Several lines of indirect evidence have linked the formation of fascin-containing cell protrusions with the status of actomyosin contractility or focal adhesions, but the processes involved, in particular the role of Rho GTPase, have remained obscure. In this study, we established, with multiple lines of evidence, that Rho activity modulates the ability of fascin-1 to interact with actin in both normal and carcinoma-derived cells. The discovery of this novel function of Rho was advanced by the development of a novel assay to measure the fascin-1/actin interaction by FRET/FLIM microscopy. In this assay, a small, actin-binding peptide, lifeact, was adopted as the FRET donor. Lifeact binds reversibly to F-actin, and thus FRET with fascin-1 takes place only when both molecules are in close proximity (<10 nm) and bound to filamentous actin. The specificity of the interaction measured was established by: 1) the lack of FRET interaction of lifeact with the non-actin-bundling form of fascin-1, fascin-1S39D; 2) the independence of fascin-1/lifeact FRET from PKC kinase activity, and 3) the correspondence of lifeact distribution and fascin-1/lifeact FRET with a subset of the F-actin structures to which phalloidin binds.
Using the combined approaches of immunofluorescence microscopy, time-lapse imaging of cells expressing GFP-fascin-1, and the novel FRET assay, we found that inhibition of either Rho or its effectors Rho kinase I and II resulted in inhibition of the fascin-1/lifeact interaction as measured by FRET/FLIM. According to the indirect immunofluorescence of endogenous F-actin or fascin-1 and the imaging of live cells expressing GFP-fascin-1, inhibition of either Rho or Rho kinases also altered cell morphology and led to the formation of more protrusions containing fascin-1 at cell edges. When analyzed in detail by confocal time-lapse microscopy, the filopodial activity of Y27632-treated cells was seen to occur within regions of increased membrane ruffling. These filopodia had distinct characteristics: GFP-fascin-1 was not localized strongly along the filopodial shaft; the filopodia were less straight, probably as a result of loss of actin-bundle rigidity caused by decreased localization of fascin-1 towards filopodial tips (Figure ), and the lifetimes of these filopodia were longer (Figure ; see Additional file 3
, movie 1; see Additional file 4
, movie 2; see Additional file 5
, movie 3).
The FRET assay studies a specific molecular interaction within the overall assembly/disassembly dynamics of protrusions. Because Rho kinase inhibition of SW480 migration on LN is fascin-1-independent [5
], it is likely that control of protrusion number has a multifactorial basis, for example, it may be linked to both actomyosin-based cell tension and chemical signaling. Inhibition of Rho or Rho kinases is well known to inhibit stress-fiber microfilament bundles within cell bodies [43
]; however, in our experiments, many cells treated with these inhibitors displayed increased association of fascin-1 with cell-body microfilament bundles. Fascin-1 and tropomyosin act as antagonists for actin binding, and previous studies of ECM-adherent cells have indicated preferential association of fascin-1 with cell-body microfilament bundles in cells under conditions of reduced focal adhesion assembly and contractility [4
]. Treatments with C3 toxin or Y27632 are likely to lead to similar conditions of reduced cell contractility in adherent cells.
By combining the new lifeact FRET assay with biochemical analyses of cell extracts, we have identified Rho-dependent regulation as a novel signaling process that affects both fascin-1 and the fascin-1/actin interaction. Importantly, the mechanistic basis for the pathway downstream of Rho and Rho kinases resides in a previously unidentified process, the interaction of LIMK1/2 with fascin-1. Both of the LIMK isoforms are expressed in many tissues, although specificity of expression patterns and subcellular localizations has been described [37
]. In this study, we identified the interaction of fascin-1 with active LIMK1/2 in SW480 cells by biochemical pull-down from cell extracts and by FRET/FLIM in intact, migrating cells. As established by the immunoblots, fascin-1 pull-down experiments, confocal microscopy, and FRET/FLIM analysis for LIMK1, the mechanism of the interaction depends on LIMK1 activation by Rho kinase phosphorylation, but is independent of the status of S39-phosphorylation of fascin-1. Both S39-phosphorylated and non-phosphorylated fascin-1 are needed for efficient migration of SW480 cells [5
], and these new data suggest that the LIMK1/fascin-1 interaction is a possible mechanism to retain S39-phosphorylated fascin-1 in proximity to actin structures at cell edges. We found that fascin-1 localized in filopodia in cells expressing wild-type or T508A mutant forms of LIMK1; however, only wild-type LIMK1 was competent to stabilize filopodia. LIMK1-T508A, which did not interact with fascin-1 in the FRET assay, did not increase filopodial persistence. Thus the LIMK1/fascin-1 interaction contributes to the stability of filopodia. As reported previously [41
], expression of kinase-dead LIMK1 reduced in decreased numbers of filopodia, and the remaining filopodia were not stabilized.
Overall, these results identify a novel pathway that regulates actin-binding by fascin-1, with functional consequences for filopodial stability (Figure ). Regulation of the fascin-1/actin interaction did not depend on activity of MLCK or on myosin ATPase activity, and is thus separable from the status of actomyosin contractility.
Figure 9 Model for the novel pathway that regulates fascin-1/actin interaction and filopodia stability. See Discussion for details. As shown in Figure 6, active p-Lin-11/Isl-1/Mec-3 kinase (LIMK)1/2 bound to both S39-phosphorylated and non-phosphorylated fascin-1. (more ...)
Similar to fascin-1, LIMK1 and LIMK2 contribute functionally to cancer cell invasion and metastasis, and are therefore of interest as therapeutic targets [36
]. Thus, fascin-1 and LIMK1/2 might co-operate to promote protrusions and cell motility if they are co-expressed in tumors. For example, LIMK1/2 overexpression is well known to increase F-actin, protrusions, and stress fibers in various cells, and correspondingly, LIMK1/2 knockdown or expression of kinase-dead LIMK1/2 decreases F-actin stability, filopodia/lamellipodia, and cell migration [41
]. We propose that the cell protrusion phenotypes observed after LIMK1/2 inhibition may, in part, be mediated by reduced actin binding and bundling by fascin-1. Interestingly, LIMK1/2 are also downstream effectors of p21-activated kinases (PAK), which, as we have previously shown regulate formation of the fascin-1/cPKC complex in colon carcinoma cells, and which also correlate clinically with metastatic progression [19
]. Thus, LIMK1/2 might represent a nexus between several pathways that regulate the ability of fascin-1 to participate in different protein complexes.
A well-characterized substrate of LIMK1/2 is cofilin, an actin-binding protein that acts to depolymerise F-actin by its actin-severing activity. Phosphorylation of cofilin on ser-3 by LIMK1/2 inhibits its severing activity, resulting in the aforementioned stabilization or reorganization of actin structures [50
]. Similar to fascin-1, the cofilin pathway has been linked to breast-cancer metastasis [50
]. It was recently shown that cofilin associates transiently with retracting filopodia, and that its actin-severing activity is enhanced on actin bundles crosslinked by fascin-1 [53
]. The relationship between the fascin-1/LIMK1/2 complex and cofilin in the dynamics of filopodia are questions of future interest. It will be important to establish if fascin-1 competes with cofilin for LIMK1 kinase availability, thereby modulating actin dynamics directly, or whether LIMK1/2 act directly or indirectly as a scaffold to maintain fascin-1 in proximity to F-actin. Alternatively, fascin-1 binding might inhibit the kinase activity of LIMK1/2, thereby affecting actin dynamics indirectly, as previously shown for nischarin [54
]. LIMK1 contributes to the transcriptional activity of serum response factor [55
], and it will be interesting to determine if this activity is modulated by fascin-1.