Retinoic acid (RA) displays pronounced anticarcinogenic activities in several types of cancer. Whereas the mechanisms that underlie this activity remain incompletely understood, tumor suppression by RA is believed to emanate primarily from its ability to regulate transcription of multiple target genes. Here, we investigated molecular events through which RA inhibits the growth of MCF-7 mammary carcinoma cells, focusing on the involvement of the two proteins that mediate transcriptional activation by RA, the nuclear hormone receptor retinoic acid receptor (RAR) and the cellular retinoic acid-binding protein (CRABP) II, in this process. RA treatment of MCF-7 cells did not affect cell cycle distribution but triggered pronounced apoptosis. Accordingly, expression array analyses revealed that RA induces the expression of several proapoptotic genes, including caspase 7 and caspase 9. Whereas caspase 7 is an indirect responder to RA signaling, caspase 9 is a novel direct target for RAR, and it harbors a functional retinoic acid response element in its second intron. In agreement with the known role of CRABP-II in enhancing the transcriptional activity of RAR, the binding protein augmented RA-induced up-regulation of caspase 9, cooperated with RA in activating both caspase 7 and 9, and amplified the ability of RA to trigger apoptosis. Surprisingly, the data indicate that CRABP-II also displays proapoptotic activities on its own. Specifically, overexpression of CRABP-II, in the absence of RA, up-regulated the expression of Apaf1 and triggered caspase 7 and caspase 9 cleavage. These observations suggest that, in addition to its known role in direct delivery of RA to RAR, CRABP-II may have an additional, RA-independent, function.
- retinoic acid
Retinoic acid (RA), the active metabolite of vitamin A, regulates multiple biological processes, including cell proliferation, differentiation, and death, and thus plays critical roles in embryonic development and in growth and remodeling of adult tissues. Natural and synthetic RA derivatives, collectively known as retinoids, are also potent inhibitors of cancer cell growth and have been shown to be efficacious in therapy and prevention of various types of cancer ( 1, 2). The pathways by which RA inhibits the growth of carcinoma cells seem to vary between cell types. It was thus reported that RA induces differentiation in embryonic tetratocarcinoma F9 cells ( 3), G1/G0 growth arrest and myeloid differentiation in HL-60 cells ( 4, 5), and postmaturation apoptosis in NB4 acute promyelocytic leukemia cells ( 6). In mammary carcinoma cell lines, RA was shown to induce growth inhibition by triggering either cell cycle arrest or apoptosis, or both ( 7– 9).
The biological activities of RA are believed to be exerted primarily through the ability of this hormone to regulate gene expression, an activity that is mediated by members of the superfamily of nuclear hormone receptors termed retinoic acid receptors (RARs; ref. 10). Like other type II nuclear receptors, RARs function as heterodimers with the retinoid X receptor (RXR). These heterodimers associate with specific DNA sequences (retinoic acid response elements, RARE) composed of two direct repeats of the consensus sequence PuG(G/T)TCA, separated by either 2 bp (DR-2) or 5 bp (DR-5; refs. 10, 11). RXR-RAR heterodimers thus bind in regulatory regions of their target genes and enhance transcriptional rates on binding of RA ( 12).
Only limited information is currently available on the identity of immediate RAR target genes that mediate the anticarcinogenic activities of RA. Three such genes were reported to be involved in RA-induced apoptosis and differentiation in NB4 promyelocytic leukemia cells (i.e., ubiquitin-activating enzyme E1–like protein, CCAAT/enhancer binding protein ε, and tumor necrosis factor (TNF)–related apoptosis-inducing ligand; refs. 6, 13, 14). Studies of breast cancer cells suggest that RA-induced apoptosis is associated with down-regulation of Bcl-2 and survivin ( 15, 16) and with up-regulation of the tumor suppressor gene PDCD4 and of SOX9, a member of the high mobility group box family of transcription factors ( 17, 18). The mechanisms underlying the RA-responsiveness of these genes and whether any of these comprise direct RAR targets remain to be clarified.
In addition to RAR, two other proteins, termed cellular retinoic acid-binding proteins (CRABP-I and CRABP-II), bind RA with high affinity and specificity ( 19, 20). CRABPs are small (∼14 kDa) soluble proteins that are members of the family of intracellular lipid binding proteins. Whereas it is generally believed that CRABPs function to solubilize and protect RA in the aqueous space of the cytosol, accumulating evidence suggests that they also play more specific roles in modulating signaling by RA. In regard to the biological functions of CRABP-II, we recently showed that this protein transports RA from the cytosol to the nucleus where it directly associates with RAR. We further showed that the resulting CRABP-II-RAR complex mediates “channeling” of RA to the receptor, thereby facilitating its ligation and enhancing its transcriptional activity ( 19, 21, 22). Additional observations showed that, as a result of the ability of CRABP-II to augment the transcriptional activity of RAR, overexpression of the protein in mammary carcinoma cells dramatically sensitizes them to RA-induced growth inhibition ( 21). Correspondingly, CRABP-II was found to inhibit mammary tumor growth both in a xenograft model and in the transgenic breast cancer mouse model mouse mammary tumor virus-neu (MMTV-neu), in which mammary tumors develop spontaneously and progress under immune surveillance ( 23). Inhibition of breast tumor development in MMTV-neu mice treated with CRABP-II was found to stem mainly from increased apoptotic rates ( 23).
The present study was undertaken to obtain further insights into the mechanisms by which RA induces growth inhibition in the mammary carcinoma MCF-7 cells and the roles of RAR and CRABP-II in this process. We show that treatment of MCF-7 cells with RA has little effect on cell cycle distribution but results in up-regulation of expression of several proapoptotic genes, and in a marked induction of apoptosis. One of the RA-induced proapoptotic genes, caspase 9, is shown to be a novel direct target for RAR and to harbor a functional RARE in its second intron. We show that CRABP-II cooperates with RAR in mediating the induction of caspase-9 expression, and that the binding protein enhances the RA-initiated apoptotic response of these cells. Additional observations indicate that, in the absence of RA, CRABP-II up-regulates the mRNA expression level of apoptotic protease activating factor 1 (Apaf1) and triggers cleavage of both caspase 7 and caspase 9. These observations imply that besides its cooperation with RAR in mediating cellular responses to RA, CRABP-II possesses additional, RA-independent, functions.
Materials and Methods
Reagents. Antibodies against caspases were purchased from Cell Signaling (Beverly, MA). Antibodies against mCRABP-II (5CRA3B3) were a gift from Pierre Chambon (IGMCB, Strasbourg, France). Anti-mouse and anti-rabbit immunoglobulin horseradish peroxidase–conjugated antibodies were from Amersham (Arlington Heights, IL).
Cells. MCF-7 cells were purchased from the American Type Culture Collection (Manassas, VA). Cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) or 10% charcoal-treated FBS.
Vectors. The putative RARE in the caspase 9 promoter and the intron DR-2 were separately cloned into the pGL3 luciferase reporter vector (Promega) using the enzyme sites KpnI and XhoI.
Proteins. Recombinant histidine-tagged RARα and RXRα lacking the amino terminal A/B domains (RARαΔAB and RXRαΔAB) were expressed in E. coli and purified as previously described ( 24).
Viruses. Adenoviruses were made by the Gene Transfer Vector Core center at the University of Iowa and stored as described ( 23).
Transactivation assays. MCF-7 cells were transfected with the indicated luciferase reporter plasmid and pCH110 (internal standard) using Fugene (Roche). Following an overnight incubation, cells were treated with RA for 24 hours and lysed. Luciferase expression was assayed using the luciferase assay system (Promega) and corrected for transfection efficiency by the activity of β-galactosidase, which was measured by standard procedures.
Flow cytometry. Cells were seeded in 60 mm plates in DMEM supplemented with 10% charcoal-treated FBS. Cells were transduced with Ad-CRABP-II or Ad-0 virus [multiplicity of infection (MOI) of 500] and grown for 18 hours before addition of RA. RA was replenished every 48 hours. Cells were scraped and collected, washed with cold PBS, lysed in hypotonic buffer (1 mg/L sodium citrate, 0.1% Triton X-100), and propidium iodide (1 mg/mL) was added. Samples were run on a Becton Dickinson FACS Calibur and the results analyzed with CellQuest version 3.3.
Western blotting. Cells were treated as indicated, scraped into the media, and collected. Pellets were washed with cold PBS and lysed in lysis buffer [10 mmol/L potassium phosphate (pH 7.5), 0.5% (w/v) Triton X-100, 10 μg/mL leupeptin, and 10 μg/mL aprotinin]. Protein concentrations in cell lysates were determined by Bradford assay (Bio-Rad) and 30 μg protein loaded in each lane. Proteins were resolved by 12% SDS-PAGE and visualized by Western blots using appropriate antibodies.
Quantitative real-time PCR. RNA was extracted using RNeasy (Qiagen, Valencia, CA). cDNA was generated using Gene Amp RNA PCR (Applied Biosystems, Foster City, CA). Quantitative real-time PCR was done in replicates using TaqMan chemistry and Assays on Demand probes (Applied Biosystems) for caspase 9 (Hs0015426_m1), caspase 7 (Hs00169152_m1), and Apaf1 (Hs00559421_m1). 18s rRNA (4319413E-0312010) was used as a loading control. Analyses were carried out using the relative standard curve method (Applied Biosystems Technical Bulletin no. 2).
Chromatin immunoprecipitation. Nearly confluent MCF-7 cells were treated with 9cRA (1 μmol/L, 20 minutes). Proteins were cross-linked to DNA (1% formaldehyde, 10 minutes). Cells were washed with PBS, scraped, collected, lysed [1% SDS, 10 mmol/L EDTA, 50 mmol/L Tris (pH 7.9), 1 mmol/L DTT, protease inhibitors (Roche)], and incubated on ice for 45 minutes. Samples were sonicated thrice, and chromatin precleared with protein A beads for 2 hours. Antibodies (3.5 μg) were added and mixtures incubated overnight at 4°C. Protein A beads were added and mixed (2 hours, 4°C). Beads were washed twice with low-salt buffer (150 mmol/L NaCl, 0.5% deoxycholate, 0.1% Nonidet P-40, 1 mmol/L EDTA, 50 mmol/L Tris-HCl), twice with high salt buffer (low salt buffer + 500 mmol/L NaCl), and twice with Tris-EDTA buffer. Cross-link was then reversed (100 mmol/L NaHCO3, 1% SDS, overnight 65°C), proteins digested with proteinase K (1 hour), and DNA purified using the nucleotide extraction kit (Qiagen). The DR-2–containing region of the caspase 9 gene was amplified by PCR using the primers 5′-tgccatgtctaccacggcacagg-3′ (forward) and 5′-tgcaccatgcctggctagtt-3′ (reverse).
Affymetrix expression array. MCF-7 cells were transduced with Ad-0 virus at an MOI of 500 for 18 hours, followed by a 4-hour treatment with 50 nmol/L RA or vehicle. Total RNA was prepared from triplicate cell cultures using RNeasy (Qiagen). Sample processing and analyses, including cDNA synthesis, cRNA synthesis, and labeling, and array applications were done at the University of Rochester Functional Genomics Center. RNA quality was assessed using the Agilent Bioanalyzer 2100 and spectrophotometric analysis before cDNA synthesis. Five micrograms of total RNA from each sample were used to generate cDNA, and 1 μg product was used in an in vitro transcription reaction containing biotinylated UTP and CTP. Twenty micrograms of full-length cRNA were fragmented and analyzed on Affymetrix U133A/B high-density oligonucleotide array. Arrays were hybridized, stained, and washed using the Affymetrix fluidics module. Detection and quantitation of target hybridization were done with a GeneArray Scanner 3000 (Affymetrix, Santa Clara, CA). Iobion's Gene traffic was used for Robust Multi-Chip Analysis and clustering of genes with similar activity by summary function. An unpaired t test giving P values of <0.1 were defined as significantly changed.
Electrophoretic mobility shift assay. DNA probes were generated by restriction enzyme digestion of the luciferase reporter construct containing the DR-2 region of the caspase 9 intron 2. Oligonucleotides were end-labeled with [32P]dCTP by filling in with Klenow fragments, and free nucleotides were removed with the Qiagen nucleotide removal kit. DNA (10,000 cpm, 16 fmol) was incubated with 1.25 pmol of receptor for 20 minutes. Cold competitor DNA was composed of 45 bp encompassing the DR-2 intron element, or a corresponding mutant DNA (DR-2, TCTGCCGGTCTGCC). Protein-DNA complexes were resolved by electrophoresis on 5% polyacrylamide gels and visualized by autoradiography.
Retinoic acid induces apoptosis in MCF-7 cells. MCF-7 cells were treated with varying concentrations of RA for 5 days and analyzed by flow cytometry. The analyses ( Fig. 1 ) showed that RA triggered DNA fragmentation, reflected by a RA dose-responsive increase of the fraction of cells in sub-G1 population. A 3-fold increase in apoptosis was observed at the highest RA concentration used (0.6 μmol/L, Fig. 1B). In contrast, RA had little effect on the distribution of cells between different cell cycle phases ( Fig. 1C). Hence, under this experimental regimen, inhibition of MCF-7 cell growth by RA is mediated primarily by induction of apoptosis.
Retinoic acid up-regulates expression of proapoptotic genes in MCF-7 cells. We thus sought to identify target genes that may mediate the proapoptotic activity of RA. To this end, we carried out an expression array analysis. MCF-7 cells were treated with 50 nmol/L RA or vehicle for 4 hours and total RNA was isolated. Probes were generated, hybridized to Affymetrix human U133 A/B arrays, which allows for monitoring the expression of more than 40,000 genes and ESTs, and differences in gene expression profiles between RA-treated and untreated cells were analyzed. Three replicates were analyzed for each condition and changes in genes that were observed in all replicates and that displayed P values of <0.1 (unpaired t test) were considered to be significant. A total of 825 genes were found to be up-regulated by at least 1.23-fold on RA treatment, with the highest fold (×18.25) observed for the well-characterized RA-responsive gene Cyp26a ( 25). RA-induced genes were clustered by similar biological functions using the GeneTraffic software system (Iobion). One such cluster was found to encompass several genes that are known to be involved in apoptotic responses ( Table 1 ). The observations that the expression of several such genes was up-regulated by RA and the relatively small magnitude of the responses suggest that induction of apoptosis by RA in MCF-7 cells may result from concerted, cumulative effects on multiple pathways. Notably, however, two genes in this cluster encode for proteins that are known to directly mediate apoptosis (i.e., caspase 7 and caspase 9). Regulation of expression of RA-responsive genes may be exerted directly (i.e., mediated by an RARE). Alternatively, responses may reflect secondary events involving RAR control of immediate target genes, which, in turn, are involved in downstream events leading to the observed modulation. Hence, an important question that arises is which of the RA-controlled proapoptotic genes in MCF-7 cells comprise direct targets for RAR. This question is particularly pertinent considering the present paucity of information on the mechanisms by which RA exerts its anticarcinogenic activities. To begin to address this issue, we examined whether either caspase 7 or caspase 9 is under direct RAR control.
Caspase 9 is a direct target for retinoic acid receptor, caspase 7 is not. To validate the array data, quantitative real-time PCR was carried out. MCF-7 cells were treated for 4 hours with RA (50 nmol/L) or vehicle, and quantitative real-time PCR was used to compare the expression levels of mRNA for caspase 9 and caspase 7 in RA-treated versus untreated cells ( Fig. 2 ). In good agreement with the Affymetrix array data ( Table 1), the expression levels of mRNA for both caspase 9 and caspase 7 increased by about 60% in response to RA treatment. To determine whether these genes are direct targets for transcriptional regulation by RA, the effect of the protein synthesis inhibitor cycloheximide on their induction by RA was studied. Cycloheximide treatment will abolish secondary events that require de novo protein synthesis but will not affect direct transcriptional responses. The analyses showed that inhibition of protein synthesis completely abolished the RA response of caspase-7 expression ( Fig. 2A). In contrast, caspase 9 mRNA was up-regulated by RA regardless of the presence of cycloheximide ( Fig. 2B). Hence, whereas the effect of RA on caspase-7 expression is a secondary response, caspase 9 is likely to be a direct target for RAR signaling.
A functional retinoic acid response element is present in the second intron of the caspase 9 gene. Consensus RAREs are composed of two direct repeats of the sequence PuG(G/T)TCA spaced by either 2 bp (DR-2) or 5 bp (DR-5). We used two programs, TransFac ( 26)1 and TESS ( 27),2 to screen the human caspase 9 gene and adjacent regulatory sequences for potential RAREs. Within a stretch of 8 kb upstream of the caspase 9 start site (1p36.3), a potential RARE composed of the noncanonical DR-2 sequence AGGTCAgcAGTTCG was found at position −1690. This element, along with 38 bp of flanking sequences on both sides, was cloned into a luciferase reporter vector and its functionality was examined by transactivation assays carried out in MCF-7 cells. The expression of the reporter did not respond to RA ( Fig. 2C), suggesting that the element does not function as an RARE. An additional potential RARE, composed of the consensus DR-2 sequence AGGTCAggAGTTCA, was found in the second intron of the caspase 9 gene, 9,461 bp downstream of the start site. This RARE and 45 bp of flanking sequences on each side were cloned into a luciferase reporter, which was used in transactivation assays. The data ( Fig. 2D) showed a dose-responsive activation of reporter expression by RA, suggesting that the element indeed comprises a functional RARE. To verify that the element can bind RAR-RXR heterodimers, electrophoretic mobility shift assays were carried out. RARα and RXRα lacking their amino terminal A/B domains (RARαΔAB and RXRαΔAB) were expressed in E. coli, purified, and examined for binding to a 90 bp oligonucleotide harboring the putative response element and its flanking sequences. The data ( Fig. 2E) showed that RAR-RXR heterodimers tightly and specifically associate with the DR-2 element of the second intron of caspase 9. Finally, to examine whether the element is occupied by RXR-RAR heterodimers in a living cell, chromatin immunoprecipitation assays were carried out. Proteins were cross-linked to chromatin in MCF-7 cells and immunoprecipitated using antibodies for RAR or RXR, or nonspecific immunoglobulin G (IgG). Precipitates were sonicated, the cross-link reversed, DNA isolated, and a 250 bp region surrounding the DR-2 in the second intron of caspase 9 amplified by PCR. The data ( Fig. 2F) showed that antibodies against either RAR or RXR specifically precipitated the intron DR-2 sequence, demonstrating that the element is occupied by the heterodimers in cells. Taken together, the observations establish that caspase 9 is a direct target for RAR signaling, and that the RARE responsible for this response is likely to be a DR-2 element located in the second intron of the gene.
Cellular retinoic acid-binding protein II augments the induction of caspase-9 expression by retinoic acid. We previously showed that the RA-binding protein CRABP-II facilitates the delivery of RA to RAR, thereby enhancing the transcriptional activity of the receptor, at least in the context of a reporter construct driven by a consensus RARE. These observations suggest that RA-induced up-regulation of direct RAR target genes will be augmented by CRABP-II. It may also be suggested that, due to dilution of the effect in subsequent steps, RA responses mediated by secondary events may be less amenable to modulation by the binding protein. The finding that caspase 9 comprises a direct target for RAR in MCF-7 cells whereas caspase 7 is an indirect responder provides an opportunity to examine the activity of CRABP-II in the context of an endogenous gene driven by a native promoter, and to test the effect of the binding protein on two genes that are differentially regulated by RA. We thus investigated the effect of overexpression of CRABP-II in MCF-7 cells on RA-induced up-regulation of the two caspases. CRABP-II was ectopically expressed using an adenovirus harboring CRABP-II cDNA ( Fig. 3A ). Initial experiments verified that viral infection resulted in a pronounced CRABP-II overexpression, which could be observed 24 hours following infection and maintained for longer than 6 days ( Fig. 3B). MCF-7 cells were infected with Ad-CRABP-II for 24 hours, treated with RA for 4 hours, and the expression of caspase-7 and caspase-9 mRNA was examined by quantitative real-time PCR. The data ( Fig. 3C) showed that CRABP-II overexpression augmented the RA-induced up-regulation of expression of the direct RAR target gene caspase 9 by about 20%, a modest but clearly significant effect which about doubled the RA induction of the gene in this experiment. CRABP-II expression had little effect on the RAR indirect target, caspase 7 ( Fig. 3D).
Retinoic acid and cellular retinoic acid-binding protein II induce cleavage of caspases 7 and 9. One of the hallmarks of the apoptotic response is the activation of caspases which results from cleavage of the inactive zymogens to their active forms. To directly examine whether up-regulation of the expression levels of caspases 7 and 9 by RA ( Fig. 2) is accompanied by their activation, cells were treated with RA and cell lysates were probed for the presence of the activated forms of the proteases using antibodies directed against cleaved caspase 7 or caspase 9. The cleaved product of caspase 7 was undetectable in untreated MCF-7 cells, and its level was slightly increased in response to RA treatment ( Fig. 4A ). The level of cleaved caspase 9 markedly increased on RA treatment and did so in a dose-responsive fashion ( Fig. 4B). Strikingly, overexpression of CRABP-II induced cleavage of both caspase 7 and caspase 9 even in the absence of RA. In the presence of RA, CRABP-II markedly augmented the ability of RA to trigger caspase cleavage ( Fig. 4A-C). Especially notable is that, whereas RA had only a small effect on the cleavage of caspase 7 on its own, it dramatically enhanced the activation of this caspase on CRABP-II overexpression, and did so in a dose-dependent manner.
Cellular retinoic acid-binding protein II up-regulates the expression of Apaf1 in the absence of retinoic acid. The observation that CRABP-II expression leads to activation of both caspase 7 and caspase 9 even in the absence of RA suggests that, in addition to its ability to augment the RA-induced activation of RAR, CRABP-II may display a RA-independent activity. These findings further suggest that this activity allows the binding protein to modulate either the expression level or the activity of component(s) in the pathway that leads to caspase cleavage. Activation of caspases on apoptotic signaling is triggered by the release of mitochondrial cytochrome c, which, in turn, binds the cytosolic adaptor protein Apaf1 to form a multiprotein complex called the apoptosome. The apoptosome, with Apaf1 at its core, cleaves procaspase 9, which then activates executioner caspases, caspase 7, and caspase 3, to propagate the apoptotic response ( 28). Considering the central role of Apaf1 in caspase activation, we examined whether CRABP-II may modulate Apaf1 expression. MCF-7 cells were infected with either Ad-0 or Ad-CRABP-II for 24 hours, and quantitative real-time PCR was used to measure Apaf1 mRNA levels. The data ( Fig. 5A ) showed that overexpression of CRABP-II per se resulted in a 2-fold increase in the level of Apaf1 mRNA. It could be argued that this effect may reflect a CRABP-II–enhanced activity of residual RA present in the cells rather than a RA-independent activity of the binding protein. This is unlikely both because the cells were depleted of vitamin A and RA stores by culturing in media containing charcoal-treated serum and because MCF-7 cells are severely impaired in their ability to synthesize RA ( 29). Nevertheless, we examined whether Apaf1 comprises a RA-responsive gene. MCF-7 cells were treated with 50 nmol/L RA for 24 hours and Apaf1 mRNA level was measured by quantitative real-time PCR. RA treatment had no effect on Apaf1 mRNA expression ( Fig. 5B), demonstrating that it is not a target for RA signaling and supporting the conclusion that CRABP-II modulates the expression of this gene by a RA-independent mechanism.
Cellular retinoic acid-binding protein II augments retinoic acid–induced apoptosis in MCF-7 cells. The finding that CRABP-II augments the RA-induced expression of caspase 9, and that it cooperates with RA in enhancing the activation of both caspase 7 and caspase 9, suggests that the binding protein is closely involved in apoptosis initiated in MCF-7 cells on RA treatment. To directly examine this possibility, flow cytometry was used to monitor the effect of CRABP-II expression on RA-induced DNA fragmentation. MCF-7 cells were infected with Ad-CRABP-II or Ad-0 for 18 hours and then treated with varying concentrations of RA for 7 days. Nuclei were isolated, stained with propidium iodide, and the fraction of cells containing fragmented DNA (sub-G1) determined. The data ( Fig. 5C) showed that overexpression of CRABP-II indeed markedly enhanced RA-induced apoptosis over a wide range of ligand concentrations.
RA displays pronounced anticarcinogenic activities in several types of cancer but the exact mechanisms that underlie these activities remain incompletely understood. The present work was undertaken to obtain insights into molecular events through which RA inhibits the growth of MCF-7 mammary carcinoma cells. Of special interest with regard to this issue is the involvement of the two proteins that mediate the transcriptional activity of RA, the nuclear hormone receptor RAR and the RA binding protein CRABP-II, in the antiproliferative activity of their ligand. The data presented above show that a 5- to 7-day treatment of MCF-7 with RA induced pronounced apoptosis but had little effect on cell cycle distribution ( Fig. 1). These findings are consistent with our previous observations that breast tumor suppression by RA and CRABP-II in the MMTV-neu mouse cancer model results from induction of apoptosis and not from effects on cell cycle progression ( 23).
Correspondingly, examination of the effect of RA on the gene expression profile of MCF-7 cells revealed that several proapoptotic genes are up-regulated on a 4-hour exposure to RA ( Table 1). At this short period of treatment, it can be expected that at least some of the responding genes comprise direct targets for RAR. Additionally, due to its ability to enhance the transcriptional activity of RAR, it may be predicted that CRABP-II will augment the RA response of genes that are under direct RAR control. We show that caspase 9 is indeed a direct target for RAR ( Fig. 2) and that the induction increases on CRABP-II overexpression ( Fig. 3). These observations comprise the first demonstration that CRABP-II enhances the RA-induced, RAR-mediated activation of an endogenous gene.
The conclusion that caspase 9 is under the direct control of RAR prompted us to search for the RARE through which the regulation is exerted. We show that the second intron of the caspase 9 gene harbors a DR-2 RARE, that the element specifically binds RXR-RAR heterodimers, that it mediates RA-induced transcriptional activation of a luciferase reporter, and that it is occupied by RXR-RAR heterodimers in MCF-7 cells ( Fig. 2). The identification of the caspase-9 RARE in an intron sequence contributes to the growing body of evidence that regulatory elements are not confined to upstream promoter regions of target genes. Additional examples include the identification of clusters of glucocorticoid response elements in introns 1 and 2 of Granzyme A and FKBP5, respectively ( 30), and of a PPAR response element within intron 3 of the FIAF gene ( 31). This notion is further supported by the recent report that only 22% of binding sites for Sp1, c-myc, and p53 on chromosomes 21 and 22 are located at the 5′ termini of protein-coding genes, with the remainder positioned within or immediately 3′ to well-characterized genes ( 32). An interesting question that arises from these observations relates to the temporal occupation of intron response elements following gene activation. Presumably, the transcription factor will be effectively displaced from the element by the traveling polymerase.
In addition to caspase 9, RA treatment of MCF-7 cells up-regulated the expression of mRNA for caspase 7. However, unlike caspase 9, the induction of caspase 7 by RA was abolished on treatment with cycloheximide ( Fig. 2), indicating that this gene is an indirect target for RA signaling. Unlike caspase 9 also, CRABP-II did not augment RA-induced up-regulation of caspase 7 ( Fig. 3). As the effect of RA on caspase 7 is likely to be mediated by another protein which, in turn, is under RA control, the ineffectiveness of CRABP-II in the context of this gene may result from dilution of the binding protein effect in secondary steps. Alternatively, it is possible that the protein that directly regulates caspase-7 expression is only needed at low levels, rendering CRABP-II function unnecessary.
Overexpression of caspases and other components of the apoptotic response has been shown to trigger apoptosis or to increase the susceptibility of cells to apoptosis-inducing agents ( 33– 35). In keratinocytes, RA treatment was reported to up-regulate the expression of several caspases. The mechanism by which RA increases caspase expression in these cells remains unknown, but it has been shown that although their overexpression does not induce apoptosis by itself, it sensitizes the cells to apoptosis induced by UVB irradiation and by doxorubicin ( 36). We show here that in MCF-7 cells, RA increases the expression of several proapoptotic genes and leads to pronounced apoptosis without the need for additional agents. It is likely that the observed RA-induced apoptosis stems from a cumulative effect of the increased expression of multiple proapoptotic genes. It is worth noting with regard to this that it has been reported that RA treatment of MCF-7 cells leads to activation of Bax and to the release of mitochondrial cytochrome c ( 37). The mechanism by which RA affects Bax remains to be clarified, but its activation, together with the increased levels of the proapoptotic factors identified here, may coalesce to trigger and execute programmed cell death in MCF-7 cells in response to RA.
Perhaps the most surprising findings of this study are that CRABP-II, a protein that is known to cooperate with RAR in regulating the transcriptional activity of RA ( 19, 21, 22), also displays RA-independent activities. The data presented here show that, on its own, CRABP-II up-regulates the expression of Apaf1 and triggers the activation of caspase 7 and caspase 9. These observations suggest that the tumor-suppressor activities of CRABP-II ( 23) may stem from two separate biological functions: the known role of this protein in the direct delivery of RA to RAR and another, RA-independent, activity. The nature of this additional function for CRABP-II remains to be delineated. Finally, we note that although CRABP-II displays proapoptotic activities, it does not induce apoptosis on its own. As the induction of cell death is likely to depend on certain thresholds of proapoptotic components, the data suggest that CRABP-II sensitizes cells to apoptosis (i.e., it cooperates with other apoptotic agents to enable a more efficient induction of cell death).
Grant support: NIH grants DK60684 and CA107013 (N. Noy), and DAMD17-03-1-0249 (L.J. Donato).
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.
- Received April 5, 2005.
- Revision received June 8, 2005.
- Accepted July 14, 2005.
- ©2005 American Association for Cancer Research.