Wei-Shou Hu, Associate Fellow

Mechanistic Study of Hepatocyte Spheroid Formation; Modeling Regulatory Networks and Cell Metabolism From a Global Perspective

Self-assembled hepatocyte spheriods with internal channels.

The study of tissue formation in vitro requires the use of noninvasive tools. To that end, these researchers used confocal microscopy in conjunction with image analysis to observe the mechanical behavior of hepatocytes during their self-assembly in round aggregates or spheroids and to probe their functions within these three-dimensional structures. The out-of-focus information—as well as the electronic noise introduced during image acquisition—can result in artifacts and impede further data analysis and volume reconstruction. Subsequently, the processed images were transferred for three-dimensional reconstruction to the SGI workstations in the Basic Sciences Computing Laboratory (BSCL).

In this work, the group developed an assay, which requires the use of confocal microscopy and image analysis, for in situ assessment of CYP2B1/2 activity in three-dimensional tissuelike structures of primary hepatocytes. The images taken at different time points and the subsequent reconstruction revealed the spatiotemporal bioactivity of hepatocytes in the aggregate. The visualization of cellular and subcellular activities during the formation of tissue-like structures is crucial in elucidating aspects of the mechanisms that cells employ during these phenomena. Better understanding of these phenomena has a great impact in tissue engineering and drug screening applications.

The different states during the course of spheroid formation are characterized by differing patterns of gene expression. With the dramatic progress in sequencing the genomes of different organisms, it is now possible to monitor the expression levels of all the genes in the genome simultaneously. One of the promising methods is carrier deoxyribonucleic acid (cDNA) microarray analysis of gene expression profiles. In this process, labeled cDNA is incubated on a DNA microarray containing thousands of spots, each corresponding to a unique DNA sequence. It is assumed that the fluorescence intensity of a spot is proportional to the initial concentration of the corresponding cDNA species in solution. Labeled DNA species have to diffuse in solution to the corresponding spot and undergo a second-order hybridization reaction with the immobilized DNA.

The research team has developed a kinetic model for the hybridization process that considers the dynamics of diffusion of labeled cDNA strands and duplex-formation reactions. They used finite element analysis to numerically solve the partial differential equations that arise from the modeling. This study’s goal is to formulate strategies for maximizing the ratio of true to false positive fluorescence intensity for species with varying abundance levels and varying degrees of regulation. The model predictions will be compared with results of hybridization experiments using defined concentration and identity of labeled cDNA species.

Another project by this group involves modeling regulatory networks and cell metabolism from a global perspective. Cell metabolism is essentially a complex network of reactions with many enzymes and reactants involved, as well as a regulatory network controlling the expression of both regulatory elements and the biochemical network. The regulatory network consists of a large number of regulatory elements of interacting genes and proteins organized in hierarchical trees. Cellular events, including physiological, differentiation, and developmental, involve the interplay of these regulatory networks.

Models of cell metabolism provide a deeper understanding and a way to organize these reactions and the interactions that occur inside cells. Dynamic models additionally provide a way to study and predict the behavior and changes over a time period that cells experience when exposed to environmental changes. This knowledge provides an essential link between functional genomics and physiomics. With the advent of the post-genomic era, a variety of large-scale gene expression profiling techniques, such as DNA microarrays, has enabled this group to survey the temporal expression pattern of the regulatory elements. Deciphering this information for reconstructing the hierarchy of the regulatory elements is becoming even more urgent.

There are several possible approaches for modeling regulatory elements. This group developed an algorithm based on the Boolean framework to reverse engineer the network. A set of parsimonious networks is predicted from expression profiles obtained by perturbing the network under different conditions. Computational simulations have been presented to substantiate these results. The group evaluated the performance of the proposed algorithms using both a set of syntheticallygenerated networks as well as an actual network such as that derived from the yeast cell mating pathway. The group generated a set of synthetic networks by varying the number of genes/nodes, connections/edges, and maximum degree of a node. Both acyclic and cyclic networks were considered. For each node, the researchers used a randomly-generated conjunctive expression to generate its Boolean function. Care was taken to ensure that the gene would not be independent of its immediate inputs. Using this approach, the group generated a number of different networks with genes varying from 10 to 100 and with varying in-degree. The programming is done entirely in c. However, exploring the entire possible network space demands many computer resources and the problem quickly scales up with the number of nodes.

The number of equations and reactions required to describe cell metabolism is large, due to the number of enzymes and reactions that are involved in this complex network. This modeling approach tends to reduce the number of equations used to accurately describe cell metabolism. First, the network is identified. Kinetic expressions based on mechanistic knowledge are used to describe each reaction. Then, the system is simplified based on the differences in time scale where each of these reactions occurs. This provides a systematic way to analyze and describe these systems and to properly examine steady-state assumptions (that are usually made for fast reactions) without eliminating the effect that the fast reactions have on the slower reactions and the whole metabolic network.

This approach requires the use of symbolic computation methods (like MAPLE or MATHEMATICA) for the analysis of the network, determination of constraints derived from the fasterscale reactions, simplification of the differential algebraic equations, and numerical solution of the simplified system.

Research Group and Collaborators

Prodromos Daoutidis, Faculty Collaborator
Shen Dong, Graduate Student Researcher
Susan Fugett Abu-Absi, Graduate Student Researcher
Chetan Gadgil, Graduate Student Researcher
Ziomara Gerdtzen, Graduate Student Researcher
George Karypis, Faculty Collaborator
Sarika Mehra, Graduate Student Researcher