Asef, a Link Between the Tumor Suppressor APC and G-Protein Signaling

See allHide authors and affiliations

Science  18 Aug 2000:
Vol. 289, Issue 5482, pp. 1194-1197
DOI: 10.1126/science.289.5482.1194


The adenomatous polyposis coli gene (APC) is mutated in familial adenomatous polyposis and in sporadic colorectal tumors. Here the APC gene product is shown to bind through its armadillo repeat domain to a Rac-specific guanine nucleotide exchange factor (GEF), termed Asef. Endogenous APC colocalized with Asef in mouse colon epithelial cells and neuronal cells. Furthermore, APC enhanced the GEF activity of Asef and stimulated Asef-mediated cell flattening, membrane ruffling, and lamellipodia formation in MDCK cells. These results suggest that the APC-Asef complex may regulate the actin cytoskeletal network, cell morphology and migration, and neuronal function.

Mutations of the tumor suppressor gene APC are responsible for familial adenomatous polyposis, a dominantly inherited disease characterized by multiple adenomatous polyps in the colon (1). The APC gene is also somatically mutated in most sporadic colorectal tumors. Consistent with its role as a tumor suppressor, overexpression of APC blocks cell cycle progression from the G1 to S phase (2). The product of the APC gene interacts with various proteins, including β-catenin, a key component of the Wnt/Wingless signaling transduction pathway that plays important roles in a number of developmental processes and in tumorigenesis (3). APC is thought to be involved in the degradation of β-catenin through its interaction with β-catenin, GSK-3β, and Axin or the closely related factor conductin/Axil (4). APC also interacts with EB1 and the human homolog of the Drosophila Discs large (hDLG) through its COOH-terminal region (5, 6). Furthermore, APC possesses an armadillo repeat domain, which is thought to be involved in protein-protein interactions. To obtain new insights into the function of APC, we attempted to identify proteins that interact with the armadillo repeat domain of APC.

We screened a human fetal brain library using the armadillo repeat domain of APC as target, and isolated a gene that we have namedAsef (for APC-stimulated guanine nucleotide exchange factor) (7). The human full-length Asef cDNA encodes a protein of 619 amino acids with motifs found in the Dbl family of proteins (Fig. 1, A and B) (8). Asef contains the Dbl homology (DH), Pleckstrin (PH), and Src homology 3 (SH3) domains. Northern blot analysis of Asef revealed a mRNA of 3.6 kb that is highly expressed in mouse brain (9).

Figure 1

(A) Predicted amino acid sequence of Asef and (B) alignment of homologous domains in Dbl family members. The Asef sequence has been deposited in GenBank (accession number AB042199). In (A), the SH3 domain is shown in bold and is boxed, the DH domain is highlighted in black, and the PH domain is shown in bold and underlined. ABR, APC-binding region; PR, proline-rich region; CH, calponin homology domain. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

To confirm that APC and Asef interact directly, we created a fusion protein (APC-arm) consisting of the armadillo repeat domain of APC and glutathione-S-transferase (GST). We examined the ability of this fusion to interact with the fragment of Asef (Asef-M in Fig. 2A) (10). The in vitro–translated APC-arm interacted with GST–Asef-M, but not with GST alone (Fig. 2B). Likewise, in vitrotranslated Asef-M interacted with GST–APC-arm, but not with the armadillo repeat domain of β-catenin fused to GST (GST–β-catenin–arm) or GST alone.

Figure 2

Association of Asef with APC in vitro and in vivo. (A) Mapping of the regions in Asef required for binding to APC. Deletion constructs of Asef were analyzed for their ability to interact with APC in the two-hybrid system. (+) Detectable activity; (−) no detectable activity; Asef-M, the fragment of Asef isolated in the initial two-hybrid screen. (B) Association of Asef with APC in vitro. In vitro–translated (IVT)35S-labeled APC-arm was incubated with GST–Asef-M or GST-Sepharose. In vitro–translated 35S-labeled Asef-M was incubated with GST–APC-arm, GST–β-catenin–arm, or GST-Sepharose. Bound proteins were analyzed by 10% SDS-PAGE and fluorography. (C) Association of Asef with APC and β-catenin in vivo. Lysates prepared from embryonic rat brain were subjected to immunoprecipitation with the indicated antibodies, then were fractionated by 6% SDS-PAGE and immunoblotted with the indicated antibodies. Pep (+) indicates that antibodies were preincubated with antigen before use in immunoprecipitation. GST–β-catenin was used to block anti–β-catenin.

To identify the region of Asef responsible for its interaction with APC, we performed two-hybrid assays using deletion fragments of Asef. Mutants lacking amino acids 73 to 126 were negative for interaction with APC, whereas a fragment containing amino acids 73 to 126 was positive (Fig. 2A). This indicates that the APC-binding region may reside in the NH2-terminal region upstream of the SH3 domain.

We next examined whether endogenous Asef associates with APC in vivo. A lysate from embryonic rat brain was subjected to immunoprecipitation with antibodies to Asef (anti-Asef) followed by immunoblotting with anti-APC (11). Asef was identified as an 85-kD protein and coimmunoprecipitated with APC (Fig. 2C). Similarly, immunoprecipitation of the lysate with anti-APC followed by immunoblotting with anti-Asef revealed an association between APC and Asef. Coprecipitation of Asef and APC was inhibited by preincubation of the antibodies with the antigens used for immunization. Immunoblot analysis of the Asef immunoprecipitates with anti–β-catenin revealed that β-catenin also coimmunoprecipitates with Asef. In addition, Asef was detected in β-catenin immunoprecipitates. Because Asef does not interact directly with β-catenin (Fig. 2B), these findings suggest that Asef, APC, and β-catenin are contained in the same complex in vivo.

Consistent with the finding that APC interacts with Asef, immunohistochemical analysis with anti-APC and anti-Asef showed that both APC and Asef are expressed in mouse colon epithelial cells (Web figure 1) (12, 13). Asef is highly expressed in the central nervous system, including the hippocampus, olfactory bulb, and cerebellum, where APC is also highly expressed (14, 15). Both APC and Asef localized in the cytoplasm of the hippocampal pyramidal neurons (Web figure 2) (12) and in the olfactory glomeruli (Web figure 1) (12). Double-labeling immunoelectron microscopy revealed colocalization of APC and Asef in the synapse of the olfactory glomerulus (Web figure 1) (12,16).

The Dbl family of proteins are guanine nucleotide exchange factors (GEFs) for specific members of the Rho family of small GTP-binding proteins (G proteins) (8). We found that Rho and Rac, but not Cdc42, bound to His6-tagged-Asef (His6-Asef) immobilized to ProBond Resin (Fig. 3A) (17). In contrast, Dbl associated with all three of these proteins, as reported previously (18). This suggests that Asef may function as a GEF for Rac and Rho, but not Cdc42. However, when full-length Asef was incubated with Rac bound to [3H]GDP, it stimulated dissociation of GDP from Rac only very weakly (Fig. 3B) (19). Because many proteins of the Dbl family exhibit oncogenic activity when their NH2-terminal regions are truncated (8), we reexamined GEF activity using a mutant form of Asef lacking the NH2-terminal region (AsefΔAPC). This region contains the APC-binding domain of Asef. As expected, AsefΔAPC strongly stimulated dissociation of GDP from Rac in a time- and dose-dependent manner (Fig. 3, B and C). In contrast, mutant AsefΔAPC that also lacked the DH domain did not show this activity (9). AsefΔAPC stimulated binding of [35S]GTPγS to Rac (9), but did not affect the GEF activity of Rho and Cdc42 (Fig. 3C). These results suggest that Asef has the potential to function as a GEF specific for Rac, and that this activity is negatively regulated by its NH2-terminal region, which contains the APC-binding domain.

Figure 3

Stimulation of Asef guanine nucleotide exchange activity by APC. (A) Binding of Asef to Rho family guanosine triphosphatases. His6-AsefΔAPC bound to ProBand Resin was incubated with RhoA, Rac1, or GST-Cdc42. The bound proteins were subjected to immunoblot analysis. As a negative control, ProBand Resin that had not been incubated with Asef was used. (B) Dissociation of [3H]GDP from Rac1 (20 nM) after 15 min of incubation at 30°C in the absence or presence of APC-arm (100 nM) and/or the indicated concentrations of Asef or AsefΔAPC (19). The amount of [3H]GDP remaining bound to Rac1 was determined, and results are expressed as percentages of the values obtained in the absence of Asef and APC. (C) Time course of [3H]GDP dissociation from Rac1 (triangles), RhoA (squares), and Cdc42 (circles) (20 nM) was measured at 30°C in the absence (closed symbols) or presence (open symbols) of AsefΔAPC (20 nM). At the indicated times, portions of the incubation mixture were removed, and the amount of [3H]GDP remaining bound to Rac1, RhoA, and Cdc42 was determined. Results are expressed as percentages of the values obtained at 0 min. (D) Dissociation of [3H]GDP from Rac1 (20 nM) after 15 min of incubation at 30°C in the absence or presence of GST (100 nM), Asef (20 nM), AsefΔAPC (20 nM), and/or the various concentrations of APC-arm (10, 20, 40, 100, 200 nM). (*) APC-arm (100 nM) was used.

We next investigated whether APC affects the Rac-specific GEF activity of Asef. When added to GEF assay reaction mixtures, GST-APC-arm but not GST alone, stimulated the activity of the full-length Asef in a dose-dependent manner and to a level comparable to that achieved with AsefΔAPC (Fig. 3, B and D). Thus, APC may activate the GEF activity of Asef by binding to its NH2-terminal region and relieving negative regulation.

Rac is involved in the reorganization of the actin cytoskeletal network, producing lamellipodia and membrane ruffling (20). We therefore examined the effect of Asef on the morphology of MDCK cells (21). When the cells were transfected with the full-length Asef, they became flattened onto the substratum and exhibited membrane ruffles and lamellipodia (Fig. 4A). However, when the APC-binding region of Asef was exogenously expressed along with the full-length Asef, these morphological changes were not observed (Fig. 4, G to J). Thus, we believe these morphological changes are induced by Asef that is activated by interaction with endogenous APC. Furthermore, when the cells were transfected with the full-length Asef along withAPC-arm, they became enlarged and exhibited even more abundant membrane ruffles and lamellipodia (Fig. 4C). Cells transfected with AsefΔAPC were also substantially larger than the parental cells and exhibited a morphology very similar to that of theAsef- and APC-arm–transfected cells (Fig. 4E). In contrast, the AsefΔAPC mutant also lacking the DH domain did not show such activity (9). Both Asef and AsefΔAPC localized in the cytoplasm and concentrated in membrane ruffles and at the edges of cells not in contact with other cells (Fig. 4, A, C, and E). Staining with anti-APC revealed that both endogenous APC and exogenously expressed APC-arm colocalized with Asef (Fig. 4, A to D). In cells not transfected with Asef, APC concentrated near the margin of protruding membrane structures as reported previously (22). However, these clusters of APC were not detected inAsef-transfected cells. These results suggest that Asef acts as a GEF for Rac in living cells and that its activity is regulated by APC.

Figure 4

Morphology of MDCK cells transfected withAsef. MDCK cells were transfected with expression plasmids encoding HA-tagged Asef (A and B), HA-tagged Asef and Myc-tagged APC-arm (C and D), HA-tagged AsefΔAPC (E), Myc-tagged APC-arm (F), HA-tagged Asef and GFP (G and H), or HA-tagged Asef and the APC-binding region of Asef (ABR in Fig. 2A) fused to GFP (GFP-ABR) (I and J). MDCK cells were double-stained with antibodies against HA (A, C, E, H, and J), APC (B), and Myc (D and F). GFP (G) and GFP-ABR (I) were visualized as green fluorescence. Arrowheads in (B) point to clusters of APC in extending membranes, and arrows in (A) to (D) indicate areas of colocalization of Asef and APC. Bar, 50 μm.

Many members of the Dbl family are oncogenic, especially when truncated (8). However, neither the full-length nor NH2-terminal–truncated derivatives of Asef show any oncogenic activity (9).

The armadillo repeat domain is the most highly conserved feature in APC, suggesting that it is essential for function (1,23). We have demonstrated here that the armadillo repeat domain of APC interacts with a Rac-specific GEF, Asef, and activates its activity. These findings raise the possibility that APC plays a role in regulating the actin cytoskeletal network, thereby affecting cell morphology, polarity, and migration. Indeed, it has been suggested that APC may be involved in epithelial cell migration (22,24). Furthermore, the colocalization of Asef and APC at the synapse of neuronal cells suggests that this complex is involved in signal transduction at the synapse in addition to neuronal cell migration. However, the identity of the signal that regulates the function of the APC-Asef complex in colon epithelial cells and neuronal cells remains to be elucidated.

Many of the mutant APCs found in colon cancers lack binding sites for microtubules, hDLG, and some of the β-catenin–binding sites, although the armadillo repeat domain itself is not always deleted (1). It is therefore interesting to speculate whether the interactions of APC with these proteins are important for localizing Asef to sites where it can properly exert its effect on the actin cytoskeletal network. Consistent with this, we found that the Asef-APC complex is associated with β-catenin. Thus, when Asef binds to mutant APCs, it may fail to be properly localized, thereby compromising its ability to stimulate migration of colon epithelial cells. This abrogation of Asef function could promote tumorigenesis by increasing the accumulation of proliferating cells in the intestinal crypt.

  • * To whom correspondence should be addressed. E-mail: akiyama{at}


View Abstract

Navigate This Article