The mutational landscape of lethal castration-resistant prostate cancer.

Characterization of the prostate cancer transcriptome and genome has identified chromosomal rearrangements and copy number gains and losses, including ETS gene family fusions, PTEN loss and androgen receptor (AR) amplification, which drive prostate cancer development and progression to lethal, metastatic castration-resistant prostate cancer (CRPC). However, less is known about the role of mutations. Here we sequenced the exomes of 50 lethal, heavily pre-treated metastatic CRPCs obtained at rapid autopsy (including three different foci from the same patient) and 11 treatment-naive, high-grade localized prostate cancers. We identified low overall mutation rates even in heavily treated CRPCs (2.00 per megabase) and confirmed the monoclonal origin of lethal CRPC. Integrating exome copy number analysis identified disruptions of CHD1 that define a subtype of ETS gene family fusion-negative prostate cancer. Similarly, we demonstrate that ETS2, which is deleted in approximately one-third of CRPCs (commonly through TMPRSS2:ERG fusions), is also deregulated through mutation. Furthermore, we identified recurrent mutations in multiple chromatin- and histone-modifying genes, including MLL2 (mutated in 8.6% of prostate cancers), and demonstrate interaction of the MLL complex with the AR, which is required for AR-mediated signalling. We also identified novel recurrent mutations in the AR collaborating factor FOXA1, which is mutated in 5 of 147 (3.4%) prostate cancers (both untreated localized prostate cancer and CRPC), and showed that mutated FOXA1 represses androgen signalling and increases tumour growth. Proteins that physically interact with the AR, such as the ERG gene fusion product, FOXA1, MLL2, UTX (also known as KDM6A) and ASXL1 were found to be mutated in CRPC. In summary, we describe the mutational landscape of a heavily treated metastatic cancer, identify novel mechanisms of AR signalling deregulated in prostate cancer, and prioritize candidates for future study.

Transcriptome sequencing to detect gene fusions in cancer.

Recurrent gene fusions, typically associated with haematological malignancies and rare bone and soft-tissue tumours, have recently been described in common solid tumours. Here we use an integrative analysis of high-throughput long- and short-read transcriptome sequencing of cancer cells to discover novel gene fusions. As a proof of concept, we successfully used integrative transcriptome sequencing to 're-discover' the BCR-ABL1 (ref. 10) gene fusion in a chronic myelogenous leukaemia cell line and the TMPRSS2-ERG gene fusion in a prostate cancer cell line and tissues. Additionally, we nominated, and experimentally validated, novel gene fusions resulting in chimaeric transcripts in cancer cell lines and tumours. Taken together, this study establishes a robust pipeline for the discovery of novel gene chimaeras using high-throughput sequencing, opening up an important class of cancer-related mutations for comprehensive characterization.

Deep sequencing reveals distinct patterns of DNA methylation in prostate cancer.

Beginning with precursor lesions, aberrant DNA methylation marks the entire spectrum of prostate cancer progression. We mapped the global DNA methylation patterns in select prostate tissues and cell lines using MethylPlex-next-generation sequencing (M-NGS). Hidden Markov model-based next-generation sequence analysis identified ∼68,000 methylated regions per sample. While global CpG island (CGI) methylation was not differential between benign adjacent and cancer samples, overall promoter CGI methylation significantly increased from ~12.6% in benign samples to 19.3% and 21.8% in localized and metastatic cancer tissues, respectively (P-value < 2 × 10(-16)). We found distinct patterns of promoter methylation around transcription start sites, where methylation occurred not only on the CGIs, but also on flanking regions and CGI sparse promoters. Among the 6691 methylated promoters in prostate tissues, 2481 differentially methylated regions (DMRs) are cancer-specific, including numerous novel DMRs. A novel cancer-specific DMR in the WFDC2 promoter showed frequent methylation in cancer (17/22 tissues, 6/6 cell lines), but not in the benign tissues (0/10) and normal PrEC cells. Integration of LNCaP DNA methylation and H3K4me3 data suggested an epigenetic mechanism for alternate transcription start site utilization, and these modifications segregated into distinct regions when present on the same promoter. Finally, we observed differences in repeat element methylation, particularly LINE-1, between ERG gene fusion-positive and -negative cancers, and we confirmed this observation using pyrosequencing on a tissue panel. This comprehensive methylome map will further our understanding of epigenetic regulation in prostate cancer progression.

DNA aptamer-conjugated lipid nanoparticle for targeted PTEN mRNA delivery to prostate cancer cells.

The use of messenger RNA (mRNA) as a cancer vaccine and gene therapy requires targeted vehicle delivery to the site of disease. Here, we designed a mRNA-encapsulating lipid nanoparticle (LNP) conjugated with anti-programmed death-ligand 1 (PD-L1) DNA aptamer that delivers mRNA encoding a tumor suppressor gene, namely phosphatase and tensin homolog (PTEN), to castration-resistant prostate cancer (CRPC) cells expressing PD-L1 on the cell surface. The DNA aptamer-conjugated LNP-based mRNA delivery system (Apt-LNP[PTEN mRNA]) mediated efficient mRNA delivery and transfection in CRPC cells than LNPs without targeting ligands. Cancer-targeted PTEN mRNA delivery using Apt-LNPs achieved significantly higher PTEN expression via aptamer-mediated endocytosis in target cancer cells compared with non-targeted LNP delivery, resulting in significant downregulation of AKT phosphorylation. This enhanced PI3K/AKT pathway regulation, and in turn reduced cell migration after two days along with a 70 % decrease in cell viability, leading to effective apoptotic cell death. In a CRPC xenograft model, Apt-LNP[PTEN mRNA] led to an approximate 60 % reduction in tumor growth, which was attributable to the effective PTEN restoration and PI3K/AKT signaling pathway regulation. PTEN expression was significantly enhanced in CRPC tumor tissues, which abolished cancer cell tumorigenicity. These findings demonstrated the potential of Apt-LNPs for targeted mRNA delivery to cancer cells, thus providing a promising tool for targeted mRNA delivery to a range of cancers and tissues using a conventional LNP systems.

Protein interaction network underpins concordant prognosis among heterogeneous breast cancer signatures.

Characterizing the biomolecular systems' properties underpinning prognosis signatures derived from gene expression profiles remains a key clinical and biological challenge. In breast cancer, while different "poor-prognosis" sets of genes have predicted patient survival outcome equally well in independent cohorts, these prognostic signatures have surprisingly little genetic overlap. We examine 10 such published expression-based signatures that are predictors or distinct breast cancer phenotypes, uncover their mechanistic interconnectivity through a protein-protein interaction network, and introduce a novel cross-"gene expression signature" analysis method using (i) domain knowledge to constrain multiple comparisons in a mechanistically relevant single-gene network interactions and (ii) scale-free permutation re-sampling to statistically control for hubness (SPAN - Single Protein Analysis of Network with constant node degree per protein). At adjusted p-values<5%, 54-genes thus identified have a significantly greater connectivity than those through meticulous permutation re-sampling of the context-constrained network. More importantly, eight of 10 genetically non-overlapping signatures are connected through well-established mechanisms of breast cancer oncogenesis and progression. Gene Ontology enrichment studies demonstrate common markers of cell cycle regulation. Kaplan-Meier analysis of three independent historical gene expression sets confirms this network-signature's inherent ability to identify "poor outcome" in ER(+) patients without the requirement of machine learning. We provide a novel demonstration that genetically distinct prognosis signatures, developed from independent clinical datasets, occupy overlapping prognostic space of breast cancer via shared mechanisms that are mediated by genetically different yet mechanistically comparable interactions among proteins of differentially expressed genes in the signatures. This is the first study employing a networks' approach to aggregate established gene expression signatures in order to develop a phenotype/pathway-based cancer roadmap with the potential for (i) novel drug development applications and for (ii) facilitating the clinical deployment of prognostic gene signatures with improved mechanistic understanding of biological processes and functions associated with gene expression changes. http://www.lussierlab.org/publication/networksignature/.

Robust methods for accurate diagnosis using pan-microbiological oligonucleotide microarrays.

BACKGROUND: To address the limitations of traditional virus and pathogen detection methodologies in clinical diagnosis, scientists have developed high-throughput oligonucleotide microarrays to rapidly identify infectious agents. However, objectively identifying pathogens from the complex hybridization patterns of these massively multiplexed arrays remains challenging. METHODS: In this study, we conceived an automated method based on the hypergeometric distribution for identifying pathogens in multiplexed arrays and compared it to five other methods. We evaluated these metrics: 1) accurate prediction, whether the top ranked prediction(s) match the real virus(es); 2) four accuracy scores. RESULTS: Though accurate prediction and high specificity and sensitivity can be achieved with several methods, the method based on hypergeometric distribution provides a significant advantage in term of positive predicting value with two to sixty folds the positive predicting values of other methods. CONCLUSION: The proposed multi-specie array analysis based on the hypergeometric distribution addresses shortcomings of previous methods by enhancing signals of positively hybridized probes.

PhenoGO: an integrated resource for the multiscale mining of clinical and biological data.

The evolving complexity of genome-scale experiments has increasingly centralized the role of a highly computable, accurate, and comprehensive resource spanning multiple biological scales and viewpoints. To provide a resource to meet this need, we have significantly extended the PhenoGO database with gene-disease specific annotations and included an additional ten species. This a computationally-derived resource is primarily intended to provide phenotypic context (cell type, tissue, organ, and disease) for mining existing associations between gene products and GO terms specified in the Gene Ontology Databases Automated natural language processing (BioMedLEE) and computational ontology (PhenOS) methods were used to derive these relationships from the literature, expanding the database with information from ten additional species to include over 600,000 phenotypic contexts spanning eleven species from five GO annotation databases. A comprehensive evaluation evaluating the mappings (n = 300) found precision (positive predictive value) at 85%, and recall (sensitivity) at 76%. Phenotypes are encoded in general purpose ontologies such as Cell Ontology, the Unified Medical Language System, and in specialized ontologies such as the Mouse Anatomy and the Mammalian Phenotype Ontology. A web portal has also been developed, allowing for advanced filtering and querying of the database as well as download of the entire dataset http://www.phenogo.org.

Genomic assessment of a multikinase inhibitor, sorafenib, in a rodent model of pulmonary hypertension.

Pulmonary hypertension (PH) and cancer pathology share growth factor- and MAPK stress-mediated signaling pathways resulting in endothelial and smooth muscle cell dysfunction and angioproliferative vasculopathy. In this study, we assessed sorafenib, an antineoplastic agent and inhibitor of multiple kinases important in angiogenesis [VEGF receptor (VEGFR)-1-3, PDGF receptor (PDGFR)-beta, Raf-1 kinase] as a potential PH therapy. Two PH rat models were used: a conventional hypoxia-induced PH model and an augmented PH model combining dual VEGFR-1 and -2 inhibition (SU-5416, single 20 mg/kg injection) with hypoxia. In addition to normoxia-exposed control animals, four groups were maintained at 10% inspired O(2) fraction for 3.5 wk (hypoxia/vehicle, hypoxia/SU-5416, hypoxia/sorafenib, and hypoxia/SU-5416/sorafenib). Compared with normoxic control animals, rats exposed to hypoxia/SU-5416 developed hemodynamic and histological evidence of severe PH while rats exposed to hypoxia alone displayed only mild elevations in hemodynamic values (pulmonary vascular and right ventricular pressures). Sorafenib treatment (daily gavage, 2.5 mg/kg) prevented hemodynamic changes and demonstrated dramatic attenuation of PH-associated vascular remodeling. Compared with normoxic control rats, expression profiling (Affymetrix platform) of lung RNA obtained from hypoxia [false discovery rate (FDR) 6.5%]- and hypoxia/SU-5416 (FDR 1.6%)-challenged rats yielded 1,019 and 465 differentially regulated genes (fold change >1.4), respectively. A novel molecular signature consisting of 38 differentially expressed genes between hypoxia/SU-5416 and hypoxia/SU-5416/sorafenib (FDR 6.7%) was validated by either real-time RT-PCR or immunoblotting. Finally, immunoblotting studies confirmed the upregulation of the MAPK cascade in both PH models, which was abolished by sorafenib. In summary, sorafenib represents a novel potential treatment for severe PH with the MAPK cascade a potential canonical target.

Information theory applied to the sparse gene ontology annotation network to predict novel gene function.

MOTIVATION: Despite advances in the gene annotation process, the functions of a large portion of gene products remain insufficiently characterized. In addition, the in silico prediction of novel Gene Ontology (GO) annotations for partially characterized gene functions or processes is highly dependent on reverse genetic or functional genomic approaches. To our knowledge, no prediction method has been demonstrated to be highly accurate for sparsely annotated GO terms (those associated to fewer than 10 genes). RESULTS: We propose a novel approach, information theory-based semantic similarity (ITSS), to automatically predict molecular functions of genes based on existing GO annotations. Using a 10-fold cross-validation, we demonstrate that the ITSS algorithm obtains prediction accuracies (precision 97%, recall 77%) comparable to other machine learning algorithms when compared in similar conditions over densely annotated portions of the GO datasets. This method is able to generate highly accurate predictions in sparsely annotated portions of GO, where previous algorithms have failed. As a result, our technique generates an order of magnitude more functional predictions than previous methods. A 10-fold cross validation demonstrated a precision of 90% at a recall of 36% for the algorithm over sparsely annotated networks of the recent GO annotations (about 1400 GO terms and 11,000 genes in Homo sapiens). To our knowledge, this article presents the first historical rollback validation for the predicted GO annotations, which may represent more realistic conditions than more widely used cross-validation approaches. By manually assessing a random sample of 100 predictions conducted in a historical rollback evaluation, we estimate that a minimum precision of 51% (95% confidence interval: 43-58%) can be achieved for the human GO Annotation file dated 2003. AVAILABILITY: The program is available on request. The 97,732 positive predictions of novel gene annotations from the 2005 GO Annotation dataset and other supplementary information is available at http://phenos.bsd.uchicago.edu/ITSS/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Evaluation of high-throughput functional categorization of human disease genes.

BACKGROUND: Biological data that are well-organized by an ontology, such as Gene Ontology, enables high-throughput availability of the semantic web. It can also be used to facilitate high throughput classification of biomedical information. However, to our knowledge, no evaluation has been published on automating classifications of human diseases genes using Gene Ontology. In this study, we evaluate automated classifications of well-defined human disease genes using their Gene Ontology annotations and compared them to a gold standard. This gold standard was independently conceived by Valle's research group, and contains 923 human disease genes organized in 14 categories of protein function. RESULTS: Two automated methods were applied to investigate the classification of human disease genes into independently pre-defined categories of protein function. One method used the structure of Gene Ontology by pre-selecting 74 Gene Ontology terms assigned to 11 protein function categories. The second method was based on the similarity of human disease genes clustered according to the information-theoretic distance of their Gene Ontology annotations. Compared to the categorization of human disease genes found in the gold standard, our automated methods can achieve an overall 56% and 47% precision with 62% and 71% recall respectively. However, approximately 15% of the studied human disease genes remain without GO annotations. CONCLUSION: Automated methods can recapitulate a significant portion of classification of the human disease genes. The method using information-theoretic distance performs slightly better on the precision with some loss in recall. For some protein function categories, such as 'hormone' and 'transcription factor', the automated methods perform particularly well, achieving precision and recall levels above 75%. In summary, this study demonstrates that for semantic webs, methods to automatically classify or analyze a majority of human disease genes require significant progress in both the Gene Ontology annotations and particularly in the utilization of these annotations.

An integrative genomic approach to uncover molecular mechanisms of prokaryotic traits.

With mounting availability of genomic and phenotypic databases, data integration and mining become increasingly challenging. While efforts have been put forward to analyze prokaryotic phenotypes, current computational technologies either lack high throughput capacity for genomic scale analysis, or are limited in their capability to integrate and mine data across different scales of biology. Consequently, simultaneous analysis of associations among genomes, phenotypes, and gene functions is prohibited. Here, we developed a high throughput computational approach, and demonstrated for the first time the feasibility of integrating large quantities of prokaryotic phenotypes along with genomic datasets for mining across multiple scales of biology (protein domains, pathways, molecular functions, and cellular processes). Applying this method over 59 fully sequenced prokaryotic species, we identified genetic basis and molecular mechanisms underlying the phenotypes in bacteria. We identified 3,711 significant correlations between 1,499 distinct Pfam and 63 phenotypes, with 2,650 correlations and 1,061 anti-correlations. Manual evaluation of a random sample of these significant correlations showed a minimal precision of 30% (95% confidence interval: 20%-42%; n = 50). We stratified the most significant 478 predictions and subjected 100 to manual evaluation, of which 60 were corroborated in the literature. We furthermore unveiled 10 significant correlations between phenotypes and KEGG pathways, eight of which were corroborated in the evaluation, and 309 significant correlations between phenotypes and 166 GO concepts evaluated using a random sample (minimal precision = 72%; 95% confidence interval: 60%-80%; n = 50). Additionally, we conducted a novel large-scale phenomic visualization analysis to provide insight into the modular nature of common molecular mechanisms spanning multiple biological scales and reused by related phenotypes (metaphenotypes). We propose that this method elucidates which classes of molecular mechanisms are associated with phenotypes or metaphenotypes and holds promise in facilitating a computable systems biology approach to genomic and biomedical research.

A comparison of single molecule and amplification based sequencing of cancer transcriptomes.

The second wave of next generation sequencing technologies, referred to as single-molecule sequencing (SMS), carries the promise of profiling samples directly without employing polymerase chain reaction steps used by amplification-based sequencing (AS) methods. To examine the merits of both technologies, we examine mRNA sequencing results from single-molecule and amplification-based sequencing in a set of human cancer cell lines and tissues. We observe a characteristic coverage bias towards high abundance transcripts in amplification-based sequencing. A larger fraction of AS reads cover highly expressed genes, such as those associated with translational processes and housekeeping genes, resulting in relatively lower coverage of genes at low and mid-level abundance. In contrast, the coverage of high abundance transcripts plateaus off using SMS. Consequently, SMS is able to sequence lower- abundance transcripts more thoroughly, including some that are undetected by AS methods; however, these include many more mapping artifacts. A better understanding of the technical and analytical factors introducing platform specific biases in high throughput transcriptome sequencing applications will be critical in cross platform meta-analytic studies.

Personalized oncology through integrative high-throughput sequencing: a pilot study.

Individual cancers harbor a set of genetic aberrations that can be informative for identifying rational therapies currently available or in clinical trials. We implemented a pilot study to explore the practical challenges of applying high-throughput sequencing in clinical oncology. We enrolled patients with advanced or refractory cancer who were eligible for clinical trials. For each patient, we performed whole-genome sequencing of the tumor, targeted whole-exome sequencing of tumor and normal DNA, and transcriptome sequencing (RNA-Seq) of the tumor to identify potentially informative mutations in a clinically relevant time frame of 3 to 4 weeks. With this approach, we detected several classes of cancer mutations including structural rearrangements, copy number alterations, point mutations, and gene expression alterations. A multidisciplinary Sequencing Tumor Board (STB) deliberated on the clinical interpretation of the sequencing results obtained. We tested our sequencing strategy on human prostate cancer xenografts. Next, we enrolled two patients into the clinical protocol and were able to review the results at our STB within 24 days of biopsy. The first patient had metastatic colorectal cancer in which we identified somatic point mutations in NRAS, TP53, AURKA, FAS, and MYH11, plus amplification and overexpression of cyclin-dependent kinase 8 (CDK8). The second patient had malignant melanoma, in which we identified a somatic point mutation in HRAS and a structural rearrangement affecting CDKN2C. The STB identified the CDK8 amplification and Ras mutation as providing a rationale for clinical trials with CDK inhibitors or MEK (mitogen-activated or extracellular signal-regulated protein kinase kinase) and PI3K (phosphatidylinositol 3-kinase) inhibitors, respectively. Integrative high-throughput sequencing of patients with advanced cancer generates a comprehensive, individual mutational landscape to facilitate biomarker-driven clinical trials in oncology.

Modeling and characterization of disease associated subnetworks in the human interactome using machine learning.

The availability of large-scale, genome-wide data about the molecular interactome of entire organisms has made possible new types of integrative studies, making use of rapidly accumulating knowledge of gene-disease associations. Previous studies have established the presence of functional biomodules in the molecular interaction network of living organisms, a number of which have been associated with the pathogenesis and progression of human disease. While a number of studies have examined the networks and biomodules associated with disease, the properties that contribute to the particular susceptibility of these subnetworks to disruptions leading to disease phenotypes have not been extensively studied. We take a machine learning approach to the characterization of these disease subnetworks associated with complex and single-gene diseases, taking into account both the biological roles of their constituent genes and topological properties of the networks they form.

Robust informatics methods for accurate diagnosis using highly multiplexed virus arrays.

To address the limitations of traditional pathogen detection methodologies in clinical diagnosis, scientists have developed oligonucleotide microarrays. However, pathogen array interpretation remains challenging because of false positive probes. We conceived a novel computational method to interpret these arrays. In an evaluation using clinical samples, our method was compared to five others. Though most methods showed high specificity and sensitivity, only the proposed method showed significant advantages in positive predictive value (2-60 fold increase).

Information-theoretic classification of SNOMED improves the organization of context-sensitive excerpts from Cochrane reviews.

The emphasis on evidence based medicine (EBM) has placed increased focus on finding timely answers to clinical questions in presence of patients. Using a combination of natural language processing for the generation of clinical excerpts and information theoretic distance based clustering, we evaluated multiple approaches for the efficient presentation of context-sensitive EBM excerpts.

Discovery of protein interaction networks shared by diseases.

The study of protein-protein interactions is essential to define the molecular networks that contribute to maintain homeostasis of an organism's body functions. Disruptions in protein interaction networks have been shown to result in diseases in both humans and animals. Monogenic diseases disrupting biochemical pathways such as hereditary coagulopathies (e.g. hemophilia), provided a deep insight in the biochemical pathways of acquired coagulopathies of complex diseases. Indeed, a variety of complex liver diseases can lead to decreased synthesis of the same set of coagulation factors as in hemophilia. Similarly, more complex diseases such as different cancers have been shown to result from malfunctions of common proteins pathways. In order to discover, in high throughput, the molecular underpinnings of poorly characterized diseases, we present a statistical method to identify shared protein interaction network(s) between diseases. Integrating (i) a protein interaction network with (ii) disease to protein relationships derived from mining Gene Ontology annotations and the biomedical literature with natural language understanding (PhenoGO), we identified protein-protein interactions that were associated with pairs of diseases and calculated the statistical significance of the occurrence of interactions in the protein interaction knowledgebase. Significant correlations between diseases and shared protein networks were identified and evaluated in this study, demonstrating the high precision of the approach and correct non-trivial predictions, signifying the potential for discovery. In conclusion, we demonstrate that the associations between diseases are directly correlated to their underlying protein-protein interaction networks, possibly providing insight into the underlying molecular mechanisms of phenotypes and biological processes disrupted in related diseases.

Use of consomic rats for genomic insights into ventilator-associated lung injury.

Increasing evidence supports the contribution of genetic influences on susceptibility/severity in acute lung injury (ALI), a devastating syndrome requiring mechanical ventilation with subsequent risk for ventilator-associated lung injury (VALI). To identify VALI candidate genes, we determined that Brown Norway (BN) and Dahl salt-sensitive (SS) rat strains were differentially sensitive to VALI (tidal volume of 20 ml/kg, 85 breaths/min, 2 h) defined by bronchoalveolar lavage (BAL) protein and leukocytes. We next exploited differential sensitivities and phenotyped both the VALI-sensitive BN and the VALI-resistant SS rat strains by expression profiling coupled to a bioinformatic-intense candidate gene approach (Significance Analysis of Microarrays, i.e., SAM). We identified 106 differentially expressed VALI genes representing gene ontologies such as "transcription" and "chemotaxis/cell motility." We mapped the chromosomal location of the differentially expressed probe sets and selected consomic SS rats with single BN introgressions of chromosomes 2, 13, and 16 (based on the highest density of probe sets) while also choosing chromosome 20 (low probe sets density). VALI exposure of consomic rats with introgressions of BN chromosomes 13 and 16 resulted in significant increases in both BAL cells and protein (compared to parental SS strain), whereas introgression of BN chromosome 2 displayed a large increase only in BAL protein. Introgression of BN chromosome 20 had a minimal effect. These results suggest that genes residing on BN chromosomes 2, 13, and 16 confer increased sensitivity to high tidal volume ventilation. We speculate that the consomic-microarray-SAM approach is a time- and resource-efficient tool for the genetic dissection of complex diseases including VALI.