Lysobisphosphatidic acid (LBPA) enrichment promotes cholesterol egress via exosomes in Niemann Pick type C1 deficient cells

Niemann-Pick Type C (NPC) disease is caused by loss of function mutations in the NPC1 or NPC2 genes, leading to aberrant trafficking of cholesterol through the late endosomal-lysosomal (LE/LY) system (1). NPC patients develop progressive neurodegeneration and early onset dementia among other neurovisceral complications (2). There is currently a single EMA approved and no FDA approved treatment for NPC (1), thus there is a strong unmet need to develop therapeutic approaches for the treatment of this devastating disease.

Following the endocytosis of low density lipoproteins, cholesterol is found localized primarily in inner membrane lamellae, known as intraluminal vesicles (ILV), within the lumen of the LE/LY compartment (3). Export of cholesterol from the compartment requires NPC2, a small soluble protein residing within the LE/LY interior, and NPC1, a polytopic transmembrane protein found in the LE/LY limiting membrane (4). In patients with NPC1 or NPC2 deficiency the efflux of cholesterol from the LE/LY is dysfunctional, and cholesterol accumulation leads to the interruption of cellular cholesterol homeostasis, secondary accumulation of sphingolipids, and aberrant LE/LY function (5).

In addition to their enrichment in cholesterol, ILV membranes also contain high levels of lysobisphosphatidic acid (LBPA), a structurally atypical phospholipid also known as bis(monoacylglycero)phosphate (BMP) (6, 7). LBPA reportedly comprises 15–20 mol% of total LE/LY phospholipids, and is not detected in other subcellular compartments (8).

We demonstrated that NPC2 and LBPA directly interact, and that LBPA dramatically enhances the cholesterol transport rates of NPC2 (9). Importantly, enrichment of NPC1-deficient fibroblasts with LBPA via incubation with its precursor phosphatidylglycerol (PG) caused a marked reduction in LE/LY cholesterol levels, while incubation with phosphatidylcholine (PC) had no effect (Fig. S1) (9). Cholesterol accumulation was also diminished in NPC1 deficient cells treated directly with LBPA (7). PG/LBPA enrichment of NPC2-deficient cells did not lead to cholesterol clearance (9), indicating that NPC2 and LBPA work together to effect cholesterol movement through the LE/LY compartment via a process that can apparently bypass NPC1. The mechanism underlying this NPC2-dependent, NPC1-independent cholesterol egress from the endo-lysosomal compartment is at present unknown. Here we demonstrate that NPC1-independent cholesterol egress in PG/LBPA enriched cells occurs, at least in part, via increased exosomal secretion.

Exosomes are a subset of extracellular vesicles (EV) that are derived from ILVs within LE (also named multivesicular bodies, MVBs), and are released after fusion of LE/MVBs with the plasma membrane (PM) (10). Exosomes can eliminate toxic protein aggregates and lipids in neurons, fibroblasts and other cell types in lysosomal storage disorders (LSD) including NPC1 disease, as well as in other neurodegenerative disorders (11, 12). LBPA has been shown to play a role in formation of the ILVs within the LE/MVBs (6, 13). This prompted us to investigate whether PG/LBPA enrichment affects exosome formation in NPC1 deficient cells, and whether this serves to increase the export of cholesterol from the cell.

We probed Western blots of exosome-enriched preparations from the culture supernatants of untreated and PG-treated human primary NPC1 fibroblasts (GM03123), and found that 24h PG treatment increased the amount of three classical exosomal markers, Alix, Flotillin-1, and CD63, in the exosome fraction by approximately 2- to 5-fold in FBS-free medium (Fig. 1A), and up to 2-fold in media containing exosome-depleted FBS (Fig. S2A), indicating an increase in exosomal release. Isolated EVs were devoid of calnexin, an abundant ER resident protein, confirming purity of the exosome fraction (Fig. 1A).

Figure 1. Treatment of NPC1 deficient cells with PG increases cholesterol secretion in exosomes.

(A) Quantitative analysis of Western Blots of exosomal markers in isolated exosomes from media of NPC1-deficient fibroblasts. Cells were grown in FBS-free medium for 24h post-PG treatment. Total protein extract (20μg) was used as Western Blot control. Bar graph represents exosomal marker levels normalized to total amount of protein in parental cells. Each individual point represents an independent exosome-enriched preparation. Data from 8–10 experiments.

(B) Electron microscopy of untreated and PG treated NPC1 fibroblasts grown in FBS-containing medium at 24h post-treatment. Representative electron micrographs of whole cells (A, B) show MVB distribution. Arrows indicate MVB. Scale bars, 2μm. C-G represent enlarged images of areas with MVBs in different cells. Scale bars, 500nm. Am- amphisome, EL- endolysosome, N- nuclei. Bar graph represents the number of peripheral MVBs/cell in N=12–13 cells/condition.

(C) Amount of cholesterol in isolated exosome fraction in WT and NPC1-deficient fibroblasts grown in serum-free medium for 24h w/wo PG. Cholesterol in each sample was normalized to the total protein in parental cells. Absolute amount of cholesterol in exosomes in untreated cells is ~0.46 ± 0.18 ug/mg protein in parental cells which constitutes 1.4% of total cellular cholesterol (~34.3 ± 3.1 ug/mg protein). See also Fig. S1D. Each individual point represents an independent exosome preparation. Data from 8 experiments.

(D) Cell mask and (E) filipin imaging of NPC1 fibroblasts treated with PG alone (24h), GW4869 (10μM) alone (24h) or GW4869 and PG together (24h). The CTR and PG only groups were additionally treated with DMSO (as vehicle for GW4869). Arrows indicate patches. Dashed green lines mark cell outlines. All images taken at the same settings. Due to the very high fluorescence intensity of filipin-positive patches at the PM in GW4869 treated cells, images were captured at low intensity, therefore the cholesterol signal in the LE/LY area is not prominent. Cropped perinuclear LE/LY areas are shown at higher magnification in insets, demonstrating the reduction in sterol accumulation upon PG treatment (as shown in (9) and Fig. S1B). Scale bar, 30μm.

(F) Cholesterol in the isolated exosome fraction in NPC1-deficient fibroblasts pretreated with 10μM GW4869 for 12h and then treated with PG for additional 24h. CTR are the same cells not treated with either GW4869 or PG. The CTR and PG only groups were additionally treated with DMSO.

(G) Time course of mRNA expression of ABCA1 cholesterol transporter in NPC1-mutant fibroblasts. N=3–6. Data from two independent experiments.

(H) LC-MS analysis of LBPA molecular species in untreated and PG-treated NPC1-mutant fibroblasts. N=2.

All graphed data show mean ± SD *: p<0.05, **:p<0.01, ***:p<0.001, compared to untreated cells (CTR) #: p<0.05 compared to PG treated cells in two-tailed Student’s test. PG concentration in all experiments 100μM.

Consistent with these biochemical results, electron microscopy (EM) revealed substantial > 60% increase in the number of LE/MVBs in close proximity to the PM in PG-treated NPC1-deficient fibroblasts (Fig. 1B). In untreated cells, by contrast, larger numbers of MVBs were, rather, found fused with autophagosomes to form amphisomes (Am), fused with LY to form endo-lysosomes (EL), or were detected in the perinuclear area (Fig. 1B).

Immunofluorescent imaging of luminal LAMP1, a marker of the endo-lysosomal compartment, in non-permeabilized NPC1 mutant fibroblasts, revealed a 6 to 10-fold increase in LAMP1-positive organelles near the cell surface at 3 and 6 hours of PG incubation, suggesting their exocytosis at early time points after treatment (Fig. S2B). Indeed, the increase in LBPA levels following PG incubation has been shown to occur at very early time points (3h) following PG incubation (14). Live-cell imaging of exocytic vesicles in PG-treated NPC1 KO HeLa cells, using the Lyn-SuperEcliptic (SE) pHluorin/mCherry reporter, a fusion of the membrane-targeted green, pH-sensitive fluorescent protein with the red, pH-insesitive mCherry, showed dynamic movement of mCherry-positive organelles and frequent exocytosis as indicated by the appearance of GFP-fluorescence at the PM (Fig. S2C), even at only 1h post-PG treatment. Collectively these data suggest that exocytosis is increased rapidly following PG treatment.

Exosomes are known to contain large amounts of cholesterol (10), thus we analyzed the cholesterol content of exosomes in control and PG-treated NPC1-deficient fibroblasts. For these experiments we used FBS-free medium to eliminate cholesterol from the serum. We found that exosomes from PG-treated cells contained on average 5-fold more cholesterol relative to the exosomes from untreated control NPC1-deficient fibroblasts (Fig. 1C). While total exosome cholesterol and exosome protein markers were comparably elevated, suggesting that PG-treated cells secrete more exosomes rather than more cholesterol per exosome, the present analysis does not allow for an assessment of variability in exosome number, size, and morphology.

GW4869 is a common inhibitor of exosome release and was used to determine whether it would affect cholesterol secretion in NPC1 deficient fibroblasts. To visualize the PM we stained live cells with the PM marker Cell Mask (Fig. 1D), and to detect cholesterol we stained fixed cells with filipin (Fig. 1E). Cell Mask staining revealed patches on the PM of GW4869-treated cells that appeared to be filipin-positive. We note that GW4869 did not affect cell viability (Fig. S2D), and therefore these highly fluorescing filipin-positive patches at the cell surface are unlikely to be adherent cellular debris. Thus, GW4849 treatment, by blocking MVB trafficking and exosome release, appears to result in a remarkable accumulation of cholesterol-laden vesicles/organelles at the PM, both in control and PG-treated cells. We also found that the amount of cholesterol is reduced in exosomes released from cells that were first pre-treated with GW4869 for 12h and then incubated with PG in the presence of GW4869 for an additional 24h, relative to cells incubated with PG but not treated with the compound (Fig. 1F). These data therefore support an exosomal route of cholesterol egress.

We further investigated the mRNA expression of ATP-binding cassette transporter A1 (ABCA1), a PM protein that promotes cholesterol efflux by a mechanism unrelated to exosome secretion (15), and found that ABCA1 mRNA was significantly reduced following 24 h of PG treatment (Fig. 1G). This result is consistent with previously reported effects of LBPA accumulation on ABCA1 levels in macrophages (16), and indicates that the PG-induced decrease in cholesterol content is ABCA1-independent.

Finally, polyunsaturated phospholipid species have been shown to promote membrane vesiculation (17), and LBPA with unsaturated acyl chains was shown to induce the formation of multivesicular structures, in contrast to saturated LBPA species (6). Notably, LC-MS analysis of LBPA molecular species in diC18:1-PG supplemented cells revealed a dramatic increase in LBPA containing the polyunsaturated fatty acid 22:6 (C40:7 or C18:1,22:6) in addition to the expected increase in diC18:1-LBPA (C36:2 in Fig. 1H).

Exosomal secretion occurs following fusion of the MVB/LE with the PM, and LBPA has been shown to play a role in the biogenesis of the ILV within MVB/LEs (6, 18). A role in exosome biogenesis is also supported by a recent study suggesting LBPA as a potential biomarker for urinary exosomes (19). Notably, we showed that LBPA enrichment is completely ineffective in NPC2 deficient cells (9), and it was recently reported that NPC2 is necessary for efficient extracellular vesicle secretion (20), supporting the importance of LBPA-NPC2 interactions in EV formation.

Total cell cholesterol levels, assessed biochemically, were modestly decreased by 22% following 24h of PG treatment (Fig. S1C), while cholesterol accumulation in the LE/LY compartment as determined by filipin staining showed a more substantial 2-fold reduction (Fig. S1A). This underscores the importance of intracellular sterol redistribution during amelioration of the NPC phenotype, and is in agreement with results observed following treatment with 2-hydroxypropyl-β-cyclodextrin (21). Using cells grown in FBS-free medium for technical reasons noted above, we found that in untreated cells the exosome fraction contains on average 1.4% of cellular free cholesterol, and that this increases by 5-fold to ~7% after 24h of PG treatment (Fig. 1C). Thus, since the decrease in total cholesterol in cells grown in serum free media was ~11% (Fig. S1D), EV appear to account for a majority of cellular cholesterol reduction at 24h of treatment. While this is a relatively small effect on the total cellular level of cholesterol and longer treatment times may show further improvement, it is noteworthy that even at 24h the PG/LBPA enrichment induces cholesterol redistribution, thus relieving the ‘traffic jam’ in NPC1 deficient cells. Indeed, exosome production is thought to be critical for maintaining the integrity of the endosomal pathway in neurodegenerative diseases (12).

These findings reveal an NPC1-independent route of cholesterol egress from the LE, via exosome secretion, in PG/LBPA enriched cells. Interestingly, we found marked increases in the amount of polyunsaturated LBPA species, even though cells were supplemented with di-oleoyl PG. As noted, phospholipids containing C22:6 are reported to promote membrane vesiculation (17). Polyunsaturation of LBPA acyl chains would increase its cone-shaped structure and thereby favor membrane invagination and the formation of ILVs. Thus, the increase in C18:1,22:6 LBPA species in PG-treated cells may contribute to the formation of ILV, leading to increased exosome secretion and, hence, increased cholesterol secretion.

In summary, PG treatment of NPC1 deficient cells leads to a reduction in LE/LY cholesterol accumulation, at least in part via exosomal egress. While we assume the actions secondary to PG treatment are due to its metabolite LBPA (or, BMP), we cannot yet exclude independent or additive effects of PG itself. Future studies will use non-hydrolysable forms of PG to distinguish effects of the PG precursor vs. those of the LBPA product.

Highlights.

• Incubation of NPC1 deficient cells with phosphatidylglycerol (PG) increases lysobisphosphatidic acid (LBPA) levels and leads to cholesterol clearance from the endolysosomal compartment.

• Marked enrichment of LBPA species containing polyunsaturated acyl chains is observed following di‐oleoyl PG treatment.

• PG treatment/LBPA enrichment stimulates the secretion of exosomes, and these contain increased cholesterol compared to untreated cells.

• PG treatment/LBPA enrichment clears cholesterol from the late endosomal/lysosomal compartment, at least in part, by stimulating the secretion of extracellular vesicles containing cholesterol.