SIRT1 mediates brain metabolic and developmental consequences of methionine synthase deficiency in inborn errors of cobalamin metabolism

Summary

Inborn errors of vitamin B12 metabolism (IECM) resulting from impaired methionine synthase (MTR) activity cause severe cognitive and neurological deficits that remain unresponsive to conventional B12 supplementation. Using a brain-specific Mtr knockout mouse model, we identify the NAD⁺-dependent deacetylase SIRT1 as a central regulator of the pathological phenotype and evaluate the therapeutic efficacy of its pharmacological activator SRT2104. MS deficiency induces profound metabolic, mitochondrial, and epigenomic alterations in the hippocampus, including promoter hypermethylation of the pyruvate dehydrogenase complex, impaired tricarboxylic acid (TCA) cycle activity, and reduced SIRT1 expression. At the functional level, we observe disrupted Wnt signaling associated with decreased neurogenesis, increased astrocytosis, and cognitive impairment. SRT2104 treatment restores mitochondrial and energy metabolism, normalizes Wnt signaling and neurogenesis markers, and rescues learning and memory performance. These findings identify SIRT1 as a therapeutic target in B12-related neurodevelopmental disorders and support the clinical repurposing of SRT2104 to alleviate persistent neurological symptoms.

Graphical abstract

Highlights

• •Brain Mtr loss triggers metabolic, mitochondrial, and epigenomic defects
• •MS deficiency impairs Wnt signaling, reduces neurogenesis, and increases astrocytosis
• •SIRT1 identified as a central regulator of hippocampal dysfunction
• •SRT2104 restores metabolic and epigenetic balance and rescues cognition

Matmat et al. show that brain-specific methionine synthase deficiency disrupts hippocampal metabolism, mitochondria, and Wnt signaling. Pharmacological activation of SIRT1 with SRT2104 restores metabolic and epigenetic balance, promoting neurogenesis and rescuing memory performance.

Introduction

The deficiency of vitamin B12 (also named cobalamin) is a global public health concern, disproportionately affecting vulnerable groups such as pregnant women, infants, and the elderly.¹ Inherited errors of cobalamin metabolism (IECM) are among the most frequent causes of inherited metabolic disorders.^1,2 Both vitamin B12 deficiency and IECM manifest in neurological impairments and megaloblastic anemia. The full spectrum of cognitive and neurological disturbances resulting from B12 deficiency and IECM remains poorly understood, despite extensive studies.^{2,3,4,5,6,7,8} Vitamin B12 plays a pivotal role in cellular functioning, primarily through its involvement as a cofactor with two vital enzymes: cytoplasmic methionine synthase (MS), which is encoded by the MTR gene and mitochondrial methylmalonyl-CoA mutase. MS catalyzes the conversion of homocysteine (Hcy) to methionine using methyltetrahydrofolate. Methionine is the precursor of S-adenosylmethionine (SAM), which serves as a universal methyl donor, critically involved in the methylation processes of nucleic acids, proteins, and numerous small molecule metabolites.⁴ This biochemical pathway underscores the extensive impact of vitamin B12 on metabolic and neurological health.⁴

Mutations in the MTR gene cause a rare but severe form of IECM known as homocystinuria megaloblastic anemia, cblG type (OMIM #250940), characterized by elevated Hcy, reduced methionine levels, and a broad spectrum of clinical manifestations.⁵ Neurological symptoms often include cerebral atrophy, cognitive impairment, demyelination, and hydrocephalus in early childhood. In adolescence and adulthood, ataxia, pyramidal signs, peripheral neuropathy, and psychiatric disturbances are more commonly observed.²

Although cblG and cblC arise from distinct genetic and biochemical defects—cblC being caused by impaired activity of MS and methylmalonyl-CoA mutase and cblG by impaired activity of MS—both conditions frequently share overlapping features, such as severe neurological involvement, megaloblastic anemia, and ocular abnormalities.^2,6 These clinical similarities may reflect converging downstream pathogenic mechanisms related to the impaired MS-dependent remethylation pathway.

The severity and reversibility of neurological symptoms in cblG are strongly influenced by the timing of treatment initiation. Current therapeutic approaches rely on high-dose hydroxocobalamin (OH-Cbl) injections to enhance residual MS activity.⁷ When administered presymptomatically, in combination with folinic acid, betaine, and methionine, OH-Cbl has been shown to normalize metabolic parameters and improve neurodevelopmental outcomes.⁸ However, delayed treatment often fails to reverse neurological or ocular damage, underlining the limited efficacy of current strategies in advanced stages^8,9 and the pressing need for novel therapeutic interventions.

Complete knockout (KO) of Mtr in mice is embryonically lethal, preventing direct study of its neurological consequences in vivo. As a result, nutritional models inducing folate and vitamin B12 deficiencies—commonly referred to as methyl donor deficiency (MDD)—have been widely used as alternative approaches. Although MDD models effectively reduce MS activity, they are limited by their inability to isolate the specific effects of B12 deficiency from the broader metabolic consequences of folate depletion. Notably, MDD induces significant epigenomic alterations affecting the expression of key genes involved in neural tube development during embryogenesis.^4,10,11

Vitamin B12 deficiency triggers endoplasmic reticulum (ER) stress and disrupts energy metabolism by inhibiting the PGC1α/PPARα pathway, as demonstrated in patient-derived fibroblasts and in MDD-induced rodent models.^12,13,14 In neuroblastoma cells, altering intracellular vitamin B12 availability via a transcobalamin-oleosin construct enhances initial neurite outgrowth while impairing cell proliferation. These effects are associated with disruptions in Erk1/2 and Akt signaling, decreased SIRT1 expression, and altered nucleocytoplasmic mRNA shuttling. The latter is linked to imbalanced methylation, acetylation, and phosphorylation of RNA-binding proteins including HuR (ELAVL1) in the hippocampus.^15,16 Interestingly, pharmacological activation of SIRT1 reverses the mislocalization of RNA-binding proteins in fibroblasts from patients with the cblG variant, highlighting a potential therapeutic mechanism.¹⁷

The hippocampus, a key brain region involved in learning and memory, continues to exhibit robust neurogenic activity during the postnatal period.^18,19 Offspring of rats exposed to MDD during gestation and lactation display long-lasting behavioral abnormalities related to hippocampal function, even when B12 and folate supplementation is provided after weaning.²⁰ Interestingly, stimulation of postnatal neurogenesis through mild transient hypoxia within the first 24 h after birth has been shown to restore cognitive performance.²¹ These findings suggest that MDD impairs neurogenesis in early life, with lasting consequences on cognitive function. Despite these insights, the brain-specific molecular consequences of impaired MS activity remain poorly understood, and the potential of SIRT1 as a therapeutic target in cblG disorders has yet to be explored.

To address this knowledge gap, we investigated the consequences of a constitutive, brain-specific knockout of Mtr (Mtr-cKO) in mice.^22,23 Using a multi-omics strategy—integrating genomics, proteomics, and metabolomics—combined with molecular, cellular, and behavioral validations, we aimed to comprehensively characterize the alterations associated with impaired MS activity. This integrative approach revealed major disruptions in pathways related to energy metabolism and neurogenesis. The decreased SIRT1 emerged as a central key player in all the changes observed in Mtr-cKO mice, leading us to hypothesize that MS deficiency exerts part of its effects through SIRT1 dysregulation. To assess the central role of decreased SIRT1 in the KO phenotype and evaluate its therapeutic relevance, we tested whether pharmacological activation of SIRT1 using SRT2104—a compound already evaluated in clinical trials—could reverse the molecular, cellular, and behavioral impairments observed in Mtr-cKO mice. These findings support SRT2104 as a promising candidate for the treatment of neurological complications associated with cobalamin metabolism disorders.

Results

Conditional brain-specific deletion of Mtr using Cre-Lox recombination resulted in marked metabolic alterations, including a significant reduction in the hippocampal SAM/SAH ratio

As expected, 21-day-old Mtr-cKO mice (Figure 1A) exhibited a strong molecular phenotype. In silico analysis confirmed the introduction of a premature stop codon (TAA) at position 192 within the Hcy-binding domain, likely producing a truncated, non-functional protein (Figure 1B). Consistently, Mtr expression was significantly decreased in the hippocampus at both the mRNA and protein levels (Figures 1C and 1D). The antibody used for MS protein detection targeted a region located downstream of the deletion site, within the SAM-binding domain, confirming the absence of full-length MS protein in KO mice. Importantly, Cre recombinase expression was restricted to the hippocampus in Mtr-cKO animals, with no detectable signal in wild-type (WT) controls (Figures 1C and 1D). To verify regional specificity, we assessed the expression of both Thy1 and Cre, revealing strong hippocampal and frontal cortex expression (Figure S1A), consistent with the brain-targeted pattern of the Thy1 promoter. Although Thy1 is also known to mark hematopoietic stem cells,²⁴ MS depletion did not appear to significantly affect hematopoiesis, as complete blood counts revealed no difference in red blood cells or platelets (Figure S1B).

Figure 1.

Conditional brain-specific deletion of Mtr induces marked metabolic alterations

(A) Strategy for Mtr excision in brain using the Thy1.2-Cre driver (exons 4–5 deleted). Created with BioRender.com.

(B) Predicted structural consequences of exon loss: intact MS protein vs. truncated form with premature stop at residue 192.

(C and D) Hippocampal expression of Mtr and Cre at mRNA (C) and protein (D) levels; α-tubulin as loading control. Statistical significance was assessed using unpaired two-tailed Student’s t test. Data are presented as mean ± SEM.

(E) Heatmap of hippocampal metabolites from one-carbon and transsulfuration pathways (Z scores per mouse).

(F) Log₂ fold changes in metabolite abundance in Mtr-cKO vs. WT mice, with adjusted p values (Benjamini-Hochberg).

(G) Ratios of methionine/homocysteine (MS activity), cystathionine/homocysteine (CBS activity), and SAM/SAH (cellular methylation potential). Statistical significance was assessed using unpaired two-tailed Student’s t test. Data are presented as mean ± SEM.

(H) Schematic of one-carbon metabolism highlighting key enzymes and metabolites.

Data are presented as mean ± SEM, with individual values shown. Sample sizes correspond to the number of points displayed in each graph. Statistical tests were selected according to data distribution and variance.

We next examined the metabolic consequences of Mtr deletion within the neurogenic hippocampus. Metabolite profiling revealed a significant reduction in methionine and SAM levels in Mtr-cKO mice (Figures 1E and 1F), while S-adenosylhomocysteine (SAH) levels remained unchanged. Accordingly, the SAM/SAH ratio, an established index of transmethylation capacity^25,26 was significantly decreased (Figure 1G). Hcy levels showed a trend toward elevation, consistent with impaired remethylation (Figure 1H). Interestingly, we observed increased levels of cystathionine and cysteine in Mtr-cKO mice (Figure 1F), suggesting a compensatory activation of the transsulfuration pathway. This shift was confirmed by a 2.5-fold increase in cystathionine β-synthase (CBS) protein expression (Figure S1C), reflecting both an accumulation of substrate (Hcy) and upregulation of enzymatic activity. Consequently, the methionine/Hcy ratio—indicative of MS activity—and the SAM/SAH ratio were both significantly reduced in the KO group, whereas the cystathionine/Hcy ratio (reflecting CBS activity) was increased (Figure 1G).

In summary, conditional deletion of Mtr in the brain resulted in profound metabolic remodeling within the hippocampus, characterized by impaired methionine and SAM synthesis, elevated Hcy, and activation of the transsulfuration pathway (Figure 1H).

Epigenomic and proteomic alterations in the hippocampus of Mtr-cKO mice reveal disrupted neurogenesis and energy metabolism

To further explore the consequences of Mtr deletion, we performed genome-wide DNA methylation and proteomic profiling of the hippocampus. While global CpG methylation levels were not significantly affected (Figure S1D), circos plots revealed no large-scale changes across chromosomes or genomic regions (Figure 2A). However, volcano plot analysis identified a set of significantly differentially methylated CpG sites (Δmethylation >25%, q value <0.01), with 54.6% hypermethylated and 45.4% hypomethylated in Mtr-cKO mice (Figure 2B). Gene Ontology (GO) analysis of these CpGs revealed significant enrichment in biological processes related to neurogenesis, nervous system development, and cell differentiation (Figure 2C). Pathway analysis using the Panther Classification System further highlighted the Wnt signaling pathway among the top enriched terms (Figure 2D). Differentially methylated regions (DMRs) were also enriched in regulatory processes such as intracellular signal transduction, response to external stimuli, and defense response (Figures 2E and 2F). Notably, we observed promoter hypermethylation in the Pdha1 gene, encoding the pyruvate dehydrogenase subunit α1 (Figure 2G), as well as in Wnt-related genes, including Xiap, an X-linked inhibitor of apoptosis (data not shown). These findings point to specific epigenetic alterations in key neurodevelopmental pathways.

Figure 2.

Genome-wide DNA methylation alterations in the hippocampus of Mtr-cKO mice

(A) Circos plot showing CpG methylation differences between Mtr-cKO and WT hippocampi (N = 6/group). Red and green dots indicate hyper- and hypomethylated sites; the inner track shows density across the genome.

(B) Volcano plot of CpG methylation changes with highlighted significant genes; boxplots display distribution of methylation differences and q values. Threshold: |Δmethylation| ≥ 25%, false discovery rate (FDR) <0.05.

(C and D) Pathway enrichment of genes associated with differentially methylated CpGs: (C) GO Biological Process; (D) PANTHER pathways. Dot size indicates gene count, color scale reflects −log₁₀(FDR).

(E) Volcano plot of DMRs with labeled genes at significant loci. Threshold: |Δmethylation| ≥ 15%, FDR <0.05.

(F) GO Biological Process enrichment of genes linked to DMRs.

(G) CpG methylation profile of the Pdha1 promoter on chromosome X; red asterisks denote significant CpGs.

Proteomic analysis identified 313 proteins differentially expressed in the hippocampus of Mtr-cKO mice, with the vast majority (95.8%) being downregulated (Figures 3A and 3B). These proteins were predominantly localized to the cytoplasm (38.3%) and mitochondria (21.1%). Functional enrichment analysis revealed strong associations with metabolic processes including NADH regeneration and canonical glycolysis (Figure 3C). Pathway mapping using the Panther system confirmed enrichment in glycolysis, tricarboxylic acid (TCA) cycle, and pyruvate metabolism (Figure 3D), indicating broad impairment of energy metabolism.

Figure 3.

Proteomic alterations in the hippocampus of Mtr-cKO mice

(A) Heatmap of differentially expressed proteins in Mtr-cKO and WT hippocampi (N = 4/group), with hierarchical clustering of proteins and samples. Protein subcellular localizations are annotated by color (GO Cellular Component). Values are scaled Z scores.

(B) Volcano plot of protein expression changes, highlighting significantly downregulated (green) and upregulated (red) proteins; boxplots show fold change and −log₁₀p value distributions. Threshold: fold change (FC) ≤ 0.9 or ≥1.1, p < 0.05.

(C and D) Functional enrichment of differentially expressed proteins: (C) GO Biological Process terms; (D) PANTHER pathways. Dot size indicates protein count; color scale reflects −log₁₀(FDR).

(E) PPI network of significantly dysregulated proteins, with edges representing interaction confidence and node colors grouping proteins by biological process.

To assess functional relationships among dysregulated proteins, we constructed a protein-protein interaction (PPI) network (Figure 3E). This network revealed clusters of proteins involved in glycolysis, TCA cycle, synaptic transmission, vesicle-mediated transport, cytoskeletal organization, and mitochondrial respiration. Proteins associated with oxidative phosphorylation formed a highly interconnected module, indicating coordinated disruption of mitochondrial bioenergetics. The central positioning of metabolic enzymes, particularly those involved in glycolysis and the TCA cycle, underscores their role as key hubs in the altered proteomic landscape of Mtr-cKO mice.

SIRT1 downregulation mediates mitochondrial metabolic dysfunction in Mtr-deficient hippocampus

Multi-omics integration reveals disruption of glycolysis and the TCA cycle

To validate the alterations observed in the omics datasets, we integrated proteomic and metabolomic analyses focused on glycolysis and TCA cycle metabolism. This analysis revealed major impairments in both pathways (Figures 4A and 4B). Proteomic quantification was used to assess the expression levels of key enzymes involved in glycolysis and the TCA cycle within the hippocampus. Notably, pyruvate levels were elevated and the lactate-to-pyruvate ratio was reduced, consistent with glycolytic dysfunction (Figure 4A). These metabolic changes coincided with reduced hippocampal expressions of LDHA and PDHA1. Oxaloacetate levels were increased, likely due to maintained pyruvate carboxylase (PC) expression, while malate accumulation was associated with reduced expression of mitochondrial and cytosolic malate dehydrogenases (MDH2 and MDH1). Although citrate and isocitrate levels remained unchanged, α-ketoglutarate levels were elevated despite decreased expression of upstream enzymes (CS, ACO2, and IDH3A), potentially suggesting compensatory anaplerotic flux through the glutamate pathway. This was supported by sustained GLUD1 expression and a trend toward glutamate levels (p.adj = 0.09). Arginine and citrulline levels remained stable, excluding a role for nitrogen metabolism in α-ketoglutarate elevation. Flux modeling under maximal energy demand confirmed disruptions in both glycolysis and the TCA cycle (Figure S2A).

Figure 4.

TCA cycle-related metabolic and proteomic alterations in the hippocampus of Mtr-cKO mice

(A) Heatmap of hippocampal metabolites related to the TCA cycle and associated pathways in WT and Mtr-cKO mice (N = 6/group). Rows represent metabolites and columns represent samples, with Z score scaling and hierarchical clustering. Group, sex, and pathway annotations are indicated.

(B) Log₂ fold changes of TCA-related metabolites in Mtr-cKO vs. WT mice; adjusted p values are shown, and color scale reflects fold change magnitude.

(C) Protein-protein interaction network of dysregulated proteins involved in energy metabolism (TCA cycle, glycolysis, pyruvate metabolism, and gluconeogenesis), highlighting a module including SIRT1, Ppargc1a, and Pparg. Clusters are color coded by function.

(D) Normalized hippocampal SIRT1 mRNA and protein levels; α-tubulin as loading control. Statistical significance was assessed using unpaired two-tailed Student’s t test. Data are presented as mean ± SEM.

(E and F) Correlation of SIRT1 protein levels with hippocampal metabolites (E) or enzymes (F) involved in energy metabolism. Dot size reflects Pearson’s r; color indicates p value; significant correlations (p < 0.05) marked with asterisks.

(G) Schematic of the TCA cycle and related pathways, with altered proteins/metabolites highlighted (green, downregulated; red, upregulated). Created with BioRender.com.

Data are presented as mean ± SEM, with individual values shown. Sample sizes correspond to the number of points displayed in each graph. Statistical tests were selected according to data distribution and variance.

SIRT1-PGC1α axis links mitochondrial gene expression to epigenomic and proteomic remodeling

We next explored whether the downregulation of TCA enzymes could reflect epigenetic consequences of MS deficiency. Promoter hypermethylation was specifically detected for Pdha1 (Figure 2G), while other TCA-related genes showed no methylation change. Using the STRING database, we identified several functional clusters among the dysregulated proteins (Figure 4C), notably involving TCA enzymes (cluster A, green) and pyruvate metabolism (cluster D, purple), connected to a regulatory module centered on SIRT1-PGC1α (Cluster C, blue). This network supports a mechanistic link between SIRT1 dysfunction and metabolic disruption, as previously described in MDD rat brains.^4,13 Consistently, we observed a marked decrease in SIRT1 expression at both mRNA and protein levels in the hippocampus of Mtr-cKO mice (Figure 4D). In line with the previously described mechanism involving HuR methylation in the regulation of SIRT1 expression,¹⁵ we observed a significant decrease in methyl-HuR levels in the hippocampus of Mtr-cKO mice (Figure S2B), along with a prominent nuclear accumulation observed by immunostaining (Figure S2C), both of which support a post-transcriptional destabilization of SIRT1 mRNA contributing to its downregulation. SIRT1 protein levels were strongly correlated with key metabolite concentrations (e.g., pyruvate, α-ketoglutarate, glutamate, malate, oxaloacetate) (Figures 4E–4G) and with expression of several TCA-related enzymes including CS, LDHA, PDHA1, IDH3A, and MDH2 (Figures 4F and 4G), supporting its role in metabolic regulation.²⁷

Mtr deletion impairs mitochondrial respiration and ATP production

We then assessed the impact of Mtr deletion on mitochondrial function using high-resolution respirometry. Mtr-cKO hippocampi exhibited a significant reduction in basal, maximal, and ATP-linked respiration (Figure 5A). Decreased maximal respiration further confirmed oxidative phosphorylation (OXPHOS) impairment (Figure 5A). Although no significant changes were observed in other respiratory parameters, we noted a reduced expression of several proteins related to complex I and ATP synthase of the mitochondrial respiratory chain (Figure S3A). Cross-referencing our proteomic data with the MitoCarta 3.0 database²⁸ identified 58 significantly dysregulated mitochondrial proteins, representing 18.5% of the altered proteome (Figures 5B and 5C). These proteins were mostly downregulated and involved in key mitochondrial functions such as energy metabolism, detoxification, OXPHOS, signaling, and structural maintenance. PPI network analysis revealed a dense core of interacting proteins centered on OXPHOS and mitochondrial metabolism, indicating coordinated dysfunction (Figure 5D). GO enrichment confirmed that these proteins were strongly associated with aerobic respiration, cellular respiration, and generation of precursor metabolites (Figure 5E). Finally, SIRT1 expression showed a strong positive correlation with maximal mitochondrial respiration (Figure S3B), reinforcing its potential role in regulating mitochondrial bioenergetics in the context of MS deficiency.

Figure 5.

Mitochondrial alterations in the hippocampus of Mtr-cKO mice

(A) Mitochondrial respiration measured by oxygraphy in hippocampal tissue, including basal, maximal, and ATP-linked respiration. Values are expressed as pmol O₂/s/mg tissue. Data are presented as mean ± SEM; statistical significance was assessed using unpaired two-tailed Student’s t test.

(B) UpSet plot showing overlap between dysregulated hippocampal proteins and mitochondrial proteins annotated in MitoCarta 3.0.

(C) Heatmap of significantly dysregulated mitochondrial proteins, annotated by functional categories (Detoxification and Homeostasis, Metabolism, Maintenance, OXPHOS, Protein Synthesis, Signaling/Dynamics). Rows represent proteins and columns represent samples, with Z score scaling and hierarchical clustering. Group and sex annotations are shown.

(D) PPI network of dysregulated mitochondrial proteins (STRING, visualized in Cytoscape), with clusters highlighted by color-coded pathways.

(E) GO Biological Process enrichment of dysregulated mitochondrial proteins; dot size indicates gene count, color scale reflects −log₁₀(FDR). Data are presented as mean ± SEM, with individual values shown. Sample sizes correspond to the number of points displayed in each graph. Statistical tests were selected according to data distribution and variance.

Wnt/β-catenin signaling is disrupted in Mtr-deficient hippocampus through a SIRT1-mTORC2-Akt-GSK3β axis

MS deficiency leads to altered Wnt signaling in the hippocampus of Mtr-cKO mice

We observed notable alterations in several key regulators of the Wnt signaling pathway in Mtr-cKO hippocampi. Specifically, Wnt7a and its downstream effector β-catenin, both involved in hippocampal neurogenesis,²⁹ were significantly reduced at the protein level (Figures 6A and 6B). The decline in β-catenin was not transcriptional, as mRNA levels were unchanged (Figure S3C), but post-translational, evidenced by elevated levels of its phosphorylated (inactive) form at Ser33/37/Thr41, resulting in an increased pβ-catenin/β-catenin ratio (Figure 6C). Concomitantly, phosphorylation of GSK3β at Ser9—a modification that inhibits its β-catenin-degrading activity—was significantly decreased, while total GSK3β was upregulated. This led to a reduced pGSK3β/GSK3β ratio, favoring β-catenin degradation and suppression of Wnt signaling (Figures 6A–6C).

Figure 6.

Mtr deletion impairs Wnt signaling, alters neuronal/glial differentiation, and affects behavior

(A) Western blot analysis of hippocampal lysates showing proteins involved in Wnt signaling, proliferation, stemness, and neuronal differentiation.

(B) Heatmap of protein expression (Z scores) with hierarchical clustering; functional roles (activator, repressor, regulator, driver), group, and sex annotations are indicated.

(C) Ratios of phosphorylated to total protein levels (β-catenin, GSK3β, AKT) from western blots; each dot represents one sample. Statistical significance was assessed using unpaired two-tailed Student’s t test or Mann-Whitney test as appropriate. Data are presented as mean ± SEM.

(D) Quantification of neuronal (NeuN), oligodendrocyte (Olig2), and astrocyte (GFAP) markers in cortical lysates by western blot. Statistical significance was assessed using unpaired two-tailed Student’s t test. Data are presented as mean ± SEM.

(E) Immunofluorescence of mature neurons (NeuN/MAP2), oligodendrocytes (Olig2/SOX10), and astrocytes (GFAP) in cortex sections from 21-day-old WT and Mtr-cKO mice; scale bars, 100 μm.

(F) Water maze test showing escape success rates over five sessions, expressed as percentage of successful escapes (9–14 mice/group); significance assessed with chi-square test with Yates’ correction.

(G) SIRT1 protein levels in hippocampal lysates from WT and Mtr-cKO mice treated with vehicle or SRT2104, with α-tubulin as loading control. Significance was assessed by two-way ANOVA with Tukey’s post hoc test. Data are presented as mean ± SEM, with individual values shown. Sample sizes correspond to the number of points displayed in each graph.

SIRT1 regulates Wnt signaling via rictor/Akt/GSK3β axis

Given the established link between SIRT1 and Wnt regulation,^30,31,32,33 and its observed downregulation in Mtr-cKO hippocampus, we investigated the involvement of SIRT1 in this signaling disruption. SIRT1 is known to modulate Akt activation via Rictor, a component of the mTORC2 complex.³⁴ Although total Akt expression remained unchanged, its active phosphorylated form (pAkt Ser473) was significantly decreased, correlating with reduced GSK3β inhibition (Figures 6A–6C). This is consistent with a marked decrease in Rictor protein levels (Figures 6A and 6B). Furthermore, SIRT1 levels were significantly associated with both β-catenin and Rictor protein expression (Figure S3D), suggesting that Mtr-dependent SIRT1 deficiency could impair the mTORC2-Akt-GSK3β signaling axis and Wnt pathway integrity.

Wnt signaling disruption leads to impaired neurogenesis and cognitive dysfunction

Western blot analyses revealed decreased expression of neural stem cell (NSC) markers (SOX2, Nestin) and neuronal progenitor marker NeuroD1 in Mtr-cKO hippocampi (Figures 6A and 6B), consistent with reduced NSC proliferation. Downregulation of Wnt target genes involved in cell cycle regulation—cyclin D1, c-Myc, and PCNA—further supported diminished neurogenic activity. In line with this, the expression of pro-neurogenic transcription factors Ascl1 and Ngn2 was also decreased at the mRNA level (Figures S4A and S4B). These molecular changes translated into altered cellular composition, as shown by decreased NeuN and Olig2 (neurons and oligodendrocytes) and increased GFAP expression (astrocytes) in the frontal cortex (Figure 6D). Immunofluorescence confirmed reduced NeuN⁺/MAP2⁺ neurons and Olig2⁺/SOX10⁺ oligodendrocytes, along with increased GFAP⁺ astrocytes in Mtr-cKO brains (Figure 6E), indicating impaired neurogenesis and astrogliosis—a hallmark of accelerated brain aging. To assess cognitive consequences, we performed a water maze test. While WT mice consistently escaped from session 2 onward, Mtr-cKO mice showed persistent learning deficits, particularly from sessions 3 to 5 (Figure 6F), consistent with hippocampal dysfunction.

Pharmacological activation of SIRT1 restores metabolic and neurogenic functions and rescues cognitive deficits in Mtr-deficient mice

SRT2104 rescues mitochondrial metabolism through normalization of key metabolites and enzymes

To test the hypothesis that SIRT1 deficiency underlies the observed metabolic alterations in Mtr-cKO mice, we treated animals with SRT2104, a selective SIRT1 activator. Treatment was administered from postnatal days 15–36, and its effects were assessed in hippocampal and cortex tissue. SRT2104 treatment robustly increased SIRT1 protein levels in Mtr-cKO mice compared to vehicle controls (Figure 6G). Given that SIRT1 activity is tightly coupled to intracellular NAD⁺ availability, we also examined the expression of NAMPT, the rate-limiting enzyme in the NAD⁺ salvage pathway and an upstream regulator of SIRT1 functionality. Our analyses revealed a non-significant decrease in NAMPT expression in Mtr-cKO mice compared to WT controls. Interestingly, SRT2104-treated mice showed a slightly non-significant increase in NAMPT levels, which may reflect a partial restoration of NAD⁺ biosynthetic capacity. These findings suggest that the pharmacological activation of SIRT1 may secondarily promote NAD⁺ salvage through NAMPT upregulation (Figures S5A and S5B). Metabolite profiling (Figure S5C) revealed normalization of pyruvate, malate, and succinate and notably, a significant reduction of α-ketoglutarate levels in SRT2104-treated Mtr-cKO hippocampi (Figure 7A). Principal-component analysis confirmed this correction, with treated Mtr-cKO mice clustering closely with WT animals (Figure 7B). On the enzymatic level, SRT2104 restored the expression of key TCA cycle enzymes (Mdh1, Pdha1, Cs, and Mdh2) that were downregulated in untreated Mtr-cKO mice (Figure 7C), demonstrating a functional rescue of mitochondrial metabolism.

Figure 7.

Pharmacological activation of SIRT1 improves metabolism, epigenetic regulation, and cognition in Mtr-cKO mice

(A) Quantification of hippocampal metabolites altered by SRT2104. Statistical significance was assessed using unpaired two-tailed Student’s t test. Data are presented as mean ± SEM.

(B) PCA of hippocampal metabolite profiles across WT-vehicle, WT-SRT2104, Mtr-cKO-vehicle, and Mtr-cKO-SRT2104 groups.

(C) Quantitative reverse-transcription PCR of metabolic enzymes (Mdh1, Mdh2, Cs, and Pdha1) in hippocampal tissue.

(D) Quantification of β-catenin and Rictor protein levels in hippocampal lysates; α-tubulin as loading control.

(E) Quantification of NeuN, Olig2, and GFAP proteins in cortical lysates; α-tubulin for NeuN/Olig2 and β-actin for GFAP.

(F) Water maze escape success rates across five sessions in WT and Mtr-cKO mice with vehicle or SRT2104 (9–14/group); significance assessed by chi-square test with Yates’ correction.

(G) Western blot of H3K4me3 levels in hippocampal lysates; total H3 as loading control.

(H) PCA of genome-wide H3K4me3 ChIP-seq profiles across groups.

(I) Genome browser tracks of H3K4me3 enrichment at neurogenic gene promoters (Runx2, Wnt3, Wnt10a, and Pax6).

(J) Quantification of normalized H3K4me3 ChIP-seq signal at selected promoters.

Data are presented as mean ± SEM, with individual values shown. Sample sizes correspond to the number of points displayed in each graph. Statistical significance was determined by two-way ANOVA with Tukey’s post hoc test, unless otherwise indicated.

SRT2104 restores rictor/β-catenin signaling and improves hippocampal-dependent cognitive function

We next investigated whether SRT2104 could restore the disrupted SIRT1-Rictor-β-catenin axis and improve neurogenic outcomes in Mtr-cKO mice. Treatment significantly increased Rictor and β-catenin protein expression in the hippocampus, reaching levels comparable to WT controls (Figure 7D). This molecular rescue was accompanied by an increased expression of NeuN and a reduction in GFAP expression in the cortex, suggesting enhanced neurogenesis and reduced astrogliosis (Figure 7E). Olig2 levels remained unchanged, indicating that oligodendrocyte populations were not significantly affected. To assess the functional consequences of these molecular changes, we evaluated spatial learning and memory using the Morris water maze. Vehicle-treated Mtr-cKO mice exhibited persistent learning deficits across sessions, while SRT2104-treated Mtr-cKO mice showed significant improvement, particularly from sessions 4 to 5 (Figure 7F). Their performance approached that of WT controls, indicating that pharmacological activation of SIRT1 not only restores molecular and cellular markers of neurogenesis but also improves hippocampal-dependent cognitive function.

SRT2104 restores hippocampal H3K4me3 levels and reactivates neurodevelopmental transcriptional programs

Finally, to further investigate the epigenetic mechanisms underlying transcriptional dysregulation, we analyzed the status of the H3K4me3 histone mark, a hallmark of active gene promoters, which is known to be sensitive to cellular SAM availability.³⁵ Western blot analysis revealed a significant reduction in global H3K4me3 levels in the hippocampus of Mtr-cKO mice. Notably, SRT2104 treatment restored H3K4me3 levels to baseline, suggesting a reversal of MS deficiency-induced epigenetic repression (Figure 7G).

To identify genes influenced by changes in H3K4me3 levels, we mapped the genomic distribution of H3K4me3 using chromatin immunoprecipitation sequencing (ChIP-seq) in hippocampal samples collected from WT, KO, and SRT2104-treated mice. Analysis of H3K4me3 ChIP-seq data identified 14,006 peaks corresponding to 17,213 promoter sites. PCA analysis based on H3K4me3 signal revealed that KO samples clustered separately from the WT group (Figure 7H). Interestingly, the SRT2104-treated KO group shifted toward the WT cluster and was indistinguishable from the WT + SRT2104-treated group. Differential signal analysis using DESeq2 revealed 72 significantly downregulated and 59 upregulated promoters in the KO vs. WT comparison. Interestingly, many genes that were downregulated in KO mice were restored to WT levels following SRT2104 treatment. Differential peak analysis showed that H3K4me3 enrichment was markedly reduced in Mtr-cKO mice at the promoters of genes involved in neurodevelopment. Among these, several genes involved in the Wnt signaling pathway, such as Sp8, Wnt3, Wnt10a, Ascl1, and Runx2, were notably rescued. Similarly, genes associated with neural differentiation, including Pax6, Pax3, and Nkx1-1, also exhibited recovery to WT expression levels upon SRT2104 treatment (Figures 7I and 7J).

Interestingly, we did not observe significant alterations in H3K4me3 enrichment at the promoters of genes involved in mitochondrial or energy metabolism, despite the marked restoration of their mRNA expression upon SRT2104 treatment. This suggests that their transcriptional regulation is likely mediated through alternative mechanisms. In line with this, PPI analysis (Figure 4C) pointed to a regulatory mechanism involving the SIRT1-PGC1α axis, whereby SIRT1-driven deacetylation of PGC1α enhances the transcription of mitochondrial genes independently of H3K4me3-mediated promoter activation. Notably, this SIRT1-PGC1α axis has also been reported in other MDD models,^12,13,14 further supporting its central role in coordinating mitochondrial function and energy homeostasis. These findings underscore the dual mechanism by which SIRT1 exerts its effects in Mtr-cKO mice: (1) by reactivating neurodevelopmental transcriptional programs via restoration of H3K4me3 marks at key promoters and (2) by enhancing mitochondrial gene expression through non-histone epigenetic regulation involving PGC1α deacetylation. These complementary regulatory modes likely underpin the broad therapeutic efficacy of SRT2104 in reversing both molecular and behavioral impairments. Moreover, H3K4me3 enrichment was not significantly altered at the promoter regions of Rictor, Sirt1, or Mtr, reinforcing the view that their regulation occurs via H3K4me3-independent mechanisms (Figure S6A). The absence of H3K4me3 signal at the Mtr promoter, which is genetically ablated in our Cre-Lox model, also serves as an internal negative control, validating the specificity and robustness of our ChIP-seq approach.

Discussion

The mechanisms underlying the variability, severity, and limited response to B12 supplementation in IECM-related neurological symptoms remain poorly understood.² The study of fibroblasts and cell models of IECM and a rat model of vitamin B12 deficiency during pregnancy and lactation highlighted dramatic genomic and epigenomic consequences related to neurodevelopmental alterations and neuroplasticity.^{15,16,17,36,37} However, the persistence of neurological and cognitive disorders in IECM patients subjected to high doses of vitamin B12 therapy is still unexplained. Elucidating the molecular mechanisms of impaired MS activity and the underlying targets is essential to design innovative therapeutic approaches in cases with severe presentations and limited amelioration by current conventional therapies. To address this issue, we developed a transgenic mouse model with selective constitutive Mtr gene deletion in the brain and performed an integrated multi-omics analysis combining methylome, histone profiling, proteomics, and metabolomics. This revealed profound alterations in brain physiology, including altered one-carbon metabolism leading to a decreased SAM/SAH ratio and activation of the transsulfuration pathway.

The silencing of MS activity led to profound disruptions of energy metabolism and neurogenesis, and to an imbalance in astrocyte vs. neuron differentiation, characterized by loss of SIRT1 expression and disruption of the SIRT1/Wnt axis, which played a central role. Indeed, we found a strong decrease in SIRT1 levels, without changes in promoter DNA methylation or H3K4me3 enrichment, suggesting regulation via a non-canonical, post-transcriptional mechanism. This appears to involve reduced HuR methylation (Me-HuR) and its increased nuclear accumulation, both of which are known to destabilize SIRT1 mRNA. These observations are supported by our results showing decreased Me-HuR expression and by immunostaining analyses revealing a higher proportion of HuR nuclear localization in KO brains. Such mechanisms align with findings in Cd320-KO mice and fibroblasts from patients with the cblG variant of homocystinuria-megaloblastic anemia.^15,17 Altogether, our data highlight the pivotal role of SIRT1 in brain energy metabolism and neurodevelopmental integrity in the context of MS silencing and support the therapeutic potential of SRT2104 in IECM.

SIRT1 is a critical regulator of stem cell maintenance and metabolism. It promotes self-renewal through deacetylation of p53, inhibition of p21, and activation of Nanog, OCT4, and SOX2. It also interacts with and acetylates OCT4 and deacetylates SOX2 through direct binding to OCT4, thereby maintaining stemness in naive pluripotent stem cells.³⁸ In our model, SIRT1-related disruption of mitochondrial energy metabolism may influence altered stem cell homeostasis in the hippocampus and cortex. The decreased proliferation, reduced neuronal differentiation, and increased astrocytic differentiation observed in Mtr-cKO mice could ultimately lead to a quiescent state of the neurogenic niche, a phenotype commonly observed in neurodegenerative diseases with hippocampal atrophy such as Alzheimer’s disease.³⁹ These findings were associated with profound mitochondrial dysfunction, characterized by reduced expression of key glycolytic and TCA cycle enzymes, accumulation of intermediary metabolites, and impaired basal and maximal respiration, ultimately compromising ATP production.

Importantly, while PDHA1 repression was linked to promoter hypermethylation, other mitochondrial genes lacked such epigenetic marks, including H3K4me3 enrichment, suggesting a broader regulatory mechanism. The loss of SIRT1 activity, along with a low SAM/SAH ratio, likely caused hypomethylation and hyperacetylation of PGC1α, impairing its role in mitochondrial gene regulation. This has already been reported in MDD models and may explain the widespread mitochondrial repression observed.¹³ Although we could not assess PGC1α modifications due to limited tissue, this warrants further investigation. Another mechanism involved in altered energy metabolism could be related to the increased transsulfuration pathway that was evidenced by increased protein expression of CBS and increased concentration of cystathionine, one of the CBS intermediate metabolites. Indeed, the negative correlation of cystathionine with maximal respiration could be related to the inhibitory effect of H₂S, an intermediate metabolite of transsulfuration produced in parallel to cystathionine, on complex IV.⁴⁰

The hippocampus is highlighted as one of the brain areas with high-energy demands due to its intense synaptic activity and metabolic requirements.⁴¹ Computational metabolic modeling suggested that the disruption of the glycolytic pathway and TCA cycle becomes dramatic in high-energy-demand conditions. The SIRT1-related disruption of mitochondrial energy metabolism may influence the altered dynamic control of stem cell homeostasis in the hippocampus and cortex of Mtr-cKO mice.⁴⁰ Proteomic and epigenomic analyses pointed out SIRT1 and Wnt/β-catenin as interacting prominent players in the consequences of Mtr silencing on hippocampus neurogenesis and neurogenic programming. This pathway allows the regulation of the NSC proliferation and differentiation into neurons.^42,43

Earlier studies have demonstrated the severe impact of B12 deficiency on cell proliferation and differentiation via the disrupted Akt pathway in neuron-type cells,^16,44 but the underlying pathological mechanisms remained unclear. Here, we showed a key role of altered SIRT1/Wnt-β-catenin axis, with decreased phosphorylation of Akt and expression of Rictor in the Mtr-cKO mice hippocampus.⁴⁵ Similar to our findings, diminished SIRT1 expression resulted in decreased Rictor levels and subsequent reduction in Akt-S473 phosphorylation in Sirt1 mutant mice.³⁴ SIRT1 is known to regulate the protein stability and transcriptional activity of β-catenin through its acetylation/deacetylation.^46,47 The impaired Wnt/β-catenin was consistent with the decreased expression of target genes of cell proliferation, decreased GSK3β-S9 phosphorylation, and decreased expression of genes involved in neurogenic programming. Our data pointed out the important role of the SIRT1/Wnt axis in maintaining the self-renewal of stem cells and neurogenic programming in hippocampus of Mtr-cKO mice.

The decreased proliferation, reduced neuronal differentiation, and increased astrocyte differentiation could lead to a quiescent state in the neurogenic hippocampus area.⁴⁸ The Wnt-related increased astrocyte differentiation is commonly found in neurodegenerative diseases with hippocampus atrophy such as Alzheimer’s disease.^49,50,51,52 The cblG and cblC myelination impairment could be also related to the reduced oligodendrocyte differentiation observed in our model.⁸ The disruption of neurogenesis and atrophy of the CA1 layer²³ of the hippocampus of Mtr-cKO mice led to severe cognitive deficits in the water maze test, similar to those reported throughout life in pups born from rats subjected to B12- and folate-deficient diet during gestation and lactation.^10,20 The impaired activity of MS produces, therefore, neurodevelopment alterations during pregnancy, which are not reversed after birth. This could explain that high-dose B12 therapy has a limited effect on part of neurological and ophthalmological manifestations in the postnatal and later life. Our data suggest, therefore, paying attention to the treatment of IECM and prevention of B12 deficiency during pregnancy.⁵³

Interestingly, we also observed upregulation of HDAC6, a key stress response regulator linked to proteostasis and ER stress.^54,55 This may reflect a compensatory mechanism to counteract increased misfolded protein burden, as supported by unpublished findings of elevated ER stress markers and HSP90 expression in our model. Such proteostasis imbalance may contribute to the cognitive decline observed in aged Mtr-cKO mice.

Taken together, our data highlighted loss of SIRT1 and disruption of the SIRT1/Wnt axis as key therapeutic targets of the consequences of MS brain silencing on energy metabolism, neurogenesis, and gliogenesis. We, therefore, treated the Mtr-cKO mice with SRT2104 pharmacological activator to assess whether the activation of SIRT1 residual activity could reverse these alterations. SRT2104 is a third-generation SIRT1 pharmacological activator, which emerged as the most potent and specific small-molecule activator of SIRT1. It demonstrated good drug tolerance at high doses (up to 3 g/day) in phase I clinical trials in healthy individuals and elderly volunteers⁵⁶ and has been evaluated in clinical studies for obesity, type 2 diabetes, ulcerative colitis, and psoriasis.^56,57,58 Preclinical evidence also supports its efficacy in disease models ranging from Duchenne muscular dystrophy—where SRT2104 showed anti-inflammatory, anti-fibrotic, and pro-regenerative effects in flies, mice, and patient-derived myoblasts—to various cancers, where it modulates metabolic pathways and suppresses tumor progression. Furthermore, SRT2104 has demonstrated neuroprotective benefits in preclinical models of Alzheimer’s, Parkinson’s, Huntington’s disease, and optic nerve degeneration via modulation of mitochondrial function, autophagy, and inflammation.^56,59,60 It is a well-tolerated compound, which has never been evaluated in inherited metabolic diseases.

The treatment with the SIRT1 activator SRT2104 reversed the main observed defects. It restored SIRT1 protein levels, normalized key metabolic enzymes, and reduced metabolite accumulation. These effects occurred despite no H3K4me3 changes at metabolic gene promoters, consistent with a chromatin-independent mechanism involving PGC1α activation. Our data support the SIRT1-PGC1α axis as a therapeutic target for mitochondrial dysfunction in MS deficiency. Notably, our findings align with recent observations in methylmalonic aciduria (MMA), another IECM. In MMA, multi-omics profiling has also revealed disruptions of TCA cycle and anaplerotic flux, particularly through glutamine metabolism. Restoration of TCA cycling with dimethyl-oxoglutarate has shown therapeutic benefit in Mmut mouse models, highlighting shared vulnerabilities and suggesting that metabolic interventions may be applicable across IECMs.⁶¹ Furthermore, gene therapy with an AAV8 vector to increase SIRT5 expression was shown to reduce hyperammonemia and methylmalonylation in MMA models, reinforcing the therapeutic potential of modulating mitochondrial regulators in inherited metabolic diseases.⁶²

SRT2104 treatment also upregulated the expression of Rictor and β-catenin in the Mtr-cKO mice, compared to the vehicle-treated counterparts. These upregulations may explain the effects of SRT2104 on neurogenesis and gliogenesis, evidenced by the increased expression of NeuN and decreased expression of GFAP, in the brain of treated mice. The upregulation in the expression of Rictor may be explained by the direct SIRT1 binding upstream of the Rictor gene supporting a positive transcriptional regulation of the Rictor gene by SIRT1.³⁴ The effects of SRT2104 on Wnt/β-catenin are also consistent with the role of SIRT1 on deacetylation and protein stability of Wnt/β-catenin and its control of progenitor cell proliferation.^38,63,64 At the epigenetic level, H3K4me3 profiling showed loss of active marks at key neurodevelopmental genes (Sp8, Pax6, Wnt3, Wnt10a, and Runx2), many of which were rescued by SRT2104. This suggests that SIRT1 also promotes neurogenic transcription via chromatin remodeling. The molecular and cellular benefits were corroborated by the recovery of cognitive abilities in the water maze test. The effect was spectacular, with a complete recovery of efficient learning of the optimal maze escape route. These results are consistent with previous data from the literature on the efficacy of increasing expression and/or activity of sirtuins to improve metabolic and cognitive outcomes.^4,17,56,62 Interestingly and given the prevalence of Wnt signaling disruption in disorders such as autism spectrum disorders, intellectual disability, and neurodevelopmental syndromes, the relevance of our findings may extend to broader classes of neuropathologies.^49,50,51,52

Taken together, our findings identify two potential complementary modes of SRT2104 action: (1) a chromatin-independent restoration of mitochondrial gene expression via PGC1α activation and (2) a chromatin-dependent reactivation of neurogenic transcription programs via H3K4me3 remodeling. These mechanisms converge on SIRT1 as a central hub controlling both metabolic and neurodevelopmental outcomes. Although we were not able to assess post-translational modifications of PGC1α, this question remains critical for future studies. Given the dysregulation of both mitochondrial and Wnt pathways in a wide range of neurodevelopmental and neurodegenerative disorders—including Alzheimer’s, Parkinson’s, and Huntington’s disease and neurodevelopmental syndromes—SRT2104 may hold promise beyond the context of IECM.

While current treatments with high dose of OH-Cbl have limited effectiveness in preventing the progression of early and later onsets of neurological and ocular symptoms,^6,7,8,9 our results on the preclinical study of SRT2104 open promising perspectives of drug repurposing and highlight the need for a clinical trial to evaluate the effects on persistent ocular and neurological symptoms in patients receiving conventional or high-dose B12 therapy. In conclusion, we explored the effects of Mtr brain-selective deletion at the postnatal stage in mice, identified SIRT1 as a target of innovative therapy, and performed a drug-repurposing preclinical study of the pharmacological activator SRT2104 to reverse the main molecular, cellular, and cognitive alterations. Mtr silencing produces a significant disruption of mitochondrial energy metabolism, neuronal proliferation/differentiation, and neurogenic programming related to impaired SIRT1/Wnt axis in the brain hippocampus. These dramatic neurodevelopmental alterations in very early life may explain why some neurological and cognitive symptoms of patients with IECM are resistant to the subsequent vitamin B12 therapy. The SRT2104 drug repurposing could represent a promising complementary treatment of neurological symptoms in IECM cases with insufficient effects of supplementation with high dose of B12. Our results support the rationale to evaluate the therapeutic effect of SRT2104 in clinical trials targeting persistent neurological and ophthalmological symptoms in B12-treated patients with inherited cobalamin disorders.

Limitations of the study

The limitations of our work include several experimental aspects that could further refine the mechanisms identified here. First, although our data implicate post-translational regulation of PGC1α in mitochondrial repression, we were not able to directly assess PGC1α acetylation or methylation due to limited tissue availability. Such analyses would help clarify the contribution of the SIRT1-PGC1α axis to metabolic recovery under SRT2104. Second, while our multi-omics profiling under basal conditions revealed extensive metabolic and epigenetic alterations, a dedicated proteomic analysis performed under SRT2104 treatment would be valuable to identify downstream pathways specifically restored by SIRT1 activation. Third, studies using patient-derived iPSC-neurons or neural progenitors would provide a human cellular framework to validate the molecular defects observed in vivo and explore inter-individual variability. Fourth, although we characterized mitochondrial respiration in untreated Mtr-cKO mice, we did not perform oxygraphy in SRT2104-treated animals, which would help quantify the extent of functional mitochondrial rescue. Finally, analyses performed at later ages or at embryonic stages may reveal earlier or delayed mechanistic events, which could help define developmental windows of vulnerability and guide potential prenatal or early-life therapeutic strategies in IECM.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Jean-Louis Guéant (jean-louis.gueant@univ-lorraine.fr).

Materials availability

This study did not generate new unique reagents. All commercial reagents, antibodies, and kits used in this work are listed in the key resources table.

Data and code availability

• •Data: This study generated methylome (GEO: GSE312263), ChIP-seq (GEO: GSE311053), and proteomics (PRIDE: PXD071487) datasets, all of which are publicly available and listed in the key resources table.
• •Code: Custom R scripts used for data processing and visualization have been deposited and are available at Zenodo: https://doi.org/10.5281/zenodo.17788832.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This project is supported by Inserm, University of Lorraine, CHRU Nancy, FHU ARRIMAGE, EpiGONE and PREDICTS ANR projects, the PIA project Lorraine Université d’Excellence (LUE) ANR-15-IDEX-04-LUE, the OMAGE grant from the region GrandEST of France, and the FEDER funding from European Union (EU). Y.A. was supported by Medical Research Council grant MR/X008517/1 and UKRI2374, Biotechnology and Biological Science Council grant UKRI1947, Leverhulm Trust grant RPG-2024-304, and Royal Society Research grant RG\R1\241024. A European patent application EP26305174 covering this work has been filed.

Author contributions

K.M. conducted and designed research, performed experiments, analyzed data, performed statistical analysis, wrote the manuscript, and had primary responsibility for final content. R.-M.G.-R. and J.-L.G. conducted and designed research, analyzed data, performed statistical analysis, wrote the manuscript, and had primary responsibility for final content. G.P., Y.A., and R.S. conducted research, performed analyses, analyzed data, and revised paper. D.C., Z.H., A.H., R.U., and J.-M.A. performed analyses, analyzed data, and revised paper. C.A., O.B., S.L., S.H., P.R., J.P., S.V.S., M.J., J.-M.C., and L.L. performed analyses. All authors have read and agreed to the published version of the manuscript.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-Mtr Polyclonal Antibody	Protein Tech	Cat#25896-1-AP; RRID: AB_2880287
Anti-Cre Recombinase (D3U7F) Polyclonal Antibody	Cell Signaling	Cat#12830; RRID: AB_2631055
Anti-Cbs Polyclonal Antibody	Abcam	Cat#Ab135626; RRID: AB_2814659
Anti-Wnt7A Polyclonal Antibody	Protein Tech	Cat#10605-1-AP; RRID: AB_2215625
Anti-β-Catenine (D10A8) XP Monoclonal Antibody	Cell Signaling	Cat#8480; RRID: AB_11127855
Anti-Gsk3β (27C10) Monoclonal Antibody	Cell Signaling	Cat#9315; RRID: AB_490890
Anti-pGsk3β (Ser9) Polyclonal Antibody	Cell Signaling	Cat#9336; RRID: AB_331405
Anti-Akt1 Polyclonal Antibody	Cell Signaling	Cat#9272; RRID: AB_329827
Anti-pAkt (Ser473) Monoclonal Antibody	Cell Signaling	Cat#4060; RRID: AB_2315049
Anti-Rictor Monoclonal Antibody	Millipore	Cat#05-1471; RRID: AB_11213687
Anti-Sirt1 (1F3) Monoclonal Antibody	Cell Signaling	Cat#8469S; RRID: AB_10999470
Anti-Sox2 (Y-17) Polyclonal Antibody	Santa Cruz	Cat#Sc-17320; RRID: AB_2286684
Anti-Nestin Monoclonal Antibody	Abcam	Cat#Ab11306; RRID: AB_1640723
Anti-Gfap Polyclonal Antibody	Abcam	Cat#Ab7260; RRID: AB_305808
Anti-C-Myc (C-19) Polyclonal Antibody	Santa Cruz	Cat#Sc-788; RRID: AB_631277
Anti-CyclinD1 (92G2) Monoclonal Antibody	Cell Signaling	Cat#2978; RRID: AB_2259616
Anti-Pcna (PC10) Monoclonal Antibody	Cell Signaling	Cat#2586S; RRID: AB_2160343
Anti-Neurod1 Polyclonal Antibody	Abcam	Cat#Ab16508; RRID: AB_470254
Anti-β-Actin Monoclonal Antibody	Abcam	Cat#ab197277; RRID: AB_449644
Anti-α-Tubulin Monoclonal Antibody	Abcam	Cat#ab185067; RRID: AB_2819060
Anti-Peroxidase-labeled anti-rabbit	Interchim	Cat#711-035-152; RRID: AB_10015282
Anti-Peroxidase-labeled anti-mouse	Interchim	Cat#715-035-152; RRID: AB_10015282
Anti-Olig2 Polyclonal Antibody	Millipore	Cat#AB9610; RRID: AB_570666
Anti-NeuN (A60) Monoclonal Antibody	Chemicon	Cat#MAB377; RRID: AB_2298772
Anti-H3K4me3 Polyclonal Antibody	ActiveMotif	Cat#39159; RRID: AB_2615077
Anti-MeHuR Polyclonal Antibody	Millipore	Cat#07-468; RRID: AB_11213528
Anti-Nampt Polyclonal Antibody	ProteinTech	Cat# 11776-1-AP; RRID: AB_2298317
Biological samples
Wild-type/Mtr-cKO brain tissues	Institut Clinique de la Souris (ICS)	Strasbourg, France
Wild-type/Mtr-cKO plasma	Institut Clinique de la Souris (ICS)	Strasbourg, France
Mtr-cKO tissues	Institut Clinique de la Souris (ICS)	Strasbourg, France
Chemicals, peptides, and recombinant proteins
ECLTM	Amersham	Cat#RPN2236
DAPI	Fisher Scientific	Cat#D1306
Ammonium persulfate (APS)	Sigma Aldrich	Cat#A-3678
TEMED	Carlo Erba	Cat#600461
Tween 20	Sigma Aldrich	Cat#P-1379
Dithiothreitol (DTT)	Sigma Aldrich	Cat#D0632
Paraformaldehyde (PFA)	Sigma Aldrich	Cat#818715
Triton X-100	Pharmacia Biotech	Cat#HFH10
Sodium Dodecyl Sulfate	Sigma Aldrich	Cat#L4509
Trypsin 25 μg/mL	Promega	Cat#V5280
SRT2104	CliniScience	Cat#orb315495
1% Phenylmethanesulfonyl fluoride	Sigma	Cat#329986
1% Sodium orthovanadate	Sigma	Cat#S6508
Phosphatase Inhibitor Cocktail	Roche	Cat#3824999390
MIR05	Bioblast	Cat#60101-01
Saponin 5mg/mL	Sigma	Cat# 8047-15-2
Pyruvate 2M	Sigma	Cat#P2256
Glutamate 2M	Sigma	Cat#6106-04-3
Succinate 1M	Sigma	Cat#123-25-1
Oligomycin 1M	Sigma	Cat#75351
FCCP 1mM	Sigma	Cat#C2920
Antimycin	Sigma	Cat#A8674
Rotenone	Sigma	Cat#557368
TMPD 15mg/mL	Sigma	Cat#T7394
KCN	Sigma	Cat#60178
Vitamin C 22mg/mL	Sigma	Cat#60178
ADP 1M	Sigma	Cat#A7506
Malate 1M	Sigma	Cat#LMDH-RO
A/G beads	Invitrogen	Cat#17822894
Critical commercial assays
BCA protein assay kit	Interchim	Cat#UP40840A
12-230 kDa Separation module	Bio-Techne	Cat#SM-W001
Anti-rabbit detection module	Bio-Techne	Cat#DM-001
Anti-mouse detection module	Bio-Techne	Cat#DM-002
Nucleospin RNA plus kit	Macherey-Nagel	Cat#DM-001
PrimeScript™ RT Master Mix	Takara	Cat#740984.50
SYBR® Premix Ex Taq™	Takara	Cat#RR390L
Qiagen’s QIAamp DNA	Qiagen	Cat#56304
Quant-iT™ PicoGreen™ dsDNA Assay Kit	Invitrogen	Cat#P7589
DNA ScreenTape Assay kit	Agilent	Cat#5067-5587
Diagenode’s Premium RRBS Kit	Diagenode	Cat#C02030036
High Sensitivity DNA Kit	Agilent	Cat#5067-4626
MiniElute PCR Purification Kit	Qiagen	Cat#28004
Deposited data
Raw and processed proteomic data	This paper	PRIDE: PXD071487
Raw and processed methylome data	This paper	GEO: GSE312263
Raw and processed ChIP-seq data	This paper	GEO: GSE311053
Experimental models: Organisms/strains
C57BL/6 mice targeted knockout of the Mtr gene specifically in the central nervous system	Institut Clinique de la Souris (ICS)	Strasbourg, France
C57BL/6 wild-type mice	Institut Clinique de la Souris (ICS)	Strasbourg, France
Oligonucleotides
Mtr exon 4–6 R/F: CCAGCCACAAACCTCTTGAC/ACACTTGGCCTACCGGATG	Eurogentec	Angers, France
Cre R/F: GCAAACGGACAGAAGCATTT/AGGCAAATTTTGGTGTACGG	Eurogentec	Angers, France
Ctnnb1 R/F: ATCAGGTCAGCTTGAGTAGCC/TCAGCTCGTGTCCTGTGAA	Eurogentec	Angers, France
Ascl1 R/F: AGAAGCAAAGACCGTGGGAG/CTTAGCCCCCTGAAACTGGG	Eurogentec	Angers, France
Ngn2 R/F: GATTCACACGAACTGCCTGC/CGTGGGGAACCTCGTAAGAC	Eurogentec	Angers, France
Sirt1 R/F: CAGCTCAGGTGGAGGAATTGT/CGGCTACCGAGGTCCATATAC	Eurogentec	Angers, France
Pdha1 R/F: GTCGGTTCCCAGTCCATCAG/GCACATGACATTTCTGTTGCG	Eurogentec	Angers, France
Cs R/F: CGGTTTGTCTACCCTTCCCC/GGCAGGATGAGTTCTTGGCT	Eurogentec	Angers, France
Mdh1 R/F: GGGAAACCCAGCCAATACGA/CATCAGCGGTTACACCGAGT	Eurogentec	Angers, France
Mdh2 R/F: CCAGTGAACTCCACCATCCC/AACGTGTTCGCTCTGACGAT	Eurogentec	Angers, France
Software and algorithms
Compass for SW software v6.0.0	Bio-Techne	https://www.bio-techne.com
ImageJ v.1.53	NIH	https://imagej.net/ij/
GraphPad Prism v10	GraphPad Software	http://www.graphpad.com/
Progenesis-Qi software v4.1	Nonlinear Dynamics	https://www.nonlinear.com/
Proteome Discoverer software v2.2	Thermo Scientific	https://www.thermofisher.com/
Mascot search engine v2.5.1	Matrix Science	https://www.matrixscience.com/
MATLAB R2021b	MathWorks, Inc	https://fr.mathworks.com/
Insight software v3.1	Shimadzu	https://www.shimadzu.com/
NIS Elements	Nikon Instruments Inc.	https://www.microscope.healthcare.nikon.com/
R Studio 2023.12.1 + 402	RStudio, Inc	https://www.r-tudio.com/fr/
Custom analysis scripts	This paper	Zenodo: https://doi.org/10.5281/zenodo.17788832

Experimental model and study participant details

Animals, treatment protocol and tissue collection

Our experiments were conducted using C57BL/6 mice with a targeted knockout of the Mtr gene (Mtr-cKO) specifically in the central nervous system. These mice were designed and produced with the Institut Clinique de la Souris (ICS) (Strasbourg, France). The knockout was achieved using the Cre/Lox system, targeting exons 4 and 5, essential for MS activity. Female mice carried “floxed Mtr exons,” while male mice expressed the Cre recombinase under the control of the Thy-1.2 promoter.^22,23 Recombination efficiency was validated using qPCR on floxed exons, and MS protein loss was confirmed by Western blot in brain tissues. The Thy-1.2 promoter was chosen for its brain-specific activity, as validated by Western blot and qPCR analyses across brain and peripheral tissues, confirming selective Mtr deletion in the CNS. No sex-specific differences were observed in any of the measured parameters, and data from male and female mice were therefore pooled for all analyses.

Additionally, an in-silico simulation was performed to model the molecular consequences of deleting exons 4 and 5. This analysis revealed a frameshift downstream of the deletion, resulting in a premature stop codon (Position 192 within the Homocysteine binding domain) and predicting a truncated, non-functional methionine synthase protein. These findings corroborate experimental data, validating the Mtr-cKO model at both experimental and computational levels. Notably, the Western blot validation utilized an antibody targeting a region downstream of the deletion, specifically within the SAM-binding domain, further confirming the reduced level of the full-length MS protein in knockout animals.

Experimental procedures spanned from Day 15 (D15) to Day 42 (D42) under standard laboratory conditions, including a 12-h light/dark cycle. Throughout the experiments, mice had ad libitum access to food and water. The treatment protocol with SRT2104 or Vehicle (DMSO-PEG) began on day 15 (D15) for the treated group until day 36 (D36), continuing for a total of 9 administrations at a frequency of three gavages per week, at a dosage of 50 mg/kg. SRT2104 was diluted using a stock solution in 20% DMSO-PEG.

Euthanasia of the mice was performed either on Day 21 (untreated mice) or Day 42 (treated Vehicle/SRT2104) as dictated by the specific experimental requirements, using an overdose of isoflurane. Tissue samples were immediately frozen in liquid nitrogen and stored at −80°C for subsequent biochemical, molecular, and LC-MS/MS analyses, or used directly for experiments requiring fresh tissues. For blood collection, samples were obtained from the sub-mandibular vein and placed directly into EDTA tubes. These blood samples were then subjected to blood count analyses using the ABC Micros 60 device by Horiba.

Sample sizes ranged from 3 to 14 animals per group, depending on the specific experimental requirements and the availability of animals, particularly for the treatment groups. This was guided by prior studies and statistical considerations to ensure sufficient power to detect significant differences. To account for potential variability introduced by the Cre/Lox system, we implemented a predefined threshold for Mtr expression, excluding mice exceeding this threshold.

All animal handling and procedures strictly adhered to the guidelines of the National Institute of Health Guide for the Care and Use of Laboratory Animals. The experiments were conducted in an accredited facility (Institut National de la Santé et de la Recherche Médicale, U1256), fully complying with the EU Directive 2010/63/EU and the French governmental decree 2013-118, under authorization number Apafis #12851.

Method details

RNA extraction and quantitative RT-PCR analysis

A total of 500 nanograms of RNA extracted from 4 hippocampus and frontal cortex tissues for each group were employed in a two-step RT-qPCR process, following the recommended protocols of the Nucleospin RNA plus kit (Macherey-Nagel). For reverse transcription, PrimeScript RT Master Mix and SYBR Premix Ex Taq (Takara, Kusatsu, Japan) were utilized. The specific primers were sourced from Eurogentec (Angers, France). The cycle threshold (Ct) values were determined for each sample, and the gene expression of interest was subsequently normalized to the expression of Tbp and Pol2 genes using the 2-ΔΔCt method.

Protein extraction and western blot/Wes Simple Protein analyses

4 to 5 Samples of hippocampus and frontal cortex for each group were frozen in liquid nitrogen and subsequently solubilized in RIPA lysis buffer for protein expression analysis. The lysis buffer was supplemented with 1% phenylmethanesulfonyl fluoride (Sigma), 1% Sodium orthovanadate (Sigma), and 0.5% Phosphatase Inhibitor Cocktail (Roche) to ensure proper protein preservation. Protein concentrations were determined using the BCA Protein Assay kit (Interchim), following the manufacturer’s guidelines. Due to the low total protein yield from the hippocampus, protein analysis was conducted using either the Capillary Western blot system (Wes Simple Protein, ProteinSimple, USA) or, when required, traditional western blot experiments.

The western blot experiments were conducted in accordance with previously described protocols.¹⁵ In these experiments, membranes were incubated overnight at 4°C with specific primary antibodies. Peroxidase-labeled anti-rabbit (711-035-152, Interchim) or anti-mouse (715-035-152, Interchim) secondary antibodies were applied at a 1:5000 dilution. The total protein per lane was normalized using alpha-tubulin (ab185067, Abcam) or beta-actin (ab197277, Abcam) and band intensity densitometry analysis was conducted using ImageJ v1.53.

For the assessment of protein expression, the Wes automated capillary-based size sorting system (ProteinSimple, USA) was also employed with previously described protocol.^22,23 Briefly, a total of 0.4 μg/μL of protein was loaded onto WES 25-well plates for separation (Protein Simple, USA), following the manufacturer’s instructions. Primary antibodies were diluted at 1:75. The total protein per lane was normalized using alpha-tubulin (2144, Cell Signaling), beta-actin (#4970, Cell Signaling), or vinculin (13901S, Cell Signaling). HRP-labeled anti-rabbit or anti-mouse secondary antibodies (ProteinSimple, USA) were used at recommended concentrations. The relative protein quantities were determined by measuring the areas under the peaks in the chemiluminescence chromatograms, employing Compass for SW software v6.0.0 (ProteinSimple, USA). The results are presented in the form of virtual blots generated by Compass for SW software.

LC-MS/MS analyses

Hippocampal tissue was subjected to a rigorous preparation protocol for subsequent analysis. Initially, mechanical homogenization was performed on the tissue using a pellet pestle motor (Kontes) within a medium of 1× PBS. Subsequently, the homogenate underwent ultrasonication for 20 min while being maintained in a cold bath at 4°C. Following ultrasonication, the homogenate was centrifuged at 12,000g for 20 min, resulting in the collection of the supernatant, which was preserved for LC-MS/MS analysis. In summary, the collected supernatants were combined with internal standards that included Dithiothreitol (DTT). Protein precipitation was achieved by introducing cold methanol and allowing the mixture to incubate on ice for 30 min. Following this, the liquid phase was diluted using a 4-fold volume of a 0.1% formic acid solution.

Analytical procedures were conducted using the Shimadzu LCMS 8045 ESI Triple Quadrupole system (Kyoto, Japan). Data analysis was performed using Insight software version 3.1 (Shimadzu). To ensure robust results, samples with concentrations hovering near the detection limits of the instrumentation were excluded, guaranteeing reliable concentration detection well within the detection limit. The metabolomic heatmaps were generated using R (v4.3) with the Pheatmap package.

To illustrate the differences in metabolite levels between groups, we performed differential analysis using the log₂ fold change (KO vs. WT) for each metabolite. Adjusted p-values were computed using multiple two-tailed unpaired Student’s T-tests with correction for multiple comparisons (e.g., Benjamini-Hochberg method). The results were visualized in a barplot generated in R using the ggplot2 package. Each bar represents the log₂ fold change, with its color intensity reflecting the magnitude of change. Annotated adjusted p-values were displayed alongside each metabolite for clarity.

Oxygraphy analysis

High-resolution respirometric measurements, specifically mitochondrial oxygen consumption assessments, were conducted using the Oxygraph-2k system developed by Oroboros Instruments. To facilitate these measurements, we employed the OROBO-POS polarographic oxygen sensor to continuously monitor oxygen levels within the Oxygraph chamber. For each studied group, five hippocampal samples were utilized. These tissues, having been previously weighed for normalization, were carefully introduced into a chamber of the Oxygraph and mixed with mitochondrial respiration medium (MIR05). The respirometry procedure was executed with ongoing agitation, achieved using an electromagnetic stir bar, and was maintained at room temperature. The oxygen consumption rates were quantified and expressed in units of picomoles per second per milligram (pmol/(s∗mg)). It is worth noting that higher oxygen consumption rates are indicative of a more active respiratory chain within the mitochondria.

In Silico simulation

To better characterize the Mtr-cKO model, we performed an in-silico simulation of the excision of exons 4 and 5 from the Mtr gene. The coding sequence of Mtr (Ensembl ID: ENSMUSG00000021311) was retrieved, and the deletion was simulated computationally by concatenating exons 3 and 6 to mimic the splicing effect. Translation of the spliced mRNA sequence revealed a frameshift resulting in a premature stop codon at position 192, within the homocysteine-binding domain (amino acids 6–326). This disruption renders the protein non-functional, as downstream domains essential for methionine synthase activity, including the SAM-binding domain, are absent. Protein domain annotations were confirmed using UniProt and Pfam, emphasizing the critical impact of the truncation. These findings validate the functional loss of methionine synthase and its role in metabolic disruption in the Mtr-cKO model.

Multi-omics and bioinformatics analyses

Methylome analysis

For methylome studies, each group consisted of six hippocampus samples obtained from 21-day-old mice. The analysis was conducted following the established protocol described previously.²² Briefly, Genomic DNA was extracted using Qiagen’s QIAamp DNA Kits following the manufacturer’s instructions. Tissue lysis was performed at 56°C overnight, and the eluted DNA was collected in a fifty μL elution buffer. DNA concentration and quality were assessed using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen) and the TapeStation instrument (4150 Agilent) with the genomic DNA ScreenTape Assay kit (Agilent). For library preparation, bisulfite conversion, and amplification, 100 ng of DNA was used, following Diagenode’s Premium RRBS Kit guidelines (Liege, Belgium). We used the methylated and unmethylated spike-in controls included in RRBS Diagenode kit to estimate the bisulfite conversion efficiency. The results indicated no over-conversion as the conversion rate of the methylated spike-in was below 2% and no under-conversion as the conversion rate of the unmethylated spike-in was above 98%. Every library pool was quantified using Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen), and the average size of DNA fragments was estimated with a 2100 Bioanalyzer instrument (Agilent) using High Sensitivity DNA Kit (Agilent). Every pool was then denatured in NaOH and diluted at 1.8 p.m. with 20% of PhiX Control v3 (Illumina). Sequencing was carried out on a NextSeq 550 machine (Illumina) using the NextSeq 500/550 High Output v2 Kit (Illumina), 75 cycles in single-end mode. After demultiplexing, Fastq files containing all the sequencing reads per sample were generated and used in downstream bioinformatics pipelines. Quality control was performed with FastQC v0.11.5. Adapters, 5′ and 3′ adjustments for possible end-repair bases and low-quality bases were removed using Trim Galore! v0.6.6 (FastQC) and Trim Galore! was downloaded from https://www.bioinformatics.babraham.ac.uk/. Bisulfite-treated reads were then aligned (using Bowtie v2.4.4) to the GRCm38/mm10 mouse reference genome followed by methylation calling using Bismark v0.22.3 22 (parameters for the mapping step: – non_bs_mm –bam –nucleotide_coverage; parameters for the methylation calls: –cytosine_report –comprehensive –merge_non_CpG).

Differential methylation analysis between Mtr-cKO and WT groups was performed using R 4.1 (RStudio v1.4.1106) and the MethylKit v1.16.1 package. The dataset was filtered for low coverage (CpGs with coverage below 10×) and extremely high coverage, normalized, and merged. Differential CpGs were identified with FDR correction (Benjamini–Hochberg), and DMR (identified within a window size of 2000 bp) were annotated using the mouse GRCm38/mm10 genome with RefSeq curated (NCBI) and GENCODE VM22 (Ensembl) databases. Enrichment analyses were performed on CpG and DMR with specific methylation changes, and the results were visualized on volcano plots.

Proteomics

Total protein extracts were prepared from tissue homogenates using RIPA lysis buffer, followed by sonication and three freeze–thaw cycles. Lysates were centrifuged at 20,000 g for 30 min at 4°C, and the clarified supernatants were collected. Protein concentration was quantified using the BCA assay (Interchim). Protein extracts (60 μg) from wild-type and Mtr-cKO mice were precipitated with acetone at −20°C for 3 h, then incubated with sequencing-grade trypsin overnight at 37°C. Peptides were desalted using ZipTip μ-C18 Pipette Tips (Millipore) and analyzed using an Orbitrap Fusion Tribrid mass spectrometer with Easy nano-LC Proxeon 1000 liquid chromatography. Chromatographic separation was achieved using C18 columns, with a solvent gradient from 95% solvent A (water, 0.1% formic acid) to 35% solvent B (acetonitrile, 0.1% formic acid) over 98 min. Peptides were analyzed in full ion scan mode at a resolution of 120,000, with HCD activation for peptide fragmentation and MS/MS data acquisition in top-speed mode. Label-free relative quantification was performed using Progenesis-Qi software 4.1 (Nonlinear Dynamics Ltd, Newcastle, UK). MS and MS/MS data were processed with Proteome Discoverer software (Thermo Scientific, v.2.2) and the Mascot search engine (Matrix Science, v.2.5.1), with mass tolerance set to 7 ppm for precursor ions and 0.5 Da for fragments. The SwissProt database (02/2017) with Mus musculus taxonomy was used for MS/MS identification, and peptide identifications were validated using a 1% FDR threshold with the Percolator algorithm. Protein abundance variations were measured with the Hi-3 label-free quantification method and validated with Anova p-values under 0.05.

ChIP-seq analysis

Samples were fixed with 1% paraformaldehyde followed by 0.125M glycine to stop crosslinking. Samples were lysed in freshly made sonication buffer (20mM HEPES, 1% SDS, 0.25 protease inhibitor cocktail (Sigma Aldrich)) and sonicated using the Bioraptor Pico (Diagenode). Fragmented chromatin was precleared using bovine serum albumin (BSA)- blocked protein A/G beads (Invitrogen) and 1.7 μg of H3K4me3 antibody (Diagenode) was added. Following overnight incubation, immune complexes were captured by incubation with BSA-blocked protein A/G beads for 1 h at 4C on a rotating wheel. Beads were washed with wash buffer 1 (2mM EDTA, 20mM tris (pH8), 150mM NaCl, 1 Triton X-100, and 0.1% SDS), buffer 2 (2 mM EDTA, 20 mM tris (pH 8), 150 mM NaCl, 1% Triton X-100, 0.1% SDS, and 500 mM NaCl), and buffer 3 (1 mM EDTA and 10 mM tris (pH 8)) for 5 min at 4C on a rotating wheel. Samples were eluted for 20 min at room temperature in 0.1 M NaHCO3 and 1% SDS. Eluted chromatin was de-crosslinked overnight at 65C using 10mg/mL proteinase K and 5M NaCL. De-crosslinked DNA was purified using MiniElute PCR Purification Kit (QIAGEN) and library preparation was performed using the KAPA HyperPrep kit (Roche) according to manufacturer’s instructions. Samples were sequenced using Illumina HiSeq 2500 sequencer [dual-indexed 150–base pair (bp) paired-end reads. FASTQ read were mapped against the mm9 reference genome using Bowtie2. High quality reads were filtered using SAMtools q > 15. Followed by PCR duplicate removal using PicardTools. Next, SAM files were converted to BAM using SAMTools. BAM files were converted to BED files using BEDTools, reads were extended to 200bp and converted to BedGraph using BEDTools. BedGraphs were converted into BigWig files using bedGraphToBigWig for visualization. Peak calling was performed using MACS2 and pvalue e−8. H3K4me3 read counts were computed using BedTools and the MACS2-identified peaks. For library size normalization and differential peak analysis, DESeq2 was performed. Differential peaks between conditions were defined using the following criteria: fold change >1.5, p-value <0.05 and minimum normalized reads per peak >40.

Bioinformatics analysis

In our study, we conducted detailed enrichment analyses for CpG sites, DMR, and proteins that exhibited significant changes in both methylation and expression levels. For CpG sites, significance was determined by a methylation difference exceeding 25% and a cumulative q-value below 0.01. DMRs were identified as significant if they included at least one significant CpG, showed a methylation difference of more than 15%, and had a cumulative q-value below 0.05. The circos plot, generated using Circlize R package, visually represents global significant methylation changes across the entire genome, highlighting the CpG methylation differences along each chromosome.

Proteins were considered significantly altered if they displayed a fold change of less than 0.9 or greater than 1.1. Raw p-values from two-tailed unpaired Student’s t-tests were used without correction for multiple testing due to the exploratory nature of this analysis. This approach was aimed at identifying potential trends, with key findings cross-validated using complementary methods to ensure their biological relevance.

To summarize our findings, we utilized the ShinyGO tool (v0.76.1) to perform enrichment analyses on the “Gene Ontology - Biological Process” and the “Panther Classification System.” We extracted the top ten significant terms based on an FDR below 0.05. To visually represent our data, a proteomic heatmap was generated using R (v4.3) with the Pheatmap package, and volcano plots were created for both methylation and proteomic analyses using R (v4.3) with the ggplot2 package.

To explore the network of protein-protein interactions among differentially expressed proteins, we used the STRING database (version 12.0, https://string-db.org/). STRING integrates known and predicted interactions from diverse sources, including experimental data, computational predictions, curated databases, and text mining (e.g., PubMed). For our analysis, we included the following interaction sources: experiments, databases, co-expression, gene fusion, and genomic neighborhood. The confidence score threshold was set at 0.400, reflecting a moderate level of interaction reliability.

To identify functionally related protein groups, we applied k-means clustering, which grouped the proteins into distinct clusters (Clusters A to E) based on their interaction patterns. The interaction networks were visualized using STRING’s graphical interface, where the thickness of edges indicates the confidence of each interaction. For enhanced visualization and integration with other data, interaction diagrams were exported and refined using Cytoscape (v3.10).

System biology metabolic modeling

Constraint-based Reconstruction and Analysis (COBRA) was used to analyze proteomic data from the hippocampus of four Mtr knockout and four control mice. Simulations were performed in MATLAB R2021b using the COBRA Toolbox and IBM CPLEX as the solver. A tissue-specific genome-scale model of the mouse brain, iBrain674-Mm, was used, representing generic neurons and astrocytes. Personalized genome-scale models were generated by mapping proteomic data to gene symbols and NCBI Gene IDs, excluding proteins not present in the model. A total of 313 proteins were mapped onto iBrain674-Mm. For each sample, the mapExpressionToReactions function assigned weights to each reaction, with reactions not linked to a protein set to 0. Reactions encoded by the Mtr gene were weighted differently for wild-type (100) and Mtr-cKO samples (20). The INIT algorithm used the mapped reactions as input. The E-Flux algorithm scaled fluxes proportionally to protein expression, with maximum forward and reverse fluxes set to ±1000 mmol∗g dry weight-1∗hr-1. Flux variability analysis computed the minimal and maximal fluxes for all reactions. CO2, H2O, and oxygen uptake rates were set to 10 mmol∗g dry weight-1∗hr-1, while other metabolites were set to 1 mmol∗g dry weight-1∗hr-1. The Wilcoxon rank-sum test identified significantly altered fluxes between Mtr knockout and control mice, with a p-value below 0.05 after FDR correction.

Immunofluorescence cell imaging

Immunofluorescence analyses were performed on 10μm sagittal brain sections obtained from a microtome, starting from the zero-plane bisecting the brain mid-sagittally. Brain structures were identified based on the Paxinos and Watson atlas for standardized slide positioning according to specific coordinates.²³ Tissue, previously fixed in Carnoy’s solution, was subsequently dehydrated and embedded in paraffin using the Leica ASP300-S following standard procedures. The paraffin-embedded sagittal brain sections were de-waxed using Histo Clear (National Diagnostics, HS-200) and rehydrated through a standard protocol. Antigen retrieval was achieved by incubating the brain slices at 95°C for 30 min in a 0.1M citrate buffer solution (pH 6). To prevent non-specific binding, brain sections were blocked with a solution containing 5% fetal bovine serum and 5% BSA (v/v) in PBS for 2 h. Subsequently, they were incubated with primary antibodies. Slides were further incubated with secondary antibodies (anti-rabbit or anti-mouse from Abcam) at a 1:1000 dilution for 2 h at room temperature. To visualize the nuclei, the nuclear fluorescent dye 4,6-diamidino-2-phenylindole (Sigma) was employed in PBS. Images were captured using a Nikon Instruments confocal microscope.

Behavioral testing

To assess learning performance, we employed a water maze test post-weaning on untreated and treated mouse groups.²³ Behavioral monitoring was conducted using a high-standard video-tracking system (Viewpoint, Lyon, France) to ensure consistency across runs. The setup consisted of a square pool filled with water (maintained at 25°C, 5 cm deep). The pool was divided into 25 equal square zones (15 × 15 cm) using gray plastic walls, with openings providing pathways between zones to establish a direct route from the starting area to the exit. Other zones allowed for potential errors. Each mouse was tested twice daily over five consecutive days (from session S1 to S5), with a 2-min time limit per trial. Learning success was measured by the percentage of successful escapes, defined as reaching the exit within the time limit and with a low number of errors (comparable to the WT group). Motor function was assessed by measuring velocity during each trial to ensure no significant differences across groups. Two habituation sessions were conducted two days before the first test session to reduce stress and variability. These sessions allowed mice to familiarize themselves with the setup, but no data was recorded. Statistical analysis was performed to evaluate learning performance. The percentage of successful or failed escapes was analyzed using a Chi-square test with Yates' correction, appropriate for categorical data. Velocity data were analyzed using independent t-tests to ensure comparability of motor function. A significance level of p < 0.05 was used to determine statistically meaningful differences. Each group consisted of 9–14 mice, with group sizes determined based on prior studies and the availability of animals, ensuring adequate statistical power. To control potential confounding variables, we implemented several measures including habituation sessions (these reduced anxiety-related behaviors, such as freezing, and variability in the data), environmental standardization (the testing environment was maintained with consistent lighting, water temperature (25°C), and minimal noise to prevent external distractions), randomization and blinding (group assignments were randomized, and data analysis was conducted by experimenters blinded to treatment and genotype to eliminate observer bias), health monitoring (mice showing signs of illness, physical impairments, or excessive stress were excluded from the study before testing).

Quantification and statistical analysis

Statistical analyses were meticulously conducted using GraphPad Prism version 10.1.0. Continuous variables, such as densitometry results, were presented as scatter dot plot that extend from minimum to maximum values, with a horizontal line indicating the mean with SEM. Data points were displayed to show the distribution within groups, providing a measure of data variability and central tendency. Pathway diagrams and schematics are created with BioRender.com (agreement number DY26V1832S). Group comparisons between wild-type and Mtr-cKO mouse models were performed using an unpaired two-tailed Student’s t test. Prior to these tests, data were assessed for normality and homogeneity of variances to ensure the appropriateness of the t test. A significance threshold of p < 0.05 was set to establish statistical significance. For analysis involving multiple groups comparisons, a two-way ANOVA was employed to explore the interaction effects and main effects, followed by a post hoc Tukey test for pairwise comparisons among the groups. A significance level of p < 0.05 was used to denote statistically meaningful differences.

Correlation analyses were performed to investigate relationships between multiple variables using Spearman’s rank correlation coefficient, suitable for assessing monotonic relationships or using Pearson Rho correlation to investigate relationships between two variables. For multiple variables correlation, results were visually represented in a correlation plot crafted using R with the ggplot2 package, including both the p-value and the correlation coefficient to clearly depict the strength and significance of the relationships observed.

Published: February 25, 2026