A plasma membrane-associated glycolytic metabolon is functionally coupled to K_ATP channels in pancreatic α and β cells from humans and mice

Graphical abstract

In brief

Ho et al. provide evidence that glycolytic enzymes, including lactate dehydrogenase, form a plasma membrane-associated metabolon with intrinsic ATP/ADP and NAD⁺/NADH cycles. The subcellular location of this complex allows both ATP-consuming and ATP-producing enzymes to locally control the ATP-sensitive K⁺ channel in human and mouse pancreatic α and β cells.

SUMMARY

The ATP-sensitive K⁺ (K_ATP) channel is a key regulator of hormone secretion from pancreatic islet endocrine cells. Using direct measurements of K_ATP channel activity in pancreatic β cells and the lesser-studied α cells, from both humans and mice, we provide evidence that a glycolytic metabolon locally controls K_ATP channels on the plasma membrane. The two ATP-consuming enzymes of upper glycolysis, glucokinase and phosphofructokinase, generate ADP that activates K_ATP. Substrate channeling of fructose 1,6-bisphosphate through the enzymes of lower glycolysis fuels pyruvate kinase, which directly consumes the ADP made by phosphofructokinase to raise ATP/ADP and close the channel. We further show the presence of a plasma membrane-associated NAD⁺/NADH cycle whereby lactate dehydrogenase is functionally coupled to glyceraldehyde-3-phosphate dehydrogenase. These studies provide direct electrophysiological evidence of a K_ATP-controlling glycolytic signaling complex and demonstrate its relevance to islet glucose sensing and excitability.

INTRODUCTION

The ATP-sensitive K⁺ (K_ATP) channel has a critical role in controlling insulin and glucagon release from pancreatic islet endocrine cells, which are essential for the regulation of blood glucose homeostasis. K_ATP channels are composed of pore-forming Kir6.2 subunits that are inhibited by ATP and regulatory SUR1 subunits that are activated by MgADP, such that elevation of the sub-plasma membrane ATP/ADP ratio (ATP/ADP_pm) leads to K_ATP channel closure.^1–3 In pancreatic β cells, closure of K_ATP channels induces membrane depolarization, resulting in the activation of voltage-sensitive Ca²⁺ channels that trigger insulin exocytosis.^4–6 In pancreatic α cells, K_ATP channels exist predominantly in the closed state, while full closure during hyperglycemic conditions inactivates voltage-dependent Na⁺ channels and subsequently suppresses Ca²⁺-stimulated glucagon release.^7,8 Thus, in both cell types, the K_ATP channel serves as a metabolic sensor for the plasma membrane that determines cellular excitability.

While it has been historically assumed that K_ATP channel closure is mediated by mitochondrial ATP generated via oxidative phosphorylation, recent evidence suggests that β cell K_ATP channels are primarily regulated by glycolytic ATP.⁹ The timing of oxidative phosphorylation, which is activated by ADP and reinforced by Ca²⁺ after membrane depolarization, does not align with the timing of K_ATP closure during pulsatile insulin secretion.^10–14 However, the activity of pyruvate kinase (PK), which converts ADP and PEP into ATP and pyruvate in the last step of glycolysis, correlates with the ATP/ADP rise preceding membrane depolarization and calcium influx.^10–14 It was recently suggested that a plasma membrane-associated pool of PK is sufficient to close K_ATP channels in human and mouse β cells.¹³ Genetic studies of PK deficiency further suggest that mitochondrially derived ATP buffers the cytosolic ATP/ADP ratio but may not be the primary driver of ATP/ADP_pm or K_ATP closure.¹⁵ It remains unclear whether compartmentalized regulation of K_ATP exists in α cells from humans or mice.

Compartmentation of glycolytic enzymes has been hypothesized to provide a highly efficient mechanism for localized regulation of the ATP/ADP ratio.^16–19 To date, the best evidence that glycolysis supports substrate channeling comes from protozoa—a special case where all nine glycolytic enzymes are contained within a membrane-bound organelle called the glycosome.^20–22 Under stress conditions, yeast are capable of assembling glycolytic enzymes into non-membrane-bound granules termed G-bodies that accelerate glucose consumption.^23,24 In mammalian cells, lower glycolysis is known to regulate a variety of plasma membrane ion channels and pumps that exist in regions of high ATP consumption, including cardiac K_ATP channels.^16,25–29 However, it is not known whether upper glycolysis is also present and participates in K_ATP regulation, leaving the concept of a glycolytic metabolon elusive.

Here, we performed electrophysiological measurements of K_ATP activity in native α and β cell plasma membranes from both humans and mice. The excised patch-clamp approach we employed not only identifies the plasma membrane localization of the endogenous enzymes but also indicates their functional relationship to the activity of native K_ATP channels. Our experiments provide direct evidence that the enzymes of upper and lower glycolysis locally regulate K_ATP channels on the plasma membrane. Glucokinase (GK) and phosphofructokinase (PFK) provide ADP to activate K_ATP channels, while the enzymes of lower glycolysis utilize substrate channeling to produce ATP that closes K_ATP. We further show the presence of two coupled reactions: (1) lactate dehydrogenase (LDH), which facilitates local NAD⁺/NADH recycling for GAPDH, and (2) the direct transfer of ADP generated by PFK to PK, which ultimately closes K_ATP channels. These results identify a plasma membrane-associated glycolytic metabolon that locally signals via K_ATP channels to orchestrate α and β cell function.

RESULTS

The plasma membrane-associated enzymes of lower glycolysis close K_ATP channels

We previously established that PK is present on the plasma membrane of human and mouse β cells where it raises ATP/ADP_pm sufficiently to close K_ATP channels.^13,15 Given that hormone secretion from both α and β cells relies on K_ATP channel regulation, we compared the effectiveness of substrate-driven K_ATP closure by PK in the two cell types. In dispersed mouse islets, the larger β cells were easily distinguished from α cells by cell size^7,30,31; however, it is likely that a small fraction of δ cells (which comprise ~5% of mouse islets³²) were present in the α cell classification. Anti-NTPDase3-coated magnetic beads were used to achieve enrichments of ~80% for human α and β cells^33–35 (Figures S1A and S1B). Staining showed only a small number of somatostatin-positive cells in either anti-NTPDase3 pull-down or flow-through fraction (Figures S1C and S1D).

As in our previous studies, we utilized the inside-out mode of excised patch clamp to study the effect of local ATP production by endogenous plasma membrane-associated PK on K_ATP channel activity (Figure 1A). K_ATP channels were identified by closure upon the addition of bath solution containing 1 mM ATP and reopening after switching to a high ADP solution containing 0.1 mM ATP and 0.5 mM ADP. In the presence of ADP, increasing concentrations of the PK substrate PEP (0.25, 1, and 5 mM) led to the dose-dependent closure of K_ATP channels in both mouse α and β cells (Figures 1B and 1C). Channel closure occurred despite the continuous replacement of 0.5 mM ADP in the bath solution, which complexes with free Mg²⁺ to form MgADP that opens the K_ATP channel.^2,3 Thus, plasma membrane-compartmentalized PK must be sufficiently close to the K_ATP channel to locally deplete MgADP and increase ATP to close the channel. An alternative interpretation of these data is that PK-derived pyruvate is oxidized by residual patch-associated mitochondria that generate ATP and close K_ATP channels independently of glycolysis.³⁶ However, pyruvate and ADP were unable to induce K_ATP channel closure, and the presence of ATP synthase inhibitor (1 μg/mL oligomycin) did not affect K_ATP channel inhibition by PEP and ADP (Figures S2A–S2F). Consistently, knockdown of β cell PK activity abolished PEP-mediated K_ATP closure.¹⁵ We conclude that glycolytic PK, rather than mitochondrial oxidative phosphorylation, is indeed responsible for channel closure in these assays. In most experiments, the PEP-dependent reduction in channel activity was caused by a lowered channel opening frequency without any change in event duration, both of which are shown for all treatments in Table S1 and the accompanying data. Significant heterogeneity in the response to PEP treatment was observed between α and β cells (Figures 1B and 1C).

Figure 1. Lower glycolytic enzymes raise ATP/ADP to close α and β cell K_ATP channels.

(A–C) PEP (0.25, 1, and 5 mM) in the presence of high ADP (0.5 mM ADP, 0.1 mM ATP) led to the production of ATP by PK (A), thereby closing K_ATP channels in excised plasma membrane from mouse β cells (B) and mouse α cells (C).

(D–F) In the high ADP condition, the chain reaction of PGM, ENO, and PK produced ATP and closed K_ATP channels (D) upon addition of 5 mM 3PG in mouse β cells (E) and 1 mM and 5 mM 3PG followed by 5 mM PEP in mouse α cells (F).

(G–K) Lower glycolytic activity (from ALDO to PK) utilized 5 mM FBP as the substrate for ALDO and 5 mM NAD⁺ and P_i (in the form of KH₂PO₄) as the substrate for GAPDH in high ADP condition to produce ATP (G), closing K_ATP channels in mouse β cells (H), mouse α cells (I), human β cells (J), and human α cells (K). Example traces show closure of K_ATP channel under 1 mM ATP, reopening after switching to high ADP condition, followed by addition of glycolytic substrates.

Graphs quantify channel activity (power) normalized to (B–C, E, and F) high ADP condition or (H–K) high ADP +5 mM FBP condition. Data are from at least 3 mice or 3 human donors and shown as mean ± SEM. #p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by ratio paired Student’s t test (E and H–K) or one-way ANOVA with post-hoc Sidak multiple comparisons (B, C, and F). Concentrations are in mM unless otherwise noted.

See also Figure S2 and Table S1.

We next tested whether the two glycolytic enzymes upstream of PK, phosphoglycerate mutase (PGM) and enolase (ENO), could locally provide PEP for PK-mediated K_ATP closure on the plasma membrane (Figure 1D). The application of the PGM substrate 3-phosphoglycerate (3PG; 5 mM) to a high ADP solution reduced K_ATP channel activity in mouse β cells (Figure 1E) with similar efficacy to 5 mM PEP (Figure 1B). In mouse α cells, we observed no effect of 1 mM 3PG but significant K_ATP closure with 5 mM 3PG (Figure 1F). The further activation of PK with 5 mM PEP had no further effect. ENO is known to be competitively inhibited by a combination of Mg²⁺, F⁻, and inorganic phosphate (P_i).^37,38 As a control, in mouse β cells, 3PG-dependent K_ATP closure was reduced from 54% (Figures 1E) to 20% in the presence of the ENO competitive inhibitor (5 mM kF, 5 mM KH₂PO₄, and Mg²⁺) (Figures S2G and S2H; p < 0.01 by unpaired t test with the same condition in Figure 1E). These findings indicate strong substrate channeling between the PGM-ENO-PK enzymes that provide substrate for PK to locally regulate K_ATP activity.

In cardiac myocytes, the glycolytic enzymes between aldolase (ALDO) and PK have been described to regulate K_ATP channels.²⁶ For comparison, we assessed K_ATP channel closure induced by the metabolism of the ALDO substrate FBP, which subsequently requires NAD⁺ and P_i at the GAPDH step in order to produce ATP in the downstream PGK and PK reactions (Figure 1G). Although the addition of 5 mM FBP alone led to variable changes in K_ATP channel activity, the addition of equimolar NAD⁺ and P_i (in the form of KH₂PO₄) led to strong inhibition of K_ATP channel activity in both mouse β and α cells (Figures 1H and 1I), indicating that ATP production by lower glycolysis is coupled with K_ATP channel closure on the plasma membrane. Bath application of FBP to human β and α cell plasma membranes also led to variable effects on K_ATP channels. Here, too, the further addition of NAD⁺ and P_i resulted in the inhibition of K_ATP channel activity in both β and α cells (Figures 1J and 1K), indicating that the enzymes encompassing ALDO to PK are present on the plasma membrane, where they function to regulate K_ATP channel activity at the ATP-producing steps.

The plasma membrane-associated enzymes of upper glycolysis open K_ATP channels

Multiple enzymes of upper glycolysis (hexokinase and glucose-6-phosphate isomerase) have been found by immunoprecipitation mass spectrometry to associate with the Kir6.2 subunits of the K_ATP channel.²⁶ However, it is not yet known whether upper glycolysis regulates K_ATP channel activity. We therefore assessed if ADP production by PFK could oppose K_ATP closure by a high ATP bath solution (1 mM ATP and 0.075 mM ADP) (Figure 2A). K_ATP activity is mostly inhibited in the high ATP condition; however, this was occasionally followed by delayed recovery. We controlled for this recovery by holding the condition for at least 200 s (after which no further changes were observed) and only collecting data at the end of each treatment. The application of fructose 6-phosphate (F6P; 10 mM) reversed K_ATP inhibition in mouse β and α cells (Figures 2B and 2C), suggesting that MgADP reached a sufficient level in the K_ATP channel microcompartment to activate the nucleotide-binding domain of SUR1.³⁹ The K_ATP channel-opening effect of F6P was found in ~40% of human β cells and ~60% of human α cells, but the level of α cell activation was approximately 3 times larger on average than β cells (Figures 2D and 2E).

Figure 2. Upper glycolytic enzymes produce ADP that activates K_ATP channels.

(A–E) In the high ATP condition (1 mM ATP, 0.075 mM ADP), addition of 10 mM F6P led to production of ADP by PFK (A), thereby opening K_ATP channels in mouse β cells (B), mouse α cells (C), human β cells (D), and human α cells (E).

(F–J) Addition of 10 mM glucose in high ATP (F) produced little effect in mouse β and α cells (G-H) but led to significant K_ATP activation through ADP production by GK in human β and α cells (I and J). Example recordings show K_ATP inhibition in the high ATP condition, followed by the addition of substrates for PFK and GK. Graphs quantify channel activity (power) normalized to high ATP condition.

Data are from at least 3 mice or 3 human donors and shown as mean ± SEM. #p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by ratio paired Student’s t test. Concentrations are in mM unless otherwise noted.

See also Table S1.

We next tested the coupling between GK-derived ADP and the K_ATP channel (Figure 2F). In the presence of a high ATP/ADP ratio (1 mM ATP and 0.075 mM ADP), the addition of 10 mM glucose to mouse β and α cell plasma membranes led to significant K_ATP channel reopening in approximately 20% of the experimental replicates (Figures 2G and 2H). In contrast, human islet cell membranes exhibited clear K_ATP opening in response to glucose, which was sufficient to overcome the high ATP/ADP ratio present in the bath solution (Figures 2I and 2J). Together, these data indicate that the upper glycolytic enzymes are indeed present on the plasma membrane and capable of activating K_ATP channels.

Although lactate activates K_ATP, the presence of plasma membrane-associated LDH does not impede the ability of PK to close K_ATP channels

Lactate is a known metabolite activator of K_ATP channels.^35,40–42 Sarcolemmal K_ATP channels in inside-out excised patches are activated by 20 mM lactate, an effect that is reversed by the addition of equimolar NAD⁺ to remove lactate by the LDH reaction.⁴¹ Because the circulatory level of lactate does not rise above 10 mM except during intense exercise,⁴³ we tested whether a lower concentration of lactate (2 mM) could produce the same effect (Figure S3A). Lactate application to mouse β cell plasma membranes in the high ATP condition (1 mM ATP and 0.075 mM ADP) had a heterogeneous effect, increasing K_ATP activity significantly in 3 of 11 experiments (Figure S3B). However, in mouse α cell membranes, we observed significant activation of K_ATP channels at 2 mM lactate (Figure S3C). The addition of excess NAD⁺ (5 mM) abolished the activating effect of lactate, indicating the presence of LDH on α cell plasma membrane. In excised patches from human islet cells, we observed dose-dependent opening of K_ATP channels after the addition of 2 and 5 mM lactate in both α and β cells; however, the addition of 5 mM NAD⁺ had no further effect (Figures S3D and S3E). In summary, exogenous lactate has a strong K_ATP channel-opening effect in both mouse and human α and β cells, and we confirmed plasma membrane LDH activity in mouse α cells (see next section for the effect of LDH in β cells).

Given the ability of lactate to open K_ATP channels and hyperpolarize mouse α cells that express MCT1 pyruvate/lactate transporters,³⁵ we next tested whether exogenous lactate at concentrations mimicking circulatory levels is potent enough to oppose PK-mediated K_ATP channel closure in mouse α cell membranes. The application of 2 mM lactate was able to reopen K_ATP channels inhibited by 2 mM PEP and 0.5 mM ADP (Figures S4A and S4B). However, it is unlikely that K_ATP channels are regulated by LDH-derived lactate since the addition of NADH (2 mM) was unable to reverse K_ATP closure by PK (Figures S4C and S4D). Pyruvate (2 mM) also activated K_ATP channels in a subset of α cell plasma membranes, and again, equimolar NADH had no additional effect (Figures S4E and S4F). These data argue that, despite the presence of LDH activity in the K_ATP microcompartment, LDH-mediated lactate production is insufficient to overcome K_ATP channel closure by PK even in mouse α cells with high LDH activity.

LDH and GAPDH facilitate a plasma membrane-associated NAD⁺/NADH cycle that supports K_ATP channel closure by PK

Regeneration of NAD⁺ is necessary for the continuous oxidation of glucose at the GAPDH step. We hypothesized the formation of a glycolytic metabolon that not only facilitates the channeling of metabolites between enzymes but also allows NAD⁺/NADH cycling within the K_ATP channel microcompartment to maintain redox balance. In order to test whether sufficient NAD⁺ can be generated from glycolytic intermediates to support ATP generation by lower glycolysis, we monitored K_ATP activity in response to FBP metabolism. In this experiment, 5 mM NADH was provided for NAD⁺ production, 5 mM pyruvate was provided as the substrate for LDH, and 5 mM P_i was provided for the GAPDH reaction (Figure 3A). The observed reduction in K_ATP channel activity in both human and mouse α and β cells (Figures 3B–3E), which we interpret to be a result of FBP metabolism by lower glycolysis leading to an increase in ATP/ADP_pm, was very similar to when FBP, NAD⁺, P_i, and ADP were provided directly (Figures 1H–1K). These results suggest that there must exist plasma membrane-bound NADH oxidases that provide NAD⁺ locally for glycolysis. To further determine whether LDH participates in NAD⁺ regeneration through the conversion of pyruvate to lactate in mouse β cells, the combination of FBP, NADH, and P_i was added first, followed by the addition of pyruvate (Figure S4G). K_ATP channel closure was enhanced upon pyruvate addition, indicating that LDH contributes to the source of NAD⁺ that is used for ATP production from FBP (Figures S4H and S4I). These findings suggest that plasma membrane-associated NADH oxidases (including LDH) and GAPDH facilitate a local NAD⁺/NADH redox cycle within the K_ATP channel microcompartment.

Figure 3. The ADP produced by PFK is used directly by PK, and NAD⁺ produced by plasma membrane-associated NADH oxidases is used directly by GAPDH to sustain ATP production.

(A–E) Addition of 5 mM pyruvate (Pyr) and NADH allows NADH oxidation to provide NAD⁺ at the GAPDH step to metabolize 5 mM FBP (substrate for ALDO) to ATP that closes K_ATP channels in mouse and human β and α cells.

(F–J) Addition of F6P (5 mM in mice, 1 mM in humans) under high ATP condition (1 mM ATP, 0.075 mM ADP) led to activation of K_ATP channels; the subsequent addition of PEP (5 mM in mice, 2 mM in humans) did not significantly alter K_ATP activity in mouse α cells (H) or human β cells (I) but led to K_ATP inhibition through PK-generated ATP in mouse β cells (G) and human α cells (J).

Example traces show K_ATP inhibition in 1 mM ATP and reopening in the high ADP condition (0.5 mM ADP, 0.1 mM ATP) (B–E) or K_ATP inhibition in the high ATP condition (G–J), followed by the addition of test substrates. Graphs quantify channel activity (power) normalized to high ADP +5 mM FBP condition (B–E) or high ATP condition (G–J). Data are from at least 3 human donors and shown as mean ± SEM. #p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by ratio paired Student’s t test. Concentrations are in mM unless otherwise noted.

See also Figures S3 and S4 and Table S1.

The ADP produced by PFK is utilized directly by PK to close K_ATP channels

As ATP/ADP_pm rise and flux through PK slows, there is a greater need to provide ADP to the PK reaction to sustain glucose oxidation. Because the GK and PFK reactions in upper glycolysis consume ATP, we hypothesized that one or both enzymes could act as a direct source of ADP substrate for PK (Figure 3F). To test this, we first drove ADP production by the PFK reaction using F6P (5 mM in mice, 1 mM in humans), resulting in the opening of K_ATP channels under a high ATP/ADP ratio (1 mM ATP and 0.075 mM ADP) in the bath solution (Figures 3G–3J). In these experiments, the F6P concentration was much lower than in Figures 2B–2E, reflecting a more physiological level. The subsequent application of PEP (5 mM in mouse, 2 mM in human) did not reduce KATP activity back to control level in mouse α cells (Figure 3H) or human β cells (Figure 3I), suggesting that neither the available 0.075 mM ADP nor the ADP produced from the PFK reaction was sufficient to drive the PK reaction. However, the subsequent addition of PEP closed KATP channels in mouse β cells and human α cells (Figures 3G and 3J), suggesting that here, the PFK and PK reactions are coupled in the regulation of KATP channels.

DISCUSSION

Our findings show that a glycolytic metabolon—reflecting a functional signaling machinery but not necessarily a unitary complex—is localized to the plasma membrane of α and β cells, where it locally controls the activity of K_ATP channels. We detected K_ATP channel regulation by the rate-controlling enzymes of upper and lower glycolysis as well as plasma membrane-associated NADH oxidases including LDH. Efficient substrate channeling of FBP was observed between the enzymes of lower glycolysis, in addition to the direct transfer of ADP between the PFK and PK reactions. Furthermore, we elucidated a plasma membrane NAD⁺/NADH cycle mediated by NAD⁺ channeling between NADH oxidases and GAPDH, including coupling between LDH and GAPDH in mouse β cells. Despite small differences, this strategy of compartmentalized K_ATP channel regulation is mostly conserved between humans and mice. In all treatments, we routinely observed variability between cells. The well-documented metabolic heterogeneity within populations of α and β cells,^44–47 as well as non-ubiquitous distribution of cellular metabolons⁴⁸ (i.e. microdomains), are both plausible explanations for the observed cell-to-cell variability.

We found that two enzymes in upper glycolysis, GK and PFK, locally activate K_ATP channels by raising the level of ADP in sub-plasma membrane space. ADP then serves as a direct substrate for ATP generation by PK. While PGK is also present and provides a source of ATP, this near-equilibrium enzyme lacks metabolic control.⁴⁹ Although glycolysis consumes 2 and generates 4 ATP, this does not necessarily imply that the enzymes of upper glycolysis are irrelevant to K_ATP channel regulation. In β cells, the coupling efficiency of upper and lower glycolysis can be modified by the siphoning of substrates. In individuals with type 2 diabetes, after GK produces ADP, glucose 6-phosphate is shuttled toward glycogen synthesis and thus decreases glucose-stimulated insulin secretion through reduced ATP output.⁵⁰ Similarly, glycerol 3-phosphate phosphatase (G3PP) has been shown to operate a glycerol shunt to curtail hyperinsulinemia by removing dihydroxyacetone phosphate (DHAP) from glycolysis, which can reduce ATP generation up to 50%.^51,52 It is yet unclear whether these metabolic shunts are present in the plasma membrane or whether they occur in another subcellular compartment.

In addition to glycolysis, NADH oxidase activity was found to be present in the K_ATP microcompartment that efficiently drove NAD⁺ regeneration to support glycolysis at the GAPDH step in both human α and β cell membranes. In mouse β cells, we showed that LDH activity contributed to this plasma membrane-associated NAD⁺/NADH cycle. The identities of other NADH oxidases were not tested directly in our assays. However, given that the combination FBP/NADH/P_i was able to close K_ATP channels, the NADH-dependent enzyme glycerol 3-phosphate dehydrogenase (GPD1) becomes a likely candidate to explain the strong K_ATP inhibition observed. Indeed, GPD1 regenerates NAD⁺ and is capable of siphoning the DHAP that is produced downstream of FBP metabolism by the ALDO reaction.^51,52 Further work is needed to ascertain whether GPD1 or other NADH oxidases are responsible for K_ATP regulation.

In support of previously published work,^35,40–42 we found that lactate opens K_ATP channels and opposes PK-mediated K_ATP closures in mouse α cells. Since lactate can reach millimolar levels in the blood, and since α cells express a high level of MCT pyruvate/lactate transporters (Slc16a1),³⁵ lactate regulation of K_ATP channels is expected to occur in vivo. However, lactate derived from glycolytic pyruvate had no impact on K_ATP channels in our experiments. In β cells, Ldha is considered a “disallowed gene.”^53–55 However, human β cells express two isoforms of LDH (A/B),⁵⁶ and ectopic expression of MCT1 is sufficient for lactate-induced insulin secretion that mediates exercise-induced hyperinsulinemic hypoglycemia.⁵⁷ Our results further provide evidence that lactate production provides NAD⁺ to support K_ATP channel closure. We previously reported glucose-dependent lactate production and lactate oscillations in mouse islet β cells,⁵⁸ showing the physiological relevance of these findings.

Limitations of the study

Pancreatic endocrine cells, like cardiac myocytes, may share K_ATP as the anchor point for lower glycolysis; however, a diversity of membrane scaffolds have been uncovered.^59–62 Disrupting this association would help the islet field determine whether stimulus-secretion coupling remains efficient without glycolytic compartmentation. In α cells, glucose is known to suppress glucagon secretion, a process that involves the low-affinity enzyme GK.^63,64 Consequently, the glycolytic regulation of K_ATP is most likely to be important in the high glucose condition, as it is in β cells. In addition, we do not yet know the relative importance of glycolytic versus mitochondrial K_ATP regulation in islet cells from individuals with diabetes. While we attempted to utilize frozen islet samples from human donors with type 2 diabetes, the thawed tissues were not viable for excised patch-clamp experiments after shipping. Nonetheless, our prior work suggests the potential of small-molecule PK activators for the treatment of diabetes based on the local coupling between PK and K_ATP channels.^13,65 Although current research in pancreatic hormone secretion and diabetes largely overlooks the distinction between metabolism in the bulk cytosol and microcompartments, partly due to a lack of suitable tools, our results highlight the need for further work on metabolic compartmentation that could become the basis for novel therapeutic strategies.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Matthew Merrins (mmerrins@medicine.wisc.edu).

Materials availability

• This study did not generate new unique reagents.

Data and code availability

• All data reported in this paper will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mouse islet preparations

C57BL/6J male mice were ordered from the Jackson Laboratory, housed in cages of 1–4 animals at 21–23°C, and maintained on a 12-h light/dark cycle (light on from 6:00 AM to 6:00 PM). Chow diet and water were provided as needed. The mice were sacrificed by CO₂ asphyxiation and cervical dislocation at 12–18 weeks old for analysis. Mouse pancreas was inflated through the common bile duct with Hanks’ balanced salt solution (HBSS, Sigma H1641) containing 0.5 mg/mL collagenase enzyme (Sigma C5138) and 0.2 mg/mL BSA (Sigma A8806). After excision, inflated pancreas was placed into a glass vial with 5 mL of the HBSS/BSA/collagenase solution and shaken using an orbital shaking water bath (ThermoFisher SHKA7000) at 300 rpm, 37°C, for 20 min. The digested pancreas was subsequently pelleted (50 g, 2 min), resuspended by mild vortex and washed three times with cold HBSS/BSA solution. Individual islets were hand-picked until clear of acinar tissues. Islets were cultured in RPMI-1640 supplemented with 10% (v/v) fetal bovine serum (ThermoFisher A31605), 1% (v/v) penicillin-streptomycin (Fisher Scientific 15070063), and 2 mM L-glutamine (Fisher Scientific 25030081). All procedures involving animals were approved by the Institutional Animal Care and Use Committees of the University of Wisconsin-Madison and the William S. Middleton Memorial Veterans Hospital, and followed the NIH Guide for the Care and Use of Laboratory Animals (eighth ed. The National Academies Press. 2011.).

Human islet preparations

Human islets of 16 male and 5 female normal donors, from 9 to 74 years of age, were obtained from the Alberta Diabetes Institute IsletCore, the Integrated Islet Distribution Program (IIDP), and ProdoLabs. Informed consent for the research use of islets was collected from the relatives of donors. This study used data from the Organ Procurement and Transplantation Network (OPTN). The OPTN data system includes data on all donor, wait-listed candidates, and transplant recipients in the U.S., submitted by the members of the OPTN. The Health Resources and Services Administration (HRSA), U.S. Department of Health and Human Services provides oversight to the activities of the OPTN contractor. The data reported here have been supplied by UNOS as the contractor for the OPTN. The interpretation and reporting of these data are the responsibility of the author(s) and in no way should be seen as an official policy of or interpretation by the OPTN or the U.S. Government. The age, sex, body mass index, and %HbA1c of each donor are provided in Table S2. All studies involving the use of human islets were approved by the University of Wisconsin Madison Institutional Review Board. Received human islets were cultured in glutamine-free CMRL supplemented with 10 mM niacinamide and 16.7 μM zinc sulfate (Sigma), 1% ITS supplement (Corning), 5 mM sodium pyruvate, 1% Glutamax, 25 mM HEPES (American Bio), 10% heat-inactivated FBS and antibiotics (10,000 units/mL penicillin and 10 mg/mL streptomycin).

METHOD DETAILS

Islet dispersion

Islets were washed briefly with VERSENE solution (ThermoFisher 15040066), and dissociated using TrypLE Select Enzyme solution (ThermoFisher 12563011) at 37°C, with occasional trituration. For human islets, the dissociated cells were purified into β cell and non-β cell fractions using mouse anti-human NTPDase3 antibodies (Ectonucleotidases antibodies hN3-B3S) and CELLection Pan Mouse IgG kit (ThermoFisher 11531D).³⁵ Cells were plated on sterilized glass shards and cultured overnight at 37°C before experiments. For mouse cells, culture media was RPMI-1640 supplemented with 10% (v/v) fetal bovine serum (ThermoFisher A31605), 1% (v/v) penicillin-streptomycin (Fisher Scientific 15070063), and 2 mM L-glutamine (Fisher Scientific 25030081). For human cells, the media was adjusted to contain 8 mM glucose.

Human dispersed cell staining

Dispersed cells from anti-NTPDase3 pull-down and flow-through fractions were fixed and double stained for insulin and glucagon (both fractions), insulin and somatostatin (pull-down fraction) or glucagon and somatostatin (flow-through fraction). Primary antibodies used were guinea pig anti-insulin (Fitzgerald #20-IP35, 1:1000), rabbit anti-glucagon (Cell Signaling #2760, 1:200), and mouse anti-somatostatin (Santa Cruz sc-55565, 1:100). Secondary antibodies used were goat anti-rabbit Alexa Fluor 488 conjugate (Cell Signaling #4412, 1:500), Cy3 AffinityPure donkey anti-guinea pig (Jackson ImmunoResearch #706–165–148, 1:500), goat anti-mouse Alexa Fluor Plus 488 conjugate (Thermofisher A32723, 1:500), goat anti-mouse Alexa Fluor 594 Affinity Pure conjugate (Jackson ImmunoResearch #115–585–020, 1:500).

Electrophysiology

A HEKA Instruments EPC10 patch-clamp amplifier was used for the registration of current. Intracellular recording electrodes made of borosilicate glass (Harvard Apparatus, Holliston, MA) were polished by a microforge (Narishige MF-830) to the final tip resistance of 4–10 MOhm. The pipette solution contained (in mM): 10 sucrose, 130 KCl, 2 CaCl₂, 10 EGTA, 20 HEPES, pH 7.2, adjusted with KOH. Recordings were made at room temperature. Briefly, gigaseals (>2.5 GOhm) were established in extracellular bath solution (in mM): 140 NaCl, 5 KCl, 1.2 MgCl₂, 2.5 CaCl₂, 0.5 glucose, 10 HEPES, pH 7.4, adjusted with NaOH and held at −50 mV before excision into inside-out configuration. Equilibrium solutions with K⁺ as the charge carrier were used for recording. The recording bath solution (control) contained (in mM): 130 KCl, 2 CaCl₂, 10 EGTA, 1.0 free Mg²⁺, 10 sucrose, 20 HEPES, pH 7.2 with KOH. ATP, ADP, and metabolites were added to the control solution and pH values were adjusted with KOH to 7.2. Data was filtered online at 1 kHz with a Bessel filter and analyzed offline using Clampfit 10.7 software (Molecular Devices). During each condition, recordings are held long enough for the activity to stabilize, such that no further changes were observed. After that, a 60-s window at the end of each condition was used to analyze K_ATP activity in terms of power, frequency, and duration, all of which were normalized to control conditions (typically the high ATP or high ADP condition) as indicated in the text. Akin to area under the curve, power is calculated as the product of the channel amplitude and single event duration, summed over all channels. Consequently, power is a proxy for the total number of ions passing through the channel, as is referred to as channel activity in the text. The characteristic K_ATP channel current amplitude was used to identify K_ATP-associated events in our analysis, and thus remain constant. Frequency reflects the number of opening events throughout the analysis window while event duration measures the average length of time before closure occurs. The latter two parameters, which are reported in Table S1, only account for events that at least open or close during the window of analysis (i.e. they do not include always-active events) and thus are underestimations for conditions that have many always-opened channels.

QUANTIFICATION AND STATISTICAL ANALYSIS

The numbers of biological replicates (from mice or human donors) are indicated in the figure legends. Statistical analysis was performed in GraphPad Prism software v9.0.0 by ratio paired t test or one-way ANOVA with post-hoc Sidak multiple comparisons. Data are shown as mean ± SEM. Data are considered statistically significant when #p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 or ns (not significant).

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse anti-human NTPDase3 antibodies	Ectonucleotidases antibodies	Cat#hN3-B3S; RRID:AB_2752250
Guinea pig polyclonal anti-insulin	Fitzgerald Industries International	Cat#20-IP35; RRID:AB_231771
Rabbit polyclonal anti-glucagon	Cell Signaling	Cat#2760; RRID:AB_659831
Mouse monoclonal anti-somatostatin	Santa Cruz	Cat#sc-55565; RRID:AB_831726
Anti-rabbit IgG (H + L), F(ab’)2 Fragment (Alexa Fluor® 488 Conjugate)	Cell Signaling	Cat#4412; RRID:AB_1904025
Cy™3 AffiniPure Donkey Anti-Guinea Pig IgG (H + L)	Jackson ImmunoResearch	Cat#706–165-148; RRID:AB_2340460
Goat anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ Plus 488	ThermoFisher	Cat#A32723; RRID:AB_2633275
Alexa Fluor® 594 AffiniPure Goat Anti-Mouse IgG (H + L)	Jackson ImmunoResearch	Cat#115–585-020; RRID:AB_2338874
Biological samples
Human donors are listed in Table S2	Alberta Diabetes Institute Isletcore, IIDP, and ProdoLabs	RRIDs are listed in Table S2
Chemicals, peptides, and recombinant proteins
Collagenase enzyme	Sigma Aldrich	Cat#C5138
Versene solution	ThermoFisher	Cat#15040066
TrypLE Select Enzyme solution	ThermoFisher	Cat#12563011
Adenosine 5′-triphosphate magnesium salt	Sigma Aldrich	Cat#A9817
Adenosine 5′-diphosphate monopotassium salt dehydrate	Sigma Aldrich	Cat#A5285
Phospho(enol)pyruvic acid monopotassium salt	Sigma Aldrich	Cat#860077
D-(−)-3-Phosphoglyceric acid disodium salt	Sigma Aldrich	Cat#P8877
D-Fructose 1,6-bisphosphate trisodium salt hydrate	Sigma Aldrich	Cat#F6803
β-Nicotinamide adenine dinucleotide sodium salt	Sigma Aldrich	Cat#N0632
D-Fructose 6-phosphate dipotassium salt	Sigma Aldrich	Cat#F1502
D-(+)-Glucose	Sigma Aldrich	Cat#G7021
Sodium pyruvate	Sigma Aldrich	Cat#P2256
β-Nicotinamide adenine dinucleotide, reduced dipotassium salt	Sigma Aldrich	Cat#N4505
Oligomycin A	Sigma Aldrich	Cat#75351
Potassium fluoride	Sigma Aldrich	Cat#402931
Sodium L-lactate	Sigma Aldrich	Cat#L7022
Critical commercial assays
CELLection™ Pan Mouse IgG Kit	ThermoFisher	Cat#11531D
Experimental models: Organisms/strains
Mice (C57BL/6J)	Jackson Laboratory	Cat#000664; RRID:IMSR_JAX:000664
Software and algorithms
GraphPad Prism v9.0.0	Graphpad Software	https://www.graphpad.com
Clampfit 10.7 Software	Molecular Devices	https://www.moleculardevices.com/products/axon-patch-clamp-system
PatchMaster v2x19.3 Software	HEKA Elektronik	https://www.heka.com
Adobe Illustrator 24.0 Software	Adobe	https://www.adobe.com/products/illustrator
Biorender	Biorender	https://www.biorender.com/

Highlights.

• K_ATP channels are regulated by a glycolytic metabolon on the plasma membrane

• Substrate channeling occurs between the consecutive enzymes of glycolysis

• Upper glycolysis produces ADP that is used directly by lower glycolysis to make ATP

• LDH and GADPH facilitate a plasma membrane-associated NAD⁺/NADH redox cycle