Role of the C‐terminus of SUR in the differential regulation of β‐cell and cardiac KATP channels by MgADP and metabolism

Key points β‐Cell KATP channels are partially open in the absence of metabolic substrates, whereas cardiac KATP channels are closed. Using cloned channels heterologously expressed in Xenopus oocytes we measured the effect of MgADP on the MgATP concentration–inhibition curve immediately after patch excision. MgADP caused a far more striking reduction in ATP inhibition of Kir6.2/SUR1 channels than Kir6.2/SUR2A channels; this effect declined rapidly after patch excision. Exchanging the final 42 amino acids of SUR was sufficient to switch the Mg‐nucleotide regulation of Kir6.2/SUR1 and Kir6.2/SUR2A channels, and partially switch their sensitivity to metabolic inhibition. Deletion of the C‐terminal 42 residues of SUR abolished MgADP activation of both Kir6.2/SUR1 and Kir6.2/SUR2A channels. We conclude that the different metabolic sensitivity of Kir6.2/SUR1 and Kir6.2/SUR2A channels is at least partially due to their different regulation by Mg‐nucleotides, which is determined by the final 42 amino acids. Abstract ATP‐sensitive potassium (KATP) channels couple the metabolic state of a cell to its electrical activity and play important physiological roles in many tissues. In contrast to β‐cell (Kir6.2/SUR1) channels, which open when extracellular glucose levels fall, cardiac (Kir6.2/SUR2A) channels remain closed. This is due to differences in the SUR subunit rather than cell metabolism. As ATP inhibition and MgADP activation are similar for both types of channels, we investigated channel inhibition by MgATP in the presence of 100 μm MgADP immediately after patch excision [when the channel open probability (P O) is near maximal]. The results were strikingly different: 100 μm MgADP substantially reduced MgATP inhibition of Kir6.2/SUR1, but had no effect on MgATP inhibition of Kir6.2/SUR2A. Exchanging the final 42 residues of SUR2A with that of SUR1 switched the channel phenotype (and vice versa), and deleting this region abolished Mg‐nucleotide activation. This suggests the C‐terminal 42 residues are important for the ability of MgADP to influence ATP inhibition at Kir6.2. This region was also necessary, but not sufficient, for activation of the KATP channel in intact cells by metabolic inhibition (azide). We conclude that the ability of MgADP to impair ATP inhibition at Kir6.2 accounts, in part, for the differential metabolic sensitivities of β‐cell and cardiac KATP channels.


Introduction
ATP-sensitive potassium (K ATP ) channels couple the metabolic state of a cell to its electrical activity. They consist of pore-forming Kir6.x subunits and regulatory sulphonylurea receptor (SURx) subunits, both of which participate in metabolic regulation of channel activity (Rorsman & Ashcroft, 2018). Kir6.2/SUR1 channels link blood glucose levels to insulin secretion from pancreatic β-cells and regulate transmitter release in neurones; Kir6.2/SUR2A channels are involved in the response to cardiac stress; and Kir6.1/SUR2B channels regulate vascular smooth muscle tone (Nichols & Lederer, 1991;Daut et al. 1994;Quayle et al. 1997;Ashcroft & Gribble, 1999;Seino et al. 2000).
Metabolic regulation of channel activity is mediated by changes in the cytosolic concentration of adenine nucleotides, which interact with nucleotide-binding sites on both subunits. Binding of ATP (or ADP) to Kir6.x produces channel inhibition, whereas interaction of MgADP (or MgATP) with the nucleotide-binding domains (NBDs) of SUR stimulates channel activity (Nichols et al. 1996;Gribble et al. 1997;Shyng et al. 1997;Tucker et al. 1997). The NBDs come together in a sandwich dimer fashion to form two nucleotide binding sites (NBS) at the interface (Lee et al. 2017). When examined in excised patches, little difference is observed in the sensitivity of Kir6.2/SUR1 and Kir6.2/SUR2A channels to inhibition by MgATP (Shyng et al. 1997;Gribble et al. 1998;Abraham et al. 2002), or in channel activation by MgADP (Hopkins et al. 1992;Nichols et al. 1996;Gribble et al. 1997;Dupuis et al. 2008;Proks et al. 2010Proks et al. , 2013. Nevertheless, these channels exhibit very different sensitivities to metabolism in intact cells. For example, in β-cells (Kir6.2/SUR1) and microvascular coronary endothelial cells (Kir6.1/SUR2B) K ATP channels open when extracellular glucose levels fall (Ashcroft et al. 1984;Langheinrich & Daut, 1997), whereas cardiac channels (Kir6.2/SUR2A) remain closed in glucose-free solutions and only open in response to severe metabolic inhibition (Nichols & Lederer, 1991;Flagg et al. 2010). Even when expressed in the same cell type, Kir6.2/SUR1, but not Kir6.2/SUR2A, channels are activated by metabolic poisoning (Dabrowski et al. 2001;Clark et al. 2010;Li et al. 2016).
The cause of these differences in metabolic sensitivity remains unclear. However, it is of important functional and clinical significance. It explains, for example, why patients or mice with gain-of-function mutations in Kir6.2 develop neonatal diabetes but have no obvious impairment in cardiac or skeletal muscle function (Clark et al. 2010(Clark et al. , 2012. In this paper, we therefore examine how SUR1 and SUR2A confer different metabolic sensitivities upon Kir6.2. We find striking differences in MgATP inhibition of Kir6.2/SUR1 and Kir6.2/SUR2A channels in the presence of MgADP. We further show that these differences in nucleotide handling can be accounted for by differences in the C-terminus of SUR. However, they only partially account for the different metabolic sensitivities of these channels.

Ethical approval
Experiments on Xenopus leavis oocytes were conducted in accordance with the policies and regulations set out in ASPA Schedule 1 in the UK and The Journal of Physiology's guidelines on animal ethics. Animals were purchased from the University of Portsmouth (Portsmouth, UK) and housed in the University of Oxford's animal facility, where they were fed twice a day and kept in filtered water. They were killed with an overdose of ethyl-m-aminobenzoate methanesulphonate (MP Biomedicals, LLC, CA, USA). Once anaesthesia was complete (as assessed by loss of reflexes), animals were killed by brain stem disruption. Oocytes were then removed under sterile conditions.

Surface expression
Surface expression assays were performed 2-3 days after injection of Kir6.2/SURx with or without a haemagglutinin (HA) label in the extracellular loop of Kir6.2, as described previously (Zerangue et al. 1999). Briefly, oocytes were incubated for 30 min in ND96 with 1% bovine serum albumin (BSA) at 4°C to block non-specific antibody binding, labelled with 1 μg/ml rat monoclonal anti-HA antibody (3F10, Boehringer Mannheim, Mannheim, Germany, in ND96−1% BSA for 40 min at 4°C), washed at 4°C for 1 h, and incubated with 2 μg/ml horseradish peroxidase-coupled secondary antibody in ND96−1% BSA for 30 min at 4°C. Cells were extensively washed, first with ND96−1% BSA and then in ND96 without BSA at room temperature (both washes for 40 min). Each individual oocyte was placed in 50 μl Power Signal Elisa (Pierce, Rockford, IL, USA) at room temperature and chemiluminescence was quantified in a Glomax 20/20 luminometer.
K ATP channels in excised patches, whether from pancreatic β-cells, mammalian cell lines or Xenopus oocytes, undergo both fast and slow rundown (reviewed by Proks et al. 2016). The nucleotide sensitivity of K ATP channels was assessed either as close to patch excision as possible ('instantaneous') or after fast rundown was complete ('rundown'). To measure the instantaneous ATP sensitivity, a single ATP concentration was applied per patch, in either the presence or the absence of 100 μM MgADP. In all cases, the current in the test solution was expressed as a fraction of the mean of that in control solution before and after ATP application.
The relationship between nucleotide concentration and K ATP current inhibition was fitted with: where I X is the steady-state K ATP current in the presence of the test nucleotide concentration [X], I 0 is the current in nucleotide-free solution obtained by averaging the current before and after application, IC 50 is the nucleotide concentration at which the inhibition is half maximal and h is the Hill coefficient. The single-channel open probability (P O ) in cell-attached patches containing a small number of channels (N < 8) was estimated from NP O , as described previously (Proks et al. 2005). For cell-attached patches containing larger number of channels, P O was estimated as: where I MEAN is the mean K ATP current in the cell-attached configuration, N is the number of active channels in the patch and i is the single-channel current (i = 4 pA at−60 mV). Following cell-attached recordings, the patch was excised and the number of active channels (N) was estimated using noise analysis from ß1 s data stretches obtained after the K ATP current had reached its maximum (Proks et al. 2010). The value of N determined by noise analysis varied between 8 and 750. There was no obvious difference between P O values determined by NP O analysis and noise analysis and there was no obvious relationship between P O and N. The latter indicates that it is unlikely that the value of P O is distorted when calculated by noise analysis.

Statistics
All values are given as mean ± SEM. Statistical significance was determined using Student's t-test.
been measured some time after patch excision. As channel activity usually runs down in excised patches, this inevitably means these studies have reported the nucleotide sensitivity of rundown or partially rundown channels. Both rundown and the decline of activation by magnesium nucleotides (DAMN) have been proposed to affect K ATP channel ATP sensitivity (Ribalet et al. 2000;Proks et al. 2010). Thus it is possible that differences in rundown, or DAMN, between Kir6.2/SUR1 and Kir6.2/SUR2A channels may explain the apparent discrepancy between nucleotide sensitivity in the intact cell and excised patch.
We therefore compared the ability of MgATP to inhibit Kir6.2/SUR1 and Kir6.2/SUR2 channels immediately after patch excision (see Methods) and after rundown ( Table 1). The IC 50 for rundown channels was similar to that reported previously (Nichols et al. 1991;Drain et al. 1998;Gribble et al. 1998;Tarasov et al. 2006;Clark et al. 2010). The IC 50 for instantaneous ATP inhibition ( Fig. 1B) was approximately twice that measured following rundown ( Fig. 1A), for both types of channel. Under both conditions, cardiac K ATP channels were slightly less inhibited by MgATP than β-cell K ATP channels.

ATP in the presence of MgADP
In the intact cell, both MgATP and MgADP are present simultaneously so that K ATP channel activity will be determined by the balance between nucleotide inhibition at Kir6.2 and nucleotide activation at SUR. We therefore next examined the effect of MgADP on both the rundown and the instantaneous MgATP concentration-response curves. We used 100 μM MgADP, which is close to that found in oocytes treated with sodium azide (97 μM; Li et al. 2016). MgADP reduced the channel sensitivity to MgATP to an extent that depended on both the type of SUR subunit and the time after patch excision at which measurements were made (Fig. 2, Table 1). MgADP (100 μM) produced an increase in the IC 50 for ATP inhibition measured after rundown that was much greater for Kir6.2/SUR1 than for Kir6.2/SUR2A (compare Fig. 2C,G). The difference was even more dramatic for the instantaneous ATP concentration-response relationship. Whereas there was a very large reduction in ATP sensitivity in the presence of MgADP for Kir6.2/SUR1 channels (Fig. 2D), no difference was observed for Kir6.2/SUR2A channels (Fig. 2H). Direct comparison of the instantaneous ATP concentration-response relationships in the presence of MgADP for Kir6.2/SUR1 vs. Kir6.2/SUR2A channels (Fig. 3B) clearly revealed that Kir6.2/SUR2A channels are almost completely closed at physiological MgATP concentrations (>1 mM), whereas ß30% of the Kir6.2/SUR1 current remains uninhibited. This result was completely missed when comparing the rundown ATP concentration-response relationships (Fig. 3A).
Cell-attached recordings in the presence of 3 mM sodium azide revealed significant on-cell channel activity in oocytes expressing Kir6.2/SUR1 (Fig. 3C,E). The mean single-channel open probability (P O ) was 0.26 ± 0.03 (n = 30). There was no obvious correlation between the estimated P O in cell-attached recordings and the number of active channels in the patch (N). Given that the maximal open probability of Kir6.2/SUR1 channels was 0.86 (as estimated by noise analysis of the peak current after excision; Table 2), this predicts that the channels remained ß30% uninhibited (in the presence of azide), which is close to that obtained in excised patches exposed to 1 mM MgATP and 100 μM MgADP (29%; Fig. 3A).
In contrast to Kir6.2/SUR1, most (62%) cell-attached patches on oocytes expressing Kir6.2/SUR2A showed no detectable channel activity, although some channel activity was observed after patch excision (Fig. 3D). In those patches where single Kir6.2/SUR2A channel currents were detected, very large currents were observed after patch excision (Fig. 3F). This suggests the existence of a small fraction of Kir6.2/SUR2A channels with reduced sensitivity to nucleotide inhibition.
Taken together, these data suggest that the difference in the metabolic sensitivity of Kir6.2/SUR1 and Kir6.2/SUR2A channels may be related to the differences in nucleotide handling.
What causes the difference in nucleotide handling? Matsuoka et al. (2000) showed that replacement of the final 42 residues of SUR2A with those of SUR1 enhanced the ability of MgADP to activate channels inhibited by 1 mM MgATP. Conversely, in the presence of 1 mM MgATP, MgADP activation was dramatically suppressed when the C-terminal 42 residues of SUR1 were replaced with those of SUR2A (Matsuoka et al. 2000).
However, Matsuoka et al. (2000) only investigated the effect of MgADP at a single ATP concentration. This paper also did not assess the instantaneous response to nucleotides, which may more closely resemble that found in the intact cell. We therefore examined the effect of exchanging the last 42 amino acids (the 'tail'; Fig. 4A) of SUR1 and SUR2A on the instantaneous MgATP concentration-response relationship in the presence of MgADP. As shown in Fig. 4B, the instantaneous concentration-response relationship was identical for SUR1 with the tail of SUR2A (SUR1-T2A) and for SUR2A. Conversely the tail of SUR1 endowed SUR2A (SUR2A-T1) with the properties of SUR1. This confirmed that the final 42 amino acids play a key role in determining the difference in nucleotide activation of SUR1 and SUR2A.

Is the tail of SUR sufficient to account for differences in metabolic sensitivity?
To determine if differences in the tails of SUR1 and SUR2A account not only for the difference in nucleotide handling by the NBDs, but also for the metabolic sensitivity of the channel, we next examined the whole-cell currents. As previously reported, Kir6.2/SUR1 but not Kir6.2/SUR2A channels were activated by metabolic poisoning with 3 mM azide (Fig. 5A,B,E). This was not due to a failure of Kir6.2/SUR2A channels to express, because the K-channel activator pinacidil produced a large current response. Surface expression assays also showed substantial expression (Fig. 5F), and a similar relationship between protein expression and current amplitude in the presence of a channel activator for both types of channel ( Fig. 5E-G).
Kir6.2/SUR1-T2A channels showed reduced metabolic activation compared to Kir6.2/SUR1 channels  (P < 0.0001), although this remained substantially greater than that of Kir6.2/SUR2A (P < 0.0001) (Fig. 5C,H): currents were expressed as a fraction of their amplitude in the presence of a K ATP channel activator to account for any differences in expression.

Complete deletion of the tail
To gain further insight into the role of the SUR tail in the K ATP channel function, we also examined the effect of deleting the last 42 amino acids of SUR1 or SUR2A (Fig. 6). For both channel types, this markedly decreased the macroscopic K ATP current amplitude in excised patches (Fig. 6A) and lowered the channel surface density ß100-fold (Fig. 6B). This is consistent with studies showing that removing a C-terminal dileucine forward trafficking motif in SUR1 impairs surface expression of the channel (Sharma et al. 1999). Tail deletion also reduced the instantaneous P O in the absence of nucleotides for both types of channel (Table 2), which provides further support for the idea that the C-terminal tail of SUR is important for the open state stability of the channel. Surprisingly, tail deletion also completely abolished MgADP activation of both types of channel (Fig. 6A,C). This result was unexpected because deleting the final 42 amino acids of SUR does not compromise the ability of the isolated NBDs to hydrolyse ATP . This suggests that nucleotide binding/hydrolysis may no longer be coupled to channel activation when SUR is

. The C-terminus modulates nucleotide interactions with the K ATP channels (data from instantaneous measurements in inside-out patches)
A, sequence alignment of the C-termini of rat SUR1, rat SUR2A and rat SUR2B. The grey box indicates the 7 amino acids previously identified by Matsushita et al. (2002). .2/SUR2A and Kir6.2/SUR2A 42 K ATP channels, plotted as relative luminescence units (RLU) (n = 6-13). * * P < 0.01; * * * P < 0.001. C, fractional currents remaining in the presence of 100 μM ADP recorded at −60 mV in inside-out patches from Xenopus oocytes expressing wild-type (WT) or truncated ( 42) K ATP channels, in the presence or absence of Mg 2+ . White bars: SUR1-containing channels; black bars: SUR2A-containing channels. Current is expressed as a fraction of that in nucleotide-free solution (n = 6-13).
truncated, and thus that the final 42 amino acids may be necessary for transduction of nucleotide binding into channel activation.

Discussion
Our data demonstrate a striking difference in nucleotide handling between Kir6.2/SUR2A and Kir6.2/SUR1 channels, and show that this involves the C-terminal region of the channel.

Reduced inhibitory effect of ATP in the presence of MgADP in excised patches
Although Kir6.2/SUR2A and Kir6.2/SUR1 channels are inhibited by ATP and activated by MgADP to similar extents and with similar IC 50 and EC 50 when these mechanisms are studied in isolation, when both nucleotides are simultaneously present channel regulation is strikingly different. This difference is particularly dramatic when the instantaneous ATP sensitivity is compared: in the presence of 100 μM MgADP, the IC 50 for ATP inhibition of Kir6.2/SUR2A is almost 10-fold less than that of Kir6.2/SUR1. As a consequence, the Kir6.2/SUR1 current at 1 mM MgATP is ß30% of that in the absence of ATP whereas the Kir6.2/SUR2A current is < 0.2%. Except in conditions of very severe metabolic deficiency, ATP levels do not fall below 1 mM, and MgADP levels are < 100 μM. Thus our results may help to explain why β-cell K ATP channels open in the absence of glucose and cardiac K ATP channels do not.
Our data demonstrate that there are intrinsic differences between SUR2A and SUR1 that account for the marked differences in the nucleotide and metabolic sensitivity when Kir6.2/SUR1 (β-cell) and Kir6.2/SUR2A (cardiac) channels are expressed in Xenopus oocytes. It is also possible that additional regulatory mechanisms contribute to the differences in metabolic sensitivity observed in native cells: for example, the levels of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) or the activity of the creatine phosphate/creatine kinase ATP buffering system (Li et al. 2002;Tarasov et al. 2006).
Previous studies have shown that in the presence of 3 mM sodium azide, ATP levels fall to ß1.2−1.4 mM in Xenopus oocytes Li et al. 2016). This corresponds to a K ATP current 20-23% of the maximum observed for Kir6.2/SUR1 in excised patches exposed to 100 μM MgADP (Fig. 3B). It is in reasonable agreement with the P O of Kir6.2/SUR1 channels, recorded from cell-attached patches in cells exposed to 3 mM sodium azide, which is ß30% of that of the maximally open channel (P O of 0.86; Proks et al. 2010).
Our results indicate that the efficacy of MgADP bound to NBS2 of SUR1 to reduce ATP inhibition at Kir6.2 is substantially impaired by a combination of rundown and DAMN following patch excision. The extent of this effect can be estimated using a simple (Monod-Wyman-Changeux, MWC) model for concerted gating of K ATP channels (Drain et al. 2004;Craig et al. 2008):  (Proks et al. 2010) and this effect can be modelled as where E ADP is the equilibrium constant for 'slow' gating when MgADP is bound to SUR and ζ is a proportionality factor that accounts for the effect of MgADP on slow gating of the channel. Assuming that the stimulatory effect of MgADP on the channel is entirely due to gating, it is possible to calculate values for E and K d,C from P O and IC 50 in the absence of MgADP (Tables 1 and 2) using eqns (3) and (4). The parameter ζ can then be determined from the value of IC 50 in the presence of MgADP (Table 1). The model predicts ζ values of 1896 and 94 for the instantaneous and rundown experimental conditions, respectively. In other words, rundown/DAMN produces a ß20-fold reduction in the efficacy of MgADP. An alternative possibility is that MgADP stimulation not only affects gating but also directly reduces ATP binding to Kir6.2 (Nichols et al. 1996;Shyng et al. 1997;Abraham et al. 2002). In this case, K d,C and K d,O could also be affected. It is not possible to distinguish formally between these two alternatives from the data.
The MWC model predicts that there is only a 4.6-fold reduction in the parameter E when MgADP binds to SUR2A if the channel has rundown. Such a weak effect on gating might perhaps explain why MgADP has no effect on the IC 50 for ATP inhibition of cardiac K ATP channels measured immediately after patch excision (Fig 2 and  Table 1), when the stimulatory effect of MgADP at SUR2A on gating may be compromised by the inhibitory effect of ADP at Kir6.2. Note that MgADP causes a small decrease (from 0.86 to 0.84) in the P O of Kir6.2/SUR2A channels immediately after patch excision ( Table 2).

Effects of tail deletion
Deletion of the last 42 residues of either SUR1 or SUR2A impaired surface expression of the K ATP channel and reduced its intrinsic open probability. It also prevented MgADP activation. Instead, MgADP inhibited both types of channel. This inhibition represents inhibition at Kir6.2, as the extent of inhibition was the same as that seen for wild-type channels in the absence of Mg 2+ . It has been previously reported that deletion of the C-terminus does not impair MgATP hydrolysis by purified NBD2 , indicating MgATP binding is unaffected in the isolated NBD. With the caveat that the NBDs may behave differently in the channel complex than in isolation, our results suggest that deletion of the tail of either SUR1 or SUR2A impairs the ability of bound MgADP to exert its stimulatory effect on channel opening. This idea is supported by the cryo-electronmicroscopy structure of the K ATP channel with Mg-nucleotides bound to the NBSs of SUR1 (Lee et al. 2017). In this structure, the C-terminal region of SUR1 lies close to the Walker A motif of NBS2, which is itself in very close proximity to Kir6.2 (20−30 nm; Lee et al. 2017). Thus, the C-terminus of SUR may be involved in coupling occupancy of the NBDs to gating of the Kir6.2 pore. For example, it may influence the ability of the NBDs to dimerise and thereby induce the conformational changes that couple MgADP binding to J Physiol 596.24 channel opening. This would explain why deletion of the tail prevents MgADP activation. If the tail of SUR2A was less effective at promoting dimerisation, it could also help explain the difference in the efficacy of MgADP at SUR1 and SUR2A in promoting channel activity.
The C-terminus contributes to functional differences in SUR1 and SUR2A Our results demonstrate that the difference in the ability of MgADP to reduce the ATP sensitivity of Kir6.2/SUR1 and Kir6.2/SUR2A channels resides in the last 42 amino acids of SUR. This is because replacing the tail of SUR1 with that of SUR2A resulted in channels that behaved like SUR2A, and vice versa. It may also explain why Kir6.2/SUR2B, in contrast to Kir6.2/SUR2A, shows a dramatic reduction in ATP inhibition in the presence of 100 μM MgADP (Table  1) -the tail of SUR2B is ß75% homologous to that of SUR1 but only 30% homologous with that of SUR2A.
Although exchanging the C-terminal tails of Kir6.2/SUR1 and Kir6.2/SUR2A channels swapped the instantaneous ATP sensitivity in the presence of MgADP, it only partially recapitulated the increase in K ATP current produced by sodium azide. Thus, stimulation of K ATP channel activity upon metabolic poisoning is likely to also involve additional regions of the channel, additional mechanisms or auxiliary proteins.

Conclusions
We conclude that different efficacies of MgADP bound at SUR1 and SUR2A to reduce ATP inhibition at Kir6.2 partially explain the different metabolic sensitivities of Kir6.2/SUR1 and Kir6.2/SUR2A channels. The underlying mechanism for this effect involves the C-terminal 42 amino acids of SUR. Deletion of these residues abolishes MgADP activation. This suggests that the C-terminus of SUR may be involved in stabilising MgADP binding, preventing closure of the NBD dimer, or decreasing the coupling efficacy between MgADP occupancy, closure of NBD dimer and gating of the Kir6.2 pore (or potentially all of these possibilities). Nevertheless, when taken together with previous studies, our results favour the view that the C-terminus of SUR is involved in transducing occupancy of the NBDs of SUR to gating of the Kir6.2 pore.