E4orf1 improves adipose tissue-specific metabolic risk factors and indicators of cognition function in a mouse model of Alzheimer's disease

Obesity, impaired glycemic control, and hepatic steatosis are each associated with increased risk of cognitive impairment and Alzheimer’s disease (AD). Insulin signaling impairments, hyperinsulinemia, and hepatic lipid accumulation are common to these conditions and may contribute to AD pathogenesis. Prior attempts to reduce AD risk via weight loss or anti-diabetic therapies have yielded limited success, and there are no approved drugs to prevent or treat hepatic steatosis. The authors hypothesized that simultaneously reducing adiposity, improving glycemic control, and lowering liver fat would attenuate cognitive decline. Given that adenoviral protein E4orf1 can bypass proximal insulin signaling to enhance distal Ras/PI3K/AKT signaling, promote GLUT4 translocation, improve glucose disposal, reduce endogenous insulin response, and prevent hepatic steatosis in mice, they tested whether adipose tissue–specific E4orf1 expression could ameliorate cognitive decline in APP/PS1 mice under diet-induced obesity.

The introduction reviews evidence linking obesity and metabolic dysfunction to dementia risk: higher BMI correlates with increased dementia and AD incidence; hyperinsulinemia is common in AD patients and predicts type 2 diabetes; obesity and hyperinsulinemia promote hepatic lipid accumulation and dysregulated hepatic metabolism that can accelerate AD. Despite metabolic improvements from weight loss, midlife weight reduction has not decreased incident AD, and trials of anti-diabetic agents (e.g., intranasal insulin, rosiglitazone, metformin-related insights) have shown limited or mixed efficacy for AD. There are currently no effective medications to reduce hepatic steatosis progression, suggesting a need for strategies that concurrently address adiposity, glycemic impairment, and liver fat. The E4orf1 protein from adenovirus 36 has been shown to enhance distal insulin signaling and glucose uptake independently of proximal insulin receptor signaling, reduce endogenous insulin response, and prevent hepatic steatosis in high-fat–fed mice, making it a candidate for testing metabolic-cognition interactions.

Design: Proof-of-concept study using a 20-week feeding protocol with doxycycline-inducible adipose-specific E4orf1 expression. Mice: Fourteen- to twenty-month-old male and female APP/PS1 mice bred with transgenic C57BL/6J mice expressing Ad36 E4orf1 in adipose tissue upon doxycycline induction. Groups: APP/PS1/E4orf1 (treatment; n=11) and APP/PS1 (control; n=7). Housing: ≤5/cage, 12-h light/dark, ad libitum food/water. Diets: 10 weeks high-fat diet with doxycycline (HFD-Doxy; 60% kcal fat; 600 ppm doxycycline/kg) followed by 10 weeks chow with doxycycline (16% kcal fat; 600 ppm doxycycline/kg). E4orf1 adipose-specific expression confirmed by immunoblot of inguinal (subcutaneous) adipose tissue. Blinding: Mice numbered; experimenter blinded to group. Measurements: - Body weight weekly; body composition at end via EchoMRI. - Glucose tolerance test (GTT): Oral glucose gavage (2 g/kg) after 4-h fast; blood glucose at 0, 15, 30, 60, 120 min via tail bleed; insulin measured at same time points by ELISA; AUC calculated. - HbA1c: Measured from 5 µl whole blood using commercial kit per manufacturer’s protocol. - HOMA-IR: Calculated as fasting glucose (mg/dL) × fasting insulin (ng/mL) × 0.072. Tissue collection: At 18–24 months of age, euthanasia by CO2 asphyxiation and cervical dislocation; collection of trunk blood (serum stored at −80°C), liver, adipose depots (inguinal, epididymal, retroperitoneal), skeletal muscle, and brain; tissues flash-frozen or stored in RNALater for protein/RNA analyses. Molecular assays: - RT-qPCR: RNA extracted (RNeasy Plus Universal kit); primers per Supplemental Table S1; relative expression by 2^-ΔΔCT method. - Western blotting: Protein lysates (RIPA buffer); densitometric analysis via ImageJ; normalization to GAPDH or total protein. - Aβ quantification: Sandwich ELISA for Aβ40 and Aβ42 in hippocampus, frontal cortex, and serum; tissue lysis in RIPA with protease inhibitors; protein normalization by BCA; 10–20 µg tissue protein and 2 µL serum loaded. Behavioral testing: Morris Water Maze (MWM) for spatial learning and memory in 17–23-month-old mice; four 60-s trials per day for two consecutive days (Trials 1–2), then Trials 3–4 conducted after a 7-day interval; probe analysis performed by removing platform to assess quadrant time. Statistics: Power calculation indicated 80% power at α=0.05 with n=4/group; to offset potential losses, >4 mice/group were used. Data presented as mean ± SEM. Group comparisons primarily by Welch’s t-test (unequal variances/sample sizes) to control type I error; significance thresholds reported per figure legends.

- Adiposity and glycemic control: Body weight did not differ between groups at study end (~18–24 months), but E4orf1 mice had significantly lower body fat percentage (p<0.05). HbA1c was significantly lower in E4orf1 mice (p<0.05). In GTT, E4orf1 mice cleared glucose significantly faster than controls (AUC p<0.001) and required significantly less endogenous insulin during the test (lower insulin AUC). Fasting and fed insulin levels were significantly lower in E4orf1 mice, and HOMA-IR was markedly reduced (p<0.001), indicating improved insulin sensitivity. - Adipose tissue signaling and metabolism: In inguinal WAT, E4orf1 increased total Ras and phospho-AKT (Ser473) protein abundance, indicating enhanced distal Ras/PI3K/AKT signaling. Adiponectin protein abundance was significantly higher. Gene expression showed reduced de novo lipogenesis (e.g., lower FASn), increased fatty acid oxidation (higher Cpt1a), improved lipid droplet biology (higher Cidea), reduced lipid synthesis/metabolism gene SCD1, and increased mitochondrial fusion gene MFN1 (no change in DRP1). - Liver lipid metabolism: E4orf1 mice had significantly reduced hepatic expression of de novo lipogenesis markers FASn, ACC, and Scd1 (protein and/or mRNA), indicating protection against hepatic lipid synthesis/accumulation. No significant differences were observed in hepatic phospho-AKT protein or fat oxidation genes. - Cognition and Aβ: In MWM, E4orf1 mice exhibited significantly improved escape latency on trial day 4 (p<0.007), indicating better spatial learning; probe trial quadrant time did not differ. Brain Aβ levels (hippocampus and cortex) for Aβ40/Aβ42 were not significantly different; serum Aβ40 was significantly reduced, while serum Aβ42 was unchanged. In cortex, E4orf1 mice showed significantly reduced RAGE expression and increased neprilysin (NEP), suggesting reduced Aβ production and/or enhanced degradation (no such changes in hippocampus). - Brain mitochondrial and neuronal markers: Hippocampus: no change in APP or phospho-AKT proteins; presynaptic gene synaptophysin increased; glycolytic gene fructokinase increased; no significant changes in mitochondrial biogenesis (Nrf1, Nrf2, PGC-1α), autophagy, fission, inflammation, or neurogenesis panels; mitochondrial fusion gene Mfn2 increased. Cortex: no change in APP or phospho-GSK3β proteins; mitochondrial fission gene Fis1 significantly reduced; neurogenesis genes NeuroD and DCX-1 increased; glycolytic gene enolase increased; no significant changes in synaptic genes, mitochondrial biogenesis, autophagy, fusion, or inflammatory genes.

The study demonstrates that adipose tissue–specific expression of E4orf1 improves systemic metabolic health in aged APP/PS1 mice—reducing adiposity, hyperglycemia, and hyperinsulinemia—while protecting against hepatic de novo lipogenesis and steatosis. These peripheral improvements were associated with better spatial learning performance in the Morris Water Maze, supporting the hypothesis that correcting obesity-related metabolic dysfunction can attenuate cognitive decline. Mechanistically, E4orf1 enhanced distal insulin signaling in adipose tissue (Ras/PI3K/AKT), increased adiponectin, shifted adipose gene expression toward reduced lipogenesis and increased oxidation, and curtailed hepatic lipogenic programs. In the brain, although tissue Aβ levels were unchanged, reduced cortical RAGE and increased NEP suggest a milieu favoring lower Aβ-mediated toxicity and enhanced Aβ degradation; serum Aβ40 was reduced. Improvements in neuronal and metabolic markers (increased synaptophysin in hippocampus; reduced Fis1 and increased NeuroD/DCX-1 in cortex; increased glycolytic enzymes) indicate preserved neuronal health and mitochondrial dynamics consistent with neuroprotection. Together, these findings link improved peripheral glycemic and lipid metabolism with indicators of improved brain function, aligning with the liver–brain and adipose–brain axis concepts. The authors caution that gene expression changes do not necessarily confirm functional outcomes and that the directionality and interdependence between peripheral and central changes remain to be fully elucidated.

This proof-of-concept study shows that targeting metabolic dysfunction—specifically glycemic impairment and hepatic steatosis—using adipose-specific E4orf1 can ameliorate obesity-related cognitive decline in an AD mouse model. E4orf1 improved glucose handling with reduced endogenous insulin, decreased adiposity, and suppressed hepatic lipogenesis, and was associated with better spatial learning and favorable brain molecular signatures (reduced cortical RAGE, increased NEP, enhanced neurogenesis-related genes). These findings identify E4orf1 as a candidate therapeutic to address obesity-related cognitive impairment and provide mechanistic insights into how peripheral metabolic impairments may contribute to AD. Future work should include longitudinal studies to define temporal relationships between peripheral and central outcomes, rigorous functional validation of brain molecular changes, evaluation across sexes with larger cohorts, and advancement of clinically relevant delivery modalities (e.g., nanoparticle-mediated E4orf1 delivery) toward preclinical development.

- Study design was cross-sectional; temporal causality between metabolic and cognitive changes cannot be established. - Aged cohort (18–24 months at endpoint) differs from many APP/PS1 behavioral studies conducted at 8–15 months; age may affect training/probe performance. - Modest sample size and potential sex effects were not fully powered to assess sex-specific outcomes. - Behavioral probe analysis may have been limited by training regimen. - Brain outcomes rely largely on gene expression; functional confirmation was not performed for all pathways. - Inability to determine whether brain changes are independent of or dependent on peripheral metabolic improvements. - Findings in transgenic mice may not directly translate clinically; delivery and safety of E4orf1 require further development.