Background
Epithelial-mesenchymal transition (EMT) is a reversible process of cellular reprogramming that plays an important role in normal development and injury response. In the last decades more interest has focused on the role of the oncogenic type of EMT in metastasis. Phenotypic changes occur during EMT, such as the loss of epithelial proteins including E-cadherin, cytokeratins and claudins, and the acquisition of mesenchymal proteins such as N-cadherin and vimentin [
1]. Following these changes at the level of gene expression, polarized epithelial cells undergo morphological changes into spindle-shaped migratory mesenchymal cells [
2]. Acquisition of a mesenchymal-like phenotype is usually associated with decreased cellular proliferation [
3] and supports the escape of cells from the primary tumor site. However, phenotypic reversion to epithelial-like along with increased proliferation is essential for disseminated tumor cells to successfully colonize distant sites. The initiation and regulation of EMT is ensured by various growth and differentiation factors that orchestrate the activation of different signaling pathways and a wide range of transcription factors. Among these, transforming growth factor beta (TGF-β) has received much attention as a major inducer of EMT during embryogenesis, fibrosis and cancer progression [
4,
5]. TGF-β-induced signaling may activate two independent pathways, Smad-dependent (transduction of signal by Smad proteins) and Smad-independent cascades including MAPK, PI3K/Akt/mTOR and Rho GTPase [
6‐
10].
Human Anterior gradient 2 protein (AGR2) belongs to the family of Anterior gradient proteins, originally identified in Xenopus laevis as a potential secretory protein that is highly expressed in
Xenopus eggs [
11]. AGR2 is classified as a member of the protein disulfide isomerase family (PDI), a group of endoplasmic reticulum (ER)-resident proteins [
12]. As a protein disulfide isomerase, AGR2 is thought to be involved in protein folding and maturation of client proteins (e.g. mucins MUC2, MUC5 and MUC1) and in maintaining ER homeostasis [
13‐
17]. ER stress caused by accumulation of misfolded proteins may stimulate the unfolded protein response that in turn increases the expression of AGR2. Following its upregulation, AGR2 participates in the attenuation of degradation processes and prevents the induction of apoptosis, leading to increased cellular survival [
12,
14,
18].
Alterations of AGR2 expression in cancer cells are reflected by the upregulation of cellular proliferation, tumor growth, inhibition of p53 and increased cellular survival, invasiveness and migration [
19‐
21]. Proteins of AGR family were originally identified as estrogen receptor regulated [
22,
23]. However, subsequent studies showed the contribution of other hormone-dependent as well as hormone-independent pathways in regulating AGR2 expression [
24‐
26]. Pro-survival oncogenic pathways responsible for regulation of AGR2 expression along with the involvement of AGR2 in cellular adhesion and interaction with the extracellular matrix indicate the important function of AGR2 in the migration and invasiveness of cancer cells [
27,
28]; however the precise mechanism remains to be elucidated.
To investigate the effect of TGF-β treatment on AGR2 expression, we used four different cancer cell lines expressing various levels of EMT markers in order to generalize the role of AGR2 in response to TGF-β treatment. Although these cells differed in classical EMT markers, AGR2 expression decreased in all tested cell lines in association with acquisition of a mesenchymal-like phenotype, as documented by changes in the levels of epithelial and mesenchymal markers. In contrast, increased expression of AGR2 was accompanied by an epithelial-like phenotype. Taken together, these data underscore the function of AGR2 in maintaining the epithelial phenotype and its role in re-establishing an epithelial phenotype during the development of metastasis.
Methods
Cell lines and reagents
Cell lines
A549 (CCL-185), H1299 (CRL-5803) (lung adenocarcinoma), BT-474 (HTB-20) and MCF-7 (HTB-22) (estrogen receptor-positive breast cancer), Panc1 (CRL-1469) (pancreatic adenocarcinoma) and HEK-293 (CRL-1573) (embryonic kidney epithelial cells) were obtained from ATCC and maintained in DMEM supplemented with 10% FBS, 1% pyruvate and L-glutamine at 37 °C in a humified atmosphere of 5% CO2. Unless otherwise stated, cells were grown to 70–80% confluence prior to treatment. TGF-β was added to a final concentration of 1 ng/ml for 24 h or as indicated. For inhibition of Erk1/2, cells were treated for 2 h with PD98059 prior to TGF-β treatment for the next 24 h.
Transfection was carried out using 2 μg of plasmid or 50 pmol of siRNA per million cells. To silence AGR2, cells were transiently transfected with siRNAs against AGR2 or untargeted siRNA as control (all from Dharmacon, ThermoFisher Scientific). The Flp-In™ System (Invitrogen) was used to generate H1299-LZ4 cells containing a single integrated Flp Recombination Target (FRT) site. The coding sequence of the human AGR2 gene was stably inserted into this site using Flp recombinase mediated site-specific DNA recombination to give H1299-LZ4-AGR2 cell line. Plasmid pcDNA3-AGR2 was used to express AGR2 in transiently transfected cells.
Antibodies
Akt2 (5B5), p-AKT (S473; 736E11), p44/42 MAPK (Erk1/2; 137F5), p-p44/42 MAPK (T202/Y204, D13.14.4E), p-Smad2 (S465/467; 138D4), Smad4, SNAI2, ZEB1 (all Cell Signaling Technology); AGR2 (K-31, in-house); AGR2 (1C3, Abnova); vimentin (V9, Dako); E-cadherin (NCH38, Dako; HECD1, Abcam; Cell Signaling); N-cadherin (3B9, Invitrogen); β-actin (C4, Santa Cruz Biotech); Alexa Fluor 488 goat anti-mouse IgG, Alexa Fluor 532 goat anti-rabbit IgG (both Abcam); α-tubulin (AA13, Sigma); HRP-conjugated swine anti-rabbit and HRP-conjugated rabbit anti-mouse (both Dako).
Reagents
Human TGF-β1 (R&D Systems), protease and phosphatase inhibitors (Sigma-Aldrich), PD98059 (Cell Signaling Technology), SB431542 (Sigma). CytoPainter Phalloidin-iFluor 594 Reagent (Abcam) was used for phalloidin staining.
CRISPR/Cas9
The oligonucleotide used to generate gRNAs for the human
AGR2 gene (AGAGATACCACAGTCAAACC) was designed using the CRISPR design web tool (
crispr.mit.edu
). The GFP-scrambled sequence (AACAGTCGCGTTTGCGACTGG) used as a control was described earlier [
29]. Oligonucleotides were cloned into the LentiCRISPR-v2 vector using Esp3I restriction cloning.
A549 cells (1 × 10
6) were transfected with 2 μg of LentiCRISPR-v2_AGR2 and 2 μg of LentiCRISPR-v2_scrambled (control), respectively. Two days after transfection, puromycin (final concentration 0.5 μg/mL) was added to the cells and after 3–4 weeks the pool of resistant cells was seeded as single colonies in 96-well plates. After 2–3 weeks, the clones were tested for AGR2 expression using Western blot. Two clones, named A549 KOAGR2 G2 and A549 KOAGR2 G9 with undetectable expression of AGR2 were selected for further experiments. For verification and identification of the inserted mutation, the genomic DNA of these clones was isolated, the target region for Cas9 in the coding sequence of
AGR2 gene was amplified and sequenced (Additional file
1: Figure S1). Mutations causing frame shift were detected in both clones.
Gene expression
Total RNA was isolated using TRIzol reagent (Invitrogen). The cDNA was synthesized by reverse transcriptase (Life Technologies). q-PCR was performed using SYBR Green MasterMix (Roche), TaqMan Universal PCR MasterMix was used for 18S rRNA (Life Technologies) representing a parallel endogenous control to GAPDH. All samples were analyzed in triplicates. The primer sequences are described in Additional file
2: Table S1.
Immunofluorescence microscopy
Cells were fixed with 3% paraformaldehyde and permeabilized with 0.1% Triton X-100. Then the cells were blocked with 1% BSA diluted in PBS supplemented with 0.1% Tween (BSA-PBS-T) for 1 h and incubated overnight with primary antibody diluted in 1% BSA-PBS-T. Next day, coverslips were washed with PBS and probed for 1 h with Alexa Fluor 488 goat anti-mouse IgG or Alexa Fluor 532 goat anti-rabbit IgG (Abcam). After washing, cells were incubated with DAPI diluted in PBS (1:2500) for 5 min. Coverslips were mounted with Vectashield and images captured using an Olympus BX41 microscope. Images were analyzed with CellSens software (Olympus).
Western blot analysis
Cell lines treated for the indicated time periods were subjected to SDS-PAGE and Western blot analysis. Prior to harvesting, cells were washed twice with cold phosphate-buffered saline (PBS) and scraped into NET lysis buffer (150 mM NaCl, 1% NP-40, 50 mM Tris pH 8.0, 50 mM NaF, 5 mM EDTA pH 8.0) supplemented with protease and phosphatase inhibitor cocktails according to the manufacturer’s instructions (Sigma). After SDS-PAGE, samples were transferred to nitrocellulose membranes which were blocked with 5% milk in PBS supplemented with 0.1% Tween (PBS-T) for 1 h at room temperature and then incubated overnight at 4 °C with primary antibodies diluted in 5% milk with 0.1% Tween. Next day, membranes were washed three times with PBS-T and probed with horseradish peroxidase-conjugated secondary antibodies (1:1000) for 1 h at room temperature. Chemiluminiscent signals were developed using ECL solution (0,5 M EDTA pH 8.0, 90 mM coumaric acid, 1 M luminol, 200 mM Tris-HCl pH 9.4, Na-perborate × 4H20, 50 mM Na-acetate pH 5.0) and visualized with GeneTools software (Syngene). Either α-tubulin or β-actin was used as a loading control and protein expression was normalized in relation to these proteins.
Invasion assay
Real-time cell analysis (RTCA) of invasion was performed on the xCELLigence Real-Time Cell Analyzer DP device (Roche) as described in the supplier’s instruction manual. Briefly, cells were added to the upper chamber of a two-chamber device separated by a porous membrane (the CIM-16 plate) through a solid matrix. Electrical impedance was displayed as a dimensionless parameter termed cell index reflecting the tumor cells’ invasive capacity. Prior to cell seeding, the surface of the upper chamber was covered with a monolayer of Matrigel. 30,000–50,000 cells/well suspended in serum-free culture medium were then seeded in the upper chamber. Cell indexes were measured every 15 min for up to 48 h with the RTCA software (version 1.2, Roche).
Scattering assay, adhesion assay, cell-matrix adhesion assay
A549 and A549 KOAGR2 cells were allowed to grow as distinct colonies by seeding at 1000 cells/10-cm plate. The effect of AGR2 knockout on cell scattering was evaluated at day 5 and photographed under a phase contrast microscope. For adhesion assays, 25,000 cells were seeded per well on 96-well plates. Cells were left to adhere for 2 h at 37 °C and then adherent cells were fixed and stained. For cell-matrix adhesion assays, each well of 24-well plates was coated with Matrigel and left to solidify for 4 h at 37 °C. Then 50,000 cells per well were seeded. After incubation for 90 min, unbound cells were removed and adherent cells were fixed and stained with crystal violet. To quantify cell attachment, each well was solubilized with 2% SDS and absorbance was measured at 595 nm using an automated plate reader.
Statistical analysis
One-way analysis of variance (one-way ANOVA) was used to determine differences between the means of studied samples. P < 0.05 was considered as significant.
Discussion
Epithelial-mesenchymal transition is widely accepted as the essential step in the initiation of metastasis development and cancer cell dissemination from primary tumors [
3,
34]. However, to accomplish metastatic growth, mesenchymal migratory cells have to be reprogrammed back to obtain the original epithelial phenotype. The reversion occurs during the process known as mesenchymal-epithelial transition, representing a second crucial step in carcinogenesis that mediates tissue colonization and the development of metastatic tumors at secondary sites [
35]. Despite considerable progress in research focused on EMT/MET crosstalk, the whole process is still not fully understood.
In recent years, the role of AGR2 in tumor development and progression is becoming more and more intensively studied [
13,
36]. The contribution of AGR2 to malignant transformation [
21], drug resistance [
19,
37,
38] and metastasis development [
30,
31] have already been reported, however the mechanism of action, as well as the scope of AGR2 functions remain unclear. In the present study, we identified AGR2 as an important factor contributing to the maintenance of the epithelial phenotype of tumor cells.
We found that AGR2 expression positively correlates with the expression of the epithelial marker E-cadherin, while TGF-β-induced reduction of AGR2 was concomitant with the classical features of mesenchymal cells such as the loss of E-cadherin, induction of N-cadherin and morphological changes arising from cytoskeleton reorganisation including diffuse cytoplasmic distribution of vimentin and re-localization of actin. These data demonstrate an important role of AGR2 in maintaining epithelial phenotypes. Interestingly, transient AGR2 silencing as well as stable knockout mimicked the initiation of EMT. Accordingly, exogenous expression of AGR2 in two different cellular models confirmed the capacity of AGR2 to revert a mesenchymal phenotype to an epithelial phenotype, indicating not only the important role of AGR2 expression in EMT, but also that elevated AGR2 expression may prevent the acquisition of a mesenchymal phenotype by cancer cells. A key role of AGR2 in EMT is also supported by the similar morphological changes in A549 cells exposed to TGF-β or with silenced AGR2, showing a shift from a classical cobblestone epithelial morphology to a fibroblast-like morphology and F-actin rearrangement. These findings representing a hallmark of epithelial cancer metastasis development are also consistent with our experiments confirming involvement of AGR2 in preventing cellular invasiveness and maintaining cell-cell junctions along with adhesion to ECM. Interestingly, reduced cell adhesion was also observed after abrogation of AGR-2 expression in prostate cancer cells [
39].
The validation of TGF-β as a potent suppressor of AGR2 expression in our panel of tumor cell lines is consistent with the previous findings that have characterised
AGR2 as a TGF-β responsive gene [
15]. The important role of Smad4 in triggering AGR2 downregulation in human pancreatic cancer cells exposed to TGF-β has recently been shown [
15]. Subsequently, AGR2 downregulation was determined by immunohistochemical staining and correlated with vimentin expression and reduced expression of membranous E-cadherin in pancreatic precursor neoplastic lesions, as well as pancreatic ductal adenocarcinomas [
40]. Additionally, our data indicate that TGF-β suppressive function on AGR2 expression is not exclusively dependent on the Smad-canonical pathways. Some previous reports suggested that MAPK signaling positively regulates AGR2 expression [
26], however TGF-β-induced MAPK activation in our cell lines led to AGR2 downregulation. Consistent with recent literature [
26,
41], we confirm the role of Erk1/2 signaling pathway in regulating AGR2 by showing that inhibition of Erk1/2 reduces AGR2 protein levels. Interestingly, although individual treatment with either Erk1/2 inhibitor or TGF-β reduces AGR2 expression, their combination results in negligible changes in AGR2 protein levels (Fig.
1). To elucidate this phenomenon, we found that inhibition of the Erk1/2 pathway in A549 cells does not affect activation of Smad signaling in response to TGF-β but rather impairs nuclear import of the Smad complex. The blockade of nuclear translocation may in turn prevent the suppressive effect of TGF-β/Smad signaling on AGR2 expression. In accordance, previously published data showed that treatment with the MAPK signaling inhibitor, PD98059, blocked the formation of Smad complex [
42] and impaired binding of Smad complexes to Smad-responsive elements [
43,
44]. Thus, we suggest that TGF-β dependent induction of the MAPK signaling cascade is essential for the transcriptional activity of the Smad-dependent pathway that results in TGF-β mediated AGR2 downregulation.
In addition to identifying the mechanism responsible for TGF-β-dependent repression of AGR2, we also revealed that AGR2 actively interferes with the EMT process to maintain an epithelial phenotype via inhibition of EMT-inducing transcription factors ZEB1 and SNAI2. Another potential mechanism by which AGR2 contributes to blocking EMT is the inhibitory effect on p38 MAPK signaling, which is frequently involved in triggering EMT [
19,
45].
Very recently, several studies showed that extracellular AGR2 promotes epithelial morphogenesis and tumorigenesis by interruption of adherens junctions, disruption of basal laminin and activation of fibroblast-associated cancer invasion [
37,
46]. That elevated levels of secreted AGR2 mediate more aggressive cellular phenotypes associated with loss of cellular polarity, remodeling of extracellular matrix and increased invasiveness may appear to be inconsistent with our results. However, it should be noted that our results describe the role of intracellular AGR2, which protects the epithelial cellular phenotype by preventing EMT induction, in accordance with other reports focused on the function of intracellular AGR2 [
39,
47]. Taken together, the role of AGR2 in EMT is dependent on its localization, which seems to be responsible for a set of different impacts on cancer cell migration, invasiveness and metastasis development. However, the precise mechanism(s) regulating the levels of intracellular and extracellular AGR2 and/or their balance remain to be further investigated. Proteins with opposite functions depending on their presence inside or outside cells were already described in detail, for example heat shock proteins, namely Hsp70, or calcium-binding proteins S100A8 and S100A9 [
48‐
50]. Moreover, the well-studied keeper of epithelial phenotype E-cadherin showed pro-metastatic function when cleaved and localized in the extracellular space [
51].
Conclusions
In summary, our findings provide new insights into the regulation of AGR2 as well as AGR2 functions with respect to TGF-β signaling, clarifying the role of AGR2 in EMT/MET. Our data confirm AGR2 as a keeper of epithelial phenotypes, showing a positive role of AGR2 in the regulation of the epithelial marker E-cadherin and a negative effect on the mesenchymal markers vimentin and N-cadherin. In primary tumors, AGR2 is predominately responsible for increased proliferation and growth of malignant cells, but later during the metastatic cascade AGR2 contributes to successful settling and adhesion of disseminated cells in secondary sites, their adaptation to the tumor milieu and stimulation of secondary tumor growth. In relation to molecular changes associated with EMT and MET phenotypes, AGR2 appears to be a key regulator of these processes indicating a dual role of AGR2 in cancer.