Autocrine TGF-β Induces Epithelial to Mesenchymal Transition in Human Amniotic Epithelial Cells

Isolation and Cultivation of Human Amniotic Epithelial Cells (hAECs)

The collection of placentas after cesarean section was approved by the Hospital Ethical Committee, and written informed consent was obtained from the donors. The fetal membranes were washed in physiological saline solution (PSS) (B. Braun Melsungen AG, Melsungen, Germany) supplemented with 50 μg/ml amphotericyn (Bristol-Myers Squibb S.L., Madrid, Spain), 48 μg/ml clotrymazol (Almirall-Prodefarma S.A., Barcelona, Spain), 50 μg/ml tobramicyn (Laboratorios Normon S.A., Madrid, Spain), and 50 μg/ml vancomicyn (Laboratorios Hospira S.L., Madrid, Spain). For the purification of hAECs, we used a standardized procedure (4,17,31) with some modifications. Briefly, AM was peeled off from the underlying chorion layer, washed extensively with phosphate-buffered saline (PBS) (Gibco, Life Technologies, Paisley, UK), and cut into small tissue fragments that were digested three times during 40 min at 37°C with 0.05% trypsin-EDTA solution (Sigma, Madrid, Spain). Each trypsin digest was inactivated with two volumes of cold complete DMEM: standard Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin, and 1% l-glutamine (all from BioWhittaker, Lonza, Basal, Switzerland). Then cells were filtered and decanted by centrifugation, and the three digests were pooled together (n = 6). At this point, the cells were labeled as freshly isolated cells (FI). The identity and purity of the population of isolated hAECs was assessed by cytokeratin flow cytometry as described by Ilancheran et al. (17). A typical purification of hAECs (n = 6) yielded more than 99% keratin-positive cells when stained for pan-cytokeratin (AE1/3) (Dako, Glostrup, Denmark). Purified hAECs were cultivated as described in the literature (4,17,31). Briefly, FI hAECs were seeded at a density of 1 × 10⁵ cells/cm² with Complete DMEM and incubated in a humidified atmosphere at 37°C with 7.5% CO₂. When reaching a confluence of 95%, the cells were detached by 0.05% trypsin/EDTA treatment. At this point, cells were called Passage 0 cells (P₀). Subsequently, hAECs were subcultured at a density of 1 × 10⁴ cells/cm² in Complete DMEM. They were labeled according to subculture as Passage 1 (P₁), Passage 2 (P₂), etc. (n = 6). Before subculturing, a phase-contrast picture of growing cells was taken with a Motic AE31 inverted optical microscope (Barcelona, Spain). We observed differences in the growing efficiency of different hAEC batches, as previously described (4), which causes variations in the time equivalence for passages of different experiments. These time variations were no more than 20-30%. For cells cultured with SB-431542 (TGF-β receptor I inhibitor; Sigma), the inhibitor at 10 μM was added immediately after cell plating and maintained until the end of the treatment (n = 4). The inhibitor was refreshed every 3 days as described (18). Cells treated this way were labeled with the passage number followed by the letters SB, P₀SB. For TGF-β assays on SB-431542-treated cells, the inhibitor was washed out to reactivate TGF-β receptors by double PBS wash and readdition of fresh medium. Then, TGF-β (2 ng/ml; Preprotech, London, UK) was added. For EMT reinduction, SB-431542 was washed out from P₁SB cells, trypsinized, and split in two, for cell expansion. Cells obtained this way were incubated in either 10 μM SB-431542 supplemented Complete DMEM or nonsupplemented Complete DMEM. The resulting next cell passages were named P₂SB or P₂SBRm, respectively (n = 3). Finally, cells were harvested after they had grown for several days. The TGF-β neutralizing antibody (TGF-β-na) (R&D Systems, Minneapolis, MN, USA) required a neutralizing concentration (20 μg/ml) that was determined following manufacturer instructions. AM freshly extracted hAECs were cultivated in the presence of 20 μg/ml of anti-TGF-β1, 2, 3 for 17 days and compared to both SB-431542-treated and nontreated cells. The antibody was supplemented every other day (n = 3).

HaCaT Cell Treatment with a Conditioned Medium from hAECs AM Treatment with SB-431542

When AM tissue fragments were used, AM was isolated following the procedure described above but excluding trypsin treatment and subsequent steps (19). For AM conditioning of the medium, an AM fragment of 4 by 4 cm was incubated in 10 ml of Complete DMEM for 24 h. A hAEC-conditioned medium was the result of 2 ml of Complete DMEM that had been conditioned by 1×10⁵ hAECs for 3 days. In both cases, the conditioned medium was fast-frozen in liquid nitrogen and stored at −80°C until use. HaCaT cells (7), a human spontaneously immortalized keratinocyte cell line, were a kind gift from Dr. Caroline S. Hill (London Research Institute, Cancer Research, UK). We assayed the biological activity of hAEC secreted TGF-β on HaCaT cells grown in Complete DMEM. Briefly, 5×10⁵ HaCaT cells were seeded in 5-cm diameter dishes and cultivated for 48 h. Twelve hours before treatment, cells were treated with 10 μM SB-431542. For conditioned medium assay, the inhibitor was washed out to reactivate TGF-β receptors by a double PBS wash (3). Then, HaCaT cells were incubated in conditioned medium for 2 h before being fixed and prepared for immunofluorescence. For AM SB-431542 treatment, AM tissue fragments were cultured in Complete DMEM with or without 10 μM of SB-431542 for 24 h. Finally, AM fragments were processed for immunofluorescence.

Cell Cycle Analysis

hAECs isolated from either AM or in culture were fixed with ice-cold 70% ethanol, washed three times with cold PBS, and stained for cell cycle analysis with propidium iodide (Sigma) following standard methods. Cells were analyzed by flow cytometry using a FACS Calibur 1 (Becton Dickinson, Madrid, Spain).

Western Blot

AM tissue sample or cells in culture were harvested and lysed with 20 mM HCl-Tris pH 7.5, 150 mM NaCl, 5 mM EDTA, 1.2 mM MgCl₂, 0.5% NP-40 (IGEPAL), and 10% glycerol, 10 mM Na-butyrate, 1 mM DTT, 25 mM NaF, and 25 mM β-glycerophosphate supplemented with phosphatase inhibitors (I and II) and protease inhibitors (all from Sigma). Protein extracts were analyzed by sodium dodecyl sulfate polyacrylamide electrophoresis (SDS-PAGE; Bio-Rad, Madrid, Spain) followed by Western blot. Images of Western blot were taken with a ChemiDoc^™ XRS System (Bio-Rad).

Antibodies and TGF-β1 ELISA

The following commercial antibodies were used: phospho-extracellular regulated kinase 1/2 (ERK1/2) and phospho-Smad2 (both from Cell Signaling Technology, Danvers, MA, USA); E-cadherin, vimentin, Snail1 (H130), and β-actin (all from Santa Cruz Biotechnology, Dallas, TX, USA); α-SMA (Sigma); pan-cytokeratin (AE1/3) and Ki-67 (both from Dako); Smad2/3 (BD Biosciences, San Jose, CA, USA); and TGF-β neutralizing antibodies (R&D Systems). Alexa fluor 594-labeled phalloidin was used to reveal actin filaments (Molecular Probes, Invitrogen, Eugene, OR, USA). For TGF-β1 detection, conditioned media from hAECs or AM was analyzed using a human-TGF-β1 ELISA commercial kit (eBioscience, San Diego, CA, USA).

Gene Expression Analysis by qPCR

For gene expression assay, RNA from AM tissue fragments or cultured cells was extracted using RNeasy Mini Kit (Qiagen, Hilden, Germany). Typically, 1 μg of RNA was retrotranscribed by GeneAmp (AppliedBiosystems, Foster City, CA, USA) or by an iScript cDNA synthesis kit (Bio-Rad), and the resulting cDNA was used for qPCR following the SYBR premix ex Taq kit (Takara-Bio, Shiga, Japan) according to manufacturer's instructions. The primers used for gene amplification were as follows: plasminogen activator inhibitor 1 (PAI1): CATGGGGCCATGGAACAAGG PAI-1-F, CTTCCTG AGGTCGACTTCAG PAI-1-R; Krüppel-like factor 4 (KLF4): CCCACACAGGTGAGAAACCT KLF4-F, ATGTGTAAGGCGAGGTGGTC KLF4-R; glyceraldehyde 3-phosphate dehydrogenase (GAPDH): ACCA CAGTCCATGCCATCAC GAPDH-F, TCCACCACC CTGTTGCTGTA GAPDH-R (all from Sigma). The rest of the genes were amplified using QuantiTect Primer Assay (www.quiagen.com/GeneGlobe): E-cadherin Hs_CDH1_1_SG (QT00080143); α-SMA Hs_ACTA2_1_SG (QT00088102); Snail1 Hs_SNAI1_1_SG (QT00010010); MMP9 Hs_MMP9_1_SG (QT00040040); metastasis-associated 1 family, member 3 (MTA3) Hs_MTA_3_SG (QT00064988); vimentin Hs_VIM_1_SG (QT00095795); TGF-β1 Hs_TGFB1_1_SG (QT00000728). Gene expression was referred to as a ratio to GAPDH expression. Gene expression values were represented as a percentage of the highest value.

Immunofluorescence

AM tissue fragments were fixed for 2 h at 4°C with 4% formaldehyde (Sigma)/PBS immediately after cesarean labor. Cultured hAECs on round cover glass were fixed for 10 min with 4% formaldehyde/PBS at room temperature (RT). Both AM and hAECs were permeabilized with 0.3% Triton X-100 (Sigma) in PBS for 15 min and then incubated with blocking buffer [0.3% bovine serum albumin (Santa Cruz), 10% fetal calf serum (Gibco), 0.1% Triton X-100 in PBS] supplemented with 5% skim milk (Difco, Sparks, MD, USA). Subsequently, samples were incubated with the appropriate primary antibody diluted in blocking buffer and washed three times with 0.1% Triton X-100/PBS. After primary antibody, samples were stained with the appropriate secondary antibodies together with Alexa fluor 594-labeled phalloidin and Hoechst 33258 (Fluka, Biochemika, Buchs, Switzerland) that was used for nuclei detection. Finally, samples were examined, and images were taken using a confocal microscope LSM 510 META (Zeiss, Jena, Thüringen, Germany).

hAECs From Term AM Proliferate in Culture and Experience a Change in Morphology

We isolated hAECs from AM and put them into culture. After 3 days, hAECs at P₀ exhibit the typical phenotype that has been previously described (4,38–40) (P₀) (Fig. 1A). A wider separation between cells was observed at P₂ (Fig. 1A). Interestingly, with further passages, cells changed morphology acquiring a more fibroblast-like appearance that was more evident at P₅ (P₃ and P₅) (Fig. 1A). We determined the proliferation rate of hAECs derived from the AM using cell cycle distribution analysis. Freshly isolated (FI) cells were mainly at the G₁ phase of the cell cycle (95.5%) (Fig. 1B). However, hAECs grown in culture for a few days (P₀) were proliferating (Fig. 1C). Consistently, cells at the amniotic epithelial layer were neither showing any mitotic event nor expressing the Ki-67 (12) antigen (Fig. 1D), while hAECs P₀ showed clear staining for Ki-67 (Fig. 1E). This is in accordance with the observation that cells at the amniotic epithelial layer do not proliferate after day 180 (25,32), whereas hAECs resumed proliferation when they were cultured (Fig. 1C, E). Our data indicate that, although hAECs were cell cycle arrested at the AM, when put into culture, they were able to proliferate and experienced a change in their morphology from cuboid to spindle cells.

OPEN IN VIEWER

Figure 1.

hAECs from term AM proliferate in culture and experience a change in morphology. (A) Morphology of primary cultured human amnion epithelial cells (hAECs) in different passages. Phase-contrast images of subconfluent hAECs are shown in different passages (P₀ to P₅) at 10× magnification (Scale bar: 100 μm). (B, C) DNA content of freshly isolated or cultured P₀ hAECs. Cells were analyzed by fluorescence-activated cell sorting (FACS). (B) The histogram shows cells at the G₀/G₁ stage (1C, single copy) and at the G₂/M stage (2C, two copies). Freshly isolated hAECs (FI) were quiescent, and all were at the G₁/G₀ phase of the cell cycle. (C) In culture, hAECs (P₀) were rescued from quiescence and entered the cell cycle. (D) A fragment of amniotic membrane (AM) was fixed and stained for Ki-67 or Hoechst 33258 to reveal nuclei. An image composition of 3 (width) by 2 (height) consecutive confocal microscopy fields is shown. Scale bar: 100 μm. (E) Cultured hAECs at P₀ were immunostained for Ki-67. Nuclei were revealed with Hoechst 33258. Scale bar: 100 μm.

hAECs in Culture Lose Their Epithelial Morphology Correlating with Gain of Mesenchymal Markers

In order to understand the nature of the changes experienced by hAECs, cell cultures were analyzed at different passages for well-established epithelial and mesenchymal cell markers. hAECs at P₀ showed a discontinuous distribution of E-cadherin at the plasma membrane that was limited to places where adjacent cells were contacting with each other (Fig. 2A). With more passages, the expression of E-cadherin became cytosolic and limited to only a few cells; finally by P₅ it had disappeared (Fig. 2A). When hAECs were put into culture, vimentin expression could already be detected at P₀ showing a radial distribution; vimentin expression increased with higher passages adopting a spindle-shaped distribution (Fig. 2B). Moreover, F-actin staining evidenced the loss of cell cohesion that was patent as early as at P₀ and increased with further passages. F-actin was cortical at P₀ and became progressively more transcytoplasmic to end up as stress fibers at P₅ (Fig. 2A, B). The decrease of E-cadherin expression together with the increase of vimentin correlated with a change in actin fiber distribution (P₃ and P₅) (Fig. 2A, B). Western blot analysis of all hAEC passages showed that E-cadherin protein levels were high at early passages and became undetectable by P₃ and P₅ (Fig. 2C), in contrast to the progressive increment of vimentin expression. Interestingly, vimentin expression from the AM tissue sample was very high (Fig. 2C). Keratins of different molecular weight were identified by Western blot and immunofluorescence in different hAEC passages (Fig. 2C, D), which is consistent with P₅ cells having an epithelial origin. Snail1 expression was very low from the AM sample and in the early passages. Strikingly, protein levels increased in the late passages (Fig. 2C). Although phosphorylation of ERK 1 and 2 was high at AM, it was decreased by P₁ and increased further with higher passage numbers (Fig. 2C). α-SMA protein was barely present at P₀ and P₁ hAECs; however, as the number of passages increased, the presence of α-SMA became more abundant and colocalized with phalloidin-stained stress fibers (Fig. 2E, F). Thus, there was a change in the organization of actin fibers in hAECs during EMT with α-SMA colocalizing with F-actin at later passages. Altogether, the above results suggest that hAECs obtained at P₅ result from an EMT process.

OPEN IN VIEWER

Figure 2.

hAECs lose epithelial cell features and gain mesenchymal cell features with increasing number of passages. hAECs in culture show cytokeratins throughout all passages. (A, B) Cultured hAECs in different passages were immunostained for E-cadherin (A) and vimentin (B) and costained for F-actin. Nuclei were revealed with Hoechst 33258. (C) Immunoblot of total cell lysates from hAECs in different passages confirmed the loss of epithelial cell markers and the gain of mesenchymal cell markers. Analyzed proteins are indicated. β-actin was used as a protein loading control. Different bands at pan-Keratin blot are due to detection of several acidic (type 1) and basic (type 2) cytokeratins. (D) Cultured hAECs in different passages were immunostained for pan-cytokeratins and costained for F-actin; nuclei were revealed with Hoechst 33258. Immunofluorescent staining revealed the presence of cytokeratins in all the passages confirming the epithelial origin of hAECs. (E) Total protein from hAECs in different passages was analyzed by Western blot for α-smooth muscle actin (α-SMA). β-actin was used as loading control. (F) hAECs grown for different passages were immunostained for α-SMA and costained for F-actin. Nuclei were revealed with Hoechst 33258. Stress fibers were noticeable in P₃ and P₅ and corresponded to a stronger expression of α-SMA. (A, B, D, F) F-actin staining shows the clear appearance of stress fibers at high passages of the culture. All cell images were adquired by confocal microscopy. Scale bar: 50 μm. Snai-1, snail1; p-ERK2, phosphorylated extracellular signal-regulated kinase 2.

During EMT hAECs Change the Expression of EMT Marker Genes

A transcription factors genetic expression program is common to different EMTs and is responsible for the loss of CDH1 (E-cadherin) expression, among others. Our study of E-cadherin showed that it was lost by P₃ (see Fig. 2A, C). Accordingly, CDH1 had its maximum expression by P₁; however, its expression was lost by P₃ (Fig. 3A). To understand CDH1 gene regulation during EMT, we studied the expression of three different known transcription factors involved in its regulation: SNAI1 (Snail1), metastasis-associated 1 family, member 3 (MTA3), and Krüppel-like factor 4 (KLF4). SNAI1 had a very poor expression at freshly isolated or early passaged hAECs, but it increased by P₃ and P₅ (Fig. 3A). However, the expression of SNAI1 was very strong at the AM, likely due to the contribution of mesodermal layer cells. The expression of MTA3 in hAECs showed a steep downregulation (Fig. 3A) that correlated with SNAI1 upregulation. KLF4 was highly expressed at AM and FI hAECs, but it was rapidly downregulated by P₃ (Fig. 3A). When looking through the passages, VIM (vimentin) expression increased with higher passage numbers (Fig. 3A).

OPEN IN VIEWER

Figure 3.

Genes associated with EMT get upregulated with high passage numbers in cultured hAECs. RNA from AM, freshly isolated hAECs (FI), or different culture passages hAECs was analyzed by qPCR for different gene expression that were represented as a ratio to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). (A) qPCR showed downregulation of the epithelial marker gene E-cadherin (CDH1) as well as a positive E-cadherin promoter regulator gene Krüppel-like factor 4 (KLF4) with an increasing number of hAEC passages. SNAI1 increased its expression with a higher number of passages together with the mesenchymal marker gene vimentin (VIM). Metastasis-associated 1 family, member 3 (MTA3) became highly expressed at hAECs P₁ and became progressively downregulated at later passages. (B) qPCR showed upregulation of plasminogen activator inhibitor 1 (PAI1) by P₀. Expression of α-SMA (ACTA2) increased with progressive hAEC passages. qPCR showed a strong expression of matrix metallopeptidase 9 (MMP9) in P₁ hAECs; in the following passages, MMP9 expression was lost. EMT, epithelial mesenchymal transition.

TGF-β occupies a central role in controlling EMT (1). In fact, abolishing its signaling is sufficient in some cell systems to prevent EMT (34). We investigated the expression of TGF-β target genes such as plasminogen activator inhibitor 1 (PAI1), ACTA2 (α-SMA), and MMP9, all three related to EMT (24). Expression of PAI1 was very low at both AM and freshly extracted hAECs. However, in P₁ its expression dramatically increased and remained high for the rest of the passages (Fig. 3B). ACTA2 expression showed a clear increase from P₂ onwards and reached its maximum by P₃ and P₅ (Fig. 3B), in accordance with its protein expression (see Fig. 2E). MMP9 only had a very acute expression at P₁ (Fig. 3B). All the above gene expression analysis data suggest that hAECs are experiencing a TGF-β-regulated EMT process.

TGF-β Is Involved in hAEC EMT

To investigate the possible involvement of TGF-β in hAEC EMT, isolated cells were put into culture with Complete DMEM supplemented with the TGF-β receptor I inhibitor SB-431542 that prevents TGF-β signaling (18) and compared to isolated cells grown in nonsupplemented Complete DMEM. Already at P₁, cells treated with SB-431542 were more compact with a cobblestonelike distribution compared to nontreated cells that exhibit gaps among them (Fig. 4A). At the subcellular level, SB-431542 treatment caused a cortical, sharp, continuous E-cadherin distribution, while untreated cells showed an irregular E-cadherin distribution (Fig. 4B). SB-431542-treated hAECs did not show any signs of vimentin in clear contrast to nontreated control cells (Fig. 4B). F-actin had a cortical distribution in SB-431542-treated cells compared to an irregular distribution in nontreated cells (Fig. 4B), where it showed the lack of intercell cohesion. Moreover, F-actin formed a ribbon located at the cells on the periphery of the islands in SB-431542-treated cells (Fig. 4B, see arrows). Cells treated with SB-431542 showed the absence of basal Smad2 phosphorylation, while hAECs grown without SB-431542 inhibitor showed a high basal level of Smad2 phosphorylation (Fig. 4C). In both cases, and upon SB-431542 washout to reactivate TGF-β receptor I, treatment with TGF-β caused a strong Smad2 phosphorylation in both SB-431542-treated and nontreated cells (Fig. 4C). Consistently, hAECs in a Complete DMEM nonsupplemented medium showed a nuclear localization of Smad2/3 that further intensified upon TGF-β treatment.

OPEN IN VIEWER

Figure 4.

The inhibition of TGF-β signaling pathway preserves the epithelial architecture of the hAECs. P₁ hAECs grown in Complete Dulbecco's modified Eagle's medium (DMEM) supplemented with 10 μM SB-431542 [transforming growth factor (TGF)-β receptor I inhibitor] had a compact epithelial morphology. (A) Phase-contrast images of P₁ hAECs cultivated in the presence or absence of SB-431542 revealed different cell morphology. Scale bar: 100 μM. (B) Cells as in A were immunostained for E-cadherin and vimentin and costained for F-actin, and nuclei were revealed with Hoechst 33258. Yellow arrows indicate ribbons of F-actin at the periphery of the cell islands. Images were taken by confocal microscopy. Scale bar: 50 μm. (C) P₁ hAECs as in A were treated with TGF-β (1 h), then total protein extracts were analyzed by Western blot for indicated proteins. Before TGF-β stimulation, the inhibitor was washed out from the cells grown in the presence of SB-431542. (D) P₁ hAECs grown in standard culture medium were treated with 2 ng/ml of TGF-β (1 h), 10 μM SB-431542 (4 h), or left untreated. Cells were immunostained for mothers against decapentaplegic homolog 2/3 (Smad2/3) and costained for F-actin; nuclei were revealed with Hoechst 33258. Images were taken by confocal microscopy. Scale bar: 50 μm. In all cases, Complete DMEM supplemented with 10 μM of SB-431542 (SB) or without it (-) is indicated.

In contrast, SB-431542 treatment prevented Smad2/3 nuclear localization (Fig. 4D). All these results suggest that TGF-β is involved in hAEC EMT.

hAECs Produce TGF-β1 in an Autocrine Way

AM is a source of TGF-β1 (15). In order to determine whether hAECs produce autocrine TGF-β1, we assayed condition medium (CM) from AM and hAECs P₀ by ELISA and found a robust production of TGF-β at P₀ hAECs (Fig. 5A). Moreover, qPCR expression analysis of AM and hAECs in different passages revealed a strong production of TGF-β1 at P₀ that decreased with further passages (Fig. 5B). HaCaT cells are human spontaneously immortalized keratinocyte epithelial cells (11) that respond to TGF-β by phosphorylating Smad2 and 3 and translocating them into the nuclei (18,37). We performed a TGF-β biological test by assaying hAECs P₁ conditioned medium (CM) on HaCaT cells. CM(P₁) induced the nuclear translocation of Smad2/3 that was inhibited by SB-431542 (Fig. 5C), thus confirming the data obtained by ELISA and qPCR. Altogether, these data indicate that hAECs produce TGF-β1 in an autocrine fashion.

OPEN IN VIEWER

Figure 5.

hAECs produce TGF-β endogenously causing the phosphorylation of Smad2 and its translocation to the nucleus. AM epithelial cells express epithelial but not mesenchymal markers. (A) Conditioned medium from AM or hAECs P₀ was analyzed for TGF-β1 content by ELISA. TGF-β1 concentration produced by 6×10⁴ cells was represented. (B) AM, freshly isolated hAECs (FI), or indicated hAEC RNA cultured passages were analyzed by qPCR for TGF-β1 gene expression that was represented as a ratio to GAPDH. (C) HaCaT cells were stimulated with conditioned medium coming from hAECs at P1, CM(P₁), and the same conditioned medium supplemented with 10 μM SB-431542. Additionally, another set of cells was stimulated with 2 ng/ml of TGF-β or TGF-β together with SB-431542. After 2-h incubation, cells fixed and immunostained with Smad2/3 antibody and costained for F-actin; nuclei were revealed with Hoechst 33258. Images were taken by confocal microscopy. Scale bar: 50 μm. (D-G) complete fragments of AM were stained for E-cadherin (D), vimentin (E), α-SMA (F), and pan-cytokeratin (G) and costained for F-actin. Nuclei were revealed with Hoechst 33258. Two overlapping distant optical planes of the same area containing either hAECs or human amniotic mesenchymal stromal cells (hAMSCs) were acquired by confocal microscopy. The amniotic epithelium (amniotic epithelium optical plane) and the amniotic mesoderm (amniotic mesoderm optical plane), labeled Amn. Epith. and Amn. Mes., respectively, are shown side by side. Scale bar: 50 μm.

Cells in the AM Epithelial Layer Express Epithelial But Not Mesenchymal Markers

Cultured hAECs at P₀ exhibited a clear expression of vimentin and α-SMA; additionally, E-cadherin showed a discontinuous distribution. In order to determine the native state of hAECs at the AM, we studied epithelial and mesenchymal markers for the epithelial and mesenchymal cells of the AM by immunostaining. Cells in the amniotic epithelial layer were forming a compact monolayer epithelium, whereas cells at the amniotic mesoderm were more disperse with a typical mesenchymal shape (Fig. 5D–G). E-cadherin had a continuous cortical distribution at cells at the amniotic epithelial layer and was absent at mesodermal layer cells (Fig. 5D). Vimentin and α-SMA proteins were abundant in cells at the mesodermal layer; however, they were virtually absent from cells at the amniotic epithelial layer (Fig. 5E, F). Additionally, cells in the AM epithelial layer showed a strong cytosolic staining for pan-cytokeratin that was lacking in cells at the amniotic mesodermal layer (Fig. 5G). In both types of cells, F-actin was present, but while F-actin in cells in the amniotic epithelial layer had a cortical distribution and was very close to E-cadherin, F-actin in cells in the mesodermal layer was evenly distributed at the whole cell (Fig. 5D, E). All these results indicate that cells in the epithelial layer of the AM constitute a bona fide epithelium. However, culturing hAECs induces an EMT that is dependent on autocrine TGF-β signaling. Inhibition of TGF-β signaling clearly prevents hAECs spontaneous EMT.

SB-431542-Treated hAECs Experienced EMT Upon Removal of the Inhibitor

hAECs experience EMT upon TGF-β autocrine production. The use of SB-431542 prevented these changes and kept cells within their original epithelial characteristics. We cultured hAECs in either an SB-431542-supplemented or normal medium until P₁. hAECs with SB-431542 at P₁ (P₁SB) were split into two groups; SB-431542 treatment was removed in one of them, whereas the inhibitor was kept in the other. Both cultures were grown for one more passage. Resulting cells were called P₂SBRm or P₂SB and were compared to P₂ cells (Fig. 6A). P₂SBRm cells lost E-cadherin neatness at cell boundaries and were more similar to P₂ cells in contrast to P₂SB cells that showed a sharp, cortical distribution of E-cadherin (Fig. 6B). Additionally, vimentin, absent in P₁SB (see Fig. 4B) and P₂SB (Fig. 6B) cells, was expressed at P₂BRm (Fig. 6B). Furthermore, Smad2/3 subcellular distribution was mainly nuclear in P₂SBRm as in P₂ cells, in contrast to P₂SB cells (Fig. 6B). Notably, P₂SBRm cells exhibited a significant basal level of Smad2 phosphorylation while it was not phosphorylated at P₂SB cells (Fig. 6C). Treatment of both P₂SBRm and P₂SB cells with TGF-β, after SB-431542 washout, induced a strong phosphorylation of Smad2 (Fig. 6C). Treatment of HaCaT cells with conditioned medium from cells where the TGF-β inhibitor had been removed, CM(P₂SBRm), localized Smad2 to the nucleus, and this was prevented by the presence of SB-431542 (Fig. 6D). The expression analysis of PAI1, MMP9, VIM, or ACTA2 showed that gene expression was prevented in P₀SB, P₁SB, and P₂SB hAECs (Fig. 6E). In contrast, P₀ and P₁ cells showed the previously described pattern of expression (see Fig. 3A, B). Strikingly, when SB-431542 was removed from the P₁SB cells, all the genes that had a low expression in P₁SB cells, turned out to be upregulated in the next passage (P₂SBRm) (Fig. 6E). All the data above suggest that removal of the ALK5 inhibitor induces the resetting of TGF-β signaling and subsequently induces EMT that could be detected at the molecular and genetic levels.

OPEN IN VIEWER

Figure 6.

Removal of SB-431542 from P₁SB hAECs induces EMT. (A) FI hAECs were grown in SB-431542 supplemented Complete DMEM or Complete DMEM for two passages. P₁SB were passaged to either 10 μM SB-431542 supplemented Complete DMEM or Complete DMEM. Cells were called P₂SB and P₂SBRm, respectively. Both types of cells were grown for 10 days before protein extraction or immunostaining (n = 3). (B) P₂SB, P₂SBRm, or P₂ hAECs were immunostained for E-cadherin, vimentin or Smad2/3 and costained for F-actin; nuclei were revealed with Hoechst 33258. Scale bar: 50 μm. (C) Either P₂SB or P₂SBRm hAECs were stimulated or not with TGF-β (1 h). Immunoblot of total cell lysates for phospho-Smad2; Smad2 and β-actin were used as loading controls. *Nonspecific band. (D) Conditioned medium from P₂SBRm hAECs [CM(P₂SBRm)] either supplemented or not with 10 μM SB-431542 was used on HaCaT cells for 2 h. Additionally conditioned medium from hAECs at P₁ [CM(P₁)] was used as a control. Moreover, HaCaT cells were treated for 2 h either with TGF-β or TGF-β together with SB-431542. Cells were immunostained for Smad2/3 and costained for F-actin; nuclei were revealed with Hoechst 33258. Scale bar: 50 μm. (E) Genes associated with EMT get upregulated after SB-431542 removal from the P₁SB hAECs. AM or different culture passage hAECs RNA were analyzed by qPCR for PAI1, MMP9, VIM, or ACTA2 genes. Gene expression is represented as a ratio to GAPDH.

A Compact Epithelium Does Not Prevent TGF-β-Induced EMT

Epithelial cells at AM form a compact epithelium (see Fig. 5D–G). To study whether a compact epithelium is sufficient to prevent autocrine TGF-β-induced EMT, hAECs were grown in medium supplemented with SB-431542, which caused them to form a P₁ confluent compact epithelium (data not shown). Then, SB-431542 was removed, and cells were cultivated for 5 more days, while in parallel a control group kept the inhibitor. In 5 days, cells where ALK5 inhibitor had been removed became bigger, flatter, less compact, and exhibited intercellular gaps compared to cells grown with SB-431542 that continued being smaller and more compact (Fig. 7A). E-cadherin distribution in SB-431542-treated cells was sharp, cortical, and neatly localized at cell boundaries, and vimentin was absent (Fig. 7B). Furthermore, Smad2/3 had a clear cytoplasmic localization (Fig. 7B). In contrast, untreated hAECs showed a notched and serrated E-cadherin distribution with cells showing clear gaps between them (Fig. 7B). Remarkably, vimentin was present in these cells (Fig. 7B). Moreover, Smad2/3 was clearly localized at the nucleus in contrast to SB-431542-treated hAECs (Fig. 7B). Therefore, even when hAECs form a tight compact epithelium similar to AM epithelium, TGF-β signaling activation prompts the cells to undergo EMT.

OPEN IN VIEWER

Figure 7.

A compact epithelium does not prevent TGF-β-induced EMT. (A) Confluent hAECs maintain a bona fide epithelial morphology when treated with SB-431542. hAECs grown in Complete DMEM either supplemented or not with 10 μM SB-431542 were examined by phase-contrast microscopy. Scale bar: 100 μm. (B) hAECs were grown to full confluence in the conditions mentioned above. Cells were immunostained for E-cadherin, vimentin, or Smad2/3 and costained for F-actin; nuclei were revealed with Hoechst 33258. Scale bar: 50 μm. Complete DMEM supplemented with 10 μM of SB-431542 (SB) or without it (-) is indicated. Immunofluorescence images were taken by confocal microscopy.

AM Epithelial Cells Are Sensitive to TGF-β

Cultured hAECs had an autocrine production of TGF-β and were sensitive to this cytokine. To investigate whether AM epithelial cells were sensitive to TGF-β, we studied AM fragments cultured in the presence or absence of SB-431542. Immunofluorescence analysis of AM fragments revealed a clear exclusion of Smad2/3 from the nuclei of AM epithelial cells treated with SB-431542, while the presence of Smad2/3 was clearly nuclear in the nontreated ones (Fig. 8A). This is in accordance with AM TGF-β production reported in the literature (15,26). It also indicates that AM epithelial cells are sensitive to TGF-β.

OPEN IN VIEWER

Figure 8.

AM produces TGF-β and is sensitive to it. (A) AM was incubated in complete DMEM or DMEM supplemented with SB-431542 for 24 h. Complete fragments of AM were stained for Smad2/3 and costained for F-actin. Nuclei were revealed with Hoechst 33258. Images were acquired by confocal microscopy. Scale bar: 20 μm. (B) AM isolated hAECs were cultivated in Complete DMEM or Complete DMEM supplemented with SB-431542 or TGF-β-neutralizing antibody (na) (indicated as α-TGF-β). Phase-contrast images (original magnification: 10×, scale bar: 100 μm) of subconfluent hAECs are shown for the three different treatments. (C) Immunoblot of total cell lysates from hAECs cultivated in Complete DMEM (-) or Complete DMEM supplemented with TGF-β-na (α-T) or SB-431542 (SB) were analyzed for the indicated proteins. Smad2 was used as a protein loading control. *A nonspecific band. Western blot bands for phospho-Smad2 were quantified and represented as a ratio to Smad2. (D) RNA from hAECs as in (C) was analyzed by qPCR for PAI1, MMP-9, or VIM genes. Gene expression is represented as a ratio to GAPDH. (E) hAECs as in (B) were immunostained for either E-cadherin, vimentin, or Smad2/3 and costained for F-actin; nuclei were revealed with Hoechst 33258. Images were acquired by confocal microscopy. Scale bar: 50 μm.

The TGF-β Neutralizing Antibody Prevents hAECs EMT

We explored further the relationship between hAECs EMT and autocrine TGF-β production by using a TGF-β1, 2, and 3 neutralizing antibody (TGF-β-na). Already at P₀, AM freshly isolated hAECs cultured in the presence of TGF-β-na exhibited an epithelial morphology similar to the SB-431542-treated cells and different from nontreated hAECs (Fig. 8B). Western blot of TGF-β-natreated hAECs revealed an important decrease in the amount of phosphorylated Smad2 (Fig. 8C). We studied the effect of the neutralizing antibody on the expression of PAI1, MMP9, and VIM at P₀ cells. All genes showed a strong expression in control nontreated cells that was either substantially decreased by TGF-β-na treatment or completely abolished by SB-431542 (Fig. 8D). At the subcellular level, the study of Smad2/3 confirmed a moderate effect of TGF-β-na on Smad2/3 translocation (Fig. 8E). However, hAECs treated with TGF-β-na had a continuous, sharp, cortical distribution of E-cadherin similar to SB-431542-treated cells but in clear contrast to nontreated cells (Fig. 8E). Moreover, F-actin staining highlighted the presence of gaps between cells in the control hAECs when compared to TGF-β-na- or SB-431542-treated ones (Fig. 8E). Finally, expression of vimentin was virtually absent from both SB-431542- and TGF-β-na-treated cells (Fig. 8E). Thus, the blockade of TGF-β with a specific neutralizing antibody had similar consequences as the inhibition of ALK5 with SB-431542. All these results indicate the direct involvement of TGF-β as the cytokine responsible for hAEC EMT.