Molecular Networks in FGF Signaling: Flotillin-1 and Cbl-Associated Protein Compete for the Binding to Fibroblast Growth Factor Receptor Substrate 2

Introduction

Fibroblast growth factor receptor substrate 2 (FRS2/FRS2α/SNT1) is a membrane linked docking protein originally identified as a protein that becomes tyrosine phosphorylated upon nerve growth factor (NGF) or fibroblast growth factor (FGF) stimulation in PC12 cells [1], [2], [3]. Together with FGF receptor substrate 3 (FRS3/FRS2β/SNT2), it belongs to the FRS adaptor protein family [4]. In this paper, we will use the name FRS2 for FRS2α/SNT1, and FRS3 for FRS2β/SNT2 for the sake of clarity.

FRS2 and FRS3 share a similar structure and 48% of amino acid sequence identity. In the N-terminus, they contain a consensus myristoylation sequence which is important for the membrane localization [1]. This sequence is followed by a phosphotyrosine binding (PTB) domain that is highly similar between the two proteins. The PTB domain binds specific peptides of certain receptor tyrosine kinases (RTKs) with or without tyrosine phosphorylated residues [5], [6].

FRS2 is ubiquitously expressed with the highest expression in brain, kidney, lung, ovary and testis and can be detected at all developmental stages of a mouse [7]. FRS2 knockout mice show embryonic lethality due to severe problems in gastrulation [8], [9], demonstrating how crucial FGF signaling is in animal development. In contrast to FRS2, the expression of FRS3 protein begins around day 9 and is restricted to tissues of neuronal origin [7], [10]. When exogenously expressed in FRS2-null mouse embryonic fibroblasts, FRS3 compensates for the loss of FRS2 by stimulating FGF induced activation of extracellularly regulated kinase (ERK), a member of the mitogen activated protein (MAP) kinase family [10].

Both FRS2 and FRS3 are tyrosine phosphorylated in response to NGF, FGF [9], [11] and glial derived neurotrophic factor (GDNF) [12], [13], but only FRS2 is threonine phosphorylated. Phosphorylation of 8 threonine residues in FRS2 occurs as a response to stimulation with FGF, epidermal growth factor (EGF), insulin and platelet derived growth factor (PDGF). This represents a negative feedback mechanism in which activated ERK inhibits further tyrosine phosphorylation of FRS2 by phosphorylating its threonine residues [14], [15]. FRS2 plays an important role in FGF dependent proliferation and migration of the cells [1] and in differentiation of PC12 cells by regulating sustained ERK activity upon FGF or NGF stimuli [1], [16], [17]. It is tyrosine phosphorylated upon insulin treatment, but its precise role in insulin signaling pathway remains largely unknown [18].

The C-terminus of FRS proteins bears multiple tyrosine phosphorylation sites (6 Tyr in FRS2), which, when phosphorylated by specific RTKs, recruit SH2-domain containing proteins such as adaptor protein Grb2 (4 Tyr in FRS2) and protein tyrosine phosphatase Shp2 (2 Tyr in FRS2) [1], [19]. The recruitment of Grb2 eventually results in a strong activation of PI3-kinase signaling and moderate activation of ERK pathway [9], [20], while phosphorylation and subsequent activation of Shp2 will result in a strong activation of ERK signaling [9].

The two members of the flotillin/reggie protein family, flotillin-1/reggie-2 (flot-1) and flotillin-2/reggie-1 (flot-2) are associated with specific membrane microdomains enriched in cholesterol and sphingolipids, also called rafts (For a review, see [21], [22]). Their membrane association is mediated by palmitoylation (both flot-1 and flot-2) and myristoylation (flot-2 only), but neither protein contains a transmembrane domain [23], [24]. Flotillins are widely expressed and well conserved between species, but their molecular function has remained somewhat enigmatic. Flotillins have been shown to participate in various signaling processes, including insulin and EGF receptor signaling [25], [26], [27], [28], in endocytosis [29], phagocytosis [30] and cell adhesion [27]. In addition, a role in neuronal regeneration has been suggested in goldfish and zebrafish [31], [32], but the evidence for such a role in mammalian animal models is lacking. Our earlier results show that flot-2 becomes Tyr phosphorylated by Src kinases and is endocytosed together with flot-1 in EGF stimulated cells [25], [27].

Although functional implications for flotillins in diverse cellular processes have been accumulating, in many cases the exact molecular mechanisms of flotillin action and the interaction partners remain to be identified. One of the known interaction partners of flotillins is the Cbl-associated protein (CAP), also known as Ponsin, which is an adaptor protein containing multiple SH3 (Src homology 3) domains and one or two sorbin homology (SoHo) domains [33]. During insulin receptor signaling, CAP is responsible for recruiting the ubiquitin ligase Cbl into membrane rafts by means of interacting with flotillin-1 [26]. Furthermore, CAP has been shown to be localized in focal adhesions and to mediate processes requiring actin remodeling [34]. Our recent findings also suggest that CAP is involved in signaling and becomes phosphorylated by the Abl kinase [35].

Here, we have identified the PTB-domain containing adaptor proteins FRS2 and FRS3 as novel interaction partners of flot-1. Membrane association and cellular localization of FRS2 was shown to be partly dependent on flot-1. We observed an interaction between flot-1 and FRS2 in mouse tissues. Intriguingly, FRS2 appears to be more Tyr phosphorylated in flot-1 knockdown cells, indicating that the interaction with flot-1 regulates the downstream signaling of FRS2. Moreover, since flot-1 binds to the PTB domain of FRS2, it competes for the binding with the FGF receptor. Furthermore, flot-1 is necessary for the signaling induced recruitment of FRS2 into lipid rafts. In addition to flot-1, FRS2 was also found to interact with CAP, which has been linked to flot-1 during insulin signaling [26]. The binding domains of these three proteins overlap, and thus flot-1 and CAP compete for the binding to FRS2. Here, we have dissected the putative role of these proteins during receptor tyrosine kinase signaling.

Results

In search of novel interaction partners of flotillins, we performed a yeast two-hybrid (Y2H) screen of a human brain library using rat flot-1 as a bait, in which several clones containing a partial sequence of FRS3 were obtained (data not shown). Since FRS3 shows a high homology to the more ubiquitously expressed FRS2, we tested the possibility of a direct interaction of flot-1 with FRS2 and characterized the interacting domains in a Y2H assay (Fig. 1A). The constructs used for this are shown in Suppl. Fig. S1A. Interaction was detected as growth of the transformed yeast colonies on nutrient deficient plates and blue colour (Fig. 1A). Full-length (FL) flot-1 was found to interact with the FL FRS2 and the PTB domain of FRS2, whereas no interaction was detected with the C-terminal part of FRS2. Interestingly, flot-2 did not interact with any of the FRS2 fragments. We also attempted to produce deletion fragments of flot-1 (see Suppl. Fig. S1A), but the C-terminal parts flot-1-CT (amino acids 226–428) and flot-1-CC (327–428) displayed autoactivity, resulting in blue colonies without any interaction partner. Furthermore, flot-1-NT (amino acids 1–253) was poorly expressed, preventing any conclusions about its interaction with FRS2. However, a deletion mutant of flot-1 missing 100 amino acids from the C-terminal end (Flot-1-STOP328) was unable to interact with any of the FRS2 fragments, indicating that the interaction domain may reside in the far C-terminus of flot-1. Suppl. Fig. S1B shows the expression in the yeast of the constructs used in Fig. 1A and a test for autoactivation in the yeast.

We next used GST-pulldown analysis to verify the interaction between FRS2 and flot-1 (Fig. 1B–C). FL FRS2, the PTB and CT domains were expressed as GST fusions and immobilized on glutathione beads (Fig. 1B). All three variants of FRS2 were capable of binding substantial amounts of flot-1 from HeLa cell lysates. To demonstrate a direct interaction, we used bacterially expressed, purified proteins. FL Flot-1-GST and Flot-1-CT-GST were immobilized on sepharose beads and tested for their binding to His-tagged FRS2 (Fig. 1C). FRS2 was clearly bound by FL and even more by the CT of flot-1. Thus, our yeast two-hybrid and pulldown data suggest that FRS2 interacts with flot-1 but not with flot-2, and the interaction involves both the PTB domain and C-terminus of FRS2 and the C-terminal part of flot-1.

To verify the interaction between flot-1 and FRS2 in vivo, we performed coimmunoprecipitation from mouse tissues (Fig. 1D). FRS2 was immunoprecipitated using same amounts of homogenized mouse tissues which express the highest amounts of FRS2 (brain, liver and kidney; see right part of Fig. 1D for relative expression levels). Coimmunoprecipitation of flot-1 with FRS2 was seen from all three tissues, whereas no precipitation of flot-1 was detected when an isotype-matched control antibody was used.

To study the possible colocalization of flot-1 with FRS2 in animal cells, we chose Hep3B cells which express high amounts of endogenous FRS2 and flotillins. Immunofluorescence staining of FRS2 and flotillins revealed that in serum-grown Hep3B cells, FRS2 was mainly localized at the plasma membrane with some diffuse and vesicle-like staining in the cytoplasm (Fig. 2). Flot-1 (Fig. 2, upper row) and flot-2 (Fig. 2, lower row) were both detected at the plasma membrane, where they colocalized with FRS2. However, little if any colocalization was detected in the endosomes, which contained high amounts of flotillins.

Our previous findings have shown that upon knockdown of flot-1 in HeLa cells, EGF-stimulated activation of MAP kinase signaling is inhibited (Amaddii et al., under revision). Furthermore, it has been shown that overexpression of FRS2 results in a dose-dependent enhancement of ERK activation upon EGF signaling [15]. Thus, we tested if overexpression of FRS2 would be able to compensate for the MAP kinase activation defects upon receptor tyrosine kinase stimulation in flot-1 knockdown cells. However, since the best characterized function of FRS2 lies within FGF receptor (FGFR) signaling, where it also mediates an increase of MAP kinase activation, we here used FGF stimulation instead of EGF. Phosphorylation of Akt and ERK2 after 5 min of FGF stimulation were found to be reduced in flot-1 knockdown cells as compared to the control cells (Fig. 3A), consistent with our previous findings (Amaddii et al., under revision). Overexpression of FRS2-CFP did not result in correction of the diminished phosphorylation of these proteins after FGF stimulation in flot-1 knockdown cells. Expression of FRS2-CFP and the knockdown efficiency of flot-1 were verified by means of Western Blot (Fig. 3A). We observed a considerable shift in the gel motility of FRS2 after FGF treatment, most likely due to phosphorylation, in both control and flot-1 knockdown cells. Thus, although the growth factor induced phosphorylation of FRS2 appears to take place in flot-1 knockdown cells, FRS2 is not able to compensate for the observed downstream signaling defects.

FRS2 binds to FGFR cytoplasmic domain by means of its PTB domain but independently of Tyr phosphorylation [6]. Since we could show that this domain also at least partly mediates the interaction to flot-1, we tested if FGFR and flot-1 would compete for the binding to FRS2 (Fig. 3B). For this, we used purified FRS2-GST immobilized on beads and lysates of Hela cells transfected with increasing amounts of myc-tagged FGFR (0.5–2 µg of DNA used for transfection, Fig. 3D). As a transfection control, the empty vector pcDNA3 was used. The binding of endogenous flot-1 to FRS2-GST in the presence of FGFR was found to decrease the more FGFR was present in the lysates (Fig. 3B), but we were unable to obtain a full loss of flot-1 binding by FRS2 due to FGFR. Significantly reduced binding of flot-1 to FRS2 was observed upon increased expression of FGFR (Fig. 3C).

For studies of the function of endogenous FRS2, which is expressed in HeLa cells only in very minor amounts, in the absence of flot-1, we generated stable Hep3B clones in which flot-2 or flot-1 were knocked-down using lentivirus-mediated RNA interference. It has to be noted that knockdown of flot-2 results in a concomitant destabilization and downregulation of flot-1 at the protein level, whereas flot-1 depletion only mildly affects flot-2 expression (Suppl. Fig. S1C), as shown by us and others for various cell lines [25], [36]. Suppl. Fig. S1D shows that very little flot-2 or flot-1 could be detected by means of immunofluorescence staining in the knockdown cells, demonstrating that a high degree of depletion of flotillins was achieved in virtually all cells of the respective clone.

To see if knockdown of flotillins affects the phosphorylation status of FRS2, we used the stable Hep3B knockdown cells. FRS2 was immunoprecipitated from serum-grown cells, and the blots were probed with anti-phospho-Tyr antibodies (Fig. 4A). Interestingly, we could detect a tendency to increased Tyr phosphorylation of FRS2 in flot-2 knockdown cells and a significantly increased phosphorylation in the flot-1 knockdown cells (Fig. 4B). Since Tyr and Thr phosphorylation of FRS2 have been described to be reciprocally regulated [14], [15], one would expect that the increased Tyr phosphorylation would result in decreased Thr phosphorylation of FRS2 in flotillin knockdown cells. We thus aimed at detecting the Thr phosphorylation of FRS2, which is mediated by ERK, by means of antibodies directed against phosphorylated recognition sites of the ERK kinase. Unfortunately, although a tendency to decreased Thr phosphorylation of FRS2 was indeed visible after depletion of flot-2 or flot-1, the signals were generally very weak and did not allow for any definite conclusions (Data not shown).

FRS2 has been described to associate with cellular membranes by means of myristoylation. Since myristoylation alone can only provide a weak membrane attachment [37], we studied by means of immunofluorescence studies if the cellular localization of FRS2 might depend on flotillins. Although FRS2 was localized at the plasma membrane in the control cells, knockdown of flot-1 or flot-2 resulted in a slight increase in the cytoplasmic staining and some loss of membrane localization of FRS2 (Fig. 4C), indicating that in the absence of flotillins, FRS2 is less membrane associated. In addition, some nuclear staining for FRS2 was detected in flot knockdown cells but not in the control. These results suggest that the membrane association of FRS2 may be facilitated by the presence of flotillins. Since flot-2 and flot-1 knockdown cells gave virtually identical results in these assays, it is likely that the changes in FRS2 membrane association are more attributable to the absence of flot-1 expression, considering that flot-2 knockdown cells also lack flot-1 (Suppl. Fig. S1B).

Previous data have shown that in some cell types, FRS2 is constitutively associated with membrane rafts [38], [39]. Thus, the interaction between flot-1 and FRS2 might facilitate the localization of FRS2 into these domains. To test this, we again used the Hep3B cells which were either serum starved or starved and then stimulated with pervanadate to enhance Tyr phosphorylation. Detergent resistant membranes (referred to as rafts below) were isolated using density gradients after detergent extraction [40]. In these gradients, rafts are mainly present in fractions 1–3, as evidenced by the enrichment of cholera toxin in these fractions (Fig. 5). Flot-1 was found to reside in fractions 2–5 in starved cells, whereas in pervanadate stimulated cells, a shift towards raft fractions was observed (Fig. 5, upper part). We were not able to detect FRS2 in the raft fractions in starved cells, whereas in pervanadate stimulated cells, some FRS2 was found in fractions 2–3 together with flot-1. Intriguingly, pervanadate did not induce a shift of FRS2 towards raft fractions in flot-1 knockdown cells (Fig. 5, lower part), implicating that flot-1 is necessary for the stimulation induced recruitment of FRS2 into detergent insoluble membranes in these cells. However, only a fraction of FRS2 becomes raft associated during signaling, contrary to previous findings from other cell types [38], [39].

Flot-1 has previously been described to interact with Cbl-associated protein (CAP). We thus tested in our raft fractions if the localization of CAP would be altered after flot-1 knockdown. In control cells, CAP displayed a broad distribution in the heavier fractions, without localizing to rafts. Although CAP underwent a shift towards lighter membranes upon pervanadate stimulation, it remained concentrated in fractions 5 and 6, without localizing to rafts (Fig. 5, upper part). In flot-1 knockdown cells, CAP was found to reside in fractions 5 and 6 both in starved and stimulated cells (Fig. 5, lower part). Thus, although CAP exhibits a flot-1 dependent localization in these gradients, flot-1 does not recruit CAP into rafts.

Previous findings have shown that during signaling of the neurotrophic receptor TrkA, flot-1, FRS2 and CAP all localized into rafts [38]. However, a direct molecular connection between these three proteins has so far not been established. Since flot-1 interacts with both FRS2 and CAP, we tested if FRS2 would be capable of interacting with CAP. Our Y2H analysis showed that FL CAP indeed interacted with FL, PTB and CT domain of FRS2, whereas the sorbin homology (SoHo) domain of CAP only showed an interaction with FRS2 PTB domain (Fig. 6A). Since both FRS2 and CAP posses multiple domains that mediate protein-protein interactions and both CT and PTB in FRS2 appeared to interact with CAP, we used GST pulldown to characterize the interaction domains more closely. Various GST fusion proteins containing CAP domains were generated (Fig. 6B). The purified GST proteins were incubated with purified FRS2-His. A strong interaction was seen between FL proteins (Fig. 6C and D). Interestingly, a CAP mutant lacking the SoHo domain (ΔSoHo) showed a similar binding to FRS2 as the FL CAP, whereas the SoHo-GST interacted less strongly (Fig. 6D). Of the three SH3 domains (src-homology domain 3) of CAP, only the third one (SH3C) was found to bind FRS2. Quantification of the binding showed that full-length CAP, ΔSoHo and SH3C all displayed a significant binding to FRS2 (Fig. 6E). These results imply that multiple domains in FRS2 and CAP are involved in their interaction. These in vitro results were verified by coimmunoprecipitation of the endogenous proteins from Hep3B cells, in which a strong coprecipitation was detected. A similar degree of coprecipitation was detected in control and flot-1 or flot-2 knockdown cells (Fig. 6F), although the coprecipitation may even be slightly increased in flot-1 knockdown cells.

Since the SoHo domain of CAP and the PTB domain of FRS2, which were found to interact, are the domains that also mediate the interaction of the respective proteins with flot-1, we tested the hypothesis that CAP and flot-1 might compete for the binding to FRS2. CAP-GST was immobilized on sepharose beads and incubated in the presence of increasing concentrations of purified FRS2-His, and the binding of endogenous flot-1 from HeLa lysates was tested (Fig. 7). The presence of increasing concentrations of FRS2-His indeed was capable of abolishing the interaction of flot-1 with CAP, demonstrating that rather than forming a trimeric complex, FRS2, flot-1 and CAP may compete for the binding of each other due to their partly overlapping interaction domains. Since the amount of FRS2 bound by CAP only modestly increases upon incubation with higher amounts of purified FRS2, whereas CAP-bound flot-1 is clearly reduced, FRS2 most likely binds to flot-1 in the lysate, prevents it from binding to CAP and keeps it in the soluble fraction.

Discussion

Here we have shown that the signaling adaptor protein FRS2 directly interacts both in vivo and in vitro with the membrane raft-associated flot-1. This interaction is mediated by the PTB domain and, to a lesser extent, the C-terminus of FRS2 and by the C-terminus of flot-1. We were able to coprecipitate flot-1 together with FRS2 from mouse tissue lysates, demonstrating that this interaction also takes place in vivo. Interestingly, flot-2, which shows a high homology to flot-1 and forms stable oligomeric complexes with it [25], does not appear to directly interact with FRS2. However, coprecipitation of flot-2 with FRS2 was detected from cell lysates, which is probably due to the strong association of flot-2 with flot-1 and not a result of a direct interaction with FRS2.

Further proof for the functional role of the interaction of flot-1 with FRS2 was provided by our results showing that depletion of flot-1 affects the cellular localization of FRS2 in that it appeared to be more soluble. Furthermore, in vanadate stimulated cells, a small fraction of FRS2 was recruited into detergent resistant membranes, which did not take place in the absence of flot-1. Even more importantly, FRS2 was not able to rescue the phosphorylation deficiency of ERK in flot-1 knockdown cells. Overexpression of FRS2 has previously been shown to result in increased ERK activation [15]. If the ERK activation pathways that flot-1 and FRS2 participate in were separate, one would expect that FRS2 overexpression would result in normal ERK phosphorylation. However, if FRS2 and flot-1 are functionally interconnected, FRS2 probably residing upstream of flot-1, no compensation would take place, as observed here. Thus, the interaction of FRS2 and flot-1 during FGFR signaling may be required for a proper activation of MAP kinases.

Flot-1 does not contain any typical PTB domain binding motifs (NPXY or NPXpY) although its interaction with FRS2 is to the most part mediated by the FRS2 PTB domain. Thus, other binding motifs must be responsible for the interaction. Consistently, the PTB domain of FRS2 can bind to both phosphorylated and nonphosphorylated sequences. It has been shown to interact with the Trk A receptor by means of a classical NPXpY motif, whereas its interaction with the juxtamembrane region of FGF receptor 1 involves an amino acid sequence lacking both Tyr and Asn residues [5], [6], [41]. However, Tyr phosphorylation was found to increase the binding of the full-length FRS2 but not of the PTB domain to FGFR1 [42]. We here observed an increased Tyr phosphorylation of FRS2 in the absence of flot-1. Previous studies have shown that Tyr and Thr phosphorylation of FRS2 are reciprocally regulated [14], [15], [39]. Unfortunately, we were not able to measure FRS2 Thr phosphorylation, which is mediated by ERK, but it is likely to be reduced. This may even be a direct consequence of the reduced ERK activity in flot-1 knockdown cells, since inhibition of ERK activity has been shown to result in decreased Thr and concomitantly increased Tyr phosphorylation of FRS2 [15].

Another novel interaction partner of FRS2 discovered in the present study is CAP, an adaptor protein with three SH3 domains. Both the PTB domain and CT of FRS2 were found to mediate this interaction, whereas in CAP, the SoHo domain, together with the third SH3 domain, seems to bind to FRS2. In our GST pulldown assays, deletion of the CAP SoHo domain did not result in a major reduction in binding to FRS2. Furthermore, both the FL CAP and the third SH3 domain showed about similar binding efficiency to FL FRS2, implicating a co-operative mode of binding. SH3 domains bind to proline rich sequences, several of which are present in FRS2, one of them even residing within the PTB sequence [2]. In the Y2H assay, the SoHo domain only bound to the PTB domain, whereas FL CAP bound to FRS2 CT as well. Thus, this would suggest that the SH3 domain of CAP may also be capable of interacting with the poly-Pro sequences in the more C-terminal part of FRS2.

Interestingly, CAP has previously been linked to flot-1 during insulin signaling [26]. Furthermore, it has been postulated that the activated NGF receptor TrkA is translocated into flotillin-containing membrane rafts. Differentiation and neurite growth of PC12 cells can be induced by NGF, which binds to its receptor TrkA and induces a signal transduction through rafts. In PC12 cells, a very transient association of TrkA with CAP, which in turn interacted with flot-1 with similar kinetics, was observed. However, CAP was detected in a constitutive complex with an adaptor protein APS [43], which has been shown to associate only with the activated TrkA [44]. This would suggest that during TrkA signaling, CAP associates with flot-1 and activated TrkA only very shortly, possibly mediating raft recruitment of TrkA, but then dissociates from them. Intriguingly, FRS2 has been shown to reside in the rafts in which the activated TrkA is recruited [38], and it is already long recognized that FRS2 is an important regulator of TrkA signaling [2], [3], [11], [14], [17], [38], [45], and may even participate in insulin signaling [18]. However, FRS2 has never been connected either with flot-1 or CAP before. Since flotillins were originally discovered as neuronal regeneration proteins [32], and knockdown of flotillins in neuronal N2A cells causes reduced neurite outgrowth [31], flotillins might indeed modulate neuronal differentiation together with FRS2 and CAP, most likely by regulating the activation of MAP kinases.

We here show that FRS2 can bind to both flot-1 and CAP, whereas previous studies have shown that CAP and flot-1 also interact with each other, and this interaction also requires the SoHo domain of CAP [26], [33]. The interaction of flot-1 with FRS2 may be required for the recruitment FRS2 to membranes or even membrane rafts, as has been suggested for the interaction of flot-1 with CAP [33]. This might in turn be necessary for a productive interaction of FRS2 with its signaling partners, such as FGF receptor. Although we cannot exclude that a trimeric complex between FRS2, flot-1 and CAP exists in the cells, the overlapping binding domains would rather suggest that these proteins compete for the binding to each other, as we also observed for the binding of FGFR and flot-1 to FRS2. Alternatively, FRS2, CAP and flot-1 may bind to each other in a successive manner, thus regulating the formation of downstream signaling complexes. In line with this, FRS2 has been shown to be connected by means of the growth factor binding protein-2 (Grb2) with the ubiquitin ligase cCbl during FGFR signaling [46]. On the other hand, cCbl regulates insulin receptor signaling by means of its interaction with CAP, which then binds to flot-1 [26]. Thus, the signalosomes that are formed by FRS2, flot-1 and CAP may vary according to the signaling receptor and the respective cellular response. Importantly, strong evidence connects all three proteins to MAPK activation, and FRS2 has been suggested to act as a scaffolder during MAPK signaling [1], [14], [34], [38], [47], [48], [49]. Our unpublished results imply that during EGF receptor signaling, flot-1 is found in complex with RAF, MEK and ERK kinases, suggesting a scaffolding function (Amaddii et al, under revision). Very interestingly, we identified another putative MAP kinase scaffolder, mitogen activated protein kinase organizer-1, MORG1 [50], [51] as a further interaction partner of flot-1 in our yeast two-hybrid screen (our unpublished results). Thus, the most likely mechanism how the interaction of FRS2 with flot-1/CAP modulates signal transduction is the regulation of MAP kinase function and the cellular activities resulting from MAP kinase activity, such as proliferation and differentiation.

Materials and Methods

Supporting Information