Abstract
The RAS/ERK pathway plays a central role in diagnosis and therapy for many cancers. ERK activity is highly dynamic within individual cells and drives cell proliferation, metabolism, and other processes through effector proteins including c-Myc, c-Fos, Fra-1, and Egr-1. These proteins are sensitive to the dynamics of ERK activity, but it is not clear to what extent the pattern of ERK activity in an individual cell determines effector protein expression, or how much information about ERK dynamics is embedded in the pattern of effector expression. Here, we evaluate these relationships using live-cell biosensor measurements of ERK activity, multiplexed with immunofluorescence staining for downstream target proteins of the pathway. Combining these datasets with linear regression, machine learning, and differential equation models, we develop an interpretive framework for immunofluorescence data, wherein Fra-1 and pRb levels imply long-term activation of ERK signaling, while Egr-1 and c-Myc indicate more recent activation. Analysis of multiple cancer cell lines reveals a distorted relationship between ERK activity and cell state in malignant cells. We show that this framework can infer various classes of ERK dynamics from effector protein stains within a heterogeneous population, providing a basis for annotating ERK dynamics within fixed cells.
Introduction
The RAS/ERK signaling pathway is a master regulator of cell fate, governing decisions such as proliferation, differentiation, and apoptosis. Rather than acting as a simple binary switch, ERK activity is characterized by complex temporal dynamics—pulses of varying frequency and duration that encode specific biological instructions. While live-cell biosensors can capture these dynamics in real-time, their application is limited in clinical and diagnostic settings, such as tumor biopsies, which rely on fixed-cell immunofluorescence (IF) snapshots. These snapshots provide a static view of protein levels but lack the temporal context of the signaling history. This study aims to bridge this gap by developing a "Rosetta Stone" to translate static protein levels back into the dynamic signaling history that created them.
Methods and Experimental Design
To map signaling history to protein states, the researchers developed a sophisticated experimental pipeline:
- Live-Cell Tracking: Using the FRET-based biosensor EKAR3.1, ERK activity was monitored in over 97,000 individual MCF10A (non-tumorigenic mammary epithelial) cells over a 19-hour period.
- Dynamic Stimulation: Cells were exposed to varying doses of epidermal growth factor (EGF) and MEK inhibitors (MEKi) to generate a diverse library of ERK activity "melodies."
- Multiplexed Staining (4i): Immediately following live imaging, cells were fixed and analyzed using iterative indirect immunofluorescence imaging (4i) to measure the levels of up to 19 protein targets in the exact same cells.
- Computational Modeling: The paired dataset was used to train linear regression models, convolutional neural networks (CNNs), and ordinary differential equation (ODE) models to find correlations and capture complex, non-linear relationships between ERK time-series and protein output.
Results: Decoding the Cellular Memory
The study identified that downstream effector proteins act as integrators of ERK activity over distinct timescales, effectively serving as cellular "memory" markers.
- Temporal Memory Spans: Proteins exhibit different "memory spans." Fra-1 and pRb were found to be long-term markers, reflecting integrated ERK activity over 12 or more hours. In contrast, Egr-1 and pERK act as short-term responders, reflecting activity within the 30–60 minutes prior to fixation.
- Non-Linear Decoding: CNN models significantly outperformed linear models for proteins like Egr-1 and pRb, suggesting that cells employ complex, non-linear decoding mechanisms for ERK temporal patterns rather than simple summation.
- Reverse Inference of Signaling History: Using a small panel of protein markers, the researchers could predict the cumulative duration of ERK activity with an R² of 0.53 at the single-cell level. A classification model successfully categorized cells into five prototypical ERK history classes (e.g., "recently activated" vs. "consistently low") with approximately 80% accuracy when considering local cell neighborhoods.
Malignancy and Signal Distortion
Analysis of cancer cell lines, including HCC827 (lung cancer), A549 (lung cancer), and MCF7 (breast cancer), revealed that the tight coupling between ERK activity and downstream targets observed in healthy cells is often distorted in malignancy. For instance, the cell cycle regulator pRb, which typically tracks ERK history closely, became insensitive to ERK inhibition in several cancer lines, indicating a decoupling of the signaling pathway from its typical physiological outputs in malignant states.
Discussion and Future Directions
This framework provides a powerful tool for annotating ERK dynamics within fixed cells, offering a path toward more informative clinical diagnostics. While the current models explain a significant portion of protein variance, remaining noise likely stems from unmeasured crosstalk with other pathways, such as PI3K or JAK/STAT. Future directions include validating this approach in human tissue biopsies and integrating additional markers or spatial transcriptomics (e.g., MERFISH) to enhance the resolution and predictive power of inferred signaling histories in complex tumor microenvironments.
Summary updated