PAK6 rescues pathogenic LRRK2-mediated ciliogenesis and centrosomal cohesion defects in a mutation-specific manner

Abstract

P21 activated kinase 6 (PAK6) is a serine-threonine kinase with physiological expression enriched in the brain and overexpressed in a number of human tumors. While the role of PAK6 in cancer cells has been extensively investigated, the physiological function of the kinase in the context of brain cells is poorly understood. Our previous work uncovered a link between PAK6 and the Parkinson's disease (PD)-associated kinase LRRK2, with PAK6 controlling LRRK2 activity and subcellular localization via phosphorylation of 14-3-3 proteins. Here, to gain more insights into PAK6 physiological function, we performed protein-protein interaction arrays and identified a subgroup of PAK6 binders related to ciliogenesis. We confirmed that endogenous PAK6 localizes at both the centrosome and the cilium, and positively regulates ciliogenesis not only in tumor cells but also in neurons and astrocytes. Notably, PAK6 rescues ciliogenesis and centrosomal cohesion defects associated with the G2019S but not the R1441C LRRK2 PD mutation. Since PAK6 binds LRRK2 via its GTPase/Roc-COR domain and the R1441C mutation is located in the Roc domain, we used microscale thermophoresis and AlphaFold2-based computational analysis to demonstrate that PD mutations in LRRK2 affecting the Roc-COR structure substantially decrease PAK6 affinity, providing a rationale for the differential protective effect of PAK6 toward the distinct forms of mutant LRRK2. Altogether, our study discloses a novel role of PAK6 in ciliogenesis and points to PAK6 as the first LRRK2 modifier with PD mutation-specificity.

Fig. 1. PAK6 interacts with ciliary proteins.

a Distribution of PAK6 candidate interactors according to their Z-score retrieved from a Human Proteome Microarray probed with recombinant full-length human PAK6. b A GO:BP analysis using gProfiler g:GOSt (https://biit.cs.ut.ee/gprofiler/gost) was performed for PAK6 candidate interactors with Z score >2.5 (left) and for PAK6 interactors annotated in PPI web-based tools PINOT, HIPPIE and MIST (PHM) (right). GO:BP terms with 2000 (array) and 1000 (PHM) term size were grouped into semantic categories. c Venn diagrams showing overlaps between the primary cilium proteome (GO:0005929, 640 genes) and the experimental (array) PAK6 interactome (left) or the literature-based (PHM) PAK6 interactome (right). d Protein network of overlapping PAK6 interactors with the primary cilium proteome (c) (including PAK6) obtained with STRING (https://string-db.org/cgi/input?sessionId=b1S4T5BW27rz&input_page_show_search=on); number of nodes: 11, number of edges: 11, average node degree: 2, average local clustering coefficient: 0.591, expected number of edges: 3, PPI enrichment P-value: 0.000502. Blue nodes are ciliary proteins present in the experimental PAK6 interactome (array) and grey nodes are those found in the literature-based PAK6 interactome. The interaction between LRRK2 and PAK6 identified in this study (blue) has been inserted manually.

Fig. 2. PAK6 is localized at centrosomes and primary cilium and regulates ciliogenesis.

a Localization of Pak6 at centrosome. Staining of mouse embryonic fibroblasts (MEFs) derived from either wild type (MEFs WT) mice or Pak6 null (MEFs KO) mice with anti-γ-tubulin (mouse) and anti-PAK6 (rabbit) antibodies. Insets show localization of Pak6 at centrosome. Scale bar = 10 μm. b Localization of PAK6 at centrosome in MCF7 cells as evidenced by immunostaining with anti-γ-tubulin (mouse) and anti-PAK6 (rabbit) antibodies. c Co-sedimentation of PAK6 with γ-tubulin in sucrose gradient fractions. Cell lysates from MCF7 cells were subjected to 40–60% sucrose density gradient ultra-centrifugation and fractionated. The resulting fractions were resolved by SDS-PAGE and analyzed by immunoblot with anti-PAK6 and anti-γ-tubulin antibodies. d PAK6 localizes to the basal body of primary cilium. Immuno-staining of PAK6 (green) localizes it to the basal body of the primary cilium in HEK293T cells. The primary cilium axoneme and basal body were respectively identified by staining with Arl13b (red) and anti-pericentrin (red). Scale bar = 2 μm. e PAK6 regulates ciliogenesis in HEK293T cells. Knock-down of PAK6 expression by PAK6 specific lentiviral-mediated shRNA downregulates ciliogenesis as shown by a decrease in the percentage of ciliated cells and a decrease in ciliary length as compared to scrambled shRNA. Cilia were stained with antibodies against Arl13b. Violin plots represent the percentage of cells per field with a primary cilium (top; N = 3 experiments: n(scramble)=16, n(Pak6 shRNA)=15 fields) and the length of cilia (bottom; n(scramble)=191, n(Pak6 shRNA)=200 cilia). Unpaired t-test, ****P < 0.0001. Scale bar = 10 μm. f Pak6 regulates ciliogenesis in MEF cells. Pak6 KO MEF cells also exhibit a deficit in the percentage of ciliated cells and ciliary length. Cilia were stained with antibodies against Arl13b. Violin plots represent the percentage of cells per field with a primary cilium (top; N = 3 experiments: n(WT) = 16, n(Pak6 15)= fields) and the length of cilia (bottom; n(WT) = 142, n(Pak6 KO) = 150 cilia). Unpaired t-test, ****P < 0.0001. Scale bar = 10 μm.

Fig. 3. PAK6 promotes ciliogenesis in SH-SY5Y cells.

a Semantic categories from GO:BP (electronic annotation) PAK6 co-expression analysis using Search-Based Exploration of Expression Compendium (SEEK) [48]. b Western blot analysis of PAK6 and phospho-PAK6 in naïve, stable LV-PAK6 and stable LV-PAK6 shRNA SH-SY5Y cells. c Example of SH-SY5Y naïve, stable LV-PAK6 and stable LV-PAK6 shRNA SH-SY5Y cells. Cilia were stained anti-Arl13b (green) and nuclei with DAPI. Arrows point to primary cilia. Scale bar 10 μm. d Quantification of (c). Violin plots represent the percentage of cells per field with a primary cilium. N = 3 experiments; n(naïve)=37, n(LV-PAK6 OE) = 36, n(shRNA PAK6) = 31 fields analyzed; one-way ANOVA with Tukey’s post-test (****P < 0.0001). Colors represent different biological replicates. e Example of SH-SY5Y transfected with 2xMyc-GFP, 2xMyc-PAK6 WT and 2xMyc-PAK6 K436M (KM). Cilia were stained anti-Arl13b (red), and GFP or PAK6 with anti-Myc (green) antibodies, and nuclei with DAPI. Arrows point to primary cilia in transfected cells. Scale bar 20 μm. Colors represent different biological replicates. f Quantification of (e). Violin plots represent the percentage of transfected cells per field with a primary cilium. N = 3 experiments; n(GFP) = 30, n(PAK6 WT) = 39, n(PAK6 KM) = 33 fields analyzed; one-way ANOVA with Tukey’s post-test (****P < 0.0001; ***P < 0.001). Colors represent different biological replicates.

Fig. 4. PAK6 promotes ciliogenesis in primary neurons and astrocytes.

a Example of primary cortical neurons isolated form Pak5/Pak6 dKO mice. Immunocytochemistry was performed with antibodies anti-AC3 (neuronal cilia marker, green), anti-MAP2 (neuronal marker, red) and with DAPI (blue). Arrows point to primary cilia (scale bar 10 µm). Scale bar of zoomed images is 1 µm. b Quantification of (a). Violin plots represent the percentage of cells per field with a primary cilium (left) and the cilia length (right). N = 3 experiments; n(WT) = 82, n(2KO) = 83; unpaired t-test (****P < 0.0001; ns P > 0.05). Colors represent different biological replicates. c Example of primary striatal astrocytes isolated form Pak5/Pak6 dKO mice. Immunocytochemistry was performed with antibodies anti-Arl13b (cilia, green), phalloidin (F-actin, magenta) and with DAPI (blue). Arrows point to primary cilia (scale bar 10 µm). Scale bar of zoomed images is 1 µm. d Quantification of (c). Violin plots represent the percentage of cells per field with a primary cilium (left) and the cilia length (right). N = 3 experiments; n(WT) = 64, n(2KO) = 52; unpaired t-test (****P < 0.0001; ns P > 0.05). Colors represent different biological replicates. e Example of Pak5/Pak6 dKO primary astrocytes transfected with 3xFlag-PAK6 WT or 3xFlag-GUS control. Immunocytochemistry was performed with antibodies anti-Arl13b (cilia, red), Flag (green) and with DAPI (blue). Arrows point to primary cilia (zoomed on the right). Scale bar 10 µm. f Quantification of (e). Percentage of transfected cells per field with a primary cilium. N = 3 experiments; n(GUS) = 30, n(PAK6 WT) = 28; unpaired t-test (***P < 0.001). Colors represent different biological replicates.

Fig. 5. PAK6 rescues G2019S but not R1441C LRRK2-associated ciliogenesis defects independently from its kinase activity.

a Example of primary astrocytes isolated from G2019S LRRK2 KI mice transfected with control 3xFlag-GUS, 3xFlag-PAK6 WT or 3xFlag-PAK6-KM and stained with antibodies against Flag (green), Arl13b (red) and with DAPI (blue). Scale bar 10 µm. Zoomed insets show representative ciliated or non-ciliated cells. b Quantification of (a). Violin plots represent the percentage of transfected cells per field with a primary cilium. N = 3 experiments; n(GUS) = 30, n(PAK6 WT) = 30, n(PAK6 KM) = 30 fields analyzed; one-way ANOVA with Tukey’s post-test (****P < 0.0001). Colors represent different biological replicates. c Example of primary astrocytes isolated from R1441C KI mice transfected as in (a). Scale bar 10 µm. Zoomed insets show representative ciliated or non-ciliated cells. d. Quantification of (c). N = 3 experiments; n(GUS) = 29, n(PAK6 WT) = 29, n(PAK6 KM) = 28 fields analyzed; one-way ANOVA with Tukey’s post-test (ns P > 0.05). Colors represent different biological replicates.

Fig. 6. PAK6 rescues G2019S but not R1441C LRRK2-associated centrosomal cohesion defects independently from its kinase activity.

a Example of A549 cells co-transfected with pCMV (EV) and flag-tagged G2019S LRRK2 or R1441C LRRK2 and treated ± MLi-2 (200 nM, 2 h), or co-transfected with flag-tagged LRRK2 and myc-tagged PAK6 WT or PAK6-KM as indicated before immunocytochemistry with antibodies against flag (green), pericentrin (red) and with DAPI (blue). Arrows point to centrosomes in transfected cells. Scale bar, 10 µm. b Quantification of the percentage of non-transfected cells (ctrl), or cells co-transfected with G2019S LRRK2 and pCMV (EV), PAK6 WT or PAK6-KM and -/+ MLi-2 treatment as indicated where duplicated centrosomes are >2.5 µm apart (split centrosomes). Bars represent mean ± s.e.m. (n = 3 experiments; *p < 0.05; **p < 0.01; ****p < 0.0001). (c) As in (b), but cells co-transfected with R1441C LRRK2. Bars represent mean ± s.e.m. (n = 3 experiments; ****p < 0.0001).

Fig. 7. PAK6 affinity to ROC-COR carrying R1441C and Y1699C mutations is dramatically decreased.

a Microscale thermophoresis of fluorescently labelled PAK6 full-length (100 nM) against MBP-fused LRRK2 Roc-COR domain was measured for R1441C (3.4 nM to 110 µM) and Y1699C (3.6 nM to 118 µM) mutations and compared with WT Roc-COR from identical experimental setting published in [35]. Upper panel: the normalized fluorescence intensities of all MST traces are plotted against different concentrations of MBP-LRRK2-Roc-COR. Lower panel: The changes of relative fluorescence during thermophoresis. The calculated KD values of PAK6 towards LRRK2-Roc-COR WT, R1441C and Y1699C are 10.2 µM ± 2.6, 51.8 µM ± 14.4 and 38.2 µM ± 11.1 respectively. Error bars show the S.D. of three measurements. The residuals between the data and the fit are shown at the bottom of the graph. b Schematic of GST-pull down (top) and alignment of PAK5 and PAK6 CRIB domains (bottom), which show 65% of amino acid identity. c Pull downs of Roc-COR and PAK5/PAK6 CRIB domains. Left panel: Coomassie gel of inputs MBP-Roc-COR purified with Amylose resin and GST, GST-CRIB-PAK5 or GST-CRIB-PAK6 purified with GSH resin. Right panel: pulldowns of GST proteins after incubation with amylose-beads bound Roc-COR in the presence of GDP or non-hydrolysable GppNHp. Roc-COR pulls down GST-CRIB-PAK6, very little GST-CRIB-PAK5 and no GST alone. d Microscale thermophoresis of fluorescently labelled Roc-COR against GST, GST-CRIB-PAK5 and GST-CRIB-PAK6. Upper panel: the normalized fluorescence intensities of all MST traces are plotted against different concentrations of GST proteins. Lower panel: The changes of relative fluorescence during thermophoresis. The calculated K_D values of Roc-COR towards GST, GST-CRIB-PAK6 and GST-CRIB-PAK5 are 446 µM ± 305, 12 µM ± 7 and 112 µM ± 32 respectively. Error bars show the S.D. of three measurements. The residuals between the data and the fit are shown at the bottom of the graph. e Left. Model obtained out of Alphafold2 modelling for LRRK2 and PAK6, colored as in [50]. Only the CRIB domain of PAK6 is shown. Right. Zoomed view on the CRIB domain of PAK6, showing interactions with ROC (green) and COR (cyan) domains of LRRK2. R1441 is highlighted in red, in close proximity with the predicted CRIB binding site. Interaction of F18 and H20 with Y1699 is highlighted.