ARTEMIN synergizes with TWIST1 to promote metastasis and poor survival outcome in patients with ER negative mammary carcinoma

Introduction ARTEMIN (ARTN) is an estrogen regulated growth factor, the expression of which promotes resistance to antiestrogen therapies and predicts poorer survival outcome of patients with estrogen receptor (ER) positive mammary carcinoma (ER+MC) treated with tamoxifen. ARTN is also expressed in ER negative mammary carcinoma (ER-MC). Herein, we determined the role of ARTN in ER-MC and defined the mechanism of action producing poor patient prognosis. Methods We modulated the expression of ARTN in two ER- (mesenchymal/claudin-low) mammary carcinoma cell lines (BT549 and MDA-MB-231) by forced expression or small interfering RNA (siRNA) mediated depletion. The effects of modulation of ARTN expression were examined by various in vitro measures of oncogenicity, including the expression of TWIST1 messenger RNA (mRNA) and protein. In vitro results were correlated to xenograft studies in immunodeficient mice. Co-expression of ARTN and TWIST1 and their association to poor survival outcome were examined in a cohort of patients with ER-MC. Pathway analysis was performed by pharmacological inhibition of phosphorylation of AKT (pAKT-Ser 473) or modulation of TWIST1 expression. Results ARTN expression resulted in ER-MC cells with enhanced mesenchymal characteristics, including increased invasion and a gene expression profile consistent with enhanced mesenchymal phenotype. ARTN stimulated ER-MC cell anchorage independent and 3D matrigel growth, endothelial cell adhesion and transmigration of ER-MC cells through an endothelial cell barrier. Forced expression of ARTN produced a larger, locally invasive tumour mass with tumour emboli that produced distant metastasis. ARTN regulated TWIST1 expression in ER-MC cells and ARTN expression was significantly correlated to TWIST1 expression in a panel of mammary carcinoma cell lines and in a cohort of patients with ER-MC. Low expression of both ARTN and TWIST1 predicted 100% relapse free and overall survival in patients with ER-MC, whereas high expression of both ARTN and TWIST1 was associated with a poor survival outcome. ARTN stimulated an increase in TWIST1 expression via increased AKT activity. siRNA mediated depletion of TWIST1 abrogated ARTN stimulated cellular behaviour associated with metastasis, and forced expression of TWIST1 abrogated the functional effects of ARTN depletion. Conclusions ARTN and TWIST1 synergize to produce a worse outcome in ER-MC and combined inhibition of ARTN and phosphatidylinositol 3-kinase/protein kinase B (PI3K/AKT) may therefore provide a novel therapeutic strategy in this subtype of mammary carcinoma.


Introduction
Progression of mammary carcinoma is a complex process that involves aberrant regulation of multiple signalling pathways [1]. Determination of estrogen receptor (ER) status of invasive carcinomas before therapeutic intervention has become standard practice in the management of mammary carcinoma. Anti-estrogen therapy has proven to be successful for the treatment of ERpositive mammary carcinoma (ER+MC), which consequently generally has a better prognosis [2]. Conversely, ER-negative mammary carcinomas (ER-MC) are more aggressive and consistently associated with poorer prognosis [2]. Furthermore, based on molecular profiles, at least four different subtypes of mammary carcinoma (luminal A, luminal B, human epidermal growth receptor (HER)2-enriched and basal-like) have been identified [3]. Another mammary carcinoma subtype that has recently been delineated, termed claudin-low, exhibits high expression of genes involved in epithelial-mesenchymal transition (EMT), such as SNAI1 and TWIST1, and cancer stem cell-like features [4,5]. Compared with the luminal, HER2-enriched and basal-like subtypes, claudin-low tumors exhibit lower expression of ER, progesterone receptor (PR), HER2 and GATA3 and higher expression of mesenchymal, cell migration and angiogenesis genes [4]. Clinicopathological analyses also revealed that claudin-low tumors have poorer prognosis and may not be managed effectively with existing chemotherapy regimens [4]. Identification of novel and targetable molecular pathways responsible for propagating the poor prognosis in ER-MC is therefore warranted. ARTEMIN (ARTN) is a member of the glial-cell linederived neurotrophic factor (GDNF) family of ligands [6]. ARTN has been reported to promote mammary [7,8], endometrial [9], lung [10] and pancreatic [11] carcinoma progression. ARTN is an estrogen-regulated gene and it has been demonstrated that ARTN reduces the efficacy of anti-estrogens in ER+MC [8]. Furthermore, ARTN expression is correlated with decreased survival of patients with ER+MC treated with tamoxifen. Importantly, depletion or inhibition of ARTN partially restores tamoxifen sensitivity in tamoxifen-resistant mammary carcinoma cells [8].
TWIST1 belongs to the family of basic helix-loophelix transcription factors originally found to modulate the expression of various target genes through canonical E-box responsive elements [12,13]. Increased expression of TWIST1 is observed in various human cancers [5,14,15] including ER-negative and claudin-low mammary carcinoma [4]. Evidence also indicates that TWIST1 contributes to cancer cell dissemination by promoting EMT and increasing cellular invasiveness [16,17]. Furthermore, a recent report suggests that TWIST1 interacts with several components of the Mi2/ nucleosome remodelling and deacetylase (Mi2/NuRD) complex including metastasis-associated protein family member 2 (MTA2). MTA2 is a corepressor of ERα and increased MTA2 expression leads to estrogen-independent growth of mammary carcinoma cells during mammary carcinoma progression and metastasis [18].
Progression from estrogen dependence to estrogen independence (anti-estrogen resistance) in mammary carcinoma involves the altered expression of one or more estrogen-regulated gene networks [19,20]. In addition to ARTN stimulation of ER-transcriptional activity and the expression of estrogen-responsive genes, we have previously demonstrated that ARTN also promotes estrogen-independent growth of ER+MC cells [8]. ARTN may therefore possess a functional role in ER-MC. Indeed, in a cohort of patients [21], not only was ARTN expression correlated with worse survival outcome in ER+MC patients [8], it also predicted worse survival outcome in ER-MC patients. Such observations prompted us to examine the role of ARTN in ER-MC.
We report herein that ARTN stimulates oncogenicity and metastasis of ER-MC cells via expression of TWIST1 and that functional interaction between ARTN and TWIST1 promote a worse survival outcome in ER-MC.

Reagents
Tetramethylrhodamine isothiocyanate phalloidin was purchased from Sigma (St. Louis, MO, USA). AKT inhibitor IV was purchased from Calbiochem (San Diego, CA, USA).

Cell function assays
Soft agar colony formation assay, 3D cell growth, cell migration and invasion assays were performed as previously described [7]. For monolayer cell proliferation, 3,000 cells were seeded into 100 μl media (2% FBS) and every 24 hours viability was quantified by alamarBlue (Invitrogen, Carlsbad, CA, USA) as described previously [8]. Colony scattering assays were performed as previously described [23]. Endothelial cell adhesion assay was performed as described previously [24], with minor modifications, as 5 μM CFMDA cell tracker green (Invitrogen, Carlsbad, CA, USA) was used to label cells. Adherent cells were observed and photographed under a fluorescence microscope (Olympus, Tokyo, Japan). Fluorescence was measured with excitation wavelength at 492 nm and emission wavelength at 517 nm. The trans-migration assay was performed as described previously [25].

Metastatic seeding assay
All animal work was conducted in accordance with a protocol approved by the institutional animal care and ethics committee. Tumor growth and metastasis assays were performed as mentioned earlier [7,26,27]. For direct tail vein injection, Balb/c nude mice (Shanghai Slaccas Co., Shanghai, China) (four to six weeks old) were used. A 1.5 × 10 6 sample of viable BT549-VEC and BT549-ARTN cells (n = 7 for each group) were washed and harvested in PBS and subsequently injected into the lateral tail vein in a volume of 0.1 ml. A 3 × 10 6 sample of viable MDA-MB-231-VEC and MDA-MB-231-ARTN cells were injected subcutaneously into the mammary fat pad of immunodeficient nude mice (n = 6 for each group). After four weeks, mice were euthanized and lungs and livers were surgically resected for histology. Tissue samples were either fixed in 4% paraformaldehyde-PBS (pH = 7.4), embedded in paraffin and 6 μm-thick sections cut (for histological studies with hematoxylin and eosin) or were frozen at -80°C in RNA-Later (Ambion, TX, USA) for RNA extraction and qPCR analysis.

Histopathological analysis
Tissue samples were collected from 94 ER-negative female mammary carcinoma patients attending the First Affiliated Hospital of Anhui Medical University (Hefei, P. R. China) presenting between 2001 and 2002. Institutional ethics committee approval for the project was obtained before commencement of the study and was in compliance with the Helsinki Declaration. Written informed consent was obtained from all patients. Of 94 patients, 33 were HER-2 positive and the remaining 61 were HER-2 negative as described in the additional file 1. All 94 patients underwent radical mastectomy. All patients had no previous diagnosis of carcinoma, no distant metastasis at the time of diagnosis, and no evidence of disease within one month after primary surgery. All patients received adjuvant chemotherapy (CMF regimen: cyclophosphamide, 600 mg/m 2 iv bolus, d1,8; methotrexate, 40 mg/m 2 iv bolus, d1,8; 5-fluorouracil, 600 mg/ m 2 iv infusion, d1,8; every four weeks for six cycles). Patients who had undergone chemotherapy or radiation therapy before surgery were excluded from this study. Of 94 patients, 76 possessed full follow-up data ranging from 5 to 64 months, with a median time of 60.0 months and a mean time of 44.7 months.
Immunohistochemical (IHC) analysis of paraffinembedded specimens was performed as described previously [7]. In brief, 3 μm thick tissue microarray (TMA) sections were deparaffinised in xylene, rehydrated in a graded series of ethanol solutions, and heated in a microwave oven in 0.01 M sodium citrate buffer (pH 6.0) for 10 minutes for antigen retrieval. Anti-ARTN antibody (rabbit) (Abcam, CAM, UK) was used at 1:100 dilution. The scoring procedure is outlined in additional file 1.

Statistics
All numerical data are expressed as mean ± standard error of the mean from a representative experiment performed in triplicate, and statistical significance was assessed by Student's t-test (P < 0.05 was considered as significant) using Microsoft Excel XP (bar graphs) unless otherwise indicated (chi-squared test). Cox regression analysis was performed to determine the association of ARTN and TWIST1 expression to the risk of relapse and death.

Forced expression of ARTN altered the morphology of ER-MC cells in vitro
To determine the functional consequences of ARTN expression in ER-MC, we generated stable cell clones with forced expression of ARTN in BT549 and in MDA-MB-231 cells (mesenchymal/claudin low cell lines) as previously described [7] [see additional file 2a]. Forced expression of ARTN in both BT549 and MDA-MB-231 cells resulted in a more scattered, elongated and mesenchymal cellular morphology in monolayer adherent culture as compared with the respective control VEC cells ( Figure 1a) and as previously reported for BT549 cells [7]. In colony-scattering assays, we observed a significantly larger proportion of scattered cells (45% and 50% more than During the progression of carcinoma toward a less differentiated and more metastatic state, cells lose their epithelial characteristics and acquire a mesenchymal morphology, together with concomitant changes in gene expression [28]. Such a phenotypic conversion is referred to as EMT. Based on the changes in cell morphology observed with forced expression of ARTN, qPCR was therefore performed to determine the effects  Table  1). The expression of TWIST1 mRNA in BT549-ARTN cells was increased 5.4-fold compared with control VEC cells and in MDA-MB-231-ARTN cells TWIST1 expression was increased by 4.5-fold compared with the VEC cells. Increased TWIST1 expression is correlated with an increased metastatic potential in mammary carcinoma cells [17,29]. Forced expression of ARTN consistently increased the mRNA level of the metastasis tumor-associated protein2 (MTA2) and NFKB1 in both cell pairs. Increased expression of MTA2 leads to estrogen independent growth of human mammary carcinoma cells [30] and increased expression of NFKB1 expression increases the invasive potential of carcinoma cells [31]. Furthermore, increased mRNA expression of NME1, MMP9 and MYC was observed in BT549-ARTN cells and MMP2 mRNA expression was increased in MDA-MB-231-ARTN cells, as compared with the respective control VEC cells. Thus, autocrine production of ARTN in ER-MC cells resulted in decreased expression of epithelial markers and increased expression of mesenchymal and metastatic related markers.

ARTN stimulates the oncogenicity and invasion of ER-MC cells in vitro
Increase in cellular invasiveness and resistance to anoikis are key components of metastasis [32]. Forced expression of ARTN significantly promoted monolayer proliferation, colony formation in soft agar [see increase in monolayer proliferation, 102% and 60% increase in soft agar colony formation, 479% and 170% increase in 3D matrigel growth, 249% and 160% increase in cell migration and 196% and 134% increase in cell invasion as compared with the control VEC cell lines, respectively and as previously reported in BT549 cells [7]. Concordantly, BT549-siARTN and MDA-MB-231-siARTN cells significantly inhibited monolayer proliferation by 23% and 17%, soft agar colony formation by 48% and 38% [see additional files 3c and 3d], 3D matrigel growth by 50% and 40%, cell migration by 31% and 27% and cell invasion by 38% and 40% as compared to the siCONT cells respectively (Figures 2d to 2f).

ARTN enhances metastatic seeding of ER-MC cells in vivo
In our xenograft model, BT549 cells failed to form consistent tumor masses. Therefore, to determine whether ARTN contributes to the metastatic ability of ER-MC cells, we injected BT549-ARTN cells into the tail vein of immunodeficient nude mice and examined their ability to form pulmonary and hepatic metastasis compared with the control VEC cell line. Metastatic nodules were readily detectable by histology in the lungs of all mice (n = 7), four weeks after injection with BT549-ARTN cells, whereas pulmonary metastases were detected in only three of seven mice injected with BT549-VEC cells. The nodules formed in the lungs of animals injected with BT549-ARTN cells gave rise to, on average, more than three nodules per lung (Figureas 3a, left panel and 3b, left panel) as compared with less than one nodule per lung on average in animals with BT549-VEC cells.
A higher frequency of liver metastasis was also observed in mice injected with BT549-ARTN cells (n = 4) compared with mice injected with BT549-VEC cells (n = 1) [see additional file 4a]. We extracted RNA from lung and liver of mice injected with BT549-ARTN cells or BT549-VEC cells and performed qPCR to quantitate the relative expression of human HPRT (hHPRT) mRNA in the respective organs. We observed that the level of hHPRT mRNA expression was increased three-fold in lung and 1.    compared with respective control VEC cells. Concordantly, BT549-siARTN and MDA-MB-231-siARTN cells exhibited decreased expression of TWIST1 protein compared with the respective siCONT cells. IF microscopy of BT549-ARTN cells also demonstrated increased TWIST1 protein expression as compared with the BT549-VEC cells with TWIST1 being predominantly nuclear localized as previously reported [33]. Concordantly, BT549-siARTN cells exhibited decreased TWIST1 protein expression by IF as compared with the siCONT cells respectively (Figure 4b).
We next performed qPCR in two immortalised but otherwise normal human mammary epithelial cells and 10 different human mammary carcinoma cell lines of differing invasive potential. qPCR gene expression data showed low expression of both ARTN and TWIST1 mRNA in the two normal mammary epithelial cell lines. Both ARTN and TWIST1 mRNA were ubiquitously expressed across different mammary carcinoma cell lines and increased ARTN and TWIST1 mRNA expression correlated with a more invasive cell phenotype (Figure 4c). To further confirm the correlation of ARTN mRNA expression with TWIST1 mRNA expression, we compared the relative expression of ARTN mRNA and TWIST1 mRNA. We observed that TWIST1 mRNA expression was positively correlated with ARTN mRNA expression (Pearson coefficient: 0.68) (Figure 4d).

Co-expression of ARTN and TWIST1 in ER-MC predicts survival outcome
We next determined ARTN and TWIST1 protein expression in a cohort of human ER-MC by IHC and examined the association of expression with survival outcome (relapse free survival (RFS) and overall survival (OS)). IHC analysis showed that both ARTN and TWIST1 protein were highly expressed in ER-MC (Figure 5a). All 94 samples were positive for TWIST1 expression. We therefore stratified the samples according to the level of TWIST1 expression into two categories (low TWIST1 and high TWIST1 expression). ARTN immunoreactivity was stratified according to level of expression into four categories (negative (-), weak (+), moderate (++) and strong (+++) expression) [see additional file 1]. Of those tumors negative for ARTN expression, 60% exhibited strong TWIST1 expression. However, with increasing expression of ARTN in tumors, there was a concordant graded increase in TWIST1 expression with 86.7% of tumors with strong ARTN expression also exhibiting high TWIST1 expression. To further confirm the correlation of expression between TWIST1 and ARTN in mammary carcinoma, we compared the relationship by Spearman's rank correlation coefficient and found a significant positive value (Spearman correlation: r s = 0.21, P = 0.03) ( Table 2). Hence, TWIST1 expression in ER-MC is partially correlated with ARTN expression.
We analyzed survival outcome of patients with tumors with either low or high ARTN or TWIST1. ARTN low tumors were defined as those with negative or weak expression of ARTN and ARTN high tumors exhibited moderate or strong expression of ARTN. Tumors with high ARTN expression tended to exhibit decreased RFS and OS of patients but this tendency did not reach significance in this cohort (Figure 5bi) [see additional file 5i]. A significant correlation of high TWIST1 expression to decreased RFS (P = 0.01) and OS (P = 0.03) of patients was observed (Figure 5bii) [see additional file 5i].
We subsequently analysed survival outcome in patients whose tumors exhibited both low ARTN and low TWIST1 expression compared with those that exhibited high ARTN and low TWIST1 expression. We observed that tumors with low TWIST1 and high ARTN were associated with decreased RFS (P = 0.004) and OS (P = 0.004) (Figure 5biii) [see additional file 5i] compared with those with low TWIST1 and low ARTN. Furthermore multivariate analysis demonstrated a significant association of ARTN high-TWIST1 low expression with decreased RFS (odds ratio: 3.4, 95% confidence interval (0.54-92.95): P = 0.03) and OS (odds ratio: 3.6, 95% confidence interval (0.55-99.84): P = 0.02) at five years compared with low ARTN-low TWIST1 expression (Table 3). Compared with tumors with low ARTN and low TWIST1 expression, tumors that exhibited both high ARTN and high TWIST1 expression were associated with worse RFS (P = 0.002) and OS (P = 0.005) (Figure 5biv (Table 3).

TWIST1 mediates ARTN-stimulated oncogenicity and invasion
To determine the functional significance of ARTN-stimulated TWIST1 expression we employed siRNA to selectively reduce TWIST1 expression in both BT549 and MDA-MB-231 cells with or without forced expression of ARTN. The siRNA targeting TWIST1 efficiently depleted TWIST1 protein expression in both cell pairs ( Figure 6a) whereas transfection of scrambled siRNA (siCONT) had no effect on TWIST1 expression. Depletion of TWIST1 reduced the basal capacity for monolayer proliferation, colony formation in soft agar, 3D matrigel growth and migration and invasion in Transwell assays of BT549-VEC and MDA-MB-23-VEC cells as previously reported for mammary carcinoma cells [34]. Depletion of TWIST1 also prevented or largely abrogated the stimulatory effects of ARTN on monolayer proliferation, colony formation in soft agar, 3D matrigel growth, migration and invasion in transwell assays in BT549-ARTN and MDA-MB-231-ARTN cells (Figures 6b to 6f).

ARTN promoted oncogenicity is mediated via activation of AKT in ER-MC cells
It has been previously reported that ARTN activates AKT to mediate its oncogenic effects in endometrial carcinoma cells [9]. The AKT signaling pathway has been reported to promote TWIST1 expression in different carcinoma cells [35,36]. Furthermore, Activation of AKT and TWIST1 in mammary carcinoma cells has previously been described [15,37]. We therefore determined total AKT protein and pAKT (Ser473) levels in BT549-ARTN and MDA-MB-231-ARTN cells compared with their control VEC cells. BT549-ARTN and MDA-MB-231-ARTN cells exhibited significantly increased phosphorylation of AKT (Ser473) compared with the respective VEC cells without alteration of the total AKT level. Concordantly, BT549-siARTN and MDA-MB-231-siARTN cells exhibited diminished AKT activation compared with the respective siCONT cells (Figure 8a).
We next evaluated whether signalling through AKT was required for ARTN stimulated oncogenicity and invasion in ER-MC cells. Treatment of BT549-VEC cells with AKT Inhibitor IV decreased the basal levels of pAKT but did not affect the total AKT level. ARTN stimulated activation of AKT was also decreased by AKT inhibitor IV without altering the expression level of total AKT (Figure 8b). AKT inhibitor IV decreased the basal level of TWIST1 expression in BT549-VEC cells and also significantly inhibited ARTN stimulated TWIST1 expression in BT549-ARTN cells. AKT Inhibitor IV reduced the basal capacity for colony formation in soft agar, 3D matrigel growth, migration and invasion in transwell assays of BT549-VEC and MDA-MB-23-VEC cells. AKT inhibitor IV also eliminated or largely abrogated the stimulatory effects of ARTN on colony formation in soft agar, 3D matrigel growth, migration and invasion in transwell assays in both BT549-ARTN and MDA-MB-231-ARTN cells (Figures  8c to 8f). Thus, ARTN utilizes AKT to increase TWIST1 expression and promote oncogenicity and invasiveness of ER-MC cells.

ARTN activation of AKT is upstream of TWIST1 mediated oncogenicity in ER-MC cells
We determined if increased TWIST1 expression could substitute for the functional effects of ARTN in ER-MC cells. We generated stable cell clones with forced expression of TWIST1 in BT549 cells (Figure 9a) using a Myc tagged-TWIST1 expression vector [34]. Western blot analysis demonstrated that forced expression of TWIST1 in BT549 cells did not alter AKT expression or ARTN expression as compared with the VEC control cells. Depletion of ARTN by siRNA in BT549-VEC cells decreased the basal level of pAKT but did not alter total AKT or TWIST1 expression as compared with BT549-VEC cells with siCONT ( Figure 9a). Depletion of ARTN by siRNA also decreased pAKT expression in BT549-TWIST1 cells, without altering AKT or TWIST1 expression as compared with siCONT cells. Thus, the activation of AKT precedes ARTN stimulated increases in TWIST1 expression.
We next analyzed the functional effects of siRNA mediated depletion of ARTN in BT549-TWIST1 cells. Forced expression of TWIST1 significantly promoted colony formation in soft agar, 3D matrigel growth, migration and invasion in transwell assays compared with the control BT549-VEC cells. BT549-TWIST1 cells exhibited a 60% increase in soft agar colony formation, 90% increase in 3D matrigel growth, 95% increase in cell migration and 110% increase in cell invasion (Figures 9b to 9e) as compared with the BT549-VEC cell line. Depletion of ARTN by siRNA decreased the basal level of colony formation in soft agar, 3D matrigel cell growth, migration and invasion in transwell assays in BT549-VEC control cells. However, in cells with forced expression of TWIST1, siRNA mediated depletion of ARTN exerted no effect. Thus, the functional effects of ARTN in ER-MC cells are mediated by increased TWIST1 expression.

Discussion
Mammary carcinoma is a heterogeneous disease, which comprises a significant fraction of ER negative subclasses [38]. Given the poor disease prognosis and survival outcomes associated with ER-MC, delineation of prognostic markers to allow better identification and treatment are essential. Herein, we have demonstrated that ER-MC cells with mesenchymal/low-claudin subtype, express high levels of ARTN, which modulate the metastatic potential of cells by promotion of migratory and invasive behavior. Although ARTN is an estrogenregulated gene [8], several reports have previously described the mechanisms of action of different growth factors to support estrogen independence during carcinoma tumor progression [39][40][41][42]. This is concordant with our previous report which demonstrated that ARTN promoted estrogen independent growth of ER +MC cells [8].
ARTN expression has previously been correlated with poorer survival outcome in mammary carcinoma overall [7] and in ER+MC patients treated with tamoxifen [8]. Herein, we have demonstrated that a lower expression of both ARTN and TWIST1 predicted a better survival outcome (100%) in ER-MC in contrast to either low expression of ARTN or TWIST1 alone, whereas high expression of both ARTN and TWIST1 predicted worse survival. That ARTN was only associated with survival outcome in ER-MC with low TWIST1 expression, and that forced expression of TWIST1 negated the effects of ARTN depletion, suggests that TWIST1 is the predominant mediator of ARTN function in ER-MC. Inhibition of ARTN would therefore not be effective as a single therapy in ER-MC with high TWIST1 expression. Acquired increased expression of TWIST1 would also negate the effects of inhibition of ARTN and presumably serve as a mechanism of resistance [43] to prolonged use of ARTN inhibitory strategies. However, based on the results presented herein, inhibition of ARTN, together with strategies to inhibit TWIST1 expression, such as inhibition of PI3K/AKT [44,45] would produce a more desirable outcome than use of either approach alone. Combined inhibition of ARTN and PI3K/AKT/TWIST1 could therefore represent a valuable therapeutic approach in ER-MC mammary carcinoma.
The prometastatic potential of TWIST1 in mammary carcinoma cells is well established [17,29,33]. Previously published reports delineating roles of transforming growth factor (TGF)-β and TWIST1 promoting EMT and metastasis [5] suggested that ARTN, a TFG-β superfamily member could also potentially regulate TWIST1 to promote EMT and metastasis [17]. TWIST1 can be regulated by multiple extracellular factors through PI3K/AKT, Wnt1 and NF-B pathways in invasive carcinoma [46][47][48] concordant with our observations that activation of AKT by ARTN leads to increased expression of TWIST1 in ER-MC cells. Elevated expression of TWIST1 has been correlated with resistance to the microtubule targeting anticancer drugs taxol and vincristine [43] and TWIST1 expression is also increased in MC stem cell-like populations [49]. That increased ARTN expression induces genes associated with EMT and stimulates metastasis of ER-MC suggests that it may regulate a cancer stem cell-like subpopulation of cells that possess an enhanced ability to metastasize to distant organs. Such a notion is concordant with a previous study [50] demonstrating that mammary carcinoma cell lines (including MDA-MB-231) contain functional cancer stem cells with metastatic capacity. In accordance, we have observed that forced expression of ARTN in ER-MC cells increased a sub-population of cells with cancer stem cell-like behavior and enhanced in vivo tumor initiating capability (manuscript in preparation). ARTN and TWIST1 may therefore functionally synergize in aspects of mammary carcinoma cell behavior other than EMT and metastasis reported herein.