Disengaging the Engine: Histone Deacetylases 1 and 2‐Mediated Acetylation of Hexokinase‐2 Regulates Energy Metabolism in Microglia Following Intracerebral Hemorrhage
Zhiwen Jiang, Heng Yang, Xinjie Gao, Zengyu Zhang, Ruiyuan Weng, Yuchao Fei, Jiabin Su, Hanqiang Jiang, Wei Ni, Yuxiang Gu
TL;DR
This study shows that inhibiting HDAC1/2 in microglia after brain hemorrhage reduces inflammation and improves brain recovery by changing energy metabolism.
Contribution
The study reveals a novel role of HDAC1/2 in regulating microglial metabolism and function through HK2 acetylation after intracerebral hemorrhage.
Findings
HDAC1/2 inhibition reduces HK2 acetylation and shifts microglial metabolism from glycolysis to fatty acid oxidation.
This metabolic shift reduces pro-inflammatory responses and enhances phagocytic activity in microglia.
HDAC1/2 inhibition also promotes mitophagy and reduces microglial proliferation and cell numbers.
Abstract
Microglia‐mediated neuroinflammation is closely associated with the pathogenesis of secondary brain injury following spontaneous intracerebral hemorrhage (ICH). However, the relationship between immune response regulation and metabolic patterns in microglia remains unclear. Histone Deacetylases 1 and 2, a class of lysine deacetylases, regulates gene transcription by modulating histone acetylation modifications and is widely involved in various cellular activities of microglia. In this study, we observed that knockout of HDAC1/2 in microglia alleviated neurological deficits caused by ICH, preserved white matter integrity, and accelerated hematoma clearance post‐ICH. Mechanistically, we found that after ICH, microglia exhibited increased expression of hexokinase 2 (HK2) and enhanced glycolysis. HDAC1/2 knockout/pharmacological inhibition affected the acetylation level of HK2, inhibited…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7
FIGURE 8
FIGURE 9
FIGURE 10- —Natural Science Foundation of Tibet Autonomous Region10.13039/501100018543
- —Ministry of Science and Technology of the People's Republic of China10.13039/501100002855
- —National Natural Science Foundation of China10.13039/501100001809
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsIntracerebral and Subarachnoid Hemorrhage Research · Neuroinflammation and Neurodegeneration Mechanisms · Amyotrophic Lateral Sclerosis Research
Introduction
1
Intracerebral hemorrhage (ICH) is a severe form of stroke that often resulted in neurological dysfunction and high disability rates [1, 2]. During the pathological process of ICH, microglia, the resident immune cells of the central nervous system, played a crucial role [3, 4, 5]. Studies showed that ICH could activate microglia, triggering a series of inflammatory responses that were key contributors to the exacerbation of brain injury [6, 7].
Histone deacetylases 1 and 2 (HDAC1/2) regulates gene transcription by altering chromatin accessibility through the modulation of histone acetylation levels, playing a critical role in the activation of microglia [8, 9, 10]. In various brain diseases and stroke models, inhibition of HDAC1/2 was associated with reduced brain damage by decreasing inflammatory responses and regulating neuronal death [10, 11, 12]. These changes were typically driven by HDAC1/2‐mediated alterations in histone acetylation, which in turn regulated the transcription of a series of genes. However, the broader involvement of non‐histone protein acetylation in this process and whether it played a significant role remained unclear.
The epigenetic regulation mediated by HDAC1/2 not only affected the inflammatory responses of microglia but also influenced cellular metabolic, like glycolysis, fatty acid oxidation and glutamine metabolism [13, 14]. Glycolysis was the primary energy source for activated microglia, and hexokinase 2 (HK2), a key enzyme in the glycolytic pathway, was responsible for phosphorylating glucose to glucose‐6‐phosphate, thus initiating glycolysis [15, 16]. In activated microglia, glycolytic activity was significantly upregulated, and HK2 expression increased accordingly [15, 17], suggesting that glycolysis played an important role in the inflammatory response triggered by ICH. In contrast, under anti‐inflammatory or reparative conditions, fatty acid oxidation became more prominent, suppressing the inflammatory phenotype of microglia [18, 19, 20]. This enhanced metabolic pathway reflected the adaptive changes of microglia in response to the stress and energy demands of disease.
Further studies revealed that HK2 could be regulated by various post‐translational modifications such as phosphorylation, ubiquitination, which affected its enzymatic activity and its interaction with VDAC1 or localization on mitochondrial [21, 22, 23]. The interaction between HDAC1/2 and HK2, as well as the role of HK2's post‐translational modifications in microglial function, remained poorly understood.
Through multi‐omics sequencing, we unexpectedly discovered that HK2 underwent significant acetylation modifications in HDAC1/2 knockout microglia. As the highly expressed hexokinase isoform in activated microglia, HK2 is crucial for maintaining glycolytic activity in these cells. The acetylation of HK2 leads to conformational changes in the enzyme, resulting in a significant reduction in its catalytic activity. This shift forces activated microglia to abandon the glycolysis‐dominated metabolic mode and transition to fatty acid oxidation. Additionally, HK2 acetylation weakens its interaction with VDAC1, partially mediating mitochondrial dysfunction in microglia. This mitochondrial impairment exacerbates oxidative stress within the cells, ultimately leading to cell death. Our study identifies HK2 as a critical target through which HDAC1/2 regulates the inflammatory and metabolic phenotypes of microglia, linking HDAC1/2‐mediated epigenetics and cellular metabolism.
Results
2
Knockout of HDAC1/2 Improved Long‐Term Neurological Dysfunction After ICH
2.1
A spontaneous Intracerebral Hemorrhage (ICH) model was established in mice via stereotactic injection of autologous blood (Figure S1A). On day 3 post‐ICH, we observed a significant increase in the expression of HDAC1 and HDAC2 in microglia surrounding the hematoma, compared to the sham group, as demonstrated by immunofluorescence co‐staining (Iba1+HDAC1/2) (Figure S1B,C,D). To further explore the role of HDAC1/2 in regulating microglial functions after ICH, we generated a microglia‐specific HDAC1/2 knockout mouse model (Figure S1E). Gene knockout was induced by intraperitoneal injection of tamoxifen in male adult mice (For WT mice, we administered the same dosage and frequency of tamoxifen to exclude its potential effects), followed by fluorescence‐activated cell sorting (FACS) to isolate cerebral microglia (Ly6G^−^CD11C^−^CD45^i^ ^n^ ^t^CD11b^+^) (Figure S1F). RT‐qPCR analysis of the sorted microglia showed that the mRNA expression of HDAC1 was reduced to 15.8% of that in wild‐type mice (Figure S1G), and the mRNA level of HDAC2 decreased to 42.6% of the wild‐type level (Figure S1H).
To investigate the potential impact of HDAC1/2 knockout on motor and cognitive functions following ICH, we performed a series of behavioral tests, including the rotarod, foot‐fault, hanging wire, and Morris‐water maze assays (Figure 1A). Following HDAC1/2 knockout, mice displayed significant improvements in motor function, as evidenced by increased time spent on the rotarod, a reduced foot‐fault rate, and higher performance in the hanging wire test till 35 days after ICH (Figure 1B–D). Moreover, Morris‐water maze results revealed that HDAC1/2 knockout alleviated long‐term memory and cognitive deficits induced by ICH, as indicated by an increased number of crossings over the platform area (Figure 1E,J), but there is no significant difference in “Time of target quadrant”, “Latency to platform” and “Distance” (Figure 1H–J). Additionally, we assessed the effect of HDAC1/2 knockout on hematoma resolution after ICH. Dynamic T2‐weighted MRI scans were conducted on days 1, 3, 7, and 35 post‐ICH to monitor hematoma volume. The data showed that, compared to WT‐ICH mice, HDAC1/2 knockout significantly accelerated hematoma clearance (Figure 1K,L).
Knockout of HDAC1/2 ameliorates neurological deficits and accelerates hematoma clearance after ICH. (A) Behavioral tests and imaging detection diagram, by figdraw. (B C D) Sensorimotor deficits were evaluated before (pre) and up to 35 days after ICH or Sham surgery by wire hanging, fault foot and rotarod test. n = 12 per group. (E) Representative image of the Morris‐water maze experiment in mice. (F) Learning curve graph of the Morris‐water maze experiment in mice. n = 5 to 12 per group. (G H I J) The escape latency, the time spent/distance in the target quadrant and the times of shuttle the platform was measured in Morris‐water maze. n = 5 to 12 per group. (K) Representative image of T2‐weighted MRI sequence after ICH in mice, as reflected by the quantification (L). n = 4 to 6 per group. All values are expressed as mean ± SD. The data were analyzed using two‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (B, C, D, F, L) or one‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (G, H, I, J). * p < 0.05, ** p < 0.01, *** p < 0.001 ns, not significant.
Knockout of HDAC1/2 Preserved the Structural and Functional Integrity of White Matter After ICH
2.2
To further investigate the protective effect of HDAC1/2 knockout in microglia on ICH‐induced white matter damage, we assessed both the structure and function of the white matter (Figure 2A). Neuronal loss was evaluated using NeuN staining 35 days after ICH. The results showed a reduction in neuronal density in the peripheral area of the hematoma in the WT‐ICH group. In contrast, the HDAC1/2 miKO‐ICH group exhibited a significant improvement in the number of neurons surrounding the hematoma (Figure 2B).
Knockout of HDAC1/2 improves the integrity of white matter structure in mice. (A) Diagram illustrating the experimental design, by figdraw. (B) Representative images of NeuN immunostaining 35 days after ICH, as reflected by the quantification. n = 5 per group. (C D) Representative images of MBP/NF‐H and (E) SMI‐32 immunostaining 35 days after ICH. (F G H I) Quantification of the fluorescence intensity and area of MBP and NF‐H in the STR. n = 5 per group. (J) Quantification of the fluorescence intensity of SMI‐32 in the STR. n = 5 per group. (K) Representative image of DTI (Diffusion Tensor Imaging) sequence scan in mice at day 35 after ICH. The red dashed circle represents the STR region. (L) Qualitative analysis results of FA, MD, AD and RD value at day 35 after ICH in mice. n = 4 to 6 per group. (M) Quantification of nerve fiber bundles with DTI sequence after ICH in mice. n = 4 to 6 per group. All values are expressed as mean ± SD. The data were analyzed using unpaired two‐ tailed Student's t test (J, L and M) or one‐ way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (B, F, G, H, I). * p < 0.05, ** p < 0.01, *** p < 0.001 ns, not significant.
Next, we performed immunofluorescence staining for MBP (a major myelin protein), NF‐H (a neurofilament heavy chain marker), and SMI‐32 (a marker for axonal degeneration) on tissue sections 35 days post‐ICH (Figure 2C–E). We observed that the intensity of MBP and NF‐H immunofluorescence around the hematoma was significantly reduced following ICH, along with a decreased area of labeling for both markers, indicating severe damage to white matter fiber tracts. However, HDAC1/2 knockout significantly improved the preservation of myelin in the tissue surrounding the hematoma (Figure 2F–I). In addition, SMI‐32 staining, which marks denuded nerve fibers after demyelination, showed a marked increase in SMI‐32 expression around the hematoma following ICH. In contrast, inhibition of HDAC1/2 resulted in a reduction in the average fluorescence intensity of SMI‐32, suggesting that HDAC1/2 inhibition alleviated demyelination (Figure 2J). Collectively, these findings indicate that HDAC1/2 deletion in microglia contributes to the preservation of white matter structural integrity after ICH.
Furthermore, we performed diffusion tensor imaging (DTI) scans 35 days post‐ICH and conducted nerve fiber tract tracing analysis in the striatum (Figure 2K). We found that HDAC1/2 knockout alleviated the loss of nerve fiber tracts induced by ICH, as evidenced by an increase in fractional anisotropy (FA) values in the region surrounding the hematoma (Figure 2L,M).
To further evaluate the protective effects of HDAC1/2 inhibition, we administered intraperitoneal injections of the selective HDAC1/2 inhibitor Romidepsin (FK228) (Figure S2A). Nerve fiber tract tracing analysis in the striatum revealed that HDAC1/2 inhibition alleviated the loss of nerve fiber tracts caused by ICH, as evidenced by an increase in fractional anisotropy (FA) values (Figure S2B–D). No significant differences were observed in the small‐world network properties among the three groups (Figure S2E). We also performed electrophysiological experiments to assess the long‐term functionality of white matter by recording compound action potentials (CAPs) in the corpus callosum (Figure S2F). The first negative peak, N1, reflects the conduction capacity of myelinated fibers, which rely on saltatory conduction through the nodes of Ranvier for rapid signal propagation. The second negative peak, N2, represents the conduction ability of unmyelinated fibers, which conduct signals more slowly due to the absence of saltatory conduction. On day 35 post‐ICH, both N1 and N2 amplitudes were significantly reduced compared to the Sham group, indicating white matter dysfunction following hemorrhage. However, Romidepsin‐treatment partially restored both N1 and N2 amplitudes, suggesting that HDAC1/2 inhibition improved white matter function after ICH (Figure S2G–K).
These findings, in line with previous histological results, further confirm that Romidepsin enhances the long‐term structural and functional integrity of white matter after ICH.
Inhibition of HDAC1/2 Promotes HK2 K866 Acetylation in Microglia
2.3
HDAC1/2, as histone deacetylases, play a pivotal role in transcriptional regulation by modulating the acetylation status of histones. Beyond histones, they also influence a variety of cellular processes through the acetylation of non‐histone proteins. To investigate the neuroprotective effects of microglia‐specific HDAC1/2 knockout/inhibition after ICH, we utilized the selective HDAC1/2 inhibitor Romidepsin to treat BV2 microglia. An in vitro ICH model using BV2 cells had previously been established, and it was shown that Romidepsin can modulate microglial polarization [24]. To further explore the mechanisms underlying HDAC1/2 inhibition in neuroprotection, we performed multi‐omics analysis, including transcriptomics, proteomics, and acetylation modification profiling (Figure 3A).
Inhibition of HDAC1/2 promotes acetylation modification of HK2 K866. (A) Multi‐omics workflow diagram of microglia, by figdraw. (B) PCA plot of transcriptomics, proteomics and acetyl‐proteomics of BV2 cells following 24 h Hemin stimulation. n = 3 to 4 per group. (C) Bar chart of differential proteins and acetylation modification sites among three groups (D) Biological and Kegg pathway enrichment of differentially acetylated proteins. (E) Enrichment analysis of the association between PTM (Post‐Translational Modifications) and proteins in biological pathway. (F) Enrichment analysis of the association between PTM (Post‐Translational Modifications) and proteins in KEGG pathway. (G) Venn diagram of differentially expressed genes between BV2+Hemin and BV2+Hemin+FK228. (H) Enrichment analysis of the differentially expressed genes between BV2+Hemin and BV2+Hemin+FK228 in KEGG pathway. (I) Volcano plot of differential acetylation modification sites. The figure shows the proteins and modification sites with fold changes within the top 5 range. (J) Acetylation modification sites map of glycolysis‐related metabolic enzymes among three groups. All values are expressed as mean ± SD. * p < 0.05, ** p < 0.01, *** p < 0.001 ns, not significant.
The multi‐omics sequencing results revealed significant differences in transcription, translation, and post‐translational modifications, particularly acetylation, among the three microglial groups (Figure 3B). Notably, the acetylation levels of proteins in the Romidepsin‐treated microglia group exhibited significant alterations compared to the other two groups (Figure 3B,C), reflecting the broad regulatory effects of this inhibitor on protein acetylation.
Enrichment analysis of differentially acetylated proteins revealed that inhibition of HDAC1/2 primarily impacts biological processes and KEGG pathways such as glycolysis, TCA cycle, and apoptosis in microglia (Figure 3D). Additionally, combined analysis of differentially expressed proteins and acetylation modifications demonstrated that HDAC1/2 inhibition regulates various biological processes, including autophagy, apoptosis, and glycerophospholipid metabolism. This inhibition also affects key signaling pathways, such as apoptosis, p53 signaling, cell cycle, autophagy, and MAPK signaling (Figure 3E,F).
Similarly, KEGG enrichment analysis of all differential genes between the BV2+Hemin+FK228 and BV2+Hemin groups revealed that HDAC1/2 inhibition significantly modulates multiple pathways, including lipid metabolism, cell cycle, apoptosis, and autophagy (Figure 3G,H). To further explore the impact of HDAC1/2 inhibition, we performed KEGG pathway enrichment analysis on the differentially expressed genes in microglia showing opposing trends after Hemin stimulation and FK228 intervention. This analysis identified that these genes were predominantly involved in pathways related to oxidative phosphorylation, phagocytosis, TNF signaling, apoptosis, and fatty acid metabolism (Figure S3A,B). Next, we focused on the differential proteins and their modification sites between the BV2+Hemin and BV2+Hemin+Romidepsin groups. Notably, we found that inhibition of HDAC1/2 led to a significant increase in acetylation at the K866 site of the key glycolytic enzyme hexokinase 2 (Figure 3I, Figure S3C). This significant acetylation modification of HK2 suggests that HDAC1/2 plays an important role in regulating microglial metabolism, particularly within the glycolytic pathway. To investigate the effect of HDAC1/2 inhibition on microglial metabolism in more detail, we analyzed the acetylation levels of glycolysis‐related enzymes. In hemin‐stimulated microglia, we did not observe significant changes in the acetylation of these enzymes. However, after HDAC1/2 inhibition, several glycolytic enzymes, including HK2, Ldha, Eno1, and Gapdh, showed significant increases in acetylation levels. Among these, the most prominent change occurred at the K866 site of HK2 (Figure 3J). To confirm these findings, we used Parallel Reaction Monitoring (PRM) technology to validate the increased acetylation at the K866 site of HK2 in microglia following HDAC1/2 inhibition (Figure S3D).
Additionally, we observed that the amino acid sequences of HK2 are highly conserved across mice, rats, and humans, indicating substantial similarity among species (Figure S3E). Motif analysis further revealed that the K866 acetylation site of HK2 is surrounded by enriched basic and hydrophobic amino acids, suggesting a specific sequence preference that may influence the functional regulation of the protein (Figure S3F). Given that HK2 is a critical enzyme in glycolysis and a microglia‐specific isoform of hexokinase, this acetylation modification warrants further investigation into whether HDAC1/2 regulates cellular glycolytic capacity and, in turn, modulates microglial function through acetylation of HK2. Furthermore, protein acetylation is typically regulated by a balance between acetyltransferases and deacetylases. However, the specific enzymes responsible for acetylating the lysine site K866 on HK2, which we recently identified, remain unclear. To elucidate the enzymes involved in HK2 acetylation, we purified HK2 from BV2 microglial cells and performed mass spectrometry to identify interacting proteins. The mass spectrometry analysis identified several known acetyltransferases and deacetylases that may regulate HK2 acetylation at the K866 site, including acetyltransferases KAT6A, KAT7, and p300, as well as deacetylases HDAC1, HDAC2, and HDAC3 (Figure S3G,H).
Knockout of HDAC1/2 Regulates the Functional Phenotype of Microglia Surrounding the Hematoma After ICH
2.4
Transcriptome analysis revealed that BV2 microglial cells treated with Hemin exhibited a pronounced pro‐inflammatory M1 phenotype, characterized by upregulation of M1‐related genes such as Tnf‐α, Nfkb1, Il6, Il1a, Fcgr3, and Fcgr2b, while M2‐related genes, including CD209, Il4, Il10, and Cd163, were downregulated. However, following Romidepsin treatment, there was a significant reduction in the expression of M1‐related genes, and an upregulation of M2‐related genes, indicating that Romidepsin promotes a shift of BV2 microglial cells toward an anti‐inflammatory M2 phenotype (Figure 4A). Notably, genes associated with microglial homeostasis displayed distinct expression patterns between the Romidepsin‐treated and Hemin groups. Some homeostasis‐related genes, such as Aif1, Mertk, and Cx3cr1, were restored following Romidepsin treatment, while others, including P2ry12, Fcgr1, and Csf1r, were further downregulated. These findings suggest that Romidepsin may selectively modulate microglial homeostasis, influencing both inflammatory and repair responses. Additionally, Gene Set Enrichment Analysis (GSEA) indicated significant enrichment in pathways related to the downregulation of LPS‐induced inflammatory responses, TGFβ1 signaling via NFIC 10hr Up, and the classical complement activation pathway (Figure S4A–C).
Knockout of HDAC1/2 regulates the functional phenotype of microglia. (A) Heatmap of transcriptomic differences in microglial markers across three states (resting, anti‐inflammatory and pro‐inflammatory state). n = 4 per group. (B) Detection diagram of microglia phenotypes and functions, by figdraw. (C) Representative images of Iba1/CD16 and Iba1/Arg1 immunofluorescence in the peri‐ hematoma STR 3 days after ICH, as reflected by the quantification (D). n = 5 per group. (E) Representative images illustrating the decreased number of Iba1+ cells in STR, as reflected by the quantification (F). n = 5 per group. (G) Representative images and Quantitative analysis of Sholl analysis for microglia at 3 days afe, as reflected by the quantification (H). n = 5 per group. (I) Representative images of Iba1/Ki67 immunofluorescence in the peri‐ hematoma STR 3 days after ICH, as reflected by the quantification (J K). n = 5 per group. (L) Representative images of Iba1/pH3 immunofluorescence in the peri‐ hematoma STR 3 days after ICH, as reflected by the quantification (M N). n = 5 per group. (O) Representative images of Iba1/dMBP immunofluorescence in the peri‐ hematoma STR 3 days after ICH, as reflected by the quantification (P Q). n = 5 per group. All values are expressed as mean ± SD. The data were analyzed using one‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (D and F) or two‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (H) or unpaired two‐ tailed Student's t test (J, K, M, N, P, Q). p < 0.05, ** p < 0.01, *** p < 0.001, ns, not significant.*
To further investigate how HDAC1/2‐miKO alters microglial functions following ICH, we performed a series of in vivo and in vitro experiments (Figure 4B). Triple immunofluorescence staining for pro‐inflammatory (CD16), anti‐inflammatory (Arg1), and microglial (Iba1) markers was used to assess microglial heterogeneity (Figure 4C). The results showed that HDAC1/2‐miKO significantly reduced the percentage of CD16^+^ microglia on day 3 post‐ICH, but had no effect on the percentage of Arg1^+^ microglia (Figure 4D). Further analysis on day 7 post‐ICH revealed similar results: HDAC1/2 knockout significantly suppressed CD16 expression in microglia, with no significant impact on Arg1 expression (Figure S4E). In addition, using the same in vitro ICH model in primary microglia, we found that Romidepsin inhibited the transcription of Cd11b and Tnf‐α while upregulating the transcription of CD206 (Figure S4F).
Microglia play a crucial role in various functions after ICH, including proliferation, phagocytosis, and inflammatory responses, all of which are essential for ICH prognosis. Since the morphological characteristics of microglia are closely linked to their functional states, we further examined microglial morphology to assess the impact of HDAC1/2 knockout on microglial activation potential. Sholl analysis revealed that HDAC1/2‐miKO did not significantly affect the morphology of microglia in a resting state. However, in the ICH model, microglia surrounding the hematoma in HDAC1/2 knockout mice displayed loss of amoeboid‐like changes with increased branching compared to WT mice (Figure 4G,H). In our in vitro ICH model, we observed significant morphological changes in microglial cells following Romidepsin treatment. Sholl analysis showed that both resting and Hemin‐stimulated microglia in the Romidepsin‐treated group exhibited more branching and longer processes. This indicates that HDAC1/2 inhibition promotes a more branched morphology, which is typically associated with a resting or surveillance state in physiological conditions but seems to extend to activated microglia in this context (Figure S5B,C).
We also found that the number of microglia around the hematoma was significantly reduced in HDAC1/2 knockout mice (Figure 4E,F), and similar results were observed in the in vitro experiment (Figure S5B,D). Microglia typically respond to stress through transcriptional regulation, particularly during the acute phase of injury. The reduced microglial population around the hematoma following HDAC1/2 knockout could be due to several factors, including decreased proliferation, increased apoptosis, or impaired migration. Transcriptomic analysis showed that several key markers of proliferation, including Cdk4, H3f3a, Mcm2, Mki67, and Cdc20, were significantly downregulated after HDAC1/2 inhibition (Figure S5A). GSEA enrichment analysis further indicated that inhibition of HDAC1/2 activates the TP53 Hras Cooperation Response down pathway (Figure S4D), a pathway known to regulate inflammation, proliferation, and apoptosis.
To assess the effects of HDAC1/2 knockout on microglial proliferation around the hematoma, we performed Ki67 and pH3 co‐staining with Iba1. The results showed that HDAC1/2 knockout significantly suppressed microglial proliferation, as evidenced by a marked reduction in the number of Ki67^+^/Iba1^+^ (Figure 4I–K) and pH3^+^/Iba1^+^ cells (Figure 4L–N). Additionally, genes related to phagocytosis were upregulated (Figure S4E), prompting us to assess microglial phagocytic capacity using myelin debris as the target. We defined two states of microglial phagocytosis of myelin debris as “Engulfed” and “No contact” (Figure S5H) [25]. Quantitative analysis revealed that HDAC1/2 knockout enhanced the phagocytic capacity of individual microglia, as evidenced by increased engulfment of myelin fragments (Figure 5O,P). However, despite the increased phagocytic capacity of individual cells, there was no significant difference in the total volume of phagocytosed myelin fragments between the groups, likely due to the overall reduction in microglial number (Figure 5Q). Moreover, HDAC1/2 knockout increased the expression of the lysosomal marker CD68 in microglia surrounding the hematoma (Figure S5F,G), suggesting enhanced phagocytic activity. HDAC1/2 knockout also upregulated the expression of Mertk, a key protein involved in clearing cellular debris and hematomas (Figure S5I,J).
Inhibition of HDAC1/2 suppresses glycolysis in microglia. (A) The cofactor pocket of HK2 (PDB: 2OU2) bound to Origin HK2 (left) and Modified HK2 (right). HK2 is shown in cartoon representation (B). (C) Diagram of how HK2 acetylation affects the glycolytic pathway. (D) The expression of HK2 in BV2 cells following 24 h of Hemin stimulation. n=3 per group. (E) The enzyme activity of HK2 purified from BV2 cells. n=3 per group. (F) Schematic of plasmid construction for the HK2K866Q mutation. (G) The enzyme activity of HK2K866Q purified from HEK 293T cells. n = 3 per group. (H I J) Glycol‐PER measurement of BV2 cells following 24 h of Hemin stimulation. n = 3 to 4 per group. (K) Heatmap of expression differences in glucose transporter family genes. n = 4 per group, by figdraw. (L) 2‐NBDG glucose uptake test of BV2 cells, as reflected by the quantification (M). n = 3 per group. (M) Heatmap of expression differences in glucose transporter family genes. n = 4 per group. (N) The level of glucose of BV2 cells. n = 3 per group. (O) The level of L‐lactic acid of BV2 cells. n = 3 per group. (P) Representative PET‐CT scan images (normalized to cerebellum) of mice brain at 3 days after ICH, as reflected by the quantification (Q). n = 3 to 4 per group. All values are expressed as mean ± SD. The data were analyzed using one‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (D to O) or unpaired two‐ tailed Student's t test (Q). * p < 0.05, ** p < 0.01, *** p < 0.001, ns, not significant.
Finally, GSEA enrichment analysis revealed that Romidepsin treatment significantly enriched pathways involved in the positive regulation of G‐protein coupled receptor signaling, purinergic nucleotide receptor signaling, regulation of cytosolic calcium ion concentration, and cyclic nucleotide metabolic processes. These pathways play essential roles in regulating microglial migration, proliferation, phagocytosis, and inflammatory responses (Figure S4G–K).
Inhibition of HDAC1/2 Promoted the Metabolic Shift in Microglia From Glycolysis to Fatty Acid Oxidation
2.5
Microglia exhibit a glycolysis‐driven metabolic profile under stress and pro‐inflammatory conditions [26]. HK2, as the first key enzyme in the glycolytic pathway, plays a critical role in this process. The upregulation of HK2 activity and expression is considered a major factor contributing to the excessive glycolytic activity observed after microglial activation [17]. Using immunofluorescence and Western blot techniques, we confirmed that HK2 expression levels are upregulated in microglia in the ICH model, showing a time‐dependent increase followed by a decrease (Figure S6A,B,E,F,I,J,N,O). HK2 exerts its glycolytic role by binding to the mitochondrial membrane channel protein VDAC1, and we also observed an increased trend in VDAC1 expression in microglia following hemin stimulation (Figure S6G,H,K,P). Additionally, by inducing M1 and M2 microglia, we found that HK2 expression was relatively higher in M1 microglia (Figure S6L–P). In this study, we observed that HDAC1/2 inhibition led to a significant increase in acetylation levels of HK2 at the K866 site in microglia. Post‐translational modifications of proteins generally affect their structure and function [23, 27], and since cellular metabolism is closely linked to function, we believe the acetylation modification at the K866 site warrants further exploration in terms of its impact on HK2 enzymatic activity.
We first used AlphaFold2 to predict the structural changes of HK2 before and after acetylation at the K866 site (Figure 5A). Additionally, we employed the DoG‐Site‐Scorer online tool (https://proteins.plus/) to predict and visualize the binding pockets of both the unmodified and acetylated HK2 proteins. The results showed that acetylation at the K866 site led to a more compact binding pocket (Figure 5B). To investigate the impact of these structural changes on the catalytic activity of HK2, we purified HK2 from BV2 cells and measured its enzymatic activity. The findings revealed that inhibiting HDAC1/2 significantly reduced HK2 enzymatic activity without affecting its expression levels in the cells (Figure 5D,E). Furthermore, overexpression of the HK2 Kac^866^ mutant in HEK 293T cells also resulted in a decrease in enzymatic activity, confirming the observed effect (Figure 5F,G).
To more accurately assess changes in cellular glycolytic capacity, we used the Seahorse XF96 platform to evaluate the metabolic state of the cells (Figure 5H). In the glycolytic stress test, BV2 cells stimulated with hemin showed elevated basal and maximum extracellular acidification rates, which were significantly suppressed by HDAC1/2 inhibition (Figure 5I,J). Additionally, we observed upregulation of various GLUT isoforms, which may represent a compensatory response by microglia to the reduced glycolytic capacity and glucose uptake (Figure 5K). HDAC1/2 inhibition reduced the uptake of the fluorescent glucose analog 2‐NBDG in BV2 cells (Figure 5L,M). Furthermore, we assessed intracellular glucose and lactate levels as indicators of glycolysis. The results showed that inhibiting HDAC1/2 consistently increased intracellular glucose and lactate levels (Figure 5N,O). In vivo, PET‐CT imaging of ^18^F‐FDG uptake in the striatum region on day 3 after ICH revealed that HDAC1/2 knockout significantly inhibited ^18^F‐FDG uptake in the cells surrounding the hematoma (Figure 5K,L). Overall, these findings suggest that HDAC1/2 inhibition significantly impairs glycolytic efficiency in activated microglia.
To further elucidate the metabolic shift in microglia and clarify the roles of other metabolic pathways under HDAC1/2 inhibition, we expanded our focus to lipid metabolism, which is particularly relevant to microglia. First, to evaluate shifts in cellular metabolic patterns, we performed non‐targeted metabolomic analysis on three groups of microglial cells: BV2, BV2+hemin, and BV2+hemin+Romidepsin (Figure 6A). PCA analysis revealed distinct clustering patterns among the groups, indicating different metabolic profiles (Figure S6B). According to HMDB compound classification, these metabolites detected were mainly lipids and lipid‐related molecules, accounting for 41.28% (Figure S7B). Comparing the BV2+hemin and BV2+hemin+Romidepsin groups, we identified 410 differential metabolites, with 251 upregulated and 159 downregulated (Figure 6C,D, Figure S7D,E), and the KEGG pathway enrichment analysis resulted to glycerophospholipid metabolism, TCA cycle, biosynthesis of unsaturated fatty acids and HIF‐1 signaling pathway (Figure S7C,S8A). We further performed KEGG pathway enrichment analysis on the differentially expressed metabolites in microglia that exhibited opposing trends after Hemin stimulation and FK228 intervention. The analysis revealed that these genes were mainly clustered in pathways related to TCA cycle, lipolysis and lipid metabolism, Additionally, key signaling pathways, including the HIF‐1, PPAR, and autophagy pathways, were also enriched, all of which are important in regulating glycolysis and fatty acid oxidation (Figure 6F). After inhibiting HDAC1/2, the RNA and protein levels of enzymes related to fatty acid oxidation and lipid synthesis in microglia were significantly altered. Overall, enzymes involved in fatty acid oxidation, such as Cpt1a and Acadm, were upregulated, while enzymes related to lipid synthesis, such as FASN and G6pdx, were downregulated (Figure 6G,H). GSEA enrichment analysis indicated that after HDAC1/2 inhibition, microglia were significantly enriched in fatty acid oxidation, mitochondrial long‐chain fatty acid β‐oxidation and cytochrome P450 oxidation pathways (Figure 6I, Figure S7A).
Inhibition of HDAC1/2 enhanced fatty acid oxidation in microglia. (A) Schematic diagram of untargeted metabolomics in BV2 cells following 24 h of Hemin stimulation, by figdraw. (B) PCA plot of untargeted metabolomics for three groups of BV2 cells. n = 6 to 8 per group. (C) Bar chart of differential metabolites diagram. RT‐qPCR of key enzymes involved in fatty acid oxidation. n = 3 per group. (D) Volcano plot of differential metabolites between BV2+Hemin and BV2+Hemin +FK228. (E) Venn diagram of the intersection of differentially expressed metabolites between BV2 vs. BV2+Hemin and BV2+Hemin vs. BV2+Hemin+FK228. (F) Enrichment analysis of the differentially metabolites between BV2+Hemin and BV2+Hemin+FK228 in KEGG pathway. (G) Heatmap of key enzymes involved in fatty acid biosynthesis and fatty acid oxidation. (H) RT‐qPCR validation of the differentially expressed genes of fatty acid oxidation and biosynthesis. n = 3 per group. (I) GSEA analysis of mitochondrial long chain fatty acid β‐oxidation. (J) Flow cytometry detection of neutral lipid content in BV2 cells (labeled with BODIPY493/503 probe) representative figure, as reflected by the quantification (K). n = 3 per group. (L) Detection of acetyl‐CoA content in BV2 cells using ELISA following 24 h of Hemin stimulation. n = 3 per group. (M) Detection of free fatty acids content in BV2 cells following 24 h of Hemin stimulation. n = 3 per group. (N) Schematic diagram of enhanced fatty acid oxidation, by figdraw. (O) Representative images of Iba1/Bodipy‐493/503 (labeling lipid droplet) immunostaining around the hematoma, as reflected by the quantification (P). n = 5 mice per group. All values are expressed as mean ± SD. The data were analyzed using one‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (H, K, L, M) or unpaired two‐tailed Student's t test (P). * p < 0.05, ** p < 0.01, *** p < 0.001 ns, not significant.
Then, we assessed the overall lipid content of the cells and found that hemin‐induced microglia exhibited some degree of lipid increase, whereas lipid content decreased following HDAC1/2 inhibition, as indicated by a reduction in Bodipy493/503 fluorescence intensity (Figure 6J,K). Acetyl‐CoA, a crucial product of fatty acid β‐oxidation, enters the mitochondria for oxidative phosphorylation, providing energy for the cell. To assess fatty acid oxidation, we measured the levels of intracellular free fatty acids and acetyl‐CoA. The results revealed that inhibition of HDAC significantly reduced the levels of free fatty acids, while simultaneously increasing the content of acetyl‐CoA (Figure 6L,M). This change suggests that microglia, following HK2 acetylation‐induced impairment in glycolytic capacity, undergo a metabolic shift favoring fatty acid oxidation for energy production (Figure 6N). We also observed a significant accumulation of Bodipy493/503‐labeled lipid droplet in microglia around the hematoma after ICH, which showed a trend of initial increase followed by a decrease (Figure S7F). Additionally, knocking out HDAC1/2 reduced the number of lipid droplet in microglia (Figure 6O,P).
Ferroptosis, a form of regulated cell death linked to lipid‐related ROS accumulation [28], was another enriched pathway, which is highly relevant in the context of the ICH model. Enhanced fatty acid oxidation is closely associated with ferroptosis. Analysis of transcriptomic data further indicated significant differences in ferroptosis‐related genes such as Slc7a11, Slc48a1, and Gss among the three groups and enrichment analysis of differential metabolites showed that inhibition of HDAC1/2 significantly enriched the ferroptosis pathway, which suggesting that HDAC1/2 inhibition influences cellular ferroptosis (Figure S8B). Using the lipid ROS probe C11‐Bodipy 581/591 to detect intracellular lipid peroxides, we found that HDAC1/2 inhibition significantly increased lipid peroxidation levels in the cells (Figure S8C,D).
HDAC1/2 Inhibition Resulted in Mitochondrial Dysfunction and Mitophagy in Microglia
2.6
Studies have shown that HK2 binding to VDAC1 plays a crucial role in maintaining mitochondrial stability by reducing outer mitochondrial membrane permeability and decreasing mitochondrial ROS production. The K866 site of HK2, located within the Hexokinase large subdomain 2, is closely associated with ATP binding, phosphate group transfer, and interaction with VDAC1. Using AlphaFold2, we predicted the binding configuration of HK2 to VDAC1 before and after acetylation at the K866 site and visualized their docking poses.
The term “Piper pose energy” refers to a type of binding energy used to assess intermolecular interactions, measuring the binding strength between molecules (such as drugs) and target proteins [29]. A lower Piper pose energy value typically indicates a more stable interaction and stronger binding [30]. In our analysis, the Piper pose energy for acetylated HK2 at the K866 site with VDAC1 was higher, suggesting reduced binding stability in this modified state (Figure 7A,B). This result was further validated by co‐immunoprecipitation experiments, which showed that HDAC1/2 inhibition also reduced the binding of HK2 to VDAC1 (Figure 7C). The binding strength between HK2 Kac^866^ and VDAC1 also decreased (Figure 7D). Interestingly, Hemin stimulation in microglia did not significantly alter HK2's binding to VDAC1 (Figure 7C, Figure S9A,B). However, we observed differences in the binding strength between HK2 and VDAC1 across different microglial subtypes, with a notably weaker interaction in M2‐type microglia (Figure S9C,D).
Inhibition of HDAC1/2 resulted to mitochondrial dysfunction in microglia. (A) Docking pose energy analysis of HK2‐VDAC1(left) and HK2K866Q‐VDAC1(right) interaction using PIPER pose energy calculations. (B) Schematic diagram of mitochondrial dysfunction mediated by HK2 acetylation modification, by figdraw. (C) Co‐Immunoprecipitation of HK2 and VDAC1 from BV2 cells, as reflected by the quantification. n = 3 per group. (D) Co‐Immunoprecipitation of HK2 and VDAC1 from HEK 293T cells, as reflected by the quantification. n = 3 per group. (E F G H I) OCR detection and measurement of BV2 cells following 24 h of Hemin stimulation. n = 3 to 4 per group. (J) The mitochondrial membrane potential of BV2 cells following 24 h of Hemin stimulation. as reflected by the quantification (K). n = 3 per group. (L) The MFI of mitoSox of BV2 cells following 24 h of Hemin stimulation, as reflected by the quantification (M). n = 3 per group. (N) Flow cytometry analysis showing Ca2+ levels in BV2 microglia using Fluo‐4 AM dye, as reflected by the quantification (O). n = 3 per groups. (P) The ATP content of BV2 cells following 24 h of Hemin stimulation. n = 3 per group. All values are expressed as mean ± SD. The data were analyzed using one‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (C, F, G, H, I, K, M, O, P) or unpaired two‐ tailed Student's t test (D). p < 0.05, ** p < 0.01, *** p < 0.001 ns, not significant.*
Mitochondria are the core of cellular energy metabolism. To verify mitochondrial functional changes mediated by HDAC1/2 inhibition‐induced acetylation of HK2 at the K866 site, we conducted a series of mitochondrial function assays. First, we assessed the oxidative phosphorylation capacity of microglia using the Seahorse XF96 analyzer. The results showed that HDAC1/2 inhibition significantly reduced basal respiration, maximal respiration, and ATP production rates in microglia, indicating severe impairment of mitochondrial oxidative phosphorylation (Figure 7E–I). We then measured mitochondrial membrane potential. Under hemin‐induced conditions, we observed some degree of mitochondrial damage. HDAC1/2 inhibition further intensified this effect, with a significant increase in mean fluorescence intensity of mito‐tracker deep red, suggesting a complete collapse of mitochondrial membrane potential (Figure 7J,K). Additionally, mitochondrial superoxide accumulation provided further evidence of mitochondrial dysfunction, as we observed a significant accumulation of superoxide in both resting and activated microglia following HDAC1/2 inhibition (Figure 7L,M). Furthermore, we used Fluo4 AM to measure intracellular calcium levels and found that HDAC1/2 inhibition significantly increased intracellular calcium (Figure 7N,O). However, HDAC1/2 inhibition led to an increase in intracellular ATP levels (Figure 7P). Mitochondrial dysfunction may impact cell proliferation and survival. GSEA enrichment analysis revealed that the Apoptosis by TGFβ1 Via MAPK1 Up pathway was significantly enriched following HDAC1/2 inhibition (Figure S10A). After inhibiting HDAC1/2, the transcription levels of genes encoding proliferation‐related transcription factors, such as Spi1, Rela, and E2f1, were significantly downregulated. Meanwhile, genes and proteins associated with proliferation and apoptosis were significantly affected, specifically resulting in inhibited proliferation and enhanced apoptosis. Therefore, we then assessed Caspase‐3 expression levels and conducted Edu assays to measure cell proliferation (Figure S10B,C). The results showed that HDAC1/2 inhibition significantly increased apoptosis in microglia (Figure S10D,E) and suppressed their proliferation (Figure S10F,G), likely due to energy deficiency caused by mitochondrial dysfunction. Apoptosis may be induced by calcium overload resulting from decreased mitochondrial membrane potential and ROS accumulation.
Mitochondrial dysfunction often triggers protective mechanisms such as mitophagy or autophagy to degrade damaged mitochondria, thereby exerting a protective effect. Studies also have shown that enhanced mitophagy during the polarization of macrophages from M1 to M2 phenotype promotes mitochondrial turnover and metabolic adaptation, resulting in a reduced inflammatory response and an anti‐inflammatory phenotype [31, 32]. Transcriptomic and proteomic analysis indicated that HDAC1/2 inhibition significantly upregulated the transcription of autophagy‐related makers (Figure 8A). HDAC1/2 inhibition also led to significant enrichment of differentially expressed mRNAs, proteins, and metabolites in the autophagy pathway (Figure 4E–I and 6F). Notably, SQSTM1 and LC3B mRNA levels were significantly increased after HDAC1/2 inhibition (Figure 8B,C).
Inhibition of HDAC1/2 enhances microglial mitophagy. (A) Heatmap representing transcriptomic and proteomic changes in autophagy, apoptosis and cell proliferation pathways in BV2. (B C) The mRNA level of SQSTM1 and MAP1LC3B of BV2 cells following 24 h of Hemin stimulation. n = 3 per groups. (D) Flow cytometry analysis showing autophagy staining of BV2 cells with MDC following 24 h of Hemin stimulation. n = 3 per groups. as reflected by the quantification (E). n = 3 per group. (F) The MFI of lysosome in BV2 cells, as reflected by the quantification (G). n = 3 per groups. (H) Representative immunofluorescence images of Lysotracker and Mitotracker in BV2 cells, as reflected by the quantification (I J). n = 3 per group. (K) Representative immunofluorescence co‐localization image of Lysotracker and Mitotracker, as reflected by the quantification (L). n = 3 per group. (M) Schematic diagram of Ad‐mCherry‐GFP AAV virus labeling LC3B, by figdraw. (N) Representative images of microglial mitophagy following 24 h of Hemin stimulation, as reflected by the quantification (O P Q R). n = 3 per group. All values are expressed as mean ± SD. The data all were analyzed using one‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test. * p < 0.05, ** p < 0.01, *** p < 0.001 ns, not significant.
To further evaluate autophagy levels, we used MDC staining and found that HDAC1/2 inhibition induced autophagy (Figure 8D,E). We then labeled lysosomes with Lyso‐Tracker Red and mitochondria with Mito‐Tracker Green to assess mitophagy. Results showed that HDAC1/2 inhibition increased the average fluorescence intensities of both lysosome and mitochondrial markers (Figure 8F–J), and fluorescence co‐localization analysis confirmed an increase in mitochondria‐lysosome autophagy following HDAC1/2 inhibition (Figure 8K,L). Additionally, by infecting cells with Ad‐mCherry‐GFP‐LC3B (Figure 8M), we observed a significant increase in LC3B expression within mitochondria after HDAC1/2 inhibition, manifested as red fluorescence accumulation in the mitochondria (Figure 8N–R).
Inhibition of HK2 Regulated the Functional Phenotype of Microglia and Alleviated Neurological Deficits After ICH
2.7
To verify the regulatory effects of weakened glycolytic capacity mediated by the acetylation modification of the HK2 K866 site on microglia, we decided to simulate this effect in vivo and in vitro. First, we evaluated the effects of the HK2 inhibitor 2‐DG, a glucose analog that competitively inhibits glucose metabolism (Figure 9A) [15]. The results showed that 2‐DG intervention could alleviate short‐term neurological dysfunction in mice following ICH (Figure 9B,C). Additionally, 2‐DG intervention reduced the number of microglia surrounding the hematoma during the acute phase (Figure 9D,F), as well as decreased the proportion of Ki67^+^ microglia, indicating that microglial proliferative activity was inhibited (Figure 9H–J). Sholl analysis revealed that the branches of microglial cells increased in the 2‐DG‐treated group (Figure 9E,G). Immunofluorescence staining showed that inhibiting HK2 significantly suppressed the pro‐inflammatory phenotype of microglial cells surrounding the hematoma, as indicated by the reduction in CD16 expression in microglial cells (Figure 8K,L). Additionally, 2‐DG intervention did not affect the average fluorescence intensity of VDAC1 in microglia (Figure S10A,D), but increased the average fluorescence intensity of Mertk and the proportion of IL10^+^ microglia (Figure S10B,C,E,F).
Inhibition of HK2 can protect against ICH‐mediated neurological deficits and suppress microglial activation. (A) Experimental timeline planning diagram. (B C) Behavioral tests at 3 days after ICH in mice treated with 2‐DG, n = 5 to 6 per group. (D) Representative images of Iba1 to mark microglia around the hematoma, as reflected by the quantification (F). n = 5 to 6 per group. (E) Representative images of Sholl‐analysis of microglia, as reflected by the quantification (G). n = 30 cells per group. (H I) Representative images of Iba1/Ki67 immunostaining around the hematoma, as reflected by the quantification (J). n = 6 per group. (K) Representative images of Iba1/CD16 immunostaining around the hematoma, as reflected by the quantification (L). n = 5 to 6 per group. (M) Schematic diagram of 3BP intervention in mouse model of ICH, by figdraw. (N) Representative images of Iba1/Ki67 immunostaining around the hematoma, as reflected by the quantification (P). n = 5 to 6 per group. (O) Representative images of Iba1/Mertk immunostaining around the hematoma, as reflected by the quantification (Q). n = 5 to 6 per group. All values are expressed as mean ± SD. The data were analyzed using one‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (B, C, F, L, P, Q) or two‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test (G) or unpaired two‐tailed Student's t test (J). * p < 0.05, ** p < 0.01, *** p < 0.001 ns, not significant.
3‐Bromopyruvate (3‐BP), a potent HK2 inhibitor, can effectively suppress the high metabolic activity of cancer cells [33]. We explored its neuroprotective effects in mice following ICH through intraperitoneal injection of 3BP (Figure 9M). After one week of continuous intraperitoneal injections of 3BP [16], we established the ICH model and assessed the functional state of microglial cells surrounding the hematoma on the third day post‐ICH. The results showed that compared to the control group, the mice that received 3BP injections exhibited significant weight loss prior to the construction of the ICH model, but their rate of weight loss after the ICH was lower than that of the control group (Figure S10H,I). Moreover, microglia surrounding the hematoma in the 3BP intervention group displayed lower proliferative activity (Figure 9N,P), reduced pro‐inflammatory phenotype (Figure S10 J–O), and higher expression of Mertk, a phagocytosis marker (Figure 9O,Q).
To investigate the role of HK2 in activated microglia in vitro, we first performed HK2 gene knockout in the BV2 cell line using SgRNA. Specifically, we incubated the BV2 cells with the plasmid (U6‐sgRNA‐EF1a‐Cas9‐FLAG‐P2A‐puro) for 36 h, followed by Western blot analysis to confirm the effectiveness of the HK2 knockout (Figure 10A). Subsequently, we examined the polarization of BV2 cells. RT‐qPCR results showed that HK2 knockout significantly suppressed the pro‐inflammatory phenotype of microglia (such as CD16, IL6, IL1β, and iNOS) while enhancing their anti‐inflammatory phenotype (such as CD206, IL10, and TGFβ) (Figure 10B,C). Additionally, we assessed the phagocytic function of BV2 cells. After co‐incubation with pHrodo Red Bioparticles for 2 h, we used flow cytometry, combined with immunofluorescence staining, to evaluate the phagocytic ability of BV2 cells. The results demonstrated that HK2 knockout enhanced phagocytosis to some extent, as indicated by an increased proportion of pHrodo Red Bioparticles positive cells (Figure 10D,E) and a higher number of pHrodo Red Bioparticles within individual cells (Figure 10F,G).
Inhibition of HK2 modulates microglia phenotype and functions in BV2 cells. (A) Western blot was used to test the effect of SgRNA‐mediated knockdown of HK2 in BV2 cells. (B) RT‐qPCR for the anti‐inflammatory genes in BV2 cells. n = 3 per group. (C) RT‐qPCR for the pro‐inflammatory genes in BV2 cells. n = 3 per group. (D) Flow cytometry was used to assess the phagocytic ability of BV2 cells with Latex beads red, as reflected by the quantification (E). n = 3 per group. (F) Representative images showing Latex beads red, engulfment in BV2 cells, with the cytoskeleton stained using Phalloidin‐488, as reflected by the quantification (G) n = 3 per group. (H) Flow cytometry analysis showing Ca2+ levels in BV2 cells using Fluo‐4 AM dye, as reflected by the quantification (I). n = 3 per groups. (J) The MFI of mitoSox of BV2 cells following 24 h of Hemin stimulation, as reflected by the quantification (K). n = 3 per group. (L) Flow cytometry analysis showing autophagy staining of BV2 cells with MDC following 24 h of Hemin stimulation. n = 3 per groups. as reflected by the quantification (M). n = 3 per group. (N) Representative immunofluorescence co‐localization image of Lysotracker(green) and TOM20(red), as reflected by the quantification (O). n = 3 per group. All values are expressed as mean ± SD. The data were all analyzed using one‐way analysis of variance (ANOVA) followed by Bonferroni's post hoc test. * p < 0.05, ** p < 0.01, *** p < 0.001, ns, not significant.
We also measured intracellular calcium ions and mitochondrial superoxide levels, and the results were consistent with the effects observed after HDAC1/2 inhibition. HK2 knockout similarly induced an increase in intracellular calcium concentration (Figure 10H,I) and mitochondrial superoxide levels (Figure 10J,K). Furthermore, we observed that HK2 knockout enhanced autophagy in BV2 cells (Figure 10L,M), as well as mitochondrial autophagy (Figure 10N,O).
Consistent with previous studies, HK2 knockout disrupted the mitochondrial function of microglia, likely due to HK2's role in stabilizing mitochondrial function by binding to VDAC1 [15]. However, we also found that HK2 knockout enhanced autophagic capacity in BV2 cells, which may mitigate some of the negative effects associated with mitochondrial dysfunction.
To further investigate the role of HK2 in the protective effects mediated by HDAC1/2 ablation, we employed a recently developed recombinant AAV11 (rAAV11) vector with high microglial specificity to package rAAV‐SFFV‐DIO‐mHK2‐3×FLAG‐WPRE‐4×miR9T (hereafter referred to as rAAV‐HK2‐OE). This construct enabled selective and efficient overexpression of HK2 in CreERT‐expressing microglia. Stereotaxic injections of rAAV‐HK2 were performed at two distinct sites in the peri‐infarct striatum, followed by tamoxifen administration one week later (Figure S12A). As a negative control, we used rAAV‐SFFV‐DIO‐mHK2‐EGFP‐WPRE‐4×miR9T (hereafter referred to as rAAV‐CTL). Immunofluorescence co‐staining of Iba1 with EGFP or Flag revealed that 70%–80% of microglia in the peri‐infarct striatum were positive for viral reporters (Figure S12B), indicating efficient microglial infection. Moreover, immunofluorescence quantification confirmed that rAAV‐HK2 successfully induced robust HK2 overexpression in HDAC1/2‐deficient microglia (Figure S12D,E). Notably, HK2 overexpression significantly reversed the phenotype observed in HDAC1/2‐deficient microglia (Figure S12F,G). Behaviorally, both rotarod and foot‐fault assays demonstrated that HK2 overexpression aggravated motor dysfunction in mice after ICH (Figure S12H,I). Consistent with these findings, susceptibility‐weighted imaging (SWI) revealed slower hematoma clearance in the rAAV‐HK2 group compared with controls (Figure S12J,K). Together, these results provide strong evidence that HK2 acts as a downstream mediator of HDAC1/2, contributing to the selective inhibition of microglial function and to the protective effects of HDAC1/2‐miKO following ICH.
Discussion
3
In this study, we explored the role of HDAC1/2 in regulating microglial function and mitochondrial metabolism following ICH. Our findings suggested that the knockout of HDAC1/2 in microglia alleviated neurological dysfunction and improved white matter integrity after ICH, largely through a shift in metabolic activity and alterations in microglial activation. Additionally, HDAC1/2 inhibition enhanced hematoma clearance, promoted neuroprotection, and reduced pro‐inflammatory responses. These results provided insights into the potential therapeutic strategies targeting HDAC1/2 to mitigate ICH‐related damage.
Neuroprotection and White Matter Integrity
3.1
One of the key findings of this study was that HDAC1/2 knockout improved white matter integrity and accelerated hematoma clearance following ICH. Using immunofluorescence staining, we demonstrated that HDAC1/2 knockout mice exhibited increased neuronal survival around the hematoma, as well as reduced demyelination and axonal damage, as evidenced by decreased SMI‐32 expression and improved MBP and NF‐H staining. These results were further corroborated by diffusion tensor imaging (DTI), which showed that HDAC1/2 knockout alleviated the loss of white matter tracts, with increased fractional anisotropy (FA) values observed in the peri‐hematoma area.
The improved white matter integrity in HDAC1/2 knockout mice was likely due to multiple factors, including reduced inflammation, enhanced microglial phagocytosis, and decreased oxidative stress. Previous studies demonstrated that microglia played a critical role in clearing debris and facilitating tissue repair following brain injury. In our study, HDAC1/2 knockout enhanced the ability of individual microglia to phagocytose myelin fragments, as indicated by increased CD68 expression and dMBP‐labeled debris uptake. This enhanced phagocytic capacity may have compensated for the reduced number of microglia observed in the knockout mice. This study found that inhibiting HDAC1/2 leads to a decrease in HK2 enzymatic activity. Previous research has also indicated that inhibition or knockout of HK2 can enhance microglial phagocytosis of Aβ and ATP production [16]. This suggests a complex mechanism by which HDAC1/2 regulates microglial function. Additionally, the reduced inflammatory response observed in HDAC1/2 knockout microglia, as indicated by decreased expression of pro‐inflammatory markers such as TNF‐α, IL‐6, and Fcgr3, further supported the protective role of HDAC1/2 inhibition in promoting tissue repair and reducing secondary brain damage after ICH.
Behavioral tests conducted in this study further confirmed the neuroprotective effects of HDAC1/2 inhibition. HDAC1/2 knockout mice exhibited significant improvements in motor function, as demonstrated by their performance in the rotarod, foot‐fault, and hanging wire tests. Moreover, these mice showed enhanced cognitive recovery with improved performance in the Morris water maze test, suggesting that HDAC1/2 inhibition not only promoted structural recovery but also restored functional outcomes after ICH.
The accelerated hematoma clearance observed in HDAC1/2 knockout mice, as seen in the T2‐weighted MRI scans, may also have contributed to the improved functional recovery. Hematoma clearance was critical in reducing mass effect and preventing further neuronal injury, and the enhanced phagocytic activity of microglia likely played a role in this process. Overall, the behavioral improvements observed in HDAC1/2 knockout mice highlighted the potential of targeting HDAC1/2 as a therapeutic strategy for enhancing recovery after ICH.
HDAC1/2 and Microglial Metabolic Reprogramming
3.2
The role of microglia in ICH has been extensively studied, with evidence supporting their dual role in either exacerbating or mitigating brain injury, depending on their activation state [10, 34]. Microglia exhibit distinct metabolic characteristics with notable subtype heterogeneity [35]. Pro‐inflammatory microglia primarily rely on glycolysis and glutamine metabolism, whereas anti‐inflammatory microglia derive their energy mainly from fatty acid oxidation and oxidative phosphorylation [19, 20, 35]. This distinct metabolic pattern provides a critical target for the precise regulation of microglial phenotypes. Current research shows that the high glycolytic activity of pro‐inflammatory microglia largely depends on hexokinase 2 (HK2), the first key enzyme in the glycolytic pathway, which initiates glycolysis by phosphorylating glucose [15, 17, 36]. Although HDAC involvement in metabolic regulation has been well established, the underlying mechanisms remain unclear.
Here, we found that HDAC1/2 knockout/inhibition leads to reduced HK2 enzymatic activity, likely due to conformational changes caused by its acetylation. The Seahorse XF96 metabolic assays revealed a significant decrease in the glycolytic proton efflux rate and a reduction in lactate, the end product of glycolysis, indicating severe glycolytic pathway disruption in microglia. We found that HDAC1/2 knockout led to a reduction in ^18^FDG‐PET signals in the striatum region after ICH. Previous studies have indicated that microglial activation and inflammation levels directly impact ^18^FDG‐PET signals in patients with neurodegenerative diseases [37]. Therefore, we can infer that the reduced ^18^FDG‐PET signal in this study partially reflects the suppression of microglial activation and the inhibition of HK2. To verify the change in enzyme activity caused by acetylation at the HK2 K866 site, we constructed a Q mutation plasmid to mimic acetylation at this site and overexpressed the protein in HEK293T cells, confirming a decrease in enzyme activity. Additionally, multi‐layered analyses of genes, proteins, and metabolites showed a marked metabolic shift in microglia following HDAC1/2 inhibition. Along with reduced glycolytic flux, we also observed enhanced fatty acid oxidation, indicative of a shift toward an anti‐inflammatory microglial phenotype, as well as complex alterations in amino acid metabolism, which remain underexplored.
The enhancement of fatty acid oxidation drives microglia toward an anti‐inflammatory phenotype. We also found significant lipid droplet accumulation in microglia around the hematoma, with the extent of accumulation decreasing over time. Furthermore, knockout of HDAC1/2 led to a significant reduction in the number of lipid droplets in microglia. Literature indicates that lipid droplet accumulation suggests a pro‐inflammatory state and phagocytic dysfunction in cells, and this phenomenon has been observed in Alzheimer's disease (AD) and aging [38, 39, 40]. In this study, the knockout of HDAC1/2 may reduce intracellular lipid droplet deposition by regulating fatty acid oxidation, thus modulating the inflammatory phenotype of microglia. This metabolic shift may result from a compensatory response triggered by decreased glycolytic capacity due to reduced HK2 activity or from transcriptional changes in metabolic enzyme‐encoding genes regulated by HDACs. Studies have found that in AD disease models, inhibiting HK2 in microglia can upregulate Lipoprotein Lipase (LPL) expression to trigger fatty acid metabolism [16]. This change in gene expression may result from upstream regulation, prompting us to examine changes in histone acetylation. The results showed that HDAC1/2 inhibition increased the acetylation level of histone H4K16, a marker of chromatin opening and active gene transcription [41], suggesting that the observed metabolic changes may be a result of transcriptional activation of specific genes. Cleavage Under Targets and Tagmentation (CUT&Tag) could further clarify which genes are directly regulated. Increased acetyl‐CoA levels and upregulation of key enzymes in the TCA cycle are also important indicators of enhanced fatty acid oxidation.
Moreover, various signaling pathways are likely involved in these profound metabolic shifts. We found that HDAC1/2 inhibition downregulated the HIF‐1 and NF‐κB signaling pathways while upregulating the PPAR and FOXO pathways [42, 43]. Collectively, this complex network of signaling pathways orchestrates the metabolic reprogramming of microglia.
Based on the reduction in activity caused by the acetylation of HK2 and the protective effects observed with HK2 inhibition/knockout in various disease models, we selected two representative HK2 inhibitors to explore their efficacy in ICH. In our animal experiments, we found that both inhibitors had a significant effect on microglial phenotype, primarily by inhibiting microglial proliferation and pro‐inflammatory phenotypes during the acute phase of ICH. This provides a new therapeutic option for the clinical treatment of ICH. Currently, 2‐DG is often used as an adjunct therapy in the radiation and chemotherapy of cancer patients (e.g., glioma, pancreatic adenocarcinoma, lung cancer, and breast cancer) to inhibit tumor cell growth, aiming to increase tumor cells' sensitivity to these treatments [44, 45, 46]. This is due to 2‐DG's ability to weaken tumor cell metabolism, making them more susceptible to the effects of radiotherapy and chemotherapy [46]. However, the widespread use of 2‐DG still faces challenges, particularly its potential toxicity and side effects on normal cells. In addition, we acknowledge that starting 2DG three days pre‐ICH is not clinically matched. We used pretreatment to ensure robust inhibition of microglial glycolysis during the acute phase and to assess off‐target effects in naïve mice; importantly, this did not change our overall conclusions. The regimen was adapted from prior studies initiating 2DG three days before inflammatory challenge [35, 47], and future work will test post‐ICH dosing. In this study, we observed significant weight loss in mice that received intraperitoneal injections of 3BP, suggesting that it affects the metabolism of normal cells, which could present some side effects if applied clinically to ICH patients. Therefore, further studies are needed to optimize the safe dosage and application methods of these two HK2 inhibitors.
Mitochondrial Dysfunction and Autophagy
3.3
Mitochondria are the powerhouse of cellular energy metabolism and are essential for cell survival. Our study demonstrates that HDAC1/2 inhibition exacerbates mitochondrial dysfunction in microglia, as evidenced by decreased mitochondrial membrane potential, increased mitochondrial superoxide production, and impaired oxidative phosphorylation, ultimately leading to intracellular calcium overload and apoptosis. After HDAC1/2 inhibition, we observed enhanced fatty acid oxidation in microglia, suggesting a shift toward an M2 phenotype. However, fatty acid oxidation inevitably increases intracellular ROS levels, thereby increasing mitochondrial stress. In this context, we detected significantly elevated lipid ROS levels in microglia, indicating that mitochondrial dysfunction may be driven by excessive fatty acid oxidation in a high ROS environment, which weakens the natural resistance of microglia to ferroptosis. Studies have shown that M1 microglia are more resistant to ferroptosis [48, 49, 50], and our research also found that hemin‐stimulated microglia exhibited significantly upregulated transcription levels of several anti‐ferroptosis genes, such as SLC7A11. However, HDAC1/2 inhibition‐induced lipid peroxidation promoted ferroptosis in microglia. Based on the subtype differences in ferroptosis resistance, we speculate that the loss of ferroptosis resistance due to HDAC1/2 inhibition is primarily reflected in the apoptosis of M1 microglia. Moreover, HK2 interacts with the mitochondrial channel protein VDAC1, coupling glycolysis and oxidative phosphorylation, while stabilizing the mitochondrial membrane potential [51]. In this study, we found that acetylation at the K866 site of HK2 weakened its interaction with VDAC1, which increased mitochondrial instability to some extent. Furthermore, compared to M2 microglia, we observed a stronger interaction between HK2 and VDAC1 in M1 microglia. Therefore, we believe that the impact of HDAC1/2 inhibition on M1 microglia, mediated through HK2 acetylation and mitochondrial membrane potential imbalance, is more pronounced. In our study, we observed elevated levels of apoptosis in microglia (with significantly increased Caspase‐3 expression), along with reduced proliferation (decreased levels of Ki67, pH3, and Edu‐positive cells). GSEA analysis also highlighted pathways enriched in TGF‐β‐mediated apoptosis and autophagy. These changes led to a reduction in the number of microglia, serving as an effective control mechanism for highly proliferative pro‐inflammatory microglia during the acute phase. Histone deacetylase inhibitors can reinstate the expression of epigenetically silenced cell death programs [52]. In our study, we validated this effect in a hemorrhagic stroke disease model and provided a new perspective on this action, namely the HK2‐mitochondrial axis, which adds a non‐epigenetic regulatory angle.
Additionally, we observed a significant increase in autophagy and mitophagy in microglia following HDAC1/2 inhibition. Autophagy may be enhanced through transcriptional pathways in response to HDAC1/2 inhibition [52, 53, 54, 55], while increased fatty acid oxidation has contributed to oxidative stress, which subsequently triggers autophagy. The enhanced autophagy and mitophagy help to clear excess lipid peroxides, damaged proteins, and mitochondria, reducing ROS accumulation in microglia, thereby suppressing inflammation and promoting the shift from M1 to M2 microglial phenotype [31, 32]. This transformation is typically accompanied by reduced inflammation and enhanced repair functions.
In summary, the inhibition of HDAC1/2 has broad effects on microglia, ultimately leading to a shift in their metabolic profile, a reduction in cell numbers, and a corresponding decrease in neuroinflammation.
Limitations
4
Our study has several limitations. First, while we observed a reduction in microglia‐mediated inflammation following HDAC1/2 inhibition, likely due to decreased microglial proliferation and increased apoptosis, this effect is complex given the heterogeneity of microglial subpopulations. Although the decrease in M1‐type microglia appears beneficial, we could not determine whether this effect is specific to particular microglial subtypes. Second, although we confirmed the protective effects of HK2 inhibition on ICH and its regulatory role in microglia through animal experiments, we were unable to verify these findings in a microglia‐specific HK2 knockout mouse model due to resource constraints. Third, HDAC1/2 significantly influences transcriptional regulation, and we observed altered expression of several metabolism‐related genes in this study. Additionally, we detected acetylation modifications at multiple histone lysine residues, such as H4K16; however, due to technical limitations, we could not employ Cut&Tag or ATAC‐seq to pinpoint the specific downstream genes affected. Additionally, in the animal experiments, we used only one modeling approach to observe the protective effects of HDAC1/2 knockout on neurological deficits following ICH, which has certain limitations. Thus, further research is necessary to validate our observations and explore their underlying mechanisms in greater depth.
Conclusion
5
In conclusion, our study demonstrated that HDAC1/2 plays a critical role in regulating microglial metabolism, mitochondrial function, and inflammatory responses following ICH. HDAC1/2 knockout, by targeting HK2, promoted a metabolic shift from glycolysis to fatty acid oxidation, inhibited microglial proliferation, alleviated neuroinflammation, preserved white matter integrity, and enhanced functional recovery. These findings suggest that inhibition of HDAC1/2 and HK2 represents a promising therapeutic target for mitigating brain damage and promoting recovery after ICH. Further research is warranted to elucidate the precise molecular mechanisms underlying HDAC1/2‐mediated metabolic regulation and its impact on microglial function.
Methods
6
Ethics Statement
6.1
The animals were housed under controlled temperature and humidity conditions with a 12 h light‐dark cycle. Food and water were provided ad libitum. All animal experiments were conducted with the approval of the Animal Care and Use Committee of Fudan University (approval number: 2021JS‐Huashan Hospital‐325) and adhered to the ARRIVE guidelines [56]. Every effort was made to reduce animal suffering and minimize the number of animals used in the study.
Experimental Animals
6.2
HDAC1/2^flox/flox^ mice were constructed by Shanghai Model Organisms Center. CX3CR1^CreER^ mice were purchased from The Jackson Laboratory. Two groups of CX3CR1^CreER(−/−)^ HDAC1/2^flox/flox^ (WT) and CX3CR1^CreER(+/−)^ HDAC1/2^flox/flox^ (HDAC1/2‐miKO) mice were bred from HDAC1/2^flox/flox^ and CX3CR1^CreER^ mice. Microglial HDAC1/2 depletion was induced in HDAC1/2‐miKO mice (8‐week‐old males) by intraperitoneal injection of tamoxifen (100 mg/kg, daily for 5 consecutive days), which was dissolved in corn oil and absolute ethanol in a ratio of 9:1 (H7904, Sigma‐ Aldrich, St. Louis, MO, USA).
Drug Preparation and Administration
6.3
For animal study, 3‐Bromopyruvate (3‐BP, Selleck, S5426) was dissolved in solvent (5% dimethyl sulfoxide (DMSO) + 30% polyethylene glycol (PEG) 300 + 5% Tween 80) and administered IP at 10 mg/kg with mouse for 7 days [16]. A 100 mg/mL solution of 2‐Deoxy‐D‐glucose (2DG, MCE, HY‐13966) was prepared by dissolving it in sterile PBS (with sonication to facilitate dissolution). The solution was then administered to mice via intraperitoneal injection at a dose of 400 mg/kg. Treatment started 3 days prior to modeling and continued daily until sacrifice [35, 47]. The HDAC1/2 inhibitor Romidepsin (FK228, Abmole, M2007) was administered intraperitoneally at 0.5 mg/kg immediately after ICH and biweekly for 2 weeks, while controls received a 10% DMSO injection [57].
Isolation and Culture of Primary Microglia
6.4
Primary microglia were isolated from neonatal C57BL/6 mice as previously described [10]. Briefly, the forebrains were dissociated with 0.125% trypsin (Thermo Fisher Scientific) and DNase (Sigma‐Aldrich), followed by centrifugation at 1500 rpm for 15 min. The cell pellet was resuspended in culture medium and filtered through a 40‐µm cell filter (Millipore). The cell mixture was then incubated in a poly‐D‐lysine‐coated flask for 10 days. Microglia were harvested by shaking the flask at 220 rpm for 1 h at 37°C. The supernatant containing microglia was collected and centrifuged at 1000 rpm for 5 min. Finally, the microglia were cultured in DMEM/F12 (Thermo Fisher Scientific) with 10% FBS.
The in Vitro ICH Model
6.5
For vitro model, the BV2 cell or primary microglia was stimulated Hemin (Sigma‐Aldrich, 51280, 20 uM) and FK228 (Abmole, M2007, 10 ng/mL) for 24 h [24]. The following assays were performed 24 h later.
Behavioral Tests
6.6
The foot‐fault test, rotarod test, and wire‐hanging test were conducted on days 1, 3, 7, 21, and 35 post‐ICH, following previously described protocols [25]. The Morris water maze test was performed between days 29 and 34 after ICH to evaluate long‐term learning and cognitive deficits [25]. All behavioral assessments were conducted by an investigator blinded to the experimental conditions.
Immunofluorescence and Image Analysis
6.7
Immunofluorescence staining was performed following established protocols (25). Briefly, frozen coronal brain sections (25 µm thick) were prepared, blocked with 5% donkey serum, and incubated with primary antibodies overnight at 4°C in a solution containing 0.3% Triton X‐100 in PBS and 1% goat or donkey serum. After thorough washing, the sections were treated with secondary antibodies at room temperature for 2 h. Following additional washes, the slices were mounted with DAPI Fluoromount‐G (36308ES20, Yeasen, Shanghai, China). Images were acquired using a Nikon AX confocal microscope and analyzed with ImageJ and Imaris software.
CAP Measurements
6.8
Compound action potentials (CAPs) were recorded from the entorhinal cortex (EC) 35 days following ICH or Sham operation. Mice were anesthetized with isoflurane and quickly decapitated for brain extraction. Coronal brain slices (350 µm thick) were prepared using a vibrating microtome (1200s, Leica). The slices were then transferred to artificial cerebrospinal fluid (aCSF) and saturated with a 95% oxygen and 5% CO_2_ gas mixture at 32°C for 30 min, followed by a 1 h recovery period at room temperature. CAPs were elicited using a concentric stimulation electrode, and recorded with a glass microelectrode (5–8 MΩ). Stimulation currents ranging from baseline up to 2 mA (in 0.25‐mA increments) were applied. The amplitudes of the N1 (myelinated fibers) and N2 (unmyelinated fibers) components of the CAPs were analyzed using pClamp 10 software (Molecular Devices, San Jose, CA, USA).
Quantitative Real‐Time PCR
6.9
Total RNA from the infarcted brain tissue was extracted using TRIzol reagent (19201ES60, Yeasen, Shanghai, China). RNA from flow‐sorted microglia was isolated with the RNeasy Plus Micro Kit (74030, Qiagen, Germany). Reverse transcription was conducted according to the manufacturer's instructions using a reverse transcription kit (K1622, Thermo‐Fisher Scientific, Pittsburgh, PA, USA) to generate cDNA. Real‐time qPCR was performed using Hieff QPCR SYBR Green Master Mix (11201ES08, Yeasen, Shanghai, China) as the detection dye. The qPCR reaction conditions were set as follows: an initial denaturation at 95°C for 5 min, followed by 40 cycles of 95°C for 10 s, 55°C for 20 s, and 72°C for 20 s. Each sample was run in triplicate, and the relative mRNA expression levels were normalized to Gapdh.
4D‐FastDIA Quantitative Acetyl‐Proteomics Analysis
6.10
Cells were ground into powder using liquid nitrogen and transferred to a 5‐mL centrifuge tube. Lysis buffer (4 times the volume of the powder) was added, followed by sonication for 3 min on ice. For PTM experiments, inhibitors were included (e.g., 3 µM TSA and 50 mM NAM for acetylation). The mixture was centrifuged at 12 000 g for 10 min at 4°C, and the supernatant was collected. Protein concentration was determined using the BCA kit. The sample was precipitated with 20% TCA, vortexed, and incubated for 2 h at 4°C. The precipitate was collected by centrifugation, washed with cold acetone three times, and dried. Proteins were dissolved in 200 mM TEAB, sonicated, and digested overnight with trypsin (1:50). The sample was reduced with dithiothreitol, alkylated with iodoacetamide, and desalted using a Strata X SPE column. Tryptic peptides were dissolved in NETN buffer and incubated overnight with antibody beads at 4°C. The beads were washed, and bound peptides were eluted with 0.1% trifluoroacetic acid. The eluted peptides were vacuum‐dried and desalted using C18 ZipTips (Millipore) before LC‐MS/MS analysis.
Tryptic peptides were dissolved in solvent A and loaded onto a home‐made reversed‐phase column (25 cm length, 100 µm i.d.). Peptide separation was done using a gradient: 0–18 min, 6%–22% B; 18–22 min, 22%–30% B; 22–26 min, 30%–80% B; 26–30 min, 80% B, at a flow rate of 450nl/min on a NanoElute UHPLC system. The peptides were analyzed using a capillary source and timsTOF Pro mass spectrometer in dia‐PASEF mode. The MS scan range was set to 100‐1700, and 8 PASEF‐MS/MS scans were acquired per cycle, with an MS/MS range of 425‐1025 and an isolation window of 25 m/z.
DDA data were processed with Spectronaut (v.17.0) software using the Pulsar search engine. Tandem mass spectra were searched against Mus_musculus_10090_SP_20230103.fasta (17132 entries) with a reverse decoy database. Carbamidomethylation on Cys was set as a fixed modification, while acetylation on the protein N‐terminal, oxidation on Met, and acetylation were set as variable modifications. The FDR for proteins, peptides, and PSMs was adjusted to < 1%. The spectral library was imported into Spectronaut for retention time prediction and DIA data search.
4D‐FastDIA Proteomics Analysis
6.11
The sample was ground with liquid nitrogen into a powder, transferred to a 5‐mL tube, and mixed with lysis buffer (4x volume of powder) containing 8 M urea and 1% protease inhibitor. The mixture was sonicated on ice for 3 min. For PTM experiments, inhibitors (e.g., TSA, NAM, phosphatase inhibitors) were added. After centrifugation (12 000 g, 10 min, 4°C), the supernatant was collected, and protein concentration was determined using the BCA kit.
The sample was mixed with 20% TCA to precipitate proteins, vortexed, and incubated at 4°C for 2 h. The precipitate was collected by centrifugation, washed with cold acetone, and dried. The protein was dissolved in 200 mM TEAB and digested overnight with trypsin (1:50 ratio). The sample was reduced with dithiothreitol (5 mM) and alkylated with iodoacetamide (11 mM). Finally, peptides were desalted using Strata X SPE columns.
Tryptic peptides were dissolved in solvent A and loaded onto a homemade reversed‐phase column. Peptides were separated using a gradient on a NanoElute UHPLC system. The peptides were analyzed by timsTOF Pro mass spectrometry in dia‐PASEF mode with a 1.75 kV electrospray voltage. The MS scan range was 300–1500, with 20 PASEF MS/MS scans per cycle (scan range: 400–850, isolation window: 7 m/z).
DDA data were processed with Spectronaut (v.17.0) and searched against Mus_musculus_10090_SP_20230103.fasta with a reverse decoy database. Carbamidomethylation on Cys was fixed, and acetylation on the N‐terminus and oxidation on Met were set as variable modifications. The FDR for proteins, peptides, and PSMs was adjusted to <1%. The spectral library was imported into Spectronaut to predict retention times and search against DIA data.
Western Blotting
6.12
Cellular proteins were extracted and quantified using the bicinchoninic acid (BCA) method. Subsequently, 40 µg of protein was separated by electrophoresis and transferred onto membranes. The membranes were blocked with 5% bovine serum albumin for 1 h, then incubated overnight at 4°C with rabbit primary antibodies. The following day, the membranes were incubated with rabbit secondary antibodies at 37°C for 1 h, followed by visualization using enhanced chemiluminescence (ECL) reagent.
Immunoprecipitation
6.13
After perfusion, peri‐hematoma tissue from the ipsilateral hemisphere was collected, homogenized, and ultrasonicated using Lysis Buffer for WB/IP Assays (Cat# P0013, Beyotime, Shanghai, China) supplemented with 1 mM PMSF to obtain whole protein extracts. Protein A/G Magnetic Beads (Cat# P2108, Beyotime) were then incubated with either anti‐target protein antibody or IgG for 10 min before being added to the clarified cell lysates. The mixture was allowed to incubate at room temperature for 1 h. Immunoprecipitated complexes were eluted using 20 mM glycine (pH 2.0) and neutralized with 1 M phosphate buffer (pH 7.4). The samples were then mixed with SDS‐loading buffer and heated at 95°C for 10 min for subsequent Western blot analysis.
MRI Measurement
6.14
For in vivo MRI, mice were initially anesthetized with 3% isoflurane and positioned securely on an animal cradle using a stereotaxic head holder. Anesthesia was maintained at a level of 1% to 1.5% isoflurane throughout the entire procedure. Vital signs, including respiration and temperature, were continuously monitored. MRI scans were performed using a 9.4T small animal MRI system (United Imaging uMR 9.4T) with an 86 mm diameter transmit volume coil and a 3‐channel receive array coil. Diffusion imaging data were collected using a gradient‐echo diffusion tensor imaging (DTI) sequence. The imaging parameters were set as follows: TR/TE = 6000/36 ms, 32 diffusion directions, field of view (FOV) = 18 × 16 mm, acquisition matrix = 88 × 80, 80 slices with a slice thickness of 0.2 mm, 8 averages, and a b‐value of 1500 s/mm^2^. DTI analysis was conducted using DSI Studio software (http://dsi‐studio.labsolver.org/). Regions of interest (ROIs) were manually delineated in a blinded fashion, covering the external capsule (EC) and striatum (STR) in both the ipsilateral and contralateral hemispheres to assess fractional anisotropy (FA) and radial diffusivity (RD). FA, MD, AD and RD maps were generated using the DSI Studio software.
An 11.7 T MRI system (Bruker BioSpec, Karlsruhe, Germany) at the Institute of Science and Technology and Brain‐inspired Intelligence in Shanghai, China, was used to scan the mouse brains on days 1, 3, and 7 after ICH. During MRI acquisition, mice were anesthetized with 3% isoflurane in air. All scans were performed using a susceptibility‐weighted imaging (SWI) sequence (repetition time/echo time = 250/5 ms; slice thickness = 0.5 mm). The field of view was 20 × 20 mm with a matrix size of 256 × 256, resulting in 25 coronal slices of 0.5 mm thickness, covering the region from the frontal pole to the brainstem. SWI lesion volumes were subsequently quantified using ImageJ software (version 4.0.1, University of Pennsylvania, USA; www.itksnap.org). Hematoma lesions were delineated along the margins of the hypointense regions, and the total lesion volume was calculated by summing the hypointense areas across all slices.
Seahorse Extracellular Flux Analysis
6.15
Mitochondrial respiration and lactate secretion in BV2 cells were evaluated by measuring the oxygen consumption rate (OCR) and the glycolytic proton efflux rate (glycoPER) using the Seahorse XFe96 extracellular flux analyzer with oxygen control (Seahorse Bioscience). XFe96 microplates (Agilent) were pre‐coated with 22 µg/mL Cell‐Tak (Corning), and BV2 cells (1–2×10^5^ cells per well) were seeded in Seahorse XF RPMI medium (Agilent) containing 2 mM L‐glutamine (Gibco), 1 mM sodium pyruvate (Sigma), and 10 mM D‐glucose (Sigma). The cells were incubated for 1 h in a CO_2_‐free incubator at 37°C before performing glycolytic and mitochondrial stress tests according to the manufacturer's protocol.
For glycolysis analysis, the basal extracellular acidification rate (ECAR) was first measured, followed by the addition of 0.5 µM rotenone (AdipoGen) and 0.5 µM antimycin A (Sigma) to inhibit mitochondrial complexes I and III. At the end of the measurements, 50 mM 2‐deoxyglucose (2‐DG, Sigma) was added to completely block glycolysis.
To assess mitochondrial respiration, basal oxygen consumption was recorded, followed by the addition of 2 µM oligomycin (Cayman Chemicals) to inhibit ATP synthase, 1.5 µM FCCP (Cayman Chemical) as an uncoupling agent, and a combination of 0.5 µM rotenone (AdipoGen) and 0.5 µM antimycin A (Sigma). Basal OCR was determined by subtracting the OCR after rotenone and antimycin A treatment from the OCR before oligomycin addition. Maximal OCR was calculated by subtracting the OCR following rotenone and antimycin A treatment from the OCR measured after FCCP addition.
Metabolomics Workflow Summary
6.16
- Metabolite Extraction: 50 mg of BV2 cells were extracted with 400 µL methanol (4:1, v/v) containing 0.02 mg/mL L‐2‐chlorophenylalanine as the internal standard. The samples were homogenized using a tissue crusher (Wonbio‐96c) at 50 Hz for 6 min and sonicated at 40 kHz for 30 min at 5°C. After protein precipitation at ‐20°C for 30 min, samples were centrifuged (13 000 g, 4°C, 15 min). Supernatants were collected for LC‐MS/MS analysis.
- Quality Control (QC): A pooled QC sample was prepared by mixing equal volumes of all samples. QC samples were analyzed similarly to ensure analytical stability and consistency.
- UHPLC‐MS/MS Analysis: LC‐MS analysis was performed using a Thermo UHPLC‐Q Exactive system. Chromatographic separation was achieved on an HSS T3 column with a gradient mobile phase. MS data were acquired in positive and negative ion modes using electrospray ionization (ESI) with data‐dependent acquisition (DDA) over a mass range of 70‐1050 m/z.
- Data Preprocessing and Annotation: Raw LC‐MS data were processed using Progenesis QI software. Internal standards and false positives were removed, and metabolites were identified using HMDB, Metlin, and Majorbio databases. Data normalization was conducted to account for sample preparation variability, and variables with RSD > 30% were excluded. The final data matrix was log10‐transformed for analysis.
- Differential Metabolite Analysis: PCA and OPLS‐DA were performed using the R package ropls for data variance analysis. Differential metabolites were identified based on VIP > 1 and p < 0.05. Metabolites were mapped to biochemical pathways using KEGG, and enrichment analysis was conducted using Fisher's exact test (scipy.stats).
Parallel Reaction Monitoring (PRM)
6.17
- Protein Extraction: Cell samples were ground into powder using liquid nitrogen, followed by the addition of lysis buffer (8 M urea, 1% Triton X‐100, 10 mM DTT, 1% protease inhibitor). The mixture was sonicated three times on ice, and debris was removed by centrifugation (20 000 g, 4°C, 10 min). Proteins were precipitated using 20% TCA at ‐20°C for 2 h, followed by centrifugation (12 000 g, 4°C, 10 min). The pellet was washed with cold acetone, redissolved in 8 M urea, and protein concentration was measured using a BCA kit.
- Trypsin Digestion: The protein solution was reduced with 5 mM DTT, alkylated with 11 mM iodoacetamide, and diluted to urea < 2 M. Trypsin was added at a 1:50 ratio for overnight digestion, followed by a second digestion at a 1:100 ratio for 4 h.
- LC‐MS/MS Analysis: Tryptic peptides were loaded onto a reversed‐phase column and separated using an EASY‐nLC 1000 UPLC system. The gradient ranged from 6% to 80% solvent B over 60 min at 700nL/min. Peptides were analyzed on a Q ExactiveTM Plus mass spectrometer with a scan range of m/z 350‐1000. Data acquisition used a data‐independent mode alternating between full MS and 20 MS/MS scans.
- Data Search: MS/MS data were analyzed using MaxQuant (v1.6.15.0). Trypsin/P was specified, allowing up to 2 missed cleavages. Fixed modification: carbamidomethyl on Cys; variable modifications: acetylation (N‐term), oxidation (Met), phosphorylation (Ser, Thr, Tyr). FDR was set to <1%.
- Data Analysis: Processed with Skyline (v3.6). Peptide settings included Trypsin [KR/P], up to 2 missed cleavages, peptide length 8‐25, and maximum 3 variable modifications. Transition settings: precursor charges 2, 3; ion charges 1, 2; ion types b, y, p; ion match tolerance 0.02 Da.
PET‐CT Scans
6.18
All PET scans were conducted using the PET/CT small animal scanner (Shanghai, China), part of the Small Animal Imaging Resource Program at the Huashan Hospital PET Center. Mice were fasted overnight, anesthetized with 2% isoflurane gas, and maintained under 1% isoflurane throughout the procedure. An intravenous catheter was inserted into the tail vein for the injection of 200–250 µCi of 18F‐FDG. Scanning started immediately after the tracer injection, with dynamic acquisition lasting 60 min. Mice were placed side‐by‐side, two per scan. PET images were analyzed using a pre‐established data analysis package for mouse brain PET studies, which utilizes a template of standard volumes of interest (VOIs). VOI regions in both the left and right hemispheres include the striatum, hippocampus, thalamus, cortex, as well as the entire cerebellum and brainstem. These template VOIs were spatially aligned to the mouse's upright CT (reoriented to have a perpendicular mid‐plane) using CT‐to‐standard CT spatial normalization parameters (i.e., skull‐to‐skull). Normalization parameters were derived using the Statistical Parametric Mapping (SPM, London) software and its normalization module. PET frames were then spatially aligned to the upright CT according to PET‐to‐CT coregistration parameters obtained through an SPM5 coregistration module that utilizes mutual‐information theory. The SUV ratio (SUVR) for each region and time point was calculated with the cerebellum as the reference region. The mean SUVR from 40 to 60 min was used as the main outcome variable. To visualize the changes in FDG‐PET uptake, averaged PET images were normalized to cerebellum values for individual mice.
EdU Cell Proliferation Assay
6.19
Cell proliferation was assessed using the EdU cell proliferation assay according to the manufacturer's protocol. Approximately 1 × 10^5^ cells were plated in 12‐well plates and cultured for 24 h before the assay. Each well received 500 µL of 10 µM EdU reagent (Beyotime, C0071S), which was incubated for 2 h to label the cells. After washing the cells three times with PBS, they were fixed in 4% paraformaldehyde for 15 min, then permeabilized with Enhanced Immunostaining Permeabilization Solution (Beyotime, P0097) for 15 min. Finally, the cells were incubated with the click‐reaction reagent for 30 min at room temperature in the dark.
Autophagy Staining Assay With MDC
6.20
Add 1 mL of MDC staining solution (Beyotime, C3018S) for every 500 000 cells, gently resuspend and disperse the cells, and incubate at 37°C in the dark for 30 min. The incubation time can be adjusted within the 30 min range based on the actual staining effect. After incubation, centrifuge at 300 g for 5 min at room temperature and remove the MDC staining solution. Wash the cells three times with Assay Buffer. After removing the Assay Buffer, the cells are ready for flow cytometry analysis.
Fluo‐4 Calcium Assay
6.21
For 5 × 10^5^ suspended cells, centrifuge at 300 g for 5 min at room temperature, discard the supernatant, wash once with PBS, and add 1 mL of Fluo‐4 staining solution (Beyotime, S1061S). Resuspend the cells into a single‐cell suspension and incubate at 37°C in the dark for 30 min. After incubation, wash the cells three times with PBS. Finally, perform flow cytometry analysis immediately.
L‐Lactate Assay With WST‐8
6.22
Centrifuge 1 × 10^6^ cells at 300 g for 5 min to collect the cells in a centrifuge tube. Discard the supernatant and remove any residual liquid. Add 100 µL of BeyoLysis Buffer A for Metabolic Assay (Beyotime, S0208S) in the appropriate proportion, gently pipette to mix, and incubate on ice for 10 min to fully lyse the cells. Centrifuge at 12 000 g for 5 min at 4°C, and collect the supernatant. Transfer 50 µL of the supernatant into the sample wells of a 96‐well plate, with control wells containing only BeyoLysis Buffer A for Metabolic Assay or Lactate Assay Buffer as blank controls. Add 50 µL of WST‐8 chromogenic working solution to each well and mix thoroughly. Incubate at 37°C in the dark for 30 min. Finally, measure the absorbance at 450 nm using a microplate reader. The L‐lactate concentration is calculated using the following formula:
A is the L‐lactate concentration (mM) determined from the standard curve, and n is the total dilution factor of the sample.
Free Fatty Acid Assay
6.23
Centrifuge 1 × 10^6^ cells at 300 g for 5 min to collect the cells into a centrifuge tube, discard the supernatant, and remove any residual liquid. Add 150 µL of BeyoLysis Buffer A for Metabolic Assay (Beyotime, S0215S) per 1 million cells, gently pipette to mix. Centrifuge at 12 000 g for 5 min at 4°C, and collect the supernatant for subsequent analysis.
Palmitic Acid Standard Curve Setup
6.24
Take 20 µL of the 1 mM palmitic acid standard solution and add 180 µL of assay buffer and lysis buffer, mix well to prepare a 100 µM palmitic acid standard solution. Then, take 0, 5, 10, 20, 30, 40, 50 µL of the 100 µM palmitic acid standard solution and add it to the standard wells of a 96‐well plate. Fill up to 50 µL in each well with BeyoLysis Buffer A for Metabolic Assay and assay buffer. Add 2 µL of enzyme mixture B to each well, mix, and incubate at 37°C in the dark for 30 min. Add 50 µL of free fatty acid detection working solution to each well, mix, and incubate at 37°C in the dark for 30 min. Finally, measure the absorbance at A570 using a microplate reader to determine the free fatty acid concentration. The calculation formula is as follows: C (µM) = A × n.
A is the free fatty acid concentration (µM) determined from the standard curve, and n is the total dilution factor of the sample.
Glucose Assay With O‐toluidine
6.25
For 5 × 10^5^ cells, add 150 µL of lysis buffer (Beyotime, glucose detection, S0201S). Use a pipette to pipette several times to ensure complete contact between the lysis buffer and the cells, and to fully lyse the cells. Afterward, centrifuge at 12 000 g for 5 min and collect the supernatant as the sample for measurement. Transfer 5 µL of the standard or sample into a PCR tube, add 185 µL of Glucose Assay Reagent to make the final volume 190 µL. Vortex to mix, then centrifuge at 5000 g for a few seconds to settle the liquid at the bottom of the tube. Heat the tube at 95°C for 8 min in a PCR machine, then cool to 4°C. After cooling to 4°C, transfer 180 µL of the liquid from each tube into a clean 96‐well plate. Measure the absorbance at 630 nm.
Glucose Uptake Fluorescence Assay With 2‐NBDG
6.26
Centrifuge 5 × 10^5^ cells at 600 g for 5 min at room temperature and discard the supernatant. Resuspend the cells in 1 mL of 2‐NBDG glucose uptake working solution (Beyotime, S0561S). Incubate the cells at 37°C in a cell culture incubator for 60 min. After incubation, centrifuge at 600 g for 5 min at 4°C to pellet the cells, and discard the supernatant. Wash the cells three times with PBS, and then analyze them using flow cytometry.
A‐CoA (Acetyl Coenzyme A) ELISA
6.27
The following procedure follows the instructions provided in the Acetyl‐CoA (A‐CoA) ELISA Kit manual (Elabscience, E‐EL‐0125). Resuspend 1 × 10^6^ cells in 200 µL of PBS and sonicate to lyse the cells. Centrifuge the extract at 1500 g for 10 min at 4°C, then collect the supernatant for detection. Set up standard wells, blank wells, and sample wells. Add 100 µL of serially diluted standard to the standard wells, 100 µL of standard & sample dilution buffer to the blank wells, and 100 µL of the sample to the remaining wells. Cover the microplate and incubate at 37°C for 90 min.
After incubation, discard the liquid in the wells without washing. Add 100 µL of biotinylated antibody working solution to each well, cover the plate, and incubate at 37°C for 1 h. Discard the liquid in the wells and dry them on clean absorbent paper. Add 350 µL of wash solution to each well, soak for 1 min, and then remove or shake off the liquid from the wells, drying them. Repeat the washing step 3 times. Then, add 100 µL of HRP conjugate working solution to each well, cover the plate, and incubate at 37°C for 30 min.
Discard the liquid in the wells and wash the plate 5 times. Add 90 µL of substrate solution (TMB) to each well, cover the plate, and incubate at 37°C in the dark for about 15 min. Add 50 µL of stop solution to each well to stop the reaction. Immediately measure the absorbance (OD value) of each well at 450 nm using a microplate reader.
Mitochondrial Superoxide Assay With MitoSO Red
6.28
The following procedure follows the instructions provided in the Mitochondrial Superoxide Assay Kit with MitoSO Red (Beyotime, S0061S). Centrifuge 5 × 10^5^ cells at 600 g for 5 min at room temperature and discard the supernatant. Add an appropriate volume of MitoSO Red staining working solution to resuspend the cells. Incubate at 37°C in a cell culture incubator for 30 min, then wash the cells three times with PBS and analyze using flow cytometry.
Intracerebral Injection of AAV11 Vector Expressing HK2
6.29
The viral vectors were constructed by Braincase Biotechnology Co., Ltd. (Shenzhen, China). In this study, we stereotaxically injected either a control virus (rAAV‐SFFV‐DIO‐EGFP‐WPRE‐4×miR9T, AAV‐CTL) or an HK2‐overexpressing virus (rAAV‐SFFV‐DIO‐mHk2‐3×FLAG‐WPRE‐4×miR9T, AAV‐HK2‐OE), both at a titer of ≥2.0×10^12^ vector genomes per milliliter. Viruses (400nl per site) were delivered into two coordinates relative to bregma: anterior–posterior (AP) = 0.14 mm, medial–lateral (ML) = 2.0 mm, dorsal–ventral (DV) = 2.5 mm; and AP = 0.14 mm, ML = 2.0 mm, DV = 3.5 mm. Following viral injection, mice were allowed to recover for one week. Subsequently, tamoxifen was administered intraperitoneally to induce Cre recombinase expression. Three weeks after viral infection, microglia exhibited HK2 overexpression, and intracerebral hemorrhage was induced in the mice.
Statistical Analysis
6.30
The results are shown as mean ± standard deviation (SD). To determine the statistical significance of the differences between the experimental groups two‐tailed unpaired or paired Student's t tests, Two‐way ANOVA and One‐way ANOVA tests were performed using the Prism 9 software (GraphPad). Sample sizes were based on experience and experimental complexity, but no methods were used to determine normal distribution of the samples. Differences reached significance with p values < 0.05 (noted in figures as ^^), p < 0.01 (^^) and p < 0.001 (^^). The figure legends contain the number of independent experiments or mice per group that were used in the respective experiments.
Author Contributions
Yuxiang Gu, Heng Yang, Wei Ni, Zhiwen Jiang conceived the project and designed experiments; Zhiwen Jiang, Xinjie Gao, Zengyu Zhang, Ruiyuan Weng, Yuchao Fei carried out experiments; Zhiwen Jiang, Xinjie Gao, Chao Gao and Jiabin Su analysed the data; Zhiwen Jiang and Xinjie Gao wrote the paper; Zhiwen Jiang Heng Yang and Xinjie Gao reviewed and edited the paper; Yuxiang Gu, Heng Yang, Wei Ni, Chao Gao and Xinjie Gao provided funding acquisition.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Supporting File 1: advs72378‐sup‐0001‐SuppMat.docx.
Supporting File 2: advs72378‐sup‐0002‐Data.xlsx.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1R. V. Krishnamurthi , A. E. Moran , M. H. Forouzanfar , et al., “The Global Burden of Hemorrhagic Stroke: A Summary of Findings from the GBD 2010 Study,” Global Heart 9 (2014): 101, 10.1016/j.gheart.2014.01.003.25432119 · doi ↗ · pubmed ↗
- 2L. Puy , A. R. Parry‐Jones , E. C. Sandset , D. Dowlatshahi , W. Ziai , and C. Cordonnier , “Intracerebral Haemorrhage,” Nature Reviews Disease Primers 9 (2023): 14, 10.1038/s 41572-023-00424-7.36928219 · doi ↗ · pubmed ↗
- 3Z. Zhang , Z. Zhang , H. Lu , Q. Yang , H. Wu , and J. Wang , “Microglial Polarization and Inflammatory Mediators after Intracerebral Hemorrhage,” Molecular Neurobiology 54 (2017): 1874–1886, 10.1007/s 12035-016-9785-6.26894396 PMC 4991954 · doi ↗ · pubmed ↗
- 4W. Ni , S. Mao , G. Xi , R. F. Keep , and Y. Hua , “Role of Erythrocyte CD 47 in Intracerebral Hematoma Clearance,” Stroke; A Journal of Cerebral Circulation 47 (2016): 505–511, 10.1161/STROKEAHA.115.010920.PMC 472965126732568 · doi ↗ · pubmed ↗
- 5X. Gao , H. Yang , W. Xiao , et al., “Modified Exosomal SIRPα Variants Alleviate White Matter Injury after Intracerebral Hemorrhage via Microglia/Macrophages,” Biomaterials Research 26 (2022): 67, 10.1186/s 40824-022-00311-4.36435797 PMC 9701394 · doi ↗ · pubmed ↗
- 6J. Magid‐Bernstein , R. Girard , S. Polster , et al., “Cerebral Hemorrhage: Pathophysiology, Treatment, and Future Directions,” Circulation Research 130 (2022): 1204–1229, 10.1161/CIRCRESAHA.121.319949.35420918 PMC 10032582 · doi ↗ · pubmed ↗
- 7J. Xu , Z. Chen , F. Yu , et al., “IL‐4/STAT 6 Signaling Facilitates Innate Hematoma Resolution and Neurological Recovery after Hemorrhagic Stroke in Mice,” Proceedings of the National Academy of Sciences 117 (2020): 32679–32690, 10.1073/pnas.2018497117.PMC 776877133293423 · doi ↗ · pubmed ↗
- 8X. Y. Sun , T. Zheng , X. Yang , et al., “HDAC 2 hyperexpression Alters Hippocampal Neuronal Transcription and Microglial Activity in Neuroinflammation‐induced Cognitive Dysfunction,” Journal of Neuroinflammation 16 (2019): 249, 10.1186/s 12974-019-1640-z.31796106 PMC 6889553 · doi ↗ · pubmed ↗