The esophageal cancer cell line KYSE510 was grown in RPMI-1640 medium (Bioroc, China) supplemented with 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin in 5% CO₂ at 37°C. The human embryonic kidney cell line HEK293 and the esophageal cancer cell line EC0156 were grown in Minimal Essential Medium (Bioroc) supplemented with 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin in 5% CO₂ at 37°C.

Genomic DNA was extracted from KYSE510 cells using the QIAamp DNA Blood Mini kit (Qiagen, Germany). The promoter fragment was synthesized with a TaqDNA polymerase mixture (BioTeke, China). Thermal cycling conditions included activation of the DNA polymerase at 94°C for 5 min followed by 30 cycles at 94°C for 30 sec, 55–65°C for 30 sec, and 72°C for 30 sec. The specific oligonucleotide primers used are shown in Table I.

The amplified promoter fragments were cloned into the pGL3-Basic vector (Promega). The various pGL3-Basic vectors were then digested with XhoI and HindIII (Takara, Japan). The promoter fragments were purified using the Wizard SV Gel and PCR clean up system (Promega), and subsequently ligated into a promoter less pGL3-Basic luciferase reporter vector. To ensure the fidelity of the cloned promoter fragments, all final constructs were sequenced using the vector-specific primers RVprimer 3: 5′-CTAGCAAAATAGGCTGTCCC-3′ and RVprimer 4: 5′-GACGATAGTCATGCCCCGCG-3′.

For the dual lucif-erase reporter assay, KYSE510 cells were seeded in a 6-well plate at a density of 2×10⁵ cells per well for at least 20 h prior to transfection. The constructed plasmids and the Renilla luciferase internal control plasmid (pRL-TK) were transfected into the cells using Lipofectamine (Invitrogen). After 24 h, the cells were treated with the lysis buffer from the Dual-Luciferase Reporter Assay System (Promega). The signals were measured using an automatic microplate reader (Synergy H1, BioTek). For knockdown of the transcriptional activator protein PURα, the cells were transfected with siRNAs (5′-CCACCUAUCGCAACUCCAUTT-3′ and 5′-AUGGAGU UGCGAUAGGUGGTT-3′) for 24 h prior to transfection with the constructed promoter and control plasmids for 24 h. The sequences negative control siRNAs are 5′-UUCUCCGAA CGUGUCACGUTT-3′ and 5′-ACGUGACACGUUCGGA GAATT-3′.

Nuclear extracts were obtained from KYSE510 cells using the ProteoExtract Subcellular Proteome Extraction kit (Calbiochem, EMD Biosciences Inc., Germany). The protein concentration of the nuclear extract was determined by Bradford method. The primers used to amplify the promoter and non-promoter sequence DNA fragments were labeled with biotin at the 5′ terminus. Streptavidin magnetic beads (Invitrogen) were washed three times with phosphate-buffered saline (PBS) before use. For affinity purification, 3 μg of each biotin-labeled DNA fragment was incubated with 30 μl of the magnetic-bead slurry for 20 min. Unbound DNA fragments were removed, and 500 μg of nuclear protein extracts was added to the streptavidin bead-biotin-labeled DNA fragments and incubated at 4°C overnight. The non-promoter control DNA fragment was used to decrease the abundance of non-specific DNA-binding proteins, such as those that bind histones, in the nuclear extracts. The bead-DNA-protein complex was washed with TBS (50 mM Tris, 300 mM NaCl) three times, and the proteins were eluted using 2% SDS. The eluted proteins were subjected to SDS-PAGE, and visualized by silver staining. Protein bands were excised and identified by in-gel trypsin digestion with subsequent analysis by MS (Q Exactive Orbitrap, Thermo Scientific). Mascot version 2.3.01 (Matrix Science Inc.) was used to analyze the data and search the databases.

The elution products from affinity purification or cell lysates were denatured in SDS-PAGE sample buffer containing 0.5 M Tris-HCl (pH 6.8), 2% SDS, 10% DTT, 10% glycerol and 0.01% bromophenol blue, boiled for 5 min, and then analyzed by SDS-PAGE followed by transfer to a PVDF membrane (Millipore, Germany). Each membrane was blocked for 1 h in PBS containing 10% nonfat milk. After blocking, each membrane was incubated overnight with rabbit polyclonal anti-E2F-1, mouse monoclonal anti-PURα, or rabbit polyclonal anti-RNA polymerase II (Pol II) (Santa Cruz Biotechnology Inc.). After washing with TBS containing 20% (w/v) Tween-20, each membrane was incubated with horseradish peroxidase-conjugated secondary antibody for 2 h and visualized with the SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific).

KYSE510 cells cultured in 10-cm dishes to 90% confluency were washed with ice-cold PBS and lysed for 30 min on ice in lysis buffer containing 1% (w/v) Triton X-100, 0.15 M NaCl, and 30 mM Tris-HCl (pH 7.5) with protease inhibitors (Roche, Germany). Lysates were sonicated and centrifuged at 10,000 × g at 4°C for 15 min. The protein concentration was determined by Bradford assay. For immuno precipitation, 1 mg of the resulting extract was incubated at 4°C overnight with anti-PURα or anti-E2F-1 and Dynabeads Protein G (Invitrogen). Immunoprecipitates were washed three times with lysis buffer and the beads were directly boiled in 1% SDS-PAGE loading buffer.

For immuno fluorescence, KYSE510 cells were fixed in 10% (w/v) paraformaldehyde on poly-L-lysine-coated slides for 30 min at room temperature and washed three times with PBS (pH 7.4). The cells were blocked with PBS (pH 7.5) supplemented with 1% (w/v) BSA and 0.1% (w/v) Triton X-100 for 30 min at room temperature. Washed cells were incubated for 30 min at room temperature with primary rabbit anti-human E2F-1 and mouse anti-human PURα (Santa Cruz Biotechnology Inc.). Then the cells were incubated in the dark for 60 min with Alexa Fluor 488-conjugated goat anti-rabbit and Alexa Fluor 594-conjugated goat anti-mouse (Life Technologies). Cells were stained with DAPI and examined using fluorescence confocal microscopy (Leica Tcs SP2, Germany).

The total mRNA was extracted using TRIzol (Invitrogen). After the quality control was examined, mRNA was transformed to cDNA by reverse transcriptase kit (Tiangen). The quantitative PCR was completed using SYBR Green Master (ROX) (Roche), and the system included SYBR Green Master (ROX) (2X) 10.0 μl, PCR forward primer (10 μM) 0.6 μl, PCR reverse primer (10 μM) 0.6 μl, Template cDNA 2.0 μl, ddH₂O ≤20.0 μl. The reaction was conducted under ABI PRISM^® 7500. The quality control and Ct values of the reaction were analyzed using SDS software.

ENCODE is a DNA elements encyclopaedia in human cell lines (26). We mainly used the ChIP-seq, histone methylation and DNA methylation data. The human cell lines represented in Fig. 1 from ENCODE include GM12878 (lymphoblastoid cells), H1-hESC (embryonic stem cells), K562 (bone marrow), HeLa S3 (cervix adenocarcinoma epithelial cells), HEP G2 (hepatocellular carcinoma cells), HUVEC (umbilical vein endothelial cells), A549 (lung carcinoma cells), IMR90 (lung fibroblasts), MCF-7 (breast cancer epithelial cells), and HESC (embryonic stem cells).

The data presented are the mean ± standard deviation. To examine differences between two groups in the luciferase reporter assay, t-tests were applied using SPSS version 17.0 (IBM software). Mann-Whitney U test was used to compare the mRNA expression of two groups in the RT-PCR experiments. P-values <0.05 were considered statistically significant.

Analysis of data from ChIP coupled with deep sequencing (ChIP-seq) in the Human Encyclopedia of DNA elements (ENCODE) database revealed a region on human chromosome 2q33.1 that contained binding sites for numerous transcription factors and thus might be involved in transcriptional regulation of CASP8. This region is located within chr2: 202,122,236 to 202,123,227 and 25 kb downstream of the previously described CASP8 promoter. Transcript of caspase-8 isoform G is adjacent to this region. To examine whether this region has promoter characteristics, we analyzed this fragment (1547 bp, termed MAX) using three promoter prediction programs; four potential promoter sequences were identified (Fig. 1A). To confirm that the fragment is transcriptionally active, we introduced MAX into a promoter-deficit luciferase reporter vector (pGL3-Basic). When transfected into KYSE510 cells, the pGL3-MAX construct resulted in considerably higher luciferase activity than the pGL3-Basic vector alone (P=0.0022, Fig. 1B).

To identify the core promoter region within the 1547-bp MAX fragment, we first constructed reporter vectors containing three different imbricating truncations of MAX that were transfected into KYSE510 cells for the luciferase activity assay (Fig. 2A). The construct containing the 5′-terminus of MAX (pGL3-M5N3) retained luciferase activity comparable to MAX, whereas constructs containing 3′-terminal fragments (pGL3-N6 and pGL3-M3N5) led to statistically significant decreases in transcriptional activity (76.2%, P=0.0004, and 74.8%, P=0.0004, respectively; Fig. 2B). These results suggested that the region contained in pGL3-M5N3 encompassed the core of this novel CASP8 promoter. Data obtained with KYSE510 cells was validated in HEK293 cells, in which we observed similar results for pGL3-M5N3, pGL3-N6 and pGL3-M3N5 (Fig. 2C).

We also examined potential epigenetic modifications of these three truncated fragments by analyzing these regions in the ENCODE database. Histone modifications, including acetylation of lysine 27 in histone H3 and trimethylation of lysine 4 in histone H3, are indicative of actively transcribed promoters. There was significant enrichment of these two kinds of histone modifications within fragment M5N3 compared with fragments M3N5 and N6 (Fig. 2D). CpG methylation status is also related to the degree of transcriptional activation of a DNA region (27). All of the CpG sites in M3N5 were methylated according to the ENCODE database, whereas the M5N3 region contained many unmethylated or partially methylated CpG sites, which also suggested that fragment M5N3 was more transcriptionally active than M3N5 and N6 (Fig. 2D). We therefore concluded that fragment M5N3 encompasses the core sequence of this novel CASP8 promoter.

The pGL3-M5N3 construct (991 bp) only decreased the luciferase activity in KYSE510 cells by 13.7% relative to the full MAX fragment (Fig. 2B). We continued to explore the parts of this fragment that were most vital for maintaining transcriptional function. Three imbricating truncations of M5N3 were introduced into pGL3-Basic (pGL3-M5P1, 359 bp; pGL3-M5P2, 404 bp; pGL3-M5P3, 313 bp) and transfected into KYSE510 cells (Fig. 3A). Comparing the activities of these three constructs to that of pGL3-M5N3, we found that pGL3-M5P1 and pGL3-M5P2 decreased the transcriptional activity significantly (P<0.0001), whereas pGL3-M5P3 retained comparable activity (Fig. 3B). The M5P3 sequence (Fig. 3C), which likely contains the essential core of this newly identified promoter, has been submitted to GenBank under accession number KF765385.

To identify which transcription factors contribute to the activity of new promoter, we developed a strategy to separate, enrich and identify the nuclear proteins that specifically bind to fragment M5N3 (Fig. 4A). The biotin-labeled fragment (BM5N3) was conjugated to streptavidin-coupled magnetic beads and then incubated with nuclear proteins extracted from KYSE510 cells. To eliminate non-specific DNA-binding proteins, such as those binding histones, we pre-incubated the nuclear proteins with a non-promoter control fragment from a conserved exon of most CASP8 transcripts in which ENCODE ChIP-seq data suggested there are no potential transcription factor-binding sites (BC8L; Fig. 4B). The bound proteins were eluted and examined by SDS-PAGE. Silver staining of the gels revealed several bands that were enriched in the BM5N3 elutions compared with the BC8L elution. Most notably, when the nuclear extracts were pre-incubated twice with the control fragment (BM5N3-2), a greatly enriched band was evident in the region in which most transcription factors are distributed (35–55 kDa; Fig. 4B).

To identify the proteins in the specific enriched bands in the promoter fragment elutions (Fig. 4B), we extracted the bands and performed LC-MS/MS. Based on the MS/MS spectra of the peptides, we identified proteins using Mascot software. After excluding proteins that were also present in the corresponding control bands, we retained the 12 promoter-specific proteins listed in Table II. The selected proteins were analyzed in terms of three parameters, peptide score, molecular mass, and subcellular localization. We excluded four proteins (KLP6, MUCL1, GNAS and FBN3) based on the fact that their actual known molecular mass fell outside the 35- to 55-kDa gel region that we extracted. Two other proteins (PDHA1 and CLEC14A) were excluded because of their low peptide scores. Of the remaining six proteins, the transcriptional activator protein PURα showed the appropriate subcellular localization and known function and thus was chosen as the candidate transcription factor to be validated.

The MS/MS spectrum showed a peptide with mass 1059.50 Da reconstituted from the doubly charged ion of 530.76 m/z. The sequence obtained from this spectrum revealed that the peptide originated from arginine 230 to lysine 239 [R(230–239)K] of PURα [National Center for Biotechnology Information (NCBI) gene identifier 5813] and was unique to PURα (Fig. 5). Mascot software calculated the probability (P-value) of the identified peptide being a random event and transformed the P-value to a peptide score (the peptide score was 20 when P=0.05). The peptide score of the unique peptide from PURα was 41 (P=0.00074), indicating the reliability of the result.

To validate the MS/MS results, we first repeated the DNA-protein affinity purification using just the core promoter region M5P3. Silver staining of the electrophoresed elution products showed a significantly enriched band at ∼40 kDa, which was identical to the band isolated with the longer M5N3 fragment (Fig. 6A). Western blotting of the eluted products confirmed the presence of PURα. We also examined E2F-1 in the western blotting (Fig. 6B) because PURα and E2F family members are known to inter-regulate (28). To determine if PURα and E2F-1 physically interact, we performed reciprocal immunoprecipitation assays. The results showed that PURα and E2F-1 directly or indirectly interacted (Fig. 6C and D). Pol II is responsible for synthesizing messenger RNA in eukaryotes and is an essential part of the transcriptional machinery (29,30). A Pol II signal was also detected in the PURα immunoprecipitates (Fig. 6D), suggesting that PURα and E2F-1 are likely components of the transcriptional complex. Immunohistochemistry with PURα and E2F-1 antibodies and confocal microscopy confirmed the colocalization of these proteins in the nucleus of KYSE510 cells (Fig. 6E).

To evaluate the function of PURα and its contribution to the transcriptional machinery, we knocked down PURα in KYSE510 cells using siRNA-mediated RNAi and then transfected the pGL3-M5N3 construction to these cells and measured promoter activity. Western blotting confirmed that the siRNA decreased PURα protein expression (Fig. 7A). The pGL3-M5N3 promoter activity was still higher than the promoter-deficient pGL3-Basic plasmid in the PURα knockdown strain (P<0.0001) but was significantly lower than for pGL3-M5N3 without PURα knockdown (P=0.0265; Fig. 7B). Based on the results of transcriptional activity, we continued to test whether PURα could influence the mRNA expression of gene CASP8. The secondary promoter locates surround the 5′-UTR of isoform G, so we designed primers specific to isoform G. We observed significant down regulation of mRNA expression of isoform G in KYSE510 cells after knowing down PURα (P=0.0152; Fig. 7C). We validated this result in another esophageal cell line (EC0156; P=0.0043; Fig. 7C). These results point to PURα as responsible for the transcriptional activity of the secondary promoter and it is able to promote the mRNA expression of isoform G.