Population Dynamics of Epithelial-Mesenchymal Heterogeneity in Cancer Cells

Abstract

Phenotypic heterogeneity is a hallmark of aggressive cancer behaviour and a clinical challenge. Despite much characterisation of this heterogeneity at a multi-omics level in many cancers, we have a limited understanding of how this heterogeneity emerges spontaneously in an isogenic cell population. Some longitudinal observations of dynamics in epithelial-mesenchymal heterogeneity, a canonical example of phenotypic heterogeneity, have offered us opportunities to quantify the rates of phenotypic switching that may drive such heterogeneity. Here, we offer a mathematical modeling framework that explains the salient features of population dynamics noted in PMC42-LA cells: (a) predominance of EpCAMhigh subpopulation, (b) re-establishment of parental distributions from the EpCAMhigh and EpCAMlow subpopulations, and (c) enhanced heterogeneity in clonal populations established from individual cells. Our framework proposes that fluctuations or noise in content duplication and partitioning of SNAIL—an EMT-inducing transcription factor—during cell division can explain spontaneous phenotypic switching and consequent dynamic heterogeneity in PMC42-LA cells observed experimentally at both single-cell and bulk level analysis. Together, we propose that asymmetric cell division can be a potential mechanism for phenotypic heterogeneity.

Epigenetic memory acquired during long-term EMT induction governs the recovery to the epithelial state

“…The degree of resolution among distinct cell-states depends on the number of biomarkers used experimentally to identify a cell population [5,6,18,44,45]. These cell-states can transition between each other either spontaneously due to factors such as stochastic gene expression and asymmetric cell division, or under microenvironmental influence such as TGFβ signalling and altered matrix stiffness [28,39,[46][47][48]. Recently, frequency of spontaneous cell-state transition has been shown to depend on the mRNA and protein half-life.…”

Population Dynamics of EMT Elucidates the Timing and Distribution of Phenotypic Intra-tumoral Heterogeneity

“…Our approach, while useful for robustly characterizing EMT at the single-cell level, is not without limitation. Phenotypically heterogeneous cells can exhibit gene expression profiles that cooperate or interfere with overall signal detection 1 , and this is further compounded by noisy celular division 21 . Although we have assumed in our analysis that populations under study were non-dividing, cell cycle scoring suggested the variable presence of division signatures across available experimental contexts as shown in Figure 6.…”

“…Notably, not all cells are responsive to the EMT-inducing signal resulting 75 in considerable phenotypic intra-tumoral heterogeneity and coexistence amongst multiple states 66 (Figure 1B). This has been reinforced by experimental observations supporting the existence of phenotypic transitions amongst EMT states as well as model-driven predictions of phenotypic heterogene-80 ity generated by noisy cell division 21,58 . Furthermore, the environmental cues that govern these cell state transitions are highly context-specific 43,62 .…”

“…This assumption is supported by the fact that CEACAM6, an inducer of EMT 6 , is the most variably expressed gene at day 0 for A549 across all treatment cases (Figures SI. [20][21][22]. For all cases, the second regime of the Markov chain was fitted from the day when the hybrid state 395 peaked to the day of treatment withdrawal (day 7).…”

Epigenetic memory acquired during long-term EMT induction governs the recovery to the epithelial state

“…The degree of resolution among distinct cell-states depends on the number of biomarkers used experimentally to identify a cell population (Bhatia et al, 2019a; Karacosta et al, 2019; Pastushenko et al, 2018; Pillai and Jolly, 2021; Ruscetti et al, 2016). These cell-states can transition between each other either spontaneously due to factors such as stochastic gene expression and asymmetric cell division, or under microenvironmental influence such as TGFβ signaling and altered matrix stiffness (Cook and Vanderhyden, 2020; Jain et al, 2022; Matte et al, 2019; Tripathi et al, 2020a; Zhao et al, 2021). The relative rates of cell-state transitions define the population distribution of cells along the E-M axis, as evident from dynamics of isolated subpopulations in vitro and in vivo (Bhatia et al, 2019b; Pastushenko et al, 2018; Yamamoto et al, 2017).…”