Modeling and estimation of dynamic EGFR pathway by data assimilation approach using time series proteomic data.

Shinya Tasaki, Masao Nagasaki, Masaaki Oyama, Hiroko Hata, Kazuko Ueno, Ryo Yoshida, Tomoyuki Higuchi, Sumio Sugano, Satoru MiyanoPubMed 2006

TL;DR

This paper presents a novel approach to building and refining complex biological pathway models by integrating time-series proteomic data with a data assimilation (DA) framework. By applying this method to the well-studied Epidermal Growth Factor Receptor (EGFR) pathway, the researchers demonstrate a semi-automated way to construct accurate in silico models, paving the way for more robust predictions of cellular behavior.

What Problem Does This Paper Address?

Understanding complex biological signaling pathways, like the EGFR pathway, is crucial for deciphering cellular processes and understanding diseases such as cancer. While extensive literature exists on these pathways, building accurate computational models that capture their dynamic behavior has been a significant challenge. Traditional modeling approaches often rely on manually curated parameters, which become intractable as pathway complexity increases. Furthermore, integrating experimental data to refine these models has been difficult due to the complexity of biological systems and the limitations of available quantitative data. This paper tackles the problem of creating more accurate and dynamic in silico models by developing a method that leverages high-throughput proteomic data and a data assimilation framework to automate parameter estimation and model refinement.

Methods

The core of this work lies in the integration of two key components:

  • Hybrid Functional Petri Net with Extension (HFPNe): This is a modeling architecture used to represent biological pathways. HFPNe extends traditional Petri nets by allowing for discrete, continuous, and generic entities and processes, enabling more detailed modeling of biological mechanisms like phosphorylation, ubiquitination, and protein interactions. The Cell Illustrator software is used as the platform for building these HFPNe models.
  • Data Assimilation (DA) Framework: This statistical framework is employed to automatically estimate kinetic parameters of the HFPNe model by fitting it to experimental time-series data. The DA approach allows for the selection of the most plausible model from a set of hypothesized models, effectively performing both parameter estimation and model selection.

To handle the complexity of the EGFR pathway, the researchers employed a strategy of dividing the model into three interconnected sub-pathways. This decomposition simplifies the parameter estimation process by allowing for sequential estimation and fixing of parameters within each sub-pathway, reducing the computational burden and improving the efficiency of the DA framework.

Datasets

The primary dataset used in this study is time-series quantitative proteomic data from EGF-stimulated HeLa cells, specifically focusing on tyrosine-phosphorylated proteins. This data was obtained using the Stable Isotope Labeling by Amino Acids in Cell Culture (SILAC) method coupled with liquid chromatography-tandem mass spectrometry (LC/MS/MS), as reported by Blagoev et al. (2004). This dataset provided measurements for 81 tyrosine-phosphorylated proteins related to EGFR signaling.

Key Insights and Ideas

  • Bridging the Gap between Proteomics and Modeling: The paper successfully demonstrates that high-throughput proteomic data, particularly time-series measurements of protein phosphorylation, can be effectively integrated into computational pathway models. This moves beyond static pathway maps to dynamic, data-driven simulations.
  • DA as a Powerful Tool for Model Refinement: The data assimilation framework proved to be a robust method for both estimating kinetic parameters and selecting the best model among several hypothesized versions. This automates a significant portion of the model building process, which was previously heavily reliant on manual tuning.
  • Decomposition for Scalability: The strategy of dividing the complex EGFR pathway into smaller, manageable sub-pathways was crucial for enabling efficient parameter estimation. This approach highlights a practical solution for tackling the inherent complexity of biological systems in computational modeling.
  • Identifying Model Deficiencies: The discrepancy in fitting the p38MAPK dynamics revealed that the initial model lacked certain regulatory mechanisms, specifically those related to downregulation. This insight directly guided the generation of new hypotheses for model refinement.
  • Biological Knowledge Remains Essential: While DA automates parameter estimation, the process of generating hypothesized models still relies on biological background knowledge, underscoring the synergistic relationship between computational and experimental approaches.

Key Results

  • The study successfully constructed a well-tuned EGFR HFPNe model using the Cell Illustrator coupled with the DA framework.
  • The model achieved excellent fitting scores for most of the measured phosphorylated proteins (EGFR, Shc, c-Cbl, Vav2, Erk2), with scores generally above 94%.
  • A significant discrepancy was observed in the fitting of p-p38MAPK, indicating a need for further model refinement.
  • Through model selection using the DA framework, the researchers identified model 6 as the best among ten hypothesized models. This model incorporated direct inactivation of MKK/3/4/6/7 and activation of phosphatase by p38MAPK, suggesting that p38MAPK inhibitory regulations might stem from its own activation rather than from Erk2.

Limitations and Future Work

  • Model Complexity and Unknown Proteins: The authors acknowledge that the proteomic data includes many proteins with unknown functions. Modeling these proteins will require further understanding of their roles in signal transduction, necessitating continued experimental investigation.
  • Scalability to Highly Complex Systems: While the DA framework was shown to work for a relatively complex system like the EGFR pathway, its applicability to even more intricate cellular events remains a challenge to be explored.
  • Future Directions: The authors suggest extending their work by combining the DA framework with Bayesian network inference. This could allow for the identification of causal relationships within signaling networks and the modeling of function-unknown proteins, further enhancing the predictive power of in silico models. They also envision a future where DA can be applied to model the detailed events within the entire cell.

Summary updated

Abstract

Cell Illustrator is a model building tool based on the Hybrid Functional Petri net with extension (HFPNe). By using Cell Illustrator, we have succeeded in modeling biological pathways, e.g., metabolic pathways, gene regulatory networks, microRNA regulatory networks, cell signaling networks, and cell-cell interactions. The recent development of tandem mass spectrometry coupled with liquid chromatography (LC/MS/MS) technology has enabled researchers to quantify the dynamic profile of a wide range of proteins within the cell. The proteomic data obtained by using LC/MS/MS has been considerably useful for introducing dynamics to the HFPNe model. Here, we report the first introduction of the time-series proteomic data to our HFPNe model. We constructed an epidermal growth factor receptor signal transduction pathway model (EFGR model) by using the biological data available in the literature. Then, the kinetic parameters were determined in the data assimilation (DA) framework with some manual tuning so as to fit the proteomic data published by Blagoev et al. (Nat. Biotechnol., 22:1139-1145, 2004). This in silico model was further refined by adding or removing some regulation loops using biological background knowledge. The DA framework was used to select the most plausible model from among the refined models. By using the proteomic data, we semi-automatically constructed a well-tuned EGFR HFPNe model by using the Cell Illustrator coupled with the DA framework.
BibTeX
@inproceedings{tasaki2006modeling,
  author = {Shinya Tasaki and Masao Nagasaki and Masaaki Oyama and Hiroko Hata and Kazuko Ueno and Ryo Yoshida and Tomoyuki Higuchi and Sumio Sugano and Satoru Miyano},
  title = {{Modeling and estimation of dynamic EGFR pathway by data assimilation approach using time series proteomic data.}},
  booktitle = {PubMed},
  year = {2006},
}
Figure