Phosphorylation of AQP4 by LRRK2 R1441G impairs glymphatic clearance of IFNγ and aggravates dopaminergic neurodegeneration

LRRK2 R1441G impairs glymphatic system clearance and decreases AQP4 polarity

The glymphatic system plays a crucial role in clearing abnormally accumulated proteins in the brain²⁵. To explore the clearance function of the glymphatic system in LRRK2 wild type (WT) and LRRK2 R1441G (R1441G) transgenic mice, fluorescein isothiocyanate [FITC]-dextran and rhodamine B were used as tracers of cerebrospinal fluid (CSF) and blood vessels (Fig. 1a). Two-photon in vivo imaging was employed to measure the clearance of FITC-dextran through the glymphatic system.In WT mice, FITC tracer appeared in the paravascular space 10 min after the injection. The fluorescence intensity increased by 15 min, peaked at 30 min, began decreasing at 45 min, and decreased further at 60 min. In contrast, while FITC-dextran was also detected in the paravascular space of R1441G mice after 10 min, the fluorescence intensity continuously increased at 15, 30, 45, and 60 min (Fig. 1b, c). These findings suggest that LRRK2 R1441G mutation impairs the clearance function of the glymphatic system.

AQP4 polarity plays a crucial role in facilitating the influx of subarachnoid CSF and the efflux of interstitial fluid (ISF) solutes to the brain parenchyma²⁶, which maintains the normal function of the glymphatic system¹¹. Under physiological conditions, AQP4 expression is predominantly polarized to the perivascular astrocytic endfeet. However, in certain disease conditions, the expression of AQP4 shifts from the endfeet to the soma²⁷. AQP4 immunofluorescence was localized in the astrocytic endfeet in WT mice. Conversely, in the R1441G group, AQP4 was depolarized and observed in the astrocytic soma (Fig. 1d). As previously elucidated, AQP4 polarity was quantified as the ratio of the low-stringency area to the high-stringency area²⁸. The low-stringency threshold encompasses the overall area of AQP4 immunoreactivity, while the high-stringency threshold corresponds to the area of AQP4 immunoreactivity in perivascular end feet (Fig. 1e). Calculation of AQP4 polarity further revealed significantly lower values in the R1441G group compared to the WT group (Fig. 1f). Taken together, R1441G mutation decreased AQP4 polarity and impaired glymphatic system clearance.

LRRK2 interacts with AQP4, and mutant LRRK2 regulates AQP4 polarity through phosphorylation

To investigate how the R1441G mutant affects the polarity of AQP4, we conducted the co-immunoprecipitation (co-IP) experiment to examine whether LRRK2 interacts with AQP4. LRRK2 successfully pulled down AQP4 using brain lysates from both WT and R1441G mice. However, no substantial difference was seen in the amount of AQP4 pulled down by LRRK2, indicating that the R1441G mutation may not affect the ability of LRRK2 to bind to AQP4 (Fig. 2a, b). To explore which domain of LRRK2 bind with AQP4, we co-transfected the different domains of LRRK2 with AQP4. AQP4 could pull down ARM, COR, Kinase and WD40 domain, indicating that AQP4 interacted LRRK2 through several different domains (Fig. 2c). Furthermore, western blot studies showed that there was no difference of total AQP4 protein level between the WT and R1441G mouse brains with or without GTPase inhibitor or kinase inhibitor treatment (Fig. 2d, e, Supplementary Fig. 1a, b).

The polarity of AQP4 depends on its overall protein expression and recruitment from intracellular stores to the plasma membrane, known as membrane compartmentalization²⁹, the latter of which is significantly influenced by AQP4 phosphorylation³⁰. Since the phosphorylation sites of AQP4 are mainly serine (Ser)^30,31, an anti-LRRK2 antibody was used to pull down AQP4 in the brain tissues from WT and R1441G transgenic mice, and then Ser phosphorylation of AQP4 (p-AQP4) was detected. We observed considerably higher levels of p-AQP4 in R1441G mice than in WT mice. Importantly, this effect was partially reversed by LRRK2 inhibitors, indicating that R1441G mutation increases AQP4 phosphorylation (Fig. 2f, g, Supplementary Fig. 1c, d).

To investigate whether the impact of mutant LRRK2 on AQP4 polarity is specific to the R1441G mutation or attributable to increased LRRK2 phosphorylation, we co-transfected LRRK2 G2019S with AQP4. The co-IP results revealed that the G2019S mutation led to an elevated expression of p-AQP4 whereas Mli-2 treatment mitigated G2019S-induced p-AQP4 (Fig. 2h, i). In summary, these findings indicate that the LRRK2 G2019S mutation could also phosphorylate AQP4, underscoring the broader impact of LRRK2 mutations on AQP4 dynamics.

The depolarizing effect of the R1441G mutation on AQP4 was further examined in an in vitro transwell system. Primary astrocytes isolated from WT or R1441G mice were seeded in the bottom wells. The endfeet of the astrocytes could pass through the filter pores to the upper wells, establishing close connections with HeLa cells (Fig. 3a). Glial Fibrillary Acidic Protein (GFAP), the most commonly used astrocytic marker was used to co-stain with AQP4 to evaluate AQP4 polarity. As depicted in Fig. 3b, AQP4 predominantly expressed on the surface membrane of astrocytic endfeet in the WT group. Conversely, in the R1441G group, AQP4 was internalized into the cytoplasm. Treatment with either GTPase inhibitor(Fx2149) or kinase inhibitor (Mli-2) successfully restored the altered distribution of AQP4 in R1441G astrocytes. This observation suggests that LRRK2 R1441G modifies AQP4 polarity through phosphorylation, and the restoration by LRRK2 inhibitors indicates potential therapeutic intervention for this modification.

To examine whether LRRK2 inhibitors can also rescue the AQP4 polarity in vivo, we administrated both GTPase inhibitor and kinase inhibitor to the WT and R1441G mice. LRRK2 inhibitors can significantly decrease AQP4 depolarization induce by R1441G mutation, indicating by AQP4 immunoactivity in endfeet of astrocyte, but not soma (Fig. 3c, d). Furthermore, Western blot study showed that LRRK2 inhibitors also alleviate the decreased expression of AQP4 in membrane caused by R1441G mutation (Fig. 3e, f, Supplementary Fig. 1e, f). The similar results were also observed with G2019S mutation (Supplementary Fig. 1g, h).These results indicated that LRRK2 inhibitors can improve AQP4 polarity. Collectively, these findings provide evidence that mutant LRRK2 affects AQP4 polarity through phosphorylation, which may impair the clearance function of the glymphatic system.

IFNγ is involved in LPS-induced neuroinflammation in mutant LRRK2 mice

LPS reportedly induces neuroinflammation and DA degeneration in mutant LRRK2 mice³². Immunofluorescence staining results showed that there was significant loss of tyrosine hydroxylase (TH) neurons in R1441G + LPS group compared with WT, WT + LPS and R1441G groups (Fig. 4a). Furthermore, the protein levels of TH in midbrain also decreased in the R1441G + LPS group (Fig. 4b, c). Similarly,the R1441G group showed no microgliosis and astrogliosis compared with the WT group. However, the reactivity of both microglia and astrocytes was more remarkable in the R1441G + LPS group than in the WT + LPS group (Fig. 4f–h). Furthermore, treatment with LRRK2 inhibitor rescued the loss of TH neurons and mitigated gliosis in the R1441G + LPS group, while there were no difference between WT + LPS group with or without inhibitor treatment. (Fig. 4a–h).

IFNγ has also been implicated in DA neurodegeneration³³. Therefore, IFNγ levels were examined. We found that the protein levels of IFNγ were markedly elevated in both WT and R1441G mice after LPS treatment. Notably, the R1441G + LPS group exhibited much higher IFNγ levels compared to the WT + LPS group (Fig. 4i, j). The administration of LRRK2 inhibitor successfully reduced the elevated levels of IFNγ (Fig. 4k, l).

Collectively, these findings suggest that the interaction between LRRK2 R1441G and LPS synergistically triggers the activation of inflammatory cells and the loss of DA neurons, with IFNγ playing a role in LPS-induced neuroinflammation.

LRRK2 R1441G impairs the clearance of IFNγ through the glymphatic system

To investigate whether IFNγ can be cleared from the brain via the AQP4-dependent glymphatic system, FITC-IFNγ was injected into the striatum of FVB mice (background strain for both LRRK2 WT and R1441G transgenic mice) treated with or without an AQP4 inhibitor (Fig. 5a). The fluorescence intensity in the brain parenchyma was measured to assess the efflux of FITC-IFNγ 2 h after the intraparenchymal injection. A notably higher dye residual in the parenchyma was observed in the FVB mice treated with the AQP4 inhibitor than in those not treated with the inhibitor (Fig. 5b, c). Previous studies have indicated that macromolecules present in the parenchyma or ISF primarily drain into deep cervical lymph nodes (dCLNs) following efflux³⁴. Remarkably, compared with control animals, animals treated with an AQP4 inhibitor displayed significantly reduced fluorescence intensity in the dCLNs (Fig. 5d, e). These findings indicate that IFNγ is cleared from the brain tissue via the AQP4-dependent glymphatic system.

Next, the efflux of IFNγ from the brain parenchyma was examined to determine whether it is affected by the R1441G mutation. Two hours after injection, FITC-IFNγ was found to be significantly higher in the parenchyma of R1441G mice than in WT mice (Fig. 5f, g). Similarly, the fluorescence intensity in the dCLNs exhibited a significant reduction in the R1441G group (Fig. 5h, i). These effects were reversed by treatment with a LRRK2 inhibitor (Fig. 5f–i). There was no difference in fluorescence intensity between WT mice treated with or without inhibitors. These findings suggest that the LRRK2 R1441G mutation impairs the clearance of IFNγ through the glymphatic system.

IFNγ upregultes the expression of IFNγ receptor-2 and LRRK2 in LRRK2 R1441G mice, further aggravating the loss of AQP4 polarity

Binding of IFNγ to its receptors (IFNGR) induces the activation of receptor-associated downstream signaling factors³⁵. IFNGR-1 and -2 are two subunits of functional IFNGR³⁶. While IFNGR-1 is essential for ligand binding, it cannot induce a biological response independently. IFNGR-2 is associated with IFNGR-1 in an IFNγ-dependent manner and plays a vital role in the transmission of functional responses³⁷. Therefore, we focused on the transduction of IFNGR-2 to determine the impact of IFNγ on LRRK2 and AQP4 in the brain.

To that end, IFNγ was injected into the striatum of WT and LRRK2 R1441G mice, and the expression of IFNGR-2 was compared 1 week later. Markedly increased expressions of both IFNGR-2 and LRRK2 were observed in the R1441G mice but not in the WT mice after IFNγ treatment (Fig, 6a–c). Notably, treatment with a LRRK2 inhibitor partially downregulated the expression of both IFNGR-2 and LRRK2 in the R1441G + IFNγ group (Fig. 6d–f). Furthermore, both reactive astrocyte and microglia were evident. However, IFNGR-2 mainly co-localized with GFAP but not with Iba-1 in the brain of R1441G mice, indicating that IFNGR-2 was mainly expressed in astrocytes and LRRK2 inhibitor could partially reverse the effects (Fig. 6g, h). Next, we assessed the polarity of AQP4 after the IFNγ treatment. AQP4 depolarization was obvious after IFNγ treatment in the R1441G mice, as evidenced by the increase in immunoactivity of AQP4 in the soma of astrocyte and the decrease of AQP4 polarity, while this effect could be rescued by LRRK2 inhibitor (Fig. 6i, j). Furthermore, western blotting study showed that AQP4 expression in the cell membrane was lower in the R1441G + IFNγ group than in the R1441G group. However, treatment with LRRK2 inhibitor partially restored the loss of AQP4 polarity (Fig. 6k–n). IFNγ treatment also decreased the expression of TH in R1441G mice, while LRRK2 inhibitor could rescue it (Fig. 6o–r). There was no significant difference in the expression of TH in the WT mice treated with or without IFNγ or LRRK2 inhibitor (Fig. 6o–r). These findings suggest that IFNγ upregulates the expression of IFNGR-2 in astrocytes, reduces AQP4 polarity and induces TH neuron loss.

Animals and LRRK2 inhibitors treatment

Animal procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of Sun Yat-sen University (Guangzhou, China). The animals were housed in a facility maintained under specific-pathogen-free (SPF) conditions and were provided with unlimited food and water. FVB/N-Khdrbs2Tg (LRRK2*R1441G)135Cjli/J (Strain #:009604) mice harbor a mutated form of the human leucine-rich repeat kinase 2 (LRRK2 R1441G) gene. To serve as wild-type (WT) controls, FVB/N-Tg (LRRK2)1Cjli/J (Strain #: 009610) mice expressing the WT human LRRK2 gene were utilized. Both the WT and R1441G LRRK2 mice were bred in the FVB background. For the experiments involving interferon-gamma (IFNγ) clearance, FVB mice (Strain #: 001800) were employed. All three types of mice were ordered from the Jackson Laboratory and bred at the Guangdong.

Three- and four-month-old WT or R1441G mice were subjected to intraperitoneal injections of either 0.9% sterile NaCl or LPS at a dose of 5 mg/kg (L2630, Sigma, USA) to induce a PD-like mouse model. For LRRK2 inhibitor treatment, GTPase inhibitor (Fx2149) was administered to mice at 10 mg/kg intraperitoneally, twice daily for three days, while kinase inhibitor (Mli-2) was administered at 10 mg/kg intraperitoneally, twice daily for ten days. Seven days after the LPS treatment, the mice were sacrificed for subsequent pathological studies. All experimental procedures were conducted in strict adherence to The Code of Ethics of the World Medical Association (Declaration of Helsinki) for animal experiments, ensuring the ethical treatment and welfare of the animals involved.

DNA constructs and CO-IP assays

Different domains of LRRK2 tagged with FLAG were constructed and then co-transfected with HA-AQP4 in HeLa cells. Cell lysates or brain tissues from mice were homogenized in protein lysis buffer (Beyotime, Shanghai, China) for quantitation of total protein and immunoprecipitation assays. The Dynabeads protein G immunoprecipitation kit (Invitrogen, Carlsbad, CA) was employed for immunoprecipitation. Briefly, 1.5 mg of Dynabeads magnetic beads were incubated with 2 μg of LRRK2-specific antibody or immunoglobulin (Ig) G for 10 min at RT to bind to the antibody. The beads were then washed and incubated with 500 μL of brain lysis buffer overnight at 4 °C to form the magnetic bead–antibody-antigen complexes. The complexes were then washed thrice, and proteins were eluted with 50 μL of elution buffer from the beads and processed for subsequent functional assays.

Primary culture of astrocytes

Newborn (1–2 days) LRRK2 WT/R1441G transgenic mice were employed to isolate primary cortical astrocytes. The protocol was modified by McCarthy and de Vellis. Briefly, the conical tubes were pre-filled with 2 mL fresh complete medium (high glucose Dulbecco’s modified Eagle’s medium +10% fetal bovine serum + 100 U/mL penicillin–streptomycin), and then the brain cortex was collected after decapitation and removal of meninges. After mechanical dissociation, samples were passaged through 70-μm nylon cell strainers (BD Biosciences). Then, the undifferentiated mouse brain cells were seeded onto 75-cm² flasks pre-coated with poly-L-lysine (Sigma-Aldrich) and 15 mL fresh medium was added. The cells were placed in a 37 °C incubator with a humidified atmosphere containing 5% CO₂. The medium was replaced every other day. To detach the microglial and precursor cells, the flasks were shaken for 2 min each time when we changed the medium during the first week. GFAP immunostaining was used to detect the purity; over 95% of the cells were astrocytes.

Co-culture on filter/cell culture inserts

The primary astrocytes were seeded onto the bottom of cell culture inserts (BD Biosciences) with a 1-μm pores (12 × 10⁴ cells/cm²) once the cells reached confluence. Subsequently, they were placed in an incubator for 6 h to facilitate cell attachment to the filter. The inserts were placed in a six-well plate filled with fresh medium. After astrocytes reached 80% confluence, which took about 7 days, HeLa cells were placed on the top of the filters (2 × 10⁴ cells/cm²)²⁹. After 7–10 days, the HeLa cells established a monolayer, and then the cells were fixed with 4% formaldehyde.

In vivo two-photon imaging of glymphatic pathway clearance

A total of 12 LRRK2 WT and R1441G mice (n = 6 for each group) were used in this study. The mice were anesthetized with 1% pentobarbital and subsequently positioned in a stable stereotaxic device. The right parietal cortex was made at lambdoid suture coordinates: mediolateral (ML) 2.0 mm, anteroposterior (AP) 1.7 mm. For imaging purposes, a slender cranial window measuring 3 mm in diameter was created.

To evaluate the PVS–ISF exchange, 10 μL of 1% FITC-dextran in artificial CSF was intracisternally injected at a rate of 1 μL/min using a microsyringe (BASi, West Lafayette, IN). Immediately before imaging, 0.2 mL of rhodamine B (1% in saline; Sigma-Aldrich) was administered via tail injection to label the vasculature. A laser scanning microscope (Leica Microsystems, Wetzlar, Germany) was used for our study. The laser was operated at 800 nm, and 25× water-immersion objective lens was utilized for imaging. To analyze the clearance of FITC-dextran from brain parenchyma, lateral images up to 100–300 μm below the cortical surface of the xyz stacks (512 × 512 pixels, 2-μm resolution) were collected at a series of time points: 10, 15, 30, 45 and 60 min after the injection.

Intraparenchymal injection of FITC-dextran

To evaluate the impact of mutant LRRK2 on the interstitial solute drainage from the brain into dCLNs, LRRK2 transgenetic mice were intraparenchymally injected with FITC-dextran. 1 μL of 1% FITC-dextran in artificial CSF was stereotactically injected into the parenchyma at a rate of 0.1 μL/min over 10 min with a syringe pump from the bregma of the brain(coordinates: AP 1.5 mm; ML 1.5 mm and dorsal/ventral −2.5 mm). The needle was held at the site for 10 min after the infusion was complete to avoid backflow. At 2 h after injection, dCLNs and brains were removed and fixed in 4% paraformaldehyde. Then, brains were cut into 50 μm-thick coronal brain slices via a vibrating blade microtome (Leica), while dCLNs(30 μm-thick) were sectioned using a freezing microtome (Leica). Images were acquired with a Leica confocal microscope and analyzed by a blinded investigator to evaluate the differences of glymphatic clearance between the control and experimental groups.

Immunofluorescence

Brain sections were taken from −80 °C and kept at room temperature (RT) for 1 h to recover. The tissues were first washed with phosphate-buffered saline and then blocked for 1 h at RT. Then, brain sections were incubated with different primary antibodies overnight at 4 °C and secondary antibody for 1 h at RT, with essential washing in the interval. The following antibodies were used: LRRK2 (MJFF2 [c41–2]; Abcam, Cambridge, UK), AQP4 (Santa Cruz Biotechnology, Dallas, TX), Alexa 488- or 555-conjugated secondary antibodies (Cell Signaling Technology, Danvers, MA). Images were captured by a Nikon microscope (Tokyo, Japan). The polarity of AQP4 in each image was characterized by the ratio of low-stringency areas (the total area exhibiting AQP4 immunoreactivity) to high-stringency areas (the region with pronounced AQP4 immunoreactivity).

Western blotting

Protein samples were prepared from the homogenized lysates of mouse brain tissue, and the concentrations were detected with the Pierc BCA Protein Assay Kit (Invitrogen). Proteins were separated by SDS-PAGE gel (Epizyme, China), transferred to a polyvinylidene fluoride membrane, and blocked with 5% milk for 1 h at RT. The membranes were then incubated with the following primary antibodies: LRRK2 (Abcam), AQP4 (Proteintect), p-Ser (Millipore), and GAPDH (Cell Signaling Technology) overnight at 4 °C and a secondary antibody (Cell Signaling Technology) for 1 h at RT. Finally, the membrane was visualized using a chemiluminescence method. All blots or gels derive from the same experiment and they were processed in parallel.

Statistical analysis

Statistical analyses were performed by blinded researchers with GraphPad Prism 9.0 (GraphPad Software, San Diego, CA). Data were analyzed with two-way ANOVA to evaluate interactions among the groups, and Tukey’s test was used to detect differences between groups and the repeated measures. One-way ANOVA followed by Tukey’s post hoc test was used for experiments with more than two groups. Independent-sample t-tests were used to analyze all other data. Statistical significance was set to p < 0.05.