Revealing cell signaling history with ERK-aeology
Summary by Allison Saul: Ram, A., Pargett, M., Choi, Y., Murphy, D., Cabel, M., Kosaisawe, N., Quon, G., & Albeck, J. (2024). Deciphering the History of ERK Activity from Fixed-Cell Immunofluorescence Measurements. bioRxiv. 2024.02.16.580760. https://doi.org/10.1101/2024.02.16.580760
Image credit: Midjourney
Signaling pathways, such as ERK/Ras, play critical roles in cell cycle decisions, movement, and differentiation by regulating gene expression. Misregulation of ERK signaling drives a variety of diseases, including cancers, where disruptions in signaling affect cell cycle regulation and proliferation [1]. Diagnostic tests to detect the pathway’s signaling effector, phospho-ERK, in fixed biopsy samples can identify disease-related signaling perturbations; however, simply measuring pERK at a single point in time does not reveal the history of ERK signaling. Clinically, understanding historical ERK dynamics could potentially inform the management and treatment of tumors driven by ERK misregulation and identify points of drug resistance, as ERK signaling dynamics drive differential gene expression.
Here, Ram et. al. uses computational modeling and pathway target protein immunostaining to infer the history of ERK signaling in fixed cells [2]. Additionally, they test the limits of their model using mathematical simulations of ERK-driven gene expression, and identify target protein predictors of long- and short-term ERK signaling.
First, the group created a dataset for training computational models. Using a live biosensor, they quantified ERK signaling histories, then assessed target protein expression at a single end time point. To vary ERK levels and dynamics, they exposed cells to an Epidermal Growth Factor (EGF) dose curve. To achieve temporal diversity in dynamics, a pathway inhibitor (MEKi) was added at specific timepoints throughout the experiment. Finally, immunofluorescence was performed, measuring protein from eight different downstream target genes of ERK, and a heatmap was generated to examine the relationship between signaling history and protein expression.
Next, Ram. et. al. performed a series of statistical modeling with this dataset. Regression modeling revealed that the duration of ERK signaling appeared to have a stronger influence on protein expression than intensity. Multiple linear regression modeling demonstrated oncogene proteins Fra-1 and pRb most accurately inferred ERK intensity and duration, whereas Egr-1 and pRb best represented ERK’s temporal dynamics. Furthermore, modeling indicated that pERK intensity at a single time point was not an accurate predictor of long-term ERK signaling history on a single-cell scale, particularly after treatment with MEKi. Consequently, the authors suggest using alternative markers such as pRb in addition to pERK to assess pathway activity history.
Finally, they trained a convolutional neural network (CNN) on their live-cell dataset. CNNs break down information, analyzes it for patterns, and creates trend predictions [3]. Ram. et. al.’s CNN model identifies the patterns of ERK signaling in a non-linear manner within fixed biopsy samples [2]. Combined with linear regression modeling, the complexities of the relationship between ERK signaling history and target protein levels can be predicted from a single biopsy.
Using computational approaches to predict signaling dynamics could be applied to questions investigating the effects of ERK signaling on fate decisions, improve clinical diagnostic techniques, and suggests a method for creating computational models of other signaling pathways relevant to disease. A broader view of historical signaling in other pathways implicated in disease provides the opportunity to improve personalized therapeutic treatments.
1. Sugiura, R., Satoh, R., & Takasaki, T. (2021). ERK: A Double-Edged Sword in Cancer. ERK-Dependent Apoptosis as a Potential Therapeutic Strategy for Cancer. Cells, 10(10), 2509. https://doi.org/10.3390/cells10102509
2. Ram, A., Pargett, M., Choi, Y., Murphy, D., Cabel, M., Kosaisawe, N., Quon, G., & Albeck, J. (2024). Deciphering the History of ERK Activity from Fixed-Cell Immunofluorescence Measurements. bioRxiv : the preprint server for biology, 2024.02.16.580760. https://doi.org/10.1101/2024.02.16.580760
3. Sarıgül, M., Ozyildirim, B. M., & Avci, M. (2019). Differential convolutional neural network. Neural networks : the official journal of the International Neural Network Society, 116, 279–287. https://doi.org/10.1016/j.neunet.2019.04.025