Int. J. Cancer: 71, 497?504 (1997) r 1997 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l?Union Internationale Contre le Cancer ACTIVITY OF RETINOIC ACID RECEPTOR-g SELECTIVELY BINDING RETINOIDS ALONE AND IN COMBINATION WITH INTERFERON-g IN BREAST CANCER CELL LINES Martin WIDSCHWENDTER1*, Gu?nter DAXENBICHLER1, Zoran CULIG2, Serge MICHEL3, Alain G. ZEIMET1, Manfred G. MO?RTL4, Andreas WIDSCHWENDTER5 and Christian MARTH1 1Department of Obstetrics and Gynecology, University of Innsbruck, Innsbruck, Austria 2Department of Urology, University of Innsbruck, Innsbruck, Austria 3CIRD Galderma, Sophia Antipolis, France 4Department of Anesthesia and Intensive Care, University of Innsbruck, Innsbruck, Austria 5Department of Medical Biochemistry, University of Innsbruck, Innsbruck, Austria Retinoids modulate several cell functions and especially inhibit the growth of a wide variety of cells including breast cancer. Retinoic acid receptor-g (RAR-g) has been shown to mediate the antiproliferative activity of retinoids. To further test this hypothesis we examined the effects of different RAR-g selectively binding retinoids (CD2325, CD2247, CD666 and CD437) on breast cancer cell lines. With exception of CD2247, all retinoids inhibited proliferation of MCF-7, SKBR-3, T47D and ZR-75-1 breast cancer cell lines, similar to the natural compound all-trans retinoic acid (ATRA). In addition, all 4 compounds were able to act synergistically with interferon-g (IFN-g) in all breast cancer cell lines including the retinoid-resistant BT-20 and 734-B lines. In functional transactivation assays we demonstrated that only in the MCF-7 cell line, TPA-mediated AP-1 activity was suppressed only by ATRA and CD2325, whereas in SKBR-3, another RA-sensitive breast cancer cell line, it was not. The synergistic antiproliferative activity involving retinoids and IFN-g could not be explained by an enhanced anti-AP-1 activity. No correlation was found between expression of RARs and cellular retinoic acid binding proteins (CRABPs) and antiproliferative effects of the retinoids. RAR-g selectively binding retinoids are potent inhibitors of breast cancer cell proliferation, alone and in combination with IFN-g. For this reason and because of a possible low toxicity, as compared with retinoic acid, we speculate that these RAR-g selective binding retinoids might be of clinical importance. Int. J. Cancer 71:497?504, 1997. r 1997 Wiley-Liss, Inc. Retinoids are known to possess antiproliferative, differentiative and immunmodulatory properties. The key molecules are the binding proteins CRABP I and II (cellular retinoic acid-binding protein), the retinoid receptors (RAR-a, RAR-b and RAR-g) and retinoid X receptors (RXR-a, RXR-b and RXR-g), which are part of the steroid/thyroid hormone receptor superfamily (Sporn et al., 1994). A growing body of evidence derived from clinical research supports the concept that retinoids are useful substances in the prevention and treatment of cancer. Retinoids partly alone, partly in combination with biological response modifiers or chemotherapy have proved to be effective in skin diseases, acute promyelocytic leukemia, cervical cancer and other malignancies (Sporn et al., 1994). When combined with tamoxifen and interferon (IFN), retinoids induce remission in heavily pretreated patients with metastatic breast tumors (Recchia et al., 1995). The pleiotropic biological activities of natural retinoids and related compounds, however, produce a number of undesirable side effects, which limit their clinical applications. In breast cancer cell lines, retinoids alone or in combination with interferons exhibit strong antiproliferative activity (Marth et al., 1986, 1993; Widschwendter et al., 1996). We have shown that IFN-g-mediated increase in RAR-g and suppression of retinoic acid (RA)-mediated CRABP II activation may play a role in synergistic inhibition of proliferation in breast cancer cell lines (Widschwendter et al., 1995). Extending these findings, we compared the antiproliferative effects of RAR-g selectively binding retinoids CD2325, CD2247, CD666 and CD437 on several breast cancer cell lines with those of natural retinoids, alone and in combination with IFN-g. Fanjul et al. (1994) have shown that retinoids may inhibit proliferation by reducing AP-1 activity. Therefore, we were interested in checking whether these retinoids were also able to reduce AP-1 activity. To compare levels of RARs and CRABPs with response to these retinoids we performed Northern blots. Modulation of CRABP is a possible mechanism influencing retinoid action. As a consequence of the previous finding that all-trans retinoic acid (ATRA) induces CRABP II, we were also interested in CRABP II promoter activation by CD2325. MATERIAL AND METHODS Reagents CD2325, CD2247, CD666 and CD437 (Fig. 1) were kindly provided by Dr. U. Reichert (CIRD Galderma, Sophia Antipolis, France), and ATRA by Dr. W. Bollag (Hoffmann-La Roche, Basel, Switzerland). For all experiments, 1-mM solutions were prepared in DMSO and further diluted in complete culture medium. To avoid degradation and isomerization of the retinoids, only fresh solutions were used. The human recombinant DNA-derived IFN-g was kindly provided by Dr. G. Adolf (E. Boehringer Institute, Vienna, Austria). The preparation was essentially pure, and exerted an antiviral activity of 2 3 107 U/mg protein. IFN-g was diluted in minimum essential medium (MEM) containing 15% fetal bovine serum (FBS) and stored at 270蚓. Cell culture The MCF-7, BT-20, SKBR-3, 734-B, ZR-75-1, T47D and Hs578T human breast cancer cell lines were cultured as described by Marth et al. (1993). Briefly, the cells were maintained in MEM containing 10% FBS (both from Eurobio, Paris, France); they were seeded into 24-well tissue culture plates (Nunc 146485, Roskilde, Denmark) in complete culture medium for evaluation of proliferation. After treatment with ATRA or CD2325 alone or in combination with IFN-g or the vehicle alone, cells were detached with the help of trypsin (0.05%)-EDTA (0.02%) in Dulbecco?s phosphatebuffered saline (PBS, Biological Industries, Kibbutz Beth Haemek, Israel), and their number was assessed using an electronic particle counter (Coulter, Dunstable, UK) as described by Marth et al. (1993). Northern blot analysis Northern blot analysis was done as described by Widschwendter et al. (1995). Briefly, total cellular RNA was extracted by the Contract grant sponsor: Jubila?umsfonds der O?sterreichischen Nationalbank, 5788. *Correspondence to: Department of Obstetrics and Gynecology, University of Innsbruck, Anichstrasse 35, A-6020 Innsbruck, Austria. Fax: 0043 512 504 3112. Received 16 December 1996 498 WIDSCHWENDTER ET AL. FIGURE 1 ? Structure, name and KD (nM)-values for RAR-a, RAR-b and RAR-g for RAR-g selectively binding retinoids CD2325, CD2247, CD666 and CD437. guanidine thiocyanate method. Ten micrograms of total RNA mixed with ethidium bromide were run on denaturing 1% agaroseformaldehyde gels and transferred to nylon membranes (Stratagen Flash Nylon Membranes, La Jolla, CA) by Northern blotting. The sheet thus prepared was fixed and photographed under ultraviolet (UV) light (to demonstrate comparable RNA levels) and hybridized with digoxigenin-labelled DNA probes as previously described. Detection of digoxigenin-labelled nucleic acids by chemiluminescence enzyme immunoassay on nylon membranes was carried out following the manufacturer?s instructions (DIG Luminescent Detection Kit, Boehringer Mannheim, Vienna, Austria). Filters were exposed to autoradiographic films (Hyperfilm, Amersham, Solna, Sweden) for 5 hr. Statistics Differences in median cell number were analyzed with the Wilcoxon U-test. Interaction of CD2325 with IFN-g was evaluated according to the recommendations of Berenbaum (1981) and Chou and Talalay (1984), using the computer program Dose-Effect Analysis With Microcomputer (Elsevier Biosoftware, Cambridge, UK). Dose-effect curves were plotted with either agent alone and with their combination. From these experiments a combination index (CI) was calculated to characterize the quality of interaction: CI 5 1 indicates an additive interaction, whereas CI . 1 represents antagonistic and CI , 1 synergistic activity. This type of calculation is very similar to isobolarogram construction and has the advantage of preventing errors caused by dose-response curves which are not simply exponential. A most widely used model for analyzing interaction of substances is the effect-multiplication criterion. Several sources of error may, however, occur, and therefore experiments with various dose combinations and evaluations of effects by appropriate mathematical methods are proposed. CAT assays MCF-7 cells were grown on 24-well plates. The CAT reporter plasmid 5 3 TRE TATA CAT (Jonat et al., 1990) was kindly provided by Dr. A.C.B. Cato (Karlsruhe, Germany), and the CAT reporter CRABPII-RARE2 (Durand et al., 1992) by Dr. P. Chambon (Strasbourg, France). Liposome-mediated transfection was performed under serum-free conditions. The DNA solution was RAR-g SELECTIVE BINDING RETINOIDS AND IFN-g IN BREAST CANCER CELLS mixed with lipofectamine transfection reagent (GIBCO BRL, Paisley, UK; 3 ml/mg plasmid DNA) and added to the cells. After an incubation period of about 10 hr, medium was changed to phenol red-free medium supplemented with 1% charcoal-treated serum, and substances were added. Twenty-four hours after supplementation, the medium was removed and the cells were frozen. CAT activity was assessed with a modified mixed-phase assay: the cells were incubated with CAT mix [1.3 mg/ml chloramphenicol, 0.25 湣i 3H-labeled (NEN, Boston, MA) and 60 nM unlabeled acetyl-coenzyme A (Sigma, St Louis, MO)] in 0.3 M Tris HCl, pH 7.0, at 55蚓 for up to 2 hr. A scintillation cocktail insoluble in water (Optiscint, Pharmacia, Uppsala, Sweden; 0.75 ml) was added to the samples, and acetyl chloramphenicol was measured with a b-scintillation counter. RESULTS Effect of ATRA and RAR-g selective binding retinoids on breast cancer cell proliferation We analyzed the effects of ATRA and CD2325, CD2247, CD666 and CD437 on the proliferation of MCF-7, ZR-75-1, T47D, SKBR-3, 734-B and BT-20 breast cancer cell lines (Fig. 2). Under our culturing conditions, ATRA significantly inhibited proliferation of MCF-7, ZR-75-1, T47D and SKBR-3 cell lines but showed no effect on 734-B and BT-20 cells. CD2325 and CD666 exhibited our antiproliferative activity similar to that of ATRA. CD437, mainly at concentrations of 1027?1026 M, demonstrating a strong antiproliferative potency on the RA-sensitive as well as RA-refractory cell lines. On the other hand, the CD2247 compound inhibited proliferation only in MCF-7 and T47D cells. Effects of RAR-g selective binding retinoids in combination with IFN-g on proliferation of breast cancer cell lines In our previous studies we were able to demonstrate a synergistic effect of RA with IFNs with regard to proliferation inhibition. To further substantiate the observation that RAR-g might be strongly involved in proliferation inhibition, we tested the RAR-g selectively binding retinoids CD2325, CD2247, CD666 and CD437 in combination with IFN-g. In order to analyze this drug interaction, we performed isobolograms and calculated the CI values (Table I). CI values ,1 are indicative of synergistic interaction. CD2325, CD2247, CD666 and CD437 in combination with IFN-g demonstrated a strong proliferation inhibition on MCF-7, SKBR-3 and even in the RA-resistant cell line BT-20. The synergistic effect was lowest with the combination of CD437 and IFN-g. In 3 other cell lines, ZR-75-1, T47D and 734B, treatment with CD2325 and IFN-g also led to synergistic proliferation inhibition. Anti-AP-1 activity of RAR-g selective binding retinoids alone and in combination with IFN-g Since antiproliferative action of retinoids may mainly be due to anti-AP-1 activity (Fanjul et al., 1994) we tested whether the RAR-g selectively binding retinoids were able to suppress AP-1 promoter activity and whether IFN-g was able to enhance an anti-AP-1 effect. As shown in Figure 3 (4 independent experiments), on MCF-7 cells, TPA-mediated AP-1 activity was suppressed by IFN-g or ATRA to 75% vs. 57%, respectively. Among the RAR-g selectively binding retinoids, only the CD2325 compound affected AP-1 activity. Only in combination with CD666 did IFN-g exert a significant additional anti-AP-1 effect. To further evaluate the effects of CD2325 on AP-1 activity, we tested 2 RA-sensitive and 2 RA-resistent cell lines. Figure 4 (3?4 independent experiments) shows that only in the 2 RA-sensitive cell lines MCF-7 and SKBR-3 did TPA activate the AP-1 promoter. Constitutive values of AP-1 activity were comparable in all 4 cell lines used (data not shown). In the 2 RA-resistant cell lines BT-20 and Hs578-T, TPA was not able to activate AP-1. Only on MCF-7 cells, RA as well as CD2325 suppressed TPA-mediated AP-1 activity. Interestingly, in SKBR-3, an estrogen receptor-negative (tested, 499 results not shown) but RA-sensitive cell line, neither ATRA nor CD2325 suppressed AP-1 activity. Effects of retinoids and IFN-g on CRABP II promoter activity In our previous report we stated that synergism of retinoids and IFNs may partly be due to IFN-mediated suppression of retinoidinduced expression of CRABP II. Figure 5 shows the results of 6 independent experiments. CD2325 was at least as potent as ATRA on CRABP II promoter activation. IFN-g only slightly suppressed the CD2325-mediated effect, but reduced ATRA-mediated CRABP II activation to 50%. Northern blots also showed CRABP II upregulation by CD2325; this effect was not suppressed by IFN-g (data not shown). Expression of RAR- and CRABP-mRNA in breast cancer cell lines To correlate growth effects on breast cancer cell lines and RAR or CRABP effects, we performed Nothern blots (Fig. 6). RAR-b and CRABP I were not expressed in the breast cancer cell lines we examined. The highest expression of RAR-a was found with MCF-7 cell lines, whereas in T47D, SKBR-3, BT-20 and 734B a comparable lower amount of this transcript was detected. Hs578T was negative for RAR-a. The ranking of RAR-g expression was as follows: MCF-7 5 T47D 5 734B , BT-20 5 Hs578T , SKBR-3. CRABP II transcripts were expressed at a very high extent in MCF-7 and T47D cells, at a lower extent in SKBR-3, BT-20 and 734B and were undetectable in the Hs578T cell lines. Comparing the growth data with RAR and CRABP expression, no correlation could be found. For instance, MCF-7 and T47D cells exhibited comparable expression patterns, while T47D cells were more RA-sensitive. Comparing the highly RA-sensitive SKBR-3 with the RA-resistant 734B cell line, we observed comparable expression patterns, except for a higher RAR-g value in 734B cells. DISCUSSION Combinations of retinoids with IFNs have been reported to be very potent inhibitors of breast cancer cell proliferation (Marth et al., 1986, 1993; Widschwendter et al., 1996). We studied the mechanisms of this synergistic action and found that IFN-g induced expression of RAR-g and suppressed RA-mediated induction of CRABP II, a molecule which may be responsible for sequestration of RA, thus limiting its distribution and biological effects (Widschwendter et al., 1995). Additional evidence for an involvement of RAR-g in retinoid signalling comes from studies with a retinoid-resistant human teratocarcinoma cell line: transfection with a vector overexpressing RAR-g restored ATRA, and induced growth inhibition and differentiation (Moasser et al., 1994). Murine F9 teratocarcinoma cells, which have lost RAR-g expression through targeted gene disruption, do not exhibit the morphologic differentiating effects of ATRA (Boylan et al., 1993). Overexpression of RAR-g in neuroblastoma cells suppresses the malignant phenotype and alters the differentiation potential (Marshall et al., 1995). The present study extends our analyses to the RAR-g selective binding retinoids CD2325, CD2247, CD666 and CD437. All these retinoids demonstrated a high selectivity for binding to RAR-g. The potency of activation of different RARs has been correlated well with binding data in the case of CD437 and CD666 (Bernard et al., 1992). Despite high selectivity for binding to RAR-g, the AC50 value (the retinoid concentration producing half maximal activation) of CD2325 for RAR-g is only 4?6-fold higher than for the 2 other RARs. Our results demonstrate that with the exception of CD2247, all RAR-g selectively binding retinoids effectively inhibit growth of breast cancer cell lines as potently as RA. These retinoids also inhibit proliferation of human melanoma cells (Schadendorf et al., 1994). Correlating the retinoid binding affinity to RARs with retinoid inhibition of growth of ER1 MCF-7 cells, Dawson et al. (1995) observed that RAR-a is the retinoid receptor involved in the inhibition of adherent cell growth. This appears to not contradict our results, since Taneja et al. (1995) demonstrated the possibility 500 WIDSCHWENDTER ET AL. FIGURE 2 ? Effects of retinoic acid and RAR-g selectively binding retinoids on proliferation of breast cancer cell lines. Cell lines were treated with all-trans retinoic acid (?), CD2325 (s), CD2247 (d), CD666 (j) or CD437 (h) in concentrations of 10210?1026 M as indicated on the abscissa. Results are presented as mean number of 6 wells counted in comparison with untreated controls (5100%). The coefficient of variation was always below 10%. For reasons of clarity it is not shown. of a functional redundancy between the various RARs: they reexpressed RAR-g or overexpressed RAR-a in RAR-g-null F9 cells and were able to restore both target-gene activation and differentiation potential. Moreover, the RAR-a-selective antagonist Ro 41-5253 inhibits SKBR-3 cell proliferation alone, but diminishes all-trans-retinoic acid-mediated growth effects while RAR-g SELECTIVE BINDING RETINOIDS AND IFN-g IN BREAST CANCER CELLS TABLE I ? EFFECTS ON PROLIFERATION OF COMBINED TREATMENT OF BREAST CANCER CELL LINES WITH RAR-g SELECTIVELY BINDING RETINOIDS (CD2325, CD2247, CD666 AND CD437) AND IFN-g, INDICATED AS A COMBINATION INDEX (CI)1 MCF-7 SKBR-3 BT-20 ZR-75-1 T47 D 734-B CD2325 CD2247 CD666 CD437 CD2325 CD2247 CD666 CD437 CD2325 CD2247 CD666 CD437 CD2325 CD2325 CD2325 IC50 IC75 IC90 IC95 0.25 0.28 0.01 0.37 0.55 0.29 0.13 0.75 0.59 0.29 0.35 0.93 0.24 0.40 0.89 0.10 0.45 0.01 0.49 0.49 0.36 0.08 0.81 0.54 0.36 0.28 0.88 0.29 0.37 0.65 0.06 0.73 0.01 0.43 0.43 0.47 0.05 0.91 0.57 0.47 0.29 0.86 0.51 0.36 0.49 0.04 0.78 0.01 0.75 0.40 0.58 0.04 0.92 0.61 0.57 0.28 0.85 0.76 0.36 0.41 1CI was calculated on the multiple drug effect equation derived by Chou and Talalay (1984), using computer software. CI , 1, 5 1 and .1 indicate synergism, additive effect and antagonism, respectively. CI values are shown for different inhibitory concentrations (IC). 501 not affecting synergistic interaction of all-trans-retinoic acid plus IFN-g (Marth et al., 1993). Various results concerning expression of RAR-b have been published (Liu et al., 1996, and references therein). We did not detect RAR-b transcripts in control or RA-treated cells (Widschwendter et al., 1995, 1996; Fig. 6), indicating that this receptor does not play a role in inhibiting proliferation in our cell culture system. Fanjul et al. (1994) state that the antiproliferative activity of retinoids may be mediated by their anti-AP-1 activity. To test this hypothesis in our system we performed transfection assays and found that AP-1 activity was stimulated only in RA-sensitive cell lines, and only by TPA. Similar data were published by van der Burg et al. (1995). In contrast to their results, we did not find high constitutive activity of AP-1 in hormone-insensitive cell lines. The SKBR-3 cell line, which is known to be estrogen receptor-negative, exhibits TPA-inducible AP-1 activity. AP-1 activity was suppressed only in the MCF-7 breast cancer cell line by ATRA and CD2325. The other cell lines were refractory to retinoid-mediated AP-1 suppression. IFN-g exhibited modest anti-AP-1 activity. Only in combination with CD666 did IFN-g have a significant additional anti-AP-1 effect. CD437 exerts its antiproliferative effects in an AP-1-independent manner (Shao et al., 1995), CD437 inducing G0/G1 arrest and apoptosis via a unique pathway which appears to FIGURE 3 ? Regulation of TRE-dependent transcriptional activity in MCF-7 breast cancer cells. MCF-7 cells were transfected with the TRE-tk-CAT reporter plasmid and subsequently treated with vehicle only or cultured in the presence of TPA (100 ng/ml) alone or in combination with IFN-g (10 ng/ml), ATRA (1 然/l) or the RAR-g selectively binding retinoids (CD2325, CD2247, CD666 and CD437) (1 然/l) alone or in combination with IFN-g (10 ng/ml) for 24 hr. CAT activity determined in untreated cells was subtracted from activity in treated cells. Bars present the mean of chloramphenicol conversion 6 SEM in TPA 6 IFN-g/retinoid-treated cells compared with TPA-treated controls (5100%). At least 4 independent experiments were done in triplicate. 502 WIDSCHWENDTER ET AL. FIGURE 4 ? Regulation of TRE-dependent transcriptional activity in different breast cancer cells. The RA-sensitive MCF-7 and SKBR-3 and the RA-resistant BT-20 and Hs578T cells were transfected with the TRE-tk-CAT reporter plasmid and subsequently treated with vehicle only (open bars), TPA (100 ng/ml) (solid bars) alone or in combination with ATRA (1 然/L) (hatched bars) or the RAR-g selectively binding retinoid CD2325 (gray bars) for 24 hr. The mean of the conversion of chloramphenicol as measured in at least 3 independent experiments is shown (6SEM), relative to that in nontreated controls (set at one). FIGURE 5 ? Regulation of CRABP II-RARE2-dependent transcriptional activity in MCF-7 breast cancer cells. MCF-7 cells were treated with ATRA (1 然/l) or with CD2325 (1 然/l) alone or in combination with IFN-g (10 ng/ml) or with ATRA (1 然/l) and IFN-g (10 ng/ml) for 24 hr. CAT activity in untreated cells was subtracted from activity in treated cells. Results are presented as the mean chloramphenicol conversion 6 SEM in treated cells in relation to untreated controls (5100%). At least 6 independent experiments were performed in triplicate. involve activation of known downstream effectors of p53 in a p53-independent manner. Treatment of MCF-7 cells with the potent AP-1 activator TPA blocks cell proliferation (Kennedy et al., 1992). In contrast, FIGURE 6 ? Expression of mRNA for RAR-a, RAR-b, RAR-g and CRABP I and II in different breast cancer cell lines. Northern blot analysis was performed on 10 痢 of total RNA as described in Material and Methods. The data are representative of 2 separate experiments. Staining of total RNA with ethidium bromide confirmed the integrity of RNA and showed comparable RNA loading in each lane (photographs not shown for reasons of clarity). retinoids that selectively inhibit AP-1 activity and do not activate transcription also inhibit proliferation (Fanjul et al., 1994). We hypothesize that the 2 major receptor activities, transcriptional activation and anti-AP-1 activity, require 2 different receptor configurations induced or stabilized by the ligand. ATRA, with its flexible polyene side chain, is apparently able to accommodate both receptor configurations (Delescluse et al., 1991). The conformationally restricted retinoids identified by Fanjul et al. (1994) favor the anti-AP-1 configuration. These retinoids fail to inhibit the proliferation of F9 cells and do not induce differentiation, whereas in the T47-D cell line they reduce proliferation as determined by cell counting, but are not as potent as ATRA. The results of Boylan et al. (1993) indicate that RAR-g is involved in the regulation of differentiation-specific genes. Therefore, induction of the antiAP-1 pathway appears to be not exclusively responsible for proliferation inhibition of breast cancer cell lines by retinoic acid. The antiproliferative effects of the RAR-g selectively binding retinoids can be significantly enhanced by IFN-g. IFN-g appears to be able to ??sensitize?? primary retinoid-insensitive cells, namely BT-20 and 734-B, for retinoid action. On the other hand, treatment of MCF-7 cells with the combination of ATRA or RAR-g selectively binding cells and IFN-g did not further enhance anti-AP-1 activity, meaning that there must be an additional factor which regulates proliferation. It is very interesting that SKBR-3, an estrogen receptor-negative breast cancer cell line, in contrast to the other estrogen-independent cell lines (BT-20, 734B), is very sensitive to retinoid-mediated antiproliferative effects. In contrast to MCF-7 cells in the SKBR-3 breast cancer cell line, TPAmediated AP-1 activation was neither suppressed by ATRA nor by CD2325. We looked for differences between SKBR-3 and other estrogen receptor-negative cell lines and found that the ras protooncogene-related protein rhoB is strongly expressed in SKBR-3 but is absent in all other estrogen receptor-negative cell lines examined (de Cremoux et al., 1994). Retinoic acid-mediated regulation of this pathway is the subject of investigations in our laboratory. In our previous work we indicated that IFN-g-mediated suppression of RA-induced CRABP II expression may be partly involved in the synergistic antiproliferative action of the combination of these 2 substances. The induction of CRABP II transcription is RAR-g SELECTIVE BINDING RETINOIDS AND IFN-g IN BREAST CANCER CELLS mediated by RAR-RXR heterodimers bound to DR1- and DR2repeated motifs (Durand et al., 1992). To test the activity of retinoids alone and in combination with IFN-g, we used a DR1-specific reporter plasmid. CD2325 is as potent as ATRA in activating this promoter. The combination of CD2325 and IFN-g only slightly reduces promoter activity, whereas ATRA-mediated activation is reduced to half the maximal values by IFN-g. We therefore hypothesize that activation of the CRABP II gene may be regulated by various pathways depending on the retinoid used, and that DR1 activation may be blocked by IFN-g. Our Northern blot analysis results reveal no correlation between the expression of the RAR- and CRABP-untreated breast cancer cell lines and the response to ATRA or the RAR-g selectively binding retinoids. MCF-7 cells exhibited a relatively high expression of RAR-a and RAR-g compared with SKBR-3; on the other hand, the SKBR-3 cell line is more sensitive to ATRA- and RAR-g selectively binding retinoid-mediated proliferation inhibition than the MCF-7 cell line. Therefore, we conclude that RAR and CRABP expression in untreated breast cancer cell lines is not a good predictor of RA-sensitivity. Finally, we cannot rule out the possibility that RAR-g selectively binding retinoids are taken up to a greater extent than ATRA, and/or are metabolized to a lesser extent, in comparison with the effectiveness of RAR-g selectively binding retinoids vs. ATRA. 503 Our data indicate that RAR-g selectively binding retinoids can be potent antiproliferative agents on breast cancer cell lines, either alone or in combination with IFN-g. Their antiproliferative potency can be explained only in part by anti-AP-1 activity; at least in part, other mechanisms (e.g., activation of differentiation-specific genes or modulation of other growth-inducing factors or receptors) appear to be responsible for growth inhibition. In conclusion, we speculate that these compounds might have clinical importance because of their possibly low toxicity compared with ATRA and other synthetic and natural (e.g., 9-cis RA) retinoids and because of their synergistic interaction with IFN-g. ACKNOWLEDGEMENTS We thank Ms. I. Gaugg for excellent technical assistance, and Dr. E. Preuss for layout. NOTE ADDED IN PROOF After this article was submitted for publication, Fanjul et al. (1996) also reported a potential role for RAR-g in inhibiting breast cancer cells by means of selective retinoids and IFNs. REFERENCES BERENBAUM, M.C., Criteria for analyzing interactions between biologically active agents. Adv. Cancer Res., 35, 269?335 (1981). BERNARD, B.A., BERNARDON, J.-M., DELESCLUSE, C., MARTIN, B., LENOIR, M.-C., MAIGNAN, J., CHARPENTIER, B., PILGRIM, W.R., REICHERT, U. and SHROOT, B., Identification of synthetic retinoids with selectivity for human nuclear retinoic acid receptor g. Bioch. biophys. Res. Comm., 186, 977?983 (1992). BOYLAN, J.F., LOHNES, D., TANEJA, R., CHAMBON, P. and GUDAS, L.J., Loss of retinoic acid receptor g function in F9 cells by gene disruption results in aberrant Hoxa-1 expression and differentiation upon retinoic acid treatment. Proc. nat. Acad. Sci. (Wash.), 90, 9601?9605 (1993). CHOU, T.C. and TALALAY, P., Quantitative analysis of dose-effect relationships: the combined effect of multiple drugs or enzyme inhibitors. Adv. Enzyme Regul., 22, 27?54 (1984). DAWSON, M.I., CHAO, W., PINE, P., JONG, L., HOBBS, P.D., RUDD, C.K., QUICK, T.C., NILES, R.M., ZHANG, X., LOMBARDO, A., ELY, K.R., SHROOT, B. and FONTANA, J.A., Correlation of retinoid binding affinity to retinoic acid receptor a with retinoid inhibition of growth of estrogen receptorpositive MCF-7 mammary carcinoma cells. Cancer Res., 55, 4446?4451 (1995). DE CREMOUX, P., GAUVILLE, C., CLOSSON, V., LINARES, G., CALVO, F., TAVITIAN, A. and OLOFSSON, B., EGF modulation of the ras-related rhoB gene expression in human breast-cancer cell lines. Int. J. Cancer, 59, 408?415 (1994). DELESCLUSE, C., CAVEY, M.T., MARTIN, B., BERNARD, B.A., REICHERT, U., MAIGNAN, J., DARMON, M. and SHROOT, B., Selective high affinity retinoic acid receptor a or b-g ligands. Mol. Pharmacol., 40, 556?562 (1991). DURAND, B., SAUNDERS, M., LEROY, P., LEID, M. and CHAMBON, P., All-trans and 9-cis retinoic acid induction of CRABP II transcription is mediated by RAR-RXR heterodimers bound to DR1 and DR2 repeated motifs. Cell, 71, 73?85 (1992). FANJUL, A.N., BOUTERFA, H., DAWSON, M. and PFAHL, M., Potential role for retinoic acid receptor-g in the inhibition of breast cancer cells by selective retinoids and interferons. Cancer Res., 56, 1571?1577 (1996). FANJUL, A., DAWSON, M.I., HOBBS, P.D., JONG, L., CAMERON, J.F., HARLEV, E., GRAUPNER, G., LU, X.-P. and PFAHL, M., A new class of retinoids with selective inhibition of AP-1 inhibits proliferation. Nature (Lond.), 372, 107?111 (1994). JONAT, C., RAHMSDORF, H.J., PARK, K.-K., CATO, A.C.B., GEBEL, S., PONTA, H. and HERRLICH, P., Antitumor promotion and antiinflammation: downmodulation of AP-1 (fos/jun) activity by glucocorticoid hormone. Cell, 62, 1189?1204 (1990). KENNEDY, M.J., PRESTIGIACOMO, L.J., TYLER, G., MAY, W.S. and DAVIDSON, N.E., Differential effects of bryostatin 1 and phorbol ester on human breast cancer cell lines. Cancer Res., 52, 1278?1283 (1992). LIU, Y., LEE, M.-O., WANG, H.-G., LI, Y., HASHIMOTO, Y., KLAUS, M., REED, J.C. and ZHANG, X.-K., Retinoic acid receptor b mediates the growth- inhibitory effect of retinoic acid by promoting apoptosis in human breast cancer cells. Mol. Cell Biol., 16, 1138?1149 (1996). MARSHALL, G.M., CHEUNG, B., STACEY, K.P., CAMACHO, M.L., SIMPSON, A.M., KWAN, E., SMITH, S., HABER, M. and NORRIS, M.D., Increased retinoic acid receptor gamma expression suppresses the malignant phenotype and alters the differentiation potential of human neuroblastoma cells. Oncogene, 11, 485?491 (1995). MARTH, C., DAXENBICHLER, G. and DAPUNT, O., Synergistic antiproliferative effect of human recombinant interferons and retinoic acid in cultured breast cancer cells. J. nat. Cancer. Inst., 77, 1197?1202 (1986). MARTH, C., WIDSCHWENDTER, M. and DAXENBICHLER, G., Mechanism of synergistic action of all-trans- or 9-cis-retinoic acid and interferons in breast cancer cells. J. Steroid biochem. mol. Biol., 47, 123?126 (1993). MOASSER, M.M., DE BLASIO, A. and DIMITROVSKY, E., Response and resistance to retinoic acid are mediated through the retinoic acid nuclear receptor gamma in human teratocarcinomas. Oncogene, 9, 833?840 (1994). RECCHIA, F., SICA, G., DE FILIPPIS, S., DISCEPOLI, S., REA, S., TORCHIO, P. and FRATI, L., Interferon-b, retinoids, and tamoxifen in the treatment of metastatic breast cancer: a phase II study. J. Interferon Cytokine Res., 15, 605?610 (1995). SCHADENDORF, D., WORM, M., JURGOVSKY, K., DIPPEL, E., MICHEL, S., CHARPENTIER, B., BERNARDON, J.-M., REICHERT, U. and CARNETZKI, B.M., Retinoic acid receptor-g-selective retinoids exert antiproliferative effects on human melanoma cell growth in vitro. Int. J. Oncol., 5, 1325?1331 (1994). SHAO, Z.-M., DAWSON, M.I., SU LI, X., RISHI, A.K., SHEIKH, M.S., HAN, Q.-X., ORDONEZ, J.V., SHROOT, B. and FONTANA, J.A., p53 independent G0/G1 arrest and apoptosis induced by a novel retinoid in human breast cancer cells. Oncogene, 11, 493?504 (1995). SPORN, M.B., ROBERTS, A.B. and GOODMAN, D.S., The retinoids: biology, chemistry and medicine (2nd ed.), Raven Press, New York (1994). TANEJA, R., BOUILET, P., BOYLAN, J.F., GAUB, M.-P., ROY, B., GUDAS, L.J. and CHAMBON, P., Reexpression of retinoic acid receptor (RAR)g or overexpression of RARa or RARb in RARg-null F9 cells reveals a partial functional redundancy between the three RAR types. Proc. nat. Acad. Sci. (Wash.) 92, 7854?7858 (1995). VAN DER BURG, B., SLAGER-DAVIDOV, R., VAN DER LEEDE, B.M., DE LAAT, S.W. and VAN DER SAAG, P.T., Differential regulation of AP1 activity by retinoic acid in hormone-dependent and -independent breast cancer cells. Mol. cell. Endocrinol., 112, 143?152 (1995). WIDSCHWENDTER, M., DAXENBICHLER, G., DAPUNT, O. and MARTH, C., Effects of retinoic acid and g-interferon on expression of retinoic acid receptor and cellular retinoic acid-binding protein in breast cancer cells. Cancer Res., 55, 2135?2139 (1995). WIDSCHWENDTER, M., DAXENBICHLER, G., BACHMAIR, F., MU? LLER, E., ZEIMET, A.G., WINDBICHLER, G., UHL-STEIDL, M., LANG, T. and MARTH, C., Interaction of retinoic acid and interferon-a in breast cancer cell lines. Anticancer Res., 16, 369?374 (1996).