|Home | About | Journals | Submit | Contact Us | Français|
Cells are known to take up molecules through membrane transport mechanisms such as active transport, channels, and facilitated transport. We report here a new membrane transport mechanism that employs neither cellular energy like active transport nor a preexisting electrochemical gradient of the free substrate like channels or facilitated transport. Through this mechanism, cells take up vitamin A bound with high affinity to retinol binding protein (RBP) in the blood. This mechanism is mediated by the RBP receptor STRA6, which defines a new type of cell-surface receptor. STRA6 is essential for the proper functioning of multiple human organs, but the mechanisms that enable and control its cellular vitamin A uptake activity are unknown. We found that STRA6-mediated vitamin A uptake is tightly coupled to specific intracellular retinoid storage proteins, but no single intracellular protein is absolutely required for its transport activity. By developing sensitive real-time monitoring techniques, we found that STRA6 is not only a membrane receptor but also catalyzes vitamin A release from RBP. However, vitamin A released from RBP by STRA6 inhibits further vitamin A release by STRA6 unless specific intracellular retinoid storage proteins relieve this inhibition. This mechanism is responsible for its coupling to intracellular storage proteins. The coupling of uptake to storage provides high specificity in cellular uptake of vitamin A and prevents the excessive accumulation of free vitamin A. We have also identified a robust small molecule-based technique to specifically stimulate cellular vitamin A uptake. This technique has implications in treating human diseases.
Vitamin A and its derivatives (retinoids) are a group of chemicals that perform critical and diverse biological functions. For example, the aldehyde form of vitamin A functions as the photoreceptor chromophore in vision (1–3) and regulates adipogenesis (4). The acid forms of vitamin A (retinoic acid) regulate gene transcription and control cell growth and differentiation (5, 6). Due to its potent biological activities, nonspecific and excessive vitamin A uptake can lead to severe toxicity (7–10). How cells control vitamin A uptake to prevent its excessive accumulation and random diffusion is unknown. Plasma retinol binding protein (RBP) is the principle carrier of vitamin A in the blood (11–13) and its excessive secretion plays a role in insulin resistance (14). STRA6, a membrane protein originally identified in cancer cells (15, 16), is the high-affinity RBP receptor that mediates cellular uptake of vitamin A (17). It is expressed in cells or tissues known to depend on vitamin A for proper function. For example, its expression in the brain is consistent with known neuronal functions of vitamin A derivatives (18, 19). Both human and animal studies have shown that loss of STRA6 can lead to severe defects in multiple organs (20, 21).
STRA6 is a multi-transmembrane domain protein that represents a new type of cell-surface receptor. STRA6’s mechanism does not depend on endocytosis (17). Its energy-independence also excludes primary or secondary active transport. Unlike channels or facilitated transporters, which are driven by the electrochemical gradient of the free substrate, STRA6’s substrate retinol is not free, has no charge, and can only be provided one molecule at a time by RBP. Its mechanism is further complicated by the high affinity interaction between retinol and RBP, its dependence on both RBP binding to deliver retinol and RBP dissociation for the next RBP to bind, and the involvement of intracellular proteins. How STRA6 mediates substrate uptake is unknown and its activity cannot be explained by known cellular uptake mechanisms (Figure 1A). We (17) and other investigators (21, 22) found that lecithin retinol acyltransferase (LRAT) stimulates STRA6’s vitamin A (retinol) uptake activity. It was also found that STRA6 and LRAT cannot take up retinylamine from the retinylamine/RBP complex (22) or retinoic acid from the retinoic acid/RBP complex (21). Why does LRAT stimulate STRA6’s retinol uptake, but not the uptake of other retinoids? Does STRA6 have any retinol uptake activity without LRAT? If it does, why can LRAT stimulate its activity? If it does not, is STRA6 only a receptor for RBP? How does STRA6 allow retinol to enter a cell while leaving RBP outside a cell? Is LRAT the only protein that can stimulate STRA6’s activity? LRAT converts retinol to retinyl ester, a storage form of vitamin A. Retinol can also be stored by binding to cellular retinol binding proteins (CRBP), such as CRBP-I (23–25) and CRBP-II (26). Can CRBP-I or CRBP-II stimulate STRA6’s retinol uptake? Is retinol the only substrate that can be taken up through STRA6? STRA6 was also found to facilitate the loading of exogenous retinol into apo-RBP in the presence of LRAT (21). Does LRAT stimulate retinol loading or retinol uptake? Can a small molecule stimulate STRA6’s retinol uptake? Answers to these questions might help to reveal STRA6’s vitamin A uptake mechanism.
We tested whether CRBP-I or CRBP-II can enhance STRA6’s vitamin A uptake and found that CRBP-I does enhance STRA6’s vitamin A uptake activity (Supplemental Figure 1). In contrast, CRBP-II is much less effective although they have similar expression levels (Figure 1B). This difference between CRBP-I and CRBP-II suggests that the ability to bind retinol is not sufficient to enhance STRA6’s vitamin A uptake. Using a 3H-retinol uptake assay, we next compared the time course of vitamin A uptake between cells expressing STRA6/LRAT, STRA6/CRBP-I and STRA6 alone and found that STRA6/CRBP-I and STRA6 cells, but not STRA6/LRAT cells, exhibit obvious saturation (Figure 1C). Although this result suggests saturation of available apo-CRBP for the CRBP-I reaction and ongoing synthesis of retinyl esters for the STRA6/LRAT reaction, the nature of vitamin A uptake by STRA6-only cells is not clear.
To overcome the complication of retinol associated with cells in the form of surface-bound holo-RBP, we added a large excess of unlabeled holo-RBP to compete with 3H-retinol/RBP bound to STRA6 on the cell surface. Surprisingly, STRA6 has little “true” retinol uptake activity after cell-surface holo-RBP is removed (Figure 1D). Since holo-RBP exists in the serum in the form of holo-RBP/transthyretin (TTR) complex (12), we also used an HPLC assay and human serum as the source of holo-RBP to confirm the finding that STRA6 by itself takes up little retinol (Figures 1E, 1F and 1G). Due to LRAT’s role in retinyl ester formation (Figure 1E), retinol-based analysis (Figure 1G) is necessary to reveal the nature of retinol uptake activity of STRA6-only cells. STRA6-only cells take up even less vitamin A in the presence of TTR, which partially inhibits STRA6’s activity (Supplemental Figure 2). The dependence of STRA6 on LRAT or CRBP-I for vitamin A uptake suggests that STRA6 is tightly coupled to these proteins.
The very low vitamin A uptake activity of STRA6-only cells suggests that STRA6 alone is seemingly ineffective in releasing retinol from holo-RBP. If retinol is not released from RBP and RBP does not get endocytosed, how does retinol enter its target cell? To resolve this paradox, we tested whether STRA6 accelerates retinol release by developing a sensitive assay for retinol release from holo-RBP. This assay is based on the dramatic enhancement of retinol fluorescence when retinol is bound to RBP. In this assay, STRA6/LRAT membranes cause a large reduction of retinol fluorescence when incubated with holo-RBP. Interestingly, STRA6 membranes also cause a significant reduction in retinol fluorescence, while membranes from control, LRAT and a STRA6 mutant (27) that has normal cell surface expression but does not bind RBP fail to cause any reduction (Figure 2A). We further fractionated the reactions into membrane and supernatant fractions and found that STRA6/LRAT causes complete retinol release from RBP in the supernatant while STRA6 causes partial retinol release (Figures 2B). Retinol in the supernatant of STRA6-treated reaction is present in the form of holo-RBP not as free retinol (Figure 2C). As expected, retinol in STRA6/LRAT membrane has been converted to retinyl esters, but retinol in STRA6 membrane remains the retinol form (Figure 2D). We also purified the RBP/TTR complex from human serum and demonstrated that STRA6 catalyzes retinol release from the RBP/TTR complex as well (Figures 3A and 3B).
Why does STRA6 stimulate retinol release in the fluorescence assay but cause little vitamin A uptake in cellular retinol uptake assays? This result can be explained by the ratio of holo-RBP to STRA6 used in these assays. The retinol fluorescence assays have much lower holo-RBP to STRA6 ratios and much higher STRA6 concentrations compared with those used in the cellular retinol uptake assay. In the fluorescence assays, STRA6 becomes more dependent on LRAT for retinol release when the holo-RBP to STRA6 ratio is increased (Figure 3C). This is also confirmed in vitamin A uptake assays in live cells (Supplemental Figure 3). One hypothesis to explain STRA6’s dependence on LRAT is that retinol released by STRA6 inhibits further retinol release, but LRAT relieves the inhibition by converting retinol to retinyl esters. To test this hypothesis, we added free retinol to STRA6 membrane and indeed observed a strong inhibitory effect on STRA6-catalyzed retinol release (Figure 3D). This inhibitory effect can be relieved by LRAT or CRBP-I.
If STRA6 depends more on LRAT for retinol release when there is more holo-RBP, what is the effect of more apo-RBP? We tested the addition of apo-RBP to STRA6 after it has catalyzed the release of retinol from holo-RBP and found that the addition of apo-RBP caused a fluorescence increase in the STRA6 reaction but not in the STRA6/LRAT reaction (Figure 4A). The increase in fluorescence suggests that retinol released by STRA6 was loaded back to RBP and that LRAT’s conversion of retinol into retinyl ester prevented the loading. Since this experiment cannot distinguish whether retinol loading started from free retinol or a retinol/STRA6 complex, we tested free retinol and found that STRA6 catalyzes very efficient loading of free retinol into RBP (Figure 4B). The initial phase of STRA6-catalyzed retinol loading is very different from the initial phase of STRA6-catalyzed retinol release (Figures 4C and 4D). In contrast to the fluorescence decay during retinol release, which does not show a downward peak, the STRA6-catalyzed fluorescence increase is characterized by a peak (Figures 4B and 4C). This biphasic property of the loading reaction can be explained by the dual functions of STRA6 in first catalyzing retinol loading and then catalyzing retinol release of the newly formed holo-RBP. Intriguingly, when STRA6 is presented with equal opportunities to catalyze retinol release or to catalyze retinol loading, the loading activity predominates initially (Figure 4E). The robust retinol loading activity of STRA6 suggests the possibility that STRA6 depends more on LRAT for retinol release when there is more holo-RBP because the released retinol inhibits further release due to the STRA6’s retinol loading activity. Consistently, STRA6-catalyzed retinol loading is inhibited by LRAT and CRBP-I (Figure 4F). STRA6’s retinol loading activity is also confirmed by HPLC analysis (Supplemental Figure 4).
CRBP-I and RBP are both retinol binding proteins. What’s the difference between STRA6-catalyzed loading of RBP versus STRA6-catalyzed loading of CRBP-I? CRBP-I and CRBP-II are also both retinol binding proteins. Why can only CRBP-I effectively couple to STRA6 to mediate vitamin A uptake from holo-RBP? To answer these questions, we developed a fluorescence resonance energy transfer (FRET) technique to monitor retinol transport to CRBP-I in real time. This technique is based on the significant overlap between the emission spectrum of retinol and the excitation spectrum of enhanced green fluorescent protein (EGFP) (Supplemental Figure 5). Using a fusion protein between CRBP-I and EGFP (EGFP-CRBP-I), we observed clear FRET signals between retinol and EGFP during STRA6-mediated retinol transfer from holo-RBP to EGFP-CRBP-I (Figures 5A and 5B). Real-time monitoring revealed an increase and saturation of FRET signals dependent on STRA6, time, and CRBP-I-concentrations; but no decay was observed unless apo-CRBP-I was added (Figure 5C and Supplemental Figure 6A). The stable signal suggests that STRA6-catalyzed retinol transfer to CRBP-I does not lead to further unloading of CRBP-I, in sharp contrast to the STRA6-catalyzed retinol loading of RBP (Figures 4B and 4C). Also in contrast to RBP, which depends on STRA6 for efficient loading with retinol (Figure 4B), CRBP-I absorbs retinol added to control membranes indistinguishably from STRA6 membranes (Supplemental Figure 6B) and is more effective in coupling to STRA6 than CRBP-II due to its higher affinity for retinol (Supplemental Figures 6C and 6D). These experiments suggest that CRBP-I’s high affinity binding to retinol distinguishes it from CRBP-II, and its STRA6-independent absorption of retinol from the membrane distinguishes it from RBP.
Are LRAT and CRBP-I the only proteins that can couple to STRA6? LRAT and CRBP-I couple to STRA6 due to their ability to restrain the free substrate intermediate, which is inhibitory to retinol release. We tested whether other proteins that can interact with the free substrate couple to STRA6. Indeed, retinol dehydrogenase (RDH), which uses free retinol as the substrate, accelerates STRA6-catalyzed retinol release by converting retinol to retinal in the presence of NADP (Figure 6A). Cellular retinoic acid binding protein-I (CRABP-I), which binds free retinoic acid, couples to STRA6 in mediating retinoic acid uptake from the retinoic acid/RBP complex (Figure 6B). LRAT and CRBP-I, which don’t interact with retinoic acid, can no longer couple to STRA6 in this context (Figure 6B).
Can small molecules also enhance STRA6’s activity by affecting the free substrate intermediate? Free unlabeled retinol is expected to enhance 3H-retinol uptake from 3H-retinol/RBP if the unlabeled free retinol inhibits STRA6’s retinol loading activity by competing with 3H-retinol. We found that this is indeed the case (Figure 7A). As the natural substrate of RBP, retinol naturally inhibits STRA6-mediated vitamin A uptake and can only enhance STRA6’s uptake activity in the context of the 3H-retinol uptake assay. Is it possible that compounds similar to retinol can enhance STRA6’s activity in all assays? We tested β-ionone, a small chemical found in fruits and vegetables, which mimics retinol in structure but is not a retinoid. Indeed, β-ionone stimulates retinol uptake from holo-RBP in a dose-dependent manner in the 3H-retinol uptake assay (Figure 7A), the retinol fluorescence assay (Figures 7B and 7C), and the HPLC-based assay using human serum as the source of holo-RBP (Figure 7F). Like CRBP-I and LRAT, β-ionone stimulates STRA6’s vitamin A uptake activity by blocking STRA6’s retinol loading activity (Figures 7C and 7E). However, the exact mechanisms are different. CRBP-I and LRAT removes free retinol, but β-ionone competes with the free retinol in the STRA6-catalyzed retinol loading reaction (Figures 7D and 7E). Remarkably, as an exogenous small molecule, β-ionone can be as effective as CRBP-I in enhancing STRA6’s retinol uptake and can even further boost retinol uptake for STRA6/CRBP-I cells (Figure 7F).
The diverse activities of STRA6 (Supplemental Figure 7 and Figure 8A), such as its abilities to catalyze retinol release, to couple to LRAT, to couple to CRBP-I, to couple to CRABP-I, to stimulate rapid retinol loading, and to enable β-ionone to greatly stimulate cellular retinol uptake, rule out alternative models of STRA6 uptake mechanism (Supplemental Figure 8) and lead to a unifying model (Figure 8B). In Step 1, STRA6 on the cell surface binds to holo-RBP with high-affinity. The source of the holo-RBP is the RBP/TTR complex in the blood. STRA6 effectively mediates vitamin A uptake from the holo-RBP/TTR complex (Figures 1E and 1G) although TTR partially suppresses STRA6’s activity (Supplemental Figure 2). In Step 2, STRA6 catalyzes retinol release from holo-RBP. STRA6’s ability to catalyze retinol release has been demonstrated and analyzed in real time (Figures 2, ,3,3, 4A, 4D, ,6A,6A, and and7B7B–7D). In Step 3, the released retinol either inhibits further retinol release through the robust STRA6-catalyzed retinol loading activity (pathway 3a) or is removed by CRBP-I and LRAT (pathway 3c). STRA6-catalyzed retinol loading has been analyzed in real time (Figures 4A, 4B, 4C, 4E, 4F, and and7E).7E). STRA6-mediated retinol uptake is coupled to CRBP-I and LRAT because they can restrain the inhibitory retinol intermediate and inhibit retinol loading. In the presence of β-ionone, pathway 3a is blocked, retinol uptake no longer depends on CRBP-I or LRAT (pathway 3b) (Figure 8B). In Step 4, apo-RBP dissociates from STRA6 and will be lost to kidney filtration due to its inability to bind TTR (12).
By combining two newly developed real-time monitoring techniques with radioactive retinoid-based and HPLC-based techniques (Supplemental Table), this study revealed the mechanism of STRA6-mediated substrate uptake and its ability to couple to distinct intracellular proteins. This study also found that no intracellular protein is absolutely required for enhanced STRA6 activity and that it is possible for a small molecule to stimulate STRA6-mediated vitamin A uptake. STRA6-catalyzed retinol release from RBP is largely responsible for the endocytosis-independent uptake of retinol, but STRA6 does not simply function as a “bottle opener”, if RBP is analogous to a “bottle” that holds vitamin A. Instead, STRA6-mediated retinol uptake is coupled to CRBP-I and LRAT (but not to the cell membranes, despite their much larger capacity to store retinol). This coupling mechanism depends on the robust STRA6-catalyzed retinol loading, which inhibits further retinol release. In simpler terms, STRA6-catalyzed retinol release and STRA6-catalyzed retinol loading largely (but not completely) cancel each other’s action. Anything (e.g., LRAT, CRBP-I, or β-ionone) that effectively blocks STRA6-catalyzed retinol loading can enhance STRA6’s vitamin A uptake. However, STRA6-mediated retinol loading shows a distinct biphasic characteristic not observed in STRA6-mediated retinol release, as revealed by real-time analyses. If STRA6 has equal opportunities to release retinol or to load retinol, the loading activity predominates initially. The distinct kinetic features of STRA6’s retinol loading and release activities suggest that RBP’s loss of retinol changes the nature of its interaction with STRA6, analogous to the differential interactions of holo-RBP and apo-RBP with TTR. This coupling mechanism’s existence also depends on a finely balanced RBP/STRA6 interaction. It would not exist if RBP dissociated from STRA6 immediately after losing retinol, in analogy to RBP’s immediate loss of binding to TTR after losing retinol (12). On the other hand, a very stable RBP/STRA6 interaction would prevent further delivery of retinol by the next RBP. This model of coupling uptake to storage is distinct from all previously published possible mechanistic models of the RBP receptor. For example, no previous model can explain the unexpected finding that a vitamin A analog can drastically enhance (not suppress) RBP receptor-mediated vitamin A uptake.
The fact that β-ionone potently stimulates STRA6-mediated vitamin A uptake suggests that CRBP-I and LRAT are not absolutely required for enhanced STRA6 activity and that small molecules can stimulate STRA6’s vitamin A uptake activity. β-ionone’s ability to enhance cellular vitamin A uptake in a STRA6-dependent manner and in a potency similar to that of CRBP-I suggests a strategy to stimulate STRA6’s vitamin A uptake without a need to increase intracellular expression of CRBP-I or LRAT. Given the specific and dose-dependent nature of β-ionone’s effect, β-ionone or related compounds can be potentially employed to treat human diseases caused by insufficient tissue retinoid levels (28–30). This advantage of this strategy is that it specifically stimulates a natural cellular vitamin A uptake mechanism in cells that naturally take up vitamin A.
The mechanism identified here is notably different from known cellular uptake mechanisms including other receptor-mediated and endocytosis-independent mechanisms (31, 32). It enables STRA6 to take up vitamin A bound with high affinity to its extracellular carrier protein without using cellular energy (like a primary or secondary active transporter) or a preexisting electrochemical gradient of the substrate (like a channels or facilitated transporter). Why did evolution come up with such a new mechanism instead of using an existing one, such as ATP-dependent transport (32) to pump vitamin A into cells? This coupling mechanism is analogous to a pressure-dependent water tap that only opens when water is needed and therefore prevents flooding. This mechanism offers exquisite control of vitamin A uptake from the cell surface, preventing vitamin A toxicity due to the potent and broad biological activities of retinoids, especially in regulating gene transcription during development and in adults (33, 34).
Methods are described in detail in the Supporting Information.
We thank B. Ribalet, K. Philipson, E. Wright, D. Bok, G. Fain, W. Hubbell, J. Nathans, R. Kaback, and B. Khakh for helpful discussion and/or suggestions on the paper, J. Parker for building the LED light source, and A. Rattner for providing the plasmid for retinol dehydrogenase. Supported by NIH grant 5R01EY018144 (H.S.). H.S. is an Early Career Scientist of the Howard Hughes Medical Institute.
Supporting Information Available: This material is available free of charge via the Internet.