INTRODUCTION
Glucagon-Like Peptide-1 (GLP-1) is produced as a product of alternative post-translational processing of proglucagon
1-3, primarily in the enteroendocrine L cells in the intestine
3-6, and preproglucagon neurons in the brain as well
7. GLP-1 has pleiotropic actions
8, including inhibition of appetite
9,10, protection of neuronal and cardiac cells
11,12, modulation of immune cells
13, and stimulation of natriuresis and diuresis
14, beyond its initial identification as an incretin for controlling glucose homeostasis
15-18. As an incretin, GLP-1 is secreted in response to high glucose during the early postprandial phase
19,20, and binds to its receptors (GLP1Rs) in the pancreas to stimulate insulin secretion, among other effects
21,22. In addition to glucose, GLP-1 secretion can also be enhanced by dietary lipids, such as triglycerides
23, 2-monoacylglycerol
24, and free fatty acids
25, as well as by proteins, including oligopeptides
26 and amino acids
27. Pharmacological doses of GLP1 have been shown to improve control of blood glucose by promoting insulin secretion in the patients of type 2 diabetes mellitus (T2DM), with little risk of hypoglycemia
28,29. GLP-1 receptor agonists (GLP1RAs) with longer half-lives than native GLP-1 have been developed
30, and are recommended as one of the first-line therapies for T2DM patients with a risk of atherosclerotic cardiovascular diseases or unable to tolerate metformin
31,32.
In addition to glucose, the ATP-sensitive potassium channels (K
ATP) play a crucial role in the release of GLP-1. K
ATP is closed in response to high glucose, leading to depolarization of the plasma membrane, which triggers secretion of GLP-1-containing vesicles
33,34. It has been shown that K
ATP inhibitors can directly promote the secretion of GLP-1 from L cell lines or primary L cells, bypassing the requirement of high glucose
33, whereas activation of K
ATP with its activators prevents glucose-induced GLP-1 secretion
5,6. Also because of the involvement of ATP in the closure of K
ATP, it has been proposed that higher levels of ATP is generated in high glucose
35, especially due to the observation that the nonmetabolizable glucose analog 2-deoxy-
d-glucose (2-DG) that cannot be converted to ATP fails to trigger GLP-1 secretion
36,37. Another critically important clue came from the indispensable role of glucokinase (GK; also known as hexokinase 4, HK4; gene name:
GCK), which has the S
0.5 of around 10 mM
38,39, and thus effectively carries out glucose catalysis when glucose levels reach higher than 10 mM, near or above
40. The GLUT2 glucose transporter (gene name:
SLC2A2), which is either present on or relocated to the apical (luminal) side of L cells
41-44 has been shown to be critical for transport of high glucose
42,43,45-47, by virtue of its high
Km close to the S
0.5 of GK
48. In summary, the current understanding on the stimulation of GLP-1 release by high glucose is that the GLUT2-GK-ATP-K
ATP axis plays a pivotal role
5,34,37,40,41, similarly to that for the secretion of insulin from pancreatic β cells
49,50.
However, there are some caveats to the above-mentioned mechanisms of GLP-1 secretion. Studies on maturity-onset diabetes of the young type 2 (MODY2) patients have shown that the GK mutations, which are defective in glucose catalysis and impair insulin secretion, do not display a deficiency in GLP-1 secretion
51,52. The sodium-glucose cotransporter 1 (SGLT1; gene name:
SLC5A2), which co-transports positively charged sodium ions along with glucose into L cells, has been elegantly shown to directly cause membrane depolarization to induce GLP-1 secretion
53-55, in parallel to the effect of membrane depolarization caused by K
ATP channels. In particular, it was shown that mice with depletion of
SGLT1 exhibit significantly impaired GLP-1 secretion as determined at 5 min after oral glucose gavage
56. In primary L cells and the L cell line GLUTag, treatment with the inhibitor of SGLT1 blocks the secretion of GLP-1 in the presence of 1 mM glucose (Fig. 5 of ref. 54). Depletion of extracellular sodium has also been shown to block GLP-1 secretion in these cells
54. Moreover, the observation that 2-DG cannot be co-transported by SGLT1, and is unable to stimulate GLP-1 secretion, was also attributed to the role of SGLT1
36,37,57. Also in support of the argument, methyl-α-
d-glucopyranoside (MDGP, α-MG), another nonmetabolizable glucose analog that can be transported by both SGLT1 and GLUT2, is able to promote GLP-1 secretion
36,57,58.
Nevertheless, SGLT1 is saturated in glucose below 8 mM at most, ranging from 0.1
53,58,59 to 8 mM
42, depending on the studies. In contrast, the secretion of GLP-1 remains unsaturated even at glucose levels exceeding 20 mM
19,42,60-62, a level commonly encountered in the intestinal lumen postprandially
42,63. Importantly, studies have indicated that high concentrations of glucose or the nonmetabolizable analog MDGP exceed the
Km of SGLT1 and hence saturate SGLT1
42,57, can lead to a greater amount of secretion of GLP-1 compared to glucose levels below the
Km of SGLT1
6,19,53,61,62, indicating that other mechanisms will account for a major portion of postprandial GLP-1 secretion.
Here, we have revealed that it is the physical binding of glucose itself to GK that directly triggers GLP-1 secretion, in the dependency of structure rather than the energy of glucose. We have shown that the nonmetabolizable glucose analog MDGP, but not 2-DG, promotes GLP-1 secretion. The difference between the two nonmetabolizable analogs is attributed to their different ability to form a steady-state binding to GK and to induce a glucose-bound conformation of the enzyme, i.e., MDGP can do this, whereas 2-DG cannot. We have also shown that, while SGLT1 is responsible for GLP-1 secretion at lower, basal glucose levels, consistent with its low Km for glucose, GK plays a dominant role in mediating high glucose to stimulate the robust postprandial GLP-1 release. Mechanistically, we have demonstrated that when occupied by glucose or its analog, GK directly interacts with the Kir6.2 (gene name: KCNJ11) subunit of KATP, leading to closure of the channel. Using a whole-cell patch clamping technique, we show that glucose or the analog stimulates GK to interact and close the KATP channel. We also performed inside-out membrane patch clamping in a reconstituted cell-free system, showing that addition of GK and glucose causes closure of the KATP channel on the membrane patch. Together, unlike the role of glucose in insulin secretion from the pancreatic β cells, glucose acts as a ligand-like stimulus, which at concentrations over the S0.5 is sensed by GK. The glucose-occupied GK in turn binds and inhibits KATP to trigger the secretion of GLP-1 from the intestinal L cells.
RESULTS
High glucose induces secretion of GLP-1 without increasing cellular ATP levels
We fasted mice for 16 h and re-fed them with a normal chow diet (67.4% saccharides) for 10 min at approximately 10 g/kg of body weight (modified based on ref. 64; Supplementary Fig. S1a)
64. Stimulation of secretion of GLP-1, assessed for combined levels of both its full-length and two truncated forms in plasma
65, was observed from 5 to 45 min after the diet was given (Fig. 1a; see also the accompanying insulin levels in Supplementary Fig. S1b), in line with previous reports
66,67. The course of the stimulation correlated positively with the concentrations of glucose in the duodenal and jejunal lumens (Fig. 1a; no change in glucose levels in the lumen of the ileum). To our surprise, we found that the levels of intracellular ATP, as well as ADP and AMP, determined by mass spectrometry, did not change in the tissues of the duodenum and jejunum during the periods of GLP-1 secretion, compared to fasting (Fig. 1a). We also did not find any increase of ATP or changes in the ratios of the adenylates in the duodenum and jejunum tissues from mice gavaged with 1.5 g/kg glucose (Fig. 1b; see also the accompanying insulin levels in Supplementary Fig. S1c), or from mice directly injected with a solution of 75 g/L glucose (10 µL/g of body weight) to the duodenal and jejunal ducts via enterostomy (Fig. 1c, d; see also the accompanying insulin levels in Supplementary Fig. S1d, e), which mimicked postprandial levels of glucose (Fig. 1b–d). In addition, in dissected duodenal and jejunal tissues exposed
ex vivo to glucose at concentrations similar to postprandial levels (above 12.5 mM; defined as “high glucose”), adenylate levels remained unchanged during GLP-1 secretion (Supplementary Fig. S1f–i). Moreover, to exclude the possibility that adenylate levels in non-L epithelial cells mask changes in adenylate levels in L cells, we performed cell sorting on intestinal epithelial cells, among which the L cells expressed green fluorescent protein (GFP) driven by the proglucagon promoter
68 (also validated in Supplementary Fig. S1j). We then determined the levels of adenylates in these primary GFP-positive L cells in high glucose, and again observed unchanged intracellular adenylate levels despite GLP-1 secretion (Fig. 1e, f). Such a lack of increase of ATP was also seen in GLP-1-secreting L cell lines, including STC-1 (Supplementary Fig. S1k, l), GLUTag (Supplementary Fig. S1m, n), and NCI-H716 (Supplementary Fig. S1o, p). Since local intracellular increase of ATP, particularly near the plasma membrane, can also lead to the closure of K
ATP channels and induce insulin secretion in pancreatic β cells
69,70, we also determined the local intracellular concentrations of ATP in STC-1 cells. For this, we expressed the indicator that detects ATP:ADP ratios, PercevalHR
71, which is fused with Lck to its N-terminus to target this indicator to the plasma membrane (Fig. 1g; validated in Supplementary Fig. S1q). As shown in Fig. 1h, we did not detect any change in local intracellular ATP:ADP ratios in these cells in high glucose. Therefore, glucose induces GLP-1 release from the intestinal L cells in an ATP-independent manner. To assess the metabolic fate of high glucose in L cells, we performed metabolic flux tracing using [U-
13C]glucose. In STC-1 cells cultured in high glucose, the proportion of glucose entering glycolysis was low (less than 20% of total intracellular glucose, as evidenced by the levels of glycolytic intermediates [U-
13C]glucose-6-phosphate (G6P) and [U-
13C]fructose-6-phosphate (F6P)) (Supplementary Fig. S1r). Furthermore, the ATP-dependent oxygen consumption rate (OCR) in L cells did not increase upon high glucose stimulation (Supplementary Fig. S1s–u). These data suggest that in L cells, high glucose is not immediately metabolized to generate ATP; instead, it accumulates inside the cells.
As a control, we also measured ATP levels in pancreatic β cells, which are increased in high glucose
72. We indeed observed an increase of approximately twofold in ATP levels in the β-TC-6 and INS-1 pancreatic cells (Fig. 1i, Supplementary Fig. S1x) when glucose concentrations were increased to 25 mM for a 2-h incubation, which sufficiently stimulated secretion of insulin (Supplementary Fig. S1v, w). We also observed an increase of local intracellular ATP:ADP ratios beneath the plasma membrane in these cells (Fig. 1j), consistent with the previous observations
69,70.
GK acts as a sensor of high glucose to trigger GLP-1 secretion
The requirement of high glucose yet without increased ATP production in L cells suggested a possibility that the presence of glucose per se may play a triggering role in GLP-1 secretion. Indeed, we found that gavaging of 1.5 g/kg nonmetabolizable MDGP into mice, which gave rise to peak concentrations of MDGP to 150 mM in the intestinal lumen after 5 min and decreased to below 10 mM after 30 min, induced GLP-1 release (Fig. 2a; see also insulin secretion in Supplementary Fig. S2a). Similarly, addition of the nonmetabolizable MDGP instead of glucose into the culture medium of primary L cells or of STC-1 cells also induced the release of GLP-1 (Fig. 2b; Supplementary Fig. S2b), indicating that the structure of glucose is critical for inducing GLP-1 secretion. Of note, 2-DG, another glucose analog, did not stimulate GLP-1 release as MDGP, the possible molecular explanation of which is stated below (e.g., Supplementary Fig. S2e, g). As SGLT1 co-transports sodium ions along with glucose, we then analyzed whether SGLT1 plays a role in this process by treating STC-1 cells with the SGLT1 inhibitor phloridzin. We found that inhibition of SGLT1 blocked GLP-1 secretion during exposure to relatively low concentrations of glucose or MDGP (below 12.5 mM; defined as the “low” range of glucose/MDGP; Fig. 2c, d), consistent with the main role of SGLT1 in transporting glucose within this low concentration range (Fig. 2e) by virtue of its low
Km for glucose
42,53,58,59. However, phloridzin had little effect on the robust GLP-1 secretion in “high” (above 12.5 mM) glucose and MDGP (Fig. 2c, d). These results suggest that SGLT1 plays an important role in GLP-1 secretion in low glucose, but not in high glucose. As GLUT2 is a major glucose transporter for the uptake of high glucose by intestinal cells (Fig. 2h; see also ref.
42,45-47), we also investigated the role of GLUT2 that transports glucose in high concentrations, and found that inhibition of GLUT2 by phloretin largely blocked high glucose-stimulated GLP-1 secretion, but not the GLP-1 secretion happening at low glucose (Fig. 2f, g).
It is intriguing that GK, also known as hexose kinase 4, has a S
0.5 of around 10 mM for glucose
38, much higher than the
Km values of other hexokinases (HK1–3)
73. We wondered whether it is GK that responds to high glucose and acts as the molecular sensor of glucose. Indeed, when
GK was knocked down in STC-1 cells (validated in Supplementary Fig. S2c) or specifically knocked out in the intestine (
GK-IKO, validated in Supplementary Fig. S2d), we observed impaired secretion of GLP-1 following incubation in high glucose or glucose gavage (Fig. 2i, j). These observations also suggested that GK controls GLP-1 secretion in L cells. In addition, we found that the total levels of cellular ATP and ADP, or their ratios, remained unchanged in the
GK-knockdown cells in high glucose (25 mM; Fig. 2k), in agreement with the findings that high glucose is not immediately metabolized to generate ATP in these cells (Fig. 2l, m). Moreover, knockdown of
GK did not lead to a decrease in ATP levels in low glucose (1 mM; Fig. 2k), indicating that other hexokinases are sufficient for glucose catabolism in this condition. The critical role of glucose concentration per se in triggering GLP-1 release is also underscored by the observation that glucose accumulated inside the cell, as a consequence of inhibition of HK2 by lonidamine, promoted GLP-1 secretion (Fig. 2n, o). As mentioned above, we found that 2-DG did not induce GLP-1 secretion as MDGP (Supplementary Fig. S2e). To investigate why the two glucose analogs MDGP and 2-DG behave differently in the stimulation of GLP-1 secretion, we compared the binding affinities of MDGP and 2-DG to GK by a steady-state fluorescence assay. We found that MDGP had an affinity for GK with a
Kd of around 25 mM (Supplementary Fig. S2f). In contrast, 2-DG did not show saturable binding in this assay (Supplementary Fig. S2g), likely due to its rapid turnover as a GK substrate
39 and its inability to induce the glucose-bound conformational state
39,74 required for GLP-1 secretion. The stark difference in binding to GK provides an explanation for why only MDGP, and not 2-DG, can mimic glucose to stimulate GLP-1 release (Fig. 2b; Supplementary Fig. S2b, e). Furthermore, the relatively lower affinity of GK for MDGP (higher
Kd or S
0.5) compared to glucose is in line with the requirement of higher concentrations of MDGP (above 12.5 mM) to stimulate GLP-1 secretion to comparable levels (Fig. 2c, d). In contrast, MDGP had a similar affinity for SGLT1 to glucose
57, and stimulated GLP-1 secretion as effectively as glucose, at the low concentration range of glucose (below 12.5 mM) — as abovementioned, SGLT1-mediated glucose response only during the low range of glucose (Fig. 2c, d). Furthermore, we found that for STC-1 cells treated with GK activators, which increase the affinity of GK towards glucose (decrease S
0.5; note that
Km values were also calculated, as it remains unclear whether GK activators alter the properties of GK as an allosteric enzyme), including dorzagliatin
75, RO-28-1675
76, globalagliatin (validated in Supplementary Fig. S2h), and AZD1656
75, the glucose concentrations needed to secrete a similar amount of GLP-1 are lower compared to untreated cells (Fig. 2p). Such an effect was not observed with the compound GK Activator 3 (Fig. 2p), which increases only catalytic activity (V
max) without affecting the S
0.5 or
Km of GK
77. As a control, knockdown of the enzymes downstream of GK, including glucose-6-phosphate isomerase (GPI), glucose-6-phosphate dehydrogenase (G6PD), or phosphoglucomutase (PGM), did not affect high glucose-stimulated GLP-1 secretion (Fig. 2q; validated in Supplementary Fig. S2i; note that glucose-6-phosphatase (G6Pase) was not targeted for knockdown or expression, as it was below detectable levels in L cells, as shown in Supplementary Fig. S2j). As an additional control, we found that GLP-1 secretion positively correlated with intracellular glucose concentrations in cultured STC-1 cells, but not with those of the glycolytic metabolites (Fig. 2r; Supplementary Fig. S3).
In humans, loss-of-function mutations of GK have been identified in MODY2 patients
78,79. Among them, the G80A mutation shows severely reduced glucose binding affinity and has an impaired catalytic activity as a result of much reduced V
max and a much increased S
0.5 (since the mutation might alter GK’s allosteric properties,
Km values were also calculated), as validated in Supplementary Fig. S4a, b. We found that, ectopic expression of G80A in wildtype (WT) STC-1 cells or re-introduction of G80A into
GK-knockdown STC-1 cells impaired GLP-1 secretion induced by high glucose (Fig. 3a, b), but had no effect on low glucose-induced GLP-1 secretion which was inhibited by SGLT1 inhibitor phloridzin (Fig. 3c). These observations support the notion that SGLT1 but not GK plays a dominant role in GLP-1 secretion in low glucose. In contrast, the L309P mutation of GK is defective in catalysis and adopts the glucose-bound conformation even in low glucose, due to a much lower V
max with unchanged S
0.5 or
Km towards glucose, as validated in Supplementary Fig. S4c. We found that this glucose-bound mutant L309P constitutively stimulated GLP-1 secretion (Fig. 3a, b). In addition, the inhibition of SGLT1 by phloridzin only marginally reduced GLP-1 secretion from cells expressing GK-L309P that mimics high glucose (Fig. 3c), reinforcing the importance of GK in GLP-1 secretion in higher range of glucose levels compared to SGLT1. Neither GK-G80A nor GK-L309P changed intracellular ATP levels in these cells (Supplementary Fig. S4e; see validation data in Supplementary Fig. S4d), similar to the result seen in
GK knockdown (Fig. 2k). The opposite effects of GK-G80A and GK-L309P on the secretion of GLP-1 were recapitulated in mice when they were knocked in into the intestines of mice with the background of intestine-specific knockout of
GK (
GK-IKO) (Fig. 3e; validated in Supplementary Fig. S4f). We also measured the plasma levels of peptide YY (PYY) and glucose-dependent insulinotropic polypeptide (GIP), two gut hormones known to also promote GLP-1 secretion
80,81, in mice with intestine-specific knockin of GK-G80A and GK-L309P. We found no significant difference in PYY levels between these knock-in mice 10 min after glucose gavage (Supplementary Fig. S4g). However, during this same time period, GLP-1 secretion was significantly higher in the GK-L309P knock-in mice compared to the GK-G80A mice (Fig. 3e), indicating that PYY does not play a role in GLP-1 secretion. We also observed that GIP levels were even higher in the GK-G80A knock-in mice in which GLP-1 secretion is impaired (Supplementary Fig. S4g). Additionally, we found no difference between these knock-in mice in the levels of cholecystokinin (CCK) (Supplementary Fig. S4g), which is also an incretin hormone and may affect GLP-1 secretion through regulating glucose absorption in the gut
82. As a control, we also examined the effects of GK-G80A and GK-L309P on insulin secretion, and found that both mutations impaired insulin secretion by reducing ATP production in high glucose when re-introduced into the
GK-knockdown β-TC-6 cells (Fig. 3d, Supplementary Fig. S4i, validated in Supplementary Fig. S4h). These data are consistent with the MODY2 patients carrying the GK-G80A and GK-L309P mutations exhibiting deficiency in insulin secretion
78,79, and with the requirement of increased ATP production for insulin secretion in high glucose
49. In addition, we identified another MODY2 patient carrying a heterozygous R397L mutation of GK (classified as pathogenic with a frequency of 1.375E–06 in gnomAD database; Supplementary Fig. S5a), which is constitutively occupied with glucose yet unable to catalyze it due to a reduced V
max (enzymatic data listed in Supplementary Fig. S5b). Furthermore, we found an additional MODY2 patient who has a heterozygous, truncated form of GK, the mRNA for which contains a repeated nucleotide sequence from position 242 to 258, and hence encodes GK with a frameshift mutation at amino acid (aa) 87 (p.M87Afs*6; Supplementary Fig. S5c; not documented in gnomAD). This frameshift alters the amino acids from positions 87 to 91 and leads to premature termination of translation at amino acid 92 (referred to as GK
Δ92-CT), yielding a truncated GK lacking the glucose-binding domain (Supplementary Fig. S5c, d). We found that GLP-1 secretion was much lower in the patient carrying the GK
Δ92-CT truncation compared to the patient with the R397L mutation, despite both patients having similar blood glucose levels after glucose intake (Fig. 3f). Both patients exhibited impaired glucose-induced insulin secretion, with insulin-to-glucose ratios around 20 pmol/mmol (Fig. 3f), which is typical for MODY2 patients, whereas healthy individuals typically have ratios close to 40 pmol/mmol
83-85. These blunting effects of GK-R397L and GK
Δ92-CT on GLP-1 secretion could also be recapitulated in STC-1 cells after introducing these mutations into the
GK-knockdown STC-1 cells (Fig. 3g; see validation data in Supplementary Fig. S5e). In addition, both mutations, when reintroduced into β-TC-6 cells, failed to rescue the knockdown of
GK in the production of ATP and insulin secretion in pancreatic β-cells (Fig. 3h; Supplementary Fig. S5f, g; see validation data in Supplementary Fig. S5e). Together, these mutational studies indicate that the binding of glucose to GK underlies GLP-1 secretion.
Glucose-occupied GK binds and inhibits KATP
We next investigated how GK in its glucose-bound state triggers GLP-1 secretion. We found that the GLP-1 secretion induced by GK-L309P could be blocked by the KATP channel openers diazoxide and tifenazoxide (NN414) (Fig. 4a; Supplementary Fig. S6a), indicating that the GK-induced GLP-1 secretion depends on the closure of KATP channels. Consistently, the KATP inhibitors glibenclamide and repaglinide could directly induce GLP-1 secretion even in the presence of the GK-G80A, effectively bypassing the dominant negative effect of this mutant on GLP-1 secretion (Fig. 4b; Supplementary Fig. S6b). These results suggest that KATP functions downstream of GK binding to glucose in GLP-1 secretion (depicted in Fig. 4c).
We also performed whole-cell patch clamping experiments to examine the effect of GK on K
ATP channel activities. We set the concentrations of ATP and ADP respectively at 2 mM and 200 μM (in the pipette buffer, as depicted in Fig. 4d, e), to resemble their respective intracellular concentrations (measured in Fig. 1f; Supplementary Fig. S1l, n), with varying glucose concentrations. We found that high concentrations of glucose (25 mM), whether introduced in the pipette (representing intracellular compartment; Fig. 4e) or bath solutions (representing extracellular compartment; Fig. 4d), led to a decrease in the endogenous whole-cell K
+ currents in L cell lines, including GLUTag cells and STC-1 cells (Fig. 4d, e). When we washed out the glucose with a low-glucose (1 mM) bath solution, the currents were restored (Fig. 4d). We also reconstituted the glucose-, rather than ATP-regulated K
ATP activity in HEK293T cells, which do not exhibit changes in ATP levels in response to different glucose levels
86, as do L cells. In HEK293T cells (which lack expression of GK
87 and K
ATP88) transfected with GK and the Kir6.2 and SUR1 (gene name:
ABCC8) subunits (the type of K
ATP expressed in L cells, as confirmed by mass spectrometry; Supplementary Fig. S6c) of the K
ATP, we observed glucose-suppressed K
+ currents (Fig. 4f, g; see validation data in Supplementary Fig. S6d). In addition, MDGP also inhibited the K
+ currents (Supplementary Fig. S6e). We also found that expression of GK-L309P could still inhibit the currents in low glucose in these cells, while GK-G80A was unable to inhibit the currents even in high glucose (Fig. 4f, g; see validation data for GLUTag cells in Supplementary Fig. S6f). We further reconstituted the regulation of GK on K
ATP by using membrane patches from HEK293T cells transfected with Kir6.2 and SUR1, and recorded currents from inside-out membrane patches. The purified bacterially expressed GK, along with 2 mM ATP, 200 μM ADP, and varying concentrations of glucose, was then added to the bath solution facing the intracellular side of K
ATP (depicted in Fig. 4h). Consistent with previous findings
89, we observed a basal open probability multiplied by the number of channels (NPo) of K
ATP in low glucose (Fig. 4h; Supplementary Fig. S6g). When high glucose was introduced, the NPo of K
ATP reduced (Fig. 4h). Returning the glucose concentration to low levels reversed the high-glucose-caused reduction in NPo, and this change could be further inhibited by the K
ATP inhibitor glibenclamide, confirming that the observed current changes specifically reflect glucose-responsive changes of K
ATP activity (Fig. 4h; see also full traces in Supplementary Fig. S6g). In addition, GK-G80A stimulated the opening of K
ATP in high glucose, while GK-L309P reduced the opening of K
ATP in low glucose (Fig. 4h). These results indicate that GK can directly modulate the activity of K
ATP.
Finally, to explore how GK regulates K
ATP, we tested if GK forms a complex with the subunits (Kir6.2 and SUR1) of K
ATP. Through immunoprecipitation, we found that when GK was ectopically expressed alongside either Kir6.2 or SUR1 in HEK293T cells, GK interacted with each of the subunits (Fig. 5a–d). We also found that the interaction between GK and the Kir6.2 subunit was significantly enhanced with the addition of high glucose in the culture medium or in cell lysates (Fig. 5a, b), whereas the interaction between GK and SUR1 was constitutive (Fig. 5c, d). In STC-1 cells, we also observed similar interaction patterns for endogenous GK–Kir6.2 and GK–SUR1 (Fig. 5e, f). These observations are consistent with several studies showing that the association between Kir6.2 and SUR1 becomes more robust when the channel is inhibited
90, and further support the notion that the interaction of GK with Kir6.2 is enhanced for the inhibition of the K
ATP channel. We also found that the GK-L309P mutant showed a robust, constitutive interaction with Kir6.2, whereas the GK-G80A mutant showed marginal interaction with Kir6.2 (Fig. 5g). Neither of the mutations affected the binding of GK to SUR1 (Fig. 5h). The glucose-induced interaction between GK and Kir6.2 could also be observed inside the living cells by FRET-FLIM analyses (Fig. 5i). The calculated GK-Kir6.2 distance (using a Förster radius of 5 nm for the GFP-mCherry pair
91) decreased from approximately 4.8 nm to 4.7 nm upon glucose addition, with a corresponding increase in FRET efficiency (Fig. 5i). As a control, we found that other hexokinases (HK1 to HK3) hardly interacted with K
ATP, at either low or high glucose (Supplementary Fig. S7a–c). Domain mapping experiments showed that the segment of aa 205–228 in Kir6.2 was required for its interaction with GK (Fig. 5j; Supplementary Fig. S7d–f). For the Kir6.2
Δ205–228 truncation that showed severe degradation (Supplementary Fig. S7f), we created a mutation in which amino acids 205–228 were replaced with alanine (Kir6.2-205–228A). This mutant did not affect the activity of K
ATP in low glucose (0 mM) and remained sensitive to glibenclamide (at the open state; Supplementary Fig. S7g); however, it blocked the GK-mediated inhibition of K
ATP in high glucose (25 mM; Supplementary Fig. S7g), and consequentially impaired the high glucose-induced secretion of GLP-1 when expressed in STC-1 cells or knocked in into the intestines of mice with the background of intestine-specific knockout of
Kir6.2 (
Kir6.2-IKO) (Fig. 5k, l; see validation data in Supplementary Fig. S7i). Together, we found that GK, when bound to glucose, interacts with and inhibits K
ATP to trigger the release of GLP-1.
DISCUSSION
We have provided multiple lines of evidence showing that glucose acts as a direct stimulus to induce the release of GLP-1 in L cells, in the sense that the glucose molecule physically triggers the secretion (Fig. 5m). First of all, we found that high glucose stimulates GLP-1 secretion in L cells without increasing ATP levels. In particular, the nonmetabolizable glucose analog MDGP, unable to generate ATP but still binds to GK, effectively inhibits the K
ATP channel and stimulates GLP-1 secretion. In support of a requirement of physical binding of glucose, the other analog 2-DG, which is phosphorylated by GK but does not stabilize the glucose-bound conformation, fails to stimulate GLP-1 release. Second, we have employed various GK mutants identified in human MODY2 patients
78,79, and found that GK mutants that adopt the glucose-bound conformation (due to a catalytic defect) can inhibit K
ATP and promote GLP-1 secretion even in low glucose, whereas GK mutants with severely reduced glucose binding affinity fail to stimulate GLP-1 secretion in high glucose. Furthermore, GK activators that enhance the affinity of GK for glucose can lower the threshold of glucose required for stimulating GLP-1 secretion. In particular, knockdown of
GK or introduction of the catalysis-defective mutants of GK does not lead to a reduction of ATP production. This is likely because other hexokinases (HK1 to HK3) which have low
Km values, are present in L cells
92 and are sufficient for producing ATP (Fig. 2l, m). In addition, ATP may be sustained from consuming non-glucose carbon sources in L cells. It was shown that L cells express high levels of monocarboxylate transporter 1 (MCT1; Supplementary Fig. S7j), which transports pyruvate and lactate into L cells for catabolism. Inhibition of pyruvate and lactate catabolism by UK5099 significantly decreased OCR in L cells in low glucose (Supplementary Fig. S7k). We also found that inhibition of fatty acid oxidation by etomoxir reduced OCR in L cells in low glucose (Supplementary Fig. S7k; note that L cells can also metabolize glutamine, albeit to a lesser extent, as demonstrated by using BPTES, an inhibitor of glutaminolysis). In contrast, pancreatic β cells primarily express GK that can only generate ATP at a high rate when glucose levels are high, due to its high S
0.5 for glucose
93. Additionally, the pancreatic β cells heavily depend on glucose for energy as they also do not express monocarboxylate transporter 1 (MCT1
94; Supplementary Fig. S7j) and have very low levels of lactate dehydrogenase
95,96, which help provide pyruvate and lactate, respectively, as substitutes for glucose to fuel mitochondrial ATP production. These differences in protein profiling may provide an explanation as to why pancreatic β cells generate higher ATP in high glucose. It is interesting to add that the strict requirement for the pancreatic β cells of glucose may safeguard against over secretion of insulin that may cause hypoglycemia, in case GLP-1 is overproduced by other stimuli such as amino acids. It is interesting to add that the ligand-like action of glucose is not unique to L cells. Glucose has also been shown to bind directly to and regulate RNA helicase DDX21
97, RNA-binding protein NSUN2
98, and transcription factor IRF6
99 to promote epidermal differentiation, and that NSUN2 upon glucose binding suppresses the cGAS-STING pathway in cancer
100.
We have also demonstrated that on binding to glucose, GK directly inhibits K
ATP channels, a critical step before the release of GLP-1. We found that GK bound with glucose, occupied by the nonmetabolizable glucose analog, or acquired the glucose-bound conformation as in the catalysis-defective mutant L309P, enhances its interaction with the Kir6.2 subunit of K
ATP channel, via the region of aa 205–228. Of note, our FRET-FLIM analyses revealed a modest but significant glucose-induced decrease in GK–Kir6.2 distance (approximately 1–2 Å; with unchanged donor-only lifetime). While the co-immunoprecipitation assays showed constitutive GK–SUR1 binding, GK–Kir6.2 association is enhanced by high glucose. These observations, together with the hetero-octameric nature of K
ATP (four Kir6.2 and four SUR1 subunits), suggest that glucose-occupied GK induces a subtle local conformational change within a pre-existing complex, rather than a binary on/off switch. However, we cannot exclude that the GK–Kir6.2 interactions involve unassembled or immature channel subunits. High-resolution structural analysis of the K
ATP complex will be required to fully resolve the allosteric regulation of the interactions. Nevertheless, the segment aa 205–228 of Kir6.2 that we identified as the GK-binding region is required for GLP-1 release stimulated by glucose. Although the E227K mutation in Kir6.2, which has been identified in cases of neonatal diabetes, is suggested to reduce the binding of ATP to K
ATP101,102, we found that the Kir6.2-205–228A variant does not affect the activity of K
ATP under physiological ATP concentrations (Fig. 4h; Supplementary Fig. S7g, h). In addition, according to the previously resolved structure, this region is distinct from the ATP binding pocket of the channel
103, which supports our finding that GK controls K
ATP independently of an increase of ATP. Moreover, the inhibition of K
ATP by GK can be recapitulated
in vitro through inside-out patch clamping in the presence of physiologically relevant concentrations of ATP. These experiments show that the glucose-bound GK reduces the activity (NPo) of K
ATP by approximately 40% (Fig. 4h), which is also consistent with the percentages observed in whole-cell patch clamping experiments (Fig. 4f). Given that the resting membrane potential of L cells in low glucose is primarily governed by K
+ efflux through K
ATP33,34,104, such a reduction in K
ATP activity can lead to a decrease in membrane potential to around –25 mV, as calculated using the Goldman-Hodgkin-Katz equation. This membrane potential exceeds the threshold (around –36 mV) necessary to trigger further membrane depolarization through voltage-dependent calcium channels, which ultimately facilitates the exocytosis of vesicles containing GLP-1
105. It is also noteworthy that the independence of the increase of ATP in L cells does not imply the lack of necessity of basal ATP for the closure of the channel, as it has been previously shown that reduction of ATP levels by mitochondrial inhibitors that grossly inhibit mitochondrial oxidative phosphorylation, prevents the K
ATP closure even in high glucose
6,33. In fact, our finding that the ATP-binding-defective mutant of K
ATP (Kir6.2-K185E/R50G/R201G
103) displays constitutive channel activity and blocks GLP-1 secretion in L cells, which cannot be overridden by the constitutively-glucose-bound GK-L309P mutant (Supplementary Fig. S7l), confirmed a requirement of a priori binding of ATP to K
ATP.
Our study cannot rule out other mechanisms that sense glucose and somehow contribute to GLP-1 secretion
in vivo. For example, we found that the intestine-specific knockout of
GK, the intestine-specific knockin of glucose-binding-defective mutant GK-G80A, or the intestine-specific knockin of Kir6.2-205–228A that cannot bind GK, all did not completely eliminate glucose-induced GLP-1 secretion (Figs. 2j, 3e, 5l). For example, efferent signals from the vagus nerve have been shown to depolarize L cells, leading to GLP-1 secretion even without direct contact with glucose
106,107. This effect is evident in L cells located in the distal gut, such as the ileum and colon
108, where GLP-1 may be released when the vagus nerve is activated by its efferent fibers in the taste bud
109,110 or in the proximal gut (such as duodenum and jejunum), which are exposed to glucose
107. In addition, as previously shown
53,54,56, and as confirmed in this study, SGLT1, by virtue of co-transporting the sodium ion, plays an important role in mediating the secretion of GLP-1 through causing membrane depolarization when glucose is present in low concentrations. It is also interesting to note that secretion of GLP-1 can be stimulated by additional stimuli, such as fat and proteins. In addition to the direct modulation of GK conformation, other nutrient-sensing pathways, such as AMPK and mTORC1, also play roles in GLP-1 secretion, in context-dependent manners. It was shown that chronic activation of mTORC1 or depletion of AMPK promotes GLP-1 synthesis and secretion in L cells
111-113. Conversely, acute activation of AMPK by metformin
114,115, lithocholic acid
116, or Cucurbitacin B
117 can trigger GLP-1 release. How these pathways integrate with the GK–K
ATP module to fine-tune GLP-1 secretion under various nutritional states warrants further investigation. Interestingly, such secreted GLP-1 does not cause hypoglycemia, for the secretion of insulin depends on a priori presence of high glucose. The distinctive mechanisms for the secretion of the two hypoglycemic hormones provide a safeguard against hypoglycemia in cases of GLP-1 secretion stimulated by non-carbohydrate stimuli. Notably, GK and K
ATP are co-expressed in several other endocrine cells, including pancreatic δ cells
118, pituitary corticotropes
40, hypothalamic arcuate glucose-excited neurons
119, and enteroendocrine K cells
120. In these cells, glucose also triggers the release of hormones (somatostatin, ACTH, neuropeptide Y, and GIP), and evidence suggests that increased ATP generation is also not necessary to induce secretion. The non-metabolizable glucose analog MDGP, which we found to mimic glucose-bound GK and stimulate GLP-1 secretion, also excites arcuate glucose-excited neurons
121. Thus, the ligand-like GK-K
ATP signaling axis may represent a conserved glucose-sensing mechanism beyond L cells.
Our findings may also have implications for the development of GK-targeted therapies. Traditional GK activators (GKAs) have faced challenges such as loss of efficacy over time
122 and adverse metabolic effects, including hepatic steatosis
123 and impaired pancreatic β-cell function
124. These adverse effects likely stem from the catalytic activation of GK, which leads to enhanced glycolysis
125. In contrast, the ligand-like mechanism of glucose for GLP-1 secretion in L cells harnesses the glucose-sensing function of GK separable from its catalytic activity. Therapeutic strategies to be designed to stabilize the glucose-bound conformation of GK, rather than to simply enhance its catalytic activity, may offer a more effective and less harmful option. Additionally, such future GKAs might offer a new avenue for developing drugs that specifically stimulate the secretion of native GLP-1 with potentially less adverse effect
126, which is especially important for patients who prioritize blood glucose control over weight loss.
This study has some limitations. To analyze GLP-1 release in L cells,
in vitro culture of cells is required. However, these cells in culture may have lost cell polarity compared to their
in vivo state, which may affect their physiological and biophysical behaviors. Nevertheless, many studies have shown that the glucose-induced GLP-1 secretion in cultured primary cells shares characteristics of cells with polarity maintained (such as those examined using the intestine
ex vivo5,60; this was also performed by us, as shown in Supplementary Fig. S1f, g). This suggests that the loss of polarity may not have a significant impact on the properties of L cells. It is reassuring that results on GLP-1 secretion obtained with cultured cells are recapitulated
in vivo in various genetic models. In addition, in this study, we analyzed the glucose-stimulated secretion of GLP-1 specifically in the lumen of the proximal gut, because the refeeding regimen we used allows glucose to accumulate only in the proximal gut (Fig. 1a; also demonstrated by others
127,128). As the primary L cells we used were derived from whole intestinal tissues and thus included L cells from both proximal and distal regions (Fig. 1f), it is reasonable to suggest that the distal intestines may also not have any increase in ATP in high glucose.
MATERIALS AND METHODS
Reagents
Rabbit anti-GK (cat. ab88056, RRID: AB_10673863; 1:1,000 dilution for IB, and 1:100 for immunoprecipitation (IP)) antibody was purchased from Abcam. Rabbit anti-Kir6.2 (cat. PA5-99440, RRID: AB_2818373; 1:1,000 for IB) and anti-SUR1 (cat. PA5-103639, RRID: AB_2852973; 1:1,000 for IB) antibodies, and Alexa Fluor 594-conjugated, goat anti-rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody (cat. A-11037, RRID: AB_2534095; 1:200 for immunofluorescence (IF)) were purchased from Thermo Fisher Scientific. Mouse anti-Myc-tag (cat. 2276, RRID: AB_331783; 1:1,000 for IB) antibody, rabbit anti-G6PD (cat. 12263, RRID: AB_2797861; 1:1,000 for IB), anti-N-cadherin (cat. 13116, RRID: AB_2687616; 1:100 for IF), and anti-PGM1 (cat. 68036; 1:1,000 for IB) antibodies, and the HRP-conjugated mouse anti-rabbit IgG (conformation-specific, cat. 5127, RRID: AB_10892860; 1:2,000 for IB) antibody were purchased from Cell Signaling Technology. Mouse anti-FLAG M2 (cat. F1804, RRID: AB_262044; 1:1,000 for IB) antibody, rabbit anti-GPI (cat. HPA024305, RRID: AB_1849922; 1:1,000 for IB) antibody, and mouse anti-FLAG M2 affinity gel (cat. A2220, RRID: AB_10063035; 1:500 for IP) were purchased from Sigma. Rabbit anti-MCT1 (cat. 20139-1-AP, RRID: AB_2878645; 1:1,000 for IB), anti-PGM2 (cat. 11022-1-AP, RRID: AB_2161422; 1:1,000 for IB), and anti-G6PC (G6Pase; cat. 66860-1-Ig, RRID: AB_2882199; 1:1,000 for IB) antibodies were purchased from Proteintech. Mouse anti-HA-tag (cat. sc-7392, RRID: AB_2894930; 1:1,000 for IB) antibody was purchased from Santa Cruz Biotechnology. The horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (cat. 115-035-003, RRID: AB_10015289; 1:5,000 for IB) and goat anti-rabbit IgG (cat. 111-035-003, RRID: AB_2313567; 1:5,000 for IB) antibodies were purchased from Jackson ImmunoResearch.
DMSO (cat. D2650), glucose (cat. G7021), methanol (cat. 646377), ethanol (cat. 1070172511 for protein MS and cat. 459836 for other usages), chloroform (cat. C7559), PBS (cat. P5493), Triton X-100 (cat. T9284), tamoxifen (cat. T5648), corn oil (cat. C8267), CsCl (cat. C4036), sodium acetate (NaAc; cat. S5636), nuclease-free water (cat. W4502), human tubal fluid (HTF) medium (cat. MR-070-D), KSOM medium (cat. MR-121-D), l-glutathione reduced (GSH; cat. G4251), mineral oil (cat. M5310), collagenase IX (cat. C7657), HEPES (cat. H4034), KCl (cat. P9333), NaCl (cat. S7653), NaHCO3 (cat. S5761), NaH2PO4 (cat. S5011), CaCl2 (cat. C5670), MgCl2 (cat. M8266), NaOH (cat. S8045), BSA (cat. A2058), MgSO4 (cat. M2643), β-mercaptoethanol (cat. M3148), hexadimethrine bromide (polybrene; cat. H9268), Trizma base (Tris; cat. T1503), EDTA (cat. E6758), EGTA (cat. E3889), sodium pyrophosphate (cat. P8135), β-glycerophosphate (cat. 50020), SDS (cat. 436143), FLAG Peptide (cat. F3290), glycerol (cat. G5516), trichloroacetic acid (cat. 91228), acetone (cat. 534064), bromophenol blue (cat. B0126), Tween-20 (cat. P9416), formaldehyde solution (formalin; F8775), d-mannitol (cat. M4125), l-methionine sulfone (cat. M0876), d-campher-10-sulfonic acid (cat. 1087520), acetonitrile (cat. 34888), ammonium acetate (cat. 73594), ammonium hydroxide solution (cat. 338818), 3-aminopyrrolidine dihydrochloride (cat. 404624), N,N-diethyl-2-phenylacetamide (cat. 384011), trimesic acid (cat. 482749), diammonium hydrogen phosphate (cat. 1012070500), ammonium trifluoroacetate (cat. 56865), ammonium bicarbonate (cat. 09830), medronic acid (cat. M0240000), thrombin (cat. T7513), ATP (cat. A6419), Mg-ATP (cat. A9187), ADP (cat. A2754), IPTG (cat. I6758), NADP+ (cat. N5755), G6PD (cat. 346774), pyruvate kinase (PK; cat. P1506), l-lactic dehydrogenase (LDH; cat. L2500), NADH (cat. N8129), PEP (cat. P0564), 5(6)-Carboxyfluorescein N-hydroxysuccinimide ester (FITC-NHS; cat. 21878), Na2HPO4 (cat. S7907), KH2PO4 (cat. P5655), 2-deoxy-d-glucose (2-DG; cat. D8375), methyl α-d-glucopyranoside (MDGP; cat. M9376), poly-l-lysine solution (cat. P8920), methanesulfonic acid (MSA; cat. 59510), N-methyl-d-glucamine (NMDG; cat. M2004), dithiothreitol (DTT; cat. 43815), iodoacetamide (cat. I6125), trypsin (for protein MS; cat. T6567), trifluoroacetic acid (cat. T6508), formic acid (cat. 5.43804), oligomycin A (cat. 75351), BPTES (cat. SML0601), etomoxir (cat. 236020), UK5099 (cat. 5.04817), dipeptidyl peptidase IV inhibitor I (diprotin A; cat. D3822), GLP-1 EIA Kit (cat. RAB0201), Glucose Assay Kit (cat. GAGO20), and BCA Protein Assay Kit (cat. 71285-3) were purchased from Sigma. Methanol (cat. A456-4, for protein MS), acetonitrile (cat. A955-4, for protein MS), formic acid (cat. A177-50, for protein MS), trifluoroacetic acid (cat. A116-50, for protein MS), Phusion High-Fidelity DNA Polymerase kit (cat. F530N), mMESSAGE mMACHINE T7 Transcription Kit (cat. AM1344), MEGAclear Transcription Clean-Up Kit (cat. AM1908), MEGAshortscript T7 Transcription Kit (cat. AM1354), Maxima SYBR Green/ROX qPCR master mix (cat. K0223), DMEM, high glucose (DMEM; cat. 12100061), glucose-free DMEM (cat. 11966025), RPMI-1640 medium (cat. 31800089), Leibovitz-15 (L-15) medium (cat. 11415064), calcium-free Hanks Balanced Salt Solution (HBSS; cat. 14170120), trypsin-EDTA (cat. 25200072), FBS (cat. 10099141C), penicillin-streptomycin (cat. 15140163), Lipofectamine 2000 (cat. 11668500), Lipofectamine 3000 (cat. L3000150), MEM non-essential amino acids solution (cat. 11140050), GlutaMAX (cat. 35050061), sodium pyruvate (cat. 11360070), ProLong Diamond antifade mountant (cat. P36970), ProLong Live Antifade reagent (cat. P36975), and Mouse Insulin ELISA Kit (cat. EMINSX5) were purchased from Thermo. Protease inhibitor cocktail (cat. 70221) was purchased from Roche. RO-28-1675 (cat. HY-10595), dorzagliatin (cat. HY-109030), Globalagliatin (cat. HY-13529), AZD1656 (cat. HY-15675), GK Activator 3 (cat. HY-144287), phloridzin (cat. HY-N0142), and KL-11743 (cat. HY-145597) were purchased from MedChemExpress. Matrigel Matrix (cat. 356230) was purchased from Corning. MinElute PCR Purification Kit (cat. 28004) was purchased from Qiagen. Mouse Ultrasensitive Insulin ELISA kit (cat. 80-INSMSU-E10) was purchased from ALPCO. Nonfat dry milk (cat. 9999) was purchased from Cell Signaling Technology. hCG (cat. 110900282) and PMSG (cat. 110904564) were purchased from Sansheng Biological Technology (Ningbo, China). WesternBright ECL and peroxide solutions (cat. 210414-73) were purchased from Advansta. BD Difco LB (Luria-Bertani) Broth Miller (LB; cat. 244620) was purchased from BD. rProtein A Sepharose Fast Flow (cat. 17127904), Protein G Sepharose 4 Fast Flow (cat. 17061806), Glutathione Sepharose 4 Fast Flow Gel (cat. 17513203), and Superdex 200 Increase 10/300 GL (cat. 28990944) were purchased from Cytiva. SP3 beads (cat. SP3001/2) were purchased from Nanomics. 3-hydroxynaphthalene-2,7-disulfonic acid disodium salt (2-naphtol-3,6-disulfonic acid disodium salt; cat. H949580) was purchased from Toronto Research Chemicals. Hexakis(1H,1H,3H-perfluoropropoxy) phosphazene (hexakis(1H, 1H, 3H-tetrafluoropropoxy) phosphazine; cat. sc-263379) was purchased from Santa Cruz Biotechnology. Polyethylenimine (PEI; cat. 23966) was purchased from Polysciences. PrimeSTAR HS polymerase (cat. R40A) was purchased from Takara. [U-13C]-glutamine (cat. 184161-19-1) and [U-13C]-glucose (cat. CLM-1396) were purchased from Cambridge Isotope Laboratories. [U-13C, 15N]AMP (cat. 123603801), [U-13C, 15N]ADP (cat. 129603801), and [U-13C, 15N]ATP (cat. 121603801) were purchased from Silantes. Seahorse XF base medium (cat. 103334) and Seahorse XF Calibrant solution (cat. 100840) were purchased from Agilent. Mouse Peptide YY (PYY) ELISA Kit (cat. JL20433-96T) and Mouse Cholecystokinin (CCK) ELISA Kit (cat. JL10451-96T) were purchased from Shanghai Jianglai Biotechnology Co., Ltd. Mouse Gastric Inhibitory Polypeptide (GIP) ELISA Kit (cat. RK02850) was purchased from ABclonal.
Mouse strains
Protocols for all rodent experiments were approved by the Institutional Animal Care and the Animal Committee of Xiamen University (XMULAC20180028 and XMULAC20220050). Unless stated otherwise, mice were housed with free access to water and standard diet (65% carbohydrate, 11% fat, 24% protein) under specific pathogen-free conditions. The light was on from 8:00 to 20:00, with the temperature kept at 21–24 °C and humidity at 40–70%. Only male mice were used in the study, and male littermate controls were used throughout the study.
Mice of WT C57BL/6J (#000664),
Gcg-
Cre (#030663; provided by Dr. Klaus Kaestner
129),
Villin-
CreERT2 (#020282; provided by Dr. Sylvie Robine
130), and mT (membrane-targeted tdTomato)/mG (membrane-targeted EGFP) (#007576; provided by Dr. Liqun Luo
131) were obtained from The Jackson Laboratory.
GCKF/F (#S-CKO-00353) mice were purchased from Cyagen, and
Kir6.2F/F (
Kcnj11F/F; #T015246) from GemPharmatech. Mice with intestine-specific
GCK knockout were generated by crossing
GCKF/F mice with
Villin-
CreERT2 mice, followed by intraperitoneal injection with tamoxifen (dissolved in corn oil) at 200 mg/kg, 4 times a week for a week. The knockout offspring were validated through genotyping (see details in Supplementary Table 1).
WT GK, along with G80A and L309P mutants, was introduced to the intestines of
GKF/F mice through the
Rosa26-LSL(LoxP-Stop-LoxP) system as described previously
132,133, followed by crossing with the
Villin-
CreERT2 mice. The removal of the intestinal
GK gene, and the
LSL cassette ahead of the introduced GK and its mutants (to trigger the expression of introduced GK), was then achieved by intraperitoneal injection of tamoxifen. To introduce GK or its mutants into
GKF/F mice, cDNA fragments encoding GK, GK-G80A, or GK-L309P were inserted into the Rosa26-CTV vector
134, followed by purification of the plasmids using the CsCl density gradient ultracentrifugation method. Some 100 μg of plasmid was then diluted with 500 μL of di-distilled water, followed by concentrating via centrifugation at 14,000×
g at room temperature in a 30-kDa-cutoff filter (UFC503096, Millipore) to 50 μL of solution. The solution was diluted with 450 μL of di-distilled water, followed by another two rounds of dilution/concentration cycles. The plasmid was then mixed with 50 μL of di-distilled water to a final volume of 100 μL, followed by mixing with 10 μL of NaAc solution (3 M stock concentration, pH 5.2). The mixture was then mixed with 275 μL of ethanol, followed by incubating at room temperature for 30 min to precipitate the plasmid. The precipitated plasmid was collected by centrifugation at 16,000×
g for 10 min at room temperature, followed by washing with 800 μL of 75% (v/v) ethanol (in di-distilled water) twice. After being dissolved in 100 μL of nuclease-free water, the plasmid, along with
SpCas9 mRNA and the sgRNAs against the mouse
Rosa26 locus, was then microinjected into the
in vitro fertilized (IVF) embryos of the
GKF/F mice. To generate the
SpCas9 mRNA, 1 ng of pcDNA3.3-hCas9 plasmid (constructed by inserting the Cas9 fragment released from Addgene #41815
135, into the pcDNA3.3 vector; diluted to 1 ng/μL) was amplified using the Phusion High-Fidelity DNA Polymerase kit on a thermocycler (T100, Bio-Rad) with the following programs: pre-denaturing at 98 °C for 30 sec; denaturing at 98 °C for 10 sec, annealing at 68 °C for 25 sec, then extending at 72 °C for 2 min in each cycle; and final extending at 72 °C for 2 min; cycle number: 33. The following primer pairs were used: 5’-CACCGACTGAGCTCCTTAAG-3’, and 5’-TAGTCAAGCTTCCATGGCTCGA-3’. The PCR product was then purified using the MinElute PCR Purification Kit following the manufacturer’s instructions. The purified SpCas9 PCR product was then subjected to
in vitro transcription using the mMESSAGE mMACHINE T7 Transcription Kit following the manufacturer’s instructions (with minor modifications). Briefly, 5.5 μL (300 ng/μL) of SpCas9 PCR product as the template was mixed with 10 μL of 2× NTP/ARCA solution, 2 μL of 10× T7 Reaction Buffer, 0.5 μL of RNase inhibitor, 2 μL of T7 Enzyme Mix, and 4.5 μL of nuclease-free water, followed by incubating at 37 °C for 2 h. The mixture was then mixed with 1 μL of Turbo DNase, followed by incubating at 37 °C for 20 min to digest the template. The mixture was then mixed with 20 μL of 5× E-PAP Buffer, 10 μL of 25 mM MnCl
2, 10 μL of 10 mM ATP, 4 μL of
E-PAP enzyme, and 36 μL of nuclease-free water, followed by incubating at 37 °C for 20 min for poly(A) tailing. The tailed product was then purified using the MEGAclear Transcription Clean-Up Kit following the manufacturer’s instructions (with minor modifications). Briefly, 20 μL of tailed RNA was mixed with 20 μL of Elution Solution, followed by mixing with 350 μL of Binding Solution Concentrate. Some 250 μL of ethanol was then added to the mixture, followed by passing the mixture through the Filter Cartridge and washing with 250 μL of Wash Solution twice. The RNA was then eluted with 50 μL of pre-warmed (at 90 °C) Elution Solution. The sgRNAs were prepared as in the SpCas9 mRNA preparation, except that: a) the gRNA Cloning Vector (Addgene, #41824
135) was used as the template, and the following programs: pre-denaturing at 98 °C for 30 sec; denaturing at 98 °C for 10 sec, annealing at 60 °C for 25 sec, then extending at 72 °C for 20 sec in each cycle; and final extending at 72 °C for 2 min; cycle number: 33; and the following primers: 5’-GAAATTAATACGACTCACTATAGGCGCCCATCTTCTAGAAAGACGTTTTAGAGCTAGAAATAGC-3’, and 5’-AAAAGCACCGACTCGGTGCC-3’; were used; b)
in vitro transcription was performed using the MEGAshortscript T7 Transcription Kit, in which the mixture containing: 7.5 μL (100 ng/μL) of purified PCR product, 2 μL of T7 10× T7 Reaction Buffer, 2 μL of T7 ATP solution, 2 μL of T7 CTP solution, 2 μL of T7 GTP solution, 2 μL of T7 UTP solution, 0.5 μL of RNase inhibitor, 2 μL of T7 Enzyme Mix, and 7.5 μL of nuclease-free water; was prepared. In addition, the poly(A) tailing assay was not performed.
The prepared Rosa26-CTV-GK,
SpCas9 mRNA, and
Rosa26 sgRNA plasmids were then microinjected into each of the zygotes of
GKF/F mice. To prepare the zygotes, the
GKF/F mice were first subjected to IVF
136 (with minor modifications
137). Briefly, the 4-week-old
GKF/F female mice were intraperitoneally injected with pregnant mare’s serum gonadotrophin (PMSG) at a dose of 10 U/mouse. At 46 h after the PMSG injection, 10 U/mouse human chorionic gonadotrophin (hCG) was intraperitoneally injected. At 12 h after the hCG injection, oocytes from the oviducts of female mice, along with sperms from the cauda epididymides and vasa deferentia of 16-week-old, proven stud
GKF/F male mice, were isolated. To isolate oocytes, oviducts were briefly left on a filter paper, followed by incubating in a human tubal fluid medium (HTF)/GSH drop on an IVF dish (prepared by placing 200 μL of HTF solution supplemented with 125 mM GSH on a 35-mm dish to form a drop, followed by covering the drop with mineral oil and pre-balancing in a humidified incubator containing 5% CO
2 at 37 °C for 0.5 h before use). The ampulla was then torn down by forceps, and the cumulus oocyte masses inside were collected and transferred to another HTF/GSH drop. To isolate sperms, the cauda epididymides and vasa deferentia were briefly left on a filter paper, followed by penetration with a 26 G needle on the cauda epididymides 5 times. Sperms were then released to an HTF drop on sperm capacitation dish (prepared by placing 200 μL of HTF solution on a 35-mm dish to form a drop, followed by covering the drop with mineral oil and pre-balancing in a humidified incubator containing 5% CO
2 at 37 °C for 12 h before use) by slightly pressing/squeezing the cauda epididymides, followed by incubation in a humidified incubator containing 5% CO
2 at 37 °C for 0.5 h. The capacitated, motile sperms (located on the edge of each HTF drop) were then collected, followed by adding to the oocyte masses soaked in the HTF/GSH drop, 8 μL per drop. The IVF dishes containing oocyte masses and sperms were then cultured in a humidified incubator containing 5% CO
2 at 37 °C for 4 h, followed by collecting and washing oocytes in a KSOM drop (freshly prepared by placing 20 μL of KSOM medium on a 35-mm dish to form a drop, followed by covering the drop with mineral oil and pre-balancing in a humidified incubator containing 5% CO
2 at 37 °C for 0.5 h) twice. The oocytes were then cultured in an HTF/GSH drop on an IVF dish for another 12 h in a humidified incubator containing 5% CO
2 at 37 °C. The presumptive zygotes (in which 2 pronuclei and an extruded, second polar body could be observed) were then picked up. Some 10 pL of DNA mixture containing
Rosa26-CTV-GK plasmid (20 ng/μL final concentration),
SpCas9 mRNA (120 ng/μL final concentration), and
Rosa26 sgRNA (100 ng/μL), was microinjected into each of the zygotes and cultured in KSOM medium at 37 °C in a humidified incubator containing 5% CO
2 for 16 h. The zygotes/embryos at the two-cell stage were picked up and transplanted into pseudopregnant ICR female mice (8–10 weeks old, > 26 g; prepared by breeding the in-estrus female with a 14-week-old, vasectomized male at a day before the transplantation), 20 zygotes/embryos per mouse, and the offspring carrying the
LSL-
GK allele were further outcrossed 6 times to C57BL/6 mice before crossing with the
Villin-CreERT2 mice. Mice with intestine-specific
Kir6.2 knockout were generated using the same method as that for the intestine-specific
GCK knockout mice.
Transgenic mice with L cells expressing GFP were generated and validated as described previously
68. Briefly, mT/mG mice, which possess
LoxP sites flanking an mT cassette (driven under the
Actin promoter, resulting in red fluorescence in all tissues and cell types), were crossed with
Gcg-
Cre mice (Cre expression under the glucagon (
Gcg) or preproglucagon promoter). This leads to the removal of the mT cassette, allowing the expression of the downstream mG cassette specifically in L cells (as well as pancreatic α cells, although only intestinal tissues were analyzed in this study).
All of the mouse strains generated above were validated through genotyping, with details listed in Supplementary Table 1.
The following ages of mice were used for the experiments: a) for analyzing glucose levels in the blood and intestinal lumen, as well as circulating GLP-1 and insulin, 6-month-old wildtype mice and GK/Kir6.2 mutant-Knockin mice were used (with tamoxifen injected at 5 months old); b) for isolating primary intestinal cells, 5-month-old mice were used; c) for other experiments, 4-month-old mice were used.
Determination of GLP-1 secretion in patients with GK mutations
Patients were recruited from the Department of Endocrinology at Zhongshan Hospital, Fudan University. Mutations in the GK gene were identified using Illumina-based whole-exome sequencing (WES) using genomic DNA isolated from EDTA-anticoagulated peripheral blood samples, achieving a mean coverage depth of over 50× and a target region sensitivity of 98%. All candidate variants were subsequently validated through Sanger sequencing. Two female patients diagnosed with GCK-MODY, aged 42 and 22 years, were included in the study; they carried GK-R397L and GKΔ92-CT mutations, respectively. Written informed consents were obtained from each participant, and the study protocol was approved by the Institutional Review Board of Zhongshan Hospital, Fudan University (B2020-180R), in accordance with the principles outlined in the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practice Guidelines.
Before the assessment, participants discontinued all anti-diabetes medications for at least two weeks. They fasted overnight, beginning at 20:00 the night before the assessments. At 08:00 the following day, they took glucose orally at a dosage of 1 g/kg, dissolved in a total volume of 250 mL of water. Insulin, GLP-1, and C-peptide were determined as described previously
138. Briefly, blood samples were collected via intravenous catheters from the antecubital vein at 0-, 30-, and 120-min post-glucose administration. Blood was drawn into ice-cold K
2EDTA spray-coated tubes (cat. 366420, BD P800 Blood Collection System) containing 6 μL of diprotin A (5 mg/mL in DMSO) and into SST II Advance Tubes (cat. 367954, BD Vacutainer), with approximately 5 mL collected for each time point. The blood samples were centrifuged at 3,000×
g for 10 min at 4 °C, followed by plasma glucose measurement using the glucose oxidase method and GLP-1 measurement using an electrochemiluminescence immunoassay, both performed at KingMed Diagnostics Co., Ltd. (Shanghai, China). The blood collected in SST II Advance Tubes was similarly centrifuged at 3,000×
g for 10 min at 4 °C. After centrifugation, serum insulin and C-peptide concentrations were determined using a two-site electrochemiluminescence immunoassay, conducted on the cobas e 801 analytical unit (Roche Modular Analytics) at the Department of Endocrinology at Zhongshan Hospital.
Data reporting
The chosen sample sizes conform to those used in this field:
n = 6–8 samples for determining the levels of GLP-1, insulin and glucose in mice
116,139-141;
n = 4–12 samples for the secretion of GLP-1 in L cells and intestine tissues
142,143;
n = 3–5 samples to evaluate the levels of metabolites in cells
86,144 and tissues
86,144,145;
n = 9–10 cells to determine the local intracellular ATP:ADP ratios
70,71,146;
n = 3 replicates to determine GK activity and the binding affinity of GK towards glucose analogs
147,148; and
n = 2–3 samples to determine the expression levels of a specific protein
149. No statistical methods were used to predetermine the sample size. All experimental findings were repeated as stated in the figure legends, and all additional replication attempts were successful. For animal experiments, mice were housed under the same conditions or in the same place. For cell experiments, cells of each genotype were cultured in the same CO
2 incubator and were seeded in parallel. Each experiment was designed and performed along with proper controls, and samples for comparison were collected and analyzed under the same conditions. Randomization was applied wherever possible. For example, during MS analyses (for metabolites and proteins), samples were processed and subjected to the MS in random orders. In cell experiments, cells of each genotype were parallel seeded and randomly assigned to different treatments. Otherwise, randomization was not performed. For example, when performing IB, samples needed to be loaded in a specific order to generate the final figures. Blinding was applied wherever possible. For example, samples, cages, or dishes during sample collection and processing were labeled as code names that were later revealed by the individual who picked and treated animals or cells, but did not participate in sample collection and processing, until assessing the outcome. Similarly, during microscopy data collection and statistical analyses, the fields of view were chosen on a random basis, and were often performed by different operators, preventing potentially biased selection for desired conditions. Otherwise, blinding was not performed, such as the measurement of GK activity
in vitro, as different reagents were added for particular reactions.
Isolation of primary intestinal cells
Primary intestinal epithelial cells were isolated from mouse intestine as described previously
34, with minor modifications. In brief, mice were sacrificed through cervical dislocation, and the small intestinal tissues were quickly excised. Tissues were flushed with ice-cold Leibovitz-15 (L-15) medium to remove luminal contents, then divided in half, cut open longitudinally, and sliced into 1–2 mm segments in a Petri dish containing the ice-cold L-15 medium. For each mouse, the tissues were digested at 37 °C in a total of 2 mL of 0.4 mg/mL collagenase XI solution (freshly prepared in calcium-free Hanks Balanced Salt Solution (HBSS) for 30 min on an orbital shaker set to 80 rpm). After digestion, the cells were centrifuged at 300×
g at room temperature. The pellets were then resuspended in 10 mL of collagenase XI for an additional 15 min of digestion before being filtered through a 70-μm cell strainer (cat. 350350; BD Falcon) and subjected to another round of centrifugation at 300×
g at room temperature. The resulting cell pellets were resuspended in Dulbecco’s modified Eagle’s medium (DMEM) containing 25 mM glucose, and rinsed twice with DMEM. Cells were then suspended in DMEM supplemented with 10% (v/v) fetal bovine serum (FBS), 2 mM
l-glutamine, 100 IU of penicillin, and 100 μg/mL of streptomycin. The cells were then seeded into a Matrigel-coated 24-well dish. To prepare the Matrigel-coated dish, Matrigel was thawed at 4 °C for 12 h, followed by dilution with ice-cold PBS to a concentration of 3 mg/mL. The diluted Matrigel was then rapidly dispensed to the wells of a pre-chilled 24-well dish at 100 µL per well, followed by a gentle swirl to ensure even coverage. The plates were incubated at 37 °C for 1 h to allow complete gel polymerization, followed by a rinse with PBS under gentle agitation. The cells were then cultured in these dishes for 2 days at 37 °C in a humidified incubator with 5% CO
2 prior to further experiments.
Primary L cells were isolated as described previously
68. Briefly, primary intestinal epithelial cells were digested from mice engineered to express GFP in L cells as described above, except that collagenase XI at 1 mg/mL was used. In addition, the digested cells were resuspended in L-15 medium containing 10% FBS. The cell suspension was then filtered using a round-bottom polystyrene test tube with a cell strainer snap cap (cat. 352235; BD Falcon), and L cells were sorted using a FACSAria Fusion flow cytometer (BD Biosciences) equipped with 5 solid-state lasers (355 nm, 15 mW; 405 nm, 85 mW; 488 nm, 50 mW; 561 nm, 50 mW; and 640 nm, 100 mW) along with a forward scatter (FSC) detector, a side scatter (SSC) detector, and an 18-channel fluorophore detector. In this experiment, the 488-nm laser and the 530/30 filter were used to excite and detect the fluorescence of FITC (GFP), while the 561-nm laser and the 610/20 filter were employed for PE-Texas Red (tdTomato). Data were collected using the FACSDiva software (v8.0.2, BD Biosciences), followed by exporting in the FCS 3.1 format. The numbers of intestinal cells of each type were quantified using FlowJo software (v10.4.0, BD Biosciences). Gating strategies were defined based on the intensity of GFP fluorescence in comparison to the non-L (GFP-negative, tdTomato-positive) intestinal epithelial cells. Specifically, intact cells from each sample were selected using FSC-A and SSC-A (left panel of Supplementary Fig. S1j, on a linear scale), followed by FSC-H and FSC-A to exclude doublets (middle, on a linear scale). The fluorescence intensities of FITC (GFP) and PE-Texas Red (tdTomato) were measured and plotted (right, on a logarithmic scale). The plot displaying PE-Texas Red-negative and FITC-positive populations identifies the L cells. Typically, the purity of the sorted L cells exceeds 98%, which was confirmed by an Axio Observer live microscope (Zeiss) using a small aliquot. Some 1× 10
5 isolated L cells were collected for each measurement and incubated in DMEM medium supplemented with 10% FBS, 2 mM
l-glutamine, 100 IU of penicillin, and 100 μg/mL of streptomycin. The cells were maintained at 37 °C for 2 h in a humidified incubator with 5% CO
2 before proceeding with further experiments. It is important to note that due to the low abundance of L cell populations in the total intestinal epithelial cells, the sorting process for each sample often exceeded 0.5 h. Consequently, at least two sorters were used during the experiment.
Determination of GLP-1 and insulin secretion in mouse and cell
Levels of GLP-1 were measured as described previously
34,142,150, with minor modifications. Briefly, mice were fasted for 16 h (from 17:00 to 9:00 of the next day) and then subjected to refeeding, glucose gavage, or glucose injection. Mice were sacrificed at different time points, and 300 μL of blood was collected from each mouse into an ice-cold K
2EDTA spray-coated tube containing 6 μL of diprotin A (5 mg/mL). The blood was then divided into two portions: one for plasma preparation to determine GLP-1 levels and the other for serum to measure insulin levels (details provided below). Plasma was then prepared by centrifuging the blood at 3,000×
g for 10 min at 4 °C. From the plasma (the supernatant), 50 μL was used to determine the levels of GLP-1, and 10 μL for the measurement of PYY, CCK, and GIP, using the respective ELISA kits according to the manufacturer’s instructions.
For measuring GLP-1 secretion in L cells, cells were allowed to grow to approximately 70% confluence in a 12-well dish. Before measurement, cells were incubated in glucose-free DMEM for 2 h, washed twice with PBS, and incubated in KRBH buffer (GLP-1 Krebs-Ringer bicarbonate-HEPES buffer; containing 4.5 mM KCl, 138 mM NaCl, 4.2 mM NaHCO3, 1.2 mM NaH2PO4, 2.6 mM CaCl2, 1.2 mM MgCl2, and 10 mM HEPES, pH 7.4, adjusted with NaOH, along with 0.1% (w/v) BSA) supplemented with 0.01 mg/mL diprotin A and the desired concentrations of glucose for 1 h. The buffer was then collected and centrifuged at 3,000× g for 10 min at 4 °C, and 500 μL of the supernatant was taken for measuring GLP-1 levels using the GLP-1 EIA Kit.
Levels of insulin were measured as described previously
144,151. Briefly, blood was collected from mice as for the GLP-1 measurement, except that 1.5-mL Eppendorf tubes were used. The blood was left at room temperature for 30 min, followed by centrifugation at 3,000×
g for 10 min at 4 °C. Some 25 μL of serum (the supernatant) was used to determine insulin levels using the Mouse Ultrasensitive Insulin ELISA kit according to the manufacturer’s instructions.
For measuring insulin secretion in pancreatic β cells, cells were allowed to grow to approximately 70% confluence in a 12-well dish, followed by incubating in glucose-free DMEM for 2 h and washing with PBS twice, and incubation in KRBH buffer (insulin KRBH buffer: 3.6 mM KCl, 135 mM NaCl, 2 mM NaHCO3, 0.5 mM NaH2PO4, 1.5 mM CaCl2, 0.5 mM MgSO4, 10 mM HEPES, pH 7.4, adjusted with NaOH, and 0.2% BSA) containing varying concentrations of glucose for 1 h. Buffer was then collected and centrifuged at 3,000× g for 10 min at 4 °C, and 500 μL of supernatant was used to measure insulin levels using the Mouse Insulin ELISA Kit.
The concentrations of each hormone were calculated using a 5-parameter logistic fitted standard curve, generated from the Arigo Biolaboratories Co. website (
arigobio.cn/ELISA-calculator). The optical density (OD) values obtained from the microplate reader were truncated to three decimal places. For the analysis of relative hormone levels, the calculated concentrations were normalized to the average concentrations of the corresponding control group.
Measurement of glucose in blood and intestinal lumen
Levels of mouse blood glucose were measured as described previously
152. Briefly, mice were individually housed for a week before the experiment. After fasting for 16 h, mice were refed with food, or given glucose through gavage or injection. Blood glucose levels were measured at different time points through tail vein bleeding, drawing approximately 20 µL of blood from each mouse using the OneTouch UltraVue automatic glucometer (LifeScan).
Levels of glucose in the lumen of different intestinal segments were determined as described previously
153, with minor modifications. Briefly, mice were sacrificed by cervical dislocation, and the intestinal tissue was quickly excised and placed in ice-cold PBS. The tissue was then divided into segments: duodenum, jejunum, and ileum. Each segment was gently flushed with 50 µL of sterile PBS using a syringe and collected in an ice-cold Eppendorf tube. The volume of each sample was adjusted to 100 µL with ice-cold PBS. Samples were centrifuged at 10,000×
g for 15 min at 4 °C to remove debris. For analysis, 2 µL of the supernatant, diluted 500-fold (for the duodenum and jejunum) or 20-fold (for the other segments), was used to determine glucose concentration using the Glucose Assay Kit.
Cell lines
In this study, no cell line used is on the list of known misidentified cell lines maintained by the International Cell Line Authentication Committee (
iclac.org/databases/cross-contaminations/). The following cell lines were utilized: HEK293T cells (cat. CRL-3216) and STC-1 cells (cat. CRL-3254), both purchased from ATCC;
LRRC8A-knockout HEK293T cells (cat. YKO-HT22252), purchased from Ubigene; GLUTag cells (cat. SCC652), purchased from Merck; NCI-H716 cells (cat. CC0517) and β-TC-6 cells (cat. CC9034), purchased from CellCook; and INS-1 cells (cat. iCell-r036), purchased from iCellBioscience. Cells were maintained in DMEM, except NCI-H716 cells that were cultured in RPMI-1640 medium, and INS-1 cells in RPMI-1640 medium supplemented with 0.2% (w/v) HEPES and 0.05 mM β-mercaptoethanol. All culture media were supplemented with 10% FBS, 100 IU of penicillin, 0.15 mg/mL of streptomycin, and NaHCO
3 (3.5 g/L for DMEM and 2 g/L for RPMI-1640, adjusted to a pH of 7.2). Cells were kept at 37 °C in a humidified incubator with 5% CO
2. All cell lines were verified to be free of mycoplasma contamination and were authenticated by STR sequencing (performed by Immocell Biotechnology Corporation in Xiamen, China). For ectopic expression, PEI was used to transfect HEK293T cells at a final concentration of 10 μM, while Lipofectamine 2000 was used at a concentration of 7 µL for 3 µg of DNA to transfect β-TC-6 cells. Total DNA to be transfected for each plate was adjusted to the same amount by using relevant empty vector. Transfected cells were harvested at 24 h after transfection.
Lentiviruses, including those for knockdown or stable expression (at close-to-endogenous levels), were packaged in HEK293T cells by transfection using Lipofectamine 2000, as described previously
149. At 30 h post-transfection, the medium (DMEM supplemented with 10% FBS, MEM non-essential amino acids, and sodium pyruvate; approximately 2 mL in volume) was collected and centrifuged at 5,000×
g for 3 min at room temperature, yielding the supernatant containing lentivirus. To infect HEK293T cells, the supernatant was mixed with 10 μg/mL (final concentration) polybrene, and added to adhered HEK293T cells that were at 40% confluence, followed by centrifugation at 3,000×
g for 30 min at room temperature (spinfection). Cells were incubated for another 12 h before further treatments. For infection of STC-1, GLUTag, and β-TC-6 cells, 60% confluence was adopted. After 12 h of infection, the cells were passaged and cultured for another 12 h before an additional round of transfection prior to subsequent treatments. Cautions were taken as follows: as STC-1 cells after lentivirus infection can become fragile in the next passage, they should not be passaged with fewer than half of the original parental cells. In addition, after the cells have attached (2–3 h post-seeding), the culture medium should be refreshed to remove any unattached dead cells. Subculturing or passaging only used cells of less than 80% confluence.
Plasmids
The BacMam expression vectors carrying cDNA for SUR1 and Kir6.2, used for transfecting K
ATP into HEK293T cells for the inside-out patch clamping experiment
154,155, were provided by Dr. Lei Chen (Peking University). Full-length cDNAs used in this study were obtained either by PCR using cDNA from MEFs, or by purchasing from Origene or Sino Biological. Mutations of GK were generated by PCR-based site-directed mutagenesis using PrimeSTAR HS polymerase. Expression plasmids for various epitope-tagged proteins were constructed in the pcDNA3.3 vector (cat. K830001; Thermo Fisher Scientific) for ectopic expression in mammalian cells via transfection, in the pBOBI vector
156 for lentivirus packaging in mammalian cells, or in the pGEX-4T-1 vector (cat. 27-1542-01; Cytiva) for bacterial expression. The lentivirus-based vector pLV-H1-EF1a-puro (cat. SORT-B19; Biosettia) was used for the expression of shRNA in STC-1 and β-TC-6 cells. PCR products were verified by sequencing (Invitrogen, China).
Escherichia coli (
E. coli) strain DH5α (cat. PTA-1977) was purchased from ATCC, and Stbl3 (cat. C737303) from Thermo Fisher Scientific. All plasmids were amplified in
E.
coli strain DH5α, except those for mutagenesis in Stbl3. All plasmids used in this study were purified by the CsCl density gradient ultracentrifugation method. All expression plasmids constructed in this study have been deposited to Addgene (
addgene.org/Sheng-cai_Lin/).
Immunoprecipitation and IB
The interaction between endogenous GK and K
ATP was determined as described previously
157, with minor modifications. Briefly, STC-1 cells of a 10-cm dish (grown to 80% confluence) were collected for immunoprecipitation of GK. Cells were lysed with 750 μL/dish of ice-cold Triton lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% (v/v) Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, with protease inhibitor cocktail), followed by sonication using a VCX 130PB sonicator (Sonics; equipped with a 5/64’’ (2 mm) stepped microtip, cat. 630-0423) for 3–5 sec at 30% maximum power on ice, and centrifugation at 4 °C for 15 min. Cell lysates were incubated with anti-GK antibody overnight. Overnight protein aggregates were pre-cleared by centrifugation at 20,000×
g for 10 min; protein A/G beads (1:250, pre-balanced with lysis buffer) were then added to the lysate/antibody mixture for another 3 h at 4 °C. The beads were centrifuged and washed 3 times each with 100 times the volume of ice-cold lysis buffer (by centrifuging at 2,000 g) at 4 °C and then mixed with an equal volume of 2× SDS sample buffer before IB.
The interaction between ectopically expressed GK and K
ATP was determined as described previously
158,159, with minor modifications. Briefly, a 6 cm-dish of HEK293T cells was transfected with various combinations of expression plasmids that include different epitope tags. At 24 h after transfection, cells were collected and lysed in 500 µL of ice-cold Triton lysis buffer, followed by sonication and centrifugation at 4 °C for 15 min. For IP of the FLAG epitope tag, anti-FLAG M2 Affinity Gel (1:200, pre-balanced in lysis buffer) was added into the supernatant and mixed for 1 h at 4 °C. The beads were washed with 200 times the volume of ice-cold Triton lysis buffer wash buffer 3 times at 4 °C, and then eluted with 30 μL of FLAG Peptide (400 μg/mL final concentration) for another 30 min at 4 °C. Some 30 μL of eluent was then collected, mixed with 7.5 μL of 5× SDS buffer, and then subjected to IB analysis. Note that samples containing K
ATP, particularly those containing the transmembrane segments, were not boiled to avoid the formation of insoluble aggregates that would fail to run into the SDS-PAGE.
To analyze the levels of GK, MCT1 and G6Pase in L cells, cells grown to 70–80% confluence in a 6-well dish were lysed with 250 μL of ice-cold Triton lysis buffer. The lysates were then centrifuged at 20,000× g for 10 min at 4 °C, and an equal volume of 2× SDS sample buffer was added into the supernatant. Except MCT1, samples were boiled for 10 min and then directly subjected to immunoblotting.
To analyze the levels of GK in intestinal tissues, a protocol designed to enhance the protein recovery rate from tissue samples was used
160, with minor modifications. Briefly, freshly excised intestinal tissues were quickly infused with 10 mL of ice-cold PBS (supplemented with a 2× protease inhibitor cocktail) using a syringe to fill their lumens. Approximately 100 mg of tissue was then transferred to a 1.5-mL Eppendorf tube, followed by homogenization on ice using an electrical disperser (T 10 basic ULTRA-TURRAX 558 from IKA, equipped with an S10N-5G Dispersing Element) in 1 mL of 2× Sample Buffer (100 mM Tris-HCl, pH 6.8, 4% (m/v) SDS, 20% (v/v) glycerol, and 4% (v/v; added freshly before use) 2-mercaptoethanol). The homogenate was briefly vortexed and then sonicated on ice. It is important to avoid generating bubbles during sonication, given the presence of SDS in the buffer, as they can disrupt the stability of proteins in the lysate. Some 50 µL of the homogenate was then transferred to a new 1.5-mL Eppendorf tube, diluted with 450 µL of distilled water, and mixed with 50 µL of ice-cold, 10× TCA solution (72% trichloroacetic acid in water). This mixture was vortexed for 30 sec and then centrifuged at 20,000×
g for 10 min at 4 °C. The resulting pellet was suspended in 1 mL of ice-cold acetone and centrifuged at 20,000×
g for 10 min at 4 °C. This step was repeated two more times. After the final centrifugation, the sediment was roughly dried on ice for 5 min and then suspended in 200 µL of 2× Sample Buffer for 2 h at room temperature, with sonication every 30 min for 2–3 sec each time until the pellet was completely dissolved. Samples were then mixed with 2 µL of bromophenol blue solution (2.5% in water), boiled for 10 min, and then subjected to immunoblotting. In this study, all samples were subjected to IB on the same day of preparation, avoiding freeze-thaw cycles.
For IB, the SDS-polyacrylamide gels were prepared in-house, as described previously
141. The thickness of the gels used in this study was 1.0 mm. Samples of less than 10 μL were loaded into wells, and the electrophoresis was run at 100 V (by PowerPac HC High-Current Power Supply, Bio-Rad) in a Mini-PROTEAN Tetra Electrophoresis Cell (Bio-Rad). In this study, all samples were resolved on 8% resolving gels, except those for GK
Δ92-CT which were run on 15% gels (prepared as those for 8%, except that a final concentration of 15% Acryl/Bis was added to the resolving gel solution). In addition, Kir6.2 and MCT1 were resolved on 10% gels. The resolved proteins were then transferred to the PVDF membrane (0.45 μm, cat. IPVH00010, Merck) as described previously
141. The PVDF membrane was then blocked by 5% (w/v) non-fat milk dissolved in TBST for 2 h on an orbital shaker at 60 rpm at room temperature, followed by rinsing with TBST (40 mM Tris, 275 μM NaCl, 0.2% (v/v) Tween-20, pH 7.6) twice, 5 min each. The PVDF membrane was then incubated with the desired primary antibody overnight at 4 °C on an orbital shaker at 60 rpm, followed by rinsing with TBST three times, 5 min each at room temperature, and then the secondary antibodies for 3 h at room temperature with gentle shaking. The secondary antibody was then removed, and the PVDF membrane was further washed with TBST 3 times, 5 min each, at room temperature. PVDF membrane was incubated in an ECL mixture (by mixing equal volumes of ECL solution and Peroxide solution for 5 min), then life with Medical X-Ray Film (FUJIFILM). The films were then developed with X-OMAT MX Developer (Carestream), and X-OMAT MX Fixer and Replenisher solutions (Carestream) on a Medical X-Ray Processor (Carestream) using Developer (Model 002, Carestream). Films were then scanned using a Perfection V850 Pro scanner (Epson) with an Epson Scan software (v.3.9.3.4), cropped using Photoshop 2023 software (Adobe), and formatted into panels using Illustrator 2022 (Adobe). Levels of total proteins and phosphorylated proteins were analyzed on separate gels, and representative immunoblots are shown. Uncropped immunoblots are provided.
Confocal microscopy
The plasma membrane localization of Lck-PercevalHR indicator was determined as described previously
145,161, with minor modifications. Briefly, cells grown to 80% confluence on coverslips in 6-well dishes were fixed for 20 min with 4% (v/v) formaldehyde in PBS at room temperature. The coverslips were rinsed twice with PBS and permeabilized with 0.1% (v/v) Triton X-100 in PBS for 5 min at 4 °C. After rinsing twice with PBS, the sections were blocked with PBS containing 5% BSA for 30 min at room temperature. Then the coverslips were incubated with anti-N-cadherin antibody (1:100, diluted in PBS) overnight at 4 °C. The cells were then rinsed 3 times with 1 mL of PBS, and then incubated with Alexa Fluor 594-conjugated, goat anti-rabbit IgG antibody (1:200 in PBS) for 8 h at room temperature in the dark. Cells were washed another 4 times with 1 mL of PBS and then mounted on slides using ProLong Diamond antifade mountant. Confocal microscopy images were taken using a STELLARIS 8 FALCON (Leica) system equipped with HyD SMD detectors and an HC PL APO CS2 63×/1.40 OIL objective (Leica). All parameters were kept unchanged between imaging. Images were taken and analyzed by LAS X Software (v. 3.0.2.16120, Leica), and formatted using Photoshop 2023 software (Adobe).
The local intracellular fluctuation of the ADP:ATP ratios near the plasma membrane was assessed using the Lck-PercevalHR indicator, as previously described
71,146. In brief, cells expressing Lck-PercevalHR were cultured in 35-mm glass-bottom dishes (cat. D35-20-10-N, In Vitro Scientific) to 60–80% confluence. Cells were starved of glucose for 2 h, followed by equilibration at 37 °C and 5% CO
2 in a humidified incubation chamber for an additional 30 min (cat. STXG-WELSX-SET; TOKAI HIT). Cells were imaged using a spinning disk confocal microscope (Olympus IXplore SpinSR, equipped with a Yokogawa CSU-W1 spinning disk with a 50-μm diameter) and a 60×/1.4 N.A. objective, utilizing the cellSens Dimension software (ver. 4.1.1). In the meantime, ProLong Live antifade reagent was added to the medium before imaging. Lck-PercevalHR was excited with 490-nm and 400-nm lasers, respectively. The emitted light was separated from the excitation light by a 490-nm short-pass dichroic mirror and collected through a 520/10 nm band-pass filter. The ratio of the intensities of the emission light at 490 nm to that at 400 nm reflects the ATP:ADP ratio.
FRET-FLIM assay
FRET-FLIM experiments were carried out as described previously
133,162, with minor modifications. Briefly, STC-1 cells stably expressing GFP-GK (donor only) or different combinations of GFP-GK (FRET donor), Kir6.2-mCherry (FRET acceptor), and mCherry-SUR1 (FRET acceptor) were cultured in 35-mm glass-bottom dishes until they reached 60–80% confluence. Cells were starved for glucose or treated with 25 mM glucose, followed by determining the fluorescence lifetime of GFP in different cells cultured in a humidified chamber with 5% CO
2 at 37 °C using a STELLARIS 8 FALCON (Leica) system equipped with HyD X and HyD SMD detectors and an HC PL APO CS2 63x/1.40 OIL objective (Leica). Cells were excited with a 482-nm laser via the systems’ tunable White Light Laser (WLL), and photon arrival times were recorded with a HyD X detector covering the GFP emission spectrum (490-550 nm). All parameters were kept unchanged between imaging. Images were taken, and the following metrics were analyzed using LAS X Software (Leica): unquenched donor lifetime (measured with donor expression only), quenched donor lifetime (specifically experiencing FRET with the acceptor), FRET efficiency, and donor-acceptor distance. In all experiments, the position of the focal plane was actively stabilized using the Leica Auto Focus Control (AFC) to prevent any focal drift or focus artifacts. The following formulas were used:
a) For quenched donor lifetime (): where represents the unquenched donor lifetime, and is the rate constant of the FRET process, both were measured in the experiments.
b) For FRET-FLIM efficiency (): .
c) For donor-acceptor distance (
):
where
is Förster radius, which is 5 nm for the GFP-mCherry pair
91.
Determination of OCRs
To measure OCRs in STC-1 cells in the presence of different concentrations of glucose, cells were plated at a density of 10,000 cells per well in a 96-well Seahorse XF Cell Culture Microplate (Agilent) in full medium (DMEM with 25 mM glucose supplemented with 10% FBS) and allowed to incubate overnight before the experiment. To specifically measure the utilization of pyruvate/lactate, fatty acids, and glutamine in these cells, UK5099 (20 μM), etomoxir (40 μM), or BPTES (20 μM) were added to the full medium, and the cells were pre-incubated for 20 h. Before measuring OCR, the medium was replaced with Seahorse XF Base Medium supplemented with 2 mM glutamine (GlutaMAX), 1 mM sodium pyruvate, and the desired concentrations of glucose. The cells were then placed in a CO2-free XF96 Extracellular Flux Analyzer Prep Station (Agilent) at 37 °C for 1 h. OCR was measured at 37 °C using an XF96 Extracellular Flux Analyzer (Agilent) equipped with a Seahorse XFe96 sensor cartridge (Agilent), which had been equilibrated in Seahorse XF Calibrant solution in a CO2-free incubator at 37 °C overnight. The assay was performed on the Seahorse XFe96 Analyzer (Agilent) at 37 °C, following the manufacturer’s instructions. Oligomycin A was used at a final concentration of 4 μM in the assay to determine ATP-dependent OCR, which was calculated by subtracting the OCR of cells treated with oligomycin from that of untreated cells. Data were collected using Wave 2.6.1 Desktop software (Agilent) and exported to Prism 9 (GraphPad) for further analysis according to the manufacturer’s instructions.
Measurement of adenylates
ATP, ADP, and AMP from intestinal and pancreatic β cell lines, as specified in the legends, were analyzed by CE-MS as described previously
86,163, with minor modifications. Briefly, each measurement required cells collected from a 10-cm dish (60–70% confluence). Cells were rinsed with 20 mL of 5% (m/v) mannitol solution (dissolved in water) and instantly frozen in liquid nitrogen. Cells were then lysed with 1 mL of methanol containing IS1 (50 µM
l-methionine sulfone, 50 µM
d-campher-10-sulfonic acid, dissolved in water; 1:500 (v/v) added to the methanol and used for standardization of the metabolite intensity and to adjust the migration time), and were scraped from the dish. The lysate was then mixed with 1 mL of chloroform and 400 μL of water by 20 sec of vortexing. After centrifugation at 15,000×
g for 15 min at 4 °C, 450 μL of aqueous phase was collected and then filtered through a 5-kDa cutoff filter (cat. OD003C34, PALL) by centrifuging at 12,000×
g for 3 h at 4 °C. In parallel, quality control samples were prepared by combining 10 μL of the aqueous phase from each sample and then filtering alongside the experimental samples. The filtered aqueous phase was then freeze-dried in a vacuum concentrator at 4 °C, and then dissolved in 100 μL of water containing IS2 (50 µM 3-aminopyrrolidine dihydrochloride, 50 µM N,N-diethyl-2-phenylacetamide, 50 µM trimesic acid, 50 µM 2-naphtol-3,6-disulfonic acid disodium salt, dissolved in methanol; used to adjust the migration time). A total of 20 μL of re-dissolved solution was then loaded into an injection vial (cat. 9301-0978, Agilent; equipped with a snap cap (cat. 5042-6491, Agilent)). Before CE-MS analysis, the fused-silica capillary (cat. TSP050375, i.d. 50 µm × 80 cm; Polymicro Technologies) was installed in a CE/MS cassette (cat. G1603A, Agilent) on the CE system (Agilent Technologies 7100). The capillary was then pre-conditioned with Conditioning Buffer (25 mM ammonium acetate, 75 mM diammonium hydrogen phosphate, pH 8.5) for 30 min, followed by balancing with Running Buffer (50 mM ammonium acetate, pH 8.5; freshly prepared) for another 1 h. CE-MS analysis was run in anion mode, during which the capillary was washed by Conditioning Buffer, followed by injection of the samples at a pressure of 50 mbar for 25 sec, and then separation with a constant voltage at –30 kV for another 40 min. Sheath Liquid (0.1 μM hexakis(1H, 1H, 3H-tetrafluoropropoxy)phosphazine, 10 μM ammonium trifluoroacetate, dissolved in methanol/water (50% v/v); freshly prepared) was flowed at 1 mL/min through a 1:100 flow splitter (Agilent Technologies 1260 Infinity II; actual flow rate to the MS: 10 μL/min) throughout each run. The parameters of MS (Agilent Technologies 6545) were set as: a) ion source: Dual AJS ESI; b) polarity: negative; c) nozzle voltage: 2,000 V; d) fragmentor voltage: 110 V; e) skimmer voltage: 50 V; f) OCT RFV: 500 V; g) drying gas (N
2) flow rate: 7 L/min; h) drying gas (N
2) temperature: 300 °C; i) nebulizer gas pressure: 8 psig; j) sheath gas temperature: 125 °C; k) sheath gas (N
2) flow rate: 4 L/min; l) capillary voltage (applied onto the sprayer): 3,500 V; m) reference (lock) masses: m/z 1,033.988109 for hexakis(1H, 1H, 3H-tetrafluoropropoxy)phosphazine, and m/z 112.985587 for trifluoroacetic acid; n) scanning range: 50-1,100 m/z; and o) scanning rate: 1.5 spectra/s. Data were collected using MassHunter LC/MS acquisition 10.1.48 (Agilent Technologies), and were processed using Qualitative Analysis B.06.00 (Agilent Technologies). Levels of AMP, ADP, and ATP were measured using full scan mode with m/z 346.0558, 426.0221, and 505.9885, respectively. Note that a portion of ADP and ATP could lose one phosphate group during in-source-fragmentation, thus leaving the same m/z ratios as AMP and ADP, and this was corrected according to their different retention times in the capillary. Therefore, the total amount of ADP is the sum of the latter peak of the m/z 346.0558 spectrogram and the former peak of the m/z 426.0221 spectrogram, and ditto for ATP. For quantification of AMP, ADP and ATP, [U-
13C,
15N]AMP, [U-
13C,
15N]ADP and [U-
13C,
15N]ATP dissolved in individual lysates were used to generate corresponding standard curves by plotting the ratios of detected labelled AMP, ADP or ATP (areas) to the products of IS1 and IS3, against the added concentrations of labelled AMP, ADP or ATP. Levels of [U-
13C,
15N]AMP, [U-
13C,
15N]ADP and [U-
13C,
15N]ATP were measured using full scan mode with m/z 351.0408, 431.0071, and 510.9735, respectively. The amounts of AMP, ADP and ATP were estimated according to standard curves, and were then divided by protein wet weight. The protein wet weight of each sample was determined by Bradford assay after dissolving the naturally dried protein sediment with 0.2 M KOH at room temperature. Cell density of 1.1 g/mL was used
164.
ATP, ADP, and AMP from intestinal and pancreatic β cell lines, as specified in the legends, as well as from intestinal tissues, were analyzed by standard HPLC-MS as previously described
165,166, with minor modifications. In brief, each measurement required cells collected from a 10-cm dish (60–70% confluence). Cells were rinsed with 20 mL of PBS and instantly frozen in liquid nitrogen. To analyze adenylates in intestinal tissues, mice were anesthetized, and the duodenum or jejunum tissues were quickly excised. The tissues were then rinsed with 20 mL of PBS and freeze-clamped. Following this, cells and tissues were lysed with 1 mL of methanol. The lysate was then mixed with 1 mL of chloroform and 400 μL water (containing 4 μg/mL [U-
13C]-glutamine used for standardization of the metabolite intensity and to adjust the migration time), by 20 sec of vortexing. After centrifugation at 15,000×
g for 15 min at 4 °C, 450 μL of aqueous phase was collected, lyophilized in a vacuum concentrator at 4 °C and then dissolved in 80 μL of 50% (v/v) acetonitrile. Some 50 μL of supernatant was loaded into an injection vial (5182-0714, Agilent Technologies; with an insert (HM-1270, Zhejiang Hamag Technology)) equipped with a snap cap (HM-2076, Zhejiang Hamag Technology). Measurements of adenylate levels were based on a previous study
167 using a QTRAP MS (QTRAP 5500, SCIEX) interfaced with a UPLC system (ExionLC AD, SCIEX). A total of 2 μL of each sample was loaded onto a HILIC column (ZIC-pHILIC, 5 μm, 2.1 × 100 mm, PN: 1.50462.0001, Millipore). The mobile phase consisted of 15 mM ammonium acetate containing 3 mL/L ammonium hydroxide (> 28%, v/v) in LC-MS grade water (mobile phase A) and LC-MS grade 90% (v/v) acetonitrile in LC-MS grade water (mobile phase B) run at a flow rate of 0.2 mL/min. Metabolites were separated with the following HPLC gradient elution program: 95% B held for 2 min, then to 45% B in 13 min, held for 3 min, and then back to 95% B for 4 min. The MS was run on a Turbo V ion source in negative mode with a spray voltage of –4,500 V, source temperature of 550 °C, gas no.1 at 50 psi, gas no.2 at 55 psi, and curtain gas at 40 psi. Metabolites were measured using the multiple reaction monitoring mode, and declustering potentials and collision energies were optimized using analytical standards. The following transitions were used for monitoring each compound: 505.9/158.9 and 505.9/408.0 for ATP; 425.9/133.9, 425.9/158.8 and 425.9/328.0 for ADP; 345.9/79.9, 345.9/96.9 and 345.9/133.9 for AMP; and 149.9/114 for [U-
13C]-glutamine. Data were collected using Analyst software (v.1.7.1, SCIEX), and the relative amounts of metabolites were analyzed using MultiQuant software (v.3.0.3, SCIEX). Similar to CE-MS analysis, a portion of ADP and ATP could lose one or two phosphate groups during in-source-fragmentation, thus leaving the same m/z ratios as AMP and ADP, which were corrected according to their different retention times in the column. For quantification of AMP, ADP and ATP, [U-
13C,
15N]AMP, [U-
13C,
15N]ADP, and [U-
13C,
15N]ATP dissolved in individual lysates were used to generate corresponding standard curves as in CE-MS analysis, except that [U-
13C]-glutamine was used as an internal standard. The following transitions were used for monitoring each compound: 520.9/158.9 for [U-
13C,
15N]ATP; 440.9/143.9 for [U-
13C,
15N]ADP; and 360.9/79.9 for [U-
13C,
15N]AMP. The amounts of AMP, ADP, and ATP were estimated according to standard curves, and were then divided by protein wet weight as described in the CE-MS analysis.
The dissolved samples were left on ice or at 4 °C for no more than 4 h before injection into the MS.
Measurement of glycolytic metabolites
Levels of FBP from L cells were analyzed using CE-MS as described previously
162. The sample preparation, CE conditions, and parameters for mass spectrometry are outlined in the CE-MS analysis section of “Measurement of Adenylates”, except that an m/z value of 338.9888 was used to quantify FBP.
The levels of 2-PG and 3-PGA in L cells were analyzed using an improved HPLC-MS method that offers enhanced sensitivity compared to standard HPLC-MS. The HPLC conditions and separation program were revised based on the chemical properties of 2-PG and 3-PGA. Briefly, samples were collected and processed following the standard HPLC-MS analysis described in the section “Measurement of Adenylates.” The lyophilized samples were re-dissolved in 60 μL of 50% (v/v) acetonitrile. Some 50 μL of supernatant was then loaded into an injection vial (5182-0714, Agilent Technologies; with an insert (HM-1270, Zhejiang Hamag Technology)) equipped with a snap cap (HM-2076, Zhejiang Hamag Technology). Measurements of 2-PG and 3-PGA levels were performed using a triple quadrupole MS (Xevo TQ Absolute; Waters) interfaced with a UPLC system (ACQUITY Premier; Waters). A total of 2 μL of each sample was loaded onto a HILIC column (Atlantis Premier BEH Z-HILIC VanGuard FIT Column, 1.7 μm, 2.1 × 100 mm, PN: 186009982; Waters). The mobile phase consisted of 15 mM ammonium bicarbonate supplemented with 2.5 µM medronic acid in LC-MS grade water (mobile phase A) and 15 mM ammonium bicarbonate containing 2.5 µM medronic acid in LC-MS grade 90% (v/v; in LC-MS grade water) acetonitrile (mobile phase B), run at a flow rate of 0.35 mL/min. Metabolites were separated using the following HPLC gradient elution program: 90% B held for 1 min, then reduced to 65% B over 8 min, held for an additional 2 min, and returned to 90% B in 30 sec, holding for 210 sec. The MS was run on a Zspray ion source in negative mode with the following parameters: a) capillary voltage: –2,000 V; b) desolvation temperature: 500 °C; c) desolvation gas flow: 500 L/h; and d) cone gas flow: 150 L/h. Metabolites were measured using the multiple reactions monitoring mode, with parameters of cone voltage and collision energy optimized using analytical standards. The following transitions were used for monitoring each compound: 185.00/97.00 for 2-PG, 185.01/97.00 for 3-PGA, and 150/114 for [U-13C]-glutamine. Data were collected, and the relative amounts of metabolites were analyzed using MassLynx (version 4.2; Waters).
Levels of other glycolytic intermediates in L cells were measured using standard HPLC-MS. The sample preparation, HPLC conditions, and parameters for mass spectrometry are outlined in the HPLC-MS analysis section of “Measurement of Adenylates”, except that 3 μL of sample was loaded onto the HILIC column. Specifically, to determine glucose uptake and catabolism in STC-1 cells using the [U-13C]-glucose tracer, the cells were starved of glucose for 2 h, then incubated with the tracer at the desired concentration for another 1 h, followed by rinsing with PBS before sample preparation. The following transitions were used for monitoring each compound: 179.1/89.1 for glucose; 259.0/96.9 for G6P; 259.0/97.0 for F6P; 169.0/97.0 for DHAP, 169.0/79.0 for G3P, 265.0/78.9 for 1,3-BPG, 166.9/78.9 for PEP, 87.0/43.0 for pyruvate, 185.0/92.0 for [U-13C]-glucose, 265.0/97.0 for [U-13C]-G6P, and 264.9/79.0 for [U-13C]-F6P. Data were collected using Analyst software (v.1.7.1, SCIEX), and the relative amounts of metabolites were analyzed using MultiQuant software (v.3.0.3, SCIEX).
Protein expression
The expression plasmids for GK and mutants were constructed by inserting respective cDNAs into pGEX-4T-1 vectors for expressing GST-tagged recombinant proteins as described previously
79,148. Briefly, plasmids were transformed into the
E.
coli strain BL21 (DE3) (cat. EC0114; Thermo Fisher Scientific), followed by culturing in LB medium in a shaker at 200 rpm at 37 °C. Expression of the proteins in the bacteria was induced with 0.1 mM IPTG at an OD
600 of 1.0. After incubating for another 12 h at 160 rpm at 16 °C, the cells were collected and homogenized in a GST binding buffer (PBS supplemented with 10 mM β-mercaptoethanol and 1% Triton X-100) on ice. The homogenates were then sonicated on ice and subjected to centrifugation at 150,000×
g for 30 min at 4 °C, followed by purification with Glutathione Sepharose 4 Fast Flow Gel (pre-balanced with GST binding buffer) at 4 °C. The Glutathione Sepharose gel was then washed with 100 times the volume of ice-cold PBS, and then incubated with 10 U/mL thrombin (dissolved in PBS, freshly prepared) at 4 °C for 8 h to cleave the GST tag from GK, releasing the GK from the Glutathione Sepharose gel. Proteins were concentrated to approximately 3 mg/mL by ultrafiltration (Millipore, UFC905096) at 4 °C, then subjected to gel filtration (Cytiva, Superdex 200), balanced with the same buffer as that used in the enzymatic assays or patch-clamp assays before proceeding with each experiment.
Enzymatic activity
Enzymatic activity of GK was determined using glucose as a substrate and ATP as a co-substrate, employing two distinct methods: one to determine the velocity of G6P formation to represent the catalytic activity of GK towards glucose, which was coupled with the G6PD enzyme
168, and the other measured the velocity of ATP consumption by coupling the PK-LDH enzymes
147, as previously described.
For the G6PD-coupled assay, 1.8 mL of G6PD Reaction Buffer (20 mM HEPES, pH 7.7, 125 mM KCl, 7.5 mM MgCl2, 5 mM ATP, 0.5 mM NADP+, 0.5 U/mL G6PD, and the desired concentrations of glucose) was added into a 4-mL cuvette on a Lambda 365 Spectrophotometer (PerkinElmer) equipped with magnetic stirrer, set to stir at 140 rpm and at a temperature of 37 °C. After a 10-min incubation, the cuvette was taken out and gently tapped to eliminate bubbles. The cuvette was then placed back on the spectrophotometer until the baseline OD340 reading became flat and stable, with the change in the value remaining less than 0.001. The reaction was initiated by addition of 6 μL of GK diluted with the Reaction Buffer to the desired concentration and pre-incubated with its activator at the chosen concentration at 37 °C for 5 min. Data were collected using UV WinLab software (v.7.1.0.68, PerkinElmer) and exported to GraphPad Prism (ver. 10.1.2) for further analysis. The concentrations of glucose consumed were calculated by the extinction coefficient for NADPH at 340 nm, which is 6,220 cm–1 M–1. The initial velocity of glucose consumption (measured within 30 sec to 1 min after the reaction was initiated), represented by the initial rates of NADPH formation, was then calculated at each glucose concentration, plotted, and the Km and Vmax values were determined based on the curves fitted to the Michaelis-Menten equation. The values of S0.5 were calculated by the Hill equation.
The PK-LDH-coupled assay was performed in a similar manner to the G6PD-coupled assay, with the use of PK-LDH Reaction Buffer (100 mM Tris-HCl, pH 7.9, 100 mM KCl, 20 mM MgCl2, 0.2 mM NADH, 1.0 mM PEP, 20 U/mL LDH, 15 U/mL PK, 5 mM ATP, and the desired concentrations of glucose). The concentrations of ATP consumed were calculated using the extinction coefficient for NADH at 340 nm, which is 6,220 cm–1 M–1. Following this, the initial rates of NADH formation were calculated to reflect the velocity of ATP consumption.
Determination of binding affinity of GK towards glucose analogs
The binding affinity of GK towards glucose analogs was determined using fluorescence-labelled differential scanning fluorimetry (FL-DSF) methods, as described previously
169. Briefly, GK was conjugated with FITC-NHS dye via the amine coupling method prior to FL-DSF analysis. Before conjugation, 1 mL of Labelling Buffer (8 mM Na
2HPO
4, 137 mM NaCl, 1.47 mM KH
2PO
4 and 2.68 mM KCl, pH 8.0) containing 20 µM GK was mixed with 1 µL of FITC-NHS dye (100 µM, prepared from a 100 mM stock solution dissolved in anhydrous DMSO). The mixture was gently rotated for 30 min at 4 °C in the dark. The reaction was then terminated by adding a Tris-HCl solution (1 M stock solution, pH 8.5) to achieve a final concentration of 10 mM. Unconjugated dye was removed using a desalting column (Zeba Spin, 7 kDa MWCO; cat. 89891; Thermo Fisher Scientific) pre-equilibrated with the Labelling Buffer. Some 10 µL of FITC-GK conjugate was then incubated with 10 µL of MDGP, 2-DG, or glucose as a control, at varying concentrations at room temperature for 5 min in a PCR tube.
FL-DSF analysis was conducted using a CFX96 Touch Real-Time PCR Detection System (Bio-Rad) with the “SYBR” program. A temperature gradient was established, ranging from 25 °C to 95 °C, with increments of 0.2 °C. The system held each temperature for 5 sec before taking measurements. Data, represented as fluorescent intensity (F), were collected at 2-s intervals. Following this, melting curves were generated using the CFX Manager software (version 3.1.1517.0823; Bio-Rad). The negative first derivative (–dF/dT) at each temperature (T) of the melting curve was calculated using the “Melt Curve Derivative Results” function of the software. The melting temperature (Tm), defined as the maximum value of –dF/dT, was obtained using the “Draw Data” function of OriginPro 2024 software. The ΔTm was calculated by subtracting the Tm of concentration of each glucose analog from the Tm of the glucose analog-free group. These ΔTm values were then plotted against the concentrations of the glucose analogs using GraphPad Prism 10.1.2 software. The binding constants (Kd) were determined using a one-site specific binding model.
Electrophysiology
Whole-cell patch clamp was performed as described previously, with minor modifications
170,171. Briefly,
LRRC8A-KO HEK293T cells were used for minimizing the current caused by chloride ion efflux through volume-regulated anion channels (VRACs), which could otherwise interfere with the K
ATP-mediated current resulting from K
+ influx when a negative voltage was applied during whole-cell patch clamping
172,173. Cells were transfected with Kir6.2 and SUR1 as described in the “Cell lines” section, except that Lipofectamine 3000 was used as the transfection reagent, and cells at a density of 10
6 cells/mL were infected with a lentivirus carrying WT GK or its G80A and L309P mutants as described in the “Cell lines” section. For the other experiments, STC-1 and GLUTag cells were used, and were directly infected with lentiviruses carrying WT GK or its mutants. These cells were then incubated in glucose-free DMEM for 2 h, followed by trypsinization. The cells were then cultured in a 12-well dish containing a poly-
l-lysine-coated coverslip in glucose-free DMEM for another 1 h in a humidified chamber with 5% CO
2 at 37 °C, for cells to attach to the coverslips. These coverslips were then transferred to a 500-μL recording chamber that was placed on the objective lens of an inverted Axio Examiner microscope (Zeiss) and filled with bath (extracellular) solution. This solution contained 10 mM HEPES, 120 mM KCl, 5 mM NaCl, 1 mM MgCl
2, 2 mM CaCl
2, and various concentrations of glucose, supplemented with NMDG and MSA to bring the pH to ~ 7.2 and final osmolarity to ~ 300 mOsm, as determined by a VAPRO vapor pressure osmometer (cat. 5600; ELITechGroup). Each bath solution was loaded into the syringe of a perfusion system driven by a multichannel peristaltic pump set to a flow rate of 2 mL/min. Borosilicate glass electrodes (micropipettes) with a tip resistance of 2–4 MΩ were pulled using a Next Generation Micropipette Puller (P-1000; Sutter Instrument) and filled with pipette (cytosolic face) buffer containing 150 mM KCl, 2 mM MgCl
2, 20 mM HEPES supplemented with 2 mM Mg-ATP, 200 μM ADP (freshly prepared), and the desired concentrations of glucose (supplemented with NMDG and MSA to bring the pH to ~ 7.2 and final osmolarity to 330 mOsm). An agar bridge filled with 3 M KCl served as the reference electrode. The plasma membrane of a cell was suctioned using the micropipette to form a gigaohm seal, followed by membrane rupture via gentle suction to achieve whole-cell access. The whole-cell K
+ current was recorded by applying a ramp protocol from –120 mV to +120 mV (for HEK293T cells) or from –140 mV to +60 mV (for STC-1 and GLUTag cells) over a duration of 400 msec, repeated every 2 sec, using an Axon Axopatch 200B Microelectrode Amplifier (Molecular Devices) equipped with a Digidata 1440A Low-noise Data Acquisition System (Molecular Devices). The glucose concentration was altered either in the bath solution using the perfusion system or in the pipette solution by using a pipette filled with different glucose concentrations. Data collection was performed using Clampex 10.7 software (Molecular Devices). All recordings were conducted at room temperature, and data were analyzed using Clampfit 10.7 (Molecular Devices) and Prism 10 (GraphPad).
Inside-out patch clamping was performed as described previously
154,174, with minor modifications. Briefly, HEK293T cells were transfected with WT Kir6.2 or its Kir6.2-205–228A mutant, as described above. After 48 h of transfection, the cells were aliquoted and cultured on the coverslips as outlined in the whole-cell patch clamping section. Borosilicate glass electrodes with a tip resistance of 6–9 MΩ were pulled. The pipette (extracellular) solution contained 140 mM KCl, 2.6 mM CaCl
2, 1 mM NaCl, 1.2 mM MgCl
2, and 10 mM HEPES (pH 7.4, by KOH), while the bath (cytosolic) solution contained 130 mM NaCl, 5 mM KCl, 2 mM CaCl
2, 1 mM MgCl
2, 10 mM HEPES (pH 7.4, by NaOH) supplemented with desired concentrations of glucose and NMDG to bring the final osmolarity of 300 mOsm. Before the experiment, the bath solution was freshly supplemented with 2 mM Mg-ATP, 200 μM ADP, and GK protein (purified as described in the “Protein Expression” section, which involved gel filtration using a column balanced and run with the bath solution). The plasma membrane of a cell was suctioned using the micropipette to form a gigaohm seal, after which the pipette was rapidly withdrawn and immersed back into the bath solution to establish the inside-out configuration. Excised patches were continuously perfused with bath solution at a flow rate of 1 mL/min. Single-channel recording was performed using a step protocol at +60 mV. Inward current (i.e., cation flux from the extracellular to the cytosolic side of the membrane) is indicated as downward deflections in all recorded traces. Signals were acquired at 20 kHz and low-pass filtered at 5 kHz. Data were analyzed with Clampfit 10.7 (Molecular Devices) software, with a filter set at 0.5 kHz.
Protein and peptide MS
Protein levels of K
ATP channel subunits expressed in STC-1 cells were quantified using liquid chromatography coupled with data-independent acquisition (DIA) mass spectrometry, as previously described
137,175, with modifications. In brief, peptides from STC-1 cells were generated using a Single-Pot Solid-Phase-enhanced Sample Preparation (SP3) protocol
176, with minor modifications. First, cells were lysed using 1% (v/v) SDS dissolved in distilled water, followed by sonication and centrifugation as described in the “Immunoprecipitation and IB” section. The lysates were then adjusted to a protein concentration of 100 µg/mL using the BCA assay. Some 100 µL of lysates were then incubated with 10 mM dithiothreitol (final concentration, dissolved in water) on a thermomixer at 56 °C, at 1,000 rpm for 30 min, followed by an incubation with 20 mM iodoacetamide (final concentration, dissolved in water) at 25 °C, at 1,000 rpm for another 30 min in the dark. The reaction was terminated by adding 20 mM dithiothreitol at 25 °C, at 1,000 rpm for an additional 10 min. Subsequently, 100 µg of SP3 beads was added to the lysate, followed by addition of an equal volume of ethanol and incubation at 25 °C, at 1,000 rpm for 30 min. The beads were washed three times with 80% (v/v) ethanol and then resuspended in 50 µL of 50 mM ammonium bicarbonate (dissolved in water), followed by overnight incubation with 0.5 µg of trypsin. The resulting peptide supernatants were mixed with 1% (v/v) trifluoroacetic acid and desalted using a C18 StageTips column (cat. MC18TB; Nanomics). Before desalting, the column was washed twice with 50% (v/v) acetonitrile and equilibrated three times with 0.1% (v/v) trifluoroacetic acid. Peptides were loaded onto the column, washed three times with 0.1% (v/v) trifluoroacetic acid, and eluted twice with 80% (v/v) acetonitrile (40 µL each). The peptide eluents were then combined and dried in a vacuum centrifugal concentrator, followed by resuspension in 20 µL of 0.1% (v/v) formic acid. Some 200 ng of peptides were then separated on a Vanquish Neo upgraded ultra-high-performance liquid chromatography (UHPLC) system (Thermo), which was tandemly equipped with a PepMap Neo Trap Cartridge (5 mm × 300 µm, 5 µm, cat. 174500; Thermo Fisher Scientific) and an EASY-Spray HPLC Column (150 µm × 150 mm, 2 µm, cat. ES906; Thermo Fisher Scientific). The mobile phase consisted of 0.1% (v/v) formic acid in LC-MS grade water (designated as mobile phase A) and 0.1% (v/v) formic acid in 80% (v/v) acetonitrile (in LC-MS grade water, mobile phase B). The peptides were eluted over a period of 13 min using gradients that ranged from 4% to 99% mobile phase B. The gradient began with 4% B for 0.3 min at a flow rate of 1.8 µL/min, followed by a linear increase to 8% B over the next 0.7 min, and then a further linear increase to 22.5% B over the subsequent 6.7 min. The gradient then continued with a linear increase to 35% B over the following 3.7 min, during which the flow rate changed to 2.5 µL/min. Next, the gradient increased linearly to 55% B within 0.4 min, followed by another linear increase to 99% B in 0.5 min, and finally held steady at 99% B for an additional 0.7 min.
Data acquisition was performed using an Orbitrap Astral mass spectrometer (Thermo Fisher Scientific) equipped with an EASY-Spray Nanosource in DIA mode. During measurements, the parameters for MS1 were set to a resolution of 240,000, covering a m/z range of 380 to 980, with a normalized automatic gain control (AGC) value of 500% and a maximum injection time (maxIT) of 5 msec. For fragmentation, peptides were processed by the astral detector using normalized higher energy collisional dissociation (HCD) set at 25%, with a normalized AGC target value of 500% and a maxIT of 3 msec. The fixed window number was set to 300, with a 2 m/z isolation window to cover the mass range from 150 to 2000 m/z. The DIA raw files were analyzed using DIA-NN software (v1.8.1)
177, and peptide and protein identifications were conducted using the UniProt database (
Mus musculus) and visualized using IBS software (v 2.0)
178. A confidence level of 99% or higher for peptide-spectrum matches (PSM) was established as the filter criterion, and a false discovery rate (FDR) validation was applied to exclude peptides and proteins with an FDR greater than 1%. For a complete set of parameters used for the identification and quantification of target proteins, please refer to the source data files deposited in the iProX partner repository accompanying the paper.
Quantification and statistical analysis
Statistical analyses were performed using Prism 9 (GraphPad Software). Each group of data was subjected to Kolmogorov-Smirnov test, Anderson-Darling test, D’Agostino-Pearson omnibus test, or Shapiro-Wilk test for normal distribution when applicable. An unpaired two-sided Student’s t-test was used to determine the significance between two groups of normally distributed data. An unpaired two-sided Mann-Whitney test was used to determine the significance between data without a normal distribution. For comparisons between multiple groups with a fixed factor, an ordinary one-way ANOVA was used, followed by Tukey, Sidak or Dunnett as specified in the legends. The assumptions of homogeneity of error variances were tested using F-test (P > 0.05). For comparison between multiple groups with two fixed factors, an ordinary two-way ANOVA was used, followed by Tukey’s, Sidak’s, Kruskal-Wallis, or uncorrected Fisher’s LSD multiple comparisons test as specified in the legends. Geisser-Greenhouse’s correction was used where applicable. The adjusted means and SEM were recorded when the analysis met the above standards. Differences were considered significant when P < 0.05.
DATA AVAILABILITY
The datasets generated and analyzed during the current study are contained within the manuscript and supplemental figures. The MS proteomics data that determined the expression levels of SUR1 and Kir6.2 in STC-1 cells (Supplementary Fig. S6e) have been deposited to the ProteomeXchange Consortium (proteomecentral.proteomexchange) through the iProX partner repository
179,180 with the dataset identifier IPX0011750000. Full immunoblots are provided as a “Full scans” file. Raw data and statistical analysis data for all graphs are in the “Source data” file. 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. All reagents, cell lines, and mouse strains generated in this study are available from the lead contact upon request.
The Author(s) 2026. Published by Higher Education Press. This is an Open Access article distributed under the terms of the CC BY license (https://creativecommons.org/licenses/by/4.0/).
