EpCAM+ Liver Cancer Stem‐Like Cells Exhibiting Autocrine Wnt Signaling Potentially Originate in Cirrhotic Patients

Abstract Hepatocellular carcinoma (HCC) is believed to originate from cancer stem cells (CSCs). While epithelial cell adhesion molecule (EpCAM) is a marker of normal hepatic stem cells (HSCs), EpCAM+ cells from HCC behave like CSCs. Since HCC mostly develops on a cirrhotic background, we sought to determine whether CSC‐like EpCAM+ cells exist in patients with advanced cirrhosis. Both flow cytometry and immunohistochemistry showed that frequency of EpCAM+ cells in advanced cirrhosis was increased as compared to control. To determine whether increased EpCAM population in advanced cirrhosis harbors any CSC‐like cells, we compared molecular and functional features of EpCAM+ cells from advanced cirrhosis (Ep+CIR; n = 20) with EpCAM+ cells from both HCC (Ep+HCC; n = 20) and noncancerous/noncirrhotic (control) (Ep+NSC; n = 7) liver tissues. Ep+CIRs displayed similarity with Ep+HCC cells including upregulated expression of stemness and Notch pathway genes, enhanced self‐renewal in serial spheroid assay and generation of subcutaneous tumors in nonobese diabetic/severe combined immunodeficiency mice. Moreover, transcriptome and miRNome of Ep+CIRs appeared closer to that of Ep+HCC cells than Ep+NSCs. Interestingly, more than 50% micro RNAs (miRNAs) and transcripts specifically expressed in Ep+HCCs were also expressed in Ep+CIRs. However, none of Ep+NSC specific miRNAs and only 7% Ep+NSC specific transcripts were expressed in Ep+CIRs. Further, according to gene expression and in vitro Wnt inhibition analysis, autocrine Wnt signaling appeared to be a distinct feature of Ep+CIR and Ep+HCC cells, which was absent from Ep+NSCs. EpCAM+ cells in advanced cirrhosis possibly include a population of CSC‐like cells which can be explored for early diagnosis of HCC development. Stem Cells Translational Medicine 2017;6:807–818


INTRODUCTION
Hepatocellular carcinoma (HCC) is the fifth most common cancer and second leading cause of cancer related deaths globally [1]. Despite therapeutic advancements in the treatment of HCC survival rates are poor due to presentation of the disease at advanced stage [2]. Recent studies indicate that HCC arises from liver cancer stem cells (CSCs) leading to tumor heterogeneity [3]. The origin of HCC in over 80% cases is in the milieu of chronic liver injury, mainly cirrhosis [4]. In cirrhosis, hepatocytes lose their replicative potential, resulting in the activation of normal hepatic stem cells (HSCs) [5]. Active division in presence of increased reactive oxygen species and inflammation, makes HSCs amenable to incorporate mutations and convert into CSCs, which eventually can lead to the development of HCC [6]. Thus, identification of CSCs can be of help in early diagnosis of HCC development in advanced cirrhotic patients.
Intriguingly, surface markers like epithelial cell adhesion molecule (EpCAM), CD133, CD90, CD13, and OV6 are used to identify HSCs as well as CSCs, making it difficult to differentiate between them [7]. Of these markers, EpCAM (MUC1) is an adhesion molecule commonly used to identify normal HSCs. It is present in developing as well as adult livers and is absent from the mature hepatocytes [8,9]. However, in HCC, EpCAM expression correlates with poor patient survival [10]. Further, EpCAM1 cells isolated from alphafetoprotein positive (AFP1) HCC tissues have been demonstrated to behave like CSCs [11,12]. Accordingly, we chose EpCAM as a marker for identification of normal stem cells (NSCs) from control and CSCs from AFP1 HCC liver tissues.
While existence of CSCs has been well demonstrated in HCC, their characterization in cirrhotic patients remains elusive, with only few reports addressing this concept [13,14]. We observed an increase in the number of EpCAM1 cells during advanced cirrhosis, warranting further studies to delineate if they have any similarity with CSCs. Since normal HSCs and CSCs express common surface markers, only certain functional characteristics can help in differentiating between them. For instance, Notch and Wnt signaling pathways are involved in the self-renewal and differentiation of normal HSCs, but, their constitutive activation has been implicated in CSCs [15][16][17]. In the present study, we therefore compared molecular and functional features of EpCAM1 cells from advanced cirrhosis, HCC and paired adjacent noncancerous/noncirrhotic (control) liver tissues along with EpCAM-cells from the same HCC tissues (Ep-HCC). Functional comparison was performed by checking the in vitro spheroid formation in nonadherent culture conditions and in vivo tumor formation in nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice. Next generation sequencing (NGS) was used to compare micro RNA (miRNA) as well as mRNA profiling in all the cell groups. NGS revealed that Wnt signaling was regulated by genes and miRNAs commonly expressed by EpCAM1 cells of both advanced cirrhosis and HCC. Owing to its well-known role in CSCs, we validated the key genes of the Wnt pathway in a larger cohort. Further, the effect of Wnt inhibitor was evaluated to determine whether any cell group is exhibiting autocrine Wnt signaling.

Patient Samples
Liver tissues were obtained from patients undergoing Living Donor Liver Transplant (LDLT) for advanced cirrhosis (n 5 20), and LDLT/resection for HCC (n 5 20), at the Institute of Liver and Biliary Sciences (New Delhi, India) during 2012-2014 years. Histologically normal appearing liver tissues (noncancerous/noncirrhotic region adjacent to tumor portion) were also taken from patients having HCC on a noncirrhotic background and served as controls (n 5 7). Patients between 18-80 years of age were enrolled in the study. Only HCC tissues that were AFP1 were included. Patients with alcohol, nonalcoholic steatohepatitis, hepatitis B virus (HBV) and hepatitis C virus (HCV) related HCC or advanced cirrhosis were included, while those having acute liver failure, acute-onchronic liver failure or any viral coinfections including Human Immunodeficiency Virus (HIV)-HBV, HIV-HCV and HBV-HCV were excluded from the study. This study was approved by the Institutional Ethics Committee and informed consent was obtained from each patient or their close relatives for inclusion in the study. About 35-40 gm of liver tissues from the patients were processed and subjected to different analyses as detailed in Supporting Information Figure 1.

Immunohistochemistry
Histopathological analyses were performed on 5 mm thick formalin fixed paraffin embedded liver tissue sections using standard pathology laboratory protocols. In brief, sections were deparaffinized with xylene and rehydrated through graded alcohol (100, 90, 80, and 70%). Sections were then stained with Hematoxylin and eosin (H&E). Reticulin staining was performed to distinguish HCC from cirrhosis as per the standard protocol [18].
For immunohistochemistry (IHC), after rehydration, sections were treated with 3% H 2 O 2 followed by antigen retrieval with sodium citrate buffer (pH 5 6.0) and blocking with 20% goat serum. Primary anti-human EpCAM, Hepatocyte Specific Antigen (HSA) or Glypican-3 antibodies were used at a dilution of 1:250, 1:500, and 1:500, respectively, for staining (Santa Cruz Biotechnologies, CA, USA, https://www.scbt.com). CD133 and CD90 antibodies were procured from Abnova, Taiwan, http://www.abnova.com and used at a dilution of 1:500 and 1:200, respectively. Detection was performed using SuperSensitive Polymer-HRP IHC Detection System/DAB large volume kit (BioGenex, CA, USA, https://store. biogenex.com), in accordance with the manufacturer's instructions. The sections were counterstained with hematoxylin to demonstrate the nuclei and were observed under the microscope. Semiquantification of EpCAM expression was defined by Quick (Q) scoring system. Q scores were calculated by multiplying the percentage of positive cells with staining intensity (0 5 no, 1 5 weak, 2 5 moderate, and 3 5 strong staining) to yield a score ranging from 0 to 300.

Enrichment of EpCAM1 Cells Using Magnet Based Cell Separation
The cell isolation procedure from liver tissues was performed by magnetic based assay with some modifications [19]. In brief, tissue explants were washed with phosphate buffer saline (PBS) and chopped into small pieces. After digestion with 0.05% type IV collagenase (Worthington Biochemical Corporation, NJ, USA, http:// www.worthington-biochem.com) for 30 minutes at 378C, tissues were again minced and the cell suspension was passed through a 70-mm cell strainer (BD Biosciences, NJ, USA, https://www.bdbiosciences.com) to remove tissue debris. Red blood cells (RBCs) were lysed by 1x RBC lysis buffer (Stem Cell Technologies, Vancouver, Canada, https://www.stemcell.com) and the cells were washed again with PBS. An aliquot of the single cell suspension was used to determine the percentage of EpCAM1 cells by flow cytometry. Viable and nonviable cells were counted by trypan blue staining and then used to isolate EpCAM1 cells by magnet assisted cell sorting. In brief, depending on the cell number, they were labeled with appropriate amount of EpCAM1 selection cocktail using EpCAM1 Cell Isolation Kit (Stem Cell Technologies, Vancouver, Canada), according to manufacturer's instructions. Purity of EpCAM1 and EpCAM-sorted cells were evaluated by flow cytometry.
Quantitative reverse-transcription polymerase chain (qRT-PCR) reaction was performed with a SYBR Green PCR Kit (Applied Biosystems, DE, USA, https://www.thermofisher.com/in/en/ home/brands/applied-biosystems.html) using an ABI PRISM 7700 Sequence Detector and ViiA 7 software (Applied Biosystems, DE, USA). The primers of selected genes were designed using Primer 3 software (Supporting Information Table 1). The gene expression level was normalized against 18S RNA (control gene). Subsequently, the relative gene expression values were determined using log of 2 2DDCT .

Spheroid Assay
Sorted cells obtained after magnetic based cell separation were suspended in 0.9% methyl cellulose in knockout Dulbecco's modified Eagle medium/nutrient mixture F-12 (DMEM/F12) media, containing 10% knockout serum replacement, 1% Penicillin-Streptomycin-Amphotericin B, 20 ng/ml human recombinant Epidermal Growth Factor and 10 ng/ml human recombinant Fibroblast Growth Factor (Life Technologies, CA, USA, https://www. thermofisher.com/in/en/home/brands/invitrogen.html) and seeded in a six-well ultralow attachment plate (Corning Inc., Corning, NY, USA, www.corning.com) at a density of about 5,000 cells per well. Plates were incubated at 378C and 5% CO 2 , for 12-14 days, during which fresh media was added every third day. When most of the spheroids attained at least 50 mm diameter, they were washed with PBS, trypsinized and repassaged under same culture conditions. Spheroids were maintained for a total of six passages. After each passage the size and number of spheroids was recorded.

Animal Studies
Twenty-five thousand or 50,000 freshly isolated EpCAM1 cells were resuspended in 200 ml of a 1:1 mixture of DMEM/F12:Matrigel (BD Biosciences, NJ, USA) and subcutaneously injected into 6 to 8 weeks old male nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice (The Jackson Laboratories, CA, USA, https://www.jax.org/). For each cell type (Ep1NSC, Ep1CIR, Ep1HCC, and Ep-HCC), three mice were injected with three different patient samples. Formation of tumor was observed till 8-10 weeks after injection of cells. Tumor volume was calculated by using the formula 1 =2[(largest diameter) 2 3 smallest diameter]. The experimental protocol was approved by Institutional Animal Ethics Committee (IAEC 298/12), National Institute of Immunology, India. All animals were treated with humane care. They were kept in individual ventilated cages and fed with autoclaved acidified water and irradiated food ad libitum.

Next Generation Sequencing for Small RNA and mRNA
Detailed protocol is given in Patients and Methods section of Supporting Information annexure.

Wnt Inhibitor Assay
EpCAM1 cells were plated at a density of 10,000 cells per well in ultra-low attachment plates with IWP12 (Inhibitor of Wnt production compound 12) (Sigma-Aldrich, Tokyo, Japan, www.sigmaaldrich.com/) at a concentration of 5 mM per well. Spheroid formation was compared in terms of both size and number with corresponding cells plated without the inhibitor.

Statistical Analysis
Data were analyzed using the statistical software Prism (version 5; GraphPad Software) and are reported as mean 6 standard error mean (SEM). For analysis of the clinical data, Kruskal-Wallis Test followed by probability adjustment by Mann-Whitney test was performed. Here p .017 was considered significant. For other experiments the unpaired Student's t test was performed to compare two mean values and p values <.05 were considered statistically significant.

Clinicopathological Features of Patients
The demographic profile of all the subjects enrolled in this study is shown in Table 1. Out of total patients with advanced cirrhosis and/or HCC, 20% were affected due to excessive alcohol intake and another 20% with HBV infection. Only one patient had HCV related HCC. Remaining 57% patients had unknown etiology. Male to female ratio in patients with advanced cirrhosis or HCC was same. There was no significant difference in the AFP levels of cirrhotic and HCC patients due to high standard deviation. Out of 20 cirrhotic patients, only one patient (5%) had AFP above the threshold value of 10 ng/ml. In case of HCC as well, only 20% had AFP values above the threshold. AST and ALT levels in both the groups were also similar. The serum bilirubin and International normalized ratio (INR) levels were higher in cirrhotic patients as compared to HCC patients and so were the MELD scores.

Histopathological Characterization of the Tissue Specimen
Histologically normal, cirrhotic and HCC liver tissues used in the present study were initially validated by a pathologist. HCC occurred in the background of cirrhosis or in the absence of cirrhosis as depicted in the gross pictures (Supporting Information Fig. 2A). Histologically normal liver tissues obtained from noncancerous/noncirrhotic region adjacent to tumor were classified on the basis of no significant appearance of fibrosis and single chord arrangement of benign hepatocytes in HE sections (Supporting Information Fig. 2B). Cirrhotic tissues were composed of regenerative parenchyma (Supporting Information Fig. 2C). HE staining of HCC specimen showed the characteristic trabecular and pseudoglandular architecture (Supporting Information Fig. 2D). Malignant hepatocytes were polygonal cells displaying varying degree of nuclear enlargement and atypia. Reticulin staining is a silver impregnation technique used to delineate reticulin fibers and is therefore useful in distinguishing HCC (which is reticulin deficient due to thick chord arrangement of cells) from cirrhosis. Reticulin staining revealed a reticulin poor HCC in Khosla, Rastogi, Ramakrishna et al.  Fig. 2E). IHC for HCC specific surface markers HSA and Glypican-3 further highlighted the tumor (Supporting Information Fig. 2F). Since, EpCAM1 cells from AFP1 HCC tissues have been shown to behave like CSCs [11], for subsequent experiments we used HCC tissues which showed AFP positivity in IHC (Supporting Information Fig. 2G).

Increased Expression of EpCAM in Cirrhotic Liver Tissues
We compared the expression of some of the most commonly used surface markers for CSC identification namely EpCAM, CD133, and CD90 in the respective tissue sections by IHC (Fig. 1A). A very weak expression of CD133 was observed in the control and cirrhotic liver sections confined to the hepatic progenitor cells. Within the HCC sections as well, weak to moderate cytoplasmic staining of CD133 was observed. In accordance with the published reports [20], CD90 expression was mainly detected in inflammatory and endothelial cells. There was not much difference in the level of CD90 expression in control and cirrhotic liver tissue. The HCC section showing maximum expression of CD90 is represented in the figure, although most of the other sections showed weak or no expression. The expression of EpCAM was detected mainly in bile ductules of control liver tissues and was completely absent from mature hepatocytes. IHC in cirrhosis revealed that EpCAM expression was increased particularly near the periportal areas in the bile ductules, oval cells, and intermediate hepatocytes. HCC tissues showed strong membranous staining of EpCAM. Flow cytometric analysis also showed an increase in number of EpCAM1 cells in advanced cirrhosis as compared to histologically normal tissue specimens (Fig. 1B). However, quantitation of both flow cytometric and IHC data demonstrated that the frequency of EpCAM1 cells was highest in HCC tissues (Fig. 1C). These results clearly indicate an increase in the EpCAM expression during progression toward HCC. However, to establish the presence of CSCs in the increased EpCAM population of cirrhotic liver tissues, their molecular and functional characterization was performed as described in the following sections.

EpCAM1 Cells from Cirrhotic Livers Display Enhanced Self-Renewal
Deregulation of various signaling pathways has been implicated in maintaining self-renewal of CSCs [21]. For instance, Notch signaling promotes proliferation of CSCs. Further, Nanog, Oct4, and Sox2 are key regulators of stemness involved in carcinogenesis [22][23][24]. Interestingly, these stemness genes are direct targets of EpCAM [25]. We therefore wanted to compare the level of expression of these genes in EpCAM1 cells isolated from control (Ep1NSC), advanced cirrhosis (Ep1CIR) and AFP1 HCC tissues (Ep1HCC). The purity of the sorted EpCAM1 cells was between 85 and 90% (Supporting Information Fig. 3). Expressions of Nanog, Oct4, and Sox2 and Notch signaling genes (Notch 1, 2, 3, Jag1, and Hey2) were significantly higher (p < .05) in Ep1CIRs and Ep1HCCs in comparison to Ep1NSC ( Fig. 2A, 2B, Supporting Information Table 2). However, there was no significant difference in the expression of these genes between Ep1CIRs and Ep1HCCs. Further, Notch 4 showed no significant difference in the expression between all the three study groups. To monitor the validity of sorted Ep1HCCs, the expression of all the above genes was also analyzed in Ep-HCC cells. All the genes showed decreased expression in Ep-HCCs as compared to Ep1HCC cells.
Self-renewal of CSCs can be established in vitro by serial spheroid assay [26]. We checked the spheroid forming ability of EpCAM1 cells by growing them in nonadherent serum free culture. A striking difference was observed in the size of the spheroids formed by Ep1NSC, Ep1CIR, and Ep1HCC cells (Fig. 2C). Violin plot clearly demonstrated that Ep1CIR spheroids were varied in size (40-126 mm). While most of them were smaller in size like Ep1NSC (40-68 mm), a small proportion were bigger like Ep1HCC (87-138 mm) spheroids (Fig. 2D). This observation affirms the existence of a heterogeneous population of EpCAM1 cells within advanced cirrhotic liver tissues. Flow cytometric analysis revealed the enrichment of EpCAM1 cells within the spheroids, thereby demonstrating their clonal expansion (Fig. 2E).
The spheroids were cultured for six passages. Number of spheroids formed by EpCAM1 cells isolated from both HCC and advanced cirrhosis increased exponentially at every passage. On the contrary, number of spheroids from Ep1NSCs increased till third passage following which it attained a stationary phase (Fig.  2F). Taken together, these results suggest that EpCAM1 cells from advanced cirrhosis have limitless self-renewal capacity unlike Ep1NSCs.

Ep1CIR Cells Possess Tumor Forming Ability
The gold standard for testing CSCs is their ability to generate tumors upon injection into mice [27,28]. We subcutaneously injected 25,000 and 50,000 Ep1NSC, Ep1CIR, Ep1HCC, and also Ep-HCC cells in flanks of NOD/SCID mice. Ep1HCC cells formed tumors in all the three mice upon transplantation of 25,000 cells/ mice (Fig. 3A). Ep1CIR cells also generated subcutaneous tumors albeit when higher cell number was used, while no tumor formation was observed with 25,000 cells (Fig. 3A, 3B). Ep1NSCs did not generate any tumor even with 50,000 cells/mice whereas EpCAM-HCC cells formed a very small tumor in one out of three mice injected with 50,000 cells. Tumor nodules formed by each cell type were confirmed by histopathological examination. Representative images shown in Figure 3C reveal characteristic histomorphological features of HCC with presence of thick trabeculae surrounded by sinusoidal cell along with focal pseudoglandular arrangement of malignant hepatocytes. These cells display enlarged nucleus with high nucleocytoplasmic ratio, conspicuous nucleoli and moderate to abundant eosinophilic cytoplasm. All tumor sections showed strong membranous immunostaining for EpCAM. The low frequency of tumor formation with cirrhotic EpCAM1 cells cannot be fully explained by number of injected cells alone. It may be affected by experimental artefacts like viability and purity of the injected cells. Further, the percentage of CSCs within the total EpCAM1 cell population from cirrhotic liver is expected to be very low and might vary from patient to patient. Nevertheless, it is reasonable to conclude from this result that EpCAM1 cells from advanced cirrhosis include a subset of CSCs.

Ep1CIR Cells Carry Ep1HCC Cell Like Signatures in Advanced Cirrhotic Livers
In order to decipher the similarity between Ep1HCC and Ep1CIR at the molecular level, NGS was performed in Ep1HCC, Ep1CIR, Ep1NSC, and Ep-HCC cell populations. Deep sequencing of transcripts and small RNAs resulted in an average of 15 million reads per sample for small RNome and 72 million reads per sample for transcriptome (Supporting Information Fig. 4A, 4B). Unique Tag distribution of small RNAs in Ep1NSCs was significantly higher as compared to the other three groups of cells (Supporting Information Fig. 4C). A total of 24,196 transcripts and 389 miRNAs were detected to be expressed collectively in the four samples (as per the criteria mentioned in the methods).

Ep1HCC Cell Specific miRNA Signatures Seen in Ep1CIR Cells.
Unsupervised hierarchical clustering of the differentially expressed miRNAs clearly showed that Ep1CIRs and EP1HCC cells clustered together while Ep-HCC and Ep1NSC miRNAs did not largely overlap with Ep1CIR (Fig. 4A).
Ep1HCC Specific Gene Signature in Advanced Cirrhosis. In parallel with the deep sequencing based miRNA studies, we also investigated the global changes in gene expression patterns associated with Ep1NSC, Ep1HCC, Ep1CIR, and Ep-HCC cell populations. Unsupervised hierarchical clustering of the differentially expressed transcripts reiterated the observation that Ep1CIR profile was closer to that of Ep1HCC as compared to Ep1NSC and Ep-HCC cells, indicating the presence of dominant molecular signature of CSCs during advanced cirrhosis (Fig. 4D).
Venn diagram depicting the comparison of Ep1HCC with Ep1NSC and Ep-HCC, and Ep1CIR with Ep1NSC showed that 1,130 upregulated and 40 down regulated genes in Ep1HCCs were also similarly differentially expressed in Ep1CIRs (Fig. 4E). A significant number of transcripts (2082) were specifically expressed only in Ep1CIR and Ep1HCC ( Fig. 4F; Supporting Information Table 4), which together with 11 Ep1HCC and Ep1CIR specific miRNAs, could be a representative signature of Ep1HCCs in the cirrhotic liver. On the contrary, very few transcripts (332) were expressed exclusively in both Ep1NSC and Ep1CIR cells.
Taken together, the results clearly indicate that Ep1CIRs are similar to Ep1HCCs and are distinct from Ep1NSC cells at the molecular level.
Pathways Regulated by miRNAs/Transcripts Commonly Expressed by Ep1CIR and Ep1HCC Cells. miRNAs and genes that were (a) specifically expressed only in Ep1HCC and Ep1CIR, (b) commonly upregulated in Ep1HCC and Ep1CIR, and (c) commonly downregulated in Ep1HCC and Ep1CIR, were subjected to significant gene ontology and pathway analysis to identify important biological events regulated by them (Fig. 5A). Wnt, Notch, mitogen-activated protein kinase, and c-Jun N-terminal kinase (JNK) were some of the important pathways regulated by commonly expressed miRNAs and transcripts of Ep1CIR and Ep1HCCs (Fig. 5B). Thorough literature survey also identified same molecular pathways involved in regulation of CSCs.
EpCAM1 Cells of Advanced Cirrhosis Patients Exhibit Autocrine Wnt Signaling. Among the commonly expressed pathways, Wnt signaling was of particular interest due to its role in stem cell self-renewal and cancer development. Further, Yamashita et al. recently reported that Wnt/b-catenin signaling augments self-renewal and inhibits the differentiation of liver CSCs by the expression of the stem cell marker EpCAM [11]. Therefore, we chose to validate the expression of Wnt pathway genes in EpCAM1 cells from control, advanced cirrhotic and HCC tissues.
Signaling molecules of Wnt pathway were >2-fold upregulated in Ep1CIR and Ep1HCC compared to Ep1NSC cells (Fig. 6A), particularly Wnt3, which is a ligand in canonical Wnt signaling. It is reported that the secretion of Wnt proteins require Evi/Wls (Evenness interrupted/Wntless), a multipass transmembrane protein. It acts as a cargo receptor for Wnt proteins, shuttling them from the Golgi to the plasma membrane where they act in an autocrine or paracrine manner to activate Wnt signaling pathways [29]. Therefore, we also checked the expression of Evi/Wls in the three study groups. Although highest in HCC, the expression of Evi/Wls was significantly increased in EpCAM1 cells from advanced cirrhotic patients than controls (Fig. 6A), indicating the existence of autocrine Wnt signaling in them. In addition, a significant increase in expression of b-catenin and two of its well-known targets c-Myc and Cyclin D1, was also noted in both Ep1CIR and Ep1HCC cells.
Chen et al. in 2009 reported that IWP compounds act on porcupine, a member of the membrane-bound O-acyltransferase (MBOAT) family and inhibit palmitoylation of Wnt proteins, which is essential for their secretion and signaling ability [30]. Therefore, we cultured the EpCAM1 cells from advanced cirrhosis and HCC patients in ultra-low attachment plates with and without the Wnt inhibitor, IWP12. The number and size of spheroids formed in the presence of IWP12 was significantly reduced with Ep1CIR and Ep1HCC cells when compared to corresponding EpCAM1 cells grown without the inhibitor (Fig. 6B, 6C). The number of spheroids formed was reduced by 44% and 54%, respectively. The responsiveness to Wnt inhibition confirmed that both Ep1CIR and Ep1HCCs exhibit autocrine Wnt signaling unlike the Ep1NSCs. Cumulatively, the observation that Ep1CIR cells showed increased expression of Evi/Wls and Wnt3 and were even responsive to Wnt inhibitor, IWP12, re-emphasized their molecular similarity to Ep1HCCs.

DISCUSSION
HCC frequently occurs in the background of cirrhosis and is believed to originate from CSCs [31]. Cirrhosis is marked by regenerative nodules resulting from localized proliferation of hepatocytes in response to liver injury. These nodules can eventually progress to become dysplastic nodules or HCC [32]. In chronic liver injury, ductular reactions become prominent due to suppressed hepatocyte proliferation, leading to the activation of stem cell response [33,34]. EpCAM1 cells have been reported as immediate progeny of the HSCs, formed due to ductular reactions [35]. In accordance with these reports, we too observed increased expression of EpCAM1 cells near the ductular regions of advanced cirrhosis liver sections. An increase in the EpCAM expression together with increased rate of cellular proliferation could possibly dictate the fate toward transformation into CSCs. Existence of CSCs during cirrhosis was earlier addressed by Sun et al. However, they used the complete cirrhotic tissue to show its tumorigenicity and did not characterize it at cellular level [13]. More recently in a mouse model, a collagenase resistant cell population was identified within cirrhotic livers which could generate tumors in mice with chronic damage and compensatory proliferation in their liver. These cells were shown to overexpress EpCAM and were called as HCC progenitor cells [14]. The present study, to our knowledge, is the first to report the existence of EpCAM1 cancer stem-like cells in advanced cirrhotic patients. All the patients included as advanced cirrhosis in this study had undergone liver transplantation.
Using molecular and functional parameters we determined CSC like traits in Ep1CIRs. Notch is a well-studied pathway implicated in the regulation of both normal HSCs and CSCs. Notch signaling was activated in HSCs during biliary regeneration, but was restricted during hepatocellular regeneration [36]. On the contrary, a previous study indicated increased Notch activity in CSCs as compared to HSCs in breast cancer [37]. Cirrhotic patients recruited in this study had primary hepatocellular injury rather than primary cholangiocyte injury (Primary Biliary Cirrhosis and Primary Sclerosing Cholangitis patients were not enrolled in the study). Yet, we observed that Jag1, Notch 1, Notch 2, Notch 3, and Hey2 were significantly upregulated in both Ep1CIR and Ep1HCC cells as compared to Ep1NSCs, indicating the appearance of CSC like phenotype during advanced cirrhosis. It is well documented that self-renewal capacity of a cell can be quantified by spheroid formation in vitro, in particular by analyzing the number and size of cells within the spheroid and the ongoing generation (self-renewal) of spheroids through repeated passages. [38]. Ep1CIR cells formed few large sized spheroids compared to the majority of smaller ones, providing direct evidence for the presence of a heterogeneous population of EpCAM1 cells within the cirrhotic livers. Increased spheroid formation over repeated passages reflected the limitless proliferation capabilities of Ep1CIR cells. Further, unlike Ep1NSCs, Ep1CIR cells were capable of generating tumors when injected subcutaneously in NOD/SCID mice. Ep1CIR cells thus exhibited enhanced self-renewal with uncontrolled proliferation and tumor formation, all of which are hallmark features of a CSC [39].
In order to determine the similarity at the molecular level between Ep1CIR and Ep1HCC cells, NGS was performed. Sequencing data deciphered that JNK, MAPK and Wnt signaling were some of the important pathways regulated by miRNA and genes commonly shared by Ep1CIR and Ep1HCC cells. JNKs were earlier considered to have pro-apoptotic functions in response to stress, inflammatory or oncogenic signals [40]. However, emerging evidence now suggests that the JNKs, especially JNK1, play an important role in the malignant transformation of cells and in tumorigenesis [41]. Studies have also shown the association of higher JNK1 activation with a poorer prognosis as well as overexpression of EpCAM, CD24, CD133, KRT19, and AFP in HCC [42]. Indeed, we too observed an over-expression of JNK1 in Ep1CIR and Ep1HCC cells as compared to Ep1NSC (data not shown).
Interestingly, 11 out of 22 Ep1HCC specific miRNAs were also expressed in Ep1CIRs. Among these, miR-511 is negatively associated with overall survival of HCC patients [43]. miR-210 promotes Abbreviations: EpCAM, epithelial cell adhesion molecule; Ep1CIR, EpCAM1 cells from advanced cirrhosis; Ep1HCC, EpCAM1 cells from HCC, Ep1NSC, EpCAM1 cells from noncancerous/noncirrhotic (control); HCC, hepatocellular carcinoma; NSCs, normal stem cells. migration and invasion of HCC cells [44]. miR-141 has been reported to confer resistance to 5-flurouracil in HCC cells [45]. Importantly, miR-18a and 224 have been proposed as serum biomarkers for diagnosis of HCC [46,47]. Thus, most of the 11 miRNAs are associated with imparting tumorigenic properties in HCC cells and their increased expression in Ep1CIRs reflects their transformed status. However, role of these miRNAs in CSCs still needs to be elucidated. Additionally, the functions of miR-548o, 561, and miR-7-1 have not yet been well explored in context of both HCC and CSCs.
Wnt signaling is an important proliferation pathway, well documented in the pathogenesis of many cancers. Importantly, Yamashita et al. showed that Wnt/b-catenin enriches the EpCAM1 CSCs by enhancing their self-renewal and inhibiting their differentiation [48]. We therefore validated the expression of Wnt pathway genes by qRT-PCR and found them to be upregulated in both Ep1CIR and Ep1HCC cells. Colon CSCs have been shown to demonstrate autocrine Wnt secretion via Evi/Wls, a transporter of Wnt [49]. Present data showed significantly increased expression of evi and wnt3 in Ep1CIR and Ep1HCC cells. Subsequently, in the presence of Wnt inhibitor (IWP 12), the spheroid forming ability of both group of cells reduced significantly. To the best of our knowledge, these results report for the first time, the existence of autocrine Wnt signaling in EpCAM1 CSCs from HCC. In addition, presence of autocrine Wnt signaling in Ep1CIR strengthen our notion of existence of CSC like cells during advanced cirrhosis.

CONCLUSION
In conclusion, we report the existence of a small population of EpCAM1 CSC like cells in advanced cirrhosis, which have the potential to develop into HCC because of their limitless selfrenewal through autocrine activation of Wnt signaling. Targeting these cells at initial stages of advanced cirrhosis can help to prevent progression toward HCC, which is often detected at a stage that is not amenable to present treatment modalities. However, to target these cells at cirrhotic stage, we need to identify an efficient biomarker which is preferentially expressed in CSCs and not NSCs. Further investigations of CSCs specific miRNAS in patient In the presence of IWP12 (an inhibitor of Wnt acyltransferase), spheroids formed by EpCAM1 cells of both advanced cirrhosis and HCC were reduced in size as well as number. There was no significant effect of the inhibitor on spheroid formation of Ep1NSCs (x200; Scale bars 5 100mm) (n 5 3 in each group). (p < .05 5 significant). Abbreviations: EpCAM, epithelial cell adhesion molecule; Ep1CIR, EpCAM1 cells from advanced cirrhosis; Ep1HCC, EpCAM1 cells from HCC, Ep1NSC, EpCAM1 cells from noncancerous/noncirrhotic (control); HCC, hepatocellular carcinoma; NSCs, normal stem cells.
serum samples may highlight their use as biomarkers for early diagnosis of HCC at cirrhotic stage.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.