ROS-responsive, brain- and M1 microglia-targeting modified ginkgetin-loaded smart liposomes ameliorate cerebral ischemia by HIF-1α-mediated negative regulation of microglia pyroptosis
Xueyuan Li, Zi Ye, Wenyang Nie, Bao Zhou, Haocheng Qin, Zhijie Zhao, Yilong Fu, Dun Liu, Shaowei Zheng, Liangyu Wang, Jun Ma, Jingying Guo, Beibei Nie, Yan Lu, Dongming Yan, Zhiwen Luo, Qingshan Wang, Meng Bian, Hui Jiang, Di Chen
TL;DR
A new smart liposome delivery system targets brain and M1 microglia to reduce stroke damage by inhibiting harmful inflammation and cell death.
Contribution
Development of ROS-responsive, brain-targeted liposomes that deliver ginkgetin to suppress microglia pyroptosis via HIF-1α regulation.
Findings
3R@Lipo/Gink reduced infarct size and improved motor performance in stroke models.
The treatment suppressed HIF-1α and NLRP3-dependent microglial pyroptosis.
Therapeutic effects were reversed by FG-4592, confirming HIF-1α pathway involvement.
Abstract
Ischemic stroke triggers a cascade of mitochondrial dysfunction, oxidative stress, neuroinflammation, and pyroptosis, ultimately leading to neuronal injury and neurological deficits. Therapeutic efficacy is often limited by inadequate blood–brain barrier penetration and off-target effects. To address these challenges, we designed 3R@Lipo/Gink, a biocompatible liposomal formulation modified with an ROS-responsive TK polymer and two functional peptides—RVG29 for enhanced brain delivery and MG1 for microglia enrichment—to enable precise transport of ginkgetin, which has been demonstrated to exert neuroprotective effects through multiple potential mechanisms, to ischemic lesions. In a middle cerebral artery occlusion/reperfusion model, 3R@Lipo/Gink markedly reduced infarct size, alleviated neuronal injury, and improved motor performance. Single-cell RNA sequencing and in vitro co-culture…
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
Figure 11
Figure 12
Figure 13Peer 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
TopicsNeuroinflammation and Neurodegeneration Mechanisms · Immune cells in cancer · Inflammasome and immune disorders
Abbreviation list
(ROS)Reactive oxygen species(mNSS)Modified neurological severity score(GK)Ginkgetin(MCAO/r)Middle cerebral artery occlusion/reperfusion(TTC) staining2,3,5-Triphenyltetrazolium chloride(HIF-1α)Hypoxia-inducible factor-1α(GSDMD)Gasdermin D(BBB)Blood-brain barrier(TEM)Transmission electron microscopy(OGD/r)Oxygen-glucose deprivation/reperfusion(T-AOC)Total antioxidant capacity(MDA)Malondialdehyde(GSH)Glutathione(EE)Encapsulation efficiency(LC)Drug loading capacity
Introduction
1
Stroke is one of the leading causes of death and disability among residents in many countries. Ischemic stroke has accounted for the vast majority of newly diagnosed cases, with the age of onset demonstrating a concerning trend toward younger populations in recent years, severely compromising patients' quality of life and survival prognosis [1,2]. It imposes an increasingly heavy burden on healthcare systems due to its characteristic features of high incidence, recurrence rate, disability rate, and mortality. While pharmacotherapy remains the primary intervention modality and is extensively employed in clinical practice, the current therapeutic paradigm is fraught with multiple limitations [3,4]. Therefore, the exploration of novel therapeutic strategies is of critical importance in the management of ischemic stroke.
Understanding the pathophysiology of ischemic stroke constitutes the cornerstone for discovering novel therapeutic targets. The pathological cascade following cerebral ischemia is initiated by the upregulation of hypoxia-inducible factor-1α (HIF-1α), accompanied by mitochondrial dysfunction, which leads to aberrant accumulation of reactive oxygen species (ROS) and consequently precipitates oxidative stress [5,6]. This oxidative milieu subsequently promotes the release of pro-inflammatory cytokines, ultimately culminating in a robust and sustained neuroinflammatory response. Throughout this process, microglia, the resident innate immune cells of the central nervous system, play a pivotal role in orchestrating the inflammatory milieu. Moreover, pyroptosis, a caspase-dependent form of programmed cell death mediated by Gasdermin D (GSDMD), is characterized by its potent pro-inflammatory nature and further amplifies the neuroinflammatory cascade [[7], [8], [9]]. Following ischemic stroke under multifactorial stress conditions, glial cells, neurons, and brain microvascular endothelial cells (BMECs) sequentially undergo pyroptosis. Pyroptotic cells release small pro-inflammatory molecules prior to membrane rupture, thereby propagating an irreversible inflammatory cascade that exacerbates blood-brain barrier (BBB) disruption. Notably, microglial pyroptosis constitutes the initiating event of this pathogenic cycle [7]. It is certain that microglial pyroptosis is a very important driving factor in the early stage of ischemic stroke. Targeted inhibition of its pyroptosis may improve the symptoms of patients by delaying the time of disease progression, suppressing neuroinflammation, and reducing neuronal damage.
Ginkgetin (GK), a natural non-toxic flavonoid isolated from Ginkgo biloba leaves, has been demonstrated to exhibit potent anti-inflammatory and neuroprotective activities [10]. Studies have revealed that GK exerts anti-inflammatory regulatory effects across multiple pathologies, including cardiac, hepatic, pulmonary, dermal, and cerebral disorders,^[^[11], [12], [13], [14], [15]^]^ extending to therapeutic interventions against aging [16]. For ischemic stroke, it can down-regulate pro-inflammatory cytokines, inhibit neuroinflammation, and protect neurons in the infarcted area, leading to functional recovery [14,17]. Furthermore, GK has also been documented to possess multi-aspect therapeutic effects, including but not limited to anticancer, antifibrotic, pro-proliferative, and anti-apoptotic properties [[18], [19], [20], [21], [22]]. These findings underscore its significant potential in the clinical management of ischemic stroke. However, challenges persist regarding its low bioavailability, suboptimal blood-brain barrier (BBB) permeability [23], and incompletely elucidated therapeutic mechanisms. In this context, the development of advanced drug delivery systems emerges as a critical imperative to overcome these pharmacological limitations.
Liposomes, as a class of nanocarrier delivery systems, demonstrate significant clinical translation potential due to their unique advantages, including excellent biocompatibility, low toxicity, and biodegradability [24]. Plus, they exhibit superior performance in cellular uptake and transport processes, with their targeting specificity and controlled drug release capabilities being markedly enhanced through various surface modifications [25]. It has been found that the RVG29 peptide demonstrates specific binding to BBB endothelial cells, enabling BBB penetration and cerebral drug accumulation via receptor-mediated endocytosis [26,27], the MG1 peptide shows high affinity for M1-polarized microglia, effectively guiding drug carriers to inflammatory microglial populations to enhance therapeutic precision [28,29], and the TK functional group confers ROS-responsive properties, allowing carrier disintegration and cargo release in pathological oxidative stress microenvironments [30,31]. This evidence suggests that constructing a liposomal system surface-engineered with RVG29, MG1, and TK, while encapsulating GK as the therapeutic payload, could represent an innovative, safe, and highly efficient therapeutic strategy for ischemic stroke.
In this research, we successfully designed and constructed a novel liposomal nanocarrier, 3R@Lipo/Gink. As depicted in Fig. 1, following intravenous administration via the tail vein, 3R@Lipo/Gink utilizes the RVG29 peptide to target ischemic brain regions. Guided by the MG1 peptide, it further selectively binds to microglia. At last, it undergoes ROS-responsive disintegration in pathological oxidative environments, thereby releasing GK. Based on single-cell sequencing and subsequent molecular analysis, the liberated GK suppresses HIF-1α expression in microglia, concurrently reducing the production of inflammatory mediators (Caspase-1, IL-1β, IL-18), inhibiting the NLRP3 inflammasome and pyroptosis executioner GSDMD to block pyroptosis, while also modulating c-Myc transcription factor expression to attenuate microglial proliferative activation. This work establishes a novel precision therapeutic strategy and mechanistic framework for ischemic stroke treatment.Fig. 1Chematic depiction of this research(The green part on the left side of the picture represents the normal oxygen condition, and the red part on the right side represents the ischemia and hypoxia condition.)3R@Lipo/Gink, surface-modified with brain-targeted RVG29 peptide and microglial cell-targeted MG1 peptide and loaded with GK, was injected into the body via the tail vein and accumulated in the ischemic brain. After the MG1 peptide binds to the M1-type microglia, in the microenvironment with mitochondrial disorders and ROS enrichment, it triggers TK-induced liposome disintegration and drug release. The released GK exerts various therapeutic effects such as inhibiting oxidative stress, microglial pyroptosis, and proliferative activation through inhibiting HIF-1α. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 1
Results
2
Characterization and brain-targeting of 3R@Lipo/Gink
2.1
As illustrated in Fig. 2A, the 3R@Lipo/Gink was fabricated via the thin-film hydration method, incorporating functionalized DSPE-TK-PEG-RVG29 and DSPE-TK-PEG-MG1 while encapsulating GK. GK is a lipid-soluble molecule that, despite its inherent solubility characteristics, can be efficiently encapsulated within liposomes via methanol dissolution followed by film hydration method. This method maximizes encapsulation efficiency by facilitating the incorporation of GK into the aqueous interior of liposomes, although it cannot avoid being encapsulated in the lipid bilayer. The actual GK injection concentration for each mouse was 1.14 mg/kg, calculated by LC/EE. Transmission electron microscopy (TEM) revealed both drug-loaded 3R@Lipo/Gink and blank 3R@Lipo carriers exhibited spherical morphology with uniform dispersion (Fig. 2B). The particle size and zeta potential of liposomes were measured using a nanoparticle size potential meter. It was found that the particle sizes of both drug-loaded and empty-loaded liposomes were around 110 nm, and the difference could be ignored. And the zeta potentials of both are positive (Fig. 2C–F). Additionally, to ensure uniformity and stability in both blood circulation and cell culture medium, the particle size of the nanoparticle was further evaluated at 24, 36, and 48 h after dissolution in PBS. The results indicated that the hydrodynamic size of the nanoparticles remained unchanged across the three different time intervals, suggesting excellent colloidal stability (Fig. S3). As shown in Fig. 2G and H, the ultraviolet spectrum and the fluorescence spectrum also show that the two are similar. Additionally, we employed UV–Vis spectrophotometry to quantify the cumulative release profiles of GK from 3R@Lipo/Gink in H_2_O_2_ and PBS environments. As shown in Fig. 2I, a significant increase in GK release was observed after 4–24 h of co-culture at different H_2_O_2_ concentration gradients. However, PBS (without ROS)-mediated release reached only 40% after 10 h and only approximately 70% after 24 h. Therefore, varying concentrations of H_2_O_2_ can significantly facilitate the release of the drug. This ROS-triggered release pattern not only confirmed the oxidative stress-responsive design conferred by DSPE-TK-PEG modification, but also demonstrated that 3R@Lipo/Gink could effectively release GK under different physiological ROS levels within the ischemic penumbra. Fig. 2J demonstrates TEM-verified structural disintegration of liposomes under oxidative conditions.Fig. 2. Characterization and Brain-Targeting of 3R@Lipo/Gink(A) Flowchart of preparation of 3R@Lipo/Gink. (B) TEM image of 3R@Lipo/Gink and 3R@Lipo (scale bar: 100 nm). (C–E) Hydrodynamic distribution of 3R@Lipo/Gink and 3R@Lipo, n = 5. (F) Zeta potential of 3R@Lipo/Gink and 3R@Lipo by nanometer particle size potentiometer, n = 5. (G–H) The ultraviolet spectra and fluorescence spectra of 3R@Lipo/Gink and 3R@Lipo. (I) The cumulative release of GK from 3R@Lipo/Gink in different concentrations of H_2_O_2_ and PBS, n = 5. (J) TEM image of lysed liposomes. (K) Photograph of 3R@Lipo/Gink. (L) IVIS images of MCAO/r mice after injection of 3R@Lipo/Gink and Cy5.5 at different time points.Fig. 2
Post in vitro characterization, Cy5.5-labeled 3R@Lipo/Gink (Fig. 2K) was intravenously administered for in vivo biodistribution analysis. After drug injection, Fig. 2L revealed enhanced cerebral accumulation of fluorescence signals in the 3R@Lipo/Gink group compared to free Cy5.5 controls, validating the brain targeting capability conferred by RVG29 surface engineering. The cause of the fluorescence accumulation in the brain regions of the mice without 3R@Lipo/Gink administration might be the disruption of the blood-brain barrier in the lesion area after MCAO surgery, facilitating the easier entry of fluorescent material from the blood into the brain parenchyma. The fluorescence colocalization analysis of Cy5.5 and IBA-1 within the peri-infarct brain tissue of ischemic mice further demonstrated the ability of 3R@Lipo/Gink to target activated microglia (Fig. S2). Collectively, these findings substantiate that 3R@Lipo/Gink achieves brain-targeted delivery and ROS-activated therapeutic payload release, thereby exhibiting substantial translational potential for cerebral ischemia.
3R@Lipo/Gink improves neuro-behavioral function in MCAO/r mice
2.2
To evaluate the therapeutic potential of 3R@Lipo/Gink on neurobehavioral outcomes in middle cerebral artery occlusion/reperfusion (MCAO/r) mice, we conducted several behavioral assessments.
Throughout the experimental period, daily body weight monitoring and mNSS evaluations were systematically performed (Fig. 3A–C). Significant intergroup weight differences emerged during postoperative days 5–7, demonstrating that 3R@Lipo/Gink promote weight loss recovery of mice with cerebral ischemia. Notably, the 3R@Lipo/Gink therapeutic cohort demonstrated substantially lower mNSS scores compared to both MCAO/r controls and blank 3R@Lipo groups by postoperative day 7, indicating more favorable neurological recovery after treatment. In addition, cylinder testing (Fig. 3D and E) revealed progressive mitigation of bilateral forelimb-use asymmetry in 3R@Lipo/Gink-treated mice, with statistically significant improvements evident from postoperative day 6 onward.Fig. 33R@Lipo/Gink Improves Neuro-Behavioral Function in MCAO/r Mice(A) Line graphs of the daily body weight of the four groups (the Sham group, the MCAO/r group, the 3R@Lipo group and the 3R@Lipo/Gink group) of mice within 8 days, n = 8, ∗comparison between 3R@Lipo/Gink group and MCAO/r group, #comparison between 3R@Lipo/Gink group and 3R@Lipo group, n = 8, ∗^/#^p < 0.05, ∗∗^/##^p < 0.01. (B) Schematic diagram of mouse weighing. (C) Line graphs of the mNSS score of the four groups from day 1 to day 7, n = 8, ∗p < 0.05, ∗∗^/##^p < 0.01. (D) Line graph of the laterality index of bilateral forelimb contact walls in the four groups of mice in the cylinder test, n = 8, ∗^/#^p < 0.05. (E) Schematic diagram of the cylinder test. (F–G) Statistical analysis of total distance and center distance in the four groups of mice in the open field test, n = 8, non-significant (ns), ∗p < 0.05, ∗∗∗∗p < 0.0001.(H) Schematic diagram of the open field test, with the red line as the center. (I) Motion trajectory diagrams of the four groups in the open field test. (J) Photo of the mouse walking along the Catwalk trail. (K) Footprint images of the four claws of four groups of mice in the CatWalk XT gait test. (L–M) Statistical analysis of average speed and the ratio of the maximum contact area between the right claw and the left claw of the four groups of mice in the CatWalk XT gait test, n = 8, non-significant (ns), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. (N) Schematic diagram of the rotarod test. (O) Statistical analysis of the residence time on the rotating bar in the rotarod test of the four groups, n = 8, non-significant (ns), ∗p < 0.05, ∗∗∗∗p < 0.0001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 3
On postoperative day 7, additional behavioral assessments were conducted, including the Open field test, CatWalk gait analysis, and Rotarod test, with adequate rest periods between each test to ensure accuracy. The Open field test, evaluating locomotor activity and exploratory behavior, revealed that compared to the Sham group, the MCAO/r group exhibited significantly reduced motor ability with decreased total travel distance and center distance. In contrast, these two parameters were markedly increased in the 3R@Lipo/Gink group compared to both the MCAO/r and 3R@Lipo groups (Fig. 3F–I). CatWalk gait analysis compared the average speed and the maximum contact area ratio between the right and left paws across the four groups. As shown in Fig. 3J–M, the MCAO/r and 3R@Lipo groups performed significantly worse than the Sham group in these aspects, whereas the 3R@Lipo/Gink group demonstrated the opposite trends, showing enhanced movement speed and improved walking ability in the impaired limbs. The Rotarod test, primarily assessing motor coordination, indicated that compared to the Sham group, the MCAO/r group showed a significant reduction in the latency to remain on the rod. In contrast, the 3R@Lipo/Gink group exhibited a significantly increased latency compared to the MCAO/r group; however, the 3R@Lipo group still demonstrated shorter latency compared to the 3R@Lipo/Gink group (Fig. 3N and O).
Overall, these tests collectively demonstrated that 3R@Lipo/Gink has a significant therapeutic effect on neurological function recovery in mice with cerebral ischemia in multiple ways.
3R@Lipo/Gink attenuated cerebral infarct volume and protected neurons in ischemic penumbra of MCAO/r mice
2.3
Initially, to demonstrate the necessity of the delivery system, we administered equal doses of free GK to the MCAO/r mice to evaluate the respective therapeutic effects. The findings indicated that treatment with the nanomaterial could significantly reduce infarct volume and alleviate neurological dysfunction when compared to free GK. This outcome offers a preliminary validation of the efficacy of this nanomaterial (Fig. S4A–D).
To investigate the therapeutic efficacy of 3R@Lipo/Gink in ischemic stroke, cerebral infarct volume was first evaluated. TTC staining on postoperative days 3 and 7 revealed extensive infarct areas (white regions) in the MCAO/r group, whereas the 3R@Lipo group showed minimal improvement. In contrast, the 3R@Lipo/Gink group demonstrated a significant reduction in infarct volume compared to both MCAO/r and 3R@Lipo groups on day 3 and day 7 respectively, confirming the therapeutic efficacy of GK within 3R@Lipo/Gink (Fig. 4A–D). TUNEL staining (Fig. 4E and F) further corroborated these findings: the MCAO/r group exhibited substantially increased TUNEL-positive cells compared to Sham controls, while 3R@Lipo/Gink treatment markedly reduced apoptotic signals. No significant improvement was observed in the 3R@Lipo group, confirming the neuroprotective effects of 3R@Lipo/Gink in ischemic penumbra regions. Nissl staining analysis of neuronal survival status and Bax immunostaining monitoring neuron death in cerebral injury areas collectively validated the neuron-protective capacity of 3R@Lipo/Gink (Fig. 4G–I). TEM (Fig. 4J) revealed ultrastructural differences of neurons in the brain tissue around the infarction: Sham group mitochondria displayed intact cristae and continuous membranes with uniform matrix density. MCAO/r and 3R@Lipo groups exhibited moderate-to-severe mitochondrial swelling, disrupted cristae with membrane discontinuity, and reduced matrix density. Remarkably, 3R@Lipo/Gink group showed preserved mitochondrial morphology with decreased vacuolization and severe swelling compared to pathological controls.Fig. 43R@Lipo/Gink Attenuated Cerebral Infarct Volume and Protected Neurons in Ischemic Penumbra of MCAO/r Mice(A-B) TTC staining of the brains of the four groups of mice on the 3 and 7 days showed that the infarcted brain tissue was white and the normal brain tissue was red. (C–D) Statistical analysis of infarct area in the four groups, n = 5, non-significant (ns), ∗p < 0.05, ∗∗∗∗p < 0.0001. (E) Representative 40 × immunofluorescence images of TUNEL staining (Dapi, blue; TUNEL, red), scale bar: 50 μm. (F) Statistical analysis of the number of TUNEL-positive cells in the four groups, n = 5, non-significant (ns), ∗p < 0.05, ∗∗∗∗p < 0.0001. (G) Nissl staining in the cerebral cortex of ischemic areas, scale bar: 100 μm. (H) Statistical analysis of Nissl bodies in the cerebral cortex of ischemic areas in the four groups, n = 5, non-significant (ns), ∗p < 0.05, ∗∗∗p < 0.001. (I) Representative images of Bax staining in the four groups within ischemic areas, scale bar: 100 μm. (J) Representative images of mitochondrial ultrastructural changes observed under the TEM, scale bar: 500 nm and 200 nm. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 4
These preliminary findings demonstrate that 3R@Lipo/Gink effectively reduces cerebral infarct volume and attenuates neuronal death within the ischemic penumbra in MCAO/r mice. Next, we used single-cell sequencing to dig the potential therapeutic mechanism of 3R@Lipo/Gink-induced neuroprotection.
Single-cell profiling landscape of ischemic stroke mouse samples
2.4
Following confirmation of 3R@Lipo/Gink's therapeutic efficacy in ischemic stroke, to mechanistically link the observed neuroprotective effects with cell-type–specific transcriptional alterations, we conducted scRNA-seq on 3R@Lipo/Gink and MCAO/r groups to investigate its mechanism of action and characterize cellular-level modifications post-intervention.
After batch correction of all the cells in the sample, they were labeled as 13 cell types according to specific marker genes. Among these, Proliferating Cells and Neurons exhibited higher G2/M scores, whereas Proliferating Cells and CPECs showed higher nCount RNA and nFeature RNA (Fig. 5A; Fig. S12B). Notably, a larger proportion of Neurons were in the G2/M phase. In Microglia, the majority of cells were in the G1 phase (>50%), followed by cells in the G2/M phase (Fig. S12A), suggesting distinct proliferative and activation states among major cell populations after ischemic injury.Fig. 5. Single-Cell Profiling Landscape of Ischemic Stroke Mouse Samples(A) The UMAP plot displayed the cell types across all samples and visualized the distribution of G2/M score, S score, nCount RNA, and nFeature RNA among all cell types. (B) S The bubble plot presented the top five DEGs for each cell type. (C) T The volcano plot showed the top five upregulated and downregulated genes in microglial and neuronal subpopulations. (D) The word clouds illustrated the differential terms between microglial and neuronal subpopulations. (E) GO–BP analysis was performed based on the DEGs between microglial and neuronal subpopulations. (F) The proportions of various cell types were compared between the MCAO/r group and the 3R@Lipo/Gink group. (G) The UMAP plot identified four annotated microglial subpopulations, and the pie charts depicted the proportions of these four subpopulations in the MCAO/r and 3R@Lipo/Gink groups. (H) The bubble plot displayed the DEGs of the four microglial subpopulations, with red indicating expression in the 3R@Lipo/Gink group and blue indicating expression in the MCAO/r group. (I) The boxplots illustrated the differences in nFeature RNA, nCount RNA, S score, and G2/M score across the four microglial subpopulations. (J) GO enrichment analysis of the four microglial subpopulations. (K) GO enrichment analysis of microglia in the MCAO/r and 3R@Lipo/Gink groups. (L–M) Differences in CXCR1 and MMP9 signaling pathway scores among microglial subpopulations. (N) Differences in CXCR1 and MMP9 pathway scores between the MCAO/r and 3R@Lipo/Gink groups. (O) The heatmap displayed six microglia-specific regulon modules. (P) Ranking of transcription factor activity scores for the four microglial subpopulations in module M6. (Q) Fraction of variance across subtypes for transcription factors in module M6. (R) The heatmap showed the top five transcription factors (TFs) in the four microglial subpopulations. (S–T) Differences in the activity scores of transcription factors E2f8, E2f7, Tfdp1, and E2f2 across the four microglial subpopulations. (U) The bubble plot highlighted the differential expression of stemness-related genes among the four microglial subpopulations and between the MCAO/r and 3R@Lipo/Gink groups. (V) The UMAP plot displayed the differential expression of HIF-1α and Myc in the four subpopulations. (W)Differential expression of Hif-1α and Myc between the MCAO/r and 3R@Lipo/Gink groups. (X) The UMAP plot illustrated differences in Casp1 and Gsdmd expression across the four subpopulations. (Y) The boxplots showed the differences in Casp1 and Gsdmd expression between the MCAO/r and 3R@Lipo/Gink groups. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 5
To examine intercellular differences, we visualized DEGs using a bubble plot (Fig. 5B). The top 5 DEGs in Neurons were Ccnd2, Rtn1, Nnat, Sox11, Tubb2b, while those in Microglia were Rrm2, Mki67, Tuba1b, Tubb5, Top2a, highlighting pronounced transcriptional remodeling in both neuronal and microglial compartments. Volcano plots were further used to illustrate the top 5 upregulated and downregulated genes in each of the 13 cell types. In Microglia, upregulated genes included Fam167b, Ptchd1, Cd34, Liph, and downregulated genes included Ldlrad3, Mfsd6, Lrrc8c, Cpne2, H3f3a. In Neurons, upregulated genes were Trim67, Tmem151b, Sox3, Cplx1, Elovl2, and downregulated genes were Rhog, Arpc1b, Serinc3, Fth1, Itm2b (Fig. 5C; Fig. S12C). Word cloud analysis highlighted functional associations: Microglia-related terms included leukocyte, production, activation, immune, immunity, apoptotic, whereas Neuron-related terms included cycle, transition, localization, neuron, chromosome, mRNA, mitotic, assembly (Fig. 5D).
Based on DEGs, GO-BP analysis revealed that Microglia were primarily enriched in leukocyte migration and positive regulation of response to external stimulus, while Neurons were enriched in mRNA processing, RNA splicing, and RNA splicing via transesterification reactions. KEGG analysis showed that Microglia were enriched in Osteoclast differentiation and lysosome, whereas Neurons were enriched in cell cycle and spliceosome pathways (Fig. S12D). Further GO enrichment indicated that Neuronal DEGs were mainly associated with glial cell differentiation, gliogenesis, cerebral cortex GABAergic interneuron differentiation, positive regulation of neuron differentiation, and regulation of epithelial cell proliferation, whereas Microglial DEGs were associated with monocyte chemotaxis, chemokine-mediated signaling pathway, glial cell migration, response to chemokine, and cellular response to chemokine (Fig. 5E). Enrichment network analysis showed that DEGs in the 3R@Lipo/Gink group were mainly enriched in cellular response to chemical stimulus (Fig. S12E).
Collectively, these results suggest that microglia are a central hub of immune and inflammatory responses following ischemic stroke, whereas neurons primarily undergo transcriptional remodeling related to cell cycle regulation and differentiation, providing a rationale for subsequent in-depth analyses of microglial heterogeneity and microglia–neuron interactions under 3R@Lipo/Gink intervention.
Characterization of microglial subpopulations and transcriptional regulatory networks
2.5
Given the prominent immune-related transcriptional signatures observed in microglia, we next focused on microglial heterogeneity at single-cell resolution. We classified high-quality microglial cells into 4 microglial subpopulations based on differential gene expression, named C0 Igf1^+^ Microglia, C1 Siglech^+^ Microglia, C2 Birc5^+^ Microglia, and C3 Cd93^+^ Microglia (Fig. 5G).
In the C0 Igf1^+^ Microglia, C1 Siglech^+^ Microglia, and C2 Birc5^+^ Microglia subpopulations, cells from the 3R@Lipo/Gink group accounted for more than three-quarters of the total, whereas in the C3 Cd93^+^ Microglia subpopulation, cells from the 3R@Lipo/Gink group comprised approximately one-half, indicating a treatment-associated redistribution of microglial states. The marker genes used for naming each subpopulation were highly expressed within their respective subtypes. The top 5 DEGs of each subpopulation were visualized using a bubble plot (Fig. 5H). For C0 Igf1^+^ Microglia, the top 5 markers were Igf1, Lpl, Timp2, Ctsd, Cd5l; for C1 Siglech^+^ Microglia, Egr1, Siglech, Ifit3, Ifit2, Ifit1; for C2 Birc5^+^ Microglia, Mki67, Birc5, Top2a, Stmn1, Tuba1b; and for C3 Cd93^+^ Microglia, Gpnmb, Lyz2, Thbs1, F13a1, S100a11. Notably, the C2 Birc5^+^ Microglia subpopulation was characterized by strong expression of cell-cycle–associated genes, suggesting a proliferative phenotype.
We further distinguished DEGs between the 3R@Lipo/Gink and MCAO/r groups. C2 Birc5^+^ Microglia exhibited the highest G2/M and S scores, whereas C2 Birc5^+^ Microglia and C3 Cd93^+^ Microglia had higher nCount RNA and nFeature RNA (Fig. S13A; Fig. 5I), further supporting the hyperproliferative nature of C2 Birc5^+^ Microglia under ischemic conditions. Volcano plots highlighted the top 5 upregulated and downregulated genes for each subpopulation (Fig. S13B). In C2 Birc5^+^ Microglia, upregulated genes included Foxm1, Bub1, Cdca3, Ccna2, Birc5, and downregulated genes included Ctss, Vsir, Unc93b1, Selenop, Cst3.
GO analysis of DEGs revealed that C2 Birc5^+^ Microglia were primarily enriched in nuclear division, chromosome segregation, mitotic nuclear division, nuclear chromosome segregation, and DNA replication (Fig. 5J). Consistently, these enrichment patterns indicate that C2 Birc5^+^ Microglia represent a proliferative microglial subset strongly activated after ischemic injury. GO analysis comparing DEGs between the 3R@Lipo/Gink and MCAO/r groups showed that 3R@Lipo/Gink-enriched genes were associated with cellular cation homeostasis, myeloid cell differentiation, positive regulation of response to external stimulus, neuron death, cell-substrate adhesion, and cell-substrate junction organization (Fig. 5K).
GSEA analysis of C2 Birc5^+^ Microglia identified enrichment in mitotic sister chromatid segregation and regulation of chromosome segregation. We further evaluated repair-associated signaling pathways, including CXCR1 and MMP9. Gene set scoring revealed that both pathways were highest in C2 Birc5^+^ Microglia (Fig. 5L and M) and were elevated in the 3R@Lipo/Gink group compared to MCAO/r (Fig. 5N), suggesting that 3R@Lipo/Gink modulates both proliferative and reparative signaling within this key microglial subpopulation.
To examine transcriptional regulation, SCENIC analysis identified regulatory modules using the connectivity specificity index (CSI). AUCell scores clustered into 6 major modules (M1–M6) (Fig. 5O). Regulon activity scores varied across microglial subpopulations, with C2 Birc5^+^ Microglia showing markedly higher activity in module M6 (Fig. 5P; Fig. S13C). TFs with high fraction of variance in M6 included E2f7, E2f8, Ybx1, E2f1, E2f2, Tfdp (Fig. 5Q). Heatmaps highlighted the top 5 TFs for each subpopulation, with E2f8, E2f7, Tfdp1, E2f2, E2f1 ranking highest in C2 Birc5^+^ Microglia (Fig. 5R). UMAP visualization confirmed higher expression of these TFs in C2 Birc5^+^ Microglia (Fig. 5S), which was further corroborated by violin plots showing their elevated expression relative to other subpopulations (Fig. 5T).
From a stemness perspective, we found that the stemness-related gene HIF-1α exhibited a relatively high percentage of expressing cells but a low average expression level in the C3 Cd93^+^ Microglia subpopulation, and its expression percentage was lower in the 3R@Lipo/Gink group. Myc showed low expression across all subpopulations, but its expression percentage was higher in the MCAO/r group (Fig. 5U). These observations prompted us to further explore whether key transcriptional regulators associated with hypoxia response and aberrant proliferation may underlie microglial dysfunction after ischemic stroke.
Subsequently, UMAP plots visualized the expression patterns of Hif-1α, which leads to ROS generation, mitochondrial damage, and the expression of pyroptosis-associated proteins NLRP3, caspase-1, and GSDMD [32], and Myc, a gene involved in regulating the cell cycle and metabolism, whose increased expression leads to cell cycle dysregulation and hyperproliferation [33], across different subpopulations (Fig. 5V). Boxplots intuitively illustrated that the median expression level of Hif-1α was higher in the MCAO/r group than in the 3R@Lipo/Gink group, indicating a higher typical expression of Hif-1α in the MCAO/r group. Meanwhile, Myc expression levels in the 3R@Lipo/Gink group were slightly lower than those in the MCAO/r group (Fig. 5W).
The pyroptosis-related genes Casp1 and Gsdmd were expressed at lower levels in the 3R@Lipo/Gink group compared with the MCAO/r group (Fig. 5X and Y), providing single-cell–level evidence that 3R@Lipo/Gink suppresses microglial pyroptosis-associated transcriptional programs. We then scored the four microglial subpopulations and different cell cycle phases based on a pyroptosis-related gene set, as shown in Fig. S13D.
Proliferation, differentiation, and cell-cell crosstalk of the key microglial C2 subpopulation
2.6
Given the prominent proliferative features and transcriptional activity of C2 Birc5^+^ Microglia identified above, we next investigated the proliferation status, differentiation trajectory, and intercellular communication patterns of microglial subpopulations at single-cell resolution.
We performed CytoTRACE2 analysis on the four microglial subpopulations and found that C3 Cd93^+^ Microglia and C2 Birc5^+^ Microglia exhibited higher CytoTRACE2 relative scores, whereas C1 Siglech^+^ Microglia had the lowest scores (Fig. 6A and B), indicating greater differentiation potential and cellular plasticity in C2 Birc5^+^ Microglia. To investigate the differentiation and stemness heterogeneity of these subpopulations, we constructed a pseudotime trajectory using Monocle. The UMAP visualization of the trajectory showed that C3 Cd93^+^ Microglia and C2 Birc5^+^ Microglia were located at the late stages of pseudotime, whereas C1 Siglech^+^ Microglia and C0 Igf1^+^ Microglia resided at the early stages (Fig. 6C), suggesting a dynamic transition of microglial states during ischemic progression.Fig. 6. Differentiation Differences and Crosstalk Pathways of Microglial Subpopulations(A–B) CytoTRACE2 relative scores revealed differences among the 4 microglial subpopulations. (C) UMAP plot showed pseudotime differences across the microglial subpopulations. (D) Proportions of each microglial subpopulation in different states were presented; pie charts illustrated the relative cell numbers of all microglial subpopulations. (E) The constructed pseudotime trajectory progressed from left to right. Distributions of different states, cell cycle phases, and microglial subpopulations along the trajectory were visualized. (F) Expression patterns of marker genes in the 4 microglial subpopulations were shown along pseudotime. (G) Ridge plots illustrated the overall distribution regions of all microglial subpopulations. (H) UMAP plot displayed the trajectory of Lineage1 across all microglial subpopulations. (I) Chord diagram depicted the number and strength of interactions between C2 Birc5^+^ Microglia and other cells. (J) Heatmap (top) showed expression of ligand–receptor proteins in outgoing and incoming signaling patterns. Heatmaps (bottom) illustrated the importance of different cells in the MK signaling pathway network. Violin plots showed expression differences of ligand–receptor proteins across cell types in the MK signaling pathway. (K–L) Crosstalk interactions were visualized in the MK signaling pathway and the Mdk-Ncl pathway, showing interactions when various cells acted as sources or targets.Fig. 6
The pseudotime trajectory was further visualized, progressing overall from right to left, with one branch point (Fig. 6E). To clearly display the distribution of each subpopulation along this trajectory, we projected all four subpopulations onto the trajectory and supplemented this with faceted plots (Fig. S13E). Ridge plots showed that C2 Birc5^+^ Microglia and C3 Cd93^+^ Microglia mainly occupied the middle and late segments, C1 Siglech^+^ Microglia was predominantly at the early segment, and C0 Igf1^+^ Microglia was distributed across the early and middle segments (Fig. 6G). These results further support a differentiation continuum from early-response microglia toward proliferative and late-stage phenotypes.
Based on the branch point, the trajectory was divided into state1–3: state1 represented the early to middle stage, state2 the middle to late stage, and state3 an intermediate stage (later than state1, earlier than state2). In state1, C1 Siglech^+^ Microglia accounted for over half of the cells. In state2, C0 Igf1^+^ Microglia and C2 Birc5^+^ Microglia were predominant, and in state3, C0 Igf1^+^ Microglia comprised the majority (Fig. 6D). These proportions correlated with cell counts: C0 Igf1^+^ Microglia (44.66%), C1 Siglech^+^ Microglia (35.5%), C2 Birc5^+^ Microglia (12.34%), and C3 Cd93^+^ Microglia (7.5%). Gene expression along pseudotime confirmed these patterns, showing consistency with marker gene distributions (Fig. 6F). Notably, the enrichment of C2 Birc5^+^ Microglia in the middle-to-late pseudotime stages is consistent with their proliferative and transcriptionally active phenotype. Additionally, MCAO/r group cells and G2/M-phase cells were enriched at the late pseudotime stages (Fig. S13F).
We further constructed a developmental trajectory (Lineage1) using Slingshot, which progressed sequentially from C1 Siglech^+^ → C0 Igf1^+^ → C2 Birc5^+^ → C3 Cd93^+^ Microglia (Fig. 6H). Marker genes (Igf1, Siglech, Birc5, Cd93) were visualized along Lineage1, showing peak expression corresponding to each subpopulation's position on the trajectory (Fig. S13G). This ordered lineage reconstruction further supports C2 Birc5^+^ Microglia as a transitional yet highly active proliferative state preceding terminal differentiation.
Given the central position of C2 Birc5^+^ Microglia along the differentiation trajectory, we next explored whether this subpopulation plays a key role in microglia–neuron communication under ischemic conditions. To identify potential signaling pathways linking key microglial subpopulations to neurons and their relevance to ischemic stroke prognosis, we first visualized all cell-cell interactions (Fig. S13H). We then specifically examined C2 Birc5^+^ Microglia-mediated signaling, revealing both the intensity and number of interactions with other cell types (Fig. 6I). Analysis of outgoing and incoming ligand-receptor expression patterns highlighted the MK signaling pathway network (Fig. 6J). In this pathway, C2 Birc5^+^ Microglia exhibited high importance as receiver and influencer, while neurons were dominant as senders. Violin plots confirmed high expression of Ncl in C2 Birc5^+^ Microglia and Mdk in neurons, suggesting a directional neuron-to-microglia communication axis mediated by MK signaling.
Finally, chord diagrams and hierarchical plots validated that C2 Birc5^+^ Microglia engaged in crosstalk with neurons via the Mdk-Ncl pathway, both at the level of overall MK signaling and specifically through secretion-mediated interactions (Fig. S13I; Fig. 6K and L), indicating that C2 Birc5^+^ Microglia act as a key communication hub linking neuronal signals to microglial activation and proliferation in the ischemic microenvironment.
Single-cell characterization of neuronal subpopulations and identification of key signaling pathways
2.7
To determine whether microglial remodeling induced by 3R@Lipo/Gink is accompanied by coordinated neuronal transcriptional changes, we next performed single-cell characterization of neuronal subpopulations. Similarly, high-quality neurons were selected and classified into 4 neuronal subpopulations based on differential gene expression, named C0 Tead2^+^ Neurons, C1 Gad1^+^ Neurons, C2 Ube2c^+^ Neurons, and C3 Nhlh2^+^ Neurons (Fig. 7A). The top 5 DEGs of each subpopulation were visualized using a bubble plot (Fig. 7B). The top DEGs in C0 Tead2^+^ Neurons were Aldoc, Dbi, Gfap, Ckb, Mt2; in C1 Gad1^+^ Neurons were Gad1, Dlx6os1, Tiam2, Nrxn3, Shtn1; in C2 Ube2c^+^ Neurons were Tpx2, Prc1, Cenpa, Ube2c, Cenpf; and in C3 Nhlh2^+^ Neurons were Spock2, Snhg11, Atp1a3, Ndnf, Reln.Fig. 7. Single-Cell Features of Neuronal Subpopulations(A) UMAP plot showed the annotation of 4 neuronal subpopulations. (B) Bubble plot displayed the top 5 differentially expressed genes (DEGs) for each of the 4 neuronal subpopulations. (C) Proportions of different neuronal subpopulations in the MCAO/r and 3R@LipoGink groups were shown. (D) UMAP visualization of G2/M score, S score, nCount RNA, and nFeature RNA across all neuronal subpopulations. (E–F) Differences in G2/M score, S score, nCount RNA, and nFeature RNA among neuronal subpopulations and between experimental groups were presented. (G) Volcano plot showed the top 5 upregulated and downregulated genes in the 4 neuronal subpopulations. (H) Word cloud illustrated DEGs in the 4 neuronal subpopulations. (I) GO-BP analysis was performed based on DEGs from the 4 neuronal subpopulations. (J) GSEA analysis of C2 Ube2c^+^ Neurons was presented. (K) Heatmap displayed the top 5 transcription factors (TFs) across the 4 neuronal subpopulations. (L) The top 5 TFs with the highest specificity scores in C2 Ube2c^+^ Neurons were shown. (M) UMAP plots showed the expression distribution of these top 5 TFs in C2 Ube2c^+^ Neurons. (N) Violin plots compared the expression differences of the above 5 TFs across all neuronal subpopulations. (O) Heatmap illustrated 5 subtype-specific regulatory modules in neuronal subpopulations. (P) Rankings of TF activity scores in modules M1–M5 were presented for each neuronal subpopulation. Additionally, for module M4, TF activity rankings were shown across two experimental groups and three cell cycle phases. The fraction of variance of TFs in M4 was compared with other modules, accompanied by a violin plot visualizing Mybl1, the top-ranked TF.Fig. 7
Analysis of subpopulation proportions revealed that in the MCAO/r group, C3 Nhlh2^+^ Neurons and C0 Tead2^+^ Neurons predominated, whereas in the 3R@Lipo/Gink group, C1 Gad1^+^ Neurons and C2 Ube2c^+^ Neurons were more abundant (Fig. 7C). C2 Ube2c^+^ Neurons exhibited the highest G2/M score, and both C0 Tead2^+^ Neurons and C2 Ube2c^+^ Neurons had higher S scores. nCount RNA was highest in C2 Ube2c^+^ Neurons, and nFeature RNA was higher in both C2 Ube2c^+^ Neurons and C0 Tead2^+^ Neurons. Moreover, the 3R@Lipo/Gink group showed higher G2/M and S scores, while the MCAO/r group had higher nCount RNA and nFeature RNA (Fig. 7D–F).
Volcano plot analysis revealed that upregulated genes in C2 Ube2c^+^ Neurons included Cdca8, Birc5, Ube2c, Cenpf, Ckap2l, whereas downregulated genes included Ypel3 (Fig. 7G). Word cloud analysis indicated that C2 Ube2c^+^ Neurons were associated with cycle, mitotic, transition, spindle, phase, chromosome, spindle (Fig. 7H). GO-BP analysis showed enrichment in chromosome segregation, nuclear division, mitotic nuclear division, nuclear chromosome segregation, and mitotic sister chromatid segregation (Fig. 7I), and GSEA analysis highlighted terms such as mitotic cell cycle process, cell division, mitotic nuclear division, reproduction (Fig. 7J).
To explore transcriptional regulation, we visualized TF activity. Heatmap analysis showed the top 5 TFs in C2 Ube2c^+^ Neurons were E2f2, E2f3, E2f7, Atf4, Mybl1 (Fig. 7K). Based on RSS rankings and UMAP visualization, these TFs were highly expressed in C2 Ube2c^+^ Neurons (Fig. 7L and M), which was further confirmed by violin plots (Fig. 7N). SCENIC analysis and CSI matrix identified regulatory modules, grouping AUCell scores into 5 major modules (M1–M5) (Fig. 7O). Regulon activity scores varied across neuronal subpopulations, with C2 Ube2c^+^ Neurons showing notably higher activity in module M4 (Fig. 7P). TFs with high fraction of variance in M4 included Mybl1, Mxd3, Elf1, Gata6, Gata3, with Mybl1 displaying prominent expression in C2 Ube2c^+^ Neurons.
CytoTRACE analysis indicated that C2 Ube2c^+^ Neurons had lower heterogeneity, whereas C3 Nhlh2^+^ Neurons exhibited higher heterogeneity (Fig. S14A–B). CytoTRACE scores were highest in C2 Ube2c^+^ Neurons and C0 Tead2^+^ Neurons, and lower in C1 Gad1^+^ Neurons and C3 Nhlh2^+^ Neurons, suggesting differentiation progression from C2 Ube2c^+^ → C0 Tead2^+^ → C1 Gad1^+^ → C3 Nhlh2^+^ Neurons (Fig. S14C–D). CytoTRACE2 analysis confirmed higher relative and absolute scores in C0 Tead2^+^ and C2 Ube2c^+^ Neurons, with 3R@Lipo/Gink cells scoring higher than MCAO/r cells (Fig. S14E–G).
Monocle-based pseudotime analysis revealed a trajectory progressing left to right with two branch points (Fig. S14H). Projection of neuronal subpopulations showed C0 Tead2^+^ Neurons mainly at the early and middle stages, C1 Gad1^+^ Neurons at the late stage, C2 Ube2c^+^ Neurons at the early stage, and C3 Nhlh2^+^ Neurons at the late stage. Based on branch points, the trajectory was divided into state1–5, with state1 and state5 as initial stages, state2 as intermediate, and state3–4 as terminal stages. C2 Ube2c^+^ Neurons predominated in state1 and state5, C0 Tead2^+^ Neurons were present in all states, and other subpopulations were enriched at later states (Fig. S14I–K). The relative proportions of the four subpopulations were 29.71% (C0 Tead2^+^), 28.99% (C1 Gad1^+^), 26.09% (C2 Ube2c^+^), and 15.22% (C3 Nhlh2^+^). Heatmap visualization confirmed expression dynamics of marker genes along pseudotime (Fig. S14L).
Slingshot analysis constructed Lineage1, spanning all neuronal subpopulations in the order C2 Ube2c^+^ → C0 Tead2^+^ → C1 Gad1^+^ → C3 Nhlh2^+^ Neurons. Expression of marker genes Tead2, Ube2c, Gad1, Nhlh2 along Lineage1 corresponded to subpopulation positions (Fig. S14N). Stemness-associated genes Bmi1, Ezh2, Ctnnb1, Sox2 in C2 Ube2c^+^ Neurons showed distinct temporal patterns: Bmi1 and Ctnnb1 remained stable, Ezh2 peaked at early to mid-stages, and Sox2 peaked at early and mid-late stages.
Intercellular communication analysis revealed potential signaling pathways linking C2 Ube2c^+^ Neurons to other cell types. Chord diagrams visualized interactions between the 4 neuronal subpopulations and all other cells (Fig. S14O). Heatmaps of the MK signaling pathway showed the importance of each cell type as sender, receiver, mediator, or influencer, with Microglia highly important as receivers and C2 Ube2c^+^ Neurons as senders (Fig. S14P). Chord diagrams confirmed the number and strength of interactions from C2 Ube2c^+^ Neurons to other cells (Fig. S14Q). Finally, hierarchical analyses of the MK and Mdk-Ncl signaling pathways suggested that C2 Ube2c^+^ Neurons may interact with Microglia via secreted proteins (Fig. S14R–S).
3R@Lipo/Gink attenuated microglia pyroptosis and oxidative stress in ischemic penumbra of MCAO/r mice
2.8
Guided by scRNA-seq data on microglia in MCAO/r mice, we experimentally validated the anti-pyroptotic effects of 3R@Lipo/Gink. As shown in Fig. 8A, the mice were divided into groups and received their respective interventions. Initiating from differentially expressed stemness genes HIF-1α and c-Myc, immunofluorescence staining revealed significantly elevated HIF-1α fluorescence intensity in MCAO/r and 3R@Lipo groups, whereas 3R@Lipo/Gink treatment substantially reduced this signal (Fig. 8B and C). After that, Western blot analysis of HIF-1α, c-Myc, and pyroptosis-related proteins (NLRP3, GSDMD, GSDMD-N, Caspase-1, cleaved Caspase-1, p20, IL-1β, cleaved IL-1β, IL-18) demonstrated upregulated expression of all 11 proteins in MCAO/r versus Sham controls. While 3R@Lipo showed negligible effects, 3R@Lipo/Gink induced marked downregulation (Fig. 8D and E). Co-staining of GSDMD/NLRP3 with Iba1 (a microglial marker) confirmed congruent suppression of their expression in the 3R@Lipo/Gink group, and this suppression was particularly evident in the Zoom-in images (Fig. 8F–J). Oxidative stress evaluation via T-AOC, MDA, and GSH assays revealed that 3R@Lipo/Gink significantly decreased MDA levels while increasing T-AOC and GSH concentrations compared to MCAO/r controls (Fig. 8K–M).Fig. 83R@Lipo/Gink Attenuated Microglia Pyroptosis and Oxidative Stress in Ischemic Penumbra of MCAO/r Mice(A) Schematic diagram of mouse grouping and intervention. (B) Representative immunofluorescence image of HIF-1α among four groups (DAPI, blue; HIF-1α, red). Scale bar: 50 μm. (C) Fluorescence intensity of HIF-1α was quantitively analyzed, n = 5, non-significant (ns), ∗p < 0.05, ∗∗∗∗p < 0.0001. (D–E) The expression levels of HIF-1α, c-myc, NLRP3, GSDMD, GSDMD-N, Caspase 1, c- Caspase 1, p20, IL-1b, c- IL-1b and IL-18 in ischemic penumbra of mice across four groups were determined by Western Blot, n = 5, non-significant (ns), ∗p < 0.05 ∗∗p < 0.01 ∗∗∗p < 0.001 ∗∗∗∗p < 0.0001. (F–G) Representative immunofluorescence image of GSDMD/NLRP3 and Iba1 (DAPI, blue; GSDMD/NLRP3, red; Iba1, green). scale bar: 50 μm. (H–J) Fluorescence intensity of GSDMD, NLRP3 and Iba1 were quantitively analyzed, n = 5, non-significant (ns), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. (K–M) Statistical analysis of corresponding standard product concentration of MDA, GSH, and T-AOC, n = 5, non-significant (ns), ∗p < 0.05 ∗∗p < 0.01 ∗∗∗p < 0.001 ∗∗∗∗p < 0.0001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 8
These multimodal findings demonstrate that 3R@Lipo/Gink effectively mitigates neuroinflammation, suppresses pyroptosis, and ameliorates oxidative stress in ischemic cerebral regions of MCAO/r mice.
3R@Lipo/Gink attenuated microglia pyroptosis and oxidative stress in OGD/r BV-2 cells
2.9
In vitro cellular experiments further validated the therapeutic efficacy of 3R@Lipo/Gink. The cells were also divided into four groups: the control group was treated with PBS, while the three OGD/r-treated groups received PBS, 3R@Lipo/Gink, or 3R@Lipo, respectively (Fig. 9A). Flow cytometric analysis using Caspase-1/PI dual staining demonstrated significantly lower BV-2 microglial mortality in the 3R@Lipo/Gink group compared to OGD/r and 3R@Lipo controls (Fig. 9D and E). Then, parallel Western blot analysis of 11 pyroptosis-related proteins (Fig. 9J–U) and immunofluorescence co-staining of Caspase-1 (Fig. 9B and C), GSDMD (Fig. 9F and 9H), and NLRP3 (Fig. S5) revealed consistent trends: OGD/r and 3R@Lipo groups exhibited enhanced pyroptotic activation versus controls, whereas 3R@Lipo/Gink treatment effectively suppressed these effects. In addition, ROS quantification showed significantly attenuated DCFH-DA fluorescence intensity in the 3R@Lipo/Gink group compared to OGD/r models (Fig. 9G and I).Fig. 93R@Lipo/Gink Attenuated Microglia Pyroptosis and Oxidative Stress in OGD/r BV-2 cells(A) Schematic diagram of cell grouping and treatment. (B–C) Representative immunofluorescence image of Caspase1 staining (DAPI blue; Caspase1, green), and fluorescence intensity of Caspase1 were quantitively analyzed, scale bar: 50 μm, n = 5, non-significant (ns), ∗p < 0.05 ∗∗p < 0.01 ∗∗∗p < 0.001. (D–E) Flow cytometry analysis of cell death and pyroptosis rate comparison in PI and Caspase1-stained BV-2 cells across five conditions, n = 5, non-significant (ns), ∗∗p < 0.01 ∗∗∗∗p < 0.0001. (F and H) Representative immunofluorescence image of GSDMD staining (DAPI blue; GSDMD, green) and fluorescence intensity of GSDMD were quantitatively analyzed, scale bar: 50 μm, n = 5, non-significant (ns), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001. (G and I) Representative immunofluorescence image of DCFH-DA staining (DCFH-DA, green). Scale bar: 50 μm. Statistical analysis of fluorescence intensity of DCFH-DA, n = 5, non-significant (ns), ∗∗p < 0.01 ∗∗∗∗p < 0.0001. (J–U) The expression levels of HIF-1α, c-myc, NLRP3, GSDMD, GSDMD-N, Caspase 1, c-Caspase 1,p20, IL-1β, c- IL-1β and IL-18 in OGD/r BV-2 cells across four groups were determined by Western Blot, n = 5, non-significant (ns), ∗p < 0.05 ∗∗p < 0.01 ∗∗∗p < 0.001 ∗∗∗∗p < 0.0001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 9
Collectively, these results confirm that 3R@Lipo/Gink robustly inhibits pyroptotic progression in OGD/r-challenged BV-2 cells while ameliorating oxidative stress.
3R@Lipo/Gink improved microglia pyroptosis through inhibiting HIF-1α accumulation
2.10
To investigate whether the HIF-1α pathway mediates the anti-pyroptotic effects of 3R@Lipo/Gink on microglia, we utilized FG-4592, an inhibitor of hypoxia-inducible factor prolyl-hydroxylase (PHD), which prevents the hydroxylation and degradation of HIF-1α [32]. As shown in Fig. 10A, BV2 cells were divided into five groups, which extended the original four groups by adding an FG-4592 group. This additional group was pretreated with FG-4592 6 h prior to OGD/r, followed by 3R@Lipo/Gink treatment post-OGD/r. Western blot analysis of 11 pyroptosis-related proteins (HIF-1α, c-myc, NLRP3, GSDMD, GSDMD-N, Caspase-1, cleaved Caspase-1, p20, IL-1β, cleaved IL-1β, IL-18) demonstrated that FG-4592 administration markedly abolished the therapeutic benefits of 3R@Lipo/Gink (Fig. 10B and C). Immunofluorescence staining of Caspase-1 (Fig. 10D and E), GSDMD (Fig. 10F and G), and NLRP3 (Fig. S6) corroborated these findings: FG-4592 reversed 3R@Lipo/Gink-mediated pyroptosis suppression, with fluorescence intensity levels approximating those in the MCAO/r group. Reactive oxygen species (ROS) analysis revealed significantly lower oxidative stress only in the 3R@Lipo/Gink group compared to OGD/r, while OGD/r, 3R@Lipo, and FG-4592 groups exhibited similarly elevated ROS levels (Fig. 10H and I). Collectively, these data demonstrate that the therapeutic efficacy of 3R@Lipo/Gink against microglial pyroptosis was offset by the inhibition of HIF-1α reduction, confirming that its mechanism depends on the HIF-1α signaling axis.Fig. 103R@Lipo/Gink Improved Microglia Pyroptosis Through Inhibiting HIF-1α Accumulation(A) Schematic diagram of cell grouping and treatment. (B–C) The expression levels of HIF-1α, c-myc, NLRP3, GSDMD, GSDMD-N, Caspase 1, c-Caspase 1, p20, IL-1β, c-IL-1β and IL-18 in OGD/r BV-2 cells across five groups were determined by Western Blot, n = 5, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. (D–G) Representative immunofluorescence image of GSDMD and Caspase1 staining (Dapi, blue; GSDMD/Caspase1, green), and fluorescence intensity of GSDMD and Caspase1 were quantitively analyzed, scale bar: 50 μm, n = 5, non-significant (ns), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001. (H) Representative immunofluorescence image of DCFH-DA staining (DCFH-DA, green). Scale bar: 50 μm. (I) Statistical analysis of fluorescence intensity of DCFH-DA, n = 5, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 10
To model whether the treatment with 3R@Lipo/Gink exerted its effect by inhibiting HIF-1α, OGD/r microglia were treated with the HIF-1α inhibitor YC-1. Our results indicated that the HIF-1α inhibitor could significantly suppress microglial pyroptosis and proliferation; however, the overall effect was less stable compared to that of the 3R@Lipo/Gink treatment, and the relevant mechanism requires further investigation (Figs. S9–10).
Next, to demonstrate that the protective effect of 3R@Lipo/Gink on neuronal cells is primarily achieved through the reshaping of the inflammatory microenvironment by inhibiting microglial proliferation and pyroptosis through inhibiting HIF-1α, we utilized microglial conditioned medium to treat ischemic and hypoxic neuronal cells. The results indicated that the collected supernatant of BV-2 (OGD/r) cells treated with 3R@Lipo/Gink could significantly alleviate the apoptosis of HT-22 (OGD/r) cells, whereas the regulatory effect of the 3R@Lipo/Gink-treated BV-2 supernatant on the apoptosis of HT-22 (OGD/r) cells after stabilizing HIF-1α was diminished (Fig. S11).
Taken together, these data suggest that 3R@Lipo/Gink exerts neuroprotective effects mainly by inhibiting the increase of HIF-1α in microglia after ischemia-reperfusion, further inhibiting the pyroptosis and proliferation of microglia, and finally remodeling the inflammatory environment around neurons.
Discussion
3
Traditional drug therapies for ischemic stroke are often limited by the blood–brain barrier and insufficient drug efficacy, leading to suboptimal clinical outcomes. The emergence of nanomaterial-based drug delivery systems offers new opportunities to overcome these barriers and enhance therapeutic performance. Among these systems, liposomes—one of the earliest clinically approved nanocarriers—have been widely used for molecular encapsulation and delivery in various diseases due to their intrinsic biocompatibility, biodegradability, and structural resemblance to cellular membranes [33]. Their phospholipid bilayer allows efficient loading of molecules with diverse polarities, reduces systemic toxicity, and enables further chemical modification to achieve cell-, microenvironment-, and immune-targeted delivery^[^[33], [34], [35]^]^.
After cerebral ischemia, ROS levels rise rapidly, driven by mitochondrial dysfunction and NADPH oxidase activation, and further intensify during reperfusion. Sustained oxidative stress leads to lipid, protein, and DNA damage, triggers microglial activation, and promotes inflammation^[^[36], [37], [38], [39]^]^. This marked elevation of ROS provides a window for selective drug release. By introducing a TK-based ROS-responsive peptide [40], our liposomal system can sense the high-ROS microenvironment of ischemic tissue and release the encapsulated drug in a controlled manner. This enhances local drug accumulation, improves therapeutic efficacy, and reduces off-target effects.
Natural compounds have attracted increasing attention in recent years for their high safety, low toxicity, and multi-target pharmacological characteristics [41]. Ginkgo biloba has long been used to treat cerebrovascular diseases, including cerebral infarction [17], and its active biflavonoid component GK exhibits antioxidant and anti-inflammatory properties. Prior studies have shown that GK suppresses ROS production, enhances antioxidant enzyme activities, and reduces pro-inflammatory cytokine release in various disease models^[^[42], [43], [44]^]^. However, its low bioavailability and poor brain-targeting capacity limit its therapeutic potential in neurological disorders.
Compared with conventional nano-biomaterial strategies, the present study addresses two major translational challenges in ischemic stroke therapy. First, it overcomes the limited brain parenchymal penetration of traditional Chinese medicine monomers by integrating ROS-responsive release with microglia-targeting delivery. Second, in the past, there have been limited studies on the treatment of ischemic stroke with GK, and its mechanism has been scarcely explored. This study incorporated single-cell sequencing technology to comprehensively explore the potential mechanism of GK in treating ischemic stroke, namely, the regulation of microglial pyroptosis. Finally, ROS-responsive TK were utilized to precisely release GK to enhance its bio-efficacy. In this study, we discovered that, through inhibiting HIF-1α, 3R@Lipo/Gink alleviated c-myc-mediated excessive microglial proliferation and NLRP3-dependent pyroptosis. Via the aforementioned two intracellular mechanisms, 3R@Lipo/Gink suppresses the expression of inflammatory factors and reduces ROS release, remodels the microglia-mediated neuroinflammatory microenvironment, and ultimately enhances the neurological function of mice following ischemic stroke.
Mechanistically, we demonstrate that 3R@Lipo/Gink alleviates ischemic injury by inhibiting HIF-1α signaling, thereby suppressing c-Myc–mediated excessive microglial proliferation and NLRP3-dependent pyroptosis. Through these two interconnected intracellular pathways, 3R@Lipo/Gink reduces inflammatory cytokine production and ROS accumulation, reshapes the microglia-driven neuroinflammatory microenvironment, and ultimately improves neurological function following ischemic stroke.
Under hypoxic conditions, HIF-1α regulates multiple genes related to antioxidant defense, but its chronic or excessive activation generates pathological feedback loops. By upregulating glycolytic enzymes, HIF-1α drives metabolic reprogramming and accelerates glycolysis for ATP production [45,46]. This metabolic shift disrupts mitochondrial redox homeostasis, causing mtROS overload and amplifying oxidative stress cascades^[^[47], [48], [49]^]^. Elevated mtROS activates the NLRP3 inflammasome, promotes caspase-1 activation, and induces the maturation and release of IL-1β/IL-18, while caspase-1-mediated cleavage of GSDMD triggers pyroptotic cell death [50]. These processes collectively exacerbate neuroinflammation and lead to neuronal dysfunction.
Consequently, the design of ROS-responsive and microglia-targeting liposomal carriers, which encapsulate GK, plays a pivotal role in drug delivery for ischemic stroke. Experimental results demonstrate that our engineered system exhibits excellent biosafety and effectively penetrates the blood-brain barrier. Through integration of microglia-targeting ligands and ROS-responsive modifications, it achieves multifaceted advantages including high therapeutic efficiency, minimal/negligible side effects, enhanced targeting specificity, and reduced dosage requirements. However, while the in vivo imaging techniques employed in this experiment have preliminarily verified the accumulation of GK, it is still necessary to conduct further dynamic monitoring of GK in the brain using other quantitative methods, such as two-photon microscopy or other real-time monitoring techniques.
Critically, this study reveals that 3R@Lipo/Gink alleviates microglial pyroptosis and over-proliferation by suppressing HIF-1α expression. For validation of the mechanism, two HIF-1α regulators were used. FG-4592 inhibits prolyl hydroxylase and prevents HIF-1α degradation, leading to its stabilization, which reversed therapeutic effect of 3R@Lipo/Gink. In contrast, YC-1 suppresses HIF-1α accumulation and transcriptional activity. These regulatory effects were used to determine whether the therapeutic benefits were dependent on the HIF-1α pathway. Furthermore, single-cell sequencing analysis also suggests that 3R@Lipo/Gink also significantly downregulates myc gene expression in microglia, thereby inhibiting their proliferation and activiation-a mechanism that fundamentally disrupts the vicious cycle of neuroinflammation and pyroptosis. Notably, we conducted in vitro HIF-1α expression inhibition experiments to simulate the therapeutic effects of 3R@Lipo/Gink. The results further confirmed that HIF-1α is crucial for 3R@Lipo/Gink-mediated suppression of microglial proliferation. When HIF-1α was inhibited, the c-myc-mediated proliferation of microglia also decreased, which indirectly demonstrated the upstream-downstream crosstalk between HIF-1α and c-myc. The scRNA-seq analysis was not performed for the Sham group and the empty carrier (3R@Lipo) group. Future studies should involve more in-depth comparisons and explorations, and detailed investigations into the regulatory mechanisms at the pathway level are required for further clarification.
Furthermore, in-depth analysis based on single-cell sequencing data further elucidates the complete logical chain—from molecular inhibition to cellular phenotypic remodeling, and ultimately to neuronal protection. It is known that the core initiating event of 3R@Lipo/Gink treatment is the targeted inhibition of the HIF-1α-myc stress-pyroptosis axis in microglia. The downstream effects of HIF-1α and myc are manifested as a systematic remodeling of microglial subpopulation structures, accompanied by transcriptional reprogramming. Cells in the treatment group were significantly enriched in the C2 Birc5^+^ microglial subpopulation, which exhibited high proliferative activity. This proliferative phenotype is specifically driven by regulatory module M6, governed by the E2f transcription factor family—key regulators of the G1/S phase transition in the cell cycle [51,52]. The normal activation of E2f may represent a compensatory or reparative proliferative response following the inhibition of aberrant myc signaling, thereby guiding microglia into an orderly proliferative state to achieve functional repair. Cell interaction analysis revealed active MK signaling pathway communication between the C2 subpopulation and neurons, a pathway involved in tissue repair, cell survival, and anti-apoptosis processes [53,54]. Neurons highly express the ligand Mdk, while C2 microglia highly express the receptor Ncl, and this pathway is overall enhanced in the treatment group. Based on these findings, it is hypothesized that the reparative C2 Birc5^+^ microglia induced by 3R@Lipo/Gink, through enhanced responsiveness to neuron-derived Mdk signals, become activated and likely contribute to neuronal survival by modulating the inflammatory microenvironment or secreting neurotrophic factors in a feedback manner, ultimately establishing a supportive environment for neuronal viability in the ischemic injury context. Whether the specific mechanism of action operates in this manner remains to be experimentally validated.
However, this research still exhibits certain deficiencies. First, the molar ratios and conjugation coefficients of DSPE-TK-PEG-RVG29 and DSPE-TK-PEG-MG1, which make up 3R@Lipo/Gink, are not available. Subsequent studies should further investigate the physical properties of these two polymers to standardize the 3R@Lipo/Gink production process. Second, although the biosafety of this material was initially assessed, with no significant toxicity observed in major organs of healthy mice, its long-term toxicological profile and immunogenic potential warrant further investigation to enhance the rigor of preclinical evaluation. The metabolic processes in vivo and the ultimate excretion pathways of 3R@Lipo/Gink are also important to further application, which also require further investigation and clarification. Thirdly, we employed the HIF-1α inhibitor YC-1 to simulate the impact of 3R@Lipo/Gink on OGD/r microglia. The findings indicated that YC-1 was less effective than GK in regulating microglial proliferation and pyroptosis. This might suggest that the key targets of GK in the treatment of ischemic stroke are not limited to HIF-1α or microglia, which warrants further investigation in future studies.
This study provides foundational insights and methodological frameworks for applying bio-nanomaterials in ischemic stroke research. Future investigations could integrate this material system with supplementary pharmaceutical agents or non-invasive therapeutic strategies to advance both fundamental discoveries and clinical translation in ischemic stroke management.
Conclusion
4
We developed a ROS-responsive liposomal formulation, 3R@Lipo/Gink, to enhance ginkgetin delivery to ischemic brain tissue. The nanoformulation accumulated effectively in the lesion area, reduced infarct size, and improved neurological function in MCAO/r mice. Mechanistic studies showed that 3R@Lipo/Gink inhibited HIF-1α, thereby downregulating c-myc–mediated microglial proliferation and suppressing NLRP3-related pyroptosis. These effects were reversed by FG-4592, supporting the role of the HIF-1α pathway. By modulating microglial activation and improving the neuronal inflammatory microenvironment, 3R@Lipo/Gink offers a promising targeted therapeutic strategy for ischemic stroke.
Methods
5
Preparation of 3R@Lipo/Gink
5.1
Take SPC, cholesterol, DSPE-TK-PEG-RVG 29, DSPE-TK-PEG-MG 1, and CY5.5, and dissolve them in 4 mL of chloroform. Dissolve ginkgetin in methanol. Mix the solutions and evaporate the solvent under reduced pressure in a sample vial to form a thin film. Add deionized water for hydration. Treat the mixture with ultrasonication and pass it through a liposome extruder (equipped with a polycarbonate membrane, pore size 100 nm). Dialyze the resulting product using a nanodialysis device (polycarbonate membrane, pore size 10 nm) and adjust the volume to 5 mL with deionized water. Add lyoprotectants and perform freeze-drying. The particle size and zeta potential of the liposomes were measured using a nanoparticle size and zeta potential analyzer (Brookhaven, USA). Their morphology was observed by transmission electron microscopy (TEM) (FEI, USA). The drug loading capacity (LC) and encapsulation efficiency (EE) were determined using a high-performance liquid chromatography (HPLC) system (Elite, China). Absorbance was measured with a UV–Vis spectrophotometer (Beijing Puxi, China).
In vitro release of ginkgetin
5.2
The in vitro release assay was carried out in a sterile incubator at 37 °C with a total volume of 10 ml for each sample, and detection was carried out without stirring. The concentration of ginkgetin was determined using UV spectrophotometry. At designated time points, the liposome dispersion in the dialysis device was collected. The liposome dispersion was treated with Triton X-100 to disrupt the lipid bilayer structure and release the encapsulated drug. The aqueous phase was then sampled, and its absorbance was measured at the maximum absorption wavelength using a UV spectrophotometer. UV–Vis wavelength range was 200–400 nm. The maximum absorption wavelength of ginkgetin is approximately 260 nm. Drug concentrations in the samples were calculated based on a standard curve derived from drug reference standards. The standard curve range was 1–50 μg/mL.The cumulative release profile of the drug was obtained by measuring its concentration at different time points.
In vivo biodistribution of 3R@Lipo/Gink
5.3
For biodistribution analysis, Cy5.5-labeled 3R@Lipo/Gink or free Cy5.5 was administered to MCAO/r mice via tail vein injection on the 24 h after MCAO/r surgery. Fluorescence signals were monitored at predetermined time points after the first injection, using the IVIS Lumina III Imaging System (Caliper, Perkin Elmer, USA) to evaluate Cy5.5 distribution in the brain. After in vivo imaging, mice were euthanized, and brains were collected for ex vivo fluorescence imaging. Immunofluorescence staining was further performed to assess Cy5.5 localization within brain tissues.
Cell culture
5.4
Mouse BV2 microglial cells were purchased from the National Certified Cell Bank (Shanghai, China). The cells were cultured in fresh complete medium [Dulbecco's Modified Eagle Medium (DMEM, Cat#: C11995500BT, Thermo Fisher Scientific) supplemented with 10% (v/v) fetal bovine serum (FBS, Cat#: VS500T, Vian-Saga) and 1% (v/v) penicillin-streptomycin (100 × , Cat#: SV30010, HyClone)], and maintained in a humidified incubator at 37 °C with 5% CO_2_.
OGD/r cell model and treatment
5.5
Normally cultured BV2 cells were transferred to an anaerobic chamber filled with a pre-mixed gas (95% N_2_ and 5% CO_2_, 37 °C) to establish OGD conditions. After 4 h of OGD treatment, the cells were transferred back to conventional DMEM medium containing glucose and 10% FBS, oxygen supply was restored, and different treatments were administered as follows: the OGD/r group received PBS only, the 3R@Lipo/Gink group was administered 3R@Lipo/Gink, and the 3R@Lipo group was given 3R@Lipo with both treatments at a concentration of 200 μg/mL. Pretreatment with FG-4592 and YC-1 was administered prior to modeling using Roxadustat (FG-4592; 50 μM, Cat#: HY-13426, MedChemExpress) or lificiguat (YC-1; 10 μmol/L; MCE, Shanghai, China) [55], respectively. Meanwhile, the Control group without OGD treatment was administered the same dose of PBS. The cells were then cultured in a normal incubator for 24 h to construct the OGD/r model.
Cell viability
5.6
Cell viability was assessed using the Cell Counting Kit-8 (CCK-8) assay. After 24-h reperfusion or before 6 h of OGD/r treatment, BV2 cells were treated with different concentrations of experimental agents (3R@Lipo/Gink or FG-4592) or PBS, respectively. Subsequently, 10% CCK-8 solution was added to each group and incubated for 1 h. Absorbance was measured to determine relative cell viability, with untreated cells serving as controls (Figs. S7 and S8).
DCFH-DA assay
5.7
Intracellular ROS levels were measured using the fluorescent probe 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA, Beyotime, China). Briefly, cells were washed three times with PBS and incubated with 10 μM DCFH-DA in a 37 °C incubator for 20 min. After washing three times with PBS again, the cells were fixed with 4% paraformaldehyde at room temperature for 30 min. Finally, cellular fluorescence was observed and recorded under a fluorescence microscope.
Flow cytometric analysis
5.8
The proportion of pyroptotic cells in each treatment group was measured by flow cytometry. Briefly, active caspase-1 was detected using the FLICA 660 Caspase-1 Assay Kit (ImmunoChemistry, Davis, CA, USA). BV2 cells were cultured in 6-well plates and treated as specified. Subsequently, cells were harvested and incubated with FLICA 660 working solution (diluted 1:50, v/v) for 45 min at 37 °C. After incubation, the cells were washed, centrifuged, resuspended, and stained with PI solution for 5 min. Finally, samples were analyzed using a FACSCalibur flow cytometer (Becton Dickinson and Company, USA).
Experimental animals
5.9
The experiment utilized adult male C57BL/6J mice weighing 24–26 g, all of which were purchased from Liaoning Changsheng Biotechnology Co., Ltd. (Liaoning, China). All experimental protocols were approved by the Institutional Animal Care and Use Committee at Zhengzhou University (Protocol ZZU-LAC20240906 [23]). Mice were housed under specific pathogen-free conditions at room temperature with a 12-h light-dark cycle and provided ad libitum access to food and water.
MCAO/r mouse model and treatment
5.10
Mice were first anesthetized via intraperitoneal injection of 2% sodium pentobarbital. After anesthesia, the mice were fixed in a supine position. A suture thread was hooked to the incisors, fully exposing the surgical area. To maintain physiological body temperature, a heating pad was used to stabilize the mouse's temperature at 37 °C. A midline incision was made along the neck to directly expose and separate the left common carotid artery, external carotid artery, and internal carotid artery. A filament was inserted through the internal carotid artery stump and advanced approximately 10 mm until slight resistance was encountered, thereby occluding the left middle cerebral artery. After the occlusion lasted for 1 h, the filament was removed to restore blood flow and achieve reperfusion. The Sham group underwent only skin and subcutaneous tissue incision and received, without vascular occlusion or other injuries. Four groups of mice received different treatments: 3R@Lipo/Gink group and 3R@Lipo group were injected with 15 mg/kg of 3R@Lipo/Gink and 3R@Lipo via the tail vein every day from 1 to 7 days after MCAO/r. The actual GK injection concentration for each mouse was 1.14 mg/kg, calculated by LC/EE; MCAO/r group and Sham group received the same dose of normal saline. Following the completion of the behavioral tests on the 7th day, all mice were euthanized, and brain samples were obtained. Brain tissues were collected from the ischemic penumbra of the mice for single-cell sequencing, molecular biological analysis, and histological analysis.
TTC staining
5.11
To assess cell viability and evaluate tissue damage, TTC staining was performed on day 3 and 7 post-surgery. After brain removal, the tissue was rapidly sliced into 1 mm-thick coronal sections along the coronal plane from anterior to posterior. The sections were immersed in 1% (w/v) TTC staining solution and incubated at 37 °C for approximately 20 min until the infarcted brain tissue appeared white while normal brain tissue remained red. The stained sections were photographed on a black background plate, and the infarct area was quantified using ImageJ software.
TUNEL staining
5.12
TUNEL staining was employed to detect neuronal death in the cortical region of the ischemic penumbra. Specifically, after deparaffinization and rehydration, paraffin-embedded brain tissue sections were incubated with Proteinase K solution at 37 °C for 25 min for antigen retrieval. Subsequently, the sections were stained with TUNEL reaction solution at 37 °C for 2 h, according to the manufacturer's instructions. Next, the nuclei were labeled by incubating with DAPI in the dark at 25 °C for 10 min. Fluorescence microscopy was used to capture the stained images. Following this, cells in the ischemic penumbra cortical region of each group were counted to determine the number of positive-labeled cells.
Nissl staining
5.13
On day 7 after surgery, the mice were executed, followed by cardiac perfusion, and the brains were first rinsed with saline and then fixed with 4% paraformaldehyde solution. Subsequently, the brain samples were dehydrated by incremental concentrations of alcohol and embedded in paraffin. 10 μm micrometer-thick serial sections were cut from the paraffin blocks and processed for Nissl staining. The sections were observed and photographed with an Olympus microscope.
Hematoxylin and eosin staining
5.14
Mice in both groups were euthanized on postoperative day 7. Subsequently, the heart, liver, spleen, lungs, kidneys, and brain were harvested and fixed in 4% paraformaldehyde. Tissue sections with a thickness of 10 μm were embedded in paraffin, dewaxed, and dehydrated through a graded series, stained with hematoxylin and eosin, and finally examined under the Olympus microscope (Fig. S1).
Transmission electron microscope (TEM)
5.15
After anesthesia perfusion of mice in each group, the ischemic penumbra of the left cerebral hemisphere of the mice was immediately resected and fixed overnight (4 °C). Then transfer the sample to 1% OsO4 for 2 h. Then, wash three times with carboxylate buffer, stain in 1% uranyl acetate for 1 h, followed by dehydration and embedding in Epon. The slices are cut on a microslicer (Leica) and placed on an EM mesh. To enhance the contrast, the sections were double-stained with uranium acetate and lead citrate. The sections were observed and examined under a Hitachi 7100 transmission electron microscope (Nikon, Japan).
Measurement of T-AOC, MDA, and GSH
5.16
An appropriate amount of ischemic brain tissue was added and homogenized using a tissue homogenizer (Servicebio, China) and centrifuged at 12,000×g for 15 min at 4 °C to obtain the homogenate. Total antioxidant capacity (T-AOC) was measured using a FRAP colorimetric assay kit (A015-3-1, Nanjing Jiancheng Bioengineering Institute, China). Malondialdehyde (MDA) levels were determined with a TBA assay kit (A003-1-2, Nanjing Jiancheng Bioengineering Institute, China), and glutathione peroxidase (GSH-Px) activity was quantified using a specific kit (A005-1-2, Nanjing Jiancheng Bioengineering Institute, China). Assays were performed according to the protocols provided by the manufacturer, and results were analyzed to evaluate oxidative stress and antioxidant capacity.
Western blot
5.17
The isolated brain tissues were homogenized in RIPA lysis buffer using a tissue homogenizer. The homogenate was centrifuged at 12,000 rpm for 10 min at 4 °C to collect the supernatant. Protein concentration was quantified using a commercial BCA protein assay kit (Beyotime Biotech, Shanghai, China) following the manufacturer's protocol. Denatured protein lysates (30 μg) were then electrophoresed on 10% SDS-PAGE gels (Epizyme Biotech, China) at a constant voltage of 120 V. The separated proteins were transferred to PVDF membranes (Millipore, USA) at a constant current of 400 mA for 1 h with ice water assistance. The membranes were blocked with 5% BSA at room temperature for 1 h, and then, the membranes were incubated with corresponding primary antibodies (Table S1) overnight at 4 °C. The following day, after three washes with TBST, the membranes were incubated with secondary antibodies: goat anti-mouse or anti-rabbit IgG horseradish peroxidase-conjugated antibody (1:5000, Yesen, China) at room temperature for 1 h. Protein blot images of each antibody were analyzed using an image analysis program (ImageJ, NIH, Bethesda, USA) to quantify protein expression based on relative image density.
Immunofluorescence
5.18
After anesthesia, mice were perfused with ice-cold normal saline via the heart. The brain tissues were quickly removed and fixed overnight at 4 °C with 4% paraformaldehyde. Dehydrate the tissue and embed it in paraffin. The embedded brain tissue was sectioned using a sectioning machine, and then these sections were dewaxed and hydrated. Sections were subjected to antigen repair in citrate buffer. After washing in PBS, permeabilize sections with 5% BSA and 0.5% Triton-X-100 and incubate with 5% normal goat serum for 1 h at room temperature. Meanwhile, BV2 cells were incubated with 4% paraformaldehyde for 1 h, washed with PBS, and then incubated with 0.15% Triton X-100 for 15 min to permeabilize the cell membrane and blocked with 5% TBST solution of skim milk for 1 h. They were then incubated with primary antibody (Table S1) overnight at 4 °C. The next day, wash the sections and incubate with the corresponding secondary antibody for 1 h at room temperature. Immunoreactivity was observed and photographed using the Olympus fluorescence microscope.
Experiments with BV-2 conditioned medium
5.19
Following the aforementioned cell modeling and treatments, the supernatant from each group of BV-2 cells was collected as BV-2 conditioned medium (BCM) and co-cultured with HT-22 cells for 24 h under the following conditions: Group 1 involved normal HT-22 cells co-cultured with control-BCM; Group 2 consisted of OGD/r HT-22 cells co-cultured with control-BCM; Group 3 included OGD/r HT-22 cells co-cultured with OGD/r-BCM; Group 4 featured OGD/r HT-22 cells co-cultured with OGD/r-3R@Lipo/Gink-BCM; Group 5 comprised OGD/r HT-22 cells co-cultured with FG-4592-OGD/r-3R@Lipo/Gink-BCM; and Group 6 contained OGD/r HT-22 cells co-cultured with YC-1-OGD/r-3R@Lipo/Gink-BCM. After 24 h of reoxygenation, the HT-22 cells were observed and analyzed.
Elisa
5.20
The concentrations of IL-1β and IL-18 in the culture medium of BV2 cells after OGD/r and treatment were measured. A standardized enzyme-linked immunosorbent assay (ELISA) was performed using the Mouse IL-1β ELISA Kit (PI301, Beyotime) and the Mouse IL-18 ELISA Kit (PI553, Beyotime) according to the manufacturer's instructions.
Cell proliferation and apoptosis
5.21
According to the manufacturer's instructions, cell proliferation and apoptosis were detected using the Cell Cycle and Apoptosis Analysis Kit (C1052, Beyotime) and the Cell Proliferation Kit (C0075L, Beyotime).
Neurobehavioral assessment
5.22
The body weight of the mice was measured every day throughout the experiment.
The neurological functional impairment in mice was systematically assessed using the modified Neurological Severity Score (mNSS) within 1–7 days after MCAO/r surgery. The mNSS system evaluates three domains: motor function, reflexes, and balance tests. One point was assigned for each uncompleted test item or absent response. Neurological impairment severity was determined by the cumulative score.
Behavioral tests
5.23
Cylinder test
5.23.1
The cylinder test was performed to evaluate forelimb motor function in mice. Mice were placed in a transparent beaker-shaped cylinder (8 cm in diameter, 18 cm in height) to observe their forelimb preference during spontaneous exploration. Motor asymmetry and recovery were analyzed by quantifying the frequency of wall contacts made by the impaired forelimb, unimpaired forelimb, or both forelimbs.
Open field test
5.23.2
The open field test (OFT) was used to assess locomotor ability and exploratory behavior. In OFT, mice were acclimatized to the environment for 1 h before testing. The room was kept at constant temperature and humidity, light and dark, and quiet. The test room was a plastic box divided into central and peripheral zones. The mice were placed in the central zone of the OFT apparatus and allowed to explore freely for 5 min, and video tracking software (SMART v.3.0 software, RWD Life Science) was used to record their move distance and distance in center.
CatWalk XT gait test
5.23.3
The CatWalk XT system (Noldus, Netherlands), a 150-cm-long small animal glass walkway, was adopted. The feature of the experimental device is that the high-speed camera installed under the glass plate can capture the changes of mouse paw prints in real time to obtain gait data. During the test, the mice were placed at the starting point of the track and allowed to cross freely. Each cross was required to be completed within 8 s, with the maximum speed variation not exceeding 60%. Record at least five average exercise values that meet the above standards for each experiment and test each mouse three times.
Rotarod test
5.23.4
In this experiment, a rotating rod (Model LE8500, Panlab SL) was used to evaluate the motor coordination of mice. During the test, the mice were asked to keep walking to prevent falling. Mice received pre-adaptation training twice a day for three consecutive days. The formal test began on the 7th day after the operation. The mice were placed on the rotating rods. Within 5 min, the rotational speed gradually increased from 4 revolutions per minute to 40 revolutions per minute. Each mouse was tested three times, and the average value of the three test results was taken for statistical analysis.
Single-cell RNA sequencing (scRNA-seq)
5.24
Processing of scRNA-seq data
5.24.1
We utilized the Seurat package (v4.3.0) to load the 10x Genomics data for each sample into R software (v4.3.0). Initially, potential doublets and low-quality cells were filtered using the DoubletFinder program (v2.0.3). Cells were retained for further analysis if they met the following criteria: 500 < nFeature-RNA <6000 and mitochondrial gene expression <25% of total gene expression. Subsequently, we performed cell type identification: the selected samples were normalized, and the top 2000 highly variable genes (HVGs) were identified from the normalized expression matrix. These genes were then normalized again and subjected to principal component analysis (PCA). Harmony (v1.2.0) was used to reduce the dimensionality of the top 30 principal components (PCs) and to cluster the cells, correcting for batch effects between samples. The results were then projected into a 2D UMAP plot to visualize distinct cell clusters. To annotate the cell types, we referenced the cell marker database (http://xteam.xbio.top/CellMarker/) to assign appropriate markers to the identified clusters, enabling the identification of different cell types and their distribution and proportion. Finally, to investigate the heterogeneity of neurons and microglia, we reclustered these two cell types separately and marked each subtype based on their unique marker genes.
Visualization of differentially expressed genes and AUCell-Based enrichment analysis
5.24.2
DEGs for each cell population and subtype were identified using the Wilcoxon rank-sum test and the default parameters of the 'FindAllMarkers' function (Log FC > 0.25). Enrichment analysis was then performed using the clusterProfiler (v4.6.2) and SCP (v0.4.8) packages to further explore the functional roles of DEGs in each cell population and subtype. Enriched pathways for each cell group and subtype were derived from KEGG, GO-BP, and GSEA databases. Additionally, we utilized AUCell to identify active genes and transcription factors in the scRNA-seq data.
Construction of pseudotime trajectories for neuronal and microglial subtypes
5.24.3
We used the Monocle package (v2.24.0) to analyze the pseudotime trajectories of neurons and microglia represented as individual cells. The pseudotime trajectory map was constructed based on scRNA-seq data. Monocle is known for its ability to identify cellular changes occurring during the differentiation process of neuronal and microglial subtypes in ischemic stroke mice.
CytoTRACE and Slingshot analysis
5.24.4
To investigate the developmental and differentiation state differences between neuronal and microglial subtypes, we first performed CytoTRACE analysis to rank the differentiation status of all neuronal and microglial subtypes. Additionally, the “getlineage” and “getCurves” functions were utilized to infer the differentiation trajectories of each neuronal and microglial subtype and to evaluate the changes in gene expression levels as a function of time. Subsequently, we employed the “Slingshot” package (v2.6.0) to construct trajectories that reflect the developmental stages and states of each subtype.
CytoTRACE2 analysis
5.24.5
We further utilized CytoTRACE2 to analyze the scRNA-seq data, predicting the potential categories and absolute developmental potential of neuronal and microglial subtypes derived from the MCAO/r and 3R@Lipo/Gink groups. The potential categories in CytoTRACE2 classify cells based on their developmental potential, and the predicted potential values provide a continuous measure of developmental potential ranging from 0 (differentiation) to 1 (pluripotency). This allows for a direct cross-dataset comparison of developmental potential in absolute spatial terms.
Cell communication
5.24.6
We predicted the potential interactions between different cell types (including various neuronal subtypes or microglial subtypes with other cells) based on scRNA-seq data using the CellChat package (v1.6.1). More importantly, in this study, we conducted a detailed visualization of the interactions between key neuronal subtypes and microglia, as well as between key microglial subtypes and neurons, to identify their critical signaling pathways.
SCENIC analysis
5.24.7
We performed SCENIC analysis, a tool designed to reconstruct gene regulatory networks from scRNA-seq data and identify stable cellular states. For this, we used the SCENIC (v1.3.1) package in Python (v3.7) with its default parameters, generating an AUCell matrix to assess the enrichment of transcription factors and the activity of regulatory factors.
Statistical analysis
5.25
Statistical analysis was performed with GraphPad Prism 9.0 and SPSS 23.0 statistical software. Significance was typically analyzed by Student's t-test, one-way ANOVA followed by post hoc LSD test, and two-way ANOVA followed by multiple t-tests. Statistical significance was defined as p < 0.05. Behavioral assessments were performed three times on the same cohort with sufficient rest between trials and averaged as one biological sample. For cell and brain histological analyses, three random visual fields were applied and averaged as one sample. For molecular biological analyses, three random tissue pieces from the peri-infarct area of each mouse were pooled as one sample. 8 samples were included in behavioral test, and 5 samples were included in histological analyses and molecular biological analyses to prevent false outcomes.
CRediT authorship contribution statement
Xueyuan Li: Data curation, Formal analysis, Investigation. Zi Ye: Data curation, Formal analysis, Investigation. Wenyang Nie: Data curation, Formal analysis, Investigation. Bao Zhou: Data curation, Formal analysis, Investigation. Haocheng Qin: Data curation, Formal analysis, Investigation. Zhijie Zhao: Investigation. Yilong Fu: Investigation. Dun Liu: Investigation. Shaowei Zheng: Investigation. Liangyu Wang: Investigation. Jun Ma: Investigation. Jingying Guo: Investigation. Beibei Nie: Investigation. Yan Lu: Investigation. Dongming Yan: Investigation. Zhiwen Luo: Conceptualization, Funding acquisition, Investigation. Qingshan Wang: Conceptualization, Funding acquisition, Investigation. Meng Bian: Conceptualization, Funding acquisition, Investigation. Hui Jiang: Conceptualization, Funding acquisition, Investigation. Di Chen: Conceptualization, Funding acquisition, Investigation.
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.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Fan J.Li X.Yu X.Liu Z.Jiang Y.Fang Y.Zong M.Suo C.Man Q.Xiong L.Neurology 1012023 e 13710.1212/WNL.0000000000207387 PMC 1035154637197995 · doi ↗ · pubmed ↗
- 2Pu L.Wang L.Zhang R.Zhao T.Jiang Y.Han L.Stroke 54202313303709403410.1161/STROKEAHA.122.040073 · doi ↗ · pubmed ↗
- 3Shehjar F.Maktabi B.Rahman Z.Bahader G.Antonisamy W.J.Naqvi A.Mahajan R.Shah Z.A.Neurochem. Int.162202310545810.1016/j.neuint.2022.105458 PMC 983965936460240 · doi ↗ · pubmed ↗
- 4Hilkens N.A.Casolla B.Leung T.W.de Leeuw F.-E.Lancet 403202428203875966410.1016/S 0140-6736(24)00642-1 · doi ↗ · pubmed ↗
- 5PrzykazaŁ.Front Immunol 12202178256910.3389/fimmu.2021.782569 PMC 863433634868060 · doi ↗ · pubmed ↗
- 6C. Iadecola, M. S. Buckwalter, J. Anrather, J Clin Invest, 130, 2777.10.1172/JCI 135530 PMC 726002932391806 · doi ↗ · pubmed ↗
- 7Long J.Sun Y.Liu S.Yang S.Chen C.Zhang Z.Chu S.Yang Y.Pei G.Lin M.Yan Q.Yao J.Lin Y.Yi F.Meng L.Tan Y.Ai Q.Chen N.Cell Death Discov.920231553716500510.1038/s 41420-023-01440-y PMC 10172388 · doi ↗ · pubmed ↗
- 8Vasudevan S.O.Behl B.Rathinam V.A.Semin Immunol 69202310178110.1016/j.smim.2023.101781 PMC 1059875937352727 · doi ↗ · pubmed ↗