A validated regulatory network for Th17 cell specification.

Th17 cells have critical roles in mucosal defense and are major contributors to inflammatory disease. Their differentiation requires the nuclear hormone receptor RORγt working with multiple other essential transcription factors (TFs). We have used an iterative systems approach, combining genome-wide TF occupancy, expression profiling of TF mutants, and expression time series to delineate the Th17 global transcriptional regulatory network. We find that cooperatively bound BATF and IRF4 contribute to initial chromatin accessibility and, with STAT3, initiate a transcriptional program that is then globally tuned by the lineage-specifying TF RORγt, which plays a focal deterministic role at key loci. Integration of multiple data sets allowed inference of an accurate predictive model that we computationally and experimentally validated, identifying multiple new Th17 regulators, including Fosl2, a key determinant of cellular plasticity. This interconnected network can be used to investigate new therapeutic approaches to manipulate Th17 functions in the setting of inflammatory disease.

A genome-wide RNA interference screen identifies new regulators of androgen receptor function in prostate cancer cells.

The androgen receptor (AR) is a mediator of both androgen-dependent and castration-resistant prostate cancers. Identification of cellular factors affecting AR transcriptional activity could in principle yield new targets that reduce AR activity and combat prostate cancer, yet a comprehensive analysis of the genes required for AR-dependent transcriptional activity has not been determined. Using an unbiased genetic approach that takes advantage of the evolutionary conservation of AR signaling, we have conducted a genome-wide RNAi screen in Drosophila cells for genes required for AR transcriptional activity and applied the results to human prostate cancer cells. We identified 45 AR-regulators, which include known pathway components and genes with functions not previously linked to AR regulation, such as HIPK2 (a protein kinase) and MED19 (a subunit of the Mediator complex). Depletion of HIPK2 and MED19 in human prostate cancer cells decreased AR target gene expression and, importantly, reduced the proliferation of androgen-dependent and castration-resistant prostate cancer cells. We also systematically analyzed additional Mediator subunits and uncovered a small subset of Mediator subunits that interpret AR signaling and affect AR-dependent transcription and prostate cancer cell proliferation. Importantly, targeting of HIPK2 by an FDA-approved kinase inhibitor phenocopied the effect of depletion by RNAi and reduced the growth of AR-positive, but not AR-negative, treatment-resistant prostate cancer cells. Thus, our screen has yielded new AR regulators including drugable targets that reduce the proliferation of castration-resistant prostate cancer cells.

An experimentally supported model of the Bacillus subtilis global transcriptional regulatory network.

Organisms from all domains of life use gene regulation networks to control cell growth, identity, function, and responses to environmental challenges. Although accurate global regulatory models would provide critical evolutionary and functional insights, they remain incomplete, even for the best studied organisms. Efforts to build comprehensive networks are confounded by challenges including network scale, degree of connectivity, complexity of organism-environment interactions, and difficulty of estimating the activity of regulatory factors. Taking advantage of the large number of known regulatory interactions in Bacillus subtilis and two transcriptomics datasets (including one with 38 separate experiments collected specifically for this study), we use a new combination of network component analysis and model selection to simultaneously estimate transcription factor activities and learn a substantially expanded transcriptional regulatory network for this bacterium. In total, we predict 2,258 novel regulatory interactions and recall 74% of the previously known interactions. We obtained experimental support for 391 (out of 635 evaluated) novel regulatory edges (62% accuracy), thus significantly increasing our understanding of various cell processes, such as spore formation.

Robust data-driven incorporation of prior knowledge into the inference of dynamic regulatory networks.

MOTIVATION: Inferring global regulatory networks (GRNs) from genome-wide data is a computational challenge central to the field of systems biology. Although the primary data currently used to infer GRNs consist of gene expression and proteomics measurements, there is a growing abundance of alternate data types that can reveal regulatory interactions, e.g. ChIP-Chip, literature-derived interactions, protein-protein interactions. GRN inference requires the development of integrative methods capable of using these alternate data as priors on the GRN structure. Each source of structure priors has its unique biases and inherent potential errors; thus, GRN methods using these data must be robust to noisy inputs. RESULTS: We developed two methods for incorporating structure priors into GRN inference. Both methods [Modified Elastic Net (MEN) and Bayesian Best Subset Regression (BBSR)] extend the previously described Inferelator framework, enabling the use of prior information. We test our methods on one synthetic and two bacterial datasets, and show that both MEN and BBSR infer accurate GRNs even when the structure prior used has significant amounts of error (>90% erroneous interactions). We find that BBSR outperforms MEN at inferring GRNs from expression data and noisy structure priors. AVAILABILITY AND IMPLEMENTATION: Code, datasets and networks presented in this article are available at http://bonneaulab.bio.nyu.edu/software.html. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Differential Changes in ACPA Fine Specificity and Gene Expression in a Randomized Trial of Abatacept and Adalimumab in Rheumatoid Arthritis.

INTRODUCTION: The biologics abatacept and adalimumab have different mechanisms of action (MoAs). We analyzed data from patients with rheumatoid arthritis treated in AMPLE (NCT00929864) to explore the pharmacodynamic effects of abatacept or adalimumab on anti-citrullinated protein antibodies (ACPAs) and gene expression. METHODS: AMPLE was a phase IIIb, 2-year, randomized, head-to-head trial of abatacept versus adalimumab. Post hoc analyses of baseline anti-cyclic citrullinated peptide-2 (anti-CCP2, an ACPA surrogate) positive (+) status and ACPA fine-specificity profiles over time, as well as transcriptional profiling (peripheral whole blood), were performed. RESULTS: Of 646 patients treated (abatacept, n = 318; adalimumab, n = 328), ACPA and gene expression data were available from 508 and 566 patients, respectively. In anti-CCP2+ patients (n = 388), baseline fine specificities for most ACPAs were highly correlated; over 2 years, levels decreased with abatacept but not adalimumab. By year 2, expression of genes associated with T cell co-stimulation and antibody production was lower for abatacept versus adalimumab; expression of genes associated with proinflammatory signaling was lower for adalimumab versus abatacept. Treatment modulated the expression of T- and B-cell gene signatures, with differences in CD8+ T cells, activated T cells, plasma cells, B cells, natural killer cells (all lower with abatacept versus adalimumab), and polymorphonuclear leukocytes (higher with abatacept versus adalimumab). CONCLUSIONS: In AMPLE, despite similar clinical outcomes, data showed that pharmacodynamic/genetic changes after 2 years of abatacept or adalimumab were consistent with drug MoAs. Further assessment of the relationship between such changes and clinical outcomes, including prediction of response, is warranted. TRIAL REGISTRATION: ClinicalTrials.gov identifier, NCT00929864.

Bayesian network analysis of panomic biological big data identifies the importance of triglyceride-rich LDL in atherosclerosis development.

Introduction: We sought to explore biomarkers of coronary atherosclerosis in an unbiased fashion. Methods: We analyzed 665 patients (mean ± SD age, 56 ± 11 years; 47% male) from the GLOBAL clinical study (NCT01738828). Cases were defined by the presence of any discernable atherosclerotic plaque based on comprehensive cardiac computed tomography (CT). De novo Bayesian networks built out of 37,000 molecular measurements and 99 conventional biomarkers per patient examined the potential causality of specific biomarkers. Results: Most highly ranked biomarkers by gradient boosting were interleukin-6, symmetric dimethylarginine, LDL-triglycerides [LDL-TG], apolipoprotein B48, palmitoleic acid, small dense LDL, alkaline phosphatase, and asymmetric dimethylarginine. In Bayesian analysis, LDL-TG was directly linked to atherosclerosis in over 95% of the ensembles. Genetic variants in the genomic region encoding hepatic lipase (LIPC) were associated with LIPC gene expression, LDL-TG levels and with atherosclerosis. Discussion: Triglyceride-rich LDL particles, which can now be routinely measured with a direct homogenous assay, may play an important role in atherosclerosis development. Clinical trial registration: GLOBAL clinical study (Genetic Loci and the Burden of Atherosclerotic Lesions); [https://clinicaltrials.gov/ct2/show/NCT01738828?term=NCT01738828&rank=1], identifier [NCT01738828].

SOMAscan Proteomics Identifies Serum Biomarkers Associated With Liver Fibrosis in Patients With NASH.

Nonalcoholic steatohepatitis (NASH) is a major cause of liver-related morbidity and mortality worldwide. Liver fibrosis stage, a key component of NASH, has been linked to the risk of mortality and liver-related clinical outcomes. Currently there are no validated noninvasive diagnostics that can differentiate between fibrosis stages in patients with NASH; many existing tests do not reflect underlying disease pathophysiology. Noninvasive biomarkers are needed to identify patients at high-risk of NASH with advanced fibrosis. This was a retrospective study of patients with histologically proven NASH with fibrosis stages 0-4. The SOMAscan proteomics platform was used to quantify 1,305 serum proteins in a discovery cohort (n = 113). In patients with advanced (stages 3-4) versus early fibrosis (stages 0-2), 97 proteins with diverse biological functions were differentially expressed. Next, fibrosis-stage classification models were explored using a machine learning-based approach to prioritize the biomarkers for further evaluation. A four-protein model differentiated patients with stage 0-1 versus stage 2-4 fibrosis (area under the receiver operating characteristic curve [AUROC] = 0.74), while a 12-protein classifier differentiated advanced versus early fibrosis (AUROC = 0.83). Subsequently, the model's performance was validated in two independent cohorts (n = 71 and n = 32) with similar results (AUROC = 0.74-0.78). Our advanced fibrosis model performed similarly to or better than Fibrosis-4 index, aspartate aminotransferase-to-platelet ratio index, and nonalcoholic fatty liver disease (NAFLD) fibrosis score-based models for all three cohorts. Conclusion: A SOMAscan proteomics-based exploratory classifier for advanced fibrosis, consisting of biomarkers that reflect the complexity of NASH pathophysiology, demonstrated similar performance in independent validation cohorts and performed similarly or better than Fibrosis-4 index, aspartate aminotransferase-to-platelet ratio index, and NAFLD fibrosis score. Further studies are warranted to evaluate the clinical utility of these biomarker panels in patients with NAFLD.

A physical sciences network characterization of non-tumorigenic and metastatic cells.

To investigate the transition from non-cancerous to metastatic from a physical sciences perspective, the Physical Sciences-Oncology Centers (PS-OC) Network performed molecular and biophysical comparative studies of the non-tumorigenic MCF-10A and metastatic MDA-MB-231 breast epithelial cell lines, commonly used as models of cancer metastasis. Experiments were performed in 20 laboratories from 12 PS-OCs. Each laboratory was supplied with identical aliquots and common reagents and culture protocols. Analyses of these measurements revealed dramatic differences in their mechanics, migration, adhesion, oxygen response, and proteomic profiles. Model-based multi-omics approaches identified key differences between these cells' regulatory networks involved in morphology and survival. These results provide a multifaceted description of cellular parameters of two widely used cell lines and demonstrate the value of the PS-OC Network approach for integration of diverse experimental observations to elucidate the phenotypes associated with cancer metastasis.

Integrated inference and analysis of regulatory networks from multi-level measurements.

Regulatory and signaling networks coordinate the enormously complex interactions and processes that control cellular processes (such as metabolism and cell division), coordinate response to the environment, and carry out multiple cell decisions (such as development and quorum sensing). Regulatory network inference is the process of inferring these networks, traditionally from microarray data but increasingly incorporating other measurement types such as proteomics, ChIP-seq, metabolomics, and mass cytometry. We discuss existing techniques for network inference. We review in detail our pipeline, which consists of an initial biclustering step, designed to estimate co-regulated groups; a network inference step, designed to select and parameterize likely regulatory models for the control of the co-regulated groups from the biclustering step; and a visualization and analysis step, designed to find and communicate key features of the network. Learning biological networks from even the most complete data sets is challenging; we argue that integrating new data types into the inference pipeline produces networks of increased accuracy, validity, and biological relevance.

Androgen receptor antagonism by divalent ethisterone conjugates in castrate-resistant prostate cancer cells.

Sustained treatment of prostate cancer with androgen receptor (AR) antagonists can evoke drug resistance, leading to castrate-resistant disease. Elevated activity of the AR is often associated with this highly aggressive disease state. Therefore, new therapeutic regimens that target and modulate AR activity could prove beneficial. We previously introduced a versatile chemical platform to generate competitive and non-competitive multivalent peptoid oligomer conjugates that modulate AR activity. In particular, we identified a linear and a cyclic divalent ethisterone conjugate that exhibit potent anti-proliferative properties in LNCaP-abl cells, a model of castrate-resistant prostate cancer. Here, we characterize the mechanism of action of these compounds utilizing confocal microscopy, time-resolved fluorescence resonance energy transfer, chromatin immunoprecipitation, flow cytometry, and microarray analysis. The linear conjugate competitively blocks AR action by inhibiting DNA binding. In addition, the linear conjugate does not promote AR nuclear localization or co-activator binding. In contrast, the cyclic conjugate promotes AR nuclear localization and induces cell-cycle arrest, despite its inability to compete against endogenous ligand for binding to AR in vitro. Genome-wide expression analysis reveals that gene transcripts are differentially affected by treatment with the linear or cyclic conjugate. Although the divalent ethisterone conjugates share extensive chemical similarities, we illustrate that they can antagonize the AR via distinct mechanisms of action, establishing new therapeutic strategies for potential applications in AR pharmacology.

DREAM4: Combining genetic and dynamic information to identify biological networks and dynamical models.

BACKGROUND: Current technologies have lead to the availability of multiple genomic data types in sufficient quantity and quality to serve as a basis for automatic global network inference. Accordingly, there are currently a large variety of network inference methods that learn regulatory networks to varying degrees of detail. These methods have different strengths and weaknesses and thus can be complementary. However, combining different methods in a mutually reinforcing manner remains a challenge. METHODOLOGY: We investigate how three scalable methods can be combined into a useful network inference pipeline. The first is a novel t-test-based method that relies on a comprehensive steady-state knock-out dataset to rank regulatory interactions. The remaining two are previously published mutual information and ordinary differential equation based methods (tlCLR and Inferelator 1.0, respectively) that use both time-series and steady-state data to rank regulatory interactions; the latter has the added advantage of also inferring dynamic models of gene regulation which can be used to predict the system's response to new perturbations. CONCLUSION/SIGNIFICANCE: Our t-test based method proved powerful at ranking regulatory interactions, tying for first out of methods in the DREAM4 100-gene in-silico network inference challenge. We demonstrate complementarity between this method and the two methods that take advantage of time-series data by combining the three into a pipeline whose ability to rank regulatory interactions is markedly improved compared to either method alone. Moreover, the pipeline is able to accurately predict the response of the system to new conditions (in this case new double knock-out genetic perturbations). Our evaluation of the performance of multiple methods for network inference suggests avenues for future methods development and provides simple considerations for genomic experimental design. Our code is publicly available at http://err.bio.nyu.edu/inferelator/.

DREAM3: network inference using dynamic context likelihood of relatedness and the inferelator.

BACKGROUND: Many current works aiming to learn regulatory networks from systems biology data must balance model complexity with respect to data availability and quality. Methods that learn regulatory associations based on unit-less metrics, such as Mutual Information, are attractive in that they scale well and reduce the number of free parameters (model complexity) per interaction to a minimum. In contrast, methods for learning regulatory networks based on explicit dynamical models are more complex and scale less gracefully, but are attractive as they may allow direct prediction of transcriptional dynamics and resolve the directionality of many regulatory interactions. METHODOLOGY: We aim to investigate whether scalable information based methods (like the Context Likelihood of Relatedness method) and more explicit dynamical models (like Inferelator 1.0) prove synergistic when combined. We test a pipeline where a novel modification of the Context Likelihood of Relatedness (mixed-CLR, modified to use time series data) is first used to define likely regulatory interactions and then Inferelator 1.0 is used for final model selection and to build an explicit dynamical model. CONCLUSIONS/SIGNIFICANCE: Our method ranked 2nd out of 22 in the DREAM3 100-gene in silico networks challenge. Mixed-CLR and Inferelator 1.0 are complementary, demonstrating a large performance gain relative to any single tested method, with precision being especially high at low recall values. Partitioning the provided data set into four groups (knock-down, knock-out, time-series, and combined) revealed that using comprehensive knock-out data alone provides optimal performance. Inferelator 1.0 proved particularly powerful at resolving the directionality of regulatory interactions, i.e. "who regulates who" (approximately of identified true positives were correctly resolved). Performance drops for high in-degree genes, i.e. as the number of regulators per target gene increases, but not with out-degree, i.e. performance is not affected by the presence of regulatory hubs.

The Inferelator 2.0: a scalable framework for reconstruction of dynamic regulatory network models.

Current methods for reconstructing biological networks often learn either the topology of large networks or the kinetic parameters of smaller networks with a well-characterized topology. We have recently described a network reconstruction algorithm, the Inferelator 1.0, that given a set of genome-wide measurements as input, simultaneously learns both topology and kinetic-parameters. Specifically, it learns a system of ordinary differential equations (ODEs) that describe the rate of change in transcription of each gene or gene-cluster, as a function of environmental and transcription factors. In order to scale to large networks, in Inferelator 1.0 we have approximated the system of ODEs to be uncoupled, and have solved each ODE using a one-step finite difference approximation. Naturally, these approximations become crude as the simulated time-interval increases. Here we present, implement, and test a new Markov-Chain-Monte-Carlo (MCMC) dynamical modeling method, Inferelator 2.0, that works in tandem with Inferelator 1.0 and is designed to relax these approximations. We show results for the prokaryote Halobacterium that demonstrate a marked improvement in our predictive performance in modeling the regulatory dynamics of the system over longer time-scales.