Female 3xTg-AD mice (stock no. 34830-JAX) and age-matched C57BL/6J (wild-type, WT) controls were obtained from the Jackson Laboratory (Bar Harbor, ME). To control for variability due to sex and age, only female mice of the same age were used across both genotypes. Each experimental group consisted of three biological replicates (n = 3 per group). Mice were housed in a temperature-controlled environment under standard laboratory conditions and maintained on a standard chow diet with ad libitum access to food and water. At 11 months of age, mice were euthanized, and brain tissue samples—including the cortex plus hippocampus— were harvested and immediately processed for RNA-sequencing (RNA-seq) analysis.

Brain cortex and hippocampus tissues were pooled, homogenized, and lysed in TRIzol reagent (ThermoFisher, GE17-0891-01). RNAs were extracted according to the manufacturer’s protocol then quantified using a Nanodrop (ThermoFisher). The isolated RNA samples were then sent to Genewiz (South Plainfield, NJ) for RNA-Seq analysis. Total RNA libraries were prepared using the Pico Input SMARTer Stranded Total RNA-Seq Kit (Takara). Briefly, 10 ng of total RNA from each sample was reversely transcribed using random priming and reverse transcriptase. Full-length cDNA was synthesized using SMART (Switching Mechanism At the 5’end of RNA Template) technology, which preserves the strand orientation of the RNA. Ribosomal cDNA was hybridized to mammalian-specific R-Probes and cleaved by ZapR. Libraries containing Illumina adapters with TruSeq HT indexes were pooled and loaded onto the Hiseq 2500 platform. Single-end reads of 75 base-pair (bp), with 30 million reads per sample, were generated for subsequent bioinformatic analysis. FASTQ files were mapped to the mouse mm10 genome using the STAR Aligner. Data analysis was performed using the statistical computing environment R, along with the Bioconductor suite of packages and RStudio. Raw data underwent background subtraction, variance stabilization, and normalization via robust spline normalization. Differentially expressed genes were identified using linear modeling and Bayesian statistics with the Limma package. For comparisons between two groups, a two-tailed Student’s t-test was used to assess statistical significance.

Transcriptomic datasets were obtained from the publicly accessible NIH-NCBI Gene Expression Omnibus (GEO) database.² Relevant datasets were systematically curated and organized for further analysis. Differential gene expression analysis was performed using the GEO2R tool, an interactive web application within the GEO database, which allow comparison between experimental groups and controls using pre-processed data.

Pathway enrichment analysis was performed using Metascape³, an integrated online tool designed for the functional annotation and interpretation of large-scale omics datasets. Differentially expressed genes (DEGs) identified from RNA-seq data were compiled and uploaded to the Metascape platform to investigate their associated biological pathways, molecular functions, cellular processes, and potential clinical relevance. The analysis incorporated gene ontology (GO) terms, KEGG pathways, Reactome pathways, and other functional categories to provide comprehensive insights into the biological significance of the DEGs.

To investigate whether the 3xTg-AD model induces transcriptomic alteration in brain EC vasculature within the cortex and hippocampus, we analyzed bulk RNA-seq data from brain tissues of 3xTg-AD mice and wild-type (WT) controls, as we recently reported (Saaoud et al., 2025). Differentially expression analysis revealed that 316 genes were significantly upregulated and 412 genes were downregulated in the 3xTg-AD group compared to WT controls, using a cutoff fold change > 1.5 and p-value < 0.05, as we previously reported (Saaoud et al., 2025). Principal component analysis (PCA) was conducted to evaluate sample variance and clustering, highlighting distinct transcriptional profiles between the two groups (Figure 1A). A curated list of 221 brain vasculature genes was obtained from the Human Protein Atlas (HPA) database⁴ for targeted analysis. As shown in Figure 1B, 3xTg-AD modulated the expression of brain EC vasculature genes, with four genes significantly upregulated. These include GIMAP4 (GTPase, IMAP family member 4), which promotes immune cell migration (Chen et al., 2022); TINAGL1 (tubulointerstitial nephritis antigen like 1), which promotes EC tube formation (Sato et al., 2022); FAM167B (family with sequence similarity 167 member B), an EC-specific gene (Feng et al., 2019); and CAV1 (caveolin 1), a gene involved in proatherogenic process, leukocyte influx, and CD4⁺ regulatory T cell (Treg) suppression (Engel et al., 2011). Conversely, several genes were significantly downregulated in the 3xTg-AD brain, including PTPRB (which regulates vascular permeability and homeostasis in inflammation) (Vestweber, 2021), CD248 (which induces a maladaptive unfolded protein response in diabetic kidney disease) (Krishnan et al., 2023), TBX2 (which defines a multipotent mesenchymal progenitor pool) (Wojahn et al., 2019), SLC38A5 (which promotes retinal vascular development and pathological neovascularization in a retinopathy model (Wang et al., 2022), TMEM102 [whose overexpression may promote chemoresistance by inhibiting the mitochondria-associated apoptotic pathway (Tai et al., 2022)], Hapln3 (a novel link protein that co-localizes with versican and is upregulated by platelet-derived growth factor in arterial smooth muscle cells) (Ogawa et al., 2004), GBP4 (which serves as a central orchestrator of immune responses to infection, inflammation, and cancer) (Tretina et al., 2019), and A2M (which neutralize cytokines and modifies hemostasis) (Lagrange et al., 2022). These findings suggest that AD brain promotes brain EC activation and inflammatory cell recruitment while inhibiting vascular development and pathological neovascularization.

To determine whether 3xTg-AD induces transcriptomic reprogramming of fibroblasts and fibrosis, we examined the expression changes of 401 fibroblast genes from the HPA database and 777 fibrosis-related genes (Gu et al., 2020) in the 3xTg-AD brain. As shown in Figure 1B, eight out of 401 fibroblast genes and 17 out of 777 fibrosis-related genes were significantly upregulated in the 3xTg-AD brain. Notably, 3xTg-AD also downregulated the expression of eleven fibroblast genes and 24 fibrosis-associated genes. Metascape pathway analysis of the 24 upregulated fibroblast and fibrosis-related genes identified the following key pathways: negative regulation of catalytic activity, regulation of hemopoiesis, blood circulation, negative regulation of cell differentiation, olefinic compound metabolic process, positive regulation of cell migration, lipid homeostasis, viral protein interaction with cytokine and cytokine receptor, and cardiocyte (cardiomyocyte) differentiation. Conversely, the key enriched pathways for the 3xTg-AD downregulated fibroblast and fibrosis-related genes included macrophage activation, macrophage differentiation, positive regulation of ROS metabolic process, phagosome, positive regulation of secretion by cell, response to angiotensin, regulation of membrane depolarization. Among the 17 upregulated fibrosis-related genes, FHIT (fragile histidine triad diadenosine triphosphatase), MPZ (myelin protein zero), CCR1 (C-C motif chemokine receptor 1), TCAP (titin-cap), PTGS2 (prostaglandin-endoperoxide synthase 2), and HSPA1B (heat shock protein family A (Hsp70) member 1B) were found to be expressed in multiple brain cell types, based on data from the Single Cell Portal at the MIT Broad Institute⁵ (Figures 1E–J). Thus, 3xTg-AD model also upregulated 6 fibrosis-related genes in multiple brain cell types in AD.

To determine whether 3xTg-AD model induces transcriptomic reprogramming in EndoMT, we analyzed the expression of 111 known EndoMT genes (Kovacic et al., 2019; Pardali et al., 2017; Piera-Velazquez and Jimenez, 2019) from the HPA database in our 3xTg-AD RNA-seq dataset. As shown in Figure 1B, 3xTg-AD modulated the expression of six EndoMT genes, with two upregulated genes—RGCC and VCAN—and four downregulated genes. RGCC (response gene to complement 32 protein, RGC-32) is known to regulate cell cycle progression and can be induced by p53 in response to DNA damage or by sublytic levels of complement system proteins, influencing cell cycle activity.⁶ RGCC is also stimulated by growth factors and cytokines, including transforming growth factor β (TGF-β), playing a role in the modulation of processes such as angiogenesis, fibrosis, cell migration, and differentiation (An et al., 2009). Notably, RGCC is essential for promoting the differentiation of T helper 17 (Th17) cells, which are implicated in multiple sclerosis, and in regulating significant transcriptomic changes in astrocytes, favoring the synthesis and secretion of extracellular matrix components, growth factors, axonal growth molecules, and pro-astrogliogenic factors (Tatomir et al., 2022). VCAN (Versican), an extracellular matrix (ECM) proteoglycan, is upregulated alongside other ECM-binding molecules, such as hyaluronan, tumor necrosis factor-α (TNF-α) induced protein 6 (TNFAIP6, TSG-6), and inter-alpha-trypsin inhibitor heavy chain 1 (ITIH1, IαI), during inflammation in various diseases, including cardiovascular, pulmonary, autoimmune diseases, and certain cancers. These interactions form stable scaffolds that act as “landing strips” for inflammatory cells as they migrate from the circulation into tissues (Wight et al., 2020). To validate the roles of RGCC and VCAN in mediating EndoMT in the brain, we examined their expression in multiple brain cell types in AD using single-cell RNA-seq data from the Single Cell Portal database at MIT-Broad Institute⁷, as shown in Figures 1K, L. Our data demonstrated that RGCC is expressed across multiple brain cell types, including ECs and vascular smooth muscle cells (VSMCs). These findings align with recent studies reporting that 14 out of 45 AD risk genes are upregulated within brain vasculature, particularly in brain ECs and VSMCs (Yang et al., 2022). Collectively, our data suggest that 3xTg-AD potentially promotes EndoMT through the upregulation of cytosol/nuclear-localized RGCC and ECM-localized VCAN.

To assess whether the 3xTg-AD model induces transcriptomic reprogramming of cell death pathways, we analyzed the expression changes of 2021 apoptotic cell death genes, based on data from the Mouse Genome Informatics (MGI) database.⁸ As shown in Table 1, the 3xTg-AD model upregulated 43 apoptosis-related genes and downregulated 36. Metascape pathway analysis revealed key pathways associated with the significantly upregulated genes, including the positive regulation of apoptotic process, regulation of apoptotic signaling pathway, regulation of EC proliferation, positive regulation of protein kinase A signaling, regulation of tumor necrosis factor production, regulation of extracellular signal-regulated kinase (ERK)1 and ERK2 cascade, cell activation involved in immune response, regulation of cellular response to stress, regulation of blood vessel EC migration. Conversely, the key pathways associated with downregulated genes included the positive regulation of apoptotic process, glial cell differentiation, positive regulation of cell migration, apoptotic signaling pathway, epithelial cell proliferation, cell chemotaxis, regulation of apoptotic signaling pathway, cellular response to lipid, regulation of cell-cell adhesion, and cellular response to angiotensin.

Ferroptosis, a newly identified form of programmed cell death, is triggered by imbalances in iron metabolism, iron-dependent lipid peroxidation, and ROS accumulation (Zhao et al., 2023). Glutathione (GSH) depletion and Glutathione peroxidase 4 (GPX4) inactivation are known to induce ferroptosis (Jiang et al., 2021). To determine whether the 3xTg-AD model affects ferroptosis, we analyzed the expression changes of 264 ferroptosis-related genes, collected from the FerrDB database⁹, among the 3xTg-AD modulated genes. As shown in Table 1, 19 ferroptosis-related genes were upregulated, including solute carrier family 1 member 5 (SLC1A5) (Su et al., 2023), DPEP1, a dexamethasone-sensitized ferroptosis gene (von Mässenhausen et al., 2022), cysteine dioxygenase type 1 (CDO1), which absorb cysteine competitively, thus restricting the process of GSH synthesis and promoting ferroptosis (Li et al., 2020), and acyl-CoA synthetase long chain family member 1 (ACSL1), which mediate ferroptosis triggered by conjugated linolenic acids (Beatty et al., 2021), and other 15 ferroptosis genes, while downregulated 7 ferroptosis-related genes, including ALOX12 [the inactivation of arachidonate 12-lipoxygenase, 12S type (ALOX12) reduces p53-driven ferroptosis triggered by ROS stress] (Chu et al., 2019), Toll-like receptor 4 (TLR4), high mobility group box 1 (HMGB1) [released by anti-inflammatory M2 macrophages, binds to TLR4 on M1 macrophages, leading to the activation of signal transducer and activator of transcription 3 (STAT3) signaling in proinflammatory M1 macrophages and promoting the inflammatory response] (Feng et al., 2024), suppressor of cytokine signaling 1 (SOCS1), neuroglobin (NGB), cytochrome B-245 beta chain (CYBB), and adenosylhomocysteinase (AHCY). The key pathways of upregulated ferroptosis genes included peroxisome proliferator activated receptor (PPAR) signaling pathway, regulation of small molecule metabolic process, positive regulation of lipid localization, response to metal ion, organic acid catabolic process, cell surface receptor protein tyrosine kinase signaling pathway, however, the downregulated pathways included positive regulation of tumor necrosis factor production and positive regulation of apoptotic process.

Moreover, we identified 299 cell death regulators across 13 different types of cell death pathways, collectively termed the cell death regulatome (Galluzzi et al., 2018; Wang et al., 2019). As shown in Table 1, eight genes were upregulated in six other cell death pathways, including acetyl-CoA acyltransferase 2 (ACAA2) in the mitochondrial outer membrane permeability programmed cell death group, translocator protein (TSPO) and caveolin 1 (CAV1) in the programmed necrotic cell death group, TSPO and ATP5G2 in the mitochondrial permeability transition (MPT)-driven necrosis group, 2,4-dienoyl-CoA reductase 1 (DECR1) in the neutrophil extracellular traps (NETotic) cell death group, polo like kinase 1 (PLK1) in the mitotic cell death group, and CAV1 in the anoikis group. Notably, CAV1 was associated with three cell death categories (apoptosis, programmed necrotic cell death, and anoikis), while TSPO was linked to two categories (programmed necrotic cell death and MPT-driven necrosis). Additionally, two genes downregulated in two other cell death pathways, including transmembrane protein 102 (TMEM102) in the mitochondrial outer membrane permeability programmed cell death group, and HMGB1 in the mitochondrial permeability transition (MPT)-driven necrosis group.

The 3xTg-AD brain also downregulated GBP4 (guanylate binding protein 4) (Figure 1B), which has been shown by us and others to regulate cytosolic recruitment and activation of pro-inflammatory caspase-4 (Drummer et al., 2021b; Wandel et al., 2020). Caspase-4 (human)/caspase-11 (mouse) plays a crucial role in the non-canonical inflammasome activation, leading to inflammatory cell death (pyroptosis) as previously reported (Drummer et al., 2023; Hagar and Miao, 2014; Sun et al., 2024). Our findings suggest that caspase-4/caspase-11 activation in brain cells including ECs may be involved in AD, which was further supported by Wyss-Coray’s team in their 2022 Nature publication (Yang et al., 2022), demonstrating the involvement of microglia and brain ECs in this process. These results indicate that the 3xTg-AD model induces caspase-4/11 activation and pyroptosis in both the AD brain and brain ECs.

Taken together, our findings reveal that the 3xTg-AD brain upregulates 43 apoptosis-related genes, 19 ferroptosis-related genes, one mitochondrial outer membrane permeabilization cell death gene, two programmed necrotic cell death genes, two MPT-driven necrosis genes, one NETotic cell death gene, one mitotic cell death gene, and one anoikis gene. This suggest that the 3xTg-AD model promotes apoptosis, ferroptosis, pyroptosis, and six other types of cell death in the AD brain (Galluzzi et al., 2018; Wang et al., 2019). Additionally, genes such as RGCC (an EndoMT and apoptotic gene), cysteine rich protein 1 (CRIP1, an apoptotic gene), dopamine beta-hydroxylase (DBH, an apoptotic gene), DNA damage-inducible transcript 4 (DDIT4, associated with both apoptosis and ferroptosis), prostaglandin-endoperoxide synthase 2 (PTGS2, involved in both apoptosis and ferroptosis), perilipin 4 (PLIN4, a ferroptotic gene), caspase-4/caspase-11 (casp4, a pyroptotic gene) (Sun et al., 2024), interleukin-1β (IL1β pyroptotic cytokine) (Sun et al., 2024), and TNF-α (TNF-α involved in apoptosis, necroptosis, and trained immunity) (Saaoud et al., 2023; Wang et al., 2019) are expressed in multiple brain cells in AD (Figures 1K, I, and Supplementary Figures 1A–G). The 3xTg-AD model thus demonstrates upregulation of several types of cell death in brain cells, including apoptosis, ferroptosis, necroptosis, pyroptosis, programmed necrotic death, and anoikis.

To explore the potential mitochondrial mechanisms underlying EndoMT, cell death, and fibrosis in the 3xTg-AD brain, we analyzed transcriptomic reprogramming within the nuclear genome-encoded mitochondrial transcriptome. As shown in Figure 2A, the 3xTg-AD brain upregulated 21 mitochondria-localized protein encoding genes, collected from the HPA database, 26 mitoCarta genes (an inventory of mammalian mitochondrial proteins and pathways, MIT-Broad Institute), and 9 genes associated with mitochondrial genetic diseases and bioenergistic metabolism (Frazier et al., 2019). These upregulated genes overlapped as shown in Figure 2B. Metascape pathway analysis of these upregulated mitochondrial genes identified the top enriched pathways (Figure 2C), including fatty acid (FA) beta oxidation (FAO), FAO using acyl-CoA dehydrogenase, mitochondrial FAO of unsaturated FAs, FA biosynthesis, PPAR signaling pathway, response to cold, lipid transport, regulation of small molecule metabolic process, and muscle structure development. Notably, three mitochondrial localized genes— solute carrier family 6 member 2 (SLC6A2), MAF BZIP transcription factor F (MAFF), troponin C1, slow skeletal and cardiac type (TNNC1)—and three mitoCarta genes— haloacid dehalogenase like hydrolase domain containing 3 (HDHD3), fragile histidine triad diadenosine triphosphatase (FHIT), interferon alpha inducible protein 27 (IFI27)—were upregulated across multiple cell types in AD brain¹⁰ (Figure 1E and Supplementary Figures 2A–E). In summary, the 3xTg-AD brain induces mitochondrial stress by upregulating 32 mitochondrial genes in brain.

To investigate the potential nuclear stress mechanisms underlying EndoMT, cell death, and fibrosis in the 3xTg-AD brain, we analyzed transcriptomic reprogramming in the nuclear gene transcriptome induced by 3xTg-AD. As shown in Figure 3A, the 3xTg-AD brain upregulated one nuclear membrane protein gene, 13 nucleolar genes, and 70 nucleoplasm genes. Additionally, the 3xTg-AD brain upregulated eight DNA damage response genes, four TFs [early growth response 3 (EGR3), FosB proto-oncogene, AP-1 transcription factor subunit (FOSB), MAFF, and recombination signal binding protein for immunoglobulin kappa J region like (RBPJL)], and two differentiation TFs (Ng et al., 2021), FOSB and mesenchyme homeobox 1 (MEOX1). Metascape pathway analysis of these upregulated nuclear genes revealed the key enriched pathways, including sulfur amino acid metabolic process, muscle tissue development, regulation of EC proliferation, fatty acid beta oxidation, response to glucocorticoids, regulation of cyclin-dependent protein serine/threonine kinase activity, positive regulation of DNA-binding TFs, lipid homeostasis, DNA-templated transcription, and others (Figure 3B).

To determine whether the upregulation of FOSB and MEOX1 contributed to the expression of 316 AD-upregulated genes, we collected FOSB target genes from the MotifMap dataset¹¹, FOSB overexpression-induced genes (the dataset ID: GSE111827), and MEOX1 knockdown-downregulated genes (GSE133927) from the NIH-NCBI GEO database². As shown in Figures 3C–E, 13 FOSB target genes, 10 FOSB overexpression-induced genes, and nine MEOX1 knockdown-downregulated genes overlapped with the 3xTg-AD upregulated genes in the brain. To explore the role of FOSB and MEOX1 in 3xTg-AD brain cells, we observed that both TFs were upregulated in multiple cell types in AD brain(see text footnote 10) Figures 3F, G. Notably, two additional TFs—EGR3 (also associated with apoptosis) and MAFF (also linked to mitochondrial-localization)—along with four nucleolar genes, including FHIT (also a mitoCarta gene), Oncostatin M Receptor (OSMR) (also a cellular response to stimulus gene), RGCC (an EndoMT and apoptotic gene), and HDHD3 (a mitoCarta gene), as well as three DNA damage response genes, cysteine rich protein 1 (CRIP1), DNA damage inducible transcript 4 (DDIT4), and Egl-9 family hypoxia inducible factor 3 (EGLN3) (all associated with apoptosis), were also upregulated in AD brain cells, as indicated by the same AD single-cell RNA-seq dataset (Figures 1E, K and Supplementary Figures 1A, C, 2B, D, 3A–C). In summary, these findings demonstrate that the 3xTg-AD brain upregulates EndoMT and cell death pathways in brain cells by enhancing the expression of nuclear genes, DNA damage response genes, TFs, and differentiation TFs FOSB and MEOX1.

To investigate the potential cellular stress mechanisms driving EndoMT, cell death, and fibrosis in the 3xTg-AD brain, we examined the transcriptomic reprogramming induced by 3xTg-AD related to cellular and oxidative stress as well as ER stress. As shown in Table 2A, 3xTg-AD mice displayed upregulation of 95 genes related to cellular response to stimulus, 32 genes related to cellular response to stress, six genes associated with oxidative stress, and seven ROS regulator genes. Metascape pathway analysis of these upregulated genes (log₁₀P < −3) revealed the key enriched pathways, including brown fat cell differentiation, fatty acid metabolic process, PPAR signaling pathway, regulation of small molecule metabolic process, response to hypoxia, regulation of extracellular signal-regulated kinase 1 (ERK1) and ERK2 cascade, positive regulation of lipid localization, lipid homeostasis, cellular response to hormone stimulus, and response to TNF.

To explore potential mechanisms by which ER stress contributes to EndoMT, cell death, and fibrosis in the 3xTg-AD brain, we further analyzed transcriptomic reprogramming induced by 3xTg-AD across nine categories of ER stress-related genes. As shown in Table 2B, this analysis revealed: (i) upregulation of five ER-localized protein genes, (ii) upregulation of three ER processing protein genes, (iii) upregulation of four genes involved in the response to ER stress, (iv) upregulation of two AD-promoting genes (Bell et al., 2016) associated with protein kinase R-like ER kinase (EIF2AK3, also known as PERK, gut microbiota generated uremic toxin trimethylamine N-oxide, TMAO (Chan et al., 2019; Saaoud et al., 2023; Sun et al., 2018) receptor (Chen et al., 2019) interaction proteins, (v) upregulation of five genes associated with AD-suppressing (Zhang et al., 2022) activating transcription factor 6 (ATF6) interaction proteins, (vi) upregulation of two genes associated with AD- exacerbating (Duran-Aniotz et al., 2017) endoplasmic reticulum to nucleus signaling 1 (IRE1) interaction proteins, (vii) upregulation of four genes in the ER associated protein degradation (ERAD) pathway (Krshnan et al., 2022), (viii) upregulation of 22 genes induced by the ER stressor thapsigargin (a plant-derived inhibitor of calcium pumps), and (ix) upregulation of 14 genes induced by the ER stressor tunicamycin (an antibiotic produced by the bacterium streptomyces lysosuperificus). Metascape pathway analysis of these upregulated ER stress genes identified the key enriched pathways (log₁₀P < −3), including positive regulation of cold-induced thermogenesis, regulation of small molecule metabolic process, PPAR signaling pathway, brown fat cell differentiation, cellular response to hormone stimulus, chaperone cofactor-dependent protein refolding, regulation of carbohydrate metabolic process, metabolism of vitamins and cofactors, triglyceride metabolic process, and regulation of cysteine-type endopeptidase activity involved in apoptotic process.

To further elucidate the role of PERK—one of the three key ER stress sensors—in the pathophysiology of AD in the brain of 3xTg mice, we analyzed two transcriptomic datasets from PERK knockout (KO) models, collected from the NCBI GEO database. As shown in Table 2C, the AGBL4 gene was significantly downregulated in the PERK KO group compared to the wild-type (WT) controls. Additionally, 22 genes were downregulated in the PERK KO group following tunicamycin treatment, relative to tunicamycin-treated WT samples. Notably, oxidative stress gene PTGS2 (prostaglandin-endoperoxide synthase 2), DDIT4 (ROS regulator, also upregulated by thapsigargin, tunicamycin, and PERK KO with tunicamycin treatment), and several other ER stress genes upregulated by PERK KO and tunicamycin treatment—such as VCAN (EndoMT gene), eukaryotic translation initiation factor 4E binding protein 1 (EIF4EBP1, cellular stress gene), perilipin 4 (PLIN4), growth arrest and DNA damage inducible gamma (GADD45G), neuronal pentraxin 2 (NPTX2), EGR3—were also upregulated across multiple brain cell types in AD (Figures 1I, L and Supplementary Figures 1C, D, 3A, 4A–C).¹⁰ Taken together, our findings demonstrate that the 3xTg-AD brain promotes EndoMT, cell death, and fibrosis in the brain and brain cells by upregulating genes related to cellular stress, oxidative stress, and ER stress.

Mitochondrial abnormalities have been reported in patients with AD, and the roles of full-length APP, and the roles of soluble APPa on mitochondrial bioenergetic metabolism have also been reported (Lopez Sanchez et al., 2019). To investigate the potential metabolic reprogramming mechanisms (via metabolites-induced post-translational modifications of histones) as we reported (Drummer et al., 2021a; Saaoud et al., 2023; Saaoud et al., 2024; Xu et al., 2024) underlying EndoMT, cell death, and fibrosis in the 3xTg-AD brain and brain cells, we examined the transcriptomic changes in metabolic pathways induced by the 3xTg-AD model. As shown in Table 3, we identified the upregulation of 25 lipid and lipoprotein metabolism genes, 8 fatty acid oxidation (FAO) genes, 7 glucose metabolism genes, and 8 OXPHOS genes in the 3xTg-AD brain. Additionally, we observed downregulation of eight lipid and lipoprotein metabolism genes, six glucose metabolism genes, and four OXPHOS genes in the 3xTg-AD brain, which aligns with previous reports of age-dependent declines in mitochondrial complex II activity in a familial AD mouse model (Emmerzaal et al., 2018), as well as reduced cytochrome c oxidase activity in AD (Morais et al., 2021). Metascape pathway analysis of AD-upregulated lipid and lipoprotein metabolism genes identified the key enriched pathways (log₁₀P < −3), including fatty acid beta oxidation, metabolism of lipids, PPAR signaling pathway, triacylglyceride synthesis, and lipid localization. The key enriched pathways associated with AD-upregulated FAO genes included metabolism of lipids, negative regulation of catalytic activity, and membrane organization. For AD-upregulated glucose metabolism genes, the top enriched pathways included glucose metabolic process, regulation of small molecule metabolic process, small molecule catabolic process, and regulation of purine nucleotide metabolic process. Notably, several genes—including LPL (lipoprotein lipase, a lipid and lipoprotein metabolism gene), PTGS2 (lipid and lipoprotein metabolism and OXPHOS gene), and APP (amyloid beta precursor protein, OXPHOS gene)—were also upregulated among multiple cell types in AD brain (Figure 1I and Supplementary Figures 5A, B).¹⁰ Taken together, our findings demonstrate that the 3xTg-AD model promotes EndoMT, cell death, and fibrosis pathways in the brain and brain cells by upregulating genes involved in lipid and lipoprotein metabolism, glucose metabolism, and OXPHOS.

Single-cell RNA seq datasets from ischemic stroke, hemorrhagic stroke, and AD models have been integrated to construct metabolic flux profiles at the single-cell level. These three disorders induce shared metabolic shifts in ECs, with altered metabolic modules primarily enriched in transporter-related pathways. These shifts are predicted to potentially reduce metabolites such as pyruvate and fumarate. Additionally, Lef1 (lymphoid enhancer binding factor 1, Elk3 (ETS transcription factor), and Fosl1 (FOS like 1, AP-1 transcription factor subunit) may function as upstream transcriptional regulators driving these metabolic changes (Guo et al., 2023). To explore the potential catabolic reprogramming mechanisms underlying EndoMT, cell death, and fibrosis in the 3xTg-AD brain and brain cells, we examined transcriptomic reprogramming of catabolic process induced by the 3xTg-AD model. As shown in Table 4, we identified the upregulation of 63 catabolic genes and the downregulation of 27 catabolic genes. Metascape pathway analysis of the upregulated catabolic genes revealed the key enriched pathways, including lipid catabolic process, regulation of small molecule metabolic process, glycerolipid catabolic process, PPAR signaling pathway, regulation of lipid localization, positive regulation of cold-induced thermogenesis, mitochondrial FAO of unsaturated fatty acids (FAs), FAO using acyl-CoA dehydrogenase, regulation of sequestering of triglyceride, and regulation of ATP metabolic process. Metascape pathway analysis of the downregulated catabolic genes identified the top five pathways, including carbohydrate derivative catabolic process, small molecule catabolic process, positive regulation of catabolic process, regulation of type II interferon production, and carboxylic acid transport. Notably, six AD-upregulated catabolic genes, including DBH (dopamine beta-hydroxylase), ADA (adenosine deaminase), FHIT (fragile histidine triad diadenosine triphosphatase), HSPA1B (heat shock protein family A (Hsp70) member 1B), and VIP (vasoactive intestinal peptide) were upregulated across multiple cell types in AD brain (Supplementary Figure 6).¹⁰ In conclusion, our findings demonstrate that the 3xTg-AD brain upregulates EndoMT, cell death, and fibrosis pathways in brain and brain cells by enhancing the expressions of catabolic genes to a greater extent than downregulating them.