The transcriptional factor YY1 upregulates the novel invasion ... - Nature

Oncogene (2005) 24, 4081–4093

& 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc

The transcriptional factor YY1 upregulates the novel invasion suppressor HLJ1 expression and inhibits cancer cell invasion Chi-Chung Wang1,2, Meng-Feng Tsai2,3, Tse-Ming Hong2,4, Gee-Chen Chang5,6, Chih-Yi Chen6,7, Wen-Ming Yang8, Jeremy JW Chen*,2,8,10 and Pan-Chyr Yang2,3,9,10 1 Graduate Institute of Life Sciences, National Defense Medical Center, Taipei, Taiwan, ROC; 2NTU Center for Genomic Medicine, National Taiwan University College of Medicine, Taipei, Taiwan, ROC; 3Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan, ROC; 4School of Pharmacy, National Taiwan University College of Medicine, Taipei, Taiwan, ROC; 5Division of Chest Medicine, Department of Internal Medicine, Taichung Veterans General Hospital, Taichung, Taiwan, ROC; 6 Institute of Toxicology, Chung Shan Medical University, Taichung, Taiwan, ROC; 7Division of Thoracic Surgery, Department of Surgery, Taichung Veterans General Hospital, Taichung, Taiwan, ROC; 8Institutes of Biomedical Sciences and Molecular Biology, National Chung-Hsing University, Taichung, Taiwan, ROC; 9Institute of Biomedical Sciences, Academia Sinica, Taipei, Taiwan, ROC

By using microarray and an invasion/metastasis lung cell line model, we identified the DnaJ-like heat shock protein 40, HLJ1, and found that the expression of HLJ1 correlates negatively with cancer cell invasion ability. Overexpression of HLJ1 can suppress cancer cell invasion in vitro. We further characterize the putative promoter region and investigate the transcriptional regulations of human HLJ1. A serial deletion of the 1.2 kb at the 50 flanking region of the human HLJ1 gene was subcloned into a vector containing reporter gene and transfected into human lung adenocarcinoma cell line CL1-0, followed by luciferase activity assay. The results indicated that the region from –232 to þ 176 could drive the basal transcriptional activity of the HLJ1 gene. Sequence analysis of the HLJ1 gene promoter region showed absence of a TATA box, but identified an inverted CCAAT box and four YY1 transcriptional factor-binding sites, which may be important in the regulation of HLJ1 expression. Co-transfection of the YY1 and HLJ1 basal promoter regions, site-directed mutagenesis, and electrophoretic mobility shift assay confirmed that YY1 could upregulate HLJ1 basal promoter activity. Furthermore, we also demonstrated that overexpression of YY1 in CL10 cells can increase HLJ1 expression and reduce cell invasive capability. The reduction of cancer cell invasive ability is, at least in part, through upregulation of Ecadherin expression. The increase in HLJ1 and Ecadherin expression, as well as the suppression of invasion ability, can be reversed specifically by HLJ1 siRNA. Oncogene (2005) 24, 4081–4093. doi:10.1038/sj.onc.1208573 Published online 21 March 2005 Keywords: DnaJ-like HSP40; invasion suppressor; promoter; YY1; E-cadherin

*Correspondence: JJW Chen, Institute of Biomedical Sciences, National Chung-Hsing University, Taichung 40227, Taiwan, ROC; E-mail: [email protected] 10 These authors co-directed the project and contributed equally Received 17 September 2004; revised 20 January 2005; accepted 20 January 2005; published online 21 March 2005

Introduction Almost all organisms synthesize a group of proteins called heat shock proteins (HSPs) in response to various stresses, for example, heat shock, exposure to heavy metals, ethanol, amino-acid analogues, sodium arsenite, and oxidative stress (Welch, 1992; Hendric and Hartl, 1993; Craig et al., 1994). HSPs have been classified into six major families by their molecular size, including Hsp100, Hsp90, Hsp70, Hsp60, Hsp40, and small HSPs. Within each gene family are members that are either constitutively expressed or inducibly regulated, targeting different compartments under various stress conditions (Caroline and Richard, 2000). The HSPs may function as molecular chaperons to protect cells from environmental stress damage by binding to partially denatured proteins, dissociating protein aggregates, and regulating correct folding (Hartl, 1996). In addition, they also participate in transporting newly synthesized polypeptides to the target organelles for final packaging, degradation or repair, and constitute a major triage system to control the protein quality (Gottesman et al., 1997). HLJ1 (Human Liver DnaJ-like protein) was first identified from the human liver cDNA library by the yeast two-hybrid method. The DNA sequence analysis showed that it contained the J and G/F domain sequence of the Hsp40 family members (Hoe et al., 1998; Ohtsuka and Hata, 2000). The amino-acid sequencing revealed that human Hsp40 is a mammalian homologue of the bacterial DnaJ HSP (Hattori et al., 1992; Ohtsuka, 1993). The main function of DnaJ is to directly interact with DnaK, followed by upregulation of ATPase activity, and act as a chaperon in conjunction with DnaK (Georgopoulos, 1992). To date, several eukaryotic homologs of DnaJ, like Hsp40 proteins, have been identified in various organisms, ranging from yeast to humans (Hamajima et al., 2002). In our previous study, with an attempt to identify the candidate genes associated with cancer invasion and metastasis, we screened an invasion lung cancer cell line

The invasion suppressor HLJ1 promoter is regulated by YY1 C-C Wang et al

4082

model (CL1-0, CL1-1, CL1-5, and CL1-5-F4 in order of increasing invasion activity) by cDNA microarray and identified a panel of metastasis-associated genes (Chen et al., 2001). HLJ1 is one of the metastasis suppressor genes selected from the above-mentioned study. We found that the HLJ1 expression level is negatively correlated with invasion ability of the lung cancer cell lines. Overexpression of the HLJ1 gene could reduce lung cancer cell invasive capability. Owing to the potential pivotal role of HLJ1 in cancer invasion, we further explore the molecular mechanisms for the transcriptional regulation of the human HLJ1 gene. We cloned its 50 -flanking region from genomic DNA. By serial deletions, luciferase transcriptional activity analysis, and functional assays, we were able to identify and characterize the basal promoter region of the HLJ1 gene. Our results also indicate that transcriptional factor YY1 expression can positively regulate HLJ1 gene expression, thus reducing lung cancer cell invasive capability. Furthermore, overexpression of HLJ1 can suppress lung cancer cell invasion through upregulation of E-cadherin expression, and the effects can be specifically reversed by HLJ1 small interfering RNA (siRNA).

Results HLJ1 gene expression negatively correlated with lung cancer cell invasive capability The cDNA microarray analysis identified that the expression of HLJ1 mRNA was negatively correlated with cell invasiveness (Figure 1a). To further confirm that the invasive phenotype of the cancer cells correlated with HLJ1 expression, HLJ1 cDNA was cloned into the pTRE2 vector (Clontech) and then transfected into CL1-5 cells previously transfected with pTet-off vector (Clontech) to establish the stable and controllable HLJ1-expressed cell lines. The expression level of HLJ1 can be downregulated in these cells by adding tetracycline. Stable HLJ1 transfectant cultured without tetracycline (2H12) revealed a 12-fold level over cells cultured with tetracycline (1 mg/ml, 2H12-Tet) and a 21-fold level over wild-type (NT) CL1-5 and mock transfection 1C10 cells, showing only a faint HLJ1 signal (Figure 1b). These three cell lines (2H12, 1C10, and CL1-5 cells) were analysed for in vitro invasion ability (Figure 1c). 2H12 cells showed significantly lower invasion capability than 2H12-Tet, 1C10, and CL1-5 cells (a ¼ 0.05, Po0.05). These results indicate that overexpression of the HLJ1 gene could reduce lung cancer cell invasive capability and may confirm that HLJ1 serves as an invasion suppressor. Cloning and sequence analysis of the human HLJ1 gene promoter To isolate the promoter region of the human HLJ1 gene, a PCR-based amplification strategy was utilized as described in experimental procedures. Using this method, Oncogene

Figure 1 HLJ1 expression levels are negatively correlated with the invasive capability of the lung cancer cell lines. (a) Close-up views of cDNA microarray images show higher expression of HLJ1 in less invasive cell lines (arrowheads). The cell lines in order of invasive activity are as follows: CL1-0oCL1-1oCL1-5. (b) The CL1-5 cells stably transfected with either pTRE2-HLJ1 (2H12) or pTRE2 vector (1C10) were established in this study. These cells were grown for 3 days in the presence (off) or in the absence (on) of 1 mg/ml tetracycline. Total RNAs were extracted and analysed for the expression levels of HLJ1 by real-time quantitative RT–PCR. (c) Overexpression of HLJ1 suppressed the in vitro invasion activity of CL1-5 cells. Suppression of HLJ1 gene expression in 2H12 cells (2H12-Tet) significantly restored the cell invasive capability in vitro (a ¼ 0.05, Po0.05). Results are means7s.d. of the data from three independent experiments

approximately 1214 bp of the 50 -flanking region was obtained, cloned into a promoterless pGL3-basic vector to give a pGL3-F2RER0 construction, and then sequenced (Figure 2). The sequence analysis performed by the ProScan program revealed no canonical TATA


4083

Figure 2 Nucleotide sequence of the 50 -flanking region of the human HLJ1 gene. A total of 1214 bp of the HLJ1 50 -flanking region was cloned and sequenced. This sequence has been submitted to GenBankTM (Accession number: AY669319). The mRNA initiation site is indicated by a horizontal arrow, marked with a bold base and designated with þ 1 in bold letters. The sequence is numbered relative to the transcription start site. Putative transcription binding sites are underlined, and the binding transcription factors are shown below the lines. Four potential YY1-binding sites were identified within the HLJ1 basal promoter region using TRANSFAC matrices (Quandt et al., 1995)

box (TATAAA) within the 1.0 kb region upstream of the transcriptional start site. Further, several potential transcriptional elements were identified by the MatInspector v2.2 program (core similarity 0.8, matrix similarity 0.9) using TRANSFAC matrices (Quandt et al., 1995) (Figure 2). These elements included the potential binding sites for YY1, GR, HiNF-A, AP-2, c-Myb, IRF-2, regulatory elements such as GC box and inverted CCAAT box, and so on. The nucleotide sequence for HLJ1 promoter reported in this paper has been deposited in the GenBankTM database with accession number AY669319. To determine the transcriptional start site(s) of the human HLJ1 gene, 50 -rapid amplification of cDNA ends (50 -RACE) was performed using RNA isolated from CL1-0 cells. Following the secondary nested PCR reaction, the amplified fragment was isolated, cloned and then sequenced. The result showed that the transcriptional start site was located on the 176 bp upstream of the translational initiation site (ATG) (Figure 2). Cell-specific promoter activity of the human HLJ1 gene To examine the promoter activity of the HLJ1 gene in human lung cancer cells, a series of promoter fragments with 50 -end deletion were generated by PCR and ligated to the pGL3-basic vector. Luciferase plasmids containing full-length or deleted HLJ1 upstream DNA fragments were transiently cotransfected with

b-galactosidase (b-gal) expression plasmid into CL1-0 cells. Luciferase activity was measured in crude cell lysate. In all experiments, promoter activity was expressed as a ratio relative to a promoter-less construct and normalized to b-gal activity. All constructs are presented in Figure 3a. As shown in Figure 3b, the plasmid containing the entire 1214 bp upstream of the initiation codon (pGL3-F2RER0 ) had higher luciferase activity in CL1-0 cells, resulting in an approximate 27fold increase in luciferase activity as compared to the pGL3-basic vector (i.e. the negative control). Sequential deletion of 661 bp from the 50 -end of the HLJ1 promoter region (pGL3-F3RER0 ) resulted in almost the same promoter activity compared to the construct containing the entire 1214 bp region (1038/ þ 176). However, deleting nucleotides from –232 to –122 (pGL3-F6RER0 ) eliminated luciferase activity, leaving only control level activity for the plasmid containing the 298 nucleotides of the cloned region (–122/ þ 176). Taken together, we considered that there might be important elements for the basal promoter activity between –232 and –122. In order to better understand the differential regulation of HLJ1 expression in different lung adenocarcinoma cell lines with varying metastasis activity (Chen et al., 2001), we used all constructs for transient transfections of the two human lung cell lines (CL1-0 and CL1-5), which endogenously expressed high and low levels of HLJ1, respectively (Figure 3c). We found that HLJ1 promoter activity patterns were similar between these two cell lines; however, the strength of the promoter activity in Oncogene


4084

Figure 3 Functional deletion mapping of the HLJ1 promoter. (a) 50 deletional constructs of the HLJ1 promoter were illustrated. The numbers on the left of each promoter deletion construct refer to the beginning of the promoter fragments. The transcription start site in the promoter fragment is indicated by an arrow and þ 1. (b) Each construct was transiently co-transfected with the pSV-b-gal vector into CL1-0 cells, and results are expressed as relative luciferase activity. Transfections were carried out in duplicate, and the individual experiment was repeated three times. The results are the mean7s.d. from three separate experiments. *a ¼ 0.05, Po0.01 compared with pGL3-F6RER0 . (c) The Northern blotting analysis of HLJ1 mRNA in CL1-0 and CL1-5 cell lines. GAPDH was as an internal control

CL1-0 cells was about 10-fold higher than in CL1-5 cells (data not shown). YY1 overexpression increases HLJ1 expression and reduces CL1-0 cell invasive activity In the sequence analysis of the putative basal promoter region of HLJ1, we found that there were four probable Oncogene

YY1-binding sites located within this region (Figure 2). To determine whether the putative transcriptional factor YY1 could regulate HLJ1 gene promoter activity, pcDNA3-YY1 construct and HLJ1 promoter constructs were co-transfected into CL1-0 cells, followed by promoter activity assay. As shown in Figure 4, the results indicated that HLJ1 basal promoter activity was positively correlated with the co-transfected amount of YY1 construct. YY1 increased the transcriptional activity of the HLJ1 basal promoter in CL1-0 cells in a dosedependent manner. However, the pGL3-F6RER0 construct, which is without these putative YY1-binding sites, had no effect in HLJ1 basal promoter activity, even under YY1 co-transfection, and identical with the pGL3-basic control (Figure 4b). To further demonstrate a function for YY1 expression in HLJ1 gene regulation, pcDNA3YY1 construct was stably transfected into CL1-0 cells, and three clones (PCY2, PCY3, and PCY5) that stably expressed YY1 were selected for further experiments. These three stable YY1-transfected clones expressed higher levels of YY1 proteins than the control clone transfected with pcDNA3 vector alone (PCC2) and CL10 cells. The result also indicated that the more YY1 expressed the more HLJ1 translated (Figure 5a). In addition, these data were also consistent with the result of RTQ-RT–PCR analysis (Figure 5b). To further investigate whether YY1 expression affected the cancer cell invasion, an in vitro invasion assay was performed. As shown in Figure 6a and b, YY1 transfectant (PCY3) with higher HLJ1 expression levels were able to reduce cell invasion capability to about 57 and 61% of levels, as compared with CL1-0 and mock transfectant (PCC2), respectively. Also, to address the functional consequence of reduced HLJ1 expression within the YY1 overexpression cell line, we evaluated the ability of PCY3 cells to invade through Matrigel-coated Transwells following transfection with HLJ1 siRNAs. The expression levels of the HLJ1 gene within siRNA-transfected cells were reduced by 70% (PCY3-siA), 64% (PCY3-siD), and 16% (PCY3-siL). The invasion of HLJ1 siRNA-transfected cells recovered to 75% (PCY3-siA) and 87% (PCY3siD), compared with PCC2 cells, respectively. However, PCY3 cells transfected with scrambled siRNA (PCY3siL) showed no effect on the invasion capability as compared with PCY3 cells. Thus, we conclude that the observed invasive suppression of YY1 stably transfecting cells is a result of HLJ1 induction and that the HLJ1 siRNAs are able to restore the invasive capability of these cells. The expression levels of HLJ1 in these cell clones are shown in Figure 6a. An intact YY1 DNA-binding domain is required for HLJ1 promoter transactivation To determine whether DNA binding by YY1 was required for HLJ1 promoter activity, we used two mutant YY1 expression plasmids disrupting the various fragments of zinc-finger domains known to be required for DNA binding in the previous study (Yao et al., 2001). We studied the ability of both the WT and zinc-finger domain mutant constructs to transactivate HLJ1 basal promoter


4085

Figure 4 Overexpression of YY1 stimulates HLJ1 promoter activity. (a) 50 deletion constructs of the HLJ1 basal promoter were illustrated. (b) CL1-0 cells were transiently co-transfected with deletion construct (1.5 mg of HLJ1 promoter F5RER0 construct or F6RER0 construct or pGL3-basic vector) and different amounts of pcDNA3-YY1 plasmid by using the LipofectAMINE method. After transfection, cells were cultured in 10% FBS medium for 44 h, and promoter activity was assayed as described in the experimental procedures. Data are the mean7s.d. from three independent experiments. YY1 overexpression increased luciferase activity in a dose-dependent manner. pGL3-basic vector was used as a negative control

construct, pGL3-F5RER0 , in transient transfection assays. Unlike WT YY1, which enhanced the activity of HLJ1 basal promoter examined, co-transfection of zincfinger domain mutants did not transactivate the pGL3F5RER0 reporter construct (Figure 7a). These data suggest that YY1 transactivate the HLJ1 promoter by directly binding to the basal promoter. Since the sequence inspection revealed that four potential YY1-binding sites located within the HLJ1 basal promoter region (Figure 2), we performed EMSA assays to determine whether YY1 could bind to these potential binding sites. Oligonucleotides encompassing the regions from site 1 (242, 218), site 2 (221, 197), site 3 (195, 171), and site 4 (164, 140), either the WT or mutants were radiolabeled, incubated with nuclear extract from CL1-0 cells, and analysed by nondenaturing polyacrylamide gel electrophoresis. All four YY1-binding sites displayed similar binding profiles. DNA–protein binding complexes were obtained with WT oligonucleotides (lane 1), whereas with the mutated oligonucleotides these complexes were not visible (lane 6). As expected, excess unlabeled WT oligonucleotides totally competed for the complex formation (lanes 2 and 3). In addition, when the potential YY1 core-binding site was mutated, the mutated YY1 oligo-

nucleotide lost its competitive ability (data not shown). Indeed, in subsequent super-shift analysis using YY1specific antibody, addition of antibodies against YY1 in the binding reaction caused disappearance of the YY1specific band (Figure 7b). The mutant sequences of potential YY1-binding sites used in EMSA assays were engineered into pGL3F41RER0 to produce the reporter plasmids pGL3-M1, pGL3-M2, pGL3-M3, pGL3-M4, and pGL3-M2/3. These constructs were then used for transient transfection assays in CL1-0 cells. When compared with the WT construct pGL3-F41RER0 , the mutants of these four potential YY1-binding sites resulted in different levels of reduction in HLJ1 promoter activity, from 17 to 34%, respectively. Furthermore, the construct with double sites mutant (pGL3-M2/3) significantly reduced HLJ1 promoter activity to 47% of WT activity (Figure 7c). HLJ1 expression inhibits cell invasion partly through upregulation of E-cadherin expression To explore the mechanism of HLJ1 in suppressing cancer cell invasion, a cDNA microarray method was performed in the tetracycline-controlled HLJ1 expression Oncogene


4086

Figure 5 YY1 positively regulates HLJ1 gene expression. (a) Western blotting analysis of HLJ1 expression was performed in CL1-0 cell, mock-transfected CL1-0 cells (PCC2), and YY1transfected CL1-0 cell clones (PCY3, PCY2, and PCY5). Equal nuclear protein extract of each cell clone was analysed by blotting with an YY1-specific monoclonal antibody (top panel) and reblotted with an HLJ1 polyclonal antibody (middle panel), and a TBP monoclonal antibody (lower panel) was the internal control. The expression of HLJ1 protein was higher in YY1-transfected clones than in mock-transfected cells and CL1-0 cells. This result was consistent with the observation of the real-time quantitative RT–PCR analysis (b). HLJ1 expression is positively correlated with YY1 expression

cell lines mentioned above to identify the differentially expressed genes that are downstream of HLJ1. One of the candidate genes was E-cadherin, which correlated positively with HLJ1 gene expression. We thus investigated whether the expression level of E-cadherin in PCY3 cells was under the control of HLJ1 gene expression. As shown in Figure 8a, E-cadherin expression in PCY3 cells was increased by 1.7- and two fold, as compared with CL1-0 and PCC2 cells, respectively. However, E-cadherin expression levels were reduced by 45% (PCY3-siA) and 41% (PCY3-siD), compared with PCY3 cells, respectively, and were close to the levels of CL1-0 and mock transfectant. On the contrary, PCY3siL had no effect on E-cadherin expression. These data are consistent with the results of Western blotting shown in Figure 8b. Furthermore, HLJ1 Tet-off cells (2H12) and mock transfectant cells (1C10) were chosen to confirm that E-cadherin expression was under the regulation of HLJ1 expression. To induce the expression of HLJ1, cells were cultured without tetracycline. As shown in Figure 9a, E-cadherin expression in 2H12 cells was increased by 1.7–2.8-fold as compared with mock transfectants (1C10 and 1C10-Tet) or CL1-5 cells. The E-cadherin expression level was approximately Oncogene

Figure 6 HLJ1 siRNAs block HLJ1 expression and increase the invasive capability of YY1-overexpressed cell lines. PCY3 cells, constitutively overexpressed YY1, were treated without HLJ1 siRNA for mock experiments or with 20 nM HLJ1 siRNA-A, -D, or scrambled siRNA-L. (a) The effects of HLJ1 siRNAs were confirmed by real-time quantitative RT–PCR analysis. RNAmediated interference (siRNA-A and -D) against HLJ1 efficiently decreases the endogenous HLJ1 RNA level in the YY1-tranfected cell line, PCY3. (b) Invasive capability of PCY3 cells was significantly enhanced by HLJ1 siRNAs. *P ¼ 0.02; **Po0.005 as compared with PCY3 cells by Student’s t-test. The data are presented as means7s.d. of the results from three independent experiments

reduced 2.8-fold in 2H12-Tet cells under the control of tetracycline (a ¼ 0.05, Po0.05). A similar result was obtained in Western blotting analysis (Figure 9b). In addition, we also found that overexpressing HLJ1 could enhance approximately three-fold activity of the E-cadherin proximal promoter in co-transfection analysis (Figure 9c). This result is consistent with the data observed in transcriptional and translational analyses. Interestingly, overexpression of HLJ1 downregulated the slug expression (Figure 9d).

Discussion Metastasis is the major cause leading to mortality for most of the cancer patients. Several molecules participating in cancer cell invasion and metastasis have been identified and characterized, such as metalloproteinases, CD44, cadherin, and tissue inhibitors of metalloproteinases (Birch et al., 1991; Liotta et al., 1991; Uleminckx et al., 1991; Sreenath et al., 1992). In our previous study, by using cDNA microarray and invasion cell line model, we have identified dozens of invasion/metastasis-related


4087

genes on a large scale (Chen et al., 2001). In this study, we characterized one of the novel invasion suppressors, HLJ1. The HLJ1 belongs to the DnaJ/Hsp40 family protein. However, the exact function in a normal and diseased state, as well as the regulation of its expression, is still unknown. Overexpression of the HLJ1 gene could reduce lung cancer cell invasive capability. To understand the molecular mechanisms controlling the expression of HLJ1 in cancer cells, we cloned and characterized the human HLJ1 promoter to facilitate

Figure 8 HLJ1 siRNAs decrease E-cadherin expression levels of YY1-transfected cell lines. (a) Real-time quantitative RT–PCR analysis and (b) Western blotting analysis of E-cadherin expression in CL1-0 cells, mock transfectant (PCC2 cells), and PCY3 cells with or without HLJ1 siRNA transfection. Equal amounts of total protein extract from each cell clone were analysed by blotting with an HLJ1-specific polyclonal antibody (top panel) and re-blotted with an E-cadherin polyclonal antibody (middle panel), and a GAPDH monoclonal antibody (lower panel) as an internal control. siRNA against endogenous HLJ1 decreases the expression level of the endogenous E-cadherin gene in PCY3 cells

Figure 7 YY1 transactivates HLJ1 promoter activity by direct binding interaction. (a) Specific induction of the HLJ1 gene promoter activity by YY1 requires its DNA-binding domain. Transient transfection assays were performed with pGL3-F5RER0 as the reporter gene. 1 mg of pGL3-F5RER0 was cotransfected with 3mg of either pCEP4F-YY1-WT or two YY1 mutants that carry different truncations in the DNA-binding domain, pCEP4FYY1(1–333) and pCEP4F-YY1(1–396). Each construct was transiently co-transfected with the pSV-b-gal vector into CL1-0 cells, and results are expressed as relative luciferase activity. *Po0.01; **Po0.05 as compared with pCEP4F-YY1-WT. (b) CL1-0 cell nuclear extracts were used in EMSA with either WT or Mut YY1 binding site 1 oligonucleotide probes as an example. The YY1immunoreactive complex is indicated by the arrow. The YY1specific complex formed on the WT probe, but not on the mutant probe. Unlabeled WT- or AP1-binding oligonucleotide was used as a specific or nonspecific competitor, respectively. The supershift assays were performed with YY1 or AP-1 antibody. The formation of YY1 complex was completely eliminated by the addition of the YY1 antibody. Competitor: þ , 100 WT; þ þ , 300 WT oligonucleotides. Antibody: þ , 0.2 mg; þ þ , 2 mg YY1 monoclonal antibody. (c) Effect of mutations in YY1 binding sites on HLJ1 promoter activity. Mutations were introduced into the pGL3-F41RER0 reporter construct to alter the four potential YY1binding sites predicted in HLJ1 promoter region. Each construct was then transiently transfected into CL1-0 cells. Luciferase activities are measured and expressed relative to the pGL3-basic vector alone. *Po0.05 as compared with pGL3-F41RER0 Oncogene


4088

Figure 9 E-cadherin expression is regulated by HLJ1. The E-cadherin expression level of stably HLJ1-transfected cells was determined by a tetracycline-off expression system. Parental cells (CL1-5), mock transfectant (1C10), and 2H12 cells that overexpressed HLJ1 under the control of tetracycline are described above. (a) Real-time quantitative RT–PCR and (b) Western blotting analysis were performed to demonstrate the expression levels of E-cadherin in these cells. Equal amounts of total protein extract from each cell clone were analysed by blotting with an HLJ1-specific polyclonal antibody (top panel) and re-blotted with an E-cadherin polyclonal antibody (middle panel), and a GAPDH monoclonal antibody (lower panel) as an internal control. (c) Transactivation of the proximal E-cadherin promoter by HLJ1. CL1-5 cells were transfected with different concentrations of HLJ1 expression plasmids (indicated at the bottom) in the presence of the E-cadherin proximal promoter (Hajra et al., 2002). The luciferase activity of E-cadherin proximal promoter construct alone was set at 1.0 and all other luciferase activities were presented relative to this value. HLJ1 overexpression increased luciferase activity in a concentration-dependent manner. (d) Real-time quantitative RT–PCR analysis of slug expression in these cells mentioned above. Reduction of slug expression in HLJ1 overexpressed cell lines (2H12) was observed

the identification of the cis-regulatory elements and the trans-regulatory factors, which are important for transcriptional regulation of HLJ1 expression. Sequence analysis of the 1.2 kb promoter region proximal to the transcriptional starting site revealed that it does not contain a consensus TATA box element but with an inverted CCAAT box (position –247, Figure 2), although the latter is more frequently found at position –70 around in promoters lacking TATA boxes (Mantovani, 1998). As GC-rich sequences are often reported in TATA-less promoters (Bird, 1986), the analysis of the HLJ1 promoter region revealed that it only contains one GC box (position 761). Promoter activity analysis showed that the core promoter region of the HLJ1 gene is located between the –232 and 122 nucleotides. In addition, we also found that there are several putative transcription factor binding sites within the basal promoter region, four potential binding sites for transcription factor YY1 especially. In contrast to the discovery that YY1 stimulates the induction of the grp78 promoter only under stress conditions (Li et al., 1997a), our study revealed that YY1 could activate HLJ1 promoter under nonstressed condition. The YY1 is a 65-kDa multifunctional zinc-finger transcription factor belonging to the human GLIKruppel family of nuclear proteins (Shi et al., 1991; Galvin and Shi, 1997; Ficzycz and Ovsenek, 2002). It Oncogene

can bind to the specific DNA consensus sequence, 50 CGCCATNTT-30 , which is present in many promoters and regulates transcriptional activity by either activation or repression (Galvin and Shi, 1997; Yao et al., 2001). It is generally believed that whether YY1 behaves as a transcriptional activator or repressor depends on its relative concentration as compared with other cell typespecific factors, and the promoter sequences surrounding the YY1-binding sites (Shi et al., 1991; Usheva and Shenk, 1994; Ficzycz and Ovsenek, 2002). In this study, four potential binding sites for transcription factor YY1 were identified within the basal promoter region. Therefore, we try to clarify the relationship between the YY1 transcription factor and HLJ1 promoter activity. After transiently or stably transfecting pcDNA3-YY1 construct into CL1-0 cells, the expression level of HLJ1 correlated positively with YY1 expression (Figures 4 and 5). Interestingly, there is a putative and reversed YY1-binding site located downstream of the basal promoter (61 to 67) within the pGL3-F6RER0 construct. Although the binding site is in reverse orientation, the transcription from this site is still activated about 2.4-fold by YY1 as compared with the pGL3-basic vector alone. Furthermore, our results indicated that the mutant YY1 proteins with deletions of YY1 DNA-binding domains were unable to stimulate the transcriptional


4089

activation of HLJ1 basal promoter-reporter construct (pGL3-F5RER0 ). We thus suggest that the intact DNAbinding domain of YY1 is necessary for the activation of HLJ1 promoter expression. Several mechanisms for YY1 action have been proposed, which may depend only on a functional DNA-binding domain. It could be sufficient for the initiator function of YY1 (Seto et al., 1991; Usheva and Shenk, 1994) or for regulating genes simply through binding to target sequences, either by displacing other regulators or through DNA bending (Natesan and Gilman, 1993; Lu et al., 1994; Meier and Groner, 1994; Zhou and Engel, 1995; Guo et al., 1997). The previous studies showed that YY1 could activate several cellular genes including rpL30, c-Myc, grp78, hGR, dihydrofolate reductase and major histocompatibility complex (MHC) or DRa genes (Hariharan et al., 1991; Azizkhan et al., 1993; Riggs et al., 1993; Hehlgans and Strominger, 1995; Li et al., 1997a). In this study, using site-directed mutagenesis, we found that four YY1-binding sites had varying degrees of effect on HLJ1 promoter activation. Although the YY1-binding sites within the HLJ1 basal promoter region differ from the ones in those genes mentioned above, these four binding sites are all functional. The effect of the individual binding site on HLJ1 promoter activation was different and mild, a decrease from 17 to 34%. Even the double binding sites mutation could only decrease the promoter activity to 47%. The results suggest that other elements may be able to compensate for the transcriptional activation while the function of one binding site is impaired. However, deletion of all four YY1-binding sites did show a 93% reduction in HLJ1 promoter activity, indicating that YY1 regulation is important for HLJ1 expression. Cadherins modulate calcium-dependent cell–cell adhesion and are important in cell aggregation and migration (Takeichi, 1991). Loss of E-cadherin and/or catenin expression are associated with invasive tumor growth in a variety of human malignancies (Berx et al., 1998; Hirohashi, 1998), including lung cancer (Bremnes et al., 2002) and breast carcinomas (Bukholm et al., 1998). In this study, our results showed that HLJ1 expression in lung cancer cells could inhibit the ability of their invasion through Matrigel (Figures 1c and 6b). Furthermore, overexpression of the HLJ1 gene is able to positively regulate E-cadherin expression in lung cancer cells (Figures 8 and 9). The loss of functional HLJ1 expression by the siRNA technique also demonstrates the regulatory effect of HLJ1 on E-cadherin expression. In the previous report (Hajra et al., 2002), the authors demonstrated that the E-box elements in the proximal E-cadherin promoter are critical in transcriptional repression of the E-cadherin gene, and slug is a likely in vivo repressor of E-cadherin in breast cancer. In this study, we found that overexpression of HLJ1 gene would downregulate slug gene expression and upregulate E-cadherin expression (Figure 9). Thus, we suggest that HLJ1 indirectly upregulates E-cadherin through inhibiting the repression effect of slug gene on Ecadherin proximal promoter. Further studies will be needed to prove this hypothesis. In addition, using

immunohistochemical staining, we also found that Ecadherin proteins are predominantly expressed in HLJ1 overexpressed lung cancer cells (data not shown). We conclude that overexpression of the HLJ1 gene is able to suppress lung cancer cell invasive activity partly through the upregulation of E-cadherin expression. However, we cannot rule out the possibility that there are other positive and/or negative regulators also involved in this invasive process, especially under the transcriptional regulation of multifunctional transcription factor YY1. In many experimental systems, YY1 can activate transcription expression, and it is uncertain how this is accomplished. Several models, including bending of DNA, the relative distance between the YY1-binding site and the transcriptional initiation site, as well as protein–protein interactions, have been proposed (Shi et al., 1997; Thomas and Seto, 1999). Increasing evidence shows that the interactions with other proteins are probably the most important factors in YY1mediated transcriptional activation. It has been suggested that interactions of YY1 with other cellular proteins or viral proteins can either disrupt the quenching activity of YY1 on other transcriptional activators or stimulate transcription with associated enzymatic activities such as HATs (Shi et al., 1991; Lee et al., 1995; Thomas and Seto, 1999). Therefore, association of YY1 with other transcription factors in the cell may be important in regulating HLJ1 gene expression. However, at the present time, little information is available about the regulation of the interplay between YY1 and HLJ1. The previous reports also indicated that YY1 might be part of a cellular stress response pathway activated by misfolded proteins (Li et al., 1997a, b). Thus, it is possible that effects on cell growth by certain forms of stress could be mediated by YY1-dependent negative regulation of cell growth. In another study (Austen et al., 1998), they demonstrated that YY1 is a potent inhibitor of c-Myc transforming activity and suggested that YY1 might have tumorsuppressing properties. HLJ1 belongs to HSP40 family, which can be induced by heat shock and serves as a cochaperon involved in protein folding and renaturation after stress. Furthermore, in our unpublished data, we also found that overexpression of HLJ1 could inhibit lung cancer cell growth. However, whether these effects are the results of negatively regulating c-Myc function is unclear so far and has to be addressed in future studies. Actually, the complex relationships among YY1, HLJ1, and E-cadherin are still unknown. In summary, the basal promoter at the 50 -flanking region of the novel invasion suppressor, HLJ1, has been identified and characterized. The transcription factor YY1 can positively regulate HLJ1 gene expression and reduce cell invasive capability, at least in part, through upregulation of E-cadherin expression. This study confirms that both the transcriptional factor YY1 and HLJ1 are involved in the regulation of the mechanisms of sophisticated cancer invasion. Further studies are necessary to elucidate the possible downstream signaling pathways of HLJ1 that suppress cancer cell invasive ability. Oncogene


4090 Materials and methods Cell culture Human lung adenocarcinoma cell lines of different invasive and metastatic capacities (CL1-0 and its sublines, CL1-1 and CL1-5) (Chu et al., 1997) were maintained at 371C in a humidified atmosphere of 5% CO2. Cells were cultured in RPMI 1640 medium (Life Technologies, Inc., Gaithersburg, MD, USA) with 10% heat-inactivated fetal bovine serum (FBS; Life Technologies, Inc.) and 1% penicillin streptomycin (Life Technologies, Inc.). Microarray analysis cDNA microarray membranes were prepared as described previously (Chen et al., 2001). Briefly, mRNA was extracted from each CL cell line and reverse-transcribed into cDNA labeled with biotin. The biotin-labeled cDNAs were used as targets and hybridized with the membranes that contained 9600 nonredundant expressed sequence tag clones selected from human cDNA libraries. After stringent washing, a color reaction toward biotin was initiated and detected by colorimetric measurement. Cloning and sequencing of the 50 -flanking region of the HLJ1 gene To clone the 50 -flanking region of the HLJ1 gene, a PCR-based method was utilized. Specific primers were designed from the 50 -end of the known HLJ1 cDNA sequence (Hoe et al., 1998) and from the bioinformatic search in GenBank. CL1-0 cell genomic DNA was isolated by a QIAamp DNA blood mini kit (Qiagen) and served as a PCR template. The sequences of the primer set used in PCR amplification are described as follows: HLJP-F2 primer, 50 -CCGCTCGAGGGATTACCTAAAAT GATATTATAGG-30 , which introduced an XhoI site (underlined), and HLJP-RER0 primer, 50 -CCCAAGCTTTTCGAAT GCCTTGAAATTAAC-30 , which introduced a HindIII site (underlined). The PCR conditions are as follows: the first cycle, 941C for 2 min 30 s, 551C for 1 min, and then 721C for 3 min; the second to 35th cycle, 941C for 40 s, 601C for 1 min; and then 721C for 3 min; and a final extension for 10 min at 721C. The amplified DNA fragment of 1214 bp was digested with XhoI/HindIII and cloned into the promoterless pGL3-Basic vector (Promega) to construct pGL3-F2RER0 . The fragment was sequenced and was found to be contiguous with the HLJ1 cDNA. Homology searches were performed using BLAST (Basic Local Alignment Search Tool) from the National Center for Biotechnology Information (NCBI) at http://www.ncbi.nlm.

nih.gov. Putative transcription factor binding elements in the HLJ1 promoter were analysed using the programs MatInspector 2.2 (Quandt et al., 1995) and SignalScan (Prestridge, 1991) (both at http://thr.cit.nih.gov/molbio/signal/ and http:// www.genomatix.de/products/index.html) using the TRANSFAC database (Heinemeyer et al., 1999). The programs GrailEXP (http://compbio.ornl.gov/grailexp/) and ProScan (http://bimas.dcrt. nih.gov/molbio/proscan) were used for CpG island and promoter region prediction. Construction of luciferase reporter gene constructs Varying lengths of the 50 -flanking region of the HLJ1 gene for luciferase assays were generated by PCR using the pGL3F2RER0 clone as the template. A common reverse primer and different forward primers, given in Table 1, were used for the amplification of various deletion fragments. XhoI and HindIII sites were introduced in the forward and reverse primers, respectively, and used for cloning of these fragments upstream of the luciferase reporter gene in promoterless pGL3-basic vector. The pGL3-Control, a positive control plasmid, was also obtained from Promega. For site-directed mutagenesis assay, the pGL3-F41RER0 construct served as a template for the generation of mutations in the YY1-binding sites. All mutant constructs were prepared by PCR using appropriate primers containing the mutations just the same as in EMSA assay. The composition of all of the constructs was confirmed by restriction endonuclease digestion and DNA sequencing. 50 -Rapid amplification of cDNA ends (RACE) Transcription start sites were identified by a 50 -rapid amplification of cDNA ends (50 -RACE) method described previously (Matz et al., 1999) with few modifications. Briefly, 10 mg total RNA isolated from CL1-0 cells was reverse transcribed by Superscript RT II (Invitrogen Life Technologies) using the T20 primer (50 -TTTTTTTTTTTTTTTTTTTT-30 ) and CapSwitch primer (50 -AAGCAGTGGTATCAACGCAGAGTACGCrGr GrG-30 ) (Matz et al., 1999). Reverse transcription PCR was first performed on a DNA cycler at 421C for 1 h and 941C for 5 min. In all, 1 ml of the first-stranded cDNA was put into 50 ml of PCR mixture with TSP primer (50 -GCAGTGGTAT CAACGCAGAG-30 ) and QHLJ1-R primer (50 -CCATC CAGTGTTGGTACATTAATT-30 ). The PCR conditions consisted of one cycle for 2 min 30 s at 941C, 1 min at 551C, and 3 min at 721C, followed by 34 cycles for 40 s at 941C, 1 min at 601C, and 3 min at 721C, and then a final extension step at 721C for 10 min was added. The reverse transcription PCR products were separated on 1% argarose gel, and the 50 -RACE products were purified, subcloned into PCRII TOPO vector

Table 1 Primer sequences used for constructions of HLJ1 promoters with different lengths Amplification primer

Primer sequence (50 to 30 )a

Promoter forward primers HLJP-F2 HLJP-F3 HLJP-F31 HLJP-F4 HLJP-F41 HLJP-F5 HLJP-F6

CCGCTCGAGGGATTACCTAAAATGATATTATAGG CCGCTCGAGTAGAATTGTCGTTCCTTTTATCTGT CCGCTCGAGGTTTAATGACTGTGATGATTT CCGCTCGAGATTTTCTCCTAGTATGGAGTACATA CCGCTCGAGTAAATATATAGTTGTGACATTCTGTG CCGCTCGAGCATTTGTCCTGTTTAATTAGGAAA CCGCTCGAGGGAAAGTGACGTCCTGTA

Promoter reverse primer HLJP-RER0

CCCAAGCTTTTCGAATGCCTTGAAATTAAC

a

Restriction enzyme cutting sites located within PCR primers are underline. XhoI site: CTCGAG; HindIII site: AAGCTT

Oncogene


4091 (Invitrogen) according to the manufacturer’s instructions and then sequenced. Transfection and luciferase assays All transfections were carried out in triplicate in six-well plates. About 2x105 cells/well were seeded for 24 h prior to transfection. Plasmids were transfected into cells using Lipofectamine reagent according to the manufacturer’s instructions (Invitrogen). The luciferase reporter constructs described above, along with the control plasmid, were cotransfected with a b-gal construct, pSV-b-Gal (Promega). The ratio of the DNA amounts for luciferase reporter constructs versus b-gal construct was 3 : 1. The cells were incubated in transfection mixture for 4 h and then harvested after 44 h in culture. CL1-0 cells were stably transfected by YY1 expression plasmid using Lipofectamine reagent. YY1 stably expressed clones (PCY2, PCY3, and PCY5) were obtained after G418 selection (400 mg/ ml). Mock transfectant (PCC2) was also included in our studies. To generate tetracycline-responsive HLJ1 transfectants (2H12), pTRE2-HLJ1, and pTK-Hyg plasmids were cotransfected into CL1-5 Tet-off cells according to manufacturer’s instruction (Clontech). At 14 days after transfection, several colonies were isolated and maintained in medium supplemented with 1 mg/ml of tetracycline in 24-well plates. To induce the expression of HLJ1, cells were cultured without tetracycline. Mock transfectant (1C10) was also obtained from the hygromycin selection. For stable transfection experiments, YY1 expression plasmid was transfected into CL1-0 cells, and HLJ1 expression plasmid was transfected into CL1-5 cells using Lipofectamine reagent and selected for growth in G418 (400 mg/ml). For co-transfection experiments, a constant amount of HLJ1 promoter–reporter luciferase plasmid or pGL3-basic vector DNA and YY1 expression plasmid in different ratios, plus 1 mg of internal control pSV-m-gal plasmid, were included. The pCEP4F-YY1 construct was generated by inserting the full-length YY1 cDNA into the pCEP4F vector. The deletion mutants of YY1 DNA-binding domain, pCEP4F-YY1(1–333) and pCEP4F-YY1(1–396), were made by restriction enzyme digestion and re-ligation of pCEP4F-YY1 (Yao et al., 2001). An aliquot of cell lysate (10– 25 ml) was used to assay luciferase activity using a luciferase assay kit (Tropix, Inc., Bedford, MA, USA). Another aliquot (10–25 ml) was used to measure m-gal activity using the Galacto-Light chemiluminescent assay kit (Tropix, Inc., Bedford, MA, USA). Luminescence was measured with a Victor2 1420 Multilabel Counter (Wallac). Transfection efficiency was normalized with b-gal activity. Each experiment was repeated at least three times. Northern and Western blot analysis Northern blot analyses were performed using the procedures described previously (Shih et al., 2001). The RNA in each lane was measured by comparing its signal intensity with that of the GAPDH probe, which was used as an internal control for RNA quantity. The details of nuclear extract preparation and Western blot analysis have been described previously (Chen et al., 2003). Total cell lysates were isolated from cells as described previously (Chen et al., 2002). HLJ1 and YY1 were detected using a 1 : 1500 dilution of mouse polyclonal antiHLJ1 (made in-house) and 1 : 1000 dilution of mouse monoclonal anti-YY1 primary antibodies (Santa Cruz Biotech), respectively, a 1 : 3000 dilution of HRP-conjugated goat anti-mouse IgG secondary antibody (Santa Cruz Biotech) using an enhanced chemiluminescence detection system (ECL, Amersham Biosciences). E-cadherin was detected using a

1 : 1000 dilution of rabbit polyclonal anti-E-cadherin antibody (Santa Cruz Biotech) and a 1 : 2000 dilution of HRPconjugated goat anti-rabbit IgG secondary antibody (Santa Cruz Biotech). TATA box-binding protein (TBP) or GAPDH was used as the internal controls for gel loading. Real-time quantitative RT–PCR HLJ1, E-cadherin, and slug mRNA expression was quantified by real-time quantitative RT–PCR (RTQ-RT–PCR). The sequences of the primer set used in RTQ-RT–PCR amplification for the HLJ1 gene are described as follows: Q-DJ-F1 primer, 50 -CCAGCAGACATTGTTTTTATCATT-30 and QDJ-R1 primer, 50 -CCATCCAGTGTTGGTACATTAATT-30 . The TaqMan probe used for HLJ1 is 50 -CAGCCACA CAATGCCTCTCGTAAACTAAT-30 . The sequences of the primer sets used in SYBR-Green RTQ-RT–PCR amplification for the E-cadherin and slug genes are described as follows: E-cad-F primer, 50 -AAGGTGACAGAGCCTCTGGAT-30 , E-cad-R primer, 50 - ATTCCCGTTGGATGACACA-30 , Slug-F primer, 50 -AGAACTCACACGGGGGAGAAG-30 , and Slug-R, 50 -CTCAGATTTGACCTGTCTGCAAA-30 . The TATA box-binding protein (TBP) was used as an internal control. The primers, probes, and detailed procedures have been described previously (Chen et al., 2003). Briefly, each amplification mixture containing 10 ng of total RNA was subjected to one cycle of reverse transcription and 40 cycles of the polymerase chain reaction. All experiments were performed in triplicate. The relative expression level of HLJ1 against that of TBP was defined as DCT ¼ [CTHLJ1– CTTBP]. The HLJ1 mRNA/TBP mRNA ratio was calculated as 2DCT K (K: constant). siRNA transient transfection Desalted siRNA duplexes were synthesized by Qiagen and were annealed following its standard protocol. The siRNA sequences targeting the human HLJ1 gene are HLJ1-A: AAC CCGGAATGAGGAGAAGAA, and HLJ1-D: AAACGCT GATGGAAGGAGTTA. A scrambled siRNA (HLJ1-L: GG ACAATGAACACGAGGAAGA) was used as the control. siRNAs were transfected using the RNAiFect Transfection Reagent (Qiagen) according to the manufacturer’s instructions. Matrigel invasion assay Invasiveness of the CL1-0 cells and siRNA transfectants were examined by using the membrane invasion culture system as described previously with some modification (Chen et al., 2001). Briefly, Transwell membranes (8 mm pore size, 6.5 mm diameter; Corning Costar Corporation, Cambridge, MA, USA) were coated with Matrigel (2.5 mg/ml; BD Biosciences Discovery Labware, Bedford, MA, USA), and dry coatings were reconstituted in RPMI for 1–2 h before cell passage. Cells were trypsinized, centrifuged, and resuspended at a density of 106 cells/ml in RPMI containing 10% FBS. Later they were seeded onto the upper chambers of precoated transwells. Lower chambers of the transwells contained the same medium. After 18 h of incubation, membranes coated with Matrigel were removed by wiping with a cotton swab, fixed with methanol, and stained with Giemza stain (Sigma) before cell counting under phase-contrast microscopy. Electrophoretic mobility shift assay (EMSA) Double-stranded oligonucleotides were prepared by heating at 801C for 20 min before slowly cooling to room temperature. Oncogene


4092 Oligonucleotides were labeled using [g-32P]ATP (3000 Ci/ mmol) and T4 polynucleotide kinase. Labeled probes were purified from unincorporated [g-32P]ATP using MicroSpin G25 columns. Nuclear extract (5 mg of protein) was incubated for 20 min at room temperature in binding buffer (4% glycerol, 1 mM MgCl2, 0.5 mM EDTA, 0.5 mM DTT, 50 mM NaCl, 10 mM Tris-HCl (pH 7.5), and 0.1 mg of poly[d(I-C)]) containing g-32P-end-labeled, double-stranded oligonucleotide in a final volume of 10 ml. Samples were then resolved by electrophoresis on 4% polyacrylamide gels at 110 V in 1 Tris/borate/EDTA buffer for 150 min at 41C. Gels were dried and placed on a phosphoimage screen overnight. For competition assays, unlabeled oligonucleotides were added to the binding reagents at a 100-fold molar excess 10 min prior to the addition of radiolabeled probe. For antibody supershift analysis, binding reactions were incubated with 2 mg of antibody for 30 min at 41C before the addition of probe. Oligonucleotides used were as follows: YY1-WT1: CATGT TAACCCATTTGTCCTGTTTA, YY1-M1: CATGTTAACC AGCAGGTCCTGTTTA, YY1-WT2: TTTAATTAGGAAA TAAAGACGGTAG, YY1-M2: TTTAATTAGGCTGCCA

AGACGGTAG, YY1-WT3: GCTAGTCGGAAAATGAA TAGTTAAT, YY1-M3: GCTAGTC-GGACTGCTAA TAGTTAAT, YY1-WT4: GGAGGAGAAAACATAGCTC AGTACT, YY1-M4: GGAGGAGAAACTGAAGCTCAGT ACT, and AP1: AAAGAATTGCTGAATCATC-ATTGCT. Mutations were introduced into the YY1 WT-binding sites (boldface and underline). Statistical analysis All experiments were performed in triplicate and analysed by ANOVA (Excel, Microsoft; Taipei, Taiwan) for significant differences. Po0.05 was considered statistically significant. Where appropriate, the data are presented as the mean7s.d. Acknowledgements This work was supported by the National Science Council and National Health Research Institutes of the Republic of China through the National Research Program for Genomic Medicine grants (NSC91-3112-P-005-008-Y, NHRI92A1NSCLC09-5, and NHRI93A1-NSCLC09-5).

References Austen M, Cerni C, Luscher-Firzlaff JM and Luscher B. (1998). Oncogene, 17, 511–520. Azizkhan JC, Jensen DE, Pierce AJ and Wade M. (1993). Crit. Rev. Eukaryot. Gene Expr., 3, 229–254. Berx G, Becker KF, Hofler H and Van Roy F. (1998). Hum. Mutat., 12, 226–237. Birch M, Mitchell S and Hart IR. (1991). Cancer Res., 51, 6660–6667. Bird AP. (1986). Nature, 321, 209–213. Bremnes RM, Veve R, Hirsch FR and Franklin WA. (2002). Lung Cancer, 36, 115–124. Bukholm IK, Nesland JM, Karesen R, Jacobsen U and Borresen-Dale AL. (1998). J. Pathol., 185, 262–266. Caroline J and Richard IM. (2000). J. Natl. Cancer Inst., 92, 1564–1572. Chen JJW, Peck K, Hong TM, Yang SC, Sher YP, Shih JY, Wu R, Cheng JL, Roffler SR, Wu CW and Yang PC. (2001). Cancer Res., 61, 5223–5230. Chen JJW, Yao PL, Yuan A, Hong TM, Shun CT, Kuo ML, Lee YC and Yang PC. (2003). Clin. Cancer Res., 9, 729–737. Chen HW, Chien CT, Yu SL, Lee YT and Chen WJ. (2002). Br. J. Pharmacol., 137, 771–781. Chu YW, Yang PC, Yang SC, Shyu YC, Hendrix MJC, Wu R and Wu CW. (1997). Am. J. Respir. Cell Mol. Biol., 17, 353–360. Craig EA, Weissman JS and Horwich AL. (1994). Cell, 78, 365–372. Ficzycz A and Ovsenek N. (2002). J. Biol. Chem., 277, 8382–8387. Galvin KM and Shi Y. (1997). Mol. Cell Biol., 17, 3723–3732. Georgopoulos C. (1992). Trends Biochem. Sci., 17, 295–299. Gottesman S, Wickner S and Maurizi MR. (1997). Genes Dev., 11, 815–823. Guo B, Aslam F, van Wijnen AJ, Roberts SGE, Frenkel B, Green MR, Deluca H, Lian JB, Stein GS and Stein JL. (1997). Proc. Natl. Acad. Sci. USA, 94, 121–126. Hajra KM, Chen DY and Fearon ER. (2002). Cancer Res., 62, 1613–1618. Hariharan N, Kelley DE and Perry RP. (1991). Proc. Natl. Acad. Sci. USA, 88, 9799–9803. Hamajima F, Hasegawa T, Nakashima I and Isobe KI. (2002). J. Cell. Biochem., 84, 401–404. Oncogene

Hartl FU. (1996). Nature, 381, 571–580. Hattori H, Liu YC, Tohnai I, Ueda M, Kaneda T, Kobayashi T, Tanabe K and Ohtsuka K. (1992). Cell Struct. Funct., 17, 77–86. Hehlgans T and Strominger JL. (1995). J. Immunol., 154, 5181–5187. Heinemeyer T, Chen X, Karas H, Kel AE, Kel OV, Liebich I, Meinhardt T, Reuter I, Schacherer F and Wingender E. (1999). Nucleic Acids Res., 27, 318–322. Hendric JP and Hartl FU. (1993). Annu. Rev. Biochem., 62, 349–384. Hirohashi S. (1998). Am. J. Pathol., 153, 333–339. Hoe KL, Won M, Chung KS, Jang YJ, Lee SB, Kim DU, Lee JW, Yun JH and Yoo HS. (1998). Biochim. Biophys. Acta., 1383, 4–8. Lee JS, Galvin KM, See RH, Eckner R, Livingston D, Moran E and Shi Y. (1995). Genes Dev., 9, 1188–1198. Li WW, Hsiung Y, Wong V, Galvin K, Zhou Y, Shi Y and Lee AS. (1997a). Mol. Cell. Biol., 17, 61–68. Li WW, Hsiung Y, Zhou Y, Roy B and Lee AS. (1997b). Mol. Cell. Biol., 17, 54–60. Liotta LA, Steeg PS and Stetler-Stevenson WG. (1991). Cell, 64, 327–336. Lu S-Y, Rodriguez M and Liao WS-L. (1994). Mol. Cell. Biol., 14, 6253–6263. Mantovani R. (1998). Nucleic Acids Res., 26, 1135–1143. Matz M, Shagin D, Bogdanova E, Britanova O, Lukyanov S, Diatchenko L and Chenchik A. (1999). Nucleic Acids Res., 27, 1558–1560. Meier VS and Groner B. (1994). Mol. Cell. Biol., 14, 128–137. Natesan S and Gilman MZ. (1993). Genes Dev., 7, 2497–2509. Ohtsuka K. (1993). Biochem. Biophys. Res. Commun., 197, 235–240. Ohtsuka K and Hata M. (2000). Cell Stress Chaperones., 5, 98–112. Prestridge DS. (1991). CABIOS, 7, 203–206. Quandt K, Frech K, Karas H, Wingender E and Werner T (MatInd and MatInspector) (1995). Nucleic Acids Res., 23, 4878–4884. Riggs KJ, Saleque S, Wong KK, Merrell KT, Lee JS, Shi Y and Calame K. (1993). Mol. Cell. Biol., 13, 7487–7495. Seto E, Shi Y and Shenk T. (1991). Nature, 354, 241–245.


4093 Shi Y, Lee JS and Galvin KM. (1997). Biochim. Biophys. Acta., 1332, F49–F66. Shi Y, Seto E, Chang LS and Shenk T. (1991). Cell, 67, 377–388. Shih JY, Yang SC, Hong TM, Yuan A, Chen JJW, Yu CJ, Chang YL, Lee YC, Peck K, Wu CW and Yang PC. (2001). J. Natl. Cancer Inst., 93, 1392–1400. Sreenath T, Matrisian LM, Stetler-Stevenson W, Gattoni-Celli S and Pozzatti RO. (1992). Cancer Res., 52, 4942–4947.

Takeichi M. (1991). Science, 251, 1451–1455. Thomas MJ and Seto E. (1999). Gene, 236, 197–208. Uleminckx K, Vackat L, Mareel M, Fiers W and Von Roy F. (1991). Cell, 66, 107–119. Usheva A and Shenk T. (1994). Cell, 76, 1115–1121. Welch WJ. (1992). Physiol. Rev., 72, 1063–1081. Yao YL, Yang WM and Seto E. (2001). Mol. Cell Biol., 21, 5979–5991. Zhou Q and Engel DA. (1995). J. Virol., 69, 7402–7409.

Oncogene