APC controls Wnt-induced β-catenin destruction complex recruitment in human colonocytes

Wnt/β-catenin signaling is essential for intestinal homeostasis and is aberrantly activated in most colorectal cancers (CRC) through mutation of the tumor suppressor Adenomatous Polyposis Coli (APC). APC is an essential component of a cytoplasmic protein complex that targets β-catenin for destruction. Following Wnt ligand presentation, this complex is inhibited. However, a role for APC in this inhibition has not been shown. Here, we utilized Wnt3a-beads to locally activate Wnt co-receptors. In response, the endogenous β-catenin destruction complex reoriented toward the local Wnt cue in CRC cells with full-length APC, but not if APC was truncated or depleted. Non-transformed human colon epithelial cells displayed similar Wnt-induced destruction complex localization which appeared to be dependent on APC and less so on Axin. Our results expand the current model of Wnt/β-catenin signaling such that in response to Wnt, the β-catenin destruction complex: (1) maintains composition and binding to β-catenin, (2) moves toward the plasma membrane, and (3) requires full-length APC for this relocalization.

division 18 . Therefore, intestinal Wnt signaling may convey positional information within the crypt and direct intracellular protein localization based on the location of the Wnt source. In the current study, we will examine the effect of a local, immobilized Wnt signal on colon epithelial cells.
Several challenges have historically limited our understanding of Wnt signaling dynamics as it relates to intestinal homeostasis in the normal and cancerous state. Due to the widespread use of the Apc Min/+ mouse model, much focus has been directed to the small intestine rather than the colon, the site of most tumor-initiating APC mutations in humans 19 . Further, very little is known about Wnt signaling in colon epithelial cells that are from non-malignant origin, as most studies utilize only cultured CRC cell lines. Finally, prior research has mostly relied on overexpression of specific Wnt pathway components or cells treated with soluble Wnt in the media, limiting the ability to elucidate endogenous β-catenin destruction complex dynamics in response to a local Wnt signal (reviewed in 3 ).
Here, we examine the response of endogenous β-catenin destruction complex components to a locally presented Wnt signal in human colon epithelial cells of both malignant and non-malignant origin. We demonstrate for the first time that a localized Wnt source can recruit the signalosome and β-catenin destruction complex in colon epithelial cells and find that this Wnt-induced recruitment requires full-length APC. Our work identifies a novel role for APC in the regulation of destruction complex movement toward the membrane following Wnt exposure.

Results
Wild-type APC, but not truncated mutant APC is recruited toward local Wnt3a in human colon cancer cells. Since APC is mutated in >80% of colorectal cancers and is a major scaffolding protein for the β-catenin destruction complex, we first asked whether APC redistributes toward a localized Wnt3a signal in a panel of human CRC cell lines. Three colon cancer cell lines, each with a different Wnt pathway status, were used: RKO (intact Wnt signaling pathway), HCT116βm (WT APC, stabilized β-catenin due to Ser45 deletion) 20 , and DLD1 cells (APC truncation at amino acid 1452, WT β-catenin, see Fig. 1a). Previous studies relied on Wnt addition to cell culture media, thus limiting the ability to examine responses to a localized stimulus. To address this issue and examine endogenous protein response to a locally applied Wnt source -we treated cells with Wnt3aconjugated or Unloaded-beads.
Cells treated for 12-14 hours with Wnt-or Unloaded-beads were fixed and stained for APC (Fig. 1b). Protein localization was scored as: (A) toward the bead, (B) away from the bead, or (C) no obvious protein polarization (Fig. 1c). Scoring results were validated with line-scan analysis. Full-length APC localized toward the Wnt3a-bead in 76% of RKO (Fig. 1d) and 80% of HCT116βm cells (Fig. 1e). In contrast, truncated APC localized toward the Wnt-bead in only 53% of DLD-1 cells (Fig. 1f). In all cases, the majority of cells treated with Unloaded-beads displayed random distribution of APC, suggesting that physical contact with an Unloaded-bead is not sufficient to re-orient APC. These data demonstrate that Wnt exposure induces APC re-localization toward the Wnt source in multiple human CRC lines. Furthermore, this process appears to be compromised in cells carrying truncated APC but remains functional in cells with an activated Wnt/β-catenin transcriptional program due to stabilized β-catenin. The phenotypic consequences of truncated APC have been previously postulated to act in a "just-right" signaling model, in which truncated APC retains partial β-catenin regulatory function to allow a specific level of Wnt activation 21 . Because truncated APC shows reduced ability to localize toward a Wnt source, it is possible that this partial localization is a method to obtain a precise level of Wnt-response. Wnt components are recruited toward localized Wnt3a in cells with an intact Wnt signaling pathway. Since APC is a key scaffold for the β-catenin destruction complex, we reasoned that additional members of the destruction complex and signalosome may also localize near a Wnt-bead. RKO cells with an intact Wnt signaling pathway were treated with beads and then stained for the scaffolding protein Axin1, kinases CK1-α and GSK-3β, the Wnt receptor FZD7, the ubiquitin ligase β-TrCP, and β-catenin. FZD7 was chosen for its defined role in gastrointestinal homeostasis [22][23][24] .
Distinct Axin1 puncta were observed near the Wnt-bead (Fig. 2a). While RKO cells contain low levels of β-catenin due to an intact Wnt pathway, they do have visible β-catenin staining which can be scored. Treatment with Unloaded-beads failed to induce relocalization of β-catenin (Fig. 2b). However, in cells treated with a Wnt-bead, β-catenin levels increase and β-catenin localized toward the bead (Fig. 2b). Of note, the ubiquitin ligase β-TrCP also localized toward the bead (Fig. 2f). These findings suggest that upon Wnt/FZD binding, the cytoplasmic destruction complex relocates to the membrane and is able to recruit β-TrCP but falls short of targeting β-catenin for destruction. This result is consistent with a model proposed by Li et al. whereby the destruction complex saturates with β-catenin, leaving newly synthesized β-catenin to control downstream Wnt target gene expression following Wnt stimulation 25 . The kinases CK1-α and GSK-3β were also detected near the Wnt-bead (Fig. 2c,d), as was the Wnt receptor FZD7 (Fig. 2e).
To test for a physical interaction between the Wnt-coated beads and destruction complex components, the beads were used to "pull down" Wnt and associated proteins from lysates of Wnt-bead-treated cells. APC and β-catenin each associated more with Wnt-beads than with Unloaded-beads (Fig. 3). Together, these data demonstrate that signalosome and destruction complex proteins concentrate near a local Wnt3a cue and do not fully disassemble in response to ~12 hour exposure to a Wnt3a bead. Furthermore, these data support a model in which β-catenin remains associated with the destruction complex following Wnt stimulation.
Stabilized β-catenin does not impact Wnt component localization in response to Wnt3a. In colorectal cancer, mutations in the APC and β-catenin encoding genes appear to be mutually exclusive. Though most CRCs have APC mutations, about half of the small subset of colorectal cancers with wild-type APC harbor mutations that activate β-catenin 26 . These mutations typically result in loss of specific β-catenin residues www.nature.com/scientificreports www.nature.com/scientificreports/ whose phosphorylation is critical for ubiquitin conjugation and proteasome-mediated destruction. To determine whether such "stabilizing" β-catenin mutations would compromise orientation of Wnt signaling components toward a localized Wnt3a, we examined HCT116βm cells. Unlike the parental HCT116 cells, which express both wild-type and mutant β-catenin, HCT116βm possess only one mutant β-catenin allele encoding a Ser45 deletion 20 . Ser45 phosphorylation by CK1-α primes the successive phosphorylation of Thr41, Ser37, and Ser33 by GSK-3β, a prerequisite for recognition and ubiquitination by β-TrCP 10-12 . Note that APC localized toward the Wnt-bead in both RKO and HCT116βm cells (Fig. 1d,e). www.nature.com/scientificreports www.nature.com/scientificreports/ As seen in RKO cells, Axin1 localized in distinct puncta near the Wnt-bead in HCT116βm cells (Fig. 4a). Despite expressing only stabilized β-catenin, HCT116βm cells exhibited both CK1-α and GSK-3β localized toward the Wnt-bead (Fig. 4c,d). This finding demonstrates that CK1-α recruitment to the destruction complex is not dependent on Ser45 of β-catenin, nor is GSK-3β recruitment dependent on Ser45 phosphorylation. FZD7 was also detected near the Wnt-bead (Fig. 4e). Notably, β-catenin also localized near the Wnt-bead (Fig. 4b), suggesting that the destruction complex maintains association with stabilized β-catenin, even in the absence of degradation. On the other hand, β-TrCP failed to localize toward the Wnt-bead in HCT116βm cells (Fig. 4f), demonstrating that β-catenin Ser45 is necessary for β-TrCP association with the destruction complex. Therefore, β-TrCP does not appear to be an inherent member of the destruction complex, but rather, is recruited following β-catenin phosphorylation. Together, these results demonstrate that an activated Wnt pathway through β-catenin stabilization is not sufficient to block orientation of core destruction complex proteins toward localized Wnt. β-TrCP failed to localize toward the Wnt cue, consistent with CK1-α phosphorylation of β-catenin as a prerequisite for β-TrCP recruitment. cells lacking full-length Apc are compromised for Wnt component localization toward Wnt. Both HCT116βm and DLD1 cells possess an activated Wnt/β-catenin transcriptional program, induced by mutation of β-catenin or APC, respectively. As described, with the exception of β-TrCP, Wnt pathway components remained able to orient toward a localized Wnt source in HCT116βm cells (Fig. 4). DLD1 cells have a mutation in the mutation cluster region that leads to a truncated APC protein lacking the Axin-binding SAMP motifs and most of the β-catenin-binding 20aa repeats (Fig. 1a). However, this truncated APC was reported to co-precipitate with Axin1, suggesting that it maintains its scaffolding properties 25 . Given that truncated APC appeared compromised for orientation toward a Wnt-bead in DLD1 cells (Fig. 1f), we asked whether other β-catenin destruction complex members also displayed diminished localization toward a Wnt3a source.
If APC only serves as a destruction complex scaffold, then we would expect normal destruction complex orientation in DLD1 cells. Instead, we found that Axin1 localization to the Wnt3a-bead was detected in only 30% of DLD-1 cells (Fig. 5a), compared to 61% of RKO cells (Fig. 2a) and was not significantly different than localization to the Unloaded-bead. GSK-3β and β-TrCP were also compromised for Wnt-bead orientation in DLD1 cells (Fig. 5d,f). Of all proteins analyzed in DLD1 cells, only CK1-α and FZD7 were localized toward the Wnt-bead more than the Unloaded-bead (Fig. 5c,e). However, even this Wnt-bead localization was observed much less frequently in DLD1 cells than in RKO or HCT116βm cells, which contain full-length APC. Subcellular β-catenin reorientation could not be assessed by visual inspection of DLD1 cells due to overall high protein levels. However, line scan analyses demonstrated that β-catenin also failed to relocate in response to a Wnt-bead (Fig. S3).
Together, these results indicate that full-length APC is necessary for optimal destruction complex orientation in response to a Wnt cue. Given that truncated APC co-precipitates with Axin1 and the destruction complex in DLD1 cells 25 , our results also suggest that APC mediates destruction complex localization by a mechanism independent of, or in addition to, its role as a scaffold for the complex. Truncated APC in DLD1 cells lacks domains that bind dynein 27 , kinesin 28 , and microtubules [27][28][29][30] . These interactions potentially contribute to movement of Figure 3. Wnt-beads pull-down APC and β-catenin. (a) RKO cells treated with Unloaded-beads or Wnt-beads were lysed and proteins pulled-down with the beads. APC and β-catenin were detected in the Wnt-bead pulldown but not the Unloaded-bead pull-down. Images are from the same gel/membrane cut between 50 and 70 kDa and probed for APC and β-catenin (above 70 kDa) or α-tubulin (below 50 kDa). Multi-channel imaging was performed for APC and β-catenin through IRDye infrared secondary antibodies to allow detection on the same membrane. (b,c) Quantification of APC and β-catenin protein pulled-down by Wnt-beads compared to Unloaded-beads from five independent experiments. Protein levels of APC and β-catenin that were pulled down by Unloaded-beads or Wnt-beads were divided by the respective input protein levels and normalized to the Unloaded-bead to demonstrate fold change of APC (b) and β-catenin (c). Error bars, SEM.
www.nature.com/scientificreports www.nature.com/scientificreports/ the destruction complex toward the Wnt source. However, because localization of FZD7 and CK1-α toward the Wnt-bead was decreased, but still present, it appears that truncated APC retains some ability to localize the complex or that alternative mechanisms may be able to localize these proteins in the absence of full-length APC. www.nature.com/scientificreports www.nature.com/scientificreports/ Apc loss impairs Wnt-induced destruction complex reorientation. Given that the β-catenin destruction complex localized toward a Wnt-bead in CRC cell lines with wild-type APC but was impaired in cells with truncated APC, we hypothesized that destruction complex localization requires full-length APC. To test the APC-dependence of Wnt-induced destruction complex localization, we utilized RKO-APC KO cells, generated through CRISPR/Cas9 by Lee and colleagues 31 . Immunofluorescent microscopy revealed reduced APC signal in RKO-APC KO cells compared to the parental RKO cell line (Fig. 6a). While parental RKO cells displayed Wnt-induced localization of β-catenin in 70% of cells scored (Fig. 2b), this localization was seen in only 25% of the RKO-APC KO cells (Fig. 6b,e). Further, Axin1 and GSK-3β failed to relocalize toward a Wnt-bead in the www.nature.com/scientificreports www.nature.com/scientificreports/ RKO-APC KO cells (Fig. 6c,f,d,g). These data demonstrate that Wnt-induced destruction complex localization in CRC cells is dependent on APC.

Wnt triggers destruction complex reorientation in normal human colonic epithelial cells in an
Apc-dependent manner. In normal colon tissue, adult stem cells facilitate the continual regeneration of the epithelial lining and rely on Wnt ligand for maintenance of the stem cell niche. Having observed Wnt-mediated re-localization of the β-catenin destruction complex in human colon cancer cell lines, we wondered if normal human colon stem cells also displayed this property and if so, whether this relocalization was APC-dependent. To address this, we sought human colonic epithelial cells (HCECs) that were isolated from normal tissue and propagated in culture under near physiological conditions (low oxygen and sera). Shay and colleagues isolated HCECs from routine colonoscopy and immortalized them with expression of cyclin-dependent kinase 4 (Cdk4) and human telomerase (hTERT) 32 . These "HCEC 1CT" cells also express endogenous stem cell markers and are able to differentiate into multiple cell lineages making them a unique and valuable model of normal colon epithelial stem cells.
Using HCEC 1CT cells, we depleted APC levels with siRNA (siAPC) and then determined localization of β-catenin, Axin1 and GSK3β. APC knock-down was efficient in HCEC 1CT and resulted in increased β-catenin protein levels, as expected (Fig. 7a). In cells treated with scrambled siRNA (siCtl), Wnt-bead exposure resulted in re-orientation of APC, β-catenin, Axin1, and GSK-3β toward the Wnt source (Fig. 7b-i). Therefore, normal non-transformed human colonic epithelial cells maintain the ability to re-orient the β-catenin destruction complex toward a Wnt cue. Remarkably, APC knock-down resulted in complete loss of β-catenin, Axin1, and GSK-3β orientation toward the Wnt-bead (Fig. 7c-e,g-i). Distinct β-catenin puncta were visualized near Wnt-beads in siCtl HCEC 1CT cells. In contrast, β-catenin levels were increased with diffuse cytoplasmic staining and prominent nuclear localization in HCEC 1CT cells depleted of APC but lacked localization toward the Wnt-bead (siAPC, Fig. 7c). These data are in agreement with the findings in RKO-APC KO cells (Fig. 6).
Like APC, Axin1 is a key scaffolding protein in the β-catenin destruction complex, and thus, might control proper localization of the complex. Using HCEC 1CT cells, we depleted Axin1 with siRNA (siAxin1) and determined the localization of APC and β-catenin. Both APC and β-catenin still localized toward the Wnt-bead in Axin1-depleted HCEC 1CT cells (Fig. 7l,m). However, this bead-induced redistribution did appear to be slightly compromised compared to that seen in siCtl HCEC 1CT cells. Given that APC-depletion resulted in complete loss of localization of destruction complex components, we conclude that APC is necessary and sufficient for Wnt-induced localization of the destruction complex, while Axin1 can enhance the localization, but is expendable.
Together, these data demonstrate that the β-catenin destruction complex retains the ability to orient toward a localized Wnt source in HCEC 1CT cells and reveal a requirement and novel function of full-length APC in destruction complex trafficking toward a Wnt cue.

Discussion
To investigate the role of APC in Wnt-directed destruction complex localization, we first explored signalosome and destruction complex response in commonly used CRC cell lines with varying Wnt pathway status. The data from these experiments provide novel findings that the β-catenin destruction complex re-localizes toward Wnt in colon epithelial cells and that this redistribution is impaired in DLD1 cells expressing a truncated form of APC. Truncated APC is present in the majority of human CRC, and results in loss of the C-terminal half of APC, which contains domains involved in destruction complex assembly, nuclear localization, and cytoskeletal interactions. Further, we demonstrate that APC loss ablates Wnt-induced destruction complex localization. Unlike the parental RKO cell line, Axin1, GSK-3β and β-catenin are unable to localize toward a Wnt-bead in RKO-APC KO cells. We extended this study to a non-transformed human colon epithelial cell line which exhibits stem cell characteristics and found that Wnt-directed destruction complex localization also occurs in colon epithelial cells of non-malignant origin. When compared to the protein distribution in DLD1 cells, which were already limited in signalosome and destruction complex reorientation in response to localized Wnt, APC depletion in the non-transformed human colon epithelial cells completely abolished Wnt-induced destruction complex localization. Combined with a previous report that the truncated APC found in DLD1 remains associated with destruction complex protein Axin1 25 , it appears that APC utilizes its function in scaffolding as well as interactions in the C-terminus to fully traffic the destruction complex toward a Wnt cue.
Numerous models of the events following Wnt receptor activation have been proposed, however the specific molecular proceedings that occur are still debated due to the use of overexpression of Wnt pathway components, global addition of Wnt to culture media, or the wide range of cell and tissue types used. Many models propose that the destruction complex is either inactivated or partially disassembled following Wnt/co-receptor interaction 3,33-36 . However, other studies have demonstrated that some of the destruction complex may be targeted to the plasma membrane following receptor activation 25,37,38 . Recently, an elegant study exposed mouse embryonic stem cells to Wnt3a-conjugated beads, resulting in APC, β-catenin, and LRP6 localization toward a Wnt-bead and also found that the local Wnt cue could trigger asymmetric cell division 39 . To our knowledge, this is the only study examining the orientation of signalosome and destruction complex components in response to localized Wnt and was performed in pluripotent cells of nonhuman origin and limited to three endogenous proteins.
Our study builds upon the current model of Wnt signal transduction and brings to light a new role for APC. Here, we report the requirement of APC for proper Wnt-induced localization of the β-catenin destruction complex in both malignant CRC cell lines and in non-transformed colon epithelial cells. In contrast to our findings, others have demonstrated that key members of the complex are degraded or endocytosed following prolonged Wnt exposure 25,31,[40][41][42] . It is possible that these differences reflect more spatially restricted Wnt contact in our study compared to global Wnt exposure. It also seems likely that the bead attached to Wnt ligand in our study was so large as to prohibit endocytosis. Potentially related, recent evidence points to a role for the central region of APC in preventing clathrin-mediated endocytosis in the Wnt-off state 31 . Over time, the complexity of potential APC functions related to Wnt signaling has been gradually revealed. Previous work by our lab and others demonstrated that APC performs roles outside of its classically defined scaffolding function in the β-catenin destruction complex. For example, APC interaction with nuclear β-catenin leads to repression of Wnt target genes through several potential mechanisms: providing access to the transcriptional corepressor CtBP or the E3 ligase β-TrCP, β-catenin sequestration from transcriptional coactivator LEF-1/TCF, or facilitating β-catenin's nuclear export [43][44][45][46][47] .
Together, our results suggest a novel mechanism of β-catenin destruction complex regulation by the APC protein (Fig. 8). In addition to its well-established role as a negative regulator of cytoplasmic β-catenin, we show that APC is also responsible for moving the β-catenin destruction complex to the cell membrane following Wnt exposure. We demonstrate for the first time that the endogenous β-catenin destruction complex reorients toward a localized Wnt signal in an APC-dependent manner in human colon epithelial cells of both normal and malignant origin. Finally, our data support a model whereby the destruction complex remains assembled and bound to β-catenin following Wnt ligand presentation. Because β-catenin accumulates in Wnt-treated cells, it appears that this β-catenin-bound complex is unable to effectively degrade β-catenin. Perhaps APC helps release modified β-catenin from the destruction complex, thereby enabling its proteasomal degradation as previously proposed 25 . Another possibility, is that Wnt stimulation leads to an APC-dependent membrane orientation of the destruction complex which results in complex inactivation. Upon removal of Wnt signal, this already assembled complex would be unlocked and able to process β-catenin for destruction. These two models are not mutually exclusive and might even be interdependent.
Future studies are necessary to clarify the precise mechanism of APC-regulated destruction complex trafficking to the membrane. Or study does not support an absolute requirement for Axin1 in Wnt-mediated destruction complex relocalization. Perhaps Axin2, a Wnt target, is able to compensate for Axin1 in this capacity. Axin has been postulated to localize to Wnt receptors following Wnt activation through a Dvl-dependent mechanism 48 . Based on the compromised destruction complex localization observed in DLD1 cells and APC-depleted HCEC 1CT cells, we suggest that APC is required for Axin docking to membrane-associated Dvl following Wnt exposure. In this light, our results strongly support the model proposed by Tacchelly-Benites et al., based on studies in Drosophila, that APC plays a major role in regulating Axin's signalosome recruitment in response to Wnt signaling by facilitating phosphorylation of Axin by GSK-3β 49 . The compromised localization observed in DLD1 cells suggests that C-terminal domains contribute to this trafficking, potentially through cytoskeletal interactions [27][28][29][30]50,51 . The mechanism by which APC moves the β-catenin destruction complex toward the signalosome is a necessary focus for future studies in order to provide insight into colon epithelial cell biology and further elucidate unknown aspects of Wnt signal transduction. Overall, our findings provide additional mechanistic details of destruction complex behavior following Wnt exposure, and uncover a novel role for APC in Wnt signal transduction.
immobilization of Wnt protein. Wnt3a was immobilized onto Dynabeads as described previously 39 .
Line scan analysis. ImageJ/FIJI (National Institutes of Health, Bethesda, MD) was used for line scan analyses to quantify protein localization in relation to a Wnt-or Unloaded-bead. Line width was set to 50, corresponding to approximately the size of a Dynabead. Lines were drawn beginning at the bead and across the cell, moving through and across the center of the nucleus. For graphical presentation in (Fig. 1c-e), each scan was set to zero by subtracting the lowest intensity value across the line. At least five representative cells were measured per condition. Refer to Supplemental Figs. S1-S3 for line scan analysis of RKO, HCT116βm, and DLD1 cells.
Statistical analysis. Blinded scoring was performed on a subset of samples using the scoring system presented in Fig. 1a. Slides were covered and numbered prior to scoring and imaging, and decoded following completion of each experiment. Experiments were performed a minimum of three times with at least 25 cells scored per condition per experiment. T-tests (two-tailed) were performed using GraphPad Prism 8 to determine statistical significance between score frequency in Wnt-bead and Unloaded-bead groups as well as between control siRNA and APC siRNA samples in the HCEC 1CT experiments.