CCAAT/enhancer-binding Protein-Delta (C/EBP-delta) Is Induced in Growth-Arrested Mouse Mammary Epithelial Cells

Abstract CCAAT/enhancer binding proteins (C/EBPs) are a highly conserved family of DNA-binding proteins that regulate cell growth and differentiation in a highly tissue-specific manner. These experiments investigated the influence of the cell cycle on C/EBP isoform expression in mammary epithelial cells (COMMA D) and fibroblasts (NIH3T3). C/EBP-δ gene expression is induced in COMMA D cells arrested in G0 by serum and growth factor withdrawal or contact inhibition. C/EBP-δ mRNA, nuclear protein content, and DNA binding activity increase during G0 growth arrest and decrease after cell cycle induction in COMMA D cells. Growth arrest is markedly delayed in COMMA D cells expressing a C/EBP-δ antisense construct. C/EBP-β is induced during G1 of the cell cycle. In contrast to COMMA D cells, C/EBP-β and C/EBP-δ mRNA levels remain relatively constant in growth-arrested and cell cycle-induced NIH3T3 cells. However, C/EBP homologous protein (CHOP10) mRNA levels markedly increase in growth-arrested NIH3T3 cells. Both COMMA D and NIH3T3 cells express growth arrest-specific (gas1) and JunD during G0. These results demonstrate that COMMA D and NIH3T3 cells achieve a common growth arrest (G0) state by cell-specific strategies that involve the induction of different C/EBP isoforms.

C/EBP gene expression is regulated at multiple levels. A variety of external stimuli, including cytokines, hormones, and growth factors influence C/EBP isoform transcription rates (5,10,12). In addition, C/EBP-␣ and C/EBP-␤ mRNAs undergo alternative AUG initiation codon usage, resulting in the translation of two protein products with different biological activities from a single mRNA (13,14). Post-translational control by phosphorylation influences the subcellular localization and DNA binding activity of C/EBP isoforms (15)(16)(17).
C/EBPs have been implicated in the regulation of growth and differentiation of a wide variety of cells including hepatocytes, adipocytes, intestinal epithelial cells, and muscle and myelomonocytic cells (2-10, 18 -21). C/EBP-␣ induces growth arrest and the expression of differentiation-specific genes in hepatocytes and 3T3-L1 cells (4 -6, 8, 9). C/EBP-␤ (NF-IL6, LAP) has been implicated in growth control of hepatocytes and the direct conversion of NIH3T3 cells to adipocytes (20,21). Another C/EBP family member, C/EBP homologous protein (CHOP10), functions in G 0 and G 1 /S phase growth control in NIH3T3 cells (22). Translocations involving CHOP10 have been identified in a high percentage of human mixoid liposarcomas (23). C/EBP-␦ has not been previously associated with growth arrest (20); however, C/EBP-␦ has been linked to early events in the hormone-dependent differentiation of 3T3-L1 cells and to the acute phase response in liver (3,10,12).
The overall goal of this study was to investigate the influence of the cell cycle on C/EBP isoform expression in COMMA D mammary epithelial cells. The mammary gland differs from most tissues in that it attains full functional capacity relatively late in life, at sexual maturation (24,25). In the adult female, the mammary gland retains the capacity to undergo hormoneinduced proliferation, differentiation, and involution throughout the reproductive years (24,25). The growth control mechanisms underlying this unique biology are poorly understood.
Much of our current understanding of growth control and the mammalian cell cycle is derived from studies with fibroblast cell lines (26). Epithelial cells, however, utilize unique transcriptional control mechanisms (27) and express some early growth response genes that differ from fibroblasts and lymphocytes (28). Epithelial cell-specific differences in the expression of growth response genes may also extend to growth arrest response genes (29). Epithelial cells are commonly implicated in human cancers (28). Understanding the growth arrest response in mammary epithelial cells is of particular significance as breast cancer is the leading cause of cancer and the second leading cause of cancer deaths among women in the United States (30). Our results demonstrate that growth arrest is associated with the expression of different C/EBP isoforms in mammary epithelial cells and fibroblasts. The specific induction of C/EBP-␦ in growth-arrested COMMA D mammary epi-thelial cells and its disappearance during the cell cycle suggest a novel tissue-specific growth regulatory role for this C/EBP isoform in mammary epithelial cells. This regulatory role is supported by a delay in growth arrest of COMMA D cells expressing C/EBP-␦ antisense.

EXPERIMENTAL PROCEDURES
Cell Culture-COMMA D, a nontransformed line of mammary epithelial cells (31,32), were cultured in complete growth medium containing Dulbecco's modified Eagle's medium with 4.5 g/liter D-glucose and supplemented with 10% fetal bovine serum, 10 ng/ml epidermal growth factor, 10 g/ml bovine insulin, 5 units/ml penicillin, and 5 g/ml streptomycin. NIH3T3 cells (ATCC CRL 1658) were cultured in similar complete growth media without the addition of epidermal growth factor. Cell cycle experiments were carried out by incubating near-confluent cultures for 72 h in growth arrest media containing Dulbecco's modified Eagle's medium with 4.5 g/liter D-glucose, 0.5% fetal bovine serum, 5 units/ml penicillin, and 5 g/ml streptomycin. After reintroducing complete media, cells were harvested at the designated times for RNA or protein analysis. Growth arrest experiments were carried out by switching near-confluent cultures from complete growth media to growth arrest media and harvesting at the designated times. To produce stable antisense cell lines COMMA D cells were transfected with a C/EBP-␦ 400-base pair partial cDNA fragment (antisense orientation) and vector controls (pcDNA3, Invitrogen, Palo Alto, CA). Stable cell lines were drug-selected, and single cell clones were isolated and expanded as described (33).
Northern and Western Blot Analysis-Total RNA was isolated using RNAzol B (Tel-Test, Inc., Friendswood TX). Northern blots were carried as described (34). Blots were probed with the following random primer labeled cDNAs: C/EBP-␣, C/EBP-␤, C/EBP-␦, CHOP10, gas1 cDNA, and histone 2B (Oncor, Gaithersburg, MD). Cyclophilin receptor protein partial cDNA was used as a constitutive probe (34). Results are representative of 2-4 experiments. Western blots were performed by standard procedures (35). Rabbit antisera against a C/EBP-␦ peptide (ARG-PLKREPDWGDGDA (2)) was used to detect C/EBP-␦. Rabbit anti-C/ EBP-␤ and anti-proliferating cell nuclear antigen antisera were purchased from Santa Cruz Biotechnology, Santa Cruz, CA. A goat anti-rabbit antibody was used as a secondary antibody, and the signal was developed with the ECL system (Amersham).

RESULTS
Cultures were synchronized in a growth-arrested (G 0 ) state by serum and growth factor withdrawal for 72 h, and the cell cycle was initiated by the introduction of complete growth media. Growth arrest (G 0 ) and synchronous cell cycle progression were verified by [ 3 H]thymidine incorporation (data not shown). C/EBP-␦ mRNA content was elevated in growth-arrested (G 0 ) COMMA D cells after 72 h of serum and growth factor withdrawal ( Fig. 1). C/EBP-␦ mRNA content decreases rapidly (within 1 h of cell cycle induction), increases slightly during G 1 /early S phase (4 -12 h after cell cycle induction), and then declines during the remainder of the cell cycle. In contrast, C/EBP-␤ mRNA content increases during early G 1 /S phase, then gradually declines to base-line levels. Growth arrest-specific 1 (gas1) mRNA, originally characterized in growth-arrested NIH3T3 cells (36,37), is highly induced in quiescent (G 0 ) COMMA D cells, declines during G 1 , and then increases somewhat through S phase (36,37). Histone 2B (H2B) mRNA content was minimal after 72 h of growth arrest, began to accumulate 8 h after cell cycle induction, peaked at 18 -20 h, and then declined.
Growth arrest (G 0 ) and cell cycle progression had little effect on C/EBP-␤ or C/EBP-␦ mRNA levels in NIH3T3 cells. gas1 mRNA levels exhibited a growth-regulated biphasic pattern (high in G 0 , low in G 1 , moderate in S), similar to that observed in COMMA D cells and previous reports from NIH3T3 cells (36,37). H2B mRNA began to accumulate to significant levels in NIH3T3 cells 8 -12 h after cell cycle induction, consistent with a synchronous cell cycle.
To investigate the association between C/EBP-␦ and the onset of growth arrest, time points were taken 1, 12, and 24 h after growth arrest induction by serum and growth factor withdrawal. In COMMA D cells, C/EBP-␦ mRNA levels began to accumulate after 12 h and persisted after 24 h of growth arrest induction (Fig. 2). CHOP10 mRNA content decreased, and C/EBP-␤ and JunD mRNA levels increased slightly after 24 h of growth arrest induction.
In NIH3T3 cells, C/EBP-␦ and C/EBP-␤ mRNA levels remain relatively unchanged 24 h after growth arrest induction (Fig.  2). In contrast to COMMA D, CHOP10 mRNA content was markedly increased 24 h after growth arrest induction in NIH3T3 cells. JunD, a previously described negative growth regulator in NIH3T3 cells (38), increased slightly after 24 h of growth arrest in NIH3T3 cells.
The growth arrest response initiated by serum and growth factor withdrawal may differ from density-dependent growth arrest, even though both induce a G 0 state (39). To investigate the influence of density-dependent growth arrest on C/EBP-␦ expression, RNA was isolated from COMMA D cells after high density plating (day 0) and grown to confluence in complete media (days 1 and 2). H2B mRNA content decreased as COMMA D cells reached confluence (day 2) (Fig. 3A). C/EBP-␦ mRNA content was low in growing and near-confluent cells (days 0 and 1). C/EBP-␦ mRNA content increased in concert with the initiation of density-dependent growth arrest and the decline in H2B mRNA content (day 2). Day 2 COMMA D cultures were refed fresh growth media, and one synchronous round of the cell cycle was initiated as evidenced by an increase in H2B mRNA content and a decrease in C/EBP-␦ mRNA content (day 3). This was followed by decreasing H2B mRNA and increasing C/EBP-␦ mRNA levels (day 4 and 5) as COMMA D cells re-enter a density-dependent, growth-arrested state.
Histone 2B mRNA content declined as NIH3T3 cells entered density-dependent growth arrest (Fig. 3B). C/EBP-␦ mRNA content, however, remained relatively high in NIH3T3 cells regardless of cell density or growth status. C/EBP-␦ mRNA was also induced in a second density-dependent growth-arrested mouse mammary epithelial cell line, NMuMG cells (data not shown).
COMMA D C/EBP-␦ protein content declines during the cell cycle. Nuclear C/EBP-␦ protein levels are elevated in 72-h growth-arrested COMMA D cells, then decline as cells synchronously progress through the cell cycle (Fig. 4A). The identity and potential function of a second lower molecular weight band in the nuclear, but not the cytoplasmic, compartment is unknown. As COMMA D cultures synchronously progress through the cell cycle, the nuclear C/EBP-␦ protein content declines, and the cytoplasmic C/EBP-␦ content increases slightly. This suggests some degree of subcellular localization of C/EBP-␦ in growing versus growth-arrested COMMA D cells. C/EBP-␤ protein content, which is predominately localized to the nucleus in COMMA D cells, increases slightly 6 and 12 h after cell cycle induction (data not shown). Proliferating cell nuclear antigen is detectable in the nucleus 12 h after cell cycle induction, consistent with normal cell cycle progression (Fig. 4B).
In contrast to the cell cycle, C/EBP-␦ nuclear protein content increases in growth-arrested COMMA D cells. Low levels of C/EBP-␦ protein are detectable in the cytoplasm, but not the nucleus, of growing COMMA D cells (Fig. 5A, lanes 1 and 2). C/EBP-␦ nuclear protein content increases significantly after 48 h of growth arrest (Fig. 5A, lanes 3 and 4). Increases in COMMA D C/EBP-␦ nuclear protein content are detectable as early as 8 h after the withdrawal of serum and growth factors (data not shown). C/EBP-␤ protein, which is also primarily localized to the nucleus in COMMA D cells, was relatively unchanged after 2 days of growth arrest (Fig. 5B, lanes 1-4). Specific binding to a C/EBP consensus site increases in nuclear extracts from 2-and 4-day growth-arrested COMMA D cells compared to day 0 (growing) cells (Fig. 6A). No binding was detected to a C/EBP mutant site (Fig. 6A). Preincubation of COMMA D nuclear extracts with anti C/EBP-␦ antisera, but not preimmune sera, decreased C/EBP binding activity (Fig. 6, B and C).
To more directly address the role of C/EBP-␦ in COMMA D cells, stable cell lines expressing a C/EBP-␦ antisense construct were produced. Growth arrest induces C/EBP-␦ protein levels in control (vector-transfected) COMMA D cells, but not in the C/EBP-␦ antisense expressing cell line, ␦3 (Fig. 7A). The influence of the C/EBP-␦ antisense expression on the growth arrest response of ␦3 cells was assessed. C/EBP-␦ antisense expression in ␦3 cells significantly delays the ␦3 growth arrest response compared with controls. [ 3 H]thymidine incorporation (indicator of DNA replication) rapidly declines (within 12 h) in control cells after serum and growth factor withdrawal of nearconfluent cultures (Fig. 7B). [ 3 H]Thymidine incorporation continues to decline to near base-line levels after 48 h of growth arrest in control cells. In ␦3 cells expressing C/EBP-␦ antisense, [ 3 H]thymidine incorporation remains elevated for 12-18 h, then begins to decline for the first 30 h of the growth arrest response. In addition, C/EBP-␦ antisense expression in ␦3 cells results in a slower decline in ␦3 histone 2B mRNA content and a slower accumulation of ␦3 gas1 mRNA over the first 30 h of the growth arrest response. DISCUSSION The mammary gland exhibits a unique tissue-specific pattern of growth and development that is well-characterized at the hormonal level but poorly understood at the molecular level. These results implicate C/EBP-␦ in the growth arrest response of COMMA D cells, a mouse mammary epithelial cell line. The results also demonstrate differences between the expression and potential function of C/EBP-␦ in COMMA D cells compared with fibroblast-derived cells. Earlier reports with FIG. 3. Northern blot analysis of COMMA D and NIH3T3 cells during density-dependent growth arrest (G 0 ). A, COMMA D cells were plated at high density in growth induction media (t ϭ 0). After sampling on day 2, cells were refed fresh growth induction media and remained in this media (days [3][4][5]. B, NIH3T3 cells were plated at high density in growth induction media (t ϭ 0). Cells remained in this medium (days 1-3). Total RNA was isolated at selected time points, and Northern blots were coprobed with C/EBP-␦ and histone 2B. 3T3-L1 cells indicated that C/EBP-␦ expression peaks during active cell division and decreases as 3T3-L1 cells withdraw from the cell cycle and terminally differentiate (40). A relatively similar temporal pattern of C/EBP-␦ expression occurs in differentiating myelomonocytic cells (19). The tissue-specific functions of C/EBP-␦ are also apparent in liver, where C/EBP-␦ expression increases early in the proliferative response following partial hepatectomy and during the acute inflammatory response (9,10,16,41). In response to inflammation, C/EBP-␦ transactivates the human complement component 3 and serum amyloid A1 genes (10,16,41). These results indicate that C/EBP-␦ is expressed in a wide range of tissues, but its pattern of induction and potential role in cell function, particularly growth regulation, is highly tissue-specific.
The association between C/EBP-␦ and growth arrest in mammary epithelial cells may be specific to treatments that induce withdrawal from the cell cycle, or G 0 . For example, serum/ growth factor withdrawal and contact inhibition both induce G 0 and both induce C/EBP-␦. If C/EBP-␦ is blocked by antisense RNA expression, G 0 induction by serum and growth factor removal is markedly delayed. In addition, cell lines expressing C/EBP-␦ antisense RNA grow to a higher cell density than the parental COMMA D cells and vector-transfected controls (data not shown). This suggests that the removal or inactivation of C/EBP-␦ inhibits the initiation of G 0 by contact inhibition. Finally, drugs that induce blocks within the cell cycle, such as hydroxyurea (G 1 /S block) or nocodazole (G 2 block), do not in-duce C/EBP-␦ (data not shown). Similar perturbations of the growth arrest response occur in hepatoma cells expressing C/EBP-␣ antisense RNA, indicating that C/EBP isoforms may function in the growth arrest response of a variety of tissues (42). It is noteworthy that cells expressing antisense RNA to the known tumor suppressor genes BRCA1 and Waf1 are also aberrant in G 0 induction and maintenance (43,44).
The in vitro induction of C/EBP-␦ in mammary epithelial cells after serum and growth factor withdrawal parallels in vivo observations during the immediate postweaning period. 2 Mammary gland C/EBP-␦ gene expression is induced early in the postweaning period in conjunction with the decline in lactogenic hormones and preceding the onset of apoptosis in the mammary epithelial compartment (45). These results suggest that C/EBP-␦ may play a role in reprogramming the mammary epithelial cell after the removal of hormone and growth factor stimulation. The specific gene(s) transactivated by C/EBP-␦ during mammary epithelial cell growth arrest are unknown. 2 A. Gigliotti and J. DeWille, manuscript in preparation. Asynchronously growing cells (t ϭ 0) were growth-arrested by switching to growth arrest media. Two hours prior to harvest at the indicated time points, cultures are pulsed with 1 Ci/ml of [ 3 H]thymidine (specific activity, 70 -90 Ci/mmol). Cells were washed in phosphate-buffered saline, lysed with ice-cold 10% trichloroacetic acid, and precipitates were collected onto glass microfiber filters (Millipore, Bedford, MA), washed, dried, and counted in a liquid scintillation counter (6 replicates/time point). Data are presented as mean Ϯ S.E.). C, Northern blot analysis. Total RNA was isolated from cells treated as described in A and B above and sequentially probed with histone 2B, gas1, and cyclophilin.
However, C/EBP isoforms have been implicated in the transactivation of the ␤-casein gene during lactation (46,47).
The rapid decline in C/EBP-␦ mRNA following cell cycle initiation suggests that C/EBP-␦ expression is tightly linked to the growth-arrested state in COMMA D cells. Similar results have been obtained in HC11 cells, a subline of the COMMA D cell line that retains the capacity for differentiation (data not shown). Preliminary experiments indicate that C/EBP-␦ mRNA has a relatively short half-life, about 30 min, in both growing and quiescent COMMA D cells (data not shown). This suggests that the rapid decrease in C/EBP-␦ mRNA content (10-fold decrease within 1 h of cell cycle induction) is due to a decrease in C/EBP-␦ gene transcription.
The role of C/EBP-␦ in mammary epithelial cell growth control may be tissue-specific. C/EBP-␣ has been consistently associated with growth arrest in adipocytes and hepatocytes (4,8,9). Under the conditions employed in these experiments, neither cell type expressed significant amounts of C/EBP-␣ (3-9). Although C/EBP-␤ has been shown to inhibit hepatoma growth, transient increases in C/EBP-␤ gene expression, nuclear localization, and DNA binding activity occur during the early G 1 phase of the cell cycle and after partial hepatectomy (9,48,49). In mammary epithelial cell cultures, C/EBP-␤ mRNA and nuclear protein content also transiently increase during G 1 . This is consistent with a role for C/EBP-␤ in the induction of immediate early gene promoters (48,49). Mammary gland C/EBP-␤ mRNA and protein levels are elevated during the proliferative phase of pregnancy/lactation (45). Murine CHOP10 shares extensive sequence homology with the hamster growth arrest DNA damage (GADD 153) gene (22,50,51). CHOP10 mRNA is constitutively expressed in adult skeletal muscle, liver, heart, and brain (22). CHOP10 mRNA is also induced by growth arrest and DNA damage, although the mechanisms of induction under these different conditions may vary (22). Microinjection experiments suggest a role for CHOP10 at the G 1 /S checkpoint (50). The rapid and persistent elevation in CHOP10 mRNA content in response to serum and growth factor withdrawal in NIH3T3 cells further supports a role for CHOP10 in the initiation and maintenance of the growth arrest response in fibroblasts. In COMMA D cells, CHOP10 mRNA content declines somewhat early in the growth arrest response, then increases in long term growth arrest, possibly as a result of glucose deprivation in long term cultures (data not shown). Although CHOP10 induction patterns differ between growth-arrested COMMA D and NIH3T3 cells, CHOP10 is induced in both cell types in response to DNA damaging agents (data not shown). Growth arrest-specific (gas) genes, which were originally described in NIH3T3 cells, are induced during growth arrest and down-regulated during G 1 (36,37). Although previous reports suggested that gas genes were not expressed in epithelial cells (29), we observed biphasic cell cycle-regulated gas1 mRNA patterns in both COMMA D and NIH3T3 cells (Fig. 1). These results indicate that gas1 plays a general role in growth control in both fibroblasts and epithelial cells. It is of interest that gas1 and C/EBP-␦ are representative of a small group of genes that are induced during G 0 . The novel induction of C/EBP-␦ during G 0 in mammary epithelial cells suggests that understanding its regulation may have important implications in aberrantly growing cells or in breast cancer.
Both C/EBP-␦ and CHOP10 form dimers with other bZip proteins and function as DNA-binding proteins (3,10,16,51). Studies are in progress to identify C/EBP-␦ dimerization partners and the genes controlled by these transcription factor complexes during mammary epithelial cell growth arrest. Studies are also in progress to define the molecular mechanisms underlying C/EBP-␦ induction during growth arrest. Since the C/EBP-␣ gene exhibits transcriptional autoregulation, similar regulatory mechanisms may also influence the expression of other C/EBP family members, such as C/EBP-␦ (5). Primer extension analysis indicates that the C/EBP-␦ transcription initiation site is similar in growing COMMA D cells, growtharrested COMMA D cells, involuting mammary gland, and NIH3T3 cells (data not shown). This suggests that divergent biological signals may converge into a single, or limited number, of transcriptional activators of the C/EBP-␦ gene.