Retinoic Acid Down-regulation of Fibronectin and Retinoic Acid Receptor Proteins in NIH-3T3 Cells

All-trans-retinoic acid (RA) markedly reduced the level of intracellular fibronectin (FN) in a time- and concentration-dependent fashion in NIH-3T3 cells, but not in NIH-3T3 cells transformed by an activated Ha-ras oncogene. Pulse/chase experiments indicated that RA affects FN biosynthesis rather than its turnover rate. Steady state levels of FN transcripts did not change after treatment of the cells with RA for various times or concentrations, suggesting that RA acts at the translational level. Similar effects were observed in other fibroblasts. In NIH-3T3 cells, RA had distinct effects on different receptors; it down-modulated retinoic acid receptor (RAR) α protein and transcript levels, it up-regulated RARβ transcripts, and it had no effect on RAR. Transformation of NIH-3T3 cells with an activated Ha-ras oncogene down-modulated RAR expression and abolished responsiveness to RA. We identified the retinoid signal transduction pathways responsible for the effects of RA on FN and RARα proteins by the use of the retinoid X receptor-selective compound, SR11237, by stable overexpression of a truncated form of the RARα gene, RARα403, with strong RAR dominant negative activity, and by overexpression of RARα. We conclude that: 1) RA-dependent FN down-modulation is mediated by RARs, 2) retinoid X receptors mediate the observed reduction of RARα by RA, and 3) the block of RA responsiveness in Ha-ras cells cannot be overcome by overexpression of RARα. These studies have defined fibronectin and RARα as targets of RA in fibroblast cells and have shown that oncogenic transformation renders the cells resistant to RA action.

subtypes may control distinct gene expression patterns by interacting with RAREs, or RXREs, in the promoter region of different responsive genes (3,4). RA has proven effective in differentiation therapy of acute promyelocitic leukemia, a disease characterized by a t (15;17) translocation with breakpoint in the RAR␣ gene (5). In vitro, overexpression of RAR␣ has been shown to suppress transformation by v-myb in monoblasts (6) and by polyoma virus in rat fibroblasts (7,8). In addition RA treatment of NIH-3T3 cells transformed by the introduction of an activated Ha-ras oncogene inhibited focus formation (9). Various reports have shown opposing effects of RA and ras on the regulation of the expression of different genes (10 -13), suggesting that an interaction between the signal transduction pathways mediated by RA and ras may take place. Therefore, studies of this interaction may provide insight into the mechanism whereby RA inhibits transformations.
FN is a large transformation-sensitive glycoprotein composed of two non identical subunits of 220 kDa. It exists in the extracellular matrix and in soluble form in the plasma. Cellular FN is produced in large amounts by fibroblasts and is implicated in a wide range of cellular processes including cell adhesion, migration, morphology, differentiation, and transformation (14,15). It is modulated by a variety of effectors including cytokines like transforming growth factor ␤ and hormones like glucorticoids and RA. Loss of cell surface FN is a hallmark of transformation, and it has been correlated with acquisition of tumorigenic and metastatic potential. This effect has been observed with many oncogenic stimuli, among which are the ras oncogenes (14,16,17).
The ras genes encodes a 21-kDa plasma membrane protein that binds guanine nucleotides and is involved in signal transduction, cell growth, and differentiation (18). Many types of tumors express mutated forms of the ras protein, resulting in constitutive activation of this protein and altered gene expression (18).
In this study, we identified fibronectin as target of RA action in NIH-3T3 cells but not in ras-transformed fibroblasts. We also identified retinoid receptors involved in this process.

EXPERIMENTAL PROCEDURES
Cell Culture-NIH-3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% calf serum, 400 units/ml penicillin, 100 g/ml streptomycin, and 2 mM glutamine. Cells were grown to subconfluence and treated with retinoids. Activated Ha-rastransfected cells were generated and cultured as described (10). All comparative studies were done with matched pair clones, i.e. all the cells were always transfected with vector plus/minus the activated ras construct. Furthermore identical results were obtained in four NIH-3T3 clones compared to their ras-transformed counterparts. C3H10T1/2 cells were obtained from ATCC (Rockville, MD) and cultured as described (19). Primary mouse skin fibroblasts were isolated and cultured as described (20).
Retinoids-RA was obtained from Sigma. The RXR-selective compound, SR11237 (21), was kindly provided by Dr. A. Levin (Hoffman La Roche). RA and SR11237 were dissolved in Me 2 SO at 10 mM and in ethanol at 1 mM, respectively. The maximum concentration of the solvent used was 0.03% for Me 2 SO and 0.1% for ethanol. Subconfluent fibroblasts were incubated in presence of the media containing the required retinoids, and these media were replaced daily. For the last 16 h the cells were serum-starved by adding serum-free media containing 0.1% bovine serum albumin, with or without the effectors.
Metabolic Labeling and Immunoprecipitation-RA-and Me 2 SOtreated cells were washed with PBS and serum-free DMEM containing 0.1% bovine serum albumin, and 10% of the normal level of methionine and cysteine was added. Cells were preincubated for 1 h before adding 80 Ci/ml [ 35 S]methionine/cysteine protein labeling mixture (DuPont NEN). After 20 min of incubation, the cell monolayers were washed with cold PBS and lysed with buffer A (40 mM Tris-HCl, pH 8.5, 100 mM NaCl, 2 mM EDTA, 0.5% sodium deoxycholate, 1.0% Nonidet P-40, 0.5% SDS, 2 mM phenylmethylsulfonyl fluoride, and 10 M leupeptin). Cell lysates were centrifuged at 15,000 ϫ g ϫ 5 min at 4°C, and the insoluble pellets were discarded. Volumes containing equal amounts of radioactivity (precipitation with 10% trichloroacetic acid) were subjected to immunoprecipitation. Polyclonal rabbit mouse FN antibodies (5 l) were added. After 2 h of incubation at room temperature, 25 l of prewashed Pansorbin (Calbiochem, La Jolla, CA) was added and the samples were incubated overnight at 4°C on a rotary shaker. The immunoprecipitates were washed three times with buffer B (20 mM Tris-HCl, pH 8.5, 150 mM NaCl, 0.5% sodium deoxycholate), solubilized in SDS loading buffer, and loaded onto 4 -15% polyacrylamide gels (22).
Pulse/Chase-The reduced level of newly synthesized FN in RAtreated cells may be due to either a reduced rate of biosynthesis or an increased turnover rate. To explore these possibilities we performed pulse/chase experiments. Fibroblasts were pulsed for 20 min with [ 35 S]methionine and [ 35 S]cysteine and chased with unlabeled amino acids. At the indicated time points, the cells were washed twice with PBS and treated with 10 g/ml trypsin for 5 min to remove extracellular and cell surface FN. The reaction was stopped by adding 2 mg/ml trypsin inhibitor. The cells were washed with cold PBS containing 2 mM phenylmethylsulfonyl fluoride, lysed in buffer B, and analyzed for labeled FN by immunoprecipitation.
Immunoblot Analysis-To determine intracellular Fn levels, retinoid-and Me 2 SO-treated cells were washed and trypsinized as described in the pulse/chase section. Cells were lysed in Laemmli buffer without reducing agent and bromphenol blue. Lysates were boiled for 5 min and centrifuged to get rid of insoluble cell debris. Protein concentration was determined by the bicinchoninic acid method (23) (Pierce). ␤-Mercapthanol and saturated solution bromphenol blue were added to the samples at 1% and 3% final concentrations, respectively. Equal amounts of protein were then loaded onto 4 -15% polyacrylamide gels. The proteins were transferred to nitrocellulose (Schleicher & Schuell) on a Bio-Rad electroblot apparatus. The membranes were stained with Ponceau stain. The blots were incubated overnight at 4°C in 5% milk in TTBS (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20). Rabbit polyclonal mouse FN antibodies (Life Technologies, Inc.) were used at 1:1250 dilution in TTBS, and blots were incubated at room temperature for 1 h. They were rinsed four times with TTBS, and incubated 1 h with horseradish peroxidase-labeled secondary antibodies (anti-rabbit IgG, Amersham Corp.) at 1:5000 dilution in TTBS. The blots were washed five times before determining immunoreactivity by a chemiluminescent method using the ECL Western blotting system (Amersham). Rabbit polyclonal mouse collagen type IV and laminin antibodies were from Becton Dickinson (Bedford, MA).
For the detection of RAR␣ and ␥ proteins, total SDS lysates were prepared from cells washed twice with ice-cold PBS, as described above, except for the trypsin treatment. Proteins were boiled and run immediately on 10% polyacrylamide gels. They were transferred to nitrocellulose membranes which were then blocked overnight in 5% milk in TBS (50 mM Tris-HCl, 150 mM NaCl). Polyclonal antibodies against the carboxyl termini of RARs from Dr. Chambon's laboratory (24,25) were diluted 1:1000 in 50 mM Tris-HCl, pH 7.4, 300 mM NaCl, 1% bovine serum albumin, and the membranes were incubated at room temperature for 2 h. The blots were rinsed and washed five times in buffer C (50 mM Tris-HCl, 500 mM NaCl, 0.1% Tween 20) and incubated at room temperature for 1 h in 5% nonfat dry milk in PBS containing a 1:3000 dilution of the horseradish peroxidase-labeled IgG. The final wash sequence was three washes with buffer C and two washes with buffer D (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.3% Tween 20) (26). Blots were developed with ECL.
Stable Transfection-For stable transfection of RAR␣ cDNA, the fibroblasts were seeded at 6 ϫ 10 5 cells/100-mm dish 24 h before transfection in regular DMEM medium. Ten g of the expression plasmid pSG5-RAR␣f (27) and 1 g of the dominant selective vector pSVneo (28) were cotransfected using the Lipofectamine (Life Technologies, Inc.) methods according to the manufacturer's instructions. Control cells were transfected with 10 g of pSG5 empty vector and 1 g of pSVneo vector. G418-resistant colonies were selected in medium containing 1 mg/ml G418, and positive clones (overexpressing RAR␣ protein) were isolated and expanded. Northern Blot Analysis-Isolation of total RNA was performed by using a Total RNA isolation kit (Tel-Test "b" Inc., Friendswood, TX). The full-length fragments of the mouse RARs were excised from the expression plasmid pSG5-RAR (29,30). The 1.4-kilobase human Fn fragment was obtained from Life Technologies, Inc. The probes were labeled with [ 32 P]dCTP using random primer labeling methods. Total RNA (40 g) was fractionated on a 1% agarose gel and blotted overnight onto Schleicher & Schuell nitrocellulose. The membranes were prehybridized for 2 h at 65°C in a buffer of 5 ϫ SSC, pH 7.0, 5 ϫ Denhardt's, 0.05 M sodium phosphate, pH 6.8, 0.1% SDS, 5 mM EDTA, 20 g/ml poly(A) n , 0.2 mg/ml denatured salmon sperm DNA, and 0.1 mg/ml denatured torula yeast RNA. The probes (2 ϫ 10 6 cpm/ml) were boiled and added to the hybridization buffer (5 ϫ SSC, 1 ϫ Denhardt's solution, 0.02 M sodium phosphate, 5 mM EDTA, 0.2% SDS, 20 g/ml poly(A) n , 0.2 mg/ml denatured sperm DNA, and 0.1 mg/ml denatured yeast RNA. The membranes were hybridized for 18 h at 65°C, followed by two washes with 1 ϫ SSC, 0.1% SDS and two with 0.1 ϫ SSC, 0.15% SDS at 65°C. Autoradiography on Kodak X-Omat AR film used double intensifying screens. In order to reprobe the membranes with labeled GAPDH, the blots were first stripped in 1% glycerol at 85°C for 5 min. Densitometric evaluation of Northern blots was performed with NIH Image 1.52 software (created by Wayne Rasband, National Institutes of Health) running on a Macintosh Centris 650 using an XC-77/77CE CCD video camera module.
Infection of NIH-3T3 with Retroviral Vectors-The retroviral vector LRAR␣403SN, in which a truncated RAR␣ gene is inserted into the retroviral vector LXSN, was a gift from Dr. S. J. Collins (31,32). Cells were seeded at 50% confluence into 100-mm dishes. The next day, they were infected with the LXSN and LXRAR␣SN retroviral vector in presence of 4 g/ml Polybrene (31). After overnight incubation, the medium was replaced and cells were grown for 36 -48 h before G418 (1 mg/ml) was added. G418-resistant cells were isolated.
Transient Transfection with Constructs Containing ␤-RARE-tk-LUC-␤-RARE-tk-LUC was a gift from Dr. S. Minucci (National Institutes of Health) (33). Fibroblasts (1.5 ϫ 10 5 /dish) were seeded in 35-mm dishes. After 24 h, cells were cotransfected with 1 g of the ␤-RAREtk-LUC and 1 g of the cytomegalovirus-␤-galactosidase reporter (34) using Lipofectamine reagent. Six hours post-transfection, the cells were incubated in DMEM containing 10% serum for 16 h. The cells were treated with 2 M RA or solvent (Me 2 SO) for 24 h. Cells were lysed with reporter lysis buffer (Promega, Madison, WI). Luciferase activity was determined in 1 g of protein from each sample using the luciferase assay system from Promega in a Monolight 2010 luminometer (Analytical Luminescence Laboratory, San Diego, CA). ␤-Galactosidase activity was quantified by the FluoReporter LacZ/Galactosidase kit (Molecular Probes, Inc., Eugene, OR).

Inhibition of FN Biosynthesis by RA in NIH-3T3 but Not in
Ha-ras NIH-3T3 Cells-FN levels were markedly reduced by RA in pSVneo, but not in Ha-ras-transfected cells (Fig. 1A). The effect was specific for FN, since neither collagen type IV nor laminin B1 and B2 chains (the only ones to be detected in NIH-3T3 cells) were affected. RA inhibition of FN was time-dependent (Fig. 1B) and concentration-dependent (Fig. 1C). A 3-fold reduction of intracellular FN was already observed after 12 h of treatment (Fig. 1B). RA caused a 90% down-regulation at 2 ϫ 10 Ϫ6 M.
Mechanism In pulse/chase experiments, the rate of disappearance of labeled FN was not altered by RA treatment of the cells (Fig.  2B). The time required to reduce the amount of 35 S-FN by 50% was 28 Ϯ 2 and 31 Ϯ 3 min in RA-and Me 2 SO-treated cells, respectively. Therefore, a reduction of FN biosynthetic rate likely accounts for the effects of RA.
Northern blots of total cellular RNA from RA-treated and control cells were hybridized to FN and GAPDH probes. The FN mRNA band intensities, which represent relative steadystate level, were normalized to GAPDH mRNA band intensities. The results (Fig. 3) reveal that the accumulated levels of FN mRNA were not altered either by different RA concentration or by times of RA treatment. RA failed to alter FN mRNA in Ha-ras NIH-3T3 cells.
RA Modulation of RARs in pSVneo NIH-3T3 but Not in Ha-ras NIH-3T3 Cells-In pSVneo NIH-3T3 cells, RAR␣ transcripts were constitutively expressed and slightly (30%) downregulated by RA (data not shown). RAR␥ mRNAs were expressed to a lower extent than RAR␣ mRNAs and were not substantially altered by RA. RAR␤ transcripts were induced in pSVneo cells to a similar extent (data not shown) as in primary mouse skin fibroblasts (36). In Ha-ras NIH-3T3 cells, the levels of RARs mRNA were generally lower than in control cells and were not responsive to RA treatment.
RAR␣ protein levels were strongly down-regulated by RA in NIH-3T3 cells, but not in Ha-ras cells (Fig. 4A). RAR␥ proteins were not altered by RA in either type of cells and could be detected only after overexposing the blots (data not shown). We were unable to detect RAR␤ proteins. The effect of RA on the levels of RAR␣ protein was time-and dose-dependent (Fig. 4B). A concentration of 0.1 nM RA was sufficient to bring about a 80% inhibition of RAR␣ protein, and a 40% inhibition could be already observed after 2 h of RA treatment.
RA Inhibition of FN and RAR␣ Proteins in Fibroblast Cells-We investigated the effects of RA on FN and RAR␣ proteins in C3H10T1/2 mouse embryo fibroblasts and in primary mouse skin fibroblasts. RA markedly reduced the level of FN and RAR␣ in these two cells (Fig. 5).
Role of RAR Signal Transduction Pathways on RA Inhibition of FN and RAR␣ Proteins-SR11237, an RXR-selective compound, specifically activates reporter genes fused to the RXRresponsive element of the CRBPII promoter, to which only RXR-RXR homodimers bind, and is unable to induce genes driven by an RAR-responsive element, like the RARE of RAR␤ and CRBPI (21). The RAR-mediated signaling pathways can be blocked by overexpressing a mutated form of the RAR␣ gene, RAR␣403, which shows strong dominant negative activity (31,32). The truncated form of the RAR␣ gene, RAR␣403, inserted into the retroviral vector LXRAR␣SN and the corresponding control retrovirus LXSN were used. We infected NIH-3T3 cells with these amphotrophic retroviruses containing the neomycin resistance gene. Neomycin-resistant cells were isolated and expanded. Northern blot analysis was performed to detect the expression of the typical 4.7-kilobase retroviral transcripts containing the RAR␣403 mRNA (Fig. 6). The dominant negative activity of such mutated receptor was confirmed in RAR␣403 NIH-3T3 cells by transient reporter assays. LXSN control and LXRAR␣403SN NIH-3T3 cells were transiently transfected with a vector carrying the luciferase reporter gene fused to two copies of the RARE of RAR␤2 gene (33), RA-dependent transactivation of luciferase was then evaluated. Cells overexpressing RARa403 transcripts showed an almost negligible 1.2-fold induction of luciferase activity in comparison to a 5-fold induction in control LXSN cells (Fig. 6B). Similar data were obtained by others in tk Ϫ NIH-3T3 cells infected with the very same retroviral vectors (31). RA failed to down-regulate the intracellular FN levels in the LXRAR␣403SN cells (Fig. 7B, left panel). Furthermore, the RXR-selective compound, SR11237, also failed to reduce FN protein (Fig. 7B, right panel). These results indicate that an RAR-mediated signaling pathway likely accounts for the RAdependent FN down-regulation.
Similar experiments were performed, looking at the RA modulation of RAR␣ (Fig. 7A). In LXRAR␣SN NIH-3T3, treatment with RA caused a marked reduction of the levels of RAR␣, ruling out the involvement of RAR signal transduction pathways. This finding was strengthened by the inhibition of RAR␣ in cells treated with SR11237 to a similar extent as that achieved by RA treatment (Fig. 7A, right panel). This suggests the involvement of RXR in this effect.
Effect of RA in NIH-3T3 Cells Overexpressing RAR␣-To further investigate the role of RARs, we generated cell lines overexpressing the RAR␣ gene. Cotransfections of NIH-3T3 cells with the plasmid pSVneo, harboring the neomycin resistance gene, and the pSG5RARaf vector, containing the fulllength RAR␣ gene, were performed. G418-resistant clones were isolated, expanded, and tested for RAR␣ protein overexpression. Two representative clones are shown in Fig. 8. Both express very high levels of RAR␣ protein, however, the levels of RAR␣ were down-modulated by RA only in clone 3A. Along the same trend, when FN sensitivity to RA was determined, RA failed to reduce the level of FN in the resistant clone.
Effect of RA on Ha-ras NIH-3T3 Cells Overexpressing RAR␣ and RAR␣403-A similar series of experiments was performed in Ha-ras NIH-3T3 cells. Since the levels of RARs were generally lower than in control cells, we generated Ha-ras NIH-3T3 cell lines, which overexpressed RAR␣. The RA effects on FN biosynthesis were evaluated. Overexpression of RAR␣ protein was not sufficient to overcome the block on RA responsiveness (Fig. 9A). Similarly, disruption of the RAR signaling pathways by the introduction of the dominant negative RAR construct, RAR␣403, into Ha-ras NIH-3T3 cells, was ineffective in altering the responsiveness of FN protein to RA. DISCUSSION We have identified FN as a molecule whose biosynthesis is down-regulated by RA in normal, but not in Ha-ras-transformed NIH-3T3 cells. The inhibition of FN biosynthesis by RA is specific, as neither collagen type IV nor laminin were affected by RA, and is RA dose-and time-dependent. Two lines of evidence suggest that RA acts on FN at a post-transcriptional level. First, the rate of newly synthesized intracellular FN is reduced by RA treatment, an event not due to an increased FN turnover rate. Second, RA did not alter the levels of FN transcripts, consistent with the absence of RARE or RXRE in the promoter region of the FN gene (37,38). Effects of RA in other cell systems have been reported. FN mRNA and protein levels were increased in primary hepatocytes from vitamin A-deficient rats, while RA treatment caused a reduction of FN mRNA and protein levels (39). In C3H10T1/2 fibroblast cells, a complex RA-dependent regulation of FN was observed as the cell surface levels increased, while intracellular FN and FN mRNA decreased after RA (19).
RA generally controls gene expression at the transcriptional level. However, lipoprotein lipase enzyme expression in 3T3-L1 adipocytes was down-regulated, but mRNA levels were not affected by RA treatment (40). RA induction of differentiation of F9 cells is achieved by controlling the expression of various genes. Early responsive genes are thought to be transcriptionally regulated, while late responsive genes may be controlled both at the transcriptional and post-transcriptional levels (41). Laminin biosynthesis is switched on, while synthesis and secretion of FN is switched off by RA treatment of F9 cells (42)(43)(44).
Transformation of NIH-3T3 cells with an activated ras caused RAR␣ and ␥ to be expressed to lower levels than in normal cells, while RA induction of RAR␤ was absent. In addition to specific mutations in the RAR␣ gene (5) the expression of RARs by transformation or in tumor derived cells has been reported. In lung cancer cells, RAR␣ and ␥ and RXRs were well expressed; however, RA induction of RAR␤ was not observed (45). Estrogen receptor-negative human breast cancer cells were insensitive to RA inhibition of cell growth, probably due to a low level of RAR␣ expression (46). Primary keratinocytes infected by v-ras Ha showed lower levels of RAR␣ and ␥ proteins, which is accompanied by a reduction in RAinduction of reporter genes fused to a RARE. 2 The molecular mechanism of this down-regulation is not known yet, but it could have important implications in explaining the variability in RA responsiveness of different tumor cells.
RAR␣ protein is down-modulated by RA in a dose-and timedependent fashion in NIH-3T3 cells. The modest inhibition (30%) of RAR␣ mRNA is in agreement with various reports in different cell lines and tissues, which, however, had exclusively focused on the effect of RA at the level of transcripts (36,47,48). More recently, RA-dependent down-modulation of RAR␣ proteins was reported in estrogen receptor-positive human breast cancer cells (46). Transcriptional regulation of retinoid receptors by RA has long been known, and one of the first RARE was found in the promoter region of the RAR␤ gene (49).
This work also shows that the RA-dependent down-modulation of intracellular FN appears to be RAR-mediated. The introduction of a mutated RAR␣ gene, which has strong dominant negative activity and blocks the RAR-mediated signaling pathways, abolished the RA inhibition of FN. Furthermore the RXR-selective compound, SR11237, was unable to mimic the action of RA on the intracellular FN. We conclude that RAR␣ down-modulation by RA is likely mediated by RXRs, since RA reduces RAR␣ levels, even when we blocked RAR-mediated signaling pathways by the overexpression of the RAR dominant negative gene, RAR␣403. In addition the RXR-selective retinoid, SR11237, is as powerful as RA in down-regulating RAR␣ protein.
In our RAR␣ overexpression studies we isolated a clone, 13A, which appeared to be insensitive to RA in that the two gene products, FN and RAR␣, were not responsive to RA. The mechanism of the observed RA resistance is not clear. We showed that action of RA on FN is mediated by RAR and is dependent on RXR/RAR heterodimer, while RAR␣ inhibition by RA is likely mediated by RXR/RXR homodimer. The fact that RA resistance is observed for both gene products argues against a defect localized entirely on the RAR signaling pathway. If this were the case, RAR␣ protein expression should have remained sensitive to RA effects. Introduction of an activated ras alters the responsiveness to RA and the level of expression of RARs. Manipulation of RAR protein levels was utilized in an attempt to correlate these two observations.
Overexpression of the RAR␣ gene in Ha-ras NIH-3T3 cells was not sufficient to overcome the block in FN and RAR␣ (data not shown) responsiveness. Failure to regain RA sensitivity after constitutively expressing RAR␤ in lung cancer cells has been reported (45). RA-dependent induction of a luciferase reporter gene fused to the RARE of the RAR␤ promoter was equally or more efficient in Ha-ras NIH-3T3 than in normal cells (data not shown). Caution should be used in evaluating the physiological relevance of reporter gene assays, because the RARE is taken out of the context of its natural promoter. Swisshelm et al. (50) showed that when a 1.5-kilobase region of the RAR␤2 promoter was used in reporter gene constructs, instead of the RARE, suppression of RA-induced activation was detected in human MCF-7 breast cancer cells. The observation that ␤-RARE-tk-LUC can be activated in Ha-ras NIH-3T3 cells suggests that necessary factors for activating ␤RARE are functional. The level of retinoid receptor expression often does not correlate with RA-responsiveness. RAR␣ was expressed in most leukemia cells whether or not they were responsive to RA (51)(52)(53). In melanomas, the level of RAR␤ and RAR␥ were similar in RA-sensitive and -resistant cells (54). These findings and the observed failure of RAR␣ overexpression to confer RA responsiveness to Ha-ras cells suggests that other factors are required to mediate RA action. These factors may be missing or not functional after ras transformation.