CCAAT/Enhancer-binding Protein δ Regulates Mammary Epithelial Cell G0 Growth Arrest and Apoptosis*

CCAAT/enhancer-binding proteins (C/EBPs) are a highly conserved family of DNA-binding proteins that regulate cell-specific growth, differentiation, and apoptosis. Here, we show that induction of C/EBPδ gene expression during G0 growth arrest is a general property of mammary-derived cell lines. C/EBPδ is not induced during G0 growth arrest in 3T3 or IEC18 cells. C/EBPδ induction is G0-specific in mouse mammary epithelial cells; C/EBPδ gene expression is not induced by growth arrest in the G1, S, or G2 phase of the cell cycle. C/EBPδ antisense-expressing cells (AS1 cells) maintain elevated cyclin D1 and phosphorylated retinoblastoma protein levels and exhibit delayed G0 growth arrest and apoptosis in response to serum and growth factor withdrawal. Conversely, C/EBPδ-overexpressing cells exhibited a rapid decline in cyclin D1 and phosphorylated retinoblastoma protein levels, a rapid increase in the cyclin-dependent kinase inhibitor p27, and accelerated G0 growth arrest and apoptosis in response to serum and growth factor withdrawal. When C/EBPδ levels were rescued in AS1 cells by transfection with a C/EBPδ “sense” construct, normal G0 growth arrest and apoptosis were restored. These results demonstrate that C/EBPδ plays a key role in the regulation of G0 growth arrest and apoptosis in mammary epithelial cells.

eration and differentiation of white blood cells, ovarian granulosa cells, and mammary epithelial cells (6,(17)(18)(19). Like C/EBP␣ and C/EBP␤, C/EBP␦ is also associated with adipocyte differentiation (2). In addition to adipocyte differentiation, C/EBP␦ also functions in the differentiation of lung epithelial cells (20) and myelomonocytic cells (21). C/EBP␦ also plays an important role in the hepatic acute phase response (22,23). C/EBP⑀, the newest member of the C/EBP family of transcription factors, regulates differentiation of granulocytes (6).
Recent reports indicate that C/EBPs play prominent roles in mammary gland development, differentiation, and programmed cell death (18, 19, 24 -26). Two recent reports investigated the role of C/EBP␤ in mammary gland biology using C/EBP␤ knockout mice (18,19). In both reports, mammary epithelial cell proliferation and differentiation were dramatically reduced in female C/EBP␤ knockout mice. C/EBP␦ also functions in mammary gland biology; however, instead of promoting mammary gland proliferation and differentiation, C/EBP␦ is associated with mammary epithelial cell G 0 growth arrest and programmed cell death (24 -26). Most reports indicate that C/EBP␣ plays a relatively minor role in mammary epithelial cell biology (19,24,25).
The mammary gland is a unique organ system in that it attains full functional capacity late in life, at sexual maturation (27). Mammary epithelial cells in the adult female cycle through intervals of quiescence, proliferation, differentiation, and programmed cell death in response to hormonal changes of the normal estrous cycle (27). These alterations in mammary epithelial cell fate are well described at the morphological level but poorly understood at the molecular level.
The overall goal of this study was to investigate the role of C/EBP␦ in mammary epithelial cell G 0 growth arrest. Even though most cells in the adult animal have exited the cell cycle and exist in G 0, few G 0 regulatory genes have been described (27,28). A better understanding of genes that regulate cell cycle exit/G 0 entry is important in understanding normal cell biology, tissue homeostasis, and cancer. Mutational inactivation of one G 0 regulatory gene, the von Hippel-Lindau (VHL) tumor suppressor gene, has been implicated in 80% of human sporadic renal cell carcinomas (29).
We previously showed that C/EBP␦ is induced in G 0 growtharrested COMMA D mouse mammary epithelial cells (24). C/EBP␦ induction occurs early in cell cycle exit/G 0 growth arrest and remains elevated throughout the time the cells remain in G 0. The present results extend this observation, demonstrating G 0 induction of C/EBP␦ in multiple mammaryderived cell lines. In addition, the induction of C/EBP␦ in mammary epithelial cells is G 0 -specific. Altering mammary epithelial cell C/EBP␦ content by transfection with C/EBP␦ antisense or over expression constructs dramatically affected both G 0 growth arrest and apoptosis in response to serum and growth factor withdrawal. These results demonstrate a key role for C/EBP␦ in the regulation of major cell fate determining pathways in mammary epithelial cells.

EXPERIMENTAL PROCEDURES
Cell Culture-The nontransformed HC 11 mouse mammary epithelial cell line was cultured in complete growth medium (CGM) consisting of RPMI 1640 medium (4.5 g/liter glucose) supplemented with 10% fetal bovine serum (FBS), 10 ng/ml epidermal growth factor, and 10 g/ml insulin. COMMA D cells were maintained as described previously (24). The NMuMG mouse mammary epithelial cell line (ATCC CRL 1636) was cultured in Dulbecco's modified Eagle's medium (4.5 g/ml glucose) supplemented with 10% FBS and 10 g/ml insulin. The mouse mammary tumor cell lines Mm5MT (ATCC CRL 1637) and MMT 060562 (ATTC CCL 51) were cultured in Dulbecco's modified Eagle's medium (4.5 g/ml glucose) supplemented with 10% FBS. NIH 3T3 cells (ATCC CRL 1658) were cultured in Dulbecco's modified Eagle's medium (4.5 g/ml glucose) supplemented with 10% calf serum. The rat intestinal epithelial cell line IEC 18 (ATTC CRL 1589) was cultured in Dulbecco's modified Eagle's medium supplemented with 5% FBS and 5 g/ml insulin. All media contained 5 units/ml penicillin and 5 g/ml streptomycin. All medium components were purchased from Life Technologies, Inc.
Generation of Cell Lines-The C/EBP␦ RNA antisense plasmid was produced by digesting MSV/EBP␦ with EcoRI-PstI, generating an ϳ300-base pair 5Ј C/EBP␦ cDNA fragment containing the ribosomal binding site. To directionally clone the C/EBP␦ 5Ј fragment in the antisense orientation in the PcDNA 3 expression vector, the C/EBP␦ fragment was first subcloned in the pGem4Z plasmid (Promega, Madison, WI), excised with EcoRI and HindIII, and ligated into PcDNA 3 (Invitrogen, Carlsbad, CA) in the antisense orientation. The C/EBP␦ overexpression plasmid was produced by ligating an EcoRI-HindIII restriction fragment containing the full-length C/EBP␦ cDNA into PcDNA 3. HC 11 cells were transfected with the various constructs or the PcDNA 3 vector using Transfectam (Promega), and selection and expansion of single cell transformants was carried out in the presence of 400 g/ml of Geneticin (Life Technologies, Inc.). Rescue cell lines were generated by cotransfecting the AS1 cell line with an ϳ300-base pair 5Ј C/EBP␦ cDNA sense construct (PcDNA 3) and a hygromycin resistant plasmid (ratio of 10:1). Cells were selected for by growth in G418 (400 g/ml) and hygromycin B (250 g/ml). Three colonies were chosen to propagate into cell lines, and the rest of the colonies were pooled together to create a bulk cell line.
Growth Arrest Experiments-80% confluent cells were washed with serum-free medium and cultured in medium supplemented with 0.1% FBS (growth arrest medium (GAM)). At the indicated times, cell were harvested for Northern or Western blot analysis. For cell cycle block experiments, cells were cultured for 36 h in GAM or in CGM containing either hydroxyurea (1 mM), nocodazole (500 g/ml), or amino aciddeficient medium (methionine-and isoleucine-free). For [ 3 H]thymidine experiments, cells were plated at 50% confluence in 96-well plates. Twenty four hours later, cells were switched to GAM. Cells were pulsed for 2 h with [ 3 H]thymidine (5 Ci/ml) (DuPont), harvested by precipitation with cold 5% trichloroacetic acid, solubilized in 0.2 N NaOH, and counted by liquid scintillation counting. Results presented are representative of three experiments with six wells per time point.
Northern Blot Analysis-Total RNA was isolated at the indicated times using RNAzol B (Tel Test, Inc. Friendswood, TX). Northern blots were performed with 30 g of total RNA as described (24,25). Filters were probed with the following 32 P-labeled cDNAs: C/EBP␦, C/EBP␤, CHOP, histone 2B (Oncor, Gaithersburg, MD), and Gas1. Cyclophilin receptor protein was used as a constitutive probe. Filters were visualized by PhosphorImager cassette and densitometry analysis performed by ImageQuant software (Molecular Dynamics) Growth Rate Determinations-10 3 cells were plated in individual wells in a 96-well plate. After 24 h (t ϭ 0), the relative number of viable cells was assessed using the CellTiter 96 aqueous cell proliferation kit (Promega). Cell monolayers were then washed with serum-free medium and cultured in CGM or incomplete growth medium consisting of RPMI 1640 medium plus 10, 2, or 0.5% FBS. Three and 6 days later, viable cell numbers were assessed. All viability assays were performed following the manufacturer's protocol. Results presented are representative of two experiments with three wells per time point.
Western Blot Analysis-Whole cell and cytoplasmic and nuclear proteins were harvested as described (24). Protease inhibitors (complete tablets, Roche Molecular Biochemicals) and kinase and phosphatase inhibitors (1 mM NaF, 1 mM NaVO 3 , 1 mM Na 2 MoO 4 , 10 nM okadaic acid) were added to protein isolation solutions. Proteins were quantified by Bradford method. 75 g of protein was separated by polyacrylamide gel electrophoresis and electroblotted to polyvinylidene difluoride membranes (Millipore, Bedford, MA). Western blots were performed by standard methods and visualized by ECL (Amersham Pharmacia Biotech). Antibodies and antisera used were as follows: C/EBP␦, C/EBP, p21, p16 and cyclin D1 (Santa Cruz, Santa Cruz, CA); p27 (Transduction Laboratories, Lexington, KY); Rb and phosphorylated Rb (New England Biolabs, Beverly, MA). Horseradish peroxidase-conjugated anti-mouse or anti-rabbit antibodies (New England Biolabs) were used to detect primary antibodies.
Apoptosis Experiments-Cells were plated at near confluence in 96well plates and grown to 100% confluence (about 24 h). Cells were washed with serum-free medium and cultured in GAM. Viable cell numbers were assayed at the indicated times using the CellTiter 96 aqueous cell proliferation kit (Promega). Results are representative of three experiment with six wells per time point. Annexin V binding followed by flow cytometry analysis was used to determine the percentage of apoptotic cells. At the indicated times, cell culture media containing both detached cells and the trypsinized monolayers were incubated with FITC-conjugated Annexin V solution (ApoAlert Annexin V FITC kit, CLONTECH, Palo Alto, CA). Approximately 1.5 ϫ 10 4 cells were assessed by flow cytometry and the number of FITC-positive cells, indicative of apoptosis, were counted. Results are representative of an experiment performed two times.

C/EBP␦ mRNA and Protein Is Induced in Serum-deprived
Mammary Epithelial Cells-We previously described C/EBP␦ induction in the G 0 growth-arrested COMMA D mammary epithelial cell line (24). To investigate whether G 0 induction of C/EBP␦ was a general characteristic of mammary cells, we investigated C/EBP␦ expression in three nontransformed mouse mammary epithelial cell lines, HC11, COMMA D, and NMuMG, and two transformed mammary epithelial cell lines, CCL 51 and Mm5MT. C/EBP␦ mRNA levels were extremely low in the growing mammary-derived cells (Fig. 1A, G). In contrast, C/EBP␦ mRNA levels were constitutively elevated in growing and 48 h growth-arrested NIH 3T3 cells. C/EBP␦ mRNA was barely detectable in growing and 48 h growtharrested IEC 18 cells. Unlike C/EBP␦, relatively high levels of C/EBP␤ mRNA were detected in all mammary-derived cell lines, regardless of growth status. C/EBP␤ mRNA levels were relatively low and also unrelated to growth status in NIH 3T3 cells. Histone 2B mRNA levels reflected growth status; histone 2B mRNA levels were elevated in growing cultures (Fig. 1A, G) and reduced in confluent, growth-arrested cultures (Fig. 1A, S).
Western blots of nuclear and cytoplasmic proteins were used to correlate C/EBP␦ protein content with mRNA data and to investigate C/EBP␦ subcellular localization. Consistent with the mRNA data ( Fig. 1A), C/EBP␦ protein was barely detectable in growing mammary-derived cells but markedly induced in 48 h growth-arrested cultures (Fig. 1B). In contrast, C/EBP␦ protein levels were elevated and unchanged in growing and growth-arrested NIH 3T3 cells. In all cell lines, the majority of C/EBP␦ protein was localized to the nucleus. The northern and Western blot data support a novel mammary-specific induction of C/EBP␦ during G 0 growth arrest.
C/EBP␦ mRNA Is Specifically Induced during G 0 -mediated Growth Arrest-To investigate cell cycle-dependent induction of C/EBP␦ mRNA, HC 11 cells were growth-arrested at the G 0 , G 1 , S, or G 2 phase of the cell cycle. HC 11 cells were cultured for 36 h in either growth arrest medium ( Fig. 2A, S;, G 0 block), amino acid-deprived medium ( Fig. 2A, AA; G 1 block), or complete medium containing hydroxyurea ( Fig. 2A, H; S block) or nocodazole ( Fig. 2A, N; G 2 block). Cell cycle blocks were verified by flow cytometry analysis of DNA content using phosphatidylinositol staining (data not shown). Compared with proliferating HC 11 cells ( Fig. 2A, G), C/EBP␦ mRNA was induced 7-fold during G 0 growth arrest. C/EBP␦ mRNA levels were relatively unchanged following growth arrest in other phases of the cell cycle ( Fig. 2). We also investigated other stress-related and growth arrest-specific genes. The stress response gene CHOP10 was induced in amino acid-deprived and hydroxyurea-treated HC11 cells. Like C/EBP␦, the growth arrest-specific gene Gas1, a marker for G 0 (30) , was induced only during serum withdrawal conditions. These data demonstrate a G 0specific induction of C/EBP␦ mRNA in mammary epithelial cells, suggestive of a role for C/EBP␦ in the initiation and/or maintenance of G 0 growth arrest.
Construction of HC 11 Antisense and Overexpression Cell Lines-To further investigate the role of C/EBP␦ in HC 11 mammary epithelial cells, we generated C/EBP␦ antisense and overexpression cell lines. Antisense 1 cell line (AS1), which had the greatest reduction in C/EBP␦ protein, was chosen for further analysis. AS1 C/EBP␦ protein levels were reduced by approximately 90% compared with control transfected cells (Fig. 3A). This level of C/EBP␦ reduction was similar to that which we previously reported in antisense-treated COMMA D mammary epithelial cells (24).
Despite multiple attempts, only one C/EBP␦-overexpressing colony survived drug selection, and therefore, only a single cell line (OV) was available for investigation. C/EBP␦ protein levels and subcellular localization was analyzed by immunoblotting of the OV cell line during normal growth. C/EBP␦ protein levels were increased 2.3-fold compared with growing control cells (Fig. 3B). The constitutively expressed C/EBP␦ protein, like endogenously expressed C/EBP␦, was localized in the nucleus.

C/EBP␦ Antisense and Overexpression Influences Mammary Epithelial Cell Proliferation under Suboptimal Growth
Conditions-We next plated control, AS1, and OV cells at low density (1,000 cells/well) and examined proliferation under varying growth conditions. After 3 days in CGM, all cell lines exhibited similar increases in cell numbers (Fig. 4A). This suggests that C/EBP␦ has little effect on proliferation under optimal growth conditions. However, in suboptimal growth conditions, there were marked differences between the AS1, OV, and control cells (Fig. 4, B-D). AS1 cell numbers increased 12-fold after 6 days in 0.5% serum (Fig. 4D). Control and OV cell numbers increased by only 1.8-and 2.5-fold, respectively, after 6 days in 0.5% serum (Fig. 4D). There were similar differences in proliferation between cell lines in media containing 10 and 2% FBS (Fig. 4, B and C). These data support a role for C/EBP␦ as a "conditional" cell cycle brake. Reducing C/EBP␦ levels (AS1 cells) increases proliferation under suboptimal growth condi-tions, whereas increasing C/EBP␦ levels (OV cells) decreases proliferation under suboptimal growth conditions.

C/EBP␦ Antisense and Overexpression Influences Cell Cycle Exit/G 0 Entry and the Expression of Cell Cycle Regulatory
Proteins-Eighty percent confluent control, AS1, and OV cells were switched from CGM to GAM to initiate G 0 . In control cells, [ 3 H]thymidine incorporation declined by 10% after 12 h and 61% after 24 h in GAM (Fig. 5) This demonstrates a marked acceleration of G 0 growth arrest in OV cells and a marked delay of G 0 growth arrest in AS1 cells.
We next investigated the influence of C/EBP␦ antisense and overexpression on phosphorylated retinoblastoma protein (P-Rb), Rb, cyclin D1, and the cyclin-dependent kinase inhibitor p27 during cell cycle exit. Western blot analysis of whole cell lysates was performed on growing (80% confluent) HC11 control, AS1, and OV cultures (t ϭ 0, Fig. 6). Cells were then switched from CGM to GAM to initiate G 0 growth arrest. Densitometric scanning analysis indicated that phosphorylated Rb levels declined 2-fold in HC11 controls after 48 h of G 0 growth arrest (Fig. 6). Cyclin D1 levels also declined in controls with the onset of G 0 growth arrest, reaching nearly undetectable levels after 48 h of G 0 growth arrest (Fig. 6). Control cell p27 levels increased during G 0 growth arrest. In contrast, phosphorylated Rb remained unchanged and cyclin D1 levels declined slightly in AS1 cells after 48 h in GAM. There was a modest increase in p27 levels in AS1 cells after 48 h in GAM. In OV cells, phosphorylated Rb declined 6-fold and cyclin D1 protein levels declined to nearly undetectable levels within 24 h of culture in GAM. OV cells also displayed a rapid increase in p27 protein levels after 24 h of culture in GAM. These data indicate that differences in the cell cycle exit rates measured by [ 3 H]thymidine incorporation between the control, AS1, and OV cells (Fig. 5) correlate with changes in cell cycle regulatory proteins.
C/EBP␦ Antisense and Overexpression Influences Mammary Epithelial Cell Apoptosis-Culturing postconfluent HC11 mammary epithelial cells in medium lacking serum and growth factors (0.1% FBS) induces an apoptotic response that parallels early events in the involuting mammary gland (25,31,32). Postconfluent control, AS1, and OV cells were cultured in medium containing 0.1% FBS, and the number of viable cells was assayed daily for 4 days. The number of viable control cells declined gradually, reaching 60% of the original cell number after 4 days (Fig. 7A). The OV cells followed a similar trend;

FIG. 1. C/EBP␦ expression in growing and growth-arrested mammary-derived cells, fibroblasts, and intestinal epithelial cells. A,
Northern blot analysis. Total RNA was isolated from nearly confluent (80%) growing cells (G). Cells were then cultured in growth arrest medium (0.1% FBS) for 48 h (serum-starved (S)). Blots were probed with the indicated 32 P-labeled cDNA probes. B, Western blot analysis. Cytoplasmic (C) and nuclear (N) proteins were isolated from cells treated as above. Proteins (75 g) were separated by 12.5% SDS-polyacrylamide gel electrophoresis. Filters were probed with a rabbit anti-mouse C/EBP␦ antibody and detected with a horseradish peroxidase-conjugated anti-rabbit secondary antibody. Blots were visualized with the ECL system. however, the decline in cell viability was more dramatic. Viable OV cells decreased to 65% of the original cell number after 1 day and 40% by day 4. In contrast, AS1 cell viability increased 23% after day 1. Even after 4 days, there was only a small decline in AS1 cell numbers compared with the t ϭ 0 starting time point.
The percentage of apoptotic cells was low (Ͻ4%) in confluent control, OV, and AS1 cells before the removal of serum and growth factors (day 0) (Fig. 7B). In control cultures, the percentage of apoptotic cells increased 4-fold after 1 day and nearly 20-fold after 4 days in 0.1% FBS. In OV cultures, the percentage of apoptotic cells increased 7-fold after 1 day and 12-fold after 2 days in 0.1% FBS. In contrast, there was only a slight (1.5-fold) increase in the number of apoptotic cells after day 1 in AS1 cells, and by day 4, there was only a 5-fold increase in apoptotic cells. There was no significant increase in apoptosis in any cell line when confluent cell lines were maintained in complete growth medium for 48 h (data not shown), consistent with previous reports in HC 11 cells (32,33). All of the cell lines had a similar apoptotic response when treated with apoptosis-inducing agent staurosporine, demonstrating that the programmed cell death response was functional in all the cell lines (data not shown).
Rescue of the AS1 Phenotype by Expression of C/EBP␦ Sense RNA-AS1 cells were stably transfected with a plasmid containing the same 300-base pair fragment of C/EBP␦ as the antisense plasmid but in the sense orientation. C/EBP␦ protein was elevated about 10-fold in rescue cell line R1 compared with AS1 parental cell line after 48 h in GAM (Fig. 8A). When

FIG. 2. G 0 -specific induction of C/EBP␦ mRNA in HC11 cells.
Total RNA was isolated from growing cells (80% confluent). Cells were then cultured in growth arrest (0.1% FBS) (S), complete growth medium without methionine (AA), or complete growth medium containing hydroxyurea (H) or nocodazole (N) for 36 h to block the cell cycle in G 0 , G 1 , S, and G 2 /M phases of the cell cycle, respectively. A, blots were probed with the indicated 32 P-labeled cDNA probes: C/EBP␦, Gas1 (growth arrest-specific 1), CHOP10, and cyclophilin (CP). Cyclophilin was used as a loading control. B, the harvested cells from the various treatments were stained with propidium iodide, and the percentage of cells in each cell cycle phase was assessed by fluorescence-activated cell sorter analysis. The results shown are representative of two independent experiments. FIG. 4. Growth curves of HC11 control, C/EBP␦ antisense, and C/EBP␦ overexpression cell lines in complete growth medium and low serum medium. HC11 control, AS1, and OV cells were split (10 3 /well) in a 96-well plate in CGM (RPMI ϩ 10% FBS, epidermal growth factor (10 ng/ml), insulin (10 ng/ml)). After 24 h, the medium was changed to CGM (A), RPMI ϩ 10% FBS (no epidermal growth factor or insulin) (B), RPMI ϩ 2% FBS (no epidermal growth factor or insulin) (C), or RPMI ϩ 0.5% FBS (no epidermal growth factor or insulin) (D). Cell numbers were quantitated using the CellTiter 96 aqueous cell proliferation kit (Promega). Cell numbers were determined 3 days later for CGM (A) and 3 and 6 days later for low serum medium (B-D). Results are representative of an experiment performed three times with triplicate wells per time point. Error bars represent S.D. parental AS1 cells and the rescue cell line R1 were cultured in GAM (0.5% FBS), the AS1 cells proliferated (similar to results in Fig. 4D) but the R1 cell line did not (Fig. 8B). Because altering C/EBP␦ levels influences both growth arrest and apoptosis, we next assessed cell survival (apoptosis) of postconfluent AS1 and R1 cultures in 0.1% FBS medium. Similar to results shown in Fig. 7B, there was only a relatively slight (18%) reduction in relative cell numbers in the AS1 cells (Fig.  8C). In contrast, there was a 61% reduction in relative cell numbers in R1 cells. These data show that restoring C/EBP␦ expression in the C/EBP␦ antisense cell line AS1 corrects defects in G 0 growth arrest and cell survival. This demonstrates that C/EBP␦ plays a key role in mammary epithelial cell G 0 growth arrest and cell survival. DISCUSSION This study investigated the growth regulatory role of C/EBP␦ in mouse mammary epithelial cells in vitro. In a previous report, we showed that C/EBP␦ gene expression and DNA binding activity is induced in the COMMA D mouse mammary epithelial cell line during G 0 growth arrest (24). In this report, we extend this observation, showing that C/EBP␦ is induced during G 0 growth arrest is a general property of mammary epithelial-derived cell lines. C/EBP␦ expression is unrelated to growth status in 3T3 cells, which express constitutively high levels of C/EBP␦, and the IEC18 rat intestinal epithelial cell line, which expresses relatively low levels of C/EBP␦. This indicates that C/EBP␦ functions in a mammary epithelial cellspecific G 0 growth control in vitro. We and others (25,26) have shown that C/EBP␦ is induced in mouse mammary gland in vivo during stage I of postweaning mammary gland involution. Because G 0 growth arrest precedes apoptosis in many cell types, C/EBP␦ may play a role in reprogramming mammary epithelial cell gene expression in preparation for apoptosis.
The extracellular ligand and the intracellular signal transduction pathway that results in G 0 induction of C/EBP␦ in mammary epithelial cells have not yet been identified. Factors that induce C/EBP␦ gene expression in other tissues and cell types do not appear to induce C/EBP␦ expression in cultured HC 11 mammary epithelial cells. For example, glucocorticoids induce C/EBP␦ expression in intestinal epithelial cells, lung epithelial cells and in adipocytes (20,33,34) but have no effect on C/EBP␦ mRNA levels in HC 11 cells (data not shown). Insulin induces C/EBP␦ mRNA in adipocytes (35); however, C/EBP␦ levels are low in mammary epithelial cells cultured in insulin-containing complete growth medium (24). Interleukin-6 and cAMP induce C/EBP␦ expression in a variety of cell types (22, 23, 36 -38), but treatment of HC 11 cells with interleukin-6 and cAMP analogues does not induce C/EBP␦ mRNA (data not shown). Other candidate molecules that may act as inducers of C/EBP␦ include additional cytokines of the interleukin and interferon families and cell adhesion molecules.
Growth rates were similar in AS1, OV, and control cell lines cultured in complete growth medium (Fig. 4A). This suggests that C/EBP␦ does not influence mammary epithelial cell growth under optimal growth conditions (presence of serum and growth factors). A similar observation has been reported for VHL, the first tumor suppressor found to function in the regulation of cell cycle exit (29). Growth rates in complete growth medium were similar between VHL-negative and VHL wild type renal carcinoma cells (29). VHL-negative cells, however, did not exit the cell cycle and enter G 0 growth arrest when cultured in low serum containing (growth arrest) medium (29). Reintroduction of wild type VHL restored appropriate G 0 growth arrest in VHL-negative cells (29). AS1 cells are not completely C/EBP␦-negative (AS1 cells express C/EBP␦ at about 10% of control levels), but AS1 cells do exhibit defective cell cycle exit/G 0 entry when cultured in low serum containing (growth arrest) medium. Reintroduction of wild type C/EBP␦ restored appropriate cycle exit/G 0 entry growth arrest and apoptosis in AS1 cells. These results suggest that C/EBP␦, like VHL, functions in the regulation of cell cycle exit.
The difficulty we encountered in the generation of a C/EBP␦ overexpressing mammary epithelial cell line supports a growth inhibitory role for C/EBP␦. Similar difficulties were not encountered in producing C/EBP␦ antisense cell lines or other C/EBP isoform expression cell lines. Once produced, however, the presence of elevated levels of C/EBP␦ (OV cells) did not directly induce cell cycle exit/G 0 growth arrest if cells were cultured in optimal growth medium. This suggests that either the individual surviving cell line (OV) had developed a compensatory mechanism to overcome the expression of C/EBP␦, or C/EBP␦ alone is insufficient to induce growth arrest. Additional factors, such as subcellular compartmentation and/or posttranscriptional modification of C/EBP␦ may be required for full function.
C/EBP␦ is regulated by subcellular localization in hepatocytes (39). In cultured mammary epithelial cells and the mammary gland in vivo, C/EBP␦ protein is primarily localized to the nucleus, regardless of growth or differentiation status (24). C/EBP␦ protein is also primarily localized to the nucleus in growing OV cell lines (Fig. 3B). This suggests that subcellular localization is not a major mechanism of C/EBP␦ regulation in mammary epithelial cells. The inability of C/EBP␦ to inhibit OV cell growth in complete growth medium could be due to a lack of phosphorylation in growing cells. Phosphorylation of C/EBP␦ is required for DNA binding during the hepatic acute phase response (40). In addition, C/EBPs bind DNA as homoand heterodimers (1-3, 8 -13). Even high levels of C/EBP␦ may be ineffective in blocking cell cycle progression if the appropriate dimerization partner is absent or inactive in cells cultured under optimal growth conditions.
Although C/EBP␦ overexpression in OV cells did not induce cell cycle exit/G 0 growth arrest in optimal growth medium, C/EBP␦ overexpression did accelerate cell cycle exit/G 0 growth arrest in suboptimal growth medium (growth arrest medium). This indicates that mammary epithelial cells with a ready supply of C/EBP␦ in the nucleus (OV cells) rapidly exit the cell cycle in response to growth arrest conditions. This suggests that C/EBP␦ may be limiting for cell cycle exit in mammary epithelial cells.
The cyclin-dependent kinase inhibitor p27 functions in G 0 growth arrest in a variety of cell types, including mammaryderived cells (41)(42)(43)(44). Basal p27 levels increased in confluent HC11 control and OV cultures following exposure to growth arrest medium and the initiation of G 0 growth arrest. In contrast, AS 1 cells cultured in growth arrest medium failed to induce p27 and exhibited a marked delay in the initiation of G 0 growth arrest. These results indicate an association between mammary epithelial cell C/EBP␦ levels, p27, and G 0 growth arrest in vitro. This association may extend to the mammary gland in vivo, as C/EBP␦ and p27 are both induced in the mammary gland during involution. 2 C/EBP␦ may influence cellular p27 levels by increasing p27 gene transcription or p27 protein stabilization. p27 overexpression is associated with apoptosis in a variety of cell lines, including breast cancer cell lines (45,46). Although p27 levels are primarily regulated by changes in protein stability (42,47), transcriptional control has recently been reported (48). C/EBP␣ and C/EBP␤ both transactivate the cyclin-dependent kinase inhibitor p21 waf1 promoter (49), and C/EBP␣ has been shown to directly stabilize the p21 protein without activating p21 waf1 gene expression (50). Consistent with previous reports, p21 and p16 were virtually undetectable in any of the HC 11-derived cell lines regardless of growth status (51).
Overexpression of cyclin D1 in MCF7 breast cancer cells is associated with cell cycle progression in low serum medium (52,53). When C/EBP␦ levels were reduced (AS1 cells), cyclin D1 levels remained elevated and cell cycle progression continued in low serum medium. When C/EBP␦ levels were increased (OV cells), cyclin D1 levels rapidly declined and cell cycle progres-2 L. Dearth, and J. DeWille, manuscript in preparation.
FIG. 8. Western blot, growth, and apoptosis analysis of C/EBP␦ antisense AS1 and C/EBP␦ rescue R1 cell lines. A, Western blot analysis. Nuclear proteins from growing and 48-h growth-arrested parental AS1 or rescue cell line R1. Filters were probed with a C/EBP␦ antibody. B, growth curve of rescue cell lines was as follows. Parental antisense (AS1) and rescue (R1) cell lines were split (10 3 cells) into a 96-well plate. 24 h later, viable cells were quantitated using the Cell-Titer 96 aqueous cell proliferation kit (Promega). The medium was then changed to medium with 0.5% FBS. Cell numbers were determined 4 days later. Results are representative of an experiment performed two times with triplicate wells/time point. Error bars represent S.D. C, viable cell assay. Postconfluent parental AS1 and R1 cells (day 0) were switched to apoptosis medium (0.1% FBS). At the indicated times, viable cells were quantitated as Fig. 7. Results are presented as a percentage of viable cells before culture in low serum medium and are representative of an experiment performed two times with three replicates per time point. sion stopped in low serum medium. These data indicate that C/EBP␦ and cyclin D1 are induced under opposing growth conditions. C/EBP␦ may influence cyclin D1 levels by acting as a transcriptional repressor of cyclin D1 in G 0 growth-arrested mammary epithelial cells. C/EBP␦ functions as a transcriptional repressor of the apolipoprotein C-III gene during the hepatic acute phase response (40). In addition, cyclin D1 levels are also tightly controlled at the posttranslational level by cell cycle regulated, calpain-mediated degradation (54).
The marked delay in the initiation of G 0 growth arrest observed in AS1 cells cultured in growth arrest medium was similar to previous studies carried out in our laboratory with C/EBP␦ antisense-expressing COMMA D mouse mammary epithelial cells (24). In both studies, reducing endogenous C/EBP␦ levels consistently delayed cell cycle exit. In this report, however, we have extended the analysis of the C/EBP␦ antisenseexpressing HC11 cells (AS1) to include cell cycle regulatory proteins and apoptosis. The results indicate that C/EBP␦ plays an important role in regulating cell cycle exit. When this role is compromised by reducing endogenous C/EBP␦ levels, regulation of cell cycle exit/G 0 entry and the execution of the programmed cell death response are delayed.
We previously reported a transient induction of C/EBP␦ during stage I of mammary gland involution, a physiological period associated with massive apoptosis in the mammary epithelial compartment (25). In this report we found that the percentage of cells undergoing apoptosis was increased in the C/EBP␦ overexpressing OV cells and percentage of cells undergoing apoptosis was reduced in C/EBP␦ antisense AS1 cells. Apoptosis in both cell lines, however, was similar in response to staurosporine (data not shown). This suggests that C/EBP␦ influences an upstream component of the apoptotic pathway that is activated by serum and growth factor withdrawal. Alternate pathways of apoptosis initiation and the downstream, common cell death pathway remain intact.
These results support a direct role for C/EBP␦ in the regulation of mammary epithelial cell fate after the withdrawal of serum and growth factors. However, our results cannot completely rule out the possibility that the observed effects of C/EBP␦ antisense or overexpression on mammary epithelial cell fate may involve other C/EBP family members, bZIP proteins, or other regulatory proteins. Most reports, however, indicate that C/EBP␣ is expressed at low levels in mammary epithelial cells and probably plays a relatively minor role in mammary epithelial cell growth regulation (19,24,25). The role of C/EBP␤ is uncertain. C/EBP␤ knockout mice exhibit defective mammary epithelial cell proliferation and differentiation (18,19); however, overexpression of C/EBP␤, or LIP, the dominant negative inhibitor of C/EBP␤, does not significantly alter HC11 growth control (data not shown). The roles of CHOP10 or other bZIP proteins in mammary epithelial growth regulation are not well understood.
Mammary epithelial cell growth, differentiation, and death is controlled by endocrine and paracrine signals (27). A better understanding of these extracellular ligands, the intracellular signaling pathways they activate, and their nuclear targets will provide a clearer picture of mammary gland growth regulation and potentially new insights into the etiology and progression of breast cancer. The growth suppressor activity of C/EBP␦ is similar to that described for the VHL gene product (29). In addition, C/EBP␦ also plays a role in mammary epithelial cell apoptosis. Many well described tumor suppressor/ growth arrest genes, such as BRCA1 (55), APC (56), p53 (57), p33 ING1 (58), CHOP (59), and the cyclin-dependent kinase inhibitor p27 (45), also induce apoptosis. Experiments are under way to characterize extracellular ligands, their receptors, and intracellular signaling pathways that induce C/EBP␦ gene transcription in mammary epithelial cells.