| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Institute of Membrane and Systems Biology, Faculty of Biological Sciences, LIGHT Laboratories, University of Leeds, Leeds, UK
1Correspondence: Institute of Membrane and Systems Biology, Faculty of Biological Sciences, LIGHT Laboratories, Clarendon Way, University of Leeds, Leeds LS2 9JT, UK. E-mail: j.b.c.findlay{at}bmb.leeds.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: apo • holo • surface plasmon resonance • mouse urinary protein • lipocalin
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
.
In mammals, RBP circulates as a monomer (21 kDa) in equilibrium with a coupled form involving a tetramer (56 kDa) of TTR. The TTR:RBP complex (
77 kDa) prevents RBP from being lost through glomerular filtration and encapsulates retinol, restricting it from freely partitioning into the plasma membrane.
A specific receptor for RBP in the plasma membrane of many eukaryotic tissues has been reported in placenta (3)
, Sertoli cells (4)
, stellate cells (5
, 6)
, peritubular pigment epithelial and embryonic carcinoma cells (7)
, choroid plexus (8)
, and a variety of tissue culture cells (9
, 10
; as reviewed in refs. 11
, 12
). The free rather than the complexed form of RBP interacts with the receptor.
Sundaram et al. (13)
first advanced the theory, based on transport studies, that this cell surface receptor mediated the transfer of retinol from RBP in the extracellular compartment to one of several structurally related cellular retinol-binding proteins (CRBPs) found within the aqueous environment of the cytosol. Since then more recent work with RBP knockout mice has supported this natural receptor-mediated retinol uptake system (14
15
16
17)
. Very recently, Kawaguchi et al. (18)
reported the identification in bovine retinal pigment epithelial cells of an integral membrane protein as a specific receptor for RBP. Mass spectrometry analysis of the RBP-receptor complex revealed that the RBP membrane receptor was a "stimulated by retinoic acid gene 6 (STRA6) homolog (mouse)," without homology to any protein of known function. They confirmed that it mediates the cellular uptake of retinol from RBP. Particularly strong expression of STRA6 occurred in cells that compose human blood-organ barriers (e.g., the brain, eye, testis, kidney, spleen, and female reproductive tract).
RBP was the first member of the lipocalin superfamily for which an X-ray structure was described (19)
. The basic framework is an eight-stranded antiparallel β-barrel (β-strands named A–H) attached to which is a carboxyl-terminal 
-helix. The amino terminus of the protein, which includes one of the three regions of highly conserved amino acids found in lipocalins, seals one end of the barrel, whereas the opposite end is the entrance to the ligand-binding cavity. The loops connecting the β-strands vary in length, with that joining the A and B strands being the longest and folding over the open end of the barrel.
All-trans-retinol binds to RBP with the isoprene tail fully extended in the β-barrel and the trimethyl cyclohexylene ring innermost in a tight steric fit in the hydrophobic binding pocket. Aromatic amino acids are abundant at the ring end of the ligand, whereas its hydroxyl is positioned at the protein surface, in the region of the entrance loops surrounding the opening of the binding cavity, where it participates in polar interactions. The high degree of complementarity in the shape of the binding site and the structure of retinol is consistent with the fact that RBP is less tolerant of changes to the ring end of the ligand structure where structural differences would be difficult to accommodate without the rearrangement of amino acids in the protein core.
Structure-function studies (20)
, using mutants of RBP, showed that loop AB was involved in the RBP-TTR interaction, whereas loops CD and EF, particularly CD, were heavily involved in RBP binding to the placental receptor. Indeed, the CD loop that guards the entrance to the binding pocket, without any real contacts with the ligand, was found to be the major determinant for receptor recognition and interaction (21)
. The hypothesis advanced in Sundaram et al. (13)
and subsequent reviews (13
, 22
, 23)
involves a high-affinity retinol binding form of RBP interacting with its receptor and releasing retinol to the transport mechanism. In so doing, RBP assumes a lower-affinity form, which can readily be replaced on the receptor. RBP isolated from serum has previously been shown to consist of two forms, one able to bind to the placental receptor with high affinity and the other with low affinity (3)
. The higher-affinity binding form is thought to be converted into the lower-affinity state on detaching from TTR, binding to the receptor, and releasing retinol. This implies some type of conformational communication between the ligand-binding site and the CD loop that interacts with the receptor. To investigate this hypothesis and to confirm the identification of the receptor binding region, we transferred this loop to another lipocalin [mouse urinary protein (MUP)] and examined its receptor binding properties using a ligand quite different from retinol.
MUP2 is one of a series of variants of the major urinary protein, which is an abundant pheromone-binding protein found in male mouse urine, where subtle recognition of a series of related compounds is thought to be involved in its biological function. The major site of MUP synthesis is the liver from which it is secreted into serum where it circulates at relatively low concentrations before being filtered by the kidney and excreted into urine. Expression of MUP mRNA is under different development and hormonal control in different tissues. Unlike RBP, MUP does not bind retinoids but a spectrum of small odorant molecules and pheromones (24)
, and its ligand-binding site is deeper and wider than the one for RBP. Among a number of odorant molecules MUP was reported to bind with the highest affinity to 2-isobutyl-3-methoxypyrazine (IBMP), methyldihydrojasmonate, methylfenchol, and a number of unsaturated cyclic aliphatic compounds (25)
. Furthermore, MUP differs from RBP in the length of the β-strands as well as the conformations and lengths of the loop regions.
Here, we explore whether the receptor-binding properties of RBP could be transferred to a quite different member of the same family, MUP, with the grafting of loop CD and using a ligand other than retinol. We further compared the ligand-binding behavior of the apo and holo forms of the chimeric protein [mouse urinary protein with grafted CD loop from RBP (MUPcd)] with RBP to examine whether there are conformationally dependent interactions and whether any functional differences could be related to structural differences. Based on crystallographic data for the apo and holo forms of MUPcd, we propose a conformation-driven model for the specific interaction of RBP with its membrane receptor initiated by the presence of the ligand facilitated by appropriate positioning of the CD loop.
The second phase of the uptake process involves transfer of retinol to CRBP. The mechanism of this transfer is unknown. To increase our understanding, studies were undertaken to determine whether CRBP also exhibited specific interactions with the membrane and, if so, to develop more fully the hypothesis of whether the receptors for RBP and CRBP were the same protein. This article describes progress on both these issues.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Expression and purification of recombinant lipocalins
cDNAs of RBP, Mup wild-type (MUPwt), and MUPcd were cloned in pQE30 vector (Qiagen), whereas CRBP [P09455 (RET1_HUMAN)] was cloned into pQE30-Xa (Qiagen). All N terminus hexa-His tagged constructs were expressed in BL21-Gold E. coli cells (Stratagene) and purified using Ni-NTA chromatography (Qiagen). Both recombinant MUPs and CRBP were soluble, whereas RBP was expressed as insoluble protein and had to be refolded and purified from inclusion bodies. Competent E. coli BL21 cells were transformed with expression vector constructs containing RBP, CRBP, MUPwt, or MUPcd. Large-scale expression of the clones was performed from overnight starter cultures, which were grown for a further 3–5 h (A600 of 0.6–0.8). Cultures were induced by the addition of isopropyl-β-D-thiogalactoside (Sigma-Aldrich, St. Louis, MO, USA) to a final concentration of 1 mM and incubated for 4 h at 30°C, and cells were harvested. To release soluble proteins (CRBP, MUPwt, and MUPcd) from the periplasmic space, cell pellets were lysed with lysozyme (Sigma-Aldrich) (0.4 mg/ml in PBS, pH 7.4) and sonicated, and the final clarified supernatant was incubated with Ni-NTA slurry for purification of tagged proteins. To release recombinantly expressed RBP from inclusion bodies after cell lysis, inclusion bodies were resuspended in a guanidinium chloride (6 M) (Sigma-Aldrich) buffer, and after a 1-h incubation at 4°C, a slow dialysis method against PBS, pH 7.4, was used for refolding. The mixture was clarified by centrifugation, and the supernatant was incubated with Ni-NTA slurry (previously equilibrated in the same buffer) for purification of recombinant RBP. All bound proteins were eluted with 0.5 M imidazole (Sigma-Aldrich) in PBS and pooled fractions were dialyzed against PBS, pH 7.4. For CRBP, dialysis was performed against Xa reaction buffer, and further Xa digestion and purification of cut protein were performed according to manufacturer’s instructions (Qiagen). The concentrations of the purified recombinant proteins were determined by measuring the absorbance at 280 nm using the molar extinction coefficients: 46,760 M–1 cm–1 for RBP, 24,070 M–1 cm–1 for CRBP, and 16,500 M–1 cm–1 for MUP and the bicinchoninic acid protein assay (Pierce Biotechnology, Rockford, IL, USA).
Membrane preparations
A human embryonic kidney cell line (HEK 293) (Clontech Laboratories, Inc., Palo Alto, CA, USA), known to express the RBP receptor (S. Wang and J. B. C. Findlay, unpublished results, 2000), was propagated as a monolayer and grown at 37°C with 5% CO2, in Dulbecco’s modified Eagle’s medium with GlutaMAX-1 supplemented with 2% penicillin/streptomycin and 10% FBS (Sigma-Aldrich). Membranes were prepared by harvesting a density of 107 cells grown in 175-cm2 flasks to 80% confluence. Harvested cells were lysed in the presence of protease inhibitors (Complete Mini, EDTA-free; Roche Diagnostics), and crude membranes were collected by ultracentrifugation at 50,000 rpm for 1 h at 4°C before resuspension in 2-(N-morpholino)ethane sulfonic acid (MES) buffer (25 mM MES/0.15 NaCl, pH 6.5) (Melford Laboratories Ltd., Suffolk, UK) containing 1% Triton-X 100 (v/v) (Sigma-Aldrich). The final supernatant collected as the solubilized membrane preparation was divided into aliquots and stored at –70°C. When Biacore analysis was performed, the aliquots were diluted 1:10 with eluent buffer before binding analysis (final concentration of Triton X-100 was 0.1%). Human red blood cell membranes were prepared according to the method of Dodge et al. (26)
. Packed red cells were washed four times with 10 vol of 150 mM NaCl and 10 mM phosphate, pH 7.2, at 4°C, and the supernatant and white cell "buffy coat" were removed by aspiration each time. The cells were lysed in 10 vol of ice-cold 5 mM sodium phosphate, pH 8.0 [containing 0.1 mM phenylmethylsulfonyl fluoride (Sigma-Aldrich)], and centrifuged at 15,000 rpm for 20 min. The supernatant was carefully removed, and the process was repeated until the membranes were light pink (three times). These membranes were suspended in ice-cold extraction buffer and treated as above before binding assays.
Placental microvilli membranes were prepared as described in ref. 3
.
Binding assays
Ligand binding
Binding of IBMP to MUP and MUPcd was assessed by isothermal titration calorimetry (ITC) using a MicroCal VP-ITC unit operating at 308 K. Before use, proteins were precipitated with ethanol to remove any endogenous ligands and then redissolved, dialyzed against PBS (pH 7.4), and degassed in a vacuum. The ligand was dissolved in degassed PBS, and all concentrations were measured by UV absorption (MUP-I 
280=16,500 M–1 cm–1; IBMP 220=4980 M–1 cm–1) immediately before titrations were started. The protein concentration was 145 µM, and the IBMP concentration was 1.5 mM. Titrations were performed in duplicate and typically comprised 25 injections (one 2-µl injection followed by 24 5-µl injections) at 4-min intervals. The initial data point was routinely deleted to allow for diffusion of ligand/receptor across the needle tip during the equilibration period. Heats of dilution for the ligand were measured independently and subtracted from the integrated data before curve-fitting in Origin 5.0 with the standard One Site model supplied by MicroCal, which is based on the Wiseman isotherm (27
, 28)
:
 |
H° is the standard enthalpy for reaction, V0 is the effective volume of the calorimeter cell, XR is the ratio of the total ligand to receptor concentrations at any given point during the titration, and r is defined by
 |
Interaction with TTR
MUPwt and MUPcd were saturated with a 5-fold molar excess of IBMP and incubated for 30 min at room temperature. Unbound ligand was removed and holo proteins were incubated with recombinant TTR (kindly provided by Professor M. J. Saraiva, IBMC, Porto, Portugal) in PBS, pH 7.4, for 30 min at 37°C at a 2:1 molar ratio. Macromolecular mixtures were subjected to size exclusion gel filtration (Superdex-75; Amersham Biosciences Corp., Piscataway, NJ) using an AKTA purifier platform system (Amersham). A 24-ml Superdex-75 column was equilibrated with PBS, pH 7.4, overnight, and 0.4-ml samples were injected (1 mg/ml) and eluted at a flow rate of 0.5 ml/min. Protein peaks were detected by monitoring absorbance at 280 nm, and the elution profiles were compared with those of pure TTR and MUP samples. The molecular mass markers used for calibration were aprotinin (6.5 kDa), cytochrome c (12.4 kDa), carbonic anhydrase (29 kDa), albumin (67 kDa), and blue dextran (2000 kDa) (Sigma-Aldrich).
Surface plasmon resonance (SPR) binding assays
Where appropriate, recombinant proteins were incubated with ligands at a 5-fold molar excess for 2 h at room temperature. All-trans-retinol (Sigma-Aldrich) was used to obtain holo-RBP or holo-CRBP and IBMP was used to obtain both holo-MUPwt and holo-MUPcd.
1. To confirm that the CD loop of RBP was the main motif responsible for its interaction with the receptor, the chimera MUPcd was compared with RBP for its ability to interact with solubilized membranes from HEK293 cells. SPR was performed using the Biacore 3000 system and NTA sensorchips (Pharmacia Biosensor, Uppsala, Sweden). The sensorchip was activated according to the manufacturer’s instructions, and the buffers used in the microfluidic system were as follows: eluent buffer (0.01 M HEPES, 0.15 M NaCl, 50 µM EDTA, and 0.005% P20, pH 7.4), regeneration buffer (0.01 M HEPES, 0.15 M NaCl, 0.35 M EDTA, and 0.005% P20, pH 8.3). Recombinant holo and apo forms of RBP, MUPwt, and MUPcd were prepared in eluent buffer and injected over the Ni-NTA-activated flow cells at a concentration typically <300 nM. After a contact time of 15 min at a flow rate of 5 µl/min, solubilized diluted membrane preparations (0.2 mg protein/ml) were injected under the same conditions, and binding events were recorded in real time in triplicate. Solubilized erythrocyte "ghost" membranes (0.2 mg protein/ml) were used as a negative control. The sensorchip surface was regenerated with regeneration buffer in a 3-min pulse injection and reactivated with nickel solution (500 µM NiCl2 in eluent buffer) before each subsequent binding experiment.
2. To test for the presence of a receptor for CRBP in placental microvilli membranes, untagged apo-CRBP was coupled onto a CM5 sensorchip by amine coupling chemistry, and SPR was performed using the Biacore 2000 system. Research grade CM5 sensor chips, N-hydroxysuccinimide (NHS)/1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) coupling reagents, ethanolamine, HBS buffer, and glycine-HCl were from Biacore AB. The sensorchip was activated according to the manufacturer’s instructions, and HBS buffer was degassed for 1 h and used as the running buffer. The flow rate was set to 5 µl/min, and the flow cells were activated for 10 min with a mixture of EDC (0.2 M) and NHS (0.05 M). Apo-CRBP was buffer-exchanged in 10 mM sodium acetate (pH 5.2) and then injected at a concentration of 0.5 mg/ml until 1,200 response units were immobilized on three different flow cells, while the fourth flow cell was used as a mock immobilized, blank control. After immobilization, ethanolamine (1 M, pH 8.5) was injected for 10 min to block the remaining activated amine groups. Glycine-HCl (10 mM, pH 2.5) was then used to wash off any nonspecifically bound ligand. Solubilized diluted membrane preparations of placental microvilli, ghosts, or phospholipid vesicles (0.2 mg protein/ml), with or without heat-treatment or addition of metal ions (K+, Ca2+, Mg2+, or Mn2+, 40 mM), were injected at a flow rate of 5 µl/min and allowed a contact time of 15 min, and binding events were recorded in real time in triplicate. Between the injections, the surfaces were regenerated by two 5-µl injections of 10 mM NaOH, after a 10-min wash with the running buffer.
3. To examine whether RBP and CRBP in their holo or apo forms could bind to the same preparation of solubilized HEK293 membranes, RBP was first immobilized onto a Ni-NTA sensorchip (holo-RBP in two flow cells and apo-RBP on the other two) as described in 1 and after an injection of solubilized membrane components (0.2 mg protein/ml), a second injection of either apo/holo-CRBP (untagged) was given at a flow rate of 5 µl/min, allowing a contact time of 15 min. Binding events were recorded in real time in triplicate.
BiaEvaluation software was used for data processing.
Competition binding assays
Binding to placental microvilli membranes was determined by a competitive assay using 35S-labeled MUPcd against a 100-fold molar excess of either unlabeled MUPcd or RBP. The competition assay was adapted from the oil-centrifugation method described in ref. 3
. In the presence or absence of ligand, in a total volume of 200 µl, 0.1 µM freshly labeled protein was incubated with 100 µl of the membrane preparation and a 100-fold molar excess of unlabeled competitor proteins.
Crystallization and data collection
Optimal conditions for crystallization of MUPcd were achieved by hanging drop vapor diffusion in 50 mM CdCl and 100 mM malate buffer (pH 4.7). Drops containing 1 µl of MUPcd (10 mg/ml) and 1 µl of reservoir solution were equilibrated against reservoir solution by vapor diffusion using the hanging drop method. Crystals of space group P43212 grew over a period of 3–7 days. Ligand soaks were performed by adding neat ligand (IBMP) to the reservoir solution to a final concentration of 0.05% (v/v) and allowing the solution to equilibrate with the drop for 24–48 h. After soaking for 1 min in a cryoprotecting solution consisting of reservoir solution with the addition of 20% (v/v) glycerol and 0.05% (v/v) IBMP, crystals were flash-frozen in liquid nitrogen. Data for apo-MUPcd and holo-MUPcd were collected at a Daresbury synchrotron source (Station 14.1) and processed in MOSFLM/SCALA (29)
. Structure refinement was accomplished with Refmac (CCP4, 1994) and manual rebuilding in Coot (30)
. Crystal coordinates have been deposited in the PDB (accession numbers 2NNE and 2NND).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
) indicated that the enthalpy of binding to MUPcd was more favorable than to that of MUPwt (H of 
–65 and of –48, respectively). However, the overall affinity did not change significantly owing to a less favorable change in entropy (Kd values of 0.3 and 0.87 µM for the wild type and the chimera, respectively). This entropy-enthalpy compensation phenomenon is typical of biomolecular associations. None of the native residues removed from MUP (from 58–63) are directly involved in ligand-binding with IBMP as detailed by Bingham et al. (31)
when the crystallographic structures for apo- and holo-MUPwt are compared. IBMP is enclosed in a hydrophobic environment formed by the side chains of Phe38, Leu40, Leu42, Ile45, Leu54, Phe56, Met69, Val82, Tyr84, Phe90, Ala103, Leu105, Leu116, and Tyr120.
|
Site-directed mutagenesis studies (20)
together with structural data (32)
demonstrated that loop EF (residues 89–99) in RBP is a crucial motif for TTR binding, whereas loops AB and CD only contribute additional contacts in macromolecular complex formation. The chromatographic elution profiles of the mixtures of TTR and recombinant MUPwt, MUPcd, and RBP revealed that there was no formation of macromolecular complexes between TTR and MUPwt as the elution profile obtained showed only two isolated peaks at the corresponding molecular weights for each of the individual proteins present (unpublished data).
Having confirmed the functional integrity of MUPcd with respect to ligand binding, the next step was to analyze the behavior of the grafted CD loop in terms of receptor interaction in the presence and absence of a ligand. Binding phenomena between lipocalins immobilized on the sensor surface and components in the solubilized membrane preparations (mobile phase) were analyzed at 37°C. The observed changes in the relative diffraction indices (response units) were recorded as a function of time. All lipocalins were N-terminally his6-tagged and were efficiently immobilized on Ni2+-activated NTA sensorchip surfaces. As an additional control for specific binding, one of the flow cells was activated, subsequently deactivated, and then examined for interaction. Specific binding data for lipocalin-membrane interactions were obtained by subtracting the control flow cell response from those of the immobilized apo/holo lipocalins.
The first set of experiments involved coupling of either apo or holo (ligand-presaturated) forms of MUPwt, MUPcd, and RBP to the same level on different flow cells and then injecting solubilized HEK293 cells membranes over the sensorchip. Figure 1
A shows the high level of response obtained for both holo-MUPcd and -RBP, which contrasted markedly with the lack of significant binding for the apo forms of these two lipocalins. To confirm this result another set of experiments was conducted by flowing a saturable amount of ligand (IBMP or retinol) over the apo proteins immobilized on the sensorchip before the injection of solubilized membranes. The results demonstrated a significant interaction comparable with those obtained for the presaturated holo proteins coupled to the sensorchip.
|
To determine whether the significant rise in resonance units was due to the specific binding of holo-lipocalins to membrane receptors or the result of nonspecific interactions with other components in the extract, we heated a pooled fraction of the membrane preparation to 80°C, solubilized it, and analyzed the binding characteristics obtained with the heat-treated preparation. The traces obtained with the heat-treated membranes were identical to the traces obtained with the blank surface, indicating that components in the heated membranes did not interact with the lipocalins (not shown).
When the binding experiment was repeated with solubilized erythrocyte ghosts [this membrane system is not thought to contain the RBP receptor (13
, 33)
], the results showed a response similar to that obtained with heat-treated HEK293 cells membranes or with the blank flow cell, offering further support for a specific receptor-dependent effect (Fig. 1B
).
Again, when MUPwt in either the apo- or holo- forms was immobilized on the sensorchip, no significant specific binding was obtained with solubilized HEK293 membranes as the mobile phase. Taken together, the results obtained indicate that the CD loop of RBP was responsible for the specificity of binding to a membrane receptor present in HEK 293 cells. This finding was evident from the observation that both holo forms of RBP and MUPcd interacted with a component in the solubilized HEK293 cell membranes, but holo-MUPwt could not (Fig. 1A
). In contrast, the apo forms of the lipocalins showed much reduced binding to the solubilized membranes, indicating that the presence of the ligand determines a conformation that is recognized by the receptor. Finally, none of the lipocalins either in the apo or holo forms exhibited interactions with red blood cell membranes, showing that HEK293 cells contain a specific receptor.
These results confirmed that the SPR signal measured with the solubilized membranes from HEK293 cells was due solely to the specific binding of lipocalins with a receptor protein in the presence of a ligand and of loop CD from RBP (grafted or native) and agree with results obtained previously (3
, 22
, 34
, 35)
with binding assays using apo and holo preparations of 125I-RBP and placental microvilli membranes. These observations were further confirmed using 35S-labeled MUPcd and placental microvilli membranes (Table 2
). The raw data showed that the interaction of labeled apo-MUPcd with membranes was nonspecific; however, both IBMP-charged MUPcd and retinol-charged RBP were able to displace radiolabeled holo-MUPcd binding to membranes. Displacement from excess unlabeled holo-RBP was 81%, whereas displacement for excess unlabeled holo-MUPcd was 76%.
|
The crystal structures for the apo and holo forms of MUPcd obtained at 1.6 Å resolution (Table 3
) reveal only small differences in the backbones of residues 45–46, 53–70, and 156–160 (Fig. 2
A, B). Perhaps significantly, the CD loop region appeared disordered, consistent with high mobility, and, therefore, these residues were deleted from the crystal structures (Fig. 2C, D
). This observation could be attributed to crystal contacts concealing differences that possibly exist in solution. The crystal structures of liganded and unliganded forms of RBP (36)
showed that only residues 35 and 36 were rearranged on ligand binding. It is clear from the functional studies, however, that in solution the apo and holo forms must have different conformations, not captured by crystal formation.
|
|
Earlier, we demonstrated that the transference of retinol from RBP to CRBP was markedly facilitated by the presence of receptor-containing membranes (13)
. We examined the possibility that this transference reflected the ability of CRBP to also bind specifically to a receptor in the membrane. Figure 3
illustrates that this is indeed the case, as revealed by SPR experiments similar to the ones described above using a different cell system, this time analyzing the behavior of immobilized CRBP and solubilized placental membranes. The analysis of binding events using membrane preparations from different sources proved that the entity interacting with CRBP was present in placental microvilli but not in erythrocyte ghosts, and that, as seen before, specific binding was greatly diminished by heat treatment. As it was previously reported in RBP binding studies with placental vesicles that divalent ions could have a role in retinol uptake (3)
, we examined the effect of the addition of certain divalent cations to placental membranes before injection. Ca2+, in particular, enhanced the binding response to CRBP. The properties of the protein receptor for CRBP in placental microvilli membranes were therefore reminiscent of those for the RBP receptor, most significantly its absence in erythrocyte membranes.
|
We then investigated whether CRBP in its holo or apo forms could bind to the same entities in solubilized HEK293 membranes as did RBP. Figure 4
illustrates the results. RBP was first attached to the sensorchip, and then solubilized HEK293 membranes were introduced. When CRBP was subsequently applied, another binding event was obtained but only with the apo form of CRBP. Holo-CRBP did not exhibit binding. Neither form of CRBP associated with a control flow cell on the sensorchip containing only RBP.
|
The key features of the system are illustrated in Fig. 5
, showing the conformationally sensitive interacting loop and in Fig. 6
, which summarizes the model proposed for the uptake of retinol based on the experimental evidence provided here and earlier.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 37)
revealing a specific mechanism for retinol release from RBP and subsequent entry into the cell via the receptor. This receptor showed high affinity for the holo form of RBP but not for the apo form (3
, 34)
. To further confirm the characteristics of the retinol uptake system, we investigated whether RBP receptor-binding properties could be transferred to another lipocalin, MUP, and whether the same ligand-sensitive conformational effects could be observed. It is apparent from the results presented here that a different lipocalin core, binding a different ligand, is capable of communicating structurally with the grafted RBP loop to effect a conformation that is recognized by the RBP receptor in a ligand-dependent fashion. This finding suggests a conformational coupling mechanism that may be common to many lipocalins and reflects a receptor-dependent biological role.
The binding of RBP to both TTR and the receptor mainly involves different loops near the entrance to the ligand-binding pocket, with TTR via loop EF (residues 89–101) and with the receptor via loop CD (residues 59–68) (20)
. The consequences of these interactions also seem to be different. Interaction with TTR has been shown to stabilize the binding of retinol to RBP. In contrast, binding to the cell surface receptor induces the release of retinol from the binding pocket. Amino acids Trp67, Phe96, Leu63, and Leu97 from RBP are flanked by side chains of TTR. Subsequent crystallographic studies on the RBP:TTR complex confirmed these observations. Also, amino acids Leu182 and Leu183 (C terminus) interact with TTR, buried in an hydrophobic patch which includes Leu82 from TTR and Val69 from RBP (32)
. The identification of the position of the TTR- and receptor-interacting sites on RBP explains why free RBP rather than the RBP:TTR complex interacts with the membrane receptor (3)
. Validating these observations, Melhus et al. (38)
using monoclonal antibodies to synthetic peptides identified the same regions on RBP as being involved in TTR and receptor interactions. We suggest that this constitutes an important mechanism for regulating retinol supply to cells.
The data obtained in this study from crystals for apo-and holo-MUPcd showed that with the exception of the grafted CD loop, there was an overall maintenance of the structure compared with that of MUPwt. The analysis of crystals of holo-MUPcd did not disclose the exact conformational change in the liganded form of the protein that allows receptor binding. The only small visible shifts were in the backbone of residues (45–46), (53–70), and (156–160). In both forms, the CD loop appeared to be disordered, which is indicative of a highly flexible region. The CD loop in RBP was also found to be disordered in lower resolution structures of mammalian RBPs determined previously (36
, 39
, 40)
. This flexibility, suggested by the reported disorder and localization, has lead to the hypothesis that it could be functionally relevant to retinol binding/release. It should be borne in mind, however, that the liganded form was obtained by soaking the ligand into existing crystals of the apo form, so the CD loop orientation could have been fixed during the crystal growth, and that during the apo/holo transition a transient conformational change is accompanied by the process of retinol binding/release.
With ligand bound, the loop is capable of adopting a high-affinity conformation recognized by the receptor. Both nuclear magnetic resonance (NMR) and crystallographic evidence for MUPwt show that this loop is part of a region that rigidifies when in the holo form (31
, 41
42
43
44)
. This finding is also true for MUPcd, being supported by NMR data that indicate differences in backbone chemical shifts of this region between apo- and holo-MUPcd (unpublished observations). In conclusion, we propose that the CD loop in the holo form is less mobile than that in the apo form and when assuming a certain orientation presents a high-affinity conformation that is recognizable by the receptor. The ligand could play an indirect role in the stabilization of loop CD by holding compact the core of MUPcd. So, the present study clearly supports the existence of a reversible conformational state involving a ligand-induced transition that triggers recognition by the membrane receptor.
We therefore put forward a conformational model (Fig. 5)
that summarizes the process by which RBP would perform its essential biological role. [i]: In the liver, before secretion, apo-RBPFree possesses a highly mobile CD loop that does not occlude access to the ligand-binding site. After encapsulation of retinol [ii], the loop assumes a position near the mouth of the ligand-binding pocket, holo-RBPFree, helping to lock the ligand in and also becoming part of the contact surface with TTR [amino acids 63 and 67]. The holo-RBP:TTR complex is then secreted into the plasma [iii], thereby reaching the tissues where specific and controlled delivery of retinol is crucial. Receptor recognition at the cell-membrane interface occurs via the CD loop [iv], so the key and so far largely structurally undefined interaction is between this loop and the membrane receptor. At face value, TTR might be expected to hinder this interaction and clearly this appears to be the case from biochemical studies (20)
. But, nevertheless, the receptor does access and bind tightly to the CD loop, suggesting a dynamic equilibrium that is reflected in the presence of both TTR-bound and -free forms of RBP. This, in turn [v], may trigger a series of structural motions in the lipocalin, "overwriting" the TTR closed conformation effect. By interacting with the receptor, the CD loop may "unlock" the binding site, provoking deeper conformational changes in RBP, which probably also involve a conformational switch of loop AB [Zanotti et al. (36)
proposed that Leu35 controlled the movement of retinol into and out of the RBP ligand-binding pocket] culminating in release/loosening from TTR and discharge of retinol to the receptor. The resultant apo-RBPFree has low affinity for both TTR and the receptor and is rapidly excreted via the kidney [vi]. Figure 5
illustrates hypothetical conformations assumed by the CD loop in the presence or absence of ligand, allowing the interaction/detachment of macromolecular complexes, such as TTR and the receptor.
The exact process by which the receptor facilitates the subsequent transfer of retinol to CRBP, however, has yet to be elucidated. One possibility is that the receptor, either on its own or with associated proteins, might shape the transmembrane mechanism through which retinol is transported into CRBP, with the hydroxyl group leading the way. Such a system would position the retinol in the binding pocket of CRBP in the proper orientation (Fig. 6)
. Therefore, retinol undergoes a reorientation when it is delivered by holo-RBP via the plasma membrane receptor to CRBP within the cytoplasm of target cells. Nevertheless, the cavity remains substantially apolar around the ligand. The hydrophobic residues lining the retinol-binding site are in close contact with the bound retinol, such that the vitamin must fit into a rather narrow and highly complementary hydrophobic binding cavity. In the cell, the poorly water-soluble retinol is stored as a retinyl ester derivative of long-chain fatty acids, whose CRBP-dependent synthesis is catalyzed by the enzyme lecithin:retinol acyltransferase (LRAT).
The results presented by van Aalten et al. (45)
suggest that the binding site of CRBP to its putative receptor would involve loops CD, EF, and helixII. The earlier study investigating the process by which retinol was transferred from plasma RBP to intracellular apo-CRBP, through the plasma membrane, indicated that the transfer was facilitated by membranes that contained the RBP receptor (13)
. In the current work, using SPR and solubilized placental membranes, CRBP exhibited specific Ca2+-dependent binding, which was greatly reduced by heat treatment of membranes. No specific binding was obtained with ghost membranes and phospholipid vesicles, implying that a specific membrane binding site for apo-CRBP might exist on the cytoplasmic side of the membrane.
We next examined whether the specific binding of CRBP was dependent on the apo form of the protein and at the same time whether there was an involvement of RBP. The experiment involved first the application of solubilized HEK293 membranes to the sensorchip containing immobilized holo-RBP and subsequently the introduction of either apo- or holo-CRBP. From the SPR traces shown in Fig. 4
, it can be seen that a receptor element first bound to the immobilized holo-RBP. Subsequent addition of apo- but not holo-CRBP, led to its interaction with the RBP:RBP receptor complex. Because RBP does not interact with CRBP or LRAT, we interpret this result to suggest that a protein component(s) that binds to RBP is also recognized by CRBP. This is the first indication that the mechanism of cellular uptake of retinol from RBP involves not only a specific RBP receptor (as recently validated in ref. 18
) but also a receptor (the same?) for the retinol recipient CRBP.
These observations are consistent with previous studies (13
, 35)
using human placental brush border membranes, in which the holo form of CRBP was ineffective for inducing the release of retinol from the receptor-bound RBP. Clearly, CRBP must also undergo the subtle conformational dynamics exhibited by RBP. Many of the details of this mechanism can now be further elucidated with the identification of the RBP receptor gene (18
, 46)
.
|
ACKNOWLEDGMENTS |
|---|
Received for publication August 15, 2007. Accepted for publication October 4, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. A. Kane, A. E. Folias, C. Wang, and J. L. Napoli Ethanol elevates physiological all-trans-retinoic acid levels in select loci through altering retinoid metabolism in multiple loci: a potential mechanism of ethanol toxicity FASEB J, March 1, 2010; 24(3): 823 - 832. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
