(Received for publication, July 26, 1996, and in revised form, December 2, 1996)
From the Laboratory of Cell Biochemistry and Biology, NIDDK, National Institutes of Health, Bethesda, Maryland 20892
To understand the subcellular localization of the vitamin D receptor (VDR) and to measure VDR content in single cells, we recently developed a fluorescent labeled ligand, 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene (BODIPY)-calcitriol. This tagged hormone has intact biological activity, high affinity and specific binding to the receptor, and enhanced fluorescent emission upon receptor binding. Using BODIPY-calcitriol, here we monitored the subcellular distribution of VDR in living cultured cells by microscopy. Time course studies showed that an equilibrium between the cytoplasmic and nuclear hormone binding developed within 5 min and was maintained thereafter. We found a substantial proportion of VDR residing in the cytoplasm, colocalized with endoplasmic reticulum, the Golgi complex, and microtubules. Confocal microscopy clarified the presence of VDR within discrete regions of the nucleus and along the nuclear envelope. There was no VDR in the plasma membrane. Low affinity BODIPY-calcitriol binding sites were in the mitochondria. Mutations in the VDR gene selectively and specifically altered BODIPY-calcitriol distribution. Defects in the hormone binding region of VDR prevented both nuclear and cytoplasmic hormone binding. Defects in the DNA binding region decreased the nuclear retention of VDR and prevented localization to nuclear foci. These results with BODIPY-calcitriol reveal cytoplasmic VDR localization in living cells and open the possibility of studying the three-dimensional architecture of intranuclear target sites.
The vitamin D receptor (VDR)1 belongs to the v-erb-A superfamily of ligand activated transcription factors. The hormonal forms of vitamin D and other steroid hormones act through their receptors to regulate the transcription of target genes and thus modulate a variety of cell functions. These hormones also exert rapid, so-called "nongenomic" actions that take place outside the nucleus. Over the last 30 years we learned many aspects of steroid receptor activation, but the subcellular distribution of the receptors remained controversial. The use of radioligands led to the classical model for steroid receptor activation, which placed the receptors in the cytoplasm (1). Later, antibodies were raised against steroid receptors, and immunocytology suggested that unactivated steroid receptors reside exclusively in the nucleus (2). In the past few years, evidence has accumulated showing that glucocorticoid, mineralocorticoid, and androgen receptors reside both in the cytoplasm and in the nucleus. Nevertheless, the consensus has remained unchanged for the nuclear localization of estrogen receptors, thyroid hormone receptors, and VDR (3). Several studies with immunocytology on aldehyde-fixed cells showed that VDR resides exclusively in the nucleus (4). However, using microwave fixation, we found a significant portion of VDR in the cytoplasm (5). Since then, cytoplasmic VDR has been found by others (6-10), but the existence of cytoplasmic VDR is still not generally accepted (11, 12).
While the role of nuclear steroid receptors in regulating transcription is clear, little is known about the function of cytoplasmic or plasma membrane receptors. The putative plasma membrane receptors have been suggested to mediate the nongenomic effects of calcitriol (13-15). Cytoplasmic VDR is another possible candidate to mediate nongenomic actions. We have demonstrated that cytoplasmic VDR mediates a rapid increase in intracellular cGMP (16, 17). Recent articles show that VDR, together with other members of the steroid receptor superfamily of proteins, associates with calreticulin, the major calcium-sequestering protein of the endoplasmic reticulum (ER) lumen (18-20). Calreticulin inhibits vitamin D signal transduction by interacting with a protein motif in the DNA-binding domain of the VDR (21, 22). This interaction could occur in the ER or in the nucleus, since calreticulin is present in both compartments. The interaction of calreticulin with VDR in the ER could mediate a rapid increase in intracellular calcium by calcitriol, triggering the inositol triphosphate receptor. Clarifying the presence of VDR in the plasma membrane or the colocalization of VDR with calreticulin in the ER could explain the mechanisms of the nongenomic actions.
The intranuclear distribution of VDR is just as controversial as its cytoplasmic localization. With immunocytology, we found VDR in discrete foci within the nucleus (5), while most other studies described a diffuse pattern. Such disagreements resulted from immunocytology data on other steroid receptors (11, 23, 24). We recently demonstrated, for the first time, a nonrandom distribution of glucocorticoid receptors within the nucleus of living cells using a green fluorescent protein chimera (25), but no such studies have been done yet on VDR in living cells.
Much of the current controversy about receptor localization is due to
technical difficulties. The use of fluorescent ligands could overcome
these obstacles. Recently, we succeeded in developing a
pharmacologically relevant fluorescently labeled steroid hormone, 3-BODIPY-calcitriol (BP-calcitriol) (26). This reagent is
stable, is resistant to endogenous esterases, freely enters living
cells, and retains most of the biologic activity of the parent hormone and its affinity to the receptor. The high quantum efficiency of the
BODIPY dye and its enhanced fluorescent emission upon receptor binding
allows the use of low hormone doses. We have established by several
criteria the specificity of BP-calcitriol binding to the receptor both
in high salt extracts from cells and in intact cells. Specificity was
supported by the fact that the relative potencies of calcitriol analogs
to compete with calcitriol were similar to their potencies to compete
with BP-calcitriol. We also demonstrated that BP-calcitriol binding is
proportional to the VDR content of cells or extracts. For these studies
we used cell lines that had a normal amount of VDR, lacked VDR, or
overexpressed VDR. We also showed that the specificity of binding is
determined by the steroid component of the labeled hormone and that
binding affinity is influenced by the position of the BODIPY label on the steroid molecule. Binding studies with BP-calcitriol and with radiolabeled calcitriol gave comparable results on VDR content in
various cell lines. Binding studies with BP-calcitriol also showed the
same VDR content whether we used spectrofluorimetry or microscopy (26).
Thus, this new reagent opens the possibility of studying the
subcellular distribution of VDR in living cells by microscopy.
Here we present evidence for the presence of VDR in the ER, along microtubules, in the Golgi, in the nuclear envelope, and in discrete foci within the nucleus. Immunocytology on fixed cells also shows colocalization of VDR with calreticulin, thus supporting a role for calreticulin in VDR actions. Real time imaging shows the time course of hormone uptake into living cells. Studies on fibroblasts from patients with hereditary resistance to calcitriol reveal that mutations in the VDR gene specifically alter VDR distribution. A DNA binding defect of VDR prevents BP-calcitriol accumulation in nuclear foci. This strongly suggests that the intranuclear foci represent specific binding sites of VDR to target genes.
Normal human skin fibroblasts and skin fibroblasts from patients with hereditary resistance to calcitriol were from biopsies as described previously (16). In fibroblasts from patients with vitamin D-dependent rickets type II, previous studies characterized the homozygous point mutations in the VDR gene and the resulting defects in VDR functions (5, 16, 17, 27-29). MCF7 human breast cancer cells were a gift from M. Lippman (Georgetown University, Washington, D. C.), LLC-PK1 porcine kidney epithelial cells were from American Type Culture Collection, and ROS 17/2.8 rat osteosarcoma cells were a gift from G. Rodan (Merck Sharp and Dohme). Cells were cultured as described earlier (5). Cell culture media were from Biofluids Inc. (Rockville, MD), and fetal bovine serum was from HyClone Research Lab Inc. (Logan, Utah). For microscopy, cells were subcultured to Lab-Tek glass tissue chamber slides (Nunc Inc., Naperville, IL), and kept for 24 h in serum-free media supplemented with an additive containing insulin, transferrin, and selenium from Collaborative Biomedical Products (Bedford, MA). Cells were used for experiments at about 80% confluence.
ChemicalsUnlabeled 1,25-dihydroxycholecalciferol
(calcitriol) was a generous gift from M. Uskokovic (Hoffmann-LaRoche)
for preliminary studies and later was purchased from Duphar B.V. Other
vitamin D analogs and steroids were from Sigma or from Hoffmann-LaRoche as described (5). BODIPY-calcitriol compounds were stored at 20 °C
either in Me2SO or ethanol in the dark. At these conditions they were stable for at least 1 year. Chemical purity and
concentrations of labeled or unlabeled steroids (including
BP-calcitriol) were tested before each experiment by high pressure
liquid chromatography and by spectrophotometry. Rhodamine 123, rhodamine B hexyl ester, NBD-ceramide,
4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acid methyl ester (BODIPY-methyl ester),
4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acid, and BODIPY-nafoxidine conjugate were from Molecular Probes, Inc.
(Eugene, OR). Nocodazole was from Aldrich (Milwaukee, WI). Colchicine,
-luminocolchicine, and cytochalasins were from Sigma. All other
reagents were of the highest purity available.
Cells were exposed to BP-calcitriol for various lengths of time (between 15 s and 72 h) and in various doses (between 0.01 nM and 1 mM), with and without a 100-fold excess of unlabeled calcitriol. Competition studies were routinely done with 10 nM BP-calcitriol for 30 min. This was the lowest concentration of BP-calcitriol giving a strong fluorescent signal at high magnification. Hormone exposure was followed by three 10-min washes with Eagle's medium. All incubations were performed at 37 °C, with shaking and shielded from light. Controls included incubation with other BODIPY-labeled steroids.
Colocalization of BP-Calcitriol with Endoplasmic Reticulum, Golgi, and Mitochondria in Living CellsMitochondria were identified with rhodamine 123 as described (30). Briefly, normal human skin fibroblasts were incubated with 1 µg/ml rhodamine 123 in assay buffer at 37 °C for 30 min either with or without 10 nM BP-calcitriol. Cells were then washed three times with a fresh buffer. Localization of the Golgi complex by NBD-ceramide was according to the protocol supplied by Molecular Probes. Microfilaments were labeled with rhodamine phalloidin in living cells according to the protocol from Molecular Probes. The ER in living cells was visualized by rhodamine B hexyl ester as described (31) and by the red fluorescent BODIPY-conjugate of brefeldin A (rBFA, a generous gift of J. W. Yedell, National Institutes of Health (NIH)) as described (32). Colocalization of the labeled cellular elements was observed by fluorescence microscopy using appropriate filter sets for a complete separation of signals.
ImmunocytologyCells were incubated with or without 10 nM BP-calcitriol at 37 °C for 30 min in buffer A (Hanks' balanced salt solution without phenol red from Life Technologies, Inc., with 2 mg/ml glucose and 1 mg/ml bovine serum albumin). After hormone exposure, cells were fixed with 2.1 s of microwave irradiation as described before (33). A primary antibody against VDR (9A7, monoclonal rat; Affinity Bioreagents, Inc., Neshanic Station, NJ) was added in a 1:800 dilution in Dulbecco's phosphate-buffered saline without calcium and magnesium (DPBS) (Biofluids Inc., Rockville, MD) with 4% human serum for 1 h, followed by a 30-min incubation with a lissamine-rhodamine labeled anti-rat second antibody in a 1:50 dilution with 1 mg/ml bovine serum albumin in DPBS. A second primary antibody against calreticulin (PA3-900, polyclonal rabbit; Affinity Bioreagents) was used in a 1:200 dilution for 1 h, followed by Cy5-conjugated affinity-purified anti-rabbit IgG in a 1:80 dilution (Jackson ImmunoResearch Laboratories Inc., West Grove, PA). In other experiments, cells were fixed with Bouin's solution (Sigma) for 5 min, washed three times with phosphate-buffered saline, and permeabilized with 0.1 mg/ml digitonin (Sigma). Anti-tubulin primary antibody (mouse monoclonal, from Sigma) was added in a 1:500 dilution in DPBS with 4% human serum for 1 h, followed by a 30-min incubation with Cy5-conjugated affinity-purified goat anti-mouse IgG (from Jackson ImmunoResearch) in a 1:80 dilution in DPBS with 1 mg/ml bovine serum albumin. Each incubation step was followed by three 5-min rinses with DPBS. All staining procedures were done at room temperature in a humidified box, shielded from light, and continuously tilted. The validity and specificity of immunostaining was established by substitution of primary antibody with nonimmune serum or ascites fluid and by including single-labeled samples together with double-labeled ones, allowing tests for proper separation of fluorescent signals.
Effects of Cytoskeleton-disruptive AgentsTo test the
effect of tubulin-disruptive agents, normal human skin fibroblasts were
preincubated with colchicine (25 µM for 1 h),
nocodazole (30 µM for 30 min), or -luminocolchicine
(10 µM for 1 h). Immunocytology with anti-tubulin
antibodies established the effectiveness of tubulin depolymerizing
agents. NBD-ceramide was used to visualize the effect of nocodazole on
dispersing the Golgi complex. Drugs affecting the microfilament system
(cytochalasin B (20 µM for 1 h) and cytochalasin D
and phalloidin (each 10 µM for 1 h)) induced
characteristic morphological changes including bulging of the nucleus
and complete disruption of the microfilament system detected by
rhodamine phalloidin staining (data not shown). Inhibitors were also
present during treatment with BP-calcitriol and during subsequent
washing.
Samples were evaluated using a Zeiss Axiovert 10 microscope equipped for epifluorescence. Images were acquired automatically, using computer-controlled illumination and a motorized stage and focus (Ludl Electronic Products, Inc.) to eliminate investigator bias and to assure limited and standard photodegradation. Filters were made by Chroma Technology Corp., optimized for BODIPY, Texas Red, CY-5, or rhodamine. Images were acquired with a high resolution, cooled charged couple device camera equipped with an electromechanical shutter (Photometrics p200). The imaging system was built around a Silicon Graphics workstation (4D310-VGX). Software for automated measurements, calibration, morphometry, and statistical analysis was developed in our laboratory, incorporating functions from a vendor-supplied library (G. W. Hannaway & Associates). Image aquisition and calibration were done as described earlier (26). Brightness values were calculated from unprocessed images and were expressed in brightness units (B). Values from hormone-treated cells were corrected for the autofluorescence (brightness of images from cells exposed to vehicle alone). For normal human skin fibroblasts, this autofluorescence varied among experiments (mean brightness values for controls were between 1 and 11 B). For competition studies, BP-calcitriol retention was calculated from the mean brightness of 100 randomly selected fields. For more detailed analysis, overlays were created on the phase contrast images over the nucleus and cytoplasm and were applied to fluorescent images for brightness calculations. The differences between brightness of samples were statistically analyzed using a paired t test or analysis of variance when appropriate.
Alternatively, for selected experiments, images were taken from a Zeiss Axioplan microscope equipped with a Bio-Rad MRC-600 confocal laser-scanning unit or from a Nikon Optiphot microscope equipped with a Bio-Rad MRC-1024 confocal laser-scanning unit. The fluorescent excitation was produced by a krypton-argon laser; for BP-calcitriol, the 488-nm line was used; for VDR immunolabeling, the 514-nm line was used, and for calreticulin, the 647-nm excitation line was used with the appropriate emission filters. When pictures were taken for spatial analysis with the high resolution CCD camera, the out of focus fluorescence and a substantial portion of the diffuse signal was removed digitally.
The unique characteristics of BP-calcitriol gave us the opportunity to study hormone uptake in living cells in real time. We studied the time course of BP-calcitriol uptake and the specificity of its retention in normal human fibroblasts.
BP-calcitriol freely entered normal human fibroblasts within seconds
and remained there after washes (Fig. 1). The amount of
retained BP-calcitriol increased in a time-dependent
manner. Steroid-free fluorochrome (BODIPY-methyl ester or the free acid form of BODIPY) also entered the cells freely, but it was completely washed out even after long incubations with high concentrations (Fig.
1D). Similarly, other cell lines from calcitriol target tissues (MCF7 human breast cancer cells, LLC-PK1 porcine kidney epithelial cells, and ROS 17/2.8 rat osteosarcoma cells) retained BP-calcitriol but did not retain the free dye (data not shown). Thus,
the steroid component of the molecule is required for the retention of
BP-calcitriol by calcitriol target cells.
It was possible for the first time to measure the retained amount of
BP-calcitriol by digital image analysis. Hormone uptake was detectable
not only in whole cells but also within cell compartments. Brightness
measurements of images from fibroblasts exposed to 10 nM
BP-calcitriol with and without unlabeled calcitriol were done as
described under "Materials and Methods." While expressing the data
in mean brightness units does not reflect the visual impression of
BP-calcitriol distribution, this calculation is the most compatible
with previous radioligand measurements of whole cell and nuclear
uptake. On the micrographs, the nucleus appears darker than the
perinuclear region. When the nonspecific binding is subtracted and the
large dark areas in the cytoplasm are also taken into account, the
resulting mean brightness values show higher specific nuclear uptake
than cytoplasmic uptake. Time course measurements showed that the
specific uptake in the cytoplasm appeared as early as 15 s (Figs.
1B and 2). However, the nuclear uptake was
not detectable until 5 min. After 5 min, the mean brightness of the
nucleus started to exceed the mean brightness of the cytoplasm, and
after 4 h, the nucleus was 1.4 times brighter than the cytoplasm. Both the cytoplasmic and the nuclear retention reached half-maximum after 1 h, and were maximal at 4 h, reflecting up-regulation
of VDR (33). The ratio of nuclear to cytoplasmic VDR was unchanged. After 12 h, we detected a temporary decrease in specific hormone uptake. Thereafter, during the second and third days, hormone retention
decreased gradually in both compartments. After 3 days, the mean
brightness of the nucleus no longer exceeded the mean brightness of the
cytoplasm, and both were one-fifth of the maximum.
These studies show that brightness measurements on BP-calcitriol-exposed cells are useful for dynamic studies of hormone uptake by target cells. With BP-calcitriol, measurements of hormone partitioning between cell compartments became possible even in single cells. These results suggest the existence of a mechanism that maintains a balance in the nuclear/cytoplasmic ratio of VDR, regardless of the time-dependent changes in VDR content.
Subcellular Distribution of BP-Calcitriol BindingStudies with BP-calcitriol allowed us to explore most of the open questions related to subcellular distribution of VDR in living cells. The debate over nuclear versus cytoplasmic partitioning has received the most attention in the past. We found a significant amount of fluorescence within the nucleus, but at least half of the signal was in the cytoplasm for all of the vitamin D target cells tested (normal human skin fibroblasts and LLC-PK1, MCF7, and ROS 17/2.8 cells).
Next, we wanted to clarify if we could detect receptors in the plasma membrane. In normal human skin fibroblasts, we reliably detected BP-calcitriol binding sites in the nucleus and in the cytoplasm, but we did not detect any BP-calcitriol accumulation in the plasma membrane. There was a possibility that the plasma membrane receptors internalized very rapidly. To rule out this possibility we tested BP-calcitriol distribution at very early time points and also in cells kept at 4 °C. At the earliest time point tested (15 s), BP-calcitriol was already in the cytoplasm, along tubulin-like fibers (Fig. 1B). At a low temperature, we could not detect plasma membrane binding sites either (data not shown).
BP-calcitriol, like the unlabeled hormone, binds to low affinity, high
capacity binding sites and to high affinity, low capacity binding sites
(VDR). To differentiate between the distribution of these two types of
binding sites, we added graded doses of BP-calcitriol in the presence
or absence of excess unlabeled calcitriol. In fibroblasts, the addition
of excess unlabeled calcitriol eliminated the BP-calcitriol signal from
the high affinity binding sites in the cytoplasm and in the nucleus
(Fig. 3B). The remaining signal was spread
diffusely in the cytoplasm and the nucleus and within cytoplasmic
speckles. The BP-calcitriol speckles (viewed through a fluorescein
filter) colocalize with rhodamine 123 signals (viewed through a
rhodamine filter) (data not shown). Since rhodamine 123 is a
cell-permeant fluorescent dye that is sequestered by active
mitochondria and does not stain the ER, we concluded that low affinity
binding sites for BP-calcitriol are located in the mitochondria.
The receptor-bound portion of the cytoplasmic BP-calcitriol was
predominantly in the perinuclear region: in the Golgi region, in
structures resembling ER (forming three-way junctions and free ended
tubules), and along fibers with a typical microtubule pattern (Figs. 1,
3, 4, and 6-9).
To further explore the presence of BP-calcitriol in the ER, we
proceeded with colocalization studies in living and in fixed cells.
Double labeling of living fibroblasts with BP-calcitriol and with rBFA,
a new dye labeling ER and the Golgi complex (32), was carried out.
BP-calcitriol (10 nM) and rBFA (0.37 µg/ml) were added
with and without unlabeled calcitriol (500 nM) for 30 min at 37 °C. Complete colocalization of specific BP-calcitriol and rBFA
signals was evident in images taken by the cooled CCD camera using
epifluorescence microscopy and was confirmed by confocal microscopy
(Fig. 4). In addition, BP-calcitriol-exposed cells were
fixed with microwave irradiation, and then the ER was labeled with
rhodamine B hexyl ester. The pattern of rhodamine-stained networks
showed close similarity to BODIPY-labeled network patterns in the
regions where the network was not too dense (data not shown). Comparing
images taken from fibroblasts labeled with rhodamine B hexyl ester and
with rBFA, we noticed that the ER in the rBFA-exposed samples sometimes
appeared to be swollen. This morphological change did not affect
colocalization with BP-calcitriol. Furthermore, recent publications
indicated a possible interaction of VDR with another ER resident
protein, calreticulin. Thus, we did double labeling of VDR and
calreticulin in cultured fibroblasts after fixation. Fig.
5A shows colocalization of VDR and
calreticulin in the ER before calcitriol exposure (yellow
signal). A 30-min exposure to 10 nM calcitriol resulted in
increased amounts of VDR fluorescence in nuclear foci (green
signal in Fig. 5B). The same calcitriol exposure did not
result in an increased amount of calreticulin signals (red
signals) in the nucleus. Thus, VDR colocalized with calreticulin in the
cytoplasm but not in the nucleus. The colocalization of BP-calcitriol
with ER and Golgi, and of calreticulin with VDR, strongly suggests the
presence of VDR in the Golgi and in the ER. To study the functional
significance of this colocalization, we exposed fibroblasts to 1 µM thapsigargin for 30 min and then together with 10 nM BP-calcitriol with or without a 50-fold excess of
unlabeled hormone for another 30 min. We found that the depletion of ER
Ca2+ stores by thapsigargin treatment completely blocked
the BP-calcitriol binding to VDR, while the low affinity binding to
mitochondria was not affected (Fig. 6; competition not
shown). Colocalization experiments in living and fixed cells
demonstrated the presence of VDR in the ER, indicating that
calreticulin interaction with VDR can take place in the ER. Experiments
with thapsigargin suggested that the regulation of ER Ca2+
stores could play a role in VDR activation.
Figs. 4 and 5B show that BP-calcitriol (green
signal) not only associates with ER membranes but also associates with
other cytoplasmic fibrillar structures. Our previous studies with
immunocytology suggested that these fibrillar structures could be
microtubules (34). Therefore, we carried out double labeling
experiments on fibroblasts using BP-calcitriol and anti-tubulin
antibodies. Fig. 7 shows that several fibrillar elements
were associated with both tubulin immunoreactivity and BP-calcitriol
(yellow signals), while both BP-calcitriol (green
signals) and tubulin immunoreactivity (red signals) occurred
at other places in the cytoplasm as well. Double labeling experiments
with rhodamine phalloidin and with BP-calcitriol did not suggest
colocalization of VDR with actin fibers (data not shown).
Tubulin-disruptive agents were used to further confirm BP-calcitriol
association with microtubules. Nocodazole treatment caused dramatic
changes in BP-calcitriol distribution in normal fibroblasts (Fig.
8). BP-calcitriol was not in the perinuclear region but was predominantly in the peripheral regions of the cytoplasm. Staining
in the ER and in mitochondria was unaffected. Treatment with colchicine
had a similar effect. The inactive isomer of colchicine (10 µM -luminocolchicine) was ineffective, as were
microfilament-destabilizing cytochalasins (data not shown). The effect
of nocodazole on the distribution of BP-calcitriol was reversible; a
wash-out after 30 min restored normal BP-calcitriol pattern.
As expected, BP-calcitriol also accumulated within the nucleus and the
nuclear envelope. The intranuclear signal had a diffuse component and
accumulation foci, as well. The discrete foci excluded the nucleoli.
This finding was confirmed by confocal microscopy both in normal skin
fibroblasts and LLC-PK1 porcine kidney epithelial cells (Fig.
9). The nuclear VDR accumulation foci were frequently in
the lower portion of the nucleus, closer to the glass attachment surface; they were not found in the nucleoli.
BP-Calcitriol Distribution in Fibroblasts with a Mutation in the VDR Gene
Skin fibroblasts from patients with hereditary
resistance to calcitriol were characterized earlier to show the
functional defects in VDR, and the mutations in the VDR gene causing
these defects (5). We used here three cell lines with different
mutations in the hormone binding region of the VDR (P8, P10, and P11)
to exclude the possibility that the high affinity BP-calcitriol binding sites in the ER and Golgi and along microtubules represent binding to
proteins other than VDR. In these mutant cells, we did not detect
BP-calcitriol binding in the ER, Golgi, or along tubulin-like fibers
(Fig. 10). The staining patterns were just like the
pattern in normal cells when they were exposed to BP-calcitriol in the presence of excess unlabeled calcitriol (diffuse and intramitochondrial signals) (compare Figs. 10 and 3). The difference between the staining pattern of normal skin fibroblasts and P8 fibroblasts (a cell line with
calcitriol binding defect) was specific for BP-calcitriol. After
exposure to 10 nM BODIPY-nafoxidine, an estrogen
derivative, we could not see any difference between the staining
pattern of normal and P8 cells (data not shown). This finding further
supports our conclusion that the BP-calcitriol signals in the
perinuclear area represent VDR locations.
The punctate pattern of nuclear BP-calcitriol accumulation suggests that these foci could represent the locations of VDR-regulated genes. To test this hypothesis, we studied BP-calcitriol distribution in skin fibroblasts from a patient with hereditary resistance to calcitriol due to a genetic defect in the DNA-binding zinc finger region of the VDR (P7). Prior studies showed in P7 cells the inability of calcitriol to induce VDR binding to DNA or to induce transcription of target genes (27).
The BP-calcitriol pattern in P7 cells was different from the pattern in normal cells (Fig. 5F). The most striking abnormality was the absence of BP-calcitriol retention in nuclear foci, while a decreased amount of diffuse signal remained in the nucleus. This lack of BP-calcitriol foci in fibroblasts with DNA-binding defects of VDR strongly suggests that the foci in normal cells represent VDR target sites. In addition, we found subtle differences in BP-calcitriol patterns outside the nucleus. Unlike in normal cells, where BP-calcitriol was predominantly in the perinuclear region, in P7 cells it was evenly distributed throughout the cytoplasm. BP-calcitriol was also found, as in normal cells, along microtubules and in the ER.
The subcellular partitioning of VDR in P7 cells was different from the partitioning in normal cells. Unlike in normal cells, in P7 cells the mean brightness of the nucleus did not exceed the mean brightness of the cytoplasm. The nucleus/cytoplasm ratio was 0.96 ± 0.04, significantly different from normal (p < 0.005). For a more detailed quantitative analysis of binding sites, we exposed normal and P7 fibroblasts to 10 nM BP-calcitriol with or without an excess of unlabeled hormone. Brightness measurements of cell compartments revealed that P7 cells have moderately decreased amounts of high affinity binding sites as compared with normal cells. In normal cells, 69.3 ± 18% of the cytoplasmic, and 74.2 ± 24% of the nuclear total brightness were displaced, whereas in P7 cells 41.6 ± 26.6% of the cytoplasmic and only 35.8 ± 29% of nuclear total brightness were displaced. These data suggest a defect in the nuclear retention of VDR because of a defect in the DNA binding region of the VDR gene.
The development of BP-calcitriol has provided an important new tool for studying the process of VDR activation in its physiological setting of living cells rather than in fixed cells. Due to the high quantum efficiency of the BODIPY dye, it became possible for the first time to follow the cytoplasm-to-nucleus translocation of VDR by time lapse videomicroscopy and to study the architecture of the intranuclear VDR binding sites by confocal laser-scanning microscopy.
To obtain BP-calcitriol, the fluorescent dye BODIPY was attached at the
A-ring of calcitriol by an ester link at the 3-position (26). We
have shown the utility of BP-calcitriol to measure hormone binding to
VDR, and the specificity of BP-calcitriol binding has been established
(26). The best feature of this reagent is that its fluorescence
emission increases when bound to VDR but does not increase when bound
to other proteins. Using spectrofluorimetry, we showed that the
hormone-free BODIPY dye is not retained by cells in suspension. Here we
show in single cells that the retention of BP-calcitriol depends on the
steroid component of the molecule. We also show that defects in the
hormone binding region of VDR result in the loss of high affinity
BP-calcitriol binding. Specific and high affinity binding of
BP-calcitriol to VDR allowed us to study hormone uptake and retention
in intact cells by microscopy.
The time course and the cytoplasmic/nuclear partitioning of specific and nonspecific binding for BP-calcitriol were in accordance with our results on whole cell and nuclear [3H]calcitriol uptake in fibroblasts (data not shown) as well as those of others (7, 35-38). The loss of fluorescence after 24 h correlated with our previous findings on the metabolism of the labeled hormone by cells (26). Our results in cultured cells show that by detecting the amount of fluorescence after BP-calcitriol exposure, we can reliably measure hormone uptake and the equilibrium between uptake and extrusion in living cells. Measuring VDR expression and intracellular hormone concentrations in single cells with BP-calcitriol could have important clinical applications.
These studies with BP-calcitriol in living cultured cells with microscopy reveal new and interesting features of VDR distribution. By measuring the number of specific BP-calcitriol binding sites in the nucleus and in the cytoplasm of living cells, we found that at least half of the VDRs reside in the cytoplasm. This observation agrees with our findings with immunocytology on microwave-fixed cells and contradicts the notion that VDRs reside exclusively in the nucleus. Exclusively nuclear localization can be generated by activating the receptor without ligand, or by technical difficulties in detecting cytoplasmic receptors. Such difficulties can occur due to fixation and permeabilization procedures, limitations in antibody sensitivity and specificity, or problems with the sensitivity and resolution of imaging equipment.
In the cytoplasm, we found VDR associated with microtubules, ER, and Golgi membranes. Our previous observations with immunocytology on microwave-fixed cells also showed the association of cytoplasmic VDR with microtubules and the sensitivity of the fibrillar cytoplasmic VDR pattern to microtubule-disrupting drugs (34). Our current studies in living cells with BP-calcitriol validated these findings. Previously, transformed glucocorticoid receptors were shown to associate with cytoplasmic microtubules (39, 40). This fact suggests that mictrotubules are involved in the activation or intracellular traffic of other steroid receptors as well. Our results are in agreement with the model of steroid receptor activation suggested by W. B. Pratt (3), which includes a role for microtubules in steroid receptor translocation.
We found specific BP-calcitriol binding sites not only along microtubules but also in the ER and in the Golgi apparatus. The localization of BP-calcitriol in the ER is consistent with the presence of VDR in the nuclear envelope, since the outer nuclear membrane is morphologically and functionally continuous with the ER. Our colocalization studies suggest that VDR in the ER and nuclear envelope may be associated with calreticulin. Calreticulin is a major calcium-binding protein of the lumen of the ER. Calreticulin is also localized to the nuclear envelope and Golgi membranes in a variety of cells (41). Recently, calreticulin has been shown to bind to the DNA binding region of glucocorticoid, retinoic acid, and androgen receptors and to inhibit their ability to activate transcription (18, 19). Calreticulin was also shown to inhibit calcitriol-mediated gene transactivation (21, 22). Our results suggest that VDR could bind to calreticulin in the ER and possibly in the Golgi. Perhaps the chaperone function of calreticulin (42) plays a role in VDR processing. However, the possibility of other roles of calreticulin in VDR functions cannot be excluded either. We used thapsigargin, a Ca2+-ATPase inhibitor, to deplete inositol 1,4,5-triphosphate-sensitive Ca2+ stores (43). We found that thapsigargin inhibits BP-calcitriol binding to VDR but not to low affinity binding sites. This finding raises the possibility that the association of VDR with calreticulin plays a role in hormone binding to VDR. Another possibility is that calreticulin is involved in the rapid calcitriol-induced increase in intracellular calcium through the inositol 1,4,5-triphosphate receptor. The nongenomic actions of calcitriol are frequently associated with a putative plasma membrane receptor. However, we could not detect any significant amount of BP-calcitriol binding in the plasma membrane, not even at very short exposure times or at low temperatures. Based on our results, we propose that some of the nongenomic effects of calcitriol may be mediated through VDR interaction with calreticulin in the ER.
While we could not find previous reports on the presence of VDR in the ER, the presence of calcitriol binding sites in the Golgi has been indicated in a study by autoradiography in oocytes (44). Localization of VDR in the Golgi could be related to calcitriol-induced Ca2+ sequestration within the Golgi (45), or it could be related to the role of calcitriol in the regulation of cell proliferation.
It is not surprising to find low affinity binding sites appearing as diffuse signals in the cytoplasm and as punctate signals within the mitochondria. The mitochondrial binding is consistent with a calcitriol metabolism by hydroxylases localized within the mitochondria.
Probably the most interesting observation with BP-calcitriol was the recognition of discrete accumulation sites in the nucleus. High levels of BP-calcitriol accumulated in a reproducible nonrandom speckled pattern within the nuclei, excluding the nucleoli. To ensure that the nuclear signal was due to the association of BP-calcitriol to VDR, several control experiments were carried out. First, incubation with excess unlabeled calcitriol displaced most of the intranuclear staining, except some of the diffuse signal (Fig. 3). Second, staining of fibroblasts with mutations in the hormone binding region of the VDR gene showed that the BP-calcitriol signals in the nucleus and in the nuclear envelope were almost completely missing (Fig. 10). The physiological significance of the specific intranuclear distribution is further supported by the fact that mutations in the DNA-binding region of the VDR gene disrupted the intranuclear architecture of VDR binding. Further characterization of the nuclear regions where VDR is concentrated could lead to a better understanding of the mechanism of gene regulation by steroid hormones.
Using the recently developed fluorescent labeled calcitriol, BP-calcitriol, we visualized cytoplasmic and nuclear VDR locations in metabolically active cells. These studies resolved the long debated question about VDR partitioning and allowed real time studies on cytoplasm-to-nucleus translocation. Using confocal microscopy, we confirmed our earlier results on colocalization of microtubules with VDR in living cells and found VDR in the ER of living as well as fixed cells. Colocalization of VDR with calreticulin in the ER, the known ability of calreticulin to block transactivation through VDR, and the effect of thapsigargin to prevent hormone binding to VDR raise the possibility that proteins regulating ER calcium stores play an important role in VDR activation. Studies on cells from patients with hereditary resistance to calcitriol made possible the correlation of both defects in the VDR gene and the resulting selective loss of receptor function with defects in intracellular distributions. Most importantly, this method of fluorescent labeling revealed the intranuclear architecture of VDR binding on target genes. Further studies may lead to a better understanding of VDR interaction with the nuclear matrix and with other nuclear proteins in three dimensions. The ability to measure VDR expression, localization, and intracellular traffic in single cells will have many more applications, such as the diagnosis and treatment of calcitriol-dependent cancers and hyperproliferative diseases.
We thank Judy Drazba (NINDS, NIH), Kenneth Springs (NHLBI, NIH), and Nancy Dwyer for help with confocal microscopy, Stephen J. Marx (NIDDK, NIH) for the cells with vitamin D receptor mutations, Jonathan W. Yewdel for the gift of rhodamine-brefeldin A, and Kendra Shih for help in editing this manuscript.