(a) SphK activities between NP–C and control were analysed in human fibroblast (n=7 per group), mouse cerebellum tissue (n=7 per group) and primary mouse PN samples (n=9 per group). SphK activity did not show passage differences between NP–C and normal fibroblasts. (b) Three days after cocultures, we measured SphK activities in PNs derived from WT and NP–C mice (n=8 per group). (c) VEGF levels were measured in CM derived from PNs with or without BM-MSCs by ELISA (n=7 per group). (d) Primary cultures of NP–C PNs were immunostained with anti-calbindin and anti-VEGF (scale bar, 50 μm). Arrowheads indicate VEGF expression by PNs. Values represent normalized fluorescence intensities of VEGF in PNs (WT PN, n=8; and NP–C PN, n=9). (e) SphK activities were measured in NP–C PNs alone (n=7) and NP–C PNs cocultured with BM-MSCs, VEGF siRNA BM-MSCs and VEGF^tg BM-MSCs (n=8 per group). (f) Effect of the PTK787 on BM-MSCs mediated SphK activation. NP–C PNs were pretreated with PTK787 at 10 μM for 1 day and cocultured for 3 days with BM-MSCs, and then SphK activity was assayed (n=7 per group). (g) Representative images of PNs stained with anti-calbindin (scale bar, 100 μm). The mean number of PNs per squared millimetre was counted (n=8 per group). (h) Effect of VEGF knockdown on SphK activity in PNs (control, n=6; and VEGF siRNA, n=8 per group). (i) Representative images and quantification of neuronal survival in normal and VEGF-knockdown PNs (scale bar, 50 μm; n=8 per group). a,h,i, Student’s t-test. b–g, one-way analysis of variance, Tukey’s post hoc test. *P<0.05, **P<0.01, ***P<0.005. All error bars indicate s.e.m.

(a) Protocol of BM-MSC treatment in NP–C mice. (b,c) SphK activity (n=7 per group; b) and VEGF levels (n=8 per group; c) were estimated in the cerebellums of WT and NP–C mice after BM-MSC treatment. (d) Cerebellar sections were stained with anti-calbindin and anti-VEGF (low-magnification scale bar, 50 μm; high-magnification scale bar, 20 μm). Values represent normalized VEGF fluorescence intensities in PCL (n=7 per group). (e) SphK activities were measured in the cerebellums of NP–C mice treated with PBS (n=6), BM-MSCs, VEGF siRNA BM-MSCs and VEGF^tg BM-MSCs (n=8 per group). (f) Left, isolation of mouse PNs using LCM (scale bar, 75 μm). Right, mRNA level of Vegf, VEGFR2 and Sphk1 on LCM-captured PNs samples (n=7 per group). (g) NP–C mice were treated daily with the PTK787 at 100 mg kg⁻¹ or PBS, starting 2 days before the BM-MSC transplantation. One day after BM-MSC treatment, SphK activity was estimated (NP–C, n=7; NP–C BM-MSC TP, n=8 per group). (h) Cerebellar sections were stained with anti-calbindin (scale bar, 50 μm), and the number of calbindin-positive PNs were quantified (n=7 per group). (i) Rota-rod scores of mice were averaged and plotted beginning 3 days after transplantation (n=15 per group). (j) Survival curve of NP–C mice (n=15 per group). Treatment with BM-MSCs and VEGF^tg BM-MSCs resulted in significantly increased survival compared with PBS treatment (P=0.0194 and P=0.0055, respectively; log-rank test). (k) Effect of VEGF knockdown on SphK activity. Left, after intracerebellar injection of control (n=7) or VEGF shRNA (n=8) in mice, SphK activities were measured in the cerebellums. Right, relative levels of Sphk1 mRNA from LCM-captured PNs samples (n=7 per group). b–i, one-way analysis of variance, Tukey’s post hoc test. k, Student’s t-test. *P<0.05, **P<0.01, ***P<0.005. All error bars indicate s.e.m.

(a–c) Primary cultures of normal and VEGF^tg PNs were transfected with control or NPC1 siRNA. Three days after transfection, we measured the levels of Npc1 (a) and Vegf (b) mRNA and secreted VEGF protein (c) in PNs (n=7 per group). (d,e) SphK activity (d), sphingosine and S1P (e) were estimated in PNs transfected with control (n=6) or NPC1 siRNA (n=7). (f–j) Four-week-old WT and VEGF^tg mice were injected with control or NPC1 shRNA into the cerebellum. Mice were sacrificed at 3 days after the injection. Npc1 (f) and Vegf (g) mRNA levels were estimated in LCM-captured PNs and VEGF protein levels (h) were measured in the cerebellums (n=7 per group). (i) Left, SphK activities were measured in the cerebellums (n=7 per group). Right, relative levels of Sphk1 mRNA from LCM-captured PNs samples (n=8 per group). (j) Sphingosine and S1P were measured in the cerebellums (n=7 per group). a–j, one-way analysis of variance, Tukey’s post hoc test. *P<0.05, **P<0.01, ***P<0.005. All error bars indicate s.e.m.

(a) Cerebellar sections from 6-week-old WT, VEGF, NP–C and VEGF/NP–C mice were immunostained with anti-calbindin and anti-VEGF (low-magnification scale bar, 50 μm; high-magnification scale bar, 20 μm). The average VEGF fluorescence intensity within the PCL was measured (WT, n=7; VEGF, n=7; NP–C, n=9; and VEGF/NP–C, n=9). (b) SphK activities were measured in cerebellums derived from 6-week-old WT, VEGF, NP–C and VEGF/NP–C mice (n=8 per group). (c) Quantitative real-time PCR for Vegf, VEGFR2 and Sphk1 mRNA in LCM-captured PNs in 6-week-old WT, VEGF, NP–C and VEGF/NP–C mice (WT, n=6; VEGF, n=6; NP–C, n=8; and VEGF/NP–C, n=8). (d) VEGF/NP–C mice were treated daily with the PTK787 at 100 mg kg⁻¹ or PBS vehicle control for 3 days before sacrifice (6-week-old), and SphK activity was estimated in cerebellums (n=7 per group). (e) Cerebellar sections were immunostained with anti-calbindin and the number of calbindin-positive PNs was quantified (WT, n=7; VEGF, n=7; NP–C, n=8; and VEGF/NP–C, n=8). (f) Left, beginning at 4 weeks of age, Rota-rod scores were averaged and plotted (n=15 per group). Right, survival curves of NP–C and VEGF/NP–C mice (P=0.0548; log-rank test, n=15 per group). (g) Cerebellar sections from WT and NP–C mice transplanted with VEGF-loaded or control microspheres were stained with anti-calbindin and anti-VEGF. The average VEGF fluorescence intensity within the PCL was measured (n=7 per group). (h) SphK activity was estimated in the cerebellums of WT and NP–C mice at one day after treatment. (i) Cerebellar sections were prepared at 2 weeks after transplantation and immunostained with anti-calbindin. The calbindin-positive PNs were counted (n=7 per group). a–i, one-way analysis of variance, Tukey’s post hoc test. *P<0.05, **P<0.01, ***P<0.005. All error bars indicate s.e.m.

(a) Western blot analysis of LC3, beclin-1, p62 and cathepsin D in primary cultured PNs derived from WT, NP–C and VEGF/NP–C mice (WT, n=5; NP–C, n=6; and VEGF/NP–C, n=6). (b) Immunocytochemistry of LC3 in WT, NP–C and VEGF/NP–C PNs (n=6 per group; scale bar, 20 μm). (c) Cathepsin D activity in primary cultured PNs (WT, n=5; NP–C, n=6; and VEGF/NP–C, n=6). (d) Western blot analysis of LC3, beclin-1, p62 and cathepsin D in the cerebellums of 6-week-old WT, NP–C and VEGF/NP–C mice (WT, n=6; NP–C, n=7; and VEGF/NP–C, n=7). (e) Cathepsin D activity in the cerebellums of WT, NP–C and VEGF/NP–C mice (WT, n=5; NP–C, n=6; and VEGF/NP–C, n=6). (f) EM images and quantification data of the cerebellum (n=5 per group; low-magnification scale bar, 1 μm; high-magnification scale bar, 200 nm). Arrow indicates autophagic vacuole. (g) Western blot analysis of Rab5 and Rab7 levels in the cerebellum (n=6 per group). (h) Cerebellar sections were immunostained with anti-active caspase-3 and the number of active caspase-3-positive cells in PCL was quantified (n=5 per group; scale bar, 50 μm). a–g, one-way analysis of variance, Tukey’s post hoc test. h, Student’s t-test. *P<0.05, **P<0.01, ***P<0.005. All error bars indicate s.e.m.

(a) Western blots of LC3, beclin-1 and p62 in PNs after VEGF knockdown (n=6 per group). (b) Immunocytochemistry of LC3 in PNs after VEGF knockdown (n=6 per group; scale bar, 20 μm). (c) Autophagic flux assay. Western blots of LC3 and p62 in PNs (n=6 per group). (d) Western blots of LC3 and p62 in cultured PNs in the presence of NH₄Cl (n=5 per group). (e) Western blots of TFEB and Lamp1 in VEGF-knockdown PNs (control, n=5 and VEGF siRNA, n=6). (f) Effect of VEGF knockdown on lysosomal pH. PNs stained with LysoTracker red (n=5 per group; scale bar, 20 μm). (g) Fluorescence analysis of autophagosomes and autolysosomes (control, n=7 and VEGF siRNA, n=8; scale bar, 10 μm). (h) Sphk activity, sphingosine and S1P levels in PNs after VEGF knockdown in the presence of curcumin (n=8 per group). (i) Western blot analysis of LC3 and p62 in VEGF-knockdown PNs treated with curcumin (control, n=7; VEGF siRNA, n=8; and VEGF siRNA/curcumin, n=8). (j) Survival of VEGF-knockdown PNs treated with curcumin (n=8 per group). (k) Left, representative traces showing intracellular [Ca²⁺] changes monitored in single fluo-4-loaded PNs. Right, maximal peak fluorescence changes were determined as the differences between basal and the maximum fluorescence (n=10 cells per group). (l) Sphk activity and sphingosine levels were measured in cultured PNs (n=8 per group). (m) Quantification of autophagosomes and autolysosomes in primary cultured PNs (WT, n=7; NP–C, n=7; and VEGF/NP–C, n=8). (n) EM analysis of the PNs (n=5 per group; low-magnification scale bar, 1 μm; high-magnification scale bar, 200 nm). (o) Survival of primary cultured PNs (WT, n=6; NP–C, n=8; and VEGF/NP–C, n=8). a,b,e,g, Student’s t-test. c,d,f,h–o, one-way analysis of variance, Tukey’s post hoc test. *P<0.05, **P<0.01, ***P<0.005. All error bars indicate s.e.m.

(a) Left, normal, hNPC-3 and hNPC-17 iPSCs generated nestin-positive neuroprogenitor cells (scale bar, 50 μm). Right, representative images of immunocytochemical staining the β-III tubulin following neural differentiation (scale bar, 50 μm). (b) hNPC-3 and hNPC-17 neurons were treated with human BM-MSCs, VEGF^tg BM-MSCs or recombinant VEGF (10 ng ml⁻¹). Three days after treatment, VEGF levels were measured in cell lysates. (c–f) SphK activity (c), sphingosine (d), S1P (e) and sphingomyelin (f) were measured in normal iPSC neurons and hNPC neurons with or without treatment. (g) Filipin staining of unesterified cholesterol in hNPC-3 neurons with or without treatment of recombinant VEGF for 3 days (scale bar, 50 μm). Quantification of filipin fluorescence intensities normalized to normal neurons. Unesterified cholesterol levels in normal iPSC neurons and hNPC neurons with or without treatment were measured (n=6 per group). (h) Effect of the VEGFR2 inhibitor on VEGF mediated sphingolipid modulation. hNPC-3 neurons were pretreated with PTK787 at 10 μM for 1 day and were treated for 3 days with 10 ng ml⁻¹ VEGF and then assayed for SphK activity, sphingosine and S1P (n=7 per group). (i,j) Effect of NPC1 knockdown on sphingolipid factors in normal iPSC neurons. (i) Three days after NPC1 siRNA transfection, we measured the levels of NPC1 mRNA and VEGF expression. (j) SphK activity, sphingosine and S1P were measured in normal iPSC neurons treated with control or NPC1 siRNA (control, n=7; NPC1 siRNA, n=9). (k) Effect of VEGF knockdown on sphingolipid factors in normal iPSC neurons. Three days after VEGF siRNA transfection, we measured the levels of VEGF mRNA, SphK activity, sphingosine and S1P in normal iPSC neurons (n=7 per group). (l) Effect of a specific SphK1 inhibitor on sphingolipid factors in normal iPSC neurons. Normal neurons were treated with or without 20 μM SK1-I for 6 h. Lipids were extracted and sphingosine, S1P and unesterified cholesterol levels were determined (n=6 per group). b–h, one-way analysis of variance, Tukey’s post hoc test. i–l, Student’s t-test. *P<0.05, **P<0.01, ***P<0.005. All error bars indicate s.e.m.

(a) Western blot analysis of LC3 and p62 in normal and NP–C iPSC-derived neurons treated with or without 10 ng ml⁻¹ recombinant VEGF (normal, n=6; hNPC, n=7; and VEGF-treated hNPC, n=7). (b) Left, representative traces showing intracellular [Ca²⁺] changes monitored in single fluo-4-loaded normal and NP–C iPSC neurons treated with or without recombinant VEGF (10 ng ml⁻¹). Right, maximal peak fluorescence changes were determined as the differences between basal and the maximum fluorescence, on addition of 200 μM GPN (n=10 cells per group). (c) Fluorescence staining and quantification of autophagosomes (mCherry⁺-EGFP⁺-LC3) and autolysosomes (mCherry⁺-EGFP⁻-LC3) in normal and NP–C iPSC neurons after recombinant VEGF treatment (normal, n=7; hNPC, n=8; and VEGF-treated hNPC, n=8; scale bar, 10 μm). (d) EM images and quantification data of normal and NP–C iPSC-derived neurons after 10 ng ml⁻¹ VEGF treatment (n=5 per group; low-magnification scale bar, 1 μm; high-magnification scale bar, 200 nm). (e) Quantification of cell viability (normal, n=5; hNPC, n=6; and VEGF-treated hNPC, n=6). (f) Model of VEGF-mediated SphK activation in NP–C neurons. (A,B) In NP–C cells, sphingosine accumulation is increased due to defective SphK activity together with decreased VEGF caused by mutated NPC1 and defective uptake via VEGFR2. (C) Abnormal sphingosine accumulation decreases calcium release from lysosomes and the reduction in calcium release causes an autophagic defect by inhibiting autophagosome–lysosome fusion. Eventually, these defects cause loss of cerebellar neurons. (D,E) When NP–C neurons are exposed to BM-MSCs or pure VEGF, the cells exhibit elevated intracellular levels of VEGF (F), which induces VEGFR2-mediated activation of SphK in the cytosol and lysosome. (G) This activation leads to decreased sphingosine accumulation and increased S1P levels. (H) Reduced sphingosine accumulation results in improved autophagosome–lysosome fusion by correction of calcium homeostasis. Finally, this restoration prevents neuronal loss in NP–C. a–e, one-way analysis of variance, Tukey’s post hoc test. *P<0.05, **P<0.01, ***P<0.005. All error bars indicate s.e.m.