Background
Niemann-Pick disease type C (NPC) is a lysosomal storage disorder caused by accumulation of unesterified cholesterol in cells of the brain, liver, etc., that occurs with an estimated frequency of 1 in 120 000 individuals [
1‐
4]. This value can be underestimated, since atypical phenotypes may not be properly diagnosed [
4]. Cholesterol is internalized by cells from serum mainly as a constituent of low density lipoprotein (LDL) by clathrin-mediated endocytosis. The LDL is directed to lysosomes where it is hydrolyzed and free unesterified cholesterol is released [
5]. The cholesterol is transported from lysosomes to the plasma membrane and the endoplasmic reticulum where it undergoes esterification. Simultaneously,
de novo synthesis of cholesterol and LDL uptake are down-regulated [
2,
6,
7].
The NPC disease is caused by mutations of
NPC1 or
NPC2 genes coding for lysosomal proteins – NPC1 and NPC2. About 95% of NPC cases are linked to mutations in the
NPC1 gene [
8,
9]. NPC1 is a transmembrane lysosomal protein while NPC2 is localized in the lumen of lysosomes [
10,
11]. The NPC1 and NPC2 proteins are engaged in transporting free cholesterol and some accompanying glycolipids from lysosomes to other cellular compartments [
6,
12,
13]. In addition to cholesterol accumulation in lysosomes its synthesis and metabolism are also affected leading to disturbances in the synthesis of steroid hormones and in the assembly of cellular membranes. Predominant symptoms of NPC disease are progressive neurodegeneration and hepatosplenomegaly. The severity of symptoms of NPC disease varies, but typically the disease leads to death in the second decade of life [
3,
14]. The neuropathological lesions in NPC patients can be reduced by application of an inhibitor of glucosylceramide synthase, the main enzyme involved in glycosphingolipid synthesis [
15].
Presently, detection of NPC disease requires skin biopsy, cultivation of fibroblasts and their staining with filipin, a fluorescent polyene antibiotic which binds free cholesterol [
3,
16]. However, this approach requires UV excitation and filipin fluorescence is prone to photobleaching, which constrains its application in NPC diagnostics [
17,
18]. Other methods of NPC diagnosis are also considered [
19]. Recently, a new approach for detection of NPC disease based on LC-MS/MS analysis of oxidized forms of cholesterol in the serum has been proposed [
20], but a wider application of this sensitive and specific method is limited by the availability of the sophisticated equipment.
Alternative visualization of cholesterol deposits in NPC cells could in principle be also based on the application of protein toxins of microbial origin which specifically recognize free cholesterol and can be used as probes for cell staining without the drawbacks of filipin. About twenty toxins produced by Gram-positive bacteria belong to the family of cholesterol-binding cytolysins [
21,
22]. Among such bacterial toxins special attention has been paid to perfringolysin O (PFO) produced by
Clostridium perfringens[
23,
24]. PFO oligomerizes upon binding to membrane cholesterol and leads to pore formation provided the cholesterol content exceeds 30 mol% [
25,
26]. A biotinylated proteolytic derivative of PFO, named BCθ, of molecular mass of 57 kDa, has been used to stain NPC cells. In those studies, cells were fixed with 4% paraformaldehyde and exposed to BCθ without cellular membrane permeabilization, yielding intracellular staining [
27,
28]. Taking into account the relatively high molecular mass of BCθ, the mechanism of its entry into cells and the nature of “the cholesterol-rich domains” detected by BCθ [
27] remain unknown.
In this study we prepared PFO fused with glutathione S-transferase (GST) and used the recombinant protein to detect cholesterol-rich organelles in Triton X-100-permeabilized NPC and healthy-donor cells. The GST-PFO-positive vesicles, found only in NPC cells, also stained with filipin, antibodies against lysosomal-associated membrane protein 1 (LAMP-1) and lysobisphosphatidic acid (LBPA), indicating that the procedure detected accumulation of cholesterol in late endosomes/lysosomes in NPC cells.
Methods
Fibroblast cultures
The studies were performed on fibroblasts derived from NPC patients and from healthy donors. The NPC patients had been diagnosed on the basis of clinical, cytochemical (Department of Genetics, Institute of Psychiatry and Neurology, Warsaw, Poland) and/or molecular parameters (NZOZ Genomed, Warsaw, Poland). To obtain fibroblast cultures, skin biopsies were dissected and cells were grown for up to 7 passages in DMEM medium containing 10% FBS, 50 U/ml penicillin, 50 μg/ml streptomycin, 0.2 μg/ml amphotericin B at 37°C, 5% CO
2. Prior to experiments cells were transferred to DMEM medium with delipidated 10% FBS and cultured for 72 h [
29]. The delipidated FBS was prepared according to Cham and Knowles [
30] using organic solvents (di-isopropyl ether and n-butanol, 40:60, v:v) and filtered throughout a 0.22 μm sterile filter. The procedure removed 96% of the original cholesterol content in FBS, as determined by the Cholesterol DST kit (Alpha Diagnostics). This study was conducted in accordance with the Helsinki Declaration. Ethical approval of the study was requested at the local Bioethical Committee of the Institute of Psychiatry and Neurology in Warsaw, Poland. According to the information obtained from the Committee, the usage of biological material taken from patients, which was excessive after diagnostics procedures, does not demand a special approval on the condition that patients signed an informed consent for such action. Patients gave their signed informed consent for the use of their cells in the scientific experiments after anonimization of the material samples.
For cholesterol depletion, cells cultured in medium containing delipidated 10% FBS were fixed, permeabilized as described further, and exposed for 1 h to 3 mM methyl-β-cyclodextrin (Sigma) in PBS at 37°C. To trigger uptake of LDL, cells were cultured in the presence of delipidated 10% FBS (72 h) and subsequently incubated with 10 μg/ml DiI-LDL (Biomed. Technology) in DMEM/delipidated 10% FBS for 5 h at 37°C.
Preparation of recombinant PFO
A synthetic gene for perfringolysin O of
Clostridium perfringens was prepared by GenScript (USA) basing on cDNA sequence No. CP000246.1 at NCBI. The sequence was optimized for expression in
E. coli. The synthesized gene was devoid of a leader sequence coding for 28 N-terminal amino acids to ensure intracellular accumulation of the expressed protein. The product was cloned by the vendor into pUC57 plasmid. To obtain PFO with a GST tag at the N-terminus, the construct was cloned into pGEX4T vector using BamHI and EcoRI sites. To allow potential removal of the GST tag, the construct was modified by introducing the sequence 5′ GAA AAC CTG TAT TTT CAG GGC 3′ encoding the ENLYFQG motif recognized by Tobacco Etch Virus (TEV) protease [
31].
The recombinant vector was transformed into BL-21 (DE3) strain of
E. coli. Bacteria were grown at 37°C in LB medium containing 100 μg/ml ampicillin to OD = 0.6, when 0.5 mM IPTG was added. The culture was continued at 18°C for 20 h, bacteria were harvested, washed in PBS and lysed in the presence of 0.35 mg/ml lysozyme (10 min, 4°C). The obtained suspension was supplemented with 1% Triton X-100 and sonicated on ice for 15 min at 0.3 cycle, amplitude 33%, using an UP200S Hielscher sonifier (Germany). The lysate was clarified by centrifugation at 20 000 × g for 40 min at 4°C and loaded onto a Glutathione-Sepharose 4B column (bioWORLD). GST-PFO was eluted from the column with 10 mM glutathione, 10 mM DTT, 50 mM Tris, pH 8.0. The GST protein was prepared as described earlier [
32]. The presence and purity of the recombinant proteins in column fractions were examined by 10% SDS-PAGE. Fractions containing highest amounts of GST-PFO or GST were pooled and filtered over Amicon Ultra-15 centrifugal filter units to remove glutathione and concentrate the protein. Samples containing about 0.45 mg/ml GST-PFO or 0.6 mg/ml GST in 5 mM DTT, 50 mM Tris, pH 8.0 were supplemented with 20% sucrose and frozen in liquid nitrogen.
Protein-lipid overlay assay
The analysis was performed essentially as described earlier [
33] with modifications. The following lipids were used: semisynthetic bovine brain sphingomyelin (SM), C16-ceramide, cholesterol, dioleoylphosphatidylcholine (DOPC), dipalmitoylphosphatidylcholine (DPPC) and dipalmitoylphosphatidylethanolamine (DPPE) (all from Sigma). The lipids were dissolved in chloroform:methanol:H
2O (1:1:0.3) and 1 μl of the solution containing 25–500 pmoles of lipid was spotted onto a 0.45-μm nitrocellulose membrane. The membrane was pressed with a hot block at 60°C for 5 s according to Taki and Ishikawa [
34], blocked with 1% gelatin and 1% polyvinylpyrrolidone and incubated for 45 min at 20°C with 1 μg/ml GST-PFO in PBS buffer containing 0.03% Tween-20. After washing, the membrane was exposed for 45 min to chicken anti-GST IgY-peroxidase (Rockland). Immunoreactive spots were visualized with the SuperSignal West Pico chemiluminescence substrate (Pierce).
Carboxyfluorescein release from liposomes
Liposomes composed of (i) cholesterol:SM:DOPC (mol% 40:30:30), (ii) cholesterol:DPPE:DOPC (mol% 40:30:30), (iii) SM:DPPC:DOPC (mol% 40:30:30) or (iv) DPPE:DPPC:DOPC (mol% 40:30:30) loaded with 10 mM 6-carboxyfluorescein were prepared as described earlier [
32] with modifications. Lipids were mixed, dried under nitrogen, resuspended in PBS containing 10 mM 6-carboxyfluorescein and sonicated in nitrogen atmosphere (15 min, 4°C, 0.3 cycle with amplitude 33% in the UP200S Hielscher sonifier). Pelleted liposomes (2000 × g, 10 min) were resuspended in PBS at a total concentration of lipids of 1 mM. Release of 6-carboxyfluorescein from the liposomes was induced by 15 μg/ml GST-PFO and measured at Em/Ex = 490/520 nm on a Spex spectrofluorimeter (Jobin-Yvone). The maximal efflux of 6-carboxyfluorescein was determined in the presence of 0.2% Triton X-100.
Surface plasmon resonance (SPR)
The analysis was performed on large unilamellar vesicles, 100 nm in diameter, containing cholesterol:DOPC, DPPE:DPPC or SM:DPPC (mol% 50:50) and prepared as described by Kulma et al. [
32]. The liposomes were deposited on the L1 chip of a BIACore X apparatus (BIACore, GE Healthcare) at a flow rate of 1 μl/min for 10 min. Binding experiments were performed at 10 μg/ml GST-PFO or 10–60 μg/ml GST in 150 mM NaCl, 30 mM Tris, pH 8.0, at a flow rate of 30 μl/min. After 10 min (binding phase) the samples were washed for another 10 min (dissociation phase).
Determination of total cellular cholesterol
Cells, 3×106/sample, were scraped from dishes after 72 h of culturing in DMEM/delipidated 10% FBS, washed twice with PBS, suspended in 200 μl of hexane:isopropanol (3:2, v:v), sonicated and incubated for 15 min at 20°C with shaking. Extracted lipids were dried under N2 and resuspended in 50 μl isopropanol (45 min, 20°C). Aliquots of 30 μl were mixed with 300 μl of Cholesterol DST reagent and processed according to the manufacturer’s instructions (Alpha Diagnostics). Protein content in “cell skeletons” remaining after lipid extraction was estimated by Bradford Ultra reagent (Expedeon Ltd) and measured at 595 nm.
Determination of chitotriosidase activity
The enzyme activity was measured in samples of blood serum using 4-methylumbelliferyl-β-D-
N,N,N’-triacetylchitotriose as a substrate (Sigma) according to [
35].
Cellular ELISA
Cells were seeded in DMEM/10% FBS on 96-well plates at 5×103 cells/well, unless indicated otherwise. After 8 h, the medium was replaced with DMEM/delipidated 10% FBS and the cultures were grown for 72 h. Then, the cells were washed, fixed in 3% paraformaldehyde in PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 4 mM MgCl2, pH 6.9), washed in PBS buffer, exposed to 50 mM NH4Cl/PBS (5 min, 20°C), washed again and permeabilized with 0.05% Triton X-100/PBS for 10 min at 20°C. In a series of experiments cells were permeabilized with 0.03-0.2% Triton X-100 (10 min, 20°C) or 0.05% digitonin (10 min, 20°C). The detergent was washed out with PBS and the cells were incubated for 45 min at 20°C in 3% fish gelatin/PBS followed by 5 μg/ml GST-PFO in PBS buffer containing 1% fish gelatin (Sigma). When indicated, prior to labeling with GST-PFO cells were incubated with 3 mM methyl-β-cyclodextrin for 60 min at 37°C. After washing 5 times in 1% fish gelatin/PBS, the cells were incubated with 2 μg/ml chicken anti-GST IgY-peroxidase in 1% fish gelatin/PBS (45 min, 20°C), washed 3 times in 1% fish gelatin/PBS and 3 times in PBS alone. The enzymatic activity of bound peroxidase was examined in the presence of 0.05 mg/ml 3,3′,5,5′-tetramethylbenzidine and 0.045% H2O2. After 30 min, the reaction was stopped with 1 M H2SO4 and the absorbance was measured at 450 nm using a Sunrise Plate Reader (Tecan Group). The results were normalized against protein content in samples estimated by Bradford Ultra at 595 nm. Samples were run in parallel with and without treatment with GST-PFO. The values of 450/595 absorbance obtained in samples not-treated with GST-PFO were subtracted from the corresponding values of samples incubated with GST-PFO.
Immunofluorescence studies
Cells (1×10
4/sample) were seeded on coverslips (15×15 mm) in DMEM containing 10% FBS and after 18 h the medium was replaced with DMEM supplemented with delipidated 10% FBS and cultured for 72 h. Cells were washed with PBS, fixed with 3% paraformaldehyde in PHEM buffer (30 min, 20°C) and processed essentially as described [
36]. Briefly, cells were washed, incubated with 50 mM NH
4Cl/PBS for 5 min at 20°C and permeabilized with 0.05% Triton X-100/PBS for 10 min at 4°C. Alternatively, in a series of experiments cells were permeabilized with 0.05% digitonin/PBS for 10 min at 20°C. After washing in PBS, cells were incubated twice with 3% fish gelatin in PBS (30 min each incubation). Next, cells were treated with 5 μg/ml GST-PFO for 45 min at 20°C and washed five times in 0.2% fish gelatin/PBS. To detect the probe, the cells were exposed to goat IgG anti-GST conjugated with biotin (Rockland) prepared in 0.2% fish gelatin/PBS (45 min, 20°C). In studies on colocalization of PFO-stained structures, the anti-GST antibody was accompanied by: (i) rabbit anti-LAMP-1 IgG (Santa Cruz Biotechnology), (ii) rabbit anti-golgin-84 IgG (Santa Cruz Biotechnology), (iii) rabbit anti-protein disulfide isomerase (PDI; Cell Signaling), (iv) mouse anti-peroxisomal membrane protein 70 (PMP70) (Sigma), (v) mouse anti-LBPA IgG (clone 6C4; Echelon), or (vi) phalloidin-FITC (Sigma). Unbound antibodies and phalloidin were washed out five times with 0.2% fish gelatin/PBS and the samples were incubated for 45 min at 20°C either with streptavidin-TRITC (Sigma) and goat anti-rabbit IgG F(ab)
2-FITC (Jackson ImmunoResearch) or goat anti-mouse IgG-FITC (ICN). In some experiments cells were exposed to 25 μg/ml filipin III (Sigma) for 45 min at 20°C in darkness and GST-PFO was detected using goat IgG anti-GST followed by donkey anti-goat-Texas Red (Jackson Immunoresearch). When living cells were exposed to 10 μg/ml DiI-LDL, they were washed after 5 h at 37°C, fixed with 3% paraformaldehyde, permeabilized with 0.05% Triton X-100, incubated with 5 μg/ml GST-PFO followed by the anti-GST-biotin antibody, as described above and exposed to streptavidin-FITC (45 min, 20°C; Sigma). After extensive washing in 0.2% fish gelatin/PBS, samples were mounted in mowiol/DABCO and examined either under a Nikon microscope equipped with a DXM1200C digital camera or under a Leica confocal microscope (TCS SP8 SMD). TRITC and FITC, TRITC and filipin, DiI-LDL and FITC were excited in the mode of sequential excitation to exclude cross-over of their fluorescence. Stacks of 8–10 confocal planes were acquired for each analyzed cell. The setting of photomultipliers were adjusted to obtain comparable ranges of pixel intensity in each channel, scan resolution was 1024×1024. Colocalization analysis was performed on single-plane confocal images using Leica Application Suite AF software which calculated the Perason’s correlation coefficient and the overlap coefficient [
32]. For both signals the intensity threshold value was set at 55% and 20% background subtraction was applied. At least 20 cells from two independent experiments were analyzed for each variant.
Electron microscopy
For the studies, NPC and control healthy fibroblasts were grown in 10-cm Petri dishes to confluence and after washing with PBS, were fixed with 3% formaldehyde/0.5% glutaraldehyde in 100 mM sodium phosphate buffer (pH 7.2) for 30 min at 20°C. Cells were washed twice with the phosphate buffer, treated with 50 mM NH4Cl in the buffer (10 min, 20°C), washed and gently scraped off the dishes. Pelleted cells (2000 × g, 2.5 min) were dehydrated in an ethanol series (20°C) followed by incubations in mixtures of LR White resin/ethanol at ratios 1:1, 2:1 and 3:1, each for 1 h. Finally, the samples were infiltrated with 100% LR White overnight, after which the resin was exchanged twice (1 h, 20°C). The samples were polymerized at 56°C for 48 h. Ultrathin sections were placed on formvar-coated nickel grids and blocked with 3% fish gelatin in PBS and next in a mixture of 3% BSA and 1% polyvinylpyrrolidone in PBS (50 min each incubation). Subsequently, the samples were incubated overnight with 10 μg/ml GST-PFO and rabbit anti-LAMP-1 antibody in 0.2% fish gelatin/PBS in a humid atmosphere. After washing five times with 0.2% fish gelatin/0.05% Tween-20/PBS, the samples were incubated with goat anti-GST IgG-biotin (3.5 h) and after washing, exposed to goat anti-biotin IgG conjugated with 6 nm gold particles and donkey anti-rabbit IgG-10 nm gold (both Aurion) prepared in 0.2% fish gelatin/0.05% Tween-20/PBS (3 h). After extensive washing: six times with 0.2% fish gelatin/0.05% Tween-20/PBS, thrice with PBS and twice with distilled H2O, the samples were counterstained with 2.5% uranyl acetate in 50% ethanol for 15 min in dark, washed with 50% ethanol and distilled H2O, and stained with lead citrate for 2 min. Finally, the samples were washed with distilled H2O, dried and examined under a JEM 1400 (Jeol) electron microscope.
Discussion
Mutations in the
NPC1 and
NCP2 genes disturb the trafficking of cholesterol in cells. These mutations are linked to a wide clinical spectrum of NPC disease and their diversity seems to contribute to the heterogeneity of NPC symptoms that range from severe to mild neurological defects, organomegaly and psychiatric symptoms [
3,
14]. At the cellular level the abnormalities in cholesterol trafficking are manifested by accumulation of free unesterified cholesterol and disturbance in the metabolism of some other lipids, particularly sphingolipids [
6,
12,
47]. At present, cholesterol deposits in NPC cells are detected by staining of cells derived from skin biopsies with the polyene antibiotic filipin [
3,
16,
39].
In this report we demonstrate that cholesterol deposits in fibroblasts from NPC patients can be conveniently visualized using a recombinant bacterial toxin, perfringolysin O fused with a GST tag. GST-PFO stained cholesterol in late endosomes/lysosomes in NPC fibroblasts permeabilized with 0.05% Triton X-100.
GST-PFO can easily be produced in large quantities in
E. coli and purified by simple one-step column affinity chromatography. The probe is highly selective for cholesterol, as we demonstrated by protein-lipid overlay assay and surface plasmon resonance analysis. The recombinant protein preserves the lytic activity of the native toxin and causes an efflux of 6-carboxyfluorescein from cholesterol-containing liposomes, but not from those devoid of cholesterol (Figure
1B). PFO binds to membranes and causes their permeabilization when the cholesterol content exceeds 30 mol% [
25,
26]. Therefore, the deposits of cholesterol in NPC cells make them preferable targets of GST-PFO. On the other hand, even though free cholesterol is present in non-NPC cells as well, its level is too low to lead to any marked staining by GST-PFO. The lytic activity of PFO does not affect the staining of cells which were fixed before the treatment with the probe. The presence of the GST tag allows detection of the probe bound to cholesterol in NPC cells with a wide array of anti-GST antibodies. Depending on the label conjugated with the anti-GST antibody various detection techniques can be used. In our hands, the probe allowed unequivocal detection of cholesterol deposits in cells by immunofluorescence and immunoelectron microscopy, and by cellular ELISA. The latter approach is suitable for screening a large number of samples and offers the possibility of a semiquantitative analysis of free cholesterol accumulated in cells. Cellular ELISA based on GST-PFO binding to cells seems to be able to distinguish and quantify “variant” biochemical phenotypes of NPC cells showing the level of cholesterol accumulation in the cells. Our preliminary data show that the GST-PFO probe is also suitable for detection of free cholesterol by FACS analysis (not shown).
The fluorescence technique is routinely used for visualization of free cholesterol deposits in NPC cells by filipin [
3,
41]. Having a high affinity toward cholesterol, filipin forms fluorescent complexes with the lipid [
17]. However, the use of filipin has some drawbacks compared to GST-PFO. The fluorescence of filipin requires UV excitation, is easily scattered, and undergoes rapid photobleaching [
18]. The use of GST-PFO for labeling does not pose such constraints, the probe can be detected in a wide range of visible or infrared wavelengths, and the fluorochromes available are fairly stable. Owing to these properties, GST-PFO also facilitates studies on the nature of vesicles harboring cholesterol deposits in NPC cells, which is another advantage of GST-PFO over filipin.
In 2003–2004 Chang’s group made an attempt to use a PFO derivative for cholesterol detection in NPC cells. The so-called BCθ probe was obtained by proteolytic digestion of PFO and biotinylation of the complex of the two resulting fragments [
48]. This probe was used to stain a Chinese hamster ovary cell line lacking the NPC1 protein, fibroblasts of NPC patients, and sections of brains of NPC mice. The authors showed that BCθ stained mainly “cholesterol-rich domains” inside the cells [
27,
28]. The probe entered the cell without prior permeabilization with detergents. Instead, it was shown that the cells became permeable to BCθ when fixed with 4% paraformaldehyde. This approach is unusual, especially taking into account the relatively high molecular mass of BCθ (ca. 57 kDa) and streptavidin-Texas Red (forms tetramers of ca. 76 kDa) used for analysis. In these conditions, both the entry of proteins across the plasma membrane, and particularly across the endosomal/lysosomal membrane and washout of excess protein, are impeded. Perhaps for these reasons the BCθ-positive structures were not well distinguished. In our studies, beside fixation, the cells were additionally permeabilized with 0.05% Triton X-100. Omitting the detergent treatment excluded any labeling of the cells with GST-PFO (ca. 78 kDa). On the other hand, 0.05%-02% Triton X-100 did not solubilize the cholesterol deposits in NPC cells, judging from the intense filipin and GST-PFO staining and cellular ELISA analysis. Permeabilization of cells with Triton X-100 increased the accessibility of cholesterol deposits to GST-PFO. Cholesterol deposits were also freely accessible to methyl-β-cyclodextrin which significantly reduced their level in the cells. This effect was not achieved in non-permeabilized cells exposed to cyclodextrin after fixation with 4% paraformaldehyde [
24]. Using permeabilized cells we were able to reveal that vesicles labeled by GST-PFO were also positive for LAMP-1 and LBPA. LAMP-1 is generally accepted as a marker of lysosomes [
49] while LBPA accumulates in the inner membranes of multivesicular bodies [
43]. In contrast to earlier reports on a high degree of colocalization of filipin-stained vesicles with LBPA [
44] in our hands vesicles labeled by GST-PFO displayed significant but not complete overlapping with LBPA staining. This could ensue from technical problems of cell permeabilization which is hard to optimize for both lipids and possibly also from the fact that we work on cells cultured for 72 h in delipidated FBS, which could affect cellular distribution of LBPA. Our data indicate that GST-PFO-positive structures bear markers of late endosomes/lysosomes but not those of endoplasmic reticulum, Golgi apparatus or peroxisomes.
Altogether, the data indicate that GST-PFO is a convenient and reliable probe for detection of cholesterol deposits in cells of NPC patients.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
KK performed confocal and electron microscopy studies, colocalization analysis, participated in drafting and editing the manuscript; EM-S initiated and performed a part of (immuno) fluorescence studies, established conditions for GST-PFO staining of cells; GT and PK cloned GST-PFO and purified the protein, GT also performed cellular ELISA studies; MM and AŁ carried out diagnostics of NPC patients, provided primary cultures of fibroblasts, guided in filipin staining of the cells; MK performed in vitro studies on binding of GST-PFO to cholesterol; AG contributed to studies on LBPA; AS conceived and designed the experiments, analyzed the data, contributed to manuscript writing. All authors read and approved the final manuscript.