Erbin belongs to the LAP (leucine-rich repeat and PDZ domain) family of scaffolding proteins that plays important roles in orchestrating cell signaling. Here, we show that Erbin functions as a tumor suppressor in colorectal cancer. Analysis of Erbin expression in colorectal cancer patient specimens revealed that Erbin was downregulated at both mRNA and protein levels in tumor tissues. Knockdown of Erbin disrupted epithelial cell polarity and increased cell proliferation in 3D culture. In addition, silencing Erbin resulted in increased amplitude and duration of signaling through Akt and RAS/RAF pathways. Erbin loss induced epithelial–mesenchymal transition, which coincided with a significant increase in cell migration and invasion. Erbin interacted with kinase suppressor of Ras 1 (KSR1) and displaced it from the RAF/MEK/ERK complex to prevent signal propagation. Furthermore, genetic deletion of Erbin in Apc knockout mice promoted tumorigenesis and significantly reduced survival. Tumor organoids derived from Erbin/Apc double knockout mice displayed increased tumor initiation potential and activation of Wnt signaling. Results from gene set enrichment analysis revealed that Erbin expression associated positively with the E-cadherin adherens junction pathway and negatively with Wnt signaling in human colorectal cancer. Taken together, our study identifies Erbin as a negative regulator of tumor initiation and progression by suppressing Akt and RAS/RAF signaling in vivo.

Significance: These findings establish the scaffold protein Erbin as a negative regulator of EMT and tumorigenesis in colorectal cancer through direct suppression of Akt and RAS/RAF signaling. Cancer Res; 78(17); 4839–52. ©2018 AACR.

Erbin is a member of the leucine-rich repeat (LRR) and PDZ domain (LAP) protein superfamily. It contains multiple protein–protein interaction modules including 16 LRRs followed by a single PDZ domain at the C-terminus. Erbin is known to localize primarily to the adherens junctions and plays a role in maintaining the structural integrity of the junction in epithelial cells (1–3). The initial discovery of Erbin identifies Erbin as an ERBB2/Her2 receptor–interacting protein that facilitates the localization of the receptor to the basolateral membrane of epithelial cells (4). In addition, it has been shown that Erbin attenuates RAF activation by disrupting Shoc2-mediated RAS/RAF interaction (5, 6). Moreover, downregulation of Erbin results in resistance to anoikis in cervical cancer cells via activation of JAK/STAT signaling (7). However, the role of Erbin in regulating cell polarity and colorectal cancer tumorigenesis and progression remains elusive.

It has been well documented that epithelial cells, including those in the gastrointestinal tract, become polarized during the differentiation process (8). The polarization process, characterized by the formation of specialized junctions between neighboring cells, also results in the segregation of two plasma membrane domains: the apical surface, facing the external medium and the basolateral surface, connected to adjacent cells and extracellular matrix (9). The apical and basolateral membranes are segregated by two highly organized junctions, tight junctions, and adherens junctions; and the many proteins that compose these junctions are assembled at the site of cell–cell contacts (10–12). Previous studies have demonstrated that loss of cell polarity, through the disruption of these junctions, is associated with increased tumorigencity and accompanied by epithelial–mesenchymal transition (EMT; refs. 13–15). Therefore, the proper establishment of epithelial polarity allows cells to sense and to respond to signals that arise from the microenvironment in a spatiotemporally controlled manner.

By facilitating the assembly of tightly controlled signaling complexes through protein–protein interactions, scaffold proteins are known to play important roles in regulating spatiotemporal responses in cell signaling (16). One of the best known examples is the signal propagation in the RAS/RAF pathway, where the step-by-step activation process from RAS to ERK is facilitated by scaffolding proteins, such as KSR and Shoc2 (17, 18). KSR1, a kinase-like protein that lacks enzymatic activity, has been shown to promote cell proliferation and oncogenic potential by enhancing signaling activation through the RAS/RAF pathway (19). KSR1 binds constitutively to MEK in the cytoplasm in unstimulated cells and translocates to the plasma membrane upon RAS activation (20, 21). At the plasma membrane, KSR1 facilitates signaling propagation by organizing the formation of RAF/MEK/ERK complex (21, 22). Although the molecular mechanism by which scaffold proteins positively regulate RAS/RAF signaling has been extensively studied, it remains largely unknown whether scaffolding proteins are involved in signaling termination.

Here, we report the identification of Erbin as a novel tumor suppressor in colon cancer. We show that the mRNA and protein expression of Erbin is markedly decreased in colorectal cancer patient specimens. Erbin negatively regulates RAS/RAF signaling by sequestering KSR1 and preventing the formation of KSR1/RAF1 complex. Functionally, knockdown of Erbin results in an increase in cell motility by inducing EMT in colon cancer cells, and deletion of Erbin gene significantly decreases the lifespan of Apc-mutant mice and accelerates the tumor progression.

Mice

All animal procedures were performed by following protocols approved by the University of Kentucky Institutional Animal Care and Use Committee (IACUC). Erbin knockout (KO) mice on C57BL6 background as previously described (23) were maintained by random intercrossing to sustain a heterogeneous mixed genetic background. To produce animals used in the experiments, Erbin+/− mice were bred with Apcf/f and Villin-cre (Vil-cre) mice (both were obtained from The Jackson Laboratory) on a C57BL6 background. These mice were then intercrossed to produce three cohorts of animals, including Apcf/+/Erbin+/+/Vil-cre (Apc/WT), Apcf/+/Erbin+/−/Vil-cre (Apc/Het), and Apcf/+/Erbin−/−/Vil-cre (Apc/KO). To monitor survival, these cohorts of mice were followed for up to 6 months.

Cells and reagents

Caco2, HT29, and SW480 cells obtained from ATCC were cultured in DMEM supplemented with 10% FBS (Sigma-Aldrich) and 1% penicillin/streptomycin. LIM2405 cells obtained from Ludwig Cancer Research Institute (New York, NY) were cultured in RPMI1640 supplemented with 10% FBS, 2 mmol/L l-Glutamine, 25 mmol/L HEPES, and 1% penicillin/streptomycin. The mutation status of common colorectal cancer–associated genes is specified in Supplementary Table S1 (24). Human colon cancer cell lines were authenticated using short tandem repeat DNA profiling and tested negative for Mycoplasma using PCR in March 2016 (Genetica). Stable Erbin and KSR1 knockdown cells were generated by lentivirus-based shRNAs as described previously (25–27). The shRNA targeting sequences for human Erbin are as follows: 5′-GCAGCCAAGTACAACCGTTAA-3′ (A1) and 5′- CGGGCTCAAGTTGCATTTGAA-3′ (A2); and for mouse Erbin are: 5′-GTTGGATTCAAATCAGATAAA-3′ (B1) and 5′-CAGTCACTACTCTAGATTATT-3′ (B2). The shRNA targeting sequence for KSR1 is 5′-CAACAAGGAGTGGAATGATTT-3′. To generate wild-type (WT) and Erbin KO mouse embryonic fibroblasts (MEF) cells, Erbin heterozygous mice were used for breeding, and individual embryos at day 13 of gestation were isolated and genotyped. MEF cells were produced from WT and Erbin-null embryos by following standard protocols (28). The primary MEF cells were immortalized using retrovirus-mediated knockdown of p53 using pBabe-puro-shp53 (Addgene).

Transwell migration and invasion assay

Transwell migration and invasion assays were performed by following previously described procedures (25). Briefly, cells grown to 50% confluency were serum starved overnight. For migration assays, 5 × 104 cells were seeded into Transwells in 0.1% BSA and allowed to migrate for 4.5 hours at 37°C toward serum-free DMEM containing collagen (15 μg/mL) and EGF (10 ng/mL). For the invasion assay, cells were seeded into the upper chambers of Transwells that were precoated with Matrigel (BD Biosciences). The cells were allowed to migrate toward 5% FBS in the lower chamber for 24 hours. At the end of the incubation period, cells migrated to the bottom of Transwells were fixed and stained with crystal violet. The numbers of cells were counted using an inverted microscope at ×20 magnification.

Time-lapse live cell imaging and analysis

Control and Erbin knockdown cells were serum starved for 4 hours and plated as single cells onto collagen (15 μg/mL)-coated glass bottom culture dishes (MatTek). Cells were stimulated with EGF (10 ng/mL), and the trajectory of moving cells was captured using a Nikon BioStation IM equipped with a CO2 incubation chamber. Time-lapse phase images were taken every 10 minutes for 6 hours with a 20× objective (25, 29). The recorded movement of the cells was analyzed using Nikon Elements AR software.

Three-dimensional cell culture and immunofluorescence staining

The polarization process of Caco2 cells was examined by following procedures described previously (30). Briefly, single-cell suspension of Caco2 cells was cultured in 3D matrix consisting of 50% growth factor–reduced Matrigel (BD Biosciences) and 50% Collagen I (Thermo Fisher Scientific). The cells were allowed to grow into acinar structures in the 3D matrix for 10 days. The phase-contrast images of the acini were captured using a Nikon Eclipse Ti-E inverted microscope. For immunofluorescence staining, the acini were fixed in 4% paraformaldehyde in PBS and permeabilized with 0.5% Triton X-100 in PBS. Actin was stained using Alexa 488–conjugated phalloidin, while the nuclei of the cells were stained with DAPI-containing mounting medium. To detect proliferating cells, the cells grown in 3D culture were treated with 5-ethynyl-2′-deoxyuridine (EdU) for 1 hour prior to fixation. The EdU-positive cells were stained using Click-iT EdU Alexa Fluor 488 Imaging Kit (Thermo Fisher Scientific). Images were taken using an Olympus FlowView FV1000 confocal laser-scanning microscope (Olympus).

Western blot analysis

Colon cancer cells or mouse tumor organoids were harvested and detergent-solubilized cell lysates were obtained as described previously (25–27, 31). Equal amounts of cell lysates were resolved by SDS-PAGE and subjected to Western blot analysis. The Erbin antibody has been reported previously (3). The phospho-AKT (p-AKT for Ser473), pan-Akt, phospho-RAF1 (p-RAF1 for Ser338), total RAF1, phospho-MEK1/2 (p-MEK for Ser217/221), total MEK1/2, phospho-ERK1/2 (p-ERK for Thr202/Tyr204), total ERK1/2, and E-cadherin antibodies were from Cell Signaling Technology. The vimentin and N-cadherin antibodies were from BD Biosciences. The β-actin and γ-tubulin antibodies were from Sigma-Aldrich.

Histologic analysis and IHC staining

Mice were euthanized at indicated time points when showing signs of intestinal neoplasia, such as hunched stature and rectal bleeding. Intestine segments were opened longitudinally onto filter paper and made into “Swiss-roll” preparations as described previously (25). For histologic analysis, H&E sections were prepared from fixed and paraffin-embedded Swiss-roll specimens by following standard techniques. The colorectal cancer tissue microarray was created by the Biospecimen Procurement and Translational Pathology Shared Resource Facility of the Markey Cancer Center (Lexington, KY), which contains 45 pairs of tumor and adjacent normal control tissues collected from patients with colorectal cancer who had undergone surgery resections at the Markey Cancer Center. For IHC staining, tissue sections were deparaffinized, rehydrated, and treated with hydrogen peroxide. Antigen retrieval was performed using Dako Target Retrieval Solution (Dako Cytomation), and IHC staining was performed as described previously (26). The stained sections were visualized using a Nikon Eclipse 80i upright microscope.

Isolation and culture of mouse tumor organoids

Intestinal tumors were isolated from three cohorts of Apc/Erbin compound mutant mice, including Apc/WT, Apc/Het, and Apc/KO, cultured in 3D Matrigel as described previously (32, 33). Dissociated tumor cells were embedded in 33% Matrigel in 3D growth medium [Advanced DMEM/F12 supplemented with 1× GlutaMAX, 1 × N-2, 1 × B-27, 1 mmol/L N-acetylcysteine, EGF (50 ng/mL) and 1% penicillin/streptomycin]. Tumor organoids were allowed to grow for 5 days and collected for protein or RNA analysis. For colony formation assays, single-cell suspensions of 1,000 cells derived from tumor organoids were seeded into 3D Matrigel as described above. The number of colonies formed after 5 days were counted using an inverted microscope.

Tumor organoids derived from Apcf/+/KrasLSL-G12D/Villin-Cre mice were generated and described previously (33). To silence Erbin expression, tumor organoids were dissociated into small cell clusters using TrypLE (Thermo Fisher Scientific) and incubated with sh-Erbin lentivirus in suspension for 6 hours in a 37°C incubator. Cells were subsequently embedded in 33% Matrigel in 3D growth medium (Advanced DMEM/F12 supplemented with 1 × GlutaMAX, 1 × N-2, 1 × B-27, 1 mmol/L N-acetyl-L-cysteine and 1% penicillin/streptomycin), and puromycin was added 2 days later to select for stable knockdown cells. For colony formation assays, tumor organoids were dissociated and single-cell suspensions were seeded into 3D Matrigel. The number of tumor organoids formed after 6 days were counted and analyzed using an inverted microscope.

Real-time PCR

Total RNA was isolated from mouse tumor organoids using the RNeasy Mini Kit (Qiagen). Equal amounts of RNA were used as templates for the synthesis of cDNA using RT2 HT First Strand Kit (Qiagen). Real-time PCR was performed using mouse Lgr5-, Axin2- Cd44-, Ccnd1-, Ki67-, Alpi-, Fabp2-, and Muc2-specific probes using StepOne Real-Time PCR system (Applied Biosysems). All values were normalized to the level of β-actin. The overall expression of β-actin mRNA remained unchanged in different treatment groups as determined by the Ct (threshold cycle) values.

Bioinformatics and statistical analysis

In experiments to assess gene or protein expression, rate of cell migration, size and number of cell grown in 3D, and EdU incorporation were summarized using bar graphs, and pairwise comparisons between different conditions were carried out using two-sample t tests. To determine the relative expression of Erbin gene in human patients with colorectal cancer, microarray and patient clinical data from two colorectal cancer studies were downloaded from the Oncomine database. The Cancer Genome Atlas (TCGA) dataset contains 192 adenocarcinoma and 22 normal samples, within which, 13 were matched pairs. Expression of Erbin in tumor versus normal samples was compared using linear mixed models. Skrzypczak and colleagues' dataset contains 81 tumors and 24 normal samples. Expression of Erbin in tumor versus normal samples was compared on the basis of 2-sample t test. All statistical analyses were performed using R (version 3.4.1).

For gene set enrichment analysis (GSEA), RNA sequencing (RNA-seq) data were obtained from the TCGA colorectal cancer study. Correlations between expressions of ERBIN and the other genes were quantified by Spearman correlation coefficient. The genes were then ordered from highest to lowest based on the correlation coefficient. This ranked list was inputted into the GSEA Desktop Application (34) to identify pathways that are associated with ERBIN expression.

Erbin is downregulated in colorectal cancer patient tumor samples

To determine whether Erbin mRNA expression (gene symbol: ERBIN; previously known as ERBB2IP) is altered in patients with colorectal cancer, we performed bioinformatic analysis of two microarray datasets of human colorectal cancer samples. The microarray and patient clinical data of the two studies (35, 36) were downloaded from the Oncomine database. The mRNA expression of Erbin was significantly reduced in tumor samples when compared with normal controls (Fig. 1A and B). Interestingly, additional analysis of Erbin expression in stage I–IV of patients with colorectal cancer revealed that Erbin was downregulated upon tumor initiation when compared with normal controls, and no further loss of Erbin mRNA was observed as tumors progressed through stage II–IV, thus suggesting that loss of Erbin expression is an early event in tumorigenesis (Supplementary Fig. S1). In addition, the expression of Erbin protein was detected along the epithelial cell–cell junction in normal human colon tissues by IHC staining, whereas the expression of Erbin was markedly reduced and mislocalized to cytoplasm in tumor tissues (Fig. 1C). The basolateral distribution of Erbin in normal tissues was consistent with those observed previously (3, 37). Quantitative results obtained from IHC staining of a colorectal cancer tissue microarray revealed that basolateral membrane localization of Erbin was lost in all colorectal cancer tumor tissues examined (Fig. 1D; Supplementary Fig. S1). Furthermore, we analyzed Erbin protein expression in matched normal and tumor tissues obtained from 7 patients with colorectal cancer using Western blot analysis. Consistent with IHC staining results, Erbin protein levels were significantly decreased in tumor tissues compared with normal controls (Fig. 1E and F; Supplementary Fig. S1). Collectively, our data obtained in patient samples provided the first evidence, suggesting that Erbin may function as tumor suppressor in colorectal cancer.

Knockdown of Erbin increases both Akt and RAS/RAF signaling

Previous studies have implicated that Erbin negatively regulates ERK signaling (5, 38). To determine the function of Erbin in colorectal cancer, we silenced Erbin expression using two shRNA lentiviral targeting constructs (A1 and A2) in SW480, LIM2405, and Caco2 colon cancer cell lines. Consistently, knockdown of Erbin resulted in an increase in phosphorylation and activation of both the Akt and MEK/ERK pathways in all three cell lines (Fig. 2A). Because both Erbin shRNA targeting constructs had similar effects on silencing Erbin expression and enhancing activation of Akt and MEK/ERK phosphorylation, the subsequent experiments were mostly performed using sh-Erbin-A2 shRNA and key findings were confirmed with sh-Erbin-A1 shRNA. To further examine the effect of Erbin loss on the temporal activation of signaling, stable control and Erbin knockdown SW480 cells were starved for 16 hours and subsequently stimulated with EGF for the indicated time (Fig. 2B). Knockdown of Erbin increased both the amplitude and the duration of signaling through the Akt and RAF/MEK/ERK pathways (Fig. 2C). Similar results were obtained in EGF-treated LIM2045 cells (Supplementary Fig. S2).

Erbin is required to maintain cell polarity

We next investigated the functional effects of Erbin downregulation using a 3D cell culture system. Control and Erbin knockdown Caco2 and SW480 cells were seeded into 3D Matrigel and allowed to grow for 10 days to form tumor spheroids. As described in previous studies (30, 39, 40), control Caco2 cells were able to form acini-like spherical structures with a single hollow lumen, which consisted of a layer of polarized epithelial cells as indicated by the apical localization of F-actin (Fig. 3A). In marked contrast, Erbin depletion altered the acinar structure by inducing the formation of multiple lumens (Fig. 3A). Moreover, although not fully polarized, control SW480 cells were able to form tumor spheroids with partially hollowed lumens; however, cell clusters formed by Erbin knockdown SW480 cells lacked lumen structure and exhibited no apical or basolateral differentiation (Fig. 3A). Together, these results suggested that Erbin plays an important role in maintaining epithelial polarity.

Furthermore, we found that knockdown of Erbin markedly increased the size of the spheroids in both Caco2 and SW480 cells (Fig. 3B). To determine whether decreased Erbin expression alters cell proliferation in 3D culture, tumor spheroids formed by control and Erbin knockdown SW480 cells were labeled with EdU to mark proliferating cells (Fig. 3C). Quantitative results showed that loss of Erbin expression increased cell proliferation (Fig. 3D). To determine the molecular mechanism by which Erbin loss induces polarity defects and promotes cell proliferation in 3D culture, control and Erbin knockdown SW480 spheroids were collected and analyzed by Western blot analysis. Interestingly, the morphologic changes observed in sh-Erbin cells were accompanied by the downregulation of E-cadherin (an epithelial cell marker) and upregulation of vimentin and N-cadherin (markers of fibroblasts/mesenchymal cells), suggesting that Erbin knockdown cells had undergone EMT (Fig. 3E). Notably, increases in both Akt and ERK activation were maintained in Erbin knockdown spheroids grown in 3D (Fig. 3E). Collectively, these data suggested that downregulation of Erbin disrupts epithelial polarity and induces EMT.

Knockdown of Erbin promotes cell migration and invasion in colon cancer

As we have observed EMT-like phenotypes in Erbin knockdown colon cancer cells grown in 3D, we next determined the role of Erbin on regulating cell motility. The induction of EMT as indicated by downregulation of E-cadherin and upregulation of vimentin was confirmed in SW480 cells grown in regular 2D culture (Fig. 4A). To monitor single-cell motility, time-lapse images of control and Erbin knockdown SW480 and LIM2405 cells were captured and the distances traveled of individual cells were analyzed using Nikon BioStation. Results showed that Erbin knockdown cells were considerably more motile and average distances traveled by Erbin knockdown cells were significantly increased compared with the control cells (Fig. 4B and C), suggesting that Erbin loss increases cell motility at the single-cell level.

In addition, control Erbin knockdown SW480 and LIM2405 cells were subjected to Transwell migration and invasion assays. We found that both Erbin knockdown SW480 and LIM2405 cells migrated significantly faster than the control cells (Fig. 4D). Similarly, knockdown of Erbin in HT29 cells increased cell migration as determined using Transwell assays (Supplementary Fig. S2), and silencing Erbin using two different targeting shRNAs had comparable effects on promoting migration in SW480 cells (Supplementary Fig. S2). Furthermore, knockdown of Erbin significantly increased the ability of SW480 cells to invade through Matrigel (Fig. 4E), thus confirming that loss of Erbin promotes cell migration and invasion. Because Erbin downregulation activates both Akt and ERK signaling, we further determined the functional contribution of these two pathways in regulating the cell motility downstream of Erbin. Interestingly, the effect of Erbin loss on promoting cell migration was completely blocked by treating cells with MEK inhibitor (PD98059), whereas Akt inhibitor (MK2206) had no effect (Fig. 4F). The effect of MEK and Akt inhibitors on suppressing ERK and Akt activation was confirmed using Western blot analysis (Fig. 4G). Collectively, these results suggested that increased cell migration observed in Erbin knockdown cells is likely a result of MEK/ERK signaling activation.

Erbin inhibits RAF/MEK/ERK signaling by disrupting the RAF1–KSR1 interaction

In our effort to further define the molecular mechanism underlying Erbin-mediated inhibition of MEF/ERK signaling, we identified KSR1 as an interacting protein of Erbin. KSR1 is known to facilitate the formation of RAF/MEK/ERK complex upon RAS activation (20). We performed coimmunoprecipitation experiments to determine whether Erbin expression affects the interaction between KSR1 and RAF1. 293T cells transfected with Flag-RAF1 in the presence or absence of CFP-KSR1 and Myc-Erbin were immunoprecipitated with the anti-Flag antibody. The overexpressed and endogenous KSR1 were found to interact with RAF1 (Fig. 5A). However, coexpression of Erbin largely reduced the amount of KSR1interacting with RAF1 (Fig. 5A and B). In addition, the interaction between endogenous Erbin and KSR1 was readily detected in LIM2405 cells (Fig. 5C). Importantly, knockdown of Erbin resulted in an increase in formation of KSR1-RAF1 complex in both Caco2 and LIM2405 cells (Fig. 5D). Furthermore, we found that Erbin loss induced increase in ERK activation was abolished in cells where KSR1 was also silenced (Fig. 5E), thus confirming the functional interplay between Erbin and KSR1 in regulating RAF/MEK/ERK signaling. Collectively, these results suggest that Erbin functions to prevent KSR1 from forming a signaling complex with RAF1 and inhibit signaling activation downstream of RAF.

Genetic deletion of Erbin promotes tumorigenesis in Apc-mutant mouse model

To determine the effect of Erbin loss on tumorigenesis of colorectal cancer in vivo, we crossed Erbin-null mice with the Apcf/f/Vil-cre mouse model (41) to investigate the susceptibility of mice deficient in Erbin to Apc-driven intestinal adenomas. Erbin KO mouse models have been used in previous studies to investigate inflammatory responses, cardiac hypertrophy, and Her2/neu–mediated tumorigenesis in breast cancer (42–44). The Erbin KO mice did not develop spontaneous tumors. We compared the proliferation and differentiation of normal intestinal epithelial cells in WT and Erbin KO mice and found no difference in the number of proliferating cells and differentiated cells of different cell linages (Supplementary Fig. S3). Consistently, the expression of genes associated with normal intestinal stem and differentiated cells remained unchanged in intestinal organoids derived from Erbin heterozygous and homozygous KO mice compared with WT mice (Supplementary Fig. S4). However, consistent with results obtained in colorectal cancer cells, we found that the amplitude and duration of EGF-stimulated Akt and Raf/Mek/Erk signaling were largely increased in Erbin KO MEF cells (Supplementary Fig. S4). Taken together, these results suggest that Erbin loss–induced activation of Akt and Erk signaling alone is not sufficient to initiate tumorigenesis in colon cancer.

To specifically assess the role of Erbin in tumor initiation and progression in intestinal epithelium, we cross Erbin KO mice onto the Apc-mutant background to generate the following three cohorts of mice: Apcf/+/Vil-Cre/Erbin+/+, Apcf/+/Vil-Cre/Erbin+/−, and Apcf/+/Vil-Cre/Erbin−/− (Apc/WT, Apc/Het, and Apc/KO, respectively). The survival studies showed that knockout of a single allele of Erbin was sufficient to markedly accelerate the tumorigenesis process, resulting in a significantly shorter lifespan when compared with Apc/WT mice. Knockout of both alleles of Erbin further accelerated this process and significantly decreased survival (Fig. 6A). Histopathologic analysis revealed that adenomas were detected in both intestine and colon regions in all three cohorts of mice. However, when Apc/WT mice were sacrificed at time points that corresponded to average lifespan of Apc/KO mice (2 months), the difference in tumor burdens was particular clear in that more than half of Apc/KO mice reached the maximum tumor burden at 2 months when no tumors were detected in Apc/WT mice (Fig. 6B). The increased tumor burden was observed in Apc/Het mice sacrificed at 4.5 months (the average lifespan of this cohort) when compared with Apc/WT mice of same age, although to a less extent (Supplementary Fig. S5). At the terminal stage of tumor development (i.e., the mice were sacrificed due to tumor burden), the total numbers and size of tumors in all three cohorts were not significantly different. Thus, deletion of each Erbin allele significantly shortened the time needed to reach maximal tumor burden in a gene dosage-dependent manner.

To determine the effect of Erbin loss on signaling activation in vivo, intestinal tumor tissues from Apc/WT, Apc/Het, and Apc/KO mice were analyzed for the phosphorylation of Akt and Erk using IHC staining. Because the tumorigenesis time course for Apc/KO mice was drastically accelerated, in which no tumors were observed in Apc/WT mice at 2 months when majority of Apc/KO mice died (Fig. 6A and B), we collected tumor tissues from different cohorts of mice when they reached maximal tumor burden. Consistently, the levels of Akt and Erk phosphorylation were markedly increased in Apc/Het and Apc/KO tumors compared with Apc/WT (Fig. 6C). In addition, tumor cells isolated from Apc/WT and Apc/KO mice were allowed to grow into tumor organoids in 3D Matrigel. These organoids comprised of adenoma cells form cystic structure without budding (Fig. 6D). To determine whether Erbin inhibits signaling in mouse adenomas, protein lysates prepared from tumor organoids were subjected to Western blot analysis. Both Akt and Erk phosphorylation were markedly increased in Erbin KO organoids (Fig. 6E). Together, these data showed that Erbin loss promotes the activation of Akt and Erk signaling and tumorigenesis in mouse models of colorectal cancer.

Loss of Erbin increases the tumor initiation potential of Apc tumor organoids

We next analyzed the tumor initiation capacity of mouse tumor organoids using the colony formation assay. Apc/WT and Apc/KO organoids were dissociated into single cells and reseeded into Matrigel. The number of tumor organoids formed was counted after 5 days in culture. Knockout of Erbin resulted in a 2-fold increase in organoid formation (Fig. 7A), suggesting increased tumor initiation capacity. In addition, we determined the profile of gene expression using quantitative RT-PCR analysis in tumor organoids. Apc/WT and Apc/KO organoids grown in 3D culture for 5 days were collected and analyzed for the expression of Wnt target genes that are known to associate with cancer stem cells (e.g., Lgr5, Axin2, and Cd44), intestinal cell differentiation (e.g., Alpi, Fabp2, and Muc2), and cell proliferation (e.g., Ccnd1 and Ki67). Tumor organoids derived from Apc/KO mice expressed higher levels of genes associated with Wnt signaling and cell proliferation, which coincided with decreased expression of genes associated with differentiated intestinal epithelial cells (Fig. 7B). Similarly, the expression of genes associated with Wnt signaling and cell proliferation were significantly increased and intestinal differentiation decreased in tumor organoids derived from Apc/Het mice (Supplementary Fig. S5), confirming that haplodeficiency of Erbin is sufficient to promote tumorigenesis.

Intriguingly, Erbin loss did not result in EMT-like phenotypes as the expression of E-cadherin remained unchanged in Apc/KO tumor tissues and organoids compared with Apc/WT (data not shown). Because human colon cancer cells used in this study contain additional oncogenic alterations (such as KRAS or BRAF mutation), we performed additional experiments using Apc/KrasG12D double mutant mouse tumor organoids to determine the effect of Erbin downregulation. Indeed, silencing Erbin using two different lentiviral shRNAs in Apc/KrasG12D tumor organoids resulted in a decrease in E-cadherin expression and an increase in the phosphorylation of Akt and Erk (Fig. 7C). In addition, consistent with results obtained in Apc/KO organoids, the ability of Erbin knockdown Apc/KrasG12D cells to form tumor organoids in 3D was increased (Fig. 7D), suggesting that decreased expression of Erbin promotes tumor initiation potential in Apc/KrasG12D tumors as well.

Furthermore, we determined whether Erbin expression is associated with cancer-related biological pathways by analyzing gene expression data from TCGA colorectal cancer RNA-seq dataset. Results from the GSEA showed Erbin expression is positively associated with the E-cadherin adherens junction pathway and negatively associated with Wnt signaling (Fig. 7E). These data support our findings that loss of Erbin promotes the disruption of epithelial polarity by reducing E-cadherin expression and enhances tumor initiation potential by increasing signaling through the Wnt pathway. Taken together, our results showed that genetic deletion of Erbin potentiates tumor formation and progression in vivo.

Loss of epithelial polarity is a hallmark of advanced malignant tumors. Emerging evidence supports the notion that disruption of polarity promotes the malignant transformation of epithelial cells (14, 45, 46). In this study, combining in vitro and in vivo analyses, we identify Erbin as a tumor suppressor in colorectal cancer. The mRNA expression of Erbin is significantly downregulated in patients with colorectal cancer. Knockdown of Erbin in colon cancer cells results in disruption of epithelial polarity, increased cell motility, and cell proliferation. Mechanistically, Erbin inhibits the activation RAF/MEK/ERK signaling by sequestering KSR1 from forming a complex with RAF1. Finally, our in vivo studies reveal that Erbin loss accelerates tumor progression in Apc-mutant mouse models.

Previous studies have suggested that Erbin inhibits the activation of ERK by disrupting Shoc2-mediated RAS/RAF interaction (5, 6). However, Shoc2 is primarily localized to the endosome compartment of the cell (47). It remains an open question how Erbin, a basolateral membrane localized protein, interferes with Shoc2-dependent activation of RAS/RAF signaling at the endosome. In our study, we show that Erbin decreases RAF/MEK/ERK signaling through directly competing with KSR1. KSR1 is known to translocate to the plasma membrane upon RAS activation (20, 21). Results from our study and others demonstrate that Erbin is localized at the basolateral membrane. Being in close proximity with receptor tyrosine kinases (such as EGFR) and the site of RAS activation, the presence of Erbin may block the access of KSR1 to RAS-bound RAF and reduces KSR1–RAF interaction. It is interesting that Erbin downregulation promotes further activation of ERK signaling cascade in colorectal cancer cells that contain KRAS or BRAF mutations. Thus, by providing a spatial control of how signaling complexes are assembled, Erbin may serve as a negative scaffold to restrict signaling output of oncogenic pathways mediated by WT or mutant KRAS and BRAF. Although the increased cell motility is mainly associated with activation of MEK/ERK pathway in Erbin knockdown cells, the activation of both Akt and MEK/ERK signaling likely contributes to increased tumorigenesis in Erbin KO mice. It has been shown recently that oncogenic KRAS promotes Wnt signaling through ERK-mediated phosphorylation of LRP6 (48). However, treating Apc/KO tumor organoids with MEK or Akt inhibitor was unable to downregulate Wnt target gene expression in our study (data not shown). It is possible that loss of Erbin expression may alter the organization of epithelial junctions that allows the dissociation of β-catenin from the cell membrane. Future studies are required to determine the mechanism by which Erbin loss induces activation of Wnt signaling.

The role of Erbin in cancer has been controversial. Although a number of studies have shown that Erbin negatively regulates cell proliferation and survival in different types of cancer cells (7, 49), other studies indicate that Erbin loss increases tumorigenesis (44, 50). Results from our study have provided several lines of evidence supporting the tumor suppressor function of Erbin in colorectal cancer: (i) analysis of human colorectal cancer gene expression datasets with large sample sizes indicates that Erbin mRNA expression is significantly downregulated in patients with colorectal cancer; (ii) Erbin protein expression is decreased in colorectal cancer patient specimens by Western blot and IHC analyses; (iii) knockout of Erbin in Apc-mutant mice promotes tumor progression and reduces survival; and (iv) tumor organoids derived from Erbin KO mice have increased tumor initiation potential. Our findings are also corroborated by the bioinformatics analysis, in which Erbin expression is found to be downregulated at early stage of colorectal cancer and associated with increased E-cadherin junctions and decreased Wnt signaling. Additional gene sets that related to receptor tyrosine kinase signaling, cell-cycle control, and protein translation are identified to be associated with Erbin expression in patients with colorectal cancer in our GSEA study (Supplementary Table S2), suggesting that Erbin loss may have a broad impact on altering cell functions in cancer. Moreover, a number of missense, nonsense, and frame-shift mutations are identified throughout the entire coding region of Erbin based on an analysis of colorectal cancer datasets available at cBioPortal (Supplementary Fig. S6; ref. 51). The mutation rate of Erbin is between 2% and 4% among different studies. However, future studies are needed to determine the mechanism by which Erbin is downregulated in colorectal cancer and how Erbin loss induces polarity defect and EMT.

In summary, our study has uncovered a pivotal role of Erbin in maintaining epithelial cell polarity and suppressing EMT in colorectal cancer. By developing novel in vivo mouse models and tumor organoid systems, we demonstrate that Erbin exerts its tumor suppressor function by negatively regulating both Akt and RAF/MEK/ERK signaling and Erbin loss promotes tumor initiation and progression. The functional interplay between Erbin–KSR1 highlights the importance of scaffolding proteins in providing the spatiotemporal control of cell signals. Future studies on Erbin-dependent inhibition of tumor progression will help to explore the potential of using Erbin as a diagnostic marker for developing personalized treatment strategies in colorectal cancer.

No potential conflicts of interest were disclosed.

Conception and design: P.D. Stevens, T. Gao

Development of methodology: P.D. Stevens, X. Xiong, T. Gao

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.D. Stevens, Y.-A. Wen, X. Xiong, Y.Y. Zaytseva, A.T. Stevens, T.N. Farmer, T. Gan, S. Marchetto, J.-P. Borg, T. Gao

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.D. Stevens, A.T. Li, C. Wang, A.T. Stevens, T.N. Farmer, H.L. Weiss, T. Gao

Writing, review, and/or revision of the manuscript: P.D. Stevens, J.-P. Borg, T. Gao

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Xiong, T. Gan

Study supervision: M. Inagaki, T. Gao

We thank Dr. Emilia Galperin (University of Kentucky) for providing the expression construct of CFP-KSR1. This work was supported by R01CA133429 (T. Gao) and F31 CA196219 (P.D. Stevens). The studies were conducted with support provided by the Biospecimen Procurement and Translational Pathology and Biostatistics and Bioinformatics Shared Resource Facilities of the University of Kentucky Markey Cancer Center (P30CA177558). J.P, Borg's laboratory is funded by La Ligue Nationale Contre le Cancer (Label Ligue J.P. Borg), Fondation A*MIDEX, and SIRIC (INCa-DGOS-Inserm 6038).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Favre
B
,
Fontao
L
,
Koster
J
,
Shafaatian
R
,
Jaunin
F
,
Saurat
JH
, et al
The hemidesmosomal protein bullous pemphigoid antigen 1 and the integrin beta 4 subunit bind to ERBIN. Molecular cloning of multiple alternative splice variants of ERBIN and analysis of their tissue expression
.
J Biol Chem
2001
;
276
:
32427
36
.
2.
Jaulin-Bastard
F
,
Arsanto
JP
,
Le Bivic
A
,
Navarro
C
,
Vely
F
,
Saito
H
, et al
Interaction between Erbin and a Catenin-related protein in epithelial cells
.
J Biol Chem
2002
;
277
:
2869
75
.
3.
Izawa
I
,
Nishizawa
M
,
Tomono
Y
,
Ohtakara
K
,
Takahashi
T
,
Inagaki
M
. 
ERBIN associates with p0071, an armadillo protein, at cell-cell junctions of epithelial cells
.
Genes Cells
2002
;
7
:
475
85
.
4.
Borg
JP
,
Marchetto
S
,
Le Bivic
A
,
Ollendorff
V
,
Jaulin-Bastard
F
,
Saito
H
, et al
ERBIN: a basolateral PDZ protein that interacts with the mammalian ERBB2/HER2 receptor
.
Nat Cell Biol
2000
;
2
:
407
14
.
5.
Huang
YZ
,
Zang
M
,
Xiong
WC
,
Luo
Z
,
Mei
L
. 
Erbin suppresses the MAP kinase pathway
.
J Biol Chem
2003
;
278
:
1108
14
.
6.
Dai
P
,
Xiong
WC
,
Mei
L
. 
Erbin inhibits RAF activation by disrupting the sur-8-Ras-Raf complex
.
J Biol Chem
2006
;
281
:
927
33
.
7.
Hu
Y
,
Chen
H
,
Duan
C
,
Liu
D
,
Qian
L
,
Yang
Z
, et al
Deficiency of Erbin induces resistance of cervical cancer cells to anoikis in a STAT3-dependent manner
.
Oncogenesis
2013
;
2
:
e52
.
8.
Bissell
MJ
,
Radisky
D
. 
Putting tumours in context
.
Nat Rev Cancer
2001
;
1
:
46
54
.
9.
Assemat
E
,
Bazellieres
E
,
Pallesi-Pocachard
E
,
Le Bivic
A
,
Massey-Harroche
D
. 
Polarity complex proteins
.
Biochim Biophys Acta
2008
;
1778
:
614
30
.
10.
Kemler
R
. 
Classical cadherins
.
Semin Cell Biol
1992
;
3
:
149
55
.
11.
Fleming
TP
,
Papenbrock
T
,
Fesenko
I
,
Hausen
P
,
Sheth
B
. 
Assembly of tight junctions during early vertebrate development
.
Semin Cell Dev Biol
2000
;
11
:
291
9
.
12.
Baas
AF
,
Smit
L
,
Clevers
H
. 
LKB1 tumor suppressor protein: PARtaker in cell polarity
.
Trends Cell Biol
2004
;
14
:
312
9
.
13.
Huber
MA
,
Kraut
N
,
Beug
H
. 
Molecular requirements for epithelial-mesenchymal transition during tumor progression
.
Curr Opin Cell Biol
2005
;
17
:
548
58
.
14.
Martin-Belmonte
F
,
Perez-Moreno
M
. 
Epithelial cell polarity, stem cells and cancer
.
Nat Rev Cancer
2012
;
12
:
23
38
.
15.
Wodarz
A
,
Nathke
I
. 
Cell polarity in development and cancer
.
Nat Cell Biol
2007
;
9
:
1016
24
.
16.
Good
MC
,
Zalatan
JG
,
Lim
WA
. 
Scaffold proteins: hubs for controlling the flow of cellular information
.
Science
2011
;
332
:
680
6
.
17.
McKay
MM
,
Morrison
DK
. 
Integrating signals from RTKs to ERK/MAPK
.
Oncogene
2007
;
26
:
3113
21
.
18.
Wimmer
R
,
Baccarini
M
. 
Partner exchange: protein-protein interactions in the Raf pathway
.
Trends Biochem Sci
2010
;
35
:
660
8
.
19.
Kortum
RL
,
Lewis
RE
. 
The molecular scaffold KSR1 regulates the proliferative and oncogenic potential of cells
.
Mol Cell Biol
2004
;
24
:
4407
16
.
20.
Stewart
S
,
Sundaram
M
,
Zhang
Y
,
Lee
J
,
Han
M
,
Guan
KL
. 
Kinase suppressor of Ras forms a multiprotein signaling complex and modulates MEK localization
.
Mol Cell Biol
1999
;
19
:
5523
34
.
21.
Morrison
DK
. 
KSR: a MAPK scaffold of the Ras pathway?
J Cell Sci
2001
;
114
(
Pt 9
):
1609
12
.
22.
Michaud
NR
,
Therrien
M
,
Cacace
A
,
Edsall
LC
,
Spiegel
S
,
Rubin
GM
, et al
KSR stimulates Raf-1 activity in a kinase-independent manner
.
Proc Natl Acad Sci U S A
1997
;
94
:
12792
6
.
23.
Tao
Y
,
Dai
P
,
Liu
Y
,
Marchetto
S
,
Xiong
WC
,
Borg
JP
, et al
Erbin regulates NRG1 signaling and myelination
.
Proc Natl Acad Sci U S A
2009
;
106
:
9477
82
.
24.
Mouradov
D
,
Sloggett
C
,
Jorissen
RN
,
Love
CG
,
Li
S
,
Burgess
AW
, et al
Colorectal cancer cell lines are representative models of the main molecular subtypes of primary cancer
.
Cancer Res
2014
;
74
:
3238
47
.
25.
Li
X
,
Stevens
PD
,
Liu
J
,
Yang
H
,
Wang
W
,
Wang
C
, et al
PHLPP is a negative regulator of RAF1, which reduces colorectal cancer cell motility and prevents tumor progression in mice
.
Gastroenterology
2014
;
146
:
1301
12
.
26.
Liu
J
,
Weiss
HL
,
Rychahou
P
,
Jackson
LN
,
Evers
BM
,
Gao
T
. 
Loss of PHLPP expression in colon cancer: role in proliferation and tumorigenesis
.
Oncogene
2009
;
28
:
994
1004
.
27.
Wen
YA
,
Stevens
PD
,
Gasser
ML
,
Andrei
R
,
Gao
T
. 
Downregulation of PHLPP expression contributes to hypoxia-induced resistance to chemotherapy in colon cancer cells
.
Mol Cell Biol
2013
;
33
:
4594
605
.
28.
Liu
J
,
Stevens
PD
,
Li
X
,
Schmidt
MD
,
Gao
T
. 
PHLPP-mediated dephosphorylation of S6K1 inhibits protein translation and cell growth
.
Mol Cell Biol
2011
;
31
:
4917
27
.
29.
Larson
Y
,
Liu
J
,
Stevens
PD
,
Li
X
,
Li
J
,
Evers
BM
, et al
Tuberous sclerosis complex 2 (TSC2) regulates cell migration and polarity through activation of CDC42 and RAC1
.
J Biol Chem
2010
;
285
:
24987
98
.
30.
Xiong
X
,
Li
X
,
Wen
YA
,
Gao
T
. 
Pleckstrin homology (PH) domain leucine-rich repeat protein phosphatase controls cell polarity by negatively regulating the activity of atypical protein kinase C
.
J Biol Chem
2016
;
291
:
25167
78
.
31.
Wen
YA
,
Li
X
,
Goretsky
T
,
Weiss
HL
,
Barrett
TA
,
Gao
T
. 
Loss of PHLPP protects against colitis by inhibiting intestinal epithelial cell apoptosis
.
Biochim Biophys Acta
2015
;
1852
(
10 Pt A
):
2013
23
.
32.
Sato
T
,
Stange
DE
,
Ferrante
M
,
Vries
RG
,
Van Es
JH
,
Van den Brink
S
, et al
Long-term expansion of epithelial organoids from human colon, adenoma, adenocarcinoma, and Barrett's epithelium
.
Gastroenterology
2011
;
141
:
1762
72
.
33.
Wen
YA
,
Xing
X
,
Harris
JW
,
Zaytseva
YY
,
Mitov
MI
,
Napier
DL
, et al
Adipocytes activate mitochondrial fatty acid oxidation and autophagy to promote tumor growth in colon cancer
.
Cell Death Dis
2017
;
8
:
e2593
.
34.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
35.
Skrzypczak
M
,
Goryca
K
,
Rubel
T
,
Paziewska
A
,
Mikula
M
,
Jarosz
D
, et al
Modeling oncogenic signaling in colon tumors by multidirectional analyses of microarray data directed for maximization of analytical reliability
.
PLoS One
2010
;
5
:
pii
:
e13091
.
36.
NCI
. 
The Cancer Genome Atlas
.
Available from
: https://cancergenome.nih.gov/.
37.
Legouis
R
,
Jaulin-Bastard
F
,
Schott
S
,
Navarro
C
,
Borg
JP
,
Labouesse
M
. 
Basolateral targeting by leucine-rich repeat domains in epithelial cells
.
EMBO Rep
2003
;
4
:
1096
102
.
38.
Rangwala
R
,
Banine
F
,
Borg
JP
,
Sherman
LS
. 
Erbin regulates mitogen-activated protein (MAP) kinase activation and MAP kinase-dependent interactions between Merlin and adherens junction protein complexes in Schwann cells
.
J Biol Chem
2005
;
280
:
11790
7
.
39.
Jaffe
AB
,
Kaji
N
,
Durgan
J
,
Hall
A
. 
Cdc42 controls spindle orientation to position the apical surface during epithelial morphogenesis
.
J Cell Biol
2008
;
183
:
625
33
.
40.
Magudia
K
,
Lahoz
A
,
Hall
A
. 
K-Ras and B-Raf oncogenes inhibit colon epithelial polarity establishment through up-regulation of c-myc
.
J Cell Biol
2012
;
198
:
185
94
.
41.
Cheung
AF
,
Carter
AM
,
Kostova
KK
,
Woodruff
JF
,
Crowley
D
,
Bronson
RT
, et al
Complete deletion of Apc results in severe polyposis in mice
.
Oncogene
2010
;
29
:
1857
64
.
42.
McDonald
C
,
Chen
FF
,
Ollendorff
V
,
Ogura
Y
,
Marchetto
S
,
Lecine
P
, et al
A role for Erbin in the regulation of Nod2-dependent NF-kappaB signaling
.
J Biol Chem
2005
;
280
:
40301
9
.
43.
Rachmin
I
,
Tshori
S
,
Smith
Y
,
Oppenheim
A
,
Marchetto
S
,
Kay
G
, et al
Erbin is a negative modulator of cardiac hypertrophy
.
Proc Natl Acad Sci U S A
2014
;
111
:
5902
7
.
44.
Tao
Y
,
Shen
C
,
Luo
S
,
Traore
W
,
Marchetto
S
,
Santoni
MJ
, et al
Role of Erbin in ErbB2-dependent breast tumor growth
.
Proc Natl Acad Sci U S A
2014
;
111
:
E4429
38
.
45.
Huang
L
,
Muthuswamy
SK
. 
Polarity protein alterations in carcinoma: a focus on emerging roles for polarity regulators
.
Curr Opin Genet Dev
2010
;
20
:
41
50
.
46.
Royer
C
,
Lu
X
. 
Epithelial cell polarity: a major gatekeeper against cancer?
Cell Death Differ
2011
;
18
:
1470
7
.
47.
Jang
ER
,
Jang
H
,
Shi
P
,
Popa
G
,
Jeoung
M
,
Galperin
E
. 
Spatial control of Shoc2-scaffold-mediated ERK1/2 signaling requires remodeling activity of the ATPase PSMC5
.
J Cell Sci
2015
;
128
:
4428
41
.
48.
Lemieux
E
,
Cagnol
S
,
Beaudry
K
,
Carrier
J
,
Rivard
N
. 
Oncogenic KRAS signalling promotes the Wnt/beta-catenin pathway through LRP6 in colorectal cancer
.
Oncogene
2015
;
34
:
4914
27
.
49.
Liu
D
,
Shi
M
,
Duan
C
,
Chen
H
,
Hu
Y
,
Yang
Z
, et al
Downregulation of Erbin in Her2-overexpressing breast cancer cells promotes cell migration and induces trastuzumab resistance
.
Mol Immunol
2013
;
56
:
104
12
.
50.
Yao
S
,
Zheng
P
,
Wu
H
,
Song
LM
,
Ying
XF
,
Xing
C
, et al
Erbin interacts with c-Cbl and promotes tumourigenesis and tumour growth in colorectal cancer by preventing c-Cbl-mediated ubiquitination and down-regulation of EGFR
.
J Pathol
2015
;
236
:
65
77
.
51.
Gao
J
,
Aksoy
BA
,
Dogrusoz
U
,
Dresdner
G
,
Gross
B
,
Sumer
SO
, et al
Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal
.
Sci Signal
2013
;
6
:
pl1
.

Supplementary data