No special-type breast cancer [NST; commonly known as invasive ductal carcinoma (IDC)] and invasive lobular carcinoma (ILC) are the two major histological subtypes of breast cancer with significant differences in clinicopathological and molecular characteristics. The defining pathognomonic feature of ILC is loss of cellular adhesion protein, E-cadherin (CDH1). We have previously shown that E-cadherin functions as a negative regulator of the IGF1R and propose that E-cadherin loss in ILC sensitizes cells to growth factor signaling that thus alters their sensitivity to growth factor–signaling inhibitors and their downstream activators. To investigate this potential therapeutic vulnerability, we generated CRISPR-mediated CDH1 knockout (CDH1 KO) IDC cell lines (MCF7, T47D, and ZR75.1) to uncover the mechanism by which loss of E-cadherin results in IGF pathway activation. CDH1 KO cells demonstrated enhanced invasion and migration that was further elevated in response to IGF1, serum and collagen I. CDH1 KO cells exhibited increased sensitivity to IGF resulting in elevated downstream signaling. Despite minimal differences in membranous IGF1R levels between wild-type (WT) and CDH1 KO cells, significantly higher ligand–receptor interaction was observed in the CDH1 KO cells, potentially conferring enhanced downstream signaling activation. Critically, increased sensitivity to IGF1R, PI3K, Akt, and MEK inhibitors was observed in CDH1 KO cells and ILC patient-derived organoids.

Implications:

Overall, this suggests that these targets require further exploration in ILC treatment and that CDH1 loss may be exploited as a biomarker of response for patient stratification.

Despite significant diagnostic and therapeutic advancements, breast cancer remains a leading cause of malignancy-associated mortality for women worldwide. Breast cancer is a heterogenous disease that can be categorized on the basis of its histopathology and molecular features. The major histological subtypes of breast cancer are No Special-Type (NST) that is often referred to as invasive ductal carcinoma (IDC: ∼80% of all cases) and invasive lobular carcinoma (ILC: ∼10%–15% of all cases; refs. 1–3). Although IDC and ILC are clinically managed as the same disease, there is an increasing appreciation of their differences in terms of molecular alterations, pathology, and prognosis (4–7). The diminished prevalence of ILC consequently has resulted in this subtype being historically understudied, and therefore implores the need to better characterize the disease to establish precision medicine approaches to treatment (4, 8). ILC tumors are largely characterized by their growth in a single file, discohesive manner with presence of stromal infiltration (5–7, 9). Approximately 95% of ILC cases present with loss of CDH1, with the majority due to genomic alterations such as loss of heterozygosity and/or truncating frameshift mutations (6, 10, 11). In addition, other molecular alterations such as PIK3CA activation, PTEN loss, and mutations in TBX3 and FOXA1 are detected at higher levels in ILC tumors compared with IDC tumors (4, 12). Enhanced Akt signaling has also been observed in several ILC cell line and mouse models (13).

As one of the activators of Akt, the IGF1 pathway has long been associated with breast cancer progression (14–18). IGF1 is an essential component in cellular proliferation and survival (15, 19). Binding of IGF1 to its receptor (IGF1R) or the insulin receptor (IR) stimulates autophosphorylation and activation of the PI3K and MAPK pathways among others (15, 18, 20). Because of high levels of circulating IGF1 being associated with poor prognosis in breast cancer (21), it is imperative to understand the dysregulation of the IGF1 pathway in tumors. Numerous IGF1R inhibitors have been evaluated in preclinical and clinical settings, including humanized monoclonal antibodies that downregulate the receptor and small-molecule tyrosine kinase inhibitors that inhibit pathway activation (20, 22–26). Unfortunately, these trials demonstrated little to no response in the recruited broad patient populations despite promising preclinical efficacy (13, 27–31), suggesting the necessity to better stratify patients to ensure efficient validation of therapies while maximizing response metrics.

To that end, we recently reported E-cadherin as a modulator of IGF1 signaling and a potential biomarker of inhibitor response (32). Briefly, transient E-cadherin knockdown sensitized cell lines to IGF1 signaling and to IGF1R inhibition (32, 33), and we further reported that ILC tumors have higher IGF1/2 expression and pIGF1R/IR activation compared with IDC tumors (13, 33). In this present study, we have expanded upon our earlier observations by generating CRISPR-mediated CDH1 knockout (CDH1 KO) IDC cell lines to elucidate the role of E-cadherin individually in more detail. Here, we report that loss of E-cadherin renders cells sensitive to IGF1, IGF2 and insulin signaling by increasing IGF1R receptor availability for ligand binding resulting in enhanced cell migration/invasion and contextual increases in sensitivity to IGF1R, PI3K, Akt, and MEK inhibitors. Given previous evidence for E-cadherin regulation of other growth factor pathways (13, 30), we also extended our analyses to additional growth factor signaling pathways. Our findings provide insights into better understanding the mechanics of E-cadherin function as well as the potential for a targeted therapeutic intervention in ILC patient populations.

Cell culture

Cell lines used in this study were obtained from the ATCC: MCF7 (RRID: CVCL_0031), T47D (RRID: CVCL_0553), ZR75.1 (RRID: CVCL_0588), MDA-MB-134-VI (RRID: CVCL_0617), MDA-MB-231 (RRID: CVCL_0062) and Asterand for SUM44PE (RRID: CVCL_3424). Cell lines were maintained in 10% FBS (Life Technologies) supplemented media (Thermo Fisher Scientific): MDA-MB-134 in 1:1 DMEM: L-15; MCF7 and MDA-MB-231 in DMEM; and T47D and ZR75.1 in RPMI. SUM44PE was maintained in DMEM/F12 with 2% charcoal stripped serum (CSS; Life Technologies) with additional supplements as previously described (5). Cell lines were cultured for less than 6 months at a time, routinely tested to be Mycoplasma free and authenticated by the University of Arizona Genetics Core (Tucson, Arizona) by short tandem repeat DNA profiling.

CDH1 knockout cell line generation

CRISPR-mediated knockout of CDH1 in MCF7 and T47D cells was performed by using the Gene Knockout Kit (V1) from Synthego as previously described (31, 34). The parental cell line was used a comparator for MCF7 and T47D, referred as wild-type (WT). For ZR75.1, a doxycycline-inducible Cas9 was used harboring either short guide RNA (sgRNA) for CDH1 or a non-targeting control (NTC) sequence and was introduced via an adenoviral vector. For ZR75.1, a total of 3 rounds of infection and Puromycin selection were performed before single cell sorting. 8 clones from each KO and NTC cells were isolated by single cell cloning and combined to generate a pool for subsequent experiments.

Immunoblotting

Protein for immunoblotting experiments were harvested and analyzed as described previously (33). Membrane blocking was performed with Intercept PBS blocking buffer (LiCOr #927–40000) for one hour at room temperature and probed with primary antibodies overnight at 4°C: pIGF1R/IR (Cell Signaling Technology #3024; RRID:AB_331253), IGF1R (Cell Signaling Technology #3027; RRID:AB_2122378), pAkt S473 (Cell Signaling Technology #4060; RRID:AB_2315049), Akt (Cell Signaling Technology #9272; RRID:AB_329827), InsR (Cell Signaling Technology #3025; RRID:AB_2280448), E-cadherin (BD Biosciences #610182; RRID:AB_397581), β-catenin (BD Biosciences #610154; RRID:AB_397555), p120 catenin (BD Biosciences; #610134; RRID:AB_397537), pEGFR Y1068 (Cell Signaling Technology #2234; RRID:AB_331701), EGFR (Cell Signaling Technology #4267; RRID:AB_2246311), α-Biotin (Cell Signaling Technology #5597; RRID:AB_10828011), non-phospho β-catenin (Cell Signaling Technology #1980; RRID:AB_2650576), pFGFR4 Y642 (Signalway #11836), FGFR4 (Cell Signaling Technology #8562; RRID:AB_10891199), pFRS2 Y196 (Cell Signaling Technology #3864; RRID:AB_2106222), pSTAT3 Y705 (Cell Signaling Technology #9131; RRID:AB_331586), and p-p44/42 MAPK (Cell Signaling Technology #4377; RRID:AB_331775). This was followed by 1 hour room temperature incubation with secondary antibodies (1:10,000; anti-mouse 680LT: LiCor #925–68020; anti-rabbit 800CW: LiCor #925–32211). Membranes were subsequently imaged on the LiCOr Odyssey CLx Imaging system, with band quantifications performed with built in software.

Immunofluorescence

Cells were plated at a density of 100,000–200,000 cells/well on glass coverslips (Fisher #12–545–80P) in 24-well plates, fixed on ice in ice-cold methanol for 30 minutes and blocked in blocking buffer (0.3% Triton X-100, 5% BSA, 1X DPBS) for 1 hour at room temperature. Primary antibody incubation was performed overnight at 4°C: IGF1R β-subunit (Cell Signaling Technology #3027; RRID: AB_2122378; 1:100), EpCAM (Cell Signaling Technology; #2929; RRID: AB_2098657; 1:100), E-cadherin (Cell Signaling Technology; #3195; RRID: AB_2291471; 1:100) and p120 catenin (BD Biosciences; #610134; RRID: AB_397537; 1:100). Secondary antibody incubation was done for 1 hour at room temperature followed by Hoechst 33342 (Thermo Fisher Scientific #62249; 1:10,000) staining. Coverslips were mounted with Aqua-Poly/Mount (Polysciences #18606–20) and images were taken on a Nikon A1 confocal microscope with a ×60 objective.

Immunofluorescence of organoids

Organoids were harvested with Cultrex Organoid Harvesting Solution (R&D Systems #3700–100–01) in an Eppendorf tube for 1 hour at 4°C. Centrifuged organoid pellets were fixed with 4% PFA followed by an antigen retrieval step. Pellets were blocked overnight at 4°C, incubated with E-cadherin antibody (BD Biosciences 610182, RRID: AB_397581) overnight at 4°C, followed by secondary antibody incubation also overnight at 4°C. Organoids were then stained with Hoechst for 30 minutes, cleared with methanol for 1 hour and mounted with ProLong Gold Antifade Mountant (Invitrogen P36930). Images were taken on a Nikon A1 confocal microscope with a ×60 objective.

Colony formation assay

Cells were plated at a density of 2,000 cells/well in 6-well plates (Thermo Fisher Scientific #08–772–1B) in either full serum (10% FBS) or low serum (0.5% FBS) with 5 nmol/L IGF1 (GroPep Bioreagents #AQU001) supplemented media. Cells were monitored every few days and media refreshed every 4 days. Cells plated in full serum media were fixed with 100% methanol on ice and stained with 0.5% Crystal Violet (Sigma-Aldrich #C0775) in 40% methanol after 14 days whereas cells grown in low serum, 5 nmol/L IGF1, were stained after 21–28 days. Wells were imaged on an Olympus SZX16 dissecting microscope and de-stained with 10% acetic acid in water and quantified by spectrophotometry at 560nm. Statistical differences were tested using two-way ANOVA.

Co-Immunoprecipitation

Co-Immunoprecipitation (Co-IP) was performed with E-cadherin (BD Biosciences #610182; RRID: AB 397581) and IGF1R (Cell Signaling Technology #3027; RRID: AB_2122378) antibodies. Cells were lysed in 20 mmol/L Tris-HCl pH 7.4 with 1% NP-40, 137 mmol/L NaCl and 5 mmol/L EDTA with fresh protease and phosphatase inhibitors (1:100) and quantified using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific #23225). 1-mg protein from each sample was pre-cleared in 20 μL of Pierce Protein G Agarose beads (Thermo Fisher Scientific #20398) and incubated in either 3 μg of E-cadherin, IGF1R or IgG antibodies (Normal mouse IgG; Millipore #12–371; RRID: AB_145840 and Normal rabbit IgG; Millipore #12–370; RRID: AB_145841) overnight at 4°C with rotation. One 4-hour incubation was performed the next day with 45 μL of Pierce Protein G Agarose beads at 4°C with rotation. Protein was eluted with Laemmli buffer and analyzed by immunoblotting.

qRT-PCR

RNA extraction was performed using the RNeasy mini kit (Qiagen #74106) and the RNA quality and amount quantified on NanoDrop. Reverse transcription to cDNA was performed with PrimeScript RT Master Mix (Takara Bio #RR036B). RT-PCR was then performed with SsoAdvanced Universal SYBR (Bio-Rad #1726275) with primers as detailed in Supplementary Materials. Results were normalized to reads from housekeeping gene RPLPO. Statistical differences evaluated using a paired t test.

FACS for anoikis resistance

Cells were stained with APC-Annexin V (BD Biosciences #550474) and Propidium Iodide (BD Biosciences #556463) in 1X Annexin binding buffer (BD Biosciences #556454) for 15 minutes at room temperature. Samples were analyzed on an LSR II Flow cytometer (BD Biosciences) and processed using BD FACSDiva and FlowJo software (BD Biosciences) as previously described (31). Live cell percentages in 2D conditions for each cell line was used to normalize the live cell percentages in ULA conditions. Statistical differences were tested using a two-way ANOVA.

Haptotaxis, migration, and invasion assays

For haptotaxis experiments, the QCM Haptotaxis Cell Migration Assay Collagen I (EMD Millipore #ECM582) kit was used as previously described (8). For Transwell migration assays, transparent 24-well PET membranes of 8 μm pore size (Thermo Fisher Scientific # 08–771–21) were used. For Collagen I invasion assays, QCM Collagen Cell Invasion Assay (EMD Millipore #ECM551) was used according to the manufacturer's protocol. For the latter two assays, cells were plated at a density of 300,000 cells/well in 300 μL 0.5% FBS media in the top chamber; all bottom chambers were filled with 0.5% FBS media ± 5 nmol/L IGF1 (GroPep Bioreagents # AQU100) or full serum (10%) media. Cells were incubated at 37°C for 72 hours. Excess cells were removed from the top chambers using cotton swabs and inserts were stained with Crystal Violet (Sigma-Aldrich #C0775) before being imaged on an Olympus SZX16 dissecting microscope and quantified with ImageJ software. Quantifications were normalized to low serum WT samples and P values calculated with one-way ANOVA.

Receptor availability assay

Cells were seeded in 6-cm plates (Thermo Fisher Scientific #08–772-E), and serum starved overnight after achieving a 70%–80% confluency. Cells were then stimulated with biotinylated IGF1 (GroPep #AQU100) for 15 minutes at 4°C to reduce receptor internalization. Following PBS washes on ice, cells were treated with 2 mmol/L BS3 crosslinker (Thermo Fisher Scientific #21580) reconstituted in PBS (pH 8.0) with 6 mmol/L KCl and 10 mmol/L EGTA for 1 hour at 4°C with occasional rocking. BS3 quenching was performed with 10 mmol/L Glycine for 15 minutes and cells were harvested for subsequent immunoblotting (30).

Dose–response assays

Cells were plated in 50 μL of media at 9,000 cells/well in 2D and ULA (Corning #3474) 96-well plates. Treatments were added 24 hours post seeding in an additional 50 μL of respective media. IGF1R inhibitors BMS-754807 (Selleckchem #S1124) and OSI-906 (Selleckchem #S1091), PI3K inhibitor Alpelisib (Selleckchem #S2814), Akt inhibitor MK-2206 (Selleckchem #S1078), MEK inhibitor U0126 (Selleckchem #S1102), and Fulvestrant (Selleckchem #S1191) were dissolved in DMSO with a final ≤0.5% DMSO concentration in treatments. Plates were collected at day 6 and measured by CellTiter-Glo (Promega #PR-G7573) following the manufacturer's protocol. Cell viability values were analyzed following blank cell deductions and normalization to vehicle readings. IC50 values for viability were calculated by nonlinear regression and statistical differences evaluated using sum-of-squares Global f-test (P < 0.05). Synergy was assessed using SynergyFinder (35).

In silico analysis

Gene expression data from the Sweden Cancerome Analysis Network–Breast (SCAN‐B) study were downloaded from Gene Expression Omnibus, accession GSE96058 (36). Differential gene expression between luminal A lobular (N = 265) and ductal (N = 1165) breast cancer samples was assessed using DESeq2 (37). Tumor purity for SCAN-B samples was estimated using the R package ESTIMATE (37). An FDR cutoff value of 0.05 was used to identify significantly differentially expressed genes. Heatmap for genes of interest was created using the R package ComplexHeatmap. Gene set variation analysis (GSVA) was performed on log transformed gene FPKM matrices using the R package GSVA (38) with default parameter “Gaussian” for kernel selection. Gene sets of interest were obtained from MSigDB version 7.4. The GSVA enrichment scores were compared between luminal A lobular and ductal breast cancer samples using the Mann–Whitney U test. Gene set enrichment analysis (GSEA version 4.1.0. Broad Institute; ref. 39) was also conducted on normalized raw counts with default parameters and customized gene sets of interest obtained from MSigDB.

Organoid culture

IDC organoids (IPM-BO-56, 75, 102, 109, 126 and 134) and ILC organoids (IPM-BO-30 and 46) were established by the Institute for Precision Medicine and maintained in media as detailed in ref. (40) with 1 nmol/L β-estradiol (Sigma-Aldrich #E8875) supplementation. For dose–response assays, organoids were dissociated to single cell suspension with Trypsin and plated in 50 μL of organoid media at 3,000 or 5,000 cells/well in 96-well round bottom plates (Falcon #353227) depending on the growth rate. Akt inhibitor MK2206 (Selleckchem #S1078; dissolved in DMSO) treatments were added 24 hours post seeding in an additional 50 μL of organoid media. Organoids were monitored every day to ensure that vehicle-treated wells were growing well and remained at an appropriate density. Media were refreshed on day 6 and plates collected on day 12 with cell viability quantified by CellTiter-Glo 3D (Promega #G9681). Dose–response assay analysis was performed as described above.

Data availability

All data generated in this study are available within the article and its Supplementary Files. SCAN-B data analyzed in this study were obtained from Gene Expression Omnibus (GEO) at GSE96058.

IGF pathway activity is enhanced in ILC patient samples and cell lines in vitro

Analysis of the publicly available Sweden Cancerome Analysis Network–Breast (SCAN‐B) study allowed a direct comparison of growth factor signaling differences between luminal A IDC (n = 1165) and luminal A ILC (n = 265) tumors. Significantly higher IGF1 and IGF2 expressions were observed in ILC tumors in addition to several other growth factor signaling related genes (Fig. 1A) such as IGFBPs and matricellular proteins such as CTGF, NOV, and CYR61. Several genes downregulated in ILC were also observed, including IGFL1, IGF1R, and WISP1. PI3K/Akt signaling was significantly higher in ILC by GSEA (Fig. 1B) and GSVA (Fig. 1C). IGF1/2 signaling activation was also higher in ILC (Fig. 1D), strengthening our hypothesis regarding IGF pathway activation in ILC over IDC tumors.

Figure 1.

IGF pathway activity is enhanced in ILC patient samples and cell lines in vitro.A, Gene expression analysis of growth factor signaling related genes comparing Luminal A IDC and ILC with significantly different fold change and FDR values displayed. B, GSEA analysis for PI3K/Akt showing significantly higher pathway activation in ILC samples obtained from the SCANB dataset. C and D, GSVA analysis of Luminal A IDC and ILC samples using respective signature sets from MSigDB with significantly higher pathway activation in ILC. E, IDC cells (MCF7, T47D, and ZR75.1) and ILC cells (MDA-MB-134-VI, SUM44PE) were stimulated with doses of IGF1 or IGF2 (0–5 nmol/L) for 15 minutes following an overnight serum starvation. IGF1R/IR and Akt signaling was assessed by Western blotting. Phosphorylation levels of IGF1R [phospho-IGF-I Receptor β (Tyr1135/1136)/Insulin Receptor β (Tyr1150/1151)] and Akt (S473) were quantified on the LiCOr Odyssey CLx Imaging system and normalized to corresponding total protein levels and loading controls. Quantified values are indicated below the Western bands. Ligand treated sample values were further normalized to respective cell line vehicle-treated samples. Representative experiment shown for all, N = 2–3 for each experiment.

Figure 1.

IGF pathway activity is enhanced in ILC patient samples and cell lines in vitro.A, Gene expression analysis of growth factor signaling related genes comparing Luminal A IDC and ILC with significantly different fold change and FDR values displayed. B, GSEA analysis for PI3K/Akt showing significantly higher pathway activation in ILC samples obtained from the SCANB dataset. C and D, GSVA analysis of Luminal A IDC and ILC samples using respective signature sets from MSigDB with significantly higher pathway activation in ILC. E, IDC cells (MCF7, T47D, and ZR75.1) and ILC cells (MDA-MB-134-VI, SUM44PE) were stimulated with doses of IGF1 or IGF2 (0–5 nmol/L) for 15 minutes following an overnight serum starvation. IGF1R/IR and Akt signaling was assessed by Western blotting. Phosphorylation levels of IGF1R [phospho-IGF-I Receptor β (Tyr1135/1136)/Insulin Receptor β (Tyr1150/1151)] and Akt (S473) were quantified on the LiCOr Odyssey CLx Imaging system and normalized to corresponding total protein levels and loading controls. Quantified values are indicated below the Western bands. Ligand treated sample values were further normalized to respective cell line vehicle-treated samples. Representative experiment shown for all, N = 2–3 for each experiment.

Close modal

We assessed the differences in IGF pathway activity in IDC (MCF7, T47D, and ZR75.1) and ILC (MDA-MB-134-VI and SUM44PE) cell lines. Both ILC cell lines showed higher pIGF1R/IR expression after IGF1 stimulation compared with the IDC cell lines, though the expression of total IGF1R/IR varied across cell models (Fig. 1E). pAkt (S473) levels, as a measure of downstream pathway activation, did not always correlate with pIGF1R/IR levels. We observed Akt activation in the absence of ligands in T47D and ZR75.1 cells, whereas MCF7 cells demonstrated robust pAkt induction despite modest pIGF1R/IR levels. These results are not entirely surprising given the existing PI3K activating mutations in MCF7 and T47D (41–43), and PTEN loss in ZR75.1 cells (43) that may contribute to the Akt activation. We further assessed the effects of insulin and again observed that ILC cell lines demonstrated enhanced activation of pIGF1R/IR, although with this ligand ZR75.1 also showed robust activation (Supplementary Fig. S1A). To test whether there was a difference between ILC and IDC cell lines in response to ligands targeting other cell surface receptors that may be regulated by E-cadherin (30, 44–46), we assessed the effects of EGF stimulation in our panel of cell lines (Supplementary Fig. S1B). Little EGFR activation could be detected in ILC cell lines suggesting cellular context-specific changes in growth factor signaling activation. Taken together, these data support increased IGF pathway activity in ILC compared with IDC in both representative patient tumors and in a cell line models.

CDH1 knockout cells as a model to study the role of E-cadherin in regulating the IGF pathway

As previously reported (32, 33), and expanded herein (Fig. 1), E-cadherin negatively regulates IGF pathway activity. To better understand this regulation, we generated CRISPR-mediated CDH1 knockout (CDH1 KO) IDC cell lines of MCF7, T47D, and ZR75.1. Both MCF7 and T47D demonstrate successful and complete knockout of CDH1 whereas ZR75.1 CDH1 KO cells retained some expression of E-cadherin though markedly reduced compared with WT cells (Fig. 2A and B). ZR75.1 cells harbor copy number amplification of CDH1 that we believe made complete knockout of CDH1 in this model challenging (42, 47).

Figure 2.

CDH1 knockout cells as a model to study the role of E-cadherin in regulating the IGF pathway. A, Western blotting and (B) qRT-PCR confirms reduction of E-cadherin in CDH1 KO MCF7, T47D, and ZR75.1 CRISPR cell lines compared with wild-type (WT) parental cells. Statistical differences evaluated using paired t test (*, P < 0.05). Representative experiment shown, N = 2 (each with two biological and 3 technical replicates). C, Representative brightfield images (×10, magnifications) of WT and CDH1 KO cell line models plated in 2D and Ultra-low attachment (ULA) plates. D, E-cadherin (red) and p120 catenin (green) staining of WT and CDH1 KO cells confirms CDH1 loss and p120 re-localization in CDH1 KO cell models by confocal microscopy (×60 objective). Inset in the ZR75.1 CDH1 KO panel shows a zoomed in image; scale bar, 50 μm. Representative experiment shown for all, N = 2–3 for each experiment.

Figure 2.

CDH1 knockout cells as a model to study the role of E-cadherin in regulating the IGF pathway. A, Western blotting and (B) qRT-PCR confirms reduction of E-cadherin in CDH1 KO MCF7, T47D, and ZR75.1 CRISPR cell lines compared with wild-type (WT) parental cells. Statistical differences evaluated using paired t test (*, P < 0.05). Representative experiment shown, N = 2 (each with two biological and 3 technical replicates). C, Representative brightfield images (×10, magnifications) of WT and CDH1 KO cell line models plated in 2D and Ultra-low attachment (ULA) plates. D, E-cadherin (red) and p120 catenin (green) staining of WT and CDH1 KO cells confirms CDH1 loss and p120 re-localization in CDH1 KO cell models by confocal microscopy (×60 objective). Inset in the ZR75.1 CDH1 KO panel shows a zoomed in image; scale bar, 50 μm. Representative experiment shown for all, N = 2–3 for each experiment.

Close modal

We first assessed cell morphology given the known role of E-cadherin in formation of cell–cell junctions. As shown in Fig. 2C, MCF7 and T47D CDH1 KO cells have a rounded morphology compared with their parental (WT) cells in 2D culture. In addition, they also demonstrated decreased cell–cell attachment as well as decreased cell–plate attachment. In ultra-low attachment (ULA) conditions, MCF7 and T47D CDH1 KO cells appeared to form less tight, more grape-like cell clusters synonymous with ILC cell aggregates (8, 48). We did not note any obvious difference in either 2D or ULA conditions between ZR75.1 WT and CDH1 KO cells. Examination of E-cadherin by immunofluorescence confirmed loss in all three cell lines models (Fig. 2D). Furthermore, similar to ILC cell lines (8, 49), p120 catenin was re-localized to the cytoplasm in the E-cadherin null CDH1 KO cells. Consistent with our earlier observations, ZR75.1 CDH1 KO cells showed residual E-cadherin expression, and p120 expression was visualized in both the cytoplasm and at the cell membrane. Because of the incomplete knockout of CDH1 in our ZR75.1 cells, we did not perform any additional experiments with this model.

As a function of the cadherin complex, E-cadherin stabilizes β-catenin on the cell membrane and prevents its release into the cytoplasm. The release of β-catenin results in free β-catenin being targeted for proteasomal degradation and/or its accumulation resulting in subsequent activation of Wnt signaling. We thus examined expression of β-catenin following treatment with a GSK-3α and GSK-3β inhibitor (CHIR-99021; Selleckchem #S2924) that functions as a Wnt/β-catenin activator. With vehicle treatment, both MCF7 and T47D CDH1 KO cells exhibited little appreciable β-catenin expression, which was only elevated following 24-hour GSK inhibitor treatment (Supplementary Fig. S1C), suggesting that CDH1 KO alone does not alter β-catenin accumulation. Together, these data support the successful generation of CDH1 KO IDC cell line models that display phenotypes consistent with loss of E-cadherin functionality, including the characteristic re-localization of p120 as seen in human ILC tumors.

Loss of E-cadherin sensitizes cells to IGF signaling pathway activation

To understand the signaling effects of E-cadherin loss, we subjected WT and CDH1 KO cells to IGF1 stimulation. We observed enhanced sensitivity in both CDH1 KO cell lines by proxy of higher pIGF1R/IR and pAkt activation (Fig. 3A). This enhanced signaling response of CDH1 KO cells to WT cells was evident to the greatest extend upon 5 nmol/L IGF1 stimulation: 2.72-fold and 1.3-fold higher signaling in MCF7 and T47D CDH1 KO cells, respectively. We next assessed the effects of E-cadherin loss on IGF2, and insulin signaling and observed enhanced IGF1R/IR activation in MCF7 CDH1 KO cells following ligand stimulation (Fig. 3B and C) that did not translate into enhanced Akt activation. T47D CDH1 KO cells showed similar sensitivity to both IGF2 and insulin when compared with WT cells, suggesting an IGF1-specific differential signaling response.

Figure 3.

Loss of E-cadherin sensitizes cells to IGF signaling pathway activation. Cells were serum-starved overnight and stimulated with (A) IGF1, (B) IGF2 or (C) Insulin (0–10 nmol/L) for 15 minutes. Cells were harvested for Western blot to assess IGF1R/IR and Akt signaling. For quantification, phosphorylated protein levels were normalized to corresponding total protein levels and loading controls. Quantified values are indicated below the Western bands. Ligand-treated sample values were further normalized to respective cell line vehicle treated samples. D, Cells were treated with 10 nmol/L IGF1 for a time course from 0 to 6 hours to assess the duration of signaling activity between WT and CDH1 KO cells. Representative experiment shown for all, N = 2–3 for each experiment.

Figure 3.

Loss of E-cadherin sensitizes cells to IGF signaling pathway activation. Cells were serum-starved overnight and stimulated with (A) IGF1, (B) IGF2 or (C) Insulin (0–10 nmol/L) for 15 minutes. Cells were harvested for Western blot to assess IGF1R/IR and Akt signaling. For quantification, phosphorylated protein levels were normalized to corresponding total protein levels and loading controls. Quantified values are indicated below the Western bands. Ligand-treated sample values were further normalized to respective cell line vehicle treated samples. D, Cells were treated with 10 nmol/L IGF1 for a time course from 0 to 6 hours to assess the duration of signaling activity between WT and CDH1 KO cells. Representative experiment shown for all, N = 2–3 for each experiment.

Close modal

To explore whether our findings were related to a general growth factor sensitivity following loss of E-cadherin, we stimulated cells with EGF but did not observe any consistent differences in pEGFR Tyr1068 activation, or pAkt levels between CDH1 KO and WT cells (Supplementary Fig. S2A). We specifically examined the effects of E-cadherin deletion on FGFR activity given recent studies on the role of FGFR4 as a driver of endocrine resistance (50), particularly in ILC (51). Cells were stimulated with a cocktail of FGF ligands (FGF 1, 2, 4, 6, 8, 17, 19, 21, and 23) resulting in strong enhancement in both raw and normalized pFGFR4 levels in both MCF7 and T47D CDH1 KO cells, despite decreased total FGFR4 levels in the CDH1 KO cells (Supplementary Fig. S2B). Surprisingly, although pFGFR4 was elevated in CDH1 KO cells, downstream activators did not display a difference between WT and CDH1 KO cells. In addition, in both MCF7 and T47D CDH1 KO cells, we observed elevated pathway activity that was sustained for up to 4 hours in the MCF7 cells and up to 2 hours in the T47D cells (Fig. 3D). These results were replicated for pAkt activation in the MCF7 CDH1 KO cells but not in the T47D CDH1 KO, warranting additional studies on the sustained receptor activity and its effects on downstream signaling response. Collectively, these results suggest that there is an enhancement of IGF1 response in IDC cell lines harboring E-cadherin loss under the conditions tested, but that this increased signaling sensitivity may not be generalizable to other cell surface receptors and is mostly applicable to the IGF1 pathway.

Loss of E-cadherin increases IGF1R availability on the membrane to allow ligand binding

To understand why and how loss of E-cadherin leads to sensitized IGF1R signaling, we first compared the expression of IGF1R in the WT and CDH1 KO cells. Using cell fractionation, minimal differences between the MCF7 WT and CDH1 KO cells were noted (Fig. 4A). Higher IGF1R expression was observed in both whole cell and membrane fractions of the T47D CDH1 KO cells compared with WT, which was consistent with increased IGF1R mRNA (Supplementary Fig. S3A). We assessed the availability of the receptor for ligand binding using a biotinylated IGF1 ligand, which was crosslinked to its receptor for subsequent complex quantification via α-biotin immunoblotting. MCF7 and T47D CDH1 KO cells showed a 2.1- and 4-fold higher ligand–receptor complex, respectively, when compared with their corresponding WT cell line controls (Fig. 4B). This finding was stronger in the T47D CDH1 KO cells, likely due to the higher baseline IGF1R expression in these cells. As E-cadherin has previously been reported to create a repressive EGFR complex (30), we examined for the presence of a physical interaction between IGF1R and E-cadherin on the cell membrane that potentially could physically and spatially prevent ligands from binding IGF1R. An immunoprecipitation performed for both IGF1R and E-cadherin did not reveal co-IP of either protein despite multiple attempts to optimize conditions (Fig. 4C). Imaging of IGF1R to assess localization on the membrane showed no significant differences between the WT and CDH1 KO cells (Fig. 4D and E). Our results suggest that increased receptor availability to bind ligand in CDH1 KO cells likely leads to the observed elevated signaling induction upon ligand binding.

Figure 4.

Loss of E-cadherin increases IGF1R availability on the membrane to allow ligand binding. A, Cell fractionation assay performed on MCF7 and T47D WT and CDH1 KO cells to assess whole-cell and membrane IGF1R expression levels. IGF1R bands were quantified and normalized to membrane control, AIF1. Quantified values are indicated below the Western bands. B, Cells were stimulated with biotinylated IGF1 (0–10 nmol/L) for 10 minutes and crosslinked to assess ligand–receptor complex levels between WT and CDH1 KO cells. C, Immunoprecipitation of IGF1R and E-cadherin in T47D cells was assessed for a co-IP of other proteins, with p-120 catenin assessed as a known interactor of E-cadherin. D and E, Cell lines were dual stained for EpCAM (green) and IGF1R (red) and imaged by confocal microscopy at a ×60 objective; scale bar, 50 μm. 12 images for each cell line were quantified and graphed (E). Representative experiment shown for all, N = 2–3 for each experiment.

Figure 4.

Loss of E-cadherin increases IGF1R availability on the membrane to allow ligand binding. A, Cell fractionation assay performed on MCF7 and T47D WT and CDH1 KO cells to assess whole-cell and membrane IGF1R expression levels. IGF1R bands were quantified and normalized to membrane control, AIF1. Quantified values are indicated below the Western bands. B, Cells were stimulated with biotinylated IGF1 (0–10 nmol/L) for 10 minutes and crosslinked to assess ligand–receptor complex levels between WT and CDH1 KO cells. C, Immunoprecipitation of IGF1R and E-cadherin in T47D cells was assessed for a co-IP of other proteins, with p-120 catenin assessed as a known interactor of E-cadherin. D and E, Cell lines were dual stained for EpCAM (green) and IGF1R (red) and imaged by confocal microscopy at a ×60 objective; scale bar, 50 μm. 12 images for each cell line were quantified and graphed (E). Representative experiment shown for all, N = 2–3 for each experiment.

Close modal

Loss of E-cadherin enhances anoikis resistance and increases cell survival

We previously showed increased anchorage-independent growth in ILC cells compared with IDC (8). In IDC cells with CDH1 KO, we similarly found an increase in ULA growth compared with WT cells in the MCF7 and T47D models (31) as has also been seen by others (52). ZR75.1 WT and CDH1 KO cells did not show any significant differences in either growth condition (Supplementary Fig. S4A), potentially due to the incomplete CDH1 deletion and a copy-number amplification (42). To explore the previously observed ability for MCF7 and T47D CDH1 KO cells to thrive in ULA conditions, we examined anoikis resistance using an Annexin V and Propidium Iodide staining technique followed by flow cytometry after a 3-day culture in 2D or ULA plates. Although the percentage of live cells in 2D were similar between WT and CDH1 KO cells, the fraction of live cells in ULA were significantly higher in the T47D CDH1 KO cells (Fig. 5A), consistent with anoikis resistance. No major differences were observed between the MCF7 WT and CDH1 KO cells in either condition.

Figure 5.

Loss of E-cadherin enhances anoikis resistance and increases cell survival. A, MCF7 and T47D cells were grown in 2D and ULA plates for 3–4 days and stained with Annexin V and propidium iodide to measure live and apoptotic cells, respectively. Live cell percentage in 2D for each cell line was used to normalize the live cell percentages in ULA plates. Representative experiment shown; N = 3 (each with three biological replicates). B and C, Cells were plated at 2,000 cells/well in full serum and low serum supplemented with 5 nmol/L IGF1 media and stained with 0.5% Crystal Violet after two to three weeks. Representative images are shown. Plates were de-stained with 10% acetic acid, absorbance quantified and graphed after normalization to WT cells of corresponding conditions. Statistical differences were evaluated using two-way ANOVA (*, P < 0.05; **, P < 0.01; and ***, P < 0.001, representative experiment shown, N = 3; each with three biological replicates).

Figure 5.

Loss of E-cadherin enhances anoikis resistance and increases cell survival. A, MCF7 and T47D cells were grown in 2D and ULA plates for 3–4 days and stained with Annexin V and propidium iodide to measure live and apoptotic cells, respectively. Live cell percentage in 2D for each cell line was used to normalize the live cell percentages in ULA plates. Representative experiment shown; N = 3 (each with three biological replicates). B and C, Cells were plated at 2,000 cells/well in full serum and low serum supplemented with 5 nmol/L IGF1 media and stained with 0.5% Crystal Violet after two to three weeks. Representative images are shown. Plates were de-stained with 10% acetic acid, absorbance quantified and graphed after normalization to WT cells of corresponding conditions. Statistical differences were evaluated using two-way ANOVA (*, P < 0.05; **, P < 0.01; and ***, P < 0.001, representative experiment shown, N = 3; each with three biological replicates).

Close modal

To further explore cell survival, we performed colony formation assays by plating cells at a low density in either full serum (10% FBS) or low serum (0.5% FBS) containing 5 nmol/L IGF1 in 2D plates. T47D CDH1 KO cells showed increased clonogenic survival in both conditions (Fig. 5B and C). CDH1 KO cells also demonstrated increased colonies in full serum compared with WT, but with no colonies observed in low serum conditions. MCF7 CDH1 KO cells showed no clear differences in full serum quantifications; however, they formed a greater number of smaller colonies than the WT cells, with similar observations in low serum conditions. The presence of enhanced cell survival phenotype in the low serum plus IGF1 condition further highlights the IGF-specific phenotype in our CDH1 KO cells. Our data support that loss of E-cadherin may play a context-dependent role in anoikis resistance and cell survival.

Loss of E-cadherin increases collagen I haptotaxis, IGF1-driven migration and serum-driven collagen invasion

We investigated whether deletion of CDH1 in IDC cell lines confers haptotactic migration toward the extracellular matrix, a phenotype that is pronounced in ILC cell lines (8). We observed a significantly higher migration toward Collagen I by MCF7 and T47D CDH1 KO cells (Fig. 6A) compared with their respective WT cells. Multiple studies have reported on E-cadherin loss permitting cell migration and invasion (53–55); conflicting reports on the requirement of E-cadherin for metastasis have also been reported (56). To explore this inconsistency while assessing the importance of IGF1, CDH1 KO cells were subjected to Transwell migration assays with chemotactic low serum (0.5% FBS) media ± 5 nmol/L IGF1 or full serum (10% FBS) media. Both CDH1 KO cell models showed increased migration toward serum compared with WT cells (Fig. 6B and C). Consistent with the IGF1 sensitive nature of these CDH1 KO cell lines, MCF7 and T47D CDH1 KO cells also showed significant migration toward IGF1, a phenotype that was successfully halted upon IGF1R/IR inhibitor, BMS-754807 addition (Supplementary Fig. S4B–S4D). Higher migration of T47D CDH1 KO cells without the presence of any chemotactic gradient highlights the migratory phenotype driven by the independent loss of E-cadherin in this model. Given the haptotactic migration toward Collagen I and migration toward IGF1 and serum, we hypothesized that CDH1 KO cells may also be able to invade through Collagen I (8) toward a gradient of IGF1 and serum. Although there was no Collagen I invasion observed with the MCF7 cells (Supplementary Fig. S4E), T47D CDH1 KO cells showed a significant Collagen I invasion toward serum (Fig. 6D). It is important to note that neither of these three cell lines are invasive by nature and the invasive phenotypes observed in T47D CDH1 KO cells are promising and requires greater investigation in the context of tumor invasion.

Figure 6.

Loss of E-cadherin increases collagen I haptotaxis, IGF1-driven migration and serum-driven collagen invasion. A, Images and quantification of Crystal Violet-stained Collagen I coated inserts from haptotaxis assays after 72 hours. Migrated cell colonies were quantified with ImageJ and plotted. B, Images and (C) quantification of Crystal Violet-stained Transwell inserts from migration assays toward the indicated attractants after 72 hours. D, Images and quantification of Crystal Violet-stained Collagen I inserts from invasion assays toward the indicated attractants after 72 hours. Graphs show representative data normalized to low serum WT cell samples from two to three independent experiments (N = 2 biological replicates). P values from one-way ANOVA statistical testing. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

Figure 6.

Loss of E-cadherin increases collagen I haptotaxis, IGF1-driven migration and serum-driven collagen invasion. A, Images and quantification of Crystal Violet-stained Collagen I coated inserts from haptotaxis assays after 72 hours. Migrated cell colonies were quantified with ImageJ and plotted. B, Images and (C) quantification of Crystal Violet-stained Transwell inserts from migration assays toward the indicated attractants after 72 hours. D, Images and quantification of Crystal Violet-stained Collagen I inserts from invasion assays toward the indicated attractants after 72 hours. Graphs show representative data normalized to low serum WT cell samples from two to three independent experiments (N = 2 biological replicates). P values from one-way ANOVA statistical testing. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

Close modal

Loss of E-cadherin sensitizes cells to Fulvestrant and IGF1R/PI3K/Akt/MEK inhibitors in a context-dependent manner

Motivated by the increased sensitivity of CDH1 KO cells to IGF ligands and the potential for clinical translation, we sought to determine whether these cells are also sensitive to IGF1R inhibitors. Only T47D CDH1 KO cells were more sensitive to the tested IGF1R inhibitors compared with WT cells (BMS-754807 and OSI-906; Fig. 7A, B, and E). MCF7 CDH1 KO cells did not show any significant differences when compared with WT cells (Supplementary Fig. S5A), although BMS754807 treatment did inhibit IGF1R and Akt signaling in these cell lines, albeit at a lower efficacy in CDH1 KO cells, potentially owing to their increased ligand sensitivity (Supplementary Fig. S5B). Given the higher incidence of PIK3CA hotspot mutations and occurrence of PTEN loss in ILC (4, 12, 57), we examined whether the loss of E-cadherin in these IDC cell lines also sensitizes them to PI3K and Akt inhibitors, Alpelisib and MK-2206, respectively. As with the IGF1R inhibitors, only T47D CDH1 KO cells showed increased sensitivity to MK2206 and an overall increased sensitivity trend to Alpelisib upon averaging statistical analysis of repeated experiments (Fig. 7CE; Supplementary Fig. S5A and S5C). Next, we assessed whether the absence of susceptibility to IGF1R and Akt inhibitors in the MCF7 CDH1 KO cells might be explained by alternative pathways being activated. We performed combination treatments of BMS-754807 with MEK inhibitor, U0126, and observed a strong additive effect where both MCF7 and T47D CDH1 KO cells were more sensitive than their corresponding WT cells (Fig. 7F and G). Although this was an expected result in the T47D cells, it is interesting to note the additive effect in the MCF7 CDH1 KO cells as this suggests the activation of alternative pathways following IGF1R, PI3K, and Akt independent inhibitions. We further assessed the synergy effect using the SynergyFinder platform (35), with the ZIP method as output (Supplementary Fig. S6A and S6B). In MCF7 and T47D WT and CDH1 KO cells, synergy scores were between −9.717 and 10.995, which validate an additive but not synergistic drug combination effect. As over 90% of ILC tumors are estrogen receptor (ER) positive, we also investigated the combination effects of Akt inhibition with an ER degrader, Fulvestrant, and observed a higher susceptibility in the T47D CDH1 KO cells (Fig. 7H; Supplementary Fig. S6C), supporting the notion of targeting the IGF pathway concurrently to targeting ER. These results support the study of combinatory therapies in preclinical studies to better target the activation of the IGF pathway following E-cadherin loss. Finally, to analyze pathway inhibition in a more physiologically relevant model, we used patient-derived ILC and IDC breast organoids (PDO) to compare Akt inhibition sensitivity. Loss of E-cadherin in ILC PDOs (PDO 30 and 46) is depicted by immunofluorescence (Fig. 7I), whereas membranous E-cadherin expression is still present in IDC PDOs (PDO 56, 75, 102, 109, 126, and 134). Comparing the IDC and ILC PDOs, ILC organoids did demonstrate a stronger trend in sensitivity to MK2206 treatment (Fig. 7J; Supplementary Fig. S6D). This was further shown by a significantly lower area under dose–response curves comparing the ILC with the IDC organoids (Fig. 7K), strengthening potential translational implications. Although our sample sizes are small given the limitations of patient-derived organoids, we are able to confidently conclude on the higher sensitivity to Akt inhibitor in ILC PDOs.

Figure 7.

Loss of E-cadherin sensitizes cells to Fulvestrant and IGF1R/PI3K/Akt/MEK inhibitors in a context-dependent manner. T47D parental and CDH1 KO cells were seeded in 96-well 2D plates and treated with IGF1R inhibitors (OSI-906 or BMS-754807), PI3K inhibitor (Alpelisib) or Akt inhibitor (MK2206) for 6 days. Conditions in the panels as follows: (A) BMS-754807; (B) OSI-906; (C) MK2206, (D) Alpelisib, and (E) shows a summary of IC50 values. CellTiter Glo assay was used to assess cell viability (relative luminescence) and data were normalized to vehicle-treated control. F and G, Cells were treated a combination of MEK inhibitor (U0126) and BMS-754807 for 6 days and viability assessed as above. H, Cells were treated a combination of Akt inhibitor (MK2206) and Fulvestrant for 6 days and viability assessed as above. IC50 values for viability were calculated by nonlinear regression and statistical differences evaluated using sum-of-squares Global f-test [P < 0.05; representative experiment shown; N = 3 (each with six biological replicates)]. I, E-cadherin immunofluorescence was performed on patient-derived IDC and ILC organoids. Green: E-cadherin and blue: Hoechst; scale bar, 50 μm. J, ILC and IDC organoids were treated with Akt inhibitor (MK2206) for 12 days and viability assessed with CellTiter Glo 3D (relative luminescence), and data were normalized to vehicle-treated cells. Statistical differences were evaluated using two-way ANOVA (**, P < 0.005; ***, P< 0.0005; ***, P < 0.0001; representative experiment shown; n = 1 (each with six biological replicates). K, AUC of IDC and ILC organoids dose response with MK2206 was calculated from (J), with statistical differences evaluated using unpaired t tests (P = 0.0056).

Figure 7.

Loss of E-cadherin sensitizes cells to Fulvestrant and IGF1R/PI3K/Akt/MEK inhibitors in a context-dependent manner. T47D parental and CDH1 KO cells were seeded in 96-well 2D plates and treated with IGF1R inhibitors (OSI-906 or BMS-754807), PI3K inhibitor (Alpelisib) or Akt inhibitor (MK2206) for 6 days. Conditions in the panels as follows: (A) BMS-754807; (B) OSI-906; (C) MK2206, (D) Alpelisib, and (E) shows a summary of IC50 values. CellTiter Glo assay was used to assess cell viability (relative luminescence) and data were normalized to vehicle-treated control. F and G, Cells were treated a combination of MEK inhibitor (U0126) and BMS-754807 for 6 days and viability assessed as above. H, Cells were treated a combination of Akt inhibitor (MK2206) and Fulvestrant for 6 days and viability assessed as above. IC50 values for viability were calculated by nonlinear regression and statistical differences evaluated using sum-of-squares Global f-test [P < 0.05; representative experiment shown; N = 3 (each with six biological replicates)]. I, E-cadherin immunofluorescence was performed on patient-derived IDC and ILC organoids. Green: E-cadherin and blue: Hoechst; scale bar, 50 μm. J, ILC and IDC organoids were treated with Akt inhibitor (MK2206) for 12 days and viability assessed with CellTiter Glo 3D (relative luminescence), and data were normalized to vehicle-treated cells. Statistical differences were evaluated using two-way ANOVA (**, P < 0.005; ***, P< 0.0005; ***, P < 0.0001; representative experiment shown; n = 1 (each with six biological replicates). K, AUC of IDC and ILC organoids dose response with MK2206 was calculated from (J), with statistical differences evaluated using unpaired t tests (P = 0.0056).

Close modal

Recent advancements have led to personalized cancer treatments; however, more progress needs to be made to effectively target tumors. Our study highlights the role of E-cadherin in regulating IGF signaling and tumorigenic properties in vitro. Genetic deletion of E-cadherin in IDC cells caused reduced cell–cell attachment and morphology transformation into looser cell clusters in suspension. Furthermore, E-cadherin loss was associated with enhanced anchorage independence and anoikis resistance, all of which are features observed in ILC tumors. In addition, CDH1 KO cells were more sensitive to IGF1, IGF2 and insulin signaling attributable to an increased receptor availability for ligand binding. This was an IGF-specific phenotype and was not observed with other growth factor signaling pathways tested albeit FGFR4 activations were higher in the CDH1 KO cells but without significant downstream signaling. Further characterization of CDH1 KO cells also showed an increased cell survival in clonogenic assays, and enhanced cell migration/invasion in the majority of models, suggesting a role for E-cadherin in controlling metastatic phenotypes. Importantly, a context-dependent increase in sensitivity to IGF1R, PI3K, Akt, and MEK inhibitors was also observed in our models. This study contributes to an increasing body of work delineating ILC as a unique breast cancer subtype and suggests a potential for targeted therapeutic approaches toward the IGF1R/PI3K/Akt axis owing to the diagnostic loss of E-cadherin in this histological subtype.

Studies have attributed spatial accessibility of IGF1R for ligand binding as a limiting factor in signaling activation (30, 58), and consistent with this notion, we found that loss of E-cadherin confers increased IGF ligand binding by IGF1R, and subsequent pathway activation. Despite previous reports on E-cadherin–repressing growth factor receptors (30, 33, 58) and co-localization (58), we were unable to find an association by co-IP despite numerous attempts, suggesting that this is not the mechanism affecting the availability of IGF1R for ligand binding when E-cadherin is present. The formation of a ternary complex involving E-cadherin, IGF1R and integrins is an ongoing topic of investigation thought to affect cell mobility (59), which could additionally complicate immunoprecipitation. Similar interactions have been reported for E-cadherin and EGFR, although we did not observe an increased sensitivity to the EGF ligand in our CDH1 KO cell lines (30, 60). We, and others, have shown that downregulation of E-cadherin by growth factors can promote EMT (58, 61); others have additionally demonstrated growth factor signaling regulation by E-cadherin, suggesting that a bidirectional feedback loop may exist. These studies were initially almost exclusively performed in the context of EGFR where multiple groups had shown that E-cadherin inhibits ligand-dependent activation of EGFR signaling (30, 62, 63). However, a recent study by Teo and colleagues (13) supported our findings by reporting on ILC cell lines generated from a p53-deficient metastatic mouse model exhibiting had PI3K/AKT pathway activation and enhanced sensitivity to pathway inhibition.

Because of the function of E-cadherin in maintaining cell–cell adhesion, its loss through EMT or genetic deletion is often thought to correlate with increased tumor invasion and metastasis (62, 64). Other studies, however, have suggested a requirement for E-cadherin in metastasis (56). Here, we show that CDH1 KO cells demonstrate increased migration toward Collagen I (8). The CDH1 KO cells also showed an enhanced migration toward serum and IGF1, with T47D CDH1 KO cells showing increased migration even in the absence of a chemoattractant. T47D CDH1 KO also showed invasion through Collagen I, providing additional in vitro evidence supporting that the loss of E-cadherin may enhance metastatic phenotypes. The in vitro nature of these findings, however, is an interpretation limitation and in vivo experimentation is needed to better understand the role of E-cadherin in metastasis and validate its effect on IGF signaling.

Inhibitors of downstream activators such as PI3K, Akt, and MEK have successfully been developed and approved for the clinic. This is paramount in malignancy with a high percentage of mutations leading to PIK3CA activation and PTEN loss such as in ILC (4), and maybe more so important given the high IGF pathway activation found in the SCAN-B dataset. Other studies with CRISPR MCF7 CDH1 KO cells lines have shown that loss of E-cadherin leads to dependency upon ROS1 and enhanced sensitivity to crizotinib (65) for which a clinical trial is currently underway. We report increased sensitivity to IGF1R/PI3K/Akt and MEK inhibitors in T47D CDH1 KO cells and an increased trend in sensitivity to an Akt inhibitor in ILC PDOs compared with IDC PDOs. Given the genetic landscape of MCF7 and T47D cells, these models are primed for hyperactive Akt signaling, which has been shown to be further enhanced by loss of E-cadherin. The helical domain mutation in MCF7 cells (E545K) is reported to have a more aggressive phenotype over the kinase domain mutation in T47D (H1047R; ref. 66), which may explain the absence of additional sensitivity in MCF7 CDH1 KO cells to compounds tested because the PIK3CA mutation may effectively mask any potential effect from E-cadherin loss. With the possibility of IGF1R inhibitor efficacy being reduced by hyperactive downstream pathways such as MEK as seen in colon carcinoma (67) and the possibilities of downstream activation inhibition being overhauled by hyperactive receptor tyrosine kinases (68), we tested and observed an additive effect of treating cells with an IGF1R inhibitor in combination with an MEK inhibitor. Furthermore, additive effects of combining Akt inhibitors with Fulvestrant was also found to be beneficial in E-cadherin deficient cells, proving novel therapeutical benefits that require further exploration in vivo. Our studies with ILC PDOs showing higher sensitivity to Akt inhibitor is very encouraging and requires additional investigation. Future studies will need to test the efficacy of inhibitors targeting the IGF pathway comparing cells ± CDH1 in vivo and investigate the potential of efficiently exploiting E-cadherin as a biomarker of therapeutic response.

In conclusion, this study delineates the role of E-cadherin in regulating enhanced IGF signaling, controlling metastatic phenotypes, and identifying use of specific tyrosine kinase inhibitors that can effectively target elevated IGF1 signaling activation. We reveal novel findings of increased receptor availability for ligand binding following the loss of E-cadherin and increased susceptibility to IGF1R/PI3K/Akt and MEK inhibitors upon CDH1 deletion. Our study adds to a growing body of evidence suggesting that CDH1 may represent a biomarker of response to kinase inhibitors and further exemplifies the need for investigation into the clinical translation of these findings. The in vitro nature of our studies in a small panel of cell lines is an important, notable limitation, but notwithstanding, we have robustly revealed important cellular- and contextual-dependent functions of E-cadherin in IGF-driven tumor phenotypes. Our findings require translation into in vivo models with the goal of validating E-cadherin loss as a biomarker of growth factor receptor inhibitor response in breast cancer as well as the utilization of E-cadherin expression as a vital patient stratification tool in clinical trials.

S. Oesterreich reports a patent for 10504-025US1 pending. A.V. Lee reports grants from NIH, Susan G. Komen, and Breast Cancer Research Foundation during the conduct of the study; as well as personal fees from Ocean Genomics outside the submitted work; and reports a patent for therapeutic combinations using IGF1R pathway inhibitors, and methods to predict anti-IGF1R therapeutic efficacy pending. No disclosures were reported by the other authors.

A. Elangovan: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. J. Hooda: Resources, investigation, methodology, writing–review and editing. L. Savariau: Resources, validation, writing–review and editing. S. Puthanmadhomnarayanan: Data curation, software, writing–original draft, writing–review and editing. M.E. Yates: Data curation, software, writing–review and editing. J. Chen: Data curation, validation, investigation, writing–review and editing. D.D. Brown: Resources, methodology, writing–review and editing. P.F. McAuliffe: Resources, methodology, writing–review and editing. S. Oesterreich: Conceptualization, resources, supervision, funding acquisition, project administration, writing–review and editing. J.M. Atkinson: Conceptualization, resources, supervision, methodology, project administration, writing–review and editing. A.V. Lee: Conceptualization, resources, supervision, funding acquisition, project administration, writing–review and editing.

The authors would like to thank Osama Shah for data acquisition and bioinformatic input and the Institute for Precision Medicine, a partnership of the University of Pittsburgh and UPMC, for providing the breast cancer patient derived organoids used in these studies. This research was supported in part by the University of Pittsburgh Center for Research Computing through the resources provided. Research funding for this project was provided in part by Susan G. Komen Scholar awards (SAC110021 and SAC160073) and the Breast Cancer Research Foundation (R01CA224909 and R01CA252378; to A.V. Lee and S. Oesterreich), and the National Cancer Institute (R01CA224909).

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.

Note: Supplementary data for this article are available at Molecular Cancer Research Online (http://mcr.aacrjournals.org/).

1.
Pestalozzi
BC
,
Zahrieh
D
,
Mallon
E
,
Gusterson
BA
,
Price
KN
,
Gelber
RD
, et al
.
Distinct clinical and prognostic features of infiltrating lobular carcinoma of the breast: combined results of 15 International Breast Cancer Study Group Clinical Trials
.
J Clin Oncol
2008
;
26
:
3006
14
.
2.
Corben
AD
.
Pathology of invasive breast disease
.
Surg Clin North Am
2013
;
93
:
363
92
.
3.
Mamtani
A
,
King
TA
.
Lobular breast cancer: different disease, different algorithms?
Surg Oncol Clin N Am
2018
;
27
:
81
94
.
4.
Ciriello
G
,
Gatza
ML
,
Beck
AH
,
Wilkerson
MD
,
Rhie
SK
,
Pastore
A
, et al
.
Comprehensive molecular portraits of invasive lobular breast cancer
.
Cell
2015
;
163
:
506
19
.
5.
Du
T
,
Zhu
L
,
Levine
KM
,
Tasdemir
N
,
Lee
AV
,
Vignali
DAA
, et al
.
Invasive lobular and ductal breast carcinoma differ in immune response, protein translation efficiency and metabolism
.
Sci Rep
2018
;
8
:
7205
.
6.
Rakha
EA
,
Ellis
IO
.
Lobular breast carcinoma and its variants
.
Semin Diagn Pathol
2010
;
27
:
49
61
.
7.
Chen
Z
,
Yang
J
,
Li
S
,
Lv
M
,
Shen
Y
,
Wang
B
, et al
.
Invasive lobular carcinoma of the breast: a special histological type compared with invasive ductal carcinoma
.
PLoS ONE
2017
;
12
:
e0182397
.
8.
Tasdemir
N
,
Bossart
EA
,
Li
Z
,
Zhu
L
,
Sikora
MJ
,
Levine
KM
, et al
.
Comprehensive phenotypic characterization of human invasive lobular carcinoma cell lines in 2D and 3D cultures
.
Cancer Res
2018
;
78
:
6209
22
.
9.
Barroso-Sousa
R
,
Metzger-Filho
O
.
Differences between invasive lobular and invasive ductal carcinoma of the breast: results and therapeutic implications
.
Ther Adv Med Oncol
2016
;
8
:
261
6
.
10.
Vos
CB
,
Cleton-Jansen
AM
,
Berx
G
,
de Leeuw
WJ
,
ter Haar
NT
,
van Roy
F
, et al
.
E-Cadherin inactivation in lobular carcinoma in situ of the breast: an early event in tumorigenesis
.
Br J Cancer
1997
;
76
:
1131
3
.
11.
Yoshiura
K
,
Kanai
Y
,
Ochiai
A
,
Shimoyama
Y
,
Sugimura
T
,
Hirohashi
S
.
Silencing of the E-cadherin invasion-suppressor gene by CpG methylation in human carcinomas
.
Proc Natl Acad Sci U S A.
1995
;
92
:
7416
9
.
12.
Desmedt
C
,
Zoppoli
G
,
Gundem
G
,
Pruneri
G
,
Larsimont
D
,
Fornili
M
, et al
.
Genomic characterization of primary invasive lobular breast cancer
.
J Clin Oncol
2016
;
34
:
1872
81
.
13.
Teo
K
,
Gomez-Cuadrado
L
,
Tenhagen
M
,
Byron
A
,
Ratze
M
,
van Amersfoort
M
, et al
.
E-Cadherin loss induces targetable autocrine activation of growth factor signalling in lobular breast cancer
.
Sci Rep
2018
;
8
:
15454
.
14.
Becker
MA
,
Ibrahim
YH
,
Cui
X
,
Lee
AV
,
Yee
D
.
The IGF pathway regulates ERα through a S6K1-dependent mechanism in breast cancer cells
.
Mol Endocrinol
2011
;
25
:
516
28
.
15.
Kim
H-J
,
Litzenburger
BC
,
Cui
X
,
Delgado
DA
,
Grabiner
BC
,
Lin
X
, et al
.
Constitutively active Type I Insulin-like growth factor receptor causes transformation and xenograft growth of immortalized mammary epithelial cells and is accompanied by an epithelial-to-mesenchymal transition mediated by NF-κB and Snail
.
Mol Cell Biol
2007
;
27
:
3165
75
.
16.
Yee
D
,
Lee
AV
.
Crosstalk between the insulin-like growth factors and estrogens in breast cancer
.
J Mammary Gland Biol Neoplasia
2000
;
5
:
107
15
.
17.
Hawsawi
Y
,
El-Gendy
R
,
Twelves
C
,
Speirs
V
,
Beattie
J
.
Insulin-like growth factor—oestradiol crosstalk and mammary gland tumourigenesis
.
Biochim Biophys Acta
2013
;
1836
:
345
53
.
18.
Baserga
R
,
Peruzzi
F
,
Reiss
K
.
The IGF-1 receptor in cancer biology
.
Int J Cancer
2003
;
107
:
873
7
.
19.
Litzenburger
BC
,
Kim
HJ
,
Kuiatse
I
,
Carboni
JM
,
Attar
RM
,
Gottardis
MM
, et al
.
BMS-536924 reverses IGF-IR-induced transformation of mammary epithelial cells and causes growth inhibition and polarization of MCF7 cells
.
Clin Cancer Res
2009
;
15
:
226
37
.
20.
Iams
WT
,
Lovly
CM
.
Molecular pathways: clinical applications and future direction of insulin-like growth factor-1 receptor pathway blockade
.
Clin Cancer Res
2015
;
21
:
4270
7
.
21.
Morgillo
F
,
De Vita
F
,
Antoniol
G
,
Orditura
M
,
Auriemma
PP
,
Diadema
MR
, et al
.
Serum insulin-like growth factor 1 correlates with the risk of nodal metastasis in endocrine-positive breast cancer
.
Curr Oncol
2013
;
20
:
e283
8
.
22.
Crudden
C
,
Girnita
A
,
Girnita
L
.
Targeting the IGF-1R: the tale of the tortoise and the hare
.
Front Endocrinol
2015
;
6
:
64
.
23.
Boone
DN
,
Lee
AV
.
Targeting the insulin-like growth factor receptor: developing biomarkers from gene expression profiling
.
Crit Rev Oncog
2012
;
17
:
161
73
.
24.
Singh
P
,
Alex
JM
,
Bast
F
.
Insulin receptor (IR) and insulin-like growth factor receptor 1 (IGF-1R) signaling systems: novel treatment strategies for cancer
.
Med Oncol
2014
;
31
:
805
.
25.
Ekyalongo
RC
,
Yee
D
.
Revisiting the IGF-1R as a breast cancer target
.
NPJ Precis Oncol
2017
;
1
:
14
.
26.
Yang
Y
,
Yee
D
.
Targeting insulin and insulin-like growth factor signaling in breast cancer
.
J Mammary Gland Biol Neoplasia
2012
;
17
:
251
61
.
27.
Pappo
AS
,
Patel
SR
,
Crowley
J
,
Reinke
DK
,
Kuenkele
K-P
,
Chawla
SP
, et al
.
R1507, a monoclonal antibody to the insulin-like growth factor 1 receptor, in patients with recurrent or refractory ewing sarcoma family of tumors: results of a phase II sarcoma alliance for research through Collaboration Study
.
J Clin Oncol
2011
;
29
:
4541
7
.
28.
Pappo
AS
,
Vassal
G
,
Crowley
JJ
,
Bolejack
V
,
Hogendoorn
PCW
,
Chugh
R
, et al
.
A phase 2 trial of R1507, a monoclonal antibody to the insulin-like growth factor-1 receptor (IGF-1R), in patients with recurrent or refractory rhabdomyosarcoma, osteosarcoma, synovial sarcoma, and other soft tissue sarcomas: results of a Sarcoma Alliance
.
Cancer
2014
;
120
:
2448
56
.
29.
Langer
CJ
,
Novello
S
,
Park
K
,
Krzakowski
M
,
Karp
DD
,
Mok
T
, et al
.
Randomized, phase III trial of first-line figitumumab in combination with paclitaxel and carboplatin versus paclitaxel and carboplatin alone in patients with advanced non–small cell lung cancer
.
J Clin Oncol
2014
;
32
:
2059
66
.
30.
Qian
X
,
Karpova
T
,
Sheppard
AM
,
McNally
J
,
Lowy
DR
.
E-Cadherin-mediated adhesion inhibits ligand-dependent activation of diverse receptor tyrosine kinases
.
EMBO J
2004
;
23
:
1739
48
.
31.
Tasdemir
N
,
Ding
K
,
Savariau
L
,
Levine
KM
,
Du
T
,
Elangovan
A
, et al
.
Proteomic and transcriptomic profiling identifies mediators of anchorage-independent growth and roles of inhibitor of differentiation proteins in invasive lobular carcinoma
.
Sci Rep
2020
;
10
:
11487
.
32.
Erdem
C
,
Nagle
AM
,
Casa
AJ
,
Litzenburger
BC
,
Wang
YF
,
Taylor
DL
, et al
.
Proteomic screening and lasso regression reveal differential signaling in insulin and Insulin-like growth factor I (IGF1) pathways
.
Mol Cell Proteomics
2016
;
15
:
3045
57
.
33.
Nagle
AM
,
Levine
KM
,
Tasdemir
N
,
Scott
JA
,
Burlbaugh
K
,
Kehm
J
, et al
.
Loss of E-cadherin enhances IGF1–IGF1R pathway activation and sensitizes breast cancers to anti-IGF1R/InsR inhibitors
.
Clin Cancer Res
2018
;
24
:
5165
77
.
34.
Chen
F
,
Ding
K
,
Priedigkeit
N
,
Elangovan
A
,
Levine
KM
,
Carleton
N
, et al
.
Single-cell transcriptomic heterogeneity in invasive ductal and lobular breast cancer cells
.
Cancer Res
2021
;
81
:
268
81
.
35.
Ianevski
A
,
He
L
,
Aittokallio
T
,
Tang
J
.
SynergyFinder: a web application for analyzing drug combination dose–response matrix data
.
Bioinformatics
2020
;
36
:
2645
.
36.
Brueffer
C
,
Gladchuk
S
,
Winter
C
,
Vallon-Christersson
J
,
Hegardt
C
,
Hakkinen
J
, et al
.
The mutational landscape of the SCAN-B real-world primary breast cancer transcriptome
.
EMBO Mol Med
2020
;
12
:
e12118
.
37.
Yoshihara
K
,
Shahmoradgoli
M
,
Martinez
E
,
Vegesna
R
,
Kim
H
,
Torres-Garcia
W
, et al
.
Inferring tumour purity and stromal and immune cell admixture from expression data
.
Nat Commun
2013
;
4
:
2612
.
38.
Hanzelmann
S
,
Castelo
R
,
Guinney
J
.
GSVA: gene set variation analysis for microarray and RNA-seq data
.
BMC Bioinf
2013
;
14
:
7
.
39.
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
.
40.
Sachs
N
,
de Ligt
J
,
Kopper
O
,
Gogola
E
,
Bounova
G
,
Weeber
F
, et al
.
A living biobank of breast cancer organoids captures disease heterogeneity
.
Cell
2018
;
172
:
373
86e10
.
41.
Ikediobi
ON
,
Davies
H
,
Bignell
G
,
Edkins
S
,
Stevens
C
,
O'Meara
S
, et al
.
Mutation analysis of 24 known cancer genes in the NCI-60 cell line set
.
Mol Cancer Ther
2006
;
5
:
2606
12
.
42.
Marcotte
R
,
Sayad
A
,
Brown
KR
,
Sanchez-Garcia
F
,
Reimand
J
,
Haider
M
, et al
.
Functional genomic landscape of human breast cancer drivers, vulnerabilities, and resistance
.
Cell
2016
;
164
:
293
309
.
43.
Hollestelle
A
,
Nagel
JHA
,
Smid
M
,
Lam
S
,
Elstrodt
F
,
Wasielewski
M
, et al
.
Distinct gene mutation profiles among luminal-type and basal-type breast cancer cell lines
.
Breast Cancer Res Treat
2010
;
121
:
53
64
.
44.
Andl
CD
,
Rustgi
AK
.
No one-way street: cross-talk between E-cadherin and receptor tyrosine kinase (RTK) signaling—a mechanism to regulate RTK activity
.
Cancer Biol Ther
2005
;
4
:
35
8
.
45.
De-Freitas-Junior
JCM
,
Carvalho
S
,
Dias
AM
,
Oliveira
P
,
Cabral
J
,
Seruca
R
, et al
.
Insulin/IGF-I signaling pathways enhances tumor cell invasion through bisecting GlcNAc N-glycans modulation. An Interplay with E-cadherin
.
PLoS One
2013
;
8
:
e81579
.
46.
Curto
M
,
Cole
BK
,
Lallemand
D
,
Liu
CH
,
McClatchey
AI
.
Contact-dependent inhibition of EGFR signaling by Nf2/Merlin
.
J Cell Biol
2007
;
177
:
893
903
.
47.
Lombaerts
M
,
Van Wezel
T
,
Philippo
K
,
Dierssen
JWF
,
Zimmerman
RME
,
Oosting
J
, et al
.
E-Cadherin transcriptional downregulation by promoter methylation but not mutation is related to epithelial-to-mesenchymal transition in breast cancer cell lines
.
Br J Cancer
2006
;
94
:
661
71
.
48.
Kenny
PA
,
Lee
GY
,
Myers
CA
,
Neve
RM
,
Semeiks
JR
,
Spellman
PT
, et al
.
The morphologies of breast cancer cell lines in three-dimensional assays correlate with their profiles of gene expression
.
Mol Oncol
2007
;
1
:
84
96
.
49.
Sarrió
D
,
Pérez-Mies
B
,
Hardisson
D
,
Moreno-Bueno
G
,
Suárez
A
,
Cano
A
, et al
.
Cytoplasmic localization of p120ctn and E-cadherin loss characterize lobular breast carcinoma from preinvasive to metastatic lesions
.
Oncogene
2004
;
23
:
3272
83
.
50.
Garcia-Recio
S
,
Thennavan
A
,
East
MP
,
Parker
JS
,
Cejalvo
JM
,
Garay
JP
, et al
.
FGFR4 regulates tumor subtype differentiation in luminal breast cancer and metastatic disease
.
J Clin Invest
2020
;
130
:
4871
87
.
51.
Levine
KM
,
Priedigkeit
N
,
Basudan
A
,
Tasdemir
N
,
Sikora
MJ
,
Sokol
ES
, et al
.
FGFR4 overexpression and hotspot mutations in metastatic ER+ breast cancer are enriched in the lobular subtype
.
NPJ Breast Cancer
2019
;
5
:
19
.
52.
Schackmann
RCJ
,
Van Amersfoort
M
,
Haarhuis
JHI
,
Vlug
EJ
,
Halim
VA
,
Roodhart
JML
, et al
.
Cytosolic p120-catenin regulates growth of metastatic lobular carcinoma through Rock1-mediated anoikis resistance
.
J Clin Invest
2011
;
121
:
3176
88
.
53.
Onder
TT
,
Gupta
PB
,
Mani
SA
,
Yang
J
,
Lander
ES
,
Weinberg
RA
.
Loss of E-cadherin promotes metastasis via multiple downstream transcriptional pathways
.
Cancer Res
2008
;
68
:
3645
54
.
54.
Valastyan
S
,
Weinberg
RA
.
Tumor metastasis: molecular insights and evolving paradigms
.
Cell
2011
;
147
:
275
92
.
55.
Petrova
YI
,
Schecterson
L
,
Gumbiner
BM
.
Roles for E-cadherin cell surface regulation in cancer
.
Mol Biol Cell
2016
;
27
:
3233
44
.
56.
Padmanaban
V
,
Krol
I
,
Suhail
Y
,
Szczerba
BM
,
Aceto
N
,
Bader
JS
, et al
.
E-cadherin is required for metastasis in multiple models of breast cancer
.
Nature
2019
;
573
:
439
44
.
57.
Michaut
M
,
Chin
SF
,
Majewski
I
,
Severson
TM
,
Bismeijer
T
,
de Koning
L
, et al
.
Integration of genomic, transcriptomic and proteomic data identifies two biologically distinct subtypes of invasive lobular breast cancer
.
Sci Rep
2016
;
6
:
18517
.
58.
Guvakova
MA
,
Surmacz
E
.
Overexpressed IGF-I receptors reduce estrogen growth requirements, enhance survival, and promote E-cadherin-mediated cell–cell adhesion in human breast cancer cells
.
Exp Cell Res
1997
;
231
:
149
62
.
59.
Canonici
A
,
Steelant
W
,
Rigot
V
,
Khomitch-Baud
A
,
Boutaghou-Cherid
H
,
Bruyneel
E
, et al
.
Insulin-like growth factor-I receptor, E-cadherin and alpha v integrin form a dynamic complex under the control of alpha-catenin
.
Int J Cancer
2008
;
122
:
572
82
.
60.
Bae
GY
,
Choi
SJ
,
Lee
JS
,
Jo
J
,
Lee
J
,
Kim
J
, et al
.
Loss of E-cadherin activates EGFR-MEK/ERK signaling, which promotes invasion via the ZEB1/MMP2 axis in non-small cell lung cancer
.
Oncotarget
2013
;
4
:
2512
22
.
61.
St John
MA
.
Inflammatory mediators drive metastasis and drug resistance in head and neck squamous cell carcinoma
.
Laryngoscope
2015
;
125
:
S1
11
.
62.
Jeanes
A
,
Gottardi
CJ
,
Yap
AS
.
Cadherins and cancer: how does cadherin dysfunction promote tumor progression?
Oncogene
2008
;
27
:
6920
9
.
63.
Fedor-Chaiken
M
,
Hein
PW
,
Stewart
JC
,
Brackenbury
R
,
Kinch
MS
.
E-Cadherin binding modulates EGF receptor activation
.
Cell Commun Adhes
2003
;
10
:
105
18
.
64.
Petrova
YI
,
Schecterson
L
,
Gumbiner
BM
.
Roles for E-cadherin cell surface regulation in cancer
.
Mol Biol Cell
2016
;
27
:
3233
44
.
65.
Bajrami
I
,
Marlow
R
,
van de Ven
M
,
Brough
R
,
Pemberton
HN
,
Frankum
J
, et al
.
E-Cadherin/ROS1 inhibitor synthetic lethality in breast cancer
.
Cancer Discov
2018
;
8
:
498
515
.
66.
Zardavas
D
,
Phillips
WA
,
Loi
S
.
PIK3CA mutations in breast cancer: reconciling findings from preclinical and clinical data
.
Breast Cancer Res
2014
;
16
:
201
.
67.
Wang
Q
,
Zhang
Y
,
Zhu
J
,
Zheng
H
,
Chen
S
,
Chen
L
, et al
.
IGF-1R inhibition induces MEK phosphorylation to promote survival in colon carcinomas
.
Signal Transduc Targeted Ther
2020
;
5
.
68.
Chandarlapaty
S
,
Sawai
A
,
Scaltriti
M
,
Rodrik-Outmezguine
V
,
Grbovic-Huezo
O
,
Serra
V
, et al
.
AKT inhibition relieves feedback suppression of receptor tyrosine kinase expression and activity
.
Cancer Cell
2011
;
19
:
58
71
.

Supplementary data