Bexarotene via CBP/p300 Induces Suppression of NF-κB–Dependent Cell Growth and Invasion in Thyroid Cancer Free

Thyroid cancers are the most common malignancy of endocrine organs (1). Well-differentiated thyroid carcinomas (DTC) derived from follicular cells account for around 80% of thyroid cancers and include papillary thyroid cancer (PTC) and follicular thyroid cancer (FTC), respectively, which are differentiated on the basis of histologic parameters. Most of these cancers have an excellent prognosis and can be effectively managed by total thyroidectomy, ablative doses of radioiodine, and suppressive treatment. The follow-up is based on measuring serum thyroglobulin (TG) and imaging with radioiodine (RI) therapeutic scans. It has been estimated that about 20% of these patients will develop a local or distant recurrence. In some cases, TG and RI scan are discordant, suggesting the presence of thyroid cancer metastases which have lost the ability to trap RI (2). This is challenging for thyroid cancer management, as anatomical localization of recurrences cannot be assessed and treatment with RI therapeutic doses is not effective, predicting a poorer outcome.

Signaling through growth factors and their receptors and defects in regulation of cell-cycle control elements are considered essential for cancer cell proliferation. Recent advances have improved the understanding of the thyroid oncogenesis. Inappropriate activation of the mitogen-activated protein kinase pathway effectors may lead to tumor initiation and transformation. Indeed, mutations or rearrangements of tyrosine kinase receptor (RET and NTRK1), BRAF, and Ras are found in most of thyroid cancers (3). Cancer progression is associated with a loss of the differentiated characteristics that define mature follicular thyroid cells with the full capacity for normal physiologic function such as iodine uptake. Iodine is transported across the basolateral plasma cell membrane of thyrocytes via the NA⁺/I⁻ (sodium iodine) symporter (NIS; ref. 4). NIS and other proteins linked to the synthesis of thyroid hormones, such as TG, thyroid peroxydase (TPO), and type I iodothyronine 5′-deiodinase (DIO1) have been reported to be poorly expressed in thyroid cancer patients (5). Reduced expression of these differentiation proteins results from Ras mutations and alterations of transcription factors involved in thyroid differentiation such as loss of TTF1 expression or chromosomal translocation of paired box 8 (PAX8) with peroxisome proliferator-activated receptor gamma (PPARγ; refs. 6, 7).

Retinoids, a group of structural and functional derivatives of vitamin A are known to regulate a large number of essential biological processes, such as cell growth, differentiation, and death. The effects are mainly mediated by 2 types of nuclear receptors—retinoic acid (RA) receptors (RARs) and retinoid X receptors (RXRs)—which act as ligand-dependent transcription factors. By qualitative mRNA detection assays, expression of retinoid receptors (RARA, RARB, RARG, and RXRA and RXRB) has been detected in normal thyrocytes, suggesting that absence of a tissue-specific retinoid receptor may participate in differentiation arrest and thyroid oncogenesis (8). Reduced expression of RARβ has indeed been observed in human thyroid carcinomas (9). RA therapy in malignant cells is based on functional alternative RA signaling pathways which, upon pharmacologic concentrations of a given retinoid, restore control of cell death, differentiation, and proliferation. Various mechanisms are involved including the transcriptional control of the tissue-specific retinoid receptor gene via endogenous receptors. We have shown the upregulation of RARβ in neuroblastoma and glioblastoma by various natural and synthetic retinoids (10), and the transcriptional control of the normal RARA gene expression by all-trans RA (ATRA) treatment in acute promyelocytic leukemia cells (11). RA efficacy in patients with thyroid cancer is supported by in vitro data, showing that retinoids induce upregulation of RARβ and differentiation of cells, accompanied by the upregulation of NIS and TG expression and DIO1 activity (12, 13). Treatment with 13-cis RA was shown to restore RI uptake in thyroid cancer patients in clinical trials (14–16). However, only one-third of the patients respond, and further trials have failed to confirm 13-cis RA efficacy in these patients (17, 18). A clinical study showed that treatment with bexarotene, a RXR agonist can induce a radioiodine uptake by metastases in some patients (19, 20). These data prompted for a better understanding of RA signaling pathways and resistance in thyroid cancer cells.

We have previously reported that lack of RA responsiveness in a FTC cell line was attributable to an altered histone acetylation pattern and could be relieved by combination of ATRA and trichostatin A, a histone deacetylase inhibitor, underscoring the ongoing clinical trials based on differentiation and transcriptional therapy combination (21). To investigate the role of RXR agonist in these patients, we took advantage of the RA-resistant thyroid cancer cell line, FTC238, derived from the metastasis of the patient whose cells, at diagnosis, allowed the establishment of the RA-sensitive FTC133 cell line (22). In this study, we show that ligands of RARs or RXRs exert different effects in thyroid cancer cells through either differentiation or inhibition of growth and invasion and may thus provide different therapeutic approaches in DTC patients. Drug effects were associated with restoration of RARB and RXRG levels. Interestingly, bexarotene inhibits the transactivation potential of NF-κB through the involvement of nuclear coactivator p300, without interfering with either nuclear translocation or binding of p65 to NF-κB–responsive elements. This study highlights in a given cancer dual mechanisms by which retinoids may target cell tumorigenicity, not only via targets of retinoids such as RARs and RXRs, as expected, but also via targets of the NF-κB pathway.

Bexarotene (LGD1069) was kindly provided by Ligand Pharmaceuticals. ATRA and 13-cis RA were purchased from Sigma-Aldrich. The powdered retinoid was dissolved in ethanol at an initial stock concentration of 10⁻² mol/L, stored protected from light at −80°C, and further diluted before use. JSH-23 was purchased from Santa Cruz Biotechnology and was dissolved in dimethyl sulfoxide at an initial stock concentration of 10⁻² mol/L.

Human FTC cell lines FTC133 and FTC238 were kindly provided by C. Schmutzler (Dusseldorf, Germany). Human breast cancer cell line MCF7 and FTC cell lines were propagated in Dulbecco's modified Eagle's medium/Ham's F12 (1:1; DMEM-F12; Invitrogen) supplemented with 10% (v/v) or 5% (v/v) heat-inactivated FBS and penicillin (100 units/mL), streptomycin (100 ng/mL), and l-glutamine (2 mmol/L). Human acute promyelocytic leukemia NB4 was propagated in RPMI-1640 medium supplemented with 10% (v/v) FBS and penicillin (100 units/mL) and streptomycin (100 ng/mL) and l-glutamine (2 mmol/L). All cultures were incubated at 37°C in a 5% CO₂ humidified atmosphere. For retinoids treatment, cells were split the day before treatment and cultured with serum-containing medium. On the next day, the medium was replaced with retinoid-containing or retinoid-free medium with serum.

Cells were plated in triplicate at 500 cells for FTC238 cells, at 1,000 cells for FTC133 and MCF7 cells and at 10,000 cells for NB4 cells per well in 96-well plates. Cell viability was measured by the Cell Titer 96 Aqueous One Solution Assay (Promega) according to the manufacturer's instructions.

The invasiveness of FTC cells was tested using the QCM Collagen Cell Invasion Assay (Chemicon; Millipore). Cells were seeded in serum-free medium containing bexarotene to the upper collagen-coated transwell. The lower chambers were filled with culture medium containing 10% FBS as a chemoattractant. After a 72-hour incubation period, the number of cells that migrated through the filters into the lower chamber was evaluated by a colorimetric method. Briefly invaded cells on the bottom of the insert membrane are incubated with Cell Stain Solution, then subsequently extracted and detected on a standard Microplate reader (560 nm).

For cell-cycle analysis, FTC238 cells were permeabilized with 0.1% triton in 1× PBS and incubated with 0.5 mL of a 50 μg/mL propidium iodide (PI) solution containing 20 U/mL RNAse-A for 30 minutes (Sigma-Aldrich). For apoptosis, cells were double stained with fluorescein isothiocyanate–conjugated Annexin V and PI for 15 minutes at 4°C in a calcium-enriched binding buffer (Beckman Coulter). All analysis was done on a FACS Calibur Flow Cytometer with Cell Quest software (Becton Dickinson, France).

Total RNA was extracted as described by the manufacturer (RNA-PLUS; Qbiogene). Reverse transcription was done on 1 μg of total RNA using the High Capacity RT Kit (Applied Biosystems). cDNAs were quantified using a fluorescence-based real-time detection method (ABI PRISM 7700 Sequence Detection System; Applied Biosystems) with the TaqMan Universal Master Mix. The primers and probe sequences for RARs and RXRs are provided as Supplementary information (Table S1). Others primers and probe were provided by Applied Biosystems (Gene Expression Assay; Courtaboeuf). For each sample, parallel PCR reactions were done for each gene of interest and the porphobilinogene deaminase (PBGD) reference gene to normalize for input cDNA. To quantify the gene expression profile, we used the comparative threshold cycle (C_t) method.

Nuclear extracts were obtained as previously described (23) and were quantified by the BCA protein assay (Pierce). Proteins were run in SDS-PAGE gels, transferred to nitrocellulose membranes. Membranes were blocked in 5% nonfat dry milk in 1× PBS at room temperature. The blots were incubated overnight at 4°C with the primary antibody anti-p65 (Upstate; Millipore). After washes with 1× PBS with 0.1% Tween 20, antigen–antibody complexes were detected by means of peroxidase-conjugated secondary antibody and an enhanced fluorochemiluminescence system (ECL-plus; Amerscham Biosciences). Equivalent loading of lanes was controlled by an anti-actin antibody (Sigma-Aldrich). Band intensities were quantified using a ChemiDoc XRS and Quantity One version 4.4.0 software.

siRNA against RXRG gene (siRXRG) and nontarget control siRNA (siCRT) were purchased from Santa Cruz Biotechnology. siRNAs were transfected into cells at a final concentration of 50 nmol/L by using the transfection reagent Lipofectamine RNAiMax (Life Technologies; Invitrogen) according to the manufacturer's instructions. Analyses were realized after 48 hours of transfection.

For cells transfections, FTC238 cells were seeded in 24-wells plates. After reaching about 80% confluence, the cells were transfected with luciferase gene in the reporter plasmid controlled by a synthetic promoter that contains direct repeats of the transcription recognition sequences for NF-κB (pNF-κB-luc; Stratagene) using Lipofectamine (Lipofectamine Reagent Plus; Invitrogen) according to the manufacturer's instructions. Cotransfection with a β-galactosidase plasmid (pSV-β-gal control vector; Promega) was used to normalize luciferase activity. After transfection, the medium was replaced by fresh normal growth medium with retinoid and the cells were incubated for 24 hours. Luciferase (Luciferase assay system; Promega) and β-galactosidase (β-Gal Reporter Gene Assay; Promega) assay were done by a standard procedure.

Assay for NF-κB binding was done with the colorimetric Universal EZ-Transcription Factor Assay (Upstate; Millipore) according to the manufacturer's instructions. Briefly, the capture probe, a double strand biotinylated oligonucleotide containing a NF-κB consensus binding site, is mixed with nuclear extract in the streptavidin-coated 96-well plate. Bound NF-κB was detected by incubation of samples with primary antibody against the p65 subunit provided with the kit. A horseradish peroxidase (HRP)-conjugated secondary antibody is then used for detection. Quantification of binding levels was evaluated by optical intensities at 450 nm with a Microplate reader. A specific unlabeled competitor oligonucleotide was included to ensure that the complex is binding the κB site in a sequence-specific manner.

Cells were fixed with 1% formaldehyde in media for 10 minutes at 37°C to cross-link DNA with proteins. Then, the cells were washed twice with PBS1X containing proteases inhibitors. Cells were resuspended in lysis buffer and sonicated (Bioruptor; Diagenode) to obtain DNA fragments to 200 to 1,000 bp. After preclearing with salmon sperm DNA/protein G agarose beads (Upstate; Millipore), the samples underwent immunoprecipitation with antibodies specific against p65, p300, acetylated histone H3, acetylated histone H4 (Upstate; Millipore), or rabbit/mouse IgG at 4°C overnight. DNA–protein complexes were sequentially washed, eluted (1% SDS, 50 mmol/L NaHCO₃), and cross-link reversed by heating at 65°C overnight. DNA was recovered by proteinase K digestion, phenol-chloroforme purification, and ethanol precipitation. DNA was amplified by real-time PCR using the Power master mix SYBR green on a ABI PRISM 7700 Sequence Detection System (Applied Biosystems). Primers sets were designed to overlap the NF-κB–binding sites of the CYR61 promoter and the RARE site of the RARB2 promoter. Specificity of the assay was tested by amplification of an intergenic region (between the GAPDH gene and the chromosome condensation–related SMC-associated protein gene) not bound by transcription factors (negative controls). The primers sequences are provided as Supplementary Information (Table S1). Data for each immunoprecipitated sample were normalized to its corresponding input (total chromatin fraction) and expressed as fold enrichment above background (IgG controls) or as fold change compared with untreated sample using the comparative C_t method.

All data were obtained from at least 3 independent experiments conducted in triplicate, and the results are presented as mean and SEM. To show statistical significance, a 2-tailed Student t test was used (* denotes a P value of at least < 0.05).

In parallel to these in vitro studies, 17 patients (8 male and 9 female, median age at diagnostic 50 years), with progressive metastatic DTC and decreased or absent RI uptake, were included in a prospective open clinical trial (CRIC99340). These patients (9 PTC and 8 FTC including 2 Hurthle and 1 insular tumor) had previously received multiple treatments: high cumulative RI dose (median 13.7 GBq), surgery of metastatic sites (#8), External Beam Therapy (#3), cimentoplasty (#1), and bisphosphonate (#1). Group 1 (n = 13) had pulmonary, bone and/or mediastinal macrometastases and group 2 (n = 4) had lung and/or mediastinal infracentimetric metastases. After informed consent, they all received 13-cis RA for 8 weeks (1 mg/kg/d), followed by RI treatment (3.7 GBq) after 4 weeks of LT4 withdrawal. Treatment was repeated when reinduction of RI uptake on posttherapy scan was obtained. In the absence of RI uptake, a further regimen with bexarotene (300 mg/m²/d) for 8 weeks was undertaken, then followed by RI treatment (3.7 GBq) after rhTSH stimulation, to avoid associated central hypothyroidism (24). To become eligible to bexarotene, patients were required to have acceptable organ function defined as follows: adequate hematologic function, defined as WBC ≥ 3,000/μL, absolute neutrophil count ≥ 1,500/μL, and platelet count ≥ 100,000/μL; normal coagulation parameters; bilirubin ≤ 1.5 times the upper limit of normal (ULN); AST/ALT ≤ 2.5 × ULN; and serum creatinine ≤ 2.5 × ULN (or creatinine clearance > 40 mL/min). In addition, only patients who had a normal baseline fasting triglyceride level were allowed to enter this study; triglycerides could be normalized before study entry with use of an antilipemic agent. Response was immediately judged on RI reinduction on posttherapy scan, and 4 to 6 months later on TG levels and tumor size by conventional imaging techniques according to RECIST criteria (25).

FTC133 and FTC238 cells provide a well-established model for the study of differential thyroid cancer retinoid sensitivity based on different criteria, such as thyroid-specific functions (DIO1 and NIS expression), cell–cell or cell–matrix interaction (intercellular adhesion molecule-1 and E-cadherin), differentiation markers (alkaline phosphatase, CD97 and TG), cell growth, and tumorigenicity (26). To characterize and compare the cell lines used for this study, we quantitatively assessed the expression of genes associated with thyrocyte differentiation [PAX8, TPO, thyroid stimulating hormone receptor (TSHR), DIO1, NIS and TG; ref. 5]. Increased expression of these differentiation-associated genes was observed in FTC133 cells after 24 hours of treatment with ATRA 10⁻⁶ mol/L. In contrast, no difference was observed in the FTC238 cell line (Fig. 1A). Thus, FTC238 remains refractory to ATRA differentiation induction contrary to the FTC133 cell line. After treatment of FTC238 cells with 13-cis RA or bexarotene, currently used retinoids in thyroid cancer patients, no upregulation of thyroid differentiation markers was noted (Fig. 1A), correlating with absence of morphologic aspects of differentiation as observed in FTC133 cells, which harbor a more homogenous and spindle-shaped aspects after ATRA or 13-cis RA treatment (Supplementary Fig. S1).

We have previously shown that FTC133 and FTC238 cell lines present, respectively, low and null expression of RARB mRNA levels (21). In this study, we examined the gene expression levels of the different RARs and RXRs by quantitative reverse transcriptase PCR (RT-PCR). RARA, RARG, RXRA, and RXRB are expressed in both cells lines, whereas RARB and RXRG are barely detectable. RARA mRNA is the major retinoid receptor mRNA detected in FTC133 cells, whereas RXRA is predominant in FTC238 cells. Globally, compared with FTC133 cells, FTC238 cells have a lower expression of RARs and a higher expression of RXRs (Fig. 1B).

In cancer cells with altered RA receptors, differentiation induced by pharmacologic concentrations of RA is associated with the upregulation of receptor expression via activation of endogenous normal receptors by the retinoids (27). Upregulation of RARs and RXRs gene expression was monitored in the 2 cells lines after incubation with ATRA 10⁻⁶ mol/L. ATRA induced a significant RARB and RXRG mRNA expression in only the RA differentiation–sensitive FTC133 cell line but not in RA-resistant FTC238 cells (Fig. 1C). Thus ATRA differentiation refractoriness of the FTC238 cell line may be related to absence of RARB and RXRG expression and its induction by ATRA.

RARs and RXRs gene expression was monitored in both cell lines after incubation with either 13-cis RA or bexarotene. These retinoids were not more effective than ATRA to induce RARs and RXRs expression in FTC133 cell line (data not shown). Interestingly, in the ATRA-resistant FTC238 cells, both retinoids allowed a significant recovery of RARB and RXRG expression levels (Fig. 2A). This upregulation of retinoid receptors by bexarotene could not be linked to induction of differentiation-associated genes (Fig. 1A) but to a decrease of cell growth in a dose-dependent manner (Fig. 2B). Inhibition of cell growth by bexarotene in FTC238 cells was not due to apoptosis, as no increase in the percentage of Annexin V–positive cells was noted after incubation with bexarotene (Fig. 2C) but, rather, to an inhibition of proliferation. Cell population in the G₁ phase increased from 75.4% ± 1.3% to 89.8% ± 5.5% upon bexarotene (P = 0.002). This increase was accompanied by a decrease of cell population in the S/G₂/M phase (13.2% ± 4.3% compared with 23.4% ± 4.2%, P = 0.01; Fig. 2D). The small increase in subG₁ population was not significant confirming the low level of Annexin V–positive cells after bexarotene. As the FTC238 cell line is derived from a pulmonary metastasis, we also investigated the invasive activity of these cells by their ability to migrate through a collagen layer. A 30% decrease of invasion was observed upon bexarotene treatment (P = 0.02; Fig. 2E). Thus, bexarotene restores RARB and RXRG expression in FTC238 cell line and inhibits cell proliferation and invasion but does not increase differentiation.

In a clinical proof-of-concept open clinical trial, 17 DTC patients with progressive metastatic DTC and decreased or absent RI uptake were first treated by 13-cis RA for 8 weeks (1 mg/kg/d), followed by RI treatment (3.7 GBq) after 4 weeks of LT4 withdrawal. Characteristics and clinical response of patients are presented in Supplementary Table S2. Clinical response was assessed on RI uptake and RECIST criteria (25). After one course of 13-cis RA, one patient had a partial response (PR), 7 patients a stable disease, and 9 patients a progressive disease. Treatment with 13-cis RA induced RI uptake in metastases of 6 of 17 (35%) patients, irrespective of the clinical response, though the patient with a PR did have also induced RI uptake (patient no. 1). All responding patients had macroscopic metastatic tumors (group 1). Of these 6 patients, 3 received a second 13-cis RA treatment, also followed by a new RI reinduction. Responses remained the same. Of note 2 of these patients had a tumor size reduction of more than 50% (patients no. 1 and 3). Bexarotene treatment was undertaken in 5 patients who did not respond to the first course 13-cis RA and were eligible for bexarotene. Two patients (patients no. 8 and 9) presented an increase in RI uptake, suggesting that resistance to 13-cis RA could be bypassed in some metastases by bexarotene as illustrated in Fig. 3. These 2 patients had macroscopic metastatic tumors (group 1). The RI uptake responses noted in group 1 with macroscopic metastases may well be explained by detection limitations. Tumor regression or stabilization was noted by conventional imaging techniques in 3 other patients treated by bexarotene but with no increased RI uptake (patients no. 10, 14, and 15). Thus, tumor response was not always correlated to RI uptake correlating with the in vitro data. Furthermore, metastasis heterogeneity led to differential metastases responses in a given patient.

To further analyze the mechanisms by which bexarotene inhibits FTC238 cell growth and invasion, we focused our studies on the NF-κB pathway. NF-κB, a transcription factor known to be a key regulator of genes involved in the control of proliferation, apoptosis, and invasion, is constitutively activated in most of cancers (28). Moreover, members of the nuclear receptor family have been shown to interfere with NF-κB signaling (29). We thus hypothesized that the activation of the RXR pathway by bexarotene could interfere with NF-κB and analyzed the expression of genes known to be downstream effectors of NF-κB (Fig. 4A). Bexarotene treatment decreased the expression of MYC, CCND1, CYR61, and CXCL1 genes involved in proliferation and invasion but not the expression of BCL2 and XIAP genes, correlating with the inhibition of proliferation and invasion and absence of apoptosis observed upon bexarotene treatment (Fig. 4B). This mechanism requires the presence of RXRG expression restored by bexarotene, as the inhibition of NF-κB target genes expression by bexarotene is not observed following downregulation of RXRG by siRNA (Fig. 4C). To further analyze NF-κB activity in FTC238 cells, we transiently transfected FTC238 cells with a luciferase reporter plasmid under the control of a synthetic promoter that contains direct repeats of the transcription recognition sequences for NF-κB. Bexarotene repressed both constitutive basal and phorbol myristate acetate (PMA)-induced NF-κB transactivation activity (Fig. 4D). Thus bexarotene, through induction of RXRG, represses transcriptional activity and target gene expression of NF-κB and results in loss of cell growth and invasion. A decrease of proliferation may also be obtained on these cells by inhibiting NF-κB pathway by a known chemical inhibitor JSH-23, confirming the biological effect of activated NF-κB pathway in these cells (Fig. 4E).

Interestingly, bexarotene targeting of the NF-κB pathway may also be observed in other thyroid cancer cells (FTC133) and in acute myeloid cells (NB4), albeit in different degrees but not in breast cancer cells (MCF7). Inhibition of cell proliferation was essentially achieved in the NB4 cell line (Supplementary Fig. S2A), whereas a decrease of invasion capacity was observed in FTC133 cell line (Supplementary Fig. S2B). Reduced expression of NF-κB proliferation and invasion target genes was noted in a cell-context manner (Supplementary Fig. S2C).

To explore the mechanism(s) whereby bexarotene induces NF-κB repression in FTC238 cells, we examined the effect of bexarotene on different steps of the NF-κB pathway (Fig. 5A). First, we analyzed the translocation of the active p65 subunit of NF-κB to the nucleus compartment. No significant change in p65 levels in nuclear extracts was noted after treatment, suggesting that bexarotene mediated its effect on NF-κB repression at the transcriptional level (Fig. 5B). However, p65 binding on a synthetic NF-κB–responsive element was not modified in nuclear extracts of treated cells in the presence or absence of both PMA (Fig. 5C). Furthermore, chromatin immunoprecipitation (ChIP) assay using anti-p65 antibody on NF-κB–binding sites of CYR61 promoter gene confirmed these results (Fig. 5D). Thus these data suggest that bexarotene represses transactivation of NF-κB genes without interfering with either the translocation or the binding of p65 to NF-κB–responsive elements.

Interestingly, by ChIP assay, the coactivator p300 was detected on the p65-binding site of CYR61 promoter gene (Fig. 5E). As expected, bexarotene induced p300 recruitment at the RARE with increased level of histones H3 and H4 acetylation (Fig. 5F). However, at p65-binding sites, bexarotene released p300 (Fig. 5E) and decreased levels of histone acetylation (Fig. 5F) resulting in reduced promoter permissiveness. Thus, in thyroid cancer cells, bexarotene represses NF-κB target gene expression through downregulation of the transactivation potential of p65 on its binding site via the bexarotene–RXR–dependent release of the nuclear coactivator p300 and resulting in histone deacetylation.

RA therapy has been attempted in thyroid cancer patients essentially to restore iodine uptake for treatment of recurrent metastases by radioactive iodine. We confirmed similar response rate on RI uptake (35%) obtained by others groups (30 to 40% refs. 14–16). Increased RI uptake in thyroid cancer may be explained by NIS expression. Induction of NIS expression by RA in breast cancer cells was reported to be mediated by genomic and/or nongenomic mechanism requiring RARs and RXRs expression (30, 31). Because bexarotene increases RARB and RXRG expression in FTC238 cells without induction of NIS, it seems that thyroid cancer cells could have other regulation mechanisms of NIS expression.

Using FTC cell lines with differential response to retinoids, we have previously identified that one of the mechanism inherent to RA resistance may lie in the nonpermissive status of the RARB gene promoter, relieved by histone deacetylase inhibitors (21). This study suggests that along with RARβ, RXRγ may equally be involved in thyroid oncogenesis, as previously suggested (32) and represents a novel biomarker of RA therapy. We identified, bexarotene, the first highly selective RXR ligand to be currently studied in clinical trial (33–35) as a growth inhibitor of the RA differentiation–resistant FTC238 cell line, to an extent not achieved with ATRA. In cutaneous lymphoma cell lines, bexarotene induces arrest of cell cycle in G₁ phase and apoptosis with activation of caspase-3 and PARP cleavage, as well as downregulation of survivin (36). A proof-of-principle clinical trial with bexarotene in patients with non–small cell lung cancer showed regression of biomarkers such as cyclin D1, cyclin D3 and epidermal growth factor receptor (37). In this study, we show that bexarotene downregulates MYC and CCND1 expression in the ATRA-resistant FTC238 cells, extending the notion that rexinoids can bypass absence of RARβ expression. Interestingly, in these cells, bexarotene upregulates RARB and RXRG which ATRA treatment fails to do. It has previously been reported, albeit in another histologic type, anaplastic thyroid cancers that the presence of RARβ and RXRγ was a prerequisite for growth inhibition by RXR ligands (38). Thus in parallel with ATRA differentiation efficacy, which requires upregulation or restoration of RARβ and RXRγ levels, bexarotene growth inhibition may also require the upregulation of both receptors. This growth inhibition cannot be attributed to apoptosis but rather to cell cycle arrest in G₁ phase in accordance with the downregulation of CCND1 expression. In line with these in vitro studies, our clinical results show that bexarotene treatment of thyroid cancer patients who failed to respond to 13-cis RA therapy leads, in some patients, not only to radioiodine uptake but also to a reduction in size and number of metastases without apparent differentiation and radioiodine uptake.

Past and current concept of metastases initiating tumor cells highlights the importance of cell adhesion proteins, proteases, and angiogenesis. Retinoids and bexarotene have been shown to inhibit angiogenesis and metastasis in lung and thyroid cancer murine models (39, 40). RA was shown to decrease secretion of VEGF in thyroid tumor cells in vitro (39). In our study, CYR61 and CXCL1, proteins known to be involved in the invasion process, were downregulated by bexarotene in FTC238 cells and may explain the observed decrease of cell invasion through a collagen layer. CXCL1 is a small cytokine belonging to the CXC chemokine family involved in angiogenesis, inflammation, and tumorigenesis. Overexpression of CXCL1 in melanocytes is associated with constitutive activation of NF-κB leading to tumor progression (41). CYR61, member of a family of secreted matrix-associated proteins, is described as a key regulator of breast cancer invasion (42). The observed rapid downregulation of CYR61 gene expression induced by bexarotene in FTC238 cells favors a direct transcriptional control. Response elements for both retinoid receptors dimers and NF-κB are present in the promoter region of the CYR61 gene. As CYR61 is involved also in pathways leading to enhance NF-κB activation via integrins/PI3K/Akt (43), we may also surmise that in these cells, an autocrine loop between CYR61 and NF-κB maintains tumorigenicity. NF-κB is known to be constitutively activated in thyroid cancer cell lines and patient samples and linked to resistance to apoptosis (44, 45). We found NF-κB constitutively activated in FTC238 cells with high levels of p65 detected in the nucleus and in protein complexes bound to NF-κB sites. Interestingly, incubation with bexarotene results in decreased transactivation of an NF-κB reporter plasmid strongly, suggesting that the inhibition of cell growth observed with bexarotene in FTC238 cells is achieved via repression of NF-κB and downregulation of genes such as MYC and CCND1. Although repression of NF-κB may results from various mechanisms at the cytoplasmic and nuclear level, in FTC238 cells, bexarotene did not inhibit p65 translocation to the nucleus nor its binding to NF-κB–responsive elements, contrary to results reported with fenretinide, a synthetic retinoid (46). Members of the nuclear receptor family, including glucocorticoid, estrogen, progesterone, and androgen receptors and, more recently, PPARγ have been shown to inhibit NF-κB activity and to physically interact with p65 in vitro (29). Furthermore, competition for DNA NF-κB response elements between NF-κB and other transcription factors may occur. This was shown for the same sequence in the promoter of the human GnRH II gene for p65 and RARα/RXRα (47). However in FTC cells, neither could we detect RXR in the p65 transcriptional complex by ChIP nor RXR/p65 interaction by coimmunoprecipitation. These results may well be explained by the low expression of endogenous RXR in these cells. However, other mechanisms may be present in FTC cells. Indeed, our data suggest that bexarotene induces CYR61 promoter repression at the p65-binding site by releasing the p300 coactivator and inducing histone deacetylation. In lipopolysaccharide-activated macrophages, Na and colleagues have also showed that retinoid-mediated suppression of the interleukin-12 production may involve both inhibition of NF-κB–DNA interactions and competitive recruitment of p300 and SRC-1 between NF-κB and RXR (48). Thus we can surmise that in FTC cells, the activation of RXR by its ligand, bexarotene, leads to the release of p300 whether by competitive recruitment or by conformational change in the p65 transcriptional complex, resulting in a histone deacetylation, promoter repression, and a downregulation of NF-κB target genes.

No potential conflicts of interest were disclosed.

The authors thank Ligand Pharmaceuticals for providing bexarotene and Dr C. Schmutzler for FTC cells, as well as Dr P. Fourreau-Moreau, Dr B. Hamon, Dr MB. Hugues, Dr M. Lepage, Pr J. Orgiazzi, Dr AF. Rueff, Pr F. Borson-Chazot, Dr JP. Caravel, Dr M. Cavarec, Dr S. Denet, Dr B. Helal, and Pr M. Schlumberger for clinical study.

This work was supported by funds from the Institut National du Cancer (INCa), the Ligue nationale contre le cancer (Comité de Paris), CRIC 99340 (principal investigator M-E. Toubert), and the Association pour la Recherche sur le Cancer (ARC n°5397, principal investigator M-E. Toubert).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.