Transcriptional amplification in tumor cells with elevated c-Myc

Abstract

Elevated expression of the c-Myc transcription factor occurs frequently in human cancers and is associated with tumor aggression and poor clinical outcome. The effect of high levels of c-Myc on global gene regulation is poorly understood but is widely thought to involve newly activated or repressed "Myc target genes." We report here that in tumor cells expressing high levels of c-Myc the transcription factor accumulates in the promoter regions of active genes and causes transcriptional amplification, producing increased levels of transcripts within the cell's gene expression program. Thus, rather than binding and regulating a new set of genes, c-Myc amplifies the output of the existing gene expression program. These results provide an explanation for the diverse effects of oncogenic c-Myc on gene expression in different tumor cells and suggest that transcriptional amplification reduces rate-limiting constraints for tumor cell growth and proliferation.

Figure 1. c-Myc overexpression leads to increased binding at promoters of active genes

A) Schematic of the inducible c-Myc expression system in P493-6 Burkitt's lymphoma cells. B) Western blot of relative levels of c-Myc protein (top) and actin (bottom) at 0hr, 1hr, and 24hr after induction. C) Left: Quantitative Western blot of c-Myc during induction. Increasing amounts of purified, recombinant his₆-c-Myc (left lanes) and lysates prepared from increasing numbers of cells (right lanes) shown at 0hr (top), 1hr (middle) and 24hr (bottom). Right: Coomasie stained gel of purified recombinant his₆-c-Myc. D) Images of three representative cells taken during c-Myc induction. From left to right, cells at 0hr, 1hr, and 24hr are shown. From top to bottom are images of bright field, CD9 stained, DAPI stained, c-Myc stained, and a merge of c-Myc and Dapi stained cells. E) Median c-Myc levels in the +/− 1kb region around transcription start sites (TSSs) of promoters ranked by increasing RNA Pol II occupancy at 0hr. Levels are in units of reads per million mapped reads per base pair (rpm/bp) for 0hr, 1hr, and 24hr (blue, black, red). Promoters were binned (50/bin) and a smoothing function was applied to median levels. F) Gene tracks of c-Myc binding at the NPM1 gene at 0hr (top), 1hr (middle), and 24hr (bottom). The x-axis shows genomic position. The y-axis shows signal of c-Myc binding in units of rpm/bp. G) Heatmap of c-Myc levels at TSS regions at 0hr, 1hr, and 24hr. Each row shows the +/−5kb centered on the TSS. TSSs for all actively transcribed genes in P493-6 cells are shown and ranked by c-Myc occupancy at 0hr. Color scaled intensities are in units of rpm/bp. H) Boxplots of c-Myc levels at 0hr, 1hr, and 24hr in the +/− 1kb centered on the TSS of all genes actively transcribed at the start of c-Myc induction. Units are in rpm/bp. Changes between mean c-Myc levels at promoters are significant (Welch's two-tailed t test) between 0hr and 1hr (p-value < 1e-16) and 1hr and 24hr (p-value < 1e-16). I) Venn diagram showing the overlap between sets of transcribed genes during c-Myc induction at 0hr, 1hr, and 24hr. Transcribed genes were defined as having an enriched region for either H3K4Me3 or RNA Pol II within 5kb of the annotated transcription start site (TSS). Each circle represents the set of transcribed genes at 0hr (blue), 1hr (black) or 24hr (red). See also Figure S1.

Figure 2. c-Myc overexpression leads to binding of active enhancers

A) Heatmap showing c-Myc occupancy levels at the top 5,000 promoters ranked by RNA Pol II levels at 0hr, the top 5,000 active enhancers ranked by H3K27Ac levels at 0hr, rRNA genes, and tRNA genes (columns) at 0hr, 1hr, and 24hr (rows). Intensities are in units of c-Myc rpm/bp. B) Heatmap of H3K27Ac and c-Myc levels at the top 5,000 active enhancers ranked by H3K27Ac levels at 0hr in P493-6 cells at 0hr and 24hr. Each row shows +/−5kb centered on the H3K27Ac peak. Rows are ordered by the net increase in c-Myc occupancy after 24hrs of c-Myc induction. Color scaled intensities are in units of rpm/bp. C) Boxplots of c-Myc levels at 0hr, 1hr, and 24hr in the +/− 1kb around the center of the top 5,000 active enhancers ranked by H3K27Ac levels at the start of c-Myc induction. Units are in rpm/bp. Changes between mean c-Myc levels at enhancers are significant (Welch's two-tailed t test) between 0hr and 1hr (p-value < 1e-16) and 1hr and 24hr (p-value < 1e-16). D) c-Myc binding motif found enriched at core promoters (left, p-value < 1e-232) or at active enhancers (right, p-value < 4e-775) 24hr after c-Myc induction. E) c-Myc occupancy at E-box sequences. Left: E-box variants (CANNTG) were ranked by c-Myc ChIP-Seq signal strength at 24hr. The signal strength relative to the canonical CACGTG E-box is indicated by the shaded color intensity adjacent to E-box sequences. Right: The percentage of all E-box motifs found in c-Myc bound regions at promoters (red) or enhancers (blue) for all E-box variants. F) Gel shift assays of purified c-Myc/Max binding to canonical CACGTG E-box sequences following competition with decreasing amounts (5, 2.5 and 1 fold excess) of unlabeled competitor sequences. The higher bands reflect the amounts of labeled canonical E-box sequences bound by c-Myc/Max and the lower band reflects the amount of unbound canonical E-box DNA. Lane 1: No competitor DNA. Lanes 2–10: The effect of adding various amounts of competitor DNA fragments containing the canonical E-box or variant E-box sequence motifs (E-box 1 and E-box 3). Bottom: Competitor DNA sequences are shown centered around the E-box (bold). Differences from the canonical sequence are highlighted in red. G) Gene tracks showing c-Myc occupancy (rpm/bp) at the NNT promoter (left) or downstream enhancer (right) after 0hr (top), 1hr (middle), and 24hr (bottom) after c-Myc induction. The x-axis shows genomic position. The y-axis shows signal strength of c-Myc binding. E-boxes proximal to the c-Myc peak at either the promoter or enhancer are depicted as boxes shaded by E-box signal strength (from Figure 2E). The NNT gene is indicated at the bottom (introns as lines, exons as boxes). See also Figure S2.

Figure 3. c-Myc overexpression leads to transcriptional amplification

A) Boxplots of the fold changes in levels of H3K4Me3 (green) and Cdk9 (pink) between 24hr and 0hr at the top 2,000 core promoters with the highest increased c-My occupancy during c-Myc induction (left), the top 2,000 enhancers with the highest increased c-Myc occupancy during c-Myc induction (center), and 2,000 enhancers without increased c- Myc occupancy during c-Myc induction (right). The differences in the levels of changes between H3K4Me3 and Cdk9 were significant (Welch's two-tailed t test) at core promoters (p-value < 2.2e-16) and at enhancers with increased c-Myc occupancy (p-value = 6.48e-07) and not significant at enhancers without increased c-Myc occupancy (p-value = 0.06). B) Western blots of RNA Pol II at 0hr, 1hr, and 24hr using antibodies specific to various forms of the enzyme. From top to bottom: RNA Pol II Ser2P specific (H5, Covance), RNA Pol II Ser2P specific (A300-654A, Bethyl), RNA Pol II Ser5P specific (H14, Covance), hypophosphorylated RNA Pol II specific (8WG16, Bethyl), total RNA Pol II (N-20, Santa Cruz). For RNA Pol II Ser2P and Ser5P antibodies, the ratio of signals vs. 0hr is displayed for each timepoint below the blot. C) Bar graph of mean +/− SEM RNA Pol II enrichment ratio between elongating and initiating regions during c-Myc induction at 0hr, 1hr, and 24hr (left, center, right) for the top 5,000 genes ranked by RNA Pol II occupancy at 0hr. The y-axis shows the ratio between the fold enrichment overbackground of RNA Pol II in the elongating region versus the fold enrichment over background of RNA Pol II in the promoter region. Changes between 0hr, 1hr, and 24hr are significant (Welch's two-tailed t test, p-value < 1e-16). D) Empirical cumulative distribution plots of RNA Pol II traveling ratios (TR) for 1,000 transcribed genes (Rahl et al., 2010). Genes were randomly selected from the pool of genes containing higher than background levels of RNA Pol II at the promoter and gene body at 0hr, 1hr, and 24hrs. Differences in the TR distribution at 0hr and 24hr are significant (Welch's two-tailed t test, p-value = 4.5e-5). E) Left: Bar graph showing quantification of total RNA levels for cells at 0hr, 1hr, and 24hr. Units are in ng of total RNA per 1,000 cells and represented as mean +/− SEM. Right: 5% TBE urea gel of ethidium bromide stained total RNA extracted from equivalent numbers of cells at 0hr, 1hr, and 24hr. Bands corresponding to the 5.8S rRNA subunit, 5S rRNA subunit, and tRNA are labeled. F) Boxplot of transcripts/cell estimations from NanoString nCounter gene expression assays for active (right) or silent (left) genes at 0hr, 1hr, and 24hr. 755 active genes (expressed > 1 transcripts/cell) at 0hr are shown (left, red). 514 silent genes (expressed < 0.5 transcripts/cell) at 0hr are shown (right, black). The number of genes with increased expression between 0hr and 1hr are significant (Wilcoxon rank sum test) for active genes (p-value < 2.2e-16) and non significant for silent genes (p-value = 0.997). See also Figure S3

Figure 4. c-Myc binds to core promoters and active enhancers when overexpressed in different cancers

A) Western blot of c-Myc protein levels in tumor cell lines compared with purified c-Myc. Increasing amounts of purified c-Myc (left lanes) and lysates prepared from increasing numbers of cells (right lanes) are shown for SCLC (left), MM (middle), and GBM (right). B) For each tumor cell line, median c-Myc levels in units of rpm/bp are plotted at promoters ranked by increasing RNA Pol II occupancy. Promoters were binned (50/bin) and a smoothing function was applied to median levels. C) For each tumor cell line, a heatmap of c-Myc levels at TSSs of actively transcribed genes is displayed. Each row represents the +/−5kb centered on the TSS. TSSs for actively transcribed genes in each tumor cell line are shown. Color scaled intensities are in units of rpm/bp. D) For each tumor cell line, a heatmap displays H3K27Ac and c-Myc levels at active enhancers. Each row represents the +/−5kb centered on the H3K27Ac peak. Rows are ordered by H3K27Ac binding in each tumor cell line. Color scaled intensities are in units of rpm/bp. E) c-Myc binding at the NOTCH1 promoter in GBM (top), SCLC (middle), and MM (bottom) tumor cell lines. The x-axis shows genomic position. The y-axis shows signal of c-Myc occupancy (rpm/bp). The NOTCH1 gene is indicated at the bottom (introns as lines, exons as boxes). Enhancer regions are shaded in light blue. The core promoter of NOTCH1 is boxed in black. F) For each tumor cell line, the c-Myc binding motif found enriched at core promoters of actively transcribed genes and active enhancers is displayed (SCLC core promoter, p < 4e-215; SCLC active enhancer, p < 9e-1504; MM core promoter, p < 6e-117; MM active enhancer, p < 5e-678; GBM core promoter, p < 1e-160; GBM active enhancer p < 4e-635).

Figure 5. c-Myc enhancer invasion and transcriptional amplification in patient-derived small cell lung carcinoma

A) c-Myc protein levels in small cell lung carcinoma for low c-Myc expressing (H128) and c-Myc overexpressing (H2171) tumor cells determined by quantitative Western Blot analysis with purified his₆-c-Myc. B) For low and high c-Myc SCLC, the median c-Myc levels in units of rpm/bp were plotted at promoters ranked by increasing RNA Pol II occupancy in H128. c-Myc levels are shown for H128 and H2171 SCLC (black, red). Promoters were binned (50/bin) and a smoothing function was applied to median levels. C) Left: Boxplots of c-Myc levels in the +/− 1kb centered on TSSs at promoters of 15,000 actively transcribed genes in H128 and H2171. Right: Boxplot representation of c-Myc levels at 15,000 active enhancers in H128 and H2171. Units are in rpm/bp. Changes between mean c-Myc levels are significant (Welch's two-tailed t test) at promoters (p-value < 2.2e-16) and enhancers (p-value < 2.2e-16). D) Western blots of RNA Pol II in H128 and H2171 cells using antibodies specific to various forms of the enyzme. From top to bottom: RNA Pol II Ser2P specific (H5, Covance), RNA Pol II Ser2P specific (A300-654A, Bethyl), RNA Pol II Ser5P specific (H14, Covance), hypophosphorylated RNA Pol II specific (8WG16, Bethyl), total RNA Pol II (N-20, Santa Cruz). For RNA Pol II Ser2P and Ser5P antibodies, the ratio of signal vs. H128 signal are displayed for each cell line below the blot. E) Bar graph of mean +/− SEM RNA Pol II enrichment ratio between elongating and initiating in H128 and H2171 cells for the top 5,000 genes ranked by RNA Pol II occupancy in H128 cells. The y-axis shows the ratio between the fold enrichment over background of RNA Pol II in the elongating region versus the fold enrichment over background of RNA Pol II in the promoter region. Changes between H128 and H2171 cells are significant (Welch's two-tailed t test, p-value < 1e-16). F) Empirical cumulative distribution plots of Pol II traveling ratios (TR) for 1,000 transcribed genes. Genes were randomly selected from the pool of genes containing higher than background levels of Pol II at the promoter and gene body in H128 and H2171 cells. Differences in the TR distribution between H128 and H2171 are significant (Welch's two-tailed t test, p-value = 5e-3). G) Left: Bar graph showing quantification of total RNA levels for H128 and H2171 cells. Units are in ng of total RNA per 1,000 cells and represented as mean +/− SEM. Right: 5% TBE urea gel of ethidium bromide stained total RNA extracted from equivalent numbers of cells from H128 and H2171. Bands corresponding to the 5.8S rRNA subunit, 5S rRNA subunit, and tRNA are labeled. H) Boxplots of transcripts/cell estimations from NanoString nCounter gene expression assays for active (right) or silent (left) genes in H128 and H2171 cells. 706 active genes (expressed > 1 transcripts/cell) in H128 cells are shown (left, red). 568 silent genes (exrpressed < 0.5 transcripts/cell) in H128 cells are shown (right, black). The number of genes with increased expression between H128 and H2171 cells are significant (Wilcoxon rank sum test) for active genes (p-value < 2.2e-16) and non significant for silent genes (p-value = 0.03). See also Figure S4.