Characterization of the Sterol and Phosphatidylinositol 4-Phosphate Binding Properties of Golgi-Associated OSBP-Related Protein 9 (ORP9)

Oxysterol binding protein (OSBP) and OSBP-related proteins (ORPS) have a conserved lipid-binding fold that accommodates cholesterol, oxysterols and/or phospholipids. The diversity of OSBP/ORPs and their potential ligands has complicated the analysis of transfer and signalling properties of this mammalian gene family. In this study we explored the use of the fluorescent sterol cholestatrienol (CTL) to measure sterol binding by ORP9 and competition by other putative ligands. Relative to cholesterol, CTL and dehydroergosterol (DHE) were poor ligands for OSBP. In contrast, both long (ORP9L) and short (ORP9S) variants of ORP9 rapidly extracted CTL, and to a lesser extent DHE, from liposomes. ORP9L and ORP9S also extracted [32P]phosphatidylinositol 4-phosphate (PI-4P) from liposomes, which was inhibited by mutating two conserved histidine residues (HH488,489AA) at the entrance to the binding pocket but not by a mutation in the lid region that inhibited cholesterol binding. Results of direct binding and competition assays showed that phosphatidylserine was poorly extracted from liposomes by ORP9 compared to CTL and PI-4P. ORP9L and PI-4P did not co-localize in the trans-Golgi/TGN of HeLa cells, and siRNA silencing of ORP9L expression did not affect PI-4P distribution in the Golgi apparatus. However, transient overexpression of ORP9L or ORP9S in CHO cells, but not the corresponding PI-4P binding mutants, prevented immunostaining of Golgi-associated PI-4P. The apparent sequestration of Golgi PI-4P by ORP9S was identified as a possible mechanism for its growth inhibitory effects. These studies identify ORP9 as a dual sterol/PI-4P binding protein that could regulate PI-4P in the Golgi apparatus.


Introduction
Lipids and sterols are exchanged between cellular organelles in transport vesicles or by soluble binding proteins that mediate monomeric transfer. Lipid and sterol binding/transport proteins are grouped into several large gene families that are structurally diverse with respect to their lipid binding folds and ancillary regulatory domains. Oxysterol binding protein (OSBP) and OSBP-related proteins (ORPs) comprise one such multigene family expressed in a variety of eukaryotic phyla [1,2]. OSBP/ ORPs are characterized by a lipid-binding fold consisting of a bbarrel with a hydrophobic interior that is capped with a flexible lid that effectively shields sterols and phospholipid acyl-chains from the aqueous environment of the cell [3,4]. In addition to the lipidbinding domain, OSBP/ORPs also transiently interact with specific protein and lipid determinants on organelles. OSBP and most ORPs have a two-phenylalanine in an acid tract (FFAT) motif that binds vesicle-associated membrane protein-associated protein (VAP), a type II endoplasmic reticulum (ER)-associated protein [5,6]. Members of the family also have N-terminal pleckstrin homology (PH) domains that bind phosphatidylinositol polyphosphates and ADP-ribosylation factor in the Golgi appara-tus and other organelles [7,8]. The domain organization of OSBP/ORPs indicates that their primary function is lipid transport and/or signalling that is mediated by sequential or simultaneous interaction with organelle membranes (reviewed in [9]).
OSBP and its mammalian and yeast homologues are known to bind oxysterols, cholesterol and ergosterol [10][11][12][13]. However the lack of direct contact between sterol hydroxyl groups and key residues in the hydrophobic b-barrel of Osh4 suggested that other lipids could be accommodated [3]. Recent structural and interactome mapping of yeast Osh proteins supports this concept. In addition to sterols, the acyl-chains of phosphatidylinositol 4phosphate (PI-4P) extend into the hydrophobic b-barrel of Osh4, with the inositol 4-phosphate headgroup contacting a pair of histidine residues at the tunnel entrance [14]. The absolute conservation of this histidine pair in OSBP homologues across all phyla suggests a common PI-4P binding activity that could be involved in exchange of sterols and PI-4P between the Golgi apparatus and other organelles [14,15]. The Osh3 b-barrel is too narrow to accommodate a sterol and exclusively binds PI-4P [4]. Similarly, Osh6 and Osh7 bind and transfer phosphatidylserine (PS) [16] but bind sterols weakly compared to other Osh proteins [11].
The mammalian OSBP/family is encoded by 12 genes, some of which have numerous splice and promoter variants that further contribute to the function diversity. OSBP, the founding member of the family, and its closest paralogue ORP4 are wellcharacterized high-affinity receptors of 25-hydroxycholesterol (25OH) and cholesterol [13,17]. Similar to Osh4, OSBP also binds PI-4P competitively with cholesterol [18], and catalyzes the reciprocal transfer of PI-4P and dehydoergosterol (DHE) between liposomes in vitro [15]. In a cellular context, OSBP could catalyzes the exchange of cholesterol and PI-4P between the ER and Golgi apparatus, respectively, with Sac1 hydrolysis of PI-4P in the ER providing energy for the transfer of cholesterol to the Golgi apparatus against a concentration gradient [15]. OSBP regulation of Golgi cholesterol and sphingomyelin homeostasis involves sterol transport activity and recruitment of ceramide transfer protein to the organelle by activation of PI 4-kinase IIa activity [19][20][21]. ORP9L binds and transfers cholesterol in vitro and is localized to the trans-Golgi and trans-Golgi network (TGN) where it regulates protein secretion and the cholesterol content of post-Golgi compartments [22]. A truncated promoter variant termed ORP9S, which is missing the N-terminal PH domain and not expressed in the Golgi apparatus, is implicated in control of Akt signalling and cell proliferation [22,23]. Unlike OSBP, ORP9L and ORP9S do not bind oxysterols in vitro nor are they involved in sphingomyelin regulation by oxysterols or cholesterol [22]. To understand the function of ORP9L and ORP9S, we investigated ligand binding properties using a Forster resonance energy transfer (FRET) assay with the fluorescent sterol cholestatrienol (CTL) and direct binding of radiolabelled phospholipids. Results show that ORP9L and ORP4S competitively extract sterols and PI-4P from liposomes, and can sequester or alter the metabolism of Golgiassociated PI-4P in cultured cells.

Cell culture and recombinant protein expression
HeLa cells were cultured in DMEM containing 10% fetal bovine serum (FBS) at 37uC in a 5% CO 2 atmosphere. ORP9L expression was silenced in HeLa cells by transfection using Trans-IT TKO transfection reagent (Mirus, Madison, WI) and a pool of three ORP9L siRNA duplexes (100 nM) or a non-targeting (NT) siRNA (100 nM) for 48 h as previously described [23]. Chinese hamster ovary (CHO) cells cultured in DMEM with 5% FCS and 34 mg/ml proline were transiently transfected with ORP9L expression vectors using Lipofectamine 2000. CHO cells expressing V5-tagged ORP9L and ORP9S under the control of the TETon transactivator were cultured and induced with doxycycline as described [22].

Sterol competition assays
Purified OSBP (20 pmol) was incubated for 2 h at 20uC with 100 nM of [ 3 H]cholesterol and increasing amounts of unlabelled cholesterol, DHE or CTL in 10 mM HEPES (pH 7.4) and 300 mM KCl (binding buffer) containing 2% (w/v) PVA and 0.05% Triton-X 100. Each assay then received 25 ml of Talon metal affinity resin (1:1, v/v) with constant mixing for 25 min. After brief centrifugation, supernatants were removed by aspiration and the Talon resin was washed 3 times with 300 ml of binding buffer at 4uC. OSBP bound to Talon resin was eluted with 100 ml of binding buffer containing 150 mM imidazole and OSBP-bound [ 3 H]cholesterol in the supernatant was measured by liquid scintillation counting.

CTL extraction from liposomes
Liposomes were prepared by combining phospholipids and CTL in a glass tube and drying under nitrogen. The lipid film was hydrated in liposome buffer (25 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM EDTA) to a final concentration of 0.5 mM and vortexed every 5-10 min for 1 h at 20uC. Unless otherwise indicated liposomes were composed of PC/PE/PS/CTL/dansyl-PE (65:20:10:2.5:2.5, mol/mol) and prepared by extrusion through a 400 nm polycarbon membrane using the Lipofast system (Avestin, Ottawa ON). Liposomes were stored at 4uC for no more than 48 h and centrifuged at 15,0006g for 5 min prior to the start of experiments.
FRET assays were conducted using a Cary Eclipse fluorescence spectrophotometer equipped with a single-cell Peltier accessory to maintain temperature at 30uC and a detector setting of 900 V. The fluorescent cholesterol analog CTL (324/370 nm, excitation/ emission) and dansyl-PE (370/520 nm, excitation/emission) were used as a FRET pair. Liposomes (0.05 mM) and proteins (0-2.5 mM) in a final volume of 60 ml of liposome buffer were combined in a micro quartz cuvette, excited at 324 nm (5 nm slit width) and emission at 520 nm (10 nm slit width) was measured at 20-30 sec intervals over 5 min. FRET emission was corrected by subtracting background (liposomes without CTL) and was expressed as a percent of maximal extraction by 1 mM methylb-cyclodextrin (CD). Extraction curves were fit using a one-phase exponential decay model (GraphPad Prism 5 Software).

Immunofluorescence microscopy
CHO or HeLa cells cultured on glass coverslips were fixed in PBS (10 mM sodium phosphate and 150 mM NaCl, pH 7.4) containing 2% formaldehyde (w/v) for 10 min and permeabilized and quenched in 2 ml of PBS containing 20 mg/ml digitonin and 100 mM glycine for 20 min at room temperature. Coverslips were blocked in PBS containing 1% BSA (w/v) for 1 h, and probed with an anti-PI-4P monoclonal antibody for 16 h at 4uC followed by a goat anti-mouse IgM Alexafluor594 secondary antibody for 1 h. Endogenous ORP9L or overexpressed ORP9 was detected with a polyclonal primary antibody and goat anti-rabbit Alexafluor488 secondary antibody. Coverslips were mounted using Mowiol 4-88. Images of HeLa cells were captured using a Zeiss Axiovert 200M inverted microscope equipped with a 63x oil emersion objective and Hamamatsu Orca R2 camera. Images of CHO cells were captured using a Zeiss LSM 510 Meta laser scanning confocal microscope (0.8 mm sections) equipped with 100x oil emersion objective. Fluorescent intensity of PI-4P immunostaining was quantified using Image J (v1.46). The 'analyze particles' command was used to quantify intensity of PI-4P staining of individual cells outlined in 16-bit images that were processed for background and threshold.

Binding of fluorescent sterols by ORP9
Previous evidence of ORP9-mediated extraction and transfer of cholesterol relied on the measurement of soluble or liposomeassociated [ 3 H]cholesterol after sedimenting the donor or acceptor liposomes [22]. Isotope-based assays are fraught with technical problems, such as protein binding to and aggregation of liposomes that limits their usefulness for measuring sterol flux in vitro. As an alternative, the fluorescent sterol analogs CTL and DHE (Fig. 1A) were evaluated as ligands for OSBP and ORP9 using a liposomebased assay and the FRET acceptor dansyl-PE. CTL mimics the membrane behaviour of cholesterol [27] and is transferred between membranes by NPC2 [28]. DHE was used in liposomal extraction and transfer assays for ORP5, Osh4p and OSBP [11,15,29]. ORP9L and ORP9S do not bind cholesterol or 25OH that is dispersed in solution and thus a direct competition assay could not be used to test their relative affinity for fluorescent sterols [22]. As an alternative, we determined the relative affinity of OSBP for CTL and DHE using an assay in which binding of [ 3 H]cholesterol (100 nM) was competed by increasing concentrations of unlabelled cholesterol, CTL or DHE dispersed in Triton X-100 (Fig. 1B FRET between donor CTL and acceptor dansyl-PE was demonstrated with liposomes containing 5 mol% CTL and 2.5 mol% dansyl-PE treated without or with cyclodextrin (CD) (Fig. 1C). In the case of untreated liposomes (black line), excitation at 324 nm and scanning the 350-600 nm interval revealed a strong dansyl-PE emission at 520 nm due to energy transfer and a CTL emission at 370 nm. Addition of 10 mM of CD (grey line) reduced the FRET emission at 520 nm and increased the CTL emission at 370 nm indicating efficient extraction of CTL from liposomes. Extraction of CTL by 10 mM CD was used to establish the maximum and minimum FRET for the calculation of percent extraction by OSBP and ORP9. The relationship between FRET emission and CTL content of liposomes is shown in Fig. 1D. FRET and CTL content of liposomes were proportional indicating that changes in the 520 nm emission directly reflect the CTL content of liposomes. CTL extraction from liposomes containing 2.5 mol% CTL and 2.5 mol% dansyl-PE was initiated by addition of purified OSBP, ORP9L and ORP9S (Fig. 1E). In the absence of OSBP or ORP9, the baseline was stable indicating negligible photobleaching. OSBP did not decrease the FRET signal but both ORP9L and ORP9S extracted CTL based on a rapid 30-40% reduction in the 520 nm emission that was complete in 60-90 s. Increasing the OSBP concentration or altering the composition of liposomes failed to improve CTL extraction. Both ORP9 variants displayed robust extraction of CTL using the FRET assay ( Fig. 2A  and B). ORP9S extracted CTL in a dose-dependent manner to a maximum of 80% at 1 mM ( Fig. 2A), while 0.25-1.0 mM ORP9L extracted a maximum of 70-90% (Fig. 2B). Compared to cholesterol-free liposomes, the inclusion of an equimolar amount of cholesterol (2.5 mol%) in liposomes caused a 50% inhibition of CTL extraction by ORP9L (Fig. 2C). ORP9L and ORP9S (0.5 mM) also extracted DHE from liposomes, although less efficiently than it did CTL (compare to 0.5 mM ORP9L and ORP9S in Fig. 2A and B) (Fig. 2D).   (Fig. 3A). Purified ORP9L and ORP9S were incubated with liposomes containing [ 32 P]PI-4P and extraction of radioactivity into the supernatant was assayed ( Fig. 3B and C). Increasing concentrations of ORP9L extracted a maximum of 9% [ 32 P]PI-4P from liposomes (Fig. 3B), which was similar to OSBP extraction of PI-4P under the same conditions [18]. ORP9S extracted slightly more PI-4P (14%) indicating the PH domain does not significantly affect this activity. Mutation of the conserved histidine pair in ORP9L-HH/AA reduced PI-4P extraction relative to wild-type ORP9L and ORP9S by .75% (Fig. 3C). Mutation of conserved residues in the b-barrel lid of ORP9L-D375-378 that reduce affinity for cholesterol [22] did not affect PI-4P extraction activity. The sterol extraction activity of ORP9L-HH/AA was tested in the CTL-based assay (Fig. 3D). ORP9L-HH/AA extracted CTL from liposomes as efficiently as the wild-type protein confirming that these residues are involved specifically in PI-4P binding.
The ability of PI-4P to compete with sterols was measured by including PI-4P in liposomes containing CTL (2.5 mol%) and measuring extraction by ORP9L and ORP9S relative to PI-4Pfree liposomes (Fig. 4). Inclusion of an equimolar amount to PI-4P (2.5 mol%) inhibited CTL extraction by both ORP9L and ORP9S ( Fig. 4A and B). Increasing the PI-4P content above 2.5 mol% did not further inhibit CTL extraction by either ORP9L or ORP9S (results not shown).
A recent study that identified Osh6 and Osh7 as PS binding and transfer proteins also suggested that several mammalian ORPs, including ORP9, might have similar activity based on the presence of a conserved a1-b2 loop at the entrance to the binding pocket and lid that interacts with phosphoserine [16]. Since CTL was extracted from liposomes containing 10 mol% PS, it was deemed unlikely to be a strong competitive ligand for ORP9. In support of this conclusion, we found that the amount of CTL extracted by ORP9L and ORP9S from PS-free liposomes (Fig. 5A) was similar to experiments using 10 mol% PS liposomes (see Fig. 1E). Direct extraction of [ 3 H]PS from liposomes by ORP9 was also measured using an assay similar to that used to measure PI-4P binding (Fig. 5B). The PS-binding protein Osh6 extracted approximately 2% of [ 3 H]PS from liposomes, which is similar to activity reported using a nonradioactive-based assay [16]. ORP9L, ORP9S, OSBP and ORP4 had similar activity, suggesting that PS is potential ligand but is relatively weak compared to sterols and PI-4P.

Expression of ORP9 sequesters cellular PI-4P
ORP9L is localized to the trans-Golgi/TGN where it affects Golgi secretion and the distribution of cholesterol in post-Golgi compartments [22]. Since PI-4P is synthesized in late Golgi compartments and has an essential role in recruitment of secretory factors [30][31][32], we tested whether reduction of ORP9L expression by siRNA silencing affected the distribution or content of PI-4P in that compartment. PI-4P was visualized in HeLa cells using a PI-4P-specific monoclonal antibody that does not recognize other phosphatidylinositol species [33]. PI-4P immunostaining was primarily in a perinuclear region that partially overlapped with the cis/medial marker giantin and the trans-Golgi network protein TGN38 ( Fig. 6A and B). In HeLa cells transfected with control non-targeting siRNA (siNT), ORP9L and PI-4P were poorly colocalization in the perinuclear Golgi compartment (Fig. 6C and  D). siRNA silencing of ORP9L reduced expression by .80% compared to siNT transfected HeLa cells (Fig. 6E) but the staining intensity and distribution of PI-4P was similar to control cells (Fig. 6C).   To determine if ORP9 could sequester or alter PI-4P distribution at the Golgi apparatus, ORP9L and ORP9S, as well as the HH/AA mutants that are defective in PI-4P binding, were transiently expressed in CHO cells and the localization of PI-4P was monitored by immunofluorescence confocal microscopy (Fig. 7). CHO cells expressing ORP9L displayed an 80% reduction in PI-4P fluorescence intensity compared to surrounding non-transfected cells (Fig. 7A). ORP9L-HH/AA was diffusely localized like the wild-type protein but the distribution and intensity of PI-4P immunostaining in expressing cells was not significantly affected. ORP9S expression also reduced PI-4P immunostaining relative to non-transfected cells (Fig. 7B). Interestingly, ORP9S-HH/AA expression increased PI-4P fluorescence intensity throughout the cell. Thus excess ORP9 shields PI-4P from detection or dramatically reduces Golgi PI-4P content ( Fig. 7A and B).
Overexpression of ORP9L and ORP9S cause disorganization of the Golgi apparatus and, in the case of ORP9S, inhibit secretion and cell proliferation [22]. This was prevented by mutations that disrupted VAP binding by the FFAT motif (ORP9-FY/AA) and sterol binding (ORP9S-D375-378). To assess if these effects could be related to sequestration of PI-4P, CHO cells expressing ORP9L and ORP9S under the control of the TET-on transactivator were cultured in the presence of doxycycline for 48 h and immunostained for PI-4P (Fig. 8). Similar to the results shown in Fig. 7, CHO cells expressing ORP9L or ORP9S were virtually devoid of PI-4P staining (see outlined areas in PI-4P panels) compared to adjacent non-transfected cells. In contrast, expression of ORP9S-FY/AA or ORP9S-D375-378 had no effect on PI-4P detection compared to adjacent non-transfected cells, indicating that the VAP and sterol binding domains of ORP9S allows it to access and interact with PI-4P in the Golgi/cytoplasmic compartment.

Discussion
Emerging evidence of divergent ligand specificities among OSBP homologues suggests regulatory functions in sterol/anionic phospholipid homeostasis and transport. OSBP, the prototypic member originally identified by its 25OH binding activity, binds PI-4P competitively with cholesterol and may exchange these lipids or regulate associated metabolic pathways at the ER-Golgi interface. ORP9 and ORP11 also associate with the Golgi apparatus where they influence the structure of and cholesterol distribution in the ER-Golgi intermediate compartment, Golgi apparatus and late endosomes [22,25,34]. ORP9 is a cholesterol binding protein but could also bind PI-4P and PS via conserved residues at the entrance to its lipid-binding pocket. Similar to OSBP, results show that Golgi-localized ORP9L, as well as the ORP9S variant lacking the PH domain, are cholesterol and PI-4P binding proteins that sequester and/or modify Golgi PI-4P content when expressed in cells.
An assay using the FRET pair CTL and dansyl-PE was developed to measure the sterol extraction and transfer activity OSBP and ORP9. Based on a competition assay between soluble sterols, CTL and DHE were both poor ligands for OSBP relative to cholesterol. The lack of extraction of CTL or DHE from liposomes by OSBP thus reflects poor affinity as well as the fact that OSBP extracts only 10-15% of the preferred ligand [ 3 H]cholesterol under similar conditions. A recent study showed that DHE can be efficiently extracted and transferred between liposomes but only when OSBP is activated by trypsin digestion or by interaction with VAP and PI-4P in donor and acceptor membranes [15]. In contrast, CTL extraction by ORP9L and ORP9S was rapid and reached a stable baseline within 90-120 s.
Despite some variability in the activity of ORP9L and ORP9S preparations purified from Sf21 cells, perhaps reflecting the presence of an endogenous ligand or modification, a 1:1 (mol/mol) ratio of ORP9 to liposomal CTL reproducibly yielded 30-40% extraction. Although we were unable to directly compare the affinity of CTL and cholesterol using the competition assays described in Fig. 1B, the robust activity measured in the FRET assay compared to previous measures of [ 3 H]cholesterol extraction from liposomes [22] indicates that ORP9L and ORP9S could remain associated with the liposomes after extraction of CTL. It seems unlikely that ORP9 has increased affinity for CTL since the addition of equimolar cholesterol to liposomes inhibited ORP9L extraction of CTL by 50% (Fig. 4C). Despite robust extraction of CTL by ORP9L and ORP9S, we were unable to demonstrate transfer to acceptor liposomes under a variety of conditions including in the presence of PI-4P and recombinant VAPA (result not shown). The reasons for this remain unclear since [ 3 H]cholesterol transfer between liposomes by ORP9L using a precipitation method was previously observed. The discrepancy could be due to differences in the sterol ligands or the type of assay. A FRET-based assay is the preferred method since it measures the delivery of sterol to the donor membrane and does not register transfer proteins that bind to membranes but do not release their ligand, which could occur in assays that rely on liposome precipitation. However, a caveat of FRET assays is the use nonphysiological ligands that have reduced affinity compared to endogenous ligands, as is the case for DHE binding by OSBP.
The results of direct binding and competition assays, and mutagenesis of essential residues indicate that ORP9L and ORP9S bind PI-4P in the C-terminal lipid-binding pocket. The ORP9 PH domain, which recognizes PI-4P in the Golgi apparatus [22], does not appear to affect this activity since ORP9L and ORP9S extracted PI-4P to a similar degree. Importantly, ORP9L-HH 488,489 AA and ORP9L D375-378 had normal CTL and PI-4P binding activity, respectively, indicating that binding of these ligands is mutually exclusive but involves interaction with different residues in the binding pocket. CTL extraction from liposomes by ORP9 was inhibited 50% by inclusion of an equimolar amount of PI-4P indicating similar relative affinities for both ligands. A PS recognition motif identified in Osh6 and Osh7 is conserved in several ORPs including ORP9 [16], and ORP9, OSBP and ORP4 displayed similar PS extraction activity as Osh6. However, inclusion of PS in liposomes did not affect CTL extraction by ORP9L or ORP9S and the relative extraction efficiently of sterols and PI-4P by ORP9 was greater, suggesting that PS is a weaker, non-preferred ligand in vitro. Whether ORP9 or other members of the family affect PS metabolism or distribution in cells requires further study.
PI-4P is enriched in the trans-Golgi/TGN where binding or transport by ORP9L could regulate its distribution or content, accounting for the inhibition of ER-Golgi protein transport and altered Golgi structure observed following ORP9L silencing in CHO cells [22]. Despite the trans-Golgi/TGN localization of ORP9L there was minimal co-localization with PI-4P, suggesting that ORP9L could prevent PI-4P accumulation in its Golgi compartment or sequester PI-4P from detection by the anti-PI-4P antibody. However, silencing of ORP9L did not effect PI-4P distribution indicating minimal impact on the metabolism and partitioning of PI-4P within the Golgi apparatus. We cannot discount the possibility that loss of ORP9L expression might alter the metabolism of PI-4P without affecting immunodetection of the Golgi pool, or be compensated for by OSBP or other Golgispecific PI-4P metabolic pathways.
Inducible expression of ORP9S inhibits ER-Golgi protein transport, fragments the Golgi apparatus and inhibits cell proliferation [22], which could be related to PI-4P sequestration. Indeed, inducible or transient expression of ORP9S or ORP9L blocked immunodetection of PI-4P, which was prevented by mutation of two histidine residues required for PI-4P binding in vitro. A prior study showed that immunodetection of PI-4P is competed by exogenous addition of the FAPP-PH domain to fixed cells [33], suggesting that the observed loss of PI-4P immunofluorescence is due to binding of PI-4P by overexpressed ORP9. However, we cannot rule out the possibility that ORP9 expression facilitates a reduction in the PI-4P content of Golgi membranes. Our finding that ORP9S sequestered PI-4P but the FFAT and sterol-binding mutants did not is consistent with the lack of growth inhibition by these two mutants [22]. It is feasible that VAP and sterol binding activity of ORP9S allows it to access and remove PI-4P from its essential secretory functions in the Golgi apparatus. In  ORP9 Is a PI-4P Binding Protein contrast, ORP9L also sequestered PI-4P but because ORP9L associates with the Golgi apparatus via the PH domain, PI-4P could still be available to promote secretion from the compartment. Thus the absence of a PH domain in ORP9S does not directly affect its ability to sequester Golgi PI-4P but could limit the subsequent availability of PI-4P.
In conclusion, we have identified another member of the mammalian OSBP family that binds sterols and PI-4P at the ER-Golgi interface. Although ORP9L and OSBP are situated in the trans-Golgi/TGN, there are fundamental differences in terms of lipid binding properties and protein partners that reveal different functional outputs. Whether there is a common underlying transport or signalling activity remains to be resolved.