Adiponectin Ameliorates GMH-Induced Brain Injury by Regulating Microglia M1/M2 Polarization Via AdipoR1/APPL1/AMPK/PPARγ Signaling Pathway in Neonatal Rats

Adiponectin (APN), a fat-derived plasma hormone, is a classic anti-inflammatory agent. Multiple studies have demonstrated the beneficial role of APN in acute brain injury, but the effect of APN in germinal matrix hemorrhage (GMH) is unclear, and the underlying molecular mechanisms remain largely undefined. In the current study, we used a GMH rat model with rh-APN treatment, and we observed that APN demonstrated a protective effect on neurological function and an inhibitory effect on neuroinflammation after GMH. To further explore the underlying mechanisms of these effects, we found that the expression of Adiponectin receptor 1 (AdipoR1) primarily colocalized with microglia and neurons in the brain. Moreover, AdiopR1, but not AdipoR2, was largely increased in GMH rats. Meanwhile, further investigation showed that APN treatment promoted AdipoR1/APPL1-mediated AMPK phosphorylation, further increased peroxisome proliferator-activated receptor gamma (PPARγ) expression, and induced microglial M2 polarization to reduce the neuroinflammation and enhance hematoma resolution in GMH rats. Importantly, either knockdown of AdipoR1, APPL1, or LKB1, or specific inhibition of AMPK/PPARγ signaling in microglia abrogated the protective effect of APN after GMH in rats. In all, we propose that APN works as a potential therapeutic agent to ameliorate the inflammatory response following GMH by enhancing the M2 polarization of microglia via AdipoR1/APPL1/AMPK/PPARγ signaling pathway, ultimately attenuating inflammatory brain injury induced by hemorrhage.


INTRODUCTION
Germinal matrix hemorrhage (GMH) is the most common type of intracranial hemorrhage to occur in preterm neonates and is caused by the rupture of blood vessels in the periventricular subependymal immature region (i.e., germinal matrix) (1). Destruction of the precursor cells within the germinal matrix and post-hemorrhagic hydrocephalus in infants suffering from GMH (2) may lead to the development of significant long-term neurocognitive sequelae, including developmental delay, seizures, and cerebral palsy (3). Unfortunately, no therapies have been shown to be effective in treating neonatal GMH, and the only preventive intervention is perinatal glucocorticoids (4). As the rates of survival in preterm births and neonates have substantially increased in recent decades, GMH-induced neurological deficits are increasingly evolving into a substantial socioeconomic burden. Therefore, safe and effective pharmacologic treatments are desperately needed.
The initial pathological damage of cerebral hemorrhage is the mechanical compression to adjacent tissues caused by hematoma (5). Secondary damage is known to be triggered by blood clots that contribute to neuroinflammation and neurological dysfunction, including post-hemorrhagic hydrocephalus (6), reactive astrocytosis (7), and microgliosis (8). Secondary brain injury has crucial roles in the prognosis of neurological deterioration in hemorrhagic stroke, including intracerebral hemorrhage, subarachnoid hemorrhage, and GMH. As therapeutic developments against primary injury have demonstrated definitive benefits in clinical trials, investigators have instead focused on exploring the mechanisms that underlie post-GMH secondary brain injury in the pursuit of novel targets for treatment.
Increasing evidence suggests that inflammatory reactions play a key role in the pathogenesis of stroke and contribute to secondary brain injury after hemorrhage (9,10). Microglia/ macrophages are the major immune cells of the central nervous system (CNS) and are critical drivers of the neuroinflammatory response after various hemorrhagic brain injuries (11). Upon activation, microglia develop into either classically activated (M1) or alternatively activated (M2) phenotypes, a process termed polarization. M1 microglia are initially present following an insult as they promote an inflammatory response. Conversely, M2 microglia mainly secrete anti-inflammatory cytokines and growth factors, and may facilitate neuronal recovery, hematoma resolution, and vascular remodeling. Additionally, in response to distinct microenvironmental cues, an already fully polarized M1 or M2 sub-population can reverse its phenotype and function (12,13). Therefore, strategies aiming to promote the phenotypic conversion of microglia from M1 to M2 might provide therapeutic potential for GMH.
Adiponectin (APN) is a plasma hormone predominantly secreted by adipocytes, and is known to exert powerful antiapoptosis/anti-inflammatory effects in acute and chronic brain injury, including ischemic or hemorrhagic stroke, mainly through the Adiponectin receptor 1 (AdipoR1) and AMPactivated protein kinase (AMPK) pathway (14). We previously found that adiponectin could exert neuroprotective effects by attenuating neuronal apoptosis via AdipoR1/AMPK signaling pathway after neonatal rats are subjected to hypoxic ischemic injury (15). Recently, emerging evidence has shown adiponectin signaling pathway has a detrimental role in regulating the immune response via microglia from the central nervous system (CNS) (16,17). Furthermore, AMPK signaling pathway activation could accelerate hematoma resolution and improve neurological outcomes after intracerebral hemorrhage (ICH) (18).
Thus, in this present study, we hypothesize that APN will attenuate neuroinflammation and accelerate hematoma clearance through the AdipoR1/AMPK signaling pathway after GMH via regulation of M1/M2 polarization.

Animals
Two hundred and fifty-five Postnatal day (P) 7 Sprague-Dawley neonatal pups (weight = 12-15 g, Harlan, Livermore, CA) of either sex were randomly assigned into either Sham (n=56) or GMH (n=199) groups. Of the 199 rats in the GMH surgery group, no pups were excluded from the study as a result of intraoperative death. All the rat pups were stored in temperaturecontrolled rooms with ad libitum access to drink and chow. The design of experiments, timeline, and number of animals within each group are depicted in Supplementary Figure S1.

Germinal Matrix Hemorrhage (GMH) Induction
GMH induction was conducted in unsexed P7 rats and was accomplished via collagenase infusion as previously described (19,20). Briefly, rat pups weighting between 12g-15g were anesthetized using 3% isoflurane, and anesthesia maintenance was conducted using 1.5% isoflurane intraoperatively. After, a stereotaxic frame was used to fix the pups in place, and the scalp was then sterilized. Next, the bregma was exposed via incision, and a burr hole (1 mm) was drilled at 1.6 mm right lateral and 1.5 mm rostral relative to bregma. A 10 µl Hamilton syringe (Hamilton Co, Reno, NV, USA) filled with 0.3 U/µl collagenase solution was fixed to an infusion pump (Harvard Apparatus, Holliston, MA, USA), and the needle was inserted 2.8 mm below the dura with the bevel facing the midline. Next, 0.3 U type VII-S collagenase from Clostridium histolyticum (Sigma Aldrich St. Louis, MO, USA) was infused at a rate of 0.3 U/ 3min, and the needle remained in place for 5 min before being removed at a rate of 1 mm/min to prevent possible leakage. Bone wax was then used to seal the burr hole, and 5-0 silk was then used to suture the incision closed. In the post-operative period, the animals were placed on a 37°C heating pad to recover and awaken from anesthesia. After recovery, the rat pups were returned to the dam. The sham surgery followed the same procedure as described above, but without collagenase infusion. The average duration of surgery for each animal was approximately 30 min.

Drug Administration
Recombinant human adiponectin protein (rh-APN) (Catalog Number: 450-24, Pepro Tech, NJ, USA) was administered by intranasally at 1 h post-GMH in three different doses (0.05 mg/ kg, 0.1 mg/kg, and 0.3 mg/kg) as described previously (15). The rh-APN was administered intranasally once daily for 3 days post-GMH in the short-term experiment or 7 days using the best dosage in the long-term experiment.

Western Blot Analysis
Brain tissues were homogenized with RIPA lysis buffer (sc-24948, Santa Cruz Biotechnology, Dallas, TX, USA) for a duration lasting at least 60 s. The homogenate was centrifuged at 15,000 rpm/min at 4°C for 20 min. The supernatant was then collected, aliquoted, and stored at -80°C. Protein concentrations were determined via DC protein assay (Bio-Rad, USA). Western blots were processed according to previously described protocol (15). Next, 30mg of protein from each sample were placed into wells of 10% gels. After which, each was run for 30 min at 80 V, followed by 90 min at 120 V. The proteins were then relocated onto 0.2mm or 0.4mm nitrocellulose membranes at 100 V for 120 min (Bio-Rad, USA). The following primary antibodies were applied to the membranes, which were allowed to incubate overnight at 4°C

Hemoglobin Assay
Spectrophotometric measurements to assess hemorrhagic volume were conducted as previously described (21). The extracted, frozen forebrain tissues were distributed into individual glass test tubes containing 3 mL of PBS, followed by homogenization for 60 s with the help of Tissue Miser Homogenizer (Thermo Fisher Scientific, Waltham, MA, USA). Next, the erythrocyte membranes were lysed via 1 min of ultrasonication. After which, the products were centrifuged for 30 min, which then allowed the supernatant to be separated from the pellets. Then, Drabkin's reagent was added (Sigma-Aldrich, USA) into aliquots of supernatant in a ratio of 4:1, which were allowed to react for 15 min. Finally, a spectrophotometer (540; Genesis 10uv; Thermo Fisher Scientific, Waltham, MA, USA) was used to calculate absorbance into a hemorrhagic volume (µl) according to the basis of a standard curve.

Neurological Tests
Neurobehavioral function was evaluated blindly in a random and unbiased protocol as previously reported (22). Short-term neurological tests, such as righting reflex and negative geotaxis tests, were performed on days 1 to 3 after GMH. Several longterm neurological tests, including water maze, rotarod, and footfault, were performed on days 21 to 28 after GMH.

Righting Reflex Test
The time for each pup to completely rollover from supine to prone was recorded. The maximum recording time was 60 s (3 trials/pup/day). The values from each of the three trials were used to calculate the average.

Negative Geotaxis Test
Pups were placed head downward onto a 45°C inclined plane, and the time to rotate 180°C was recorded. The maximum duration of each trial was 20 s (3 trials/pup/day). The values from each of the three trials were used to calculate the average.

Water Maze Test
Morris water maze tests were conducted to assess animal memory and learning capacity as previously described. The apparatus consisted of a metal pool (110 cm in diameter) accompanied by a small platform (11 cm in diameter), in which the pups could climb to escape the water. Swim distance, latency, and velocity were digitally recorded and analyzed using a tracking software (Noldus EthoVision, USA). In the cued test, pups were manually guided to the platform if they had difficulty in locating it, and the platform location was moved every other trial. In the hidden tests, the platform was submerged 1 cm below the water, and the duration of time spent in probe quadrant was recorded.

Quantitative Real-Time PCR (qRT-PCR) Analysis
To assess gene expression levels, total RNA was isolated from brain tissue using RNeasy kit (Catalog Number: 74106, Qiagen) and cDNA were synthesized using GoScript Reverse Transcriptase (Catalog Number: PRA5000, Promega) following the manufacturer's instructions as previously described (19). The RNA expression was measured by qRT-PCR with specific primers ( Table 1) using iQ ™ 5 real-time PCR detection systems (Bio-Rad, USA).

Nissl Staining
On day 28 after GMH, we euthanized the rats, harvested the brains, and fixed them in 10% formalin for histology preparation. The brains were sliced into 16mm-18mm thick slices (Leica Microsystems, LM3050S) after being embedded with Optimal Cutting Temperature (OCT, Fisher Scientific, Waltham, MA, USA). Then, the coronal brain slices were stained with 0.5% Cresyl violet, followed by microscopic imaging (Olympus-BX51). Ventricular volume, white matter loss, and cortical thickness were evaluated as reported previously (19).

Statistical Analysis
All data are presented as mean ± SD. Statistical analysis was performed using SPSS v.24.0 (IBM Corp., USA). Statistical differences were performed using one-way ANOVA with Dunnet's post-hoc test for multiple comparison and two-tailed Student's t test for two group comparisons. If the difference was P ≤ 0.05, it was considered statistically significant.

RESULTS
Time Course and Spatial Expression of Endogenous Adiponectin, AdipoR1, and AdipoR2 After GMH

Intranasal Administration of rh-APN Ameliorated Neurological Deficits and Attenuated Inflammation in 72 h After GMH
To explore and assess the translational treatment regimen of APN, three doses (0.05mg/kg, 0.1mg/kg, 0.3mg/kg) were administered in a single treatment intranasally 1 h post-GMH. More time was spent flipping to the prone position and rotating to the head upward position in vehicle-treated groups when compared to the Sham group at days 1 and 2 after GMH (Figures 2A, B). Improved short-term neurological function was observed in all three dosages of rh-APN-treated pups. Interestingly, the medium dose of rh-APN-treated pups demonstrated the best results in both righting reflex ( Figure 2A) and negative geotaxis tests ( Figure 2B), which was compared to sham group as early as 2 d post-GMH. Collectively, 0.1mg/kg of rh-APN treatment was used for each of the following experiments. Anti-human Adiponectin antibody was used to assess exogenous rh-APN levels in the brain after intranasal treatment. As shown in Figure 2C, more rh-APN were observed in brain tissue when compared to vehicle treated group. A key feature of stroke involves the secretion of inflammatory cytokines (23). The anti-inflammatory effects of APN have been reported in other studies of intracerebral hemorrhage (ICH) rat models (24). Herein, the protein expression levels of pro-inflammatory (IL-1b and IL-6) and anti-inflammatory cytokines (IL-10) in perihematomal brain tissue were detected. Pro-inflammatory cytokine expression dramatically increased, but anti-inflammatory cytokine expression levels were slightly increased 72 h after GMH ( Figure 2C). However, with the treatment of rh-APN, the marked increase of pro-inflammatory cytokines (IL-1b and IL-6) were blunted, which was further validated by immunostaining of IL-6 in the choroid plexus ( Figure 2D). Simultaneously, antiinflammatory cytokines, IL-10 ( Figure 2C) and Arginase-1 mRNA expression ( Figure 5H) by qRT-PCR were significantly promoted.

Rh-APN Treatment Improved Long-Term Neurological Function at 21 to 28 Days After GMH
To evaluate and explore rh-APN treatment effects on long-term GMH-induced neurological impairments, water maze, foot-fault, and rotarod tests were conducted 4 weeks post-GMH to evaluate and assess neurological function. In the Morris Water Maze evaluation, vehicle-treated rats swam significantly longer ( Figure 3A), spent a longer duration seeking the platform ( Figure 3B), and spent less time within the defined quadrant zone ( Figures 3C, D) when compared to the sham group. Meanwhile, rh-APN-treated rats had a significantly greater performance than vehicle-treated rats as demonstrated by shorter travel distance, decreased time to escape, and more time spent in the target quadrant zone (Figures 3A-D). However, there was no significant difference in swimming velocity among these three groups ( Figure 3E). Rats in the vehicle groups had a significantly shorter latency to fall at both of 5RPM and 10RPM acceleration compared to the sham group in the rotarod test. Rh-APN treatment significantly reduced the falling latency compared to the vehicle, However, there were no significant differences when compared with sham ( Figure 3F). Additionally, rh-APN treatment also significantly improved sensorimotor function after GMH in the foot fault test compared to the vehicle group ( Figure 3G). Regarding the growth profile, there was significantly slowed normal growth from 14 d to 28 d after GMH was observed in vehicle-treated animals compared to sham group, but rh-APN restored normal body weight ( Figure 3H).

GMH-Induced Microglial M1 Phenotype Polarization Was Prevented by rh-APN via AdipoR1 Signaling Pathway
There is a beneficial role of alternative polarization of microglia in ameliorating brain hemorrhage-induced inflammation, including GMH (25,26). To explore whether the microglia M1/M2 polarization involved in attenuating inflammation mediated by rh-APN treatment after GMH, the M1 phenotype marker (CD68) and M2 phenotype marker (CD206) were examined by immunostaining assay. Data showed that the percentage of CD68 and Iba1 double-positive cells increased at 72 h after GMH, but significantly decreased following rh-APN administration ( Figures 4A, B). Simultaneously, rh-APN treatment significantly increased the percentage of CD206 and Iba1 double-positive cells after GMH when compared with the vehicle group ( Figures 4A, C). Importantly, the ratio of M1-like (CD68 + Iba1 + ) cells/M2-like (CD206 + Iba1 + ) cells decreased with administration of rh-APN (Supplementary Figure S3E). However, the rh-APN-mediated effects in M1/M2 polarization, as described above, were reversed by knockdown of AdipoR1 with si-AdipoR1 ( Figures 4A-C). In sham animals, microglia remained in their ramified state with or without rh-APN treatment ( Figures 4A-C). Consistent with the immunofluorescence staining results, Western blot analysis showed that the expression of M1 marker, CD68, was significantly increased and M2 marker D206 expression was slightly increased at 72 h post-GMH. However, administration of rh-APN significantly decreased CD68 expression and promoted CD206 expression, and the effects were significantly reversed by si-AdipoR1 ( Figures 4D-F Values are expressed as mean ± SD. ANOVA, Dunnett. n = 6-9 for each group. *P < 0.05 compared to sham, # P < 0.05 compared to GMH + Vehicle, @ P < 0.05 compared to GMH + rh-APN (0.05mg/kg), & P < 0.05 compared to GMH + rh-APN (0.1mg/kg). (C) Western blot data showed that intranasal administration of rh-APN after GMH could be delivered successfully into brain tissue, and it reduced proinflammatory cytokine levels of IL-1b and IL-6 while increasing expression levels of anti-inflammatory cytokine, IL-10, within the brain. ANOVA, Dunnett. n = 6 for each group. *P < 0.05 compared to sham, # P < 0.05 compared to GMH + Vehicle. (D) Immunofluorescence staining assay showed IL-6 immunoreactivity in the choroid plexus was dramatically increased in pups that were subjected to GMH when compared with sham, and rh-APN treatment significantly reduced the IL-6 immunoreactivity compared to GMH.
increase population size via proliferation, a transformation from ramified to amoeboid morphology, and chemokine secretion (27). To explore whether treatment with rh-APN affected the post-GMH microglial response, pan microglia marker (Iba1), and activated microglia/macrophage marker (CD11b/c+) were detected in the periventricular area. Iba1 positive-stained microglia in both vehicle and rh-APN-treated groups showed significant morphological changes when compared to sham group at 72 h after GMH ( Figure 5A).
Because microglial proliferation exerts post-hemorrhage protective effects, Ki67 + /Iba1 + microglia, as well as BrdU + / Iba1 + microglia were evaluated in the periventricular area at 72 h post-GMH. As shown in Figures 5A-D, the number of Ki67 + /Iba1 + and BrdU + /Iba1 + microglia were significantly increased in rh-APN-and vehicle-treated pups compared to sham. However, there is no significant difference between rh-APN and vehicle groups with a slight increase (Figures 5B, D). Interestingly, the number of activated microglia (CD11b/c + ) was significantly increased in the periventricular area in rh-APN group when compared with the vehicle group ( Figures 5E, F). Importantly, immunofluorescence staining showed an increased accumulation of AdipoR1+/CD206+ cells ( Figure 5G) and CD206+ cells (Figures 6E, F) in the periventricular area with rh-APN treatment, indicating that the M2 microglia may account for the large increase in activated microglia.

rh-APN Administration Enhanced Hematoma Resolution and Reduced Ventricular Dilation by Increasing M2 Microglia in the Periventricular Region After GMH
Peroxisome proliferator-activated receptor gamma (PPARg) is a pivotal transcription factor and has a key role in the upregulation of CD36 expression (28). Multiple studies have shown that PPARg is a key participant in promoting polarization of microglia to the M2 phenotype in various experimental stroke models (29,30). Hence, expression of PPARg and CD36 expression were evaluated via Western blot analysis. The results revealed that exogenous rh-APN administration further augmented PPARg and CD36 expression at 72 h post-GMH ( Figures 4G, H).
Since a previous study demonstrated PPARg-induced upregulation of CD36 could enhance hematoma resolution and ameliorate secondary brain injury after GMH (21), hemoglobin assay time-course experiments were performed at 24 h, 72 h, and 7 d post-GMH to further explored the effects of rh-APN treatment on hematoma clearances. At 24 h after GMH, all groups had significantly greater hemoglobin content in the ipsilateral brain compared to sham, but there were no significant differences between GMH groups either with/ without treatment of rh-APN or si-AdipoR1 ( Figure 6B). However, rh-APN treatment had significantly reduced hemoglobin volume when compared with vehicle controls at 72 h after GMH, which had been reversed by AdipoR1 knockdown with si-AdipoR1 RNA (Figures 6A, C). Additionally, the vehicle-and rh-APN-treated groups had greater hemoglobin content compared to sham at 7 d after GMH, but the rh-APN-treated group had significantly less hemoglobin content compared to the vehicle group ( Figure 6D).
GMH. There were no CD206+ microglia observed in the sham group, but vehicle-treated pups demonstrated a subtle increase in the number of CD206+ microglia. Importantly, with the knockdown of AdipoR1 by si-AdipoR1 RNA, we observed fewer CD206-positive microglia in si-AdipoR1 group when compared to vehicle groups ( Figures 6E, F). By contrast, treatment of rh-APN significantly increased CD206+ microglia compared to vehicle, whereas this increase was abrogated by knockdown of AdipoR1 ( Figures 6E, F). Furthermore, double immunostaining showed that CD206 was co-located with hemoglobin (marker of hematoma) in the periventricular area in pups after GMH with rh-APN treatment, indicating rh-APN promoted hematoma resolution by CD206+ microglia mediated phagocytosis ( Figure 6G).
Additionally, to evaluate whether rh-APN alleviated the severity of ventriculomegaly, the ventricular volume was assessed at 28 d post-GMH. As shown in Supplementary Figure S3A, significant ventricular dilation occurred in the vehicle group when compared with sham. However, rh-APN treatment significantly reduced the ventricular volume when compared with the vehicle-treated pups (Supplementary Figures S3A, B). White matter loss was markedly increased in the vehicle group, but was significantly restored with administration of rh-APN (Supplementary Figures S3A, D). Concurrently, the cortical thickness was markedly decreased in the vehicle-treated pups, whereas rh-APN treatment significantly reduced the cortical loss (Supplementary Figures S3A, D).

rh-APN Attenuates Neuroinflammation and Promotes Hematoma Resolution via AdipoR1/APPL1/LKB1/AMPK Signaling Pathway at 72 h after GMH
To evaluate and assess whether AdipoR1/APPL1/LKB1/AMPK signaling protect neonatal brain against hemorrhage after GMH, the specific AdipoR1 siRNA, APPL1 siRNA, LKB1 siRNA, and their negative control (scramble siRNA) were intracerebroventricularly injected 24 h before GMH induction. Western blot analysis revealed that AdipoR1, APPL1, p-AMPK, PPARg, CD36, CD206, and IL-10 expression were each significantly increased in the vehicle group when compared with sham group (Figures 7A-C, E-H, K). Administration of rh-APN further upregulated their expression compared to the vehicle-treated group. In contrast, the M1 marker CD68 (Figures 7A, I) and proinflammatory cytokine IL-1b ( Figures 7A, G) expression was dramatically increased, whereas rh-APN treatment decreased their expression when compared with the vehicle group. Knockdown of AdipoR1 with AdipoR1 siRNA (si-AdipoR1) significantly decreased AdipoR1 expression and abolished the effects of rh-APN ( Figures 7A-K). As shown in Figure 7A, silencing of AdipoR1 expression was associated with decreased APPL1, AMPK phosphorylation, PPARg, CD36, CD206, and IL-10 expression, but was associated with increased CD68 and IL-1b expression after GMH with rh-APN treatment. Furthermore, knockdown of APPL1 and LKB1 expression showed similar trends as knockdown of AdipoR1 was associated with decreased AMPK phosphorylation and PPARg, CD36, CD206, and IL-10 expression ( Figures 7A, F-H, K), but with increased CD68 and IL-1b expression ( Figures 7A, I, J).

Selective Inhibition of AMPK/PPARg Signaling in Activated Microglia Cells Abolished the Effect of rh-APN on Inhibition of Neuroinflammation and Promotion of Hematoma Resolution
To specifically assess whether microglia AMPK/PPARg signaling mediated a protective effect of rh-APN post-GMH, we administered a specific AMPK inhibitor and PPARg antagonist into microglia cells intracerebroventricularly with Lipo-Dorsomorphin (1mg/g, 2ml) and Lipo-GW9662 (2mg/g, 2ml) 24 h Values are expressed as mean ± SD. ANOVA, Dunnett. n = 6 for each group. *P < 0.05 compared to sham, # P < 0.05 compared to GMH + Vehicle, @ P < 0.05 compared to GMH + rh-APN or GMH + rh-APN + Scramble siRNA.
before GMH was induced as described in our previous study (19). Green fluorescence-labeled liposomes were mainly found in activated IBA+ microglia at 72 h post GMH, whereas no labeled liposomes were observed in Neurons (Red, NeuN+) and Astrocytes (Red, GFAP+) ( Figure 8A). The addition of Lipo-Dorsomorphin and Lipo-GW9662 to rh-APN had no effect on AdipoR1 expression (Figures 8B, C). Phosphorylation of AMPK (Figures 8B, D) and PPARg ( Figures 8B, E) expression were significantly reduced in Lipo-Dorsomorphin-treated pups when compared to rh-APN with or without Lipo-PBS-treated pups. In addition, Lipo-GW9662 inhibited PPARg expression, but exerted no effect on AMPK phosphorylation ( Figures 8B, E, F), indicating that PPARg is a downstream kinase of AMPK after GMH. Administration of the two liposome-encapsulated inhibitors significantly increased CD68 expression but decreased CD36 and CD206 expression when compared with rh-APN or rh-APN + Lipo-PBS treated pups ( Figures 8B, F-H). Collectively, AMPK/PPARg inhibition in activated microglial cells restored post-GMH inflammation, indicating a key role of microglial AMPK/PPARg signaling to protect the neonatal brain from damage induced via hemorrhage.

DISCUSSION
Germinal matrix hemorrhage (GMH) is the most commonly occurring neurological disorder associated with premature Western blot data showed that rh-APN significantly decreased CD68 expression, whereas Lipo-Dorsomorphin and Lipo-GW9662 reversed the inhibitory effect of rh-APN. (All samples of GMH + rh-APN in Western blot were from the same animals which were euthanized after short-term neurobehavioral tests). Values are expressed as mean ± SD. ANOVA, Dunnett. n = 6 for each group. *P < 0.05 compared to sham, # P < 0.05 compared to GMH + Vehicle, @ P < 0.05 compared to GMH + rh-APN or GMH + rh-APN + Lipo-PBS.
newborns, with an incidence of 3.5 per 1000 live births (31), and is defined as blood vessel rupture of subependymal immature near the ganglionic eminence (32). Herein, we investigated the protective effects of APN on post-GMH brain injury and shed light on mechanisms that could be involved. We found that rh-APN could improve short-and long-term neurobehavioral outcomes by attenuating GMH-induced inflammation. Mechanistically, rh-APN decreased proinflammatory cytokines and increased anti-inflammatory cytokine secretion by preventing GMH-induced microglial polarization to the M1 phenotype while promoting the acquisition of the M2 phenotype. Furthermore, rh-APN enhanced hematoma resolution by promoting accumulation of M2 microglia with enhanced phagocytic ability in the periventricular regions. Finally, Western blot data revealed that knockdown of AdipoR1, APPL1, and LKB1, as well as selectively inhibited AMPK phosphorylation and PPARg in activated microglia reversed APN's effects. Taken together, in the present study, we found that APN facilitated the conversion of the M1 to the M2 microglial phenotype after GMH. AdipoR1-mediated activation of the APPL1/LKB1/AMPK/PPARg pathway had a critical role in the M1 to M2 transformation, as well as in the process of hematoma resolution. Growing evidence suggests that marked inflammation has crucial roles in GMH-induced secondary brain injury (33,34). Therefore, we evaluated inflammatory-related cytokines after GMH. Hemorrhagic brain injury in the setting of GMH induces a substantial increase in the production of proinflammatory cytokines (IL-1b and IL-6) within the brain, which may consequently progress to hydrocephalus. The post-GMH inflammatory response aggravated GMH-induced hydrocephalus, consequently incurring damage to tissues, disruption of the blood-brain barrier (BBB), and massive brain cell death. Following hemorrhagic stroke, the damaged neurons and astrocytes release various pro-inflammatory molecules, resulting in the activation of microglia to further increase the release of inflammatory cytokines (35,36). Microglia are capable of acquiring various levels of activation, which reflects the features responsible for their contribution to vital components of recovery after GMH, such as neuroinflammation, tissue repair, or immunomodulation. Classically activated (M1, CD68+) microglia reportedly release destructive pro-inflammatory mediators, whereas the alternatively activated (M2, CD206+) microglia participate in tissue debris clearance via phagocytosis, as well as participate in the release of several protective and trophic factors. M1 microglia polarization mainly act in the acute phase after GMH, resulting in substantial pro-inflammatory cytokine release (26). Consistently, microglia primarily polarized to the M1 phenotype with few M2 phenotype microglia at 72 h after GMH in our study.
Microglia/macrophages are characterized by remarkable plasticity and versatility, such as being able to switch from one phenotype to another. After including ICH in mice using collagenase, the switch from the M1 to the M2 microglial phenotype was observed on days 1 to 3 post-ICH (37). Stimulation of cannabinoid receptor 2 (CB2R) suppressed neuroinflammation via regulation of M1/M2 polarization via cAMP/PKA signaling pathway in an experimental rat model of GMH (26). Recently, we reported that GW9508-mediated GPR40 activation ameliorates neuroinflammation and improves neurological function through PAK4/CREB/KDM6B signaling pathway after GMH (20). Our data from this study showed that intranasal administration of rh-APN promoted M2 microglia polarization with increased anti-inflammatory cytokine (IL-10 and Arginase-1) release.
Moreover, we investigated adiponectin's effects on sham animals, which revealed that adiponectin did not influence the polarization of microglia in the absence of GMH. Combined with IHC staining, it was shown that rh-APN did not affect activation of microglia in sham animals. It was demonstrated that rh-APN drives microglial M2 polarization on the basis of an activated state. Furthermore, it was revealed that knockdown of AdipoR1 reversed the effect of rh-APN and increased expression of AdipoR1 on alternatively activating (M2, CD206-positive) microglia after rh-APN treatment. In conjunction, it is very likely that the necessary transition from the M1-activated state to the M2 phenotype using rh-APN relies on upregulation of AdipoR1 expression.
Proliferation and accumulation of microglia have detrimental roles pertaining to GMH pathology in the immature preterm brain (38). Tang et al. reported that CB2R activation inhibited thrombin-induced microglial proliferation, as well as the inflammatory response. We found increased microglia proliferation following GMH and rh-APN treatment had no effect on the proliferation of GMH. However, the increased accumulation of activated microglia or macrophages (CD11b/c +) in the periventricular region were observed post-GMH with rh-APN treatment. Meanwhile, administration of rh-APN significantly increased M2 (CD206 + ) microglia, which indicates that the activated and accumulated microglia in the periventricular region might be predominantly M2 (CD206 +) microglia.
Peroxisome proliferator-activated receptor-g (PPARg) is a ligand-activated transcription factor that belongs to the superfamily of nuclear hormone receptors. Activation of PPARg mediated the conversion of microglia phenotype and phagocytic capabilities of peripheral M2 polarized macrophages in various CNS diseases (39,40), including hemorrhagic stroke (21). Flores et al. revealed that the PPARg-induced CD36 upregulation in microglia increased M2 microglia polarization and was crucial in enhancing post-GMH hematoma resolution.
PPARg agonists enhanced myelination, reduced inflammation and hydrocephalus, and promoted neurological recovery in newborns with intraventricular hemorrhage (IVH) (41). More recently, studies have shown that activation of adiponectin receptor with adipoRon could boost PPARg expression and inhibit pro-inflammatory microglia responses, therefore ameliorating hyperperfused cognitive deficits (40). Accordingly, we demonstrate that APN promote phenotypic change of microglia from "classically" M1 activated to "alternatively activated" M2 states and enhances hematoma clearance, thereby reducing hydrocephalus by activation of PPARg via AdipoR1 signaling. Emerging evidence has shown that the activation of AMPK promoted microglial M2 polarization, thereby inhibiting neuroinflammation after stroke (42)(43)(44). Moreover, APN ameliorated intracerebral hemorrhageinduced neuroinflammation through the AdipoR1-AMPK pathway (45). Since APPL1 and LKB1 were involved in APN/ AdipoR1-mediated AMPK activation (15), AMPK might participate in APN-induced upregulation of PPARg. Intraventricular administration of specific siRNA-targeted APPL1 and LKB1 significantly reversed AMPK activation and increased PPARg expression induced by rh-APN/AdipoR1. In addition, CD36 and IL-10 expression were decreased, while IL-1b and IL-6 expression were increased after knockdown of APPL1 and LKB1 was followed with APN treatment, suggesting that APN alleviated GMH-induced inflammation by upregulation of PPARg via AdipoR1/APPL1/LKB1/ AMPK signaling.
Because AdipoR1 expression can be found on various brain cells and the role of microglial AdipoR1/AMPK/PPARg signaling in neuroinflammation in vivo remains unexplored, Lipo-Dorsomorphin (AMPK inhibitor) and Lipo-GW9662 (PPARg antagonist) was used to selectively inhibit AMPK and PPARg activation in microglia, as previously described (19). With the inhibition of AMPK activation and decreased PPARg expression, the CD36 and IL-10 expressions were completely reversed with rh-APN treatment. By contrast, expression of IL-1b and IL-6 were significantly increased, suggesting that rh-APN attenuated neuroinflammation may be primarily mediated through the AMPK/PPARg pathway in microglia. However, our present study mainly focused on the effects of APN, which were mediated via microglial signaling after GMH. The APN effects on neurons or astrocytes will be investigated in future studies.
In the present study, GMH was induced via stereotaxic injection of sterile collagenase into the ganglionic eminence of newborn rats (P7). However, there are several potential limitations of this model. First, the rapid progression of developmental events following the P7 developmental stage in the rat brain differs from the neonatal human brain, which has been reviewed by others (46). Although researchers have considered the P7 rat to be equivalent to the human term neonate for decades (47), there is evidence that P7 may actually represent an earlier gestation (48,49). Second, the GMH collagenase model may not only produce hemorrhage, but may also induce a significant inflammatory reaction (46). The increased inflammation may account for a proportion of the enlarged ventricular dilation when compared to that of the GHM model using direct blood injection. Third, the hemorrhage in the brain was not proven to initiate from the germinal matrix. Thus, further investigations are needed to demonstrate the process of hematoma formation following infusion of collagenase into the periventricular ganglionic eminence. Fourth, all pups suffered from the procedure without mortality across the study. This is different from human findings following high-grade GMH, which have a higher rate of mortality (30%-50%) in clinical settings (50).However, the neurological consequences following GMH induced by collagenase successfully mimic the ventricular dilation and the motor deficits seen in human preterm infants.

CONCLUSIONS
To conclude, this study elucidates the neuroprotective effect of APN in reducing proinflammatory cytokine release and in enhancing hematoma resolution via promotion of M2 microglial polarization and accumulation of M2 microglia in the periventricular area, thereby alleviating neurological deficits in an experimental rat model of GMH (Figure 9). The neuroprotective effects of APN were associated with the AdipoR1/APPL1/AMPK/PPARg signaling pathway.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The animal study was reviewed and approved by Loma Linda University Institutional Animal Care and Use Committee.  compared to sham, # P < 0.05 compared to GMH + Vehicle, @ P < 0.05 compared to GMH + rh-APN or GMH + rh-APN + Scramble siRNA.