A Network of 17 Microtubule-Related Genes Highlights Functional Deregulations in Breast Cancer

Abstract

A wide panel of microtubule-associated proteins and kinases is involved in coordinated regulation of the microtubule cytoskeleton and may thus represent valuable molecular markers contributing to major cellular pathways deregulated in cancer. We previously identified a panel of 17 microtubule-related (MT-Rel) genes that are differentially expressed in breast tumors showing resistance to taxane-based chemotherapy. In the present study, we evaluated the expression, prognostic value and functional impact of these genes in breast cancer. We show that 14 MT-Rel genes (KIF4A, ASPM, KIF20A, KIF14, TPX2, KIF18B, KIFC1, AURKB, KIF2C, GTSE1, KIF15, KIF11, RACGAP1, STMN1) are up-regulated in breast tumors compared with adjacent normal tissue. Six of them (KIF4A, ASPM, KIF20A, KIF14, TPX2, KIF18B) are overexpressed by more than 10-fold in tumor samples and four of them (KIF11, AURKB, TPX2 and KIFC1) are essential for cell survival. Overexpression of all 14 genes, and underexpression of 3 other MT-Rel genes (MAST4, MAPT and MTUS1) are associated with poor breast cancer patient survival. A Systems Biology approach highlighted three major functional networks connecting the 17 MT-Rel genes and their partners, which are centered on spindle assembly, chromosome segregation and cytokinesis. Our studies identified mitotic Aurora kinases and their substrates as major targets for therapeutic approaches against breast cancer.

Keywords: Aurora kinases; Systems Biology; biomarker; kinesins; mitotic defects; prognostic value; therapeutic targets.

Figure 1

Expression of 17 MT-Rel genes in breast tumors and normal tissues. (A) Boxplots of mRNA expression level of the 17 MT-Rel genes in 112 breast tumors (bold boxplot) and their adjacent normal tissues (thin boxplot) from the TNMplot database (tnmplot.com). (B) Histograms of fold change of gene expression between tumor and normal breast tissues. Dotted line indicates a fold change value of 1.4. (C) Proportion of tumor samples showing higher expression of a given gene compared to normal samples using the third quartile as cutoff value. (D) Proportion of tumor samples showing lower expression of a given gene compared to normal samples using the first quartile as cutoff value. (E) Probeset intensities for each indicated gene in breast tumors from the REMAGUS02 cohort [20] classified according to their molecular subtype; ER+ (●), HER2+ (▲), TNBC (▼). A blue line indicates the median value. * p < 0.05; *** p < 0.001; **** p < 0.0001.

Figure 2

Co-regulation of expression of 17 MT-Rel genes in breast tumor and normal tissues. (A) Heat map hierarchical clustering of the 17 MT-Rel genes in normal (red) and tumor (blue) breast samples. Dendogram at the bottom shows the clustering of normal samples in red (cluster 3), and two clusters of tumors in blue (cluster 1) and green (cluster 2). (B) Heat map of Pearson correlation coefficient (r) in normal (left) and tumor (right) breast samples.

Figure 3

Survival curves of breast cancer patients according to MT-Rel gene expression level. (A) Overall survival curves of breast cancer patients according to KIF4A (218355_at) or MTUS1 (212096_s_at) probeset intensities from KMplotter (kmplot.com). (B) Relapse-free survival curves as in (A).

Figure 4

Effect of MT-Rel gene silencing on breast cancer cell viability. Cell viability was measured following silencing of each MT-Rel gene by siRNA transfection (96 h) into MDA-MB-231 (A) and MDA-MB-468 (B) breast cancer cell lines. Shown are mean values and standard deviation from four independent experiments. Significant pvalues are indicated in red; * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. (C) Scattered dot plot of gene dependency score (Chronos). A lower Chronos score indicates a higher likelihood that the gene of interest is essential in a given cell line. A score of 0 indicates that a gene is non-essential. The purple lines indicate the median values.

Figure 5

Systems Biology-derived networks connecting 17 MT-Rel genes. (A). The 17 MT-Rel genes (in orange) and their partners are linked by edges displayed in red for inhibitions, green for activations and black for protein–protein interactions. Node colors were assigned as follows: orange for the 17 MT-Rel genes differentially expressed in the three considered breast cancer transcriptomic datasets, blue for 17 MT-Rel genes differentially expressed in two of these datasets, light blue for genes with normal expression in breast cancer but present in the initial list of 280 MT-Rel genes and light purple for genes not present in the list of 280 MT-Rel genes. Using the tnmplot.com site, we annotated the fold change between normal and breast cancer tissues on the node: red contour if the gene is up-regulated, and green contour if the gene is down-regulated. The thicker is the border, the higher the fold change. (B) Sub-networks extracted from the network shown in (A) are associated with following enriched GO terms “spindle organization” (GO:0007051, p = 6 × 10⁻¹⁶) (left), “Mitotic sister chromatid segregation” (GO:0000070, p = 2 × 10⁻¹⁷) (middle) and “cytokinesis” GO:0000910, p = 3 × 10⁻¹¹) (right panel).

Figure 6

Schematic representation recapitulating the functional impact of MT-Rel genes in breast cancer.