A New cis Element Is Involved in the HER2 Gene Overexpression in Human Breast Cancer Cells

Introduction

The HER2/c-erbB2 gene encodes a 185-kDa transmembrane tyrosine kinase receptor belonging to the EGFR 3 family (1) . No direct ligand of HER2 has been identified so far, but recent studies suggest that HER2 is a pan-HER subunit of the high affinity heterodimeric receptors for growth factors of the EGF superfamily (2 , 3) . HER2 is expressed at low levels in a variety of healthy human adult epithelial cells (4) . Overexpression of the p185^HER2 protein in at least 30% of primary human mammary adenocarcinomas (5 , 6) is correlated with poor prognosis (7) and indicates a poor response to chemotherapy (8) and endocrine therapy (9) .

We are interested in characterizing the molecular modifications responsible for HER2 gene overexpression in breast adenocarcinoma cells. Our laboratory (10) and others (11 , 12) have shown that the accumulation of abnormal levels of HER2 mRNA is not the consequence of the stabilization of the messenger, but results from transcriptional deregulation. The transcription factors AP-2 (11 , 13 , 14) , PEA3 (15 , 16) , and RBP-Jκ (17 , 18) contribute to HER2 gene overexpression in human breast tissues. They all act through recognition sequences located in the 250-bp-long proximal HER2 promoter.

In a previous study, we have identified multiple positive and negative cis-acting elements along a 6-kb fragment of the normal human HER2 promoter. Differences between the populations of trans-acting factors of several mammary epithelial cell lines, expressing different levels of HER2 mRNA, were detected. Interestingly, a 219-bp region of the HER2 promoter greatly stimulated the LUC reporter gene expression in BT-474 mammary cells. This suggested the presence of a particular trans-activator in these cells (19) .

In this study, we describe the identification of a 17-bp-long cis sequence located in this 219-bp-activating fragment, which is recognized by a transcription factor that we named HTF. HTF has an estimated molecular weight of M_r 50,000 and is abundant in some mammary tumor cell lines overexpressing the HER2 mRNA. The recognition sequence for HTF is present, with one mismatch, in the EGFR promoter. The binding capacity of HTF to these related cis sequences has been compared.

Materials and Methods

Cell Lines

The HBL-100, MCF-7, T-47D, MDA-MB-453, BT-474, SK-BR-3, and ZR-75-1 human mammary epithelial cells were purchased from American Type Culture Collection (Manassas, VA) and cultured in the recommended media supplemented with 10% fetal bovine serum, 2 mM glutamine, and 100 μg/ml penicillin/streptomycin.

Plasmids

pAαs-KS⁻.

The 219-bp HER2 PvuII-SmaI-activating fragment (−716 to −497, relative to the CAP site) was isolated from the p2069-LUC plasmid (19) , blunted using the Klenow fragment of DNA polymerase (Boehringer Mannheim, Mannheim, Germany), and inserted, in antisense orientation, in the SmaI site of the pBluescriptII KS⁻ (Stratagene, La Jolla, CA).

p-C4→537, p←C4-537, and p←3C4-537.

The C4 oligonucleotide (−523 to −479, relative to the CAP site) containing BamHI and SmaI additional restriction sites at its ends (Life Technologies, Inc., Bethesda, MD) was inserted between the corresponding sites in the p537-LUC plasmid (19) . The resulting p-C4→537 and p←C4-537 plasmids contain one copy of the C4 fragment, respectively, in sense and antisense orientation 5′ of the 537-bp HER2 promoter. The p←3C4-537 plasmid contains three copies of the C4 fragment in the antisense orientation 5′ of the 537-bp HER2 promoter.

EMSA

Nuclear proteins were extracted from subconfluent cells, according to Dignam (20) . The following regions overlapping the 219-bp-long HER2-activating fragment were used as probes: a 251-bp XE fragment from the pAαs-KS⁻ containing the 219-bp-long HER2 fragment flanked by 32 bp of the polylinker, and the PCR-generated fragments C1 (−734 to −479), C2 (−558 to −479), and C4 (−523 to −479).

The mutated C4 oligonucleotides are: C4(G): 5′-TTCAAAGATTCCAGAAGATATGCGCCGGGGGTCCTGGAAGCCAC-3′, C4(A): 5′-TTCAAAGATTCCAGAAGATATGCACCGGGGGTCCTGGAAGCCAC-3′, and C4(T): 5′-TTCAAAGATTCCAGAAGATATGCTCCGGGGGTCCTGGAAGCCAC-3′. The sequence of the EGFR oligonucleotide corresponding to a fragment (−226 to −271, relative to the CAP site) of the EGFR promoter (21) is: 5′-CGGCCGCGCTGCGCCGGGGGCTGCCCGGACGTCTAGCTCGCGCGGG-3′. The mutated EGFR oligonucleotide is: 5′-CGGCCGCGCTGCCCCGGGGGCTGCCCGGACGTCTAGCTCGCGCGGG-3′.

These DNAs were end-labeled using the Klenow fragment of DNA polymerase or the T4 polynucleotide kinase (Boehringer Mannheim) in the presence of [α-³²P]dCTP or [γ-³²P]ATP (ICN, Costa Mesa, CA), respectively. The radiolabeled fragments (2 × 10⁵ CPM) were incubated with 10 μg of nuclear proteins, 1 μg of poly(dG-dC)-poly(dG-dC) (Pharmacia, Uppsala, Sweden), and, in some cases, a 50–200 molar excess of cold DNA competitor in 20 μl of “high salt” binding buffer [10 mM Tris-HCl (pH 7.5), 200 mM NaCl, 2.5 mM DTT, 10% glycerol, 0.5% NP40, 2.5 mM EDTA, and 5 mM MgCl₂]. The reaction mixtures were incubated for 20 min at room temperature and resolved through a 5% polyacrylamide gel (acrylamide:bisacrylamide, 30:0.8, wt/wt) in 1X TEA buffer [7 mM Tris-HCl (pH 7.5), 3 mM sodium acetate, and 1 mM EDTA]. The gel was dried and analyzed by autoradiography. The signals were quantitated by PhosphorImager (Molecular Dynamics). The NSC was Msp1-digested pGEM3Z DNA.

Transient Transfection Assays

BT-474 cells (7 × 10⁵) were plated on 35-mm tissue culture dishes and grown until 60% confluent. The cells were rinsed with serum containing medium and incubated for 3 h in 1 ml of complete medium containing a 2-μg DNA/4 μl PEI (ExGen500; Euromedex, Strasbourg, France) mixture. The transfection medium was replaced by complete medium, and the cells were further incubated for 45 h. Cells were harvested and lysed, and the LUC enzymatic activities were measured using the Luciferase Reporter Gene Assay kit (Boehringer Mannheim). The LUC activities, measured using a luminometer (Berthold 9501; Wildbad, Germany), were expressed in relative light units. The protein contents of the cell extracts were measured using the Micro BCA Protein Assay Reagent (Pierce Chemical Co., Rockford, IL). Each reporter vector was transfected in triplicate, and the transfection experiments were repeated twice.

Results

HER2 Promoter Binding Activity of HTF.

A 251-bp XE fragment containing the previously described (19) 219-bp HER2-activating sequence and additional sequences from the polylinker (Fig. 1A ⇓ ; see “Materials and Methods”) was incubated with crude nuclear extracts from BT-474 cells. In a first series of experiments, three PCR-generated fragments (C1, C2, and C3) were used as competitors (Fig. 1A) ⇓ . The retarded complex was displaced by the competitors XE, C1, and C2, but not by C3 (Fig. 1B) ⇓ . Next, EMSA was performed using the C2 fragment as a probe and fragments C4 and C5 as competitors (Fig. 1A) ⇓ . The C2 and C4 fragments displaced the DNA/protein complex, whereas C5 and the NSC did not (Fig. 1C) ⇓ . The C3 fragment did not give rise to a retarded complex in EMSA (data not shown). EMSA using the C4 fragment as a probe confirmed the presence of the activating sequence in the 3′ end of the initial activating fragment (Fig. 1D) ⇓ . To further confirm the presence of the activating sequence in the C4 fragment, EMSA was performed using the C1 fragment as a probe and C4 and C5 fragments as competitors. One protein/DNA complex was observed, which was abolished by the C1 and the C4 fragments and not by the C5 fragment (Fig. 1E) ⇓ . Taken together, these results suggest that the cis-activating sequence is located in the 3′ half of the 44-bp-long C4 fragment. We named HTF the BT-474 protein recognizing this sequence. Chemical footprinting experiments were done to identify the precise cis sequence, in the C4 fragment, recognized by HTF (data not shown). A 17-bp-long sequence located between the −506 and −489 positions, relative to the transcription start site, was protected (Fig. 1F) ⇓ . The presence of known cis sites in this sequence was determined by the use of the TFD sites database. The protected sequence overlaps closely related consensus core cis sequences for the transcription factors AP-2 and NF-κB (Fig. 1F) ⇓ . To find out whether HTF binds to the AP-2-related cis sequence, we mutated the C* (Fig. 1F) ⇓ , which is well conserved among the AP-2 recognition sequences (22) . The binding of HTF to the different mutants was assessed by EMSA. The replacement of the C by an A abolished the binding, whereas the replacement by a G or a T did not significantly modify the binding (Fig. 2A) ⇓ . The molecular weight of HTF was first estimated to be M_r 120,000 by Southwestern assays. However, control experiments revealed that the 120-kDa band was, in fact, the PARP protein (23) . We then performed UV cross-linking with BT-474 nuclear extracts depleted of PARP by immunoprecipitation. One major UV cross-linked product was detected with an apparent molecular weight of M_r 85,000 (data not shown). Once this size was adjusted for the approximately 35-kDa contribution of the DNA, the molecular weight of HTF was estimated at M_r 50,000.

Analysis of the Core Binding Sequence of HTF.

The HTF core binding site, GCCCCGGGG, containing one mismatch (second C replaced by a G), is present between positions −236 and −244 in the EGFR promoter (22) . We compared, by EMSA, the binding of HTF to the C4 sequence and the EGFR oligonucleotide described by Kageyama et al. (21) . The retarded bands had a similar electrophoretic mobility, but the intensity of the “C4 complex” was higher than that of the “EGFR complex” (Fig. 2B ⇓ , Lanes 1 and 8). Interestingly, the C4 complex was only weakly displaced by a 50-fold excess of the cold EGFR oligonucleotide. In contrast, the EGFR complex completely disappeared with a 50-fold excess of the C4 competitor (Fig. 2B) ⇓ . Thus, HTF seems to bind to the EGFR cis sequence with a lower affinity than to the C4 cis sequence.

We, thus, compared the binding affinity of HTF to C4 and EGFR oligonucleotides. First, EMSA was performed using a fixed amount of BT-474 nuclear proteins and increasing concentrations of probes of identical specific activities. Fig. 2C ⇓ shows that HTF efficiently bound C4 and EGFR oligonucleotides at a concentration of 1 pmol μl⁻¹ and that a weak binding could already be observed at a concentration of 0.2 pmol μl⁻¹. The signals were quantitated, and the binding ratio of C4:EGFR was calculated for each probe concentration. The results indicated that the binding of HTF to the C4 oligonucleotide was 3-fold higher than to the EGFR oligonucleotide. The affinity of a protein for its ligand is a function of the rates of association and dissociation. We compared these rates for HTF binding to the C4 and EGFR oligonucleotides. The association rate was measured by incubating the BT-474 nuclear proteins with the oligonucleotides for 1–16 min and detection of the bound fraction by EMSA. The association rate was similar and extremely rapid for both oligonucleotides and, therefore, difficult to measure precisely (data not shown). To measure the dissociation rate, the BT-474 proteins were first incubated for 20 min with the labeled C4 or EGFR oligonucleotide. A 1000-fold excess of the respective unlabeled oligonucleotide was added, and the mixture was incubated for 0–16 min. The samples were analyzed by gel electrophoresis (Fig. 2D) ⇓ . The decrease in the intensity of the shifted band reflected the dissociation rate of the protein/DNA complex. The EGFR complex was less stable since the cold specific DNA displaced the probe after 1 min of incubation. In contrast, the C4 probe formed a more stable complex with HTF because the cold specific DNA displaced the probe after 6 min of incubation. Therefore, the greater dissociation rate of the EGFR complex than that of the C4 complex suggests that HTF has a higher affinity for C4.

We investigated the contribution of the C/G base difference between the two core binding sites of C4 and EGFR to the difference in the binding affinity seen above. The wild-type C4 and EGFR oligonucleotides were used as probes in EMSA. The wild-type and mutated (m) oligonucleotides were used as competitors. The mC4 and mEGFR oligonucleotides were mutated at the divergent position (C to G and G to C mutations in the C4 and EGFR fragment, respectively). When the C4 oligonucleotide was used as a probe, the C4 and the mEGFR oligonucleotides completely displaced HTF from the complex. Moreover, the C4 oligonucleotide was a better competitor than the mC4. Similarly, the mEGFR oligonucleotide was a better competitor than the EGFR. When the EGFR was used as a probe, the retarded complex was completely displaced by the C4 and mEGFR competitors, as well as by the mC4 and EGFR oligonucleotides (Fig. 2E) ⇓ . Taken together, these results show the importance of the second C, in the HTF cis sequence, for the binding affinity of HTF.

In Vivo Stimulatory Effect of the C4 Fragment.

We confirmed the stimulatory effect of the C4 fragment, containing the HTF recognition sequence, on the HER2 basal promoter. The C4 fragment inserted in both orientations induced a 5–8 fold increase in the transcriptional activity of the 537-bp HER2 promoter in the BT-474 cells (Fig. 3A) ⇓ . The stimulatory effect of C4 was equivalent to that obtained with the p756-LUC plasmid containing the whole 219-bp-activating sequence. In contrast, the C4 fragment induced a slight, but statistically not significant, modification of the transcription from the heterologous tk promoter (data not shown). The C to A mutation, which abolished the binding of HTF to C4 (Fig. 2A) ⇓ , introduced in p-C4M→537 and p756M-LUC reporter vectors, decreased the transcriptional activation induced by C4 (Fig. 3A) ⇓ . These results demonstrate that when the binding of HTF to the C4 fragment is diminished, the activation level through the C4 fragment is inhibited.

Finally, we measured HTF binding activity in seven mammary epithelial cell lines expressing different levels of the HER2 transcript to determine whether there was a parallelism between HTF binding activity and HER2 mRNA expression level. Equal amounts of C4 probe and nuclear proteins from each cell line were tested in EMSA (Fig. 3B) ⇓ . A major retarded complex, with similar electrophoretic mobility, was detected in all the mammary cells. Densitometric scanning of the retarded bands was performed. The cells could be divided in two groups according to the HTF binding activity. The highest HTF activity was observed in nuclear extracts from BT-474, ZR-75-1, and SK-BR-3 cells overexpressing the HER2 mRNA. The second group including MCF-7, T-47D, MDA-MB-453, and HBL-100 cells contained low levels of HTF activity. Interestingly, MDA-MB-453 cells showed a low level of HTF binding activity, whereas they overexpress the HER2 transcript. In conclusion, the presence of HTF in different mammary tumor cell lines and its binding activity is increased in the majority of the cells, overexpressing the HER2 mRNA, tested.

Discussion

BT-474 breast cancer cells overexpress the HER2 transcript 80 times when compared with healthy breast cells. In this study, we describe the identification of a 44-bp DNA fragment of the human HER2 promoter necessary for the increased transcription of the HER2 gene in BT-474 cells. HTF, interacting with this fragment, has an approximate molecular weight of M_r 50,000. It binds to a 17-bp-long cis sequence overlapping related cis sequences for AP-2 and NF-κB transcription factors. The nucleotides, included in the core binding site for HTF, essential for the binding and activity of HTF have been determined. HTF also binds to a cis sequence of the EGFR promoter, but with a lower affinity than to the C4 cis sequence.

The C4 fragment is responsible for the high transcriptional activity of the previously observed 219-bp fragment of the HER2 promoter (19) . Indeed, the stimulatory effect of this fragment was similar to that of the original activating region of the p756-LUC plasmid. The C4 fragment stimulated the transcription in an orientation-independent manner.

The 17-bp-long cis sequence recognized by HTF overlaps related consensus cis sequences for the transcription factors AP-2 and NF-κB. Supershift experiments and competition of the HTF/C4 complex with an oligonucleotide recognized by the NF-κB complex excluded the implication of this factor in our system. We are currently testing the possibility that HTF could be an AP-2-related protein.

The HTF core recognition sequence is present, with one mismatch, in the EGFR promoter (21) . This prompted us to analyze the interaction of HTF with C4 and EGFR cis sequences. HTF bound to the C4 fragment with a higher affinity than to the EGFR promoter fragment. The C/G difference between the two cis sequences partially explained the difference in the binding affinity of HTF for C4 and for EGFR sequences. Indeed, the G to C shift increased the binding capacity of HTF to the EGFR fragment, whereas the C to G shift reduced the binding capacity of HTF to the C4 fragment. The importance of the second C in the core binding site for HTF was also pointed out by in vitro and in vivo experiments. Indeed, the substitution of an A for the C abolished the binding of HTF and significantly diminished the stimulating activity of the C4 fragment. Similarly, Scott et al. (16) have shown that minor changes in the ets response element of the HER2 promoter reduced the interaction with nuclear proteins and the promoter activity in MDA-MB-453 cells. However, the surrounding regions could also probably play an important role in the DNA/HTF interaction, as demonstrated by Johnson (25) for the AP-2 recognition sites present in the EGFR promoter. The last question addressed in this study is whether there is a correlation between the HTF binding activity and the HER2 mRNA expression in breast cancer cells. We have tested seven mammary cell lines for HTF binding capacity to the C4 fragment. The nuclear extracts from the BT-474, ZR-75-1, and SK-BR-3 HER2-overexpressing cell lines contain the highest levels of binding activity to the HTF cis sequence. The MCF-7, T-47D, and MDA-MB-453 showed a lower HTF activity. The nontumoral HBL-100 cells showed a weak HTF activity. These results suggest that HTF could also be important for HER2 overexpression in other tumor mammary cells. The fact that MDA-MB-453 cells showed a low level of HTF binding activity, whereas they overexpress the HER2 transcript, confirms our previous results indicating that a different mechanism might be responsible for HER2 overexpression in MDA-MB-453 cells (19) . Thus, HTF might contribute to the HER2 overexpression with other transcription factors like AP-2 (13 , 14) and PEA3 (15 , 16) proteins. These last transcription factors interact with cis sequences located in a 250-bp-long region immediately upstream the CAP site. HTF probably plays a major role in HER2 overexpression because the interaction of HTF with its cis sequence stimulates the transcriptional activity of this 250-bp promoter by a factor of 30 (19) .