Hyperglycemia associated brain dysfunction: rodent models for studying diabetic cognitive deficits

Abstract

Diabetes and chronic hyperglycemia exert a profound impact on the central nervous system, significantly increasing the risk of cognitive impairment and neurodegenerative disease. This neurological decline is driven by a complex interplay of molecular and cellular mechanisms, including dysregulated insulin signaling, impaired cerebral glucose metabolism, excitotoxic glutamate accumulation, oxidative stress, and neuroinflammation. Collectively, these metabolic stressors compromise synaptic integrity and plasticity, promoting pathological hallmarks such as amyloid-beta accumulation, tau hyperphosphorylation, and neuronal atrophy. Structural and functional abnormalities are particularly prevalent in the hippocampus and prefrontal cortex, regions essential for learning and memory. Researchers utilize a diverse array of rodent model systems, including chemically induced, diet-induced, genetic, and autoimmune models, to study specific disease facets. No single model, however, perfectly replicates the complexity of human diabetes; instead, models vary in their face, construct, and predictive validity. While some systems offer high translational relevance, others provide superior mechanistic clarity. Consequently, the strategic selection of animal models is crucial for accurately characterizing diabetes-associated neurological dysfunction and facilitating the translation of preclinical findings into effective human clinical interventions. In this review we discuss the major diabetic rodent models and synthesize the current evidence linking diabetes and hyperglycemia to neurological dysfunction in the context of the rodent model(s) best suited for investigating specific impairments. Understanding how metabolic disease influences neuronal and molecular pathways that are critical for cognition and being able to study them in a translational model is essential for identifying early therapeutic targets and reducing the growing neurological burden of diabetes.

Introduction

Diabetes, long characterized primarily by its damage to peripheral tissues, is now recognized to have a major impact on the central nervous system (CNS). Consequently, the cognitive impairments associated with metabolic dysfunction are gaining significant attention in modern diabetes research. According to the CDC, over 38 million individuals in the U.S. were living with diabetes as of 2019, a figure projected to climb to 60 million by 2060. Furthermore, more than one-third of the American population is currently classified as prediabetic. These alarming statistics underscore the urgent need to investigate the long-term consequences of chronic metabolic disease on brain health.

Insulin is the central hormone which regulates carbohydrate metabolism and cellular glucose uptake. Diabetes mellitus is a chronic metabolic disease characterized by elevated blood glucose levels due to a disruption in insulin signaling. There are two types of diabetes, both disrupt systemic energy balance but develop through different mechanisms. Type 1 Diabetes (T1D) is an autoimmune disorder where a combination of genetic predisposition and environmental factors trigger the immune system to destroy the pancreatic β-cells, resulting in a severe and absolute insulin deficiency. Conversely, Type 2 Diabetes (T2D) is driven by peripheral insulin resistance where target tissues fail to respond to insulin, alongside a progressive decline in β-cell function. T2D is triggered by a mix of genetic risk and lifestyle factors including obesity, poor diet, and inactivity.

Growing clinical and preclinical evidence links both type 1 and type 2 diabetes to structural and functional brain abnormalities, resulting in deficits across several domains: learning and memory, executive function, mood regulation, and an increased risk of dementia (Galicia-Garcia et al. 2020). Brain regions critical for higher-order cognition, such as the hippocampus and prefrontal cortex, exhibit are particularly vulnerable to diabetic pathology (He et al. 2025). However, the precise cellular and molecular mechanisms translating chronic metabolic dysfunction into neuronal and neural circuit impairment remain poorly understood. To bridge this knowledge gap, researchers rely heavily on rodent models, which have a long history of reliability in modeling neurological and cognitive disorders.

Methodological advantages of rodent models

Mice and rats are indispensable tools in biomedical research due to their profound anatomical, physiological, and genetic alignment with humans. Rodents and humans each possess approximately 30,000 genes, sharing nearly 95% of their genome (Bryda 2013). Crucially, the glucose handling and insulin signaling pathways in rodents closely mirror human physiology. This conservation allows researchers to faithfully recapitulate core human diabetic phenotypes including: hyperglycemia, insulin resistance, and impaired glucose tolerance (King and Bowe 2016).

Beyond biological similarities, rodents offer distinct practical advantages for efficient, highly controlled experimental design. Their small size, low resource requirements, short gestation periods, and large litter sizes facilitate rapid experimental timelines (Bryda 2013). The availability of transgenic, knockout, and conditional genetic models allows for precise spatial and temporal manipulation of metabolic and signaling pathways. Controlled environments and selective breeding eliminate confounding variables common in human cohorts, significantly enhancing reproducibility.

Rodent models are particularly well-suited for tracking metabolic parameters across disease progression. Researchers can monitor fluctuations using tools ranging from basic glucometers to highly sophisticated diagnostic techniques, such as glucose tolerance tests, insulin tolerance tests, and continuous glucose monitoring (Hahn et al. 2024). These tools enable high-resolution tracking of how genetic, environmental, or pharmacological interventions alter metabolic outcomes.

These biological, practical, and translational strengths make rodent models foundational for uncovering diabetic disease mechanisms and potential links to cognitive decline. This review systematically addresses the intersection of metabolic and neurological dysfunction. We first introduce the major rodent models of diabetes, highlighting their induction methods, metabolic profiles, and translational relevance to specific human phenotypes. We then evaluate the specific neurological deficits driven by chronic hyperglycemia and metabolic stress. Finally, we pair distinct cognitive and structural impairments with the specific rodent models best optimized to study them, providing a framework for targeted experimental design.

While diabetes presentation involves various comorbidities, this synthesis focuses explicitly on the roles of insulin dysregulation, glucose metabolism, excitotoxicity, oxidative stress, brain atrophy, and the development of Alzheimer’s disease-like pathology. Ultimately, decoding how metabolic failure impacts the molecular and cellular pathways vital for cognition is critical for identifying early therapeutic targets and mitigating the expanding global burden of diabetic cognitive decline.

Rodent models of diabetes

The ideal animal model is characterized by face validity, construct validity, and predictive validity. Face validity refers to the extent to which the animal exhibits observable behavioral, anatomical, and physiological features that resemble the human disease state. Construct validity describes whether the model shares the same underlying biological mechanisms as the human condition. Predictive validity reflects the how well the model responds to therapeutic interventions in a manner consistent with humans. While it is rare for a single animal model to fully satisfy all three criteria, models that meet a greater number of these benchmarks are generally considered more reliable and translationally relevant. We address how well the current rodent models of diabetes meet these three criteria (Table 1).

Table 1.

Degree of validity for current rodent models of diabetes

Experimental model	Face validity	Predictive validity	Construct validity
Chemical
Streptazocin (STZ)	+++	+++	+
Alloxan	+++	+++	+
Dietary
High Fat Diet (HFD)	+++	+++	+++
Combinatorial
HFD + STZ	+++	+++	++
Genetic
db/db mouse	+++	+++	+
ob/ob mouse	+	+++	+
ZDF rat	+++	+++	+
GK rat	++	+++	+++
Akita mouse	+++	+++	+
NOD mouse	+++	+++	+++

Rodent models of diabetes can be broadly categorized based on their mechanism of disease induction and the aspect of human diabetes they are intended to model. Current diabetic rodent models generally fall into three categories, each offering distinct experimental advantages:

• Chemically induced models: Widely utilized due to their reliability, reproducibility, and experimental flexibility.

• Genetic models: Ideal for uncovering inherited metabolic dysfunctions and the specific molecular pathways driving disease progression.

• Diet-induced models: Crucial for replicating the lifestyle-driven metabolic stress characteristic of type 2 diabetes mellitus.

Each category captures distinct features of diabetic pathology, such as pancreatic beta cell loss, insulin resistance, or progressive metabolic dysfunction. Because human diabetes is a disease with multiple etiologies and stages, no single animal model fully captures and mimics all aspects of the condition. Instead, the choice of model should be guided by the specific biological question being addressed, and the use of multiple models is often necessary to more fully represent the complexity of the disease. The following sections evaluate the mechanisms and applications of each model in detail.

Chemically induced models of diabetes

Streptozotocin

Streptozotocin (STZ) is the most widely utilized pharmacological agent for inducing diabetes in laboratory rodents, serving as a highly robust and versatile tool in preclinical research. A primary strength of the STZ model lies in its procedural flexibility; depending on the chosen dosing strategy and administration protocol, STZ can effectively model key features of both type 1 and type 2 diabetes (Ghasemi and Jeddi 2023). Furthermore, its efficacy across both mice and rats enhances its utility across diverse experimental paradigms.

STZ is a naturally occurring antibiotic derived from the bacteria Streptomyces achromogenes that functions as a DNA alkylating agents(King and Austin 2017). Its diabetogenic effects arise from its selective uptake into pancreatic beta cells via the highly expressed glucose transporter GLUT2. Once internalized, STZ triggers a series of cytotoxic events including rapid DNA alkylation, leading to structural fragmentation and eventual beta cell death (Athmuri and Shiekh 2023). STZ also dramatically elevates the production of reactive oxygen species (ROS), compounding cellular damage (Nahdi et al. 2017). The resulting targeted destruction of insulin-producing beta cells culminates in sustained hyperglycemia and profound glucose dysregulation.

The metabolic severity and resulting diabetic phenotype are highly dependent on the chosen dosing regimen, allowing investigators to tailor the model to specific experimental timelines. Administration of a single, concentrated dose of STZ induces rapid and near total cell destruction. This severe insulin deficiency closely mimics the clinical presentation of type 1 diabetes (T1D). By contrast, repeated, lower doses of STZ provoke gradual cell damage and partial insulin deficiency to mimic type 2 diabetes (T2D). This dose dependent flexibility allows investigators to tailor the model to specific experimental goals, including studies of disease progression, metabolic dysfunction, or diabetic complications.

There are several strengths of the STZ model. The induction is rapid, allowing for quick and predictable significant shifts in glycemic status. STZ is versatile, capable of modeling both T1D and T2D phenotypic traits. STZ is highly effective in both mice and rats. There are some known limitations of STZ. STZ accumulation can cause unintended non-pancreatic damage, particularly hepatic and renal toxicity (Eleazu et al. 2013). There can be higher animal mortality, compared to genetic models, due to systemic toxicity. Chemical ablation fails to capture the complex autoimmune pathophysiology driving human T1D. Regardless, STZ induced diabetes models remain a staple of experimental diabetes research.

The STZ model has high face validity as it successfully replicates core clinical phenotypes of human diabetes, including sustained chronic hyperglycemia, progressive weight loss, and severe disruptions to lipid and glucose metabolism. The model demonstrates strong predictive value, as STZ-treated rodents respond predictably to standard glucose-lowering therapies that improve glycemic control in human patients. The underlying mechanism of beta-cell destruction significantly diverges, however, from human disease. Human T1D is an autoimmune pathology characterized by progressive, immune-mediated destruction of beta-cells, whereas the STZ model relies on acute chemical necrosis and oxidative stress (Eleazu et al. 2013). Despite this mechanistic divergence from human autoimmune pathogenesis, the STZ model has exceptional face validity, a rapid induction timeline, and the phenotypic flexibility ensure it remains a foundational asset for investigating the systemic and neurological consequences of chronic hyperglycemia.

Alloxan

Alloxan is another commonly used cytotoxic agent to induce an experimental type 1 diabetic state in laboratory rodents. Effective in both mice and rats, alloxan selectively targets pancreatic beta cells. Its diabetogenic efficacy relies on its structural profile, which allows it to infiltrate beta cells via glucose transporters, triggering an intracellular redox cycle that culminates in beta cell death. The destruction of beta cells by alloxan is a multi-step, highly cytotoxic process driven by two primary mechanisms. Once internalized, alloxan is rapidly reduced by cytosolic reducing agents (such as glutathione) into dialuric acid. It then undergoes autoxidation back into alloxan (Szkudelski 2001). This continuous cycling generates massive quantities of reactive oxygen species (ROS) (Elsner et al. 2006), including superoxide radicals and hydrogen peroxide, inducing severe DNA fragmentation and fatal oxidative stress. Alloxan also exhibits an incredibly high affinity for sulfhydryl groups. It directly targets and deactivates essential functional proteins, including glutathione, cysteine, and the key glucose sensor glucokinase (Brito et al. 2011). This dual action, combining localized oxidative damage with the systemic inactivation of metabolic enzymes, profoundly impairs insulin production and rapidly induces beta cell necrosis.

Alloxan quickly establishes hyperglycemia, providing an efficient turnaround for acute studies. Alloxan is also highly economical and straightforward to prepare, reducing the financial barriers of large-scale animal cohorts. Because it induces a reliable, severe hyperglycemic baseline, it is frequently used to investigate secondary diabetic complications, most notably painful diabetic neuropathy driven by progressive, glucose-mediated peripheral nerve damage (Ivanova et al. 2025). There are a couple notable drawbacks. Alloxan features an extremely tight dosing window. Minor over-dosing can trigger acute systemic toxicity and high animal mortality rates, requiring meticulous calibration. Significant accumulation can occur in non-target tissues, frequently causing severe nephrotoxicity (kidney damage) and hepatic steatosis (fatty liver), both of which risk confounding experimental outcomes (Ader et al. 1998). While alloxan is an accessible alternative to STZ, its use requires a strict cost-benefit analysis due to its volatile pharmacological window.

Alloxan has high face validity as it faithfully replicates core clinical symptoms of human type 1 diabetes, including persistent severe hyperglycemia, rapid weight loss, and altered lipid and carbohydrate metabolism. Alloxan also has robust predictive capacity, as alloxan-induced diabetic states respond predictably to standard clinical interventions, such as glucose-lowering agents and targeted antioxidant therapies (Shah and Khan 2014). Like STZ, alloxan exhibits clear constructive limitations. It induces rapid, necrotic cell death via chemical toxicity and oxidative stress, completely bypassing the progressive, immune-mediated autoimmune destruction that characterizes human type 1 diabetes pathogenesis. While its narrow dosing margin and off-target toxicity require careful experimental management, the low cost, rapid induction timeline, and reliable generation of severe hyperglycemia make alloxan an extremely valuable model for studying downstream diabetic complications.

Diet induced models of diabetes

High fat diet

High fat diet (HFD) feeding is one of the most common rodent models of type 2 diabetes and metabolic syndrome. In this model, mice or rats are transitioned from a standard chow diet to one where a massive proportion of total caloric intake is derived from fat—typically shifting lipid composition from ~ 10% to ~ 60% of total calories (Surwit et al. 1998). Over time, this chronic nutritional surplus provokes systemic metabolic alterations that closely mirror human T2D, including obesity, insulin resistance, and impaired glucose regulation (Stott and Marino 2020). Because this model mimics the effects of long-term dietary and lifestyle factors, it is often used to study metabolic dysfunction associated with obesity and metabolic syndrome.

The metabolic disruption triggered by HFD feeding stems primarily from the gradual development of systemic insulin resistance, which progresses through a well-characterized physiological cascade. Chronic exposure to excess dietary fats drives severe adiposity. The expanding adipose tissue triggers the accelerated release of pro-inflammatory cytokines and lipotoxic lipid metabolites (Heydemann 2016). These circulating inflammatory mediators and lipids interfere with intracellular insulin signaling cascades in peripheral target tissues, most notably skeletal muscle and the liver, reducing glucose transporter efficiency and suppressing glucose uptake. In response to rising peripheral resistance, pancreatic beta-cells initially compensate by increasing insulin secretion, leading to chronic hyperinsulinemia (Mosser et al. 2015). Over time, this compensatory mechanism becomes unsustainable. Beta-cell function begins to decline, resulting in impaired glucose tolerance and the gradual manifestation of overt hyperglycemia.

The choice to utilize an HFD model requires balancing its high physiological fidelity against its extended experimental timelines and phenotypic variability. Unlike chemical models (such as STZ or alloxan) that induce abrupt, artificial beta-cell death, the HFD model captures the progressive, insidious onset of metabolic failure driven by chronic caloric excess. Beyond glucose dysregulation, HFD-fed rodents develop a constellation of comorbidities that mimic human metabolic syndrome, including systemic inflammation, dyslipidemia, and visceral adiposity (Deal et al. 2020). The gradual disease onset provides an excellent experimental window for investigating the prediabetic state and early molecular biomarkers of metabolic decline. One of the main limitations of the HFD model is the prolonged induction period. Establishing a robust diabetic phenotype requires a significant temporal investment, often necessitating several weeks to months of continuous dietary exposure. Additionally, the severity of the resulting hyperglycemic state is highly variable and sensitive to confounding factors, including the specific rodent strain, exact macronutrient composition of the diet, and environmental housing conditions (Burcelin et al. 2002). In many cohorts, rodents develop profound insulin resistance and obesity without progressing to severe, sustained hyperglycemia, potentially limiting the utility of the HFD model for questions requiring a high-glucose baseline.

The HFD model has excellent across the board validity. The model comprehensively replicates the clinical presentation of human T2D, faithfully mirroring core features such as obesity, insulin resistance, low-grade metabolic inflammation, and impaired glucose tolerance. The underlying biological mechanisms, specifically the adipose tissue expansion prompting inflammatory cascades that corrupt insulin receptors, match the precise etiological pathways observed in human diet-induced obesity. The model exhibits high predictive value; pharmacological agents utilized clinically to enhance human insulin sensitivity or optimize glucose homeostasis produce highly parallel metabolic improvements in HFD-fed rodents.

While investigators must account for extended timelines and strain-dependent variability, the unparalleled physiological accuracy and high construct validity of the HFD model make it a robust and indispensable paradigm for decoding the cellular mechanics underlying type 2 diabetes and metabolic syndrome.

Combination induced models of diabetes

High fat diet + low dose STZ

The combination of a high-fat diet (HFD) with low-dose streptozotocin (STZ) administration represents a highly robust and widely utilized bifunctional rodent model for type 2 diabetes (T2D). Rather than relying on a single pathological trigger, this sequential paradigm elegantly recapitulates the multi-stage progression of human T2D. Animals are first fed an HFD to establish a baseline of systemic insulin resistance, followed by a sub-lethal dose of STZ to induce partial pancreatic beta-cell dysfunction. The resulting phenotype exhibits both impaired insulin signaling and compromised insulin secretion, closely mirroring the clinical pathogenesis of advanced human T2D.

The combined model achieves its high physiological fidelity by sequentially targeting peripheral tissues and the endocrine pancreas. Chronic consumption of the HFD drives visceral adiposity and low-grade chronic inflammation. This lipotoxic and inflammatory environment disrupts intracellular insulin signaling cascades, impairing glucose uptake in peripheral tissues (Abbasi and Khodadadi 2025). Subsequent administration of low-dose STZ induces mild, targeted damage to pancreatic beta-cells via DNA alkylation and moderate oxidative stress (Zhang et al. 2008). Unlike high-dose regimens that completely ablate the endocrine pancreas, this low-dose titration causes only partial loss of insulin-producing cells. The synergy between established peripheral insulin resistance and a newly restricted insulin secretory capacity prevents the pancreas from compensating, rapidly culminating in sustained hyperglycemia and progressive metabolic decline.

The dual induction strategy of this model offers distinct experimental benefits but introduces unique methodological variables that require careful management. It successfully captures the transition from insulin resistance to pancreatic cell damage within a single experimental system. It produces a stable, severe diabetic phenotype much faster than the HFD alone, significantly reducing husbandry time and resource costs. Unlike the HFD model, where rodents often maintain normal glucose levels through massive insulin compensation, this model guarantees sustained, chronic hyperglycemia. Although highly robust, there are some considerations before choosing to use this combination model. Managing a two-step protocol involving dietary modifications and precise chemical injections increases operational complexity compared to single-intervention designs. The exact severity of the diabetic state is highly sensitive to confounding factors, including variations in HFD macronutrient composition, STZ batch potency, and baseline rodent strain vulnerabilities (Srinivasan et al. 2005). While it effectively simulates the downstream consequences of beta-cell failure, the use of a cytotoxic chemical agent bypasses the natural, chronic metabolic exhaustion that drives beta-cell burnout in human patients.

One of the primary advantages of this model is its outstanding across the board validity. The model demonstrates excellent face validity, faithfully replicating core clinical hallmarks of advanced human T2D, including persistent hyperglycemia, insulin resistance, and severely impaired glucose tolerance. Construct validity is moderate. While the model successfully integrates the two primary biological pillars of human T2D (insulin resistance and secretory deficit), the underlying mechanism of beta-cell loss is driven by acute chemical necrosis rather than progressive metabolic wear. The model exhibits high predictive value; standard clinical anti-diabetic therapeutics (such as insulin sensitizers and secretagogues) produce highly parallel improvements in glucose regulation when administered to these rodents (Tang et al. 2024).

By layering chemically induced beta-cell stress onto a background of diet induced insulin resistance, the combined HFD/low-dose STZ model serves as an invaluable translational tool. It bridges the gap between purely chemical and purely dietary models, offering an efficient and reliable framework for studying T2D progression and testing novel therapeutic interventions.

Genetic models of type 2 diabetes

db/db mouse

The db/db mouse is one of the most widely utilized genetic models of T2D. This model is defined by a homozygous point mutation in the leptin receptor gene (Lepr), which completely disrupts normal signaling of leptin. Leptin is an adipose-derived hormone essential for regulating appetite, satiety, and energy homeostasis (Wang et al. 2014a). Deprived of functional leptin signaling, these animals display profound hyperphagia, rapid weight gain, and morbid obesity, which inevitably culminate in severe insulin resistance and chronic hyperglycemia.

The cellular and behavioral pathology of the db/db mouse originates entirely from the inability of cells to respond to the hormone leptin. Under normal physiological conditions, circulating leptin crosses the blood-brain barrier to signal the hypothalamus, suppressing appetite and elevating energy expenditure when lipid stores are sufficient (Dornbush and Aeddula 2026). Because the mutated leptin receptors are incapable of transducing this signal, db/db mice exist in a state of perceived starvation, driving relentless, compulsive overeating (Guest and Rahmoune 2019). The resulting massive caloric surplus accelerates adipose tissue expansion, triggering systemic lipotoxicity, severe peripheral insulin resistance, and an eventual breakdown of glycemic control.

One of the major strengths of the db/db model is the severity and consistency of its metabolic phenotype, though its monogenic nature requires careful consideration when translating findings to human populations. These mice reliably and uniformly develop obesity, insulin resistance, and severe hyperglycemia on a predictable timeline, providing a highly stable baseline for longitudinal studies (Burke et al. 2017). Because the disease is genetically driven, it eliminates the confounding variables and unpredictable success rates associated with chemical inductions or dietary compliance. Unlike the STZ or alloxan models, the db/db mouse allows for the study of diabetic complications without the risk of off-target chemical toxicity in hepatic or renal tissues. Despite these advantages, the db/db model has some notable limitations. The severity of the obesity and metabolic dysfunction in db/db mice is profoundly more extreme than what is typically observed in the vast majority of human T2D patients. Human T2D is fundamentally a polygenic disease heavily influenced by complex gene-environment interactions, whereas the db/db phenotype is driven entirely by a single, rare neuroendocrine mutation.

The db/db mouse is a model highly optimized for studying the physiological consequences of metabolic stress, despite its artificial etiology. The db/db model demonstrates high face validity by accurately mirroring the clinical presentation of human T2D, including morbid obesity, profound insulin resistance, leptin resistance, and sustained, chronic hyperglycemia. Construct validity is somewhat limited. While the downstream cellular consequences (e.g., receptor desensitization, metabolic exhaustion) align with human disease, the initiating cause does not represent the typical lifestyle- and polygenic-driven pathogenesis of human T2D. The model exhibits high predictive value, as standard clinical glucose-lowering medications and insulin sensitizers demonstrate parallel efficacy in improving metabolic profiles and glycemic control in db/db cohorts (Li et al. 2010).

The db/db mouse remains a foundational genetic tool in preclinical research. While its monogenic nature oversimplifies human etiology, its phenotypic reliability and severe metabolic profile make it an invaluable system for examining advanced diabetic complications and testing novel therapeutic agents.

ob/ob mouse

The ob/ob mouse is a well-established genetic model widely utilized to investigate the mechanisms of obesity, metabolic syndrome, and the early stages of T2D. This model is characterized by a homozygous nonsense mutation in the leptin gene (Lep), which completely ablates the production of functional leptin. Deprived of this essential endocrine signal, ob/ob mice exhibit profound hyperphagia and a rapid, early-onset accumulation of adipose tissue, serving as a foundational system for studying obesity-related insulin resistance (Zhang et al. 1994).

In healthy physiology, leptin acts within the hypothalamus to suppress appetite and stimulate energy expenditure (Hu et al. 2025). Because ob/ob mice lack functional leptin, the central nervous system fails to perceive existing energy stores, locking the animal into a state of perpetual neuroendocrine hunger that drives massive food intake. The resulting morbid obesity alters lipid clearance, triggering dramatic elevations in circulating insulin levels (hyperinsulinemia) and severe peripheral insulin resistance (Suriano et al. 2021).

The ob/ob model provides a highly predictable phenotype for obesity research, but its unique hormonal and metabolic profile introduce distinct experimental limitations. These animals reliably develop severe obesity and insulin resistance at a very young age, offering a highly consistent timeline for longitudinal metabolic studies. Like the db/db mouse, the model relies on a single, well-defined genetic mutation, thus eliminating the experimental noise and phenotypic variability common in diet-induced models. The model is well-suited for tracking how massive adiposity impacts peripheral insulin pathways and alters energy expenditure. While ob/ob mice exhibit severe obesity and profound insulin resistance, they feature robust beta-cell compensation. Consequently, their hyperglycemia is often mild, or transient compared to more severe T2D models. The mutation causes significant comorbidities, including profound hyperlipidemia and infertility, which can complicate long-term breeding strategies and confound specific experimental outcomes (Lindström 2007). Congenital leptin deficiency is exceedingly rare in human populations. Therefore, the initiating mechanism of metabolic failure in ob/ob mice does not accurately reflect the polygenic, lifestyle-driven etiology of typical human T2D.

When considering the translational validity of the ob/ob mouse, it is clear the model has strength as an obesity model rather than a complete model of advanced diabetes. The model displays high face validity regarding obesity, systemic insulin resistance, and metabolic dysregulation. However, its face validity is more limited as a model of advanced T2D due to the absence of progressive beta-cell failure and sustained, severe hyperglycemia. Construct validity is also limited. The syndrome stems entirely from a single, monogenic neuroendocrine defect rather than the intricate web of polygenic vulnerabilities and environmental factors (such as diet and sedentary lifestyle) that drive human disease. The model demonstrates robust predictive capacity; clinical interventions designed to enhance insulin sensitivity or alleviate metabolic inflammation yield highly parallel therapeutic improvements in ob/ob mice (Ozcan et al. 2015).

While the ob/ob mouse falls short of capturing the full spectrum of late-stage diabetic hyperglycemia and beta-cell exhaustion, its phenotypic consistency makes it an invaluable, high-fidelity tool for decoding the molecular links between adipose tissue expansion and systemic insulin resistance.

Zucker Diabetic Fatty Rat (ZDF)

The Zucker Diabetic Fatty (ZDF) rat is a premier genetic model used to investigate type 2 diabetes (T2D) and metabolic syndrome. This model features a spontaneous, missense mutation in the leptin receptor gene (Lepr), functionally homologous to the mutation found in the db/db mouse, which disrupts normal leptin signaling pathways (Phillips et al. 1996). Consequently, the ZDF rat reliably develops an array of metabolic phenotypes including obesity, insulin resistance, and profound hyperglycemia, making it an invaluable tool for tracking the longitudinal progression of diabetic complications.

The metabolic breakdown in the ZDF rat follows a predictable physiological timeline driven by central leptin resistance. Impaired leptin receptor signaling disrupts the hypothalamic regulation of satiety and energy expenditure, inducing chronic overeating and rapid weight gain (Unger 1997). The resulting severe adiposity triggers systemic lipid accumulation, which impairs insulin signaling cascades in skeletal muscle and hepatic tissues. While pancreatic beta-cells initially compensate for peripheral resistance by hypersecreting insulin, this response is unsustainable. Over time, the combined stress of glucotoxicity and lipotoxicity induces progressive beta-cell failure, leading to a sharp decline in insulin production and the onset of persistent, severe hyperglycemia (Yokoi et al. 2013).

The ZDF rat model provides a framework for studying advanced diabetes, though its phenotypic intensity requires careful consideration. Beyond metabolic dysfunction, the ZDF rat naturally develops secondary complications analogous to human clinical presentations, including dyslipidemia, nephropathy, and vascular/cardiovascular complications (Pick et al. 1998). Unlike other genetic rodent models (such as the ob/ob mouse) that maintain robust pancreatic compensation, the ZDF rat exhibits true pancreatic failure, closely mirroring the advanced stages of human T2D. The phenotype is highly robust and predictable, particularly in male rats, allowing for tightly controlled longitudinal studies. The extreme, rapid-onset obesity observed in ZDF rats, however, can act as a confounding variable, making it difficult to separate the direct effects of hyperglycemia from the systemic impacts of morbid adiposity. Like other genetic models, the disease is initiated by a single receptor mutation and oversimplifies the polygenic architecture and lifestyle-driven dynamics of typical human T2D development.

The ZDF rat demonstrates a high degree of utility for macrovascular and microvascular diabetes research. The ZDF model demonstrates high face validity by accurately replicating the full clinical constellation of advanced human T2D, including progressive insulin resistance, pancreatic exhaustion, sustained hyperglycemia, dyslipidemia, and microvascular decline. Construct validity is limited by its monogenic nature rather than the gene-environment interactions that dictate human pathogenesis. The model displays high predictive value; standard clinical anti-diabetic therapeutics, including metformin and insulin sensitizers, produce highly parallel metabolic and glycemic improvements in ZDF cohorts (Ferreira et al. 2020).

The ZDF rat is an excellent translational model that bridges the gap between purely metabolic studies and diabetic complication research. Because it features authentic beta-cell failure alongside cardiovascular decline, it remains a high value platform for evaluating late-stage T2D pathologies and novel therapeutics.

Goto-Kakizaki rat (GK)

The Goto-Kakizaki (GK) rat is a unique, non-obese genetic model of type 2 diabetes (T2D) developed through the selective inbreeding of Wistar rats displaying mild glucose intolerance (Goto et al. 1976). Unlike most rodent paradigms for T2D, the GK rat establishes chronic hyperglycemia in the complete absence of obesity. This distinctive trait provides researchers with a powerful, unconfounded experimental system to isolate and study pancreatic beta-cell failure and secretory deficits independent of excess adiposity and severe lipotoxicity.

The diabetic phenotype of the GK rat is characterized by a primary defect in insulin secretion coupled with subtle peripheral metabolic alterations. The hallmark of the GK model is a profound defect in glucose-stimulated insulin secretion. Pancreatic β-cells fail to adequately respond to elevated blood glucose levels, driving persistent postprandial and fasting hyperglycemia (Portha et al. 2001). Over time, these rats exhibit a progressive loss of β-cell mass, architectural disruption of the pancreatic islets, and localized fibrosis (Homo-Delarche et al. 2006). While the model is fundamentally defined by a secretory deficit, GK rats also develop a secondary, moderate level of insulin resistance and altered glucose utilization within skeletal muscle and hepatic tissues (Guest 2019).

One of the major advantages of the GK rat model is the absence of obesity. By eliminating morbid obesity as a variable, researchers can evaluate the direct cellular and molecular mechanisms of glucose toxicity on beta-cells without the confounding influence of massive visceral adiposity (Ostenson and Efendic 2007). Human T2D is predominantly a polygenic disease. Because the GK rat phenotype is driven by multiple naturally selected genetic loci rather than an artificial monogenic mutation (like Lepr or Lep), it more accurately replicates human genetic complexity. The model also avoids the systemic shock and off-target organic toxicities associated with chemical induction agents like STZ or alloxan. The severity of diabetes in GK rats is relatively mild and stable compared to more aggressive chemical or monogenic models, which may limit its utility for studies requiring advanced, catastrophic metabolic collapse. Because these animals remain lean, the GK model fails to capture the full spectrum of obesity-driven complications, such as profound dyslipidemia, severe hepatic steatosis, and metabolic syndrome-associated systemic inflammation.

Given its polygenic nature, the GK rat model is replicates the genetic alignment rather than holistic lifestyle replication. The GK model demonstrates moderate face validity. It faithfully recapitulates impaired glucose tolerance, defective insulin secretion, and chronic hyperglycemia. However, its presentation diverges from the typical human T2D demographic, where many patients exhibit concurrent obesity. The construct validity of the GK rat is exceptionally high. Its polygenic origin aligns far better with the multifactorial, inherited genetic susceptibilities driving human T2D than any monogenic knock-out or knock-in model. The model displays high predictive value; standard clinical therapeutics aimed at glucose regulation improve metabolic outcomes (Zhou et al. 2024).

The GK rat remains an invaluable and distinct translational platform in T2D research. By decoupling pancreatic failure from obesity, it provides a framework for decoding the specific molecular vulnerabilities that govern beta-cell exhaustion and impaired insulin kinetics.

Genetic and autoimmune models of type 1 diabetes

Akita mouse

The Akita mouse is a well-characterized spontaneous genetic model of type 1 diabetes (T1D) defined by a progressive loss of pancreatic beta-cells. Unlike chemical models, diabetes in the Akita mouse is driven by a single point mutation in the insulin 2 gene (Ins2) (Yoshioka et al. 1997). This structural mutation prevents the proper folding of the proinsulin molecule, triggering an aggressive intracellular stress cascade that culminates in beta-cell death, severe insulin deficiency, and sustained, chronic hyperglycemia.

The molecular etiology of diabetes in Akita mice presents a classic case of cellular protein toxicity. The conformational defect prevents the formation of essential disulfide bonds within the insulin molecule, rendering it unable to exit the endoplasmic reticulum (ER). Misfolded proinsulin progressively accumulates within the lumen of the ER, leading to ER stress. (Oyadomari et al. 2002). Severe, chronic ER stress persistently activates the unfolded protein response. When homeostatic adaptations fail to clear the protein aggregation, the pathway shifts to activate pro-apoptotic signaling molecules driving localized beta-cell apoptosis. The progressive loss of beta cells results in a decline in endogenous insulin production, elevated blood glucose, and the development of diabetes.

One of the primary strengths of the Akita mouse model is its genetic consistency and reproducibility. Because the Ins2 mutation is inherited in a predictable, autosomal dominant manner, animals uniformly develop hyperglycemia without requiring volatile chemical injections or long-term dietary manipulation. Due to the pervasive nature of the underlying ER stress, the remaining beta-cells lack the capacity to regenerate. This lack of endogenous recovery makes the Akita mouse a gold-standard model for evaluating the long-term efficacy of islet transplantation therapies (Schmidt et al. 2009). By avoiding systemic cytotoxic agents like STZ or alloxan, researchers can study diabetic microvascular and macrovascular complications without the confounding variable of off-target tissue damage. The diabetic phenotype, however, exhibits severe sex-linked variability. Male Akita mice develop rapid, life-threatening hyperglycemia that frequently requires exogenous insulin titration to ensure survival, whereas females present with a significantly milder, delayed phenotype (Oyadomari et al. 2002). The model also lacks the characteristic autoimmune markers (e.g., autoantibodies) that drive human T1D, limiting its utility for immunotherapeutic research.

The Akita mouse model demonstrates strong face validity, accurately mirroring the clinical endpoints of T1D in humans, including progressive beta-cell mass reduction, drastic insulinopenia, severe polydipsia, polyuria, and sustained hyperglycemia. Construct validity is limited. Human T1D is an organ-specific autoimmune disease where the immune system selectively obliterates healthy beta-cells. In contrast, the Akita mouse develops diabetes via an intrinsic, non-immune-mediated endoplasmic reticulum protein-misfolding pathology. The model displays high predictive value; standard clinical protocols, including exogenous insulin therapy, islet transplantation, and targeted glucose-lowering regimens, yield highly predictable improvements in the animal’s overall metabolic and glycemic profiles (Miyata et al. 2020).

While the Akita mouse does not replicate the autoimmune etiology of human T1D, its reliable, non-chemical induction protocol and irreversible beta-cell loss make it an indispensable translational tool for examining the systemic effects of chronic hypoinsulinemic hyperglycemia and testing advanced therapeutic strategies.

NOD mouse (NOD)

The Non-Obese Diabetic (NOD) mouse is a cornerstone of T1D research, offering one of the closest experimental mirrors to human pathogenesis. Autoreactive T cells actively infiltrate the pancreatic islets (Kachapati et al. 2012). This localized inflammatory process drives the progressive destruction of insulin-producing pancreatic beta-cells. As beta-cell mass diminishes, insulin production plummets, ultimately resulting in persistent hyperglycemia and severe insulin deficiency.

One of the major advantages of the NOD mouse is its strong relevance to the autoimmune nature of human type 1 diabetes. It closely mimics the autoimmune nature of human T1D, making it an invaluable tool for studying T-cell activation, inflammatory signaling, and immune tolerance (Welzen-Coppens et al. 2013). Because the disease develops naturally rather than being artificially induced by chemicals or radiation, it provides an authentic window into both the initiation and progression of autoimmune diabetes. For this same reason, disease onset and incidence are not uniform; they fluctuate significantly based on sex and environmental conditions. Further, while the underlying immunology aligns well with humans, regulatory differences exist. For example, NOD mice exhibit severe insulitis across nearly all pancreatic islets, whereas human insulitis is typically less pronounced and spares many islets (Driver et al. 2011).

The NOD mouse has exemplary validity across all three criteria. The clinical presentation is an excellent match. Mice naturally develop the core hallmark symptoms of human T1D: progressive beta-cell loss, insulin deficiency, and chronic hyperglycemia. The underlying biological mechanism mirrors that seen in humans. Both the murine model and human disease share the exact same root cause: T-cell-mediated autoimmune destruction of beta cells. It translates well to therapeutics. Immunotherapies capable of altering disease progression in humans show similar efficacy in modifying diabetes development in NOD mice (Chen et al. 2013).

Neurological deficits associated with hyperglycemia

Brain insulin signaling and cerebral glucose metabolism

Insulin is a central hormone in carbohydrate metabolism. Secreted by pancreatic beta cells, it facilitates cellular glucose uptake and systemic energy balance. Diabetes mellitus is a chronic metabolic disease characterized by elevated blood glucose levels resulting from impaired insulin production, insulin resistance, or both (Galicia-Garcia et al. 2020).

• Type 1 Diabetes (T1D): Characterized by the autoimmune destruction of pancreatic beta cells, leading to severe insulin deficiency. Its etiology involves genetic predisposition, autoimmune dysfunction, and environmental triggers like viral infections (Atkinson et al. 2014).

• Type 2 Diabetes (T2D): Driven by a combination of genetic risk and lifestyle factors (e.g., poor diet, obesity, inactivity, smoking). It features progressive beta-cell dysfunction and peripheral insulin resistance, where target tissues fail to respond to insulin (Cerf 2013).

While T1D and T2D differ in origin, both expose the central nervous system to sustained metabolic stress. This environment increases the vulnerability of neurons and glia to structural and functional damage, linking disrupted insulin signaling and metabolic dysfunction to an increased risk of neurological impairment (Yoon et al. 2023).

Beyond its classic role in peripheral glucose regulation, insulin also functions as an important neuromodulator within the central nervous system. Insulin receptors are widely expressed in brain regions key for cognition, including the hippocampus, prefrontal cortex, and entorhinal cortex (Pomytkin et al. 2018). Binding of insulin to these receptors activates intracellular signaling pathways, including the phosphoinositide 3-kinase (PI3K)/Akt pathway, that regulate both neuronal metabolism and synaptic function (Gabbouj et al. 2019). Notably, prolonged hyperglycemia can decrease the amount of insulin receptors at the BBB, reducing the ability for insulin to enter the brain (Kullmann et al. 2016).

The majority of glucose uptake in neurons is largely mediated by insulin-independent GLUT1 and GLUT3 transporters (Uemura and Greenlee 2006), however, GLUT4 provides an additional insulin-sensitive mechanism in response to heightened metabolic demand (Pearson-Leary and McNay 2016). In the hippocampus, insulin-induced PI3K/Akt activation triggers the translocation of GLUT4 to the plasma membrane, allowing neurons to scale up energy utilization during periods of increased activity (Huang and Czech 2007). Consequently, impaired insulin signaling impairs PI3K/Akt activation, disrupting GLUT4 trafficking which limits glucose availability during periods of increased neuronal activity, contributing to how diabetes compromises cerebral glucose utilization and impacts synaptic function and plasticity (van Gerwen et al. 2023).

In addition to metabolic functions, insulin directly regulates synaptic function and plasticity. Activation of the PI3K/Akt pathway facilitates long-term potentiation (LTP), the cellular basis of learning and memory (Guo et al. 2024), by modulating the phosphorylation and trafficking of NMDA and AMPA receptors (Skeberdis et al. 2001; Arnold et al. 2018). Disruption of insulin receptor signaling reduces synaptic density (Chiu et al. 2008), demonstrating that insulin signaling is necessary for normal neuronal and circuit function.

Insulin also serves as a critical brake on neuroinflammation via glycogen synthase kinase 3 beta (GSK3β). Under physiological conditions, activation of the PI3K/Akt pathway inhibits GSK3β activity. Insulin resistance, however, disinhibits GSK3β leading to sustained activity (Affuso et al. 2024) that promotes the production of proinflammatory cytokines which contribute to chronic neuroinflammation (Hajeforoosh et al. 2026). Thus, impaired neuronal insulin signaling in diabetes may compromise cognition through both reduced metabolic support, neuroinflammation, and direct disruption of synaptic plasticity.

Importantly, insulin resistance within the brain can occur independently of peripheral insulin resistance, rendering neuronal pathways selectively vulnerable to metabolic stress (Kim and Feldman 2015). A reduction in insulin receptor activation and downstream signaling is observed in both diabetic patients and experimental models, resulting in impaired synaptic plasticity and altered neuronal excitability (Ferrario and Reagan 2018). This distinct central pathology has given rise to the concept of Alzheimer’s disease as “type 3 diabetes”, a state where brain-specific insulin resistance drives progressive cognitive decline and accelerates neurodegeneration (De La Monte and Wands 2008). Although this concept of “type 3 diabetes” remains somewhat controversial, it highlights the potential for insulin dysregulation to directly impair neuronal function and accelerate neurodegenerative processes independently of peripheral insulin resistance. Together, these findings suggest that the neuro-metabolic bridge represents a critical mechanistic link between metabolic dysfunction and synaptic failure in the diabetic brain.

At the systems level, impaired insulin signaling disrupts cerebral energy metabolism and compromises glucose utilization. Peripheral insulin resistance is associated with regional hypometabolism, particularly within cognitive cortical regions (Kim and Feldman 2015; Willette et al. 2015). For instance, Higher Homeostatic Model Assessment of Insulin Resistance (HOMA-IR) scores correlate with reduced glucose metabolism across several key regions, including the ventral prefrontal cortex, cingulate cortex, temporal cortex, insula, and posteromedial cortex (Willette et al. 2015).

Because neuronal signaling and synaptic plasticity impose substantial energetic demands, this metabolic deficit compounds the direct synaptic effects of disrupted insulin signaling, contributing to cognitive dysfunction in diabetes. There are strikingly similar patterns of regional hypometabolism in AD, providing an additional point of mechanistic overlap between metabolic disease and neurodegeneration (Deery et al. 2024). To investigate these mechanisms, researchers rely on rodent models that accurately replicate insulin deficiency or insulin resistance (Table 2). Together, these complementary models provide essential experimental frameworks for isolating how disrupted insulin signaling drives synaptic failure and neurodegenerative vulnerability in the diabetic brain.

Table 2.

Rodent models of impaired insulin signaling

Model	Mechanism	Translational relevance	Reference
High-Dose STZ	rapid destruction of pancreatic β-cells.	Models severe insulin deficiency and acute features of T1D.	(Ferrario and Reagan 2018)
Low-Dose STZ + HFD	partial β-cell dysfunction.	Models insulin resistance alongside progressive metabolic decline.	(Ferrario and Reagan 2018)
db/db mouse	Homozygous mutation in the leptin receptor gene.	Replicates severe obesity, chronic hyperglycemia, and systemic IR to study long-term cognitive decline.	(Kim and Feldman 2015)
HFD	Prolonged high-fat dietary intake.	Replicates the gradual emergence of insulin resistance associated with obesity and metabolic syndrome.	(Kim and Feldman 2015)

Hyperglycemia, energy metabolism, and excitotoxicity

The brain is uniquely vulnerable to the metabolic stress induced by chronic hyperglycemia. Unlike many peripheral tissues, the central nervous system relies almost exclusively on glucose for ATP production (Brown et al. 2019). Consequently, precise glucose regulation is mandatory for proper brain function. In a healthy adult brain, glucose is transported across the blood brain barrier via glucose transporters (GLUTs). Once inside, glucose undergoes cellular respiration to generate the ATP required to fuel the exceptionally high energy demands of neurons (Koepsell 2020). This substantial metabolic requirement directly supports critical physiological tasks, including the propagation of action potentials, the synthesis of neurotransmitters, and the active mitigation of cellular oxidative stress. Under chronic hyperglycemic conditions, this localized glucose metabolism is compromised, driving an increased risk of cognitive impairment and progressive neurological deficits.

Chronic hyperglycemia also disrupts cerebral amino acid metabolism, particularly pathways involving glutamate, glutamine, and GABA (Zhu et al. 2022). Because glutamate functions both as the primary excitatory neurotransmitter in the brain and as an important metabolic intermediate, altered glucose utilization disrupts the tightly coupled glutamine-glutamate/GABA cycle between neurons and astrocytes (Schousboe et al. 2013). Within this cycle, glutamine serves as a glutamate precursor, which is subsequently converted to the inhibitory neurotransmitter GABA by glutamic acid decarboxylase (GAD) to maintain neurotransmitter homeostasis (Bak et al. 2006).

Imbalance of this cycle impairs cognitive performance, as interference with either glutamatergic or GABAergic signaling produces spatial learning and memory deficits (Ylinen et al. 1995; Liu et al. 2014). Clinical findings support this link: T2D patients testing positive for GAD autoantibodies, which impair GAD function and reduce GABA synthesis, exhibit a greater risk of cognitive decline than patients who test negative for these antibodies (Takagi et al. 2013). Together, diabetes-associated alterations in glutamate and GABA metabolism may disrupt excitatory/inhibitory balance, impair neurotransmission, and contribute to cognitive dysfunction (Zhou et al. 2023).

While glutamate is indispensable for normal synaptic transmission under physiological conditions, its accumulation becomes highly neurotoxic. Excess glutamate serves as a primary driver of neuronal death in both acute brain injuries and chronic neurodegenerative diseases (Zhou and Danbolt 2014). Under hyperglycemic conditions, astrocytes exhibit a significantly reduced capacity to clear extracellular glutamate from the synaptic cleft, leading to pathological accumulation (Rivera-Aponte et al. 2015). Hyperaccumulation drives excessive signaling through ionotropic AMPA and NMDA glutamate receptors leading to a massive influx of intracellular calcium. This calcium overload triggers mitochondrial dysfunction and severe oxidative stress, ultimately culminating in neuronal death. In cultured cortical neurons, seven days of sustained exposure to elevated glutamate concentrations triggers excitotoxicity and cell death (Sasaki-Hamada et al. 2022). Because this excitotoxic loop directly alters neuronal excitability and destabilizes synaptic signaling, it represents a critical mechanistic bridge linking systemic metabolic dysfunction to the cognitive deficits seen in diabetes.

The STZ model has specific experimental advantages for isolating and investigating the metabolic and cellular disturbances tied to excitotoxic neuronal injury. A major advantage of the STZ protocol is its rapid and highly reliable induction of chronic insulin deficiency and marked blood glucose elevation. STZ models consistently replicate key hallmarks of human diabetic encephalopathy, including altered glutamate signaling, impaired synaptic plasticity, and distinct deficits in hippocampal function (Otsuka et al. 2026). Because STZ-induced diabetes develops relatively rapidly, this framework is particularly advantageous for isolating and characterizing the early pathological alterations that precede widespread hyperglycemia-induced neuronal death. This model thus provides a highly valuable translational system for investigating how chronic metabolic stress unravels neuronal communication and cellular homeostasis.

Oxidative stress and molecular damage

Oxidative stress occurs when the production of reactive oxygen species (ROS) overwhelms endogenous antioxidant defenses. Oxidative stress is consistently elevated in both diabetic patients and experimental animal models of diabetes (Son 2012). Under physiological conditions, antioxidant enzymes including superoxide dismutase (SOD), catalase, and glutathione peroxidase (GPx) work together to neutralize ROS and protect cells from oxidative damage (Bajaj and Khan 2012). In diabetes, however, the activity of these antioxidant systems becomes compromised, reducing the capacity of cells to effectively counteract increased ROS production (Bajaj and Khan 2012). This disruption is observed in the diabetic brain, where changes in glutathione-dependent antioxidant enzymes and antioxidant capacity coincide with elevated lipid and protein oxidation (Ulusu et al. 2003; Nakhaee et al. 2010). Thus, the oxidative imbalance associated with diabetes results from both increased ROS production and an impaired ability to adequately neutralize these reactive species.

Chronic hyperglycemia drives this redox imbalance through several interconnected pathways, prominently including mitochondrial dysfunction and the formation of advanced glycation end products (AGEs). Because neurons depend on mitochondrial oxidative phosphorylation to meet their exceptionally high energetic demands, they are especially vulnerable to redox imbalance (Wang et al. 2021). Under chronic hyperglycemic conditions, excess glucose impairs electron transport efficiency, which increases ROS production (Giri et al. 2018). In turn, persistent mitochondrial ROS production damages mitochondrial proteins, lipids, and DNA, further disrupting mitochondrial function and promoting additional ROS release (Liao et al. 2025). This creates a positive feedback loop in which metabolic dysfunction and oxidative stress progressively reinforce one another, contributing to neuronal dysfunction and neurodegeneration (Quan et al. 2011).

AGEs are generated when excess blood glucose glycates peripheral and central proteins. This process directly damages protein structure and indirectly triggers intracellular cascades that spike ROS production. Fueled by hyperglycemia, the relationship between these biomarkers is reciprocal and self-sustaining: ROS accelerates the production of AGEs, which in turn stimulates further ROS generation. Among the various pathways disrupted by metabolic stress, the ROS pathway is a primary mediator of cellular damage and neuroinflammation. For example, high-glucose environments in cultured human endothelial cells activate downstream ROS signaling and NF-κB via AGE pathways, ultimately upregulating the expression of proinflammatory cytokines and triggering cellular apoptosis (Ho et al. 2006). On top of NF-κB promoting a proinflammatory environment in the brain, activated NF-κB directly suppresses the Nrf-2 pathway, downregulating critical antioxidant genes including SOD, HO-1, and glutathione s-transferase (GST) (Wakabayashi et al. 2010). This repression weakens the natural antioxidant defenses of the brain (Ganesh Yerra et al. 2013). Together, these mechanisms establish a self-perpetuating cycle between oxidative stress and neuroinflammation, wherein increased ROS promotes inflammatory signaling and neuroinflammation disrupts antioxidant protection to exacerbate oxidative damage.

The molecular consequences of diabetes-induced oxidative stress are widespread throughout the central nervous system, as demonstrated across multiple experimental models. In STZ induced diabetes in rats, protein conjugation with 4-hydroxynonenal (HNE), a cytotoxic byproduct of lipid peroxidation, is increased throughout the hippocampus (Reagan et al. 2000). HNE directly compromises neuronal survival by impairing glucose transport, inhibiting glutamate uptake, and inducing apoptosis in hippocampal neurons. Additionally, key oxidative stress markers such as malondialdehyde, xanthine oxidase, and nitric oxide were significantly elevated across multiple brain regions in type 2 diabetic patients (Calabrese et al. 2012).

Because oxidative stress is fundamentally tied to diabetes, chemically induced diabetic models are particularly effective for studying these mechanisms (Table 3). Together, these complementary rodent models offer vital experimental frameworks for deciphering how oxidative stress intersects with neuronal dysfunction, synaptic impairment, and accelerated neurodegeneration in the diabetic brain.

Table 3.

Diabetic rodent models of oxidative stress

Model	Mechanism	Translational relevance	Reference
STZ	Rapidly induces insulin deficiency and chronic hyperglycemia. STZ inherently promotes ROS generation during the destruction of pancreatic β cells.	Replicates elevated ROS, mitochondrial dysfunction, and acute oxidative damage. Highly reproducible timeline allows researchers to align oxidative damage with early synaptic and cognitive deficits.	(Reagan et al. 2000; Quan et al. 2011)
HFD or HFD + STZ	Induces chronic, low-grade metabolic stress that drives a gradual accumulation of cellular ROS over an extended timeline.	Provides strong translational relevance to human Type 2 Diabetes (T2D) and metabolic syndrome by mimicking the progressive, long-term metabolic burden on the brain.	(Calabrese et al. 2012)

Neurovascular dysfunction

Diabetic cognitive dysfunction is increasingly recognized as involving not only intrinsic neuronal abnormalities but also widespread dysfunction of the neurovascular unit (NVU). Composed of neurons, astrocytes, endothelial cells, pericytes, and vascular smooth muscle cells, the NVU maintains blood-brain barrier (BBB) integrity and regulates cerebral blood flow in response to neuronal activity (Zlokovic 2011; Iadecola 2017). This dynamic, activity-dependent regulation, termed neurovascular coupling, is ensures that increases in neuronal activity are accompanied by proportional increases in local oxygen and metabolic substrate delivery (Iadecola 2017).

Chronic hyperglycemia disrupts this system, primarily by impairing cerebral endothelial function. Because endothelial cells regulate vascular tone, cerebral perfusion, and the integrity of the BBB, their impairment represents an important mechanism of altered brain homeostasis in diabetes (Faraci 2011). Specifically, diabetes-associated oxidative stress, inflammation, and metabolic dysfunction compromise endothelial signaling and damage the cerebral microvasculature (van Sloten et al. 2020). As endothelial cells form the core cellular interface of the BBB, this microvascular damage can also disrupt the structural integrity and increase BBB permeability (Bogush et al. 2017).

Blood-brain barrier breakdown disrupts the tightly controlled extracellular environment required for optimal neuronal function, permitting circulating inflammatory mediators and other neurotoxic blood-derived factors to enter the brain parenchyma (Bogush et al. 2017). These factors further stimulate reactive glial responses and exacerbate localized oxidative stress. This creates another pathway through which vascular and metabolic dysfunction converge to promote neuronal injury (Zlokovic 2011).

Beyond structural BBB disruption, diabetes-induced microvascular damage impairs neurovascular coupling itself. Reduced microvascular reactivity limits the capacity of the cerebral circulation to adjust perfusion dynamically during neuronal activation (van Sloten et al. 2020; Barloese et al. 2022). Disruption of neurovascular coupling therefore creates a mismatch between neuronal energetic demand and substrate delivery, even when systemic glucose availability is elevated (Mondal et al. 2024). Because synaptic transmission and plasticity are energetically demanding processes, this persistent neuroenergetic deficit contributes to neuronal dysfunction before overt structural degeneration becomes apparent (Yu et al. 2019).

Collectively, endothelial dysfunction, BBB disruption, and impaired neurovascular coupling reduce the ability of the cerebral microcirculation to maintain an optimal metabolic environment. Over time, these pathological changes induce chronic cerebral hypoperfusion and sustained neuroenergetic dysregulation. Consequently, neurovascular dysfunction represents an active, early contributor to diabetic cognitive impairment rather than merely a downstream consequence of neuronal degeneration (Zlokovic 2011; Iadecola 2017).

Astrocytes and neuron-glia metabolic coupling

Although neurons have very high energetic demands, neuronal metabolic processes do not operate in isolation. Astrocytes play a central role in cerebral metabolic homeostasis by coordinating substrate availability with activity-dependent neuronal energy demands and providing metabolic and antioxidant support to neurons (Mulica et al. 2021). Positioned at the neurovascular interface, astrocytic end-feet processes envelope cerebral capillaries where perisynaptic processes directly interact with neuronal synapses. This unique spatial architecture enables astrocytes to bridge vascular nutrient delivery with neuronal energy utilization. Through glycolysis, astrocytes process glucose into lactate, which is transported to neighboring neurons as an activity-dependent metabolic substrate (Mulica et al. 2021). This coordinated exchange ensures that astrocytes respond to changes in neuronal activity and support the energetic demands required for synaptic transmission and plasticity.

Astrocyte-neuron metabolic coupling is closely integrated with neurotransmitter recycling. Following synaptic release, extracellular glutamate is cleared by surrounding astrocytes, preventing excessive glutamate accumulation within the synaptic cleft. Within astrocytes, glutamate is converted to glutamine by glutamine synthetase and subsequently transported back to neurons for the resynthesis of glutamate or GABA (Bak et al. 2006; Schousboe et al. 2013). Astrocytes therefore simultaneously protect neurons from excitotoxic glutamate accumulation and recycle neurotransmitter precursors necessary for continued synaptic transmission. Under hyperglycemic conditions, however, astrocytic glutamate clearance is impaired (Rivera-Aponte et al. 2015). This deficit disrupts neurotransmitter homeostasis and leads to the toxic the glutamate accumulation which promotes excitotoxic neuronal injury in diabetes.

Insulin signaling represents another important regulator of astrocytic function. Astrocytes express functional insulin receptors, demonstrating that the action of central insulin is not restricted to neurons. Selective loss of astrocytic insulin receptors impairs glucose handling by decreasing GLUT1 expression, glycolytic activity, and lactate release, thereby reducing the capacity of astrocytes to sense and respond to changes in nutrient availability (García-Cáceres et al. 2016). Furthermore, astrocytic insulin signaling also influences communication with neighboring neurons: insulin receptor loss impairs insulin-stimulated ATP release and downstream purinergic signaling, directly impacting neuronal activity (Cai et al. 2018). Compromised astrocytic insulin signaling also disrupts mitochondrial function, elevates ROS production, and uncouples cerebral blood flow from glucose uptake (Fernandez et al. 2022). These findings place astrocytes at an important intersection between insulin signaling, cellular energy metabolism, oxidative stress, and neurovascular regulation.

Astrocytic abnormalities are prominently observed in experimental models of diabetes. In STZ-induced diabetic rats, hippocampal astrocytes display progressive upregulation in glial fibrillary acidic protein (GFAP) and S100β accompanied by morphological alterations, consistent with sustained reactive astrogliosis (Nagayach et al. 2014). These changes occur alongside microglial activation, cellular degeneration, and deficits in learning and memory, demonstrating that glial abnormalities develop within brain regions particularly vulnerable to diabetes-associated cognitive dysfunction(Nagayach et al. 2014).

Under chronic metabolic stress, progressive astrocytic failure compounds neuronal injury through multiple convergent mechanisms: diminished glucose metabolism starves neurons during high demand states, impaired glutamate clearance drives excitotoxicity, and mitochondrial and oxidative abnormalities further compromise cellular homeostasis. Thus, neuronal dysfunction in diabetes may reflect not only cell autonomous metabolic impairment within neurons, but also a breakdown in the metabolic and homeostatic interactions between astrocytes and neurons.

Cellular pathophysiology

At the cellular level, diabetes induced neurological dysfunction stems from coordinated impairments across neurons, astrocytes, and microglia. Rather than resulting from isolated neuronal injury, diabetes-associated brain dysfunction reflects interconnected metabolic and cellular abnormalities across multiple cell types within the brain (Biessels and Despa 2018). Due to high energetic demands and limited metabolic flexibility, neurons are especially sensitive to disruptions in glucose availability (McDonald et al. 2023). Chronic hyperglycemia compromises neuronal ion homeostasis, calcium handling, and mitochondrial respiration, ultimately impairing action potential firing, synaptic transmission, and overall plasticity (Otero et al. 2025). As primary regulators of central glucose metabolism, glutamate clearance, and antioxidant defense, astrocytes undergo severe functional decline. Under hyperglycemic conditions, their capacity to buffer extracellular glutamate and support baseline neuronal energy requirements is drastically reduced (Kim et al. 2019). In response to persistent metabolic stress, microglia shift toward a chronic pro-inflammatory phenotype. This activation amplifies local oxidative damage and further exacerbates synaptic dysfunction (Gao et al. 2023).

Compromised cellular crosstalk directly manifests as measurable synaptic pathology, particularly within brain regions critical for cognitive processing. In STZ-induced diabetic rats, dendritic spine density is significantly depleted in both the frontal cortex and the hippocampus compared to age-matched controls (Wang et al. 2014b). Expression of synaptophysin, a presynaptic protein critical for functional synaptic vesicle trafficking, is decreased in these same regions, signaling a loss of structural synaptic integrity (Wang et al. 2014b). These alterations are accompanied by abnormal brain cholesterol metabolism. The brain is the most cholesterol-rich organ and disrupting this pathway severely compromises synaptogenesis and the structural maintenance of neurotransmitter rich active zones (Wang et al. 2014b).

Despite clear evidence of these cellular and synaptic alterations, several critical questions remain. It remains unclear whether neuronal dysfunction in diabetes is primarily a direct result of intrinsic metabolic impairment or an indirect consequence of glial failure and neuroinflammation. The precise degree to which insulin signaling directly modulates neuronal excitability, synaptic plasticity, and ion channel function in vivo is still poorly characterized. Most studies evaluate isolated pathologies (e.g., oxidative stress, excitotoxicity, or tau phosphorylation) rather than deciphering how these interconnected pathways converge within vulnerable structures like the hippocampus. It is currently unknown whether hyperglycemia induces reversible functional shifts or triggers permanent, irreversible structural degeneration at the cellular level. Addressing these gaps is essential for identifying early therapeutic windows and distinguishing mechanistic pathways of hyperglycemia induced neurodegeneration.

To address these outstanding questions, researchers may use complementary rodent models that target distinct facets of multicellular brain dysfunction (Table 4). Collectively, these models provide complementary systems for studying the complex cellular interactions underlying cellular pathophysiology of diabetic brain dysfunction.

Table 4.

Diabetic rodent models of neurodegeneration

Model category	Specific systems	Primary pathological focus	Translational relevance
Chemical	STZ	Chronic insulin deficiency, altered neuronal excitability, disrupted dendritic morphology, and impaired synaptic plasticity.	Ideal for isolating early pathological changes tied directly to acute hyperglycemia.
Genetic	db/db Mouse Zucker Diabetic Fatty Rat	Chronic metabolic dysfunction, progressive obesity, sustained glial activation, and long-term neuroinflammation.	Useful for tracking how prolonged metabolic stress alters cellular signaling over extended timelines.
Combination	High-Fat Diet (HFD) + Low-Dose STZ	Simultaneous obesity-associated inflammation, peripheral insulin resistance, and partial beta-cell dysfunction.	Replicates the complex, multi-system metabolic environment characteristic of human Type 2 Diabetes progression.

Hippocampal atrophy and structural brain changes

Hippocampal atrophy is a well-documented consequence of diabetes, driven by a complex interplay of molecular, structural, and vascular alterations. Rather than suffering from generalized, uniform brain aging, the hippocampus serves as a primary, highly vulnerable target for diabetes-related injury. Diabetes affects the transcriptome within the hippocampus by altering the expression of histone and histone-related genes. Because many of these genes regulate synaptic function and neuroplasticity, their disruption undermines cognitive processes (Thomas et al. 2013). Animal models reinforce these molecular findings. In STZ-induced diabetic rats, hippocampal neurons display a distinct retraction and simplification of apical dendrites, alongside heightened oxidative damage (Magariños and McEwen 2000). Clinical data indicates that midlife, but not late-life, diabetes is heavily associated with hippocampal atrophy. This timeline suggests that cumulative damage is fueled by decades of exposure to glucose toxicity, abnormal insulin signaling, and microvascular dysfunction (Mayo Clinic, 2014).

A key mechanism through which these vascular abnormalities may promote hippocampal degeneration is chronic cerebral hypoperfusion (CCH), a sustained, long-term reduction of cerebral blood flow (Rajeev et al. 2023). In diabetes, the convergence of endothelial dysfunction, structural vascular remodeling, and impaired cerebrovascular reactivity compromises the capacity of the brain to appropriately regulate regional perfusion (Feng and Gao 2024).

This persistent reduction in cerebral blood flow restricts the delivery of oxygen and metabolic substrates, particularly glucose, resulting in cellular energy deficits that can impair mitochondrial function, increase oxidative stress, and ultimately promote neuronal injury (Kimura et al. 2025). The hippocampus is acutely susceptible to CCH because of its high basal metabolic rate and sensitivity to disruptions in oxygen and glucose availability; consequently prolonged reductions in cerebral blood flow are central drivers of progressive hippocampal dysfunction and degeneration (De Jong et al. 1999).

Compared to non-diabetic controls, individuals with diabetes exhibit significantly lower volumetric ratios across multiple metrics: total brain volume to intracranial volume (TBV: ICV), hippocampal volume to intracranial volume (HV: ICV), and hippocampal volume to total brain volume (HV: TBV). Crucially, longer diabetes duration directly correlates with more severe reductions in these ratios. Because the HV: TBV ratio disproportionately shrinks, hippocampal degradation is an independent, targeted pathology rather than a simple byproduct of global brain atrophy. Human neuroimaging further substantiates these patterns, demonstrating clear, localized atrophy within both the hippocampus and the amygdala of diabetic patients (Hirabayashi et al. 2016). Taken together, this multi-level evidence underscores the unique susceptibility of the hippocampus to the chronic metabolic and oxidative stressors imposed by diabetes, tracing a clear path from early transcriptomic disruptions to structural decay.

Diabetic ketoacidosis and acute neurodegenerative risk

While chronic hyperglycemia drives gradual, progressive neuronal injury, acute metabolic crises like diabetic ketoacidosis (DKA) produce severe, long-lasting neurological damage. Beyond its acute systemic risks, DKA is increasingly recognized as a potent accelerator of neurodegenerative pathways (Bencharfa et al. 2025). Intriguingly, epidemiological data reveals that individuals with type 2 diabetes (T2D) who experience DKA carry a significantly higher risk of subsequently developing Alzheimer’s disease (AD), but notably not other forms of dementia (Chen et al. 2019). This specificity underscores a unique, shared pathophysiology between acute diabetic crises and AD-related neurodegeneration.

DKA places the brain into a hostile metabolic environment, mimicking the physiological profile of hypoxic or ischemic injury through reduced cerebral blood flow, increased concentrations of energy phosphates, and profound acidosis. This environment triggers distinct molecular alterations. Experimental rodent models demonstrate that DKA induces extensive tau hyperphosphorylation across multiple specific residues: T205, T212, S214, S262, S396/404, and S422 (Basurto-Islas et al. 2024). These specific modifications are heavily implicated in the initiation of neurofibrillary tangles. This pathological shift is driven by a two-pronged mechanism: elevated activity of stress-responsive kinases, particularly c-Jun N-terminal kinase (JNK) proteins, and concomitant downregulation of PP2A, the primary phosphatase responsible for maintaining tau homeostasis (Basurto-Islas et al. 2024).

DKA profoundly disrupts the broader mTOR/Akt axis, which is vital for neuronal survival and protein homeostasis (Basurto-Islas et al. 2024). DKA is associated with increased mTOR phosphorylation and elevated Akt phosphorylation at the Ser473 residue. These shifts occur alongside the downregulation of GSK-3beta (another major tau kinase), illustrating widespread signaling destabilization. Synaptic plasticity markers, such as synapsin-1 and PSD-95, are elevated following DKA. Rather than indicating preserved synaptic integrity, this likely reflects a compensatory structural remodeling response to acute injury. The structural fallout of DKA is visible in neuroimaging, particularly within vulnerable pediatric cohorts. Imaging studies of children with T1D and recurrent DKA have increased white matter volume, reduced grey matter volume, and lower cognitive performance(Jaser and Jordan 2021). These clinical insights prove that DKA is not merely a transient metabolic emergency, but a source of cumulative, permanent neurodevelopmental and neurocognitive alterations.

To investigate how acute DKA fuels chronic neurodegeneration, the best experimental rodent paradigm is STZ induction followed by insulin withdrawal to provoke ketoacidosis. By selectively destroying pancreatic beta cells, STZ establishes chronic hyperglycemia. Subsequent insulin withdrawal precipitates the acute, volatile metabolic state characteristic of clinical DKA including ketosis, acidosis, dehydration, and restricted cerebral perfusion. This multi-phase model accurately mirrors human cerebral metabolic disturbances, making it an invaluable tool for decoupling acute metabolic shock from baseline diabetic progression in the CNS.

Both chronic and acute metabolic instability converge to promote long-term brain injury. DKA stands out as a critical, underrecognized driver of diabetes-related neurological deficits, embedding an Alzheimer’s-like molecular signature in the brain long after the acute crisis has resolved.

Links to AD-like pathology

Alzheimer’s disease (AD) is classically defined by three hallmark pathologies- the extracellular accumulation of amyloid-beta plaques, intracellular neurofibrillary tangles (NFTs) composed of hyperphosphorylated tau, and chronic neuroinflammation. (Yin et al. 2021). Emerging evidence reveals that diabetes can independently replicate these pathologies in the brain, even in the absence of traditional genetic risk factors for AD (Biessels and Despa 2018). These findings highlight metabolic dysfunction as an important contributor to the development of AD-like neurodegenerative pathology.

As discussed previously, experimental models demonstrate that severe metabolic disturbances such as diabetic ketoacidosis (DKA) induce robust tau hyperphosphorylation and activation of multiple tau kinases in the rodent brain (Basurto-Islas et al. 2024). Conversely, while AD-like pathologies can arise independently of AD-associated genetic mutations in experimental models of diabetes, experimental models of AD also demonstrate alterations in brain insulin signaling and glucose utilization, highlighting a potential point of convergence between metabolic dysfunction and AD (Polis et al. 2022).

In diabetic brains, Aβ accumulation in vulnerable regions like the hippocampus and prefrontal cortex is driven by accelerated deposition and impaired clearance (Liu et al. 2008; Wang et al. 2014b). In STZ diabetic models, levels of both Aβ₁₋₄₀ and Aβ₁₋₄₂ spike significantly within the hippocampus and temporal cortex, directly correlating with cognitive decline. The accumulation does not occur just as a result of increased production; the clearance of Aβ fails (Stanciu et al. 2020). Diabetes impairs the expression and function of P-glycoprotein (P-gp), a critical efflux transporter at the blood brain barrier (Liu et al. 2008). Without functional P-gp, Aβ cannot be efficiently cleared out of the brain, leading to progressive, toxic accumulation.

Glycogen synthase kinase-3 (GSK-3) serves as the central mechanistic bridge between metabolic instability and neurodegeneration (Lauretti et al. 2020). GSK-3 orchestrates both sides of AD pathology: promoting Aβ production while simultaneously accelerating tau hyperphosphorylation (Alhassan et al. 2025).

In the healthy brain, GSK-3 activity is kept in check by upstream Akt signaling. However, the diabetic environment disrupts this balance. Hyperglycemia suppresses Akt signaling, effectively removing the brakes from GSK-3 (Qu et al. 2014). This unchecked GSK-3 hyperactivation drives the dual molecular phenotype of AD pathology, mimicking sporadic Alzheimer’s disease through purely metabolic distress.

When evaluating experimental models to study the intersection of diabetes and AD (Table 5), the non-genetic STZ rodent model remains the preferred choice. Unlike artificial genetic mutations, the STZ model provides evidence that metabolic trauma alone can initiate and accelerate the molecular and structural changes characteristic of Alzheimer’s disease.

Table 5.

STZ model of diabetes and Alzheimer’s disease phenotypes

Feature	Familial AD models	STZ diabetic model
Pathological Driver	Inherited genetic mutations (rare in humans)	Purely metabolic and glycemic dysfunction
AD Hallmarks	plaques, Tau pathology	Aβ accumulation Tau hyperphosphorylation, Neuroinflammation Oxidative stress
Intracellular Signaling	Variable baseline signaling	Recapitulates Akt/GSK-3 axis dysregulation
Clinical Relevance	Models rare familial AD	Models sporadic AD driven by metabolic syndrome

Summary

Diabetes and chronic hyperglycemia exert a heavy toll on the central nervous system, driving progressive cognitive decline and elevating the risk of neurodegenerative disease. This damage is structurally and functionally concentrated within the brain regions most critical for learning and memory.

At the molecular and cellular level, diabetic brain dysfunction is not caused by a single isolated pathway, but rather by an intricate, interconnected network of metabolic stressors. Dysregulated insulin signaling and impaired cerebral glucose metabolism starve neurons of efficient energy. Chronic metabolic strain leads to excitotoxic glutamate accumulation, triggering severe oxidative stress and chronic neuroinflammation. Together, these stressors compromise synaptic integrity, impair synaptic plasticity, and promote classic Alzheimer’s-like pathology—including amyloid-beta (Aβ) accumulation, tau hyperphosphorylation, and profound neuronal atrophy. While chronic hyperglycemia slowly harms the brain, acute metabolic crises like diabetic ketoacidosis (DKA) are accelerators. DKA induces acute signaling instability that rapidly provokes pathways associated with Alzheimer’s-like neurodegeneration.

Human diabetic brain dysfunction is highly complex. However, because no one animal model perfectly replicates human diabetes, researchers must choose models based on their experimental goals (Fig. 1). When selecting a rodent model to study diabetes mediated changes in neural structure and function, researchers must consider three distinct axes of validity:

Fig. 1.

Rodent model flow chart

• Face Validity: How closely the model mirrors human clinical symptoms.

• Construct Validity: How accurately the model replicates the underlying biological and molecular causes (e.g., insulin resistance vs. absolute insulin deficiency).

• Predictive Validity: How reliably therapeutic responses in the model predict clinical success in human trials.

Table 6 highlights how rodent models of metabolic dysfunction and neurodegeneration differ across their underlying mechanisms, rate of cognitive decline, reproducibility, and clinical relevance. Chemical models, such as high-dose STZ, induce rapid pancreatic -cell destruction, severe insulin deficiency, and acute oxidative stress, providing high experimental reproducibility and a rapid onset of cognitive deficits well-suited for modeling Type 1 Diabetes (T1D) and acute features of sporadic Alzheimer’s disease (AD). In contrast, dietary models (HFD) and combination protocols (low-dose STZ + HFD) induce chronic, low-grade metabolic stress and peripheral insulin resistance. These models exhibit a more gradual progression of cognitive decline over extended timelines, closely mirroring the multi-system burden of human Type 2 Diabetes (T2D) and metabolic syndrome.

Table 6.

Comparison of rodent models discussed in this review

Model	Mechanisms reproduced	Cognitive deficit progression	Reproducibility	Translational relevance
High-Dose STZ Chemical	• Rapid destruction of pancreatic β-cells• severe insulin deficiency• ROS generation• altered neuronal excitability• disrupted dendritic morphology• Aβ accumulation• Tau hyperphosphorylation• Akt/GSK-3 axis dysregulation.	Rapid onset; aligns early synaptic and cognitive deficits with acute hyperglycemia and oxidative damage.	High predictable timeline allows precise alignment of acute damage with early deficits.	Models severe insulin deficiency, acute features of Type 1 Diabetes (T1D), and sporadic Alzheimer’s disease (AD) driven by metabolic dysfunction.
Low-Dose STZ + High-Fat Diet	• Partial β-cell dysfunction• peripheral insulin resistance• obesity-associated inflammation• gradual cellular ROS accumulation under chronic low-grade metabolic stress	Progressive decline; tracks gradual cognitive burden driven by prolonged metabolic stress over extended timelines.	Moderate to High; provides a consistent multi-system metabolic environment.	High translational value for human Type 2 Diabetes (T2D), mimicking the long-term metabolic burden and multi-system progression.
High-Fat Diet	• chronic low-grade metabolic stress• gradual ROS accumulation• insulin resistance	Slow emergence of deficits associated with sustained metabolic syndrome and obesity.	Moderate; timeline depends on diet duration and subject susceptibility.	Replicates the gradual emergence of insulin resistance, obesity, and metabolic syndrome seen in human clinical populations.
db/db Mouse / Zucker Diabetic Fatty Rat	• Homozygous leptin receptor gene mutation• severe obesity• chronic hyperglycemia• systemic insulin resistance• sustained glial activation• long-term neuroinflammation	Long-term, progressive cognitive decline linked to sustained metabolic stress and altered cellular signaling over extended timelines.	High; reliable genetic onset and phenotypic consistency across subjects.	Replicates severe obesity and chronic metabolic dysfunction to study long-term neurodegenerative impact and cellular signaling changes.
Familial AD Models	• Inherited genetic mutations• amyloid plaque deposition• tau pathology	Progressive neurodegeneration directly driven by genetic mutation rather than metabolic status.	High; well-established timelines for genetic plaque and tangle development.	Translates specifically to rare familial forms of Alzheimer’s disease rather than metabolic or sporadic AD.

Genetic models further delineate metabolic versus inherited drivers of neurodegeneration. Genetic diabetic rodents (db/db mice and Zucker Diabetic Fatty rats) utilize receptor mutations to reliably produce chronic neuroinflammation, severe obesity, and long-term cognitive deterioration driven by sustained metabolic stress. By comparison, traditional familial AD genetic models focus on inherited mutations to reproduce amyloid plaques and tau pathology, making them highly reproducible for studying rare familial AD. Overall, diabetic rodent models provide vital experimental tools for isolating how systemic metabolic impairment and glucose dysregulation contribute to the pathogenesis of cognitive decline associated with hyperglycemia.

Ultimately, while the STZ model remains an excellent choice for isolating the direct impact of hyperglycemia on neuronal function, capturing the full spectrum of human diabetes-associated neurological decline frequently requires the strategic, complementary use of multiple experimental systems.

Author contributions

B.M.B. drafted the manuscript; B.M.B. and D.H.B. edited and reviewed the manuscript. All persons designated as authors qualify for authorship and all those who qualify for authorship are listed.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.