A SNP in the flt-1 promoter integrates the VEGF system into the p53 transcriptional network

Abstract

The VEGF system is essential for angiogenesis. VEGF overexpression frequently correlates with increased microvascularity and metastasis and decreased spontaneous apoptosis. Although a precise mechanism has not been established, studies suggest that VEGF expression is negatively regulated by p53, a master regulator and tumor suppressor. There are no reports of additional components of the VEGF signal transduction pathway being part of the p53 transcriptional network. A target of VEGF, the VEGF receptor 1/flt-1, can regulate growth and migration of endothelial cells and modulate angiogenesis. VEGF appears to be up-regulated in various cancers in which flt-1 may have a role in tumor progression and metastasis. We identified a C-to-T SNP upstream of the transcriptional start site in ≈6% of the people examined. The SNP is located within a putative p53 response element. Only the promoter with the T SNP (FLT1-T) was responsive to p53 when examined with reporter assays or by endogenous gene expression analysis in cell lines with different SNP status. In response to doxorubicin-induced DNA damage, there was clear allele discrimination based on p53 binding at the FLT1-T but not FLT1-C promoters as well as p53-dependent induction of flt-1 mRNA, which required the presence of FLT1-T. Our results establish that p53 can differentially stimulate transcription at a polymorphic variant of the flt-1 promoter and directly places the VEGF system in the p53 stress-response network via flt-1 in a significant fraction of the human population. We suggest that the p53-VEGF-flt-1 interaction is relevant to risks in angiogenesis-associated diseases, including cancer.

VEGF is a multifunctional cytokine that is pivotal to blood vessel formation, including angiogenesis and vasculogenesis (1), and is expressed in a variety of cells (2), including smooth muscle, endothelial, and epithelial cells, monocytes/macrophages, T lymphocytes, and polymorphonuclear neutrophils. Alternative splicing of the human VEGF gene results in at least four isoforms, of which VEGF165 is the most prevalent. VEGF exerts its biological effects through two high-affinity receptors: Fms-like tyrosine kinase 1 (flt-1) receptor 1 (VEGFR-1) and VEGF receptor 2 (VEGFR-2/FLK-1/KDR) (1). These VEGF receptors are found at the surface of endothelial cells, hematopoietic stem cells, leukocytes, osteoblasts, and many tumor cells (3).

FLK-1 is considered the principal receptor of VEGF-dependent angiogenic signals, with an important role in normal vessel growth (1) and hematopoiesis, as well as other functions (4). The precise function of flt-1 is still under debate (5). Park et al. (6) initially proposed that flt-1 might not be primarily a receptor transmitting a mitogenic signal, but rather a “decoy” receptor, able to negatively regulate VEGF activity on the vascular endothelium by preventing VEGF binding to VEGFR-2. In addition, flt-1 is implicated in up-regulation of tissue factor, urokinase-type plasminogen activator, and plasminogen activator inhibitor 1 (7) in endothelial cells. However, direct proliferative, migratory, or cytoskeletal effects mediated by flt-1 receptor remain to be demonstrated (9). In other cell types flt-1 has additional and different roles, such as inducing tissue factor and chemotaxis in monocytes and enhancing matrix metalloproteinase expression in vascular smooth muscle (10, 11). Recent evidence indicates that flt-1 is not only involved in lung-specific metastasis by matrix metalloproteinase 9 induction (12), but is also involved in ligand-induced pathological angiogenesis in tumor cells (13).

Flt-1^-/- mice die at embryonic day 8.5 (14). The endothelial cells form normally but fail to assemble correctly into organized blood vessels, apparently as a consequence of increased commitment of mesenchymal cells to becoming hemangioblasts, the common precursors of both blood and endothelial cells (15).

Levels of flt-1 expression can vary considerably across tissues (e.g., lung, kidney, heart, liver, and brain) and in response to stimuli (e.g., hypoxia) (16). Given the association of flt-1 function with cell growth, development, and various pathologies (see Discussion), variations in expression are expected to contribute to its cellular impact. In this work we searched for possible genetic sources of variation in flt-1 expression focusing on 1,196 bp of the flt-1 proximal promoter. A SNP was found in ≈6% of the population sampled. Surprisingly, the SNP occurred in a putative p53 response element (RE) sequence, with the less frequent SNP resulting in an RE predicted to be more p53-responsive (17). We characterized the influence of the SNP on promoter activity by reporter gene assays as well as endogenous flt-1 gene expression in cell lines heterozygous for the SNP. Analysis of transactivation, promoter occupancy, and responses to genotoxic stress show that only one, the FLT1-T allele, was p53-responsive. Thus, the VEGF system is directly placed in the p53 stress-response transcriptional network via flt-1 in a significant fraction of the human population. For those individuals with the uncommon polymorphic promoter variant, p53-mediated stress responses could markedly influence the role of flt-1 in various biological functions.

Results

Identification of a SNP in the flt-1 Promoter. We used a combination of molecular analysis, custom bioinformatics applications, and in vivo functional analyses to identify and characterize potential SNPs in a 1,196-bp region of the 5′ regulatory region of the flt-1 gene that includes the transcription start site (TSS). By using single-strand conformation polymorphism (SSCP) analysis, only one polymorphic variant was detected among 150 individuals sampled (Fig. 1A). Based on sequencing, this variant, referred to as FLT1-T, had a T rather than a C at a position 677 bases upstream from the transcription start (named C519T according to GenBank accession no. D64016). The frequency of FLT1-C/T heterozygotes in our sample of healthy individuals is 6%. We also genotyped a panel of 19 cancer-derived human cell lines plus the HMEC (human microvascular endothelial cells) endothelial cell line using a simple restriction fragment length polymorphism (RFLP) assay (Table 1, which is published as supporting information on the PNAS web site) because the C/T SNP alleles can be distinguished by Nsp1 digestion (Fig. 1B). The colon carcinoma cell line HCT116 and the lung carcinoma cell line CaLu-1 were heterozygous, and the other cells were C/C homozygous (Fig. 1B and Table 1).

Fig. 1.

Identification of a SNP in the flt-1 promoter region. (A) SSCP analysis of a 200-bp flt-1 promoter fragment (positions 465-665; GenBank accession no. D640169) from healthy volunteers (lanes 1-8). An extra band is seen in lane 2. Sequence analysis revealed a single heterozygous C/T nucleotide polymorphism (C519T) located 677 bases upstream of the TSS (position 519, GenBank accession no. D64016). Bioinformatic analysis predicted that this SNP falls within a potential p53 RE. (B) RFLP analysis of HMEC, MCF-7, and HCT116 cells revealed that only HCT116 cells harbor a heterozygous C/T allele, whereas HMEC and MCF-7 cells are homozygous for the C allele. The T allele SNP results in an NSP1 site that enables the T (+) and C (-) to be distinguished. A DNA sample from an individual with heterozygous C/T alleles was used as a positive control. (C) Schematic representation of the putative p53 consensus RE containing the C/T SNP in the promoter sequence of the FLT-1 gene. The asterisk indicates the location of SNP c519t, and the underlined region corresponds to the spacer sequence. The second half site contains three mismatches from the consensus (lowercase letters).

Bioinformatic analysis predicted that the SNP falls within a putative RE for the tumor suppressor gene and transcription factor p53 (p53RE; Fig. 1C). The p53 consensus sequence is RRRCWWGYYYnRRRCWWGYYY, where R is purine, Y is pyrimidine, W is C or T, and n is a spacer of 0-13 nt between the two half sites (18). The SNP is located at the second W (C/T) of the first half site in the sequence GGACA(c/T)GCTccctgGGACcTGagC, where the two half sites are separated by 5 bases (ccctg). The infrequent T variant would result in a good match to the first consensus half site and might increase responsiveness to wild-type p53 (wtp53). However, the second half site contains three mismatches from consensus (lowercase letters) and would be expected to interact only weakly with p53.

Transactivation Capacity of wtp53 Toward flt-1 Alleles. To assess the potential of the FLT1-T promoter variant to respond to p53-induced transactivation in human cells, the 1,196-bp promoter region of flt-1 described above containing either the FLT1-T or FLT1-C SNP was cloned into a pGL3 basic luciferase reporter plasmid (Fig. 2A). The constructs were transfected first into two different p53^+/+ cell lines (HMEC and MCF7), both known to express endogenous flt-1. There was an at least 8-fold difference in transcriptional activity of the FLT1-T promoter when compared with FLT1-C (Fig. 2B).

Fig. 2.

Characterization of the promoter activity of FLT1-C and FLT1-T alleles from different cell lines. (A) A description of the T and C alleles (FLT1-T and FLT1-C) of the flt-1 promoter placed into the luciferase reporter vector pGL3-basic. (B) To assess the potential of the T allele within the flt-1 promoter within different human cell lines, the ability of the FLT1-T and FLT1-C promoter alleles to support transcription was determined after transfections into HMEC and MCF-7 cells (homozygous for the C allele). There was an at least 8-fold increase of transcriptional activity of the T allele promoter (FLT1-T) when compared with FLT1-C in these p53^+/+ cell lines. (C) MCF-7 cells were also transiently cotransfected with vectors containing wtp53 or a truncation of the human p53 (Q331stop) under the control of a CMV promoter. Cotransfection with pSV-β-galactosidase control vector was carried out to standardize for transfection efficiency. Luciferase activity of each FLT1-C or FLT1-T reporter construct was compared with the empty pGL3-promoter vector (control). Data represent the averages and standard deviations for three experiments.

To further examine the impact of p53 expression on transactivation from the flt-1 promoter, MCF-7 cells were transfected with the FLT1-C or FLT1-T reporter constructs along with expression plasmids encoding either wtp53 or a truncation mutant (Q331stop, as shown in Fig. 2C). The addition of the wtp53 plasmid increased expression from the FLT1-T plasmid an additional 4-fold, whereas the truncated p53 had little if any effect. There was only minimal luciferase activity associated with the FLT1-C allele even when wtp53 was overexpressed.

We used a SaOS2 p53-null cell line to investigate more directly a role for p53 in transactivation of the FLT1-T promoter variant (Fig. 3). Cells were transfected with the FLT1-C and FLT1-T reporter plasmids along with the p53-expressing plasmid. In the absence of p53 (Fig. 3A) there was no induction of transcription from either the FLT1-C or FLT1-T plasmids (Fig. 3B). In contrast, cotransfection with a wtp53 expression plasmid led to an increase in transcriptional activity from the FLT1-T, but not the FLT1-C promoter (Fig. 3B). The efficiency of p53 transactivation from the FLT1-T promoter was ≈25% of that from a reporter containing the strong p21-5′-p53 RE. (A direct comparison may not be possible because the p21 RE was not inserted in the context of the flt-1 promoter region.)

Fig. 3.

Transactivation of FLT1-C/T alleles by transfected wtp53 in SaOS2 cells. The ability of the FLT1-T and FLT1-C alleles to support transactivation by p53 was examined after transfection into SaOS2 p53^-/- cells. (A) SaOS2 cells were transiently cotransfected with a vector containing wtp53, and proteins were determined 24 h after transfection by Western blot. (B) Luciferase activity from the FLT1-T and FLT1-C luciferase reporter plasmids relative to the empty pGL3-promoter vector in the presence or absence of p53. Cells were also transformed with pSV-RLV40 vector to standardize for transfection efficiency. The P21-5′-p53 RE-luciferase reporter is a positive control for p53 induction. Presented are the averages and standard deviations for three experiments. (C) p53 occupancy on FLT1-C and FLT1-T promoters. p53-deficient SaOs2 cells were transfected with the empty vector (control), wild-type FLT1-C and FLT1-T reporter constructs, and a p53-producing plasmid. ChIP was performed with antibodies (Ab) against p53. p53 binding was determined by using flt-1-specific primers in the PCR analysis. “Input” corresponds to total lysate used as a control for PCR amplification. “Ab” indicates samples immunoprecipitated with p53 antibody. “No Ab” corresponds to the amplification from cell extracts that were immunoprecipitated with only the control anti-mouse secondary antibody. (D) RFLP analysis of chromatin immunoprecipitated DNA from SaOS2 cells. Digestion of DNA samples with NspI restriction enzyme (+) is presented. Restriction-digested FLT1-C and FLT1-T promoters were used as internal controls.

The difference in p53-mediated transactivation from the flt-1 promoter alleles could be due to differences in binding. Therefore, lysates from SaOS2 cells that had been cotransfected with the reporter plasmids carrying the FLT1-C or -T alleles along with the wtp53 expression vector were subjected to chromatin immunoprecipitation (ChIP) with p53 antibody. As shown in Fig. 3C, the DNA from cells transfected with the FLT1-T promoter p53 RE exhibited p53 occupancy, whereas cells containing the FLT1-C promoter had little, if any, p53-specific binding. Moreover, RFLP analysis using the Nsp1 restriction enzyme confirmed that the “chipped” DNA were in fact the transfected FLT1-T constructs and not the endogenous FLT1-C homozygous alleles (Fig. 3D). Overall, these results demonstrate that the FLT1-T SNP can result in FLT-1 being incorporated into the p53 master regulatory system.

p53-Dependent Up-Regulation of the Endogenous FLT1-T Allele by Genotoxic Stress. The results with transfected SaOS2 cells led us to pursue a direct role for p53 in the differential induction of endogenous FLT1-T and FLT1-C alleles. The isogenic pair of cell lines HCT116 p53^+/+ and HCT116 p53^-/- afforded us a unique opportunity because they were heterozygous for the uncommon FLT1-T allele (Fig. 1B and Table 1).

Cells were exposed to the topoisomerase inhibitor doxorubicin (0.3 μg/ml for 24 h), a well established inducer of p53, as exemplified in Fig. 4A, and the consequences on induction of the endogenous flt-1 alleles were assessed. As shown in Fig. 4B, doxorubicin results in a 7-fold increase in flt-1 mRNA in the p53^+/+ cells. This increase contrasts with the lack of flt-1 induction in the p53-null cell line. The specific role of p53 in transactivation of flt-1 was assessed through ChIP analysis, described in Fig. 4C. There was a 4-fold increase in flt-1 DNA precipitated with the p53 antibody from extracts of doxorubicin-treated wtp53 cells, whereas there was essentially no flt-1 DNA brought down from the p53-deficient cells. As expected, there was strong binding to the p21 promoter DNA (used as positive control) after doxorubicin treatment in the wtp53 cells (Fig. 2C). Based on ability to cut the DNA with NspI (>90%), there was clear FLT1-T allele-specific binding by the doxorubicin-induced p53 (Fig. 4D). In a related set of experiments with a p53-null lung adenocarcinoma CaLu-1 cell line (also heterozygous FLT1-C/T), transient transfection with a p53-expressing plasmid resulted in a 3.5-fold increase in flt-1 mRNA (Fig. 5A, which is published as supporting information on the PNAS web site), which was related to p53 recruitment to the promoter (Fig. 5B). In contrast, in p53-positive breast cancer MCF-7 (homozygous FLT1-C/C) cell line even transient p53 overexpression failed to induce flt-1 mRNA expression (Fig. 5A).

Fig. 4.

The endogenous FLT1-T allele in HCT116 (FLT1-C/T) is specifically induced by doxorubicin via p53. (A) Western blots of p53 from HCT116 p53^+/+ and p53^-/- cell lines that are heterozygous for the FLT1-C/T alleles after 24 h of doxorubicin treatment (0.3 μg/ml). Expression is normalized to β-actin levels. (B) flt-1 mRNA is regulated by p53 in HCT116 cells harboring a heterozygous C/T allele. Shown are the relative flt-1 mRNA expression levels as analyzed by real-time PCR analyses after 24 h of doxorubicin treatment (0.3 μg/ml). Expression is normalized to GAPDH levels. p53-dependent flt-1 mRNA expression was compared in HCT116 p53^+/+ and HCT116 p53^-/- cells. The mean of the control gene expression level was set as 1. Each bar represents the mean of six independent experiments ± SD. (C) After exposure to doxorubin, ChIP was performed on extracts with antibodies against p53. Total lysate was used as a control for PCR amplification (input). PCR was performed with gene-specific primers. p53 binding was shown to be much greater to the T allele than to the C allele in the promoters of the endogenous genes. There was no flt-1 or p21 DNA recovered by ChIP from HCT116 p53^-/- cells. Presented are the averages and standard deviations for three independent experiments. (D) RFLP analysis of the PCR products from the ChIP experiment involving p53^+/+ cells in C. Presented as controls (first three columns) are restriction analyses of PCR products of the FLT1-T and FLT1-C reporter constructs with and without treatment by NspI as well as PCR products of the endogenous flt-1 promoter region (HCT116). +, NspI restriction enzyme.

To establish that differences between the HCT116 cell lines in terms of ability to transactivate the FLT1-T vs. FLT1-C alleles were strictly due to differences in p53 status, cells were exposed to doxorubicin after transfection with the FLT1-T and FLT1-C luciferase reporter plasmids. Only the FLT1-T plasmid exhibited expression, and this was restricted to the p53^+/+ cell line (Fig. 6A, which is published as supporting information on the PNAS web site). Furthermore, transfection of p53 plasmid into the p53-null cell line also resulted in strong expression of the FLT1-T, but not the FLT1-C luciferase reporter (Fig. 6B), demonstrating that this cell line has the potential to be induced by p53 from the FLT1-T allele.

We conclude that there is strong endogenous FLT1-C/T allele discrimination by damage-induced as well as overexpressed p53.

Discussion

Variation in targeted p53 RE sequences can alter transcriptional control and may contribute to transactivation specificity in response to stress signals, which in turn could result in biological diversity within a population. Reliable identification of cis-acting promoter genetic variants that can affect gene regulation is still a challenge in genomics because of the plasticity of promoter architecture or the difficulty to predict which regulatory SNP can affect recruitment of transcription factors to REs (17, 19-22). Such regulatory genetic variants can be expected to act as low-penetrance modifiers of disease risk, and their identification is important especially for the molecular characterization of complex traits (19). Ultimately, the transcriptional consequences of the variants need to be assessed.

Here we report a functionally distinct genetic variation in the promoter of the flt-1 gene in humans. The infrequent FLT1-T variant results in the inclusion of this gene into the network coordinated by the master transcription regulator p53, a tumor suppressor gene that is highly responsive to a variety of stress conditions.

Having observed significant variation in flt-1 expression between cell types and individuals, we investigated underlying genetic sources as contributors to expression differences. Published information on genetic variants (dbSNP, Human Gene Mutation Database, Human Genic Bi-Allelic Sequences, Ensembl) revealed only one SNP in the promoter region. However, that SNP (rs17086745: G/C, chr 13: 27967272) was not confirmed by subsequent genotyping (23). Therefore, we searched for allelic variants in the flt-1 promoter and 5′ UTR region by SSCP analysis. Subsequent DNA sequence analysis revealed a C→T SNP 677 bases upstream of the TSS. The frequency of C/T heterozygous individuals was 6%.

Our results establish that only the flt-1 promoter with the T-SNP is responsive to p53 and can recruit p53 proteins based on ChIP assays, which was determined by using both transient transfection assays and analysis of the endogenous flt-1 gene in cell lines that are heterozygous for the newly identified regulatory SNP. The use of various human cell lines with differences in p53 status, together with the ectopic expression of functionally distinct p53 alleles or with the activation of p53 by using different genotoxic stresses, allowed us to confirm the specific responsiveness of the FLT1-T promoter, but not FLT1-C, to p53.

In particular, we took advantage of the HCT116 cell line derived from colon cancer, which is heterozygous for the FLT1-T SNP. The availability of a p53-null variant obtained by gene targeting (24) provided a close-to-isogenic control cell line to assess the role of p53 on flt-1 expression under controlled conditions. In HCT116, we compared responses from the endogenous FLT1-C/T alleles and from transfected FLT1-T or FLT1-C promoter fragments. There was allele discrimination in both cases either by p53 stabilized through doxorubicin treatment or by ectopic p53 overexpression. ChIP analysis followed by flt-1 promoter amplification and RFLP established that p53 bound specifically to the FLT1-T allele.

Our results demonstrate the potential for wtp53 to stimulate transcription at the flt-1 promoter that depends on an infrequent SNP that generates a consensus half-site RE within an overall weak p53 RE. These results are consistent with our recent genome-wide search of functionally distinct p53 REs that was based on an experimentally derived sequence and structure definition of what constitutes a p53-responsive RE and on predictions of functional changes caused by SNPs in REs (17). We had identified nearly 200 polymorphic p53 REs in novel putative p53 targets that were predicted to be functionally distinct and directly tested six of these. Those six could be induced by genotoxic stress or could be activated directly by transfection with p53 cDNA. The present report broadens the general conclusions from that study on interindividual variation in the p53 transcriptional network. These and other cis-acting polymorphisms in REs (25) represent an important class of genetic variation that has implications for disease such as cancer. Interestingly, sequences like the flt-1 RE that contain a long spacer between half sites and multiple mismatches were excluded from our previous genome-scale search because of their predicted overall weak activity. The present results with the FLT1-T SNP clearly indicate that this class of REs with a well matched half site can be responsive to p53.

p53 is a prominent tumor suppressor gene that coordinates cellular responses to stress, primarily acting as a sequence-specific transcription factor. Besides genotoxic stresses, other cellular perturbations are known to activate p53, including hypoxia. The long and growing list of established p53 direct transactivation targets include genes involved in angiogenesis such as matrix metalloproteinase 2, PAI-1, and BAI (26-28).

The inclusion of the flt-1 gene and, therefore, the VEGF pathway into the p53 transcriptional network, but only in a fraction of the human population, adds complexity to the spectrum of possible biological outcomes of p53-mediated responses to genotoxic damage, including chemotherapy. Indeed, flt-1 overexpression has been associated with pathological angiogenesis, tumor progression, cell survival, proliferation, migration, invasion, and metastasis (1, 29-34). Although we have established a direct interaction between p53 and the FLT1-T allele, an assessment of the biological consequences of the newly identified SNP in the flt-1 promoter is beyond the scope of our study. Cell culture and animal model studies will be required to determine the impact of the SNP and p53 activation on flt-1-induced biological responses. At present, genotype/phenotype relationships in human cells are limited by the lack of isogenic cell line models that allow direct interpretation of the biological consequences SNP alleles (17).

There is a general agreement that FLK-1 is the major mediator of the mitogenic and angiogenic effects of VEGF during transformation and tumorigenesis (4). However, recent reports demonstrate that flt-1 is present and functional on different human cancer cells and that activation of flt-1 by VEGF can activate processes involved in tumor progression and metastasis (12, 33, 35). Thus, we suggest that the FLT1-SNP allele may be a significant risk factor in the development of angiogenesis-driven diseases. The present findings on differences in p53-responsiveness to the FLT1-T and FLT1-C alleles should be considered in epidemiological studies that address related health issues such as cancer susceptibility and therapy, especially given the prominent role of p53 in cancer.

Materials and Methods

Use of Human Materials. This investigation conforms with the principles outlined in the Declaration of Helsinki. DNA was obtained from 150 Caucasian individuals and different endothelial (HMEC), bone-related (SaOS2 and U2OS), colon (HCT15, HCT116, HT29, LoVo, RKO, SW1417, and SW480), breast (BT20, BT474, Cal51, Cama1, H184, HCC1397, TD47D, ZR751, and MCF-7), and lung (CaLU1) tumor cell lines.

DNA Extraction and Screening of the FLT-1 Promoter for Genetic Variations. Information about genetic variations was obtained by database search in the National Center for Biotechnology Information's dbSNP (36), Human Gene Mutation Database, or Human Genic Bi-Allelic Sequences or was experimentally determined by SSCP analysis.

SSCP analysis and RFLP analysis were performed by using standard methods (see Supporting Materials and Methods, which is published as supporting information on the PNAS web site).

In Silico Sequence Analysis for Potential cis Elements. Promoter sequences with either C (FLT1-C) or T (FLT1-T) allele at position -912 (i.e., 912 nt upstream of the TSS) in the FLT-1 promoter were analyzed for potential cis elements by using matinspector analysis software and transplorer with the database TRANSFAC (37) (version 3.5, Gesellschaft für Biotechnologische Forschung, Braunschweig, Germany).

Cloning of the Two Polymorphic Variants of the FLT-1 Promoter from Genomic DNA. A 1,196-nt fragment of the he FLT-1 promoter upstream of TSS was PCR-amplified by using the following primers: 5′-GCGCACGCGTGTGGCAACTTTGGGTTACCCAAC-3′ and 5′-GCGCAAGCTTTGTGAGAAGCAGACAGCTGAGCA-3′. The two primers contain, respectively, HindIII and MluI restriction sites at their 5′ ends. PCR products were purified, digested with HindIII and MluI, and cloned into the luciferase reporter vector pGL3basic (Promega). The identity of the inserts was confirmed by DNA sequencing (ABI PRISM 377, Applied Biosystems).

Transcriptional Regulation of FLT-1. To examine the putative functional significance of the SNP within the FLT-1 promoter, we compared the activity of promoter fragments of the two SNP alleles cloned in luciferase reporter vectors in different tumor cell lines. The activity of the predicted p53 RE in the FLT1-T promoter was characterized by cotransfections by using the plasmids pCMV-WTp53 or pCMV-(Q331stop)-p53, which express the human wtp53 cDNA or the truncation mutant Q331Stop, respectively, and the control vector pCMV-Neo (38).

Characterization of FLT1 Promoter Activity by Luciferase Assays in Transient Transfection Experiments. For each cell line 1-2.5 × 10⁵ cells were seeded in 12-well plates 24 h before transfection. Cells were transfected by using FuGENE 6 (Boehringer Mannheim) according to manufacturer's instructions at ≈80% confluence [with 350 ng of FLT-1 promoter-luciferase constructs or the control plasmid pGL3-basic (Promega)]. When appropriate, 0.5 μg of the p53 expression and control plasmids were cotransfected. Cotransfection with 0.5 μg of pSV-β-galactosidase control vector (Promega) was carried out to standardize for transfection efficiency in MCF-7 and HMEC cells, and 150 ng of Renilla reniformis luciferase pSRLV40 plasmid (Promega) was used as normalizing control in SaOS2 and HCT116 cells. Total plasmid DNA per well was adjusted to an equal level by adding the empty vector pCMV-Neo. For doxorubicin treatments, medium was removed 30 h after transfection, and fresh medium was added in the presence of the indicated dose of the drug. In all experiments cells were harvested 48 h after transfection by using the reporter lysis buffer (Promega). We analyzed firefly and Renilla luciferase or β-galactosidase activity at room temperature using a Lumat LB 9501 (Berthold Technologies) or a Victor Wallac (PerkinElmer) multilabel plate reader according to the manufacturers' instructions using the Dual-Luciferase Reporter Assay System (Promega) and/or Galacto-Light (Applied Biosystems). For each construct, relative luciferase activity is defined as the mean value of the firefly luciferase/β-galactosidase or firefly luciferase/Renilla luciferase ratios obtained from three independent experiments. A two-tailed t test was performed to determine statistical significance.

Transcription Analysis of the Endogenous flt-1 Gene. TaqMan real-time PCR analysis was used to study induction of endogenous flt-1 mRNA expression in HCT116 p53^+/+ and the HCT116 p53^-/- cell lines, which are heterozygous for the SNP in the flt-1 promoter. The MCF-7 p53^+/+ cells which are homozygous C/C in flt-1 were used as a negative control for p53-responsiveness. Briefly, the comparative CT method was used to determine the ratio of target to β-actin endogenous control. Total cellular RNA was isolated from transfected cells by using TRIzol reagent (Invitrogen). For RT-PCR analysis, first-strand cDNA was synthesized as described in ref. 39.

TaqMan analysis was carried out according to the manufacturer's instructions, with the use of an Applied Biosystems 7700 system (PerkinElmer). Amplification conditions and primer sequences are described in Supporting Materials and Methods and in Table 2, which is published as supporting information on the PNAS web site. Specificity of the flt-1 RT-PCR products was confirmed by sequencing using an automated sequencing device (ABI PRISM 377, Applied Biosystems). Alternatively, primers on demand (Applied Biosystems) for flt-1 (Hs00176573), P21 (Hs00355782), and GAPDH (Hs99999908) as endogenous housekeeping gene were also used, and PCR was assessed as described in ref. 17.

Western Blot Analysis. Cell extracts were adjusted to equal protein levels (determined in triplicate according to Bradford with BSA as a reference standard), resolved by 4-10% BisTris NuPage, and transferred to polyvinylidene difluoride membranes (Invitrogen) with a semidry electroblotter (Owl Separation Systems). Membranes were probed with monoclonal antibodies specific for p53 (pAb1801 and DO-1, Santa Cruz Biotechnology). The quality as well as the equal loading and transfer of protein blots were determined by Ponceau S staining using a monoclonal antibody against β-actin (Sigma). The M_r values of the immunoreactive bands were determined by using molecular weight markers. After washing, blots were incubated with anti-mouse IgG-conjugated peroxidase antibody (Santa Cruz Biotechnology), and immune complexes were visualized by using ECL reagent (Amersham Pharmacia).

ChIP Assay. ChIP assays were performed with an established protocol and minor modifications (40) by using 1 μg of mouse monoclonal anti-p53 antibody DO7 (Pharmingen). Amplification, primers, and PCR conditions are described in Supporting Materials and Methods.

Cell Culture Conditions. The human MCF-7 cells were cultured in DMEM (Biochrom) supplemented with 100 units/ml penicillin plus 100 μg/ml streptomycin (Biochrom), 10% estrogen-deficient FCS (Biochrom), and 2 mmol/liter l-glutamine (Biochrom). HMEC were cultured according to our recent protocol (8). Human osteosarcoma p53-null cell line SaOS2 was maintained in DMEM with 10% standard FBS and antibiotics. HCT116 p53^+/+ and its p53^-/- derivative (gifts from B. Vogelstein, The Johns Hopkins University, Baltimore), as well as the human osteosarcoma U2OS cell line and lung adenocarcinoma CaLu1 cell lines, were grown in McCoy's A5 medium supplemented with 10% FBS and antibiotics. All cells were incubated at 37°C with 5% CO₂.

Statistical Analysis. Data analysis was performed by using spss 11.0 for Windows (SPSS) and sigmaplot 2002 8.0 for Windows (Systat). Values are given as means ± SD if not otherwise indicated.