All-trans-Retinol Generated by Rhodopsin Photobleaching Induces Rapid Recruitment of TIP47 to Lipid Droplets in the Retinal Pigment Epithelium

Abstract

purpose. To investigate the effect of light stimulation on lipid droplets (LDs) and LD proteins in the retinal pigment epithelium (RPE).

methods. Dark-adapted mouse eyes were exposed to intense flashes of light, and ARPE-19 cells were treated with all-trans-retinol. The two specimens were labeled with BODIPY493/503 for LDs and with antibodies for three LD proteins: adipocyte differentiation-related protein (ADRP), TIP47, and Rab18. The labeling intensity in fluorescence microscopy was quantified by image analysis. Localization of mutated TIP47 was also examined. Immunoelectron microscopy was performed for ADRP in mouse RPE. Expression of TIP47 in ARPE-19 cells was knocked down by RNA interference (RNAi), and its effect on retinyl ester storage was measured by HPLC.

results. Both flashes of light on mouse eyes and all-trans-retinol on ARPE-19 cells caused rapid translocation of TIP47 from the cytosol to LDs, whereas ADRP distributed constitutively in LDs. The density of LDs did not show visible changes by any treatment. The localization of TIP47 to LDs was abolished when either the amino-terminal or the carboxyl-terminal half of the molecule was deleted, but was enhanced by a short deletion in the carboxyl terminus. Manipulation of TIP47 expression by RNAi or cDNA transfection did not affect the retinyl ester amounts in ARPE-19 cells significantly.

conclusions. All-trans-retinol generated by photobleaching in the retina induces rapid translocation of TIP47 to LDs in the RPE.

Lipid droplets (LDs) are ubiquitous cytoplasmic structures found in virtually all cells. They are composed of the core of lipid esters and the surface of the phospholipid monolayer. ^{1 2} Recent studies have revealed that there are several functional proteins in LDs. The proteins found in LDs or in LD-rich fractions include caveolins, ^{3 4 5} enzymes of eicosanoid synthesis, ⁶ enzymes of cholesterol synthesis, ⁷ Rab proteins, ⁸ and signaling proteins ⁹ among others. ^{10 11 12 13} The presence of these proteins as well as other results suggested that LDs play important roles in a variety of cellular activities, including intracellular lipid trafficking, lipid metabolism, signal transduction, membrane biogenesis, and protein degradation. ^{14 15 16 17} Besides the proteins listed herein, LDs harbor PAT proteins, which are named after perilipin, adipocyte differentiation-related protein (ADRP), and TIP47. ¹⁸ Among the PAT proteins, expression of perilipin, S3-12, and MLDP are limited to adipocytes and steroidogenic cells, whereas ADRP and TIP47 are found prevalently in many cell types. Perilipin has been shown to shield lipid esters from lipases in the resting condition and, at the same time, to be essential for stimulated lipolysis (for a recent review, see Ref. ¹⁹ ). Despite the similarity in the amino acid sequence, and also in the putative three-dimensional molecular structure, ²⁰ the functions of other PAT proteins in LDs have not been clear. 

Recently, a structure called the retinosome, or retinyl ester storage compartment, was shown to harbor retinyl esters in the retinal pigment epithelium (RPE), and was proposed to be involved in the retinoid cycle in the retina. ²¹ The retinosome was similar to LDs in harboring ADRP. But, in contrast to the round shape of LDs in other cell types, the retinosome has been described by confocal microscopy as an ellipsoid with its longer axis running along the cell’s apicobasal aspect and has been observed by electron microscopy as a vacuole-like structure near the cell–cell boundary. ²¹ These unique characteristics of the retinosome prompted us to study whether other LD-associated proteins are expressed and related to the retinyl ester storage function in RPE. 

In the present study, we examined three LD proteins—ADRP, TIP47, and Rab18—in the RPE. In the first part, we examined mouse RPE in vivo, and found that intense light caused rapid translocation of TIP47 to ADRP-positive structures in dark-adapted eyes. Immunoelectron microscopy showed that the ADRP-positive structure has the morphologic characteristics of canonical LDs. In the second part, we used ARPE-19 cells as an in vitro model of RPE, and showed that all-trans-retinol caused rapid translocation of TIP47 to LDs. These two results indicated that all-trans-retinol generated by photobleaching of rhodopsin in the retina induced the redistribution of TIP47 in RPE. The molecular domains of TIP47 responsible for its distribution to LDs, and the effect of its knockdown or overexpression on the retinyl ester storage were also examined in ARPE-19 cells. 

Materials and Methods

Cell Culture

Human ARPE-19 cells were obtained from American Type Culture Collection. The cells were cultured in a 1:1 mixture of Dulbecco’s modified Eagle’s medium (DMEM) and nutrient mixture F12, supplemented with 10% fetal bovine serum (FBS), 50 U/mL penicillin, and 0.05 mg/mL streptomycin in 5% CO₂/95% air at 37°C. When appropriate, all-trans-retinol (Sigma-Aldrich, St. Louis, MO) was added to the culture medium. Retinol was bound to methyl-β-cyclodextrin in DMEM in a 1:1 molar ratio and diluted to 1 or 3 mM by the culture medium. 

Animals

All animal experiments conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Eight-week-old male albino mice, ddy, and BALB/c strains, were fed water and food ad libitum. They were kept in an illuminated room for light-adapted eyes and in total darkness for more than 24 hours for dark-adapted eyes. Mice were anesthetized by inhalation of diethyl ether and then by subcutaneous injection of pentobarbital, and the eyes were enucleated. Dark-adapted eyes were processed in the dark room under dim red light until fixation. To examine changes caused by light, dark-adapted mice were exposed to intense flashes of light from a photographic flash unit (DCR-PC105; Sony, Tokyo, Japan), and processed 30 minutes later. The amount of rhodopsin was measured as described. ²²  

Antibodies

Rabbit anti-human TIP47 antibody ²³ and rabbit anti-Rab18 antibody ⁸ were raised and purified as described previously. Rabbit anti-mouse TIP47 antibody was raised using a peptide of mouse TIP47 segment (309-322), and affinity-purified using a peptide column. The specificity of the antibody was examined by Western blot analysis. Rabbit anti-mouse ADRP, mouse anti-ZO-1, and mouse anti-lysobisphosphatidic acid antibodies were donated by Tom Keenan (Department of Biochemistry, Virginia Polytechnic Institute and State University, Blacksburg, VA), Mikio Furuse (Department of Cell Biology, Kobe University Faculty of Medicine, Kobe, Japan), and Toshihide Kobayashi (Institute of Physical and Chemical Research [RIKEN], Saitama, Japan) respectively. Mouse anti-human ADRP (Progen, Heidelberg, Germany), Lamp1 (Developmental Studies Hybridoma Bank, the University of Iowa, Iowa City, IA), EEA1 (BD Transduction Laboratories, Lexington, KY), and TGN46 (Serotec, Oxford, UK) were purchased. Biotinylated horse anti-goat IgG antibody (Vector Laboratories, Burlingame, CA), anti-IgG antibodies and streptavidin conjugated with fluorochromes (Invitrogen-Molecular Probes, Eugene, OR), fluoronanogold-conjugated anti-rabbit IgG antibody (Nanoprobes, Yaphank, NY) were also purchased. 

Immunofluorescence Microscopy

ARPE-19 cells cultured on coverslips were fixed with a mixture of 3% formaldehyde and 0.025% glutaraldehyde for 10 minutes, permeabilized with 0.01% digitonin for 30 minutes, and treated with 3% bovine serum albumin (BSA) for 10 minutes. The cells were incubated with antibodies to ADRP, TIP47, or Rab18, and then with Cy3-conjugated secondary antibodies for visualization. LDs were stained using BODIPY493/503 (Invitrogen-Molecular Probes). Nuclei were visualized using DAPI (4′,6′-diamino-2-phenylindole). Images were acquired with confocal laser-scanning microscope (Pascal) or a fluorescence microscope (Axiophot2) equipped with a digital camera (AxioCam; all from Carl Zeiss Meditec, Jena, Germany). Quantitative analysis of the ADRP and TIP47 labeling was performed by using ImageJ software (available by ftp at zippy.nimh.nih.gov/ or at http://rsb.info.nih.gov/ij/; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD). ¹⁷ A threshold was set to visualize the LD labeling alone, and the proportion of positive pixels in the cell area was measured. The obtained value thus reflects the LD area positive for the respective proteins. 

For the mouse RPE preparation, the anterior segments and vitreous body were removed, the neural retina was peeled off, and the remaining eyecups were fixed with a mixture of 3% formaldehyde and 0.025% glutaraldehyde for 10 minutes. After permeabilization with 0.01% digitonin for 2 hours, the samples were quenched with 1 mg/mL sodium borohydride for 10 minutes and treated with 3% BSA for 30 minutes. They were labeled with antibodies and BODIPY493/503 in a manner similar to labeling of the cultured cells. This protocol labels TIP47 in LDs and the trans-Golgi network, but hardly in other organelles. ²³ The samples were placed on glass slides and observed en face by the laser confocal scanning microscope. 

Immunoelectron Microscopy

Immunoelectron microscopy of mouse eyecups was performed by both post- and pre-embedding methods. In the postembedding method, they were fixed in a mixture of 3% formaldehyde and 0.01% glutaraldehyde in 0.1 M PIPES (pH 7.4; piperazine-N-N′-bis(2-ethanesulfonic acid)) for 30 minutes, infiltrated with 2.3 M sucrose in the same buffer, and rapidly frozen by plunging them into liquid propane (−170°C) in a cryofixation unit (KF 80; Reichert, Vienna, Austria). The samples were then immersed in 1.5% uranyl acetate dissolved in anhydrous methanol (–90°C), replaced with Lowicryl HM20 at −45°C, and polymerized as described. ²⁴ Ultrathin sections were processed for immunogold electron microscopy using rabbit anti-mouse ADRP antibody followed by colloidal gold–conjugated secondary antibody. For the pre-embedding method, the eyecup was fixed in the same manner and permeabilized by 0.01% digitonin for 30 minutes. After blocking, they were incubated with rabbit anti-mouse ADRP antibody followed by fluoronanogold-conjugated goat anti-rabbit IgG antibody, fixed, and treated with a solution for autometallographic enhancement (GoldEnhance; Nanoprobes). 

cDNA Transfection and RNA Interference

Full-length and various truncation mutants of human TIP47 were amplified by PCR and cloned to the pcDNA3.1-TOPO-V5/His vector (Invitrogen, Carlsbad, CA). Mutants with several regions substituted with alanines were generated by a mutagenesis kit (QuickChange Site-Directed Mutagenesis Kit; Stratagene, La Jolla, CA), and cloned similarly. ²⁵ The vectors were transfected to ARPE-19 cells (Lipofectamine 2000; Invitrogen) and processed for immunofluorescence microscopy 2 days later. The expression of proteins of expected sizes was confirmed by Western blot analysis. 

Small interfering (si)RNA duplexes (Smart pool siGENOME duplexes; Dharmacon Research, Boulder, CO) were used to knockdown the expression of TIP47 and ADRP. A control RNA duplex (siControl Non-Targeting siRNA) was also obtained from Dharmacon Research. The siRNAs were transfected into cells (RNAiFect; Qiagen, Valencia, CA), and the cells were harvested 3 days later for analyses. 

Western Blot Analysis

ARPE-19 cells were lysed in 2.5% SDS, and 15 mM Tris-HCl [pH 8.0], by heating at 70°C for 3 minutes. Protein concentration was measured by bicinchoninic acid (BCA) assay (Pierce, Rockford, IL) and an equal amount of proteins (30 μg) were electrophoresed in 15% polyacrylamide gels. They were electrotransferred to nitrocellulose membrane and probed by rabbit anti-human TIP47 and mouse anti-human ADRP antibodies. After incubation with horseradish peroxidase (HRP)–conjugated second antibodies (Pierce), the blots were developed in substrate (Super Signal West Dura Substrate; Pierce). 

Quantification of Retinoids in ARPE-19 Cells

ARPE-19 cells confluent in a 100-mm culture dish were dispersed by trypsinization, pelleted, and extracted with organic solvents as described previously. ²⁶ The extract was subjected to high-performance liquid chromatography, and the amount of retinol and retinyl ester was measured. ²⁷ The retinoids were identified on the basis of the retention time, and the spectra of the peaks were monitored with a diode-array detector (L-2450; Hitachi, Ltd., Tokyo, Japan). The ratio of absorbances at 330 and 350 nm were compared with those of retinoid standards. The extraction and the analyses of retinoids were performed under dim red light. 

Results

LD Proteins in Mouse RPE In Vivo

The anti-ADRP and anti-Rab18 antibodies used were characterized in previous studies. ^{8 28} The specificity of the anti-mouse TIP47 antibody was confirmed by Western blot analysis (Fig. 1A) : it reacted with a 47-kDa band, but not with a 52-kDa band that was recognized by an anti-mouse ADRP antibody. The anti-mouse TIP47 antibody did not bind to overexpressed EGFP-mouse ADRP. 

Distribution of LD proteins in mouse RPE in vivo was examined by a confocal laser scanning microscope. When the eyecup was prepared from mice kept in normal light conditions, LDs stained by BODIPY493/503 were observed as prominent dots in the RPE cytoplasm. Both ADRP and TIP47 were labeled in virtually all LDs, whereas Rab18 was not detected (Fig. 1B) . In accordance with our previous observation, there were a few LDs that lacked labeling for ADRP or TIP47. By double labeling with other organelle markers that are listed in Materials and Methods, TIP47 was occasionally seen in the trans-Golgi network, but the labeling intensity was much weaker than that in LDs. BODIPY493/503-stained LDs and the labels for ADRP and TIP47 were concentrated near the cell–cell boundary in many cells, but they were also seen in other parts of the cytoplasm. By examining confocal micrographs (e.g., Fig. 1B ), 66.1% (n = 548) of the ADRP-positive dots were distributed within 2 μm from the edge of the cell, and the rest were seen deeper in the cytoplasm. 

To study whether the distribution of LD proteins changes after stimulation with light, we exposed mice kept in total darkness for more than 24 hours to intense flashes of light, and the eyes were fixed 30 minutes later. Retinas of dark-adapted ddy mice contained 0.16 nM (n = 2) of rhodopsin, but it became undetectable after stimulation, indicating that virtually all rhodopsin was bleached by the procedure. The eyes of mice kept in the dark until death were examined for comparison. In the dark-adapted eyes, LDs were observed in a density similar to that seen in light-adapted eyes and were labeled positively for ADRP (Fig. 2A) . In contrast, labeling for TIP47 in LDs was minimal in dark-adapted eyes (Fig. 2B) . Thirty minutes after light stimulation, LDs and ADRP labeling did not show a change (Fig. 2A) . In contrast, the TIP47 labeling around LDs increased drastically. In this sample, virtually all LDs became positive for TIP47 (Fig. 2B) , indicating that ADRP and TIP47 were localized to the same LDs. An increase in TIP47 expression could not be excluded as a cause of the increased LD labeling, but considering the result in the ARPE-19 cells (discussed later), we speculate that TIP47 translocated from the cytosol to the LDs on stimulation with light. These results were obtained in both ddy and BALB/c mice. 

By post-embedding immunoelectron microscopy of Lowicryl sections, labeling of ADRP was localized to round, electron-lucent structures (Fig. 3A) . The structures had a smooth contour, did not have a limiting membrane, and appeared identical with LDs observed in other cell types. We also performed pre-embedding immunoelectron microscopy using nanogold-labeled antibody and gold enhancement. By visualizing the nanogold particles by gold enhancement, 66.6% of the labeling was found around canonical LDs (Fig. 3B) . The rest of the labeling was seen in a scattered manner in the cytoplasm and was not associated with any specific structure. Virtually all the canonical LDs labeled positively for ADRP under these conditions. Most of the ADRP-positive LDs were usually seen near the cell boundary, which was consistent with the observations by confocal laser microscopy. Some specimens were fixed with a stronger fixative, embedded in Epon, and subjected to morphologic observation, which confirmed the presence of canonical LDs near the cell boundary (Fig. 3C) . By both immunoelectron microscopy and conventional thin-section electron microscopy and irrespective of the direction of sectioning (i.e., either horizontal or vertical to the RPE monolayer), the LDs appeared as spherical structures. This result implies that most ADRP-positive structures in mouse RPE are canonical LDs, but it does not exclude the possibility that some ADRP could be present in other structures that were not preserved well by the present protocol. 

LD Proteins in ARPE-19 Cells

To analyze the mechanism of TIP47 redistribution in RPE, we adopted human ARPE-19 cells as an in vitro model system. Because the antibodies used for the mouse RPE did not react with human proteins, a separate set of antibodies that were raised against human ADRP and TIP47 were used. ^{23 28}  

ARPE-19 cells are known to retain structural and functional properties of RPE in vivo and to show polarized epithelial morphology when grown on permeable supports. ²⁹ The polarity of the cell may be compromised when cultured on coverslips, but the cells still grew as a monolayer showing a cobblestone appearance. The formation of tight junctions was confirmed by labeling with ZO-1 (data not shown). In the initial stage of experiments, we compared ARPE-19 cells cultured on cell-migration filters (Transwell; Corning, Inc., Corning, NY) with those grown on coverslips, but distribution of LDs and LD proteins was not different by immunofluorescence microscopy (data not shown). 

In ARPE-19 cells cultured in the normal culture medium, LDs were observed as small dots throughout the cytoplasm. By double labeling, ADRP was seen around most LDs (Fig. 4) . In contrast, TIP47 was observed around LDs in less than 10% of cells, and LDs in the rest of the cells were devoid of TIP47 labeling (Fig. 4) . Even in cells showing TIP47-positive LDs, there were usually other LDs lacking TIP47. The absence of TIP47 in most LDs in quiescent cells has been observed in other cell types. ^{25 30} Rab18 was not labeled positively in ARPE-19 cells (data not shown). 

In the mouse retina in vivo, intense light bleaches rhodopsin and causes a rapid increase in all-trans-retinol, which is then supplied to RPE and converted to retinyl esters. ³¹ We assumed that all-trans-retinol induced the redistribution of TIP47 after light stimulation, because fatty acids have been shown to induce a similar translocation of TIP47 as well as lipid ester formation. ^{25 30} To test this assumption, all-trans-retinol (1 or 3 mM) was given to ARPE-19 cells, and its effect was examined after 10 minutes and 2 hours. As described later, quantification of retinyl esters confirmed that the administered all-trans-retinol was esterified in ARPE-19 cells. 

Even after the addition of all-trans-retinol, ADRP persisted in virtually all LDs and the labeling intensity did not show a significant change (Fig. 5A) ; quantification of the labeling intensity showed a slight increase in ADRP present in LDs at 10 minutes after the 3 mM all-trans-retinol administration, but not in the other samples (Fig. 5C) . In contrast, the proportion of LDs positive for TIP47 began to increase as early as 10 minutes after the addition of all-trans-retinol and further increased 2 hours after the addition (Fig. 5B) . Quantification showed significant changes in all the samples, and the labeling intensity reached 26 times as much as the untreated control at 2 hours after the administration of 3 mM all-trans-retinol (Fig. 5C) . Because the labeling of TIP47 in the trans-Golgi network was negligible in ARPE-19 cells, the increase caused by the all-trans-retinol administration can primarily be attributed to the labeling in LDs. 

To examine the possibility that all-trans-retinol upregulated TIP47 expression and thus caused the increased labeling in LDs, total TIP47 expression was estimated by Western blot analysis. The result indicated that the expression of TIP47 in ARPE-19 cells did not change significantly by administration of all-trans-retinol (Fig. 5D) . Altogether, the results indicated that all-trans-retinol induced the rapid accumulation of TIP47 in LDs and that this effect was caused by its translocation from the cytosol. 

Domains Necessary for TIP47 Localization to LDs

To identify the molecular domain of TIP47 necessary for LD localization in RPE, several deletion mutants were constructed, and their cDNAs were transfected into ARPE-19 cells. The mutants as well as the full-length TIP47 were expressed with a V5 tag at the carboxyl terminus, and transfected cells were labeled with an anti-V5 antibody. The expression of the proteins was monitored by Western blot analysis (Supplementary Fig. S1). With the present labeling protocol, cells expressing V5-tagged constructs were identified by diffuse cytoplasmic labeling. The ratio of cells showing LD localization among transfected cells with or without the all-trans-retinol treatment was counted for each mutant (Fig. 6) . The result showed that deletion of either the amino- or the carboxyl-terminal half completely abolished localization to LDs. In contrast, deletion of short segments in the carboxyl terminus, or replacement of hydrophobic amino acids that were shown to make the putative hydrophobic cleft ²⁰ caused the LD localization, even without the addition of all-trans-retinol. The result was in accordance with the result that we obtained in other cell types using nontagged TIP47 mutants. ²⁵ In the same study, we found that the V5-tagged molecules occasionally showed intense cytoplasmic labeling that was never observed for endogenous TIP47 or nontagged TIP47 mutants. ²⁵ We speculate that the addition of a V5 tag partially perturbed the targeting mechanism, and this explains why the V5-tagged, full-length TIP47 showed a relatively high ratio of LD localization even in control cells. Nevertheless, the overall result indicates that TIP47 in RPE is targeted to LDs by using the same signal used in other cell types. 

Retinol Esterification and TIP47

The results of the mouse RPE in vivo and in ARPE-19 cells inferred that TIP47 is translocated to LDs when a bolus of all-trans-retinol is supplied to RPE. We speculated that the rapid redistribution of TIP47 may be involved in some aspects of retinyl ester formation and/or storage. To explore this possibility, the expression of TIP47 in ARPE-19 cells was manipulated by RNA interference or cDNA transfection, and its effect on the amount of retinyl ester in the cell was examined. 

In control cells cultured in the normal medium, the amount of retinoids was minimal as expected (Fig. 7A) . When ARPE-19 cells were treated with 1 or 3 mM all-trans-retinol, the amounts of retinol and retinyl ester increased drastically (data not shown). When treated with 1 mM all-trans-retinol for 30 minutes or with 3 mM all-trans-retinol for 2 hours, the increase in retinol and retinyl esters in cells transfected with the control siRNA was similar to that seen in nontransfected cells (Fig. 7A) . When cells were treated with siRNA for TIP47 knockdown, the protein expression of TIP47 was reduced to <30%, as estimated by Western blot analysis (Fig. 7B) . In those cells, the amount of retinol and retinyl ester after the all-trans-retinol administration (1 mM, 30 minutes; 3 mM, 2 hours) was decreased compared with levels in the control, but the difference was not statistically significant (Fig. 7A) . A similar result was obtained in three independent experiments. The inconclusive result could be partly caused by a large amount of TIP47 existing as a soluble protein in the cytoplasm, and even after the significant reduction of TIP47 by RNAi, the residual amount may be sufficient to sustain its function in LDs. Consistent with this assumption, when the expression of TIP47 was increased more than 10-fold by cDNA transfection, the amount of retinyl esters did not change significantly from that in control cells (data not shown). 

Discussion

TIP47 and LD in RPE

Light causes photobleaching of rhodopsin and isomerizes 11-cis-retinal to all-trans-retinal, which is then reduced to become all-trans-retinol. ³¹ When dark-adapted eyes are exposed to intense light, a large amount of all-trans-retinol is generated in photoreceptor cells, and it is taken up and esterified in RPE before reisomerization to the cis form for further rounds of the retinoid cycle. Imanishi et al. ²¹ showed that retinyl esters accumulate rapidly in specific structures of RPE when dark-adapted eyes are exposed to intense light. 

In other cell types such as fibroblasts and hepatocytes, we and others ^{25 30} have observed that addition of fatty acids increases the lipid ester storage in LDs and, at the same time, causes changes in LD proteins. In these latter instances, glycerols, including monoacyl- and diacylglycerols, and cholesterol should be used for synthesis of triacylglycerides and cholesterol esters, respectively, whereas in RPE, fatty acids should be mobilized to synthesize retinyl esters. Despite differences in what was supplied exogenously and what was mobilized from endogenous pools, the increased storage of lipid esters and the translocation of TIP47 to LDs were observed similarly. 

The mechanism that drives the rapid translocation of TIP47 from the cytosol to LDs is not known. Apparently, the presence of TIP47 in LDs is not related to the size of LDs, ²⁵ and thus is not determined by the static amount of esters in LDs. Consistent with this, although the number and size of LDs in RPE did not appear to change by dark adaptation, TIP47 was hardly observed in those LDs. In contrast, intense labeling of TIP47 was observed in LDs in the light-adapted eyes where continuous recycling of retinoids should be occurring. The result indicates that turnover of lipids, including the increased influx of all-trans-retinol and its esterification, is a likely factor that induces the translocation of TIP47 to LDs in the RPE. 

The putative hydrophobic cleft in TIP47 deduced from the three-dimensional crystallographic analysis has been proposed to bind lipids. ²⁰ In the present study as well as in a separate study, ²⁵ we found that disruption of the hydrophobic cleft by deletion of the carboxyl terminus or replacement of critical hydrophobic residues by alanines caused constitutive localization of TIP47 to LDs. These results imply that the hydrophobic cleft of TIP47 constitutes the on–off switch that regulates its LD localization. In RPE, direct binding of all-trans-retinol, fatty acids, or other lipids used to synthesize retinyl esters may induce the redistribution of TIP47 on light stimulation. 

The function of TIP47 as an adaptor molecule for the mannose-6-phosphate receptor recycling from late endosomes to the trans-Golgi network has been reported. ^{32 33} A ternary complex made of TIP47, mannose-6-phosphate receptor, and Rab9 was shown to be essential for efficient trafficking. ³³ In contrast, the identification of the functions of TIP47 in relation to LDs has been elusive. In contrast to perilipin and ADRP, which are constitutively present on LDs, TIP47, S3-12, and MLDP are recruited to LDs on de novo synthesis of lipid esters. A possible function of the latter proteins may be related to the delivery of nascent lipid esters to preexisting LDs where ADRP and/or perilipin are already present. ³⁴ In addition, as shown in keratinocytes, TIP47 may be involved in shielding lipid esters from cytoplasmic lipases. ³⁵ However, in the present study, even when TIP47 expression was suppressed below one-third of the control level, the amount of stored retinyl esters was not affected significantly. This result could be interpreted in several different ways. First, TIP47 may not be critical for retinyl ester storage or for transport of all-trans-retinol and/or its binding partner. Rather, we think that the residual amount of TIP47 remaining after the RNAi procedure was sufficient for its function related to retinoid recycling. In fact, most TIP47 exists as a soluble protein in the cytosol in untreated cells, ³² and even when TIP47 was induced to localize in LDs, the amount of TIP47 in the soluble fraction did not change significantly (Ohsaki Y and Fujimoto T, unpublished observation, 2006). In contrast, it is also possible that retinyl esters may exist in non-LD structures and that storage in those locations may not be influenced by manipulation of TIP47. 

The present results indicate that TIP47 in RPE in vivo is likely to recycle between the cytosol and LDs, depending on the light environment. This distributional change is synchronized with retinoid metabolism in the retina and should occur repeatedly each time animals experience light and dark adaptation. Reduction of TIP47 by RNAi did not cause significant changes in ARPE-19 cells, but the consequence of TIP47 knockdown may become apparent only after a certain time or after several rounds of light–dark transition in vivo. Studies of gene-targeted animals may be necessary to analyze the physiological function of TIP47 and to study whether it plays a critical role in the retinoid cycle in the eye. 

The Structure that Stores Retinyl Esters in RPE

It is noteworthy that LDs that were stained with BODIPY493/503 and positive for ADRP were present in the RPE in dark-adapted mouse eyes and in ARPE-19 cells in normal culture conditions. Neither the intensity of the BODIPY493/503 labeling nor the LD number changed significantly by light stimulation of mouse eyes or by all-trans-retinol addition to ARPE-19 cells. Imanishi et al. ²¹ reported that autofluorescence of retinyl esters became significantly increased by light stimulation of dark-adapted eyes. They also showed that the retinyl ester fluorescence colocalized with ADRP. Combined with our result, these observations indicate that retinyl esters were incorporated to preexisting ADRP-positive LDs where other lipid esters already existed. In contrast, the same paper proposed that vacuolar structures that were seen near the cell-cell boundary correspond to retinyl ester storage sites, or retinosomes. ²¹ But in the present study, we show that most of the ADRP labeling in mouse RPE was localized to canonical LDs that show round morphology with a smooth contour. This result does not exclude the presence of retinosomes, but suggests that they may not be the major structures that harbor ADRP in RPE. Furthermore, we cannot exclude the fact that functionally heterogeneous LD populations may coexist in RPE, although neither the ultrastructural nor cytochemical analyses suggested the possibility. Altogether, and in conjunction with previous reports, ^{36 37 38} our results suggest that LDs with morphology similar to those in other cell types are also the sites where retinyl esters are stored in the RPE. 

Abnormalities in LD-related proteins may lead to diseases in the eye. In this context, it is notable that a point mutation in Nir2 causes its constitutive localization to LDs in human cells, ³⁹ and that mutation of the corresponding site in the Nir2 homologue retinal degeneration B induces retinal degeneration in Drosophila melanogaster. ⁴⁰ Diseases involving TIP47 have not been reported, but the possibility that some disorders may occur when its LD targeting mechanism is perturbed remains. Our result would provide a solid basis for analyzing possible abnormalities in relation to TIP47. 

⁴  Present affiliation: Hakodate National College of Technology, Hakodate, Hokkaido, Japan.

 Supported by Grants-in-Aid for Scientific Research (KM, TF) and the 21st Century COE Program, “Integrrated Molecular Medicine for Neuronal and Neoplastic Disorders,” (TF) of the Ministry of Education, Culture, Sports, Science, and Technology of the Japanese Government.

 Submitted for publication July 7, 2006; revised January 9, 2007; accepted April 6, 2007.

 Disclosure: E. Tsuiki, None; A. Fujita, None; Y. Ohsaki, None; J. Cheng, None; T. Irie, None; K. Yoshikawa, None; H. Senoo, None; K. Mishima, None; T. Kitaoka, None; T. Fujimoto, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A) Specificity of the anti-mouse TIP47 antibody. When the mouse eye tissue lysate was examined by Western blot analysis, the anti-mouse TIP47 antibody reacted with a 47-kDa band alone, which was clearly different from the 52-kDa band recognized by the anti-ADRP antibody. The specificity of the anti-mouse TIP47 antibody was further confirmed by the lack of reaction with EGFP-mouse ADRP that was overexpressed in human Huh7 cells. The endogenous human TIP47 did not react with the anti-mouse TIP47 antibody. (B) Confocal laser scanning microscopy of mouse RPE in vivo. Mice were kept in normal light conditions until time of death. The eyecup specimens were labeled with anti-ADRP, anti-TIP47, or anti-Rab18 antibodies (red) and by BODIPY493/503 (green), and observed en face. ADRP and TIP47 colocalized with LDs, which were distributed in the cytoplasm without any particular concentration. Rab18 was not detected in the mouse RPE, but the reactivity of the antibody with mouse Rab18 was confirmed by the positive labeling of BALB/c 3T3 cells (inset). Bars, 10 μm.

Figure 2.

 

Mice were kept in total darkness for more than 24 hours, and the eyes were fixed in the dark or 30 minutes after exposure to intense flashes of light. (A) LDs and ADRP. The number and distribution of LDs did not show any difference between the two samples. ADRP was observed around the LDs. (B) LDs and TIP47. TIP47-positive LDs were scarce in the dark-adapted eyes, but they increased drastically after the light exposure, and virtually all the LDs became positive for TIP47. Bars, 10 μm.

Figure 3.

 

Immunogold electron microscopy of ADRP in mouse RPE. (A) Postembedding labeling of Lowicryl-embedded sections. The eyecup from a dark-adapted ddy mouse obtained 30 minutes after exposure to intense light is shown. Immunogold labeling specific for ADRP (arrows) was seen along the periphery of LDs (blue asterisk) that show electron lucency, round shape, and smooth contour. (B) Pre-embedding labeling with nanogold-conjugated antibodies and gold enhancement. The eyecup obtained from a light-adapted BALB/c mouse was used. Most of the labeling was seen around LDs (blue asterisks), which were preferentially distributed near the cell–cell boundary (green dots: the intercellular space). Some labeling was seen in the non-LD cytoplasm, but no specific structure was labeled. (C) Conventional thin section electron microscopy of RPE obtained from a light-adapted BALB/c mouse. Canonical LDs (blue asterisks) were observed near the cell–cell boundary.

Figure 4.

 

ARPE-19 cells grown in the normal medium were labeled for ADRP and TIP47 using specific antibodies (red) and for LDs by BODIPY493/503 (green). ADRP localized in most LDs, whereas TIP47 was seen in only a minor population of LDs. Bars, 10 μm.

Figure 5.

 

ARPE-19 cells were treated with all-trans-retinol for 10 minutes or 2 hours, and distribution of ADRP or TIP47 (red) and LDs (green) was observed. (A) ADRP and LDs. Neither ADRP nor LDs showed visible changes after all-trans-retinol treatment. (B) TIP47 and LDs. The labeling of TIP47 in LDs increased markedly after the treatment. The increase was already apparent at 10 minutes and became more significant at 2 hours. (C) The labeling intensity of ADRP and TIP47 was quantified by image analysis. Intensity of ADRP did not show a consistent change. In contrast, the TIP47 labeling increased significantly both dose and time dependently, and reached approximately 26 times the levels in the control after 2 hours of the 3 mM all-trans-retinol treatment. (*P < 0.005, Student’s t-test). (D) The protein expression of TIP47 was examined by Western blot analysis. An equal amount (30 μg) of protein was loaded into each lane. The increase in TIP47 by the all-trans-retinol treatment was not significant. Scale bars: (A, B) 10 μm.

Figure 6.

 

V5-tagged TIP47 mutants were expressed in ARPE-19 cells transiently and the ratio of cells showing distribution in LDs was quantified with or without 3 mM all-trans-retinol treatment for 2 hours. The ratio is the average of results in three independent experiments (±SD).

Figure 7.

 

(A) Quantification of retinol and retinyl esters in ARPE-19 cells. Control cells were sampled without any treatment. Cells transfected with a random-sequence control siRNA or TIP47 siRNA were treated with 3 mM all-trans-retinol for 2 hours before sampling. Both retinol and retinyl ester levels were increased drastically by the all-trans-retinol treatment, showing effective incorporation and processing in ARPE-19 cells. Neither retinol nor retinyl esters showed significant differences between cells treated with control and TIP47 siRNA. The result was similar, even when cells were treated with 1 mM all-trans-retinol for 30 minutes. (B) Western blot analysis showed that the expression of TIP47 was effectively suppressed by the RNAi procedure. The same amount of protein (30 mg) was loaded in each lane.

Supplementary Materials

Supplementary Figure S1 - (PDF) 

ARPE-19 cells were transfected with full-length and mutant TIP47 cDNAs with a V5 tag at the carboxyl terminus. The cell lysates (30 μg) were subjected to SDS-PAGE and Western blotting using an anti-V5 antibody. There was some variation in the expression level, but all the constructs showed the expected mobilities. 

The authors are grateful to Kumi Tauchi-Sato and Tetsuo Okumura for technical assistance.