Simple Fluorescence Labeling Method Enables Detection of Intracellular Distribution and Expression Level of Retinoid X Receptors

Abstract

There is no straightforward method to visualize the intracellular distribution of nuclear receptors, such as retinoid X receptors (RXRs), which are trafficked between the cytosol and nucleus. Here, in order to develop a simple fluorescence labeling method for RXRs, we designed and synthesized compound 4, consisting of an RXR-selective antagonist, CBTF-EE (2), linked via an ether bond to the fluorophore nitrobenzoxadiazole (NBD). Compound 4 is nonfluorescent, but the ether bond (-O-NBD) reacts with biothiols such as cysteine and homocysteine to generate a thioether (-S-NBD), followed by intramolecular Smiles rearrangement with an amino group such as that of lysine to form a fluorescent secondary amine (-NH-NBD) adjacent to the binding site. Fluorescence microscopy of intact or RXR-overexpressing MCF-7 cells after incubation with 4 enabled us to visualize RXR expression as well as nuclear transfer of RXR induced by the agonist bexarotene (1).

Bioimaging technologies are essential to elucidate the functions and localization of intracellular proteins, and fluorescence imaging of suitably labeled proteins is a representative approach. Many methods for fluorescently labeling a target protein with a fluorescent probe have been reported.¹⁻⁶ Low-molecular-weight fluorophores are of particular interest, because they are expected to have minimal effects on the activity of target proteins.

Nuclear receptors play key roles in various biological processes. For example, retinoid X receptors (RXRs) are involved in energy regulation in vivo, and their ligands are candidates for the treatment of type 2 diabetes,⁷ Alzheimer’s disease,⁸ and Parkinson’s disease.⁹ Therefore, in this work, we focused on RXRs.^10,11 RXRs exist in vivo as homodimers or heterodimers with other nuclear receptors (RARs, LXRs, PPARs, and others).^12,13 Some of these heterodimers exhibit “permissive regulation”, which means that the target genes of the partner receptor can be transcriptionally regulated by an RXR ligand alone.^14,15 Thus, RXRs are an attractive drug target. Nevertheless, their diverse physiological functions are not yet fully understood, and therefore further studies are needed to promote drug discovery research.

Labeling studies have shown that binding of the RXR agonist 9-cis retinoic acid (9cRA)¹⁶ or bexarotene (1)¹⁷ to unliganded RXR in the cytosol induces trafficking to the nucleus (Figure 1A). However, previous labeling methods have various disadvantages. For example, GFP labeling of intracellular RXR requires overexpression, which may alter RXR’s function or location, while immunofluorescence staining can only be performed on fixed cells.

Figure 1.

(A) Structures of RXR agonist bexarotene (1), RXR antagonist CBTF-EE (2), and fluorescent RXR antagonist CBTF-EE-NBD (3). (B) Molecular basis of the RXR labeling strategy employed in this research. (C) Molecular design of the proposed RXR labeling ligand.

Therefore, we aimed to develop a simple method to evaluate the expression level and intracellular distribution of RXR under physiological conditions by means of direct and selective fluorescence labeling of RXR.

To achieve RXR-selective direct fluorescence labeling, we focused on the direct chemical modification of proteins with the fluorophore nitrobenzoxadiazole (NBD), which emits fluorescence at 530 nm with a maximum excitation wavelength of 480 nm.¹⁸ Yamaguchi et al. labeled the glycoprotein avidin with NBD and stated that the target amino acid residue is Lys.¹⁹ Chen et al. also reported that a compound containing NBD with an ether bond reacts with biothiols such as cysteine (Cys) and homocysteine (Hcy).^20,21 This compound is nonfluorescent, but reaction of the ether bond (-O-NBD) with a thiol group (-SH) affords a thioether (-S-NBD), which undergoes intramolecular Smiles rearrangement with an amino group such as that of lysine (Lys) to generate a fluorescent secondary amine (-NH-NBD) (Figure 1B).

In order to target NBD to RXR, we focused on CBTF-EE (2) (Figure 1A), an RXR-selective antagonist reported by Watanabe et al.²² Takioku et al. have shown that CBTF-EE-NBD (3) (Figure 1A), in which NBD is linked to the end of the alkyl ether side chain of 2, retains RXR antagonist activity.²³ Taking all of the above results together, we designed CBTF-ONBD (4), in which NBD is linked to the RXR-targeting moiety by an ether bond in place of the amino group of 3 (Figure 1C).

CBTF-ONBD (4) was synthesized by using the previously reported 7 as a starting material (Scheme 1). Tosyl compound 6 was obtained in 90% yield by stirring diethylene glycol (5) and TsCl in dry CH₂Cl₂ in the presence of DMAP. Compounds 6 and 7 were reacted in DMF in the presence of K₂CO₃ to afford 8 in 83% yield. The carboxylic acid ester of 8 was hydrolyzed to afford 9, but direct reaction of 9 with NBD-F did not proceed, probably because the acidity of the carboxylic acid is higher than that of the hydroxyl group in 9. We therefore decided to block the carboxylic acid moiety with a protecting group. Reaction of 9 with MOMCl in dry CH₂Cl₂ in the presence of DIPEA gave 10 in 93% yield. Reaction of 10 with NBD-F in the presence of DIPEA in dry CH₂Cl₂ gave 11 in 37% yield. Although NBD-Cl might be preferable to NBD-F as a reagent, we found that no reaction took place under similar conditions. Deprotection of the MOM group of 11 and recrystallization of the product from CH₂Cl₂/n-hexane afforded the final compound 4 in a yield of 42%.

Scheme 1.

To examine the RXR labeling ability of 4, a mixture of 4 and RXRα-LBD was incubated at room temperature in the dark, and the fluorescence intensity was monitored (excitation 480 nm, emission 530 nm). The fluorescence emission increased with time and reached a maximum at 24 h. The time-to-half-maximum fluorescence intensity (t_1/2) was approximately 6 h (Figure 2A). Thus, 4 fluorescently labels RXRα-LBD. To confirm the RXR selectivity of 4, a similar experiment was run using PPARγ-LBD. The resulting fluorescence was low and did not increase time-dependently, supporting the RXR selectivity of 4 (Figure 2A). Next, to confirm that the fluorescence intensity change in the presence of RXR was due to the binding of 4 to the ligand binding pocket (LBP) in RXRα-LBD, we performed labeling after pretreatment with the RXR agonist bexarotene (1). RXRα-LBD was incubated first with 1 (10 equiv to 4), then 4 was added, and the fluorescence intensity after further incubation for 6 h was compared with that in the absence of pretreatment with 1 (Figure 2B). Pretreatment with 1 for 2 h or even just 1 h significantly decreased the fluorescence intensity, indicating that 1 competes with 4 (Figure S1). We confirmed that 4 showed no increase in fluorescence intensity in the absence of RXRα-LBD (Figure S2). These results indicate that 4 binds to the ligand binding pocket. An RI binding assay using [³H]9-cis retinoic acid ([³H]9cRA) yielded a K_i value of 18.09 nM for 4, compared with a value of 201 nM²³ for 1 (Figure 2C).

Figure 2.

Results of fluorescence intensity measurement (λ_ex, max = 480 nm, λ_em, max = 530 nm) and RI binding assay. (A) Time courses of change in fluorescence intensity during incubation of 4 (0.5 μM) with RXRα-LBD (0.5 μM, black) or PPARγ-LBD (0.5 μM, orange) in Hepes buffer (10 mM, pH 7.9, 1% ACN) at room temperature in the dark. The inset shows an expanded view up to 6 h for PPARγ-LBD. (B) Comparison of fluorescence intensity after incubation of RXRα-LBD with 4 for 6 h at room temperature in the dark with or without pretreatment with 1. Mean ± SD (N = 3), t test, ****; p < 0.0001. A mixture of 1 (5.0 μM) and RXRα-LBD (0.5 μM) in Hepes buffer was incubated at room temperature in the dark for 16 h, and then 4 (0.5 μM) was added. (C) RI binding assay using [³H]9cRA. K_i = 18.09 nM. K_i was calculated by substituting the IC₅₀ of the competition curve of [³H]9cRA (K_d = 37 nM) and 4 into the Cheng–Prusoff equation.

RXRα-LBD can form a tetramer in the absence of a ligand,²⁴ and to inhibit tetramer formation, reducing agents such as dithiothreitol (DTT) are used. However, DTT cannot be used with 4, because 4 reacts with thiol. Furthermore, when tris(2-carboxyethyl)phosphine (TCEP), which does not contain a thiol group, was used, no fluorescence intensity change was observed, suggesting that this reducing agent also cannot be employed with 4 (Figure S3A,B). Then we conducted native PAGE, where the upper bands correspond to tetramers and the lower bands correspond to dimers (Figure S3C).²⁵ A comparison of lanes 1 and 2 indicates that TCEP has little effect on the formation of RXR-LBD tetramers, while a comparison of lanes 1 and 4 indicates that the tetramer of 4 is increased. It has been reported that 2 forms a tetramer when bound with RXR-LBD.²² Therefore, it seems likely that 4, which has the same skeleton as 2, has similar properties.

The binding of NBD to a protein is expected to increase the molecular weight by 163. This molecular weight change is too small to detect by SDS-PAGE (Figure S4). However, MALDI-TOF-MS (MALDI-8030, Shimadzu, Kyoto, Japan) using sinapinic acid as the matrix showed a peak corresponding to a molecular weight increase of about 170 from the peak of RXRα-LBD at 27,109 (Figure 3). This result suggests that RXRα-LBD covalently binds a single NBD molecule. Other matrixes such as α-cyano-4-hydroxycinnamic acid were not effective (Figure S5). NBD bound to amino groups in the peptide chain has a molar absorption coefficient of 26,000 M^–1·cm^–1 at 475 nm.²⁶ Based on this molar absorption coefficient, the labeling efficiency of RXRα-LBD and 4 at a concentration ratio of 1:1 was calculated from the absorbance as 25% (Figure S6).

Figure 3.

MALDI-TOF-MS data for RXRα-LBD alone or labeled with 4. RXRα-LBD (10 μM) without labeling is shown in blue, and RXRα-LBD (10 μM) + 4 (10 μM) is shown in red. Conditions: MALDI-8030, linear mode, positive-ion mode, pulsed extraction 27000. Matrix: sinapinic acid (SA) 20 mg/mL in 50% acetonitrile. Preparation: on-target mix, 1:1 (0.5 μL each).

We next examined the binding site of 4 by means of docking simulation using AutoDock Vina. The RXRα-LBD cocrystal structure of LG100754 (Figure 4B), an RXR antagonist (PDB ID: 3A9E),²⁷ was used as the receptor structure. We found that 4 binds to RXRα-LBD in a very similar conformation to LG 100754 (Figure 4A). Furthermore, Cys and Lys residues are located close to the NBD moiety (Figure 4C). This is consistent with the putative fluorescence labeling mechanism. Comparison of the MS data of trypsin-treated peptides labeled with 4 or not labeled suggested that the NBD-labeled position is Lys 436 of RXRα-LBD (Tables S2 and S3).

Figure 4.

Docking structures obtained by simulation using AutoDock Vina. (A) Conformations of RXRα-LBD (3A9E) and ligands LG100754 (yellow) and 4 (green). (B) Molecular structure of LG100754. (C) Amino acids near the binding site. Cys is shown in yellow, and Lys is shown in red.

NBD labeling with 4 generates the OH derivative 9 (Scheme 1) as a residue. Our RXR binding assay method²³ using the fluorophore BODIPY (λ_ex, max = 485 nm, λ_em, max = 535 nm) revealed that 9 shows a lower binding affinity than 1 (Figure S7A). Since 4 reacts with RXR-LBD and produces fluorescent NBD, whose fluorescence emission overlaps with that of BODIPY, the binding affinity of 4 could not be measured. Compound 9 showed RXR antagonist activity in a reporter assay (Figure S7B).

Next, we examined whether 4 can fluorescently label RXRs in living cells. We selected MCF-7 cells (cultured human breast cancer cells) because of their high endogenous expression of RXR.²⁸ RXR has been reported to translocate into the nucleus in the presence of agonists.^16,17 Thus, after labeling RXR-LBD with 4, 1 was added and the nuclear translocation of the fluorescent label was evaluated with an all-in-one fluorescence microscope (BZ-X810, KEYENCE, Osaka, Japan). Quantitative evaluation was performed using an image cytometer (BZ-H4XI). Transfer of the label to the nucleus was observed, as shown in Figure 5A, where the Hoechst-stained nuclei appear in blue, and fluorescence originating from NBD is shown in green. However, translocation into the nucleus in the absence of 1 (Figure S8) or even in its presence (Figure 5A) was not complete, possibly due to the effect of 9 formed during the labeling reaction.

Figure 5.

Fluorescence photographs of MCF-7 cells labeled with 4. NBD fluorescence derived from 4 appears in green, and Hoechst fluorescence in the nucleus appears in blue. (A) Incubated overnight with 4 (1.0 μM), for 2 h with 1 (10 μM), and for 30 min with Hoechst (0.4 μM). Objective lens ×20. (B) Incubated with 4 (1.0 μM) overnight. Objective lens ×4. (C) Plot of the numbers of labeled cells detected in B. Mean ± SD (N = 6), t test, ****; p < 0.0001. TF–, untransfected cells; TF+, cells transfected with human RXRα.

We next transfected MCF-7 cells with human RXRα. The transfected cells showed significantly greater NBD-derived fluorescence than untransfected cells, confirming that RXR was selectively labeled within the cell (Figure 5A, B, and C). In addition, nuclear translocation of the fluorescent label was confirmed. As noted above, residual fluorescence outside the nucleus was observed after treatment with 1, possibly due to the presence of 9. Alternatively, the nucleus may not be able to accommodate all of the overexpressed RXR.

In this work, we designed and synthesized compound 4, consisting of an RXR-selective antagonist, CBTF-EE (2), linked via an ether bond to the fluorophore NBD, in order to develop a simple fluorescence labeling method for RXRs. Compound 4 reacts specifically at the ligand binding site of RXRα-LBD and subsequently forms a fluorescent secondary amine (-NH-NBD). Fluorescence microscopy of intact or RXR-overexpressing MCF-7 cells incubated with 4 in assay buffer successfully visualized both RXR expression and nuclear transfer of RXR induced by the agonist bexarotene (1). This simple labeling technique can be performed in an extremely short time compared to methods using genetic manipulation or antibodies and may prove useful for studying the role of RXR agonists as candidate drugs to treat various diseases, as well as probing permissive heterodimer pathways involving RXR. A similar strategy using a specific targeting moiety to deliver NBD may also be applicable to visualize the intracellular distributions of other receptors.

Glossary

Abbreviations

ACN

acetonitrile

BODIPY

boron-dipyrromethene

DIPEA

N,N-diisopropylethylamine

DMAP

4-dimethylaminopyridine

DMF

N,N-dimethylformamide

DTT

dithiothreitol

Em

emission

EtOAc

ethyl acetate

Ex

excitation

GFP

green fluorescent protein

hRXR

human RXR

LBD

ligand-binding domain

LBP

ligand-binding pocket

LC-MS/MS

liquid chromatography with tandem mass spectrometry

LXR

liver X receptor

MALDI-TOF-MS

matrix assisted laser desorption/ionization-time-of-flight mass spectrometry

MOMCl

chloromethyl methyl ether

NBD

nitrobenzoxadiazole

9cRA

9-cis retinoic acid

PAGE

polyacrylamide gel electrophoresis

PPAR

peroxisome proliferator activated receptor

TsCl

p-toluenesulfonyl chloride RAR, retinoic acid receptor

RI

radioisotope

RXR

retinoid X receptor

SD

standard deviation

SDS

sodium dodecyl sulfate

TCEP

tris(2-carboxyethyl)phosphine

TF

transfection

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.4c00033.

• Synthetic and NMR data for compounds, absorption and fluorescence spectra data, and binding data (PDF)

Author Contributions

H.K. conceived and designed the project. Y.Tanaka synthesized compounds. M.K. and S.N. produced hRXRα-LBD. M.F. performed reporter gene assays. Y.Tanaka and Y.Takamura performed the RI binding assay. Y.Tanaka and Y.Takamura analyzed the UV and fluorescence spectra. All authors analyzed and discussed the data. The manuscript was written by Y.Tanaka and H.K.

This work was partially supported by Okayama Foundation for Science and Technology (to H.K.) and The Tokyo Biochemical Research Foundation (TBRF) (to H.K.).

The authors declare no competing financial interest.