Sp1 and c-Myc regulate transcription of BMI1 in nasopharyngeal carcinoma

Authors

  • Hong-Bo Wang,

    1. State Key Laboratory of Oncology in South China, Sun Yat-sesn University Cancer Center, Guangzhou, China
    2. State Key Laboratory of Biocontrol, School of Life Sciences, Sun Yat-sen University, Guangzhou, China
    3. Department of Experimental Research, Sun Yat-Sen University Cancer Center, Guangzhou, China
    Search for more papers by this author
    • These authors contributed equally to this work
  • Gui-Hong Liu,

    1. Department of Oncology, the Second Affiliated Hospital of Guangzhou Medical College, Guangzhou, China
    Search for more papers by this author
    • These authors contributed equally to this work
  • Hua Zhang,

    1. State Key Laboratory of Oncology in South China, Sun Yat-sesn University Cancer Center, Guangzhou, China
    2. Department of Experimental Research, Sun Yat-Sen University Cancer Center, Guangzhou, China
    Search for more papers by this author
  • Shan Xing,

    1. State Key Laboratory of Oncology in South China, Sun Yat-sesn University Cancer Center, Guangzhou, China
    Search for more papers by this author
  • Li-Juan Hu,

    1. State Key Laboratory of Oncology in South China, Sun Yat-sesn University Cancer Center, Guangzhou, China
    2. Department of Experimental Research, Sun Yat-Sen University Cancer Center, Guangzhou, China
    Search for more papers by this author
  • Wei-Feng Zhao,

    1. State Key Laboratory of Oncology in South China, Sun Yat-sesn University Cancer Center, Guangzhou, China
    Search for more papers by this author
  • Bo Xie,

    1. Zhongshan School of Medicine, Sun Yat-Sen University, Guangzhou, China
    Search for more papers by this author
  • Man-Zhi Li,

    1. State Key Laboratory of Oncology in South China, Sun Yat-sesn University Cancer Center, Guangzhou, China
    2. Department of Experimental Research, Sun Yat-Sen University Cancer Center, Guangzhou, China
    Search for more papers by this author
  • Bo-Hang Zeng,

    1. Department of Oncology, the Second Affiliated Hospital of Guangzhou Medical College, Guangzhou, China
    Search for more papers by this author
  • Yingqiu Li,

    1. State Key Laboratory of Biocontrol, School of Life Sciences, Sun Yat-sen University, Guangzhou, China
    Search for more papers by this author
  • Mu-Sheng Zeng

    Corresponding author
    1. Department of Experimental Research, Sun Yat-Sen University Cancer Center, Guangzhou, China
    • State Key Laboratory of Oncology in South China, Sun Yat-sesn University Cancer Center, Guangzhou, China
    Search for more papers by this author

Correspondence

M.-S. Zeng, State Key Laboratory of Oncology in Southern China, Sun Yat-sen University Cancer Center, 651 Dongfeng Road East, Guangzhou 510060, China

Fax: +86 20 8734 3171

Tel: +86 20 8734 3191

E-mail: zengmsh@mail.sysu.edu.cn

Abstract

B-lymphoma mouse Moloney leukemia virus insertion region 1 (Bmi1), a member of the polycomb group, has elevated expression and is involved in the pathogenesis of various aggressive cancers, including nasopharyngeal carcinoma (NPC). To date, the mechanisms underlying the high expression of Bmi1 in NPC remain obscure. To gain new insights into the transcriptional regulation of BMI1, we cloned and characterized the promoter region of BMI1. Luciferase reporter assays demonstrated that the region from −783 to +375 showed significant promoter activity. With the use of a series of 5′-deletion and 3′-deletion promoter constructs in luciferase reporter assays, the +167/+232 and −536/−134 regions were found to be sufficient for full promoter activity. Transcriptional activity of the BMI1 promoter was dependent on the Sp1 binding site cluster (+181/+214) as well as the E-box elements (−181), and was abolished after mutation of the two cis-elements. Electrophoretic mobility shift assays and chromatin immunoprecipitation assays demonstrated that Sp1 bound to the region from +181 to +214 within the BMI1 promoter. In addition, gain-of-function and loss-of-function analyses revealed that Sp1 augmented Bmi1 expression. Further investigations using immunohistochemistry and quantitative RT-PCR disclosed a significant positive correlation between the expression of Sp1 and Bmi1 in normal nasopharyngeal epithelial cells, NPC cells, and NPC tissue specimens. In addition, Myc, the known transcription factor for BMI1 in neuroblastomas, also activated the transcription of BMI1 through binding to the E-box element (−181) within its promoter, and showed a positive correlation with the mRNA level of BMI1 in NPC. In conclusion, these findings provide valuable mechanistic insights into the role of Sp1 and c-Myc in BMI1 transcription in NPC, and suggest that targeting of Sp1 or c-Myc may be a potential therapeutic strategy for NPC.

Abbreviations
Bmi1

B-lymphoma mouse Moloney leukemia virus insertion region 1

ChIP

chromatin immunoprecipitation

EMSA

electrophoretic mobility shift assay

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

MITA

mithramycin A

NF-κB

nuclear factor-κB

NPC

nasopharyngeal carcinoma

NPEC

normal nasopharyngeal epithelial cell

qRT-PCR

quantitative RT-PCR

SEM

standard error of the mean

siRNA

small interfering RNA

TSS

transcription start site

Introduction

B-lymphoma mouse Moloney leukemia virus insertion region 1 (Bmi1), a member of polycomb-repressive complex 1, is a transcription repressor that plays essential roles in the regulation of stem cell self-renewal, embryogenesis, cell proliferation, and senescence [1-7]. Emerging evidence indicates that Bmi1 is upregulated in various malignancies, and promotes tumor progression by inhibiting the transcription of tumor suppressors, such as p53, p16INK4a, and p19Arf [8-11]. In addition, elevated expression of Bmi1 is correlated with an advanced stage and/or unfavorable prognosis in malignancies originating in the breast, neuroblast, colon, and esophagus, indicating an oncogenic role for Bmi1 in tumor progression [12-14].

To date, growing evidence has revealed the regulatory mechanisms of Bmi1. E2F1 and N-Myc promoted tumor proliferation through transactivation of BMI1 in neuroblastomas [13]. Hypoxia-induced Twist1 directly activated BMI1 transcription, and cooperatively acted with Bmi1 to induce an epithelial–mesenchymal transition and stemness properties in head and neck squamous cell carcinoma [15, 16]. Other transcription factors, such as SALL4 and c-Myb have also been identified as activators of BMI1 transcription in both hematopoietic and leukemic cells, while nuclear factor-κB (NF-κB) and the sonic hedgehog-activated Gli1 are potent transcription factors in Hodgkin lymphoma and medulloblastoma cells, respectively [17-21]. Although there are many studies addressing Bmi1 regulation, a detailed promoter analysis has yet to be conducted, and the underlying mechanisms remain elusive; therefore, these issues deserve to be investigated in more detail.

Nasopharyngeal carcinoma (NPC) is one of the most common cancers in southern China and Southeast Asia, with the highest incidence rate (40 per 100 000 persons per year) being among the Cantonese-speaking subpopulation in Guangdong province, China [22-24]. It is consistently associated with infection with Epstein–Barr virus, and occurs with striking geographical and ethnic differences in incidence [22]. Therefore, NPC is a malignancy with a complex etiology involving both genetic and environmental factors. In the previous two studies, we reported that the expression of Bmi1 is elevated in tumor tissues, and that high expression of Bmi1 correlates with poor prognosis in NPC. Bmi1 might induce immortalization and an epithelial–mesenchymal transition in human nasopharyngeal epithelial cells, indicating an essential role for Bmi1 in the pathogenesis of NPC [25, 26]. However, the molecular mechanisms underlying the high expression of Bmi1 in NPC remain largely unknown.

BMI1 is located on chromosome 10p11.23 (NC_000010.10), extends over 10 kb (chr10: 22 610 13–22 620 414), and comprises 10 exons. Its genomic location does not undergo frequent allelic losses and amplifications in NPC [24, 27, 28]. Although there are many CpG islands within its 5′-flanking region, the DNA methylation inhibitor 5-Aza-2′-deoxycytidine has no effect on its expression, suggesting that deregulated transcription rather than allelic amplifications and methylation may account for BMI1 gene overexpression in NPC [29].

To explore the mechanisms responsible for BMI1 transcription, the luciferase reporter system was first used to characterize the core promoter region of BMI1. The regions from +167 to +232 and from −536 to −134 were identified as the potential BMI1 gene promoter core regions whose transcriptional activity is dependent on the cooperative effect of the Sp1 binding site cluster (+181/+214) and the E-box elements (−181). Electrophoretic mobility shift assays (EMSAs), chromatin immunoprecipitation (ChIP) assays and gain-of-function and loss-of-function analyses suggested Sp1 as a functional transcription activator for BMI1. In addition, the expression of Bmi1 was positively correlated with that of Sp1 in normal nasopharyngeal epithelial cells (NPECs), NPC cells, and NPC tissue specimens. Furthermore, c-Myc was also shown to bind to the E-box element of the BMI1 promoter and promote the transactivation of BMI1 in NPC. Taken together, our results highlighted the essential roles of Sp1 and c-Myc in BMI1 transcription, and indicated a potential application of them in NPC therapy.

Results

Identification of the proximal promoter regions of BMI1

Phylogenetic conservation analysis revealed that the region from −500 to +500 of the BMI1 gene is highly conserved in vertebrate species [the first base of BMI1 cDNA, known as the transcription start site (TSS), was assigned as +1] (Fig. S1). To better elucidate the minimal promoter regions of BMI1 in NPC, the promoter activities of the region ~ 2000 bp upstream of the TSS were investigated. Two 5′-flanking regions, from −2063 to −783 (denoted as −2063/−783) and from −783 to +375 (denoted as −783/+375), were cloned upstream of the firefly luciferase reporter gene in the pGL3-basic vector (Fig. 1A), and this was followed by transfection into NPC-derived CNE2 and HNE1 cells. A dual-luciferase reporter system showed that the −783/+375 region had distinctly increased luciferase activity as compared with the control in the CNE2 (~ 120-fold increase) and the HNE1 (~ 100-fold increase) cells (Fig. 1B). These results defined an essential transcription motif within the −783/+375 region of BMI1.

Figure 1.

Characterization of the BMI1 promoter. (A) Schematic diagram of the BMI1 promoter fragments spanning from −2063 to −783 and from −783 to +375. The −2063/−783 and −783/+375 regions were cloned upstream of the firefly luciferase reporter gene in the pGL3-basic vector. (B) Transcriptional activity analysis of the potential BMI1 promoter fragments in CNE2 and HNE1 cells. (C) Schematic diagram of the 5′-truncated and 3′-truncated fragments spanning from −783 to +375 of BMI1. (D) Promoter activity analysis of the 5′-truncated and 3′-truncated fragments spanning from −783 to +375. For (B) and (D), CNE2 and HNE1 cells were transfected with either the pGL3-basic vector or the pGL3-basic vector containing the indicated fragments of the potential BMI1 promoter. pRL-SV40, which expresses Renilla luciferase, was cotransfected to calibrate the differences in both transfection and harvesting efficiency. Luciferase activity was detected 24 h post-transfection, and presented as relative luciferase units (RLU) (firefly luciferase/Renilla luciferase). All data are presented as the mean ± SEM from triplicate measurements. A representative result of three independent experiments is shown. For (A), (C) and (E), the numbers on the schematic diagram indicate the location of the fragments relative to the TSS, which was assigned as +1. ***P < 0.001. (E) The fragment from +167 to +232 in the 5′-flanking region of the BMI1 is highly conserved in humans and mice. Sequence alignment of the human BMI1 promoter region (from +167 to +232) with that of the mouse is shown. The sequence of the human BMI1 promoter region (from +167 to +232) was blast-searched (NCBI database).

In order to specifically determine the proximal promoter elements of BMI1, a series of 5′-truncated and 3′-truncated constructs were generated (Fig. 1C) and transfected into CNE2 and HNE1 cells for promoter activity assays. As shown in Fig. 1D, the −536/+232 and −783/+232 fragments displayed the first and second highest promoter activity among the six fragments. The promoter activity of the −134/+232 region (with a deletion of 402 nucleotides at the 5′-end of the −536/+232 region) decreased to 20% (CNE2 cells) or 27% (HNE1 cells) of that of the −536/+232 region. In addition, the promoter activity of the −783/+167 region (with a deletion of 65 nucleotides at the 3′-end of the −783/+232 region) decreased to 7% (CNE2 cells) or 10% (HNE1 cells) of that of the −783/+232 region. These findings suggested that both the +167/+232 region and the −536/−134 region were required for the full transcription for BMI1, whereas the +167/+232 region might play a critical role in the basal transcription of BMI1 in NPC cells.

We next analyzed the conservation of the newly identified promoter region +167/+232, using comparative sequence analysis, as the −1070/+53 fragment of BMI1 has been reported to be a potent promoter in neuroblastomas [30]. As shown in Fig. 1E, the +167/+232 region was highly conserved in humans and mice, suggesting that the +167/+232 region may be a basal promoter of the human BMI1 gene with evolutionarily conserved transcriptional regulatory elements.

Characterization of transcription factors essential for BMI1 transactivation

To comprehensively identify the transcription factors binding to the promoter region of BMI1 in NPC, the region from −536 to +375 was analyzed. As shown in Fig. 2, the region from −536 to +375 lacks both an apparent TATA-box and CAAT elements, and is a GC-rich element. Putative transcription factors and their binding sites within the BMI1 promoter region −536/+232 were predicted with the online software alibaba2.1 and tess, and were selectively highlighted. Among them, the Sp1 cluster (from +181 to +214) and the ETF-Egr cluster (from +126 to +179) were newly found in the present study, and the E-box region (−181) and two E2F1 binding sites (−319 and −122) have been reported to be essential in the transactivation of BMI1 in neuroblastomas.

Figure 2.

Schematic representation of the selected potential transcription factor binding sites within the region from −536 to +375. Prediction of the transcription factor binding sites in the promoter region was performed with the online software alibaba2.1 and tess.

To validate their roles in the regulation of BMI1 transcription in NPC cells, the above motifs in pGL3-BMI1−783/+375 were subjected to either site-directed mutagenesis or deletion (labeled as m or Δ in Fig. 3A). Point mutations that changed the sequence of E2F1 (−122) binding sites from TGTGGCGC to TGCGTAGC, changed the sequence of E2F1 (−319) binding sites from TTTGGAGC to CTCGGATC and changed the sequence of E-box regions (−181) from CACGTG to TTTGTG were introduced (mutations shown in bold). As shown in Fig. 3A,B, no significant change in the promoter activity was observed in the elements carrying mutant E2F1 binding sites (−122 or −319 mE2F1) or missing the ETF-Egr cluster (+126/+179 ΔETF-Egr), but an approximately 50% decrease in the promoter activity was observed in the elements carrying a mutant E-box region (−181 mE-box) or missing the Sp1 binding site cluster (+164/+241 ΔSp1), as compared with the wild-type region −783/+375 (Fig. 3B). These results suggest that the E-box (−181) and the Sp1 binding site cluster (+164/+241) are essential for BMI1 transcription.

Figure 3.

Essential roles of the Sp1 binding sites and the E-box element in BMI1 transcription. (A) Schematic of potential binding sites for the transcription factors within the region from −783 to +375. The E-box element (−181) and the binding sites for E2F1 (−122) and E2F1 (−319) were subjected to site-directed mutagenesis, whereas the ETF-Egr binding site cluster (+126/+179) and Sp1 binding site cluster (+164/+241) were deleted. The mutation and deletion are labeled as ‘m’ and ‘Δ’, respectively. The indicated fragments with mutation or deletion were cloned upstream of the firefly luciferase reporter gene in the pGL3-basic vector. (B) Promoter activity analysis of the potential transcription factor binding sites with point mutations or partial deletions in CNE2 cells. (C) Schematic of the E-box elements and Sp1 binding sites within the fragment from −783 to +232. The E-box elements (−181) or Sp1 binding sites (+181/+214) were subjected to site-directed mutagenesis. The indicated fragments with a single mutation or double mutations were cloned upstream of the firefly luciferase reporter gene in the pGL3-basic vector. (D) Promoter activity analysis of the potential transcription factor binding sites with the indicated mutations in CNE2 and HNE1 cells. For (B) and (D), NPC cells were transfected with either the pGL3-basic vector or pGL3-basic vector containing the indicated fragments of the potential BMI1 promoter. All data are presented as the mean ± SEM from triplicate measurements. A representative result of three independent experiments is shown. **P < 0.01; ***P < 0.001. RLU, relative luciferase units; WT, wild type.

To further verify whether the putative Sp1 binding sites mediated the transcriptional activity of the BMI1 promoter, we introduced 11 point mutations that changed the sequence from CGAGGAGGAGGAGGAGGAGGCCCCGGAGGAGGAGGCG to CGAGCACGATGCGAAGCAGCCCCCGGTGCATGACGCG, designated as mSp1 (+181/+214) (Fig. 3C). The plasmid pGL3-BMI1−783/+232, containing mutations at either the E-box region (−181) or the Sp1 binding site cluster (+181/+214), was transfected into CNE2 and HNE1 cells for 24 h, and luciferase assays were then conducted. As shown in Fig. 3D, as compared with the wild-type control, pGL3-BMI1−783/+232 containing mutations at either the E-box (−181 mE-box) or the Sp1 binding site cluster (+181/+214 mSP1) led to approximately 50% and 70% decreases in the promoter activity, respectively. In addition, pGL3-BMI1−783/+232 containing double mutations of the above two sites led to an additional (~ 90%) decrease in promoter activity, suggesting that the E-box element and the Sp1 binding site cluster are required for the full transcription of BMI1.

Binding of Sp1 to the BMI1 promoter in vitro and in vivo

We next examined whether Sp1 regulates BMI1 transcription, as the E-box region has been extensively reported to be responsible for the transcription of BMI1. To verify whether Sp1 binds to the Sp1 binding site cluster within the BMI1 promoter, we conducted EMSAs by incubating CNE2 cell nuclear extracts with the radiolabeled probe in the presence of antibody against Sp1 or FLAG (as a negative control). Probes were γ-32P-labeled double-stranded oligonucleotides containing either the wild-type region spanning from +181 to +214 of the BMI1 promoter or the corresponding mutant region, and designated as γ-32P-labeled wt Sp1 probe or γ-32P labeled mut Sp1 probe. A nonspecific band was determined by adding γ-32P-labeled mut Sp1 probe to CNE2 cell nuclear extracts. A supershifted band was observed when CNE2 cell nuclear extracts were exposed to the antibody against Sp1 (Fig. 4A, lane 3), but not when they were exposed to the antibody against FLAG (Fig. 4A, lane 2). Notably, the signal of the shifted band became faint in the presence of antibody against Sp1 (Fig. 4A, lane 3) as compared with that against FLAG (Fig. 4A, lane 2). In addition, we investigated whether Sp3 bound to Sp1 sites in the BMI1 promoter by supershift assay with a specific antibody against Sp3, as Sp1 and Sp3 have similar structures and their DNA-binding domains are highly homologous. No supershifted band was observed with antibody specific for Sp3 (Fig. S2, lane 5). Collectively, these findings indicate that Sp1 might selectively bind to the BMI1 promoter core regions in vitro.

Figure 4.

Sp1 interacts with the BMI1 promoter in vitro and in vivo. (A) EMSA verified Sp1 as a potential nuclear protein interacting with the BMI1 promoter. The analysis was performed in the presence (+, lanes 1–4) of CNE2 nuclear extracts. Shift bands were present in the binding reaction that included a γ-32P-labeled wild-type (wt) probe comprising the Sp1 binding site (+181/+214) (lanes 1–3), but were absent in the reaction that included the labeled mutant (mut) probe comprising the mutant Sp1 binding site (lane 4). A supershifted band was present when the antibody against Sp1 was added to the reaction (lane 3), but was absent when the antibody against FLAG was added (lane 2). A representative result of three experiments is shown. (B) Scheme of the location of the ChIP amplicons in the E-box and the binding sites for Sp1. (C) ChIP assay showed that Sp1 directly interacted with the Sp1 binding sites, but not with the −239/−87 E-box element within the BMI1 promoter. Formaldehyde-crosslinked chromatin isolated from CNE2 cells was precipitated with antibody against Sp1 or FLAG (negative control), respectively. The isolated chromatin complexes were subsequently amplified by PCR with 2× GC Hotstar Taq buffer with primer sets that encompassed the E-box region (from −239 to −87) or the predicted Sp1 binding sites (from +90 to +324). A specific strong band of the expected size was detected in the input DNA (lane 3). The fragment containing the Sp1 binding sites was detected (lane 2, top) whereas the fragment containing the E-box region was significantly faint in DNA precipitated by antibody against Sp1 (lane = 2, bottom). No band was detected in the chromatin complex precipitated by antibody against FLAG (lane 1).

Subsequently, to investigate whether Sp1 could actually interact with the BMI1 promoter in vivo, ChIP assays were first performed. As shown in Fig. 4B,C, antibody against Sp1 crosslinked to the Sp1 binding site cluster (the ChIP primer-amplified region from +90 to +324), but showed significantly lower affinity for the E-box region (the ChIP primer-amplified region from −239 to −87) within the BMI1 promoter in CNE2 cells. In addition, antibody against FLAG immunoprecipitated neither of the above two sites. These results indicate that Sp1 might specifically and directly bind to the BMI1 promoter in vivo.

Mithramycin A (MITA) treatment repressed promoter activity and expression of Bmi1

We further investigated the role of Sp1 by using MITA, a US Food and Drug Administration-approved drug that inhibits transcriptional activation by preventing Sp1 from binding to GC-rich regions [31-34]. CNE2 and HNE1 cells were treated with 125 nm MITA for 72 h. As shown in Fig. 5, the promoter activity was significantly attenuated in MITA-exposed cells as compared with dimethylsulfoxide-treated cells. Consistent with this, the Bmi1 protein level was suppressed after treatment with MITA in a dose-dependent manner (Fig. S3), suggesting that Sp1 might be crucial for the transcription of BMI1 through directly binding to the BMI1 promoter region. Taken together, these results demonstrate that Sp1 could enhance the expression of Bmi1 through activating its promoter.

Figure 5.

MITA treatment suppressed the promoter activity and expression of Bmi1. (A) MITA treatment repressed the promoter activity of the fragment from −783 to +232 within the BMI1 promoter. CNE2 and HNE1 cells were cotransfected with both pGL3-basic containing the fragment (−783/+232) and pRL-SV40 for 16 h, exposed to 125 nm MITA for 48 h, and then subjected to luciferase assays. ***P < 0.001. (B) MITA treatment downregulated the expression of Bmi1. CNE2 and HNE1 cells (4 × 105) were placed in six-well plates for 16 h, and then treated with 125 nm MITA for 72 h before western blotting. GAPDH served as an internal control. RLU, relative luciferase units.

Ectopic expression of Sp1 elevated the promoter activity of BMI1

To confirm the role of Sp1 in the regulation of BMI1 promoter activity, Sp1 was first overexpressed in CNE2 and HNE1 cells through transfection with pCDNA6-myc-HisB-SP1, which contained the entire Sp1 coding sequence and a Myc-tag. Transgene expression was confirmed by western blotting analysis with a mAb against Myc-tag. As shown in Fig. 6A, the level of Myc-tag-labeled Sp1 was distinctly increased in pCDNA6-myc-HisB-SP1-transfected cells as compared with vector-transfected cells. To determine whether the introduction of exogenous Sp1 improved the transcriptional activity of the BMI1 promoter, CNE2 and HNE1 cells were cotransfected with pGL3-BMI1−783/+232, pRL-SV40 and either pCDNA6-myc-HisB-SP1 or pCDNA6-myc-HisB for 48 h, and then subjected to luciferase assay. Ectopic expression of Sp1 elevated the luciferase activities of not only the firefly luciferase, but also the Renilla luciferase, consistent with the report that Sp1 may regulate the transcription of SV40. However, the relative luciferase units (firefly luciferase/Renilla luciferase) were significantly augmented in Sp1-overexpressing CNE2 and HNE1 cells (Fig. 6B). In addition, we also used pRL-TK-Luc as control or the protein concentration to normalize the activity, and found that the activity of the BMI1 promoter upregulated by Sp1 was more significant (data not shown). Collectively, these results indicate that ectopic expression of Sp1 augmented BMI1 promoter activity.

Figure 6.

Ectopic expression of Sp1 elevated BMI1 promoter activity. (A) Sp1 was ectopically expressed in CNE2 and HNE1 cells. (B) Sp1 overexpression induced transcriptional activity of the BMI1 promoter. CNE2 and HNE1 cells were cotransfected with 10 ng of pRL-SV40, 0.1 μg of the firefly luciferase reporter plasmids and 1.5 μg of either the pCDNA6-myc-HisB vector or the pCDNA6-myc-HisB-SP1 plasmid in each reaction. Cell lysates were harvested 48 h post-transfection, and subjected to western blotting (A) and luciferase assay (B). A representative result of three experiments is shown. ***P < 0.001. RLU, relative luciferase units.

Effect of Sp1 overexpression or knockdown on the Bmi1 level in NPC cells

To investigate the role of Sp1 in Bmi1 expression, gain-of-function studies were first performed. As shown in Fig. 7A,B, both the protein and mRNA levels of Bmi1 were obviously increased in Sp1-overexpressing HNE1 cells as compared with the control cells, suggesting that Sp1 may induce the expression of Bmi1 in NPC cells. To verify the above findings, loss-of-function analyses were performed in both CNE2 and HNE1 cells with a small interfering RNA (siRNA) duplex targeting Sp1 (siSp1), which effectively abrogated Sp1 expression. As compared with negative control (siNC)-transfected cells, siSp1 transfectants showed reduced Bmi1 expression at both the protein and mRNA levels in both CNE2 and HNE1 cells (Fig. 7C,D). We further investigated the effect of Sp3 on the expression of Bmi1 to eliminate the possibility that Sp3 might regulate the expression of Bmi1. SiSp3, which distinctly attenuated Sp3 expression, had no effect on Bmi1 expression, indicating that Sp1 induced the expression of Bmi1 through specifically crosslinking to its binding sites. Collectively, these findings indicate that Sp1 is a strong inducer of Bmi1 expression, and is also required for the full expression of Bmi1 in NPC cells.

Figure 7.

Effect of Sp1 overexpression or knockdown on the Bmi1 level in NPC cells. (A, B) Sp1 overexpression enhanced the protein (A) and mRNA (B) levels of Bmi1 in HNE1 cells. Seventy-two hours post-transfection, HNE1 cells that were successfully transfected with pEGFP-N1 and either the empty vector (pCDNA6-myc-HisB) or the SP1 overexpression plasmid (pCDNA6-myc-HisB-SP1) were sorted on the basis of enhanced green fluorescent protein expression with fluorescence-activated cell sorting. The sorted cells were harvested and subjected to western blotting (A) and qRT-PCR (B). (C, D) Knockdown of Sp1 reduced the endogenous Bmi1 level. CNE2 and HNE1 cells were transfected with the indicated siRNA duplex for 72 h, and this was followed by western blotting (C) and qRT-PCR (D). GAPDH served as an internal control. A representative result of three experiments is shown. *P < 0.05; **P < 0.01; ***P < 0.001.

Positive correlation between Bmi1 and Sp1 expression in NPECs, NPC cells, and NPC tumor tissues

On the basis of the above observations, we then examined whether Bmi1 and Sp1 were expressed concomitantly in various NPECs and NPC cells. Linear regression analysis showed a positive correlation between the mRNA levels of both BMI1 and SP1 in the cell lines (P < 0.045, r = 0.730; Fig. 8A,B). In addition, both Bmi1 and Sp1 showed a similar trend in the protein expression level among eight NPEC and NPC cell lines (Fig. 8C). To confirm these findings in in situ NPC tumors, we further evaluated the correlation between the expression of SP1 and BMI1 in human NPC specimens by using qRT-PCR. Linear regression analysis showed a significant positive correlation between the mRNA levels of SP1 and BMI1 in 69 NPC tissues (P < 0.0001, r = 0.807; Fig. 8D). We then determined the protein levels of Bmi1 and Sp1 in 22 NPC specimens by immunohistochemistry. As shown in Fig. 8E,F, there were 13 specimens that showed high expression of both Sp1 and Bmi1, and five specimens that showed low expression of both Sp1 and Bmi1. Evaluation of the data with Fisher's exact test revealed that the protein levels of Bmi1 and Sp1 were significantly correlated in NPC specimens (P = 0.011). Collectively, these results suggest that both the mRNA and protein levels of Bmi1 and Sp1 are positively correlated in NPECs, NPC cells, and NPC tissue specimens.

Figure 8.

Expression of Bmi1 and expression of Sp1 are positively correlated in NPECs, NPC cells, and NPC tumor specimens. (A–C) Bmi1 and Sp1 showed a similar expression trend in the mRNA (A, B) and protein levels (C) among eight NPECs/NPC cells. The arrow indicates Sp1 protein, and the asterisks indicate the possible isoforms or proteolytic degradation products of Spl. (D) A significant positive correlation was found between the mRNA levels of BMI1 and SP1 in NPC tissues. The mRNA levels of BMI and SP1 in 69 NPC tissue specimens were examined by qRT-PCR. The correlation was analyzed with Spearman's correlation coefficient. (E) The protein expression levels of Bmi1 and Sp1 in NPC tumors were statistically correlated. The expression levels of Bmi1 and Sp1 in 22 NPC tumors were analyzed by immunohistochemical staining. The overall staining intensity and percentage were quantified with the Nuance multispectral imaging system. A value designated as the immunohistochemical score (H-score) was derived by summing percentages of cells stained at each intensity multiplied by the weighted intensity of staining. An H-score of ≤ 1 was considered to be low, and an H-score of > 1 was considered to be high. (F) Representative immunohistochemical staining of Bmi1 in consecutive NPC tumor sections with low (biopsy 1) and high (biopsy 2) Sp1 expression.

C-Myc promoted the expression of Bmi1 in NPC cells

As shown in Fig. 3B, the BMI1 promoter core region spanning from −536 to −134, which contains a Myc-binding E-box element (−181), possessed transcriptional activity in NPC cells, suggesting that this region was also required for the transcription of BMI1. To explore the exact effect of c-Myc on the expression of Bmi1 in NPC cells, ChIP assay was first used to determine whether c-Myc binds to the E-box region in the BMI1 promoter. As shown in Fig. 9A,B, c-Myc was crosslinked to the E-box region (the ChIP primer-amplified region from −239 to −177), but showed significantly lower affinity for the adjacent region (the ChIP-primer-amplified region from −118 to −26) within the BMI1 promoter in CNE2 cells. Similar results were obtained in C666-1 cells, suggesting that c-Myc may bind to the E-box region within the BMI1 promoter in NPC cells.

Figure 9.

Effect of c-Myc on the expression of Bmi1. (A) A scheme of the location of the ChIP amplicons for the E-box and the adjacent region is shown. (B) ChIP assay showed that c-Myc directly interacted with the −239/−177 region where the E-box region located, but not with the adjacent −118/−26 region. Formaldehyde-crosslinked chromatin isolated from CNE2 and C666-1 cells was precipitated with antibody against c-Myc or FLAG, respectively. The isolated chromatin complexes were subsequently amplified by qRT-PCR with primer sets that encompassed the E-box region (from −239 to −177) or the adjacent region (from −118 to −26). Relative enrichment was calculated by normalizing the threshold cycle (CT) value of DNA coprecipitated by antibody against Myc against the CT value of DNA coprecipitated by antibody against FLAG control (IgG). (C, D) Ectopic expression of c-Myc augmented the expression and promoter activity of Bmi1 in HNE1 cells. HNE1 cells were cotransfected with 10 ng of pRL-SV40, 0.1 μg of the firefly luciferase reporter plasmids and 1.5 μg of either the pCDNA6-myc-HisB vector or the pCDNA6-myc-HisB-MYC plasmid in each reaction. Cell lysates were harvested 48 h post-transfection, and subjected to western blotting (C) and luciferase assay (D). (E, F) Knockdown of c-Myc reduced Bmi1 expression in CNE2 cells. After transfection with either siMyc or NC for 72 h, the expression levels of Bmi1 or c-Myc in CNE2 cells were analyzed by western blotting (E) and qRT-PCR (F). (G) There was a positive correlation between the mRNA levels of BMI1 and MYC in NPC tissues. The mRNA levels of BMI1 and MYC in 69 NPC tissue specimens were examined by qRT-PCR. The correlation was analyzed with Spearman's correlation coefficient. GAPDH served as an internal control. A representative result of three experiments is shown. *P < 0.05; **P < 0.01; ***P < 0.001. RLU, relative luciferase units.

Subsequently, gain-of-function analysis was performed to investigate the role of c-Myc in the expression of Bmi1. As shown in Fig. 9C, c-Myc was ectopically expressed in HNE1 cells through transfection with pCDNA6-myc-HisB-MYC, which contained the entire c-Myc coding sequence. As compared with the control cells, c-Myc-overexpressing HNE1 cells showed elevated expression of Bmi1 and the promoter activity of the −783/+232 region of the BMI1 gene (Fig. 9C,D). To confirm the above results, loss-of-function analyses were performed with an siRNA duplex targeting MYC (siMyc), which effectively abrogated c-Myc expression. As compared with siNC-transfected cells, siMyc transfectants showed reduced Bmi1 expression at both the protein and mRNA levels in CNE2 cells (Fig. 9E,F). In addition, linear regression analysis showed a positive correlation between the mRNA levels of BMI1 and MYC in NPC tissues (P = 0.001, r = 0.483; Fig. 9G). Collectively, these findings indicate that c-Myc is an essential transcription factor for BMI1.

Discussion

Although the expression of Bmi1 has been reported to be elevated and significantly correlated with the clinical outcome in NPC, the precise mechanisms underlying Bmi1 regulation remain elusive [25]. In the present study, we defined the regions from +167 to +232 and from −536 to −134 as the core promoter regions for the BMI1 gene. In addition, Sp1 and c-Myc were identified as transactivators of BMI1. This conclusions are supported by the following lines of evidence: (a) a series of deletions revealed that the regions spanning from +167 to +232 and from −536 to −134 possessed essential transcriptional activities; (b) Sp1 binding sites and the E-box region were mapped to the promoter region of the BMI1 gene, and mutations of these sites caused a dramatic decrease in BMI1 promoter activity; (c) Sp1 and c-Myc directly interacted with the BMI1 promoter in vitro and in vivo; (d) knockdown of either Sp1 or c-Myc significantly attenuated BMI1 promoter activity and endogenous Bmi1 expression; and (e) Sp1 and c-Myc expression statistically correlated with Bmi1 expression in NPECs, NPC cells, and/or NPC tumor tissues. Taken together, these findings show that Sp1 and c-Myc serve as important transcription factors for BMI1.

Sp1, a zinc finger transcription factor, is ubiquitously expressed in normal tissue and upregulated in tumor cells [35-37]. As an anchor for TATA-box binding protein-associated factors, Sp1 recruits the basal transcriptional machinery to initiate transcription in TATA-less promoters [38-40]. In the present study, the identified promoter (from −536 to +375) of BMI1 lacks an apparent TATA-box and CAAT elements, and is a G/C-rich region. Therefore, Sp1 may bind to G/C-rich regions within the BMI1 promoter to help establish a transcription preinitiation complex.

Multiple GC-binding transcription factors such as Egr1, Egr2 and ETF have been predicted to interact with the identified BMI1 promoter (from +126 to +214). However, knockdown of the transcription factors did not affect Bmi1 expression (data not shown), whereas knockdown of Sp1 attenuated the transcription of BMI1. Sp3 could bind to the Sp1 cluster, as Sp1 and Sp3 have similar structures and their DNA-binding domains are highly homologous [37, 41]. To eliminate the possibility that Sp3 regulated the transcription of BMI1, a supershift assay was performed with a specific antibody against Sp3. No supershifted band was observed with antibody specific for Sp3. In addition, knockdown of Sp3 had no effect on Bmi1 expression in either CNE2 or HNE1 cells. Collectively, these findings suggest that Sp1 is a specific and crucial transcription factor for BMI1.

To date, the −1070/+53 fragment has been reported to be a potent promoter of BMI1 in neuroblastomas. In addition, transcription factors such as N-Myc and E2F1 have been shown to bind to it [30]. In the present study, we found that E2F1 had no effect on BMI1 transcription, and confirmed that the BMI1 promoter core region spanning from −536 to −134, which contains a Myc-binding E-box element (−181), also possessed transcriptional activity in NPC cells. ChIP assays showed that c-Myc could be crosslinked to the BMI1 promoter in vivo. Consistent with these findings, knockdown of c-Myc attenuated Bmi1 expression, suggesting that c-Myc is also an effective transcription factor in NPC.

We defined the +167/+232 region as a basal promoter, and identified Sp1 as a novel transcription factor for BMI1. In addition, pGL3-BMI1−783/+232, containing double mutations of the E-box region and Sp1 binding site cluster, led to an additional (~ 90%) decrease in the promoter activity. Sp1 and c-Myc are enriched in the active promoter, and cooperatively regulate the transcription of human telomerase catalytic subunit and phosphatidylserine synthase-1 [42, 43]. Therefore, both Sp1 and c-Myc may be required for the full transcription of BMI1 and may have a cooperative effect on BMI1 transcription.

In the present work, Sp1 enhanced the expression of Bmi1 through activating its promoter. It is noteworthy that the levels of Bmi1 and Sp1 showed significant correlation in NPECs, NPC cells, and NPC tumor tissues, suggesting a potential value of Sp1 inhibitors as a therapeutic strategy for the treatment of NPC. MITA, a US Food and Drug Administration-approved drug that inhibits the transcriptional activity of Sp1, is an effective anticancer chemotherapeutic agent for advanced cases in various malignancies, such as bone and prostate carcinoma [33, 34, 44]. We found that MITA treatment effectively suppressed promoter activity and diminished Bmi1 expression in NPC cells, implying that MITA may have promise as an improved treatment for NPC.

It is well known that gene expression is regulated not only at the transcriptional level but also at the post-transcriptional level. As shown in Fig. 8B,C, the mRNA and protein levels of Bmi1 are not well correlated, suggesting that post-transcriptional regulation may also contribute to Bmi1 deregulation in cancer cells [45-47]. It has been reported that Bmi1 expression can be suppressed by microRNAs, such as miR-15a, miR-16, miR-128, miR-200c, miR-203, and miR-194, in various carcinomas, suggesting that the expression level of Bmi1 may be regulated in multiple ways [48-51].

In conclusion, we have characterized the core promoter regions of BMI1, and identified Sp1 and c-Myc as important transcription factors that directly bind to the BMI1 promoter region. Our findings provide new insights into the transcriptional regulation of BMI1, and may be helpful in the exploration of novel therapeutic strategies for advanced NPC cases.

Experimental procedures

Tissue specimens and cell lines

A total of 22 paraffin-embedded sections of NPC tissue were collected from the archives of the Department of Pathology at the Cancer Center of Sun Yat-Sen University in Guangzhou, PR China. Sixty-nine RNA samples from fresh biopsies of the nasopharynx, which were pathologically confirmed as NPCs, were collected from the Department of Nasopharyngeal Carcinoma at the Cancer Center. Two fresh biopsies of the nasopharynx, which were pathologically exclusive of NPC, were collected from the Department of Head and Neck Surgery at Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University. None of the patients had received local or systemic anticancer treatment before the operation. This study was approved by the Institute Research Ethics Committee at the Cancer Center. Written informed consent was obtained from each patient.

Primary cultures of NPECs were established as described previously, and grown in keratinocyte serum-free medium (Gibco, Grand Island, NY, USA) [25]. All human NPC cell lines were cultured in RPMI-1640 medium (Gibco) supplemented with 10% fetal bovine serum (Gibco). Cells were grown in a humidified 5% CO2 incubator at 37 °C, and passaged with standard cell culture techniques.

RNA interference

RNA oligoribonucleotides were purchased from Dharmacon (Chicago, IL, USA) or GenePharma (Shanghai, China). The siRNAs targeting the mRNA of human SP1 (GenBank accession no. NM_138473.2), human SP3 (NM_003111.4) and MYC (NM_002467.4) were denoted as siSp1 (L-026959-00-0003; Dharmacon), siSp3 (L-023096-00-0003; Dharmacon) and siMyc (GenePharma), respectively. The siRNAs for Sp1 and Sp3 are ON-TARGET plus SMART pool siRNA duplexes. The NC (D-001220-01-20; Dharmacon) RNA duplex was nonhomologous to any human genome sequences. CNE2 and HNE1 cells (1 × 105) were seeded on six-well plates for 16 h, and this was followed by transfection with 50 nm RNA duplex with Lipofectamine RNAiMAX (Invitrogen, Grand Island, NY, USA), according to the manufacturer's instructions.

Plasmid constructs

The potential BMI1 promoter regions (−2063/−783 and −783/+375) and 5′-truncated or 3′-truncated fragments were PCR-amplified with the genomic DNA from NPEC1-Bmi1 cells (immortalized nasopharyngeal epithelial cells), pGL3-BMI1−783/+375 or pGL3-BMI1−783/+232 as the template, respectively, and were then inserted into the HindIII–KpnI sites upstream of the firefly luciferase in the pGL3-Basic vector (Promega, Madison, WI, USA). All of the constructs were named on the basis of the location of the promoter fragments relative to the TSS. Point and deletion mutagenesis were performed with overlap extension PCR methods. The full-length cDNA sequence of c-Myc was PCR-amplified with the cDNA from CNE2 cells, and cloned into the pcDNA6-myc-HisB vector. This construct was named pCDNA6-myc-HisB-MYC. An SP1 expression vector named pCDNA6-myc-HisB-SP1 was generated in our previous study [52]. All constructs were confirmed by direct sequencing. The primers are listed in Table 1.

Table 1. Sequences of primers
NameSense primer (5′- to 3′)Antisense primer (5′- to 3′)Amplicon (bp)
Primers for cloning
−2063/−783 GGGTACCGGATTTAATGGGAAAGGGGCATCA CCCAAGCTTCAAACCCAGGAGATCGCATCGT 1280
−783/+375 GGGGTACCACGATGCGATCTCCTGGGTTTG CCCAAGCTTTGCAGACGCCGCTGTCAATGG 1158
−783/+167 GGGGTACCACGATGCGATCTCCTGGGTTTG CCCAAGCTTTCCTCCCGCTCCGCACTC 950
−783/+53 GGGGTACCACGATGCGATCTCCTGGGTTTG CCCAAGCTTATTGTCTCGCCCCGATCTCTGC 836
−783/+232 GGGGTACCACGATGCGATCTCCTGGGTTTG CCCAAGCTTCCGCCTCGACCTCCAACG 1015
−536/+232 GGGGTACCGTTGTTTCCGCGAGCCCAACTTG CCCAAGCTTCCGCCTCGACCTCCAACG 768
−134/+232 CCCAAGCTTATTGTCTCGCCCCGATCTCTGC CCCAAGCTTCCGCCTCGACCTCCAACG 366
Primers for deletion
+126/+179 ΔETF-Egr CCGATCGCGCCCGAGGAGGAGGAGGAGGAGGCCCCGGAGGAGG CCTCCTCCTCCTCCTCCTCGGGCGCGATCGGTTTTATTCTGCGGGGCTGG 1109
+164/+241 ΔSp1 CGAGTGCGGAGCGGGAAGGAGGAGGCCGAGGCGCCGGAGGAGG GCCTCGGCCTCCTCCTTCCCGCTCCGCACTCGGGGGAGGGC 1083
Primers for mutation
−319 mE2F1 GGAGCAGGGCGGCGCGTGCGTAGCTGTGGAGAAATGTC GACATTTCTCCACAGCTACGCACGCGCCGCCCTGCTCC 1015
−122 mE2F1 CGGACTCCGAGCCATCTCGGATCCGGTGTCAGTTTCCAC GTGGAAACTGACACCGGATCCGAGATGGCTCGGAGTCCG 1015
−181 mE-box CCTTAAGGAATGAGGGGAGTTTGTGACCCGCTGGGGG CCCCCAGCGGGTCACAAACTCCCCTCATTCCTTAAGG 1015
+181 to +214 mSp1 GGGGTACCACGATGCGATCTCCTGGGTTTG R1: CGTCATGCACCGGGGGCTGCTTCGCATCGTGCTCGGCCGCCGCCGCCTCCTCCCGCTCC 1015
R2: CCCAAGCTTCCGCCTCGACCTCCAACGCGTCATGCACCGGGGGCTGCTTCGCATCG  
Primers for ChIP
−239/−87 E-box CACGGGCCTGACTACACCGACACT CACCGCTGAAGGCAGAGTGGAAAC 152
+90/+324 Sp1 CACCTCCCAGCCCCGCAGAA CGCCACGCTCGTCTCATGCC 234
Primers for qPCR
−118/−26 GGTGTCAGTTTCCACTCTGCCTTC GCCGTGTGTGTGCGTGCGAGC 93
−239/−177 E-box CACGGGCCTGACTACACCGACACT ACGTGCTCCCCTCATTCCT 62
BMI1 TGGCTCGCATTCATTTTCTG TGTGGCATCAATGAAGTACCCT 185
SP1 TCCAGACCATTAACCTCAGTGC TGTATTCCATCACCACCAGCC 141
MYC GGAGGCTATTCTGCCCATTTG CGAGGTCATAGTTCCTGTTGGTG 175
GAPDH CTCCTCCTGTTCGACAGTCAGC CCCAATACGACCAAATCCGTT 111

Plasmid transfections

All plasmid transfections were performed with Fugene HD (Roche, Indianapolis, IN, USA), according to the manufacturer's instructions. HNE1 cells (3 × 105) were seeded on six-well plates for 16 h, and this was followed by transfection with the pCDNA6-myc-HisB vector, the pCDNA6-myc-HisB-SP1 plasmid, or the pCDNA6-myc-HisB-MYC vector, as indicated, at a final concentration of 3 μg per well.

Luciferase reporter assay

CNE2 and HNE1 cells (1 × 105) plated in a 24-well plate for 16 h were cotransfected with 10 ng of the pRL-SV40 vector and 0.5 μg of the firefly luciferase reporter, unless otherwise indicated. The cells were then subjected to luciferase activity analysis with the Dual-Luciferase reporter assay system (Promega), according to the manufacturer's instructions. pRL-SV40, which expresses Renilla luciferase, was cotransfected as an internal control to correct for differences in both transfection and harvesting efficiencies.

Quantitative RT-PCR (qRT-PCR) analysis

Total RNA was extracted with the TRIzol reagent (Invitrogen). cDNA was synthesized from 2 μg of the total RNA with a reverse transcriptase kit (Invitrogen). The mRNA level was evaluated by qRT-PCR with Power SYBR Green qPCR SuperMix-UDG (Invitrogen), and was analyzed on an ABI Prism 7900 HT sequence detection system (Applied Biosystems, Foster City, CA, USA). Relative expression was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression, which yielded a 2−ΔΔCt value. The primers for qRT-PCR are listed in Table 1.

Western blotting analysis

Western blotting analysis was performed as previously described [53]. The blots were probed with rabbit polyclonal antibodies against Bmi1 (Proteintech Group, Wuhan, China), Sp1 (#07-645; Millipore, Temecula, CA, USA), Sp3 (A302-484A; Bethyl Laboratories, Montgomery, TX, USA), and c-Myc (#04-216; Millipore). The same membranes were then stripped and reprobed with mouse mAbs against GAPDH (Santa Cruz Biotechnology, Santa Cruz, CA, USA), to confirm equal loading of the samples.

Immunohistochemistry

Paraffin-embedded samples were cut into 4-μm sections and processed for immunohistochemistry as previously described [25]. Following incubation with rabbit polyclonal antibodies against either Bmi1 (Proteintech Group) or Sp1 (#07-645; Millipore), immunostaining was performed with the Zymed Histostain TM−Plus Kits (Zymed, South San Francisco, CA, USA) [54]. To define the expression of Bmi1 and Sp1, five random fields were captured under ×200 magnification with a computerized system that included a Digital Sight DS-Fi1 camera installed on a Nikon Eclipse 80i light microscope (Nikon, Melville, NY, USA). Images were analyzed with the Nuance Multispectral Imaging System (Cambridge Research and Instrumentation, Woburn, MA, USA) according to the uniform settings. Peritumoral nasopharyngeal epithelial cells, inflammatory cells and stromal cells were not included in the analysis. The evaluations were recorded as percentages of positively stained tumor cells in each of four intensity categories, which were assigned as follows: 0, no staining; 1, weakly positive; 2, moderately positive; and 3, strongly positive. For each slide, a value designated as the immunohistochemical score (H-score) was derived by summing the percentages of cells stained at each intensity multiplied by the weighted intensity of staining [55]. An H-score of ≤ 1 was considered to be low, and an H-score of > 1 was considered to be high.

EMSA

EMSA was performed as described previously [52]. Briefly, complementary oligonucleotide pairs containing either the putative binding site for Sp1 (Sp1 probe) or the mutant binding site for Sp1 (mutant Sp1 probe) were labeled with [32P]ATP[γP] (PerkinElmer Life and Analytical Sciences, Wellesley, MA, USA). CNE2 cell nuclear extracts were incubated with antibody against either Sp1 (sc-59; Santa Cruz), Sp3 (#07-107; Millipore), or FLAG, and this was followed by incubation with labeled probe. For competition analysis, a 100-fold excess of unlabeled double oligonucleotides was incubated with CNE2 cell nuclear extracts for 1 h at 4 °C, and then for an additional 30 min at room temperature in the presence of labeled probes. DNA–protein complexes were separated on a pre-electrophoresed nondenaturing polyacrylamide gel. The gel was then vacuum-dried, exposed to a Storage Phosphor Screen, and read with a Typhoon 9400 phosphorimager (GE Healthcare, Piscataway, NJ, USA). The sequences of the probes used in this study are shown in Table 2.

Table 2. Sequences of the oligonucleotides for siRNA assays and EMSAs
NameSequence (5′- to 3′)
siRNA duplexes
siSp1 CUACAGAGGCACAAACGUA
GGGCAGACCUUUACAACUC
GAAGGGAGGCCCAGGUGUA
GCCAAUAGCUACUCAACUA
siSp3 GGUAUUCACUCUAGCAGUA
GAAAUUUGUUUGUCCAGAA
GAUAGGAACUGUUAAUACU
GCGAGAUGAUACUUUGAUU
siMyc CACCUAUGAACUUGUUUCA dTdT
EMSA
Sp1 probe GGAGGAGGAGGCCCCGGAGGAGG
Mutant Sp1 probe TGCGAAGCAGCCCCCGGTGCATG

ChIP assay

A total of 2 × 106 CNE2 cells were crosslinked with formaldehyde, and the nuclei were isolated and sonicated to shear the DNA into fragments of 500 bp to 1 kb. The chromatin complexes were subjected to immunoprecipitation with 5 μg of antibody against Sp1 (#07-645; Millipore), c-Myc (#04-216; Millipore), or FLAG (F1804; Sigma-Aldrich, St Louis, MO, USA), as indicated. The input DNA was isolated from the sonicated lysates before immunoprecipitation. The immune complexes were collected with Protein A/G Plus agarose (Pierce, Rockland, IL, USA), and the crosslinks were then reversed by heating the samples at 65 °C for 4 h. The DNA was purified with the Wizard SV gel and PCR clean-up system (Promega), and was then subjected to PCR, as indicated in the figure legends. PCR performed with LA Taq DNA Polymerase (Takara, Dalian, China) and 2× GC Buffers was used to detect the GC-rich fragment. Primers for the ChIP assay are listed in Table 1.

Statistical analysis

The data analysis was conducted with graphpad prism version 4.0 (GraphPad Software, San Diego, CA, USA) and spss 16.0 (SPSS, Chicago, IL, USA). The results were representative of at least three independent experiments. Data were presented as the mean ± standard error of the mean (SEM) obtained with triplicate samples. Analysis of the differences between groups was determined with the two-tailed Mann–Whitney test. Correlations between the levels of Sp1 and Bmi1 were assessed with Spearman's correlation coefficient, unless otherwise indicated. A P-value of 0.05 was used as the criterion of statistical significance.

Acknowledgements

We are particularly grateful to M. Li for his expert advice on EMSA experiments. We thank Q. Zhong and J.-P. Zhang (SunYat-Sen University Cancer Center) for the careful editing of the manuscript. This study was supported by grants from National Natural Science Funds for Distinguished Young Scholars (81025014), the Ministry of Science and Technology of China (2012CB967003), and the National Natural Science Foundation of China (81161120408 and 91019015).

Ancillary

Advertisement