Regulation of myeloid cell phagocytosis by LRRK2 via WAVE2 complex stabilization is altered in Parkinson’s disease

LRRK2 Regulates WAVE2 Expression.

We first determined if LRRK2 status alters the levels of the WAVE proteins in different mammalian cell types. In the BV2, murine microglia-like cell line, Wave2 protein levels were decreased upon Lrrk2 knockdown. However, the levels of Wave1 and Wave3 remained unchanged, as did the transcript levels of Wave2 (SI Appendix, Fig. S1A). Primary microglia and bone marrow-derived macrophages (BMDMs) from Lrrk2 KO mice displayed a reduction in Wave2 protein (Fig. 1 A and B). Conversely, microglia and BMDMs from hyperkinase-conferring Lrrk2–G2019S knockin (KI) mice showed an increase (Fig. 1 C and D). Furthermore, BMDMs from BAC Tg mice overexpressing LRRK2 also display increased Wave2 protein levels (Fig. 1E). The Wave2 transcript was unchanged in Lrrk2 KO or G2019S microglia (SI Appendix, Fig. S1B). Intriguingly, Wave2 protein levels were unchanged in other cell types tested, including primary astrocytes, cortical neurons, and mouse embryonic fibroblasts (SI Appendix, Fig. S1 C–E). To confirm if Wave2 levels were altered in vivo, tissue punches of the SNc, striatum (STR), and whole brain (WB) lysates from Lrrk2–G2019S KI mice were examined. Wave2, but not Wave1 or Wave3, protein was increased in SNc, STR, and WB (SI Appendix, Fig. S1 F–H).

LRRK2 Deficiency Impairs Phagocytic Activity in Myeloid Cells.

Given that LRRK2 status modulates WAVE2 levels in myeloid cells and WAVE2 is a central protein in regulating actin dynamics and phagocytic activity (17, 25), we examined LRRK2’s potential role in phagocytosis by conducting systematic engulfment assays in myeloid cells. We found that Lrrk2 KO microglia display impaired uptake of latex beads or Escherichia coli bioparticles, both basally and after treatment with various TLR agonists: LPS, polyI:C, or zymosan (Fig. 2A and SI Appendix, Fig. S2A). Furthermore, phagocytic efficiency was also altered in Lrrk2 deficiency (Fig. 2 B and C). A similar phagocytic impairment was observed in Lrrk2 KO BMDMs (Fig. 2D). Importantly, this impairment was reversed by exogenous WAVE2 expression (Fig. 2D). We then directly injected pH-sensitive beads, which fluoresce upon lysosomal-fusion, into the midbrain of mice as previously described (26). There was a significant reduction in bead uptake in the midbrain of Lrrk2 KO mice (Fig. 2E). Decreased WAVE2 is predicted to reduce the ratio of filamentous (F) to free globular (G) actin. Therefore, we performed F/G-actin assays and determined that Lrrk2 KO microglia have a decreased F/G-actin ratio (Fig. 2F). Additionally, we observed a decrease in F/G-actin ratio in Lrrk2 KO BMDMs that was rescued by Wave2 expression (Fig. 2G).

LRRK2–G2019S Increases Phagocytic Activity in Myeloid Cells.

Similar to our assessment of phagocytic activity in Lrrk2 KO myeloid cells, we determined whether LRRK2–G2019S myeloid cells displayed alterations in engulfment. In contrast to Lrrk2 KO animals, G2019S-KI microglia and BMDMs display increased engulfment of either latex beads or E. coli bioparticles basally and in a TLR-independent manner (Fig. 3A and SI Appendix, Fig. S2 B–D). Moreover, phagocytic efficiency in G2019S-KI microglia was elevated in both basal and LPS treatment (Fig. 3 B and C). There was also an increase in engulfment of pH-sensitive beads in G2019S-KI mice (Fig. 3D) and a significant increase in F/G-actin in G2019S-KI microglia (Fig. 3E).

To determine if Wave2 mediates the increase in phagocytosis in G2019S cells, we down-regulated Wave2 by shRNAi in both G2019S BMDMs and mice. Wave2 reduction in G2019S-KI BMDMs (Fig. 3F) and mice (Fig. 4 and SI Appendix, Fig. S3) curbed the hyperphagocytic activity to WT levels, suggesting that G2019S-mediated phagocytosis is regulated via an increase of Wave2 protein. Additionally, we tested if exogenous LRRK2 could modify phagocytosis in macrophages. Consistently, BMDMs from WT hLRRK2 BAC Tg mice displayed increased bead or bioparticle engulfment (SI Appendix, Fig. S2 E and F).

Finally, we extended our observations to PD patients carrying the G2019S mutation (SI Appendix, Table S1). We obtained monocyte-derived M1 and M2 macrophages from venous blood of LRRK2–G2019S PD patients and healthy controls. Both M1 and M2 macrophages displayed increased WAVE2 protein levels in LRRK2–G2019S PD patients compared with controls (Fig. 5 A and B). However, the protein levels of WAVE1 remained unchanged. Both M1 and M2, G2019S human macrophages exhibited an increase in bead engulfment (Fig. 5 C and D). Taken together, these results demonstrate that LRRK2 is required for regulating phagocytic activity via WAVE2 and this action is enhanced by the G2019S mutation.

WAVE2 Is Directly Coupled to LRRK2.

We next explored how WAVE2 is regulated by LRRK2. First, we conducted coimmunoprecipitation studies to determine if Lrrk2 and Wave2 form a complex. Lrrk2 endogenously forms a complex with Wave2 that is not detectable in Lrrk2 KO BMDMs (SI Appendix, Fig. S4 A and B). Additionally, Lrrk2–Wave2 complexes are observed in BV2 cells but not in LRRK2 null astrocytes, which did not show observable engulfment defects (SI Appendix, Fig. S4 C and D), further supporting their cell-specific relationship. Finally, we conducted a proximity ligation assay (PLA) capable of detecting close proximity (<40 nm) protein–protein interactions in a cell. We found that PLA spots were significantly increased in G2019S-KI BMDMs that were specific to Lrrk2–Wave2 as opposed to Lrrk2–Wave1 (Fig. 6A and SI Appendix, Fig. S5 A and B), and were undetectable in Lrrk2 KO cells (SI Appendix, Fig. S5C). Furthermore, we assessed if the LRRK2–WAVE2 complex is localized in the phagosome or with phagocytosed beads. PLA spots were colocalized with phagocytosed beads and Rab5a, an early endosome marker that regulates fusion between phagosome and endosome and stimulates phagosome-phagosome fusion (Fig. 6 A and B and SI Appendix, Fig. S6A) (27). Moreover, phagocytosed beads were also colocalized with Rab5a protein (SI Appendix, Fig. S5D). However, the intensity of Rab5 was not altered by the G2019S mutation, suggesting that LRRK2–WAVE2 complexes might affect to endosome–phagosome fusion or Rab5a activity.

To ascertain that LRRK2 directly interacts with WAVE2, a GST pull-down assay was performed with recombinant Flag-tagged full-length (FL) hLRRK2 proteins (WT and G2019S) and bacterially generated recombinant GST-Wave2. Wave2 interacts with hLRRK2-WT, as determined by immunoblot with an anti-Flag antibody, compared with GST alone (Fig. 6C and SI Appendix, Fig. S6B). Moreover, this interaction was enhanced by the G2019S mutation. These results further indicate that the increased kinase activity reported for the G2019S mutation may be crucial for the interaction between LRRK2 and Wave2.

LRRK2 Directly Phosphorylates WAVE2, Thereby Regulating WAVE2 Stability.

LRRK2 is a serine/threonine kinase (28). Therefore, we assessed whether LRRK2 directly phosphorylates Wave2. We examined phosphorylation of recombinant GST-Wave2 using an in vitro kinase assay by either fluorescent-labeled ATP or antiphosphothreonine (pThr) and antiphosphoserine (pSer) antibodies. GST-Wave2 was directly phosphorylated by Flag-tagged FL hLRRK2-WT, which was further increased by Flag-tagged FL hG2019S. In contrast, the GST-tagged kinase dead hD1994A (970–2,527 amino acids; N-terminal truncated form) did not phosphorylate Wave2 (Fig. 6D and SI Appendix, Figs. S6C and S7A). These results suggest that Wave2 is a substrate of LRRK2 kinase.

To determine the Wave2 sites phosphorylated by LRRK2, we combined in vitro kinase assays with mass spectrometry-based phosphopeptide mapping. This identified two Wave2 phosphopeptides that are consistently enriched following incubation with hLRRK2 (WT and G2019S), but not hD1994A. Both the phosphorylated residues are Thr sites, T96 and T470, in the Wave2 sequence (SI Appendix, Fig. S7B). We found then determined the threonine phosphorylation state of Wave2 in the cell. A significant decrease, but not absence, of pThr of Wave2 was observed in Lrrk2 KO BMDMs, with or without LPS treatment, which was rescued by adenoviral (Ad)-hLRRK2-WT infection. Furthermore, pThr levels were enhanced by the G2019S mutation compared with pThr levels in non-Tg BMDMs (control) (SI Appendix, Fig. S8).

To further determine which phosphorylation site is associated with regulation of phagocytosis, Ad-Wave2 mutants were transduced into non-Tg or G2019S-KI BMDMs. Overexpression of WT-Wave2 increased phagocytic activity in both Lrrk2 genetic backgrounds, an observation mimicked by T470E overexpression. In contrast, phagocytosis was significantly reduced in both non-Tg and G2019S-KI BMDMs with overexpression of the T470A mutant but not by T96E or A mutants (Fig. 6E). Furthermore, the F/G-actin ratio was suppressed by transduction of the T470A mutant (Fig. 6F). Collectively, these results suggest that direct phosphorylation of Wave2 at T470 by LRRK2 is an important regulator of phagocytosis. We then confirmed that in vitro phosphorylation of Wave2 by LRRK2 is dramatically reduced in a T470A mutant (Fig. 6G and SI Appendix, Fig. S6D). Taken together, these data suggest that T470 is the likely candidate for LRRK2-mediated phosphorylation.

To determine how LRRK2 affects the levels of Wave2 we hypothesized that phosphorylation by LRRK2 affects Wave2 protein stability. We evaluated the stability of Wave2 following treatment with the protein synthesis inhibitor, cycloheximide (CHX). Lrrk2 KO or G2019S-KI BMDMs display a marked reduction or increase in Wave2 stability, respectively (SI Appendix, Fig. S9 A and B). Additionally, proteasomal inhibition increased endogenous levels of Wave2, particularly in the absence of Lrrk2 (SI Appendix, Fig. S9C). Protein levels of expressed Flag-WT-Wave2 were also increased with expression of hG2019S and reduced with hD1994A compared with hLRRK2-WT expression. Under all three conditions, proteasomal inhibition increases levels of expressed Wave2 (SI Appendix, Fig. S9D). Although these experiments are confounded by the fact that both CHX and MG132 affect the quantity of Lrrk2 we attempted to link the identified phosphorylation sites of Wave2 to its stability. Flag-WT-Wave2 and Flag-Wave2 point mutants were cotransfected into HEK293T cells with GFP. Importantly, T470E was more stable than that of WT-Wave2 and the T470A mutant displayed the greatest reduction in protein levels by CHX. The observed differences were abolished with MG132 treatment (SI Appendix, Fig. S9 E and F). In contrast, mutations at Wave2 T96 did not significantly affect Wave2 protein levels (SI Appendix, Fig. S9G), suggesting that LRRK2-mediated Wave2 phosphorylation at T470 is crucial for protein stability.

LRRK2 Kinase Inhibition Decreases Phosphorylation of WAVE2, Phagocytosis, and Actin Assembly.

We examined if inhibition of LRRK2 kinase activity, using LRRK2 kinase inhibitors affected the LRRK2–WAVE2 pathway described above. We first assessed the effects of the LRRK2 inhibitors on Wave2 immunoprecipitated from non-Tg or G2019S-KI BMDMs. Both inhibitors significantly reduced Wave2 phosphorylation in both genetic backgrounds and suppressed the Lrrk2–Wave2 interaction in a manner similar to the pThr reduction rate. Furthermore, the level of Wave2 protein was reduced by LRRK2 kinase inhibition in both non-Tg and G2019S backgrounds (SI Appendix, Fig. S10A). Our findings suggest that LRRK2 kinase-mediated phosphorylation of WAVE2 is important for regulation of phagocytic activity. Thus, we predicted that LRRK2 kinase inhibitors would inhibit phagocytic function. Consistent with this model, both basal and LPS-induced bead engulfment were suppressed by kinase inhibition in both non-Tg and G2019S-KI microglia (SI Appendix, Fig. S10B). Importantly, LRRK2 kinase inhibitors had no effect in Lrrk2 KO microglia, indicating negligible off-target effects of the inhibitors. The F/G-actin ratio was similarly decreased by LRRK2 kinase inhibition (SI Appendix, Fig. S10C). Taken together, these results provide pharmacological data to bolster the model that LRRK2 kinase activity is critical for LRRK2 mediated WAVE2 phosphorylation, WAVE2 stability, and the concomitant increase in actin polymerization/phagocytosis.

LRRK2 Regulation of WAVE2 Determines DA Neuron Health.

To evaluate whether the LRRK2–WAVE2 pathway in myeloid cells affects the death of DA neurons, we conducted coculture studies of WT midbrain DA cultures with either Lrrk2 WT or KO BMDMs infected with Ad-GFP control or Ad-GFP/Flag-Wave2. BMDMs were used as opposed to primary microglia because more cells could be cultured and infection caused less toxicity. Tyrosine hydroxylase-positive (TH⁺) neuronal loss was suppressed when cocultured with Lrrk2 KO BMDMs compared with WT BMDMs. This suppression was reversed when Lrrk2 KO BMDMs were transduced with Ad-GFP/Flag-Wave2 (Fig. 7A and SI Appendix, Figs. S11A and S12A). To determine if soluble mediators secreted by myeloid cells could cause cell death, we incubated neuronal cultures with conditioned media from Lrrk2 WT and KO BMDMs and failed to observe TH⁺ cell loss (Fig. 7B). Conversely, G2019S-KI BMDMs enhanced the loss of TH⁺ neurons. Moreover, this loss was rescued by inhibition of Wave2 (Fig. 7C and SI Appendix, Figs. S11B and S12B). Conditioned medium from non-Tg and G2019S-KI BMDMs did not lead to loss of TH⁺ neurons (Fig. 7D). To ascertain if LRRK2-mediated T470 phosphorylation of Wave2 influences the loss of DA neurons, midbrain cells were cocultured with non-Tg or G2019S-KI BMDMs infected with Ad-GFP or Ad-GFP/Flag-T470A. We found that TH⁺ neuronal loss conferred by the G2019S mutation was ameliorated by T470A expression (Fig. 7E and SI Appendix, Fig. S11C). Together, our data suggest that LRRK2–WAVE2 actions in myeloid cells have the potential to directly impact neuronal survival. This impact likely requires direct cell–cell contact, as use of Arp2/3 complex inhibitor II, CK-869, prevented neuronal loss in our coculture model (Fig. 7F).

To confirm that our DA-BMDM coculture system could be recapitulated by DA-microglia, we cocultured midbrain DA neurons with primary microglia from non-Tg and G2019S mice. Consistently, the loss of DA neurons was exacerbated by G2019S microglia compared with coculture of DA neurons with non-Tg microglia (Fig. 7G). This exacerbation was ameliorated by treatment of CK-869, similar to that shown in Fig. 7H. Moreover, conditioned medium from primary microglia did not cause DA neuronal loss (Fig. 7H). Taking these data together, we find that this result suggests that the phagocytic response of microglia and BMDMs are modified by LRRK2 status that may affect the loss of DA neurons. While microglia and macrophages have different phagocytic responses in diverse conditions, these results suggest our BMDM–DA midbrain cocultures are an appropriate model for microglia in this context (29–31). Finally, we were unable to observe the canonical cell death marker, annexinV, in our coculture paradigm, suggesting the phagocytosis of cells without phosphatidylserine membrane presentation at this time point (SI Appendix, Fig. S12C).

WAVE2 Mediates DA Loss in Flies and Mice.

To demonstrate if the LRRK2 G2019S–WAVE2 pathway can mediate DA cell death in vivo, we directly injected the SNc of LRRK2-G2019S KI mice with shWAVE2 or control adenovirus followed 3 d later with LPS into same injection site. Similar to findings in the LRRK2 BAC G2019S rats (11), the G2019S KI mice displayed enhanced TH⁺ cell loss after stereological measurement (Fig. 8A). Importantly, TH⁺ cell loss was ameliorated by viral-mediated down-regulation of WAVE2. To support these findings genetically, we returned to the Drosophila model. We examined whether LRRK2–G2019S expression in immune cells could lead to DA cell death by utilizing the GAL4-UAS system to express LRRK2 specifically in Drosophila ensheathing glia (EG). EG are the established primary phagocyte of the Drosophila CNS (32). We ectopically expressed LRRK2–G2019S in EG under the mZ0709-GAL4 (MZ-GAL4) promoter, which results in survival deficits compared with control flies that can be ameliorated by inhibition of Scar in EG (Fig. 8B). These flies also display a decrease in activity that can be rescued by Scar knockdown (Fig. 8C). Finally, we determined if these same flies are subject to TH⁺ cell loss. In both the PPL1 and PPM1/2 clusters of the Drosophila CNS we observed a reduction in TH⁺ cells in flies expressing LRRK2–G2019S in EG, which was rescued by Scar down-regulation (Fig. 8D). We found no change in the total neuronal content in aged flies (30 d) by measuring levels of the pan-neuronal marker embryonic lethal abnormal vision (elav) (SI Appendix, Fig. S12E). Taken together, these data support a model by which LRRK2 mediates DA neuronal death by its function in immune cells via its relationship with WAVE2/Scar.

Midbrain Culture and Coculture with BMDMs or Microglia.

Primary mouse midbrain neuron culture and coculture systems were prepared following the published protocols (58, 59). The primary DA neurons were maintained in Neurobasal medium (Invitrogen) supplemented with B27, N2, 0.5 mM l-glutamine and penicillin/streptomycin for 14 d. Briefly, the mesencephalon of LRRK2 KO or G2019S and littermate control mice at embryonic day 14.5 were dissected and brain segments were dissociated by trypsin solution and mechanical triturating. After dissociation, the isolated cells were seeded at 1 × 10⁶ cells per well to 12-well plates precoated with poly-l-ornithine (20 μg/mL) and laminin (10 μg/mL), and maintained in complete neurobasal medium for 14 d. Adenovirus of WAVE2 or shWAVE2-infected BMDMs were collected via centrifugation. The BMDMs and primary microglia were then resuspended in complete neurobasal medium. Next, 1.5 × 10⁵ cells of BMDMs were seeded to 14-d-old midbrain cultures. After coculture, cells were stained with TH antibody to count for TH⁺ cells.

Human Macrophages.

Peripheral blood mononuclear cells were isolated from the blood of PD patients with G2019S mutations and controls in the families of the two groups using Ficoll-Paque PLUS (GE Healthcare) according to the manufacturer’s instructions and detailed in SI Appendix, SI Materials and Methods.

In Vitro and in Vivo Phagocytosis Assay.

For in vitro phagocytosis assay, cells were exposed to latex beads (Sigma) at a ratio of 1:15 (G2019S cells:beads) or 1:50 (KO cells:beads) ratio, or 0.05 mg/mL pHrodo Green E. coli bioparticles (Invitrogen) (SI Appendix, Table S2). For the in vivo phagocytosis assay, pH-sensitive latex beads were made by coupling 3-μm latex beads with CypHer5E mono N-hydroxysuccinimide (NHS) ester (GE Healthcare). Details are provided in SI Appendix, SI Materials and Methods.

Drosophila Studies.

All flies were maintained on a standard cornmeal/agar medium at ambient 25 °C when under experimental conditions. LRRK2 flies were characterized previously (60). UAS-hLRRK-WT and UAS-LRRK2-G2019S flies were a gift from Bingwei Lu, Stanford University, Stanford, CA. mz0709-GAL4 flies were graciously provided by Marc Freeman, Oregon Health & Science University, Portland, OR. Scar-RNAi was obtained from the Vienna Drosophila Research Centre. Details are provided in SI Appendix, SI Materials and Methods.

Human and Animal Ethics.

Human biological samples were collected in accordance with written informed consent and under Good Clinical Practice guidelines and approved by the ethics committee at the University Health Network, Canada. All procedures involving mice were approved by the University of Ottawa Animal Care Committee and were maintained in accordance with the Guidelines for the Use and Treatment of Animals described by the Animal Care Council of Canada and endorsed by the Canadian Institutes of Health Research.

Statistical Analysis.

All values are expressed as means ± SEM. The statistical significance of difference between two groups was determined by unpaired two-tailed Student’s t test. To compare multiple groups, statistical significance was determined by one-way ANOVA and Tukey’s post hoc test or two-way ANOVA followed by Bonferroni post hoc test using Graph Pad Prism 6 (GraphPad Software).