Information-dependent enrichment analysis reveals time-dependent transcriptional regulation of the estrogen pathway of toxicity.

The twenty-first century vision for toxicology involves a transition away from high-dose animal studies to in vitro and computational models (NRC in Toxicity testing in the 21st century: a vision and a strategy, The National Academies Press, Washington, DC, 2007). This transition requires mapping pathways of toxicity by understanding how in vitro systems respond to chemical perturbation. Uncovering transcription factors/signaling networks responsible for gene expression patterns is essential for defining pathways of toxicity, and ultimately, for determining the chemical modes of action through which a toxicant acts. Traditionally, transcription factor identification is achieved via chromatin immunoprecipitation studies and summarized by calculating which transcription factors are statistically associated with up- and downregulated genes. These lists are commonly determined via statistical or fold-change cutoffs, a procedure that is sensitive to statistical power and may not be as useful for determining transcription factor associations. To move away from an arbitrary statistical or fold-change-based cutoff, we developed, in the context of the Mapping the Human Toxome project, an enrichment paradigm called information-dependent enrichment analysis (IDEA) to guide identification of the transcription factor network. We used a test case of activation in MCF-7 cells by 17ß estradiol (E2). Using this new approach, we established a time course for transcriptional and functional responses to E2. ERα and ERß were associated with short-term transcriptional changes in response to E2. Sustained exposure led to recruitment of additional transcription factors and alteration of cell cycle machinery. TFAP2C and SOX2 were the transcription factors most highly correlated with dose. E2F7, E2F1, and Foxm1, which are involved in cell proliferation, were enriched only at 24 h. IDEA should be useful for identifying candidate pathways of toxicity. IDEA outperforms gene set enrichment analysis (GSEA) and provides similar results to weighted gene correlation network analysis, a platform that helps to identify genes not annotated to pathways.

Molecular mechanisms of multiple toxin-antitoxin systems are coordinated to govern the persister phenotype.

Toxin-antitoxin systems are ubiquitous and have been implicated in persistence, the multidrug tolerance of bacteria, biofilms, and, by extension, most chronic infections. However, their purpose, apparent redundancy, and coordination remain topics of debate. Our model relates molecular mechanisms to population dynamics for a large class of toxin-antitoxin systems and suggests answers to several of the open questions. The generic architecture of toxin-antitoxin systems provides the potential for bistability, and even when the systems do not exhibit bistability alone, they can be coupled to create a strongly bistable, hysteretic switch between normal and toxic states. Stochastic fluctuations can spontaneously switch the system to the toxic state, creating a heterogeneous population of growing and nongrowing cells, or persisters, that exist under normal conditions, rather than as an induced response. Multiple toxin-antitoxin systems can be cooperatively marshaled for greater effect, with the dilution determined by growth rate serving as the coordinating signal. The model predicts and elucidates experimental results that show a characteristic correlation between persister frequency and the number of toxin-antitoxin systems.

Evolution of a genome-encoded bias in amino acid biosynthetic pathways is a potential indicator of amino acid dynamics in the environment.

Overcoming the stress of starvation is one of an organism's most challenging phenotypic responses. Those organisms that frequently survive the challenge, by virtue of their fitness, will have evolved genomes that are shaped by their specific environments. Understanding this genotype-environment-phenotype relationship at a deep level will require quantitative predictive models of the complex molecular systems that link these aspects of an organism's existence. Here, we treat one of the most fundamental molecular systems, protein synthesis, and the amino acid biosynthetic pathways involved in the stringent response to starvation. These systems face an inherent logical dilemma: Building an amino acid biosynthetic pathway to synthesize its product-the cognate amino acid of the pathway-may require that very amino acid when it is no longer available. To study this potential "catch-22," we have created a generic model of amino acid biosynthesis in response to sudden starvation. Our mathematical analysis and computational results indicate that there are two distinctly different outcomes: Partial recovery to a new steady state, or full system failure. Moreover, the cell's fate is dictated by the cognate bias, the number of cognate amino acids in the corresponding biosynthetic pathway relative to the average number of that amino acid in the proteome. We test these implications by analyzing the proteomes of over 1,800 sequenced microbes, which reveals statistically significant evidence of low cognate bias, a genetic trait that would avoid the biosynthetic quandary. Furthermore, these results suggest that the pattern of cognate bias, which is readily derived by genome sequencing, may provide evolutionary clues to an organism's natural environment.

PeakDecoder enables machine learning-based metabolite annotation and accurate profiling in multidimensional mass spectrometry measurements.

Multidimensional measurements using state-of-the-art separations and mass spectrometry provide advantages in untargeted metabolomics analyses for studying biological and environmental bio-chemical processes. However, the lack of rapid analytical methods and robust algorithms for these heterogeneous data has limited its application. Here, we develop and evaluate a sensitive and high-throughput analytical and computational workflow to enable accurate metabolite profiling. Our workflow combines liquid chromatography, ion mobility spectrometry and data-independent acquisition mass spectrometry with PeakDecoder, a machine learning-based algorithm that learns to distinguish true co-elution and co-mobility from raw data and calculates metabolite identification error rates. We apply PeakDecoder for metabolite profiling of various engineered strains of Aspergillus pseudoterreus, Aspergillus niger, Pseudomonas putida and Rhodosporidium toruloides. Results, validated manually and against selected reaction monitoring and gas-chromatography platforms, show that 2683 features could be confidently annotated and quantified across 116 microbial sample runs using a library built from 64 standards.

Phenotypes and tolerances in the design space of biochemical systems.

One of the major unsolved problems of modern biology is deep understanding of the complex relationship between the information encoded in the genome of an organism and the phenotypic properties manifested by that organism. Fundamental advances must be made before we can begin to approach the goal of predicting the phenotypic consequences of a given mutation or an organism's response to a novel environmental challenge. Although this problem is often portrayed as if the task were to find a more or less direct link between genotypic and phenotypic levels, on closer examination the relationship is far more layered and complex. Although there are some intuitive notions of what is meant by phenotype at the level of the organism, it is far from clear what this term means at the biochemical level. We have described design principles that are readily revealed by representation of molecular systems in an appropriate design space. Here, we first describe a generic approach to the construction of such a design space in which qualitatively distinct phenotypes can be identified and counted. Second, we show how the boundaries between these phenotypic regions provide a method of characterizing a system's tolerance to large changes in the values of its parameters. Third, we illustrate the approach for one of the most basic modules of biochemical networks and describe an associated design principle. Finally, we discuss the scaling of this approach to large systems.

Automated construction and analysis of the design space for biochemical systems.

MOTIVATION: Our recent work introduced a generic method to construct the design space of biochemical systems: a representation of the relationships between system parameters, environmental variables and phenotypic behavior. In design space, the qualitatively distinct phenotypes of a biochemical system can be identified, counted, analyzed and compared. Boundaries in design space indicate a transition between phenotypic behaviors and can be used to measure a system's tolerance to large changes in parameters. Moreover, the relative size and arrangement of such phenotypic regions can suggest or confirm global properties of the system. RESULTS: Our work here demonstrates that the construction and analysis of design space can be automated. We present a formal description of design space and a detailed explanation of its construction. We also extend the notion to include variable kinetic orders. We describe algorithms that automate common steps of design space construction and analysis, introduce new analyses that are made possible by such automation and discuss challenges of implementation and scaling. In the end, we demonstrate the techniques using software we have created. AVAILABILITY: The Design Space Toolbox for MATLAB is freely available at http://www.bme.ucdavis.edu/savageaulab/ CONTACT: masavageau@ucdavis.edu

Mechanistic Modeling of Biochemical Systems without A Priori Parameter Values Using the Design Space Toolbox v.3.0.

Mechanistic models of biochemical systems provide a rigorous description of biological phenomena. They are indispensable for making predictions and elucidating biological design principles. To date, mathematical analysis and characterization of these models encounter a bottleneck consisting of large numbers of unknown parameter values. Here, we introduce the Design Space Toolbox v.3.0 (DST3), a software implementation of the Design Space formalism enabling mechanistic modeling without requiring previous knowledge of parameter values. This is achieved by using a phenotype-centric modeling approach, in which the system is first decomposed into a series of biochemical phenotypes. Parameter values realizing phenotypes of interest are subsequently predicted. DST3 represents the most generally applicable implementation of the Design Space formalism and offers unique advantages over earlier versions. By expanding the Design Space formalism and streamlining its distribution, DST3 represents a valuable tool for elucidating biological design principles and designing novel synthetic circuits.

Unrelated toxin-antitoxin systems cooperate to induce persistence.

Persisters are drug-tolerant bacteria that account for the majority of bacterial infections. They are not mutants, rather, they are slow-growing cells in an otherwise normally growing population. It is known that the frequency of persisters in a population is correlated with the number of toxin-antitoxin systems in the organism. Our previous work provided a mechanistic link between the two by showing how multiple toxin-antitoxin systems, which are present in nearly all bacteria, can cooperate to induce bistable toxin concentrations that result in a heterogeneous population of slow- and fast-growing cells. As such, the slow-growing persisters are a bet-hedging subpopulation maintained under normal conditions. For technical reasons, the model assumed that the kinetic parameters of the various toxin-antitoxin systems in the cell are identical, but experimental data indicate that they differ, sometimes dramatically. Thus, a critical question remains: whether toxin-antitoxin systems from the diverse families, often found together in a cell, with significantly different kinetics, can cooperate in a similar manner. Here, we characterize the interaction of toxin-antitoxin systems from many families that are unrelated and kinetically diverse, and identify the essential determinant for their cooperation. The generic architecture of toxin-antitoxin systems provides the potential for bistability, and our results show that even when they do not exhibit bistability alone, unrelated systems can be coupled by the growth rate to create a strongly bistable, hysteretic switch between normal (fast-growing) and persistent (slow-growing) states. Different combinations of kinetic parameters can produce similar toxic switching thresholds, and the proximity of the thresholds is the primary determinant of bistability. Stochastic fluctuations can spontaneously switch all of the toxin-antitoxin systems in a cell at once. The spontaneous switch creates a heterogeneous population of growing and non-growing cells, typical of persisters, that exist under normal conditions, rather than only as an induced response. The frequency of persisters in the population can be tuned for a particular environmental niche by mixing and matching unrelated systems via mutation, horizontal gene transfer and selection.

The Human Toxome Collaboratorium: A Shared Environment for Multi-Omic Computational Collaboration within a Consortium.

The Human Toxome Project is part of a long-term vision to modernize toxicity testing for the 21st century. In the initial phase of the project, a consortium of six academic, commercial, and government organizations has partnered to map pathways of toxicity, using endocrine disruption as a model hazard. Experimental data is generated at multiple sites, and analyzed using a range of computational tools. While effectively gathering, managing, and analyzing the data for high-content experiments is a challenge in its own right, doing so for a growing number of -omics technologies, with larger data sets, across multiple institutions complicates the process. Interestingly, one of the most difficult, ongoing challenges has been the computational collaboration between the geographically separate institutions. Existing solutions cannot handle the growing heterogeneous data, provide a computational environment for consistent analysis, accommodate different workflows, and adapt to the constantly evolving methods and goals of a research project. To meet the needs of the project, we have created and managed The Human Toxome Collaboratorium, a shared computational environment hosted on third-party cloud services. The Collaboratorium provides a familiar virtual desktop, with a mix of commercial, open-source, and custom-built applications. It shares some of the challenges of traditional information technology, but with unique and unexpected constraints that emerge from the cloud. Here we describe the problems we faced, the current architecture of the solution, an example of its use, the major lessons we learned, and the future potential of the concept. In particular, the Collaboratorium represents a novel distribution method that could increase the reproducibility and reusability of results from similar large, multi-omic studies.

Qualitatively distinct phenotypes in the design space of biochemical systems.

Although characterization of the genotype has undergone revolutionary advances as a result of the successful genome projects, the chasm between our understanding of a fully characterized gene sequence and the phenotypic repertoire of the organism is as broad and deep as it was in the pre-genomic era. There are two fundamental unsolved problems that provide the context for the challenges in relating genotype to phenotype. We address one of these and describe a generic method for constructing a system design space in which qualitatively distinct phenotypes can be identified and counted, their relative fitness analyzed and compared, and their tolerance to change measured.