PMC

Western blots of (A) ELK1 and ERK2 levels in MCF10A cells starved for EGF and treated with siGAPDH or siELK1 and (B) ELK1 and ERK2 phosphorylation levels in MCF10A cells following EGF treatment for the indicated times. Arrows indicate slower migrating phosphorylated ELK1 species. (C) Summary of microarray analysis of gene expression changes caused by ELK1 depletion in either EGF-starved or EGF-stimulated cells. The numbers of genes which become up- or down-regulated upon ELK1 depletion are indicated. (D) Overlap between genes which show a siELK1-induced change in mRNA levels at 0 min and 30 min of EGF stimulation.

(A) Network formed by proteins encoded by actin cytoskeleton/migration-related genes associated with “unique” ELK1-bound regions (each protein denoted by a circle). Asterisks mark genes tested in (B) (PFN1 is associated with a redundant region). Grey circles indicate that gene expression is changed upon ELK1 depletion. See for details of gene names. (B) mRNA levels of nine actin cytoskeleton- and migration-associated genes were measured by RT-qPCR in serum starved MCF10A cells transfected with siELK1 and normalised to an siGAPDH-transfected control; bars show average values from three biological repeats with standard deviations. Levels of ELK1 mRNA indicate the efficiency of depletion; RBL2, EFR3A are negative controls which do not associate with ELK1. * P<0.05, ** P<0.01 (Student's paired t-test).

(A) Overlap between regions bound by ELK1 (MCF10A cells) and GABPA (Jurkat cells; ). (B) Distribution of distances between peak summits and the nearest TSS for regions bound uniquely by ELK1 (“unique”-top) or redundantly with GABPA (“redundant” -bottom)(limited to within 2 kb up- or downstream). (C) TFBS logos obtained through STAMP-assisted visualisation of Weeder-derived position weight matrices of motifs overrepresented in “unique” (top) and “redundant” (bottom) ELK1-bound regions. Only the logos most closely resembling transcription factor binding sites present in the JASPAR database are shown. The second and third logos shown for the “unique” regions were identified by sequentially masking out the ETS and SRF binding motifs respectively. (D) Distributions of the number of ETS motifs found in the “unique” and “redundant” ELK1-bound regions. Occurrences of non-overlapping unique hexamers, corresponding to exact matches to derivatives of the octameric motifs shown above the graphs were counted. “Background” represents the distribution of an inverted ETS motif. (E) Occurrence of an ETS-SRF module in “unique” and “redundant” ELK1-bound regions. A module is defined as containing an ETS and SRF site within 100 bp of the ELK1 peak summit and a distance between the ETS and SRF motifs of less than 50 bp. (F) Overlap between the indicated categories of regions bound by ELK1 in MCF10A cells and regions bound by FOS in HeLa-S3 cells .

(A) Summary of ChIP-seq data validation (see ). Box plots show the median (horizontal lines) and the distribution of the enrichment levels, in qPCR validation of ELK1 or IgG ChIP samples. Binding regions were randomly selected from the high or low confidence ChIP-seq datasets. As a comparison, the values from regions which do not detectably bind ELK1 are shown (Neg). Data are from three biological repeats. (B) Genomic distribution of ELK1 ChIP-seq regions from the high confidence dataset (left) compared to a random distribution (right). Sectors corresponding to the promoter (up to 10 kb upstream from the TSS), the downstream region (up to 10 kb downstream from the TTS) and the UTRs are indicated. (C) Overlap between genes assigned to ELK1 ChIP-seq regions and genes which show a change in mRNA levels upon ELK1 depletion under the indicated conditions of EGF stimulation. Z-scores were obtained by comparing with an overlap of the ChIP-seq gene list with 10,000 random lists containing the same numbers of genes. (D) Overlap between numbers of genes which are associated with ELK1 binding events and show a siELK1-induced change in mRNA levels at 0 min and 30 min of EGF stimulation.

(A) K-means clustering of expression levels of genes associated with ELK1-bound regions that show a significant response to siELK1 and/or EGF treatment. Individual clusters (1–8) are separated by dotted lines. (B) Summary profiles of the target genes in clusters 2, 4, 7 and 8. The data are presented as changes of individual (grey) and average (black) expression values of genes in each cluster under each of the four experimental conditions. For each gene, the mean of the signals across all four conditions was set as zero, and expression levels (z-transformed) presented are relative to this value. (C) Distribution of “unique” and “redundant” region-associated ELK1 target genes in the clusters identified in (A), with Chi square or Fisher Exact test-derived P-values. (D) Distribution of genes up- and down-regulated by ELK1 depletion in the “unique” (U) and “redundant” (R) ELK1 target gene datasets after 0 or 30 mins EGF stimulation. (E) Heatmap summary of changes in expression of three genes from the “redundant” ELK1 target dataset upon treatment with siELK1 (from microarrays) or siGABPA (by RT-PCR) (relative to a siGAPDH-transfected control), at 0 and 30 minutes of EGF stimulation. (F) Enrichment of the indicated GO categories in expression clusters identified in (A); P-values were obtained in a Chi-square test and refer to the cumulative enrichment of clusters which cross the dashed line vs. those that do not. (G) Heatmap of the distribution of GO terms identified for “unique” and “redundant” ELK1-associated genes. Each GO term is scored by −log₁₀(P-value). Lines on the left mark terms related to gene expression (grey), actin/migration (green) and cell survival (red).

(A) Control and ELK1-depleted MCF10A cells were starved for EGF for 48 h, subsequently stimulated with EGF for 24 hours and stained with phalloidin to visualise the actin cytoskeleton. Nuclei are stained with Hoechst (blue). Red arrows – subcortical actin, white arrowheads – membrane protrusions. (B) Quantification of the percentage of cells exhibiting membrane protrusions from siGAPDH- and siELK1-transfected cells. Bars show average values from three biological repeats with standard deviations; three fields were scored for each repeat. (C) Representative images of wounds created in monolayers of siGAPDH- or siELK1-transfected MCF10A cells, at t = 1 h and t = 15 h post-stimulation with EGF. Lines show borders of areas which were used for quantification. (D) Areas of wounds in monolayers of MCF10A cell treated as in (C) were measured at one-hour intervals of live imaging experiments; data are shown for the siELK1-treated population as a percentage increase of area compared to control siGAPDH treated cells. Graph shows average of three biological repeats with standard deviation. (E) Quantification of MCF10A cells which detached from clusters in the siGAPDH and siELK1-treated populations between t = 60 min and t = 460 min after EGF treatment. The experiment was performed in three biological repeats and in each case the fate of 20 cells was determined. (F) Migratory trajectories of MCF10A cells transfected with siGAPDH or siELK1, manually tracked between t = 1 h and t = 7 h of EGF stimulation. Each coloured line represents the path travelled by an individual cell. (G) Box plots of the distributions of trajectory lengths for MCF10A cells transfected with the indicated siRNA species, manually tracked between t = 1 h and t = 7 h after EGF stimulation. The experiment was performed in 3 biological repeats, in each case 10 cells were tracked. The grey shaded area corresponds to the second and third quartiles of the migratory range covered in control (GAPDH siRNA transfected) cells. Statistical significance was determined in Student's paired t-tests (B and F) or P-values were obtained from Kolomogorov-Smirnov tests (E and G). * P<0.05 ** P<0.01 *** P<0.005.