(419c) A Hidden Markov Model Reconstruction of a Cellular Signaling Network in Embryonic Fibroblasts from Time-Course Gene Expression Profiles Reveals the Mechanism of the SPRY2 Tumor-Suppressor
- Conference: AIChE Annual Meeting
- Year: 2015
- Proceeding: 2015 AIChE Annual Meeting Proceedings
- Group: Emerging Frontiers in Systems and Synthetic Biology
Tuesday, November 10, 2015 - 3:51pm-4:09pm
A hidden Markov model reconstruction of a cellular signaling network in embryonic fibroblasts from time-course gene expression profiles reveals the mechanism of the SPRY2 tumor-suppressor.
Mark F. Ciaccio and Neda Bagheri. Department of Chemical & Biological Engineering, McCormick School of Engineering, Northwestern University.
Introduction: Cellular phenotype is controlled, in part, by the abundance of signaling proteins as well as the dynamic parameters governing their activation. For this reason, time-course gene expression data is invaluable for understanding the molecular regulatory mechanisms that underlie cancer behavior. Sprouty Homolog 2 (SPRY2), a known tumor suppressor, not only attenuates the amplitude of Receptor Tyrosine Kinase (RTKs) signaling but also decreases the duration of activity . However, the mechanism by which SPRY2 inhibits RTK-signaling is poorly understood. We performed RNA-sequencing at six time points on cell lysates after stimulation with Fibroblast Growth Factor (FGF) in Sprouty-null Mouse Embryonic Fibroblasts (MEFs) and wild-type controls. We trained a hidden Markov model on a transcription-factor/DNA binding database , and used it to predict the transcription factors responsible for bifurcation points where clusters of time-course transcriptional profiles diverge. The differential activity of transcription factors informed how the signaling network is rewired in the absence of SPRY2. Background: SPRY2 is down-regulated in many types of cancers including those of the colon and breast . Understanding the mechanism of SPRY2 in signal transduction can lead to new therapies for cancer. Materials and Methods: We quantified the expression of gene transcripts using RNA-seq at six time points in SPRY-knockout vs. wild-type MEF cells. A refined version of the Dynamic Regulatory Events Miner (DREM) algorithm  was used with a protein-DNA binding database to infer the most likely active transcription factors at each time point. Differences in edge confidences between signaling networks inferred from the changes in activity of transcription factors were used as the basis to infer the location of SPRY2 within the signaling network. Results and Discussion: We found SPRY2 increased binding of the ubiquitin ligase, CBL-B, to several RTKs including FGFR. This suggests that the Sprouty-proteins locate ubiquitin ligases to the cellular membrane, thereby marking the receptors for degradation. Conclusions: Hidden Markov models are an efficient means to infer the mechanism of proteins from time-course gene expression profiles. This study suggests the importance of the regulation of RTK-degradation in the etiology of cancer and highlights the importance of incorporating regulation of protein stability into computational models of cancer.