Peiminine‐Induced Selective Autophagy of AIM2 Inflammasomes Rescues Cerebral Ischemic Injury

Abstract

Background

Inflammasome hyperactivation drives neuroinflammation following cerebral ischemia, yet pharmacological strategies inducing their selective clearance remain largely unexplored. Selective autophagy holds particular promise as a therapeutic strategy by eliminating detrimental substrates while avoiding the side effects of excessive autophagy. However, few compounds are currently available to achieve this selective regulation.

Methods

Transient middle cerebral artery occlusion was used to screen the active compounds from Fritillaria, combined with TTC staining and neurological scores. Postischemic apoptosis, inflammation, and autophagy were evaluated via immunofluorescence and immunoblotting. RNA sequencing of the ischemic penumbra revealed transcriptomic changes. Cell‐specific effects were confirmed using cell coculture and glial‐deficient transgenic mice. Compound‐inflammasome interactions were studied by biolayer interferometry and molecular dynamics.

Results

Peiminine demonstrates robust neuroprotection after cerebral ischemia. RNA sequencing and immunoblotting analyses revealed that peiminine markedly suppresses postischemic inflammation. Correspondingly, peiminine exerts its neuroprotective effect primarily through actions on microglia but not astrocytes, indicated by neuron–glia cocultures and transgenic mice models with targeted cell‐type depletion. Biolayer interferometry and molecular dynamics simulations indicated that peiminine selectively binds to AIM2, showing no interaction with NLRP1 (NLR family pyrin domain containing 1) and approximately 100‐fold higher affinity compared with NLRP3. Notably, peiminine specifically activates the selective autophagic clearance of AIM2 inflammasomes within microglia without increasing overall autophagic flux or affecting other inflammasomes. Selective blockage of autophagy in microglia abolished the neuroprotective effect of peiminine.

Conclusions

Peiminine exhibited robust neuroprotection against cerebral ischemia by selectively promoting the autophagic clearance of AIM2 inflammasomes to mitigate neuroinflammation.

Nonstandard Abbreviations and Acronyms

DEGs

differentially expressed genes

OGD‐R

oxygen–glucose deprivation and reoxygenation

tMCAO

transient middle cerebral artery occlusion

Research Perspective.

What Is New?

• Peiminine, a compound derived from Fritilliflora, exhibited robust neuroprotective effects against cerebral ischemia along with antineuroinflammatory activity.

• Peiminine mitigates neuroinflammation by selectively promoting the autophagic clearance of AIM2 inflammasomes within microglia, without increasing overall autophagic flux or affecting other types of inflammasomes.

What Question Should Be Addressed Next?

• Given the direct binding of peiminine with AIM2 and few impacts of peiminine on autophagosome biogenesis machinery, we hypothesize that peiminine may function as a molecular glue, enhancing the interaction between AIM2 and specific E3 ubiquitin ligase or other components of the autophagic machinery; further studies will likely need to apply proteomics approaches to comprehensively map protein interactome of peiminine, thereby fully unraveling its mechanism of action.

Stroke, a leading cerebrovascular disorder with limited effective treatments, represents a critical global health burden with high rates of mortality and disability. Emerging evidence underscores the pivotal role of neuroinflammation in ischemic pathophysiology, where sustained release of inflammatory mediators and damage‐associated molecular patterns exacerbates neuronal apoptosis and neurological deficits. ¹ , ² , ³ Mechanistically, damage‐associated molecular pattern‐triggered assembly of inflammasomes, which are multiprotein complexes orchestrating innate immunity, drives pathological progression through proteolytic activation of proinflammatory cytokines IL‐1β (interleukin‐1β) and IL‐18. ⁴ , ⁵ Therapeutic development has yielded various inflammasome‐targeting inhibitors, with several candidates progressing to clinical evaluation. ⁶ However, their clinical translation remains hampered by systemic immunosuppressive effects and off‐target sequelae. ⁷ , ⁸ This therapeutic impasse highlights the pressing demand for precise pharmacological interventions capable of selective inflammasome modulation. Most current inhibitors function by preventing the assembly of inflammasomes. ⁶ , ⁹ Promoting the degradation of inflammasomes may offer an alternative strategy to inhibit inflammation, which remains largely unexplored.

As an evolutionarily conserved lysosomal degradation pathway responsible for intracellular substrate clearance, autophagy serves as a critical homeostatic regulator under ischemic conditions. ¹⁰ , ¹¹ However, autophagy’s influence on ischemic injury is complex, acting as a double‐edged sword. On one hand, excessive autophagic flux can exacerbate ischemic damage. The overproduction of autophagic vesicles following ischemia may overwhelm lysosomal function, causing autolysosome accumulation within neurons and worsening synaptic dysfunction. ¹² , ¹³ This uncontrolled escalation of autophagic activity ultimately contributes to neuronal demise. ¹⁰ , ¹⁴ Conversely, selective autophagic clearance tend to offer a protective mechanism. Targeted removal of damaged mitochondria by mitophagy mitigates ischemic injury by reducing mitochondria‐released cytochrome c and subsequent apoptosis. ¹⁵ , ¹⁶ , ¹⁷ Similarly, pexophagy suppresses hypoxia‐induced oxidative stress by selectively eliminating peroxisomes. ¹⁸ , ¹⁹ These substrate‐specific clearance mechanisms achieve therapeutic benefits without inducing global autophagic overactivation, suggesting a strategic paradigm for ischemic intervention. Despite these findings, drug development is still limited by the lack of compounds that can selectively target specific types of autophagy without disrupting normal autophagic processes. ²⁰

Natural products play a crucial role as lead compounds in drug discovery. Fritillaria, widely used in traditional Chinese medicine, ²¹ is renowned for its anti‐inflammatory, antibacterial, and analgesic properties by suppressing NF‐κB (nuclear factor kappa B) signaling pathway. ²² , ²³ However, its impacts on both ischemic brain injury and autophagy remain underexplored. Here, we assessed major compounds derived from Fritillaria and surprisingly found that peiminine offered protection against ischemic brain injury with autophagic substrate selectivity. Specifically, peiminine mitigates neuroinflammation by selectively promoting the autophagic clearance of AIM2 inflammasomes within microglia, without increasing overall autophagic flux or affecting other types of inflammasomes. These results underscore the importance of selective AIM2‐inflammasome autophagy in cerebral ischemia and identify peiminine as a promising candidate for developing novel therapeutic strategies targeting ischemic stroke via selective autophagy.

METHODS

Data supporting the figures shown in this article are available on request from the authors.

Middle Cerebral Artery Occlusion Surgery

All animal experiments were carried out following the National Institutes of Health guidelines, and all procedures were approved by the local ethics review committee of Zhejiang Chinese Medical University (No. 15399). Personnel performing surgeries (middle cerebral artery occlusion [MCAO]), drug administrations (vehicle/ peiminine), and outcome assessments (infarct volume measurement, behavioral tests) were blinded to group assignments throughout the study. Mice were randomly assigned to experimental groups using a computer‐generated random number sequence produced in Microsoft Excel. Specifically, within each predefined body weight stratum (±2 g), a random number was generated for each mouse using the “RAND ()” function. The mice were subjected to 24 hours of food deprivation, after which they were anesthetized with a 1% solution of sodium pentobarbital. A small incision was performed in the common carotid artery, and a 6‐0 nylon monofilament suture (HUAWEI, 20142020178) was blunted to a hemispherical shape using a soldering pencil tip under microscopic guidance. The diameter of the hemispherical tip was measured using a digital micrometer screw gauge. Only filaments with a tip diameter falling within the range of 0.24 to 0.26 mm were used for model construction. The filaments were coated with 0.1% poly‐l‐lysine (Sigma‐Aldrich, Cat# P1399) overnight. The filament (actual length: 12 mm) was advanced into the internal carotid artery while continuous monitoring of regional cerebral blood flow was performed using laser Doppler flowmetry (moorVMS‐LDF2; probe position: anteroposterior= −2.0 mm, midline= +5.0 mm). Insertion was halted immediately upon observing a sharp regional cerebral blood flow drop (typically at ~10.0 depth from carotid bifurcation), consistent with successful MCA origin occlusion. A 70% reduction in blood flow, as indicated by the flowmeter readings, was considered a successful occlusion. Timing commenced upon the sharp regional cerebral blood flow drop, with filament withdrawal 1 hour later. Reperfusion was induced by removing the filament, then peiminine was administered into the lateral ventricle (anteroposterior= −0.5 mm, midline= +1.0 mm, dorsal‐ventral= −2.0 mm) along with wortmannin immediately (Selleck, S2758; 50 nM) and MG‐132 (Selleck, S2619; 10 μM).

Cell Culture and Oxygen–Glucose Deprivation Reoxygenation Treatment

Pregnant mice at 14 to 16 days of gestation were subjected to sterilization procedures, and the fetuses were extracted from the abdomen. The fetal heads were carefully trimmed, and brain tissues were dissected out. The cerebral cortices were stripped away, and the brain tissue was digested with trypsin at 37 °C for 30 minutes. Using a Pasteur pipette, the tissue was gently triturated to create a single‐cell suspension. The isolated cells were then cultured in Neurobasal medium (Gibco, 21103‐049) supplemented with B27 (Gibco, A1895601) and glutamine (Gibco, 35050‐061) for 7 to 10 days until neuronal maturity.

Primary microglia and astrocytes were obtained from newborn mice aged 0 to 3 days postpartum, following similar extraction steps as described for neurons. These glial cells were cultured in DMEM medium enriched with serum and allowed to mature over 10 to 14 days. Once mature, the neurons were seeded in the lower chambers of Transwell inserts, and different types of glial cells were placed in the upper chambers, allowing coculture for 24 hours.

BV2 (microglial cell line) and primary microglia cells were maintained in high‐glucose DMEM (Gibco, 11995065) containing 10% FBS (ExCell, FCS500) at 37 °C with 5% CO₂ for 24 hours. Following this, the cells were rinsed with glucose‐free DMEM (Gibco, 11966‐025) and transferred into hypoxic chambers containing 5% CO₂ and 95% N₂ for culturing at 37 °C for 4 hours. Subsequently, they were treated with peiminine with a purity of 99% (Yuanye, B21854), using concentrations of 10, 20, 40, and 80 μM.

After oxygen–glucose deprivation and reoxygenation (OGD‐R), BV2 cells were incubated in high‐glucose DMEM without FBS in a 37 °C incubator with 5% CO₂ for 18 hours.

Longa’s Score

Neurological deficits were scored 18 hours after ischemia induction using the following scale: (1) no neurological deficits: 0; (2) inability to extend the paralyzed forepaw completely: 1; (3) turning toward the paralyzed side while walking: 2; (4) falling to the paralyzed side during ambulation: 3; and (5) inability to walk spontaneously, accompanied by a loss of consciousness: 4.

TTC Stain

Mice were decapitated 18 hours after reperfusion. Their brains were sectioned into 1 mm slices and stained with a 0.25% solution of TTC (Sigma‐Aldrich, T8877). The brain slices were incubated in the TTC solution at 37 °C for 25 minutes, followed by fixation in 4% paraformaldehyde for 1 hour. Infarcted areas were analyzed using Image Pro Plus 6.0. The formula for calculating the percentage of cerebral infarction volume=(Sum of the uninfarcted areas on the contralateral hemisphere−Sum of the uninfarcted areas on the ipsilateral hemisphere)/Sum of the uninfarcted areas on the contralateral hemisphere×100%.

Modified Neurological Severity Score Rotarod Test

The mice were randomly divided into model group and peiminine group (Table1). For motor function assessment, the animals were placed on an accelerating rotarod apparatus, and the duration they remained on the rod was recorded. The speed of the rotarod gradually increased from 5 to 40 rpm within 2 minutes. Each trial concluded when the mice fell off the rod. Three days before transient MCAO (tMCAO) surgery, the animals underwent training sessions on the rotarod. Baseline performance was established as the average of 8 test runs conducted 1 day before tMCAO surgery, with each run separated by 20 minutes. Motor test data were presented as the mean duration that mice stayed on the rotarod across 8 trials.

Table 1.

Modified Neurological Severity Scores

Degree of nerve damage	Score
Motor tests
Raising rat by the tail	3
Flexion of forelimb	1
Flexion of hindlimb	1
Head moved >10° to vertical axis within 30 s	1
Placing rat on the floor (normal=0; maximum=3)	3
Normal walk	0
Inability to walk straight	1
Circling toward the paretic side	2
Falling down to the paretic side	3
Sensory tests	2
Placing test (visual and tactile test)	1
Proprioceptive test (deep sensation, pushing the paw against the table edge to stimulate limb muscles)	2
Beam balance tests (normal=0; maximum=6)	6
Balances with steady posture	0
Grasps side of beam	1
Hugs the beam and 1 limb falls down from the beam	2
Hugs the beam and 2 limbs fall down from the beam, or spins on beam (>60 s)	3
Attempts to balance on the beam but falls off (>40 s)	4
Attempts to balance on the beam but falls off (>20 s)	5
Falls off: no attempt to balance or hang on to the beam (<20 s)	6
Reflexes absent and abnormal movements	4
Pinna reflex (head shake when touching the auditory meatus)	1
Corneal reflex (eye blink when lightly touching the cornea with cotton)	1
Startle reflex (motor response to a brief noise from snapping a clipboard paper)	1
Seizures, myoclonus, myodystony	1

Maximum points 18, 1 is awarded for the inability to perform the tasks or for the lack of a tested reflex; 1–6, mild injury; 7–12, moderate injury; 13–18 indicates severe injury.

Desorption Electrospray Ionization Mass Spectrometry Imaging

Mass spectrometry imaging was performed using a Synapte XS Q‐TOF system equipped with a desorption electrospray ionization source. Before sample analysis, mass calibration was carried out in electrospray ionization mode using 200 ng/mL leucine enkephalin. Samples were analyzed in separate runs for positive and negative ion modes. The mass scanning range was set from 50 to 1200 m/z. Key instrumental parameters were as follows: ion source temperature, 120 °C; capillary voltage, 0.52 kV; spray solvent composed of methanol/water (98:2, v/v) containing 0.1% formic acid; nitrogen gas pressure, 0.7 MPa; solvent flow rate, 4 μL/min; spray incidence angle, 75°; spatial resolution, 50 μm; and scan speed, 200 μm/s. Following system setup and validation, tissue sections were scanned under the optimized conditions to acquire spatially resolved mass spectra across the sample surface.

TUNEL Stain

The brain frozen sections were fixed with 4% paraformaldehyde for 60 minutes and then washed twice with PBS for 10 minutes each time. Following this, the sections were incubated with PBS containing 0.5% Triton X‐100 at room temperature for 5 minutes to enhance permeabilization. After 2 washes with PBS, the TUNEL reaction mixture (Beyotime, C1086) was applied, and the sections were incubated at 37 °C in a dark environment for 60 minutes. Subsequently, the sections were washed with PBS 3 times, for 5 minutes each time. Finally, after applying an antifade mounting medium, the sections were examined under a fluorescence microscope.

Western Blotting

Tissue preparation: Add lysis buffer at a ratio of 1 mg of tissue to 10 μL of lysis buffer, then homogenize the mixture. Cell preparation: Seed cells in 6‐well plates. When collecting cells, gently rinse them with ice‐cold PBS and collect the cells into a centrifuge tube using a cell scraper. Centrifuge the aforementioned tissue or cell samples at 4 °C and 4000g for 2 minutes. Discard the supernatant and resuspend the cell pellet in RIPA lysis buffer. After lysing on ice for 20 minutes, centrifuge again at 4 °C and 12 000g for 20 minutes, collect the supernatant, and determine the protein concentration in the lysate using the BCA Protein Assay Kit (Beyotime, P0012). For SDS‐PAGE electrophoresis, load 40 μg of protein from each sample group. After separating the proteins, transfer them onto a nitrocellulose membrane. Block nonspecific binding sites with PBS containing 5% skim milk, then incubate with the primary antibody. The antibodies used in this study include cleaved‐caspase‐3 (CST, 9661), caspase‐3 (CST, 9662), caspase‐1 (Abclone, A0964), IL‐1β (CST, 63124), IL‐18 (CST, 57058), NLRP1 (NLR family pyrin domain containing 1; Affinity, DF13187), AIM2 (CST, 63660), NLRP3 (CST, 15101), NLRC4 (NLR family CARD domain‐containing protein 4; Abclone, A24336), p‐ULK1 (p‐Unc‐51 like autophagy activating kinase 1; Affinity, AF2301), ULK (CST, 8054), Beclin (Abclone, A21191), p‐AMPK (p‐AMP‐activated protein kinase; CST, 2535), AMPK (CST, 5831), p‐Akt (p‐protein kinase B; CST, 4060), Akt (Abclone, A18120), Atg5 (Abclone, A19677), Atg7 (autophagy‐related 7; Abclone, A19604), ASC (CST, 67824), SQSTM1 (sequestosome 1; MBL, PM045, China), LC3‐B (Sigma, L7543, USA), and β‐actin (Abclone, AC026, China).

Immunofluorescence

Frozen sections or cell slides were washed 3 times with PBS for 5 minutes each time. They were then permeabilized with PBS containing 0.3% Triton X‐100 at room temperature for 15 minutes. After discarding the permeabilization solution, the slides were blocked with PBS containing 1% BSA and 5% donkey serum at room temperature for 1 hour. Following the removal, the blocking solution, the slides were incubated with the primary antibody at 4 °C overnight. The next day, the primary antibody was removed, and the slides were washed 3 times with PBS, each washing lasting 5 minutes. The slides were then incubated with the secondary antibody at room temperature for 2 hours, followed by washes with PBS. Finally, the slides were mounted with an antifade mounting medium and observed under a fluorescence microscope. The antibodies used in this study include NeuN (ABclonal, A19086), Iba1 (ionized calcium‐binding adaptor molecule 1; CST, 17198), GFAP (glial fibrillary acidic protein; CST, 34001).

Virus Injection

The mice were anesthetized with 1% sodium pentobarbital and then secured in a stereotaxic instrument. The height of the anterior and posterior fontanelles was adjusted to ensure proper alignment. The coordinates for the target brain regions were as follows: anteroposterior= −0.5 mm, midline= +1.0 mm, dorsal‐ventral= −4.7 mm. Using a microsyringe pump, virus injections were performed with a single injection volume of 200 nL at 1 mm intervals upward, for a total of 4 injections. Each injection lasted 5 minutes, followed by a 5‐minute pause before the next injection. Lentivirus (LV‐eflα‐DIO‐taCasp3‐T2A‐TEVp‐P2A‐mCherry‐WPREs) was used to specifically infect microglia, and adeno‐associated virus (AAV2/5‐GFaABC1D‐taCasp3‐P2A‐WPREs and AAV2/5‐GFaABC1D‐taCasp3‐T2A‐TEVp‐WPREs) was used to target astrocytes. After 2 weeks for viral expression, the corresponding model or behavioral tests were conducted. Post experiment, sections were examined to verify the accuracy of the injection site, and western blot analysis was used to detect the expression of the corresponding proteins.

Cell Viability

MTT assay was performed to detect the cell viability. Primary neurons were cultured 7 days before treatment in 96‐well plate at 1×10⁵ cells per well at 37 °C. OGD‐R treatment was performed for 2 hours, and drug administration was performed upon reoxygenation and continued for 18 hours. MTT (5 mg/mL, Sigma, M2128, USA) was added to the well plate, shake the plate slightly to mix the liquid thoroughly and incubated at 37 °C for 4 hours. After the incubation, the supernatant was removed and 150 μL DMSO (LEAGENE, CC0188, China) was added to each well. The plates were shaken slowly for 10 minutes, and measured at 570 nm.

Toluidine Blue Staining

The frozen sections were fixed in 4% paraformaldehyde for 30 minutes, then washed with PBS for 5 minutes. Following the wash, the sections were immersed in a 1% toluidine blue staining solution (Aladdin, T104232) for 18 minutes. They were then rinsed in ddH₂O for 1 minute and differentiated in 75% ethanol for 1 minute. Subsequently, the sections were soaked twice in 85% ethanol for 2 minutes each time, followed by two immersions in 95% ethanol for 2 minutes each. This was followed by two 2‐minute immersions in absolute ethanol. The sections were then placed in xylene for 5 minutes. Finally, the sections were sealed using a mixture of neutral gum and xylene (1:1 ratio) and observed under a microscope.

Correlative Light Microscopy‐Electron Microscopy

BV2 cells were cultured in gridded glass bottom dishes. The AIM2‐GFP (green fluorescent protein) and mCherry‐LC3 plasmids were transfected into the cells as described above, and live‐cell images were captured by laser scanning confocal microscopy at 24 to 48 hours post‐transfection. Subsequently, the cells were fixed with a mixture of paraformaldehyde and glutaraldehyde. After 24 hours of fixation, the samples were prepared following the electron microscopy sample preparation protocol, and ultrathin slices were taken using a scanning electron microscope according to the grid coordinates.

Transfection

Transfection of BV2 cells was performed using liposome‐mediated transfection with the jetPRIME reagent (Polyplus, 101000046). Cells were seeded in 6‐well plates. After 24 hours to allow for cell adherence, a transfection mixture was prepared as follows: add 200 μL of transfection buffer to a nuclease‐free centrifuge tube; add 2 μg of plasmid DNA, vortex for 10 seconds, and briefly centrifuge; add 4 μL of jetPRIME transfection reagent, vortex again for 10 seconds, and briefly centrifuge. Allow the mixture to stand at room temperature for 10 minutes. The transfection mixture was then added dropwise to the cells. After 4 hours, the culture medium was replaced with fresh medium containing 5% FBS.

RNA‐Sequencing

Total RNA was extracted from ischemic hemisphere using Total RNA Kit I (Omega, USA) following the manufacturer’s instructions. The raw data were first processed with FastQC to filter out adapters and low‐quality sequences. Then, the clean reads were mapped to the “Mus musculus.GRcm39” genome using Star. Differential gene expression analysis was conducted using DESeq2 to identify genes differentially expressed between the Veh group and the peiminine group. The gene expression levels in stressed samples were compared with those in control samples in order to identify the differentially expressed genes (DEG) by DESeq2. The DEGs were defined by the criteria of an adjusted P value (Padj) <0.05 and an absolute log2 fold change (|log2FoldChange|) >1. Kyoto Encyclopedia of Genes and Genomes enrichment analyses for the DEGs were performed using the clusterProfiler version 3.8.

Molecular Docking

The 3‐dimensional structure of AIM2, which was obtained from the RCSB Protein Data Bank (PDB IDs: 3rn5 and 3vd8), was determined using X‐ray crystallography and used as the receptor. The 3‐dimensional structure of peiminine was downloaded from the PubChem database (CID: 167691). During molecular docking, up to 15 binding poses were generated for each run. The energy difference between the lowest and highest binding energy poses was set to 2 kcal/mol. The docking mode employed was semiflexible docking.

Molecular Dynamic

Molecular dynamics simulations were conducted using GROMACS version 2023.2. The initial structure of the complex was input into the Solution Builder module of CHARMM‐GUI to construct a solvent box with at least 10 Å distance from the complex boundary. The system was solvated with Transferable Intermolecular Potential with 3 Points (TIP3) water molecules, and Na⁺ and Cl⁻ ions were added to achieve a physiological ion concentration of 0.15 mol/L, balancing the overall charge of the system. Periodic boundary conditions were applied, and particle mesh Ewald‐fast Fourier transform grid information was automatically generated. CHARMM36m force field parameters were used to generate topology files for both the proteins and small molecules. Energy minimization was performed using the steepest descent method for 5000 steps, ensuring that the maximum force (emtol) on any atom in the system fell below 1000 kJ/mol per nm. Long‐range electrostatic interactions were calculated using the particle mesh Ewald method. Following energy minimization, a 125 ps preequilibration (NPT (the number of particles (N), pressure (P), and temperature (T) of the simulated system are all kept constant) ensemble) was conducted at a fixed volume and constant temperature to achieve thermal equilibrium. The final molecular dynamics simulation spanned 100 ns with a time step of 2 fs. Once the system stabilized, the protein‐ligand intermolecular binding free energy was calculated using the molecular mechanics Poisson–Boltzmann surface area method.

Quantitative Polymerase Chain Reaction

Pretreated cells were placed on ice, and lysis buffer was added (Beyotime, R0026). RNA extraction was performed according to the manufacturer’s instructions. Reverse transcription was carried out using MonScript RTIII (Monad, MR05101). Primers were designed, and real‐time polymerase chain reaction was performed using the CFX Connect Real‐Time PCR Detection System.

Quantitative polymerase chain reaction experiments were conducted in triplicate. The primer sequences used were as follows:

• Aim2‐F: GTCACCAGTTCCTCAGTTGTG

• Aim2‐R: CACCTCCATTGTCCCTGTTTTAT

• Gapdh‐F: AGGTCGGTGTGAACGGATTTG

• Gapdh‐R: TGTAGACCATGTAGTTGAGGTCA

Biolayer Interferometry

Biolayer interferometry experiments were conducted using an Octet R8 instrument at 25 °C in PBS containing 0.02% Tween‐20. antinitrilotriacetic acid biosensors were preequilibrium 10 minutes in buffer, then loaded 10 ug/mL of protein NLRP1, AIM2, and NLRP3 for 10 minutes. Equilibrium dissociation constant were determined from binding data obtained with at least 6 concentrations of peiminine. The data were analyzed using the global fitting algorithm provided by Octet Analysis Studio 13.0.3.52.

Statistical Analysis

All data were collected and statistically analyzed in a double‐blind manner. Data are presented as the mean±SEM. To analyze differences among 3 or more groups, 1‐way ANOVA was used, followed by Tukey’s multiple comparison test for post hoc analysis. Statistical comparisons between 2 groups were performed using an unpaired Student’s t test. Survival curve comparison was used to compare the results of survival rates with log‐rank test. P<0.05 was considered to indicate statistical significance.

RESULTS

Peiminine Confers Neuroprotective Effect Against Cerebral Ischemia

To evaluate the therapeutic potential of Fritillaria in cerebral ischemia, we tested its main alkaloids (including peimine, peimisine, and peiminine) using a tMCAO mouse model. Among the compounds tested, only peiminine exhibited significant neuroprotective effects, as evidenced by reduced infarct volumes (Figure 1A). Further dose–response analyses identified 3 μg as the optimal dose of peiminine, which significantly mitigated both acute and chronic ischemic symptoms, as demonstrated by improved neurobehavioral scores, Modified Neurological Severity Score and rotarod test performance, as well as reduced postischemic mortality (Figure 1B through 1D). To assess the diffusion range of these chemicals in the brain parenchyma, we performed desorption electrospray ionization mass spectrometry imaging 24 hours post injection. We found that peimine showed limited diffusion into the parenchyma, whereas peimisine and peiminine diffused broadly and persisted for >24 hours in the parenchyma, encompassing the ischemic penumbra (Figure S1). Furthermore, peiminine’s neuroprotective effects were sustained for up to 7 days post administration shown by the improved performance in the rotarod test and increased survival rate (Figure 1C and 1D). Consistently, peiminine reduced ischemia‐induced apoptosis, corroborated by reduced TUNEL‐positive cells in ischemic tissue and decreased cleaved caspase‐3 expression (Figure 1E and 1F). Collectively, these findings demonstrate peiminine’s robust neuroprotective effects in ischemic injury.

Figure 1. Peiminine confers neuroprotective effect against cerebral ischemia.

Mice were subjected to tMCAO. Peimine, peimisine, or peiminine were intracerebroventricularly injected upon reperfusion. A, Representative brain slices with TTC staining at 24 h post‐tMCAO. Columns display infarct volumes as determined by TTC staining. B, Columns display neurological performance as determined by Longa scoring. Curves display neurological performance as determined by Modified Neurological Severity Score. C, Curves display motor coordination as determined by rotarod test. D, Survival curves within 7 d post tMCAO. n=8–11 in each group. E, Representative brain slices after TUNEL and NeuN staining. Scale bar, 10 μm. F, The expression level of cleaved‐caspase‐3 and total‐caspase‐3 in the penumbra were determined by western blot (n=4 mice per group). Data are expressed as mean±SEM from at least 3 independent experiments. Casp3 indicates caspase‐3; PM, peimine; PMN, peiminine; PMS, peimisine; and tMCAO, transient middle cerebral artery occlusion.

Peiminine Inhibits Neuroinflammation After Ischemia

To investigate the underlying mechanisms of the protective effects of peiminine, we performed transcriptomic profiling of ischemic penumbra tissues in tMCAO models through RNA sequencing. The results demonstrated significant changes in the brain transcriptome induced by peiminine compared with the tMCAO group, with 161 DEGs identified (90 upregulated and 71 downregulated). Kyoto Encyclopedia of Genes and Genomes pathway analysis revealed that the majority of these DEGs were associated with inflammatory processes (Figure 2A). Consistently, the heatmap analysis displayed significantly increased expression of many anti‐inflammatory genes (eg, Tnfaip3, Nfkbid, Il1r2, Siglece, and Nfkbiz) (Figure 2B). To further validate the sequencing results, we assessed inflammatory markers following ischemia using tMCAO mouse model and OGD‐R. Immunofluorescence staining revealed that peiminine administration upon reperfusion significantly inhibited glial activation including both microglia and astrocytes (Figure 2C), further suggesting a potential anti‐inflammatory effect of peiminine treatment. Immunoblotting analysis further confirmed these findings, demonstrating that peiminine treatment suppressed the expression of key proinflammatory markers, including cleaved‐caspase‐1, cleaved‐IL‐1β, and cleaved‐IL‐18 (Figure 2D). These findings suggest that the protective effect of peiminine may be related to its ability to modulate inflammatory processes within the brain. Correspondingly, peiminine did not directly promote neuronal survival in cultured cortical neurons following OGD‐R (Figure 3A), as neurons are not the primary cell type responsible for inflammatory regulation in the brain.

Figure 2. Peiminine inhibits neuroinflammation after ischemia.

Mice were subjected to tMCAO and peiminine was intracerebroventricularly injected upon reperfusion. A, Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis of potential targets. B, The heatmap displays the expression levels of genes identified as significantly enriched in proinflammatory and anti‐inflammatory gene set. C, Representative images showing microglial and astrocyte morphology after ischemia. Scale bar, 50 μm. Columns display the degree of microglial and astrocyte activation as determined by the suprathreshold GFP‐positive area in the cortex. n=4 mice per group. D, The expression levels of cleaved‐caspase‐1, cleaved‐IL‐18, and cleaved‐IL‐1β in the penumbra were determined by western blot (n=4 mice per group). Data are expressed as mean±SEM from 3 independent experiments. Casp3 indicates caspase‐3; GFP, green fluorescent protein; GFAP, glial fibrillary acidic protein; IL‐1β, interleukin‐1β ; JAK‐STAT, Janus kinase/signal transducer and activator of transcription; PMN, peiminine; and tMCAO, transient middle cerebral artery occlusion.

Figure 3. Peiminine depends on microglia to confer neuroprotective effect.

A–C, Cells were exposed to OGD‐R for 2 h, and treated with vehicle (PBS) or peiminine upon reoxygenation. A, Neurons were treated with peiminine (10, 20, 40, and 80 μM) upon reoxygenation. Cell viability of neurons was determined by MTT assay. B, Neurons were cultured in the lower chamber and the indicated types of cells were cocultured in the upper transwell inserts. Peiminine (20 μM) was added in the upper transwell upon reoxygenation and cell apoptosis in the lower chamber was assessed by TUNEL staining. Scale bar, 50 μm. Columns represent the proportion of TUNEL‐positive (TUNEL+) cells in the lower chamber. C, The expression level of cleaved‐caspase‐3 and total caspase‐3 in the lower chamber at 18 h post reoxygenation was determined by western blot (n=4 mice per group). D, Timeline of virus injection, surgery, and drug injection. E, Mice were subjected to tMCAO and peiminine was intracerebroventricularly injected upon reperfusion. LV‐eflα‐DlO‐taCasp3‐P2A‐mCherry‐WPRE (DIO‐mCherry) and LV‐eflα‐DIO‐taCasp3‐T2A‐TEVp‐P2A‐mCherry‐WPREs (DIO‐mCherry‐Casp3) were injected into the right hemispheres of Cx3cr1‐Cre ^ERT2 transgenic mice to selectively deplete microglia in the brain. n=8 mice per group. F, AAV2/5‐GFaABC1D‐taCasp3‐P2A‐WPREs (GFa‐mCherry) and AAV2/5‐GFaABC1D‐taCasp3‐T2A‐TEVp‐WPREs (GFa‐mCherry‐Casp3) were injected into the right hemispheres of wild‐type mice to selectively deplete astrocytes in the brain. N=8 mice per group. Scale bar, 200 μm. Representative brain slices with toluidine blue staining. Columns display infarct volume as determined by toluidine blue staining. Data are expressed as mean±SEM from 3 independent experiments. Casp3 indicates caspase‐3; OGD‐R, oxygen–glucose deprivation reoxygenation; PMN, peiminine; tMCAO, transient middle cerebral artery occlusion; veh, vehicle; and WT, wild type.

Peiminine Depends on Microglia to Confer Neuroprotective Effect

The brain’s response to ischemic injury involves a complex interplay among various cell types, including neurons, glial cells, and other cells. ¹ We hypothesized that peiminine’s protective effects are not directly mediated through neurons but instead rely on interactions with glial cells or other cells, which in turn exerts indirect neuroprotective effects. To test this hypothesis, we used a Transwell coculture system, pairing neurons in the lower chamber with different cell types in the upper chamber. The results showed that peiminine exhibited significant suppressive effect on neuronal apoptosis only under conditions of microglia coculture. By contrast, peiminine failed to exert neuroprotection when neither astrocytes, macrophages, nor neurons were cocultured in the upper chamber (Figure 3B and 3C). To validate this observation in vivo, we injected caspase‐3‐expressing virus (LV‐eflα‐DIO‐taCasp3‐T2A‐TEVp‐P2A‐mCherry‐WPREs) into the right hemispheres of Cx3cr1‐Cre ^ERT2 transgenic mice to selectively deplete microglial cells in the brain. The results showed that eliminating microglia completely abolished peiminine’s protective effects (Figure 3E). By contrast, the ablation of astrocytic populations in the brain by injecting AAV2/5‐GFaABC1D‐taCasp3‐T2A‐TEVp‐WPREs into the right hemispheres of wild‐type mice failed to affect the neuroprotective effect conferred by peiminine (Figure 3F). Taken together, these results suggest that peiminine depends more on microglia to confer neuroprotective effect.

Peiminine Selectively Binds to AIM2 and Induces the Degradation of AIM2 Inflammasomes

To investigate the molecular mechanism underlying the anti‐inflammatory effect of peiminine, we initially focused on its impact on key inflammasomes, which act as molecular “switches” in inflammatory cascades. ²⁴ , ²⁵ Biolayer interferometry was used to detect the molecular interactions between peiminine and various inflammasome components (NLRP1, AIM2, and NLRP3) in the brain (Figure 4A). His‐tagged NLRP1, AIM2, and NLRP3 were expressed and immobilized onto the nitrilotriacetic acid sensor. Following this, a concentration gradient series of peiminine solutions was introduced to assess binding characteristics. The results showed distinct binding affinities among the tested interactions. Specifically, peiminine exhibited robust concentration‐dependent binding to AIM2, characterized by an equilibrium dissociation constant of 4.6 μM. In contrast, peiminine binds to NLRP3 with an equilibrium dissociation constant value ∼100 times higher than that observed for AIM2, indicating a much weaker interaction. Notably, peiminine displayed negligible binding to NLRP1 under the same experimental conditions (Figure 4B). To further verify the interaction between peiminine and AIM2 inflammasome, we conducted molecular docking and molecular dynamics simulations. Analysis of the AIM2 (PDB ID: 3vd8) model revealed strong binding energies between the peiminine and AIM2, achieving a binding energy of −6.8 kcal/mol. The primary binding sites involved in this interaction were identified as GLU291, LYS295, ARG316, and LEU311 (Figure 4C). The results indicated that the carbonyl group in peiminine can form a strong hydrogen bond with the protein, thereby contributing to its protective role. In parallel tests of similar compounds, we found that other alkaloids lacking carbonyl groups did not exhibit protective effects following ischemia (Figure 1A). The results of the molecular dynamics simulations indicated that the docking site of peiminine established a stable, long‐term noncovalent interaction with AIM2, and the protein’s conformation remained relatively stable throughout the simulation (Figure 4D; Figure S2A and S2B). Consistent with the molecular docking results, the root mean square fluctuation analysis revealed that amino acids 275 to 285 and 302 to 325 may frequently interact noncovalently with the small molecule (Figure S2B). The radius of gyration analysis showed that the spatial variation range of the protein was very small (Δradius of gyration<0.2 nm), indicating that the protein remained in a relatively conservative folding state (Figure S2C). During the entire simulation process, 1 to 2 hydrogen bonds were intermittently formed between the small molecules and the protein (Figure S2D). These interactions were conducive to the binding of ligands and receptors and helped maintain the conformational stability of the system. The solvent accessible surface area results indicated that the complex remained stable after 25 ns (Figure S2E). Molecular mechanics Poisson–Boltzmann surface area analysis revealed that the total molecular binding free energy was −92.998 kJ/mol. Specifically, the electrostatic energy and van der Waals energy were −110.975 and −0.031 kJ/mol, respectively. These findings suggest a strong noncovalent interaction between the molecules, particularly driven by the electrostatic potential energy (Figure S2F). Therefore, peiminine demonstrated a high potential to bind within the AIM2’s docking pocket. To further validate the impact of peiminine binding to AIM2, we assessed AIM2 expression in both tMCAO animal models and OGD‐R cell models. Immunoblotting analysis revealed that peiminine specifically reduced AIM2 levels both in vivo and in vitro after ischemia, without significantly affecting NLRP1, NLRP3, or NLRC4 inflammasomes (Figure 4E and 4F; Figure S3A and S3B). To elucidate the mechanism behind the reduction in AIM2 levels, we performed real‐time polymerase chain reaction analyses, which revealed no changes in AIM2 mRNA expression (Figure S4). Thus, the decrease in AIM2 involves posttranslational mechanisms.

Figure 4. Peiminine selectively binds to AIM2 and induces the degradation of AIM2 inflammasomes.

The interaction between AIM2 inflammasome and peiminine. A, Schematic representation of inflammasome structures and binding interactions of peiminine assessed by BLI. The inflammasomes were NLRP1 (PDB ID: 6x6c), AIM2 (PDB ID: 3vd8), and NLRP3 (PDB ID: 6npy). B, Representative BLI data for NLRP1, AIM2, and NLRP3 inflammasomes binding to peiminine are shown. C, Binding mode of AIM2 and peiminine from cocrystal structure. D, Snapshots of the coassembly process for AIM2 and peiminine molecules in water at waiting times of 0, 10, 20, 40, and 60 ns. E, Mice were subjected to tMCAO and peiminine (3 μg) was intracerebroventricularly injected upon reperfusion. The expression levels of indicated proteins in the penumbra were determined by western blot (n=4 mice per group). F, Microglia cells were subjected to OGD‐R and treated with peiminine (20 μM) upon reoxygenation. The expression levels of indicated proteins at 18 h post reoxygenation was determined by western blot (n=4). BLI indicates biolayer interferometry; Casp3, caspase‐3; Ctrl, control; NLRP1, NLR family pyrin domain containing 1; NLRP2, NLR family pyrin domain containing 2; NLRP3, NLR family pyrin domain containing 3; NTA, nitrilotriacetic acid; OGD‐R, oxygen–glucose deprivation reoxygenation; and PMN, peiminine.

Peiminine Induces Selective Autophagy of AIM2 Inflammasomes

Given that protein degradation primarily occurs via the ubiquitin‐proteasome system or lysosomal pathways, ²⁶ we investigated whether either pathway was responsible for AIM2 degradation. We treated cells with the autophagy inhibitor wortmannin and the proteasome inhibitor MG‐132. The inhibition of autophagy, but not the proteasome pathway, led to AIM2 accumulation, indicating that AIM2 is degraded through the autophagy pathway (Figure 5A). To further confirm the activation of AIM2‐inflammasome autophagy, we used correlated light and electron microscopy. Autophagic vesicles and AIM2 inflammasomes were fluorescently labeled, and scanning electron microscopy revealed AIM2 colocalizing with double‐membrane autophagic vesicles, indicating selective autophagic degradation of AIM2 inflammasome (Figure 5B). Consistently, peiminine treatment significantly enhanced the colocalization of AIM2 with autophagic vesicles label by AIM2‐GFP and mCherry‐LC3 (Figure 5C). Notably, the total number of LC3 puncta was unaltered (Figure 5C), which is consistent with unaltered LC3‐II expression by peiminine (Figure 5D and 5E). Similarly, peiminine also failed to affect the expression of autophagosome initiation‐related proteins, including pULK1, Beclin, AMPK, pAkt, ATG5, and ATG7 (Figure S5). Taken together, these results suggest that peiminine activates selective AIM2‐inflammasome autophagy without affecting enhancing autophagy flux.

Figure 5. Peiminine induces selective autophagy of AIM2 inflammasomes.

A, Mice were subjected to tMCAO and peiminine (3 μg), wortmannin (50 nM), and MG‐132 (10 μM) were intracerebroventricularly injected upon reperfusion. The expression levels of indicated proteins in the penumbra were determined by western blot (n=4 mice per group). B, C, BV2 cells were transfected with mCherry‐LC3 and AIM2‐GFP plasmids and subjected to OGD‐R. Peiminine (20 μM) was treated upon reoxygenation. B, mCherry and GFP fluorescence were detected by confocal microscopy and the labeled regions were imaged by scanning electron microscopy subsequently. Representive images taken by confocal microscopy and scanning electron microscopy are shown. Scale bar, 10 μm. ROI, scale bar, 400 nm. C, Representative fluorescent images showing autophagosomes (red) and AIM2 inflammasomes (green) in BV2 cells. Scale bar, 10 μm. Columns display the number of LC3 puncta per cell, the number of AIM2 inflammasomes per cell, and the percentage of AIM2‐GFP puncta colocalized with mCherry‐LC3 relative to total AIM2‐GFP. D, Mice were subjected to tMCAO and peiminine (3 μg) was intracerebroventricularly injected upon reperfusion. The expression level of LC3‐II and p62 in the penumbra were determined by western blot (n=4 mice per group). E, Microglia cells were subjected to OGD‐R and treated with peiminine (20 μM) upon reoxygenation. The expression level of LC3‐II and p62 in the penumbra were determined by western blot (n=3). F, AAV2/9‐CBG‐DIO‐EGFP‐miR30shRNA(NC)‐WPRE and AAV2/9‐CBG‐DIO‐EGFP‐miR30shRNA(Atg7)‐WPRE (DIO‐GFP‐shAtg7) were injected into the right hemispheres of Cx3cr1‐Cre ^ERT2 transgenic mice to selectively deplete Atg7 in microglia. Scale bar, 20 μm. Representative brain slices with toluidine blue staining. Columns display infarct volume as determined by toluidine blue staining (n=8). Data are expressed as mean±SEM from 3 independent experiments. Ctrl indicates control; GFP, green fluorescent protein; OGD‐R, oxygen–glucose deprivation reoxygenation; PMN, peiminine; ROI, region of interest; tMCAO, transient middle cerebral artery occlusion; and WOT, wortmannin.

To determine whether peiminine’s protective effects depend on microglial inflammasome autophagy, we injected AAV2/9‐CBG‐DIO‐EGFP‐miR30shRNA (Atg7)‐WPRE in the right hemisphere of Cx3cr1‐Cre ^ERT2 transgenic mice to selectively depleting the autophagy‐related gene Atg7 in microglia (Figure 5F). Toluidine blue staining revealed that selective autophagy deficiency in microglia abolished peiminine’s protective effects (Figure 5F). These findings establish that peiminine exerts anti‐ischemic effects by selectively activating AIM2‐inflammasome autophagy in microglia.

DISCUSSION

In this study, peiminine, a compound derived from Fritilliflora, exhibited robust neuroprotective effects against cerebral ischemia with autophagic substrate selectivity. peiminine mitigates neuroinflammation by selectively promoting autophagic clearance of AIM2 inflammasomes. Critically, it achieves this without increasing overall autophagic flux or affecting other inflammasome types. As an important component of the innate immune system, AIM2 are activated by DNA fragments through the oligonucleotide/oligosaccharide‐binding domain to initiate inflammasome activation. The degradation of excessive AIM2 through the autophagy pathway has been implicated in tumors and autoimmune diseases. ²⁷ , ²⁸ However, the role of AIM2‐inflammasome autophagy in the brain remains unexplored. Emerging evidence supports the activation of AIM2 inflammasome by ischemia‐induced DNA damage. ²⁹ NLRP1 primarily responds to pathogenic invasion, which is generally not a primary concern in the context of cerebral ischemia. ³⁰ NLRP3 is activated by extracellular ATP and reactive oxygen species, both of which are extensively released during the cell rupture occurring in the acute phase of ischemia. ³¹ Although reactive oxygen species generation can also be triggered by reperfusion, it predominantly occurs within an early time window immediately following oxygen restoration. ³² Notably, DNA damage may persist long after the initial ischemic event, ³³ continuously stimulating the AIM2 inflammasome and sustaining inflammatory responses over extended periods post ischemia. ³⁴ , ³⁵ In our study, we observed that peiminine reduced the expression levels of AIM2 and other inflammasome components without affecting NLRP1 or NLRP3 expression (Figure 4E and 4F). This effect stems from the activation of AIM2‐inflammasome autophagy, as evidenced by unaltered AIM2 transcript levels (Figure S4A) and the restoration of AIM2 upon autophagy inhibition. Image data further confirmed the colocalization of AIM2 and autophagosomes (Figure 5B). The enhanced autophagic clearance of AIM2 inflammasomes was sufficient to confer significant anti‐inflammatory (Figure 2D) and prosurvival (Figure 1D) effects following cerebral ischemia. The selectivity of peiminine toward autophagic substrates was further reflected by its lack of effect on autophagosome biogenesis initiation (Figure 5D and 5E; Figure S5A and S5B). As reported, inhibiting autophagy by 3‐MA or knocking out autophagy‐related gene can exert a protective effect against cerebral ischemia. ³⁶ , ³⁷ The neurotoxic effect of excessive autophagy flux is supposed to result from aggravated lysosomal dysfunction ¹² , ¹³ and cellular structural damage due to nonselective clearance by autophagosomes. ³⁸ Although peiminine does not enhance autophagic flux, it promotes the binding of autophagosomes to the AIM2 inflammasome, thereby facilitating its selective degradation. This selective orientation toward autophagic flux may ensure efficient degradation of critical substrates while avoiding side effects of excessive autophagy. Thus, our study reveals the potential of peiminine as a selective lead compound for AIM2 inflammasome‐targeted autophagy. These findings provide new insights for the development of selective autophagy‐based therapeutic strategies against ischemic injury.

Previous studies showed that total alkaloids of Fritillaria inhibit the inflammatory response by regulating NF‐κB signaling pathway. ³⁹ , ⁴⁰ NF‐κB integrates signals from >20 receptors and its inhibition can be a consequence of broader anti‐inflammatory effects. Existing reports merely observed phenomenological NF‐κB suppression (eg, reduced p65 phosphorylation or suppressed IκB [inhibitor of kappa B] phosphorylation) without identifying direct molecular triggers. Our work identifying a molecular the interactor of peiminine by showing their direct binding. Additionally, AIM2 inflammasome ablation has also been reported to suppress NF‐κB activation. ⁴¹ Here, by using biolayer interferometry technique, we proved that peiminine exhibited selective binding ability with AIM2 but not NLRP1, NLRP3, and NLRC4 (Figure 4B; Figure S3). Studies have shown that ubiquitin‐dependent pathway plays a critical role of selective inflammasome autophagy in the peripheral tissues. The ubiquitin‐specific peptidase 5 promotes the autophagic degradation of the NLRP3 inflammasome through E3 ligase MARCHF7 to alleviate the inflammatory response. ⁴² Pristimerin enhances the K63 ubiquitin chain at the PYCARD K158/161 site of AIM2 to promote autophagy and alleviate tendinopathy in a USP50‐dependent manner. ²⁷ However, the molecular mechanisms underlying selective AIM2‐inflammasome autophagy in the central nervous system remain unclear. The AIM2 inflammasome in CD4⁺ T cells is degraded by autophagy under the action of the E3 ligase TRIM11, ⁴³ which was expressed at a lower level in the brain than in the periphery. ⁴⁴ Given the direct binding of peiminine with AIM2 and few impacts of peiminine on autophagosome biogenesis machinery, we hypothesize that peiminine may function as a molecular glue, enhancing the interaction between AIM2 and specific E3 ubiquitin ligase or other components of the autophagic machinery. This could promote the trafficking of autophagosomes to the AIM2 inflammasome without affecting overall autophagic flux. Therefore, using peiminine as a probe for target identification could likely uncover key molecules involved in AIM2 inflammasome autophagy within the central nervous system, potentially revealing potential therapeutic targets for cerebral ischemic injury. A key limitation is that subsequent studies will likely need to apply proteomics approaches to comprehensively map protein interactome of peiminine, thereby fully unraveling its mechanism of action.

Conclusions

In conclusion, we identified peiminine as a small molecule capable of selectively facilitating AIM2‐inflammasome autophagy to protect against cerebral ischemia. Our study constitutes the pioneering exploration of the pathological significance of AIM2‐inflammasome autophagy in central nervous system and provides the leading compound for its selective regulation.

Sources of Funding

This project was supported by grants the National Natural Science Foundation of China (82273903, 82473900, and 82405045) and Zhejiang Provincial Natural Science Foundation (LD25H090002).

Disclosures

None.

Supporting information

Figures S1–S5

Unedited Gels