Comprehensive metabolic profiling across five lifespan stages in murine hippocampus and cortex reveals sex-related variation in age-related cognitive decline

Abstract

This study employs Barnes maze behavioral assessments, untargeted liquid chromatography-mass spectrometry metabolomics, and ¹³C₆-glucose isotopic tracing to systematically investigate cognitive function and metabolic profiles in hippocampal and cortical tissues of male and female mice across five distinct age-ranges. Behavioral analyses reveal significant cognitive decline in both sexes by 16-months-of-age, with females exhibiting more severe impairment by 23-months, demonstrating a sex-related variation. ¹³C₆-glucose tracing analyses reveals that glucose is rapidly and preferentially metabolized toward the Tricarboxylic acid cycle over glycolysis and the pentose phosphate pathway (PPP), with metabolism rates increasing from juvenility to meet developmental demands and maintaining homeostasis into pre-elderly. Surprisingly, glucose metabolism continues to rise in elderly males but declines in females. Developmental shifts from purine biosynthesis to degradation display sex-related variation, highlighting sustained synthesis in elderly males versus degradation in aging females. Finally, age and sex- related differences in amino acids, neurotransmitters, histidine-derived antioxidants, and the arginine-urea cycle further underscore complex metabolic reprogramming in the CNS. Overall, our study elucidates from a metabolic perspective the molecular basis of sex-related variation in age-related cognitive decline by characterizing sex-related variation in reprogramming of glucose, purine, and amino acid metabolic networks.

Integrated behavioural & metabolomic profiling across the mouse lifespan reveals sex-specific brain metabolic reprogramming underlying divergent trajectories of age-related cognitive decline.

Introduction

The mechanisms of cognitive decline, a central feature of brain aging, have not been fully elucidated, especially the gender-specific differences. Clinical observations have shown that women exhibit higher susceptibility to neurodegenerative diseases such as Alzheimer’s disease and faster cognitive decline¹. However, current research on age-related cognitive decline tends to ignore gender as a key biological variable, leading to significant gaps in understanding the mechanisms underlying gender differences in cognition with increasing age.

The brain, one of the most complex organs in mammals, is characterized by its unique anatomical regions, each integral to brain development and function throughout lifetime. The cerebral cortex and the hippocampus are key regions of CNS that control cognitive functions and memory². The cerebral cortex is responsible for processing and interpreting incoming sensory information, and integrating and transmitting this information to the hippocampus, which encodes and stores memory³. Conversely, during memory recall, the hippocampus feeds information back to the cerebral cortex to guide cognitive and behavioral responses during the memory retrieval and consolidation phases. In general, the cerebral cortex is more responsible for higher cognitive functions, while the hippocampus is more focused on learning and memory⁴. Together, they support the complex cognitive and behavioral performance of all mammals. The well-regulated communication between the cortex and hippocampus through intrinsic synaptic connections, neurotransmitter release, and neural network activity is essential for learning and memory during early juvenile development. In contrast, aging-dependent cognitive and memory decline is a multifaceted process, associated with factors like hippocampal long-term potentiation deficits⁵, oxidative protein damage in the CNS⁶, dopaminergic dysfunction⁷, and neuroinflammation⁸. Without intervention, age-related impairments in learning, memory, motor functions, and language⁹ progress to various neurodegenerative diseases, including Alzheimer’s Disease (AD) and Parkinson’s Disease (PD), conditions with major impact in the growing elderly population worldwide¹⁰. Notably, cognitive functional changes during development and cognitive decline during aging (including AD) exhibit sex-specific variations^11,12. In the United States, approximately two-thirds of AD patients are female¹³ and the lifetime risk of developing AD for women is twice that of men^12,13. Although substantial effort and research have been conducted, it remains a long-term puzzle why women have a higher chance of developing age-dependent cognitive decline and AD¹⁴. Remarkable research in mice has revealed a substantial detailed snapshot of the mammalian CNS from the perspective of genomics, and transcriptomics^15,16. Considerable research is underway to identify cellular and molecular mechanisms to prevent or treat aging-related cognitive decline and the progression to neurodegenerative diseases. There is a particular need for a better understanding of the metabolic changes underlying the decline in cognitive development from adolescence to the elderly, with a special focus on sex-related differences, which has often been overlooked in research and relatively underexplored.

Emerging evidence indicates that aging-related cognitive decline and dementia are frequently associated with metabolic impairment. Glucose as the primary fuel in the CNS is completely oxidized via TCA cycle to generate sufficient ATP to maintain neuronal function and survival under aerobic conditions^17,18 or metabolized anaerobically to lactate¹⁹. Besides glycolysis, glucose undergoes metabolism through PPP to counteract oxidative stress and provide key metabolites for nucleotide biosynthesis²⁰. Substantial evidence indicates that even only a few minutes of glucose and oxygen deprivation triggers significant synaptic dysfunction and cognitive impairment^21,22, highlighting the CNS’s intolerance to any disruptions in glucose and oxygen supply²³. In the dynamic process of CNS metabolism, amino acids (AAs) play a pivotal role, not only as basic components for protein synthesis but also as key regulators of neurotransmission and metabolic pathways²⁴. For example, neurotransmitters such as glutamate and γ-aminobutyric acid (GABA), as well as aspartate and N-acetyl aspartate, play key roles in bidirectional information transfer through complex neural circuits and synchronized oscillations between neurons²⁵. Additionally, neurotransmitter modulators including purinergic metabolic components such as adenosine (signaling via specific receptors) affect the excitability and inhibition of specific neurons in the CNS and modulate neuronal plasticity, learning, memory, motor functions, and various physiological processes²⁶. Taken together, the intricate and complex glucose, AAs, and purine metabolic reprogramming significantly impacts brain health and function. However, the in vivo glucose metabolic dynamics, comprehensive metabolic networks and sex-specific differences throughout lifetimes in the CNS remain unknown.

Here, we employed a multi-dimensional research approach to systematically investigate age-dependent sex-related variation in cognitive function. Initially, we conducted comprehensive cognitive assessments of both male and female mice across five developmental stages (1, 2, 10, 16, and 23 months) using the Barnes maze test. Building upon these behavioral findings, we subsequently employed advanced mass spectrometry and isotopic ¹³C₆-glucose flux analysis coupled with precise amino acid quantification to elucidate metabolic dynamics in cognition-related brain regions (cerebral cortex and hippocampus). This integrated strategy from behavioral phenotyping to metabolic mechanism investigation not only established a functional-metabolic correlation map but more importantly, revealed the critical role of metabolic reprogramming in shaping sex-related variation during cognitive development and decline. Our findings provide novel experimental evidence and a theoretical framework for understanding sex differences in age-related cognitive impairment.

Results

Cognitive function in mice declines with age showing significant sex-related variation only at 23 months of age

To systematically investigate age-related cognitive changes in mice of both sexes, we conducted a longitudinal analysis using the Barnes maze test across five key developmental stages (juvenile [1 month], adult [2 months], middle-aged [10 months], pre-elderly [16 months], and elderly [23 months]) (Fig. 1a). The results demonstrated a progressive decline in spatial learning and memory performance with aging in both male and female mice. Notably, 16 months of age emerged as a critical turning point, where mice first exhibited significant cognitive impairment without showing sex differences. As aging progressed to 23 months, cognitive decline accelerated and began to display marked sex-related variation (Figs. 1c, d and S1a). Specifically, 23-month-old female mice required significantly more time to locate the target box in the escape latency test, indicating more severe impairment in spatial learning and memory compared to age-matched males. This study reveals the stage-dependent characteristics of age-related cognitive decline, with 16 months marking the onset of significant deterioration and 23 months showing evident sex differences.

Fig. 1. Cognitive function in mice declines with age showing significant sex-related variation only at 23 months of age.

a, b Study design schematic illustrating the cognitive testing paradigm across five age groups of male and female mice, followed by collection of hippocampus and cortex at 0.5, 1.5, and 2.5 h after ¹³C₆-glucose injection for metabolomic analysis. This figure was created using BioRender (https://BioRender.com/a7lnqo2). c Escape latency for individual male and female mice during the fifth day of Barnes maze testing. Data represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. P value was assessed by unpaired two-tailed t-test. The experimental design comprised five age cohorts of biologically independent animals: 1-month-old mice (n = 13; 8 female, 5 male), 2-month-old mice (n = 16; 8 female, 8 male), 10-month-old mice (n = 16; 8 female, 8 male), 16-month-old mice (n = 15; 8 female, 7 male), and 23-month-old mice (n = 15; 7 female, 8 male). d Representative movement traces of 23-month-old male and female mice in the Barnes maze.

To rule out whether the age-dependent cognitive decline is linked with the reduced peripheral moving velocity during aging, we carefully measured the average movement speed in both sexes during aging. Specifically, at 16 months of age, male mice exhibited a significant reduction in locomotor activity, whereas their female counterparts showed only a mild decline. Despite this disparity in motor function, both sexes displayed a comparable and significant degree of cognitive impairment at this stage (Figs. 1c and S1b), indicating a notable dissociation between motor and cognitive capabilities at the 16-month-old age. Furthermore, no progressive decline of the motor speed at 23-month-old compared to 16-month-old male mice, while the 23-month-old female mice displayed a significant reduction of the motor velocity, reaching similar levels as those of 23-month-old male mice (Figs. 1c and S1b). Altogether, the reduced motor ability is not the decisive factor for the aging-induced cognitive decline, and the sex-related variation in cognitive deterioration at 23-month-old mice is independent of the decreased peripheral movement velocity.

Rapid glucose metabolism in the cortex and hippocampus

Our initial behavioral assessments revealed that while cognitive decline begins at 16 months in both sexes, female mice develop significantly more severe cognitive impairment by 23 months, demonstrating clear sex-related variation in age-related cognitive deterioration. Given the crucial role of glucose as the brain’s primary energy substrate, we sought to investigate whether sex-specific metabolic differences underlie these cognitive variations. To comprehensively map glucose metabolic dynamics in cognition-related brain regions across the lifespan, we performed detailed ¹³C₆-glucose tracing in the hippocampus and cortex of 10 male and 10 female mice at five developmental stages: Juvenile (1 M), Adult (2 M), Middle-aged (10 M), Pre-elderly (16 M), and Elderly (23 M). Using intravenous administration of ¹³C₆-glucose followed by tissue isolation at three time points (0.5, 1.5 and 2.5 h post-injection), we precisely characterized in vivo glucose metabolic reprogramming dynamics throughout the lifetime in both sexes (Fig. 1b). This experimental design enabled systematic examination of potential sex-divergent metabolic patterns that may contribute to the observed differences in cognitive aging trajectories.

First, we observed that isotopically labeled glucose was rapidly metabolized into glycolytic, TCA cycle, and PPP intermediates, yet nearly all of these labeled metabolites declined to very low or undetectable levels by 2.5 h post-injection across all age groups and in both sexes, in both the hippocampus and cortex (Fig. S2a–c and Table S1). In contrast, at 1.5 h, a substantially higher abundance of ¹³C-labeled metabolites was detected (Fig. S3a–c). Among these, labeled lactate (M + 3) and lactate (M + 2)—the end products of glycolysis and the PPP, respectively—were already prominent at 0.5 h (Fig. 2c, e), indicating a rapid initial metabolic flux. Furthermore, analysis of the 0.5-h time point revealed that glucose was largely and rapidly directed into the TCA cycle relative to glycolysis and the PPP in both sexes throughout life, as evident from the rapid generation of a series of ¹³C-labeled TCA cycle intermediates (citrate, succinate, fumarate, malate) (Fig. 2b–i). Taken together, these results demonstrate that glucose catabolism through glycolysis, the PPP, and the TCA cycle is a rapid and dynamic process in the hippocampus and cortex of both sexes from development to aging. This process was characterized by a swift rise in metabolite flux, which peaked early (by 0.5 h) and was subsequently followed by a sharp decline to very low levels by 2.5 h (Fig. S3d).

Fig. 2. Sex-related variation in glucose metabolism dynamics and reprogramming from development to aging.

a Schematic diagram of ¹³C₆-glucose labeling to intermediates in glycolysis, PPP and TCA cycle. This figure was created using BioRender (https://BioRender.com/g9dk9hs). b, c ¹³C₆-glucose incorporation into glycolytic metabolites (pyruvate and lactate). d ¹³C₆-glucose incorporation into amino acids (serine and alanine). e ¹³C₆-glucose incorporation into PPP. f The ratio of lactate(M + 3)/lactate(M + 2). g ¹³C₆-glucose incorporation into AMP. h, i ¹³C₆-glucose incorporation into TCA cycle metabolites, citrate, succinate, fumarate, malate, glutamate, glutamine and aspartate. j Schematic diagram of sex-related variation in energy metabolism in the hippocampus and cortex. This figure was created using BioRender (https://BioRender.com/i1stj2t). The experimental cohorts comprised five age groups of biologically independent animals. For hippocampal analyses: 1-month-old mice (n = 12; 6 female, 6 male), 2-month-old mice (n = 9; 4 female, 5 male), 10-month-old mice (n = 16; 8 female, 8 male), 16-month-old mice (n = 12; 6 female, 6 male), and 23-month-old mice (n = 14; 8 female, 6 male). For cortical analyses: 1-month-old mice (n = 12; 6 female, 6 male), 2-month-old mice (n = 10; 4 female, 6 male), 10-month-old mice (n = 16; 8 female, 8 male), 16-month-old mice (n = 12; 6 female, 6 male), and 23-month-old mice (n = 14; 8 female, 6 male). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were analyzed by one-way ANOVA with Tukey’s post hoc test for multiple comparisons. PDH pyruvate dehydrogenase, AMP adenosine monophosphate.

Sex-related variation in glucose metabolism dynamics and reprogramming from development to aging

The sex- and age-dependent differences in the percentage of labeled glucose-derived intermediates were most pronounced at 0.5 h after injection, suggesting distinct metabolic vigor in glycolysis, the PPP, and the TCA cycle across the lifespan between females and males. We therefore selected this time point for detailed comparison of glucose metabolic reprogramming in both sexes. At 0.5 h, plasma levels of ¹³C₆-glucose (M + 6) also exhibited marked sex-related variation : in females, the tracer concentration peaked at 16 months and declined significantly by 23 months, whereas in males, it increased progressively with age (Fig. S3e). This indicates that age-related changes in whole-body glucose disposal or distribution differ substantially between sexes²⁷.

¹³C₆-glucose is metabolized into pyruvate (M + 3) through glycolysis, which can then be converted into lactate (M + 3) via lactate dehydrogenase (LDH) activity (Fig. 2a). We observed significant elevations in labeled pyruvate and lactate in the hippocampus and cortex of female mice from juveniles to the pre-elderly stage, with a notable decline in the elderly almost to similar levels as juveniles (Fig. 2b, c). Similarly, the production of labeled serine from 3-phosphoglyceric acid (3-PG) and alanine from pyruvate in female mice rose significantly from Juvenile to adult and decreased markedly in the hippocampus and cortex of elderly females (Fig. 2d).

In addition to glycolysis, glucose may be metabolized through the PPP, forming lactate (M + 2). As with lactate (M + 3) from glycolysis, we found that lactate (M + 2) from the PPP continuously increased from juveniles until pre-elderly but dropped at the elderly stage in the hippocampus and cortex of females (Fig. 2e). Unexpectedly and unlike females, all of the labeled metabolites of glycolysis and PPP were relatively higher in males than in females at juvenile stages and continued up without dropping at the elderly stage in males (Fig. 2b–e). Moreover, we compared the ratio of lactate (M + 3) produced from glycolysis to lactate (M + 2) from the PPP to define the metabolic reprogramming of glucose between glycolysis and PPP. We found that the ratio of glycolysis to PPP was relatively higher in males than females at most time points across the lifespan (Fig. 2f). Particularly noticeable was the significant drop in elderly females (23 M) in both the hippocampus and cortex. These results indicate an increased flux through PPP in females at the 1 month and 23 month time points in both hippocampus and cortex.

Besides generating NADPH to combat oxidative stress, the PPP also generates cytoplasmic PRPP for the synthesis of nucleotides. Consistent with this view, we found that AMP (M + 5) levels were significantly and relatively higher in the hippocampus and cortex at these two stages (1 month and 23 months) in females than in males (Fig. 2g), supporting the view that female glucose metabolism is wired toward a relatively higher PPP over glycolysis than males in juveniles and the elderly stage, allowing for rapid synthesis of AMP. Thus, we provide direct in vivo evidence that females have a lower glucose flux to glycolysis and increased flux toward PPP than males at the juvenile stage. Females gradually increase their glucose flux to achieve a similar rate of glycolysis and PPP as males from adult to pre-elderly stages, but eventually show a decreased glucose flux capacity and shift to PPP (Fig. 2f), accompanied with increased AMP (Fig. 2g).

To further investigate how glucose is fully oxidized toward TCA in vivo across the lifetime, we examined the flux changes of ¹³C₆-glucose into TCA cycle intermediates (Fig. 2h). Pyruvate (M + 3), aside from being converted into lactate (M + 3) via anaerobic oxidation, is also transformed into acetyl-CoA by the action of pyruvate dehydrogenase (PDH), subsequently generating citrate (M + 2). Male and female mice exhibit distinct patterns of glucose oxidation toward the TCA cycle. In the cortical and hippocampal regions of female mice, the production of glucose-derived TCA cycle metabolites (citrate, succinate, fumarate, and malate) was notably lower than in males at the juvenile stage, then increased to reach similar levels from the adult to pre-elderly stage, but eventually declined at the elderly stage (Fig. 2h). Conversely, in male mice, the generation of glucose-derived TCA cycle metabolites continuously increased from juvenile to the elderly. Additionally, aspartate, glutamate and glutamine, three amino acids that are closely linked to the TCA cycle, mirror the trend observed with the direct TCA cycle products (Fig. 2i). Furthermore, the glucose metabolic reprogramming from juvenile to elderly in both the cortex and hippocampus is similar (Fig. 2b–i). Altogether, we delineated sex-related variation in glucose metabolism and distinct glucose metabolic reprogramming from juvenile to elderly: (i) glucose is rapidly metabolized toward glycolysis, PPP and TCA cycle; (ii) the rate of glucose metabolism by glycolysis is lower in females than males at the juvenile stage, then increases to reach similar levels in both sexes, but eventually declines in elderly females but not in males (Fig. 2j).

Sex and age-dependent comprehensive metabolic landscapes of the cortex and hippocampus

Following our focused analysis on glucose metabolism we utilized untargeted metabolomics to assess metabolic changes at five distinct life stages. To ensure the stability and reliability of the metabolomics data a rigorous quality control (QC) procedure was implemented (see “Methods” for details). This QC strategy allowed us to detect and correct for any potential instrumental drift or batch effects during the long sequence of data acquisition. The PCA plots of the samples versus the QCs showed a high degree of overlap of the QC samples, indicating the high quality and consistency of the data acquisition (Fig. S4b and Table S2). Our study generated a comprehensive, dataset analyzed at three major dimensions: (1) development and age-related changes, (2) sex differences, and (3) brain region specificity (Fig. S4a).

The untargeted metabolomic profiling of the cortical and hippocampal brain regions identified 306 metabolites, categorized into eight chemical classes based on their respective quantities and percentages (Fig. S4c). Pathway enrichment analyses revealed the top 25 metabolic pathways affected (Fig. S4d). The results show that AA pathways as well as purine metabolism are among most affected pathways. These are considered in more detail in later sections.

Gender-based comparisons across all samples revealed significant metabolic differences as illustrated in the PLS-DA plot (Fig. S4e). The volcano plot also revealed over two hundred metabolites with sex-specific differences (Fig. S4f). Notably, these gender disparities were most pronounced in adolescent mice (Fig. S4g), diminishing through adulthood and pre-elderly stages, with elderly mice showing the second highest level of sex differences following juvenility. These findings underscore a marked sex-related variation in brain metabolism, which exhibits age-dependent variations.

We also examined the impact of sex and age on different brain regions. The PLS-DA plots revealed that aging distinctly influenced the metabolome of each brain region, exhibiting unique metabolic signatures across lifetimes (Fig. S5a). Juvenile and elderly mice displayed metabolomic profiles that were distinguishable from other age stages, while adult, middle-aged, and pre-elderly stages demonstrated similar metabolic characteristics. Quantifying the number of age-induced differential metabolites, we observed the most significant changes in the hippocampus and cerebral cortex from juvenile to adulthood and from juvenile to elderly (Fig. S5b). Juvenile female mice exhibited alterations in 221 and 207 metabolites in the hippocampus and cortex, respectively, compared to 116 and 69 in males, indicating a greater impact in juvenile females than males. Additionally, there were regional differences in age effects with female hippocampi showing greater metabolomic alterations at each age stage compared to the cortex, whereas in males, the cerebral cortex exhibited more pronounced changes during aging (Fig. S5c). Consistently, subsequent analyses by metabolite category revealed extensive metabolic remodeling, especially in AA and purine metabolism, in the hippocampus and cerebral cortex from juveniles to adults and from juveniles to old age (Fig. S5d). Altogether, we revealed that the impact of age on hippocampus and cortex metabolomics exhibits both sex and regional specificity.

Age and sex-dependent snapshots of amino acid changes

Our untargeted metabolomics revealed significant age-related alterations in AAs in both brain regions. First, the differential concentration distribution of AAs in the brain showed alanine and glutamate exhibiting the highest levels, followed by aspartate and glycine (Figs. 3a and S6d; Table S3). Alanine, a direct derivative of pyruvate, serves as a crucial energy metabolism molecule, while glutamate, aspartate, and glycine are key neurotransmitters in the cortex and hippocampus. Second, most AAs demonstrated significantly higher levels at the adolescence and elderly stages in females than males (Figs. 3a and S6a–c). The levels of glycine, methionine, phenylalanine, tyrosine, lysine, aspartate, glutamate, glutamine and arginine were much higher in juveniles than in the adult stage in females but not in males. However, histidine, proline and isoleucine/leucine were higher in both female and male mice at this stage (Fig. S6c). Conversely, only alanine was distinctly higher in juvenile male mice. With age progression, especially from adulthood to middle age, no significant AA fluctuations were observed in either sex. However, old age precipitated a notable rise in several AAs such as methionine, phenylalanine, proline, tyrosine, and lysine in both sexes, while glycine, aspartate and arginine were only elevated in elderly female mice. Hence, our study elucidated that both juvenile and elderly mice exhibit elevated AA levels with significant sex differences during these two phases in the hippocampus and cortex. The heightened AA levels during the juvenile period reflect a state related to enhanced protein synthesis, neurotransmitter function or precursors like aspartate and glycine, and the energy required for development to sustain overall cellular activity. In contrast, aberrant accumulations in old age could represent a decompensatory state related to increased protein degradation, less protein and neurotransmitter synthesis, enhanced oxidative stress and reduced cognitive function and memory (Fig. S6e).

Fig. 3. Sex-related variation in aspartate and glutamate as well as related neurotransmitter metabolism during development and aging.

a Alterations in quantitative glutamate, aspartate, and glycine levels in the developing and aging hippocampus and cortex. b, c Alterations in neurotransmitter (GABA, glutamine, NAAG, NAA) levels in the developing and aging hippocampus and cortex. d Changes of NAA/Asp, NAAG/NAA, NAAG/Glu, and Gln/Glu ratio. e Summary chart of neurotransmitter metabolism in the cortex and hippocampus. This figure was created using BioRender (https://BioRender.com/cua2ou1). Data represented as mean ± SEM. The experimental cohorts comprised five age groups of biologically independent animals. For hippocampal analyses: 1-month-old mice (n = 17; 9 female, 8 male), 2-month-old mice (n = 18; 9 female, 9 male), 10-month-old mice (n = 18; 9 female, 9 male), 16-month-old mice (n = 18; 9 female, 9 male), and 23-month-old mice (n = 18; 9 female, 9 male). For cortical analyses: 1-month-old mice (n = 18; 9 female, 9 male), 2-month-old mice (n = 17; 8 female, 9 male), 10-month-old mice (n = 18; 9 female, 9 male), 16-month-old mice (n = 18; 9 female, 9 male), and 23-month-old mice (n = 18; 9 female, 9 male). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were analyzed by two-way ANOVA with Tukey’s multiple comparisons test. NAA N-acetylaspartate, NAAG N-acetylaspartylglutamate, GABA γ- aminobutyric acid, Gln Glutamine, Glu Glutamate.

Sex-related variation in aspartate and glutamate as well as related neurotransmitter metabolism during development and aging

Among all of the quantified AAs, we observed an obvious sex-related variation in aspartate, glutamate and glutamine (Fig. 3a, b). This prompted us to further investigate these AAs and their related neurotransmitter synthesis pathways. During the transition from juvenile to adulthood, we observed a significant decline in aspartate and gamma amino butyric acid (GABA) levels within the hippocampus and cortex, while glutamate levels substantially increased only in female mice (Fig. 3b). Concurrently, the N-acetylaspartate (NAA) to Asp ratio was increased in females, suggesting enhanced enzymatic activity catalyzing Asp to NAA during developmental stages (Fig. 3d). Conversely, the N-acetylaspartylglutamate (NAAG) to Glu ratio (hippo, cortex), Gln/Glu (hippo, cortex), and NAAG/NAA (cortex) ratios were decreased (Fig. 3d). NAA is pivotal for myelinogenesis. Thus, during development, the female has a sophisticated metabolic reprogramming to coordinately reach the increased demand for NAA in neuronal energy supply and myelin synthesis support.

With aging, in female mice, there was a notable decline in NAAG and NAA levels, coupled with a significant increase in aspartate (Fig. 3a–c). Moreover, the reduction in the NAA/Asp ratio in aged female mice signified a constrained NAA biosynthesis pathway, concurrently leading to reduced NAAG levels. Both NAA and NAAG are broadly acknowledged for their neuroprotective roles, and their downregulation could promote neural dysfunction, while excess aspartate might exert toxic effects through N-Methyl-D-aspartic acid (NMDA) receptor activation (Fig. 3e). Moreover, the diminished NAAG/NAA and NAAG/Glu ratios in the aged male mouse cortex suggest an impediment in NAAG synthesis in the aged cerebral cortex. Hence, we identified the downregulation of the neuropeptides NAA and NAAG alongside an abnormal rise in aspartate levels with aging, more predominantly in females than males, highlighting the intricate dynamics of these metabolites in age-associated neurological alterations and more profound changes in females than males.

Histidine metabolic reprogramming from development to aging

We observed significant alterations in histidine levels during the developmental and aging phases, prompting a detailed examination of the histidine metabolic pathway with age (Fig. 4a). Notably, histidine levels exhibited a higher level in both regions in females than males but with a similar pattern in both sexes, decreasing initially with significant downregulation during adulthood compared to juveniles. However, its levels significantly rose again in old age (Fig. 4a).

Fig. 4. Histidine metabolic reprogramming from development to aging.

a Alterations in quantitative histidine levels in the developing and aging hippocampus and cortex. b–d Changes of histidine metabolism pathway metabolites (carnosine, homocarnosine, methylhistidine, methylhistamine). e Schematic diagram of histidine metabolism pathway. This figure was created using BioRender (https://BioRender.com/zclfw3x). Data represented as mean ± SEM. The experimental cohorts comprised five age groups of biologically independent animals. For hippocampal analyses: 1-month-old mice (n = 17; 9 female, 8 male), 2-month-old mice (n = 18; 9 female, 9 male), 10-month-old mice (n = 18; 9 female, 9 male), 16-month-old mice (n = 18; 9 female, 9 male), and 23-month-old mice (n = 18; 9 female, 9 male). For cortical analyses: 1-month-old mice (n = 18; 9 female, 9 male), 2-month-old mice (n = 17; 8 female, 9 male), 10-month-old mice (n = 18; 9 female, 9 male), 16-month-old mice (n = 18; 9 female, 9 male), and 23-month-old mice (n = 18; 9 female, 9 male). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were analyzed by one-way ANOVA with Tukey’s multiple comparisons test.

Histidine catabolism yields carnosine and homocarnosine, all metabolites possessing antioxidative properties. Our findings indicated a significant decline in homocarnosine and carnosine levels during the developmental phase in both female and male mouse hippocampi and cortex, implicating less need to generate those anti-oxidants during development (Fig. 4b, c). In contrast, from adulthood to pre-elderly age, carnosine and homocarnosine levels significantly increased, potentially reflecting early aging compensatory mechanisms in both females and males to counteract increased oxidative stress. In the elderly stage, carnosine levels showed a trend of reduction in these two regions in both females and males, suggesting a potential decrease in antioxidative activity and transition into a decompensatory phase. In contrast, methylhistidine levels were notably elevated in male mice during middle and old age (Fig. 4b, c). Additionally, an increase in methylhistamine, a histidine metabolite, was observed in the hippocampi of aged female mice (Fig. 4d), possibly indicating an enhanced histidine-to-histamine pathway, with reduced carnosine synthesis. In summary, histidine and its metabolites show significant changes in concentration with age in the hippocampus and cerebral cortex (Fig. 4e).

Reprogramming of arginine metabolism during development and aging

We observed significant alterations in arginine levels during developmental and aging phases, particularly within the hippocampus and cortex of female mice (Fig. 5a). Consequently, we investigated changes in metabolites related to the urea cycle pathway. The urea cycle serves as a primary metabolic route for converting highly toxic ammonia into less toxic urea, thereby reducing the accumulation of toxic ammonia that contributes to neurodegenerative changes. Our analyses revealed that, in the juvenile period, metabolites associated with the urea cycle—arginine, ornithine, citrulline, and polyamines derived from ornithine (putrescine, spermidine, and spermine) were significantly higher in female mice, indicating heightened metabolic activity of the urea cycle in adolescent female mice (Fig. 5b, c). From adulthood to the pre-elderly, levels of citrulline significantly declined in the hippocampus but not the cortex in female mice, while putrescine levels notably increased. In contrast, levels of citrulline and urea significantly rose in male mice (Fig. 5b, c). In the aging phase, a marked increase in arginine levels was observed in the cortex and hippocampus of female mice, with a significant downregulation of ornithine, suggesting a disruption in the urea cycle due to potentially decreased activity of arginase 1 (ARG1). Simultaneously, an increase in the downstream product, putrescine, was noted. Citrulline levels significantly rose in the hippocampus of both male and female mice, while levels of argininosuccinic acid significantly decreased in the cortex and hippocampus of aged male mice, indicating a potential downregulation of argininosuccinate synthase 1 (ASS1) in aged male mice. Overall, we identified a high metabolic activity of the urea cycle in both male and female mice during adolescence. However, with advancing age, a disruption in the urea cycle was observed in both sexes during old age, leading to aberrant ammonia detoxification in the brain (Fig. 5d).

Fig. 5. Reprogramming of arginine metabolism during development and aging.

a Alterations in quantitative arginine levels in the developing and aging hippocampus and cortex. b, c Changes of urea cycle metabolites (ornithine, citrulline, argininosuccinic acid, urea, putrescine, spermidine, spermine). d Schematic diagram of urea cycle. This figure was created using BioRender (https://BioRender.com/5xasanm). Data represented as mean ± SEM. The experimental cohorts comprised five age groups of biologically independent animals. For hippocampal analyses: 1-month-old mice (n = 17; 9 female, 8 male), 2-month-old mice (n = 18; 9 female, 9 male), 10-month-old mice (n = 18; 9 female, 9 male), 16-month-old mice (n = 18; 9 female, 9 male), and 23-month-old mice (n = 18; 9 female, 9 male). For cortical analyses: 1-month-old mice (n = 18; 9 female, 9 male), 2-month-old mice (n = 17; 8 female, 9 male), 10-month-old mice (n = 18; 9 female, 9 male), 16-month-old mice (n = 18; 9 female, 9 male), and 23-month-old mice (n = 18; 9 female, 9 male). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were analyzed by one-way ANOVA with Tukey’s multiple comparisons test.

Sex-related variation in purine metabolic reprogramming as a function of age

Purine metabolism changed significantly with regard to age, brain region and sex. Adenosine is not only the major component of the purinergic signaling cascade but also a key neuromodulator in the CNS regulating neuronal metabolism, function, and survival²⁸. Thus, we quantified adenosine concentrations, followed by integrating these data with untargeted purine metabolomic pathways (Fig. 6a). First, accurate quantification of adenosine levels revealed the major sexual difference in adenosine changes that appeared mainly at the juvenile and the elderly stages in both brain regions (Fig. 6b). Specifically, adenosine levels were increased at the adult stage, maintained a similar level from adult to pre-elderly stages, then increased at the elderly stage, while its level was significantly lower in females than males in both regions at the juvenile and elderly stages (Fig. 6b). These results indicate that adenosine, as an important neuromodulator, displays a sex-related variation at the juvenile and the elderly stages but such a difference is diminished from adult to pre-elderly stage (Fig. 6b).

Fig. 6. Sex-related variation in purine metabolic reprogramming as a function of age.

a Schematic diagram of purine metabolism. This figure was created using BioRender (https://BioRender.com/20sonc1). b, c Sex-related variation in purine metabolism (adenosine, inosine) during development and aging in the cortex and hippocampus. d mRNA expression levels of purine catabolic enzymes adenosine deaminase (ADA), purine nucleoside phosphorylase (PNP), and xanthine dehydrogenase (XDH) in 23-month-old male and female mice. Data represented as mean ± SEM. N = 4 in each group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. P value was assessed by unpaired two-tailed t-test. e, f Sex-related variation in purine metabolism (ADP, AMP, IMP, GMP, GDP, guanosine, hypoxanthine, xanthine) during development and aging in the cortex and hippocampus. g Schematic diagram of sex-related variation in purine metabolism in the hippocampus. This figure was created using BioRender (https://BioRender.com/mj2px3t). Data represented as mean ± SEM. The experimental cohorts comprised five age groups of biologically independent animals. For hippocampal analyses: 1-month-old mice (n = 17; 9 female, 8 male), 2-month-old mice (n = 18; 9 female, 9 male), 10-month-old mice (n = 18; 9 female, 9 male), 16-month-old mice (n = 18; 9 female, 9 male), and 23-month-old mice (n = 18; 9 female, 9 male). For cortical analyses: 1-month-old mice (n = 18; 9 female, 9 male), 2-month-old mice (n = 17; 8 female, 9 male), 10-month-old mice (n = 18; 9 female, 9 male), 16-month-old mice (n = 18; 9 female, 9 male), and 23-month-old mice (n = 18; 9 female, 9 male). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were analyzed by two-way ANOVA with Tukey’s multiple comparisons test. ADP adenosine diphosphate, AMP adenosine monophosphate, IMP inosine monphosphate, GMP guanosine monophosphate, GDP guanosine diphosphate.

Adenosine not only acts as an important neurotransmitter modulator but also functions as a key purine metabolism component. Thus, we further examined changes in adenosine metabolism using untargeted purine metabolomic profiling. We found that during development, only females displayed purine metabolic reprogramming toward purine synthesis featured with elevated adenosine-AMP-ADP and inosine-hypoxanthine-xanthine-GMP-GDP (Fig. 6e, f). Similar to adenosine levels, purine metabolic intermediates reached similar levels in females and males in the adult, middle-age, and pre-elderly, indicating no sex-related variation during these stages. However, in the elderly stage, sex-related variation appeared again in some cases. Notably, in the female elderly stage, adenine nucleotide metabolism was programed for degradation associated with increased adenosine-inosine-hypoxanthine-xanthine and decreased AMP, ADP, and GDP in both brain regions. Therefore, we examined the mRNA expression levels of three critical enzymes in the purine metabolic pathway, adenosine deaminase (ADA), purine nucleoside phosphorylase (PNP), and xanthine dehydrogenase (XDH). The mRNA expression levels of ADA, PNP, and XDH were significantly higher in both the cerebral cortex and hippocampus of aged female mice compared to age-matched males. The observed accumulation of metabolites in the purine degradation pathway in aged female mice may be attributed to this sex-related elevation in the expression of these catabolic enzymes (Fig. 6d). In contrast, in the male elderly stage, purine metabolism was switched from degradation to synthesis characterized by elevated adenosine, AMP, GMP, and GDP in both regions (Fig. 6f). Altogether, our studies led to three conclusions regarding the unique sex-related variation in adenosine and purine metabolic reprogramming during development and aging: (1) Adenosine, as the key CNS neuromodulator, is higher in males than females in the hippocampus and cortex in juveniles and the elderly but this sex-related variation is diminished from adult to pre-elderly stages. (2) In females, purine metabolism is reprogrammed toward purine synthesis during development, while during aging, purine metabolism is switched to decompensatory degradation; (3) However, in males, significant purine metabolic reprogramming is not observed during development, but purine synthesis persisted with aging (Fig. 6g).

Spatiotemporal patterns of cognitive sex-related variation revealed by integrated cortical and hippocampal metabolomic profiling

This study reveals significant sex-related variation and its associated metabolic reprogramming mechanisms during cognitive aging in mice. Assessment by the Barnes maze test revealed that although mice of both sexes showed cognitive decline at 16 months of age, female mice exhibited more severe cognitive impairment by 23 months of age. Further metabolomic analyses of the hippocampus and cortex of mice at five different ages revealed a clear age- and sex-dependent metabolic profile of these brain regions. Our metabolomic analyses showed that while both young and aging male and female mice maintain efficient glycolytic capacity to produce pyruvate and lactate and participate in both PPP and TCA cycling pathways, glucose metabolism kinetics exhibited significant age and sex differences. Most importantly, our data indicate that female mice develop metabolic dysregulation earlier in aging, manifested by dysregulation of the coordination of four key metabolic axes—(i) glucose homeostasis, (ii) amino acid metabolism, (iii) purine metabolism, and (iv) antioxidant defenses—suggesting that there is a fundamental sex difference in metabolic resilience during brain aging. The sex-specific reprogramming of these metabolic networks, moreover, explains from a metabolic perspective why female aged mice exhibit poorer cognitive function than males, and the metabolic perspective elucidates the underlying mechanisms of sex differences in cognitive aging.

Discussion

Sex-related variation characterizes the trajectory of cognitive aging, yet the underlying mechanisms predisposing females to a higher incidence of age-related neurodegenerative cognitive impairment remain poorly understood. While mouse models have revealed that cognitive decline emerges around 16 months in both sexes, females exhibit more pronounced deficits by 23 months—a divergence linked to disturbances in cerebral glucose metabolism and enhanced purine degradation in females, alongside preserved metabolic compensation in males. Critically, this cognitive divergence cannot be attributed to general locomotor decline, as longitudinal assessment of movement speed reveals a dissociation between motor and cognitive aging; for instance, while male mice exhibit a significant motor decline by 16 months that stabilizes thereafter, their cognitive function shows a progressive deterioration, and females develop severe cognitive deficits by 23 months without a commensurate worsening of motor speed^29,30. Thus, even though elderly male mice have reduced motor speed relative to females, they nevertheless are able to find the escape box faster than females. Sex-specific drivers of cognitive aging remain understudied in human cohorts, despite clinical epidemiology indicating a greater female susceptibility to conditions such as Alzheimer’s disease. Together, these observations suggest that sex-related variation metabolic alterations in the hippocampus and cortex, rather than peripheral motor impediments, may contribute significantly to disparities in cognitive aging.

Our large-scale in vivo murine isotopic labeled glucose tracing analyses led to new and compelling discoveries of sex-related variation in glucose metabolism and distinct glucose metabolic reprogramming from juvenile to elderly mice in the hippocampus and cerebral cortex. The rate of glucose metabolism was significantly lower in females than males at the juvenile stage, then increased to reach similar levels in both sexes during adulthood, middle age, and pre-elderly. However, in the elderly group glucose metabolism declined in females, whereas it continued to increase in males as a compensatory response to counteract aging. Then, integrating comprehensive metabolomics with the accurate AAs quantification led us to discover substantial sex-related variation in AA metabolism, including the synthesis of neurotransmitters, antioxidants and urea cycle metabolites across lifespan. Intriguingly, the untargeted metabolomic profiling and accurate adenosine quantification further revealed significant sex-related variation. Specifically, adenosine (a key CNS neuromodulator) is higher in males than females in the hippocampus and cortex in juveniles and the elderly, a sex-related variation that was not observed during mid-life from adult to pre-elderly stages. Purine metabolism is reprogrammed toward purine synthesis during development and is switched to decompensatory degradation in the elderly only in females. However, in males purine synthesis persisted with aging. Overall, we identified substantial sex-related variation affecting the metabolism of glucose, purines and AAs, compounds which are essential for energy metabolism and the synthesis of neurotransmitters, neuromodulators, antioxidants and detoxification in CNS. Our findings provide a metabolic basis for the long-standing observation that females are more susceptible to age-related cognitive decline. The male brain’s sustained glucose utilization and purine synthesis may confer resilience against neurodegeneration, whereas females’ metabolic decompensation-particularly in hippocampal energy and purine salvage pathways-could predispose them to dysfunction. Our work underscores the necessity of sex-specific approaches in understanding brain aging and designing interventions for neurodegenerative diseases.

It is well acknowledged that glucose metabolism plays a pivotal role in cognitive function³¹, and that disruption of glucose metabolism induces cognitive impairment and progression to neurodegenerative diseases such as Alzheimer’s³¹ and Parkinson’s disease^32–34. Glucose may be metabolized through glycolysis to generate energy or through the PPP to produce PRPP for use in nucleotide biosynthesis or to generate NADPH which is needed to produce GSH to combat oxidative stress. In the CNS oxidative stress is believed to significantly contribute to the aging-related cognitive decline and progression of AD³⁵. Thus, it is critical to define the in vivo glucose metabolic utilization and reprogramming in the CNS across the lifetime, a daunting experimental challenge. Here we overcame this difficulty and took advantage of using a well-accepted murine experimental model across lifetimes and conducted large-scale in vivo isotopically labeled glucose tracing analyses to elucidate the variances in glucose utilization across different ages and sexes in the hippocampus and cortex, two key regions for cognition and memory. We isolated brain tissues at three different time points following ¹³C₆-glucose injection and discovered rapid in vivo glucose metabolism dynamics in the CNS with a near completion by 2.5 h. We found that the 0.5 h time point was the most reliable in allowing us to determine the fate of the injected ¹³C₆-glucose. Thus, tracing metabolites from labeled glucose at 0.5 h provided us a time window to get a detailed and comprehensive view of glucose metabolic fate in response to metabolic reprogramming. We delineated sex-related variation in glucose dynamics and metabolic reprogramming from development to aging. We were able to trace glucose metabolism toward glycolysis, PPP and the TCA cycle across the lifetimes. We found that females are featured with a lower glucose flux rate at the juvenile stage and a decline in glucose flux at the elderly stage, features distinctly different from what we observed in males. Hence, our discovery reveals a sex-related variation in glucose metabolism dynamics and reprogramming that provides a potential metabolic mechanism underlying the puzzle of the higher prevalence of age-related cognitive dysfunction and progression to AD in women and suggests potential preventive and therapeutic strategies to mitigate such neurodegeneration.

AAs are basic components for protein synthesis, neurotransmitter synthesis and metabolic pathways. However, the precise quantification of AA metabolism in the CNS during development and aging is lacking. Interestingly and unlike glucose flux and purine metabolism, both adolescent and elderly mice exhibit elevated AA levels with significant sexual differences characterized by higher amounts of AAs in females than males in the hippocampus and cortex at both phases. The heightened AA levels in juvenile females reflects a state of enhanced protein and neurotransmitter synthesis and overall cellular activity during development and provides a potential metabolic basis for earlier CNS development in females than males during the juvenile period^36,37. In contrast, increased protein degradation, less protein and neurotransmitter synthesis, and declined cognitive function and memory, implicating the higher prevalence of aging-related cognitive decline in females than males.

Supporting this speculation, we observed an obvious sex-related variation in aspartate, glutamate and glutamine and their derivatives of neurotransmitters including NAA and NAAG. These amino acid derivatives are synthesized in neurons, present at high concentrations within the CNS, serve as neuronal markers and are considered as indicators of neuronal health and function^38,39. We discovered that during development, female metabolic reprogramming coordinatively reaches the increased demand for NAA in neuronal energy supply and myelin synthesis by metabolizing aspartate. However, with aging, the downregulation of the neuroprotectants NAA and NAAG alongside an abnormal rise in aspartate levels was more predominant in females than males. NAA is crucial for myelinogenesis, and its diminished levels in females may reflect a reduction in neuronal quantity or function, frequently correlating with cognitive decline. NAAG modulates glutamatergic transmission and may play roles in neuroprotection⁴⁰ and synaptic plasticity⁴¹. These metabolic imbalances are often associated with various neurological diseases. As the most abundant neuropeptide in the CNS, reduced NAAG levels observed in females may lead to decreased stimulation of metabotropic glutamate receptor 3 (mGluR3), associated with cognitive impairments^42,43. In the adult mammalian brain, glutamate and GABA function as the primary excitatory and inhibitory neurotransmitters, respectively⁴⁴. The elevated levels of aspartate in the hippocampus and cerebral cortex of aged female mice could potentially result in excitotoxicity due to excessive NMDA receptor activation. Thus, our studies highlight the intricate dynamics of AA metabolism in age-associated neurological alterations and more profound changes in females than males.

Notably, histidine, a semi-essential amino acid, epitomizes the intricate metabolic alterations in the brain throughout development and aging. Histidine catabolism yields carnosine and homocarnosine, all metabolites possessing antioxidative properties. Our investigation into histidine metabolism unveiled fluctuating patterns that mirror the brain’s physiological transitions. From adolescence to adulthood, histidine levels initially decline, followed by a significant resurgence in old age. In contrast, from adulthood to pre-elderly age, carnosine and homocarnosine levels significantly increased, potentially reflecting early aging compensatory mechanisms in both females and males to counteract increased oxidative stress.This pattern, observable in both sexes and key brain regions such as the cerebral cortex and hippocampus, underscores the pervasive impact of aging on AA metabolism, suggesting a protective role against oxidative stress, a hallmark of aging⁴⁵. Our findings indicate subtle variations in the trajectories of these metabolites: a significant reduction during development, a transient increase in early aging as a potential compensatory response, and an eventual decrease in old age, suggesting the depletion of antioxidative defense mechanisms.

Mounting evidence implicates arginine and polyamine metabolism in aging and neurodegenerative processes⁴⁶. L-arginine, a semi-essential amino acid, plays a pivotal role as a precursor to numerous biologically active metabolites. Both arginine and ornithine, key metabolites in the urea cycle, are crucial for converting highly toxic ammonia into less toxic urea, a primary detoxification pathway. During adolescence, mice exhibit heightened activity in the urea cycle, leading to the generation of abundant polyamine derivatives, including putrescine, spermidine, and spermine. These molecules play pivotal roles in cellular proliferation and differentiation, DNA, RNA, and protein synthesis, protein phosphorylation, signal transduction, and the regulation of neurotransmitter receptors⁴⁷. In AD patients’ brains, elevated levels of toxic ammonia and urea have been detected⁴⁸. In aged female mice, a significant increase in hippocampal arginine levels and a marked decrease in ornithine levels were observed, suggesting a disruption in the urea cycle potentially due to reduced activity of ARG1, which mediates the conversion of arginine to urea and ornithine, the latter being the final product of the urea cycle. Inhibition of ARG1 disrupts this detoxification process, limiting the clearance of toxic ammonia, potentially leading to the activation of inducible nitric oxide synthase (NOS), production of peroxynitrite, and further nitrosylation and neurodegenerative changes in adjacent neurons. Similarly, an increase in putrescine production may indicate altered activity of ornithine decarboxylase 1 (ODC1), with studies suggesting that inhibition of ODC1 may prevent AD-related symptoms⁴⁹. However, in aged male mice, an increase in hippocampal citrulline levels and a significant decrease in argininosuccinate suggest a reduction in the activity of argininosuccinate synthase 1 (ASS1). Overall, our findings reveal a high metabolic activity of the urea cycle in both male and female mice during adolescence; With aging, a disruption of the urea cycle occurs in both genders, leading to abnormal ammonia detoxification in CNS, thereby impairing cognitive function during aging.

Throughout brain development, processes such as cell division, growth, migration, and differentiation require sufficient energy and cell building blocks. Purine nucleotides play a pivotal role in these activities. For example, adenosine, a key neurotransmitter modulator and a major component of purinergic signaling, functions via specific receptors in CNS to modulate neuronal plasticity, learning, memory, motor functions, and various physiological activities⁵⁰. Disruptions in purinergic signaling can impair neural development, resulting in a spectrum of adverse long-term outcomes ranging from mild cognitive deficits to severe neurological and psychiatric disorders⁵¹. However, the changes in adenosine and comprehensive purine metabolic reprogramming in the hippocampus and cortex from development to aging in both sexes has not been previously examined. Our analysis revealed that adenosine levels are higher in males than females in the hippocampus and cortex in juveniles and the elderly. However, females exhibit a significant upsurge during the transition from juveniles to adults to reach similar levels as the males during much of adulthood, until the elderly stage. Surprisingly, in females, purine metabolism is reprogrammed toward purine biosynthesis during development, while it is switched to purine degradation in the elderly stage. However, no significant purine metabolic reprogramming is observed during development in males, with purine synthesis persisting with aging. Thus, we revealed a previously unrecognized sex-related variation in adenosine levels and purine metabolic reprogramming most obviously at the juvenile and the elderly stages in both the hippocampus and cortex brain regions. These results suggest a role for purinergic signaling and metabolic reprogramming in modulating CNS energy, cognitive function and memory from development to aging, and highlight an important sex-related variation in purine metabolism in the hippocampus and cortex.

Prior studies have reported region-specific alterations in neurotransmitter metabolism in mice with cognitive impairments induced by a diabetes model⁵². However, reports on concurrent sex and region-specific investigations across lifetimes are scarce. Here we not only defined the substantial sex-related variation featured with more profound metabolic changes in females during development as well as the elderly stage but also revealed that the impact of both development and aging on the metabolic changes in the hippocampus is more pronounced in female mice, whereas male mice exhibit fewer profound changes in cortex during development but more significant changes in the cortex during aging. To a certain extent, the brain’s metabolic response to development and aging is orchestrated through a network of interactions among different brain regions. Thus, sex and region-specific analyses have demonstrated that the male and female mouse hippocampus and cortex exhibit divergent metabolic responses to development and aging, providing a metabolic foundation for sex-related variation in cognitive development and decline from young to old, highlighting the imperative of incorporating sex as a factor in developing preventative and therapeutic strategies for age-related diseases leveraging metabolic intermediates.

In conclusion, our study demonstrates that age-related cognitive decline initiates similarly at 16 months but progresses to female-exacerbated impairment by 23 months, coinciding with sex-specific metabolic reprogramming in the cortex and hippocampus. While both brain regions show parallel metabolic shifts during aging, we identify fundamental sex differences in glucose flux, TCA cycle activity, and neurotransmitter metabolism (glutamate/aspartate), with females exhibiting earlier metabolic decompensation across purine cycling, antioxidant defenses (histidine), and ammonia detoxification (arginine). These findings establish that sex-specific metabolic adaptations-particularly in hippocampal energy homeostasis and cortical neuroprotection-underlie divergent cognitive aging trajectories, fundamentally advancing our understanding of neural aging while highlighting the necessity of sex-based approaches in neurodegeneration research.

Methods

Animals

C57BL/6J mice (No.AM03118) were purchased from Jiangsu Aniphe Biolaboratory Inc and housed in individually ventilated cages under controlled conditions (40–60% humidity, 20–24 °C, 12-h light/dark cycle) with food and water ad libitum. All procedures were approved by the Ethics Review Committee of Xiangya Hospital, Central South University (CSU-2023-0336). The animal protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of Central South University. All procedures strictly followed the welfare and ethical principles of laboratory animals. We have complied with all relevant ethical regulations for animal use.

Barnes maze tests

The Barnes maze is a well-established behavioral test for assessing spatial learning and memory in rodents. A stationary escape pod is set up over an unchanging hole and the mice are allowed to navigate based on cues from negative stimuli (bright lights). On day 1, each mouse was placed in the escape pod for two minutes, then taken out and placed in the center of the platform for 180 s of free exploration, and guided by the researcher to the escape box for 90 s of adaptation in the absence of negative stimuli before being taken out and placed back in the cage. Next, during the training period from the second day to the fourth day, a two-rounds-per-day training process was performed, which was the same as the acclimatization process, but with bright light stimulation. The mice were placed in the center of the training platform and allowed to explore freely, the latency to find the target area was recorded and the search trajectory was recorded during 180 s probe trials to complete the 3-day, six-trial acquisition experiment, and the learning curves were obtained from the latencies of the acquisition trials during the daily training period. The fifth day was the testing phase, which was the same as the training period, in which the mice were placed in the center of the training platform and the latencies of the mice to find the escape pods were recorded to complete the training test of learning memory.

¹³C₆-glucose tracing, tissue collection, and flux data normalization

Mice received tail vein injections of 5% U-¹³C₆-Glucose (5 ml/kg) at 0.5, 1.5, and 2.5 h prior to terminal sample collection. For collection, mice were deeply anesthetized with an intraperitoneal injection of sodium pentobarbital (30 mg/kg) and subsequently perfused transcardially with PBS (pH 7.4) for 10 min. The brain was placed in culture dishes containing PBS at 4 °C for subsequent dissection of brain regions, including the hippocampus and cerebral cortex. All tissues were promptly frozen in liquid nitrogen and stored at −80 °C until further analysis. The normalized ¹³C enrichment (fractional abundance, F) was calculated from raw mass spectral intensity data using the following formula:

$$F_{\mathrm{n}}=I_{\mathrm{n}}/\sum \left(I_{\mathrm{o}}+I_1+I_2+...+I_{\mathrm{m}}\right)$$

where I_n represents the signal intensity of the ¹³C-labeled isotopologue containing n ¹³C atoms (denoted as M + n). The denominator corresponds to the sum of the intensities of all detected isotopologues for that metabolite, ranging from the unlabeled form (M + 0) to the fully labeled form (M + m). Here, m indicates the maximum number of carbon atoms in the metabolite that can be replaced by ¹³C, i.e., the total number of carbon atoms that are theoretically able to be labeled.

Sample preparation

Metabolomic profiling was performed on hippocampal and cerebral cortical tissues from a cohort of C57BL/6J mice across five age groups (1, 2, 10, 16, and 23 months old) with a balanced sex ratio. Tissues were collected from ten mice per group. After allocating one mouse per group to other experimental protocols and the accidental loss of one hippocampal sample from a 1-month-old male mouse and one cortical sample from a 2-month-old female mouse during microcentrifuge tube transfer, the final sample sizes for analysis were as follows: hippocampus (n = 8 for 1-month-old males, n = 9 for all other groups) and cerebral cortex (n = 8 for 2-month-old females, n = 9 for all other groups). 3–10 mg brain tissue was added to pre-chilled Lysis Buffer (Methanol: Acetonitrile: Water, 5:3:2, v/v/v) to a final concentration of 15 mg/ml and then homogenized. After 30 min of vortex at 4 °C, suspensions were then centrifuge at 18,300 × g for 10 min 4 °C. The supernatant containing metabolites was applied for metabolomics. For absolute quantification of adenosine and amino acids, the Lysis Buffer was supplemented with final concentrations of 2 μM isotopic labeled amino acid mixture and 1 μM ¹³C₅-adenosine⁵³.

UHPLC-MS metabolomics and quality control (QC)

Metabolomics analysis was conducted using a ThermoFisher Dionex UltiMate 3000 UHPLC system coupled with a Q Exactive HF MS. A Kinetex C18 column (2.1 × 150 mm, 1.7 μm) was employed with a flow rate of 0.45 ml/min, column oven temperature set at 45 °C, and sample maintained at 4 °C with an injection volume of 10 μl. For positive ion mode liquid chromatography, the mobile phases consisted of solvent A (0.1% formic acid in water) and solvent D (0.1% formic acid in acetonitrile), with a gradient elution as follows: 0–0.5 min, 5% D; 0.5–1.1 min, 5–95% D; 1.1–2.75 min, 95% A; 2.75–3 min, 95–5% A; 3–5 min, 5% A. For negative ion mode liquid chromatography, the mobile phases consisted of solvent B (95% acetonitrile in water with 1 mM ammonium acetate) and solvent C (5% acetonitrile in water with 1 mM ammonium acetate), with a gradient elution as follows: 0–0.5 min, 100% C; 0.5–1.1 min, 100% B; 1.1–2.75 min, 100% B; 2.75–3 min, 100–0% B; 3–5 min, 100% C. Mass spectrometry parameters were set as follows: Resolution 60,000, Scan Range 65–900 m/z, Maximum injection time 200 ms, Automatic gain control (AGC) 3 × 10⁶ions, Sheath gas 45, Auxiliary gas 15, Sweep gas 0. A quality control (QC) protocol was implemented to ensure data stability and reproducibility throughout the metabolomic profiling process. A pooled QC sample was prepared from a previously established human plasma pool collected and archived by the research group. This homogeneous reference material was systematically analyzed at regular intervals throughout the acquisition sequence to monitor instrumental stability and ensure data reproducibility across batches. Specifically, one QC sample was injected after every 10 experimental samples within each analytical batch to monitor instrumental performance and signal stability. This approach facilitated the detection and correction of technical variations, including instrumental drift and batch-related artifacts, ensuring robust and reproducible metabolic phenotyping.

Adenosine and amino acid quantification

Add stable isotope internal standards of quantitative metabolites (adenosine, glutamate, aspartate, glycine, phenylalanine, tyrosine, isoleucine/leucine, valine, alanine, lysine, methionine, proline, serine, threonine, histidine, and arginine) into the extraction solution (methanol:acetonitrile:water = 5:3:2) at −20 °C pre-cooling; place hippocampal and cortical samples on ice to thaw and weigh, then transfer them to 1.5 ml EP tubes; add pre-cooled extraction solution containing stable isotope internal standards; subsequent steps are the same as non-targeted metabolomics. Original data are converted to mzXML format files using RawConverter software, imported into Maven software for peak extraction based on the mass differences of stable isotopes, and peak areas (area) are quantitatively calculated.

Quantitative RT-PCR

Total RNA was extracted from the cortex and hippocampus tissues using TRIzol reagent (Invitrogen). cDNA was synthesized using a SuperScript IV Reverse Transcriptase kit. Quantitative real-time PCR was performed in triplicate using SYBR Green PCR Master Mix (Qiagen) on a LightCycler 480 Instrument II (Roche). Primer sequences were as follows: Ada (forward: 5′-CTATGGACTTGGCTGGGGAT-3′, reverse: 5′-CCATTCTTTACTGCGCCCTC-3′); Pnp (forward: 5′-ATGACCGGGATATGAGGCAG-3′, reverse: 5′-GCCACAGTCTCAAAGTTGGG-3′); Xdh (forward: 5′-CCGCCTTCAGAACAAGATCG-3′, reverse: 5′-CCTTCCACAGTTGTCACAGC-3′).

Data processing and analysis

The raw data were converted to mzXML format files using RawConverter software and then imported into Maven software for peak extraction. Systematic error removal by CordBat method was employed to correct for batch effects or instrument signal drifts⁵⁴. Subsequent analysis was conducted on the MetaboAnalyst 5.0 website (http://www.metaboanalyst.ca). Following normalization, logarithmic transformation, and missing value imputation (if applicable), Student’s t test was performed for each metabolite. Metabolites were considered differentially abundant if they satisfied both *p* <0.05 (statistical significance) and a fold change (FC) greater than or less than 1, corresponding to FC > 1 for significant up-regulation or FC < 1 for significant down-regulation. Pathway enrichment analysis and pathway topology analysis were further performed on the identified differential metabolites, wherein the metabolites were enriched into pathways for analysis.

Statistics and reproducibility

Sample sizes for animal studies were determined based on established standards in the field, with predetermined group sizes of 5–9 for behavioral tests and 8–9 for metabolomic analyses. These numbers were selected to ensure adequate statistical power while adhering to the 3 R (Reduction, Refinement, Replacement) principles. Different cohorts of C57BL/6 mice were used for distinct experimental endpoints: qPCR assays were performed on tissue from three independent biological replicates (i.e., three distinct animals per condition), while behavioral and metabolomic data were each collected from separate groups of mice, representing independent biological replicates per assay. Although behavioral and metabolomic experiments were not repeated across independent experimental batches due to their inherent design (e.g., stress accumulation in behavioral paradigms and the high-throughput nature of metabolomics), reproducibility was ensured through the use of statistically sufficient biological replicate numbers, strictly standardized protocols, and rigorous internal quality controls. Data are presented as mean ± SEM. All statistical analyses were performed using GraphPad Prism version 10. Statistical significance was defined as a two-tailed p < 0.05, and denoted in figures as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant. Specific analyses were conducted as follows: comparisons between sexes within the same age group were assessed by unpaired two-tailed t-test; differences across age groups within the same sex were analyzed by one-way ANOVA followed by Tukey’s post hoc test for multiple comparisons; and the effects of both age and sex, along with their interaction, were evaluated using two-way ANOVA followed by Tukey’s multiple comparisons test.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

X.L. designed and conducted the experiments, collected the data and analyzed the experimental data, drew the figures and wrote the manuscript; C.H.C. took charge of infusing [U-¹³C₆] glucose in vivo and collecting brain tissue samples; Q.G. and Y.D.W. managed the collection and preliminary processing of brain tissues; W.P.L. handled the initial data processing and normalization; F.Y. and Y.J.Z. guided the process of metabolomics data analysis; R.E.K. provided critical feedback on the manuscript and expertise in methods; Y.X. oversaw the design of experiments and interpretation of results, the writing and organization of the manuscript and did final editing.

Peer review

Peer review information

Communications Biology thanks Sreemathi Logan and the other anonymous reviewer(s) for their contribution to the peer review of this work. Primary handling editors: Rosie Bunton-Stasyshyn and Benjamin Bessieres. A peer review file is available.

Data availability

All other data generated in this study are presented in the article and its supplementary information. The numerical source data for all graphs and charts in this study are available in Supplementary Data 1–3. All other data supporting the findings of this study are available from the corresponding author upon reasonable request. Plasma metabolic flux data supporting the findings in this study have been published previously²⁷.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09527-9