Lysosome-targeted multifunctional lipid probes reveal the sterol transporter NPC1 as a sphingosine interactor

Abstract

Lysosomes are catabolic organelles involved in macromolecular digestion, and their dysfunction is associated with pathologies ranging from lysosomal storage disorders to common neurodegenerative diseases, many of which have lipid accumulation phenotypes. The mechanism of lipid efflux from lysosomes is well understood for cholesterol, while the export of other lipids, particularly sphingosine, is less well studied. To overcome this knowledge gap, we have developed functionalized sphingosine and cholesterol probes that allow us to follow their metabolism, protein interactions, and their subcellular localization. These probes feature a modified cage group for lysosomal targeting and controlled release of the active lipids with high temporal precision. An additional photocrosslinkable group allowed for the discovery of lysosomal interactors for both sphingosine and cholesterol. In this way, we found that two lysosomal cholesterol transporters, NPC1 and to a lesser extent LIMP-2/SCARB2, bind to sphingosine and showed that their absence leads to lysosomal sphingosine accumulation which hints at a sphingosine transport role of both proteins. Furthermore, artificial elevation of lysosomal sphingosine levels impaired cholesterol efflux, consistent with sphingosine and cholesterol sharing a common export mechanism.

Keywords: lysosomal storage diseases; organelle-targeted probes; photocrosslinking; protein–lipid interaction; sphingolipids.

Fig. 1.

Synthesis and characterization of lysosome-targeted probes. (A) Synthesis of lysosome-targeted coumarin (lysocoumarin, 4), lysosome-targeted pacSphingosine (lyso-pacSph, 5), and lysosome-targeted pacCholesterol (lyso-pacChol, 6). Annotation of the functional groups by color-coded legend. (B and C) Lysosomal localization of lysoprobes. Confocal images of coumarin fluorescence in HeLa cells pulsed with lyso-pacSph or lyso-pacChol (10 µM) for 1 h and 45 min, respectively, chased overnight, and incubated with LysoTracker Red (100 nM) 30 min previous to imaging. (Scale bar, 10 µm.) Lysoprobe: cyan and LysoTracker Red: magenta (D and E) Thin layer chromatography of HeLa cells pulsed with lyso-pacSph or lyso-pacChol (10 µM) for 1 h and 45 min, respectively, chased overnight, harvested immediately, and/or irradiated with a 405-nm UV light for 90 s and further incubated at 37 °C for indicated times. Cellular lipids were extracted, labeled with 3-azido-7-hydroxy coumarin, spotted, and developed on a silica plate.

Fig. 2.

Application of lyso-pacChol and lyso-pacSph for identification of protein–lipid interactions. (A) Scheme illustrating the workflow used to capture protein interactors. Created with Biorender.com (B) Differential identification of proteins cross-linked to pacChol (10 µM) vs. lyso-pacChol (10 µM) and pacSph (1 µM) vs. lyso-pacSph (5 µM). Biotinylated protein–lipid complexes were visualized using fluorescently labeled streptavidin. Proteins identified with lysoprobes but not with globally distributed probes are marked with red arrows. (C) Time-dependent identification of FLAG-tagged ceramide synthase 5 (CerS5) by lyso-pacSph compared to globally distributed pacSph. HeLa SGPL1^−/− cells were labeled with lyso-pacSph (5 µM) or pacSph (2 µM for 1 h) uncaged and either cross-linked immediately or incubated 15 min before cross-linking and subjected to the workflow in (A) featuring cross-linking (+UV) or no cross-linking (−UV) steps. The inputs (I, 10%), flow-throughs (FT, 10%), and eluates (E, 100%) were immunoblotted against FLAG tag. (D) Chemoproteomic analysis of lyso-pacChol (10 µM) and lyso-pacSph (5 µM) interactors. Volcano plot showing the results of a differential abundance analysis using the limma package (moderated t test and P values estimated by the fdrtool package) of proteins cross-linked to the respective probes. Log₂ fold change of cross-linked over noncrosslinked (x axis) and negative log₁₀ P values (y axis) of protein interactors. Hit proteins (red annotated dots) displayed a false discovery rate of  ≤0.01 and a fold change of >1.5.

Fig. 3.

Lyso-pacChol and lyso-pacSph metabolism in WT, NPC1 KO, and LIMP2 KO cells by TLC. (A) Postlysosomal metabolism of lyso-pacChol (A). HeLa WT, NPC1 KO, or LIMP-2/SCARB2 KO cells were labeled with lyso-pacChol (10 µM) or for 45 min and chased overnight. Upon uncaging, cells were chased and lipids extracted at indicated times, clicked with 3-azido-7-hydroxycoumarin, and visualized by TLC. (B) Quantification of cholesterol esterification in WT, NPC1, and LIMP2-deficient HeLa cells. The intensity of the ester corresponding to pacChol ester divided by the sum of pacChol and pacChol ester is displayed as percentage. (C) TLC analysis of lipid metabolites arising from incubation of HeLa WT, NPC1 KO or LIMP-2/SCARB2 KO cells with lyso-pacSph (10 µM) for 1 h, overnight chase, uncaging and extraction at the indicated times. (D) Quantification of lysosomal sphingosine (Sph) export in WT, NPC1, and LIMP2-deficient HeLa cells. Sph is readily metabolized not only to ceramide (Cer) and sphingomyelin (SM) but also to phosphatidylcholine (PC) via the SGPL1 breakdown pathway. Intensity of the Sph is expressed as percentage compared to the sum of all labeled lipids. Data are shown as mean of three independent experiments ± SE.

Fig. 4.

Visualization of lysosomal lipid egress in WT, NPC1 KO, LIMP2 KO, and NPC1/LIMP2 DKO cells. Subcellular localization of lyso-pacChol (A) and lyso-pacSph (B) in WT, NPC1, LIMP2, and NPC1/LIMP2-deficient HeLa cells. Confocal images of cells labeled with lyso-pacChol or lyso-pacSph (10 µM) for 45 min and 1 h, respectively, and chased overnight. Upon uncaging, cells were chased for the indicated times, cross-linked, fixed, and functionalized with AlexaFluor 594 picolyl azide (B) or AlexaFluor 488 picolyl azide (A). (Scale bar, 10 µm.) (C and D) Quantification of lysosomal lipid egress. Pearson’s R value of nonthresholded images from lipid channel vs. LAMP1 immunofluorescence calculated for each time point (n ≥ 45) using the Coloc 2 feature from Fiji (ns P value > 0.5, *P ≤ 0.05, **P  ≤ 0.01, ***P value ≤ 0.001, and ****P value ≤ 0.0001).

Fig. 5.

Sphingosine accumulation is a direct effect from loss of NPC1. (A) Confocal images of lyso-pacSph in cellular models of NPC1. HeLa NPC1 KO cells (NPC1), HeLa WT cells treated with 0.5 µg/mL U18666A 24 h prior to imaging (U18666A), and HeLa NPC1 KO cells incubated with lipoprotein-deficient medium 48 h prior to imaging (starvation) were labeled with lyso-pacSph (1.25 µM) for 1 h and chased overnight. Upon uncaging, cells were chased from 0 to 30 min, cross-linked, fixed with methanol, and functionalized with AlexaFluor 594 picolyl azide. (Scale bar, 10 µm.) (B) Quantification of lysosomal lipid egress. Pearson’s R value of nonthresholded images from lipid channel vs. LAMP1 immunofluorescence calculated for each time point (n ≥ 42) using the Coloc 2 feature from Fiji (ns P value > 0.05. (C) Confocal images of lyso-pacSph in cellular models of NPC1. HeLa 11ht NPC1 KO cells (NPC1 KO) with NPC1-Flag induced on an NPC1 KO background (NPC1-WT) or NPC1 (L472P)-Flag induced on an NPC1 KO background (NPC1 L472P) were labeled with lyso-pacSph (10 µM) for 1 h and chased overnight. Upon uncaging, cells were chased from 0 to 30 min, cross-linked, fixed with methanol, and functionalized with AlexaFluor 594 picolyl azide. (Scale bar, 10 µm.) (D) Quantification of lysosomal Sph egress in cellular models of NPC1. Pearson's R value of nonthresholded images from lipid channel vs. LAMP1 immunofluorescence calculated for each time point (n ≥ 73) using the Coloc 2 feature from Fiji (ns P value > 0.05 and ****P value ≤ 0.0001).

Fig. 6.

Elevation of lysosomal sphingosine levels can recreate an NPC1-like phenotype. (A) Synthesis of lysosome-targeted sphingosine (lyso-Sph, 7) and lysosome-targeted cholesterol (lyso-Chol, 8). (B) Illustrative model of the experiment. (C and D) Confocal images from HeLa WT cells labeled with either lyso-pacSph (1.25 µM) and lyso-Chol (25 µM) (C) or lyso-pacChol (750 nM) and lyso-Sph (15 µM) (D) for 1 h and chased overnight. Upon uncaging, chase cross-linking experiments were performed from 0 to 30 min. Cells were fixed with methanol and functionalized with AlexaFluor 594 picolyl azide (C) or AlexaFluor 488 azide (D). (Scale bar, 10 µm.) (E and F) Quantification of lysosomal lipid egress in C and D. Pearson’s R value of nonthresholded images from lipid channel vs. LAMP1 immunofluorescence calculated for each time point (n ≥ 42) using the Coloc 2 feature from Fiji (ns P value > 0.05, *P value ≤ 0.05, **P value ≤ 0.01, ***P value ≤ 0.001, and ****P value ≤ 0.0001).