Adult C57BL/6J (B6) mice of both sexes (10–12 weeks old) were purchased from Jackson Laboratory (Bar Habor, Maine, US) (Stock No. 000664, RRID : IMSR_JAX:000664). All animals were handled in accordance with the National Institutes of Health and the Association for Research in Vision and Ophthalmology, and all experimental procedures and the use of animals were approved and monitored by the Animal Care Committee of Schepens Eye Research Institute of Massachusetts Eye and Ear.

Baicalein (Cat. No. 70610, Cayman Chemical Company, USA) was dissolved in 20% (2-hydroxypropyl)-beta-cyclodextrin (Cat. No. H107, Sigma-Aldrich, St. Louis, MO, USA) to the desired concentrations and prepared freshly before use. Vehicle control solution 20% (2-hydroxypropyl)-beta-cyclodextrin (βcd) was dissolved in PBS and prepared freshly before use.

Mouse microglia BV2 cell line was purchased from the ATCC (Manassas, Virginia, US) (Cat. No. CRL-2469, RRID : CVCL_5745) and cultured in DMEM medium (Cat. No. 11885084, Thermo Fisher Scientific Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Cat. No. F4135-500ML, Sigma-Aldrich) and 50 ng/ml recombinant mouse colony stimulator factor 1 (rmCSF1) (Cat. No. 315-02, PeproTech, Cranbury, NJ, USA). The medium was changed every 3 days. Cells were seeded into six-well plates and stimulated with 1 μg/ml LPS (Cat. No. L6529-1MG, Sigma) for 6 h followed by vehicle (LPS + Veh) or 10 μM baicalein (LPS + Ba). After 48 h of culturing, the supernatants were collected for cytokine arrays (N = 2), and cells were fixed for immunostaining or RNAs were extracted for semiquantitative PCR (qPCR) (N = 6).

The human microglial cell line HMC3 was purchased from the ATCC (Cat. No. CRL-3304, RRID : CVCL_II76) and cultured in Eagle’s minimum essential medium (ATCC, Cat. No. 30-2003) with 10% FBS (Cat. No. F2442, Sigma-Aldrich). A total of 2,000 cells per well were seeded in a 48-well plate. Activation of HMC3 cells was induced by 100 ng/ml LPS (Cat. No. L6529, Sigma-Aldrich) for 6 h followed by 5 mM adenosine triphosphate (ATP) (Cat. No. R0441, Thermo Fisher Scientific) for 30 min (17). PBS, vehicle, or baicalein (10 μM) was then added after the cells were stimulated with 100 ng/ml LPS for 6 h plus an additional 30 min of ATP treatment. After 48 h of incubation, RNAs were extracted for qPCR (N = 5) and supernatants from HMC3 cell cultures were collected for cytokine array assays (N = 4).

Cytokine arrays were conducted as previously described (18). Both HMC3 and BV2 microglia cells were seeded in six-well plates overnight before the experiments. HMC3 cells were divided into four groups: 1) control, 2) 100 ng/ml LPS + 5 mM ATP, 3) LPS + ATP + Veh, and 4) LPS + ATP + 10 μM Ba. After incubating cells with LPS for 6 h followed by 30 min of ATP stimulation as described previously (17), vehicle or 10 μM baicalein was added. After 48 h of incubation, the supernatants of the culture were collected for cytokine array assay following the manufacturer’s instructions [human cells: Proteome Profiler Human Cytokine Array Kit, R&D Systems, (Minneapolis, US) Cat. No. ARY005B; mouse cells: Proteome Profiler Mouse Cytokine Array Kit, Panel A, R&D Systems, Cat. No. ARY006]. Briefly, the mixture of cell culture supernatant and cytokine detection antibody cocktail was incubated with cytokine array membrane overnight at 4°C. After that, the membranes were incubated with streptavidin–horseradish peroxidase (HRP) for 30 min followed by a Chemi Reagent mix. The signals on the membrane were detected either by X-ray film or iBright CL1500 (Thermo Fisher Scientific). The expression level of cytokines on the membrane was quantified by measuring the pixel density of spots using ImageJ. The value of each dot was normalized to the maximum value for each cytokine. With that, a ratio between 0 and 1 was calculated for each cytokine and used for heatmap generation using GraphPad Prism.

Total RNAs were extracted from mouse retinas or cultured cells using ZYMO Research (Irvine, CA, US) Quick-RNA Microprep Kit (Cat. No. R1051, Zymo Research) following the manufacturer’s instructions. RNA was reverse-transcribed using PrimeScript™ RT Master Mix [Cat. No. RR036A, Takara Bio (Kusatsu, Shiga, Japan)]. A mixture of master mix containing cDNA, 2× Master Mix from KAPA SYBR Fast qPCR kit (Cat. No. 07959397001, Kapa Biosystems, USA), and 10 μM of specific primers was used to detect specific gene expression using the Mastercycler ep realplex real-time PCR system (Eppendorf, Westbury, NY, USA). Relative gene expression level was presented in fold changes after being normalized to the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (Gapdh) using the ΔΔCt method. The results were presented as the relative fold change normalized to the naive retina or cells. The primers’ sequences are listed in Table 1 .

For immunostaining of cultured cells, BV2 mouse microglia were first seeded onto a Nunc^® Lab-Tek™ II Chamber Slide™ System (Cat. No. 154534PK, Thermo Fisher Scientific) and cultured overnight in DMEM (Cat. No. 11885084, Thermo Fisher Scientific) with 10% FBS and 50 ng/ml rmCSF1 (PeproTech, Cat. No. 315-02) at 37°C with 5% CO₂. After that, the samples were fixed in 4% paraformaldehyde (Sigma) for 10 min followed by permeabilization with 0.05% Triton X-100 in PBS for 10 min, and then blocked with 3% normal donkey serum [Cat. No. 017-000-121, RRID : AB_2337258, Jackson ImmunoResearch Labs (West Grove, PA, US)] for 1 h at room temperature. The cells were then incubated with primary antibodies, rabbit anti-pho-NFκB [1:500, Cell Signaling Technology, (Danvers, MA, US) Cat. No. 3033T, RRID : AB_331284] or rabbit anti-pho-PI3K (1:500, Cell Signaling Technology, Cat. No. 4228S, RRID : AB_659940). After washing three times with washing buffer (0.1% Triton X-100, 0.1% Tween-20 in PBS), they were probed by secondary antibodies including Cy2 donkey anti-rabbit (1:1,000, Jackson ImmunoResearch Labs, Cat. No. 711-225-152, RRID : AB_2340612) and Cy3 donkey anti-rabbit (1:1,000, Jackson ImmunoResearch Labs, Cat. No. 711-165-152, RRID : AB_2307443). The glass slides were mounted with Fluoroshield Mounting Medium with DAPI [Abcam, (Cambridge, UK) Cat. No. ab104139]. Imaging was taken by a Leica SP8 confocal microscope.

For immunostaining of tissue, mice were euthanized by inhalation of carbon dioxide. The eyeballs were collected and fixed in 4% paraformaldehyde overnight at 4°C. Retinas were then dissected out and blocked with Mojito buffer (10% normal goat serum, 3% normal donkey serum, 1% BSA, 0.5% Tween-20, 0.5% Triton X-100, and 0.1% sodium citrate buffer) for 2 h at room temperature. Then, the retinas were incubated with primary antibodies, including mouse anti-BRN3a [1:500, Millipore, (Burlington, MA, US) Cat. No. MAB1585, RRID : AB_94166] and rabbit anti-IBA-1 [1:250, Wako (Minato City, Tokyo, Japan), Cat. No. 019-19741, RRID : AB_839504], at 4°C overnight. After washing the retinas three times in washing buffer (0.1% Triton X-100 in PBS), they were incubated with biotin-conjugated antibodies at 1:250 dilution followed by streptavidin-conjugated Alexa Fluor secondary antibody. Biotin-conjugated antibodies used included goat anti-mouse biotinylated [1:250, Vector Laboratories (CA, US), Cat. No. BA-9200, RRID : AB_2336171]. After rinsing three times in washing buffer (0.3% Triton X-100 in PBS), they were incubated with secondary antibodies including streptavidin, Alexa Fluor 546 conjugate (1:1,000, Invitrogen (Waltham, MA, US), Cat. No. S11225, RRID : AB_2532130), and Cy2 donkey anti-rabbit (1:1,000, Jackson ImmunoResearch, Cat. No. 711-225-152, RRID : AB_2340612) at room temperature for 2 h. After washing three times with washing buffer, the glass slides were mounted with Fluoroshield Mounting Medium with DAPI (Cat. No. ab104139, Abcam). Imaging was taken by the Leica SP8 confocal microscope. For IBA-1 imaging in the whole-mounted retina, only the images of IBA-1⁺ microglia in BRN3a⁺ RGC layer were taken with the depth of around 6–8 μm. ImageJ software (National Institutes of Health, Bethesda, MD, USA) was used to quantify the number of BRN3a⁺ RGCs and IBA-1⁺ microglia/macrophage, as well as the measurement of the process length of microglia and cell body size. All quantification procedures were conducted by two researchers in a masked fashion.

BV2 cells from different groups were lysed with Pierce™ RIPA lysis and extraction buffer (Cat. No. 89900, Thermo Fisher Scientific) and supplemented with 1× Halt protease and phosphatase inhibitor (Cat. No. 1861261, Thermo Fisher Scientific). The samples were then sonicated on ice by a Q55 Sonicator (Qsonica, NY, USA; four pulses of 22 kHz, 5 s each at 20% power output). The lysates were centrifuged at 17,000×g for 5 min. The protein concentration of the lysates was determined using Qubit 4 Fluorometer (Thermo Fisher Scientific). A 30-μg protein was loaded per lane and separated by SDS-PAGE (4%–20% Mini-PROTEAN polyacrylamide gel, Cat. No. 4561096, Bio-Rad). The proteins were electrophoretically transferred to 0.45-μm pore nitrocellulose membranes by Trans-Blot Turbo Transfer System (Bio-Rad). The membranes were blocked with 3% BSA (A7096, Sigma-Aldrich) at room temperature for 1 h and then incubated overnight with the primary antibodies at 4°C, including rabbit anti-phospho-NFκB (1:1,000, Cell Signaling Technology, Cat. No. 3033T, RRID : AB_331284), rabbit anti-phospho-PI3K (1:1,000, Cell Signaling Technology, Cat. No. 4228S, RRID : AB_659940), and mouse anti-β-actin (1:4,000, Thermo Fisher Scientific, Cat. No. MA5-15739, RRID : AB_10979409). After washing with PBS-T buffer (PBS with 0.1% Tween 20) for three times, the blots were incubated with HRP-conjugated 1:2,000 secondary antibody in 3% BSA in PBS-T [Bio-Rad Laboratories (Hercules, CA, US), goat anti-rabbit IgG, Cat. No. 170-6515, RRID : AB_11125142; goat anti-mouse IgG, Cat. No. 172-1011, RRID : AB_11125936] for 1 h at room temperature. Signals were developed with ECL using a Super Signal West Pico kit (Thermo Fisher Scientific) and detected with iBright CL1500 (Thermo Fisher Scientific). The densitometric analysis was performed with ImageJ software using 8-bit grayscale positive chemiluminescent membrane images. All the quantification results were averaged from six independent blots per group and expressed as the mean ratio of the values (target protein/housekeeping protein) ± SEM.

Proteins were extracted from cultured BV2 cells as described (19). Briefly, cells were subjected to different treatments, including naive, LPS-treated BV2 for 6 h followed by the addition of vehicle or 10 μM baicalein. After a total of 48 h, cells were washed three times with PBS to remove FBS in the culture medium. Then, the cells were homogenized by 100 μl 5% SDS lysis buffer. Then, samples were centrifuged at 16,000×g for 30 min at 4°C. The supernatant was recovered for protein concentration measurement using Pierce™ Rapid Gold BCA Protein Assay (Cat. No. A53225, Thermo Fisher Scientific), according to the manufacturer’s protocol. Protein (50 µg) from each sample was reduced with 20 mM dithiothreitol (Cat. No. D0632, Sigma-Aldrich) and then alkylated with 40 mM iodoacetamide (Cat. No. I1149, Sigma-Aldrich) at room temperature in the dark for 10 min. The SDS lysate was acidified to a final concentration of 1.2% aqueous phosphoric acid. The solution was then diluted with the S-Trap protein binding buffer (90% methanol, 0.1 M TEAB, pH 7.55), at 6:1 (ratio, v/v), with S-Trap Mini Spin Column [ProtiFi(Farmingdale, NY, US), Cat. No. C02-mini-40]. Proteins were enzymatically digested with trypsin, 1:25 (w/w, trypsin:protein), according to the manufacturer’s protocol. The eluted peptides were subsequently dried up at 4°C in a vacuum centrifuge (Labconco, MO, USA), and the pellet was redissolved in 0.1% formic acid. The eluted peptide concentration was measured using colorimetric peptide assay (Cat. No. 23275, Thermo Fisher Scientific). Lastly, the peptides (2 µg) in each sample were subjected to nano-liquid chromatography tandem mass spectrometry (LC-MS/MS).

The hybrid quadrupole time-of-flight TripleTOF^® 6600 mass spectrometer (Sciex, MA, USA) was employed for target-free Sequential Window Acquisition of all Theoretical Mass Spectra (SWATH-MS) acquisition. SWATH-MS is a data-independent acquisition (DIA) that allows an in-depth, accurate, and consistent label-free quantification of proteins, particularly intracellular proteins. First, 2 µg tryptic peptides were pooled from all samples to establish protein identification and spectral library using data-dependent acquisition (DDA) (20). Then, 2 µg of peptide from each individual sample was injected into LC-MS/MS equipped with a SilicaTip electrospray emitter (New Objective, FS360-20-10-N-20-C12, 10 μm) handled by nano-flow liquid chromatography (NanoLC 415, Eksigent) for SWATH-MS acquisition. The raw data were collected and protein mapping was processed by the Paragon™ algorithms using ProteinPilot 5.0 (Sciex) for protein identification with a false discovery rate (FDR) ≤1%. Quantitative SWATH data were processed using PeakView 2.2 (Sciex) for spectral extraction followed by normalization and statistical analysis using MarkerView 1.3 (Sciex). The statistically significant differential proteins were evaluated with Welch’s t-test at P <0.05 with at least 2 quantifiable peptides per protein, represented as the mean of fold change of all biological samples (N = 6 per group).

The pathway analysis and upstream regulator prediction were performed by QIAGEN Ingenuity Pathway Analysis (IPA). Differentially expressed proteins (DEPs) between the LPS + Veh and LPS + Ba groups as defined by statistical significance (FDR < 0.01) were selected for gene set overrepresentation analysis by Mammalian Phenotype Ontology (MPO) (http://bioinfo.vanderbilt.edu/webgestalt/) (21) and functional enrichment analysis by Gene Ontology (GO), using the online tool of DAVID analysis (http://david.ncifcrf.gov/) (22).

Retinal ischemia was induced in adult mice unilaterally as previously described (5). Mice were anesthetized with a mixture of 120 mg/kg ketamine and 20 mg/kg xylazine in sterile saline (1:1:4). The pupil was dilated by 1% tropicamide eye drop followed by administrating a drop of topical anesthesia 0.5% proparacaine hydrochloride on the cornea. An easy entry hole for a glass micropipette was made by a 30-gauge needle at the peripheral cornea. The glass micropipette connected to a sterile saline bag by an extension tube was cannulated to the anterior chamber through the entry on the cornea. By elevating the saline bag to 1 m above the mouse eye level, the intraocular pressure (IOP) was raised to 75 mmHg. After 50 min of elevated IOP, the glass micropipette was withdrawn from the anterior chamber. Antibiotic ointment was applied on the cornea, and the anesthetized mice were left to recover on a warm pad. Mice were sacrificed at 1 and 4 weeks following I/R injury for qPCR and immunostaining of IBA-1⁺ microglia and at 4 weeks after I/R injury for immunostaining of BRN3a⁺ RGC and IBA-1⁺ microglia. For the quantification of microglia morphology changes, N = 4–5 mice per group; for qPCR measurements of mRNA levels of microglia activation markers and pro-inflammatory cytokines, N = 6 mice per group; for identifying T-cell population in retina or LNs using flow cytometry, N = 5 mice per group; and for RGC quantification and retinal function tests, N = 5–8 mice per group. All normal eyes used in this study were from naive mice.

Intravitreal administration of vehicle or baicalein was performed weekly post-I/R injury. The first administration was intravitreally given to mice immediately after the completion of retinal I/R model induction. Adult mice were anesthetized by i.p. injection of ketamine (120 mg/kg)/xylazine (20 mg/kg) mixture. A drop of proparacaine HCl (0.5%; Baush & Lomb Incorporated, Tampa, FL, USA) was applied to numb the eye. A hole was punctured on the limbus using a 30-gauge needle. Two microliters of Veh or 10 μM Ba was injected into the vitreous via a glass micropipette. Care was taken to avoid damaging the lens or retina. Antibiotic ointment was applied on the entry site on the limbus before leaving the anesthetized mice to recover on a heating pad.

T subsets as defined by their cytokine expression in the eye’s draining (superior cervical) lymph nodes (LNs) and retina were measured (10). For retinal cell flow cytometry, anesthetized mice were transcranially perfused with 15 ml saline. The retinas were dissected out and digested by papain. After incubating for 15 min at 37°C, papain digestion was stopped by ovomucoid protease inhibitor. Superior cervical LNs were dissected, and cells were mechanically dissociated using two forceps. Cell aggregates were separated by filtration through a 70-µm nylon cell strainer [Corning (Glendale, AZ, US), 431751]. For analyzing the frequencies of CD4⁺ T cells that expressed IFN-γ (Th1) and IL-17 (Th17), isolated lymphocytes were washed in IsoFlow (Beckman Coulter Inc., Brea, CA, USA) and stained with FITC-conjugated anti-mouse CD4 antibody (IgG2b, clone GK1.5, Biolegend, Cat. No. 100406, San Diego, CA, USA). Thereafter, cells were permeabilized with perm/wash buffer (Invitrogen, Cat. No. 00-8333-56) and stained with PE-labeled anti-mouse IFN-γ antibody (IgG1, clone XMG1.2, Biolegend, Cat. No. 505808, RRID : AB_315402) and PE-labeled anti-mouse IL-17A (IgG1, clone TC11-18H10.1, Biolegend, Cat. No. 506901, RRID : AB_315461) to detect Th1 and Th17 cells, respectively. The antibody-stained cells were analyzed with BD LSR II Flow Cytometer (BD Biosciences), and data were analyzed using FlowJo (FlowJo LLC, Ashland, OR, USA).

Mice were dark adapted for at least 8 h before the tests. Adult mice were anesthetized by i.p. injection of ketamine (120 mg/kg)/xylazine (20 mg/kg) mixture using a 25-gauge needle. Both pupils were dilated by topical application of 1% tropicamide followed by a drop of Genteal to keep the cornea moist. The mouse was placed on a heated platform kept at 37°C. The reference and ground electrodes were inserted beneath the skin over the forehead and tail base, respectively. Two gold-ring recording electrodes were gently placed on the center of the corneas with a drop of artificial tear (Genteal, Alcon) (Fort Worth, TX, US) covered. The positive scotopic threshold responses (pSTR) were recorded by averaging 40 responses following flash stimulation intensities at 1.7E−4 (cd s/m²). The amplitude of pSTR was measured from the baseline to the peak of the positive deflection. Data were processed by the software included in the electroretinography (ERG) recorder system (Espion Electroretinography System; Diagnosys LLC, Lowell, MA, USA). After the ERG recording, antibiotic ointment and artificial tear were applied to the mice cornea, and mice were left on a heated pad until recovery.

Optomotor response was recorded as previously described (23). Briefly, the mouse was placed on a small platform with stimuli presented on the computer monitors. The luminance of individual stripes was measured using a 371 R Optical Power Meter (Graseby Optronics, Orlando, FL, USA) positioned at the eye level. A constant speed of movement (black and white stripes) with adjustable contrast and the width of stripes on the monitors were presented. The contrast of a given spatial frequency was calculated using the formula (L _max − L _min)/(L _max + L _min), where L _max is the luminance (in cd/m²) of the white stripe, and L _min is the luminance of the black stripe (in cd/m²). For visual acuity measurements, spatial frequency was determined at a constant speed of 12 degree/s. For contrast sensitivity measurements, the spatial frequency was set to 0.186 cyc/degree at the same speed. The threshold of visual acuity and contrast sensitivity was determined based on the tracking of movements of mouse head in response to stimuli.

All statistical analysis was performed using GraphPad Prism for Windows, version 9.0 (GraphPad Software Inc, San Diego, CA, USA). Statistical analysis was conducted by either Student’s t-test or one-way ANOVA with Tukey’s multiple comparisons test. Two-way ANOVA with Šídák’s multiple comparisons test was used for comparison of multiple time points. P <0.05 was considered statistically significant. Data were expressed as mean ± SEM. N refers to the number of mice or the number of biological replicates in cell cultures. Each dot in each figure represented individual replicates.

As the primary resident immune cells in the CNS, the microglia play a critical role in surveilling and maintaining the homeostasis and functionality of neurons (24, 25). To determine if baicalein has direct effects on microglial homeostasis, we first investigated in LPS-primed mouse BV2 microglia cultures. To comprehensively examine cytokine inductions following LPS and baicalein treatment, we performed pro-inflammatory cytokine screening using cytokine arrays followed by validation at the mRNA levels using qPCR. BV2 cells were incubated with or without LPS (dissolved in PBS), followed by the addition of PBS control, 20% βcd (solvent for dissolving baicalein) as vehicle (Veh) control, or baicalein (Ba) ( Figure 1A ) and cultured for total 48 h. As expected, cytokine arrays of the supernatants detected drastically increased production of all pro-inflammatory cytokines in LPS-treated cultures compared with non-treated controls. Among LPS-primed groups, PBS and vehicle-treated cells showed similar induction of cytokines, suggesting minimal effects of vehicle on LPS-primed BV2 cells. In contrast, the addition of baicalein after LPS stimulation significantly suppressed the production of the most pro-inflammatory cytokines examined, such as IL-6, IL-1β, TNF-α, IL-17, and CXCL10, compared with the vehicle group ( Figure 1A ). The results of qPCR confirmed the induction of mRNAs of the corresponding pro-inflammatory cytokines by LPS and the inhibitory effects of baicalein in BV2 microglia ( Figure 1B ).

The effects of baicalein were also verified in the human microglia HMC3 cell line primed by LPS and ATP (17). After a total of 48 h of incubation, LPS and ATP stimulation induced increased production of pro-inflammatory cytokines, such as IL-1β, IL-17A, IL-1α, CXCL10, and CCL2 as shown by cytokine arrays, compared with the untreated control group ( Figure 1C ). The Veh or PBS-treated group showed a pattern of increased pro-inflammatory cytokine expression similar to that seen in the LPS and ATP-primed group ( Figure 1C ), confirming that vehicle control did not affect microglial activation. Notably, treatment with baicalein brought the majority of inflammatory cytokines nearly to their baseline levels ( Figure 1C ). The qPCR data confirmed that baicalein suppressed the mRNA levels of pro-inflammatory cytokines, such as Il-6, Il-1β, and Tnfα, in LPS and ATP-primed HMC3 cells ( Figure 1D ). These results revealed a direct anti-inflammatory and pro-homeostatic effect of baicalein on microglia.

Phosphorylation of PI3K and its downstream signal NFκB pathway has been reported to be key intracellular events associated with the pro-inflammatory responses of the microglia (26, 27). To determine if baicalein suppresses the PI3K/NFκB axis, we examined the phosphorylation of PI3K (pho-PI3K) and NFκB (pho-NFκB) in LPS-primed BV2 microglia with immunohistochemistry ( Figure 2A ). Because LPS-induced phosphorylation of PI3K and NFκB (28, 29) is most prominent at 3 h post-LPS stimulation, we pretreated BV2 microglia with vehicle or baicalein followed by LPS stimulation. Increased phosphorylation of PI3K and NFκB induced by LPS was detected in BV2 cells at 3 h post-LPS stimulation, whereas cultures pretreated with baicalein exhibited attenuated pho-PI3K and pho-NFκB levels after LPS priming compared with the LPS and Veh + LPS groups ( Figure 2A ). This result was confirmed by Western blot assays ( Figure 2B ). As expected, Western blot detected significant increases of pho-PI3K and pho-NFκB post-LPS stimulation in vehicle-treated BV2 cells (Veh + LPS) compared with controls ( Figures 2B-D ). In contrast, treatment with baicalein (Ba + LPS) significantly attenuated LPS-induced PI3K and NFκB phosphorylation compared with the Veh + LPS-treated group ( Figures 1B-D ). These results suggest that baicalein is capable of suppressing LPS-induced PI3K and NFκB signaling axis and serving as a pro-homeostatic agent in mouse microglia.

To comprehensively study the intracellular signaling changes and explore the underlying mechanisms of baicalein-mediated microglial responses, we employed the systemic proteomic approach using SWATH-MS acquisition. BV2 cells were treated with LPS followed by the addition of vehicle or baicalein, and they were cultured for a total of 48 h before being collected for proteomic analysis. Equal amounts of reduced peptides from each sample were ionized and electrosprayed into MS for quantification ( Figure 3A ). MS detected 2,445 unique proteins that were mapped to mouse gene IDs with 1% FDR. Among these proteins, we detected 742 DEPs between the LPS + Veh and untreated control cultures, at a cutoff of P <0.05 and FC >1.2 ( Figure 3B ). The top 10 upregulated DEPs detected in LPS + Veh cells included LCN2, ACOD1, PTGS2, CSTA2, SLC7a11, IL1rn, SQSTM1, UPP1, SERPINB2, and GBP2. Enriched GO analysis identified biological processes related to oxidation–reduction (e.g., PTGS2, PLOD3, SDHC), lipid metabolic process (e.g., IL-1rn, PTGS2), and immune system process (e.g., ACOD1, LCN2, TLR7) ( Figure 3B and Figure S1A ). Enriched MPO analysis showed LPS-induced abnormal phenotypes of adaptive and innate immunity and professional antigen-presenting cell (pAPC) physiology ( Figure S1B ). Among the 742 DEPs induced by LPS stimulation, 28 LPS-induced DEPs (P < 0.05 and FC > 1.2) were reversed by the addition of baicalein at 6 h after LPS stimulation ( Figure 3C ). Notably, 4 of the top 10 most significantly upregulated DEPs in the LPS + Veh group, namely, PTGS2, LCN2, SQSTM1, and SERPINB2, were significantly suppressed by baicalein ( Figure 3C ). GO biological processes also linked the baicalein-induced DEPs to oxidation–reduction, lipid metabolic and apoptotic processes, and inflammatory responses, suggesting the reversal of LPS-initiated processes by baicalein ( Figure S1C ). A similar suppression of innate immunity and pAPC was identified in the baicalein-treated group ( Figure S1D ). Canonical pathway analysis of baicalein (LPS + Ba)-induced DEPs performed by IPA further revealed their association with the endoplasmic reticulum (ER) stress and neuroinflammation, specifically through inhibition of the IL-17 pathway (with z-score = 2; Figure 3D ). IL-1β, TLR4, and chemical-kinase inhibitors (e.g., PD98059 and U0126) were predicted as upstream regulators induced by baicalein ( Figures 3E, F ). Significant downregulation of PTGS2, LCN2, SERPINB2, and SPP1 further linked the upstream regulator IL-1β to this regulation process ( Figure 3F ). These results, together with the observed inhibition of IL-1β and IL-17A expression as shown in the above cytokine array and qPCRs ( Figures 1A-D ), strongly support that baicalein inhibits LPS-induced activation of the microglia and IL-17 pathway, via IL-1β and TLR4 signaling.

To evaluate the anti-inflammatory effect of baicalein in microglia in vivo, we induced acute retinal I/R in mice followed by weekly intravitreal administration of vehicle or baicalein ( Figure 4A ). Our previous studies (5, 18) reported that acute retinal I/R leads to neuroinflammation characterized by microglial activation and subsequent infiltration of CD4⁺ T cells into the retina (5, 18). Consistent with our previous findings, we detected robust morphological changes of the microglia on the RGC layer. It featured increased numbers of IBA-1⁺ microglia/macrophages, enlargement of IBA-1⁺ cell body size, and shortened cellular processes at 1 week post-I/R insult ( Figures 4B–E , and Figure S2 ). At the mRNA level of retina post-I/R injury, we detected increased expression of activated microglial markers (including Iba-1, Ptgs2, Nox2, Tlr4, and Ym1, Figures 4F-J ) and pro-inflammatory cytokines (including Il-1β, Il-1α, Il-6, and Ccl2, Figures 4K–N ), compared with the naive group. Remarkably, baicalein treatment retained the morphological features of the resting microglia, as shown by a lower IBA-1⁺ cell density, smaller cell body size, and longer cellular processes in the retina at 1 and 4 weeks post-I/R insults ( Figures 4B–E , and Figure S2 ). In addition, baicalein significantly suppressed the expression of activated microglial markers including Iba-1, Ptgs2, Nox2, Tlr4, and Ym1 at 1 week after I/R insult, but not at 4 weeks post-injury ( Figures 4F–J ). Significant downregulation of pro-inflammatory cytokines, including Il-1β, Il-1α, Il-6, and Ccl2, was also revealed at 1 and 4 weeks post-I/R injuries ( Figures 4K–N ). These results suggest that weekly administration of baicalein effectively suppressed the I/R-induced retinal microglial activation and neuroinflammation.

Despite the eyes being regarded as an immune-privileged site, it has been shown that the compromise of the blood retinal barrier under stress allows for the recruitment of peripheral immune cells to the lesion site and triggers adaptive immune responses, ultimately aggravating neuronal death (5, 30). We previously reported that retinal infiltration of CD4⁺ T cells peaked at 2 weeks post-I/R injury (5, 18). In the present study, we showed that baicalein significantly suppressed Th17 (CD4⁺ IL-17⁺) cell infiltration into the retinas when examined at 2 weeks post-I/R injury compared with vehicle-treated retinas ( Figures 5A, B ). Moreover, since priming of T-cell responses usually occurs in the secondary lymphoid tissues, such as the draining LNs (10), we investigated the T-cell responses in the eye’s draining LNs and defined the proportions of CD4⁺ T-cell subsets in the superior cervical LNs after I/R injury. Compared with the vehicle-treated retinas with I/R injury, baicalein effectively suppressed the activation of Th17 (IL-17⁺) and Th1 (IFN-γ⁺) cells in the superior cervical LNs at 2 weeks post-I/R injury ( Figures 5C-F ). The results indicate that baicalein suppresses CD4⁺ T-cell infiltration into the retina, particularly Th17 cells, as well as their activation in the superior cervical LN after retinal I/R damage.

Lastly, we examined if baicalein presents a therapeutic effect on protecting retinal neurons and visual functions post-I/R injury. Surviving neurons were quantified by immunostaining of retinal flat mounts for a specific RGC marker, BRN3a, at 4 weeks post-I/R ( Figure 6A ). The RGC density in the I/R-injured mice was reduced to 2,369 ± 383 cells/mm² (39.3% loss), compared with that of the normal retina (3,901 ± 145 cells/mm²) ( Figures 6A, B ). Weekly administration of baicalein significantly rescued RGCs from I/R-induced injury, with 3,600 ± 137 cells/mm² (7.7% loss) RGC density ( Figures 6A, B ).

To investigate the effects of baicalein on RGC physiological function and spatial vision following I/R injury, ERG and optomotor response (OMR) assessments were conducted respectively at 1 and 4 weeks post-I/R injury. pSTR is a widely used indicator of RGC function as previously described (5, 31, 32). In vehicle-treated eyes, significant and progressive reduction of pSTR amplitude was detected ( Figures 6C, D ). In contrast, baicalein-treated eyes showed significant improvement on pSTR at 1 and 4 weeks after I/R injury ( Figures 6C, D ). Compared with the reduced contrast sensitivity and visual acuity in vehicle-treated mice following I/R injury, baicalein-treated mice exhibited significant improvements in both contrast sensitivity and visual acuity 4 weeks after I/R insults ( Figures 6E, F ). Taken together, these results indicated that weekly intravitreal injections of baicalein can effectively improve RGC survival and enhance RGCs’ physiological function as well as spatial vision.