NFκB1/p50 Is Not Required for TNF-stimulated Growth of Primary Mammary Epithelial Cells: Implications for NFκB2/p52 and RelB

Abstract

NFκB plays an important role in mammary gland development and breast cancer. We previously demonstrated that TNF stimulates growth of mammary epithelial cells (MEC) in a physiologically relevant 3D primary culture system, accompanied by enhanced DNA-binding of the NFκB p50 homodimer. To further understand the mechanism of TNF-stimulated growth of primary MEC, the requirement for NFκB1/p50 and the role of cyclin D1 in TNF-stimulated growth were examined. TNF induced the formation of DNA-binding complexes of p50 and p52 with their coactivator bcl3 in MEC nuclear extracts. Concomitantly, TNF increased the binding of NFκB proteins to the κB site on the cyclin D1 promoter, and increased expression of cyclin D1 mRNA and protein. Using MEC from p50 null mice, we found that p50 was not required for TNF-induced growth nor for up-regulation of cyclin D1. However, TNF induced a p52/RelB NFκB DNA-binding complex in p50 null MEC nuclear extracts. In addition, we found that in wild type MEC, TNF stimulated the occupancy of p52 and RelB on the cyclin D1 promoter κB site, while p50 was present constitutively. These data suggest that in wild type MEC, TNF stimulates the interaction of bcl3 with p50 and p52, and the binding of p52, as well as RelB, to cyclin D1 promoter κB sites, and as a consequence stimulates the growth of MEC. In the absence of p50, p52 and RelB can compensate for p50 in TNF-stimulated growth and cyclin D1 induction in MEC.

INTRODUCTION

Breast cancer is the most common type of cancer in US women. In order to gain a greater understanding of its initiation and progression, it is necessary to identify those factors that regulate the growth and differentiation of the mammary epithelium. Previous studies in our laboratory suggested that tumor necrosis factor (TNF) is a potential regulator of normal mammary gland development ((1), and unpublished). Normal mammary epithelial cells (MEC) express TNF protein, as well as TNF and TNF receptor mRNAs in a developmentally regulated manner (1). Moreover, in vivo studies using TNF null mice found that ductal morphogenesis of the mammary epithelium is inhibited during puberty (manuscript in preparation). In a physiologically relevant primary MEC culture model established in our laboratory (2–4), we previously demonstrated that TNF induced the growth as well as extensive branching morphogenesis of rat MEC (5–7), effects that are mediated through the p55 TNF receptor (1) and matrix metalloproteinase-9 (7). TNF was also found to inhibit functional differentiation (1,5,6,8,9). More recently, we found that in contrast to the TNF-induced cell death or stasis in breast cancer cell lines, TNF increased the growth of mammary tumor cells derived from MNU-induced rat (10) and erbB2-overexpressing mouse (unpublished) mammary tumors in three dimensional primary culture. The proliferation of both normal and transformed rat MEC was accompanied by increased DNA-binding activity of the NFκB p50 homodimer (10).

NFκB is a pleiotropic transcription factor, controlling inflammation, cell survival, transformation and oncogenesis (11). There are five members in the NFκB/Rel family: p65 (RelA), p50/105 (NFκB1), p52/p100 (NFκB2), c-rel, and RelB, which bind to DNA as homo- or heterodimers. All the NFκB proteins share a 300-residue N-terminal Rel homology domain that mediates the DNA binding, dimerization and nuclear targeting functions, as well as interaction with IκB (11). The mature DNA-binding proteins p50 and p52 are generated by proteolytic processing of p105 and p100 precursors, respectively (12). Among the five NFκB proteins, p65, c-Rel and RelB contain transcriptional activation domains, whereas p50 and p52 do not (12). However, a member of the IκB family, Bcl3, can function as a transcriptional coactivator for p50 and p52 homodimers (13,14). NFκB signaling consists of both canonical and non-canonical pathways (reviewed in reference (11,15)). In the canonical pathway, upon stimulation with cytokines, the IκB proteins that sequester NFκB proteins in the cytoplasm are phosphorylated by an IκB kinase (IKK) complex and rapidly degraded, releasing NFκB to translocate to the nucleus. In the non-canonical pathway, NFκB-inducing kinase (NIK) and IKKα are stimulated through receptors including the B cell-activating factor receptor (BAFF), CD40, and the lymphotoxin βκreceptor (LTβR), resulting in an induction of the processing of p100 to p52 and facilitating the nuclear translocation of RelB/p52 dimers into the nucleus (15). In contrast to the tight regulation of p52 processing, the proteolysis of the precursor protein p105 to p50 is constitutive in unstimulated cells (11). Several studies have demonstrated that p50 and p52, the two highly homologous NFκB subunits, have distinct but also many redundant functions in bone development and B-cell differentiation (16–19).

NFκB plays a key role in mammary gland development (20) and breast cancer progression (21). Abnormal proliferation and branching of the mammary epithelium is observed in mice lacking the IκBα gene (22). Elevated NFκB DNA-binding activity has been reported in both mammary carcinoma cell lines and primary breast cancer cells of human and rodent origin (23–26). Moreover, interruption of NFκB activation restored the sensitivity of endocrine-resistant breast cancer to tamoxifen (27). A number of studies have been carried out to understand the contribution of different NFκB members to breast tumorigenesis (24,28,29). Interestingly, while p50/p65 is the major increased NFκB dimer in breast cancer cell lines, NFκB dimers composed of p50 or p52, as well as their coactivator, bcl3, are the selectively up-regulated NFκB proteins in human breast tumor tissue, suggesting a different activation pattern in human breast tumors when compared to breast cancer cell lines (24,29). More recently, Zhou et al. suggested that the activity of the p50 homodimer might be used as a prognostic marker in a subset of ER positive breast cancer patients (27,30). However, little is known about the role of p50 in regulating the growth of mammary epithelial cells.

In the studies reported here, we investigated the role of NFκB1/p50 in primary MEC, with the objective of determining whether p50 was required for TNF stimulation of growth and cyclin D1 expression. We found that TNF increased the expression of cyclin D1 in proliferative primary MEC associated with an activation of NFκB1/p50. However, NFκB1/p50 was not required for the proliferation or the up-regulation of cyclin D1 by TNF in MEC. In contrast, NFκB2/p52 and RelB appear to play important roles in TNF-stimulated, NFκB-mediated growth of primary MEC.

MATERIALS AND METHODS

Animals

Sprague-Dawley CD female rats were purchased from Charles River Laboratories (Wilmington, MA). NFκB/p50 null mice (B6, 129-Nfkb1^tm1Bal) or age-matched wild type mice (B6129PF2/J) were obtained from Jackson Laboratories (Bar Harbor, ME). Mice and rats were maintained in microisolator cages in a temperature and humidity controlled environment with a 12 hr light-dark cycle, and were given chow and water and libitum. The care and use of the animals was in accordance with NIH guidelines and Institute Animal Care and Use Committee regulations.

Antibodies

The following antibodies were used in these studies. Anti-p50 (sc-114X), p52 (sc-298X), p52 (sc-848X), p65 (sc-109X), c-Rel (sc-70X), Rel B (sc-226X) and anti-bcl3 (sc-185) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-Hsc70 (SPA-820) and anti-p50 (KAP-TF112) were from Stressgen Biotechnologies Corporation (Victoria, BC Canada). Anti-cyclin D1 (#2926) was purchased from Cell Signaling Technology Inc. (Danvers, MA). Rabbit anti-p50 serum 1157 (NCI 1157) and anti-p52 serum 1495 (NCI 1495) were kindly provided by Dr. Nancy Rice and Mimi Ernst at NCI, or the Biological Resources Branch, DCTD, NCI-Frederick Cancer Research and Development (Frederick, MD). Detection of p50 by western blot was performed using the NCI 1157 antibody, unless specified otherwise. HRP conjugated donkey anti-mouse and donkey anti-rabbit secondary antibodies were purchased from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA).

Isolation and primary culture of mammary epithelial organoids

Isolation and primary culture of mammary epithelial organoids were performed according to the protocol developed by our lab (31,32). Briefly, rats at 56–66 days of age or mice at 8–10 weeks of age were sacrificed, then mammary glands removed, minced and digested with collagenase III (Worthington Biochemical Corporation, Lakewood, NJ) at 172.8 U/ml (rat) or 108 U/ml (mouse) and with dispase (Roche Applied Science, Mannheim, Germany) at 1.2 mg/ml (rat) or 0.7 mg/ml (mouse) for 12 hrs to isolate mammary epithelial organoids. After enrichment (31,32), the rat or mouse organoids were embedded within an EHS-derived reconstituted basement membrane (RBM) (prepared from the EHS sarcoma as described previously (2,3)) and cultured in an EGF-free serum-free medium (SFM) consisting of phenol red-free DMEM-F12 (50:50, v:v) with 10 μg/ml insulin, 1 μg/ml progesterone, 1 μg/ml hydrocortisone, 1 μg/ml prolactin, 5 μg/ml transferrin, 5μM ascorbic acid, 1 mg/ml fatty acid-free BSA, and 50 μg/ml gentamycin. Recombinant human or mouse TNF (40 ng/ml) (Biosource International, Camarillo, CA) were used for TNF-treated rat or mouse MEC respectively. Ovine prolactin (NIDDK-oPRL-21) was obtained from Dr. A. F. Parlow at the National Hormone & Pituitary Program, Harbor-UCLA Medical Center, CA.

For EMSA and western blot, ~ 6 rats were used to generate MEC organoids for each experiment. At least 12 rats were used for each DAI assay. In the CHIP assay, 12–16 rats were used to generate MEC organoids for each time point. For the cell growth, EMSA and western blot studies with p50 null and wild type mice, 20–30 mice per group were used for each experiment.

Morphological analysis, MTT and ³H-thymidine incorporation assays

Morphological development of the MEC organoids from p50 null or wild type mice was photographed under the 4 × objective on days 1, 4, 9 and 15 of primary culture using a Nikon FX-35A camera mounted on an Olympus CK2 inverted microscope. To evaluate growth of the epithelial organoids, MEC (5 × 10⁵) from age-matched p50 null or wild type mice were cultured in a 24-well plate within the EHS RBM in 1 ml SFM with or without TNF (40 ng/ml) for the times indicated. The viable cell number was measured by the MTT assay as previously described (5) and is presented as optical density. For determination of ³H-thymidine incorporation, wild type mouse MEC (5 × 10⁵ per well of a 24-well plate) organoids were cultured with or without TNF (40 ng/ml) for 6 days, pulsed labeled with ³H-thymidine (5 μCi/well) for 4 hr, and ³H-thymidine incorporation determined as described previously (6). CPM data were normalized to cell number using the MTT assay.

Cytosolic and nuclear extraction

Cellular fractionation was performed according to Brasier et al (33) with modification. MEC were isolated from the EHS RBM after incubation with 5 U/ml dispase (BD Biosciences, Franklin Lakes, NJ) in Hank’s Balanced Salt Solution (HBSS) for 1 hr at 37°C, and washing with cold HBSS. Cell pellets were resuspended in Buffer A (50 mM HEPES, pH 7.4, 10 mM KCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol (DTT), 0.1 μg/ml phenylmethylsulfonyl fluoride (PMSF), 1 μg/ml pepstatin A, 1 μg/ml leupeptin, 10 μg/ml soybean trypsin inhibitor, 10 μg/ml aprotinin, and 0.5% Nonidet P-40). After 10 min on ice, the samples were sheared ten times through a 22-gauge needle and centrifuged at 500 × g for 4 min at 4°C. The supernate was further centrifuged at 100,000 × g for 1hr at 4°C and the clear fraction was saved for the cytosolic extract. The nuclear pellet was resuspended in Buffer B (Buffer A with 1 M sucrose) and centrifuged at 14,000 × g for 30 min at 4°C. The nuclei were purified by washing twice with cold PBS with 1 mM DTT and then incubated in Buffer C (10% glycerol, 50mM HEPES, pH 7.4, 400 mM KCl, 1 mM EDTA, 1 mM EGTA, 1mM DTT, 0.1 μg/ml PMSF, 1 μg/ml pepstatin A, 1 μg/ml leupeptin, 10 μg/ml soybean trypsin inhibitor, 10 μg/ml aprotinin) with vigorous vortexing for 30 min at 4°C. After centrifugation at 14,000 × g for 10 min at 4°C, the supernate was collected as nuclear extract. As determined by immunoblotting with anti-lactate dehydrogenase (Fitzgerald Industries International Inc., Concord, MA) or anti-lamin B (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), no cross-contamination was seen between cytosolic or nuclear fractions (data not shown).

Western blot

Twenty to fifty micrograms of protein were separated on 10% SDS-PAGE gels, transferred to PVDF membranes and blotted with a 1:200 dilution of anti-bcl3 or a 1:1,000 dilution of anti-p50, p52 or cyclin D1. The washed membranes were then incubated with horseradish peroxidase-conjugated donkey anti-rabbit or donkey anti-mouse antiserum (1:5000 dilution). Immunoreactive bands were visualized using the ECL detection system (Amersham Bioscience, Piscataway, NJ). The membranes were then stripped and reprobed with anti-Hsc 70 for an internal control. The density of each band was first normalized to that of the internal control and then fold inductions were determined as the ratio of relative intensities of TNF-treated versus control samples.

Isolation of RNA and Northern blot

To isolate RNA from primary cultured MEC, media were removed from the 100 mm culture dishes and 3 ml TRIzol reagent (Life Technologies, Grand Island, NY) was added to each dish. Total RNA was isolated according to the manufacturer’s protocol. Northern blot was performed as previously described (8). The density of cyclin D1 mRNA was first normalized by that of the internal control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and then fold inductions were determined as the ratio of relative intensities of TNF-treated versus control samples. The plasmid containing cyclin D1 cDNA was kindly provided by Dr. Yan Dong (Roswell Park Cancer Institute, Buffalo, NY). The recombinant human GAPDH cDNA probe was purchased from Clontech (Palo Alto, CA).

DNA Affinity Immunoblotting (DAI) assay

The DAI assay was performed according to Liu et al (34). In brief, 200 μg cytosolic or nuclear protein from control or TNF-treated MEC was incubated with 5′ end-labeled biotin-consensus NFκB oligo [5′-AGTTGAGGGGACTTTCCCAGGC-3′] (Integrated DNA Technologies, Coralville, IA) at 4°C for 30 min. Biotin-NFκB oligo/protein complexes were captured by incubation with streptavidin magnetic beads (Promega Corporation, Madison, WI). The beads were pelleted by a strong magnet and the protein complex was eluted using 2X SDS sample loading buffer and resolved on SDS-PAGE. The proteins bound to the NFκB oligo were detected by western blot using specific antibodies as indicated.

Electrophoretic mobility shift analysis (EMSA)

EMSA was performed as described previously (10) except that 10 μg nuclear protein was used for each reaction and antibodies were incubated at 4°C overnight for the supershift analyses. NFκB oligonucleotide containing consensus NFκB binding site (5′-AGTTGAGGGGACTTTCCCAGGC-3′) and mutant NFκB oligo (5′-AGTTGAGGCGACTTTCCCAGGC-3′) were ordered from Santa Cruz Biotechnology (Santa Cruz, CA). NFκB oligonucleotide containing the κB site on the rat cyclin D1 promoter (5′-ACTACAGGGGAGTTTTGTTG-3′) was purchased from Integrated DNA Technologies (Coralville, IA).

Chromatin Immunoprecipitation Assay (ChIP)

The ChIP protocol was adapted from previous reports (35,36) with modifications. After culturing MEC organoids in serum-free medium for five days, cells were treated with or without TNF (40 ng/ml) for the times indicated. MECs were then isolated from the EHS RBM and fixed with 1% formaldehyde at room temperature for 10 min. Cross-linking was stopped by the addition of glycine to a final concentration of 125 mM for 5 min. Fixed MECs were washed in cold PBS. Pellets from approximately 10⁸ cells were resuspended in 2 ml cell lysis buffer (5 mM PIPES [piperazine-N, N′-bis(2-ethanesulfonic acid); pH 8.0], 85 mM KCl, 0.5% NP-40, 1 μg/ml leupeptin, 1 μg/ml aprotinin and 1 mM PMSF) and incubated on ice for 10 min. Cell membranes were broken by shearing the MEC suspension ten times through a 22-gauge needle. Nuclei were pelleted by centrifugation at 2,000 × g for 5 min. Using nuclei counts to estimate cell number, nuclear pellets from control or TNF-treated groups were resuspended in nuclei lysis buffer (50 mM Tris-HCl [pH 8.1], 10 mM EDTA, 1% SDS with the same protease inhibitors as in the cell lysis buffer) to the same final concentration of ~ 2 × 10⁸ per ml and incubated on ice for 10 min. Chromatin was sheared to an average length of 100 to 400 bp DNA fragments by sonication of the nuclear lysate on ice for twenty-five 10-sec pulses using the Sonic Dismembrator (Fisher Model 60, Pittsburgh, PA) at a power setting of 2 to 3. Debris was cleared by centrifugation at 14,000 × g for 15 min at 4°C. A part of the sonicated lysate from each time point (50-100 μl) was saved to use for input control. The lysate from 2 × 10⁸ MEC was diluted 10 fold with ChIP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl [pH 8.1], 167 mM NaCl with the same protease inhibitors as in cell lysis buffer) and pre-cleared with 80 μl of salmon sperm DNA/Protein A agarose (Upstate Cell Signaling Solutions, Lake Placid, NY) for 30 min at 4°C followed by centrifugation at 500 × g for 5 min. Aliquots of pre-cleared lysate were incubated with 10 μg of anti-p50 (sc114X), anti-p52 (sc-848X), anti-RelB (sc-226X) or rabbit IgG for 16 hrs at 4°C before the addition of 60 μl of salmon sperm DNA/Protein A agarose beads for another 4 hr incubation to collect the antibody/NFκB protein complex. The agarose beads were centrifuged at 500 × g for 5 min and washed sequentially with Low Salt, High Salt, LiCl Immune Complex Wash Buffer and TE buffer as described in the protocol provided with the beads. Cross-linked DNA-protein complexes were eluted from the beads by adding 250 μl of freshly prepared elution buffer (1% SDS, 0.1 M NaHCO₃), vortexing, then incubating the complexes at room temperature for 15 min with rotation. Elution was repeated with another 250 μl of elution buffer and both eluates were combined. The DNA-protein crosslink was reversed by incubating the above eluent or input lysate first with 40 μg/ml RNase and 200 mM NaCl at 65°C for 16 hr and then with 10 mM EDTA, 40 mM Tris-HCl and 160 μg/ml of protease K at 45°C for 1 hr. DNA was extracted with phenol-chloroform-isoamyl alcohol (25:24:1, v:v:v) and then precipitated with 1/10 volume of 3 M NaOAc (pH 5.3), 20 μg glycogen and 2.5 volumes of EtOH at −80°C overnight. DNA was collected by centrifugation at 14,000 × g for 30 min at 4°C and resuspended in 30 μl of H₂O. Input samples were diluted 50 fold before PCR. Each PCR reaction mixture contained 1 μl of immunoprecipitate or diluted input sample, 1 × PCR buffer (Roche), 200 μM dNTP (Roche) and 1 U Taq DNA polymerase (Roche) in a total volume of 50 μl. PCRs were performed by denaturing at 94°C for 20 s, annealing at 59°C (κB site) or 55°C (non-κB site) for 30 s, and extending at 72°C for 1 min for 35 cycles. Half of each PCR product was electrophoresed on a 2% agarose gel and visualized using ethidium bromide. The primers that specifically amplify a 105 bp DNA fragment containing the κB site on the rat cyclin D1 promoter were 5′-CTC CTT TTT CTC TGC CCG GCT T-3′ (sense) and 5′-CCC TCT GGA GGC TGC AGG ACT T-3′ (antisense). The primers that amplify a 146 bp DNA fragment without a κB site on the rat cyclin D1 promoter were 5′-CAA CGA AGC CAA TCG GGA AG-3′ (sense) and 5′-GAG AAA AAT AAA TCT TTG AAG-3′ (antisense).

Statistics

The MTT data were analyzed by one way ANOVA at each time point, using the Holm-Sidak method for pairwise multiple comparison. A P value of <0.05 was considered statistically significant. ³H-Thymidine incorporation between control and TNF-treated groups was analyzed using Student’s t-test.

RESULTS

Characterization of TNF-induced NFκB-DNA binding complexes in rat MEC in primary culture

Our laboratory previously reported that TNF increases the proliferation of rat MEC in primary culture associated with an increase in the DNA-binding activity of the NFκB1/p50 homodimer (10). To further understand the role of p50 in regulating the growth of MEC, we used antibodies against all five members of the NFκB family to examine the composition of consensus NFκB DNA-binding complexes in the nuclear extracts of rat MEC treated with or without 40 ng/ml TNF. We chose this TNF concentration in all of our present studies because it was shown previously to modify MEC growth and differentiation, while higher levels of TNF did not have additional effects (5). Using EMSA, two TNF-stimulated NFκB-binding complexes were detected in rat MEC nuclear extracts, a minor slower migrating band (labeled complex 1), and a major band (labeled complex 2), which migrated with higher mobility (Figure 1). The majority of complex 1, and all of complex 2 were depleted by the p50 antibody (sc114x), leading to the formation of two supershift bands. Complex 1 was also depleted by the p65 antibody, suggesting that this minor DNA-binding complex contained the p65/p50 heterodimer. Additionally, two supershift bands were detected with the p52 antibody (NCI 1495), concomitant with a slight decrease in the intensity of the upper band; together with partial depletion of the upper band with the RelB antibody, this suggests that the complexes that comprise the upper band may also include p52 and RelB. These data therefore demonstrate that while the major NFκB-DNA binding complex in rat MEC in primary culture is composed of a p50/p50 homodimer, additional complexes of p65/p50, as well as complexes containing p52 and RelB are also present and induced by TNF.

FIGURE 1.

Characterization of the TNF-induced NFκB DNA binding complexes in primary rat MEC. MEC were cultured in the absence (control, C) or presence of TNF for 10 days, and nuclear extracts prepared for EMSA. Two TNF-inducible NFκB DNA binding complexes were detected in rat MEC (arrows), a minor complex denoted as 1, and a major complex denoted as 2. In supershift analysis using the samples from the TNF-treated cells (lanes 3–7), subsequent to the incubation of nuclear extract with ³²P-labeled NFκB oligo, antibodies against the different NFκB subunits, p50 (sc-114X), p52 (NCI 1495), p65, c-Rel and RelB, were added to the reaction mixture for further incubation before loading. M, binding of nuclear extracts from TNF-treated MEC to a mutant NFκB probe. Cold, competition of nuclear extracts from TNF-treated MEC with cold consensus NFκB oligo. The asterisk indicates free probe. The results are representative of two independent experiments using this p52 antibody, and of more than three independent experiments using each of the other antibodies.

In order to understand the mechanism by which the NFκB-DNA binding activity was increased, western blot analysis was used to determine whether TNF altered the expression or nuclear translocation of p50 and its coactivator bcl3. p52 was also examined because of its close homology to p50. The expression of p50 was modestly increased by TNF in both cellular fractions on days 3 and 8, and more extensively on day 12 (Figure 2A). In addition to the 50 kDa p50 band, 1–3 bands varying from 51 to 55 kDa were seen in both cellular fractions although this heterogeneity was more pronounced in the nuclear fraction. The characteristics of these more slowly migrating proteins are not known, but they may represent phosphorylated species. Interestingly, p105, the precursor of p50, was found in the nucleus as well as the cytoplasm, although the difference in mobility (cytoplasmic p105 migrated more slowly) suggests a difference in their posttranslational modification. TNF did not have a major effect on the levels of p105, although the expression of the slower migrating form of p105 was increased in the nucleus at both days 8 and 12, and the more rapid form at day 12 in the TNF groups. The presence of p105 in the nucleus, as well as the TNF-induced increase in a putative posttranslationally modified species, raises the possibility that nuclear p50/p105 complexes may form and be biologically significant.

FIGURE 2.

Effect of TNF on the expression, localization and interaction of p50 and p52 with their coactivator bcl3 in rat MEC in primary culture. A. Expression of p50, p52 and bcl3 proteins in cytosolic (C) and nuclear extracts (N) of rat MEC cultured with or without TNF for 3, 8, or 12 days was determined by western blot. Twenty μg protein were loaded in each lane. Hsc70 was used as a loading control. B. Identification and localization of proteins bound to the NFκB consensus oligo in MEC (DAI assay). Proteins binding to the biotinylated NFκB oligonucleotide in cytosolic or nuclear extracts of MEC (200 μg protein per sample) were pulled down by streptavidin magnetic beads and analyzed by western blot. In A and B, the membranes were blotted sequentially with antibodies to bcl3, p50 (KAP-TF112) and p52 (sc-298X). Proteins of interest are indicated by arrows. The results are representative of three independent experiments.

Similar to p50, the expression of p52 and p100 was increased by TNF in MEC although this increase was most dramatic in the cytosolic fraction and only p100 was increased in the nucleus (Figure 2A). Finally, we found bcl3 in both cytoplasmic and nuclear fractions of rat MEC, and moreover, it was TNF-inducible at later time points in culture (Figure 2A). Bcl3 migrated as a heterogeneous set of 1–4 bands of 60–77 kDa with distinct patterns noted between cytoplasmic and nuclear fractions.

Since bcl3 has previously been shown to be required for transactivation of the p50 and p52 homodimers (13,37), we next investigated the effect of TNF on the interaction of p50 and p52 with bcl3 using the DAI assay (Figure 2B). This assay allows detection of proteins which bind to the NFκB oligo (e.g. p50, p52, or their precursors), as well as proteins such as bcl3 which form complexes with NFκB family members. After immunoblotting proteins bound to the biotin-labeled NFκB oligo, we found that TNF induced nuclear NFκB-DNA binding complexes containing p50, p52, p105, and/or p100. Bcl3 was found to be associated with these nuclear DNA-binding complexes. TNF also increased the DNA-binding activity of cytoplasmic p50 and p100. Together, these data show the existence of TNF-inducible DNA-binding complexes composed of bcl3 with p50 and/or p52 in this physiological MEC model system. The data also suggest the possibility that bcl3 may interact with p105 or p100 in a DNA-binding complex, although to our knowledge this has not previously been described.

NFκB1/p50 is not required for TNF-stimulated growth or morphogenesis of MEC

Our previous studies (10), together with Figures 1 and 2, suggested that p50 might play an important role in the stimulation of MEC growth by TNF. To determine whether p50 is required, we compared the effect of TNF on the growth and morphogenesis of freshly isolated MEC from p50 null and wild type mice. For each p50 null mouse used in the assays, the genotype was verified (16); we also confirmed their inability to synthesize the p50 protein. As shown in Figure 3, no p50 or p105 was detected in the whole cell lysate of p50 null mammary glands or in cultured MEC using the p50 antibody NCI 1157. Figure 3 also shows that total cellular levels of p50 and p105 in wild type mouse MEC do not change after 4 or 8 days of TNF treatment. In a separate experiment, TNF did not alter nuclear or cytoplasmic levels of p50 or p105 in wild type mouse MEC (evaluated on day 8; data not shown). Importantly, we found that TNF significantly increased the growth of MEC from both wild type and p50 null MEC to a similar extent (~2.8 fold on day 15) (Figure 4). This increase in cell number resulted, at least in part, from an increase in DNA synthesis, since in an experiment with wild type MEC, ³H-thymidine incorporation was increased 7-fold (P<0.001) by TNF (data not shown). Additionally, MEC from the wild type and p50 null groups grew from end-bud like organoids to multi-lobular alveolar colonies, and this morphological development was stimulated by TNF to a similar extent in both groups (Figure 5). However, in contrast to rat MEC (10), no ductal branching morphogenesis was observed in TNF-treated mouse MEC. From the data in Figures 4 and 5, we conclude that p50 is not required for TNF-stimulated growth or morphogenesis of MEC.

FIGURE 3.

Confirmation that mammary glands and MEC from p50 null mice do not express NFκB1/p50. NFκB1/p50 was detected by western blot using the NCI 1157 antibody in the whole cell lysates of mammary gland tissue (A) and MEC cultured with or without TNF (B). Arrows indicate the positions of p50 and its precursor p105. The same blots were stripped and reblotted with anti-Hsc 70 as a loading control. The results are representative of at least two independent experiments.

FIGURE 4.

TNF increases viable cell number of p50 null and wild type mouse MEC grown in primary culture. The viable cell number of control or TNF-treated (40 ng/ml) MECs from p50 null (KO) or wild type (WT) mice was determined by MTT assay, and is presented as optical density from that assay. The OD reading shown in the graph is the mean of triplicate samples for each time point (±SEM). TNF-stimulated growth is significantly increased on days 12 and 15 in both wild type and p50 null groups, and is significantly increased on day 8 in the p50 null group. This result is representative of two independent experiments.

FIGURE 5.

Wild type and p50 null MEC organoids underwent similar morphological development in primary culture in the absence or presence of TNF. Mouse MEC organoids grew from simple end bud-like organoids (arrowheads) to multilobular alveolar organoid colonies (arrows). This figure is representative of three independent experiments.

NFκB2/p52 and RelB may compensate for p50 in p50 null mice

In order to clarify the mechanism by which TNF stimulated the growth of MEC in the absence of p50, we used EMSA to compare the consensus NFκB DNA-binding activity in cultured MEC nuclear extracts from p50 null and wild type mice after 6 days of treatment with or without TNF. As seen in Figure 6A, two TNF-stimulated DNA-binding complexes were detected in wild type MEC; in contrast, only a single complex was detected in p50 null MEC, which migrated at the same position as the upper band (complex 1) in wild type nuclear extracts.

FIGURE 6.

Identification of NFκB DNA-binding complexes in wild type and p50 null mouse MEC. MEC were grown in primary culture with or without TNF (40 ng/ml) for six days before the nuclear extracts were harvested for EMSA. A. The DNA-binding activity of consensus NFκB oligo was compared between wild type and p50 null mice MEC. The arrows indicate two NFκB DNA-binding complexes, 1 and 2, detected in wild type MEC. Only complex 1 was detected in p50 null MEC. B. Competition and supershift analysis using the nuclear extracts of TNF-treated MEC from wild type or p50 null mice. The competition experiment was performed by pre-incubating nuclear extract (lane 1) with a 100-fold excess of unlabeled consensus (cold, lane 2) or mutant (M) (lane 3) NFκB oligo prior to the addition of the ³²P-labeled NFκB oligo. Antibodies against p65, p50 (sc-114X), p52 (sc-848X) and RelB were used for supershift analysis. A non-specific supershift band detected with the p50 antibody (sc-114X) is labeled (*) in the p50 null group (lane 5, see text for more details). C. Supershift analysis of NFκB2/p52. Two different p52 antibodies were incubated with TNF-treated (lanes 1–3) or untreated (lanes 4–6) nuclear extracts from wild type or p50 null mice in supershift analysis. This experiment is representative of two independent experiments.

Using antibodies against NFκB p65, p50 (sc-114X), p52 (sc-848X), and RelB, supershift analysis was performed to characterize the composition of the NFκB DNA-binding complexes in the nuclear extracts of TNF-treated MEC from both groups (Figure 6B). In wild type MEC, incubation with anti-p65 depleted complex 1, whereas incubation with anti-p50 decreased the intensity of complex 1 and depleted complex 2. A modest decrease in complex 1 was seen with the p52 and RelB antibodies. These data suggest that the majority of complex 1 in wild type MEC is composed of the p50/p65 heterodimer, and that complex 2 is the p50 homodimer. In contrast, in p50 null MEC nuclear extracts, complex 1, the sole NFκB DNA-binding species, was not shifted or decreased in intensity by the p65 antibody, but was completely depleted with anti-p52 (sc-848X) (lower panel of Figure 6B, lane 6). The density of the NFκB DNA-binding complex in p50 null mice was also partially decreased by the RelB antibody (Figure 6B, p50 null, lane 7). These data suggest that complex 1 in p50 null mice may contain both a p52 homodimer and a p52/RelB heterodimer.

The apparent supershift with the p50 antibody in the p50 null extracts (lower panel of Figure 6B, lane 5, asterisk) is non specific. Indeed, while undertaking this study, we found that the sc-114X p50 antibody, in addition to detecting p50 in wild type MEC, also detected a protein of 52–53 kDa in both wild type and p50 null nuclear extracts (data not shown; see also (38)). This suggests that this antibody may cross-react with p52, a supposition supported by the ability of the p52 antibody to supershift complex 1 in the p50 null mice. We used sc-114X anti-p50 in our supershift studies because among the p50 antibodies we have tested, it is the only commercially available antibody that can supershift p50 in the EMSA. It is important to note that although this antibody is not specific for p50, the absence of the complex 2 NFκB DNA-binding species in p50 null mice confirms that complex 2 in wild type MEC consists of the p50 homodimer.

To confirm our finding that p52 was present in the NFκB-DNA binding complex in p50 null MEC, we used another anti-p52 serum (NCI 1495) for supershift analysis in TNF-treated wild type and control p50 null MEC nuclear extracts (Figure 6C). The specificity of NCI 1495 to detect p52 was first tested by western blot using whole cell lysates from p52 null and wild type mouse embryonic fibroblasts, as well as liver and spleen from a p52 null mouse (data not shown). Consistent with the sc-848X-p52 antibody, incubation with the NCI 1495 p52 antibody abolished the NFκB DNA-binding complex 1 in the p50 null MEC nuclear extract with the exception that two supershift bands were detected (Figure 6C, lane 6, arrowheads). These supershift bands were also seen in the wild type MEC EMSA when probed with the NCI 1495 antibody (Figure 6C, lane 3), although not with the sc-848X antibody (Figure 6C, lane 2). Both sc848X and NCI 1495 decreased the density of complex 1 in wild type MEC. Figure 6C confirms the data of Figure 6B in that the sole NFκB DNA-binding complex detected in p50 null mouse MEC (complex 1) was primarily composed of p52. It also suggests that in addition to p50/65 and p50/p50 dimers, a p52-containing NFκB complex may also be present in wild type MEC.

In order to determine if the increased DNA-binding activity of p52 was due to increased expression of p52 protein, we used western blot to detect the level of p52 in whole cell lysates of cultured MEC from both groups treated with or without TNF. As shown in Figure 7, by immunoblotting with two different p52 antibodies, we found that TNF increased the levels of p52 protein in both p50 null and wild type MEC to a similar extent.

FIGURE 7.

TNF increases the expression of NFκB2/p52 protein in p50 null and wild type mouse MEC. Whole cell lysates were harvested from primary MEC cultured with or without TNF at the times indicated. p52 levels were detected by western blot using two different p52 antibodies as indicated. Arrows show the positions of the precursor p100 and p52. The same blot was stripped and reprobed with anti-Hsc 70 for loading control. This result is representative of two independent experiments.

Effect of TNF on NFκB-mediated expression of cyclin D1 in MEC primary culture

The TNF-stimulated growth of MEC, in association with activation of the DNA-binding activity of the p50 homodimer in wild type MEC, and the p52/RelB heterodimer in p50 null MEC, suggested that NFκB-mediated gene transcription was important in MEC growth. Since cyclin D1 is a direct transcriptional target of NFκB (39–41), we undertook a study to determine whether TNF-stimulated growth was associated with increased cyclin D1 expression, and if so, to determine which NFκB proteins were recruited to the cyclin D1 promoter in response to TNF treatment. MEC isolated from the rat mammary gland were used for these studies in order to ensure adequate sample size. We initially determined the time-dependent effect of TNF on the expression of cyclin D1 mRNA and protein in MEC in primary culture. TNF increased both cyclin D1 mRNA (Figure 8A) and protein (Figure 8B) expression, but not until after 2 days of culture (Figure 8). No significant change was detected at times less than 24 hr (data not shown).

FIGURE 8.

TNF increased the expression of cyclin D1 mRNA and protein in rat MEC. Fold induction of TNF vs. control (T/C) for each time point is shown under each figure. A. Cyclin D1 mRNA (Northern blot). GAPDH was used as the RNA loading control on the same blot. B. Cyclin D1 protein (Western blot). Hsc 70 was used as the protein loading control on the same blot. This result is representative of at least two independent experiments.

To determine if the up-regulation of cyclin D1 by TNF is mediated by activation of NFκB, the DNA-binding activity of NFκB to the κB site on the rat cyclin D1 promoter was examined. This NFκB binding site is conserved among human, mouse and rat cyclin D1 promoters (40,42,43). Figure 9 shows that two NFκB DNA-binding complexes were detected on the κB site of the cyclin D1 promoter, and both were TNF-inducible after 4 and 8 days of culture (Figure 9, lanes 1–4). Supershift analysis performed on the nuclear extract from TNF-treated MEC showed that the major complex formed with the cyclin D1 oligo was the p50/p50 homodimer (complex 2) (Figure 9, lane 7). The minor complex (complex 1) appeared to consist of the p50/p65, p50/RelB and p52/RelB heterodimers.

FIGURE 9.

TNF induced the DNA-binding activity of the NFκB oligo containing the κB site on the cyclin D1 promoter in rat MEC. EMSA was carried out with the nuclear extract from rat MEC treated with or without TNF for 4 or 8 days. Arrows point to two NFκB DNA-binding complexes on the κB site of the cyclin D1 promoter. Competition (lane 5) and supershift analysis (lane 6–9) were performed with nuclear extract from TNF-treated MEC (Day 8). The supershift band detected with the p50 antibody (sc-114X) is indicated by an arrowhead. The p52 antibody used in lane 8 was sc-848X.

Because the major NFκB protein bound to the cyclin D1 κB site-containing oligo is p50, we next asked whether TNF stimulated cyclin D1 expression in p50 null mouse MEC. This was found to be the case. Thus, as seen in Figure 10, cyclin D1 protein levels were increased in whole cell lysates prepared from both wild type and p50 null MEC treated with TNF, demonstrating that p50 is not required for TNF induction of cyclin D1.

FIGURE 10.

TNF induces expression of cyclin D1 protein in MEC from p50 null and wild type mice. MEC isolated from p50 null or wild type mice were cultured with or without TNF at the times indicated. Cyclin D1 protein was detected in whole cell lysates of MEC by western blot. Fold induction of TNF vs. control (T/C) for each time point is shown under the figure. The loading of protein was normalized by the amount of Hsc 70 on the same blot. This result is representative of at least two independent experiments.

To further understand why p50 is not required in TNF-up-regulated cyclin D1 transcription and to determine if other NFκB proteins are involved, the ChIP assay was used to examine the effect of TNF on the occupancy of different subunits of NFκB on the κB site of the rat cyclin D1 promoter in intact rat MEC. Figure 11 shows that although the κB site was occupied by NFκB1/p50 on days 3 and 7 of MEC primary culture, the addition of TNF did not change the amount of p50 bound to the κB site. A similar result was seen at earlier time points including 15 and 48 hr (data not shown). We next determined whether p52, RelB or p65 bound to the κB site. Similar to p50, p65 also bound to the cyclin D1 κB site, and its levels were not changed by TNF in rat MEC (Figure 11). In contrast, TNF induced the binding of p52 to the same κB site at both 3 and 7 days of culture. Binding of RelB was also detected on day 7 of TNF treatment (day 3 was not tested; Figure 11). The specificity of the CHIP assay was monitored in immunoprecipitated samples using two negative controls, rabbit IgG and a non-κB site DNA fragment on the rat cyclin D1 promoter as shown in Figure 11.

FIGURE 11.

The binding of p52 and RelB to the cyclin D1 promoter, but not that of p50 or p65, was induced by TNF in rat MEC in primary culture. Rat MEC were cultured with or without TNF for 3 or 7 days before crosslinking. The binding of NFκB p50, p52, RelB and p65 to the κB site on the rat cyclin D1 promoter was determined by the CHIP assay. Antibodies against p50 or p52 were sc-114X or sc-848X, respectively. This result is representative of two separate experiments.

DISCUSSION

TNF is one of the important cytokines that regulate the development of the mammary gland (1). In a physiologically relevant primary mammary epithelial cell model (2–4), TNF was shown to increase proliferation and branching morphogenesis, and inhibit the functional differentiation of MEC (1,5–9). Following our previous report that TNF-stimulated proliferation of MEC is associated with the activation of the NFκB p50 homodimer (10), we extend this in the present study to demonstrate that TNF induces nuclear DNA-binding complexes of both NFκB1/p50 and NFκB2/p52, and that these active complexes are associated with their coactivator bcl3. TNF-stimulated growth of MEC was also shown to correlate with the up-regulation of cyclin D1 and the DNA-binding of NFκB on the cyclin D1 promoter. Importantly, however, in contrast to our initial hypothesis, p50 was not required for either TNF-stimulated growth or up-regulation of cyclin D1 in MEC. Instead, our data strongly suggest that p52 and RelB compensate for p50 in the growth of MEC when p50 is absent, and that in wild type MEC the TNF-stimulated up-regulation of cyclin D1 is mediated through the transactivation of both p50 and p52/RelB.

TNF stimulates DNA-binding activity of p50 and p52, and their interaction with bcl3 in MEC

Previous studies demonstrated that the activation and translocation of p50-containing NFκB dimers is regulated through cytokines such as TNF and activation of the IKK complex (canonical pathway), while p52-containing NFκB dimers are regulated through ligands for receptors such as CD40 and activation of NIK and IKKα (non-canonical pathway) (15). In MEC, TNF modestly increased the levels of nuclear NFκB1/p50 protein and induced its binding to 20–22 base pair oligonucleotides containing a consensus NFκB site or the κB site on the cyclin D1 promoter. TNF also increased expression of p52 and its precursor p100 in MEC cytoplasmic fractions, as well as the expression of p100 in the nucleus. Unexpectedly, however, nuclear p52 levels were not increased by TNF, even though TNF increased binding of p52 to the biotinylated NFκB consensus oligonucleotide in the cell-free DAI assay, and to the κB site of the cyclin D1 promoter in the intact cell ChIP assay. This apparent discrepancy may reflect an activation of the DNA-binding activity of p52 by TNF, although the mechanism of this effect remains to be determined.

Our DAI data also demonstrated that TNF-stimulated the DNA-binding activity of both p105 and p100. The significance of this observation is not known, however binding complexes containing these precursors may exclude certain NFκB dimers from κB binding sites, as has been shown for RelB (44,45) and/or affect the transcriptional activity of the complexes. With respect to the former, it is tempting to speculate that the marked induction of p100 by TNF may play a role in the virtual exclusion of the canonical p65/p50 heterodimer from κB binding sites in rat MEC. Dejardin et al (29) previously demonstrated a physical interaction between p65 and p100. Finally, the TNF induction of p100 and its processing to p52 suggests a role for the non-canonical pathway of NFκB activation in response to TNF treatment of MEC, as has been reported in other cell types (44,45).

Bcl3 was previously shown to specifically interact with and transactivate p50 or p52 homodimers in the nucleus (13,14). Its expression has been documented in breast tumors but not in adjacent tissues (24), although its presence is controversial in breast cancer cell lines (24,46). Importantly, our data show for the first time that bcl3 is expressed in normal MEC and that it is localized to both the cytoplasm and the nucleus. Moreover, TNF-inducible DNA-binding complexes of bcl3 with activated p50 and p52 were found in the nuclei of proliferating MEC. Bcl3 was detected as heterogeneous bands in MEC nuclear extracts, suggesting the presence of different phosphorylated forms of bcl3 as shown in human HepG2 cells (47). Our data suggest that bcl3 acts as a coactivator of p50 and p52 NFκB complexes in MEC although the overall transcriptional effects may depend on the relative ratio of phosphorylated bcl3 to p52 (48).

NFκB1/p50 is not required for TNF-stimulated growth of MEC nor for NFκB-mediated cyclin D1 up-regulation: roles for p52 and RelB

The studies reported herein do not support our initial hypothesis that NFκB1/p50 is required for TNF-stimulated growth or up-regulation of cyclin D1 in MEC primary culture. Using MEC organoids from p50 null and wild type mice, we found that TNF stimulated their growth and morphological development to a similar extent. Moreover, cyclin D1 expression was increased by TNF in both p50 null and wild type MEC.

Although a number of genes undoubtedly contribute to the TNF-induced growth response, we chose cyclin D1 for investigation because of its regulation by NFκB and its critical role in mammary gland development (49,50). NFκB-mediated cyclin D1 transcription contributes to cell cycle progression in mouse mammary carcinoma cells (40) and immortalized human breast epithelial cells (41). In primary cultured MEC, we showed that TNF induces the expression of cyclin D1 at both mRNA and protein levels suggesting the importance of cyclin D1 in regulating TNF-stimulated proliferation of primary MEC. The role of NFκB in cyclin D1 expression was determined using EMSA cell-free and ChIP intact cell assays to examine the effect of TNF-induced occupancy of the κB site on the cyclin D1 promoter by NFκB. In previous studies using nuclear extracts of serum-stimulated NIH 3T3 cells, both the p50/p65 heterodimer and the p50 homodimer were detected on the κB site of this promoter (39,42). Using EMSA, we detected TNF-stimulated binding of both dimers to this promoter in MEC nuclear extracts, although significantly more p50 homodimer bound than the p65/p50 heterodimer. With the more physiological ChIP assay, we discovered that the p50 occupancy of the cyclin D1 κB site was constitutive at all time points (15, 48, 72 hr and day 7) tested, and was not changed after TNF treatment. In contrast, TNF markedly stimulated the binding of both p52 and RelB to the κB site on the promoter. Together, these data suggest that: 1) cyclin D1 is important in the proliferation of MEC in primary culture; 2) activated transcription factors within a structurally-related family such as p50 and p52 can recognize identical consensus sites, as has been reported (51); 3) the up-regulation of cyclin D1 is mediated through NFκB in MEC primary culture.

A number of reports have documented an important role for p52 in regulating cyclin D1. In human breast epithelial cells, a p52 homodimer/bcl3 complex was found to bind to the NFκB site and directly activate the cyclin D1 promoter (41). In untransformed murine mammary cells, RelB/p52 induced cyclin D1 promoter activity (52). Moreover, a RelB/p52 complex rescued a delay in mammary gland ductal branching and an induction of cyclin D1 during early pregnancy, in transgenic mice expressing the IκBα super-repressor (53). These same investigators also reported that the anchorage-independent growth of breast tumor cells was blocked through inhibition of RelB and the accompanying suppression of cyclin D1 (53). Our observation that TNF induced the occupancy of p52 and RelB on the cyclin D1 promoter, together with the TNF stimulated increase in the expression of p52 and cyclin D1 in p50 null mice, suggests that p52 and RelB play more important roles than p50 in cyclin D1 transcription in MEC. Moreover, the insensitivity of p52/RelB to IκBα may allow sustained activation of NFκB (51) and could account for the prolonged effect of TNF on NFκB-mediated transcription in our MEC model.

Compensation of p52 for p50

NFκB1/p50 and NFκB2/p52 are distinct NFκB members, but are similar in structure, lack transactivation domains and undergo proteolysis from precursors. They have distinct functions in organogenesis and inflammation, but are redundant for osteoclastogenesis (18) and lymph node formation (54). Functional compensation within the NFκB family has been demonstrated in various NFκB null cell lines (55). In particular, Hoffmann and coworkers found that p52 could compensate for p50 in p50 null fibroblasts while c-Rel could compensate for p65 (55). In addition, TNF-induced expression of some genes, including RANTES and IP-10 was not affected by p50 or p52 deficiency but was only defective in cells lacking both proteins. Similarly, in our present study we found that in p50 null MEC, TNF increased expression of p52, and induced an NFκB DNA-binding complex containing p52. However, in contrast to NFκB1/p50 null fibroblasts in which p52 formed a DNA-binding complex with p65, RelB complexed with p52 in p50 null MEC. These data suggest that the compensation between NFκB proteins is a common but cell type specific event. This needs to be taken into consideration when studying the function of an individual NFκB-regulated gene or when developing NFκB-targeted therapeutics.

Future studies will be necessary to determine whether p50 can compensate for p52 in p52 null MEC, and/or whether TNF stimulated growth of MEC and the up-regulation of cyclin D1 are prevented in MEC which are deficient in both p50 and p52. These questions are not addressable in breast cell lines, since the majority of human and rodent cell breast cell lines that we have examined are not growth stimulated by TNF, and for most of them, the major TNF-inducible complex is the p65/p50 heterodimer rather than the NFκB dimers that we and others have identified in tissue samples and primary culture.

In summary, we conclude that NFκB1/p50 protein is not required for TNF-stimulated growth of MEC nor the up-regulation of cyclin D1. However, the induction by TNF of its DNA-binding activity in MEC from both wild type rats and mice, as well as the TNF-induced formation of a complex of p50 with its transactivator bcl3, suggest that it may contribute to TNF-stimulated growth, even though it is dispensable. When p50 is absent, p52 and RelB appear to compensate. The present study provides a basis for understanding the mechanism by which TNF stimulates the growth of malignant breast epithelial cells. Endogenous TNF within the tumor environment may enhance tumor development and spread, at least in a subset of patients. With respect to this consideration, studies have been initiated to examine the efficacy of therapies that target TNF in women with breast cancer (56). Understanding the mechanism by which TNF stimulates the growth of mammary tumor cells may help in the development of a new treatment for breast cancer.