Retinoids regulate gene transcription through activating retinoic acid receptors (RARs)/retinoic X receptors (RXRs). Of the three RAR receptors (α, β, and γ), RARβ has been considered a tumor suppressor gene. Here, we identified a novel RARβ isoform-RARβ5 in breast epithelial cells, which could play a negative role in RARβ signaling. Similar to RARβ2, the first exon (59 bp) of RARβ5 is RARβ5 isoform specific, whereas the other exons are common to all of the RARβ isoforms. The first exon of RARβ5 does not contain any translation start codon, and therefore its protein translation begins at an internal methionine codon of RARβ2, lacking the A, B, and part of C domain of RARβ2. RARβ5 protein was preferentially expressed in estrogen receptor-negative breast cancer cells and normal breast epithelial cells that are relatively resistant to retinoids, whereas estrogen receptor-positive cells that did not express detectable RARβ5 protein were sensitive to retinoid treatment, suggesting that this isoform may affect the cellular response to retinoids. RARβ5 isoform is unique among all of the RARs, because a corresponding isoform was not detectable for either RARα or RARγ. RARβ5 mRNA was variably expressed in normal and cancerous breast epithelial cells. Its transcription was under the control of a distinct promoter P3, which can be activated by all-trans-retinoic acid (atRA) and other RAR/RXR selective retinoids in MCF-7 and T47D breast cancer cells. We mapped the RARβ5 promoter and found a region -302/-99 to be the target region of atRA. In conclusion, we identified and initially characterized RARβ5 in normal, premalignant, and malignant breast epithelial cells. RARβ5 may serve as a potential target of retinoids in prevention and therapy studies.
The biological effects of retinoids are mainly mediated by two families of nuclear receptors: retinoic acid receptors (RARs) and retinoic X receptors (RXRs), each consisting of three receptor subtypes (α, β, γ; refs. 1, 2). In addition, each RAR gene generates multiple isoforms by either alternative splicing or differential usage of two promoters (1, 2). RARs/RXRs belong to the superfamily of nuclear receptors that mediate the transcriptional effects of steroid hormones, vitamin D, and thyroid hormone (3). RARs preferentially dimerize with RXRs to form RAR-RXR heterodimers that are thought to be obligatory intermediates in the effects of RAR ligands on gene expression (4). RXRs also can homodimerize to form transcriptionally active complexes (5). Homo- and heterodimeric retinoid receptor complexes bind to distinct retinoid response elements embedded in the regulatory regions of retinoid-responsive genes (6). Although there is considerable variability in the sequence and structure of the retinoid response elements in retinoid-regulated genes, they conform to a general canonical sequence in which two directly repeated receptor-binding hexanucleotide motifs [consensus (A/G)G(G/T)TCA] are separated by a variable number of intervening nucleotides (6).
RARβ itself is a retinoid target gene and believed to play a role as a tumor suppressor gene in tumorigenesis (7). The human RARβ gene was first identified from hepatocellular carcinoma in 1987 (8), followed by the identification of retinoic acid response element (RARE) in its promoter region (9). In the mice, the RARβ gene generates four distinct transcripts: splice variants RARβ1 and RARβ3 from transcription at promoter P1, and RARβ2 and RARβ4 from the RARE-containing P2 promoter (2, 10). In the human, only RARβ2 and RARβ4 transcripts have been identified in normal adult cells (11). Human RARβ1 is expressed in fetal tissues and some small cell lung carcinoma cell lines (12); whereas a human homologue of the RARβ3 isoform has not been detected (7). The RARβ2 and RARβ4 transcripts differ only in the content of their 5′-most exon, a result of alternative splicing (10). On the basis of homology with other members of the steroid hormone receptor superfamily, six distinct domains (A–F) have been identified within RARs and RXRs (2). Thus far, all of the identified RAR isoforms are only different at their unique A domain and are derived from two promoters and alternative splicing (11). Isoforms of a given RAR gene generally contain identical protein sequences B–F (11).
We’ve been interested in RAR/RXR alteration in the process of breast tumor progression with a MCF10 model (13). During the characterization of RARβ expression in the MCF10 series of cell lines (benign MCF10A, premalignant MCF10AT, and malignant MCF10CA1a cell lines; ref. 14), we identified a novel RARβ isoform, which we named RARβ5. RARβ5 mRNA expression is under the control of a distinct promoter P3 and is mediated by all-trans-retinoic acid (atRA) and other RAR/RXR selective ligands in breast cancer cells. It was detected in both normal and breast cancer cells. We also cloned and initially characterized the promoter region of RARβ5. The same protein isoform was previously associated with RARβ4 transcript (11, 15) and then termed RARβ′ (7). In this study, we have identified RARβ5 at both gene and protein level.
MATERIALS AND METHODS
The MCF10A cell line was received from Karmanos Cancer Institute (Detroit, MI) and cultured as described previously (13). Normal human mammary epithelial cells (HMEC) were purchased from Clonetics (Santa Rosa, CA) and cultured in MEGM with supplements (Clonetics). MCF-7, T47D, and MDA-MB435 cell lines were purchased from American Type Tissue Collection (Manassas, VA). BCA-1 to BCA-11 breast carcinoma cell lines were from breast cancer patients and established in our laboratory (16). These cell lines are still at their early passages (passage number at 7-10), and their characteristic features are summarized in Appendix 1 as supplemental material. All of these cell lines were cultured in MEM supplemented with 100 units/mL penicillin, 100 μg/mL streptomycin, and 10% fetal bovine serum, 200 μmol/L l-glutamine and 100 μmol/L MEM nonessential amino acids. The atRA was purchased from Sigma (St. Louis, MO). The 9-cis–retinoic acid (9-cisRA), 4-hydroxyphenyl retinamide (4-HPR), and LGD1069 were obtained from the repository of the National Cancer Institute (Bethesda, MD). Am80 (RARα/β selective ligand) was a generous gift from Dr. Koichi Shudo (ITSUU Laboratory, Tokyo, Japan).
Rapid Amplification of cDNA 5′-Ends and cDNA Cloning.
Rapid amplification of cDNA 5′-ends (5′-RACE) was done with SMART RACE cDNA Amplification Kit (Clontech Laboratories, Inc., Palo Alto, CA) according to the User Manual. RACE PCR was done with a Universal Primer and two RARβ2 gene-specific reverse primers located at 1754 bp (RARβ2-1754RP, GGACTGTGCTCTGCTGTGTTCCCACTT) and 1216 bp (RARβ2-1216RP, GGTCTGCGATGGTCAAGCCAGTGAA) 3′ of the RARβ2 transcription site. PCR products were cloned into pCR4-TOPO vector (Invitrogen, Carlsbad, CA) and sequenced with ABI PRISM 377 DNA Sequencer (Applied Biosystems, Foster City, CA).
RT was done in a final volume of 20 μL with 2 μg total RNA and 100 units of MuLV reverse transcriptase (Invitrogen) at 42°C for 50 minutes. Conventional PCR was mainly used to qualitatively detect gene expression. PCR was done with 1 μL RT product with PCR Supermix (Invitrogen). PCR primer pairs are RARβ2 (475FP, GACTGTATGGATGTTCTGTCAG; and 730RP, ATTTGTCCTGGCAGACGAAGCA) and RARβ5 (14FP, CTGGAAGGTCGTACACAGTGA; and 343RP, GGACATTCCCACTTC-AAAGC). β-actin (FP, GTCACCAACTGGGACGACA; and RP, TGGCCATCTCTTGC-TCGAA) was used as an internal control. Real-time PCR was done with 1 μL RT product with 7900HT Sequence Detection System (ABI, Applied Biosystems) and ABI 2 × SYBR Green PCR Master Mix (ABI# 4309155) according to recommended guidelines of ABI. Primer pairs for real-time PCR were RARβ2 (584FP, GATTGACCCAAACCGAATGGCAGCA; and 730RP) and RARβ5 (15FP, GGAAGGT-CGTACACAGTGAATTTCTCTGAG; and RARβ2-730RP); real-time PCR data were analyzed with a software package (ABI Prizm SDS2.1) provided with the instrumentation system.
Expression Vector and In vitro Translation.
The RARβ5 expression vector was generated by PCR-cloning with pcDNA3.1/V5-His TOPO TA Expression Kit (Invitrogen). The open reading frame (ORF) of RARβ5 was isolated by PCR of 5′-RACE cDNA from MDA-MB435 cells with primers containing RARβ5 start and stop codon (RARβ5-start, CAGAAGAATATGATTTACACTTGTCACCG; and RARβ5-stop, GTCTTATTGCACGA-GTGGTGACTG). RARβ2 expression vector pTag (RARβ2 β′-; RARβ2 with a mutation to knockout a downstream translation start site) was a generous gift from Dr. Karen Swisshelm (Department of Pathology, University of Washington, Seattle, WA; ref. 11). RARβ2 insert was cut from pTag (RARβ2 β′-) and ligated into the BamH1 sites of pcDNA3.1 vector (Invitrogen) to generate RARβ2 expression vector-pcDNΑ3.1(RARβ2). The pcDNA3.1(RARβ2) was used for both in vitro translation to generate RARβ2 protein as positive control for Western blot and cotransfection to test the effects of RARβ2 expression on RARβ5 promoter activity. In vitro translation was done with TNT Quick Coupled Translation kit (Promega, Madison, WI).
When cells grew to 50 to 70% confluence, cell lysates were prepared and subjected to Western blot analysis as described previously (17). Two antibodies were used to detect RARβ isoforms, one recognizing amino acids 430-447 in the COOH-terminus of RARβ2 (sc-552, Santa Cruz Biotechnology Inc., Santa Cruz, CA), the other one recognizing amino acids 407-423 in the COOH-terminus of RARβ2 (Geneka Biotechnology Inc., Montreal, Quebec). Another two antibodies recognizing COOH-terminus of RARα (sc-551, Santa Cruz Biotechnology Inc.) and RARγ (sc-550, Santa Cruz Biotechnology Inc.) were used to detect RARα and RARγ isoforms.
The 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium Bromide Assay for Cell Growth.
Cell proliferation was examined by colorimetric [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; MTT] assay. MTT is a pale yellow substrate that is cleaved by living cells to yield a dark blue formazan product. This process requires active mitochondria, and even freshly dead cells do not cleave substantial amounts of MTT. Briefly, cells (500 cells per well) were seeded in 96-well plates and cultured overnight. Then cells were incubated with 1 μmol/L retinoids, and the media were changed every second day. After 7-day treatment, 0.01 mL of MTT solution (5 mg/mL) was added to each well, mixed gently, and incubated with the cells at 37°C for 2 to 3 hours. The media were carefully removed, 0.1 mL of DMSO was added to each well, and plates were assayed for cell proliferation as described previously (18).
RARβ5 Promoter-Luciferase Reporter Plasmids.
The 1-kb 5′ flanking region (P-1000/+33 relative to the transcription start site) of RARβ5 was first isolated by PCR from genomic DNA extracted from MDA-MB435 breast cancer cells with Advantage 2 PCR kit (Clontech). Primer pairs were designed to contain a Kpn restriction site at the 5′ end of the forward primer and a Xho restriction site at the 5′ end of the reverse primer. The PCR product was first cloned to the pCR4-TOPO vector, and then subcloned to the Kpn/Xho sites of the promoterless PGL3 basic vector (Promega). Orientation and sequence of all of the constructs were verified by direct sequencing. All of the other promoter deletion mutation constructs (P-428/+33, −323/+33, −302/+33, −177/+33, −99/+33) were cloned in the same way with PGL3-P-1000/+33 as a template.
Cell Transfections and Luciferase Assay.
MCF-7 and T47D cells were plated at 1 to 1.2 × 105 cells per well in 12-well plates. After overnight incubation, the media were replaced by MEM containing 2% fetal bovine serum. Transient transfection was done in the same media with Lipofectamine 2000 (Invitrogen). Cells were transfected with 0.5 μg/well promoter constructs (0.5 μg/well for PGL3-P-1000/+33, the amount of other deletion mutants was correspondingly adjusted to make each well contain the same amount of the plasmids) with or without 0.5 μg/well pcDNA3.1 empty vector or 0.64 μg/well pcDNA3.1(RARβ2) expression vector. A 20 ng/well pCMVβgal vector (Clontech) was cotransfected as an internal control for transfection efficiency. After 5 hours incubation, the medium was replaced with a fresh one containing 1 μmol/L atRA or other retinoids or DMSO (solvent control, 1 μL/10 mL media), and cells were incubated for an additional 24 hours. Luciferase and β-galactosidase activities were assayed with Luciferase Reporter Assay Kit and Luminescent β-gal detection kit II (Clontech).
A Novel RARβ Transcript in MCF10A Breast Epithelial Cells.
During characterization of RARβ2 expression in MCF10A series of cell lines (14), with primers recognizing RARβ2 coding region, we detected RARβ2 transcript by reverse transcription (RT)-PCR; but we failed to detect RARβ2 transcript with primers recognizing RARβ2 5′-untranslated region (UTR). Hence, to examine the 5′ region of RARβ2 in MCF10A cells, we did a 5′-RACE analysis of the MCF10A total RNA with two RARβ2 specific primers (Fig. 1,A). Using these two primers, we failed to detect the expected RARβ2 fragments with size ∼1.8 kb and ∼1.3 kb, respectively, but did consistently detect a band with a smaller size (∼0.6 kb shorter). The 5′-RACE product was cloned and sequenced. The Blast search of GenBank suggests it to be a novel RARβ isoform that has not yet been reported, henceforth it is referred to as RARβ5. The 5′ end sequence of RARβ5 cDNA is presented in Fig. 1,B. Only the first exon (Fig. 1; exon 6, 58 bp, bold) is RARβ5 specific. We also used another primer set designed in terms of the RARβ5-specific sequence (RARβ5-14FP) and the 3′-UTR sequence of RARβ2 (RARβ2-1827RP) to amplify a fragment spanning the whole coding region of RARβ5. Cloning and sequencing analysis of this PCR product showed a unique first exon of the RARβ5, whereas all of the downstream exons are common to all of the isoforms of RARβ. Alignment of the first exon of RARβ5 to bacterial artificial chromosome (BAC) clones RP11-421F9 and RP11-733H11 (GenBank accession nos. AC133141.2 and AC098477.2) shows that the first exon of RARβ5 is located ∼29.4 kb downstream of the first exon (exon 5) of RARβ2 and ∼2.8 kb upstream of the second exon of RARβ2. Therefore, this novel exon is numbered exon 6, and numeration for the downstream exons is updated from what was reported previously (ref. 11; Fig. 1,C). The 5′-UTR of RARβ5 mRNA is 237 nucleotides long and contains 2 upstream ORFs (uORFs; Fig. 1,B). We note in this respect that both RARβ2 and RARβ4 contain multiple uORFs (8, 11) that could play a role in tightly controlling translation efficiency. Differing from RARβ2, the first exon of RARβ5 does not contain any translation start codon, the 5′-most AUG of the RARβ5 transcript with the RARβ5 coding sequence is located at nucleotide 238 of RARβ5 mRNA (within the third exon of RARβ5) and corresponds to an internal methionine codon at amino acid 113 of the RARβ2 protein (Fig. 1, B and C). This AUG is within an appropriate nucleotide context for translation initiation (19) and would result in a protein of 336 amino acids with an estimated molecular mass of ∼37 kDa (Fig. 1,C). In vitro translation of RARβ5 expression vector confirmed that this AUG is a functional translation initiation codon (Fig. 1,D). It should be noted that in vitro translation generated multiple protein bands, which is consistent with a previous report and might not happen in the cells (11), probably because of lack of the whole UTR region in the expression vector and lack of natural chromatin environment. This RARβ5 protein product is identical to a truncated RARβ2 or RARβ4 protein reported previously (7, 11). The predicted amino acid sequence is given in Fig. 1 E.
AtRA Mediated Expression and Regulation of RARβ5.
Identification of RARβ5 raised a question as to whether its expression is mediated by atRA. To directly examine the presence of RARβ5 in comparison with RARβ2 in patients, we used the human breast cancer cells derived from the patients and being in early in vitro (<10) passages. In addition, we also examined the expression of both RARβ isoforms in established breast cancer cell lines and normal human mammary epithelial cells. RT-PCR analysis showed that all of the examined breast cancer cells expressed detectable RARβ5 mRNA, but its expression was differentially regulated by atRA treatment (Fig. 2,A). It was up-regulated by atRA in BCA-1, 3, and 4 cells, whereas BCA-8 showed a slight down-regulation of RARβ5 by atRA. Similarly, the level of RARβ2 was up-regulated by atRA in some cells (BCA-1, 3, 4, 8, 9, and 10), whereas in others it remained unaltered. However, no correlation between RARβ2 and RARβ5 expression could be established. Since in breast cancer, estrogen receptor (ER) plays a critical role in its response to various chemotherapeutic agents and to quantitate the expression of RARβ5 and RARβ2 mRNA, we did real-time PCR with HMEC, ER-positive breast cancer cell lines MCF-7 and T47D, and the ER-negative breast cancer cell line MDA-MB435 that expresses a high level of RARβ2 mRNA (11). Real-time PCR clearly showed that RARβ5 was preferentially up-regulated by atRA in MCF-7 cells, whereas RARβ2 was preferentially up-regulated by atRA in T47D cells (Fig. 2,B). ER-negative MDA-MB435 cells expressed a high level of RARβ2 but a low level of RARβ5 relative to HMEC cells. RARβ5 and β2 were consistently expressed at a low level in ER-positive MCF-7 and T47D cells relative to normal HMEC cells (Fig. 2,B). Because other retinoids function through a similar mechanism in the cells, it is reasonable to expect that other retinoids might also mediate RARβ5 expression. By using conventional RT-PCR, we confirmed that 4-HPR and 9-cisRA also differentially up-regulated RARβ5 expression in MCF-7, T47D, and BCA-3 cells. Fig. 2,C presents data showing real-time RT-PCR analysis of RARβ5 expression mediated by different RAR/RXR selective ligands in T47D cells. In addition to atRA, Am80 (RARβ/α selective ligand) and LGD1069 (RXR selective ligand) also significantly up-regulated RARβ5 expression, whereas 4-HPR (a weak RARγ ligand) and 9-cisRA (RAR/RXR ligand) had relatively low efficacy in mediating RARβ5 expression (Fig. 2 C).
RARβ5 Protein Expression in Correlation to Cellular Resistance to atRA.
We did Western blot to analyze RARβ protein expression in a panel of breast epithelial cells (normal HMEC; ER-positive MCF7 and T47D; ER-negative MDA-MB231, MDA-MB435, human breast carcinoma BCA-2 and BCA-8, MCF10A benign, and MCF10AT premalignant breast epithelial cells) with two RARβ polyclonal antibodies that were raised against amino acids at the COOH-terminal (a region common among human RARβ isoforms). A protein band with the expected molecular mass (∼37 kDa) was detected in HMEC, MDA-MB231, BCA-2, MCF10A, and MCF10AT cells by both antibodies (Fig. 3, A and B), suggesting this protein could be RARβ5. An RARβ2 protein band (∼55 kDa) was only detected from the in vitro translation product (Fig. 3 D). None of the cell lines expressed detectable RARβ2. In another experiment, we were not able to detect RARβ2 protein expression in all of the BCA (-1, -2…, -11) cell lines, but we detected RARβ2 protein in HMEC cells that were from a different source (Cambrex Bio Science Inc., Walkersville, MA)4 with the same antibody (C-19, Santa Cruz Biotechnology), indicating that RARβ2 protein expression could be cell-type specific. The ∼37 kDa protein detected in this study could be identical to the RARβ protein isoform (termed RARβ4, ∼40 kDa) identified previously in breast cancer cells (11), because same antibody was used for the detection, and the molecular size is also very close considering the 10% margin of error for our molecular mass standards. This RARβ protein isoform seemed to be preferentially expressed in ER-negative normal and cancerous breast epithelial cells, but it was not detectable in ER-positive breast cancer line MCF-7 and T47D.
To evaluate whether RARβ5 expression is associated with cellular resistance to retinoids, cell lines expressing different level of RARβ5 protein (Fig. 3, A and B) were selected to assess their sensitivity to retinoids with MTT assay. Immortalized benign MCF10A cells, which express high level of RARβ5, were resistant to both atRA and 4-HPR; ER-negative MDA-MB231 cells, which also express RARβ5 protein, showed resistance to atRA, but 4-HPR effectively inhibited the proliferation of MDA-MB231 cells. ER-positive MCF-7 and T47D cells in which RARβ5 protein expression was not detectable were sensitive to both atRA and 4-HPR (Fig. 3 E). In addition, MDA-MB435, MCF10AT, and BCA-2 cells, which express detectable RARβ5 protein, were relatively resistant to atRA (data now shown). These results suggest that RARβ5 might contribute to cellular resistance to atRA, which functions through receptor-dependent pathway. RARβ5 did not have much influence on the effect of 4-HPR, which functions through both receptor-dependent and independent pathway (20).
Sequence Analysis of 5′ Flanking Region of RARβ5.
Although the RARβ5 regulation pattern by atRA was more or less similar to that of RARβ2, in some cells such as BCA-2 and BCA-8, MCF-7 and T47D, the expression pattern was significantly different. Sequence alignment showed that the first exon of RARβ5 is far away (∼30 kb) from the P2 promoter, suggesting that RARβ5 and RARβ2 use different promoters. Therefore, we cloned and sequenced the 1-kb 5′ flanking region of RARβ5. Fig. 4 shows the 480-bp 5′ flanking sequence of RARβ5. The 5′ flanking sequence (−1000/−59) of RARβ5 was analyzed with several promoter identification programs including Proscan (21), Promoter 2.0 (22), BDGP: Neural Network Promoter Prediction (http://www.fruitfly.org/seq_tools/promoter.html), McPromoter (http://genes.mit.edu/McPromoter.html), PromoterInspector (http://www.genomatix.de/). None of these programs was able to predict this promoter. The region close to the putative transcription start site lacks the canonical TATA and CCAAT boxes, but a TATA-like box (TATAATT) is present 42 bp upstream of the transcription start site. Additional analysis with MatInspector (23) and TFSEARCH (http://www.cbrc.jp/research/db/TFSEARCH. html) identified DNA binding sites for AP-1, GATA, SRY, CEBP/β, NFAT, OCT1, and so forth at the region −500/+1 (Fig. 4). In addition, Promo (24) identified a binding site for RXRα at 312 bp upstream of the transcription start site. However, these are the potential binding sites, and tests for functionality need to be done to confirm their importance.
Transcription Start Site of RARβ5.
To determine the transcription start site of RARβ5, we did 5′-RACE analysis on mRNA from MDA-MB435 and MCF10A cells. The 5′ regions of both RARβ2 and RARβ5 were cloned from MDA-MB435 cells and sequenced. Because only a single transcription start site in the human RARβ2 gene has been defined (9), RARβ2 cDNA cloning and sequencing served as a good control for mapping the transcription start site of RARβ5 with 5′-RACE analysis. In total, six RARβ5-positive clones from MCF10A 5′-RACE products and four RARβ5-positive and two RARβ2-positive clones from MDA-MB435 were sequenced. Sequence analysis showed a single transcription start site of RARβ5 in both MCF10A and MDA-MB435 cells (Table 1). Interestingly, a single nucleotide (A) deletion close to transcription start site was found in five of the six clones from MCF10A cells and in all four clones from MDA-MB435 cells. However, as no deletion was found in the corresponding DNA region in MDA-MB435 cells, it seems that the deletion happened during the transcription process. Two transcription start sites [+1 and −11; +1 identifies the first nucleotide of the putative transcription start site based on GenBank sequence data (NM_000965)] of RARβ2 were identified in MDA-MB435 cells; it seems that RARβ2 has multiple transcription start sites in this cancer cell line.
RARβ5 Promoter Activity.
A series of RARβ5 promoter-luciferase reporter vectors were constructed. When these constructs were transfected into MCF-7 and T47D cells (which are ER-positive and responsive to atRA) and assayed for reporter gene activity, although differential promoter activity was observed in the two cell lines, the region −99/+33 consistently showed negligible or very low promoter activity in the two cell lines, suggesting either the existence of a negative regulatory element within this region or the presence of a strong activator in the region between −177 and −99 (Fig. 5, A and B). In MCF-7 cells, PGL3–1000/+33, 428/+33, −323/+33, −302/+33, and −177/+33 exhibited significant basal promoter activity (relative to the empty PGL3 Basic vector control). The atRA treatment additionally increased promoter activity by 2- to 5-fold in MCF-7 cells but only 2- to 3-fold in T47D cells, which was in agreement with real-time PCR results (Fig. 2 B). Deletion of region −302/−177 significantly decreased promoter activity induced by atRA in MCF-7 cells. On the basis of transfection assay data from both cell lines, it seems that the promoter region −302/−99 is the target region for atRA stimulation, whereas no RARE/RXRE was identified in this region. Therefore, either the target binding site has not been identified, or the stimulatory effect caused by atRA is an indirect effect.
Because atRA functions through receptor-dependent pathway, we hypothesized that expression of RARβ2 could affect RARβ5 promoter activity mediated by atRA. To test this hypothesis, RARβ2 expression vector (pcDNA3.1-RARβ2) was cotransfected with RARβ5 promoter constructs into MCF-7 and T47D cells. After transfection, the cells were incubated in the presence or absence of atRA for 24 hours, and luciferase assay was done for promoter activity. Surprisingly, cotransfection of the empty vector pcDNA3.1 also greatly increased RARβ5 promoter activity, whereas cotransfection of RARβ2 expression vector pcDNA3.1-RARβ2 did not cause a significant increase in promoter activity relative to the control (Fig. 5, C and D); however, in the presence of atRA, RARβ2 expression did significantly increase promoter activity compared with that of empty vector-transfected cells in MCF-7 cells (Fig. 5 C), suggesting that activation of RARβ5 promoter activity by atRA is receptor dependent.
We also examined RARβ5 promoter activity in the presence of other RAR/RXR selective ligands in both MCF-7 and T47D cells. As shown in Fig. 6, A and B, all of the tested RAR/RXR selective ligands differentially increased RARβ5 promoter activity, whereas Am80 showed the highest efficacy. These data also confirmed our RT-PCR analysis (Fig. 2 C).
Is RARβ5 a Unique Isoform among RARs?
The identification of RARβ5 raises the question as to whether corresponding isoform exists in the other two RARs. Sequence analysis revealed the possibility of existence of a corresponding isoform in both RARα and RARγ, because a similar internal start codon is present in both RARα and RARγ at similar positions. In addition, we also observed a similar protein band at the position of ∼37 kDa in MCF-7 and MDA-MB231 cells through Western blot analysis with RARα- and RARγ-specific antibodies that are not cross-reactive with each other and recognize the COOH-terminus of the corresponding RAR isoform. Therefore, we first 5′-RACE–analyzed the mRNA extracted from MCF-7 and MDA-MB231 cells with RARα-specific primers but failed to detect any expected cDNA band with a smaller size corresponding to the putative truncated RARα. Only a single RARα band of an expected size corresponding to full-length RARα was detected (data not shown). Because these receptor isoforms degrade quickly, we therefore hypothesized that the observed protein band of low molecular size for RARα and RARγ might be a fragment generated from protease cleavage. We used cell permeable proteasome inhibitor MG132, which can inhibit the degradation of RARα and RARγ (25). MG132 treatment completely blocked the generation of this fragment (Fig. 7), showing that these bands are products of protease cleavage of RARα and RARγ. The biochemical and biological properties of these RARα and RARγ fragments are not clear at present. When cells were treated with MG132, a corresponding equivalent band of RARβ5 was observed. Whether the expression of RARβ5 protein is because of inhibition of protein degradation or induction by MG132 treatment still needs additional in-depth studies.
Because a major level of regulatory control by retinoids is post-translational, we examined effect of atRA treatment on RARβ5 protein expression with and without proteasomal inhibition. Because proteasomal inhibitors are generally cytotoxic, a period of 8.5 hours was found not to trigger significant cell death in the cells examined. Cells were treated for 8 and 24 hours with 1 μmol/L atRA, and then treated with or without 40 μmol/L MG132 for the final 8.5 hours. We did not observe significant alteration of RARβ5 protein level in MDA-MB435 cells; however, no RARβ protein was detectable in T47D cells in either condition (data not shown). Nevertheless, MG132 effectively blocked atRA-induced RARα degradation in both cell lines as observed in MCF-7 cells (25) .
Genomic Structure of RARβ5 in Comparison to RARβ2.
On the basis of the identified RARβ5 cDNA sequence, the known RARβ2 sequence, and the published Human Genome Project Data (http://www.ncbi.nlm.nih.gov/genome/guide/human/), we were able to elucidate the complex organization of the RARβ5 and RARβ2 genes. BLAST search permitted us to align the first three exons of RARβ5 to a 70560-bp BAC clone RP11-421F9 (GenBank accession no. AC133141) mapped to chromosome 3p24. Similarly, the remaining exons were precisely aligned within another 189308-bp BAC clone RP11-659P16 (GenBank accession no. AC093416). Then the first exon of RARβ2 was aligned within the 198468-bp BAC clone RP11-733H11 (GenBank accession no. AC098477). Additional alignment of these three clones showed a 4145-bp overlap between clones RP11-733H11 and RP11-421F9 and a1862-bp overlap between clones RP11-421F9 and RP-659P16 (Fig 8,B), showing the continuity of the gene sequence in these BAC clones. An analysis of these three BAC clones revealed that the RARβ5 gene spans over 130-kb of DNA, whereas the RARβ2 gene spans over 160 kb of DNA. All of the splice junctions conform to the GT/AG rule for splice donor and acceptor sites (ref. 26; Fig. 8,A). Fig. 8 B summarizes our analysis on the genomic structures of hRARβ5 and hRARβ2 genes, and a new numeration for their exons is proposed herewith.
The major biological effect and gene expression induced by retinoids are believed to be mediated by nuclear receptors RARs/RXRs. Because RARs and RXRs are primary effectors of retinoid signaling, they themselves seem to be targets for disruption in tumorigenesis. RARβ has been extensively studied in human carcinomas, and several studies have suggested that it might play a role in tumor suppression (27, 28, 29). Therefore, RARβ has been considered a target molecule for retinoids in chemoprevention and therapeutic studies.
In this study, we identified RARβ5, a novel RARβ isoform directed by a distinct promoter P3. We also provided the first evidence showing that RARβ5 protein is identical to RARβ4 or RARβ′, a previously identified truncated RARβ protein (7, 11). In 1999, Sommer et al. (11) identified a 40 kDa RARβ protein isoform, which was interpreted as RARβ4. This RARβ protein isoform was found to be elevated in human breast tumor cells, especially in cytoplasm relative to RARβ2 protein. In 2002, Chen et al. (7) showed an antagonistic role for this RARβ protein isoform in signaling by retinoic acid and termed it RARβ′, and its expression was interpreted as “leaky scanning.” In the same year, another group (15) showed that the expression of this protein isoform is associated with cellular resistance in response to retinoids. They also interpreted it as RARβ4 and tried to link the protein expression to RARβ4 mRNA expression. These data seemed correct, but the interpretation on the generation of that RARβ (RARβ4 or RARβ′) protein isoform seems questionable. There has been no evidence showing that the protein isoform is translated from endogenous RARβ2 or RARβ4 transcripts. The interpretation on the generation of that RARβ protein isoform was based on transfection and in vitro translation experiments, in which the expression vectors generally do not contain full-length 5′ and 3′UTR, and the reporter genes are not in a natural chromatin environment. In addition, the existence of multiple uORFs in the long 5′UTR region of RARβ2 and RARβ4 could also inhibit leaky scanning (30). Some cells such as MCF10A series of cell lines do not express detectable RARβ4 mRNA,5 the same protein isoform could only be from RARβ5 transcript in these cells. The identification of RARβ5 in breast epithelial cells suggests that RARβ′ is the primary translation product from ORF of RARβ5 transcript. Should leaky scanning occur with RARβ2 or RARβ4, it might result in RARβ′ at a very low level (30, 31). Moreover, the existence of multiple uORFs and the long leader sequence in RARβ isoform mRNAs could be a signal that their translations are under tight control.
In most of the previous studies, measurements of RARβ expression are preferentially made at the mRNA levels (25), leading to certain level of complexity in understanding its function. All the more, in most of breast cancer cell lines, RARβ2 protein was not detected by Western blotting, although its mRNA was detectable (Fig. 2). We have carried out experiments to address these concerns to a certain extent and to study the expression of RARβ5 both at the RNA and protein levels in various patient-derived primary breast cancer cells, established breast cancer cell lines, and immortalized benign MCF10A, premalignant MCF10AT cell lines, and normal human breast epithelial cells. At the mRNA level, RARβ5 is expressed in normal human breast epithelial cells as well as in benign, premalignant, and tumor cell lines. In the presence of atRA, the level of RARβ2 mRNA is preferentially elevated in contrast to RARβ5 in T47D cells. At the protein level, we failed to detect endogenous RARβ2 protein by Western blotting but did detect a corresponding RARβ5 band in MDA-MB231 and HMEC cells. In agreement with our studies, Tanaka et al. (25) also failed to detect RARβ2 protein under their experimental conditions. Hence, either RARβ2 protein is not stable, or its expression is too low to be detected in these cells.
RARβ5 identification also defines a new type of RARβ isoform, which is under the control of a distinct promoter P3, and the protein lacks the A, B, and part of the C domain (the first zinc finger) of other RARβ isoforms. The loss of DNA binding ability while retaining the capability to form heterodimers with RXR makes RARβ5 act as a trans-dominant–negative regulator of RARβ function (7). We note in this respect that similarly truncated RARα isoforms lacking all of the sequence located NH2-terminal to the second zinc finger have been identified previously (32). Most analogous to RARβ5 is the progesterone C mRNA that encodes a NH2-terminally truncated progesterone receptor (33, 34). RARβ5 protein expression, localization, and function were characterized previously as a truncated RARβ protein isoform (7, 11, 15). Unlike some other reported dominant-negative nuclear receptors (35, 36), RARβ5 does not bind cis-acting DNA elements and therefore cannot directly inactivate gene transcription. RARβ5 likely represses by stoichiometric competition, away from the RARE, against other transcription factors within the cell (e.g., RARα, RARβ, and RARγ) for transcription cofactors (7). Although RARβ4 protein (identical to RARβ5 protein) was reported to be elevated in breast cancer cells (11, 15), our data show that both RARβ5 mRNA and protein are expressed in normal HMEC cells, indicating that RARβ5 is not a tumor-specific isoform; it could be a regulatory factor for RARβ target genes in both normal and tumor breast epithelial cells. We could not detect RARβ5 protein in ER-positive breast cell lines that are sensitive to retinoids, whereas it can be detected in ER- negative breast cancer cells and normal breast epithelial cells that are relatively resistant to retinoids, indicating that this isoform may contribute to cellular resistance to retinoids. In the metastatic atRA-resistant M-4A4 cell line derived from MDA-MB435 cells, RARβ4 protein (identical to RARβ5) was elevated in comparison with the isogenic nonmetastatic NM-2C5 cell line, and its protein expression was also up-regulated by atRA after 4- and 6-day treatment (15), which is consistent with our RT-PCR and MTT data and in agreement with the conclusion that it plays a negative role in RARβ2 function (7).
Analysis of the RARβ5 5′ flanking region by computer program failed to predict the P3 promoter, indicating that P3 is not a typical promoter. The functionality of the TATA-like box 42-bp upstream of transcription start site remains unclear. In this respect, another noncanonical TATA box (TATATTA) has been reported in the P2 promoter of RARβ (9). However, cloning and transfection studies of the RARβ5 5′ flanking region confirmed the presence of P3 promoter. The atRA target promoter region (−302/−99) lacks any canonical RXRE/RARE elements. The magnitude of activation of the RARβ5 promoter by atRA seemed to be cell-type specific. Cotransfection of empty vector pcDNA3.1 resulted in a significant increase in reporter gene activity, the reason is currently unknown; transfection experiment can generate artifacts, and a control (empty) vector must be included for comparison to see the function of the transfected gene. The effect of atRA seems to be at least partially RARβ2 dependent, whereas RARβ2 overexpression itself might not have a significant effect on RARβ5 promoter activity in the absence of ligand-atRA. Other RAR/RXR selective retinoids also differentially increased RARβ5 promoter activity, indicating that both RARs and RXRs can be involved in the RARβ5 transcriptional activation. Because only two promoters have been previously identified in all of the RAR genes (11), the identification of this promoter has biological significance. Both RARβ2 and RARβ5 can be transactivated by atRA in the same cells, whereas their functions seem to be different, revealing a mechanism of fine-tuning atRA-induced transcription.
The corresponding isoform of RARβ5 in RARα and RARγ gene seems not to be present, although fragments cleaved from RARα and RARγ protein were detected. Whether the fragments are functional or not and their biochemical properties still need to be determined. It seems RARβ5 isoform is unique among all of the RARs and not a cleaved product from other RARβ isoforms, which suggest that RARβ gene might have different function from other RARs. The expression and regulation of RARβ protein is an important issue for functional research of this receptor. We did not observe significant regulation of RARβ5 protein by atRA after 24 hours treatment with or without proteasomal inhibition, suggesting that the translational regulation is independent of transcriptional regulation under the experimental condition. The RARβ post-translational regulation by retinoids will be addressed in depth in the future studies.
In summary, we have identified a novel, unique RARβ isoform (RARβ5) and mapped its promoter region. We also initially characterized its expression and transcriptional regulation in normal and cancerous breast epithelial cells. RARβ5 identification reveals an additional layer of complexity to retinoid signaling, and this isoform may serve as a potential target of retinoids in breast cancer prevention and therapy studies. Future study on RARβ function should include the analysis of RARβ5 isoform in both normal and tumor cells and its response to retinoids. Effective inhibition of RARβ5 might be necessary for the prevention and treatment of breast and other cancers by retinoids.
Grant support: The United States Army Breast Cancer Research Program DAMD17-99-9221 (K. Christov) and the Illinois Department of Public Health Penny Sevens Breast and Cervical Cancer Research Fund (X. Peng).
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.
Note: Supplementary data for this article may be found at Cancer Research Online at http://cancerres.aacrjournals.org. The sequences reported in this article have been deposited in the GenBank database (accession nos. AY501390-AY501391).
Requests for reprints: Konstantin Christov, Department of Surgical Oncology, University of Illinois at Chicago, 840 South Wood Street (M/C 820), Chicago, IL 60612. Phone: (312) 996-5347; Fax: (312) 996-9365; E-mail: firstname.lastname@example.org
X. Peng, D. Yun, K. Christov, unpublished data.
X. Peng, R. G. Mehta, D. A. Tonetti, K. Christov, unpublished data.
|hRARβ5 transcriptional start site sequence .||.||.|
|hRARβ5 transcriptional start site sequence .||.||.|
We thank Dr. Karen Swisshelm for generously providing RARβ2 expression plasmid, Dr. Zhenyu Li (Department of Pharmacology, University of Illinois at Chicago, Chicago, IL) for expert technical assistance in making DNA constructs, Dr. Koichi Shudo (ITSUU Laboratory) for providing Am80 RARβ/α selective retinoid, Marcia Dawson (Burnham Institute, La Jolla, CA) for providing RARβ agonists, and Scott Kenndy for editorial assistance.