Gene regulatory networks (GRNs) research reveals complex life phenomena from the perspective of gene interaction, which is an important research field in systems biology. Traditional Bayesian networks have a high computational complexity, and the network structure scoring model has a single feature. Information-based approaches cannot identify the direction of regulation. In order to make up for the shortcomings of the above methods, this paper presents a novel hybrid learning method (DBNCS) based on dynamic Bayesian network (DBN) to construct the multiple time-delayed GRNs for the first time, combining the comprehensive score (CS) with the DBN model. DBNCS algorithm first uses CMI2NI (conditional mutual inclusive information-based network inference) algorithm for network structure profiles learning, namely the construction of search space. Then the redundant regulations are removed by using the recursive optimization algorithm (RO), thereby reduce the false positive rate. Secondly, the network structure profiles are decomposed into a set of cliques without loss, which can significantly reduce the computational complexity. Finally, DBN model is used to identify the direction of gene regulation within the cliques and search for the optimal network structure. The performance of DBNCS algorithm is evaluated by the benchmark GRN datasets from DREAM challenge as well as the SOS DNA repair network in Escherichia coli, and compared with other state-of-the-art methods. The experimental results show the rationality of the algorithm design and the outstanding performance of the GRNs.

Tamoxifen is an effective anti-estrogen treatment for patients with estrogen receptor-positive (ER+) breast cancer, however, tamoxifen resistance is frequently observed. To elucidate the underlying molecular mechanisms of tamoxifen resistance, we performed a systematic analysis of miRNA-mediated gene regulation in three clinically-relevant tamoxifen-resistant breast cancer cell lines (TamRs) compared to their parental tamoxifen-sensitive cell line. Alterations in the expression of 131 miRNAs in tamoxifen-resistant vs. parental cell lines were identified, 22 of which were common to all TamRs using both sequencing and LNA-based quantitative PCR technologies. Although the target genes affected by the altered miRNA in the three TamRs differed, good agreement in terms of affected molecular pathways was observed. Moreover, we found evidence of miRNA-mediated regulation of ESR1, PGR1, FOXM1 and 14-3-3 family genes. Integrating the inferred miRNA-target relationships, we investigated the functional importance of 2 central genes, SNAI2 and FYN, which showed increased expression in TamR cells, while their corresponding regulatory miRNA were downregulated. Using specific chemical inhibitors and siRNA-mediated gene knockdown, we showed that both SNAI2 and FYN significantly affect the growth of TamR cell lines. Finally, we show that a combination of 2 miRNAs (miR-190b and miR-516a-5p) exhibiting altered expression in TamR cell lines were predictive of treatment outcome in a cohort of ER+ breast cancer patients receiving adjuvant tamoxifen mono-therapy. Our results provide new insight into the molecular mechanisms of tamoxifen resistance and may form the basis for future medical intervention for the large number of women with tamoxifen-resistant ER+ breast cancer.

MYCN, an oncogenic transcription factor of the Myc family, is a major driver of neuroblastoma tumorigenesis. Due to the difficulty in drugging MYCN directly, revealing the molecules in MYCN regulatory networks will help to identify effective therapeutic targets for neuroblastoma therapy. Here we perform ChIP-sequencing and small RNA-sequencing of neuroblastoma cells to determine the MYCN-binding sites and MYCN-associated microRNAs, and integrate various types of genomic data to construct MYCN regulatory networks. The overall analysis indicated that MYCN-regulated genes were involved in a wide range of biological processes and could be used as signatures to identify poor-prognosis MYCN-non-amplified patients. Analysis of the MYCN binding sites showed that MYCN principally served as an activator. Using a computational approach, we identified 32 MYCN co-regulators, and some of these findings are supported by previous studies. Moreover, we investigated the interplay between MYCN transcriptional and microRNA post-transcriptional regulations and identified several microRNAs, such as miR-124-3p and miR-93-5p, which may significantly contribute to neuroblastoma pathogenesis. We also found MYCN and its regulated microRNAs acted together to repress the tumor suppressor genes. This work provides a comprehensive view of MYCN regulations for exploring therapeutic targets in neuroblastoma, as well as insights into the mechanism of neuroblastoma tumorigenesis.

Retinoblastoma, a very aggressive cancer of the developing retina, initiatiates by the biallelic loss of RB1 gene, and progresses very quickly following RB1 inactivation. While its genome is stable, multiple pathways are deregulated, also epigenetically. After reviewing the main findings in relation with recently validated markers, we propose an integrative bioinformatics approach to include in the previous group new markers obtained from the analysis of a single cell line subject to epigenetic treatment. In particular, differentially expressed genes are identified from time course microarray experiments on the WERI-RB1 cell line treated with 5-Aza-2'-deoxycytidine (decitabine; DAC). By inducing demethylation of CpG island in promoter genes that are involved in biological processes, for instance apoptosis, we performed the following main integrative analysis steps: i) Gene expression profiling at 48h, 72h and 96h after DAC treatment; ii) Time differential gene co-expression networks and iii) Context-driven marker association (transcriptional factor regulated protein networks, master regulatory paths). The observed DAC-driven temporal profiles and regulatory connectivity patterns are obtained by the application of computational tools, with support from curated literature. It is worth emphasizing the capacity of networks to reconcile multi-type evidences, thus generating testable hypotheses made available by systems scale predictive inference power. Despite our small experimental setting, we propose through such integrations valuable impacts of epigenetic treatment in terms of gene expression measurements, and then validate evidenced apoptotic effects.

Human bladder cancer tumors have been shown to contain a subpopulation of cells with stem-like characteristics that may trigger tumor growth, recurrence, and metastasis. These cells, known as tumor-initiating cells (TICs), would be effective diagnostic tools and valuable therapeutic targets. Here, we report the isolation of TICs from seven bladder cancer cell lines and show that TICs from different sources vary on their ability to form tumorspheres in vitro and generate xenografts in vivo, which suggest they are remarkably heterogeneous. We used the Affymetrix PrimeView™ Human Gene Expression Array to analyze gene expression profiles of bladder TICs, which may help understand their tumorigenic capacities and develop novel treatments specifically targeted toward these cells. We then constructed a transcription factor-gene regulatory network that includes three key transcription factors that are involved in cell survival, differentiation, proliferation, and apoptosis. We validated our findings by analyzing mRNA expression of the key genes in this network in 24 clinical tissues. Our results suggest that this transcription factor-gene regulatory network could be useful in the development of clinical diagnostic tools and therapy approaches for bladder cancer.

Altered functioning of the biological clock is involved in cancer onset and progression. MicroRNAs (miRNAs) interact with the clock genes modulating the function of genetically encoded molecular clockworks. Collaborative interactions may take place within the coding-noncoding RNA regulatory networks. We aimed to evaluate the cross-talk among miRNAs and clock genes in colorectal cancer (CRC). We performed an integrative analysis of miRNA-miRNA and miRNA-mRNA interactions on high-throughput molecular profiling of matched human CRC tissue and non-tumor mucosa, pinpointing core clock genes and their targeting miRNAs. Data obtained in silico were validated in CRC patients and human colon cancer cell lines. In silico we found severe alterations of clock gene-related coding-noncoding RNA regulatory networks in tumor tissues, which were later corroborated by the analysis of human CRC specimens and experiments performed in vitro. In conclusion, specific miRNAs target and regulate the transcription/translation of clock genes and clock gene-related miRNA-miRNA as well as mRNA-miRNA interactions are altered in colorectal cancer. Exploration of the interplay between specific miRNAs and genes, which are critically involved in the functioning of the biological clock, provides a better understanding of the importance of the miRNA-clock genes axis and its derangement in colorectal cancer.

It has been difficult to elucidate the structure of gene regulatory networks under anticancer drug treatment. Here, we developed an algorithm to highlight the hub genes that play a major role in creating the upstream and downstream relationships within a given set of differentially expressed genes. The directionality of the relationships between genes was defined using information from comprehensive collections of transcriptome profiles after gene knockdown and overexpression. As expected, among the drug-perturbed genes, our algorithm tended to derive plausible hub genes, such as transcription factors. Our validation experiments successfully showed the anticipated activity of certain hub gene in establishing the gene regulatory network that was associated with cell growth inhibition. Notably, giving such top priority to the hub gene was not achieved by ranking fold change in expression and by the conventional gene set enrichment analysis of drug-induced transcriptome data. Thus, our data-driven approach can facilitate to understand drug-induced gene regulatory networks for finding potential functional genes.

The major part of the genome that was previously called junk DNA has been shown to be dynamically transcribed to produce non-coding RNAs. Among them, the long non-coding RNAs (lncRNA) play diverse roles in the cellular context and are therefore involved in various diseases like cancer. LncRNA transcript profiling of glioblastoma (n = 19) and control brain samples (n = 9) identified 2,774 and 5,016 lncRNAs to be upregulated and downregulated in GBMs respectively. Correlation analysis of differentially regulated lncRNAs with mRNA and lncRNA identified several lncRNAs that may potentially regulate many tumor relevant mRNAs and lncRNAs both at nearby locations (cis) and far locations (trans). Integration of our data set with TCGA GBM RNA-Seq data (n = 172) revealed many lncRNAs as a host as well as decoy for many tumor regulated miRNAs. The expression pattern of seven lncRNAs- HOXD-AS2, RP4-792G4.2, CRNDE, ANRIL, RP11-389G6.3, RP11-325122.2 and AC123886.2 was validated by TCGA RNA-Seq data and RT-qPCR. Silencing ANRIL, a GBM upregulated lncRNA, inhibited glioma cell proliferation and colony growth. Cox regression analysis identified several prognostic lncRNAs. An lncRNA risk score derived from five lnRNAs-RP6-99M1.2, SOX21-AS1, CTD-2127H9.1, RP11-375B1.3 and RP3-449M8.9 predicted survival independent of all other markers. Multivariate cox regression analysis involving G-CIMP, IDH1 mutation, MGMT promoter methylation identified lncRNA risk score to be an independent poor predictor of GBM survival. The lncRNA risk score also stratified GBM patients into low and high risk with significant survival difference. Thus our study demonstrates the importance of lncRNA in GBM pathology and underscores the potential possibility of targeting lncRNA for GBM therapy.

The identification of overexpressed miRNAs in multiple myeloma (MM) has progressively added a further level of complexity to MM biology. miRNA and gene expression profiles of two large representative MM datasets, available from retrospective and prospective series and encompassing a total of 249 patients at diagnosis, were analyzed by means of in silico integrative genomics methods, based on MAGIA2 and Micrographite computational procedures. We first identified relevant miRNA/transcription factors/target gene regulation circuits in the disease and linked them to biological processes. Members of the miR-99b/let-7e/miR-125a cluster, or of its paralog, upregulated in t(4;14), were connected with the specific transcription factors PBX1 and CEBPA and several target genes. These results were validated in two additional independent plasma cell tumor datasets. Then, we reconstructed a non-redundant miRNA-gene regulatory network in MM, linking miRNAs, such as let-7g, miR-19a, mirR-20a, mir-21, miR-29 family, miR-34 family, miR-125b, miR-155, miR-221 to pathways associated with MM subtypes, in particular the ErbB, the Hippo, and the Acute myeloid leukemia associated pathways.

Triple-negative breast cancer (TNBC) is an aggressive breast cancer subtype. Genome-scale molecular characteristics and regulatory mechanisms that distinguish TNBC from other subtypes remain incompletely characterized.

Mutation accumulation and epigenetic alterations in genes are important for carcinogenesis. Because leukemogenesis-related signal pathways have been investigated and microarray sample data have been produced in acute myeloid leukemia (AML), myelodysplastic syndromes (MDS) and normal cells, systems analysis in coupling pathways becomes possible. Based on system modeling and identification, we could construct the coupling pathways and their associated gene regulatory networks using microarray sample data. By applying system theory to the estimated system model in coupling pathways, we can then obtain transductivity sensitivity, basal sensitivity and error sensitivity of each protein to identify the potential impact of genetic mutations, epigenetic alterations and the coupling of other pathways from the perspective of energy, respectively. By comparing the results in AML, MDS and normal cells, we investigated the potential critical genetic mutations and epigenetic alterations that activate or repress specific cellular functions to promote MDS or AML leukemogenesis. We suggested that epigenetic modification of β-catenin and signal integration of CSLs, AP-2α, STATs, c-Jun and β-catenin could contribute to cell proliferation at AML and MDS. Epigenetic regulation of ERK and genetic mutation of p53 could lead to the repressed apoptosis, cell cycle arrest and DNA repair in leukemic cells. Genetic mutation of JAK, epigenetic regulation of ERK, and signal integration of C/EBPα could result in the promotion of MDS cell differentiation. According to the results, we proposed three drugs, decitabine, genistein, and monorden for preventing AML leukemogenesis, while three drugs, decitabine, thalidomide, and geldanamycin, for preventing MDS leukemogenesis.

There are many stages in the development and metastasis of colorectal cancer (CRC). In this study, we compared the differential expression genes in different stages of metastatic CRC. Then, we screened the continuously up-regulated genes and the continuously down-regulated genes that were associated with the development and metastasis of CRC. After analyzing the intersection of differential expression genes in each stage, we screened the continuously up-regulated genes and deviated genes in the extracellular matrix and the continuously down-regulated genes and deviated genes in the mitochrondia of CRC. Then, we performed gene ontology enrichment analysis of the deviated genes in different phases, and we found that key molecular events occurred in the period extending from stage II to III (early stage of metastasis) of CRC. Furthermore, in this period we found that the chemotaxis of inflammatory cells had decreased in the extracellular matrix. On the other hand, the aerobic respiration had increased in the mitochondrion. Then, we constructed protein-protein interaction network of deviated genes in the extracellular matrix and mitochondrion. We used the network module and hub network to analyze the protein-protein interaction network. The network module analysis showed that the protein complex of VEGFA and CCL7-CCR3 is the key node in the extracellular matrix, while MAPK1 is the key node in the mitochondrion. The hub network analysis showed that the signal transmission chain FN1→SPARC→COL1A1→MMP2 is the key regulatory pathway for extracellular signal transmission. Furthermore, it also showed that CAV1→MAPK3→RAF1→NR3C1→MAPK1→ESR1 is the key regulatory pathway for signal transmission in mitochondrion.

Herein, we aimed at identifying global transcriptome microRNA (miRNA) changes and miRNA target genes in lung adenocarcinoma. Samples were selected as training (N = 24) and independent validation (N = 34) sets. Tissues were microdissected to obtain >90% tumor or normal lung cells, subjected to miRNA transcriptome sequencing and TaqMan quantitative PCR validation. We further integrated our data with published miRNA and mRNA expression datasets across 1,491 lung adenocarcinoma and 455 normal lung samples. We identified known and novel, significantly over- and under-expressed (p ≤ 0.01 and FDR≤0.1) miRNAs in lung adenocarcinoma compared to normal lung tissue: let-7a, miR-10a, miR-15b, miR-23b, miR-26a, miR-26b, miR-29a, miR-30e, miR-99a, miR-146b, miR-181b, miR-181c, miR-421, miR-181a, miR-574 and miR-1247. Validated miRNAs included let-7a-2, let-7a-3, miR-15b, miR-21, miR-155 and miR-200b; higher levels of miR-21 expression were associated with lower patient survival (p = 0.042). We identified a regulatory network including miR-15b and miR-155, and transcription factors with prognostic value in lung cancer. Our findings may contribute to the development of treatment strategies in lung adenocarcinoma.

The rapid accumulation of cancer-related data owing to high-throughput technologies has provided unprecedented choices to understand the progression of cancer and discover functional networks in multiple cancers. Establishment of co-expression networks will help us to discover the systemic properties of carcinogenesis features and regulatory mechanisms of multiple cancers. Here, we proposed a computational workflow to identify differentially co-expressed gene modules across 8 cancer types by using combined gene differential expression analysis methods and a higher-order generalized singular value decomposition. Four co-expression modules were identified; and oncogenes and tumor suppressors were significantly enriched in these modules. Functional enrichment analysis demonstrated the significantly enriched pathways in these modules, including ECM-receptor interaction, focal adhesion and PI3K-Akt signaling pathway. The top-ranked miRNAs (mir-199, mir-29, mir-200) and transcription factors (FOXO4, E2A, NFAT, and MAZ) were identified, which play an important role in deregulating cellular energetics; and regulating angiogenesis and cancer immune system. The clinical significance of the co-expressed gene clusters was assessed by evaluating their predictability of cancer patients' survival. The predictive power of different clusters and subclusters was demonstrated. Our results will be valuable in cancer-related gene function annotation and for the evaluation of cancer patients' prognosis.

Despite efforts for extensive molecular characterization of cancer patients, such as the international cancer genome consortium (ICGC) and the cancer genome atlas (TCGA), the heterogeneous nature of cancer and our limited knowledge of the contextual function of proteins have complicated the identification of targetable genes. Here, we present Aberration Hub Analysis for Cancer (AbHAC) as a novel integrative approach to pinpoint aberration hubs, i.e. individual proteins that interact extensively with genes that show aberrant mutation or expression. Our analysis of the breast cancer data of the TCGA and the renal cancer data from the ICGC shows that aberration hubs are involved in relevant cancer pathways, including factors promoting cell cycle and DNA replication in basal-like breast tumors, and Src kinase and VEGF signaling in renal carcinoma. Moreover, our analysis uncovers novel functionally relevant and actionable targets, among which we have experimentally validated abnormal splicing of spleen tyrosine kinase as a key factor for cell proliferation in renal cancer. Thus, AbHAC provides an effective strategy to uncover novel disease factors that are only identifiable by examining mutational and expression data in the context of biological networks.

Despite improvements in surgical procedures and chemotherapy, pancreatic cancer remains one of the most aggressive and fatal human malignancies, with a low 5-year survival rate of only 8%. Therefore, novel strategies for prevention and treatment are urgently needed. Here, we investigated the mechanisms underlying the anti-pancreatic cancer effects dihydroartemisinin (DHA). Microarray and systematic analysis showed that DHA suppressed proliferation, inhibited angiogenesis and promoted apoptosis in two different human pancreatic cancer cell lines, and that 5 DHA-regulated microRNAs and 11 of their target mRNAs were involved in these effects via 19 microRNA-mRNA interactions. Four of these microRNAs, 9 of the mRNAs and 17 of the interactions were experimentally verified. Furthermore, we found that the anti-pancreatic caner effects of DHA in vivo involved 4 microRNAs, 9 mRNAs and 17 microRNA-mRNA interactions. These results improve the understanding of the mechanisms by which DHA suppresses proliferation and angiogenesis and promotes apoptosis in pancreatic cancer cells and indicate that DHA, an effective antimalarial drug, might improve pancreatic cancer treatments.

LINE-1 retrotransposons encode the reverse transcriptase (RT) enzyme, required for their own mobility, the expression of which is inhibited in differentiated tissues while being active in tumors. Experimental evidence indicate that the inhibition of LINE-1-derived RT restores differentiation in cancer cells, inhibits tumor progression and yields globally reprogrammed transcription profiles. Newly emerging data suggest that LINE-1-encoded RT modulates the biogenesis of miRNAs, by governing the balance between the production of regulatory double-stranded RNAs and RNA:DNA hybrid molecules, with a direct impact on global gene expression. Abnormally high RT activity unbalances the transcriptome in cancer cells, while RT inhibition restores "normal" miRNA profiles and their regulatory networks. This RT-dependent mechanism can target the myriad of transcripts - both coding and non-coding, sense and antisense - in eukaryotic transcriptomes, with a profound impact on cell fates. LINE-1-encoded RT emerges therefore as a key regulator of a previously unrecognized mechanism in tumorigenesis.

To develop and evaluate the long-term prophylactic treatment for chronic diseases such as osteoporosis requires a clear view of mechanism at the molecular and systems level. While molecular signaling pathway studies for osteoporosis are extensive, a unifying mechanism is missing. In this work, we provide experimental and systems-biology evidences that a tightly connected top-level regulatory network may exist, which governs the normal and osteoporotic phenotypes of osteoblast. Specifically, we constructed a hub-like interaction network from well-documented cross-talks among estrogens, glucocorticoids, retinoic acids, peroxisome proliferator-activated receptor, vitamin D receptor and calcium-signaling pathways. The network was verified with transmission electron microscopy and gene expression profiling for bone tissues of ovariectomized (OVX) rats before and after strontium gluconate (GluSr) treatment. Based on both the network structure and the experimental data, the dynamical modeling predicts calcium and glucocorticoids signaling pathways as targets for GluSr treatment. Modeling results further reveal that in the context of missing estrogen signaling, the GluSr treated state may be an outcome that is closest to the healthy state.

MiRNA expression abnormalities in adenocarcinoma arising from pancreatic ductal system (PDAC) and Vater's papilla (PVAC) could be associated with distinctive pathologic features and clinical cancer behaviours. Our previous miRNA expression profiling data on PDAC (n=9) and PVAC (n=4) were revaluated to define differences/similarities in miRNA expression patterns. Afterwards, in order to uncover target genes and core signalling pathways regulated by specific miRNAs in these two tumour entities, miRNA interaction networks were wired for each tumour entity, and experimentally validated target genes underwent pathways enrichment analysis. One hundred and one miRNAs were altered, mainly over-expressed, in PDAC samples. Twenty-six miRNAs were deregulated in PVAC samples, where more miRNAs were down-expressed in tumours compared to normal tissues. Four miRNAs were significantly altered in both subgroups of patients, while 27 miRNAs were differentially expressed between PDAC and PVAC. Although miRNA interaction networks were more complex and dense in PDAC than in PVAC, pathways enrichment analysis uncovered a functional overlapping between PDAC and PVAC. However, shared signalling events were influenced by different miRNA and/or genes in the two tumour entities. Overall, specific miRNA expression patterns were involved in the regulation of a limited core signalling pathways in the biology landscape of PDAC and PVAC.

Long non-coding RNAs (lncRNAs) play key roles in diverse biological processes. Moreover, the development and progression of cancer often involves the combined actions of several lncRNAs. Here we propose a multi-step method for constructing lncRNA-lncRNA functional synergistic networks (LFSNs) through co-regulation of functional modules having three features: common coexpressed genes of lncRNA pairs, enrichment in the same functional category and close proximity within protein interaction networks. Applied to three cancers, we constructed cancer-specific LFSNs and found that they exhibit a scale free and modular architecture. In addition, cancer-associated lncRNAs tend to be hubs and are enriched within modules. Although there is little synergistic pairing of lncRNAs across cancers, lncRNA pairs involved in the same cancer hallmarks by regulating same or different biological processes. Finally, we identify prognostic biomarkers within cancer lncRNA expression datasets using modules derived from LFSNs. In summary, this proof-of-principle study indicates synergistic lncRNA pairs can be identified through integrative analysis of genome-wide expression data sets and functional information.