Unlocking pancreatic metabolic memory: Can early interventions reverse obesity and block diabetes before it strikes?

Abstract

This article synthesizes current evidence to elucidate the role of pancreatic macrophage metabolic memory in linking obesity to type 2 diabetes. Through an integrated analysis of recent studies, we describe how metabolic stress reprograms pancreatic macrophages toward a spectrum of inflammatory phenotypes and establishes persistent memory via epigenetic and exosomal pathways. This memory sustains pro-inflammatory signaling even after metabolic insults cease, leading to impaired β-cell function and dedifferentiation through altered transcriptional regulation and paracrine communication. We conclude that targeting this macrophage memory axis offers a promising therapeutic strategy, with interventions such as exercise, natural products, and epigenetic modulators showing potential to reverse maladaptive polarization. Future research using single-cell omics will be essential to decode macrophage heterogeneity and advance memory-targeted therapies. During the transition from obesity to type 2 diabetes, the islets are often accompanied by persistent inflammation. The metabolic memory formed by pancreatic macrophages may be the core cell-intrinsic mechanism underlying the long-term maintenance of this phenomenon. That is, macrophages, through epigenetic reprogramming, “remember” prior metabolic stress. Even after the external stimuli subside, their pro-inflammatory phenotype continues to be sustained, thereby driving the macroscopically observed state of persistent inflammation. These two phenomena constitute a causal relationship. This concept is distinct from classical “trained immunity”, which refers to the enhanced response of immune cells to a heterologous secondary stimulus, focusing on the boosting of defensive functions. This study primarily utilized the PubMed database for retrieval, employing keywords such as “pancreas”, “metabolic memory” and “macrophage” for subject term searches. It screened for mechanism-based articles and review papers concerning metabolic memory in pancreatic macrophages.

Key Words: Diabetes; Obesity; Metabolic memory; Pancreas

Core Tip: Sustained exposure of pancreatic macrophages to metabolic stress (e.g., obesity, type 2 diabetes) triggers epigenetic reprogramming, leading to the establishment of a persistent “metabolic memory”. This memory maintains chronic local inflammation in the pancreas even after external stimuli subside, resulting in continuous deterioration of β-cell function and thereby driving the progression of type 2 diabetes. Targeting this maladaptive memory through lifestyle interventions or epigenetic modulators represents a highly promising early intervention strategy to halt the disease progression.

INTRODUCTION

The pathological legacy of obesity: Metabolic memory as a driver of pancreatic dysfunction

Numerous epidemiological studies have confirmed that obesity is a central risk factor for the development of type 2 diabetes (T2DM). Its pathological progression originates from obesity-induced insulin resistance and pancreatic dysfunction[1]. On one hand, factors such as adipose tissue dysfunction, chronic inflammation, lipotoxicity, and ectopic fat deposition induce peripheral insulin resistance; on the other hand, insufficient compensatory insulin secretion leads to pancreatic dysfunction, ultimately resulting in the loss of glycemic control and progression to clinical T2DM[2]. Notably, even after weight control is achieved, impairments induced by obesity may persist or even exacerbate.

This phenomenon indicates that the systemic metabolic disturbances and chronic inflammatory states caused by obesity possess a significant pathological legacy effect, termed “metabolic memory”[3]. The core mechanism lies in the fact that metabolic stress during the obese state can induce persistent alterations within pancreatic cells, such as epigenetic reprogramming and mitochondrial impairment, thereby driving the long-term deterioration of pancreatic function[4]. Therefore, pancreatic metabolic memory represents a critical pathological link connecting obesity to the pathogenesis and complications of T2DM, providing a novel theoretical perspective for understanding the long-term metabolic hazards of obesity.

Pancreatic macrophages: From innate immune surveillance to metabolic memory regulation

Under physiological conditions, pancreatic macrophages perform innate immune surveillance functions. They clear apoptotic β-cell debris via phagocytosis, preventing autoimmune responses and maintaining a low-inflammatory state within the islets. Simultaneously, they indirectly regulate β-cell function by secreting cytokines that promote insulin secretion[5]. Under pathological conditions such as obesity or T2DM, pancreatic macrophages play a central role in metabolic memory. Signals like lipotoxicity and hyperglycemia drive their polarization toward a pro-inflammatory (M1-like) phenotype, leading to the secretion of cytokines that impair β-cell function and exacerbate insulin resistance. Coupled with the slow turnover of macrophages, this results in a persistent and largely irreversible inflammatory response, rendering them key cellular carriers and regulators of metabolic memory[6].

This functional reprogramming represents a stress response involving transcriptional, epigenetic, and metabolic rewiring following metabolic disturbance[7]. Coupled with the slow turnover kinetics of pancreatic macrophages[8], the response becomes persistent and largely irreversible, continuously exacerbating inflammation and tissue damage. Consequently, pancreatic macrophages serve as both critical cellular carriers and key regulators of metabolic memory.

Rationale for focusing on pancreatic macrophages in metabolic memory research

Due to their central roles in sensing metabolic signals, regulating islet inflammation, and modulating β-cell function, pancreatic macrophages have become an indispensable focus in metabolic memory research. Mechanistically, as the most abundant immune cells surrounding β-cells[8], pancreatic macrophages rapidly detect metabolic disturbances through specific pathways. They subsequently secrete cytokines such as interleukin-1 beta (IL-1β) and tumor necrosis factor-alpha (TNF-α) to precisely regulate β-cell dedifferentiation, dysfunction, or apoptosis, thereby governing the evolution of the islet inflammatory microenvironment[6]. From a therapeutic perspective, interventions targeting pancreatic macrophages offer unique advantages. Compared to systemic anti-inflammatory approaches, gene-editing-based cellular reprogramming technologies enable precise local modulation of the disease microenvironment. This strategy holds promise for more accurately reversing islet inflammatory memory and protecting β-cells while maintaining systemic immune homeostasis[9].

PANCREATIC MACROPHAGE HETEROGENEITY AND FUNCTIONAL PLASTICITY IN OBESITY-ASSOCIATED METABOLIC STRESS

Phenotypic spectrum of pancreatic macrophages: M1/M2 polarization and beyond

In classical immunology, the most commonly used classification method for macrophages is to categorize them into pro-inflammatory M1 type and anti-inflammatory M2 type. However, this dichotomy only reveals the existence of two distinct types, M1 or M2, and does not encompass intermediate states or other potential phenotypes[10]. Although this classification is classic and serves as an important framework for studying obesity and diabetes, its binary nature also limits the investigation into other macrophage phenotypes within the pancreas and their interactions with other cell types.

Furthermore, another study found no significant change in the M1/M2 ratio of macrophages within the islets of obese mice, suggesting that in T2DM, islet-resident macrophages may undergo a more refined phenotypic shift beyond the simple M1/M2 dichotomy[11]. For a more comprehensive summary of pancreatic macrophage subtypes, please refer to Table 1.

Table 1 Macrophage subtypes and their distribution characteristics.

Macrophage subtype names	Specific distribution in the pancreas	Surface markers and signature molecules	Species	Functional characteristics	Ref.
M1 macrophages	Pancreatic acinar cells	Gpr18, Fpr2	Mouse	pro-inflammatory	[70,71]
M2a macrophages	Pancreatic acinar cells	IL-4, IL-13	Mouse	Mediate tissue repair and wound healing
M2b macrophages	bone marrow-derived Mφ	LPS, IL-1β	Mouse	Modulate immune responses	[72,73]
M2c macrophages	bone marrow-derived Mφ	IL-10	Mouse	Anti-inflammatory
M2d macrophages	bone marrow-derived Mφ	IL-10, VEGF	Mouse	Tumor-associated macrophages	[74]
F4/80+ CD11c+ macrophages	Near the blood vessels within the islets	F4/80+ CD11c+	Mouse	Captured secretory granules and antigen presentation from β-cells	[75]
F4/80+ CD11b+ CD206+ macrophages	Pancreatic stromal macrophages	IL-10, Fizz1	Mouse	Anti-inflammatory	[76]
F4/80+ CD11b+ CD206- macrophages	Peripheral duct cells of the exocrine portion of the pancreas	Ym1, Fizz1	Mouse	Anti-inflammatory

Gpr18: G-protein coupled receptor 18; Fpr2: Formyl peptide receptor 2; IL: Interleukin; LPS: Lipopolysaccharide; VEGF: Vascular endothelial growth factor; Fizz1: Found in inflammatory zone 1; Ym1: Y-macrophage 1.

Metabolic stress (hyperlipidemia/hyperglycemia)-induced reprogramming of pancreatic macrophages

Macrophage reprogramming refers to the dynamic and plastic transformation of a macrophage’s phenotype, function, and metabolic state in response to local microenvironmental signals. This process enables macrophages to adapt to diverse physiological or pathological conditions. A sustained pathological microenvironment (e.g., high-fat, high-sugar) can induce long-term, stable alterations in macrophages, which itself constitutes a form of pathological metabolic memory[12]. Within the pancreas, this plasticity is crucial for maintaining islet homeostasis and responding to tissue injury.

Under conditions such as T2DM, the pancreas is subjected to persistent metabolic stress. Core stressors like hyperglycemia and hyperlipidemia act synergistically to drive the polarization of islet macrophages towards a pro-inflammatory (M1-like) phenotype. Not only pancreatic resident macrophages, but also a substantial number of macrophages are recruited to adipose tissue in conditions such as T2DM, where they subsequently undergo polarization. These macrophages promote inflammation while accelerating glucose production, reducing insulin secretion, and increasing insulin resistance. This process, in turn, further drives the recruitment and polarization of adipose tissue macrophages (ATMs), thereby promoting the progression of T2DM[13]. This imbalanced reprogramming breaks pancreatic immune homeostasis. The resulting chronic, low-grade inflammation directly impairs islet β-cell function, exacerbates insulin resistance, and promotes pathological changes such as fibrosis. Consequently, metabolic stress-induced macrophage reprogramming represents a central link connecting metabolic dysregulation with pancreatic immune inflammation.

Crosstalk between pancreatic macrophages and β-cells: A bidirectional regulatory loop

Pancreatic macrophages and β-cells are closely linked from the developmental stage onwards. In osteopetrotic (op/op) mouse models deficient in macrophage colony-stimulating factor, mice exhibit significantly reduced β-cell numbers, proliferation defects, and abnormal islet morphology, indicating that macrophages are crucial for islet development and insulin secretion[14].

Under metabolic stress induced by a high-fat diet (HFD), macrophages rapidly accumulate within the islets, accompanied by increased β-cell proliferation[15]. In long-term HFD-fed diabetic mouse models, the number of islet-resident macrophages increases significantly. These islet-resident macrophages are anatomically closely opposed to β-cells, providing a structural basis for functional interactions[16,17]. Their interaction is primarily mediated through direct contact and soluble factors[18]. Under diabetic conditions, the expression of the inhibitory receptor Siglec-7 on β-cell surfaces is downregulated, impairing its negative regulatory capacity mediated via the immunoreceptor tyrosine-based inhibition motif domain and thereby exacerbating inflammation. Conversely, overexpression of Siglec-7 can alleviate β-cell inflammation and help maintain function[19]. These findings elucidate the central role of intra-islet cellular communication in the pathology of diabetes.

MECHANISMS OF METABOLIC MEMORY FORMATION IN PANCREATIC MACROPHAGES AND THEIR PATHWAYS MEDIATING β-CELL DYSFUNCTION

Memory establishment: Intrinsic reprogramming of pancreatic macrophages

Environmental stimuli can induce epigenetic and functional reprogramming in pancreatic resident macrophages, forming a “memory” state that enables them to mount faster and stronger responses upon re-encountering the same or similar stimuli. This process plays a key role in the initiation and progression of metabolic diseases. Research indicates that this reprogramming is environment-dependent. For example, following pancreatic β-cell death, apoptotic cells can reprogram macrophages, driving their transformation into a repair-oriented phenotype characterized by the expression of insulin-like growth factor-1[20]. In pancreatic cancer, this microenvironment-driven immunometabolic reprogramming mechanism similarly exists. In pancreatic ductal adenocarcinoma, tumor cells can directly interact with M1-like macrophages and reprogram them epigenetically, metabolically, and functionally through mechanisms such as DNA methylation. This leads to a shift from the anti-tumor M1-like phenotype to a pro-tumor M2-like phenotype, thereby promoting cancer development[21].

This phenomenon, mediated by epigenetic and metabolic reprogramming and leading to persistent functional alterations in immune cells, is termed “trained immunity”. Studies have shown that in mice switched from a HFD back to a normal diet, while systemic cytokine levels induced by the diet return to normal within four weeks, the innate immune response of myeloid cells retains the enhanced efficacy characteristic of the HFD period and exhibits qualitative changes[22]. After metabolic normalization, the clearance of lipid accumulation within macrophages often lags. These retained lipids are not merely inert energy storage molecules but can act as signaling molecules or inflammasome activators, continuously disrupting normal macrophage function and maintaining their pro-inflammatory state[23]. The polarization of pancreatic macrophages exhibits plasticity and can be reshaped by factors such as epigenetic regulation, inflammation, and the tissue microenvironment[24]. Non-coding RNAs play a significant regulatory role in this process. For instance, the long non-coding RNA (lncRNA) M2-type macrophage peroxidase can inhibit pancreatic inflammation caused by M1 macrophage polarization[25].

Memory effects: Direct attack on β-cells by macrophage persistent secretory profiles

In obesity and T2DM, the function of islet-resident macrophages undergoes a fundamental shift: From homeostatic regulators that maintain β-cell integrity and support physiological insulin release to key drivers of chronic islet inflammation[26]. As illustrated in Figure 1, this phenotypic transition establishes a persistent inflammatory microenvironment that forms the core component of pancreatic metabolic memory.

Figure 1 Mechanisms of macrophage-driven β-cell impairment in obesity/T2DM islets. Activated macrophages release interleukin-1 β (inducible nitric oxide synthase/nitric oxide axis), galectin-3 (Ca²⁺ dysregulation), prostaglandin E2 (cyclic adenosine monophosphate reduction), and exosomal miR-212-5p (sirtuin 2 suppression), and form tunneling nanotubes to deplete insulin granules, collectively impairing Glucose-stimulated insulin secretion and β-cell viability. (by figdraw.com). IL-1β: Interleukin-1β; iNOS: Inducible nitric oxide synthase; NO: Nitric oxide; Gal3: Galectin-3; PGE2: Prostaglandin E2; SIRT2: Sirtuin 2; GSIS: Glucose-stimulated insulin secretion.

Key pro-inflammatory mediators released by activated M1-like macrophages constitute the first line of direct attack. Cytokines such as IL-1β (a master regulator) and emerging factors like galectin-3 are crucial. IL-1β activates the nuclear factor kappa B and c-Jun N-terminal kinase signaling pathways within β-cells, inducing the expression of inducible nitric oxide (NO) synthase and excessive NO production. This disrupts mitochondrial adenosine triphosphate (ATP) synthesis and insulin exocytosis[27]. Galectin-3, secreted primarily by M1-like macrophages, accumulates in obese islets and impairs the calcium transients required for effective glucose-stimulated insulin secretion (GSIS)[28]. Lipid mediators represent another axis of immunometabolic assault: Elevated saturated free fatty acids in the obese islet microenvironment promote macrophage production of bioactive lipids such as prostaglandin E2. This exacerbates β-cell lipotoxicity, reduces cyclic adenosine monophosphate levels, and impairs first-phase insulin secretion[29].

Complex forms of intercellular communication further amplify this assault. extracellular vesicles, particularly exosomes derived from M1 macrophages, can deliver miRNAs such as miR-212-5p to β-cells. This downregulates sirtuin 2 and inhibits the protein kinase B/glycogen synthase kinase-3β/β-catenin pathway, which is essential for insulin synthesis and secretion[30]. Furthermore, lncRNAs at the intrinsic β-cell level also contribute to this destructive network. For example, lncRNA βlinc2, which is upregulated in islets of T2DM models, shows expression levels positively correlated with body weight, blood glucose, and insulin resistance. Conversely, downregulated lncRNA βlinc3 is negatively correlated with body weight and blood glucose. Overexpression of βlinc2 or knockdown of βlinc3 can increase β-cell apoptosis[31].

In summary, these soluble cytokines, lipid mediators, vesicular carriers, physical contacts, and regulated intrinsic β-cell lncRNAs, collectively produced by memory macrophages, form a self-reinforcing network of attack. This network persistently damages β-cell function and drives the long-term pathological progression of T2DM.

Memory transmission: Epigenetic cross-regulation between macrophages and β-cells

Islet-associated macrophages (IAMs) acquire specific chromatin alterations, most notably persistent histone 3 lysine 4 trimethylation at promoters of proinflammatory genes, consistent with the paradigm of trained immunity. This macrophage-intrinsic memory governs the duration and intensity of pathological cues transmitted to neighboring β-cells, thereby extending dysfunction even after acute inflammatory triggers subside.

The metabolic-epigenetic axis centered on the histone demethylase lysine demethylase 5A provides a key mechanistic link between macrophage chromatin state and islet endocrine function. lysine demethylase 5A regulates the balance between glycolysis and oxidative phosphorylation in IAMs, and disruption of this axis can lock macrophages into a metabolically constrained state that imposes a persistent brake on GSIS[32]. Furthermore, efferocytosis of apoptotic β-cells constitutes a pivotal encoding mechanism for this memory. In obesity, chronic efferocytic burden and the resulting sustained transforming growth factor β signaling can paradoxically impair GSIS[33]. Simultaneously, reprogrammed M1-like macrophages secrete exosomes enriched in specific miRNAs. Upon uptake by β-cells, these miRNAs silence key transcriptional regulators, thereby imprinting macrophage inflammatory memory onto β-cells through a non-genetic yet heritable mechanism that perpetuates dysfunction[34]. The key mechanisms underlying pancreatic macrophage metabolic memory are summarized in Table 2.

Table 2 Key mechanisms and outcomes of pancreatic macrophage-mediated metabolic memory.

Mechanism pathway	Triggering factors	Macrophage state	Key mediators	β-cell outcome	Evidence model or study type	Ref.
Epigenetic modifications in pancreatic macrophages	High glucose, palmitic acid, obesity, diabetic complications (e.g., ischemia-reperfusion)	Impaired M2 polarization; enhanced pro-inflammatory M1-like phenotype; increased inflammatory signaling and phagocytic activity; activation of apoptosis pathways	DNMT1, peroxisome PPARγ1, Dnm3os, MALAT1	Improved systemic insulin sensitivity	Obese mouse models, in vitro macrophages, patient samples	[77-79]
Non-coding RNAs in pancreatic macrophages	HG and palmitic acid stimulation	Upregulation of E330013P06 and Dnm3os promotes inflammation and foam-cell formation; downregulation of the anti-inflammatory lncRNA mist accelerates inflammation	E330013P06, Dnm3os, mist	No direct impact on β-cells explicitly stated in the source	Mouse models; monocyte/macrophage experiments	[80]
Metabolic memory of pancreatic macrophages	History of hyperglycemia; effects persist even after normalization of metabolic parameters	Sustained pro-inflammatory state and metabolic dysfunction mediated by “trained immunity”	DNA methylation, histone modifications (e.g., H3K4me3, H3K9ac), non-coding RNAs (e.g., miRNA, lncRNA)	Islet β-cell dysfunction, insulin secretion defects; potential β-cell damage due to chronic inflammation	In vitro cell experiments, animal models, human cohort studies	[81]
Macrophage-derived cytokines and lipid mediators	Obesity, hyperglycemia, growth factors, oxidative stress, inflammatory cytokines	Enhanced pro-inflammatory M1-like phenotype	Exosomal miR-212-5p	Impaired insulin secretion	Mouse models, in vitro cell experiments	[82-85]
Epigenetic interactions	hIAPP aggregation and exposure	Modulates inflammatory phenotype and polarization; establishes long-term epigenetic memory (“trained immunity”); regulates secretion of factors (e.g., tPA, TGF-β)	tPA	Protects β-cells by reducing hIAPP-aggregate-induced toxicity	In vitro studies	[86-89]

DNMT1: DNA methyltransferase 1; PPARγ1: Peroxisome proliferator-activated receptor gamma 1; Dnm3os: Dynamin 3 opposite strand; MALAT1: Metastasis-associated lung adenocarcinoma transcript 1; HG: High glucose; H3K4me3: Histone H3 lysine 4 trimethylation; H3K9ac: Histone H3 lysine 9 acetylation; miRNA: MicroRNA; lncRNA: Long non-coding RNA; tPA: Tissue plasminogen activator; TGF-β: Transforming growth factor-β; hIAPP: Human islet amyloid polypeptide.

TARGETING PANCREATIC MACROPHAGE METABOLIC MEMORY FOR EARLY INTERVENTION AGAINST OBESITY-RELATED T2DM

Lifestyle interventions (aerobic exercise): Reversing macrophage polarization and epigenetic memory

Aerobic exercise is increasingly recognized not merely for caloric expenditure but as a potent inducer of molecular reprogramming within the immune system. It acts as a physiological countermeasure to pancreatic metabolic memory by promoting a shift in macrophages from a pro-inflammatory (M1) to a reparative (M2) phenotype, thereby dampening the chronic inflammation that sustains β-cell dysfunction[35]. This reversal involves systemic signaling; exercise attenuates insulin resistance in tissues like muscle and fat by modulating the miR-221-3p/Janus kinase/signal transducer and activator of transcription axis[36]. By downregulating miR-221-3p, it relieves suppression on /Janus kinase/signal transducer and activator of transcription pathways, facilitating macrophage polarization toward an anti-inflammatory state. This reduces circulating cytokines (e.g., TNF-α, IL-6), alleviating inflammatory pressure on islets and promoting β-cell recovery[37].

Significantly, exercise challenges pathological metabolic memory at the epigenetic level. It induces “beneficial trained immunity” in macrophages, involving genome-wide alterations in chromatin accessibility and histone modifications (e.g., histone 3 lysine 4 trimethylation redistribution). This reprograms macrophages to suppress pro-inflammatory genes and enhances anti-inflammatory pathways, increasing their metabolic flexibility from glycolysis (M1) toward oxidative phosphorylation (M2)[38]. The efficacy of this remodeling is intensity-dependent, with higher-intensity exercise sustaining M2 polarization longer, likely due to greater metabolic demands forcing immune reprogramming[39]. These benefits can occur independent of significant weight loss, suggesting exercise directly remodels the macrophage epigenome to counteract obesity’s sequelae, offering a non-pharmacological strategy to hinder T2DM progression[40].

Acupuncture-based neuromodulation: Indirect targeting of pancreatic macrophage metabolic memory

Acupuncture interventions [manual acupuncture and electroacupuncture (EA)] provide a non-pharmacological strategy to modulate macrophage inflammation and metabolic reprogramming in obesity and T2DM. In targeting pancreatic macrophage metabolic memory for early intervention, acupuncture primarily induces systemic immunometabolic changes that indirectly influence islet macrophages by reducing chronic low-grade inflammation driven by ATMs and promoting anti-inflammatory phenotypes[41]. Recent studies (2023-2025) further highlight its capacity to bias macrophage polarization toward M2-like states, particularly in adipose tissue, thereby benefiting pancreatic islets through reduced cytokine spillover and neuro-immune modulation[41,42].

Electroacupuncture-driven macrophage reprogramming in obesity models consistently suppresses adipose inflammatory signaling and improves whole-body metabolic homeostasis. EA at ST36 or multi-acupoint protocols (e.g., ST25, ST36, SP6) similarly shift macrophage metabolism away from glycolysis-dominant proinflammatory states toward oxidative phosphorylation, counteracting trained immunity–like persistent adaptations. This metabolic rewiring attenuates systemic inflammation and supports improvements in insulin sensitivity and glucolipid metabolism[43,44].

Neuro-immune modulation of pancreatic islets by electroacupuncture offers an indirect route to restrain local macrophage activation and preserve islet structure. EA at ST25 can activate the pancreatic intrinsic nervous system via transient receptor potential vanilloid 1 channels, increasing substance P and calcitonin gene-related peptide and thereby restoring islet architecture, reducing vacuolation, and dampening local macrophage activation and cytokine production[45]. Pharmacological comparators such as curcumin (nuclear factor kappa B inhibition) and resveratrol (sirtuin 1 activation) can directly engage inflammatory pathways but remain constrained by bioavailability, whereas acupuncture leverages neuro-immune crosstalk without pharmacokinetic limitations[46].

Protocol parameters (intensity, waveform, and acupoint combinations) critically determine the durability of macrophage polarization shifts and metabolic outcomes. Dense–disperse waveforms and combined abdominal/Lower-limb point regimens have been associated with sustained M2 skewing and longer-lasting metabolic benefits. Therapeutic potential therefore centers on dampening proinflammatory dominance while enhancing resolution programs, with the broader goal of interrupting the feed-forward loop linking islet inflammation to β-cell impairment in early obesity-related T2DM. Clinical evidence supports acupuncture as an adjunctive approach, and emerging mechanistic data on islet-relevant neural targeting warrant further translational validation[47].

Epigenetic modulators (histone deacetylase inhibitors, miRNA mimics/inhibitors): Erasing pancreatic macrophage functional memory

While lifestyle interventions and natural products offer systemic benefits, the pursuit of pharmacological erasure of metabolic memory has led to the exploration of direct epigenetic modulators. Specifically, histone deacetylase inhibitors (HDACi) function by altering the acetylation status of histone tails, thereby remodeling chromatin accessibility. In the context of diabetic islet inflammation, specific histone deacetylases (HDACs) (such as HDAC3) act as negative regulators of anti-inflammatory pathways or maintain the open chromatin state of proinflammatory genes essential for the M1 phenotype. Pharmacological inhibition of these enzymes can disrupt the persistent epigenetic marks (e.g., stabilizing histone 3 lysine 27 acetylation at M2-associated loci) that constitute the physical trace of metabolic memory. For instance, isoform-selective inhibition of HDAC3 has been shown to reprogram macrophage responsiveness, reducing the secretion of toxic cytokines (IL-1β, TNF-α) and protecting β-cells from apoptosis[48]. Unlike broad-spectrum immunosuppression, HDACi therapy aims to restore the epigenetic plasticity of macrophages, allowing them to reset from a sensitized state back to a tolerogenic surveillance phenotype[49].

Parallel to histone modification, miRNA-based therapeutics offer a strategy to intervene at the post-transcriptional level. Macrophage polarization is strictly governed by a network of regulatory miRNAs. In obesity-related T2DM, proinflammatory miRNAs such as miR-155 are chronically upregulated, locking macrophages into an aggressive state. Conversely, resolution-promoting miRNAs like miR-146a are often suppressed. The administration of miR-155 inhibitors (antagomirs) or miR-146a mimics can effectively override these dysregulated feedback loops[50]. Although challenges regarding tissue specificity and delivery systems (e.g., nanoparticle encapsulation) remain, the application of epigenetic modulators represents a significant paradigm shift. It moves the therapeutic goal from merely managing hyperglycemia to actively revising the immune history of the pancreas, potentially halting the vicious cycle of β-cell failure at its epigenetic root[51]. The mechanisms of action, specificity for memory erasure, and supporting evidence for each intervention are systematically summarized in Table 3.

Table 3 Interventional strategies for pancreatic macrophage metabolic memory.

Intervention strategy	Representative agents/methods	Mechanism of action	Specificity for memory erasure	Evidence type	Ref.
Lifestyle intervention	Aerobic exercise	Activates SIRT1 pathway, promotes macrophage M2 polarization, downregulates TNF-α/IL-6, upregulates IL-10	High Rationale: > 8 weeks via H3K4me3/H3K27ac epigenetic remodeling; pancreatic macrophage direct evidence	Preclinical + clinical	[90,91]
		Improves pancreatic macrophage metabolic reprogramming, enhances PKB2-mediated insulin signaling		Obese mice (n = 10/group): Reduces HOMA-IR, upregulates PBMC SIRT1
		H3K27ac, inhibits HDAC3, modulates epigenetic memory		Obese humans (n = 89): Endurance exercise enhances insulin sensitivity
Natural products	Resveratrol, curcumin	Resveratrol: Activates adenosine monophosphate - AMPK/SIRT1/Nrf2, inhibits NF-κB, enhances IL-10 promoter H3K27ac; regulates macrophage polarization (concentration-dependent)	Moderate Rationale: 2 weeks; histone modification; pancreatic macrophage direct evidence	Preclinical + clinical	[92,93]
		Curcumin: Blocks TLR4/MyD88/NF-κB, inhibits NLRP3 inflammasome, reverses IL-6 promoter DNA methylation; alleviates pancreatic β-cell oxidative stress		Cells (3 independent experiments): Resveratrol reduces LPS-induced NO/TNF-α, curcumin inhibits IL-1β
				Human trial (n = 40): 1 g/day resveratrol increases SHBG, improves metabolism
Epigenetic modulators	HDACi, miR-10a mimics	HDAC inhibitors: Inhibit HDAC1/2/3, reduce H3K27me3, activate Nrf2, enhance IL-10 expression	Extremely high Rationale: 4-8 weeks; reverses obesity-induced epigenetic abnormalities; pancreatic macrophage direct evidence	Preclinical + clinical	[94]
		miR-10a mimics: Promotes pancreatic macrophage OXPHOS, increases acetyl-CoA/H3K18ac, inhibits HDAC3, modulates metabolic memory		Mice: Improves glucose/insulin resistance, reduces macrophage infiltration
				Lymphoma patients: MS-275 lowers TNF-α/IL-6
				Human pancreatic macrophages: SAHA upregulates IL-10
miRNA-based intervention	miR-10a mimics, miR-146a mimics, miR-155 antagonists	miR-10a: Regulates metabolic reprogramming, enhances H3K18ac, blocks inflammatory memory	Extremely high Rationale: 6-8 weeks; reverses epigenetic abnormalities; pancreatic macrophage direct evidence	Preclinical	[91,95]
		miR-146a: Targets HDAC2-PI3K axis, inhibits M1 polarization		Mice: MiR-10a reduces M1/M2 ratio, miR-146a improves insulin sensitivity
		miR-155 antagonists: Rescues PDX1, associated with pancreatic β-cell function		Cells: MiR-146a upregulates IL-10, miR-155 antagonists inhibit TNF-α
Other	DMI	Mevalonate: Induces innate immune memory via inflammatory gene H3K27ac/H3K4me3	Moderate Rationale: > 3 weeks; reverses partial inflammatory memory; pancreatic macrophage direct evidence	Preclinical	[91]
		DMI: Enriches TNF/IL-6 promoter H3K4me3, reduces H3K9me3, regulates metabolic memory, inhibits inflammatory responses in pancreatic macrophages		Cells: Mevalonate enhances secondary inflammatory response, DMI inhibits LPS-induced TNF-α
				Animal experiment: DMI improves obesity-related inflammation

SIRT1: Sirtuin 1; TNF-α: Tumor necrosis factor-α; IL-10: Interleukin-10; IL-6: Interleukin-6; PKB2: Protein kinase B2; H3K27ac: Histone 3 lysine 27 acetylation; HDAC3: Histone deacetylase 3; AMPK: AMP-activated protein kinase; Nrf2: Nuclear factor erythroid 2-related factor 2; NF-κB: Nuclear factor kappa B; TLR4: Toll-like receptor 4; MyD88: Myeloid differentiation factor 88; NLRP3: NOD-like receptor pyrin domain-containing protein 3; H3K27me3: Histone 3 lysine 27 trimethylation; OXPHOS: Oxidative phosphorylation; H3K18ac: Histone 3 lysine 18 acetylation; PI3K: Phosphoinositide 3-kinase; PDX1: Pancreatic and duodenal homeobox 1; DMI: Dimethyl itaconate; H3K4me3: Histone 3 lysine 4 trimethylation; H3K9me3: Histone 3 lysine 9 trimethylation; HOMA-IR: Homeostasis model assessment of insulin resistance; PBMC: Peripheral blood mononuclear cell; LPS: Lipopolysaccharide; NO: Nitric oxide; SHBG: Sex hormone-binding globulin; SAHA: Suberoylanilide hydroxamic acid; HDACi: Histone deacetylase inhibitors.

CHALLENGES AND FUTURE DIRECTIONS

Dissecting tissue-specificity of pancreatic macrophage metabolic memory

Ontogenic and functional heterogeneity of tissue-resident macrophages constitutes a major barrier to translating metabolic memory into clinical therapy. Many studies extrapolate observations from bone marrow-derived macrophages (BMDMs) or ATMs to the pancreas; however, IAMs display a distinct transcriptional and epigenetic identity shaped by their specialized microenvironmental niche. Tissue context may therefore impose fundamentally different rules for memory encoding in IAMs, making pancreas-specific mechanistic dissection a prerequisite for rational therapeutic intervention[51].

The pancreatic islet microenvironment creates a specialized immune-endocrine interface that imprints IAMs with a niche-specific epigenetic landscape. Constant exposure of IAMs to locally high concentrations of insulin, ATP, and gamma-aminobutyric acid (GABA) released by active β-cells differs markedly from the lipid overflow-dominated cues that shape adipose macrophages[52]. Broad epigenetic reprogramming strategies (e.g., non-selective HDAC inhibition) may therefore restore adipose macrophage homeostasis yet inadvertently disrupt essential physiological functions of resident IAMs or fail to engage the precise chromatin loci modified within the islet niche[53].

Tissue-dependent variation in trained immunity introduces an additional layer of complexity in defining pancreatic macrophage metabolic memory. Hyperglycemia can induce a systemic trained phenotype in circulating monocytes (central memory), whereas the pancreatic persistence of memory (peripheral memory) appears to be sustained by local niche signals. Evidence suggests that IAMs rely predominantly on self-renewal under steady-state conditions, but T2DM-associated recruitment generates a chimeric macrophage population within islets[18]. Determining whether pathological memory primarily resides in long-lived resident macrophages that retain cumulative tissue history or in newly recruited cells remains a critical hurdle, because mis-targeting the dominant carrier population could be ineffective or harmful. For instance, blocking monocyte recruitment may reduce incoming inflammatory pressure yet leave entrenched epigenetic dysfunction in resident macrophages uncorrected despite continued proximity to β-cells[54].

Uniform systemic epigenetic modulation carries substantial off-target risk and motivates the development of pancreas-selective targeting strategies. Systemic delivery of metabolic reprogrammers could impair macrophage populations in the lung (alveolar macrophages) or liver (Kupffer cells), increasing infection susceptibility - a vulnerability already elevated in diabetic patients. Identification of IAM-exclusive surface markers and niche-restricted metabolic dependencies should therefore guide targeted delivery platforms (e.g., ligand-conjugated nanoparticles) designed to edit pancreatic immune memory without compromising systemic host defense.

Ontogenic and functional heterogeneity of tissue-resident macrophages complicates translation of “metabolic memory” into therapy. Many mechanistic insights come from BMDMs or ATMs, but direct extrapolation to pancreatic IAMs can mislead. IAMs carry a niche-imprinted transcriptional and epigenetic program, and may encode memory by rules distinct from cultured BMDMs. Although BMDMs exposed to diabetic cues (high glucose/insulin/palmitate) acquire a “metabolically activated” ATM-like state, tissue macrophages in vivo - including IAMs - retain chromatin architectures and proteomic signatures not recapitulated in vitro[55]. The islet microenvironment differs sharply from adipose tissue: Β-cells constitutively secrete insulin together with ATP and GABA, forming an immune-endocrine interface rather than a lipid-overflow inflammatory milieu. IAMs are tuned to these signals; they express purinergic receptors and sense ATP co-released with insulin, linking activation to β-cell secretory activity. β-cell-derived GABA may bias macrophages toward an anti-inflammatory IL-10 + phenotype (pnas.org). Therefore, broad epigenetic reprogramming approaches effective in adipose inflammation (e.g., non-selective HDAC inhibition) may fail in islets if they do not engage IAM-specific loci - or may disrupt homeostatic IAM functions shaped by the islet epigenetic landscape. Pancreas-specific mechanistic dissection is thus required rather than one-size-fits-all strategies[56].

Ontogeny further stratifies islet macrophage “training.” Islets harbor long-lived embryonically derived resident IAMs and adult bone marrow-derived recruited IAMs. In healthy pancreas, yolk sac-derived resident IAMs dominate (> 90%), self-renew locally, reside near vessels/β-cells, and support development and maintenance by clearing apoptotic cells and producing VEGF, insulin-like growth factor-1, IL-10, and transforming growth factor β; macrophage deficiency causes structural defects and reduced β-cell mass. Recruited monocyte-derived IAMs are scarce at baseline but rise with obesity/diabetes via lymphocyte antigen 6 complex + monocyte influx, localize peri-islet, and adopt a pro-inflammatory program (CD11c, inducible NO synthase, IL-1β, TNF-α, IL-6) that impairs β-cells and amplifies chemokine-driven recruitment[57]. Disease progression is accompanied by dynamic reshaping of the islet macrophage compartment. Early obesity is characterized predominantly by expansion of resident IAMs through local proliferation with minimal monocyte contribution, whereas prediabetes and T2DM are associated with progressively greater recruitment and accumulation of monocyte-derived macrophages. In late-stage T2DM, monocyte-derived IAMs may comprise about 60%-80% of the islet macrophage pool, consistent with pronounced CD68 + macrophage infiltration reported in both human and mouse islets. Studies differ regarding the extent of C-C chemokine receptor 2 dependence, likely reflecting differences in model severity and disease timing. From a therapeutic perspective, inhibiting recruitment (e.g., C-C chemokine receptor 2 antagonism) may attenuate acute inflammatory influx yet leave preconditioned resident macrophages - and their putative “memory” - largely intact, whereas systemic epigenetic “resetting” carries a risk of broad immunosuppression. Accordingly, pancreas-selective strategies that exploit IAM-enriched surface markers or niche-imposed metabolic dependencies, coupled with ligand-directed nanoparticles for targeted delivery of reprogramming payloads, may offer a more precise means to reverse maladaptive memory while preserving systemic host defense[11].

Clinical translation bottlenecks in macrophage-targeted therapies

Despite the compelling efficacy of macrophage-targeted interventions in preclinical rodent models - ranging from specific depletion via colony-stimulating factor 1 receptor (CSF1R) inhibitors to phenotypic reprogramming using metabolic modulators - the translation of these strategies into human clinical practice for T2DM has faced substantial impediments. Notably, the CSF1R signaling pathway plays a central role in metabolic homeostasis, underpinning its theoretical relevance as a therapeutic target for diabetes[58]. The progression from preclinical success to clinical application is currently stalled, primarily due to challenges in tissue specificity, interspecies heterogeneity, and the delicate balance required to modulate inflammation without compromising systemic immunity[59].

The most significant hurdle remains delivery specificity and off-target toxicity. In murine studies, systemic administration of CSF1R inhibitors (e.g., PLX3397) effectively depletes islet macrophages and restores β-cell function. While depleting macrophages may alleviate inflammation, it can concurrently disrupt essential paracrine support for β-cell function. A comprehensive study in lean mice demonstrated that CSF1R inhibition with PLX5622 depletes islet macrophages and directly impairs glucose-stimulated insulin secretion, a defect linked to the loss of macrophage-derived IL-1β, which acts as a physiological insulin secretagogue[58]. Furthermore, the same study found that systemic CSF1R inhibition improved hepatic insulin sensitivity, and these immune and metabolic perturbations were reversible upon drug withdrawal, a crucial consideration for chronic disease therapy. Evidence from non-pancreatic compartments underscores this tissue-specificity. For instance, PLX3397 robustly depletes ATMs without improving glucose homeostasis in diet-induced obese mice, suggesting that benefits are not guaranteed by macrophage reduction alone[60]. Conversely, in a model of aging, PLX5622-induced modulation of hypothalamic microglia and ATMs, when combined with environmental enrichment (a lifestyle mimic), additively improved metabolic outcomes[58]. In humans, CSF1R is ubiquitous across all tissue-resident macrophage populations[61]. Systemic inhibition risks depleting Kupffer cells in the liver and alveolar macrophages in the lungs, leading to hepatotoxicity and increased susceptibility to infections - adverse events that are unacceptable for the management of a chronic, non-fatal condition like T2DM[62,63]. Consequently, while CSF1R inhibitors have utility in oncology, their safety profile has largely precluded their development as diabetes therapeutics.

Furthermore, there is a profound discordance in inflammatory signatures between rodent models and human pathology. Genetic or diet-induced obesity models in mice often exhibit exaggerated, aggressive macrophage infiltration and a clear M1/M2 dichotomy that simplifies the disease mechanism. In contrast, human islet inflammation is low-grade, heterogeneous, and chronic. Human islet macrophages often display an intermediate activation state rather than a polarized phenotype, and their contribution to β-cell dysfunction is more subtle, involving impaired efferocytosis rather than massive cytokine release[61]. This discrepancy explains why potent anti-inflammatory biologics, such as IL-1β antagonists (e.g., anakinra, canakinumab), demonstrated only modest glycemic improvements in clinical trials and failed to provide durable preservation of β-cell mass comparable to animal data[62].

Emerging technologies aim to circumvent these barriers through precision nanomedicine. Strategies utilizing liposomes or exosomes as biomimetic carriers to deliver reprogramming agents (e.g., miRNAs or anti-inflammatory drugs) specifically to macrophages are gaining traction. By conjugating these carriers with ligands that target macrophage-specific receptors (e.g., mannose receptors or scavenger receptors), researchers hope to achieve high local concentrations within the pancreas while minimizing systemic exposure[64]. However, the dense extracellular matrix of the fibrotic islet in T2DM poses a physical barrier to nanoparticle penetration. Moreover, the lack of validated biomarkers to stratify patients based on their specific islet inflammatory status makes it difficult to identify the subpopulation that would benefit most from immunomodulation.

Future success will likely depend on moving away from broad-spectrum depletion toward precise modulation strategies. Integrating macrophage-modulating agents with established therapies - such as glucagon-like peptide-1 receptor agonists, which have secondary anti-inflammatory effects - may offer a synergistic pathway to restore islet homeostasis without dismantling the host’s immune defense[65].

Future Outlook: Deciphering heterogeneity and spatial memory via single-cell multi-omics

The advent of single-cell technologies has fundamentally shifted the understanding of immunometabolism, moving beyond the M1/M2 dichotomy to reveal pancreatic IAMs as a continuous spectrum of activation states. A critical frontier in T2DM research is to identify the specific subpopulation that harbors the epigenetic imprints of metabolic memory[66]. Current single-cell atlases have identified distinct IAM subsets, including a novel “lipid-associated macrophage”-like population in obese islets, characterized by a transcriptional signature geared toward lipid metabolism and phagocytosis (e.g., triggering receptor expressed on myeloid cells 2, Trem2, cluster of differentiation 9). Future studies must integrate scRNA-seq with single-cell ATAC-seq to profile the chromatin accessibility of these subsets. This multi-omics approach is essential to pinpoint which genomic loci remain accessible in specific macrophage clusters long after glycemic normalization, thereby physically mapping the location of persistent metabolic memory[67].

Furthermore, metabolic memory possesses a crucial spatial dimension. The precise localization of a macrophage within the islet architecture - whether interacting with β-cells in the core or with other cells at the periphery - dictates its functional trajectory and memory signature, as emerging Spatial Transcriptomics data reveal[68]. To definitively establish causality, the field must transition to dynamic tracking, employing pseudotime trajectory analysis combined with lineage tracing systems. This will determine whether “memory macrophages” are long-lived residents or derive from continually conditioned monocytes. Ultimately, answering these questions is key to developing subset-specific therapies that can selectively eliminate pathogenic clones carrying deleterious memory while preserving the homeostatic functions of resident macrophages[69].

CONCLUSION

In summary, metabolic memory in pancreatic macrophages represents a crucial mechanism linking obesity to the onset and progression of T2DM. Metabolic stress imprints a lasting pro-inflammatory phenotype on these immune cells. Even after the correction of hyperglycemic and hyperlipidemic conditions, the resulting inflammation and dysregulation of lipid metabolism continue to impair β-cell function, thereby creating a vicious cycle that fuels the progression of diabetes.

While the concept of “pancreatic macrophage metabolic memory” provides a novel perspective for understanding T2DM, significant limitations remain in this field, which also define critical avenues for future research. The primary current limitations include: (1) Scarcity of human data. Most existing evidence originates from murine models. Due to the challenge of obtaining live or sequential human pancreatic samples, direct evidence is lacking to confirm the existence and epigenetic features of an analogous “metabolic memory” in human pancreatic macrophages; (2) Substantial heterogeneity of macrophages. The pancreas harbors multiple macrophage populations with spatiotemporally dynamic phenotypes. It remains unclear which subset is the primary carrier of “metabolic memory” and how different subsets interact; and (3) Lack of tissue-specific resolution. The literature predominantly focuses on macrophage function at a systemic level. There is a paucity of studies specifically investigating macrophages within defined tissues like the pancreas or comparing differences between macrophages across distinct tissue niches. To address these gaps, future research should leverage advanced technologies. For instance, integrating single-cell multi-omics to analyze pancreatic samples from organ donors across the spectrum from health to obesity, prediabetes, and T2DM will be crucial. This approach can map the epigenetic and transcriptional landscape of human pancreatic macrophages during disease progression and enable cross-species validation with murine models. Subsequently, identifying and validating the key metabolites, receptors, or epigenetic modifiers that drive the macrophage transition from “transient activation” to a “persistent memory” state is essential.

Targeting this mechanism - through lifestyle interventions, natural products, and epigenetic modulation - offers new therapeutic avenues for early intervention against obesity-related T2DM. Future research should employ single-cell omics and other advanced technologies to further decipher the heterogeneity of pancreatic macrophages and the specificity of their memory formation, thereby overcoming the translational barriers from basic research to clinical application. Ultimately, reprogramming the “memory” of the pancreatic immune microenvironment holds promise for opening a new, fundamental pathway for the prevention and treatment of diabetes.