Outgrowth of BT-474 human breast cancer cells in immune-deficient mice: a new in vivo model for hormone-dependent breast cancer.

The effect of co-inoculation of basement membrane matrix, Matrigel and two human breast cancer cell lines, BT-474 and SK-BR-3, was tested in immune-deficient mice. Both cell lines strongly overexpress c-ErbB-2 protein, whereas only BT-474 is reported to be oestrogen receptor positive. Co-inoculation of Matrigel and BT-474 cells but not of Matrigel and SK-BR-3 cells resulted in tumour formation in bg-nu-xid mice. Oestrogen supplementation greatly enhanced tumorigenicity, but did not seem to be an absolute requirement. In vivo, BT-474 cells grow as a poorly differentiated adenocarcinoma with a doubling time of 9.4 +/- 1.1 days after inoculation into the neck region. A high proliferative activity appears to be compensated by a relatively high rate of cell loss, as BT-474 tumours contain many cells with the typical morphology of apoptotic cell death. Wild-type p53, known to participate in the induction of apoptosis, is absent from the tumours, whereas Bcl-2, known to inhibit apoptosis, is expressed at intermediate levels. BT-474 tumours tend to metastasise to the regional lymph nodes and are capable of forming micrometastatic lesions in the lung. Flow cytometrical analysis of DNA ploidy demonstrated no change in tumours compared with the cell line. Immunohistochemical and flow cytometrical detection of a number of hormone and growth factor receptors, transcription factors, cell adhesion molecules and proteins involved in proliferation and cell death demonstrated no major changes in ploidy and phenotype of tumours compared with the cell line. High expression of the cell-surface molecules c-ErbB-2 and episialin make it a potentially useful model for research in immune therapy. ImagesFigure 2

mice. Otrogen s ementation greatly hanced tumorigecty, but did not seem to be an absolute In vivo, BT474 cells grow as a poorly differentiated with a doubling timc of 9.4 1.1 days after inoculation into the nek regkoL A high proliferative activity appears to beco by a relatvely high rate of cdl loss, as BT-474 tumours contam many ceils with the typical morphology of apoptotic cell death. Wild-type p53, known to participate in the inductin of apoptosis, is absent from the tumours, whereas Bd-2, known to inhibit apoptosis, is expresse at interediate kvels, BT-474 tumours tend to mttasase to the regonal lymph nodes and are of formming eastatic lesions m the hmg. Flow cytometical nalysis of DNA ploidy no change in tmmours compared with the cell hne.
Immunohistochemical and flow cytometical detection of a number of hormone and growth factor rceptors, cell adhson molecules and proteins invohled in proliferation and cal death demonstrated no major chang in ploidy and phenotype of tumours compared with the cell ine. High expression of the cell-surface molcules c-ErbB-2 and epialin make it a potentially useful model for research in immune therapy.
werwoi. brast cancer, immune-deficnt mice; Matngel; tumongemcity; oestrogen receptor, c-erbB-2 Heterotransplantation of human cancers into immunedeficient mice can be used to test various anti-cancer therapies (Giovanella et al., 1978;Sebesteny et al., 1979). For a number of tumour types, such as colon cancer, melanoma and pancreatic cancer, orthotopic implantation of primary tumours in these animals has been reported to improve significntly both take rate and expression of metastatc phenotype (Cornil et al., 1989;Fu et al., 1991Fu et al., , 1992. Primary breast cancer, however, lacks tumorigenicity, even when implanted orthotopically. Primary tumours that do grow in immune-deficient mice are predominantly of the hormone-independent phenotype. Furthermore, only a minonty of established breast cancer cel lines are oestrogen receptor (ER) positive (Fogh et al., 1977). At present, in vivo/in vitro studies of hormone-dependent breast cancer largely depend on the MCF-7, T47D and ZR-75-1 cell lines (Shafie andLiotta, 1980, Leung andShiu, 1981;Weckbecker et al., 1992;Yue and Brodie, 1993). These cell lines have been established from pleural effusions and are tumorigenic in oestrogen-suppkmented immundicient mice. Recently it has been reported that co-inoculation of dispersed primary tumours or of cultured tumour cels with Matrigel or fibroblasts enhas tumour take and growth of various types of cancer in immune-deficient mice (Horgan et al., 1987;Fridman et al., 1990Fridman et al., , 1991Fridman et al., , 1992Chung, 1991;Pretlow et al., 1991;Albini et al., 1992;Noel et al., 1992;Paaniti et al., 1992;Mehta et al., 1993;Sterling-Levis et al., 1993;Topley et al., 1993;Yue and Brodie 1993;Bao et al., 1994). The aim of the present study was therefore to evaluate whether this approach would increase the take rate of breast cancer cell lnes known to show little or no tendency to produce tumours in these mice. For this purpose two human breast cancer cell lines were selected, SK-BR-3 and BT-474, the latter (isolated in 1976 from a sold invasive ductal carcinoma;Lasfargues et al., 1978) reported to be ER positive (Lupu and Lippman, 1993). A emnarkable feature of these cell ies is that they show strong c-erbB-2 oncogene overexpresson; c-erbB-2 is overexpr 2to 30-fold in 30% of human breast cancers (Elledge et al., 1992) and has been reported to be an important prognostic factor in node-negative breast ancer (Pavelic et al., 1992). Stable and exclusive c-erbB-2 overexpression by malignant cells (Niehans et al., 1993) and profound effects of signal transduction through the c-ErbB-2 pathway (Harwerth et al., 1993;Lupu and Lippman 1993) make this receptor protein an interesting target for immunotherapy (Hynes, 1993). Therefore, we believe that hormone-reponsive o ivo models with c-erbB-2-overexpressing human breast cancer cells would be useful in the development of novel therapies for breast cancer.
In this report we described formation of tumours from BT-474 cells in immune-deficient mice supplemented with oestrogen. We compared take rates of orthotopically and subcutaneously inoculated tumour cells and evaluated growth kinetics and metastatic behaviour. Apart from expression of c-ErbB-2 protein and oestrogen receptor in BT-474 cells, we analysed DNA ploidy, cell cycling activity and apoptosis and expression of fed with gamma-irradiated rodent food and water ad libitmn and handled under laminar flow biocontainment to prevent contamination.

Cel lins
The SK-BR-3 and BT-474 cell lnes were kindly provided by  (1994). Briefly, for staining of membrane-associated antigens, cells were fixed with 1/% paraformaldehyde and permeabilised with 40 jag of L<lysophosphatidykholine (Sigma) per 10' cels (10 min at 4-C). Depending on loalisation of the antigen at the inner or outer membrane, cells were fixed and permeabilsed before or after incubation with primary and secondary antibody respectively. For ining of antis locaised in the cytosol BT 474 kud cr~e gml p mdIin b=P gfk nk HJ van Skbn etFi 23 or nucleus, cells were fixed with 1% paraformaldehyde (5 min at 4-C) followed by 100% methanol (10 min at -20C). Incubation with primary monoclonal antibodies (30 min on ice), was followed by another 30 min incubation with fluorescein isothiocyanate (FITC)or R-phycoerythrin (RPE)-labelled isotype-specific, goat anti-mouse antibody (GAM; Souther Biotechnology Associates, Birmingham, AL, USA). Monoclonal antibodie used are listed in Table  IH. Control samples were incubated with PBS alone and isotype-secific secondary antibodies alone. Samples were measured on a FACScan flow cytometer (Becton Dickinson, San Jose, CA, USA). FITC and propidium iodide (Sigma) fluorescene was measured using, r ely, a 530/30 nm (FLI) and a 585/42 mm (F12) bandpass filter and a minimum of 10 000 events were counted. The CellFit software was used for double-fluorescence measurement of DNA and MIB-1 or PCIO antigens respectively.

DNA ploidy
The DNA index from frozen sections and paraffin-embedded tissue was determined using the methods described by Vindelov et al. (1983) and Hedley et al. (1983). Chicken red blood cells (CRBCs) were added to each sample as a reference. The DNA index (defined as human aneuploid G0.//human diploid Goll) of BT-474 cells was calculated using the ratio human diploid G11 peak/CRBC. Samples were measured on a FACScan flow cytometer (Becton Dickinson) and analysed using MODFIT software (Verity Software House, Topsham, ME, USA).

Immunohistochemistry
To compare expression of antigens in BT-474 tumours with flow cytometical analysis of the cell line, frozen sections of tumour were fixed according to the protocol used for flow cytometry. Standard DAB staining with biotinylated goat anti-mouse bridging antibody (Becton Dickinson) and streptavidin-biotinylated horseradish peroxidase complex (Becton Dickinson) was used for detection of bound primary antibodies. Methyl green dye was used for background staining.
a b livn _ rapid outgrowth of BT-474 tumours with a mean doubling time of 9.4 ± 1.1 days. Remarkably, this was only observed after inoculation into the neck region; only one of six animals also developed a small, but not progressively growing, tumour in the posterior fat pad. None of the animalg inoculated with SK-BR-3 cells (with or without hormone supplemetation) developed a progressively growing tumour in the neck; only one out of ten animals developed a nonprogressively growing tumour with a vohlme of 280 mm3 in the mammary fat pad (Table I).
Because co-inoculation of various types of tumour cells with fibroblasts has been reported to increase take rate in immune-deficient mice, two groups of four animals were inoculated with an equal mixture of tumour cdls and tuLmour-derived fibroblasts in Matrigel (Table I and Figure  1c), resulting in two progressively growing  None of the animals inoculated with SK-BR-3 cells developed a tumour.
Effect of inulatin dose and site on BT-474 ntnour formation Tlhree groups of four oestrogenised animls were inoculated with I0, 10 and 103 BT-474 cells premixed with Matrigel. Animals were inoculated subcutaneously in the neck and anterior mammary fat pad (Table H). Decreasing the number of cells inoculated resulted in a longer latency period, but had no effect on growth rate of tumours. Orthotopic inoculation of BT-474 cells in the mammary fat pad impaired tumour formation and growth rate compared with inoculation in the neck. Even though progvel growing tumours developed in the neck region of ten animals, none of the tumours growing in the mammary fat pad reached a size of more than 400 mm3.  = 2.52). The percentage of cells in Sphase as determined from these histograms was almost identical in BT474 cells in vitro and in vivo, 27.2% and 24.94% respectively. The percentage of diploid mouse cells in tumours varied between 11% and 40%.
We used a panel of 19 antibodies to further characterise BT-474 tumours and to detect possible phenotypic changes in vivo compared with cultured BT-474 cells. Immunohistochemical staining of frozen and paraffin sections from a BT-474 tumour growing progressively in the neck was used for characterisation of BT474 cells growing in vivo (Table  III). Because BT-474 adhered very weakly to various tissue Effect of BT-474 inoculum size on tumour take rate. Approximately 1 x 105, 1 x 10 or 1 x 103 cells were premixed with Matrigel and injected in both the neck and anterior mammary fat pad of animals receiving oestrogen supplementation. 3Tumours were all > 100 mm3 and <400 mm3. bOne animal developed two tumours a section slides used for immunohistochemical staining of cultured cells, we used flow cytometry to determine expression of antigens on the cells. Results of flow cytometric analysis are shown in Figure 4. For most antigens a good correlation exists between expression in vitro and in vivo. BT474 cells showed high expression of PgR, c-ErbB-2 (Figures 2d and 4), vitronectin receptor (VnR), Ep-CAM (EGP40), episialin (MUC1), p53, Bcl-2 and transferrin receptor. Expression of EGFR and c-Myc in vivo was very low, even though there was a clear expression of these antigens in vitro (Figure 4). Furthermore, 32% of cells growing in vitro expressed vimentin (Figure 4), and a population of 32% of cells expressed lower levels of Ep-CAM (Figure 4) in vitro. BT474 cells expressed high levels of the proliferation markers Ki-67 and PCNA both in vivo (48.47 and 99.02% positivity respectively) and in vitro (99.29 and 97.09% positivity respectively).

Oestrogen receptor expression on BT-474 cells grown in vivo and in vitro
We could not detect the presence of ER in BT474 tumours using immunohistochemistry on frozen sections or on paraffin sections using the H222 or LHl antibody. This seemed to be in contrast with the high expression of progesterone receptor (PgR), significant growth inhibition by anti-oestrogen in vitro (data not shown) and growth stimulation by oestrogen in vivo. Subsequently, enzyme-linlked immunoassay (EIA) using the LHl antibody and ligandbinding asssay were performed. Again BT474 tumours were found to be negative using both EIA and ligand-binding assay, but the BT474 cell line, in vitro, was found to be ER negative by EIA (3 fmol mg-' protein) and ER positive by ligand-binding assasy (31.7 fmol mg-' protein).  Cell cycle kinetics: comparison of tumours growing in the neck and in the mammary fat pad We investigated whether differences in growth rate between tumours growing in the neck and mammary fat pad were correlated with differences in growth fraction or apoptotic fraction. These fractions were determined by immunohistochemistry with the MIB-1 antibody against Ki-67 and in situ end-labelling of DNA strand breaks respectively. Growth fractions were quantitated by counting a total of 1000 cells in every section. Apoptotic fractions were determined with an automated image analysis system. Surprisingly, no significant differences were found in growth fractions or apoptotic fractions of tumours growing in the neck compared with the mammary fat pad of the same mouse (Table IV).

Dimss
While useful xenotransplantation models have been developed for other primary tumours Cornil et al., 1989;Fu et al., 1991Fu et al., , 1992, this approach remains very difficult for primary hormone-dependent breast cancer. The only hormone-responsive cell lines for which successful xenotransplantation has been described are the MCF-7, ZR-75-1 and T-47D cell lines (Shafie and Liotta, 1980;Leung and Shiu, 1981;Weckbecker et al., 1992;Yue and Brodie, 1993), which are all derived from pleural effusions. In this study we show that a hormone-responsive cell line derived from a primary breast carcinoma, BT-474 (Lasfargues et al., 1978), can grow in bg-nu-xid mice when cells are co-inoculated with Matrigel. The increase in tumorigenicity caused by mixing cells with Matrigel is so dramatic that even inoculation of approx- IgG2a ND 'Summarised from flow cytometric data (Figure 4). b32% of the cells expressed lower or higher levels respectively. 9Determined on paraffin sections. dResults stated in text. ND, not determined. Antigen expression in frozen sections and cell lines was scored semiquantitatively, scores ranging from negative (-) to high expression ( + 5). +1 is +, +5is +++++.
imately 1000 cells results in formation of tumours. Because the main component of Matrigel is the basement membrane protein laminin, the capability of BT-474 cells to bind to this protien using receptors such as 4f4-integrin, is likely to contribute to this effect of Matrigel on tumorigenicity. It has been hypothesised that Matrigel, and especially proteolytic fragments of laminin, plays a role in stimulating angiogenesis (Fridman et al., 1992), and therefore is responsible for the outgrowth of inoculated tumour cells. Furthermore, matrixassociated growth factors (e.g. fibroblast growth factors; tissue plasminogen activator) within Matrigel may also enhance tumorigemncity.
Although it does not seem to be an absolute requirement, oestrogen supplementation greatly enhances tumorigenicity. Removal of oestrogen pellets from animals with established BT-474 tumours resulted in tumour regression accompanied by marked apoptosis (GT Colbern, personal communication), indicating that BT474 tumours are dependent on oestrogen for their growth. Similarly, oestrogen-dependent MCF-7 cells have been reported to retain the ability to activate a programmed cell death pathway following oestrogen ablation (Kyprianou et al., 1991). Our findings are therefore consistent with the hypothesis that oestrogen not only stimulates proliferation, but may also serve as a strong survival factor for breast cancer cells. Addition of fibroblasts, reported to increase tumongemcity of several cell lines in immune-deficient mice (Horgan et al., 1987;Chung, 1991;Noel et al., 1992), had no additional effect. A number of reports described an increased take rate of breast tumour cells after inoculation into the mammary fat pad (Miller et al., 1981;White et al., 1982;Miller and Mclnerney, 1988;Price et al., 1990;Elliott et al., 1992), but in the present study we observed a higher take and growth rate of BT-474 cells in the neck region than in the mammary fat pad. Because no difference was observed in expression of the cellular proliferation marker Ki-67 between two tumours derived from the neck and mammary fat pad respectively, a difference in cell loss was likely to be responsible for this phenomenon. However, we did not find a significant difference in the number of apoptotic cells detected by in situ end-labelling of DNA strand breaks, making it difficult to explain observed differences in tumour growth rate. We cannot exclude the possibility that physical differences such as subcutaneous  'Tumour growing in the mammary fat pad. Growth fracton was defined as the percentage of Ki-67 positive cells. A total of 1000 cells were counted in every section. Apoptotic fraction was defined as area in the sections staining brown after in situ end-labelling of DNA stand break.s. space and blood supply have a negative effect on tumour growth in the mammary fat pad. In vivo BT-474 cells grow as a poorly differentiated adenocarcinoma, metastasising to lymph nodes and capable of forming micrometastatic lesions in lung. No major differences were seen in histology of primary BT-474 tumours and lymph node metastases. Flow cytometric analysis of the DNA content of BT-474 cells demonstrated the DNA ploidy of tumours growing in bg-nu-xid mice to be identical to that of the cell line. To determine if the phenotype had changed and to characterise BT-474 tumours further, we evaluated the presence of various markers. For most antigens tested, a good correlation between expression in vitro and in vivo was found, indicating that no major phenotypic changes had occurred. Because tumours were not routinely passaged in immune-deficient mice, no data are available on long-term stability of BT-474 tumours. However, the high tumorigenicity of BT-474 cells in this model allows induction of many tumours from the same passage of cultured cells if a sufficient number of cells are stored in liquid nitrogen. Additional advantages of inoculation of cultured cells are constant inoculum size and the absence of heterogeneity caused by inter-and/or intra-tumour heterogeneity often present in passaged tumours.
The BT-474 cell line was originally reported to be ER negative, but more recent studies have reported it to be ER positive, as detected by radioligand-binding assay (Koga et al., 1990;Lupu and Lippman, 1993). In our laboratory we were not able to detect ER expression on BT-474 tumours using immunohistochemistry, although the biological behaviour of BT-474 cells in vivo and in vitro was oestrogen responsive. Subsequently, a ligand-binding assay clearly demonstrated the presence of intermediate levels of ER in the BT-474 cell line, while an enzyme immunoassay performed on the same samples faied to detect significant amounts of ER. The low ER expression detected in tumour using the ligand-binding assay may have been caused by high level of endogenous oestrogen owing to hormone supplementation. Recent studies have demonstrated, in both breast tumour cell lnes and primary breast tumours, presene of variant ER mRNAs, probably resulting from alternative splcing (Castes et al., 1993). Tlhrefore, a possible explanation for the conflicting results in our study may be that, although ER is functional, it cannot be detected immunohistochemically owing to an alteration of the epitope recognised by the H222 and LHI antibodies. The functionality of the ER is dearly demonstrated by sensitivity of this cell ine to anti-oestogens (data not shown), by high expression of PgR, which is thought to be regulated by oestrogen (Horwitz, 1993), and by tumour gression after oestrogen deprivation (GT Colbem, personal communication).
A remarkable feature of the BT-474 cell line is its overexpression of c-erbB-2, an oncogene overexpressed in 30% of human breast cancers (Elliedge et al., 1992), almost half of which are also ER positive (Pavelic et al., 1992;Hynes, 1993). Interestingly, evidence is increasing that c-erbB-2 overexpression is associated with tumour cell resistance to NK-cell activity as well as treatment with anti-oestrgens (Hudziak et al., 1988;Hynes, 1993;Lichtenstein et al., 1993;Wiltschke et al., 1994). In this context it is interesting to note that studies are emerging demonstrating a negative feedback loop between sign transduction through ER and c-ErbB-2 (Warri et al., 1991;Dati et al., 1993;Read et al., 1993). Ligation of ER with oestrogen inhibits the expression of c-ErbB-2. In contrast, the c-ErbB-2 ligands gp3O and p75 have been reported to down-regulate in a dose-dependent manner the expression of ER in BT474 and MCF-7 cels, when using hormone-depeted medium (Colomer et al., 1992). Moreover, growth of BT474 cells has been reported to be stimulated by gp3O in oestrogen-depleted and inhibited in oestrogen-suplemented medium (Grunt et al., 1994). Signal transduction via c-ErbB-2 may thus be responsible for escape from anti-oestrogen treatment in a simila way as has been hypothesised for other growth factor receptors, such as epidermal growth factor receptor (EGFR) and insulin-like growth factor I (IGF-I) recptor (Arteaga et al., 1989, Murphy andDotzlaw, 1989;Gill et al., 1991;Ignar-Trowbridge et al., 1992;Long et al., 1992). Also, down-regulation of ER mediated by c-ErbB-2 ligands may dcrease the availability of ER in vivo, giving a possible explanation for our failure to detect ER in BT474 tumours. EGFR could be clearly detected in BT-474 cell monolayers, but not in frozen sections from solid tumours. A similar discrepancy was found for the presence of c-Myc protein, although this could have been caused by rapid degradation of this protein after excision of tumours (Hann and Eisenman, 1984). The aJ4rintegrin was present in frozen sections, but could not be detected in cultured BT474 cells using flow cytometry.
As ited by various proliferation markers and by high S-phas dete by DNA flow cytometry, BT-474 tumours show high proliferative activity. However, compared with the number of proliferating cells, the doubling time of BT474 tumours was relatively long, inditing the existenm of a high rate of cell turnover. Use of in situ end-labelling of fragmented DNA revealed that many cells displayed the classic morphology of apoptosis, with chromatin condensadon, nuclear fragmentation and cell shinage. Other cells had a more pycnotic appearance with round, condensed nuclei and little cytoplasm. Most strikgly, some apoptotic cells were engulfed by other tumour cells in a process known as tumour emperpoesis (Tsunoda et al., 1992).
The high rate of apoptosis, even in rapidly growing BT-474 tumours with high PCNA and Ki-67 indices, prompted us to analyse some factors that could be involved in this process.
The product of the c-myc oncogene has been reported to drive both proliferation and apoptosis (Evan and Littlewood, 1993), whereas increased expression of the bcl-2 gene could possibly enhance survival of cells otherwise doomed to die. In many cell types, Bcl-2 has been reported to counteract the induction of cell death by various treatments (Kamesaki et al., 1993;Reed, 1994), including apoptosis induced by overexpression of c-myc (Bissonnette et al., 1992). The intermediate lvel of Bcl-2 expression in BT-474 cells raises the question of whether the level of Bcl-2 is not sufficient to inhibit apoptosis effectively in BT-474 tumours, or whether this type of apoptotic cell death is simply not modulated by Bcl-2, but by one of its recently discovered variants (Boise et al., 1993;Oltvai et al., 1993).
Another factor possibly involved in the induction of apoptosis is the product of the p53 tumour-suppressor gene. Mounting evidence shows that wild-type p53 protein is important for the induction of apoptosis in cells with a signifant amount of DNA damage, e.g. as a result of Xradiation or chemotherapy (Lowe et al., 1993). Moreover, wild-type p53 has recently been reported to play a role in the induction of apoptosis by growth factor deprivation (Zhu et al., 1994), possibly by down-regulating Bcl-2 and upregulating Bax protein expresson (Haldar et al., 1994;Miyashita et al., 1994). It has been reported that BT-474 cells bear a missense mutation in the p53 gene and show loss of heterozygosity (Bartek et al., 1990) at this locus. The effect of absence of wild-type p53 in this cell lie on cell death regulation therefore needs further clarification.
In conclusion, BT-474 cells offer an interesting opportunity to investigate various aspects of growth regulation and dissemination in oestrogen-dependent breast cancer. Moreover, the capacity of this cell lne to grow and metastasise in immune-deficient mice in combination with high expression of oell-surface proteins such as c-ErbB-2 and esialin make it a potentially usefu in vitro/in vivo model for research in immune therapy.