Elevated hexosylceramides in Parkinson's disease cause gene upregulations in neurons mimicking responses to pathogens

Abstract

Parkinson's Disease (PD) is driven by pathological aggregates of alpha-synuclein (αSyn), whose formation is facilitated by impaired glycosphingolipid metabolism via acidic glucocerebrosidase (GCase). We investigated glucosylceramide (GlcCer) accumulation in human, mouse, and cellular PD models. Lipidomic analyses revealed elevated plasma GlcCer, especially GlcCer24:1, and a shift in phosphatidylcholine (PC) species in PD patients. PD patient skin fibroblasts accumulated more GlcCer under lysosomal stress. GlcCer and sulfatides (SHexCer) were increased in Pink1^-/-SNCA^A53T PD mouse brains, and HT22 neurons exposed to preformed αSyn fibrils accumulated GlcCer and ceramides. GlcCer24:1 enhanced fibril toxicity, but had no direct or indirect effect on G-protein coupled receptors. RNAseq of GlcCer24:1-treated dorsal root ganglion neurons showed upregulation of glycolipid response genes, similar to pathogen-related signaling. These data indicate extracellular GlcCer is elevated in PD and triggers innate immune responses in sensory neurons.

Fig. 1. Plasma lipidomic analyses in PD patients versus healthy controls (HC).

Patients (n = 16 f, 34 m) and controls (n = 25 f, 25 m) were age-matched <60-70+ years old at the time of blood sampling. Demographic details are shown in Supplementary Table 1. A Volcano plots show the Log2(fold change) on the x-axis versus the negative Log10 of the t-test P-value on the Y-axis. Increased lipids in PD are in red, reduced in blue. Lipids with statistically significant P-value but below the threshold for fold change are in orange. The Volcano plots reveals changes in FA, HexCer and PC which are detailed in (B–D). Please note that hexosylceramides (HexCer) represent GlcCer and GalCer. Hex2Cer mostly LacCer. B Fatty acids (FA) female and male PD patients and controls. C Hexosylceramides obtained by targeted and untargeted (TOF) lipidomic analyses. D Phospahtidylcholines (PC). The line is the mean, the whiskers show the SD. Each scatter is a subject. Data were submitted to 2-way ANOVA and subsequent posthoc analysis for each lipid species using a adjustment of alpha according to Sidak. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. CAR carnitines, CER ceramides, CE cholesterol ester, DG diglycerides, FA fatty acids, HexCer hexosylceramides, LPC lysophosphatidylcholines, LPE lysophosphatidylethanolamines, LPG lysophosphatidylglycerols, LPI lysophosphatidylinositols, PC phosphatidylcholines, PE phosphatidylethanolamines, PD phosphatidylglycerols, PI phosphatidylinositols, SM sphingomyelins, ST sterols, TG triglycerides, UbiQ ubiquitin, –O ether bound.

Fig. 2. Ceramides and hexosylceramides in PD and HC primary fibroblasts upon stimulation with pimozide.

Primary fibroblast cultures were obtained from 3 mm skin biopsies of the lower leg from n = 12 (9 f, 3 m) HC and n = 13 (4 f, 9 m) PD patients. Sphingolipids were analyzed by targeted LC-MS/MS from 2.5 ×105 primary human fibroblasts per sample, and concentrations in ng/ml were auto-scaled to have an common mean and variance of 1 (Z-score). Sub-confluent cultures were stimulated with 12.5 µM pimozide (PIM 12.5 µM = 1/5^th of EC50) or vehicle (1:10000 DMSO) and harvested at 24 h. A Box/scatter plots of sphingolipid z-scores. Pimozide treatment raised Cer 24:1 and GlcCer’s and reduced sphingosine (SPH d18:1) predominantly in PD-PHF. The box shows the interquartile range, the line is the median, whiskers show minimum to maximum, scatters are individual subjects. Statistics: 2-way ANOVA and subsequent t-tests with adjustment of alpha according to Sidak for the between subject factor (4 groups) *P < 0.05, **P < 0.01, ***P < 0.001. B Score plots of a canonical discrimination analyses using sphingolipid concentrations as input. Clusters of vehicle treated PHF are overlapping, but pimozide treated PD-PHF differ from HC-PHF. C Paired analysis of for the most abundant GlcCer 16:0 and GlcCer24:1. PD fibroblasts show a stronger PIM evoked increase. D Violin plots reveal stronger variance of sphingolipid levels in PHF of PD patients than controls. Data as in (A).

Fig. 3. Ceramides and hexosylceramides in PD mouse brain.

A Ceramides and hexosylceramides in brain tissue (cortex, subcortex, midbrain) in Pink1^−/−SNCA^A53T and Sv129FVB wildtype control mice (age 50–60 weeks) analyzed by targeted LC-MS/MS analysis. Each scatter is one mouse. Data were submitted to 2-way ANOVA using the between subject factor “genotype” and the within subject factor “ceramide”, followed by posthoc t-tests with adjustment of alpha according to Sidak for genotype. *P < 0.05, ****P < 0.0001. B Ceramides and hexosylceramides in brain tissue (cortex, subcortex, midbrain) in Pink1^−/−SNCA^A53T and Sv129FVB wildtype control mice (age 50–60 weeks) analyzed by untargeted UHPLC-MS/MS lipidomic screen. Data were analyzed using 2-way ANOVA with the between subject factor “genotype” and the within subject factor “ceramide”, followed by posthoc analysis for each lipid using the false discovery rate for adjustment of alpha. Heatmaps of top 50 regulated lipids are presented in Supplementary Fig S3.

Fig. 4. Preformed alpha synuclein fibrils (αSyn-PFF) increase ceramides in HT22 cells.

A Ceramide and hexosylceramide concentrations in picograms per 100,000 HT22 mouse hippocampal neurons as assessed by targeted LC-MS/MS analysis. Cells were seeded in the absence or presence of PFF at 2 µg/ml or 6 µg/ml in culture flasks and grown for 48 h. Control cells were treated with the respective volume of DMSO or were left untreated (naïve HT22). Each group consisted in 8 replicates. Data were submitted to 2-way ANOVA and subsequent posthoc t-test using an adjustment of alpha according to Sidak for the between subject factor “treatment” (i.e. 4 groups). *P < 0.05, **P < 0.01. High PFF increased all ceramides. B Heatmap of top 50 regulated lipids in HT22 mouse hippocampal neurons as assessed by UHPLC-MS/MS lipidomic analysis. Cells were treated as explained in (A). AUC/IS values were square root transformed and auto-scaled to have a common mean and variance of 1 (z-scores). Lipids were clustered according to Euclidean distance metrics using the Ward method. C Principal Component Analysis of HT22 lipidomic data showing a 3D-score scatter plot of the first three PC. Each scatter shows one sample. The circles are 90% confidence elipsoids.

Fig. 5. Cellular uptake of preformed alpha synuclein fibrils (αSyn-PFF) in HT22 cells.

Immunofluorescent studies show the uptake of AF555-labeled αSyn preformed fibrils in HT22 mouse hippocampal neurons. Pretreatment with GlcCer24:1 (1 µM) had no significant impact on the PFF-AF555 fluorescence uptake in HT22 cells (pink). Wheat germ agglutin WGA-AF488 (yellow) is used as counterstain of plasma membranes (early time point) and endolysosomes (late time point). Hoechst-33342 is used as nuclear stain. Additional zoom-in images are shown in Supplementary Fig. S5.

Fig. 6. GlcCer18:1 and GlcCer24:1 do not activate heterologous expressed G-protein coupled receptors.

A Beta arrestin-based screening of GPCR activity in a heterologous expression model in COS cells upon stimulation with GlcCer 18:1 versus vehicle at 1, 5 and 10 µM. The heatmap shows the mean of four replicates for candidate GPCR selected from an initial screen (330 GPCRs, 8 neg and 4 × 2 pos controls). Details in Supplementary Fig. 1. B As in A but stimulation with GlcCer 24:1 C: Dynamic Mass Redistribution (DMR) analysis of candidate GPCRs stimulated with GlcCer 18:1 or GlcCer 24:1 at 1 µM. DMR measures a ligand-induced shift in resonant wavelength in picometers. Three candidate GPCR were tested with GlcCer 18:1 (upper left) and 14 with GlcCer 24:1. Carbachol activation of muscarinic receptor CHRM1 was used as positive control (right Y-axis). MCHR1 melanin-concentrating hormone receptor 1, FFA3 free fatty acid receptor 3 (GPR41), CXCR7 CXC-type chemokine receptor 7, GPR4 G-protein coupled receptor 4, pH-sensing, GPRC5A orphan GPCR, alias: retinoic acid-induced protein 3 (RAI3), GPR77 C5a anaphylatoxin chemotactic receptor, GPR82 and GPR85 both orphan, P2RY8 P2Y purinoceptor, NPSR1 neuropeptide S receptor 1, GRM5 metabotropic glutamate receptor, GPPR gastrin releasing peptide receptor, GPR123 adhesion GPCR, GPR143 ocular albinism type 1 (OA1), receptor for tyrosine, L-dopa and dopamine, BDKRB2 bradykinin receptor B2, AVPR2 arginine vasopressin receptor 2, ADRA1B alpha-1B adrenoreceptor.

Fig. 7. GlcCer24:1 of primary neurons triggers upregulation of genes involved in GO “response to glycolipid”.

A Volcano plots of regulation of mRNA expression assessed by RNAseq in primary sensory neurons of the dorsal root ganglia from adult mice treated with 1 µM GlcCer24:1 versus vehicle. The x-axis shows the Log2(Fold change) of TMM-normalized counts. The y-axis shows the negative logarithm of the t-test P-value. The data are from n = 4 cultures per condition. Upregulated genes are shown in red, downregulated in blue. B Candidate genes were selected according to the FDR-adjusted P-value and are sorted according to abundance and presented as TMM-normalized reads (trimmed mean of m values). C Candidate genes were submitted to GO and pathway enrichment analysis using DAVID, STRING and Panther consistently showing an enrichment of genes involved in “response to glycolipid”.