The Adenomatous polyposis coli (APC) protein is an important tumour suppressor in the colonic epithelium 1. A key function of this highly conserved protein is to antagonize Wnt signalling, by constitutively downregulating the transcriptional activity of β-catenin/Armadillo, a key effector of the Wnt signalling pathway 2. Loss of this function is thought to be critical in the initiation of colorectal tumorigenesis as it causes a transcriptional switch in the intestinal epithelium towards actively dividing crypt progenitor cells 345. APC proteins are highly conserved among vertebrates and flies, and flies encode two APC proteins with overlapping roles in Wnt signalling during development 67.

However, APC proteins have additional functions in connection with the actin and microtubule cytoskeletons that appear to be separate from their function in controlling Wnt signalling 89. One of these functions is a role of APC in facilitating cellular adhesion, as indicated by studies in Drosophila tissues 10 and in mammalian colorectal cancer cells 11. This function in cellular adhesion is likely to be conferred by the subcellular pool of APC protein that is associated with adherens junctions (AJs) in Drosophila 1213 and in polarised mammalian cells 14. The mechanism by which APC facilitates cellular adhesion is unknown.

In order to explore this mechanism, we asked whether Drosophila E-APC (also called dAPC2) might have a structural role at AJs. If so, E-APC would be expected to be stably associated with AJs, similarly to the structural components of the adhesive complex. As in mammalian epithelia 1516, the main functional components of this complex in Drosophila are the transmembrane protein E-cadherin, a calcium-dependent trans-membrane adhesion molecule, and the catenins (Armadillo and α-catenin) that link E-cadherin to the actin cytoskeleton at the cytoplasmic side 171819202122. We thus conducted photobleaching experiments on live embryos expressing E-APC or structural AJ components tagged with green fluorescent protein (GFP) 232425, to compare their relative mobility. These experiments revealed that GFP-E-APC is less stably associated with AJs than their structural components. We also found that GFP-E-APC is remarkably mobile in neurons.

We used the GAL4 system to express GFP-E-APC throughout the embryo, and found that its subcellular distribution is very similar to that of endogenous E-APC in fixed embryos. In particular, GFP-E-APC is concentrated underneath the plasma membrane in apicolateral regions of embryonic epithelial cells (Fig. 1a; Fig. 2b). These regions form the zonula adherens (ZA) which contains the AJs 26; they can also be visualised with antibody staining against α-catenin 27 (Fig. 1b), and we observe a remarkably close coincidence of GFP-E-APC and α-catenin. Similar results were obtained by Akong at el. who examined the same GFP-E-APC transgene in the embryo 7, and who also showed that GFP-E-APC is distributed similarly as endogenous E-APC in larval neuroblasts 28.

Next, we conducted fluorescence recovery after photobleaching (FRAP) experiments in live embryos, to examine how stably GFP-E-APC is associated with adherens junctions. We bleached the fluorescence in a defined square centred over the junctional region of an epithelial cell with a short laser pulse, and examined the recovery of the fluorescence within this square over time (Fig. 2a). This revealed a relatively fast rate of recovery of most of the fluorescence within a few minutes (Fig. 3a) [see additional file 1]. Quantitative analysis shows that nearly 80% of the initial fluorescence is recovered within ~220 seconds (with a half-time value of ~60 seconds) (Fig. 4a,4e). This indicates that the association of E-APC with the ZA is dynamic rather than stable. The fluorescence recovery observed in these FRAP experiments could be due to movement of E-APC between the cytoplasm and the plasma membrane, but also to sideways movement along the plasma membrane (see also below).

We also conducted FRAP experiments with structural AJ components, namely E-cadherin-GFP, Armadillo-GFP and α-catenin-GFP. In these cases, we can only recover a small fraction of the initial fluorescence within the time frame of the experiment (Fig. 4a,4b,4c,4d,4e; note that these experiments cannot be extended beyond ~6 minutes, due to the extensive cell shape changes during this developmental stage). Furthermore, the rate of recovery is slower than that observed with GFP-E-APC, with estimated half-times of >3 minutes (α-catenin-GFP and of E-cadherin-GFP; Fig. 4b,4d,4e). This also appears to be true for Armadillo-GFP (Fig. 4c,4e), though we cannot estimate its half-time of recovery with confidence, given that its fluorescence levels are considerably lower than that of the other GFP-tagged protein examined in this study.

We conclude that E-APC is significantly more mobile than the structural AJ components. This suggests that E-APC shuttles either within the cortex, along the zonula adherens, or that it shuttles from the cytoplasm to the plasma membrane (as previously proposed; 29). Interestingly, the observed rates of recovery of GFP-E-APC were much slower than the estimated rate of free diffusion (e.g. 30; the rate of recovery of GFP alone was <10 seconds, i.e. too fast to be measured by our experimental setup). This suggests that the movements of GFP-E-APC are primarily determined by the kinetics of its binding to ligands. One of these could be Axin which associates with E-APC in Drosophila cells to from large dot-like structures 31. Similarly, Axin associates with APC in mammalian cells to form large molecular weight protein complexes 32. Our observations argue against a structural role of E-APC in cellular adhesion. However, they are consistent with a catalytic role of E-APC in facilitating cellular adhesion, for example by maintaining the junctional pool of Armadillo 102933. In support of this, recent evidence suggests that there is rapid exchange of β-catenin within the junctional cadherin complex, and that APC is required for this process 34.

In late embryonic stages, GFP-E-APC forms striking patches underneath the apical plasma membrane of epidermal cells that are in the process of forming denticle extrusions (Fig. 5). These striking 'pre-denticle' patches are also seen in embryos stained with antibody against E-APC, and overlap with actin patches 25. They may thus represent an actin-dependent association of E-APC as that seen in the cortex of earlier epithelial cells and at the ring canals between nurse cells within the egg chambers 2533. FRAP experiments revealed that the presence of GFP-E-APC in these pre-denticle patches is also dynamic, with an estimated half-time of fluorescence recovery of 200–300 seconds (Fig. 5). Again, E-APC is therefore unlikely to have a structural role in these patches.

It has been reported that E-APC and Armadillo are required for anchoring mitotic spindles in the cortex of dividing blastoderm cells in the early Drosophila embryo 25. We cannot measure the kinetics of GFP-E-APC association with the cortex in these early embryonic cells, because of insufficient expression levels at this stage. However, assuming that these kinetics do not change radically during embryonic development, our observations from the later embryos (Fig. 3,4,5) suggest that E-APC has a catalytic role in capturing microtubules in the cellular cortex, rather than providing a structural tether 25.

We also expressed GFP-E-APC in eye imaginal discs, to examine its subcellular distribution within a larval epithelial sheet. We thus noticed striking puncta of green fluorescence within the axons of the optic stalk that connects these discs to the larval brain (Fig. 6). These puncta resemble the E-APC/Axin-GFP dots that we observe in embryonic cells 31 and in these axons (not shown), and also the E-APC/Armadillo dots that Peifer and colleagues observed in early embryos 25. They may thus represent the Axin destruction complex 31. We performed FRAP experiments, bleaching a 6 μm wide strip perpendicularly across the axons and monitoring the recovery of the fluorescence into the bleached section. This revealed that the GFP-E-APC puncta are remarkably dynamic: they re-appear within the bleached area within a minute, with an estimated half-time of ~100 seconds (Fig. 6) [see additional file 2]. Many of these puncta seem to re-appear from other focal planes, so we cannot be absolutely certain that they represent movement of existing puncta. However, some of the puncta re-appearing in the bleached area can clearly be traced as moving puncta within the same focal plane (e.g. see isolated axon, left-hand side of [additional file 2]). The movement of these GFP-E-APC puncta may be due to tracking (e.g. along microtubules), although we cannot see uni-directionality of movement (i.e. the movement appears to be up and down the axons). The movement we observe in these axons is somewhat reminiscent of that observed with a GFP-tagged truncation of Xenopus APC that misses its C-terminus (thus resembling the overall structure of E-APC): this truncation, despite lacking the putative microtubule-interacting domain within its C-terminus, forms large puncta that can track along microtubules in Xenopus tissue culture cells 35.

GSK3 is inhibited during Wnt signalling 2, and GSK3 mutants in Drosophila (shaggy/zeste white3, or sgg, mutants) therefore mimic constitutive and sustained Wingless signalling 36. The normal level of Wingless signalling in the embryonic epidermis does not appear to change the subcellular distribution of bulk E-APC protein 12, although it does cause a re-location of Axin-GFP/E-APC complexes to the plasma membrane 31. However, a reduction of cortical E-APC has been observed in early sgg mutant embryos 1225. Likewise, in older sgg mutant embryos, the levels of membrane-associated GFP-E-APC are also noticeably reduced (Fig. 7a,7b,7c). However, this does not appear to be due to a change in mobility of GFP-E-APC since the kinetics of fluorescence recovery between wild-type and sgg null mutant embryos were comparable (Fig. 4f). Instead, it is due to a reduction of the overall E-APC protein levels in these mutants: Western blot analysis of 2–16 hour old embryos revealed that the total levels of GFP-E-APC protein were much lower in sgg mutant embryos compared to the wild type (Fig. 7d). The same is true for endogenous E-APC whose levels are also substantially reduced in sgg mutants (Fig. 7e). This indicates that sgg is required for the stability of E-APC protein, and it suggests that sustained Wingless signalling may destabilise E-APC. Similarly, phosphorylation by GSK3 is required for the stability of mammalian Axin, a functional binding partner of APC 37, and the levels of Drosophila Axin in embryos are also reduced after prolonged Wingless signalling 38. Destabilisation of the main components of the Axin complex (Axin and APC) during Wnt signalling may be a positive feedback mechanism resulting in the amplification of the signalling level.

The subcellular distribution of E-APC and its accumulation at the adherens junctions is unchanged in other mutants of the Wingless signalling pathway (including wg, axin, dsh and signalling-defective arm mutants; 31; F. Hamada, X. Yu and M. B., unpublished observations). We thus did not expect any of these mutants to affect the shuttling behaviour of GFP-E-APC to and from the plasma membrane. In support of this, preliminary FRAP experiments indicated that the kinetics of fluorescence recovery are unaffected in dsh null mutant embryos (not shown). Taken together with our results from the sgg mutants, this suggests that the kinetic association of GFP-E-APC with the plasma membrane is unaffected by Wingless signalling.

Our FRAP experiments provided evidence that E-APC is a cytoplasmic shuttling protein whose association with the adherens junctions is highly dynamic. The speed of its shuttling to and from the plasma membrane appears to be constitutive and does not require GSK3 activity. The dynamic association of E-APC with the plasma membrane is consistent with a catalytic role of E-APC, and argues against a structural or tethering role in the cell cortex.

A.C. developed and conducted most of the FRAP experiments; J.M. completed the analysis of GFP-E-APC in live and fixed sgg mutant embryos, including the Western blots; M.B. directed the study, helped with the microscopy and drafted the manuscript. All authors read and approved the final manuscript.