMAX denotes the maximum (intraburst) open probability, MgCl2, 2.6 mM CaCl2, and 10 mM Hepes, pH 7.4 with KOH. The AS denotes the contribution of interburst closures to the channel internal (bath) solution contained 107 mM KCl, 1 mM K SO , open probability, and the parameter ␭ is the difference in the 2 4 Ͻ Ͻ 10 mM EGTA, and 10 mM Hepes, pH 7.2 with KOH, and contribution of the individual channel subunit j (1 j 4) to the nucleotides as indicated). All experiments were carried out in the energy of the open state (GO, j) between the mutant (GO,M, j) and absence of Mg2ϩ to prevent the activating effects of nucleotides the wild-type (GO,WT, j) channel: mediated by SUR1 (3). Rapid exchange of internal solutions was ␭ ϭ exp((G Ϫ G )͞RT). [6] achieved with a sewer pipe system. Experiments were conducted O,M, j O,WT, j

at 20–24°C. The macroscopic slope conductance, G, was mea- The values of ␭ (0.62), PMAX (0.86), and AS (0.62) were calcu- sured between Ϫ20 and Ϫ100 mV (9). ATP concentration– lated from Eqs. 4–6 by using measured parameters for wild-type response curves were fit with the Hill equation: G͞Gc ϭ 1͞(1 ϩ and homomeric mutant channels; the values of IC50 and PO for h ([ATP]͞IC50) ), where [ATP] is the ATP concentration, IC50 is all individual species of heteromeric channels were determined the ATP concentration at which inhibition is half maximal, and from Eqs. 4 and 5 and for the heterozygous channel population h is the slope factor (Hill coefficient). To control for possible from Eqs. 1 and 2. rundown, Gc was taken as the mean of the conductance in control solution before and after ATP application. Results and Discussion Single-channel currents were recorded from small inside-out To examine the functional differences between the two classes of membrane patches at Ϫ60 mV, filtered at 5 kHz, and sampled at mutations, we expressed wild-type or mutant human Kir6.2, 20 kHz. Open probability was determined (in the absence of together with SUR1, in Xenopus oocytes (9). We first compared ATP) from current records of Ϸ1-min duration shortly after the effect of mutations on the properties of homomeric channels patch excision as I͞iN, where I is the mean macroscopic current, composed of a single type of Kir6.2 subunit. We then simulated i is the single-channel current, and N is the number of channels heterozygosity by coxpressing a 1:1 mixture of wild-type and mutant Kir6.2 mRNAs with SUR1, which is expected to result (taken as the maximum number of open levels). For heterozy- in a mixed population of channels containing different ratios of gous channels (Fig. 1C), the mean P represents the average of O wild-type and mutant subunits (Fig. 1C): We refer to this ͗P ͘ all types of channels ( O ). population as heterozygous channels. Data Analysis of Heterozygous Channels. For analysis of heterozy- Unlike Wild-Type Channels, Mutant KATP Channels Are Not Closed by gous channels, we assumed independent mixing of wild-type and Resting ATP Levels. When expressed in Xenopus oocytes, wild-type mutant subunits. Thus, the relative numbers of KATP channels KATP channels are normally closed but they are activated by with different subunit compositions (Fig. 1C) were expected to metabolic inhibitors, such as azide, which lower intracellular follow the binomial distribution as described in refs. 2 and 10. ATP (Fig. 2). In contrast, significant resting whole-cell Kϩ The values of IC50,H and PO,H (the intrinsic open probability; i.e., currents are present in oocytes expressing homomeric Q52R, the open probability in the unliganded state) for the heterozy- V59G, or R201C (homQ52R, homV59G, and homR201C, re- gous channel population can then be obtained from the follow- spectively) (Fig. 2 and Table 1). This result suggests that ing equations: metabolism causes less block of mutant KATP channels than wild-type channels. Whole-cell homV59G and homQ52R cur- ͑͞ ϩ ͓ ͔͞ ͒ 16 1 ATP IC50,H rents were not increased further by azide application, as if these channels are fully activated at rest and insensitive to metaboli- 4 cally induced changes in nucleotides. However, homR201C ϭ ͸ ͑͑͞ Ϫ ͒ ͑ ϩ ͓ ͔͞ ͒͒ 4! 4 i ! i! 1 ATP IC50,i [1] channels were enhanced by azide, suggesting that they are more iϭ0 ATP-sensitive than homV59G and homQ52R channels. For heterozygous channels (hetV59G, hetQ52R, and hetR201C), and hetV59G and hetQ52R resting whole-cell currents were of similar amplitude, constituting Ϸ65% of their homomeric values, 4 but were far larger than wild-type currents (Fig. 2B). These ⅐ ϭ ͸ ͑͑͞ Ϫ ͒ ͒ 16 PO,H 4! PO,i 4 i ! i! , [2] resting currents were further activated by azide, reaching a final iϭ0 amplitude close to that of both homomeric and wild-type channels. Resting hetR201C currents were smaller than Ͻ where the index, i, denotes the number of mutant subunits (0 hetQ52R and hetV59G currents but still significantly larger than Ͻ i 4) and IC50,i and PO,i stand for the IC50 and PO of channels wild-type (Fig. 2 and Table 1). These differences in resting containing i mutant subunits. Assuming that the energy of the current probably account for the difference between the mild channel open state, O, is simply the sum of the contributions of and severe forms of the disease because they may be expected to the four individual subunits in the channel (GO,i): hyperpolarize the patients’ cells and thereby influence electrical activity and cell function to differing extents. 4 ϭ ͸ GO GO, j. [3] The Molecular Basis of Disease Severity. To explore the molecular jϭ1 basis of the different metabolic sensitivities, we examined the

17540 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0404756101 Proks et al. Downloaded by guest on September 25, 2021 PHYSIOLOGY

Fig. 1. Functional role of KATP channels and location of mutations. (A) Schematic of how metabolic regulation of KATP channel activity sets the cell membrane potential. (B) Homology model of Kir6.2 (7), indicating the loca- Fig. 2. Effects of mutations on whole-cell KATP currents. (A) Whole-cell tion of residues R201, Q52R, and V59G associated with neonatal diabetes. For currents recorded from Xenopus oocytes coexpressing SUR1 and either wild- clarity, only two subunits are shown, and the intracellular and transmembrane type or mutant Kir6.2, as indicated, in response to voltage steps of Ϯ20 mV ϩ domains are from separate subunits. ATP (black) is shown in its binding site. (C) from a holding potential of Ϫ10 mV in 90 mM K . The start of azide applica- Schematic of the mixture of channels with the different subunit compositions tion is indicated by arrows. The initial decrease in KATP current is due to a small expected when wild-type and mutant Kir6.2 are coexpressed (as in the het- block by azide that is mediated by an interaction with SUR1. (B) Mean erozygous state). The relative numbers of the channel types expected if steady-state whole-cell currents evoked by a voltage step from Ϫ10 to Ϫ30 mV wild-type and mutant subunits segregate independently (i.e., follow a bino- before (control) and after application of 3 mM azide. The number of oocytes mial distribution) are indicated above the figure. is 8–12 in each case.

ability of ATP to block wild-type and mutant channels in human Kir6.2 clone. We found that ATP produced a half- inside-out patches (Fig. 3). All homomeric mutant channels were maximal block of homomeric R201H channels at 298 Ϯ 25 ␮M substantially less sensitive to intracellular ATP than wild type, (n ϭ 5), and of the heterozygous channel population at 12.5 Ϯ the order of potency being WT Ͼ Q52R Ͼ R201C Ͼ V59G 1.1 ␮M(n ϭ 5). The latter value is close to that found for (Table 1). The reduced ATP sensitivity is expected to contribute hetR201C, which also causes mild disease, and significantly less to the larger resting whole-cell KATP currents. Although all than that found for mutations associated with severe disease. homomeric mutant channels showed markedly reduced ATP In pancreatic beta cells, an increase in KATP current will sensitivity, significant differences were observed for heterozy- hyperpolarize the plasma membrane and suppress electrical gous channels (Fig. 3B). In particular, hetQ52R and hetV59G activity, which will lead to a reduction in insulin secretion and, channels were half-maximally blocked by ATP concentrations of thereby, diabetes (11). Although the ATP concentration causing 20–30 ␮M, whereas the ATP sensitivity of hetR201C channels half-maximal block of hetR201C channels is only slightly greater was higher (11 ␮M) (Table 1). The different ATP sensitivities of than that found for wild-type channels (11 ␮M vs. 7 ␮M) (Table the heterozygous channels parallel the variation in resting whole- 1), the magnitude of the relative current at 0.1 mM ATP is cell currents observed for these channels in oocytes and underlie Ϸ2.5-fold larger (0.1 Ϯ 0.02, n ϭ 5, compared with 0.04 Ϯ 0.01, the difference between the mild and severe forms of disease. n ϭ 6, respectively) (Fig. 3B). Because [ATP]i is unlikely to fall In previous studies using the mouse Kir6.2 clone, we were below 0.1 mM in beta cells, even when extracellular glucose is unable to detect a significant change in the ATP sensitivity of the low (12), the difference in wild-type and mutant KATP current at heterozygous channel population with the R201H mutation (4). this (and higher) ATP concentration probably explains why the We therefore tested the effect of the R201H mutation in the R201C and R201H mutations result in neonatal diabetes.

Proks et al. PNAS ͉ December 14, 2004 ͉ vol. 101 ͉ no. 50 ͉ 17541 Downloaded by guest on September 25, 2021 Table 1. Macroscopic and single-channel properties of PNDM mutations

Mutations Irest, ␮A(n)IC50 ATP, ␮M(n) PO(0) (n) i,pA(n)

Wild-type 0.07 Ϯ 0.01 (8) 7.0 Ϯ 1.1 (6) 0.53 Ϯ 0.02 (8) 4.1 Ϯ 0.2 (3) homQ52R 5.49 Ϯ 0.47 (10)* 84 Ϯ 12 (6)* 0.83 Ϯ 0.01 (8)* 4.2 Ϯ 0.2 (6)† homV59G 6.42 Ϯ 0.24 (10)* 7,400 Ϯ 1,500 (6)* 0.83 Ϯ 0.01 (8)* 4.0 Ϯ 0.1 (6)† homR201C 3.73 Ϯ 0.12 (12)* 106 Ϯ 12 (6)* 0.60 Ϯ 0.03 (9)† 3.9 Ϯ 0.3 (6)† hetQ52R 3.66 Ϯ 0.39 (9)* 23 Ϯ 3 (6)* 0.70 Ϯ 0.03 (9)* ND hetV59G 4.19 Ϯ 0.32 (9)* 26 Ϯ 6 (6)‡ 0.70 Ϯ 0.03 (9)* ND hetR201C 1.22 Ϯ 0.13 (12)* 11 Ϯ 2 (5)§ ND ND

Mean values of resting whole-cell KATP current (Irest), ATP concentration producing half-maximal block of the channel (IC50), intrinsic open probability (i.e., in the absence of ATP) PO(0), and single channel current at Ϫ60 mV (i). Statistical significance against wild-type is indicated: *, P Ͻ 0.001; ‡, P Ͻ 0.01; §, P Ͻ 0.05; †, not significant. ND, not determined.

Accumulating evidence implicates small changes in KATP with reduced ATP sensitivity, which also have a normal current with impaired insulin secretion (11). In particular, a ECG (29). 4-fold reduction in the ATP sensitivity of Kir6.2 produces severe Our results suggest that, in addition to their neurological neonatal diabetes in transgenic mice (13). Mutations in meta- features, patients with severe mutations might have greater bolic genes, such as glucokinase (GCK), also result in a larger impairment of beta cell function and thus more severe diabetes KATP current (14), reduced insulin release (15) at a given glucose than those with mild mutations. Because subjects with mild concentration, and maturity onset diabetes of the young in mutations, such as R201H, do not show any insulin increment humans (16, 17). It is also pertinent that a common polymor- with i.v. glucose or glucagon (4), it would not be possible to phism (E23K) in KCNJ11 is associated with an increased risk of detect a more severe defect. However, consistent with this idea, in humans (18). The functional effects of this R201H subjects do produce insulin in response to sulfonylureas polymorphism remain controversial (19, 20), but our finding that (4, 6), but no insulin increment was seen in either of the two neonatal diabetes (a far more dramatic phenotype) is produced patients with neurological features who have been tested (E. R. by only a subtle shift in the ATP concentration–inhibition curve Pearson and A. T. Hattersley, personal communication). suggests that the effects of a Kir6.2 gene variant predisposing to diabetes in later life may be hard to discern experimentally, Effects of PNDM Mutations on Single-Channel Kinetics. Why does the particularly in the heterozygous state. Q52R mutation cause a greater shift in the ATP sensitivity of All mutations cause neonatal diabetes, but only some of them the heterozygous channel than the R201C mutation, despite the are also associated with developmental delay, muscle weakness, similar ATP sensitivities of the homomeric channels? To answer and epilepsy (4–6). Our results show that disease severity is this question, we recorded single-channel currents in inside-out correlated with the extent of KATP channel block by ATP and membrane patches in ATP-free solution, where the intrinsic with the magnitude of the whole-cell KATP current under resting gating of the channel can be assessed. Fig. 4 shows that the open conditions. Kir6.2 is not expressed only in the beta cell; it is also probabilities of the mutant channels in the unliganded state, found in other endocrine cells, skeletal muscle, cardiac muscle, PO(0), were strikingly different. Thus, the PO(0) of homomeric and neurones throughout the brain (21–23). In many of these channels containing the R201C mutation was not significantly tissues, KATP channels are normally silent and open only under different from wild-type, whereas that of homQ52R and metabolic stress. Thus, a greater reduction in ATP sensitivity is homV59G channels was substantially greater (Table 1). It is well required to increase the resting KATP current sufficiently to documented that an increase in PO(0) reduces the ability of ATP influence electrical activity. This may explain why mutations that to close the KATP channel (30, 31). Thus, mutations at residue produce a small decrease in the ATP sensitivity of heterozygous 201, which lies within the putative ATP-binding site (8), probably channels (R201C and R201H) result in neonatal diabetes alone, act by reducing ATP binding per se, whereas the Q52R and V59G whereas those causing a greater reduction (Q52R and V59G) are mutations appear to decrease ATP sensitivity indirectly, by associated with severe disease. favoring the open conformation of the channel. Because V59G Kir6.2 is strongly expressed within the hippocampus, cerebel- channels have a much lower ATP sensitivity than Q52R chan- lum, and basal ganglia and at lower levels in cortex (23, 24). nels, despite a similar PO(0), the V59G mutation is also likely to Further studies at the organ͞whole-animal level are needed to affect ATP binding and͞or the mechanism by which ATP explain precisely how the severe mutations in Kir6.2 lead to binding is transduced into channel closure. The data are also epilepsy, developmental delay, muscle weakness, and dysmor- consistent with a role for the slide helix, within which residue 59 phic features. However, clonic-tonic epilepsy, like that found in resides, in Kir channel gating. severe disease, is usually associated with enhanced cortical or Lipids, such as PIP2, increase the PO(0) and reduce the ATP hippocampal activity (25). In our case, it is most simply explained sensitivity of the KATP channel (32). To determine whether the as a result of overactivity of KATP channels in GABAergic V59G mutation affects PO(0) directly or indirectly by means of interneurones (26), which results in a reduced inhibitory tone an increase in PIP2 binding, we used neomycin, a polycation that and a predisposition to seizure. Consistent with this idea, KATP binds to PIP2, closes KATP channels and has been used to channels in the hippocampus are expressed in most inhibitory evaluate PIP2 affinity (33). In 3͞3 patches, 100 ␮M neomycin did interneurones but only in a minority of excitatory pyramidal not alter Kir6.2-V59G͞SUR1 currents over 5–10 min (data not neurones (27), and the KATP channel opener diazoxide inhibits shown). This finding is consistent with the idea that the V59G firing of interneurones but not pyramidal cells (28). The ob- mutation affects PO(0) directly, rendering gating insensitive to served muscle weakness could be of neurological or muscle PIP2 modulation. origin, because Kir6.2 is expressed in both skeletal muscle and Given that the KATP channel pore is composed of four Kir6.2 peripheral nerve. Interestingly, although Kir6.2 is expressed in subunits (2), in the heterozygous state most channels will contain the heart, no marked effects were observed in the ECG (4), a mixture of wild-type and mutant subunits (Fig. 1C). Because which agrees with studies on transgenic mice expressing Kir6.2 ATP binding to a single subunit is sufficient to close the channel

17542 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0404756101 Proks et al. Downloaded by guest on September 25, 2021 Fig. 4. Effects of mutations on single-channel currents. Single KATP channel currents were recorded at Ϫ60mV from inside-out patches excised from oocytes coinjected with mRNAs encoding SUR1 plus either wild-type or mu- tant Kir6.2, as indicated.

(10), channels will exhibit a substantially reduced ATP sensitiv- ity only when all four subunits are mutant (i.e., in Ϸ1͞16th of channels in the heterozygous state, if the number of mutant subunits in the tetramer is binomially distributed). This finding explains why mutations in the ATP-binding site, like R201C,

cause only a small shift in the ATP concentration–inhibition PHYSIOLOGY curve of heterozygous channels. An empirical indication of the extent of intrinsic channel opening in the mixed population of channels that is expected when wild-type and mutant cDNAs are coexpressed is given by the mean PO(0) (͗PO(0)͘; see Materials and Methods). As Table 1 shows, ͗PO(0)͘ was significantly higher for hetQ52R and hetV59G channels than wild type. These mutations therefore reduce the free energy of the open state in the heterozygous channel population. We simulated an ensemble of Kir6.2 tet- ramers containing a binomial mix of wild-type and mutant subunits exhibiting free energy additivity in their effects on gating (see Materials and Methods and Supporting Text). The ͗PO(0)͘ of hetQ52R channels is predicted to be 0.73, in excellent Fig. 3. Effects of mutations on KATP channel ATP sensitivity. (A)KATP current recorded in response a succession of voltage ramps from Ϫ110 to agreement with the experimental observation (0.70) (Table 1), ϩ100 mV to an inside-out patch excised from a Xenopus oocyte coexpress- and the predicted half-maximal inhibitory [ATP] is 22 ␮M ing SUR1 and either wild-type or mutant Kir6.2, as indicated. The dotted (observed 23 ␮M). Thus, the lower ATP sensitivity of hetQ52R line indicates the zero current level. ATP was applied to the intracellular channels is consistent with an additive energetic influence of membrane face as indicated. (B)(Upper Left) Mean relationship between mutant and wild-type subunits on channel gating and, thereby, [ATP] and KATP conductance, G, expressed relative to the conductance in ATP sensitivity. In the case of the V59G channels, simulations ͞ ϭ the absence of nucleotide, Gc, for Kir6.2 SUR1 (open blue circles, n 6), suggest that the mutation must affect both gating and ATP- and heterozygous (filled red circles, n ϭ 6) and homomeric (filled black ͞ squares, n ϭ 6) Kir6.2-R201C͞SUR1 channels. The smooth curves are the binding transduction and indicate that the latter effect is largely suppressed in heterozygous channels containing wild-type sub- best fit to the Hill equation. For wild-type, IC50 ϭ 6.6 ␮M and h ϭ 1.1. For heterozygous R201C, IC50 ϭ 10.4 ␮M and h ϭ 1.0. For homomeric R201C, units (as is the case for hetR201C). These findings explain why IC50 ϭ 102 ␮M and h ϭ 1.3. (Upper Right) Mean relationship between [ATP] hetQ52R and hetV59G channels have similar ATP sensitivities and KATP conductance expressed relative to the conductance in the absence despite the very different ATP sensitivities of the homomeric of nucleotide for Kir6.2͞SUR1 (open blue circles, n ϭ 6) and heterozygous channels. (red filled circles, n ϭ 6) and homomeric (black filled squares, n ϭ 6) The results we have described thus far predict that the second ͞ Kir6.2-V59G SUR1 channels. The smooth curves are the best fit to a mod- mutation at V59 (V59M), which is also associated with severe ified Hill equation containing 1͞16 of homomeric channels. For wild-type, disease, will have an increased intrinsic open probability, IC50 ϭ 6.6 ␮M and h ϭ 1.1. For heterozygous V59G, IC50 ϭ 26 ␮M and h ϭ 1.18. For homomeric V59G, IC50 ϭ 8.1 mM and h ϭ 0.75. (Lower Left) Mean whereas the R201H mutation, which, like R201C, causes PNDM relationship between [ATP] and KATP conductance expressed relative to the alone, will not. Consistent with this idea, the PO(0) of homV59M conductance in the absence of nucleotide for Kir6.2͞SUR1 (open blue channels was 0.83 Ϯ 0.1 (n ϭ 5), significantly different from circles, n ϭ 6) and heterozygous (filled red circles, n ϭ 6) and homomeric wild-type (P Ͼ 0.001), whereas there was no significant differ- (filled black squares, n ϭ 6) Kir6.2-Q52R͞SUR1 channels. The smooth curves ence in the PO(0) between wild-type and R201H channels. are the best fit to the Hill equation. For wild-type, IC50 ϭ 6.6 ␮M and h ϭ 1.1. For heterozygous Q52, IC ϭ 21 ␮M and h ϭ 1.2. For homomeric Q52R, 50 Mutations May Differentially Alter Sulfonylurea Efficacy. Our finding IC ϭ 83 ␮M and h ϭ 1.7. (Lower Right) Mean relationship between [ATP] 50 that Kir6.2 mutations produce neonatal diabetes by different and KATP conductance expressed relative to the conductance in the absence of nucleotide for wild type (open blue circles) and hetR201C (filled red molecular mechanisms may have implications for therapy. We circles), hetQ52R (filled black squares), and hetV59G (filled green hexa- observed that 500 ␮M tolbutamide blocked azide-activated gons) channels. whole-cell currents by 89 Ϯ 1% (n ϭ 12), 65 Ϯ 5% (n ϭ 9), and

Proks et al. PNAS ͉ December 14, 2004 ͉ vol. 101 ͉ no. 50 ͉ 17543 Downloaded by guest on September 25, 2021 41 Ϯ 2% (n ϭ 9) for hetR201C, hetQ52R, and hetV59G channels result in neonatal diabetes alone, whereas those that channels, respectively. This difference is consistent with previous produce a greater reduction in ATP-sensitivity are associated reports that mutations that enhance intrinsic KATP channel with additional symptoms. The molecular mechanism by which opening reduce the inhibitory efficacy of sulfonylureas (30). a mutation affects the channel ATP sensitivity determines the Thus, our results indicate that, whereas sulfonylureas may pro- severity of its effect in the heterozygous state, with those vide a valuable alternative to insulin injections for patients with mutations that influence gating producing larger effects on ATP mild mutations (like R201) (4, 6), those patients with mutations sensitivity, and a more severe disease phenotype, than those that affecting intrinsic gating may require more intensive drug ther- lie in the putative ATP-binding site. Our results also suggest that the efficacy of sulfonylurea therapy may vary with genotype. apy or a combination of drug and insulin therapy.

Conclusions We thank Chris Miller for helpful discussion. This work was supported by grants from the Royal Society, the , and Christ Church Our results demonstrate that KCNJ11 mutations that cause a College, Oxford. F.M.A. is a Royal Society GlaxoSmithKline Research small decrease in the ATP sensitivity of heterozygous KATP Professor. A.T.H. is a Wellcome Trust Research Leave Fellow.

1. Seino, S. & Miki, T. (2003) Prog. Biophys. Mol. Biol. 81, 133–176. 17. Hattersley, A. T., Turner, R. C., Permutt, M. A., Patel, P., Tanizawa, Y., Chiu, 2. Shyng, S. & Nichols, C. G. (1997) J. Gen. Physiol. 110, 655–664. K. C., O’Rahilly, S., Watkins, P. J. & Wainscoat, J. S. (1992) Lancet 339, 3. Tucker, S. J., Gribble, F. M., Zhao, C., Trapp, S. & Ashcroft, F. M. (1997) 1307–1310. Nature 387, 179–183. 18. Gloyn, A. L., Weedon, M. N., Owen, K. R., Turner, M. J., Knight, B. A., 4. Gloyn, A. L., Pearson, E. R., Antcliff, J. F., Proks, P., Bruining, G. J., Hitman, G., Walker, M., Levy, J. C., Sampson, M., Halford, S., et al. (2003) Slingerland, A. S., Howard, N., Srinivasan, S., Silva, J. M., Molnes, J., et al. Diabetes 52, 568–572. (2004) N. Engl. J. Med. 350, 1838–1849. 19. Schwanstecher, C., Meyer, U. & Schwanstecher, M. (2002) Diabetes 51, 5. Vaxillaire, M., Populaire, C., Busiah, K., Cave, H., Gloyn, A. L., Hattersley, 875–879. A. T., Czernichow, P., Froguel, P. & Polak, M. (2004) Diabetes 53, 2719– 20. Riedel, M. J., Boora, P., Steckley, D., de Vries, G. & Light, P. E. (2003) Diabetes 2722. 52, 2630–2635. 6. Sagen, J. V., Raeder, H., Hathout, E., Shehadeh, N., Gudmundsson, K., Baevre, 21. Inagaki, N., Gonoi, T., Clement, J. P. T., Namba, N., Inazawa, J., Gonzalez, G., H., Abuelo, D., Phornphutkul, C., Molnes, J., Bell, G. I., et al. (2004) Diabetes Aguilar-Bryan, L., Seino, S. & Bryan, J. (1995) Science 270, 1166–1170. 53, 2713–2718. 22. Sakura, H., Ammala, C., Smith, P. A., Gribble, F. M. & Ashcroft, F. M. (1995) 7. Antcliff, J. F., Haider, S., Proks, P., Sansom, M. S .P. & Ashcroft, F. M. (2004) FEBS Lett. 377, 338–344. EMBO J., in press. 23. Karschin, C., Ecke, C., Ashcroft, F. M. & Karschin, A. (1997) FEBS Lett. 401, 8. John, S. A., Weiss, J. N., Xie, L. H. & Ribalet, B. (2003) J. Physiol. 552, 23–34. 59–64. 9. Gribble, F. M., Ashfield, R., Ammala, C. & Ashcroft, F. M. (1997) J. Physiol. 24. Liss, B., Bruns, R. & Roeper, J. (1999) EMBO J. 18, 833–846. 498, 87–98. 25. McCormick, D. A. & Contreras, D. (2001) Annu. Rev. Physiol. 63, 815–846. 10. Markworth, E., Schwanstecher, C. & Schwanstecher, M. (2000) Diabetes 49, 26. Dunn-Meynell, A. A., Rawson, N. E. & Levin, B. E. (1998) Brain Res. 814, 1413–1418. 41–54. 11. Ashcroft, F. & Rorsman, P. (2004) Hum. Mol. Genet. 13, R21–R31. 27. Zawar, C., Plant, T. D., Schirra, C., Konnerth, A. & Neumcke, B. (1999) 12. Detimary, P., Dejonghe, S., Ling, Z., Pipeleers, D., Schuit, F. & Henquin, J. C. J. Physiol. 514, 327–341. (1998) J. Biol. Chem. 273, 33905–33908. 28. Griesemer, D., Zawar, C. & Neumcke, B. (2002) Eur. Biophys. J. 31, 467–477. 13. Koster, J. C., Marshall, B. A., Ensor, N., Corbett, J. A. & Nichols, C. G. (2000) 29. Koster, J. C., Knopp, A., Flagg, T. P., Markova, K. P., Sha, Q., Enkvetchakul, Cell 100, 645–654. D., Betsuyaku, T., Yamada, K. A. & Nichols, C. G. (2001) Circ. Res. 89, 14. Sakura, H., Ashcroft, S. J., Terauchi, Y., Kadowaki, T. & Ashcroft, F. M. (1998) 1022–1029. Diabetologia 41, 654–659. 30. Trapp, S., Proks, P., Tucker, S. J. & Ashcroft, F. M. (1998) J. Gen. Physiol. 112, 15. Terauchi, Y., Sakura, H., Yasuda, K., Iwamoto, K., Takahashi, N., Ito, K., 333–349. Kasai, H., Suzuki, H., Ueda, O., Kamada, N., et al. (1995) J. Biol. Chem. 270, 31. Enkvetchakul, D., Loussouarn, G., Makhina, E., Shyng, S. L. & Nichols, C. G. 30253–30256. (2000) Biophys. J. 78, 2334–2348. 16. Froguel, P., Vaxillaire, M., Sun, F., Velho, G., Zouali, H., Butel, M. O., 32. Shyng, S. L. & Nichols, C. G. (1998) Science 282, 1138–1141. Lesage, S., Vionnet, N., Clement, K., Fougerousse, F., et al. (1992) Nature 33. Krauter, T., Ruppersberg, J. P. & Baukrowitz, T. (2001) Mol. Pharmacol. 59, 356, 162–164. 1086–1093.

17544 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0404756101 Proks et al. Downloaded by guest on September 25, 2021