Phosphatidylinositol 4,5-Biphosphate (PIP2) Modulates Interaction of Syntaxin-1A with Sulfonylurea Receptor 1 to Regulate Pancreatic β-Cell ATP-sensitive Potassium Channels*

Background: PIP2 actions on activating KATP channels are not only on Kir6.2 but may be also on syntaxin-1A, to modulate syntaxin-1A actions on SUR1. Results: PIP2 disrupts Syn-1A·SUR1 interactions by reducing syntaxin-1A availability to inhibit of KATP channels. Conclusion: PIP2 modulates syntaxin-1A·SUR interactions. Significance: Membrane phospholipid composition in health and diabetes profoundly affect β-cell KATP channels by several mechanisms to influence insulin secretion. In β-cells, syntaxin (Syn)-1A interacts with SUR1 to inhibit ATP-sensitive potassium channels (KATP channels). PIP2 binds the Kir6.2 subunit to open KATP channels. PIP2 also modifies Syn-1A clustering in plasma membrane (PM) that may alter Syn-1A actions on PM proteins like SUR1. Here, we assessed whether the actions of PIP2 on activating KATP channels is contributed by sequestering Syn-1A from binding SUR1. In vitro binding showed that PIP2 dose-dependently disrupted Syn-1A·SUR1 complexes, corroborated by an in vivo Forster resonance energy transfer assay showing disruption of SUR1(-EGFP)/Syn-1A(-mCherry) interaction along with increased Syn-1A cluster formation. Electrophysiological studies of rat β-cells, INS-1, and SUR1/Kir6.2-expressing HEK293 cells showed that PIP2 dose-dependent activation of KATP currents was uniformly reduced by Syn-1A. To unequivocally distinguish between PIP2 actions on Syn-1A and Kir6.2, we employed several strategies. First, we showed that PIP2-insensitive Syn-1A-5RK/A mutant complex with SUR1 could not be disrupted by PIP2, consequently reducing PIP2 activation of KATP channels. Next, Syn-1A·SUR1 complex modulation of KATP channels could be observed at a physiologically low PIP2 concentration that did not disrupt the Syn-1A·SUR1 complex, compared with higher PIP2 concentrations acting directly on Kir6.2. These effects were specific to PIP2 and not observed with physiologic concentrations of other phospholipids. Finally, depleting endogenous PIP2 with polyphosphoinositide phosphatase synaptojanin-1, known to disperse Syn-1A clusters, freed Syn-1A from Syn-1A clusters to bind SUR1, causing inhibition of KATP channels that could no longer be further inhibited by exogenous Syn-1A. These results taken together indicate that PIP2 affects islet β-cell KATP channels not only by its actions on Kir6.2 but also by sequestering Syn-1A to modulate Syn-1A availability and its interactions with SUR1 on PM.

Pancreatic ␤-cell regulates glucose-stimulated insulin secretion through association with ATP-sensitive potassium channels (K ATP channels). 4 The K ATP channel is a hetero-octamer with four Kir6.2 (inward rectifier K ϩ 6.2) subunits forming a conduction channel surrounded by four regulatory SUR1 subunits (1). ␤-Cell plasma membrane (PM) excitability and insulin secretion are set by concentration of nucleotides, ATP, and ADP (2,3). The physiologic ␤-cell secretagogue is glucose, which, upon entry and metabolism in ␤-cells, increases ATP production, causing K ATP channel closure leading to cellular depolarization (4), which activates L-type voltage-dependent Ca 2ϩ channels, with resulting Ca 2ϩ influx triggering exocytotic fusion of insulin granules with PM (4,5). Conversely, when plasma glucose levels fall, increase in ADP and decrease in ATP concentrations lead to K ATP channel activation, with ensuing PM hyperpolarization, which reduces insulin release.
In addition to adenine nucleotides, K ATP channels are regulated by other endogenous factors in ␤-cells, particularly phosphatidylinositol 4,5-biphosphate (PIP 2 ). PIP 2 , comprising only 1% of PM phospholipids, stimulates activity of ATP-sensitive and -insensitive Kir channels by increasing channel open probability (6). PIP 2 is an indispensable membrane phosphoinositide that participates in a wide variety of other cellular func-* This work was supported by Canadian Institutes of Health Research Grant tions, including production of second messengers, endo-and exocytosis, and regulation of ion channels, transporters, and actin cytoskeleton (7)(8)(9)(10)(11). For K ATP channel regulation, a negatively charged inositol triphosphate headgroup of PIP 2 interacts electrostatically with positively charged amino acid residues in N-and C-terminal cytoplasmic domains of Kir6.2 (12,13). However, several studies demonstrated that the SUR subunit plays an essential role in stabilizing PIP 2 -Kir6.2 interaction. For instance, a brief application of PIP 2 shifted ATP inhibition of Kir6.1/SUR1 channels compared with Kir6.2 alone, and this PIP 2 recovery was more stable when SUR1 was present, thus indicating that SUR increases PIP 2 binding and stimulation on Kir6.2 (14,15). A recent report showed that a mutation in SUR1-TMD0 induces spontaneous Kir6.2 current decay and was reversed with exogenous PIP 2 (16). PIP 2 therefore plays versatile roles in controlling ␤-cell K ATP channel activities and insulin exocytosis (12,17). Lin et al. (18) showed that disrupting K ATP channel and PIP 2 interaction by overexpressing PIP 2 -insensitive Kir6.2 mutants caused cellular depolarization and elevated basal insulin secretion. Conversely, up-regulation of PIP 2 expression causing activation of K ATP channels resulted in cellular hyperpolarization, which reduced insulin secretion despite the presence of high glucose (18).
Besides the aforementioned actions of PIP 2 on various ion channels, PIP 2 also interacts with various components of the exocytotic fusion machinery, including CAPS, synaptotagmins, rabphilin, and Syn-1A (19 -23). Syn-1A is one of three SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) proteins that form the ternary complex constituting the minimal membrane fusion machinery in neurons and neuroendocrine cells (24). Besides its role in membrane fusion machinery, Syn-1A appears to play additional roles in the secretory process, effectively regulating various calcium and potassium channels involved in both initiating and terminating exocytosis (25).
In a body of work, we have shown that Syn-1A could act as an endogenous regulator exhibiting potent inhibitory action on ␤-cell K ATP channels (25)(26)(27)(28). We identified specific conserved motifs within the nucleotide binding domains of SUR1 that functionally interact with Syn-1A (28). Syn-1A contains highly charged, polybasic juxtamembrane motif in which neutralizing mutations abrogated Syn-1A-PIP 2 electrostatic interaction, causing a reduction in exocytosis by influencing the clustering of Syn-1A molecules on PM required for efficient membrane fusion (21,22). In this study, we investigated the hypothesis that these actions of PIP 2 on Syn-1A could influence Syn-1A interactions with SUR1 to affect K ATP channel activities in ␤-cells.
In Vitro Binding Assay and Western Blotting-In vitro binding assays were performed as described (34). Briefly, 250 pmol of GST (control) and GST-Syn-1A (aa 1-265) or GST-Syn-1A-5RK/A (aa 1-265), both containing only the cytoplasmic domain bound to glutathione-agarose beads, were incubated with lysate extract of HEK293 cells (400 g of protein) co-transfected with SUR1 and Kir6.2 in lysis buffer in the presence of increasing concentrations of PIP 2 or other indicated phospholipids (Echelon Biosciences Inc.) at 4°C for 2 h with constant agitation. Beads were washed three times, and samples were separated on 10% SDS-PAGE, transferred to nitrocellulose membrane, and identified with anti-SUR1 antibody (1:1,000; gift from J. Ferrer, Barcelona, Spain).
Electrophysiology-K ATP channel recordings were performed on INS-1E cells using the inside-out patch clamp technique (35) and on rat ␤-cells and HEK293 cells using the whole-cell patchclamp technique. Pipette resistance when filled with solution was 1.0 -1.5 megaohms. GST, GST-Syn-1A, ATP (Sigma-Aldrich) and PIP 2 (Sigma-Aldrich) were perfused onto the cytoplasmic side of excised membrane patches. Membrane patches were held at Ϫ50 mV to evoke inward currents. For ␤-cell, HEK293, and INS-1 cell voltage-clamped whole-cell studies, membrane potential was held at Ϫ70 mV, and a pulse of Ϫ140 mV (500 ms) was given every 10 s to monitor K ATP current magnitude. Pipette resistance was 2-4 megaohms. Bath solution contained 140 mM NaCl, 4 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 10 mM HEPES, 2 mM glucose, pH 7.3. Pipette solution contained 140 mM KCl, 1 mM MgCl 2 , 1 mM EGTA, 10 mM HEPES, pH 7.25. GST, GST-Syn-1A, and PIP 2 were added to intracellular solution for dialysis into cells via patch pipette. Tolbultamide (0.1 mM; Tolb) was perfused into bath solution after maximum current reached to completely inhibit and verify the K ATP current. All recordings were carried out at 22-24°C using an EPC10 amplifier with Pulse version 8.77 acquisition software (HEKA Electronik, Lambrecht, Germany). Data were sampled at 1 kHz.
FRET Imaging-As described previously (33), FRET study by total internal reflection fluorescence microscopy (TIRFM) assesses molecular interactions on the surface of PM, avoiding contamination from intracellular FRET signals. HEK293 cells were transfected with different combinations of plasmids 2 days prior to the experiment, where EGFP fused with SUR1 was used as the FRET donor, and mCherry fused with full-length Syn-1A or full-length Syn-1A-5RK/A was used as the FRET acceptor; Kir6.2 co-infected to express functional K ATP channels localized correctly on PM. For FRET analysis, four images, including donor excitation/donor emission (Dd), donor excitation/acceptor emission (Da), acceptor excitation/acceptor emission (Aa), and acceptor excitation/donor emission (Ad), were acquired under same conditions. Donor-only and acceptoronly samples were acquired at the beginning of each experiment for bleed-through calculation. FRET efficiency was used to indicate interaction of the two proteins, calculated with the equation, FRET efficiency % ϭ (((FRET raw Ϫ (CoB ϫ Dd FRET ) Ϫ (CoA ϫ Aa FRET ))/Dd FRET ) ϫ 100%, where CoB is the amount of donor bleed-through in the absence of acceptor, and CoA is the amount of acceptor bleed-through in the absence of donor.
After baseline FRET images were taken, the cells were permeabilized with digitonin (10 g/ml in intracellular buffer, 5 min, 37°C). FRET images were then taken again, followed by perfusion with the indicated lipids for another 5 min, and then we waited for another 7-10 min before the final FRET images were captured. Intracellular buffer contained 20 mM HEPES, 5 mM NaCl, 140 mM potassium gluconate, and MgCl 2 , pre-equilibrated with O 2 /CO 2 ϭ 95:5, pH 7.4, at 37°C. PIP 2 , powder dissolved in double-distilled H 2 O, was sonicated for 30 -45 s to a stock concentration of 920 mM (per the manufacturer's instructions). This PIP 2 stock was diluted to the indicated concentrations in intracellular buffer and sonicated again for 30 s before adding to the cells.
Statistical Analysis-For statistical analysis of FRET efficiency, regions of interest were drawn around entire areas of the PM surface expressing any FRET signal (blue, green, or red; see the pseudocolor bar in Figs. 5, 7, and 9) as indicated, not including the purple areas having no FRET signal. From these regions of interest, FRET efficiency was calculated as mean Ϯ S.E., and values were compared using the Mann-Whitney test by SigmaStat version 3.1 (Systat Software Inc., Chicago, IL). For electrophysiological experiments, data analysis was done using SigmaPlot version 11.0 (Systat Software Inc.). Data obtained from concentration-response curves were fitted to the drug responsiveness equation, , A1 is K ATP current before PIP 2 application, A2 is maximal K ATP current activated by PIP 2 , X denotes [PIP 2 ] applied to membrane patches, X0 denotes [PIP 2 ] that produced half-maximal K ATP channel activation, and p is slope of the curve. Curve fitting was performed by Origin version 6.0. Inside-out electrophysiological data were analyzed using each cell as its own control. Whole-cell electrophysiological data are presented as mean Ϯ S.E., expressed as current normalized to cell capacitance (pA/ pF). For multiple groups, channel activity was compared using one-way analysis of variance, followed by the Student-Newman-Keuls post hoc test. For the binding assay and Western blotting, blots were quantified by densitometry scanning followed by analysis with Scion Image (beta-4.0.2, Scion Corp.).
Data were compared using Student's t test. We considered p Ͻ 0.05 as a significant difference.

PIP 2 Dose-dependently Inhibits Syntaxin-1A Binding to SUR1-
A Substantial body of evidence indicates that highly negatively charged membrane phosphatidylinositol polyphosphates interact with positively charged residues on the N and C termini of the cytoplastic domains of Kir channels (14,35,36). In addition, it is well established that highly charged, polybasic juxtamembrane regions of Syn-1A interact with PIP 2 (29); thus, the SNARE fusion machinery itself may be a target of regulation by phosphoinositides (22,29,37). Although there is strong evidence for the role of PIP 2 on Kir6.2 and SNARE protein interactions, we here explored the possibility that PIP 2 could activate K ATP channels in a manner contributed by its actions on Syn-1A, which, in turn, perturbs Syn-1A binding to SUR1.
PIP 2 Activation of Kir6.2/SUR1 Channels Is Reduced by Syn-1A-We examined PIP 2 dose-response activation of the K ATP channel on INS-1E cells in the absence and presence of 1 M GST-Syn-1A using an inside-out patch clamp technique. Membrane patches were held at Ϫ50 mV to induce inward currents. Fig. 2, A and B, shows representative traces of the protocol utilized for PIP 2 in the absence and presence of Syn-1A. Membrane patches were initially exposed to 0, 1, and 3 mM ATP K int solution to characterize K ATP channels and verify the ATP sensitivity of recorded currents. In Fig. 2A, after patch excision, channels rapidly run down in the absence of ATP; however, subsequent exposure to 5, 10, and 20 M PIP 2 gradu-ally recovered the currents to the level observed immediately after patch excision. Consistent with previous reports, prior to PIP 2 applications, both 1 and 3 mM ATP inhibited K ATP channels; however, three subsequent applications of PIP 2 decreased ATP sensitivity and completely abolished the inhibitory effect of 1 mM ATP, as reported previously (14). Of note, Fig. 2B shows that concomitant application of GST-Syn-1A and PIP 2 did not completely recover the channels after rundown. In addition, in the presence of GST-Syn-1A, reapplication of ATP after PIP 2 activation did not produce any less current inhibition. In Fig.   2C, adding GST-Syn-1A greatly reduced PIP 2 -mediated channel activation, causing a rightward shift of the dose response. EC 50 values for PIP 2 in the absence and presence of GST-Syn-1A are 2.38 Ϯ 0.81 (n ϭ 5) and 9.64 Ϯ 0.14 (n ϭ 3), respectively. Our results indicate that exogenously added GST-Syn-1A inhibits K ATP channels through direct binding and interaction with SUR1 at its cytoplasmic nucleotide-binding folds (NBF-1 and NBF-2), as we reported previously (34), and that Syn-1A regulates K ATP channels through PIP 2 interactions. For the latter finding, our results suggest that the exoge- We performed the converse experiment of inside-out recordings of INS-1E cells perfused with GST-Syn-1A (1 M; indicated by black circles above the current traces) in the presence or absence of PIP 2 (10 M; indicated by solid bars). The summarized results are expressed as percentages of maximum current elicited at 0 mM ATP K int solution determined in each patch. Fig. 3Ai shows the representative K ATP current tracing of GST with PIP 2 , and the corresponding summary data are shown in Fig. 3Aii (n ϭ 5). Here, K ATP currents underwent spontaneous decay in the absence of ATP; subsequent perfusion of PIP 2 in the continuous presence of GST increased channel activities. When comparing K ATP currents between GST alone (87.9 Ϯ 3.62% of control maximal current) and subsequent perfusion of GST and PIP 2 , PIP 2 exposure led to a pro-nounced increase in K ATP current (108.06 Ϯ 7.77%), which appeared larger but not significantly different from the initial current immediately after patch excision (0 mM ATP K int ) (Fig.  2B). These results indicate that after patch excision, K ATP channels undergo spontaneous rundown, and subsequent PIP 2 exposure led to recovery of channel activity. In Fig. 3, Bi and Bii), administration of GST-Syn-1A alone reduced maximum control K ATP currents by ϳ65.4% (n ϭ 5). When comparing K ATP currents between GST-Syn-1A alone and subsequent Membrane patches were initially exposed to 0, 1, and 3 mM ATP K int solution to characterize K ATP channels and verify the ATP sensitivity of the currents recorded. C, PIP 2 dose-dependent activation of K ATP current in the absence (n ϭ 5) and presence of Syn-1A (n ϭ 3). Results are mean Ϯ S.E. (error bars).

Syn-1A-PIP 2 Interaction in SUR1/K ATP Regulation
concomitant perfusion of GST-Syn-1A and PIP 2 , PIP 2 recovered Syn-1A-inhibited currents to 89.6 Ϯ 10.52% of maximal currents (n ϭ 5), but the currents never reached that seen immediately after patch excision. These results taken together indicate that PIP 2 activation of K ATP currents exceeds PIP 2mediated recovery of Syn-1A inhibition of K ATP currents. This is probably because of PIP 2 direct actions on the Kir6.2 subunit (12,13). PIP 2 Acts on Syn-1A to Functionally Disrupt Syn-1A⅐SUR1 Interactions in Rat ␤-Cells-To delineate the physiological relevance of our findings in INS-1E, we employed rat islet ␤-cells using whole-cell patch clamp recordings. 1 M GST, 1 M GST-Syn-1A, or a combination of 1 M GST-Syn-1A and 10 M PIP 2 was dialyzed into ␤-cells via a patch pipette. In the presence of 1 M GST only (Fig. 4, Ai and B), K ATP currents gradually developed in ␤-cells, reaching a maximum current density of 118.97 Ϯ 13.79 pA/pF (n ϭ 6). In contrast, dialyzing 1 M GST-Syn-1A (Fig. 4, Aii and B) (29). The postulate is that if Syn-1A-5RK/A mutant inhibitory actions on K ATP channels would not be disrupted by PIP 2 , then PIP 2 activation of K ATP channels may be reduced. Furthermore, the contribution of Syn-1A inhibitory action on SUR1 may be able to oppose the direct actions of PIP 2 on Kir6.2 in opening the channel.
The previous studies in Figs. 1-4 examined the effects of exogenously added GST-Syn-1A (containing only the cytoplasmic domain) and PIP 2 on K ATP channels. Live cell FRET imaging analysis enables the examination of full-length Syn-1A-mCherry versus full-length Syn-1A-5RK/A-mCherry and SUR1-EGFP expressed in HEK293 cells. Here, we assessed whether PIP 2 can disrupt their in vivo (thus physiological) interactions on PM by TIRF imaging (Fig. 5), which optically isolates the PM surface (see "Experimental Procedures"). HEK293 cells were permeabilized with digitonin to permit entry of PIP 2 (10 M). Of note, the addition of this physiologic PIP 2 concentration seemed to increase the fluorescence intensity of the Syn-1A-mCherry hotspots (indicated by arrows), suggesting an increase in Syn-1A-mCherry cluster formation in these PM areas (Fig. 5A, top images). Remarkably, in the same experiment (Fig. 5A, bottom images), the 10 M PIP 2 disrupted Syn-1A⅐SUR1 complexes (indicated by arrowheads), shown as a reduction of FRET efficiency from 30.69 Ϯ 2.2 to 15.381 Ϯ 1.3 (Fig. 5, A and C). This was a ϳ50% reduction, which, when taken with the similar disruption of GST-Syn-1A⅐SUR1 complexes in the protein-binding study (Fig. 1A), suggests that PIP 2 disruption of Syn-1A⅐SUR1 complexes seems to "free" more Syn-1A molecules to participate in the formation of Syn-1A clusters, the latter also promoted by PIP 2 . Remarkably, PIP 2 did not disrupt Syn-1A-5RK/A⅐SUR1 complexes (indicated by arrowheads in Fig. 5B; no PIP 2 , 37.14 Ϯ 3.1; with PIP 2 , 35.42 Ϯ 1.7 (Fig. 5C)). We noted larger areas of Syn-1A-5RK/A⅐SUR1 FRET fluorescence (Fig. 5B) than Syn-1A⅐SUR1 fluorescence (Fig. 5A, bottom). Thus, we calculated the fluorescent area against total PM area (Fig. 5D) and found that the Syn-1A-RK/ A⅐SUR1 FRET area occupied 31.9 Ϯ 4.9%, which is 198% of the Syn-1A⅐SUR1 FRET area of 16.1 Ϯ 3.7%. Along with the binding studies in Fig. 1A, it seems that the Syn-1A-5KR/A mutation increased its abundance on the PM from increased formation of Syn-1A-5RK/A⅐SUR1 complexes. Last, we noticed that the locations of Syn-1A⅐SUR1 FRET signals (Fig. 5A, bottom; indicated by arrowheads) were mostly not colocalized with the areas with abundant Syn-1A-mCherry (Fig. 5A, top and bottom; indicated by arrows), an important point that we discuss further below along with the results in Fig. 9. In these studies, we were a little surprised by these strongly positive results because we had initially expected that exogenously added PIP 2 would not incorporate substantially into the PM (30), thus exhibiting lesser effects on the Syn-1A⅐SUR1 interactions in the PM. The most likely explanation is that PIP 2 incorporates into PM through lipid tails over time, which would occur with greater frequency the longer duration PIP 2 in solubilized solution is exposed to the interior surface of the cell. Another explanation is that digitonin (used for permeabilization) affects PM cholesterol in a manner that could influence PIP 2 incorporation into PM or increase the sensitivity to PIP 2 promotion of Syn-1A cluster formation in the PM (21,22). Digitonin permeabilization (prior to the addition of PIP 2 ), however, did not independently affect Syn-1A cluster formation or Syn-1A⅐SUR1 FRET interactions (Fig. 5A). Alternatively, PM permeabilization might have led to inadvertent depletion of some cytosolic factors that can influence the state of Syn-1A (free versus complexed) or the sensitivity of Syn-1A⅐SUR1 complex disruption by PIP 2 .
We then determined the functional implications of these binding studies by examining whether PIP 2 activation of K ATP channels in INS-1 cells would be perturbed by the PIP 2 -insenstive mutation of Syn-1A. As shown in Fig. 6A (analysis in Fig.  6C), dialyzing INS-1 cells with standard pipette solution displayed a K ATP channel current density of 100.3 Ϯ 18.7 pA/pF (Ϫ140 mV stimulation, used as control in Fig. 6C). Under identical conditions, overexpression of full-length Syn-1A-WT plasmid in INS-1 cells (Fig. 6A, left traces) reduced K ATP chan-nel current density to 29% (29.1 Ϯ 4.2 pA/pF; Fig. 6C) of control. Application of 1 M PIP 2 (a physiologically lower concentration than 10 M PIP 2 used in previous binding and functional studies) blunted the overexpressed Syn-1A inhibition to 60% of control (60.2 Ϯ 5.4 pA/pF). In contrast, overexpression of Syn- 1A-5RK/A (Fig. 6A, right traces; analysis in Fig. 6C) into INS-1 cells reduced K ATP channel current to 18.6% of control (18.6 Ϯ 2.1 pA/pF), which was a further 36% reduction from that caused by overexpressed Syn-1A (p Ͻ 0.05). This more potent inhibition of K ATP channel activity by Syn-1A-5RK/A is consistent with increased Syn-1A-5RK/A⅐SUR1 complex formation in PM (Figs. 1 and 5D). Remarkably, the addition of 1 M PIP 2 could not increase K ATP current inhibited by overexpressed Syn-1A-5RK/A (14.9 Ϯ 3.1 pA/pF; Fig. 6A, right traces; analysis in Fig. 6C).
The above results suggest that under physiologically low PIP 2 concentration (ϳ1 M), PIP 2 modulation of Syn-1A⅐SUR1 complex was sufficient to modulate K ATP activity, which seemed more dominant over the direct actions of PIP 2 on Kir6.2 to open channels. In this experiment, the exogenous PIP 2 was acting directly on the cytoplasmic PIP 2 -binding domain of Syn-1A and not likely to have significantly affected Syn-1A cluster formation in PM. Importantly, a 1 M PIP 2 concentration had no detectable effect on Syn-1A⅐SUR1 complex assembly, at least in vitro, which may in part be due to differences in the sensitivity of the assay (Fig. 1A). When we raised PIP 2 concentration to 10 M (which disrupted in vitro Syn-1A⅐SUR1 assembly in Fig. 1A), we found no differences in K ATP channel activities between control, Syn-1A-WT, and Syn-1A-5RK/A ( Fig.  6B; analysis in Fig. 6D). These results indicate that at higher PIP 2 concentration, direct effects of PIP 2 on the Kir6.2 subunit predominate over the effects of PIP 2 on Syn-1A-5RK/A(and Syn-1A-WT) and SUR1 interaction to open K ATP channels, although this higher PIP 2 dosage remained unable to disrupt the abundant Syn-1A-5RK/A⅐SUR1 complexes (Figs. 1A and 5, C-E). These results modified our original thinking to suggest several modes by which PIP 2 activates K ATP channels in insulin-secreting ␤-cells: one at physiologic low PIP 2 concentration acting on Syn-1A at its PIP 2 -binding site that finely modulates Syn-1A⅐SUR1 interactions and one at higher PIP 2 concentrations, which act on K ATP channels by two mechanisms, first by directly binding Kir6.2 and second by sequestering Syn-1A molecules into Syn-1A clusters, which reduces the availability of free Syn-1A molecules to bind SUR1 and could disrupt Syn-1A⅐SUR1 complexes.

Specificity of Cellular Inositol Phospholipids in Modulating Syn-1A⅐SUR1 Complex Disassembly and K ATP Channel Activity-
We next assessed whether other abundant cellular phospholipids might similarly affect Syn-1A⅐SUR1 complexes (Fig. 7) to modify K ATP channel activity (Fig. 8) and whether this is mainly attributable to the negative charge of phospholipids purported to bind positively charged juxtamembrane polybasic residues of Syn-1A (14,35,36). FRET imaging assessment showed that 10 M phosphatidylcholine (0 net charge) and phosphatidyl-L-serine (less negative charge than PIP 2 ) did not affect Syn-1A⅐SUR1 complex formation (Fig. 7, A, B, and D) or HEK293 K ATP channel activities (Fig. 8, A, B, C, and E). Inositol trisphosphate (IP 3 ), which has a larger negative charge compared with PIP 2 (38), caused only a minor disruption (24%) in Syn-1A⅐SUR1 complexes (Fig. 7, C and D; without IP 3 , 33.05 Ϯ 2.9; with IP 3 , 24.91 Ϯ 3.9) but did not significantly affect K ATP channel activity (Fig. 8, D and E). These results indicate that the PIP 2 effects on Syn-1A⅐SUR1 interactions that modulate K ATP channel activity are specific.
Effects of PIP 2 Depletion from PM on Syn-1A⅐SUR1 Complex Formation and K ATP Channels-Our experiments above so far have used exogenous PIP 2 to alter Syn-1A⅐SUR1 interactions. Synaptojanin-1 is a polyphosphoinositide 5-phosphatase, which, when overexpressed using the construct HA-IPP1-CAAX, was reported to deplete endogenous PIP 2 from the PM, causing disruption of the Syn-1A clusters on the PM (30,31). Consistently, expressing this synaptojanin-1 construct to deplete endogenous PIP 2 (in Fig. 9A), we saw a reduction of the larger hotspots (suggesting clusters, indicated by arrowheads in the top images) of Syn-1A-mCherry fluorescence (Syn-1A-mCherry expressed in HEK cells) to small fluorescence spots (indicated by arrowheads in the bottom images), indicating a dispersion of Syn-1A molecules from Syn-1A clusters; this is similar to the previous report using this construct (31). Further intensity profile analysis of cross-sections of the indicated regions of these images shows that in the absence of synatojanin-1 (Fig. 9B, pink line), there was sustained high intensity Syn-1A-mCherry fluorescence suggesting Syn-1A clusters, whereas in the cell treated with synaptojanin-1, Syn-1A-mCherry fluorescence appeared as narrow spikes, indicating either dispersed Syn-1A molecules or much smaller clusters of Syn-1A molecules (Fig. 9C, blue  line). These results are consistent with Fig. 5A (top images), where the addition of exogenous PIP 2 appeared to increase Syn-1A-mCherry clustering on PM. Synaptojanin-1-induced deple-tion of endogenous PIP 2 resulted in increased FRET signals (59.3 Ϯ 6.2%; Fig. 9D) compared with the absence of synaptojanin-1 (35.6 Ϯ 7.1%; note more green to red in ϩsynpatojanin-1 versus blue to green in Ϫsynaptojanin-1 in Fig. 9A), indi-

Syn-1A-PIP 2 Interaction in SUR1/K ATP Regulation
cating increased Syn-1A⅐SUR1 complex formation in the PM. Of note, in Ϫsynaptojanin-1 cells, the FRET signals (indicated by arrows) were mostly located away from the Syn-1A-mCherry clusters (indicated by arrowheads; Fig. 9A), as was similarly observed in Fig. 5A. In contrast, in ϩsynaptojanin-1 cells, the FRET signals were mostly in small Syn-1A-mCherry hotspots (arrows and arrowheads point to the same hotspots). These results led us to further strengthen our thinking that PIP 2 depletion releases Syn-1A molecules from the large Syn-1A clusters to migrate away from the cluster to other PM sites where SUR1 molecules are located to then form Syn-1A⅐SUR1 complexes.
We then examined the functional implications of the endogenous PIP 2 depletion that we showed to increase Syn-1A⅐SUR1 complex formation. In Fig. 10, A and B, INS-1 cells transfected with synaptojanin-1 caused a 63% reduction of K ATP current, from 148.35 Ϯ 23.2 pA/pF (control, n ϭ 7) to 55.38 Ϯ 11.2 pA/pF (synaptojanin-1, n ϭ 9; p Ͻ 0.01). The application of exogenous GST-Syn-1A (1 M) to the synaptojanin-1-transfected cells showed a K ATP current of 51.54 Ϯ 9.3 pA/pF (n ϭ 7; p Ͻ 0.05), which is a 65% reduction compared with control. This is very similar to the effects of synaptojanin-1 treatment alone, indicating that exogenously added GST-Syn-1A could not further reduce the K ATP current that was already inhibited by the PIP 2 depletion caused by synaptojanin-1 treatment. Taken together, these results suggest that the enhanced "endogenous" Syn-1A⅐SUR1 complex formation caused by PIP 2   FEBRUARY 28, 2014 • VOLUME 289 • NUMBER 9 depletion must have had all of the SUR1 sites occupied by endogenous Syn-1A to have resulted in optimal inhibition of K ATP channels, leaving no available SUR1 sites for the exogenous GST-Syn-1A to bind and further inhibit K ATP channels.

DISCUSSION
In this work, we showed that the actions of PIP 2 on activating pancreatic islet ␤-cell K ATP channel are contributed by alteration of Syn-1A interactions with SUR1, in addition to the known actions of PIP 2 on Kir6.2 (14 -18, 35, 36). Specifically, we showed that in vitro binding of recombinant GST-Syn-1A (containing only the cytoplasmic domain) and SUR1 was dose-dependently disrupted by increasing PIP 2 concentrations (Fig.  1A). Although these results suggest that exogenous GST-Syn-1A and PIP 2 could sequester each other from acting on K ATP channels, our FRET study showed that PIP 2 could also disrupt in vivo FRET interactions of SUR1 (-EGFP) and fulllength Syn-1A (-mCherry) (Fig. 5). PIP 2 effects on Syn-1A⅐SUR1 interactions were relatively specific at physiologic concentrations, with similar charged PI(3,5)P 2 having reduced effects and more (PIP 3 ) or less negatively charged phospholipids having little to no effect on Syn-1A⅐SUR1 complex assembly or K ATP channel activity (Figs. 1 (B and C), 7, and 8). The functional implication of GST-Syn-1A⅐SUR1 sequestration was demonstrated by electrophysiological studies on rat islet ␤-cells, INS-1E cells, and SUR1/Kir6.2-expressing HEK293 cells, which uniformly showed that efficacy of PIP 2 activation of K ATP channels could be reduced by the addition of GST-Syn-1A (Figs. 2-4). All of these effects of PIP 2 on Syn-1A⅐SUR1 complex formation and consequent K ATP channel activity could be abrogated by the PIP 2 -insensitive Syn-1A mutant, Syn-1A-5RK/A. Importantly, we demonstrated multiple modes by which PIP 2 activates ␤-cell K ATP channel, whereby modulation of Syn-1A⅐SUR1 complex formation by physiologic low PIP 2 concentration sufficient to alter K ATP channel activity seemed to be more sensitive than the direct actions of PIP 2 (at higher concentration) on Kir6.2 (Fig. 6). These results suggest three mechanisms for the actions of the exogenous PIP 2 in opening K ATP channels. First, PIP 2 binds and sequesters exogenous GST-Syn-1A from binding SUR1. Second, some PIP 2 is incorporated into the PM to disrupt Syn-1A⅐SUR1 complexes in the PM. Third, PIP 2 acts on Kir6.2. The release of PM-bound Syn-1A molecules from PIP 2 -induced disruption of Syn-1A⅐SUR1 complexes seemed to contribute to the availability of Syn-1A to participate in the increase in Syn-1A clustering on the PM (Fig. 5A). This was assessed more critically with experiments whereby we depleted endogenous PIP 2 from the PM with synaptojanin-1, reported to reduce Syn-1A clustering on PM (31). Here, endogenous PIP 2 depletion appeared to release Syn-1A molecules from the PM clusters (Fig. 9). The "freed" Syn-1A molecules could then migrate to, find, and bind SUR1 molecules to form tight Syn-1A⅐SUR1 complexes (Fig. 9), which effected optimal inhibition of K ATP channels, leaving no available SUR1 molecules for additional exogenous GST-Syn-1A to bind and further inhibit K ATP channels (Fig. 10).
In our in vitro binding studies employing GST-Syn-1A binding to expressed SUR1, the GST-Syn-1A (without the transmembrane domain) would not be expected to form physiologic Syn-1A clusters, but nonetheless the PIP 2 -binding site at the cytoplasmic aa 260 -265 domain directly binds SUR1, and the resulting GST-Syn-1A⅐SUR1 complex could be disrupted by exogenously added physiologic PIP 2 concentrations. These results raise the question of how PIP 2 modulation of Syn-1A interactions with SUR1 could influence K ATP channel opening kinetics. We recently reported that Syn-1A binds SUR1 domains at the W B motif of NBF-1 and W A and W B motifs of NBF-2 (33). This suggests that Syn-1A might be binding the NBF-1/-2 dimer rather intimately, perhaps as a scaffolding protein, and this configuration and as yet undefined sites within these binding interfaces of SUR1 with Syn-1A might be putative sites for PIP 2 disruption. Although we do not know what the PIP 2 -sensitive sites are in SUR1, the PIP 2 -sensitive site in Syn-1A is known (29). We directly tested the latter by employing PIP 2 -insensitive mutations of the juxtamembrane basic residues in Syn-1A (Syn-1A-5RK/A) (29), and indeed, PIP 2 could not disrupt Syn-1A-5RK/A⅐SUR1 complexes. This was demonstrated by in vitro and in vivo (FRET) binding studies. The fact that abrogating mutations of the PIP 2 -binding sites actually increased Syn-1A⅐SUR1 binding indicates that the Syn-1A can bind SUR1 in a PIP 2 -independent manner via other H3 domains (27). Consistently, Syn-1A-5RK/A inhibition of SUR1/ Kir6.2 K ATP channels was rendered more resistant to PIP 2 activation, indicating that the actions of PIP 2 on SUR1 via Syn-1A binding have an important contribution to channel opening kinetics. Moreover, we found that PIP 2 at 1 M, which did not disrupt Syn-1A⅐SNARE complexes could already modulate Syn-1A⅐SUR1 complex actions on ␤-cell K ATP channel activity. This was lower than the PIP 2 concentrations required to disrupt Syn-1A⅐SUR1 complexes, and PIP 2 with this lower concentration could also directly act on Kir6.2 to induce channel activation. Interestingly, a recent study showed that the N-terminal TMD0 domain of SUR1 profoundly influenced Kir6.2 channel sensitivity to PIP 2 (16), which, however, is a domain that does not bind Syn-1A 5 but which nonetheless could still affect the more distant NBF domains' interactions with Syn-1A.
Hence, we conclude that the PIP 2 effects on SUR1 proteins that influence K ATP channel gating are both direct and indirect. The direct effects we showed in this work are PIP 2 actions on Syn-1A binding to SUR1, probably at the NBF domains (26,28). The indirect effects would be PIP 2 actions on Syn-1A clusters on the PM, whereby a physiologic increase or reduction of PM PIP 2 levels will determine how many Syn-1A molecules will be released from the Syn-1A clusters to bind adjacent SUR1 molecules on the PM. K ATP channels play a most important role in regulating insulin secretion from islet ␤-cells (4,5). Thus, any factor that acts directly or indirectly on ␤-cell K ATP channels could have consequential effects on insulin secretion, which we now show to include PIP 2 actions on Syn-1A binding to SUR1. PM and cytosolic PIP 2 levels are likely to be perturbed in diabetes, metabolic syndrome, and lipid disorders. Syn-1A levels are severely reduced in islets of type-2 diabetic patients (39). Because the electrostatic binding between PIP 2 and Syn-1A might enable each to sequester the other (21,22), the excess or deficiency of either one in these pathologic conditions could profoundly influence the availability of Syn-1A to bind SUR1 or the availability of PIP 2 to bind Kir6.2. Thus, this combined perturbation of ␤-cellular lipids and Syn-1A levels would probably contribute to perturbation of K ATP channels that in turn could partly account for the deficient biphasic insulin secretion in diabetes.