Main

Epithelial–mesenchymal transition (EMT), paradigmatic of the cell plasticity concept, is involved in embryonic development as well as in several pathological conditions, including organ fibrosis and cancer progression.1 During EMT, epithelial cells loose polarization, disassemble cell–cell junctions, and switch to a mesenchymal-type gene expression program. Cells undergoing EMT start to synthesize fibronectin, collagen type I, and vimentin acquiring a fibroblast-like phenotype. In some cases, cells undergoing EMT express α-smooth muscle actin (α-SMA) assuming the myofibroblastic phenotype. Several signaling factors, including TGFβ1, trigger this process.1, 2, 3

It has been shown that TGFβ1 downregulates the spermidine synthase gene4 and that fibroblasts treated with difluoromethylornithine (DFMO), a specific inhibitor of the enzyme ornithine decarboxylase (ODC), responsible in mammalian cells of ornithine decarboxylation to putrescine, exhibit a significant increase in c-myc, matrix metalloproteinase (MMP)1, MMP2, TGFβ1, and collagen type I mRNAs,5 suggesting that intracellular polyamine levels affects transcription of EMT-related genes.

Polyamines are organic polycations present in all living organisms. Higher polyamines are synthesized by the sequential addition of aminopropyl groups to putrescine by a reaction involving S-adenosyl methionine decarboxylase (SAMDC) as well as spermidine and spermine synthases. ODC and SAMDC are considered rate-limiting enzymes in polyamine biosynthesis. In addition to these multi-level controls of synthesis, uptake and efflux into extracellular environment co-operate in the fine-tuning of intracellular levels of polyamines. The availability of specific inhibitors of ODC and SAMDC has allowed to establish the essential role of polyamines in many processes, including cell growth,6, 7, 8, 9 differentiation,10, 11 cell adhesion,12 induction of apoptosis,13 and signal transduction.14, 15, 16 On this basis, DFMO administration has been suggested for the treatment of infectious diseases,17 protozoal parasitism18 or, with modest results, cancer19 and hyper proliferative diseases.20

Polyamines may act as ligands of DNA, RNA, phospholipids, nucleotide triphosphates, or proteins and act as effectors of gene expression and signal transduction. Polyamines are also substrates for transglutaminase-2 (TG-2), which catalyses the covalent incorporation of several low molecular weight amines into protein in the form of γ-carboxyl group of a peptide-bound glutamic acid. Key cellular processes are modulated by the amount of both free and protein-conjugated polyamines formed by TG activity.21

The aim of this study was to elucidate whether polyamines have a role in EMT modulation of Madin-Darby canine kidney (MDCK) cells derived from distal tubular epithelial cells.

MATERIALS AND METHODS

Cell Culture

MDCK cells (ATCC, Middlesex, UK) were cultured in Dulbecco's modified Eagle medium (DMEM; Sigma, Saint-Louis, USA) and supplemented with 10% fetal bovine serum (FBS). Polyamine-related experiments were instead performed in RPMI 1640 (Gibco BRL, Paisley, UK) supplemented with 10% horse serum (HS) with cells cultured in 150 cm2 flasks. Addition of HS to all experimental conditions instead of FBS is an established procedure aimed at minimizing cytotoxic effects due to high levels of amine-oxidases in FBS, which, on the contrary, is very low in HS. Addition of HS instead of FBS is necessary because of the experimental protocol design that, in some conditions, requires addition of polyamines (i.e. to counteract effect of DFMO). FBS high-level content of amino-oxidases can in fact generate significant amounts of hydrogen peroxide, ammonia, and aldehydes through polyamine degradation, leading to cell injury and death.22 Cells were plated and pre-treated, when scheduled, alternatively with 3 mM DFMO for 24 h, leading to the depletion of endocellular polyamines, or DFMO+ a mix of polyamines (P, containing 10 μM putrescine and 10 μM spermidine) to validate the specific effect of DFMO. Medium was changed after 24 h, and cells were exposed to the following experimental conditions: control (RPMI+10% HS), DFMO (RPMI+10% HS+3 mM DFMO), TGFβ1 (RPMI+10% HS+10 ng/ml TGFβ1), DFMO+TGFβ1 (3 mM DMFO and 10 ng/ml TGFβ1, respectively, in RPMI+10% HS) or DFMO+TGFβ1+P (3 mM DMFO and 10 ng/ml TGFβ1, respectively, plus polyamine mix, in RPMI+10% HS). Cells were allowed to grow up to 96 h and processed for western blot or immunofluorescence. Cells pre-treated with 3 mM DMFO for 24 h were used to evaluate proteins linked to polyamines that were assessed through the administration of 100 μM EZ-link penthylamine-biotin (BPA, Pierce, Rockford, USA) for further 6 h. Cells were then finally processed for two-dimensional (2D) electrophoresis or immunofluorescence, in this case, BPA administration was performed in the absence or presence of a specific siRNA against TG-2.

Assessment of TGFβ1 mRNA and Protein in Polyamine-Depleted Cells

TGFβ1 mRNA presence was routinely assessed through RT-PCR in control and DFMO-treated cells. TGFβ1 protein was evaluated with two different methods: secreted protein was evaluated on 5-day treated cells pre-treated on day 4 with Brefeldin A (Sigma-Aldrich), this to prevent cytokine secretion. Cells were then fixed, stained with an anti-TGFβ1 antibody (Clone TB21, Serotec GmbH, Dusseldorf, Germany) and detected with corresponding FITC-conjugated secondary antibody. Cells were then observed through a FACS Excalibur (Becton & Dickinson, Mountain View, USA). TGFβ1 in the supernatant was dosed at day 5 of treatment through application of a specific ELISA test (Human TGFβ1 Instant ELISA, Bender MedSystems, Wien, Austria). In both cases, positive control was obtained by treatment of MDCK cells with 4.2 μM Cyclosporin as described earlier by Feldman et al.23

TG-2 Knockdown

The TG-2 knockdown was performed by the administration of a specific siRNA addressed to the known incomplete canine transcript sequence (accession N° DR103953) targeting a 23 nucleotide long stretch starting at 573 nucleotide. The sequences of siRNA were: CAAGCAGGCTTTAGAGCCACAAA (sense strand) and UGUGGCUCUAAAGCCUGCUUGAA (antisense strand). The two oligonucleotides were dissolved in 20 mM Tris-HCl buffer, 100 mM NaCl, pH 7.5 and then annealed by slow cooling after 1 min denaturation in boiling water bath. The siRNA was administered to cultures in a 12-well plate at 100 nM final concentration using a lipoplexed vehicle GeneSilencer (GenLantis, San Diego CA, USA) according to the manufacturer's instructions in serum-free medium 8 h later, cells were then washed and used for BPA administration experiments. As control, the siRNA addressed to fluorescent protein GFP was administered to parallel cultures. Sense strand: CAAGCUGACCCUGAAGUUCUAA; antisense strand: AGAACUUCAGGGUCAGCUUGAA.

Matrix Metalloproteinase Assay

MMP evaluation was performed at 120 h on extracellular medium of cells submitted exactly to the above-described experimental protocol. Medium was changed 24 h before medium sampling. Briefly, samples were loaded onto a standard 7.5% acrylamide gel containing 0.1% gelatin and run at 120 V for 2 h. Gelatin digestion was counterstained with Coomassie Blue R-250 (Sigma-Aldrich). Zymograms were loaded based on cellular protein concentration as evaluated by BCA assay.

Polyamine Content Evaluation

Cells were scraped in 0.2 M HClO4 and subsequently centrifuged at 15 000 g. Supernatant was dansylated and analyzed by HPLC as described earlier.24 Detection was performed with a fluorimeter (Waters 2475, Waters Corporation, Watford, UK).

Apoptosis Assessment in Polyamine-Depleted Cells

Apoptosis was assessed in control and polyamine-depleted cells through Annexin V/propidium iodide labeling and subsequent FACS analysis (FACS Excalibur).

2D Electrophoresis and Western Blot

Cells collected for western blot and 2D were trypsinized, counted, and washed in TBS. For western blot, 4 × 104 cells/μl were resuspended in Laemmli sample buffer-containing phosphatase and protease inhibitor cocktail. Samples were then boiled for 10′ and proteins concentration was determined using BCA assay.

Western blot samples (10–50 μg) were load onto a SDS-PAGE (T% 8–16) in a Protean II XI system (Bio-Rad, Hercules, CA, USA), transferred onto nitrocellulose membrane (Protean BA, Schleicher & Schuell, Dassel, Germany) with a Novablot semidry system, 3% w/v BSA saturated in TBS and incubated separately with e-cadherin (Santa Cruz Biotechnologies, CA, USA, 1:1000), fibronectin (Santa Cruz, 1:1000), α-SMA (kindly provided by Professor G Gabbiani, 1:750), β-tubulin (AbCam, 1:1000), GRP78 (Santa Cruz Biotechnology, 1:1000), GRP94 (Sigma, 1:1000), HSP90 α/β (Santa Cruz Biotechnology, 1:1000) α-enolase (AbD Serotec, 1:1000) or vimentin (Novocastra, 1:1000) in 3% w/v BSA in TBS-Tween 0.15% v/v (TBS-T). Membranes were then rinsed in TBS-T and incubated with corresponding secondary antibodies. Chemioluminescence was used for immunodetection. Images were digitalized by mean of VersaDoc 4000 (Bio-Rad) and analyzed with QuantityOne software (Bio-Rad).

For analytical (100 μg) and preparative (400 μg) 2D electrophoresis, samples were solubilized in the reduction/alkylation solution, ie 8 M urea, 4% CHAPS, 5 mM tributyl-phosphine, 20 mM iodoacetamide (IAA), 40 mM Tris, and 0.1 mM EDTA pH 8.5. Before isoelectric focusing (IEF), samples were incubated in this solution for 3 h, to allow proper reduction and alkylation. To prevent over alkylation during the IEF step, excess of IAA was neutralized by adding an equimolar amount of DTT. Samples were then cleaned using an ice-cool solution consisting of tri-n-butyl-phosphate: acetone: methanol (1:12:1). Fourteen milliliters of this mixture were added to each sample to reach a final acetone concentration of 80% (v/v) and incubated at 4°C for 90′. The precipitate was pelleted by centrifugation at 2800 g for 20′ at 4°C. After washing with the same de-lipidizing solution, it was centrifuged again and then air- dried. Finally, samples were dissolved in the focusing solution, ie 7 M urea, 2 M thiourea, 4% CHAPS, and 50 mM dithioerythritol (DTE) and load onto home-made non-linear pH 3–10 strips (18 cm long, 3 mm wide, 0.5 mm thick).25 Strip re-swelling was carried out overnight at room temperature in the focusing solution, ie 7 M urea, 2 M thiourea, 2% (w/v) CHAPS, 15 mM DTE, and a 0.6% (v/v) carrier ampholyte cocktail, containing 40% of the pH 3.5–10 and 60% of the pH 4–8 intervals.

The proteins were focused at ≤50 mA per strip at 20°C, using progressively increasing voltage for a total of 80 000 Vh. Equilibration step was carried out for 30′ in the equilibration buffer, ie 6 M Urea, 50 mM Tris-HCl pH 8.8, 2% w/v SDS, 30% v/v glycerol, and 0.01% w/v bromophenol blue. In second dimension was used a SDS-PAGE (T% 8–16) in a Protean II XI system (Bio-Rad).

Blue silver colloidal staining26 or silver staining were used for preparative or analytical proteins detection. Images were digitalized by mean of GS-800 (Bio-Rad) and analyzed with PDQuest software (Bio-Rad).

Procedure described above was also adopted for 2D western blot of BPA-labeled proteins (BPA+). Briefly, proteins were transferred to nitrocellulose membrane, saturated in 3% w/v BSA in TBS, and incubated with neutravidin HRP-conjugated (Pierce Biotech., Rockford, USA) in 1% w/v BSA in TBS-T. Chemioluminescence was used for immunodetection. Images were analyzed with PDQuest software (Bio-Rad); 2D experiments were performed in triplicate.

Statistical Analysis

Densitometric values of 2D gel and western blot were analyzed using non-parametric Mann–Whitney U test. In the case of 2D gels, spots with at least twofold increase relative to control with corresponding two-tailed P-value <0.05 were accepted as statistically significant. Mass spectrometry identification was limited to BPA+ spots retaining these characteristics. Densitometric western blot differences were considered statistically significant at two-tailed P-value <0.05.

Tryptic Digestion and Protein Identification by LC-ESI MS-MS/MS

Protein spots were excised from gel, rinse in 50% v/v acetonitrile (ACN) in 5 mM ammonium bicarbonate pH 8.9 until full destaining, rinse twice in 100% v/v ACN, and briefly rinse in 1 mM CaCl2 and 100 mM ammonium bicarbonate pH 8.9. Enzymatic digestions were performed using trypsin in 100 mM ammonium bicarbonate buffer pH 7.8 overnight at 37°C. After incubation, the reaction was quenched by the addition of formic acid to pH 2.

All mass spectrometric measurements were performed using a LTQ linear ion trap mass spectrometer (Thermo Electron, San Jose, USA) coupled to a HPLC Surveyor (Thermo Electron) and equipped with a Jupiter C18 column 250 × 1 mm (Phenomenex). Peptides were eluted from the column using an acetonitrile gradient, 5% B for 6′ followed by 5 to 90% B within 109′ (eluent A: 0.1% formic acid in water; eluent B: 0.1% formic acid in acetonitrile) at flow rate of 50 μl/min. The column effluent was directed into the electro spray source. The spray voltage was 5.0 kV. The ion trap capillary was kept at 200°C and the voltage at 2.85 V. Spectra were obtained in automated MS/MS mode: each full MS scan (m/z 400–1800) was followed by five MS/MS of the most abundant ions. The ions analyzed this way were automatically excluded for 30′′.

The raw data acquired by the mass spectrometer were converted in a peak list file for database search using Extract_msn in Bioworks 3.3.1 Sp1 with default parameters for LTQ spectra.

Protein identification was performed using SEQUEST software 3.3.1 from Thermo Electron, operating on a 10-processor computer cluster (AETHIA, Torino, Italia) and searched against a Canis familiaris protein database (no. of protein entries 41427) downloaded on 11/2008 by NCBI using the taxonomy browser tool. Peptide MS/MS assignments were filtered following very high stringent criteria: Xcorr ≥1.9 for the singly charged ions, Xcorr≥2.2 for doubly charged ions, and Xcorr≥3.7 for triply charged ions, peptide probability ≤0.01, Delta Cn≥0.1, and Rsp ≤4 according to the HUPO criteria.27 The mass tolerance for precursor ions was set to 2 amu and the mass tolerance for fragment ions was 1 amu. To identify the largest panel of peptides, the option no enzyme was used for the in silica digestion of human databases, so every combination of human peptides was evaluated. Moreover, the search parameters no included variable or fix modifications.

Immunofluorescence Analysis

Indirect immunofluorescence analysis was performed as described earlier.28 Briefly, cells were gently rinsed in warm RPMI/2% HEPES buffer and fixed for 20′ at 37°C in 1% paraformaldehyde dissolved in RPMI/2% HEPES buffer. Cells were then rinsed in PBS and stained for e-cadherin (1:100), fibronectin (1:100), zonula occludens 1 (ZO-1, 1:100), TG-2 (clone CUB 7402, Thermo Fisher Scientific, Fremont, USA), GRP78 (ABcam, Cambridge, USA) or α-SMA (1:50). Detection was performed with corresponding Alexa 488 or Alexa 568 conjugated secondary antibody. Nuclei were counterstained with DAPI. Images were acquired on a confocal Apotome Axiovert microscope (Carl Zeiss AG, Jena, Germany).

RESULTS

DFMO Administration Reduces Intracellular Polyamine Content in MDCK Cells

DFMO administration reduces MDCK intracellular levels of polyamines (Figure 1).

Figure 1
figure 1

Bar graph showing the effect of difluoromethylornithine (DFMO), TGFβ1, DFMO+TGFβ1, or DFMO+TGFβ1+ administration of exogenous polyamines on total (ie including putrescine, spermidine, and spermine) intracellular polyamine levels. DFMO decreased polyamines to ∼20% compared with controls; TGFβ1 alone did not statistically influence polyamine content, whereas coupled administration of DFMO and TGFβ1 decreased polyamines. Exogenous polyamine addition to the culture medium restored values comparable to control. ***P<0.001 values are referred compared with control conditions.

Full size image

DFMO administration reduced intracellular polyamine content to ∼20% of controls values (ie 5.922±0.296 nmol/mg of total protein extract in DFMO-treated cells vs 29.215±1.416 nmol/mg in control cells, P<0.001). TGFβ1 alone did not statistically affect polyamine content, whereas combined administration of DFMO and TGFβ1 reduced intracellular polyamines to 9.635±0.482 nmol/mg (P<0.001 vs control value). Exogenous polyamine addition to the culture medium restored values comparable to those detected in control cells (31.179±1.559 nmol/mg).

Polyamine Depletion Enhances EMT in MDCK Cells

Polyamine-depleted MDCK cells showed a significant morphological rearrangement into large polygonal cells (Figure 2b, phase contrast panel), associated with a marked redistribution of E-cadherin staining (decreased at plasma membrane levels and increased in the cytoplasm) (Figure 2g). Immunofluorescence also showed some scattered cells exhibiting positivity for fibronectin and an increase in α-SMA positive staining that were limited to sparse cytoplasmic rod-like structures (Figure 2n and s). Combined administration of DFMO and TGFβ1 resulted in a morphological re-arrangement at confluence (Figure 2c); very similar to the one induced by TGFβ1 alone. Treatment up to 96 h resulted in a reduced expression of the epithelial markers E-cadherin, as shown by western blot (Figure 3, panel A; Supplementary Figure 1), and ZO-1 (data not shown). Immunofluorescence analysis revealed that combined treatment resulted in the downregulation of plasma membrane E-cadherin (Figure 2i) and acquisition of a dense network of fibronectin (Figure 2p) associated to an increased expression of α-SMA in stress fibers in each cell (Figure 2u), whereas the treatment with TGFβ1 alone resulted in an increased positive staining for α-SMA detectable in only in 20 to 30% of cells, this feature being mainly limited to the external edge of MDCK cell islets. Exogenous polyamine addition to the medium prevented all these phenomena: E-cadherin expression on plasma membrane was maintained (Figure 2l), whereas fibronectin and α-SMA expression was negligible (Figure 2q and v, respectively). Addition of polyamines alone showed no influence on expression of all considered markers, whereas addition of polyamines to TGFβ1 showed a moderate inhibition of TGFβ1-stimulated phenotypical modulation (Supplementary Figure 1).

Figure 2
figure 2

Phase contrast (a–e) or immunofluorescence (f–v) of MDCK cells stained for e-cadherin (f–l), fibronectin (m–q) or SMA (r–v) treated with DFMO, TGFβ1 (TGFβ), DFMO+TGFβ1 (D+T), or DFMO+TGFβ1+ exogenous polyamines (D+T+P). MDCKs treated with D+T showed decreased expression of e-cadherins and increased expression of fibronectin and α-SMA compared with TGFβ1 alone. Polyamine addition to the medium inhibited cell phenotypical modulation induced by D+T. Original magnification 200 × (a–e) and 630x (f–v, bar=20 μm).

Full size image
Figure 3
figure 3

Representative western blot (a) or for E-cadherin, fibronectin or α-SMA of MDCK cells treated with DFMO, TGFβ1 (TGFβ), DFMO+TGFβ1 (D+T), or DFMO+TGFβ1+ exogenous polyamines (D+T+P). (b) Zymogram for the same conditions. Polyamine depletion decreased epithelial marker e-cadherins and increased expression of mesenchymal markers fibronectin and SMA while at the same time increasing activity of MMP2, pro-MMP9, and MMP9. β-Tubulin was charged as loading control.

Full size image

Zymograms (Figure 3, panel B) showed that polyamine depletion resulted in an increased activity of MMP2, pro-MMP9, and MMP9; moreover, polyamine depletion once again synergized with TGFβ1 and was partially prevented by polyamine addition to the medium (ie by restoring levels similar to those observed after TGFβ1 alone).

Polyamine Depletion Induces Limited TGFβ1 Secretion but not Apoptosis

RT-PCR showed the presence of mRNA for TGFβ1 in control and polyamine-depleted cells (Supplementary Figure 2aA). FACS analysis of labeled cells (Supplementary Figure 2B) showed expression of TGFβ1 in polyamine-depleted cells equal to 5.6% of cell population whereas percentage elevated to 29.7% compared with control in the case of MDCK treated with 4.2 μM Cyclosporin. TGFβ1 dosed in the supernatant collected at day 5 (Supplementary Figure 2C) displayed a concentration of TGFβ1 equal to 450 pg/ml in polyamine-depleted cells, 900 pg/ml of TGFβ1 in Cyclosporin-stimulated cells, whereas unstimulated control cells showed an endogenous TGFβ1 secretion of 90 pg/ml of TGFβ1. Apoptosis, assessed through Annexin V/propidium iodide labeling and subsequent FACS analysis showed no apoptosis induction in polyamine-depleted cells (Supplementary Figure 2, panel D).

Proteomics

2D proteomics showed, as expected, a great difference in the proteomic profile of TGFβ1 treated cells compared with controls (Figure 4, panels A and B). The proteomic profile of control, TGFβ1 treated, or polyamine-depleted TGFβ1 treated cells showed, respectively, 1584±6, 1848±6, or 1857±4 spots. 1506±7 spots were matched across the three groups, whereas 1847±4 spots were matched when TGFβ1 and polyamine-depleted TGFβ1 treated were compared. Difference between polyamine-depleted TGFβ1 and TGFβ1 alone was limited to 30 differentially expressed spots (Figure 4, panels B and C), among them, 27 showed a statistically significant twofold upregulation compared with TGFβ1 alone. Spots were subsequently identified with LC-Mass spectrometry (see Table 1), α-enolase (spots 4, 5, 6), vimentin (spots 7, 8, 9), glucose-regulated protein 78 kDa (GRP78, spots 10, 11, 12), glucose-regulated protein 94 kDa (GRP94, spot 13), or HSP90 α/β (spots 16, 17, 18).

Figure 4
figure 4

Silver staining of representative 2D electrophoresis showing the effect of TGFβ1 (TGFβ) or DFMO+TGFβ1 (D+T) on proteomic profile of MDCK cells. (a) Untreated cells, (b) TGFβ1, or (c) DFMO+TGFβ1. Statistically differentially expressed spots between condition TGFβ and D+T were labeled with numbers. Correspondence with mass spectrometry identified proteins is reported in Table 1.

Full size image
Table 1 Identification by LC-tandem mass spectrometry of penthylamine-biotin (BPA) positive proteins in MDCK whole lysate and statistically significantly spots with a twofold change intensity relative to TGFβ1 alone (2 × )
Full size table

Each gel was performed in triplicate and analyzed using PDQuest Advanced software (Bio-Rad). Average gels were created from the three gels for each type of condition. Normalization and statistical analysis of the results were then performed to compare each condition.

Functional Assessment Using a Biotinilated Polyamine

BPA was efficiently incorporated in MDCK cells as revealed by immunofluorescence (Figure 5a). Competition with exogenous polyamines reduced BPA incorporation within cells, whereas administration of a TG-2 siRNA (Figure 5c) notably decreased TG-2-dependent covalent bond formation of polyamines on cellular target proteins. BPA is thereby covalently linked only in the presence of active TG-2. Moreover, 2D blots of BPA-labeled proteins showed that BPA is covalently linked on a limited subset of cellular proteins (Figure 5b). Statistically significant twofold upregulated proteins were all BPA+. A complete list of BPA+ proteins is displayed in Table 1.

Figure 5
figure 5

(a) Penthylamine-biotin (BPA) incorporation in polyamine-depleted MDCK cells or in presence of exogenous polyamines (poly) (b). BPA is actively incorporated in MDCK cells. Competition with exogenous polyamines drastically reduced BPA staining. (b) 2D blots of BPA-labeled proteins revealed with neutravidin-HRP showed incorporation of BPA on several cellular proteins, labels are reported in Table 1. (c) Transfection of siRNA designed against transglutaminase-2 (TG-2) drastically reduced TG-2 expression hence reducing BPA incorporation in MDCK cells, siRNA for GFP was administered as control.

Full size image

Silencing of TG-2 Induces upregulation of Endoplasmic Reticulum-Stress Proteins and Enhanced EMT

Silencing of TG-2 in the presence of polyamines (Figure 6), added 4 h later, showed a response of cell to TGFβ1 quite similar to administration of TGFβ1 in combination with DFMO. In both conditions, D+T or TG-2 silenced cells, both GRP78 and α-SMA were upregulated compared with TGFβ1 alone. These evidences suggest that TG-2-dependent polyamine covalent bond formation on endoplasmic reticulum (ER)-stress proteins is a relevant post-translational modification for homeostasis of ER-stress proteins. When polyamines are depleted from cells or their covalent bond formation on ER-stress proteins is inhibited through silencing of TG-2, cells increase expression of GRP78 and GRP94 (data not shown) thereby increasing cells susceptibility to TGFβ1.

Figure 6
figure 6

Silencing of TG-2 or polyamine depletion enhance EMT in MDCK cells. Cells silenced for TG-2 (g) and subsequent treated with TGFβ1 in the presence of polyamines showed an increased expression of GRP78 (h) and α-SMA (i) similar to polyamine-depleted cells (in e, f, respectively) treated with control GFP siRNA (d) compared with cells treated with TGFβ1 alone (b and c, respectively) and silenced with control GFP siRNA (a). GRP78 polyamination is TG-2 dependent and modulate EMT in MDCK cells. Original magnification 630 × , bar=20 μm.

Full size image

BPA Positive Protein Expression is Changed on DFMO or Polyamine Administration

GRP78, GRP94, HSP90α/β, α-enolase, or vimentin showed to be statistically upregulated in TGFβ1-treated cells compared with control (P≤0.05 in all cases). GRP78, GRP94, HSP90α/β, α-enolase showed an upregulation in polyamine-depleted TGFβ1 treated cells. Exogenous polyamines addition partially restored expression levels of these proteins (Figure 7).

Figure 7
figure 7

Representative western blots and corresponding bar graphs showing effect of TGFβ1, DFMO+TGFβ1 (D+T), or DFMO+TGFβ1+exogenous polyamines (D+T+P) on cellular expression of GRP78, GRP94, HSP90α/β, α-enolase, or vimentin expression.

Full size image

DISCUSSION

This study supports for the first time a major role for polyamines in EMT in MDCK cells (for review see Hay and Zuk29). EMT is a complex process that regulates major events during early development and, in adult life, has a pivotal function in tissue remodeling during wound healing, organ fibrosis, and cancer progression.2 Although a full comprehension of EMT is still far from being achieved, it is generally accepted that epithelial cells undergo a number of defined phenotypic and functional changes that include first loss of epithelial markers and then acquisition of mesenchymal-like traits. The final cell emerging from the EMT program expresses features distinctive of the myofibroblast, such as tensile properties, conferred by the presence of α-SMA cytoplasmic filaments. Transformation of MDCKs to mesenchymal-like cells has been reported earlier in vitro by using antibodies to E-cadherin.

As stated above, EMT is often regarded as a process of de-differentiation in which epithelia loose their specific markers acquiring a mesenchymal phenotype. Newly differentiated cells exhibit activation of MMPs, enzymes that cleave almost all extracellular matrix components, and activate growth factors and growth factor receptors.30

The results of this study indicate that polyamine depletion enhances TGFβ1-induced EMT in MDCKs by downregulating epithelial markers ZO-1 and E-cadherin as well as by increasing expression of mesenchymal markers fibronectin and α-SMA. At the same time, polyamine depletion increases activity of MMP2, pro-MMP9, and MMP9, all enzymes involved in the break down of cell–cell and cell–ECM interactions, thus triggering EMT process.30, 31, 32 Polyamine depletion and TGFβ1 stimulation show a synergic effect on EMT-related markers, as also confirmed by the fact that exogenous polyamine addition, which restores intracellular polyamine content, prevents these changes. Although a limited TGFβ1 secretion is induced by the sole administration of DFMO, this is in any case insufficient to induce the observed enhanced EMT. Levels of TGFβ1 induced by DFMO are in fact the half of those induced by 4.2 μM Cyclosporin administration that proved earlier23 to be capable to alter paracellular permeability but not EMT. The irrelevance of DFMO-induced TGFβ1 secretion allowed us to look for a possible alternative explanation for the observed enhanced EMT in polyamine-depleted cells. For this reason, a proteomic approach was applied. 2D proteomic analysis of TGFβ1-treated and polyamine-depleted MDCKs vs TGFβ1-treated MDCKs showed a statistically significant variation of 30 spots among the two conditions: we identified enhanced expression of α-enolase, vimentin, HSP90α/β, GRP78, and GRP94. A growing body of literature indicates an essential function for GRP78 and GRP94 proteins in preserving ER function. In particular, the 78 kDa molecular chaperone GRP78 is ubiquitously expressed in mammalian cells, and it has been reported to bind to hydrophobic patches on nascent polypeptides in the ER. This is to prevent aggregation of misfolded proteins and to contribute to the development of proper secondary structure of mature proteins. In addition to preventing the physical aggregation of misfolded proteins, GRP78 is one of the initial components in the signaling cascade of the so-called unfolded protein response (UPR). Similarly to GRP78, GRP94, identified as a 94-kDa protein also referred to as ERp9933 or endoplasmin,34 is an abundant ER glycoprotein. It shares 50% amino acid identity to HSP90α/ß, a well-characterized 90-kDa member of the heat-shock protein family. Recent evidence suggests that GRP94 may also possess protein chaperoning activity. Its synthesis is induced by the accumulation of misfolded proteins in the ER.35, 36 In the ER, GRP78 and GRP94 act in tandem on folding intermediates of the newly synthesized immunoglobulin chains, with GRP94 being involved at a more advanced step in their processing.37

Upregulation of both GRP78 and GRP94, together with α-enolase and vimentin, has been validated in our model by western blot analysis performed at day 3 after TGFβ1 administration. We substantially confirmed earlier immunohistochemical and electron microscopy findings by Parkkinen et al.38 These authors showed in fact redistribution, whorl formation, and autophagy of the ER in polyamine-deprived BHK cells. This is of particular interest in view of recent in vitro studies39, 40 demonstrating that disturbances of the ER homeostasis results in a EMT-like process characterized by cell morphology changes and extensive reorganization of the actin cytoskeleton.2, 39, 40

In our model, endocellular polyamine depletion dramatically enhances TGFβ1-induced EMT, by increasing at the same time the expression of typical ER-stress proteins like GRP78 and GRP94, thus suggesting a direct involvement of ER-stress in the overall and polyamine depletion-dependent scenario.

Preliminary experiments, performed with a rat hepatoma cell line (HTC) clonally selected to be defective for polyamine transport across cell membrane (data not shown, manuscript in preparation), show that endocellular polyamine availability is crucial. In these mutant cells, exogenous polyamines are ineffective in counteracting induction of fibronectin by combined DFMO and TGFβ1 treatment, thus excluding a direct effect of polyamines on TGFβ1 binding on its receptor.

Results of the experiments performed by loading MDCKs with a biotin-linked cadaverine (BPA), in the presence or in the absence of DFMO, suggest that polyamines might be directly involved in the regulation of both GRP78 and GRP94. BPA, that is a mono-functional polyamine carrying only one free -NH2 function, is efficiently up taken by MDCK cells, as demonstrated by immunofluorescence. Within the cell, BPA is than covalently linked to a specific and limited number of cellular proteins. These experiments revealed GRP78 and GRP94 as putative protein targets for polyamine covalent bond formation.

Availability of endocellular polyamines is thereby critical for the regulation of GRP78 and GRP94 and, possibly, for the regulation of ER-stress and subsequent induction of an EMT-like process. Results showed that cell response to TGFβ1 demonstrated to be very similar in polyamine-depleted cells as well as in cells where TG-2 was previously silenced in the presence of polyamines. In all cells, both ER-stress protein GRP78 and α-SMA were upregulated compared with TGFβ1 alone, suggesting that polyamination, through covalent bond formation induced by TG-2, may be a relevant post-translational modification for ER-stress proteins homeostasis. In our experimental settings, TG-2-dependent polyamine covalent bond formation onto ER-stress proteins may represent an intracellular regulatory mechanism able to counteract TGFβ1 induced EMT. In this respect, it is noteworthy that a recent in vivo study has reported that TG-2 inhibition ameliorates experimental fibrosis.41 Moreover, further studies are needed to fully comprehend relevance of polyamine biology in the cell, in particular related to TGFβ1 pathway. Moderate inhibition of TGFβ1-stimulated phenotypical modulation by the sole addition of polyamines, suggests in fact involvement of polyamines also in other cellular pathways.

In conclusion, our work provides the first evidence for a key role of polyamines in EMT. 2D proteomic approach allowed suggesting a direct link between polyamine depletion, activation of ER-stress, and enhancement of EMT, thus reinforcing the concept that ER-stress may be crucial for the epithelial cell fate and the triggering of an EMT-like response.