Modulation of Phosphorylation of Tocopherol and Phosphatidylinositol by hTAP1/SEC14L2-Mediated Lipid Exchange

The vitamin E derivative, alpha-tocopheryl phosphate (αTP), is detectable in cultured cells, plasma and tissues in small amounts, suggesting the existence of enzyme(s) with α-tocopherol (αT) kinase activity. Here, we characterize the production of αTP from αT and [γ-32P]-ATP in primary human coronary artery smooth muscle cells (HCA-SMC) using separation by thin layer chromatography (TLC) and subsequent analysis by Ultra Performance Liquid Chromatography (UPLC). In addition to αT, although to a lower amount, also γT is phosphorylated. In THP-1 monocytes, γTP inhibits cell proliferation and reduces CD36 scavenger receptor expression more potently than αTP. Both αTP and γTP activate the promoter of the human vascular endothelial growth factor (VEGF) gene with similar potency, whereas αT and γT had no significant effect. The recombinant human tocopherol associated protein 1 (hTAP1, hSEC14L2) binds both αT and αTP and stimulates phosphorylation of αT possibly by facilitating its transport and presentation to a putative αT kinase. Recombinant hTAP1 reduces the in vitro activity of the phosphatidylinositol-3-kinase gamma (PI3Kγ) indicating the formation of a stalled/inactive hTAP1/PI3Kγ heterodimer. The addition of αT, βT, γT, δT or αTP differentially stimulates PI3Kγ, suggesting facilitated egress of sequestered PI from hTAP1 to the enzyme. It is suggested that the continuous competitive exchange of different lipophilic ligands in hTAPs with cell enzymes and membranes may be a way to make these lipophiles more accessible as substrates for enzymes and as components of specific membrane domains.


Introduction
The vitamin E derivative, alpha-tocopheryl phosphate (aTP), is formed in small amounts from alpha-tocopherol (aT) in cultured cells, plasma and animal tissues and is present in foods and tissues in amounts of nmol/g of extracted material [1,2,3,4]. For the phosphorylation reaction a putative aT kinase, and for the dephosphorylation reaction an aTP phosphatase or esterase can be postulated, and both activities have been detected in cells in culture or in tissues [1,2,3,5,6].
The negative charge of aTP renders it more similar to phosphorylated messenger lipids such as phosphatidylinositol phosphates, with possibly increased ability to modulate specific and non-specific protein-membrane interactions (reviewed in [7]). However, although regulatory effects of tocopheryl phosphate esters on enzymes have been reported early on [8], the in vivo biological function of aTP is not clear to date. aTP may act as an active cofactor for specific enzymatic reactions (reviewed in [9]), it may be a ligand of a receptor or transcription factor, or act as ''second messenger'' in the membrane capable of exerting regulatory effects [2]. In vivo, atherosclerotic lesions of hypercholesterolemic rabbits are more efficiently reduced by supplementation with aTP when compared to a-tocopheryl acetate (aTA), resulting from reduced cytokines and scavenger receptor expression [10,11]. The often higher potency of aTP when compared to aT is either due to a better uptake and cellular retention of the molecule e.g. by organic anion transporters (OAT) [12], to its intracellular hydrolysis by esterases [1,2,3,5,6], or to its preferential direct interaction with specific proteins and cellular structures such as with protein kinase C alpha (PKCa) [13], or similar to atocopheryl succinate (aTS) with Bcl-xL/Bcl-2 or mitochondrial succinate oxidase [14,15,16].
Since aT and to a lesser extent aTP are hydrophobic molecules located mainly in membranes, specific lipid transfer proteins (LTP) may be required to make them more accessible to kinases and phosphatases or to transport them to specific proteins, membrane domains and organelles. For the intracellular transport of aT, several proteins such as the microsomal triglyceride transfer protein (MTTP), the Niemann-Pick C1-like 1 protein, the atocopherol transfer protein (a-TTP) and three tocopherol associ-ated proteins (hTAPs) (hTAP1, hTAP2, hTAP3 or hSEC14L2, hSEC14L3, hSEC14L4, respectively) have been identified (reviewed in [17]). The three hTAPs are highly homologous and related to the Saccharomyces cerevisiae SEC14p protein, which is the prototype of a large eukaryotic family of proteins carrying a SEC14-lipid binding domain playing a role in lipid metabolism, signalling and membrane trafficking (reviewed in [18,19,20,21]). It has been postulated that these proteins stimulate signaling reactions by either directly transferring their ligands (e.g., phosphatidylinositol, phosphatidylcholine, squalene) to specific enzymes (e.g., PI3K, PI4K, phospholipase C, squalene epoxidase), by supplementing the membrane system occupied by these enzymes and regulating their activity by increasing their accessibility to further reactions [20,22]. More recently the LTP have been suggested to sense the lipid environment and regulate enzymes by obligatory homotypic or heterotypic lipid-exchange which enables lipid presentation to the catalytic center in enzymes where they react in a temporally and spatially coordinated manner [23].
The relatively large binding pocket of hTAPs (10262 Å 3 for hTAP1 [24]) can accommodate several different hydrophobic ligands that within cells may form a group of lipids competing for the same binding site. One group of lipids able to bind to hTAPs is related to vitamin E (a-tocopherol), encompassing the four natural tocopherol and tocotrienol analogues (a-, b-, c-, d-) as well as some derivatives such as a-tocopheryl quinone (aTQ) and a-tocopheryl succinate (aTS) [16,25,26,27]. An intracellular tocopherol transport function of these proteins is supported by the finding that the cellular uptake of aT and aTS is increased by hTAP1 overexpression [14,16], that the in vitro aT transport to mitochondria is augmented by hTAP1 [28], and that mitochondria-mediated apoptosis is induced by aTS in hTAP1-overexpressing mesothelioma cells and in prostate cancer cells [14,15,16].
In addition to tocopherol analogues and derivatives, hTAPs bind in vitro several other ligands, such as squalene, phosphatidylinositol (PI), phosphatidylinositol-3,4,5-phosphate, phosphatidylcholine (PC) and phosphatidylserine, suggesting transport of these ligands to specific enzymes or intracellular sites (reviewed in [19]). The competition with these ligands for a common binding site and their exchange could affect phospholipid-dependent transport and signalling pathways. Accordingly, aT stimulates in vitro squalene epoxidase and phosphatidylinositol-3-kinase gamma (PI3Kc) activity possibly by forcing the release of squalene or phosphatidylinositol, respectively, and/or facilitating their presentation to the enzymes [14,25,26,29,30]. Thus, aT may act as competing heterotypic ligand to PI or squalene as proposed for PC in stimulating the phosphorylation of PI by PI kinases [20].
In this study, we describe the existence of aT phosphorylation activity present in primary human coronary artery smooth muscle cells (HCA-SMC). To assess the substrate specificity of the putative aT kinase, we evaluate whether other vitamin E analogues can become phosphorylated as well. Moreover, we check whether hTAP1 can bind aT and aTP and whether aT phosphorylation and PI3Kc activity can be modulated by hTAP-mediated lipid exchange.

Cell proliferation assay
THP-1 cells were plated into 96-well microtiter plates (10,000 cells/well), treated with aTP and cTP and grown for 0, 28 and 52 h. Treatments in 96 well microtiter plates with aTP and cTP were done using working stock dilutions prepared in 1% ethanol in order to keep total ethanol concentrations in the cell culture medium below 0.1%. Compounds diluted for the working stock dilutions were assessed by thin layer chromatography (TLC) and no loss was observed as a result of dilution (e.g. as result of precipitation). Cell numbers were assessed using the CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Madison, WI), and measurements were done using a GLOmax absorbance reader (Promega) at 490 nm after assay duration of 4 h.

Purification of recombinant hTAP1 from Escherichia coli
Recombinant hTAP1 containing an amino-terminal Histidine tag was expressed and purified as previously described [26,27].

Binding of aTP to recombinant hTAP1
The binding of aT and aTP to recombinant hTAP1 was assessed using Isoelectric Point Mobility Shift (IPMS) assay essentially as previously described [27]. In this assay, the native hTAP1 protein migrates on an isoelectric focusing polyacrylamide gel until it has a net charge of 0 (what occurs at the calculated isolectric point of recombinant hTAP1 at pH 7.9 [27]), and the mobility of hTAP1 is changed upon ligand (PI) binding, until the PI-hTAP1 complex reaches again a net charge of 0.

UPLC assay
The a-tocopheryl phosphorylation assay was performed as above, the products separated on TLC, and a control spot for aTP and two labelled sample spots scraped and extracted with ethanol. The samples were analyzed on a Waters UPLC fitted with a 1.7 mm 2.16100 mm C18 bridged ethane linked hybrid column. The solvent chosen was A) water containing 4.0 g/L ammonium bicarbonate, and B) methanol. The flow rate was at 0.4 ml/min and column temperature was 40uC. The following gradient was used (Table 1).

Phosphatidylinositol kinase assay
The in vitro phosphatidylinositol kinase assay was performed using recombinant PI3Kc/p110c basically according to the protocol supplied by the manufacturer (Alexis Biochemicals, San Diego, CA). Briefly, sonicated phosphatidylinositol (100 mM), tocopherols (50 mM) and the recombinant hTAP1 protein (100 nM) were preincubated for 10 min in a total volume of 100 ml reaction buffer (20 mM Tris-HCl (pH = 7.4), 4 mM MgCl 2 , 100 mM NaCl) containing 20 mM cold ATP and 10 mCi -[c-32 P]-ATP (Amersham Biosciences). The reaction was started by adding 0.2 mg PI3Kc/p110c (50 nM) and incubated at 37uC for 20 min. The reaction was stopped with 150 ml of 1 M HCl, and the phospholipids extracted with 400 ml chloroform/methanol (1:1), separated by TLC, exposed to film and quantified as previously described [26].
Transfection THP-1 cells (1.5610 6 cells per ml) were grown in 12 well plates (1 ml per well) overnight, transfected with pCGCG-luc (a reporter plasmid containing 3169 bp of the human VEGF promoter in front of the Firefly luciferase gene (kindly provided by S. J. Prior, University of Maryland, Baltimore, MD [35])), and with the Renilla internal control plasmid pRL-TK (Promega, Madison, WI), for 3 h using Fugene (Promega) as transfection reagent, and then treated with aT, cT, aTP or cTP (all 20 mM) for additional 21 h. Extracts were prepared, and promoter activities were measured using the Dual-Luciferase assay kit (Promega) using a GLOmax luminometer (Promega). The VEGF promoter-Firefly luciferase activities were normalized to the thymidine kinase promoter-Renilla luciferase activities, and the activities of the control transfections were set to 100%.
Western blotting THP-1 monocytes (1.5610 6 cells in 10 ml media per dish) were grown overnight and then treated with aT, aTP and AS-605240 for 24 h as indicated in the figure legend. The cells were harvested, centrifuged, washed with ice cold PBS, incubated at 4uC for 5 min in 0.5 mL cell lysis buffer (20 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM b-glycerol phosphate, 1 mM sodium orthovanadate, 1 mM PMSF, 1/1000 diluted TABS protease inhibitor cocktail (Roche, Indianapolis, IN)), homogenized 10 times using a G26 needle and centrifuged for 10 min at 16000 rcf at 4uC. The protein concentration was measured using the BCA kit (Pierce, Rockford, IL). Immunoblots were done according to standard methods using 30 mg of extract per lane and separated by 10% SDS-PAGE. The level of Akt phosphorylation was determined using primary anti-phospho-Akt(Ser473) antibody, primary anti-Akt antibody (both from Cell Signalling Technology, Danvers, MA), and horseradish peroxidase coupled donkey antirabbit IgG secondary antibody (Amersham Biosciences, Piscataway, NJ). Proteins were visualized with an enzyme-linked chemiluminescence detection kit (Immun-Star HRP) according to the manufacturer's instructions (Biorad, Hercules, CA). Chemiluminescence was monitored by exposure to film (Kodak BioMax), and the signals were analyzed using a Fluorchem 8900 workstation and the AlphaEaseFC software (AlphaInotech).

Statistical analysis
All values are expressed as the mean 6 standard error of the mean (SEM) as explained in the figure legends. The median fluorescence intensity was determined for FACS analysis and the mean 6 SEM calculated as described in the figure legends. Student's t-test was used to analyze the significant differences between two conditions. A p,0.05 was considered as significant and indicated by * or # in the graphs.

Phosphorylation of aT in primary human coronary artery smooth muscle cells
In preliminary studies we have shown that small amounts of aT can become phosphorylated in vitro by HMC-1 human mast cells and primary human coronary artery cells [2], as well as in NIH-3T3-L1 adipocytes and in rat livers upon feeding 14 C-aT [3]. To characterize the enzymatic reaction involved, an in situ aT phosphorylation assay measuring the production of aTP from aT and [c-32 P]-ATP was performed with primary human coronary artery smooth muscle cells (HCA-SMC), the newly formed aTP extracted and separated by Thin Layer Chromatography (TLC) ( Figure 1A). After that, the labelled spot corresponding to aTP was scraped from the TLC plate and analyzed by Ultra Performance Liquid Chromatography (UPLC) as described in Materials and Methods. In the UPLC graph, the peaks from the isolated spots corresponded with the aTP control peaks, clearly showing that aTP is synthesized in our in vitro assay system ( Figure 1A). A very weak spot was observed in the absence of added aT, reflecting some aT in the serum. The phosphorylation of aT occurred in a concentration dependent manner. No phosphorylation occurred with aTP suggesting that the pyrophosphate is not formed. The phosphorylation of a-tocopheryl quinone (aTQ) and cT was measured as well, although with lower efficiency ( Figure 1B and C). By comparing the intensity of the radioactive aTP spots with spots obtained from diluting 59-[c-32 P]-ATP (6000 Ci/mmol) of known concentration, it was calculated that ,168 molecules/cell/hour were synthesized in this assay.

Comparison of cellular activities of aTP and cTP
Since both aTP and cTP are formed in the in vitro assay, it was interesting to determine whether the two compounds affect cells with different potency, what could contribute to the activity differences seen with aT and cT in THP-1 monocytic leukaemia cells [4,36] and other experimental systems despite a generally lower cT level (reviewed in [37]). When THP-1 cells were incubated with either aTP or cTP at increasing concentrations for 4, 28 or 52 h, cTP inhibited their proliferation more efficiently than aTP (Figure 2A); concentrations of cTP above 20 mM led to cell loss due to cytotoxic/apoptotic effects, what occurred with aTP only at concentrations above 46 mM [31]. Similar to that, cTP inhibited CD36 scavenger receptor surface exposition stronger than aTP ( Figure 2B). It remains to be shown whether the higher activity of cTP contributes to the higher activity of cT when compared to aT observed in a number of experimental models such as apoptosis, cell proliferation, gene expression, cancer and inflammation (reviewed in [38]).
Binding of a-tocopherol and a-tocopherol phosphate to human tocopherol associated protein 1 (hTAP1) is associated with release of bound phosphatidylinositol hTAP1 can bind several uncharged hydrophobic ligands (such as tocopherols, tocotrienols, phosphatidylcholine, phosphatidylserine and squalene), but also charged ligands (such as aTS, phosphatidylinositol (PI) and phosphatidylinositol-3,4,5-phosphate) (reviewed in [19]). It was therefore important to check whether it can also bind aTP, which with calculated pK a values of 6.07 and 1.64, is expected to carry two negative charges with physiological condition and occurs in solution as di-sodium salt [39]. Indeed, when assayed in vitro by Isoelectric Point Mobility Shift (IPMS) assay [27], aT could compete with PI for binding to recombinant hTAP1 suggesting that the two ligands bind and depending on their concentration can exchange each other at an overlapping binding site ( Figure 3A). Since 50% displacement of PI (125 mM) was observed with aT at 50 mM, the affinity of aT to the binding pocket of hTAP1 is stronger than that of PI. When compared to aT ( Figure 4A), the competition with aTP was slightly weaker ( Figure 3B). As negative control, another hydrophobic molecule, ritonavir, was not able to compete, showing the specificity of this assay ( Figure 3A).

The a-tocopheryl phosphorylation reaction is stimulated by recombinant hTAP1
Having established that aT and aTP both can bind to hTAP1, it was important to assess whether hTAP1 facilitates the phosphorylation reaction of aT. Indeed, the addition of recombinant hTAP1 (3 and 15 mg/2.5 ml ICB) stimulated the aT phosphorylation reaction in a concentration dependent manner ( Figure 3C).

The phosphatidylinositol-3-kinase gamma activity is stimulated in vitro by aT and aTP in an hTAP1-dependent manner
In a previous study, recombinant hTAP1 reduced the in vitro activity of the phosphatidylinositol-3-kinase gamma (PI3Kc); the addition of aT stimulated PI3Kc, e.g. by forcing egress of PI from hTAP1 to the enzyme and/or by inducing conformational changes leading to activation of PI3Kc [14,26]. To assess whether different tocopherol analogues influence PI3Kc activity with different potency, we measured in vitro the activity of recombinant PI3Kc in the presence of aT, bT, cT and dT. All four tocopherols stimulated PI3Kc activity with similar efficiency ( Figure 4A).
In the presence of recombinant hTAP1, in vitro PI3Kc activity was reduced (to 38624%, n = 3, P,0.05) what could be the result of direct hTAP1/PI3Kc interaction and/or formation of an inactive/stalled complex ( Figure 4A) [26]. In the presence of hTAP1 the different tocopherol analogues showed different potency to stimulate PI3Kc ( Figure 4A), suggesting that hTAP1 not only reduces PI3Kc activity, but also gives a certain selectivity to the tocopherols to activate PI phosphorylation by PI3Kc, e.g. as a result of different binding affinity, ligand exchange rate or ligand induced conformational changes.
Since aTP can also bind hTAP1 ( Figure 3B), it was important to determine whether hTAP1 can influence the ability of aTP to stimulate PI3Kc activity. aTP stimulated PI3Kc stronger than aT ( Figure 4B). In the presence of hTAP1 the fold induction of PI3Kc activity seen with aTP was even higher, despite having a slightly lower ability to compete with PI ( Figure 3A and 3B), what may aT, aTP, cT, and cTP differentially up-regulate vascular endothelial growth factor promoter activity in THP-1 cells We previously reported that aTP stimulates the PI3K/Akt signal transduction pathway, leading to the induction of a number of genes including the vascular endothelial growth factor (VEGF) [4]. To assess whether PI3Kc is the PI3K isoform regulating VEGF expression in THP-1 monocytes, these cells were treated with AS-605240 (1 mM), an inhibitor specific for PI3Kc [40,41]. As measured by Western blotting, Akt(Ser473) phosphorylation was strongly inhibited by AS-605240 (to 34.1615.7, n = 3, p, 0.05) and the stimulation by aTP was blocked ( Figure 5A), suggesting that PI3Kc is a predominant PI3K isoform present in these cells [40], and therefore is involved in regulating Akt and VEGF by aTP [4]. To assess whether aT, aTP, cT, and cTP differentially up-regulate VEGF promoter activity in THP-1 cells, a reporter construct containing the human VEGF promoter in front of the luciferase gene was transfected into THP-1 monocytes and VEGF promoter activity measured. Both aTP and cTP significantly activated the VEGF promoter with similar potency, whereas aT and cT had no significant effect in these cells ( Figure 5B).

Discussion
We show here that aT and cT is phosphorylated in HCA-SMC suggesting presence of enzyme(s) with aT phosphorylation activity in these cells. The aT phosphorylation activity is stimulated by recombinant hTAP1 which binds aT and aTP thus facilitating the transport and presentation of the substrate (aT) to the putative aT kinase and/or the removal of the product (aTP) away from it, thus increasing the enzyme's catalytic turnover. The in vitro activity of PI3Kc is inhibited by hTAP1 indicating the formation of an inactive hTAP1/PI3Kc heterodimer [26]. The binding of PI to hTAP1 is reversed by aT and aTP leading to stimulation of PI3Kc activity, suggesting that aT and aTP promote dissociation of the inactive complex and/or the release of sequestered PI from hTAP1 for subsequent presentation to the kinase by means of a heterotypic lipid exchange mechanism. Although it is possible that presentation of aT to the putative aT kinase occurs by homotypic exchange, it should be noted that aT phosphorylation activity was assayed in the presence of permeabilized cells possibly allowing heterotypic ligand exchange when encountering the cellular plasma membranes [20]. Analogous lipid-exchange mechanisms were recently visualized with the crystal structures of the closest SEC14p homolog -the Saccharomyces cerevisiae Sfh1a [42], as well as proposed for human a-TTP [43,44]. While our in vitro and cell culture studies are focusing only on PI3Kc and aT kinase, other enzymes such as PI4K, phospholipases, squalene epoxidase, fatty acid synthase, choline-phosphate cytidyltransferases could be modulated by hTAPs-mediated lipid exchange as well [30]. At a molecular level, the transfer of PI/PC by Saccharomyces cerevisiae SEC14p function has been mainly linked with activation of PI4K, secretion and trafficking of lipid raft proteins [42,45]. However, none of the three hTAP1/2/3 proteins was able to complement for SEC14p function in yeast [26], and direct interaction of the hTAPs with PI3K and modulation of its activity in vitro and in vivo in mice and humans suggests that these proteins are performing a regulatory function different from yeast SEC14p [14,26]. hTAPs may affect gene expression in a tocopherol-and/or tocopheryl phosphate-dependent manner, e.g. by affecting the PI3K/Akt signal transduction pathway by transporting these ligands to specific enzymes such as cytosolic PI3Kc, or to membrane sites accessible for regulating PI3K, Akt, and PHLPP1 [4,46,47,48]. Whether similar signalling events also contribute to the regulation of the biosynthesis of cholesterol by TAP1/SEC14L2 by regulating squalene epoxidation via stimulating squalene transport and presentation to squalene epoxidase remains to be investigated [29,30].
The exchange of hTAP ligands may be a way to make these lipophiles more accessible as substrates for enzymes and as components of specific membrane domains (lipid rafts, vesicles, organelles) ( Figure 6A and 6B). Each hTAP may show different preferences for specific lipids and enzymes what determines which lipids are exchanged and which reaction is catalysed. It has to be kept in mind that the activity measured in our assay represents the sum of many binary on/off switches at individual hTAP1/PI3Kc molecules and their response to aT and aTP. In cells, in which hTAP1/PI3Kc interaction occur dynamically in time and space, hTAP1 may act as sensor for lipid information (location, type and amount of lipid, lipid gradients) and generate a self-organizing system able to respond to changes in extra-and intracellular lipids and transmit this information into responses of PI3K-mediated signalling and gene expression. In fact, the higher concentration of vitamin E in plasma membrane domains (e.g. lipid rafts) is In vitro PI3Kc activity is modulated by recombinant hTAP1 (4 mg) in a tocopherol analogue specific manner. PI3Kc activity was assessed as described in materials and methods and the mean6 SEM results plotted (n = 3, *P,0.05 relative to untreated control (c) without hTAP1; # P,0.05 relative to aT in the presence of hTAP1). aT, bT, cT, dT: a-, b-, c-, d-tocopherols, respectively. W: wortmannin. (B) In vitro PI3Kc activity is inhibited by wortmannin (W) (1 mM), and stimulated by aT (50 mM) and more by aTP (50 mM). Recombinant hTAP1 (4 mg) inhibits PI3Kc activity possibly by forming a stalled/ inactive complex; addition of aT or aTP reverts the inhibition by hTAP1, possibly by promoting dissociation of the inactive complex and/or competing with bound phosphatidylinositol allowing its egress from the hTAP1 binding site and the transfer to the enzyme. PI3Kc activity was assessed as described in materials and methods, the control set to 100% and the mean6SEM plotted (n = 3, *P,0.05 relative to control (c)). doi:10.1371/journal.pone.0101550.g004 compatible with specific mechanisms for tocopherol insertion and removal from membranes, vesicles and organelles [49,50] and may define the sites at which aT or aTP mediated lipid exchange and signaling can occur. Enhanced tocopherol delivery to membranes may be required in cells at risk for tocopherol depletion, such as in epithelial duct cells of secretory glands or airway ciliated epithelial cells exposed to high levels of oxygen, in which the hTAPs are abundantly expressed [28,51]. Enrichment of aT, aTP, or PI3P/ PI4P at specific sites may also contribute to the aT-mediated increase of hexosaminidase secretion in rat mast cells [52], and/or determine the identity of vesicles required for intracellular trafficking [53]. It is noteworthy that the related Saccharomyes cerevisiae SEC14p mediates vesicle formation and secretion from endosomes and the trans-Golgi Network (TGN) to the vacuole [54], and it can be speculated that the lumen of the epithelial ducts represents a cellular space that is homologous to the yeast vacuole.
As described in this study, aTP facilitates better lipid exchange in hTAPs when compared to aT, thus stimulating better hTAPsdependent reactions by enhancing the egress and presentation of heterotypic ligands to enzymes in the correct spatial orientation. The cellular response to aT and aTP may depend on whether they enable catalysis of PI by PI3K or PI4K, or the conversion of aT to aTP by a putative aT kinase or vice versa by an aTP phosphatase in a given tissue and cell type [4]. We find that aTP (and more so cTP) is more potent than aT in reducing cell proliferation, and in normalizing oxLDL-induced CD36 mRNA and protein expression [31] as well as CD36 cell surface exposition [4]. Since CD36 mediates signal transduction and gene expression of ligands that increase VEGF expression (e.g. oxLDL and oxidized lipids [55,56,57]), these changes in CD36 expression and localization may further contribute to the regulatory effects of aT or aTP on VEGF expression [58,59]. Interestingly, one ligand of CD36, thrombospondin, negatively regulates VEGF expression and angiogenesis [60] as well as myristic acid uptake and signalling [61], and it remains to be determined whether aT or aTP uptake and interaction with and internalization of CD36, and possibly interference with thrombospondin binding, play some role in stimulating VEGF expression.
The production of aTP in vascular smooth muscle cells (VSMC) may instruct neighbouring pericytes/endothelial cells or invading monocytes/macrophages to produce VEGF leading to an increase of vascular permeability and/or adaptive formation of new vessels [62,63], e.g. during post-infarction wound healing [64] or during development acting as tubulogenic morphogen during vasculoand/or nephro-genesis [65]. Whether activation of PI3K/Akt/ VEGF and angiogenesis/vasculogenesis by aTP mediates the essential function of aT to prevent fetal resorption and ischemia/ reperfusion injury in placenta, embryo, brain and muscle remains to be further investigated [66]. It appears possible that usage of aT/aTP-induced PI3Kc-mediated signalling to enhance angiogenesis e.g. during placentation and embryogenesis has evolved to ensure sufficient amounts of aT to prevent free radicals damage upon formation of functional oxygen-transporting blood vessels. In the mammary gland epithelium, an increase in VEGF expression by aTP could also assist in the expansion of the vascular and lobulo-alveolar system during pregnancy and lactation, and increase capillary permeabilization required to increase the production of milk with sufficient VEGF and aT to nurse pups [67,68]. It is interesting to note that in the mammary gland TAP proteins are expressed specifically in epithelial duct cells where they may take part in regulating these events [28].
In summary, the activities described with hTAP1/aT-kinase and hTAP1/PI3Kc fit well into a model proposed for Saccharomyces cerevisiae SEC14p-related proteins [20,23,69,70]. However, whether these reactions play a role for in vivo signalling function of aT and aTP requires the cloning of an aT kinase as well as an aTP phosphatase. Ligand exchange for sites within hTAPs could be a way to enhance PI3Kc (or other enzymes) -dependent lipid reactions and increase their specificity in time and space. In doing so, hTAP/PI3K can act as sensor for cellular lipid information (location, type and amount of lipid) and translate it into responses in signalling and gene expression. It can also be envisioned that hTAPs catalyze lipid reactions not only at nano-scale for lipid transfer and signalling in cells, but also at larger scale for applications in biotechnology. Further research is required to identify the putative aT kinase, to establish the biological function Figure 6. Hypothetical molecular model for hTAPs in lipid transport and enzyme regulation. (A) hTAPs transfer lipids from/to cellular import/export sites or between different membranes and membrane domains such as lipid rafts, e.g. between membranes of the Golgi, endoplasmic reticulum, mitochondria, vesicles or membranes of cilia in airway epithelia [28,51]; in secretory cells lipid transfer may be polarized. (A and B) hTAPs mediated lipid transport may change lipid composition and membrane curvature and in this way influence signal transduction and secretion. (B) hTAPs bring lipid substrates (S) to specific enzymes (E), present them in the correct orientation and timing, and/or remove the lipid products (P) from the enzyme, thus enhancing lipid turnover at the catalytic center (CC). Lipid exchange may occur preferentially upon interaction of hTAPs with membranes, thus confining lipid presentation by hTAPs and subsequent lipid modification to enzymes located to membranes. Moreover, the affinity of different ligands to the ligand binding pocket can influence lipid exchange rate thus influence lipid-specificity to stimulate enzyme activity. The carboxy-terminal GOLD (G) domain in hTAPs may confine the exchange activity to certain sites and thus further increase the reaction specificity. doi:10.1371/journal.pone.0101550.g006 SEC14L2-Mediated Lipid Exchange and Signalling by Vitamin E PLOS ONE | www.plosone.org of aTP and cTP and the role of the three hTAPs in lipid transport, signal transduction and gene expression.