Research Article

A Membrane Receptor for Retinol Binding Protein Mediates Cellular Uptake of Vitamin A

See allHide authors and affiliations

Science  09 Feb 2007:
Vol. 315, Issue 5813, pp. 820-825
DOI: 10.1126/science.1136244

Abstract

Vitamin A has diverse biological functions. It is transported in the blood as a complex with retinol binding protein (RBP), but the molecular mechanism by which vitamin A is absorbed by cells from the vitamin A–RBP complex is not clearly understood. We identified in bovine retinal pigment epithelium cells STRA6, a multitransmembrane domain protein, as a specific membrane receptor for RBP. STRA6 binds to RBP with high affinity and has robust vitamin A uptake activity from the vitamin A–RBP complex. It is widely expressed in embryonic development and in adult organ systems. The RBP receptor represents a major physiological mediator of cellular vitamin A uptake.

Vitamin A and its derivatives are essential for vision (1) and many other biological processes, because they are involved in the proliferation and differentiation of many cell types throughout life (2, 3). The majority of dietary vitamin A is stored in the liver. The principal physiological carrier of vitamin A (retinol) in the blood for delivery to other organs is retinol binding protein (RBP) (4). RBP is responsible for a well-regulated transport system that provides an evolutionary advantage by helping vertebrates adapt to fluctuations in vitamin A in natural environments (5). RBP also functions as a signal in insulin resistance (6). Loss of RBP makes mice extremely sensitive to vitamin A deficiency, because the hepatic vitamin A store can no longer be mobilized (7). Even with a nutritionally complete diet, RBP knockout mice have dramatically lower serum vitamin A concentrations, similar to the concentrations in the later stages of vitamin A deficiency in humans. Given the role of vitamin A in immune regulation and the susceptibility of vitamin A–deficient children to infection before visual symptoms (8), it is likely that the immune system is compromised in RBP knockout mice even under vitamin A–sufficient conditions. Indeed, the circulating immunoglobulin amount in these mice is half of that in wild-type mice, even under vitamin A sufficiency (9). In humans, loss of RBP function causes a dark adaptation defect and progressive atrophy of the retinal pigment epithelium (RPE) at young ages (10). Under conditions of vitamin A deficiency at which wild-type mice behave normally, RBP knockout mice have severe developmental defects in embryos (11) and rapid vision loss in adults after merely a week of vitamin A deficiency (7). In addition, these mice rapidly develop testicular defects (12).

It was first shown in the 1970s that there exists a specific cell surface receptor for RBP on the RPE and on intestinal epithelial cells (1316). During the past 3 decades, evidence has also accumulated for the existence of the RBP receptor on other tissue or cell types, including the placenta (1720), choroid plexus (18, 21), testis (18, 22), and macrophages (23). The cell surface RBP receptor not only specifically binds to RBP but also mediates vitamin A uptake from vitamin A–loaded RBP (holo-RBP) (1517, 20, 22, 23). Here, we report the identification of the RBP receptor as STRA6, a widely expressed multitransmembrane domain protein. STRA6 met all three criteria expected of the RBP receptor. First, it conferred RBP binding to transfected cells. Second, it mediated cellular uptake of vitamin A. Third, it was localized to the cellular locations expected of the RBP receptor in native tissues.

Identification of the RBP receptor as STRA6. Potential obstacles to purifying the RBP receptor included the fragility of the receptor protein and the transient nature of binding. We designed a strategy to stabilize the RBP-receptor interaction, and this strategy permitted high affinity purification of the RBP-receptor complex. The strategy combined the 6Xhistidine tag (His tag)–nickel system and a bifunctional cross-linker with an amine-reactive group and a photoreactive group (Fig. 1A). We used bovine RPE cells as the starting material because they are known to express the RBP receptor at high concentrations (13). Because N-terminal alkaline phosphatase (AP)–tagged RBP was able to specifically bind to its receptor on RPE cells with a staining pattern expected for basolateral membranes (Fig. 1B), we created His-tagged RBP (His-RBP) by tagging at the N terminus of RBP. After binding purified His-RBP protein conjugated with the cross-linker to RPE membranes, ultraviolet (UV) cross-linking, membrane solubilization, and nickel resin purification, we observed a covalent protein complex of about 80 kD containing RBP and its putative receptor (Fig. 1C). Mass spectrometry analysis revealed that the RBP binding protein in the complex was STRA6, a protein that has 11 putative transmembrane domains but has no homology to any protein with known function (fig. S1). STRA6 is a retinoic acid–induced gene in P19 embryonic carcinoma cells (24).

Fig. 1.

(A) Schematic diagram of the strategy used to identify the RBP receptor. The structure of the aminoreactive and photoreactive cross-linker is shown on the top and illustrated as red symbols in the diagram. The letter H above the symbol for RBP represents the 6XHis tag at the N terminus. Shaded rectangles represent membrane. (B) Binding of AP-RBP to freshly dissected bovine RPE cells (top). AP reaction product is deep purple. Presence of 10-fold excess RBP abolished the labeling (bottom). The hexagonal shape of an RPE cell is outlined in red dotted lines in each picture. Scale bars indicate 10 μm. (C) Immunoblot analysis using an antibody against RBP. Molecular weight marker is shown on the left (in kD). The His-RBP monomer band represents His-RBP that was bound to RPE membrane but not cross-linked to the receptor. Untagged RBP competition in lane 4 prevented His-RBP from binding to RPE membrane and therefore lowered signals for both the complex and the monomer band.

STRA6 mediates RBP binding and vitamin A uptake from holo-RBP. To test whether STRA6 could confer RBP binding to transfected cells, we transfected bovine STRA6 cDNA into COS-1 cells. Bovine STRA6–transfected but not untransfected cells bound to AP-RBP with high affinity (Kd = 59 nM) (Fig. 2, A and B). An excess of unlabeled RBP blocked this binding activity (Fig. 2B). To test whether STRA6 could enhance cellular vitamin A uptake from holo-RBP, we performed both 3H-retinol–based assays (Fig. 2C) and high-performance liquid chromatography (HPLC)–based assays (Fig. 2, D to H). STRA6 expression dramatically increased cellular vitamin A uptake from holo-RBP. Vitamin A is preferentially stored as retinyl esters after uptake from holo-RBP (25). Lecithin retinol acyltransferase (LRAT) is the enzyme that converts retinol into retinyl ester (26) and plays an important role in vitamin A storage (27, 28). Thus, we included LRAT expression in most vitamin A uptake assays. By using an assay based on 3H-retinol-RBP, we found that STRA6 activity was most evident in the presence of LRAT (Fig. 2C), because LRAT makes it possible to store vitamin A at high concentration in the form of retinyl esters, the more stable derivative of vitamin A. In these assays, nearly all specific vitamin A uptake activity, which was sensitive to blocking by an excess of unlabeled RBP, was due to STRA6. Vitamin A uptake by COS-1 cells transfected with STRA6 and LRAT was highly efficient. When the concentration of 3H-retinol-RBP was 3 nM in the medium and cell density was 1.8 × 106 cells/ml, about 25% of the total radioactivity was taken up by cells in 1 hour, and about 50% of the total radioactivity was taken up by cells in 2 hours. The amount of retinol taken up by cells in 1 hour is about 4.6 times the moles of RBP bound to the cell surface at steady state. We also used HPLC to analyze vitamin A uptake mediated by STRA6 (Fig. 2, D to H). In this assay, untranfected COS-1 cells accumulated little retinyl ester. COS-1 cells transfected with LRAT alone did accumulate a small amount of retinyl esters, but cells cotransfected with STRA6 and LRAT showed a 15-fold higher amount of retinyl ester accumulation than that of cells transfected with LRAT alone (Fig. 2, D to F). At the saturating retinol-RBP concentration (0.44 μM) used for the HPLC assays, the amount of retinol taken up by cells in 1 hour is about 14 times the moles of RBP bound to the cell surface at steady state. Holo-RBP in the blood is in complex with the thyroxine binding protein, transthyretin (TTR). Although TTR blocks the vitamin A exit site in the holo-RBP-TTR complex (29), the fact that tissues take up vitamin A from holo-RBP-TTR in the blood in vivo suggests that TTR cannot completely inhibit tissue vitamin A uptake from holo-RBP. When we assayed vitamin A uptake by using human serum as the source of the holo-RBP-TTR complex, STRA6 could still efficiently take up vitamin A from this complex (Fig. 2, G and H).

Fig. 2.

RBP binding and vitamin A uptake activity of STRA6. (A) Binding of RBP to live COS-1 cells transfected with STRA6 (top) but not to untransfected cells (bottom). The AP reaction product is deep purple. Scale bar, 20 μm. (B) Concentration dependence of the binding of AP-RBP to live COS-1 cells transfected with STRA6. Controls are LRAT-transfected cells, untransfected cells, and STRA6-transfected cells with 50 × unlabeled RBP during binding. The plots for these three controls are superimposed on this graph. (C) Comparison of vitamin A uptake activities from 3H-retinol/RBP. U, S, L, and S/L denote untransfected cells and cells transfected with STRA6, LRAT, and both, respectively. RBP, 100 × unlabeled holo-RBP. The activity of S/L cells is defined as 100%. Error bars indicate SD. (D to H) HPLC assays of vitamin A uptake for COS-1 cells transfected with STRA6 and LRAT (green) or LRAT alone (red) and for untransfected cells (blue). AU, absorbance units. (D) HPLC profile of hexane extracts of COS-1 cells after vitamin A uptake from holo-RBP. Arrowheads indicate peaks representing retinyl ester. Peaks representing hexane-soluble peptides (asterisks) serve as an internal control. (E) Overlay of the retinyl ester peaks in (D). (F) Overlay of absorption spectra of the retinyl ester peaks in (D). (G) Overlay of the retinyl ester peaks from similar HPLC runs as shown in (D) but using 25% human serum as the source of holo-RBP. (H) Overlay of absorption spectra of the peaks in (G).

Because retinoic acid up-regulates STRA6 in certain cancer cell lines such as WiDr human colon adenocarcinoma cells (24, 30), we used WiDr cells as an independent model to study vitamin A uptake mediated by STRA6. We increased STRA6 expression in WiDr cells by retinoic acid treatment and decreased its expression by RNA interference (RNAi). Consistent with STRA6's function in mediating vitamin A uptake from holo-RBP, its increased expression by retinoic acid treatment enhanced vitamin A uptake activity, whereas its decreased expression by specific RNAi knockdown suppressed vitamin A uptake activity (Fig. 3A). In addition, we performed RNAi knockdown experiments on primary bovine RPE cultures. Again a decrease in STRA6 expression suppressed RPE'svitamin A uptake activity from holo-RBP (Fig. 3B).

Fig. 3.

STRA6 RNAi in native cell types and STRA6 mutagenesis. (A) Retinoic acid (RA) treatment of WiDr cells increases the expression of STRA6 and the vitamin A uptake activity of WiDr cells, whereas STRA6 RNAi treatment suppresses STRA6 expression and vitamin A uptake activity. NC is negative control RNAi oligonucleotide (oligo). S-1, S-2, and S-3 are RNAi oligos for human STRA6. Error bars indicate SD. (B) STRA6 RNAi treatment suppresses STRA6 expression and vitamin A uptake activity of primary bovine RPE culture. S-4, S-5, and S-6 are RNAi oligos for bovine STRA6. (C) Comparison of vitamin A uptake activities of COS-1 cells transfected with bovine STRA6 wild-type (WT) or mutants Gln310→His310/Leu315→Pro315 (M1), Leu256→His256/Tyr336→His336 (M2), and Leu297→Arg297 (M3). Activity of cells transfected with WT STRA6 is defined as 100%. (D) Comparison of AP-RBP binding activities of COS-1 cells transfected with bovine STRA6 WT or mutant M1, M2, or M3. Activity of cells transfected with WT STRA6 is defined as 100%. (E) Surface expression of wild-type or mutant STRA6 as assayed by immunostaining of transfected live COS-1 cells. Scale bar, 10 μm.

Because STRA6 is not homologous to any protein of known function and has never been characterized at the structural level, we used an unbiased strategy to test the effects of mutations on STRA6 function. We produced and characterized 50 random missense mutants of STRA6 and found that 3 of them have substantial loss of vitamin A uptake activity (Fig. 3, C to E). The locations of these mutations in the putative transmembrane topology model of STRA6 are shown in fig. S1B. Mutants 2 and 3 (M2 and M3) were not expressed at the cell surface, which may be sufficient to account for the dramatic loss of RBP binding and vitamin A uptake activity from holo-RBP. Mutant 1 (M1) was normally expressed at the cell surface but had substantially reduced RBP binding and vitamin A uptake activity. The loss of RBP binding for M1 may account for its lower vitamin A uptake activity.

Characterization of STRA6-mediated vitamin A uptake from holo-RBP. We found that STRA6-mediated vitamin A uptake is specific to RBP. STRA6 could not enhance cellular uptake of vitamin A when vitamin A was bound to bovine serum albumin (BSA) or β-lactoglobulin (Fig. 4A). STRA6-mediated vitamin A uptake was saturable with regard to retinol-RBP concentration (Fig. 4B and fig. S2). A variety of studies from the past 3 decades had indicated that endocytosis is not required for RBP receptor–mediated vitamin A uptake from holo-RBP (1417, 20, 22, 25). In agreement with these studies, metabolic inhibitors that effectively inhibited endocytosis did not inhibit STRA6-mediated vitamin A uptake by STRA6-transfected COS-1 cells or retinoic acid–stimulated WiDr cells (Fig. 4C and 4D). Membranes prepared from transfected cells have vitamin A uptake activity similar to that of live cells (Fig. 4E). The robust activity of this cell-free system also argues against an endocytosis-based mechanism. We also designed an assay that monitored both RBP and vitamin A that remains bound to RBP in the culture medium after cellular vitamin A uptake. We purified the RBP in the cell culture medium after cellular uptake of vitamin A and measured the absorption spectrum of the purified RBP (Fig. 4F). A preferential decrease in the vitamin A peak for RBP taken from the medium of STRA6- and LRAT-transfected cells again argued against an endocytosis-based mechanism.

Fig. 4.

Characterization of STRA6-mediated vitamin A uptake from holo-RBP. (A) STRA6 mediates 3H-retinol uptake from 3H-retinol bound to RBP, but not 3H-retinol bound to BSA or β-lactoglobulin. “Transfected” indicates transfection with STRA6 and LRAT. Retinol uptake activity of transfected cells from holo-RBP is defined as 100%. Error bars indicate SD. (B) STRA6-mediated vitamin A uptake is saturable with regard to retinol-RBP concentration. The highest retinol uptake activity at 60 min is defined as 100%. (C) MI (a mixture of metabolic inhibitors 2-deoxyglucose and sodium azide) effectively inhibits horseradish peroxidase (HRP) endocytosis of COS-1 cells (left) but does not inhibit the vitamin A uptake activity of COS-1 cells transfected with STRA6 and LRAT (S/L) (right). L, LRAT transfected cells. The activity of S/L cells without MI is defined as 100%. (D) MI effectively inhibits HRP endocytosis by WiDr cells (left), but does not inhibit vitamin A uptake activity of WiDr cells (right). The activity of WiDr cells treated with 40 mM retinoic acid but without MI is defined as 100%. (E) Vitamin A uptake activity of membrane fractions of human embryonic kidney 293 cells. U, S, L, and S/L denote membranes from untransfected cells, cells transfected with STRA6, LRAT, and both, respectively. R, 100 × holo-RBP. The activity of S/L cells is defined as 100%. (F) RBP remains outside the cell after STRA6-mediated vitamin A uptake from holo-RBP for 8 hours. Absorption spectra of RBP that remains in the supernatant of untransfected cells (left) and cells transfected with STRA6 and LRAT (right) are shown.

Localization of STRA6 is consistent with its function as the RBP receptor. The tissue localization of STRA6 has been studied previously (24). STRA6 is expressed during embryonic development and in the adult brain, spleen, kidney, female genital tract, and testis (and at lower quantities in heart and lung). Furthermore, STRA6 is highly enriched in the RPE in the adult eye (24). To further study the location of STRA6 in adult organs, we produced polyclonal antibodies against bovine STRA6 and performed immunohistochemistry on several adult organs. In the RPE, STRA6 is localized to the basolateral membrane (Fig. 5, A to C), a location consistent with its role in interacting with RBP in the choriocapillaris blood (14). In contrast to its absence in the endothelial cells of the choriocapillaris (Fig. 5, A and B), STRA6 was expressed in retinal blood vessels, another location of the blood-retina barrier (Fig. 5, D and E). Holo-RBP from retinal blood vessels is a potential source of vitamin A for Müller cells in the retina (31). In addition to blood-brain barriers as observed in a previous study (24), we found STRA6 expression in astrocyte perivascular endfeet that surround blood vessels negative for STRA6 (Fig. 5F). In accord with National Center for Biotechnology Information (NCBI)'s expressed sequence tag (EST) tissue expression profile for STRA6, we observed STRA6 expression in the placenta (Fig. 5G) and the spleen (Fig. 5H).

Fig. 5.

Immunolocalization of STRA6. Green signal is STRA6. Red signal is peropsin, an apical RPE marker (A) or GSL I–isolectin B4, an endothelial cell marker (B, D, F, G, and H). Blue signal is nuclear stain 4′,6′-diamidino-2-phenylindole (DAPI). Scale bars represent 10 μmin(A) to (C) and 40 μm in (D) to (H). CH, choroid; OS, photoreceptor outer segments; ONL, outer nuclear layer; and INL, inner nuclear layer. (A and B) STRA6 is localized to the basolateral membrane of the RPE. Arrowheads indicate signals on the lateral membrane of RPE cells. (C) A tangential section of the RPE layer. (D and E) STRA6 signals are most enriched in the RPE layer of the eye but are also present in the blood vessels in retina. (E) is an overexposed picture of (D) to show the weaker signals in retina blood vessels. (F) Localization of STRA6 in hippocampus. Examples of large blood vessels positive for STRA6 (large arrowhead) and negative for STRA6 (small arrowhead). The STRA6-negative vessel is surrounded by STRA6-positive astrocyte perivascular endfeet. (G) Localization of STRA6 in placenta. (H) STRA6 is highly expressed in a subset of cells in the spleen.

Discussion. The existence of an RBP receptor is supported by a large body of evidence. The absence of vitamin A from abundant erythrocytes and serum albumin, which can bind vitamin A, argues against simple diffusion of retinol from the RBP to the cell membrane as the mechanism of vitamin A delivery by RBP (32). The fact that free vitamin A can diffuse through membranes is not a good argument against the existence of an RBP receptor. First, nearly all vitamin A in blood is bound to the soluble RBP-TTR complex, which cannot freely pass through cell membranes. Second, there are known examples of molecules that can diffuse through membranes but still depend on membrane transporters to facilitate their transport. For example, prostaglandin can pass the membrane by simple diffusion, but prostaglandin transporter greatly facilitates its transport across membranes (33). In addition, there is strong experimental evidence supporting the existence of an RBP receptor that mediates cellular vitamin A uptake (1323). This study has identified STRA6 as the RBP receptor. The RBP-STRA6 system represents a small-molecule delivery mechanism that involves an extracellular carrier protein but does not depend on endocytosis. Given the potent biological effects (including toxicity) of vitamin A and its derivatives (24), a controlled release of vitamin A into cells from holo-RBP through STRA6 has an evolutionary advantage over nonspecific diffusion of vitamin A. This mechanism makes it possible to achieve high efficiency and specificity for vitamin A delivery to organs distant from the liver such as the eye, brain, placenta, and testis.

STRA6's localization in adult brain is consistent with vitamin A's roles in regulating adult brain activities (34), such as synaptic plasticity (35) and cortical synchrony during sleep (36). STRA6 expression in the placenta and testis is consistent with the role of RBP in delivering vitamin A to these tissues (11, 12). STRA6 expression in the spleen and thymus is consistent with a role for vitamin A in immune regulation and in prevention against infectious disease (8, 37). According to NCBI's EST analysis, STRA6 is also expressed in human skin and lung, both known to depend on vitamin A for proper function (38, 39). Undifferentiated human skin keratinocytes were found to have the highest RBP binding activity of any cell or tissue types tested (18). The absence or very low expression of STRA6 in the liver(24) makes physiological sense. Liveris the major site of production of vitamin A–loaded RBP to deliver vitamin A to peripheral organs. If STRA6 was highly expressed in the liver and absorbed vitamin A from holo-RBP produced by liver itself, a short circuit would be created.

STRA6 was originally found to be a retinoic acid–stimulated gene in cancer cell lines (24, 30). However, it is possible that there are certain noncancer cell types that respond to retinoic acid to stimulate STRA6 expression. For example, vitamin A combined with retinoic acid increases retinol uptake in the lung in a synergistic manner (40), consistent with the ability of retinoic acidtostimulate STRA6 expression. Onestudy found that STRA6 was overexpressed up to 172-fold in 14 out of 14 human colorectal tumors relative to the normal colon tissue (30). Thus, the RBP-STRA6 system not only functions as a physiological mechanism of vitamin A uptake, but also potentially participates in pathological processes such as insulin resistance (6) and cancer.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1136244/DC1

Materials and Methods

Figs. S1 and S2

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article