1. Introduction
Neonatal hypoxic-ischemic encephalopathy (HIE) denotes hypoxic-ischemic brain damage (HIBD) resulting from perinatal asphyxia. HIE is one of the most harmful and common diseases occurring in the neonatal period and often causes neonatal death and neurological developmental disorders. Currently, there are no fundamental prevention and treatment measures for HIBD [
1]. Different organs exhibit varying susceptibility to hypoxia, with brain cells being the most sensitive [
2]. The pathogenesis of HIBD is complex and diverse, including energy metabolism disorders, inflammatory responses, oxidative stress, mitochondrial dysfunction, and apoptosis [
3]. The development of drugs to treat HIBD has been slow. Urgent efforts are required to elucidate the mechanisms underlying HIE development, identify disease targets, develop potent targeted therapies, and understand the pharmacological mechanisms of these treatments.
Some traditional Chinese medicine injections, such as compound Danshen injection (also known as Xiangdan injection), have shown positive results in the adjuvant treatment of HIE in some hospitals [
4], effectively improving treatment efficacy with a low side effect incidence [
5]. Salvianolic acid C (SAC) is mainly derived from
Salvia miltiorrhiza, which has antioxidant, anti-inflammatory, and antiapoptotic activities [
6]. SAC, one of the active monomers of the compound Danshen injection [
9], can regulate inflammatory cytokine levels in stroke and reduce microglial polarization [
7,
8]. Although evidence suggests that SAC plays a neuroprotective role, its activity in the treatment of HIE has not been studied, and the specific mechanisms underlying its effects remain unclear.
Most existing studies on human brain diseases have focused on animal experiments [
10], with limited resulting progress in HIE. Human brain organoids represent a superior model for these studies [
11]. The culture of brain organoids begins with pluripotent stem cells (PSCs), which are differentiated into many different cell types and mature structures through the addition of various cytokines [
12]. Importantly, these organoids exhibit cortical structures close to those of the human brain [
13], such as the subventricular zone (SVZ) characterized by the marker T-box brain protein 2 (TBR2), which is also the region most sensitive to hypoxia [
14]. The use of three-dimensional (3D) brain organoids provides a bridge for studying the environmental and genetic factors involved in brain injury during neonatal development in clinical medicine [
15].
Single-nucleus RNA sequencing (snRNA-seq) enables the characterization of cellular heterogeneity in the brain by delineating the transcriptomic landscape and dynamic properties of various cell types at single-cell resolution [
16]. This approach enables enhanced comprehension of natural product mechanisms, detailing normal, pathological, and post intervention cellular diversity and molecular changes, offering new insights into disease, drug targets, and signaling pathways [
17].
Therefore, we comprehensively evaluated the pharmacodynamic effects of SAC on HIE in rats and human brain organoids. To explore the dynamic mechanisms of HIBD and SAC therapy, we collected rat brain samples at different time points after hypoxia-ischemia (HI) for snRNA-seq. Our findings provide theoretical insights into the complexity of HIBD mechanisms and the potential clinical application of SAC in HIE treatment.
2. Methods
2.1. Animal experiments
Sprague-Dawley (SD) rat litters were acquired from Shanghai SLAC Laboratory Animal Co., Ltd., China. The rats were maintained in a controlled environment with a temperature range of 24-26 °C and humidity levels suitable for their well-being. The rats were maintained on a 12 h light/dark cycle and were given unrestricted access to fresh water and a standard diet. A modified Rice-Fannuzzi model was used [
18], which is the most frequently employed HIBD model. Seven-day-old suckling rats were anesthetized and maintained with isoflurane. After ligation of the right common carotid artery, the suckling rats were returned to their mothers for 1.5-2 h and then placed in a 37 °C airtight hypoxic incubator (8% oxygen and 92% nitrogen) for 2 h. The sham group underwent neck skin incision without ligation or hypoxia.
The rats were then administered daily treatments as follows. The sham and model group rats were intraperitoneally injected with the same concentration of normal saline (0.01 mL∙g−1). The low-dose SAC group was administered an intraperitoneal injection of SAC 10 mg∙kg−1 daily, and the high-dose SAC group was administered an intraperitoneal injection of 20 mg∙kg−1 SAC daily. SAC (purity ≥ 98.0%, batch number: DST240125-010) was acquired from Desite Biological Technology Co., Ltd., China, and dissolved in saline solution. The drug was administered to the rats immediately after hypoxia induction, followed by a 24 h interval between subsequent doses. Finally, the rats were euthanized.
All animal experiments were approved by the Animal Care and Use Committee of the Zhejiang University School of Medicine, China (approval number: ZJU20240849).
2.2. Culture of human PSC (hPSC) and brain organoids
H9 human embryonic stem cells (hESCs; WA09) were purchased from the WiCell Institute, USA. hiPSC-B1 cells (CTCC-001-0650) were purchased from Zhejiang Meisen Cell Technology Co., Ltd., China, and hiPSC-DYR0100 cells were obtained from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences, China. hPSCs were maintained on Matrigel (354277; Corning, USA)-coated cell culture plates using mTeSR1 medium (85850; STEMCELL Technologies, Canada). Undifferentiated hPSC aggregates were passaged using Gentle Cell Dissociation Reagent (100-0485; STEMCELL Technologies), and the aggregates were scraped with a cell scraper. All cell lines underwent mycoplasma testing every three weeks, and no contamination was observed.
Human brain organoids were generated from H9 hESCs, hiPSC-B1 cells, and hiPSC-DYR0100 cells using the STEMdiffTM Brain Organoid Kit (08570; STEMCELL Technologies), per the manufacturer’s instructions. Embryoid bodies (EBs) were formed from hPSCs. On day 5 of culture, the EBs were transferred to the induction medium. On day 7 of culture, the EBs were embedded in Matrigel to facilitate neuroepithelial outgrowth. After day 10 of culture, the brain organoids were maintained for a prolonged duration using the STEMdiffTM Brain Organoid Maturation Kit (08571; STEMCELL Technologies).
2.3. Human brain organoid hypoxia and drug exposure treatment
On day 50 of in vitro differentiation, human-derived brain organoids from three hPSCs were transferred to an oxygen-controlled CO2 incubator (95% N2, 5% CO2, ≤ 1% O2, 37 °C; Thermo Fisher Scientific, USA). The medium was previously equilibrated at approximately 1% O2, 5% CO2, and 37 °C for 12 h. After 48 h, these brain organoids were transferred to an incubator containing 21% O2 and 5% CO2 for 24 h of reoxygenation with SAC exposure.
2.4. Cell culture
Rat brain astrocytes (CTX-TNA2), human oligodendrocyte cell lines (MO3.13), and mouse microglial cells (BV2) were obtained from Cellverse Co., Ltd., China, Shanghai Jinyuan Biotechnology Co., Ltd., China, and the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences, respectively. Cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin mixture (P/S; Gibco). The cells were incubated at 37 °C in a 5% CO2 atmosphere and subcultured upon reaching 80% confluency.
For CTX-TNA2 cell transfection, small interfering RNA (siRNA) was transfected using Lipofectamine 2000 transfection reagent (11668019; Invitrogen, USA). The small interfering RNA targeting signal transducer and activator of transcription 3 (siStat3) and negative control (NC) sequences are listed in Table S1 in Appendix A.
BV2 cells were treated with 10 ng∙mL−1 of interleukin (IL)-4 (HY-P70644; MedChemExpress, USA) and 10 ng∙mL−1 of IL-13 (HY-P70460; MedChemExpress) for 24 h to generate anti-inflammatory microglia (MG).
2.5. Oxygen glucose deprivation/reperfusion (OGD/R)
OGD/R was induced as previously described [
19]. CTX-TNA2, MO3.13, and BV2 cells were rinsed with new medium twice and grown in DMEM without glucose and FBS in an anaerobic chamber (95% N
2 and 5% CO
2; Billups-Rothenberg, USA)
. The cells were then transferred to an aerobic environment (95% air and 5% CO
2) for reoxygenation. CTX-TNA2 and MO3.13 cells were exposed to hypoxia for 6 h, then reoxygenated for 24 h. BV2 cells were exposed to hypoxia for 3 h, then reoxygenated for 6 h. After OGD/R treatment, CTX-TNA2 cells were treated with 50 μmol∙L
-1 SAC, MO3.13 cells were administered 10 or 30 μmol∙L
-1 SAC or 10 or 50 ng∙mL
−1 neuregulin 3 (NRG3) (HY-P75944; MedChemExpress), and BV2 cells were treated with 100 or 200 μmol∙L
-1 SAC.
2.6. 2,3, 5-Triphenyltetrazolium chloride (TTC) staining
Twenty-four hours post HI, brain tissue was collected from neonatal P8 rats, incubated at −20 °C for 30 min, and cut into coronal sections approximately 2 mm in thickness. The sections were stained with 0.25% TTC solution (G3005; Solarbio, China) in the dark for 30 min. The brain slices were then submerged in 4% paraformaldehyde. The volume of cerebral infarction was measured using Image-Pro Plus software.
2.7. Hematoxylin and eosin (H&E) and Nissl staining
To obtain rat tissue samples, the rats were anesthetized with isoflurane, and their hearts were perfused with normal saline, followed by 4% paraformaldehyde. The brain tissues were submerged in 4% paraformaldehyde and encapsulated in paraffin. For brain organoid sample preparation, the organoids were washed with phosphate-buffered saline (PBS), immersed in 4% paraformaldehyde, and stored at 4 °C.
Paraffin-embedded samples were sliced into 4 µm thick sections. The sections underwent deparaffinization, hydration, and staining with H&E (HK2053; Haoke Biotechnology, China) or Nissl solution (G1036; Servicebio, China). The histological staining results were evaluated and recorded using a Jiangfeng white light scanner (KF-PRO-120). The results were analyzed using the ImageJ software.
2.8. Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) staining
In situ detection of apoptosis was performed using the One-Step TUNEL Assay Kit (HKI0011; Haoke Biotechnology) following the manufacturer’s instructions. After reaction with the working solution, the slides were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) before mounting and were sealed with an anti-fluorescence quenching mounting agent (HKI0007-1; Haoke Biotechnology).
2.9. Neurological function scoring experiment
2.9.1. Righting reflex
Each suckling rat was positioned supine on a plate, ensuring its head, neck, and back were in contact with the surface, and the duration required for the rat to transition from the supine posture to an upright stance on all four limbs was documented.
2.9.2. Cliff avoidance
Two-thirds of the body of each rat was placed at the platform’s edge, and the latency for the rat to retreat or move away from the edge was documented.
2.9.3. Negative geotaxis experiment
Each rat was positioned on a rough board inclined at a 45° angle to the horizontal plane, toward the upper extremity of the board. The rat was initially placed with the head facing down, and the latency of each rat to start the ascent was recorded [
20].
2.10. Blood biochemistry and blood routine analysis
Serum biochemical indices in rat serum samples, including alanine aminotransferase (ALT) and aspartate aminotransferase (AST), were analyzed using a Catalyst Dx Biochemical Analyzer (Idexx Laboratories, USA). Routine whole blood indices, including the number or proportion of white blood cells, red blood cells, hemoglobin, lymphocytes, monocytes, and neutrophils, were measured using a ProCyte Dx automatic blood cell analyzer (Idexx Laboratories).
2.11. Immunofluorescence staining and immunohistochemistry
Rat brain tissue specimens were sliced into 4 µm paraffin slices as described above. The brain organoid cultures were fixed in 4% paraformaldehyde overnight and subsequently transferred to 30% sucrose for dehydration. The organoids were then immersed in a gelatin solution and snap-frozen. Subsequently, 16 µm thick slices were obtained. These sections were then rewarmed, dried, fixed, and washed. Ethylenediaminetetraacetic acid (EDTA) antigen repair buffer was used for antigen repair in the microwave, and hydrogen peroxide was added to inhibit endogenous peroxidase activity. Next, 3% bovine serum albumin (BSA) was added to block serum. Primary antibodies were added, and the samples were incubated at 4 °C. The following day, the appropriate secondary antibody was added, and the samples were incubated in the dark. The antibodies are listed in Table S2 in Appendix A.
For immunofluorescence staining, nuclei were counterstained with DAPI after treatment with signal amplification reagents. Finally, sections were washed and blocked with an anti-fluorescence quenching agent (HKI0007-1; Haoke Biotechnology). For immunohistochemistry, sections were washed, and 3-3′ diaminobenzidine (DAB) color solution (PR30010; Proteintech, China) was added dropwise until a brown-yellow color was observed. The sections were then washed with pure water, counterstained with hematoxylin, differentiated, stained blue, dehydrated, and treated with xylene. Finally, the sections were sealed with neutral gum. After staining, images were captured and analyzed using a white light scanner (KF-PRO-120).
2.12. Luxol fast blue (LFB) staining
The rat brain slices were gradually deparaffinized in xylene, rehydrated in a graded series of ethanol, and washed in distilled water. The sections were stained overnight at 60 °C with 0.1% LFB (HK1050; Haoke Biotechnology), washed with tap water, and differentiated by alternating between 70% ethanol and 0.05% lithium carbonate under microscopic guidance until optimal gray matter fading was achieved. Subsequently, sections were counterstained with 0.1% eosin, rinsed, air-dried, cleared in xylene, and mounted with neutral balsam.
2.13. Micro-electrode array (MEA) recording
As previously outlined [
21,
22], brain organoids were seeded in six-well MEA plates (Axion Biosystems, USA). Recordings were conducted with the Maestro MEA system and AxIS software (Axion Biosystems). The MEA plates were allowed to rest in the Maestro setup for 3 min, after which the data were recorded for 5 min. The neural metric tool AxIS was used to identify spikes, and electrodes were considered “active electrodes” if they recorded at least five spikes per min. An interspike interval (ISI) threshold of ≤ 100 ms was used to identify bursts in data from each electrode, necessitating ≥ 5 spikes. At the same ISI, a network burst required at least ten spikes. The synchronization index utilized a cross-correlogram window of 20 ms.
2.14. snRNA-seq
2.14.1. Single-nucleus processing of the rat brain
The cerebral ischemic hemispheres of rats in the sham, model, and SAC groups (
n = 4 rats per group) were collected at 24 h and 7 d after HI. The tissue was added to the lysate and PBS mixture and sheared into small pieces. Tissue from four rats per group was pooled to generate a single nuclear suspension per group. To minimize RNA degradation during nuclei isolation, the lysis buffer was precooled and supplemented with RNase inhibitors. Gentle homogenization was performed, avoiding damaging the nuclei, and the lysis time was carefully optimized. The samples were then prelysed in precooled lysate on ice. The reaction mixture was diluted with precooled 4% BSA, and the reaction was terminated. After washing and centrifugation, the cells were resuspended in the lysate containing 4% BSA. After each washing step, the morphology of the nuclei was examined under a microscope to ensure integrity. Viability was assessed using Trypan Blue staining to determine the proportion of viable nuclei. A Miltenyi Biotec kit SOP (130-109-398; Miltenyi, Germany) was used to remove debris. Nuclei were resuspended in 1× PBS, 2% BSA, and 0.1% RNase inhibitor. After filtering through a 20 µm cell sieve, the nuclei were counted, and the concentration was modified to 700-1200 nuclei∙µL
−1 [
23].
2.14.2. Single-cell library preparation and quality control
Per the 10× Genomics Chromium Single-Cell 3 Kit instructions, the single-nuclei suspension was loaded, and 8000 single cells were captured. Complementary DNA (cDNA) library preparation was subsequently performed. Libraries were sequenced using the Illumina NovaSeq6000 sequencing system (paired-end multiplexing run, 150 bp; USA), with at least 20 000 read depths per cell.
For quality control, sequencing data were demultiplexed and converted to the FASTQ format using Illumina bcl2fastq software (version 5.01). Raw gene expression matrices were produced with the Cell Ranger (version 6.0.1) pipeline aligned with the rat reference genome and subsequently analyzed with the Seurat package (version 4.4.0) in R software (version 4.1.3). The quality control criteria included the exclusion of low-quality cells with fewer than 500 identified genes and cells with mitochondrial encoded transcripts over 25%. Doublet Finder (version 2.0.4) was employed to exclude multicellular elements from the dataset.
We computed the gene expression values for visualization employing the log-normalization technique via Seurat software. The cells were further clustered according to the appropriate resolution. Uniform manifold approximation and projection (UMAP) was used for two-dimensional (2D) visualization.
2.14.3. Bioinformatics analysis
Seurat was employed to identify differentially expressed genes (DEGs). The criteria for DEGs were as follows: min.pct = 0.25,
P_val < 0.05, avg_log
2FC ≥ 0.25, and min.diff.pct = 0.1. ClusterProfiler and Metascape were used for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses to determine the key biological processes and signaling pathways, and significantly enriched terms were selected for visualization (
P < 0.05) [
24].
Monocle2 was employed to delineate cellular developmental trajectories, and alterations in cellular states were assessed utilizing the GeneTest function. The plot_cell_trajectory function was subsequently employed to sort the cells in pseudo-time.
CellChat was employed to examine the expression of ligand-receptor (LR) pairs between cells, reveal the interaction network between cells, and compare the number and strength of interactions between different cell subsets.
SCENIC was combined with snRNA-seq data to construct a gene regulatory network, identify transcription factors (TFs) and target gene co-expression modules, infer direct target genes, and calculate regulon activity and cell type specificity scores.
Partial bioinformatics analysis findings were derived utilizing the OmicStudio tools available.
2.15. Serum inflammatory factor chip detection
Serum inflammatory cytokines in rats were measured utilizing Luminex liquid suspension chips (Wayen Biotechnologies, China). The measurements were conducted with a Bio-Plex Pro Rat Cytokine 23-plex Kit (12005641; Bio-Rad, USA) per the manufacturer’s instructions. Briefly, the chip was incubated with the sample for 1 h, after which the appropriate antibody was added, followed by streptavidin-R-phycoerythrin. Detection was performed using a Bio-Plex array reader (Bio-Rad).
2.16. Real-time quantitative reverse transcription polymerase chain reaction PCR (RT-qPCR)
Total RNA was extracted utilizing TRIZOL (R0016; Beyotime, China). Reverse transcription was performed with the HiFiScript cDNA Synthesis Kit (CW2569M; Cwbio, China). The SYBR Green PCR Kit (11184ES; Yeasen, China) was applied in a CFX-Touch™ 96 real-time PCR system (Bio-Rad) with the necessary primers (Sangon, China) for assessing gene expression. The specific primer sequences are listed in Table S3 in Appendix A. Relative expression was calculated using the ΔΔCt method and normalized. The experiment was repeated three times.
2.17. Western blotting
Proteins were separated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. The membranes were then blocked with 5% skim milk for 1 h at room temperature and thereafter exposed to the primary antibody overnight at 4 °C. The membranes were then incubated with horseradish peroxidase-labeled secondary antibody (1:5000 in Tris-buffered saline with Tween 20 (TBST) with 5% non-fat milk) for 1 h at room temperature. Subsequently, the membranes were washed three times in TBST buffer, incubated with enhanced chemiluminescence reagent, and exposed to detect the protein bands [
25]. The antibodies used are listed in Table S2.
2.18. Lactate dehydrogenase (LDH) release assay
Cytotoxicity was assessed utilizing an LDH Kit (c0016; Beyotime). The supernatants of the brain organoid cultures were collected from each well and incubated with the detection working solution. Absorbance was measured with an Infinite M1000 Pro microplate reader (Tecan, Switzerland).
2.19. Enzyme-linked immunosorbent assay (ELISA)
IL-6 and IL-1β levels were quantified in CTX-TNA2 cells supernatants (IL-6) or lysates (IL-1β) using ELISA kits (KE20024, KE20005; Proteintech) in accordance with the manufacturer’s guidelines. Samples and standards were incubated on pre-coated plates containing capture antibodies. IL-6 was detected using horseradish peroxidase (HRP)-conjugated antibodies, whereas IL-1β was detected using biotinylated antibodies followed by streptavidin-HRP. 3,3′,5,5′-Tetramethylbenzidine (TMB) substrate was introduced, and absorbance was recorded at 450 nm (reference 630 nm). Cytokine concentrations were calculated using a four-parameter logistic curve fit.
NRG3 levels in BV2 cell supernatants were determined using an ELISA kit (SYP-M2228-T1; Ubingbio Technology, China) following the manufacturer’s protocol. Briefly, samples and standards were added to microtiter plates pre-coated with capture antibodies specific for NRG3. After incubation, HRP-conjugated detection antibodies were introduced. Other steps were similar to those described above.
2.20. Nuclear protein extraction
Nuclear proteins were extracted from CTX-TNA2 cells per the kit manufacturer’s directions (EX1470; Solarbio). Briefly, cells were washed with ice-cold PBS and lysed with Reagent A. The lysates were then gently shaken and centrifuged. The pellet was resuspended in Reagent B, supplemented with protease and phosphatase inhibitor cocktails. Subsequently, nuclear extracts were obtained by centrifugation. The supernatant containing nuclear proteins was collected and stored at −80 °C.
2.21. Liquid chromatography-mass spectrometry (LC-MS) analysis
The SAC distribution in the brain of postnatal day 7 (P7) rat pups was evaluated via targeted LC-MS quantitative analysis following intraperitoneal administration. Rat pups were administered SAC at doses of 20, 40, and 80 mg∙kg−1 (with 20 mg∙kg−1 serving as the experimental dose). At 30 min post administration, the rats were deeply anesthetized, and their brain tissues were collected. The tissue samples were stored at −80 °C until analysis. For sample preparation, frozen brain tissue was homogenized in 1 mL precooled 80% methanol, sonicated for 30 min, and centrifuged; the supernatant was collected, and the pellet was re-extracted with an additional 1 mL of 80% methanol. The combined supernatants were vacuum-dried and reconstituted in 200 µL of 50% methanol. Samples were analyzed using an ACQUITY UPLC BEH C18 column (1.7 µm, 2.1 mm × 100 mm; Waters, USA) on the Nexera X2 UHPLC system (Shimadzu, Japan) coupled to a 5500 QTRAP mass spectrometer (AB SCIEX, USA) operating in negative ion mode. The mobile phase consisted of water (A) and acetonitrile (B) at a flow rate of 300 µL∙min−1, with the following gradient: 0-4 min, B from 30% to 90%; 4-6 min, hold at 90% B; 6-7 min, B from 90% to 30%; 7-10 min, hold at 30% B. Mass spectrometric detection was performed in multiple reaction monitoring (MRM) mode, monitoring the quantifier transition m/z 491.2 to 293 and the qualifier transition m/z 491.2 to 311. Qualitative and quantitative analyses of the chromatographic peak areas were performed using Analyst software (version 1.6.3).
2.22. Statistical analyses
All data are expressed as the mean ± standard deviation of the mean (SD). Statistical significance was assessed using one-way analysis of variance (ANOVA) when more than two groups were analyzed. An unpaired Student’s t test was employed to compare two groups. The significance level was established at P < 0.05. Data analysis was conducted utilizing GraphPad Prism 8.0.
3. Results
3.1. SAC alleviates brain injury in HIBD rats
To determine the efficacy of SAC, we employed TTC staining to measure cerebral infarction volume in the sham, model, and SAC (10 and 20 mg∙kg
−1) groups at 24 h after HI. SAC at 10 and 20 mg∙kg
−1 markedly decreased the cerebral infarction volume in a dose-dependent manner (
Figs. 1(a) and
(b)). Consequently, we selected a dosage of 20 mg∙kg
−1 for the subsequent studies. The body weights of suckling rats in the SAC group were considerably greater than those of the rats in the model group (
Fig. 1(c)). Brain morphology 7 d post HI was compared between the groups. Severe edema and liquefaction occurred in the model group, and the ischemic hemisphere was significantly reduced, which was significantly improved after SAC treatment (
Fig. 1(d)). In the HIBD study, the non-ischemic hemisphere of the rat brain was considered to have a normal weight. Therefore, the ratio of right to left hemisphere weight was used to measure brain mass loss caused by HIBD [
26]. SAC treatment significantly reduced brain mass loss in HIBD rats (
Fig. 1(e)). Furthermore, the behavioral tests revealed that SAC alleviated HI-induced neurological damage (
Fig. 1(f)).
Pathological staining indicated that the cerebral ischemic hemisphere in the model group had substantial damage, the cerebral cortex and hippocampal cells were disordered and reduced, and Nissl body arrangement and structure were abnormal (
Figs. 1(g) and
(h)). After SAC treatment, neuronal morphology and quantity improved, and the structural integrity of the brain was reinstated (
Figs. 1(g) and
(h)). SAC alleviated the apoptosis induced by HI (
Figs. 1(i) and
(j)). The infarct core, penumbra, and non-damaged areas could be clearly identified in global staining sections at low magnification (Fig. S1(a) in Appendix A). The infarct core was characterized by severe structural disintegration and dense TUNEL-positive (TUNEL
+) apoptotic cells; the penumbra area showed partial neuronal loss, disordered H&E staining, fragmented Nissl bodies, and scattered TUNEL
+ cells; the non-damaged regions showed intact cellular architecture.
In addition, we evaluated the toxicity of 20 mg∙kg
−1 SAC. Seven days after HI and SAC administration, the heart, liver, spleen, lungs, and kidneys of the SAC group showed no apparent damage (Fig. S1(b) in Appendix A). In comparison to the sham group, there were no significant alterations in ALT and AST levels in the model and SAC groups (
Fig. 1(k)). Moreover, blood routine analysis showed that SAC reduced the HI-induced elevation in the quantity of inflammatory cells, particularly neutrophils (Fig. S1(c) in Appendix A). Therefore, no evident toxicity was observed after 7 d of 20 mg∙kg
−1 SAC administration.
3.2. SAC rescues neuronal apoptosis and neural network integrity in HIE brain organoids
We next assessed the pharmacodynamic effects of SAC therapy on human brain organoids. The brain organoid model was established as shown in
Fig. 2(a). Immunofluorescence staining revealed abundant cell types and structures in these brain organoids (Fig. S2(a) in Appendix A).
The HIE model was then established using human brain organoids, as shown in
Fig. 2(b). To validate the HI injury, we rigorously monitored pathophysiological parameters at multiple levels. Western blotting and immunofluorescence staining revealed the expression of hypoxia-inducible factor-1α (HIF-1α), the gold standard indicator of hypoxia, in brain organoids after hypoxia and reoxygenation (Figs. 2(c)-(f)). Pathological staining showed that both cell morphology and numbers changed significantly after HIE, confirming HI-induced brain organoid damage (Figs. S2(b) and (c) in Appendix A). We further confirmed that HIE enhanced different forms of cell death (Figs. S2(d)-(g) in Appendix A). In addition, by adding antiapoptotic and pro-apoptotic inhibitors, we demonstrated that our HIE brain organoids could be used for drug evaluation (Figs. S2(h)-(j) in Appendix A). These results are consistent with existing reports on hypoxic brain organoid models [
14,
27,
28].
Three hPSC-derived cell lines (H9 hESC, hiPSC-B1, and hiPSC-DYR0100) were used to evaluate the efficacy of SAC treatment in HIE (Fig. S3(a) in Appendix A). The concentrations of 1, 5, and 10 µmol∙L
-1 were designated as low (SAC L), medium (SAC M), and high (SAC H) SAC doses, respectively, in the treatment groups. Before HIE modeling, all brain organoids were randomly allocated into three groups, and statistical results indicated no significant differences in the area of brain organoids among the groups, ensuring the accuracy of the experiment (
Fig. 2(g)). After HIE modeling and SAC administration, an assessment of the brain organoid area and the cell debris in the culture medium revealed that 5 µmol∙L
-1 SAC reduced cell death and damage in HIE brain organoids to the most significant extent (
Fig. 2(g), Fig. S3(b) in Appendix A). Furthermore, 5 µmol∙L
-1 SAC significantly reduced HIE-induced cytotoxicity (
Fig. 2(h)). Therefore, the concentration of 5 µmol∙L
-1 SAC was chosen for the subsequent studies.
We next assessed
TBR2 expression and found that HIE modeling resulted in a significant reduction in
TBR2 messenger RNA (mRNA) levels in the brain organoids, indicating impaired neurogenesis, as TBR2 is a critical TF expressed in intermediate progenitor cells during neurogenesis [
14]. This TF is crucial in the shift from radial glial cells to mature neurons. SAC treatment significantly restored
TBR2 expression levels (
Fig. 2(i)), suggesting a potential neuroprotective effect after HIE injury. Immunofluorescent staining revealed that SAC significantly reduced apoptosis (
Figs. 2(j) and
(k)), which primarily occurred in neurons as identified by previous marker genes (Fig. S2(a)).
In vivo experiments using rat brain tissues validated these findings. Ki67/doublecortin (DCX) co-staining, neuron-specific class III β-tubulin (TUJ-1), and growth-associated protein 43 (GAP43) immunohistochemistry showed that SAC treatment reversed the decrease in positive cells due to HIBD (Figs. S3(c)-(e) in Appendix A), indicating enhanced neurogenesis. Western blotting revealed that SAC significantly reversed the HI-induced upregulation of cleaved caspase-3 and B-cell lymphoma-2 (Bcl-2)-associated X protein (BAX), along with the downregulation of Bcl-2 (Figs. S3(f)-(i) in Appendix A), consistent with the apoptosis findings in human brain organoids.
Subsequently, in human brain organoids, electrophysiological tests revealed that the model group exhibited decreased discharge, active electrodes, and network bursts in comparison with those in the control group, reflecting neuronal weakness and network synchronization impairment. SAC treatment reversed these effects, enhancing neuronal signaling and network integrity (Figs. 2(l)-(n), Figs. S3(j)-(m) in Appendix A).
Collectively, our comprehensive investigations employing molecular, histological, and functional approaches confirmed that the HIE brain organoids model recapitulates key pathophysiological hallmarks of HIE, including hypoxia-induced molecular responses, structural damage, aberrant neurogenesis, and impaired neuronal network activity. Furthermore, these findings provide compelling evidence for the therapeutic potential of SAC in mitigating HIE-induced damage and promoting neuroprotection.
3.3. Experimental process and single-nucleus transcriptome landscape of SAC treatment in HIBD rats
To understand the transcriptome landscape and mechanism of HIBD and SAC treatment, we performed snRNA-seq using 10× genomics on brain tissues from four representative rats in each group (sham, model, and SAC) at 24 h and 7 d after HI. In HIBD, the first 24 h post HI is generally recognized as the acute phase, during which acute neuronal death and inflammation occur [
29]. By day 7, the disease transitions into the subacute phase, with a focus on neural repair and regeneration. At this stage, the rat brain recovery processes begin [
30]. The experimental procedure is illustrated in
Fig. 3(a). After mass filtration, 59 282 cells were detected: the 24 h_sham, 24 h_model, 24 h_SAC, 7 d_sham, 7 d_model, and 7 d_SAC groups yielded 9997, 12 141, 10 546, 9294, 8465, and 8839 cells, respectively (Fig. S4(a) in Appendix A). The data underwent dimensionality reduction at a resolution of 0.8 and were visualized via UMAP (Fig. S4(b) in Appendix A).
To confirm the identity of the 30 clusters, we identified ten cell lines using classical marker genes (
Fig. 3(b)). The proportion of each cell lineage varied widely across the groups (
Fig. 3(c)). The cells included astrocytes (ACs; identified by the presence of aquaporin-4 (
Aqp4), glial fibrillary acidic protein (
Gfap), tenascin C (
Tnc), and solute carrier family 1 member 2 (
Slc1a2)), Cajal-Retzius cells (reelin (
Reln), tumor protein p73 (
Tp73), and neuron-derived neurotrophic factor (
Ndnf)), endothelial cells (MDS1 and EVI1 complex locus (
Mecom), fms-related receptor tyrosine kinase 1 (
Flt1), cysteine/tyrosine-rich 1 (
Cyyr1), and adhesion G protein-coupled receptor L4 (
Adgrl4)), excitatory neurons (cAMP-regulated phosphoprotein 21 (
Arpp21) and solute carrier family 17 member 7 (
Slc17a7)), inhibitory neurons (glutamate decarboxylase 1 (
Gad1) and glutamate decarboxylase 2 (
Gad2)), MG (cathepsin S (
Ctss), C-X3-C motif chemokine receptor 1 (
Cx3cr1), LYN proto-oncogene, Src family tyrosine kinase (
Lyn), and dedicator of cytokinesis 8 (
Dock8)), neural progenitor cells (NPCs; nucleolar and spindle associated protein 1 (
Nusap1), DNA topoisomerase II alpha (
Top2a), and centromere protein F (
Cenpf)), oligodendrocytes (OLs; myelin basic protein (
Mbp), myelin-associated oligodendrocyte basic protein (
Mobp), SRY-box transcription factor 10 (
Sox10), and UDP glycosyltransferase 8 (
Ugt8)), oligodendrocyte precursor cells (OPCs; myelin transcription factor 1 (
Myt1), multiple EGF like domains 11 (
Megf11), and platelet derived growth factor receptor alpha (
Pdgfra)), and pericytes (collagen type III alpha 1 chain (
Col3a1), ATP binding cassette subfamily C member 9 (
Abcc9), and collagen type I alpha 1 chain (
Col1a1)) (
Figs. 3(d) and
(e), Fig. S4(c) in Appendix A). In comparison with that in the sham group, the proportion of excitatory neurons was markedly diminished in the model group, whereas the number of ACs increased at various time periods (
Fig. 3(c)), suggesting neuronal damage and death due to HIBD. These results indicate changes in cell distribution resulting from HI, and these changes were partially recovered after SAC treatment. Interestingly, we noted a substantial rise in the percentage of MG from 24 h to 7 d post HI. This finding is consistent with previous reports [
31,
32], which suggested that the nervous system of rat pups is actively developing during the first two weeks after birth, leading to a gradual increase in MG numbers, with notable morphological diversity. Moreover, the development of axons and the formation of synapses may also stimulate the proliferation of MG during this period [
33].
3.4. SAC reduces the proportion of Stat3+ ACs and attenuates acute inflammation
ACs are critical in the pathophysiology of HIBD due to their multifaceted roles in regulating inflammatory responses and responding to injury [
3]. To investigate the heterogeneity of ACs in depth, we clustered these cells into eight subsets according to their characteristic genes (
Fig. 4(a), Fig. S5(a) in Appendix A). We observed that variations in these subsets occurred mainly at 24 h (
Fig. 4(b)). Using the AC activation score from the Gene Set Enrichment Analysis (GSEA) dataset, we found that AC activation increased post HI but declined following SAC treatment (Fig. S5(b) in Appendix A).
Notably, the most pronounced change in ACs was a sharp increase in the
Stat3+ AC subset ratio in the model group 24 h post HI, becoming the major subset, accounting for 43.02%, compared with 0.26% in the sham group (
Fig. 4(b)). The characteristic genes of
Stat3+ AC were enriched in the acute inflammatory response and regulatory cell response to stress (Fig. S5(a)). Based on the GSEA dataset,
Stat3+ AC exhibited the highest acute inflammatory response score (
Fig. 4(c)). Furthermore, this subset highly expressed canonical proinflammatory AC marker genes, such as
Gfap,
S100b, and
Stat3 (
Fig. 4(d)).
HI upregulated the expression of most acute inflammation-related genes in
Stat3+ AC, whereas SAC reversed this effect (
Fig. 4(e)). In addition, we examined the activity of TFs related to the inflammatory response (aryl hydrocarbon receptor (AHR), proto-oncogene
c-Rel (REL), B-cell lymphoma 3 protein (BCL3), forkhead box O3 (FOXO3), PBX/knotted 1 homeobox 1 (PKNOX1), interferon regulatory factor 3 (IRF3), and basic helix-loop-helix family member e41 (BHLHE41)) in the
Stat3+ AC subset and found a general decrease in their levels after SAC treatment (
Fig. 4(f)). Particularly, REL is involved in various inflammatory responses and nuclear factor kappa B (NF-κB) signaling transduction [
34,
35]; BCL3 acts as a transcriptional activator in the nucleus, promoting the transcription of NF-κB target genes [
36,
37]. The specific upregulation of inflammation-related TFs in
Stat3+ AC, along with selective suppression after SAC treatment (Fig. S5(c) in Appendix A), highlights the importance of this cell population in the pathogenesis of HIBD and the therapeutic effects of SAC. Consistent with these findings, HI injury increased the expression of multiple proinflammatory cytokine genes (Fig. S5(d) in Appendix A).
Our findings support a mechanism by which HI-induced nuclear translocation of REL and BCL3 in ACs activates NF-κB signaling and
Stat3 expression, driving proinflammatory cytokine release that amplifies acute inflammation. To verify this mechanism, we further examined the rat serum inflammatory cytokine profile using a protein chip at 24 h after HI, revealing that SAC inhibits the release of multiple proinflammatory factors induced by HI (
Fig. 4(g)). ELISA and immunofluorescence staining also revealed that the expression changes of inflammatory cytokines (IL-6, IL-18, and macrophage colony-stimulating factor (M-CSF)) in rat brain tissues across groups were consistent with the serum inflammatory cytokine profiles (Figs. S5(e)-(g) in Appendix A). Immunofluorescence quantification further validated the specific inhibition of SAC in
Stat3+ AC expansion post HIBD (
Figs. 4(h) and
(i)), which was mainly distributed in the infarct core and penumbra of the ischemic hemisphere (Fig. S5(h) in Appendix A). Furthermore, multiplex immunofluorescence staining of rat brain sections found that within
Stat3+ AC, the expression of REL and BCL3 in the nucleus was significantly increased in the model group compared to the sham group, indicating enhanced activity of these inflammation-related TFs (Figs. S5(i) and (j) in Appendix A). Additionally, SAC treatment reversed this effect. These findings, combined with the snRNA-seq data showing reduced expression of proinflammatory genes within
Stat3+ AC (
Fig. 4(e), Fig. S5(c)), strongly suggest that SAC directly modulates neuroinflammation within the affected central nervous system (CNS) compartment.
In vitro validation using CTX-TNA2 ACs demonstrated that OGD/R induced the upregulation of
Stat3,
Bcl2, and
Rel mRNA, and increased REL protein in the nucleus, whereas SAC exhibited an inhibitory effect (
Fig. 4(j), Figs. S5(k) and (l) in Appendix A), consistent with the changes in
Stat3+ subgroup numbers and TF activation. ELISA and immunofluorescence staining confirmed that SAC significantly improved the expression of inflammatory factors, including IL-6 and IL-1β, in ACs (
Fig. 4(k), Figs. S5(m) and (n) in Appendix A). Western blotting revealed that the model significantly activated the STAT3 pathway in ACs, whereas SAC effectively inhibited the phosphorylation of this protein (
Figs. 4(l) and
(m)). Furthermore, the changes in total STAT3 protein were consistent with the aforementioned experimental results (Fig. S5(o) in Appendix A). Transient transfection knocking down
Stat3 gene expression in CTX-TNA2 cells significantly reduced the release of inflammatory factors in ACs after OGD/R modeling (
Figs. 4(n) and
(o)), and SAC may have concurrent STAT3-independent synergistic anti-inflammatory mechanisms. These
in vitro results provided mechanistic support for the direct anti-inflammatory effects of SAC on ACs, a major cell type involved in neuroinflammation within the CNS [
3].
To further validate the conserved mechanisms observed across species, we investigated
Stat3+ AC activation in human brain organoids. Immunofluorescence staining of human brain organoid sections revealed a significant increase in STAT3 expression in regions populated by GFAP-expressing ACs, following HIE induction (
Figs. 4(p) and
(q)). This finding indicated an elevated number of STAT3
+GFAP
+ cells in the HIE brain organoids. Notably, SAC administration significantly attenuated the HIE-induced surge of
Stat3+ AC in human brain organoids (
Figs. 4(p) and
(q)). This cross-species validation aligned with and substantiated previous snRNA-seq findings from rat brain tissues, confirming conserved STAT3-mediated pathogenic mechanisms in astrocytic responses to ischemic injury across species.
Collectively, we performed in vitro and in vivo functional analyses, including cytokine analysis, Stat3 knockdown, and other molecular biology experiments, which provided strong evidence for the proinflammatory role of the Stat3+ AC subset in acute-phase HIE pathology and validated Stat3 as a key pharmacological target of SAC. These findings highlight the therapeutic potential of targeting AC-mediated inflammatory responses in CNS injury.
3.5. SAC treatment improves microglial immune homeostasis
MG, the intrinsic immune cells of the CNS, play crucial roles in the inflammatory cascade following HIBD, undergoing distinct molecular and functional alterations during pathology [
38]. Through comprehensive clustering analysis, we identified eight microglial subpopulations with unique genetic signatures (
Fig. 5(a), Fig. S6(a) in Appendix A). The GO enrichment analysis of the top 100 DEGs for the functional annotations of each subset is shown in
Fig. 5(b). Notably, the
Cped1+ MG subset exhibited a dramatic reduction at 24 h post HI that persisted through 7 d (
Fig. 5(b)), characterized by high expression of chemotaxis-related genes (chymotrypsin like elastase 1 (
Cela1),
Cx3cr1, C-C motif chemokine receptor 5 (
Ccr5), slit guidance ligand 2 (
Slit2), semaphorin 6A (
Sema6a), dedicator of cytokinesis 4 (
Dock4), integrin subunit alpha 9 (
Itga9), ectonucleoside triphosphate diphosphohydrolase 1 (
Entpd1), integrin subunit alpha 6 (
Itga6), astrotactin 1 (
Astn1), protocadherin 7 (
Pcdh7), and selectin P ligand (
Selplg);
Fig. 5(c)) and the biological functions of chemotaxis (
Fig. 5(b)). This depletion suggested compromised microglial surveillance and impaired injury response mechanisms, as chemotaxis is crucial for targeted migration to lesion sites [
39]. Based on the GSEA dataset, we observed SAC treatment attenuated microglial activation compared with the model group (Fig. S6(b) in Appendix A). Among all the subsets,
Grial1+ MG displayed the lowest activation state (Fig. S6(c) in Appendix A), occupying the starting position in differentiation trajectories (
Fig. 5(d)), consistent with a resting phenotype necessary for biological functions that maintain CNS homeostasis through synaptic regulation (
Fig. 5(b)) [
40]. According to the ratio, SAC effectively preserved resting MG populations and maintained chemotactic capacity across post HI timepoints (
Fig. 5(b)).
During acute HI phases, we observed expansion of the proliferative
Pola1+ MG subset associated with the biological function of DNA replication (
Fig. 5(b), Fig. S6(d) in Appendix A), reflecting characteristic microglial hyperplasia in early CNS injury responses. Proliferative MG are a key target in cerebral ischemic diseases, such as stroke [
41]. The DEG analysis revealed that SAC may promote
Pola1+ MG to release cytokines and modulate neutrophil chemotaxis and T cell proliferation, enhancing the immune response (
Fig. 5(e), Fig. S6(e) in Appendix A), while suppressing HI-induced ferroptosis and autophagy pathways (
Fig. 5(e)) that could impair cellular function [
42].
The
Mx1+ MG and phagocytic
Kif26b+ MG proportions gradually increased as HIBD progressed (
Fig. 5(b)).
Mx1+ MG was found to highly express multiple interferon response-related genes (poly (ADP-ribose) polymerase family member 14 (
Parp14), poly(ADP-ribose) polymerase family member 9 (
Parp9), interferon regulatory factor 1 (
Irf1), interferon induced with helicase C domain 1 (
Ifih1), and 2′-5′-oligoadenylate synthetase 2 (
Oas2)) and participate in the positive control of classical NF-κB signaling (
Fig. 5(d), Fig. S6(f) in Appendix A). The heatmap and gene set scoring show the pro- and anti-inflammatory functions of
Mx1+ MG and
Cd163+ MG, respectively (
Fig. 5(f), Figs. S6(g) and (h) in Appendix A). Notably, the marker genes employed were well-established markers of pro- and anti-inflammatory MG phenotypes, all of which have been extensively documented in previous studies [
43,
44]. Furthermore, our results showed that the
Cd163+ MG subset highly expresses IL-10
in vivo, further confirming its anti-inflammatory properties (Fig. S6(i) in Appendix A). Thus, we inferred that MG progressively move from the initial stages of the immune response to the regulatory stages of the inflammatory response with increasing ischemia duration. Furthermore, our findings imply that SAC may enhance the regulation of
Cd163+ MG driver genes (such as recombination signal binding protein for immunoglobulin kappa J region (
Rbpj)) expression by increasing the activity of TFs such as ELL-associated factor 6 (EAF6), myocyte enhancer factor 2C (MEF2C), and Krüppel-like factor 13 (KLF13), thereby exerting the anti-inflammatory effect (
Fig. 5(g), Fig. S6(j) in Appendix A). Immunofluorescence analyses demonstrated that SAC reduced M1 polarization at 7 d post HI (
Figs. 5(h) and
(i)), aligning with the pseudotime analysis results (Fig. S6(k) in Appendix A).
Moreover, the
Npl+ MG subset displayed dynamic temporal changes linked to the biological function of membrane lipid metabolism (
Fig. 5(b)). Early HI suppressed lipid transport functions, whereas later
Npl+ MG expansion in the model group suggested involvement in neurorepair through bioactive lipid mediator production and debris clearance [
45,
46]. The SAC-mediated stabilization of
Npl+ MG levels correlated with reduced inflammation and tissue damage (
Fig. 5(b)), supporting microglial metabolic reprogramming as a therapeutic target.
Herein, we propose that SAC has immunomodulatory capacity for MG, referring to the ability of SAC to dynamically regulate microglial activation states, balance pro- and anti-inflammatory phenotypes, restore homeostasis, regulate migration and proliferation, and promote reparative functions.
In conclusion, SAC regulates the activation of various MG subsets at different stages of HIBD in a multi-modality manner, from early proliferative and injury response to later inflammatory control and repair. This finding highlights the therapeutic potential of SAC in modulating microglial heterogeneity to restore CNS homeostasis after HIBD.
3.6. SAC enhances neurorestorative capacity by preserving protocadherin related 15+ (Pcdh15+) NPCs and MER proto-oncogene tyrosine kinase+ (Mertk+) OPCs
In addition to the inflammatory response, the dynamic changes in endogenous progenitor cells during the pathological process of HIBD deserve attention. Therefore, we systematically analyzed the dynamics of progenitor cells post HI.
First, we clustered NPCs into five subpopulations with their own specific genes (
Fig. 6(a), Fig. S7(a) in Appendix A). We also described the differences in subgroup proportions across the groups (
Fig. 6(b)). GSEA-based differentiation scoring revealed that
Pcdh15+ NPC demonstrated strong differentiation abilities (
Fig. 6(c)).
Pcdh15+ NPC specifically expressed the TFs SOX10 (
Fig. 6(d)), essential for oligodendrogenesis [
47]. The RNA rate analysis further corroborated their differentiation trajectory toward oligodendroglial phenotypes (
Fig. 6(d)).
We noted that
Pcdh15+ NPC exhibited elevated expression of neural development regulators (deleted in colorectal carcinoma (
Dcc) and roundabout guidance receptor 1 (
Robo1)), and
Pdgfra (Fig. S7(b) in Appendix A), a key driver of OPC proliferation. SAC preserved
Pcdh15+ NPC populations across post HI timepoints (
Fig. 6(b)), restored endocytic function, and stabilized metabolic pathways critical for intracellular homeostasis (Fig. S7(c) in Appendix A), countering HI-induced progenitor depletion observed in ischemic conditions.
We clustered the OPCs into five subpopulations (
Fig. 6(e), Fig. S7(d) in Appendix A) and showed differences in their proportions (
Fig. 6(f)). The
Mertk+ OPC subset uniquely expressed the immunomodulatory tyrosine kinase
Mertk (
Fig. 6(e)), which coordinates glial differentiation and immune surveillance [
48]. GO enrichment analysis also demonstrated the role of
Mertk+ OPC in the immunological response (Fig. S7(e) in Appendix A). OPCs can control CNS homeostasis and sense and respond to a variety of neuroinflammatory reactions [
49,
50]. Notably,
Mertk+ OPC demonstrated very low expression of the OPC markers
Pdgfra and chondroitin sulfate proteoglycan 4 (
Cspg4) and higher expression of differentiation-committed OL precursors (COP), namely G protein-coupled receptor 17 (
Gpr17) and bone morphogenetic protein 4 (
Bmp4) (
Fig. 6(g)), indicating an intermediate stage of OL precursor to OL transition [
51]. Pseudotemporal trajectory analysis confirmed this transitional state (Fig. S7(f) in Appendix A). SAC enhanced COP expansion (
Fig. 6(f)), evidenced by increased BMP4
+ oligodendrocyte transcription factor 2
+ (Olig2
+) cells at 7 d after HI (
Figs. 6(h) and
(i)), suggesting accelerated OL maturation.
Our findings establish SAC-modulated progenitor cell dynamics post HI, maintaining neural progenitor pools with oligodendrogenic potential via Pcdh15+ NPC preservation and promoting COP-mediated differentiation commitment in the OL lineage. These coordinated effects suggest that SAC enhances endogenous repair processes by facilitating OL regeneration, potentially supporting white matter reconstruction and functional recovery in HIBD.
3.7. SAC promotes OL differentiation and maturation and improves communication between anti-inflammatory MG and OLs
We clustered the OLs directly involved in myelin regeneration and obtained seven transcriptionally distinct subpopulations (
Fig. 7(a), Fig. S8(a) in Appendix A). Intriguingly, the proportion of OL subpopulations showed a dynamic shift with postnatal development in neonatal rats (
Fig. 7(b)). According to the signature genes enrichment results, nidogen 1
+ (
Nid1+) OL, oligodendrocytic myelin paranodal and inner loop protein
+ (
Opalin+) OL, and family with sequence similarity 124 member A
+ (
Fam124a+) OL were associated with myelination and OL differentiation (
Fig. 7(b)). According to GSEA dataset scoring,
Opalin+ OL and
Fam124a+ OL had stronger differentiation abilities, and SAC further enhanced these 7 d post HI (
Fig. 7(c), Figs. S8(b) in Appendix A). Pseudotemporal trajectory analysis positioned these subsets at terminal differentiation stages (
Fig. 7(d)), marked by the elevated expression of myelin maturation markers (
Mbp, myelin regulatory factor (
Myrf), and 2′,3′-cyclic nucleotide 3′-phosphodiesterase (
Cnp);
Fig. 7(e)). SAC treatment counteracted HI-induced acute mature OL depletion at 24 h and promoted their sustained recovery, especially that of the
Fam124a+ OL subset (
Fig. 7(b)). These findings were corroborated by immunofluorescence (
Figs. 7(f) and
(g)). LFB staining confirmed that SAC ameliorated the structural damage and pathological changes in the myelin sheath caused by HI (Fig. S8(c) in Appendix A). The MBP immunofluorescence staining and LFB staining directly demonstrated the neuroprotective effects of SAC on OLs maturation and myelin sheath repair, indicating that SAC promotes OLs differentiation and maturation.
As previously reported, anti-inflammatory MG are essential for the remyelination function of OLs [
52]. These MG provide support for myelin regeneration by inhibiting inflammatory responses, removing myelin debris, and promoting OL maturation [
53]. Thus, we used CellChat to examine cellular communication between
Cd163+ MG and
Nid1+ OL,
Opalin+ OL, and
Fam124a+ OL in different groups 7 d post HI. HI diminished the signal number and strength originating from
Cd163+ MG, suggesting intercellular crosstalk dysfunction, whereas SAC administration restored signaling complexity (
Fig. 7(h)).
Furthermore, significant LR pairs were identified, elucidating the crosstalk between
Cd163+ MG and
Nid1+ OL,
Opalin+ OL, and
Fam124a+ OL (
Fig. 7(i)). The communication probabilities of NRG3-ErbB4 and neuregulin 2 (NRG2)-ErbB4 pairs were markedly downregulated in the model group, particularly for NRG3-ErbB4, whereas this trend was reversed after SAC treatment (
Fig. 7(i)). We also analyzed the expression levels of these LR pairings among the various groups (Fig. S8(d) in Appendix A). NRGs are widely distributed throughout the nervous system and are involved in multiple neurodevelopmental processes. In MG, NRG signaling pathways inhibit inflammatory responses, reduce neuronal injury [
54], and promote the regeneration of nerve axons and myelin sheaths [
55]. NRG3, a member of the NRG family, can bind to ErbB4 receptors [
56].
Erbb4 is highly expressed in OLs and encodes a transmembrane receptor tyrosine kinase. Binding to the ligand, ErbB4 plays a crucial role in remyelination under pathological conditions [
57]. Immunofluorescence colocalization firstly confirmed the NRG3-ErbB4 pair interaction between anti-inflammatory MG and myelinating OLs (
Fig. 7(j)).
We validated our findings
in vitro in BV2 and MO3.13 cells. The experimental workflow for BV2 cells is depicted in
Fig. 8(a). BV2 cells were stimulated with IL-4 and IL-13 to induce an anti-inflammatory phenotype. Immunofluorescence staining revealed a significant upregulation of CD163 expression in stimulated BV2 cells (
Figs. 8(b) and
(c)), and Western blotting confirmed a marked increase in ARG1 expression following stimulation (
Fig. 8(d)), consistent with the expression profile of anti-inflammatory MG markers identified in our sequencing data. Following OGD/R induction and SAC treatment in these polarized cells, ELISA assays demonstrated a significant reduction in NRG3 protein secretion in the OGD/R group compared to the control (
Fig. 8(e)). This finding indicated that OGD/R impaired NRG3 ligand secretion from anti-inflammatory MG. SAC administration reversed this effect in a concentration-dependent manner (
Fig. 8(e)). Our MO3.13 cell studies further confirmed that OGD/R modeling significantly reduced ErbB4 receptor activation in OLs and MBP expression, which were reversed by NRG3 protein and SAC administration in a concentration-dependent manner (Figs. 8(f)-(i)). These findings demonstrate that SAC promotes NRG3 protein secretion and enhances its interaction with the ErbB4 receptor on OLs, thereby activating receptor-mediated signaling to facilitate myelin repair.
Taken together, our results suggest that SAC may promote myelin restoration and regeneration by alleviating mature OL injury and remodeling the cellular communication microenvironment, particularly through the NRG3-ErbB4 signaling axis between anti-inflammatory MG and mature OLs.
4. Discussion
4.1. Overview of study innovations
Compared with 2D cell cultures, organoids exhibit more complex and diverse cell types, along with structural features akin to
in vivo cells, mirroring cell-extracellular environment interactions [
58]. Thus, organoid models address the integrity and heterogeneity limitations of cellular models [
58,
59]. Importantly, the establishment of the HIE brain organoids model was validated by monitoring HIF-1α, a gold-standard biomarker for assessing hypoxia-related cellular responses. HIF-1α expression was quantitatively analyzed at both hypoxia and reoxygenation phases using protein-level assays, ensuring the robustness of the pathological mimicry in our organoid system. The human brain organoid model used in this study recapitulates key features of the neonatal HIE [
14,
60], including the expression changes in key factors in neonatal hypoxia-vulnerable brain regions, hypoxia-induced hallmark molecular responses, multiple types of neuronal death, and neural network dysfunction. Thus, this model is a valuable tool for studying the pathogenesis of HIE and evaluating the efficacy of potential therapeutic interventions. We employed HIE human brain organoids to assess the efficacy of natural products for cerebral ischemic diseases, mitigating species-specific variations in animal models, thereby facilitating a more comprehensive and precise elucidation of pharmacological effects.
HIBD, a prevalent perinatal neural injury, has long posed a challenge in neuroscientific research [
61]. This study precisely tracked the temporal dynamics of single-nucleus transcriptomes post HI injury, elucidating for the first time the pathology of HIBD and its restoration process under drug intervention at the single-cell level [
62]. Mild hypothermia therapy has been considered a common treatment method for HIE; however, its therapeutic effect is limited [
63]. Combining this approach with drug injections has become increasingly popular [
5], highlighting the necessity of this study. Compared to other disease studies, the use of snRNA-seq in this study offers distinct advantages, including pioneering application in neonatal brain disease studies, technical superiority for brain tissue analysis, and minimized technical defects, enabling a more thorough comprehension of HIE pathophysiology and the effects of SAC treatment. Our observations fill the gap in single-cell data for neonatal rat brains and uncover many prospective biomarkers that can be used as diagnostic or adjunct therapy targets for HIE.
4.2. In-depth analysis of results
The time points of 24 h and 7 d were chosen in this study to capture the distinct phases of HIBD progression. The 24 h time point reflected the acute phase. Specifically, this phase was characterized by immediate cellular responses such as excitatory neuron death (
Fig. 3(c)), neuroinflammation (
Fig. 4), and breakdown of glial-immune homeostasis (
Fig. 5). These processes mirror the clinical observations in early HIE pathophysiology involving energy failure, excitotoxicity, and acute inflammation [
29]. The 7 d time point represented the subacute phase. During this phase, the brain began to initiate recovery processes, including the differentiation of NPCs and the maturation of OLs, which are essential for myelin regeneration. The inflammation microenvironment transitioned to a regenerative state. Key processes included OL differentiation and maturation (
Fig. 6,
Fig. 7), microglial reprogramming (
Fig. 5), and initiation of remyelination (Fig. S8(c)). These findings correspond to the clinical subacute phase [
30], where endogenous repair processes dominate but are often insufficient. Regarding dose selection, the 20 mg∙kg
−1 SAC dose for rodent studies was chosen in alignment with previous neuroprotection reports in cerebral ischemia models [
7] and our preliminary tests comparing 10 and 20 mg∙kg
−1 doses. The intermediate 5 µmol∙L
-1 dose for brain organoids studies optimally attenuated HIE-induced injury and demonstrated a balance between efficacy and toxicity.
LC-MS analysis confirmed that intraperitoneally injected SAC could penetrate the blood-brain barrier (BBB) and distribute in the brain. A distinct SAC peak (retention time = 1.45 min) was detected in all SAC-treated groups (20, 40, and 80 mg∙kg−1), but not in the saline control (Fig. S9 in Appendix A). The peak areas increased with the administered dose, indicating dose-dependent brain accumulation. These results provide direct evidence for the BBB penetration of SAC.
Persistent brain inflammation is a key mechanism leading to impaired brain function during development [
64]. Our findings reveal temporally distinct glial responses. We found that in the early stage after HI, ACs primarily participated in the acute inflammatory response, with a surge in
Stat3+ AC, while MG proliferated extensively. The increase in proinflammatory cytokine levels in peripheral blood may exacerbate nerve damage caused by HI [
65]. It has been reported that astroglial STAT3 is involved in the inflammatory response after HIBD [[
66], [
67], [
68], [
69]], possibly via the regulation of IL-6 signaling [
66] and IL-1β expression [
67]. The brain tissues of HIE patients exhibit upregulated STAT3-regulated inflammatory markers (such as C3) [
66], while AC-specific
Stat3 knockout animal models demonstrated reduced IL-1/IL-6 responsive ACs, attenuated neuroinflammation, and functional recovery [
66]. The findings of the aforementioned
in vivo experiments are consistent with our results. Targeted STAT3 inactivation may become a therapeutic strategy for HIE [
70]. Nevertheless, the effect of SAC on this target has not yet been determined.
In our study, SAC significantly inhibited the generation of
Stat3+ AC subsets and dampened their proinflammatory TF activity, thereby removing inflammatory mediators and mitigating acute inflammation. Critically, the reduction of
Stat3+ AC following SAC treatment in human brain organoids further supports the translational relevance, and the consistency of the findings with our rat brain tissue snRNA-seq data reveals conserved mechanisms across species. Existing literature suggests that REL interacts with inhibitor of kappa B (IκB), and inflammatory signals trigger the activation of the IκB kinase complex, leading to IκB phosphorylation and degradation, thereby activating the classical NF-κB pathway [
71]. Upon nuclear translocation, REL directly activates proinflammatory genes, further activating STAT3 [
71]. BCL3 can bind to inhibitory p50 homodimers, converting them into transcriptional activation complexes [
72] or removing inhibitory p50 dimers, enhancing NF-κB pathway activity and amplifying the inflammatory response [
73]. Furthermore, crosstalk exists between the NF-κB and STAT3 pathways [
74], resulting in a synergistic amplification effect that further enhances the inflammatory response [
75]. In our study, SAC was found to suppress the expression levels of
Rel and
Bcl3 (
Fig. 4(j)) and the nuclear localization of REL and BCL3 TFs (Figs. S5(i)-(l)), potentially blocking NF-κB activation and reducing the synergistic amplification effect between the NF-κB and STAT3 pathways, reducing the release of inflammatory factors and alleviating the inflammatory response. SAC attenuated the expression of proinflammatory genes, providing direct evidence for the regulatory effects of SAC on neuroinflammation within the CNS. Our
in vitro gene-silencing approaches confirmed that
Stat3 is the primary pharmacological target of SAC. This functional validation in cellular models further supports the mechanistic link between
Stat3 inhibition and the SAC-driven attenuation of neuroinflammatory cascades. Regarding the result that SAC may have concurrent STAT3-independent synergistic anti-inflammatory mechanisms, the regulation of TF activation by SAC could potentially inhibit the NF-κB pathway.
Stat3 silencing only partially alleviated inflammation, which could be due to incomplete
Stat3 knockdown and potential off-target effects using siRNA. The early modulation of SAC in neuroinflammation indirectly influenced the regenerative processes in the subacute phase by creating a more favorable environment for OL differentiation and maturation [
76,
77].
In addition, we found that the ratio of
Mx1+ MG increased gradually with disease duration, which expressed higher levels of M1-like microglial markers. Previously, it was reported that the number of M1 MG increases in the first 14 d after cerebral ischemia [
78]. The protective MG were found to fill the ischemic core 24 h after brain injury. On day 7, a sharp increase in CD68 expression appeared in the border area and ischemic core, accompanied by gradual enhancement of microglial phagocytic function and decreased expression of protective microglial marker genes [
78]. In
Cd163+ MG, it was implied that SAC may enhance the regulation of some driver genes (such as
Rbpj) by increasing the activity of TFs such as KLF13. Prior research has demonstrated that the absence of
Klf13 induces alterations in the immunological response of the nervous system, culminating in neurological impairments [
79]. The increased expression of
Rbpj may regulate cell differentiation by mediating the neurogenic locus notch homolog protein 1 (Notch1) signaling pathway, alleviating cerebral hemorrhage injury [
80]. Therefore, it was inferred that SAC might activate the TF KLF13 to increase the expression of
Rbpj, thereby mediating the Notch1 signaling pathway, regulating the inflammatory response, and alleviating HI-induced brain injury.
Previous studies evaluating drug efficacy post HI injury have predominantly focused on neuroprotection, with limited reports on neurorepair and regeneration, and the neurogenic response in neonates following HI remains poorly understood [[
81], [
82], [
83]]. In this study, we observed a strong increase in
Pcdh15+ NPC following SAC treatment, which was associated with cell adhesion and synaptic structure maintenance. NPCs may indirectly participate in the process of cell adhesion by secreting various growth factors and extracellular matrix constituents [
84]. In the mammalian CNS, endogenous NPCs, mainly existing in the SVZ of the lateral ventricles and the subgranular zone (SGZ) of the dentate gyrus [
85], proliferate and differentiate into functional neurons and glial cells in the damaged areas, completing regeneration and repair [
86].
Myelin, the protective sheath of the CNS, is crucial for neural signaling. Dysplasia or damage can disrupt conduction, potentially resulting in long-term consequences in children [
87]. In newborns, HI in the critical period of white matter development damages white matter OPCs and immature OLs, resulting in myelination disruption [
88]. Thus, abnormal myelination or demyelination is a major pathological manifestation of HIE [
89]. However, specific therapeutic strategies for myelin regeneration remain limited. Our study elucidated the process of OL lineage changes after cerebral ischemia, from the proliferation of OL precursors to the differentiation of mature OLs. As expected, we noted a progressive rise in the number of mature OLs postnatally in rat pups, indicating ongoing myelin development. HI injury led to a near-complete loss of mature OLs, whereas SAC treatment exhibited significant protective effects. The rescue of mature OLs (such as
Fam124a+ OL) during the acute phase not only mitigated early demyelination but also provided a cellular foundation for rapid remyelination in the subacute phase, which is crucial for the restoration of myelin integrity and overall neural function.
Additionally, SAC alleviated the symptoms of HIBD by promoting communication between MG and OLs. SAC may facilitate the secretion of NRG3 by anti-inflammatory MG. When NRG3 activates the ErbB4 receptors on OLs, a series of downstream signaling pathways is activated, contributing to myelin repair and neuroprotection. From this perspective, our findings also provide novel insights for clinical treatment related to myelin regeneration in HIE.
4.3. Limitations of the study and prospects
The current study has several limitations. The brain organoids we established lack vascular structures and immune cells [
90], which are crucial components of the brain microenvironment and play significant roles in ischemic diseases. The lack of vascularization and innate immune components may affect nutrient and oxygen diffusion within the organoids, influencing cell survival and response to injury, and limit studies on neuroinflammation and immune pathway interactions [
91,
92]. Although there have been many breakthroughs in related research in recent years [
93], further optimization is still required. Due to the challenges in extracting and isolating various glial cells from brain organoids, we did not extensively validate pathways and targets within glial cells using this model. Moreover, we did not investigate the long-term pathological changes in HIBD or the long-term pharmacological effects of SAC in detail.
To further validate the mechanisms, future studies should focus on
in vivo target validation using conditional knockout rats, longitudinal studies to assess long-term neurodevelopmental outcomes, and biomarker validation in clinical samples, such as cerebrospinal fluid or blood. The vascularization and immune system improvement in human brain organoids may be addressed by introducing coculture systems with endothelial progenitor cells or vascular organoids [
94], developing bioengineered organoid-on-a-chip platforms [
95], and integrating MG derived from hPSCs [
96]. Integrating single-cell and spatial transcriptomics analyses of vascularized human brain organoids containing MG will strengthen cross-species validation by confirming conserved molecular signatures and cellular interactions observed in the animal model [
97,
98], bridging critical gaps between preclinical animal studies and human clinical outcomes. Addressing key translational hurdles, such as pharmacokinetic studies combined with mass spectrometry imaging and spatial metabolomics, formulation development, and clinical trial design, is also essential prior to clinical application. In particular, given that the mild hypothermia treatment combined with other neuroprotective strategies is the current standard of care for HIE [
63], the potential of combining mild hypothermia treatment with SAC warrants further clinical investigation for improved treatment outcomes. Such investigations will further validate the therapeutic potential of SAC and inform dosing regimens for clinical translation.
5. Conclusions
In this study, we performed a comprehensive snRNA-seq analysis to examine the transcriptome dynamics mechanism of HIBD pathology and SAC treatment. Our findings demonstrated that SAC effectively alleviated Stat3+ AC-mediated acute inflammatory responses, in addition to helping maintain glial immune homeostasis. During the subacute phase, SAC further promoted the differentiation and maturation of OLs, improved the interaction between glial cells, and promoted myelin regeneration. According to our findings, SAC exerted broad neuroprotective effects in HIBD. Therefore, this study emphasized the importance of SAC treatment for HIBD and proposed potential novel biomarkers for HIE diagnosis. Our dataset provided resources for in-depth investigations into HIE pathophysiology and pharmacological development.
CRediT authorship contribution statement
Xuan Mou: Writing - original draft, Visualization, Validation, Project administration, Formal analysis, Data curation. Lu Li: Writing - original draft, Project administration, Methodology, Funding acquisition, Conceptualization. Xinyue Liu: Investigation, Data curation. Aolin Zhang: Visualization, Methodology, Data curation. Tao He: Visualization, Methodology, Investigation. Baofeng Rao: Data curation. Jiatian Zhang: Methodology, Data curation. Renjie Chen: Visualization, Software. Malte Spielmann: Writing - review & editing. Chi Chiu Wang: Writing - review & editing, Funding acquisition, Conceptualization. Bin Cong: Writing - review & editing, Methodology, Conceptualization. Xiaohui Fan: Writing - review & editing, Methodology, Funding acquisition, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This study was supported by the Zhejiang Province Traditional Chinese Medicine Science and Technology Project (GZY-ZJKJ-24076), the High-level Talents Special Support Program of Zhejiang Province (2024-KYY-GXJS-0026), the ‘Pioneer’ Research and Development Program of Zhejiang (2023C03004), the Transverse Research Project of Zhejiang University (2023-KYY-A070350007), the Theme-base Research Scheme, Research Grants Committee (T13-602/21-N), and the Starlit South Lake Leading Elite Program (2023A303005). We would like to thank the LC-Bio Technologies Co., Ltd., China for the technical support of snRNA-seq.
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2025.09.010.
PDF
(8277KB)
