Towards control of cellular decision-making networks in the epithelial-to-mesenchymal transition

Abstract

We present the epithelial-to-mesenchymal transition (EMT) from two perspectives: experimental/technological and theoretical. We review the state of the current understanding of the regulatory networks that underlie EMT in three physiological contexts: embryonic development, wound healing, and metastasis. We describe the existing experimental systems and manipulations used to better understand the molecular participants and factors that influence EMT and metastasis. We review the mathematical models of the regulatory networks involved in EMT, with a particular emphasis on the network motifs (such as coupled feedback loops) that can generate intermediate hybrid states between the epithelial and mesenchymal states. Ultimately, the understanding gained about these networks should be translated into methods to control phenotypic outcomes, especially in the context of cancer therapeutic strategies. We present emerging theories of how to drive the dynamics of a network toward a desired dynamical attractor (e.g. an epithelial cell state) and emerging synthetic biology technologies to monitor and control the state of cells.

Figure 1:. The TGFβ-induced EMT/metastasis network includes a core regulatory network as well as multiple cross-talks and feedback loops.

TGFβ mediated EMT drives the transcription factors SNAIL1 and TWIST to induce the first layer of the core regulatory network. ZEB1 is then induced by SNAIL1 and TWIST, which jointly leads to loss of E-cadherin expression, increased expression of N-cadherin (mesenchymal marker), and an increased expression of matrix metalloproteinases (MMPs), which promote invasiveness and metastasis. MicroRNAs act to repress these transcription factors for preventing EMT, which are themselves repressed by SNAIL1 and ZEB1. The MAP Kinase pathway, mediated by tyrosine receptor kinases (TRK), interacts with the core network to indirectly increase SNAIL1 expression. This pathway also participates in a feedback loop with BACH1 and RKIP. Circles represent proteins and transcription factors. MicroRNAs are indicated by stem-loops. Connecting edges between intermediates indicate co-facilitation of an interaction while dashed interactions are composed of multiple sequential interactions. The MAPK pathway is designated by the dashed box in the cytoplasm.

Figure 2:. In vitro models investigate various steps of the metastatic cascade.

Cellular dissociation from the primary tumor and subsequent migration can be studied through adhesion assays, and migratory assays such as wound healing and zone exclusion assays. Degradation of the basement membrane as well as translocation across the ECM is readily recapitulated in 3D cultures. Intravasation and extravasation, wherein cells must squeeze through confined spaces such as gaps in endothelial cell walls, and circulation lends itself to study through microfluidic models. The growth of cancer cells at secondary sites may be recapitulated through 3D cultures such as spheroids.

Figure 3.. Synthetic biological EMT modifiers.

(A) Schematic illustration of a genetic engineering tool example (e.g. Cas9) and outcomes for knocking out or knocking in a specific EMT gene of interest. (B) Biological schematic of the core regulatory network controlling epithelial-mesenchymal transition. Network control via perturbing each toggle switch as critical areas with synthetic microRNAs or anti-sense oligonucleotides. (C) Direct perturbation of an endogenous EMT gene locus with transcriptional tools such as dCas9-VP64 or dCas9-KRAB.

Figure 4.. Network modeling of the epithelial-to-mesenchymal transition (EMT).

(A) Simplified version of the EMT network model of Steinway et al. The network includes multiple pathways, TFs, and crosstalks that are known to be involved in EMT and are used in multiple mathematical models reviewed here. (B) The model has an epithelial E and a mesenchymal M stable steady state in the absence of external signals. The colored valleys illustrate the attractor landscape. (C, D) External signals or attractor control interventions such as stable motif (SM) control or feedback vertex set (FVS) control, described in detail in the next section and Figure 5, can modify the attractor landscape so that only the E or M state is a stable steady state. (E) Hybrid epithelial/mesenchymal states arise in the model under certain perturbations (e.g. combined TGFβ induction and SMAD KO). (F) A signaling pathway can underlie the EMT decision making process through a noncanonical TF (blue edges and node outlines) and bypassing the canonical (known) TF core.

Figure 5.. Model-dependent and mode-independent control strategies using stable motif control and feedback vertex set control.

Stable motif (SM) control and feedback vertex set (FVS) control are attractor-control strategies that can steer the dynamics of the system towards one of its dynamical attractors (e.g. steady states). (A) Example network topology underlying a continuous or discrete mathematical model. Due to the multiple feedback loops in this network, it supports multistability. (B-C) The FVS of a network (panel B) is a set of nodes that, if removed, break all feedback loops of the network (panel C). (D) SM control makes model-dependent predictions; it uses the parameters of a model to determine which feedback loops are sufficient for steering the model towards any of its attractors. FVS control makes model-independent predictions; it uses solely the topology of a model to determine nodes (the FVS) that can steer any model that shares the topology towards any if its attractors. In the representation of the attractors yellow stand for a high value of the state variable and blue indicates a low value of the state variable.

Figure 6.. Synthetic Biology Detectors.

(A) Schematic illustration of gene circuit integrated in a cell population of interest allowing lineage tracking of cellular profiles among a population of heterogeneous cell types that are isogenic. (B) Schematic illustrations for higher order gene circuits that could be recruited to serve as sensors for multiple EMT markers to initiate detection and actuation in real time. (C) Schematic illustration of an example gene circuit that could allow robust and wide dynamic control of gene expression over a wide range of stimulus to juxtapose ON/OFF states of overexpression/knockout. (D) Schematic illustration of a synthetic biology gene circuit that can be used to observe controlled switching between epithelial and mesenchymal states by overexpressing targets that promote EMT in one state and targets that inhibit EMT in the alternative state.

Figure 7.. Synthetic Biology Actuators.

(A) Schematic illustration of genetic circuit library that could serve as sensors for various EMT genes in singlet and respond with appropriate response output (e.g. inducing apoptosis genes). (B) Schematic illustration of a synthetic biology gene circuit such as a decoder that could analyze the spectrum of cellular states relating to EMT to contrast classifications of single markers. (C) Schematic illustration of gene circuit controlling spatial induction of EMT to study how spatial parameters affect the development of transition between epithelial and mesenchymal cell types. Such a system could be accomplished with a digital mirror device (DMD). (D) Schematic illustration of immunoengineering as an outlet for post- EMT control by killing of mesenchymal cells by engineered T cells.