Cholesterol and sphingolipid-rich membrane microdomains or rafts have been shown to be involved in signaling through many growth factor receptors but the molecular details of these processes are not well understood. The reggie/flotillin proteins are ubiquitously expressed proteins with a poorly characterized function. They are constitutively associated with membrane rafts by means of acylation and oligomerization. Previous studies have implicated reggies in signaling, regulation of actin cytoskeleton and in membrane transport processes. In this study, we analyzed the putative role of reggie-1/flotillin-2 in signaling through the epidermal growth factor receptor. We show that reggie-1 becomes phosphorylated by Src kinase at several tyrosines upon stimulation of cells with epidermal growth factor. In addition, Src and reggie-1 are present as a molecular complex. Epidermal growth factor stimulation of cells results in a Tyr163-dependent translocation of reggie-1 from the plasma membrane into endosomes. We also show that reggie-1 is capable of enhancing the spreading of cells, again in a tyrosine-dependent manner, and knockdown of reggie-1 interferes with spreading. Thus, we reveal a new function for reggie-1 in the regulation of cell adhesion and actin dynamics and in growth factor signaling.
Signaling through growth factor receptors such as the epidermal growth factor receptor (EGFR) involves complex molecular networks of proteins and multiple phosphorylation events, including the autophosphorylation of EGFR, which in many cases are mediated by tyrosine kinases (Bromann et al., 2004). Tyrosine kinases of the Src family are involved in transmitting signals downstream of the receptors, and plenty of substrates for Src kinases are known, the phosphorylation of which plays an essential role in signaling (Ahn et al., 2002; Tu et al., 2003). One characteristic feature of these signaling pathways is that they tend to split, and signaling through a certain receptor can often proceed through different routes. Such signaling pathways can use distinct molecular chains of events for inducing a certain response of the cell, such as proliferation or migration. Src kinases and many of their substrates are also involved in mediating the reorganization of actin cytoskeleton and modulation of cell adhesion in response to growth factor signaling.
Many proteins involved in signal transduction, including some Src kinases, are localized in specialized cholesterol-rich membrane microdomains or rafts (Simons and Toomre, 2000). Microdomains have also been implicated in signal transduction through membrane receptors, and some growth factor receptors have even been found at least transiently localized in rafts. This indicates that membrane microdomains play an important role in cellular signaling. Rafts have been suggested to provide a platform in which signaling proteins such as Src kinases are concentrated and can form molecular complexes with other proteins, providing means of efficiently regulating signaling. However, in many cases the molecular details of how membrane rafts are involved in growth factor signaling are not yet clarified.
Reggie-1/flotillin-2 and reggie-2/flotillin-1 are evolutionarily highly conserved, widely expressed proteins associated with non-caveolar rafts (Langhorst et al., 2005; Stuermer et al., 2001). Reggies were originally described as neuronal proteins upregulated in goldfish retinal ganglion cells during axonal regeneration (Lang et al., 1998; Schulte et al., 1997). Reggies do not contain any clear functional domains other than the prohibitin homology (PHB) domain, the function of which is unclear (Morrow and Parton, 2005; Tavernarakis et al., 1999). In addition, reggies bear very limited similarity to other proteins, and their molecular function has thus remained elusive. Although reggies are tightly associated with membrane rafts, they do not contain any transmembrane domains and thus do not traverse the membrane bilayer. We have earlier shown that reggie-1 associates with membranes by means of myristoylation of the N-terminal glycine and multiple palmitoylation in Cys4, Cys19 and Cys20 (Neumann-Giesen et al., 2004). Mutation of Gly2 to Ala results in a non-myristoylated and non-palmitoylated protein, which is soluble and not capable of associating with membranes. Although the lipid modifications are necessary and sufficient for membrane association, they alone are not sufficient to target reggie-1 to membrane rafts or detergent-insoluble membranes. Efficient raft association is mediated by homooligomerization, which requires the C-terminal portion of reggie-1, together with the lipid modifications (Neumann-Giesen et al., 2004). Reggie-2 is associated with membranes by means of a single palmitoylation in Cys34 (Morrow et al., 2002). However, a hydrophobic stretch close to the N-terminus, which also includes Cys34, is necessary for raft association, whereas plasma membrane localization is mediated by a different hydrophobic sequence (Liu et al., 2005). In addition, raft association of flotillin-1 was shown to be independent of Cys34 (Liu et al., 2005), although this contradicts other results (Morrow et al., 2002).
In epithelial cells, reggie-1 is mainly localized at the plasma membrane and enriched in cell-cell contact sites (Neumann-Giesen et al., 2004), although it can also be detected in endosomal compartments. By contrast, reggie-2 often has a more exclusively endosomal localization (Liu et al., 2005). Ectopic expression of reggie-1-EGFP results in generation of numerous filopodia-like protrusions that contain actin (Hazarika et al., 1999; Neumann-Giesen et al., 2004). In these protrusions, reggie-1 is often found concentrated at the tips, consistent with its localization in filopodia in neuronal growth cones (Lang et al., 1998). These findings suggested a role for reggie-1 in the modulation of actin cytoskeleton.
Recently, several studies have suggested that reggies are involved in various cellular signaling and membrane transport events, including phagocytosis, T-cell signaling and signaling through insulin and IgE receptors (Baumann et al., 2000; Dermine et al., 2001; Kato et al., 2006; Morrow and Parton, 2005; Rajendran et al., 2003; Roitbak et al., 2005). In the absence of phosphatidylinositol 3-kinase activity, signals from insulin receptor can be directed through a pathway which involves reggie-2 (Baumann et al., 2000). Within a trimeric complex consisting of reggie-2, cbl and cbl-associated protein (CAP), reggie-2 binds directly to the sorbin-homology (SoHo) domain of CAP, thus mediating the recruitment of the complex into rafts. Recent findings also suggest the involvement of the reggie-2/CAP complex in actin remodeling (Liu et al., 2005), and association of reggie-2 with another SoHo protein, ArgBP2, seems to play a role in neurite outgrowth (Haglund et al., 2004).
Although many findings indicate a role for reggies in cellular signaling, nothing is known about their putative modifications, such as phosphorylation, which might play a role in signaling. As several tyrosine residues in reggie-1 are predicted to be phosphorylated, we analyzed the tyrosine phosphorylation of reggie-1 in growth-factor-stimulated epithelial cells.
Reggie-1 contains eight tyrosine residues, six of which are predicted by the NetPhos 2.0 program to become phosphorylated. To study whether endogenous reggie-1 is phosphorylated, an immunoprecipitation (IP) with an anti-phospho-tyrosine (Tyr-P) antibody was performed because the available antibodies specific for reggie-1 are poorly applicable to immunoprecipitation but work in western blots. HeLa and PC12 cells were incubated with pervanadate to prevent dephosphorylation of tyrosine residues, after which tyrosine-phosphorylated proteins were precipitated. Stringent washing conditions were used in order to prevent co-precipitation of non-phosphorylated proteins. Reggie-1 was detected by means of western blot and was found to be present in the immunoprecipitates from both HeLa and PC12 cells (Fig. 1A, lanes PY). A control IP (lanes C) was used to demonstrate the specificity of the precipitation and showed no precipitation of reggie-1.
To verify that reggie-1 is phosphorylated itself, we analyzed the phosphorylation of reggie-1-EGFP (R1-EGFP) fusion proteins (Neumann-Giesen et al., 2004), which were expressed in HeLa cells and immunoprecipitated with anti-EGFP antibodies. Western blot was performed with an anti-Tyr-P antibody. R1-EGFP was found to be phosphorylated in cells incubated with pervanadate, whereas no phosphorylation was detected without pervanadate incubation (Fig. 1B, left panel), indicating that reggie-1 undergoes a dynamic phosphorylation and dephosphorylation. To show that reggie-1 and not EGFP, which also contains tyrosine residues, was phosphorylated, a truncated R1-EGFP construct (SH-R1-EGFP) (Neumann-Giesen et al., 2004), which contains the first 30 residues of reggie-1 fused with EGFP, was used. This results in a fusion protein that is associated with membranes by means of myristoylation and palmitoylation within the N-terminal sequence of reggie-1 (Neumann-Giesen et al., 2004). As a further control, EGFP alone was used. Neither SH-R1-EGFP nor EGFP were found to be tyrosine phosphorylated (Fig. 1B, middle panel), demonstrating that the phosphorylation takes place within the reggie-1 sequence. Membrane association of reggie-1 seems to be a prerequisite for phosphorylation, because the full-length reggie-1 G2A mutant (G2A-R1-EGFP), which is soluble owing to lack of myristoylation and palmitoylation, was not phosphorylated (Fig. 1B, right panel). The level of overexpression of R1-EGFP and its mutant variants in this and successive experiments was only modest and very similar, with about 0.7 to 1.3 times the amount of endogenous reggie-1, as determined by western blot of cell lysates with a reggie-1-specific antibody (see supplementary material Fig. S1). In general, unless otherwise stated, all experiments presented in this paper were performed at least twice with very similar results.
Membrane rafts are involved in signal transduction by many growth factor receptors, including the EGFR (Liu et al., 1996; Pike and Casey, 1996; Stehr et al., 2003; Waugh et al., 1999). Thus, the phosphorylation of R1-EGFP after stimulation of cells with EGF was analyzed. As shown in Fig. 1C, R1-EGFP was tyrosine phosphorylated after EGF treatment of cells in the absence of pervanadate, demonstrating that reggie-1 is phosphorylated under physiological conditions. Phosphorylation of reggie-1 was first detectable after 2 minutes of EGF treatment, highest after 5 minutes and began to decline after 10 minutes of EGF stimulation. The induction of reggie-1 phosphorylation was shown to be dependent on the activation of EGFR, since the EGF-induced tyrosine phosphorylation of R1-EGFP could be prevented with tyrphostin AG1478, which inhibits the autophosphorylation of EGFR (Fig. 1D). These results demonstrate that reggie-1 is tyrosine phosphorylated in an EGF-dependent manner and is thus likely to be involved in signal transduction through EGFR.
The Src family kinases Fyn (Stuermer et al., 2001; Stuermer et al., 2004) and Lyn (Kato et al., 2006) have been previously shown to colocalize and coimmunoprecipitate with reggies, and Src kinases are thus good candidates for tyrosine phosphorylation of reggie-1. Therefore, the effect of Src kinase inhibitors on the phosphorylation of R1-EGFP was next studied. The Src inhibitors PP1 and PP2 were found to inhibit tyrosine phosphorylation of R1-EGFP to a large degree, whereas a non-inhibiting analog PP3 had no effect on the phosphorylation of reggie-1 (Fig. 2A). Quantitative analysis of three independent experiments demonstrated that 5 μM PP1 resulted in 85.4% (s.d. 7.4%) and 5 μM PP2 in 88.6% (s.d. 5.1%) reduction of phosphorylation compared with control with PP3. The EGF-stimulated tyrosine phosphorylation of reggie-1 in the absence of pervanadate was also shown to be inhibited by Src kinase inhibitors, as shown in Fig. 2B for PP2 and for the highly specific Src inhibitor SU6656. Coexpression of R1-EGFP with Src or the constitutively active Y527F-Src mutant resulted in reggie-1 phosphorylation in the absence of pervanadate or EGF treatment, whereas the kinase-dead mutant of Src had no effect (Fig. 2C). Coexpression of Fyn kinase also resulted in considerable phosphorylation of R1-EGFP, indicating that also other Src kinases could mediate the tyrosine phosphorylation of reggie-1. The dependency of reggie-1 phosphorylation on the activity of Src kinases was also demonstrated by the fact that R1-EGFP was not phosphorylated in embryonic fibroblasts derived from mice deficient for the three ubiquitous Src kinases (Src, Yes and Fyn). However, coexpression of Src, but not of kinase-dead Src, resulted in rescue of tyrosine phosphorylation of R1-EGFP in these cells (Fig. 2D).
To determine whether reggie-1 and Src kinase are present as a molecular complex, coimmunoprecipitations were performed. HeLa cells transfected with R1-EGFP and Src, Y527F-Src or kinase-dead Src were subjected to immunoprecipitation with anti-EGFP, and all three forms of Src were found to coimmunoprecipitate with reggie-1 (Fig. 3). However, the degree of co-precipitation was dependent on the activity of Src, as the constitutively active form Y527F-Src was coprecipitated more efficiently than Src or the kinase-dead form. R1-EGFP was also found to coimmunoprecipitate with Y527F-Src when Src antibodies were applied for IP (data not shown). Similar results were obtained when reggie-1-myc and Y527F-Src were subjected to co-precipitation (data not shown), indicating that Src kinase and reggie-1 form a signaling complex, consistent with our previous data (Roitbak et al., 2005).
To identify the tyrosine residues of reggie-1 that are phosphorylated by Src, all eight Tyr residues in reggie-1 were next mutated to Phe either as single (Y24, Y27, Y124, Y241) or double substitutions (Y158+Y163, Y348+Y358) and analyzed for their phosphorylation status. Interestingly, all of the produced mutants were phosphorylated to a similar degree as the wild-type protein (Fig. 4A). A slightly reduced phosphorylation was only observed for the Y27F mutant. However, in western blots with the anti-GFP antibody, this mutant demonstrated a strong band with a slightly lower molecular size than the full-length protein, probably because of proteolytic processing. We have previously shown that this processing, which also takes place in wild-type reggie-1 to a smaller degree, results in release of the protein from the membranes (Neumann-Giesen et al., 2004). Some findings indicate that this processing might be mediated by calpain (Mairhofer et al., 2002) and could regulate the function of reggie-1 (Neumann-Giesen et al., 2004). The lower band was not phosphorylated in any of the experiments performed, and thus the increased processing could explain the reduced phosphorylation observed with the Y27F mutant.
The similar phosphorylation levels of the tyrosine mutants and the wt reggie-1 would indicate that reggie-1 is phosphorylated in multiple tyrosine residues. Therefore, seven of the eight tyrosines, excluding Tyr27, were mutated within a single construct. The mutant protein carrying these substitutions (7xYF-R1-EGFP) was found to be unphosphorylated when cells were stimulated with EGF (5 minutes) in the absence of pervanadate, in contrast to the phosphorylated wild-type protein (Fig. 4B). Since the cellular localization of the Y163F mutant was altered compared with the wild-type reggie-1 (see below), we generated a mutant in which all six other tyrosine residues except Tyr163 and Tyr27 were mutated to Phe within one construct (6xYF-R1-EGFP). Upon EGF stimulation, this mutant was phosphorylated, in contrast to the 7xYF mutant (Fig. 4B), clearly demonstrating that Y163 is a phosphorylation site because the only difference between these mutants is the presence of Tyr163 in the 6xYF mutant. However, single mutation of Tyr163 resulted in phosphorylation level comparable to the wild-type protein (Fig. 4C), indicating that other tyrosine residues in reggie-1 must be phosphorylated.
We have previously shown that reggie-1 is involved in reorganization of the actin cytoskeleton and induces filopodia-like protrusions upon overexpression (Neumann-Giesen et al., 2004). To study the effect of phosphorylation on the localization and function of reggie-1, the tyrosine mutant proteins were expressed in HeLa cells and analyzed by means of fluorescence microscopy. In a typical experiment, 87% of the R1-EGFP-transfected cells showed the localization depicted in Fig. 5, namely at the plasma membrane and in endosomes. In accordance with our previous data, R1-EGFP was found to induce filopodial protrusions. Most of the tyrosine mutants show a highly similar localization and phenotype as the wild-type protein (data not shown). However, in accordance with the appearance of the processed form, Y27F showed a more soluble localization with only a weak filopodial induction. Interestingly, in 85% of the transfected cells, the Y163F-R1-EGFP and 7xYF-R1-EGFP (Fig. 5) were localized at the plasma membrane but not in endosomes. In 15% of the cells, a weak vesicular localization was seen in addition to the prominent plasma membrane staining. Furthermore, both mutants showed a more efficient induction of filopodia than the wild-type R1-EGFP (Fig. 5, lowermost row), resulting in a `hairy' appearance with numerous actin-containing filopodia evident after phalloidin staining. These results suggest that phosphorylation of Tyr163 might regulate the cellular localization and function of reggie-1.
To study the effect of EGF-induced phosphorylation on the localization of reggie-1, cells were starved overnight and thereafter stimulated with EGF for 15 minutes. After starvation, R1-EGFP (Fig. 6A) and reggie-1-myc (6E, R1-myc) were localized mainly at the plasma membrane, and less endosomal staining was observed than steady-state localization (compare Fig. 6A with Fig. 5). Intriguingly, EGF stimulation resulted in translocation of R1-EGFP (Fig. 6B) and R1-myc (6F) into intracellular, endosome-like compartments. For quantification of the endocytosis, 200 transfected cells each from two independent experiments were scored. The translocation of R1-EGFP was detected in 37% of the transfected cells after EGF stimulation, whereas less than 1% of the starved cells showed a similar localization. By contrast, Y163F mutant protein remained at the plasma membrane after EGF stimulation and was incapable of translocating into endosomes (Fig. 6C,D). In the case of the Y163F mutant, less than 1% of cells showed translocation from the plasma membrane in both starved and EGF-treated cells.
Endocytosis of R1-EGFP upon EGF stimulation was also studied by means of live imaging. Administration of EGF resulted in movement of wild-type R1-EGFP containing vesicles away from the plasma membrane towards endosomes (see supplementary material Movies 1 and 2; supplementary material Movie 2 shows a section of Movie 1 with vesicle tracks marked). However, the Y163F mutant remained at the plasma membrane and very little vesicle movement could be detected (see supplementary material Movie 3). Taken together, these data further suggested an important role for phosphorylation of Tyr163 in the regulation of reggie-1 localization.
To determine the identity of the compartment that reggie-1 is endocytosed into, HeLa cells transfected with R1-EGFP were starved, stimulated with EGF for 15 minutes and subsequently fixed and stained for endogenous LAMP-3/CD63, a well-established late endosomal marker protein. Colocalization in endosomes was quantified by counting the percentage of endosomal structures that were positive for both proteins. As above, R1-EGFP was translocated to an intracellular compartment (Fig. 7A) in which a significant colocalization (40% of the LAMP-3-positive structures) was observed between LAMP-3 and reggie-1 in late endosomes. Since the internalization of reggie-1 is triggered by EGF stimulation and EGFR has been shown to localize to rafts during signaling, we analyzed whether reggie-1 and EGFR could be found in the same compartment. Indeed, reggie-1 and EGFR partially colocalized (Fig. 7B, colocalization in 37% of EGFR-positive structures) in a vesicular compartment probably representing late endosomes. We also performed colocalization experiments of R1-EGFP with EEA1 (Fig. 7C) and transferrin receptor (TfnR, Fig. 7D), which are markers for early/recycling endosomes, but very little colocalization (in about 5% of EEA1/TfnR-positive structures) could be observed.
The EGF-induced phosphorylation and endocytosis of reggie-1 raised the interesting possibility that reggie-1 might play a role in signaling from the EGFR to actin cytoskeleton. To analyze this, we knocked down reggie-1 in HeLa cells by means of RNA interference, which resulted in a considerable degree of reduction of reggie-1 (see Fig. 8). The cells were then starved and stimulated with EGF for 5 minutes to induce remodeling of the actin cytoskeleton, which was detected by phalloidin staining. Control cells (Fig. 6G) show very intense ruffling and generation of a large number of filopodia in a polarized fashion, whereas the reggie knockdown cells (Fig. 6H) have a much lower number of actin protrusions and ruffles and show no polarized phenotype. These data indicate that reggie-1 might indeed play an important role in the signaling of EGFR to the actin cytoskeleton.
A role for reggie-1 in actin-dependent processes is proposed (Haglund et al., 2004; Hazarika et al., 1999; Hazarika et al., 2004; Neumann-Giesen et al., 2004). To gain insights into functional aspects of reggie-1 phosphorylation, spreading assays of cells transfected with reggie-1 were performed. Cells were allowed to spread on fibronectin-coated glass coverslips for 25 minutes, and the spreading was analyzed by means of fluorescence microscopy. The cells were scored as spread (flat appearance with lamellipodia), half-spread (non-flat with filopodial protrusions) or non-spread (attached but round without filopodia). Examples of each category are shown in Fig. 8A. At least 200 cells were scored per sample, and a minimum of three independent assays was performed. Compared with the control cells transfected with EGFP, reggie-1 overexpression was found to considerably enhance cell spreading (Fig. 8B). Interestingly, the soluble, non-acylated G2A mutant functioned in a dominant-negative fashion and inhibited the spreading of cells. Furthermore, the Y27F mutant, which is also more soluble than the wild-type protein, exhibited a slightly inhibitory effect on spreading, further suggesting that the membrane localization and processing of reggie-1 regulates its function. Most of the other mutants were found to be similar to the wild-type protein in that they enhanced spreading (data not shown). Unexpectedly, even the nonphosphorylated 7xYF mutant was found capable of promoting cell spreading. However, Y24F, Y124F and Y163F mutants were inactive in this spreading assay, showing neither an inhibitory nor an enhancing effect and being similar to EGFP control, indicating a loss of function.
To further demonstrate a role for reggie-1 in cell-matrix adhesion, we again knocked down reggie-1 by 85-90% using two different siRNA duplexes (Fig. 8D) and performed spreading assays on fibronectin. Reggie-1 knockdown cells were found to spread poorly compared with cells transfected with control siRNA duplexes (Fig. 8C), demonstrating that reggie-1 is important for spreading of cells on fibronectin substrate.
In this study, we have for the first time provided direct molecular evidence for a role of reggie-1 in signaling through growth factor receptors. Our data demonstrate that reggie-1 becomes tyrosine phosphorylated at several residues after stimulation of cells with EGF, and that this phosphorylation is mediated by Src kinases. In addition, we have here revealed a novel function of reggie-1 in the regulation of cell adhesion and spreading which also seems to depend on the phosphorylation of reggie-1 at specific tyrosine residues.
Like many substrates of Src kinases, reggie-1 seems to be multiply phosphorylated at several tyrosine residues because all the single tyrosine mutants showed a phosphorylation comparable to the wild-type protein. Similar results have also been shown for other multiply phosphorylated proteins. For example, hepatocyte-growth-factor-regulated tyrosine kinase substrate (Hrs) contains at least two tyrosine residues, Tyr329 and Tyr334, which become phosphorylated upon EGF stimulation (Urbe et al., 2003). However, when either of these residues is mutated, the single mutants show similar phosphorylation efficiencies as the wild-type Hrs, whereas a double mutant exhibits a severely reduced phosphorylation. This suggests that although phosphorylation of both residues is possible, normally either Tyr334, which was found to be the preferential phosphorylation site in Hrs, or Tyr329 becomes phosphorylated. However, phosphorylation of further tyrosine residues is also likely to occur. In this study, we obtained very similar results for reggie-1, suggesting that the decision on which residues become phosphorylated in EGF-stimulated cells is under complex regulation. Thus, it would be interesting to characterize the phosphorylation pattern of endogenous reggie-1 by means of mass spectrometry but such analysis is hampered by the fact that the reggie-1 antibodies do not work well for immunoprecipitation.
EGF stimulation of cells and the consequent phosphorylation of reggie-1 by Src were found to affect the localization of reggie-1 in that it became endocytosed from the plasma membrane into LAMP-3/CD63-positive endosomes and even partially colocalized with the endocytosed EGF receptor. Although earlier studies have suggested that reggies are localized in endosomal compartments in some cell types (Dermine et al., 2001; Liu et al., 2005; Stuermer et al., 2001), this is the first demonstration of endocytosis of reggie-1 dependent on growth factor signaling. Interestingly, this translocation was found to be dependent on a specific tyrosine residue, Tyr163, which is embedded in a KxxxDxxxY163 consensus phosphorylation motif typical of tyrosine kinases (Edgar and Polak, 2001). Exchange of Tyr163 resulted in inhibition of EGF-stimulated endocytosis and exclusive steady state plasma membrane localization. In addition, the induction of filopodia-like protrusions was found to be even more efficient than in the case of the wild-type protein, suggesting that endocytosis of reggie-1 might negatively regulate its function in actin remodeling. Although it is tempting to speculate that endocytosis of reggie-1 would be dependent on phosphorylation of Tyr163 by Src, we have recently identified a putative cholesterol binding motif (VxxxxxY163xxxxxK) in reggie-1 which also involves Tyr163 (Roitbak et al., 2005). However, the fact that endocytosis of reggie-1 is induced by EGF and the 6xYF mutant containing Tyr163 is phosphorylated would rather suggest that phosphorylation is indeed involved. Thus, further experiments will be necessary to clarify whether reggie-1 can indeed bind cholesterol and whether the endocytosis of reggie-1 is dependent on cholesterol binding and/or on phosphorylation at Tyr163. It will also be of interest to determine which pathway mediates the uptake of reggie-1 from the plasma membrane. Our preliminary data suggest that endocytosis of reggie-1 is independent of clathrin and does not require caveolin or dynamin-2, suggesting the involvement of membrane rafts (M.A. and R.T., unpublished results). Recent findings have also suggested that reggie-2 might be involved in a clathrin- and caveolin-independent endocytosis pathway (Glebov et al., 2006).
Our data show that reggie-1 is involved in regulating the spreading of cells on fibronectin substrate, a process that is dependent on the integrin family of cell adhesion receptors. Overexpression of wild-type reggie-1 resulted in enhanced cell spreading, whereas reggie-1 depletion inhibited spreading. Thus, our results here indicate a novel function for reggie-1 in integrin signaling. These findings are in agreement with the previously suggested role of reggie-1 in the remodeling of the actin cytoskeleton, which is tightly linked with the formation of focal contacts and integrin signaling upon cell spreading. Phosphorylation of several substrates, e.g. focal adhesion kinase and vinculin, by Src has been shown to affect cell spreading (Hanks and Polte, 1997; Zhang et al., 2004). Cross talk between EGFR and integrins seems to occur because EGFR is activated upon plating of cells on fibronectin (Miyamoto et al., 1996; Moro et al., 1998). Furthermore, some integrins seem to be associated with and signal through membrane rafts (Claas et al., 2001; Decker et al., 2004), providing a potential link from phosphorylation of reggie-1 to cell motility.
Interestingly, we found that mutations of specific tyrosine residues (Tyr24, Tyr124 and Tyr163) rendered reggie-1 inactive and incapable of enhancing cell spreading, whereas most other tyrosine mutations showed no effect, suggesting that phosphorylation of these specific residues might regulate the function of reggie-1 in this process. Surprisingly, the unphosphorylated 7xYF mutant was found to be comparable to the wild-type protein in the spreading assay. However, there are examples in the literature showing similar behavior to other phosphorylated proteins involved in cell spreading. One possible explanation for this phenomenon could be that the function of reggie-1 depends on a crucial combination of phosphorylated and non-phosphorylated tyrosines. Phosphorylation of some tyrosines (e.g. Tyr24, Tyr124 and Tyr163) might be necessary to facilitate spreading, whereas the phosphorylation of others might provide an inhibitory signal. The absence of these inhibitory signals in the 7xYF mutant of reggie-1 might explain why it is still capable of enhancing cell spreading. Similar observations have been shown for the actin-binding protein villin, which is also a substrate of Src and regulates cell motility in a phosphorylation-dependent manner (Tomar et al., 2004). Phosphorylation of specific tyrosine residues in villin was shown to promote cell migration whereas phosphorylation of other tyrosines inhibited it. Another explanation for the ability of the unphosphorylated reggie-1 mutant to enhance cell spreading could be that a conformational switch of the protein is necessary for the activation, and that the active conformation can be more readily adopted by the 7xYF mutant, abrogating the necessity for phosphorylation. For example, vinculin and ezrin (Srivastava et al., 2005; Zhang et al., 2004) undergo a conformational change that is necessary for their activation. Such conformational changes can be facilitated by the tyrosine substitutions at specific sites, which might explain the observed enhancement of cell spreading by the 7xYF mutant of reggie-1. Thus, it is plausible that phosphorylation and protein conformation together regulate the function of many proteins, including reggie-1, in cell spreading.
In the case of reggie-1, cell spreading was considerably inhibited by the knockdown of reggie-1 by means of RNA interference and by the G2A and Y27F mutants. Since the G2A mutant is soluble and Y27F mutant partially soluble, the non-membrane associated forms of reggie-1 seem to act as antagonists or dominant-negative proteins. This might have a physiological role in the regulation of the function of reggie-1 in signaling. Earlier findings have suggested that in activated platelets, reggie-1 is released from the membranes after being processed by calpain (Mairhofer et al., 2002), which is also the major protease involved in regulating the dynamics of focal adhesions (Glading et al., 2002). We have observed similar processing in epithelial cells (Neumann-Giesen et al., 2004) (our unpublished findings). Thus, dual regulation of reggie-1 function in cell spreading might be accomplished by phosphorylation and proteolytic processing, in that processing and release from the membranes of reggie-1 by calpain could result in inhibition of reggie-1 and thus in negative regulation of cell adhesion.
Although our data show that Src activity is necessary for the phosphorylation of reggie-1 and that Src kinase can be coimmunoprecipitated with reggie-1, it is not clear if the interaction between Src and reggie-1 is direct or if other proteins are involved. Reggie-1 does not contain any typical Src interaction motifs such as SH2 domains, and we could not detect phosphorylation of immunoprecipitated R1-EGFP in an in vitro assay with recombinant Src kinase (our unpublished results). Thus, the interaction might be facilitated by other proteins, which act as bridging adaptors between Src and reggie-1. Good candidates for such adaptor proteins are the SoHo domain containing proteins, especially CAP, which has been shown to interact with reggie-2, and also mediate further interactions by means of its SH3 domains (Kioka et al., 2002). Since CAP is involved in actin remodeling and is capable of interacting with actin binding proteins such as vinculin, it might be a member of a signaling complex comprising also reggie-1 and Src. Thus, we propose a model according to which activation of EGFR leads to activation and recruitment of Src kinase to reggie-1, possibly by means of CAP or other SoHo proteins. Phosphorylation of reggie-1 by Src in turn could result in signaling to actin cytoskeleton and changes in cell adhesion/spreading. In addition, reggie-1 can be internalized from the plasma membrane to endosomes, which might modulate or downregulate this signaling. Interestingly, it has been shown that cell detachment results in endocytosis of raft domains (del Pozo et al., 2004). Although recent findings implicate caveolae in this process in fibroblasts (del Pozo et al., 2005), it is possible that the reggie microdomains, which are different from caveolae, might also be involved in the regulation of integrin signaling, especially in many cell types devoid of caveolae.
Taken together, our results here indicate that reggie-1 might play a role in the regulation of raft-mediated signaling processes, including growth factor receptor signaling and cell-matrix adhesion. Regulation of such signaling events is of key importance for normal cell adhesion and proliferation, and aberrant signaling can lead to malignant transformation of cells and cancerogenesis. Owing to their ubiquitous expression, reggies are good candidates for regulators of raft-mediated signaling. Intriguingly, a recent study suggested that reggie-1 might be upregulated in the course of melanoma progression, and overexpression of reggie-1 was shown to result in increase in the proliferation rate and invasive potential of melanoma cell lines (Hazarika et al., 2004). Our findings here provide a putative molecular explanation for the function of reggie-1 in cell adhesion and migration, which seems to be regulated by means of phosphorylation, endocytosis and proteolytic processing. Our recent data also imply that knockdown of reggies results in formation of aberrant focal contacts (T. Babuke and R.T., unpublished results), providing further evidence for an important role of reggies in the regulation of cell adhesion. However, further studies will be necessary to reveal the exact molecular chain of events that takes place during the course of these signaling processes.
Materials and Methods
Plasmid cloning and mutagenesis
Full-length rat reggie-1 cloned into the plasmid pEGFP-N1 (Clontech), SH-R1-EGFP and G2A-R1-EGFP have been described earlier (Neumann-Giesen et al., 2004). Further point mutations were generated using the QuikChange Site-Directed Mutagenesis Kit (Stratagene). Reggie-1 was cloned into pCMV-Tag5B vector (Stratagene), resulting in a C-terminal myc fusion tag. Y527F-Src, kinase-dead Src, Src and Fyn constructs were a kind gift from I. Dikic (University of Frankfurt, Germany).
Cell culture and transfection
HeLa and SYF (Src, Yes and Fyn deficient mouse embryonic fibroblasts; American Type Culture Collection) cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS; Invitrogen). PC12 cells were grown in RPMI-1640 supplemented with 10% FCS. All cell lines were transiently transfected with plasmids using Metafectene (Biontex, Munich, Germany) according to the manufacturer's instructions.
Knockdown of reggie-1 with siRNAs
HeLa cells grown to a confluency of 50% were transfected with siRNA using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Commercial control and reggie-1-targeting stealth siRNA duplexes (Invitrogen) were used. Three days after transfection, the cells were used for experiments and the knockdown was verified by western blotting of equal protein amounts of lysates from control and reggie-1 knockdown cells as well as by immunofluorescence.
Growth factor and inhibitor treatment
Stimulation of cells and treatment with inhibitors were generally carried out 24-48 hours post transfection. For detection of phosphotyrosines, the cells were, when indicated, incubated for 25 minutes at 37°C with a freshly prepared pervanadate solution (10 mM Na3VO4 and 300 mM H2O2), which was diluted in DMEM to a final concentration of 100 μM. In the cases where pervanadate treatment was omitted, the cells were alternatively treated with epidermal growth factor (EGF, 100 ng/ml, Sigma) for 5 minutes to monitor phosphorylation (when indicated). To inhibit Src family kinases the cells were incubated with 5 μM or 10 μM (as indicated) PP1, PP2 or the control compound PP3 or 10 μM SU6656 (all Calbiochem) in DMEM for 25 minutes. When indicated, the cells were subsequently incubated with both the inhibitor and pervanadate for a further 25 minutes. The EGF receptor was inhibited by treating the cells with 5 μM AG1478 (Biosource) in DMEM for 45 minutes at 37°C and subsequent incubation with AG1478 and pervanadate (+/–EGF) for additional 25 minutes. In the case of growth factor stimulation, cells were treated with 100 ng/ml EGF for the indicated duration.
For fluorescence microscopy analysis, reggie-1-transfected HeLa cells were starved in DMEM without FCS for 16 hours and then stimulated with 100 ng/ml EGF for either 5 or 15 minutes (see figure legends).
The ESA/flotillin-2 antibody was purchased from Transduction Laboratories and diluted 1:1000 for western blotting. A monoclonal Src antibody (Santa Cruz) was applied at 1:500 for western blotting and 1:100 for immunoprecipitation. For immunoprecipitation of EGFP fusion proteins, a polyclonal GFP antibody (Clontech) was applied (1:200), whereas for western blotting, a monoclonal GFP antibody (Roche Applied Sciences) was used (1:1000). The monoclonal phospho-Tyr antibody (Cell Signaling) was applied at 1:1000 for western blotting and 1:75 for immunoprecipitation. For detection of GAPDH in western blots a monoclonal antibody from Abcam was used and diluted 1:5000. For immunofluorescence analysis of myc-tagged reggie-1, a monoclonal myc antibody (Cell Signaling) was applied (1:500). For the detection of endogenous LAMP-3, a monoclonal LAMP-3/CD63 antibody (Santa Cruz) was applied (1:150). The monoclonal antibody (MAb108) for endogenous EGF receptor was a kind gift from I. Dikic (University of Frankfurt) and was applied at a final concentration of 10 ng/μl for immunofluorescence staining. The monoclonal antibody against EEA1 (Transduction Laboratories) was applied at 1:75 and the monoclonal antibody against transferrin receptor (Zymed) was diluted 1:100. The monoclonal primary antibodies used for immunofluorescence were detected with a Cy3-conjugated anti-mouse antibody (Jackson ImmunoResearch, 1:300).
Cells expressing EGFP fusion proteins were fixed with 3.7% paraformaldehyde and embedded in Gelmount (Biomeda) with 50 mg/ml 1,4-diazadicyclo(2,2,2)octane (Fluka). For immunofluorescence analysis, the cells were fixed with paraformaldehyde, permeabilized with 50 μg/ml digitonin, incubated in blocking solution (3% bovine serum albumin), labeled with primary antibodies (myc, EGF receptor or LAMP-3) for 1 hour and stained with a Cy3-conjugated secondary antibody for 45 minutes. Thereafter the cells were embedded as described above. To visualize the actin cytoskeleton, cells were fixed and permeabilized as described above and incubated with Alexa Fluor 594-conjugated phalloidin (Invitrogen) diluted 1:50 in blocking solution. The analysis was performed using a confocal laser-scanning microscope (Zeiss LSM510 Meta).
For detection of phosphorylated proteins, cells were lysed in immunoprecipitation buffer (10 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 0.5% Triton X-100, 60 mM N-octylglucoside, pH 8.0) supplemented with Protease Inhibitor Cocktail (Sigma) and pervanadate (1 mM) on ice for 30 minutes. Lysates were incubated by rolling at 4°C for 16 hours with 30 μl Dynabeads (Dynal) which had been coupled with the antibody. The beads were washed with 0.5 ml each of Neufeld buffer (10 mM Tris-HCl, 600 M NaCl, 0.1% SDS, 0.05% Nonidet P40, pH 8.5), IMM buffer (1% Triton X-100 and 0.5% sodium deoxycholate in PBS), IMM with 2 M KCl and 0.1× PBS. The precipitated proteins were solubilized into 1× electrophoresis loading buffer by incubation for 3 minutes at 90°C and analyzed by SDS-polyacrylamide gel electrophoresis.
For coimmunoprecipitation, cells were lysed in CoIP buffer (50 mM Tris-HCl, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P40, pH 7.4) supplemented with Protease Inhibitor Cocktail (Sigma) on ice for 30 minutes and the lysates were sonicated twice for 10 seconds. The lysates were precleared by rolling with 50 μl of Pansorbin beads (Calbiochem) twice each for 15 minutes at 4°C. After immunoprecipitation, the beads were washed five times with TBST (10 mM Tris, 150 mM NaCl, 0.05% Tween 20). Otherwise the samples were processed as described above.
Cell spreading assay
Cell spreading assays were carried out on four-well Lab-Tek II Chamber Slides (Nalgene Nunc International), which were coated with 20 μg/ml human fibronectin (Sigma) for 1 hour at room temperature and then blocked with 0.5% bovine serum albumin. HeLa cells were washed with PBS and resuspended in PBS with 0.05% EDTA. 100,000 cells were seeded per well and allowed to spread for 25 minutes at 37°C. The cells transfected with EGFP plasmids were washed with PBS, fixed with 3.7% paraformaldehyde and embedded in Gelmount (Biomeda) with 50 mg/ml 1,4-diazadicyclo(2,2,2)octane (Fluka). The cells transfected with siRNA duplex oligoribonucleotides were washed and fixed as described and thereafter stained with phalloidin to visualize the actin cytoskeleton and embedded as described above. The cells were scored as either non-spread, half-spread or spread.
For live imaging, HeLa cells were transfected with EGFP fusion proteins and seeded on thin-bottom glass chamber slides (Lab-Tek II, Nalge Nunc, Naperville, IL). Cells were starved of FCS for 16 hours and then stimulated with 100 ng/ml EGF in DMEM with 20 mM HEPES pH 7.4. To monitor the EGF-induced endocytosis of R1-EGFP (wild type or Y163F mutant), pictures were taken once every second (total of 300 frames) with a confocal laser-scanning microscope (Zeiss LSM510 Meta) using pinhole settings that correspond to an optical thickness z=1.5 μm. The imaging plane in the lower middle of the cells was chosen to make sure that endosomes are present in this section. Owing to the technical set-up, the starting point of imaging was about 2 minutes after addition of EGF. Movies were generated using the ImageJ program.
This study was supported by a Deutsche Forschungsgemeinschaft (DFG) SFB628 grant to R.T. We would like to thank Ivan Dikic (Frankfurt) for the generous gift of Src constructs and EGFR antibody and Metello Innocenti and Mika Ruonala for help with the construction of the movies.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/120/3/395/DC1
- Accepted November 13, 2006.
- © The Company of Biologists Limited 2007