A c-erbB-2 Promoter-specific Nuclear Matrix Protein from Human Breast Tumor Tissues Mediates NF-κB DNA Binding Activity*

The c-erbB-2 gene overexpression plays a major role in the pathogenesis of breast cancer. Binding studies detected a nuclear matrix protein (NMP) in human breast tumor tissues that recognizes a matrix attachment region (MAR) in the immediate vicinity of the c-erbB-2 gene promoter. This NMP is expressed in breast tumor tissues and cell lines along withc-erbB-2, but is not found in corresponding normal tissues. Furthermore, when NMP purified from the breast tumors by its affinity to the MAR sequence is added to nuclear extracts of breast cancer cells, it selectively stimulates the binding of the NF-κB transcription factor to DNA. A model is suggested in which the association of the MAR-like sequence with the nuclear matrix raises the local concentration of the specific NMP, which in turn interacts with the nuclear factor NF-κB to increase its local level. Such a complex could explain at a molecular level the “increase in NF-κB DNA binding activity” often observed in c-erbB-2- andBRCA1-positive human breast tumors. The increased NF-κB activity could thereby contribute to breast cancer progression.

Invasive breast cancer is the most common serious malignancy and a leading cause of death among women. It is generally believed that the overexpression of the c-erbB-2 gene (also known as HER-2 and Neu; Refs. 1-3) leads to abnormal growth, cellular transformation, and neoplasia (4 -6); but the function of c-erbB-2 and the role of overexpression in tumor progression are still obscure.
Some functional hints have come from the detection of a c-erbB-2 promoter-specific DNA-binding nuclear protein that is present only in malignant human breast tissues (9) and induces mitogenesis and cell surface expression of the c-erbB-2 protein in resting NIH/3T3 cells. Many cellular factors are involved in such specific gene regulation (e.g. Refs. 7-9) and can, in some cases, elevate the expression of c-erbB-2 at the transcriptional level by a transactivating effect at a promotervicinal DNA sequence in the gene. To better understand the progression of breast tumorigenicity in c-erbB-2-expressing tumors, an attempt was made to study its mechanism within the innermost nuclear structure, the nuclear matrix, through DNA-protein interaction. Nuclear matrix is believed to play critical roles in regulating many key biological reactions in the nucleus such as: gene transcription, DNA replication, DNA organization, and RNA splicing and processing. In this study, we show that a nuclear DNA-binding protein is a breast tumorspecific nuclear matrix protein (NMP) 1 that recognizes a nonconventional matrix attachment region (MAR) in the c-erbB-2 gene promoter. The NMP protein stimulates the binding of NF-B to DNA in nuclear extracts from breast cancers, suggesting a possible role in breast cancer progression.

MATERIALS AND METHODS
Human Breast Tissue Samples-Human normal (benign) and malignant (tumor) breast tissues were obtained from Pathology Department of Wayne State University Medical School, Detroit, MI. Only confirmed sets of c-erbB-2-expressing as well as negative breast tissue samples were used.
Preparations of Nuclear Matrix (NM) and NMP-Nuclear matrices from breast tissues and cell lines were prepared by a standard procedure (33) with certain modifications. Tissues were powdered under liquid N 2 , and nuclei were purified according to the protocol. Nuclei were extensively treated with 200 g/ml DNase I enzyme for 2 h with constant agitation, at room temperature. DNase I-treated residual nuclear pellets were extracted several times with 2 M NaCl containing high-salt buffer and finally washed with RSB (10 mM NaCl, 10 mM Tris-HCl, pH 7.5, 3 mM MgCl 2 , 0.5 mM phenylmethylsulfonyl fluoride)-0.25 M sucrose buffer.
For the NMP isolation, the method of Fey and Penman (29) was used with certain modifications. The nuclear matrix were solubilized in 8 M urea containing disassembly buffer for 30 min at room temperature and then dialyzed in a renaturation buffer (100 mM KCl, 25 mM HEPES, pH 7.6, 5 mM MgCl 2 , 2 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, 0.125 mM EGTA, and 2% glycerol) overnight at room temperature. The dialyzed material was centrifuged at high speed to remove the intermediate filaments (the pelleted material). The clear supernatant thus obtained is the solubilized NMP pool, which was further dialyzed in low salt buffer (25 mM HEPES, pH 7.6, 100 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, and 0.2 mM phenylmethylsulfonyl fluoride) and used for DNA-protein gel binding and SouthWestern assays as well as for affinity purification.
Oligonucleotide Probes-Corresponding to different regulatory regions of the c-erbB-2 gene as well as BRCA1, NF-B binding sequences were chemically synthesized, gel-purified, annealed to make doublestranded, and 5Ј-end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase, according to standard protocols.
Preparation of Nuclear Extracts-Nuclear extracts from breast tissues and cell lines were prepared according to a method described earlier (9). Nuclear extracts were finally dialyzed against buffer (25 mM HEPES, pH 7.6, 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, and 0.2 mM phenylmethylsulfonyl fluoride) and stored at Ϫ70°C in small aliquots.
MAR Binding Assay-MAR binding assays were performed according to a standard procedure (33) with some modifications. A DNase I-treated, high-salt-extracted residual nuclear pellet was washed three times in RSB-0.25 M sucrose buffer and once in MAR binding buffer. The resulting pellet (nuclear matrix) was mixed with 5Ј-end-labeled oligonucleotide probes corresponding to regions: AϩT (Ϫ450/Ϫ390), GGA-rich (Ϫ79/Ϫ22), Ϫ300/Ϫ280, and Ϫ22/ϩ9 (34) of the c-erbB-2 gene in 100 l of MAR binding buffer for 4 h at room temperature, with constant agitation. After three washings with MAR binding buffer the bound DNAs were further processed and gel electrophoresed as described in Ref. 33. In most of MAR binding assays, 5A 260 of the NM and 50,000 cpm of 5Ј-DNA probes in the presence of 150 g/ml Escherichia coli DNA were used, unless otherwise mentioned.
Southern Blot (DNA) Analysis-DNA isolated from 5A 260 units of nuclear matrix were electrophoresed on agarose gel, transferred to the nitrocellulose paper, and hybridized with GGA-specific (Ϫ79/Ϫ22) and nonspecific 32 P-nick-translated probes according to standard protocol.
Electrophoretic Mobility Shift Assay (EMSA) and SouthWestern Blot Assay-EMSA reactions were carried out with NMPs and nuclear ex-tracts and with 32 P-labeled double-standard oligonucleotide probes in the presence of 3 g of poly[d(I⅐C)] and KCl containing binding buffer, exactly as described earlier (8,9). For competition studies, 50-fold molar excess of unlabeled double-stranded oligonucleotides were added. In SouthWestern blot assay, 10 g of NMPs were usually resolved on a 10% SDS-polyacrylamide gel, electrotransferred onto nitrocellulose membrane, and renatured as described (8,9). The membrane was hybridized with 2.5 ϫ 10 6 cpm of 32 P-labeled oligonucleotide probes for 15 h at room temperature, with constant agitation. The rest of the steps were as described in the standard protocol.
Affinity Purification of Breast Tumor-specific Nuclear Matrix Protein-An affinity resin with Ϫ79/Ϫ22, GGA-rich, MAR-like sequences of c-erbB-2 was generated as described earlier (9). Complementary nucleotides corresponding to Ϫ79 to Ϫ22 sequences of c-erbB-2 promoter were annealed, phosphorylated, ligated, and coupled to CNBr-activated Sepharose 4B.
Large pools of NMPs from breast tumor tissues were used for the purification purposes. All the steps of the purification protocols were exactly as described earlier (9). Briefly, solubilized bulk NMP pools from breast tumor tissues were first mixed with total 200 g of salmon sperm DNA for 15 min and then mixed with the nonspecific, Ϫ22/ϩ9, DNA affinity resin for 6 h, all at 4°C, with shaking. The low-salt flow-through from nonspecific affinity column was then mixed with the GGA-rich (Ϫ79/Ϫ22) DNA affinity column, and the rest of the purification protocol was followed as described (9).
Western Blot (Protein) Analysis-Twenty micrograms of nuclear matrix from normal and tumor tissues were resolved on a 10% SDS- polyacrylamide gel and electrotransferred on to nitrocellulose membrane. Filters were incubated with NF-B antibodies (anti-p65 and anti-p50 subunits of NF-B) and anti-human nuclear matrix 45. The rest of the procedure followed was as suggested by the manufacturers (Santa Cruz Biotechnology Inc. (Santa Cruz, CA) and Upstate Biotechnology, Inc. (Lake Placid, NY).

RESULTS
To better understand the progression of breast tumorigenicity in c-erbB-2-expressing tumors, an attempt was made to study the possible involvement of protein interactions with DNA within the innermost nuclear structure, the nuclear matrix. To test for novel factor(s) in the nuclear matrix that bind to a specific segment of the c-erbB-2 gene regulatory sequences, MAR binding studies were conducted with many distinct segments of 5Ј-upstream regulatory regions of c-erbB-2 enhancerpromoter ( Fig. 1A) and nuclear matrix preparations from human normal, c-erbB-2-expressing tumor breast biopsy tissues and breast cancer cell lines (33). Results (Fig. 1B) indicate that of the four regions tested, only one specific region, rich in GGA repeats (Ϫ79/Ϫ22) and in the immediate vicinity of the c-erbB-2 gene transcription start site, has the specific binding affinity observed only with the nuclear matrices of tumor tissues (Fig. 1B, lane 1) and not with the normal tissues (Fig. 1B,  lane 2). This observation was further confirmed in the breast tumor cell line BT-20 (Fig. 1B, lane 3). The upstream region that contains a stretch of AT-rich sequences (conventionally a part of MAR) binds less effectively to the nuclear matrix than the GGA-rich region (Fig. 1B, lane 4). Breast tumor and BT-20 cell nuclear matrices binding with the GGA-rich DNA element were tested with more stringent conditions by using increasing amounts of a competitor E. coli DNA, where the GGA-rich DNA element still appears to be the dominant bound material (Fig.  1C, lane 3, breast tumor NM, lane 4, BT-20 NM with 200 g/ml E. coli DNA versus lanes 1 and 2 of same NMs with 100 g/ml E. coli DNA). A Southern blot analysis with DNAs from tissue nuclear matrices further demonstrates a strong binding with GGA-rich DNA probe (Fig. 1D, lane 2), not observed with the nonspecific (Ϫ22/ϩ9) DNA probe. DNA from normal breast nuclear matrix failed to show any affinity with either probes (Fig. 1D, lane 1). These results (Fig. 1B-D) demonstrate the presence of a MAR-like DNA element within the immediate upstream of c-erbB-2 promoter. These sequences are rich in GGA and not in AT, which traditionally is part of, but not a prerequisite for, MAR. This GGA-rich region (Ϫ79/Ϫ22) has also been mentioned as an activator of gene promoter and an alternate transcription start site region of the c-erbB-2 gene (35,36). Other groups have reported the presence of MAR in chicken oviduct, chicken lysozyme gene, avian-globin gene, mouse immunoglobulin K chain, and 5Ј-upstream regulatory region of human H4 histone gene promoter (25,31,32,37,38).
To reaffirm the results of Fig. 1, whether this DNA attachment onto the nuclear matrix of breast tumors is mediated by any specific protein factor(s), we performed DNA-protein binding gel shift assays with NMPs from breast biopsy tissues. Solubilized NMPs devoid of intermediate filaments (29) from several normal (benign), tumor breast biopsy tissues and breast cancer cell line, BT-20, were mixed with the radiolabeled GGA-rich DNA probe, in DNA binding assays. The DNA binding EMSA clearly demonstrates a strong DNA-protein complex formation only with the NMPs of breast tumors (Fig. 2A, panels  a-c, lanes 2, 4, 6) and not with NMPs of their normal breast tissue counterparts (Fig. 2A, panels a-c, lanes 1, 3, 5). The tumor cell line BT-20 displays the same complex as the tumor tissue NMP (Fig. 2A, panel d, lane 8), while no such complex is observed with a normal breast cell line MCF-10A (Fig. 2A,  panel d, lane 7). Different groups have reported the presence of MAR-binding proteins, such as: SATB1, OCT-1, nucleolin, and recently from human breast carcinomas (26,28,(37)(38)(39)(40).
To test whether this binding activity of NMPs from breast tumors is a specific phenomenon, we examined the status of NMP binding with other region of c-erbB-2 (Ϫ22/ϩ9). Again the  1, 2, and 3 (cytoplasmic, nuclei, and nuclear extracts, respectively). gel shift assay from two more sets of breast tumor, but not breast normal NMPs clearly demonstrates the formation of a specific DNA-protein complex only with the GGA-rich, Ϫ79/ Ϫ22, DNA probe (Fig. 2B, panel a, lanes 2 and 4 versus lanes 1  and 3; BT-20 NMP, lane 6 versus MCF10A NMP, lane 5). No binding is observed in these NMP extracts with other DNA (Ϫ22/ϩ9) probe (Fig. 2B, panel b). These results were supported by an SouthWestern blot assay (Fig. 2C), probed with the same two DNA probes. The MAR-like GGA probe (Ϫ79/ Ϫ22) identifies a dominant nuclear matrix protein of 68 kDa, seen only in the breast tumor NMPs and not in their normal counterparts (Fig. 2C, lanes 2 and 4 over lanes 1 and 3 as well as in BT-20 NMP, lane 5). The second DNA probe (Ϫ22/ϩ9) fails to demonstrate any binding. Furthermore, we investigated whether this binding activity of GGA with breast tumor NMPs is localized in the nuclear matrix domain only or is a general phenomenon, found in other compartments of the cell. Results (Fig. 2D)  A slower migrating intense DNA-protein complex seen only in the nuclei (lane 2) and nuclear extract (lane 3) could probably be due to CAAT and TATA (as well as some other unknown)binding proteins, since the specific probe (Ϫ79/Ϫ22) also contains CAAT and TATA binding sequences. This complex is totally absent from the nuclear matrix compartment (lanes 4 and 5), affirming the fact that the nuclear matrix is devoid of these high-salt-sensitive nuclear factors.
Together, the results of Figs. 1 and 2 clearly demonstrate the presence of a MAR-like non-conventional (rich in GGA and not in AT) element in the close vicinity of the c-erbB-2 gene transcription start site, whose attachment to the breast nuclear matrix is mediated by a sequence-specific DNA-binding NMP that is expressed only in the malignant tissues and breast cancer cell line BT-20 and not in the normal breast tissues.
To elucidate a functional role(s) of the identified NMP factor, we affinity-purified the protein from the breast tumor NMP pools on the GGA-rich DNA (Ϫ79/Ϫ22) affinity column. The purified NMP from breast tumors displays a dominant polypeptide of 68 kDa (Fig. 3A, panel a, lane 2). This purified NMP binds the GGA-rich (specific) probe in a sequence-specific manner, as demonstrated by SouthWestern (Fig. 3A, panel b) and EMSA (Fig. 3B) assays. GGA-rich (specific) probe observes a strong band with 50 ng of purified NMP and 10 g of breast tumor NMP pools (Fig. 3A, panel b, lanes 2 and 1, respectively), while the nonspecific (Ϫ22/ϩ9) DNA probe fails to bind these NMPs. The EMSA result (Fig. 3B) with 50 ng of purified NMP displays the same DNA-protein complex (Fig. 3B, lane 3) as in breast tumor NMP pools (10 g, lane 2), which is competed out with specific (GGA-rich, Ϫ79/Ϫ22) cold DNA (lane 4), but not with nonspecific (Ϫ22/ϩ9) cold DNA (lane 5).
To ascertain a functional role that this specific NMP factor may be contributing, a functional test was performed. The purified NMP was added to the nuclear extracts from various tumor cell lines in an EMSA binding reaction, using defined DNA probes from c-erbB-2, BRCA1, exon 1 regulatory sequences, and NF-B binding sequences (see "Materials and Methods," "Oligonucleotide Probes"). Addition of increasing amounts of purified NMP into MDA-MB231 and BT-20 nuclear extracts and with DNA probe (Ϫ22/ϩ9) of c-erbB-2 does not have any effect on the DNA-protein complex formation (Fig. 4A,  lanes 2-4 versus lane 1). Similarly addition of NMP into these extracts and with DNA probe (Ϫ78/Ϫ42) of BRCA1 also does not influence the DNA-protein complex (Fig. 4B, lanes 2-4  versus lane 1). Interestingly, addition of purified NMP to these nuclear extracts and with NF-B binding DNA probe appears to have a selective stimulatory effect on the DNA binding activity of nuclear factor NF-B ( This increase in NF-B binding activity appears to be highly selective, which is not observed with other two probes (of Ϫ22/ϩ9 of c-erbB-2 and Ϫ78/Ϫ42 of BRCA1) tested (Fig. 4, A  and B). Purified NMP factor alone does not bind to B or any other probe except its specific GGA-rich recognition sequence (Ϫ79/Ϫ22).
To further evaluate whether NF-B is present in the nuclear matrix of breast tissues, we performed a Western blot analysis of the nuclear matrix preparations from normal and tumor breast tissues, which were subsequently probed with anti-p65 and anti-p50 antibodies of NF-B. As per our prediction, reasonable amounts of NF-B (p65 as well as p50) subunits were found preferentially associated with the breast tumor nuclear matrix (Fig. 5A, panel a; lane 2 in both panels) and not with the normal breast nuclear matrix (Fig. 5A, panel a; lane 3 in both panels). Lane 1 in both panels are standard control p65 and p50 peptides. To demonstrate that equal amounts of proteins were used, we silver-stained a parallel gel with these samples and also performed a Western blot assay with a control anti-human nuclear matrix antibody. The results of the silver stain (Fig.  5A, panle b) and Western blot analysis (Fig. 5A, panel c) confirm that equivalent amounts of protein were used from the tumor (lane 2) and normal (lane 3) breast tissue nuclear matrices. Additionally, we also performed a NF-B-specific Western blot analysis with the total nuclear content along with the nuclear matrix from the breast tumor tissue. The results of Fig.  5B clearly demonstrate that a much smaller yet a reasonable fraction of total nuclear NF-B is associated with the nuclear matrix (Fig. 5B, panel a, lane 3 versus lane 2). A silver stain of the nuclear extract (lane 2) and nuclear matrix (lane 3) are shown underneath (Fig. 5B, panel b) to show that equivalent amounts of protein were used for the analysis. DISCUSSION The proteinaceous network of the NM is believed to be involved in DNA organization, DNA replication, gene transcription, and RNA splicing and processing (17)(18)(19)(20). The protein components (NMPs) of the NM provide the structural framework for loop domains of DNA, attached at MARs (15,21,22).
Several groups (12-14, 25-28, 41) have demonstrated that sequence-specific DNA-binding proteins can be components of nuclear matrix attachment sites. In recent years, intensive studies have also suggested possibly related role(s) of the nuclear matrix in tumor progression (10 -16). In cancer cells, some transforming proteins appear to be associated with the matrix (23,24), and there are also indications of specific alterations in the protein composition of the matrix as cells undergo differentiation (25) and during the invasion and proliferation of tumors (14,15,24,25,42,43). Could a "tumor-specific" NMP be involved in a transition associated with tumor progression?
The results reported here may provide such an instance, based on a sequence-specific DNA-binding nuclear matrix protein exclusively present in human breast tumor tissues. Concerning the association with DNA, the protein is unlikely to be part of one of the integrated structures that generates loop domains of chromatin. Such binding sequences, found in DNase I-sensitive sites near the bases of the loops, are generally rich in AT sequences. In contrast, the sequence-specific DNA-binding protein studied here mediates the attachment of a GGA-rich enhancer region of c-erbB-2 to the nuclear matrix. This is the other recognized mode of attachment to the nuclear matrix: transient and based on enhancer regions upstream (5Ј) of actively transcribed genes (31,32).
Based on these findings, the juxtaposition of DNA binding factors and DNA regulatory elements at the matrix may increase the local concentration of various transcription factors (25,29,30). We suggest that the specific binding of the c-erbB-2 DNA element to the breast tumor nuclear matrix thereby concentrates nuclear factor NF-B at those sites. It can thereby rationalize the increased NF-B DNA binding activity observed in malignant breast and other solid tumors that express c-erbB-2 (and BRCA1) genes (44,45). 2 Increased transcription of this tumorigenic protein could then be a factor in tumor progression; but one must now approach the difficult question of how the specific NMP is expressed and regulated in breast tumors.