Cantu syndrome-associated SUR2 (ABCC9) mutations in distinct structural domains result in KATP channel gain-of-function by differential mechanisms

The complex cardiovascular disorder Cantu Syndrome arises from gain-of-function mutations in either KCNJ8 or ABCC9, the genes encoding the Kir6.1 and SUR2 subunits of ATP-sensitive potassium (KATP) channels. Recent reports indicate that such mutations can increase channel activity by multiple molecular mechanisms. In this study, we determine the mechanism by which KATP function is altered by several mutations in distinct structural domains of SUR2: D207E in the intracellular L0-linker and Y985S, G989E, M1060I, and R1154Q/W in TMD2. Mutations were engineered at their equivalent position in rat SUR2A (D207E, Y981S, G985E, M1056I and R1150Q/W) and functional effects were investigated using macroscopic rubidium (86Rb+) efflux assays and patch clamp electrophysiology. The results show that D207E increases KATP activity by increasing intrinsic stability of the open state, whilst the cluster of Y981S/G985E/M1056I mutations, as well as R1150Q/W, augment Mg-nucleotide activation. The response of mutant channels to inhibition by the sulfonylurea drug glibenclamide, a potential pharmacotherapy for CS, was also tested. There was no major effect on glibenclamide sensitivity for the D207E, Y981S, G985E or M1056I mutations, but glutamine and tryptophan substitution at R1150 resulted in significant decreases in potency.


Introduction
ATP-sensitive potassium (K ATP ) channels are potassium-selective ion channels formed by obligate co-assembly of pore-forming Kir6.x subunits and regulatory sulfonylurea receptors (SURx), in a 4:4 stoichiometry [1][2][3][4]. Channel opening is dynamically regulated by intracellular nucleotides and membrane phospholipids, and thereby couples the membrane potential of excitable cells to their metabolic state [5]. By binding to, and stabilizing, closed states of the Kir6.x subunit, ATP decreases channel open probability whilst magnesiumnucleotide complexes (MgADP and MgATP) bind to the nucleotide binding domains (NBDs) of SURx subunits to activate the channel [6,7].
In cardiac, smooth, and skeletal muscle, SUR2 subunits (of which there are two main splice variants, SUR2A and SUR2B) co-assemble variously with Kir6.1 (as in vascular smooth muscle) or Kir6.2 (as in ventricular and skeletal muscle) [8,9]. In the heart, Kir6.2/SUR2A K ATP channels have been proposed to be critical for ischemic pre-conditioning, whilst in skeletal muscle Kir6.2/SUR2A channels may provide a "brake" to hyperpolarize the membrane potential despite elevations in intracellular calcium during periods of exercise and increased metabolism [9,10]. In smooth muscle, Kir6.1/SUR2B K ATP activity is a key determinant of electrical excitability and consequent contractility in blood and lymphatic vessels, as well as in bladder and uterine muscle [11][12][13][14][15] There have now been multiple reports of mutations in the ABCC9 and KCNJ8 genes (which encode for SUR2 and Kir6.1 respectively) associated with the complex heritable disorder, Cantu Syndrome (CS) [16][17][18][19][20][21]. CS patients exhibit diverse cardiovascular features including: dilated and tortuous vessels, cardiomegaly, electrophysiological alterations in the cardiac conduction system, decreased neuro-vascular coupling and persistence of fetal circulation [16][17][18][20][21][22][23][24][25]] . An emerging model for the molecular basis of CS is that missense mutations in ABCC9 or KCNJ8 result in increased K ATP channel activity, and consequent reduced smooth muscle excitability and contractility [26,27]. CS-associated mutations in SUR2 have previously been shown to result in K ATP channel gain of function (GoF) by distinct mechanisms, including enhanced Mg 2+ -nucleotide activation and increased intrinsic open probability with consequent decreases in ATP inhibition [18,19]. Here we examined the functional effects of previously uncharacterized CS mutations that are predicted to cluster together at the link between NBD1 and TMD2: Y981S (human Y985S), G985E (G989E) and M1056I (M1060I) (Fig. 1), and compared the molecular consequences to those of D207E (D207E), located in the intracellular L0-linker, between TMD0 and TMD1 ( Fig. 1). In addition, the sensitivity of mutant channels to the sulfonylurea K ATP -inhibitor, glibenclamide was tested. Glibenclamide holds promise as a potential treatment for CS, however numerous K ATP GoF mutations which reduce sulfonylurea sensitivity have previously been reported [28][29][30]. Therefore, determining sulfonylurea sensitivity for specific mutations may be required for future individualized therapy. The results are interpreted alongside structural insights from recently reported high resolution cryo-EM structures of K ATP channel complexes [3,4] to provide further detail of the molecular basis of K ATP channel GoF in CS.

Molecular biology and cell culture
Mutations were introduced into a rat SUR2A (pCMV_rSUR2A) cDNA construct using site- 2.5 mg/ml oligomycin and 1 mM 2-deoxy-D-glucose to induce metabolic inhibition (MI) and incubated at room temperature for a further 10 minutes. Cells were then washed three times with Ringer's solution (either with or without MI supplements) before the experiment was commenced. Ringer's solution was added to each well, collected, and replaced at the defined time points (2.5, 5, 12.5 and 22.5 and 37.5 minutes). After the experiment, cells were lysed with 2% SDS to attain the remaining intracellular 86 Rb + and sample radioactivity was determined by scintillation counting.
The cumulative 86 Rb + efflux at each time point was calculated from the total counts from each sample (including the 86 Rb + remaining post-cell lysis). Apparent K ATP -independent efflux rate constants (k 1 ) were obtained from GFP-transfected cells using the equation: (1), and K ATP -dependent efflux rate constant (k 2 ) was obtained from K ATP -transfected cells using the equation: where k 1 was obtained from GFP-transfected cells (equation 1). The number of active channels was assumed to be proportional to k 2 . In MI conditions a time-dependent divergence from a mono-exponential efflux is observed. This is attributed to inactivation of background efflux mechanisms over time, therefore in this condition rate constants were derived from exponential functions fit to early time points only (2.5 -12.5 min). Data shown represents the mean ± S.E.M. from at least 3 independent experiments each with multiple replicates (N ≥ 3, n ≥ 7). Statistical significance was determined using Mann-Whitney U tests with a p value < 0.05 deemed statistically significant.

Inside-out excised patch clamp recordings
Where the current in K int = I max = 1, I min is the normalized minimum current observed in ATP/MgATP/glibenclamide, [X] refers to the concentration of ATP/MgATP/glibenclamide, IC 50 is the concentration of half-maximal inhibition and H denotes the Hill coefficient.
Data were tested for statistical significance using the Mann Whitney U test, and presented as mean ± S.E.M.

Case history of subject with SUR2[Y985S] mutation
The subject is the fourth child of healthy, unrelated Caucasian parents, with no family history of relevance to her condition. The pregnancy was complicated with raised nuchal translucency at twelve weeks and polyhydramnios at thirty-two weeks gestation. At thirtyeight weeks of gestation, labour was induced, with uncomplicated vaginal delivery. Birth weight was 5.3kg (>99 centile). There was no significant delay in early development, but language skills developed slowly. At birth, hypertrichosis was evident, with a full head of dark hair with low anterior hairline, shoulders, arms, legs and back were covered with long, thick and dark hair. At three years of age, facial features were rather coarse, with mild epicanthic folds and down slanting palpable fissures with full lips and a broad face. The forehead was extremely low with fine hair in front of the ears, extending over her chin and lanugo over her neck and chest. The heart was slightly enlarged, but there was no overt evidence of cardiomyopathy.
At age 5, the subject presented with recurrent respiratory infections and required hospital admission for pneumonia, leading to tonsillectomy and adenoidectomy, which improved severe snoring and obstructive sleep apnoea. Height was on the 50th centile, weight on the 91st centile and her head circumference was on the 98th centile. Facial features remained coarse with down slanting palpable fissures, full cheeks, broad tip to the nose with mild thickening of the alae nasae and a low columella. Significant joint laxity was evident in the hands, with deep palmar creases and soft skin on the palms and generous fetal finger pads.
This subject thus exhibited most of the features typically found in individuals with CS [18,24]. Sequencing of ABCC9 coding regions revealed a heterozygous mutation (c.2954A>C, p.Y985S) that was absent in genomic DNA from either parent. Heterozygous de novo mutations (p.G989E; p.M1060I) were also identified in two additional diagnosed CS subjects, for whom clinical details are not available. The three mutated residues are predicted to cluster in a similar location within the SUR2 protein ( Fig. 1). In the following study we therefore analyzed the molecular consequences of these mutations, and compared them to the consequences of the most common CS mutation (p.R1154Q), and another uncharacterized CS mutation p.D207E [16], located in distinct SUR2 domains.

Cantu Syndrome mutations result in gain of function of K ATP channel in intact cells
To determine the effect of mutations on K ATP channel function, SUR2A constructs were co- suggesting that the number of active channels was similar.
All known CS patients are heterozygous and we modeled heterozygous conditions by co-expressing Kir6.2 together with WT SUR2A and mutant SUR2A subunits at a 1:1 ratio.
The resultant channels were assayed by monitoring 86 Rb + efflux. Only very minor increases in basal efflux rate were observed for D207E, G985E and M1056I, whilst a moderate, statistically significant increase was observed for Y981S channels (Fig. 3). Taken together, these data demonstrate that whilst all tested mutations result in K ATP gain-of-function, in heterozygous conditions the effect is subtle under basal conditions.

D207E in the L0 linker increases open-state stability and decreases ATP-inhibition
Increased basal K ATP channel activity could arise from changes in multiple distinct nucleotide sensing mechanisms, including increased Mg-nucleotide activation, or decreased ATP inhibition, either as a result of decreased binding affinity or decreased efficacy due to increase in intrinsic open state stability [19]. To investigate the effects of the above mutations on nucleotide sensitivity, we used inside-out patch clamp recordings. As shown in Fig. 4 indicating that sensitivity to Mg 2+ -activation was unaffected by this mutation (Fig. 4F), and suggests that GoF results from decreased sensitivity to inhibitory ATP itself. Considering the location of this residue, predicted to be in close proximity to the ATP binding site on Kir6.1 in the octameric K ATP complex [3,4], this could conceivably arise from altered binding affinity.
Alternatively, decreased ATP sensitivity could be the result of enhanced open-state stability of channels [19,31]. To test the latter directly, we measured the response of channels to PIP 2 perfused onto the intracellular surface of excised membrane patches (Fig. 5). PIP 2 increases open probability (Po) to ~1, and the ratio of initial current levels to the activated level in PIP 2 provides an estimate of the initial 'intrinsic' Po [28]. As shown in Fig. 5, the intrinsic Po was increased from ~0.4 in WT to ~0.7 in D207E channels. Therefore, the D207E mutation within the L0 linker results in K ATP gain of function by increasing the intrinsic open probability of channels, rather than by decreasing inhibitory ATP binding affinity.

Mutations within the Y981/G985/M1056 cluster increase Mg 2+ -nucleotide activation
The disease-associated mutations Y981S, G985E and M1056I are all clustered togetheron transmembrane helices 12 and 13 in TMD2 (Fig. 1). In comparison to WT SUR2A the IC 50 for ATP inhibition in the presence of Mg 2+ was significantly increased by each of these mutations although, in contrast to D207E, there was no effect on ATP sensitivity in the absence of Mg 2+ (Fig. 6). This is further demonstrated by the increase in for all mutants (Fig. 6G), indicating that the mutations in this cluster of residues linking NBD1 to TMD2 increase channel activity by enhancing Mg-nucleotide activation.

R1150Q and R1150W in TMD2 also enhance Mg 2+ -nucleotide activation
Having established that the TMD2 Y981S, G985E and M1056I mutations enhance Mg 2+nucleotide activation we sought to test whether this mechanism was conserved for other TMD2 mutations, the most common CS-associated mutations R1150Q and R1150W. In agreement with a previous report [16], we show that R1150Q causes a large increase in MgATP IC 50 whilst R1150W has a more modest effect (Fig. 7). In contrast, R1150Q and R1150W caused only slight increases in ATP IC 50 (Fig. 7), again reflected by increased IC 50 [MgATP]/IC 50 [ATP] for R1150Q and R1150W (Fig. 7F), and thus both R1150Q and R1150W cause gain of function predominantly by enhancing Mg 2+ -nucleotide sensitivity.

The effect of CS GoF mutations on glibenclamide sensitivity
Glibenclamide (glyburide) inhibits K ATP channels in a biphasic manner, with high-affinity inhibition arising from interaction with the SUR subunit occurring at nanomolar to micromolar concentrations and low-affinity inhibition due to interaction with the Kir6.x subunit [32]. To specifically measure high affinity inhibition we applied glibenclamide up to 10 µM.

Cantu Syndrome-associated ABCC9 mutations all cause K ATP GoF
To date, the few analyzed Cantu Syndrome-associated mutations in ABCC9 (SUR2), have been shown to result in gain-of-function of K ATP channels in the presence of Mg 2+ nucleotides, which can arise either from decreased sensitivity to inhibitory ATP, or augmented activation by Mg 2+ -nucleotides [16,19] mutations will be conserved irrespective of the pore-forming subunit. In addition, although the dependence of channel activity on intracellular nucleotides differs quantitatively between the two major SUR2 splice variants (SUR2A and SUR2B, which differ only in their C-terminal exon) [33,34], it is anticipated that, qualitatively, the changes observed for mutant SUR2Acontaining channels will be common for SUR2B-containing channels. Interestingly, many CS features such as vascular dilatation and lymphedema likely arise from smooth muscle dysfunction, whilst effects on cardiac electrophysiology and skeletal muscle are less obvious [24]. This may suggest that the biophysical effects of mutations are more severe in SUR2Bthan in SUR2A-containing channels. Alternatively, it is possible that the predominantly smooth muscle consequences of CS arise due to the physiological context of K ATP function in smooth muscle rather than unique biophysical effects on SUR2B compared to SUR2A.

The mechanistic basis of GoF varies between mutations
Here, we compared the sensitivity of wild type and mutant channels to ATP in the absence and presence of Mg 2+ , to separate Mg-independent inhibitory effect of ATP from the activating effect of MgATP. This analysis shows that the D207E mutation reduces ATP inhibition itself whilst the Y981S/G985E/M1056I mutations all increase activity by enhancing MgATP activation. The D207E mutation is found in the intracellular L0 linker between the two transmembrane domains TMD0 and TMD1 (Fig. 1) In contrast, we show that the Y981S, G985E and M1056I mutations all act by increasing K ATP channel activation by MgATP (Fig. 6). These residues are all predicted to lie in close proximity to each other in a cluster within TMD2; Y981 and G985 are found at the Nterminal end of TM12, immediately following the NBD1-TMD2 linker, whilst M1056I is situated on the opposing TM13 (Fig. 1). The location of the Y981/G985/M1056 cluster at the link between the NBDs and the TMDs (Fig. 1) is appropriate for transduction of movements between the intracellular and transmembrane domains of SUR2. Biochemical analyses of SUR and related ABC proteins indicate that MgADP or MgATP binding to the NBDs of SUR may act to stabilize dimerization of NBD1 and NBD2 [7,38,39]. However, how binding or NBD dimerization is coupled to gating of the channel pore remains poorly understood.
In addition, we show that the previously reported common CS mutations R1150Q and R1150W (located in TM15 of TMD2) also enhance MgATP activation (Fig. 7), demonstrating that multiple transmembrane regions of TMD2 are involved in the conformational changes associated with Mg 2+ -nucleotide activation.
Notably, the gain-of-function induced by each mutation is quite subtle when mutant and WT SUR2A subunits are co-expressed to mimic the clinically relevant heterozygosity (Fig. 2). Recent reports of GoF mutations in Kir6.2 and SUR1 that underlie neonatal diabetes demonstrate that even subtle biophysical effects can result in disease [40], suggesting that dramatic changes may not be necessary. On the other hand, since SUR2Bs may be the more pathologically relevant splice variant, it is possible that these mutations will have a greater effect on channels containing SUR2B. In addition, the channel activity measured in 86 Rb + experiments under basal conditions may not fully recapitulate the metabolic and physiological context for K ATP channels in muscle or other differentiated cell types, and so we cannot rule out a more significant activating effect under other conditions.

Consequences for sulfonylurea sensitivity
Previous studies have demonstrated that second generation sulfonylureas such as glibenclamide inhibit SUR2-containing KATP channels, albeit with lower potency than SUR1containing channels [41]. As such, glibenclamide, or other sulfonylureas, represents a potential pharmacotherapy for CS. However, there are multiple reports of neonatal diabetes mutations in the Kir6.2/SUR1 K ATP subunits which reduce sulfonylurea sensitivity [28,29], and as we have recently demonstrated, the CS-associated mutation V65M in Kir6.1 profoundly reduces glibenclamide inhibition of recombinant channels [30]. Therefore, it is important to assess the effect of SUR2 CS mutations on inhibitor sensitivity. Here, we show that the D207E, Y981S, G985E and M1056I mutations do not obviously affect glibenclamide sensitivity (Fig. 8), as determined in the absence of nucleotides in excised patch clamp experiments. It has been reported that sulfonylurea inhibition of SUR2-containing channels is affected by nucleotide regulation [42,43], and so it is possible that these mutations may alter SU sensitivity under more complex physiological regulation, but this remains to be established.
A decrease in glibenclamide potency was observed in both the R1150Q and R1150W mutations (Fig. 9). Interestingly, R1150 lies in TM15 (Fig. 1) and previous studies have demonstrated that TMs14-16 are critical for high-affinity SU binding to SUR subunits [44,45]. Indeed, serine to tyrosine substitution of a single residue in TM16 (predicted to lie within ~ 15 Å of R1150 on the cytoplasmic extensions of the TM helix) is sufficient to confer SUR2-like sensitivity to SUR1, and vice versa [44,45]. This raises the possibility that the R1150 mutations may directly decrease glibenclamide sensitivity via disruption of the drug binding site. The R1150W mutation exhibited a more pronounced effect than the glutamine mutation at the same site, perhaps due to a greater steric effect of the bulky tryptophan sidechain. Regarding the relevance to treatability of disease, it is important to note that glibenclamide sensitivity was evaluated in a "homozygous" context where all SUR2 subunits were mutated, whilst CS patients identified so far are heterozygous for the gain of function mutations and therefore the effect of the R1150 mutations on SU sensitivity may be moderated in the patients.
Taken together, the present results provide further evidence for K ATP gain of function consequences of SUR2 mutations in Cantu Syndrome. For several mutations clustered in TM12-13, the results illustrate a common mechanism (enhanced MgATP activation) without marked effect on sulfonylurea sensitivity. The results provide novel insights into the function of K ATP channel complexes, can be useful for linking CS genotype to phenotype in this complex disorder, and will inform the consideration of therapeutic approaches to CS.