Introduction
Tumours become malignant when cancer cells migrate into neighbouring tissues and survive at these ectopic sites1. Invasion and metastasis are the hallmarks of cancer malignancy, although invasion is not unique to cancer: cells invade tissues during embryonic development, in immune surveillance and in non-cancerous diseases. Identification of the molecules involved specifically in cancer invasion could allow development of therapies to inhibit the spread of cancer.
Invasion requires cell migration, which involves changes to cell–cell adhesion, cell–matrix adhesion, membrane dynamics and proteolytic degradation of the extracellular matrix. The plethora of molecules reported to be involved in cancer migration and invasion includes epidermal growth factor receptor (EGFR), a tyrosine kinase receptor that is mutated and/or upregulated in many epithelial tumours. Several pathways downstream of EGFR are implicated in tumour cell migration and invasion, including phospholipase C
and phosphatidylinositol 3-kinase (PI(3) kinase)2. A paper by Sabe and colleagues on page 85 of this issue3 describes a new pathway, identified through a series of elegant experiments, demonstrating that when stimulated, EGFR interacts specifically with GEP100/BRAG2 — a guanine nucleotide-exchange factor (GEF) for the small GTP-binding protein Arf6 — to induce tumour invasion. This signalling pathway is likely to be relevant in vivo, at least in some breast cancers, as expression of EGFR and GEP100 is higher in more malignant breast tumours, whereas decreasing GEP100 expression significantly reduces the metastasis of mammary tumour cells in mice.
Arfs belong to the Ras superfamily of GTP-binding proteins and are widely expressed and conserved in all eukaryotes. Their primary role is in membrane trafficking and they are anchored to membranes by amino-terminal myristoylation. As with most members of the Ras superfamily, Arfs cycle between a GTP-bound (active) form and a GDP-bound (inactive) form. This cycle is tightly regulated by two groups of proteins: GEFs, which facilitate the exchange of GDP for GTP, and GTPase-activating proteins (GAPs), which promote the hydrolysis of bound GTP4. There are six mammalian Arfs, each of which regulates specific steps of membrane trafficking. Arf6 is mainly involved in endosomal membrane traffic and structural organization at the plasma membrane5. Previous studies have also shown that Arf6 is expressed at higher levels in more invasive breast cancer cells and also mediates invasion6, 7.
For most proteins of the Ras superfamily, there are more GEFs and GAPs than GTP-binding proteins and a major challenge in the field is to define how each GEF and GAP contributes to the various responses regulated by each GTP-binding protein. Sabe and colleagues used RNA intereference (RNAi) to knock down each of the ten Arf-GEFs expressed in an invasive breast cancer cell line, MDA-MB-231. Surprisingly, they found that only one, GEP100, is required for EGF-stimulated invasion. Next, they demonstrated that GEP100 is required for EGF-stimulated Arf6 activation and identified an unusual mechanism of GEP100 recruitment to EGFR, where the pleckstrin-homology (PH) domain of GEP100 binds directly to two phosphorylated tyrosines (Tyr 1068/1086) after EGFR stimulation. Phosphotyrosines are generally considered to recruit proteins containing either SH2 or phosphotyrosine-binding (PTB) domains, and indeed Tyr 1068/1086 also recruit the adaptor protein Grb2 to EGFR through its SH2 domain. Grb2 in turn binds through Gab1 to PI(3) kinase, which is also implicated in tumour cell migration8. The GEP100 PH domain seems to be specific for Tyr 1068/1086 and related Arf-GEF PH domains do not bind to these residues. PH domains are often studied because of their ability to bind to phosphoinositides, and indeed some Arf-GEFs are regulated by binding of their PH domains to 3´-phosphorylated phosphoinositides. However, several PH domains are known to bind to proteins and the PH domain structure is related to that of the PTB domain9. The PH domain of pleckstrin itself seems to bind directly to phosphotyrosine residues. It will be interesting to know the structural basis for the binding of GEP100 PH domains to phosphotyrosines, and whether it can also bind to phosphoinositides.
As Arf6 needs to cycle between GTP- and GDP-bound forms for its normal function, EGFR would be expected to recruit a GAP as well as GEP100. A likely candidate is centaurin-
2, which is recruited to the plasma membrane in a PI(3) kinase-dependent manner after EGF stimulation and specifically functions as a GAP for Arf6 (ref. 10).
So how could recruitment of Arf6 to the EGFR contribute to EGF-induced cell invasion? The specific contribution of Arf6 is likely to derive from its role in endosome recycling. In migrating cells, this recycling could be essential for the delivery of membrane to the leading edge to allow forward protrusion11. However, the spatio-temporal distribution of active Arf6 will be critical in determining what it recycles. For example, in epithelial cells Arf6 localizes in cell–cell junctions and activated Arf6 induces endocytosis of E-cadherin, resulting in loss of adherens junctions. EGFR associates with E-cadherin at cell–cell junctions5 and GEP100-induced Arf6 recruitment could thus induce endocytosis of E-cadherin (Fig. 1). Indeed, Sabe and colleagues show that E-cadherin is lost from cell–cell adhesions in EGF-stimulated MCF7 cells overexpressing GEP100 and Arf6 (ref. 3). GEP100 also interacts directly with -catenin, which is part of the E-cadherin complex12, and thus it has the potential to function as a bridge between -catenin and EGFR. As downregulation of E-cadherin contributes to invasion in several cancers1, analysing the relationship between GEP100, EGFR, and E-cadherin endocytosis will be an important area of research.
Figure 1: Schematic model of the potential mechanisms for EGFR/GEP100/Arf6-induced invasion of epithelial cancers.
Step 1: in the absence of EGF, cell–cell contacts are maintained and E-cadherin is in adherens junctions. Low levels of Arf6–GTP are attributable to Arf–GEFs other than GEP100. Step 2: On EGRF stimulation, the EGFR intracellular domain becomes tyrosine-phosphorylated and recruits GEP100 through its PH domain. This interaction stimulates the recruitment and activation of Arf6 (a). Arf6–GTP could stimulate endocytosis of EGF/EGFR/GEP100, as well as E-cadherin, resulting in the loss of adherens junctions and consequently loss of cell–cell interactions (b). Step 3: after cells lose epithelial cell–cell contacts, E-cadherin is targeted for degradation. In response to EGF, cells become polarized and migrate. EGF stimulation induces EGFR endocytosis, and thus GEP100 could signal to Arf6 on endosomes as well as at the plasma membrane (a). Arf6 activation contributes to the formation of invadopodia, possibly by stimulating integrin recycling and/or delivery of proteases to degrade the extracellular matrix (b). EGFR/GEP100/Arf6 contributes to extension of the leading edge during invasion, probably by stimulating delivery of membrane and integrins through endosomal recycling, as well as by inducing reorganization of the actin cytoskeleton together with the Rho GTPases Rac and Rho (c).
Full size image (137 KB)Cell adhesion to the extracellular matrix (ECM) is also controlled by Arf6 through the recycling of 
1-integrins, and interestingly, this process is dependent on GEP100 (ref. 13). As this process involves the delivery of membrane vesicles to the plasma membrane, Rab GTPases could function in coordination with Arf6 in this pathway. In particular, Rab25 contributes to invasion through a direct interaction with the cytoplasmic tail of 1-integrin, thereby promoting the delivery of vesicles to the plasma membrane at pseudopodial tips, as well as the retention of a pool of cycling 51 at the cell front during migration in a 3D matrix14. As with Arf6 and GEP100, Rab25 expression is higher in aggressive breast cancers. It would therefore be interesting to see whether vesicles positive for Rab25 also contain Arf6 when EGF is stimulated. ARAPs are PI(3) kinase-regulated GAPs for both Arf6 and the small GTP-binding protein RhoA that affect cell spreading15 and might act together with GEP100 to coordinate integrin recycling and signalling.
Arf6 may also promote invasion by regulating EGFR endocytosis and recycling (Fig. 1), thereby determining the proportion of EGFR that is active on the plasma membrane when compared with endosomes, and thus which pathways it activates16. EGF often stimulates cell growth and proliferation as well as invasion; in this study, however, GEP100 had only a small effect on the growth of tumours at the primary site of injection in mice3. Future research should therefore address how GEP100 and Arf6 affect EGFR endocytosis, localization and signalling.
As well as inducing endocytosis, Arf6 can affect the actin cytoskeleton either directly or through regulation of Rho family GTP-binding proteins, which are well known coordinators of cell migration11. For example, Arf6 functions together with Rac1 to activate phospholipase D and PtdIns(5) kinase, leading to the accumulation of phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2), a regulator of several actin-binding proteins. Arf6-induced phospholipase D activation functions in coordination with Rac1 to promote cell migration. In epithelial cells, the Arf6–GEF ARNO promotes Rac1 activation by recruiting the Rac1–GEF complex DOCK180–ELMO5.
Arf6 contributes not only to migration, but also specifically to invasion by regulating the formation of invadopodia, which are actin-rich protrusions at the plasma membrane where matrix degradation occurs in 3D environments6, 7. Invasion generally requires the production and activation of matrix-degrading proteases, including matrix metalloproteinases (MMPs), and these are concentrated in invadopodia17. Although Arf6 and GEP100 do not seem to alter the activity of MMP2 and MMP9 in breast cancer cells3, 6, they could still regulate other proteases. Whether GEP100 mediates EGFR-induced assembly and/or turnover of invadopodia is not yet known.
The specificity of GEP100 for EGF-induced Arf6 activation and invasion suggests that this pathway could be targeted in cancer therapy in patients with EGFR mutations and/or overexpression of EGFR. As the binding of the GEP100 PH domain to two specific phosphotyrosine residues in EGFR is necessary for the activation of Arf6, the authors suggest that therapeutics aimed at inhibiting this interaction might be developed. This could be a useful method to treat patients whose cancers acquire resistance to EGFR inhibitors2, although whether GEP100 affects cancer growth as well as invasion needs to be established. Apart from the therapeutic perspective, the work of Sabe and colleagues emphasizes that individual GEFs function locally to transduce a specific extracellular stimulus to activate a GTP-binding protein. It is likely that the activation of Arf6 is coordinately linked to other invasion-promoting proteins recruited at the same time to the receptor signalling complex. It will be interesting to determine how general the involvement of the EGFR/GEP100/Arf6 pathway is in cancer invasion and also whether other GEFs for Arf6 have similar roles in response to other stimuli.