Background

Parkinson’s disease (PD) is one of the most common neurodegenerative disorders. Clinically, it is characterized by a triad of classical symptoms: resting tremor, rigidity or hypokinesia. At pathological level, the hallmarks of PD include a predominant loss of midbrain dopaminergic (mDA) neurons in the substantia nigra pars compacta (SNpc) and the presence of Lewy bodies containing aggregated α-SYNUCLEIN filaments, and/or TAU hyperphosphorylation [1, 2]. Current treatments for PD are symptomatic and do not affect the progressive nature of the degenerative process. Efforts aiming at develo** more effective therapies capable of stop** or delaying disease progression currently involve gaining a better understanding of the function of proteins involved in PD. Ninety percent of the patients suffer from sporadic idiopathic forms of PD [3]. In last decades, several genetic forms of PD, accounting for only 10% of the cases, have been identified. Notably, several proteins that are implicated in genetic forms of PD, such as PARKIN, TAU, α-SYNUCLEIN, PINK1 and DJ-1, have been also associated with sporadic PD [4, 5].

Autosomal-dominant mutations in Park8 gene encoding Leucine-rich repeat kinase2 (LRRK2) is one of the most prominent risk factors for sporadic PD with a mutation frequency of 2-40% in different populations [3, 6, 7]. Interestingly, these patients exhibit typical features of idiopathic, late-onset PD, indicating that even LRRK2-mediated disease requires aging [1, 5]. LRRK2 is a large, multi-domain protein composed of 2527 amino acids (289 kDa). It contains a kinase domain sequence, a Ras of complex protein domain (ROC) and the C-terminal of COR (COR) domain that are predicted to bind and hydrolyze GTP similarly to the ROCO protein family [8]. These three domains are considered the catalytic core of LRRK2. Additionally, LRRK2 has Ankyrin repeats, Leucine-rich repeats (LRR) and a WD40 domain that predominantly serve as binding sites for protein-protein interactions and structural scaffolds for different signaling processes, which is another important function of LRRK2. However, the precise function of LRRK2 in cell signaling remains to be defined.

WNTs (from Win gless/ Int egration) are a large family of 19 secreted lipid-modified glycoproteins that serve multiple functions in development, tissue homeostasis, regeneration and disease [9]. In mammalian nervous system, WNT morphogens play a crucial role in formation and modulation of neuronal circuits [10]. Three main WNT signaling pathways have been described: WNT/β-catenin, WNT/Planar Cell Polarity (PCP) and WNT/Calcium (Ca2+), the two latter also referred to as non-canonical or WNT/β-catenin-independent signaling pathways. Upregulation of WNT/PCP signaling usually inhibits the WNT/β-catenin signaling and vice versa [11, 12]. We previously found that proper levels of WNT/β-catenin and WNT/PCP signaling are required for correct mDA neuron development and function [13,14,15,16].

A few studies have provided evidence that WNT/β-catenin signaling components interact with “PD proteins”. We previously found that PARKIN, an ubiquitin E3 ligase forming a complex with PINK and DJ-1 [ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/; version 20141001; 85,893 protein sequences) with additional sequences from cRAP database (downloaded from http://www.thegpm.org/crap/). Mass tolerance for peptides and MS/MS fragments were 10 ppm and 0.05 Da, respectively. Oxidation of methionine and deamidation (N, Q) as optional modification and two enzyme miss cleavages were set for all searches. Propionamidation of C as optional modification was set for MS database searches after in-gel digestion. Percolator was used for post-processing of Mascot search results. Peptides with false discovery rate (FDR; q-value) < 1%, rank 1 and with at least 6 amino acids were considered. Proteins matching the same set of peptides were reported as protein groups. Protein groups/proteins were reported only if they had at least one unique peptide. Label-free quantification using protein area calculation in Proteome Discoverer was used (“top 3 protein quantification”). Two analyses of on-beads digests and analyses for all gel area within single gel line were searched as one dataset.

Database search

Tandem mass spectra were extracted using Raw2MGF (KI in-house software), and the resulting mascot generic files from each lane were searched against the mouse SwissProt protein database using the Mascot Daemon 2.3.02 search engine (Matrix Science Ltd.), which was set to search the SwissProt protein database (selected for Homo s. sapiens, version 2012.03) using trypsin and two missed cleavage sites. Peptide mass tolerance was set to 10 ppm and 0.05 Da for the fragment ions. Carbamidomethylation of cysteine was specified as a fixed modification, whereas oxidation of methionine and deamidation of asparagine and glutamine were defined as variable modifications. The lists of identified proteins were exported from the .dat files using the following criteria: significance threshold of 0.05, MudPit protein scoring, required red bold, and including same-set proteins. Mascot score threshold was set to 50.

MS/MS data mining

We performed a literature study on identified proteins using UniProt [34], PubMed and Web of Science databases. Overexpression studies were performed in HEK293T, therefore the protein IDs had to be converted into mouse Ids, and not all the hits were present in the mouse genome. The conversion of the protein ID was done by biological DataBase network (bioDBnet).

Results

Unbiased identification of WNT/PCP proteins as candidate LRRK2 binding partners

Proteomic approaches such as tandem mass spectrometry (MS/MS) has become a powerful tool to discover unknown binding partners of any protein of interest. Nevertheless, the stoichiometry of protein-protein interactions and their preservation in a test tube is highly dependent on the cell type and sample processing [35,36,37]. To capture the diversity of candidate LRRK2 interactions, we used three different conditions: (1) NP40 buffer and in-gel tryptic digestion of human HEK293T; (2) NP40 buffer and in-gel tryptic digestion of a mouse substantia nigra cell line (SN4741) [28] expressing endogenous LRRK2; (3) Sodium deoxycholate buffer (SDOCH) and digestion on beads using SN4741 cells (Fig. 1a). Only proteins absent in the IgG immunoprecipitation control, but present in the LRRK2 pull-down, were considered as candidate LRRK2 binding partners. LRRK2 was never present in IgG controls (Additional file 3: Figure S2A, Additional file 4: Table S2 and Additional file 5: Table S3), but was one of the most abundant hits in our MS/MS analysis after pull-down with antibodies against C-MYC (for LRRK2 overexpression in human cells) or endogenous LRRK2 in mouse cells, confirming the good quality of our IP protocols. This strategy allowed us to identify about 500 proteins: 122 proteins interacting with overexpressed LRRK2 in HEK293T cells (in gel digestion), 120 proteins with endogenous LRRK2 in SN4741 cells (in gel digestion), and 283 proteins with endogenous LRRK2 in SN4741 cells (in solution digestion). The full list of proteins is provided in Additional file 4: Table S2 and Additional file 5: Table S3. Comparison of the 3 data sets (Fig. 1b) revealed only 2 proteins in common: BEN domain-containing protein 3 (BEND3), a repressor of transcription, and GIPC1, a PDZ domain containing protein (Fig. 1c). Notably, 15 proteins were found to bind endogenous LRRK2 in SN4741 dopaminergic cells in the two conditions analyzed (Fig. 1b-c). Analysis of the identified LRRK2 binding partners revealed the presence of 5 proteins that play an important role in WNT/PCP signaling, either via endocytic trafficking of the core WNT receptors or by direct modulation of WNT/PCP activity (asterisk in Fig. 1c, Additional file 6: Table S4). In addition, each of our conditions detected proteins that had been previously reported to bind LRRK2 (Table 1), confirming thus the validity of our approach.

Fig. 1
figure 1

Large-scale immunoprecipitation-coupled MS/MS screening of LRRK2 binding partners revealed proteins belonging to WNT/PCP signaling pathway. a Scheme of the experimental workflow. b Venn diagram showing results from 3 different MS/MS data sets. c A list of identified proteins which were detected at least in 2 experiments. Proteins that have been linked to WNT signaling are marked with a star, stronger evidence is denoted by with two stars

Full size image
Table 1 Previously reported LRRK2 binding partners from the Overexpressed LRRK2 MS/MS data set were confirmed our “Overexpressed LRRK2 in-gel” data set, proving the quality and validity of the experiment
Full size table

In vitro and in vivo validation of GIPC1 and ILK, two novel WNT/PCP binding partners of LRRK2

We subsequently validated some of the candidate LRRK2 binding partners in the SN4741 substantia nigra cell line (Fig. 2a-b) and in embryonic day E18.5 in ventral midbrain tissue (Fig. 2c) using LRRK2 immunoprecipitation and immunoblotting. These experiments revealed that LRRK2 binds to GIPC1 and ILK in vitro (Fig. 2a) and in vivo (Fig. 2c). To further verify these results, we used three different monoclonal antibodies to compare (Additional file 1: Figure S1A-B) a control GFP SN4741 cell line to a CRISPR/Cas9 LRRK2 knock-down (KD) SN4741 cell line with monoallelic frameshift mutation (Additional file 7: Figure S3A-B). We found that the pull down of GIPC1 or ILK with LRRK2 was reduced in LRRK2 KD SN4741 cells compared to control SN4741 cells, indicating the specificity of the interaction (Fig. 2b). These experiments therefore confirmed that LRRK2 physically binds to: (1) GIPC1, an adaptor protein for G protein coupled receptors (GPCRs) that controls endocytic trafficking of the WNT/PCP receptor VANGL2 [21]; and (2) Integrin-linked kinase (ILK), an Ankyrin repeat containing serine-threonine kinase that regulates WNT/PCP [22, 23] and WNT/β-catenin [38, 39] signaling via remodeling extracellular matrix and cell-cell adhesion. Another interesting candidate from our analysis, Lipoma-preferred partner (LPP) bound to LRRK2 in SN4741 cells and in VM tissue (Fig. 2a, c) but we failed to confirm this interaction in our knock-down LRRK2 cell line (Fig. 2b). Thus our results identify two novel interactors of endogenous LRRK2 in dopaminergic cells: the WNT/PCP signaling components, GIPC1 and ILK.

Fig. 2
figure 2

Verification of the MS results in vitro and in vivo in mouse ventral midbrain. (a-c) Western blot validation of LRRK2 binding partners using specific antibodies. ILK and GIPC1 but no other WNT/PCP signaling components interact with LRRK2 in SN4741 (a-b) and mouse ventral midbrain of E18.5 old embryos (c). b We knocked down LRRK2 using CRISPR/Cas9 methodology and generated clonal cell lines with either normal (WT) or decreased levels of LRRK2 (KD). SN4741-LRRK2-KD served as a negative control for our co-IP experiments. d-f Western blot analysis of co-IP of transiently overexpressed candidates shows that LRRK2 binds to FLOTILLIN-2 (d) and CULLIN-3 (e) in HEK293T. JIP3 (f), a previously published LRRK2 binding partner, served as positive control. g-i IF confirmed that LRRK2 co-localizes with FLOTILLIN-2 (g), CULLIN-3 (h), and JIP3 (i) in HEK293T cells. Nuclear staining by Dapi is in blue. N ≥ 3. Scale bars indicate 20 μm

Full size image

FLOTILLIN-2 and CULLIN-3 are novel WNT binding partners of LRRK2

Our in silico analysis of the identified LRRK2 interactors in HEK293T cells revealed 6 functional group of proteins (Additional file 3: Figure S2B), including a group of proteins potentially involved in WNT signaling, such as FLOTILLIN-1/2, CULLIN-1 and SHROOM-3 (Additional file 3: Figure S2C). These candidates were validated by transient overexpression and immunoprecipitation of LRRK2. We used JIP3 as positive control for LRRK2 co-IP [40]. While we did not detect any interaction of LRRK2 with CULLIN-1 or SHROOM-3, and only a weak interaction with FLOTILLIN-1 (data not shown), we confirmed that LRRK2 interacts with FLOTILLIN-2 (Fig. 2d) and JIP3 (Fig. 2f). We also found an interaction of LRRK2 with CULLIN-3 (Fig. 2e), a protein that was not identified by MS/MS, but it is known to be down-regulated by Lrrk2 knock-down [41], and to inhibit WNT/β-catenin signaling [42]. In addition, we also found that LRRK2 immunofluorescence is present in cell lamellipodia, together with FLOTILLIN-2 (Fig. 2g), or in the cytoplasm of SN4741 cells together with CULLIN-3 and JIP3 (Fig. 2h-i), reinforcing the concept that LRRK2 interacts with several proteins regulating WNT signaling in dopaminergic neurons.

LRRK2 interacts with PRICKLE1 and CELSR1 but not with other core components of WNT/PCP pathway

To further determine whether there is a possible crosstalk between LRRK2 and WNT/PCP signaling, we performed co-IP and immunoblotting experiments in HEK293T cells after overexpression of LRRK2 and some WNT/PCP signaling components, such as Ror2, Vangl2, Prickle1, and Celsr1. We found that LRRK2 specifically interacts with CELSR1 (Fig. 3a) and PRICKLE1 (Fig. 3b) but not with other key components of the WNT/PCP signaling pathway, such as VANGL2 (Fig. 3c) or ROR2 (Fig. 3d). LRRK2-CELSR1 interaction was further confirmed by using different vectors and antibodies for pull-down (Additional file 8: Figure S4A). The efficiency of the pull-down increased when more LRRK2 and less CELSR1 was transfected (Additional file 8: Figure S4B). LRRK2 immunofluorescence was found in the plasma membrane and cell-cell contacts, together with CELSR1 (Fig. 3h-i), while it was detected in cytoplasmic punctate structures together with PRICKLE1 (Fig. 3j-l), following a pattern similar to that of polymerized DVL [19, 43,44,45,46]. We also examined whether the physical interactions between LRRK2 and CELSR1 or PRICKLE1 were modified by some of the most common PD-related LRRK2 mutations, such as Y1699C, R1441C and G2019S (Fig. 3e-g). However, we did not observe any alteration in the physical interaction of these proteins, indicating that LRRK2 mutations do not affect the capacity of LRRK2 to bind core WNT/PCP proteins.

Fig. 3
figure 3

LRRK2 interacts with CELSR1 and PRICKLE1, two key components of WNT/PCP signaling. a-d Western blotting analysis of co-IP of transiently overexpressed LRRK2 with WNT/PCP signaling components in HEK293T. LRRK2 physically interacts with CELSR1 (a) and PRICKLE1 (b), but neither with VANGL2 (c) nor ROR2 (d). e-i IF of proteins overexpressed in HEK293T. Under normal conditions, LRRK2 is evenly distributed in the cytoplasm (i). CELSR1 alone is usually polarized in cells (f), whereas PRICKLE1 tends to form puncta (h). e Once co-expressed, LRRK2 partially co-localizes with CELSR1 in the cytoplasmic membrane and in cell-cell contacts (arrowheads). LRRK2 robustly changes its localization when it is co-expressed with PRICKLE1, and together they form puncta structures in the cytoplasm (g). Arrowheads show co-localizations, whereas arrows point out that there is no leakage of the fluorescent signal in PRICKLE1-positive and LRRK2-negative cells. Nuclear staining by Dapi is in blue. N ≥ 3. Scale bars show 20 μm. j-l Physical interaction between LRRK2 and CELSR1 (j) or PRICKLE1 (k-l) is not modified in the most common LRRK2 mutations. l Analysis of immunoblot band intensity shows the relative binding of LRRK2 WT and mutants to PRICKLE1. No difference between LRRK2 WT and mutants was detected. Signal was adjusted to the background and normalized to the corresponding input (N = 3, SD)

Full size image

Localization of LRRK2 into PRICKLE1-induced puncta is independent of endosomal processes and form signalosome-like structures

It has been previously shown that LRRK2 interacts with several RAB family proteins and regulates the endosomal-lysosomal pathway [47,48,49,50]. We thus examined whether LRRK2-PRICKLE1 puncta are present in endosomal and/or lysosomal compartments in transiently transfected HEK293T cells. We found that LRRK2 does partially co-localize with RAB5a (marker of early endosomes) and very sporadically with RAB11 (marker of recycling endosomes), but does not co-localize with RAB7, a marker of late endosomes (Additional file 9: Figure S5A-C). We also found that PRICKLE1 as well as DVL2 strongly co-localize with RAB7 (Additional file 8: Figure S4C-D), while LRRK2-PRICKLE1 puncta neither co-localize with this late endosomal marker, nor with the lysosomal marker LAMP1, despite LRRK2 alone co-localized with LAMP1 in the cytoplasmatic membrane (Fig. 4a-b, Additional file 9: Figure S5). These results suggest that LRRK2-PRICKLE1 puncta, unlike LRRK2 or PRICKLE1 alone, are not localized in the endosomal or lysosomal compartments.

Fig. 4
figure 4

PRICKLE1-induced LRRK2 puncta are not in the endosomal compartment but rather in signalosome-like structures, together with DVL2 polymers. a-e Immunofluorescence analysis of the subcellular localization of transiently overexpressed LRRK2. a Areas double positive for LRRK2-PRICKLE1 are negative for the lysosomal marker LAMP1. Arrows point to the partial co-localization of LRRK2 with LAMP1 in the membrane. Arrowheads point to LAMP1, which does not localize with LRRK2-PRICKLE1 puncta. b A late endosomal marker, RAB7, co-localizes with PRICKLE1 (arrowheads), but not with LRRK2 puncta (arrows). c LRRK2 forms puncta with DVL2. d PRICKLE1 and DVL2 are found in close proximity, but do not co-localize. PRICKLE1 is surrounded by 1 to 3 DLV2 complexes. e In the presence of LRRK2, DVL2 and PRICKLE1 partially co-localize (arrows). The co-localization of LRRK2, DVL2 and PRICKLE1 reveals the capacity of these proteins to form signalosome-like structures. Arrowheads show the localization of PRICKLE1 in close proximity to DVL2 puncta and LRRK2. Nuclear staining by Dapi is in blue. N ≥ 3. Scale bars show 10 μm

Full size image

We next examined whether LRRK2-PRICKLE1 localize in DVL puncta, a WNT signaling compartment formed by DVL polymers [43,44,45,46]. First, we demonstrated that LRRK2 localizes in DVL2 puncta (Fig. 4c), as previously reported for DVL1 and DVL3 [19]. Strikingly, PRICKLE1 was found in very close proximity to DVL2, but they did not co-localize (Fig. 4d). However, in the presence of LRRK2, then PRICKLE1 and DVL2 co-localized (Fig. 4e), suggesting that LRRK2 facilitates this process. Combined, our results show that LRRK2-PRICKLE1 puncta are not in endosomes, but rather in DVL2 signalosomes, which suggests a possible regulation of WNT signaling.

LRRK2 inhibits WNT/β-catenin signaling pathway, while PRICKLE1 turns LRRK2 from an inhibitor to an activator of WNT/β-catenin signaling

There are no sensitive in vitro assays to measure the activity of WNT/PCP signaling at present. Nevertheless, it’s known that the activation of WNT/PCP signaling inhibits the WNT/β-catenin signaling [25, 42]. Here we examined whether LRRK2 and its deletion mutants (Fig. 5a) can alone or in combination with the WNT/PCP signaling component PRICKLE1 modulate the WNT/β-catenin TOPFlash reporter assay in HEK293T cells. We first found that expression of LRRK2 alone decreased the basal level of TOPFlash activation up to 40% (Fig. 5b, d). This effect was maintained after deletion of the armadillo domain (dHEAT) and lost by deletion of COR domain (KINASE, WD40 and LRRS LRRK2 mutants) (Fig. 5b) pointing out that the RocCOR domains are likely important for WNT/PCP signaling activation and/or WNT/β-catenin inhibition. We also found that PRICKLE1 in combination with LRRK2 induced TOPFlash/β-catenin activation (Fig. 5c). This activation was LRRK2-dependent, as it was reduced by LRRK2 truncations, being the ARMADILLO-ANKYRIN and the LRR domains essential, with a contribution of the WD40 domain (Fig. 5c). LRRK2 most common PD mutations did not inhibit TOPFlash in a significant manner (Fig. 5d). These data suggest that multiple domains of LRRK2 are necessary to regulate WNT/β-catenin signaling and that this regulation is context dependent, since PRICKLE1 is capable of turning LRRK2 from an inhibitor into an activator of this pathway.

Fig. 5
figure 5

LRRK2 inhibits the activity of WNT/β-catenin pathway – an indirect read-out of WNT/PCP signaling activity. a Scheme of the Lrrk2 truncated vectors. b TOPFlash assay in HEK293T cells overexpressing LRRK2-WT or truncations. LRRK2 suppresses the basal activity of WNT/β-catenin signaling. This effect is lost with the most severe LRRK2 truncations, LRRK2-KINASE, WD40 and LRRS, that lack RocCOR domain. c LRRK2 together with PRICKLE1 increases the TOPFlash activity. This effect is lost in mutants that lacked the LRRs domain and the Ankyrin repeats, but is partially maintained in the mutant missing only the Armadillo domain (dHEAT). d LRRK2 PD mutants did not significantly inhibit TOPFlash (Mann Whitney t test). e DVL2 alone and together with LRRK2 strongly activates WNT/β-catenin signaling. PRICKLE1 down regulates the DVL2-dependent activation even in presence of LRRK2. ANOVA with Holm-Sidak multi-comparison test and Mann Whitney T-test (for the PD mutants) was used for statistical analysis. Data show mean (N>3) and standard deviation

Full size image

PRICKLE1 reduces the activation of WNT/β-catenin signaling by competing with DVL2 in the presence of LRRK2

It has been previously shown that DVL increases WNT/β-catenin signaling activity [51,52,53], and that LRRK2 binds to DVL [19], and together they further increase the TOPFlash/β-catenin activity by 2-5 folds depending on the DVL isoform [20]. On the other hand, PRICKLE1 can bind DVL, which is then ubiquitinated and targeted for degradation, leading to the downregulation of WNT/β-catenin signaling [54]. This interaction has been proposed to be a mechanism by which PRICKLE1 regulates asymmetric localization of FZD and DVL across cell-cell contacts from the proximal to the distal side of the cell [55,56,57]. We therefore examined how LRRK2 and PRICKLE1 modulate the DVL2-induced activation of WNT/β-catenin signaling (Fig. 5e). Our results confirm that DVL2 induces the activation of WNT/β-catenin signaling (~60 fold). This activation was not modified by PRICKLE1, but was increased by LRRK2 (~130 fold compared to control) as previously described [51]. However, the DVL2-LRRK2 dependent activation of WNT/β-catenin signaling was reduced by PRICKLE1 down to ~80 fold. Constitutively active beta-catenin (ca-B-CATENIN) was used as a positive control (~500 fold) to show that the effect of DVL2-LRRK2 was not saturated. These data suggest that the co-localization of LRRK2, PRICKLE1 and DVL2 in signalosome-like structures (Fig. 4c-e) allows PRICKLE1 to inhibit the increase in WNT/β-catenin signaling induced by LRRK2-DVL2.

Combined, our data indicate that the LRRK2-PRICKLE1 protein complex acts as a modulator of WNT/β-catenin signaling, increasing or decreasing its activity in the absence or the presence of DVL2, respectively. This observation is consistent with several studies showing that protein levels, temporal events and the localization of LRRK2 within the cell are important determinants for the function and regulation of LRRK2 activity [20, 58,59,60,61].

Overexpression of LRRK2 in Xenopus laevis embryos causes inhibition of WNT/β-catenin signaling and convergence and extension defects

We next examined whether LRRK2 in addition of inhibiting WNT/β-catenin signaling in vitro, also does that in vivo. For this purpose we took advantage of Xenopus laevis, a well characterized model organism with well-established biological read-outs to examine the activation of WNT signaling pathways [62]. We first co-injected lrrk2 mRNA (1-2 ng) with the 14XTOPFlash and Renilla reporters’ DNA into dorsal marginal cells at 4-8 cell-stage embryos, and measured luciferase activity at stage 10.25 (Fig. 6a). These experiments revealed that overexpression of lrrk2 inhibits the WNT/β-catenin pathway in vivo (Fig. 6b), confirming thus our in vitro observations.

Fig. 6
figure 6

LRRK2 inhibits WNT/β-catenin signaling in vivo and mediates WNT/PCP signaling during Xenopus early development. a-b TOPFlash assay in X. laevis embryos. a Schematic drawing of the experiment. 1 or 2 ng of lrrk2 mRNA was co-injected with 200 pg 14XTOPFlash and 25 pg Renilla reporters DNA into dorsal marginal cells at 4 to 8 cell-stage of Xenopus laevis embryos. Firefly and Renilla luciferase were measured at stage 10.25. b Overexpression of lrrk2 in Xenopus laevis embryos lead to significant reduction of WNT/β-catenin activity in a dose-dependent manner. Two tailed T-test, N = 3. c-d Overexpression of lrrk2 mediates convergence and extension movement’s defects in Xenopus embryos. c Typical phenotype of the lrrk2 overexpressing embryo. 180 pg of lrrk2 (left) or β-galactosidase (right) DNA was injected in dorsal marginal cells at the 4 to 8-cell stage. d Bar chart of convergent extension defects in lrrk2 (180 pg DNA), β-galactosidase (180 pg DNA) and celsr1 (1 ng mRNA) injected embryos. e Categories used for evaluating the convergent extension defects. Grade 1: >0.95; Grade 2: 0.85-0.95; Grade 3: 0.75-0.85; Grade 4: 0.65-0.75; Grade 5: 0.55-0.65; Grade 6: 0.45-0.55; Grade 7: <0.45

Full size image

Our in vitro findings also provided evidence that LRRK2 inhibits WNT/β-catenin signaling through the activation of the WNT/PCP pathway. It has been shown that the WNT/PCP signaling controls the convergent extension movements [63, 64]. We therefore examined whether lrrk2 controls these processes during the early development of Xenopus embryos. Overexpression of lrrk2 was performed in the dorsal marginal zone at 4-8 cell stage embryos, and the embryos were subsequently analyzed at stage 32. Interestingly, LRRK2 misexpression resulted in a mild convergent extension defects compared to uninjected or β-galactosidase controls (Fig. 6c). In order to perform a more detailed analysis, we used celsr1 as positive control [26, 65] and evaluated the different grades of the convergent extension defects caused by WNT/PCP pathway deregulation. For that, we measured the length of anteroposterior axis in each embryo, and normalized it by the average length of uninjected embryos. Based on the analysis of numerous embryos (79 uninjected, 29 β-galactosidase, 46 lrrk2 injection and 27 celsr1) (Fig. 6d), we divided the WNT/PCP phenotypes into 7 categories (Fig. 6e). Our results show that 56.5% of lrrk2-injected embryos displayed a grade 2 phenotype and 30.4% a grade 3, whereas the phenotype of the uninjected and the β-galactosidase group were grade 1 in 92.4% and 72.4% of the cases, respectively. On the other hand, celsr1, a strong regulator of WNT/PCP pathway, exhibited more severe phenotypes (grade 2 = 18.5%, grade 3 = 55.6%, grade 4 = 14.8%, grade 5 = 7.4%, grade 6 = 3.7%). Thus, our in vivo data demonstrate that LRRK2 is a novel regulator of the WNT/PCP signaling pathway, as well as an inhibitor of the WNT/β-catenin signaling pathway.

Discussion

Identification of novel LRRK2 binding partners: GIPC1 and ILK

Although many LRRK2 substrates have been proposed, the identification of true LRRK2 binding partners at physiological levels and their function remains to be determined. Our IP-coupled MS/MS approach using 2 different in vitro cell types and 2 biochemical methods in combination with our candidate-based approach has provided valuable information about what proteins can bind to LRRK2 after overexpression and at endogenous levels. Here we identify a number of novel interactors that are functionally linked to the WNT/PCP signaling pathway, which have been validated by other biochemical methods. These include GIPC1, ILK, PRICKLE1, CELSR1, FLOTILLIN-2 and CULLIN-3. Interactions of LRRK2 with GIPC1 and ILK were further verified by immunoblotting in vitro and in vivo, in the ventral midbrain of E18.5 mouse embryos.

One of the novel LRRK2 interactors that we identify in our study is GIPC1, an adaptor protein for GPCRs that is involved in vesicle trafficking of multiple receptors [66]. GIPC1 interacts with VANGL2, a key WNT/PCP membrane protein. Disruption of GIPC1 function leads to defects in hair cell maturation and hair follicle orientation, a typical phenotype displayed by WNT/PCP deregulation [21]. Moreover, it has been shown that GIPC1 interacts with dopamine D2 and D3 receptors [67,68,69,70] and its mutation results in a reduction of DA neurons leading to locomotion defects in D. melanogaster [71]. It is thus conceivable that alterations in the interaction between LRRK2 and GIPC1 may be of importance for the function of mDA neurons, a possibility that remains to be investigated.

There is evidence that LRRK2 binds and phosphorylates focal adhesion kinase (FAK) by which it regulates motility of reactive microglia and its response to brain injury, both of which were impaired in G2019S mutants [72]. The role of LRRK2 in regulating cytoskeletal rearrangements is further supported by our observation that LRRK2 binds ILK, a serine-threonine kinase that remodels the cytoskeleton and the extracellular matrix, leading to changes in cell shape, cell polarity and cell motility [23, 73]. It has been shown that overexpression of ILK activates WNT/β-catenin signaling [38, 39], while down-regulation of ILK leads to planar cell polarity defects and failure in hair-follicle development [23]. Moreover, ILK induces WNT/PCP pathway when it binds to the proline-rich motifs of DVL and subsequently promotes RHOA/CDC42/RAC1-mediated cell motility [22, 74, 75]. This dual activity of ILK towards distinct WNT signaling pathways correlates well with our finding that its binding partner, LRRK2, also modulates the activity of WNT/β-catenin and WNT/PCP signaling depending on the exact composition of the interacting complex and its protein levels. Our findings thus provide a mechanistic basis for how LRRK2 could mediate such effects.

LRRK2 interacts with two core WNT/PCP signaling components: PRICKLE1 and CELSR1

Using a candidate-based approach we also discovered that LRRK2 specifically interact with PRICKLE1 and CELSR1, two key players of the WNT/PCP signaling pathway. Notably, the localization of LRRK2 in the cell varied depending on its binding partners. CELSR1 and LRRK2 were found at the plasma membrane, corresponding to oligomerized, autophosphorylated LRRK2 which displays high kinase activity [76]. On the other hand, PRICKLE1 and LRRK2 were found in cytoplasmatic puncta, a localization similar to the functional complexes formed by LRRK2 and DVL1/3 [19]. Interestingly, we demonstrate that PRICKLE1-LRRK2 localizes in DVL2 signalosome-like structures and functionally interacts. Indeed, PRICKLE1 turned LRRK2 from an inhibitor to an activator of basal WNT/β-catenin signaling, but it inhibited the DVL2-induced activation of this pathway even in the presence of LRRK2. These findings suggest that not only the localization, but also the functionality of LRRK2, is regulated by its interaction with distinct WNT/β-catenin and WNT/PCP signaling components.

While the reason for the differential localization is unknown, LRRK2 shares several functions with PRICKLE1 and CELSR1. LRRK2 regulates axon-dendrite polarity [77], endocytosis and synaptogenesis [50, 78,79,80,81], and neurite branching in human dopaminergic neurons [82]. Similarly, CELSR1 also regulates dendrite and axonal growth in cooperation with FZD, VANGL and DVL [83,84,85], and PRICKLE1 promotes neurite and axon outgrowth [86]. Moreover, mutations in human PRICKLE1 have been associated to seizures [87], progressive myoclonus epilepsy [88, 89] and autism [90]. These findings suggest that LRRK2, by interacting with CELSR1 or PRICKLE1 may contribute to the regulation of typical WNT/PCP-controlled functions such as synaptogenesis, vesicle trafficking and synaptic plasticity [91,92,93,94].

LRRK2 regulates both WNT/β-catenin and WNT/PCP pathways

It has been previously shown that LRRK2 regulates WNT/β-catenin signaling [19, 20, 95], a pathway that is in most cases negatively regulated by signaling components of the WNT/PCP pathway [25, 42]. LRRK2 has a GTPase activity and binds active RAC1 [96], a small GTPase that (besides other functions) acts downstream of WNT5A-dependent WNT/PCP signaling activation [46, 84, 97]. Interestingly, G2019S and R1441C mutations fail to activate RAC1 [96], and also result in altered levels of WNT/β-catenin signaling [20]. Moreover, a recent study [98] has shown that loss of function of LRRK2 increases WNT/β-catein signaling in cultured fibroblasts and in adult mice in vivo. These mice also exhibit increased bone density and altered tibia shape, a phenotype suggestive of WNT/PCP signaling, which plays a crucial role in limb development [99,100,101]. Indeed, PRICKLE1 is known to control the growth of long bones and its mutation causes severe skeletal malformations resulting in shorter and wider bones [102]. These data together with our results indicate that alterations in LRRK2 structure and/or function are likely to modify the balance between WNT/β-catenin and WNT/PCP signaling. Strikingly, our results show that LRRK2 has the capacity to either inhibit or upregulate WNT/β-catenin signaling depending on its interacting partners and levels of WNT signaling. Moreover, overexpression of LRRK2 in human cells demonstrated that LRRK2 alone inhibits WNT/β-catenin signaling via its RocCOR domain. This result was confirmed in vivo, in Xenopus embryos, where LRRK2 not only inhibited the WNT/β-catenin signaling but also regulated the WNT/PCP signaling pathway, as shown by alteration of convergence-extension movements, the process by which cells intercalate and the embryo elongates. Thus our results provide first evidence for a role of LRRK2 in WNT/PCP signaling pathway.

Core WNT/PCP signaling components such as PRICKLE1 [54] and CELSR1 [27], have been previously found to inhibit WNT/β-catenin signaling. Surprisingly, PRICKLE1 in complex with LRRK2 increased basal WNT/β-catenin signaling, but reduced the DVL-induced activation of WNT/β-catenin signaling. Thus our findings suggest a role of PRICKLE1 and LRRK2 as dual regulators of WNT signaling by switching between WNT/β-catenin and WNT/PCP signaling activity.

Conclusions

Our study shows for the first time that LRRK2 interacts with WNT/PCP signaling pathway through its physical binding to multiple WNT/PCP regulatory components. By performing a proteomic screening we discovered several candidate LRRK2 interactors, and verified a few proteins, such as GIPC1 and ILK, which bind to endogenous LRRK2 in dopaminergic cells and in mouse ventral midbrain. We also demonstrate that LRRK2 can bind to FLOTILLIN-2, CULLIN-3, CELSR1 and PRICKLE1, which lead to changes in LRRK2 localization. While LRRK2 has been previously shown to activate the WNT/β-catenin pathway, our data suggest that the regulation of WNT signaling is more complex. We found that the formation of specific protein complexes can trigger WNT/PCP signaling and inhibit the WNT/β-catenin pathway both in vitro and in vivo, in X. laevis. Our results thus show that LRRK2 works as a dual regulator of WNT/β-catenin and WNT/PCP signaling (as PRICKLE1 and ILK), and suggest a possible additional role of LRRK2 in controlling biological processes regulated by the WNT/PCP pathway and deregulated in PD, such as axon-dendritic polarity and synaptic function.