Computational systems biology of epithelial-hybrid-mesenchymal transitions

https://doi.org/10.1016/j.coisb.2017.02.004Get rights and content

Highlights

  • EMT and MET are not binary; cells can attain hybrid epithelial/mesenchymal state(s).

  • Mechanism-based and data-based computational models can quantify the EMT/MET spectrum.

  • These models can also investigate a role of EMT in other traits such as stemness.

Abstract

Metastasis accounts for more than 90% of cancer-related deaths, and is fueled by fine-tuned transitions among many cellular phenotypes. Transitions among epithelial (strong cell–cell adhesion, no or little migration), mesenchymal (no cell–cell adhesion, high migration), and hybrid epithelial/mesenchymal (both cell–cell adhesion and cell migration) phenotypes are considered to be a hallmark of metastasis. Recent years have witnessed rapid progress in mapping the regulatory networks underlying these transitions. This progress has enabled the capability to develop computational systems biology models to characterize how various intracellular and extracellular signals can drive these transitions. Here, we discuss how different mathematical models have contributed to elucidating the underlying principles of these transitions and guided further experiments to address key unanswered questions concerning metastasis.

Introduction

Metastasis causes above 90% of all cancer-related deaths. Metastasis involves primary tumor cells leaving the home organ, entering the bloodstream, disseminating throughout the body, and forming secondary tumors (metastases) at distant organs [1]. The ‘metastatic cascade’ is extremely challenging for cancer cells. Therefore, only 0.01% of cells entering the circulation are estimated to be able to form metastases [2].

An intriguing aspect of metastasis emerges from the fact that more than 80% of cancers are carcinomas, i.e. cancers beginning in epithelial organs such as breast, prostate, and lung. Carcinoma cells adhere tightly to their neighbors in highly organized 3-D structures and lack the innate ability to invade surrounding tissue. Standard thinking in the field suggests that to metastasize, the cells shed some epithelial traits of cell polarity and E-cadherin based cell–cell adhesion, and pick up mesenchymal features of migration and invasion 3, 4, 5. After reaching the distant organ, these migratory cells often lose migration and regain cell–cell adhesion, reverting to an epithelial phenotype. Thus, in many cases, these reversible bidirectional transitions among epithelial and mesenchymal phenotypes – EMT (Epithelial–Mesenchymal Transition) and its reverse MET (Mesenchymal–Epithelial Transition) – form the cornerstone of cancer metastasis [6]. We would be remiss in not mentioning the raging controversy within the EMT field regarding recent claims that EMT is dispensable for metastasis but required for emergence of drug resistance 7, 8. These recent reports have raised important issues such as whether these results hold true only for specific genetically engineered mouse models (GEMMs), whether the markers used are truly indicative of EMT, and whether the knockdown of one transcription factor ablated EMT fully 2, 9.

Recent experimental and computational studies have suggested that these transitions are rarely ‘all-or-none’, instead cells can exhibit a spectrum of intermediate or hybrid epithelial/mesenchymal (E/M) phenotypes ∗∗10, 11, ∗∗12, ∗∗13, 14, ∗15, 16, 17. Such hybrid E/M cells can both adhere to their neighbors and migrate, thereby leading to tumor budding and/or clustered migration of Circulating Tumor Cells (CTCs) 18, 19. These clusters, although quite rare, can form up to 50-times more secondary tumors as compared to individually migrating CTCs, suggesting the enhanced metastatic potential of a hybrid E/M phenotype ∗∗20, 21. Furthermore, the presence of clusters in patients can predict poor survival [20]. Therefore, understanding how cells transition among epithelial, mesenchymal and hybrid E/M phenotypes can offer novel insights into halting metastatic progression.

Section snippets

Regulatory networks underlying EMT/MET

EMT can be induced by many signaling pathways such as TGFβ, HGF, EGF, FGF, Wnt, Notch, NF-kB, Hedgehog, JAK/STAT, and Hippo [22], hypoxia [23], and mechanical factors such as extracellular matrix (ECM) density [24]. These signals often activate one or more of EMT-inducing transcription factors (EMT-TFs) such as ZEB1/2, SNAIL1/2, TWIST1, Goosecoid – all of which can, directly or indirectly, repress epithelial genes including E-cadherin – the inhibition of which is considered as a hallmark of EMT

Mechanism-driven ‘bottom-up’ models of EMT

Mechanism-based ‘bottom-up’ models represent experimentally identified interactions among a set of small number of core players that regulate EMT/MET, and characterize the emergent dynamics of the regulatory network. The first two ‘bottom-up’ attempts to model EMT focus on interactions between two microRNA families miR-34, miR-200 and two EMT-TF families ZEB and SNAIL ∗∗10, 11. These small-scale models capture the kinetics of individual reactions in the network, i.e. similar to the

Data-driven ‘top-down’ models of EMT

The data-driven ‘top-down’ models start with omics level data – typically gene expression – and apply various statistical techniques to identify patterns and/or trajectories for these transitions. For instance, Zadran et al. [46] characterize the changes in gene expression profiles for TGFβ1-induced EMT in A549 lung cancer cells on a free energy landscape to predict that EMT proceeds via a stable intermediate state that may correlate with metabolic shifts. Similarly, Chang et al. [47] collected

Conclusion

As discussed above, computational dynamical and statistical models of EMT have contributed significantly by providing quantitative frameworks to characterize the landscape of genetic and biophysical changes associated with EMT in multiple contexts. They have highlighted that EMT is rarely an ‘all-or-none’ process, and yielded specific falsifiable hypotheses that can be tested experimentally to understand regulation of EMT in a more nuanced way. Furthermore, they have provided tractable schemas

Acknowledgements

This work was supported by National Science Foundation (NSF) Center for Theoretical Biological Physics (NSF PHY-1427654) and NSF DMS-1361411.

References (60)

  • X. Zheng et al.

    Epithelial-to-mesenchymal transition is dispensable for metastasis but induces chemoresistance in pancreatic cancer

    Nature

    (2015)
  • K.R. Fischer et al.

    Epithelial-to-mesenchymal transition is not required for lung metastasis but contributes to chemoresistance

    Nature

    (2015)
  • W. Li et al.

    Probing the fifty shades of EMT in metastasis

    Trends Cancer

    (2016)
  • M. Lu et al.

    MicroRNA-based regulation of epithelial-hybrid-mesenchymal fate determination

    Proc Natl Acad Sci USA

    (2013)
  • X.-J. Tian et al.

    Coupled reversible and irreversible bistable switches underlying TGFβ-induced epithelial to mesenchymal transition

    Biophys J

    (2013)
  • M.K. Jolly et al.

    Stability of the hybrid epithelial/mesenchymal phenotype

    Oncotarget

    (2016)
  • S.N. Steinway et al.

    Combinatorial interventions inhibit TGFβ-driven epithelial-to-mesenchymal transition and support hybrid cellular phenotypes

    npj Syst Biol Appl

    (2015)
  • M. Yu et al.

    Circulating breast tumor cells exhibit dynamic changes in epithelial and mesenchymal composition

    Science

    (2013)
  • F. Andriani et al.

    Conversion to stem-cell state in response to microenvironmental cues is regulated by balance between epithelial and mesenchymal features in lung cancer cells

    Mol Oncol

    (2016)
  • A.J. Armstrong et al.

    Circulating tumor cells from patients with advanced prostate and breast cancer display both epithelial and mesenchymal markers

    Mol Cancer Res

    (2011)
  • A. Grigore et al.

    Tumor budding: the name is EMT. Partial EMT

    J Clin Med

    (2016)
  • M.K. Jolly et al.

    Implications of the hybrid epithelial/mesenchymal phenotype in metastasis

    Front Oncol

    (2015)
  • K.J. Cheung et al.

    Polyclonal breast cancer metastases arise from collective dissemination of keratin 14-expressing tumor cell clusters

    Proc Natl Acad Sci USA

    (2016)
  • B. De Craene et al.

    Regulatory networks defining EMT during cancer initiation and progression

    Nat Rev Cancer

    (2013)
  • K. Lundgren et al.

    Hypoxia, snail and incomplete epithelial–mesenchymal transition in breast cancer

    Br J Cancer

    (2009)
  • J.A. Somarelli et al.

    Mesenchymal–epithelial transition in sarcomas is controlled by the combinatorial expression of miR-200s and GRHL2

    Mol Cell Biol

    (2016)
  • B.-T. Preca et al.

    A self-reinforcing CD44s/ZEB1 feedback loop maintains EMT and stemness properties in cancer cells

    Int J Cancer

    (2015)
  • U. Burk et al.

    A reciprocal repression between ZEB1 and members of the miR-200 family promotes EMT and invasion in cancer cells

    EMBO Rep

    (2008)
  • M. Lu et al.

    Tristability in cancer-associated MicroRNA-TF chimera toggle switch

    J Phys Chem B

    (2013)
  • S.-M.M. Park et al.

    The miR-200 family determines the epithelial phenotype of cancer cells by targeting the E-cadherin repressors ZEB1 and ZEB2

    Genes Dev

    (2008)
  • Cited by (25)

    • The fundamentals of phenotypic plasticity

      2020, Phenotypic Switching: Implications in Biology and Medicine
    • Hybrid epithelial/mesenchymal phenotypes promote metastasis and therapy resistance across carcinomas

      2019, Pharmacology and Therapeutics
      Citation Excerpt :

      A cell line derived from this mouse model also contained all 3 subpopulations – E, M, and hybrid E/M – as revealed by flow cytometry analysis (Ruscetti et al., 2016). The subpopulations within the cell line also spontaneously interconverted between the three states, validating the mathematical models that predict the existence of multi-stability in EMT networks and consequent phenotypic switching, even in the absence of an EMT/MET inducer (Burger et al., 2017; Jolly & Levine, 2017). In clinical samples, a hybrid E/M phenotype was reported in prostate cancer CTCs metastases from men with metastatic castration-resistant prostate cancer (CRPC) co-expressing epithelial and mesenchymal biomarkers such as EpCAM, cytokeratins, E-cadherin, N-cadherin, and vimentin (Armstrong et al., 2011).

    • Discrete dynamic network modeling of oncogenic signaling: Mechanistic insights for personalized treatment of cancer

      2018, Current Opinion in Systems Biology
      Citation Excerpt :

      Evasion of cell death and uncontrolled growth (and how to prevent them) are the most studied outcomes in models of oncogenic signaling, yet the interplay of oncogenic signaling in invasion and metastasis is equally important and less understood. Several groups have recently developed mathematical models of the biological processes underlying cancer cell invasion and metastasis [14,22,23,56,57,61–63], with a particular emphasis on the epithelial-to-mesenchymal transition (EMT), a developmental process hijacked by cancer that is thought to be necessary for the initiation of metastatic spreading. On the discrete network modeling side, one of the initial efforts was by Steinway et al., who constructed a model of EMT in the context of hepatocellular carcinoma [56,62].

    • Epigenetic factor competition reshapes the EMT landscape

      2022, Proceedings of the National Academy of Sciences of the United States of America
    View all citing articles on Scopus
    View full text