Cytoplasmic sphingosine-1-phosphate pathway modulates neuronal autophagy

Abstract

Autophagy is an important homeostatic mechanism that eliminates long-lived proteins, protein aggregates and damaged organelles. Its dysregulation is involved in many neurodegenerative disorders. Autophagy is therefore a promising target for blunting neurodegeneration. We searched for novel autophagic pathways in primary neurons and identified the cytosolic sphingosine-1-phosphate (S1P) pathway as a regulator of neuronal autophagy. S1P, a bioactive lipid generated by sphingosine kinase 1 (SK1) in the cytoplasm, is implicated in cell survival. We found that SK1 enhances flux through autophagy and that S1P-metabolizing enzymes decrease this flux. When autophagy is stimulated, SK1 relocalizes to endosomes/autophagosomes in neurons. Expression of a dominant-negative form of SK1 inhibits autophagosome synthesis. In a neuron model of Huntington's disease, pharmacologically inhibiting S1P-lyase protected neurons from mutant huntingtin-induced neurotoxicity. These results identify the S1P pathway as a novel regulator of neuronal autophagy and provide a new target for developing therapies for neurodegenerative disorders.

Figure 1. SK1 enhances autophagy in primary neurons.

(a) Autophagy was effectively induced in primary cortical neurons by overexpression of untagged SK1. Two neuronal cohorts were transfected with a red marker of morphology, mApple (bottom panel), a marker of autophagy, beclin1-GFP (top panel), and cotransfected either with an empty plasmid (mock; left panel) or untagged SK1 (right panel). Live neurons were analyzed with epifluorescence. Note the changes in beclin1-GFP localization, when SK1 is coexpressed. Arrow depicts puncta in the soma. Also note beclin1-GFP puncta in neurites. Bar, 20 μm. (b) The puncta index was estimated by measuring the standard deviation of beclin1-GFP fluorescence intensity in a region corresponding to the neuronal soma and major neuritis in mock-, SK1-transfected neurons, and SK1-transfected neurons treated with wortmannin (20 nM). *P < 0.001 (Dunnett’s test). (c) The puncta index was estimated by measuring the standard deviation of mApple fluorescence intensity in a region corresponding to the neuronal soma and major neurites in mock-, SK1-transfected neurons, and SK1-transfected neurons treated with wortmannin (20 nM). N.S.—non-significant (Dunnett’s test). (d) Photoswitchable protein Dendra2 as a surrogate for protein turnover. Brief irradiation with short-wave-length visible light causes Dendra2 to undergo an irreversible conformational change (“photoswitch”) and emit red fluorescence that can be tracked until altered molecules are cleared. Optical pulse-chase of a primary neuron expressing Dendra2 with an automated microscope. Note the decay of red fluorescence. Bar, 50 μm. (e) The single-cell half-life of Dendra2-LC3 was significantly reduced by SK1 and beclin1 expression, and increased by S1PP and S1PL expression. Beclin1 was used as a positive control. The change in red fluorescence intensity over time was used to calculate the half-life of Dendra2-LC3, and the half-life of Dendra2-LC3 when SK1, beclin1, S1PP or S1PL is overexpressed. *P < 0.01 (ANOVA). (f) Probability plot of the half-lives measurements from individual neurons. *P < 0.01 (Kolmogorov-Smirnov test).

Figure 2. Pharmacological stimulation of autophagy causes the formation of SK1-positive puncta in primary neurons.

(a) In a significant percentage of primary rat cortical neurons, ectopically expressed SK1-GFP exhibits a punctate appearance (left panel), whereas the distribution of a co-transfected marker of morphology mCherry (right panel) is diffuse. Bar, 50 μm. (b) Percentage of neurons with SK1-GFP- and mCherry-positive puncta. *P < 0.01 (t-test). (c) Stimulation of autophagy with 5 μM 10-NCP promotes the formation of SK1-GFP-positive puncta. Live neurons transfected with SK1-GFP and mCherry were observed before (left panel) and after (right panel) a 4-h incubation with 10-NCP. Bar, 50 μm. See bottom panel for higher magnification image of SK1 puncta. Puncta before and after the treatment are depicted with arrows. (d) SK1-GFP-positive puncta were efficiently formed in cortical neurons following treatment with 5 μM 10-NCP or with 5 μM fluphenazine or by incubation in insulin-free medium as reflected by the puncta index. The puncta index was estimated by measuring the standard deviation of SK1-GFP fluorescence in a region corresponding to the neuronal soma and major neurites before and after treatment with 10-NCP (5 μM, 4 h) or with fluphenazine (FPZ; 5 μM, 4 h) or after incubation in insulin-free media (-ins; 4 h). *P < 0.001 (Dunnett’s test). N.S.—non-significant (t-test). (e) The puncta index was estimated by measuring the standard deviation of mCherry fluorescence in the neuronal soma and major neurites. N.S.—non-significant (t-test). (f) Stimulation of autophagy with 10-NCP increases the number of SK1-GFP puncta not only in the soma, but also in neurites. *P < 0.001 (t-test).

Figure 3. SK1 partially colocalizes with autophagosomal organelles in neurons.

(a) Stimulation of autophagy with 5 μM 10-NCP promotes the formation of SK1-GFP-positive puncta. Neurons transfected with SK1-GFP and RFP-LC3 were treated with 10-NCP (right) or vehicle (left), fixed, and analyzed by confocal microscopy. In non-treated cells, few RFP-LC3 puncta were observed (left). In 10-NCP-treated cells, most SK1-GFP and RFP-LC3 puncta colocalized. Bar, 5 μm. (b) The puncta index was measured by analyzing the standard deviation of SK1-GFP or RFP-LC3 fluorescence in a region corresponding to neuronal processes. *P < 0.001 (t-test). This confirms that autophagy in neurons is upregulated. (c) Colocalization of SK1-GFP and RFP-LC3 signals was measured with the Coloc_2 plugin for the image processing program ImageJ/Fiji. The colocalization algorithms were run as the SK1-GFP signal against the RFP-LC3 signal (SK1-GFP x RFP-LC3) and the RFP-LC3 signal against the SK1-RFP (RFP-LC3 x SK1-GFP). *P < 0.01, **P < 0.0001 (ANOVA), N.S., non-significant. (d) SK1-GFP colocalizes with autophagosomal organelles in neurons. An example of a neurite in which an organelle (arrow) is both positive for SK1-GFP and RFP-LC3. Two sharp arrows depict a possible fusion event between a SK1-GFP-positive organelle and an RFP-LC3-positive organelle. Some organelles are clearly positive for SK1-GFP but not RFP-LC3 and, therefore, are not autophagosomal. Bar, 2 μm. (e) The dominant negative form of SK1 inhibits autophagosome formation. The expression of dnSK1-GFP prevents the formation of RFP-LC3 puncta in neurons stimulated with 10-NCP. Neurons transfected with dnSK1-GFP and RFP-LC3 were treated with 10-NCP, fixed, and analyzed with confocal microscopy. Bar, 20 μm. (f) The puncta index was measured by analyzing the standard deviation of RFP-LC3 fluorescence in a region corresponding to the neuronal soma and major neurites in neurons co-expressing SK1-GFP and dnSK1-GFP. The S1PP or S1PL constructs were used as controls. Neurons were stimulated with 10-NCP. *P < 0.001 (Dunnett’s test). Note that the puncta index is lower in dnSK1-GFP-expressing neurons and in neurons that express S1P-degrading enzymes.

Figure 4. SK1 colocalizes with endosomal markers in neurons stimulated with an autophagy enhancer.

(a) Stimulation of autophagy with 10-NCP promotes the formation of SK1-GFP-positive puncta which colocalize with early endosomes (marked by EEA1 and Rab5 staining; top panels) and with late endosomes (marked by Rab7 staining; lower middle panel). In contrast, SK1-GFP was significantly less colocalized with a marker of lysosomes (LAMP1; bottom panel). Neurons transfected with SK1-GFP were treated with 5 μM 10-NCP (4 h), fixed, stained with an antibody against EEA1, Rab5, Rab7 or LAMP1, and analyzed by confocal microscopy. Bar, 50 μm. (b) Colocalization of SK1-GFP and EEA1, Rab5, Rab7 or LAMP1 signals was measured with the Coloc_2 plugin for the image processing program ImageJ/Fiji. The colocalization algorithms were run as the SK1-GFP signal against an organelle marker signal (SK1-GFP x a marker). (c) SK1-RFP does not colocalize with the lysosomal dye Lysotracker Green DND-26. Neurons transfected with SK1-RFP were treated with 5 μM 10-NCP (2 h), then treated with 75 nM Lysotracker Green DND-26 for 30 min and imaged. Arrows indicate distinct SK1-RFP-positive and lysotracker-positive organelles, which are located nearby. Bar, 50 μm.

Figure 5. Many endosomes make contacts with the ER.

(a) Electron micrographs of primary cortical neurons treated with a vehicle or 10-NCP (5 μM, 4 h). ER, the endoplasmic reticulum; E, endosome. Note that in an autophagy-stimulated neuron, a large endosome is surrounded by the ER. Bar, 100 nm. (b) In neurons stimulated with 10-NCP, SK1-RFP-positive structures make contacts with EYFP-WIPI1 structures. Neurons transfected with SK1-RFP and WIPI1-YFP were treated with 5 μM 10-NCP (4 h), fixed, and analyzed with confocal microscopy. Bar, 2 μm. Bar in the zoomed image, 0.5 μm. (c) Two S1P-metabolizing enzymes, S1PP and S1PL, are located on the ER. Neurons transfected with S1PP-mApple and Green-ER or with S1PL-mApple and Green-ER were fixed and analyzed by confocal microscopy. Bar, 5 μm. (d) S1PP and S1PL are located on the ER in 10-NCP-treated neurons. Neurons transfected with S1PP-mApple and Green-ER or with S1PL-mApple and Green-ER were treated with 10-NCP (5 μM, 2 h), fixed and analyzed by confocal microscopy. Bar, 5 μm.

Figure 6. SK1 and S1PL modulate the half-life of a substrate of autophagy, Htt^ex1-Q₄₆-Dendra2.

(a) Optical pulse labelling of striatal neurons expressing Htt^ex1-Q₄₆-Dendra2. The single-cell half-life of Htt^ex1-Q₄₆-Dendra2 was reduced by SK1 expression, and increased by S1PL expression. Neurons were transfected with either with Htt^ex1-Q₄₆-Dendra2 and an empty plasmid or Htt^ex1-Q₄₆-Dendra2 and SK1 or Htt^ex1-Q₄₆-Dendra2 and S1PL. The change in photoswitched red fluorescence intensity over time was used to calculate the half-life of Htt^ex1-Q₄₆-Dendra2 in three neuronal cohorts. *P < 0.01 (ANOVA). (b) Probability plot of the half-lives measurements from individual neurons. *P < 0.01 (Kolmogorov-Smirnov test).

Figure 7. An inhibitor of S1PL, THI, enhances survival of neurons expressing Htt^ex1-Q₇₂-GFP.

(a) An example of survival analysis. Primary cortical neurons transfected with mApple (a morphology and viability marker) and Htt^ex1-Q₇₂-GFP were tracked with an automated microscope. Images collected 24 h after transfection and 24 h thereafter demonstrate the ability to return to the same field of neurons and follow them over time. Each segmented image is a montage of non-overlapping images captured in one well of a 24-well plate. Arrows indicate randomly chosen neurons (labelled 1 and 2). These two neurons are enlarged on the right panel. Scale bar on the left panel is 400 μM. Scale bar on the left panel is 20 μm. (b) Cortical neurons transfected with mApple and Htt^ex1-Q₇₂-GFP were treated with 0.5 μM THI or 0.5 μM 10-NCP or vehicle. Cumulative risk for death was calculated with JMP software. 10-NCP (a positive control) and THI reduced the risk for death (i.e., improved survival) of neurons expressing mutant Htt^ex1. *P < 0.001 (Log-Rank test). (c) Striatal neurons transfected with mApple and Htt^ex1-Q₇₂-GFP were treated with 0.5 μM THI or 0.5 μM 10-NCP or vehicle. Cumulative risk for death was calculated with JMP software. 10-NCP and THI reduced the risk for death of neurons expressing mutant Htt^ex1. *P < 0.001 (Log-Rank test).

Figure 8. A possible mechanism of SK1-mediated autophagy.

Pharmacological stimulation of autophagy results in activation of SK1 (1), which then relocalizes to endosomes (2). An endosome then may make contacts with the ER surface, where WIPI1 and beclin1 protein complexes are assembled (3), resulting in the biogenesis of an pre-autophagosomal structure (4). Once formed (5), a LC3-positive autophagosome (5) may fuse with the endosome to form SK1- and LC3-positive amphisomes (6). S1PP and S1PL are localized to the ER, ensuring that unwanted S1P can be locally metabolized to stop S1P signaling (7).