Persistent BBB dysfunction with increased paracellular barrier permeability is present in chronic phases after stroke, questioning the recovery of the BBB (Fig. 1). To investigate the molecular mechanisms contributing to BBB recovery and detect potential causes of the limited BBB recovery, we profiled the DNA methylome and transcriptome of isolated microvessels from ischemic hemispheres 7 days after the induction of thromboembolic (TE) stroke via injection of thromboembolic suspension in two experimental groups: young mice aged 6 months and old mice aged 18 months. A diagram of the experimental flow is shown in Fig. 1.
DNA methylome profile of the poststroke BBB recovery
Global DNA methylation, determined by the percentage of 5-mC content, was not significantly altered following TE stroke in 6-month-old mice, although the TE stroke group showed slight hypomethylation comparing with sham control (control) group (Fig. 2a). As global methylation only captures large-scale changes to the DNA methylome, CpG-level changes were investigated through reduced representation bisulfite sequencing (RRBS), comparing the DNA methylome of brain microvessels from post-TE stroke mice to age-matched controls. During poststroke BBB recovery, we detected 9,818 differentially methylated regions (DMRs, 1000 bp in size) with a percent change in methylation greater than 25% and a q-value < 0.01, with 4,638 hypermethylated and 5,180 hypomethylated DMRs (Fig. 2b). DMRs were primarily intronic (48.2%) and exonic (18.7%), while only 8.9% of DMRs were located in promoter regions (Fig. 2c). Strikingly, less than 1% of DMRs resided within CpG islands (Fig. 2c).
Categories of differentially methylated genes in post-stroke BBB recovery were identified through gene over-representation analysis. Overall, post-TE stroke changes to the DNA methylome largely affected genes encoding cell structural proteins (e.g., cell junction, and cell polarity, actin cytoskeleton, extracellular matrix, membrane microdomain), transporters and channels (e.g., potassium, organic anion and inorganic cation, calcium ion transport), and proteins involved in endothelial cell processes (e.g., angiogenesis and vasculogenesis, cell signaling and transcription regulation) (Fig. 2d, 2e).
Due to their well-defined function in the context of transcription, DMRs within gene promoters and non-promoter regions were individually assessed, with non-promoter regions including the 5’ and 3’ untranslated regions (UTRs), introns, exons, and 1-5kb upstream of transcription start sites. Out of the 1,020 DMRs in gene promoters, 474 DMRs were hypermethylated and 546 DMRs were hypomethylated. (Fig. 3a). There were 8,798 non-promoter DMRs, with 4,164 DMRs hypermethylated and 4,634 DMRs hypomethylated (Fig. 3b).
Prominent categories of differentially methylated genes in post-stroke BBB recovery were cell junction, polarity, and actin cytoskeleton. Significant cell junction GO terms included tight junction (q-value = 0.0025) and adherens junction organization (q-value = 0.00012). Notable differentially methylated genes with prevalent hypermethylation included junctional adhesion molecule-A (F11r), while prevalent hypomethylation occurred in occludin (Ocln) and claudin-5 (Cldn5) (Fig. 3c). Establishment or maintenance of cell polarity, a process required for proper localization of TJ complexes, was significant (q-value = 0.0025), with differential methylation observed in the Wnt signaling ligand genes, Wnt5a (hypomethylated mostly in promotor region) and Wnt7b (hypermethylated in gene body), the ARF6-regulating genes Cyth3, (hypomethylated in promotor region) and Frmd4a (hypermethylated in promoter and gene body), the endothelial polarity regulator Amotl2, as well as a regulator for establishing cell polarity Prkcz, (hypermethylation in promoter and gene body). Among the significant methylation changes in genes encoding actin cytoskeleton included those involved in actin filament organization (q-value = 0.0058) and positive regulation of cell projection organization (q-value = 2.85E-08), particularly within the actin-interacting genes Ezr and Myo3b (Fig. 3c).
In relation to poststroke BBB recovery, there were changes in a spectrum of genes associated with angiogenesis (q-value = 7.29E-06), endothelium development (q-value = 0.0084), regulation of endothelial cell migration (q-value = 0.0086) and endothelial cell differentiation (q-value = 0.0091). Significant differential methylation was present on the promoter and gene body of endothelial transcription factor genes include Sox17, Gata2 and Gata3 (all hypomethylated) while Robo4, involved in angiogenesis and endothelial barrier maintenance, had hypomethylated promoter DMRs. Strikingly, hypermethylation of the angiogenic inhibitor Angptl4 was present but not Hmgb2, a promoter of endothelial cell proliferation and migration (Fig. 3d).
Altered methylation patterns were also found in genes encoding proteins involved in endothelial cell processes, such as intracellular signaling cascades and transcription (Fig. 2e). Examples include canonical Wnt signaling pathway (q-value = 8.74E-06), with differential methylation of the inhibitor Gsk3b but not the Wnt ligand receptor Fzd4, as well as Ras protein signal transduction (q-value = 1.74E-05), with differentially methylated (hypomethylated) genes including the GTPase Cdc42ep4 and Rap2a, in both promotor and gene body DMRs. Other notable signaling cascade genes with hypomethylation include Mapk4, which regulates proinflammtory cytokines expression. Regarding transcriptional regulation, the DNA-binding transcription activator activity (q-value = 0.0018) and transcription regulator complex (q-value = 0.005) genes, such as Foxo3 and Grhl2, were hypomethylated predominately in their gene bodies (Fig. 3e).
Transporter, receptor, and channel genes also had striking changes in methylation. For example, differentially methylated genes involved in passive transmembrane transporter activity (q-value = 0.00028) included the ion channel-encoding genes Glra1, Kcnt2 and Scn3a. Lrp1b and Lrp5, low-density lipoprotein receptor family members, were hypomethylated predominantly in their gene bodies (Fig. 3f).
Extracellular matrix (ECM) genes (q-value = 6.74E-06) comprise the final category of differentially methylated genes, including genes involved in the regulation of cell-substrate adhesion (q-value = 1.28E-06), cell-substrate junction assembly (q-value = 0.003) and focal adhesion (q-value = 0.0099). Prominent promoter and gene body hypomethylation was present in Fn1 (fibronectin), the fibronectin receptor Itgb5, and the focal adhesion molecule Tns2 (Fig. 3g).
Effect of methylome changes on the transcriptome profile of the poststroke BBB recovery
The methylomic changes are best understood within the context of transcription, manifested as transcriptional repression due to promoter region methylation, while gene body methylation can lead to either transcriptional activation or repression (29, 30). Thus, we performed a parallel analysis of transcriptomic changes in brain microvessels during post-stroke BBB recovery in young (6-months old) mice. Compared to age-matched controls, post-stroke BBB recovery resulted in 2,740 differentially expressed genes (DEGs), defined as log2 fold change > 0.58 and an adjusted p-value < 0.05, with 1,546 upregulated and 1,194 downregulated (Fig. 4a). Gene ontology analysis revealed upregulated DEGs were involved in cell junction organization, such as cell-substrate junction (q-value = 0.001), establishment or maintenance of cell polarity (q-value = 0.009), ATP metabolic process (q-value = 5.66E-07), mitochondrial transport (q-value = 0.0010) and signaling cascades, such as ERK1 and ERK2 cascade (q-value = 0.0026) and small GTPase mediated signal transduction (q-value = 0.0028). Other enriched GO terms for upregulated DEGs included angiogenesis (q-value = 9.10E-06) and regulation of DNA-binding transcription factor activity (q-value = 0.0091). Downregulated DEGs demonstrated an enrichment of GO terms related to cell junctions, such as cell junction assembly (q-value = 0.0011) and cell-cell adhesion via plasma membrane adhesion molecules (q-value = 0.0012), ion channel activity (q-value = 1.63E-08), passive transmembrane transporter activity (q-value = 3.84E-08), and signaling pathways, such as G protein-coupled receptor activity (q-value = 1.62E-10) (Fig. 4b).
As promoter and gene body methylation directly affect the transcriptome, we assessed the overlap between changes to the DNA methylome and the transcriptome. Only 94 DEGs had altered promoter methylation, while 588 DEGs had differential methylation of non-promoter regions (Fig. 4c). However, the correlation between promoter methylation and gene expression demonstrated a non-significant trend toward a negative correlation (R = -0.045, p = 0.67). Despite that, there are groups of genes with significant promoter hypomethylation (q-value) and transcript upregulation (p.adjust), like the RhoGTPases, Cdc42ep4 (p.adjust = 0.0004, q-value = 7.94E-09) and Rhoa (p.adjust = 0.0236, q-value = 4.28E-12), focal adhesion molecule Tns2 (p.adjust = 0.00255, q-value = 3.74E-20), the regulator of angiogenesis and endothelial barrier establishment Robo4 (p.adjust = 0.0051, q-value = 3.15E-09), integrin β5 (Itgb5, p.adjust = 0.032, q-value = 0.00014), extracellular matrix protein fibronectin (Fn1, p.adjust = 0.032, q-value = 3.34E-29) and the angiogenic factors thrombospondin 4 (Thbs4, p.adjust = 0.04, q-value = 2.14E-17) and Alkbh5 (p.adjust = 6.56E-05, q-value = 1.66E-13). Promoter hypermethylation and transcript downregulation was observed in the regulator of VEGF-induced angiogenesis and glycine-mediated vascular reconstruction Glra1 (p.adjust = 0.043, q-value = 3.29E-05), the transcriptional repressor Scm-like with four MBT domains protein 2 (Sfmbt2, p.adjust = 0.017, q-value = 7.69E-08), the regulator of mechanotransduction Ttn (p.adjust = 0.001, q-value = 1.28E-26), cell polarity protein Cyth3 (p.adjust = 0.0124, q-value = 1.16E-11) and Epha2 (p.adjust = 0.031, q-value = 2.19E-05) (Fig. 4d, left).
Similar to promoter changes, methylation in non-promoter regions showed a non-significant negative correlation with transcriptome expression (R = -0.019, p = 0.52), although a positive correlation exists between gene expression and non-promoter methylation for a group of genes. Genes with hypomethylated non-promoter regions associated with significant transcript upregulation include the TJ protein occludin (Ocln, p.adjust = 0.032, q-value = 1.11E-07) adherens junction protein plakoglobin (Jup, p.adjust = 0.00068, q-value = 2.9E-13), actin cytoskeleton linker protein ezrin (Ezr, p.adjust = 0.0056, q-value = 2.28E-10), the regulator of the Wnt signaling and BBB maintenance Sox17 (p.adjust = 0.01, q-value = 1.80E-07), the angiogenic transcription factor Erg (p.adjust = 0.038, q-value = 2.94E-37), along with signaling molecules that promotes angiogenesis and barrier permeability, such as Mapk4 (p.adjust = 1.15E-05, q-value = 1.03E-07), and Igfbp3 (p.adjust = 0.00024, q-value = 7.72E-13), and the angiogenic inhibitor Notch4 (p.adjust = 0.00016, q-value = 2.46E-05). Furthermore, hypermethylation of non-promoter regions is associated with downregulation of genes encoding cell adhesion and mechanotransduction Cdh23 (p.adjust = 9.57E-07, q-value = 1.63E-15), actin cytoskeleton and mechanotransduction myosin 3b (Myo3b, p.adjust = 0.037, q-value = 1.37E-05), the transcription factor and regulator of Wnt/b-catenin signaling pathways Sox14 (p.adjust = 0.0032, q-value = 0.0002), the potassium transporter Scn3a (p.adjust = 0.0084, q-value = 1.17E-07) and the regulator of cell projection and morphogenesis Grhl2 (p.adjust = 0.036, q-value = 8.72E-16; Fig. 4d, right). Taken together, the methylome and transcriptome profile of BBB recovery indicated extensive remodeling of barrier properties mirrored by structural alterations (TJ protein expression, actin cytoskeleton remodeling, reestablishing cell polarity), and a restoration of the extracellular matrix and transporter systems. The brain endothelial cells display a more proangiogenic phenotype with activation of angiogenic transcription factors and Wnt-β-catenin signaling pathways for remodeling barrier properties. In addition to Wnt-β-catenin, other prominent signaling pathways that can alter recovery outcomes include Rho GTPase and MAPK.
Effect of aging on the BBB DNA methylome and transcriptome profile in poststroke recovery
Aging plays a critical role in the epigenetic alteration of the brain endothelial cells function and consequently on barrier properties (55). Aged mice (18 months) had larger infarct sizes with profound BBB leakiness 7 days after TE stroke onset (Fig. 1). Analyzing DNA methylome profile in post-stroke BBB recovery in aged mice, we found no changes in the global methylation level (Fig. 5a). However, RRBS analysis revealed that aging post-TE stroke brain microvessels had 11,287 DMRs, 5,005 hypermethylated and 6,282 hypomethylated, compared to age matched controls (Fig. 5b). The genomic regions containing the highest percentages of DMRs were introns, exons, and 1–5 kb upstream of the transcriptional start site, respectively while only 8.9% of DMRs were located within promoter regions (4.0% hypermethylated and 4.9% hypomethylated). When investigating DMR location in relationship to CpG islands, most (86.2%) were located within the open sea, while only 1.1% of DMRs were located within CpG islands (Supplementary Fig. 1a, b).
Similar to poststroke BBB recovery in young mice, poststroke BBB in aging mice had gene enrichment in clusters of cell junctions, actin cytoskeleton, angiogenesis, signaling pathways and transcription factors, transporters, and channels, as well as the extracellular matrix (Fig. 5c, 5d). The promoter regions showed 1,154 DMRs (521 hypermethylated and 633 hypomethylated) (Fig. 6a), while non-promoter regions had 10,133 DMRs, 4,484 being hypermethylated and 5,649 being hypomethylated (Fig. 6b). The significant methylation pattern in promoter and non-promoter regions were present in the tight junction cluster (q-value = 0.0015), with hypermethylation of genes that encode Cldn5, Tjp2, and Ocln and hypomethylation of F11r, actin cytoskeleton (e.g. regulation of actin filament organization, q-value = 0.0004; cell projection assembly, q-value = 0.0005; and actomyosin structure organization, q-value = 0.0031) with notable hypomethylation of actin cytoskeleton-related genes that encode Ezr, and filament-associated protein Cnn3. Two genes that regulate the BBB recovery and maintenance, Wnt5b and Wnt7a, showed hypermethylation and hypomethylation, respectively, predominantly in gene body (Fig. 5c and 6c).
Another cluster with significantly altered methylation related to angiogenesis and endothelial cell function. That included regulation of angiogenesis (q-value = 0.00999), blood circulation (q-value = 0.00044), and circulatory system processes (q-value = 0.00041). Hypomethylation was detected in genes encoding VEGF receptors, Flt1 and Kdr, and transcription factors involved in endothelial cell function (e.g., Foxo1, Hes1, Klf8 and Erg), while other essential endothelial cell transcription factors showed more hypermethylated pattern (e.g., Crip2 and Elk3) (Fig. 6d). In the cluster of signaling pathways and transcription factors, significant hypomethylation occurred in genes encoding canonical Wnt signaling pathway (q-value = 3.35E-07) with Fzd9 having a hypomethylated promoter, RhoGTPAse (q-value = 0.00016) with hypomethylated promoter DMRs within Cdc42ep1 and Rhoa, and Protein kinase C encoding genes (belonging to regulation of peptidyl-tyrosine phosphorylation, q-value = 0.00047) including Prkce and Prkcz. Other notable changes were present in the regulation of BMP signaling pathway (q-value = 0.0011) and response to transforming growth factor beta (q-value = 0.0091), with Smad6 containing DMRs (Fig. 6e). Differentially methylated genes encoding transporters, channels, and receptors were calcium ion transmembrane transport (q-value = 0.00098) and calcium channel activity (q-value = 0.009996), with DMRs located within Ryr2, potassium channel activity (q-value = 0.009996), the voltage-gated potassium channel subunit Kcnh7m and the calcium-binding protein S100a11 (Fig. 6f). Another significant category was regulation of ion transmembrane transport (q-value = 7.14E-05) and transmembrane transporter complex (q-value = 3.59E-07) (Fig. 6f). The extracellular matrix cluster was also significantly enriched with genes that regulate cell-matrix adhesion (q-value = 3.08E-05), cell-substrate junction organization (q-value = 0.00068), negative regulation of cell-substrate adhesion (q-value = 0.0059), and extracellular matrix organization (q-value = 0.00010). Genes of interest with altered methylation include Fn1, which encodes fibronectin (hypermethylation), along with its binding partner Fbln1, Coll6a1, Sparcl1 and integrin-β5 (Itgb5) (Fig. 6g).
Unique to old post-stroke BBB recovery included enrichment of DMRs in genes involved in endothelial to mesenchymal transition (EndMT), inflammation, and mitochondria function (Fig. 5c, 5d). Interestingly, EndMT genes with differential methylation were Tgfb1 and its receptor, Tgfbr2, transcription factors such as Ctbp1 and Snai1 as well Sox9, Sox2, another SRY-related HMG-box family member, Grhl2 and Hmga1 (Fig. 6h). Among inflammatory mediators, prominent DMRs were located within in genes of complement system (e.g., C1qa and C3), various cytokines and chemokines (e.g., Cxcl14, Ccl12 and Cxcl10), toll like receptor 2 (Tlr2), lysosomal function (e.g., Grn), and histocompatibility complexes (e.g., H2-D1; Fig. 6i). Mitochondrial-associated genes also demonstrate altered methylation patterns; for example, the ADP:ATP antiporter, Slc25a4 (hypomethylation), cytochrome-encoding gene Cyb5r3, and metabolic enzyme Clybl and Glud1 (Fig. 6j).
The altered methylation pattern was analyzed in the context of transcriptomic changes. RNA-seq revealed that poststroke BBB recovery in old mice have global transcriptomic changes compared to age-matched controls: out of the 6,999 total DEGs, 3,723 were upregulated, and 3,276 were downregulated (Fig. 7a). Most of the upregulated genes were enriched in the following categories: angiogenesis (q-value = 7.69E-08), cell-substrate junction (q-value = 0.0097), actin filament organization (q-value = 0.00014), regulation of apoptotic signaling pathway (q-value = 3.87E-0), Ras protein signal transduction (q-value = 0.00046), regulation of ERK1 and ERK2 cascade (q-value = 0.0050), inflammatory processes, such as regulation of leukocyte migration (q-value = 2.57E-06), and metabolism, including ATP metabolic processes (q-value = 9.08E-20) and oxidative phosphorylation (q-value = 1.98E-17) (Fig. 7b). Intriguingly, upregulated DEGs were enriched for GO terms relating to translational control, which was unique to the old poststroke group. This included cytoplasmic translation (q-value = 8.02E-10) and ribosome (q-value = 1.04E-37) (Fig. 7b). Furthermore, enriched GO terms for downregulated DEGs included calcium ion transmembrane transporter activity (q-value = 2.79E-06), potassium ion transport (q-value = 0.0037), calcium channel complex (q-value = 0.0008), and transmembrane transporter complex (q-value = 1.32E-14) (Fig. 7b).
Comparing the overlap between DEGs and DMRs revealed only 294 DEGs with altered methylation within their promoter regions, and 1,594 DEGs with altered methylation within non-promoter regions (Fig. 7c). Similar to the poststroke condition in young mice, expression of select DEGs negatively correlated with their promoter region methylation (R = -0.0082, p = 0.88); however, methylation of non-promoter regions positively and significantly correlated with gene expression (R = 0.042, p = 0.027). (Fig. 7d). Despite this positive correlation, there are groups of genes that have a negative correlation between their expression and methylation pattern. Hypomethylated genes with upregulated transcript expression included the actin cytoskeleton protein ezrin (Ezr, p.adjust = 6.75E-05, q-value = 8.26E-10) and actin-related protein Cnn3 (p.adjust = 0.00028, q-value = 2.68E-10), regulators of angiogenesis like Vegf receptors Flt1 (p.adjust = 0.031, q-value = 1.76E-14) and Kdr (p.adjust = 0.0232, q-value = 5.68E-07), angiogenic transcription factors Erg (p.adjust = 0.0117, q-value = 2.81E-06) and Hes1 (p.adjust = 0.0013, q-value = 2.48E-06) and the cell cycle regulatory axis Ccnd1 (p.adjust = 3.24E-05, q-value = 1.22E-06)/Igfbp3 (p.adjust = 0.0008, q-value = 3.83E-35). Members of the Wnt canonical pathway and regulators of the BBB repair, Wnt7b and Fzd9, were hypomethylated (Wnt7b q-value = 8.83E-13, and Fzd9 q-value = 1.47E-07) with significant upregulation of gene expression (Wnt7b p.adjust = 0.0022; Fzd9, p.adjust = 0.0188). Besides Wnt signaling pathways, promoter hypomethylation and transcript upregulation was observed for Rhoa (p.adjust = 0.0008, q-value = 3.68E-14) and Cdc42ep1 (p.adjust = 0.0014, q-value = 5.47E-05) (Fig. 7d).
Intriguingly, old poststroke BBB recovery also resulted in the significant hypomethylation and subsequent upregulation of a repressor of angiogenesis, the transcriptional factor Foxo1 (p.adjust = 0.0034, q-value = 2.37E-06) and hypermethylation/downregulation of two angiogenic factor genes, the transcription factor Klf8 (p.adjust = 0.0011, q-value = 8.31E-28) and Epha6 (p.adjust = 0.0010, q-value = 6.13E-09) (Fig. 7d). Ultimately, altered methylation and expression of these genes could affect the outcome of poststroke angiogenesis. Regarding structural proteins responsible for building the barrier, hypermethylation of a non-promoter region in Slit1 is associated with its decreased transcript expression (p.adjust = 0.0483, q-value = 1.27E-14), while promoter hypermethylation of Cldn5 was associated with upregulated transcript expression (p.adjust = 0.0055, q-value = 2.66E-09) (Fig. 6b, 7d). For old poststroke mice, a unique methylation pattern exists in genes encoding endothelial to mesenchymal transformation. Hypomethylation within non-promoter regions and increased transcript expression is present for genes that encode Snai1 (p.adjust = 0.0086; q-value = 2.84E-07), Sox9 (p.adjust = 0.0006, q-value = 7.76E-09), Tgfb1 (p.adjust = 6.90E-05, q-value = 8.91E-63) and Tgfbr2 (p.adjust = 0.0047, q-value = 1.80E-09).
In summary, the transcriptional and methylome profile of poststroke BBB recovery in aging mice implies that angiogenesis and Wnt-related pathways drive the BBB recovery process, although the repair process is affected by the activation of genes involved in endothelial to mesenchymal transformation and angiogenic repression, limiting the final outcomes.
Common and unique transcriptomic and DNA methylome profiles of BBB poststroke recovery in young and aging mice
As the severity of poststroke BBB injury differs between young and old mice, we compared the DNA methylome and transcriptome changes to highlight cellular processes contributing to the discrepancy in poststroke BBB recovery. Remarkably, there are only 1,138 significant DMRs common to poststroke BBB recovery in both old and young mice (Fig. 8a). This lack of overlap is still observed when assessing methylation within specific genomic regions, as there are 112 common DMRs within promoters and 1,026 DMRs within non-promoter regions (Supplementary Fig. 2a, 2b). Common DMRs identified in poststroke BBB recovery of old and young mice do not significantly correlate (R = 0.0038, p = 0.134), which is also true when assessing the correlation of common non-promoter DMRs (R = 0.0018, p = 0.5227) (Fig. 8a, Supplementary Fig. 2d). A significant correlation, however, is observed for the common promoter region DMRs (R = 0.0188, p-value = 0.00055) (Supplementary Fig. 2c). Common DMRs with methylation changes in the same direction (e.g. hypermethylated in both groups) belong to signaling pathways (e.g., positive regulation of MAPK cascade, q-value = 0.03; GTPase regulator activity, q-value = 0.015; and G protein-coupled receptor activity, q-value = 0.041) (Supplementary Fig. 2e). DMRs with methylation changes in opposite directions (e.g. hypermethylated in old and hypomethylated in young) are located within genes involved in endocytosis (e.g. clathrin-coated endocytic vesicle, q-value = 0.0187; endocytic vesicle, q-value = 0.039) and b-catenin binding (q-value = 0.038) (Supplementary Fig. 2f). Comparing the transcriptomic changes in BBB recovery across the young and old mice reveals 2,042 significant DEGs common to both experimental groups (Fig. 8b). The common DEGs between old and young poststroke BBB recovery have a strong positive correlation (R = 0.704, p < 2.2E-16), with only 8 DEGs regulated in the opposite direction (e.g., upregulated in old post-TE stroke mice and downregulated in young post-TE stroke mice) (Fig. 8b).
Furthermore, we identified genes of interest with similar changes in methylation (e.g., hypermethylated or hypomethylated in both experimental groups) for both young and old mice. These genes encode proteins involved in actin-binding signaling pathway activity, transcriptional regulation, and protection from oxidative stress. For example, actin-binding proteins Parva and Ezrin, and the Ets family transcription factor Erg were hypomethylated with increased transcript expression in the poststroke microvessels of both young and old mice. Other examples included the activator of MAPK signaling Alk (hypermethylated and decreased transcript expression) and the negative regulator of MAPK signaling Dusp3 (hypomethylated and increased transcript expression), indicating the same trend in the regulation of MAPK kinase in BBB recovery (Fig. 8c). Genes with differential methylation patterns (e.g. hypermethylated in old and hypomethylated in young) across the BBB of old and young poststroke mice include the signaling molecule Prkce, involved in actin cytoskeleton function (migration, adhesion) and actin cytoskeleton modulator Arf6 (both genes hypomethylated in old mice and hypermethylated in young mice), while Cdc42ep4 was hypermethylated in old and hypomethylated in the young post-stroke BBB. Conversely, increased transcript expression of Arf6 and Cdc42ep4 was observed in old and young post-stroke BBB (Fig. 8c).
Finally, there is a unique DNA methylome and transcriptome profile for both experimental groups. These genes belong to categories relevant to endothelial cell biology, such as cell junctions, angiogenesis, and signaling pathways. The unique pattern for poststroke BBB recovery in young mice was characterized by alterations in methylation and transcriptome expression of cell junction and polarity complex regulators such as Ocln, Epha2, Tns2, and Cyth3; the angiogenic transcriptional regulators Sox14, Foxo3, Klf4, and Gata2 (Fig. 8d). Other unique patterns are observed in regulators of MAP kinase (e.g., Mapk4), transporters (e.g., Glr2), the extracellular matrix (e.g., Spock3), and inflammation (e.g., Il6) (Supplemental Fig. 3a).
The effects of aging on BBB recovery results in a unique transcriptome and DNA methylome profile that includes the cell junction and polarity complex-encoding genes Cldn5, Cldn11, Tjp2 and Dlg5; the angiogenic transcription factors Ets2, Hes1, and Foxo1; the Wnt signaling pathway genes Fzd9, Wnt7a, and Wnt7b; the Rho kinase pathway gene Cdc42ep3; the integrins Itgb2, Itgb4 and Itgb1; and the transporters Lrp10 and Atp1a2. The intriguing profile of aging poststroke BBB recovery is also characterized by a profound alteration in methylation and gene expression of regulators of endothelial to mesenchymal transformation (Snai1, Smad6, Tgfbr2, and Sox9) and epigenetic regulators (Sirt2 and Kdm6a) (Fig. 8e and Supplemental Fig. 3b).
Thus, DNA methylation has an important role in BBB recovery, directing some of the critical processes involved in restoring the structural and functional BBB, such as angiogenesis, junctional proteins, establishment of polarity, actin cytoskeleton reorganization, extracellular matrix, as well as transporter system reestablishment. Nevertheless, DNA methylation could also contribute to the limited BBB recovery in young mice, mostly through the activation of specific signaling pathways (e.g., Rho GTPases), while in aging mice the limited BBB recovery could be due to the repression of structural protein expression (e.g., claudin-5), as well as activation of genes involved in endothelial to mesenchymal transformation, repression of angiogenesis, and epigenetic regulation.