Prognostic and Functional Significant of Heat Shock Proteins (HSPs) in Breast Cancer Unveiled by Multi-Omics Approaches

Abstract

Heat shock proteins (HSPs) are a well-characterized molecular chaperones protein family, classified into six major families, according to their molecular size. A wide range of tumors have been shown to express atypical levels of one or more HSPs, suggesting that they could be used as biomarkers. However, the collective role and the possible coordination of HSP members, as well as the prognostic significance and the functional implications of their deregulated expression in breast cancer (BC) are poorly investigated. Here, we used a systematic multi-omics approach to assess the HSPs expression, the prognostic value, and the underlying mechanisms of tumorigenesis in BC. By using data mining, we showed that several HSPs were deregulated in BC and significantly correlated with a poor or good prognosis. Functional network analysis of HSPs co-expressed genes and miRNAs highlighted their regulatory effects on several biological pathways involved in cancer progression. In particular, these pathways concerned cell cycle and DNA replication for the HSPs co-expressed genes, and miRNAs up-regulated in poor prognosis and Epithelial to Mesenchymal Transition (ETM), as well as receptors-mediated signaling for the HSPs co-expressed genes up-regulated in good prognosis. Furthermore, the proteomic expression of HSPs in a large sample-set of breast cancer tissues revealed much more complexity in their roles in BC and showed that their expression is quite variable among patients and confined into different cellular compartments. In conclusion, integrative analysis of multi-omics data revealed the distinct impact of several HSPs members in BC progression and indicate that collectively they could be useful as biomarkers and therapeutic targets for BC management.

Keywords: HSPs; breast cancer; data mining; expression; miRNAs; prognosis; proteomics.

Figure 1

Gene expression analysis of HSP members between normal and breast cancer samples analyzed by UALCAN database. (A) Box plots of HSP genes up-regulated in BC compared to normal tissues. (B) Box plots of HSP genes down-regulated in BC compared to normal tissues. p < 0.01 was considered significant.

Figure 1

Gene expression analysis of HSP members between normal and breast cancer samples analyzed by UALCAN database. (A) Box plots of HSP genes up-regulated in BC compared to normal tissues. (B) Box plots of HSP genes down-regulated in BC compared to normal tissues. p < 0.01 was considered significant.

Figure 2

Analysis of genetic alterations in HSP members differentially expressed in BC. The percentage of alterations in HSP genes was extracted by using the OncoPrint tool of cBioPortal containing sequencing data of 2509 patients. Red and blue represent amplification, and deep deletion, respectively, green and grey represent missense and truncating mutations. Asterisk * indicates the corresponding HSP members showing a percentage of alteration higher than 3%. Histograms showing the percentage of genetic alterations of HSP genes up-regulated (A), and down-regulated (B) in BC.

Figure 3

Analysis of promoter methylation status in HSP members differentially expressed in BC. For each HSP member the levels of promoter methylation between normal and breast cancer were retrieved from UALCAN database. Beta value cut-off indicates hyper-methylation [beta value: 0.7–0.5] or hypo-methylation [beta-value: 0.3–0.25]. HSP members with hyper-methylated promoters are marked in green; HSP members with hypo-methylated promoters are marked in black. Box plots showing the promoter methylation status between normal and breast primary tumors in HSPs up-regulated (A) and down-regulated in BC (B). p < 0.01 was considered significant.

Figure 4

Survival curves of HSP members significantly associated with prognosis. The analysis is shown for Relapse Free Survival (RFS) Overall Survival (OS) and Distant Metastasis Free Survival (DMFS) using the Kaplan-Meier Plotter database. Patients were divided into two groups by using the best cut-off of probe expression. Survival curves of HSP-cluster associated with poor (A) and good (B) prognosis.

Figure 5

Enrichment pathway analysis of HSPs co-expressed genes significantly correlated with prognosis. Significantly enriched biological pathways were ranked by p-value using the FunRich 3.0 software. (A,B) The top six biological pathways in which the HSPs co-expressed genes up-regulated in poor and good prognosis, derived from GOBO, were significantly involved. (C,D) Venn diagrams of overlapping HSPs co-expressed genes up-regulated in poor and good prognosis, derived from GOBO, UALCAN, and bc-GenExMiner. (E,F) Protein-protein interaction network of the common HSPs co-expressed genes derived from the three cross-platforms GOBO, UALCAN and bc-GenExMiner using STRING database. The network was performed with medium confidence (combined score > 0.4) and disconnected nodes were deleted.

Figure 6

Proteomic expression of HSP members and prognostic significance. (A) Survival analysis of HSPs protein expression evaluated by immunohistochemistry (IHC) staining, using The Human Protein Atlas (HPA) database. The database was queried for each HSP member, and only significant analyses were reported; (B) Representative proteomic map of a breast cancer tissue. The 2-D separation was performed on IPG gel strips (18 cm, 3.0–10 NL) followed by the SDS-PAGE on a vertical linear-gradient slab gel (9–16%T). HSP protein members identified by Maldi-Tof spectrometry are marked with the Gene name. When present, different isoforms of the same protein were jointly labeled; (C) Quantitative analysis of HSP proteins, given as box-plot graph, in breast cancer tissues. Quantification was deduced by the 2D gels, analyzed by Image-Master software. In the ordinate are the values of N%V; (D) Representative IHC images for individual HSP proteins (HSPB1, HSPD1, and HSPA8) in breast cancer tissues showing weak, moderate and strong staining, respectively retrieved by the HPA database. The antibody immune reactivity was observed in several cell compartments, such as cytoplasm, plasma membrane as well as in nucleus.