Receptor-selective Retinoids Implicate Retinoic Acid Receptor

The effect of retinoids on malignant cells and embryos indicates that retinoids influence the expression of growth factors or alter the response of cells to growth factors. The bone morphogenetic proteins, Bmp-2 and Bmp-4, are candidates for such growth factors because retinoic acid (RA) treatment of F9 embryonal carcinoma cells induced Bmp-2 mRNA, while simultaneously repressing Bmp-4 levels. Also, recombinant Bmp-2 affected the growth and differentiation of these cells. Regulation of each gene was concentration dependent and required continuous RA treatment. The short half-lives of the Bmp-2 (75 ± 1 1 mm) and Bmp-4 (70 ± 4 mm) mRNAs suggest that their abundance is primarily controlled at the transcriptional level. To determine which RA receptor (RAR) controls bmp-2 and bmp-4 expression, F9 cells were exposed to various receptor-selective retinoids. RAR aand y-selective retinoids induced Bmp-2 and repressed Bmp-4 equally as well as all-trans RA. In contrast, a RAR fl-selective retinoid had little effect on Bmp-2 induction but repressed Bmp-4. A RAR aselective antagonist inhibited all-trans RA stimulation of Bmp-2, although not as dramatically as a RAR IJ selective antagonist. No differences were observed between Bmp levels in all-trans RA and 9-cis RAtreated cells, indicating that the RXRs play little part in controlling these genes. The results are consistent with RAR a and ‘y-controlled Bmp-2 and Bmp-4 regulation. Introduction The acid form of vitamin A, RA, is an important regulator of cell proliferation and differentiation in a broad array of tissues Received 9/1 1/95; revised 1 1/3/95; accepted 11/3/9S. 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 mdicato this fact. 1 This work was supported by Grant F93USF-1 from the American Cancer Society, Florida Division, Inc., the Seymour Kramer Award from the Loukomia Research Foundation, Inc., and Grant R29 HD31 1 1 7 from the National Institute of Child Health and Human Development. 2 To whom requests for reprints should be addressed, at Department of Biology, L1F136, University of South Florida, 4202 East Fowlor Avenue, Tampa, FL 33620. Phone: (813) 974-2623; Fax: (813) 974-3263; E-mail: [email protected]. 3 The abbreviations used are: RA, retinoic acid; RAR, RA receptor; RXR, (i). RA-regulated gene expression is mediated by nuclear retinoid receptors that have been shown to act as liganddependent transcription factors (2). These receptors are encoded by six different genes. Numerous isoforms arising from differential promoter usage and alternate splicing have been identified, and the expression of these isoforms is developmentally regulated (3). Furthermore, the receptors act as homodimers and heterodimers, each with unique characteristics (Ref. 4 and references therein). This plethora of receptors and gene pathways may begin to explain the multiple effects retinoids have on differentiation. To ameliorate the toxic and teratogenic effects of retinoids, extensive effort has been devoted to the synthesis of retinoids that activate a subset of retinoid receptors (5). Both receptor-selective agonists and antagonists have been identifiod. These retinoids have enormous potential clinical value. They are also useful research tools, as they may regulate a subset of the target genes (e.g., bmp-2 and bmp-4) regulated during RA-inducod differentiation. Bmps are growth factors that can induce cartilage and bone growth in vivo (6, 7). Bmp-2 and Bmp-4 are closely related (92% at the amino acid level) members of the transforming growth factor 3 family. Recombinant Bmp-2 and Bmp-4 induce bone in ectopic bone assays (8) and mimic the induction of dental mesenchyme by dental epithelium (9). These growth factors are potential targets for the teratogenic effects of retinoids because they can be regulated by RA (iO). Bmp-2 and Bmp-4 play many roles in addition to bone and tooth induction. Furthermore, the bmp-2 and bmp-4 genes and genes regulating Bmp signal transduction are functionally conserved between mammals, amphibians, and insects (1 1 , 12). Mutations and embryo manipulations indicate that normal Bmp levels are required for normal axial patterning from invertebrates to vertebrates (i i , i 3-i 5). Since retinoids are known to perturb development at these early stages (reviewed in Ref. i 6), one mechanism may be perturbation of Bmp levels. The research cited above characterized the effects of RA, Bmp-2, and Bmp-4 on mesoderm and axis formation. As these occur relatively late in mammalian development, it was surprising to observe that RA administration at pregastrulation stages induced malformations in developing mouse embryos (i 7). F9 embryonal carcinoma cells provide an ideal experimental system for dissecting the role of RA in regulatrotinoid X receptor; Bmp, bone morphogenotic protein; RACT, a treatment of RA, dibutyryl cAMP, and theophyllino; DAB, 5,6-dichloro-1-13-D-ribofuranosylbenzimidazole. 116 bmp-2 and bmp-4 Regulation in F9 Cells quired to induce Bmp-2 ANA to maximal levels. Replacing ing Bmp-2 and Bmp-4 in very early mouse differentiation. F9 cells are the malignant stem cells of a mouse teratocarcinoma. Monolayer cultures of F9 cells treated with RA differentiate into benign cells resembling the primitive endoderm of the mouse blastocyst. If the cells are treated with RACT, the cells become biochemically indistinguishable from parietal endoderm and express parietal endoderm markers such as laminin and tissue plasminogen activator. This RA-responsive system was used to show that Bmp-2 and Bmp-4 are RA-regulated genes. Additionally, recombinant Bmp-2 affected the morphology, growth, and gene expression of differentiating F9 cells, suggesting that regulation of Bmp-2 and Bmp-4 may affect the loss of malignancy in these cells (10). This work furthers our understanding of how bmp-2 and bmp-4 are regulated by RA in F9 cells. The mechanism by which RA regulates these growth factors was addressed by determining the half-life of the mRNAs under different treatment conditions. Finally, the effect of receptor-selective retinoids on bmp-2 and bmp-4 expression and F9 cell differentiation was determined. Results RA Regulation of bmp-2 and bmp-4 in F9 Embryonal Carcinoma Cells. Previous experiments showed that the abundance of Bmp-2 transcripts was increased, and the abundance of Bmp-4 transcripts was decreased in F9 cells treated with 0.5 t i all-trans RA, 250 M dbcAMP, and 500 p.M theophylline (1 0). To determine the optimal conditions for Bmp-2 and Bmp-4 expression, F9 cells were treated with 1 0_6 M, 10 M, 1 08 M, and 1O M all-trans RA with CT (250 j. M dibutyryl cAMP and 500 .tM theophylline) for i 02 h. ANA was prepared for Northern blot analysis and Bmp-2, Bmp-4, and laminin Bi mANA levels were measured. Fig. iA shows the autoradiographs of these blots, and Fig. lB shows the data normalized to the constitutive message 36B4 (1 8). As shown in Fig. 1 , 106 M all-trans RA (Lane 6) induced Bmp-2 twice as effectively as 10 M RA (Lane 7). Bmp-4 levels were undetectable at 106 M all-trans RA, while 36% of the Bmp-4 RNA measured in CT-treated cells remained in 10 M alltrans RA-treated cells. Higher concentrations of RA were not tested, because 1 0_6 M all-trans RA is generally considered the upper limit of physiological concentrations. Cell morphology and the abundance of the parietal endoderm marker, laminin Bi , indicated that the cells were completely differentiated. Thus, high Bmp-2 RNA levels and the absence of Bmp-4 RNA correlate with the parietal endoderm phenotype. Some transcripts are regulated very rapidly by RA (e.g., HoxAl within 2 h; Ref. 19), whereas others require several days of RA treatment (e.g., laminin Bi). A time course of Bmp-2 and Bmp-4 expression was obtained by isolating ANA from cells treated with 106 M all-trans RA and CT for 12, 24, 48, 72, and 102 h (Fig. 1 , Lanes 2-6). Bmp-2 required 3 days of drug treatment to reach maximum levels. Bmp-4 was completely repressed by 2 days of treatment. This mdicates that these growth factors are regulated late in the stepwise progression of differentiation. Furthermore, continuous all-trans RA treatment was redrug-containing media with drug-free media at 12, 24, 48, or 72 h, followed by harvesting at 102 h (Fig. i , Lanes 11-14), induced Bmp-2 to loss than 40% of the amount observed in continuously treated cells. Interestingly, the Bmp-4 mRNA abundance in cells treated with i 06 M RA and CT (Fig. i, Lane 2) for 12 h was 2.5-fold higher than that observed in CT-treated cells (Fig. 1 , Lane 10). This increased level was maintained for nearly 4 days following removal of drug (Fig. i , Lane 1 1), suggesting that the transient Bmp-4 induction was not due to the recent change of media. The data shown in Fig. 1 indicate that optimal conditions to induce Bmp-2 and repress Bmp-4 include continuous i06 M RA and CT treatment for at least 3 days. Bmp-2 and Bmp-4 mRNAs Are Short-Lived. Two main mechanisms by which RNA abundance is regulated are transcription rate and message stability. To analyze the relative stability of these transcripts under different conditions, F9 cells were grown in the absence of drug (undifferentiated), in 1 .0 LM all-trans RA, or in 1 .0 M all-trans RA and CT (fully differentiated) for 3 days. The cells were then treated with the transcriptional inhibitors DAB or actinomycin D. The concentrations used, 25 g/ml DAB or 2 g/ml actmnomycmn D, have been shown to inhibit 5,6-rH]uridine incorporation by 9095% (20). The results of two representative experiments are shown in Fig. 2A. Since Bmp-2 mANA is induced by RA or RACT and is undetectable in undifferentiated cells, only the data regarding the retinoid-treated cells is shown. Likewise, since Bmp-4 is detectable only in undifferentiated and RAtreated cells, only this data is shown. All transcript measurements were normalized to the level of the constitutive message 36B4, which remained constant during the 6 h of transcription inhibition. The data is plotted (Fig. 2B) as a percentage of the ANA abundance in cells not treated with inhibitor (time zero). Within 3 h of DAB or actinomycin D addition, Bmp-2 and Bmp-4 mRNA levels fell to iO% of those observed at time zero. In contrast, laminin Bi mANA levels in RACT-treated cells were 54% of time zero levels after 6 h of DAB treatment (data not shown). A straight line was obtained when the data shown in Fig. 2B was plotted as the log of RNA amount versus time, indicating first-order decay kinetics. The Bmp-2 message half-life was 75 ± ii mm, and the Bmp-4 message half-life was 70 ± 4 mm. Thus, Bmp-2 and Bmp-4 have relatively unstable messages, suggesting that the regulation of these genes occurs primarily at the transcriptional level. Regulation of Bmp-2 and Bmp-4 by Receptor-selective Retinoids. Six unique retinoid receptors, RAR a, 3, y and AXA a, 13, “ mediate the transcription of AA-regulated genes. A number of natural and synthetic retinoids have been identified that specifically activate all or a subset of these receptors. All-trans RA and 9-cis-AA are naturally occurring retinoids. All-trans AA activates all three AAAs, while 9-cisAA activates all three AAAs and all three RXAs. Aol 3-7410 (aka TTNPB) is a nonmetabolized activator of AAA a, 3, and 7. Three retinoids that preferentially activate individual RARs are Ao40-6055 (AAA a), CD2Oi9 (RAR j3), and CD437 (RAR .y). Two antagonists are available that prevent RAR a (Ro4i 5253) activation or AAA p and y (CD2665) activation.