GGA3 Functions as a Switch to Promote Met Receptor Recycling, Essential for Sustained ERK and Cell Migration

Summary

Cells are dependent on correct sorting of activated receptor tyrosine kinases (RTKs) for the outcome of growth factor signaling. Upon activation, RTKs are coupled through the endocytic machinery for degradation, or recycled to the cell surface. However, the molecular mechanisms governing RTK recycling are poorly understood. Here, we show that Golgi-localized gamma-ear containing Arf-binding protein 3 (GGA3) interacts selectively with the Met/Hepatocyte Growth Factor RTK when stimulated, to sort it for recycling in association with “gyrating”-clathrin. GGA3 loss abrogates Met recycling from a Rab4 endosomal subdomain, resulting in pronounced trafficking of Met towards degradation. Decreased Met recycling attenuates ERK activation and cell migration. Met recycling, sustained ERK activation and migration require interaction of GGA3 with Arf6 and an unexpected association with the Crk adaptor. The data show that GGA3 defines an active recycling pathway and support a broader role for GGA3-mediated cargo selection in targeting receptors destined for recycling.

INTRODUCTION

Receptor Tyrosine Kinases (RTKs) control many aspects of cell behavior including proliferation, survival, differentiation and migration in response to their environment. Upon ligand binding, RTKs become catalytically active and tyrosine phosphorylated enabling the recruitment of signaling proteins to initiate downstream signaling cascades. This process is balanced by the simultaneous recruitment of endocytic proteins, which enhance RTK internalization, allowing for their removal from the cell surface and subsequent signal termination (Sorkin and von Zastrow, 2009). However, it is now recognized that internalization, in addition to regulating signal termination, is an integral part of signaling, controlling strength, spatial and temporal restrictions to RTK signals (Gould and Lippincott-Schwartz, 2009; Sorkin and von Zastrow, 2009). Thus a molecular understanding of the processes that regulate entry of RTKs into endocytic compartments is key to our understanding of a biological response.

The Hepatocyte growth factor (HGF) and its receptor, Met, are potent regulators of epithelial-mesenchymal transitions, cell scatter and invasion (Peschard and Park, 2007). During development, their action is essential for the growth and survival of placental trophoblasts, outgrowth of motor neurons and migration of muscle precursor cells (Bladt et al., 1995; Maina and Klein, 1999; Schmidt et al., 1995; Uehara et al., 1995). In the adult they coordinate wound healing in various organs such as the liver, heart and kidney (Borowiak et al., 2004; Huh et al., 2004; Kawaida et al., 1994; Nakamura et al., 2000). The chronic activation of Met is associated with several human tumors (Birchmeier et al., 2003). One mechanism involves mutations that impair trafficking of Met by limiting its access to the degradative compartment and resulting in sustained signaling (Abella et al., 2005; Kong-Beltran et al., 2006; Lee et al., 2000; Peschard et al., 2001). Since defects in cargo trafficking have now emerged as a common feature associated with several human diseases, a full understanding of the pathways that regulate RTK trafficking is essential.

Following ligand activation, RTKs, including Met, are internalized through clathrin-dependent or -independent mechanisms (Hammond et al., 2001; Orth et al., 2006; Sigismund et al., 2005), eventually converging to deliver cargo to early endosomes (Sorkin and von Zastrow, 2009). From here, RTK cargo is diverted towards one of two fates to be routed to late endosomes/lysosomes for degradation, or to be recycled back to the plasma membrane. Many studies have provided molecular insights into the details of how RTKs such as the EGFR (Haglund et al., 2003; Huang et al., 2006; Raiborg et al., 2002), and Met receptor (Abella et al., 2005; Hammond et al., 2003; Peschard et al., 2001) are targeted towards the degradative pathway, however, mechanisms that regulate and coordinate recycling pathways remain unclear.

Recycling of RTKs to the cell surface can occur either directly from the early endosome via a “fast route”, or indirectly through a “slow route”, traversing the endocytic recycling compartment (Grant and Donaldson, 2009). In general, control of vesicle trafficking depends on the Rab and ADP-ribosylation factor (Arf) small GTPases and their binding proteins (D’Souza-Schorey and Chavrier, 2006). Although activation of RTKs leads to activation of Arf and Rab GTPases (Kimura et al., 2006; Palacios et al., 2001), the mechanisms by which these proteins are coupled to, and regulate, RTK trafficking are incompletely understood.

The Golgi-localized, gamma-ear-containing, Arf-binding proteins (GGAs) are adaptor proteins, evolutionarily conserved from yeast to humans. The GGA family is comprised of three proteins in humans, GGA1, 2, 3 (Bonifacino, 2004). GGA proteins promote clathrin assembly and mediate intracellular transport of cargo, such as mannose-6-phosphate receptor (M6PR) and sortilin, as well as plasma membrane trafficking of Gag proteins required for HIV release (Nielsen et al., 2001; Puertollano et al., 2001a). Despite detailed structural data on the modular domains of the GGA proteins, less is known about the dynamics of GGA complexes that mediate transport events. GGA proteins have been observed on early endosomes (Puertollano and Bonifacino, 2004) and dynamic clathrin-coated structures positive for the transferrin receptor (TfR) (Zhao and Keen, 2008), yet the functional significance of this localization is poorly understood. These observations raise the question of whether GGA proteins regulate vesicle transport for a broader range of cargo than initially proposed.

Here, we report that GGA3 defines a recycling compartment for the Met RTK. GGA3 is essential for functional recycling of the Met RTK, sustained Met-dependent ERK activation and cell migration. Selective recruitment of GGA3 with an activated Met RTK occurs through the Crk adaptor protein following Met internalization and trafficking to a Rab4/Rab5-positive compartment. These results identify a key function for GGA3 and provide mechanistic insight into selective Met RTK recycling.

RESULTS

GGA3 is Recruited to an Activated Met RTK During Endocytosis

To investigate the relevance of GGA proteins to Met RTK trafficking, we first examined the ability of GGA proteins to associate with the Met RTK. Endogenous GGA3 was found to co-immunoprecipitate with endogenous Met RTK upon HGF stimulation of HeLa cells as early as 5 min; maximal association occurred by 15 min (Figure 1A). No association of Met was observed with endogenous GGA1 or GGA2, although lower levels of endogenous GGA1 were detected (Figure 1A). Consistent with recruitment of GGA3 to a Met RTK once internalized, GGA3 failed to localize with the Met RTK in the absence of HGF (Figure 1B). In response to HGF, a 1.4-fold increase in localization of GGA3 to endosomes was observed and at this time, GGA3 localized with Met positive punctae, typical of early endosomes, (Figures 1B and 1C).

Figure 1. HGF Regulates Recruitment of GGA3 to the Met RTK.

(A) 500μg of HGF-stimulated HeLa lysates were immunoprecipitated (IP) with anti-Met and IB was as shown. 50μg of cell lysate (Input) were similarly detected. (B) HeLa cells pre-treated with cycloheximide (CHX) for 2h were left unstimulated or stimulated 15 min with HGF and processed for immunofluorescence (IF) to localize Met (red) and GGA3 (green). Bar = 10μm. (C) HeLa cells transfected with GFP-Rab4 and stimulated 15 min HGF followed by IF processing. Bar = 10μm. Insets shows higher magnification of intracellular vesicles (Arrows indicate regions of co-localization). (D) Co-localization quantification of Met and GFP-Rab4 and GFP-Rab5 −/+ 15 min HGF. (E) Co-localization quantification of GGA3 and GFP-Rab4 and GFP-Rab5 −/+ 15 min HGF. (F) Quantification of GGA3 vesicles −/+ 15 min HGF. All values from 3 independent experiments (N=15). Student’s t-test; *P<0.02. See also Figure S1.

GGA3 Localizes with Met to a Rab4 Compartment

To better understand the GGA3-Met association, we sought to determine more precisely the intracellular localization of GGA3 within the endocytic network. A subset of GFP-tagged Rab proteins were expressed and used as markers to distinguish subdomains of endocytic compartments (Stenmark, 2009). Quantitative co-localization analysis revealed that GGA3 localizes predominantly with early Rab5 and Rab4 positive vesicles (~43.6% and ~67.7%, Figure S1A), while less co-localization was observed with the late markers Rab7 or Rab11 (~36.5%, and ~35.1%, Figure S1A). At 15 min post-HGF stimulation, Met localizes to both Rab4 and Rab5 positive vesicles (~55.3% and ~54.6% respectively) (Figures 1C, 1D and S1B), consistent with overlap of these Rab microdomains on early endosomes (Stenmark, 2009). Moreover, in response to HGF, GGA3 shows a preferential increase to Rab4 over Rab5-positive vesicles (1354 versus 913/ per cell, Figures 1E, 1F and S1C), giving an overall 1.48-fold increase of Met and GGA3 to Rab4 over Rab5-positive vesicles. Together this supports a possible role for GGA3 as an endocytic adaptor for Met.

GGA3 Knockdown Promotes Rapid Degradation of the Met Receptor

To understand the role of GGA3 on Met receptor function, the effect of RNAi-mediated knockdown (KD) of GGA3 on the trafficking and degradation of Met was analyzed. In HeLa cells depleted for GGA3, the half-life of the Met receptor was 50% that of control cells, (T/2 ~ 54.5 min compared to 106.7 min, Figures 2A, 2B). Comparable results were obtained using three independent siRNA duplexes, confirming specificity of the KD (Figure S2A). Following 15 min of HGF, Met internalizes into endosomes, as visualized by immunofluorescence (Figures 1B and S2B). At this time point, no detectable differences were observed in Met localization in GGA3 KD cells versus control cells (Figure S2B). However, consistent with the observed increased rate of Met degradation, by 60 min post HGF stimulation, a decrease in Met protein signal is apparent in GGA3 KD cells compared to control cells (Figure S2B). The increase in Met degradation following GGA3 KD is not due to increased Met protein turnover under basal conditions, as treatment with the protein translation inhibitor, cycloheximide had no effect (data not shown). Additionally, synthetic transport of Met to the plasma membrane was similarly not affected by GGA3 KD, using a 20°C temperature block and release to follow TGN export (Figure S3A). Moreover, no significant differences in Met receptor ubiquitination (ratio Ub/Met of 2.1 and 1.0 in control versus 2.6 and 0.9 in GGA3 KD cells at 5 min and 30 mins post-HGF stimulation) was observed in GGA3-depleted cells following HGF stimulation, arguing against an increase in ubiquitination as a cause for enhanced Met RTK degradation (Figure 2C).

Figure 2. GGA3 KD Enhances HGF-Induced Met Protein Degradation.

(A) HeLa control (CTL) or GGA3 KD cells were stimulated with HGF in the presence of CHX as indicated. IB was as shown. See also Figure S1. (B) Densitometric analysis of Met blots from 4 independent experiments like those shown in (A). Values were used to fit to an exponential decay and half-life value (t_1/2) was calculated (Microsoft Excel). Results expressed as mean values ± SEM. (C) 500μg HGF-stimulated CTL and GGA3 KD HeLa cell lysates were immunoprecipitated (IP) with anti-Met and IB was as shown for IP and 50μg for input. Densitometric ratio of Ub/Met levels is indicated below blot. (D) CTL and KD cells transfected with GFP-Rab7 stimulated 15 min with HGF and processed for IF using anti-Met. Quantification of Met and Rab7 co-localization from at least 3 independent experiments (N=30). Bar = 10μm. See also Figures S2 and S3.

In contrast to GGA3 KD, depletion of the ESCRT component, Tsg101 resulted in enhanced stabilization of Met protein levels in response to HGF (Figure S2C). This is in agreement with previous studies for a role of ESCRT complexes promoting Met degradation (Hammond et al., 2003), dependent on Met ubiquitination (Abella et al., 2005). Hence, under these conditions, GGA3 KD has an opposite effect to Tsg101 KD on Met stability, supporting a role for GGA3 to abrogate Met trafficking to the degradative compartment. In support of this, by 15 min of HGF treatment, the co-localization between the Met RTK and the late endosomal marker, Rab7, increased by 26% in GGA3 KD cells versus control cells (Figure 2D). Taken together, this data supports enhanced entry of the Met RTK into the canonical degradative pathway in the absence of GGA3.

GGA3 Knockdown Decreases Met Recycling

The HGF-dependent co-immunoprecipitation of endogenous Met and GGA3 proteins at 5 min, and their co-localization to Rab4/Rab5 positive vesicles, suggests that GGA3 could serve as an adaptor to recruit Met into the Rab4 positive tubular domains of the early endosome to promote recycling. Hence, we addressed whether GGA3 KD decreases entry of Met into recycling vesicles. To establish that Met recycles, we employed a thiolcleavable amine-reactive biotinylation reagent to label and follow the recycling pool of Met. Cells were surface-labeled with Sulfo-NHS-SS-biotin at 4°C, and internalization initiated by incubating cells for 7 min with HGF at 37°C to allow Met to accumulate in early endosomes. During cell surface biotinylation and chase, no significant differences in the amount of labeled internalized Met receptors were observed following HGF-treatment (7min), indicating that Met internalization occurs at a similar efficiency in GGA3 KD and control cells (data not shown). In contrast, GGA3 KD cells displayed reduced levels of Met returning to the cell surface (~9% compared to ~32% in control cells, Figure 3A), supporting a role for GGA3 in Met recycling.

Figure 3. GGA3 Mediates Recycling of the Met RTK.

(A) CTL and GGA3 KD cells were surface-labeled on ice with Sulfo-NHS-SS-biotin, stimulated 7 min with HGF at 37°C, and biotin from remaining cell surface receptors was removed by MesNa treatment at 4°C. Cells were then rewarmed to 37°C for the indicated times to allow recycling, followed by a second reduction with MesNa. The amount of recycled Met receptor is expressed as the percentage of the pool of biotinylated Met during the internalization period as described in experimental procedures. Values are mean ±5 SEM of 5 independent experiments. Student’s t-test; **P< 0.006, *P<0.05. Representative IB is indicated below graph. (B) IF of CTL or KD cells pulsed for 5 min with HGF at 37°C to internalize Met receptors in early endosomes (0’chase), rapidly washed at 4°C to remove unbound ligand and chased for 20 min to allow recycling. Met (white) and DAPI (blue). Bar = 10μm. (C) Quantification of (B). Bar graph represents cells showing a greater ratio of endosomal (EN) over plasma membrane staining of Met. Student’s t-test; *, P < 0.01. KD levels as assessed by IB. See also Figure S4. (D) Live cell imaging of association of GFP-GGA3 structures (green) with internalized Alexa555-HGF (red). Continuous image streams of the boxed regions (see Movie S1) display examples of internalized HGF with one or more surrounding dynamic GGA3 structures. See also Movie S2. (E) Quantitation of Alexa555-HGF signal intensity during a chase period (15 min load, 4-6 min chase) in structures with or without nearby dynamic GFP-GGA3 (16 spots each from four different experiments); slope of decrease in structures with nearby GFP-GGA3 is −0.34 ± 0.13 SEM.

To further address a role for GGA3 in Met recycling, an immunofluorescence-based assay, adapted from those previously established for the TfR (Driskell et al., 2007), was performed. Following a brief 5 min pulse with HGF, washout, and 20 min chase, the majority of the Met receptor was localized at the plasma membrane or in small endosomes in control cells (Figure 3B). By contrast, GGA3 KD cells showed accumulation of the Met receptor in large endosomes, some of which had already reached a perinuclear compartment, consistent with decreased entry into a recycling compartment (Figures 3B and 3C). Cell surface levels of Met were also reduced following HGF pulse/chase (~46%) in GGA3 KD cells, compared to control cells (~64%), when measured using flow cytometry (Figure S4A). Under similar pulse/chase conditions, KD of Rab4A also altered the distribution of the Met receptor towards larger endosomes, albeit to a lesser extent than GGA3 KD (Figure S4B). This supports a role for Rab4 in Met recycling and is in keeping with Met accumulation to Rab4 but not Rab11 positive recycling compartments when Rab proteins were over-expressed (Figure S4C).

To probe the relationship between structures containing Met and GGA3, we used live cell microscopy. COS1 cells transiently expressing GFP-GGA3 were incubated with Alexa555-HGF for 20 min to track Met receptor positive vesicles. Numerous examples of vesicular Alexa555-HGF with coincident or nearby GFP-GGA3-coated membranes were observed (Figure 3D). Examination of these regions using simultaneous two-color streaming (continuous) imaging reveals fast-moving GFP-GGA3 structures around most of the Alexa555-HGF spots (Movie S1). Notably, the Alexa555-HGF content in endosomes with overlapping dynamic GFP-GGA3 structures declined over time, while in isolated endosomal structures devoid of these GGA3-positive structures the Al555-HGF intensity remained steady or increased over time (Figure 3E). Moreover, the majority of fast-moving GFP-GGA3 structures (92%, total of 257 GFP-GGA3 spots counted) contain detectable mCherry-tagged clathrin (Movie S2), though not all clathrin spots (e.g., coated pits) contain GGA3. Thus, the GFP-GGA3 structures appear analogous to the “gyrating” clathrin- and GGA1-containing structures implicated in TfR recycling observed previously (Zhao and Keen, 2008), and support the concept that the dynamic GGA3 structures function as recycling tubules in association with endocytic HGF-Met complexes. Together these multiple approaches provide evidence for GGA3 functioning as an adaptor involved in Met recycling.

GGA3 Knockdown Attenuates Met-Dependent ERK Signaling and Cell Migration

Activation of Met leads to the formation of intracellular signaling complexes and induces cell motility, invasion, and branching tubulogenesis (Maroun et al., 2000). Thus, altered Met recycling due to GGA3 KD may impact on downstream signaling. To test this, we examined the phosphorylation status of Akt and ERK1/2 in response to HGF, as a readout for activation of the PI3K and MAPK pathways, respectively. While no significant differences were observed in p-Akt levels, the duration of p-ERK1/2 was markedly attenuated in GGA3 KD cells during 4h of HGF treatment, (Figures 4A and 4B). This supports the observed rapid degradation of Met under these conditions (Figure 2A and 2B) and our previous data that Met can activate ERK1/2 but not PI3K from an endosomal compartment (Abella et al., 2005). Consistent with only transient activation of ERK1/2, reduced nuclear localization of p-ERK1/2 was observed after 60 min of HGF stimulation in GGA3 KD cells (Figure 4C). Sustained ERK1/2 signaling is a prerequisite for HGF-induced cell migration (Maroun et al., 2000). Significantly, GGA3 KD, reduced cell migration to 26% of control cells in response to HGF (Figures 4D and 4E), and correlated with reduced localization of Met towards actin-rich membrane ruffles (Figure 4F). This is in accordance with defects in Met recycling (Figure 3A) and our observations that ERK1/2 activation is required for lamellipodia formation and cell migration downstream of HGF (Maroun et al., 2000).

Figure 4. Loss of GGA3 Attenuates HGF-Induced ERK Signaling and Migration of HeLa Cells.

(A) CTL and KD cells treated with HGF and CHX as indicated. IB was as shown. (B) Representative Odyssey infrared (IR) analysis of p-ERK1/2 levels (top panel) and p-Akt levels (bottom panel). (C) CTL and KD treated 60 min with HGF then fixed and stained for p-ERK1/2 (white) and DAPI (blue). Bar = 10μm. (D) Representative phase-contrast images (10x) of migration assays using CTL and KD cells −/+ HGF. (E) Quantification of experiments (N=3) shown in (D) using Scion Image. Student’s t-test; *, P < 0.007. Inset shows typical IB of GGA3 expression levels. (F) CTL and GGA3 KD cells either left untreated (left) or stimulated with HGF 60 min then stained for Met, phallodin (actin) and DAPI. Arrows point to leading edge. Bar = 10μm.

Arf6 is Required for GGA3-mediated Met Recycling

To establish the mechanism through which GGA3 regulates Met recycling, we utilized a structure function approach to uncouple GGA3 from specific binding proteins. GTP-bound Arf proteins interact with GGA proteins (Boman et al., 2000; Dell’Angelica et al., 2000). To test a requirement for GGA3-Arf interactions in Met recycling, stable HeLa cell lines expressing either siRNA-resistant GFP-GGA3, or GFP-GGA3 N194A, a mutant that specifically uncouples GGA from interaction with Arf-GTP proteins (Puertollano et al., 2001b), were generated. Upon siRNA transfection and KD of endogenous GGA3, expression of GFP-GGA3, but not GGA3-N194A, restored Met recycling back to the plasma membrane (Figure 5A and 5B). Moreover, restoration of Met trafficking to a recycling compartment, resulted in enhanced Met stability, prolonged ERK1/2 phosphorylation and cell migration in response to HGF (Figures 5C, 5D and S5A). Hence, an interaction between GGA3 and Arf-GTP is required for the regulation of Met RTK recycling by GGA3.

Figure 5. GGA3-mediated Met Recycling Requires Arf6.

(A) Stable HeLa cell lines expressing siRNA-resistant GFP-GGA3 (WT) or GFP-GGA3 N194A left untreated or pulse/chased with HGF were processed for IF using anti-Met. Bar = 10μm (B) IB of KD and rescue levels of stable HeLa cell lines compared to parental cells. Bar graph represents cells showing a greater ratio of endosomal (EN) over plasma membrane staining of Met for CTL, GGA3 KD (KD), GGA3 WT rescue (WT) or GGA3 N194A rescue (N194A) following IF recycling assay. Student’s t-test; *, P < 0.0001. (C) (Top panel) CTL, KD, WT and N194A cells were stimulated with HGF and used for IB as indicated. (Bottom panel). Representative Odyssey IR imaging quantification of levels shown in Top panel. (D) Migration assay of CTL, KD, WT and N194A cells −/+ HGF, N=3. See also Figure S5A. (E) HeLa cells stimulated with HGF used for GGA pulldown assay (GGA-PD) followed by IB for Arf6 to assess GTP-loaded Arf6. 10% input shown. (F) HeLa cell lysates stimulated 15 min HGF were used in GGA3 pulldowns using either WT or N194A GST-GGA proteins followed by IB for Arf6 and GST levels. 10% input shown. (G) HeLa cells transfected with GFP-Rab4 and HA-Arf6 WT stimulated 5 min HGF then processed for IF using anti-HA and anti-Met. Regions highlighted by boxes a and b are shown at higher magnification. Bar = 10μm. See also Figure S5.

In order to identify which Arf family member is responsible for GGA3-mediated recycling of the Met RTK, specific siRNA-depletion of Arf1, Arf3 and Arf6 was performed and cells were assessed for their ability to support Met recycling (Figure S5B and S5C). KD of Arf1 resulted in accumulation of the Met receptor in a cycloheximide-sensitive intracellular compartment, suggestive of a secretory-defect from the Golgi (Figure S5B). This is in agreement with previous data that Arf1 can affect secretory traffic (Volpicelli-Daley et al., 2005). No observable change in Met transport in response to HGF was observed following Arf3 KD (Figure S5B). In contrast, upon HGF treatment following Arf6 KD, Met localized predominantly to endocytic vesicles rather than recycling to the plasma membrane as observed in control cells (Figure S5B), while KD of Arf1 and Arf3 did not result in endosomal accumulation of the Met RTK under similar conditions (Figure S5B). In support of a role for Arf6 in Met trafficking, Met localized with both WT-Arf6 and constitutively active Arf6Q67L in tubulo-vesicular structures (Figure S5D). Moreover, activation of endogenous Arf6 was observed by 5 to 15 min post HGF stimulation (Figure 5E) coincident with the ability of WT but not the GGA3 N194A mutant, to interact with GTP-Arf6 (Figure 5F). Furthermore, endosomal structures triple-labelled for Met, Arf6 and Rab4 were observed in response to HGF (Figure 5G). Taken together these data support that GGA3 and Arf6 function together to enhance recycling of Met from a Rab4 positive compartment.

GGA3 binds the Crk Adaptor

Previously identified GGA cargo, such as the M6PR, contain acidic cluster dileucine motifs (DXXLL), which directly interact with the VHS domain of GGA proteins (Bonifacino, 2004). However, the intracellular domain of Met lacks putative, consensus dileucine motifs. To understand mechanisms through which GGA3 could be recruited to the Met RTK, we analyzed the protein sequence of GGA3 for predicted binding sites of known proteins recruited to Met. Using Scansite (Obenauer et al., 2003), two putative Crk SH3 domain proline-rich binding sites (prolines 404 and 463) were identified within the hinge segment of GGA3 (Figure 6A). Consistent with this prediction, endogenous GGA3 co-immunoprecipitates with Crk in the absence of HGF stimulation (Figure 6B). Hence, at steady-state these proteins can exist in a complex. Furthermore, glutathione-s-transferase (GST) fused Crk, Crk-SH3 domain, but not GST protein alone, was able to pull-down GGA3 protein (Figure S6A and data not shown). While mutagenesis and substitution of either predicted Crk proline binding site for alanine residues failed to significantly abrogate this association (Figure 6C), substitution of both prolines significantly decreased the association between GGA3 and Crk (Figure 6C and S6B). This identifies prolines 404 and 463 as components of Crk binding sites on GGA3.

Figure 6. Crk Association with GGA3 is Required for Met Association and Downstream Biological Processes.

(A) Schematic diagram of GGA3 protein domains and putative proline-rich Crk SH3 binding sites. (B) 1mg protein lysates from HeLa cells −/+ 15 min HGF were used for IP using anti-Crk (IP: Crk; Ab= antibody present). 50μg input shown. See also Figure S6. (C) Co-IP of HA-Crk from 500μg HEK293 cells transfected with HA-Crk and GFP-tagged GGA3 constructs as indicated followed by IB. 50μg input shown. (D) HeLa cells transfected with CTL or Crk siRNA, −/+ 15 min HGF and used to IP (1mg) using anti-Met. 50μg input shown. (E) IB of ΔCrk stable expression levels. (F) (Top panels) Control, GGA3 KD or ΔCrk stable cells stimulated with HGF. (Bottom panels) Representative Odyssey IR imaging quantification. (G) Migration assay shown in (F) −/+ HGF, N=3. See also Figure S6C.

A GGA3-Crk interaction is required for Met recycling, ERK activation and cell migration

In support of a requirement for Crk in the recruitment of GGA3 to a Met complex, reduced association between Met and GGA3 was observed in Crk KD cells following HGF treatment (Figure 6D). To test the specific requirement of GGA3-Crk interactions, stable cell lines expressing siRNA-resistant GGA3 containing the double proline mutant (referred to as ΔCrk) were generated (Figure 6E). Importantly, expression GGA3ΔCrk was unable to compensate for the depletion of GGA3 in promoting sustained ERK activation, Met stability and cell migration in response to HGF (Figures 6F, 6G and S6C). Hence this data demonstrates a requirement for a specific GGA3-Crk complex, in acting as a conduit to the Met receptor, thereby promoting Met recycling, sustained ERK activation and cell migration.

Arf6 and Crk co-operate to recruit GGA3 to Met –positive endosomal membranes

As published previously, GGA3-N194A, which fails to bind GTP Arfs is diffusely cytosolic and does not detectably localize to endosomal membranes (Puertollano et al., 2001b). However in response to HGF, a proportion of GGA3-N194A could associate with endosomes (Figure 5A), suggesting that additional factors besides Arf GTP binding can promote endosome recruitment of GGA3. As GGA membrane recruitment has been proposed to occur via multiple low-affinity interactions between Arfs and cargo (Wang et al., 2007), we tested the a requirement for GGA3-Crk binding to act as such a factor during GGA3 recruitment in response to HGF. To this end, the subcellular localization of GGA3-N194A, GGA3ΔCrk and a GGA3-N194A/ΔCrk double mutant to Met-positive endocytic vesicles were scored in response to HGF (Figure 7A and 7B). While HGF dependent recruitment to Met-positive vesicles of the GGA3-N194A mutant was decreased by ~44%, recruitment of GGA3-N194A/ΔCrk was reduced by ~88%. Thus, in response to HGF, the Crk-binding sites of GGA3 can compensate for Arf-binding and both Arf-GTP and Crk binding are required for full recruitment of GGA3 to endosomes. These data support a model whereby dual recruitment of Arf-GTP and Crk promote HGF dependent targeting of GGA3 to Met-positive endosomes and Met recycling (Figure 7C).

Figure 7. Arf6 and Crk co-operate to recruit GGA3 to HGF-induced endosomal membranes.

(A) HeLa cells transfected with the indicated GFP-GGA3 constructs were pre-treated with CHX (2h), then stimulated with HGF (15 min) and processed for IF. Box region was used to perform line scan analysis of fluorescence intensity versus distance. (B) Quantification of cells exhibiting endosomal (EN) co-localization of GFP and Met, from 3 independent experiments (N=75). Student’s t-test; **, P < 0.005. (C) Model for GGA3-mediated recycling of Met RTK.

DISCUSSION

Although it is generally considered that a portion of internalized RTKs recycle following ligand-dependent internalization, the mechanisms that regulate entry of RTKs and other cargo into the recycling compartment, rather than the degradative compartment are poorly understood. Here we identify an active RTK recycling pathway by which GGA3 functions as a specific cargo adaptor to target the Met RTK into recycling tubules. This was established by analyzing Met internalization, recycling and degradation under conditions in which GGA3 levels were depleted and rescued with various GGA3 mutants, and coupling the outcomes with Met dependent signaling and migration. These multiple approaches yielded quantitative, complementary results that support a model whereby GGA3 is recruited to an activated Met RTK cargo complex present within the early tubular endosomal network via Crk and Arf6. The formation of a GGA3-Met complex localized to a Rab4 enriched endosomal compartment, would promote access of Met into a recycling pathway while decreasing entry of Met into the degradative pathway. GGA3-dependent entry of Met into the recycling pathway, promotes sustained ERK1/2 activation and relocalization of Met towards the leading edge to initiate localized signaling required for cell migration.

At steady-state, GGA3 is predominantly localized to the trans-Golgi network, (Boman et al., 2000; Dell’Angelica et al., 2000), however a subpopulation of GGA3 localizes to vesicles (Wakasugi et al., 2003), previously defined as early endosomal in nature (Puertollano and Bonifacino, 2004). This is consistent with our observations that GGA3 is enriched in Rab4, and to a lesser extent, Rab5 -positive early endosomes and co-localizes with endocytosed Met cargo in these endosomes. Rab4 is a regulator of recycling vesicle formation at the early endosome (Sonnichsen et al., 2000; van der Sluijs et al., 1992), consistent with our functional assignment of GGA3 as an early recycling adaptor for Met. Furthermore, GGA3 may spatially restrict Met accessibility within the early endosome by modulating the Rab-based protein machinery. In this regard, GGA3 binds Rabaptin-5 (Mattera et al., 2003), which can complex with Rab4 and Rab5 (Vitale et al., 1998) as well as the GDP/GTP exchange factor, Rabex-5 (Horiuchi et al., 1997). Interestingly, GGA1 has been observed in association with clathrin on dynamic and rapid recycling structures (Zhao and Keen, 2008). Similarly in live cells, we observe GGA3 and clathrin in dynamic structures surrounding endocytosed HGF-Met complexes. Although we failed to observe GGA1 recruitment to Met, given the high degree of structural homology of GGA family proteins, these results support a role for other GGA family proteins in rapid recycling pathways.

The Met RTK lacks traditional DXXLL GGA-binding motifs. We identified an alternative mechanism through which GGA3 is recruited to Met, involving the constitutive interaction of GGA3 with the Crk adaptor. Although Crk does not have direct binding sites on Met, the major substrate for Met, the scaffold protein Gab1, contains six Crk SH2 domain phosphotyrosine binding sites and robustly recruits Crk to Met in response to HGF (Lamorte et al., 2002). GGA3 recruitment via Crk provides a mechanism for ligand-dependent specificity of engagement with the Met RTK complex. As neither GGA1 nor GGA2, contain these Crk binding proline rich motifs, and fail to associate with Crk (data not shown), this provides an explanation why neither of these proteins were observed to be recruited to a Met complex.

Key to the mechanism by which GGA3 regulates Met recycling is the finding that coupling of GGA3 to both Crk and Arf6 is necessary for efficient Met recycling. Arf6 is activated downstream from the Met RTK (Palacios and D’Souza-Schorey, 2003) which we show corresponds to the time kinetics of GGA3 recruitment and Met recycling. Additionally, Arf6 KD attenuates recycling of Rac-positive endosomes to the plasma membrane in response to HGF (Palamidessi et al., 2008). The finding that the uncoupling of GGA3 from Arf (N194A mutant) results in partial recruitment of GGA3 (~50%) to endosomes, indicates that initial membrane recruitment of GGA3 to Met can also occur in an Arf-independent manner. These data therefore support a model whereby activation of Arf6 by Met could aid in retaining a GGA3-Crk-Met complex in endosomal recycling membranes or serve to recruit other factors required for mediating Met recycling.

Internalized RTKs can continue to signal from endosomal compartments and it is now recognized that the ability of endosomes to serve as an intracellular signaling platform is an important component of the RTK signaling cascade (Gould and Lippincott-Schwartz, 2009). Sustained ERK1/2 signaling is required for Met induced cell migration (Maroun et al., 2000). Our results point to a mechanism through which Met promotes prolonged ERK1/2 activation and cell migration. GGA3 dependent entry of Met into a recycling network, rather than the degradative pathway, allows for prolonged activation of ERK1/2 from endosomes. GGA3 dependent recycling also localizes Met to regions of the plasma membrane that are required for actin dynamics and cell motility. Somewhat similar to HGF, the bacterial protein, InlB can activate Met to trigger actin remodeling and Listeria monocytogenes internalization (Hamon et al., 2006), which requires several endocytic proteins (Veiga and Cossart, 2005) including GGA3 (personal communication, Esteban Veiga and Pascale Cossart). Thus in the context of Listeria entry, GGA3 may be required to recycle Met signaling complexes to sites of bacterial entry for phagocytic uptake.

Unlike Met, GGA3 KD was associated with a delay in degradation of internalised EGF (Puertollano and Bonifacino, 2004). In a similar manner to the EGFR, ligand-dependent degradation of Met is dependent on ubiquitination of the Met receptor (Abella et al., 2005; Peschard et al., 2001). In contrast, GGA3 KD, promotes ligand-dependent degradation of the Met RTK, supporting a distinct role for GGA3 in Met trafficking and recycling. The difference observed between the role for GGA3 in trafficking of these two RTKs may thus reflect their differential ability to recruit Crk and undergo recycling. In addition to the indirect recruitment via Grb2 (Lock et al., 2002) of the major Crk binder, Gab1, Met contains a direct binding motif for Gab1 which may enhance the ability of Met to engage with a GGA3-Crk complex. In a similar manner to Met, a role for GGA3 in the exocytosis of retroviral Gag proteins is independent in the ability of GGA3 to bind ubiquitin but requires the ability of GGA3 to bind Arf proteins (Joshi et al., 2008).

The unexpected observation that GGA3-N194A can still be recruited to Met-positive endosomes lead us to test whether GGA3-Crk binding was involved in the endosomal association of GGA3. Precedence for coincident detection between clathrin adaptors and their interactors in mediating membrane recruitment has previously been established (Wang et al., 2007). As mutation of both the Arf-GTP and Crk binding sites were necessary to abolished recruitment of GGA3 to endosomes, this identifies Crk as a key player in GGA3 endosomal recruitment.

Our results clearly establish that, GGA3 coordinates the recycling, signaling and degradative fates of the Met RTK. Recycling in recent years has emerged as a mechanism to spatiotemporally coordinate localized signaling complexes, actin dynamics and directed cell movement downstream of motogenic stimuli and their receptors, such as Met (Jekely et al., 2005; Palamidessi et al., 2008; Wang et al., 2006). Given the importance of Met and RTKs in cancer progression, it will be important to assess the role of GGA3 in these processes.

Experimental Procedures

Chemicals, DNA constructs, Antibodies and Cells

Detailed list of chemicals, antibodies and DNA constructs is described in Supplemental Experimental Procedures. HeLa, HEK293 and COS1 cells cultured in DMEM containing 10% FBS. Transient transfections in HEK293 and HeLa cells were performed using Lipofectamine Plus according to manufacturer’s instructions (Invitrogen).

Biochemical Assays

Lysis and immunoblotting (IB) were as described in (Parachoniak and Park, 2009). For Met degradation assays, cells were stimulated with 0.5nM HGF containing cycloheximide at 37°C for the indicated time points. For blots requiring quantification, membranes were blocked with LI-COR blocking buffer (LI-COR Biosciences), incubated with primary antibodies as above, followed by incubation with IR-conjugated secondary antibodies prior to detection and analysis on the Odyssey IR imaging system (LI-COR Biosciences).

Colocalization Studies

IF assays were performed as described in (Parachoniak and Park, 2009). For co-localization quantification, MetaMorph software was used for object based co-localization measurements. Images were smoothed with a 3×3 lowpass filter and endosomes identified and counted using size estimates and intensity thresholds in each image set using the “Count Nuclei” application. Binary images were created for each set of endosomal spots and combined pair wise using logical AND operation to give only the “co-localized” spots. These spots were then counting using the “Count Nuclei” module. The minimum spot size was set so as to remove any small spots due to partial, and likely random, overlap of spots. Results were logged into Excel for analysis. Values for all analyses including co-localization and vesicle counting represent mean value ± standard error of the mean (SEM).

Recycling Assays

For biotinylation assay, cells were serum starved and pretreated with low levels of 10nM lactacystin and 100nM concanamycin inhibitors for 1h before chilling on ice and biotinylated for recycling assay as described previously (Hammond et al., 2003). After biotinylation, cells were stimulated with 0.5nM HGF at 37°C in the presence of inhibitors for 7 min to allow internalization. Cells were placed on ice, stripped with reducing reagent (100 mM sodium 2-mercaptoethanesulfonic acid (MESNA) in 50 mM Tris-HCl pH 8.6, 100 mM NaCl, 1 mM EDTA, and 0.2% BSA), to remove non-internalized biotinylated proteins, followed by returning cells to 37°C. To determine percentage of internalized proteins that recycled, cells were returned to ice, subjected to a second reduction with MesNa prior to lysis and recovery with NeutrAvidin-agarose beads and immunoblot for detection of Met levels. Percent recycled Met was determined by quantifying IB and subtracting the amount recovered from a similarly processed sample that did not undergo a second round of reduction (representing total pool of internalized Met) divided by the total pool of internalized Met.

IF assay was performed as described previously (Driskell et al., 2007). Cells grown on glass coverslips were pulsed with prewarmed (37°C) 0.5nM HGF for 5 min, washed 5x with Leibovitz-15 Medium containing 0.2% BSA at 4°C, and chased at 37°C for 20 min, fixed and processed for IF. Cells were scored on ratio of endosomal (EN) over plasma membrane staining of Met and reported as mean + SEM, N=4. In each experiment, a minimum of 20 fields were scored.

siRNA transfection

HeLa cells were seeded at 2.0×10⁵ in 6-well dishes and transfected with 50nM siRNA using Hiperfect as per manufacture’s instructions (Qiagen). All experiments performed 72 h post-transfection. siRNA sequences are described in Supplemental Experimental Procedures.

Arf-GTP Assay

1.0×10⁶ HeLa cells were serum starved overnight then stimulated with 0.5nM HGF for indicated times and subjected to pulldown assays using GST-GGA3 (1-316) domain as described previously (Anton et al., 2006).

Migration Assay

Equal number of HeLa cells (5 × 10⁴) were seeded directly onto 6.5-mm Corning Costar transwell chambers for migration assays as described previously (Paliouras et al., 2009). All bar graphs represents mean ± SEM.

Live-cell imaging

Imaging was performed as described previously (Zhao and Keen, 2008) using a Zeiss Axiovert 200 microscope and Olympus 150X/1.45 NA objective. Additional details can be found in Supplemental Experimental Procedures.