The Scribble–SGEF–Dlg1 complex regulates E-cadherin and ZO-1 stability, turnover and transcription in epithelial cells Free

Most internal organs consist of a monolayer of polarized epithelial cells surrounding a central lumen, which functions as a barrier that segregates the internal medium from the outside environment (Rodriguez-Boulan and Macara, 2014). The establishment of cell polarity is regulated by the coordinated action of three highly conserved protein complexes: PAR, Crumbs and Scribble (Bilder et al., 2003). The Scribble complex, which comprises Scribble, Dlg1 (Discs Large) and Lgl (Lethal Giant Larvae), was initially identified in Drosophila as a crucial regulator of epithelial polarity (Bilder and Perrimon, 2000; Gateff and Schneiderman, 1974; Mechler et al., 1985; Woods and Bryant, 1991). It was later shown to be involved in the regulation of other cellular processes, including cell–cell adhesion, asymmetric cell division, vesicular trafficking, cell migration and planar cell polarity (Elsum et al., 2012). In mammalian cells, the Scribble complex also plays key roles in the regulation of cell adhesion and polarity (Bonello and Peifer, 2019; Stephens et al., 2018). Notably, dysregulation of the Scribble complex is commonly observed in human cancers and is associated with tumor progression (Elsum et al., 2012). With no known catalytic activity, the proteins in the Scribble complex are believed to function as scaffolding platforms to recruit other binding partners, including the Rho GTPases and their regulators such as RhoGEFs and RhoGAPs, to build spatially distinct signaling complexes (Bonello and Peifer, 2019; Iden and Collard, 2008; Mack and Georgiou, 2014; Ngok et al., 2014; Stephens et al., 2018). However, it is not known which downstream signaling pathways are regulated by the Scribble complex.

Previous work from our laboratory has shown that SGEF (also known as ARHGEF26), a RhoG-specific GEF, forms a ternary complex with two members of the Scribble polarity complex: Scribble and Dlg1 (Awadia et al., 2019). SGEF is targeted to the apical junctional complex in a Scribble-dependent fashion and functions in the regulation of barrier function at tight junctions (TJs), and the formation and maintenance of E-cadherin mediated adherens junctions (AJs) (Awadia et al., 2019). The most notable effect of silencing SGEF expression in epithelial cells is a significant downregulation of the protein levels of E-cadherin in epithelial cells. Additionally, the protein levels of ZO-1 (also known as TJP1) are also decreased (Awadia et al., 2019). The precise mechanisms by which this novel Scribble–SGEF–Dlg1 complex regulates E-cadherin and ZO-1 stability and/or expression levels are not known.

Here, we describe a novel signaling pathway governed by the Scribble–SGEF–Dlg1 ternary complex. Our results show that the three members of the complex are involved in regulating the stability of both TJs and AJs. At TJs, the ternary complex regulates ZO-1 levels and junction permeability, a process that relies on the integrity of the ternary complex and the exchange activity of SGEF. In contrast, only SGEF is involved in regulating E-cadherin protein levels, where its catalytic activity is also essential. Our results demonstrate that silencing SGEF disrupts the E-cadherin–catenin complex at the membrane, which allows for its internalization. This triggers a positive feedback loop, which exacerbates the downregulation of E-cadherin through transcriptional repression, in a process that involves β-catenin signaling and the transcriptional repressor Slug (also known as SNAI2).

Previously, we have shown that SGEF forms a ternary complex with Scribble and Dlg1, and that silencing SGEF expression leads to a reduction in the levels of junctional proteins such as E-cadherin, afadin and ZO-1 in epithelial cells (Awadia et al., 2019). However, the specific roles of Scribble and Dlg1 and the relative contribution of each complex member to the regulation of E-cadherin and ZO-1, were not fully understood.

To determine the contribution of Scribble and Dlg1 in the regulation of E-cadherin, we generated Scribble and Dlg1 knockout (KO) MDCK cell lines using CRISPR/Cas9 (Fig. 1A). We then analyzed the protein levels of various junctional proteins by western blotting (WB), and their localization by immunofluorescence (IF). Interestingly, unlike SGEF KD cells, Scribble and Dlg1 KO cells showed a less pronounced downregulation of E-cadherin, which was still statistically significant compared to both control (CTRL) and SGEF KD cells (Fig. 1A,B). In contrast, the localization of E-cadherin was notably affected in Scribble and Dlg1 KO cells, with weaker junctional staining and increased intracellular punctate structures (Fig. 1D,E, arrowheads). Other AJ proteins, like p120-catenin and β-catenin, showed a similar re-localization pattern without changes in the total protein levels (Fig. 1F; Fig. S1A,B, arrowheads).

These findings suggest that SGEF is crucial for regulating E-cadherin protein levels, whereas Scribble and Dlg1 mainly affect its localization. Note that for the experiments described in Fig. 1, we used a previously established CTRL shRNA cell line (Awadia et al., 2019). To confirm that the observed differences were not due to the different systems used for generating KO versus KD cells, we also generated KO CTRL cells and compared them to shRNA CTRL cells for E-cadherin and ZO-1 levels, as well as general cell morphology. Our results show no significant differences between these control groups (Fig. S2A–C).

Given that Scribble and Dlg1 have been associated with TJ formation, we analyzed the localization of the TJ marker ZO-1 in Scribble KO and Dlg1 KO cells (Elsum et al., 2013; Ivanov et al., 2010; Yates et al., 2013). Interestingly, both Scribble and Dlg1 KO cells phenocopied SGEF KD cells, showing less ZO-1 at the junctions, and a more linear TJ phenotype compared to the curvilinear phenotype in CTRL MDCK cells (Fig. 1D,E; Fig. S1C). This decrease in ZO-1 intensity correlated with decreased protein levels in all conditions (Fig. 1A,C). This contrasts with the downregulation of E-cadherin, which was more pronounced in SGEF KD cells compared to Scribble or Dlg1 KO.

To further characterize this phenotype, we analyzed cell shape parameters defined by the TJ contour (area, circularity, axial and feret ratio) and tortuosity index. Principal component analysis (PCA) revealed that Scribble KO and Dlg1 KO cells resembled SGEF KD cells in junction shape and TJ tortuosity index, with larger apical areas, more regular shapes and less tortuous TJs (Fig. S1D,E).

To assess the functional implications of these phenotypes, we analyzed the ability of CTRL, SGEF KD, Scribble KO and Dlg1 KO cells to form a monolayer and establish an impermeable barrier by measuring the transepithelial electrical resistance (TEER) which measures the charge-selective permeability of small solutes in confluent monolayers grown on semi-permeable filters and provides an indication of TJ barrier function (Anderson and Van Itallie, 2009). As previously reported, SGEF KD cells showed a significant TEER reduction compared to CTRL MDCK cells (Awadia et al., 2019). Similarly, Scribble KO and Dlg1 KO cells exhibited an ∼50% lower TEER compared to CTRL cells, similar to SGEF KD cells (Fig. 1G).

To summarize, Scribble, SGEF and Dlg1 are crucial for regulating ZO-1 levels, TJ architecture and barrier function. However, only SGEF plays a significant role in the regulation of E-cadherin levels.

Our previous results have demonstrated that a catalytic dead (CD) mutant of SGEF fails to rescue the downregulation of E-cadherin in SGEF KD cells, suggesting that the guanine exchange activity of SGEF is necessary for the regulation of E-cadherin levels (Awadia et al., 2019). To better understand the role of SGEF targeting by the ternary complex and its catalytic activity in the regulation of E-cadherin levels, we designed two constructs. The first encoded the catalytic domain of SGEF fused to GFP (DH-PH–GFP). This construct lacks any known targeting information and cannot bind to Scribble or Dlg1. The second encoded the catalytic domain of SGEF fused to the VSV-G protein (VSV-G–DH-PH–GFP), which targets it to the basolateral membrane independently of Scribble and Dlg1 (Fig. 2A) (Keller et al., 2001).

We used these constructs in SGEF KD cells to address two potential scenarios: (1) SGEF activity alone is sufficient to rescue E-cadherin downregulation, regardless of its localization; and (2) SGEF activity localized at the basolateral membrane is needed to rescue E-cadherin protein levels. We stably expressed each construct in SGEF KD cells and analyzed their ability to rescue E-cadherin levels by WB and immunofluorescence. As controls, we also analyzed the previously characterized Rescue wild-type (WT) and Rescue CD stable cell lines (Awadia et al., 2019).

Our results showed that E-cadherin levels were significantly rescued only when the catalytic activity was targeted to the basolateral membrane (Rescue WT and Rescue VSV-G–DH-PH) (Fig. 2B,C). Similarly, ZO-1 protein levels were partially rescued by VSV-G–DH-PH, but not by DH-PH (Fig. 2B,C). The rescue levels for ZO-1 were not as robust as those seen for E-cadherin, which could be attributed to the contribution of Scribble and Dlg1 to the regulation of ZO-1 levels. This is supported by a small but reproducible increase in ZO-1 protein levels when SGEF KD cells were rescued with a CD SGEF, which can restore the ternary complex but not the exchange activity (Fig. 2C) (Awadia et al., 2019).

Immunofluorescence staining showed that targeting the catalytic activity of SGEF to the membrane using VSV-G not only restored the protein levels of E-cadherin and ZO-1, but also their proper junctional localization (Fig. 2D). Quantification revealed that junctional intensity of E-cadherin and ZO-1 in VSV-G–DH-PH-expressing cells was comparable to, or higher than that in CTRL cells (Fig. 2E). This suggests SGEF and RhoG activation might contribute not only to the regulation of E-cadherin and ZO-1 protein levels, but also to targeting and/or stabilizing these proteins at junctions. This small discrepancy with the WB results might be attributed to the fact that this is a mixed population of cells expressing different levels of VSV-G–DH-PH (see Fig. 2D). Notably, E-cadherin was not recruited to the junctions where a cell expressing VSV-G–DH-PH contacted a non-expressing cell (Fig. 2D, arrowheads), consistent with our observations in mixed populations of SGEF KD cells rescued with SGEF WT (Fig. S3A). These findings underscore the role of the Scribble–SGEF–Dlg1 complex in targeting SGEF activity to the basolateral membrane, where its catalytic function is essential for regulating E-cadherin turnover.

We also confirmed that RhoG activity, which is reduced in SGEF KD cells (Awadia et al., 2019), is crucial for rescuing E-cadherin and ZO-1 levels. Our results show that overexpression of a constitutively active form of RhoG (Q61L) rescued both E-cadherin and ZO-1 levels. IF and WB analysis showed that the protein levels and the localization at junctions were restored to WT levels for both E-cadherin and ZO-1 (Fig. S3B–F). Similar to what was seen for VSV-G–DH-PH rescue cells, E-cadherin localized at AJs only when adjacent cells expressed RhoG(Q61L) (Fig. S3B, arrowheads). Overall, our results demonstrate an essential role for the catalytic activity of SGEF in the regulation of both E-cadherin and ZO-1 levels and suggest that the downstream activation of RhoG is a crucial step.

We then investigated the molecular mechanisms controlling the SGEF-dependent downregulation of E-cadherin. Our initial hypothesis was that E-cadherin degradation was enhanced when SGEF expression was silenced, and inhibiting the proteolytic system involved would rescue E-cadherin levels. To test our hypothesis, we used the commercially available inhibitors MG132 and chloroquine to inhibit proteasomal and lysosomal degradation, respectively, and analyzed E-cadherin protein levels by WB and IF. Our results showed a small and reproducible increase in E-cadherin levels in cells treated with the proteasome inhibitor MG132 (Fig. 3A,B). In contrast, inhibition of lysosomal degradation caused no appreciable difference. However, even after overnight treatment with MG132, E-cadherin levels were still almost 50% lower in SGEF KD cells when compared to CTRL, suggesting that inhibiting the proteasome is not sufficient to restore normal levels of E-cadherin in the absence of SGEF (Fig. 3B). This was supported by a post-hoc analysis, which showed that overall, SGEF KD cells treated with the two inhibitors are statistically the same as non-treated cells (Fig. 3B). Additionally, IF analysis showed no significant changes in the localization of E-cadherin when the proteasome and lysosome systems were inhibited (Fig. 3C,D). Overall, our results suggest that the downregulation of E-cadherin observed in SGEF KD cells cannot be attributed to an increase in protein degradation.

Given that the E-cadherin downregulation in SGEF KD cells could not be explained by an increase in protein degradation, we analyzed E-cadherin mRNA levels in CTRL, SGEF KD, Rescue WT and Rescue CD cells using real-time PCR (Fig. 4A). Our results showed that E-cadherin mRNA levels were significantly downregulated in SGEF KD cells, and the re-expression of SGEF WT (Rescue WT), but not SGEF CD (Rescue CD), restored mRNA levels back to normal. This underscores the essential role for the catalytic activity of SGEF in the transcriptional regulation of E-cadherin.

E-cadherin transcriptional downregulation has been widely studied in the context of epithelial-to-mesenchymal transition (EMT), where the Snail and Zeb families of transcriptional repressors play a key role (Batlle et al., 2000; Cano et al., 2000; Comijn et al., 2001; Conacci-Sorrell et al., 2003). To understand the transcriptional downregulation of E-cadherin in SGEF KD cells, we analyzed the expression levels of the transcription factors known to be associated with E-cadherin repression. As a positive control, we treated the cells with hepatocyte growth factor (HGF), which is known to induce the expression of the Snail family members and induce EMT in MDCK cells (Lee et al., 2011; Leroy and Mostov, 2007). We found that Slug, a Snail family member, was significantly upregulated (up to 10-fold) in SGEF KD cells (Fig. 4B,C; Fig. S4A). Other known E-cadherin repressors, including Snail (also known as SNAI1) and ZEB1 showed no change (Fig. S4B,C). In addition, other EMT markers, like the intermediate filament protein vimentin and N-cadherin showed no change in protein levels between CTRL and SGEF KD cells (Fig. S4D,E), suggesting that SGEF KD cells have not undergone complete EMT.

We then used our previously established cell lines to understand the contributions of the catalytic activity of SGEF in the regulation of Slug. Similar to what we observed for E-cadherin, the catalytic activity of SGEF was essential for the regulation of Slug protein levels, as they were restored to CTRL levels only when SGEF WT was re-expressed (Rescue WT), but not with SGEF CD (Rescue CD) (Fig. 4B,C).

Slug can be regulated both at the mRNA and protein level (Kim et al., 2014; Moon et al., 2021; Vallin et al., 2001; Wang et al., 2009). Our results showed that the changes observed in Slug protein levels closely correlated with a significant increase of Slug mRNA in SGEF KD cells, suggesting that Slug protein levels are regulated at the level of transcription (Fig. 4D). Slug mRNA levels are rescued by re-expressing SGEF WT, but not SGEF CD, confirming the role of SGEF catalytic activity in regulating Slug expression through transcription. We also analyzed the contribution of the proteolytic systems in regulating Slug levels, given that Slug and other Snail family members are short-lived proteins that are regulated by polyubiquitylation and degradation by the 26S proteasome system (Diaz et al., 2014). Our results showed that inhibiting the proteasomal or lysosomal systems had no significant effect on Slug protein levels in SGEF KD cells (Fig. S4F,G).

Interestingly, we saw a small but reproducible increase in Slug levels when we knocked out Scribble and Dlg1 (Fig. 4E). Although not significantly different from CTRL (Fig. 4F), this increase correlated with the small decrease in E-cadherin observed in Scribble and Dlg1 KO cells (Fig. 1B) and might reflect an intermediate phenotype. This prompted us to test whether this intermediate phenotype was also correlated with a decrease in SGEF levels. Owing to the lack of reliable antibodies, we were only able to analyze SGEF mRNA levels using quantitative (q)PCR (Fig. 4G). Our results show a slight decrease in SGEF mRNA levels in both Scribble and Dlg1 KO cells, with levels in between those of CTRL and SGEF KD cells.

In summary, we show that E-cadherin is downregulated at the transcriptional level in SGEF KD cells, which is accompanied by the upregulation of the transcriptional repressor Slug. The regulation of both E-cadherin and Slug levels depends on the catalytic activity of SGEF.

To confirm that the downregulation of E-cadherin in SGEF KD cells is mediated by the upregulation of Slug levels, we stably silenced the expression of Slug in SGEF KD cells using lentivirus to generate SGEF and Slug double KD (dKD) cells. We hypothesized that silencing Slug in SGEF KD cells would rescue E-cadherin protein levels. Our results showed that Slug expression was efficiently silenced in the stable dKD cells, with Slug protein levels comparable to those in CTRL MDCK cells (Fig. 5A,B). Importantly, when compared to SGEF KD cells, E-cadherin levels were rescued in the dKD cells to a level that was even higher than those of CTRL cells (Fig. 5A,C). Surprisingly, silencing Slug did not restore ZO-1 levels, which were also downregulated in SGEF KD cells, as previously shown (Awadia et al., 2019) (Fig. 5A,D). Analysis of ZO-1 mRNA levels by qPCR shows a significant decrease in SGEF KD cells when compared to CTRL cells (Fig. 5E). These results suggest that ZO-1 is regulated at the transcriptional level, independently of Slug.

IF analysis of E-cadherin and ZO-1 distribution revealed that E-cadherin levels were rescued in dKD cells. However, E-cadherin was not exclusively confined to lateral junctions and was also present in the cytosol and basal membrane (Fig. 5F, arrowheads). In contrast, there was no apparent recovery in the ZO-1 signal at junctions in the dKD cells (Fig. 5F). Quantification of E-cadherin and ZO-1 intensity at the junctions corroborated the WB results. It also provides additional information suggesting that, in the absence of SGEF, restoring E-cadherin protein levels is not sufficient to restore a normal junction morphology. First, E-cadherin is more diffuse at the junctions in the dKD cells (Fig. 5G), suggesting possible defects in targeting, stability or retention at the membrane. Second, there was a small but reproducible increase of ZO-1 intensity at the junctions in the dKD cells (Fig. 5G). Given that ZO-1 protein levels did not increase when compared to SGEF KD cells (Fig. 5A,D), this might reflect improved targeting of ZO-1 to TJ, which might be indirectly regulated by the increase of E-cadherin at AJs, as TJs and AJs are highly interdependent (Campbell et al., 2017).

Overall, our results demonstrate that E-cadherin downregulation in SGEF KD cells is driven by an upregulation of Slug and suggests that ZO-1 expression is regulated through a different mechanism. In addition, they suggest that rescuing E-cadherin protein levels without SGEF is not sufficient to restore its proper localization.

The β-catenin signaling pathway has been reported to be a positive stimulator of EMT, increasing cell invasion and metastasis (Novak and Dedhar, 1999). Interestingly, β-catenin activation functions as a transcriptional co-activator for the T-cell factor (TCF) and lymphoid enhancer-binding factor (LEF) transcription factors (Valenta et al., 2012). To determine whether the β-catenin pathway was responsible for the upregulation of Slug levels observed in SGEF KD cells, we treated cells with iCRT3, an inhibitor of the β-catenin-responsive transcription (CRT) (Gonsalves et al., 2011). Importantly, iCRT3 inhibits β-catenin transcriptional coactivator activity but has no effect on its interaction with E-cadherin (Gonsalves et al., 2011). Our results show that treating SGEF KD cells with iCRT3 reduced Slug protein to CTRL levels (Fig. 6A,B), and rescued E-cadherin to levels that were higher than those in CTRL cells (Fig. 6A,C). Surprisingly, and in contrast to the results obtained with the SGEF and Slug dKD cells, the decrease in ZO-1 levels observed in SGEF KD cells was also restored to CTRL levels after iCRT3 treatment (Fig. 6A,D). This suggests that β-catenin signaling pathway is regulating ZO-1 expression via a transcriptional repressor other than Slug.

We then compared the localization of endogenous E-cadherin and ZO-1 between CTRL, untreated SGEF KD and iCRT3-treated SGEF KD cells. We found that E-cadherin intensity was rescued when SGEF KD cells were treated with iCRT3. In iCRT3-treated SGEF KD cells E-cadherin localized both at the lateral and basal membranes, as opposed to the typical lateral membrane localization observed in CTRL cells (Fig. 6E, arrowheads). The redistribution of E-cadherin was similar to that observed in SGEF and Slug dKD cells (Fig. 5E). In addition, iCRT3-treated cells showed increased ZO-1 intensity and more cytosolic E-cadherin (Fig. 6E, xz cross section). Quantification of E-cadherin and ZO-1 intensity at the junctions supports our WB results, showing a rescue in signal intensity at the junctions (Fig. 6F).

To assess the functional relevance of the β-catenin signaling pathway in SGEF KD cells, we analyzed the ability of CTRL, SGEF KD, SGEF and Slug dKD and iCRT3-treated cells to form a monolayer and establish an impermeable barrier, using TEER. Here, we found that inhibiting β-catenin signaling pathway in SGEF KD cells was sufficient to restore the TEER to levels comparable to those in CTRL cells (Fig. 6G). By contrast, TEER was not rescued in SGEF and Slug dKD cells, most likely because silencing Slug in SGEF KD cells did not rescue ZO-1 levels (Fig. 5A,D). Supporting these findings, we saw a significant increase in the tortuosity of the iCRT3-treated cells, but not in SGEF and Slug dKD cells when compared to SGEF KD cells (Fig. 6H). However, this increased tortuosity in iCRT3-treated cells was still not a complete rescue when compared to the tortuosity of CTRL cells. These results suggest that the decrease in barrier function observed in SGEF KD can be attributed, at least in part, to the loss of ZO-1.

In summary, we show that E-cadherin and ZO-1 levels are regulated by a β-catenin-mediated signaling pathway. Interestingly, E-cadherin is regulated by Slug, whereas ZO-1 is independently regulated by a yet-to-be-characterized different factor. Finally, we conclude that restoring ZO-1 protein levels is sufficient to restore barrier function in the absence of SGEF.

As cell–cell adhesions are established, E-cadherin molecules from neighboring cells initially form weak monomeric trans interactions. These monomeric E-cadherin trans-pairs can then interact in cis to form oligomeric arrays or clusters, which help to strengthen and stabilize the interaction as the AJ matures (Troyanovsky et al., 2021). The ability of E-cadherin to cluster, and the dynamics of these clusters can also be regulated by multiple factors, including the interaction of the cadherin complex with the actin cytoskeleton, actomyosin contractility, other cis-interactions and additional adhesion proteins (Troyanovsky et al., 2021; Yap et al., 2015). As a result, the mobility of E-cadherin at mature junctions is very limited when compared to nascent or immature junctions (Adams et al., 1998; de Beco et al., 2009; Sako et al., 1998; Yamada et al., 2005). We hypothesized that the Scribble–SGEF–Dlg1 ternary complex plays a role in the regulation of AJ stability and that disrupting the complex would result in increased E-cadherin mobility at the membrane.

To study the diffusion of E-cadherin in CTRL and SGEF KD cells, we tagged endogenous E-cadherin with mScarlet using a CRISPR knock-in (KI) system (Fig. 7A,B) (Bollen et al., 2022). As expected, the total protein levels of the endogenous-tagged E-cadherin are significantly lower in SGEF KD cells, which suggests that the KI did not interfere with the normal regulation of E-cadherin levels. We then used fluorescence recovery after photobleaching (FRAP) in CTRL and SGEF KD E-cadherin–mScarlet KI cell lines to study E-cadherin dynamics. Given that it is difficult to perform FRAP experiments in cells grown in Transwell filters, we adapted a previously described protocol that allowed us to grow cells on the bottom of an inverted Transwell filter for 6 days after confluency (Miyazaki et al., 2023). Before the experiment, the filter was mounted on a 3D printed support on a glass bottom dish that positioned the cells at a distance from the glass that could be imaged with a 63× oil objective. We then photobleached mature junctions and analyzed their recovery. Fig. 7C shows representative pre-bleach, bleach and post-bleach frames from CTRL and SGEF KD cells (see Movies 1 and 2). The fluorescence recovery along junctions was analyzed as a function of time and displayed as kymographs and average recovery graphs (Fig. 7D,E). The kymograph analysis showed that in CTRL cells the bleached area remained uniform with no significant fluorescence recovery within the first minute after bleaching, whereas in SGEF KD cells it recovered rapidly (Fig. 7D). FRAP quantification (Fig. 7E) confirms that the speed at which the signal recovered after the bleaching event was faster for SGEF KD cells, with a half-time of 106.6 s compared to 140.2 s for CTRL. In addition, the overall fluorescence recovery showed a striking difference between the immobile fractions of the two cell lines, with CTRL cells having a higher immobile fraction (55.5%) than SGEF KD cells (44.4%). Overall, this data shows that in SGEF KD cells, the basolateral pool of E-cadherin is more mobile compared to that in CTRL cells, suggesting that E-cadherin is less stable.

Our results suggest that the members of the Scribble–SGEF–Dlg1 complex, particularly SGEF, function to stabilize E-cadherin at the membrane. E-cadherin stability at the basolateral membrane has been shown to be dependent on its interaction with p120-catenin. Dissociation of the E-cadherin–p120-catenin complex or reducing p120-catenin levels results in a significant increase in E-cadherin internalization (Davis et al., 2003; Ireton et al., 2002; Ishiyama et al., 2010).

We did not observe a decrease in p120-catenin or β-catenin levels upon SGEF silencing, but their localization was affected (Fig. 1F; Fig. S1A,B) (Awadia et al., 2019). Based on these results, we hypothesized that silencing SGEF might lead to dissociation of the E-cadherin–p120-catenin complex, which would stimulate E-cadherin internalization and the subsequent release and translocation of β-catenin to the nucleus. To test this hypothesis, we co-immunoprecipitated the E-cadherin–p120-catenin complex in CTRL and SGEF KD cells, using anti-p120-catenin antibodies and blotting for E-cadherin. Our result showed a significant decrease in the amount of co-immunoprecipitated E-cadherin in SGEF KD cells (Fig. 8A). The reciprocal co-immunoprecipitation, E-cadherin immunoprecipitation and blotting for p120-catenin showed similar results, with reduced levels of p120-catenin being precipitated in SGEF KD cells (Fig. 8B). These results suggest that SGEF plays a role in the formation and/or stabilization of the E-cadherin–p120-catenin complex.

One prediction from these results is that E-cadherin internalization should be increased in SGEF KD cells, and that inhibiting endocytosis should restore E-cadherin protein levels. To test this, we used two endocytosis inhibitors: dynasore (for clathrin-mediated endocytosis, CME) and MβCD (for clathrin-independent endocytosis) (Fig. 8C,D). Our results showed that dynasore treatment, but not MβCD, restored E-cadherin protein levels almost completely in SGEF KD cells, confirming the role of endocytosis in the downregulation of E-cadherin (Fig. 8C-E). E-cadherin localized to AJ after dynasore treatment in SGEF KD cells, but the pattern was slightly more diffuse compared to that in CTRL cells (Fig. 8E; Fig. S5B, C). We also confirmed these results using a different CME inhibitor, namely Pitstop 2 (Fig. S5A).

The results observed after inhibiting CME can be interpreted in two ways: (1) that E-cadherin is still transcriptionally repressed and newly synthesized protein accumulates gradually at the membrane when endocytosis is inhibited; or (2) that inhibiting endocytosis prevents the E-cadherin–catenin complex from being internalized, thus preventing β-catenin release, its translocation to the nucleus, and restoring E-cadherin transcription. Using the same inhibitors, we immunoblotted for Slug as a β-catenin signaling indicator. Surprisingly, we saw a striking reduction of Slug levels in the SGEF KD cells treated with dynasore, supporting our second hypothesis/scenario (Fig. 8F,G). In agreement, immunofluorescence analysis showed that β-catenin is more tightly associated with junctions in dynasore-treated SGEF KD cells when compared to the diffuse localization of non-treated SGEF KD cells or cells treated with MβCD (Fig. S5B,D).

Overall, our findings demonstrate that silencing SGEF disrupts the E-cadherin–p120-catenin complex, promoting E-cadherin internalization and subsequent β-catenin release. Inhibiting CME rescues E-cadherin levels and localization, highlighting the crucial role of SGEF in maintaining E-cadherin stability at the membrane.

The establishment of apicobasal identity is essential for function of epithelial cells, and the loss of cell polarity is a hallmark of many diseases, including cancer (Rodriguez-Boulan and Macara, 2014). The development of an apicobasal axis in epithelial cells and tissues is regulated by the coordinated action of three multi-protein complexes: PAR, Crumbs, and Scribble (Bilder et al., 2003). In mammalian cells, the Scribble complex plays key roles in the regulation of cell adhesion and polarity (Bonello and Peifer, 2019; Stephens et al., 2018). The function of the Scribble complex has also been associated with the regulation of other cellular processes, such as cell migration and planar cell polarity (Dow et al., 2007; Montcouquiol et al., 2003; Osmani et al., 2006). Initially characterized in Drosophila, the Scribble complex includes Scribble, Dlg1 and Lgl (Bilder and Perrimon, 2000). However, each of these three proteins encodes multiple protein–protein interaction domains and function as scaffolding platforms to recruit other binding partners. Several interaction partners have been identified for the Scribble complex, including junctional proteins and Rho GTPase regulators, but the molecular mechanisms controlling its function are still poorly understood (Stephens et al., 2018).

We have previously shown that the RhoG-specific GEF SGEF interacts simultaneously with Scribble and Dlg1, forming a ternary complex, which regulates the assembly and function of AJ and TJ in both 2D and 3D (Awadia et al., 2019). Importantly, silencing SGEF results in a significant downregulation of E-cadherin, a central component for cell–cell adhesion, epithelial morphogenesis and homeostasis (Hartsock and Nelson, 2008). E-cadherin downregulation is also an important hallmark of EMT, which can be observed during development and cancer. In epithelial cancers, E-cadherin is frequently downregulated, increasing invasion in vitro and cancer progression in vivo (Bruner and Derksen, 2018; Thiery, 2002; Thiery and Sleeman, 2006).

This study characterizes the role of the Scribble–SGEF–Dlg1 ternary complex in the downregulation of E-cadherin levels and junction integrity. Our work shows that SGEF is the only member that regulates E-cadherin levels. Importantly, SGEF is the only complex member that has a catalytic function, and its catalytic activity is essential for the regulation of E-cadherin (Awadia et al., 2019). In contrast, Scribble and Dlg1 KO cells showed an altered AJ architecture but not E-cadherin protein levels, which suggests a role in the regulation of E-cadherin targeting and/or stability at the AJ, consistent with previous reports (Choi et al., 2019; Dow et al., 2007; Firestein and Rongo, 2001; Ivanov et al., 2010; Laprise et al., 2004; Lohia et al., 2012; Qin et al., 2005; Yates et al., 2013). The difference between SGEF KD and Scribble and Dlg1 KO could potentially be attributed to compensation by other family members; Scribble belongs to the leucine-rich repeat and PDZ protein (LAPP family), which has three other members in mammals, Lano (also known as LRRC1), Erbin and Densin (also known as LRRC7) (Santoni et al., 2002), and there are four Dlg members in mammals (Funke et al., 2005). However, this appears unlikely as it was recently shown that KO of the three LAPP members expressed in epithelial cells had no effect on AJs (Choi et al., 2019). Similarly, even though the other Dlg family members are expressed in epithelial cells, they localize to different parts of the cell, suggesting that they have independent roles (Van Campenhout et al., 2011).

Our results demonstrate that the Scribble- and Dlg1-mediated targeting of SGEF at the basolateral membrane and the downstream activation of RhoG are crucial for E-cadherin regulation. Given that previous studies show that SGEF exists in an autoinhibited state, we believe its association with the complex might function to release the autoinhibition and activate SGEF locally (Zhang et al., 2021). Furthermore, we show that all members of the Scribble–SGEF–Dlg1 complex play a role in the regulation of TJ architecture and barrier function, and display similar changes in cells morphology when their expression is silenced. This is consistent with previous work showing that Scribble and Dlg1 play a role in the regulation of TJ architecture and function (Ivanov et al., 2010; Stucke et al., 2007).

The effects of perturbing the Scribble–SGEF–Dlg1 complex on TJs appear to be mediated by ZO-1, as silencing SGEF, Scribble or Dlg1 results in a significant decrease of junctional ZO-1. The ZO family of TJ proteins is involved in defining the characteristic ‘wavy’ pattern of MDCK cells, and silencing their expression results in a straightening of the junctions and an increase in permeability (Choi et al., 2016; Fanning et al., 2012; Tokuda et al., 2014; Van Itallie et al., 2009). Interestingly, Scribble interacts with ZO-1 (and ZO-2), and this interaction is important for TJ function, potentially stabilizing ZO-1 (Ivanov et al., 2010; Metais et al., 2005). This is supported by a report that shows that overexpression of Scribble in MCF10a cells, which lack an organized TJ structure, restores ZO-1 localization to junctions and barrier function (Elsum et al., 2013). Our results show that the regulation of ZO-1 levels and TJ function is more complex than the downregulation of E-cadherin and is not only influenced by the catalytic activity of SGEF, but also by Scribble and Dlg1. Interestingly, we show that KO of Scribble and Dlg1 promote a reduction in SGEF mRNA levels, which suggests some of the observed effects might be due to reduced SGEF activity. We cannot yet conclude from our results whether the three members of the complex control ZO-1 through the same pathway (e.g. by regulating the activity of SGEF) or through parallel pathways mediated by Scribble and/or Dlg1. The fact that rescuing with just the catalytic activity of SGEF (VSV-G–DH-PH) does not completely restore ZO-1 levels, supports a synergistic role between SGEF and Scribble and/or Dlg1.

Our results demonstrate that E-cadherin downregulation in SGEF KD cells is a result of a decrease in its mRNA levels and is mediated by the transcriptional repressor Slug, which is regulated by the β-catenin signaling pathway. The catalytic activity of SGEF is essential for the regulation of Slug levels, as the catalytic dead mutant cannot rescue the phenotype. In contrast, Scribble and Dlg1 KO cells displayed only a minor increase in Slug levels, confirming the key role of SGEF in the regulation of E-cadherin levels. Interestingly, a Slug and SGEF dKD rescues E-cadherin levels, but not its proper localization, suggesting that an intact Scribble–SGEF–Dlg1 complex stabilizes E-cadherin at the membrane. Our results also suggest that proteasomal degradation of E-cadherin might contribute to its downregulation, but apparently not at a significant level. With stable SGEF KD cells, we cannot distinguish whether E-cadherin proteasomal degradation occurs early, when SGEF is being actively downregulated, and then stops when transcription is repressed, or if it is not a significant factor. Using inducible SGEF KD to analyze these early time points would help us answer this question.

Our results show that ZO-1, like E-cadherin, is regulated at the transcriptional level in SGEF KD cells in a process mediated by β-catenin, but in a Slug-independent fashion. A potential candidate is ZEB1, a transcriptional repressor also regulated by the β-catenin signaling pathway, that is associated with E-cadherin repression (Eger et al., 2005; Sanchez-Tillo et al., 2011). ZEB1 can also regulate the expression of ZO-1 indirectly via a MAPK-ERK pathway (Liu et al., 2018). Interestingly, Scribble plays a role in the regulation of ZEB1 levels, also via MAPK-ERK (Elsum et al., 2013). We did not detect changes in ZEB-1 expression upon SGEF KD, which suggests ZO-1 might be regulated by a different pathway. It is also possible that the SGEF regulates ZEB-1 phosphorylation and not its expression levels, as ZEB-1 transcriptional repression activity has been shown to be regulated by phosphorylation (Llorens et al., 2016). We plan to explore these possibilities in future studies.

The amount of E-cadherin at the plasma membrane is determined by a dynamic equilibrium between synthesis, degradation, endocytosis and recycling (Kowalczyk and Nanes, 2012). E-cadherin can be endocytosed through different mechanisms, including clathrin-dependent, clathrin-independent and caveolae-mediated endocytosis (Bryant and Stow, 2004; Ivanov et al., 2004; Le et al., 1999; Lohia et al., 2012; Qin et al., 2005). Once internalized, E-cadherin can be recycled back to the plasma membrane, sequestered in internal compartments, or sent for degradation (Bryant and Stow, 2004). Endocytosis of E-cadherin is triggered by the disassembly of the E-cadherin–p120-catenin complex (Davis et al., 2003; Ireton et al., 2002; Ishiyama et al., 2010). Previous work has shown that the Scribble complex might play a role in endocytosis, sorting and recycling in both Drosophila and mammals, including in the internalization of E-cadherin, NMDA receptors and GPCRs (Lahuna et al., 2005; Lohia et al., 2012; Piguel et al., 2014). Our results suggest that E-cadherin downregulation in SGEF KD cells is initiated by the dissociation of the E-cadherin–p120-catenin complex, followed by E-cadherin endocytosis and eventually degradation, potentially by the proteasome (Fig. S6). Interestingly, inhibiting endocytosis rescued E-cadherin and Slug levels, placing the complex dissociation and internalization of E-cadherin as early events in its downregulation.

In this study, we show that the activation of RhoG by SGEF at the basolateral membrane is essential for the regulation of E-cadherin stability and expression, as well as of ZO-1 levels and barrier function. However, we did not explore the downstream effectors of RhoG. Perhaps the most likely candidate is the RhoG-specific effector engulfment and cell motility 2 (ELMO2), which has been involved in the recruitment of E-cadherin to initial cell–cell adhesion sites in MDCK cells (Katoh and Negishi, 2003). The SGEF–RhoG axis might be involved in the recruitment of ELMO2 to cell junctions to locally activate Rac1. ELMO2 can activate Rac1 by forming a complex with the Rac1-GEF dedicator of cytokinesis 1 (DOCK1) (Toret et al., 2014a). Interestingly, ELMO2 has also been shown to function in Rab11-mediated recycling of E-cadherin, together with the integrin-linked kinase (ILK) (Chen et al., 2013; Tan et al., 2001; Wu et al., 1998). In addition, ELMO2, RhoG and ILK can form a ternary complex (Ho and Dagnino, 2012; Jackson et al., 2015). Finally, silencing ELMO2 affects E-cadherin localization and protein levels, but not as drastically as silencing SGEF (Toret et al., 2014a,b). This might be a result of incomplete silencing or might alternatively suggest that SGEF and RhoG modulate more than one downstream output, including ELMO2 and additional effectors. Future efforts will be devoted to dissecting the pathway that controls E-cadherin dynamics and turnover downstream of SGEF and RhoG.

An interesting observation from this study was that rescuing E-cadherin without restoring the catalytic activity of SGEF (e.g. SGEF and Slug dKD or iCRT3 treatment) was not sufficient to restore proper localization E-cadherin at AJs, suggesting that SGEF plays an additional role stabilizing E-cadherin at the membrane. As junctions mature, E-cadherin–catenin complexes in AJs form clusters driven by cis- and trans-interactions in the cadherin ectodomain. The stability of these clusters is further enhanced by the binding to F-actin and other interacting proteins (Mege and Ishiyama, 2017; Troyanovsky, 2023; Yap et al., 2015). As E-cadherin assembles into these clusters, it becomes less mobile at the membrane, showing little to no membrane diffusion along mature junctions (de Beco et al., 2009). Our results show that, when SGEF is silenced, E-cadherin localization at junctions is more diffuse, and more mobile at the membrane with a faster diffusion or exchange and lower immobile fraction. Recently, Troyanovsky and colleagues have shown that Scribble can mediate the formation of a specific population of E-cadherin clusters. This involves cis-interactions between E-cadherin complexes and associated proteins like Scribble, which has been shown to interact with β-catenin (Ivarsson et al., 2014; Troyanovsky et al., 2021; Zhang et al., 2006).

Interestingly, although Scribble and Dlg1 KO mice are non-viable (Caruana and Bernstein, 2001; Murdoch et al., 2003), SGEF KO mice appear normal (Samson et al., 2013). There are several potential explanations for this discrepancy. First, it is not unusual for RhoGEFs to elicit a compensation response by other GEFs when downregulated. Second, similarly, the loss of E-cadherin might be compensated in the mouse by other cadherins. In MDCKs we showed an upregulation of K-cadherin in SGEF KD cells (Awadia et al., 2019). Third, it is also likely that Scribble and Dlg1 have functions that are independent of SGEF, as they are expressed in a wide range of tissues (Caruana and Bernstein, 2001; Murdoch et al., 2003). Finally, the SGEF KO mice expresses a truncated form of SGEF encoding the N-terminus (Samson et al., 2013), which should retain some of the functionality of the Scribble–SGEF–Dlg1 complex (Awadia et al., 2019). Future work will focus in understanding the mechanisms that control SGEF activation at junctions, the spatiotemporal regulation of RhoG activity, and the signaling pathway downstream of SGEF and RhoG activation both in vitro and in vivo.

MDCK II cells were a gift from Ian G. Macara (Vanderbilt University, Nashville, TN). MDCK cells were grown in DMEM (Gibco) containing 10% FBS (SH3039603; Cytiva) and antibiotics (penicillin-streptomycin; SV30010; Cytiva). Cell lines were grown at 37°C and 5% CO₂. All experiments were conducted with early passage cells that were passaged no more than 20 times. Mycoplasma was tested regularly by staining with Hoechst 33342 (AnaSpec).

The commercial antibodies against the following proteins were used throughout this study: RhoGDI (SC-360, rabbit polyclonal) 1:10,000 WB; p120-catenin (SC13957, rabbit polyclonal) 1:1000 WB, 1:100 IF; p120-catenin (SC23873, mouse monoclonal) 1:1000 WB, 1:100 IF, from Santa Cruz Biotechnology; tubulin (T9028, mouse monoclonal) 1:50,000 WB, from Sigma-Aldrich; β-catenin (A302-010, rabbit polyclonal) 1:5000 WB, 1:500 IF from Bethyl laboratories; mNeonGreen (32F6, mouse monoclonal) 1:200 IF, from ChromoTek; E-cadherin (24E10, rabbit mAb) 1:1000 WB, 1:200 IF; Slug (C19G7 rabbit mAb) 1:500 WB; Snail (C15D3 rabbit mAb) 1:500 WB; ZEB1 (D80D3 rabbit mAb) 1:500; vimentin (D21H3, rabbit mAb) 1:1000 WB; N-cadherin (D4R1H, rabbit mAb) 1:1000, all from Cell Signaling Technology; ZO-1 (339100, mouse monoclonal) 1:1000 WB, 1:100 IF; Alexa Fluor 488- and Alexa Fluor 594- conjugated anti-mouse-IgG and anti-rabbit-IgG secondary antibodies (A11008, A11001, A11005, A32733 and R37117); and HRP-conjugated anti-mouse-IgG, and anti-rabbit-IgG secondary antibodies from Jackson Immunoresearch (715-035-151, 711-035-152). Alexa Fluor 647 (A22287) conjugated to phalloidin was from Thermo Fisher Scientific

The following reagents were also used throughout this study. Incubations were done overnight (∼16 h) unless indicated otherwise. HGF (R&D-P14210) was used at 100 ng/ml. For endocytosis experiments, dynasore hydrate (Sigma-D7693), Pitstop 2 (MedChemExpress-HY-115604/CS-0103973) and MβCD (Sigma-C4555) were used at 100 µM, 25 µM and 2 mM, respectively. For degradation experiments (R)-MG132 (Cayman Chemical-13697) and chloroquine diphosphate salt (Sigma-Aldrich) were used at 20 µM and 25 µM, respectively. For β-catenin inhibition, iCRT3 (EMD Millipore-219332) was used at 50 µM.

CTRL, SGEF KD, Rescue WT and Rescue CD MDCK cell lines were established in our previous publication (Awadia et al., 2019). The shRNA used to silence Slug in SGEF KD was cloned into pLKO.1-Hygro (Addgene #24150). To generate Scribble and Dlg1 CRISPR/Cas9 KO cell lines we used pLentiCRISPR v2-Blast (Addgene #83480). DH-PH and VSV-G–DH-PH Rescue constructs were cloned using gateway recombination (Thermo Fisher Scientific) into pLenti CMV Hygro DEST (Addgene #17454). We generated stable cells lines using these constructs via lentiviral delivery and antibiotic selection. For KO cell lines, single cell clones were identified and validated. Details for the constructs used in this study, including shRNA- and gRNA-targeting sequences utilized are listed in Table S1.

Lentiviral particles were packed using HEK293FT and a standard calcium phosphate transfection method as described previously in (Awadia et al., 2019). Packaging plasmids pMD2.G, pSPAX2, and the lentivirus plasmid were mixed in a molar ratio 1:1:1 (total 10 µg). pMD2.G and pSPAX2 were a Addgene plasmids #12259 and #12260 (deposited by Didier Trono, EPFL, Lausanne, Switzerland). HEK293FT culture medium was changed 24 h after transfection, and lentivirus particles were harvested 48 h after transfection.

MDCK cells were infected with lentivirus particles overnight. The following day, the infection medium was replaced with complete medium and 500 ng/ml hygromycin to select transformed cells. For some viruses, single-cell colonies were isolated by serial dilution and tested for the corresponding phenotype.

To tag the endogenous E-cadherin gene (Cdh1) from MDCK cells, we followed the protocol described by Bollen and collaborators (Bollen et al., 2022). The gRNA used was 5′-GAGGTGGCGAGGACGACTAG-3′, which targets the region between D882 and the STOP Codon at the C-terminal portion of E-cadherin. This gRNA was cloned into a Cas9 D10A backbone (PX462) (Addgene #62987). The upstream and downstream homology arms, both 500 bp, were cloned into the plasmid TVBB C-term-mScarlet (Addgene #169219) with the purpose of knocking in mScarlet in frame with D882. Both constructs were then electroporated into MDCK cells using the Neon^TM transfection system (Thermo Fisher Scientific). KI positive populations were enriched by serial dilutions.

Cells cultured for 4 days at a starting density of 7×10⁴/cm² were lysed in TRIzol (Ambion) for 5 min and transferred to an RNA-free microcentrifuge tube following manufacturer recommendations. Cell lysates were clarified, and then total RNA extracted with the Direct-zol RNA miniPrep kit (Zymo). cDNA was synthesized using Superscript IV RT polymerase (Invitrogen), RNA pol inhibitor (Applied Biosystems), and 2000 ng of RNA as indicated by the manufacturer. Finally, real-time PCR was carried out using 100–200 ng of template, 200 nM of primers and Power SYBR™ Green PCR Master Mix (Applied Biosystems) in a Bio-Rad thermocycler. Ct values were then normalized by ΔΔCt formula and relative expression compared across at least three independent experiments. For this analysis, we used a two-tailed paired Student's t-test as ΔΔCt normalization lacks variability for CTRL conditions and an ANOVA analysis could be misleading.

Cells cultured for 4 days at a starting density of 7×10⁴/cm² were rinsed with PBS and then scraped into a lysis buffer containing 50 mM Tris-HCl pH 7.4, 10 mM MgCl₂, 150 mM NaCl, 1% Triton X-100 and EZBlock protease inhibitor cocktail (BioVision). The lysates were then centrifuged for 10 min at 20,000 g, the supernatant collected, and protein concentration measured using the DC protein assay (Bio-Rad). For immunoblotting, lysates were boiled in Laemmli buffer, and 20–60 μg of protein was resolved by SDS-PAGE.

For co-immunoprecipitation of the endogenous E-cadherin–catenin complex 1 mg of protein lysate was incubated either 2 µg with anti-E-cadherin (Cell Signaling 4A2 mouse monoclonal, 14472) or anti-p120-catenin (Santa Cruz Biotechnology 6H11 mouse monoclonal, sc-23873) antibody for 1 h at 4°C on a rotor. Next, 20 µl of 20% slurry Protein G Mag Sepharose (Cytiva; 10 μg/ul binding capacity) was added and incubated for 30 min. Immune complexes were recovered by centrifugation (10,000 g for 30 s) and washed three times with lysis buffer. Finally, samples were resuspended in Laemmli buffer, boiled, and resolved by SDS-PAGE.

In both instances, the proteins were transferred onto PVDF membranes and immunoblotted with the indicated antibodies. Immunocomplexes were visualized using the SuperSignal West Pico PLUS Chemiluminescent HRP substrate (Thermo Fisher Scientific). Full uncropped images of blots in this paper are shown in Fig. S7.

To measure TEER, the cells were plated onto a 0.4 μm polyester membrane, at 10⁵/cm² (Corning). Resistance measurements were performed 4 days later in triplicate using an EVOM Epithelial Volt/ohm meter (World Precision Instruments). The values represent the average of three replicates minus the background resistance. Values were normalized to those of CTRL cells (100%).

The immunofluorescence (IF) assays were performed as described before (Garcia-Mata et al., 2003). Briefly, MDCK cells grown on coverslips or Transwell filters (Corning) were fixed for 10 min with 4% paraformaldehyde and quenched with 10 mM ammonium chloride. Cells were then permeabilized with 0.1% Triton X-100 in PBS for 10 min. The coverslips were then washed with PBS and blocked with PBS, 2.5% goat serum (S26; Sigma), 0.2% Tween 20 for 5 min, followed by 5 min of blocking with PBS, 0.4% fish skin gelatin and 0.2% Tween 20. Cells then were incubated with the primary antibody for 1 h at room temperature. Coverslips were washed five times with PBS, 0.2% Tween 20 followed by 5 min of blocking with PBS, 0.4% fish skin gelatin (G7765; Sigma), 0.2% Tween 20 and 5 min with PBS, 2.5% goat serum, 0.2% Tween 20. Secondary antibody diluted in the blocking solution was then added for 45 min, washed five times with PBS, 0.2% Tween 20, and mounted on glass slides using ProLong Diamond Antifade Mountant (P36965; Thermo Fisher Scientific).

Confocal images were collected using a Leica Stellaris 5 laser scanning confocal microscope equipped with an HC PL APO 63×/1.40 OIL CS2 objective, HyD detectors and a Tokai Hit STX stage top incubator (set to 37°C and 5% CO2 for live imaging).

All quantification of junctional E-cadherin and ZO-1 was carried out on confocal z-stacks acquired under Nyquist parameters with a HC PL APO 63×/1.40 OIL CS2 objective.

For segmentation of MDCK monolayers, we used the TJ marker ZO-1. Briefly, we used the plugin Tissue Analyzer from ImageJ (Aigouy et al., 2010) and a 3 µm maximum projection aligned using the brightest point of ZO-1 staining as reference. Segmentation was manually corrected, and junctions exported as regions of interests (ROIs) to the ImageJ ROI manager.

To understand the protein localization at the junctions, we used a perpendicular analysis measuring the staining intensity across. For this, we used ImageJ and did a 3 µm apical projection defined by the brightest point of ZO-1 staining. Then, we manually drew perpendicular lines to the junctions (40 px wide) and measured the intensity for E-cadherin and ZO-1. Then, we used RStudio (https://posit.co/download/rstudio-desktop/) and the Tidyverse package (https://www.tidyverse.org/) to wrangle the data. The brightest point of ZO-1 was used to center all measurements and avoid bias. Data was normalized for each experiment using the min-max feature scaling and graphed using the ggplot library (https://ggplot2.tidyverse.org/). At least three fields from at least three independent experiments were used for quantification unless otherwise indicated.

For the tortuosity index measurements, we defined tortuosity index as the coefficient between the cell perimeter and the minimum polygon connecting the cell tricellular junctions. To achieve this, we segmented cells using the plugin Tissue Analyzer as described above and exported whole cell segmentations as ImageJ ROIs. Then, we manually traced the minimum polygon connecting tricellular junctions for each cell.

To measure E-cadherin dynamics in live cells, we cultured cells in inverted filters following the guidelines from Miyazaki and collaborators (Miyazaki et al., 2023). MDCK cells expressing E-cadherin–mScarlet were plated onto inverted 0.4 μm polyester membrane, at 10⁵/cm² (Corning), and they were incubated for ∼2 h until they attached. Then, filters were placed in their corresponding vessel and grown for 6 days with medium being changed every other day. To be able to image the inverted filters, we designed a spacer which was 3D printed and could be sterilized and reused. This spacer fits in an Ibidi 35 mm glass bottom µ-dish and allows the cells to be close to the glass bottom for FRAP microscopy.

FRAP data was collected using the FRAP module in LAS X software. For acquisition, laser output was kept constant at 4% and gain adjusted to achieve proper exposure in the low E-cadherin SGEF KD cells. Three frames every 30 s were collected pre bleach. Post bleach acquisition was performed in three different timeframes. The first 10 s were imaged every second (10 frames). The next 50 s were imaged every 5 s (10 frames). Finally, we continued imaging every 30 s until the experiment reached 15 min (25 frames). For bleaching, laser intensity was kept at 90% and multiple bleach events had to be performed for proper signal reduction; bleaching areas dimensions were kept constant.

FRAP data was then analyzed using RStudio and the Tidyverse library. At least 35 junctions per condition per experiment, from four independent experiments were analyzed. The data was then normalized using the min-max feature scaling, a mathematical regression calculated to get the immobile fraction and t_1/2 recovery and graphed using the ggplot library.

Technical replicates are defined in this study as individual junctions/cells (IF) or pipetting replicates (qPCR) whereas biological replicates are defined as the averages of technical replicates from independent experiments. Western blot data does not have technical replicates, only biological replicates. Error bars represent the s.e.m. unless otherwise indicated.

All the statistical analysis was carried out in RStudio. Here, processed data was tested for normal distribution and homogeneity of variance using Shapiro–Wilks and Levene tests, respectively. Means from data that confirmed these two assumptions were then tested using a one-way ANOVA with a Tukey post-hoc analysis. If data did not pass these two assumptions, a Kruskal–Wallis and a Wilcoxon post-hoc analysis were used. See Table S2 for statistical tests used and post-hoc P-values.

For qPCR, data was assumed to have normal distribution (Fay and Gerow, 2013), but this was not formally tested. P<0.05 was considered statistically significant. Statistical significance is denoted as *P<0.05; **P<0.005 and ***P<0.001.

The authors would like to thank Martijn Gloerich and Kai Simons for providing reagents. We would also like to thank the members of the Furuse lab and Yildirim-Ayan lab for helping with the FRAP setup. Lastly, we like to thank Andrei Ivanov, Stephan Huveneers, Andrew Goryachev and the members of the Garcia-Mata lab for critical comments and valuable discussions.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.262181.reviewer-comments.pdf