Ilias Tagkopoulos

Predictive Behavior Within Microbial Genetic Networks

The homeostatic framework has dominated our understanding of cellular physiology. We question whether homeostasis alone adequately explains microbial responses to environmental stimuli, and explore the capacity of intracellular networks for predictive behavior in a fashion similar to metazoan nervous systems. We show that in silico biochemical networks, evolving randomly under precisely defined complex habitats, capture the dynamical, multidimensional structure of diverse environments by forming internal representations that allow prediction of environmental change. We provide evidence for such anticipatory behavior by revealing striking correlations of Escherichia coli transcriptional responses to temperature and oxygen perturbations—precisely mirroring the covariation of these parameters upon transitions between the outside world and the mammalian gastrointestinal tract. We further show that these internal correlations reflect a true associative learning paradigm, because they show rapid decoupling upon exposure to novel environments.

An Arabidopsis gene regulatory network for secondary cell wall synthesis

The plant cell wall is an important factor for determining cell shape, function and response to the environment. Secondary cell walls, such as those found in xylem, are composed of cellulose, hemicelluloses and lignin and account for the bulk of plant biomass. The coordination between transcriptional regulation of synthesis for each polymer is complex and vital to cell function. A regulatory hierarchy of developmental switches has been proposed, although the full complement of regulators remains unknown. Here we present a protein–DNA network between Arabidopsis thaliana transcription factors and secondary cell wall metabolic genes with gene expression regulated by a series of feed-forward loops. This model allowed us to develop and validate new hypotheses about secondary wall gene regulation under abiotic stress. Distinct stresses are able to perturb targeted genes to potentially promote functional adaptation. These interactions will serve as a foundation for understanding the regulation of a complex, integral plant component.

Evolutionary potential, cross‐stress behavior and the genetic basis of acquired stress resistance in Escherichia coli

Bacterial populations have a remarkable capacity to cope with extreme environmental fluctuations in their natural environments. In certain cases, adaptation to one stressful environment provides a fitness advantage when cells are exposed to a second stressor, a phenomenon that has been coined as cross‐stress protection. A tantalizing question in bacterial physiology is how the cross‐stress behavior emerges during evolutionary adaptation and what the genetic basis of acquired stress resistance is. To address these questions, we evolved Escherichia coli cells over 500 generations in five environments that include four abiotic stressors. Through growth profiling and competition assays, we identified several cases of positive and negative cross‐stress behavior that span all strain–stress combinations. Resequencing the genomes of the evolved strains resulted in the identification of several mutations and gene amplifications, whose fitness effect was further assessed by mutation reversal and competition assays. Transcriptional profiling of all strains under a specific stress, NaCl‐induced osmotic stress, and integration with resequencing data further elucidated the regulatory responses and genes that are involved in this phenomenon. Our results suggest that cross‐stress dependencies are ubiquitous, highly interconnected, and can emerge within short timeframes. The high adaptive potential that we observed argues that bacterial populations occupy a genotypic space that enables a high phenotypic plasticity during adaptation in fluctuating environments.

An integrative, multi‐scale, genome‐wide model reveals the phenotypic landscape of Escherichia coli

Given the vast behavioral repertoire and biological complexity of even the simplest organisms, accurately predicting phenotypes in novel environments and unveiling their biological organization is a challenging endeavor. Here, we present an integrative modeling methodology that unifies under a common framework the various biological processes and their interactions across multiple layers. We trained this methodology on an extensive normalized compendium for the gram‐negative bacterium Escherichia coli, which incorporates gene expression data for genetic and environmental perturbations, transcriptional regulation, signal transduction, and metabolic pathways, as well as growth measurements. Comparison with measured growth and high‐throughput data demonstrates the enhanced ability of the integrative model to predict phenotypic outcomes in various environmental and genetic conditions, even in cases where their underlying functions are under‐represented in the training set. This work paves the way toward integrative techniques that extract knowledge from a variety of biological data to achieve more than the sum of their parts in the context of prediction, analysis, and redesign of biological systems.

Transcriptional Network Analysis Identifies BACH1 as A Master Regulator of Breast Cancer Bone Metastasis

Background: The transcriptional network governing cancer metastasis is largely unexplored. Results: BACH1 regulates multiple metastasis genes and promotes breast cancer metastasis to bone. Conclusion: BACH1 is a master regulator of breast cancer bone metastasis and transcriptional network reverse engineering is helpful to identify novel functional genes of metastasis. Significance: This study provides a systems biology approach to identify master regulators of complicated biological processes. The application of functional genomic analysis of breast cancer metastasis has led to the identification of a growing number of organ-specific metastasis genes, which often function in concert to facilitate different steps of the metastatic cascade. However, the gene regulatory network that controls the expression of these metastasis genes remains largely unknown. Here, we demonstrate a computational approach for the deconvolution of transcriptional networks to discover master regulators of breast cancer bone metastasis. Several known regulators of breast cancer bone metastasis such as Smad4 and HIF1 were identified in our analysis. Experimental validation of the networks revealed BACH1, a basic leucine zipper transcription factor, as the common regulator of several functional metastasis genes, including MMP1 and CXCR4. Ectopic expression of BACH1 enhanced the malignance of breast cancer cells, and conversely, BACH1 knockdown significantly reduced bone metastasis. The expression of BACH1 and its target genes was linked to the higher risk of breast cancer recurrence in patients. This study established BACH1 as the master regulator of breast cancer bone metastasis and provided a paradigm to identify molecular determinants in complex pathological processes.

SBROME: a scalable optimization and module matching framework for automated biosystems design.

The development of a scalable framework for biodesign automation is a formidable challenge given the expected increase in part availability and the ever-growing complexity of synthetic circuits. To allow for (a) the use of previously constructed and characterized circuits or modules and (b) the implementation of designs that can scale up to hundreds of nodes, we here propose a divide-and-conquer Synthetic Biology Reusable Optimization Methodology (SBROME). An abstract user-defined circuit is first transformed and matched against a module database that incorporates circuits that have previously been experimentally characterized. Then the resulting circuit is decomposed to subcircuits that are populated with the set of parts that best approximate the desired function. Finally, all subcircuits are subsequently characterized and deposited back to the module database for future reuse. We successfully applied SBROME toward two alternative designs of a modular 3-input multiplexer that utilize pre-existing logic gates and characterized biological parts.

Multi-omics integration accurately predicts cellular state in unexplored conditions for Escherichia coli.

A significant obstacle in training predictive cell models is the lack of integrated data sources. We develop semi-supervised normalization pipelines and perform experimental characterization (growth, transcriptional, proteome) to create Ecomics, a consistent, quality-controlled multi-omics compendium for Escherichia coli with cohesive meta-data information. We then use this resource to train a multi-scale model that integrates four omics layers to predict genome-wide concentrations and growth dynamics. The genetic and environmental ontology reconstructed from the omics data is substantially different and complementary to the genetic and chemical ontologies. The integration of different layers confers an incremental increase in the prediction performance, as does the information about the known gene regulatory and protein-protein interactions. The predictive performance of the model ranges from 0.54 to 0.87 for the various omics layers, which far exceeds various baselines. This work provides an integrative framework of omics-driven predictive modelling that is broadly applicable to guide biological discovery.

Automatic design of synthetic gene circuits through mixed integer non-linear programming.

Automatic design of synthetic gene circuits poses a significant challenge to synthetic biology, primarily due to the complexity of biological systems, and the lack of rigorous optimization methods that can cope with the combinatorial explosion as the number of biological parts increases. Current optimization methods for synthetic gene design rely on heuristic algorithms that are usually not deterministic, deliver sub-optimal solutions, and provide no guaranties on convergence or error bounds. Here, we introduce an optimization framework for the problem of part selection in synthetic gene circuits that is based on mixed integer non-linear programming (MINLP), which is a deterministic method that finds the globally optimal solution and guarantees convergence in finite time. Given a synthetic gene circuit, a library of characterized parts, and user-defined constraints, our method can find the optimal selection of parts that satisfy the constraints and best approximates the objective function given by the user. We evaluated the proposed method in the design of three synthetic circuits (a toggle switch, a transcriptional cascade, and a band detector), with both experimentally constructed and synthetic promoter libraries. Scalability and robustness analysis shows that the proposed framework scales well with the library size and the solution space. The work described here is a step towards a unifying, realistic framework for the automated design of biological circuits.

A synthetic biology approach to self-regulatory recombinant protein production in Escherichia coli

BackgroundRecombinant protein production is a process of great industrial interest, with products that range from pharmaceuticals to biofuels. Since high level production of recombinant protein imposes significant stress in the host organism, several methods have been developed over the years to optimize protein production. So far, these trial-and-error techniques have proved laborious and sensitive to process parameters, while there has been no attempt to address the problem by applying Synthetic Biology principles and methods, such as integration of standardized parts in novel synthetic circuits.ResultsWe present a novel self-regulatory protein production system that couples the control of recombinant protein production with a stress-induced, negative feedback mechanism. The synthetic circuit allows the down-regulation of recombinant protein expression through a stress-induced promoter. We used E. coli as the host organism, since it is widely used in recombinant processes. Our results show that the introduction of the self-regulatory circuit increases the soluble/insoluble ratio of recombinant protein at the expense of total protein yield. To further elucidate the dynamics of the system, we developed a computational model that is in agreement with the observed experimental data, and provides insight on the interplay between protein solubility and yield.ConclusionOur work introduces the idea of a self-regulatory circuit for recombinant protein products, and paves the way for processes with reduced external control or monitoring needs. It demonstrates that the library of standard biological parts serves as a valuable resource for initial synthetic blocks that needs to be further refined to be successfully applied in practical problems of biotechnological significance. Finally, the development of a predictive model in conjunction with experimental validation facilitates a better understanding of the underlying dynamics and can be used as a guide to experimental design.

From Data to Optimal Decision Making: A Data-Driven, Probabilistic Machine Learning Approach to Decision Support for Patients With Sepsis

Background A tantalizing question in medical informatics is how to construct knowledge from heterogeneous datasets, and as an extension, inform clinical decisions. The emergence of large-scale data integration in electronic health records (EHR) presents tremendous opportunities. However, our ability to efficiently extract informed decision support is limited due to the complexity of the clinical states and decision process, missing data and lack of analytical tools to advice based on statistical relationships. Objective Development and assessment of a data-driven method that infers the probability distribution of the current state of patients with sepsis, likely trajectories, optimal actions related to antibiotic administration, prediction of mortality and length-of-stay. Methods We present a data-driven, probabilistic framework for clinical decision support in sepsis-related cases. We first define states, actions, observations and rewards based on clinical practice, expert knowledge and data representations in an EHR dataset of 1492 patients. We then use Partially Observable Markov Decision Process (POMDP) model to derive the optimal policy based on individual patient trajectories and we evaluate the performance of the model-derived policies in a separate test set. Policy decisions were focused on the type of antibiotic combinations to administer. Multi-class and discriminative classifiers were used to predict mortality and length of stay. Results Data-derived antibiotic administration policies led to a favorable patient outcome in 49% of the cases, versus 37% when the alternative policies were followed (P=1.3e-13). Sensitivity analysis on the model parameters and missing data argue for a highly robust decision support tool that withstands parameter variation and data uncertainty. When the optimal policy was followed, 387 patients (25.9%) have 90% of their transitions to better states and 503 patients (33.7%) patients had 90% of their transitions to worse states (P=4.0e-06), while in the non-policy cases, these numbers are 192 (12.9%) and 764 (51.2%) patients (P=4.6e-117), respectively. Furthermore, the percentage of transitions within a trajectory that lead to a better or better/same state are significantly higher by following the policy than for non-policy cases (605 vs 344 patients, P=8.6e-25). Mortality was predicted with an AUC of 0.7 and 0.82 accuracy in the general case and similar performance was obtained for the inference of the length-of-stay (AUC of 0.69 to 0.73 with accuracies from 0.69 to 0.82). Conclusions A data-driven model was able to suggest favorable actions, predict mortality and length of stay with high accuracy. This work provides a solid basis for a scalable probabilistic clinical decision support framework for sepsis treatment that can be expanded to other clinically relevant states and actions, as well as a data-driven model that can be adopted in other clinical areas with sufficient training data.