Research Article: An Integrated Framework to Model Cellular Phenotype as a Component of Biochemical Networks

Date Published: November 29, 2011

Publisher: Hindawi Publishing Corporation

Author(s): Michael Gormley, Viswanadha U. Akella, Judy N. Quong, Andrew A. Quong.


Identification of regulatory molecules in signaling pathways is critical for understanding cellular behavior. Given the complexity of the transcriptional gene network, the relationship between molecular expression and phenotype is difficult to determine using reductionist experimental methods. Computational models provide the means to characterize regulatory mechanisms and predict phenotype in the context of gene networks. Integrating gene expression data with phenotypic data in transcriptional network models enables systematic identification of critical molecules in a biological network. We developed an approach based on fuzzy logic to model cell budding in Saccharomyces cerevisiae using time series expression microarray data of the cell cycle. Cell budding is a phenotype of viable cells undergoing division. Predicted interactions between gene expression and phenotype reflected known biological relationships. Dynamic simulation analysis reproduced the behavior of the yeast cell cycle and accurately identified genes and interactions which are essential for cell viability.

Partial Text

Efforts to develop therapeutics for complex disorders such as cancer, infectious disease, and autoimmune disease require an understanding of the specific pathways through which networks of molecular interactions influence cellular function. Due to the complexity of biochemical pathways, a combinatorially large number of experiments that can simultaneously measure the changes in gene or protein expression such as a microarray or an LCMS-based proteomics are required in order to fully characterize normal and disease-producing mechanisms [1]. An iterative approach, using computational biology to complement high-throughput experimentation, may increase the efficiency by which data can be gathered by eliminating redundant or irrelevant experiments and suggesting hypotheses to build optimally upon current knowledge [2–4]. Development of gene expression microarray platforms enables the collection of expression data on a genome-wide scale sufficient for the derivation of gene-gene interactions and reverse engineering of system’s scale models of gene networks [5, 6]. However, computational models of biological systems often disregard cellular phenotype data. Phenotype should be explicitly incorporated in computational gene network models to contextualize perturbations according to their effect on the desired change in cellular phenotype. This not only allows for a seamless coupling between computation and experimentation but also enables a guided search to identify molecules, complexes, and pathways that regulate disease-specific processes such as migration, proliferation, differentiation, or cell death [2, 4].

In this paper, we describe the development and analysis of a fuzzy logic model of cell cycle in Saccharomyces cerevisiae relating the expression of seventeen cell cycle genes to the budding phenotype. The structure of the model and semi-quantitative rules describing regulatory interactions between genes and between genes and phenotype were derived from a time series gene expression microarray dataset using an exhaustive search method. Best fit models for each gene and phenotype were identified and interpreted based on agreement with known interactions from the literature. In addition, node-specific models were integrated into a composite network model, and a simple iterative scheme was used to approximate the dynamic behavior of the system. The dynamic model converges to two alternative self-consistent states, matching hypotheses developed from experimental investigation. The composite network model was then analyzed to identify essential genes, that is, genes which are necessary for viability, and to predict synthetic lethal and synthetic rescue phenotypes in silico. This work represents a proof of concept demonstrating the feasibility of integrating phenotype information into mechanistic transcriptional models and the value of this approach in guiding hypothesis generation.

Molecular profiling technologies such as gene expression microarrays have enabled the quantification of molecular abundance at a genome-wide scale in high throughput. As technologies have matured, goals of data analysis have grown from the identification of conserved expression patterns across samples or conditions to the comprehension of gene function in the context of complex regulatory and functional networks. In this study, we have extended a fuzzy logic-based modeling approach to derive a transcriptional network consisting of seventeen genes known to be important for cell cycle regulation in yeast and a network element representing a phenotypic observation (the fraction of budding cells). Both the topology (i.e., interactions) and regulatory traits (e.g., stimulation, inhibition) of relationships between genes and between genes and phenotype are derived from publicly available gene microarray expression data and phenotype data using a bounded exhaustive search of the potential interaction space. Comparison of inferred gene regulatory interactions with known interactions in the literature provides confidence in the biological relevance of model predictions. Through the analysis of best fit fuzzy logic signaling models, genes with direct and indirect effects on phenotype were identified. In addition, we used our model to predict the effects of gene knockdown on cellular viability. In this manner, the methodology we have developed provides a direct link between computational analysis of molecular profiling data and experimental observations. We envision coupling this computational modeling method with experimentation in an iterative fashion to incrementally build understanding of regulatory mechanisms that control poorly understood cellular functions. This approach allows for systematic characterization of gene function in the context of disease-related, functional mechanisms that will facilitate rational design of targeted therapies.




Leave a Reply

Your email address will not be published.