Melatonin Regulates Glymphatic Function to Affect Cognitive Deficits, Behavioral Issues, and Blood–Brain Barrier Damage in Mice After Intracerebral Hemorrhage: Potential Links to Circadian Rhythms

Male C57BL/6 mice (8–10 weeks, 22–24 g) from Beijing Vital River Laboratory were housed in a temperature‐controlled room with a 12‐h light/dark cycle and ad libitum access to food and water. Procedures were approved by Tianjin Medical University General Hospital's Animal Ethics Committee.

ICH model was prepared as described previously [26]. Using a microinjection pump (Shenzhen Ruiwode Life Science Technology Co. Ltd., China), 0.65 μL of Collagenase IV (Solarbio, C8160, 0.3 mg/mL) was injected at a rate of 0.5 μL/min. On the first day post‐surgery, mice without motor deficits were excluded from the study.

Melatonin (L132758, Aladdin) was used to reset the circadian rhythm, with studies showing that a dose of 10 mg/kg is the most effective and safe with no side effects [21]. Luzindole (L132758, Aladdin, 30 mg/kg), a melatonin receptor antagonist, assessed circadian rhythms and responses. Mice were divided into four groups: Sham, ICH + Vehicle, ICH + Mel (ICH + Melatonin), and ICH + Mel+Lu (ICH + Melatonin+Luzindole) (Figure 1). Collagenase was injected to establish the ICH model except for the Sham group, which received saline. Melatonin and Luzindole were dissolved in DMSO and diluted with saline. ICH + Mel and ICH + Mel+Lu groups received daily intraperitoneal injections of melatonin or Luzindole, respectively. Sham and ICH groups received saline injections.

Neurological function was assessed using the mNSS and corner test to evaluate motor, sensory, reflex, and balance functions [3]. The mNSS scores range from 0 (normal) to 18 (severe impairment). The corner test measures sensory, motor and postural asymmetry based on left turn percentages. Memory was evaluated with the novel object recognition (NOR) test [27, 28], where exploration times for familiar and novel objects were recorded and analyzed to compute the discrimination index. This can be expressed by the following equation: Discrimination index = (Tnovel − Tfamiliar)/(Tnovel + Tfamiliar).

Brain water content was measured using a wet/dry method [3]. On days 1, 3, 5, 7, and 14 post‐ICH, brains were divided into the ipsilateral hemisphere, contralateral hemisphere, and cerebellum. The wet weight was measured, then the brains were dried at 100°C for 24 h. Brain water content = [(Wet Weight − Dry Weight)/Wet Weight] × 100%.

As previously described [29], CBF was assessed using the PeriCam PSI system (Perimed AB, Sweden) on the day before and on days 1, 3, 7, and 14 post‐ICH. Perfusion data were analyzed with PIMsoft 1.2 software, selecting five specific locations as regions of interest (ROI) for each sample. Average perfusion values were calculated.

As described in previous research [3], mouse brains were collected on days 1, 3, 7, and 14 post‐ICH and sectioned into 1‐mm‐thick coronal slices. Hematoma volume was calculated using ImageJ software with the formula V = t × (S1 + S2 + … + Sn), where V represents the hematoma volume, t is the slice thickness, and S represents the hematoma area.

To assess BBB permeability, two methods were employed: Evans blue (EB) extravasation absorbance measurement and fluorescence imaging [29, 30]. On days 1, 3, 5, 7, and 14 post‐modeling, 2% EB was injected via the tail vein and allowed to circulate for 2 h. Mice were then perfused with saline, and their brains were collected. The brain tissues were homogenized, centrifuged, and the absorbance of the supernatant was measured at 610 nm using a spectrophotometer. The concentration was quantified using a standard curve and normalized to tissue weight (μg/g).

For fluorescence imaging, 30‐μm thick coronal sections were prepared using a cryostat and stained with DAPI (blue). Fluorescence images of EB (red) were then captured using an inverted fluorescence microscope (1.25×/4×/10×). Finally, the permeability of the BBB was assessed by quantifying the percentage of the area occupied by red fluorescence (EB) relative to blue fluorescence (DAPI) using ImageJ software.

As previously described [28, 31], on day 3 post‐ICH, Hamilton 25 μL syringes and 33‐gauge needles were used to inject 10 μL of 2.5% 70 kDa rhodamine B isothiocyanate dextran (RITC‐dextran) at 1 μL/min into the cisterna magna, followed by a 30‐min circulation. Mice were then perfused, and brains were fixed overnight in 4% paraformaldehyde at 4°C. Coronal sections (100 μm thick) were prepared using a Leica Biosystems CM 1950 microtome. Fluorescence images were captured with an Olympus IX 73 inverted fluorescence microscope after DAPI staining. Quantitative analysis was conducted using ImageJ software.

To assess the clearance of interstitial metabolites in the brain [13], on day 3 post‐ICH, a Hamilton 5 μL syringe and a 33‐gauge needle were used to inject 0.5 μL of 2.5% 70 kDa RITC‐dextran at a rate of 0.2 μL/min into the hippocampus (2.0 mm posterior to bregma, 1.5 mm lateral to the right, 2 mm below the skull). One hour later, the mice were perfused with cold PBS, and the brains were extracted and fixed overnight in 4% paraformaldehyde at 4°C. Coronal sections (100 μm thick) were prepared, and fluorescent images were captured after DAPI staining. Quantitative analysis was performed using ImageJ software.

As previously described [26], on days 1, 3, 5, 7, and 14 post‐ICH, mice were anesthetized, and 8 μL of 2% EB was injected into the cerebellomedullary cistern at a rate of 2 μL/min. Cervical tissues were then dissected to expose dCLN, which were subsequently collected. Representative images of EB drainage from the dCLN were captured using an Olympus SZX‐7 microscope. We collected all dCLNs from both the left and right sides of each mouse. If a mouse had more than one dCLN on either side, all dCLNs from that side were pooled together and treated as a single data point for analysis. The dCLNs were homogenized, centrifuged, and the supernatant was transferred to a 96‐well plate. EB concentration was measured at 610 nm, and actual EB concentrations were calculated from a standard curve using Microsoft Excel.

To further investigate the drainage of dCLN on day 5 post‐ICH, bilateral dCLN were extracted 30 min after the injection of RITC‐dextran. The dCLN were stained with DAPI, and representative fluorescence images were captured using an Olympus IX 73 microscope [32]. The dCLN were then sectioned into 10 μm thick frozen slices, restained with DAPI, and imaged again. Fluorescence areas of RITC‐dextran (red) and DAPI (blue) were quantified using ImageJ software, and the percentage of the red area relative to the blue area was calculated using Microsoft Excel.

As previously described [31], on day 3 post‐ICH, brain sections were analyzed by immunofluorescence. Following extraction, brain tissue was dehydrated in a gradient series of 15% and 30% sucrose solutions. The tissue was then embedded in OCT compound (Sakura Finetek USA) and sectioned into 10 μm coronal slices. Sections were blocked for 1.5 h with PBS containing 3% bovine serum albumin, 0.2% Triton X‐100, and 0.05% Tween 20, and then incubated overnight at 4°C with primary antibodies: rabbit anti‐AQP4 (1:800, 59678, Cell Signaling), mouse mAb GFAP (1:500, 3670S, Cell Signaling), and goat anti‐CD31 (1:500, AF3628, Bio‐Techne). Subsequently, slices were incubated for 1 h with species‐specific fluorescently labeled secondary antibodies. After DAPI staining, fluorescence images were captured and analyzed using ImageJ software.

AQP4 polarization was assessed on day 3 post‐ICH [12, 33]. ImageJ was used to quantify AQP4 polarization levels in blood vessels and surrounding regions, with ROI chosen manually in a blinded manner. The fluorescence intensity of AQP4 labeling was measured along a 100 μm‐long axis perpendicular to the direction of the blood vessel, generating a linear fluorescence profile extending from the brain tissue into the vessel and back into the surrounding brain tissue. Threshold analysis was used to measure the percentage area of AQP4 immunofluorescence greater than or equal to that surrounding the blood vessel (AQP4% area). Polarization was represented as the percentage area where AQP4 immunoreactivity was lower than that surrounding the vessel (1 − AQP4% area).

As previously described [28], protein samples from the brain tissue and hippocampus were collected on day 3 post‐ICH. The proteins were transferred to polyvinylidene fluoride (PVDF, Millipore) membranes, which were then blocked with 5% non‐fat milk. The membranes were incubated overnight at 4°C with the following primary antibodies: AQP4 (1:1000, sc‐32739, Santa Cruz), GFAP (1:1000, Rabbit mAb #12389S, Cell Signaling), and β‐Actin (1:2000, TA‐09, ZSGB‐BIO). Following this, the membranes were incubated with secondary antibodies for 1 h. Protein bands were visualized using the ChemiDoc Touch Imaging System and quantified using ImageJ software.

All data were statistically analyzed using GraphPad Prism 10.3. Data are presented as mean ± standard error of the mean (SEM). The normality of data distribution was evaluated using the Shapiro–Wilk test. If the data were normally distributed, One‐way analysis of variance (ANOVA) was used to assess group differences. For data that were not normally distributed, non‐parametric tests (Kruskal–Wallis test) were applied. Two‐way ANOVA and Tukey's post hoc test were employed to evaluate the differences among multiple groups with repeated measurements. A p‐value less than 0.05 was considered statistically significant.

Early behavioral deficits and cognitive impairments are common following ICH and are associated with poor outcomes [34, 35]. To assess the effects of biological regulation on motor function in ICH mice, we performed the mNSS (Figure 2A) and the Corner Test (Figure 2B). Significant behavioral deficits were observed in all experimental groups within 24 h post‐ICH. The mNSS showed that melatonin treatment accelerated recovery compared to the ICH + Vehicle group, with differences evident from days 3 to 14, and the most pronounced on day 5. Luzindole inhibited the therapeutic effects of melatonin. Although trends were similar, the mNSS appeared to be more sensitive than the Corner Test in assessing behavioral deficits.

To evaluate the impact of melatonin and luzindole treatment on cognitive impairments in ICH mice, we conducted the NOR test (Figure 2C–F). Results indicated that the ICH + Vehicle group spent significantly less time around the novel object compared to the ICH + Mel group (Figure 2F). Melatonin treatment significantly increased the time spent exploring the novel object. However, pre‐treatment with luzindole negated this improvement (Figure 2F). Overall, these data suggest that melatonin has neuroprotective effects on cognitive impairments induced by ICH, but luzindole can reverse these effects.

Brain edema plays a key role in the pathophysiology following ICH and is closely associated with disturbances in GS function [36]. To explore how melatonin regulates brain edema through its effects on the GS, we specifically measured brain water content (Figure 2G–I). The results showed that changes in brain water content were consistent with behavioral assessments, with significant differences observed between the melatonin and vehicle groups from days 3 to 7, peaking on day 5. Luzindole treatment reversed these effects. The most pronounced changes were seen in the ipsilateral brain (Figure 2G), followed by the contralateral brain (Figure 2H), while no significant differences were noted in the cerebellum (Figure 2I). These findings indicate a strong association between brain edema and behavioral impairments.

Additionally, since the maintenance of GS and BBB functions is closely related to CBF [6], we employed laser speckle imaging to evaluate CBF changes before and after ICH (Figure 2J–Q). Baseline CBF measurements were taken before ICH. CBF significantly decreased on day 1 post‐ICH and began to recover by day 3. The ICH + Mel group exhibited faster recovery compared to the ICH + Vehicle group, with significant differences noted on days 3, 7, and 14. Luzindole treatment moderately inhibited this recovery (Figure 2L–Q). By day 14, CBF in the ICH + Mel group was higher than in the sham group, possibly due to vascular proliferation, while the sham group also experienced mild damage from vehicle injection. Overall, melatonin treatment improved cortical CBF following ICH, whereas luzindole diminished this therapeutic effect.

We assessed hematoma absorption in ICH mice by measuring hematoma volume in brain slices at days 1, 3, 7, and 14 post‐ICH (Figure 3A–F). BBB damage and repair were evaluated by EB extravasation using a spectrophotometer (Figure 3G–I) and EB fluorescence staining (Figure 3J–M). Mice in the ICH + Mel group showed smaller hematoma volumes on days 3 and 7 compared to the ICH + Vehicle group, indicating that melatonin significantly enhanced hematoma absorption. No significant difference was observed between the ICH + Mel and the ICH + Mel + Lu groups, suggesting luzindole did not strongly inhibit melatonin's effects. EB extravasation results showed that melatonin treatment reduced EB leakage on days 3 and 5 compared to the ICH + Vehicle group (Figure 3G,H). On day 5, luzindole significantly inhibited melatonin's positive effects (Figure 3H), though not on day 3 (Figure 3G). This suggests that melatonin may significantly promote BBB repair in the early stages, while the inhibitory effect of luzindole may require time to accumulate and become more pronounced in the later stages, thereby suppressing this repair process. EB fluorescence staining on day 3 showed significant differences between the ICH + Mel and the ICH + Mel + Lu groups (Figure 3L,M), indicating its sensitivity compared to spectrophotometric measurements, making it easier to detect earlier changes.

To investigate the changes in CSF influx following ICH and the effects of melatonin and luzindole treatment on this function, RITC‐dextran was injected into the cisterna magna on day 3 post‐ICH (Figure 4A,B). The functionality of GS was assessed by observing tracer accumulation. Quantitative analysis (Figure 4C) revealed a significant reduction in CSF tracer penetration into the brain. In the ICH + Mel group, CSF tracer influx was significantly increased, whereas a significant statistical difference was also observed between the ICH + Mel + Lu and ICH + Vehicle groups, indicating that melatonin treatment improved GS influx function, while luzindole treatment reversed this effect.

We also compared the influx function across different brain sections (Figure 4D), finding significant regional differences in recovery. In the hematoma sections (Figure 4E) and hippocampal sections (Figure 4F), the ICH + Mel group exhibited significantly more fluorescence in the ipsilateral hematoma sections compared to the ICH + Vehicle group, while luzindole inhibited this therapeutic effect. Melatonin treatment significantly affected both hemispheres in the hippocampal sections (Figure 4H), but in the hematoma sections, it had a significant effect only on the ipsilateral side with no significant difference on the contralateral side (Figure 4G). This indicates that melatonin treatment effectively promotes recovery of influx function in the ipsilateral hematoma region and significantly improves influx function in both hemispheres of the hippocampus.

To investigate the impact of melatonin and luzindole treatment on CSF influx following ICH, we injected RITC‐dextran into the hippocampus and assessed interstitial solute clearance by comparing residual tracer levels across different experimental groups (Figure 5A,B). Quantitative analysis (Figure 5C) revealed that the ICH + Vehicle group showed a significantly increased tracer retention in brain tissues, whereas the ICH + Mel group exhibited a substantial reduction in tracer retention. Luzindole partially reversed the beneficial effects of melatonin. Significant differences in solute clearance recovery were observed across various brain sections (Figure 5D), indicating variable contributions of brain regions to solute clearance.

Recent studies highlight the dCLN as key sites for brain solute clearance. To quantitatively assess CSF solute clearance, we performed EB absorbance assays on dCLN (Figure 5E). Results showed a significant decrease in EB excretion starting from day 1 post‐ICH (Figure 5F). By days 3, 5, and 7, the ICH + Mel group exhibited faster EB excretion recovery compared to the ICH group, with the most pronounced difference on day 5 (Figure 5G–I). Luzindole significantly inhibited this effect (Figure 5H). By day 14 post‐ICH, EB excretion in the untreated ICH + Vehicle group remained significantly different from the sham group, while no significant differences were observed between the sham group and either the ICH + Mel or ICH + Mel + Lu groups (Figure 5J). Melatonin treatment showed a trend of faster recovery (Figure 5K). To further validate these results, we collected dCLN from mice 30 min after RITC‐dextran injection into the cisterna magna on day 5 post‐ICH and performed immunofluorescence detection (Figure 5L,M). Results confirmed that the ICH + Mel group had higher fluorescence values compared to the ICH + Vehicle group, while luzindole treatment significantly suppressed this effect (Figure 5N). In summary, melatonin treatment significantly accelerated solute elimination, whereas luzindole partially counteracted this effect.

Evidence indicates that the function of GS depends on astrocytic end‐feet AQP4, and co‐staining of AQP4 and GFAP is commonly used to assess reactive astrocyte proliferation. Our results (Figure 6A,B) show that on day 3 post‐ICH, GFAP expression is significantly higher in the ICH + Vehicle group compared to the sham group. Melatonin treatment significantly reduces GFAP levels, while luzindole partially reverses this effect. Similar results were observed in GFAP Western blot experiments (Figures 6C,D and S1A–D), although no significant difference was found between the ICH + Mel and ICH + Mel + Lu groups. AQP4 polarization analysis (Figure 6E–G) indicates that melatonin treatment restores AQP4 polarization around blood vessels, whereas luzindole partially inhibits this trend. Western blot analysis (Figures 6H,J and S2A–D) shows that melatonin significantly restores AQP4 accumulation in peri‐hematoma tissues, while luzindole inhibits this therapeutic effect. No significant differences were observed in AQP4 Western blot results among the groups in the hippocampus (Figures 6I,K and S3A–D).

The data that support the findings of this study are available from the corresponding author upon reasonable request.