Mitochondrial Problems in Insulin-Producing Cells: Causes and Possible Treatments

Pancreatic β-cells are highly specialized for glucose-stimulated insulin secretion (GSIS) and rely critically on mitochondrial metabolism to translate glucose sensing into exocytosis. Glucose enters β-cells and is metabolized via glycolysis to pyruvate, which fuels the tricarboxylic acid (TCA) cycle and oxidative phosphorylation. In the mitochondria, oxidation of pyruvate generates NADH and FADH₂, driving electron transport, ATP synthesis and the generation of metabolic coupling factors (such as ATP, GTP, NADPH and intermediate metabolites) that trigger insulin granule exocytosis [1,2,3,4]. Calcium uptake into mitochondria further stimulates TCA cycle dehydrogenases, amplifying ATP production and insulin release [5]. β-cell mitochondria also supply reducing equivalents (NADPH) and ATP to the endoplasmic reticulum (ER) to support proinsulin folding and maturation [6,7]. Indeed, the ER and mitochondria operate in redox balance: proinsulin oxidative folding in the ER requires NADPH from mitochondria, and mitochondrial function depends on ER redox homeostasis [8]. In short, β-cells funnel glucose-derived carbons and energy through mitochondria for insulin secretion, and any disruption of mitochondrial ATP or coupling-factor production impairs GSIS [9]. The interplay between β-cell mitochondrial bioenergetics and glucose metabolism is therefore not only important, but necessary, for normal insulin secretion and glucose homeostasis [10].

Although oxidative phosphorylation (OXPHOS)-derived ATP has long been viewed as the principal metabolic signal driving GSIS, emerging data challenge this traditional model. Recent studies suggest that glycolytic ATP production and cytosolic ATP gradients may serve as key coupling factors for insulin exocytosis, particularly during the initial glucose response. At the plasma membrane, local pyruvate Kinase (PK), converts phosphoenolpyruvate (PEP) to ATP, thereby closing K_ATP channels and causing Ca²⁺ influx independently of mitochondrial oxidative phosphorylation. Genetic and pharmacologic activation of PK increases cytosolic ATP/ADP and insulin secretion [11,12]. Furthermore, direct electrophysiological recordings have demonstrated the presence of a plasma membrane associated glycolytic metabolon that locally channels substrates through phosphofructokinase and pyruvate kinase to generate ATP and modulate K_ATP ATP channel activity, reinforcing the concept of compartmentalized glycolytic ATP signaling in β-cell excitability [13]. This has led to a growing debate over whether OXPHOS-derived ATP or glycolytically generated ATP plays the dominant role in β-cell stimulus–secretion coupling.

However, this compartmentalized glycolytic ATP model does not fully negate the established role of mitochondrial metabolism in sustaining insulin secretion. While glycolytic ATP may cause K_ATP closure and membrane depolarization, mitochondrial oxidative phosphorylation remains essential for sustaining the secretory response through the continuous provision of ATP, NADPH, and anaplerotic intermediates that fuel the TCA cycle [14,15]. In fact, studies using mitochondrial inhibitors demonstrate that β-cell depolarization may occur in the absence of OXPHOS, but the second phase of insulin release collapses when mitochondrial metabolism is impaired, emphasizing a sequential interplay between glycolytic initiation and mitochondrial amplification [16].

These findings suggest that β-cell bioenergetics are spatially and temporally compartmentalized rather than mutually exclusive, glycolytic ATP may act locally at the plasma membrane, while mitochondrial ATP and coupling factors regulate global energy homeostasis and redox signaling [17]. Thus, the current debate reflects not a replacement of the mitochondrial model, but a refinement toward a dual-pathway framework in which both glycolytic and oxidative ATP production cooperate to support GSIS. Understanding this integrated metabolic crosstalk is crucial, as mitochondrial dysfunction in diabetes likely disrupts not only oxidative ATP supply but also the downstream coordination of glycolytic flux and cellular excitability.

Further, despite progress toward a dual-pathway model, several key mechanistic questions remain unresolved, and merit targeted experimental investigation. First, what is the relative spatiotemporal contribution of glycolytic versus mitochondrial ATP to K_ATP channel closure in intact islets under physiological glucose oscillations? This question could be addressed using genetically encoded ATP sensors with subcellular targeting (e.g., plasma membrane-targeted versus mitochondria-targeted PercevalHR or iATPSnFR2) combined with simultaneous electrophysiology and live-cell imaging in human islets [18,19]. Such experiments would reveal whether local ATP gradients at the plasma membrane are sufficient to fully explain first-phase insulin secretion or whether mitochondrial ATP contributes even at this early stage. Second, how do different glucose concentrations (5 mM vs. 10 mM vs. 20 mM) shift the relative dependence on glycolytic versus mitochondrial pathways? Dose–response studies using selective inhibitors (e.g., UK5099 to block mitochondrial pyruvate transport, or oligomycin to inhibit ATP synthase) alongside glycolytic enzyme modulators (e.g., TEPP-46 to activate PKM2) would delineate the glucose concentration thresholds at which each pathway becomes rate-limiting [11,20]. Third, does the glycolytic metabolon at the plasma membrane physically interact with mitochondria via mitochondria-associated ER membranes (MAMs) or other contact sites? High-resolution proximity labeling (e.g., TurboID or APEX2 targeted to plasma membrane K_ATP channels or glycolytic enzymes) followed by mass spectrometry could identify whether mitochondrial proteins participate in plasma membrane metabolic signaling complexes [21]. Fourth, can mitochondrial dysfunction be rescued by enhancing glycolytic ATP production, or does loss of mitochondrial coupling factors (NADPH, GTP, anaplerotic substrates) represent an insurmountable barrier? β-cell-specific overexpression of plasma membrane-targeted pyruvate kinase isoforms in mitochondrial transcription factor A (TFAM) knockout or mtDNA-depleted models would test whether amplifying local glycolytic ATP can compensate for mitochondrial failure [12,22]. Finally, how does chronic hyperglycemia reprogram the balance between glycolytic and mitochondrial ATP signaling in human T2DM islets? Single-cell metabolic flux analysis using ¹³C-glucose tracing combined with single-cell RNA-seq and ATP imaging in islets from diabetic donors could reveal whether the shift toward glycolytic dependence represents a compensatory adaptation or a maladaptive metabolic rewiring [23]. Resolving these questions will require integrating cutting-edge imaging, metabolomics, and genetic tools in physiologically relevant β-cell models and will ultimately determine whether therapeutic strategies should aim to restore mitochondrial function, enhance glycolytic capacity, or rebalance both pathways.

That said, under physiologic conditions, β-cell mitochondria maintain an optimal balance between energy production and reactive oxygen species (ROS) generation. Low physiological levels of ROS can act as signalling molecules, but excessive ROS from mitochondrial respiration or nutrient excess induces oxidative damage to mitochondrial DNA, proteins and lipids. β-cells have relatively low antioxidant defences, making them especially vulnerable to oxidative stress. Chronic nutrient overload (glucotoxicity and lipotoxicity) leads to mitochondrial overwork and ROS accumulation. In vitro studies show that under conditions of high glucose and/or fatty acids, β-cell apoptosis is driven by mitochondrial ROS: antioxidants can prevent β-cell apoptosis in this context, implying a causal role of mitochondria in the damage [1]. Disruption of the mitochondrial membrane potential under stress further impairs ATP synthesis, blunting GSIS. Moreover, ER stress and chronic misfolding of proinsulin also perturb mitochondrial redox balance. ER stress releases Ca²⁺ into the cytosol and generates additional ROS, overloading mitochondria and creating a vicious cycle of dysfunction [24]. Thus, mitochondrial dysfunction in β-cells, whether due to intrinsic defects or extrinsic stress, leads to defective insulin secretion and β-cell failure.

Mitochondrial morphology and turnover are tightly regulated to match metabolic needs. Mitochondria constantly undergo fission and fusion: fusion (mediated by Mitofusin 1 and 2 (MFN1/2) and Optic atrophy 1 (OPA1)) allows mixing of mitochondrial contents, whereas fission (mediated by Dynamin-related protein 1 (Drp1)) helps segregate damaged segments [25]. Under stress, β-cells tend to show fragmented mitochondria, reflecting enhanced fission or impaired fusion. This fragmentation affects mitochondrial function and can also act as a signal for initial general autophagy, allowing the cell to degrade damaged organelles and maintain homeostasis [26,27,28]. For example, exposure to high glucose or cytokines downregulates Mfn2 [29] and upregulates Drp1 in β-cells, promoting fragmentation and isolating damaged portions for autophagic removal. Fragmented mitochondria are often less efficient in ATP production [30,31,32].

While excessive fragmentation is indeed a hallmark of stressed β-cells, it is important to note that mitochondrial fission also serves essential physiological roles [33]. Recent evidence indicates that β-cell mitochondrial fission dynamically responds to metabolic changes such as hyperglycemia, starvation, or ATP synthase inhibition, highlighting that fragmentation can occur as a physiological adaptation rather than solely under stress [27]. Similarly, a study using β-cell-specific Drp1 knockout mice exhibited highly fused mitochondria, abnormal Ca²⁺ handling, and blunted GSIS despite normal oxygen consumption, indicating that fission is required for physiological responsiveness to glucose as well [34].

Mitophagy, the selective autophagy of mitochondria, is a critical quality-control mechanism in β-cells. Under normal conditions, basal mitophagy helps eliminate old or mildly dysfunctional mitochondria, maintaining a healthy pool [35,36,37]. Under stress, enhanced mitophagy clears severely damaged organelles. The canonical PINK1-Parkin pathway initiates mitophagy: PINK1 accumulates on depolarized mitochondria and recruits Parkin, which ubiquitinates outer membrane proteins to tag the organelle for autophagosome engulfment [38,39]. Further, β-cells exposed to diabetogenic cytokines (IL-1β, IFN-γ) upregulate mitophagy as a compensatory response to nitrosative/oxidative damage [33]. Mitophagy-deficient β-cells in the islets of Mt Keima mice accumulated fragmented, dysfunctional mitochondria and exhibited worsened hyperglycemia under stress. Further, overexpression of CLEC16A [40], a T1DM susceptibility gene and known mitophagy regulator, reduced β-cell apoptosis [33]. Thus, mitophagy is cytoprotective, removing injured mitochondria, preventing cytosolic release of proapoptotic factors, and sustaining ATP supply. Other studies have also previously implicated macroautophagy in the cytoprotection of β-cells [41,42].

A β-cell-specific ubiquitin-dependent complex plays a pivotal role in regulating mitophagy. The CLEC16A gene encodes an E3 ubiquitin ligase that partners with NRDP1 and the deubiquitinase USP8 to form a tripartite complex essential for mitochondrial quality control [40]. CLEC16A facilitates non-degradative ubiquitin tagging of itself and NRDP1, stabilizing the complex and creating a scaffold for the recruitment of mitophagy effectors such as Parkin [40]. Disruption of this complex, either pharmacologically, as with lenalidomide [43], or metabolically, as seen under glucolipotoxic stress [44,45], impairs mitophagy, diminishes mitochondrial respiration, and reduces insulin secretion. These findings link dysfunction of the CLEC16A-NRDP1-USP8 axis to β-cell failure and highlight the importance of ubiquitin signalling in maintaining mitochondrial integrity in diabetes.

In 2025, Walker et al. [46] uncovered a conserved mitochondrial retrograde signaling pathway that links mitochondrial quality control failure to loss of cellular maturity across metabolic tissues, including pancreatic β-cells. Using complementary genetic models targeting mitophagy (Clec16a knockout), mitochondrial genome integrity (Tfam knockout), and mitochondrial fusion (Mfn1/2 double knockout), they demonstrated that defects in these pathways converge on electron transport chain (ETC) and oxidative phosphorylation (OXPHOS) impairment, triggering activation of the integrated stress response (ISR). This was characterized by eIF2α phosphorylation, ATF4 stabilization, and induction of canonical ISR effectors such as Ddit3/CHOP, Atf3, Cebpβ, and Gdf15, independent of ER stress. In β-cells, this retrograde signaling led to downregulation of key identity genes (Ins2, MafA, Ucn3, Slc2a2/GLUT2) and induction of dedifferentiation markers (Aldh1a3, Cd81), without evidence of apoptosis or progenitor reversion. Mechanistically, early ETC/OXPHOS defects, not oxidative stress, were shown to initiate ISR activation, as restoration of NADH oxidation or treatment with the OXPHOS enhancer (–)-epicatechin suppressed ISR signaling. Single-nucleus ATAC-seq further revealed loss of chromatin accessibility at β-cell identity loci and enrichment of ATF4 and BACH2 motifs, suggesting chromatin remodeling as a downstream consequence of mitochondrial distress. Crucially, pharmacologic inhibition of the ISR with ISRIB restored β-cell mass, Ucn3 expression, and glucose tolerance, confirming that mitochondrial stress–induced dedifferentiation is reversible. Collectively, these findings position mitochondrial integrity as a primary determinant of β-cell identity and reveal the ISR as a key therapeutic target for preserving β-cell maturity under metabolic stress.

While these findings establish a retrograde signaling route by which mitochondrial dysfunction can reprogram β-cell fate, anterograde regulators (e.g., FOXO1, PGC1α, PRDM16, ERRγ, and PDX1) have been shown to influence mitochondrial metabolism and maturity. However, their effects are tissue-specific and do not uniformly regulate both identity and mitochondrial function [47,48,49,50,51]. For example, FOXO1 promotes β-cell identity but enhances brown adipose tissue (BAT) differentiation [52,53], while PGC1α is critical for BAT but dispensable in β-cells [49,54]. These limitations in tissue specificity and pathway overlap have made it difficult to establish whether mitochondrial regulation alone is sufficient to govern cellular maturity.

Overall, diabetic β-cells exhibit a downward spiral: nutrient and inflammatory stresses increase mitochondrial ROS and Ca²⁺ load, triggering fragmentation and mitophagy. If clearance is insufficient or signalling is defective, dysfunctional mitochondria accumulate, bioenergetics collapse, and apoptotic pathways are activated (e.g., cytochrome-c release, caspase-3). The net result is impaired GSIS and loss of β-cell mass, hallmarks of both T1DM and T2DM pathogenesis [55,56]. This is captured in Figure 1 below.

Nuclear transcription factors and genetic variants influence mitochondrial biogenesis and maintenance. First, transcription factor PDX1, a key β-cell differentiation gene, directly controls the mitochondrial transcription factor TFAM. TFAM is essential for mitochondrial DNA (mtDNA) maintenance and transcription of the 13 protein-coding genes of oxidative phosphorylation (OXPHOS). Pdx1 deficiency in β-cells (as in MODY4 or conditional knockout mice) reduces TFAM expression and leads to mitochondrial DNA depletion, impaired respiratory chain function, and defective insulin secretion [57]. Notably, restoring TFAM expression in PDX1-deficient β-cells rescues mtDNA copy number, OXPHOS activity, ATP production and GSIS, highlighting how PDX1–TFAM signalling tightly couples mitochondrial bioenergetics to the nuclear β-cell identity program. A parallel mechanism of β-cell–mitochondrial coordination is exerted through TFB1M, a mitochondrial rRNA methyltransferase that stabilizes the 12S rRNA component of the mitoribosome [58]. Although originally described as a transcription factor, TFB1M is now known to be critical for post-transcriptional modification of mitochondrial ribosomes and the fidelity of mitochondrial protein synthesis. Mice with β-cell-specific Tfb1m deletion (Tfb1m^β−/−) develop progressive insulin secretory failure and overt diabetes [59]. These islets exhibit markedly reduced levels of mitochondrial-encoded proteins, diminished ATP production, and impaired oxygen consumption [60,61]. Importantly, human carriers of a common TFB1M risk variant show reduced islet TFB1M expression and decreased insulin secretion [62], mirroring the mouse phenotype and providing direct evidence that even modest genetic variation in mitochondrial translation can predispose to β-cell failure and T2DM. Further, a study using ChIP-Seq profiling in human β-cell models (EndoC-βH1, primary islets, and hPSC-derived β-like cells) revealed that HNF4α target genes are enriched in pathways related to cytoskeletal remodelling and mitochondrial redox metabolism [63]. While actin dynamics and morphogenesis were common across cell types, insulin secretion and cAMP signalling were exclusively enriched in β-cells, suggesting HNF4α's mitochondrial effects are embedded within β-cell specific transcriptional programs.

Beyond the direct transcriptional control of mitochondrial genes, emerging evidence indicates that mitochondrial metabolites themselves exert epigenetic regulation of β-cell identity genes through metabolic-epigenetic crosstalk. Acetyl-CoA, generated primarily through mitochondrial pyruvate oxidation and citrate export to the cytosol, serves as the obligate substrate for histone acetyltransferases (HATs). In β-cells, glucose stimulation increases mitochondrial acetyl-CoA production, which is then transported to the nucleus where it drives histone H3 acetylation at key β-cell gene loci, including INS, PDX1, and MAFA [64]. This acetylation opens chromatin structure and enhances transcription of genes essential for β-cell function and identity. Pharmacological or genetic disruption of ATP citrate lyase (ACLY), the enzyme that converts cytosolic citrate to acetyl-CoA, reduces histone acetylation and impairs insulin gene expression in β-cells [65]. Similarly, α-ketoglutarate (α-KG), a tricarboxylic acid (TCA) cycle intermediate, functions as an essential cofactor for the TET family of DNA demethylases and Jumonji-domain containing (JMJD) histone demethylases. α-KG availability directly influences the activity of these dioxygenases: under conditions of glucose stimulation, elevated mitochondrial α-KG export supports TET-mediated DNA demethylation at insulin enhancer regions and JMJD3-mediated removal of repressive H3K27me3 marks at β-cell maturity genes [66,67]. Conversely, when mitochondrial function is impaired, such as in models of TFAM deficiency or chronic glucolipotoxicity, α-KG levels decline, leading to hypermethylation of CpG islands and accumulation of repressive histone marks that silence β-cell identity programs [68]. This metabolic-epigenetic axis creates a direct link between mitochondrial bioenergetics and chromatin state, positioning mitochondrial dysfunction as not only a bioenergetic failure but also an epigenetic reprogramming event that drives β-cell dedifferentiation. Future studies employing stable isotope tracing combined with ChIP-seq and bisulfite sequencing in β-cells under metabolic stress will be essential to map the full scope of metabolite-driven epigenetic remodeling in diabetes pathogenesis.

Beyond transcriptional regulation, the translational capacity of β-cell mitochondria has emerged as a key regulator of β-cell performance. Recent evidence suggests that expression of mitochondrial ribosomal protein genes (MRPs) is downregulated in human T2DM islets [69], implicating impaired mitochondrial protein synthesis in disease pathogenesis. In particular, the mitoribosomal subunit CRIF1 (also known as MRPL59) is partially reduced in diabetic islets. Mice with β-cell-specific haploinsufficiency of Crif1 (Crif1^β+/−) maintain normal glucose tolerance under basal conditions but exhibit selective loss of first-phase insulin secretion and accelerated islet failure when challenged by high-fat diet or aging [70]. Biochemically, CRIF1 is essential for inserting nascent mitochondrial-encoded peptides into the inner mitochondrial membrane. Its reduction leads to decreased OXPHOS complex assembly and ATP production, consistent with the insulin secretion defect observed [71,72].

Other β-cell-expressed genes with mitochondrial roles have emerged from genetic studies. For instance, CCDC66, LARS2, UCP2, and PPARGC1A are implicated in mitochondrial dynamics, biogenesis, and coupling efficiency. CCDC66, though originally characterized for its role in ciliary function, has been found to regulate mitochondrial morphology in metabolically active tissues, including islets, where its loss leads to aberrant cristae structure and decreased respiratory capacity. LARS2, which encodes mitochondrial leucyl-tRNA synthetase, is required for accurate mitochondrial translation. Mutations in LARS2 impair the incorporation of leucine into nascent mitochondrial proteins, thereby affecting the assembly and function of oxidative phosphorylation (OXPHOS) complexes. In β-cells, this can limit mitochondrial ATP production, leading to suboptimal insulin granule exocytosis [73,74]. UCP2 (uncoupling protein 2) modulates the proton gradient across the inner mitochondrial membrane, thus influencing the balance between ATP synthesis and heat generation. Although mild UCP2 expression can serve a protective antioxidant function, its overexpression has been shown to reduce ATP output and impair GSIS in rodent islets [75,76,77]. Notably, UCP2 levels are elevated in human T2DM islets, and polymorphisms in the UCP2 promoter region have been associated with altered T2DM risk and insulin secretion capacity in population studies [78,79,80]. Studies over the past decade indicate that UCP2 can function as a mitochondrial metabolite transporter, facilitating the export of C4 metabolites such as malate and aspartate. This activity influences redox balance rather than directly dissipating the proton gradient, highlighting a more nuanced role in fine-tuning substrate flow and redox coupling in β-cell metabolism [81].

Despite these mechanistic insights, the physiological role of UCP2 in β-cells remains controversial, and several unresolved questions limit therapeutic translation. First, does UCP2 primarily function as a proton uncoupler, a metabolite transporter, or both, and does this depend on metabolic context? While biochemical studies demonstrate C4 metabolite transport activity [81], electrophysiological recordings in reconstituted systems also confirm proton leak conductance [82]. It remains unclear whether these activities occur simultaneously or are regulated by post-translational modifications such as glutathionylation, which has been shown to toggle UCP2 between transport and uncoupling modes [83]. Second, why do Ucp2 knockout studies yield inconsistent results across different models? While Ucp2^−/− mice on an ob/ob background show improved insulin secretion and glycemic control [75], other studies report unchanged or even impaired β-cell function in Ucp2 knockout mice on different genetic backgrounds or under different dietary conditions [84,85]. These discrepancies may reflect compensatory upregulation of other UCPs (UCP1, UCP3) or genetic background-specific differences in baseline mitochondrial coupling efficiency. Third, is the elevated UCP2 expression observed in human T2DM islets a cause or a consequence of metabolic dysfunction? Longitudinal studies tracking UCP2 levels during progression from prediabetes to overt diabetes are lacking, making it impossible to determine whether UCP2 upregulation is a primary driver of β-cell failure or a compensatory (albeit insufficient) antioxidant response to chronic oxidative stress [78,86]. Fourth, can UCP2 inhibition be safely targeted therapeutically without disrupting its cytoprotective antioxidant functions? Genipin, a natural UCP2 inhibitor, enhances insulin secretion in rodent models but has not been systematically evaluated in human islets or in vivo human studies, and its off-target effects on other mitochondrial proteins remain poorly characterized [87,88]. Resolving these controversies will require conditional, inducible β-cell-specific Ucp2 knockout models combined with real-time imaging of both proton gradient dissipation and metabolite flux, as well as human islet studies correlating UCP2 expression with functional outcomes across a range of metabolic states. Complementing these findings, transcriptional coactivator PGC-1α regulates both NRF1 and NRF2, promoting transcription of nuclear-encoded genes required for ETC assembly [89]. In β-cells, PGC-1α must be tightly regulated. While it is crucial for adapting to increased energetic demands in other tissues, its overexpression in rodent islets suppresses insulin secretion and disturbs mitochondrial coupling efficiency. Genetic variants in PPARGC1A, such as the Gly482Ser polymorphism [90,91], have been linked to altered islet function, reduced insulin secretion, and increased T2DM risk in multiple populations [91,92,93,94,95,96,97,98]. While each factor contributes incrementally, they are best discussed together because their coordinated regulation is necessary for maintaining mitochondrial output during insulin secretion. Disruption at any point, whether through defective mitochondrial protein synthesis (LARS2), altered membrane coupling (UCP2), or dysregulated biogenesis (PPARGC1A), can weaken β-cell energetic capacity. Table 1 summarizes all known transcriptional regulators linking mitochondrial function, or lack thereof, to β-cell physiology, including but not limited to those discussed above. Meanwhile, Figure 2 summarizes the structural features of key transcriptional regulators.

Chronic nutrient overload in diabetes drives β-cell mitochondrial stress. High glucose and free fatty acids increase substrate flux into mitochondria, hyperpolarizing the respiratory chain and generating excess superoxide [112,113]. Elevated ROS damages mitochondrial DNA, respiratory chain proteins, and lipid membranes [114]. In vitro, rodent β-cell lines (such as MIN6, INS-1E) and human islets exposed to glucolipotoxic conditions (glucose and palmitate) show elevated mitochondrial superoxide and H₂O₂ production [115,116,117], triggering apoptosis that is preventable by antioxidants [1]. In vivo, db/db and KK-Ay mice models show oxidative inactivation of key metabolic enzymes (e.g., aconitase, PDH) in β-cells, undermining ATP production [118,119].

Persistent oxidative stress also activates stress kinases (JNK, p38) that inhibit insulin gene expression and sensitize β-cells to apoptosis [120,121,122]. In a mouse model with β-cell-specific JNK activation (MKK7D), JNK impaired GSIS without affecting islet morphology or insulin content, indicating disrupted insulin signalling rather than cell loss [123]. However, this signalling defect included impaired AKT phosphorylation and PDX1 mislocalization, both essential for insulin gene expression, supporting JNK's inhibitory role [124,125]. While JNK activation alone did not induce apoptosis, it sensitized β-cells to dysfunction, suggesting that in the presence of oxidative stress or inflammatory cues, JNK could contribute to β-cell apoptosis [123,126,127].

Under glucotoxic conditions, mitochondrial uncoupling often increases (via upregulation of UCP2 and related proteins), further reducing ATP yield and fuelling ROS as electrons leak [80]. This mechanism was confirmed in Ucp2-deficient mice, which exhibited higher islet ATP levels and potentiated insulin secretion in response to glucose. In another ob/ob mouse model, where UCP2 is upregulated, deletion of UCP2 restored first-phase insulin secretion and improved glycaemic control [75]. Reduced ATP generation means impaired closure of K_ATP channels and a weaker insulin response to glucose [128]. For example, in the β-cell Tfam KO model, glucose failed to depolarize mitochondria and raise ATP or Ca²⁺ levels [99] demonstrating how oxidative phosphorylation collapse translates to secretion failure [129]. Furthermore, oxidative stress can damage mtDNA, leading to mutations and deletions that further compromise mitochondrial function [130]. One study in diabetic human islets found evidence of mtDNA depletion and mutation accumulation with age, suggesting a feed-forward decline in mitochondrial capacity [131]. Though such mitochondrial DNA defects are not primary causes of common diabetes, they exacerbate β-cell dysfunction under chronic stress. Apart from nutrients, inflammatory cytokines (as in T1DM or late-stage T2DM) cause nitric oxide (NO) and ROS production in β-cells. NO reacts with superoxide to form peroxynitrite, which irreversibly damages mitochondrial electron carriers [132,133]. Indeed, proinflammatory cytokines induce β-cell mitochondrial fragmentation, loss of membrane potential, and mitophagy [33]. If the resulting damaged mitochondria are not removed, bioenergetic failure and cell death ensue.

New facets of the β-cell oxidative response should be noted. The NRF2 transcription factor is now recognized as a master antioxidant regulator in β-cells. Upon oxidative stress, NRF2 induces detoxifying enzymes to mitigate ROS damage [134]. Conversely, mitochondrial redox enzymes are themselves regulated by metabolic sensors. As noted above, SIRT3 activation increases MnSOD activity and lowers intracellular ROS. β-cells lacking SIRT3 have higher oxidative stress and poorer survival under glucolipotoxic conditions [135]. These cross-talk pathways suggest that antioxidant defenses are coupled to nutrient sensing. Emerging death pathways further expand this view. For instance, ferroptosis, an iron-dependent form of lipid peroxidation, can occur in β-cells when GPX4 is deficient, linking lipid oxidative stress to cell loss [136]. Incorporating NRF2 and sirtuin-mediated antioxidant control thus complements the classical view of ROS in β-cell failure.

The downstream effects of mitochondrial dysfunction on β-cell physiology are profound. The immediate impact is on GSIS [137]. Loss of mitochondrial ATP production precludes the closure of ATP-sensitive K⁺ channels, preventing membrane depolarization and Ca²⁺ influx that trigger insulin granule exocytosis [138]. This mechanism is well-illustrated by the Tfam knockout model: mutant β-cells showed reduced mitochondrial membrane hyperpolarization, blunted Ca²⁺ signalling and markedly lower insulin release to glucose [99]. Likewise, Crif1^β+/− mice lost the rapid first-phase insulin secretion that is normally triggered by the initial glucose rise [70]. In human T2DM islets, transcriptomic analyses reveal downregulation of virtually all mitochondrial gene sets (OXPHOS, translation, etc.) relative to non-diabetic islets, consistent with a global mitochondrial deficit that translates into secretion failure [139]. Beyond ATP, mitochondria generate coupling factors (e.g., NADPH, GTP, malonyl-CoA) that amplify GSIS through alternative signalling pathways [140]. For example, mitochondrial NADPH is necessary for the amplification pathway of insulin secretion and for sustaining ER redox balance [141]. Loss of these factors further impairs sustained insulin output during chronic glucose exposure. Even intermittent mitochondrial uncoupling (e.g., via UCP2) can lower ATP and dissipate metabolic signals, shifting β-cells towards a non-secretory state [142]. As mitochondrial impairment worsens, β-cells undergo apoptosis or necrosis.

Chronic oxidative and nitrosative stress triggers mitochondrial permeability transition, characterized by cytochrome-c release and caspase activation [143]. Cytokine-treated β-cells demonstrate mitochondrial fragmentation and outer membrane permeabilization if mitophagy is blocked [33,144]. At the tissue level, this cell death reduces β-cell mass. Simultaneously, surviving β-cells often dedifferentiate or enter a senescent state, further reducing functional β-cell mass [145]. Notably, β-cells in long-standing diabetes show hallmarks of mitochondrial aging with fewer and damaged mitochondria, enlarged cristae and DNA deletions, analogous to aging neurons [146,147]. In non-human primate models, developmental exposure to maternal Western-style diet increased mitochondrial fragmentation in β-cells without altering maturity marker expression, implying that nutrient-driven morphological changes may occur independently of β-cell identity [148]. This aligns with observations in diabetes, where mitochondrial fragmentation and swelling often accompany β-cell failure. Inflammation and immune cell infiltration exacerbate this loss. Inflammatory mediators in both T1DM and T2DM impose additional stress on β-cell mitochondria (captured in Table 2 below). For example, sustained IL-1β exposure leads to iNOS induction and excess NO, which blocks complex IV and generates peroxynitrite in mitochondria [149]. This reinforces β-cell destruction in autoimmune diabetes. In T2DM, islet macrophage activation and NLRP3 inflammasome signals amplify mitochondrial stress [150,151]. Thus, mitochondrial dysfunction not only impairs insulin secretion but also sensitizes β-cells to immune and metabolic insults, creating a feed-forward loop of β-cell failure [24]. Table 2 summarizes examples of mitochondrial processes and proteins disrupted in diabetic β-cells, and their effects on insulin secretion, while Figure 3 shows mitochondrial events during insulin resistance, prediabetes, and overt diabetes. In general, impairments in mitochondrial dynamics or mitophagy lead to the accumulation of defective organelles, increased oxidative stress and cell death signals.

Given the central role of mitochondria in β-cell health, strategies that protect or restore mitochondrial function hold therapeutic promise. Pharmacological and lifestyle interventions may target mitochondrial bioenergetics, redox balance and quality control in β-cells. Metformin is an established clinical therapy, considered by the majority as the first-line treatment for non-insulin dependent T2DM, with extensive evidence gathered from human studies across the years [154]. Further, in patients with new-onset T2DM, metformin has been shown to improve mitochondrial fitness in peripheral blood cells (PBCs), increase markers of mitophagy and reduce mitochondrial ROS and membrane polarization [155]. Although metformin's primary site of action is the liver, these findings suggest that it may also exert beneficial effects on β-cell mitochondrial quality. Metformin activates AMP-activated kinase (AMPK) by inducing subtle changes in the ATP/ADP and ATP/AMP ratios, without significantly changing the total ATP levels [156]. Metformin acts primarily by activating AMPK in the liver at therapeutic doses (~40–80 μM in the portal vein, or ~50 mg/kg/day in high-fat diet (HFD)-fed mice). This activation suppresses gluconeogenesis, reduces hepatic glucose production and improves insulin sensitivity. In HFD-fed C57BL/6J mice, these doses of metformin enhance mitochondrial function by increasing complex I activity, ATP production, mitochondrial membrane potential, and by promoting mitochondrial fission. This supports mitophagy, maintains mitochondrial quality and increases nutrient oxidation. These effects are AMPK-dependent, in liver-specific AMPKα1/2 KO mice, metformin fails to improve glucose control or stimulate mitochondrial function. By contrast, suprapharmacological doses (≥500–1000 μM or 150–500 mg/kg/day), often used in in vitro studies or animal models, can suppress mitochondrial respiration by depleting cellular ADP levels, but this is not relevant to metformin's clinical mechanism of action [1,157,158]. While hepatic AMPK activation is central to metformin's glucose-lowering effects, AMPK also plays a critical role in pancreatic β-cells. To better understand this, researchers investigated the physiological consequences of completely ablating both catalytic isoforms of AMPK in β-cells [159]. Interestingly, in isolated islets from βAMPK double knockout (βAMPK.dKO) mice, GSIS was paradoxically enhanced. This was associated with increased K_ATP channel activity at low glucose concentrations, possibly due to altered subunit trafficking and more efficient granule translocation to the plasma membrane, likely resulting from the loss of AMPK-mediated inhibition via kinesin light chain phosphorylation [160,161,162]. However, in vivo studies revealed a different outcome: βAMPK.dKO mice exhibited markedly impaired insulin secretion following intraperitoneal glucose injection, leading to defective glucose homeostasis and hyperglycaemia [159]. This discrepancy highlights the context-dependent nature of AMPK signalling, suggesting that while β-cell-intrinsic AMPK constrains insulin secretion in vitro, its absence disrupts coordinated endocrine responses in vivo. Importantly, these defects were not due to reduced β-cell mass but may involve disrupted neuroendocrine regulation, as RIP.Cre-expressing neurons may influence β-cell function. Additionally, activation of AMPK specifically in β-cells (via βAMPK.CA transgenic mice) suppressed GSIS, consistent with pharmacological studies using metformin [1,157,163]. Beyond AMPK activation, other mitochondria-targeted interventions have emerged. Activating sirtuin–NAD⁺ axes is one promising strategy. Pharmacologic SIRT1/3 activators or NAD⁺ precursors (NR, NMN) have been shown to improve mitochondrial metabolism and β-cell function in preclinical diabetes models [164]. Some NAD⁺ boosters, such as NR and NMN, have entered early-phase clinical trials in humans for metabolic disorders, whereas specific SIRT1/3 activators remain at the preclinical stage [165,166,167,168,169]. Beyond that, antimiR therapy against miR-146a-5p rescued β-cell mitochondrial respiration and insulin secretion in vitro. These miRNA-based approaches have been investigated in non-human trials and have been showing positive results [170,171]. Enhancing mitophagy represents another approach [172]. In a β-cell Clec16a-knockout mouse with excessive Parkin activity and defective mitophagy, overexpressing the E3 ligase RNF41 restored Parkin balance and normalized mitochondrial morphology and insulin release [134]. Mitophagy modulation remains in its preclinical stages, with compounds such as urolithin A now advancing into early human studies for mitochondrial diseases [173,174]. Finally, novel cell-based therapies are on the horizon. Coculturing islets with mesenchymal stem cells markedly improved graft performance in vitro, partly because mesenchymal stem cells (MSCs) donated functional mitochondria to β-cells [175]. Indeed, co-culture of murine and human islets with adipose-derived MSCs improves β-cell survival, viability, and glucose-stimulated insulin secretion in vitro, and preconditioning via co-culture enhances engraftment and glycemic control in vivo after transplantation of marginal islet masses [176]. Mechanistically, these effects are attributed to mitochondrial transfer, paracrine secretion of trophic factors, gap-junctional contacts, and anti-inflammatory signaling, although direct visualization of mitochondrial donation in intact in vivo grafts remains limited [177,178,179]. Mitochondrial transfer and MSC-based approaches are currently in early preclinical stages, however MSC transplantation has been previously tested in clinical trials for diabetes [180,181,182,183]. Collectively, sirtuin/NAD⁺ boosters, miRNA inhibitors, mitophagy modulators, and mitochondrial transplantation illustrate the therapeutic breadth beyond classic antioxidants or UPR-targeting drugs.

Emerging evidence also implicates GLP-1 signalling in mitochondrial biogenesis and antioxidant defences in β-cells. GLP-1 increases mitochondrial calcium and ATP in MIN6 β-cells via cAMP-mediated activation of IP₃ and ryanodine receptors. PKA mediates the IP₃ pathway, while cAMP-GEFII regulates the ryanodine receptor pathway [184]. GLP-1 and exendin-4 upregulate PGC-1α, increasing mitochondrial mass, membrane potential and oxygen consumption [185]. Once upregulated, PGC-1α activates both NRF1 and NRF2, promoting transcription of nuclear-encoded genes required for ETC assembly [89]. That said, impaired eIF2α phosphorylation disrupts ER proteostasis, leading to proinsulin accumulation and ER stress that contribute to β-cell dysfunction and death under metabolic stress. This ER stress exacerbates mitochondrial dysfunction by impairing mitochondrial biogenesis and ATP production, reducing the β-cell energy supply needed for insulin secretion. GLP-1R agonists mitigate ER stress through PKA-dependent modulation of the UPR, restoring eIF2α phosphorylation balance, which supports mitochondrial function by enhancing ATP synthesis and biogenesis [186,187]. Mitophagy activators and mild mitochondrial uncouplers both offer promising strategies to protect β-cells under diabetogenic stress [173,188,189,190,191]. Urolithin A, an activator, promotes Parkin-dependent mitophagy, facilitating the selective clearance of damaged mitochondria and maintaining mitochondrial integrity via CLEC16A-NRDP1-USP8-mediated signalling [173,190]. Mild uncouplers (e.g., niclosamide ethanolamine or BAM15) transiently depolarize mitochondrial membrane potential (Δψm), triggering mitophagy without causing bioenergetic collapse [192,193]. This controlled depolarization regulates mitochondrial turnover, reduces ROS and preserves β-cell function. Overall, therapeutic focus on β-cell mitochondria is promising but challenging. Table 3 summarizes examples of mitochondria-targeted interventions and their rationale.

Despite significant advances in our understanding of mitochondrial dysfunction in β-cells, several major gaps remain that offer compelling directions for future research and therapeutic development. Addressing these gaps will not only offer new mechanistic insight but may also redefine how we preserve β-cell health across the spectrum of diabetes.