Title

Probabilistic dynamic modeling of the ErbB signaling pathways.

Investigators

Peter Sorger, Dennis Vitkup

Status

In Progress

Abstract

ErbB pathways, involved in cell proliferation, survival and motility, are among the best studied mammalian signaling networks (Citri 2006). Dysregulation of the ErbB signaling has been implicated in a variety of human cancers (Slamon 1987, Rusch 1993). ErbB receptors and downstream targets are also a major focus of current pharmaceutical research (Paez 2004, Niepel 2009). The combinatorial complexity of the pathway, both at the level of ErbB1-4 receptors and at the level of their multiple downstream targets, makes computational modeling essential in understanding ErbB-related signaling (Lazzara 2009). Kinetic models of ErbB pathways are usually based on ordinary differential equations, where core interactions are represented as elementary reactions with appropriate rate laws (Kholodenko 1999, Schoeberl 2002, Hatakeyama 2003, Hendrinks 2005, Birtwistle 2007). Although helpful, the predictive power of these models is limited by the presence of a large number of unmeasured and uncalibrated parameters. Currently, there is no principled way to account for parameter uncertainty in existing models.

This project represents an experimental-computational collaboration between two groups actively involved in analysis of dynamic cellular pathways. The Sorger lab and its collaborators have been involved in modeling ErbB pathways for over a decade (Hendrinks 2005, Chen 2009). The Sorger lab is also an established leader in the collection and analysis of dynamic multi-factorial data, using a variety of approaches including high throughput micro-ELISA assays, flow cytometry, fixed and live-cell imaging combined with perturbations by RNAi and small-molecules drugs (Albeck 2006). The Vitkup lab pioneered the application of the integrative probabilistic approaches to genomic data integration in the context of metabolic (Hsiao 2009) and protein-protein disease networks (Feldman 2008). The main goal of the project is to assemble a comprehensive collection of quantitative multi-factorial data for model calibration of immediate-early signaling in ErbB pathways and to predict ErbB pathway dynamics using a new principled Bayesian formalism. Specifically, the probability distributions of parameters, consistent with the measured experimental data, will be obtained using advanced Markov Chain Monte Carlo (MCMC) simulations. Our main focus will be the understanding of the core biological processes controlling the ErbB pathways and the use of probabilistic methods to infer relevant pathway perturbations. The model prediction will be tested and refined by inhibiting different components of the pathway, using either RNAi or available small-molecules. We will use experimental perturbations to test the importance of feedback loops within and between various pathway cascades. The likely effects of frequent coding mutations in cancer cell lines will be also investigated.

Progress

We are using a variety of methods to collect data on ErbB signaling. The simplest involve multiplex biochemical assays of phospho-proteins and have provided data on 10-12 distinct proteins at multiple time points (used for the first round of modeling described below). We are currently collecting quantitative mass spectrometry data to determine absolute protein abundances (in molecules per cell) and to measure the modification states of additional kinases. One promising approach involves binding kinases to biotin-modified cross-linkable ATP, purifying bound proteins on streptavidin followed by MS3 mass spectrometry. This has the potential to yield detailed phospho-dynamics on ~100 kinases. We are also working on single-cell measurements which are better suited to elucidate the asynchronous cell response to ErbB ligands, compared to population-average data. Fixed cell imaging makes it possible to assay ~6 kinases, 5 cell cycle regulators and 5 ErbB-regulated transcription factors in sets of two or three proteins per assay. This provides information on distributions of protein activities that will be invaluable for modeling. We are also establishing dynamical single-cell assays for receptor oligomerization and for Erk and Akt signaling.

Using a differential equation model of the ErbB signaling we performed MCMC simulations in the space of 174 model parameters. The simulated parameters included protein concentrations and kinetic rates of all relevant model reactions. Multiple MCMC chains were run in parallel using adaptive step size adjustment. The chain convergence was evaluated using the Gelman-Rubin criterion. The probability distributions of the model parameters were compared with estimates available in the literature. The literature-based estimates generally agree with the simulated distributions for most proteins. Using the same approach we also studied how several key phosphatases acting on EGFR, ERK, and MEK are essential for controlling ErbB network dynamics. Specifically, we simulated inhibition of the key phosphatases in the model at various times after EGF treatment and were able to experimentally validate the model predictions. More generally, we used the Kullback-Leibler distance measure to determine all model parameters essential for long-term signal response in the ErbB network. For each parameter, we compared its distribution when the model was fitted only with early time points (

References

Albeck JG, MacBeath G, White FM, Sorger PK, Lauffenberger DA, Gaudet S. Collecting and organizing systematic sets of protein data. Nat Rev Mol Cell Biol. 2006;7:803-12.

Birtwistle MR, Hatakeyama M, Yumoto N, Ogunnaike BA, Hoek JB, Kholodenko BN. Ligand-dependent responses of the ErbB signaling network: experimental and modeling analyses. Mol Sys Biol. 2007;3:144.

Chen WW, Schoeberl B, Jasper PJ, Niepel M, Nielsen UB, Lauffenburger DA, Sorger PK. Input-output behavior of ErbB signaling pathways as revealed by a mass action model trained against dynamic data. Mol Syst Biol. 2009;5:239.

Citri A, Yarden Y. EGF-ERBB signalling: towards the systems level. Nat Rev Mol Cell Biol. 2006;7:505-16.

Feldman I, Rzhetsky A, Vitkup D. Network properties of genes harboring inherited disease mutations. Proc Natl Acad Sci U S A. 2008;105(11):4323-8.

Hatakeyama M, Kimura S, Naka T, Kawasaki T, Yumoto N, Ichikawa M, et al. A computational model on the modulation of mitogen-activated protein kinase (MAPK) and Akt pathways in heregulin-induced ErbB signalling. Biochem J. 2003;373:451-63.

Hendrinks BS, Orr G, Wells A, Wiley HS, Lauffenberger DA. Parsing ERK activation reveals quantitatively equivalent contributions from epidermal growth factor receptor and HER2 in human mammary epithelial cells. J Biol Chem. 2005;280:6157-69.

Hsiao TL, Revelles O, Chen L, Sauer U, Vitkup D. Automatic policing of biochemical annotations using genomic correlations. Nat Chem Biol. 2009;6(1):34-40.

Kholodenko BN, Demin OV, Moehren G, Hoek JB. Quantification of short term signaling by the epidermal growth factor receptor. J Biol Chem. 1999;274:30169-81.

Lazzara MJ, Lauffenburger DA. Quantitative modeling perspectives on the ErbB system of cell regulatory processes. Exp Cell Res. 2009;315(4):717-25.

Niepel M, Spencer SL, Sorger PK. Non-genetic cell-to-cell variability and the consequences for pharmacology. Curr Opin Chem Biol. 2009;13(5-6):556-61.

Paez JG, Janne PA, Lee JC, Tracy S, Greulich H, Gabriel S, et al. EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science. 2004;304:1497-500.

Rusch V, Baselga J, Cordon-Cardo C, Orazem J, Zaman M, Hoda S, et al. Differential expression of the epidermal growth factor receptor and its ligands in primary non-small cell lung cancers and adjacent benign lung. Cancer Res. 1993;53:2379-85.

Schoeberl B, Eichler-Jonsson C, Gilles ED, Muller G. Computational modeling of the dynamics of the MAP kinase cascade activated by surface and internalized EGF receptors. Nat Biotechnol. 2002;20:370-5.

Slamon DJ, Clark GM, Wong SG, Levin WJ, Ulrich A, McGuire WL. Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science. 1987;235:177-82.