Cholesterol and its numerous oxygenated derivatives (oxysterols) profoundly affect the biophysical properties of membranes, and positively and negatively regulate sterol homoeostasis through interaction with effector proteins. As the bulk of cellular sterols are segregated from the sensory machinery that controls homoeostatic responses, an important regulatory step involves sterol transport or signalling between membrane compartments. Evidence for rapid, energy-independent transport between organelles has implicated transport proteins, such as the eukaryotic family of OSBP (oxysterol-binding protein)/ORPs (OSBP-related proteins). Since the founding member of this family was identified more than 25 years ago, accumulated evidence has implicated OSBP/ORPs in sterol signalling and/or sterol transport functions. However, recent evidence of sterol transfer activity by OSBP/ORPs suggests that other seemingly disparate functions could be the result of alterations in membrane sterol distribution or ancillary to this primary activity.
The intracellular pathways for cholesterol synthesis, uptake and efflux are subject to negative and positive feedback by direct interaction of cholesterol or its oxygenated derivatives (oxysterols) with regulatory components, including the SREBP (sterol-regulatory-element-binding protein)–SREBP-cleavage-activating protein–Insig complex [1,2], HMG-CoA (3-hydroxy-3-methylglutaryl-CoA) reductase  and LXRs (liver X receptors) [4,5]. These reside in relatively cholesterol-poor compartments that are subject to acute changes in local cholesterol and oxysterol content. Although there is substantial information on sterol interaction with this regulatory machinery, the proximal events that control delivery of cholesterol and oxysterols to relevant compartments are poorly understood. On the basis of evidence that differentiates vesicular from protein-mediated cholesterol transport, it is proposed that the latter accounts for the majority of cholesterol delivery between organelles [6,7]. However, identification of soluble cholesterol/oxysterol transport proteins has been problematic due to functional redundancy, inadequate probes or end-points to monitor cholesterol movement in vivo and difficulties in the identification of organelle-specific targeting mechanisms. Recent interest has focused on START [StAR (steroidogenic acute regulatory) protein-related lipid transfer] and the OSBP (oxysterol-binding protein)/ORP (OSBP-related protein) families that are encoded by 27 genes comprising almost 0.1% of the entire human genome [8,9]. Members of these protein families have verified cholesterol/oxysterol-binding motifs adjacent to domains with regulatory and membrane-targeting functions. The present review will focus on the OSBP/ORP family in the context of integration of sterol binding and transport functions with effects on signalling, vesicular trafficking and lipid metabolism. Where necessary, structural and functional parallels with yeast OSHs (OSBP homologues) will be identified (but see also the recent comprehensive reviews on this subject [10,11]).
PHYLOGENETIC DISTRIBUTION AND GENE EXPRESSION
Data mining of EST (expressed sequence tag), cDNA and genomic databases revealed that the human OSBP family consists of OSBP and ORP1–11 (also termed OSBP1, OSBP2 and OSBP-like 1–3 and 5–11) [12–14]. Mammalian ORPs are grouped into six sub-families based on sequence identity (Figure 1): subfamily I, OSBP and ORP4; subfamily II, ORP1 and 2; subfamily III, ORP3, 6 and 7; subfamily IV, ORP5 and 8; subfamily V, ORP9; and subfamily VI, ORP10 and 11 . The defining feature of this gene family is a C-terminal sterol-binding or OH (OSBP homology) domain that harbours a signature sequence (EQVSHHPP) that is conserved in orthologues from other species. With the exception of ORP2, full-length or L (‘long’) gene products have N-terminal PH (pleckstin homology) domains that interact with PIPs (phosphoinositide phosphates). All OSBP/ORPs, with the exception of subfamily IV and VI, also contain a FFAT motif [two phenylalanines (FF) in an acidic tract] that binds VAP [VAMP (vesicle-associated membrane protein)-associated protein] on the ER (endoplasmic reticulum). This function seems to be fulfilled in subfamily IV by a unique C-terminal transmembrane domain that mediates ER localization. The functional significance of these domains will be elaborated below; however; the initial impression is of a family of sterol-binding proteins capable of interaction with the ER and other organelle membranes.
Structural organization of the human OSBP/ORP family
ORP genes also give rise to truncated or S (‘short’) variants with unique expression and functional properties as a result of alternative promoter and splice-site usage. For example, an alternate promoter prior to exon 3 in the ORP4 gene gives rise to a short (ORP4S) variant missing the PH domain  (Figure 1). Alternative promoters before exons 16 and 6 of the ORP1 and ORP9 genes respectively result in expression of ORP1S and ORP9S [16,17]. As many as eight ORP3 isoforms are produced as a result of alternative splicing of exons 9, 12 and 15 .
OSBP orthologues are broadly distributed across eukaryotic phyla. The yeast genome encodes homologues OSH1–7 that are sorted into four subfamilies . Entire or partial genomic complements of OSBP homologues have been identified in flies (Drosophila melanogaster) , plants [21,22], worms (Caenorhabditis elegans)  and slime mould (Dictyostelium discoideum) . At one extreme, the parasitic protist Cryptosporidium parvum expresses only two ORPs, a short-type containing only the ligand-binding domain and a longer variant with an N-terminal PH domain , whereas at the other end of the spectrum Arabidopsis thaliana and mammals have 12 genes . Phylogenetic relationships between 120 taxa revealed support for clustering on the basis of type rather than taxonomic distribution, suggesting a set of early evolutionary ancestors [14,25]. Groupings of 2–3 mammalian and fungal OSBP/ORPs within each cluster also indicate more recent duplication events. Interestingly, the presence of OSBP/ORPs in D. melanogaster and C. elegans, neither of which synthesize sterols, implies a lack of involvement in regulation of endogenous synthesis . However, this does not preclude a role in uptake or transport as these organisms require exogenous sterols for survival.
Owing to the size of the gene family and diversity of promoter and splice variants, ORPs have highly complex expression patterns that have not been correlated with tissue-specific functions. Studies credited with the initial profiling of the human and murine ORP gene family reported the tissue distribution of mRNA transcripts by Northern blotting, hybridization with human-tissue-expression filter arrays [12,13] and RT (reverse transcription)–PCR analysis [13,14,27]. Although OSBP/ORP mRNAs are expressed to some extent in most human tissues, transcript levels vary significantly between tissues and within subfamilies. For instance, OSBP (subfamily I) is expressed in a variety of human and mouse tissues, but predominantly in brain, heart, kidney and liver, whereas transcripts for ORP4L and ORP4S (also subfamily I) are expressed almost exclusively in brain and heart [13,15,27]. Similarly, protein and mRNA expression profiles of subfamily III members are tissue-specific and non-overlapping: ORP3 is expressed in kidney, lymph nodes and thymus; ORP6 is in brain and skeletal muscle; and ORP7 is restricted to the gastrointestinal tract . It is also notable that OSBP/ORP gene expression profiles are not shared among tissues with a prominent role in sterol metabolism and, with the exception of ORP6  and ORP1 , are not regulated by altering cellular sterol levels. The lack of acute regulation of OSBP/ORP gene expression suggests essential housekeeping function(s).
STRUCTURAL ORGANIZATION OF OSBP/ORPs
The defining feature of the OSBP/ORP gene family is the aptly named C-terminal 400 amino acid OH domain that binds cholesterol and oxysterols [29–32] (Figure 1). The sterol-binding activity of OSBP/ORPs are routinely assayed by dispersing oxysterols or cholesterol in aqueous buffers or with low concentrations of detergent respectively [29,31]. Under these conditions, OSBP binds 25-OH (25-hydroxycholesterol) (Kd of 10 nM) and cholesterol (Kd of 173 nM) with relatively high affinity [15,33–35]. Competition-based assays revealed that OSBP binds other oxysterols with reduced affinity and does not bind steroid hormones . Similarly, ORP4L binds 25-OH (Kd of 10 nM) in vitro , an activity that is inhibited by mutations in the putative lid region of the binding fold . Despite the binding of cholesterol (Kd of 267 nM) and 25-OH (Kd of 48 nM) by a GST–ORP4S fusion protein, encompassing the FFAT and sterol-binding domain , ORP4S expressed in CHO (Chinese-hamster ovary) cells was devoid of sterol-binding activity perhaps due to a cellular inhibitory factor . ORP2 binds 25-OH (Kd of 3.9 μM), (22R)-OH [(22R)-hydroxycholesterol] (Kd of 14 nM), 7-ketocholesterol (Kd of 140 nM) and cholesterol [32,36]. ORP1L binds (22R)-OH and 25-OH [32,37] and the OH domain of ORP8 bound 25-OH but not (24S)-hydroxycholesterol or 7-oxocholesterol .
The sterol-binding activity of OSBP/ORPs has also been demonstrated in cultured cells by using photo-activatible 6-azi-cholesterol and -25-OH . The OH domains of OSBP and ORPs 1–5, 7 and 10 expressed in COS cells formed adducts with photo-25-OH as did full-length ORPs. With the exception of ORP9 and ORP4, all family members formed adducts with photo-cholesterol. Although the binding constants and ligand specificity are not available for the entire family, it is apparent that OSBP/ORPs bind cholesterol and oxysterols in the low nM range and display preference for side-chain hydroxylated sterols, such as 25-OH. As the apparent affinity for oxysterols and cholesterol differs only 3–4-fold and cholesterol is more than 103-fold more abundant than oxysterols, the endogenous ligand for OSBP/ORP could be cholesterol. However, the difference in solubility of these sterols suggests that OSBP/ORPs could preferentially bind to oxysterols in the cytoplasmic compartment and to cholesterol in membranes.
The structure of the OH domain of the yeast Osh4p has been solved in complex with cholesterol, ergosterol or oxysterols . The general features of the binding fold, which accommodates a single sterol ligand in a hydrophobic tunnel that is capped by an α-helical lid, is consistent with a sterol transport function. The ligand-binding core is formed by a 180 amino acid β-barrel consisting of a 19-strand, anti-parallel β-sheet that forms a hydrophobic tunnel with a hydrophilic exterior. The tunnel is capped by a flexible N-terminal lid, followed by a two-stranded β-sheet and three α-helices that form a wall of the β-barrel, plugging the end of the tunnel distal to the lid. The lid of Osh4p also contains a curvature-sensing activity similar to the Arf (ADP-ribosylation factor)-GAP (GTPase-activating protein) lipid packing sensor that might assist in selective binding to Golgi membranes . A 125 residue C-terminal region localized to the distal end of the β-barrel is implicated in membrane binding and sterol transfer . Ergosterol, cholesterol, 7-hydroxycholesterol, 20-hydroxycholesterol and 25-OH are positioned with the 3-hydroxy groups at the bottom of the tunnel and the sterol iso-octyl group interacting with side chains in the lid by van der Waals interactions. Quite unexpectedly, sterol hydroxy groups do not make direct contact with residues lining the tunnel, but are hydrogen-bonded to polar residues via water molecules. Two water molecules are bound to the sterol 3-hydroxy group and Gln96, which is in a cluster of hydrated residues (Trp46, Tyr97, Asn165 and Gln181), at the bottom of the tunnel. The 25-hydroxy side chain of 25-OH interacts with Lys109 at the mouth of the tunnel via two water molecules, but other oxysterol hydroxy groups do not demonstrate such ordered interactions. Interestingly, Osh4p could only be crystallized either as a deletion mutant, lacking the N-terminal lid, or in a ligand-bound form, indicating that the N-terminal region is highly flexible and disordered in the apo-state . Molecular dynamics simulations of the binding and release of cholesterol suggest that Osh4p undergoes significant conformational changes between bound and apo forms, consistent with opening and closing of the lid [42,43].
Full-length ORPs account for approx. 4% of all PH domain-containing proteins encoded by the human genome . Despite limited sequence similarity, all PH domains consist of a conserved seven stranded β-sandwich capped by an α-helix. Opposite to the helical cap, conserved basic residues in loops connecting the β-strands mediate interaction with compositionally diverse PIPs found in eukaryotic membranes. However, less than 10% of PH domains bind PIPs with high affinity and specificity  and only 30% of PH domain-containing proteins in yeast display membrane targeting .
PH domains are distinguished from other PIP-binding domains by their specific high-affinity binding to PIPs with two vicinal phosphate groups: PtdIns(3,4)P2, PtdIns(4,5)P2 or PtdIns(3,4,5)P3 . This specificity results in targeting of some eukaryotic PH domain proteins to the PM (plasma membrane) where PIP levels are increased as a result of signalling events. Among a handful of outliers in this group are the closely related OSBP, Osh1p and Osh2p PH domains that bind PtdIns4P and PtdIns(4,5)P2 with similar, moderate affinity in vitro. In vivo, however, OSBP and Osh1p PH domains specifically localize to the Golgi apparatus in a PtdIns4P-dependent manner, a feature that has led to their use as diagnostic tools for qualitative measurement of PtdIns4P in the Golgi apparatus and other membranes [46,47]. Golgi localization of the OSBP and Osh1p PH domains in yeast is dependent on Arf1p and PtdIns4P synthesis by Pik1p (phosphatidylinositol kinase 1) . It was later confirmed that mammalian ARF1 is required for membrane targeting of the OSBP PH domain . Conserved basic residues in the variable β-loops of the OSBP PH domain are required for PIP binding; however, a second site for Golgi interaction in the Osh1p PH domain involves a histidine residue at position 79 in the β7 strand . This surface of the PH domain is positioned away from the plane of the membrane and thus probably interacts with a protein factor, possibly Arf1. The presence of an arginine residue at this position in the Osh2p PH domain results in affinity for PtdIns4P at the Golgi  and PM . Interestingly, His79 is conserved in OSBP and ORP9, both of which bind PtdIns4P and are localized to the Golgi apparatus [29,51], suggesting a conserved two-site interaction model for Golgi-specific PH domains that involves lipid and protein receptors.
The PH domains of other mammalian ORPs also bind PIPs but are poorly characterized with respect to ligand specificity and in vivo membrane targeting. The PH domain of ORP3, which binds PtdIns(3,4)P2 and PtdIns(3,4,5)P3 with moderate affinity , is necessary, but not sufficient, for PM localization [28,52]. The ORP1L PH domain binds liposomes containing PtdIns(3,4)P2, PtdIns(3,5)P2 and PtdIns(3,4,5)P3 with low affinity and specificity, however PH domain-dependent localization is not observed .
Interestingly, several short ORPs lacking PH domains display in vitro PIP-binding activity. Residues 205–314 of Osh4p, corresponding to β-strands 9–17 of the sterol-binding fold, bound PtdIns(4,5)P2in vitro with affinity similar to the full-length Osh4p . Mutations in this region that inhibit PIP-binding also prevent Golgi localization of Osh4p in yeast cells . A similar region of ORP1S binds a variety of PIPs, but, unlike full-length ORP1S and Osh4p, this region was not sufficient to inhibit Sec14 (secretory 14)-dependent Golgi secretion in yeast . Other short ORPs, including Osh6p, ORP2, ORP10S and ORP9S, non-specifically bind PIPs and other anionic phospholipids that are immobilized on membrane supports [56–58]. In addition, Osh4p also catalyses the extraction and transfer of PtdIns(4,5)P2 from liposomes . This activity is less efficient than sterol extraction and transfer, but is unaffected by mutations that block these activities, suggesting that a second site is involved that is distant from the sterol-binding region. The aforementioned PtdIns(4,5)P2-binding region of Osh4p comprises two positively charged loops opposite the sterol-binding pocket that mediate second-site binding of anionic lipids required for sterol transfer by a mechanism involving simultaneous contact between adjacent membranes .
FFAT motif and VAP binding
Mammalian ORPs of subfamilies I, II, III and IV contain a FFAT motif (E-F/Y-F/Y-DAxE) that binds mammalian VAP-A and VAP-B or, in the case of yeast Osh proteins, Scs2p (suppressor of choline sensitivity 2) [15,17,60,61] (Figure 1). VAP is a type II integral membrane protein that is anchored to the ER by a C-terminal hydrophobic domain that resembles that of tail-anchored proteins that are inserted post-translationally . The cytoplasmic domain consists of an N-terminal MSP (major sperm protein)-like motif that binds FFAT and a coiled-coiled domain involved in multimerization. The combination of FFAT and PH domains provides most OSBP/ORPs with dual membrane-targeting activity. However, this relationship does not strictly apply: ORP2 and ORP9S lack PH domains but have FFAT motifs; ORP1S and ORP4S lack both; and ORP10 and ORP11 have only PH domains. In lieu of a FFAT motif, subfamily IV ORPs are anchored to the ER by C-terminal transmembrane domains . Interestingly, 14 out of 17 eukaryotic proteins containing perfect FFAT motifs have been implicated in lipid metabolism, indicating that VAP is an important regulator of lipid homoeostasis in the ER and other organelles [61,63]. For a more in-depth analysis of other biological functions of VAP see .
Ankyrin domain targeting to membrane contacts
ORP1, Osh1p and Osh2p each contain three ankyrin motifs (Figure 1), a common 33-residue structural domain comprised of two antiparallel α-helices separated by a β-hairpin, which forms a protein–protein interaction surface (reviewed in ). Osh1p is uniquely localized to the NVJ (nuclear–vacuolar junction), a site of piecemeal autophagy of the nucleus , and the TGN (trans-Golgi network) . The sequestration of Osh1p to the NVJ requires interaction of the three ankyrin repeats with Nvp1 (nuclear vacuolar protein 1), which anchors the outer nuclear membrane to the vacuole through interactions with Vac8 (vacuole-related 8) [50,67]. Although Osh1p is not required for NVJ formation, its partitioning between the Golgi and NVJ appears to be important for raft-dependent trafficking of the tryptophan transporter, Tat2p . Osh2p also has a three-ankyrin-repeat motif, but does not localize to the NVJ, nor does an Osh1p chimaera containing the Osh2p ankyrin repeats . The only mammalian ORP with three ankyrin repeats, ORP1L, does not localize to NVJs but could form cholesterol-dependent ER–endosome contact sites . The ankyrin motifs in ORP1L interact with Rab7 to regulate endosomal positioning on microtubule tracks [53,70].
BIOLOGICAL FUNCTIONS OF OSBP/ORPS
Sterol transport by OSBP/ORPs
Mammalian OSBP/ORPs have a common cholesterol and/or oxysterol-binding activity, yet individual members of the family have seemingly disparate functions, such as regulation of lipid metabolism, signal transduction, vesicular transport and cytoskeletal regulation. These activities could be secondary to a common sterol transport activity or primary signalling functions that result from conformational changes induced by cholesterol or oxysterol binding. Owing to the size of the OSBP/ORP family and the potential diversity of sterol ligands, these proposed functions need not be mutually exclusive. However, recent evidence demonstrating in vitro and in vivo sterol transfer by OSBP/ORPs and Osh proteins, and supporting structural evidence, suggests that sterol transfer is a primary function which effects a variety of downstream targets by altering membrane environment
Sterol transfer in yeast
The absence of other putative transfer proteins in yeast, such as those of the START family, suggests that OSH proteins have key sterol regulatory or transfer functions. Indeed, deletion of all seven OSH genes (OSHΔ) is lethal but bypassed by the expression of any one OSH gene indicating common/related function(s) . The substantive accumulation of ergosterol and other sterols in OSHΔ cells implied an important role in sterol regulation, but it was not until the structure of Osh4p was solved that it became apparent that the family might have a conserved transfer function . Thereafter, Raychaudhuri et al.  showed that Osh4p catalysed the non-vesicular transport of cholesterol and ergosterol between liposomes in vitro, an activity that was enhanced 2- to 4-fold by inclusion of PtdIns(4,5)P2 in donor and/or acceptor liposomes. All OH domains of the OSH family are capable of transferring cholesterol in vitro, but Osh2p, Osh4p and Osh5p are the most active when phosphatidylserine was included in the donor and acceptor liposomes . Osh4p displayed in vivo transfer activity in an esterification-based assay, which monitored cholesterol and ergosterol transfer from the PM to the ER, but it was evident that other Osh proteins were also involved, particularity Osh3p and Osh5p . Even though Osh4p is not localized to the ER or PM , its expression can partially restore PM-to-ER sterol transport. This suggests that individual Osh proteins are loosely sequestered at one or more organelles, but can function outside these constraints to compensate for the loss of other family members.
ER-to-PM cholesterol transport in yeast has a non-vesicular component that involves equilibration of non-raft-associated sterols between the two compartments . Although sphingolipids control this bidirectional pathway by influencing the free sterol pool in the PM, an OSHΔ strain had a 5-fold reduction in ergosterol transport to the PM [72,73]. It was speculated that a sterol transfer function was not consistent with the reported roles of Osh proteins in vesicular trafficking associated with polarized cell growth. Inactivation of OSH genes results in lack of Rho1p and Cdc42p localization and defective vesicle fusion at bud sites, a defect that is rescued by expression of any OSH gene . However, it is conceivable that Osh proteins regulate the localized sterol environment at bud sites, and other regions of the PM, by direct or indirect sterol transfer and thus influence the recruitment and activity of lipid-anchored GTPases. The prevailing evidence that Osh proteins bind and transfer sterols in vitro, collectively transport sterols between the PM and ER, and that disruption of Osh4p sterol- or phospholipid-binding activity abrogates rescue of the OSHΔ strain, supports the concept that the yeast family are sterol transfer proteins [10,41,59].
Sterol transfer by mammalian OSBP/ORPs
Recent evidence has also suggested that OSBP and several ORPs have sterol transfer activity that could underlie the observed effects of these proteins on lipid and sterol homoeostasis. OSBP was originally hypothesized to inhibit cholesterol synthesis by interacting with 25-OH. However, stable expression in CHO cells did not affect 25-OH-mediated suppression of HMG-CoA synthase, HMG-CoA reductase or LDLR [LDL (low-density lipoprotein) receptor] mRNAs . Moreover, silencing of OSBP expression by RNAi (RNA interference) failed to affect sterol-regulatory events at the ER [75,76].
OSBP localizes to the ER and Golgi apparatus via FFAT domain interaction with VAP  and PtdIns4P binding by the PH domain [30,46] respectively. Within these compartments, OSBP stimulates SM (sphingomyelin) biosynthesis in response to cellular cholesterol and oxysterol levels. Overexpression and silencing studies show that OSBP is required for 25-OH and cholesterol-dependent stimulation of SM synthesis [75,77]. Activation occurs via recruitment of the CERT (ceramide transfer protein) to the Golgi apparatus, resulting in increased ceramide delivery to SM synthase . Defective SM synthesis in OSBP-depleted cells was not rescued by expression of OSBP with mutations in the PH, FFAT or sterol-binding domains, indicating a mechanism involving contacts between the ER and Golgi . This initially suggested that OSBP binds cholesterol or oxysterol in the Golgi and/or ER and uses this signal to increase SM synthesis as a means of co-ordinating the level of these two raft-associated lipids. This relationship is further strengthened by the observation that depletion of SM or cholesterol at the PM promotes dephosphorylation and translocation of OSBP to the Golgi apparatus .
The lack of interaction between OSBP and CERT indicates an indirect mechanism of activation, suggestive of sterol-dependent signalling. However, OSBP could also function as a sterol transfer protein between the ER and Golgi that modifies the membrane environment and Golgi recruitment of CERT (Figure 2). Consistent with this possibility, we recently showed that OSBP not only binds sterols presented in aqueous and detergent dispersion, but also extracts cholesterol from liposomes and mediates its transfer between donor and acceptor liposomes . Cholesterol transfer is stimulated by PtdIns4P binding by the PH domain, reproducing at least one stage of cholesterol transfer between the Golgi apparatus that involves membrane-specific targeting . Unlike Osh4p, PtdIns4P stimulated cholesterol transfer by OSBP when included only in donor liposomes, both donor and acceptor liposomes, but not in acceptor liposomes alone. This could reflect a requirement for both cholesterol and PtdIns4P in the same membrane and shows that PtdIns4P imposes some degree of directionality on cholesterol transfer. Demonstration of sterol transfer activity by OSBP provides a mechanistic basis to explain CERT activation as follows: OSBP interaction with VAP in the ER would facilitate binding of cholesterol (or oxysterols), followed by targeting to the Golgi apparatus and release of sterol ligand; the subsequent enrichment of sterols in the Golgi apparatus would alter the membrane environment leading to CERT recruitment (Figure 2). Current evidence points towards a sterol- and OSBP-dependent activation of PtdIns4P synthesis in the Golgi, leading to increased CERT recruitment via its PH domain [79,80].
Functional consequences of OSBP- and ORP9L-dependent sterol transfer between the ER and Golgi apparatus
OSBP in the ER/Golgi also appears to directly or indirectly affect sterol-dependent activities in other areas of the secretory pathway. Cholesterol efflux at the PM by the ABC (ATP-binding cassette) A1 transporter protein is increased 3-fold in CHO cells depleted of OSBP . Down-regulation of ABCA1 by OSBP overexpression was prevented by disruption of cholesterol binding, but not by mutations that abolished localization to the Golgi apparatus or ER . As OSBP regulation of CERT activity requires the PH, FFAT and sterol-binding domains, inhibition of ABCA1 expression appears to be independent of its role in sterol-activated SM synthesis in the Golgi apparatus. Similarly, OSBP expression is negatively correlated with processing of APP (amyloid precursor protein) and secretion of amyloidogenic Aβ (amyloid β-peptide) . OSBP overexpression caused accumulation of APP–Notched dimers in the Golgi apparatus, an effect that was reversed by addition of 25-OH. As Aβ levels are positively correlated with cholesterol [83,84], OSBP could reduce the sterol composition of membranes where APP processing occurs and attenuate Aβ production. Although it remains to be determined where and how OSBP exerts these effects, these results implicate OSBP in the control of cholesterol or oxysterol distribution in Golgi and post-Golgi compartments, thus regulating the activity of cholesterol-sensitive functions, such as secretases and ABCA1 (Figure 2).
Similar to OSBP, ORP9L localizes to the trans-Golgi/TGN and ER via its PH domain and FFAT motif respectively, and catalyses PtdIns4P-dependent cholesterol transfer between liposomes [17,51]. The sterol-binding and transfer activity of ORP9L was not linked to SM metabolism, and Golgi localization of ORP9L was not affected by oxysterols or altered cholesterol homoeostasis [17,51]. Silencing of ORP9L expression in CHO cells resulted in accumulation of cholesterol in filipin-positive endosomes/lysosomes but did not affect cellular cholesterol mass or synthesis, suggesting that cholesterol movement between the ER and Golgi apparatus is ultimately affecting other downstream compartments. Interestingly, ORP9L depletion also fragmented the Golgi apparatus. The TGN is an important delivery site for endocytosed cholesterol  and cholesterol overload or depletion strongly inhibits secretory activity [86–88]. Thus transport of cholesterol between the trans-Golgi/TGN and ER by ORP9L could be required to maintain optimal cholesterol levels in the secretory pathway (Figure 2).
Initial overexpression experiments showed that ORP2 increased cholesterol efflux to acceptors in the medium and decreased cholesterol esterification and increased synthesis, possibly due to Golgi localization [57,90]. However, ORP2 was recently shown to undergo sterol-dependent movement between the cytoplasm and lipid droplets; an ORP2 sterol-binding mutant was constitutively associated with lipid droplets but 22-hydroxycholesterol inhibited the association . Silencing of ORP2 expression by RNAi decreased TAG (triacylglycerol) hydrolysis and increased cholesterol ester formation, but the latter effect was only observed when lipid droplet formation was enhanced by oleate treatment . The sterol-dependent localization of ORP2, coupled with VAP- and PIP-binding activity, is suggestive of a transfer activity that controls neutral lipid synthesis in lipid droplets.
Mechanisms of sterol transport by the OSBP gene family
As OSBP/ORPs have the potential to interact with donor and acceptor membranes by independent targeting functions, an important question is how the sterol-loaded protein transverses the gap between membranes [91,92]. This could involve: (i) complete disengagement of OSBP/ORPs from membranes after binding and releasing sterol ligands (Figure 3A) or (ii) transfer while OSBP/ORPs are simultaneously engaged with both donor and acceptor membranes at contact sites (Figures 3B and 3C). The membrane contact site model is particularly alluring as most OSBP/ORPs have FFAT and PH domains that could mediate simultaneous contact with two closely apposed membranes. Sterol transfer is then predicted to occur by flip-flop of the sterolbinding domain between donor and acceptor membranes. Close apposition of the ER and trans-Golgi cisternae, where OSBP and ORP9L are predicted to localize, as well as other organellar membranes, has been identified by electon microscopy ultrastructure studies [93,94]. With the exception of ORP1L (see below), OSBP/ORPs have not been localized to, or directly implicated in, the formation of membrane contact sites, nor is there evidence that sterol transport occurs at contact sites. In the absence of PH and FFAT domains, OH domains from all Osh proteins form simultaneous contacts between two liposomal membranes enriched in anionic lipids that, in the case of Osh4p, greatly enhanced sterol transfer . Mechanistically this could involve dual membrane contacts at the entrance of the sterol-binding pocket and a distal lipid-binding site that would facilitate sterol uptake, followed by disengagement or ‘pivoting’ to deliver ligand to the acceptor membrane (Figure 3B). Even if membranes are sufficiently separated to prevent simultaneous contact, yeast or mammalian OSBP/ORPs could be concentrated at these sites where collisional transfer would be favoured. Figure 3(C) shows a potential variation of the contact site model wherein the transmembrane domain of ORP5 or 8 anchor the proteins to the donor site and the PH domain interacts with PIPs on the target membrane.
Mechanisms of sterol transfer and integration with sterol sensing functions
Another important consideration is the directionality of sterol transfer between donor and acceptor membranes. As an example, vectoral transfer of ceramide by CERT between two membranes is regulated by phosphorylation-dependent conformational changes. CERT contains a serine-rich phosphorylation site, adjacent to the PH domain, which negatively regulates its interaction with Golgi PtdIns4P . Initial phosphorylation of CERT by PKD (protein kinase D) in the Golgi apparatus is followed by phosphorylation at nine adjacent serine residues by CK1 (casein kinase 1) γ2, resulting in reduced Golgi localization, ceramide transport and SM synthesis [95,96]. In the ER, protein phosphatase PP2Cϵ (protein phosphatase 2Cϵ) interacts with VAP and dephosphorylates CERT leading to increased Golgi localization and SM synthesis . Collectively, this explains the ATP-dependence of ceramide transport  and shows that an ER and Golgi phosphorylation cycle drives the delivery of ceramide from the ER to the Golgi apparatus .
Given its close metabolic relationship to CERT it is feasible that OSBP is under a similar control by phosphorylation. Indeed, OSBP is also Golgi-localized and dephosphorylated in response to depletion of cellular SM and cholesterol [78,99]. The relevant OSBP phosphorylation site has multiple CK1 sites but is adjacent to the FFAT motif and does not contain a PKD consensus phosphorylation site. Independent of phosphorylation, OSBP binding of exogenous oxysterol in cultured cells results in Golgi localization, suggesting a conformational change that exposes the PH domain and increases interaction with Golgi PtdIns4P and other determinants [29,99] (Figure 3A). This could be a general mechanism as an ORP3 N-terminal/OSBP C-terminal chimaera showed increased localization to the PM in cells treated with 25-OH . It remains to be determined how phosphorylation and sterol-induced conformational changes are related to sterol-transport between membranes.
Sterol-dependent signalling by OSBP/ORPs
It is difficult to strictly segregate putative sterol-transport functions of OSBP/ORPs from those that are signalling-related. One could envision a scenario whereby sterol binding by OSBP/ORPs during transport to a target membrane also induces conformational signals that promote activation of ancillary signalling pathways (Figure 3). This would conveniently integrate cholesterol and oxysterol ‘sensing’ by growth regulatory pathways with intracellular cholesterol transport and distribution. Whereas conformational changes are induced by sterol binding to Osh4p , these are more consistent with the opening and closing of the binding pocket during ligand binding, a conclusion supported by molecular dynamic modelling [42,43]. Although there are differences in the lid region of Osh4p bound to side-chain oxysterols compared with cholesterol, sterol ligands of Osh4p are buffered from direct interaction with amino acid residues by water . This argues against a signalling mechanism that differentiates between apo and holo forms of the sterol-binding domain, but the possibility that sterol binding triggers subtle conformation effects that promote independent signalling cascades cannot be excluded.
An example of this apparent dual function is OSBP, which was identified as a sterol-sensing scaffold that regulates the dephosphorylation and inactivation of ERK (extracellular-signal-regulated kinase), a component of the MAPK (mitogen-activated protein kinase) pathways [34,100]. The cholesterol-bound form of OSBP is a scaffold for the serine/threonine phosphatase PP2A and HePTP (haematopoietic tyrosine phosphatase), both of which dephosphorylate and inactivate ERK. Removal of cholesterol or binding of 25-OH to OSBP causes the phosphatase complex to dissociate, allowing hyperphosphorylation and activation of ERK . Deletion mapping and pulldown experiments showed that PP2A and HePTP bind at non-overlapping sites in the C-terminal 400 amino acids of OSBP . A model was proposed wherein cholesterol extraction from membranes by a CRAC (cholesterol recognition/interaction amino acid consensus) motif , which encompasses the α-helical lid and β-sheets 1–3 of the predicted binding fold , in conjunction with the N-terminal glycine/alanine-rich region and PH domain, induced a conformation favourable for phosphatase recruitment. This process could be envisioned as one step in a sterol transfer pathway (Figure 3A).
Most of the experiments linking OSBP to ERK regulation were performed in vitro or by overexpression in cultured cells. However, phosphorylation of ERK was also decreased in liver and cultured hepatocytes infected with an adenovirus encoding OSBP, and was increased by silencing OSBP expression in hepatocytes . Adenoviral OSBP overexpression in murine liver elevated plasma VLDL (very-low-density lipoprotein), TAG and mRNA for Insig-1 and SREBP-1c, whereas RNAi suppressed insulin-mediated SREBP-1c processing and decreased TAG synthesis. This suggests that OSBP could regulate insulin-dependent TAG metabolism by modulating ERK activity.
In an apparently unrelated role that involves a similar scaffolding mechanism, stimulation of pro-atherogenic profilin-1 expression in endothelial cells by 7-oxocholesterol involves JAK2 (Janus kinase 2) phosphorylation of Tyr394 in OSBP and subsequent recruitment and phosphorylation of STAT3 (signal transducer and activator of transcription 3) . This provides a mechanism for oxysterol-mediated dysfunction of the cytoskeleton in endothelial cells but seems at odds with the relatively low affinity of OSBP for side-chain hydroxylated sterols [104,105].
In addition to a role in regulation of ER–Golgi cholesterol transport, ORP9L has been implicated in the regulation of Akt activity, raising the possibility that sterol transfer could be integrated with this important proliferative pathway. ORP9L and ORP9S contain a PDK2 (phosphoinositide-dependent kinase 2) phosphorylation site, termed the hydrophobic motif, that is required for activation of AGC-type kinases (protein kinase A/protein kinase G/protein kinase C-family kinases), such as Akt . In bone marrow mast cells, the hydrophobic motifs of ORP9S and Akt were co-ordinately phosphorylated, and RNAi-depletion of ORP9L in HEK (human embryonic kidney)-293 cells resulted in a 3-fold increase in phosphorylation of the Akt PDK2 site, but not the PDK1 site. . These results suggest that ORP9 is a negative regulator of Akt phosphorylation, but whether this is a direct interaction or secondary to altered sterol distribution is unknown.
Both ORP1L and ORP8 affect expression of ABCA1 and other LXR target genes by a possible signalling-related mechanism. Gene silencing of ORP8, by RNAi, increased ABCA1 expression in macrophages via up-regulation of LXR promoter activity . As ORP8 is an ER transmembrane protein, this could occur by transport at membrane contact sites, via an intermediary sterol carrier(s) or by altered synthesis of LXR ligands in the ER (Figure 3C). Interestingly, ABCA1 and ORP8 are reported to co-immunoprecipitate suggesting that interaction in the ER could be negatively regulating ABCA1 expression and activity . Overexpression of ORP1L enhanced [53,70] or repressed  the activity of LXR and expression of some target genes, indicating a cell-specific effect at the levels of sterol-transfer or signalling functions.
Evidence for involvement of OSBP/ORPs in cell proliferation, survival and differentiation has also been derived from other single and multicellular animal and plant models. A D. melanogaster OSBP homologue (OSBP-Dm) prevented Wee1p inhibition of p34cdc2 and cell-cycle progression when expressed in Schizosaccharomyces pombe . As D. melanogaster does not synthesize sterols, it was speculated that OSBP-Dm has a cell cycle/cell signalling function. In C. elegans, a truncated ORP missing the PH domain was identified as a novel regulator of the TGF-β (transforming growth factor-β) signalling pathway based on interaction with human BRAM [BMP (bone morphogenic protein) receptor-associated protein] . Inhibition of BIP (BRAM-interacting protein) expression by RNAi phenocopied TGF-β deletion and affected body length, suggesting that this ORP might have a broader role in C. elegans TGF-β signalling pathways. During D. discoideum culmination (the transition from the slug to a fruiting body) the expression of Dd-OSBPa was increased and enriched in the periphery of pstO cells (the rear two-thirds of a slug) . The formation of slugs was normal in OSBPa-null cells, however the culmination–fruiting body switch was defective. A screen linked to signalling functions lead to the identification of PiORP1 (Petunia inflata oxysterol binding-protein-related protein 1) as an interacting partner and substrate for receptor-like kinase PRK1 (pollen-expressed receptor-like kinase), an essential kinase in pollen development . PiORP1 and PRK1 expression coincide at the pollen development stage, but persist into later stages of plant growth. A salt-inducible OSBP homologue from soyabean (Glycine max) termed GmOSBP is expressed highly during middle–late phases of cotyledon senescence and could be involved in growth and the stress response . Similarly, the expression of a potato OSBP homologue lacking a PH domain is up-regulated following exposure to Phytophthora infestans, the cause of late blight . These studies point to important roles for OSBP/ORPs in diverse organisms but require further investigation to identify ligands and differentiate signalling from lipid transport/regulatory functions.
OSBP/ORP interaction with the cytoskeleton
An emerging theme in this field is the interaction of OSBP/ORPs, as well as their cognate binding partner VAP, with the three classes of cytoskeletal elements: actin, intermediate filaments and microtubules. Ultrastructural analysis of mouse hippocampal slices revealed that VAP was associated with the ER and microtubules and was frequently positioned between these two structures . Interaction of VAP with microtubules was required for regulation of ER membrane protein lateral diffusion  and bouton formation at neuromuscular junctions in D. melanogaster . The regulation of microtubule organization at neuromuscular junctions is particularly interesting because of the recent identification of a rare form of motor neuron disease/ALS8 (amyotrophic lateral sclerosis 8) that is caused by a VAP-B P56S mutation [111–113]. The physical interaction of VAP with microtubules appears to involve the FFAT domain, but could occur directly or indirectly through other FFAT-containing proteins, such as Nir2 [109,114].
A series of studies show that ORP1L is a link between VAP and the microtubule network. ORP1L recruits microtubule motors on to LEs (late endosomes)/lysosomes through ankyrin interactions with the GTP-bound form of Rab7 to promote clustering of LE/lysosomal compartments [53,70]. On the basis of FRET (fluorescence resonance energy transfer) and pulldown experiments, Rab7 recruits RILP (Rab7-interacting lysosomal protein) and ORP1L to LE/lysosomes to form a tripartite complex . The complex then recruits p150Glued, a component of the dynein–dynactin motor complex, which in turn drives minus-end microtubule transport of the LE/lysosomes. The cholesterol content in the outer membranes of LEs regulates the recruitment of the dynein–dynactin–p150glued complex by a sensing function of the ORP1L sterol-binding domain . When endosomal cholesterol is elevated, ORP1L binds cholesterol and promotes minus-end transport and clustering of endosomes around the microtubule-organizing centre, a situation that is observed in NPC (Niemann–Pick C) lysosomal cholesterol storage disease. Depletion of endosomal cholesterol enhances the interaction of ORP1L with its ER partner VAP, subsequently displacing dynein–dynactin–p150glue and dispersing LEs. The interaction between ORP1L and VAP seems to occur at ER–LE contact sites (Figure 3B), but an increase in contact sites under cholesterol-depleted conditions and the lack of effect of ORP1L on LE cholesterol levels lead the authors to conclude that ORP1L is primarily a cholesterol sensor .
Overexpressed ORP10 co-localizes with microtubules and induces a cable-like morphology that was dispersed by depolymerizing agents . Silencing of ORP10 in hepatoma cells increased cholesterol and TAG synthesis and VLDL secretion, cellular functions that could be related to its genetic linkage to high TAG traits in a Finnish cohort . This prompted speculation that the ORP10 interaction with microtubules could be linked to organelle positioning and secretion.
The intermediate filament protein vimentin is implicated in esterification and transport of LDL-derived cholesterol [116,117]. When overexpressed in CHO cells, ORP4S caused the vimentin network to collapse into thick bundles and inhibited esterification of LDL-cholesterol, suggesting that vimentin–ORP4 interactions could be involved in cholesterol delivery to the ER . The vimentin-binding site is within the sterol-binding domain of ORP4, but a leucine repeat N-terminal of the FFAT motif, which is absent in ORP4S, maintains the integrity of the intermediate filament network . OSBP with mutations in the leucine repeat also collapsed the vimentin network, but this was via dimerization with ORP4L and hence indirect. These novel interactions suggest that the lipid regulatory and/or transportation functions of certain ORPs can be mediated through interactions with the intermediate filaments.
ORP3 affects the organization of the actin cytoskeletal network through interactions with R-Ras . ORP3 and R-Ras co-immunoprecipitated and are concentrated at filopodial protrusions of HEK-293 cells. ORP3 overexpression inhibited integrin activity and enhanced polarized cell protrusions, which was dependent on PM-targeting by the PH domain. Other actin-regulated functions were impaired by ORP3 overexpression in primary macrophages, including cell spreading and podosome formation. ORP3 silencing increased integrin activity, resulting in enhanced spreading and extracellular matrix adhesion, a phenotype accompanied by reorganization of the actin cytoskeleton. Thus ORP3 seems to regulate cell adhesion, spreading and migration by inhibiting the Ras signalling pathway, which controls integrin activity and the actin cytoskeleton.
Interactions of ORPs with the cytoskeleton seem to fall into two classes: direct physical interactions that affect the network organization (ORP4 and ORP10) and those that are secondary to signalling-related mechanisms (ORP3 and ORP1L). It is conceivable that the cytoskeleton is a scaffold on which OSBP/ORPs transfer sterols between membranes that are contacted by these networks. Alternatively, sterol-binding/transfer activity could be integrated with these interactions as a means to impart sterol-dependent regulation of the cytoskeleton (Figure 3). More studies are required to determine whether this is a general feature of the OSBP/ORP family, restricted to specific subfamilies or members, or mediated by interaction with VAP.
INVOLVEMENT OF OSBP/ORPs IN HUMAN DISEASE
As OSBP/ORPs are implicated in regulation and transport of cholesterol and oxysterols, which mislocalize or accumulate in neurological and cardiovascular diseases, altered ORP expression or activity might be expected to accompany disease onset and/or progression. To date, human diseases that result from mutations in OSBP/ORP genes have not been identified. This is not surprising as functional redundancy within the family could prevent the expression of phenotypes associated with mutations in a single OSBP/ORP gene. However, evidence that OSBP/ORPs are involved in dyslipidaemias continues to grow, indicating a primary or secondary role in cellular sterol and lipid deposition.
OSBP/ORPs are frequently linked to LXR regulation of cholesterol efflux and disposal, a pathway that affects formation of arterial lesions resulting from accumulation of lipid-laden macrophage-derived foam cells. Differentiation of human monocytes into macrophages results in a >100-fold increase in the ORP1L mRNA and appearance of the protein in the endosomes/lysosomes. . OSBP and other ORPs are not similarly activated. Cholesterol-loading of human macrophages with acetyl-LDL further increased ORP1L expression by 75%  and increased ORP6 mRNA by 2-fold . The influence of ORP1L expression on macrophage cholesterol metabolism was examined by bone marrow transplant from macrophage-specific ORP1L transgenics into LDLR−/− mice. Following the feeding of a western-type diet, LDLR−/− mice expressing macrophages with a 15-fold increase in ORP1L expression had a 2.1-fold increase in aortic root lesion size despite having plasma lipid levels that were similar to controls . This was accompanied by reduced expression of ABCG1, ABCG5 and ApoE (apolipoprotein E), increased PLTP (phospholipid transfer protein) and interleukin-1β, and reduced efflux of macrophage cholesterol to HDL (high-density lipoprotein). ORP1L silencing in RAW224.7 cells also inhibited the up-regulation of ABCG1 expression by (22R)-OH treatment. Interestingly, overexpression of ORP1L in COS or HEK-293 cells increased LXRα reporter gene activity in response to (22R)-OH and non-sterol agonists, indicating that cell- and oxysterol-specific responses are modulated by ORP1L . As ORP1L is a sterol-dependent regulator of endosomal positioning , it could be involved in cholesterol egress from this compartment for conversion into the oxysterol activators of LXR.
ORP8 is also involved in atherosclerotic lesion formation as immunohistochemical analysis showed prominent staining of ORP8 in lesional CD68+ macrophages compared with those from a healthy arterial wall . RNAi silencing of ORP8 in macrophages increased ABCA1 expression and cholesterol efflux to ApoA1 (apolipoprotein A-I), was synergistic with a synthetic LXR agonist and inhibited by mutations in the LXR-response element DR4 (direct repeat 4) and the E-box. The transcriptional regulation of LXR target genes by ORP8 could involve sequestration of an endogenous sterol activator.
Studies have described altered OSBP/ORP expression in a variety of disease conditions, but, for the most part, functional links are absent. A microarray screen for monocyte genes affected by immune complexes containing oxLDL (oxidized LDL) identified OSBP as a potential mediator of pro-survival and inflammatory responses . A miRNA (microRNA) screen in oxLDL-stimulated monocytes identified an up-regulated ORP9-related miRNA that inhibited ORP9 expression, lipid uptake and inflammatory cytokine secretion . Altered OSBP/ORP expression occurs in a variety of seemingly unrelated pathological and experimental conditions: ORP1 is associated with genomic imprinting clusters in Beckwith–Wiedemann syndrome [121,122]; expression of ORP8 is a response to antipsychotic drug treatment ; increased ORP6 expression is observed during during neural lineage choice ; and ORP4 has been identified as a blood dissemination marker for tumour metastasis .
SNPs (single nucleotide polymorphisms) in subfamily VI (ORP10 and 11) have been implicated in dyslipidaemic disease. ORP11 expression is up-regulated in adipose tissues of men with a high risk of cardiovascular disease . Further investigation associated a number of SNPs in the ORP11 gene with cardiovascular disease risk factors, including hypertension, LDL-cholesterol plasma levels and hyperglycaemia . Polymorphisms in the ORP10 gene are linked to high TAG serum levels in Finnish dyslipidaemic subjects . The effect of these polymorphisms on ORP10 function is unknown, but RNAi silencing in Huh7 cells resulted in increased TAG and cholesterol biosynthesis, and secretion of ApoB100 (apolipoprotein B-100) and ApoA1 in the culture medium. These effects were not correlated with expression of known lipid metabolic enzymes, but could be related to organelle and lipid transfer mediated through interaction with microtubules.
VAP and OSBP have been implicated in the replication of HCV (hepatitis C virus), a single-strand RNA virus that assembles on ER membranes and lipid droplets . NS5A (non-structural protein 5A) and the RNA-dependent RNA polymerase NS5B (non-structural protein 5B) both interact with VAP-A and VAP-B on the ER to form a stable replication complex [129–131]. HCV replication in cells was significantly diminished by VAP silencing [129,132] or hyperphosphorylation of NS5A, which disrupted its interaction with VAP-A . OSBP also interacts with NS5A and is required for HCV replication in human hepatoma cells, an activity that appeared to involve targeting to the Golgi apparatus . VAP and OSBP interact with the N-terminus of NS5A, possibly to enhance viral replication by a scaffolding or membrane-targeting mechanism, or by facilitating the creation of an optimal membrane environment for viral assembly. The p48 protein of the single-strand RNA-encoded Norwalk virus (Norovirus) also interacts with VAP-A suggesting it is a commonly used host factor for viral replication .
Altering the expression of OSBP/ORP family members, or measuring gene expression in response to specific cues or agents, has identified seemingly diverse and sometimes discordant functions for this large gene family. However, recent evidence that mammalian and yeast ORPs can transfer sterols in vitro and in vivo, coupled with their conserved sterol transport-like domain, suggests that many of the observed functional outputs are the consequences of this fundamental activity. The major challenge in upcoming years will be to identify where OSBP/ORPs transfer sterols, the mechanisms that control directionality, endogenous ligand(s) and how the transfer is linked to a sterol-sensing function, either at the protein or membrane level.
amyloid precursor protein
BMP (bone morphogenic protein) receptor-associated protein
ceramide transfer protein
casein kinase 1
two phenylalanines (FF) in an acidic tract
hepatitis C virus
human embryonic kidney
haematopoietic tyrosine phosphatase
liver X receptor
non-structural protein 5A
- OH domain
OSBP homology domain
D. melanogaster OSBP homologue
Petunia inflata oxysterol binding-protein-related protein 1
protein kinase D
protein phosphatase 2
single nucleotide polymorphism
StAR (steroidogenic acute regulatory) protein-related lipid transfer
transforming growth factor-β
VAMP (vesicle-associated membrane protein)-associated protein
The work of our laboratory is supported by the Canadian Institutes for Health Research [grant number 62916] and a grant-in-aid from the Heart and Stroke Foundation of Canada (Nova Scotia).