A Bioinformatic Pipeline Places STAT5A as a miR-650 Target in Poorly Differentiated Aggressive Breast Cancer

Abstract

Breast cancer (BRCA) is a leading cause of mortality among women. Tumors often acquire aggressive features through genomic aberrations affecting cellular programs, e.g., the epithelial to mesenchymal transition (EMT). EMT facilitates metastasis leading to poor prognosis. We previously observed a correlation between an amplification of miR-650 (Amp-650) and EMT features in BRCA samples isolated from Mexican patients. In this study, we explored the cBioportal database aiming to extend that observation and better understand the importance of Amp-650 for BRCA aggressiveness. We found that Amp-650 is more frequent in aggressive molecular subtypes of BRCA, as well as in high grade poorly differentiated tumors, which we confirmed in an external miRNA expression database. We performed differential expression analysis on samples harboring Amp-650, taking advantage of gene target prediction tools and tumor suppressor gene databases to mine several hundreds of differentially underexpressed genes. We observed STAT5A as a likely putative target gene for miR-650 in aggressive poorly differentiated BRCA. Samples with both Amp-650 and low expression of STAT5A had less overall survival than samples with either or none of the alterations. No target gene has been described for miR-650 in BRCA, thus, this bioinformatic study provides valuable information that should be corroborated experimentally.

Keywords: STAT5A; amplification; bioinformatic; breast cancer; miR-650; target.

Figure 1

Frequency of Amp-650 in samples with different clinical classifications. The y axis indicates the amplification frequency of miR-650 in samples classified by (a) ethnicity, (b) molecular subtype, (c) HER2 status of Luminal B samples (only samples having both molecular and HER2 data were included). (d) Neoplasm Histologic Grade (NHG), (e) Nottingham Prognosis Index (NPI), (f) tumor stage, and (g) differential expression analysis made with GSE86278 dataset over miR-650. Values above each bar indicate the number of samples in each group. Prognosis categorization into good, moderate, and poor, was made based on the NPI thresholds proposed by Albergaria, A. et al. [9] and poor prognosis samples were combined with moderate prognosis samples to increase the group size: poor (n = 199) and moderate (n = 1101). Likewise, tumor staging 0 (n = 12) with 1 (n = 695) and 3 (n = 323) with 4 (n = 30) were merged due to the small sample size. Significant differences are indicated with an asterisk (Z > 1.645).

Figure 2

Schematic representation of the workflow followed to mine genes from differential expression analysis of putative targets of miR-650. Red boxes indicate genes that were carried to subsequent analysis. Aggressive breast cancers (BRCA) are a compilation of basal, HER2-enriched, and NHG 3, while nonaggressive are a compilation of luminal subtypes and NHG 2.

Figure 3

Pathway enrichment analysis of putative target genes and their expression correlation. (a) Protein–protein interaction networks. Red squares denote the miR-650 putative target genes used as input in the STRING software. (b) Correlation matrix of microarray expression data from METABRIC using samples of basal, HER2-enriched, or NHG 3 subtypes and harboring Amp-650 (n = 63). Blue and red colors indicate positive and negative correlations, respectively, the larger the circle the closer the correlation to 1:1, and crossed circles denote a not significant correlation. A similar analysis with TCGA gene expression data also does not support a correlation between these genes (data not shown). Correlation plots generated with corrplot package [19].

Figure 4

(a) Z-score was used to set high expressing samples (above + 1) or low expressing samples (below -1) of STAT5A. (b) Prognostic significance of Amp-650 in combination with the expression level of STAT5A. There were only two samples with high STAT5A (STAT5A above + 1) + Amp-650. Dashed lines indicate median OS (Months): High STAT5A = 203.7, Low STAT5A = 122.17, and Low STAT5A + Amp-650 = 99.55. Statistical differences are indicated with a line accompanied by its p-value. Samples without amplification (w/o Amp-650). Both plots were constructed with downloaded Z-score distribution of Illumina Human v3 microarray data from METABRIC (n = 1980).

Figure 5

Amp-650 leads to overexpression of miR-650, which causes downregulation of STAT5A. Low expression of STAT5A contributes to the formation of BRCA aggressive subtypes due to facilitating EMT-related features. Closeup circle shows miR-650 seed region and STAT5A sequence complementarity according to TargetScan (http://www.targetscan.org/vert_72/). The main drawing was made with the free version of BioRender (https://biorender.com/).