During cell migration, cells respond to environmental cues with a variety of morphological changes, including leading edge, or lamellipodial, extension, and retraction of the rear of the cell. To coordinate these behaviors in time and space, many signaling pathways are engaged, principal among them the Rho GTPases, which regulate the actin cytoskeleton, vesicular trafficking, and cellular polarity, among other functions (Burridge and Wennerberg 2004; Jaffe and Hall 2005; Ridley 2006). Biochemical assays, loss-of-function assays, and gain-of-function assays have all contributed to our understanding of the functions of these Rho GTPases, particularly the major family members RhoA, Rac1, and Cdc42. Each of these GTPases interacts with unique downstream effectors to alter cellular behaviors (Jaffe and Hall 2005), but how they are coordinated in both space and time to yield productive cellular movement has remained difficult to assess, except at a gross biochemical level. This difficulty arises in part because these GTPases must become active to initiate downstream signaling, and perhaps only 5% of any GTPase is active at any time within the cell. Further, the subcellular location of the GTPase alone does not reflect the location of the activity (Kraynov et al. 2000; Pertz et al. 2006). Since it is known that the location of the active form of these GTPases is critical for the resultant cellular behavior, it is important to correlate GTPase activation levels with cellular behaviors.
However, high-resolution measurements of multiple signaling activities simultaneously have been impaired by the spectral constraints of currently available biosensors, limiting imaging to one sensor per experiment. Through collaboration with the Danuser laboratory at Harvard, we developed two new approaches that allowed us to
resolve the coordination of the three Rho GTPases, RhoA, Rac1, and Cdc42, with submicron and seconds resolution. First, our collaborators developed a “computational multiplexing” technique, where the activities of individual Rho GTPases, determined in separate experiments, are related locally to cell morphological activities as a common reference. Using this methodology, we were able to show that despite the complexity and heterogeneity of protrusion behavior from cell to cell, the relationships between the GTPases were remarkably consistent and could be inferred indirectly by correlation analysis of multiple separate imaging experiments.
Our analyses show that RhoA is activated at the cell edge synchronous with edge advancement, whereas Cdc42 and Rac1 are activated 2m behind the cell edge with a delay of 40 s relative to initiation of protrusion. Thus, Rac1 and RhoA operate at the leading edge antagonistically through spatial separation and precise relative timing. RhoA plays a role in the initial events of protrusion, while Rac1 and Cdc42 signals activate pathways implicated in reinforcement and stabilization of newly expanded protrusions.
N.B.: For the journal article describing this part of the work, there were three co- first-authors, including myself. Thus, the data represented herein are a collaboration of our efforts which are not easily dissected. Dr. Louis Hodgson and I generated the imaging data, while all the computational work was performed on that data by Dr. Matthias Machacek. Where possible, I have used my own examples and data, but many of the graphics are compilations of data from our joint efforts. I have attempted to indicate specifically my contributions to this work rather than claim ownership of all data and data analysis in the following sections. Additionally, I generated an
intermolecular Cdc42 sensor for this paper, though the data I generated was not incorporated for technical reasons. As a consequence, none of the Cdc42 data herein is my work.
We also focus on a little known but widely expressed protein called TEM4. TEM4 (tumor endothelial marker 4) was identified as a gene whose expression was upregulated in endothelial cells during tumor-induced angiogenesis (St Croix et al. 2000) and is a member of the Dbl family of Rho guanine nucleotide exchange factors (RhoGEFs) (Rossman et al. 2005). Our collaborators have shown that TEM4 associates with microtubules and promotes microtubule-dependent FA disassembly by engaging the RhoC/mDia1 signaling pathway. TEM4 also regulates protrusion dynamics of the leading edge and directionality of cellular migration upstream of RhoC/mDia1. Here, we focus on a particular aspect of TEM4 regulation of RhoA activity. Curiously, even though TEM4 has catalytic activity toward all three Rho isoforms in vitro, knockdown of TEM4 expression results in an elevation of RhoA activity. To determine the nature of this RhoA activation, we visualized RhoA activity in live cells with or without TEM4, demonstrating that upon TEM4 knockdown, RhoA becomes active all along the cell periphery, rather than solely in protrusions, suggesting that TEM4, through its regulation of RhoC, suppresses RhoA activation at non-protruding regions of the cellular edge and contributes to microtubule mediated focal adhesion disassembly. Thus, we expand the role for RhoA at the leading edge by examining one of its regulators, TEM4.
Lastly, we examine the role of Cdc42 in the formation of another type of leading edge protrusional structure: filopodia. As has been stated above, actin-based structures are responsible for the majority of cytoskeletal changes observed in cells, including the
formation of filopodia, which contain tightly bundled parallel actin filaments (Davenport et al. 1993). These structures perform many functions, including a sensing function in neurons (Lendvai et al. 2000; Sabatini and Svoboda 2000), but their formation is not well understood. The bundled nature of the filopodia has given rise to a number of hypotheses about how these structures might form, and which proteins might be involved, such as the VASP family, formins, and N-WASP/Arp2/3, each of which is controversial (Svitkina et al. 2003; Lebrand et al. 2004; Steffen et al. 2006; Korobova and Svitkina 2008). From these data, two models have been proposed: the convergent elongation and de novo nucleation models, but no good methods exist to resolve these models. Recent work by Kirschner et al. (Lee et al. 2010b) used a reconstituted lipid bilayer system to propose a new model, the clustering-outgrowth model, wherein molecules are locally recruited to initiate linear outgrowth of filopodia from those sites. Here, in a cellular system, we attempt to discern how Cdc42 regulates filpodia formation, and whether its functions are consistent with any of these three models. Our results demonstrate that Cdc42 is activated locally at the sites of filopodia formation and is maintained in an active form during the course of filopodia elongation. The activity of Cdc42 is regulated by a novel GAP whose activity is modulated by Src activity at these filopodia.