Dynamic GLUT trafficking at high glucose levels enhances insulin secretion: Dysregulation leads to decreased insulin secretion during type 2 diabetes

Significance

Glucose uptake into pancreatic β cells is a fundamental step for glucose-stimulated insulin secretion, which is facilitated by glucose transporters (GLUTs). Our study uncovered a novel mechanism of GLUT trafficking in β cells, highlighting their dynamic response to elevated glucose. At higher glucose concentrations, GLUTs are recruited readily to the plasma membrane and there is constant recycling and therefore more endocytosis. However, the dynamic regulation of GLUT trafficking in response to high glucose is impaired in human islets with type 2 diabetes, resulting in reduced number of docked insulin granules which indicates lesser insulin secretion. GLUTs contribute to the process of insulin secretion by colocalizing with a small subset of primed granules that release immediately upon stimulation. GLUT dynamics during glucose sensing leads to insulin secretion and maintains glucose homeostasis.

Abstract

Glucose transporters (GLUT1/2) facilitate glucose uptake in pancreatic β cells, triggering insulin secretion. The availability of GLUTs at the plasma membrane (PM) is governed by its expression, delivery to the membrane, and endocytosis. Little is known about the dynamics of GLUT trafficking in response to glucose that triggers glucose-stimulated insulin secretion. In our study, we found that the recruitment of GLUTs to the PM of β cells correlates with increasing glucose concentrations. The recruitment of GLUTs to PM is coupled with endocytosis driven by clathrin. During endocytosis, GLUT actively translocated to clathrin pits, triggering internalization. Disruption of endocytosis with dominant negative dynamin2-K44A or pitstop2 leads to altered GLUT dynamics in response to high glucose levels, resulting in reduced glucose uptake. Augmenting this finding, GLUT distribution and endocytosis in response to glucose are impaired in human islets from type 2 diabetes (T2D) donors, further confirmed by single-cell RNA-seq from β cells of nondiabetic and T2D donors. This leads to limited availability of insulin granules at higher glucose concentrations in T2D islets. Moreover, our data suggest that the glucose-dependent spatial association between GLUT and a subset of insulin granules which are primed via Munc-13.1. In line with this, we see our results highlight the importance of GLUTs delivery to the PM and clathrin-mediated endocytosis, leading to sustained availability of GLUTs for glucose uptake and subsequent insulin secretion in response to glucose. Notably, impaired insulin secretion during T2D might be reversible by modulating GLUT trafficking.

Diabetes is a metabolic disorder characterized by high levels of blood glucose. It has been listed among the top 10 leading causes of death (WHO) and is estimated to increase approximately 10.7% (700 million) of the world’s population by 2045 (1, 2). Insufficient islet hormone secretion is one of the major causes of impaired glucose management during diabetes (3, 4). Glucose is a primary energy source for living organisms, and due to its polar nature, entry into the cells is regulated by specific transport mechanisms (5). Glucose transport is facilitated by a group of membrane proteins called glucose transporters (6). Based on the mechanism of action, they are classified into two types: (I) sodium-coupled glucose transporters, which transport glucose via a secondary active mechanism dependent on sodium, and (II) facilitated transporters (GLUTs), which facilitate glucose transport depending on its concentration gradient (7, 8). There are 14 types of glucose transporters (GLUTs) expressed in mammals, of which GLUT1 and GLUT4 play a crucial role in maintaining glucose homeostasis in humans, whereas in the mouse, GLUT2 and GLUT4 are involved (9–13).

GLUTs in the pancreatic β cells play an important role in insulin secretion by facilitating glucose uptake into the cells (14–16). Upon entry, glucose is fed into glycolysis, followed by the TCA cycle and oxidative phosphorylation to ultimately result in ATP generation. This leads to an increase in cytosolic ATP levels, which results in the closure of the ATP-dependent potassium channels. The resulting depolarization of the cells leads to insulin release through exocytosis. This mechanism is referred to as glucose-stimulated insulin secretion (GSIS) (10, 17–19). The secreted insulin further regulates glucose homeostasis by promoting glucose uptake by recruiting GLUT4 to the PM in peripheral tissues, such as muscle and adipocyte cells. Additionally, as a key mechanism for regulating the membrane protein turnover, recruitment to membrane followed by endocytosis plays a crucial role in maintaining a dynamic equilibrium in GLUT trafficking and glucose uptake (20–22).

As a principal component involved in maintaining glucose homeostasis, alterations in GLUT levels have been closely associated with the etiology of diabetes (23). GLUT1 and GLUT2 are the principal glucose transporters expressed in β cells (24). Previous reports have shown that downregulation in gene expression of GLUT1 and GLUT2 in β cells has defects in the first phase of insulin secretion (25). A subset of docked insulin granules which undergo exocytosis in response to glucose stimulation are referred to as primed insulin granules. Previous studies have shown that insulin granules enriched with Munc-13.1 are preferentially the primed ones (16, 26, 27). Deletion of Munc-13.1 resulted in defects in the phasic insulin secretion (28, 29). Despite these findings, the dynamics of GLUT trafficking in the context of glucose uptake leading to insulin secretion of primed granules in β cells remains unexplored. Investigating the mechanistic role of GLUTs during insulin granule exocytosis in response to glucose would open broad avenues in understanding the pathophysiology of T2D.

In the present study, we evaluate the trafficking of GLUTs in pancreatic β cells in response to different glucose concentrations using live cell imaging. We observe an increase in the rate of recruitment of GLUTs to the PM through exocytosis with increasing glucose concentrations. After its delivery to the membrane, GLUTs generally traffic between the PM and endosomes. Therefore, we investigated the effect of glucose on GLUT internalization. We found that GLUTs internalize when cells are exposed to higher concentrations of glucose and are driven by colocalization with clathrin. During endocytosis, GLUTs are actively moved toward the clathrin pits, where they are subsequently internalized. To further investigate the role of endocytosis on the behavior of GLUTs in response to glucose, we disrupted the GLUT endocytosis using dynamin2-K44A and pitstop2, which resulted in altered GLUT dynamics and impaired glucose uptake in response to high glucose. Furthermore, in a physiological setting such as human or mouse pancreatic β cells, the distribution and colocalization of GLUTs with clathrin were similar to the results we observed in INS1 cells. However, GLUT distribution and endocytosis in response to glucose are impaired in human islets isolated from T2D donors. Part of the reason might be due to reduced expression of GLUT, exhibited in T2D donor cells in single-cell RNA-seq, apart from trafficking defects that compromise GLUT recycling due to reduced expression of clathrin as well. In accordance with the above observations, we also observed limited availability of insulin granules at higher glucose concentrations in T2D islets. This highlights the importance of GLUT trafficking in facilitating GSIS, a process that is disrupted in T2D.

Results

GLUT Recruitment to the Plasma Membrane Increases with Increasing Glucose Concentration.

The expression levels of GLUTs in β cells increase with higher circulating glucose concentrations (30–32). As glucose uptake is the primary step in GSIS, we assessed whether the high glucose concentration affects the trafficking of GLUTs. For this, we exploited total internal reflection fluorescence (TIRF) microscopy to quantify the recruitment of GLUTs to the PM. INS1 cells were transfected with GLUT1-EGFP and exposed to various glucose concentrations for 20 min as mentioned in the figure. Cells were subsequently imaged using TIRF microscopy. GLUT1-EGFP shows a punctate distribution at the PM (Fig. 1A). The number of puncta was quantified using Image J function called find maxima and was normalized to the area of the cell (Fig. 1B) (details are described in the Materials and Methods). Puncta density at the PM showed an increase with an increase in glucose concentration (Fig. 1A). The density of GLUT1 puncta at the PM under no glucose, 3 mM, 5 mM, 7 mM, 10 mM, and 20 mM glucose concentrations were 0.001 μm⁻², 0.054 μm⁻², 0.058 μm⁻², 0.072 μm⁻², 0.084 μm⁻², and 0.089 μm⁻², respectively. Thus, cells treated with different concentrations of glucose showed a significant difference in the density of GLUT1 puncta at the PM compared to no glucose condition. Similar to GLUT1 response to glucose, we observed an increase in GLUT2-EGFP puncta density at the PM with increasing glucose concentrations (Fig. 1 C and D), with the density under no glucose, 3 mM, 5 mM, 7 mM, 10 mM, and 20 mM glucose concentrations being 0.003 μm⁻², 0.013 μm⁻², 0.020 μm⁻², 0.035 μm⁻², 0.058 μm⁻², and 0.062 μm⁻², respectively. To exclude the possibility of osmotic effects, we treated cells with similar concentrations of mannitol, as suggested in the previous literature (33–35). The density of GLUTs remained the same with different concentrations of mannitol (SI Appendix, Fig. S1 A–H).

Fig. 1.

Density and colocalization of GLUTs with endocytic protein clathrin under various concentrations of glucose in INS1 cells. (A) GLUT1 localization at the plasma membrane in INS1 cells with glucose concentrations as listed in the figure. (B) Density of GLUT1 at the plasma membrane under the conditions A. Data represented as mean ± SEM (n = 47 cells– No glucose; n = 34 cells– 3 mM glucose concentration; n = 43 cells– 5 mM glucose concentration; n = 30 cells– 7 mM glucose concentration; n = 30 cells– 10 mM glucose concentration; n = 30 cells– 20 mM glucose concentration). 2 to 4 independent experiments for each condition. ***P < 0.001. (C) Same as A for GLUT2. (D) Density of GLUT2 at the plasma membrane for conditions mentioned in C. Data represented as mean ± SEM (n = 58 cells– No glucose; n = 54 cells – 3 mM glucose concentration; n = 39 cells – 5 mM glucose concentration; n = 34 cells – 7 mM glucose concentration; n = 31 cells – 10 mM glucose concentration; n = 21 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. ***P < 0.001. (E) INS1 cells expressing the GLUT1 in green channel and clathrin in red channel and corresponding overlay images at various concentrations of glucose as mentioned in the figure. (F) Percentage of colocalization of GLUT1 and clathrin for conditions mentioned in E. Data represented as mean ± SEM (n = 15 cells – No glucose; n = 15 cells – 3 mM glucose concentration; n = 15 cells – 20 mM glucose concentration) two independent experiments for each condition. ***P < 0.001. (G) Same as E for GLUT2. (H) Percent of colocalization of GLUT2 and clathrin for conditions mentioned in G. Data represented as mean ± SEM (n = 15 cells – No glucose; n = 15 cells – 3 mM glucose concentration; n = 15 cells – 20 mM glucose concentration) two independent experiments for each condition. ***P < 0.001, **P < 0.01. For all images, (Scale bar, 1 μm.) (I) Schematic representation of GLUT endocytosis under lower glucose concentration. (J) Schematic representation of GLUT endocytosis under higher glucose concentration.

Further, we observed an overall increase in the fluorescence intensity of GLUT puncta apart from the increase in the number of clusters after the addition of glucose, suggesting that GLUT is trafficking to the PM upon glucose treatment (SI Appendix, Fig. S1 I–P). The above effects were limited to D-glucose since the addition of L-glucose showed no recruitment of GLUTs to the PM (SI Appendix, Fig. S2 A–D). Similarly, 2-deoxy-D-glucose, a nonmetabolized form of glucose, did not induce GLUT recruitment to the PM (SI Appendix, Fig. S3 A–D). These findings suggest that GLUT exocytosis increases with increasing glucose concentration, which is specifically dependent on D-glucose and its metabolism.

GLUTs Colocalize with Clathrin at the Plasma Membrane in a Glucose-Dependent Manner.

Our previous results showed that high glucose concentrations increase the recruitment of GLUTs to the PM, which could result from the effect of glucose on GLUT trafficking driven by exocytosis to the membrane. GLUTs are internalized subsequently by clathrin-mediated endocytosis (CME) (21). To analyze the role of CME in regulating GLUT density on the PM, INS1 cells were cotransfected with clathrin-mCherry and GLUT1-EGFP and imaged using TIRF microscopy under different glucose concentrations (Fig. 1E). The percentage of colocalization of GLUT1-EGFP with clathrin-mCherry was analyzed using metamorph software (details are mentioned in the Materials and Methods). We observed that more than 60% of GLUT1-EGFP puncta colocalizes with clathrin-mCherry puncta (Fig. 1E). The percentage of colocalization rose from 20 (no glucose condition) to 44.8 (3 mM glucose) and to 66.5 (20 mM glucose) (Fig. 1F). These results revealed that the punctate distribution of GLUT1 is colocalized with clathrin as glucose concentration rises. Similar experiments were performed by transfecting cells with GLUT2-EGFP as well. GLUT2 colocalizes with clathrin in a similar manner as GLUT1 (Fig. 1G), with the percentage of colocalization increasing from 10.39 (no glucose) to 47.97 (3 mM glucose) and 63 (20 mM glucose) (Fig. 1H). Together, these data indicate that CME regulates GLUT trafficking and homeostasis at the PM in a glucose-dependent manner.

GLUT Clusters Get Captured at Clathrin Pits Before Endocytosis.

To understand the process of GLUT endocytosis happening at the PM, we assessed the dynamics of GLUT movement in real time. For this, GLUT and clathrin plasmids were cotransfected to INS1 cells. Time-lapse imaging was performed at a frame rate of 1 frame/10 seconds. As described in the Materials and Methods, we marked GLUT and clathrin puncta (Fig. 2 A–F) and the movement of GLUT and clathrin at the PM was assessed using image J Pugin MTrack J. Our results showed that GLUTs exhibited higher mobility at the PM compared to clathrin, which demonstrated significantly reduced movement (Fig. 2 D and E). GLUT migrates toward clathrin-coated regions as it undergoes endocytosis, which is marked by the disappearance of both GLUT and clathrin puncta in the TIRF field (Fig. 2F). Average and total distances traveled by GLUT (Average distance traveled - 0.49 μm⁻² for N = 1, 0.38 μm⁻² for N = 5 and total distance traveled- 1.73 μm⁻² N = 1, 1.2 μm⁻² for N = 5), are significantly greater than those of clathrin (Average distance traveled - 0.14 μm⁻² for N = 1, 0.14 μm⁻² for N = 5 and total distance traveled- 0.26 μm⁻² N = 1, 0.35 μm⁻² for N = 5) (Fig. 2 G–J). This suggests that GLUT is more dynamic in its movement across the PM, while clathrin remains more localized at endocytic sites. As illustrated in (Fig. 2 K and L), once GLUT reached clathrin sites, a sudden disappearance of both GLUT and clathrin puncta was observed, indicating that GLUT undergoes endocytosis. This was further supported by a rapid decline in the fluorescence intensity of both GLUT and clathrin puncta during the endocytic process (Fig. 2 M and N). Overall, the data suggest that GLUT puncta localize to clathrin sites before undergoing endocytosis after the addition of high glucose.

Fig. 2.

GLUT puncta recruited to the sites of clathrin puncta during endocytosis. (A) TIRF representative image for INS1 cells expressing GLUT. (B) TIRF representative image for INS1 cells expressing clathrin. (C) TIRF representative image for overlay images of INS1 cell expressing both clathrin and GLUT. (D) Trajectories of individual GLUT molecules movement represented with a straight line at various time points. (E) Trajectories of individual clathrin molecules movement represented with circles at various time points. (F) Overlay of GLUT and clathrin trajectories with a straight line (GLUT) and circles (clathrin). (G) Bar graph representing the average distance traveled by GLUT and clathrin in a single cell. Data represented as mean ± SEM (n = 10 Puncta, N = 1 cell). ***P < 0.001. (H) Bar graph representing the total distance traveled by GLUT and clathrin in a single cell. Data represented as mean ± SEM (n = 10 puncta, N = 1 cell). ***P < 0.001. (I) Bar graph representing the average distance traveled by GLUT and clathrin. Data represented as mean ± SEM (n = 58 puncta, N = 5 cells) 2 to 4 independent experiments. ***P < 0.001. (J) Bar graph representing the total distance traveled by GLUT and clathrin. Data represented as mean ± SEM (n = 58 puncta, N = 5 cells) 2 to 4 independent experiments. ***P < 0.001. (K) Time lapse strips showing the GLUT movement toward the clathrin further undergoing endocytosis. (Scale bar, 0.2 μm.) (L) Time lapse strips showing the clathrin localization during GLUT endocytosis. (Scale bar, 0.2 μm.) (M) Scatter plot showing the GLUT puncta intensity during endocytosis at various time points. (N) Scatter plot showing the clathrin puncta intensity during endocytosis at various time points. For all images, (Scale bar, 1 µm.) unless it is specified.

Disruption of CME Alters the Dynamics of GLUTs at the Plasma Membrane in Response to Glucose.

Given that GLUT display and organization at the PM play a crucial role in glucose uptake and subsequent insulin secretion, we investigated how alterations in clathrin-mediated GLUT endocytosis affect GLUT dynamics and behavior at the PM. To examine this, we used a dominant-negative strategy and an inhibitor-based approach to disrupt CME. The dominant-negative dynamin2-K44A inhibits the fission of endocytic buds during CME (36). In contrast, pitstop2 inhibits the binding of BAR domain proteins to clathrin, which is essential for membrane curvature generation and subsequent recruitment of dynamin during CME (37). Cells were transfected with GLUT1-EGFP alone or cotransfected with either dynamin2-K44A construct/wild type (WT) dynamin2-mCherry and GLUT1-EGFP. GLUT1 puncta density at the PM was assessed under no glucose or at high glucose conditions (Fig. 3A). Our results show that GLUT1-EGFP and WT dynamin2 with GLUT1-EGFP (control) cells exhibit less GLUT1 density (0.064 μm⁻² -GLUT1-EGFP and 0.058 μm⁻² -WT dynamin2 with GLUT1-EGFP) during no glucose condition. However, dynamin2-K44A-mCherry transfected cells showed a significantly higher density of GLUT1-EGFP puncta during no glucose (0.275 μm⁻²) irrespective of cells being present in low or high (20 mM) glucose (0.304 μm⁻²). In contrast, control cells exhibited a significant increase in GLUT1 density after adding 20 mM glucose (0.304 μm⁻²-GLUT1-EGFP and 0.327 μm⁻²-WT dynamin2 with GLUT1-EGFP) (Fig. 3B). When similar comparisons were done in GLUT2, we saw a higher density of GLUT2 puncta in dynamin2-K44A in a no glucose state (0.229 μm⁻²) compared to control cells (0.034 μm⁻²-GLUT2-EGFP and 0.020 μm⁻²-WT dynamin2 with GLUT2-EGFP) under the same conditions. In cells transfected with dynamin2-K44A-mCherry, no significant increase was observed at 20 mM glucose (0.223 μm⁻²). Control cells, on the other hand, demonstrated a significant increase in GLUT2 puncta density at 20 mM glucose (0.261 μm⁻²-GLUT2-EGFP and 0237 μm⁻²-WT dynamin2 with GLUT2-EGFP) (Fig. 3 C and D).

Fig. 3.

GLUT Response to glucose is altered under clathrin-mediated endocytosis disruption. (A) GLUT1 localization at the plasma membrane in INS1 cells expressing WT dynamin2 and dynamin2-k44A under glucose concentrations as listed in the figure. The images from dynamin2 expressing cells are shown in SI Appendix, Fig. S5 A and B. (B) Density of GLUT1 at the plasma membrane under the conditions mentioned in A. Data represented as mean ± SEM (n = 18 cells – No glucose and n = 19 cells – 20 mM glucose concentration in GLUT1 control; n = 16 cells – No glucose and n = 16 cells – 20 mM glucose concentration in WT dynamin2; n = 15 cells – No glucose; n = 15 cells – 20 mM glucose concentration in dynamin2-K44A). 2 to 4 independent experiments for each condition. P < 0.001, P > 0.05. (C) Same as A for GLUT2. The images from dynamin2 expressing cells are shown in SI Appendix, Fig. S5 C and D. (D) Density of GLUT2 at the plasma membrane under the conditions mentioned in C. Data represented as mean ± SEM (n = 15 cells – No glucose and n = 15 cells – 20 mM glucose concentration in GLUT2 control; n = 15 cells – No glucose and n = 15 cells – 20 mM glucose concentration in WT dynamin2; n = 23 cells – No glucose; n = 25 cells – 20 mM glucose concentration in dynamin2-K44A). 2 to 4 independent experiments for each condition. P < 0.001, P > 0.05. (E) GLUT1 localization at the plasma membrane in INS1 cells treated with pitstop2 under glucose concentrations as listed in the figure. (F) Density of GLUT1 at the plasma membrane under the conditions represented in E. Data represented as mean ± SEM (n = 15 cells – No glucose and n = 16 cells – 20 mM glucose concentration in no serum control; n = 17 cells – No glucose and n = 15 cells – 20 mM glucose concentration in DMSO control; n = 17 cells – No glucose; n = 16 cells – 20 mM glucose concentration in pitstop2). 2 to 4 independent experiments for each condition. P < 0.001, P > 0.05. (G) Same as E for GLUT2. (H) Density of GLUT2 at the plasma membrane under the conditions represented in G. Data represented as mean ± SEM (n = 17 cells – No glucose and n = 16 cells – 20 mM glucose concentration in no serum control; n = 15 cells – No glucose and n = 16 cells – 20 mM glucose concentration in DMSO control; n = 16 cells -No glucose; n = 18 cells – 20 mM glucose concentration in pitstop2). 2 to 4 independent experiments for each condition. P < 0.001, P > 0.05. (I) INS1 cells expressing GLUT1-eGFP under various glucose concentrations at time points specified in the figure. (J) GLUT1 density at various time points as specified in I, for different glucose concentrations. Data represented as mean ± SEM (n = 6 cells, - No glucose; n = 6 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. (K) Same as I for GLUT1 and dynamin2 WT expressing cells. The images from dynamin2-expressing cells are shown in SI Appendix, Fig. S5A. (L) GLUT1 density at various time points for different glucose concentrations as mentioned in K. Data represented as mean ± SEM (n = 6 cells – No glucose; n = 6 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. (M) Same as I for GLUT1 and dynamin2-K44A cells. The images from dynamin2-K44A expressing cells are shown in SI Appendix, Fig. S5B. (N) GLUT1 density at various time points for different glucose concentrations as mentioned in N. Data represented as mean ± SEM (n = 9 cells – No glucose; n = 10 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. (O) Same as I for GLUT2. (P) GLUT2 density at various time points for different glucose concentrations as specified in O. Data represented as mean ± SEM (n = 8 cells – No glucose; n = 6 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. (Q) Same as O for GLUT2 and dynamin2 WT expressing cells. The images from dynamin2 expressing cells are shown in SI Appendix, Fig. S5C. (R) GLUT2 density at various time points for different glucose concentrations as specified in Q. Data represented as mean ± SEM (n = 8 cells – No glucose; n = 6 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. (S) Same as O for GLUT2 and dynamin2-K44A cells. The images from dynamin2-K44A expressing cells are shown in SI Appendix, Fig. S5D. (T) GLUT2 density at various time points for different glucose concentrations as specified in S. Data represented as mean ± SEM (n = 10 cells – No glucose; n = 10 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. For all images, (Scale bar, 1 μm.) (U) Bar graph representing glucose uptake by INS1 cells in treated (pitstop2) and untreated conditions, cultured at 3 mM and 20 mM glucose media. Data represented as mean ± SEM (n = 3 trials). **P < 0.01. (V) Schematic representation of disruption of GLUT endocytosis affects the glucose uptake.

We validated the above results by using the endocytosis inhibitor pitstop2. Cells were treated with DMSO (vehicle control) and pitstop2. The density of GLUT1 puncta under no glucose and 20 mM glucose in DMSO, pitstop2-treated cells (0.047 μm⁻², 0.266 μm⁻² and 0.237 μm⁻², 0.249 μm⁻²) respectively (Fig. 3 E and F). Consistent with observations in dynamin2-K44A experiments, pitstop2-treated cells showed a significantly higher density of GLUT1 puncta under no glucose compared to vehicle control, and no significant increase was seen with 20 mM glucose. In line with these results, GLUT2 transfected cells demonstrated similar outcomes (Fig. 3G). The density of GLUT2 puncta during no glucose and 20 mM glucose for DMSO (0.047 μm⁻²-no glucose; 0.134 μm⁻²-20 mM glucose) respectively. Such an increase was not seen in pitstop2-treated cells, and the density remained approximately the same (0.165 μm⁻²-no glucose and 0.151 μm⁻²-20 mM glucose) (Fig. 3H). Since pitstop2 exerts its effects under serum-free conditions (37) we evaluated whether the lack of serum introduces additional effects. The GLUT1 (0.039 μm⁻²-no glucose and 0.237 μm⁻²-20 mM glucose) and GLUT2 (no glucose 0.042 μm⁻² compared to 0.182 μm⁻² at 20 mM glucose) show similar effects in 20 mM glucose, like in control (Fig. 3 F and H). We, therefore, did not see any effect of serum on the density of GLUT1 or GLUT2. Overall, these observations suggest that endocytosis regulates GLUT behavior in response to glucose at the PM. Any alterations in the endocytosis lead to changes in GLUT dynamics and, consequently glucose uptake.

Trafficking of GLUTs to the Plasma Membrane Is Glucose-Dependent.

It has been demonstrated in earlier reports that insulin-stimulated GLUT trafficking is essential for peripheral tissues to absorb glucose, and impaired GLUT trafficking is a characteristic feature of diabetes in these tissues (38, 39). It would be interesting to study the kinetics of GLUT trafficking to PM in tissues that secrete insulin, where glucose uptake by the pancreatic β cell is the first step in triggering GSIS. To study the dynamics of GLUT trafficking in pancreatic β cells, we transfected INS1 cells with GLUT1-EGFP/ GLUT2-EGFP. After 24 h of transfection, cells were treated with 20 mM glucose, which would increase the GLUT density at the PM in the pancreatic β cells, as observed in the previous results. The recruitment of GLUTs to the PM was recorded at various time points after the addition of 20 mM glucose (Fig. 3 I and O). The density of punctate structures at different time points was calculated using metamorph software and normalized to the area of the cell. Normalized density of GLUTs was calculated in single cells, and an average of many such cells was calculated. (Fig. 3 J and P). Under no glucose condition, there was no significant GLUT1/ GLUT2 puncta recruitment to PM over time. Interestingly, we observe an almost fourfold increase in the trafficking of GLUT1/ GLUT2 s to PM upon glucose stimulation within 6-8 mins after treatment (Fig. 3 J and P). Overall, the data suggest that enhanced trafficking of GLUTs to the PM upon glucose stimulation thus promoting glucose uptake, leading to insulin secretion.

Further, we analyzed the kinetics of GLUT trafficking to the PM by disrupting the endocytosis using dynamin2-K44A. Cells expressing WT dynamin2 and GLUT1-EGFP/ GLUT2-EGFP showed a similar GLUT response as in the case of control, with GLUT trafficking to the PM saturated within 6 to 8 min (Fig. 3 K, L, Q, and R). In contrast, cells expressing dynamin2-K44A, GLUT1 showed no response to glucose, and the density of GLUT1 puncta at the PM is significantly higher in the no glucose condition compared to control cells (Fig. 3 M and N). Likewise, experiments are performed with GLUT2 and dynamin2-K44A, our results are consistent with GLUT1 (Fig. 3 S and T). Thus, the above results suggest endocytosis plays a role in the dynamics of GLUTs trafficking in response to glucose, which is essential for insulin secretion.

GLUT Endocytosis Increases Glucose Uptake Into Pancreatic β Cells.

To gain insights into the uptake of glucose in a β cells population, we performed a bulk assay such as GOD-POD (Glucose oxidase and peroxidase) assay, that measured the uptake of glucose. This assay relies on glucose oxidase, which catalyzes the conversion of glucose to gluconic acid and produces hydrogen peroxide as a byproduct. Another enzyme, peroxidase, catalyzes the reaction of H₂O₂‚ with chromogenic substrates forming a colored complex which has an absorption maxima at 505 nm (40). Cells cultured in 20 mM glucose media showed a glucose uptake of 7.1 mM (Fig. 3U). To assess whether GLUT recycling, where the GLUT transporter is recruited to the membrane and endocytosed; is important for glucose uptake, we used an endocytosis inhibitor pitstop2. Pitstop2-treated cells showed a reduced uptake of 2.6 mM. Our results showed that there was a significant reduction in the glucose uptake in pitstop2-treated cells, where the GLUT endocytosis is inhibited at 20 mM glucose concentration compared to the control. However, at lower glucose concentration, no noticeable difference in glucose uptake was observed between pitstop2-treated and control cells. Overall, these results suggest a crucial role of GLUT endocytosis on glucose uptake at higher glucose concentration. The results are true both at the level of a single cell or in bulk assays.

Exocytosis of GLUTs Impaired in Human Type 2 Diabetic Pancreatic β Cells.

To investigate the GLUTs behavior within the physiological condition, we tested the glucose-dependent effect on GLUTs in isolated mouse pancreatic β cells as well. The Islets were treated with various glucose concentrations, later fixed and immuno-stained for GLUT1 and GLUT2, and imaged using TIRF, (SI Appendix, Fig. S6 A and C). The density of GLUT1 and GLUT2 puncta under no glucose and high glucose conditions were 0.01 μm⁻², 0.04 μm⁻² and 0.02 μm⁻², 0.014 μm⁻², respectively (SI Appendix, Fig. S6 B and D). We observe a significant difference in the density of GLUT2 at higher glucose concentration when compared with no glucose condition but not in the case of GLUT1. This might be because GLUT2 is the principal glucose transporter that responds to glucose in the mouse pancreatic β cells.

In order to comprehend the changes that occur during human T2D, it is important to explore the trafficking of GLUTs in human pancreatic β cells. GLUT1 is the principal glucose transporter in the human pancreatic β cells (11, 12). Therefore, we checked the recruitment of GLUT1 to the PM in human pancreatic β cells isolated from both nondiabetic (ND) and clinically diagnosed T2D donors. We treated human islets with low and high glucose concentrations, fixed, and immuno-stained the islets with the GLUT1 antibody. Then TIRF microscopy was used to study the trafficking of GLUTs to the PM. We observe the punctate structures in cells stained for GLUT1 near the PM (Fig. 4 A and C). The number of puncta was estimated using the find maxima plugin in the Image J software and normalized to the area of the cell, as mentioned before. The density of GLUT1 puncta in ND donors and T2D donors during low and high glucose concentrations were 0.024 μm⁻², 0.068 μm⁻² and 0.02 μm⁻², 0.03 μm⁻², respectively (Fig. 4 B and D). In ND β cells, there was a significant increase in the density of GLUT1 puncta at the PM under a high concentration of glucose compared to no glucose condition. This is similar to previous results shown in the INS1 cells and mouse pancreatic β cells (Fig. 1 B and D and SI Appendix, Fig. S6A). Whereas in the T2D condition, no significant difference in the density of GLUT1 puncta was observed under high glucose concentration compared to no glucose condition. Further, we observed a drastic reduction in the overall density of GLUT puncta in T2D pancreatic β cells compared to ND human pancreatic β cells under high glucose concentrations. Our data revealed that GLUT1 recruitment to the PM increases with increasing glucose in ND pancreatic β cells whereas, in T2D pancreatic β cells, recruitment of GLUT1 is disrupted.

Fig. 4.

Clathrin mediated endocytosis of GLUT1 represented by its colocalization with clathrin is dysregulated in human type 2 diabetic pancreatic β cells, leading to reduced insulin granules at the plasma membrane. (A) GLUT1 localization at the plasma membrane in nondiabetic human pancreatic β cells under various glucose concentrations as listed in the figure. (B) Density of GLUT1 puncta at the plasma membrane for conditions as in A. Data represented as mean ± SEM (n = 20 cells – No glucose; n = 20 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. **P < 0.01. (C) GLUT1 localization at the plasma membrane in type 2 diabetic human pancreatic β cells under various glucose concentrations as listed in the figure. (D) Density of GLUT1 puncta at the plasma membrane for concentrations as in C. Data represented as mean ± SEM (n = 20 cells – No glucose; n = 20 cells – 20 mM glucose concentration) 2 to 4 independent experiments for each condition. P > 0.05. (E) Pancreatic β cells from nondiabetic human samples immunostained with GLUT1 antibody in red channel, clathrin antibody in the green channel, and corresponding merged images in overlay during no glucose and high glucose states. (F) Percent of colocalization of GLUT1 and clathrin for conditions mentioned in E. Data represented as mean ± SEM (n = 20 cells – No glucose; n = 20 cells – 20 mM glucose concentration) two independent experiments for each condition. ***P < 0.001. (G) Pancreatic β cells from type 2 diabetic human sample immunostained with GLUT1 antibody in red channel, clathrin antibody in the green channel and corresponding merged images in overlay during no glucose and high glucose states. (H) Percent of colocalization of GLUT1 and clathrin for conditions mentioned in G. Data represented as mean ± SEM (n = 20 cells – No glucose; n = 20 cells – 20 mM glucose concentration) two independent experiments for each condition. P > 0.05. (I) Gene expression analysis for genes SCL2A1, SCL2A2, PDX1, PRKAA1, CTLA, CTLB, and CTLC from nondiabetic (n = 3) and type 2 diabetic (n = 3) human pancreatic islets samples. (J) Pancreatic β cells from nondiabetic human samples transfected with insulin granule marker (NPY- mCherry) during low and high glucose conditions. (K) Density of insulin granules in at the plasma membrane for conditions mentioned in J. Data represented as mean ± SEM (n = 28 cells – 3 mM glucose concentration, n = 31 cells – 10 mM glucose concentration), two independent experiments for each condition. ***P < 0.001. (L) Pancreatic β cells from type 2 diabetic human samples transfected with insulin granule marker (NPY- mCherry) during low and high glucose conditions. (M) Density of insulin granules in at the plasma membrane for conditions mentioned in L. Data represented as mean ± SEM (n = 28 cells – 3 mM glucose concentration, n = 31 cells – 10 mM glucose concentration, type 2 diabetic condition) two independent experiments for each condition. *P < 0.05. (N) INS1 cells expressing the GLUT1 in green channel and NPY in red channel and corresponding overlay images at various concentrations of glucose as mentioned in the figure. (O) Percentage of colocalization of GLUT1 and NPY for conditions mentioned in N. Data represented as mean ± SEM (n = 15 cells – No glucose; n = 16 cells – 3 mM glucose concentration; n = 15 cells – 20 mM glucose concentration) two independent experiments for each condition. P > 0.05. (P) Same as N for GLUT2. (Q) Percentage of colocalization of GLUT2 and NPY for conditions mentioned in P. Data represented as mean ± SEM (n = 15 cells – No glucose; n = 15 cells – 3 mM glucose concentration; n = 15 cells – 20 mM glucose concentration) two independent experiments for each condition. P > 0.05. (R) INS1 cells expressing the GLUT1 in red channel and Munc-13.1 in green channel and corresponding overlay images at various concentrations of glucose as mentioned in the figure. (S) Percentage of colocalization of GLUT1 and Munc-13.1 for conditions mentioned in R. Data represented as mean ± SEM (n = 15 cells – No glucose; n = 15 cells – 3 mM glucose concentration; n = 15 cells – 20 mM glucose concentration) two independent experiments for each condition. ***P < 0.001, **P < 0.01. (T) Same as R for GLUT2. (U) Percentage of colocalization of GLUT2 and Munc-13.1 for conditions mentioned in T. Data represented as mean ± SEM (n = 15 cells – No glucose; n = 15 cells – 3 mM glucose concentration; n = 15 cells – 20 mM glucose concentration) two independent experiments for each condition. ***P < 0.001, **P < 0.01. For all images, (Scale bar, 1 µm.)

Glucose-Dependent Clathrin-Mediated Endocytosis Is Disrupted in Type 2 Diabetes.

To elucidate the endocytosis of GLUT at the physiological aspect, we checked the colocalization of GLUT with clathrin in isolated mouse pancreatic β cells. Mouse islets were treated under various glucose concentrations as mentioned in the figure (SI Appendix, Fig. S6 E and G). Subsequently, the islets were immuno-stained for GLUT1 and clathrin/GLUT2 and clathrin and imaged using TIRF microscopy. An overlap of GLUT1 or GLUT2 puncta with clathrin was observed, similar to results from INS1 cells. The percentage of colocalization of clathrin with GLUT1 and GLUT2 under no glucose and 20 mM glucose were 19.5, 38.6, and 16.3, 67.6 respectively (SI Appendix, Fig. S6 F and H). Consistent with our results in the INS1 cells, GLUT2 showed a significant increase in colocalization with clathrin at high glucose concentration but not GLUT1. The above results suggest that endocytosis of GLUT2 in response to glucose might be indispensable for GLUT2 dynamics near the PM, which helps in glucose uptake.

To better understand the alterations taking place during T2D, we examined the glucose-dependent colocalization of GLUT1 and clathrin in human pancreatic β cells from ND and T2D donors. To assess this, human islets were treated with different concentrations of glucose as mentioned in the figure (Fig. 4 E and G). The islets were immuno-stained for GLUT1 and clathrin. TIRF imaging was performed to check the colocalization of GLUT1 and clathrin at the PM under high and low glucose concentrations. We observed GLUT1 and clathrin as punctate structures at PM. Colocalization of GLUT1 with clathrin was analyzed using metamorph software (details are mentioned in the Materials and Methods section). The percentage of GLUT1 colocalization with clathrin in ND donors and T2D donors during low and high glucose concentrations were 7.9, 20.5 and 4.1, 6.6, respectively (Fig. 4 F and H). In ND islets, there is a considerable increase in the colocalization of GLUT1 with clathrin at the PM under high glucose concentration when compared to the no glucose condition. This is consistent with the results observed in the case of INS1 and mouse pancreatic islets (Fig. 1 F and H and SI Appendix, Fig. S6F). In the T2D condition, no measurable increase in colocalization of GLUT1 with clathrin was seen when subjected to high glucose concentration compared to no glucose condition. We also observed very little colocalization of GLUT1 with clathrin in T2D compared to ND human pancreatic β cells under high glucose concentrations. These results indicated that GLUT endocytosis is reduced during the T2D condition, which is important for GLUT dynamics in β cells near the PM.

The above results led us to study the gene expression of GLUTs, associated transcription, metabolic factors, and endocytosis-related genes in the pancreatic β cells of ND and T2D donors. Using the Seurat R package, we analyzed the expression of SCL2A1, SCL2A2, PDX1, CLTA, CLTB, CLTC, and PRKKA1 specific to β cells from scRNA-seq data of the pancreatic islets of T2D and ND donors. To perform gene expression analysis for our gene of interest, specifically in β cells, we created clusters of each cell type of the pancreatic islet using cell-specific markers (SI Appendix, Fig. S7 A and B). Further, we performed gene expression analysis to check the change in expression levels between T2D and ND donors. SLC2A1, PDX1, and PRKAA1 were observed to be downregulated in β cells of T2D donors compared to ND donors, which correlated with the endocytosis genes, i.e., clathrin light chain and heavy chain (CLTA, CTLB, and CTLC) which were downregulated in T2D (Fig. 4I). Q-PCR results further confirm lower expression of GLUT1 in T2D β cells (SI Appendix, Fig. S7C). These data demonstrated that in addition to GLUT trafficking and endocytosis, expression of these related genes is also disrupted in T2D.

Reduced Density of Docked Insulin Granules During Type 2 Diabetes.

Given that the density of docked insulin granules recruited to the β cell membrane is glucose-dependent, which is significantly reduced in T2D (41–43), we evaluated whether this effect is mediated via GLUTs. In order to accomplish this, we performed adenovirus-mediated transduction to the human pancreatic β cells from ND and T2D donors with NPY-mCherry, which labels the insulin granules. TIRF imaging was performed to study the density of insulin granules at PM in the presence of 3 mM and 10 mM glucose concentrations. Under high glucose treatment, cells were treated with diazoxide to prevent the exocytosis of insulin granules (41). We observed NPY-mCherry positive puncta that label the insulin granules in human pancreatic β cells (Fig. 4 J and L). Insulin granule density was estimated using the find maxima function of ImageJ and normalized to the area of the cell. The density of insulin granules in pancreatic β cells of ND donors and T2D donors during low and high glucose concentrations were 0.56 μm⁻², 0.75 μm⁻² and 0.25 μm⁻², 0.32 μm⁻², respectively (Fig. 4 K and M).

In the ND condition, we observed a significant increase in the density of NPY-mCherry puncta under high glucose concentration compared to the low glucose concentration. On the contrary, in T2D β cells, glucose had an effect on the density of NPY-mCherry puncta at high glucose concentration. Nevertheless, the effect is considerably reduced compared to the ND condition. The above results suggest that the glucose-dependent effect on insulin granules correlates well with the recruitment and endocytosis of GLUT transporters under high and low glucose conditions. The parallel response pattern observed with GLUT trafficking and insulin granule density in response to glucose may serve as a key regulatory mechanism.

To further investigate this mechanistic link between insulin granules and GLUTs in a spatial context, we analyzed their colocalization at the PM using TIRF microscopy. Cells were cotransfected with GLUT1-EGFP/GLUT2-EGFP and NPY-mCherry and imaged after 24 h of transfection. Cells were pretreated with various glucose concentrations as mentioned in the figure (Fig. 4 N and P). Surprisingly, a subset of insulin granules colocalized with GLUTs. The level of colocalization did not change with increasing glucose concentrations (Fig. 4 O and Q). Since only a fraction of the docked pool of granules colocalized with GLUTs, we evaluated the functional pool of insulin granules at the plasma membrane. One such pool is known as primed granules, characterized by the presence of Munc-13.1 (27, 44–47), which undergoes rapid fusion upon glucose stimulation. To determine whether GLUT specifically associates with the primed insulin granules, we assessed the colocalization with Munc-13.1. Cells were cotransfected with GLUT1-mRuby2/GLUT2-mCherry and Munc-13.1-EGFP and treated with different glucose concentrations as listed in the figure (Fig. 4 R and T). Our results showed a glucose-dependent increase in colocalization between GLUT and Munc-13.1-EGFP (Fig. 4 S and U). This suggests that the GLUT trafficking promotes the priming of insulin granules in response to glucose. Altogether, our data infer that GLUT trafficking is crucial for maintaining the density of insulin granules at the PM in a glucose-dependent manner, thereby facilitating GSIS.

Discussion

Membrane proteins and channels often undergo activity-dependent and signaling-dependent trafficking, which is essential for regulating their function (48). GLUT is a membrane protein; it traffics between the endosomes and the PM in response to stimuli. GLUT trafficking mediated glucose uptake is well reported in peripheral tissues such as adipose and muscle tissues (20, 49). In these tissues exocytosis of facilitated glucose transporter 4 (GLUT4) from GLUT4 storage vesicles to the PM is stimulated by insulin (21, 50, 51). To achieve the dynamic equilibrium of GLUT4 at PM, it undergoes endocytosis based on the presence or absence of insulin (21). In peripheral tissues, GLUT undergoes endocytosis via clathrin-mediated (CME) and dynamin-dependent fashion (52, 53). The metabolic regulation of GLUT1 trafficking via CME is disrupted in glucose deprived condition in cancer (54). On the contrary, the trafficking of GLUTs in response to metabolic conditions in the insulin secreting pancreatic β cells remains elusive. β cells predominantly express GLUT1 and GLUT2, while the expression levels of GLUT4 in the β cells are relatively very low (13, 55). In β cells GLUTs are primarily localized at the plasma membrane (56). In this study, we examined the trafficking of facilitated glucose transporters (GLUT1 and GLUT2) in the endocrine pancreatic β cells that mediate the uptake of glucose into these cells resulting in insulin secretion. We further examined how this mechanism is affected in T2D human islets. This will pave the way to understanding the dynamics of GLUT trafficking in response to glucose and its role in GSIS and how these responses are disrupted during T2D.

Our study evaluated the surface distribution of GLUTs at the PM in response to changes in the glucose concentration. To investigate this aspect, we used the Total Internal Reflection Fluorescence Microscopy (TIRF-M) technique because of its ability to record the surface dynamics of GLUTs (41, 57, 58). In the TIRF field GLUT appeared as discrete puncta, with each puncta representing a cluster of approximately 10-25 GLUT molecules (59). Our results show that the GLUT exocytosis to the PM increases at higher glucose concentrations compared to the no glucose condition and this response is D-glucose dependent. In order to investigate the effect of glucose on GLUT endocytosis, we examined the colocalization of clathrin and GLUT under various concentrations of glucose. Interestingly, we observed that the colocalization of GLUT with clathrin is more at high glucose concentrations when compared to low glucose concentrations (Fig. 1 I and J). Previously, it was shown that in nonsecretory cells, clathrin primarily mediates the endocytosis of GLUT4 from preexisting clusters at the PM (21) In contrast to this, our study in the pancreatic β cells (secretory cells) shows that during endocytosis, GLUT at the PM travels toward the clathrin pits and subsequently undergoes internalization.

Further, we evaluated whether the GLUT distribution at the PM is affected due to inhibition of CME using (I) pitstop2, which has the ability to inhibit the membrane curvature during endocytosis (37) or (II) dynamin2-K44A, dominant negative dynamin, which inhibits the pinching of membrane during endocytosis (36). Surprisingly, there were more GLUT puncta at the PM in the above conditions under no glucose condition compared to control but there was no significant change in the number of puncta under higher glucose concentrations. Overall, glucose-dependent dynamics of GLUT puncta were disrupted when CME was blocked with (I) and (II).

In addition to the trafficking of GLUTs in response to glucose, our study examined the spatiotemporal distribution of GLUTs with respect to time upon stimulation. We found that the dynamics and distribution of GLUTs at the PM changes in a time-dependent manner upon exposure to glucose. GLUT puncta composed of multiple single molecules of GLUTs begin to appear approximately 6-8 mins after glucose addition. This temporal latency likely reflects the involvement of additional regulatory processes required for glucose sensing, consistent with observations in peripheral tissues (60).

We further investigated how endocytosis impacts the recruited GLUTs. Previous studies on the dynamics of clathrin-mediated endocytosis give insights into the timing and sequence of protein recruitment during different stages of clathrin-coated pit formation, and vesicle budding (61). We disrupted the CME using dynamin2-K44A, which impacted the GLUT dynamics and distribution in response to glucose, leading to a reduction in glucose uptake (Fig. 3V). This is in line with the data where dynamin deficient β cells show reduced insulin secretion (62). Our results suggest that GLUT endocytosis is mediated via dynamin, which explains the defects during glucose uptake leading to GSIS (62). Such a phenomenon where endocytosis mediates glucose uptake has been previously seen in fibroblast cells as well (63).

To study the same in a physiological setting, we examined the distribution and colocalization of GLUT with clathrin in mouse islets, which shows a similar trend to that of INS1 cells. Previous studies involving immunostaining of pancreatic β cells suggested that GLUT1 is the primary glucose transporter and plays a major role in GSIS (11, 12). In our study, we evaluate the trafficking of GLUT1 in human islets. In the nondiabetic (ND) human islets the glucose-dependent changes were similar to what was observed with cell lines and mouse islets. In contrast, in the T2D human pancreatic islets, we identified dysregulation of GLUT1 trafficking to the PM in response to glucose. Additionally, the expression of GLUTs was also downregulated in T2D conditions further contributing to reduced recycling of GLUTs during T2D.

Since glucose uptake is the primary input for GSIS for (I) stimulating insulin granule docking at the PM and (II) further driving insulin exocytosis in the first and second phases (41, 42, 64–66), we evaluated insulin granule docking in ND and T2D islets. We observed that the glucose-dependent insulin granules density is highly disrupted in T2D islets. Surprisingly, the time course of glucose-dependent recruitment of GLUT to the PM and endocytosis correlates with glucose-dependent insulin granule docking to the PM (27). From the above observations, we infer that the regulated exocytosis and endocytosis of GLUTs in response to glucose play an important role in maintaining proper trafficking and docking of insulin granules at the PM, hence regulating GSIS. A subpopulation of docked insulin granules, which are primed to undergo rapid exocytosis upon stimulation with glucose and plays a crucial role in the first phase of insulin secretion (16, 26, 27). Munc-13.1 is a key priming factor that is specifically associated with priming granules. Previous studies have demonstrated that the downregulation of Munc-13.1 impairs the phasic of insulin secretion (28, 29). Our results showing colocalization of GLUT with Munc-13.1, suggest that GLUTs probably spatially regulate the insulin granule priming in response to glucose. Our study emphasizes GLUT behavior in response to glucose and its regulation. This is crucial for regulating insulin granule density and therefore, represents an attractive and potential therapeutic approach for developing glucose-metabolizing drugs for the treatment of diabetes.

Materials and Methods

Cells.

INS1 832/13 cells (67) used in this study were maintained in RPMI1640, containing 10% FBS, 1% penicillin-streptomycin, Sodium pyruvate (1 mM), 0.1% β-mercaptoethanol, and 1% glutamine at 37 °C with 5% CO₂.

Mouse islets were isolated from 13 to 17-wk-old C57BL/6 mice. Islet isolation details are provided in the SI Appendix, Supplementary Materials and Methods, once isolated they were cultured in RPMI 1640 culture medium containing 16 mmol/L glucose, 10% fetal bovine serum (FBS), streptomycin (100 U/mL), and penicillin (100 U/mL) at 37 °C with 5% CO₂ (68).

Human pancreatic tissue was obtained from human cadaveric donors by the Nordic Network of Clinical Transplantation (41) or the ADI Isletcore at the University of Alberta, Canada (Ethics details in the SI Appendix) (69). Written consent from the donor and concerned families for each of the samples. The obtained tissue was cultured overnight, and islets were isolated. The islets were cultured in CMRL 1066 media supplemented with 5.5 mM glucose and other conditions as described above for INS1 cell. Trypsinization and plating of the isolated islets were performed to obtain single cells and cells cultured in the same media. Detailed protocol in SI Appendix, Supplementary Materials and Methods. The ethical approvals for both human and mouse tissue related experiments are there in the study approvals section below.

Transfection and Transduction.

Transient transfection was performed by seeding the cells on poly-L-lysine coated coverslips. Seeded cells were transfected using lipofectamine 2000®/jetPRIME® with a dilution of 0.5 to 1 µg plasmid DNA construct:1 to 3 µL reagent. The reaction was terminated after 4 to 8 h, and imaging was done 28 to 36 h after transfection.

Adenovirus particles (adNPY-mCherry) were added to the culture media at a concentration of 1:1000 and cultured for 24 to 36 h before imaging similar to protocols used in previous studies (41).

Constructs.

GLUT1-EGFP (Addgene plasmid # 18729), GLUT1-mRuby2 (Cloning was done as described in SI Appendix, Supplementary Materials and Methods) GLUT2-EGFP (Cloning was done as described in SI Appendix, Supplementary Materials and Methods), GLUT2-mCherry, Clathrin-mCherry (gifted from C J Merrifield, Institute for integrative biology of the cell, France), Dynamin2-mCherry, Dynamin2-K44A-mCherry constructs (gifted by Thomas J Pucadyil, IISER-Pune, India) (36), PIP2-EGFP (70), NPY-mCherry constructs (gifted by Sebastian Barg, Uppsala University, Uppsala, Sweden), and Munc-13.1-EGFP (gifted by Jens Rettig, Saarland University, Saarbrücken, Germany) were used in this study.

Solutions.

Cells were imaged using extracellular buffer (EC buffer) containing 138 mM NaCl, 5.6 mM KCl, 1.2 mM MgCl2, 26 mM CaCl2, 5 mM HEPES, (pH 7.4 adjusted using NaOH) and D-glucose/ Mannitol/ L-glucose/ 2-Deoxy-D-glcuose. Further, the osmolarity of the EC buffer was checked using OSMOMAT 030, which measures the osmolarity based on the freezing point depression of the sample solution. To assess endocytosis, EC buffer comprises glucose and either DMSO (vehicle control) or 20 µM pitstop2 was used.

For quantifying insulin granule density or colocalization cells were incubated with EC buffer containing 200 µM Diazoxide (27).

RNA Extraction and qPCR.

Total RNA was prepared using the Trizol method. Real-time PCR (qPCR) was performed with a SYBR GREEN SUPERMIX according to the manufacturer’s protocol. The final reaction volume was 20 μL. More details provided in SI Appendix, Supplementary Materials and Methods.

Glucose Uptake Assay.

To assess the effect of GLUT endocytosis on glucose uptake in the pancreatic β cells, which leads to insulin secretion, we performed glucose oxidase and peroxidase (GOD-POD) assay using a glucose assay kit (E-BC-K234-S, Elabscience Biotechnology, China). Cells were incubated on glucose media for 2 h along with pitstop2, followed by incubation with 3 mM glucose + pitstop2 and 20 mM glucose + pitstop2 for 45 min. Assay was performed according to the manufacturer’s protocol, additional details are provided in SI Appendix, Supplementary Materials and Methods.

Immunostaining.

Islets were incubated with desired glucose concentrations and fixed with 4% paraformaldehyde, permeabilized using 0.3% triton X and blocked in buffer containing 1% BSA in PBS. This was followed by incubation with primary antibody, GLUT1/GLUT2/Clathrin (dilution 1:200 in PBS), and later incubation with a secondary fluorescent antibody - Alexa Flour 594/568/Alexa Flour 488. After an hour, islets were washed with PBS and imaged using total internal reflection microscopy (TIRF). Additional details are provided in SI Appendix, Supplementary Materials and Methods.

Microscopy.

For primary cell culture experiments with the mouse or human islets, imaging was done using total internal fluorescence reflection (TIRF) microscope based on an Axio Observer Z1 with a 63x/1.45 objective. Scaling was maintained at 160 nm. Cells were imaged at 100 ms exposure time, the laser power was maintained at 561 (0.5mW) and 115 491 (1mW) unless specified otherwise.

INS1 cells were imaged using total TIRF microscope based on a Nikon Ti2 Eclipse with a 100x/1.45 objective. Scaling was maintained at 0.07 μm/pixel. Further details in SI Appendix, Supplementary Materials and Methods.

Image Analysis.

Puncta/granule density at the plasma membrane of the cell was calculated using the “find maxima” function in Image J software (41) or metamorph function (details in SI Appendix, Supplementary Materials and Methods). Colocalization analysis was performed using metamorph software (71). MTrackJ plugin in Image J was used to track the movement of GLUT and clathrin puncta at the plasma membrane (72, 73).

Gene Expression Analysis.

Single-cell RNA sequencing (scRNA-seq) data were obtained from the Gene Expression Omnibus (GEO; GSE221156). For our study, three T2D samples and three ND samples were included in the analysis. Details for the bioinformatic analysis are provided in SI Appendix, Supplementary Materials and Methods.

Statistics for Image Analysis Data.

Data are presented as mean ± SEM unless otherwise stated. Statistical significance was assessed using one-way ANOVA for Figs. 1 B, D, F, and H and 4 O, Q, S, and U. All other data were analyzed using Student’s t test. Significant difference is indicated by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001). Statistical tests were realized with MS-Excel/Prism.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All other data are included in the manuscript and/or SI Appendix. Previously published data were used for this work (74, 75).