Lipidomics analysis of a LTP knockout library revealed known and candidate controllers of cellular lipid levels.

(A) Overview of the arrayed gene knockout library targeting LTPs. NT: non-targeting. (B) Western blotting of the selected wells showing the efficiency of the LTP knockout library on MelJuSo cells. gNT: Non-targeting, control cells. (C) Summary plot of the lipidomics analysis: Z-scores calculated for 17 lipid classes are plotted. Each data point represents the mean Z-score of an LTP knockout cell line; size and color of data points represent the standard deviation of Z-scores for 3 experiments on average. PC: phosphatidylcholine, LPC: lyso-phosphatidylcholine, PE: phosphatidylethanolamine, LPE: lyso-phosphatidylethanolamine, PS: phosphatidylserine, PI: phosphatidylinositol, PA: phosphatidic acid, PG: phosphatidylglycerol, DCer: dihydroxyceramide, Cer: ceramide, SM: sphingomyelin, HexCer: hexosylceramide, LacCer: lactosylceramide, DG: diacylglycerol, TG: triacylglycerol, CE: cholesterol ester, FFA: free fatty acids.

NPC1 and NPC2 knockout cells accumulate sphingomyelin in lysosomes.

(A) GLTP knockouts show decreased glucosylceramide levels compared to control cells. GlcCer: glucosylceramide (B) NPC1 and NPC2 knockout cells have increased sphingolipids levels compared to control cells. (C) Increased sphingolipid levels in NPC1 and NPC2 knockouts are represented in all sphingolipid classes. (D) Immunofluorescence images of NPC1 and NPC2 knockout cells permeabilized and stained with recombinant GFP-EQT show the accumulation of sphingomyelin in the lysosomes compared to the parental MelJuSo (MJS) cells. Pixel intensities of the images were adjusted evenly. Scale bar 10µm. (E) Flow cytometry analysis of cells stained with EQT-GFP, CTB-Alexa568 and CD63-V450 on the cell surface. Analysis demonstrates the accumulation of sphingomyelin (EQT-GFP) and the glycosphingolipid GM1 (CTB-Alexa568) on the cell surface without affecting the surface protein levels. (F) Mean fluorescence intensities for each staining normalized to control cells from 3 experiments. Red lines and error bars correspond to mean and standard deviations from 3 experiments, respectively. Approximately 2000 cells were analyzed per experiment.

CERT, ORP9 and ORP11 knockout cells demonstrate reduced sphingomyelin levels.

(A) CERT, ORP9 and ORP11 knockouts demonstrate decreased sphingomyelin and increased glucosylceramide levels. (B) Schematic representation of de novo sphingomyelin synthesis pathway of mammalian cells. (C) Decreased sphingomyelin and increased glucosylceramide levels observed in knockout cells are present in various sphingolipid subspecies. (D) Similar to CERT knockouts, ORP9 and ORP11 knockouts are sensitive to methyl-β-cyclodextrin treatment. Red lines correspond to mean values from 4 experiments. (E) Western blot of knockout cells showing that the loss of ORP9 or ORP11 does not affect CERT protein levels. (F) ORP9 and ORP11 localize at the Golgi apparatus. Scale bar 10µm. (G) Domain architecture of CERT, ORP9 and ORP11. All proteins contain PH domains for Golgi localization. CERT and ORP9, but not ORP11, contain a FFAT motif for interacting with VAP proteins.

ORP9 dimerization is critical for ER localization of ORP11.

(A) AlphaFold revealing coiled coils in ORP9 and ORP11. (B) AlphaFold-Multimer shows that coiled coils of ORP9 and ORP11 interacting with each other. (C) ColabFold protein complex prediction for full length ORP9 and ORP11 shows the dimerization via coiled coils (arrow). (D) Position aligned error for panel C. Note that the interaction of coils has a lower score (yellow arrows). (E) Coiled coils of ORP9 and ORP11 is sufficient to define their localization as ORP11 coiled coils colocalizes with the mitochondria-targeted ORP9 coiled coils at mitochondria. Scale bar: 10µm. (F) Colocalization analysis of immunofluorescent images corresponding to panel E. Red lines correspond to mean values from 3 experiments; n is number of analyzed cells. (G) Co-immunoprecipitation confirming that the coils of ORP9 and ORP11 interact with each other. (H) Immunofluorescence images of PH domain localizations. Compared to individual PH domains of ORP9 and ORP11, ORP9-ORP11 chimera demonstrates better affinity towards the Golgi. Scale bar: 10µm. (I) Colocalization analysis of immunofluorescent images corresponding to panel H. Red lines correspond to mean values from 3 experiments; n is number of analyzed cells. (J) Co-immunoprecipitation analysis of VAPA, ORP9 and ORP11 from over-expressing cells. Despite lacking a FFAT motif, ORP11 interacts with the ER-resident VAPA protein. This interaction is facilitated by the FFAT motif of ORP9 as a VAPA mutant unable to interact with FFAT motifs (VAPA**) was unable to co-precipitate ORP9 or ORP11.

ORP9 and ORP11 dimerization is critical for their Golgi localization.

(A) Immunofluorescence images of cells stained for ORP9. ORP9 fails to localize at the Golgi as efficiently in ORP11 knockout cells. (B) Colocalization analysis of immunofluorescent images corresponding to panel A. Red lines correspond to mean values from 3 experiments; n is number of analyzed cells. (C) Immunofluorescence images of cells stained using an ORP11 antibody. Similar to ORP9, ORP11 fails to localize at the Golgi as efficiently in ORP9 knockout cells. Note that the effect of ORP9 loss on ORP11 localization is more dramatic than vice versa. (D) Colocalization analysis of immunofluorescent images from panel C. Red lines correspond to mean values from 3 experiments; n is number of analyzed cells. All scale bars are 10µm.

ORP9 and ORP11 are essential for PS and PI(4)P levels in the Golgi apparatus.

(A) Phylogenetic tree of human OSBP-related domains showing that ORP9 and ORP11 belong to the PS transporter branch. (B) Schematic representation of FRET-based lipid transfer assay. Rhodamine in the donor vesicles quenches NBD fluorescence unless NBD-labeled lipid is transferred to acceptor liposomes. (C) PS transfer activity of ORP9-ORD and ORP11-ORD (ORD9 and ORD11, respectively) is promoted by addition of PI(4)P to acceptor liposomes, suggesting counter transfer of PI(4)P. (D) The PI(4)P sensor OSBP-PH localizes more prominently to the Golgi area in ORP9 and ORP11 knockout cells. (E) Golgi quantification of the PI(4)P sensor confirms the accumulation of PI(4)P at this organelle. This phenotype was rescued by reconstitution of the missing protein in knockout cells (shown with a dagger). Red lines correspond to mean values from 3 experiments; n is number of analyzed cells. (F) The PS sensor Lact-C2 localizes less prominently to the Golgi in ORP9 and ORP11 knockout cells. (G) Golgi quantification of the PS sensor RFP-LactC2 indicates reduced PS levels in this organelle. This phenotype could be rescued by reconstitution of the missing protein in knockout cells (shown with a dagger). Red lines correspond to mean values from 3 experiments; n is number of analyzed cells.

ORP9 and ORP11 are needed for de novo sphingomyelin synthesis in the Golgi apparatus.

(A) Thin-layer chromatography readout of the enzymatic activity assay performed in lysates reveals the unreduced sphingomyelin synthesis capacity of knockout cells. (B) Quantification of the enzymatic activity assays corresponding to panel A. Note the increased GlcCer production capacity in the knockout cells. Red lines and error bars correspond to mean and standard deviation from 5 experiments, respectively. (C) Graphic representation of experiments using palmitic acid alkyne for monitoring de novo sphingomyelin synthesis. Alkyne modified sphingolipids are “clicked” with azido-coumarin before thin-layer chromatography (TLC) analysis. (D) TLC readout of de novo sphingomyelin synthesis assay in intact cells using palmitic acid alkyne. Knockout cells demonstrate reduced conversion of palmitic acid to sphingomyelin. FA: fatty acid, SM: sphingomyelin, LPE: lyso-O-phosphatidylethanolamine (E) Quantification of de novo sphingomyelin synthesis assay from panel C. Red lines and error bars correspond to mean and standard deviation from 5 experiments, respectively. (F) Simplified representation of Golgi-targeted lipidomics and enzymatic activity assays. (G) Western blot confirming the immunomagnetic isolation of trans Golgi membranes. Cross-reactivity to Protein G and/or Golgin-97 IgG were labeled with asterisks. (H-J) Lipidomics analysis of Golgi isolates show lowered PS levels in the Golgi of ORP9 and ORP11 knockout cells. Same Golgi fractions have reduced sphingomyelin but increased ceramide levels. Each colored line corresponds to a lipid species. Data points and the error bars correspond to mean and standard deviations from 4 experiments. (K) TLC readout of enzymatic activity assay performed in isolated Golgi from knockout cells. Golgi of ORP11 and especially that of ORP9 demonstrate lowered capacity to synthesize sphingomyelin. (L) Quantification of sphingomyelin synthesis activity assay performed in Golgi isolates corresponding to panel I.

Model of ORP9-ORP11 mediated promotion of sphingomyelin synthesis at the ER-trans Golgi contact site.

ORP9 and ORP11 require dimerization with each other to localize at ER-trans Golgi membrane contact sites, where they exchange PS for PI(4)P. Consequently loss of either protein causes PS and PI(4)P imbalances in the Golgi apparatus. Since only ORP9 contains a FFAT motif, its loss leads to a more pronounced effect on PS, PI(4)P and sphingolipid levels. The same contact site also accommodates OSBP and CERT for the anterograde trafficking of cholesterol and ceramide, the latter is used for sphingomyelin production. Cholesterol, sphingomyelin and PS are trafficked to the plasma membrane by vesicular means to be asymmetrically distributed between the leaflets of the plasma membrane.

Plasmids

Antibodies

Glucosylceramide is the main hexosylceramide in MelJuSo cells.

(A) MelJuSo cells silenced for glucosylceramide synthase (siUGCG) or galactosylceramide synthase (siUGT8) or treated with the glucosylceramide synthase inhibitor (PDMP) were fed with the fluorescent ceramide analog BODIPY-C5-ceramide. Lipids were isolated and separated by thin-layer chromatography. Note that the fluorescent band corresponding to monohexosylceramide is affected only when glucosylceramide synthesis is perturbed. (B) Quantifications corresponding to panel A. Bars and error bars correspond to mean and standard deviations from 4 experiments, respectively.

Overview of lipidomics analysis performed for LTP knockout library.

(A) Percentage distribution of 17 lipid classes in LTP knockout cells; each data point represents the mean percentage of an LTP knockout cell line. (B) Box plots of Z-scores of NT control cells. Note that Z-scores for NT control cells do not exceed the absolute value of 2. PC: phosphatidylcholine, LPC: lysophosphatidylcholine, PE: phosphatidylethanolamine, LPE: lysophosphatidylethanolamine, PS: phosphatidylserine, PA: phosphatidic acid, PG: phosphatidylglycerol, PI: phosphatidylinositol, DCer: dihydroceramide, Cer: ceramide, SM: sphingomyelin, HexCer: hexosyl-ceramide, LacCer: lactosylceramide, DG: diacylglycerol, TG: triacylglycerol, CE: cholesterol ester, and FFA: free fatty acid.

GLTP, NPC1 and NPC2 knockout cells do not display major phospholipid imbalances.

(A, B) Lipidomics analysis of GLTP knockouts show lowered glucosylceramide levels that is represented in various acyl chain species. (C) GLTP knockouts do not present altered phospholipid levels. (D) Lipidomics analysis of NPC1 and NPC2 knockout cells do not display major changes in their phospholipid repertoire. (E) Coomassie Blue staining of recombinant EQT-GFP.

Sphingolipid imbalances in CERT, ORP9 and ORP11 knockouts are represented in multiple acyl chain species.

(A) CERT, ORP9 and ORP11 knockouts do not display major phospholipid imbalances. (B) Reduced sphingomyelin and increased glucosylceramide levels in CERT, ORP9 and ORP11 knockout cells is observed in multiple acyl chain species. Lipid levels are normalized to the POPC (16:0/18:1) levels.

CERT, ORP9 and ORP11 knockouts do not display defects in Golgi morphology.

(A) Immunofluorescence staining for cis and trans Golgi markers, GM130 and Golgin-97 respectively, do not show any changes in ORP9, ORP11 and CERT knockouts. Scale bar 10µm. (B) Electron microscopy images for ORP9 and ORP11 knockouts do not display any changes in Golgi morphology. Scale bar 250nm.

ORP9 and ORP11 interact with each other via their coiled coils.

(A) AlphaFold prediction of ORP9 and ORP11 reveals secondary structures between their PH and ORD domains. (B) PCOILS predicts coiled coils in the same of both proteins. Residues corresponding to coiled coils are highlighted. (C) AlphaFold-multimer / ColabFold prediction for ORP9-ORP11 coils dimerization. (D) Predicted local distance difference test (pLDDT) and position aligned error plots for panel C. (E) Predicted local distance difference test for figure 4C. (F) Unique peptide counts from BioID experiments performed with VAPA, VAPB, MOSPD1, MOSPD2, and MOSPD3 from Cabukusta et al. 2020. Both ORP9 and ORP11, despite latter lacking a FFAT motif, detected in BioID experiments with VAPA and VAPB.

ORP9 and ORP11 are needed for maintaining phospholipid levels in the Golgi.

(A) Coomassie Blue staining of recombinant OSBP-related domains of ORP9 and ORP11. (B) ORP9 and ORP11 contain the critical residues for PS transfer. (C) NBD/Rhodamine FRET-based lipid transfer assay using the OSBP-related domains of ORP9 and ORP11, ORD9 and ORD11 respectively, demonstrate the PS trafficking ability between vesicles in vitro. (D) Intracellular localization of GFP-OSBP-PH and RFP-Lact-C2 in MelJuSo cells.

Loss of ORP9 or ORP11 does not affect sphingomyelin synthase levels.

(A) Western blot analysis showing that the protein two human sphingomyelin synthases, SMS1 and SMS2, are not affected in ORP9 and ORP11 knockouts. (B) qRT-PCR showing that mRNA levels of SMS1 and SMS2 are not affected dramatically in knockout cells. Two primer sets per protein each was used (shown with roman numbers). Red lines correspond to mean values. (C) Thin-layer chromatography analysis visualizing the conversion of palmitic acid alkyne to sphingomyelin. To define the band corresponding to sphingomyelin, cells were treated with PDMP or with a cocktail of fumonisin B1 and myriocin prior to alkaline hydrolysis.

CERT is not less present in the Golgi of ORP9 and ORP11 knockouts.

(A) Immunofluorescent staining of ORP9, ORP11 and CERT knockout cells using two antibodies against CERT. Note that antibodies fail to stain the Golgi in CERT knockouts. (B) Quantifications corresponding to panel A. Loss of ORP9 or ORP11 do not reduce CERT localization at the Golgi. Note that loss of ORP11 increases CERT localization at the Golgi, shown by two antibodies. Red lines correspond to mean from 3 experiments.

Immunomagnetic isolation yields to intact Golgi membranes.

(A) Sphingomyelin synthase enzymatic activity assay performed on immunomagnetic Golgi isolates shows that the isolated membranes are intact and enzymatically active. Sphingomyelin synthases contain six transmembrane helices and require intact membrane for activity (Tafesse et al. 2006 “The Multigenic Sphingomyelin Synthase Family” J Biol Chem Volume 281, Issue 40). (B) Western blotting of immunomagnetic Golgi isolations from knockout cells using Golgin-97 antibody. (C) Amount of lipids isolated from samples in the absence or presence of the trans Golgi specific antibody Golgin-97. The plot is based on 16 data points obtained from n=4 experiments. Data points and error bars correspond to mean and standard deviations, respectively. (D) Distribution of PS species. (E) Distribution of sphingomyelin species. (F) Distribution of ceramide species.

Golgi localization of sphingomyelin synthases is not reduced in ORP9 and ORP11 knockout cells.

(A) Immunofluorescence images showing localization of sphingomyelin synthases in ORP9 and ORP11 knockout cells. (B) Quantification of images corresponding to panel A. Note that the localization of SMS1 or SMS2 is not reduced in the knockout cells. An increased localization for SMS1 in the Golgi apparatus in ORP9 knockout cells is detected by two antibodies. Red lines correspond to mean from 3 experiments.