Here we have examined a potential role for endocytosis in Wnt signal transduction. Both Wnt-3A (a mammalian Wnt) and Wg (a Drosophila Wnt) stimulate high levels of β-catenin accumulation in L cells (Figure 3a), suggesting the utility of this cell line for dissecting the role of trafficking in the Wnt pathway. Incubation of L cells for 1 hour at 37°C with Wg-conditioned medium, but not control-conditioned medium (Figure 1a), leads to the appearance of a punctate immunostaining pattern as assessed by confocal microscopy (Figure 1b). Suggestive of internalization into vesicles, this punctate staining pattern is not observed when cells are incubated with Wg at 4°C (Figure 1c), a temperature known to inhibit vesicular trafficking pathways. We next sought to determine whether the internalization of Wnt is mediated by clathrin-coated vesicle formation, by far the most well-characterized mechanism for regulated endocytosis . There are a variety of ways to perturb clathrin-mediated internalization, including the use of small molecules (e.g. monodansylcadaverine (MDC) and chlorpromazine (CPZ)), osmotics (e.g. hypertonic sucrose), dominant-interfering mutant proteins (e.g. dynamin), and loss of function reagents (e.g. siRNAs). MDC is believed to inhibit clathrin-mediated endocytosis by stabilizing nascent clathrin-coated vesicles and preventing uncoating , presumably reducing the availability of free clathrin to assemble at the plasma membrane. Hypertonic sucrose has been shown to prevent the proper assembly of clathrin lattices at the plasma membrane . Moreover, CPZ blocks clathrin-mediated endocytosis through a mechanism whereby adaptor complex 2 (AP2) and clathrin are redistributed away from the plasma membrane, making clathrin unavailable for assembly at the cell surface . Dynamin, a GTPase recruited to clathrin-coated pits, is an additional component of endocytosis which is thought to regulate the final step of vesicle biogenesis by catalyzing the scission of budding plasma membrane . Mutation of lysine 44 (K44E) has been shown to decrease the affinity of dynamin for GTP, and expression of this mutant in cells dramatically inhibits endocytosis from the plasma membrane without affecting vesicular transport processes elsewhere in the cell . Internalization of Wg is markedly reduced when cells are incubated in the presence of MDC (Figure 1d), hypertonic sucrose (Figure 1e), and CPZ (Figure 1f), with most of the staining in these cases being confined to the cell surface. Further, while L cells transiently transfected with either GFP alone or GFP in combination with wild-type dynamin internalize Wg into intracellular vesicles, cells expressing GFP in combination with the K44E mutant dynamin display predominantly cell surface staining (Figure 1g–l). By contrast, Wg internalizes quite robustly in adjacent non-transfected cells surrounding those harboring the K44E-mutated dynamin (Figure 1l). Whereas approximately 70–80% of cells transfected with GFP alone or in combination with the wild-type dynamin internalize Wg, only 20–25% of those expressing the K44E mutant protein exhibit comparable levels of endocytosis (Figure 1m). Consequently, the internalization of Wg into L cells occurs on a time-frame consistent with regulated endocytosis (minutes to an hour), and is sensitive to various manipulations which inhibit clathrin assembly and dynamin-mediated scission of budding plasma membrane.
We next sought to characterize the subcellular localization of internalized Wg. To identify potential intracellular compartments containing the trafficked Wnt, we performed Wg internalization assays in the presence of Cy3-conjugated transferrin. Receptors for transferrin constitutively cycle among the plasma membrane, early sorting endosomes, and perinuclear recycling endosomes, and have served as a classic system for the study of clathrin-mediated endocytosis [9,14]. Incubation of L cells with transferrin for 1 hour at 37°C leads to a steady-state distribution which is both punctate and perinuclear (Figure 2e and 2f), suggestive of the constitutive cycling of the protein from the plasma membrane via vesicles to perinuclear recycling endosomes. Internalized Wg exhibits partial overlap with transferrin, particularly in the perinuclear region of the cell (Figure 2d and 2f). By contrast, co-incubation of cells with Wg and transferrin at 4°C leads to an overlapping accumulation of the two proteins at the cell surface (Figure 2a and 2c). Hence, both Wg and transferrin not only internalize through similar endocytic pathways, but also have a similar steady-state localization in a perinuclear recycling endosome. The lack of complete co-localization of Wg and transferrin suggest that while the two proteins enter the cell through similar compartments, they ultimately have somewhat distinct fates. Indeed, it has previously been demonstrated that Wg transits through multivesicular bodies to the lysosomes for degradation following signal transduction .
Given that Wg is internalized within a time-frame relevant to Wnt signaling in L cells, we sought to determine whether the two processes are functionally coupled. Wnt signaling is typically measured through Wnt-stimulated accumulation of cytoplasmic β-catenin, as well as downstream target gene expression. L cells express very low basal levels of β-catenin and are quite responsive to exogenous mammalian Wnt-3A and Drosophila Wg (Figure 3a), making them a powerful system for measuring Wnt-induced increases in β-catenin. We assayed Wnt-stimulated accumulation of β-catenin in the presence or absence of the same mechanistically-distinct inhibitors which abolish Wg endocytosis. As demonstrated in Figure 3a, both Wnt-3A and Wg induce substantial stabilization of β-catenin over a 3-hour time-course, without altering the levels of a control cytoplasmic protein (GSK3β). By contrast, incubation of cells with MDC (Figure 3b), hypertonic sucrose (Figure 3c), and CPZ (Figure 3d) abrogates the Wnt-3A-stimulated increase in β-catenin. However, none of these inhibitors alter the levels of the control protein GSK3β. To assess the potential requirement for endocytic trafficking of the more downstream process of Wnt target gene expression, we measured Wnt-3A-stimulated expression of a luciferase reporter fused to a promoter containing TCF/LEF binding sites ("TOPFLASH"). As depicted in Figure 3e, vehicle-treated L cells stably expressing reporter constructs yield an approximately 50-fold increase in TOPFLASH activity relative to co-expressed LacZ following a 5-hour stimulation with Wnt-3A. By contrast, pre-incubation of these cells with MDC, hypertonic sucrose, or CPZ completely abolishes Wnt-induced reporter gene activity (Figure 3e). To order the endocytic trafficking step in the pathway, we assessed the sensitivity of various downstream pathway activators to these endocytosis inhibitors. Activation of the pathway through lithium-mediated inhibition of GSK3β is completely impaired when endocytosis is blocked (Figure 3f), a surprising observation which will be discussed later in this report. By contrast, the constitutive expression of cytoplasmic β-catenin in SW480 cells, which harbor an inactive version of APC, is resistant to perturbation of endocytosis (Figure 3g). Taken together, these results suggest that Wnt signaling is heavily dependent upon clathrin-mediated endocytosis, and that the explicit internalization event occurs between the steps of GSK3β inhibition and APC inactivation.
In addition to perturbing clathrin assembly at the plasma membrane, we also examined the effect of the dynamin mutant on Wnt signaling. We measured both Wnt-3A- and Wg-stimulated accumulation of β-catenin on a cell-by-cell basis by immunofluorescence microscopy (Figure 4). Wnt-3A-stimulated β-catenin stabilization was measured in L cells transfected with GFP alone (Figure 4a–4d) or GFP in combination with either wild-type (Figure 4e–4h) or K44E-mutated dynamin (Figure 4i–4l). As observed in the western blot experiments (Figure 3), very little β-catenin is detected in the L cells in the absence of Wnt-3A (Figure 4a,4e,4i). Addition of Wnt-3A, by contrast, leads to prominent immunostaining of nuclear β-catenin in cells expressing GFP alone or in combination with wild-type dynamin (Figure 4c and 43 g where transfected cells are indicated by arrows and non-transfected cells by arrow-heads). However, as demonstrated in Figure 4k, Wnt-3A-stimulated accumulation of β-catenin is completely abolished in cells expressing the K44E-mutated dynamin (identified by GFP in Figure 4l and by arrows in Figure 4k), whereas neighboring non-transfected cells stabilize β-catenin comparably to control cells (Figure 4k where non-transfected cells are indicated by arrow-heads). As depicted in the quantitations (Figure 4m–4n), cells expressing GFP alone or in combination with wild-type dynamin demonstrate Wnt-dependent increases in β-catenin in approximately 70–90% of cells, with only approximately 20–30% of the K44E dynamin-expressing cells conferring a similar level of responsiveness. Taken together with the previous results using MDC, hypertonic sucrose, and CPZ, the ability of K44E-mutated dynamin to block Wnt signaling suggests that endocytosis is a key mechanistic step in the pathway.
The various methods used to inhibit clathrin-mediated endocytosis, including small molecules, osmotics, and dominant interfering mutants, all potently abolish Wnt internalization and signal transduction. In order to further examine the requirement of internalization in Wnt signaling, we have performed loss of function experiments using siRNAs directed against clathrin heavy chain. Although clathrin is composed of both light and heavy chains, it is the latter which is truly critical for establishing the lattice coat necessary for membrane budding, as siRNA-mediated silencing of this subunit has previously been shown to greatly attenuate transferrin endocytosis . Consequently, we generated siRNA oligonucleotides corresponding to the region of clathrin heavy chain which has previously been shown to silence the protein . As illustrated in Figure 5a, siRNAs against clathrin heavy chain selectively ablate this protein without affecting levels of the control protein GSK3β. Moreover, siRNA-mediated silencing of clathrin heavy chain substantially attenuates Wnt-3A-stimulated reporter activity (Figure 5b).