Berberine targets the PI3Kα LYS672 site to suppress the PI3K/Akt/NF-κB axis and modulate microglial functional states after intracerebral hemorrhage

Abstract

Microglia-driven neuroinflammation contributes to secondary injury after intracerebral hemorrhage (ICH). Modulating microglial functional states represents a promising therapeutic strategy. Berberine (BBR) has neuroprotective effects, but its role in ICH remains unclear. This study investigated whether BBR alleviates ICH-induced neuroinflammation by modulating microglial responses after ICH via the phosphatidylinositol 3-kinase (PI3K)/protein kinase B (Akt)/nuclear factor kappa-B (NF-κB) axis. Network pharmacology predicted BBR targets. An ICH mouse model was used to assess BBR's effects on hematoma volume, neurological function and neuroinflammation. Transcriptomics identified differentially expressed genes and enriched pathways. Immunofluorescence assessed glial activation, cell death and extracellular matrix remodeling. In lipopolysaccharide-stimulated BV2 cell, microglial functional states and PI3K/Akt/NF-κB signaling were evaluated by Western blotting and immunofluorescence, with the mechanism clarified via molecular docking, inhibitors, and an activator. Network pharmacology identified 208 common targets enriched in inflammatory and extracellular matrix pathways. In vivo, BBR reduced hematoma volume, improved neurological outcomes, attenuated neuronal damage, and suppressed glial activation and cell death. Transcriptomics revealed BBR reversed ICH-induced inflammatory and extracellular matrix-related gene expression. In vitro, BBR shifted microglia from pro-inflammatory to anti-inflammatory phenotype and rebalanced cytokine expression. Molecular docking identified PI3K as a primary target, and Western blotting confirmed BBR suppressed PI3K/Akt/NF-κB signaling, validated by inhibitors. Notably, the PI3K agonist failed to reverse BBR's inhibition, reinforcing PI3K as a key functional target. BBR targets the PI3Kα LYS672 site to inhibit downstream NF-κB, thereby regulating microglial functional states and alleviating post-ICH neuroinflammation. These findings support BBR as a potential ICH therapeutic.

Graphical abstract

BBR binds to the LYS672 site of PIK3CA, thereby inhibiting PI3K catalytic activity and subsequently blocking Akt phosphorylation. The resulting decrease in Akt activity then suppresses downstream NF-κB signaling and modulates STAT3 activity through pathway crosstalk. This cascade leads to the coordinated downregulation of pro-inflammatory associated genes (IL-1β, TNF-α, IL-6 and iNOS) and promotes repair-associated microglial states, ultimately alleviating post-ICH neuroinflammation and conferring neuroprotection.

Introduction

Intracerebral hemorrhage (ICH) is an acute cerebrovascular disease characterized by high disability and mortality rates, contributing disproportionately to stroke-related disability and death despite its relatively low incidence [1,2]. Recovery of neurological function remains extremely limited even with advances in hematoma evacuation and supportive care [3], largely due to secondary brain injury by excessive microglial activation in the regions surrounding the hematoma [4,5]. Despite extensive understanding of the pathophysiological mechanisms, specific therapeutic interventions targeting this cascade remain lacking [6]. Therefore, modulating neuroinflammation represents a potentially effective approach to improving neurological outcomes following ICH.

Neuroinflammation is a hallmark of secondary brain injury following ICH and is increasingly recognized as a key determinant of neurological outcomes [7,8]. The intensity and duration of neuroinflammation directly affect patient prognosis, with short-term exacerbation of brain edema and long-term contribution to cognitive and motor deficits [9,10]. Thus, Precise regulation of the neuroinflammatory response represents a critical therapeutic approach to improving post-ICH neurological outcomes. The progression and resolution of neuroinflammation are largely shaped by the functional states of resident microglia, which are the primary responders to injury signals in the brain [[11], [12], [13]]. Understanding the heterogeneity of microglial responses, rather than viewing them as a uniform population, is essential for developing effective therapeutic strategies.

Microglia are highly plastic cells that adopt diverse phenotypic states depending on the stage of injury and local microenvironmental cues [14,15]. Depending on differences in microenvironmental signals, activated microglia can adopt diverse phenotypic states, ranging from pro-inflammatory to repair-associated profiles, rather than adhering strictly to a binary M1/M2 classification [11,16,17]. Recent single-cell transcriptomic studies have revealed that microglia in the ICH brain comprise multiple functionally distinct subpopulations, each associated with specific tasks such as sensing damage, amplifying inflammation, phagocytosing debris or promoting tissue repair [[18], [19], [20]]. These subpopulations are not static but dynamically shift between pro-inflammatory and repair-associated phenotypes across different phases of ICH, allowing microglia to coordinate both the onset and resolution of neuroinflammation [21]. Such functional diversity underscores the importance of targeting microglial subpopulations in a context-dependent manner, rather than relying solely on the classical M1/M2 polarization framework [22]. Modulating the balance of these functional states may therefore offer a more precise therapeutic strategy for limiting secondary brain injury and improving recovery after ICH.

Berberine (BBR), an extract from the traditional Chinese herb Coptis chinensis, has long been used for its anti-inflammatory properties [23,24]. BBR has recently attracted interest for its neuroprotective effects in ischemic stroke and Alzheimer's disease models [[25], [26], [27]]. BBR also modulates neuroinflammation through complex multi-target mechanisms, including indirect regulation via microbiota-gut-brain axis [28]. Despite the recognized neuroprotective and anti-inflammatory properties of BBR, its role in ICH and the molecular mechanisms by which it regulates microglial functional states remain largely unknown.

This study integrates network pharmacology, molecular docking, and multi-level experimental validation to demonstrate that BBR regulates microglial functional states via the Phosphatidylinositol 3-kinase/Protein Kinase B/Nuclear factor kappa B (PI3K/Akt/NF-κB) signaling axis, providing a theoretical basis for its application in ICH treatment.

Materials and methods

Network pharmacology

Using the canonical SMILES and 2D structures of the drug, we predicted the targets of BBR in the SwissTargetPredictions and PharmMapper databases, respectively. We then combined these results with data from GeneCards and OMIM to identify disease targets for ICH. The intersection between ICH targets and BBR targets was determined using Venny 2.1.0, identifying potential BBR targets for ICH therapy. Protein-protein interaction (PPI) network was constructed using the STRING database and visualized with Cytoscape 3.9.1. Nodal topological features, the degree centrality (DC), betweenness centrality (BC), and closeness centrality (CC) were calculated using the CytoNCA plugin within the constructed PPI network. Targets with DC, BC, and CC values all above the median were defined as core targets. Subsequent Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were carried out using R packages “ggplot2”, “clusterProfiler”, “enrichplot” and “org.Hs.eg.db”, with data visualization via “ggplot2”.

Chemical reagents

Detailed information on all chemical reagents is provided in Table 1.

Table 1.

List of chemical reagents.

Reagent	Company	Cat. Ref.
Berberine	Macklin (Shanghai, China)	B875003
Delivector Avertin (mice)	Dowobio (Shanghai, China)	DW3100
Collagenase type VII	Sigma (St Louis, MO, USA)	C0773
Isoflurane	RWD (Shenzhen, China)
RIPA lysis buffer	Beyotime (Shanghai, China)	P0013B
Protease inhibitor cocktail	Beyotime (Shanghai, China)	P1045-2
Phosphatase inhibitor cocktail	Beyotime (Shanghai, China)	P1045-1
BCA protein assay kit	GlpBio (CA, USA)	GK10009
0.45 μm Polyvinylidene Fluoride	Millipore (MA, USA)
Ultrahigh sensitivity ECL kit	GlpBio (CA, USA)	GK10008
4 % paraformaldehyde	Servicebio (Wuhan, China)	G1101
Triton X-100	Beyotime (Shanghai, China)	P0096
Antigen retrieval buffer	Servicebio (Wuhan, China)	G1202
DAPI medium	Solarbio (Beijing, China)	S2110
TUNEL assay kit	Vazyme (Shanghai, China)	A112-03
Trizol reagent	Thermo Fisher Scientific (DE, USA)	15596026
Total RNA Extraction kit	BioFlux (Beijing, China)	BSC52M1
PrimeScript RT master Mix	TaKaRa (Kyoto, Japan)	RR036A
SYBR green master Mix	TaKaRa (Kyoto, Japan)	RR820A
Dulbecco's modified Eagle's medium	Procell (Wuhan, China)	PM150210
Fetal bovine serum	Procell (Wuhan, China)	164210
CCK-8 reagent	GlpBio (CA, USA)	GK10001

Ethics statement

C57BL/6 J mice (6–8 weeks old, 20–25 g) were used from Beijing Huafukang Biotechnology Co., Ltd. Five mice were housed per cage under a 12 h light-dark cycle with free access to food and water in a temperature and humidity-controlled environment. Health status was monitored daily, and no unexpected adverse events occurred. Group assignment was performed using a random number generator. All animal experiments were conducted in accordance with the ARRIVE guidelines and approved by the Ethics Committee of Zhengzhou University (Approval No. 2022013).

ICH mouse models

We induced an ICH model via collagenase injection as previously prescribed [29]. Briefly mice were anesthetized intraperitoneally with tribromoethanol (18 μL/g). After shaving the fur and disinfecting the scalp with povidone-iodine, a midline incision was performed. The subcutaneous tissue was dissected to expose the skull, clearly revealing the anterior and posterior fontanelles. The mouse was then secured in a stereotaxic apparatus (RWD, Shenzhen, China) and a 0.6 mm burr hole was drilled at the right striatum (0.2 mm posterior to bregma, 2.0 mm lateral to the midline, 3.5 mm deep). A total of 0.75 μL of type VII collagenase (0.1 U/μL) was slowly injected at a rate of 0.1 μL/min using a micropipette. After injection, the needle was left in place for 5 min to prevent backflow, and the wound was then sealed with biological tissue glue. Mice were kept on a heating pad to maintain normothermia during the recovery period and then returned to clean cages after awakening. For the sham group, the same procedure was performed, but an equal volume of saline was injected instead of collagenase.

Animal experiment design

BBR was dissolved in a vehicle consisting of 5% DMSO, 40% PEG 200, 5% Tween 80, and 50% saline, and administered orally at a dose of 100 mg/kg/day based on a previous study [30,31]. One week after acclimatization, mice were randomly divided into the following three experimental groups: (1) Sham group: sham surgery + oral gavage of saline; (2) ICH group: ICH model + oral gavage of solvent; (3) ICH + BBR group: ICH model + oral gavage of BBR. BBR administration commenced 2 h post-ICH and was given once daily for 3 or 7 days, depending on the specific outcome measures. A total of 104 mice were used. Animal allocation was as follows: MRI, HE staining, and Nissl staining: 4 mice per group (12 mice total). Neurological function assessment: 6 mice per group (54 mice total), and a subset of these animals was subsequently used for Western blot analysis (6 mice per group, 18 mice total). Transcriptomic sequencing: 3 mice per group (9 mice total). Iba1/MPO/GFAP/TUNEL staining: 4 mice per group (12 mice total). MMP9 and laminin staining: 4 mice per group (12 mice total). Five additional mice were excluded due to surgical failure or absence of hematoma formation.

Behavioral assessment

Neurological disorder assessment

Neurological function was evaluated on day(s) 1, 3, and 7 post-surgery using the modified neurological severity score (mNSS, range 0–18). Each mouse was assessed independently by two experienced investigators blinded to the experimental groups, and the average of their scores was used for statistical analysis. The scoring system included spontaneous activity, limb symmetry, forelimb extension, climbing ability, proprioception, and balance. A higher score indicates a more severe neurological deficit.

Grip strength test

Mice were placed on a horizontal wire, and the time they could hang was recorded. Longer hanging times indicated better neuromuscular strength [2]. All tests were conducted by two independent observers blinded to group allocation.

mNSS and grip strength were measured repeatedly in the same mice at each designated time point.

Magnetic resonance imaging (MRI) acquisition and analysis

T2-weighted imaging was performed on a 4.7 T small-animal magnetic resonance imaging system (MR Solutions MRS-4717, UK) on day 3 following ICH. Prior to scanning, mice were anesthetized with inhaled isoflurane and placed on the scanning table. Mice were maintained under continuous isoflurane anesthesia during image acquisition. The scanning parameters were as follows: repetition time (TR) 5000 ms, echo time (TE) 51 ms, slice thickness 0.5 mm, slice spacing 0 mm, and field of view (FOV) 25 mm. Image J software was used to outline and calculate the area of brain injury in each slice. Hematoma volume was calculated using Image J software by outlining the hyperintense area on each slice and applying the formula: hematoma volume = (sum of areas) × slice thickness.

Western blot analysis

Perihematomal brain tissues or treated BV2 cells were harvested and lysed in RIPA buffer supplemented with protease and phosphatase inhibitors. Samples were homogenized on ice, incubated for 20 min, and centrifuged at 12,000×g for 15 min at 4 °C. The supernatant was collected as total protein, and protein concentrations were quantified using BCA assay kit. Equal protein samples were combined with 5 × loading buffer, heat-denatured in a boiling water bath (100 °C) for 10 min, and immediately cooled on ice. Proteins were separated on a 10% SDS-PAGE gel until the bromophenol blue reaches the bottom of the gel. Following electrophoresis, proteins were transferred onto PVDF membranes using a wet transfer system assembled as follows: sponge–filter paper–gel–PVDF membrane–filter paper–sponge. Membranes were blocked with protein-free rapid blocking buffer at room temperature for 1 h with gentle shaking, followed by overnight incubated with primary antibodies at 4 °C with shaking. After washing, membranes were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature, then the signal was visualized using ECL reagent on Amersham imager 600 (GE Life Sciences, MA, USA). Grayscale values were analyzed using ImageJ software. Detailed information on the antibodies used is provided in Table 2.

Table 2.

List of primary antibodies.

Antibody (host)	Dilution ratio	Company	Cat. Ref.
Western blot analysis
IL-1β (goat)	1:1000	R&D	AF-401-NA
TNF-α (Rabbit)	1:1000	abcam	ab183218
IL-6 (Rabbit)	1:1000	abcam	ab290735
iNOS (Rabbit)	1:1000	abcam	ab178945
MMP9 (Rabbit)	1:1000	abcam	ab283575
Phospho-PI3K p85 (Tyr458)/p55 (Tyr199) (Rabbit)	1:1000	Affinity	AF3242
PI3K p85 alpha (Rabbit)	1:1000	Affinity	AF6241
Phospho-AKT (S473) (Rabbit)	1:5000	HUABIO	ET1607-73
AKT1 (Rabbit)	1:5000	HUABIO	ET1609-47
Phospho–NF–kappaB p65 (Ser536) (Rabbit)	1:1000	Cell signaling technology	3303 S
NF-kappaB p65 (Rabbit)	1:1000	Cell signaling technology	8242 S
Phospho-Stat3 (Rabbit)	1:1000	Cell signaling technology	9145
GAPDH (Rabbit)	1:10000	HUABIO	ET1601-4
β-Actin (Rabbit)	1:10000	Proteintech	81115-1-RR
Goat anti-Rabbit IgG HRP	1:10000	abcam	ab205718
Donkey anti-goat IgG	1:5000	Proteintech	SA00001-3
Immunofluorescence staining
IBA1 (mouse)	1:500	abcam	ab283319
GFAP (mouse)	1:200	Cell signaling technology	3670 S
NeuN (mouse)	1:100	abcam	ab279295
Myeloperoxidase (Rabbit)	1:100	abcam	ab208670
CD16+CD32 (Rabbit)	1:400	abcam	ab223200
CD206 (Rabbit)	1:200	Proteintech	18704-1-AP
Arg-1 (Rabbit)	1:200	Cell signaling technology	93668 S
Laminin (Rabbit)	1:200	abcam	ab11575
MMP9 (Rabbit)	1:100	abcam	ab283575
Dylight 488 (goat anti-Rabbit IgG)	1:200	abbkine	A23220
DyLight 488 (goat anti-mouse IgG)	1:200	abbkine	A23210
Dylight 594 (goat anti-mouse IgG)	1:200	abbkine	A23410
Dylight 594 (goat anti-Rabbit IgG)	1:200	abbkine	A23420

Immunofluorescence staining of brain sections

Following cardiac perfusion, brain tissues were harvested, fixed in 4% paraformaldehyde with 20 ml saline followed by 20 ml 4% paraformaldehyde for 48 h in room temperature, dehydrated through graded ethanol, cleared with xylene, and embedded in paraffin. Coronal sections of 5 μm thick were prepared using a microtome. After dewaxing the sections with xylene and rehydrating through graded ethanol, and subjected to antigen retrieval. The sections were placed in a sodium citrate antigen retrieval solution and treated using a microwave method (100 °C for 3 min, 40 °C for 15 min). Following room temperature cooling, sections underwent three PBS washes, 15 min permeabilization with 0.2% Triton X-100, and 1 h blocking with 5% goat serum at room temperature. Overnight incubation with primary antibody incubation was performed at 4 °C in a humidified chamber. The next day, sections were exposed to the matching fluorescently labeled secondary antibody for 1 h at room temperature in the dark. Finally, sections were mounted with an anti-fade mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI). Details of the antibodies used are listed in Table 2. Images were acquired using an upright microscope (BX53, Olympus) under identical settings. Three non-consecutive sections were taken from each animal. For each section, three random fields of view (200 × or 400 × ) were captured in the perihematomal region or at the center of the hematoma for quantitative analysis. Quantitative analysis of positive cells was performed using Image J software. For structural analysis, the continuity index (total skeleton length/number of branches) was calculated using the skeleton function in Image J.

Hematoxylin and eosin (H&E) staining

After fixation and paraffin embedding, brain tissues were sectioned coronally at 5 μm. Sections were first dewaxed and rehydrated, then stained with hematoxylin for 5 min. After removal of excess dye under running water, they were stained with eosin solution for 2 min. After dehydration in graded ethanol, sections were mounted with neutral resin. Under a light microscope, we examined the perihematomal region for histopathological changes, including neuronal damage, vacuolization, inflammatory infiltration, and tissue integrity loss.

Nissl staining

Fixed and paraffin-embedded brain tissues were cut into 5 μm sections. Following deparaffinization and rehydration, sections were immersed in 0.1% toluidine blue for 10 min and rinsed with distilled water. Sections were differentiated in 95% ethanol until the background became clear, dehydrated in graded ethanol, cleared with xylene, and mounted with neutral resin. Nissl body morphology in the perihematomal region was examined under a light microscope. In normal neurons, Nissl bodies are abundant in the cytoplasm and appear as blue-violet granular structures; in damaged neurons, Nissl bodies are reduced, fragmented, or absent, and the cell bodies appeared shrunken.

Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) staining

Cell death was assessed using a TUNEL assay kit, which labels DNA fragmentation. After permeabilization, tissue sections were incubated with the TUNEL reaction mixture at 37 °C for 60 min protected from light. After DAPI counterstaining, TUNEL-positive cells were counted in three random selected 400 × fields per section in the perihematomal region.

Reverse transcription quantitative PCR (RT-qPCR)

Total RNA was extracted using the BioFlux Total RNA Extraction Kit. For perform reverse transcription, a 10 μL reaction mixture was prepared comprising 2 μL 5 × PrimeScript RT Master Mix, 500 ng RNA, and RNase-free ddH₂O to reach the final volume. The reaction was performed using PrimeScript RT Master Mix according to the manufacturer's instructions. Reaction conditions: 37 °C for 15 min, 85 °C for 5 s, then hold at 4 °C. The synthesized cDNA was preserved at −20 °C for subsequent analysis. Real-time PCR amplification was performed using SYBR Green Master Mix on a QuantStudio 5 real-time PCR system (Thermo Fisher Scientific, USA). Reaction mixture (10 μL): 5 μL SYBR Green Master Mix, 0.8 μL each of forward and reverse primers, 1 μL cDNA, 0.2 μL ROX Reference Dye II, and 3 μL ddH₂O. Reaction conditions: 30 s pre-denaturation at 95 °C, 5 s denaturation at 95 °C, 34 s annealing and extension at 60 °C, for a total of 40 cycles. Three replicates were set up for each sample. GAPDH was used as an internal control, and relative mRNA expression levels were calculated using the 2^{ˆ(−ΔΔCt)} method. The primer sequences used in this experiment are shown in Table 3.

Table 3.

Primers sequences for RT-qPCR.

Gene	Primer	Sequence
GAPDH	F	ACCCAGAAGACTGTGGATGG
	R	CACATTGGGGGTAGGAACAC
IL-1β	F	GCAACTGTTCCTGAACTCAACT
	R	ATCTTTTGGGGTCCGTCAACT
TNF-α	F	CCCTCACACTCAGATCATCTTCT
	R	GCTACGACGTGGGCTACAG
IL-10	F	GGCTGGACGAGAGCCGAACG
	R	CCCGGGGTGTAGGCACCACT
IL-6	F	TAGTCCTTCCTACCCCAATTTCC
	R	TTGGTCCTTAGCCACTCCTTC
Arg-1	F	CTCCAAGCCAAAGTCCTTAGAG
	R	AGGAGCTGTCATTAGGGACATC
CD206	F	TACTTGGACGGATAGATGGAGG
	R	CATAGAAAGGAATCCACGCAGT
CD86	F	TGTTTCCGTGGAGACGCAAG
	R	TTGAGCCTTTGTAAATGGGCA

Transcriptome sequencing and analysis

On the day 3 post-ICH, brain tissue surrounding the hematoma was collected from mice in the ICH group and the ICH + BBR group. Total RNA was extracted using Trizol reagent. RNA concentration was quantified using a Nanodrop, and RNA integrity was assessed using an Agilent 4200 TapeStation. Taking advantage of the polyA tail structure of eukaryotic mRNA, mRNA was enriched using Oligo (dT) magnetic beads. After random fragmentation with divalent cations, the fragmented mRNA served as a template for reverse transcription to synthesize the first strand of cDNA, followed by synthesis of the second strand. The purified double-stranded cDNA was subjected to end repair, A-tail addition, and sequencing adapter ligation. Fragments of 200–300 bp were selected for PCR amplification and purification to complete library construction. The library samples were sequenced using Illumina in a PE150 sequencing mode.

After obtaining the raw data, we used cutadapt (version 4.3) to remove raw sequencing data, yielding high-quality clean reads. The featureCounts software (version 2.0.4) was used to quantify reads mapped to gene regions and expression levels were normalized as FPKM (fragments per kilobase of transcript per million mapped reads). Differential expression analysis was performed using edgeR (version 3.40.2), with threshold of p value < 0.05 and |log2(FoldChange)| > 1. The topGO software package (version 2.50.0) was employed for GO enrichment analysis of differentially expressed genes (DEGs). Meanwhile, the clusterProfiler software (version 4.6.2) was utilized to conduct KEGG pathway enrichment analysis and gene set enrichment analysis (GSEA) on the selected DEGs.

Cell culture

The BV2 microglial cell line was sourced from Wuhan Procell Life Science Technology Co., Ltd. (Wuhan, China). Cells were cultured in high-glucose Dulbecco's Modified Eagle Medium (DMEM) containing 10% fetal bovine serum, 100 U/mL penicillin, and 100 μg/mL streptomycin. Incubation was carried out at 37 °C in a humidified atmosphere containing 5% CO₂. Upon reaching 80–90% confluence, the cells were subcultured using 0.25% trypsin-EDTA. All experiments were performed with cells harvested using cells in the logarithmic growth phase.

Cell viability assay

Cell viability was evaluated using a CCK-8 assay. BV2 cells were seeded in 96-well plates at 5 × 10³ cells per well and exposed to different concentrations of BBR (0, 5, 10, 20, 40, 80 μM) for 24 h. Thereafter, 10 μL of CCK-8 reagent was added to each well, and the plates were incubated at 37 °C for 1–2 h. Absorbance was read at 450 nm using a SpectraMax M5/M5e microplate reader. Viability was calculated as (OD value of experimental group – OD value of blank group)/(OD value of control group – OD value of blank group) × 100%.

Microglial response assay

BV2 cells were divided into three groups: control group, LPS group (100 ng/mL, 24 h), and LPS + BBR group (10 μM). In the LPS + BBR group, cells were pretreated with BBR for 2 h followed by treated with LPS for 24 h. After treatment, we assessed microglial functional states by immunofluorescence staining of Iba1/CD16 (pro-inflammatory) and Iba1/CD206 (repair-associated). The shift in the balance of microglial phenotypes was further quantified. In parallel, the mRNA levels of classical phenotypic markers and pro-inflammatory cytokines were determined by RT-qPCR.

Immunofluorescence staining of BV2 cells

BV2 cells grown on confocal dishes received the designated treatments. After fixation with 4% paraformaldehyde for 15 min, the cells were permeabilized with 0.3% Triton X-100 for 10 min, and blocked with 5% goat serum for 30 min. Subsequently, cells were incubated overnight at 4 °C with the primary antibodies (see Table 2 for details). After washing the next day, cells were incubated with the corresponding fluorophore-conjugated secondary antibodies for 1 h at room temperature, protected from light. Finally, the cells were mounted with an anti-fade medium containing DAPI and observed and photographed under a confocal microscope (Leica STELLARIS 5, China). Four fields of view were randomly captured per dish under identical acquisition parameters. Fluorescence intensity of CD16⁻and CD206-positive cells was quantified using Image J software.

Molecular docking

Retrieve the target protein from the UNIPROT database to obtain its sequence information, and predict the protein's 3D structure using ALPgFold3. Simultaneously, retrieve molecular information for BBR from the PubChem database and download its 3D structure file. Molecular docking was performed using AutoDock Vina software, and the conformation with the optimal binding free energy (affinity) was selected as the most likely binding mode for subsequent analysis. We utilized both PyMOL and Discovery Studio software for the visualization and analysis of the binding mode. In PyMOL, we focused on analyzing and highlighting the polar interactions formed between BBR and the protein's amino acid residues. Additionally, we employed Discovery Studio to conduct a comprehensive receptor-ligand interaction analysis and generated a two-dimensional ligand interaction diagram, which visually illustrates the key interaction features of the three-dimensional binding site.

Signal pathway validation

To further validate the molecular mechanism of BBR, the following experimental groups were designed:

Inhibitor validation: (1) Control group; (2) LPS group; (3) LPS + BBR group; (4) LPS + MK-2206 (10 μM, Akt inhibitor) group; (5) LPS + BAY 11–7082 (10 μM, NF-κB inhibitor) group. For inhibitor groups, MK-2206 was administered 2 h before LPS stimulation, and BAY 11–7082 was administered 1 h before LPS stimulation. Cells were harvested 30 min post-LPS stimulation, and phosphorylation levels of signaling pathway proteins were detected by Western blotting.

Rescue experiments: (1) Control group; (2) LPS group; (3) LPS + BBR group; (4) LPS + BBR+740 Y–P (20 μg/mL, PI3K agonist) group. In the rescue group, cells were co-treated with BBR and 740 Y–P for 2 h, followed by LPS stimulation for 30 min. PI3K and Akt phosphorylation were then evaluated by Western blotting.

Statistical analysis

All statistical analyses were performed using GraphPad Prism 8.0. Data are presented as mean ± standard deviation (SD). For multiple comparisons, one-way analysis of variance (ANOVA) with Tukey's post hoc test or two-way ANOVA with Bonferroni's correction was applied. mNSS and grip strength data were analyzed using two-way ANOVA. For other experiments involving independent group comparisons, one-way ANOVA was applied. Student's t-test was used for comparisons between two groups. A p-value of <0.05 was considered statistically significant.

Results

Network pharmacology predicts potential targets and pathways of BBR against ICH

To systematically investigate the molecular mechanisms underlying the therapeutic effects of BBR (Fig. 1A and B) in ICH, a network pharmacology analysis was first performed. Database searches identified 481 BBR-associated targets and 1764 ICH-associated targets, with an intersection of 208 common targets (Fig. 1C and Table S1). Key targets included TP53, AKT1, STAT3, SRC, EGFR, JUN, PIK3CA, RELA, MAPK3, and CTNNB1 (Fig. 1D). A PPI network was constructed based on interaction strengths from the STRING database (Fig. 1E).

Fig. 1.

Network pharmacology predicts potential targets and pathways of BBR against ICH:-A, B. Chemical structure of berberine. Two-dimensional (A) and three-dimensional (B) structures of BBR. C. Identification of overlapping targets. Venn diagram showing the intersection of BBR-associated targets (481) and ICH-associated targets (1,764), yielding 208 common targets. D. Core target screening. Key targets include TP53, AKT1, STAT3, SRC, EGFR, JUN, PIK3CA, RELA, MAPK3, and CTNNB1. E. PPI network. The PPI network of 208 common targets was constructed using the STRING database. Nodes represent target proteins, and edges represent interactions. Node size and color depth are proportional to degree centrality. F. GO enrichment analysis. Bar chart displaying the top 10 enriched biological processes, cellular components, and molecular functions, sorted by -log₁₀ (p-value). G. KEGG pathway enrichment analysis. Bubble chart illustrating the top 10 enriched pathways, including the PI3K/Akt, apoptosis, HIF-1 and FoxO signaling pathway.

GO enrichment analysis indicated that these targets are primarily involved in biological processes such as inflammatory responses, cell proliferation, and apoptosis regulation (Fig. 1F). KEGG pathway enrichment analysis showed that the overlapping targets were significantly enriched in the PI3K/Akt, apoptosis, HIF-1 and the FoxO signaling pathway (Fig. 1G). The component-target-pathway network diagram visually demonstrates that BBR may exert its therapeutic effects through multiple targets and pathways.

BBR alleviates ICH-induced neurological deficits and brain injury

To evaluate the therapeutic effects of BBR on ICH, an ICH mouse model was established and BBR treatment was administered employing a comprehensive approach that included MRI, neurological function scoring, HE staining, and Nissl staining. The experimental timeline is shown in Fig. 2A. MRI scans revealed that BBR treatment significantly reduced hematoma volume on days 3 post-surgery (Fig. 2B and C).

Fig. 2.

BBR alleviates ICH-induced neurological deficits and brain injury:-A. Experimental timeline. Schematic diagram showing the time course of ICH model establishment, BBR administration, and subsequent behavioral and histological assessments. B, C. Hematoma volume. Representative T2-weighted MRI images (B) and quantitative analysis of hematoma volume (C) on day 3 after ICH (n = 4 per group, Scale bar = 5 mm). D. Neurological function. Modified neurological deficit scores assessed on day(s) 1, 3, and 7 post-ICH (n = 6 per group). E. Grip strength. Forelimb grip strength measured at baseline and on day(s) 1, 3, and 7 post-ICH (n = 6 per group). F, G. Histopathological damage. Representative HE staining images of brain tissue surrounding the hematoma in Sham, ICH, and ICH + BBR groups and quantitative analysis of lesion area (mm²) (n = 4 per group, Scale bar = 1 mm/50 μm). H, I. Neuronal loss. Representative Nissl staining images of brain tissue surrounding the hematoma in Sham, ICH, and ICH + BBR groups and quantitative analysis of Nissl-positive neurons (cells/field). (n = 4 per group, Scale bar = 1 mm/50 μm). Differences between groups were analyzed using one-way ANOVA or two-way ANOVA. Data are presented as mean ± SD. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Modified neurological severity score assessment indicated that neurological scores in the ICH group were significantly elevated in the ICH group compared with the Sham group at all postoperative time points. Compared with the ICH group, the BBR treatment group exhibited significantly lower neurological scores on day(s) 1, 3, and 7 postoperatively, suggesting improved neurological function (Fig. 2D). Baseline grip strength was comparable among all groups prior to surgery. On postoperative day(s) 1, 3, and 7, grip strength in the ICH group was significantly lower than that in the Sham group, suggesting that ICH induced a marked decrease in contralateral limb muscle strength. Notably, grip strength in BBR-treated group was significantly higher than that in the ICH group on both postoperative day(s) 1, 3, and 7, indicating that BBR improves post-ICH muscle strength recovery (Fig. 2E).

HE staining revealed that the cortical and basal ganglia regions of the Sham group mice exhibited intact tissue structures, normal cell morphology, and distinct nuclei, with no obvious vacuolization or inflammatory cell accumulation. In the ICH group, the perihematomal brain tissue displayed marked tissue destruction, vacuolization, and inflammatory infiltration. In contrast, in the ICH + BBR group showed substantially reduced tissue damage and partial recovery of cell morphology on day 3 post-surgery (Fig. 2F). For quantitative analysis of HE staining, the lesion area (mm²) in each section was measured using Image J (Fig. 2G). Nissl staining revealed abundant Nissl-positive neurons in the cortical and basal ganglia regions of the Sham group brains, with distinct blue-purple Nissl bodies in the cytoplasm. In contrast, the perihematomal region of ICH brains showed a marked reduction in Nissl-positive cells. The remaining Nissl bodies appearing fragmented, dissolved, and faintly stained, indicating severe neuronal injury. BBR administration markedly elevated the count of Nissl-positive neurons, improved neuronal morphology, and partially restored Nissl bodies, demonstrating a neuroprotective effect on day 3 post-surgery (Fig. 2H). Nissl-positive neurons in the perihematomal region were counted in three random fields per section (400 × ) using Image J (Fig. 2I).

Transcriptomic sequencing reveals that BBR regulates the expression of genes associated with inflammatory responses and extracellular matrix (ECM) remodeling

To further investigate the molecular mechanisms of BBR, transcriptomic sequencing was performed on brain tissues from mice in the ICH group and ICH + BBR group on day 3 post-surgery. The volcano plot showed that, compared with the ICH group, the ICH + BBR group exhibited 246 upregulated and 224 downregulated genes (Fig. 3A). The clustering heatmap visually illustrates the regulatory effects of BBR on DEGs (Fig. S1A). A PPI network constructed from DEGs revealed extensive interactions among the encoded proteins (Fig. S1B).

Fig. 3.

Transcriptomic sequencing reveals that BBR regulates the expression of genes associated with inflammatory responses and ECM remodeling:-A. Identification of differentially expressed genes (DEGs). Volcano plot showing DEGs between the ICH + BBR group and the ICH group. B. GO enrichment analysis. Genes downregulated in the ICH + BBR group are primarily enriched in inflammatory response and immune response; genes upregulated in the ICH + BBR group are primarily enriched in ECM organization and synaptic transmission. C. KEGG pathway enrichment analysis. Dot plot showing significantly enriched pathways, including the PI3K/Akt, Toll-like receptor, IL-17, NOD-like receptor, Cytokine-cytokine receptor interaction and ECM-receptor interaction signaling. D-I. GSEA analysis. GSEA enrichment plots showing that the PI3K/Akt signaling (D, E), IL-17 signaling (F, G), and Cytokine-cytokine receptor interaction (H, I) are significantly downregulated in the ICH + BBR group compared with the ICH group. (n = 3 per group).

According to GO enrichment analysis, genes downregulated in the ICH + BBR group relative to the ICH group were mainly enriched in inflammatory responses and immune responses (Fig. 3B), suggesting that BBR may exert neuroprotective effects through multi-level regulation of biological events. Genes upregulated in the ICH + BBR group were predominantly linked to ECM organization and synaptic transmission, indicating that BBR may promote neural tissue repair (Fig. 3B). KEGG enrichment analysis revealed that the identified genes were prominently associated with several signaling cascades, including the PI3K/Akt, Toll-like receptor, IL-17 and NOD-like receptor, as well as ECM receptor interactions (Fig. 3C). These pathways are strongly implicated in neuroinflammation, cell survival, and vascular basement membrane integrity, thereby providing direction for subsequent mechanistic studies.

GSEA analysis further revealed the overall regulatory trends of these pathways. The results demonstrated significant enrichment of GO gene sets related to inflammatory responses and cytokine production. At the pathway level, in the ICH + BBR group exhibited marked downregulation of several pathways, including the PI3K/Akt, IL-17, Cytokine-cytokine receptor interaction, Toll-like receptor and NOD-like receptor signaling pathways (Fig. 3D–I and S1C-D). Notably, BBR suppressed the expression of genes associated with ICH-induced inflammatory responses and immune responses, primarily involving the PI3K/Akt signaling pathways. Furthermore, the enrichment of the extracellular matrix receptor interaction pathway suggests that BBR may participate in vascular protection by regulating ECM remodeling.

BBR alleviates neuroinflammation and cell death following ICH

The effects of BBR on neuroinflammation were evaluated using immunofluorescence staining and Western blot analysis. Immunofluorescence staining was employed to examine microglia and astrocytes activation, as well as neutrophil infiltration. Iba1 and GFAP staining revealed marked accumulation of both microglia and astrocytes in the perihematomal region of the ICH mice, whereas BBR treatment substantially reduced both cell populations (Fig. 4A, B, E and F).

Fig. 4.

BBR alleviates neuroinflammation and cell death after ICH:-A. Microglial activation. Representative immunofluorescence images of Iba1 (green) staining in the perihematomal region (n = 4 per group, Scale bar = 50 μm). B. Astrocyte activation. Representative immunofluorescence images of GFAP (green) staining in the perihematomal region (n = 4 per group, Scale bar = 100 μm). C. Cell death. Representative images of TUNEL staining (green) in the central region of the hematoma (n = 4 per group, Scale bar = 50 μm). D. Neutrophil infiltration. Representative immunofluorescence images of MPO (green) staining in the central region of the hematoma (n = 4 per group, Scale bar = 50 μm). E-H. Quantification of positive cells. Quantitative analysis of Iba1-positive cells (E), GFAP-positive cells (F), TUNEL-positive cells (G), and MPO-positive cells (H). I. Inflammatory protein expression. The protein levels of MMP-9, TNF-α, and IL-1β in brain tissue were analyzed using Western blot (n = 6 per group). J-L. Quantification of Western blot. Quantitative analysis of MMP9 (I), TNF-α (K), and IL-1β (L). M, N. Inflammatory cytokine mRNA expression. RT-qPCR analysis of TNF-α (M) and IL-1β (N) mRNA levels in brain tissue (n = 4 per group). Differences between groups were analyzed using one-way ANOVA. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Microglial morphological analysis revealed that Iba1⁺ microglia in the Sham exhibited a branched morphology with small cell bodies, slender processes and abundant branching. In the ICH group, microglial morphology underwent significant changes, characterized by enlarged cell bodies, shortened and thickened processes, and an amoeboid activated morphology. In contrast, microglial morphology in the ICH + BBR group partially recovered, tending toward a quiescent state (Fig. 4A). Regarding astrocytic morphology, astrocytes in the Sham group exhibited a stellate morphology with small cell bodies and slender, radially distributed processes. In the ICH group, astrocytes were markedly activated, with enlarged cell bodies and thickened, elongated processes. In contrast, the ICH + BBR group showed reduced astrocyte activation with smaller cell bodies and processes tending toward a normal morphology (Fig. 4B). Quantitative analysis showed that BBR treatment significantly reduced the number of Iba1-positive and GFAP-positive cells in the perihematomal region compared with the ICH group (Fig. 4E-F).

TUNEL staining revealed markedly increased cell death in the ICH group, an effect that was substantially attenuated by BBR treatment (Fig. 4C and G). Myeloperoxidase (MPO) staining showed significant neutrophil infiltration in the ICH group, which was markedly attenuated by BBR treatment (Fig. 4D and H). Western blot analysis demonstrated significantly elevated protein levels of matrix metallopeptidase-9 (MMP-9), tumor necrosis factor α (TNF-α), and interleukin-1β (IL-1β) proteins in brain tissue from the ICH group, and these elevations were substantially reversed following BBR administration (Fig. 4I-L). RT-qPCR analysis further confirmed that BBR treatment significantly downregulated TNF-α and IL-1β mRNA levels in ICH mice (Fig. 4M-N).

BBR protects basement membrane integrity and inhibits ECM degradation after ICH

Given the transcriptomic indication that BBR may regulate ECM-related pathways, we performed a focused analysis on genes involved in ECM remodeling. GO term enrichment analysis by GSEA revealed significant differences between the ICH group and the ICH + BBR group in the “ECM organization” and “Collagen metabolic process” pathways (Fig. 5A and B). BBR treatment reversed ICH-induced aberrant expression of ECM-related genes, including members of the matrix metalloproteinase, collagen, and laminin families (Fig. 5C and D). Further analysis revealed significant alterations in the “Collagen fibril organization” and “ECM structural constituent conferring” pathways, suggesting profound remodeling of the structural integrity and mechanical properties of the collagen network following ICH (Fig. S2A–D).

Fig. 5.

BBR protects basement membrane integrity and inhibits ECM degradation after ICH:-A, B. Enrichment analysis of ECM-related pathways. GO term enrichment analysis by GSEA showing significant differences between the ICH and ICH + BBR groups in “ECM organization” (A) and “Collagen metabolic process” (B). C, D. Heatmap of ECM-related gene expression. Heatmaps showing ICH-induced abnormalities in the expression of ECM-related genes, including matrix metalloproteinases, collagens, and laminins, and their reversal by BBR treatment. E. MMP9 expression. Representative immunofluorescence images of MMP-9 (green) in the central region of the hematoma (n = 4 per group, Scale bar = 100 μm). F. Laminin expression. Representative immunofluorescence images of laminin (red) in the perihematomal region (n = 4 per group, Scale bar = 50 μm). G. Quantification of MMP9-positive cells. Quantitative analysis of MMP-9 positive cell numbers. H. Quantification of laminin continuity index. Quantitative analysis of the laminin continuity index. Differences between groups were analyzed using one-way ANOVA. Data are presented as mean ± SD. ∗∗p < 0.01, ∗∗∗∗p < 0.0001.

Immunofluorescence staining revealed that a marked increase in MMP-9 positive cells in the ICH group, accompanied by disruption and fragmentation of the continuous linear laminin network. In contrast, BBR treatment reduced MMP-9 expression and partially restored laminin continuity (Fig. 5E and F). Quantitative analysis confirmed that BBR significantly lowered the number of MMP-9 positive cells (Fig. 5G) and markedly elevated the laminin continuity index (Fig. 5H), indicating that BBR protects the integrity of the basement membrane and inhibits ECM degradation.

BBR shifts BV2 microglia from a pro-inflammatory toward a repair-associated phenotype

To investigate whether BBR influences microglial phenotype switching, we established an LPS-induced BV2 cell model. The timeline of the cellular experiments is shown in Fig. 6A. CCK-8 assays showed that BBR at concentrations up to 10 μM did not affect BV2 viability, therefore, 10 μM was selected for subsequent experiments (Fig. 6B).

Fig. 6.

BBR shifts BV2 microglia from a pro-inflammatory toward a repair-associated phenotype:-A. Experimental timeline. Schematic diagram showing the time course of BBR pretreatment, LPS stimulation, and subsequent cellular assays. B. Cell viability. CCK-8 assay to evaluate the effects of different concentrations of BBR (0, 5, 10, 20, 40, and 80 μM) on BV2 cell viability (n = 4 per group). C, D. Immunofluorescence staining of microglial phenotypes. Representative images of BV2 cells double-stained for Iba1 (green) and either the pro-inflammatory marker CD16 (red) or the anti-inflammatory marker CD206 (red). E, F. Quantification of fluorescence intensity. Quantitative analysis of CD16 (E) (n = 4 per group, Scale bar = 50 μm) and CD206 (F) (n = 5 per group, Scale bar = 50 μm) fluorescence intensity in different treatment groups. G-M. Inflammatory cytokine and phenotypic marker mRNA expression. RT-qPCR analysis of IL-1β (G) (n = 4 per group), TNF-α (H) (n = 5 per group), IL-6 (I) (n = 4 per group), CD86 (J) (n = 4 per group), CD206 (K) (n = 4 per group), IL-10 (L) (n = 6 per group) and Arg1 (M) (n = 5 per group) mRNA levels. Differences between groups were analyzed using one-way ANOVA. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Immunofluorescence staining revealed that, compared with the control group, the LPS group exhibited significantly higher CD16 fluorescence intensity and lower CD206 fluorescence intensity. After BBR treatment, CD16 fluorescence intensity was markedly reduced compared with the LPS group, while CD206 intensity was significantly increased, suggesting that BBR inhibits pro-inflammatory microglial phenotype and favors the acquisition of a repair-associated phenotype (Fig. 6C–F). RT-qPCR analysis revealed that LPS stimulation significantly upregulated the mRNA levels of pro-inflammatory markers CD86, IL-1β, TNF-α, and IL-6, while downregulating the repair-associated markers Arg1, CD206, and IL-10 (Fig. 6G-M). BBR treatment reversed these changes, suppressing the pro-inflammatory signature and restoring the expression of repair-related genes.

Molecular docking predicts that BBR targets the PI3K/Akt/NF-κB axis

To further explore how BBR regulates neuroinflammation, we performed molecular docking based on network pharmacology and transcriptomic data to predict the binding affinity and interaction modes with key targets. According to the molecular docking analysis, BBR formed stable binding complexes with all five key targets examined (Fig. 7A and B). Fig. 7A and B depict the 3D and 2D structures of BBR at its optimal docking positions with the core targets, respectively.

Fig. 7.

Berberine targets PI3K and inhibits the downstream PI3K/Akt/NF-κB signaling axis:-(A-B) Molecular docking analysis. (A) 3D and (B) 2D structures showing the binding modes of berberine (BBR) with PIK3CA, AKT1, RELA, NF-κB1, and STAT3. Hydrogen bonds, pi-alkyl, pi-cation, and van der Waals interactions are indicated. (C) Binding affinity comparison. Binding energies (kcal/mol) of BBR with the five core targets. (D) Representative Western blots. Protein expression levels of p-PI3K, PI3K, p-Akt, Akt, NF-κB, p–NF–κB, p-STAT3, IL-1β, TNF-α, IL-6, iNOS and β-actin in BV2 cells under different treatments. (E–O) Quantification of relative gray value. (E, G) Total PI3K and Akt protein levels show no significant differences among groups (n = 6 per group). (F, H–K) LPS stimulation significantly increases p-PI3K (F), p-Akt (H), NF-κB (I), p–NF–κB (J), and p-STAT3 (K) levels (n = 6 per group), whereas BBR treatment effectively inhibits these phosphorylation events. (L–O) The expression of NF-κB and STAT3 target genes, including IL-1β (L), TNF-α (M), IL-6 (N), and iNOS (O) (n = 6 per group). Differences between groups were analyzed using one-way ANOVA. Data are presented as mean ± SD. ^ns p > 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Specifically, BBR stably binds to the LYS672 site of PIK3CA via hydrogen bonds. In AKT1, BBR primarily forms hydrogen bonds with GLN218 and pi-alkyl interactions with LYS183. In RELA, BBR forms pi-cation interactions with ARG84. In NF-κB, BBR forms non-covalent interactions with TRP828 and others. In STAT3, BBR binds to TRP43, ALA44, and SER113 via van der Waals forces. BBR exhibits the lowest binding energy with PIK3CA (the p110α catalytic subunit of PI3K) at −7.7 kcal/mol, which is significantly stronger than its interactions with AKT1 (−6.7 kcal/mol), RELA (−5.9 kcal/mol), NF-κB1 (−6.6 kcal/mol), and STAT3 (−6.7 kcal/mol) (Fig. 7C). These findings indicate that PI3K is likely the principal target of BBR within the AKT/NF-κB inflammatory axis. The stable binding of BBR to the LYS672 site of PIK3CA may inhibit PI3K catalytic activity by occupying this site.

BBR inhibits LPS-induced activation of the PI3K/Akt/NF-κB pathway

Based on molecular docking results, we assessed the impact of BBR on the PI3K/Akt/NF-κB pathway and its associated signaling components, including STAT3, by Western blotting (Fig. 7D). LPS stimulation markedly elevated the phosphorylation of p-PI3K, p-Akt, p–NF–κB, and p-STAT3 in BV2 cells, whereas BBR treatment effectively inhibited these phosphorylation events (Fig. 7F and H-K). Furthermore, BBR significantly reduced the expression of the NF-κB and STAT3 downstream targets IL-1β, IL-6, TNF-α, and iNOS (Fig. 7L-O). Meanwhile, total PI3K and Akt protein levels remained unchanged across all groups (Fig. 7E and G). These results indicate that BBR inhibits the PI3K/Akt signaling axis, thereby suppressing downstream NF-κB activation and modulates STAT3 signaling through pathway crosstalk.

BBR inhibits LPS-induced nuclear translocation of NF-κB

To morphologically assess BBR-mediated inhibition of NF-κB pathway, we examined the subcellular localization of NF-κB using immunofluorescence. In the control cells, NF-κB was predominantly cytoplasmic and exhibited minimal colocalization with DAPI-stained nuclei (Fig. 8A). After 30 min of LPS stimulation, NF-κB underwent significant nuclear translocation, as evidenced by a strong increase in nuclear NF-κB fluorescence. In contrast, BBR pretreatment significantly inhibited this nuclear translocation, with NF-κB remaining largely confined to the cytoplasm.

Fig. 8.

BBR blocks NF-κB nuclear translocation and suppresses PI3K/Akt/NF-κB signaling: evidence from inhibitor and rescue experiments:-(A) NF-κB nuclear translocation. Representative immunofluorescence images showing subcellular localization of NF-κB in BV2 cells (n = 5 per group, Scale bar = 50 μm). (B–D) Inhibitor validation. (B) Protein levels of p-AKT and p–NF–κB following treatment with the AKT inhibitor MK-2206 and the NF-κB inhibitor BAY 11–7082. (C–D) Quantitative analysis. MK-2206 significantly inhibits p-AKT while partially suppressing p–NF–κB; BAY 11-7082 markedly suppresses p–NF–κB without affecting p-AKT (n = 5/6 per group). (E–G) Rescue experiments with PI3K agonist 740 Y–P. (E) Protein levels of p-PI3K and p-AKT following treatment with the PI3K-specific agonist 740 Y–P. (F–G) Quantitative analysis. Phosphorylated protein levels are significantly reduced in the BBR-treated group; however, the addition of 740 Y–P does not restore these levels (n = 5 per group). Differences between groups were analyzed using one-way ANOVA. Data are presented as mean ± SD. ^ns p > 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Pharmacological inhibition and rescue experiments validate that BBR targets the PI3K/Akt/NF-κB axis

To validate BBR-mediated inhibition of the AKT and NF-κB pathways, we employed the selective inhibitors MK-2206 (AKT) and BAY 11–7082 (NF-κB). As shown in Fig. 8B–D, MK-2206 significantly inhibited p-AKT levels while partially inhibiting p–NF–κB levels. BAY 11-7082 markedly suppressed p–NF–κB levels but did not significantly affect p-AKT. The inhibitory effect of BBR resembled the combined action of both inhibitors, indicating that AKT functions upstream of NF-κB.

Rescue experiments revealed that the PI3K agonist 740 Y–P failed to reverse the inhibitory effects of BBR on p-PI3K and p-AKT (Fig. 8E–G). Compared with the LPS group, the levels of phosphorylated proteins were significantly reduced in the BBR-treated group. However, the addition of 740 Y–P did not restore these phosphoproteins. Based on molecular docking result that BBR stably binds to the LYS672 residue of the p110 catalytic subunit, we speculate that BBR may induce a conformational change in p110α, thereby uncoupling the signal relay from the p85 regulatory subunit to the catalytic core and preventing 740 Y–P-mediated activation. This finding supports that BBR binds to PIK3CA with high specificity and may exert sustained regulatory effects through allosteric inhibition.

Schematic diagram of the mechanism by which BBR regulates microglial functional states

Based on these findings, a schematic summarizing the proposed mechanism by which BBR modulates microglial functional states is presented in Fig. 9. BBR binds to the LYS672 site of PIK3CA, thereby inhibiting PI3K catalytic activity and subsequently blocking Akt phosphorylation. The resulting decrease in Akt activity then suppresses downstream NF-κB activation and modulates STAT3 signaling through pathway crosstalk. This cascade leads to the coordinated downregulation of pro-inflammatory associated genes (IL-1β, TNF-α, IL-6 and iNOS) and promotes repair-associated microglial phenotype, ultimately alleviating post-ICH neuroinflammation and conferring neuroprotection.

Fig. 9.

Schematic diagram of the mechanism by which BBR regulates microglial functional states:-BBR binds to the LYS672 site of PIK3CA, thereby inhibiting PI3K catalytic activity and subsequently blocking Akt phosphorylation. The resulting decrease in Akt activity then suppresses downstream NF-κB signaling and modulates STAT3 activity through pathway crosstalk. This cascade leads to the coordinated downregulation of pro-inflammatory associated genes (IL-1β, TNF-α, IL-6 and iNOS) and promotes repair-associated microglial states, ultimately alleviating post-ICH neuroinflammation and conferring neuroprotection.

Discussion

Secondary brain injury following ICH arises from complex mechanisms, with microglial-mediated neuroinflammatory responses playing a pivotal role [21,32]. Modulating the phenotypic transition of microglia from classically activated toward a repair-associated state has become a key therapeutic strategy for ICH [16,33,34]. This study demonstrates that BBR targets the PI3K/Akt axis to suppress downstream NF-κB signaling while modulating STAT3 activity through pathway crosstalk, thereby shifting microglial functional states and alleviates post-ICH neuroinflammation.

BBR possesses neuroprotective effects and multi-target properties [35,36]. Network pharmacology predicted 208 potential BBR targets in ICH, with significant enrichment in multiple inflammation-related pathways, including PI3K/Akt, NF-κB, and JAK/STAT, consistent with the “multi-target, multi-pathway” characteristics of BBR [37,38]. In vivo, BBR treatment significantly reduced hematoma volume, preserved neurological function, attenuated neuronal damage, and inhibited microglial/astrocytic activation and cell death. Transcriptomic sequencing further revealed that BBR reversed ICH-induced inflammatory gene expression and significantly enriched ECM-related pathways, suggesting that BBR may reduce ECM and basement-membrane damage [39].

The PI3K/Akt axis serves as the upstream hub through which BBR regulates inflammation [40]. Molecular docking revealed that BBR binds most strongly to PIK3CA (−7.7 kcal/mol) via the LYS672 residue, suggesting that PI3K as a primary target. In inhibitor validation experiments, MK-2206 potently inhibited p-AKT and partially suppressed p–NF–κB, whereas BAY 11–7082 specifically inhibited p–NF–κB without affecting p-AKT. These results collectively indicate that AKT acts upstream of NF-κB, a signaling hierarchy that closely aligns with the mode of action of BBR.

Notably, the PI3K agonist 740 Y–P, which activates PI3K by binding to the SH2 domain of the p85 subunit [41], failed to reverse BBR-mediated suppression of p-PI3K. Given that BBR stably binds to the LYS672 site of the p110α catalytic subunit, we propose that BBR may disrupt signal relay from p85 to p110α by inducing a conformational change in p110α. This would render p110α catalytically inactive even when 740 Y–P binds p85 and recruits PI3K to the membrane [42,43]. Interestingly, LPS stimulation still elevates p-PI3K levels, suggesting an activate p85-independent mechanism [44], which BBR also blocks, further supporting PIK3CA as a key functional target.

NF-κB, a master regulator of inflammation, drives the transcription of multiple pro-inflammatory genes, including IL-1β, TNF-α, IL-6 and iNOS [45,46]. STAT3, which can be activated via JAK-mediated phosphorylation at Tyr705 or modulated by PI3K/Akt through acetylation [47,48], functions as an interconnected signaling node that may cooperate with NF-κB in regulating pro-inflammatory gene expression [49]. STAT3 has also been shown to participate in the regulation of pro-inflammatory state and can form transcriptional complexes with NF-κB at the promoters of various pro-inflammatory genes, thereby producing a synergistic amplification effect [[50], [51], [52]]. BBR concurrently suppressed p–NF–κB and p-STAT3, suggesting a “dual brake” effect on pro-inflammatory microglial activation. By targeting the upstream PI3K/Akt hub, BBR attenuates downstream NF-κB signaling, while also modulating STAT3 activity through pathway crosstalk, a strategy that may be more effective than inhibiting either pathway alone.

The functional link between PI3K/Akt/NF-κB signaling and microglial phenotypic changes is supported by consistent evidence across models. In BV2 microglia, BBR suppressed pathway phosphorylation and concurrently shifted cells from a pro-inflammatory to a repair-associated phenotype. In vivo, BBR reduced microglial activation and NF-κB-driven cytokine production. These coordinated changes suggest that modulation of the PI3K/Akt/NF-κB axis directly contributes to the altered microglial functional states observed after BBR treatment.

ECM preservation represents a novel dimension of the mechanism of BBR. Transcriptomic sequencing revealed significant enrichment of ECM-related genes, and immunofluorescence further confirmed that BBR inhibits MMP-9 expression and preserved laminin integrity [53]. Given that ECM integrity is crucial for neuronal survival, axonal regeneration, and vascular homeostasis [54,55], this effect constitute another neuroprotective axis.

We acknowledge that the exclusive use of male mice and the BV2 cell model are limitations, as BV2 cells do not fully recapitulate the post-ICH microenvironment. However, our in vivo experiments support the translational relevance of our findings, and future studies using primary microglia and female animals are warranted to validate our findings.

Collectively, our findings demonstrate that BBR targets the PI3K/Akt axis to suppress downstream NF-κB signaling and modulate STAT3 through pathway crosstalk, shifting microglial functional states from a pro-inflammatory toward a repair-associated phenotype, thereby alleviating post-ICH neuroinflammation and improving neurological recovery. This study provides a new theoretical foundation for BBR in ICH treatment and offers insights for neuroinflammatory intervention strategies based on signaling network regulation.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors contributions

QLW: Conceptualization, Methodology, Validation, Formal analysis, Main experimental work, Data Curation, Writing - original draft. XLL: Validation, Formal analysis, Data Curation, Visualization. YL, RXW: Methodology, Investigation, Preliminary data collection. ZL, RS, PPG, YC, HZZ, LC: Preliminary data collection, Investigation. V. Wee Yong, MZX: Writing - Review & Editing, Supervision. All authors read and approved the final manuscript.

Funding

This study was supported by grants from the National Natural Science Foundation of China (No. U25A2065, W2541024); Zhengzhou University Youth Student Basic Research Project (No. ZDBJ202540); Henan Provincial Key Research and Development and Promotion Special (No. 232102311086); Henan Province Youth and Middle-aged Health Science and Technology Innovation Excellent Youth Talent Cultivation Project (No. YQRC2023012).

Declaration competing interests

The authors declare that they have no competing interests.

Abbreviations

Akt

Protein kinase B

BBB

Blood-brain barrier

BBR

Berberine

BC

Betweenness centrality

CC

Closeness centrality

DC

Degree centrality

DEGs

Differentially expressed genes

DAPI

4′,6-diamidino-2-phenylindole

ECM

Extracellular matrix

TNF-α

Tumor necrosis factor α

GO

Gene ontology

GSEA

Gene set enrichment analysis

H&E

Hematoxylin and eosin

ICH

Intracerebral hemorrhage

IL-1β

Interleukin-1β

KEGG

Kyoto encyclopedia of genes and genomes

LPS

Lipopolysaccharide

MMP-9

Matrix Metallopeptidase 9

mNSS

modified neurological severity score

MRI

Magnetic resonance imaging

MPO

Myeloperoxidase

NF-κB

Nuclear factor kappa-B

PI3K

Phosphatidylinositol 3-kinase

PPI

Protein-protein interaction

RT-qPCR

Reverse transcription quantitative PCR

STAT3

Signal transducer and activator of transcription 3

TUNEL

Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling

Appendix A. Supplementary data

The following is the Supplementary data to this article.