ARTEMIN Promotes Oncogenicity and Resistance to 5-Fluorouracil in Colorectal Carcinoma by p44/42 MAPK Dependent Expression of CDH2

ARTEMIN (ARTN), one of the glial-cell derived neurotrophic factor family of ligands, has been reported to be associated with a number of human malignancies. In this study, the enhanced expression of ARTN in colorectal carcinoma (CRC) was observed; the expression of ARTN positively correlated with lymph node metastases and advanced tumor stages and predicted poor prognosis. Forced expression of ARTN in CRC cells enhanced oncogenic behavior, mesenchymal phenotype, stem cell-like properties and tumor growth and metastasis in a xenograft model. These functions were conversely inhibited by depletion of endogenous ARTN. Forced expression of ARTN reduced the sensitivity of CRC cells to 5-FU treatment; and 5-FU resistant CRC cells harbored enhanced expression of ARTN. The oncogenic functions of ARTN were demonstrated to be mediated by p44/42 MAP kinase dependent expression of CDH2 (CADHERIN 2, also known as N-CADHERIN). Inhibition of p44/42 MAP kinase activity or siRNA mediated depletion of endogenous CDH2 reduced the enhanced oncogenicity and chemoresistance consequent to forced expression of ARTN induced cell functions; and forced expression of CDH2 rescued the reduced mesenchymal properties and resistance to 5-FU after ARTN depletion. In conclusion, ARTN may be of prognostic and theranostic utility in CRC.


INTRODUCTION
Colorectal carcinoma (CRC) is the fourth most diagnosed cancer accounting for 8.5% of cancer related death globally (1). Metastasis, especially hepatic metastases, remains the predominant cause of CRC-related death in patients (2,3). CRC patients with distant metastases exhibit dismal 5-year survival rates of 14% (4). Surgery, radiation, chemotherapy and targeted therapy, or combinations of these therapies are currently available treatment options for CRC patients. The introduction of chemotherapy dramatically improved the median survival of metastatic CRC (mCRC) patients when 5fluorouracil (5-FU, an analog of fluoropyrimidine) was first applied in the treatment of mCRC (5). Combination application of fluoropyrimidine-based therapy with other agents such as oxaliplatin and irinotecan also improve treatment efficacy and reduce toxicity (5,6). Targeted therapies were later utilized in treatment of CRC with or without chemotherapy. Targets for biomarker-oriented therapies are involved in signaling pathways including receptor tyrosine kinases (RTK), Phosphoinositide-3 Kinase (PI3K)-AKT, Transforming Growth Factor b (TGF-b)-SMAD, WNT-b-Catenin and Tumor Protein 53 (TP53), and also mutations in genes including RAS, RAF, PTEN (Phosphatase and Tensin homolog) and MYC (7)(8)(9). The RAS-RAF-MEK-p44/42 Mitogen-Activated Protein Kinase (p44/42 MAPK) cascade, as a common downstream mediator of RTKs such as Epidermal Growth Factor Receptor (EGFR), Vascular Endothelial Growth Factor Receptor (VEGFR) and Human Epidermal Growth Factor Receptor 2 (HER2), has been a focus for the development of targeted therapies (9)(10)(11)(12)(13)(14)(15). There are currently nine FDAapproved targeted agents for CRC of which 2 are anti-EGFR, 4 are anti-VEGF/VEGFR and 3 are immune checkpoint inhibitors (7). There is also a growing list of targeted agents currently in clinical trial (7). However, due to the high heterogeneity of CRC (16)(17)(18), complex feedback and crosstalk of targeted biomarkers (7,(19)(20)(21)(22)(23)(24) and the dynamic expression status of the targets during targeted treatment (25,26), acquired resistance to therapies and disease progression are observed even if the patients responded well to the initial treatment (27). Most CRC patients have very limited further choices for therapy (28). Therefore, the development of novel validated oncogenic targets for prognosis and therapy is of urgent importance to ameliorate CRC treatment outcomes.
The mRNA expression of the ARTN gene is detected in multiple fetal and adult human organs (29,36,37) associated with peripheral nerve innervation. In the adult, ARTN is expressed in peripheral tissues including prostate, pancreas, heart, kidney, pituitary gland, lung, ovary, small intestine, colon, testis and blood vessels (29,36,38) and at low levels in the central nervous system (29). Elevated expression of ARTN has been observed in multiple human cancers including endometrial (39), lung (40), mammary (41), liver (42) and pancreatic (43) carcinoma. In these cancers, ARTN expression is positively correlated with tumor size, invasiveness and metastasis (39,40,44,45). Increased ARTN expression in mammary carcinoma also promotes cancer stem cell-like behavior and acquired resistance to both chemo-and targetedtherapies (33,46). Furthermore, a recent study has indicated that splenic Ter-cells also secrete ARTN to promote hepatocellular carcinoma progression (42). Herein, the functional role of ARTN in CRC progression has been determined and demonstrated that ARTN promotes progression of CRC by p44/42 MAPK dependent expression of CADHERIN 2 (CDH2, also known as N-CADHERIN) which promotes epithelial-to-mesyenchymal transition and cancer cell metastasis (47).

Cell Culture and Transfection
Caco2, DLD1, HCT116, HT29, Lovo, and SW480 cell lines used in this study were obtained from American Type Culture Collection and cultured as recommended. Transfection reagent FuGENE 6 (Roche Diagnostics GmbH, Mannheim, Germany) was used for plasmid transfection in Caco2, DLD1 and HCT116 cells. According to the selection markers of each plasmid, 2-4 weeks of antibiotic selection following transfection was applied as described previously (48,49). For siRNA transient transfection, Lipofectamine RNAiMax was used as described previously (48,50). The cells were incubated with the plasmid/ siRNA-transfection reagent mixture for 4-6 hours in serum free media. Equal amount of media with 10% FBS were added after this incubation period. The transfected cells were then subjected to assays at 24-48 hours after transfection.

Plasmid and siRNAs
pIRESneo3-ARTN and pSilencer-siARTN constructs were used as previously described (41). In brief, the pIRESneo3 empty vector and pIRESneo3-ARTN constructs were used to generate stable forced expression of ARTN in Caco2, DLD1 and HCT116 cells; the pSilencer-CK (scrambled sequence) and pSilencer-siARTN constructs were used to generate stable depletion of ARTN in DLD1 and HCT116 cells. The CDH2 expression construct was kindly provided by Dr. Keith R. Johnson from Weizmann Institute of Science (51). Two siRNAs s2771 and s2772 (Life Technologies, MA, USA), targeting human CDH2 mRNA (GenBank: M34064.1) exon5 and exon 10, separately, were used to deplete the endogenous expression of CDH2 in DLD1 and HCT116 cells.

Reverse Transcription Polymerase Chain Reaction (RT-PCR) and PCR Primers
Total RNA isolation, RT-PCR and quantitative PCR were performed as previously described (52)(53)(54). Primer sequences used were as previously described (39,52). Total RNA of cell samples was extracted with RNeasy Mini Kit (QIAGEN, Stockach, Germany) and converted to cDNA with SuperScript ® VILOTM cDNA Sythesis Kit (Invitrogen, Carlsbad, USA) for quantitative Real-time PCR (polymerase chain reaction) or semi-quantitative PCR (i.e. RTPCR, reverse transcription PCR). Real-time PCR was performed with SYBR ® Green PCR Master mix (Applied Biosystem, Invitrogen, Carlsbad, USA) on the ABI 7900HT Real-Time PCR system.

Cell Functional Assays
Apoptotic analysis was carried out as described previously (50,58). Total cell number, MTT, BrdU incorporation, anchorageindependent growth (soft agar and suspension culture), threedimensional morphogenesis, in vitro cell motility (migration, invasion and wound healing) assays were carried out as previously described (48,50,55): (1) Total cell number: single cells were seeded on day 0 in media with 10% FBS or 1% FBS, and the number of live cells (Trypan blue negative cells) was counted every two days in the following 8-10 days. (2) Alarma blue assay: Alarma blue assay: Alamar Blue ® reagent was added into the cell culture at the ratio of 1:10 and incubated at 37°C for 2-4 hours. The cell viability as reflected by the fluorescence of the media was quantified by plate reader with the excitation wavelength of 560 nm and emission wavelength of 590 nm. (3) BrdU incorporation: Cells were incubated in media with 10% or 1% FBS in the presence of BrdU for a period of time before fixation. The detection of incorporated BrdU was quantified by The BrdU Cell Proliferation Assay Kit (Merck Millipore, Darmstadt Germany). (4) Anchorage-independent growth: single cells were suspended and seeded into 0.35% of soft agar with media containing 1% FBS; the number of colonies ≥50 µm in diameter were recorded after 14 days of incubation. (5) Three-dimensional morphogenesis: single cells were seeded in media with 2% FBS and 10% Matrigel and the cell viability was quantified with Alamar Blue assay after 7 days. (6) In vitro cell motility: single cells were seeded into the upper chamber of Transwell inserts with (invasion assay) or without (migration assay) a pre-coated layer of Matrigel in medium with 0.5% FBS which is placed in the medium with 10% FBS in the lower chamber, and allowed to migrate through the bottom of the inserts for 24-72 hours; the number of cells on the lower surface of the insert was counted. Colony scattering assay was performed as previously described (59). Cells were seeded from a low cell density of 500 cells/well and allowed to growth for 14 days at the end of which the numbers of compact, loose and scattered colonies were recorded.

ALDEFLUOR Assay
The ALDEFLUOR (measuring ALDH1 activity) assay was performed with ALDEFLUOR ™ Kit (STEMCELL Technologies, Vancouver, Canada) (60). Cells were incubated with BAAA (BODIPY ® -aminoacetaldehyde) in the presence or absence of the ALDH inhibitor DEAB at 37°C away from light for 30 min before subjected to flowcytometry analysis. The ALDH bright cells were gated according to the DEAB gating control. The percentages of the ALDH bright cells are stated on the gates (50).

Histopathological Analysis
Tissue specimens were collected from 109 patients, 89 of whom were diagnosed with colorectal carcinoma attending the First Affiliated Hospital of Anhui Medical University (Hefei, P. R. China) between 2002 and 2004. Institutional ethics committee approval for the project was obtained before commencement of the study and followed the Helsinki Declaration.
Rabbit anti-ARTN antibody (Abcam, CAM, UK) was used at the dilution of 1:100 in IHC staining for ARTN expression in paraffin-embedded specimens. Analysis of the IHC staining and clinical parameters of the patients were determined as previously described (41).

Animal Studies
All animal work was performed according to the animal care protocol USTCACUC1301013, which was approved by The Institutional of Animal Care and Ethics Committee of The University of Science and Technology of China. Tumor xenograft and metastasis assays were performed as previously described (48,55). Tumor xenografts were established by subcutaneous flank injection with DLD1 and HCT116 stable cells (1x10 6 cells on each flank), individually. Injected mice were monitored for the formation and growth of the subcutaneous tumors for3-5 weeks. At the end of the monitoring period, the mice were sacrificed and the subcutaneous tumors were harvested for Ki67 staining and TUNEL assay. Metastasis assays were performed by injecting 2X10 6 cells of each stable cell line into the lateral tail vein per mouse. After six weeks, the injected mice were euthanized, and the livers and lungs were harvested and counted for the numbers of metastatic nodules. Harvested tumors, livers and lungs were fixed in 4% paraformaldehyde-PBS and subjected to histological analysis including hemoatoxylin and eosin staining, BrdU assay and TUNEL immunostaining.

Statistics
Numerical data are presented as mean ± S.E.M. (Standard Error of the Mean). Statistical analyses were assessed by Student's t test using Microsoft Excel XP or one-way ANOVA using SPSS Statistics software as appropriate. Kaplan-Meier censored survival analysis was performed to determine the association of ARTN expression to the risk of relapse and death of the colorectal patient cohort. P < 0.05 was considered as significant.

Increased Expression of ARTN in CRC Associated With Poor Prognosis
The expression of ARTN in noncancerous colorectal and CRC tissue specimens was determined using IHC analyses. Only 30% of noncancerous colorectal tissue specimens exhibited ARTNpositive immunoreactivity ( Table 1) with marginally positive staining in the myoblasts around the crypts but not in the crypt cells ( Figure 1A, left). In contrast, 65.2% of CRC tissue specimens exhibited strong ARTN immunoreactivity compared to noncancerous colorectal specimens ( Table 1). Higher intensity of ARTN expression was observed in CRC tissue specimens, both in the myoblasts and the carcinoma cells ( Figure 1A, right).
The association between ARTN expression and clinicopathological parameters in a group of 89 CRC patients was next examined ( Table 2). The expression of ARTN was positively correlated with lymph node metastasis (p = 0.001) and advanced disease stages (p = 0.018). Additionally, Kaplan-Meier analysis of this CRC patient cohort showed that ARTN-positive expression was significantly associated with shorter relapse-free survival (p = 0.024, Figure 1B, left). It was also observed that CRC patients with ARTN-positive cancer tended to have shorter overall survival time, albeit this observation did not reach statistical significance ( Figure 1B, right). These data suggested ARTN expression may serve as a prognostic marker in CRC.

Forced Expression of ARTN Enhances Oncogenic Behaviour of CRC Cells
Caco2, DLD1, HCT116, HT29, Lovo and SW480 cells were screened for endogenous mRNA and protein levels of ARTN, GFRa1, GFRa3 and RET using RT-PCR and western blot as shown in Figures S1A, B. Caco2, DLD1, HCT116, and HT29 cells were observed to express moderate levels of ARTN mRNA. All cell lines exhibited endogenous expression of GFRa1, GFRa3 and RET. To determine the functional effects of ARTN in CRC progression, Caco2, DLD1 and HCT116 cells were stably transfected with the pIRESneo3-ARTN construct (-ARTN) or with pIRESneo3 as the vehicle control (-Vec). DLD1 and HCT116 cells were also stably transfected with a pSilencer-siARTN construct (-siARTN) to deplete ARTN expression, or with pSilencer as the vehicle control (-siVec). The forced or depleted expression of ARTN protein in the CRC cell lines were demonstrated using western blot ( Figures S2A and S3A).
The forced expression of ARTN in DLD1 cells significantly increased cell proliferation compared to vehicle control cells as indicated by total cell number assay in either 1% or 10% FBS ( Figure 2A). DLD1-ARTN cells also exhibited increased S-phase entry compared to DLD1-vector cells (DLD1-Vec) as determined with a BrdU incorporation assay ( Figure 2B). Hoechst 33258 staining also revealed a significant reduction in serum deprivation induced-apoptosis in DLD1-ARTN cells compared to DLD1-Vec cells ( Figure 2C). Similar directional changes in cell proliferation, S-phase entry, and apoptosis were observed in HCT116 cells with forced expression of ARTN ( Figures S2B-D).  Resistance to anoikis and loss of contact inhibition critically contribute to cancer progression (61,62), which is examined by soft agar colony formation and foci formation assay, respectively. The forced expression of ARTN in DLD1 cells increased the number and size of colonies formed in soft agar ( Figure 2D) and also the capacity for foci formation ( Figure 2E). Additionally, forced expression of ARTN enhanced the growth of DLD1 cells in three-dimensional (3D) Matrigel ( Figure 2F); and colonies with forced expression of ARTN exhibited visible cell protrusions projecting into the surrounding matrix ( Figures 2F and S2G, as indicated with red arrow heads). Similarly, forced expression of ARTN also enhanced soft agar colony formation, foci formation and 3D Matrigel growth of Caco2 and HCT116 cells ( Figures S2E-G).
The effect of ARTN expression on the properties of epithelialmesenchymal transition (EMT), migration and invasion of CRC cells were next examined in colony scattering assays, DLD1-ARTN cells formed higher proportion of loose and scattered colonies, and a lower proportion of compact colonies than those formed by the vector control cells ( Figure 2G). Monolayercultured DLD1-ARTN cells developed pseudopodia-like protrusions on the edge of cells, indicative of increased mesenchymal characteristics as compared to the epithelial-like morphology of DLD1-Vec cells ( Figures S2I). Visualization of the subcellular distribution of filamentous actin (F-actin) using fluorescence microscopy revealed accumulation of F-actin at the cell periphery of DLD1-ARTN, intermingled with the leading edges of the cells; in contrast, the DLD1-Vec cells exhibited more prominent stress fibers across the cells with relatively less accumulation of F-actin at the cell periphery ( Figure 2H). Enhanced cell migratory potential was observed in DLD1-ARTN cells in wound healing ( Figure 2I) and transwell migration assays ( Figure 2J). DLD1-ARTN cells also exhibited significantly increased invasion through the Matrigel layer in transwell invasion assays ( Figure 2K). Similar morphological changes and enhanced migratory and invasive potential were also observed in Caco2 and HCT116 cells with forced expression of ARTN ( Figures S2H-M). Thus, the forced expression of ARTN in CRC cells stimulated cell proliferation, cell survival, oncogenicity, migration and invasion.

Depletion of ARTN Inhibits Oncogenic Behaviors of CRC Cells
Caco2, DLD1, HCT116, and HT29 cells were found to be positive for ARTN expression as demonstrated using RT-PCR assay; whereas, Lovo or SW480 cells did not express ARTN mRNA under similar RT-PCR conditions ( Figure S1A). All six CRC cell lines were observed to express mRNA of GFRa1, GFRa3, SDC3 and RET, except HCT116 cells which did not express GFRa1 mRNA under similar RT-PCR conditions ( Figure S1A). The protein levels of RET, GFRa1 and GFRa3 and ARTN in Caco2, DLD1 and HCT116 stable cells with forced expression of ARTN was analysed using western blot analysis ( Figure S1B). Based on the expression level of ARTN and that of the GFRa1, GFRa3, SDC3 and RET receptors in the screened CRC cell lines, we next generated stable clones of DLD1 and HCT116 cells with depletion of endogenous ARTN using siRNA-ARTN construct. In contrast to forced expression of ARTN, siRNA-mediated depletion of endogenous ARTN in DLD1 cells significantly reduced total cell number ( Figure 3A) by decreasing cell proliferation as determined in the BrdU incorporation assay ( Figure 3B) and inducing apoptosis as determined by Hoechst 33258 staining ( Figure 3C). The DLD1-siARTN cells exhibited markedly reduced soft agar colony formation assay ( Figure 3D). DLD1-siARTN cells also formed less colonies and exhibited significantly reduced cell viability in 3D Matrigel as compared to the vector control DLD1-siVec cells ( Figure 3E). DLD1-siARTN cells exhibited more compact colony morphology as compared to DLD1-siVec ( Figure S3F). Moreover, reduced cell migration and invasion was observed with depletion of endogenous ARTN in DLD1 cells ( Figures 3F, G). Similar directional changes in decreased oncogenicity were observed in HCT116 cells after depletion of endogenous ARTN expression ( Figures S3B-F). Hence, the depletion of endogenous ARTN in CRC cells decreased cell proliferation, survival, oncogenicity, migration and invasion.

ARTN Promotes Oncogene and Suppresses Tumor Suppressor Gene Expression
The expression of genes mediating various cellular functions in these stable cell lines was determined using quantitative real-time PCR analysis. Consistent with ARTN mediation of CRC cell proliferation and survival, DLD1 cells with forced expression of ARTN exhibited significantly elevated expression of cell cycle regulatory genes such as CCND1, CCNE1, CHEK2, CDK4 and CDC25A, and anti-apoptotic genes such as BCL2, CFLAR and TERT, but significantly reduced expression of the pro-apoptotic genes CASP7 and BAD (DLD1-ARTN vs. DLD1-Vec, Table 3). Similarly, the expression of mesenchymal phenotype associated genes such as CDH2, VIM, MMP1 and MMP9 was enhanced with forced expression of ARTN, while that of epithelial phenotype associated genes such as CDH1 and OCLN were decreased, in accordance with the EMT-inducing, promigratory and pro-invasive function of ARTN in CRC cells. In contrast, depletion of endogenous ARTN in DLD1 cells resulted in decreased expression of CCND1, CHEK2, BCL2, VIM, CDH2 and increased expression of BAX, CDH1 and OCLN (DLD1-siARTN vs. DLD1-siVec, Table 3). The expression of the key EMT markers in Caco2, DLD1 and HCT116 cells was also verified by western blot analysis. ARTN expression in CRC cells decreased protein levels of OCLN and increased protein levels of CDH2 and SNAIL: a small increase in expression of VIM with forced expression of ARTN was observed only in Caco2 and DLD1 cells but not in HCT116 cells ( Figure 3H and Figure S6); whereas the opposite expression pattern of OCLN, CDH2 and SNAIL was observed with depletion of endogenous ARTN in CRC cells ( Figure 3I).

Forced Expression of ARTN in CRC Cells Promotes Xenograft Growth and Metastasis
The xenograft growth of CRC cells with forced expression of ARTN was examined as previously described (48). DLD1-ARTN cell-derived tumors exhibited significantly enhanced tumor volume compared to DLD1-Vec cell-derived tumors ( Figure 4A). As determined by Ki67 staining and TUNEL assay, DLD1-ARTN cell-derived tumor specimens exhibited significantly higher Ki67-positive nuclei ( Figure 4B) and lower apoptotic positive nuclei ( Figure 4C) compared to DLD1-Vec cell-derived tumor specimens. On the contrary, tumors generated by DLD1-siARTN cells exhibited reduced tumor volume ( Figure 4D) with decreased Ki67-positive nuclei ( Figure 4E) and increased apoptotic positive nuclei ( Figure 4F), compared to DLD1-siVec cells generated tumor specimens. Similar directional changes in xenograft growth was observed with HCT116 cell generated tumors with either forced or depleted expression of ARTN ( Figures S4A-F). Thus, ARTN expression in CRC cells promoted xenograft growth.

Forced Expression of ARTN in CRC Cells Promotes Metastasis
Metastasis assays with tail vein injection using DLD1 cells with and without forced expression of ARTN were also performed as previously described (55). Mice were examined for lung and liver metastases, as described in Materials and Methods.
Hemoatoxylin and Eosin (H&E) staining showed that mice injected with DLD1-ARTN cells developed more metastatic nodules in lung compared to those mice injected with DLD1-Vec cells (Figures 5A, B). Metastatic burden was verified by the determination of the mRNA expression of human HPRT gene (encoding human hypoxanthine phosphoribosyl transferase) which is distinguishable from the endogenous mouse HPRT gene, and therefore may be used for quantification of cells of human origin in mice (63,64), i.e. metastases in this study, relative to mouse b-ACTIN mRNA using real-time qPCR ( Figure 5C). Furthermore, only one mouse injected with DLD1-ARTN cells was observed to develop metastatic nodules in the liver (n=1/6) whereas no metastatic nodules in liver were Gene expression is presented in fold change relative to the individual control cells, respectively. Fold change values are representative of three independent biological experiments. To compensate for potential differences between markers, the relative expression was computed, based on the efficiency (E), normalized by three housekeeping genes b-ACTIN, HPRT, and GAPDH. P values in bold are considered as significant (p < 0.05). observed in the group of mice injected with DLD1-Vec cells ( Figure 5G). Consistently, DLD1-siARTN cells generated a smaller number of lung metastatic nodules compared to DLD1-siVec cells (Figures 5D, E). Concordantly, markedly lower human HPRT mRNA expression was observed in lungs from mice injected with DLD1-siARTN cells as compared to the lungs of mice injected with DLD1-siVec cells ( Figure 5F). No metastatic nodules were observed in the livers of mice injected with either DLD1-siVec or DLD1-siARTN cells ( Figure 5H). Thus, ARTN expression in CRC promoted metastatic dissemination of CRC cells.

Forced Expression of ARTN in CRC Cells Stimulates a Cancer Stem-Like Cell (CSC) Phenotype
The functional effects of ARTN expression on the CSC phenotype were assessed using DLD1 and HCT116 cells with either forced or depleted expression of ARTN. DLD1-ARTN cells formed an increased number of larger-size colonospheres in suspension culture compared to DLD1-Vec cells. Conversely, DLD1-siARTN cells exhibited a reduced number of colonospheres compared to DLD1-siVec cells ( Figure 6A). The numbers of DLD1 colonospheres increased with each successive generation, indicating the enrichment of self-renewable CSC phenotype. DLD1-ARTN cells maintained higher colonosphere formation capacity while DLD1-siARTN cells formed less colonospheres as compared to their respective control cells after 3 passages (P1-P3, Figure 6A). The mRNA levels of ARTN and CSC associated genes were also assessed in DLD1 cells after culture on either monolayer or in suspension. DLD1 cell generated colonospheres exhibited increased mRNA ( Table 4) and protein ( Figure 6B) levels of ARTN compared to DLD1 cells cultured on monolayer. Consistently, increased mRNA levels of CSC-phenotype associated genes, ALCAM, BMI1, CD133, CD44 and ALDH1, were also observed in DLD1 colonospheres compared to DLD1 cells cultured on monolayer ( Table 4).

ARTN Stimulates Its Functions Through p44/42 MAPK/CDH2 Signaling in CRC Cells
The mechanistic basis of ARTN-stimulated oncogenic behaviors in CRC cells was further investigated. Amongst the reported RET mediated pathways, increased p44/42 MAPK activity significantly contributes to progression of CRC (65)(66)(67). Herein, a significant increase in the phosphorylation of p44/42 MAPK was also observed in CRC cells with forced expression of ARTN. Conversely, depleted expression of ARTN markedly reduced the phosphorylation of p44/42 MAPK in CRC cells (DLD1, Figure 7A; HCT116, Figure S5A).
As observed in Figures 3H, I, the levels of CDH2 protein in CRC cells positively correlated with ARTN expression. The levels of CDH2 protein also positively correlated with p44/42 MAPK phosphorylation in CRC cells ( Figure 7B), while the expression changes for other investigated mesenchymal markers were not as significant as that of CDH2. Transient forced expression of CDH2 in DLD1 cells did not affect ARTN expression nor p44/ 42 MAPK phosphorylation ( Figure 7B). Furthermore, transient depletion of endogenous CDH2 expression in DLD1 cells partially abolished the ARTN-stimulated 3D Matrigel growth ( Figure 7C) and cell invasion ( Figure 7D), whereas transient forced expression of CDH2 partially rescued 3D Matrigel growth ( Figure 7C) and cell invasion ( Figure 7D). Similar directional changes in ALDH1 activity was also observed in DLD1 stable (ARTN or siARTN) cells with either forced or depleted expression of CDH2. ( Figure 7E).
We also demonstrated that the pharmacological inhibition of p44/42 MAPK phosphorylation using MEK inhibitor, U0126, markedly reduced the protein levels of CDH2 in DLD1 cells, without affecting the endogenous protein levels of ARTN ( Figure 7F). Additionally, the pharmacological inhibition of p44/42 MAPK using U0126 significantly abrogated the ARTNenhanced invasion ( Figure 7G) and 3D Matrigel growth ( Figure 7H) of DLD1 cells. Hence, ARTN stimulated oncogenic activities in CRC cells was partially mediated in a p44/42 MAPK-N-CADHERIN-dependent manner. Gene expression is represented in fold change relative to the individual control cells, respectively. Fold change values are representative of three independent biological experiments. To compensate for potential differences between markers, the relative expression was computed, based on the efficiency (E), normalized by three housekeeping genes b-ACTIN, HPRT, and GAPDH. P values in bold are considered as significant (p < 0.05). Gene expression is presented as fold change of colonosphere culture relative to monolayer culture. Fold change values are representative of three independent biological experiments. To compensate for potential differences between markers, the relative expression was computed, based on the efficiency (E), normalized by three housekeeping genes b-ACTIN, HPRT, and GAPDH. P values in bold are considered as significant (p < 0.05). transiently depleted in DLD1 5-FU-R cells. The depleted expression of ARTN in both control or 5-FU-R DLD1 cells decreased monolayer cell viability ( Figure 8D) and growth in 3D Matrigel ( Figure 8E). Consistently, depleted expression of CDH2 in DLD1-5-FU-R cells further reduced cell viability ( Figure 8F) and 3D Matrigel growth ( Figure 8G) after treatment with 5-FU. Combined treatment of DLD1-5-FU-R cells with 5-FU and U0126 significantly reduced the cell viability to the level of control cells ( Figure 8H). These data indicate that both inhibition of p44/42 MAPK by U0126 and depletion of CDH2 can partially re-sensitize ARTN-dependent resistance to 5-FU in DLD1 cells.

DISCUSSION
In this study, the positive association between ARTN expression and advanced characteristics of CRC ( Table 2) suggested that increased ARTN expression in CRC cells promotes the metastatic process. It was observed that high tumor expression of ARTN in CRC also predicted relapes and overall survival rates. Given that it has also recently been described that ARTN is also produced by splenic Ter cells, and secreted to serum to promote hepatocellular carcinoma progression (42), then examination of tumor ARTN expression will potentially underestimate the full contribution of ARTN to relapse and survival. Hence, estimation of serum ARTN may provide a more accurate determination of the contribution of ARTN to CRC progression if splenic Ter cells are also responsive to CRC. Regardless, the data herein establishes ARTN as a potential theranostic target for metastatic CRC. Enhanced Epithelial-to-Mesenchymal Transition (EMT) is observed with forced expression of ARTN in the examined CRC cells. Forced expression of ARTN promoted morphological changes such as the laminopodial-like plasma membrane protrusions in Caco2-ARTN and DLD1-ARTN cells, and mesenchymal-like cell morphology in HCT116-ARTN cells. In addition, DLD1 and HCT116 cells with forced expression of ARTN exhibited a reduced F-actin network, which is associated with cell transformation (68,69) and leading to morphological changes and enhanced motility (70,71). In DLD1-ARTN and HCT116-ARTN cells, the reduced proportion of compact colonies reflects loss of tight intercellular contact; while the accelerated wound closing rate and increased cell migration indicate increased cell motility. Increased growth and cell protrusions in 3D Matrigel, and increased cell invasion indicate the enhanced invasive capacity of these cells. This data supports a motogenic function of ARTN that is associated with an enhanced mesenchymal phenotype (72,73). Examination of the mRNA and protein expression of EMT markers/regulators revealed an ARTN-promoted mensenchymal profile: forced expression of ARTN in examined CRC cells reduced the expression of the epithelial maker OCLN but increased that of the mesenchymal markers CDH2 and VIM ( Table 3). Whilst the reduction of CDH1 expression is acknowledged as one marker for EMT, a reduction of CDH1 mRNA was observed in CRC cells with forced expression of ARTN; however, the protein expression of CDH1 exhibited little reduction in DLD1-ARTN or HCT116-ARTN cells. EMT exists in a continuum (74) and it is possible that high ARTN producing DLD-1 cells are in initial stages of the EMT process, in which CDH1 is still required for the maintenance of intercellular adhesion during migration. A similar observation has been reported that autocrine hGH (human growth hormone) promotes EMT with no change in CDH1 (75).
The EMT process has also been reported to promote stemness in cancer cells (72,(76)(77)(78). Together with an enhanced mesenchymal phenotype, forced expression of ARTN rendered DLD1 cells with enhanced stem-cell like properties. This would contribute to increased viability of the detached mesenchymal-like cancer cells (as observed by enhanced anchorage independent growth), which together with enhanced invasiveness would mediate ARTN promoted metastasis (79)(80)(81). Indeed, forced expression of ARTN promoted CRC cell metastasis. Furthermore, the enhanced cell viability and survival associated with mensenchymal transition would promote resistance to chemotherapeutic agents (81). Indeed, reduced sensitivity to 5-FU was associated with elevated expression of ARTN in DLD1 cells, which was reversed by depletion of endogenous ARTN. Studies of underlying mechanisms have indicated that EMT, stem cell-like properties and chemoresistance are interrelated with each other and share phenotypic and signaling profiles with mesenchymal and stem-like cells (82)(83)(84). In this study, it was demonstrated that ARTN enhanced the expression of EMT markers including CDH2 and SNAIL, both of which play key roles in the activation of WNT/b-Catenin pathway (85)(86)(87). The mRNA level of other EMT markers such as MET, MMP and MMP9 were also increased with forced expression of ARTN and reduced with depletion of endogenous ARTN in DLD1 cells. In addition, forced expression of ARTN enhanced the mRNA levels of cell survival genes and cell cycle regulators including CCND1, BCL2, MYC and NFkB as well, but reduced that of BAX ( Table 3).
The functions of ARTN herein are observed to be dependent on the activation of p44/42 MAPK and the subsequent increased expression of CDH2 in DLD1 cells. The expression of SNAIL exhibited similar directional changes as that of ARTN in the examined CRC cells. SNAIL is well-known for its regulatory role in the CDH1-to-CDH2 switch (switch of expression) during epithelial-to-mesenchymal transition (88,89), as SNAIL not only suppresses the expression of Cadherin 1 (CDH1, also known as E-cadherin) (88) but also activates the expression of CDH2 in cooperation with other transcriptional activators such as TWIST1 (90). As SNAIL is reported to be induced by p44/42 MAPK signaling during EMT (91)(92)(93), SNAIL could be the signaling mediator in between p44/42 MAPK and CDH2. There may be alternative mechanism for the p44/42 MAPK dependent regulation on CDH2. The work by Li and colleagues has suggested SP1, one of the downstream targets of p44/42 MAPK, as a potential transcription factor for the transcriptional activation of the CDH2 gene (94,95). We performed preliminary analyses on the sequence (-2000 to -12000kb) upstream of the first exon of the CDH2 gene with transcription factor binding site analysis programs TRANSFAC BLAST and TESS (as incorporated as supplementary files 1 and 2) and revealed binding sites for potential transcription factors, among which NFkB and AP-1 are known to be direct targets of activated p44/42 MAPK (96,97) and demonstrated to be associated with CRC progression (98)(99)(100)(101). Enhanced CDH2 expression is well known to associate with cancer cell motility, invasiveness and metastasis in different cancers (102)(103)(104)(105). CDH2 mediates the trans-endothelial migration (TEM) of cancer cells by enhancing the endothelia attachment of circulating cancer cells (106). The heterotypic adhesion between cancer cells and endothelia also promotes signaling by CDH2 binding molecules such as p120 and b-Catenin, leading to the dissolution of VE-Cadherin-mediated endothelial junction and facilitating the TEM of the cancer cells (107)(108)(109). Finally, a function of CDH2 in chemoresistance is reported herein, as depletion of CDH2 expression partially abrogated chemoresistance.
In summary, this study observed an association between tumor ARTN expression and clinical outcome for CRC patients. A functional role of ARTN in promoting colon cancer cell oncogenicity and metastasis with CSC-like behavior and chemoresistance via p44/42 MAPK dependent expression of CDH2 was further demonstrated. Hence, ARTN may be of prognostic and therapeutic utility in CRC. Furthermore, targeting CDH2 may be an alternative or combination strategy to inhibit progression of ARTN dependent cancer.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Institutional Medical Ethics ReviewBoard of the First Affiliated Hospital of Anhui Medical University (Hefei, P. R. China). The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by The Institutional of Animal Care and Ethics Committee of The University of Science and Technology of China.

AUTHOR CONTRIBUTIONS
Q-SZ and PL designed the research. Z-SW performed ARTN expression analysis on clinical samples. Q-SZ performed cellular experiments. XS and MZ conducted animal experiments. Q-SZ wrote the paper. TZ, VP, and PL edited the paper. All authors contributed to the article and approved the submitted version.