Report

Dishevelled 2 Recruits ß-Arrestin 2 to Mediate Wnt5A-Stimulated Endocytosis of Frizzled 4

See allHide authors and affiliations

Science  05 Sep 2003:
Vol. 301, Issue 5638, pp. 1391-1394
DOI: 10.1126/science.1082808

Abstract

Wnt proteins, regulators of development in many organisms, bind to seven transmembrane–spanning (7TMS) receptors called frizzleds, thereby recruiting the cytoplasmic molecule dishevelled (Dvl) to the plasma membrane.Frizzled-mediated endocytosis of Wg (a Drosophila Wnt protein) and lysosomal degradation may regulate the formation of morphogen gradients. Endocytosis of Frizzled 4 (Fz4) in human embryonic kidney 293 cells was dependent on added Wnt5A protein and was accomplished by the multifunctional adaptor protein β-arrestin 2 (βarr2), which was recruited to Fz4 by binding to phosphorylated Dvl2. These findings provide a previously unrecognized mechanism for receptor recruitment of β-arrestin and demonstrate that Dvl plays an important role in the endocytosis of frizzled, as well as in promoting signaling.

The secreted glycoprotein signaling molecules of the Wnt family (Wingless or Wg in Drosophila) are conserved in evolution and play important developmental roles (1). Wnt activity is mediated by interaction with the frizzled 7TMS receptors, which signal by stabilizing β-catenin, thereby enhancing the activity of LEF/Tcf transcription factors. The most proximal signaling intermediate in this pathway is a cytoplasmic molecule, Dvl, which is recruited to the plasma membrane by Fz (2, 3). In Drosophila, Wg is endocytosed in a frizzled-dependent manner, and spatially restricted differences in the rates of Wg endocytosis and lysosomal degradation correlate with the formation of Wg morphogen gradients (4).

β-arrestins are ubiquitous multifunctional adaptor proteins that universally regulate numerous aspects of 7TMS receptor function (5). When receptors are activated, they are rapidly phosphorylated by G protein–coupled receptor kinases (GRKs) (6), and then bind β-arrestin 1 or 2. β-arrestins desensitize second-messenger generation by sterically blocking receptor–G protein interaction (5); mediate endocytosis of the receptors in clathrin-coated pits by binding clathrin, AP-2, and other elements of the endocytic machinery (7, 8); and serve as scaffolds that link the receptors to other signaling pathways (911). Here we have investigated the mechanisms responsible for Fz internalization and the possible involvement of β-arrestins in this process.

To visualize Fz4 receptors in live cells, we constructed a Fz4–green fluorescent protein (GFP) molecule and expressed it in human embryonic kidney 293 (HEK293) cells (12). It was present predominantly at the cell surface (Fig. 1A) with a few green puncta also observed in the cytosol, possibly due to the presence of partially processed Fz4-GFP. When cells were stimulated for 30 min with Wnt5A conditioned medium (Wnt5A), which activates signaling through Fz4 (13), no internalization of Fz4-GFP was observed (Fig. 1B). Similarly, activation of protein kinase C (PKC) with 1 μM phorbol myristoyl acetate (PMA) for 30 min resulted in no apparent receptor internalization (Fig. 1C). However, when cells were stimulated concurrently with both Wnt5A and an activator of PKC such as PMA (Fig. 1D) or substance P, which activates receptors coupled to the G protein Gq (Fig. 1E), Fz4-GFP was internalized and the receptor was present in intracellular vesicles.

Fig. 1.

βarr2 mediates Wnt5A- and PKC-induced Fz4-GFP internalization. (A to E) Confocal images of Fz4-GFP expressed alone (A to D) or with substance P receptor (SPR) (E) in HEK293 cells. Cells were left untreated (A) or treated with Wnt5A (B), 1 μM PMA (C), Wnt5A and 1 μM PMA (D), and Wnt5A and 100 nM substance P (SP) (E) at 37°C for 30 min. (F to I) HEK293 cells were transfected with control siRNA (F and G) or siRNA against βarr2 (H and I) and then with Fz4-GFP. Cells were treated with Wnt5A and 1 μM PMA for 0 min (F and H) and 30 min (G and I) at 37°C. Scale bar, 10 μm.

The prototypic 7TMS receptor, the β2-adrenergic receptor (β2-AR), is internalized by a classic clathrin-coated pit mechanism that uses β-arrestin as an adaptor to link to various elements of the endocytic machinery (14). To compare the mechanisms of Fz4 and β2-AR internalization, we expressed Fz4-GFP and β2-AR–red fluorescent protein (15) in the same HEK293 cells and stimulated them for 30 min with, simultaneously, Wnt5A (in the presence of PMA) and isoproterenol. Both receptors were internalized and appeared in the same population of intracellular vesicles (fig. S1). Moreover, treatment of Fz4-GFP–expressing cells with the clathrin inhibitors sucrose (0.45 M) or monodan-sylcadaverine (200 μM) inhibited Wnt5A- and PMA-induced internalization of Fz4 (16). Taken together, these data suggest that Fz4 internalization is mediated by a clathrin-mediated process similar to that used for internalization of the β2-AR.

Because the β2-AR and numerous other 7TMS receptors use β-arrrestins as essential adaptors in clathrin-mediated endocytosis (14), we tested whether β-arrestins were involved in Fz4 internalization. We used small interfering RNA (siRNA) to reduce the expression of βarr2 in HEK293 cells (fig. S2) (17). Internalization of Fz4 induced by Wnt5A and PMA was completely ablated in cells expressing βarr2 siRNA (Fig. 1, F to I). In contrast, knock-down of βarr1 had no effect on Fz4 internalization (16).

We next studied the effects of Fz4 stimulation on the spatial distribution of βarr2-GFP. When βarr2-GFP was expressed in HEK293 cells, it was diffusely distributed in the cytoplasm (Fig. 2A). This distribution was unaffected when Fz4 was expressed along with βarr2-GFP, even when the cells were stimulated with Wnt5A for up to 12 hours (Fig. 2B). Because βarr1 interacts with Dvl (18), we coexpressed βarr2-GFP together with Dvl2, but this also did not change the distribution of βarr2-GFP (Fig. 2C). However, coexpression of Fz4, Dvl2, and βarr2-GFP led to redistribution of βarr2 to the plasma membrane, where it was found in a punctate pattern (Fig. 2, D and E). This pattern is similar to that previously observed with the stimulated β2-AR (19). Our inability to detect βarr2 recruitment to Fz4 in the absence of Dvl transfection presumably is due to inadequate amounts of endogenous Dvl to interact with the larger amounts of transfected βarr2-GFP. When βarr1-GFP rather than βarr2-GFP was used, no recruitment to Fz4 was observed, even in the presence of Dvl2 (16). Thus, βarr2-GFP translocation to the plasma membrane required both Dvl2 and Fz4. When 50 or 200 ng of Fz4 plasmid was coexpressed with 200 ng of Dvl2 and βarr2-GFP plasmid, ≤20% or ≥70% of βarr2-GFP–expressing cells demonstrated βarr2-GFP translocation, respectively (n = 3). On the basis of these findings, Fz4 might be expected to directly recruit Dvl2 to the plasma membrane independent of βarr2. Indeed, when Dvl2-GFP was coexpressed with Fz4, it was redistributed smoothly and evenly to the plasma membrane (fig. S3B), whereas it was diffusely distributed throughout the cytoplasm when expressed alone (fig. S3A) (2, 13).

Fig. 2.

Recruitment of βarr2 to the plasma membrane requires both Fz4 and Dvl2. (A to E) Confocal images of βarr2-GFP expressed alone (A) or with Fz4 (B), Myc-Dvl2 (C), or Fz4 and Myc-Dvl2[(D), middle section of the cell; (E), bottom section of the cell] in HEK293 cells. Cells were left untreated [(A), (C), (D), and (E)] or treated with Wnt5A at 37°C for up to 12 hours (B). (F) Interaction of Fz4 with Dvl2. Cell extracts from HEK293 cells overexpressing Myc-Dvl2 with vector or FLAG-Fz4 were immunoprecipitated with antibody to FLAG (anti-FLAG). Immunoprecipitates were immunoblotted with anti-Myc to detect associated Dvl2. (G) Interaction of Dvl2with βarr2. Proteins were immunoprecipitated from cell extracts of HEK293 cells overexpressing Myc-Dvl2with vector or FLAG-βarr2 with anti-FLAG. Immunoprecipitates were immunoblotted with anti-Myc to detect associated Dvl2. (H) Dvl2 mediates βarr2 binding to Fz4. His-βarr2 was overexpressed with vector, FLAG-Fz4, or FLAG-Fz4 and Myc-Dvl2 in HEK293 cells. Cells were left untreated or treated with Wnt5A at 37°C for 3 hours, and cell extracts were immunoprecipitated with anti-FLAG. Immunoprecipitates were immunoblotted with anti-His. Whole-cell lysates were immunoblotted with the respective antibodies. Scale bar, 10 μm.

Interactions among Fz4, Dvl2, and βarr2 were also examined by coimmunoprecipitation experiments. When expressed in HEK293 cells, Fz4 tagged with a FLAG epitope coimmunoprecipitated with Dvl2 modified to contain a Myc epitope (Fig. 2F). FLAG-βarr2 also coimmunoprecipitated with Myc-Dvl2 (Fig. 2G). Interaction of His-βarr2 with FLAG-Fz4 was enhanced when Myc-Dvl2 was also expressed (Fig. 2H). This interaction was not further enhanced by stimulation of the cells with Wnt5A. These data suggest that Fz4 can directly recruit Dvl2, but perhaps not βarr2, to the plasma membrane and that it is Dvl2 that is responsible for recruiting βarr2 to the plasma membrane to bind Fz4. This represents a new mode of regulation by βarr2, which in other cases is recruited to 7TMS receptors directly after receptor phosphorylation, without the involvement of other adaptor proteins (20).

In cells coexpressing FLAG-Fz4, Myc-Dvl2, and βarr2-GFP, endogenous clathrin and βarr2-GFP were distributed in an overlapping punctate pattern at the plasma membrane (Fig. 3, A to C). This suggests that βarr2, recruited to Fz4 by Dvl2, is capable of targeting Fz4 to clathrin-coated pits for internalization. When cells expressing Fz4, Dvl2, and βarr2 were stimulated with Wnt5A for 30 min, no further membrane recruitment of βarr2-GFP was observed (Fig. 3, D and E); however, Fz4-GFP was internalized, appearing in intracellular vesicles (Fig. 3, F and G). Thus, as it does with the β2-AR (19, 21), βarr2 dissociates from Fz4 and does not internalize with it. At least when Fz4, Dvl2, and βarr2 are all coexpressed, the constitutive activity of Fz4 appears to be sufficient to promote Dvl2-mediated βarr2 recruitment, but not Fz4 internalization, which requires the additional signaling of Wnt5A stimulation.

Fig. 3.

βarr2 targets Fz4 to clathrin-coated pits by way of Dvl2 for Wnt5A-induced receptor internalization. (A to C) Confocal images of endogenous clathrin (A) and βarr2-GFP (B) in the same HEK293 cell overexpressing FLAG-Fz4, Myc-Dvl2, and βarr2-GFP. βarr2-GFP colocalizes with endogenous clathrin on the plasma membrane as shown with arrows in the merged image (C). (D and E) Confocal images of βarr2-GFP in HEK293 cells overexpressing Fz4, Myc-Dvl2, and βarr2-GFP. (F and G) Confocal images of Fz4-GFP in HEK293 cells overexpressing Fz4-GFP, Myc-Dvl2, and FLAG-βarr2. Cells were treated with Wnt5A for 0 min (D and F) and 30 min (E and G) at 37°C. Scale bar, 10 μm.

Our findings indicate that Fz4 internalization requires, in addition to Dvl2 and βarr2, the activity of PKC and also the presence of Wnt5A agonist bound to frizzled. Thus, PKC activity might be required to assemble the complex of Fz4, Dvl2, and βarr2. Accordingly, we tested the effect of a PKC inhibitor on these phenomena. When Fz4 and Dvl2-GFP were coexpressed, Dvl2-GFP was found at the plasma membrane with Fz4. This association was not affected by a 1-hour treatment with the PKC inhibitor GFX109203 (4 μM) (Fig. 4, A and B). However, when Fz4, Dvl2, and βarr2-GFP were coexpressed in cells, βarr2-GFP was found at the plasma membrane (Fig. 4C), and this association was disrupted by the PKC inhibitor (Fig. 4D). Thus, PKC activity is necessary for the Dvl2-mediated recruitment of βarr2 to Fz4. The data in Fig. 4, E to G, further support the idea that PKC phosphorylates Dvl2 to increase its interactions with βarr2. Endogenous PKC isoforms α, β, and γ coimmunoprecipitated with Myc-Dvl2 (Fig. 4E), and addition of [γ-32P]ATP to such immunoprecipitates led to phosphorylation of Dvl2 by endogenous kinase(s) associated with Dvl2, presumably including PKC (Fig. 4F) (2224). Moreover, as shown by mobility shifts on SDS–polyacrylamide gel electrophoresis, endogenous Dvl2 appears to be phosphorylated after PMA stimulation of cells (Fig. 4G). Finally, purified PKCα phosphorylated immunoprecipitated Dvl2, and this enhanced its binding to βarr2 (Fig. 4F).

Fig. 4.

Phosphorylation of Dvl2by PKC enhances the complex assembly of Fz4, Dvl2, and βarr2. (A and B) Confocal images of Dvl2-GFP in HEK293 cells expressing Fz4 and Dvl2-GFP. (C and D) Treatment-dependent distribution of βarr2-GFP in a HEK293 cell expressing Fz4, Dv12, and βarr2-GFP. Cells were treated with the PKC inhibitor GFX109203 (4 μM) for 0 min (A and C) and 1 hour (B and D) at 37°C. In (A) and (B) the intensities of membrane-bound Dvl2-GFP are similar, as are the average intensities of βarr2-GFP in the cytosol in (C) and (D). Plasma membrane puncta were defined as accumulations at least 20 units above the cytosolic background on a relative scale of 0 to 255. An untreated cell (C) had 33 puncta (combined intensity, 2871 ± 33 above background). After treatment with GFX109203, only 15 puncta (total intensity, 716 ± 15) remained (D). (E) Interaction of Dvl2 with PKC isoforms. Cell extracts from HEK293 cells overexpressing vector or Myc-Dvl2were immunoprecipitated with anti-Myc. Immunoprecipitates were immunoblotted with antibody against PKCα, PKCβ, or PKCγ (Anti-pen PKC, Upstate) (upper). Whole-cell lysates were immunoblotted with antibodies to Myc or PKC (middle, lower). (F) PKC phosphorylation of Dvl2 stimulates βarr2 binding. HEK293 cells were transfected with Myc-Dvl2, and cell extracts were immunoprecipitated with anti-Myc. Immunoprecipitated Dvl2 was phosphorylated in the presence of [γ-32P]ATP by either endogenous Dvl2-associated kinase(s) or purified PKCα. Immunoprecipitated Dvl2 was processed for autoradiography (upper) or incubated with cell lysates expressing βarr2 for more than 3 hours at 4°C. Dvl2 immunoprecipitates were immunoblotted with anti–β-arrestin (A1CT) (lower). (G) Phosphorylation of endogenous Dvl2 by PKC. HEK293 cells were unstimulated (lane 1) or stimulated with PMA (1 μM) for 15 min at 37°C without (lane 2) or with preincubation of 0.5 μM PKC inhibitor staurosporine for 20 min at 37°C (lane 3). The cell extracts (lanes 1 to 3) were subjected to treatment with alkaline phosphatase (lanes 4 to 6). Proteins from cell lysates were immunoblotted with anti-Dvl2. Alk. Phosphatase, alkaline phosphatase. Scale bar, 10 μm.

The role of Dvl2—to recruit βarr2 and thereby promote internalization of Fz4— reinforces Dvl's central role in coordinating regulation of frizzled receptor activity. The requirement for concurrent Wnt5A stimulation and PKC activation for Fz4 internalization suggests a complex mechanism in which signals besides those resulting from direct Fz stimulation may be necessary to trigger Fz endocytosis. The ultimate purpose of βarr2-mediated internalization of Fz4 remains unknown and might relate to Wnt gradient formation, signaling, or receptor down-regulation.

Supporting Online Material

www.sciencemag.org/cgi/content/full/301/5638/1391/DC1

Materials and Methods

Figs. S1 to S3

References

References and Notes

View Abstract

Navigate This Article