Rho-family GTPases regulate various aspects of cell function by controlling cytoskeletal changes; however, their spatial regulation within the cells remains largely unknown. To understand this regulation, we have studied the spatiotemporal activity of Rho-family GTPases in migrating cells and growth factor-stimulated cells by using probes based on the principle of fluorescence resonance energy transfer. In migrating fibroblasts and epithelial cells, the level of RhoA activity is high both at the contractile tail and at the leading edge, whereas Rac1 and Cdc42 activities are high only at the leading edge. In cells stimulated with epidermal growth factor or nerve growth factor, activities of Rac1 and Cdc42 were transiently elevated in a broad area of the plasma membrane, followed by a localized activation at nascent lamellipodia. In contrast, on epidermal growth factor stimulation, RhoA activity decreased diffusely at the plasma membrane. Notably, RhoA activity persisted at the tip of growth factor-induced membrane ruffles and, in agreement with this finding, RhoA is required for membrane ruffling. These observations suggest that the activities of Rho-family GTPases are elaborately regulated in a time- and space-dependent manner to control cytoskeletal changes and that the basic mechanism of controlling cell shape via Rho-family GTPases is common to various cell types.
Rho-family GTPases function as critical molecular switches in transducing extracellular signals to the cytoskeleton [1,2]. It is widely accepted that RhoA triggers actin stress fibre formation, that Rac1 induces lamellipodia, thin-membrane protrusions and membrane ruffles and that Cdc42 elicits the formation of filopodia [3,4]. These observations suggest that the site demonstrating such morphological change correlates with the activation site of each Rho-family GTPase; however, due to the lack of a proper technique to visualize the activity of Rho-family GTPases, this assumption was not examined until the recent development of probes based on the principle of FRET (fluorescence resonance energy transfer).
FRET is a process by which radiationless transfer of energy occurs from a fluorophore of a donor to an acceptor molecule placed in close proximity to the donor [5–7]. The donor and acceptor fluorophores must be placed within a distance of approx. 5 nm from each other (in the case of autofluorescent proteins) in order to observe detectable FRET. In addition to this distance, the relative orientation of the donor and the acceptor also affects the FRET efficiency. Thus, by labelling proteins of interest with donor and acceptor fluorophores, the conformational change of a protein or interaction of a pair of proteins can be monitored based on this principle. The recent development of a variety of autofluorescent proteins as donors and/or acceptors has been accelerating the development of such FRET-based probes.
Development of FRET probes for Rho-family GTPases
The activity of Rho-family GTPases is detectable either directly or indirectly with FRET-based probes (Figure 1). The direct measurement of activated Rho-family GTPases is achieved by means of probes comprising their cognitive effector molecules and a pair of the donor and acceptor fluorophores [8–10]. The binding of endogenously activated Rho-family GTPases to the probe is monitored by the change in efficiency of FRET. The indirect method detects conformational change of fluorophore-labelled Rho-family GTPases by means of FRET [9–11]; this method therefore assumes that the fluorophore-labelled Rho-family GTPases and the endogenous ones are similarly regulated by GEF (guanine nucleotide-exchange factor) and GAP (GTPase-activating protein) for Rho-family GTPases. The advantages and disadvantages of both methods have been fully described elsewhere . In our experience, the direct method using the Rho-binding domains of effector proteins frequently results in cell damage, probably due to their dominant-negative effect against endogenous Rho-family GTPases. Thus we will mainly discuss the results obtained by the indirect method in this paper.
Two types of FRET probes for Rho-family GTPases
Spatial regulation of Rac1, Cdc42 and RhoA in migrating cells
Cell migration, an essential process in embryonic development, wound repair, immune surveillance and tumour cell metastasis, is regulated by many extracellular cues, including growth factors, cytokines and the extracellular matrix. Rho-family GTPases function as critical molecular switches in transducing such extracellular signals to both the actin and microtubule cytoskeleton [1,2]. For the directional migration of cells, the activity of Rho-family GTPases must be co-ordinately regulated both in space and time (see Figure 2) .
Spatial pattern of the activities of Rho-family GTPases in migrating cells
The activities of Rac1 and Cdc42 increase in a gradient from the tail to the leading edge in migrating cells. This observation agrees substantially with the proposed role of Rac1 and Cdc42, which are known to induce lamellipodia and filopodia respectively at the leading edge [9,11]. Interestingly, the level of increase towards the edge of lamellipodia is steeper for Cdc42 than for Rac1 . Thus Cdc42 may function particularly at the tip of lamellipodia to extend further the filopodia, whereas Rac1 may function to yield a dense mesh-like actin structure by accelerating Arp2/Arp3 complex formation not only at the lamellipodia but also at the cell body . In HeLa and MDCK (Madin–Darby canine kidney) cells migrating at a low cell density, RhoA is activated both at the contractile tail and at the leading edge, whereas RhoA is activated only at the leading edge in MDCK cells migrating as a monolayer sheet (K. Kurokawa and M. Matsuda, unpublished work). It has been proposed that, in migrating leucocytes, Rho functions at the tail of the cells by promoting actomyosin contraction, thereby inducing retraction of the cytoplasmic tail . Furthermore, the RhoA-induced activation of ROCK (Rho-associated kinase) is known to be essential for cell detachment at the tail of migrating leucocytes . The high RhoA activity at the uropod of migrating HeLa cells and MDCK cells at a low cell density agrees with these previous observations. However, we have not detected high RhoA activity at the rear end of MDCK cells migrating as a monolayer sheet, nor have we observed any inhibitory effect of the ROCK inhibitor on MDCK cell migration in the wound-healing assay (K. Kurokawa and M. Matsuda, unpublished work). This observation demonstrates that the Rho-ROCK pathway is not involved in the migration of monolayer cells, and is in agreement with a previous report showing that ROCK activity is not necessary for the migration of monolayer REF cells . Thus the requirement of the RhoA-ROCK pathway at the tail of migrating cells appears to be cell context-dependent.
We have found that Cdc42 activity is particularly high at the tip of membrane ruffles and filopodia of migrating cells. However, Nalbant et al.  have proposed that Cdc42 activity is suppressed in the filopodia by using a probe named MeroCBD. This discrepancy is probably caused by the difference in the design of probes. Our probe, Raichu-Cdc42, indirectly monitors Cdc42 activity by measuring the balance between GEFs and GAPs for Cdc42. However, the sensitivity of Raichu-Cdc42 to GAPs might not be the same as the sensitivity of the authentic Cdc42, which could result in a false positive. On the other hand, MeroCBD directly detects endogenous Cdc42-GTP. Nevertheless, MeroCBD could succumb to a false negative result, because it might not be able to bind to Cdc42-GTP that is tightly associated with the Cdc42 effector(s) in the filopodia. This discrepancy may not be solved until the development of probes based on some other principle.
Spatial regulation of RhoA, Rac1 and Cdc42 in EGF (epidermal growth factor)-stimulated fibroblasts and epithelial cells
Growth factors elicit drastic morphological change in cells; therefore it is not surprising that many growth factors increase or decrease the activity of Rho-family GTPases. In growth factor-stimulated epithelial cells, the activities of Rac1 and Cdc42 increase transiently and diffusely at the plasma membrane . After a few minutes, activation of Rac1 and Cdc42 are confined to narrow areas exhibiting membrane ruffles. In clear contrast, RhoA activity decreases diffusely at the plasma membrane in a manner dependent on Rac1 (K. Kurokawa and M. Matsuda, unpublished work) . Importantly, however, a FRET probe has revealed that RhoA activity remains high at the membrane ruffles in nascent lamellipodia (K. Kurokawa and M. Matsuda, unpublished work). The observed decrease in RhoA activity agrees with the proposed role of RhoA in stress fibre formation, because growth factors frequently decrease the level of stress fibres. The persistent RhoA activity at the cell periphery also implicates RhoA in the regulation of membrane ruffling, the induction of which is a typical phenotype of activated Rac. Furthermore, these observations clearly show that RhoA, Rac1 and Cdc42 are all activated at membrane ruffles. In agreement with this observation, we have found that the dominant-negative mutants of RhoA and Cdc42 as well as that of Rac1 inhibit EGF-induced membrane ruffling. Furthermore, we have found that inhibitors of actin polymerization also abrogate the activation of Cdc42 and the induction of membrane ruffling. Thus the membrane ruffling is driven by a number of molecules, including Rho-family GTPases, which comprise a positive feedback loop.
Spatial regulation of RhoA, Rac1 and Cdc42 in neuronal cells
Neurite extension is another example of the morphological changes regulated by Rho-family GTPases . We have examined the activity changes of Rho-family GTPases in PC12 cells stimulated with NGF (nerve growth factor), which are frequently used as a model system for the neuritogenesis. Immediately after the addition of NGF, Rac1 and Cdc42 are transiently activated in broad areas of the plasma membrane [22,23]. As in EGF-stimulated epithelial cells, RhoA activity decreases rapidly and diffusely at the plasma membrane in NGF-stimulated PC12 cells. Thereafter, during the course of neurite extension, a repetitive activation and inactivation cycle of all of RhoA, Rac1 and Cdc42 is observed at the motile tips of protrusions. These observations indicate that each of RhoA, Rac1 and Cdc42 co-operatively functions to induce neurites during neuronal development, as has been seen in the induction of membrane ruffles.
FRET imaging of Rac1 and Cdc42 activity possesses a unique advantage over the conventional biochemical methods in that time-dependent change and spatial distribution can be traced in individual cells. A biochemical study of PC12 cells has already detected a single peak of Rac1 activation immediately after the addition of NGF . However, in the single-cell recordings using the FRET probes, we observed intermittent activation of Rac1 and Cdc42 following NGF treatment, which has not been detectable by analyses using the bulk of cells. Immediately after NGF addition, Rac1 and Cdc42 are transiently activated in broad areas at the cell periphery, and thereafter, activation localized at the neurite tips is recurrently observed. This finding of localized activation during the extension process may help to explain why expression of constitutively active mutants of Rac1 and Cdc42 inhibit NGF-induced neurite outgrowth . A similar failure of the induction of neurite outgrowth by the constitutively active mutants of Rac1 and Cdc42 has been reported in the absence of ligands in PC12 cells [25,26] and other neuronal cells . We propose that localization signals, including the localized Rac1/Cdc42 activity shown here, are necessary for neurite formation, as has been discussed previously . Another possible explanation is that the Rho-GTPase signalling pathway has a cyclic mode of action, and constitutively active mutants may block this cycle , although the underlying mechanisms have not yet been described.
To scrutinize further the role of Rho-family GTPases in neurite extension, we have examined the activity change of Rho-family GTPases during the laminin-induced growth cone advance of dorsal root ganglion cells and N1E-115 neuroblastoma cells (Figure 3) (T. Nakamura, K. Aoki and M. Matsuda, unpublished work). Rac1 and Cdc42 activities are high in the P-domain (peripheral domain) of growth cones. Active Rac1 is uniformly detected throughout the P-domain, whereas Cdc42 activity increases gradually towards the growth cone edge. RhoA activity is higher in the P-domain than in the central domain and axon shaft, and a high level of RhoA activity is maintained in the extending part of growth cones. Thus RhoA activation in the shaft and cell body results in neurite retraction, whereas high RhoA activity in the P-domain is necessary to retain the spread morphology of the nerve growth cone.
Schematic representation of the activity of Rho-family GTPases in neurites
After examining the spatiotemporal regulation of Rho-family GTPases in fibroblasts, epithelial cells and neuronal cells, it has become clear that the spatial activity of Rho-family GTPases is closely correlated with dynamical changes in cytoskeletal structure. An increasing gradient of Rac1 activity is formed towards the lamellipodial membrane protrusion in epithelial cells and in the growth cone of neuronal cells. Similar to Rac1, Cdc42 activity also shows an increasing gradient, but it is more restricted to the tip of extending plasma membranes. RhoA activity is high not only at the tip of extending plasma membranes, but also in regions where the plasma membrane shrinks, i.e. the uropod of migrating epithelial cells and the shaft of shrinking neurites. These observations suggest that the basic mechanism underlying the morphological change is common among different cell types.
FRET-based probes have revealed the elaborate spatiotemporal regulation of Rho-family GTPases. The next step will be to identify the molecules regulating Rho-family GTPases at specific times and places. A single cell contains a number of regulatory molecules for Rho-family GTPases. Therefore, for a full understanding of the signalling cascade upstream of Rho-family GTPases, we next need to know when and where each regulatory molecule functions within the cells. Equally important as this question is the need to examine when and where effectors of Rho-family GTPases are activated, because Rho-family GTPases seem to select specific subsets of effectors depending on the cellular context. It is necessary to develop probes that can address these challenging questions for a comprehensive understanding of the role of Rho-family GTPases in the cell structure.
Localization and Activation of Ras-like GTPases: Focused Meeting held at the Royal Agricultural College, Cirencester, U.K., 21–23 March 2005. Organized and Edited by A. Ridley (Ludwig Institute of Cancer Research, London, U.K.) and M. Seabra (Imperial College London, U.K.).