Pathogenic LRRK2 variants are gain-of-function mutations that enhance LRRK2-mediated repression of β-catenin signaling

Abstract

Background: LRRK2 mutations and risk variants increase susceptibility to inherited and idiopathic Parkinson's disease, while recent studies have identified potential protective variants. This, and the fact that LRRK2 mutation carriers develop symptoms and brain pathology almost indistinguishable from idiopathic Parkinson's disease, has led to enormous interest in this protein. LRRK2 has been implicated in a range of cellular events, but key among them is canonical Wnt signalling, which results in increased levels of transcriptionally active β-catenin. This pathway is critical for the development and survival of the midbrain dopaminergic neurones typically lost in Parkinson's disease.

Methods: Here we use Lrrk2 knockout mice and fibroblasts to investigate the effect of loss of Lrrk2 on canonical Wnt signalling in vitro and in vivo. Micro-computed tomography was used to study predicted tibial strength, while pulldown assays were employed to measure brain β-catenin levels. A combination of luciferase assays, immunofluorescence and co-immunoprecipitation were performed to measure canonical Wnt activity and investigate the relationship between LRRK2 and β-catenin. TOPflash assays are also used to study the effects of LRRK2 kinase inhibition and pathogenic and protective LRRK2 mutations on Wnt signalling. Data were tested by Analysis of Variance.

Results: Loss of Lrrk2 causes a dose-dependent increase in the levels of transcriptionally active β-catenin in the brain, and alters tibial bone architecture, decreasing the predicted risk of fracture. Lrrk2 knockout cells display increased TOPflash and Axin2 promoter activities, both basally and following Wnt activation. Consistently, over-expressed LRRK2 was found to bind β-catenin and repress TOPflash activation. Some pathogenic LRRK2 mutations and risk variants further suppressed TOPflash, whereas the protective R1398H variant increased Wnt signalling activity. LRRK2 kinase inhibitors affected canonical Wnt signalling differently due to off-targeting; however, specific LRRK2 inhibition reduced canonical Wnt signalling similarly to pathogenic mutations.

Conclusions: Loss of LRRK2 causes increased canonical Wnt activity in vitro and in vivo. In agreement, over-expressed LRRK2 binds and represses β-catenin, suggesting LRRK2 may act as part of the β-catenin destruction complex. Since some pathogenic LRRK2 mutations enhance this effect while the protective R1398H variant relieves it, our data strengthen the notion that decreased canonical Wnt activity is central to Parkinson's disease pathogenesis.

Keywords: LRRK2; Osteoporosis; Parkinson’s disease; Wnt signaling; β-catenin.

Fig. 1

Lrrk2 knockout mice display increased tibial cortical bone strength. Tibiae from 7 female Lrrk2 knockout (KO) and 7 age and sex-matched wild-type (WT) mice displayed no differences in a gross morphology or b trabecular bone architecture. However, micro-computed tomographic analysis of tibial cortical bone revealed c increased second moment of area around the minor axis, I_min; d increased second moment of area around major axis, I_max; and e increased predicted resistance to torsion, J. f Shows a graphical heatmap displaying statistical differences along the tibia between wild-type and knockout mice, using 1-way ANOVA followed by Tukey-Kramer post-hoc analysis (blue = n/s, yellow, p < 0.05, green p < 0.01, red, p < 0.001). Note that in addition to the parameters I_min, I_max and J, other bone parameters were studied, all of which showed differences (see main text for explanation and Additional file 3: Figure S2 and Additional file 4: Figure S3 for raw data). These altered parameters indicate increased bone strength in Lrrk2 null animals

Fig. 2

Lrrk2 knockout cells display increased canonical Wnt activity. a-d Wild-type and Lrrk2 knockout cells were co-transfected with TOPflash or FOPflash and TK-renilla. a In the absence of canonical Wnt pathway activation, TOPflash activity 24-h post-transfection was significantly higher in Lrrk2 knockout cells than wild-type controls (n = 36; T-test, p < 0.001). No difference was detected in FOPflash values (n = 30). b Overnight transfection followed by 6-h treatment with 50 ng/ml recombinant Wnt3a elicited a further increase in TOPflash activity in Lrrk2 knockout cells relative to wild-type (1-way ANOVA: n = 9; F = 19.143, p < 0.001; Bonferroni post-hoc analysis: p < 0.001 versus all other conditions). No significant changes in FOPflash values were detected (n = 6, F = 0.40, p = 0.989). c Overnight transfection followed by incubation for 6 h in the presence of 30 mM NaCl or LiCl. 1-way ANOVA (n = 9; F = 93.414, p < 0.001) followed by Bonferroni post-hoc analysis revealed increased LiCl-driven TOPflash activity in Lrrk2 knockout cells (p < 0.001 relative to all other conditions). No significant changes in FOPflash values were detected (n = 6; F = 2.391, p = 0.1). d Co-transfection with FLAG-β-catenin or empty vector revealed a marked activation of TOPflash by β-catenin in Lrrk2 knockout cells relative to wild-type cells (n = 9; 1-way ANOVA (F = 128.490, p < 0.001; Bonferroni post-hoc analysis, p < 0.001 versus all other conditions). No significant changes in FOPflash values were detected (n = 9, F = 0.165, p = 0.919)

Fig. 3

Lrrk2 knockout cells display increased Axin2 promoter activity. Wild-type and Lrrk2 MEFs were co-transfected with a reporter plasmid containing the murine Axin2 promoter upstream of luciferase and TK-renilla. The relative luciferase activity of resultant cell lysates was measured 24 h post-transfection. a In the absence of canonical Wnt pathway activation, Axin2 promoter activity was significantly higher in Lrrk2 knock-out cells than wild-type controls (n = 18; T-test, p < 0.001). b Co-transfection with FLAG-β-catenin or empty vector revealed a marked activation of the Axin2 reporter by β-catenin in Lrrk2 knockout cells (n = 9; 1-way ANOVA (F = 8.037, p < 0.001) followed by Bonferroni post-hoc analysis, p < 0.01 versus all other conditions)

Fig. 4

Lrrk2 knockout mice display elevated β-catenin levels in the brain. Whole brain lysates from aged-matched Lrrk2 knockout (KO) and wild-type (WT) male mice were resolved by SDS-PAGE and blotted, as indicated, for Lrrk2, β-catenin and β-actin as a loading control. a Shows representative images for 3 wild-type and 3 knockout mice. b Mean β-catenin/β-actin ratios for ten wild-type and 11 knockout mice, ± the standard error of the mean. These calculations revealed significantly increased β-catenin levels in Lrrk2 knockout mouse brains (t-test, p = 0.015). c Graphical illustration of ECT assays. Immobilised E-cadherin cytosolic tail protein (green) bound to beads can be used to affinity purify free β-catenin (red) from complexed β-catenin in lysates. d, e ECT assays of lysates from male and female Lrrk2 knockout and heterozygous brains reveal increased free β-catenin in knockout brains with intermediate levels in heterozygotes. f Analysis of all ECT data by 2-way ANOVA followed by Bonferroni post-hoc testing reveals significant effects of Lrrk2 deficiency on free β-catenin levels in the brain

Fig. 5

Effect of LRRK2 kinase inhibition on canonical Wnt signalling. a HEK293 cells transfected with TOPflash and FLAG-DVL1 were treated with DMSO or LRRK2-in-1, TAE684, or CZC-25146 for 20 h at the indicated concentrations. Resultant luciferase values are expressed relative to DMSO controls. b-d The effect of kinase inhibion with these compounds was further analysed by comparing effects in wild-type and Lrrk2 knockout (KO) cells. b At concentrations used LRRK2 phosphorylation at serine-935 is reduced. c Using conditions identical to panel A, the three compounds also display differing effects in wild-type and Lrrk2 knockout cells. d The values produced for each inhibitor in panel C were expressed as a ratio of wild-type cells over knockout cells, indicating that LRRK2 inhibition weakens DVL1-driven canonical Wnt signalling. 1-way ANOVA reveals a significant effect of treatment (n = 12, F = 8.628, p < 0.001), with 2-sided Dunnett’s post-hoc analysis indicating significant differences between all treatments relative to control (LRRK2-in-1, p < 0.05; TAE684, p < 0.05; 1 μM CZC-25146, p < 0.01; 5 μM CZC-25146, p < 0.001)

Fig. 6

Effect of LRRK2 mutations on basal Wnt signalling. Lrrk2 knockout (KO) cells were transfected with TOPflash or FOPflash plus wild-type LRRK2 or the indicated LRRK2 mutant. a 1-way ANOVA (n = 9-15, F = 3.296, p < 0.05) followed by 2-sided Dunnett’s post-hoc analysis indicate that pathogenic LRRK2 mutations weaken canonical Wnt signalling relative to wild-type LRRK2 (RG, p < 0.05; YC, p < 0.05; GS, p < 0.05). RC is decreased relative to wild-type but does not reach significance. b 1-way ANOVA (n = 9, F = 13.388, p < 0.001) followed by 2-sided Dunnett’s post-hoc analysis indicate the pathogenic LRRK2 mutation GR weakens basal canonical Wnt signalling relative to wild-type LRRK2 (p < 0.01), whilst the protective RH mutation enhances canonical Wnt activity (p < 0.01). Note that the pathogenic mutations NH and RP decreased relative to wild-type but neither reaches significance. Mutations used: RC = R1441C, RG = R1441G, YC = Y1699C, GS = G2019S, NH = N1437H, RP = R1628P, GR = G2385R; RH = R1398H