Impaired α-TTP-PIPs Interaction Underlies Familial Vitamin E Deficiency

See allHide authors and affiliations

Science  31 May 2013:
Vol. 340, Issue 6136, pp. 1106-1110
DOI: 10.1126/science.1233508

Vitamin E Out

Familial vitamin E deficiency is caused by mutations in the α-tocopherol transfer protein (α-TTP) gene. Kono et al. (p. 1106, published online 18 April; see the Perspective by Mesmin and Antonny) studied natural mutations in α-TTP. α-TTP bound phosphatidylinositol polyphosphates (PIPs), especially PI(4,5)P2, and a disease-related missense mutation abolished PIP binding but not α-tocopherol binding. The x-ray crystal structure of the α-TTP–PIP complex suggested that PIP binding opens the lid of the α-tocopherol–binding pocket to facilitate the release of α-tocopherol. Thus, PIP binding to α-TTP at the target membrane may facilitate the release of α-tocopherol in the hydrophobic pocket of α-TTP to the lipid bilayer of the target membrane, providing a mechanism for the transfer of lipids from the lipid-transfer protein to the target membrane.


α-Tocopherol (vitamin E) transfer protein (α-TTP) regulates the secretion of α-tocopherol from liver cells. Missense mutations of some arginine residues at the surface of α-TTP cause severe vitamin E deficiency in humans, but the role of these residues is unclear. Here, we found that wild-type α-TTP bound phosphatidylinositol phosphates (PIPs), whereas the arginine mutants did not. In addition, PIPs in the target membrane promoted the intermembrane transfer of α-tocopherol by α-TTP. The crystal structure of the α-TTP–PIPs complex revealed that the disease-related arginine residues interacted with phosphate groups of the PIPs and that the PIPs binding caused the lid of the α-tocopherol–binding pocket to open. Thus, PIPs have a role in promoting the release of a ligand from a lipid-transfer protein.

Intracellular lipid transport is required for numerous cellular events (1, 2). Lipids are transported between organelles by vesicles or are delivered by lipid-transfer proteins (3). Some lipid-transfer proteins possess specific organellar-targeting motifs to assure precise lipid transport from donor to acceptor organelles (1, 2). However, many other lipid-transfer proteins have no known organellar-targeting domains, and the molecular bases underlying intracellular lipid transport by these proteins are largely unknown.

α-Tocopherol transfer protein (α-TTP), which specifically binds α-tocopherol (α-Toc), the most abundant form of vitamin E in mammals, is expressed in the liver where it regulates the amount of α-Toc secreted into the plasma (4). Heritable mutations in the α-TTP–encoding gene result in ataxia with vitamin E deficiency (AVED), an autosomal recessive disorder associated with low circulating vitamin E concentrations and neurodegenerative pathology (5). More than 20 mutations in the α-TTP gene have been identified in AVED patients (5). α-TTP is a member of the Sec14-like protein family (6) and has a lipid-binding domain, the Sec14 domain. Because α-TTP has no known organellar-targeting domain, we focused on the mutations in AVED patients to investigate the molecular mechanism underlying α-TTP–mediated α-Toc transport in liver cells. Of the nine disease-associated missense mutations (Fig. 1A), three (R59W, R192H, and R221W) are located in one region on the α-TTP protein surface, which is distinct from the α-Toc binding site (7) (Fig. 1B). The R59W and R221W mutations give rise to the severe, early-onset form of the disease (8), indicating that these arginine residues are critical for α-TTP function.

We assessed the α-Toc binding and intermembrane transfer activities of recombinant R59W α-TTP (R59W) in vitro. α-[3H]Toc comigrated with the wild type or R59W on gel filtration, and the peaks of α-Toc of R59W and the wild type were almost the same (Fig. 1C). The intermembrane α-Toc transfer activity of R59W was somewhat higher than that of the wild type (Fig. 1D). The wild type stimulated secretion of α-Toc from hepatoma cells, but R59W did not stimulate it at all (9) (Fig. 1E). Thus, R59W can bind and transfer α-Toc in vitro, but cannot stimulate α-Toc transport in cells.

Fig. 1 R59W α-TTP mutant impairs α-Toc secretion without affecting α-Toc binding and transfer.

(A) Distribution of α-TTP mutations. Blue and green boxes indicate CRAL_TRIO_N (N) and Sec14 domain, respectively. Missense mutations are shown below the panel, and insertions, deletions, and splicing mutations are indicated above the panel. (B) A loop representation of α‐TTP (PDB: 1R5L). α-Toc is in yellow. The side chains of R59, R192, and R221 are depicted as stick models. (C) α-[3H]Toc binding assay of wild type (WT) or R59W. Immunoblot of α-TTP in each fraction is also indicated. (D) Intermembrane α-Toc transfer activity of WT or R59W. (E) α-Toc secretion from McA-RH7777 cells stably transfected with WT or R59W. Protein amounts of α-TTP and α-tubulin were evaluated by immunoblot. Error bars indicate SE (n = 3 biological replicates). Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

Using beads derivatized with phosphatidylinositol 3,4-bisphosphate [PI(3,4)P2], α-TTP bound PI(3,4)P2 as expected (10), but R59W did not (Fig. 2A). Native polyacrylamide gel electrophoresis (PAGE) analysis of the α-TTP-PI(3,4)P2 mixture revealed that the PIPs increased the mobility of α-TTP (Fig. 2B), which also suggests that α-TTP interacts with PI(3,4)P2. The mobility shift of α-TTP was increased more by PI(4,5)P2 than by PI(3,4)P2. It was also slightly increased by PI(4)P and PI(3,4,5)P3 (Fig. 2, C and E, and fig. S1). None of the above PIPs affected the mobility of R59W (Fig. 2, D and F, and fig. S1). Thus, α-TTP binds PI(4,5)P2 and PI(3,4)P2 and the disease-associated R59 is involved in the α-TTP-PIPs interaction. Gel filtration chromatography of a mixture of α-TTP and [3H]PI(4,5)P2 showed that PI(4,5)P2 induced the formation of α-TTP tetramers (Fig. 2G). [3H]PI(4,5)P2 comigrated with both the monomer and the tetramer. Sedimentation velocity analysis also confirmed the tetramer formation by PI(4,5)P2 (fig. S2). The stoichiometry between α-TTP and PI(4,5)P2 was ~1:1.

Fig. 2 R59W α-TTP mutant impairs PIPs binding.

(A) Binding of WT and R59W to PI(3,4)P2-derivertized beads. (B) Native PAGE analysis of α-TTP-PI(3,4)P2 interaction. Gel was stained with silver. U, unshifted; S, shifted. (C) Same as in (B), except that PI or various PIPs were tested. (D) Native PAGE of WT or R59W with PI or various PIPs at a concentration of 2.5 μM. (E) Same as in (C), except that gels were stained with SYPRO ruby and band shifts were quantified. (F) Same as in (E), except that R59W was used. (G) [3H]PI(4,5)P2 binding assay. Immunoblot of α-TTP in each fraction was also indicated. For (E) and (F), error bars indicate SE (n = 3 independent reactions).

Crystals of the ternary complex consisting of α-TTP, α-Toc, and either PI(3,4)P2 or PI(4,5)P2 were analyzed by x-ray crystallography at 2.6 or 2.0 Å resolution, respectively (table S1). The crystallographic asymmetric unit contained four α-TTP molecules related by noncrystallographic 222 symmetry. Each α-TTP monomer bound both α-Toc and PIPs (Fig. 3A and fig. S3A). α-Toc was located deep in the hydrophobic core of α-TTP, as expected (7, 11). PIPs lay in close proximity to the bound α-Toc. The inositol phosphate (IP) head group was bound in the positively charged cleft (Fig. 3B and fig. S3B). The head groups of PI(3,4)P2 and PI(4,5)P2 were clearly defined in the electron density maps, whereas the diacyl moieties of the PIPs were not visible. Only the glycerol backbone and several carbon atoms of the acyl chains were visible between the hydrophobic groove formed by α9 (residues 165 to 185) and α10 (residues 198 to 221, termed the "lid") in the α-TTP-PI(3,4)P2 complex. The positively charged cleft of α-TTP, which accommodates the negatively charged IP head group of PIPs, was composed of three distinct regions in the amino acid sequence of α-TTP (residues 58 to 68, 184 to 191, and 214 to 222; Fig. 3A and fig. S3A). The side chains of R59, R68, D185, K190, R192, T215, K217, and R221 formed the inner wall of the cavity (Fig. 3C and fig. S3C). The three disease-associated arginine residues (R59, R192, and R221) all interacted with PIPs. The side chain of R59 formed a salt bridge with D185 and attracted negatively charged phosphates of PIPs. The side chains of R192 and R221 interacted with the 4-phosphate of the PIPs. Like the side chains of R59, R192, and R221, the side chain of K217 interacted with the 5-phosphate of PI(4,5)P2 or the 3-phosphate of PI(3,4)P2 (Fig. 3C and fig. S3C). Substituting K217, which is not a disease-related residue, with alanine did not affect in vitro α-Toc binding and intermembrane transfer activities (Fig. 3, E and F). However, K217A mutant did not stimulate α-Toc secretion in hepatoma cells (Fig. 3G). None of the PIPs caused a significant mobility shift of K217A in native PAGE (Fig. 3, H and I, and fig. S1). Thus, as inferred from the crystal structure, K217 is also required for both PIPs binding and the cellular α-Toc transport activities of α-TTP. Taken together, α-TTP binds PIPs through the positively charged cleft where the disease-associated arginine residues are clustered and PIPs binding is required for the cellular transport of α-Toc.

Fig. 3 Crystal structure of α-TTP-PI(4,5)P2 complex.

(A) Stereo view of the structure of α-TTP in complex with α-Toc and PI(4,5)P2. PI(4,5)P2 and α-Toc are shown as stick models: C in yellow, O in red, and P in orange. Some α helices are indicated with their labels, and some amino acid residues with their numbers. The "lid" helix is in blue, and three amino acid segments forming a positively charged cavity are differently colored: residues 58 to 68 in salmon, 184 to 191 in magenta, and 214 to 222 in cyan. (B) The structure of the PI(4,5)P2 complex in electrostatic potential surface representation, and residues of interest, are indicated. Positive and negative potentials are in blue and red, respectively. (C) PI(4,5)P2 and residues surrounding the head group as stick models. Carbon atoms of protein residues are colored as in (A). Residues of interest and phosphate groups of PI(4,5)P2 are indicated. The 2FO FC difference electron density map for PI(4,5)P2 is contoured at 1.5σ level in gray mesh. (D) Superposition of the open (blue, PDB: 1OIZ), closed (gray, PDB: 1OIP), and PI(4,5)P2-bound (green) structures is shown in Cα models. (E) α-[3H]Toc binding assay of WT or K217A. Immunoblot of α-TTP in each fraction is shown below. (F) Intermembrane α-Toc transfer activity of WT or K217A. (G) α-Toc secretion from McA-RH7777 cells stably transfected with WT or K217A. Protein amounts of α-TTP and α-tubulin were evaluated by immunoblot. Error bars indicate SE (n = 3 biological replicates). (H) Native PAGE analysis of WT or K217A with PI or various PIPs. Gel was stained with silver. U, unshifted; S, shifted. (I) Same as in (H), except that gels were stained with SYPRO ruby and band shifts were quantified. Error bars indicate SE (n = 3 independent reactions).

In the α-TTP-PI(4,5)P2 complex, the hydrophobic groove between α9 and α10 had a volume of 1147 Å3, which is large enough to accommodate the acyl chains. The groove was lined with residues with hydrophobic side chains in α9 and α10 (Fig. 3B). Hence, the two acyl chains are probably confined to the groove and are in contact with these residues through van der Waals interactions. Crystal structures of α-TTP in complex with α-Toc [Protein Data Bank (PDB) ID: 1OIP and 1R5L] and in complex with Triton X-100 (PDB ID: 1OIZ) are known (7, 11). In the α-TTP-α-Toc complex, bound α-Toc is isolated from the solvent by closing the lid (closed conformation). In contrast, in the α-TTP-Triton X-100 complex, α-TTP adopts a conformation with an open lid (open conformation). A superposition of the open, closed, and PIPs-bound structures of α-TTP (Fig. 3D and fig. S3D) suggests that the α-TTP-PIPs complex is intermediate between the closed and open conformations. Thus, binding of the acyl chains of PIPs to α-TTP may open the lid of the hydrophobic groove that contains α-Toc.

In vitro, addition of PI(4,5)P2 or PI(3,4)P2 to the acceptor liposomes increased the transfer of α-Toc by the wild type (Fig. 4A). Other negatively charged lipids such as cardiolipin or phosphatidylserine did not increase the transfer (fig. S4). The PIPs did not promote the α-Toc transfer by R59W and K217A (Fig. 4B). PIPs in the donor liposomes decreased the α-Toc transfer by the wild type (Fig. 4A). The antibiotic neomycin, which strongly binds to PI(4,5)P2, suppresses PI(4,5)P2-related events by masking PI(4,5)P2 (12, 13). Indeed, neomycin inhibited the in vitro α-TTP-PI(4,5)P2 interaction (Fig. 4C). Neomycin treatment inhibited α-Toc efflux in α-TTP–expressing hepatoma cells, but did not significantly affect α-Toc efflux in the cells that did not express α-TTP (Fig. 4D), suggesting that PI(4,5)P2 is involved in α-TTP–mediated α-Toc efflux from hepatocytes.

Fig. 4 PIPs promote α-TTP–mediated α-Toc transfer.

(A) Intermembrane α-Toc transfer assay of α-TTP using donor or acceptor liposomes doped with PIPs. (B) Intermembrane α-Toc transfer assay of WT, R59W, or K217A using acceptor liposomes doped with PIPs. α-Toc transfer activity was presented as a percentage of control (no PIPs). (C) Native PAGE of α-TTP with PI(4,5)P2 in the presence of neomycin (Neo). Gel was stained with silver. (D) McA-RH7777 cells or McA-RH7777 cells stably transfected with WT α-TTP were assayed for α-Toc secretion in the absence or presence of 10 mM neomycin. Expression levels of α-TTP and α-tubulin were evaluated by immunoblot (below). Error bars indicate SE (n = 3 biological replicates). *P < 0.05 (unpaired t test, two-tailed). (E) Scheme of α-Toc/PI(4,5)P2 exchange assay. (F) α-TTP–dependent gain or loss of α-Toc and PI(4,5)P2 in the precipitated liposomes. ppt, precipitate; lip., liposomes.

The yeast sterol transfer protein Osh4p binds PI(4)P and acts as a sterol/PI(4)P exchanger in vitro (14). The similarity between Osh4p and α-TTP in the mode of PIPs binding and the effect on the transfer activity is notable. Moreover, like Osh4p, α-TTP acted as an α-Toc/PI(4,5)P2 exchanger: α-TTP bound α-Toc and PI(4,5)P2 in a competitive manner (fig. S5) and transferred them between liposomes in the opposite direction (Fig. 4, E and F). Gel filtration analysis of α-TTP with PI(4,5)P2 and α-[3H]Toc showed that α-[3H]Toc bound mainly to the monomer and, to a lesser extent, to the tetramer (fig. S6). Given that the tetramer consists of PI(4,5)P2-bound α-TTP, the result suggests that α-TTP can bind both α-Toc and PI(4,5)P2 simultaneously. The α-TTP-α-Toc-PIPs complex in the crystal structures may represent a transient intermediate between the α-Toc–bound and PIPs-bound forms. We thus suggest a potential mechanism of intracellular α-Toc transport by α-TTP: α-TTP containing α-Toc first interacts with the head group of PIPs in the target membrane through the positively charged cleft. The PIPs is then transferred to α-TTP, which leads to opening the lid of the α-Toc–binding pocket and stimulates the transfer of α-Toc to the target membrane. As a result, α-TTP exchanges α-Toc for PIPs.

PI-binding motifs have been proposed from the crystal structure of the yeast Sec14 homolog 1 (Sfh1)–PI complex (15) and are considered to be biologically important in humans, as well as in yeast (16). We have shown that the disease-associated basic residues in the motif interact with PIPs and that this interaction is essential for the function of α-TTP. Several Sec14-like proteins bind PIPs (10, 1720). At least three of the basic amino acid residues that interact with the phosphate groups of PIPs in α-TTP are conserved (fig. S7). R103 and R234 in cellular retinaldehyde binding protein (CRALBP) are mutated in hereditary retinopathy and are also present in the predicted PI-binding motifs (21). R103 and R234 of CRALBP correspond to R59 and K190 of α-TTP, respectively (fig. S7), suggesting that binding of PIPs to these amino acid residues has a role in transporting 11-cis-retinal.

In liver cells, α-TTP is thought to catalyze the transfer of endocytosed α-Toc to the plasma membrane (PM) where a transporter secretes it into the circulation (22). Given that α-TTP interacted preferentially with PI(4,5)P2 and PI(3,4)P2, both of which are concentrated in the PM, we hypothesize that α-TTP transfers α-Toc to the PM by targeting these PIPs. Thus, PIPs binding may be required for α-TTP both to target the PM and to stimulate the release of α-Toc.

Supplementary Materials

Materials and Methods

Figs. S1 to S7

Table S1

References (2333)

References and Notes

  1. Acknowledgments: We thank P. T. Hawkins for providing PI(3,4)P2-derivatized beads; K. Yamamoto for providing Ricinus communis agglutinin 120; the beamline staffs at Photon Factory for their assistance with data collection; and J. Raymond for English proofreading. This work was supported by the Core Research for Evolutional Science and Technology, Japan Science and Technology Agency (CREST, JST) (to H.A.); the Program for Promotion of Basic and Applied Researches for Innovations in Bio-oriented Industry (to H.A.); Grants-in-aid from the Japanese Ministry of Education, Culture, Sports, Science, and Technology (20370045 to H.A.); and the Japanese Ministry of Health, Labor, and Welfare (to H.A.). The coordinate and structure factor data of α-TTP-PI(3,4)P2 and α-TTP-PI(4,5)P2 complexes have been deposited to the Protein Data Bank under the accession codes 3W67 and 3W68, respectively.
View Abstract

Navigate This Article