Systems Genomics Reveals Age- and Sex-Dependent Metabolic Dysregulation from Glo1 Reduction in Mice

Abstract

Objectives:

Glyoxalase 1 (Glo1) detoxifies reactive dicarbonyl compounds such as methylglyoxal, a precursor of advanced glycation end products (AGEs), which contribute to metabolic disorders. However, the contribution of AGE-independent mechanisms to Glo1-related metabolic dysfunction remains unclear.

Methods:

We conducted a longitudinal study in male and female Glo1 heterozygous knockdown (Glo1^+/−) mice (~50% Glo1 expression). Metabolic phenotypes, including body weight, adiposity, glycemic control, and plasma lipid levels, were assessed over time. Atherosclerotic burden, AGE levels, and gene expression profiles in liver, adipose, muscle, kidney, and aorta were examined to identify pathway alterations and regulatory genes affected by Glo1 reduction.

Results:

Partial Glo1 loss resulted in obesity, hyperglycemia, dyslipidemia, and altered lipid metabolism in an age- and sex-dependent manner, with most phenotypes emerging after ~14 weeks. Glo1^+/− females exhibited impaired glycemic control and elevated triglycerides, along with perturbations in adipogenesis, PPARγ signaling, insulin signaling, and fatty acid metabolism in liver and adipose tissue. Glo1^+/− males displayed increased skeletal muscle mass and visceral adiposity with changes in lipid metabolic pathways. Methylglyoxal-derived AGE accumulation was altered only in male skeletal muscle and did not explain broader phenotypes. Transcriptomic analyses suggest altered glucose and lipid metabolism may be partially driven by alternative detoxification of methylglyoxal to metabolites such as pyruvate. Transcription factor analysis identified Hnf4a (across tissues) and Arntl (in aorta, liver, and kidney) as female-biased regulators altered by Glo1 deficiency.

Conclusions:

Glo1 reduction disrupts metabolic health through sex- and age-dependent pathways largely independent of AGE accumulation, involving tissue-specific metabolic reprogramming and transcriptional regulation.

NEW & NOTEWORTHY

This study reveals that partial deficiency of glyoxalase 1 (Glo1) leads to age- and sex-specific metabolic dysfunction in mice through transcriptional and regulatory changes independent of advanced glycation end products (AGEs). Transcriptomic profiling and integrative genomics identified female-biased transcription factors and GWAS-linked metabolic genes as key mediators. These findings uncover novel, AGE-independent regulatory pathways linking Glo1 to metabolic disease risk and emphasize the importance of sex-specific analysis in metabolic genomics research.

Graphical Abstract

1. INTRODUCTION

The global prevalence of obesity and related metabolic disorders, including cardiovascular disease (CVD) and type 2 diabetes (T2D), continues to rise despite existing preventive and therapeutic strategies. Improved mapping of disease mechanisms is therefore essential for identifying novel biological pathways and therapeutic targets. Glyoxalase 1 (Glo1) has been linked to obesity [1, 2], glycemic control [3], insulin sensitivity [4], aortic endothelial dysfunction [5] ,non-alcoholic fatty liver disease (NAFLD) [6], skeletal muscle dysfunction [7], and CVD [8], highlighting its critical role in maintaining metabolic health [9, 10]. Although advanced glycation end products (AGEs) have traditionally been considered the main mediators of Glo1-related dysfunction, AGE-independent mechanisms remain underexplored. Glo1 is a key component of the glyoxalase system, which consists of the ubiquitous enzymes Glo1 and Glyoxalase 2 (Glo2) and a catalytic amount of reduced glutathione (GSH) [11]. This system serves as a defense mechanism against the accumulation of reactive dicarbonyls such as methylglyoxal (MG), a cytotoxic byproduct of glycolysis [12]. MG and GSH spontaneously form a hemithioacetal, the substrate for Glo1. Glo1 then catalyzes its conversion to S-D-lactoylglutathione, which is subsequently hydrolyzed by Glo2 to produce D-lactate and regenerate GSH. Thus, Glo1 initiates the detoxification of MG into a non-toxic and stable product.

Impairment of the glyoxalase system leads to the accumulation of MG, resulting in dicarbonyl stress. MG readily reacts with DNA, lipids, and protein residues particularly arginine and lysine to irreversibly generate AGEs, making MG a major precursor of these compounds. Among MG-derived AGEs, N-(5-hydro-5-methyl-4-imidazolon-2-yl)-ornithine (MGH1) and carboxyethyl-lysine (CEL) are the most abundant. The interaction between AGEs and their receptor (RAGE) plays a key role in the pathogenesis of various diseases, including cancer, Alzheimer’s disease, NAFLD, insulin resistance, and diabetes, as well as its micro- and macrovascular complications such as retinopathy, nephropathy, and neuropathy [4, 13–16]. When the glyoxalase pathway is compromised, alternative detoxification systems can partially compensate for MG clearance, although not as effectively. These enzymes include aldehyde dehydrogenases (ALDHs) and aldo-keto reductases (AKRs), which metabolize MG to pyruvate and hydroxyacetone, respectively [17].

Despite the established role of the MG-AGE pathway in obesity and metabolic dysfunction, previous studies in fish and mice have shown that complete Glo1 knockout does not always lead to elevated MG, AGEs, or observable phenotypes, likely due to compensatory activity of ALDH and AKR enzymes [18]. In contrast, Glo1 knockouts in Drosophila and C. elegans display elevated MG and AGEs [19, 20], highlighting species-specific differences in compensatory mechanisms. Although Glo1 knockout models implicate the MG-AGE axis in metabolic disease, complete loss of Glo1 activity is rare in humans, and the underlying physiological mechanisms of Glo1 function remain incompletely understood [21]. To model a more physiologically relevant state, we investigated the effects of partial Glo1 reduction using Glo1^+/− mice on a C57BL/6 background, which exhibit a 45–65% decrease in enzymatic activity. We examined the mechanisms by which reduced Glo1 expression influences obesity and metabolic dysfunction, and evaluated the impact of sex and age over a six-month period.

Our results revealed that female Glo1^+/− mice displayed more severe metabolic phenotypes including elevated glucose levels and impaired lipid metabolism in comparison to males. Interestingly, AGEs were increased only in male Glo1^+/− skeletal muscle compared to controls, suggesting that AGE-independent mechanisms may underlie the female-biased phenotypes. Gene expression, metabolic pathway, and network analyses identified female-biased transcription factors such as Hnf4a and metabolic intermediates such as pyruvate, generated from MG detoxification, as contributors to metabolic reprogramming in females. Furthermore, integration of tissue-specific gene expression signatures with human genome-wide association study (GWAS) data provided additional mechanistic and clinical insights linking Glo1 to multiple metabolic traits and diseases. Collectively, these findings highlight a sex- and age-dependent role for Glo1 in regulating metabolism through transcriptional and metabolic networks that operate largely independently of AGEs.

2. METHODS

2.1. Heterozygous Glo1 Knockdown Mice

The Glo1^+/− mice were originally generated in Dr. Michael Brownlee’s laboratory by injecting a Glo1-targeting shRNA lentivirus into C57BL/6 mouse embryos [22, 23]. Mice carrying a single copy of the viral insert were identified by Southern blotting and used to establish founder lines. Glo1 mRNA and protein expression were assessed by quantitative PCR and Western blot analysis, and enzymatic activity was further confirmed by functional assays. Heterozygous offspring from the founder line exhibited a 45–65% reduction in tissue Glo1 activity and were used in all experiments [22]. Glo1^+/− embryos were kindly provided by Dr. Abraham Palmer’s laboratory and re-derived at the University of California, Los Angeles (UCLA) by the Division of Laboratory Animal Medicine (DLAM) through implantation into surrogate C57BL/6 females. In our laboratory, male Glo1^+/− mice were mated with wildtype females to expand the colony. Genotypes were confirmed by PCR as described below. All animal experiments involving Glo1^+/− mice and their littermate controls were conducted in accordance with the United States National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the UCLA Animal Research Committee.

2.2. Genotyping and Husbandry

Genomic DNA was extracted from mouse ear punches (n = 10–13 per sex) and analyzed by PCR using the KAPA Mouse Genotyping Kit (Kapa Biosystems, Wilmington, MA, USA). Glo1 knockdown primers were as follows: forward 5′-GCTTCTCCCACAAGTCTGTG-3′ and reverse 5′-GGTACAGTGCAGGGGAAAGA-3′. Gapdh primers served as the internal control: forward 5′-AACTTTGGCATTGTGGAAGG-3′ and reverse 5′-ACACATTGGGGGTAGGAACA-3′. Mice were provided ad libitum access to a standard chow diet (Newco Distributors Inc., Rancho Cucamonga, CA, USA) and water. Food and water intake were recorded weekly to monitor caloric intake across experimental groups.

2.3. Glo1 Enzyme Activity Assay

Protein extracts were prepared from 50 mg of liver and kidney tissue to determine Glo1 enzymatic activity according to the manufacturer’s instructions (Sigma-Aldrich, St. Louis, MO, USA). Briefly, tissue lysates were incubated with the substrate and co-substrate for 20 minutes, precipitated with perchloric acid on ice, vortexed, and centrifuged at 14,000 rpm for 5 minutes. Supernatants were transferred to a 96-well UV plate in duplicate, and absorbance was measured at 240 nm using a microplate reader (BioTek, USA). A blank control was processed in parallel without protein. Glo1 activity was quantified based on the production of S-lactoylglutathione, where one unit of enzyme activity corresponds to the amount of enzyme that converts 1.0 μmol of S-lactoylglutathione from methylglyoxal and reduced glutathione per minute at pH 6.6 and 25 °C. Final enzyme activity values (U/L) were calculated using the formula provided by the manufacturer.

2.4. Body Weight and Composition

For both Glo1^+/− and littermate control mice (n = 10–13 per genotype per sex), body weight was recorded weekly beginning at 3 weeks of age and continuing until 28 weeks. Starting at 5 weeks of age, fat and lean mass were measured biweekly until 27 weeks using nuclear magnetic resonance (NMR; Bruker, Madison, WI, USA). Body weight was obtained using a calibrated scale. Statistical analysis was performed using two-way ANOVA with repeated measures, followed by Bonferroni post hoc correction.

2.5. Intraperitoneal Glucose Tolerance Test (IPGTT)

Glucose tolerance tests were conducted at 5, 12, 23, and 33 weeks of age. Mice were fasted overnight for 14 hours before intraperitoneal (IP) injection of a 20% glucose solution (Sigma-Aldrich, St. Louis, MO, USA) at a dose of 2 g glucose per kg body weight. Blood samples (<5 μL) were collected from a tail incision at 0, 15-, 30-, 60-, and 120-minutes post-injection for glucose measurements using a handheld glucometer.

2.6. Plasma Lipid, Glucose, and Insulin Quantification

Blood plasma was collected at 7, 12, and 28 weeks of age for lipid, glucose, and insulin quantification. Mice were fasted overnight for 14 hours and anesthetized with isoflurane prior to retro-orbital blood collection using a microcapillary tube. Blood was collected into K₂-EDTA tubes, placed on ice, and centrifuged at 1,500 × g for 10 minutes. Plasma was separated and stored at −80 °C until analysis. Plasma triglycerides (TG), total cholesterol (TC), unesterified cholesterol (UC), free fatty acids (FFA), high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, glucose, and insulin were measured using enzymatic colorimetric assays as previously described [24]. Very-low-density lipoprotein (VLDL) cholesterol was calculated using the equation VLDL = TG / 5.

2.7. Characterization of Atherosclerotic Lesions

Mouse hearts were embedded in optimal cutting temperature (OCT) compound (VWR, Radnor, PA, USA) and stored at −80 °C for histological analysis. Using a stereomicroscope, hearts were sectioned longitudinally at 10 μm thickness and fixed in 80% 2-propanol. The aortic sinus was stained with Oil Red O (Abcam, Cambridge, MA, USA) for lesion quantification. Sections were fixed in 10% formalin for 10 minutes, rinsed three times in 1× PBS, incubated in Oil Red O solution for 10 minutes, rinsed with tap water, and counterstained with hematoxylin for 1 minute. Lesions were visualized using an inverted phase-contrast microscope (Eclipse TE300; Nikon Co., Tokyo, Japan).

2.8. Total RNA Extraction, cDNA Synthesis, and qPCR

Liver, gonadal adipose, and skeletal muscle tissues from both male and female mice of 28 weeks of age were flash-frozen in liquid nitrogen immediately after euthanasia. Total RNA was extracted using the RNeasy Mini Kit (Qiagen, Germantown, MD, USA) following the manufacturer’s protocol. RNA concentration and purity were determined using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Complementary DNA (cDNA) was synthesized using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA, USA). Gene expression of Ager, Akr1a1, Aldh1a1, Lipin1, Acc1, Fasn, Elovl6, Scd1, Srebp1c, Dgat1, and Dgat2 was quantified by qPCR using an Applied Biosystems Real-Time PCR instrument with SYBR Green Master Mix (Thermo Fisher Scientific, Waltham, MA, USA). Primer sequences are listed in Supplementary Table 1.

2.9. Protein Extraction from Metabolic Tissues

Protein was extracted from frozen liver, gonadal adipose, and skeletal muscle for ELISA and Western blot analyses according to the Abcam ELISA tissue preparation guide. Approximately 50 mg of frozen liver or skeletal muscle tissue and 100 mg of frozen gonadal adipose tissue were homogenized in extraction buffer containing 1% Triton X-100 (Thermo Fisher Scientific, Hudson, NH, USA), 100 mM Tris (pH 7.4), 150 mM NaCl, 1 mM EGTA, 1 mM EDTA, 0.5% sodium deoxycholate, and a protease/phosphatase inhibitor cocktail (Abcam, Cambridge, MA, USA). Homogenates were centrifuged at 13,000 rpm for 20 minutes at 4 °C, and the supernatant containing soluble protein was collected. Total protein concentration was determined using the Bicinchoninic Acid (BCA) Protein Assay (Pierce, Rockford, IL, USA). Lysates were diluted to 2000 μg/mL for liver and skeletal muscle and to 500 μg/mL for gonadal adipose tissue for ELISA. For Western blotting, protein extraction from liver and kidney was performed as described above, and lysates were diluted to 80 μg total protein per sample.

2.10. ELISA Quantification of AGEs

The two major methylglyoxal-derived AGEs, N-(5-hydro-5-methyl-4-imidazolon-2-yl)-ornithine (MGH1) and carboxyethyl-lysine (CEL), were quantified using protein extracts from liver, gonadal adipose, and skeletal muscle. MGH1 levels were measured using the OxiSelect^™ Methylglyoxal (MG) Competitive ELISA Kit (Cell Biolabs Inc., San Diego, CA, USA), and CEL levels were determined using the OxiSelect^™ Nε-(Carboxyethyl) Lysine (CEL) Competitive ELISA Kit (Cell Biolabs, San Diego, CA, USA). Absorbance was measured using a microplate reader (Bio-Rad, Hercules, CA, USA), and data were analyzed with Gen5 software.

2.11. Western Blotting for Glo1 and RAGE Quantification

Liver and kidney tissue lysates (80 μg total protein per sample) were loaded onto SDS–PAGE gels following the manufacturer’s instructions (General Protocol for Western Blotting; Bio-Rad). After electrophoresis and transfer, membranes were incubated overnight at 4 °C with primary antibodies. The following day, HRP-conjugated secondary antibodies were applied for 1 hour at room temperature. Protein signals were detected using the SuperSignal^™ West Pico PLUS Chemiluminescent Substrate (Thermo Scientific, REF 34580) and visualized on a Bio-Rad imaging system using Image Lab^™ 3.0.1 software.

2.12. Total RNA Isolation, Microarray Profiling, and Identification of Differentially Expressed Genes (DEGs)

Liver, gonadal adipose, kidney, and aorta tissues from 34-week-old female mice were flash-frozen in liquid nitrogen. Approximately 10–15 mg of liver, kidney, and aorta tissue, and 30 mg of gonadal adipose tissue, were homogenized and processed for RNA isolation using the RNeasy Mini Kit (QIAGEN GmbH, Hilden, Germany). RNA concentration and integrity were determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific, MA, USA) and a Bioanalyzer (Agilent Technologies, CA, USA). A total of 23 samples passing quality control (RIN > 7.0) were sent to the UCLA Neuroscience Genomics Core for labeling and hybridization using the Illumina MouseRef-8 v2.0 array. The dataset included 11 samples from C57BL/6 WT control mice (3 per tissue, except aorta = 2 samples) and 12 samples from Glo1^+/− mice (3 per tissue). Gene expression data were deposited in the Gene Expression Omnibus (GEO) under accession number GSE118034. Data processing and normalization were performed using the lumi Bioconductor package in R [24]. Variance-stabilizing transformation was applied, followed by robust spline normalization. Differentially expressed genes (DEGs) between groups were identified using a linear model. False discovery rates (FDRs) were calculated using the q-value Bioconductor package [25]. FDR (q-value) < 0.1 were used as cutoffs to determine significant DEGS for downstream pathway, and suggestive DEGs with p < 0.05 were used in integrative genomics analyses.

2.13. Functional Annotation of Glo1^+/− DEGs

DEGs identified from the Glo1^+/− microarray analysis were annotated for their potential biological functions using canonical pathways and functional categories from multiple databases, including the Kyoto Encyclopedia of Genes and Genomes (KEGG) [26], BioCarta (http://www.biocarta.com/genes/index.asp), Reactome [27], and gene sets derived from Gene Ontology (GO) [28]. Fisher’s exact test was used to calculate enrichment p-values for each pathway or functional category among up- and down-regulated DEGs, followed by multiple testing corrections using Storey’s method [23]. Additionally, gene set enrichment analysis (GSEA) was performed to identify tissue-specific biological pathways and to assess the distribution of up- and down-regulated genes within those pathways.

2.14. Identification of Enriched Transcription Factor Downstream Targets Among Glo1^+/− DEGs

To identify potential upstream regulators of Glo1^+/− DEGs, we evaluated whether transcription factor (TF) downstream target genes were enriched among the DEGs using the Binding Analysis for Regulation of Transcription (BART) computational tool. TFs with Irwin–Hall p-values < 0.01 were considered significantly enriched.

2.15. Overlap of Glo1 DEGs with Human Sex- and Age-Biased Transcriptomic Programs

To examine whether Glo1-regulated transcriptomic changes overlap with sex-biased or age-related gene expression programs in humans, we curated tissue-specific sex-biased and age-associated DEGs Gene by Tissue Expression (GTEx) study and canonical human aging genes from GenAge [29, 30]. Because the reference sets are human, mouse Glo1 DEGs were mapped to human orthologs (Ensembl/MGI) prior to overlap testing. For each comparison, the background included only genes expressed in both our dataset and the reference dataset for the matched tissue. Enrichment of overlap between Glo1 DEGs and each human gene set was evaluated with Fisher’s exact test. Multiple tests across tissues and reference sets were controlled by Benjamini–Hochberg FDR. Enriched/overlapping genes were functionally annotated using Gene Ontology/Reactome and GeneCards/NCBI Gene.

2.16. Assessing Enrichment of Glo1^+/− DEGs for Genes Mapped to Human Metabolic Trait Loci

To test whether genes identified in our study are linked to human metabolic traits, we curated 15 publicly available genome-wide association study (GWAS) summary datasets, containing disease association p-values for all tested genetic variants (single nucleotide polymorphisms, SNPs) from large-scale GWAS of metabolic traits. These traits included obesity (adult BMI [31], hip circumference [32], waist circumference [32] , and waist-to-hip ratio [32]), type 2 diabetes (T2D) [33], coronary artery disease (CAD) [34], glucose homeostasis traits such as HbA1c [35], fasting glucose [36], fasting insulin [36], homeostatic model assessment for β-cell function (HOMA-B) [37], insulin resistance (HOMA-IR) [37], and lipid profiles (total cholesterol, triglycerides, LDL, HDL [38]. For each GWAS dataset, SNPs with a minor allele frequency (MAF) < 0.05 were excluded. Among SNPs in linkage disequilibrium (r² > 0.5), only the variant with the strongest disease association was retained. Marker Set Enrichment Analysis (MSEA) was then used to test whether genes affected in the Glo1^+/− model were enriched for disease-associated SNPs in humans. MSEA evaluates whether a defined gene group (in this case, DEGs) shows enrichment for disease-associated SNPs beyond random expectation [39]. Reported GWAS SNPs within ±50 kb of each gene were mapped to that gene. For each DEG set, MSEA tested enrichment using a chi-like statistic. The null distribution was estimated by permuting gene labels to generate 10,000 random gene sets (matched for size to each DEG set) while preserving SNP–gene mappings. Enrichment p-values were calculated from a Gaussian distribution fitted to the permuted and observed statistics. Finally, Benjamini–Hochberg false discovery rate (FDR) values were computed across all DEG sets tested for each GWAS.

3. RESULTS

3.1. Reduction of Glo1 Gene Expression and Enzymatic Activity in Glo1^+/− Mice

To confirm genotypes and establish experimental groups, all mice were genotyped by PCR at 3 weeks of age, as previously described. PCR results showed significantly reduced Glo1 band in Glo1^+/− mice (Supplementary Figure 1A). At 28 weeks of age, Western blot analysis revealed reduced Glo1 protein expression in liver and kidney tissues of Glo1^+/− mice compared to wild-type controls. Enzymatic activity assays further showed a downward trend in Glo1 activity in Glo1^+/− mice relative to wild-type mice at 28 weeks in both liver and kidney tissues (p = 0.0675; Supplementary Figure 1B). Consistent with these findings, microarray-based expression profiling confirmed reduced Glo1 expression by 46% in adipose tissue (p = 3.2 × 10⁻³), 55% in aorta (p = 1.3 × 10⁻³), and 46% in liver (p = 1.5 × 10⁻⁴), validating partial Glo1 knockdown at 34 weeks of age (Supplementary Figure 1C).

3.2. Glo1 Reduction Affects Body Composition

As shown in Figure 1A, a schematic overview of the experimental design outlines the timeline of phenotypic data collection from week 3 to week 34. To investigate the onset and progression of phenotypic changes associated with Glo1 reduction, body weight was monitored weekly from weaning (3 weeks) through 28 weeks of age (n = 10–13 per group; Figure 1B). Fat and lean mass were assessed biweekly from 5 to 27 weeks of age using NMR analysis (Figure 1B; two-way ANOVA with repeated measures). Compared to wildtype (WT) littermates, female Glo1^+/− mice began exhibiting significantly higher body weight at 17 weeks, while male Glo1^+/− mice showed an increase beginning at 15 weeks. This trend persisted through 28 weeks of age. Both male and female Glo1^+/− mice displayed significantly elevated fat mass starting at 15 weeks, which remained higher through the end of the study (Figure 1C). Interestingly, male Glo1^+/− mice also exhibited a significant increase in muscle mass beginning at 11 weeks and continuing through 28 weeks, whereas female mice showed no change in muscle mass compared to controls (Figure 1D). Adiposity, calculated as fat mass relative to total body weight, was significantly increased in both sexes from 15 to 28 weeks (Figure 1E). Concurrent monitoring of food and water intake showed no consistent differences between groups across the 27-week period, suggesting that the observed changes in body composition were not due to altered caloric intake (Supplementary Figure 2A–D).

Figure 1. Characterization of body composition phenotypes in Glo1^+/− mice.

(A) Experimental design showing data collection time points. Body weight for (B) female (n = 12–13 per group) and male (n = 11–14 per group) mice was monitored weekly from weaning (3 weeks) to 28 weeks of age. (C) Fat mass and (D) muscle mass was measured biweekly from 5 to 27 weeks of age by NMR. (E) Adiposity (fat-to-muscle mass ratio) was assessed biweekly over the same period. At 28 weeks, (F) female and (G) male mice were euthanized for tissue collection and weighing of liver, kidney, heart, bilateral hind-limb skeletal muscles, and adipose depots (gonadal, subcutaneous, retroperitoneal, mesenteric, and perirenal). Data are presented as mean ± SEM and analyzed by two-way ANOVA with repeated measures to assess the main effects of genotype, time, and their interaction. Post-hoc Bonferroni’s multiple-comparison tests were performed to evaluate differences between genotypes at individual time points. *p < 0.05, **p < 0.01, ***p < 0.001.

At 28 weeks of age, metabolic tissue weights including liver, kidney, heart, skeletal muscle, and fat depots (gonadal, subcutaneous, retroperitoneal, mesenteric, and perirenal) were compared between groups (Figure 1F–G). Female Glo1^+/− mice exhibited significantly increased kidney and subcutaneous adipose tissue weights compared to controls (Figure 1F). In males Glo1^+/− mice, liver, kidney, heart, and multiple adipose depots (gonadal, retroperitoneal, and mesenteric) were significantly heavier compared to controls (Figure 1G). Lean mass measured by NMR was significantly increased in male Glo1^+/− mice, although the dissected skeletal muscle from the lower hindlimbs did not differ between male groups (Figure 1D, G). This discrepancy may be attributed to the fact that only the lower hindlimb skeletal muscles were dissected when skeletal muscle tissue was weighed (Figure 1G), as oppose to NMR analysis (Figure 1D), which reflects total body lean mass, including muscle from all regions (skeletal muscle, smooth muscle and cardiac muscle). Together, these findings indicate that partial loss of Glo1 leads to post-developmental obesity and altered body composition in both female and male Glo1^+/− mice.

3.3. Glo1 Reduction Induces Glucose Intolerance in Female but Not Male Mice

To assess the impact of Glo1 reduction on glucose tolerance over time, we performed IPGTT experiments on 14-hour fasted mice at 7, 12, 23, 28, and 33 weeks of age (Figure 2). Additional mice from both sexes (n = 5/group) were exempt from the 28-week endpoint to allow further assessment of glycemic phenotypes at 33 weeks, although only female cohorts showed differences in glucose tolerance over time (Figure 2A). We observed significantly increased glucose AUC in female Glo1^+/− mice at 23 weeks (n = 8/group) and 33 weeks (n = 5/group) compared to WT (Figure 2A, D–E), but no differences in male Glo1^+/− mice (Supplementary Figure 3A-E). Moreover, female Glo1^+/− mice showed no differences between groups when comparing fasting plasma glucose or insulin levels (Supplementary Figure 4A-B), whereas male Glo1^+/− mice exhibited significantly decreased plasma glucose levels at 12 weeks of age without differences in insulin levels (Supplementary Figure 4C-D). These data demonstrate that Glo1 reduction results in the development of glucose intolerance in female but not male mice.

Figure 2. Characterization of glucose metabolism phenotypes in Glo1^+/− mice.

(A) The AUC was quantified for female mice at 5, 12, 23, and 33 weeks of age. (B) Baseline IPGTT were performed on fasted female mice at 5 weeks of age (n = 13). Follow-up IPGTTs were conducted at (C) 12 weeks (n = 13), (D) 23 weeks (n = 13), and (E) 33 weeks (n = 5). IPGTT data were analyzed using two-way ANOVA with repeated measures, followed by Bonferroni’s post hoc multiple-comparison tests to determine the statistical significance of genotype, time, and their interaction. AUC data were analyzed using one-way ANOVA followed by Bonferroni’s multiple-comparison tests. *p < 0.05, **p < 0.01, ***p < 0.001. Male data are shown in the Supplementary Figures, as no significant changes were observed in male cohorts at any time point.

3.4. Glo1 Reduction Results in Altered Lipid Profiles in a Sex-Dependent Manner

To assess the impact of Glo1 reduction on lipid phenotypes over time, we measured plasma TG, TC, UC, FFA, HDL, LDL, and VLDL levels after a 14-hour overnight fast (Figure 3, Supplementary Figure 5). Hyperlipidemia, characterized by significantly elevated TG (Figure 3A) and VLDL (Figure 3D) levels, was observed in female Glo1^+/− mice at 28 weeks compared to WT counterparts. Interestingly, male Glo1^+/− mice exhibited a contrasting lipid phenotype, showing significantly decreased TG (Figure 3E), TC (Figure 3F), and HDL (Figure 3G) levels at 28 weeks, with no significant changes in other lipid classes (Supplementary Figure 5A-F). Although both female and male Glo1^+/− mice demonstrated altered circulating lipids at 28 weeks, these effects were sex-dependent. Female Glo1^+/− mice showed greater susceptibility to increases in circulating lipids compared to WT mice, suggesting potential cardiovascular risk, whereas male Glo1^+/− mice displayed decreased plasma lipids relative to controls. Collectively, these findings indicate that Glo1 reduction affects plasma lipid profiles in a sex- and age-dependent manner.

Figure 3. Plasma lipid profiling in Glo1^+/− mice.

Plasma lipid levels of (A) triglycerides (TG), (B) total cholesterol (TC), (C) high-density lipoprotein (HDL), and (D) very-low-density lipoprotein (VLDL) were quantified in fasted female mice at 7, 12, and 28 weeks of age (n = 9 per group). Corresponding plasma lipid profiles for (E) TG, (F) TC, (G) HDL, and (H) VLDL were also measured in the same female cohorts. Data are presented as mean ± SEM. Statistical significance between groups was determined using Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001.

3.5. Reduced Glo1 Levels Do Not Result in Atherosclerosis

To determine whether Glo1^+/− mice develop cardiovascular phenotypes, particularly in light of the altered circulating lipid profiles, we assessed the presence of atherosclerotic lesions in the aortic arch using Oil Red O staining. No evidence of aortic lesions was observed in 34-week-old female or male Glo1^+/− mice (n = 4–5/group), suggesting that partial loss of Glo1 alone does not induce atherosclerosis. These findings are consistent with the known natural resistance of C57BL/6 mice to atherosclerosis, as our model was not crossed with atherosclerosis-prone strains such as ApoE⁻/⁻ or Ldlr⁻/⁻, nor were the mice exposed to an atherogenic diet.

3.6. Glo1^+/− Exhibits Perturbed Lipid Pathways in Metabolic Tissues

To determine whether lipid pathways were altered in parallel with the lipid phenotypes observed in Glo1^+/− mice, we examined the expression of key fatty acid and triglyceride metabolism genes in the liver, gonadal adipose tissue, and skeletal muscle of both sexes using qPCR for Lipin1, Acc1, Fasn, Elovl6, Scd1, Srebp1c, Dgat1, and Dgat2 (Figure 4). We identified both sex-specific alterations and shared expression patterns in the fatty acid and triglyceride pathways (Figure 4B), with some genes showing consistent changes across sexes and others exhibiting opposite regulation between males and females (Figure 4A).

Figure 4. Expression of lipid metabolism genes across liver, gonadal adipose and skeletal muscle tissues in Glo1^+/− mice.

(A) Gene expression of key lipid metabolism regulators (Lipin1, Acc1, Fasn, Elovl6, Scd1, Srebp1c, Dgat1, and Dgat2) was assessed in liver, gonadal adipose, and skeletal muscle tissues from 28-week-old female and male mice (n=4 per group). (B) Schematic representation of the lipid metabolism pathway highlighting significant transcriptional changes observed in Glo1^+/− mice. Heatmaps depict tissue-specific expression patterns, with red arrows indicating altered expression in female Glo1^+/− mice and blue arrows indicating changes in males. Upward arrows denote significantly upregulated mRNA expression, whereas downward arrows represent significantly downregulated expression. Genes encoding enzymes within the lipid metabolism pathway that were not assayed (Acs, Gpat, Agpat) are shown in gray. Data were analyzed using unpaired t-tests with Welch’s correction. Statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001.

Across all tissues, both female and male Glo1^+/− mice showed decreased expression of Acc1, which encodes the rate-limiting enzyme in fatty acid synthesis. A similar trend was observed for Lipin1, a gene involved in the conversion of phosphatidate to diacylglycerol during triglyceride synthesis. Although the reduction of Acc1 and Lipin1 in gonadal adipose tissue does not explain the increased adiposity observed in Glo1^+/− mice, it strongly suggests adipose tissue dysfunction, potentially representing either a protective or maladaptive response. In contrast, Fasn, encoding a key enzyme in de novo lipogenesis, and Dgat2, encoding the enzyme that catalyzes the final step in triglyceride synthesis, were upregulated in gonadal adipose tissue in both female and male Glo1^+/− mice (though Dgat2 did not reach statistical significance in males, p = 0.079). The concurrent upregulation of Fasn and Dgat2 supports the obesogenic phenotype observed in both sexes of Glo1^+/− mice.

Genes that exhibited clear sex-specific alterations included Fasn, Dgat1, and Scd1 across metabolic tissues. Dgat1, which also catalyzes triglyceride formation, was significantly upregulated in female gonadal adipose tissue (p = 0.0394) but not in males (p = 0.73), despite the increased adipose depot weight in male Glo1^+/− mice (Figure 1G). In contrast, male Glo1^+/− mice displayed significantly increased Fasn and Scd1 expression in the liver and skeletal muscle, whereas no such changes were observed in females. Scd1 encodes the rate-limiting enzyme in the biosynthesis of monounsaturated fatty acids and is known to protect against adiposity when deleted in mice [40]. Thus, upregulation of Fasn and Scd1 in male Glo1^+/− mice may contribute to lipid accumulation within these tissues, consistent with their increased liver weight (Figure 1G) and higher muscle mass (Figure 1D). Although, female Glo1^+/− mice did not show changes in Scd1 expression but exhibited significantly increased Fasn in skeletal muscle. Elevated Fasn expression in skeletal muscle has been linked to impaired insulin sensitivity and glucose intolerance, consistent with the glucose intolerance phenotype observed exclusively in female Glo1^+/− mice [41]. Collectively, these findings reveal sex-specific alterations in lipid metabolism across multiple tissues that may underlie the divergent metabolic phenotypes observed in Glo1^+/− mice (Figure 4B), while also providing mechanistic insight into female-specific susceptibility to glucose dysregulation.

3.7. Glo1^+/− Mice Show Minimal Evidence for AGE Accumulation or Rage Signaling Activation

AGEs have been implicated as mediators of Glo1-related metabolic dysfunction [1, 4, 10]. To determine whether the phenotypes and lipid pathway alterations observed in Glo1^+/− mice were mediated by methylglyoxal (MG)–derived AGEs, we quantified major AGE species, namely MGH1and CEL as well as related components in metabolic tissues using ELISA, qPCR, and microarray analyses (Figure 5). MGH1 and CEL are two of the most abundant MG-derived AGEs that bind to RAGE and are well-established biomarkers of AGE accumulation [42–44]. Quantification of MGH1 (n = 6–8/group) and CEL (n = 5/group) in the liver, gonadal adipose tissue, and skeletal muscle of both sexes revealed no significant differences in any tissue of female Glo1^+/− mice. In contrast, MGH1 levels were significantly elevated in the skeletal muscle of male Glo1^+/− mice compared with controls, whereas CEL levels remained unchanged (Figure 5B). Overall, these data suggest that metabolic disturbances resulting from Glo1 reduction particularly in females, are unlikely to be primarily mediated by AGE accumulation.

Figure 5. Quantification of AGEs and AGE-related components across liver, gonadal adipose and skeletal muscle tissues in Glo1^+/− mice.

(A) Schematic overview of MG detoxification pathway and the formation of AGEs. (B) Quantification of the MG-derived AGEs CEL and MGH1 by ELISA in 28-week-old male and female mice across metabolic tissues, including liver, gonadal adipose, and skeletal muscle (n = 6–7 per group). (C) Gene expression of Ager, Akr1a1, and Aldh1a1 quantified by qPCR in liver, gonadal adipose, and skeletal muscle tissues (n = 5–6 per group). (D) Microarray-based expression profiling of Akr and Aldh subtypes in female liver, gonadal adipose, aorta, and kidney at 34 weeks of age. (E) GSEA plots showing coordinated expression changes of Akr and Aldh subtypes in the same tissues. Data were analyzed using unpaired t-tests with Welch’s correction. Statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001.

To further evaluate the potential contribution of AGE–RAGE signaling, expression of Ager, the gene encoding RAGE, was examined in the liver, gonadal adipose tissue, and skeletal muscle (Figure 5C). Ager was significantly downregulated in the skeletal muscle of both sexes and trended downward in gonadal adipose tissue, suggesting that RAGE-mediated signaling is unlikely to be a major contributor to the observed phenotypes. To corroborate these findings, downstream targets of RAGE signaling, including genes involved in reactive oxygen species (ROS) production, NF-κB activation, and apoptosis were examined by microarray in the liver, gonadal adipose tissue, and kidney (Supplementary Figure 6). Expression of these downstream signaling genes was not consistently or significantly altered between Glo1^+/− and control mice (Supplementary Figure 6A-D), further supporting the absence of AGE–RAGE pathway activation.

Given the limited evidence for AGE accumulation or RAGE activation, we next investigated whether compensatory detoxification pathways might prevent AGE buildup in Glo1^+/− mice. Gene expression of Akr1a1 and Aldh1a1, key genes that encode enzymes that detoxify MG via alternative metabolic routes [17], was analyzed by qPCR (Figure 5C). Akr1a1 was significantly upregulated in the skeletal muscle of male Glo1^+/− mice, while Aldh1a1 was elevated in the liver of males and skeletal muscle of females. Neither enzyme was altered in gonadal adipose tissue. Complementary transcriptomic analyses of multiple Akr and Aldh isoforms in female liver, gonadal adipose tissue, aorta, and kidney (Figure 5D) revealed significant upregulation of several subtypes such as Akr1b3, Akr1c12, Aldh1a7, Aldh8a1, Aldh3b1, and Akr7a5, particularly in the kidney, as validated by GSEA plots (Figure 5E). This pronounced induction in the kidney suggests enhanced MG detoxification capacity in this tissue.

Collectively, these findings, namely the lack of AGE accumulation in most tissues, downregulation of Ager, and upregulation of Glo1-compensatory enzymes do not support a major role for AGE–RAGE signaling in mediating the metabolic effects of Glo1 reduction. Instead, increased activity of compensatory detoxification enzymes, especially in the female kidney, may facilitate MG clearance by converting it to downstream metabolites such as pyruvate, a key metabolic intermediate that can influence systemic energy regulation.

3.8. Glo1^+/− Exhibit Tissue-Specific Transcriptomic Alterations in Multiple Metabolic Tissues

To uncover the molecular basis of the obesity and glucose intolerance in female Glo1^+/− mice, we profiled the transcriptome of metabolic tissues including gonadal adipose, aorta, and liver in 34-week-old female mice. At FDR < 0.1, a total of 93, 347, 421, 232 differentially expressed genes (DEGs) were found in adipose tissue, aorta, liver, and kidney respectively (Figure 6A), suggesting large-scale transcriptomic alterations in Glo1^+/− female mice.

Figure 6. Summary of DEG signatures in Glo1^+/− mice.

(A) Venn diagram showing the number of significant DEGs (FDR < 10%) identified in adipose, aorta, and liver tissues. (B) Heatmap of individual sample expression changes for genes involved in adipocyte differentiation, adipocytokine signaling, and insulin signaling, showing genes with significant expression differences (p < 0.05). Fold change was calculated by comparing expression values in Glo1^+/− mice to the mean expression of wild-type controls. Red and blue represent increased and decreased expression, respectively. (C) Canonical pathways significantly enriched among DEGs (FDR < 5%), determined using Fisher’s exact test.(D) Associations between Glo1 regulated genes in female mice and human metabolic traits, as determined by MSEA. Asterisks denote significant associations (FDR < 5%). CAD, coronary artery disease; T2D, type 2 diabetes; BMI, body mass index; HIP, hip circumference; WC, waist circumference; WHR, waist-to-hip ratio.

Glo1^+/− gene signatures exhibited strong tissue specificity when we compared the DEGs across tissues (Figure 6A). For adipose tissue, aorta, liver, and liver, 71.0%, 83.9%, 81.7%, and 73.7% of signatures were unique for each respective tissue. Two genes were consistently down-regulated in adipose, liver, and aorta, including Nqo2 (N-Ribosyldihydronicotinamide: Quinone Reductase) and Tuba1a (Tubulin Alpha 1a). Nqo2 is an antioxidant that may protect against diabetes-induced endothelial dysfunction [45], whereas Tuba1a is the gene responsible for producing cytoskeleton protein tubulin, a known target for glyoxalase system defect [5]. The coordinated downregulation of these genes suggests that reduced Glo1 expression may promote oxidative stress and compromise cellular structural integrity.

3.9. Glo1 Reduction Perturbs Metabolic Pathways Systemically

To investigate the molecular mechanisms underlying the metabolic phenotypes observed in Glo1^+/− females, we performed pathway analyses across multiple tissues. In adipose tissue (Figure 6B), several regulators of adipocyte differentiation, adipocytokine signaling, and insulin signaling displayed significant expression changes. Moreover, key genes involved in lipid and glucose metabolism including Elovl6, Fasn, Lep, Ppargc1a, Mapk9, and Pck1 were differentially expressed. Across tissues (Figure 6C), functional annotation revealed significant enrichment of pathways associated with insulin resistance, insulin-like growth factor (IGF) binding, and monocarboxylic acid metabolism, consistent with impaired glucose homeostasis [46–49]. Dysregulation of lipid metabolic pathways was detected across multiple tissues, including AMPK signaling, IGF1–mTOR, and glucocorticoid receptor pathways in adipose tissue; fatty acid metabolism, lipid biosynthesis, and unsaturated fatty acid metabolism in the aorta; PPAR signaling, fatty acid metabolism and degradation, and gluconeogenesis in the kidney; and MAPK signaling and biological oxidations in the liver. Collectively, these results demonstrate that Glo1 reduction perturbs metabolic regulation systemically, leading to widespread disruption of glucose and lipid metabolism across tissues, which may underlie the obesity, hyperlipidemia, and glucose intolerance phenotypes observed in Glo1^+/− females.

3.10. Glo1 Reduction Affects Genes Genetically Linked to Human Metabolic Diseases

Although Glo1 has been previously implicated as a candidate gene for human metabolic disorders [1, 8–10], the mechanisms linking its dysregulation to disease susceptibility remain unclear. To extend our DEG findings (Figure 6A–6C) and assess their relevance to human health, we integrated tissue-specific Glo1^+/− DEG signatures with large-scale human genome-wide association study (GWAS) data using Marker Set Enrichment Analysis (MSEA) from the Mergeomics platform [39]. We found that Glo1-regulated genes exhibit significant enrichment for human metabolic traits (Figure 6D). The strongest and most consistent associations were observed for lipid-related traits, including LDL, HDL, and total cholesterol (TC), followed by anthropometric traits such as body mass index (BMI), height, and waist-to-hip ratio (WHR). Furthermore, liver-specific signatures correlated with human glycemic traits including fasting glucose, HbA1c, and HOMA-IR, are consistent with the hyperglycemia and impaired glucose tolerance observed in Glo1^+/− females. Notably, significant associations were also detected for coronary artery disease (CAD) and type 2 diabetes (T2D), reinforcing the translational relevance of our mouse findings. Collectively, these results demonstrate that Glo1 reduction alters genes overlapping with human loci linked to dyslipidemia, insulin resistance, and cardiometabolic disorders, providing mechanistic insight into how impaired detoxification capacity may contribute to systemic metabolic disease risk.

3.11. Identification of Transcription Factor (TF) Hotspots Underlying Glo1^+/− Signature

To better understand how Glo1 reduction leads to the perturbations of DEGs in individual tissues, we explored transcription factors (TF) that may mediate the gene expression changes induced by Glo1 knockdown. We utilized the computational tool BART to predict TF enrichment for Glo1^+/− signatures. Interestingly, PPARg, the master regulator of adipogenesis, is among the top perturbed adipose TF hotspots, as its target genes are highly enriched among the adipose DEGs (Table 1). Moreover, several other TFs implicated directly or indirectly in adipogenesis were also highly ranked in adipose TF hotspots, including CEBPB/A, PPARA, HNF4A, and RXRA. These findings align with the increased adiposity in female Glo1^+/− mice. In aorta, we identified perturbations in a number of TFs involved in circadian clock regulation (CRY1, CRY2, PER2) [50], vascular inflammation (PPARG, NR3C1) [51], and those linked in coronary artery disease, including the estrogen receptor, ESR2. In the liver, several TFs are related to circadian clock regulation (CLOCK, CRY2, NR1D1), liver metabolism (NR1D1, RORA) and liver development (KDM6A, MYC, HDAC3), indicating the potential of Glo1 to alter a wide spectrum of physiological activities in the liver. Lastly, in the kidney following a similar trend to the liver we highlight various circadian TFs (CLOCK, CRY2, ARNTL, NR1D1), kidney development TFs (POLR2B, HNF4A, HDAC3, MYC, KDM6A) and those involved in glucose and lipid metabolism (NR1H2, PPARA, RORA). With sex differences being present in our phenotypic results, we looked to highlight at the TF level if there are any known TFs that have a female bias to potentially explain these differences. Of interest, when comparing our results to known sex biased TFs from the GTEx study [52], we highlight that across each tissue that a minimum of 22% of our TFs have evidence of being female biased: adipose (16/60, 27%), aorta (14/64, 22%), liver (11/44, 25%) and kidney (9/38, 24%). Importantly, a number of these female biased TFs are related to circadian rhythm (ARNTL, NR1D1), tissue development (HNF4A) and metabolism (PPARA, HNF4A), which collectively may in part explain female-specific phenotypic differences including poorer glucose tolerance. These tissue-specific TFs likely mediate the effects of Glo1 reduction on the molecular pathways altered in individual tissues.

Table 1.

Top 15 transcription factors whose downstream targets are enriched for Glo1^+/ DEG signatures in adipose tissue, aorta, liver and kidney. Significant enrichment p values are shown.

Gene Symbol	Liver	Adipose	Kidney	Aorta
HDAC3	6.32E-04	1.97E-05	4.77E-04	2.19E-05
RXRA	1.19E-03	4.54E-05	9.92E-05	9.31E-05
NR1D1	2.14E-04		9.55E-06	1.66E-04
CRY2	1.01E-03		2.70E-04	2.24E-04
PPARA		3.52E-05	4.42E-04	1.24E-05
NR1H2		6.11E-04	2.35E-04	2.82E-04
HNF4A		8.97E-04	7.13E-05	9.92E-05
POLR2B	1.34E-04		6.64E-05
XBP1	2.04E-04		1.19E-04
KDM6A	3.21E-04		5.70E-04
RORA	4.09E-04		1.84E-04
MYC	5.51E-04		4.77E-04
CLOCK	6.98E-04		4.94E-04
NR3C1		1.97E-06		8.30E-06
PPARG		3.03E-06		1.12E-07
CEBPB		1.41E-04		3.63E-04
CEBPA		4.26E-04		5.13E-04
CREB1	6.11E-04
H2AZ	1.49E-04
KMT2C	4.59E-04
PHF8	7.21E-04
GTF2B	8.70E-04
EP300		1.84E-04
STAT5B		6.11E-04
STAT5A		8.97E-04
NCOR1		9.52E-04
ESR1		1.16E-03
RXRG		1.26E-03
ARNTL			1.84E-04
ARID1A			3.35E-04
PRDM16				2.14E-04
ESR2				3.08E-04
PER2				3.08E-04
CRY1				3.63E-04

3.12. Enrichment of Glo1 DEGs for Sex-Biased and Aging Gene Programs

To further examine whether Glo1-regulated transcriptional programs overlap with known human sex-biased or age-related genes, we compared Glo1 DEGs (q ≤ 0.05) with tissue-matched sex- and age-biased genes from GTEx and the curated human aging gene set from GenAge (Supplementary Table 2). Compared to sex-biased genes, we found that kidney Glo1 DEGs were significantly enriched for human kidney sex-biased genes (OR = 3.74, Fisher’s exact p = 0.0074), with overlapping genes (HMGCS2, ETFDH, SLC25A20, OPLAH) pointing to fatty acid oxidation as potential sex-influenced processes modulated by Glo1. In contrast, liver and aorta DEGs showed non-significant (p=0.08 and 0.3 respectively) overlap with human sex-biased genes, and adipose DEGs had no overlap with human sex-biased genes. Compared to age-related genes, liver Glo1 DEGs showed a significant overlap (OR=2.96, p=0.033) with GenAge genes, including IGF1, FGF21, IGFBP2, GSTP1, and HSPA8, regulators of growth hormone/IGF signaling, proteostasis, and stress adaptation [29]. Similarly, aorta Glo1 DEGs were significantly enriched for GenAge genes (OR=2.25; p=0.044) such as FOXO1 and CDKN1A, key transcriptional and cell-cycle regulators in aging pathways, but not significantly enriched for GTEx age-associated genes in aorta despite an overlap of 18 genes (e.g., CXCL9, DDIT4, CYR61, CTSZ), several of which are linked to vascular inflammation and stress responses. Kidney had three shared DEGs with GenAge genes (PCK1- gluconeogenesis enzyme, EGR1- stress response, CEBPB - immune and metabolic regulation) and adipose DEGs showed only a single overlap with age-related DEGs (IGFALS, an insulin-like growth factor binding protein with roles in growth and metabolism). Together, these analyses support that Glo1 intersects with human age- and sex-biased genes in a tissue-specific manner, with stronger convergence in kidney for sex-biased genes and in liver and aorta for age-related genes.

3.13. Inference of the potential role of MG metabolite pyruvate in regulating Glo1^+/− signatures

To further explore regulatory molecules that reprogram the transcriptome, we focused on MG metabolites produced by compensatory enzymes that we found to be upregulated in our gene expression analysis. In particular, pyruvate can be produced from MG detoxification which can further regulate gene expression. Indeed, we found that the genes involved in gluconeogenesis (Pcx, Pck, Enol, Pgam1, Pgk1, Gapdh, Fbp1 and G6pc) were mostly upregulated in female Glo1^+/− kidney and to a lesser extent in female Glo1^+/− liver, key tissues known to be involved in gluconeogenesis (Figure 7). These results suggest a potential role of MG metabolites in regulating gene expression and glucose production. Although a few gluconeogenesis genes were also found to be altered in the aorta (Pcx, Fbp1) and gonadal adipose (Pck) tissues (Supplementary Figure 7), these changes likely reflect metabolic reprogramming and feedback mechanisms rather than gluconeogenesis.

Figure 7. Schematic summary of the gluconeogenesis pathway illustrating the potential role of the MG metabolite pyruvate in regulating Glo1^+/− signatures.

Schematic overview depicting the detoxification of MG by Aldh in the kidney, leading to the generation of pyruvate. The resulting pyruvate can enter the gluconeogenesis pathway, contributing to glucose synthesis. Asterisks denote enzymes showing significantly increased gene expression in Glo1^+/− mice. Statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001.

4. DISCUSSION

To explore the role of Glo1 in metabolic disorders, we conducted a systems biology study of Glo1^+/− mice (Figure 8). We performed extensive metabolic phenotyping in both sexes at various ages and complemented these analyses with transcriptomic and AGE profiling of key metabolic tissues. This was followed by network and integrative genomics analyses to elucidate the underlying regulatory cascades and their relevance to human disease.

Figure 8. Schematic summary of the effects of Glo1 reduction in female and male mice.

Schematic overview of the study design illustrating metabolic phenotypes induced by partial Glo1 deficiency in female and male mice over time, with emphasis on the more pronounced female phenotypes. The proposed mechanisms underlying these sex-specific metabolic alterations are also depicted.

From this study, we found that Glo1^+/− mice demonstrated metabolic dysregulation, including increased body weight and fat mass in both sexes in adulthood. Alterations in metabolism was also observed in a sex-dependent manner, including increased whole body muscle mass, and decreased plasma lipids such as TC, TG and HDL in male Glo1^+/− mice, whereas female Glo1^+/− mice demonstrated glucose intolerance and elevated plasma lipids such as TG and VLDL. The metabolic phenotypes observed in Glo1^+/− female mice were paralleled by large-scale tissue-specific expression alterations in metabolic genes and pathways across tissues. Select lipid metabolic pathway alterations were also confirmed by qPCR experiments in both males and females. The tissue-specific molecular signatures were further tied to numerous tissue-specific TFs that regulate adipogenesis, insulin signaling, aortic endothelial functions, and key liver functions, and collectively demonstrated significant association with multiple human metabolic traits. To explain the source for the observed systemic effects of Glo1 reduction, we quantified the most prominent AGEs (MGH1 and CEL) in metabolic tissues by ELISA and found AGE levels did not differ between Glo1^+/− mice and WT controls in most tissues. Our findings agree with those from the Wortmann et al. study which did not detect a difference in AGEs in the liver and kidney between Glo1^+/− and WT by ELISA [53]. In contrast, Giacco et al. did detect significantly elevated AGEs in the kidney by immunohistochemistry (IHC) between Glo1^+/− and WT [23]. We did not quantify kidney AGEs or use IHC. The differences in results may arise from different techniques used to acquire data and should be further assessed in future studies. Our results led to our assessment of potential changes in AGE receptor (Ager) gene expression, which did not show significant differences between Glo1^+/− mice and WT mice (Figure 5C). Further evaluation of Rage signaling in liver, gonadal adipose and kidney in terms of ROS signaling genes, Nfkb signaling genes and apoptosis signaling genes also failed to show consistent and significant alterations between Glo1^+/− mice and controls. Overall, the phenotypes observed in our Glo1^+/− mice cannot be explained by the AGE pathway as a major factor in our study.

Our observation that Glo1^+/− exhibited increased weight and adiposity is consistent with the previous linkage study that revealed Glo1 resided under a QTL for body weight in mice [54]. In females, the body weight increases in Glo1^+/− mice was largely due to growth of various adipose depots since lean mass remained on par with control mice, whereas increased body weight in male Glo1^+/− mice were attributed to both fat and muscle mass increase. In parallel, Glo1^+/− female mice also had elevated circulating TG and VLDL levels. Lipid metabolism genes Fasn and Dgat2 were also upregulated in female and male Glo1^+/− mice in gonadal adipose, supporting the obesogenic phenotypes observed (Figure 1, Figure 4, Figure 8). Along with the obesity and hyperlipidemia phenotypes, female Glo1^+/− mice also demonstrated comorbidities such as glucose intolerance. As shown in our data (Figure 7), generation of metabolites such as pyruvate that arise from MG clearance by compensatory enzymes such as Aldh may promote reprogramming of transcriptome and may also contribute to the dysregulated glucose phenotypes in female Glo1^+/− mice.

Surprisingly, male Glo1^+/− mice showed no difference in glucose tolerance and a decrease in plasma glucose at 12 weeks of age but not 7 or 28 weeks of age when compared to WT. Our results align with a study conducted by Schumacher et al. that used a complete Glo1 knockout mouse model in male mice and observed no change in blood glucose when compared to controls [17]. Male Glo1^+/− mice also showed significantly lower plasma TG, TC, and HDL levels, which are generally associated with a reduced cardiovascular risk profile. For example, reduced HDL is generally associated with greater risk of atherosclerotic cholesterol accumulation [55], although its causal role has not been established. Similarly, very low TG levels may indicate underlying conditions such as impaired hepatic lipid synthesis or export, malnutrition, or systemic illness [56], although low TG is also associated with a reduced risk profile for cardiovascular disease [57]. Both lower circulating lipids and the preserved glucose tolerance observed in male Glo1^+/− mice could be due to their increased total muscle mass, likely reflecting a compensatory adaptation in which greater muscle mass in male Glo1^+/− mice facilitates systemic glucose homeostasis. The transcriptional profile of male skeletal muscle marked by elevated Scd1 and Dgat2 expression, suggests remodeling of lipid metabolism. Increased Scd1 expression in muscle has been associated with increased polyunsaturated fatty acids, metabolic function, exercise capacity, and lipid oxidation [58]. Upregulation of Dgat2, a key enzyme in triglyceride synthesis, supports increased lipid storage capacity within skeletal muscle. Notably, male Glo1^+/− mice also exhibited increased fat mass across multiple depots such as gonadal, retroperitoneal, and mesenteric adipose tissue along with increased Fasn and Dgat2 expression in adipose tissue (Figure 8). Together, these findings suggest that differential lipid metabolism in muscle and adipose tissue may underlie the distinctive metabolic phenotype observed in male Glo1^+/− mice compared to female mice.

Interestingly, many of the perturbed metabolic profiles were observed after 12 weeks of age, suggesting an age-dependent effect in Glo1^+/− mice. Previous studies such as Prevenzano et al. and Wortmann et al. assessed Glo1 knockdown mice primarily in males [53, 59] and found a lack of metabolic phenotypes in younger mice. Prevenzano et al. additionally found compromised glucose tolerance in Glo1 males at 10-month age along with an increasing trend in body weight but not at an earlier age. In contrast, our study utilizes a longitudinal approach with both male and female cohorts, tracking metabolic outcomes from early post-weaning to late adulthood. Our data reveal more female-biased metabolic dysregulations at mid to late age. The sex and age differences may explain the different phenotypic observations. The interaction of Glo1 with age and sex may explain the limited evidence from human studies and previous animal model studies since age- and sex-stratified analysis is necessary to reveal the associations.

To further understand the sex-biased mechanisms, investigation of the female adipose transcriptome revealed alterations in numerous pathways and key regulatory genes involved in adipocyte differentiation, adipocytokine signaling, insulin signaling, fatty acid and triglyceride biosynthesis, and PPAR signaling (Figure 6B) collectively indicative of a state that promotes adipogenesis and increased obesity risk [60, 61]. This is further supported by the identification of adipogenic master regulators, including PPARG, TP hotspots for the DEGs altered in Glo1^+/− female mice. The liver gene signatures were significantly enriched for pathways involved in fatty acid metabolism, bile acid metabolism, and metabolism of lipids (Figure 6C), consistent with the altered plasma TG and VLDL levels (Figure 3A, D) observed in Glo1^+/− females.

As males exhibited decreased lipid profiles and lower fasting glucose levels, we examined genes involved in lipid metabolism in both sexes using qPCR and found distinct, tissue-specific changes in several genes (Fasn, Scd1, Srebp1c, among others) between sexes, providing molecular support for the divergent phenotypic manifestations. However, we observed consistent alterations in genes such as Acc1 (fatty acid synthesis) and Lipin1 (triglyceride synthesis). The differing lipid profiles between sexes likely reflect collective, sex-dependent differences in the activity of lipid metabolic pathways across multiple tissues (Figure 4A).

Our characterization of sex-specific responses in metabolic traits yielded novel results suggesting that female mice are more vulnerable to Glo1 reduction, particularly in maintaining glucose and lipid homeostasis. Four out of seventeen genes (Ebp, Lss, Hsd17b7, and Nsdhl) in the steroid synthesis pathway showed altered expression exclusively in female adipose tissue (fold change = 6.3, p = 2.6e-4) and estrogen receptor ESR2 was one of TFs enriched for Glo1 DEGs. As steroid synthesis and estrogen signaling are critical pathways regulating sexually dimorphic adiposity, they may serve as mediators of the effects of reduced Glo1 on adiposity in females. However, attributing the phenotypic differences between female and male mice to dysregulation of steroid biosynthesis, in the absence of measuring circulating estradiol or testosterone and examining gonads, is premature. Our interpretation of the observed sex differences is based on inference of the steroid biosynthesis pathway and transcription factors such as estrogen receptor are speculative and direct measurement of sex hormones in the circulation and investigation of the gonads will be necessary to substantiate this hypothesis.

We also note that the full transcriptome was examined only in females, while select lipid-related genes were tested in both sexes. Thus, further investigation is warranted to evaluate additional sexually dimorphic pathways. Some aspects of the female-specific vulnerability to glucose intolerance may also be explained by transcription factors with known female-biased activity. These include regulators involved in circadian rhythms which influence glucose metabolism, and others (e.g., Hnf4a, a core TF identified across all tissues analyzed) which play dual roles in lipid and glucose metabolism. Notably, multiple circadian TFs were found to be enriched in the liver, kidney, and aorta, such as CLOCK, ARNTL, CRY1, and CRY2. These TFs likely contribute to aging- and sex-dependent phenotypes because circadian programs in these tissues are known to undergo age-related reprogramming and exhibit sex differences [62–68].

Glo1 has been previously implicated as a key regulator for CAD [8]. We did not observe direct evidence for atherosclerotic lesions in 34-week-old Glo1^+/− mice, likely due to the natural resistance of mice for atherogenesis without extreme genetic and dietary perturbations. However, the transcriptomic profiling of aorta provided supportive links between Glo1 and atherosclerosis. Pathway analysis of the aorta DEGs showed significant alterations related to lysosome and lipid metabolism, processes implicated in atherogenesis [69, 70]. Additionally, TF prediction for the aorta DEGs identified TFs that are related to circadian rhythms and endothelial inflammation, which are also important processes for atherogenesis [51, 71]. Furthermore, the Glo1 DEG signatures from multiple tissues examined were also strongly associated with lipid profiles (TC, LDL, and HDL) and CAD in human GWAS. The negative atherosclerotic phenotype in Glo1^+/− mice could be due to insufficient reduction of Glo1 activity, natural resistance to the development of atherosclerotic lesions and no enforced genetic or environmental stress in our study. Previous studies on the role of Glo1 in atherosclerosis have also yielded inconsistent results. Geoffrion et al. showed that Glo1 transgene or Glo1^+/− did not affect the progression of atherosclerosis in the ApoE^−/− mice at 22 weeks of age, even under diabetic conditions [72]. On the other hand, Jo-Watanabe et al. showed that Glo1 transgenic mice on Apoe^−/− background under a high-fat diet went through a reduction of glycation and oxidative stress while preventing age-related endothelial dysfunction by the prevention of eNOS inactivation [73].Future studies that focus on the long-term effects in atherosclerosis susceptible models with significant genetic or environmental challenges could yield more power in elucidating the potential causal role of Glo1 in atherosclerosis.

A unique finding from our study is that in the presence of a compromised glyoxalase system such as in our Glo1^+/− mice, compensatory mechanisms and enzymes such as Akr and Aldh increase and may reduce the accumulation and effects of MG in the body and explain the lack of prevalent changes in AGEs. While initially, this wards off any immediate effects and consequences of accumulated MG, long-term effects resulting from the alternative methods of detoxifying MG may lead to increased metabolites such as pyruvate, which feeds into gluconeogenesis, may further reprogram metabolism.

To our knowledge, this is the first study to conduct a comprehensive study to show evidence for both sex- and age-dependent effects that result from Glo1 reduction in mice. In addition, our results highlight numerous TFs and DEGs that overlap with sex- and age-associated genes cardiometabolic diseases in humans, which may in part explain the sex- and age dependent phenotypes observed in Glo1^+/− mice and confer human relevance of our findings.

In summary, our findings support the role of Glo1 reduction in modulating obesity and metabolic dysfunction in a sex- and age- specific manner without strong evidence for the involvement of AGEs (Figure 8). Our discoveries regarding tissue-specific genes, pathways and transcription factor hotspots involved in metabolic regulation downstream of Glo1 in multiple metabolic tissues support broader molecular functions of Glo1 beyond its known role in the glyoxalase system. Moreover, the application of these findings to human disease further supports the importance of exploring the physiological functions and pathogenic potential of Glo1.

Supplementary Material

Supplemental Figs. S1-S7: http://datadryad.org/share/LINK_NOT_FOR_PUBLICATION/Qp5VGcPKOO_EsBlwO8yzH8nAYEJN2Q-doIz5YjG1vDs

Supplemental Tables S1–S2: http://datadryad.org/share/LINK_NOT_FOR_PUBLICATION/Qp5VGcPKOO_EsBlwO8yzH8nAYEJN2Q-doIz5YjG1vDs