Glycation-lowering compounds inhibit ghrelin signaling to reduce food intake, lower insulin resistance, and extend lifespan

SUMMARY

Non-enzymatic reactions in glycolysis produce methylglyoxal (MGO), a reactive precursor to advanced glycation end-products (AGEs), which has been hypothesized to drive obesity, diabetes, and aging-associated pathologies. A combination of nicotinamide, α-lipoic acid, thiamine, pyridoxamine, and piperine (Gly-Low) lowers the deleterious effects of glycation by reducing MGO and the MGO-derived AGE, MG-H1, in mice. Gly-Low supplementation in the diet reduces food consumption, decreases body weight while preserving muscle mass, improves insulin sensitivity, and increases survival in leptin receptor-deficient (Lepr^db) and wild-type C57B6/J mice. Transcriptional, protein, and functional analyses demonstrate that Gly-Low inhibits appetite stimulation through ghrelin and AMP-activated protein kinase (AMPK) signaling pathways in the hypothalamus, leading to reduced hunger responses. Consistent with these molecular findings, Gly-Low inhibits ghrelin-mediated hunger responses. As a late-life intervention, Gly-Low slows hypothalamic aging signatures, improves glucose homeostasis and motor coordination, and increases lifespan, suggesting its potential benefits in ameliorating age-associated decline.

In brief

Wimer et al. identify Gly-Low, a combination of glycation-reducing compounds, as a promising therapeutic to reduce food intake and extend lifespan in mice by targeting glycolytic detoxification and appetite-regulating pathways. These findings suggest a strategy for combating obesity and age-associated decline, offering an alternative to traditional caloric restriction.

Graphical Abstract

INTRODUCTION

Despite the substantial efforts of public health, the incidence of obesity is growing worldwide.¹ Obesity reduces life expectancy by increasing the risk of several diseases.² Lifestyle changes have gained popularity to combat increasingly sedentary lifestyles and excess caloric intake, but dietary improvements remain challenging for most individuals.^3,4 Part of this challenge is due to homeostatic mechanisms governing food intake and energy expenditure to resist body weight loss.⁵ Ghrelin and leptin are hormones that regulate food intake by promoting appetite and satiety, respectively.^6–8 Chronic dietary excess can disrupt leptin and ghrelin signaling impairing homeostatic regulation of food intake and thus promoting obesity.⁹

Food overconsumption and obesity are contributing factors to chronic hyperglycemia, which can alter glycolytic flux and thus increase the production of reactive α-dicarbonyls, such as methylglyoxal (MGO).^10,11 MGO is an unavoidable byproduct of anaerobic glycolysis and is generated through the non-enzymatic degradation of glycolytic intermediates, dihydroxy-acetone phosphate (DHAP) and glyceraldehyde-3-phosphate (G3P).¹² Once formed, MGO reacts non-enzymatically with biomolecules such as proteins, lipids, and DNA to form advanced glycation end-products (AGEs).^13,14 These covalent adducts have many deleterious consequences, such as impairing protein function, disrupting tissue architecture by forming extracellular crosslinks, activating inflammatory signaling cascades, altering cell-cell communication, and damaging nucleic acids.¹¹ Accumulation of these adducts, which occurs slowly with age or rapidly in hyperglycemic and obese individuals, drives many age-, obesity-, and hyperglycemia-associated pathologies. Cellular protection against AGEs occurs by endogenous glyoxalase enzymes, which detoxify MGO and thereby prevent downstream AGE formation.^10,15 We recently demonstrated using Caenorhabditis elegans that accumulation of AGEs also increases food intake.¹⁶ We therefore hypothesize that therapeutically enhancing detoxification of AGEs may protect against obesity and obesity-associated pathologies.

RESULTS

A natural compound screen identifies compounds that reduce glycation stress

Hyperglycemia, as often occurs in diabetes, accelerates AGE formation and its accumulation in various tissues, contributing to pathologies such as nephropathy, retinopathy, cardiomyopathy, neuropathy, and vascular injury.¹⁷ In C. elegans, the glod-4 mutant, which lacks the endogenous MGO detoxification pathway, develops diabetic-like manifestations, including peripheral neuropathy and reduced lifespan.¹⁸ To identify interventions protective against AGE-associated pathologies, we previously conducted a high-throughput screen of 640 natural compounds (TimTex, NPL640) using the glod-4 model. We further tested 11 hits from our screen, as well as compounds from relevant glyoxalase-associated literature (thiamine¹⁹ and pyridoxamine²⁰) in a cell culture model of glycation stress, as measured by rescuing neurite length retraction of rat dopaminergic (N27) neurons following exposure to MGO (Figure S1A). We found that a combination of five compounds: α-lipoic acid, nicotinamide, piperine, pyridoxamine, and thiamine, termed Gly-Low, conferred better protection against MGO relative to the single compounds (Figure S1B).

To test Gly-Low’s therapeutic potential, we tested it in the diabetic leptin receptor-deficient mouse model, Lepr^db, which rapidly develops obesity, hyperglycemia, and glucose intolerance, alongside increased glycation precursors and their adducts in diabetes-relevant tissues such as the kidneys, heart, and liver.²¹ We started 8-week-old male Lepr^db mice on a regular, low-fat diet supplemented with Gly-Low for 16 weeks (Figure 1). LC/MS analysis of plasma demonstrated that Gly-Low significantly lowered absolute levels of both MGO (61% decrease) and its protein-bound arginine adduct, MG-H1 (41% decrease), compared to controls (Figure 1A), confirming its glycation reducing effects in vivo.

Figure 1. Glycation-lowering compounds (Gly-Low) lower glycolytic byproducts as well as rescue hyperphagia and obesity-associated pathologies in a diabetic, leptin receptor-deficient (Lepr^db) mouse model.

(A) Absolute levels of glycolytic byproduct, MGO (left), and its arginine adduct, MG-H1 (right), were reduced in the plasma of male mice treated with Gly-Low. n = 5 per treatment group, measured by LC/MS.

(B and C) Weekly food consumption (B) and body weights (C) of male Lepr^db mice fed diets containing varying concentrations of Gly-Low were altered in a dose-dependent manner. Food measurements of n = 3 cages of group-housed animals per treatment (n = 5 in the control group). Body weights of n = 12 mice per treatment group (n = 17 in the control group). Bar graphs (right) show food consumption differences (B) and percent change in body weight from starting diets

(C) at their respective timepoint relative to control.

(D) Percent change from baseline in fat (left) and lean mass (right) of Lepr^db mice on their respective diet.

(E) Wet weights of inguinal fat deposits and livers of Lepr^db mice.

(F) Representative images of H&E-stained liver sections from control and 1× Gly-Low-treated mice showing the presence of large lipid vacuoles (arrowhead). Quantification of liver lipid vacuoles, n = 3–4 livers per treatment group, 4 fields^# of view per animal.

(G) Random (non-fasted) and fasted (16-h) blood glucose levels up to a max of 600 mg/dL. n = 12 per treatment group (n = 17 in the control, with deaths throughout the study [1 death before 8 weeks, 8 deaths before 16 weeks]).

(H) Survival curves of control mice and Gly-Low-treated mice prior to their experimental endpoint. p = 0.007 (log rank Mantel-Cox), p = 0.007 (Gehan-Breslow-Wilcoxon). Significance: ns (not significant), *p < 0.05, **p < 0.005, ***p < 0.0005. Statistical analyses performed by unpaired t test. a, b, and c designate statistical significance (p < 0.05) between the control group and 1×, 0.5×, and 0.25× Gly-Low treatments, respectively. Data are represented as mean ± SEM.

Glycation-lowering compounds reduce food intake and diabetic pathologies in Lepr^db mice

All concentrations of Gly-Low (0.25×, 0.5×, and 1×) reduced food consumption during the first 7 weeks on the diet, with a dose-dependent effect observed for the first 6 weeks (Figure 1B). The loss of statistical significance for lower doses (0.5× and 0.25×) after 6 weeks may be attributed to the increased variability in food consumption among control fed mice later in the study (Figure S2A). Gly-Low had an early dose-dependent effect on body weight gain that remained significant only for the highest dose (1×) over time (Figure 1C). By 8 weeks of age, which is the age at which treatment began, Lepr^db mice were already over-weight. During the treatment period, control-fed Lepr^db mice continued gaining weight, reaching an average of 45 grams by 8 weeks of treatment, roughly 50% heavier than wild-type (C57B6/J) mice, reflecting severe obesity. In contrast, 1× Gly-Low-treated mice lost an average of 22% of their starting weight, stabilizing at 30 grams, which is comparable to wild-type male mice of the same age.²² Mice fed the 0.5× Gly-Low diet maintained their weight, while those on 0.25× Gly-Low continued gaining weight, failing to prevent pathological obesity.

To determine whether weight loss was due to fat or lean mass reduction, we performed dual X-ray absorptiometry (DXA), which directly measures bone and fat mass and calculates lean mass. After 8 weeks of treatment, control-fed mice gained an average of 10% fat mass while 1× Gly-Low-treated mice lost an average of 13% (Figure 1D, left). Only the highest dose (1×) of Gly-Low preserved lean mass, while lower doses failed to prevent lean mass loss (Figure 1D, right). To assess whether increased activity or metabolic rate contributed to weight loss of 1× Gly-Low-treated mice, we performed metabolic cage testing, which revealed no significant differences in activity or metabolic rate between groups (Figures S2B and S2C).

Given the pronounced reduction in food intake with 1× Gly-Low treatment, we performed several tests to assess potential food aversion. A formal conditioned taste aversion (CTA) test was not feasible due to incompatible solubility of Gly-Low compounds for IP injection. Instead, we used food restriction and re-feeding experiments and tested Gly-Low in palatable diets. In wild-type (C57B/6J) mice, re-feeding rates after an 18-h fast were statistically similar whether mice were re-fed a control diet or a Gly-Low-supplemented standard or high-fat (HFD) diet (Figures S3A and S3B). Additionally, 0.5× Gly-Low reduced food consumption in both wild-type and Lepr^db mice (Figure S3C). However, when 0.5× Gly-Low was incorporated into a palatable HFD, it had no effect on food consumption in wild-type mice (Figure S3C). These findings, combined with long-term treatment data (∼10 months) showing converging food consumption rates over time (Figure 2B), suggest that Gly-Low does not induce taste aversion. However, we acknowledge that formal CTA tests remain the gold standard for taste aversion testing and therefore cannot rule out the possibility that taste aversion contributes to Gly-Low’s effect on food intake. Although Gly-Low affected feeding behavior and body weight across multiple concentrations, we focused subsequent in vivo analyses on the highest dose due to its effectiveness and robust reductions in circulating AGE levels.

Figure 2. Gly-Low and its constituents reduce body weights and food consumption in male and female C57B6/J mice, independent of major pituitary hormones and hypothalamic leptin signaling.

(A and B) Body weights (A) and food consumption (B) of male wild-type mice chronically (24 weeks) treated with Gly-Low were reduced compared to control mice.

(C and D) Body weights (C) and food consumption (D) of male wild-type mice acutely (1 week) treated with Gly-Low were reduced compared to control mice.

(E and F) Body weights (E) and food consumption (F) of female wild-type mice acutely (1 week) treated with Gly-Low were reduced compared to control mice.

(G and H) Body weights (G) and food consumption (H) of male wild-type mice acutely (1 week) treated with individual compounds of Gly-Low compared to control mice. Boxplots to the right show changes from starting to ending weights within the respective treatment group.

(I and J) Plasma levels of growth hormone (GH), thyroid-stimulating hormone (TSH), adrenocorticotropic hormone (ACTH), and prolactin (PRL) from male mice acutely treated with Gly-Low constituents (I) and from mice acutely treated with Gly-Low (J).

(K) Food consumption rates (kcal) of wild-type mice fed a control diet or a Gly-Low diet were reduced following an injection of leptin compared to those given a saline vehicle control. Bar graphs (right) show differences in food consumption between treatment groups at 1 and 32 h after an injection of leptin compared to their saline food consumption. Significance: ns p > 0.05, *p < 0.05, **p < 0.005, ***p < 0.0005. Statistical analyses: paired t test for (G); unpaired t test for all other panels. Data are represented as mean ± SEM.

Consistent with the reduced adiposity observed by DXA analysis, Gly-Low-treated mice had significantly smaller inguinal fat pads (Figure 1E). Liver weights were also significantly lower in Gly-Low-treated mice, and histological analysis showed 89.8% reduction in large lipid vacuoles, indicating improved lipid homeostasis (Figures 1E and 1F).

In agreement with reduced glycation burden, Gly-Low-treated mice displayed significantly improved glycemic control, with 9.9% lower fasted glucose and 28.4% lower random blood glucose after 8 weeks, which persisted throughout treatment (Figure 1G). This was accompanied by reduced polyuria, measured by weighing home cages to determine cage soiling (Figure S2D). Polyuria in diabetes results from excessive glucose in the urine due to excessive blood sugar levels.²³ In mouse models of diabetes, excessive urine output can contribute to differences in cage weight due to excessive soiling of bedding and nesting material and thus serve as a proxy for changes in polyuria.²⁴ Notably, water intake was unchanged between Gly-Low- and control-fed mice, confirming that reduced soiling was not due to lower fluid consumption (Figure S2E). Furthermore, Gly-Low-treated mice exhibited reduced proteinuria, a marker of diabetic kidney damage.²⁵ Control-fed mice displayed increased urinary protein levels over time, indicative of worsening kidney function, whereas, Gly-Low-treated mice had reduced urinary protein, suggesting a protective role within the kidney (Figure S2F).

The benefits of Gly-Low in reducing hyperphagia and diabetes-associated pathology translated to a complete rescue of early mortality (Figure 1H). Previous studies report a median life-span of 349 days for male Lepr^db mice, with continuous mortality after 16 weeks of age.²⁶ In our study, 18% of control-fed mice died by 23 weeks of age (15 weeks on diet), and chance of survival dropped to 52.9% by 26 weeks. In contrast, Gly-Low-treated mice had no recorded deaths during this period. Collectively, these findings in leptin receptor-deficient mice highlight the therapeutic potential of Gly-Low in treating various obesity- and diabetes-associated conditions by reducing overconsumption, associated glycation burden, and glycation-associated pathologies.

Gly-Low reduces food intake in a leptin-independent manner

Following experimentation in a leptin receptor-deficient mouse model (Lepr^db), we tested the highest dose of Gly-Low in 4-month-old wild-type (C57BL/6J) mice. Long-term treatment of male wild-type mice resulted in similar reductions and subsequent maintenance of body weights (Figure 2A) and food consumption (Figure 2B). After 6-month treatment, Gly-Low-treated mice lost an average of 13.2% of their body weight, while control-fed mice gained an average of 16.9%. As observed in Lepr^db mice, metabolic cage testing confirmed that weight loss was primarily due to reduced food intake rather than changes in activity or metabolic rate (Figures S3D and S3E). To test for sex specificity, we conducted a 1-week acute treatment in age-matched, wild-type males and females, both of which showed similar reductions in body weight and food consumption (Figures 2C–2F). After treatment, young (3-month-old) males lost an average of 11.7% of their body weights, while females lost an average of 11.1%, indicating that Gly-Low’s effects on food consumption and body weight are not sexually dimorphic.

Next, we determined which Gly-Low constituents drive its effects on food consumption and/or body weight. Each compound comprising Gly-Low was incorporated into a standard low-fat diet at the same dose as in the 1× Gly-Low diet, and young (6-month-old) wild-type male mice were acutely treated for 1 week. At the end of treatment, α-lipoic acid and nicotinamide treatment caused significant weight loss, while thiamine caused significant weight gain (Figure 2G). Notably, α-lipoic acid had the strongest effect on food consumption, with mice consuming roughly 32% less food than control-fed mice and driving the most significant weight loss (Figure 2H). Interestingly, this effect was less pronounced in Gly-Low-treated mice, suggesting that other components temper α-lipoic acid’s appetite-suppressing effects. These findings indicate that α-lipoic acid is the primary driver of appetite suppression, while nicotinamide and thiamine contribute to weight change through additional mechanisms.

To determine how Gly-Low and its individual components influence feeding behavior and body weight, we measured circulating hormone levels of the hypothalamic-pituitary axis, which directly regulate body weight maintenance or feeding behavior. Using a multiplex hormone panel, we analyzed growth hormone (GH), thyroid-stimulating hormone (TSH), adrenocorticotropic hormone (ACTH), and prolactin (PRL). GH secretion decreases during feeding.²⁷ GH levels were significantly elevated in piperine-treated mice and trended lower in nicotinamide-treated mice (Figure 2I, left). However, GH levels were unchanged in Gly-Low-treated mice (Figure 2J, left), suggesting that piperine’s effects are tempered by co-administration with other components of Gly-Low. Hypothyroidism is often observed during reduced TSH production and release, leading to decreased basal energy expenditure and increased body weights.²⁸ Surprisingly, TSH levels trended lower in nicotinamide- and α-lipoic acid-treated mice (Figure 2I, middle left) but were unaffected in Gly-Low-treated mice (Figure 2J, middle left). ACTH regulates cortisol and androgen production.²⁹ Following feeding, plasma levels of ACTH are significantly reduced.³⁰ All Gly-Low components trended toward lower ACTH levels, and the full Gly-Low cocktail caused a significant reduction in circulating levels of ACTH (Figures 2I and 2J, middle right panels). These findings suggest additive effects of the individual components. PRL levels were significantly lower in nicotinamide-treated mice and trended lower in α-lipoic acid-, pyridoxamine-, and thiamine-treated mice. However, PRL levels were unchanged in Gly-Low-treated mice. These findings indicate that Gly-Low’s individual components influence feeding-related hormones differently, with additive or negating effects when combined. α-lipoic acid appears to be the primary driver of appetite suppression, while nicotinamide may regulate body-weight independently of food consumption. This aligns with observations in Lepr^db mice, where nicotinamide alone reduced body weight without affecting food intake (Figure S2G). These results suggest that similar mechanisms underlie Gly-Low’s effects across different models.

To determine whether Gly-Low acts independently of leptin signaling in wild-type mice, we injected exogenous leptin into 6-month-old wild-type males and measured food consumption compared to saline-injected controls. Exogenous leptin stimulates pSTAT3 signaling within hypothalamic neurons, and a readout of this signaling can be measured by a reduction in feeding behavior. Both Gly-Low- and control-fed mice showed similar reductions in food consumption following leptin injections, indicating intact leptin signaling (Figure 2K). Gly-Low-treated mice showed a greater initial increase in food consumption (1 h post-injection), but differences between groups disappeared by 32 h post-injection. pSTAT3 immunostaining in hypothalamic neurons within the arcuate nucleus confirmed that Gly-Low-treated mice exhibited a comparable leptin response to controls (Figure S3F). These findings demonstrate that Gly-Low reduces food intake through a leptin-independent mechanism.

Gly-Low reduces glycolysis and enhances cellular detoxification pathways in the hypothalamus

The hypothalamus surrounds the third ventricle, allowing it to sense circulating nutrients and hormones from the periphery to regulate body-weight homeostasis.^31,32 Thus, we interrogated the hypothalamus for molecular changes induced by Gly-Low. We assessed changes in hypothalamic transcripts using bulk RNA-sequencing and performed protein analysis on hypothalamic lysates from 3-month-old wild-type mice untreated or acutely treated (1 week) with Gly-Low.

We investigated how Gly-Low influenced the expression of pathways relevant to glycation production and clearance. We found that Gly-Low-treated mice showed a general reduction in the expression of glycolytic genes, except for Tpi and Pgk1, whose increased activity is predicted to lower MGO production^33,34 (Figures 3A and S4A). In contrast, genes directly responsible for MGO detoxification (glyoxalase genes) and a larger set of general cellular detoxification genes showed increased expression in Gly-Low-treated mice (Figures 3A and S4A). Consistent with our transcriptomic findings, targeted plasma metabolomics data revealed decreased glycolytic metabolites and elevated pentose phosphate pathway (PPP) metabolites. (Figure S4B). The PPP regenerates NADPH, which is necessary for replenishing glutathione.^35,36 This suggests that Gly-Low increases the potential for MGO-specific and generalized cellular detoxification. We hypothesize that Gly-Low may reduce glycation burden by reducing its production and enhancing its clearance. Furthermore, these findings in the hypothalamus of wild-type mice complement our findings that Gly-Low reduces systemic burden of the glycation precursor, MGO, and its predominant AGE, MGH1, in the blood of Lepr^db mice.

Figure 3. Gly-Low treatment alters genes responsible for appetite stimulation and blunts hypothalamic ghrelin and downstream AMPK signaling.

(A) Heatmaps showing transcript fold changes of genes involved in MGO production (glycolysis) and clearance (glyoxalase) in the hypothalamus of wild-type male mice acutely (1 week) treated with Gly-Low.

(B) Heatmap showing transcript fold changes of negative and positive appetite regulators in the hypothalamus of mice acutely (1 week) treated with Gly-Low.

(C) Plasma levels of acylated ghrelin were unchanged with Gly-Low treatment.

(D and E) Plasma levels of ghrelin-receptor antagonist, LEAP-2 (D), and ghrelin-stimulated growth factor, IGF-1 (E), were reduced with Gly-Low treatment.

(F and G) Exogenous ghrelin increased food consumption compared to saline injected controls in control (F) but not Gly-Low-treated (G) mice.

(H) Schematic depicting activation or inhibition of mTOR signaling and ribosomal translation by hormone/growth factor or ghrelin-dependent GHSR signaling, respectively.

(I) Food consumption rates (kcal/g BW) following IP injection of AMPK activator, AICAR, compared to saline vehicle controls were increased in control-fed mice but significantly decreased in Gly-Low-fed mice.

(J) Western blots of hypothalamic lysates from mice acutely (1 week) treated with Gly-Low. Protein expression of phosphorylated AMPK (top), phosphorylated AKT (middle), and phosphorylated S6 kinase (bottom) relative to their un-phosphorylated protein levels with quantification (right). Uncropped western blot images with ladder are included in supplemental information.

(K) Transcripts from ribosomal genes were significantly upregulated in hypothalamic lysates of mice acutely (1 week) treated with Gly-Low compared to control mice. Significance: ns p > 0.05, *p < 0.05, **p < 0.005, ***p < 0.0005. Statistical analyses performed by unpaired t test. a and b designate statistical significance (p < 0.05) between AICAR-injected and saline-injected groups fed either a control diet or Gly-Low diet, respectively. Data are represented as mean ± SEM.

To investigate how Gly-Low may impact feeding behavior, we analyzed the 1,407 genes differentially expressed in the hypothalamus of Gly-Low-treated vs. control mice for changes to canonical feeding behavior genes. Interestingly, two well-characterized regulators of hunger and satiety, Agrp and Pomc, were consistently changed in the direction typically observed in hungry mice³⁷ (Figure 3B). Analysis of this set of positive and negative regulator feeding genes in Gly-Low-treated mice indicates that no one particular pathway was expressed at higher or lower levels than expected (Figure S4C).

Gly-Low inhibits ghrelin signaling and alters activation of hypothalamic AMPK

The stomach-derived hormone ghrelin is a common signaling molecule to both the Agrp and Pomc pathways. Ghrelin is a peptide produced in the stomach during fasting and times of hunger.^38,39 Following production, the circulating hormone binds its receptor, the GH secretagogue receptor (Ghsr),⁴⁰ within the hypothalamus to engage signaling pathways that stimulate food intake.^41,42 To determine whether Gly-Low affected ghrelin production, we measured levels of acylated ghrelin in the plasma of acutely treated 4-month-old male wild-type mice. We found no significant difference in ghrelin levels between the treatment groups (Figure 3C). Ghrelin signaling is inhibited by competitive binding of the ghrelin receptor by the liver-derived hormone LEAP2⁴³ or when ghrelin production is reduced during feeding.^44,45 However, LEAP-2 levels were significantly lower in Gly-Low-treated mice (40% lower; Figure 3D), suggesting the absence of competitive Ghsr binding. Ghrelin signaling stimulates the release of GH,³⁹ which in turn stimulates the release of IGF-1 in a well-studied GH/IGF-1 endocrine axis within the hypothalamus and pituitary gland.⁴⁶ To test if this ghrelin-related pathway is disrupted in Gly-Low-treated mice, we measured insulin-like growth factor (IGF-1) in the plasma of control and Gly-Low-treated mice. Consistent with disrupted ghrelin signaling, IGF-1 levels were significantly lower in Gly-Low-treated mice relative to control mice (10.5% lower; Figure 3E). To test if ghrelin signaling is indeed disrupted in Gly-Low-treated mice, we subjected control and Gly-Low-treated mice to exogenous acylated ghrelin, known to activate ghrelin signaling, increase appetite, and induce feeding behavior.⁴⁷ In control mice, those injected with ghrelin consumed significantly more food (average of 75% increase) relative to PBS-injected controls post-IP injection (Figure 3F). In contrast, Gly-Low-treated mice failed to respond to ghrelin injections with increased food consumption (an average of 23% less food consumed relative to PBS-injected controls; Figure 3G). These findings suggest a role for impaired ghrelin signaling in the reduced food intake observed upon Gly-Low treatment.

Upon ghrelin signaling, AMPK is phosphorylated and activated⁴⁸ (Figure 3H). To determine whether Gly-Low suppresses hypothalamic AMPK activation, we challenged 5-month-old male mice with the AMPK activator, 5-aminoimidazole-4-carboxamide ribonucleoside (AICAR), after an 18-h fast. IP-injections of AICAR increased food consumption in mice adapted to a control diet but decreased food consumption in mice adapted to a Gly-Low diet (Figure 3I). To assess hypothalamic AMPK activity, we measured ACC phosphorylation (pACC), a direct AMPK substrate in hypothalamic lysates. In mice adapted to a control diet, pACC levels significantly increases with fasting and trended higher 1 h post-AICAR injection (Figure S4D). However, this response was reversed in mice adapted to a Gly-Low diet. Together, these findings indicate that Gly-Low inhibits AMPK activation in response to both energy depletion (high AMP:ATP ratio) associated with fasting and in response to AMP analogs like AICAR.

Gly-Low increases ribosomal protein S6 phosphorylation in the hypothalamus

In parallel with Ghsr-dependent AMPK signaling, growth factors, such as GH and IGF-1, counteract ghrelin by binding their respective hypothalamic receptors, promoting AKT phosphorylation (pAKT; Figure 3H). pAKT stimulates mechanistic target of rapamycin (mTOR), leading to S6 kinase phosphorylation, enhanced protein synthesis, and appetite suppression.⁴⁹ Importantly, activated (phosphorylated) AMPK directly and indirectly inhibits mTOR activity, highlighting the antagonistic relationship between these two pathways.^50–52 Consistent with impaired ghrelin signaling and reduced AMPK activity, pAMPK levels were significantly lower in hypothalamic lysates from 4-month-old Gly-Low-treated mice, while phosphorylated S6 (pS6) levels were increased (Figure 3J). To further assess mTOR/S6 pathway activation, we analyzed hypothalamic RNA-seq data, which revealed a widespread upregulation of ribosomal transcripts (Figure 3K). This suggests enhanced ribosomal biogenesis and protein translation, hallmarks of mTOR/S6 signaling. Supporting this, Fcf1 and Nip7, key regulators of ribosomal rRNA processing, were also significantly upregulated in Gly-Low-treated mice (Figure S4E). Interestingly, despite increased pS6 levels, pAKT was significantly reduced in the hypothalami of Gly-Low-treated mice (Figure 3J), consistent with lower IGF-1 levels. This suggests that mTOR activation in Gly-Low-treated mice likely results from reduced inhibition by AMPK rather than AKT stimulation. These findings suggest that Gly-Low reduces food consumption by impairing appetite-stimulating ghrelin signaling and modulating AMPK-associated appetite-stimulating signaling.^53,54

Gly-Low rescues aging-associated dysregulations in glucose homeostasis, motor coordination, and extends lifespan as a late-life treatment

Glycation stress increases with age, accelerating aging-associated conditions.^55–57 To test whether Gly-Low (1× dosage) mitigates age-related decline, we assessed glucose metabolism and motor function in middle-aged (12 months old; 8 months of treatment) and aged (24–5 months old; 4–5 months of treatment) wild-type male mice. Like young wild-type mice, aged mice treated with Gy-Low exhibited a rapid reduction in food intake and body weight relative to control-fed counterparts. While food intake was significantly reduced during the first 12 weeks of treatment, and body weight declined initially, then weight stabilized and remained lower than controls for the remainder of the experiment (Figures S4F and S4G). Similar to improved glucose homeostasis observed in Lepr^db mice, Gly-Low-treated middle-aged mice showed improved glucose metabolism during a glucose tolerance test (Figure 4A). Following an injection of exogenous glucose, Gly-Low-treated mice displayed a 47.0% lower maximum blood glucose level and a 45.4% lower area under the curve (AUC) compared to controls. Additionally, insulin response testing revealed that Gly-Low-treated mice had a 62.7% greater reduction in blood glucose from baseline, compared to 33.5% in controls (Figure 4B). With age, wild-type C57B6/J mice develop dysregulated glucose homeostasis, reflected in increased fasted and non-fasted blood glucose levels.⁵⁸ Gly-Low significantly reduced fasting blood glucose levels in both middle-aged and aged mice (Figures 4C and 4E). To assess age-related motor function decline, we performed rotarod testing, commonly used to evaluate sensorimotor function and neuromuscular function, which declines with age.^59–61 Gly-Low-treated mice outperformed controls in both middle-aged and aged cohorts (Figures 4D and 4F), indicating preserved motor coordination with treatment.

Figure 4. Gly-Low rescues aging-associated dysregulation in glucose homeostasis, motor coordination, and extends lifespan as a late-life intervention.

(A) Gly-Low-treated mice had significantly reduced surges in glucose levels during a glucose tolerance test (GTT) compared to control mice. AUC analysis (right).

(B) Glucose levels were decreased to a greater degree in mice treated with Gly-Low than control mice when administered a bolus of insulin during insulin tolerance testing (ITT). AUC analysis (right).

(C and E) Fasted blood glucose levels were reduced in middle-aged (12 months) (C) and aged (25 months) (E) wild-type male mice treated with Gly-Low compared to control mice.

(D and F) Rotarod performance was improved in middle-aged (12 months) (D) and aged (25 months)

(F) Gly-Low-treated mice compared to control mice.

(G) Kaplan-Meier curve showing lifespan extension as a late-life intervention (beginning at 24 months of age) in wild-type male Gly-Low-treated compared to control mice.

(H) Scatterplot showing the log2 fold changes of genes significantly (p < 0.05) altered in the hypothalamus of aged (25 months) Gly-Low-treated vs. aged (25 months) control and aged (25 months) control vs. young (3 months) control mice. The regression line is shown in red. The Pearson correlation coefficient, r, is shown in the top right quadrant. Significance: *p < 0.05, **p < 0.005, ****p < 0.00005. Statistical analyses performed by unpaired t test. Data are represented as mean ± SEM.

Given its effects on functional aging, we next examined whether Gly-Low extends lifespan as a late-life intervention. A cohort of 24-month-old male C57BL/6J mice was placed on a Gly-Low diet, with natural lifespan and deaths recorded, in which mice were only removed from the study based on stringent health parameters. Survival analysis (Kaplan-Meier curve) showed that control mice had a median lifespan of 825 days (105 days post-intervention), whereas Gly-Low-treated mice lived a median of 888 days (173 days post-intervention; Figure 4G). This represents an 8.25% increase in median lifespan (p = 0.0199, Mantel-Cox) and a 5.23% increase in maximum life-span when treatment started at 24 months of age. This also translates to a 60.7% increase in survival time post-intervention. Unlike caloric restriction, which does not extend lifespan when initiated late in life, Gly-Low’s effects are likely not solely driven by reduced food intake.^62,63

Since hypothalamic aging drives whole-body aging,^64–68 we assessed transcriptional differences in the hypothalamus of aged (25 months) Gly-Low-treated and control mice. From the differentially expressed genes found in this dataset, we performed linear regression against the genes identified when we compared aged control mice to young (3-month) control mice (Figure 4H). This regression shows a negative correlation (r = 0.8, p < 0.0001) between the two comparisons, suggesting that some changes in hypothalamic transcription seen in normal aging are reversed with a late-life intervention of Gly-Low.

DISCUSSION

Here, we highlight Gly-Low as a potential therapeutic for obesity- and diabetes-associated pathologies using various mouse models. We selected the compounds that make Gly-Low based on their ability to protect against glycation stress. For that reason, we chose to test Gly-Low in the well-studied, glycation-burdened Lepr^db mouse model of hyperphagia and obesity.²¹ As hypothesized, Gly-Low significantly reduced systemic levels of glycation stress, reducing levels of the glycation precursor, MGO, and its arginine-adduct AGE, MG-H1. Our findings, along with previous work in C. elegans, support the potential of glycation lowering compounds as a therapeutic strategy against diabetic pathologies.^18,69 MGO is a well-known target for aging and age-related diseases,^55–57 but it has been difficult to target pharmacologically.⁶⁹ Here, we show that Gly-Low likely reduces MGO and MG-H1 through multiple mechanisms, including lowering glycolytic production and enhancing cellular detoxification pathways.

Beyond reducing glycation stress, we found that Gly-Low suppresses appetite, leading to reduced caloric intake without loss in relative muscle mass. Given that Lepr^db mice are both glycation burdened and hyperphagic, Gly-Low expectedly improved multiple pathological phenotypes. However, Gly-Low also had health-promoting effects in young, middle-aged, and aged wild-type mice, suggesting broader therapeutic potential. Our data suggests that Gly-Low’s health benefits likely stem from both its glycation-lowering and calorie-reducing effects, which are closely linked. We previously found that MG-H1 increases feeding in C. elegans,¹⁶ while others have shown that MGO-modified bovine serum albumin induces insulin resistance, weight gain, and shortens lifespan in mice.⁷⁰ In a clinical trial, glyoxalase-activating compounds that detoxify MGO reduced body weight and improved insulin sensitivity.²⁷ We propose that Gly-Low influences feeding behavior through multiple mechanisms, including suppression of the appetite-stimulating ghrelin pathway and activating the appetite-suppressing mTOR pathway in the hypothalamus. While our data supports mTOR pathway activation, further studies measuring translation flux and ribosomal biogenesis are needed to fully validate the extent of mTOR activation.

Caloric restriction (CR) is one of the most potent and widely conserved interventions for increasing healthspan and lifespan across species.⁷¹ While some of Gly-Low’s benefits may come from calorie reduction, our findings suggest that it acts through distinct mechanisms. One key difference is Gly-Low’s ability to extend lifespan even when initiated late in life. CR’s lifespan benefits depend on the age at which CR is initiated, with late-life CR showing little or no effect on longevity.^62,63 Lipman et al. reported no median lifespan extension in rats subjected to 33% CR starting at 18 months⁶² and demonstrated that CR rats had increased mortality rates compared to ad libitum (AL) fed rats (61% and 47%, respectively; p = 0.6), concluding that there may be a stage in the aging process after which CR no longer increases longevity.⁶² Additionally, Hahn et al. compared large cohorts of mice (800 animals) chronically (beginning at 12 weeks of age) fed an AL diet to those chronically fed an AL diet prior to switching to a dietary restriction (DR) diet at 24 months of age.⁶³ They reported that late on-set DR caused no measurable increase in survival in a large fraction of old animals, with observed increases being largely driven by a single breeding cohort.⁶³ In contrast, we found that Gly-Low treatment at 24 months increased both median and maximum lifespan, with mice voluntarily eating 13.4%–29.6% less than controls. While our study differs from traditional CR paradigms, these findings warrant further investigation. We hypothesize that Gly-Low promotes metabolic health and anti-aging effects by altering hypothalamic signaling, which, in turn, regulates both feeding behavior and systemic aging phenotypes.^64–67 Our transcriptional analysis of the hypothalamus in aged Gly-Low-treated mice indicates that many age-related gene expression changes are reversed by treatment (Figure 4H).

While CR is known to improve metabolic health and slow aging,⁷² long-term CR is difficult to maintain in humans.⁷³ We propose that lowering glycation stress with therapies such as Gly-Low may offer a sustainable alternative to enforced CR by providing voluntary calorie reduction, direct tissue protection, and slowing of aging-related outcomes linked to glycation stress due to methylglyoxal and related precursors.

Limitations of the study

Although the current study indicates that Gly-Low inhibits hypothalamic ghrelin and AMPK signaling to mechanistically contribute to the observed reduction in feeding behavior, we cannot fully exclude the potential contribution of taste aversion. In support of our hypothesis, we provide data that Gly-Low did not affect immediate feeding behavior in mice that were food restricted for 18 h and then re-introduced to either a control diet or a diet containing Gly-Low, with alterations in food intake only emerging after several hours (Figures S3A and S3B). However, standard CTA protocols were not conducted in our cohorts. Future studies employing rigorous CTA assessments will be necessary to distinguish between Gly-Low’s effect on feeding behavior and potential aversive responses. Additionally, while our data highlight Gly-Low’s impact on mitigating age-associated decline in male mice, we did not collect data from aged female mice. Given the significant influence of hormonal changes during aging in females, future studies should evaluate Gly-Low’s efficacy in female aging models. Finally, the individual contributions of Gly-Low’s components to glucose metabolism, ghrelin signaling, and aging remain to be fully defined. While we assessed the effect of individual compounds on food consumption rates, body weights, and abbreviated hormonal analysis, future studies testing individual components in varied combinations will be essential to determine whether certain combinations produce synergistic effects for specific outcomes.

RESOURCE AVAILABILITY

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Dr. Pankaj Kapahi (pkapahi@buckinstitute.org).

Materials availability

This study did not generate new unique reagents or mouse lines.

Data and code availability

• FASTQ files for raw bulk RNA-seq were deposited at the Gene Expression Omnibus (GEO) and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.

• Raw metabolomics files were deposited at the Mass Spectrometry Interactive Virtual Environment (MassIVE) and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact and or corresponding author upon request.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse Akt (pan) (40D4)	Cell Signaling	Cat# 2920; RRID: AB_1147620
Rabbit monoclonal pAKT Ser473 (D9E)	Cell Signaling	Cat# 4060; RRID: AB_2315049
Rabbit monoclonal S6 Ribosomal protein (5G10)	Cell Signaling	Cat# 2217; RRID: AB_331355
Rabbit monoclonal pS6 Ribosomal protein (Ser 240/244) (D68F8)	Cell Signaling	Cat# 5364; RRID: AB_10694233
Rabbit monoclonal AMPKα	Cell Signaling	Cat# 2532; RRID: AB_330331
Rabbit monoclonal pAMPKα (Thr172) (D4D6D)	Cell Signaling	Cat# 50081; RRID: AB_2799368
Rabbit monoclonal mTOR (7C10)	Cell Signaling	Cat# 2983; RRID: AB_2105622
Rabbit monoclonal β-Actin (13E5)	Cell Signaling	Cat# 4970; RRID: AB_2223172
Rabbit monoclonal Acetyl-CoA Carboxylase (ACC) (3662)	Cell Signaling	Cat# 3662; RRID: AB_2219400
Rabbit polyclonal Phospho-Acetyl-CoA Carboxylase (pACC) (Ser79) (D7D11)	Cell Signaling	Cat#11818; RRID: AB_2687505
Chemicals, peptides, and recombinant proteins
Alpha-lipoic acid	Sigma	Cat# 62320–25G-F
Nicotinamide	Sigma	Cat# 72345–50G
Thiamine hydrochloride	Sigma	Cat# T4625–10G
Pyridoxamine dihydrochloride	Chem-IPEX INT’L	Cat# 1461
Piperine	Sigma	Cat# 1003442421
Insulin	Sigma-Aldrich	Cat# I2643–50MG
D-Glucose	Sigma-Aldrich	Cat# G8270–100G
Ghrelin peptides	Phoenix Pharmaceuticals	Cat #031–30
Methylglyoxal (MGO)	Sigma Aldrich	Cat #M0252
AICAR	Toronto Research Chemicals	Cat# A611700
Critical commercial assays
LEAP-2 (38–77) (Human)/LEAP-2 (37/76) (Mouse) ELISA Kit	Phoenix Pharmaceuticals	Cat #075–40
Rat/Mouse Ghrelin active ELISA Kit	Sigma Aldrich	Cat #EZRGRA-90K
Mouse IGF-1 ELISA Kiit	Crystal Chem	Cat #80574
MILLIPLEX MAP Mouse Pituitary Magnetic Bead Panel - Endocrine Multiplex Assay	Millipore Sigma	Cat# MPTMAG-49K-06
Deposited data
Raw and analyzed bulk RNA seq data	This paper	GEO: GSE234483
Raw and analyzed metabolomics data	This paper	MassIVE MSV000098844
Experimental models: Organisms/strains
Mouse: C57BL/6J Sex and age: Males, 3 months (individual compound experiments) Sex and age: Males, 3 months (individual compound experiments) Sex and age: Males, 4 months (hypothalamic young sequencing control) Sex and age: Males, 4 months (ghrelin response experiments) Sex and age: Males, 6 months (leptin response experiments) Sex and age: Males, 6 months (AICAR response experiments) All sample sizes are described in respective figure legends. Gender disparities within given experiments have been described in the “Limitations of the Study”, and authors acknowledge the limited female cohort in our studies.	Jackson Laboratories	Cat # 000664
Mouse: Leptin receptor deficient homozygous for Leprdb, wild type for Dock7m Sex and age: Males, 2 months (Gly-Low dose response experiments) All sample sizes are described in respective figure legends. Gender disparities within given experiments have been described in the “Limitations of the Study”, and authors acknowledge the limited female cohort in our studies.	Jackson Laboratories	Cat # 000642
Mouse: Aged C57BL/6J Sex and age: Males, 24 monhts (lifespan experiments) All sample sizes are described in respective figure legends. Gender disparities within given experiments have been described in the “Limitations of the Study”, and authors acknowledge the limited female cohort in our studies.	National Institutes of Aging, Bethesda, MD	N/A
Software and algorithms
ImageJ	Schneider et al.74	N/A
GraphPad Prism	GraphPad	Prism v. 10.3.1 (464)
RStudio	RStudio Team (2020). RStudio: Integrated Development for R. RStudio, PBC, Boston, MA	N/A
Shimadzu LC (MGO Quantification)	Shimadzu Corporation	N/A
AB SCIEX 4500 QTRAP (MGO Quantification)	Sciex	N/A
Agilent 1260 (LC) (Metabolomics) Agilent 1260 SL LC-6520 Quadrupole-Time of Flight (Q-TOF) (MS) (Metabolomics)	Agilent	N/A
Progenesis Qi Software v. 2.2.5826.42898 (Metabolomics)	Nonlinear Dynamics, Newcastle, UK	N/A
MONA database (Metabolomics)	MassBank of North America, Davis, California	N/A
Other
Experimental Diet: Gly-Low Standard Chow (21% fat (kcal), 60% carbohydrate (kcal))	Envigo Teklad	TD.200742
Experimental Diet: Standard Chow (21% fat (kcal), 60% carbohydrate (kcal))	Envigo Teklad	TD.200743
Experimental Diet: Gly-Low High Fat Chow (60% fat (kcal), 21% carbohydrate (kcal))	Envigo Teklad	TD.200300
Experimental Diet: Standard High Fat Chow (60% fat (kcal), 21% carbohydrate (kcal))	Envigo Teklad	TD.200299

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

Studies included the use of male and female C57BL/6J (Jackson Laboratories #000664), and male leptin receptor deficient mice (Jackson Laboratories #000642, homozygous for Lepr^db, wild-type for Dock7^m) (National Institutes of Aging, Bethesda, MD). All mice were communally housed and age-matched with ad libitum access to water and diet in a pathogen and temperature-controlled room with a 12h light-dark cycle beginning at 06:00 a.m. All procedures were conducted per NIH Guidelines for Care and Use of Animals and were approved by the Institutional Animal Care and Use Committees at Buck Institute for Research on Aging and the University of California San Francisco.

Mice were fed either a standard low-fat chow diet (21% fat (kcal), 60% carbohydrate (kcal) Envigo: TD.200743), a standard high fat chow diet (60% fat (kcal), 21% carbohydrate (kcal), Envigo: TD.200299), a standard low-fat chow diet supplemented with our Gly-Low compound cocktail (21% fat (kcal), 60% carbohydrate (kcal) Envigo: TD.200742), or a standard high fat chow diet supplemented with our Gly-Low compound cocktail (60% fat (kcal), 21% carbohydrate (kcal).

A combination of supplemental grade compounds, safe to be consumed in set dosages, were prepared and incorporated into a modified pre-irradiated standard AIN-93G mouse chow diet from Envigo. The cocktail consists of alpha lipoic acid (20.19%), nicotinamide (57.68%), thiamine hydrochloride (4.04%), piperine (1.73%), and pyridoxamine dihydrochloride (16.36%), and is supplemented in the diet to achieve a daily consumption rate in mg/kg of body weight/day. For the 1× (full-dose) diet, these percentages translate to 3 g/kg alpha lipoic acid, 8.57 g/kg nicotinamide, 0.6 g/kg thiamine hydrochloride, 0.26 g/kg piperine, and 2.43 g/kg pyridoxamine dihydrochloride. Noted vocabulary where 1× is equal to a full dose, 0.5× is equal to a half dose, and 0.25× is equal to a quarter dose of the compound cocktail.

METHOD DETAILS

Lifespan

19-month-old male C57BL/6J mice received from National Institutes of Aging (Bethesda, MD) and maintained on vivarium chow until they reached 24 months of age. At 24 months of age, mice were randomly assigned to begin either a control chow diet or Gly-Low diet. Health checks were performed 5 times a week and mice reached either a humane endpoint or died of natural causes.

Measuring food intake and energy metabolism

Food consumption and body weight were measured once weekly. Chow weight of communally caged mice was recorded once weekly and individual intake was measured as change in weight over the week divided by the number of mice per cage.

Metabolic parameters in mice were assessed using a Promethion Metabolic Caging System housed in the Buck Institute Mouse Phenotyping Core. Mice were singly housed and received water and food ad libitum. Cages were maintained at 20^◦–22^◦C under a 12h light-dark cycle beginning at 06:00 a.m., and mice were acclimated to single housing 24h before being studied. The cages continuously weighed food for each mouse, and daily intake was measured as a change in food weight over 24h periods. Metabolic data were collected by the respiration rates of each mouse and were normalized to individual mouse body mass. X, Y, and Z beam breaks quantified total activity and steps taken during the metabolic cage run.

Body composition quantification

Echo MRI and DXA scans were used to analyze body composition in anesthetized (isoflurane) immobilized mice. Water weight and bone mass were excluded from body weight to quantify lean and fast mass.

Glucose testing

Random (non-fasted) and fasted blood glucose levels were determined for each mouse after 8 weeks and 16 weeks of treatment. Non-fasted blood glucose levels were assessed between 08:00 a.m. and 10:00 a.m. by a collection of blood by tail nick and the use of a handheld glucometer (AccuCheck). Fasted blood glucose was determined after a 16h fast between 8:00 a.m. and 10:00 a.m. by a collection of blood by tail nick and the use of a handheld glucometer (AccuCheck). Handheld Accucheck glucometer had a maximum read capacity of 600 mg/dL. Therefore, any values above max read capacity were listed as 600 mg/dL.

Glucose and insulin tolerance testing

Mice underwent GTT testing at 12 months of age, following 8 months of Gly-Low treatment. Food was removed from control-fed and Gly-Low treated mice for 14 h before testing of glucose tolerance from 6:00 p.m. to 8:00 a.m. Mice received a single IP injection of D-glucose (2 g/kg), followed by a single tail nick to collect blood for blood glucose monitoring by handheld glucometer (AccuCheck). Total glucose AUC was measured by GraphPad Analysis.

Mice underwent ITT testing at 12 months of age, following 8 months of Gly-Low treatment. Food was removed from control-fed and Gly-Low treated mice 4 h before testing of insulin tolerance from 8:00 a.m. to 12:00 p.m. Mice received IP injections of insulin (0.75U/kg), followed by a single tail nick to collect blood glucose monitoring by handheld glucometer (AccuCheck). Mice that had a blood glucose reading <40 mg/dL were immediately removed from the study and injected with 100 μL of 2 g/kg D-glucose. Total glucose AUC was measured by GraphPad Analysis. Handheld Accucheck glucometer had a maximum read capacity of 600 mg/dL. Therefore, any values above max read capacity were listed as 600 mg/dL.

Leptin injections

Mice underwent leptin sensitivity testing at 4 months of age, following 1 month of Gly-Low treatment. Food was removed from control-fed and Gly-Low treated mice 16 h before testing of leptin sensitivity from 7:00 a.m. to 7:00 p.m. Mice received a single IP injection of either saline vehicle or leptin (1 g/kg). Food consumption rates were collected by metabolic cage food measurements. Mice were allowed four days of rest before undergoing another single IP injection of either saline vehicle or leptin (1 g/kg) to serve as their control. Total food consumption was measured by GraphPad Analysis.

Leptin injection and pSTAT3 signaling

Mice underwent histological leptin sensitivity testing at 4 months of age, following 1 month of Gly-Low treatment. Food was removed from control-fed and Gly-Low treated mice 16 h before testing of leptin sensitivity from 7:00 a.m. to 7:00 p.m. Mice received a single IP injection of either saline vehicle or leptin (1 g/kg). Within 30 min of receiving the injection, mice were euthanized by CO₂ asphyxiation and cervical dislocation. Dissections were performed and mouse brains were removed and washed with PBS before being postfixed in 4% PFA overnight with agitation at 4C. Afterward, a brain matrix (BrainTree) was used to isolate sections containing the hypothalamus. This section was immediately embedded in OCT, frozen on dry ice, and stored at −80C. Next, 4-μm sections were cut on a cryostat, blocked for 1 h with 5% BSA containing 0.1% Triton X-100, and incubated with pSTAT3 (1:200, Cell Signaling). Adequate secondary antibody was used for the HRP-diaminobenzidine reaction. The HRP-diaminobenzidine reaction was performed using the ABC Kit (Vector Laboratories), using biotin-labeled goat anti-rabbit IgG. Images were acquired using a Zeiss AxioImager brightfield microscope.

Ghrelin responsiveness

Ghrelin peptides (Catalog #031–30) (Phoenix Pharmaceuticals, Inc.). Mice were injected with reconstituted ghrelin (0.1 mg/kg) by subcutaneous injection. Following injection, mice were singly housed, and individual food consumption was recorded over 90 min post-injection.

AICAR experiments

6-month-old wild-type C5BLl6/J male mice were adapted to a control (n = 12) or 1× Gly-Low (n = 12) diest for 3 days. With the exception of unfasted (FED) control groups, all mice were fasted for 18 h prior to IP-injections with saline or of 250 mg/kg AICAR dissolved in saline then re-fed their respective diets. Fasted controls were dissected immediately following the 18-h fast as were unfasted controls and hypothalamai flash frozen on liquid nitrogen. Saline- and AICAR-injected groups were dissected 1 h post-injections, and hypothalamai flash frozen on liquid nitrogen. All tissues were stored at −80^◦C until processed for Western blot analysis.

6-month-old wild-type C5BLl6/J male mice were pair-housed and adapted to a 1× Gly-Low diet for 5 days. Mice were fasted for 18 h then injected with saline (n = 12) or 250 mg/kg AICAR (n = 12) and re-fed Gly-Low diet. Food intake was measured at the 4 h and 6 h timepoint post-injections. All mice were then allowed to adapt to a control diet for 1 week. Mice were fasted for 18 h then injected with saline (n = 12) or 250 mg/kg AICAR (n = 12) and re-fed control diet. Food intake was measured at the 4 h and 6 h timepoint post-injections.

Hormone quantification

Blood samples were collected by cardiac puncture when mice were euthanized for dissection. Blood samples were collected in heparin lined tubes and left on ice for 30 min. Afterward, samples were centrifuged for 15 min at 2200 g to isolate plasma. Ghrelin, LEAP2, and IGF-1 were measured in plasma by ELISA kits according to manufacturer’s instructions.

LEAP2 ELISA (Catalog #075–40) (Phoenix Pharmaceuticals, Inc.)

Ghrelin ELISA (EZRGRA-90K) (Sigma Aldrich)

IGF-1 ELISA (Catalog #80574) (Crystal Chem)

Hypothalamic RNA sequencing

Hypothalamic transcripts were analyzed from male C57BL/6J mice at three ages under different treatment paradigms: 1) young (4-month-old) male C57BL/6J mice fed a control or glycation-lowering (Gly-Low) diet for 1 week, 2) aged (25-month-old) male C57BL/6J mice fed a control or glycation-lowering (Gly-Low) diet for 5 months and 3) young (3-month-old) male C57BL/6J mice fed a control diet for 1 week. Aged 19 month-old male C57BL/6J mice were ordered from the National Institutes of Aging (Bethesda, MD). Young mice were acquired from Jackson laboratories (#000664). Mice were fed vivarium chow (Envigo Teklad 2018) before starting either a control diet or Gly-Low diet. Mice were sacrificed via CO₂ asphyxiation followed by cervical dislocation. The brain was rapidly dissected and the hypothalamus was removed with tweezers, flash-frozen, and stored at —80^◦C. RNA was isolated using Zymo research quick RNA miniprep kit (cat # 11–328) according to the manufacturer’s recommendations. Isolated RNA was sent for library preparation and sequencing by Novogene Corporation Inc. where RNA was poly-A selected using poly-T oligo-attached magnetic beads, fragmented, reverse transcribed using random hexamer primers followed by second strand cDNA synthesis using dTTP for non-directional library preparation. Samples underwent end repair, A-tailing, adapter ligated, size selected, amplified, and purified. Illumina libraries were quantified using Qubit and qPCR and analyzed for size distribution using a bioanalyzer. Libraries were pooled and sequenced on an Illumina Novoseq 6000 to acquire paired-end 150 bp reads. Data quality was assessed and adaptor reads and low quality reads were removed. Reads that passed the quality filtering process were mapped paired-end to the reference genome (GRCm38) using Hisat2 v2.0.5. featureCounts v1.5.0-p3 was used to count reads that mapped to each gene. Differential expression analysis was performed using DESEq2 (1.20.0). Where indicated, bootstrapping was performed using R (R version 4.1.2) program ‘boot’ (1.3–28.1). To determine the expected mean and standard deviation, n = i log2 fold changes were randomly selected 1000 times, in which i is the number of genes in the gene set.

Metabolomics

Samples were prepared and analyzed by Northwest Metabolomics Research Center according to the following:

LC-MS Conditions: Acquisition Mode MS1; threshold count: 100; m/z 60–1000; Gas temp 325^◦C; Drying Gas 10L/min; Nebulizer 35 psi; TOF Fragmentator 120 V; Skimmer 65 V; 4 spectra/s; 250 ms/spectrum; Acquisition Mode MS2: range m/z 20–1000; 4 spectra/s; 250 ms/spectrum; CE 20 eV; 5 precursor per cycle, MS2 threshold 200 counts; active exclusion after 3 spectra; static exclusion rang m/z 60–100; Abundance dependent accumulation target 50000 counts/spectrum; exclusion list from blank injection enabled; reference mass correction enabled.

LC System: Agilent 1260 as the Mobile Phase (pump 2) and reference solution (pump 1)

MS System: Agilent 1200 SL LC-6520 Quadrupole-Time of Flight (Q-TOF) MS.

LC column: WATERS XBridge BEH Amide (15 cm × 2.1 mm; 2.5 μm)

Buffer A: 10 mM ammonium acetate and 0.2% acetic acid in 95% H2O + 3% ACN +2% MeOH;

Buffer B: 10 mM ammonium acetate and 0.2% acetic acid in 5% H2O + 93% ACN +2% MeOH;

Injection: 5 μL (+)-ESI and 10 μL (—)-ESI; Wash: 95%ACN+5%H2O for 10 s; flow rate (mL/min): 0.3.

LC Column Chamber Temp: 40^◦C; ESI mode: (±)

Worklist: each sample was injected in both positive and negative ESI mode (labeled _POS and _NEG). Blank of sample preparation labeled as blank_Prep. Plasma and tissue samples were combined to make QC samples from plasma (QCp) and tissue (QCt).

Gradient operation (Separation)
min	B%
0	95
3	95
8	50
12	50
13	95
35	95

Plasma sample preparation

Samples were thawed at 4^◦C, vortexed 10 s 28 μL plasma plus 22 μL water was transferred to a 2 mL Eppendorf vial. 250 μL Methanol was added then the sample was vortexed 10 s. Samples were incubated at −20^◦C for 20 min and centrifuged at 14000 rpm at 4^◦C for 15 min 150 μL supernatant was transferred to a new 2 mL Eppendorf vial and dried completely using a Vacufuget at 30^◦C for about 1.5 h. Samples were reconstituted with 250 μL HILIC solvent, vortexed 10 s then centrifuged for 5 min at 14000 rpm at 4^◦C. 250 μL supernatant was transferred into LC vials for MS analysis.

Data processing

Data processing was performed using Progenesis Qi software v. 2.2.5826.42898 (Nonlinear Dynamics; Newcastle; UK). Peak alignment was carried out taking samples as reference. Peak-peaking was performed using sensitivity and chromatographic peak width at 1 (3 for (+)-ESI) and 0.01min; respectively. The retention time limit was set 1.0–12.0 min. Possible adduct ions were defined as follows: [M + H]+; [M+Na]+; [M + NH4]+; [M + K]+; M+⋅; [M+ACN+H]+; in (+) ESI mode; and [M–H]–; [M + HCOO–H]–; [M + Cl]–; and [2M + CH3COO–H]– in (—) ESI mode. The adduct ions were grouped into mass features through peak deconvolution. Putative peak annotation was performed by searching metabolites from MONA Database using accurate m/z measurements from the full scan data and MS2 spectra. We set the m/z tolerance of 20 ppm (MS1) and 30 ppm (MS2).

Liver histology

Dissections were performed and mouse livers were removed and washed with PBS before being postfixed in 4% PFA overnight with agitation at 4C. Afterward, livers were moved to 70% EtOH before being paraffin embedded and sectioned. The liver was sectioned by microtome in a coronal orientation at a thickness of 4 μm. H&E staining and trichrome staining were used for the identification and quantification of large lipid vacuoles.

Western blotting

Proteins were extracted from hypothalamic samples in TPER buffer (Thermo Fisher) containing a protease inhibitor cocktail (Sigma). Protein extracts were denatured at 70^◦C for 15 min prior to running on a 5–12% Bis-Tris gel. The transfer was completed using iBlot (Thermo Fisher) and blocked in 5% BSA for 1 h. Primary antibodies were incubated overnight at 4^◦C with agitation. Adequate HRP-conjugated antibodies were incubated at room temperature for 1 h prior to imaging.

Akt (Cell Signaling) (40D4) 1:1000

pAKT Ser473 (Cell Signaling) (D9E) 1:1000.

S6: (Cell Signaling) (5G10) (2217) 1:1000

pS6 Ser 240/244: (Cell Signaling) (D68F8) 1:1000.

AMPK: (Cell Signaling) (CST-4181) 1:1000

pAMPK: (Cell Signaling) (CST-2531)

mTOR: (Cell Signaling) (7C10): 1:1000.

Actin (Cell Signaling) (13E5) 1:1000.

Mass spectrometry MGO quantification

Quantification of dicarbonyls

200 μL of 80:20 MeOH:ddH2O (−80^◦C) containing 50 pmol ¹³C3-MGO was added to 10 μL of serum and extracted at −80^◦C overnight. Insoluble protein was removed via centrifugation at 14,000 × g for 10 min at 4^◦C. Supernatants were derivatized with 10 μL of 10 mM o-phenylenediamine for 2 h with end-over-end rotation protected from light.⁷⁵ Derivatized samples were centrifuged at 14,000 × g for 10 min, and the supernatant was chromatographed using a Shimadzu LC system equipped with a 150 × 2mm, 3μm particle diameter Luna C18 column (Phenomenex, Torrance, CA) at a flow rate of 0.450 mL/min. Buffer A (0.1% formic acid in H2O) was held at 90% for 0.25 min, then a linear gradient to 98% solvent B (0.1% formic acid in acetonitrile) was applied over 4 min. The column was held at 98% B for 1.5 min, washed at 90% A for 0.5 min, and equilibrated to 99% A for 2 min. Multiple reaction monitoring (MRM) was conducted in positive ion mode using an AB SCIEX 4500 QTRAP with the following transitions: m/z 145.1→77.1 (MGO); m/z 235.0→157.0 (3-DG); m/z 131.0→77.0 (GO); m/z 161.0→77.0 (HPA); m/z 148.1→77.1 (¹³C3-MGO, internal standard).^76,77

Quantitation of PTMs (QuarkMod)

Protein pellets from the dicarbonyl quantifications (above) were resuspended in 65 μL of 50 mM NH4HCO3, pH 8.0. Samples were spiked with 10 μL of a master mix containing internal standards (see table). Proteins were digested by adding 5 μL of sequencing grade trypsin (0.1 mg/mL) (Promega) for three h at 37^◦C. Trypsin was denatured by boiling at 95^◦C for 10 min, and the samples were cooled to room temperature. Aminopeptidase M (Millipore, 15 μg in 10 μL) was added, and samples were incubated overnight at 37^◦C. Aminopeptidase was denatured via heating at 95^◦C for 10 min, and samples were again cooled to room temperature. 15 μL of heptafluorobutyric acid (1:1 in H2O) was added to each sample, and debris was removed via centrifugation at 14,000 × g for 10 min. Clarified supernatants were chromatographed using a Shimadzu LC system equipped with a 150 × 2.1mm, 3.5 mm particle diameter Eclipse XDB-C8 column (Agilent, Santa Clara, CA) at a flow rate of 0.4500 mL/min. Mobile phase A: 10 mM HFBA in water; mobile phase B: 10 mM HFBA in ACN. The following gradient was used: 2 min, 1% B; 6 min, 50% B; 6.5 min, 95% B; 9 min, 95% B; 9.5 min, 1% B. The column was equilibrated for 3 min at 5% B. MRM was conducted in positive mode using an AB SCIEX 4500 QTRAP. The MRM detection window was 50 s with a target scan time of 0.75 s. The following parameters were used for detection and as previously described:

Species	Q1 (m/z)	Q3 (m/z)	CE (V)
Lys	147.1	84.1	29
13C615N2 Lys	155.1	90.1	29
Arg	175.1	70.1	47
13C615N4 Arg	185.1	75.1	47
Leu	132.1	86.1	17
13C615N Leu	139.1	93.1	17
MG-H1	229.2	70.1	53
13C-MG-H1	230.2	70.1	53
CEA	247.2	70.1	55
13C-CEA	248.2	70.1	55
CEL	219.2	84.1	41
CEL-d4	223.2	88.1	41
CML	205.0	84.1	38
CML-d4	209.0	88.1	38

Quantification of free MGH1

10 μL of serum was added to 200 μL of 80:20 MeOH:ddH2O (−80^◦C) containing ten pmol ¹³C-MG-H1 and extracted at −80^◦C overnight. Insoluble protein was removed via centrifugation at 14,000 × g for 10 min at 4^◦C, and supernatants were transferred to a new tube. 15 μL of heptafluorobutyric acid (1:1 in H₂O) was added to each sample, and debris was removed via centrifugation at 14,000 × g for 10 min. Samples were analyzed as described above (QuARKMod).

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical details of experiments can be found in the figure legends. As the figure legends indicate, all data are expressed as mean ± SEM. Statistical tests were selected based on appropriate assumptions with respect to data distribution and variance characteristics. For normally distributed data, statistical significance was determined using an unpaired T test. Statistical significance was determined using a one-sample T test for data normalized to the vehicle control group. All statistical analyses were performed using GraphPad Prism. Significant differences are indicated: * p ≤ 0.05, ** p ≤ 0.005, *** p ≤ 0.0005, **** p < 0.0001.

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116422.

Highlights.

• Gly-Low reduces food intake and weight gain while preserving muscle mass

• Gly-Low works independently of traditional caloric restriction to extend lifespan in mice

• Gly-Low treatment improves insulin sensitivity and glucose homeostasis in mice

• Gly-Low acts through hypothalamic signaling to elicit appetite and weight control