Effects of retinoic acid and fenretinide on the c-erbB-2 expression, growth and cisplatin sensitivity of breast cancer cells.

We investigated the effects of all-trans retinoic acid (ATRA) and fenretinide (4-HPR) on c-erbB-2 expression in SK-BR-3, BT-474 and MCF-7 breast cancer cells and on the growth, differentiation, apoptosis and cisplatin (CDDP) sensitivity of SK-BR-3 cells. It has been reported that oestrogen inhibits c-erbB-2 in oestrogen receptor-positive breast cancer cells. Using ELISA, Western and Northern analysis we have demonstrated that ATRA and 4-HPR exert similar effects down-regulating c-erbB-2 protein and mRNA in c-erbB-2-overexpressing SK-BR-3 and BT-474 and in normally expressing MCF-7 cells. Both retinoids inhibit SK-BR-3 cell growth. ATRA induces cellular enlargement and flattening, suggesting epithelial differentiation. 4-HPR causes nuclear and cytoplasmic condensation, DNA fragmentation and externalization of phosphatidylserine, indicating apoptosis. c-erbB-2 expression/activity has been linked to sensitivity against CDDP. Therefore, combinations of ATRA or 4-HPR with CDDP were tested for their anti-proliferative activity. Retinoid-conditioned cells were either exposed to retinoid and CDDP (schedule I, 'continuous retinoid treatment') or to CDDP alone (schedule II, 'retinoid pretreatment'). This retinoid-conditioning followed by CDDP +/- retinoid yields stronger growth inhibition compared with unconditioned cells, which were exposed to CDDP +/- retinoid (schedule III, 'no retinoid pretreatment'). The inefficacy of schedule III indicates that retinoid-conditioning is essential for the improvement of the antiproliferative effect. The interactions in schedules I and II are synergistic for ATRA and CDDP, but slightly antagonistic for 4-HPR and CDDR However, 4-HPR + CDDP is more effective in growth inhibition than each drug alone.

inhibit carcinogenic transformation and the growth of established tumours. The antiproliferative effects of retinoids are frequently associated with cell differentiation and/or programmed cell death (Bollag et al, 1994;Krupitza et al. 1995). Retinoids have come under the scrutiny of oncologists to assess their potential in cancer prevention and therapy. AlU-trans retinoic acid (ATRA) is effective against acute promyelocytic leukaemia (for review see Fenaux et al. 1997) and 13-cis retinoic acid is effective against cervical cancer and squamous cancer of the skin. The clinical use of retinoids is compromised, however, by the high hepatotoxicity.
Promising results concerning therapeutic efficacy and toxicity have been reported for N-(4-hydroxyphenyl) retinamide (fenretinide, 4-HPR) (Veronesi et al, 1996), which accumulates in the mammary gland and which is currently in clinical trials for the prevention of breast cancer, oral cancer and basal cell carcinoma (Costa et al, 1995). Retinoids bind and activate nuclear retinoic acid receptors (RARs) and/or retinoid X receptors (RXRs), which represent transcription factors that control retinoid-responsive genes. These genes regulate cell growth and differentiation. Compared with ATRA. 4-HPR reveals differential and weaker RARIRXR transactivation (Fanjul et al, 1996). which might explain its low hepatotoxicity. It is possible that 4-HPR activates additional, as yet undefined, signalling pathways (Kazmi et al. 1996). In addition. both retinoids inhibit the AP-1 transcription factor (Fanjul et al, 1994(Fanjul et al, , 1996, which becomes activated upon growth factor signalling. Therefore, a negative interaction between retinoid and growth factor signalling seems to occur. Progression of carcinomas has been linked to the expression of oncogenes, such as c-mvc and c-erbB-2 (also referred to as HER-2 or neu) (Somay et al. 1992;Grunt et al. 1995). At present. c-erbB-2 represents one of the most important oncogenes in breast cancer. c-erbB-2 amplification/overexpression occurs in approximately 25% of breast carcinomas and is associated with an unfavourable clinical outcome. It codes for a 1 85-kDa protein. which belongs to the membrane-anchored type 1 (epidermal growth factor receptorrelated) receptor tyrosine kinases and which becomes indirectly activated by epidermal growth factor-like ligands (for review see Grunt and Huber, 1994). c-erbB-2 can be inhibited by steroids and cytokines (Read et al. 1990: De Bortoli et al. 1992Marth et al. 1992;Kalthoff et al. 1993;Nehme et al. 1995). We and others have demonstrated a negative interaction between the oestrogen receptor and c-erbB-2 (Read et al. 1990;Grunt et al. 1995;Saceda et al, 1996). Recently, we have also shown that c-erbB receptor activation elevates the expression of RAR-a in SK-BR-3 cells (Flicker et al, 1997).
Here, we investigated the effects of retinoids on c-erbB-2 expression in SK-BR-3. BT-474. MCF-7 and MDA-MB-468 breast cancer cells. The responses to ATRA and 4-HPR were 79 further analysed in SK-BR-3 cells with respect to morphology and growth rate. In addition, evidence suggesting that c-erbB-2 expression/activity is associated with alterations of the sensitivity against cytotoxic drugs, such as cisplatin (CDDP) (Hancock et aL 1991;Benz et al, 1993;Arteaga et al, 1994;Pietras et al, 1994) prompted us to examine the effect of ATRA and 4-HPR on CDDP-mediated cytotoxicity in SK-BR-3 cells.
After a 3-day incubation, the test compounds were added.
Test compounds Stock solutions of ATRA (Sigma, St Louis, MO, USA), 4-HPR (gift from Janssen-Cilag, Vienna, Austria), taxol (Sigma) and etoposide (Sigma) were prpared in dimethyl sulphoxide (DMSO). The final concentration of DMSO in the cultures did not exceed 0.1% (vtv). CDDP (kindly provided by Bristol Myers-Squibb, Vienna, Austria) was reconstituted according to the manufacturer's recommendation to a concentration of 1 mg ml-l 0.9% sodium chloride. Stocks were stored light-protected at -80C.

Enzyme-linked immunosobent assay
For quantitative determination of c-erbB-2 protein, the Human neu Quantitative Enzyme-Linked Immunosorbent Assay System (Oncogene Science, Manhasset, NY, USA) was applied using the manufacturer's protocols. Briefly, 1-10 x 105 cells per well were plated in six-well plates (Costar, Cambridge, MA, USA) and grown for 3 days followed by 3 days of steroid depletion.
Subsequently, the cultures were exposed to ATRA or 4-HPR. The protein content in the lysates of trypsinized cells was detrmined according to Bradford (Bio-Rad Laboratories, Munich, Germany) and 0.5-10 jig of total protein was subjected to the assay. The optical densities were determined in a microplate reader and the amount of c-erbB-2 protein was given in arbitrary human neu units (HNU) jig-' total protein.

Western blotting
Cells were plated at 1 x 101 per well in 24-well plates (Costar) and were grown in 1 ml of standard medium to subconfluence followed by 3 days of steroid depletion. Subsequently, the cells were exposed to 10 gm ATRA for 24 h. Preparation of protein samples, electrphoretic separation and transfer were performed as described . pI85c1bB-2 was detected using mouse monoclonal anti-c-erbB-2 (Oncogene Science), whereas tyrosine-speific protein phoshorylation was det ined by mouse monoclonal anti-phosphotyrosine (Upstate Biotechnology, Lake Placid, NY, USA) (1 jg ml-', 4 h, room temperature).

Ntmher blotng
Cells were grown to subconfluence in standard medium in T75 tissue culture flasks (Falcon, Franklin Lakes, NJ, USA). After 3 days of steroid-depletion, the cells were exposed to ATRA or 4-HPR. RNA was extracted with RNAzol B (Cinna/Biotecx, Houston, TX, USA). Processing of the samples, electrphoresis in 1% formaldehyde-containing agarose gels, transfer onto Immobilon S membranes (Millipore, Bedford, MA, USA) and detection of specific mRNAs using random primer-labelled biotinylated cDNA probes were performed as described . A 0.48-kb EcoRl-HindlI fiagment from the c-erbB-2 cDNA inserted into pGEM-3 was used for the detection of c-erbB-2 transcripts and a 1.3-kb EcoRI-Hindi fragment from the GAPDH cDNA inserted into pSP65 was used as internal standard.
Annexin V Approximately 5 x 10' cells were grown for 3 days in 12-well plates (Costar), steroid depleted for 3 days and exposed to ATRA, 4-HPR, taxol or etoposide (10 jiM, 48 h). Detection of phosphatidylseine on the cell surface was performed with annexin V-FITC and evaluated by flow cytomety as described by the manufacturer (Clontech, Palo Alto, CA, USA).
Data analysis for combination treatment Synergism, additivity or antagonism of the drugs was determined by calculating the combination index (CI) using the equation: Clx = (D),/(Dx), + (D),/(Dx), + alD),(D),/(Dx),(Dx), where Clx represents the CI value for x% effect (Dx), and (Dx), are the doses of agents 1 and 2 required to exert x% effect alone, whereas (D), and (D), represent the doses of agents 1 and 2 that elicit the same x% effect in combination with the other agent respectively. a describes the type of interaction: a = 0 for mutually exclusive drugs (similar modes of action), a = 1 for mutually non-exclusive drugs (independent modes of action) (Sacks et al, 1995). The CI values were determined for 50% growth inhibition, and the equation was solved for a = 0 and for a = 1. CI = 1 indicates additivity, CI < 1 synergism and CI > 1 antagonism. In addition, the geometric isobologram method was applied for drug concentrations causing 50% growth inhibition (ICO). The IC,O values of the retinoids and of CDDP were plotted on the x or v axis, respectively, and a line connecting these two points was drawn. Synergism is encountered if the experimental point falls below that line, whereas antagonism occurs if the point lies above it (Sacks et al. 1995).

Expression of c-erbB-2 Spontaneous expression
A c-erbB-2-specific ELISA was used to compare the baseline expression of the c-erbB-2 protein in four mammary carcinoma cell lines (Table 1). Overexpression. defined by > 10 HNU gg-' protein (Nugent et al, 1992), was found in SK-BR-3 and BT-474 cells, whereas MCF-7 cells contained normal levels of c-erbB-2. MDA-MB-468 cells were negative for this oncoprotein.
The effects of ATRA and 4-HPR Exposure of SK-BR-3, BT-474 and MCF-7 cells to 10 giM ATRA for 24 h reduced the c-erbB-2 protein to 40%-60 relative to (C) Sustained downregulation by 10 gm ATRA Upper panels, 30 9g of total RNA was probed against c-erbB-2. Lower panels; stripped filters were rehybrnkzed against GAPDH protein detected by ELISA was proven by immunoblotting demonstrating an ATRA-mediated decrease of pl85c-bB2 in all cell lines tested. This was accompanied by a decreased level of tyrosine-phosphorylated proteins (Figure 1). Inhibition of c-erbB-2 was examined in more detail in SK-BR-3 cells. Cells exposed for 48 h to a concentration as low as 10 nm of ATRA had aleady down-regulated c-erbB-2 protein. Doses between 1 and 10 gm yielded relatively similar degrees of inhibition (49-46% or 58-55 HNU gg-' protein) relative to control (118 HNU ig-' protein) Figure 4 Kintics of 4-HPRF-exiated reduction of c-erbB-2 protein (A, EUSA) and mRNA (B, Nortemrn analysi). (B) Cels were exposed to vehile (-) or to 10 gw 4-HPR (+) (data not shown). In time course experiments, using 1 jM ATRA, the first signs of c-erbB-2 down-regulation were discernible after 24 h and proceeded during the observation period ( Figure 2A). This down-regulation was stable even after removal of ATRA from the culture ( Figure 2B). SK-BR-3 cells express large amounts of the 4.8-kb c-erbB-2 mRNA, which was down-regulated by ATRA in a dose-and timedependent manner relative to GAPDH ( Figures 3A and B). Inhibition of c-erbB-2 mRNA by ATRA occurred as early as 8 h after retinoid addition ( Figure 3B) and remained depressed even after removal of ATRA from the culture ( Figure 3C).
In analogy to ATRA, 4-HPR reduced c-erbB-2 protein and mRNA in SK-BR-3 cells in a dose-(data not shown) and timedependent manner, as demonstrated by ELISA and Northern blotting ( Figures 4A and B).
Morphology SK-BR-3 cells grew as loosely packed monolayers never reaching 100% confluence. One proportion of the cells spread and presented a flattened shape, whereas the other proportion remained rounded ( Figure 5A). ATRA-teated cells increased in size, spread further and demonstrated a flattened shape with multiple cytoplasmic extensions, representing a more mature British Journal of Cancer (1998) 78(1) FLgure 5 Morphology of SK-BR-3 cells exposed (4 days) to vehicle (A), 10 jim ATRA (B) or 10 gm 4-HPR (C). Scale bars 20 jm. Note, ATRA-induced differentiation causes flaitening and spreading (B), 4-HPR-meriated apoptosis causes nuclear and cytoplasnic condensation and celular roundi-p (C) phenotype ( Figure SB). The cells revealed large lacy nuclei that contained large nucleoli and that were surrounded by sizeable flat cytoplasms. Multinucleated cells were frequently seen in these cultures. In contrast, 4-HPR-treated cells rounded up and showed reduced adherence to the substrate ( Figure 5C). Nuclear and cytoplasmic condensation, cellular partition into membrane boundvesicles (apoptotic bodies) and chromatin aggregation at the nuclear membrane was observed in these cultures. These phenotypes were stable for at least 2 weeks after retinoid removal (data not shown).

DNA frmentaton
Control cultures were devoid of cytoplasmic DNA fragments ( Figure 6, lane 3). ATRA-treated cells contain small amounts of DNA fragments (lane 4), which range in size from approximately nucleosomal length fragmentation of cytoplasmic DNA (ladder) was found at molecular sizes of 200 bp and multiples of this unit (lane 5). DNA smears may be induced by necrotic cell loss. whereas DNA 'laddering' represents the hallmark of apoptosis (Trauth et al, 1989).
Annexin V 4-HPR-induced morphology and DNA 'laddering' correlated with the appearance of phosphatidylserine on the cell surface, as demonstrated by staining with annexin V-FITC (Table 2), which represents a marker for apoptosis . In vitro cell growth Treatment with the single agents Cytostasis of SK-BR-3 cells was obtained with l09 M ATRA and complete growth arrest occurred with > I0-1 M ATRA ( Figure 7A). The dose window for ATRA-mediated growth inhibition was fairly wide ranging from 10-9 to 10-5 M. In contrast, 4-HPR caused a sharp decline in cell numbers within 106-61-5 M ( Figure 7B). Inhibition of proliferation occurred slowly with 1O-5 M ATRA, whereas it was immediate (after 1 day) and higher for a similar dose (8 x 106 M) of 4-HPR Cornbined treatment with retinoids and CDDP Two of tiree treatment protocols of combinatons of ATRA and CDDP enhanced the growth-inhibiting effect compared with each drug alone. Strongest inhibition occurred if the cultures had been reated for 2 days with ATRA alone followed by ATRA and CDDP for 3 days ('continuous retinoid tratment', Figure 8A).
Interestingly, a 2-day ATRA preincubation was sufficient to condition the cells for CDDP -even without the concurrent presence of ATRA ('retinoid prentatment', Figure 8B). In contras, simulaneous application of ATRA and CDDP without a preceding exposure to ATRA did not improve the effect of CDDP ('no reinoid petreatment', data not shown). Tberefore, ATRA-conditioning seems to be important for the elevation of the CDDP-mediated growth reduction.
The same protocols were applied for combinations of 4-HPR with CDDP. Again, continuous exposure for 5 days to 4-HPR including CDDP co-treatment for the last 3 days induced the strongest responses (Figue 8C). Some improvement of the CDDP effect was obtained by separate application of 4-HPR followed by CDDP ( Figure 8D), whereas combination of both drugs without 4-HPR prereatment was not superior to CDDP alone (data not shown).
The type of interaction (synergism vs antagonism) was determined for similar and independent mechanisms of drug action.
The CI values for 50% growth inhibition indicate that continuous ATRA treatment and ATRA pretreatment synergistically elevate (CI < 1), whereas 4-HPR slightly antagonizes the CDDP effect (CI > 1) ( Table 3). Analysis using the geometric isobologram method yielded equivalent results (inserts in Figure 8A-D), supporting the conclusions drawn from the CI values. Evaluation of the third teatment schedule (no retinoid pretratment) was not feasible, as no improvement of the CDDP effect was observed.

DISCUSSION
Retinoids inhibit cell proliferation, induce differentiation or trigger apoptosis. The actual response depends on the given retinoid, the type of cells and the growth conditions (Grunt et al, 1991(Grunt et al, , 1992aKrupitza et al, 1995). Two different mechanisms of retinoid action are known. Interaction with RARs/RXRs induces wansactivation of responsive genes and/or inhibition of the AP-1 transcription factor (Fanjul et al, 1994(Fanjul et al, , 1996. Retinoid receptors act as liganddependent transcription factors and reveal striking homologies to the steroid receptors. The molecular processes triggered by retinoid/steroid receptors are different from those induced by c-erbB-2, which transduces protein phosphorylation signals via the mitogen-activated protein(MAP)-kinase cascade to transcription factors such as AP-1. Yet, both signalling pathways control cell growth and differentiation. Both retinoid/steroid receptors and c-erbB-2 membrane receptor tyrosine kinases represent important target structures for antineoplastic intervention. In breast cancer, activation of c-erbB-2 inhibits the oestrogen receptor and, vice versa, stimulation of the oestrogen receptor down-regulates c-erbB-2, demonstrating a negative interaction between these pathways Saceda et al, 1996;Tang et al, 1996). In contrast, activation of c-erbB-2 stimulates the expression of RARa in SK-BR-3 cells (Flicker et al, 1997). These oestrogen receptornegative, c-erbB-2-overexpressing cells (Hynes et al, 1989) contain RARs and are sensitive to retinoids (Pellegrini et al, 1995).
Here, we have demonstated that ATRA and 4-HPR inhibited c-erbB-2 protein and mRNA and protein tyrosine phosphorylation in SK-BR-3, BT-474 and MCF-7 cells, indicating that both agents reduced the malignant characteristics of the cells. Corresponding results have been obtained by Bacus et al (1990) andPellegrini et al (1995). No retinoic acid response element has been identified so far, whereas AP-1, AP-2 and SP-l sites have been found in the regulatory region of c-erbB-2.
D'Souza and colleagues (1993) (Bacus et al, 1990) and of ovarian carcinoma cells (Grunt et al, 1991(Grunt et al, , 1992b(Grunt et al, , 1993. In contrast, the 4-HPR-induced phenotype is reminiscent of apoptosis . Experimental and clinical data indicate that c-erbB-2 expression/activity is associated with altered sensitivity against immunological, endocrine and chemotherapeutic intervention (Hancock et al, 1991;Benz et al, 1993;Kalthoff et al, 1993;Tsai et al, 1993;Arteaga et al, 1994;Pietras et al, 1994;Yu et al, 1996). It has been established that c-erbB-2 overexpression correlates with multidrug resistance of non-small-cell lung cancer (Tsai et al, 1993). In breast cancer, some investigators have reported that c-erbB-2 overexpression/hyperactivation confers resistance against CDDP (Benz et al, 1993), against cyclophosphamide, methotrexate and fluorouracil (Paik, 1992), against tamoxifen (Benz et al, 1993) and against paclitaxel . Others, however, have demonstrated that c-erbB-2 activation elevates the sensitivity of c-erbB-2-overexpressing cells against CDDP, which might be caused by receptor-mediated inhibition of DNA repair enzymes, such as DNA-polymerase-a and -0 (Hancock et al, 1991;Arteaga et al, 1994;Pietras et al, 1994). Interestingly. a DNA repair enzyme activity has been described for the epidermal growth factor receptor (Mroczkowski et al, 1984). Therefore, we wondered whether retinoid-mediated inhibition of c-erbB-2 alters CDDP sensitivity. Retinoids   CDDP in ovarian cancer (Formelli andCleris. 1993: Caliaro et al. 1997). in head and neck cancer (Shalinsky et al. 1995) and in cervical carcinoma (Rustin. 1994). Here, we have demonstrated synergy between ATRA and CDDP. but slight antagonism between 4-HPR and CDDP. However. the antiproliferative response to combinations of 4-HPR and CDDP was stronger than that induced by each drug alone. Preincubation with retinoids was essential for elevated growth inhibition by CDDP. whereas simultaneous application of retinoid and CDDP without retinoid pretreatment did not improve the cell response. ATRA-mediated differentiation and 4-HPR-induced apoptosis were accompanied by reduced c-erbB-2 expression. demonstrating that both processes deliver converging signals for target gene regulation. However. both retinoids differed in their potency to modulate CDDP sensitivity, indicating that additional mechanisms might be responsible for the potentiating effect of ATRA. This is supported by work from Caliaro et al (1997), who suggest that retinoid-mediated alteration of the glutathione-S-transferase activity accompanied by changes in platinum-DNA adduct formation and in epidermal growth factor receptor expression could account for the potentiation of CDDP cytotoxicity in ovarian cancer cells. Retinoids not only represent promising drugs for single-agent anti-cancer treatment. but may be even more beneficial when given in combination with chemotherapeutics. Application of such protocols could bypass the development of resistance and limiting toxicities of retinoids and CDDP.