Redox compartmentalization drives functional heterogeneity of mature insulin secretory vesicles in pancreatic β-cells

Abstract

Pancreatic β-cell function requires precise regulation of insulin secretory vesicles (ISVs), yet the redox heterogeneity within mature ISVs remains poorly defined. Here, we implement a novel oxidation-sensing system using NPY-fused DsRed1-E5 (Timer) targeted to mature ISVs in INS-1E and human Endoc-βH5 β-cell models. Leveraging Timer's oxidative color transition from green (Low-oxidative) to yellow-red (High-oxidative), supported by independent measurements using the established redox sensor Grx1-roGFP2, we resolve distinct ISV subpopulations. Strikingly, Krebs-Ringer Bicarbonate HEPES (KRBH) Buffer treatment amplified ISV redox heterogeneity through increasing cytosolic oxidation. Factor screening identified glutamine deprivation as the principal driver of this diversification. Spatial analysis revealed Low-oxidative ISVs predominantly docked peripherally (0–1 μm from plasma membrane), while High-oxidative ISVs localized deeper (>1 μm) and exhibited 1.7-fold higher mobility. TIRF microscopy and volumetric imaging both demonstrated superior glucose-responsive secretion from Low-oxidative ISVs during both first and second phases of glucose-stimulated insulin release. Lysotracker co-localization showed High-oxidative ISVs were preferentially targeted for lysosomal degradation (2.3-fold higher association). These findings establish an oxidation-based taxonomy for mature ISVs, linking redox states to distinct functional fates: secretion-competent Low-oxidative vesicles versus degradation-prone High-oxidative vesicles, redefining ISV heterogeneity as a fundamental organizational principle in β-cell physiology and its dysregulation in metabolic stress.

Highlights

• •ISV redox heterogeneity is proven across species and imaging platforms.
• •Glutamine mediates KRBH-induced redox changes in mature ISVs.
• •Low- and high-oxidative mature ISVs differ in distribution and kinetics.
• •Intraluminal redox level predicts individual mature ISV fate.

1. Introduction

Pancreatic β-cells maintain glucose homeostasis through insulin secretion from distinct insulin secretory vesicles (ISVs) subtypes. Multiple lines of evidence support the classification of ISVs into distinct subtypes. From a spatial distribution, ISVs are categorized into the Readily Releasable Pool (localized near the plasma membrane) and the Reserve Pool (stored in deeper cytoplasmic regions) [[1], [2], [3]]. Based on secretory competence, ISVs can be further classified as newcomer vesicles (released without prior docking), pre-docked vesicles (released after docking), and incompetent secretory vesicles [[3], [4], [5], [6], [7], [8], [9], [10]]. Regarding intraluminal crystalline status, ISVs are stratified into mature ISVs (characterized by dense cores), immature ISVs (lacking dense cores), and others (e.g., vesicles with rod-like crystalline cores) [11,12]. Additionally, compositional heterogeneity suggests that ISVs can be divided into subtypes enriched with differential membrane proteins [13], which may also exhibit distinct lipid, protein profiles, calcium affinities and fusion characteristics [5,13,14].

Following the discovery of DsRed-E5 (Timer) [15], it has been extensively implemented across diverse biological systems, including cultured cells (C2C12 [[16], [17], [18], [19]], HEK293 [[20], [21], [22]], MEF [19,22,23], INS-1 [23,24], hiPSC [21]), mouse tissues (heart [16,18], skeletal muscle [18,19,25], pancreatic islet [24], etc.), Drosophila [26] and plants [27]. Timer undergoes a time-dependent color transition from green to red through oxidation [26,28,29], a property that has been exploited for monitoring intracellular secretory vesicle age [24,30,31] and mitochondrial turnover [[16], [17], [18],22,23,25,32]. Moreover, Timer's optical properties have been demonstrated to remain stable against physiological variations, as they have been reported to be mainly dependent on oxidation and temperature [15,26]. Timer has been detected with fluorescent signals after expressed in vivo for about 6 h that longer than the maturation time (about 3 h [33,34]) of ISVs, indicating that Timer may mainly tag mature ISVs rather than immature ISVs [15]. In 2020, Syncollin-Timer was employed to determine the relative age of individual ISVs and identify regulatory factors influencing the ISVs' age distribution in β-cells [24,35]. However, Timer has also been utilized as an indicator of cumulative redox and oxygenation status in other researches [19,26,36]. Previous studies utilizing Timer-labeled mature ISVs have primarily focused on age analysis [30], leaving the effects of redox dynamics on mature ISVs subtyping and functional behaviors unexplored.

Reactive oxygen species (ROS) are primarily generated from mitochondrial activity. These ROS have been shown to impair insulin secretion through two distinct mechanisms: reduction of ATP synthesis in mitochondrial [[37], [38], [39], [40], [41]] and promotion of insulin oxidation processes [42] that contribute to β-cell dysfunction and peripheral insulin resistance [43]. Although classified as a non-essential amino acid, glutamine (Gln) critically supports β-cell function. Zhang et al. (2020) demonstrated that reductive glutamine metabolism promotes insulin secretion in β-cells [44]. Conversely, glutamine deprivation elevates mitochondrial and cytosolic ROS, correlating with β-cell dysfunction and exacerbated lipotoxicity [45]. Clinical observations reveal that glutamine levels are reduced in type 2 diabetes (T2D) patients [45], suggesting its potential role in modulating intracellular and intravesicular ROS levels, which may consequently influence ISVs subtyping and the related physiological processes.

There exists a significant knowledge gap regarding the relationship between redox levels and ISVs subtyping. In this study, we propose to address this gap by utilizing the NPY-Timer labeled β-cell system combined with established imaging technologies.

2. Results

2.1. Mature ISVs redox heterogeneity is widely observed and KRBH-inducible across multiple labelling systems and species

To investigate mature ISV redox heterogeneity, we generated NPY tag fused Timer fluorescent protein (Fig. 1A), building on its established application in mitochondrial redox studies [19,36]. Validation in INS-1E cells (immortalized rat pancreatic-β cell model) confirmed precise localization and time dependent color-shift (Supplementary Figs. S1–S2). KRBH treatment (0.2 % BSA, 1h) significantly elevated ISV redox heterogeneity (Fig. 1B, Supplementary Video.1) and luminal oxidation versus controls, evidenced by: (1) reduced 561nm/488 nm PCC (Fig. 1C); (2) increased 561nm/488 nm ratio (Fig. 1D). For cross-validation, we employed VAMP2-Grx1-roGFP2 [46,47] in INS-1E cells, which similarly showed elevated ISV redox heterogeneity (Fig. 1E) and increased 405nm/488 nm ratios (Fig. 1F) under KRBH. Parallel experiments in NPY-Timer expressed Endoc-βH5 cells (non-proliferative human pancreatic-β cell model [48]) confirmed these trends (Fig. 1G–I). Three ROS probes (DHE, CellROX Orange, H₂DCF-DA) supported KRBH-induced cytosolic ROS elevation in INS-1E cells (Supplementary Fig. S3), potentially helping to drive ISV redox heterogeneity. MIN6 cells (immortalized mouse pancreatic-β cell model) showed analogous ROS increases (Supplementary Fig. S4). Collectively, we have tried two redox sensors targeting ISV and pancreatic-β cell model from three species to prove the existence of mature ISVs redox heterogeneity, the KRBH's elevation impact on ISV's redox level and the cytosolic ROS level via three widely-used probes.

Fig. 1.

Mature ISVs redox heterogeneity is widely observed and KRBH-inducible across multiple labelling systems and species. (A) Schematic diagram for the hypothesis and workflow of using NPY-Timer system to classify ISVs redox heterogeneity in pancreatic β-cell. (B) Representative images for NPY-Timer labeled INS-1E cell under Medium/KRBH (0.2 % BSA, 30 min) group, Scale bar, 5 μm. (C) Histogram of the Pearson's Coefficient between 561nm/488 nm channels under Medium/KRBH (0.2 % BSA, 30 min) treatment in NPY-Timer labeled INS-1E cells. p < 0.0001(Mann Whitney test), n = 39/42, respectively. (D) Histogram of the 561nm/488 nm Ratio for total ISVs under Medium/KRBH (0.2 % BSA, 30 min) treatment in NPY-Timer labeled INS-1E cells. p < 0.0001(Mann Whitney test), n = 2771/3163, respectively. (E) Representative images for VAMP2-Grx1-roGFP2 labeled INS-1E cell under Medium/KRBH (0.2 % BSA, 30 min) group, Scale bar, 2 μm. (F) Histogram of the 405nm/488 nm Ratio for total ISVs under Medium/KRBH (0.2 % BSA, 30 min) treatment in VAMP2-Grx1-roGFP2 labeled INS-1E cells. p < 0.0001(Mann Whitney test), n = 1159/885, respectively. (G) Representative images for NPY-Timer labeled Endoc-βH5 cell under Medium/KRBH (0.2 % BSA, 30 min) group, Scale bar, 5 μm. (H) Histogram of the Pearson's Coefficient between 561nm/488 nm channels under Medium/KRBH (0.2 % BSA, 30 min) treatment in NPY-Timer labeled Endoc-βH5 cells. p < 0.0001(t-test), n = 12/12, respectively. (I) Histogram of the 561nm/488 nm Ratio for total ISVs under Medium, KRBH (0.2 % BSA, 30 min) treatment in NPY-Timer labeled Endoc-βH5 cells. p < 0.0001(Mann Whitney test), n = 5899/10981, respectively. Data are presented as mean ± SEM.

Supplementary data related to this article can be found online at https://doi.org/10.1016/j.redox.2025.103847

The following are the Supplementary data related to this article:

2.2. Glutamine mediates KRBH-induced mature ISVs redox heterogeneity through regulating mitochondrial ROS

To screen out the key factors under KRBH treatment on mature ISVs redox heterogeneity, we firstly performed comparative analysis of KRBH and culture medium components (reducing agents, nitrogen/carbon sources; Fig. 2A) revealed glutamine's critical role. Glutamine deprivation recapitulated KRBH's phenotype on mature ISVs redox heterogeneity, while supplementation reversed them (Fig. 2B), validated by 561nm/488 nm PCC quantification (Fig. 2C–D). Also, cytosolic ROS detection via DHE staining confirmed glutamine's key regulatory role from KRBH treatment (Fig. 2E–F, Supplementary Fig. S4). For testing whether the elevated ROS is produced from mitochondria under KRBH treatment, we also determined the mitochondrial ROS level with a well-established probe (MitoSOX Red). The results are consistent with DHE staining (Fig. 2G–H), supporting the hypothesis that the elevated ROS of KRBH group is mainly derived from mitochondria. Mito-TEMPO is mitochondria-targeted superoxide dismutase mimetic and has been reported to reduce mitochondrial ROS level [49]. We treated NPY-Timer labeled INS-1E cells with KRBH plus 100 μM mito-TEMPO for 1 h, which could remarkably alleviated KRBH-induced mature ISVs redox heterogeneity (Figs. 2I), 561nm/488 nm PCC (Fig. 2J) as well as the 561nm/488 nm ratio change (Fig. 2K). Mechanistically, glutamine has been regarded as a highly active amino acid central to multiple metabolic processes. It supports mitochondrial antioxidant capacity by supplying glutamate for glutathione(GSH) synthesis [50], fueling mitochondrial energy metabolism and cytosolic NADPH production [44], and regulating protective proteins [51], collectively reducing oxidative damage to the cell [52]. Consistent with these multifaceted roles in maintaining redox homeostasis, our data suggest that glutamine depletion as KRBH's key mechanism for modulating mature ISVs redox heterogeneity via mitochondrial ROS regulation.

Fig. 2.

Glutamine mediates KRBH-induced mature ISVs redox heterogeneity through regulating mitochondrial ROS. (A) Table of the component differences between Medium and KRBH buffer, and the corresponding key factor screening plan, respectively. (B) Representative images of the NPY-Timer labeled INS-1E cells under Medium (50 μM β-ME, 1 mM Pyruvate, 11.1 mM Glucose, 2 mM Gln), Medium - β-ME (No β-ME, 1 mM Pyruvate, 11.1 mM Glucose, 2 mM Gln), Medium - Glutamine (No Gln, 50 μM β-ME, 1 mM Pyruvate, 11.1 mM Glucose), KRBH (0.2 % BSA), KRBH + Carbon sources (Carbon, 0.2 % BSA, 1 mM Pyruvate, 11.1 mM Glucose), KRBH + Glutamine (K + Gln, 0.2 % BSA, 2 mM Gln), 30 min incubation, respectively. Scale bar = 5 μm. (C) Histogram of the Pearson's correlation coefficient under Medium, Medium - β-ME, Medium - Glutamine (No Gln) in NPY-Timer labeled INS-1E cell. p > 0.9999/p < 0.0001(Dunn's multiple comparisons test), n = 60/7/18, respectively. (D) Histogram of the Pearson's correlation coefficient under KRBH, KRBH + Carbon sources (Carbon), KRBH + Glutamine (K + Gln) in NPY-Timer labeled INS-1E cell. p = 0.1137/p < 0.0001(Dunnett's multiple comparisons test), n = 60/8/7, respectively. (E) Representative images of the DHE stained INS-1E cells under Medium, Medium - Glutamine (No Gln), KRBH, KRBH + Glutamine (K + Gln), respectively. Scale bar = 50 μm. (F) Histogram of the DHE intensity under Medium, Medium - Glutamine (No Gln), KRBH, KRBH + Glutamine (K + Gln). p < 0.0001/p < 0.0001(Dunn's multiple comparisons test), n = 18/21/19/13, respectively. (G) Representative images for MitoSOX Red stained INS-1E cell under Medium, Medium - Glutamine (No Gln), KRBH, KRBH + Glutamine (K + Gln), respectively. Scale bar = 30 μm. (H) Histogram of the MitoSOX Red intensity under Medium, Medium - Glutamine (No Gln), KRBH, KRBH + Glutamine (K + Gln). p < 0.0001/p = 0.3864/p < 0.0001 (Šídák's multiple comparisons test), n = 6/6/6/6, respectively. (I) Representative images of the NPY-Timer expressed INS-1E cells under Medium, KRBH, KRBH + mito-TEMPO (K + TEP, 0.2 % BSA, 100 μM mito-TEMPO) for 30 min, respectively. Scale bar = 5 μm. (J) Histogram of the Pearson's correlation coefficient under Medium, KRBH, KRBH + mito-TEMPO (K + TEP), p < 0.0001/p < 0.0001(Dunnett's multiple comparisons test), n = 21/16/14, respectively. (K) Histogram of the 561nm/488 nm Ratio for total ISVs under Medium, KRBH, KRBH + mito-TEMPO (K + TEP), p < 0.0001/p < 0.0001(Dunn's multiple comparisons test), n = 945/1611/1015, respectively. Data are presented as mean ± SEM.

2.3. Low-oxidative mature ISVs are different from High-oxidative mature ISVs in spatial distribution and kinetics

Glutamine deprivation-induced redox heterogeneity provides an analytical framework for distinguishing ISVs subtypes. Three-dimensional segmentation classified mature ISVs into Low-oxidative and High-oxidative subtypes based on 561nm/488 nm ratio (Fig. 3A). Distance distribution curves confirmed Low-oxidative ISVs proximity to plasma membrane (PM), while High-oxidative ISVs exhibit a right-shift trend (Fig. 3B–D). Similar spatial analysis in VAMP2-Grx1-roGFP2 labeled INS-1E cells also supported that Low-oxidative ISVs locate to the PM-proximal regions, rather High-oxidative ISVs (Fig. 3E–F). Tracking analysis demonstrated oxidative status-dependent mobility: track length of individual ISV is likely to correlative with the red-shift coloration of its spectrum (Fig. 3G), consistent with prior reports of increased motility by vesicle aging [53]. Velocity distribution curves exhibited left-shifted peaks for Low-oxidative ISVs versus High-oxidative subtypes (Fig. 3H). Given comparable spatial/kinetic profiles of Red/Yellow ISVs, it also strongly convincing us about the taxonomy standard (Fig. 1A) by ISV redox level (561nm/488 nm ratio), and support that ISVs can be separated into two subtypes: Low-oxidative (Green) and High-oxidative (Red/Yellow). Together, our results suggest that Low-oxidative ISVs may preferentially dock near PM, while High-oxidative subtypes occupy internal regions with enhanced trafficking activity, indicating potentially distinct functional pathways and fate determination.

Fig. 3.

Low-oxidative mature ISVs are different from High-oxidative mature ISVs in spatial distribution and kinetics. (A) Representative images of raw data (left panel) and segmentation (right panel) of NPY-Timer labeled INS-1E cell under Glutamine starvation, Scale bar = 3 μm. Insulin secretory vesicles could be classified into 2 subtypes (Low-oxidative: Green, High-oxidative: Yellow & Red). (B) Frequency distribution of the Distance to PM for 2 subtypes (Low-oxidative: Green, High-oxidative: Yellow & Red) under Glutamine starvation in NPY-Timer labeled INS-1E cells. (C) Histogram of the Distance to PM for 2 subtypes (Low-oxidative, High-oxidative) under Glutamine starvation in NPY-Timer labeled INS-1E cells. p < 0.0001(Mann Whitney test), n = 909/655, respectively. (D) Histogram of the Percentage in each ISV subtype near PM (Distance ≤ 1 μm) under Glutamine starvation in NPY-Timer labeled INS-1E cells. p < 0.0001(Paired t-test), n = 25/25. (E) Histogram of the Distance to PM for 2 subtypes under Glutamine starvation in VAMP2-Grx1-roGFP2 labeled INS-1E cells. p < 0.0001(Mann Whitney test), n = 601/350, respectively. (F) Histogram of the Percentage in each ISV subtype near PM (Distance ≤ 1 μm) under Glutamine starvation in VAMP2-Grx1-roGFP2 labeled INS-1E cells. p = 0.0034(Paired t-test), n = 10/10. (G) Representative images of raw data (left panel) and the tracks (right panel) of NPY-Timer labeled INS-1E cell under Glutamine starvation, Scale bar = 3 μm. Insulin secretory vesicles could be classified into 2 subtypes (Low-oxidative: Green, High-oxidative: Yellow & Red). (H) Frequency distribution of the Mean velocity for 2 subtypes (Low-oxidative: Green, High-oxidative: Yellow & Red) under Glutamine starvation in NPY-Timer labeled INS-1E cells. Data are presented as mean ± SEM.

2.4. Oxidative level is strongly correlated to the fate of individual mature ISV within the population

To assess functional differences between redox subtypes, we performed TIRF imaging in NPY-Timer-labeled INS-1E cells during glucose-stimulated insulin secretion (GSIS). Mature ISV subtypes were quantified at three timepoints: Starvation (2.8 mM glucose, 30 min), Phase1 (16.7 mM glucose, 5 min), and Phase2 (16.7 mM glucose, 30 min) (Fig. 4A). Total ISVs decreased after 16.7 mM glucose stimulation (Fig. 4B), primarily due to reduction of Low-oxidative ISVs (Fig. 4C), not High-oxidative ISVs (Fig. 4D). Frame-by-frame analysis captured fusion events exclusively in Low-oxidative ISVs (Fig. 4E), supported by fluorescence intensity changes along time (Fig. 4F), consistent with prior reports of younger vesicles’ higher secretion capacity [5,24]. Furthermore, volumetric imaging in NPY-Timer-labeled Endoc-βH5 cells during GSIS (Fig. 4G, left) evaluated PM-proximal ISV subtypes (Distance-to-PM ≤1 μm) (Fig. 4G, right). Low-oxidative ISVs predominated during secretion (Fig. 4H), with >50 % localized near PM versus <30 % of High-oxidative ISVs (Fig. 4I–J). Low-oxidative ISVs replenished during Phase2 (Fig. 4H–I). The volumetric fluorescent imaging of NPY-Timer-labeled INS-1E cells showed similar patterns (Supplementary Fig. S5).

Fig. 4.

Oxidative level is strongly correlated to the fate of individual mature ISV within the population. (A) Representative TIRF images of NPY-Timer labeled INS-1E cell during glucose-stimulated insulin secretion (GSIS) process. Starvation (2.8 mM glucose for 30 min), Phase1(16.7 mM glucose for 5 min), Phase2(16.7 mM glucose for 30 min). Scale bar = 3 μm. (B) Histogram of the total ISV number change from the TIRF images of NPY-Timer labeled INS-1E cells. p < 0.0001/p = 0.0477(Paired, Dunnett's multiple comparisons test), n = 20/20/20, respectively. (C) Histogram of the Low-oxidative ISV number change from the TIRF images of NPY-Timer labeled INS-1E cells. p < 0.0001/p = 0.0027(Paired, Dunnett's multiple comparisons test), n = 20/20/20, respectively. (D) Histogram of the High-oxidative ISV number change from the TIRF images of NPY-Timer labeled INS-1E cells. p = 0.07819/p = 0.5567(Paired, Dunnett's multiple comparisons test), n = 20/20/20, respectively. (E) Representative TIRF images for the fusion event of one Low-oxidative ISV and Non-fusion property of High-oxidative ISV in NPY-Timer labeled INS-1E cells. 1.3s per frame. Scale bar = 100 nm. (F) Fluorescent intensity curve of each ISV subtype along time from Fig. 4E. (G) Representative volumetric images of raw data (left panel) of NPY-Timer labeled Endoc-βH5 cell during GSIS, Scale bar = 2 μm. 3D cartoon model (right panel) shows how to measure “Distance to PM” for Low-oxidative ISVs and High-oxidative ISVs. (H) Histogram of the relative number change of Low-oxidative ISVs and High-oxidative ISVs near PM (Distance ≤ 1 μm) normalized to the Starvation group. Starvation (2.8 mM glucose for 30 min), Phase1(16.7 mM glucose for 5 min), Phase2(16.7 mM glucose for 30 min). n = 13/13/13, respectively. (I) Frequency distribution of the number change of Low-oxidative ISVs during GSIS in NPY-Timer labeled Endoc-βH5 cell during GSIS. Over 50 % of Low-oxidative ISVs are locate near PM (Distance ≤ 1 μm). (J) Frequency distribution of the number change of High-oxidative ISVs during GSIS in NPY-Timer labeled Endoc-βH5 cell during GSIS. Less than 30 % of High-oxidative ISVs are locate near PM (Distance ≤ 1 μm). (K) Representative images of the Low/High-oxidative ISVs distribution under Glutamine starvation for 1h/2h/3h within NPY-Timer labeled INS-1E cell, Scale bar, 5 μm. (L) Curve of the Distance to PM of Low-oxidative ISVs and High-oxidative ISVs under Glutamine starvation for 1h/2h/3h within NPY-Timer labeled INS-1E cell. p = 0.0383(Paired t-test), n = 15/15/15 cells, respectively. (M) Representative colocalization images of the Low/High-oxidative ISVs to lysosomes under Glutamine starvation for 1h/3h within NPY-Timer and Lysotracker co-labeled INS-1E cell. 1h group at the top row, and 3h group at the bottom row. Scale bar, 5 μm. (N) Histogram of the Pearson correlation coefficient for quantifying the colocalization level of the Low-oxidative ISVs and High-oxidative ISVs to lysosomes under Glutamine starvation for 1h/3h within NPY-Timer and Lysotracker co-labeled INS-1E cell. For Low-oxidative ISV's 1h vs 3h: p = 0.1975 (Šídák's multiple comparisons test), n = 22/70 cells, respectively. For High-oxidative ISV's 1h vs 3h: p < 0.0001 (Šídák's multiple comparisons test), n = 22/70 cells, respectively. Data are presented as mean ± SEM.

Given High-oxidative ISVs near PM decreased without replenishment, we further investigated their possible fate. Glutamine deprivation induced perinuclear clustering of High-oxidative ISVs (Fig. 4K), quantified by Distance-to-PM measurements (Fig. 4L). Lysotracker colocalization analysis showed time-dependent accumulation of High-oxidative ISVs with lysosomes (Fig. 4M−N), consistent with nutrient depletion-induced degradation [[54], [55], [56]].

These findings establish-oxidative level as a determinant of mature ISVs fate: Low-oxidative ISVs exhibit glucose-responsive secretion, while High-oxidative counterparts undergo lysosomal degradation (Fig. 5). This redox-based dichotomy suggests oxidative status may regulate insulin vesicle turnover, with implications for β-cell functional maintenance.

Fig. 5.

A proposed model for the redox heterogeneity of mature ISV pool inside pancreatic β-cell. (A) The whole picture for the redox heterogeneity of ISVs population inside pancreatic β-cell. (B) The ROI1 from Panel A shows that High-oxidative ISVs (Red/Yellow) inside lysosome undergo digestive enzyme mediated degradation. (C) The ROI2 from Panel A shows that Low-oxidative ISVs (Green) near plasma membrane undergo exocytosis.

3. Discussion

This study leverages the color transition property of the Timer fluorescent protein in response to oxidation, coupled with VAMP2-Grx1-roGFP2 redox sensor, establishing Timer's utility as a sensor for the redox status of insulin secretory vesicles in pancreatic β-cells. We identified significant redox heterogeneity within the ISV pool and successfully categorized ISVs into distinct Low-oxidative and High-oxidative subtypes. These subtypes exhibited divergent spatial distributions and kinetic behaviors. Crucially, our investigations revealed a functional association: Low-oxidative ISVs displayed a preferential propensity for exocytosis, whereas High-oxidative ISVs were predominantly linked to degradation processes. In summary, we established a novel system for detecting redox heterogeneity in mature ISVs and classified them into two functionally relevant subtypes based on their redox status, which correlates with differential fate decisions. Future studies could focus on elucidating the significance of mature ISVs redox heterogeneity in β-cell dysfunction during diabetic conditions and exploring potential underlying mechanisms, ultimately aiding in the evaluation of approved drugs and the screening of novel therapeutics for T2D.

To explore the pathological significance of ISV redox subtyping, our preliminary data indicate that glucolipotoxicity (GLT) treatment elevates ISV oxidative levels (561nm/488 nm ratio, Supplementary Fig. S6). This observation aligns with elevated cytosolic and mitochondrial ROS levels in diabetic β-cells reported previously [46]. Accumulation of High-oxidative ISVs likely impairs ISV functional homeostasis and secretion capacity, potentially contributing to β-cell failure. More work is required to investigate GLT's impact on ISV homeostasis and screen therapeutics for β-cell functional restoration.

Oxidation plays an important role in insulin secretory vesicle maturation and β-cell function [[57], [58], [59]] Literature analysis reveals Timer fluorescent signals emerge ∼6 h post-expression - exceeding typical ISV maturation time (∼3 h [33,34]) - indicating Timer may primarily tags mature ISVs over immature vesicles [15]. This confirms NPY-Timer-labeled ISVs represent mature populations. It has been reported that NOX4-mediated cytosolic H₂O₂ production correlates with K_ATP channel closure regulating GSIS [57]. Although cytosolic pro-oxidative states facilitate glucose-stimulated ISV exocytosis, its correlation to redox-based ISV subtyping remains unexplored. Further studies will be necessary to assess the NOX4 or K_ATP channel's impact on ISV redox heterogeneity.

While our data support mitochondrial ROS regulation of ISV redox heterogeneity (Fig. 2G–K), the mechanism underlying redox propagation to ISVs requires further investigation. We propose testable hypotheses by priority: (1) Inter-organelle interactions: Evidence suggests close interactions between mitochondria and ISVs, which could involve the formation of direct contact sites. Such Mitochondria-ISV contact sites might facilitate direct ROS transfer. This possibility is supported by our TEM data (Supplementary Fig. S7) and advanced imaging studies (soft X-ray [60,61], cryo-ET [2]) demonstrating organellar proximity. Furthermore, a recent study identified mitochondria-peroxisome contacts maintaining redox homeostasis via PTPIP51-ACBD5 mediated ROS transfer [62]. This finding not only strengthens the possibility of direct contact site formation between mitochondria and diverse organelles, including ISVs, but also provides a valuable reference blueprint for exploring the potential molecular mechanisms of ROS transfer at mitochondrial-ISV interfaces. (2) Vesicle membrane permeability: Distinct membrane compositions among ISV subtypes [13,14] may lead to differential ROS permeability. Oxidative alterations in High-oxidative ISVs may modify membrane properties, potentially enhancing ROS permeability and establishing a self-amplifying oxidative damage cycle. Further research is warranted to rigorously test the hypothesis and the potential molecular mechanisms during this process.

4. Conclusions

In β-cell biology, reactive oxygen species (ROS) have been established as significant regulators of insulin secretion, mitochondrial activity, and β-cell integrity. Our study has identified redox heterogeneity within insulin secretory vesicles (ISVs), correlating it with ISVs subtyping and fate determination. Notably, glutamine reduction has been documented to elevate mitochondrial ROS levels, a phenomenon also observed in type 2 diabetes (T2D) patients, which has further been confirmed to enhance ISVs redox heterogeneity in β-cells in this study. Consequently, investigating the relationship between ISVs redox heterogeneity under T2D conditions and assessing its impact on current T2D therapeutic agents’ efficacy in modulating ISVs redox heterogeneity presents a compelling avenue for future research.

5. Methods

5.1. Cell culture

Rat β-cell model (INS-1E) was a gift from Professor Pierre Maechler (University of Geneva, Switzerland). The cells were cultured in RPMI 1640 culture medium (Invitrogen, 11875119) containing 2 mM l-Glutamine (Hyclone, SH30034.01), 11.1 mM glucose (Invitrogen, A2494001), 10 % FBS (Invitrogen, 10091148), 50 mg/ml Penicillin-Streptomycin (Invitrogen, 15140122), 10 mM HEPES (Invitrogen, 15630080), 1 mM Sodium Pyruvate (Invitrogen, 11360070), and 50 μM β-Mercaptoethanol (β-ME, Invitrogen, 21985023). The cells were maintained at 37 °C in a humidified atmosphere with 5 % CO₂.

Murine β-cell line (MIN6) was a gift from the mammalian core of iHuman institute. The cells were cultured in DMEM medium containing 2 mM l-Glutamine, 25 mM glucose, 15 % FBS, 50 mg/ml Penicillin-Streptomycin, 20 mM HEPES, 1 mM sodium pyruvate, and 50 μM β-Mercaptoethanol (β-ME).

Human β-cell model (Endoc-βH5) were purchased from the Human Cell Design (HCD) Corporation. Endoc-βH5 cells were cultured by Ulti-β1(HCD, UB1-100-BSA) culture media on β-Coat (HCD, β-Coat) precoated dishes according to the commercially established protocol. Then cells were used for transfection and stimulation.

To generate the INS-1E cell line stably expressed NPY-Timer, plasmids without endotoxin were prepared by the EndoFree Plasmid Maxi Kit. Wildtype INS-1E cells were transfected using Neon™ NxT Electroporation System (Invitrogen, NEON1S) with the plasmids, carrying the NeoR gene which confers resistance to G-418 (MCE, HY-17561). After 48 h transfection, the medium was re-placed with fresh complete medium supplemented with G-418 at 200 μg/mL and a 15 days selection allowed obtaining a stable polyclonal cell population expressing the plasmids. Flow cytometry was applied for sorting positive cells by BD Influx. After selection, G-418 concentration was reduced at 50 μg/mL for stable cells maintenance.

Before imaging, 6e5 to 8e5 INS-1E/MIN6 cells were seeded into a 29 mm glass-bottom dish pre-coated with Poly-l-ornithine (Sigma, P4957) for 30 min and cultured for 48 h–72 h before data collection.

5.2. Molecular cloning

NPY-Timer plasmid was generated by the recombination reaction kit between the NPY fragment from NPY-mApple plasmid (Addgene, 83498) and Timer fragment from RIP-Timer plasmid (Addgene, 15109). VAMP2-Grx-roGFP2 plasmid was generated by the recombination reaction kit between the VAMP2 fragment from VAMP2-pHmScarlet plasmid (Weikigene, 340) and the Grx-roGFP2 fragment from pEIGW Grx1-roGFP2 plasmid (Addgene, 64990). After sequenced, the right plasmids were amplified and extracted with Endotoxin-free plasmid extraction kit. NPY- pHluorin2 plasmid was generated from the recombination reaction kit between the NPY fragment from NPY-mApple plasmid (Addgene, 83498) and pHluorin2 fragment (a gift from Prof. Xiaoqing Zhang and Prof. Yujiang Fang (Tongji University, China)).

The Primer sequences are as follows:

NPY-F: GACCGGTGGATCCCGGGCC.

NPY-R: GCCATACCACATTTGTAGAGG.

Timer-F: GCCCGGGATCCACCGGTCCCAGCCCTAACTCTAGAG.

Timer-R: CTACAAATGTGGTATGGCTACAAATGTGGTATGGCT.

VAMP2-F: GAGCTGTACAAGTAAGCGGCCGCGACTCTAG.

VAMP2-R: AAACTCTTGAGCCATCATGGTGGGATCCCCGC.

Grx1-roGFP2-F: ATGGCTCAAGAGTTTGTGAACTGC.

Grx1-roGFP2-R: TTACTTGTACAGCTCGTCCATGCC

pHluorin2-F: CCACCGGTCGCCACCATGGTGAGCAAGGGCGA

pHluorin2-R: GAGTCGCGGCCGCTATCACTTGTACAGCTCGTCCATG.

5.3. Dye staining

For Dihydroethidium (DHE) staining, wildtype INS-1E/MIN6 cells were seeded and incubated with the corresponding stimuli (Culture Medium, Culture Medium without l-Glutamine (Invitrogen, 21870076), KRBH (Coolaber, SL65501) with 0.2 % BSA (Sigma, B2064), KRBH with 0.2 % BSA plus 2 mM l-Glutamine) and stained with 10 μM DHE (MCE, HY-D0079) for 30 min. Then data were collected by DV-Ultra DeltaVision microscopy with a sCMOS camera as well as a 1.522 oil-immersion 100x objective lens.

For CellROX Orange staining, wildtype INS-1E cells were seeded and incubated with the corresponding stimuli (Culture Medium, Culture Medium without l-Glutamine, KRBH with 0.2 % BSA, KRBH with 0.2 % BSA plus 2 mM l-Glutamine) and stained with 5 μM CellROX Orange (Yeasen, 50103ES50) for 30 min. Then data were collected by DV-Ultra DeltaVision microscopy with a sCMOS camera as well as a 1.522 oil-immersion 100x objective lens.

For Lysotracker Deep Red staining, NPY-Timer stably expressed INS-1E cells were seeded and incubated under KRBH with 0.2 % BSA for 1 h or 3 h and stained with 0.5 μM Lysotracker Deep Red (Invitrogen, L12492) for 10 min. Then data were collected by Leica Thunder Live Cell Imager with a sCMOS camera as well as a 1.516 oil-immersion 100x objective lens.

For H2DCF-DA staining, wildtype INS-1E cells were seeded and incubated with the corresponding stimuli (Culture Medium, KRBH with 0.2 % BSA) and stained with 10 μM H2DCF-DA (MCE, HY-D0940) for 30 min. Then data were collected by Nikon Spinning Disk Microscopy with a 1.516 oil-immersion 60x objective lens.

For MitoSOX Red staining, wildtype INS-1E cells were seeded and incubated with the corresponding stimuli (Culture Medium, Culture Medium without l-Glutamine, KRBH with 0.2 % BSA, KRBH with 0.2 % BSA plus 2 mM l-Glutamine) for 30 min to 1h and stained with 2.5 μM MitoSOX Red (MCE, HY-D1055) for 10 min. Then data were collected by Leica Thunder Live Cell Imager with a sCMOS camera as well as a 1.516 oil-immersion 63x objective lens.

5.4. Epi-microscopy imaging

For detecting the 561nm/488 nm ratio change of Timer-labeled insulin vesicles in a long duration, INS-1E cells seeded on glass-bottom dishes were transfected with NPY-Timer with a confluency about 70 %, then imaged for 3D whole cell volume in specific timepoints (0/12/24/36/48/72/96/120/144/168/192/216 h, respectively) by DV-Ultra DeltaVision microscopy with a sCMOS camera as well as a 1.522 oil-immersion 100x objective lens. Each cell is about 10 μm thickness, Z-interval is 0.2 μm. Images were processed by the algorithm of Deconvolution to enhance the contrast of insulin secretory vesicles.

For detecting the 561nm/488 nm ratio change of Timer-labeled insulin vesicles, INS-1E cells seeded on glass-bottom dishes were transfected with NPY-Timer with a confluency about 70 %, then imaged for 3D whole cell volume in specific timepoints by DV-Ultra DeltaVision microscopy with a 1.522 oil-immersion 100x objective lens.

For 2D imaging of triple labelling assay, NPY-Timer expressed INS-1E cells were seeded glass-bottom dishes and co-labeled with Lysotracker Deep Red. Then data were collected by Leica Thunder Live Cell Imager with a sCMOS camera as well as a 1.516 oil-immersion 100x objective lens. Images were processed by the algorithm of Instant Computational Clearing (ICC) to enhance the contrast of insulin secretory vesicles. ICC parameters include Strength as 90 % and Feature scale as “1500 nm”.

For 3D volumetric imaging during GSIS, NPY-Timer expressed INS-1E cells or Endoc-βH5 cells were seeded glass-bottom dishes. Then cells were starved under KRBH buffer supplemented with 2.8 mM glucose and 0.2 % BSA, and the images were captured as Starvation (2.8 mM glucose for 30 min), Phase1(16.7 mM glucose for 5 min), Phase2(16.7 mM glucose for 30 min). Related data were collected by Leica Thunder Live Cell Imager with a sCMOS camera as well as a 1.516 oil-immersion 100x objective lens. Images were processed by the algorithm of Instant Computational Clearing (ICC) to enhance the contrast of insulin secretory vesicles. ICC parameters include Strength as 90 % and Feature scale as “1500 nm”.

5.5. Immunofluorescence

INS-1E cells cultured on coverslips were fixed with 4 % PFA (Coolaber, SL1830) for 10 min, and then permeabilized with 0.1 % Triton (Sigma, X100) for 10 min at room temperature. After blocking with 10 % donkey serum (Solarbio, SL050) for 1 h, cells were incubated with the indicated primary insulin-antibody (CST, 8138S, 1/800) diluted in 1 % donkey serum for overnight at 4 °C. After washing 3 times with PBS, cells were incubated with fluorophore-tagged secondary antibody (Invitrogen, A32787, 1/500) for 2 h at room temperature. Coverslips were washed by PBS and mounted by ProLong™ Gold antifade (Invitrogen, P36930), finally protected from light and dried overnight. Then the samples were imaged under Nikon SORA spinning disk microscopy with a 1.518 oil-immersion 100x objective lens.

5.6. Transmission electron microscopy

INS-1E or Endoc-βH5 cells were seeded onto coverslips and fixed in 0.1 M cacodylate-buffered (pH 7.4, EMS, 11652), 2 % paraformaldehyde (Coolaber, SL1830) and 2.5 % glutaraldehyde (Sigma, G6403) for 1 h, then post-fixed in 1 % osmium tetroxide (Thermo, 18450) and 1.5 % (w/v) K₃Fe (CN)₆ (Merck, 14459-95-1) for 1 h. Following this, the samples were stained with 2 % uranyl acetate (EMS, 22400) for 2 h, dehydrated with gradient ethanol, and embedded in Epoxy Resin (SPI Supplies, 02659-AB, 02827-AF, 02828-AF, 02823-DA). The embedded samples were freshly cut onto TEM-grids (Quantifoil, Q57481) and imaged.

For Mice islets, following a similar workflow, samples were isolated and fixed in 0.1 mM phosphate-buffered solution (Macklin, P885754) containing 2.5 % glutaraldehyde (Sigma, G6403), followed by post-fixed in 1 % osmium tetroxide (Thermo, 18450). Ultrathin sections (60 nm) required re-staining with 2 % uranyl acetate (EMS, 22400) before dehydration and embedding. Digital images were acquired by Zeiss GeminiSEM 460 using accelerating voltage of 5 KeV at Molecular Imaging Core in Shanghaitech University and analyzed via Fiji (1.54f).

5.7. Intraluminal acidification detection of ISV

NPY- pHluorin2 [63] expressed INS-1E cells were seed onto black 96well-plates with flat bottom and treated with corresponding stimulations for 30 min: Culture Medium, KRBH (0.2 % BSA), 100 nM Bafilomycin A1 (MCE, HY-100558). The acidification level was subsequently evaluated using Flexstation 3.0 (Molecular Devices) with SoftMax Pro (v5.4.5.000, Molecular Devices), with the excitation filters of 405 nm and 485 nm, and emission filter of 535 nm (for both excitation wavelengths.

5.8. Total internal reflection fluorescence (TIRF) imaging

NPY-Timer expressed INS-1E cells were prepared with a confluency about 70 % and starved for 30 min under KRBH buffer supplemented with 2.8 mM glucose and 0.2 % BSA. Time lapse data were collected every 1.3 s for 20 min using Nikon Ti2-E TIRF microscopy with a 1.518 oil-immersion 60x objective lens in PSF mode.

For TIRF imaging during GSIS, NPY-Timer expressed INS-1E cells were prepared with a confluency about 70 % and starved under KRBH buffer supplemented with 2.8 mM glucose and 0.2 % BSA. Then the images were captured as Starvation (2.8 mM glucose for 30 min), Phase1(16.7 mM glucose for 5 min), Phase2(16.7 mM glucose for 30 min) using Nikon Ti2-E TIRF microscopy with a 1.518 oil-immersion 60x objective lens in PSF mode.

5.9. Image analysis

For 3D segmentation, images were loaded into Imaris (10.2, Oxford Instrument) and segmented with the Spots module. Plasma Membrane (PM) was manually drawn for measuring the shortest distance of ISVs subtypes to PM. Representative images were captured in Imaris and data were exported to GraphPad prism for the statistical analysis.

For colocalization analysis, images were loaded into Fiji (1.54f) and analyzed by the plugin Coloc2 to calculate the Pearson Correlation Coefficient, and exported to GraphPad prism 10.4.2 for the statistical analysis.

For the intensity detection of DHE/Cell ROX Orange/H2DCF-DA/MitoSOX Red, images were loaded into Fiji (1.54f) and masked by thresholding. Then the images were measured in the ROI manager and exported to Excel for data arrangement and then transferred to GraphPad prism 10.4.2 for the statistical analysis.

5.10. Quantification and statistical analysis

All histograms and curve were exported from GraphPad prism 10.4.2 For most histograms, individual data points are presented unless the overcrowd points block the trend display. The results are presented as mean ± SEM. t-test or One-way ANOVA was used in statistical analysis. Each result was collected at least three independent batches, and each batch includes at least one technological repeat.

CRediT authorship contribution statement

Lu Zhuang: Writing – review & editing, Writing – original draft, Visualization, Validation, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Yuwei Zhao: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation. Qikai Qin: Writing – review & editing, Visualization. Kejia Xiong: Writing – review & editing, Investigation, Formal analysis. Zhen Qian: Writing – review & editing, Visualization, Formal analysis. Yan Liu: Writing – review & editing, Visualization, Validation, Supervision, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data will be made available on request.