Mutant p53 cooperates with the SWI/SNF chromatin remodeling complex to regulate VEGFR2 in breast cancer cells

Abstract

Mutant p53 impacts the expression of numerous genes at the level of transcription to mediate oncogenesis. We identified vascular endothelial growth factor receptor 2 (VEGFR2), the primary functional VEGF receptor that mediates endothelial cell vascularization, as a mutant p53 transcriptional target in multiple breast cancer cell lines. Up-regulation of VEGFR2 mediates the role of mutant p53 in increasing cellular growth in two-dimensional (2D) and three-dimensional (3D) culture conditions. Mutant p53 binds near the VEGFR2 promoter transcriptional start site and plays a role in maintaining an open conformation at that location. Relatedly, mutant p53 interacts with the SWI/SNF complex, which is required for remodeling the VEGFR2 promoter. By both querying individual genes regulated by mutant p53 and performing RNA sequencing, the results indicate that >40% of all mutant p53-regulated gene expression is mediated by SWI/SNF. We surmise that mutant p53 impacts transcription of VEGFR2 as well as myriad other genes by promoter remodeling through interaction with and likely regulation of the SWI/SNF chromatin remodeling complex. Therefore, not only might mutant p53-expressing tumors be susceptible to anti VEGF therapies, impacting SWI/SNF tumor suppressor function in mutant p53 tumors may also have therapeutic potential.

Keywords: 3D culture; SWI/SNF; VEGFR2; breast cancer; chromatin; mutant p53; transcription.

Figure 1.

Mutant p53 promotes VEGFR2 expression in breast cancer cells. (A) MDA-468.shp53 cells were grown in 3D culture conditions for 8 d with (+DOX) and without (−DOX) doxycycline to induce an shRNA targeting mutant p53. Total VEGFR2 transcript was assayed by qRT–PCR and normalized to the −DOX condition. (**) P < 0.001 by one-tailed t-test. Below is the related immunoblot showing levels of the indicated proteins. (B) MDA-231 cells were grown in 3D culture conditions and assayed for VEGFR2 expression following depletion of mutant p53 with two different siRNAs as described in the Materials and Methods. Expression was normalized to control siRNA. (*) P < 0.05; (**) P < 0.01 by one-tailed t-test. The immunoblot at right shows the indicated protein levels with control or p53 siRNAs. (C) MDA-468.shp53 cells were selected to stably express mutant p53 hot spot mutant R175H, R245S, or R248W that lacks the short hairpin sequence used target endogenous mutant p53 R273H. A control cell line containing empty vector or the cells expressing the indicated p53 hot spot mutants were grown in 3D culture in the presence of doxycycline to deplete the endogenous mutant p53 R273H. Total VEGFR2 messenger RNA was analyzed by qRT-PCR and normalized to the −DOX control condition. Corresponding immunoblot of p53 proteins with actin loading control is shown below. In A–C, error bars represent the standard error. In each experiment, at least three biological replicates were performed. Endogenous VEGFR2 was detected with anti-VEGFR2 antibody, and mutant p53 was identified with a mixture of mABs 1801 and DO-1. (D–F) The Cancer Genome Atlas (TCGA) breast cancer RNA sequencing (RNA-seq) version 2 data set analysis stratified by p53 mutational status (wild type, truncation mutation, hot spot missense mutation, or non-hot spot missense mutation as indicated). RNA expression of VEGFR2 (D), VEGFA (E), and HIF1A (F) is presented as a box plot, where the box contains the interquartile range. The central line represents the median gene expression. Median expression values are delineated for the truncation mutant category in D and the wild-type p53 category in E and F. RNA expression values were analyzed as upper quartile-normalized RNA-seq by expectation maximization (RSEM) of reads. (*) P < 0.05 by Welch's one-tailed t-test in D.

Figure 2.

VEGFR2 inhibition phenocopies loss of mutant p53. MDA-231 cells (A) and MDA-468 cells (B) were transfected with two independent siRNAs to deplete mutant p53 or VEGFR2 and then grown in 3D culture conditions for 8 d. Representative differential interference contrast images were acquired at 10× magnification on live imaging. Relative cell areas of an average of at least 95 colonies per condition for three independent replicates were calculated and are shown in the corresponding bar graphs. Error bars represent the standard deviation. Bar, 100 μm. (*) P < 0.01 by one-tailed t-test.

Figure 3.

Mutant p53 gain of function is mediated by VEGFR2 and may predict response to bevacizumab. (A) MDA-231.shp53 cells were engineered to express control vector, VEGFR2, or VEGFR2 tyrosine phosphorylation mutant Y1059F as described in the Materials and Methods and then grown in 3D culture conditions for up to 8 d. Where indicated, cells were grown in the presence of doxycycline (DOX; low Mut p53) to deplete endogenous mutant p53. Differential interference contrast images were acquired at 10× magnification on live imaging. Bar, 100 μm. (B) Immunoblot of the indicated proteins from A. (C) Relative cell areas of an average of at least 85 colonies per condition among four independent replicates were analyzed. Error bars represent the standard deviation. (*) P < 0.001 calculated by one-tailed t-test. (D) For wound migration analysis, MDA-231 cells were transfected with control siRNA and two independent siRNAs each to deplete mutant p53 or VEGFR2 and then seeded to confluency in tissue culture plates containing inserts. Representative differential interference contrast images (Supplemental Fig. 3A) were acquired immediately upon removal of the insert (0 h) and 48 h later. Relative migration was calculated by dividing the total distance migrated of each sample by the total migration in the siControl sample. At least three images were quantitated per sample. The data are an average of four biological replicates. Error bars represent the standard deviation. (*) P < 0.01; (**) P < 0.001 by two-sided t-test. (E,F) Response ratio showing reduction in tumor volume in TP53 wild-type tumors (E) and TP53 mutated tumors (F) treated with chemotherapy alone (Chemo) or chemotherapy plus bevacizumab (Chemo + Bev). Each data point represents one patient's response to the indicated treatment, which was calculated as the tumor volume of the residual tumor divided by the initial tumor volume. Data are plotted as a box plot, and the sample size is indicated by “n”. P-value was derived from Kruskal-Wallis test. Median values of the chemotherapy-only cohorts are delineated.

Figure 4.

Mutant p53 associates with the VEGFR2 promoter and leads to promoter remodeling. MDA-468.shp53 cells were cultured for 8 d in 3D culture in the presence (−Mut p53; black) and absence (+Mut p53; red) of doxycycline (DOX). Cells were treated with formaldehyde to cross-link chromatin and subjected to the indicated procedures. (A) Scanning ChIP for mutant p53 was performed along 4 kb surrounding the VEGFR2 TSS. ChIP was performed in the presence and absence of doxycycline for mutant p53 and also in the absence of antibodies to p53 using primers corresponding to the indicated data points. Immunoprecipitated chromatin was subjected to qPCR, and the percent input-normalized signal between −DOX and +DOX samples was plotted relative to the peak binding signal at the −150-bp VEGFR2 site. Error bars represent the standard error of the three independent experiments shown in Supplemental Figure S4A–C. (B) For micrococcal nuclease (MNase), PCR chromatin was digested with MNase, and mononucleosome-sized DNA fragments were isolated. qPCR was performed for six amplicons averaging 66 bp along 446 bp of the VEGFR2 promoter from −390 bp to +56 bp relative to the TSS, with the signal normalized to amplicon 1. Error bars represent the standard error of three independent experiments. (*) P < 0.05 by one-tailed t-test. (C) In vivo DNase I footprinting by ligation-mediated PCR (LM-PCR) was performed at the VEGFR2 promoter between approximately −160 bp and +5 bp of the TSS. Densitometry analysis of the relative DNase I hypersensitivity signal is represented by a histogram ([red] +Mut p53; [black] −Mut p53).

Figure 5.

Mutant p53 is found in protein complexes with members of the SWI/SNF CRC at the VEGFR2 promoter. Extracts of MDA-468 (A,C), SK-BR-3 (B), or MDA-231 (D) cells were subjected to immunoprecipitation (IP) with anti-p53 antibodies (mAb DO-1; A,B) or anti-BAF155 antibody (C,D) followed by immunoblotting (IB) with anti-p53 (mAbs DO-1 [A,B] or DO-1 and 1801 [C,D]), anti-BAF53A (A,B), or anti-BAF155 (C,D) antibodies. Inputs represent 5% (A,B,D) or 3.3% (C) of total extract. (E) ChIP–re-ChIP (sequential ChIP) was performed in MDA-468.shp53 cells by performing initial ChIP for IgG or mutant p53 followed by re-ChIP with BAF170 or IgG antibodies. qPCR was performed at the VEGFR2 promoter at site −150 bp from the TSS. Signal is shown as percent of input material. Error bars represent the standard error of two independent experiments. (F–H) Immunodepletion ChIP was performed in MDA-468.shp53 by immunoprecipitating cross-linked cell extract with IgG or anti-p53 mAbs (DO-1/1801/PAb421). ChIP was then performed on the immunodepleted extracts with antibodies to BAF170 (F) or BAF155 (G). qRT–PCR was performed at the VEGFR2 promoter as in E. Signal is shown as fold signal over ChIP for IgG. Error bars represent the standard error of two independent experiments. (H) The immunoblot for mutant p53 with histone 2A as a loading control corresponds to F and G.

Figure 6.

SWI/SNF is required for maximal VEGFR2 expression, nucleosomal remodeling, and expression of other mutant p53-dependent genes. (A,B) MDA-468.shp53 cells were grown for 5 d in 2D cell culture in the presence (−Mut p53; black) and absence (+Mut p53; red) of doxycycline (DOX) and then treated with formaldehyde and prepared for scanning ChIP to detect occupancy of BAF155 (A) or BAF170 (B). IgG was used as a control in either case. Immunoprecipitated chromatin was subjected to qRT–PCR using primers that spanned the length of the VEGFR2 gene (from −2.35 kb to +30 kb downstream from the coding and the 3′ UTR). The percent input-normalized signal between −DOX and +DOX samples was plotted relative to the peak binding signal at the −150-bp VEGFR2 site. Error bars represent the standard error of three independent experiments. (C) Immunoblot of the indicated proteins in A and B. (D) ChIP for mutant p53 in MDA-468.shp53 cells grown in 2D culture was performed in the presence and absence of BRG1 and BRM. Negative site corresponds to +30 kb downstream from the VEGFR2 3′ UTR. (**) P < 0.01 by one-tailed t-test. (E) MNase-assisted ChIP was performed on MDA-468.shp53 cells grown in 2D culture with control siRNA or siRNA to BAF170. Cells were fixed with formaldehyde, nuclei were extracted and incubated with micrococcal nuclease, ChIP was performed using antibodies to histone H3 and IgG, and immunoprecipitated mononucleosomal-sized DNA was purified by agarose gel electrophoresis. qRT–PCR was performed using MNase-PCR primers at the proximal promoter (−78 to −10 bp from the TSS; amplicon 6 [red]) and normalized to the distal promoter (−390 bp to −330 bp from the TSS; amplicon 1 [blue]). Error bars represent the standard error of three independent experiments. (F–I) MDA-468.shp53 cells grown in 2D culture were transfected with two independent siRNAs (20 nM) to deplete BRM (F), BRG1 (G), BAF155 (H) , or BAF170 (I). Total VEGFR2 transcript was assayed by qRT–PCR and normalized to control siRNA (Ctrl). Error bars represent the standard error of three independent experiments. Corresponding immunoblots are shown in Supplemental Figure S6D. (J–N) MDA-468.shp53 cells grown in 2D culture were transfected with a mixture of 50 nM siRNA to codeplete BRM and BRG1 or with control siRNA (Ctrl). VEGFR2 protein (J) and RNA (K) are shown. Three other mutant p53 transcriptional targets—IGFBP5 (L), ceruloplasmin (M), and mammaglobin-A (N)—were also assayed by qRT–PCR and normalized to RPL32 internal control. Error bars represent the standard error of three independent experiments. (*) P < 0.05; (**) P < 0.01 by two-tailed t-test.

Figure 7.

The SWI/SNF complex mediates mutant p53-dependent transcription at many mutant p53-responsive genes. (A) RNA-seq was performed on two independent replicates of MDA-468.shp53 cells grown for 4 d with either control siRNA, siRNA to deplete mutant p53 (Mut p53 knockdown), or siRNAs to codeplete BRG1 and BRM (SWI/SNF knockdown). The top affected genes compared with siControl, using a false discovery rate value <0.01, were analyzed. The total number of up-regulated and down-regulated genes for each knockdown condition are depicted. The numbers of co-up-regulated or co-down-regulated genes in both Mut p53 knockdown and SWI/SNF knockdown conditions are demonstrated by a Venn diagram. Immunoblots for both replicates are shown in Supplemental Figure S7A. (B) The table lists the number of coregulated genes (common genes) from the RNA-seq data. Percent of coregulated genes was calculated by dividing the number of coregulated genes by the number of genes affected in the knockdown condition. (C) Proposed model depicting how mutant p53 interacts with SWI/SNF at mutant p53-responsive genes to promote transcription. Mutant p53 is recruited by SWI/SNF to SWI/SNF-regulated genes. Mutant p53 recruits other transcription factors (TFs), histone acetyltransferases (HATs), or other chromatin modifiers that promote SWI/SNF-dependent promoter remodeling.