TALK-1 channels control β cell endoplasmic reticulum Ca²⁺ homeostasis

Abstract

Ca²⁺ handling by the endoplasmic reticulum (ER) serves critical roles controlling pancreatic β-cell function and becomes perturbed during the pathogenesis of diabetes. ER Ca²⁺ homeostasis is determined by ion movements across the ER membrane, including K⁺ flux through K⁺ channels. Here, we demonstrated that K⁺ flux through ER-localized TALK-1 channels facilitated Ca²⁺ release from the ER in mouse and human β-cells. We found that β-cells from mice lacking TALK-1 exhibited reduced basal cytosolic Ca²⁺ and increased ER Ca²⁺ concentrations, suggesting reduced ER Ca²⁺ leak. These changes in Ca²⁺ homeostasis were presumably due to TALK-1-mediated ER K⁺ flux, because we recorded K⁺ currents mediated by functional TALK-1 channels on the nuclear membrane, which is continuous with the ER. Moreover, overexpression of K⁺-impermeable TALK-1 channels in HEK293 cells did not reduce ER Ca²⁺ stores. Reduced ER Ca²⁺ content in β-cells is associated with ER stress and islet dysfunction in diabetes, and islets from TALK-1-deficient mice fed a high-fat diet showed reduced signs of ER stress, suggesting that TALK-1 activity exacerbated ER stress. Our data establish TALK-1 channels as key regulators of β-cell ER Ca²⁺, and suggest that TALK-1 may be a therapeutic target to reduce ER Ca²⁺ handling defects in β-cells during the pathogenesis of diabetes.

INTRODUCTION

Pancreatic β-cell Ca²⁺ influx triggers insulin secretion, and endoplasmic reticulum (ER) Ca²⁺ (Ca²⁺_ER) handling plays a key role in this process (1). Ca²⁺_ER serves many essential functions in β-cells, such as controlling protein processing and metabolism, and defects in Ca²⁺_ER homeostasis can trigger the unfolded protein response (UPR) (2). The importance of precise β-cell Ca²⁺_ER handling is evident in type-1 and type-2 diabetes mellitus (T2DM), during which Ca²⁺_ER homeostasis is disrupted, leading to β-cell dysfunction and eventual destruction (1–8). Impaired Ca²⁺_ER handling also causes defects in glucose-stimulated insulin secretion (GSIS), contributing to hyperglycemia (3, 9). Therefore, treatments which reduce ER stress in the context of β-cell dysfunction improve glucose tolerance (10–12). However, while it is known that β-cell Ca²⁺_ER concentrations are perturbed in diabetes (2, 4–7), the molecular determinants which set β-cell Ca²⁺_ER are poorly understood.

Maintenance of Ca²⁺_ER homeostasis requires that Ca²⁺ movement across the ER membrane is balanced with a simultaneous K⁺ flux in the opposite direction (13–15). Without this K⁺ countercurrent, Ca²⁺ release from the ER would rapidly generate a negative charge on the inside of the ER membrane, inhibiting further Ca²⁺_ER release. To date, only a few ER K⁺ channels have been identified, including TRIC-A channels, which regulate Ca²⁺_ER stores in myocytes (16, 17); TRIC-B channels, which control Ca²⁺_ER homeostasis in alveolar epithelial cells and osteoblasts (18, 19); and SK Ca²⁺-activated K⁺ channels, which modulate Ca²⁺_ER uptake in neurons and cardiomyocytes (20). Genetic ablation or pharmacological inhibition of these channels impairs Ca²⁺_ER handling. For example, knockout of TRIC-A or TRIC-B channels results in increased Ca²⁺_ER stores, presumably due to the loss of a K⁺ countercurrent which regulates the ability of Ca²⁺ to exit the ER (15, 16, 18). Despite the importance of K⁺ countercurrents in maintaining Ca²⁺_ER homeostasis, nothing is known about the mediators or functions of β-cell ER K⁺ countercurrents.

ER localization has been reported for several K2P channels, including TASK-1 (21), TASK-3 (22), TASK-5 (23), TWIK-2 (24, 25), and THIK-2 (24). Although the subcellular localization of TALK-1 channels has not been reported, a protein interactome study has determined that a majority (>60%) of the proteins interacting with TALK-1 are ER-resident proteins (26). Similarly, a human pancreatic islet cDNA library generated and screened in a membrane yeast-two-hybrid assay to identify islet TALK-1 interacting proteins detected multiple ER-resident proteins that interact with TALK-1 (27). In accordance with these observations, TALK-1 shows substantial intracellular staining in human and mouse pancreatic β-cells (28). Although these findings suggest that TALK-1 channels may serve an intracellular role, investigations of intracellular K2P channels have focused primarily on elucidating the factors that enable their functional expression on the plasma membrane, and an ER function for K2P channels has not been identified.

In β-cells, TALK-1 contributes to plasma membrane potential (V_m) hyperpolarization, thereby regulating cytosolic Ca²⁺ (Ca²⁺_c) influx and insulin secretion (28). TALK-1 is distributed in pancreatic islets as well as gastric somatostatin cells (29, 30), and is the most abundant islet K⁺ channel at the transcriptional level (31–33). A primary physiological function of β-cell TALK-1 channels is to limit glucose-induced islet electrical and Ca²⁺_c oscillations, controlling second-phase pulsatile insulin secretion (28). Furthermore, a non-synonymous polymorphism in TALK-1 (rs1535500) associated with T2DM (34–36) causes a gain-of-function in TALK-1 activity (28), which may impair Ca²⁺ _c oscillations and pulsatile insulin secretion. However, the molecular mechanisms underlying TALK-1 regulation of islet Ca²⁺_c oscillations remain unclear.

Here, we tested the hypothesis that TALK-1 channels were functional in the ER and mediated ER K⁺ countercurrents which support β-cell Ca²⁺_ER homeostasis. By measuring Ca²⁺_ER, Ca²⁺_c, and single-channel K2P currents on the ER membrane, we demonstrated that TALK-1 channels conducted ER K⁺ countercurrents which enhanced Ca²⁺_ER leak in mouse and human β-cells. We found that TALK-1 control of β-cell Ca²⁺_ER modulated islet Ca²⁺_c dynamics, which has important implications for understanding the regulation of Ca²⁺_c oscillations that underlie pulsatile insulin secretion. Moreover, we showed that other ER-localized K2P channels, such as TASK-1, could function in an identical manner. Inhibition of K⁺ currents through either TALK-1 or TASK-1 increased steady-state Ca²⁺_ER concentrations, demonstrating that the K⁺ channel function of these proteins was essential for their effects on Ca²⁺_ER homeostasis. Moreover, islets from mice lacking TALK-1 channels showed reduced signs of ER stress induced by chronic exposure to a high-fat diet (HFD), suggesting that defects in TALK-1 channel activity can perturb ER health and contribute to islet dysfunction in T2DM. Overall, these findings identify an intracellular function of K2P channels, and reveal TALK-1 channels as a possible therapeutic target to modulate Ca²⁺_ER homeostasis to reduce β-cell ER stress in diabetes.

RESULTS

TALK-1 activity promotes Ca²⁺_ER leak

TALK-1’s prominent intracellular staining pattern (28) and physical association with several ER-resident proteins (26) suggested that TALK-1 could be localized to the ER. To determine the subcellular localization of TALK-1, we performed immunofluorescence staining of mouse pancreas sections, and detected co-localization of TALK-1 with the ER marker calreticulin (Fig. 1A). Additionally, co-expression of a TALK-1/mCherry fusion protein and an ER-targeted indicator (37) in mouse islet cells revealed TALK-1 in the ER (Fig. S1).

Figure 1. TALK-1 channels modulate β-cell Ca²⁺_ER homeostasis.

(A) Representative images of a mouse pancreas section stained for TALK-1 and calreticulin. Scale bar is 10 µm. Images are representative of results obtained from 3 mice. (B) β-cell Ca²⁺_ER measurements made with the genetically encoded Ca²⁺_ER indicator D4ER. Cells were perfused with solutions containing indicated glucose concentrations and 50 µM CPA (N = 3 mice per genotype). (C) Area under the curve (AUC) analysis of Ca²⁺_ER under low (2 mM) and high (11 mM) glucose conditions from (B). (D) CPA-induced reduction in Ca²⁺_ER, presented as percent of maximum Ca²⁺_ER of wild-type (WT) β-cells from (B). (E) WT and TALK-1 KO β-cells were perfused with the indicated solutions; 11 mM glucose (G) and 125 µM diazoxide (Dz) were present throughout the experiment. (F) Fold increase in Ca²⁺ in response to the indicated treatments. (G) Ca²⁺ AUC for the period following addition of 2.5 mM Ca²⁺ to the extracellular buffer (t=1000–1750 s) (N = 5 mice per genotype for E to G). Statistical significance was determined by Student’s t-test; *P<0.05, **P<0.01, ***P<0.005.

While TALK-1 conducts K⁺ currents on the plasma membrane in β-cells (28), it has not yet been determined whether K2P channels in the ER, such as TALK-1, are functional. To test if TALK-1 channel function could affect Ca²⁺_ER homeostasis, we first directly measured β-cell Ca²⁺_ER from control (wild-type) and TALK-1 KO islets (28) with the ER-targeted, genetically encoded Ca²⁺ indicator D4ER (38) (Fig. 1B). Under both low and high glucose conditions, TALK-1 KO β-cells had significantly higher Ca²⁺_ER concentrations (Fig. 1, B and C). Inhibition of sarco-endoplasmic Ca²⁺ ATPases (SERCAs) with cyclopiazonic acid (CPA) produced a greater decrease in Ca²⁺_ER in KO β-cells compared with controls (Fig. 1D). Absolute Ca²⁺_ER concentrations in KO β-cells remained above wild-type β-cells after application of CPA. These findings suggest that inhibition of SERCAs was insufficient to completely empty β-cell Ca²⁺_ER stores, as observed in neurons (39), and implied that TALK-1 channels were a critical determinant of β-cell steady-state Ca²⁺_ER concentrations. As a slight reduction in Ca²⁺_ER stimulates β-cell proliferation (40), we tested if TALK-1 activity affected β-cell number or proliferation. The absence of TALK-1 did not alter islet cellular composition, nor did it impair adaptive proliferation (as determined by BrdU incorporation) in response to a short term (one week) HFD stimulus (41) (Fig. S2, A to F), suggesting that inhibition of TALK-1 did not influence islet cell composition.

To further confirm that TALK-1 modulates Ca²⁺_ER, we quantified Ca²⁺_ER indirectly in wild-type and KO β-cells by measuring Ca²⁺_c in response to multiple stimuli. Treating β-cells with the Ca²⁺ ionophore ionomycin in the absence of extracellular Ca²⁺ resulted in more Ca²⁺ release from KO than wild-type β-cells (Fig. S3, A and B), suggesting increased intracellular Ca²⁺ stores. We next perfused isolated wild-type and TALK-1 KO β-cells with Ca²⁺-free buffer containing diazoxide to selectively monitor Ca²⁺_c independently of Ca²⁺ entry through plasma membrane channels (Fig. 1E). Under these conditions, TALK-1 KO β-cells exhibited lower basal Ca²⁺_c, and the addition of CPA produced a larger increase in Ca²⁺_c (Fig. 1, E and F), suggesting reduced Ca²⁺_ER leak and increased Ca²⁺_ER stores. Following washout of CPA, addition of Ca²⁺ to the extracellular buffer led to a similar amount of Ca²⁺_c influx in wild-type and TALK-1 KO cells, showing that activation of store-operated Ca²⁺ entry (SOCE) was not impaired in TALK-1 KO β-cells (Fig. 1, E and F). However, the reduced basal Ca²⁺_c observed without external Ca²⁺ was maintained in the presence of extracellular Ca²⁺ in TALK-1 KO β-cells (Fig. 1, E and G).

As TALK-1 is also detected in human β-cells (28), we next examined whether TALK-1 was present in the ER of human β-cells. Immunofluorescent staining of human pancreas sections revealed co-localization of TALK-1 with the ER marker calreticulin (Fig. 2A). To assess TALK-1-mediated regulation of human β-cell Ca²⁺_ER, we measured CPA-induced Ca²⁺_ER release in β-cells expressing a dominant-negative TALK-1 (TALK-1 DN) construct (Fig. 2B). The TALK-1 DN contains a pore mutation which inhibits K⁺ conductance when it interacts with endogenous TALK-1 (28). The TALK-1 DN construct also contains a P2A element between the sequences encoding TALK-1 and an mCherry reporter, allowing us to detect cells expressing the TALK-1 DN by mCherry fluorescence; β-cells were identified by post-staining for insulin (28). Inhibition of TALK-1 with the TALK-1 DN caused a significant increase in CPA-induced Ca²⁺_ER release (Fig. 2C), demonstrating that TALK-1 modulates Ca²⁺_ER homeostasis in human β-cells.

Figure 2. TALK-1 channels modulate human β-cell Ca²⁺_ER homeostasis.

(A) Representative image of a human pancreas section stained for TALK-1 and calreticulin. Scale bar is 10 µm. (B) Representative recordings of intracellular Ca²⁺ in human β-cells transfected with either TALK-1 dominant negative mutant (DN) or mCherry control. 11 mM glucose (G), 0 mM Ca²⁺, 125 µM diazoxide (Dz), and 1 mM EGTA were present throughout. (C) Quantification of the fold change in the Ca²⁺ AUC in response to treatment with CPA in human β-cells. The number of β-cells per donor are indicated on the graph. Statistical significance was determined by Student’s t-test and one-way ANOVA followed by Bonferroni’s multiple comparison test; *P<0.05, **P<0.01, ***P<0.005.

To test whether TALK-1 activity reduced Ca²⁺_ER storage, we examined the effects of TALK-1 expression on Ca²⁺_ER in HEK293 cells. Two channel-forming isoforms of TALK-1 (1b and 1a) (42) co-localized with an ER-targeted YFP marker in these cells (Fig. 3A). We next assessed the consequences of TALK-1 expression on Ca²⁺_ER homeostasis in these cells. To minimize potential deleterious effects of protein overexpression, we used the TALK-1 DN mutant as a control, which permitted dissociation of the effects of TALK-1-mediated K⁺ conductance and protein-protein interactions on Ca²⁺_ER homeostasis. Overexpression of K⁺-conducting wild-type TALK-1 yielded a substantial increase in basal Ca²⁺_c and a concomitant reduction in total Ca²⁺ released from the ER during CPA-induced inhibition of SERCAs when compared to the non-K⁺-conducting TALK-1 DN channels (Fig. 3, B through D). These observations suggest that TALK-1 channel activity promotes Ca²⁺_ER leak. Addition of Ca²⁺ to the extracellular buffer produced a larger increase in Ca²⁺_c in wild-type-TALK-1 compared to TALK-1 DN-expressing cells, and similar to CPA-induced Ca²⁺_ER release, stimulation of IP3-triggered Ca²⁺_ER release with carbachol elicited a greater response in cells expressing TALK-1 DN (Fig. 3, B and E). Thus, the K⁺-channel function of TALK-1 is sufficient to alter Ca²⁺_ER homeostasis.

Figure 3. The K⁺ channel function of TALK-1 contributes to its regulation of Ca²⁺_ER homeostasis.

(A) TALK-1b and -1a co-localize with the ER marker ER-YFP. Images are representative of 3 independent experiments. Scale bar is 10 µm. (B) Representative recordings of HEK293 cells expressing either WT TALK-1 or TALK-1 DN and perfused with the indicated solutions; 10 mM glucose was present throughout the experiment. (C) Normalized Ca²⁺ AUC for the period during treatment with CPA (t=250–600 s). (D) Ca²⁺ AUC for the period following addition of 2.5 mM Ca²⁺ to the extracellular buffer (t=1000–1750 s). (E) Fold increase in Ca²⁺ in response to treatment with the muscarinic receptor agonist carbachol (N = 3 independent experiments for B to D). Statistical significance was determined by Student’s t-test; *P<0.05, **P<0.01, ***P<0.005.

To confirm the specificity of the effect of TALK-1 on Ca²⁺_ER, we compared Ca²⁺_ER storage in cells stably and inducibly expressing different K2P channels (43) (Fig. 4A, Fig. S4, A and B). Similar to transfected cells, TALK-1 induction reduced Ca²⁺_ER stores (Fig. 4A). We also found that expression of TASK-1 (Fig. 4A) and TASK-3 (Fig. S5, A and B) channels (21, 22) also caused a reduction in Ca²⁺_ER. We confirmed the Ca²⁺_ER reduction in TALK-1- and TASK-1-expressing cells by directly measuring Ca²⁺_ER using the genetically encoded Ca²⁺_ER indicator T1ER (44) (Fig. 4B). However, not all K2P channels influenced Ca²⁺_ER, as demonstrated by the absence of a Ca²⁺_ER phenotype following induction of TREK-2 or TREK-1 channels (Fig. 4A, Fig. S5 C through F).

Figure 4. Pharmacological manipulation of K2P channel activity can alter steady-state Ca²⁺_ER concentrations.

(A) Representative recordings of CPA-induced Ca²⁺_ER release in cell lines with tetracycline-inducible expression of the indicated K2P channels. Ca²⁺ AUC in response to CPA is shown to the right (representative of N = 3 independent experiments; NI: not induced). (B) Direct quantification of Ca²⁺_ER concentration in HEK293 cells with inducible expression of TALK-1, TASK-1, TREK-2, TREK-1 and the Ca²⁺_ER indicator T1ER (N = 3 independent experiments). (C and D) Treatment of TASK-1-expressing cells with ML365 restores Ca²⁺_ER to pre-channel expression concentrations (C); AUC quantification (D) (N = 3 independent experiments). (E) Mouse α-cells were treated with ML365 in the presence of 11 mM glucose and 125 µM diazoxide (representative of N = 3 independent experiments). The response to CPA is quantified in (F) (N = 3 independent experiments). Statistical significance was determined by Student’s t-test; *P<0.05, **P<0.01, ***P<0.005.

This finding prompted us to examine whether pharmacological modulation of K2P channels could be used to manipulate Ca²⁺_ER storage. Because specific pharmacology for TALK-1 channels does not presently exist, we tested if selective TASK-1 inhibition with ML365 (a small molecule antagonist of TASK-1 (45), and a partial TASK-3 inhibitor) could influence Ca²⁺_ER. Inhibition of TASK-1 channel activity with ML365 treatment significantly restored Ca²⁺_ER loss caused by TASK-1 channel induction (Fig. 4, C and D). As ML365 also partially blocks TASK-3, treatment of TASK-3-expressing cells with this compound caused a modest increase in Ca²⁺_ER (Fig. S5, A and B). However, ML365 did not influence Ca²⁺_ER in cells expressing TREK-1 or TREK-2 (Fig. S5, C through F). These data further support the hypothesis that K⁺ flux through ER K2P channels enhances Ca²⁺_ER leak.

Although these observations suggested that pharmacological modulation of K2P channels could be used to regulate Ca²⁺_ER, immortalized cell lines expressing TASK-1 may not recapitulate Ca²⁺_ER handling of primary tissues. Thus, we tested whether pharmacological blockade of TASK-1 could alter Ca²⁺_ER in primary islet α-cells, where they regulate glucagon secretion (46). ML365 treatment increased α-cell Ca²⁺_ER stores (Fig. 4, E and F). These data demonstrate that TASK-1 affects α-cell Ca²⁺_ER homeostasis, and that pharmacological inhibition of Ca²⁺_ER-modulating K2P channels could be used to control Ca²⁺_ER in primary islet cells.

TASK-1 mutations have been implicated in pulmonary arterial hypertension (PAH), one of which (G203D) is a dominant-negative mutation that directly impairs K⁺ conductance through TASK-1 (47). Expression of TASK-1 G203D produced a significantly greater increase in Ca²⁺_ER stores compared to that of control TASK-1 channels (Fig. S5, G and H), an effect similar to that of expressing the TALK-1 DN mutant, which increased Ca²⁺_ER stores compared to cells expressing wild-type TALK-1. These observations imply that defects in TASK-1 K⁺ conductance may inappropriately increase Ca²⁺_ER or impair physiologically important Ca²⁺_ER fluxes.

TALK-1 and TASK-1 form functional channels across the ER membrane

During Ca²⁺_ER release, K⁺ moves across the ER membrane to maintain ER electroneutrality and sustain the driving force for Ca²⁺_ER release (14, 16, 48, 49). To directly assess whether TALK-1 functions as an ER K⁺ channel, we used nuclear patch clamp electrophysiology (50) to measure channel activity on the outer nuclear membrane, which is continuous with the ER (Fig. 5, A and B). In nuclei from cells expressing TREK-2, which does not affect Ca²⁺_ER, we did not detect TREK-2 channel activity (Figure 5C). However, nuclei from cells expressing TALK-1 (Fig. 5D) or TASK-1 (Fig. 5E) exhibited single-channel openings consistent with their respective biophysical profiles, suggesting that TALK-1 and TASK-1 form functional channels on the ER membrane.

Figure 5. Functional TALK-1 and TASK-1 channels are present in the ER membrane.

(A) Nuclear patch clamp of the outer nuclear membrane (ONM) permits detection of ER ion channels. (B) Representative image of isolated mouse islet nuclei with patch pipette positioned on nucleus. (C) Recordings obtained from the nucleus of a TREK-2-expressing HEK293 cell (representative of 5 nuclei). (D) Current trace obtained from the nucleus of a TALK-1 expressing HEK293 cell. Representative current amplitude histograms at right (representative of 8 nuclei). (E) As in D, but recorded from the nucleus of a TASK-1 expressing cell (representative of 7 nuclei). (F) Representative current traces obtained from WT mouse nuclei; patches held at −50 mV (representative of 42 nuclei) (G) Single-channel current-voltage relationships from nucleus recordings obtained from TALK-1- (N = 8) and TASK-1- (N = 7) expressing HEK293 cells and WT islet-cells (N = 42). (H) Percent of nuclei with K2P-channel-like channel activity detected in WT and TALK-1 KO β-cells (N = 42 nuclei; 4 mice per genotype). Statistical significance was determined by Student’s t-test; *P<0.05.

These results implied that TALK-1 and TASK-1 regulate Ca²⁺_ER homeostasis by allowing K⁺ flux across the ER membrane. We further tested whether TALK-1 modulation of Ca²⁺_ER release depended on K⁺ flux by manipulating the cytosolic K⁺ concentration. Using digitonin-permeabilized HEK293 cells expressing wild-type or DN TALK-1 and the genetically encoded Ca²⁺_ER indicator G-CEPIA1er (37), we examined Ca²⁺_ER leak in response to SERCA inhibition with CPA. In the presence of K⁺, Ca²⁺_ER leak was faster in cells expressing wild-type TALK-1 compared to TALK-1 DN (Fig. S6, A through D). Therefore, K⁺ flux through TALK-1 supports the movement of Ca²⁺ across the ER membrane. We also examined whether TALK-1 functions as an ER K⁺ channel in primary cells by performing nuclear patch clamp recordings on nuclei isolated from wild-type and TALK-1 KO islets. We detected single channel openings (Fig. 5F) with a current amplitude comparable to cloned TALK-1 in 55.6 ±6.3% of wild-type islet cell nuclei (Fig. 5G). However, only 31.2 ±2.7% of nuclei from TALK-1 KO islets displayed K2P-like channel openings (Fig. 5H). Together, our findings suggest that TALK-1 and TASK-1 form functional channels on the ER membrane, allowing for a K⁺ countercurrent that supports Ca²⁺_ER leak and helps to set Ca²⁺_ER.

TALK-1 regulation of β-cell Ca²⁺_ER handling modulates islet Ca²⁺ oscillations

To dissect the role of TALK-1 modulation of Ca²⁺_ER during β-cell Ca²⁺ influx, we controlled β-cell Ca²⁺_c influx with K⁺-induced depolarization of diazoxide-treated cells, in the presence or absence of the SERCA inhibitor thapsigargin (51) (Fig. 6A). Under these conditions, the role of plasma membrane TALK-1 channels is effectively dissociated from its intracellular functions: diazoxide circumvents the depolarizing effects of glucose by activating K_ATP channels, and K⁺ depolarization activates voltage-gated Ca²⁺ channels (VDCCs) independently of K⁺ channel activity (52). Subtracting the control trace from the thapsigargin-treated trace revealed the Ca²⁺_ER contribution to the Ca²⁺_c signal (Fig. 6B). During Ca²⁺ influx, Ca²⁺_ER uptake was observed (Figure 6B, downward deflection), whereas Ca²⁺_ER release occurred following the depolarizing K⁺ pulse (Figure 6B, upward component). We found reduced Ca²⁺_ER release in KO β-cells (Fig. 6, B and C), in accordance with our finding that TALK-1 channel activity promotes Ca²⁺_ER release.

Figure 6. TALK-1 regulates Ca²⁺_ER handling during plasma membrane Ca²⁺ influx in β-cells.

(A) Intracellular Ca²⁺ oscillations in response to pulses of 45 mM K⁺ (K45) for 40 seconds in the presence or absence of thapsigargin (1.25 µM). Recordings were performed in the presence of 11 mM glucose (G), 2.5 mM Ca²⁺, and 125 µM diazoxide (Dz). (B) Subtraction of the thapsigargin-treated trace from the control trace in A reveals the kinetics of Ca²⁺_ER uptake and release. (C) Quantification of average Ca²⁺_ER uptake and release in WT and TALK-1 KO β-cells (N = 3 mice per genotype). (D) Effect of CPA on glucose-stimulated Ca²⁺ influx in WT and KO islets. (E) Area under the curve (AUC) analysis of glucose-stimulated Ca²⁺ influx for periods corresponding to low glucose (2G), high glucose (11G), and CPA (11G + CPA) (N = 49 WT and 53 TALK-1 KO islets). Statistical significance was determined by Student’s t-test; *P<0.05, **P<0.01, ***P<0.005.

β-cell Ca²⁺_ER release has been implicated in the activation of hyperpolarizing Ca²⁺-activated K⁺ currents (53–55). When stimulated with glucose, KO islets show accelerated Ca²⁺ oscillations (28), which may be due to changes in Ca²⁺_ER control of the V_m and VDCC activity. We tested this by depleting Ca²⁺_ER using CPA in wild-type and TALK-1 KO islets undergoing glucose-stimulated Ca²⁺ oscillations (Fig. 6D). Under low glucose conditions, basal Ca²⁺_c concentrations were modestly lower in KO islets (Fig. 6E). Upon stimulation with high glucose, Ca²⁺ influx was significantly greater in KO islets (Fig. 6E). However, Ca²⁺_ER depletion with CPA normalized Ca²⁺_c concentrations in KO islets to similar to those in wild-type islets (Fig. 6E). As depletion of Ca²⁺_ER removes the contribution of the ER from the glucose-stimulated Ca²⁺_c signal, this finding suggested that TALK-1 influences β-cell Ca²⁺_c by modulating Ca²⁺_ER handling, which in turn regulates plasma membrane K⁺ currents and Ca²⁺ influx. Therefore, we proceeded to test the relationship between TALK-1 regulation of Ca²⁺_ER release and β-cell Ca²⁺-activated K⁺ currents.

The termination of each electrical oscillation is triggered by a slowly activating, Ca²⁺-dependent K⁺ current termed K_slow, which is mediated by intermediate-conductance K_Ca channels (IK), apamin-insensitive small conductance (SK) K_Ca channels, and K_ATP channels (55–57). The β-cell ER can release Ca²⁺ close to the plasma membrane (58), and K_slow activity is sensitive to Ca²⁺_ER release (53–55). In KO islets, V_m repolarization is reduced by approximately 50% at the termination of each electrical oscillation (28), suggesting that K_slow may be impaired in KO islets. We tested this notion by measuring K_slow in wild-type and KO β-cells. K_slow amplitude (Fig. 7A, inset) was reduced in KO β-cells by 48 ±17% relative to wild-type β-cells (Fig. 7, B and C). TALK-1 is not activated by Ca²⁺_c in oocytes (59), and we also found that TALK-1 activity in HEK293 (Fig. 7D) or β-cells (Fig. 7E) was insensitive to Ca²⁺_c, making it unlikely that TALK-1 is a constituent channel of K_slow. These findings suggest that TALK-1 may modulate β-cell K_slow indirectly through control of Ca²⁺_ER homeostasis. To assess whether modulation of K2P channels activity affects depolarization-induced Ca²⁺_ER uptake and release, we inhibited TASK-1 in α-cells with ML365. We found that TASK-1 channel inhibition reduced Ca²⁺_ER release induced by V_m depolarization (51) (Fig. 7, G and H), suggesting that TASK-1 facilitates α-cell Ca²⁺_ER release.

Figure 7. Reduced K_slow currents are associated with altered Ca²⁺_ER dynamics.

(A and B) Representative K_slow currents recorded from WT (A) and TALK-1 KO (B) β-cells. The peak of the K_slow tail current is indicated by the arrow. (C) Quantification of K_slow currents recorded from WT and TALK-1 KO β-cells (N = 26 cells; 4 mice per genotype). (D and E) Average whole-cell currents recorded in HEK293 cells expressing TALK-1 with intracellular buffer containing low Ca²⁺ (50 nM, black line) or high Ca²⁺ (5 µM, green line) in HEK293 (D, N = 11 cells per condition) and mouse β-cells (E, N = 15 (50 nM Ca²⁺) and 13 cells (5 µM)). (F) Depolarization-induced Ca²⁺ influx in mouse α-cells treated with vehicle or ML365. (G) AUC analysis of rising (rise) and decaying (fall) phase of Ca²⁺ influx in α-cells suggests reduced Ca²⁺-induced Ca²⁺_ER release in ML365-treated α-cells (N = 11 cells per condition). Statistical significance was determined by Student’s t-test; *P<0.05, ***P<0.005.

TALK-1 channel activity exacerbates islet ER stress

Reduced β-cell Ca²⁺_ER content is associated with ER stress and islet dysfunction in diabetes (4, 60–63). Our results indicated that TALK-1 channels were a determinant of β-cell Ca²⁺_ER concentrations, and suggested that TALK-1 channel activity could exacerbate Ca²⁺_ER depletion which leads to ER stress. Therefore, we determined whether the absence of functional TALK-1 channels impacted islet responses to the metabolic stress of a HFD. After one week of HFD feeding, the expression of genes involved in ER stress signaling did not differ between wild-type and TALK-1 KO islets (Fig. 8A). SERCA abundance is reported to change as a function of Ca²⁺_ER content (64, 65); however, we did not detect differences in the expression of mRNAs encoding SERCA2b and SERCA3 in wild-type and TALK-1 KO islets. However, after prolonged (20 weeks) exposure to a HFD, TALK-1 KO islets exhibited decreased expression of multiple ER stress genes, as well as significantly decreased expression of mRNA encoding SERCA2b and SERCA3 (Fig. 8B). The decreased SERCA expression may represent a compensatory mechanism to reduce Ca²⁺_ER overloading and maintain β-cell Ca²⁺_ER concentrations in an optimal range.

Figure 8. TALK-1 channel activity exacerbates ER stress.

(A) Reverse-transcribed RNA from islets isolated from wild-type (WT) and TALK-1 KO mice fed a HFD for 1 week was subjected to quantitative real-time PCR (qRT-PCR) to measure total Xbp1, spliced Xbp1, CHOP, BiP, Atp2a2 (SERCA2b), and Atp2a3 (SERCA3) expression (N = 4–5 mice per genotype). (B) Reverse-transcribed RNA from islets isolated from wild-type (WT) and TALK-1 KO mice fed a HFD for 20 weeks was subjected to qRT-PCR to measure total Xbp1, spliced Xbp1, CHOP, BiP, Dnajc3, Hsp90, Atp2a2 (SERCA2b), and Atp2a3 (SERCA3) expression (N = 3–4 mice per genotype). (C) INS-1 cells co-transfected with TALK-1 DN mutant, wild-type TALK-1 (WT), or TALK-1 A277E and an ATF6 promoter luciferase reporter (p5×ATF6-GL3) were treated with vehicle (VHL) (DMSO, 0.0125% v/v) or tunicamycin (Tm) (0.25 µg/mL) for 16–20 hours prior to cell lysis and luciferase assay (N = 4 independent experiments). (D) INS-1 cells were co-transfected with TALK-1 WT or TALK-1 A277E and pCMV-D4ER to measure basal Ca²⁺_ER concentrations in 11 mM glucose (N = 4 independent experiments). Statistical significance was determined by Student’s t-test; *P<0.05, **P<0.01.

We also assessed whether the T2DM-linked gain-of-function polymorphism (rs1535500) encoding TALK-1 A277E (28) interfered with ER function. ATF6 transcriptional activation occurs in response to ER Ca²⁺ depletion and protein misfolding (65, 66), which we measured in INS-1 cells (with a luciferase reporter containing five tandem repeats of ATF6 binding sites (67, 68)) after application of tunicamycin, which inhibits protein glycosylation and causes protein misfolding and ER stress. INS-1 cells expressing wild-type TALK-1 or TALK-1 A277E were significantly more susceptible to tunicamycin-induced ATF6 activation than cells expressing the non-conducting TALK-1 DN. Furthermore, TALK-1 A277E activated ATF6 to a significantly greater extent than wild-type TALK-1 (Fig. 8C). However, co-expression of TALK-1 A277E with the TALK-1 DN reduced ATF6 activation to amounts comparable to that induced by expression of TALK-1 DN alone (Fig. 8C). INS-1 cells expressing TALK-1 A277E showed reduced Ca²⁺_ER concentrations compared to cells expressing wild-type TALK-1 (Fig. 8D). Together, our findings indicate that TALK-1 channels control Ca²⁺_ER fluxes in the islet, which may regulate plasma membrane ion channel activity, electrical excitability, and Ca²⁺_ER concentrations important for protein processing (Fig. S7).

DISCUSSION

Tight regulation of β-cell Ca²⁺_ER is required to sustain insulin synthesis, metabolism, as well as intracellular Ca²⁺ signaling, and perturbations in Ca²⁺_ER handling contribute to diabetes pathogenesis. β-cell Ca²⁺_ER is controlled by several proteins including SERCAs (38, 62), IP3Rs (69), RyRs (2, 69, 70), and the translocon (71). However, the ubiquitous distribution of most Ca²⁺_ER handling proteins precludes their clinical use in treating diabetes. Here, we demonstrated that pharmacological manipulation of K2P channels can be used to control primary cell Ca²⁺_ER. Moreover, our data indicated that inhibiting TALK-1 channel activity could protect islets from ER stress induced by prolonged exposure to a HFD. These observations suggest the exciting potential of utilizing K2P channels such as TALK-1 as a therapeutic target to reduce β-cell ER dysfunction under diabetic conditions.

Ca²⁺_ER is determined by a balance of SERCA activity, Ca²⁺_ER release, and Ca²⁺_ER buffering. Ca²⁺_ER release shifts the ER membrane potential (V_m(ER)) towards the Ca²⁺_ER reversal potential (E_Ca²⁺(ER)), where net Ca²⁺_ER efflux would stop. However, the K⁺ countercurrent across the ER membrane maintains V_m(ER) positive of E_Ca²⁺(ER), facilitating Ca²⁺_ER release (16, 20, 49, 72). Our data indicate that ER TALK-1 K⁺ currents support the electrochemical driving force for Ca²⁺_ER release, consistent with our observation that TALK-1 overexpression decreased Ca²⁺_ER storage by increasing Ca²⁺_ER leak. Conversely, inhibition of TALK-1 should move the V_m(ER) closer to E_Ca²⁺(ER), resulting in reduced Ca²⁺_ER leak and increased Ca²⁺_ER (15), a prediction in accordance with the phenotype of TALK-1 KO β-cells.

Although a diminished contribution of Ca²⁺_ER leak to bulk Ca²⁺_c might be predicted to impair GSIS, TALK-1 KO islets exhibit increased glucose-stimulated Ca²⁺ influx and insulin secretion (28). These observations suggest that β-cell Ca²⁺ influx through VDCCs is increased following loss of TALK-1 channels. Ca²⁺_ER can play important roles in controlling plasma membrane channels that tune VDCC activity. For example, β-cell Ca²⁺_ER release influences insulin secretion through modulation of currents which hyperpolarize the V_m, such as K_slow, thereby indirectly controlling VDCC activity. In TALK-1 KO β-cells, reduced Ca²⁺_ER release presumably results in less K_slow activation, leading to V_m depolarization and more persistent electrical activity, culminating in a net increase in Ca²⁺_c and enhanced GSIS. Ca²⁺_ER release serves an important role in regulating 2^nd-phase Ca²⁺_c influx and insulin secretion, which are increased when SERCAs are inhibited pharmacologically and is also observed in islets lacking SERCA3 (38, 73, 74). Similarly, we find that both 2^nd-phase Ca²⁺_c and insulin secretion are increased in the absence of functional TALK-1 channels (28). These observations suggest that TALK-1 modulation of the Ca²⁺_ER handling which control 2^nd-phase insulin secretion is a key physiological function of β-cell TALK-1 channels.

As mentioned above, TALK-1 KO islets exhibit increased insulin secretion, a higher frequency of Ca²⁺_c oscillations, and increased plateau fraction (the fraction of time spent in electrically excitable periods) (28). The present study helps resolve the molecular mechanisms underlying these phenotypes. A rationale for the increased plateau fraction and oscillation frequency in TALK-1 KO islets is that ER TALK-1 channels sustain Ca²⁺_ER release, which results in greater K_slow activity and V_m hyperpolarization. Because this Ca²⁺_ER release is reduced in TALK-1 KO β-cells, K_slow activation is diminished, resulting in an increased plateau fraction. This is in accordance with observations that depletion of Ca²⁺_ER inhibits K_slow and accelerates Ca²⁺_c oscillations (53). Ca²⁺_ER concentrations also determine the activation of depolarizing store-operated currents, as demonstrated by the acceleration of V_m and Ca²⁺_c oscillations after treatment of islets with SERCA inhibitors (75–77), presumably due to inhibition of K_slow and activation of SOCE. These observations highlight that SOCE serves an important role in shaping islet Ca²⁺_c oscillations, and future studies are required to dissect the relationship between TALK-1 modulation of Ca²⁺_ER stores, SOCE, V_m and Ca²⁺_c oscillations, and insulin secretion.

ER K⁺ channels are also important for Ca²⁺_ER uptake, as demonstrated by the influence of ER-localized SK channels in regulating neuronal and cardiomyocyte ER and SR Ca²⁺ uptake (20). SK channels are predicted to preserve ER pH homeostasis through activation of an ER K⁺/H⁺ antiporter that promotes ER H⁺ entry to balance SERCA-mediated H⁺ loss during Ca²⁺ uptake (20). Although we cannot exclude a role for K2P channels in modulating SERCA function, we found reduced basal Ca²⁺_ER when TALK-1 was heterologously expressed, and conversely found increased Ca²⁺_ER in TALK-1 KO β-cells. These findings suggest that if TALK-1 controls Ca²⁺_ER uptake, it would presumably do so by inhibiting SERCA function, in contrast to SK channels, which enhance SERCA function. Any effects of TALK-1 on SERCA activity could be through indirect mechanisms modulated by Ca²⁺_ER, such as mitochondrial ATP production which energizes the β-cell SERCA pump (78).

Our observations suggested that TALK-1 activity controls Ca²⁺_ER, which impacts many aspects of β-cell function in health and disease. In addition to controlling Ca²⁺_c signals, another essential function of β-cell Ca²⁺_ER handling is to maintain insulin production and processing, which is impaired under conditions of β-cell stress induced by insulin resistance or decreased β-cell mass. A hallmark of ER stress is increased Ca²⁺_ER leak (2), which can be caused in β-cells by reductions in the abundance of proteins which affect Ca²⁺_ER such as SERCA2b (60, 62) or sorcin (68). Our data indicate that TALK-1 activity aggravated islet ER stress under conditions of increased systemic insulin demand. Moreover, our finding that T2DM-associated TALK-1 A277E channels exacerbated Ca²⁺_ER leak and ER stress responses suggests that defects in TALK-1 channel activity could contribute to islet ER dysfunction in diabetes. TALK-1 transcript abundance is reduced under conditions that cause ER stress in diabetes (such as palmitate or inflammatory cytokine treatment (79)), which may be a protective mechanism to preserve β-cell Ca²⁺_ER homeostasis by reducing Ca²⁺_ER leak. It will be important to determine how TALK-1 participates in the cellular response to other diabetes-associated ER stressors.

Mutations in other K2P channels have also been associated with various disorders that may be associated with defects in Ca²⁺_ER handling. For example, dominant-negative mutations in TASK-1 or TASK-3 result in PAH or Birk-Barel syndrome, respectively (47). We found that a PAH-linked mutation in TASK-1 (G203D) enhanced Ca²⁺_ER stores relative to wild-type TASK-1. Thus, in patients with the TASK-1 G203D mutation, disruptions in ER/SR Ca²⁺ handling may contribute to PAH (80). In pulmonary arterial smooth muscle cells, impaired Ca²⁺_ER transfer to mitochondria leads to pulmonary vascular remodeling, a defect that can be targeted with clinically used chemical chaperones to ameliorate PAH (80, 81). Inhibition of TASK-1 using the specific inhibitor A293 causes pulmonary vascular remodeling and PAH in rats, and pharmacological activation of TASK-1 protects from the development of PAH (82). TASK-3 also controls Ca²⁺_ER, and a mutation in KCNK9 (which encodes TASK-3, G236R) causes Birk-Barel syndrome, which is characterized by intellectual disability, hypotonia, and facial dysmorphism. TASK-3 is also implicated in mitochondrial function (83), highlighting the importance of determining the relationship between TASK-3 modulation of Ca²⁺_ER handling and mitochondrial function. As we found that pharmacological regulation of TASK-1 could control primary cell Ca²⁺_ER, K2P channels such as TASK-1, TASK-3 or TALK-1 could be targeted for cell-selective therapies to reduce ER dysfunction.

Not all K2P channels regulate Ca²⁺_ER, as demonstrated by our finding that neither TREK-1 channels nor TREK-2 channels affected Ca²⁺_ER homeostasis. These findings could be due to localization of these channels; TREK-1 channels are found primarily on the plasma membrane (84), whereas the subcellular localization of TREK-2 channels has not been determined. However, all K⁺ channels are assembled in the ER prior to their delivery to the plasma membrane (85). It remains to be determined how certain K2P channels (specifically, TALK-1, TASK-1, and TASK-3) regulate Ca²⁺_ER whereas others (specifically, TREK-1 and TREK-2) do not. As many K⁺ channels, such as K_ATP, require a physical interaction with the plasma membrane lipid phosphatidylinositol 4,5-bisphosphate (PIP₂) to conduct K⁺, K2P channels that are dependent on PIP₂ for activity may not function in the ER membrane. This property could be due to intrinsically low PIP₂ concentrations on the ER membrane, which limits K⁺ channel activity until they are transported to the plasma membrane (86). TREK-1 is highly sensitive to PIP₂, whereas both TASK-1 and TASK-3 channels, both of which affect Ca²⁺_ER, are insensitive to PIP₂ (87). Future studies are needed to better understand the regulatory mechanisms underlying K2P channel activity in the ER, and how these impact Ca²⁺_ER homeostasis.

In conclusion, we demonstrate that TALK-1 in the ER regulates Ca²⁺_ER handling, thus controlling K_slow activity, Ca²⁺ influx, and insulin secretion. These findings highlight a physiological function of K2P channels in the regulation of Ca²⁺_ER. K2P channels may provide cell-selective targets to modulate Ca²⁺_ER to treat the many diseases characterized by dysfunctional Ca²⁺_ER handling.

MATERIALS AND METHODS

Chemicals

Unless otherwise specified, all reagents were obtained from Sigma-Aldrich.

Mouse Models

The mice used in this study were 8–12 week-old males on a C57Bl6/J background. The generation of Kcnk16^−/− (TALK-1 KO) mice has been previously described (28). For experiments using mouse α-cells, transgenic mice expressing tdRFP specifically in α-cells were used (46). All mice used in this study were handled in compliance with protocols reviewed and approved by the Vanderbilt University Institutional Animal Care and Use Committee, according to guidelines set forth by the NIH.

Islet isolation and culture

Mouse islets were isolated using collagenase P (Roche) digestion of the pancreas and density gradient centrifugation (5). Human islets from adult nondiabetic donors (donor information is provided in Table S1) were obtained through isolation centers organized by the Integrated Islet Distribution Program. In experiments using D4ER, cells were transduced with Ad-D4ER (38) 48 hours prior to imaging. In human β-cell experiments, cells were transfected with TALK-1 DN- or mCherry-expressing plasmids (28). Islets and dispersed cells were cultured for 24–48 hours prior to experimentation (28).

Cell culture and luciferase assays

The development of TREx-293 cells with inducible expression of K2P channels has been previously described (43). To measure TALK-1 and TASK-1 expression in induced cells, lysates from TALK-1- or TASK-1-TREx-293 cells treated for 24 hours with or without tetracycline (1 µg/mL) induction were run on 4−12% Bis-Tris polyacrylamide gels (Invitrogen). The protein was then transferred to a nitrocellulose blotting membrane (BioRad) which was probed with TALK-1 (Novus Biologicals #NBP1-83071) or TASK-1 (Abcam #49433) antibodies. Equal loading of wells was assessed by stripping and reprobing membranes with a β-actin antibody (Cell Signaling Technologies #4970). Representative blots are shown in Fig. S4. For experiments comparing the effects of wild-type TALK-1 and TALK-1 DN on Ca²⁺_ER handling, HEK293 cells were transfected with pCDNA3.1 plasmids encoding these channels using Lipofectamine 3000 (Thermo Fisher) according to the manufacturer’s instructions, and imaged 48 hours post-transfection.

INS-1 (832/13) cells were cultured in RPMI 1640 supplemented with 15% FBS and penicillin-streptomycin. INS-1 cells were transfected with p5×ATF6-GL3 (Addgene #11976) and plasmids encoding wild-type TALK-1, TALK-1 A277E, or TALK-1 DN (28). Cells were incubated overnight with vehicle (DMSO) or tunicamycin (0.25 µg/mL) for 16–20 hours prior to performing a luciferase assay using the Steady-Glo Luciferase Assay System (Promega) according to the manufacturer’s instructions.

Patch clamp electrophysiology

An Axopatch 200B amplifier (Molecular Devices) was used to measure whole-cell K⁺ channel currents in the voltage-clamp mode; currents were digitized using a Digidata 1440, lowpass filtered at 1 kHz and sampled at 10 kHz. For K_slow recordings, pipettes were filled with an intracellular solution (57) containing (in mM) 28.4 K₂SO₄, 63.7 KCl, 11.8 NaCl, 1 MgCl₂, 20.8 HEPES, 0.5 EGTA (pH 7.22 with KOH) and ~0.05 mg·ml⁻¹ amphotericin B. Nuclear patch clamp experiments were preformed using the approach described by Mak and colleagues (50). Nuclei were patched in a solution containing (in mM): 150 KCl, 10 HEPES, 0.5 EGTA, 0.36 CaCl₂ (pH 7.3 with KOH). Patch electrodes were pulled to a resistance of 8–10 MΩ, loaded with recording solution, and coated with Sigmacote. Single-channel currents were lowpass filtered at 1 kHz and sampled at 50 kHz. When intracellular [Ca²⁺] was clamped, cells were recorded using the whole-cell configuration using electrodes filled with a solution containing (in mM) 140 KCl, 5 HEPES, 4 Mg·ATP, 1 EGTA, 137 µM (for 50 nM Ca²⁺ final) or 946 µM (for 5 µM Ca²⁺ final) CaCl₂, (pH 7.22 with KOH). [Ca²⁺] was determined using MAXCHELATOR software. The extracellular buffer used for islet-cells (A modified Krebs-Ringer buffer, KRB) contained (in mM) 119 NaCl, 2.5 CaCl₂, 4.7 KCl, 25 HEPES, 1.2 MgSO₄, 1.2 KH₂PO₄, 11 glucose (pH 7.35 with NaOH). The extracellular buffer used for HEK293 cells (HEK buffer) contained (in mM) 150 NaCl, 5 KCl, 2 MgCl₂, 2.5 CaCl₂, 10 HEPES, and 10 glucose (pH 7.35 with NaOH). When assessing the Ca²⁺ sensitivity of TALK-1 in β-cells, the extracellular buffer was supplemented with a cocktail of K⁺ channel inhibitors including 200 µM tolbutamide (MP Biomedicals), 10 mM tetraethylammonium chloride (Acros Organics), 100 nM apamin (Alomone Labs), 100 nM iberiotoxin (Alomone Labs), 100 nM TRAM-34 (Alomone Labs), and 10 µM nifedipine to inhibit voltage-gated Ca²⁺ channels. Cells were recorded with a voltage-clamp protocol used to assess K2P channel currents (28). Recordings were analyzed using Clampfit 10 (Molecular Devices) and Microsoft Excel software.

Calcium imaging

Mouse and human β-cells were loaded with 2 µM Fura-2 AM (Molecular Probes) and imaged as previously described (88). Cyclopiazonic acid (CPA; Alomone Labs) was used at a concentration of 50 µM; ionomycin was used at a concentration of 5 µM (Alomone Labs). Human β-cells were post-stained for insulin (88). In all experiments, cells were perfused with a flow of 2 mL·min⁻¹ at 37 °C. For analysis of mouse β-cell Ca²⁺_ER uptake and release (51), Fura-2 loaded cells were incubated for 10 minutes in KRB supplemented with 11 mM glucose, 125 µM diazoxide (Enzo) and 1.25 µM thapsigargin (Alomone Labs) or vehicle. In high-[K⁺] stimulus buffer, NaCl was reduced accordingly to maintain osmolarity. For experiments using D4ER, cells were incubated for 20 minutes in KRB containing 2 mM glucose prior to imaging.

For assays comparing the effects of expression of K2P channels in stably transduced TREx-293 cells, 30,000 cells/well were seeded to 384-well black-wall, clear-bottom, amine-coated plates (BD Biosciences). Channel expression was induced with 1 µg·ml⁻¹ tetracycline in culture medium, and the cells were cultured overnight in a 5% CO₂ incubator at 37 °C. The cells were washed using an ELx405CW plate washer (Bio-Tek Instruments, Inc.), and loaded with 4 µM Fluo-4 AM (Molecular Probes) for 45 minutes in a 5% CO₂ incubator at 37 °C. The cells were then washed with buffer supplemented with 1 mM EGTA and incubated in a 5% CO₂ incubator at 37 °C for 8 minutes before imaging. Plates were then loaded into a whole-plate kinetic-imaging Functional Drug Screening System (FDSS 6000, Hamamatsu, Bridgewater, NJ) and imaged at 37 °C as previously described (43).

When assessing the effects of TASK-1 or TASK-3 channel blockade, cells were loaded three hours before imaging with 500 nM ML365 (Tocris) or DMSO vehicle in culture medium, and in α-cells, the culture medium also contained 125 µM diazoxide. ML365 was present throughout the experiment. For high-speed imaging of α-cell Ca²⁺ influx, cells were loaded with 5 µM Fluo-4 AM for 25 minutes, followed by washing with KRB (11 mM glucose). α-cells were then patched according to the perforated patch clamp protocol described above on a Nikon Eclipse TE2000-U microscope equipped with an X-Cite 120Q widefield fluorescence light source (Excelitas Technologies) and a D-104 microscope photometer (Photon Technologies Inc.). Upon obtaining a low-leak, GΩ seal, the fluorescence light source was activated, and plasma membrane currents were recorded using the K_slow voltage-clamp protocol (57) simultaneously with Fluo-4 fluorescence. Currents and photometer signal were digitized and sampled at 10 kHz. For analysis of the effects of TALK-1 on the CPA-induced Ca²⁺_ER leak rate, TREx-293 cells transfected with wild-type TALK-1 or DN and CEPIA1-ER (37) (Addgene #58215) were permeabilized for 4 minutes in a 5% CO₂ incubator at 37 °C in an intracellular buffer containing (in mM): 140 potassium gluconate or 140 Tris base (K⁺-free), 10 HEPES, 1 EGTA, 0.432 CaCl₂, 3 Mg·ATP, with sucrose added as needed to match osmolarity (pH 7.24), and supplemented with 50 µg·ml⁻¹ digitonin (Santa Cruz). Cells were then washed for an additional 5 minutes in the appropriate buffer without digitonin prior to the start of imaging. To determine the rate constant of CPA-induced Ca²⁺ leak, the normalized data was fit to a one-phase exponential decay model using GraphPad Prism7 software. Data were analyzed using Nikon Elements, Microsoft Excel, Clampfit 10 and GraphPad Prism7 software.

Site-directed mutagenesis

The TASK-1 G203D point mutation was generated using a previously described approach (28). The sequences of oligonucleotide primers (Integrated DNA Technologies) used to create the TASK-1 G203D mutant were ACCACCATCGGCTTCGACGACTACGTGGCGCTGCAGA (forward) and TCTGCAGCGCCACGTAGTCGTCGAAGCCGATGGTGGT (reverse). PCRs were performed in 50 µL with Q5 high-fidelity DNA polymerase (New England Biolabs) with 100 ng of pCDNA3.1-KCNK3 plasmid. DNA was then incubated with 1 µL DpnI for two hours at 37 °C. Clones were sequenced to confirm mutagenesis.

Quantitative real-time PCR (qPCR)

qPCR of cDNA obtained from mouse islets was performed according to the approach described by Tong and colleagues (62). A list of primers can be found in Table S3.

Immunofluorescence

Processing and staining of paraffin-embedded mouse and human pancreas sections was performed as previously described (human donor information is provided in Table S2) (28). Sections were stained using primary antibodies against TALK-1 (Novus Biologicals #NBP1-83071; 1:175) and calreticulin (Santa Cruz #N-19; 1:125); secondary antibodies used were Alexa Fluor 488-conjugated donkey anti-rabbit (Jackson Immunoresearch #711-546-152; 1:300) and DyLight 650-conjugated donkey anti-goat (Thermo Fisher #SA5-10089; 1:250). HEK293 cells co-transfected with TALK-1a or TALK-1b (28) and ER-targeted EYFP (Addgene #56589) were washed twice with cold phosphate-buffered saline (PBS), then fixed in 4% paraformaldehyde (Electron Microscopy Sciences) for 30 minutes at 4 °C. Cells were then incubated in PBS supplemented with 0.2% bovine serum albumin (BSA), 2% normal donkey serum (NDS; Jackson Immunoresearch), and 0.05% Triton X-100 for one hour, followed by incubation in PBS containing primary antibodies against TALK-1 (1:175) and GFP (Novus Biologicals NB600-597; 1:300), 0.2% BSA, 1% NDS, and 0.1% Triton X-100, overnight at 4 °C. Following removal of the primary antibody solution, the cells were subjected to two 10-minute PBS washes, then incubated in the dark for one hour at room temperature in PBS containing 1% NDS and secondary antibodies: Alexa Fluor 488-conjugated donkey anti-rabbit (1:300) and Alexa Fluor 647-conjugated goat anti-mouse (Thermo Fisher A21237; 1:300). The secondary antibody solution was removed and the cells were subjected to three 8-minute PBS washes prior to imaging. All images were obtained using a Zeiss LSM 710 or Zeiss LSM 780 confocal laser scanning microscope. Images were analyzed using ImageJ software.

For analysis of islet cell numbers, paraffin embedded were processed as described above, and stained using primary antibodies against insulin (Dako #A0564; 1:500), somatostatin (Santa Cruz Biotechnology sc-7819: 1:250), and glucagon (Proteintech #15954-I-AP: 1:500); secondary antibodies used were Alexa Fluor 488-conjugated donkey anti-rabbit (Jackson Immunoresearch #711-546-152; 1:500), DyLight 650-conjugated donkey anti-goat (Thermo Fisher #SA5-10089; 1:250), Cy3-conjugated donkey anti-guinea pig (Jackson Immunoresearch #706-165-148; 1:500).

For analysis of high-fat diet induced islet-cell proliferation, age-matched wild-type and TALK-1 KO were placed on a high-fat diet (60% kcal/fat; Research Diets #D12492) for 10 days. Four days prior to sacrifice, mice were provided with drinking water containing BrdU (0.8 mg·ml⁻¹) supplemented with Splenda artificial sweetener (20 mg·ml⁻¹). Paraffin embedded pancreata were processed as described above, and were subjected to antigen retrieval performed in 1× NaCitrate pH 6.0, for 14 minutes in a microwave at high power, followed by cooling at room temperature in 1× NaCitrate solution for 25 minutes. Following antigen retrieval, slides were washed for 10 minutes in ddH₂O, followed by two 2-minute washes in PBS. Sections were the stained using primary antibodies against insulin (Dako #A0564; 1:500), somatostatin (Santa Cruz Biotechnology sc-7819: 1:250), glucagon (Proteintech #15954-I-AP; 1:500), and BrdU (Developmental Studies Hybridoma Bank #G3G4; 1:50). Secondary antibodies used were Alexa Flour 647-conjugated goat anti-mouse (Life Technologies #A21237: 1:250), DyLight 488-conjugated Donkey anti-mouse (Thermo Scientific #SA5-10166; 1:300), and Alexa Flour 594-conjugated Donkey Anti-Guinea Pig (Jackson Immunoresearch #706-586-148; 1:400), DyLight 650-conjugated donkey anti-goat (Thermo Fisher #SA5-10089; 1:250), Alexa Fluor 488-conjugated donkey anti-rabbit (Jackson Immunoresearch #711-546-152; 1:500). Blocking was done in a dark humidity chamber for one hour using Dako Blocking Solution (Ref # X0909). Primary antibodies were diluted to above concentrations in DAKO Antibody Diluent Solution (Ref#S3002) and incubated on the sections overnight at 4 °C. Following primary antibody incubation, slides were washed for 5 minutes in PBS twice. Secondary antibodies were diluted to the above concentrations in PBS supplemented with 5% NDS and incubated on slides in the dark for 2 hours at room temperature. Sections were then washed twice for five minutes in PBS and DAPI was added (1:1000 for 2 minutes). Following DAPI staining, sections were washed for 5 minutes in ddH20 and then mounted with a coverslip.

All sections were imaged with an Aperio ImageScope and analyzed using an algorithm developed with Aperio IndicaLabs- CytoNuclear FLv1.2 software. The algorithm is designed to take into account factors such as nuclear staining, cytoplasm radius, nucelar size, nuclear roundness, and dye fluorescence wavelength (Cy2, Cy3, or Cy5), to identify, differentiate, and count β, δ, and α cells. The algorithm was also used to count the number of β-, δ-, and α- cells on the slides labeled with Brd-U, which was further analyzed using ImageJ software and Microsoft Excel.

Statistics

The data is presented as recordings that are averaged or representative of results obtained from at least three independent cultures. All values presented are the mean ± SEM. Statistical differences between means were assessed using Student’s t-test or one-way ANOVA, as appropriate. The significance of all experimental findings presented as fold changes was assessed by performing statistical tests on log-transformed data. P<0.05 was considered as significant.

Supplementary Material

Supplemental Figures

Fig. S1. TALK-1 exhibits ER localization.

Fig. S2. Islet cell number and proliferation are not modulated by TALK-1 activity.

Fig. S3. Intracellular Ca²⁺ stores are increased in TALK-1 KO islet cells.

Fig. S4. Tetracycline-inducible expression of TALK-1 and TASK-1.

Fig. S5. TASK-3 and TASK-1 K2P channel activity alter Ca²⁺_ER concentrations.

Fig. S6. Ca²⁺_ER leak is accelerated by TALK-1 channels.

Fig. S7. Hypothetical model depicting potential molecular mechanisms of TALK-1 channel modulation of β-cell Ca²⁺_ER handling and Ca²⁺_C oscillations.

Table S1. Human islet donor characteristics

Table S2. Human pancreas donor characteristics

Table S3. Primers used for qRT-PCR