Expression and function of epithelial cell adhesion molecule EpCAM: where are we after 40 years?

The epithelial cell adhesion molecule (EpCAM) has first been described in 1979 as a humoral antigen expressed on colon carcinoma cells [1]. Here, we have summarized the impressive progress in the biology of EpCAM in the past four decades. Expression patterns, regulation, and the multiple functions of EpCAM in normal epithelia, in carcinoma, and in pluripotent stem cells are reviewed. Furthermore, the clinical implications and applications related to EpCAM are discussed. Lastly, the most recently discovered involvement of EpCAM in the regulation of epithelial-to-mesenchymal transition, which is of paramount importance during metastases formation and therapy resistance, is delineated.

The human EPCAM gene is encoded on the plus strand of chromosome 2p21 and consists of 9 exons covering 41.88 kilobases (kb). Exon 1 encodes the 5′-untranslated region and the signal peptide, exon 2 the EGF-like motif, exon 3 the thyroglobulin domain, exons 4–6 the cysteine-poor part of the domain, exon 7 the transmembrane domain, exon 8 parts of the intracellular domain, and exon 9 the remaining intracellular domain and the 3′-untranslated region [27]. A 1.1-kb fragment of the EPCAM promoter sufficient to drive gene expression and confers epithelial specificity was cloned [28, 29]. The promoter can be further subdivided in a gene proximal part composed of 570 base pairs (bp) and a distal part of 550 bp that act synergistically in EPCAM expression and are negatively regulated by nuclear factor kappa B (NF-κB) [29]. Sankpal et al. further described the usage of an extracellular-regulated kinase 2 (ERK2) binding site within the EPCAM promoter [30], while Yamashita et al. reported on the regulation of the EPCAM promoter by a Wnt-β-catenin-Tcf4 complex in hepatocellular carcinoma cells [31]. Furthermore, the EMT-inducing transcription factor Zeb1 represses EPCAM expression in zebrafish [32].

EpCAM is a transmembrane protein with a single membrane-spanning domain (23-aa) that connects the larger extracellular domain (265-aa) to a short intracellular domain (26-aa) (Fig. 3). The extracellular domain contains a signal peptide, an EGF-like, cysteine-rich domain, and a thyroglobulin-like domain, which was initially referred to as a second EGF-like repeat [36], followed by a cysteine-poor region [37]. Mass spectrometry and Edman sequencing of the extracellular domain of EpCAM demonstrated the cleavage of the signal peptide after aa 23, resulting in an N-terminus starting with a modified pyroglutamate [38]. Disulfide bonds were mapped to Cys²⁷–Cys⁴⁶, Cys²⁹–Cys⁵⁹, Cys³⁸–Cys⁴⁸, Cys¹¹⁰–Cys¹¹⁶, and Cys¹¹⁸–Cys¹³⁵ (Fig. 3) [38].

N-Glycosylation of EpCAM has been reported with no evidence of O-glycosylation [2]. N-Glycosylation sites have been mapped to Asn⁷⁴, Asn¹¹¹, and Asn¹⁹⁸ of EpCAM [38]. Initially complete glycosylation of Asn¹¹¹, partial glycosylation of Asn⁷⁴, and no glycosylation at Asn¹⁹⁸ were reported [38]. However, single and dual mutations of Asn⁷⁴ and Asn¹¹¹ revealed that mutations retained a certain degree of glycosylation, which was lost when all positions including Asn¹⁹⁸ were mutated [39]. Loss of glycosylation at this site resulted in severe reduction of protein stability and half-life at the membrane from 21 to 7 h [39]. Glycosylation of EpCAM may impact on EpCAM stability and expression in tumor cells, as it was found increased in cancerous versus healthy tissue [40].

Crystal structure analysis of the extracellular domain of EpCAM, termed EpEX, at 1.86 Å resolution revealed a three-partite fold in amino-terminal (puroGlu²⁴–Leu⁶²), thyroglobulin-like (Ala⁶³–Arg¹³⁸), and carboxy-terminal (Val¹³⁹–Lys²⁶⁵) domains (Fig. 3). The respective ND, TY, and CD domains are each in contact with the other two domains, forming a triangle [3]. EpEX molecules form cis-dimers with strongest interactions between the TY loop of one EpEX molecule with the βC sheet of a second molecule [3]. Coarse grain modelling of the intramembrane domain demonstrated a dimeric structure in which two helices of the membrane-spanning parts are symmetrically arranged and cross each other between Val²⁷⁶ and Val²⁸⁰ [3]. There is so far no structural information of the intracellular domain of EpCAM termed EpICD.

Initial characterization of EpCAM functions was conducted in murine fibroblasts and L153S mammary carcinoma cells, which are characterized by loss of cell adhesion and the adoption of a spindle-shaped morphology. Ectopic expression resulted in increased intercellular adhesion and cell aggregation in suspension, along with the segregation of EpCAM-positive and EpCAM-negative cells, and a reduced capacity of fibroblast to grow invasively [4, 41]. Membrane-proximal thyroglobulin-like domains, initially referred to as a second EGF-like domain, mediate lateral interactions of EpCAM in cis on one cell. Membrane-distal EGF-like repeats are required for interactions of EpCAM in trans on adjacent cells [42]. Taken together, formation of functional EpCAM tetramers as the initiating event in the formation of cell adhesion complexes was proposed [42].

Surprisingly, overexpression of EpCAM in epithelial cell lines that depend on cadherin-mediated cell-cell connections decreased adhesion by impairing functional adherens junctions through disruption of E-cadherin, α-catenin, and F-actin interactions [43, 44]. EpCAM’s capacity to inhibit cadherin-mediated adhesion in breast epithelial cells depended on phosphoinositide 3-kinase (PI3K) and its shifted interaction with N-cadherin to EpCAM [45].

Knockout of EpCAM in mice supported its role in orchestrating the structure and functionality of epithelium in the intestinal tract [46‐48]. EpCAM knockout induced an abnormal placental development associated with death in utero at gestation day E12.5 [46]. Subsequent knockouts did not reproduce an embryonic lethality; however, mice died of severe intestinal erosion and hemorrhagic diarrhea shortly after birth [44, 45]. Intestinal defects were reminiscent of human CTE where mutations cause a loss of EpCAM expression at the plasma membrane [22, 49]. Despite agreeing on the observed phenotype, Lei et al. [48] and Guerra et al. [47] disagreed on the molecular mechanisms. Lei et al. observed a loss of tight junction formation owing to substantially reduced recruitment of murine claudin 7 to tight junctions in the absence of EpCAM [48]. These findings are supported by the direct interaction of EpCAM with claudin 7 [50], which was demonstrated to support tumor progression in colorectal cancer [51, 52]. Guerra et al. demonstrated a dysregulation of E-cadherin and ß-catenin functions leading to partial disruption of adherens junctions [47]. A cooperation of E-cadherin and EpCAM in epithelial adhesion regulation is further supported by work in zebrafish, where EpCAM knockout impaired the integrity of the skin periderm through reduced cell surface levels of E-cadherin and increased levels of tight junction protein 1 (Tjp1) [53].

More discrepancies exist on EpCAM’s role in cell-cell adhesion. Gaber et al. [54] could not confirm intercellular homo-oligomers, despite formation of cis-dimers [54], leading to the conclusion that EpCAM’s role in adhesion is not assumed as a homophilic cell adhesion molecule. Furthermore, neither regulated intramembrane proteolysis of EpCAM nor EPCAM knockout in cell lines had any measurable impact on cell-matrix and cell-cell adhesion in cancer cells [33]. Hence, EpCAM’s molecular function in adhesion has not been satisfactorily resolved yet. EpCAM is undeniably involved in maintenance of epithelial integrity in various animal models and human conditions where it acts in concert with established cell adhesion molecules. Therefore, EpCAM might support cell adhesion primarily mediated by other molecules such as claudins and cadherins, and/or might preferentially play a role in adhesion of normal but not tumor cells.

EpCAM received considerable attention as a prognostic marker based on its strong expression in various carcinomas and their metastases as compared to normal epithelia of the same localization [11‐13]. EpCAM is highly and frequently expressed in the vast majority of carcinomas that have been analyzed [11, 12] and its expression in metastases frequently correlates with levels in the corresponding primary tumors (Fig. 4 and [13, 14, 21, 55]). As such, EpCAM bears potential as a prognostic and therapeutic marker in carcinomas and during cancer progression and metastasis formation. However, EpCAM’s prognostic value varies depending on the tumor entity. High expression in primary carcinomas is associated with poor prognosis in breast [56, 57], colorectal [58], prostate [59], gallbladder [60], ovarian [61], bladder [62], pancreas [63, 64], and adenoid cystic carcinomas [65]. Oppositely, high expression of EpCAM is associated with better prognosis in colonic [12], esophageal [66], renal [67, 68], gastric [69], endometrial [70], thyroid [71], and head and neck carcinomas [72]. Multiple cellular functions of EpCAM may deploy in dependency of tumor types and localization, and might differently affect single cells within tumors. High expression of EpCAM can promote sustained proliferation and tumor/metastatic growth and might thereby be associated with poor prognosis. EpCAM also represents a marker of epithelial differentiation and might therefore associate with a more differentiated, less migratory/invasive phenotypes, with reduced resistance to irradiation and chemotherapy, and hence with better survival. Therefore, EpCAM’s association with differential clinical outcome is complex and might vary depending on the origin of the tumor or even the stage of tumor progression. This unsolved discrepancy in the prognostic value of EpCAM expression remains poorly understood and requires further investigation.

Another essential aspect is that cancer-related lethality is primarily caused by therapy-resistant cells and frequently untreatable metastases. Numerous studies on EpCAM’s prognostic value have initially concentrated on primary tumors rather than disseminated cells [12, 57‐65, 67‐73]. However, primary tumor biopsies represent a specific region of the tumor at a given time point. Based on growing evidence of intratumor heterogeneity and temporal variation in tumor antigen expression [74‐80], a singular measurement of primary tumors might suffer from bias. Consequently, numbers of disseminated tumor cells, which are considered primary sources of relapse and metastases, have been implemented as additional prognostic marker [81‐83]. Here, tumor cells found in the blood are termed circulating tumor cells (CTCs) and tumor cells in distant organs, for example, in the bone marrow, are termed disseminated tumor cells (DTCs). Clinical assessment of CTCs in the blood is minimally invasive and allows for longitudinal measurements [82, 84] and aims at the prediction of metastasis formation [19]. Measurement of CTCs in the blood of metastatic breast cancer (MCB) patients was performed using EpCAM-specific antibodies for the enrichment of rare circulating tumor cells, which were subsequently further characterized by the expression of cytokeratins, presence of a nucleus, and lack of white blood cell marker CD45. CTC numbers in the blood showed prognostic value and patients with five CTCs or more per 7.5 mL blood had decreased overall and progression-free survival [18]. More recently, 3173 patients with non-metastatic (stages I–III) breast cancer were analyzed and a CTC threshold of ≥ 1 cell per 7.5 mL of blood correlated with decreased OS, disease-free survival, distant disease-free survival, and cancer-specific survival [85]. The prognostic value of EpCAM-positive CTCs has been confirmed for further tumor entities, including lung cancer [86], advanced ovarian cancer [87], gastric cancer [88], colorectal cancer [89], and head and neck squamous cell carcinoma (HNSCC) [90]. Hence, the enrichment of rare EpCAM-positive systemic tumor cells in the blood in the frame of liquid biopsies has the potential to evaluate disease outcome including the patients’ burden of metastatic cells [20, 21, 85].

EpCAM is highly and frequently expressed in the vast majority of carcinomas, in tumor-initiating cells, and in disseminated tumor cells, which qualified EpCAM as a potential target for cancer therapies [15, 91, 92]. Several publications demonstrated that EpCAM-specific antibodies can eliminate cancer cells by various mechanisms [93‐102] and detect DTC in the bone marrow of patients [103, 104]. Monoclonal antibody 17-1A was clinically tested alone and in combination with γ-interferon for the treatment of gastrointestinal and metastasized colorectal carcinomas [105, 106]. Induction of complete remissions by mAb 17-1A were reported in metastasized colorectal cancer [107, 108]. Positive overall survival data with mAb 17-1A in a pivotal study with metastasized colorectal cancer patients [96, 99] resulted in approval of the antibody in Germany under the trade name Panorex®. However, when compared to standard care with 5-fluorouracil-based chemotherapy in a later trial [109], edrecolomab was found inferior, leading to its market withdrawal. EpCAM-specific humanized antibodies ING-1 and 3622W94, bearing higher binding affinity than edrecolomab, were tested in clinical trials but were discontinued because of low tolerability. While edrecolomab might have suffered from an insufficient binding affinity, ING-1 and 3622W94 displayed too high affinity that no longer allowed to distinguish normal and malignant cells, leading to pancreatitis [110, 111].

Micromet Inc. developed fully human IgG1 EpCAM-specific antibody MT201 (adecatumumab) that had an intermediate binding affinity for EpCAM in an attempt to improve therapeutic efficacy [112]. Adecatumumab was investigated as monotherapy in a phase II clinical trial in MBC patients. Adecatumumab did not induce measurable tumor regression; however, patients treated with high-dose antibody and expressing high levels of EpCAM retrospectively showed reduced development of new metastases (3/18 versus 14/29 patients) [113]. In a dose-escalating phase I trial enrolling prostate cancer patients, reduction of prostate-specific antigen levels was observed [114]. While having a favorable safety profile, the antibody program is no longer pursued because it did not reach its clinical endpoints.

EpCAM was also targeted with T cell-engaging antibodies designed to connect cytotoxic T cells with cancer cells for redirected lysis. Antibody catumaxomab (Removab®) developed by Trion Pharma has an EpCAM- and a CD3-binding arm and its Fc domain binds Fc gamma-receptor expressing antigen-presenting cells [115]. Catumaxomab was approved by the European Medicines Agency (EMA) to treat malignant ascites [116], but was eventually terminated. MT110 (solitomab) is a tandem single-chain antibody construct comprised of an EpCAM- and CD3-specific binding domain developed by Micromet Inc. [117]. Pharmacokinetics, tolerability, safety, and anti-tumor activity of MT110 were assessed in a dose-escalating phase I clinical trial in metastatic colorectal, gastric, and lung carcinomas (129) with disease stabilization in 7/19 patients. Due to gastrointestinal toxicities, solitomab could not be escalated to more efficacious dose levels, and the program was abandoned.

Recently, EpCAM-specific antibody EpAb2-6 was developed, which binds to an epitope within the thyroglobulin domain and demonstrated inhibitory potential [118]. EpAb2-6 induces apoptosis, inhibits tumor growth in mouse models, and represses EpCAM cleavage to form the signaling moiety EpICD in pancreatic and colon carcinoma cells [118, 119]. Thus, EpAb2-6 is an anti-EpCAM antibody that inhibits central EpCAM signaling functions, which might represent a novel approach to treatment methods targeting EpCAM in carcinomas. Furthermore, EpCAM-specific antibodies were conjugated with bouganin (citatuzumab bogatox) [120], Pseudomonas exotoxin A (oportuzumab monatox) [121] and alpha-amanitin [122], with interleukin-2 to activate T cells in the vicinity of EpCAM-positive tumor cells (huKS-IL2) [123], and with encapsulated inhibitory RNAs [124‐126]. In 2015, a phase I clinical trial of patients with EpCAM-positive metastatic cancers aimed at evaluating the maximum tolerated doses, pharmacokinetics, and immunogenicity of the anti-EpCAM-based immunotoxin MOC31PE that is coupled to Pseudomonas exotoxin A. MOC31PE was intravenously combined with the immunosuppressant cyclosporin in n = 63 metastatic patients and revealed a safe profile with a half-life in plasma of 3 h and a reduction of [127]. Subsequently, MOC31PE was applied to address peritoneal metastasis in colorectal and ovarian cancers within the ImmunoPeCa phase I/II. Based on the promising cytotoxicity profile and a low systemic uptake, MOC31PE is currently further evaluated in a clinical phase II study [128, 129].

Further clinical trials including the usage of chimeric antigen receptor T cells (CAR T cells) targeting EpCAM are ongoing in order to address their potential for the treatment of recurrent and treatment-resistant solid tumors (https://​clinicaltrials.​gov/​ct2/​show/​NCT04151186). Currently, a total of 64 clinical trials are listed by the US national library of medicine that involve EpCAM as a biological using CAR T cells, monoclonal antibodies, immunotoxins and immunocytokines, and EpCAM-based capture methods to enrich CTCs (https://​clinicaltrials.​gov/​ct2/​results?​cond=​&​term=​EpCAM&​cntry=​&​state=​&​city=​&​dist=​). While EpCAM expression on most carcinoma is a highly attractive feature for tumor-associated antigens, expression on healthy epithelia—mostly of the gastrointestinal tract—will limit the therapeutic window of EpCAM-targeted therapies and call for potential side effects. A next generation of EpCAM-targeted drugs that are selectively activated in the tumor microenvironment may finally allow to leverage this target antigen [130].

Finally, EpCAM serves as a target for image-guided surgery approaches that aim at improving resection margins. Residual tumor cells following incomplete tumor removal at the resection margin are a major source of local recurrence in solid tumors [131]. The employed strategies include the systemic application of EpCAM-specific antibodies labeled with fluorescent dyes such as fluorescein isothiocyanate (FITC) and near-infrared fluorescence dye (NIRF) IRDye800CW [132, 133]. Fluorescence-labeled anti-EpCAM antibodies allow the detection of residual tumor nodules in the millimeter size range using intraoperative imaging systems [133, 134].

In the future, we expect further insight in the role of EpCAM in the regulation of cell fate in health and disease, which would eventually revive its usage as therapeutic target. The discovery of RIP of EpCAM and the generation of the signaling-active fragments EpEX and EpICD paved the way for a novel class of EpCAM inhibitors that target signaling in cancer and the metastatic cascade. Antagonizing antibodies such as EpAb2-6 and small molecule inhibitors that compete with EpCAM RIP or with EpICD functions could develop into promising candidate drugs for the treatment of EpCAM-positive carcinomas and the formation of lethal metastases.

EpCAM will retain a central role as of anchor molecule in the enrichment of CTCs that harbor metastatic potential in advanced cancer patients. Here, EpCAM has dual potential as a quantifier of systemic tumor cells in the frame of liquid biopsies and as a potential target for adjuvant therapy of residual tumor cells with specific biologicals.

Clinical studies on the predictive potential of EpCAM for carcinoma entities associated with a high degree of treatment resistance are required for the use of EpCAM as biomarker in clinical decision-making. For example, prospective studies on the expression of EGFR and EpCAM might shed light on the power of EpCAM expression to predict treatment response in carcinoma entities that implement a therapy with the anti-EGFR monoclonal antibody cetuximab or small molecule inhibitors of EGFR.

In the future, the role of EpCAM in pluripotent stem cells might gain momentum. Especially, the implication of EpCAM and EpICD in the generation of iPS is of relevance. Additionally, the use of EpEX in combination with the reprogramming factors Klf4 or Oct3/4 was instrumental in the generation of iPS and might therefore experience (pre-)clinical application.

Lastly, it can be anticipated that basic research on the expression dynamics and the molecular functions of EpCAM in cancer and healthy cells will further support the abovementioned clinical applications of this highly versatile molecule.

Over the past four decades, EpCAM has evolved from a humoral antigen expressed on the majority of carcinoma cells to a complex signaling molecule involved in central aspects of cell fate such as proliferation, pluripotency, differentiation, and organ integrity. Signaling by intact EpCAM and the proteolytic fragments EpEX and EpICD as well as signaling-independent functions of EpCAM serve these various purposes and represent novel handles to tackle metastatic malignancies. Furthermore, EpCAM remains an attractive target for antibody-based cancer therapies and as a biomarker for patient stratification. Importantly, EpCAM is the central target molecule for the enrichment and characterization of systemic tumor cells with prognostic and metastatic potential. Eventually, EpCAM became a multifaceted protein with a long-standing history in molecular oncology and in stem cell biology.