TL;DR
This paper introduces a novel theoretical framework to disentangle "stable" (persistent) and "unstable" (transient) sources of variation in gene expression levels within cell populations. By applying this framework to T cell receptor (TCR) expression, they reveal that genetic diversity contributes significantly to stable variation in polyclonal T cells, while even genetically identical T cells exhibit substantial transient variation.
What Problem Does This Paper Address?
Understanding the sources of variation in gene expression is fundamental to biology. While much research has focused on "noise" (stochastic fluctuations) in gene expression, it's been unclear how much of the observed differences in expression levels between cells are truly transient versus persistent. This is particularly relevant for multicellular organisms where stable differences can arise from differentiation, cell lineages, or even intraclonal structure. Existing methods often struggle to quantitatively distinguish these sources, relying on qualitative comparisons of histograms. This paper aims to provide a rigorous quantitative framework to dissect these stable and unstable components of expression variation.
Methods
The core of this work is a theoretical framework that models cell populations as mixtures of subpopulations, each with a characteristic mean expression level. They derive a model where the total variance in expression can be decomposed into a "stable" component (due to differences in mean expression between subpopulations) and an "unstable" component (due to within-subpopulation noise).
Key methodological contributions include:
- Theoretical Framework: Development of a quantitative model that partitions expression variance into stable ($R^2$) and unstable components.
- Estimation Strategy: A method to estimate $R^2$ and the characteristic timescale of unstable dynamics ($\tau_\tau$) by following the temporal evolution of the mean expression levels of isolated cell cohorts (e.g., high and low expressors sorted via FACS). This method relies on the observation that the difference in means between two cohorts decays exponentially over time.
- Stochastic Model: A detailed stochastic model for protein expression within a single cell, incorporating production rate fluctuations and degradation, which forms the basis for simulations and theoretical derivations.
Datasets
The paper uses experimental data from two types of mouse CD4+ T cell populations:
- Polyclonal Population: Derived from wild-type mice, exhibiting genetic diversity in TCR loci due to V(D)J recombination.
- Monoclonal Population: Derived from TCR-transgenic mice (Marilyn strain), where all T cells express the same TCR, serving as a genetically uniform control.
No new benchmark datasets were introduced; rather, existing experimental techniques (flow cytometry) were used to generate data for analysis with the novel framework.
Key Insights and Ideas
- Stable vs. Unstable Variation: The framework provides a clear distinction between persistent differences (stable) and transient fluctuations (unstable) in gene expression.
- Genetic Diversity Drives Stability: In polyclonal T cell populations, genetic diversity is a significant contributor to stable differences in TCR expression levels.
- Surprising Stability in Isogenic Cells: Even in genetically identical TCR-transgenic T cells, a substantial portion of TCR expression variation is surprisingly stable, suggesting non-genetic mechanisms (e.g., epigenetic modifications) play a role.
- Universal Timescale: The characteristic timescale for the unstable dynamics of TCR expression was found to be approximately 32 hours in both polyclonal and monoclonal populations, suggesting a conserved underlying molecular process.
- Robustness of Method: The method for estimating $R^2$ is robust to typical day-to-day variations in flow cytometry measurements and to random measurement errors, as it relies on differences in means.
Key Results
- Polyclonal T cells: Approximately 70% of TCR expression variance is attributed to the stable component ($R^2 \approx 71\%$).
- Monoclonal T cells: Approximately 80% of TCR expression variance is attributed to the unstable component ($R^2 \approx 17\%$).
- Characteristic Timescale ($\tau_\tau$): For both populations, the timescale of unstable dynamics was estimated to be around 32 hours.
Limitations and Future Work
- Model Assumptions: The framework is tailored for long-tailed, unimodal expression distributions and assumes independent subpopulations at fixed density. It may require adaptation for multimodal distributions or more complex cell flow dynamics.
- Experimental Design: The accuracy of the method relies heavily on an adequate experimental design that ensures independent measurements for sorting and quantification to avoid biases from regression to the mean.
- Mechanistic Interpretation: While the framework quantifies stable and unstable components, the precise molecular mechanisms underlying the stable variation in isogenic cells (beyond genetic variation) remain to be fully elucidated. The authors speculate about epigenetic modifications.
- Future Directions: The authors suggest extending the framework to analyze multimodal distributions, incorporate gene regulatory networks, and apply it to analyze quantitative traits and responses in heterogeneous cell populations. They also note that cell division can affect the timescale of fluctuations and that further work could explore the impact of cell division on $R$ and $\tau_\tau$.
Summary updated