Crk associates with ERM proteins and promotes cell motility toward hyaluronic acid

Cell migration is a well organized process regulated by the extracellular matrix-mediated cytoskeletal reorganization. The signaling adaptor protein Crk has been shown to regulate cell motility, but its precise role is still under investigation. Herein, we report that Crk associates with ERM family proteins (including ezrin, radixin, and moesin), activates RhoA, and promotes cell motility toward hyaluronic acid. The binding of Crk with ERMs was demonstrated both by transient and stable protein expression systems in 293T cells and 3Y1 cells, and it was shown that v-Crk translocated the phosphorylated form of ERMs to microvilli in 3Y1 cells by immunofluorescence and immunoelectron microscopy. This v-Crk-dependent formation of microvilli was suppressed by inhibitors of Rho-associated kinase, and the activity of RhoA was elevated by coexpression of c-Crk-II and ERMs in 3Y1 cells. In concert with the activation of RhoA by Crk, Crk was found to associate with Rho-GDI, which has been shown to bind to ERMs. Furthermore, upon hyaluronic acid treatment, coexpression of c-Crk-II and ERMs enhanced cell motility, whereas the sole expression of c-Crk-II or either of the ERMs decreased the motility of 3Y1 cells. These results suggest that Crk may be involved in regulation of cell motility by a hyaluronic acid-dependent mechanism through an association with ERMs.


INTRODUCTION
The extracellullar matrix (ECM) plays an important role in various cellular responses (1)(2)(3). ECM drives the spatio-temporal reorganization of the cytoskeleton which is involved in physiological cell migration, tumor cell invasion, and metastasis. Multiple cell surface molecules have been shown to participate in this ECM-dependent signalling mechanism.
One of the major molecules is CD44, a transmembrane receptor for hyaluronic acid (4,5), which associates with the actin cytoskeleton through the ERM family proteins (ERMs) including ezrin, radixin, and moesin. The cleavage of CD44 at the extracellular domain by membrane-associated metalloproteinases plays a crucial role in efficient cell detachment during cell migration (6,7). The binding of CD44 and ERMs is controlled by the threonine-phosphorylation of ERMs through Rho-associated kinase (ROCK), and also by N-terminal phospholipid modification of ERMs (8).
The signalling adaptor protein Crk, which is composed of an SH2 and two SH3 domains, is considered to be involved in the cytoskeletal regulation. Crk has been shown to interact with components of focal adhesion, such as p130 Cas and paxillin (9,10), which were tyrosine-phosphorylated mainly by integrin stimulation. Crk transmits signals to downstream effecters through Crk-SH3 binding proteins, C3G and Dock180, which exert a guanine-nucleotide exchange factor (GEF) activity on Rap-1/R-Ras and Rac, respectively (11)(12)(13). Thus, Crk may regulate cytoskeletal movement thorough these GEFs and small GTPases. In fact, studies of C3G knockout mice have suggested the regulation of cell adhesion by C3G through Rap-1 (14). The phagocytosis, membrane ruffling, and lamellipodia formation has been shown to be regulated by a Dock180-ELMO-Rac-dependent mechanism (15, 16).
Besides the identification of the activation of Rap-1 or Rac, we and others previously by guest on March 24, 2020 http://www.jbc.org/ Downloaded from for 1h. For Rac assay, cells were lysed with lysis buffer composed of 1% NP40, 25 mM HEPES pH7.4, 150 mM NaCl, 10% glycerol, 1 mM EDTA, 10 mM MgCl 2 , 1 µg/ml aprotinin, and 1 mM PMSF. Lysates were clarified by 12,000 rpm at 4 for 1 min, and the supernatant was incubated with 10 µg of purified GST-PAK2-RBD and glutathione-beads at 4 for 1h. In both Rho and Rac assays, the beads were washed three times with each lysis buffer and subjected to SDS-PAGE with a 12% gel. Precipitated RhoA or Rac1 was detected by immunoblotting using anti-RhoA or Rac1Ab.
Immunoelectron microscopy-Analysis was performed by the pre-embedding method with double immunostainig. v-Crk-induced 3Y1 cells were fixed with 0.1% glutaraldehyde in 0.1 M cacodylate buffer (CB) for 5 min on ice, and first incubated with a mixture of primary rat mAbs for pERM and a mouse mAb for v-Crk for 3 days at 4 . After washing with PBS, the specimens were incubated with 10 nm gold-labeled anti-mouse Ig Ab for 1hr, followed by incubation for 1 hr with biotin-labeled anti-rat Ab, which was further reacted with peroxydase-labeled streptavisin. After re-fixation for 5 min, the enzyme reaction was visualized by using diaminobenzidine (DAB) as substrate. Cells were re-fixed with 2% OsO4 in 0.1M phosphate buffer for 50 min and then embedded in epon. Cells in epon block were sectioned into 1-µm thicknesses and stained with 1% toluidinblue for confirmation of 8 measured. To examine the mechanism of Crk-mediated cytoskeletal movement, the association of Crk and ERMs was examined, because we reported the Crk-dependent activation of RhoA and the cleavage of CD44 (19), and ERMs are known to bind to CD44 regulating actin cytoskeleton. First, we found that anti-Crk antibody coprecipitated transiently expressed ERMs with endogenous Crk in human embryonal kidney To analyze the binding mechanism of Crk and ERMs, we transiently transfected mutants of c-Crk-II with ERMs in 293T cells. Although the association of c-Crk-II and radixin/moesin seemed to be weakened when we expressed the SH2 or SH3 mutants of c-Crk-II (Fig. 1C, lanes 9, 10, 12, 13), we could not detect a remarkable suppression of the binding of SH2-or SH3-mutants of c-Crk-II and ERMs (Fig. 1C). Mutational analyses using SH2 or SH3 mutants of v-Crk and CrkL were also performed, and similar results were obtained (data not shown). To examine the mechanism of Crk-induced Rho activation, we focused on the association of Crk and Rho-GDI because no known Rho GEF has been reported to bind to Crk. In fact, Rho-GDI contains possible Crk-interacting sequences such as the YXXP motif for the SH2 domain and proline-rich region for the SH3 domain. The colocalization of these proteins was observed in 3Y1 cells by confocal microscopy (Fig. 2B) and force-expressed Crk was coprecipitated with Rho-GDI by using anti-Rho-GDI antibody in 293T cells (Fig. 2C, arrowhead). The association of ERMs and Rho-GDI was also observed, as reported previously (Fig. 2C, asterisk) (21). It should be noted that we failed to detect the association of endogenous Crk and Rho-GDI (data not shown).

Association of Crk and pERMs and induction of microvilli formation-As ERM
proteins were known to be regulated by phosphorylation, the binding of Crk and the phosphorylated form of ERMs (pERMs) were examined by using a v-Crk-inducible 3Y1 cell line (clone 21-2-1) (19). In the presence of v-Crk, the association of Crk with pERMs was detectable in the cytoplasmic fraction of 3Y1 cells (Fig. 3A).
We then analyzed the subcellular localization of pERMs in a v-Crk-inducible 3Y1 cell line. pERMs were observed diffusely in the cytoplasm and partially at the edge of the cytoplasm of 3Y1 cells without v-Crk ( Fig. 3B-a). However, with v-Crk induction, pERMs were demonstrated to translocate to cellular microvilli, and co-localization of v-Crk and pERMs was shown by a merged image (Fig. 3B, b-d). In addition, co-localization of v-Crk and pERMs was also detected as dotted patterns in the cytoplasm (Fig. 3B, b-d, arrowheads).
To confirm the involvement of ROCK, which was known to phosphorylate ERMs in v-Crk-dependent microvilli formation of pERMs, we utilized the ROCK inhibitor Y27632, and found that this reagent inhibited the localization of pERMs to microvilli, while the remaining co-localization of v-Crk and pERMs was still detectable in the cytoplasm ( To confirm the immunofluorescence study, we performed immunoelectron microscopy using the double staining method. pERMs were visualized by using diaminobenzidine, in which they are recognized as electron-dense black substances by transmission electron microscopy (TEM), and the presence of anti-v-Crk Ab was demonstrated by 10 nm-gold particle-labeled secondary antibody. pERM labeled with DAB was recognized in the cytoplasm of vCrk expressing 3Y1 cells by light microscopy (Fig. 4a).
Immunoelectron microscopy demonstrated that vCrk was colocalized with pERM at the microvilli, cytoplasmic edge, and filamentous structure in the cytoplasm of the cells ( Fig.   4b-g). recovered the motility to the levels of wild type 3Y1 cells, but did not provide significant enhancement of motility higher than that of wild type 3Y1 cells (data not shown).

The cleavage of CD44 in 3Y1 cells expressing both Crk and ERMs-To
To confirm the involvement of the CD44 cleavage in v-Crk-regulated cell motility, we examined the effect of PI-3 kinase inhibitors because PI-3 kinase was known to up-regulate CD44 cleavage (23,24). Wound healing assay demonstrated that PI-3 kinase inhibitors such as LY294002 and Wortmannin tend to suppress the motility of 3Y1 cell lines; however, this suppressive function was most prominently found in 3Y1 cells stably expressing Crk and ezrin (Fig. 5C). PI-3 kinase inhibitors did not affect the motility of cells expressing Crk and moesin (Fig. 5C).

DISCUSSION
Signalling adaptor protein Crk was originally found as an avian sarcoma encoding oncoprotein v-Crk(25). Since human c-Crk-II, the homologue of v-Crk, was isolated, the identification of Crk targets has suggested that Crk links between tyrosine phosphorylated proteins and guanine-nucleotide exchange factors for small GTPases, and regulates cytoskeletal reorganization. In particular, under fibronectin stimulation, the integrin-provoked signal has been shown to be mediated by Crk and transmitted to the downstream effecter Dock180, leading to Rac activation. However, the mechanism of Crk-mediated cell migration or tumor cell invasion has still been under investigation.
In this study, we have found a novel interaction of Crk and the ERM family of proteins that is involved in activation of Rho and hyaluronic acid-CD44 dependent regulation of cell motility (Fig. 6) According to our previous results, v-Crk activated RhoA in fibroblasts and coexpression of Crk, and ERMs enhanced the activity of RhoA in 293T cells. As no known Rho-GEF was found to bind to Crk, the mechanism of Crk-dependent activation of Rho was the missing link. Rho-GDI has been shown to bind to the N-terminal FERM domain of ERMs (21), and these data led us to hypothesize that upon ECM stimulation, Crk binds to the negative regulator of RhoA such as Rho-GDI, inactivates of Rho-GDI, and leads to the activation of RhoA shown as Fig. 6. Thus, we examined the association of Rho-GDI and Crk. In this study, the association of force-expressed Crk and Rho-GDI was observed in 293T cells, but we failed to show the association of endogenous Crk and Rho-GDI (data not shown). In 293T cells, we did not examine the inhibition of the function of negative regulator Rho-GDI, because the simple expression of Rho-GDI did not significantly suppress the activity of RhoA measured by pull-down assay. Furthermore, we also tried to test the effect of Crk on another negative regulator of RhoA, Rho-GAP. However, we did not observe a significant activation of RhoA by the double expression of Crk and Rho-GAP (data not shown). Establishment of a deficient cell line for the negative regulator of RhoA may reveal the Crk-dependent activation mechanism of Rho in future studies.
As it is known that ERMs were phosphorylated in the cytoplasm and translocated to membrane, and that the phosphorylated form of ERMs (pERMs) link between CD44 and actin cytoskeleton, we analyzed the localization of pERMs in v-Crk inducible fibroblasts. In 3Y1 cells, v-Crk translocated pERMs and induced microvilli formation by a ROCK-dependent signalling mechanism. Although we expected the induction of v-Crk-induced phophorylation of ERMs, we failed to demonstrate such increased phosphorylation of ERMs by Crk in our system (Fig. 3C). We speculated the relatively high levels of pERMs in the cytoplasm of wild type 3Y1 cells may mask further phosphorylation of ERMs.
In this study, we showed that the association of Crk and ERMs was involved in the hyaluronic acid-CD44 signalling mechanism to promote cell motility. Considering the mechanism of Crk-dependent enhancement of CD44 cleavage, Crk may also regulate the transcriptional levels of matrix-metalloproteinases (MMPs). As Crk is also known to activate PI-3 kinase (26)