Background
The difficulties encountered in obtaining sufficient supply of transplantable β-cells is a major problem in cell therapy of type I diabetes. Liver may be a potential source of cells for β-cell engineering. Indeed, liver and pancreas derive from the same endodermal region during embryogenesis 1 and hepatocytes and β-cells share similar built-in glucose-sensing systems.
Among transcription factors involved in pancreatic β-cell specification, Pdx-1 plays a central role. All progenitors of the endocrine as well as the exocrine pancreas express Pdx-1 23. In the adult, Pdx-1 expression is mainly restrained to β-cells where it regulates important β-cell functions like insulin transcription. Several in vitro studies, using viral or stable plasmid gene transfer, show that Pdx-1 expression in hepatic cells results in reprogrammation into insulin producing cells 4567. Fusion of Pdx-1 to the VP16 activation domain from Herpes simplex virus (Pdx-1-VP16) leads to more efficient liver-to-pancreas conversion than Pdx-1 alone 89101112. Stable mouse Pdx-1 or Pdx-1-VP16 gene transfection initiates conversion of rat epithelial liver stem-like WB cell line into pancreatic endocrine precursor cells 1314. In these cells, long-term high glucose (HG) culture in vitro is necessary for further pancreatic endocrine differentiation.
Safety of gene therapy remaining a prime concern, protein transduction offers a more secure alternative to induce stem cell differentiation. Indeed, protein transduction domains (PTD) allow proteins to translocate across the cytoplasmic membrane. Due to an Antennapedia-like PTD in its structure, Pdx-1 protein can permeate different cell types and induces insulin expression in pancreatic ducts 1516. However, in human embryonic stem cells, the adjunction of a PTD domain derived from the HIV TAT protein (TAT) is necessary for Pdx-1 cell transduction 17. Up to now, no study using protein transduction has achieved liver-to-endocrine pancreas conversion in vitro.
Here, we evaluate if Pdx-1(-VP16) proteins have the capacity to induce a pancreatic endocrine shift in WB cells. To achieve this, we treat WB cells with Pdx-1 or Pdx-1-VP16 proteins, containing their own PTD, or fused to the PTD of TAT.
Methods
Recombinant proteins synthesis
Full-length mouse Pdx-1 and Pdx-1-VP16 open reading frames were cloned into pET28b-TAT-v2-1 expression plasmid containing the HIV TAT protein PTD (kind gift from S. Dowdy), in order to construct Pdx-1, Pdx-1-VP16, TAT-Pdx-1 and TAT-Pdx-1-VP16. PTDPdx-1-eGFP was constructed by fusing the PTD of Pdx1 (RHIKIWFQNRRMKWKK) to eGFP and subsequent cloning into pET28b-TAT-v2-1. TAT-eGFP was constructed by insertion of eGFP into pET28b-TAT-v2-1. eGFP was cloned into pET21a(+) expression vector (Novagen, WI, USA). See Additional data 1 for more details.
<p>Additional file 1</p>
Supplementary materials and methods.
Click here for file
Recombinant proteins were produced according to Studier's method of auto-induction 18. Proteins were purified by Ni2+ affinity chromatography on Protino Ni-TED resin (Macherey-Nagel, France), diafiltrated on Centricon-Plus-20 centrifugal filter devices (Millipore, France), and stored at -20°C in PBS/pH8.0/25%glycerol.
Luciferase assay
The RIP2-reporter gene was constructed by cloning the [-683 bp, +11 bp] 5' flanking region of rat insulin-II gene into the EcoRV-site of pGL4.10 [luc2] (firefly luciferase) (Promega, France).
18 × 103 HepG2 cells/well were seeded onto a 96-well plate and grown 24 h in 10% FCS DMEM (5% CO2, 37°C). 0,12 μg of RIP2-reporter and 0,13 μg of HSV-TK-hRluc control vector (renilla luciferase, pGL4-74, Promega) were co-transfected in these cells using JetPEI reagent (Polyplus-Transfection, France). 12 h after transfection, the medium was replaced with medium containing 5 μM protein or storage buffer for negative controls. 36 h later, cells were assayed for luciferase activities using Dual-Glo-Assay-System (Promega) and a scintillation counter (MicroBeta-Trilux, Wallac/Perkin-Elmer, France).
WB cell culture and protein treatment
WB-F344 cells, kindly provided by N. Malouf 19, were grown in low glucose (LG) medium: 10% FCS RPMI-1640 (11 mM glucose, Invitrogen, France) (5% CO2, 37°C).
To evaluate transduction efficiency, 50 × 103 WB cells/well were seeded onto a 24-well plate. 24 h later, the medium was replaced with fresh medium containing 15 μM eGFP, PTDPdx-1-eGFP, or TAT-eGFP proteins and incubated for another 24 h before confocal microscopy analysis.
To evaluate effects of Pdx-1(-VP16) proteins, 25 × 103 WB cells/well (passage 15) were seeded onto a 96-well plate with medium adjusted to HG concentration (25 mM D-glucose, Sigma-Aldrich). 12 h later, the medium was replaced by HG medium containing 1 μM protein (Pdx-1 (n = 4), TAT-Pdx-1 (n = 4), Pdx-1-VP16 (n = 5), or TAT-Pdx-1-VP16 (n = 5)) or storage buffer for HG control cells (n = 5). The medium was replaced every 3–4 days. Control cells in LG medium (n = 5) were also grown. To favour differentiation rather than proliferation, WB cells were treated for 2 weeks without being trypsined. Then mRNAs were collected and analysed by RT-PCR. Each sample (n) represents one culture well.
Laser scanning confocal microscopy
Cells were treated 10 minutes with trypsin/EDTA, washed with PBS, stained with Vybrant-CM-Dil (1:200 dilution in PBS, Molecular Probes), and washed twice. eGFP signal (λexc 488 nm, λem 507 nm) and CM-Dil staining (λexc 553 nm, λem 570 nm) were examined by laser scanning confocal microscopy (Nikon TE-2000, France).
RT-PCR analysis
mRNAs were isolated with the Dynabeads-mRNA-Direct Kit (Invitrogen). First-strand cDNA was synthesized using M-MuLV reverse transcriptase (Promega) and random pentadecamer primers. The resulting cDNA was amplified for 35 cycles [94°C 30 sec, 59/60°C 30 sec, 72°C 40 sec] with RedTaq DNA Polymerase (Sigma-Aldrich) on a 9700 thermocycler (Applied-Biosystems, France) using primers listed in Additional data 2.
<p>Additional file 2</p>
List of primer information for RT-PCR.
Click here for file
Results
Pdx-1 and TAT PTDs transduced HepG2 and WB cells
To assess PTDPdx-1- or TAT-mediated protein transduction in liver cells in vitro, we evaluated the ability of Pdx-1 and TAT PTDs to deliver fusion proteins into WB and HepG2 cells. To achieve this, we produced three recombinant eGFP fusion proteins: PTDPdx-1-eGFP, TAT-eGFP and eGFP alone as a negative control (Fig. 1a). Intracellular eGFP fluorescence was detected in cells treated with PTDPdx-1-eGFP or TAT-eGFP – as punctuated or diffused signals (Fig. 1b). No cellular uptake was detected in cells treated with eGFP protein lacking PTD (Fig. 1b). These results indicated clearly (i) that the efficiency of HepG2 and WB cell transduction was dependant on the presence of a PTD domain, and (ii) that PTDPdx-1efficiently delivered fusion proteins into cells.
<p>Figure 1</p>
PTDPdx-1 and TAT fusion protein synthesis and their ability to transduce HepG2 and WB cells
PTDPdx-1 and TAT fusion protein synthesis and their ability to transduce HepG2 and WB cells. (a) Schematic structure (top panel) and purity (bottom panel) of fusion proteins. PTDPdx-1 and TAT represent protein transduction domains of the Pdx-1 protein and the HIV TAT protein, respectively. His represents hexahistidine tag used to purify proteins by His-tag affinity chromatography. Purified proteins were run on a SDS-PAGE gel (8%) stained with Coomassie blue. Molecular weights are 28 kDa for eGFP, 31 kDa for PTDPdx-1-eGFP and TAT-eGFP. (b) Observation by confocal microscopy of eGFP fluorescence in HepG2 and WB cells treated for 24 hours with 15 μM PTDPdx-1-eGFP, TAT-eGFP, or eGFP protein lacking PTD. Treated cells were observed by confocal microscopy without being fixed in order to exclude artifactual protein uptake 24. CM-Dil was used to visualize cytoplasmic membrane (red staining). Scale bars = 50 μm.
![]()
Pdx-1(-VP16) protein transduction increased insulin promoter activity
In order to convert WB cells into insulin producing, cells we produced and purified four Pdx-1 proteins: Pdx-1, Pdx-1-VP16, TAT-Pdx-1 and TAT-Pdx-1-VP16 (Fig. 2a). Rat insulin-II promoter activity was significantly increased by treatment with any of the Pdx-1(-VP16) proteins (Fig. 2b). These results showed that these four Pdx-1(-VP16) proteins transduced hepatic cells and, in part, went to the nucleus and activated a β-cell specific promoter. Fusion of TAT to these proteins, which consequently contain two PTDs in their structure, increased insulin promoter activity in an equal manner as PTDPdx-1 alone. Pdx-1 fusion to VP16 activation domain did not further enhance insulin promoter activity.
<p>Figure 2</p>
Synthesis and biological activity of Pdx-1, TAT-Pdx-1, Pdx-1-VP16 and TAT-Pdx-1-VP16 fusion proteins
Synthesis and biological activity of Pdx-1, TAT-Pdx-1, Pdx-1-VP16 and TAT-Pdx-1-VP16 fusion proteins. (a) Schematic structure (top panel) and purity (bottom panel) of Pdx-1, TAT-Pdx-1, Pdx-1-VP16 and TAT-Pdx-1-VP16 fusion proteins. PTDPdx-1 represents protein transduction domain of the Pdx-1 protein, which is naturally present in the four Pdx-1(-VP16) fusion proteins. TAT represents protein transduction domain of the VIH TAT protein. VP16 is the activation domain of Herpes simplex virus I. His represents hexahistidine tag used to purify proteins by His-tag affinity chromatography. Purified proteins were run on a SDS-PAGE gel (8%) stained with Coomassie blue. Molecular weights are 45 kDa for Pdx-1 and TAT-Pdx-1, 60 kDa for Pdx-1-VP16 and TAT-Pdx-1-VP16. (b) Insulin promoter activity after Pdx-1(-VP16) protein transduction. Twelve hours after transient transfection of insulin promoter luciferase plasmid, HepG2 cells were treated with 5 μM Pdx-1(-VP16) fusion proteins for 36 hours. Then luciferase activity was measured. Data are expressed as mean + SEM (Control n = 14, Pdx-1 n = 13, TAT-Pdx-1 n = 7, Pdx-1-VP16 n = 12, TAT-Pdx-1-VP16 n = 19). Insulin promoter activity of controls was arbitrarily set at 1. Statistical comparisons were performed using a non parametric Mann-Withney test, *: p < 0.05, ***: p < 0.001.
![]()
Pdx-1, Pdx-1-VP16, TAT-Pdx-1 and TAT-Pdx-1-VP16 proteins induced the expression of insulin and pancreatic-related genes in WB cells
Whereas LG control cells expressed amylase and weakly insulin 2, but none of the other pancreatic genes tested here, HG control cells expressed transcription factors implicated in endocrine pancreatic differentiation: Ngn3, Nkx2.2 and Pdx-1 (Fig. 3). Expression of Kir6.2 was also detected. See additional data 3 for original electrophoresis gel images.
<p>Figure 3</p>
Gene expression profile of WB-cells treated with Pdx-1(-VP16) or TAT-Pdx-1(-VP16) proteins in high glucose culture
Gene expression profile of WB-cells treated with Pdx-1(-VP16) or TAT-Pdx-1(-VP16) proteins in high glucose culture. RT-PCR analysis was performed after two weeks of protein treatment. cDNA from rat pancreas served as control for pancreatic gene expression, and cDNA from rat liver served as positive control for albumin expression (*). Each sample is represented by a single column and corresponds to an independent biological repeat of the experiment (LG (n = 5), HG (n = 5), Pdx-1 (n = 4), TAT-Pdx-1 (n = 4), Pdx-1-VP16 (n = 5), TAT-Pdx-1-VP16 (n = 5)). Each gene is represented by a single row of coloured boxes. Black colouring represents RT-PCR positive samples, whereas white colouring represents negative samples. LG: Low Glucose control WB cells; HG: High Glucose control WB cells. (See additional data 3 for original electrophoresis gel images).
![]()
<p>Additional file 3</p>
Electrophoresis gel of RT-PCR analysis of WB cells treated with Pdx-1(VP16) proteins.
Click here for file
Taking advantage of the commitment of WB cells towards an endocrine phenotype in HG culture, Pdx-1(-VP16) protein treatment was performed on WB cells grown in HG medium. Proteins induced the expression of a wide panel of β-cell genes after two weeks of culture: NeuroD, Pax4, Nkx6.1, insulin 1, Glut-2. Expression of Pancreatic Polypeptide (PP), a non-β endocrine hormone, was also detected. Among different samples, some expressed all of these β-cell genes (Ngn3, NeuroD, Pax4, Nkx2.2, Nkx6.1, Pdx-1, insulin 1, Glut-2, Kir6.2) and others only some of them. Although insulin 1 was expressed, insulin 2 expression was not induced by Pdx-1(-VP16) protein treatment. Glucokinase (GK) was not detected despite using primers amplifying pancreatic as well as hepatic GK isoforms.
No pancreatic endocrine gene expression was observed at earlier time points (day 6 and 9 of treatment, data not shown).
The hepatic marker Albumin was absent in all conditions tested, including LG control cells.
Discussion
In the present study, we demonstrate the aptitude of Pdx-1 or Pdx-1-VP16 protein transduction to reprogram hepatic stem-like cells into β-like cells in vitro.
First, we verified that the PTD of Pdx-1 and TAT allow transduction into hepatic WB and HepG2 cell lines. Intracellular localization of fusion proteins revealed by confocal microscopy analysis indicate a true uptake of proteins and not mere adherence to the cell surface. Furthermore, transduced recombinant Pdx-1, TAT-Pdx-1, Pdx-1-VP16 and TAT-Pdx-1-VP16 proteins exert biological activity on an insulin promoter reporter system. Pdx-1 fusion to VP16 activation domain does not further enhance insulin promoter activity consistent with previous findings 10. Our results confirm reports of Pdx-1 transduction 1520 and are the first demonstration of Pdx-1-VP16 protein transduction.
Pancreatic differentiation experiments were conducted in an HG environment. In fact, long-term HG culture furthers liver cell commitment towards a pancreatic fate 46131421. Previous studies of Pdx-1(-VP16) expression in WB cells do not distinguish between respective contributions of transgene expression and HG culture on differentiation. Here, we show that HG culture alone converts WB cells into pancreatic endocrine precursor cells. In contrast to LG cultures, HG cultures express four pancreatic endocrine genes: Ngn3, Nkx2.2, Pdx-1 and Kir6.2. These results concur with Yang et al. study, where confluent culture of hepatic oval stem cells for 2 months in HG medium induces conversion into insulin-producing cells 21. Kir6.2 expression is at odds with previous reports where Kir6.2 was detected in WB cells overexpressing Pdx-1 or Pdx-1-VP16 genes only (i) after 2–3 months in HG culture 14, (ii) 40 days post-transplantation into diabetic mouse 13, or (iii) after Pax4 co-expression 22.
Pdx-1, TAT-Pdx-1, Pdx-1-VP16, or TAT-Pdx-1-VP16 proteins further induce expression of NeuroD, Pax4, Nkx6.1, insulin 1 and Glut-2 after two weeks of treatment. Some samples express just a part of these markers or display RT-PCR expression patterns similar to HG controls. Quantitative analysis would help to further nuance pancreatic gene expression in-between these samples, in particular up-regulation of the endogenous Pdx-1 gene. The heterogeneity between samples may be a consequence of: (i) infrequent liver to endocrine pancreas conversion, leading to few pancreatic gene positive cells which may be difficult to detect, (ii) different kinetics of gene expression between wells. According to Tang et al. study, our experiments do not reveal more efficient differentiation following Pdx-1-VP16 protein treatment compared to Pdx-1 protein treatment 14. TAT-mediated transduction does not lead to more advanced differentiation suggesting that containing two PTDs (PTDPdx-1 and TAT) does not increase transduction efficiency. Surprisingly, insulin 1 and Pax4, two of the pancreatic genes expressed after our protein treatments, are not detected in a previous in vitro study on WB cells transduced with Pdx-1 or Pdx-1-VP16 genes, even after 3 months of HG culture 14. Here, despite insulin 1 expression, neither insulin 2 nor glucokinase are detected pointing at the possible need for long-term culture in HG medium to obtain mature β-like cells 14. Moreover, Ngn3 expression in wells scoring positive for all other pancreatic genes, including insulin 1, suggests the presence of remaining subpopulations of immature pancreatic precursor cells.
Overexpression of Pdx-1 or Pdx-1-VP16 genes in hepatic cells leads to exocrine as well as a range of endocrine cell types 4781314. In our study, Pdx-1(-VP16) or TAT-Pdx-1(-VP16) protein treatments result in expression of PP, but not of glucagon suggesting that Pdx-1(-VP16) transduction in hepatic stem-like cells may lead to endocrine β-cell and non-β-cell phenocopies. Amylase, expressed in rat liver or exocrine pancreatic cells, is detected in all conditions tested here 23.
Recently, Koya et al. demonstrated that Pdx-1 protein delivery into diabetic mice restores euglycemia mainly through pancreatic β-cell regeneration. The authors observed β-cell gene expression and insulin synthesis in the pancreas and in the liver of treated mice suggesting hepatic insulin contribution to euglycemia 20. In complement, our findings provide the first direct evidence that Pdx-1(-VP16) protein transduction in conjunction with HG culture reprograms hepatic stem-like cells into cells displaying similarities with β-cells in vitro. At a point where strategies for targeted β-cell differentiation begin to surface, our study illustrates how simple exposition to Pdx-1(-VP16) protein in the surrounding medium triggers short-term pancreatic endocrine conversion. This study may contribute to the development of protein transduction therapy, a new concept to induce β-cell differentiation.