Fei Hua

Postdoctoral Associate
Sept.2002-Sept.2004, Center for Cancer Research, Advisor: Luk Van Parijs
Oct. 2004- Oct. 2005, Department of Biological Engineering, Advisor: Douglas A. Lauffenburger

EDUCATION

Ph.D. in Physiology, Cornell University, Ithaca, NY
B.S. in Biological Sciences, Tsinghua University, Beijing, P.R.China PRIMARY

COLLABORATORS

Sampsa Hautaniemi, Lauffenburger Lab, MIT Cindy Stokes, Entelos Inc.

RESEARCH SUMMARY

Programmed cell death or apoptosis is a critical process for maintaining the homeostasis of multicellular organisms. Fas (CD95/Apo-1) receptor is an important trigger for apoptosis. Defects in this receptor-triggered signaling can lead to the development of autoimmune diseases and cancers. We have applied two different types of computational approaches to study the regulation of this pathway.

As the first approach, we created an ordinary differential equation based mathematical model, which integrated current information concerning the signaling network downstream of the Fas activation. Since the mechanisms for Bcl-2 to block one branch of the Fas pathway is unclear, we used the model to test different inhibitory interactions of Bcl-2 with various components on the pathway. By comparing dynamic changes of the outcome (caspase-3 activation) in response to Bcl-2 up-regulation between different model simulations and experiments, we find Bcl-2 binding to both molecule Bax and truncated Bid, instead of Bax, Bid or truncated Bid alone, gives the best agreement between the two. Moreover, although Bcl-2 overexpression dramatically slows down caspase-3 activation, decreasing Bcl-2 level only has a negligible effect on the outcome in both model simulation and experiments. These results demonstrate a general model finding that varying the expression levels of signal molecules frequently has asymmetrical effects on the outcome.

Since signaling networks governing cell functional behaviors are highly multivariate and interconnected, as the second approach, we developed a series of machine-learning approaches to understand the regulation of the Fas pathway with different combinations of multiple protein levels. This methodology included creating an ODE model for the Fas pathway, generating a large simulation dataset with combination of varied initial conditions, clustering the dataset based on model outcome, applying decision tree model to the clustered dataset, and calculating a matrix of transition cost between different rules based on the decision tree. Using this approach, we were able to identify that similar outcome can be generated from different sets of multivariate protein states, moreover different perturbations may be required to alter similar phenotypic behaviors.

CDP PUBLICATIONS

Hua F, Cornejo MG, Stokes CL and Lauffenburger DA. (2005) Effects of Bcl-2 levels on Fas signaling-induced caspase-3 activation: molecular genetic tests of computational model predictions. (Journal of Immunology, accepted with minor revision).