SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

The phenotypic homology of fibroblasts and mesenchymal stem cells (MSCs) has been recently described. Our study investigated the in vitro potential of human skin fibroblasts to differentiate into mesodermal (osteocyte and adipocyte) and endodermal (hepatocyte) cell lineages by comparison with human bone marrow (hBM) MSCs. The endodermal potential of fibroblasts was then explored in vivo in a mouse model of liver injury. Fibroblasts were able to acquire osteocyte and adipocyte phenotypes as assessed by cytochemistry and gene expression analyses. After exposure to a specific differentiation cocktail, these cells presented hepatocyte-like morphology and acquired liver-specific markers on protein and gene expression levels. Furthermore, these fibroblast-derived hepatocyte-like cells (FDHLCs) displayed the ability to store glycogen and synthesize small amounts of urea. By gene expression analysis, we observed that fibroblasts remained in a mesenchymal-epithelial transition state after hepatocyte differentiation. Moreover, FDHLCs lost their hepatocyte-like phenotype after dedifferentiation. In vivo, human fibroblasts infused directly into the liver of hepatectomized severe combined immunodeficient (SCID) mice engrafted in situ and expressed hepatocyte markers (albumin, alpha-fetoprotein, and cytokeratin 18) together with the mesodermal marker fibronectin. Despite lower liver-specific marker expression, the in vitro and in vivo differentiation profile of fibroblasts was comparable to that of mesenchymal-derived hepatocyte-like cells (MDHLCs). In conclusion, our work demonstrates that human skin fibroblasts are able to display mesodermal and endodermal differentiation capacities and provides arguments that these cells share MSCs features both on the phenotypic and functional levels. (HEPATOLOGY 2007;46:1574–1585.)

Fibroblasts and mesenchymal stem cells (MSCs) are mesodermal in status and have been isolated from various tissues. However, unlike fibroblasts, MSCs were recently demonstrated to possess broad differentiation capacities.1–3 Recent studies suggested the difficulty discriminating fibroblasts from MSCs based on phenotype (morphology and specific protein markers) or growth capacity.4–6 These works proposed pluripotency as an exclusive feature of MSCs or identified relationships between cell phenotype and gene expression profile. Conversely, previous data reported the ability of fibroblasts to differentiate into osteocytes,7–14 or chondrocytes.15, 16 While adipose differentiation was thought to be reserved for embryonic fibroblasts or 3T3-L1 cell line,17, 18 Feldon et al. recently obtained adipocytes from human orbital fibroblasts.19 However, there is to date no evidence that fibroblasts possess endodermal differentiation potential. In this work, we analyzed the in vitro mesodermal (osteocyte and adipocyte) and endodermal (hepatocyte) differentiation capacities of human skin fibroblasts. We then confirmed the endodermal differentiation potential of these cells in vivo after transplantation into hepatectomized severe combined immunodeficient (SCID) mice. We used human bone marrow (hBM) MSCs as a control for all experiments.

In parallel, we explored whether fibroblasts present an epithelial-mesenchymal transition (EMT) expression profile throughout differentiation and dedifferentiation processes. As reported for hepatic stellate cells,20 we observed that fibroblast-derived hepatocyte-like cells (FDHLCs) still expressed mesodermal marker α-smooth muscle actin (ASMA) whereas dedifferentiated fibroblasts (DFs) lost the acquired hepatocyte-like phenotype and neo-expressed slug gene, confirming that in both processes cells remained in an EMT state.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Ethical Committee.

This study was approved by the local institutional human ethical committee.

Cell Isolation and Culture.

Human fibroblasts (n = 5) were collected by skin biopsy (medio-anterior side of the forearm) of 8-year-old to 35-year-old volunteers after written informed consent. Cells were grown on culture dishes with AmnioMax-C100 Basal Medium containing C100 Supplement (Invitrogen). After reaching confluence, cells were detached using 0.05% trypsin–1 mM ethylene diamine tetraacetic acid (EDTA) (Invitrogen) and replated at 0.5 × 104 cells/cm2 in αMEM medium (Invitrogen) containing 10% fetal bovine serum (FBS) (Perbio) and 1% penicillin/streptomycin (Invitrogen) (defined as growth medium). Supernumerary cells were suspended in FBS containing 10% dimethyl sulfoxide (VWR) and frozen in 1-mL vials in liquid nitrogen. When necessary, frozen aliquots were thawed, plated in culture flasks at 1 × 103 cells/cm2 and handled as fresh cultures. Bone marrow (BM) samples (n = 13) were collected by aspiration of vertebrae of postmortem donors. BM aspirates were collected into heparinized syringes containing 10% Hank's balanced salt solution (Invitrogen) and processed within 48 hours. Mononuclear cells (MNCs) were separated by centrifugation at 400g for 20 minutes at 20°C after carefully loading BM on Ficoll-Paque PLUS (Amersham Biosciences). Cells were then washed twice with phosphate-buffered saline (PBS, Invitrogen). MNCs were seeded in culture flasks at 1 to 6 × 106 cells/cm2 in growth medium. First medium change was performed after 48 to 72 hours and then twice weekly. Passaging of both cell types was performed at 70% confluence with 0.05% trypsin–EDTA and cells were replated at 0.5 × 104 cells/cm2 in growth medium. Cells were cultured at 37°C in a humidified atmosphere containing 5% CO2. Fibroblast colony forming unit (CFU-F) assays were performed by plating 0.5 to 10 cells/cm2 in growth medium. Plates were controlled after 10 days for presence of colonies. Cumulative population doublings were calculated using the following equation: ([log10(NH) − log10 (NI)]/log10 (2)), where NI is inoculum number and NH is cell harvest number, as described elsewhere.21

Transmission Electron Microscopy.

Skin biopsy samples were fixed with 2.5% electron microscopy grade glutaraldehyde (Agar Scientific) buffered in 0.1 M sodium cacodylate for 48 hours, postfixed in 1% osmium tetroxide (Agar Scientific), and embedded in Epoxy Embedding Medium (Fluka Chemie, Buchs, Switzerland). Semithin sections were contrasted with uranyl acetate and lead citrate before examination in transmission electron microscopy with a Zeiss EM109 (Carl Zeiss Inc., Oberkochen, Germany).

Mesodermal Differentiation.

Cells were plated at 1.5 × 104 cells/cm2 in uncoated 6-well dishes. To obtain osteogenic differentiation, cells were incubated with growth medium containing 0.1 μM dexamethasone, 0.05 mM ascorbate and 10 mM β-glycerophosphate (all from Sigma), and changed twice a week. After 4 weeks, calcium deposition was evaluated by von Kossa and alizarin red stainings. For adipogenic differentiation, cells were incubated with growth medium containing 1 μM dexamethasone, 0.5 mM isobutyl-methylxanthin, 0.2 mM indomethacin (all from Sigma) and 10 μg/mL insulin (Lilly) with medium change twice a week. After 4 weeks, lipid vesicles were revealed by Oil Red-O staining.

Hepatocyte Differentiation.

For hepatocyte differentiation, cells were plated at 1.5 × 104 cells/cm2 on collagen type I coated dishes. Cells at passage 1 to 3 were incubated with 20 ng/mL HGF (Peprotech), 10 ng/mL FGF4 (Peprotech), 1% insulin-transferrin-selenium (ITS premix, Invitrogen) and 0.61 g/L nicotinamide (Sigma) in serum-free Iscove's modified Dulbecco's medium (IMDM, Invitrogen) for 10 days (referred as induction medium), and then with 20 ng/mL OSM (Peprotech), 1 μM dexamethasone (Sigma) and 1% ITS premix in the same basal medium for ≥ 20 days (maturation medium). For negative control, cells were incubated with serum-free IMDM.

Flow Cytometry.

Trypsinized cells were suspended at a concentration of 0.5 to 1 × 103/μL in PBS and then incubated for 30 minutes at 4°C with fluorescence-conjugated monoclonal antibodies and control mouse isotypes as specified in Table 1. Cells were then washed and resuspended in Isoton (Beckman Coulter) for reading with a FACS Calibur Flow Cytometer (Becton Dickinson). Intracytoplasmic human albumin staining was performed as follows: 2 × 105 cells were incubated in 200 μL Cytofix (BD Bioscience) for 20 minutes at room temperature (RT), washed with 1 mL 1× Perm/Wash solution (BD Bioscience) and centrifugated for 5 minutes at 805 g. Cell pellets were suspended in 50 μL Perm/Wash and incubated with 1 μL fluorescein isothiocyanate (FITC)-labeled rabbit anti-human albumin antibody (Cedarlane) or 1 μL FITC-labeled IgG1 control isotype (BD Pharmingen) for 25 minutes at RT. After washing cells with 1 mL PBS and centrifugating for 5 minutes at 805 g, cells were fixed in 250 μL Cellfix (BD Bioscience) and read on a FACSCalibur flow cytometer.

Table 1. Description of Fluorescence-Conjugated Monoclonal Antibodies Used for Flow Cytometry
Antigen specificityFluorescenceIsotypeProvenance (MoAb – isotypes)
  1. Abbreviations: APC, allophycocyanin; FITC, fluorescein isothiocyanate; MoAb, monoclonal antibody; PE, phycoerythrin; PE-Cy5, phycoerythrin-cyanin 5. Provenance of monoclonal antibodies and control isotypes is specified as follows: (1) Becton Dickinson, (2) Immunotech, (3) Dako Systems, (4) Beckman Coulter, (5) Serotec, (6) BD Pharmingen, (7) Cedarlane.

Mesenchymal lineage markers
CD105FITCIgM2 – 2
CD73PEIgG11 – 1
CD90PE-Cy5IgG11 – 6
CD29PEIgG11 – 1
CD44PEIgG2b3 – 5
CD49bFITCIgG13 – 5
CD49eFITCIgG2b3 – 5
CD71PEIgG13 – 1
Hematopoietic lineage markers
CD45APCIgG11 – 1
CD34APCIgG11 – 1
CD14PE-Cy5IgG2a1 – 5
CD13PEIgG14 – 1
CD117PE-Cy5IgG13 – 6
Immunoreactivity markers
CD80PEIgG13 – 1
CD86PEIgG11 – 1
HLA-DRFITCIgG2a1 – 6
HLA-ABCPEIgG2a4 – 5
Hepatocyte markers
CD26PEIgG13 – 1
AlbuminFITCIgG17 – 1

Immunocytochemistry.

Cells were fixed with 3.5% formaldehyde for 15 minutes, then permeabilized with 1% Triton X-100 (Sigma) in tris-buffered saline (TBS) buffer (pH 7.5) for 15 minutes and rinsed with TBS. After 1 hour incubation with blocking solution (3% dry milk in TBS), cells were rinsed with TBS and then incubated for 1 hour at RT with primary mouse anti-human monoclonal antibodies as follows: 1/50 vimentin, 1/50 albumin (Sigma); 1/50 ASMA, 1/50 HepPar1 (Dako Cytomation); 1/50 CK18 (Progen); 1/50 CK8, 1/200 connexin-32 (CX-32) (Chemicon); 1/20 dipeptidylpeptidase-IV (DPPIV) (Ancell); 1/50 CD200 (GeneTex); 1/100 E-cadherin (BD Pharmingen) and rabbit polyclonal anti-human antibodies as follows: 1/50 laminin, 1/100 fibronectin, 1/100 αFP (Dako Cytomation). Cells were washed 5 times and primary antibodies were revealed with 1 hour incubation with Cy3-coupled anti-mouse IgG (Jackson Immunoresearch Laboratories) and FITC-coupled anti-rabbit IgG (Sigma). After rinsing, nuclei were revealed by 30 minutes staining with 4,6-diamidino-2-phenylindol (DAPI, Sigma). For immunoperoxydase staining, cells were incubated after formaldehyde fixation with 30% hydrogen peroxide (Sigma) for 3 minutes. They were washed with distilled water and then permeabilized with 1% Triton X-100. After 1 hour blockade in 1% bovine serum albumin (BSA, Sigma), cells were incubated for 1 hour at RT with primary antibodies in 0.1% BSA, washed with PBS, and exposed for 30 minutes to anti-rabbit or anti-mouse Dako Envision Horseradish Peroxidase (HRP) antibody (Dako Cytomation). Staining was performed by 3 minutes exposition to diaminobenzidine (Sigma). After rinsing, nuclei were colored with Mayer's hematoxylin (10 minutes) and cells were mounted. Each antibody was tested on fresh (<24 hours) cultures of human primary hepatocytes (HHs) as positive control and MSCs or fibroblasts as negative controls.

Reverse-Transcription Polymerase Chain Reaction.

Total RNA was extracted from cultured cells using the TriPure isolation reagent (Roche) and complementary DNA was generated using a reverse transcription kit (Thermoscript RT, Invitrogen), according to the manufacturer's instructions. Polymerase chain reaction (PCR) amplifications were performed using specific primers (Supplementary Table 1) and polymerase elongase (Invitrogen) in a final volume of 25 μL. Amplifications are specified in Supplementary Table 1. All PCR experiments were performed at 30 cycles. Samples were electrophoresed on a 1% agarose gel and nucleic acids were visualized by ethidium bromide staining.

Glycogen Staining.

After 3.5% formaldehyde fixation, cells were incubated for 10 minutes in 1% periodic acid (Sigma), washed with distilled water and incubated with Schiff's reagent (Sigma) for 15 minutes. After rinsing 10 minutes with tap water, nuclei were stained 10 minutes with Karazi's hematoxylin. Cells were then washed with tap water and mounted. Freshly cultured (<24 hours) mouse hepatocytes served as positive control. Fibroblasts and MSCs were used as negative controls.

Urea Assay.

Urea concentrations of culture supernatants were screened by a colorimetric assay (Gentaur), handled as per manufacturer's instructions and analyzed with a HTS7000PLUS absorbance reader (PerkinElmer). Serum-free IMDM incubated with freshly cultured (<24 hours) mouse hepatocytes served as positive control. Serum-free IMDM served as negative control.

Albumin Assay.

Human albumin concentrations of culture supernatants were screened by an enzyme-linked immunosorbent assay (ELISA, Cygnus Technologies) per the manufacturer's instructions and analyzed with a Multiskan EX absorbance reader (Thermo). Serum-free IMDM incubated with freshly cultured (<24 hours) HHs served as positive control. Serum-free IMDM served as negative control.

Estimation of Cell Rates.

Cell rates estimation was performed by counting positive cells (for hepatocyte-like morphology or glycogen storage) in 3 different random fields (200× magnification) of each 3 different culture samples. Total cell counts are provided for each experiment.

Cell Transplantation.

Cell transplantations were performed on 8-week-old to 12-week-old female SCID mice. The animals were randomly assigned into 2 groups. In the first group, undifferentiated fibroblasts (UFs) were injected into the spleen of hepatectomized mice. In the second group, fibroblasts were injected into the liver of hepatectomized mice. Hepatectomy was performed on left or anterior liver lobes when associated respectively to spleen or liver infusion and cells were infused after the surgical procedure. Each group of mice was constituted of 3 control mice receiving 100 μL PBS and 6 sample mice receiving injections of 1 × 106 cells suspended in PBS. After 3 weeks, serum was collected and tissues (liver and spleen) were removed and paraffin enrobed for further analyses.

Immunohistochemistry.

Liver sections (5 μm) were deparaffinized with Histosafe, isopropanol, and methanol solutions. Epitope retrieval was performed by incubating sections for 75 minutes in 0.02 M citrate buffer (pH 5.8) heated to 97°C. Endogenous peroxydase was inactivated with 0.3% hydrogen peroxide in methanol for 15 minutes at RT. Unspecific binding was blocked by incubating sections for 30 minutes at RT in PBS with 1% BSA. Tissues were stained with mouse monoclonal anti-human albumin (1/250) antibody using avidin-biotin technique with DakoCytomation ARK as per manufacturer's instructions. Polyclonal anti-human antibodies were incubated overnight at RT at the following concentrations: 1/3000 rabbit anti-human fibronectin, 1/1000 rabbit anti-human αFP and 1/200 guinea pig anti-human CK18 (Progen). Sections were rinsed with PBS and then incubated for 30 minutes at RT with anti-rabbit Dako Envision HRP antibody or for 45 minutes with HRP-conjugated goat anti-guinea pig antibody (Progen). Sections were washed and immunoreactivity was revealed by 3 minutes exposition with diaminobenzidine. After rinsing, nuclei were counterstained with Mayer's hematoxylin for 10 minutes and then sections were mounted. Controls for albumin and fibronectin staining were performed on human or mice liver sections.

Statistics.

Results are expressed as mean ± standard deviation (SD) and statistically (*P < 0.05, **P < 0.01, ***P < 0.001) significant differences were assessed by unpaired Student t test.

Results

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

We first compared fibroblasts and MSCs at the phenotype level. Although both cell types displayed similar in vitro spindle-shaped morphology (Fig. 1A,B), fibroblasts had greatest proliferation capacity with significantly superior growth rate (doubling time of 3.8 ± 1.5 days for fibroblasts (n = 3) versus 8.0 ± 3.7 days for MSCs (n = 3) within the first 10 passages, *P < 0.05). Furthermore, fibroblast cultures performed up to 52 population doublings (PD) within 15 passages whereas MSCs stopped proliferating at passage 8 and grewed up to 34 PD. As mesodermal lineages, fibroblasts and MSCs displayed comparable expression profile of mesodermal antigens (vimentin, fibronectin, ASMA, laminin) as demonstrated by immunofluorescence (Fig. 1C-J). Similarly, equivalent immunophenotype was observed in both cell types by flow cytometry with high expression of CD44, CD29, CD73, CD90, CD13 and HLA-ABC, moderate expression of CD105, CD71, CD49e and low expression of CD45, CD34, CD14, CD117, CD49b, HLA-DR and CD26 (Fig. 1M,N). Both cell types were negative for CD80 and CD86 (data not shown). Fibroblasts cultures were ascertained for the absence of contaminating epidermal stem cells by sampling biopsies in follicle-free skin region (medio-anterior side of the forearm) and by analysing these biopsies using optic and electron microscopy which confirmed the lack of follicula structures in these samples (Supplementary Fig. 1). These data were corroborated by showing lack of CD200 expression in the cultured cells (Fig. 1K).22

thumbnail image

Figure 1. Comparative fibroblasts and MSCs characterization. Morphological aspect of fibroblasts (A, day 13) and MSCs (B, day 15) in first passage growth culture. Immunofluorescence assay showing stainings of fibroblasts (C-F) and MSCs (G-J) for vimentin (C,G), fibronectin (D, H), ASMA (E,I) and laminin (F,J) (n =3 each). Absence of CD200 expression in fibroblasts (K) as shown by immunocytofluorescence assay whereas this marker was positively detected in human neuroblastoma SH-SY5Y cells (n = 3). Nuclei were revealed by DAPI staining. Pictures magnifications are (A-J) 400× and (K,L) 200×. (M) Example of representative flow cytometric profile of fibroblasts at first passage. Analyzed epitopes are indicated in the histograms. Bars indicate the fluorescence level of corresponding isotype. (N) Comparative cytometric phenotype of MSCs and human skin fibroblasts (both cells from passages 2 to 4). Fibroblasts are indicated by (-f) suffixes.

Download figure to PowerPoint

After having observed phenotypic homology of fibroblasts and MSCs, we compared mesodermal differentiation capacity of these cells to further delineate their pluripotency. By classical mesodermal differentiation assays, we observed that fibroblasts up to passage 3 were able to acquire morphological characteristics of osteocytes as assessed by von Kossa and alizarin red stains revealing deposition of mineral matrix (Fig. 2A,D). Fibroblasts could also accumulate intracytoplasmic lipid-rich droplets, a feature of adipocytes demonstrated by Oil Red O staining (Fig. 2G). These features were acquired in a similar fashion than MSCs. We confirmed these results by analysing the gene expression profile of fibroblasts-derived mesodermal lineages by RT-PCR (Fig. 2J). Fibroblast-derived osteocytes (FDOs) neo-expressed osteocyte-specific genes (osteocalcin, bone sialoprotein, collagen type 1-α1) that were not detected in UFs. Up-regulation of PPAR-γ2 was observed in fibroblasts after adipocyte differentiation whereas this marker was slightly expressed after osteocyte differentiation and in UFs suggesting less specificity of this marker, as confirmed elsewhere.23 However, fibroblast-derived adipocytes (FDAs) displayed neo-expression of PPAR-γ1, lipoprotein lipase and adipsin. We obtained similar results on cytochemistry and RNA expression assays after mesodermal differentiation of MSCs (data not shown). However, we were not able to differentiate fibroblasts aged more than 3 passages as described in Fig. 3. To compare differentiation potential of fibroblasts cultured less or more than 3 passages, we used fresh (<P3) cells and thawed cryopreserved fibroblasts for subculture (>P3). We noticed that cell phenotype of the 3 cell populations (fresh and ≤P3 or >P3 cryopreserved fibroblasts) was comparable as assessed by flow cytometry (Fig. 3A) and cell morphology (Fig. 3B). However, subcultured (>P3) fibroblasts were not able to synthesize mineral matrix or intracytoplasmic lipid-rich droplets after mesodermal differentiation as confirmed by cytochemistry and RT-PCR assays (Fig. 3B,C).

thumbnail image

Figure 2. Comparative mesodermal differentiation of fibroblasts and MSCs. Aspect of extracellular calcium mineralization in (A,D) fibroblasts and (C,F) MSCs after osteocyte differentiation revealed by (A-C) von Kossa and (D-F) alizarin red stains. Intracytoplasmic lipid-rich droplets stained by Oil Red O assay in (G) adipocyte differentiated fibroblasts and (I) MSCs. No staining could be observed in undifferentiated fibroblasts (B,E,H). (J) RT-PCR analysis showing comparable gene expression of osteocyte (osteocalcin, bone sialoprotein and collagen type I-α1) and adipocyte (PPARγ1, PPARγ2, lipoprotein lipase, and adipsin) markers in differentiated fibroblasts and MSCs. Whereas PPARγ2 was slightly expressed in UFs and SaOs, all analyzed markers were neo-expressed in differentiated cells. Abbreviations: FDAs, fibroblasts-derived adipocytes; FDOs, fibroblasts-derived osteocytes; HAs, human adipocytes; MDAs, mesenchymal-derived adipocytes; MDOs, mesenchymal-derived osteocytes; SaOs, osteosarcoma cells; UFs, undifferentiated fibroblasts. Pictures were taken at (A-F, H) 100× and (G,I) 400× magnifications.

Download figure to PowerPoint

thumbnail image

Figure 3. Comparison of mesodermal differentiation potential of fresh, cryopreserved (≤P3) and subcultured (>P3) cryopreserved fibroblasts. (A) Analysis of cell phenotype (hematopoietic, mesenchymal and immunological markers) by flow cytometry revealing no statistically significant difference between fresh, cryopreserved and subcultured (>P3) fibroblasts populations. (B) Pictures show comparable morphology of undifferentiated fibroblasts cultivated less or more than 3 passages and loss of mesodermal differentiation (calcium matrix deposition or intracytoplasmic lipid-rich droplets) potential of subcultured (>P3) cryopreserved fibroblasts. (C) RT-PCR analysis confirmed the lack of mesodermal differentiation of >P3 fibroblasts as these cells did not express most of the osteocyte-specific and adipocyte-specific genes. For each group, data are representative of 3 independent experiments. Abbreviations: FDAs, fibroblast-derived adipocytes; FDOs, fibroblast-derived osteocytes; HAs, human adipocytes; UFs, undifferentiated fibroblasts; SaOs, osteosarcoma cells. Unless indicated, pictures were taken at 400× magnification.

Download figure to PowerPoint

After having assessed the mesodermal differentiation potential of fibroblasts, we assayed these cells for endodermal hepatocyte differentiation. We observed that in vitro conditioned fibroblasts acquired a morphology defined as round or polygonal shaped cells reduced in size and containing cytoplasmic granulations and central nucleus with prominent nucleolus (Fig. 4A). FDHLCs morphology appeared in the maturation step as for mesenchymal-derived hepatocyte-like cells (MDHLCs) (Fig. 4A). Similar counts of hepatocyte-like cells were estimated in FDHLCs and MDHLCs populations (respectively 61.6 ± 7.6% [n = 675] and 61.3 ± 1.2% [n = 921]). This phenotype was observed on each sample induced to differentiate and could be maintained in culture up to 10 weeks. Using immunocytochemistry, we performed comparative analysis of hepatocyte-specific antigen expression in FDHLCs and MDHLCs and observed highly similar profiles. Both cell types positively stained for albumin, αFP, DPPIV and epithelial marker CX-32 but were negative for CK8, 18 and HepPar-1 (Fig. 4A). All these markers were expressed by HHs whereas undifferentiated fibroblasts did not stain positive for hepatocyte-specific protein markers except for DPPIV in small proportion (Fig. 4A) as confirmed by flow cytometry analysis of CD26 expression. By comparison, DPPIV staining was strongly enhanced in FDHLCs (Fig. 4A). The average of albumin-positive cells in the FDHLCs population was estimated by flow cytometry at 65.2 ± 13.3% (Fig. 4B) while MDHLCs population contained 73.4 ± 17.1% albumin positive cells and UFs were negative for albumin staining (0.9 ± 0.4 %) (n = 3 for each cell type).

thumbnail image

Figure 4. Comparative immunostaining of hepatocyte-specific antigens expression in FDHLCs and MDHLCs. (A) Pictures show that after 35 days of in vitro hepatocyte differentiation, FDHLCs and MDHLCs acquired a polygonal-shaped morphology with intracytoplasmic granulations. The comparison of hepatocyte marker expression of both cell types revealed similar positivity for albumin, αFP, DPPIV, and CX-32 whereas those cells were negative for CK8, CK18, and HepPar-1 epithelial markers. Positive and negative controls of corresponding immunomarking were performed respectively on human liver cells and UFs. Pictures were taken at 400× magnification. (B) Representative examples of flow cytometric albumin immunostaining in FDHLCs, MDHLCs, and UFs. For each sample, positivity value is indicated in the corresponding histogram. Abbreviations: αFP, alpha-1-fetoprotein; CX-32, connexin-32; CK, cytokeratin; DPPIV, dipeptidylpeptidase IV; FDHLCs, fibroblast-derived hepatocyte-like cells; HHs, human hepatocytes; MDHLCs, mesenchymal-derived hepatocyte-like cells; UFs, undifferentiated fibroblasts.

Download figure to PowerPoint

Hepatocyte-specific markers expression analysis by RT-PCR revealed that FDHLCs expressed albumin, αFP, α1-antitrypsin, G6Pase, TDO, GS and HGF receptor c-met (Fig. 5). However, FDHLCs lacked expression of some other immature or mature hepatocyte markers as HNF4, PEPCK, TAT, CYP3A4, CYP2B6 and also epithelial markers as CK8 and CK18. Comparatively, MDHLCs expressed larger amount of hepatocyte-specific markers and also the epithelial marker CK8 but not CK18. While MSCs naturally expressed HGF receptor c-met, fibroblasts lineages needed in vitro conditioning to express this marker. Both MDHLCs and FDHLCs displayed persistent expression of mesodermal marker vimentin. According to what we observed after mesodermal differentiation of subcultured (>P3) fibroblasts, we were not able to induce the expression of hepatocyte-specific markers within these cells on the protein or mRNA level as depicted in Fig. 6.

thumbnail image

Figure 5. Compared gene expression analysis in FDHLCs and MDHLCs by RT-PCR. The figure shows concomitant immature (αFP) and mature (α1AT, G6Pase, TDO, and GS) hepatocyte marker expression in both FDHLCs and MDHLCs whereas other markers are not expressed (HNF4, CYP2A3, CYP2B6). MDHLCs expressed some liver-specific markers lacking in FDHLCs (CK8, PEPCK, TAT). Although MSCs constitutively expressed HGF receptor c-met, fibroblasts needed in vitro conditioning to express this transcript. Both MDHLCs and FDHLCs displayed persistence of mesodermal marker vimentin. Abbreviations: FDHLCs, fibroblasts-derived hepatocyte-like cells; LCS, human liver cell suspension; MDHLCs, mesenchymal-derived hepatocyte-like cells; MSCs, mesenchymal stem cells; UFs, undifferentiated fibroblasts. Abbreviations for oligonucleotides are listed in Table 1.

Download figure to PowerPoint

thumbnail image

Figure 6. Comparison of hepatocyte differentiation potential of fresh, cryopreserved (≤P3) and subcultured (>P3) cryopreserved fibroblasts. (A) Although hepatocyte-like morphology was similarly acquired in ≤P3 and >P3 cultured fibroblasts, these latter cells were not able to express albumin and αFP as shown by immunocytochemistry assay. (B) RT-PCR confirmed that no difference could be observed in albumin, αFP, and α1AT expression after hepatocyte differentiation of fresh or cryopreserved fibroblasts. However, differentiation of >P3 cryopreserved fibroblasts did not allow the expression of these hepatocyte-specific markers. For each group, data are representative of 3 independent experiments. Abbreviations: α1AT, alpha-1 antitrypsin; αFP, alpha-fetoprotein; Cryo, cryopreserved (<P3) fibroblasts; FDHLCs, fibroblast-derived hepatocyte-like cells; Fresh, fresh (<P3) fibroblasts; LCS, liver cell suspension; UFs, undifferentiated fibroblasts. Pictures were taken at 400× magnification.

Download figure to PowerPoint

At the functional level, FDHLCs displayed ability to store glycogen and we estimated 21.7 ± 5.1% of Periodic Acid Schiff (PAS) positive cells in these populations (n = 804) versus 21.1 ± 5.3% PAS-positive cells in the MDHLCs population (n = 1041) (Fig. 7). Comparatively, 75.5 ± 6.9% of the HHs population (n = 1134) presented glycogen storage (Fig. 7C) while no staining was noticed in the MSCs population (Fig. 7D). Supernatant analysis reported that urea secretion level of FDHLCs was not significantly different from MDHLCs or human hepatocytes (respectively 0.62 ± 0.31, 1.09 ± 0.23 and 1.37 ± 0.68 mM/30,000 cells*day) but that these levels were significantly higher to control urea level (0.08 ± 0.06 mM/30,000 cells*day) (Fig. 7). Human albumin was not detected by ELISA assay in the supernatants of FDHLCs or MDHLCs suggesting an negligeable amount of albumin producing cells in these populations.

thumbnail image

Figure 7. Comparison of hepatocyte-like functionality of FDHLCs and MDHLCs. (A-D) Comparable glycogen cytoplasmic deposition was noticed in (A) FDHLCs and (B) MDHLCs by PAS staining. Positive and negative control stainings are presented respectively by (C) HHs and (D) UFs. (E) Urea production levels of FDHLCs (n = 4), MDHLCs (n = 4), and mouse hepatocytes (n = 4) were not statistically different whereas urea concentrations were significantly higher in FDHLCs, MDHLCs, and mouse hepatocyte supernatants than in IMDM controls (n = 6). Data represent mean ± SD of n independent experiments. **P < 0.01 and ***P < 0.001 compared with control. Abbreviations: FDHLCs, fibroblast-derived hepatocyte-like cells; HHs, human hepatocytes; MDHLCs, mesenchymal-derived hepatocyte-like cells. Pictures were taken at 400× magnification.

Download figure to PowerPoint

To investigate whether human fibroblasts were able to engraft and differentiate into hepatocyte-like cells in vivo, we injected 1 × 106 UFs into the spleen or the liver of hepatectomized SCID-mice. Three weeks later, livers were harvested and analysed by immunohistochemistry for human mesodermal and endodermal proteins. No engraftment of human fibroblast could be observed when these cells were injected into the spleen (data not shown). However, when cells were transplanted directly into the left liver lobe of SCID mice, we observed that fibroblasts engrafted in mice livers as small cell clusters expressing both mesodermal marker fibronectin and hepatocyte markers albumin, αFP and CK18 (Fig. 8). When analyzed on adjacent serial sections, these cell clusters showed co-expression of human albumin, αFP and CK18 or of fibronectin and albumin (Supplementary Fig. 2). Although there was a low repopulation rate, the in vivo engraftment and cell expression profile was comparable to what we observed with hBM-MSCs transplanted into SCID mice.24 Estimation of the engrafted cell number is provided in Supplementary Table 2. In order to seek out the hepatocyte-like functionality of in vivo engrafted cells, we evaluated the production of human albumin in mice sera by ELISA but did not detect this protein. We noticed analogous albumin ELISA results in sera of mice transplanted with hBM-MSCs.

thumbnail image

Figure 8. In vivo hepatocyte differentiation of human skin fibroblasts. Pictures show immunohistochemical stains for human markers fibronectin, αFP, albumin, and CK18. Analysis of livers of transplanted SCID-mice revealed that fibroblasts engrafted as small clusters of cells presenting combined expression of mesenchymal marker fibronectin and liver-specific markers αFP, albumin, and CK18. Positive and negative control stains are presented, respectively, on human and mouse livers. Abbreviations: αFP, alpha-1-fetoprotein. Pictures were taken at 400× magnification.

Download figure to PowerPoint

Finally, we explored whether fibroblasts presented an EMT status during hepatocyte differentiation or dedifferentiation processes. For that purpose, we studied the expression of EMT markers (as snail, slug, notch-1, and e-cadherin) on UFs, FDHLCs, and dedifferentiated fibroblasts (DFs) by comparison with the expression profile of MCF-7 cells (human breast adenocarcinoma cell line) and human hepatocytes. DFs were obtained after incubating FDHLCs for 1 week with growth medium. These cells lost hepatocyte-like morphology and restructured their cytoskeleton to regain a spindle-shaped form (Fig. 9A,B). They restarted to proliferate and reached confluency in one week after being plated at 1 × 106/cm2. Furthermore, DFs lost the expression of hepatocyte-specific mRNA markers albumin, αFP and α1AT (Fig. 9C). We observed that UFs expressed snail, notch-1 and e-cadherin which expression was down-regulated after hepatocyte differentiation (Fig. 9D). Slight neo-expression of slug and e-cadherin was observed in DFs whereas FDHLCs expressed none of these markers. Moreover, fibroblasts maintained ASMA expression throughout differentiation/dedifferentation processes (Fig. 9D). E-cadherin protein expression assay confirmed the RT-PCR results as UFs and DFs but not FDHLCs positively stained for this marker (Fig. 9E–H).

thumbnail image

Figure 9. Epithelial-mesenchymal transition analysis in differentiating fibroblasts. Comparison of morphological aspect of (A) FDHLCs and (B) DFs. Panel (C) shows the loss of albumin, αFP, and α1AT mRNAs expression in DFs. (D) RT-PCR showing snail, notch-1 and e-cadherin expression in UFs while these markers were down-regulated after hepatocyte differentiation. DFs neo-expressed low levels of slug and e-cadherin. Panels (E–H) show immunocytochemical staining for human e-cadherin in (E) UFs, (F) FDHLCs, (G) DFs, and (H) HHs. It confirms that epithelial marker e-cadherin was expressed in UFs, DFs, and HHs but not in FDHLCs. Abbreviations: DFs, dedifferentiated fibroblasts; FDHLCs, fibroblast-derived hepatocyte-like cells; HHs, human hepatocytes; LCS, liver cells suspension; MCF7, human breast adenocarcinoma cell line; UFs, undifferentiated fibroblasts. Pictures were taken at (A,B,E-G) 400× and (H) 200× magnifications.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

In this work, we present for the first time the endodermal differentiation potential of human skin fibroblasts. We initially demonstrated the phenotypic homology between fibroblasts and MSCs evoked elsewhere,5 and then compared both cell types for their potential to differentiate into mesodermal lineages. Classically, this differentiation potential is considered as a constitutive feature of MSCs and ultimately warrants for their phenotype.3 Distinction between MSCs and fibroblasts is usually based on this MSCs-specific potential although recent reports identified lineage-specific gene expression profiles.4, 6 Our data were consistent with previous works describing the potential of fibroblasts to differentiate into osteocytes7–14 and adipocytes.19 Discrepancies within the literature regarding mesodermal differentiation ability of fibroblasts could be explained by cell diversity depending on tissue source, donor age, culture passaging or in situ cell diversity.25 Accordingly, we were not able to differentiate fibroblasts aged more than 3 passages (Fig. 3). Together, these data attest that fibroblasts and MSCs share equivalent phenotypic features and mesodermal differentiation capacities and suggest that the corresponding assays are inefficient in distinguishing both these cell types. Our observations raise caution about the MSCs identity and have vast implications as tissue sampling for MSCs and fibroblasts are largely equivalent. It underlines the critical need for development of cell lineage characterization based on other features than phenotype (for example, gene array or metabolic assays).

Because MSCs were recently shown to possess hepatocyte differentiation capacity,26, 27 we studied the endodermal potential of fibroblasts. The data reported that even if MDHLCs displayed slightly broader marker expression, both cell types acquired similar morphological changes and liver-like functional features (urea secretion, glycogen storage). Conversely, whereas MDHLCs expressed CK8 RNA marker, none of these F/MDHLCs were able to acquire cytokeratin 8 or cytokeratin 18 expression on the protein level. However, both cell types positively stained for CX-32 suggesting a partial acquisition of the liver epithelial phenotype. The loss of pluripotency of fibroblasts after passage 3 which was also observed with MSCs after passage 8 (data not shown) reflects probably the deleterious effect of in vitro cell proliferation on native cell biology and functionality. Differences in the occurrence of this loss of functionality could depend on the robustness of these cell types.

We confirmed the hepatocyte differentiation potential of fibroblasts in vivo in a model of liver-injured SCID mice. These cells were able to engraft and express hepatic markers as albumin and αFP together with mesodermal markers. Furthermore, the engrafted cells displayed the expression of CK18 which evoked a positive role of the murine microenvironement on the acquisition of an epithelial phenotype. However, no human albumin could be detected in mice serum suggesting that the functionality of differentiated fibroblasts did not improve in vivo in our experimental conditions. Nevertheless, the in vivo data were similar to what could be observed with MSCs confirming the homology of both cell types.

Because EMT is a phenomenon of growing interest that has been described in normal organogenesis and in various physiopathological situations like fibrosis or neocarcinogenesis,28 we studied the EMT status of fibroblasts after endodermal differentiation and dedifferentiation. Our results suggested that FDHLCs behave like some described liver progenitors (for example, hepatic stellate cells) as they co-express mesodermal and endodermal markers.20 We confirmed this result after in vivo differentiation by showing coexpression of fibronectin and albumin in engrafted fibroblasts (Fig. 8B). Furthermore, we showed that fibroblasts are susceptible to lose the phenotype they acquired after differentiation when they are replaced in a growth factor-free medium. Together these data confirm that fibroblasts remain in a EMT status throughout differentiation/dedifferentiation processes by co-expressing mesodermal and endodermal markers and by presenting EMT-specific markers while recovering an undifferentiated state.

In conclusion, whereas fibroblasts were recently described as phenotypicaly similar to hBM-MSCs, our observations suggest that both cell types also share in vitro and in vivo differentiation properties. The comparison of hepatocyte potential of fibroblasts and hBM-MSCs argues that these latter cells possess more robust differentiation capability as fibroblasts could only be differentiated at early passages. As recently reported,29, 30 we confirmed the limitation of hepatocyte differentiation on the functional level in both cell types and the need for in vitro developments of cell culture conditions to improve the hepatocyte-like phenotype. However, regarding the ease of sampling and the advantageous proliferation capacities of fibroblasts, these cells should be further considered in the hepatocyte differentiation field.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

We are grateful to Floriane André and Nawal Jazouli for excellent technical assistance, to professor Jean-Christophe Renauld for providing SCID mice, and to Alberte Lefevre, Christine Galant, and Christine Sempoux for handling electron microscopy.

References

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Supplementary material for this article can be found on the H EPATOLOGY Web site ( http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html ).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.