High content screening and proteomic analysis identify a kinase inhibitor that rescues pathological phenotypes in a patient-derived model of Parkinson's disease

Abstract

Combining high throughput screening approaches with induced pluripotent stem cell (iPSC)-based disease modeling represents a promising unbiased strategy to identify therapies for neurodegenerative disorders. Here we applied high content imaging on iPSC-derived neurons from patients with familial Parkinson's disease bearing the G209A (p.A53T) α-synuclein (αSyn) mutation and launched a screening campaign on a small kinase inhibitor library. We thus identified the multi-kinase inhibitor BX795 that at a single dose effectively restores disease-associated neurodegenerative phenotypes. Proteomics profiling mapped the molecular pathways underlying the protective effects of BX795, comprising a cohort of 118 protein-mediators of the core biological processes of RNA metabolism, protein synthesis, modification and clearance, and stress response, all linked to the mTORC1 signaling hub. In agreement, expression of human p.A53T-αSyn in neuronal cells affected key components of the mTORC1 pathway resulting in aberrant protein synthesis that was restored in the presence of BX795 with concurrent facilitation of autophagy. Taken together, we have identified a promising small molecule with neuroprotective actions as candidate therapeutic for PD and other protein conformational disorders.

Fig. 1. Identification of BX795 by high content screening of a kinase inhibitor library.

a Directed differentiation of Pax6+ (green)/Nestin+ (red) neural precursor cells (NPCs; DIV 0, left) into TUJ1+ (red)/ TH+ (green) neurons (DIV 21, right). The differentiation protocol and timeline of analysis are shown in the drawing in the middle. FG2 and FGF8, fibroblast growth factors 2 and 8; SHH Sonic Hedgehog, AA ascorbic acid, Scale bar represents BDNF brain-derived neurotrophic factor, GDNF glial cell-derived neurotrophic factor (GDNF), cAMP cyclic AMP. Scale bars, 50 μm. b Scatter plot showing the ratio of TH versus TUJ1 fluorescence intensity in duplicate upon treatment with 273 small molecule kinase inhibitors. The dots inside the green square correspond to the 4 hit compounds showing significant increase of TH versus TUJ1 fluorescence ratio as compared to the DMSO controls (blue dots). The red arrow indicates BX795. c Representative images of patient-derived p.A53T-neurons immunolabelled for TH in 384-well plates. Upper micrograph shows control DMSO-treated cells while lower micrograph represents BX795-treated cells. Scale bar represents 150 μm. d Tests of the four hit compounds in a dose-response format. Data are presented as mean ± SEM. (one-way ANOVA, *P < 0.05, n = 3 independent experiments).

Fig. 2. Rescue of neuropathological features in patient-derived p.A53T neurons by BX795.

a BX795 has a positive effect on neurite length of p.A53T-neurons. Representative confocal images of healthy control (ctl) and p.A53T-neurons immunostained for TH and quantification of total neurite length of TH+ cells. Data represent mean ± SEM. (Comparisons by ANOVA with Tukey correction *P < 0.05, **P < 0.01, n = 4 independent experiments with at least 50 cells analyzed in each experiment). Scale bar, 50 μm. b BX795 alleviates axonal neuropathology in p.A53T-neurons. Higher magnification at the right (upper, DMSO-treated cells; lower, BX795-treated cells) shows neurites with swollen varicosities or fragmented processes (arrows). Scale bar, 30 μm. Quantification of axonal degeneration is estimated in the accompanying graph by measuring the ratio of TUJ1+ spots over the total TUJ1+ area in untreated (p.A53T) or BX795-treated p.A53T-neurons. Data represent mean ± SEM.(Comparisons by ANOVA with Tukey correction, *P < 0.05, **P < 0.01, n = 20 randomly selected fields for each condition; data was from three independent experiments). c BX795 reduces protein aggregates in p.A53T-neurons. Representative confocal images showing protein aggregates in p.A53T TUJ1+ neurons (Scale bar, 10μm) and quantification in untreated or BX795-treated TUJ1+ cells (Data was from three independent experiments. Mann–Whitney test; n = at least 30 cells per group; ****P < 0.0001). d Detection and quantification of p(Ser129)αSyn by Western blot in control and p.A53T neurons in the absence or presence of BX795, as indicated; Actin shows equal protein loading. Data represent mean ± SEM (t test, *P < 0.05, n = 4 independent experiments).

Fig. 3. BX795 reduces protein aggregates in a gene-edited p.A53T line of mature human iPSC-derived TH neurons.

a Representative confocal images of wild-type (ctl) and isogenic p.A53T iCell Dopa neurons immunolabelled for Nuclei, TUJ1, MAP2 and TH. Scale bar, 30 μm. b Representative confocal images of wild-type (ctl) and isogenic p.A53T iCellDopa neurons showing immunostaining for tyrosine hydroxylase (TH green) and protein aggregates (red). p.A53T cells were treated or not with BX795, as indicated. Scale bar, 5 μm. c Quantification of aggregates in TH+ neurons. Data represent mean ± SEM. (Comparisons by ANOVA with Tukey correction, ****P < 0.0001, n = at least 50 randomly selected TH+ cells for each condition).

Fig. 4. Bioinformatics analysis of dysregulated proteins in p.A53T-neurons that are restored by BX795.

a Hierarchical clustering of 118 upregulated proteins in patient-derived p.A53T-neurons that are restored upon treatment with BX795 (one-way ANOVA analysis). Columns in the different groups (control, p.A53T-neurons and p.A53T-neurons treated with BX795) correspond to individual samples tested and rows represent single proteins (blue, low expression; red, high expression; n = 3 for control and p.A53T; n = 2 for p.A53T + BX795). b GO enrichment analysis for biological processes, molecular function and cellular compartments was performed using DAVID software (p < 0.01). c Pathway analysis using Reactome software (p < 0.01).

Fig. 5. Protein network of pathways and processes restored by BX795 treatment.

a Heatmaps illustrating specific proteins upregulated in p.A53T-neurons that are involved in RNA metabolism, protein synthesis, protein modification and transport and response to stress, which are restored after BX795 treatment. High expression is in red and low expression is in blue. b STRING-analysis representation of the protein-protein interaction network of the 118 upregulated proteins in p.A53T-neurons that are restored by BX795. Each circular node depicts one protein and the different colors represent the different pathways/processes as indicated. Connecting lines represent protein-protein associations and line intensity represents the confidence score of a functional association.

Fig. 6. Restoration of disease-associated proteins by BX795 in p.A53T-neurons.

a Heatmap of proteins associated with neurodegeneration that are restored after BX795 treatment. High expression is in red and low expression is in blue. b Disease-associated proteins that are modified by BX795 are either known or associated genetic risk factors for neurodegenerative diseases as revealed by human genetic studies. c STRING network analysis of the neurodegeneration-associated proteins restored by BX795 in p.A53T-neurons and their interaction with αSyn. Each αSyn interactor is shown as a colored circle and connecting lines between proteins represent protein-protein associations. The intensity of lines represents the confidence score of a functional association.

Fig. 7. BX795 affects the mTORC1 signaling pathway in p.A53T patient-derived and iCell Dopa neurons to attenuate protein synthesis.

a Western blot showing increased levels of p-RPS6 in p.A53T-patient iPSC-derived neurons and a notable reduction in the presence of BX795. GADPH shows equal protein. b Quantification of p-RPS6 levels in p.A53T-patient iPS-derived neurons. Data represent mean ± SEM (Comparisons by ANOVA with Bonferroni’s multiple comparisons test. **P < 0.01, n = 3 independent experiments). c Representative confocal images of control (ctl) and isogenic gene-edited p.A53T iCellDopa neurons, either non-treated or treated with BX795. Cells were immunolabeled for phosphorylated RPS6 (green) andmicrotubule associated protein 2 (MAP2; red)(upper panel) and labeled for total protein synthesis (protein label, green)(lower panel). Nuclei are seen with Hoechst dye (blue). Scale bar, 30 μm. d BX795 reduces phosphorylated RPS6 levels and reduces total protein synthesis in p.A53T-neurons. Quantification of fluorescence intensity in control, untreated p.A53T or BX795-treated p.A53T neurons. Data represent mean ± SEM (Comparisons by ANOVA with Tukey correction, ***P < 0.001 ****P < 0.0001, n = 100 randomly selected cells for each condition).

Fig. 8. BX795 restores neurodegenerative phenotypes via the mTORC1 signaling pathway.

a Representative images of SH-SY5Y cells stably transduced to express DsRed only or DsRed and human pA53T-αSyn. After neuronal differentiation, cells were immunolabled for αSyn (SNCA), TUJ1 and pRPS6. b Western blot showing that the presence of mutant SNCA in differentiated p.A53T-transduced SH-SY5Y cells, results in an increase in the levels p-mTOR and p-RPS6. Actin shows equal protein loading. Data represent mean ± SEM (t test, *P < 0.05, n = 3 independent experiments). c Western blot showing an acute reduction in the levels of p-mTOR and p-RPS6 in the above stably transduced and differentiated SH-SY5Y cells, in the presence of BX795. Actin shows equal protein loading. Data represent mean ± SEM (Comparisons by ANOVA with Tukey correction, *P < 0.05, ****P < 0.0001, n = 3 independent experiments). d BX795 inhibits the kinase activity of pure recombinant p70 S6K (25 ng/assay) in a dosedependent manner. Rapamycin (Rapa) was used as a positive control. Data represent mean ± SEM (Comparisons by ANOVA with Tukey correction, *P < 0.05, **P < 0.01, n = 3). e Western blot showing the effect of BX795 treatment alone and in combination with mTOR activator MHY-1485 on the p(Ser129)-αSyn levels in SH-SY5Y cells stably transduced to express DsRed only or DsRed and human pA53T-αSyn. f Representative confocal images show protein aggregates in control (ctl) and patient p.A53T neurons, either non-treated (DMSO) or treated with BX795 alone or in combination with the mTOR activator MHY-1485.

Fig. 9. BX795 facilitates autophagy in an inducible SH-SY5Y cell line expressing human p.A53T-αSyn.

a Schema illustrating that cytosolic LC3 is cleaved to yield LC3-I, which is subsequently conjugated to phosphatidylethanolamine (PE) to form membrane-bound LC3-II (green circles). Pre-autophagosomal structures engulfing protein cargo and organelles destined for degradation close to form double membrane spherical autophagosomes. These fuse with lysosomes to yield autolysosomes and their contents are degraded. Bafilomycin blocks autophagic flux by inhibiting autophagosome-lysosome fusion, which results in accumulation of LC3-II+ autophagosomes. b Representative immunoblot showing steady-state levels of LC3-II and p62 in lysates of inducible SH-SY5Y cells expressing the human p.A53T-αSyn (-Dox) and quantification relative to actin. Data represent mean ± SEM, t est, *P < 0.05, n = 3 independent experiments. c Representative confocal images of individual p.A53T SH-SY5Y cells (-Dox) transfected with GFP-LC3 that were treated or not with bafilomycin A1 in the absence or presence of BX795. d Quantification of GFP-LC3 puncta per cell. Comparisons by ANOVA with Tukey correction. *P < 0.05, n = 72 cells (control DMSO), n = 79 cells (BX795), n = 67 cells (Bafilomycin A1), n = 68 cells BX795 + Bafilomycin A1. Data are representative of three independent experiments). e Assessment of autophagic flux using mCherry-GFP-LC3 color change between autophagosomes and autolysosomes. Autophagic flux is induced when the GFP:mCherry ratio is reduced. f, g. Representative confocal images of individual cells [inducible SH-SY5Y cell line expressing p.A53T-αSyn (-Dox)] transfected with GFP-mCherry-p62 that were treated with DMSO (control) or BX795 and quantification of the ratio of GFP +/mCherry+ puncta (t test, n = 60 (control DMSO), n = 53 (BX795) **P < 0.01 Data are representative of three independent experiments). h Representative immunoblot showing steady-state levels of p62 in cells [inducible SH-SY5Y cell line expressing p.A53T-αSyn (-Dox)] treated or not with BX795, and quantification relative to actin. Data represent mean ± SEM, t test, n = 3 independent experiments. i Representative electron micrographs showing control and patient-derived p.A53T cells. Multilamellar autophagic vacuoles (arrows) are depicted in p.A53T cells; in this case they seem to be engulfed by an additional membranous structure. Scale bar 500 nm. j Representative confocal images showing autophagosome accumulation (brightly labeled green spheroids) in patient-derived p.A53T neurons as compared to either control (ctl) or p.A53T cells treated with BX795. Scale bar, 30 μm.