Weighted frequent gene co-expression network mining to identify genes involved in genome stability

Abstract

Gene co-expression network analysis is an effective method for predicting gene functions and disease biomarkers. However, few studies have systematically identified co-expressed genes involved in the molecular origin and development of various types of tumors. In this study, we used a network mining algorithm to identify tightly connected gene co-expression networks that are frequently present in microarray datasets from 33 types of cancer which were derived from 16 organs/tissues. We compared the results with networks found in multiple normal tissue types and discovered 18 tightly connected frequent networks in cancers, with highly enriched functions on cancer-related activities. Most networks identified also formed physically interacting networks. In contrast, only 6 networks were found in normal tissues, which were highly enriched for housekeeping functions. The largest cancer network contained many genes with genome stability maintenance functions. We tested 13 selected genes from this network for their involvement in genome maintenance using two cell-based assays. Among them, 10 were shown to be involved in either homology-directed DNA repair or centrosome duplication control including the well-known cancer marker MKI67. Our results suggest that the commonly recognized characteristics of cancers are supported by highly coordinated transcriptomic activities. This study also demonstrated that the co-expression network directed approach provides a powerful tool for understanding cancer physiology, predicting new gene functions, as well as providing new target candidates for cancer therapeutics.

Figure 1. Workflow to mine frequent co-expression network using QCM from cancer and normal tissue microarray datasets.

Blue ovals indicate gene members shared by different networks, ovals in other colors indicate genes unique to each network.

Figure 2. Comparison of networks identified from multiple cancer vs. normal tissue microarray datasets.

Top 13 networks (ranked by size) were shown. The size of each circle represents the relative size of each network. The numbers inside the circles indicate the size of the network. The numbers above the connection line indicate the numbers of common genes shared by the two networks. Different top-enriched biological functions in each network were assigned with different colors. ECM: extracellular matrix construction. Parameter settings are: β = 0.8, γ = 0.8, λ = 2.0, t = 1.0 (for networks from cancer datasets); β = 0.8, γ = 0.7, λ = 2.0, t = 1.0 (for networks from normal tissue datasets).

Figure 3. Validated protein-protein interactions on genes from networks identified from cancer datasets using IPA.

The edges represent validated protein-protein interactions obtained from Ingenuity Knowledge Base. The nodes are gene members. Only members with connection to other members are shown. A: Validated protein-protein interactions on genes from Cancer Network 1 (cell proliferation/cell cycle control network) using IPA. The red circles indicate the genes further selected for genome stability function assays using RNAi. B: Validated protein-protein interactions on genes from Cancer Network 6 (extracellular matrix network) using IPA.

Figure 4. Kaplan-Meier curve of breast cancer, glioblastoma (GBM) and ovarian cancer (OV) using network genes identified from cancer datasets.

The p-values are computed using Log- rank test with 100 repeats. A: using Network 1 genes on NKI mixed cohort; B: using Van't Veer 70-gene signature on NKI mixed cohort; C: using Network 1 genes on NKI LN+ cohort; D: using van't Veer 70-gene signature on NKI LN+ cohort; E: using Network1 genes on NKI ER− cohort; F: using Van't Veer 70-gene signature on NKI ER− data. G: using Network 18 genes on TCGA GBM dataset; H: using 23-gene signature on TCGA GBM cohort . I: using Network 17 genes on TCGA OV cohort. J: using 19-gene signature on TCGA OV dataset . Blue lines: good survival outcome group; Red lines: poor survival outcome group. LN+: lymph node positive. ER−: estrogen receptor negative.

Figure 5. Cell-based assays to test gene involvement in the genome stability maintenance using RNAi.

Cells transfected with firefly gene GL2 siRNA were used as the negative control for both assays. A: HR assay on HeLa cells depleting target gene expression by siRNA. Error bar represents standard error. Asterisks indicate the results with statistically significant decreased activity upon siRNA depletion using Student's test (p<0.05). Black line represents the 80% of HR activity in the control sample as a cutoff. B: Centrosome assay on HeLa cell line depleting target gene expression by siRNA. Error bar represents standard error.