Aquaporin‐4 deficiency reduces TGF‐β1 in mouse midbrains and exacerbates pathology in experimental Parkinson's disease

Abstract Aquaporin‐4 (AQP4), the main water‐selective membrane transport protein in the brain, is localized to the astrocyte plasma membrane. Following the establishment of a 1‐methyl‐4‐phenyl‐1,2,3,6‐tetrahydropyridine (MPTP)‐induced Parkinson's disease (PD) model, AQP4‐deficient (AQP4−/−) mice displayed significantly stronger microglial inflammatory responses and remarkably greater losses of tyrosine hydroxylase (TH+)‐positive neurons than did wild‐type AQP4 (AQP4+/+) controls. Microglia are the most important immune cells that mediate immune inflammation in PD. However, recently, few studies have reported why AQP4 deficiency results in more severe hypermicrogliosis and neuronal damage after MPTP treatment. In this study, transforming growth factor‐β1 (TGF‐β1), a key suppressive cytokine in PD onset and development, failed to increase in the midbrain and peripheral blood of AQP4−/− mice after MPTP treatment. Furthermore, the lower level of TGF‐β1 in AQP4−/− mice partially resulted from impairment of its generation by astrocytes; reduced TGF‐β1 may partially contribute to the uncontrolled microglial inflammatory responses and subsequent severe loss of TH+ neurons in AQP4−/− mice after MPTP treatment. Our study provides not only a better understanding of both aetiological and pathogenical factors implicated in the neurodegenerative mechanism of PD but also a possible approach to developing new treatments for PD via intervention in AQP4‐mediated immune regulation.

mediators in neuroinflammatory responses. 11,12 Activated microglia are commonly seen within the substantia nigra pars compacta (SNpc) of PD brains investigated at autopsy; these cells directly induce significant, highly detrimental neurotoxic effects by excessive production of a large array of cytotoxic factors such as interleukin-1β (IL-1β), tumour necrosis factor-α (TNF-α), interleukin-6 (IL-6) and nitric oxide (NO). [13][14][15] Moreover, as an antigenpresenting cell (APC), activated microglia express costimulatory molecules, such as cluster of differentiation 40 (CD40), cluster of differentiation 80 (CD80; B7-1) and cluster of differentiation 86 (CD86; B7-2), that promote APC-dependent T-cell activation. [16][17][18] Subsequently, activated T cells injure neurons by cell contact-dependent mechanisms that involve Fas ligand (FasL) and/or release of cytotoxic factors. 19 Attenuation of microglial activation can protect up to 90% of DNs in PD animal models. [20][21][22] Aquaporin-4 (AQP4), originally known as a mercurial-insensitive water channel, is most strongly expressed in astrocytes throughout the brain and spinal cord, as well as in ependymal cells lining the brain ventricles; it is involved in the regulation of water homeostasis in the brain. [23][24][25] Recently, AQP4 expression has been reported to be involved in the pathology of the development of PD in a 1-methyl-4phenyl-1,2,3,6-tetrahydropyridine (MPTP)-induced mouse model. In our previous studies, compared with AQP4 +/+ mice, AQP4 −/− mice were significantly more prone to MPTP-induced neurotoxicity and subsequently exhibited significantly stronger microglial responses in the midbrain and more severe PD symptoms. [26][27][28] However, the mechanisms underlying hyperactive microglial responses and more severe clinical symptoms in PD after administration of MPTP in AQP4-deficient mice remain unclear.
In this study, significantly decreased transforming growth factor-β1 (TGF-β1) led to increased neuroinflammatory responses in the midbrain and more severe PD pathology in AQP4-deficient mice after MPTP intoxication. Our findings suggest a novel role for AQP4 in brain neurodegeneration and an opportunity for the development of new therapeutic approaches to treat neurodegenerative diseases.

| Transgenic mice
AQP4-deficient mice were generated as previously described. 27 AQP4 −/− mice were maintained on the CD1 background. AQP4 +/+ CD1 mice were used as wild-type (WT) control animals. Mice were identified by polymerase chain reaction (PCR) analysis of tail samples at post-natal day 5 and by Western blot analysis of the cerebral cortex. Mice were bred and maintained under environmentally controlled conditions (ambient temperature, 22°C; humidity, 40%) on a 12-hour light/dark cycle with access to food and water. All experiments were performed on age-and weight-matched littermates (20-28 g). Mouse breeding was performed to achieve timed pregnancy with an accuracy of ±0.5 days.
All experiments were approved by IACUC (Institutional Animal Care and Use Committee of Nanjing Medical University). All efforts were made to minimize animal suffering and to reduce the number of animals used for the experiments.

| Acute MPTP treatment
Sixteen-week-old male AQP4 +/+ and AQP4 −/− mice were injected intraperitoneally (i.p.) four times with MPTP-HCl (20 mg/kg of free base; Sigma Chemicals, St. Louis, MO) in saline or saline alone for controls at 2-hour intervals. The total dose per mouse was 80 mg/ kg, and the mice were killed 7 days after the last injection. Mortality rates in acute MPTP-treated AQP4 −/− mice (41%, 35/85) were twofold higher than in AQP4 +/+ mice (21%, 17/80, P < 0.05). As the mice of MPTP group underwent an obvious mortality after MPTP treatment, only mice that went through all 7 days procedure were included in the following statistical analyses. MPTP handling and safety measures were in accordance with published guidelines. 29

| Chronic MPTP/probenecid treatment
Sixteen-week-old male AQP4 +/+ and AQP4 −/− mice received a total of 10 doses of MPTP-HCl (20 mg/kg in saline, subcutaneously [sc]) in combination with an adjuvant, probenecid (250 mg/kg in dimethyl sulfoxide [DMSO], ip). Mice were treated in a similar manner with probenecid and saline as controls. The 10 doses were administered on a 5-week schedule, such that injections were given with an interval of 3.5 days between consecutive doses. Animals were killed 7 days after the last injection. AQP4 −/− and AQP4 +/+ mice after chronic MPTP treatment showed no mortality. Probenecid was purchased from Sigma Chemical Co. Probenecid was used to inhibit the rapid clearance and excretion of MPTP from the brain and kidney following each injection. Neither probenecid nor DMSO at the concentrations used in this study produced any significant effect on striatal dopamine (DA) contents. 30

| Mesencephalic primary neuron culture and treatment
Primary mesencephalic neuronal cultures were prepared from the ventral mesencephalon of gestational 16-to 18-day-old AQP4 +/+ and AQP4 −/− mouse embryos. Mesencephalic tissues were dissected and maintained in ice-cold Ca2 + -free Hanks' Balanced Salt Solution (HBSS) (Gibco, Grand Island, NY) and then dissociated in HBSS containing trypsin-0.25% ethylenediaminetetraacetic acid (EDTA) for 20 minutes at 37°C. Dissociated cells were then plated at equal density (2 × 10 6 cells) in a 25-cm 2 flask precoated with 1 mg/mL poly-D-lysine (Sigma). Cultures were maintained in a chemically defined medium consisting of neurobasal medium fortified with B-27 supplements, 500 μg/mL of glutamine, 100 IU/mL penicillin and 100 μg/mL streptomycin (Invitrogen, Carlsbad, CA). The cells were maintained in a humidified CO 2 incubator (5% CO 2 and 37°C) for 24 hours. Half of the culture medium was replaced every 2 days. Seven-day-old cultures were used for the experiments. Primary mesencephalic DNs were exposed to 10 μmol/L MPP + (Sigma) for 24 hours.
Primary cultures of mixed glia from day 1-2 newborn mice were prepared. Briefly, following the removal of meninges, brain tissues were minced and incubated in a rocking water bath at 37°C for 30 minutes in HBSS (Gibco) in the presence of 0.25% trypsin (Sigma). Enzyme-digested dissociated cells were triturated with 0.25% of foetal bovine serum (FBS, Gibco), followed by a wash and centrifugation (300× g for 10 minutes). The pellet was resuspended in HBBS and passed through 100-μm nylon mesh, followed by a second wash and centrifugation (300× g for 10 minutes). Following dilutions with astrocyte-specific medium (Dulbecco's Essential Medium containing 1% penicillin-streptomycin, 10% FBS), the cells were plated and allowed to adhere for 1 day in a humidified CO 2 incubator at 37°C. After 24 hour, any non-adherent cells were removed, and fresh astrocyte-specific medium was added. Adherent cells were maintained in astrocyte-specific medium for 10 days with a medium change every 3-4 days. The microglia population peaked at 12-14 days in these cultures. Microglia-enriched cultures were thoroughly agitated in an orbital incubator shaker (250 rpm for 2 hours at 37°C) to remove any cells adherent to the astrocyte monolayer.
Immediately following agitation, all cells suspended in the culture medium were collected and centrifuged at 300× g for 5 minutes at 4°C. The cell pellet contained microglia that were resuspended and diluted with fresh astrocyte-specific medium, bringing the cells to a final concentration of 8 × 10 5 cells/mL until assayed. The original flasks in which the microglia had been shaken were maintained with astrocyte-specific medium for subsequent experiments. Primary astrocytes were seeded at 1 × 10 6 cells per well in 6-well plates and incubated with phosphate buffered saline (PBS) or MPP + (50 μmol/L) for 48 hours in 0.1% serum-supplemented medium. The culture medium was collected and centrifuged at 300 g for 5 minutes, then the volume of each supernatant was adjusted to the same volume (to standardized preparations) and immediately stored at −80°C until used for TGF-β1 assay by ELISA using commercial kits.

| BV-2 cell culture
The immortalized microglial cell line BV-2, derived from raf/myc-immortalized murine neonatal microglia, was kindly provided by Prof.
BV-2 cells in medium without TGF-β1/anti-TGF-β1 served as controls. After injection, the mice were kept in cages with a constant temperature (25°C) and humidity. They were exposed to a 12:12-hour light-dark cycle and had unrestricted access to tap water and food. Mice were killed for further study at 6 days after the MPTP injection. and CellQuest software (BD Biosciences); data were collected on 10 000 cells per condition.

| Immunohistochemistry staining
At the end of the experiments, the mice were perfused with 4% paraformaldehyde (PFA, Sigma). Brain samples were collected and post-fixed in 4% PFA at 4°C overnight. They were transferred to 15% sucrose in PBS overnight and then to 30% sucrose overnight until the brain sunk to the bottom of the tube.

| Statistical analysis
Data analyses and graphs were performed using spss software for  Figure 1A and B, compared with incubation with PBS treatment controls, incubation with brain homogenates from either saline-or MPTP-injected AQP4 +/+ or AQP4 −/− mice significantly induced increases in the expression of costimulatory molecules major histocompatibility complex II (MHCII), CD80, and CD40 in BV-2. Compared with brain homogenates from salinetreated AQP4 +/+ mice (or AQP4 −/− mice), brain homogenates from MPTP-treated AQP4 +/+ mice (or AQP4 −/− mice) induced a much higher increase in the costimulatory molecules MHCII, CD80, and CD40. Similarly, compared with PBS treatment controls, treatment with brain homogenates from either saline-or MPTP-injected AQP4 +/+ or AQP4 −/− mice increased both mRNA ( Figure 1C) and protein ( Figure 1D) expression levels of pro-inflammatory TNF-α, IL-1β, and IL-6 cytokines in BV-2. In addition, compared with brain homogenates from MPTP-treated AQP4 +/+ mice, brain homogenates from MPTP-treated AQP4 −/− mice induced higher increases in TNF-α but not in TGF-β1 cytokines in BV-2 cells. These results suggest that some factors in AQP4 −/− mouse brains facilitate the higher expression of pro-inflammatory cytokines but lower levels of TGF-β1 after MPTP treatment.
These results suggested that after MPTP treatment, AQP4 deficiency results in enhanced astrocyte activation but no detectable difference in α-syn protein levels in astrocytes between AQP4 +/+ and AQP4 −/− mice.
These results suggest that α-syn was mainly derived from neurons. However, in addition to neurons, there are many other sources of α-syn.

| AQP4 −/− mouse midbrains expressed higher levels of α-syn with or without acute or chronic MPTP treatment
To further investigate the possible role of α-syn in the more severe

| MPTP treatment failed to increase TGF-β1 production in AQP4 −/− mice
Transforming growth factor-β1 plays a critical role in the downregulation of microglial responses to minimize brain inflammation and efficiently restricts the exacerbation of brain damage in both human PD and MPTP-induced mouse models of PD. 43-48 AQP4 +/+ and AQP4 −/− mice showed similar levels of serum TGF-β1 without MPTP treatment. 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine treatment resulted in an increase in serum TGF-β1 in AQP4 +/+ mice only ( Figure 5A). Consistently, after MPTP treatment, a more significant increase in the level TGF-β1 was shown in the midbrain of AQP4 +/+ but not AQP4 −/− mice ( Figure 5B,C).
Midbrain TGF-β1 is mainly expressed by microglia and astrocytes. [49][50][51] Since AQP4 was expressed in astrocytes but not in microglia, 28,35-37 we further investigated the expression of TGF-β1 in astrocytes. The results ( Figure 5D) showed that compared with PBS treatment, MPP + treatment resulted in increased TGF-β1 levels in AQP4 +/+ but not in AQP4 −/− astrocytes. These results indicated that AQP4 deficiency in mouse astrocytes resulted in the failure to increase TGF-β1 production in response to MPP + treatment; this issue F I G U R E 4 Expression of α-synuclein (α-syn) in the midbrains of AQP4 +/+ and AQP4 −/− mice after acute or chronic 1-methyl-4-phenyl-1,2,3,6tetrahydropyridine (MPTP) intoxication. Quantitative real-time PCR (Qrt-PCR) analysis of α-syn mRNA expression in AQP4 +/+ and AQP4 −/− mice after acute (A) or chronic (B) MPTP administration. The mRNA expression was measured individually and normalised to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). Western blot analysis of α-syn expression in both AQP4 +/+ and AQP4 −/− mouse midbrains after either acute (C,D) or chronic (E,F) MPTP treatment. β-Actin served as a loading control. The histograms represent the normalized levels of α-syn expressed as a ratio. Data represent the mean ± SEM from five mice per group and are representative of three independent experiments. *P < 0.05 compared with saline-treated AQP4 −/− mice may at least partially contribute to the more severe microgliosis and neuronal damage.

| Injection of TGF-β1 in the striatum significantly reduced neuronal damage and microglial activation in MPTP-treated AQP4 −/− mice
To investigate whether lower levels of TGF-β1 may be responsible for the more severe inflammation and pathology in AQP4 −/− mouse brains, we increased TGF-β1 in AQP4 −/− mice by stereotactic injection 24 hours after the last MPTP injection. The results in Figure 6A and B show that there were no significant differences in the stereological counts of TH + SNpc DNs between saline-injected AQP4 −/− mice and their AQP4 +/+ littermates without MPTP treatment. After treatment with MPTP, there were remarkably greater losses of TH + neurons in the SNpc in AQP4 −/− mice than in AQP4 +/+ mice. However, TGF-β1 stereotactic injection efficiently decreased the loss of TH + DNs in both AQP4 +/+ and AQP4 −/− mice. More importantly, TGF-β1 rescued many more TH + DNs in AQP4 −/− mice.
Since MPTP induces a robust microglial response, 52 we characterized microglial activation in mice after MPTP treatment.
These data suggest that TGF-β1 more effectively attenuates MPTP-induced microglial activation and dopaminergic neuronal death.

| In vitro increases or decreases in TGF-β1 significantly regulated mouse brain homogenatestimulated BV-2 activation
To further confirm that the lower level of TGF-β1 in AQP4 −/− mouse midbrains contributed to more severe hyperactive microglial cell responses, we added TGF-β1 to the AQP4 −/− mouse midbrain homogenate or used anti-TGF-β1 to neutralise TGF-β1 in the AQP4 +/+ mouse midbrain homogenate. The addition of TGF-β1 to the brain homogenate from MPTP-treated AQP4 −/− mice resulted in a significant decrease in costimulatory molecules MHCII, CD80, and CD40 ( Figure 7A,B), as well as pro-inflammatory cytokines IL-1β, TNF-α and IL-6 in BV-2. In contrast, adding TGF-β1 neutralizing antibodies to F I G U R E 5 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-treatment failed to increase transforming growth factor-β1 (TGF-β1) production in AQP4 −/− mice or in AQP4 −/− astrocytes. A, TGF-β1 in the peripheral serum of AQP4 +/+ and AQP4 −/− mice after acute MPTP intoxication was measured with ELISA. Data were from five mice per group and are representative of three independent experiments. B, Western blot detection of TGF-β1 in the midbrains of AQP4 +/+ and AQP4 −/− mice after acute MPTP intoxication. One representative experiment of three is shown. C, Values are presented as the mean ± SEM for five mice per group from three independent experiments and were normalized to the protein concentrations of cell extracts. D, TGF-β1 detection by ELISA in the supernatants collected from midbrain astrocytes incubated for 48 h with 50 μmol/L of MPP + . **P < 0.01 compared with MPTP-treated AQP4 −/− mice; *P < 0.05 compared with saline-treated AQP4 +/+ mice; #P < 0.05 compared with MPTP-treated AQP4 +/+ mice; &P < 0.01 compared with phosphate buffered saline (PBS)-treated AQP4 +/+ astrocytes; +P < 0.01 compared with MPP + -treated AQP4 +/+ astrocytes the brain homogenates from MPTP-treated AQP4 +/+ mice resulted in an increasing trend in pro-inflammatory cytokines IL-1β, TNF-α, and IL-6, but there were no significant differences ( Figure 7C). These data further suggested that the lower TGF-β1 level might be one of reasons in AQP4 −/− mice contributed to stronger microglial activation, which might subsequently result in more dopaminergic neuronal death and more severe PD disease after MPTP intoxication, and certain other factor(s) in brain homogenates might also be responsible for the more microgliosis and neuronal damage after MPTP treatment.

| D ISCUSS I ON
Aquaporin-4 is a predominant water channel protein in mammalian brains that is mainly localized to the astrocyte plasma membrane. 25 In our previous study, AQP4 −/− mice showed significantly stronger microglial responses and exhibited significantly more severe neuronal pathology after administration of MPTP. 28 However, the mechanisms remain unclear. In this study, for the first time, we revealed that the significantly reduced TGF-β1 in Post-mortem studies and animal experiments linked microglia-mediated neuroinflammation to losses of TH + neurons and the pathogenesis of PD. Overactivation of microglial cells may cause severe brain tissue damage in various neurodegenerative diseases. 53,54 However, many studies have indicated that AQP4 is not expressed on Mac-1 + microglia. It might be helpful to potentially exclude the possibility of AQP4 deficiency directly leading to differential microglial responses and neuronal damage between AQP4 +/+ and AQP4 −/− mice after MPTP treatment.
In addition, the level of endogenous α-syn was much higher in AQP4 −/− mice than in AQP4 +/+ mice without MPTP treatment.
Thus, these results indicate that high levels of α-syn alone may be insufficient to induce neuronal pathology. Instead, previous studies showed that α-syn may play an adjunctive role in PD by enhancing microglial activation-mediated neuronal pathology, which is triggered by certain factors that are necessary for the onset of PD (eg, MPTP). Previous studies 57 and our studies suggested the high level of α-syn may make AQP4 −/− mice more susceptible to MPTP, but the mechanism still needs to be further improved. A decreased number of CD4 + CD25 + regulatory T (Treg) cells promote the increased severity of PD in AQP4-deficient mice. 28 TGF-β1 signalling is required for the generation of the peripheral Treg cell subset by inducing the expression of the transcription factor fork head box P3 (Foxp3). 58 In addition, TGF-β1 plays a critical role in the down-regulation of microglial responses by suppressing the activation, proliferation and production of IL-1, IL-6, and TNF-α, thereby minimizing brain inflammation. 43 Moreover, TGF-β1 elicits the neurotrophic activity of glial cell-derived neurotrophic factor (GDNF) and contributes to the survival of midbrain DNs to protect against the toxic effects of MPP + . 59 In this study, for the first time, we showed that AQP4 deficiency in mice resulted in the failure to increase TGF-β1 production in the midbrain and peripheral blood after MPTP treatment; these changes might account for the hyperactive microglial neuroinflammatory responses and enhanced loss of DNs by MPTP-induced neurotoxicity. Our observation appears to support the notion that reduced TGF-β1 signalling in the striatum contributes to the loss of DNs in the substantia nigra. 48 The primary cell types expressing TGF-β1 mRNA in the adult brain are microglia and astrocytes. 49,51 According to previous research, astrocytes but not microglia express AQP4. 25,28,[35][36][37] In this study, we further demonstrated that MPP + treatment failed to increase TGF-β1 production in AQP4 −/− astrocytes. However, since TGF-β1 can be produced by cells other than astrocytes in the brain, further studies are needed to explore how AQP4 controls astrocytes and/or other cells to generate TGF-β1.

| CON CLUS IONS
Our study illustrated that the TGF-β1 production in astrocytes was impaired in AQP4 −/− mice; this alteration may contribute to the hyperactive microglial neuroinflammatory responses and subsequent enhancement of the loss of DNs by MPTP-induced neurotoxicity.

ACK N OWLED G EM ENTS
This work was supported by a grant from the National Natural Science Foundation of China to Chuan Su (No. 81430052). We thank Jianhua Ding and Yan Zhou (Nanjing Medical University Dept Pharmacology) for mouse housing, breeding and scientific support.
We thank Ming Lu (Nanjing Medical University Dept Pharmacology) and Yong Li (Nanjing Medical University Dept Pathogen Biology and Immunology) for technical support in histology.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.