MCF7 cells were transfected with either the empty vector pcDNA3.2 (Invitrogen, Carlsbad, CA, USA) or a plasmid containing the BP1 open reading frame under control of the cytomegalovirus promoter. Plasmid-containing cell lines were selected in 800 μg/ml G418. Cells were maintained in RPMI 1640 supplemented with 10% fetal bovine serum, penicillin/streptomycin, 500 μg/ml G418, and 2 mM glutamine. MTT assays were performed to measure cell viability. Cells were seeded in triplicate in 96-well plates, and were cultured in normal growth media containing 20 ng/ml human TNFα (Sigma-Aldrich, St Louis, MO, USA) or were left untreated. After 72 hours, samples were incubated with 5 mg/ml MTT at 37°C for 4 hours. Formazan crystals were dissolved in dimethylsulfoxide (Sigma-Aldrich). Samples were read at 570 nm with a Versamax microplate reader (Molecular Devices, Sunnyvale, CA, USA).

Cell lines were cultured at 3 × 10⁵ cells/well in six-well plates, and were cultured in normal growth media containing 20 ng/ml TNFα for 18 hours or were left untreated. Cells were labeled with a 1:100 dilution of Annexin V–FITC conjugate and 5 μg/ml propidium iodide according to the manufacturer's instructions (Trevigen, Gaithersburg, MD, USA). Each sample was analyzed using a Nikon Eclipse TE300 inverted epifluorescence microscope (Nikon Instruments Inc, Melville, NY, USA) with filter sets for FITC and TRITC. Early apoptotic cells were distinguished by the presence of green staining in the plasma membrane and the absence of red nuclear staining.

Complementary sequences spanning 2,555 to 2,513 nucleotides upstream of the bcl-2 ATG start site were annealed and 5'-end-labeled with γ-³²P-ATP using T4 kinase (Invitrogen). The Wheat Germ Coupled Transcription/Translation kit (Promega, Madison, WI, USA) was used to generate BP1 protein from the plasmid pGEM7 containing the BP1 open reading frame. Unlabeled competitor oligonucleotides were added at 500× or 1,000× molar excess to binding reactions. For supershift analyses, binding reactions included BP1 antibody 8. The following sequences were used as probes and competitors: bcl-2, 5'-ACGGTGGGCCTGAAAGTTACTATATGGAAGTCCTCATCGTGTA-3'; mutant bcl-2, 5'-ACGGTGGGCCTGAAAGTTAGCTCGACGAAGTCCTCATCGTGTA-3'; negative control, 5'-TCTTAGAGGGAGGGCTGAGGGTTTGAAGTCCAACTCCTAAGCC-3'.

A construct containing the bcl-2 P1 promoter region linked to a luciferase reporter gene (LB170) was a kind gift from Dr Linda Boxer (Stanford University, Stanford, CA, USA). Cells were transfected with 2.5 μg LB170 and 1 μg plasmid encoding β-galactosidase, using Fugene 6 Transfection Reagent (Roche, Indianapolis, IN, USA) at a 3:2 ratio of Fugene:DNA according to the manufacturer's instructions. Forty-eight hours post transfection, β-galactosidase activity was measured using the Beta-Galactosidase Enzyme Assay System (Promega), and the luciferase reporter activity was assayed using the Luciferase Assay System (Promega). Luciferase activity output was given in relative light units. The relative light unit value for each sample was divided by the β-galactosidase activity to normalize differences in transfection efficiencies. Each transfection was performed three times in duplicate.

Using LB170 as a template, mutation of the BP1 binding site was performed using the Quik Change II XL Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA, USA). HPLC-purified complementary primers (Invitrogen) were designed to delete a seven-nucleotide region of the BP1 consensus binding site (underlined): 5'-'GGTGGGCCTGAAAGT TA

CTATATG

Total RNA was extracted using Trizol Reagent (Invitrogen) according to the manufacturer's instructions. Reverse transcription of mRNA was performed using the iScript cDNA Synthesis Kit (Biorad, Hercules, CA, USA). TaqMan analyses of BP1 and 18S were performed using QPCR Master Mix Plus reagent (Eurogentec, San Diego, CA, USA). For SYBR Green analyses of bcl-2, the reactions were performed using iTaq SYBR Green Supermix with ROX (Biorad). The cycling conditions were as follows using the ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA, USA): 50°C for 2 minutes, then 95°C for 10 minutes, followed by 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. SYBR Green analyses also included a dissociation protocol.

The ABI Prism software was used to perform an automatic cycle threshold analysis and to generate a standard curve for extrapolation of the sample data. Mean values of each gene were normalized to the corresponding mean value for 18S. The following sequences were used for primers and probes: 18S primers, 5'-GCCGCTAGAGGTGAAATTCTTG-3' and 5'-CATT CTTGGCAAATGCTTTCG-3'; 18S probe, 5'-ACCGGCGCAAGACGGACCAG-3'; BP1 primers, 5'-CCTCCCCCAATTTGTCCTACTC-3' and 5'-GGTTGCTGGCAGGACAGGTA-3'; BP1 probe, 5'-AGCCAGCGAACCCCGGAGACTC-3'; bcl-2 primers, 5'-TGGGATGCCTTTGTGGAACT-3' and 5'-GAGACAGCCAGGAGAAATCAAAC-3'.

Cell lysates were prepared in ice-cold RIPA lysis buffer (50 mM Tris, pH 7.5, 2 mM ethylenediamine teraacetic acid, 100 mM NaCl, 1% NP-40) containing 1× Complete Mini protease inhibitor cocktail (Roche). Proteins were separated by SDS-PAGE and were transferred to a polyvinylidene difluoride membrane. Blots were probed overnight at 4°C with rabbit anti-BP1 (Novus Biologicals, Littleton, CO, USA) at a 1:5,000 dilution, or with a 1:1,000 dilution of mouse anticaspase-7 and anticaspase-8 antibody, rabbit anticaspase-9 and anti-PARP (anti-Poly(ADP-Ribose) Polymerase) antibody (Cell Signaling, Danvers, MA, USA) or mouse anti-Bcl-2 antibody (Santa Cruz, Santa Cruz, CA, USA). After washing, blots were incubated with either horseradish peroxidase-linked goat anti-mouse (1:500 dilution) or donkey anti-rabbit secondary antibodies (1:15,000 dilution). Signals were detected using SuperSignal West Dura Extended Duration Substrate (Pierce, Rockford, IL, USA). Relative band intensities were quantitated using the Kodak Image Station 2000 MM and the Kodak ID software (version 3.6.4; Scientific Imaging System, Eastman Kodak Co., Rochester, NY, USA) and by standardizing protein levels against β-actin.

Statistical tests comparing mean levels were performed with SAS software based on a priori analysis of variance contrasts. Each replicate was treated as an independent observation. Except where noted, contrasts involving MCF7/EV cells were based on averaging across EV1 and EV2. Luciferase values were log-transformed and the percentage of positive cells stained with Annexin V was arcsine-transformed for significance testing. Results are declared significant at α = 0.02, two-sided.

Three MCF7 cell lines were generated that stably express increased levels of BP1 protein (MCF7/BP1-1, MCF7/BP1-2, and MCF7/BP1-4), as well as two control cell lines containing the empty vector (MCF7/EV1 and MCF7/EV2) (Figure 1a). We first compared the viability of MCF7/EV and MCF7/BP1 cell lines that were grown in the presence or absence of TNFα. As shown in Figure 1b, an average of 43% of MCF7/EV cells survived 3 days post TNFα treatment, whereas all three BP1-overexpressing cell lines displayed an approximately twofold increase in viability (74%, 90%, and 80% for BP1-1, BP1-2, and BP1-4, respectively; P < 0.0001 for each comparison). Furthermore, MCF7/BP1 cells exposed to TNFα showed a twofold to threefold decrease in Annexin V binding compared with MCF7/EV cell lines (Figure 1c, P < 0.0001 for all three overexpressing cell lines), indicating that increased BP1 expression decreases the ability of MCF7 cells to undergo apoptosis.

We then examined whether constitutive BP1 expression affected TNFα-mediated cell death through modulation of caspase pathways. Upstream initiator caspase-8 and caspase-9, as well as the downstream effector caspase-7 and its substrate PARP, were analyzed by Western blot analysis in MCF7/EV and MCF7/BP1 cell lines treated with TNFα for various times (Figure 2a). MCF7 cells are deficient in caspase-3 due to a genomic deletion in exon 3 17, so this caspase was not examined. Processed fragments of each caspase are readily apparent after 12 and 24 hours of exposure to TNFα. Across each cell line, processing of PARP is also seen by 12 hours, with the amount of cleaved protein accumulating through 24 hours. In each MCF7/BP1 cell line, however, there is a clear reduction in cleavage of every caspase as well as of PARP. Analyses of band intensities of each fragment revealed a ≥50% decrease in caspase and PARP cleavage in cells overexpressing BP1, relative to the levels of cleaved products in MCF7/EV cells. Strikingly, untreated MCF7/BP1 cells showed a threefold to fourfold increase in levels of full-length PARP relative to MCF7/EV cells (Figure 2b). In addition, MCF7/BP1 cells show a 1.6-fold to 2.0-fold downregulation of procaspase-8 (Figure 2c), indicating that BP1 may affect the early stages of apoptosis. Together, our findings demonstrate a role for BP1 in caspase-dependent pathways of TNFα-mediated cell death.

We next sought to define transcriptional targets of BP1 that might explain why its overexpression results in increased cell viability in the presence of TNFα. bcl-2, a well-established antiapoptotic oncogene, is often associated with resistance to various cell-death-inducing agents 18. The bcl-2 gene contains two promoters: P1, located 1,386 to 1,423 bp upstream of the translational start site; and P2, located 1.3 kb downstream of P1 19. The sequence 5'-TACTATATG-3' matches a consensus binding site for BP1 protein 8 and is located upstream of the P1 promoter at -2539 bp relative to the ATG translational start site.

An electrophoretic mobility shift assay was used to demonstrate that BP1 protein can specifically bind to a dsDNA oligonucleotide probe containing this site (Figure 3a). A shifted band was observed in the presence of in vitro transcribed and translated BP1 protein (lane 2), while a faint band was observed at this location when wheatgerm extract alone was mixed with the bcl-2 probe (lane 1). Specificity of the interaction was evidenced by the loss of the shifted band upon addition of 500× or 1,000× molar excess of competitor DNA of the same sequence as the bcl-2 probe (lanes 3 and 4). Addition of excess negative control DNA that lacks a BP1 binding site did not reduce the intensity of the band (lanes 5 and 6). In the presence of anti-BP1 antibody we observed both a decrease in the shifted band as well as the appearance of a supershifted band (arrow, lanes 7 and 8), verifying that BP1 protein bound to the bcl-2 probe DNA. These data indicate that the bcl-2 gene is a potential target for regulation by BP1.

In support of this finding, a comparison of bcl-2 expression levels in MCF7/EV and MCF7/BP1 cells by western blot analysis and by real-time PCR revealed a twofold increase in both bcl-2 protein (Figure 3b) and mRNA (Figure 3c, black bars; P < 0.0001 comparing the average of untreated EV1 and EV2 with BP1-1, BP1-2, and BP1-4).

Constitutive expression of bcl-2 abrogates cell death in MCF7 cells exposed to TNFα 2021. To examine whether regulation of bcl-2 by BP1 is associated with the observed increase in MCF7/BP1 cell viability, bcl-2 mRNA expression was analyzed in TNFα-treated cells. Although bcl-2 mRNA was downregulated by TNFα in MCF7/EV cells (Figure 3c, gray bars; P = 0.004 and P < 0.0001 for EV1 and EV2, respectively), BP1-overexpressing cells showed no significant change in bcl-2 mRNA after treatment. Consistent with these data, Bcl-2 protein levels are not reduced by TNFα treatment, in contrast to the empty vector control (Figure 3d).

We next determined whether increased levels of bcl-2 expression in MCF7/BP1 cells could be attributed to direct regulation of the bcl-2 gene by BP1 protein. A schematic diagram of the promoter region of bcl-2 is shown in Figure 4a. MCF7/EV and MCF7/BP1 cell lines were transfected with LB170 (a gift from Dr Linda Boxer, Stanford University), a construct containing the bcl-2 P1 promoter region and the 5'-flanking sequence 22, including the BP1 binding site, linked to the luciferase reporter gene. MCF7/BP1-1 and MCF7/BP1-4 consistently showed a fivefold activation of the P1 promoter, whereas MCF7/BP1-2 showed up to an 11-fold increase, compared with levels seen in MCF7/EV control cells (Figure 4b, black bars; P < 0.0001 for all three overexpressing cell lines). These results show that BP1 overexpression increased transcriptional activation through the bcl-2 promoter. The results do not, however, distinguish between a direct effect, caused by binding of BP1 protein to the promoter, and an indirect effect by BP1, due to regulation of other factors that bind and activate transcription of bcl-2. Site-directed mutagenesis and deletion of the BP1 consensus binding site were carried out to differentiate these possibilities.

Using the LB170 construct as a template, a two-step site-directed mutagenesis procedure was performed. First, seven nucleotides of the nine-nucleotide sequence in the BP1 binding site were deleted to generate delLB170, followed by insertion of the mutated sequence, described in 23, to create mutLB170 (Figure 4a). An electrophoretic mobility shift assay was performed to determine whether this mutation could inhibit binding of BP1 to bcl-2 (Figure 4c). As before, BP1 protein (WG/BP1) bound to the bcl-2 probe, as indicated by the shifted band (lane 2, arrow). No protein binding to the bcl-2 probe was observed at this location using the wheatgerm extract (lane 1). Competition with 500× and 1,000× molar excess of unlabeled probe DNA (bcl-2) resulted in the loss of the shifted band signal (lanes 3 and 4). If excess competitor DNA containing a seven-nucleotide mutation of the BP1 binding site was added (mbcl-2), however, little competition for binding was observed (lanes 5 and 6). A negative control DNA also did not compete for binding (lanes 7 and 8). This mutation is thus sufficient to disrupt binding of BP1 protein to DNA.

MCF7/EV and MCF7/BP1 cell lines were then transiently transfected with the wild-type LB170, delLB170, or mutLB170. Notably, deletion of the BP1 binding site resulted in an average 45% to 51% decrease in bcl-2 promoter activation across all cell lines (Figure 4b, grey bars; P < 0.05). Mutation of this site caused an average 37% to 49% reduction in activation of the bcl-2 promoter, which was statistically significant for BP1-1 (white bars, P = 0.02) but not for BP1-2 or BP1-4, perhaps due to residual BP1 binding to the mutant site (Figure 4c). We thus conclude that BP1 protein can bind to the bcl-2 promoter and directly contribute to activation of its expression in MCF7 cells.