The transcription factor NRF2 (NFE2L2) is a pivotal activator of genes encoding cytoprotective and detoxifying enzymes that limit the action of cytotoxic therapies in cancer. NRF2 acts by binding antioxidant response elements (ARE) in its target genes, but there is relatively limited knowledge about how it is negatively controlled. Here, we report that retinoic X receptor alpha (RXRα) is a hitherto unrecognized repressor of NRF2. RNAi-mediated knockdown of RXRα increased basal ARE-driven gene expression and induction of ARE-driven genes by the NRF2 activator tert-butylhydroquinone (tBHQ). Conversely, overexpression of RXRα decreased ARE-driven gene expression. Biochemical investigations showed that RXRα interacts physically with NRF2 in cancer cells and in murine small intestine and liver tissues. Furthermore, RXRα bound to ARE sequences in the promoters of NRF2-regulated genes. RXRα loading onto AREs was concomitant with the presence of NRF2, supporting the hypothesis that a direct interaction between the two proteins on gene promoters accounts for the antagonism of ARE-driven gene expression. Mutation analyses revealed that interaction between the two transcription factors involves the DNA-binding domain of RXRα and a region comprising amino acids 209-316 in human NRF2 that had not been defined functionally, but that we now designate as the NRF2-ECH homology (Neh) 7 domain. In non–small cell lung cancer cells where NRF2 levels are elevated, RXRα expression downregulated NRF2 and sensitized cells to the cytotoxic effects of therapeutic drugs. In summary, our findings show that RXRα diminishes cytoprotection by NRF2 by binding directly to the newly defined Neh7 domain in NRF2. Cancer Res; 73(10); 3097–108. ©2013 AACR.

The human body is continuously threatened by reactive oxygen species (ROS) and electrophiles that are generated by metabolism and by environmental agents. The NF-E2 p45-related factor 2 (NRF2) is a cap'n'collar (CNC) basic-region leucine zipper (bZIP) transcription factor, which plays a major role in protecting cells from prooxidants and electrophiles because it regulates basal and inducible expression of genes that contain antioxidant response element (ARE) sequences in their promoter regions. NRF2-target genes include those encoding antioxidant and detoxication enzymes such as aldo-keto reductase (AKR), heme oxygenase-1 (HO-1), glutathione S-transferase (GST), glutamate-cysteine ligase, and NADP(H):quinone oxidoreductase-1 (NQO1; refs. 1–3).

The ubiquitin ligase substrate adaptor kelch-like ECH-associated protein 1 (KEAP1) is a major repressor of NRF2. Under normal conditions, NRF2 is constantly degraded via the ubiquitin-proteasome pathway in a KEAP1-dependent manner. Under stressed conditions, ROS or eletrophiles modify cysteine residues in KEAP1 causing loss of its adaptor activity, and in turn failure to ubiquitylate NRF2. Upon inactivation of KEAP1, NRF2 accumulates in the nucleus where it heterodimerizes with small Maf proteins and activates ARE-driven genes (4, 5). Recent studies have shown that KEAP1-dependent ubiquitylation of NRF2 can be prevented by protein-protein interactions: these include the binding of p21 to NRF2 or the binding of p62/sequestosome-1 to KEAP1 (6, 7).

In contrast to the wealth of knowledge about the activation of NRF2, far less is known about the mechanisms by which cells turn off or downregulate NRF2 once it has been activated. Importantly, several transcriptional repressors of NRF2 have been identified, such as Bach1, P53, activating transcription factor 3 (ATF3), and estrogen-related receptor beta (ERRβ; refs. 8–11), suggesting that NRF2 activity is strictly regulated even when KEAP1 is inactivated. Nonetheless, constitutive upregulation of NRF2 has been observed in various tumors, including non–small cell lung cancer (NSCLC), and those of breast, head and neck, and gallbladder (12–14). Indeed, deregulation of NRF2 has been shown to contribute to tumorgenesis and drug resistance (14–18). Thus, NRF2 is emerging as a new molecular target for the treatment of certain cancers. It is therefore important to understand the molecular mechanisms by which NRF2 activity can be suppressed because this might provide novel strategies for therapeutic intervention.

In a previous study, we reported that retinoic acid receptor alpha (RARα) antagonizes NRF2 activity (19). Retinoid X receptor (RXR) is the obligatory heterodimerization partner for RARα (20, 21), but its role in regulating the function of NRF2 has not been investigated to date. In the present study, we discovered that RXRα can inhibit the transcriptional activity of NRF2 through a physical interaction between the 2 factors. A RXRα-binding region, which is located between the NRF2-ECH homology (Neh) 5 and Neh6 domains of human NRF2, has been identified. Mutation analyses have revealed that the DNA-binding domain (DBD) of RXRα is required for the interaction with NRF2. Moreover, we have provided evidence that RXRα is capable of interacting with NRF2 on ARE sites in gene promoters, showing a previously unrecognized mechanism by which the CNC-bZIP factor can be inhibited. Downregulation of NRF2 via forced expression of RXRα in NSCLC A549 cells, where the CNC-bZIP factor is constitutively active, increased sensitivity to therapeutic drugs. Thus, our data show a novel mechanism by which RXRα can suppress drug resistance.

Chemicals and cell culture

Unless otherwise stated, all chemicals were from Sigma-Aldrich Co., Ltd., and all antibodies were from Santa Cruz Biotechnology. Antibody against mouse Nrf2 (H300) (sc-13032, Santa Cruz) was used for this study. Actin and β-tubulin antibodies were purchased from Sigma. Alexa Fluor 488 goat anti-rabbit IgG(H+L) was obtained from Invitrogen. HEK293 (human embryonic kidney-293), MCF7 (human breast carcinoma), Caco2 (human colon cancer), and NSCLC A549 cell lines were from the American Type Culture Collection.

Animals

Five-week-old male C57BL/6 male mice were used in this study. The mice (n = 8) were given butylated hydroxyanisole (BHA) by intragastric gavage (i.g.) at 200 mg/kg daily for 3 days. The equivalent volume of corn oil (vehicle) was given to the control mice (n = 8). Mice were sacrificed and tissues were processed as described previously (22). All animal procedures were performed in accordance with the approval of the Laboratory Animals Ethics Committee of Zhejiang University (Zhejiang, China).

Plasmids

Plasmids encoding mouse (m) Nrf2 were provided by Dr Mike McMahon (Medical Research Institute, University of Dundee, Scotland, United Kingdom): these included pcDNA3.1/V5-mNrf2 (full-length), pcDNA3.1/V5-mNrf2ΔETGE (lacking amino acids 79-82) and pcDNA3.1/V5-mNrf2ΔDIDLID (lacking amino acids 17-32; ref. 23). pHyg-EF-hNRF2 encoding EGFP tagged full-length human (h) NRF2 was kindly provided by Dr. Masayuki Yamamoto and Dr. Ken Itoh (Institute of Basic Medical Sciences, University of Tsukuba, Japan). pSG5-mRXRα encoding full-length mRXRα was generously provided by Dr. Pierre Chambon (Institut de Génétique et de Biologie Molèculaire et Cellulaire, CNRS/INSERM/ULP, France). We generated a series of plasmids expressing tagged hNRF2, mNrf2, or mRXRα wild-type or mutants (see Supplementary Methods): as shown in Fig. 2C, 6 plasmids expressing GST-tagged hNRF2 mutants and 7 plasmids encoding GFP-tagged hNRF2 or mNrf2 mutants were created; as shown in Fig. 3A, 4 plasmids expressing GST-tagged mRXRα mutants were generated. All plasmids were verified by DNA sequencing.

Transfections and luciferase reporter gene activity

Lipofectamine 2000 (Invitrogen) was used for transfection (24). The siRNA against hRXRα (RXRα-siRNA) or nontargeting negative control siRNA (scrambled-siRNA) were synthesized by TaKaRa Biotechnology. The sequences for RXRα-siRNA were 5′-GGAGAUGCAUCUAUUUUAATT-3′ (forward) and 5′-UUAAAAUAGAUGCAUCUCCTG-3′ (reverse). Empty vectors were used as negative controls for transfection experiments with plasmids. Twenty-four or 48 hours after transfection, the transfected cells were treated with xenobiotics for 6 hours to 24 hours before being harvested for further analysis. The ARE-luciferase reporter plasmid pGL-GSTA2.41bp-ARE was used and the dual luciferase activities were determined as described elsewhere (25). Stable cell lines A549-mRXRα and A549-EGFP overexpressing GFP-mRXRα and GFP, respectively, were generated as described in Supplementary Methods.

Real-time quantitative PCR

Total RNA isolation and real-time (RT)-PCR was conducted as described previously (24).

Western blot analysis, GST pull-down assay, and immunoprecipitation

Preparation of protein samples, SDS-PAGE gels and immunoblotting was carried out using standard protocols (24). Immunoblotting with antibody against actin or β-tubulin was conducted to confirm equal loading for whole-cell extracts and nuclear extracts, respectively. GST pull-down assay and immunoprecipitation was conducted to detect the interaction between NRF2 and mRXRα mutant GST or GFP fusion proteins. The procedures are provided in Supplementary Methods.

Fluorescently tagged proteins and immunofluorescence

Cells were seeded on glass coverslips, and transiently transfected with the indicated fluorescently labeled proteins. Forty-eight hours later, cells were fixed, processed, and examined as described previously (4, 5). Anti-RXRα antibody was used to detect endogenous RXRα, followed by staining with Texas Red goat anti-rabbit IgG(H+L). Counterstaining with 4′,6-diamidino-2-phenylindole (DAPI) was used to verify the location and integrity of nuclei. The fluorescence images were observed with a Zeiss LSM510 Meta laser-scanning confocal microscope (Carl Zeiss, Inc.).

Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) assays were conducted as described previously (24). The relative binding of NRF2 or RXRα to ARE sites were calculated by quantification of band intensity with an Odyssey Infrared Imaging System (LI-COR Biosciences) normalized to that of the input.

Biotinylated ARE-binding assay

The preparation of nuclear extracts and the Bio-ARE pull-down assay was conducted as described previously (19). A double-stranded 5′-biotinylated ARE probe, representing the 41 bp of nucleotides -682 to -722 in the rat GSTA2 gene promoter, was synthesized by TaKaRa Biotechnology. NRF2 predepletion was conducted by 1 hour incubation of cell lysates with antibody to NRF2 before the Bio-ARE pull-down procedures. The pulled down mixture was analyzed on SDS–PAGE followed by immunoblotting with antibody against RXRα or NRF2.

Cytotoxicity assay

Cytotoxicity was determined as described previously (24). The IC50 and the combination index for determining synergism were calculated as described elsewhere (24).

Statistical analysis

Statistical comparisons were conducted by unpaired Student t tests. P < 0.05 was considered statistically significant.

RXRα inhibits basal and inducible ARE-driven gene expression

To evaluate the effect of RXRα on NRF2 and ARE-driven gene expression, we transfected MCF7 cells with siRNA to knockdown RXRα; immunoblotting confirmed successful knockdown of RXRα in these cells (Fig. 1A, top). In MCF7 cells cotransfected with the ARE-driven reporter plasmid pGL-GSTA2.41bp-ARE, knockdown of RXRα was found to increase basal luciferase reporter activity about 1.5-fold, and increased induction of the reporter gene activity by 20 μmol/L tert-butylhydroquinone (tBHQ) from 4- to 6-fold. To test whether this increase in ARE-driven gene expression was a general effect, we also examined colon cancer Caco2 cells (Fig. 1B, top). Knockdown of RXRα in Caco2 cells resulted in a 2-fold increase in the basal levels of mRNA for endogenous AKR1C1 and HO-1, both of which are NRF2-target genes. Treatment of Caco2 cells with 20 μmol/L tBHQ in which RXRα had been knocked down further increased AKR1C1 mRNA from 10- to 14-fold, and HO-1 mRNA from 5- to 7-fold (Fig. 1B, bottom). These data indicate that loss of RXRα increases NRF2 activity.

We next tested whether overexpression of RXRα might suppress ARE-driven gene expression in Caco2 cells using the pEGFP-C1-mRXRα expression vector; transient expression of exogenous GFP-mRXRα was confirmed by immunoblotting with a specific antibody against RXRα (Fig. 1C). As anticipated, both basal and inducible NRF2-target gene expression were inhibited by forced overexpression of RXRα: the basal AKR1C1 and HO-1 mRNA levels were reduced by 20% and the induction of AKR1C1 and HO-1 mRNA levels by tBHQ was decreased from 11-fold to just 2- and 4-fold, respectively (Fig. 1D). These data show that NRF2-regulated genes are the targets of RXRα-mediated repression, and RXRα can suppress the expression of ARE-driven genes in a ligand-independent manner.

Antagonism of ARE-driven gene expression by RXRα is independent of KEAP1

It is well established that KEAP1 is a major repressor of NRF2 activity (5). To investigate whether KEAP1 plays any role in the antagonism of NRF2 by RXRα, we carried out a further study in the A549 NSCLC cell line, which contains a loss-of-function mutation in KEAP1 (14). RXRα-siRNA was transfected into A549 cells, and knockdown of RXRα was confirmed by immuoblotting (Fig. 1E, left). Consistent with our observations in MCF7 and Caco2 cells, knockdown of RXRα in A549 cells increased AKR1C1 mRNA 2-fold and HO-1 mRNA 4-fold (Fig. 1E, right). It also increased AKR1C1 and HO-1 protein levels significantly (Fig. 1E, left panel). These findings suggest that RXRα inhibition of ARE-driven transcription occurs independently of KEAP1.

RXRα and NRF2 physically interact in vitro

To investigate the mechanism by which RXRα represses NRF2, we examined the localization of the CNC-bZIP transcription factor and its abundance after overexpression of RXRα. Using Caco2 and A549 cells, we found that RXRα altered neither the nuclear accumulation of NRF2 nor its abundance (Fig. 1C & Fig. 6A). We next considered whether RXRα-mediated repression of NRF2 might be a consequence of a direct interaction between the 2 proteins. To test this possibility, we generated a GST-tagged NRF2 construct, and conducted GST-pull-down experiments that tested its ability to interact with RXRα (Fig. 2A and B). The recombinant full-length NRF2 (GST-hNRF2) interacted strongly with His-tagged full-length RXRα protein (Fig. 2B, lane 1). In contrast, the GST control did not bind specifically to RXRα (lane 2). An inverse GST-pull-down assay with recombinant GST-RXRα and His-tagged NRF2 confirmed that the 2 proteins interact specifically (Fig. 3B, lane 2). Thus, our data indicate that NRF2 and RXRα can form a complex in vitro.

To determine the region of NRF2 that is required to interact with RXRα, a series of NRF2 truncated proteins tagged with GST (see Fig. 2C) were expressed, and their abilities to interact with purified recombinant His-RXRα were tested by GST-pull-down assay. We found RXRα interacted with the N-terminal NRF217-338 protein (Fig. 2D, lane 2). In contrast, RXRα failed to interact with the C-terminal NRF2339-605 protein (Fig. 2D, lane 7), suggesting that the Neh6, Neh1, and Neh3 domains of NRF2 are not required for the interaction between the 2 factors. In addition, RXRα failed to interact with the NRF217-110 or NRF2109-219 proteins, which contain Neh2 or Neh4 plus Neh5, respectively (Fig. 2D, lanes 4 and 5), suggesting that these individual domains are not sufficient to enable NRF2 to bind RXRα. Remarkably, RXRα interacted with NRF2109-338 and NRF2209-316 (lanes 3 and 6), whereas deletion of the amino acids 209-316 completely abolished the interaction (Supplementary Fig. S1A, lanes 2 and 3). These results indicate residues 209-316 of NRF2 are sufficient to support an interaction with RXRα in vitro.

To further confirm that amino acids 209-316 of NRF2 are required for the interaction with RXRα, we created a series of expression vectors encoding GFP-tagged NRF2 mutants (Fig. 2C). The analyses were conducted using purified His-RXRα to coimmunoprecipitate material from lysates of HEK293 cells that had been previously transiently transfected with the various NRF2-mutant expression plasmids. Consistent with the GST-pull-down assay results, an antibody against RXRα could coimmunoprecipitate the mNrf2ΔETGE, NRF217-316, and NRF2109-316 mutant proteins, all of which contain amino acids 209-316 of NRF2 (Fig. 2E, lanes 1, 2, and 6). The deletion of amino acids 209-316 abolished this interaction (Supplementary Fig. S1B, lane 3). Moreover, residues 201-329 of mNrf2, which are orthologous to residues 209-316 of hNRF2, could also be immunoprecipitated from the cell extracts (Fig. 2E, lane 7). As a further control, we tested HEK293 cell extracts that expressed GFP alone, and showed that RXRα could not coimmunoprecipitate GFP (Fig. 3D and E). Collectively, these results indicate that amino acids 209-316 of human NRF2 are necessary for its interaction with RXRα in vitro.

The RXRα DBD is required for interaction with NRF2

RXRα comprises 3 major domains: the N-terminal AF-1 domain, the well-conserved DNA-binding domain (DBD), and the ligand-binding domain (LBD) that is responsible for dimerization. It also contains the relatively small AF-2 region with a ligand-dependent transactivation function (refs. 26–28; Fig. 3A). To delineate which region of RXRα is necessary for interaction with NRF2, 4 recombinant GST-RXRα mutants (GST-ΔRXRα) were expressed and purified, and their abilities to interact with purified His-NRF2 were tested by GST-pull-down assay (Fig. 3A). Mutant RXRα230-467, which contains the C-terminal ligand binding and dimerization domains, failed to interact with NRF2 (Fig. 3B, lane 4). Likewise, NRF2 was unable to bind to RXRα1-139, which represents the AF-1 region (lane 6). In contrast, mRXRα1-229, representing the N-terminal amino acid half of RXRα, interacted well with NRF2 (lane 3). Importantly, the mutant RXRα140-205 protein that comprises the DBD domain interacted strongly with NRF2 (lane 5). Accordingly, the GST-luc and GST controls did not bind NRF2 (lanes 1 and 7). Taken together, our data indicate that the DBD is sufficient to interact with NRF2.

RXRα and NRF2 directly interact in vivo

To show whether a physical interaction occurs between RXRα and NRF2 in vivo, we examined whether both proteins colocalize in cells. Plasmids encoding GFP-tagged NRF2 and RFP-tagged mRXRα were transfected into HEK293 cells. Images obtained by confocal laser scanning microscopy revealed that singly expressed GFP-NRF2 and RFP-RXRα were predominately localized in the nucleus of the transfected cells (Fig. 4A, image a–d). As a control, we monitored the cellular localization of the GFP and RFP proteins, both of which were distributed uniformly in the cytoplasm and the nucleus (Supplementary Fig. S2). When GFP-NRF2 and RFP-RXRα were coexpressed, a substantial portion of both proteins coexisted in small foci-like structures within the nucleus (Fig. 4A, image e–g). To confirm these observations, we next tested whether the ectopically expressed GFP-NRF2 could localize with endogenous RXRα. After pHyg-GFP-hNrf2 was transiently transfected into A549 cells alone, the cellular localization of endogenous RXRα was examined using indirect immunofluorescence. Confocal laser scanning microscopy showed a similar nuclear colocalization of the 2 proteins (Fig. 4B, image c). Thus, these results support our contention that NRF2 interacts with RXRα in the nucleus. Moreover, we conducted coimmunoprecitation experiments with total lysates prepared from COS7 cells transfected with the expression vector for V5-tagged mouse Nrf2 with a deletion of residues 17–32 in its Neh2 domain (mNrf2ΔDIDLID-V5). When precipitated with an anti-RXRα antibody, a strong mNrf2 band was detected that suggested both transcription factors coexist in an immunoprecitable complex (Supplementary Fig. S3, top blot, lane 4).

To examine whether an interaction occurs between endogenous NRF2 and endogenous RXRα, we exposed MCF7 cells to 20 μmol/L tBHQ for 24 hours. Immunoprecipitation using antibodies against NRF2 and RXRα revealed the presence of an immuno-complex in the MCF7 cell lysates between endogenous NRF2 and RXRα, which could be increased further by tBHQ treatment [Fig. 4C (a) and 4C (b), lanes 3 and 4]. The abundance of NRF2 (Fig. 4C, c, lane 2) and the expression of its target gene AKR1C (Supplementary Fig. S4, lane 2) was increased as reported previously (25). We next carried out similar studies with cell extracts prepared from mouse tissues. Strikingly, we found that the 2 transcription factors interacted strongly in both small intestine and liver [Fig. 4D (a) and 4D (b), lane 3]. When the mice were treated with BHA, expression of the Nrf2-target genes Nqo1 and Gstm1 was increased in the small intestine and liver (Supplementary Fig. S5A and S5B, lane 2) as expected (1, 22, 29, 30). Again, the interaction between Nrf2 and RXRα was increased by BHA [Fig. 4D (a) and 4D (b), lane 4], suggesting that RXRα is important in regulating the function of Nrf2 in these tissues. Taken together, these results establish that RXRα and NRF2 interact directly in vivo.

RXRα is recruited to the ARE in a NRF2-dependent manner

To assess whether the interaction between RXRα and NRF2 occurs when the factors are bound to DNA, we conducted a ChIP assay with antibody against RXRα or NRF2. As expected, increased NRF2 bound to ARE sequences in the promoters of HO-1 and AKR1C1 after MCF7 cells had been exposed to 20 μmol/L tBHQ for 6 hour (Fig. 5A, lanes 3 and 4). Interestingly, the ChIP assays revealed that RXRα was also able to associate with ARE sites (Fig. 5A, lanes 5 and 6), though it was observed that the DNA bands representing RXRα associated with ARE sequences were much weaker than those representing NRF2. Upon exposure to tBHQ, the amount of RXRα that associated with ARE sequences increased approximately 1.5-fold (Fig. 5A, lanes 5 and 6), correlating with the increase in NRF2 on ARE sites (Fig. 5A, lanes 3 and 4); this occurred in spite of the fact that the abundance of RXRα protein remained unchanged, as seen in MCF7 cells (Fig. 5B, bottom blot, lanes 1 and 2). These results are specific to ARE-containing gene promoters; a DNA sequence in the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) promoter was not detected in complexes immunoprecipitated with either antibody (Fig. 5A). Our findings suggest that the association of RXRα with ARE sequences possibly requires the presence of NRF2, which in turn is able to recruit RXRα to the cis-element.

To further evaluate whether RXRα requires NRF2 to recognize AREs, we carried out a biotinylated-ARE (Bio-ARE) oligonucleotide pull-down assay on nuclear extracts from MCF7 cells treated with 20 μmol/L tBHQ. Although tBHQ did not influence the nuclear levels of RXRα (Fig. 5B, bottom blot, lanes 1 and 2), only RXRα from tBHQ-treated MCF7 cells bound strongly to Bio-ARE beads (Fig. 5B, bottom blot, lane 8). Such binding was concomitant with nuclear accumulation of NRF2 and its increased binding to Bio-ARE beads (Fig. 5B, top blot, lanes 2 and 8). Importantly, depletion of NRF2 from the nuclear extracts after preincubation with an antibody against NRF2 (Fig. 5C, lane 3), not only abolished the binding of NRF2 to Bio-ARE as expected (Fig. 5C, top blot, lanes 8 and 9), but also significantly reduced the binding of RXRα to Bio-ARE (Fig. 5C, bottom blot, lanes 8 and 9). As a negative control, an unrelated biotin-labeled double-stranded oligonucleotide failed to pull down NRF2 or RXRα from the nuclear extracts (Fig. 5B, lane 16; 5C, lane 7). These results suggest that RXRα can be tethered onto DNA by forming a heteromeric protein-protein complex with NRF2.

To examine whether RXRα might form inactive complexes with NRF2 on ARE sequences, we generated a stable MCF7 cell line, designated MCF7-RXRα, which overexpresses RXRα from a pEGFP-mRXRα plasmid. Immunoprecipitation confirmed that both the endogenous RXRα and exogenously derived GFP-mRXRα associated with NRF2 in these stably transfected cells (Supplementary Fig. S6, lane 4). In agreement with the observation following transient transfection of RXRα (Fig. 1D), induction of HO-1 by tBHQ (20 μmol/L) was nearly completely blocked in MCF7-RXRα cells (Fig. 5D, lanes 3 and 4). We found comparable levels of nuclear NRF2 and its binding to ARE sequences stimulated by tBHQ (Fig. 5D and Supplementary Fig. S7, lanes 5–8). However, a ChIP assay revealed that the loading of CREB (cAMP response element binding protein) binding protein (CBP) and RNA polymerase II (RNA Pol II) onto ARE-sites upon tBHQ treatment were markedly attenuated in MCF7-mRXRα cells (Fig. 5E, lanes 7, 8, 11 and 12). This is in contrast to the situation in MCF7 cells, where the levels of CBP and RNA Pol II on the ARE site were dramatically increased under the same conditions (Fig. 5E, lanes 5, 6, 9, and 10). Taken together, our data suggest that the lack of transcriptional activity of NRF2 caused by RXRα overexpression is likely due to direct negative interference by RXRα on ARE-sites, leading to the disruption of the recruitment of CBP and RNA Pol II to the promoters.

Overexpression of RXRα in A549 cells downregulates NRF2 and increases sensitivity to anticancer drugs

Recent studies have provided evidence that high constitutive expression of NRF2 occurs in many cancer cells (31), and RNAi knockdown of the CNC-bZIP factor can sensitize such cells to chemotherapeutic drugs (24, 32, 33). To test whether the repression of NRF2 by RXRα has similar biologic consequences, we generated a cell line named A549-RXRα in which RXRα is stably expressed. As A549 cells carry a somatic KEAP1 mutation, it contains supranormal levels of NRF2 and its target genes are constitutively overexpressed, which in turn increases cell proliferation and resistance to anticancer drugs (14, 32). Immunoblotting confirmed transgene expression (Fig. 6A, lane 2). As expected, expression of the NRF2-target gene HO-1 was diminished in A549-RXRα cells. Cytotoxicity revealed that the IC50 of A549-RXRα to the anticancer drug oxaliplatin was about 40 μmol/L, compared with an IC50 of 100 μmol/L in A549 cells that stably expressed GFP, named A549-GFP (Fig. 6B). The A549-RXRα cells also displayed increased sensitivity to doxorubicin (Fig. 6C). Our results therefore suggest that the downregulation of cytoprotective genes regulated by NRF2 contributes to the increased sensitivity of A549 cells to these drugs by overexpression of RXRα.

Regulation of NRF2 by KEAP1 has been a subject of intense study and it is relatively well characterized. However, little is known about other mechanisms by which NRF2 activity is controlled. Herein, we have presented the first evidence that RXRα functions as a repressor of ARE-driven gene expression. We found RXRα-mediated repression of ARE-driven genes is ligand- and redox-independent, and requires the presence of NRF2 to recruit RXRα to the promoters of target genes. Importantly, we have discovered that amino acids 209–316 of human NRF2 are necessary for its interaction with RXRα, and as this region has not previously been shown to possess functional importance we have designated it the Neh7 domain. Repression of NRF2 by RXRα was observed in 5 different cell types, as well as mouse small intestine and liver, thereby indicating its general significance. To our knowledge, it has not been reported previously that RXRα attenuates ARE-driven gene transcription by directly targeting NRF2.

RXRs play an essential role in the regulation of multiple nuclear hormone-signaling pathways through their ability to dimerize with other nuclear receptors. RXRs mediate retinoid signalling through forming a heterodimer with RAR and by forming a homodimer (21, 34). In addition, RXRs form heterodimers with many other members of the subfamily of nuclear receptors, including peroxisome proliferator-activated receptor, liver X receptor (LXR), pregnane X receptor (PXR) and constitutive androstane receptor (CAR; refs. 21, 34). Heterodimerization of RXR with its partners dramatically enhances its DNA-binding activity (21, 34). Upon binding DNA, some nuclear receptors repress transcription of target genes through their interaction with transcriptional co-repressors in the absence of ligands. Furthermore, ligand binding by a transcriptional agonist causes conformational changes in corepressors, allowing dissociation of transcriptional co-repressors and association of transcriptional coactivators (35). Herein we found that RXRα did not influence nuclear translocation of NRF2 or its binding to ARE sequences. Instead, RXRα associated with ARE-bound NRF2, suggesting that inhibition of NRF2 by RXRα is likely due to the direct interference of recruitment by the CNC-bZIP factor of coactivators to gene promoters. Significantly, we found that an RXRα mutant lacking its ligand binding and heterodimerization domains was as efficient at interacting with NRF2 as the wild-type protein, showing that the nuclear receptor is able to repress NRF2 in a ligand-independent manner.

Previously, we reported that RARα mediates inhibition of NRF2 by all trans retinoic acid through an undefined protein–protein interaction (19). Herein, we have described physical and functional interactions between RXRα and NRF2. Although RXRα and RARα can heterodimerize, our in vitro GST-pull-down study indicates that RXRα binds NRF2 directly and that RARα is not necessary for the interaction between RXRα and NRF2. Specifically, the physical interaction involves the DBD of RXRα and the Neh7 domain of NRF2, and we failed to detect any interaction between RXRα and the CNC-bZIP Neh1 domain of NRF2. This is distinct from other interactions between nuclear hormone receptors and bZIP proteins such as jun and BZLF1 (36, 37) in which the zinc finger DNA-binding domain of the receptor interacts with the bZIP domain of the transcription factor. Thus, our studies show that both RARα and RXRα are repressors of NRF2 activity and presumably act via distinct mechanisms.

Previous work has shown NRF2 contains 6 functional domains, named Neh1-Neh6 (5). It is well known that the stability of NRF2 is controlled through protein–protein interactions between its Neh2 domain and KEAP1. The stability of NRF2 is also controlled by its Neh6 domain, and recently it has been found that this involves β-TrCP-mediated ubiquitylation and phosphorylation of residues in Neh6 by GSK-3 (23, 38). In contrast, the Neh4 and Neh5 regions, which lie adjacent to each other, were found by Katoh and colleagues to act together as a transactivation domain (39); Neh4 and Neh5, both individually and cooperatively, bind CBP, and are indispensable for maximal NRF2 transactivation. Subsequently, Zhang and colleagues (40) reported that the Neh4 and Neh5 domains of NRF2 recruit BRG1, a catalytic subunit of SWI2/SNF2-like chromatin-remodeling complexes, to HO-1 enhancers for transcription initiation. In the present study, we identified a RXRα interaction domain in NRF2 (now called Neh7), which abuts the Neh5 domain of NRF2. We found overexpression of RXRα reduced the loading of CBP and RNA Pol II onto ARE sites, suggesting that the binding of RXRα may disrupt the binding of CBP to the Neh4 and Neh5 domains of NRF2, thereby suppressing transcriptional initiation. We therefore propose a model in which protein–protein contact between RXRα and NRF2 prevents a productive interaction between the transactivation domains of NRF2 and the basal transcription machinery.

In adult mammalian liver, RXRα is the most abundant among the 3 RXR isoforms (i.e., RXRα, -β, and -γ) and is an obligatory partner of two major xenobiotic receptors, CAR and PXR (41). Previous studies have revealed that Nrf2 is also highly expressed in the liver (22), and plays a key role in regulating the expression of phase II drug-metabolizing enzymes and drug-efflux pumps (30, 33, 42, 43). In the present study, we showed that Nrf2 and RXRα interact in vivo. In agreement with our observation, Dai and colleagues (44) reported that in hepatocyte-specific RXRα knockout mice, the expression of Gsta1 and/or Gsta2, Gstm1 and/or Gstm3, Gstm2 and Gstm4, all of which are regulated by NRF2 (29, 30, 44), were increased compared with their levels in the liver of wild-type mice, presumably due to loss of NRF2 suppression. Moreover, the expression of these Gst subunits from hepatocyte-specific RXRα knockout mice was further enhanced by acetaminophen, whose hepatotoxic metabolite was able to directly activate the KEAP1-NRF2 pathway (45), showing RXRα represses NRF2/ARE signalling pathway in vivo. Taken together, we hypothesize that RXRα plays a key role in linking xenobiotic metabolism with the NRF2-ARE cytoprotective signaling pathway. Furthermore, several nuclear receptors besides RXRα repress NRF2 (11, 46, 47). It is interesting to speculate that antagonism of NRF2 by RXRα occurs because at some level the CNC-bZIP protein and nuclear receptors are functionally incompatible, and that it might be advantageous to attenuate NRF2 activity.

Recent studies have revealed that NRF2 exhibits abnormal increased activity in several types of tumor due to oncogene activation, or somatic mutation of KEAP1 or NRF2 (15, 17). The overactivation of NRF2-ARE signaling in cancer cells promotes drug resistance and cell proliferation (31, 48). It has been reported that RXRα is downregulated in many tumors (49–51). On the basis of our study, it seems plausible that a reduction of RXRα will up-regulate the NRF2-ARE signalling pathway, thereby contributing to tumorigenesis and drug resistance. In this study, we have shown that forced expression of RXRα in NSCLC A549 cells downregulated the NRF2/ARE cytoprotective pathway and sensitized them to anticancer drugs, suggesting that transcriptional repression by RXRα may be used as a mechanism to attain an appropriate level of gene expression in NRF2 overactive cell types. Here, we describe the Neh7 domain in Nrf2 as a potential target by which the CNC-bZIP factor can be inhibited.

In summary, in this report we describe the interaction of NRF2 with RXRα introducing a new dimension to our understanding of the transcriptional hierarchy in ARE-driven gene regulation. Our studies, therefore, suggest a novel, and as yet unrecognized, function of RXRα as a putative transcriptional co-repressor of NRF2.

No potential conflicts of interest were disclosed.

Conception and design: H. Wang, K. Liu, P. Gao, Y. Hai, Y. Li, J.D. Hayes, X.J. Wang, X. Tang

Development of methodology: H. Wang, K. Liu, M. Geng, P. Gao, X. Wu, Y. Hai, Y. Li, X.J. Wang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Wang, K. Liu, M. Geng, P. Gao, X. Wu, Y. Hai, Y. Li, Y. Li, L. Luo, X.J. Wang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Wang, K. Liu, M. Geng, P. Gao, X. Wu, Y. Hai, Y. Li, X.J. Wang, X. Tang

Writing, review, and/or revision of the manuscript: H. Wang, J.D. Hayes, X.J. Wang, X. Tang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Wang, K. Liu, M. Geng, X. Wu, Y. Li, X.J. Wang

Study supervision: H. Wang, K. Liu, Y. Li, X.J. Wang, X. Tang

The authors thank Professor Roland Wolf for his generosity in providing reagents and Ms. Ai Xin for her technical assistance.

This work was supported by NSFC (31170743 and 81172230), ZJKJT (2010C33156 and 20112308), the 111 Project (B13026), and ZJNSFC (LZ12H16001).

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.

1.
Itoh
K
,
Chiba
T
,
Takahashi
S
,
Ishii
T
,
Igarashi
K
,
Katoh
Y
, et al
An Nrf2/small Maf heterodimer mediates the induction of phase II detoxifying enzyme genes through antioxidant response elements
.
Biochem Biophys Res Commun
1997
;
236
:
313
22
.
2.
Wu
KC
,
McDonald
PR
,
Liu
JJ
,
Chaguturu
R
,
Klaassen
CD
. 
Implementation of a high-throughput screen for identifying small molecules to activate the Keap1-Nrf2-ARE pathway
.
PLoS ONE
2012
;
7
:
e44686
.
3.
Yates
MS
,
Kensler
TW
. 
Chemopreventive promise of targeting the Nrf2 pathway
.
Drug News Perspect
2007
;
20
:
109
17
.
4.
Dinkova-Kostova
AT
,
Holtzclaw
WD
,
Cole
RN
,
Itoh
K
,
Wakabayashi
N
,
Katoh
Y
, et al
Direct evidence that sulfhydryl groups of Keap1 are the sensors regulating induction of phase 2 enzymes that protect against carcinogens and oxidants
.
Proc Natl Acad Sci U S A
2002
;
99
:
11908
13
.
5.
Itoh
K
,
Wakabayashi
N
,
Katoh
Y
,
Ishii
T
,
Igarashi
K
,
Engel
JD
, et al
Keap1 represses nuclear activation of antioxidant responsive elements by Nrf2 through binding to the amino-terminal Neh2 domain
.
Genes Dev
1999
;
13
:
76
86
.
6.
Chen
W
,
Sun
Z
,
Wang
XJ
,
Jiang
T
,
Huang
Z
,
Fang
D
, et al
Direct interaction between Nrf2 and p21(Cip1/WAF1) upregulates the Nrf2-mediated antioxidant response
.
Mol Cell
2009
;
34
:
663
73
.
7.
Komatsu
M
,
Kurokawa
H
,
Waguri
S
,
Taguchi
K
,
Kobayashi
A
,
Ichimura
Y
, et al
The selective autophagy substrate p62 activates the stress responsive transcription factor Nrf2 through inactivation of Keap1
.
Nat Cell Biol
2010
;
12
:
213
23
.
8.
Brown
SL
,
Sekhar
KR
,
Rachakonda
G
,
Sasi
S
,
Freeman
ML
. 
Activating transcription factor 3 is a novel repressor of the nuclear factor erythroid-derived 2-related factor 2 (Nrf2)-regulated stress pathway
.
Cancer Res
2008
;
68
:
364
8
.
9.
Faraonio
R
,
Vergara
P
,
Di Marzo
D
,
Pierantoni
MG
,
Napolitano
M
,
Russo
T
, et al
p53 suppresses the Nrf2-dependent transcription of antioxidant response genes
.
J Biol Chem
2006
;
281
:
39776
84
.
10.
Hoshino
H
,
Kobayashi
A
,
Yoshida
M
,
Kudo
N
,
Oyake
T
,
Motohashi
H
, et al
Oxidative stress abolishes leptomycin B-sensitive nuclear export of transcription repressor Bach2 that counteracts activation of Maf recognition element
.
J Biol Chem
2000
;
275
:
15370
6
.
11.
Zhou
W
,
Lo
SC
,
Liu
JH
,
Hannink
M
,
Lubahn
DB
. 
ERRbeta: a potent inhibitor of Nrf2 transcriptional activity
.
Mol Cell Endocrinol
2007
;
278
:
52
62
.
12.
Nioi
P
,
Nguyen
T
. 
A mutation of Keap1 found in breast cancer impairs its ability to repress Nrf2 activity
.
Biochem Biophys Res Commun
2007
;
362
:
816
21
.
13.
Ohta
T
,
Iijima
K
,
Miyamoto
M
,
Nakahara
I
,
Tanaka
H
,
Ohtsuji
M
, et al
Loss of Keap1 function activates Nrf2 and provides advantages for lung cancer cell growth
.
Cancer Res
2008
;
68
:
1303
9
.
14.
Singh
A
,
Misra
V
,
Thimmulappa
RK
,
Lee
H
,
Ames
S
,
Hoque
MO
, et al
Dysfunctional KEAP1-NRF2 interaction in non-small-cell lung cancer
.
PLoS Med
2006
;
3
:
e420
.
15.
DeNicola
GM
,
Karreth
FA
,
Humpton
TJ
,
Gopinathan
A
,
Wei
C
,
Frese
K
, et al
Oncogene-induced Nrf2 transcription promotes ROS detoxification and tumorigenesis
.
Nature
2011
;
475
:
106
9
.
16.
Morrow
CS
,
Smitherman
PK
,
Diah
SK
,
Schneider
E
,
Townsend
AJ
. 
Coordinated action of glutathione S-transferases (GSTs) and multidrug resistance protein 1 (MRP1) in antineoplastic drug detoxification. Mechanism of GST A1-1- and MRP1-associated resistance to chlorambucil in MCF7 breast carcinoma cells
.
J Biol Chem
1998
;
273
:
20114
20
.
17.
Shibata
T
,
Kokubu
A
,
Gotoh
M
,
Ojima
H
,
Ohta
T
,
Yamamoto
M
, et al
Genetic alteration of Keap1 confers constitutive Nrf2 activation and resistance to chemotherapy in gallbladder cancer
.
Gastroenterology
2008
;
135
:
1358
68
.
18.
Tew
KD
. 
Glutathione-associated enzymes in anticancer drug resistance
.
Cancer Res
1994
;
54
:
4313
20
.
19.
Wang
XJ
,
Hayes
JD
,
Henderson
CJ
,
Wolf
CR
. 
Identification of retinoic acid as an inhibitor of transcription factor Nrf2 through activation of retinoic acid receptor alpha
.
Proc Natl Acad Sci U S A
2007
;
104
:
19589
94
.
20.
Chambon
P
. 
A decade of molecular biology of retinoic acid receptors
.
FASEB J
1996
;
10
:
940
54
.
21.
Mangelsdorf
DJ
,
Evans
RM
. 
The RXR heterodimers and orphan receptors
.
Cell
1995
;
83
:
841
50
.
22.
McMahon
M
,
Itoh
K
,
Yamamoto
M
,
Chanas
SA
,
Henderson
CJ
,
McLellan
LI
, et al
The Cap'n'Collar basic leucine zipper transcription factor Nrf2 (NF-E2 p45-related factor 2) controls both constitutive and inducible expression of intestinal detoxification and glutathione biosynthetic enzymes
.
Cancer Res
2001
;
61
:
3299
307
.
23.
McMahon
M
,
Thomas
N
,
Itoh
K
,
Yamamoto
M
,
Hayes
JD
. 
Redox-regulated turnover of Nrf2 is determined by at least two separate protein domains, the redox-sensitive Neh2 degron and the redox-insensitive Neh6 degron
.
J Biol Chem
2004
;
279
:
31556
67
.
24.
Tang
X
,
Wang
H
,
Fan
L
,
Wu
X
,
Xin
A
,
Wang
XJ
. 
Luteolin inhibits NRF2 leading to negative regulation of the NRF2/ARE pathway and sensitization of human lung carcinoma A549 cells to therapeutic drugs
.
Free Radic Biol Med
2011
;
50
:
1599
609
.
25.
Wang
XJ
,
Hayes
JD
,
Wolf
CR
. 
Generation of a stable antioxidant response element-driven reporter gene cell line and its use to show redox-dependent activation of nrf2 by cancer chemotherapeutic agents
.
Cancer Res
2006
;
66
:
10983
94
.
26.
Lee
MS
,
Kliewer
SA
,
Provencal
J
,
Wright
PE
,
Evans
RM
. 
Structure of the retinoid X receptor alpha DNA binding domain: a helix required for homodimeric DNA binding
.
Science
1993
;
260
:
1117
21
.
27.
Rastinejad
F
,
Wagner
T
,
Zhao
Q
,
Khorasanizadeh
S
. 
Structure of the RXR-RAR DNA-binding complex on the retinoic acid response element DR1
.
EMBO J
2000
;
19
:
1045
54
.
28.
Zechel
C
,
Shen
XQ
,
Chen
JY
,
Chen
ZP
,
Chambon
P
,
Gronemeyer
H
. 
The dimerization interfaces formed between the DNA binding domains of RXR, RAR and TR determine the binding specificity and polarity of the full-length receptors to direct repeats
.
EMBO J
1994
;
13
:
1425
33
.
29.
Higgins
LG
,
Hayes
JD
. 
Mechanisms of induction of cytosolic and microsomal glutathione transferase (GST) genes by xenobiotics and pro-inflammatory agents
.
Drug Metab Rev
2011
;
43
:
92
137
.
30.
Wu
KC
,
Cui
JY
,
Klaassen
CD
. 
Effect of graded Nrf2 activation on phase-I and -II drug metabolizing enzymes and transporters in mouse liver
.
PLoS One
2012
;
7
:
e39006
.
31.
Hayes
JD
,
McMahon
M
. 
NRF2 and KEAP1 mutations: permanent activation of an adaptive response in cancer
.
Trends Biochem Sci
2009
;
34
:
176
88
.
32.
Singh
A
,
Boldin-Adamsky
S
,
Thimmulappa
RK
,
Rath
SK
,
Ashush
H
,
Coulter
J
, et al
RNAi-mediated silencing of nuclear factor erythroid-2-related factor 2 gene expression in non-small cell lung cancer inhibits tumor growth and increases efficacy of chemotherapy
.
Cancer Res
2008
;
68
:
7975
84
.
33.
Anwar-Mohamed
A
,
Degenhardt
OS
,
El Gendy
MA
,
Seubert
JM
,
Kleeberger
SR
,
El-Kadi
AO
. 
The effect of Nrf2 knockout on the constitutive expression of drug metabolizing enzymes and transporters in C57Bl/6 mice livers
.
Toxicol In Vitro
2011
;
25
:
785
95
.
34.
Kastner
P
,
Mark
M
,
Chambon
P
. 
Nonsteroid nuclear receptors: what are genetic studies telling us about their role in real life?
Cell
1995
;
83
:
859
69
.
35.
Xu
L
,
Glass
CK
,
Rosenfeld
MG
. 
Coactivator and corepressor complexes in nuclear receptor function
.
Curr Opin Genet Dev
1999
;
9
:
140
7
.
36.
Pfahl
M
. 
Nuclear receptor/AP-1 interaction
.
Endocr Rev
1993
;
14
:
651
8
.
37.
Pfitzner
E
,
Becker
P
,
Rolke
A
,
Schule
R
. 
Functional antagonism between the retinoic acid receptor and the viral transactivator BZLF1 is mediated by protein-protein interactions
.
Proc Natl Acad Sci U S A
1995
;
92
:
12265
9
.
38.
Rada
P
,
Rojo
AI
,
Chowdhry
S
,
McMahon
M
,
Hayes
JD
,
Cuadrado
A
. 
SCF/{beta}-TrCP promotes glycogen synthase kinase 3-dependent degradation of the Nrf2 transcription factor in a Keap1-independent manner
.
Mol Cell Biol
2011
;
31
:
1121
33
.
39.
Katoh
Y
,
Itoh
K
,
Yoshida
E
,
Miyagishi
M
,
Fukamizu
A
,
Yamamoto
M
. 
Two domains of Nrf2 cooperatively bind CBP, a CREB binding protein, and synergistically activate transcription
.
Genes Cells
2001
;
6
:
857
68
.
40.
Zhang
J
,
Ohta
T
,
Maruyama
A
,
Hosoya
T
,
Nishikawa
K
,
Maher
JM
, et al
BRG1 interacts with Nrf2 to selectively mediate HO-1 induction in response to oxidative stress
.
Mol Cell Biol
2006
;
26
:
7942
52
.
41.
Mangelsdorf
DJ
,
Borgmeyer
U
,
Heyman
RA
,
Zhou
JY
,
Ong
ES
,
Oro
AE
, et al
Characterization of three RXR genes that mediate the action of 9-cis retinoic acid
.
Genes Dev
1992
;
6
:
329
44
.
42.
Chanas
SA
,
Jiang
Q
,
McMahon
M
,
McWalter
GK
,
McLellan
LI
,
Elcombe
CR
, et al
Loss of the Nrf2 transcription factor causes a marked reduction in constitutive and inducible expression of the glutathione S-transferase Gsta1, Gsta2, Gstm1, Gstm2, Gstm3 and Gstm4 genes in the livers of male and female mice
.
Biochem J
2002
;
365
:
405
16
.
43.
Cheng
Q
,
Taguchi
K
,
Aleksunes
LM
,
Manautou
JE
,
Cherrington
NJ
,
Yamamoto
M
, et al
Constitutive activation of nuclear factor-E2-related factor 2 induces biotransformation enzyme and transporter expression in livers of mice with hepatocyte-specific deletion of Kelch-like ECH-associated protein 1
.
J Biochem Mol Toxicol
2011
;
25
:
320
9
.
44.
Dai
G
,
Chou
N
,
He
L
,
Gyamfi
MA
,
Mendy
AJ
,
Slitt
AL
, et al
Retinoid X receptor alpha regulates the expression of glutathione S-transferase genes and modulates acetaminophen-glutathione conjugation in mouse liver
.
Mol Pharmacol
2005
;
68
:
1590
6
.
45.
Copple
IM
,
Goldring
CE
,
Jenkins
RE
,
Chia
AJ
,
Randle
LE
,
Hayes
JD
, et al
The hepatotoxic metabolite of acetaminophen directly activates the Keap1-Nrf2 cell defense system
.
Hepatology
2008
;
48
:
1292
301
.
46.
Ansell
PJ
,
Lo
SC
,
Newton
LG
,
Espinosa-Nicholas
C
,
Zhang
DD
,
Liu
JH
, et al
Repression of cancer protective genes by 17beta-estradiol: ligand-dependent interaction between human Nrf2 and estrogen receptor alpha
.
Mol Cell Endocrinol
2005
;
243
:
27
34
.
47.
Ikeda
Y
,
Sugawara
A
,
Taniyama
Y
,
Uruno
A
,
Igarashi
K
,
Arima
S
, et al
Suppression of rat thromboxane synthase gene transcription by peroxisome proliferator-activated receptor γ in macrophages via an interaction with NRF2
.
J Biol Chem
2000
;
275
:
33142
50
.
48.
Kensler
TW
,
Wakabayashi
N
. 
Nrf2: friend or foe for chemoprevention?
Carcinogenesis
2010
;
31
:
90
9
.
49.
Brabender
J
,
Danenberg
KD
,
Metzger
R
,
Schneider
PM
,
Lord
RV
,
Groshen
S
, et al
The role of retinoid X receptor messenger RNA expression in curatively resected non-small cell lung cancer
.
Clin Cancer Res
2002
;
8
:
438
43
.
50.
Brabender
J
,
Lord
RV
,
Metzger
R
,
Park
J
,
Salonga
D
,
Danenberg
KD
, et al
Role of retinoid X receptor mRNA expression in Barrett's esophagus
.
J Gastrointest Surg
2004
;
8
:
413
22
.
51.
Brabender
J
,
Metzger
R
,
Salonga
D
,
Danenberg
KD
,
Danenberg
PV
,
Hölscher
AH
, et al
Comprehensive expression analysis of retinoic acid receptors and retinoid X receptors in non-small cell lung cancer: implications for tumor development and prognosis
.
Carcinogenesis
2005
;
26
:
525
30
.