Global microRNA level regulation of EGFR-driven cell-cycle protein network in breast cancer

Abstract

The EGFR-driven cell-cycle pathway has been extensively studied due to its pivotal role in breast cancer proliferation and pathogenesis. Although several studies reported regulation of individual pathway components by microRNAs (miRNAs), little is known about how miRNAs coordinate the EGFR protein network on a global miRNA (miRNome) level. Here, we combined a large-scale miRNA screening approach with a high-throughput proteomic readout and network-based data analysis to identify which miRNAs are involved, and to uncover potential regulatory patterns. Our results indicated that the regulation of proteins by miRNAs is dominated by the nucleotide matching mechanism between seed sequences of the miRNAs and 3'-UTR of target genes. Furthermore, the novel network-analysis methodology we developed implied the existence of consistent intrinsic regulatory patterns where miRNAs simultaneously co-regulate several proteins acting in the same functional module. Finally, our approach led us to identify and validate three miRNAs (miR-124, miR-147 and miR-193a-3p) as novel tumor suppressors that co-target EGFR-driven cell-cycle network proteins and inhibit cell-cycle progression and proliferation in breast cancer.

Figure 1

Whole-genome miRNA (miRNome) screen with a quantitative proteomics readout. (A) Flow chart of the study. MDA-MD-231 cells were transfected in 6-well format in an automated system with the miRNA library containing 810 mimics. Cells were lysed after 48 h of transfection and incubated with siRNA-validated antibodies for RPPAs. Subsequent to primary data analysis, screen results were compared with target prediction algorithms for each downregulated network protein, and network analysis of the EGFR/cell-cycle proteins at the miRNome level was done using the more-than-random co-regulation approach. (B) RNAi-based antibody validation. For each network proteins, siRNAs were used to validate the specificity/sensitivity of the antibodies before they were incubated with the lysates of the miRNA screen. The x axis of the heatmap shows siRNAs, and the y axis represents the antibodies used to quantify the abundance of proteins. (C) Pearson's correlation coefficients between the two biological replicates in the screen for each protein analyzed. (D) Heatmap showing the effect of whole-genome miRNAs on the EGFR/cell-cycle protein network. MicroRNAs are given in rows and proteins in columns. While red rectangles show upregulation, blue ones show downregulation of proteins for given miRNAs.

Figure 2

Whole-genome miRNA regulation of the EGFR/cell-cycle protein network. (A) Histogram of the effects of the whole-genome set of miRNAs on the given protein PLCG1 (for the histogram of the other proteins measured in this study, see Supplementary Figure S4). While the x axis demonstrates normalized z-scores of the expression change of PLCG1 upon miRNA transfections, y axis shows the frequency (count) of miRNAs. (B) Edge numbers for different z-score thresholds. With increasing stringencies of z-score, number of edges decreases rapidly. Two commonly used significance thresholds, P<0.05 and P<0.001 (equivalent to ∣z∣>1.96 and ∣z∣>3.29, respectively), were shown with red dots on the curve. (C) Dense miRNA–protein network at the absolute z-score threshold of 1.96. Blue circles represent the proteins and black circles indicate the miRNAs. While green edges between an miRNA and a protein show downregulation of that protein by miRNA, red edges show the upregulation of the protein by the given miRNA. miRNAs that regulate more than one protein are located on the circle and those miRNAs that regulate only one network protein are located outside of the circle. (D) miRNA regulation of the EGFR network at a more stringent threshold (∣z∣>3.29). The resulting network is much less dense compared with the graph shown in (C).

Figure 3

Regulatory network showing only the miRNAs that are both predicted to target the given gene and downregulating the protein in our screen (with absolute z-score >1.96). This interaction is shown with a black edge between the miRNA and protein. The validated miRNA/protein interactions in other independent studies are shown with a green edge between the miRNA and the protein. The yellow circles indicate the miRNAs, which are identified to regulate EGFR-driven cell-cycle network in this study. The full list of miRNAs and proteins is given in Supplementary Table S7.

Figure 4

The match between miRNA seed sequences and 3′-UTR of the genes explains the reduction in the expression of most of the proteins in the EGFR/cell-cycle network. The Kolmogorov–Smirnov (K–S) test was used to assess whether distributions of z-scores of two groups (predicted versus non-predicted miRNAs) differ significantly. For most of the proteins with predicted targeting miRNAs, a significant negative aberration of expression distributions was observed (18 proteins out of 25, with two-sided P<0.05). Black lines demonstrate the z-score distribution of the non-predicted miRNAs against that of predicted ones. Two target prediction algorithms were used shown in red (TargetScan, conserved), yellow (TargetScan, non-conserved) and blue (MicroCosm) lines.

Figure 5

Co-regulation of EGFR/cell-cycle network proteins by whole-genome miRNAs. (A) Principles of the ‘more-than-random co-regulation’ approach. (1) Bipartite network consisting of proteins A, B, C, D and miRNAs 1, 2, 3, 4. Green edge between an miRNA and a protein indicates that expression of the protein is reduced by the miRNA whereas a red edge shows that expression of the protein is increased by the miRNA. (2) Protein co-regulation network where the numbers on the edges indicate the number of miRNAs which co-regulate the given protein pairs. Protein pairs A–B was co-downregulated two times, while protein pairs B–C were co-upregulated once. C–D, B–C, and A–C proteins were inversely regulated once by miRNAs, which is shown with a blue directed edge. (3) The significant co-regulations can be identified by the ‘more-than-random’ model. The model compares for each protein pair the number of miRNAs that co-regulate them in the experimentally observed data with the number of miRNAs co-regulating them in the random bipartite network graphs. (4) The statistical-significance test was performed by generating 10 000 randomized networks and computing for each protein pair the fraction of randomized networks in which their co-regulation appeared at least as often as in the observed bipartite network. The labels on the protein co-regulation network indicate the obtained P-values. (5) In a thresholding step, a co-regulation with P<0.05 t₂ was considered statistically significant. (B) Consensus co-regulation graph of EGFR/cell-cycle proteins by miRNAs. Edges shown in the graph are obtained from the most stringent consensus graph. The green edges between the proteins show the co-downregulation of two proteins while the red edges show the co-upregulation. The directed blue edges indicate that the node where the edge originates from is downregulated whereas the node where the edge is directed to is upregulated. The numbers on the edges indicate the number of miRNAs having such regulation patterns in the most stringent consensus graph. Cell-cycle proteins that are regulated by miRNAs in a coordinated manner are marked by the gray box.

Figure 6

Regulatory graphs for miR-124, miR-147 and miR-193a-3p and Western blot validation of RPPA results. (A) Rank of miRNAs based on their frequency in the consensus graphs. Only the first top 50 miRNAs are drawn. The miRNAs (miR-124, miR-147 and miR-193a-3p) that were chosen for experimental validations are shown with red dots. (B) Co-regulation of EGFR/cell-cycle proteins is shown for three miRNAs, which were among the miRNAs having the highest number of more-than-random regulations. Cell-cycle proteins are drawn as dark gray nodes and shown in a light gray background. Color codes of the edges are as in Figure 5B. (C) Western blot validation of the effects of these three miRNAs on the protein expression the EGFR/cell-cycle components. β-Actin was used as loading control.

Figure 7

miR-124, miR-147 and miR-193a-3p are identified as novel tumor suppressors in breast cancer and glioblastoma cell lines by targeting EGFR/cell-cycle proteins. (A) The effects of miRNAs on the cell-cycle progression. After overexpressing the miRNAs in MDA-MB-231 cells, total DNA content (7-AAD) and S-phase population (BrdU) were determined by FACS after 48 h. The red rectangles show the percentage of proliferating cells in S-phase. The results are shown as an average of three biological and three technical replicates. (B) The effects of miRNAs on the viability of the cells. After overexpressing the miRNAs in MDA-MB-231 cells, viable cells were counted by a luciferase-based viability assay after 72 and 96 h. The results are shown as average of two biological and five technical replicates. (C) Validation of direct targeting of genes by miRNAs. A luciferase reporter assay was performed after co-transfecting MDA-MB-231 cells with the luciferase vector harboring the 3′-UTR of the gene of interest and miRNA mimics. Luciferase activities were measured after 48 h of transfection, and results are presented as average of four biological and three technical replicates. P1, P2 and P3 denote the parts of 3′-UTRs analyzed for the genes with very long or difficult to clone 3′-UTRs. (D) Quantification of effects of miR-124, miR-147 and miR-193a-3p on the mRNA level by qRT–PCR. The results are presented as average of two biological and three technical replicates. (E) The effects of miRNAs on the cell-cycle progression of MCF-7 and U87 cell lines as described in (A). (F) The effects of miRNAs on the viability of MCF-7 and U87 cell lines as described in (B). Two asterisks (^**) denote the P-value <0.01, three asterisks (^***) denote the P-value <0.001.