Oncogene (2004) 23, 8310–8319 & 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

Expression of the NF-jB-responsive BTG2 is aberrantly regulated in breast cancer

Hirofumi Kawakubo1, Jennifer L Carey1, Elena Brachtel2, Vandana Gupta1, Jeffrey E Green3, Paul D Walden4 and Shyamala Maheswaran*,1

1Department of Surgical Oncology, Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114, USA; 2Department of Pathology, Massachusetts General Hospital, Boston, MA 02114, USA; 3Laboratory of Cell Regulation and Carcinogenesis, NIH, Bethesda, MD 20892, USA; 4Department of Urology, NYU Medical Center, New York, NY 10016, USA

BTG2, a -inducible antiproliferative gene, is stimu- rat PC12 cell line during neuronal differentiation lated in breast cancer cells by activation of nuclear factor (Bradbury et al., 1991), and in mouse 3T3 cell line by kappa B (NF-jB). In rat mammary glands, BTG2 is the tumor promoter tetradecanoyl phorbol acetate expressed in epithelial cells and levels decreased during (Fletcher et al., 1991). The rat and mouse homologs of pregnancy and lactation but recovered during involution. BTG2 are designated TIS21 and Pc3, respectively Estrogen and progestin suppress BTG2 expression, (Fletcher et al., 1991). BTG2 is expressed predominantly suggesting that these steroids, which stimulate prolifera- in the epithelium of many tissues (Melamed et al., 2002) tion and lobuloalveolar development of mammary epithe- and can be regulated in response to DNA damage lial cells, may downregulate BTG2 in the mammary gland induced by UV, ionizing radiations and adriamycin, and during pregnancy. Consistent with the report that BTG2 by NGF, interleukin-6, 12-O-tetradecanoylphorbol-13- inhibits expression, suppression of BTG2 acetate, epidermal growth factor (EGF), serum and mRNA in the mammary gland during gestation, and by cAMP (Matsuda et al., 2001; Tirone, 2001). The estrogen and progestin, correlated with stimulation of expression of BTG2 is inducible by p53(Rouault et al., cyclin D1. Ectopic expression of BTG2 inhibited breast 1996), and the presence of binding sites for nuclear cancer cell growth by arresting cells in the G1 phase, an factor kappa B (NF-kB), AP1 and GATA1 in the effect reversed by cyclin D1. BTG2 expression was very promoter of BTG2 (Duriez et al., 2002) suggests that it low or undetectable in human breast cancer cell lines may also be responsive to stimuli that can activate compared with nontumorigenic mammary epithelial cells, transcription factors other than p53. and nuclear expression of BTG2 was absent in 65% of There is a strong correlation between BTG2 expres- human breast tumors compared with adjacent matched sion and negative regulation of growth: BTG2 is normal glands. Spontaneous mammary tumors arising in a expressed during the G0/G1 transition and levels decline mouse model with targeted expression of the early region as cells enter the (Rouault et al., 1996; of the SV40 large tumor Ag demonstrated loss of BTG2 Iacopetti et al., 1999). Stable expression of BTG2 in very early during the tumorigenic process. Thus cells impairs G1 to S progression (Montagnoli et al., deregulation of BTG2 may be an important step in the 1996; Lim et al., 1998a; Guardavaccaro et al., 2000). development of mammary tumors. Conversely, BTG2-deleted cells continue to proliferate Oncogene (2004) 23, 8310–8319. doi:10.1038/sj.onc.1208008 in response to DNA damage in comparison with control Published online 20 September 2004 cells, which growth arrest (Rouault et al., 1996). Furthermore, the detection of BTG2 in tumors was Keywords: BTG2; breast cancer; cell cycle; cyclin D1; associated with decreased aggressiveness (Cmarik et al., NF-kB 1994). Recently, Kuo et al. (2003) demonstrated that BTG2 mRNA and other family members including BTG1 and BTG3were strongly induced by the tumor suppressor p19Arf (p14 in humans) in NIH3T3 cells conditionally expressing the P19Arf transgene1. Introduction BTG2 prevents proliferation through Rb-dependent (Guardavaccaro et al., 2000) and -independent (Lim BTG2 was originally identified as an immediate early et al., 1998a) mechanisms. It represses the transcription gene stimulated by the nerve growth factor (NGF) in the of cyclin D1, an effect that lowers the kinase activity of the CDK4/cyclin D1 complex leading to hypopho- *Correspondence: S Maheswaran, Department of Surgical Oncology, sphorylation of Rb (Guardavaccaro et al., 2000). Jackson 904, Massachusetts General Hospital and Harvard Medical Furthermore, its ability to physically interact with School, 55 Fruit Street, Boston, MA 02114, USA; E-mail: [email protected] CDK1 and CDK4 modifies the activity of CDK4/cyclin Received 24 March 2004; revised 17 June 2004; accepted 29 June 2004; E complex, which is critical for G1 to S transition in Rb- published online 20 September 2004 negative cells (Lim et al., 1998a). The antiproliferative BTG2 expression in breast cancer H Kawakubo et al 8311 function of BTG2 can also be explained by its interaction with PMRT1 (protein arginine methyltrans- ferase of type I) (Lin et al., 1996; Lim et al., 1998b), which through methylation of may be involved in the transduction of antiproliferative and differentia- tion signals, and Caf1 (a homolog of the yeast Caf1/ Pop2, a component of the CCR4 transcription com- plex), a regulator of a number of involved in cell cycle regulation and progression (Rouault et al., 1998). BTG2 also associates with HoxB9 and enhanced HoxB9-dependent transcription of antiproliferative genes such as NCAM (Prevot et al., 2000; Sakaguchi et al., 2001). The BTG2 gene localizes to chromosomal 1q32 (Duriez et al., 2002), and loss of heterozygosity at 1q23– 32 appears to be one of the lesions associated with Figure 1 Activation of NF-kB induces BTG2. (a) T47D cells and development of breast cancer (Chen et al., 1989). T47D cells expressing the dominant-negative IkB-a transgene were However, little is known about the role of BTG2 in treated with 35 nM MIS and total RNA (7.5 mg) was analysed by mammary gland development and tumorigenesis, and so Northern blot. (b) T47D cells and those expressing dominant- far no mutations in the BTG2 gene have been reported negative IkB-a were treated with TNF-a for 2 h and total RNA (7.5 mg) was analysed for BTG2 expression. Hybridization to 18S is in breast cancer. In this paper, we demonstrate that shown activators of the NF-kB pathway such as the Mullerian inhibiting substance, a member of the transforming growth factor (TGF) b superfamily, and tumor necrosis activity induced by MIS (Segev et al., 2000). TNF-a,a factor-a (TNF-a) induce BTG2 expression in breast cytokine that robustly activates NF-kB signaling, also cancer cells. In the rat mammary gland, BTG2 expres- induced BTG2 expression in breast cancer cells through sion dramatically declined during pregnancy, while an NF-kB-dependent mechanism (Figure 1b). cyclin D1, a gene reported to be suppressed by BTG2, was upregulated. Estrogen and progestin suppressed BTG2 expression in the mammary gland is suppressed BTG2 mRNA in vitro, suggesting that BTG2 expression during pregnancy in the mammary gland may be regulated by these steroids, which are involved in growth and differentia- Since BTG2 expression is associated with regulation of tion of the breast. Consistent with its antiproliferative growth and differentiation, we analysed the expression activity, BTG2 suppressed breast cancer cell growth, of BTG2 in the mammary gland during postnatal and its expression was deregulated in mammary tumors. morphogenesis. During pregnancy, under the influence Thus BTG2 may be a negative regulator of mammary of luteal and placental sex steroids and other hormones, epithelial cell growth. the epithelial cells undergo extensive proliferation resulting in a marked increase in ductal, lobular and acinar growth. At the completion of lobuloacinar growth, terminal differentiation of the secretory cells Results leads to synthesis and secretion of milk components. Activation of NF-kB induces BTG2 expression in breast During weaning, the glandular epithelial cells of the cancer cells mammary gland undergo extensive apoptosis, and the mammary gland returns to a histologic state similar to BTG2 was identified as a target gene induced by MIS in the resting state (Nandi, 1958). human breast cancer cells using an HG-U95Av2 BTG2 mRNA was readily detectable in the mammary oligonucleotide array. Northern blot analysis of RNA glands of virgin rats and expression was gradually isolated from T47D cells confirmed the induction of downregulated in the rat mammary gland during BTG2 by MIS (Figure 1a). Since we had previously pregnancy and there was no detectable BTG2 expression demonstrated that MIS induces NF-kB DNA binding in the lactating glands but levels recovered during activity (Segev et al., 2000, 2001, 2002), and the BTG2 involution (Figure 2a). Phosphorimaging of band promoter is reported to have an NF-kB consensus DNA intensities from three animals per time point demon- binding element (Duriez et al., 2002), we tested whether strated a 50% decline on gestational day 5 (G5) to 30% stimulation of BTG2 by MIS was mediated through on G11 and to 15% on G20 compared to that in virgin activation of the NF-kB pathway. As shown in animals (set at 100%; Figure 2b, n ¼ 3). Figure 1a, upregulation of BTG2 mRNA by MIS was The localization of BTG2 protein in the mammary mitigated by the expression of dominant-negative IkB-a, gland determined by immunohistochemical analysis was suggesting that activation of the NF-kB pathway was consistent with the results obtained by Northern required for this process (Figure 1a). We had previously analysis. As seen in Figure 2c, BTG2 protein expression demonstrated that expression of dominant-negative was detectable in the nuclei of myoepithelial and luminal IkB-a in T47D cells abrogates NF-kB DNA binding epithelial cells of the ducts and lobules of the mammary

Oncogene BTG2 expression in breast cancer H Kawakubo et al 8312

Figure 2 Regulation of BTG2 expression during postnatal morphogenesis of the mammary gland. (a) Total RNA (7.5 mg) isolated from mammary glands of 8-week-old virgin, pregnant (G: gestation; G5–G21), lactating (PD: postdelivery; PD0–PD10) and weaned (pups removed 2 days after lactation; PD3–PD10) rats (n ¼ 1 for each sample) was analysed for BTG2 expression by Northern blot. (b) Total RNA (7.5 mg) isolated from mammary glands of rats during indicated periods of development was analysed for BTG2 expression (n ¼ 3). To measure changes in BTG2 expression, band intensities were quantified using phosphorImager and iQMac data analysis software. (c) Immunohistochemical detection of BTG2 protein in the rat mammary gland. Paraffin-embedded mammary tissue sections from various stages of postnatal morphogenesis shown above were stained with an anti-BTG2 antibody (original magnification  125). The slides were counterstained with hematoxylin

gland. While 100% of the nuclei of epithelial cells lining could inhibit breast cancer cell growth using colony the ducts of virgin mammary glands were positive for inhibition assays. A human BTG2 cDNA expressing the BTG2 staining, the percentage of epithelial cell nuclei protein (Figure 3a) was cotransfected into MCF7 cells expressing the BTG2 protein gradually decreased during along with a plasmid encoding hygromycin resistance. pregnancy with almost no expression during late The PCDNA vector (V) construct was transfected into pregnancy (G20) or in the lactating mammary gland. cells as control. Cells were grown in the presence of Expression recovered in the mammary glands of weaned hygromycin for 3weeks and drug-resistant colonies were animals. counted. Expression of BTG2 in cells resulted in about 70% reduction in drug-resistant colonies compared to vector-transfected cells (n 3; Figure 3b). Cyclin D1 expression overcomes BTG2-mediated ¼ To test whether inhibition of breast cancer cell growth inhibition of breast cancer cell growth by BTG2 was associated with cell cycle arrest, MCF7 Since BTG2 expression is associated with antiprolifera- cells were transfected with constructs encoding BTG2 tive activity, we tested whether expression of BTG2 along with a plasmid encoding the cell surface marker

Oncogene BTG2 expression in breast cancer H Kawakubo et al 8313 CD20 as described by van den Heuvel and Harlow MB-468, MDA-MB-435 and MDA-MB-231 breast (1993). Analysis of CD-20-positive transfectants by cancer cells (Figure 4A). fluorescence-activated cell sorter (FACS) analysis We then analysed the expression of BTG2 in human for DNA content demonstrated a 10–12% increase breast cancer by immunohistochemical analysis of in the fraction of cells in the G1 phase of the cell breast tissue resected from 23breast cancer patients. cycle compared to cells transfected with the vector These included 14 poorly differentiated (histologic grade (Figure 3c). 3), five moderately differentiated (histologic grade 2) Since the antiproliferative activity of BTG2 has been and four well-differentiated (histologic grade 1) ductal reported to result from its ability to inhibit cyclin D1 adenocarcinomas. The specificity of the antibody to expression (Guardavaccaro et al., 2000), we tested detect human BTG2 protein by immunohistochemistry whether coexpression of cyclin D1 can overcome was confirmed by preincubating the antibody with COS BTG2-mediated inhibition of growth. As seen pre- cells expressing a transiently transfected human BTG2 viously, BTG2-transfected cultures had a lower number expression construct. Normal human breast tissue was of colonies compared to vector-transfected controls, analysed by immunohistochemistry following incuba- while expression of cyclin D1 alone resulted in a tion with the anti-rabbit BTG2 antibody, anti-BTG2 statistically significant increase in colony numbers. antibody preincubated with vector-transfected COS Coexpression of cyclin D1 and BTG2 counteracted the protein lysate or anti-BTG2 antibody preincubated with effect of the other (Figure 3d). BTG2-transfected COS protein lysates. As seen in In the mammary glands of rats, unlike BTG2, which Figure 4B, the anti-BTG2 antibody and the anti-BTG2 gradually declines during pregnancy, cyclin D1 mRNA antibody preincubated with vector-transfected COS cell sharply increased at early gestation compared with that protein detected the BTG2 protein in the epithelial and in virgin animals. Through late gestation, cyclin D1 myoepithelial cells of normal ducts and lobules. How- mRNA was maintained at levels higher than that in ever, preincubation of the antibody with BTG2-expres- virgin rats but declined at lactation. During the early sing COS cell protein extracts abolished staining. days of involution, coincident with a strong induction of Staining was also absent when the tissue was incubated BTG2, cyclin D1 mRNA decreased further and was with affinity-purified rabbit immunoglobulins (data not undetectable (Figure 3e, upper panel). Immunohisto- shown). chemical analyses of cyclin D1 protein in the mammary In 15 of 23breast cancer samples analysed, nuclear glands of rats were in agreement with the above results expression of BTG2 was absent in the tumor but very (Figure 3e, middle panel). Since the expression of BTG2 strong BTG2 staining was detectable in the nuclei of the and cyclin D1 in the mammary glands of rats demon- epithelium of ducts and lobules in the surrounding strated a striking negative correlation during early normal tissue. These consisted of 11 tumors of histologic periods of gestation, we analysed their expression in a grade 3, three turmors of histologic grade 2 and one larger cohort of timed-pregnant female rats (n ¼ 3). The tumor of histologic grade 1. In the remaining eight inverse correlation between BTG2 and cyclin D1 cases, BTG2 expression was observed in the nuclei of expression during early gestation (Figure 3e, lower both the uninvolved ducts and in the tumor, with no panel) was consistent with the finding that BTG2 can discernible difference in levels (Figure 4C). suppress cyclin D1 (Guardavaccaro et al., 2000). Moreover, estrogen and progesterone, important BTG2 expression is suppressed during mammary gland factors in stimulating breast growth during pregnancy, tumorigenesis in mice suppressed BTG2 and induced cyclin D1. In MCF7 cells, estrogen-mediated suppression of BTG2 correlated In order to determine changes in BTG2 expression with induction of cyclin D1 expression (Figure 3f, left during tumor progression, we studied BTG2 expression panel). Similarly, suppression of BTG2 expression in in mammary tumors, which arise spontaneously in T47D breast cancer cells by the progestin R5020 was transgenic mice, that express the early region of the also associated with an increase in cyclin D1 (Figure 3f, SV40 large tumor Ag under the regulation of the rat right panel). TNF-a, an activator of the NF-kB path- prostatic steroid binding protein promoter (Green et al., way, also suppressed cyclin D1 expression in T47D cells 2000). In this mouse mammary carcinoma model, (Figure 3g). almost all transgenic female mice develop atypical ductal hyperplasia by 8 weeks, nodular atypical BTG2 expression is aberrantly regulated in breast tumors hyperplasia by 12 weeks and invasive carcinomas by 16–20 weeks (Green et al., 2000). This provided us with Since BTG2 expression in breast cancer cells inhibited a model to study the expression of BTG2 during growth, we wished to determine whether BTG2 expres- mammary tumor progression. sion would be deregulated in breast cancer. Analysis of BTG2 protein expression during mammary gland BTG2 expression in human breast cancer cell lines tumorigenesis in this mouse model was analysed by demonstrated that compared to levels observed in immunohistochemistry. In nontransgenic mice, BTG2 184A1 and MCF10A, two immortalized normal human protein localized predominantly to the nucleus of mammary epithelial cell lines, expression was much epithelial cells in normal ducts and lobules of the lower in breast cancer cell lines and was almost mammary gland (Figure 5a). In the transgenic mice, undetectable in the estrogen receptor-negative MDA- epithelial cells expressing BTG2 were interspersed with

Oncogene BTG2 expression in breast cancer H Kawakubo et al 8314

Oncogene BTG2 expression in breast cancer H Kawakubo et al 8315 nonexpressing cells in ducts exhibiting atypical hyper- putative DNA binding sites for many transcription plasia (Figure 5b and d). Thus loss of BTG2 expression factors including AP1, GATA1 and SP1 (Duriez et al., may represent an early event in neoplastic transforma- 2002). The kinetics of BTG2 induction is stimulus tion of the mammary gland. Furthermore, BTG2 was dependent. EGF and NGF transiently induce BTG2 in consistently absent in small nodules of tumors observed PC12 cells by 1 h (Bradbury et al., 1991). However, in in 12-week-old animals but present in the epithelial cell MCF7 cells maximal induction of BTG2 by g-irradia- nuclei of nontumor tissue (Figure 5c and d). BTG2 tion, adriamycin and UV occurred around 3–6 h and expression was also absent in advanced mammary persisted for longer periods of time (Cortes et al., 2000). adenocarcinomas of older animals but was expressed MIS and TNF-a induce BTG2 mRNA in T47D cells in the adjacent nontumor tissue (Figure 5e–h). after 2 h of treatment and the levels are maintained above basal for up to 24 h. The dynamic expression pattern of BTG2 during Discussion postnatal morphogenesis of the mammary glands suggests that it may play an important role in regulating The decline in BTG2 expression in the mammary gland growth and differentiation processes in this tissue. The during gestation, deregulation of its expression in mammary gland is a target tissue for estrogens, mammary tumors and inhibition of breast cancer cell depending on it for hormonal stimulation of epithelial growth by ectopic expression of BTG2 suggest that cell proliferation. During pregnancy, circulating estro- BTG2 may have a tumor suppressor function in gen is high and, by means of a complex hormonal mammary epithelial cells. BTG2 was originally de- interplay with progesterone, prolactin and growth scribed as a p53-inducible gene and contains six hormone, stimulates mammary epithelial cells to pro- consensus p53DNA binding sites in the 5 0 flanking liferate and undergo lobuloalveolar development. The region and the first intron. The induction of BTG2 in ability of estrogen and progestin to suppress BTG2 T47D cells, which contain a transactivation-deficient expression suggests that these steroids may be involved p53allele mutated at codon 194 (Nigro et al., 1989; in downregulating BTG2 mRNA in the mammary gland Wosikowski et al., 1995), suggests that BTG2 expression during pregnancy. may be regulated through p53-independent mechanisms BTG2 and cyclin D1 expressions demonstrate as well. The presence of an NF-kB site in the promoter an inverse correlation in the rat mammary gland during region of BTG2 (Duriez et al., 2002) renders it early gestation, and estrogen and progestin treatment of responsive to signals such as MIS and TNF-a, which breast cancer cells suppressed BTG2 expression with an robustly induce NF-kB. The partial abrogation of associated increase in cyclin D1. These observations BTG2 induction by MIS in T47D cells expressing are consistent with the report that BTG2 down- dominant-negative IkB-a suggests that other transcrip- regulates cyclin D1 expression (Guardavaccaro et al., tion factors besides NF-kB may also be involved in 2000). MIS-mediated induction of BTG2. In fact characteriza- Cyclin D1 has been identified as a target gene induced tion of the BTG2 promoter revealed the presence of following activation of the NF-kB pathway (Horwitz

Figure 3 BTG2 inhibits breast cancer cell growth by preventing cell cycle progression. (a) COS7 (left panel) and MCF7 (right panel) cells were transfected with 3 mg of PCDNA vector, PCDNA-BTG2 or PCDNA-antisense BTG2 constructs and total protein was analysed by Western blot using a rabbit anti-BTG2 antibody. Position of the BTG2 protein is shown. NS represents a nonspecific band in MCF7 immunoblots. (b) Equal numbers of MCF7 cells were stably transfected with 0.5 mg of hygromycin resistance plasmid and either 5 mg of PCDNA vector or 5 mg of BTG2 cDNA. Colonies, which grew in medium containing 100 mg/ml hygromycin for 3weeks, were stained with crystal violet. Colonies in plates transfected with vector were set at 100%. (c) BTG2 inhibits cell cycle progression of breast cancer cells. MCF7 cells were transfected with either 4 mg of vector or a BTG2 expression construct and 1 mg of a plasmid encoding the cell surface marker CD20. After 24 h, cells were harvested and cell cycle distribution of CD20-positive cells was analysed by FACS. Percentage of cells in the G1 phase of the cell cycle is shown (n ¼ 3). (d) Antagonistic growth effects of cyclin D1 and BTG2 on breast cancer cells. Equal numbers of MCF7 cells were stably transfected with 0.5 mg of hygromycin resistance plasmid and (1) 5 mg of vector, (2) 5 mg of BTG2, (3) 5 mg of BTG2 þ 5 mg of cyclin D1, or (4) 5 mg of cyclin D1. Total amount of DNA transfected was equalized with PCDNA vector. Colonies, which grew in medium containing 100 mg/ml hygromycin for 3weeks, were stained with crystal violet. Colonies in plates transfected with vector were set at 100%. (e) Cyclin D1 and BTG2 expressions demonstrate an inverse correlation in the mammary glands of rats. Upper panel: The blot containing total RNA isolated from the mammary glands of rats described in Figure 2a and probed for BTG2 was reanalysed for cyclin D1 expression. Hybridization to 18S is shown in Figure 2a. Middle panel: Immunohistochemical detection of cyclin D1 protein in the rat mammary gland. Paraffin-embedded mammary tissue sections from various stages of postnatal morphogenesis shown above were stained with an anti-cyclin D1 antibody (original magnification  125). The slides were counterstained with hematoxylin. Lower panel: Inverse correlation between cyclin D1 and BTG2 expression in the mammary glands during early gestation. Total RNA isolated from the mammary glands of rats (n ¼ 3) was analysed for BTG2 and cyclin D1 expression by Northern blot. Positions of cyclin D1 and BTG2 are shown. Hybridization to 18S is shown to control for loading. (f) Suppression of BTG2 expression in breast cancer cells by estrogen and progestin treatment is associated with stimulation of cyclin D1. Left panel: MCF7 cells grown in estrogen-deprived medium for 4 days were treated with 1 nM estrogen for 2 and 4 h. Total RNA (7.5 mg) was analysed for BTG2 and cyclin D1 expression. Right panel: T47D cells grown in charcoal-stripped serum for 4 days were treated with 10 nM of the progestin R5020 for 2 and 4 h. Total RNA was analysed for BTG2 and cyclin D1 expression. Quantification of transcript levels by phosphorimaging is shown for both blots. Expression in the untreated sample was set at 1. Hybridization to 18S is shown to control for loading. (g) T47D cells were treated with 1 ng/ml TNF-a for 2, 6 and 24 h and total RNA was analysed for BTG2 and cyclin D1 expression

Oncogene BTG2 expression in breast cancer H Kawakubo et al 8316 et al., 1997; Guttridge et al., 1999; Hinz et al., 1999; 2002; Rocha et al., 2003). In HeLa cells, suppression of Whitehead et al., 1999). However, our results demon- NF-kB activity with the dominant-negative IkB-a strate that TNF-a, a potent activator of the NF-kB impaired proliferation but did not result in any change pathway, induced BTG2 and suppressed cyclin D1 in in cyclin D1 levels (Kaltschmidt et al., 1999). Induction T47D cells. These results are consistent with the of p53suppresses cyclin D1 by regulating Bcl-3levels observation that NF-kB activation by TNF-a was not and enhancing the association of p52 with HDAC1. sufficient to induce cyclin D1 (Hinz et al., 1999). Since BTG2 is target of p53, it is interesting to speculate Modulation of cyclin D1 by NF-kB demonstrates a that suppression of cyclin D1 by BTG2 may involve a complex regulation that is dependent on the context of similar mechanism (Rocha et al., 2003). cell and stimulus, and Rel/NF-kB dimers with different Cyclin D1 is overexpressed in 50% of mammary activities (Kaltschmidt et al., 1999; Nakamura et al., carcinomas (Gillett et al., 1996), and overexpression of cyclin Dl in mammary glands results in abnormal mammary cell proliferation including the development of mammary adenocarcinomas (Wang et al., 1994) suggesting that increased levels of cyclin D1 may contribute to neoplastic transformation of the breast. Coexpression of cyclin D1 has been shown to reverse BTG2-mediated inhibition of cell cycle progression (Guardavaccaro et al., 2000). Colony inhibition assays to determine the effects of cyclin D1 and BTG2 on breast cancer cell growth revealed that cyclin D1 expression in MCF7 cells significantly enhanced growth while ectopic expression of BTG2 suppressed growth. A combination of BTG2 and cyclin D1 counteracted the growth regulatory effects of each other. Thus BTG2 may be one of the factors involved in regulating the expression and function of cyclin D1 in the breast. Several lines of evidence suggest an important role for BTG2 in the suppression of carcinogenesis. Expression analysis of BTG2 during prostate carcinogenesis de- monstrated that reduction or loss of BTG2 expression was coincident with malignant progression. In high- grade prostatic intraepithelial neoplasia, which are considered to be probable precursor lesions for a subset of prostate cancers, BTG2 expression was either absent or limited to weak cytoplasmic expression suggesting that loss of BTG2 expression may be an early event in prostate cancer progression. BTG2 expression was also undetectable in higher grade tumors and in bone metastases (Ficazzola et al., 2001). Furthermore, thymic

Figure 4 BTG2 expression is deregulated in breast cancer. (A) Expression of BTG2 in human breast cancer cell lines. Total RNA from (7.5 mg) MCF10A, 184A1, A549, T47D, MCF7, MDA-MB- 231, MDA-MB-435 and MDA-MB-468 was analysed by Northern for expression of BTG2. (B) Specificity of the anti-BTG2 antibody. Equal amounts of total protein lysates from COS cells transfected with either PCDNA vector or BTG2 expression construct were incubated with anti-BTG2 antibody for 1 h at room temperature. Normal human breast tissue was analysed by immunohistochem- istry with the preadsorbed anti-BTG2 antibody. Immunohisto- chemical analysis with anti-BTG2 antibody that was not preincubated with COS cell proteins is shown as positive control (C) (a–d) Nuclear BTG2 expression is absent in breast tumors. Nuclear BTG2 expression is absent in tumor (b) but observed in adjacent uninvolved glands (c). Hematoxylin–eosin stains of tumor (a) and uninvolved glands (d) are shown. Original magnifications  200. (e–h) Nuclear BTG2 expression is present in tumor and in uninvolved glands. Nuclear BTG2 expression is observed in tumor (f) and in adjacent, uninvolved glands (g). Hematoxylin–eosin stains of tumor (e) and uninvolved glands (h) are shown. Original magnifications  200

Oncogene BTG2 expression in breast cancer H Kawakubo et al 8317 carcinoma tissues derived from SV40T-Ag harboring compared to normal mouse mammary glands (Desai transgenic mice did not express BTG2 (Lim et al., 1995). et al., 2002). Thus deregulation of BTG2 expression may Recently, Struckmann et al. (2004) demonstrated that be a crucial point in the pathogenesis of some types of BTG2 expression was suppressed in clear cell renal cancer. carcinoma suggesting that downregulation of BTG2 The dynamic pattern of BTG2 expression in the may be an important step in renal cancer development. mammary gland during pregnancy and its loss early in Our results demonstrate that nuclear BTG2 expression the tumorigenic process suggests that BTG2 may play was absent in 65% of human breast tumors compared an important role in transformation and/or differentia- with surrounding matched normal tissue. Moreover, tion of the mammary gland. The interaction of BTG2 analysis of BTG2 expression in spontaneous mammary with the cell cycle machinery and its physical association tumors that arise in transgenic mice expressing the early with several growth regulatory proteins may present it region of the SV40 large tumor Ag (Green et al., 2000) to be an ideal candidate for inactivation during the suggests suppression on BTG2 may be an early event tumorigenic process. Animal models in which BTG2 during neoplastic progression of the mammary gland. expression is targeted to the mammary gland may Interestingly, mammary tumors arising in these mice are provide further insight into the role of BTG2 in the reported to have higher levels of cyclin D1 and cyclin E tumorigenic process in the breast.

Figure 5 BTG2 expression during mammary tumor progression in mice. Paraffin-embedded mouse mammary tissue derived from 6- week-old nontransgenic animals (a) and from transgenic mice of 6, 12, 18 and 24 weeks old (b–h) was analysed for BTG2 protein expression by immunohistochemistry. The slides were counterstained with hematoxylin. BTG2 stainings of nontumor and tumor tissue shown in the figure were derived from the same tissue section on the slide. BTG2 is expressed in the nuclei of epithelial cells lining the ducts and lobules of nontumor tissue in the surrounding area but expression is lost in regions showing atypical hyperplasia and in the tumors

Oncogene BTG2 expression in breast cancer H Kawakubo et al 8318 Materials and methods 6 mg of either PCDNA vector or PCDNA-BTG2 was transiently transfected into COS cells using FUGENE-6 and Cell culture proteins were extracted with RIPA buffer 48 h after transfec- tion. The rabbit anti-BTG2 antibody was preincubated with Human breast cancer cell lines T47D, MCF7, MDA-MB-231, either the vector- or BTG2-expressing COS protein lysate for MDA-MB-435 and MDA-MB-468 and A549 were grown in 1 h at room temperature. Normal human breast tissue sections Dulbecco’s modified medium supplemented with 10% female were analysed by immunohistochemistry (see below) using the fetal bovine serum, glutamine and penicillin/streptomycin. The anti-BTG2 antibody, BTG2 antibody preincubated with nontumorigenic MCF10A cells (ATCC) were grown in vector-transfected COS cell proteins or BTG2 antibody mammary epithelial growth medium (MEGM, Clonetics) preincubated with BTG2-transfected COS cell proteins. supplemented with 100 ng/ml cholera toxin (Calbiochem) and Immunohistochemistry was performed on formalin-fixed, the 184A1 cells (a kind gift from Dr Martha Stampfer) were paraffin-embedded 5 mm tissue sections according to standard grown in MEGM supplemented with transferrin and isopro- procedures. Deparaffinized tissue sections were treated with terenol. Human recombinant MIS (rhMIS) was collected from 3% hydrogen peroxide for 10 min to inhibit endogenous growth media of Chinese hamster ovary cells transfected with peroxidase. After the sections had been microwaved in a the human MIS gene and purified as described (Ragin et al., 10 mM citrate buffer solution at 100 W for 25 min, they were 1992). T47D cells stably expressing IkB-a-DN were generated incubated with 10% normal goat serum for 30 min to block as described by Hoshiya et al. (2003). TNF-a was purchased any nonspecific protein binding. For BTG2 detection, sections from Promega. were incubated with rabbit antibodies against human BTG2 at room temperature overnight at a final antibody concentration Western blot analysis of 1 mg/ml diluted in blocking serum solution. After incubation The characterization of the rabbit anti-BTG2 antibody has with biotinylated link antibody and peroxidase-labeled strep- been described (Ficazzola et al., 2001). Expression of BTG2 tavidin, the staining was developed by reaction with diamino- protein in COS and MCF7 cells was analysed by Western blot benzidine (DAB) or Nova Red for 5 min. Detection using a rabbit anti-BTG2 antibody according to the protocol chemistries were all purchased from Vector Laboratories described (Ha et al., 2000). (Burlingame, CA, USA), and used according to the manu- facturer’s instructions. The sections were lightly counterstained with hematoxylin and then mounted with a coverslip. Sections Animals were stained with similar concentrations of affinity-purified All animals were cared for and experiments were performed in rabbit immunoglobulins as controls. Immunohistochemical this facility under AAALAS approved guidelines using stains were evaluated semiquantitatively and ranged from protocols approved by the Institutional Review Board- negative to positive nuclear expression. Institutional Animal Care and Use Committee of the The mouse monoclonal anti-human cyclin D1 antibody Massachusetts General Hospital. BTG2 expression analysis (NCL-CYCLIN D1-GM) was purchased from Novocastra in the rat mammary gland during postnatal morphogenesis Laboratories. The protocol used for immunohistochemical was carried out using Sprague–Dawley rats. Mammary glands detection of cyclin D1 was essentially similar to that described were harvested bilaterally from each animal for RNA isolation above, except for the following steps. The sections were treated and immunohistochemical analysis. with 0.3% hydrogen peroxide to inhibit endogenous peroxide The transgenic mouse model for spontaneous mammary and incubated with 40 mg/ml of primary antibody for 1 h carcinoma resulting from tissue-specific targeted expression of at 371C. the early region of the SV40 large tumor Ag using the promoter of the rat prostatic steroid binding protein has been Growth inhibition assays described (Green et al., 2000). After establishing a mouse colony, mammary glands were resected from 6-, 12-, 18- and To measure the ability of BTG2 to inhibit cellular prolifera- 24-week-old animals. Tissue was fixed in 4% paraformalde- tion, MCF7 cells were transfected with empty PCDNA vector, hyde and embedded in paraffin for immunostaining. BTG2 and cyclin D1 expression constructs along with a hygromycin resistance plasmid. Cells were grown in the hygromycin-containing medium for 2–3weeks, stained with RNA analysis crystal violet and drug-resistant colonies were counted. The EST clones to detect the expression of human and mouse BTG2 and the rat cyclin D1 were purchased from Incyte Cell cycle analysis Genomics Inc. The human cyclin D1 expression construct was To analyse the effect of BTG2 on cell cycle progression, MCF7 a kind gift from Dr Phil Hinds. For Northern blot analysis, cells in logarithmic growth phase were transiently transfected equal amounts of RNA were separated on a formaldehyde gel, with either vector or BTG2 expression construct along with a transferred to Hybond-N membrane (Amersham) and probed plasmid encoding the cell surface marker CD20 as described with either human/mouse BTG2 or human/rat cyclin D1. (van den Heuvel and Harlow, 1993). Cells were fixed for 24 h after transfection and stained for CD20 expression and for Immunohistochemical analysis DNA content using propidium iodide. The cell cycle distribu- A total of 23cases of human breast carcinoma with adjacent tion of CD20-positive cells was determined by FACS analysis. uninvolved breast parenchyma were selected from the files of the Massachusetts General Hospital (MGH) Pathology Acknowledgements Department according to protocols approved by the Human We thank Drs Emmett Schmidt, Toshi Shioda, Paul Harkin, Research Committee at MGH. Leif Ellisen and Jose Teixeira for critically reading this To establish the specificity of the BTG2 antibody, the manuscript. We thank Drs Donahoe and MacLaughlin for antibody was preincubated with COS cell protein lysates providing the MIS used in these experiments. We thank derived from either vector- or BTG2-transfected cells. Briefly, Dr Clayton Naeve, Director of the Hartwell Center for

Oncogene BTG2 expression in breast cancer H Kawakubo et al 8319 Bioinformatics and Biotechnology at St Jude Children’s PDW), and by the Breast Cancer Research Grant from the Research Hospital for DNA microarray analysis. This work Massachusetts Department of Public Health, the Avon Breast was supported by the Department of Defense Breast Cancer Cancer Pilot Project Grant, the Claflin Distinguished Scholar Research Grant DAMD17-03-1-0407 (to VG), NIH CA84441 Award, partial support from the Dana-Farber Harvard Breast (to PDW) and Department of Defense grant PC030351 (to Cancer SPORE and from NIH/NCI grant CA89138 (to S.M.).

References

Bradbury A, Possenti R, Shooter EM and Tirone F. (1991). Lim IK, Park TJ, Kim S, Lee HW and Paik WK. (1998b). Proc. Natl. Acad. Sci. USA, 88, 3353–3357. Biochem. Mol. Biol. Int., 45, 871–878. Chen LC, Dollbaum C and Smith HS. (1989). Proc. Natl. Lin WJ, Gary JD, Yang MC, Clarke S and Herschman HR. Acad. Sci. USA, 86, 7204–7207. (1996). J. Biol. Chem., 271, 15034–15044. Cmarik JL, Herschman H and Colburn NH. (1994). Mol. Matsuda S, Rouault J, Magaud J and Berthet C. (2001). FEBS Carcinogen., 11, 115–124. Lett., 497, 67–72. Cortes U, Moyret-Lalle C, Falette N, Duriez C, Ghissassi FE, Melamed J, Kernizan S and Walden PD. (2002). Tissue Cell, Barnas C, Morel AP, Hainaut P, Magaud JP and Puisieux 34, 28–32. A. (2000). Mol. Carcinogen., 27, 57–64. Montagnoli A, Guardavaccaro D, Starace G and Tirone F. Desai KV, Xiao N, Wang W, Gangi L, Greene J, Powell JI, (1996). Cell Growth Differ., 7, 1327–1336. Dickson R, Furth P, Hunter K, Kucherlapati R, Simon R, Nakamura Y, Grumont RJ and Gerondakis S. (2002). Mol. Liu ET and Green JE. (2002). Proc. Natl. Acad. Sci. USA, Cell. Biol., 22, 5563–5574. 99, 6967–6972. Nandi S. (1958). J. Natl. Cancer, 21, 1039–1063. Duriez C, Falette N, Audoynaud C, Moyret-Lalle C, Bensaad Nigro JM, Baker SJ, Preisinger AC, Jessup JM, Hostetter R, K, Courtois S, Wang Q, Soussi T and Puisieux A. (2002). Cleary K, Bigner SH, Davidson N, Baylin S, Devilee P, Gene, 282, 207–214. Glover T, Collins FS, Weston A, Modali R, Harris CC and Ficazzola MA, Fraiman M, Gitlin J, Woo K, Melamed J, Vogelstein B. (1989). Nature, 342, 705–708. Rubin MA and Walden PD. (2001). Carcinogenesis, 22, Prevot D, Voeltzel T, Birot AM, Morel AP, Rostan MC, 1271–1279. Magaud JP and Corbo L. (2000). J. Biol. Chem., 275, Fletcher BS, Lim RW, Varnum BC, Kujubu DA, Koski RA 147–153. and Herschman HR. (1991). J. Biol. Chem., 266, Ragin RC, Donahoe PK, Kenneally MK, Ahmad MF and 14511–14518. MacLaughlin DT. (1992). Protein Expr. Purif., 3, 236–245. Gillett C, Smith P, Gregory W, Richards M, Millis R, Peters G Rocha S, Martin AM, Meek DW and Perkins ND. (2003). and Barnes D. (1996). Int. J. Cancer, 69, 92–99. Mol. Cell. Biol., 23, 4713–4727. Green JE, Shibata MA, Yoshidome K, Liu ML, Jorcyk C, Rouault JP, Falette N, Guehenneux F, Guillot C, Rimokh R, Anver MR, Wigginton J, Wiltrout R, Shibata E, Kaczmarc- Wang Q, Berthet C, Moyret-Lalle C, Savatier P, Pain B, zyk S, Wang W, Liu ZY, Calvo A and Couldrey C. (2000). Shaw P, Berger R, Samarut J, Magaud JP, Ozturk M, Oncogene, 19, 1020–1027. Samarut C and Puisieux A. (1996). Nat. Genet., 14, 482–486. Guardavaccaro D, Corrente G, Covone F, Micheli L, Rouault JP, Prevot D, Berthet C, Birot AM, Billaud M, D’Agnano I, Starace G, Caruso M and Tirone F. (2000). Magaud JP and Corbo L. (1998). J. Biol. Chem., 273, Mol. Cell. Biol., 20, 1797–1815. 22563–22569. Guttridge DC, Albanese C, Reuther JY, Pestell RG and Sakaguchi T, Kuroiwa A and Takeda H. (2001). Mech. Dev., Baldwin Jr AS. (1999). Mol. Cell. Biol., 19, 5785–5799. 104, 113–115. Ha TU, Segev DL, Barbie D, Masiakos PT, Tran TT, Segev DL, Ha TU, Tran TT, Kenneally M, Harkin P, Jung M, Dombkowski D, Glander M, Clarke TR, Lorenzo HK, MacLaughlin DT, Donahoe PK and Maheswaran S. (2000). Donahoe PK and Maheswaran S. (2000). J. Biol. Chem., J. Biol. Chem., 275, 28371–28379. 275, 37101–37109. Segev DL, Hoshiya Y, Hoshiya M, Tran TT, Carey JL, Hinz M, Krappmann D, Eichten A, Heder A, Scheidereit C Stephen AE, MacLaughlin DT, Donahoe PK and Mahes- and Strauss M. (1999). Mol. Cell. Biol., 19, 2690–2698. waran S. (2002). Proc. Natl. Acad. Sci. USA, 99, 239–244. Horwitz BH, Scott ML, Cherry SR, Bronson RT and Segev DL, Hoshiya Y, Stephen AE, Hoshiya M, Tran TT, Baltimore D. (1997). Immunity, 6, 765–772. MacLaughlin DT, Donahoe PK and Maheswaran S. (2001). Hoshiya Y, Gupta V, Segev DL, Hoshiya M, Carey JL, Sasur J. Biol. Chem., 276, 26799–26806. LM, Tran TT, Ha TU and Maheswaran S. (2003). Mol. Cell. Struckmann K, Schraml P, Simon R, Elmenhorst K, Mirlacher Endocrinol., 211, 43–49. M, Kononen J and Moch H. (2004). Cancer Res., 64, Iacopetti P, Michelini M, Stuckmann I, Oback B, 1632–1638. Aaku-Saraste E and Huttner WB. (1999). Proc. Natl. Acad. Tirone F. (2001). J. Cell. Physiol., 187, 155–165. Sci. USA, 96, 4639–4644. van den Heuvel S and Harlow E. (1993). Science, 262, Kaltschmidt B, Kaltschmidt C, Hehner SP, Droge W and 2050–2054. Schmitz ML. (1999). Oncogene, 18, 3213–3225. Wang TC, Cardiff RD, Zukerberg L, Lees E, Arnold A and Kuo ML, Duncavage EJ, Mathew R, den Besten W, Pei D, Schmidt EV. (1994). Nature, 369, 669–671. Naeve D, Yamamoto T, Cheng C, Sherr CJ and Roussel Whitehead IP, Lambert QT, Glaven JA, Abe K, Rossman KL, MF. (2003). Cancer Res., 63, 1046–1053. Mahon GM, Trzaskos JM, Kay R, Campbell SL and Der Lim IK, Lee MS, Lee SH, Kim NK, Jou I, Seo JS and Park CJ. (1999). Mol. Cell. Biol., 19, 7759–7770. SC. (1995). J. Cancer Res. Clin. Oncol., 121, 279–284. Wosikowski K, Regis JT, Robey RW, Alvarez M, Buters JT, Lim IK, Lee MS, Ryu MS, Park TJ, Fujiki H, Eguchi H and Gudas JM and Bates SE. (1995). Cell Growth Differ., 6, Paik WK. (1998a). Mol. Carcinogen., 23, 25–35. 1395–1403.

Oncogene