Aquaporin 4 and its isoforms regulation ameliorate AQP4 Mis-localization-induced glymphatic dysfunction in ischemic stroke

Graphical abstract

Schematic diagram of the proposed mechanism. In ischemic stroke, impaired cerebral fluid dynamics leads to edema formation, which upregulates AQP4 protein expression and altered its isoforms distribution. This compromises AQP4 localization at astrocyte end-feet, resulting in metabolic waste accumulation and glymphatic system dysfunction. AQP4 inhibition reduces AQP4 mis-localization, thereby enhancing glymphatic fluid clearance and partially reversing metabolic disturbances.

Highlights

• •The functionality of the glymphatic system after ischemic stroke is closely linked to the resolution of edema.
• •TGN-020 corrects AQP4 mis-localization via isoform regulation, preserving glymphatic function.
• •AQP4-M23 reduces AQP4 mis-localization, while AQP4-M1 worsens stroke-induced edema.
• •SNTA1 expression correlates with AQP4 isoform dynamics.
• •Regulation of AQP4 isoforms expression may be a principal mechanism for controlling glymphatic structure.

Abstract

Introduction

The glymphatic system, a brain waste clearance pathway, is impaired during ischemic stroke-induced edema, although the underlying mechanisms remain unclear.

Objectives

This study investigates the temporal dynamics of glymphatic dysfunction post-stroke and the roles of aquaporin 4 (AQP4), its isoforms, and syntrophin alpha 1 (SNTA1) in AQP4 polarization.

Methods

Using a transient middle cerebral artery occlusion (tMCAO) mouse model, glymphatic function was assessed via cisterna magna contrast injection and magnetic resonance imaging. The AQP4 antagonist TGN-020 was administered to elucidate edema’s role in glymphatic dysfunction. AQP4 isoforms viral vectors and SNTA1 modulation were used to study AQP4 polarization and glymphatic function. Techniques included western blotting, q-PCR, immunofluorescence, TEM and behavioral tests. Transcriptomic and metabolomic analyses were performed to assess gene expression and metabolic changes.

Results

Cerebrospinal fluid (CSF) flow decreased during the hyperacute phase, recovering with edema resolution. By administering TGN-020 to reduce edema, distinct alterations in the localization of AQP4 were observed. Specifically, there was a notable increase in AQP4 localization within the astrocyte end-feet. Consequently, CSF inflow and interstitial fluid (ISF) drainage were restored. Transcriptomic sequencing was used to analyze ubiquitination-related channels in tMCAO mice. Metabolic sequencing showed that TGN-020 therapy protected the metabolic stability. Our findings highlight the critical role of AQP4 isoforms in the polarized distribution of AQP4. The upregulation of the AQP4-M1 isoform exacerbated edema and motor dysfunction, whereas the AQP4-M23 isoform corrected the mis-localization of AQP4. Inhibition of AQP4 not only restored the polarized integrity of AQP4 in astrocyte end-feet but also alleviated the metabolic disruptions caused by tMCAO. Furthermore, overexpression of SNTA1 enhanced AQP4 polarity by modulating the expression of AQP4 isoforms.

Conclusion

Cerebral edema disrupts AQP4 localization and glymphatic function following stroke. TGN-020 modulates AQP4 polarization through regulation of AQP4 isoforms and restores glymphatic dysfunction. AQP4-M23 isoform emerges as a key regulator of AQP4 polarization, providing new insights into ischemic stroke pathophysiology.

Introduction

Ischemic stroke is one of the most prevalent neurological disorders [1]. Stroke disrupts intercellular endothelial adhesion molecules, leading to the breakdown of the blood–brain barrier (BBB), displacement of astrocyte end-feet, and a pronounced inflammatory response [2]. Cerebral edema, traditionally regarded as a critical acute complication resulting from increased BBB permeability, remains a major challenge in stroke management [2]. Current non-surgical treatments, such as intravenous hypertonic solutions, often result in undesirable complications and exhibit limited efficacy [3]. Recent studies suggest that cerebrospinal fluid (CSF) inflow into brain tissue following ischemic stroke significantly contributes to acute tissue swelling, implicating its correlation with the glymphatic system [4].

The glymphatic system facilitates the exchange and flow of CSF and interstitial fluid (ISF), thereby promoting waste clearance and maintaining brain homeostasis [5]. It is a complex physiological process [6]. However, recent studies have demonstrated that ischemic stroke induces pathological alterations in perivascular spaces and mis-localization of Aquaporin 4 (AQP4) on astrocyte end-feet [7,8]. These disruptions impair glymphatic function, resulting in cerebral edema, neuronal loss, and inadequate clearance of neurotoxic substances. Furthermore, post-stroke glymphatic dysfunction may increase the risk of dementia and other neurological complications [9]. It has been reported that changes in AQP4 polarization (localization of AQP4 on astrocyte end-feet) and astrocyte proliferation following stroke contribute to increased glymphatic deterioration, including metabolic and inflammatory factors, thereby exacerbating brain destruction [10]. Although current studies indicates that glymphatic system impairment during stroke may depend on edema formation, the underlying molecular mechanism—particular the relationship between parenchymal fluid exchange and AQP4 polarization —remain unclear. Therefore, it is particularly important to study the repair of glymphatic dysfunction after stroke.

AQP4, the most abundant member of the aquaporin family, plays a pivotal role in regulating water balance and CSF flow [10]. AQP4 polarization is essential for efficient glymphatic function. TGN-020, a specific AQP4 inhibitor, has been shown in preclinical studies to reduce brain edema and infarct volume when administered shortly before or after ischemic stroke [11,12], offering a potential therapeutic approach for managing stroke-induced edema. However, TGN-020 has also been observed to reduce AQP4 polarity localization in models of neurodegenerative diseases [13,14], thereby exacerbating glymphatic dysfunction. Several studies indicate that TGN-020 treatment can enhance AQP4 polarization in astrocytes following stroke [12]. Nevertheless, current studies have shown that glymphatic system impairment stroke may depend on edema formation, while the precise molecular mechanisms underlying post-stroke glymphatic dysfunction remain poorly understood.

AQP4 is expressed in two isoforms, AQP4-M1 and AQP4-M23, depending on the location of translation initiation, each with distinct characteristics and functions [15]. The AQP4-M1 isoform, which moves freely within the plasma membrane, primarily facilitates CSF flow, whereas AQP4-M23 tends to form orthogonal array particles (OAPs), that restrict astrocyte migration [16,17]. An increase in the ratio of AQP4-M1 to AQP4-M23 can disrupt OAPs, which play a crucial role in the polarized distribution of AQP4 [18,19]. Syntrophin alpha 1 (SNTA1), as a membrane-associated structural protein, can bind to AQP4 and anchor it to the astrocyte end-foot. Therefore, loss of SNTA1 has been found to influence the normal polarity distribution of AQP4 [20]. This may also be one of the mechanisms affecting the loss of glymphatic system function after ischemic stroke.

Given these findings, elucidating the roles of AQP4 and its isoforms in post-stroke glymphatic system dysfunction is crucial. In this study, we first investigated the development of glymphatic system function during the acute phases of ischemic stroke. Next, we investigated the effects of AQP4 inhibition and the regulation of AQP4 isoforms on glymphatic system recovery. Finally, the importance of SNTA1 in polarity localization of AQP4 was discussed. These findings have important implications for the development of new therapeutic strategies to address glymphatic system dysfunction induced by ischemic stroke and its associated complications.

Materials and methods

Animals

Adult male C57BL/6 mice (22–25 g, 7 weeks old) were obtained from GemPharmatech Co., Ltd. (China). The mice were maintained on a 12-h light/dark cycle at a controlled temperature with ad libitum access to water and food. After a 7-day acclimated period, the mice were randomly assigned to experimental groups.

Ethics statement

All animal experiments were approved by the Institutional Animal Use and Care Committee at the First Affiliated Hospital of University of Science and Technology of China (Approval No. 2024-N(A)-0194) and conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

Surgery and Drug treatment

The transient middle cerebral artery occlusion (tMCAO) model was induced using an intraluminal filament occlusion method [21]. Briefly, mice were anesthetized with 2 % isoflurane (RWD Life Science, China) and placed in a supine position. The right internal carotid artery (ICA), external carotid artery (ECA), and common carotid artery (CCA) were exposed, and a suture (1620A4, Beijing Cinontech Co., Ltd., China) was inserted into the ICA to occlude the middle cerebral artery. After 60 min of occlusion, reperfusion was initiated. Laser Speckle Doppler Flowmetry (PeriCam PSI Z; Perimed, Sweden) confirmed successful occlusion, defined as a＞75 % reduction in cerebral blood flow relative to baseline. The tMCAO mice were randomly divided into different groups based on different reperfusion time points. Sham-operated mice underwent anesthetized and vessels isolation without blocked.

The AQP4 inhibitor TGN-020 (0.1 mg/20 g, HY-W008574, MedChemExpress,) was dissolved in 0.5 % CMC-Na (HY-Y1889A, MedChemExpress) and administered via intraperitoneal injection every 12 h, starting 10 min post-occlusion, as previously described [22]. Sham-operated mice received CMC-Na at the same time points.

Intracerebral injection of AAV Vector

All adeno-associated virus (AAV) were constructed by BrainVTA Co., Ltd. (Wuhan, China) to modulate the expression of AQP4 isoforms and SNTA1. We injected a control virus (rAAV-GfaABC1D-mCherry-WPRE-SV40, AAV-CTRL), a virus overexpressing AQP4-M1 (rAAV-GFAABC1D-AQP4-M1-3XFlag-P2A-mCherry-WPREs, AAV-AQP4-M1), and a virus overexpressing AQP4-M23 (rAAV-GFAABC1D-AQP4-M23-3XFlag-P2A-mCherry-WPREs, AAV-AQP4-M23). rAAV-CMV-SNTA1-EGFP-WPRE-hGH polyA (AAV-SNTA1) was overexpressed with SNTA1, and its corresponding control was rAAV-CMV-EGFP-WPRE-hGH-polyA (AAV-CTRL). rAAV-U6-shRNA (SNTA1)-CMV-EGFP-SV40-polyA (shSNTA1) was a knockdown of SNTA1 expression, and its corresponding control group is rAAV-U6-shRNA(scramble)-CMV-EGFP-SV40-polyA (shCTRL). All viruses had a titer ≥ 3.00 × 10¹² vg/mL.

The viruses (100 nl/min, 600 nl per site) were delivered via stereotaxic injection into three sites [bregma: anteriorposterior (AP) 0.6 mm, mediallateral (ML) −2 mm, dorsoventral (DV) 2/1.6/1.2 mm]. Transfection efficiency was confirmed via immunostaining, and AQP4 isoform/SNTA1 expression was analyzed using quantitative real-time polymerase chain reaction (qRT-PCR), Western blot (WB), and immunofluorescence.

Mice magnetic resonance imaging (MRI)

Mice were anesthetized with 4 % isoflurane and maintained at 1.5 % isoflurane during imaging acquisition. Physiological parameters (heart rate: 80–120 bpm, temperature: 37 ± 0.5 °C) were monitored. A 9.4 T MRI system (uMR 9.4 T, Wuhan United Imaging Life Science Instrument Co., Ltd., China) was to used. To visualize ischemic lesions, axial 2D T2-weighted images (T2WI) and diffusion-weighted images (DWI) were obtained for each mouse. The imaging parameters were as follows: For fast spin echo T2WI, repetition time (TR)/echo time (TE) = 3000 ms/44 ms, field of view (FOV) = 20 mm × 20 mm, matrix = 256 × 256, slice thickness = 0.5 mm, interslice distance = 0.3 mm, number of slices = 22. For DWI, TR/TE = 7355 ms/46 ms, FOV = 13 mm × 11 mm, slice thickness = 0.5 mm, interslice distance = 0.3 mm, number of slices = 32, b values = 0, 1000, 2000 s/mm². Contrast-enhanced MRI (DCE-MRI) employed a 3D T1-weighted imaging (T1WI) sequence to evaluate glymphatic function within the brain. The T1WI sequences were acquired before and after the administration of Gd-DTPA (GC36099, GlpBio, USA) into the cerebellar cisterna at a rate of 1 μ L /min over 10 min, as described in a previous study [23]. The following parameters were used: TR = 3.56 ms, TE = 1.35 ms, flip angle = 6°, FOV = 18 × 8 × 12 mm³, matrix = 176 × 80 × 118, and slice thickness = 0.1 mm. The 3D T1WI sequences were acquired before injection (baseline) and at 20 min post-injection, with scanning performed every 6 min for a total of 210 min.

MRI Postprocessing and data analysis

T2WI and DWI data were analyzed using MR Vision (MR Vision Co., Menlo Park, USA). Brain injury metrics were quantified based on established methods [23,24]. Ischemic lesions volume was expressed as a percentage of the total cerebral hemisphere volume (%HLV), while brain swelling severity was assessed using the brain swelling volume percentage (%BSV). The formula were as follows:

$$\%\text{HLV}=[\text{contralateral hemisphere area}-(\text{total ipsilateral area}-\text{infarct area})]/\text{contralateral hemisphere area}\times \text{1}00\%$$ (1)

$$\%\text{BSV}=(\text{total hemisphere ipsilateral area}-\text{contralateral hemisphere area})\times \text{1}00\%$$ (2)

Time-series 3D DCE data were aligned to the standard reference template for head motion correction. Time-series images were spatially normalized to the SIGMA standard space via ANTs (v2.4.3) to derive deformation fields. Deformation field inverse transformations were applied to standard space ROIs, generating individual space ROIs. Voxel-by-voxel signal intensity changes in contrast-enhanced images were calculated by subtracting and dividing each time-point image by the baseline. SI normalization was implemented in MATLAB (R2020a, The MathWorks Inc., Natick, USA). Five bilateral ROIs (left/right hemisphere, pituitary fossa, olfactory bulb, and whole brain) were defined using MRIcroGL software. The average values of the time course within each ROI were extracted, and the area under the curve (AUC) was computed. Subsequently, semi-quantitative kinetic parameters (influx/efflux rate) were computed for contrast agent infusion and tissue clearance. Formula were as follows:

$$\text{Influx rate}=[(\text{SIpeak}-\text{SIpre})/(\text{SIpre}\times \text{Tpeak})]\times \text{1}00(\%)$$ (3)

$$\text{Efflux rate}=[(\text{SIpeak}-\text{SIend})/(\text{SIpre}\times \text{Tend})]\times \text{1}00(\%)$$ (4)

SIpeak: signal intensity at peak enhancement, SIpre: signal intensity before injection, SIend: signal intensity at Tend, Tpeak: time to peak enhancement, Tend: time at the end of acquisition.

Then, the temporal profile of each ROI was fitted using a model with two exponentials as described previously. First, the mean value for each ROI was extracted and then calculated using the following two exponentials model formula:

$$f(\text{t})={\text{c}}_{\text{1}}·(\text{1}-{\text{e}}^{-\text{t}/\tau \text{in}})+{\text{c}}_{\text{2}}·({\text{e}}^{-\text{t}/\tau \text{out}}-\text{1})$$ (5)

c₁ and c₂: gain constants, τin: influx time constant, τout: efflux time constant.

Intracisternal tracer infusion

CSF flow was assessed through cisterna magna injection of fluorescent tracers. Mice were anesthetized and the cisterna magna was exposed. A catheter (#BB31695-PE/1, Smiths Medical Co., Ltd., USA) connected to a Hamilton syringe was inserted into the cisterna magna using a 24G needle. Fluorescent tracers (3-kDa dextran, D34682, Invitrogen; 45-kDa ovalbumin, O34782, Invitrogen) were dissolved in artificial CSF (R22153, Yuanye Medical Co., Ltd., China) and infused at 1 μL/min for 10 min. After injection, the needle was kept in place for 10 min. Mice were euthanized at 30 min and 60 min post-infusion. Brains were collected and fixed in 4 % paraformaldehyde (PFA) for 24 h. 100-μm sections were examined for tracer inflow using laser scanning confocal microscopy (FV3000, Olympus). Tracer distribution was quantified blindly using ImageJ (National Institutes of Health, Bethesda, USA).

Interstitial solute drainage

Interstitial solute drainage was evaluated by intracerebral injection of tracers to assess ISF clearance capacity. Fluorescent tracers including 3-kDa dextran (D34682, Invitrogen) or 45-kDa ovalbumin (O34782, Invitrogen) dissolved in artificial CSF were injected into brain parenchyma lesions using a stereotaxic device. The injection speed was 0.2 μ L/min, and the injection coordinates were AP −1.8 mm, ML −1.5 mm, DV 2.0 mm relative to bregma. After injection, the needle was left in place for 10 min before slow withdrawal to prevent reflux. Thirty minutes post-injection, the mice were anesthetized and perfused. Brains were fixed at 4 % PFA for 24 h. Tracer efflux was then observed using laser confocal microscopy (FV3000, Olympus).

Immunofluorescence (IF)

Mice were deeply anesthetized and perfused with phosphate-buffered saline (PBS) followed by 4 % PFA fixation. The brains were post-fixed in PFA overnight at 4 °C, then cryoprotected through graded sucrose dehydration (20 % and 30 %). Subsequently, brains were embedded in Tissue-Tek O.C.T compound (Sakura Finetek, USA) and sectioned coronally at 20 μm thickness. The sections were incubated with the following primary antibodies: glial fibrillary acidic protein (GFAP, 1:500, 3670S, CST, USA), anti-AQP4 (1:300, ab128906, Abcam, USA), anti-CD31 (1:100, ab281583, Abcam, USA) and SNTA1 (1:200, sc-166635, SANTA CRUZ, USA). The next day, sections were incubated with secondary antibodies for 1 h at room temperature: donkey anti-rabbit and anti-mouse antibodies conjugated with Alexa Fluor 488/594/647 (1:100, Abcam). The sections were then stained with DAPI (1:1000, MK411A, Biomiky Co., Ltd., China). IF staining was visualized using laser confocal microscopy (FV3000, Olympus). Semi-quantitative analysis of the images was performed using ImageJ software (National Institutes of Health, Bethesda, MD, USA). The assessment of AQP4 expression and polarization according to previous method. AQP4 polarization was evaluated by the ratio of low-threshold to high-threshold AQP4-positive areas.

Meningeal Preparation and staining

Mouse heads were dissected as previously described and fixed overnight [25]. Next day, meninges were carefully peeled off and incubated with the primary antibody (LYVE1, 1:1000, 28321–1-AP, Proteintech Co., Ltd., China) mix overnight at 4°C. After washing with PBS at room temperature, the tissues were incubated with a fluorophore-conjugated secondary antibody for 2 h, followed by further washing in PBS. Imaging was conducted as soon as possible.

WB

Mouse cortex tissues were extracted and centrifuge, and the light supernatant was collected. Proteins (20 μL per lane) were separated using 10–12 % SDS-PAGE electrophoresis and transferred onto PVDF membranes (Billerica Millipore, MA, USA). The membranes were incubated overnight at 4 °C with the following primary antibodies: anti-AQP4 (1:1000, 16473–1-AP, Proteintech Co., Ltd., China) and anti-AQP4 isoforms (1:500, sc-390488, Santa Cruz, USA). The next day, the blots were incubated with secondary antibodies for 1 h at room temperature. Protein bands were detected using enhanced chemiluminescence (BL523B, LABGIC Co., Ltd., China). The intensity of immunoreactive bands was quantified using ImageJ analysis software (National Institutes of Health, MD, USA).

Co-immunoprecipitation (CO-IP) was used to detect protein interactions, and antibodies were mixed with proteins and incubated overnight at 4°C. The antibodies used include normal mouse IgG (4340, Cell signalling technology, USA). Next day, Protein A/G magnetic beads (Nanjing Vazyme Co., Ltd., China) were added and incubated at 4°C overnight. Then, the beads were collected by magnetic adsorption and subjected to protein immunoblotting.

HE staining

Slices were dehydrated through graded ethanol concentrations and HE staining was performed a HE staining kit (WASCI Co., Ltd., China). The stained sections were examined under a light microscope (KF-DPS-120, Jiang Feng Co., Ltd., China).

Electron Microscopic Morphological analysis

Samples were sequentially fixed with glutaraldehyde and osmium tetroxide. After dehydration, the samples were sectioned into 70 nm sections and examined using a JEM1400 Transmission Electron Microscope (JEOL, Japan).

Behavioral tests

The behavior of mice was tested according to the experimental schedule. The Morris water maze (MWM) test was conducted following a previous protocol to assess the spatial learning and memory of the mice [27]. Briefly, mice were trained to find platforms four times a day for five days. On the sixth day, the platform was taken away. ANY-maze video tracking software (Stoelting, USA) was used to record and analyze the escape latency, the time spent in the target quadrant and the number of crossing platform within 60 s of mice free swimming and crossed the platform. The novel object recognition (NOR) test was performed to evaluate recognition memory in mice according to established protocols [28]. After 3 days of habituation to the empty arena, mice were exposed to two identical square objects for 5 min (training phase). Following a 24 h interval, one square object was replaced with a novel circular object (testing phase). Exploratory behavior (sniffing or whisker contact within 2 cm) was monitored for 5 min. Recognition memory was quantified using the discrimination index. Stroke motor deficits were assessed using the classical rotarod test [29]. Mice were placed on the rotating rod under quiet conditions. The rotating rod slowly accelerated to a certain speed. The speed of the rotating rod and the duration of stay were recorded when the mouse finally dropped.

qRT-PCR analysis

Tissue samples were processed using a commercial RNA isolation kit (New Cell & Molecular Biotech Co., Ltd., China). RNA concentration was determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA), with the A260/A280 ratio maintained between 1.8 and 2.0 for all samples. Reverse transcription was carried out on purified RNA with the RevertAid cDNA synthesis reagents (Vazyme Co., Ltd., China), following the supplier's recommended procedures. Gene amplification analysis was conducted in 25 μL reactions incorporating UltraSYBR Green master mix (Vazyme Co., Ltd., China) and processed through the Light-Cycle 96 platform (Roche, Switzerland). Expression levels were normalized to the housekeeping gene GAPDH and quantified through the comparative threshold cycle method. Corresponding amplification primers are detailed in Supplementary Table 1.

RNA transcriptional sequencing analysis

RNA transcriptome sequencing was conducted using brain tissue samples from tMCAO and control mice. Differentially expressed genes were identified with P-value < 0.05 and false discovery rate < 0.05.

Liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis

LC-MS/MS analysis was performed using an ultra-high-performance liquid chromatography system (Thermo Fisher Scientific, USA). Differential peaks were isolated and identified using an LC-MS/MS database.

Statistics and Reproducibility

Each experiment was independently repeated at least three times with consistent results. Statistical analyses were conducted using GraphPad Prism 8.0 (La Jolla, USA) and SPSS 24.0 (Armonk, USA). All data were tested for normality and expressed as mean ± SD. Differences among multiple groups were analyzed by one-way analysis of variance (ANOVA) with Tukey’s post hoc test. For MWM, escape latency was analyzed by two-way repeated-measures ANOVA with Tukey’s post hoc test. Glymphatic system function dynamics assessed by MRI at multiple time points were analyzed by two-way ANOVA with Dunnett's post hoc test. In all analyses, P < 0.05 was considered statistically significant.

Results

Brain injury in tMCAO mice at different time points

In this study, we initially assessed the extent of brain injury in tMCAO mice at various time points (Fig. 1A). Infarct area and brain swelling were assessed by T2WI and DWI at 2 h, 1 d, 3 d, and 7 d post-stroke, as shown in Fig. 1B. Our results revealed that the infarct volume in the tMCAO group at 2-hour was significantly smaller than the volumes observed at 1-day post-tMCAO (Fig. 1C; %HLV at 2 h: 8.713 ± 4.113, 1 d: 52.93 ± 9.406, 3 d: 47.34 ± 12.36, 7 d: 45.15 ± 8.799). Regarding brain swelling, we detected swelling at all time points as illustrated in Fig. 1B. Brain swelling increased at 1-day and peaked at 3-day (Fig. 1C; %BSV at 2 h: 104.2 ± 5.154, 1 d: 118.2 ± 6.868, 3 d: 126.1 ± 12.12, 7 d: 107.8 ± 9.918). However, no significant difference in brain swelling was observed between 2-hour and 7-day post-reperfusion. These findings indicate that swelling begins to subside after 3-day after tMCAO. This suggests that both the infarct area and brain swelling fluctuate during the acute phase of ischemic stroke, providing critical insights into the progression of brain injury following ischemic events.

Fig. 1.

Dynamic Evaluation of tMCAO Brain Injury and Glymphatic System Function in Mice Over Time. A Overview of the experimental process and group allocations. B Representative T2WI and DWI scans of the sham group and ischemic stroke group at various time points. C Ischemic lesion and brain swelling volumes in sham and tMCAO group at indicated time points (n = 8 mice per group). D Representative Western blots of AQP4 protein expression. E Quantification of relative AQP4 protein expression normalized (n = 8 mice per group). F Representative sagittal MRI images showing the flow of the contrast agent Gd-DTPA. G-I Quantitative analysis of signal-to-noise ratio variations and AUC in three ROIs across a 210-minute DCE Sequence (n = 4 mice per group). All data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. sham;^P < 0.05,^^P < 0.05,^^^P < 0.001,^^^^P < 0.0001 vs. tMCAO 2 h; #P < 0.05, ##P < 0.01 vs. tMCAO 1 d; &&&P < 0.001 vs. tMCAO 3 d). All data were compared by one-way ANOVA with Tukey’s post hoc test.

We also examined edema-associated AQP4 protein levels in each group by Western blotting (Fig. 1D). In the tMCAO 1-day group, the AQP4 content was significantly higher than in the tMCAO 2-hour and 7-day groups (Fig. 1E; sham: 0.412 ± 0.2145, tMCAO 2 h: 1.783 ± 0.5261, tMCAO 1 d: 2.88 ± 0.8266, tMCAO 7 d: 1.761 ± 0.7153).

Differences in Spatiotemporal distribution of Gd-DTPA in tMCAO mice

To evaluate glymphatic transport efficacy, we investigated the dynamics of the contrast agent Gd-DTPA in tMCAO mice and compared it with the sham group. Using DCE-MRI, we monitored the contrast agent’s distribution in the brain over a period ranging from 20 to 210 min following injection into the cisterna magna (Fig. 1F). The contrast agent was observed to diffuse through the perivascular spaces, initially along the superficial brain surface followed by gradual penetration into deeper parenchymal regions. This spatial–temporal distribution pattern is consistent with previous neuroimaging findings [23]. High signal intensity in the cisterna magna extended through Virchow-Robin spaces and other perivascular routes, including the cerebellum, pituitary recess, and olfactory bulb, with a particularly pronounced signal intensity in the cerebellum.

To discern differences in contrast agent distribution between brain regions after stroke, we analyzed signal intensity in five regions of interest (ROIs): the whole brain, left hemisphere, right hemisphere, olfactory bulb, and pituitary recess (Fig. 1G-I and Supplementary Fig. 1A-B). In the sham group, the signal intensity curves for all five ROIs exhibited an initial increase followed by a gradual decrease. In contrast, the tMCAO 2-hour group showed a less pronounced rate of signal increase. In the tMCAO 1-day, 3-day, and 7-day groups, the signal peaked at earlier time points compared to the sham group. Notably, the signal intensity in the tMCAO 7-day group was comparable to that in the sham group, indicating partial restoration of glymphatic flow. Quantitative analysis of the area under the AUC demonstrated significantly reduced and temporally delayed contrast agent distribution in all five examined brain regions at both 2-hour and 1-day time points following tMCAO, when compared with the sham group. Comparative analysis revealed significantly lower influx/efflux rates in tMCAO versus sham mice across most perivascular pathways (Supplementary Fig. 2A-B). The internal flow rate in the tMCAO 7-day group was significantly increased (P < 0.05), although it remained lower than in the sham group. This suggests that the recovery of contrast agent signal intensity corresponds with the resolution of edema.

We conducted a comprehensive comparison among different experimental groups to assess variations in the dynamics and kinetics of glymphatic transport. We extracted the time-course data from each animal and fitted it to a double exponential model, which provided parameters for the time constants representing the influx and efflux of the contrast agent Gd-DTPA. Significant differences in τin were observed between the experimental groups, specifically in the right brain region between 2-hour and 1-day after tMCAO, and between 2-hour and 3-day (Supplementary Fig. 2C). However, no significant differences were noted in τout in the olfactory bulb and pituitary recess among the tMCAO mice (Supplementary Fig. 2D). These results suggest that glymphatic transport dynamics are altered in tMCAO mice, with varying patterns of influx and efflux at different post-stroke time points, indicating potential for recovery over time. The τin and τout derived from our model are consistent with previously observed phenomena.

Signal intensity analysis across five ROIs is shown in Supplementary Fig. 3A. The results revealed no significant differences between the left and right hemispheres or the whole brain within each group (Supplementary Fig. 3B-F). The olfactory bulb consistently exhibited the highest signal intensity, indicating that contrast agents accumulate in this area (Supplementary Fig. 3G). These findings demonstrate comparable metabolic activity between the ipsilateral and contralateral hemispheres, independent of the presence or severity of brain injury.

Effects of AQP4 inhibition on cerebral cortex injury in tMCAO mice

In examining changes in stroke-related injury over time, we noted that the tMCAO 1-day group exhibited severe cerebral edema and glymphatic dysfunction. Consequently, we selected 1-day post-injury for further investigation.

The overall experimental process and grouping for this part of the study are depicted in Fig. 2A. TGN-020 was administered as a treatment to observe its effects on brain injury via MRI (Fig. 2B). Compared with the tMCAO + vehicle group, TGN-020 significantly reduced both infarct area and brain swelling in tMCAO mice (Fig. 2C: %HLV: 14.77 ± 5.789 vs 45.88 ± 11.94, P = 0.0001; %BSV: 100.8 ± 3.896 vs 118.2 ± 13.80, P = 0.0244). These findings suggest that TGN-020 may provide neuroprotection by mitigating the severity of cortical damage in ischemic stroke.

Fig. 2.

TGN-020 Improves Brain Injury and Glymphatic System Dysfunction in tMCAO Mice. A Overview of the second part of the experimental process and group allocations. B Representative T2WI and DWI images of three distinct mouse groups. C Ischemic lesion volumes and brain swelling volumes in each group (n = 5 mice per group). D Schematic of intracerebral injection procedure in mice. Representative fluorescence microscopy images show Ovalbumin (45-kDa, red) inflow into CSF at 30- and 60-minute. The white arrow indicates the residual tracer deposition. Scale bar: 1000 μm. E Quantification of Ovalbumin inflow at 30- and 60-minute in brain sections (n = 4 mice per group). F Brain slices from each group show dextran (3-kDa, green) and ovalbumin (45-kDa, yellow) inflow into the brain cistern. Scale bar: 1000 μm. G Quantification of Glucan and Ovalbumin inflow in brain sections from each group (n = 4 mice per group). H Visual depiction of interstitial drainage (highlighted in red) following the introduction of the Ovalbumin tracer into the infarcted side. Scale bar: 1000 μm. I Quantification of the residual Ovalbumin tracer-covered area fraction in each group (n = 5 mice per group). All data are shown as mean ± SD. *P < 0.05, ****P < 0.0001 vs. sham; #P < 0.05, ###P < 0.001, ####P < 0.0001 vs. tMCAO + vehicle. One-way ANOVA with Tukey’s multiple comparisons test was performed in (C) and (I). Two-way repeated-measures ANOVA with Tukey’s post hoc test in (E) and (G).

Impact of AQP4 inhibition on glymphatic dysfunction and AQP4 isoforms in tMCAO mice

The glymphatic system of tMCAO mice was evaluated by using cisterna magna CSF and intracerebral parenchyma with tracer injection to observe the degree of diffusion. First, 45-kDa fluorescent tracers were injected into the cisterna magna, and brain tissue samples were collected at 30 min and 60 min to observe tracer distribution (Fig. 2D). Quantitative analysis showed greater tracer accumulation in sham group cortex compared to other groups at 30 min post-injection (Fig. 2E). The ischemic side of the tMCAO + vehicle group displayed no tracer presence, indicating delayed tracer inflow. However, following treatment with TGN-020, a modest amount of tracer inflow was observed in the ischemic side cortex, suggesting that TGN-020 significantly enhanced CSF influx. At 60 min post-injection, tracer accumulation in the infarct area was significantly higher in the tMCAO + vehicle group compared to both the sham group and the tMCAO + TGN-020 group. To further understand tracer dynamics in CSF, we quantified the diffusion of tracers of different molecular sizes (Fig. 2F, Green and Yellow). 3-kDa tracers showed greater diffusion than 45-kDa in all groups (Fig. 2F). These results show that the flow is faster for smaller molecules in ISF space (Fig. 2G). Additionally, we assessed clearance effect of the glymphatic system after intracerebral parenchyma tracer injection (Fig. 2H). The results revealed significantly fewer remaining traces in the treated group compared to the untreated group (Fig. 2I). These findings suggest that TGN-020 effectively improves the drainage of CSF and ISF in the brain parenchyma of tMCAO mice.

To investigate the polarization status of AQP4 in the brain, we first examined the localisation of AQP4 and CD31 around the infarcted cortex. AQP4 showed polarized distribution around blood vessels, confirming its high signal expression in perivascular regions (Supplementary Fig. 4).

Subsequently, we conducted a comprehensive analysis of AQP4 and GFAP localization patterns (Fig. 3A). In the tMCAO + vehicle group, we observed a distinct disruption of AQP4 polarization, characterized by diffuse intracellular distribution predominantly co-localized with the astrocyte cell bodies. In contrast, the tMCAO + TGN-020 group maintained near-normal polarization patterns, with AQP4 predominantly localized to astrocytic end-feet processes. Quantitative results demonstrated a significant reduction in AQP4 polarity in the tMCAO + vehicle group. Notably, TGN-020 treatment effectively protected AQP4 polarization, maintaining its normal perivascular localization (Fig. 3B).

Fig. 3.

TGN-020 Improves the Structure of Intracerebral Transport in the Glymphatic and Meningeal Lymphatic Systems. A Comparison of GFAP-positive areas (green) and AQP4 polarization (red) in the peri-infarct cortex. Scale bar, 50 μm. B Quantification of AQP4 polarization across the three groups (n = 6 mice per group). C-E Representative Western blot bands and densitometric quantification of AQP4-M1 and AQP4-M23 in the peri-infarction area (n = 6 per group). F Perivascular space structure observed by HE staining. TEM images showing perivascular astrocyte end-feet (green) in each group. H Representative images of meningeal lymphatic vessels stained with LYVE1. The enlarged view shows meningeal lymphatic vessels in the cortical subarachnoid space (COS) region. H-I Quantification of LYVE1 distribution and vessel diameter in the three groups (n = 6 mice per group). All data are shown as mean ± SD. *P < 0.05, ****P < 0.0001 vs. sham; ##P < 0.01 vs. tMCAO + vehicle. ALL data were compared by one-way ANOVA with Tukey’s post hoc test. AC, astrocytes; drAC, detached retracted astrocytes; EC, endothelial cell; RBC, red blood cell.

To validate the role of AQP4 isoforms changes in the glymphatic system, we administered TGN-020 and performed densitometric analysis (Fig. 3C-E). The results revealed that the expression levels of AQP4-M1 and AQP4-M23 were significantly higher in the tMCAO group compared to the TGN-020 group (AQP4-M1: P = 0.0446; AQP4-M23: P = 0.0354).

Changes in the distribution of AQP4 isoforms can lead to sparse distribution of astrocyte end-feet, a critical component of the perivascular space [30]. We explored perivascular space and astrocyte ultrastructural changes using HE staining and TEM imaging. HE staining revealed that the perivascular space around the infarcted cortex was enlarged in both the tMCAO + vehicle and tMCAO + TGN-020 groups, potentially affecting CSF flow in the brain (Fig. 3F). Additionally, electron microscopy showed that astrocytes in the tMCAO + vehicle group were severely swollen, rounded, and detached, whereas the structural integrity of astrocytes was restored following TGN-020 treatment (Fig. 3F). These results suggest that increased AQP4 isoforms distribution after tMCAO leads to AQP4 mis-localization and astrocyte rupture, which can be mitigated by AQP4 inhibition (TGN-020).

Meningeal lymphatics (mLVs) serve as a crucial link between the central nervous system and the periphery. Located in the dural membrane of mice, mLVs drain extracellular fluid (CSF and ISF) from the CNS to peripheral lymph nodes and represent the next step in the efflux of the glymphatic system. mLVs were specifically labeled with lymphatic endothelial receptor 1 (LYVE1) (Fig. 3G). Immunostaining results confirmed that mLV coverage was significantly increased in the tMCAO + vehicle group, accompanied by mLV expansion. Inhibition of AQP4 alleviated the expansion of mLVs (Fig. 3H-I).

AQP4 inhibition alleviates behavioral disorders in tMCAO mice

To evaluate motor skills and anxiety-like behaviors in each group, we conducted behavioral tests using the Morris Water Maze (MWM) and rotarod experiments. In the MWM spatial navigation test, mice in the tMCAO + vehicle group exhibited a significantly longer escape latency compared to those in the sham group, indicating a decline in spatial learning and memory abilities (Fig. 4A). During the MWM spatial exploration test, the tMCAO + vehicle group showed a marked reduction in platform crossings compared to the sham group. Conversely, the TGN-020 group exhibited an increase in the number of platform crossings and spent more time in the fourth quadrant compared to the tMCAO + vehicle group (Fig. 4B-C). In contrast, tMCAO mice treated with TGN-020 demonstrated a significantly reduced escape latency compared to the tMCAO + vehicle group, suggesting improved spatial learning and memory following treatment (Fig. 4D). These findings further support the notion that TGN-020 enhances spatial learning and memory impairments caused by acute ischemic stroke.

Fig. 4.

Effects of TGN-020 on cognitive dysfunction and metabolic pattern in tMCAO mice. A Representative image of swim traces from the Morris water maze test. B-D Data from the MWM test analyzing spatial learning and memory abilities of tMCAO mice (n = 10 mice per group). E Representative thermal imaging from the probe trial in the novel object recognition test. F Relative cognitive index with the novel object analyzed to evaluate memory ability following tMCAO (n = 10 mice per group). G Duration and speed in the rotarod test used to assess motor function in each group (n = 10 mice per group). H Transcriptomic changes in tMCAO mice. Red indicates down-regulated genes, blue dots indicate up-regulated genes, and gray dots indicate no significant changes. There was no significant change in the expression of AQP4 and SNTA1. I GSEA shows the ubiquitin-activity pathway. J Workflow for metabolomics analysis. K Volcano plot comparing differential metabolites between the tMCAO + vehicle and tMCAO + TGN-020 groups. L Pie chart depicting the proportions of different metabolite categories between the tMCAO + vehicle and tMCAO + TGN-020 groups. M Heatmap of the top 20 differentially expressed metabolites. N Analysis of affected metabolic pathways in comparisons between the tMCAO + vehicle and tMCAO + TGN-020 groups. All data are shown as mean ± SD. ****P < 0.0001 vs. sham; #P < 0.05, ##P < 0.01, ####P < 0.0001 vs. tMCAO + vehicle. One-way ANOVA with Tukey’s multiple comparisons test was performed in (B), (C) and (F). Two-way repeated-measures ANOVA with Tukey’s post hoc test in (D).

The NOR test assessed memory function (Fig. 4E). The tMCAO + vehicle group had a significantly lower object recognition index compared to the sham group, indicating a diminished ability to distinguish between new and familiar objects, as well as reduced exploratory behavior towards novel objects (Fig. 4F). However, mice in the tMCAO + TGN-020 group exhibited a significantly improved object recognition index compared to the tMCAO + vehicle group, reflecting a restoration of memory and exploratory behavior. Overall, these results suggest that TGN-020 can mitigate cognitive and motor deficits induced by acute ischemic stroke.

In the rotarod test, the tMCAO + TGN-020 group demonstrated improved rotation speed and endurance compared to the tMCAO + vehicle group (Fig. 4G). This suggests that TGN-020 may enhance balance and motor resilience in mice following neurofunctional damage induced by acute ischemic stroke.

Ubiquitination degradation occurs in tMCAO mice

To investigate the genomic changes in mice after cerebral ischemia, we analyzed array sequencing data of young adult male mice on day 1 after tMCAO. Differential gene analysis revealed no change in the expression of AQP4 and SNTA1 mRNA (Fig. 4H). However, comparative WB quantification identified statistically significant upregulation of AQP4 protein expression in our study (Fig. 2D). GSEA results showed that the ubiquitin–proteasome pathways were up-regulated after tMCAO (Fig. 4I).

AQP4 inhibition Protects against tMCAO-Induced metabolic Perturbations

We used metabolomics to analyze the therapeutic mechanism of TGN-020 in tMCAO mice (Fig. 4J). The results revealed clear separation between the tMCAO + vehicle and tMCAO + TGN-020 groups based on their metabolic profiles. Compared with the tMCAO + vehicle group, 123 metabolites were upregulated and 141 were downregulated in the tMCAO + TGN-020 group (Fig. 4K, Supplementary Table 2). The differentially expressed metabolites were predominantly lipids and lipid-like molecules (24.47 %), and organoheterocyclic compounds (17.99 %; Fig. 4L). TGN-020 treatment of tMCAO mice, which restored AQP4 polarization, led to downregulation of organic acids and derivatives (Fig. 4M). The bubble plot revealed that restored AQP4 polarization was associated with regulation of glycolysis or gluconeogenesis, taurine and hypotaurine metabolism, glycine, serine, and threonine metabolism, alpha-linolenic acid metabolism, and methane metabolism (Fig. 4N).

AQP4-M23 isoform enhances brain injury recovery and glymphatic function in tMCAO mice

Our previous findings indicated that changes in AQP4 isoforms expression are closely related to AQP4 polarity. We next explored the potential benefits of regulating AQP4 isoforms in mice with ischemic stroke. The experimental flow chart is shown in Fig. 5A. Initially, q-PCR analysis was performed to assess the efficacy of AQP4 isoforms adeno-associated virus injection in inducing AQP4 isoforms overexpression across the experimental groups (Fig. 5B). MRI assessed brain injury in each group (Fig. 5C). Although there were no significant differences in infarct area among the three groups, the condition of edema was significantly improved in the tMCAO + AAV-AQP4-M23 group compared to the tMCAO + AAV-AQP4-M1 group (P = 0.0039, Fig. 5D BSV%).

Fig. 5.

Different AQP4 Isoforms Affect Brain Injury and Glymphatic Function in tMCAO Mice. A Overview of the third part of the experimental process and group allocations. B Relative mRNA expression levels of two isoforms of AQP4 in the peri-infarction area after tMCAO (n = 6 mice per group). C Representative T2 and DWI images of the three distinct groups of mice. D Ischemic lesion volumes and brain swelling volumes in each group (n = 6 mice per group). E Description of brain injection techniques and representative brain slice staining from three groups showing dextran (3-kDa, green) inflow into the brain cistern (above) and parenchyma (down) at 30 min. Scale bar: 1000 μm. F Quantification of Glucan inflow into the cistern (n = 3 mice per group). G Quantification of the residual Glucan in each group (n = 3 mice per group). H Comparison of GFAP-positive areas (green), AQP4 (red), and CD31 (grey) in the peri-infarct cortex. Scale bar: 100 μm. I Quantification of AQP4 polarization across the three groups (n = 4 mice per group). J Representative thermal imaging from the probe trial in the NOR test. K Relative cognitive index with the novel object analyzed to evaluate memory ability (n = 10 mice per group). L Duration and speed in the rotarod test used to assess motor function in each group (n = 10 mice per group). All data are shown as mean ± SD. *P < 0.05, **P < 0.01, ****P < 0.01 vs. tMCAO + AAV-CTRL; #P < 0.05, ##P < 0.01, ####P < 0.0001 vs. tMCAO + AAV-AQP4-M1. All data were compared by one-way ANOVA with Tukey’s post hoc test.

We evaluated CSF flow and ISF diffusion by injecting a 3-kDa fluorescent tracer into the cisterna magna and brain parenchyma (Fig. 5E). Thirty minutes after injection, the tracer distribution in the cortical region was prominent in the tMCAO + AAV-CTRL group, whereas distribution in the tMCAO + AAV-AQP4-M1 group was limited to the bottom of the section (Fig. 5E Above). This indicates a delay in tracer flow into the brain following the increase in AQP4-M1 (Fig. 5F). Subsequently, a 3-kDa tracer was injected into the brain parenchyma to assess intracerebral clearance (Fig. 5D Below). The results showed a significant reduction in residual tracer in the tMCAO + AAV-AQP4-M23 group compared to the tMCAO + AAV-AQP4-M1 group (Fig. 5G).

Immunofluorescence analysis of AQP4 localization was performed (Fig. 5H). Quantitative analysis revealed that AQP4 polarization was significantly higher in the tMCAO + AAV-AQP4-M23 group compared to the tMCAO + AAV-AQP4-M1 group (Fig. 5I).

To evaluate memory, we conducted the NOR test (Fig. 5J). The tMCAO + AAV-AQP4-M1 group exhibited a significantly lower object recognition index compared to the tMCAO + AAV-AQP4-M23 group, suggesting reduced ability to differentiate between new and familiar objects and diminished exploratory behavior towards novel objects (Fig. 5K). In the rotarod test, the tMCAO + AAV-AQP4-M1 group showed lower rotation speed and endurance compared to the tMCAO + AAV-CTRL group, indicating that increased AQP4-M1 impaired balance and increased motor fatigue following ischemic stroke-induced neurofunctional damage (Fig. 5L). These results suggest that the AQP4-M23 isoform may provide protective effects by enhancing astrocyte polarization and reducing edema in the context of acute ischemic stroke.

SNTA1 overexpression improves glymphatic function in tMCAO mice

We utilized AAV to overexpress and knockdown SNTA1 to assess its effect on glymphatic function. The experimental flow chart is shown in Fig. 6A. Western blot results revealed a significant decrease in the level of AQP4-M23 and AQP4-M1 in AAV-SNTA1 mice after tMCAO compared to the shSNTA1 group (Fig. 6B-D). Densitometric analysis revealed that the levels of SNTA1 and AQP4-M23/AQP4-M1 in the cerebral cortex were downregulated in the AAV-SNTA1 + tMCAO group compared with the shSNTA1 + tMCAO group (Fig. 6E-F).

Fig. 6.

SNTA1 overexpression improves glymphatic clearance in tMCAO mice. A Overview of the last part of the experimental process and group allocations. B-F Representative Western blot bands and densitometric quantification of AQP4-M23, AQP4-M1, SNTA1 and AQP4-M23/AQP4-M1 in the peri-infarction area (n = 8 per group). G Co-IP of SNTA1 with AQP4 from tMCAO mice brain precipitated by the SNTA1 antibody. H Description of brain injection techniques and representative brain slice staining showing ovalbumin (45-kDa, green) inflow into the brain cistern (left) and parenchyma (right) at 30 min. Scale bar:1000 μm. I Quantification of Ovalbumin inflow into the cistern (n = 3 mice per group). J Quantification of residual Ovalbumin-covered area fraction (n = 3 mice per group). K Comparison of SNTA1(green), AQP4 polarization (red), and GFAP (grey) in the peri-infarct cortex. Scale bar: 200 μm. L Quantification of AQP4 polarization across the four groups (n = 3 mice per group). All data are shown as mean ± SD. **P < 0.01, vs. tMCAO + shCTRL; #P < 0.05, ##P < 0.01 vs tMCAO + AAV-CTRL. All data were compared by one-way ANOVA with Tukey’s post hoc test.

Next, we verified the interaction between SNTA1 and AQP4 by co-immunoprecipitation (Fig. 6G). The results showed that AQP4 was co-precipitated with SNTA1, but not with IgG. In conclusion, reduced SNTA1 levels post-tMCAO may cause AQP4 mis-localization.

We evaluated CSF flow and ISF diffusion by injecting a 45-kDa fluorescent tracer into the cisterna magna and brain parenchyma (Fig. 6H). Thirty minutes after injection into the cisterna magna, SNTA1 overexpression + tMCAO group significantly enhanced tracer levels in the brain compared with AAV-CTRL after tMCAO group (Fig. 6I). This indicates an improvement in glymphatic function. Subsequently, tracer was injected into the brain parenchyma to assess intracerebral clearance (Fig. 6H right). The results showed a significant reduction in residual tracer in the tMCAO + AAV-SNTA1 group compared to the tMCAO + AAV-CTRL group and tMCAO + shSNTA1 group. (Fig. 6J-K).

Immunofluorescence analysis of AQP4 localization was performed (Fig. 6K). Quantitative analysis showed that AQP4 polarization in the tMCAO + AAV-SNTA1 group was significantly higher than that in the tMCAO + shSNTA1 group, indicating that overexpression of SNTA1 can help AQP4 polarization localization (Fig. 6L).

Discussion

The whole experiment demonstrated that in tMCAO mice, TGN-020 mediated the regulation of AQP4 isoforms expression, or modulation of the AQP4-M23 isoform, significantly enhanced the perivascular localization of AQP4 in astrocytes. This regulatory mechanism improves glymphatic dysfunction and mitigates ischemic stroke pathologies. A comprehensive schematic diagram of the proposed mechanism is presented in the Graphical Abstract.

Previous research has highlighted varying degrees of damage to the glymphatic system across several neurological disorders [26]. Our research is primarily structured into four main components. First, we systematically investigated the temporal dynamics of CSF circulation during post-stroke edema formation. Subsequently, we discovered that the modulation of AQP4, a protein implicated in edema, not only ameliorates brain damage but also enhances the efficacy of CSF circulatory system. Throughout this investigation, we also observed notable alterations in the expression patterns of AQP4 isoforms. Building upon these insights, in the third and fourth segments of our research, we delve deeper into examining the impacts of selectively regulating AQP4 isoforms and the membrane localization protein SNTA1 on the functionality of the glymphatic system post-ischemic stroke.

Specifically, we initially assessed the temporal progression of cerebral edema following tMCAO using MRI. We then investigated the dynamic changes in glymphatic system dysfunction with an MRI contrast agent. Cerebral edema can be categorized into two phases based on BBB integrity. In the early stage, edema is cytotoxic and ionic, occurring without BBB damage; in the later stage, vasogenic edema develops as the BBB becomes compromised. Our data align with previous reports, which indicate that the peak of vasogenic edema occurs 1–3 days post-reperfusion [31,32]. Zhu et al. suggested that after stroke, the opening of the BBB facilitates fluid influx into the brain parenchyma rather than outflow, primarily due to hydrostatic forces and osmotic gradients [32]. Notably, the five regions we observed that ischemic stroke affects CSF flow not only in the infarcted hemisphere but also in the contralateral hemisphere and throughout the brain. This may be attributed to alterations in overall cerebral hemodynamics and the blood–brain barrier disruption. Unilateral brain injury in traumatic brain injury studies has similarly impaired global CSF inflow along glymphatic pathways, suggesting an interdependence of the glymphatic system across both hemispheres [33]. These findings imply that restoring glymphatic function in the contralateral hemisphere could be a therapeutic target for enhancing stroke prognosis.

Subsequently, we employed TGN-020, an inhibitor of AQP4, a protein critically involved in edema pathogenesis, to mitigate edema formation and observe functional and structural alterations within the glymphatic system. We observed alterations in AQP4 polarity and isoforms distribution, underscoring the link between glymphatic system dysfunction and cerebral edema development post-stroke. TGN-020 alleviates glymphatic system burden by mitigating astrocyte structural changes and AQP4 mis-localization. Prior studies have documented the neuroprotective effects of inducing AQP4, and AQP4 inhibitors have been shown to restore neurological function in ischemic stroke [34,35]. Cognitive impairments and other neurological deficits are common after ischemic stroke, with lymphatic system dysfunction potentially exacerbating these issues [36]. Our results found that enhancing the brain parenchymal channel structure of the glymphatic system can substantially improve neural function prognosis. Clinical studies have also linked enlarged perivascular spaces to post-stroke depression, suggesting that maintaining a healthy lymphatic system is crucial for recovery [37,38]. Brain tissue sequencing revealed no differences in AQP4 and SNTA1 gene expression but showed ubiquitination pathway alterations. This may be one of the mechanisms affecting glymphatic system dysfunction in ischemic stroke. Although TGN-020 is proposed as a direct pore blocking inhibitor of the AQP4, emerging evidence challenges its specificity in regulating AQP4 expression [39,40]. It suggests potential interactions with additional molecular pathways. These findings indicate that TGN-020 exhibits off-target effects beyond its direct interaction with AQP4, potentially involving other aquaporins, ion channels, or indirect influence on AQP4 expression through signaling molecules.

Different AQP4 isoforms have distinct functions: the AQP4-M1 isoform primarily regulates water transport, while the AQP4-M23 isoform predominantly contributes to the structural integrity of OAPs [41]. OAPs are closely related to AQP4 polarity, and the distribution of AQP4 isoforms changes following ischemic stroke. An increase in the AQP4-M1 to AQP4-M23 ratio is associated with AQP4 mis-localization. Si et al. observed a decrease in the expression and proportion of AQP4-M23 and AQP4-M1 in Parkinson’s disease mice, which affected the extent of AQP4 coverage along the perivascular area and led to glymphatic system dysfunction [30]. This observation suggests a novel mechanism through which AQP4 redistribution can influence glymphatic system function and provide a potential approach to improve ischemic stroke outcomes.

As SNTA1 regulates AQP4 localization, co-immunoprecipitation revealed their interaction, in which AQP4-M23 isoform has a stronger effect, and studies have shown that SNTA1 deletion mainly affects AQP4-M23 isoform. Overexpression of SNTA1 can reverse glymphatic system dysfunction and is also related to enhanced polarity of AQP4, which is better reflected in the 3D reconstruction image, which also provides a new idea for treatment.

This study has several limitations. Firstly, although TGN-020 has been extensively utilized in research concerning the glymphatic system's function in neurological disorders, akin to the majority of fundamental animal disease studies, our research evaluated the efficacy of this compound over a relatively brief timeframe. Consequently, there is a paucity of data regarding the long-term therapeutic potential of TGN-020 in disease contexts. Secondly, we note that while TGN-020 has been widely used as an AQP4 inhibitor in previous studies, its pharmacological specificity may not be exclusive. Some of its effects could be attributed to interactions with other molecular targets, including the differential expression of various AQP4 isoforms. This complexity in its mode of action calls into question the pure AQP4 − mediated nature of its effects. To more accurately verify its impact on the glymphatic system, transgenic animal models with AQP4 gene knockout can be employed. These models would help isolate the role of AQP4 and provide more conclusive evidence regarding the compound's mechanism of action. Lastly, due to the technical difficulties associated with directly evaluating the glymphatic system in human subjects, there is insufficient evidence to confirm the role of AQP4 in the post-stroke recovery of the human glymphatic system. This clearly emphasizes the need for continuous research into the underlying mechanisms of human glymphatic system repair. Such research is crucial for translating pre − clinical findings to clinical applications and developing more effective therapeutic strategies for stroke patients.

In conclusion, our study demonstrates that modulating AQP4 isoforms can influence glymphatic system function, presenting a potential therapeutic pathway for ischemic stroke. These findings enhance our understanding of stroke pathophysiology and suggest new directions for future research and clinical applications.

Conclusion

Following ischemic stroke, glymphatic system dysfunction develops dynamically cerebral edema formation, wherein AQP4 mis-localization is a key contributor to glymphatic system impairment. This study elucidates that edema-induced upregulation of AQP4 isoform expression constitutes a fundamental molecular mechanism underlying AQP4 mis-localization in the tMCAO model. We demonstrate that AQP4 pharmacological inhibition or isoform modulation restores AQP4 polarity, enhances glymphatic clearance function, provides neuroprotection and metabolic homeostasis in tMCAO mice. These results suggest that therapeutic strategies combining AQP4 inhibition with AQP4 isoforms regulation may represent a promising approach for restoring glymphatic function in ischemic stroke pathology.

Compliance with Ethics Requirements.

All Institutional and National Guidelines for the care and use of animals (fisheries) were followed.

CRediT authorship contribution statement

Hanhong Zhang: Methodology, Data curation, Formal analysis, Writing – original draft. Jinjing Wang: Methodology, Formal analysis. Siyuan Zhang: Data curation, Formal analysis. Dingyi Yan: Validation. Yiran Dong: Formal analysis. Pan Zhang: Writing – review & editing. Wen Sun: Conceptualization, Project administration. Xinfeng Liu: Project administration, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

The following are the Supplementary data to this article: