PIP4K2A regulates intracellular cholesterol transport through modulating PI(4,5)P2 homeostasis

The transport of LDL-derived cholesterol from lysosomes to peroxisomes is facilitated by membrane contacts formed between the lysosomal protein synaptotagmin VII and the peroxisomal lipid phosphatidylinositol 4, 5-bisphosphate [PI(4,5)P2]. Here, we used RNA interference to search for regulators of PI(4,5)P2 and to study the effects of altered PI(4,5)P2 homeostasis on cholesterol transport. We found that knockdown of phosphatidylinositol 5-phosphate 4-kinase type-2 α (PIP4K2A) reduced peroxisomal PI(4,5)P2 levels, decreased lysosome-peroxisome membrane contacts, and increased accumulation of lysosomal cholesterol in human SV-589 fibroblasts. Forced expression of peroxisome-localized, kinase-active PIP4K2A in the knockdown cells reduced cholesterol accumulation, and in vitro addition of recombinant PIP4K2A restored membrane contacts. These results suggest that PIP4K2A plays a critical role in intracellular cholesterol transport by upregulating PI(4,5)P2 levels in the peroxisomal membrane. Further research into PIP4K2A activity may inform future therapeutic interventions for managing lysosomal storage disorders.

Here, we performed a small-scale RNAi screening to search for regulators of peroxisomal PI(4,5)P 2 . We found that disruption of PIP4K2A, a PI(5)P-kinase, caused robust cholesterol accumulation in lysosomes. Meanwhile, reduced lysosome-peroxisome membrane contacts (LPMCs) were detected in PIP4K2A-knockdown cells as revealed by immunostaining, proximity-dependent biotinylation assay, and in vitro reconstitution assay. Interestingly, PIP4K2A deficiency also reduced the peroxisomal PI(4,5)P 2 level and this impairment was successfully reverted by reexpression of the wild-type or peroxisome-anchoring form of PIP4K2A. Taking these data together, we conclude that PIP4K2A regulates LMPC and cholesterol transport through modulating the homeostasis of PI(4,5)P 2 on peroxisomes.

Reagents
The anti-LAMP1 (H4A3) antibody was purchased from Developmental Studies Hybridoma Bank. The anti-PMP70 antibody, filipin, and D-biotin were purchased from Sigma. ALLN was purchased from Calbiochem. Fluorophore-conjugated secondary antibodies were purchased from Invitrogen. The anti-PI(4,5)P 2 antibody and PI(4,5)P 2 standards were purchased from Echelon Biosciences.
Cell culture SV589 and HEK293T cells were cultured in DMEM supplemented with 10% FBS and 100 units/ml penicillin and 100 g/ml streptomycin sulfate. Cells were grown at 37°C with 5% carbon dioxide.

Immunofluorescence
Cells grown on coverslips were washed with PBS and fixed with 4% paraformaldehyde for 30 min at room temperature. Cells were then permeabilized with 0.1% Triton X-100 for 10 min, blocked in 3% BSA in PBS for 1 h, and incubated overnight at 4°C with 3% BSA in PBS containing primary antibodies in 1:1000 dilution (for anti-LAMP1 antibody) or 1 g/ml (for anti-PMP70 antibody). Secondary antibodies were applied in 1:1000 dilution (2 g/ml) for 1 h at room temperature. Slides were coverslipped with FluorSave mounting medium (Millipore) and dried at room temperature (11).

Filipin staining
Cells were washed and fixed as indicated previously (12). Fixed cells were incubated with PBS containing 10% FBS, 50 g/ml filipin, and primary antibodies for 1 h at room temperature. Cells were then incubated with secondary antibodies diluted in PBS containing 10% FBS and 50 g/ml filipin for 1 h at room temperature. Slides were coverslipped as previously described.

Generation of CRISPR-Cas9-mediated PIP4K2A-knockout cell line
Guide RNA targeting the first exon of the human PIP4K2A gene (sequence TGGCGACCCCCGGCAACCTA) was designed using the CRISPR Design website (http://crispr.mit.edu) and cloned to pX330-U6-Chimeric-bb-CBh-SpCas9 vector. The guide RNA-containing constructs were cotransfected with a puromycin resistant expression plasmid. Cells were selected with 2 g/ml puromycin for 4 days and seeded onto 96-well plates. Colonies from single cells were expanded after 10 days. Genomic regions flanking the targeted regions were amplified by PCR and sequenced.

Analysis of SREBP-2 cleavage
Wild-type and PIP4K2A-knockout SV-589 cells were seeded in 60 mm petri dish in triplicate. On the second day, cells were switched to medium A (DMEM containing 5% lipoproteindeficient serum, 1 M lovastatin, and 10 M mevalonate) for 16 h. Cells were then treated with LDL at indicated concentrations for 4 h, followed by 1.5 h incubation in the presence of 25 g/ml ALLN as described previously (13). Cells were harvested in RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% SDS, 1.5% NP-40, 0.5% deoxycholate, and 2 mM MgCl 2 ) plus protease inhibitors. Cell lysates were subjected to standard Western blotting.

Measurement of PM cholesterol
The measurement of plasma membrane (PM) cholesterol was performed as previously described (8) with minor modifications. In brief, wild-type and PIP4K2A-knockout SV-589 were seeded in 6-well plates in duplicate. On the second day, cells were switched to medium A and grown for 16 h. Cells were treated with DMEM plus 3.5% (w/v) hydroxypropyl--cyclodextrin for 15 min and then cultured in LDL-containing medium A for 4 h. Cells were incubated with or without 1 U/ml cholesterol oxidase. Cholesterol was extracted with chloroform-methanol (2:1), dried under nitrogen, and measured with an Amplex Red Cholesterol Assay Kit (Invitrogen) according to the manufacturer's instructions.

Generation of shPIP4K2A-stably expressing cell line
Human PIP4K2A-targeting oligo (sequence GCACTTCG-TAGCGCAGAAAGT) was inserted into pLKO.1 vector. HEK293T cells were cotransfected with pLKO.1-shPIP4K2A, psPAX2, and pMD2.G in a mass ratio of 4:3:1. The medium was replaced on the second day of transfection, and the supernatant was collected 48 h posttransfection. The lentiviral particle-containing supernatant was mixed in an equal volume with fresh medium to infect HeLa cells. Cells stably expressing shRNA cassette were selected and maintained in culture medium containing 1 g/ml puromycin.

Proximity-dependent biotinylation assay
The BirA* fragment carrying an R118G mutation was described previously (14). The plasmid encoding NPC1-BirA*-EGFP fusion protein was constructed with standard cloning methods using the pEGFP-N1 plasmid as a backbone. HEK293T cells stably expressing NPC1-BirA*-EGFP fusion protein were transfected with indicated siRNAs using RNAiMAX (Invitrogen). Forty-eight hours after transfection, cells were incubated in 10% FBS/DMEM either with or without 50 M biotin for 24 h. Total proteins were extracted in RIPA buffer containing protease inhibitors. Whole cell lysate was dialyzed against RIPA to remove residual free biotin. Biotinylated proteins were precipitated with NeutrAvidin agarose beads (Invitrogen). Pellet fractions were isolated by boiling beads in Laemmli buffer and separated by SDS-PAGE.

Lipid dot blot
Peroxisomes and lysosomes from SV589 cells transfected with indicated siRNAs were purified with either a peroxisome or lysosome isolation kit (Sigma Aldrich) according to the manufacturer's instructions. Acidic lipids from organelles were extracted with methanol-12N HCl (10:1) followed by adding chloroform at a volumetric ratio of 2:1. The organic phase was separated by centrifugation at 13,200 g for 10 min at room temperature, collected, and dried under nitrogen. Pellets were resuspended for lipid blot analysis. A Hybond-C nitrocellulose membrane was spotted with extracted lipids, dried, blocked with 3% BSA, and blotted with the antibody against PI(4,5)P 2 .

In vitro reconstitution of LPMC
The in vitro reconstitution assay was performed as previously described (8). In brief, lysosomes and peroxisomes were isolated from HeLa cells stably expressing PEX3-EGFP-His 6 transfected with either scramble shRNA or PIP4K2A-targeting shRNA. Peroxisomes were pulled down with Ni-NTA, washed with a reconstitution buffer (250 mM sucrose, 1 mM DTT, 1 mM MgCl 2 , 50 mM KCl, and 20 mM HEPES pH 7.2) containing 2 mM EGTA, and incubated with lysosomes in the presence or absence of 1 mg/ml cytosol or 50 g/ml purified Flag-PIP4K2A at 37°C for 30 min. Ni-NTA beads were spun down, boiled with sample buffer, and subjected to Western blotting.

Purification of recombinant PIP4K2A
HEK293T cells were seeded in 15 cm petri dish and transfected with 12 g Flag-PIP4K2A plasmid per dish. Forty-eight hours after transfection, cells were scraped, washed with ice-cold PBS, and resuspended in immunoprecipitation (IP) buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, and 1% Triton X-100) containing protease inhibitors. The lysate was cleared by centrifugation at 13, 200 g at 4°C for 10 min. Supernatant was incubated with anti-Flag M2 resin (Sigma) on a rotator at 4°C for 4 h. Then the M2 beads were spun down and washed extensively with IP buffer. Bound proteins were competitively eluted with 0.1 mg/ml 3×Flag peptide, and eluate was collected and dialyzed against PBS. Protein concentration was determined by BCA assay (Pierce), and protein purity was assessed by SDS-PAGE followed by Coomassie Brilliant Blue staining.

PIP4K2A is required for intracellular cholesterol transport
The homeostasis of PI(4,5)P 2 is controlled by kinases and phosphatases as shown in Fig. 1A. To identify key enzyme(s) regulating peroxisomal PI(4,5)P 2 and subsequently, cholesterol transport, we individually knocked down each gene using siRNA in SV589 cells ( Table 1). The RNAi efficiency is shown in Fig. 1B. Among the PI(4,5)P 2metabolizing genes, only cells expressing siRNA against PIP4K2A exhibited significant perinuclear cholesterol accumulation, resembling the phenotype induced by NPC1 or ABCD1 deficiency (Fig. 1C, D) (8). These cholesterolrich puncta colocalized with late endosome/lysosome marker LAMP1 (Fig. 1E). PIP4K2A catalyzes the conversion of PI(5)P to PI(4,5)P 2 . According to previous studies, PIP4K2A is distributed mainly in the cytosol and partially in the nucleus (15). It has the highest catalytic activity among all three isoforms. PIP4K2A is responsible for clearance of PI(5)P upon oxidative stress (16) and its mutations are associated with leukemia (17)(18)(19) and schizophrenia (20)(21)(22).
To further confirm that lysosomal cholesterol accumulation resulted from loss of PIP4K2A, we generated PIP4K2Aknockout SV589 cells using the CRISPR/Cas9 technique. Similarly, these PIP4K2A-knockout cells showed drastic cholesterol accumulation (Fig. 1F). To analyze the effects on downstream cholesterol transport, we monitored the inhibition of SREBP-2 cleavage. The PIP4K2A-knockout cells exhibited delayed inhibition of SREBP-2 cleavage after LDL addition, suggestive of a lowered level of cholesterol arriving at the endoplasmic reticulum (ER) (Fig. 1G). In addition, the cholesterol level on the PM was decreased in PIP4K2A-knockout cell upon LDL treatment (Fig. 1H).

PIP4K2A deficiency reduces LPMCs
Because peroxisomal PI(4,5)P 2 is required for LPMC, we next sought to investigate whether PIP4K2A deficiency decreased lysosome-peroxisome association. Indeed, a more than 30% reduction in LPMCs was observed in PIP4K2A-knockdown cells ( Fig. 2A, B). We further developed a proximity-dependent biotinylation assay to analyze the interorganellar contacts. BirA* is a biotin ligase that covalently attaches biotin to the nearby lysine residues when cells are supplemented with extra biotins (14). We thus generated an HEK293T cell line stably expressing NPC1 fused with BirA* and enhanced green fluorescent protein (EGFP), allowing the spatial distance between lysosome and peroxisome to be determined by monitoring the biotinylation of peroxisomal proteins (Fig. 3A). The NPC1-BirA*-EGFP fusion protein was successfully targeted to lysosomes, as evidenced by the colocalization of EGFP with LAMP1 (Fig. 3B). We first validated the peroxisome biotinylation assay in SYT7-knockdown cells where the close apposition between lysosomes and peroxisomes is known to be impaired (8). PMP70 is a peroxisome membraneintegrated protein and belongs to the superfamily of ABC transporters. PMP70 (also named ABCD3) is one of the major components of peroxisomal membranes and is responsible for transport of long chain fatty acyl-CoA across the peroxisomal membrane (23). We then used biotinylated PMP70 as a readout of the proximity between lysosomes and peroxisomes. We observed reduced levels of biotinylated PMP70 when SYT7 was deficient, suggestive of an increased spatial distance between lysosomes and peroxisomes. The biotinylation of gp78, an ER membraneanchored protein, remained unchanged, suggesting that the contacts between lysosomes and ER were not affected (Fig. 3C). With this technique, we then analyzed the effect of PIP4K2A on LPMC. The results showed that silencing PIP4K2A reduced the biotinylated PMP70 level to about one-third of control cells without altering the biotinylation of the NPC1-BirA*-GFP fusion protein itself (Fig. 3D, E). These data suggest that PIP4K2A regulates the membrane contacts between lysosomes and peroxisomes in vivo.

Peroxisomal PIP4K2A is required for LPMCs in vitro
To directly show that PIP4K2A in peroxisomes is required for LPMC, we purified peroxisomes and lysosomes from control or PIP4K2A-knockdown cells and performed the in vitro reconstitution assay as shown previously (8) (Fig. 4A). In the presence of cytosol and ATP/GTP, lysosomes and peroxisomes from control shRNA-expressing cells formed tight associations with each other (Fig. 4B, lane 1). However, the peroxisomes from PIP4K2A-knockdown cells failed to pull down lysosomes (Fig. 4B, lane 2). On the contrary, the lysosomes from PIP4K2A-knockdown cells were still in association with peroxisomes (Fig. 4B, lane 3). Furthermore, we purified recombinant Flag-tagged PIP4K2A from HEK293T cells (Fig. 4C) and applied it to the in vitro reconstitution system. Addition of the recombinant protein successfully rescued LPMC when the peroxisomes from PIP4K2Aknockdown cells were incubated with the lysosomes from control cells (Fig. 4D, compare lane 5 with lane 4). Together, these results demonstrate that PIP4K2A in peroxisomes, rather than in lysosomes, is required for LPMC formation.
We next transfected PIP4K2A-deficient cells with constructs encoding various forms of PIP4K2A followed by filipin staining (Fig. 5B). Reexpressing of the wild-type PIP4K2A tagged with green fluorescent protein (GFP), as well as the peroxisome-anchoring, kinase active PEX3-PIP4K2A-GFP, completely reversed cholesterol accumulation in PIP4K2Aknockout cells. However, delivery of the construct harboring a kinase-dead mutation failed to rescue the phenotype. Interestingly, a schizophrenia-associated, kinase-active mutation N251S (24) did not reverse cholesterol accumulation. Notably, cells expressing PIP4K2A fused to the carboxyl terminus of NPC1 still exhibited robust lysosomal cholesterol accumulation, proving that PIP4K2A does not modulate cholesterol transport by altering lysosomal PI(4,5)P 2 level. Collectively, these data strongly support that PIP4K2A is required for LPMC and intracellular cholesterol transport through modulating peroxisomal PI(4,5)P 2 level.

DISCUSSION
In this study, we discovered that the PI(5)P kinase PIP4K2A is actively involved in the regulation of intracellular cholesterol transport. There are several lines of evidence supporting that PIP4K2A is required for cholesterol transport from lysosome to peroxisome. First, disruption of PIP4K2A results in cholesterol accumulation in lysosomes (Fig. 1). Second, knockdown of PIP4K2A reduces LPMC as evidenced by fluorescent microscopy (Fig. 2), proximitydependent biotinylation assay (Fig. 3), and in vitro reconstitution assay (Fig. 4). Third, knockdown of PIP4K2A greatly reduces peroxisomal, but not lysosomal, PI(4,5)P 2 levels. Reexpression of peroxisome-anchored, but not lysosome-anchored, PIP4K2A ameliorates cholesterol accumulation in PIP4K2A-knockdown cells (Fig. 5). Fourth, Fig. 3. Silencing PIP4K2A reduced lysosome-peroxisome membrane contacts as measured by proximity-dependent biotinylation assay. A: Schematic illustration of proximity biotinylation assay. B: Colocalization of NPC1-BirA-GFP (green) and lysosome marker LAMP1 (red). C: Validation of proximity-dependent biotinylation assay. HEK-293T cells stably expressing NPC1-BirA-GFP fusion protein were transfected with indicated siRNAs and incubated with biotin-containing culture medium. Then the cells were lysed and biotinylated proteins were pulled down with avidin beads and subjected to Western blotting. D: Proximity-dependent biotinylation assay in control and PIP4K2A-silencing cells. Experiments were performed as described in C. E: Quantification of gray density of PMP70 in D. Triplicate pellet fractions were normalized with input and quantified by Image J.
peroxisomes, but not lysosomes, purified from PIP4K2Aknockdown cells displayed reduced LPMC in vitro (Fig. 4). These data collectively suggest that PIP4K2A is involved in intracellular cholesterol transport through regulating PI(4,5)P 2 homeostasis on peroxisomes.
PIP4K2A has been identified as one of the three isozymes responsible for the conversion of PI(5)P to PI(4,5)P 2 . In fact, the catalytic activity of PIP4K2A is far more active than that of the 2B and 2C isoforms (15,25). The PIP4K2B and PIP4K2C modulate PI(5)P by recruiting high-activity isoform PIP4K2A (15). Interestingly, we did not observe noticeable cholesterol accumulation upon PIP4K2B or PIP4K2C knockdown, suggesting that PIP4K2A-mediated peroxisome PI(4,5)P 2 generation is independent of the other two isoforms. The genetic polymorphisms of PIP4K2A are associated with chronic lymphoblastic leukemia (CLL) and schizophrenia. These genetic variants can alter the binding of transcriptional factors and reduce the protein level of PIP4K2A (17). We hereby find that lowered PIP4K2A decreases LPMC and impairs cholesterol transport. In addition, it has been reported that CLL patients exhibit lower serum cholesterol (26). The N251S variant of PIP4K2A was found to associate with schizophrenia without affecting the enzyme activity. We showed that PIP4K2A(N251S) failed to reverse cholesterol accumulation phenotype in PIP4K2A-knockdown cells (Fig. 5B). Taken together, these data suggests it is possible that cholesterol might play a role in the development and progression of CLL and schizophrenia.
NPC disease is one of the lysosome storage disorders (LSDs)with massive cholesterol accumulation in lysosomes. It is also known that peroxisomal disorders are accompanied with cholesterol accumulation in lysosomes (8,9), demonstrating that peroxisome plays an important role in cholesterol transport from lysosomes. Although the regulation of PIP4K2A is not well characterized, there are some clues in previous studies suggesting PIP4K2A may be involved in LSDs. For example, reactive oxygen species are increased in the cells from LSDs (27,28). Reactive oxygen species can impair PIP4K2A expression through modulating nuclear factor erythroid 2-related factor 2 and its downstream genes (16), and therefore exacerbates the lysosomal storage phenotypes.
In summary, our work demonstrates that PIP4K2A is involved in cholesterol transport from lysosomes to peroxisomes through modulating the PI(4,5)P 2 homeostasis on peroxisomes. Enhancing PIP4K2A activity might serve as a potential therapeutic approach for treating LSDs such as NPC disease. Fig. 4. Peroxisomal PIP4K2A was required for lysosome-peroxisome membrane contacts in vitro. A: Schematic illustration of in vitro reconstitution assay. Peroxisomes or lysosomes from HeLa cells stably expressing either scramble or PIP4K2A-targeting shRNA were purified as previously described (8). Peroxisomes were pulled down with Ni-NTA and further incubated with purified lysosomes. Ni-NTA resin was spun down, washed, and the associated lysosomes were detected by Western blotting. B: In vitro reconstitution of lysosome-peroxisome membrane contacts. Lysosomes and peroxisomes were purified from indicated cells and subjected to in vitro reconstitution illustrated in A. C: Purified recombinant Flag-PIP4K2A from HEK-293T cells was resolved on SDS-PAGE and stained with Coomassie Brilliant Blue. D: In vitro lysosome-peroxisome membrane contacts rescued by the recombinant Flag-PIP4K2A protein. Reconstitution assay was performed with or without 1 mg/ml cytosol or 50 g/ml Flag-PIP4K2A protein. Experiments were performed as described in A.

Fig. 5.
PIP4K2A affected peroxisomal PI(4,5)P 2 level. A: PI(4,5)P 2 dot blot. Lipids were extracted from indicated fractions of cells. Extracts containing equal amounts of protein from either lysosomes or peroxisomes were dotted in 2-fold dilution on a Hybond membrane. PI(4,5)P 2 standards were dotted as indicated. Blots were detected with the anti-PI(4,5)P 2 antibody. B: Rescue experiments using various PIP4K2A constructs. PIP4K2A-knockout cells were transfected with indicated plasmids and stained with filipin. Cells were outlined in white dashed lines. Representative pictures from each experiment were shown.