Differential Regulation of Elafin in Normal and Tumor-Derived Mammary Epithelial Cells Is Mediated by CCAAT/Enhancer Binding Protein β Free

The C/EBP (CCAAT/enhancer binding protein) family of transcription factors is characterized by the basic region/leucine zipper structure (1). One member of this family, C/EBPβ, has recently proven to be important for the development, differentiation, and proliferation of mammary epithelial cells (2–5). Knock-out of the C/EBPβ gene in mice leads to defective mammary gland differentiation and failure to ovulate (6–8), providing evidence that C/EBPβ regulates mammary differentiation through specific genes.

The alternative translation of the intronless C/EBPβ gene produces three different protein isoforms (9, 10). Liver-enriched activator protein 1 (LAP1) and LAP2, initiated at the first and second in-frame methionines, respectively, activate transcription. On the contrary, the truncated isoform of C/EBPβ, liver-enriched inhibitory protein (LIP), is translated from the third in-frame methionine and lacks most of the NH₂-terminal transactivation domain. Hence, LIP is unable to activate gene transcription and inhibits the LAP activity as a transcriptional repressor by forming LAP/LIP dimerization (6, 7). Thus, the LAP/LIP ratio is an important indicator of C/EBPβ transcriptional activity and plays a crucial role in the various biological consequences through regulating gene expression.

Although a number of reports indicate that C/EBPβ are directly or indirectly involved in many changes in gene expression that lead to cellular differentiation or transformation, specific downstream targets for C/EBPβ that would negatively regulate mammary epithelial cells from tumorigenesis are not well characterized. Recently, we have focused on a serine protease inhibitor, elafin, as a candidate gene that is regulated by C/EBPβ in mammary epithelial cells. A variety of serine proteases, including urokinase-type plasminogen activator, plasmin, and polymorphonuclear leukocyte elastase (PMN-E) plays a pivotal role not just in inflammation, but also in the processes of tumor cell invasion and metastasis in human breast cancer (11–13). PMN-E mediates the degradation of extracellular matrix proteins, such as elastin, collagen, and proteoglycans (14). Recently, the prognostic potential of PMN-E has been extensively examined in primary breast cancer patients (15–18). A study measuring PMN-E in primary breast tumor tissues revealed that increased levels of PMN-E have been correlated with a poor metastasis-free survival and overall survival in a large series of primary breast cancer patients (15). Elevated expression of PMN-E in breast tumor tissue also predicts poor responsiveness to chemotherapy in patients with metastatic breast cancer (16, 19).

Activation of elastase is counterbalanced by elafin, a serine protease inhibitor. Elafin is also known as skin-derived anti-leukoproteinase (SKALP), an epidermal serine proteinase inhibitor (20). A number of studies have shown that SKALP/elafin is specifically expressed in inflammatory skin diseases such as psoriasis (21). Whereas initial studies have associated elafin with inflammation and protection against tissue damage, our recent interest has focused on its possible roles in tumor suppression. Zhang et al. identified elafin by subtractive hybridization, comparing genes expressed in normal mammary epithelial cells with those from breast carcinomas. Elafin was shown to be differentially expressed between normal cells and tumor cells (22). This finding raises a possibility that the loss of expression of elafin may contribute as importantly to carcinogenesis of breast epithelial cells as to the overexpression of their proteinase. However, the molecular mechanism underlying the regulation of the elafin gene in mammary epithelial cells still remains unclear. In this study, we provide evidence that elafin is a novel substrate of C/EBPβ. Our results show that the elafin gene is transcriptionally down-regulated through its promoter in most breast tumor cell lines. This expression pattern seems to be correlated with that of the C/EBPβ-LAP/LIP ratio. Functional analysis of the elafin promoter identified the multiple C/EBPβ binding sites as cis-elements involved in transcriptional activation of normal cells. Electrophoretic mobility-shift assay (EMSA) and chromatin immunoprecipitation (ChIP) assay revealed that C/EBPβ protein specifically interacts with these C/EBPβ binding sites in a different manner between 76NE6 normal cells and MDA-MB-231 tumor cells. We also show that the expression of the elafin gene is regulated by C/EBPβ in a negative feedback loop manner. Furthermore, the increase in C/EBPβ levels (LAP/LIP ratio) can activate the transcription of the endogenous elafin gene through its promoter in MDA-MB-231 tumor cells. Collectively, our data show that the mechanism of down-regulation of elafin in breast cancer cells is through lack of its transcriptional activation by C/EBPβ.

Cell culture. Serum was from Atlanta Biologicals, Inc., and cell culture medium was from HyClone. The immortalized mammary epithelial cells, 76NE6, 76Nf2v, 76NY54H, 76NE7, and MCF-10A, were each immortalized by different means. 76NE6 was generated by transfection of the E6 gene of the human papillomavirus (HPV) into the mortal 76N cell. The mechanism by which E6 mediates immortalization is, in part, through p53 degradation via the E6AP (E6-associated protein), a ubiquitin ligase. E6AP protein binds to p53 and mediates its degradation (23, 24). 76Nf2v and 76NY54H represent two cell lines that were transfected with different E6 mutants (f2v and Y54H), which are unable to induce degradation of p53, but are still able to immortalize the 76N mammary epithelial cells (23). The 76NE7 immortalized cell line was generated by transfection of 76N cells with the E7 oncogene of HPV (25). The immortalization of E7 is through the inactivation of the pRb pathway as E7 binds to pRb and inactivates its function by targeting ubiquitin-mediated degradation (26). We also evaluated the MCF-10A immortalized cells, which were generated by the continuous culturing of the parental mortal cell (MCF-10M) in serum-containing medium for about 800 days, at which point they spontaneously generated two immortal sub-lines, MCF-10A (attached cells) and MCF-10F (floating cells; ref. 27). Both were near diploid with a shared t(3;9) translocation and have been used as normal immortalized breast epithelial controls for studies of human breast cell lines (28). The culture conditions for each cell line were described previously (29–32). All cells were cultured and treated at 37°C in a humidified incubator containing 6.5% CO₂–93.5% air.

Patient samples. Tumor and adjacent normal tissue were obtained at the time of surgical intervention from 152 patients with stage I, II, or III carcinoma of the breast. Patients signed informed consent for participation in this study, which was approved by the Institutional Review Board of the M.D. Anderson Cancer Center. Following surgical resection of the primary tumor and regional lymph nodes, specimens were examined in pathology, and fresh tumor and normal tissue were collected by the pathologist and processed in the laboratory to obtain protein lysates. Clinical and pathologic information were prospectively collected on each of the patients, including lymph node status, estrogen receptor (ER) status, progesterone receptor (PR) status, and HER-2/neu status. Patients were followed for the long-term clinical end points of recurrence, development of metastatic disease, and overall survival.

Plasmid preparation. The human elafin-luciferase fusion reporter plasmid, termed pSPL1000, and its deletion mutants, termed pSPL440 and pSPL290, were gifts from Dr. Arno Pol (University Medical Center St. Radboud, the Netherlands; ref. 33). Construct pSPL1000 was used as a template in the PCR synthesizing the 185-bp promoter fragment (position −172/+13; all nucleotide positions in this paper are given relative to the start site of translation). Two primers (forward: 5′-ATCCCGGGTAATCCTGAGGGAAAGCCCC-3′, position −172/−153, reverse: 5′-TGCTGGCCCCCATGGTGTCA-3′, position −7/+13) were used to introduce a SmaI and a NcoI sites at the 5′ and the 3′ site of the PCR product, respectively. A 185-bp SmaI/NcoI fragment was directionally subcloned in frame with the firefly luciferase gene in pSLA4 vector, a modified pSLA3 vector, directly fused to the ATG of the luciferase gene through the NcoI site. The resulting clone was designated as pSPL172. Similarly, the 107-bp promoter fragment (position −94/+13) was generated using construct pSPL1000 as a template. Two primers (forward: 5′-ATCCCGGGTAAATACCACAGACCCGCCC-3′, position −94/−74, reverse: 5′-TGCTGGCCCCCATGGTGTCA-3′, position −7/+13) were used to introduce a SmaI and a NcoI sites at the 5′ and the 3′ site of the PCR product, respectively. This 107-bp SmaI/NcoI fragment was subcloned into pSLA4 vector as mentioned above. The resulting clone was designated as pSPL94. Reporter plasmids with mutation(s) in the C/EBPβ or activator protein 1 (AP-1) binding sites were generated by site-directed mutagenesis using the Quick Change XL Site-Directed Mutagenesis Kit (Stratagene) with synthesized oligonucleotides as follows: pSPL440m1: sense, 5′-GGAGAAACACCTTGGTTTTGGCCGCAAGACTGGATCTACCAGTG-3′; and antisense, 5′-CACTGGTAGATCCAGTCTTGCGGCCAAAACCAAGGTGTTTCTCC-3′; pSPL440m2: sense, 5′-GACTGGATCTACCAGTGACGGTCTGAATAACCTTCGGTGATTCC-3′; and antisense, 5′-GGAATCACCGAAGGTTATTCAGACCGTCACTGGTAGATCCAGTC-3′; pSPL440m3: sense, 5′-CTCTTCTTGGGTCTCACTGTCGGTCAAAACATGAAGAATTTCATTG-3′; and antisense, 5′-CAATGAAATTCTTCATGTTTTGACCGACAGTGAGACCCAAGAAGAG-3′; pSPL440m4: sense, 5′-GAATTTCATTGTAATGTTACCGCCGAAGTGAGCCAGCACTTCTAC-3′; and antisense, 5′-GTAGAAGTGCTGGCTCACTTCGGCGGTAACATTACAATGAAATTC-3′; pSPL440m5: sense, 5′-GGGACAATCAGAGATGATGTGATGGCCGGTCCATTAGTTCTTCC-3′; and antisense, 5′-GGAAGAACTAATGGACCGGCCATCACATCATCTCTGATTGTCCC-3′; pSPL440m6: sense, 5′-CCCTCCCAGAATGGGGTGGATCCCTACCAATACAGCTAAGG-3′; and antisense, 5′-CCTTAGCTGTATTGGTAGGGATCCACCCCATTCTGGGAGGG-3′; pSPL440 mAP-1: sense, 5′-GTAATGTCCATTAGTTCTTCCTGGTAGTCGTCCTGAGGGAAAGCCCC-3′; and antisense, 5′-GGGGCTTTCCCTCAGGACGACTACCAGGAAGAACTAATGGACATTAC-3′. Each sequence is identical to that of pSPL440 except for the underlined sequence. The generated constructs were confirmed by sequencing.

To construct C/EBPβ overexpressing plasmid, a 1,310-bp fragment of C/EBPβ cDNA (GenBank NM_005194) was amplified by reverse transcription-PCR (RT-PCR) using Ready-To-Go You-Prime First-Strand Beads (Amersham Biosciences) and Taq DNA polymerase (Qiagen). Total RNA was extracted from 81N normal epithelial cells as mentioned below. Following the first strand synthesis, amplification was carried out by 34 cycles of PCR at 94°C for 30 s, 60°C for 30 s, and 72°C for 2 min using the following two primers: 5′-CCAGCCACCAGCCCCCTCACTAATA-3′ and 5′-AAGTGCCCCAGTGCCAAAGTTTTGC-3′. Gel-purified RT-PCR product was sequenced and subcloned into pCR3.1 plasmid (Invitrogen and Life Technologies) to generate pCR3.1-C/EBPβ. Correct integration of C/EBPβ cDNA into pCR3.1 vector was confirmed by digestion with restriction enzyme and sequencing.

Northern blot analysis. Total RNA was isolated from cells grown in a 150-mm-diameter dish using a guanidinium isothiocyanate and cesium chloride. Northern blot analysis was done as previously described (34). Twenty micrograms of total RNA per lane was used for Northern blot analysis. The DNA probes were prepared by using Random Primed DNA Labeling Kit (Roche) with [α-³²P] dCTP. 36B4 mRNA and 28S rRNA bands on ethidium bromide–stained gel were used as a loading control.

Protein isolation and Western blot analysis. Cell lysates were prepared and subjected to Western blot analysis as described previously (35). Briefly, 50 μg of protein were subjected to electrophoresis on SDS-PAGE and transferred to Immobilon P overnight at 4°C at 35 mV constant voltage. The blots were blocked overnight at 4°C in BLOTTO [5% nonfat dried milk in 20 mmol/L Tris, 137 mmol/L NaCl, and 0.05% Tween (pH, 7.6)]. After being washed, the blots were incubated in primary antibodies for 2.5 h. Primary antibodies used were elafin (HM2063; HyCult Biotechnology B.V.), C/EBPβ (C-19; Santa Cruz Biotechnology), and actin (Chemicon International, Inc.). Blots were then incubated with goat anti-mouse or anti-rabbit immunoglobulin–horseradish peroxidase conjugate at a dilution of 3:5,000 in BLOTTO for 1 h, finally washed and developed by the Renaissance chemiluminescence system (Perkin-Elmer Life Sciences, Inc.). Actin was used to standardize equal protein loading. Expression of C/EBPβ isoforms (LAP/LIP) was quantified by IPLab Gel Quantitation Software and normalized by actin.

Transient transfection and luciferase assay. Cells were transfected with the elafin promoter plasmid and the pRL-TK vector by using Genejuice Transfection Reagent (Novagen) according to the manufacturer's instructions. At 24 h after transfection, the cell lysates were collected for a luciferase assay. Cells were washed twice with PBS and resuspended in lysis buffer (Amersham Biosciences). Three cycles of freeze-thawing were done in liquid nitrogen/water at 37°C. Because the pRL-TK vector expresses the Renilla luciferase, which can be measured separately as a control for transfection efficiency, the firefly luciferase activity reflecting the elafin promoter from each sample was normalized by its Renilla luciferase activity.

Preparation of nuclear extracts and EMSA. Nuclear extraction for EMSAs was done using standard protocols (36). The sequences of oligonucleotides used for the probes and cold competitors are listed in Table 2. Annealed oligonucleotides were labeled with [α-³²P] dCTP using the Klenow fragment of Escherichia coli DNA polymerase and were used as the probe. The reaction mixture contained 8 mmol/L Tris-HCl (pH, 7.9), 24 mmol/L HEPES-KCl (pH, 7.9), 120 mmol/L KCl, 24% glycerol, 2 mmol/L EDTA, 2 mmol/L DTT, 1.5 μg of poly(deoxyinosinic-deoxycytidyric acid) (poly(dIdC); Amersham Pharmacia Biotech, Inc.), and 10 μg of nuclear extract. After preincubation for 5 min, the indicated cold competitors or antibodies were added to the mixture, and the binding reaction was allowed to proceed at 4°C for 30 min. The reaction mixture was further incubated for 30 min in the presence of [³²P]-labeled probe DNA. Anti-C/EBPβ and anti-C/EBPα antibody were from Santa Cruz Biotechnology. After separating the binding complexes on a 6% polyacrylamide gel (containing 0.3× Tris-borate EDTA) at 4°C for 5 h, the gel was dried, and X-ray film was exposed for detection of complexes.

ChIP assay. ChIP assays were done using the ChIP assay kit (Upstate Biotechnology) according to the manufacturer's instructions. Cells were fixed in normal culture medium with formaldehyde at a final concentration of 1% for 10 min at 37°C. Sonication was done to achieve an average DNA length of 500 to 1,000 bp. The following antibodies were used for the immunoprecipitation of the cross-linked chromatin: anti-C/EBPβ COOH-terminal (C-19, Santa Cruz Biotechnology), anti-phospho-C/EBPβ (Thr²¹⁷)-R (Santa Cruz Biotechnology), and anti-C/EBPα (N-19; Santa Cruz Biotechnology). DNA was amplified by PCR using sequence-specific primers that amplify the region between nucleotides −265 and −123 (position −265/−123) within the elafin promoter containing C/EBPβ-5 site, position −331/−242 containing C/EBPβ-4 site, and position −376/−302 containing C/EBPβ-2 site. PCR was carried out as follows: 1 cycle at 95°C for 15 min; 35 cycles at 94°C for 30 s, 64°C for 30 s, 72°C for 1 min; and 1 cycle at 72°C for 10 min, using primers 5′-AGTGAGCCAGCACTTCTACTCTGTG-3′ and 5′-AATATCCACCCCATTCTGGGAGGGA-3′ for position −265/−123, 1 cycle at 95°C for 15 min; 36 cycles at 94°C for 30 s, 66°C for 30 s, 72°C for 1 min; and 1 cycle at 72°C for 10 min, using primers 5′-CCTTTCTCTTCTTGGGTCTC-3′ and 5′-ACAGAGTAGAAGTGCTGGCT-3′ for position −331/−242, 1 cycle at 95°C for 15 min; 35 cycles at 94°C for 30 s, 64°C for 30 s, 72°C for 1 min; and 1 cycle at 72°C for 10 min, using primers 5′-CAAGACTGGATCTACCAGTG-3′ and 5′-GAAATACAGTGAGACCCAAG-3′ for position −376/−302. As another option, anti-C/EBPβ COOH-terminal antibody was preincubated overnight with the respective blocking peptide in control reactions before immunoprecipitation. The amplified DNA fragments were separated on 3% agarose gel and visualized with ethidium bromide. The detected band was confirmed by sequencing.

Small interfering RNA transfection. A series of small interfering RNA (siRNA) was synthesized by Ambion. The sequences used for C/EBPβ siRNA and elafin siRNA were as following: C/EBPβ siRNA: sense, 5′-GAUGAAUGAUAAACUCUCUTT-3′; and antisense, 5′-AGAGAGUUUAUCAUUCAUCTG-3′. elafin siRNA-1: sense, 5′-GGAGUUCAAAGCCCAAUUUTT-3′; and antisense, 5′-AAAUUGGGCUUUGAACUCCTC-3′. elafin siRNA-2: sense, 5′-GGACAAGUUUCAGUUAAAGTT-3′; and antisense, 5′-CUUUAACUGAAACUUGUCCTT-3′. LacZ siRNA was used for control siRNA: sense, 5′-AGAUGAAACGCCGAGUUAATT-3′; and antisense, 5′-UUAACUCGGCGUUUCAUCUTT-3′. siRNA transfection was done according to the manufacturer's instructions. Briefly, 1 × 10⁵ of 76NE6 were plated in six-well dish. siRNA (100 pmol) and 8 μL of siPORT amine transfection agent (Ambion) were mixed with 100 μL of Opti-MEM (Life Technologies). After 10 min, siRNA and siPORT Amine Transfection Agent were mixed together and incubated for 10 min at room temperature. The mixture was then added to the cell culture. After overnight incubation, the culture medium was changed with fresh complete medium. The cells were harvested at 72 h (for elafin siRNA) and 96 h (for C/EBPβ siRNA) after transfection for Western blot analysis.

RT-PCR analysis. Total cellular RNA was prepared from 100-mm dishes using RNeasy Mini kit (Qiagen). A 1-μg portion of total RNA was reverse transcribed as described above. Following the first strand synthesis, amplification was carried out by 35 cycles of PCR at 94°C for 30 s, 65°C for 30 s, and 72°C for 1 min (for elafin) or by 35 cycles of PCR at 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min [for glyceraldehyde-3-phosphate dehydrogenase (GAPDH)] using the following primers: GAPDH: 5′-CCAGCCACCAGCCCCCTCACTAATA-3′ and 5′-AAGTGCCCCAGTGCCAAAGTTTTGC-3′; Elafin: 5′-TTGGTATGGCCTTAGCTCTT-3′ and 5′-CTGAATGGGAGGAAGAATGG-3′. PCR products were separated on agarose gels and visualized by ethidium bromide staining.

Expression profile of C/EBPβ in normal breast epithelial and breast tumor cells. The gene encoding human C/EBPβ gives rise to three protein products, including LAP1, LAP2, and LIP, with molecular weights of 55, 45, and 21 kDa, respectively (9, 10). We used an antibody, which recognizes an epitope at the COOH terminus of the C/EBPβ protein, and detected all three C/EBPβ isoforms in normal and tumor-derived mammary epithelial cells. The cell lines used for each of the C/EBPβ protein levels (i.e., both LAP and LIP) include a panel of normal mammary epithelial cells (mortal; 81N, 76N, immortal; 76NE6 Y54H, 76NE6 f2v 76NE6, 76NE7 and MCF-10A), and breast cancer cell lines (T47D, BT20T, ZR75T, MCF-7, MDA-MB-157, MDA-MB-231, MDA-MB-436, and HBL 100). Western blot analysis revealed that LAP1 is highly expressed in all of the normal cells as compared with that in most tumor cell lines (Fig. 1A). Particularly, 76N normal mortal cells and MCF-10A immortalized cells express LAP1 and not the other C/EBPβ isoforms. The other normal cell lines, including 81N, 76NE6 and its variant cells (76NE6 Y54H, 76NE6 f2v), and 76NE7, express all three C/EBPβ isoforms. In contrast, LIP is dominant in most tumor cell lines, except for T47D cells.

We then quantified the expression levels of LAP1, LAP2, and LIP and calculated the LAP/LIP ratio in each cell line. Data shown in the scatter diagram (Fig. 1B) represent the amount of LAP1 alone or that of total LAP (i.e., LAP1 + LAP2) expression as the ratio to LIP expression. Our analysis reveals that the LAP1/LIP ratio is much lower in most tumor cells (BT20T, ZR75T, MCF-7, MDA-MB-157, MDA-MB-231, and MDA-MB-436) than that of normal cells. The total amount of LAP expression as a ratio to LIP is also lower in most tumor cells as compared with normal cells, corresponding to the LAP1/LIP ratio.

The expression of C/EBPβ in breast tumor tissue is altered as compared with normal adjacent tissue. We collected tumor and normal adjacent breast tissue from 152 patients with stage I, II, or III breast cancer. The median age was 56.5 years (range, 26–84 years), and the median follow-up was 61.6 months (range, 3.8–80 months). The tumors were positive for ER in 110 (72.85%) patients, positive for PR in 100 (66.23%) patients, and positive for HER-2/neu in 28 (18.54%) patients. A total of 76 (50%) patients had positive lymph nodes on pathologic assessment. At the time of last follow-up, 17 (11.18%) patients had developed recurrent disease. Western blot analysis was done to assess LAP and LIP levels in each of the tumor and normal tissue samples. Figure 1C shows expression of C/EBPβ protein isoforms in 20 patients (representative cohort of the 152 patients) with breast cancer. Whereas LAP1 as well as LAP2 are highly expressed both in normal and tumor tissues in each patient, LIP is dominant specifically in tumor tissues from several patients (Fig. 1C). For instance, tumor tissue from patients 1, 2, 3, 5, 6, 10, 14, 15, and 18 have a high level of LIP as compared with normal tissue. Additionally, a high LIP/LAP ratio correlated strongly with negative ER status (P = 0.01) and positive HER-2/neu status (P = 0.054) in the primary tumor (Table 1).

The elafin gene is transcriptionally down-regulated through its promoter in most breast tumor cells. We next asked whether the elafin gene is a substrate of C/EBPβ, accounting for the potential involvement of elafin in tumor suppression. Computer analysis of the elafin promoter revealed the presence of several known transcription factor-binding sites including nuclear factor-κB (NF-κB), AP-1, C/EBPβ, and Sp1 (Fig. 2A). In particular, there are seven potential C/EBPβ binding sites, including transcription start site within 440 bp of upstream region in the elafin promoter (33), raising a possibility that C/EBPβ is targeting the elafin promoter as a major transcriptional regulator. To correlate the expression of elafin with that of C/EBPβ in normal and tumor cells, we examined the elafin mRNA levels in a panel of normal mammary epithelial cells, immortalized mammary epithelial cells, and breast cancer cell lines (Fig. 2B). Northern blot analysis reveals a difference in the expression pattern of the elafin mRNA in normal versus tumor cells, with higher levels of elafin in normal and immortalized epithelial cells and barely detectable levels in most tumor cell lines except for MDA-MB-157 and MDA-MB-436 (Fig. 2B). The expression pattern of the elafin protein is very parallel with that of its mRNA (data not shown). These data suggest that the elafin gene is transcriptionally silenced or inactivated in most tumor cells, and its expression profile is well correlated to that of C/EBPβ transcription factor (LAP/LIP ratio).

We next set out to determine the mechanism through which the expression of elafin is transcriptionally down-regulated in breast tumor cells. To this end, we initially ruled out the influence of genetic alteration on the elafin gene expression using Southern blot analysis and PCR sequencing of the entire elafin gene (data not shown). We also found that the elafin mRNA is stable in both normal cells (76NE6/MCF-10A) and tumor cells (MDA-MB-157/MDA-MB-436), suggesting that the differences in elafin mRNA stability among cells might not contribute to its down-regulation in some of the tumor cell lines (data not shown).

To investigate whether decreased transcriptional activity of the elafin promoter can account for lack of elafin expression in tumor cells, we compared the promoter activity of a wild-type elafin promoter-luciferase fusion reporter plasmid (pSPL1000) in immortalized breast epithelial cell lines (76NE6 and MCF-10A) with that in breast cancer cell lines (ZR75T, MCF-7, MDA-MB-231, and MDA-MB-436). The elafin promoter activity is 3- to 10-fold higher in immortalized breast epithelial cells than that of tumor cell lines (Fig. 2C). Additionally, PCR sequencing by amplifying the elafin promoter region revealed that there were no mutations in the elafin promoter even in tumor cells (data not shown). These results suggest that the elafin promoter activity may fully account for the differences in elafin mRNA expression in normal versus tumor cells.

Deletion mapping of the human elafin promoter. A series of 5′ deletions in the promoter region was generated to map the elafin promoter. The resulting plasmids were transiently transfected into 76NE6 cells, and the luciferase activities were measured. As shown in Fig. 3A, stepwise deletion from −290 to −172 resulted in a significant decrease in elafin promoter activity to 36%. Further deletion of pSPL94 decreased in the promoter activity down to 12% (Fig. 3A). The elafin promoter activity in MCF-10A cells also decreased stepwise from −290 to −94 (data not shown). Taken together, these results suggest that the region between −290 and −94 in the elafin promoter might include the transactivating elements.

Mutational analysis of the human elafin promoter. Of the seven potential C/EBPβ binding sites within 440 bp of upstream region in the elafin promoter, the 290-bp region harbors three C/EBPβ binding sites, C/EBPβ-4, C/EBPβ-5, and C/EBPβ-6 (Fig. 2A). However, there still remains a possibility that C/EBPβ binding sites between −440 and −290, C/EBPβ-1, C/EBPβ-2, and C/EBPβ-3 (Fig. 2A), are also involved in the promoter activity in normal cells. Hence, we mutated each of these binding sites using the deleted construct, pSPL440, and determined which C/EBPβ binding sites are responsible for the transactivation of the elafin promoter (Fig. 3B).

The single mutation at C/EBPβ-4 (pSPL440m4) and that at C/EBPβ-5 (pSPL440m5) significantly reduced the promoter activity to 43% and 37% from that seen in wild-type pSPL440, respectively (Fig. 3B). Furthermore, the double mutation at both C/EBPβ-4 and C/EBPβ-5 (pSPL440m4m5) had more suppressive effect on the promoter activity, decreasing it to 21%. These results suggest that both C/EBPβ-4 and C/EBPβ-5 sites could be specifically involved in the transactivation of the elafin promoter. To further clarify the contribution of C/EBPβ-1, C/EBPβ-2, and C/EBPβ-3 sites between −440 and −290 to the transactivation, we generated a triple mutation into pSPL440 by mutating C/EBPβ-1, C/EBPβ-2, and C/EBPβ-3 sites. The mutations at these C/EBPβ sites also significantly reduced the promoter activity to 20% as compared with pSPL440, and only single mutation at C/EBPβ-2 (pSPL440m2) significantly reduced the promoter activity to 56% (Fig. 3B). However, no suppressive effect on the elafin promoter activity was seen in the mutation at AP-1 site (pSPL440mAP-1; Fig. 3C), suggesting the specific role of C/EBPβ sites in the elafin promoter activity.

To rule out the involvement of p53 in the transcriptional regulatory mechanism of the elafin promoter, we also used p53 wild-type immortalized mammary epithelial cells, MCF-10A and 76NY54H cells, in the mutational analysis of the elafin promoter. The results observed in these p53 wild-type cell lines were very similar to those seen in p53 deleted cells (76NE6) as mentioned above (Fig. 3D and E).

Taken together, the results from functional analysis of the elafin promoter suggest that both C/EBPβ-4 and C/EBPβ-5 sites may be necessary for the transactivation of the elafin promoter, and C/EBPβ-2 sites could cooperate with C/EBPβ-4 and C/EBPβ-5 sites. The mutational analysis using p53 wild-type cell lines also suggest that the transcriptional regulation of the elafin promoter is independent of p53 phenotype.

C/EBPβ interacts with the specific C/EBPβ binding sites in the elafin promoter. To address if C/EBPβ can directly bind to specific regions on the elafin promoter, we took two different approaches. We initially did EMSA using specific sequences for C/EBPβ-2, C/EBPβ-4, and C/EBPβ-5 binding sites from the elafin promoter (sequences are listed in Table 2) and nuclear extracts from either 76NE6 normal cells (Fig. 4A,–C, lanes 2–8) or MDA-MB231 tumor cells (Fig. 4A,–C, lanes 9–15). As shown in Fig. 4A, a similar oligonucleotide-protein binding pattern was observed between 76NE6 and MDA-MB-231 cells with wild-type C/EBPβ-5 probes (lanes 2 and 9). The specificity of the generated signal (band 1) was determined using competitive EMSA. The results revealed that band 1 was completely competed out by excess unlabeled wild-type C/EBPβ-5 (lanes 3 and 4, lanes 10 and 11), but not by mutant C/EBPβ-5 (lanes 5 and 6, lanes 12 and 13), suggesting that band 1 is specific. Additionally, supershift experiments in the presence of anti-C/EBPβ antibody showed that band 1 was slightly diminished and supershifted in 76NE6 cells (lane 7). However, the supershifted band was not so remarkable in MDA-MB-231 cells as that seen in 76NE6 cells (lane 14).

The EMSA was also done with sequences specific for C/EBPβ-2 and C/EBPβ-4, and the results revealed that the same specific band 1, which was also supershifted by anti-C/EBPβ antibody, was shifted (Fig. 4B and C). Band 1 was the only one shifted when the probe was incubated with the nuclear extract from 76NE6 cells (lane 2). However, this band was strongly competed out by both wild-type (lanes 3 and 4) and mutant cold competitors (lanes 5 and 6). In contrast to 76NE6 cells, two bands were seen when the C/EBPβ-2 or C/EBPβ-4 probe was incubated with the nuclear extract from MDA-MB-231 cells (Fig. 4B  and C, lane 9). Competitive EMSA revealed that the faster mobility band (band 3) was more specifically competed out than the slower one (band 2), suggesting that band 3 may be more specific. Band 3 was still supershifted in the presence of anti-C/EBPβ antibody (bottom, lane 14). In summary, these results suggest that C/EBPβ may preferentially interact with the C/EBPβ-5 binding site in the elafin promoter of 76NE6 cells. Whereas C/EBPβ is bound to the C/EBPβ-5 site in 76NE6 cells, other isoforms of C/EBPβ may be associated with the C/EBPβ-2 and C/EBPβ-4 sites in MDA-MB-231 cells.

In the second approach to examine if C/EBPβ is associated with the elafin proximal promoter region in vivo, ChIP assays were done with antibodies against C/EBPβ COOH-terminal, phospho-C/EBPβ (Thr²¹⁷) and anti-C/EBPα (Fig. 4D,, top; E, top; and F, top and middle). Initially, the precipitated DNA from 76NE6 cells was subjected to PCR using specific primers that amplify sequences spanning the C/EBPβ-5 (Fig. 4D), C/EBPβ-2 (Fig. 4E), or C/EBPβ-4 site (Fig. 4F) of the elafin promoter region. Consistent with our EMSA results, C/EBPβ was specifically associated with the endogenous elafin promoter containing the C/EBPβ-5 site in 76NE6 cells, whereas phospho-C/EBPβ and C/EBPα were not (Fig. 4D,, top). Addition of blocking peptide against C/EBPβ antibody resulted in a reduced PCR signal, suggesting the specificity of the ChIP signal in 76NE6 cells (Fig. 4D,, bottom, lane 2 versus lane 3). However, the PCR signal was not significantly reduced even in the presence of the blocking peptide in MDA-MB-231 cells, suggesting that the PCR signal obtained by immunoprecipitation with C/EBPβ antibody is nonspecific, and that C/EBPβ is not specifically associated with the elafin promoter region containing the C/EBPβ-5 site in vivo in MDA-MB-231 cells (Fig. 4D,, bottom, lane 5 versus lane 6). In contrast, the association of C/EBPβ with C/EBPβ-2 or C/EBPβ-4 site was observed in MDA-MB-231 cells, but not in 76NE6 cells (Fig. 4E and F). The results from the ChIP assays are consistent with the EMSA results, suggesting that the C/EBPβ-5 site in the elafin promoter plays a central role in transcriptional activation of the elafin promoter in 76NE6 cells.

Overexpression of C/EBPβ transactivates the elafin gene through its promoter in MDA-MB-231 tumor cells. Next, we asked if C/EBPβ could directly activate the transcription of elafin gene in breast cancer MDA-MB-231 cells. Either wild-type elafin promoter-luciferase fusion reporter plasmid or a deletion plasmid lacking all of the C/EBPβ binding sites was cotransfected with an C/EBPβ expression plasmid. The luciferase activity of the full-length pSPL1000 construct was activated with increasing amounts of C/EBPβ by up to 4.1-fold compared with 1.9-fold with the deletion plasmid pSPL94, suggesting that C/EBPβ can transactivate the elafin promoter (Fig. 5A).

Because DNA binding analyses of the elafin promoter suggest that the interaction of C/EBPβ with C/EBPβ-2, C/EBPβ-4 and C/EBPβ-5 sites could be differentially regulated in tumor cells, we investigated the role of these C/EBPβ binding sites in the elafin promoter activation by C/EBPβ overexpression (Fig. 5B). A series of wild-type or mutant reporter plasmids were transiently cotransfected with empty plasmid or C/EBPβ expression plasmid into MDA-MB-231 cells, and their luciferase activities were analyzed. Whereas the luciferase activity from wild-type pSPL440 was activated 3.6-fold by C/EBPβ, fold activation by C/EBPβ overexpression using a single mutation at C/EBPβ-2 or C/EBPβ-4 decreased to 2.0-fold. The single mutation at C/EBPβ-5 had little response to C/EBPβ overexpression. Furthermore, the double mutation at both C/EBPβ-4 and C/EBPβ-5 had more suppressive effect on the fold activation by C/EBPβ, decreased to 1.1-fold (Fig. 5B). These results indicate that each of these C/EBPβ binding sites is necessary for the activation of the elafin promoter in response to C/EBPβ.

Next, we determined whether overexpression of C/EBPβ enhances elafin mRNA expression in MDA-MB-231 tumor cells. To this end, the expression of C/EBPβ isoforms in MDA-MB-231 tumor cells transiently transfected with C/EBPβ-expressing plasmid was initially characterized by Western blot analysis. Because the C/EBPβ-expressing plasmid used contains three conservative ATG translation start sites, three isoforms, including LAP1, LAP2, and LIP, are encoded by alternative usage of different start codons. The results indicate that an activating isoform of C/EBPβ, LAP1, was predominantly overexpressed at 24 h following transfection with C/EBPβ-expressing plasmid, and all isoforms of C/EBPβ were up-regulated at 48 h after transfection (Fig. 5C,, top). However, the increase in LAP was greater than that in LIP, resulting in an increasing LAP1/LIP ratio (from 0.5 in control cells to 1.5 in cells transfected with C/EBPβ-expressing plasmid). The total amount of LAP expression (LAP1 + LAP2) as a ratio to LIP was also higher in cells transfected with C/EBPβ-expressing plasmid as compared with control cells (Fig. 5C , bottom).

Northern blot analysis after transfection showed that overexpression of C/EBPβ in MDA-MB-231 cells stimulates the expression of the elafin mRNA compared with those transfected with empty plasmid 48 h after transfection (Fig. 5D,, top). Similar results were confirmed by RT-PCR (Fig. 5D , bottom). Thus, the altered expression ratio between these C/EBPβ isoforms was associated with alterations in the elafin mRNA, suggesting a role of LAP/LIP ratio in mediating the transactivation of elafin gene.

Taken together, we conclude that overexpression of C/EBPβ transactivates the elafin gene through C/EBPβ-2, C/EBPβ-4, and C/EBPβ-5 sites in its promoter in MDA-MB-231 tumor cells.

Cross-talk of C/EBPβ and elafin in 76NE6 breast epithelial cells. We investigated whether direct down-regulation of C/EBPβ using siRNA would have an impact on the elafin expression in normal cells. To this end, we transfected 76NE6 breast epithelial cells with siRNA to C/EBPβ. We noted a marked decrease in the elafin expression following treatment with the C/EBPβ-specific siRNA sequence as determined by Western blot analysis, whereas the control siRNA (LacZ siRNA) did not affect elafin expression (Fig. 5E,, top). To further investigate the regulatory circuit between C/EBPβ and elafin, 76NE6 cells were transfected with two kinds of elafin siRNA, and the elafin and C/EBPβ levels were examined. Transfection of both of elafin siRNA resulted in complete suppression of elafin levels and remarkable up-regulation of LAP1 levels in 76NE6 cells as compared with LacZ siRNA controls (Fig. 5E , bottom). These data strongly suggest that the expression of elafin gene is mediated by C/EBPβ, and the circuit between C/EBPβ and elafin is working both ways in the regulation of elafin gene expression in normal breast epithelial cells.

C/EBPβ regulates mammary gland development, proliferation, and differentiation through specific gene expressions (6, 7). C/EBPβ is an intronless gene whose transcript encodes three in-frame translational products in human breast epithelial cells (LAP1, LAP2, and LIP). Several groups have reported on the differences in expression of LAP/LIP between normal and breast cancer cells. For instance, LAP1 is present exclusively in normal mammary epithelial cells, whereas LIP is often characterized as a biological predictor of the aggressiveness in breast cancer, and its overexpression causes epithelial proliferation and the formation of hyperplasia in cultured mammary epithelial cells (5). These suggest that a LIP-initiated growth cascade may be susceptible to additional oncogenic stress, which could result in the initiation and progression of neoplasia.

We initially characterized normal versus tumor tissue samples from breast cancer patients as well as cell lines for the expression of C/EBPβ proteins and assessed the ratio of LAP/LIP. We found a distinct expression pattern of C/EBPβ protein in normal versus tumor cells, and the LAP/LIP ratio in tumor cells tends to be lower than that in normal cells. Our next question is which downstream genes accounting for the pathogenesis of breast cancer are targeted by C/EBPβ. Normal epithelial cells have an anti-inflammatory mechanism through which cells protect themselves from the action of proteases by producing cytokines. For instance, the attack of neutrophil elastase induces a rapid increase in the levels of C/EBPβ, leading to the enhancement of interleukin-6 (IL-6) mRNA with its anti-inflammatory and anti-protease effect (37). C/EBPβ has been previously shown to activate transcription of IL-6 gene (38, 39). These studies suggested that C/EBPβ could mediate an anti-protease effect by directly targeting protease inhibitors, such as elafin. In fact, there are a number of potential binding sites for the C/EBPβ in the elafin promoter. Furthermore, the elafin gene is transcriptionally down-regulated in most breast tumor cell lines in which LAP/LIP ratio is low. These findings raised a possibility that protease inhibitor elafin is a direct substrate of C/EBPβ. In this study, we examined how the expression of the elafin gene is regulated in breast epithelial cells, and in doing so, have established a novel pathway linking C/EBPβ and elafin. We focused on the transcriptional regulation of the elafin gene and found that inactivation of the elafin promoter is responsible for the down-regulation of the elafin gene in tumor cells.

Our functional analysis of the elafin promoter using a series of C/EBPβ mutants suggest that several C/EBPβ binding sites could be involved in the transactivation of the elafin promoter in normal breast epithelial cells. In particular, the results from our EMSA suggest that C/EBPβ is specifically associated with the C/EBPβ-5 site in 76NE6 normal cells in vitro. Although the mutation at the C/EBPβ-2 and C/EBPβ-4 sites decreased the elafin promoter activity in 76NE6 cells, the competitive EMSA indicates that the affinity of C/EBPβ to these mutated C/EBPβ-2 and C/EBPβ-4 sites are not significantly altered. Additionally, ChIP assays confirmed that C/EBPβ interact with the C/EBPβ-5 site, but not with C/EBPβ-4 or C/EBPβ-2 site, in the endogenous elafin promoter region in 76NE6 cells. Although the roles of the C/EBPβ-4 or C/EBPβ-2 site in normal cells are still unclear, these results suggest that the C/EBPβ-5 site is the most important for the transcriptional activation of the elafin gene in normal cells.

In contrast, our ChIP assay reveals that the C/EBPβ-5 site in MDA-MB-231 cells is not significantly associated with C/EBPβ in vivo. Furthermore, we found remarkable differences in the bindings of transcription factor(s) to the C/EBPβ-2 and C/EBPβ-4 sites between 76NE6 cells and MDA-MB-231 cells. As shown in our EMSA, another oligonucleotides-protein complex at the C/EBPβ-2 and C/EBPβ-4 sites was detected in MDA-MB-231 tumor cells, but not in 76NE6 normal cells. Using ChIP analysis, we also confirmed that the transcription factor(s) immunoprecipitated by C/EBPβ antibody are present in the elafin proximal promoter region containing the C/EBPβ-2 and C/EBPβ-4 sites in MDA-MB-231 cells in vivo. These findings raise a possibility that the C/EBPβ-2 and C/EBPβ-4 sites may be occupied with other C/EBPβ isoform(s), such as LIP, leading to a decrease in LAP/LIP ratio in these sites and partially contributing to transcriptional suppression of elafin gene in MDA-MB-231 tumor cells. However, because the C/EBPβ COOH-terminal antibody used in our EMSA and ChIP assay recognizes all of the isoforms and the C/EBPβ NH₂-terminal antibody specific for LAP1 is not commercially available, the identity of this faster mobility band still remains unknown.

Recently, several reports have shown that some transcriptional co-repressors, including ETO/MTG8 and SMRT, are interacting with C/EBPβ and acts as an inhibitor of C/EBPβ (40, 41). These data raise the possibility that transcriptional co-repressors may still be present near the C/EBPβ binding sites to suppress elafin transactivation. We also showed that C/EBPβ mutant elafin promoters do not fully respond to C/EBPβ overexpression plasmid. This finding suggests that all of these binding sites are functional in transcriptional activation of the elafin gene, supporting a model in which the C/EBPβ-2, C/EBPβ-4, and especially C/EBPβ-5 sites synergistically act in switching on the transactivation of the elafin gene in response to C/EBPβ in MDA-MB-231 tumor cells (Fig. 6).

We have two lines of evidence that shows the elafin and C/EBPβ are involved in a cross-talk in normal breast epithelial cells. First, the depletion of C/EBPβ by siRNA resulted in suppression of the elafin expression in 76NE6 cells. Second, the down-regulation of elafin led to the up-regulation of LAP1 level, suggesting that LAP1 could be induced to compensate for elafin loss. These results support a model in which the endogenous elafin gene is tightly controlled under the negative feedback loop between elafin and an activating form of C/EBPβ in normal breast epithelial cells (Fig. 6). Furthermore, overexpression of C/EBPβ in tumor cells establishes a correlation between the increase in LAP/LIP ratio and elafin re-expression, which may also support our hypothesis that the differences in the LAP/LIP ratio between normal and tumor cells have impact on elafin transactivation.

Maspin has already been established as a tumor-suppressing serpin family of protease inhibitors initially isolated from normal human mammary epithelial cells (42). Functional studies have shown that maspin inhibits tumor invasion and motility of human mammary tumor cells in cell culture (43), as well as tumor growth and metastasis in the nude mice assay (42). Similarly to the elafin, the expression of maspin is specific in normal mammary epithelial cells and is controlled at the transcriptional level. However, the maspin promoter is regulated by the combination of elements including Ets and AP-1 binding sites, but not by C/EBPβ (44), suggesting that the regulation by C/EBPβ is unique for the elafin promoter.

The expression of elafin has been investigated in squamous cell carcinoma of the human oral tissues, esophagus, and lung, showing that increased elafin expression was observed in well-differentiated squamous cell carcinoma (45–47). However, most of the poorly differentiated tumor cells showed the decrease in elafin expression and high elastase activity (48). These reports suggest that elafin expression is possibly involved in the cell differentiation program of human squamous cell carcinoma, and progressive loss of elafin expression could facilitate tumor invasion through the elastase activity.

We suggest that inhibition of elafin expression plays a key role in breast cancer progression and metastasis. We hypothesize that elafin may be a novel tumor suppressor gene, and exploration of mechanisms for the differential regulation of elafin in normal versus tumor cells may be of great significance in the determination of the malignant phenotype in breast cancer, as well as for improved prognostication and treatment of breast cancer.

Grant support: CA87458 from the NIH (to K. Keyomarsi), P50CA116199 from NCI (to G. Hortobagyi; project to K. Keyomarsi), by the Odyssey Program and the Kimberly Clark Foundation Endowment for New and Innovative Research at the University of Texas M. D. Anderson Cancer Center, by YASUDA Medical Research Foundation in Japan (to T. Yokota), and by the Susan G. Komen Foundation (to K.K. Hunt).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We gratefully acknowledge Dr. Linda Sealy (Vanderbilt University School of Medicine), Dr. Krishna P. Bhat (The University of Texas M.D. Anderson Cancer Center), and Dr. Peter F. Johnson (National Cancer Institute Frederick) for technical advice and helpful discussions.