Intracellular nanovesicles mediate α5ß1 integrin trafficking during cell migration.

Membrane traffic is an important regulator of cell migration through the endocytosis and recycling of cell surface receptors such as integrin heterodimers. Intracellular nanovesicles (INVs) are transport vesicles that are involved in multiple membrane trafficking steps, including the recycling pathway. The only known marker for INVs is tumor protein D54 (TPD54/TPD52L2), a member of the TPD52-like protein family. Overexpression of TPD52-like family proteins in cancer has been linked to poor prognosis and an aggressive metastatic phenotype, which suggests cell migration may be altered under these conditions. Here, we show that TPD54 directly binds membrane and associates with INVs via a conserved positively charged motif in its C terminus. We describe how other TPD52-like proteins are also associated with INVs, and we document the Rab GTPase complement of all INVs. Depletion of TPD52-like proteins inhibits cell migration and invasion, while their overexpression boosts motility. We show that inhibition of migration is likely due to altered recycling of α5ß1 integrins in INVs.

Combinatorial mathematical modelling approaches to interrogate rear retraction dynamics in 3D cell migration.

Cell migration in 3D microenvironments is a complex process which depends on the coordinated activity of leading edge protrusive force and rear retraction in a push-pull mechanism. While the potentiation of protrusions has been widely studied, the precise signalling and mechanical events that lead to retraction of the cell rear are much less well understood, particularly in physiological 3D extra-cellular matrix (ECM). We previously discovered that rear retraction in fast moving cells is a highly dynamic process involving the precise spatiotemporal interplay of mechanosensing by caveolae and signalling through RhoA. To further interrogate the dynamics of rear retraction, we have adopted three distinct mathematical modelling approaches here based on (i) Boolean logic, (ii) deterministic kinetic ordinary differential equations (ODEs) and (iii) stochastic simulations. The aims of this multi-faceted approach are twofold: firstly to derive new biological insight into cell rear dynamics via generation of testable hypotheses and predictions; and secondly to compare and contrast the distinct modelling approaches when used to describe the same, relatively under-studied system. Overall, our modelling approaches complement each other, suggesting that such a multi-faceted approach is more informative than methods based on a single modelling technique to interrogate biological systems. Whilst Boolean logic was not able to fully recapitulate the complexity of rear retraction signalling, an ODE model could make plausible population level predictions. Stochastic simulations added a further level of complexity by accurately mimicking previous experimental findings and acting as a single cell simulator. Our approach highlighted the unanticipated role for CDK1 in rear retraction, a prediction we confirmed experimentally. Moreover, our models led to a novel prediction regarding the potential existence of a 'set point' in local stiffness gradients that promotes polarisation and rapid rear retraction.

Membrane Tension Orchestrates Rear Retraction in Matrix-Directed Cell Migration.

In development, wound healing, and cancer metastasis, vertebrate cells move through 3D interstitial matrix, responding to chemical and physical guidance cues. Protrusion at the cell front has been extensively studied, but the retraction phase of the migration cycle is not well understood. Here, we show that fast-moving cells guided by matrix cues establish positive feedback control of rear retraction by sensing membrane tension. We reveal a mechanism of rear retraction in 3D matrix and durotaxis controlled by caveolae, which form in response to low membrane tension at the cell rear. Caveolae activate RhoA-ROCK1/PKN2 signaling via the RhoA guanidine nucleotide exchange factor (GEF) Ect2 to control local F-actin organization and contractility in this subcellular region and promote translocation of the cell rear. A positive feedback loop between cytoskeletal signaling and membrane tension leads to rapid retraction to complete the migration cycle in fast-moving cells, providing directional memory to drive persistent cell migration in complex matrices.

Modeling Rho GTPase Dynamics Using Boolean Logic.

Rho GTPases such as the canonical Rac1 and RhoA are embedded within complex networks requiring the precise spatiotemporal balance of GEFs, GAPs, upstream regulators, growth factors, and downstream effectors. A modeling approach based on Boolean logical networks is becoming an increasingly relied-upon tool to harness this complexity and elucidate further details regarding Rho GTPase signaling. In this methods chapter we describe how to initially create appropriately sized networks based on literature evidence; formalize these networks with reactions based on Boolean logical operators; implement the network into appropriate simulation software (CellNetAnalyzer); and finally perform simulations and make novel, testable predictions via in silico knockouts. Given this predictive power, the Boolean approach may ultimately help to highlight potential future avenues of experimental research.

Rationalizing Rac1 and RhoA GTPase signaling: A mathematical approach.

Precise spatiotemporal dynamics of Rho GTPases are essential for efficient cell migration. Manipulating Rac1 and RhoA signaling is thus a potential intervention strategy to abrogate harmful cell invasion and subsequent metastasis; however GTPase signaling can be extremely complicated due to crosstalk and the multitude of upstream regulators and downstream effectors. Studying Rho GTPase networks in a formal mathematical setting can therefore be of great use. We recently built a predictive model based on Boolean logic which identified a negative feedback loop critical for RhoA and Rac1 activity. Here, we discuss the value and potential pitfalls of different mathematical approaches which have been used to study Rho GTPase dynamics, and highlight the importance of choosing the correct approach given the data available and outputs desired. Overall, a mathematical approach, particularly when combined iteratively with in vitro experiments, can be of great use in deriving new biological insight to further harness the activity of Rho GTPases.

Modelling GTPase dynamics to understand RhoA-driven cancer cell invasion.

Metastasis, initially driven by cells migrating and invading through the local environment, leads to most cancer-associated deaths. Cells can use a variety of modes to move in vitro, all of which depend on Rho GTPases at some level. While traditionally it was thought that Rac1 activity drives protrusive lamellipodia at the leading edge of a polarised cell while RhoA drives rear retraction, more recent work in 3D microenvironments has revealed a much more complicated picture of GTPase dynamics. In particular, RhoA activity can dominate the leading edge polymerisation of actin to form filopodial actin-spike protrusions that drive more invasive cell migration. We recently described a potential mechanism to abrogate this pro-invasive localised leading edge Rac1 to RhoA switch via manipulation of a negative feedback loop that was revealed by adopting a logical modelling approach. Both challenging dogma and taking a formal, mathematical approach to understanding signalling involved in motility may be vital to harnessing harmful cell migration and preventing metastasis in future research.

A MAPK-Driven Feedback Loop Suppresses Rac Activity to Promote RhoA-Driven Cancer Cell Invasion.

Cell migration in 3D microenvironments is fundamental to development, homeostasis and the pathobiology of diseases such as cancer. Rab-coupling protein (RCP) dependent co-trafficking of α5ß1 and EGFR1 promotes cancer cell invasion into fibronectin (FN) containing extracellular matrix (ECM), by potentiating EGFR1 signalling at the front of invasive cells. This promotes a switch in RhoGTPase signalling to inhibit Rac1 and activate a RhoA-ROCK-Formin homology domain-containing 3 (FHOD3) pathway and generate filopodial actin-spike protrusions which drive invasion. To further understand the signalling network that drives RCP-driven invasive migration, we generated a Boolean logical model based on existing network pathways/models, where each node can be interrogated by computational simulation. The model predicted an unanticipated feedback loop, whereby Raf/MEK/ERK signalling maintains suppression of Rac1 by inhibiting the Rac-activating Sos1-Eps8-Abi1 complex, allowing RhoA activity to predominate in invasive protrusions. MEK inhibition was sufficient to promote lamellipodia formation and oppose filopodial actin-spike formation, and led to activation of Rac and inactivation of RhoA at the leading edge of cells moving in 3D matrix. Furthermore, MEK inhibition abrogated RCP/α5ß1/EGFR1-driven invasive migration. However, upon knockdown of Eps8 (to suppress the Sos1-Abi1-Eps8 complex), MEK inhibition had no effect on RhoGTPase activity and did not oppose invasive migration, suggesting that MEK-ERK signalling suppresses the Rac-activating Sos1-Abi1-Eps8 complex to maintain RhoA activity and promote filopodial actin-spike formation and invasive migration. Our study highlights the predictive potential of mathematical modelling approaches, and demonstrates that a simple intervention (MEK-inhibition) could be of therapeutic benefit in preventing invasive migration and metastasis.