Pancreatic and duodenal homeobox gene 1 induces hepatic dedifferentiation by suppressing the expression of CCAAT/enhancer-binding protein β

Authors

  • Irit Meivar-Levy,

    1. The Endocrine Institute, Sheba Medical Center, Tel-Hashomer, Israel
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Tamar Sapir,

    1. The Endocrine Institute, Sheba Medical Center, Tel-Hashomer, Israel
    2. Life Sciences, Bar-Ilan University, Ramat-Gan, Israel
    Search for more papers by this author
    • This work was performed in partial fulfillment of the requirements for the Ph.D. degree for T.S., S.G.-H., and V.A.

    • These authors contributed equally to this study.

  • Shiraz Gefen-Halevi,

    1. The Endocrine Institute, Sheba Medical Center, Tel-Hashomer, Israel
    2. Life Sciences, Bar-Ilan University, Ramat-Gan, Israel
    Search for more papers by this author
    • This work was performed in partial fulfillment of the requirements for the Ph.D. degree for T.S., S.G.-H., and V.A.

  • Vered Aviv,

    1. The Endocrine Institute, Sheba Medical Center, Tel-Hashomer, Israel
    2. Sackler School of Medicine, Tel-Aviv University, Ramat-Aviv, Israel
    Search for more papers by this author
    • This work was performed in partial fulfillment of the requirements for the Ph.D. degree for T.S., S.G.-H., and V.A.

  • Iris Barshack,

    1. The Institute for Pathology, Sheba Medical Center, Tel-Hashomer, Israel
    Search for more papers by this author
  • Nicholas Onaca,

    1. Rabin Medical Center, Beilinson Campus, Petah-Tiqva, Israel
    Search for more papers by this author
  • Eytan Mor,

    1. Rabin Medical Center, Beilinson Campus, Petah-Tiqva, Israel
    Search for more papers by this author
  • Sarah Ferber

    Corresponding author
    1. The Endocrine Institute, Sheba Medical Center, Tel-Hashomer, Israel
    2. Sackler School of Medicine, Tel-Aviv University, Ramat-Aviv, Israel
    • The Endocrine Institute, Sheba Medical Center, Tel-Hashomer 52621, Israel
    Search for more papers by this author
    • fax: 972-3-5302083


  • Potential conflict of interest: Nothing to report.

Abstract

It is believed that adult tissues in mammals lack the plasticity needed to assume new developmental fates because of the absence of efficient pathways of dedifferentiation. However, the well-documented ability of the transcription factor pancreatic and duodenal homeobox gene 1 (PDX-1) to activate pancreatic lineage development and insulin production following ectopic expression in liver suggests a surprising degree of residual plasticity in adult liver cells. This study seeks a mechanistic explanation for the capacity of PDX-1 to endow liver cells with pancreatic characteristics and function. We demonstrate that PDX-1, previously shown to play an essential role in normal pancreatic organogenesis and pancreatic β-cell function and to possess the potential to activate multiple pancreatic markers in liver, can also direct hepatic dedifferentiation. PDX-1 represses the adult hepatic repertoire of gene expression and activates the expression of the immature hepatic marker α-fetoprotein. We present evidence indicating that PDX-1 triggers hepatic dedifferentiation by repressing the key hepatic transcription factor CCAAT/enhancer-binding protein β. Hepatic dedifferentiation is necessary though not sufficient for the activation of the mature pancreatic repertoire in liver. Conclusion: Our study suggests a dual role for PDX-1 in liver: inducing hepatic dedifferentiation and activating the pancreatic lineage. The identification of dedifferentiation signals may promote the capacity to endow mature tissues in mammals with the plasticity needed for acquiring novel developmental fates and functions to be implemented in the field of regenerative medicine. (HEPATOLOGY 2007.)

Recent studies have demonstrated that pancreatic-specific transcription factors, previously known to control organ differentiation in the embryo, also possess instructive roles in diverting the developmental fate of adult liver cells along the pancreatic lineage.1–11 This process represents a novel and yet unexplained pathway of regeneration capacity in mammals. The first studied pancreatic transcription factor possessing such capacities is the pancreatic and duodenal homeobox gene 1 (PDX-1; also known as insulin-promoting factor 11). PDX-1 plays a central role in pancreatic development in the embryo12, 13 and in β-cell function in the adult pancreas.14 The role of PDX-1 in inducing a developmental redirection of the liver along the pancreatic lineage has been demonstrated both in vivo and in vitro, in Xenopus,5 mouse1, 2, 6–10 and human.4, 11 It has been suggested that liver cells that ectopically express PDX-1 may undergo a transdifferentiation process that results in an irreversible switch of one differentiated cell type to another.3, 5, 11 However, the mechanism by which PDX-1 modifies the hepatic phenotype and function is unknown.

Here we demonstrate that to acquire the alternate functional pancreatic lineage, adult liver cells are obliged to undergo a dedifferentiation process manifested by a loss of adult markers and the expression of the immature hepatic marker α-fetoprotein (AFP). PDX-1, but none of several other pancreatic transcription factors analyzed, induced hepatic dedifferentiation. The capacity of PDX-1 to affect a wide array of hepatic gene expression can be explained, at least in part, by its capacity to repress specifically the expression of the key hepatic transcription factor CCAAT/enhancer-binding protein β (C/EBPβ). It has been previously demonstrated that C/EBPβ induces pancreas-to-liver transdifferentiation.15–17 Our findings document the mechanistic characterization of the reverse process, which activates pancreatic lineage in liver and identifies proactive roles for PDX-1 and C/EBPβ as key components in this process.

Abbreviations

AAT, alpha-1 antitrypsin; Ad-INS, replication-deficient recombinant adenovirus that encodes human proinsulin complementary DNA under the control of the cytomegalovirus promoter; Ad-PDX-1, replication-deficient recombinant adenovirus that encodes rat PDX-1 complementary DNA under the control of the cytomegalovirus promoter; ADH1, alcohol dehydrogenase-1; ADH1B, alcohol dehydrogenase-1β; AFP, α-fetoprotein; ALB, albumin; C/EBPβ, CCAAT/enhancer-binding protein β; FOXA2, forkhead box A2; G6PC, glucose 6-phosphatase; GLUL, glutamate synthase; HNF, hepatocyte nuclear factor; LAP, liver-enriched activating protein; LIP, liver-enriched inhibitory protein; NGN3, neurogenin 3; PDX-1, pancreatic and duodenal homeobox gene 1; RT-PCR, reverse-transcription polymerase chain reaction.

Materials and Methods

Human Liver Cells.

Adult human liver tissues were obtained from 3 different liver transplantation surgeries from children 4-10 years old and 8 individuals over 40 years old. Liver tissues were used with approval from the Committee on Clinical Investigations (the institutional review board).

The isolation of human liver cells was performed as described.4, 18 The cells were cultured in Dulbecco's minimal essential medium (1 g/l of glucose) supplemented with 10% fetal calf serum (20 ng/ml; Cytolab, Ltd., Israel) and nicotinamide (10 mM; Sigma) and kept at 37°C in a humidified atmosphere of 5% CO2 and 95% air.

Viral Infection.

Liver cells were infected with recombinant adenoviruses at 500 moi (multiplicity of infection) for 5 days.

The adenoviruses used in this study were as follows: Ad-CMV-PDX-1, Ad-CMV-INS, Ad-RIP-β-GAL,19 and Ad-CMV-GFP (Clontech, BD Biosciences, United States); Ad-CMV-NEUROD1 (a generous gift from M. Walker and G. Ridner, Weizmann Institute, Israel); Ad-CMV-NKX6.1 (a generous gift from C.B. Newgard, Duke University); Ad-CMV-NGN3 (a generous gift from M.S. German, University of California at San Francisco); and Ad-CMV-LIP and Ad-CMV-LAP (a generous gift from H. Sakaue and M. Kasuga, Kobe University, Japan).

Animals.

Balb/c mice (8-9 weeks old, 18-19 g) were housed and treated with recombinant adenoviruses as described.1, 2

RNA Isolation and Real-Time Reverse-Transcription Polymerase Chain Reaction (RT-PCR).

The total RNA was isolated, and complementary DNA was prepared and amplified as described.2, 4 The primer pairs and annealing temperatures are listed in Supplementary Data 1. The TaqMan fluorogenic probes and the Assay-On-Demand (Applied Biosystems, Foster City, CA) used in this study are listed in Supplementary Data 2.

C-Peptide Detection.

C-peptide secretion was measured in primary cultures of adult liver cells 3-5 days after the initial exposure to the viral treatment as described.4 C-peptide secretion to the medium was measured by a radioimmunoassay with a human C-peptide radioimmunoassay kit (Linco Research, St. Charles, MO; <4% cross-reactivity to human proinsulin). C-peptide secretion was normalized to the total cellular protein measured by a Bio-Rad protein assay kit.

Immunofluorescence.

Human liver cells treated by Ad-PDX-1 for 5 days were plated on glass cover slides in 6-well culture plates. Forty-eight hours later, the cells were fixed and stained as described.4 The antibodies used in this study are listed in Supplementary Data 3. The slides were analyzed with a laser scanning confocal microscope (LSM-1024, Bio-Rad).

Western Blot Analyses.

Protein extraction, separation, and western blot analyses were conducted as described.19 For protein immunoprecipitation, 500 mg of whole-cell extracts was incubated with 5 mg of rabbit polyclonal anti–rat C/EBPβ antibodies (Santa Cruz Biotechnology) at 4°C for 16 hours. Protein A–Sepharose CL-4B (Amersham Bioscience, Sweden) was added to the samples, which were incubated for an additional 2 hours at 4°C. Agarose beads were collected through 1 minute of centrifugation at 12,000g and washed 3 times with a lysis buffer. The samples were separated on sodium dodecyl sulfate–polyacrylamide gel electrophoresis and immunoblotted with mouse monoclonal anti–human C/EBPβ antibodies (1:1000; Santa Cruz Biotechnology).

Histology and Staining.

Livers were fixed and stained as described.1, 2

Flow Cytometry.

Cells (5 × 105) were collected, washed twice in phosphate-buffered saline (5 minutes, 1000g), and resuspended in 400 μl of phosphate-buffered saline containing 1 mg of ribonuclease per milliliter and 20 μl of a propidium iodide solution (2 mg/ml). Flow cytometry was performed (FACSCalibur, Becton Dickinson, Heidelberg, Germany) with the CellQuest program.

Statistical Analyses.

Statistical analyses were performed with a 2-sample Student t test assuming unequal variances.

Results

PDX-1 Suppresses the Expression of Adult Hepatic Genes.

The ability of PDX-1 and other pancreatic transcription factors to activate pancreatic markers in the liver has been extensively documented.1–11 Because different transcription factors may have distinct effects on this process, we analyzed the effects of pancreatic transcription factors on the hepatic repertoire. Using recombinant adenovirus–mediated gene delivery, we expressed PDX-1, neurogenic differentiation 1 (NEUROD1), NK6 transcription factor related, locus 1 (NKX6.1), or neurogenin 3 (NGN3; also known as NEUROG3) in primary cultures of adult human liver cells and evaluated their effects on the characteristic markers of differentiated liver cells. RT-PCR analyses revealed that Ad-PDX-1 reduced the expression of the mature liver cell markers; albumin (ALB), alcohol dehydrogenase-1β (ADH1B), glucose 6-phosphatase (G6PC), and glutamate synthase (GLUL) genes, but activated the expression of the immature hepatic marker AFP. None of the other pancreatic transcription factors tested had significant effects on hepatic markers (Fig. 1A,B). These effects of PDX-1 were confirmed at the protein level through western blot analysis (Fig. 1C). A clear inhibitory effect of PDX-1 on the host repertoire of markers was manifested with double immunofluorescence analyses (Fig. 2A,B); each of the PDX-1–expressing cells exhibited a reduction in adult hepatic markers, whereas PDX-1–negative cells remained unaffected (Fig. 2A,B). In agreement with results obtained by RT-PCR and immunoblotting, PDX-1, but none of the other pancreatic transcription factors, increased the staining observed for the immature hepatic marker AFP (Figs. 1 and 2C,D).

Figure 1.

PDX-1 suppresses adult hepatic marker gene expression in human liver cells in vitro. RT-PCR analyses of human liver cells treated with Ad-PDX-1, Ad-NEUROD1, Ad-NKX6.1, or Ad-NGN3 for ALB, ADH1B, G6PC, GLUL, and AFP gene expression. Ad-INS–infected cells serve as both the viral infection and the produced proinsulin control. (A) Representative results of ethidium bromide staining of agarose-separated polymerase chain reaction products. (B) Quantitative real-time RT-PCR analyses presented as the relative levels of the mean ± standard deviation versus Ad-INS–treated liver cells (n ≥ 8 in 4 different experiments; *P < 0.005, **P < 0.01). (C) Western blot analyses of ALB, AAT, AFP, and PDX-1 proteins in control untreated (lane 1), Ad-INS–treated (lane 2), and Ad-PDX-1–treated (100 and 500 moi, lanes 3 and 4, respectively) cells. Mitogen-activated protein kinase served as the protein load control. Representative results (n = 4).

Figure 2.

PDX-1 represses hepatic markers in adult human liver cells without inducing cell proliferation. Double immunofluorescence analyses for (A,B) hepatic AAT (green) and PDX-1 (red), (C,D) AFP (green) and PDX-1(blue), (E,F) PDX-1 (green) and Ki67 (blue), and (G,H) C-peptide (green) and Ki67 (blue). Parts A, C, E, and G present control treated liver cells, and parts B, D, F, and H present Ad-PDX-1–treated cells. Cell nucleoli were stained by propidium iodide (red). The scale bar is 10 μm. (I) Cell cycle analyses at 2, 4, and 5 days after the Ad-PDX-1 treatment in adult human liver cells versus control Ad-INS–treated cells. The data are presented as the mean ± standard deviation (n ≥ 4 in 2 different experiments).

Importantly, PDX-1–induced hepatic dedifferentiation was not associated with increased cell proliferation (Fig. 2E,F). PDX-1 neither increased the number of cells present in the S phase nor increased the number of apoptotic cells in comparison with control Ad-INS–treated cells (Fig. 2I). These data strongly suggest that the PDX-1–mediated activation of the pancreatic repertoire in the liver represents a bona fide liver-to-pancreas transdifferentiation process.

PDX-1 Suppresses Adult Hepatic Markers In Vivo.

The capacity of PDX-1 to activate the pancreatic lineage in the liver was demonstrated originally in vivo.1, 2 Moreover, several studies have suggested that prolonged ectopic expression of PDX-16, 7 but not NEUROD1 in most cells in the liver may cause hepatic dysfunction.6 To analyze whether some of the PDX-1 effects on hepatic function are attributable to its capacity to repress the hepatic repertoire of gene expression, we examined the effect of PDX-1 on mature hepatic markers in mice. As previously reported, ectopic PDX-1 expression delivered by a first-generation recombinant adenovirus is transient and peaks at day 5.1, 2 Indeed, the PDX-1 effect on the expression of hepatic markers displayed similar temporal characteristics, with maximal repression of adult hepatic markers on the same day (Fig. 3A). However, 30 days after the PDX-1 treatment, as the levels of the transgene decreased, hepatic gene expression was restored, and the expression of ALB, alcohol dehydrogenase-1 (ADH1), and alpha-1 antitrypsin (AAT) genes was comparable to that in age-matched control adenovirus-treated mice (Fig. 3A). This time course suggests that the effect of PDX-1 on the hepatic repertoire depends on its continuous expression in the liver. Double immunohistochemistry analyses of liver sections, 5 days after Ad-PDX-1 administration, revealed an inverse correlation between PDX-1 and ALB; all ALB-positive cells were negative to PDX-1 nuclear staining, whereas PDX-1–positive cells had very low ALB levels (Fig. 3B). As in the primary culture in vitro, PDX-1 expression in vivo did not induce accelerated cell proliferation, as determined by Ki67 staining (data not presented).

Figure 3.

PDX-1 suppresses adult hepatic markers in mice livers, in vivo. (A) Quantitative real-time RT-PCR analyses for ADH1, ALB, and AAT gene expression levels. Ad-INS or Ad-β-gal served as viral infection controls. The results are presented as the relative levels of the mean ± standard deviation versus untreated control liver cells (n ≥ 6; *P < 0.005, **P < 0.05). (B) Double immunohistochemical analysis for PDX-1 (nuclei, arrow) and ALB (cytoplasm, headless arrow) in mice liver sections 5 days after Ad-PDX-1 administration.

These data suggest that PDX-1 has the capacity to repress the adult hepatic repertoire in mice in vivo, as it does in human liver cells in vitro.

PDX-1 Suppresses C/EBPβ Expression in Liver Cells.

It seems unlikely that PDX-1 could directly repress the expression of multiple genes in the liver through a direct effect.20 Therefore, we tested whether hepatic transcription factors could mediate the wide effects of PDX-1 on hepatic markers. PDX-1, but none of the other pancreatic transcription factors, decreased only the C/EBPβ expression (Fig. 4A; the data are not presented for the rest). The C/EPBβ transcript is a single 1.4-kilobase messenger RNA with 4 distinct translation initiation sites. However, at the protein level, multiple C/EBPβ isoforms have been reported. The full-length isoform (38/40 kDa) C/EBPβ and the 34-kDa isoform liver-enriched activating protein (LAP) are identical in their activity, whereas the liver-enriched inhibitory protein (LIP), which encodes a 14-kDa protein, lacks the trans-activation domain, and serves as a dominant negative inhibitor of C/EBPβ/LAP activity.21, 22

Figure 4.

PDX-1 inhibits C/EBPβ without affecting additional hepatic transcription factor gene expression. (A) Quantitative real-time RT-PCR of hepatic transcription factors: C/EBPβ, C/EBPα, FOXA2, HNF1α, HNF4, and HNF6 gene expression in Ad-PDX-1–treated or Ad-NKX6.1–treated adult human liver cells. The results are presented as the relative levels of the mean ± standard deviation versus Ad-GFP–treated liver cells (n ≥ 8 in 4 different experiments; *P < 0.01). (B) Western blot analyses for C/EBPβ (45 kDa) and its variant LAP (35 kDa) and PDX-1 in control untreated (lane 1) and Ad-PDX-1–treated cells (100 and 500 moi, lanes 2 and 3, respectively). Mitogen-activated protein kinase served as the protein load control. Double immunofluorescence analyses for C/EBPβ (blue) and PDX-1 (green) in (C) control treated liver cells and (D) Ad-PDX-1–treated cells. Part D is presented by an RGB filter split: (D1) red, (D2) green, and (D3) blue. Cell nucleoli were stained by propidium iodide (red). The scale bar is 10 μm.

Ectopic PDX-1 expression in adult human liver cells resulted in diminished C/EBPβ and LAP levels (Fig. 4). PDX-1 reduced the protein levels of C/EBPβ and LAP in a dose-dependent manner (Fig. 4B). Moreover, double immunofluorescence for PDX-1 and C/EBPβ revealed that although all control liver cells exhibited strong nuclear C/EBPβ staining in culture, PDX-1–expressing cells exhibited diminished C/EBPβ levels(Fig. 4C).

This raises the possibility that the effect of PDX-1 on hepatic dedifferentiation could be mediated by its capacity to decrease C/EBPβ levels.

Hepatic Dedifferentiation Is Obligatory for PDX-1–Induced Activation of the Pancreatic Lineage in the Liver.

To further analyze the possible role of C/EBPβ activity in mediating the effects of PDX-1 on the transdifferentiation process, we manipulated the C/EBPβ activity in adult human liver cells in a PDX-1–independent manner, using ectopic expression of LAP and LIP. Indeed, LIP, the dominant negative inhibitor of C/EBPβ activity, repressed the expression of adult hepatic markers, whereas LAP, which mimics C/EBPβ function, restored adult liver characteristics, as manifested by decreased levels of AFP and increased levels of adult hepatic markers, including that of C/EBPα (Fig. 5 and Supplementary Fig. 1).

Figure 5.

CEBPβ prevents the effect of PDX-1 on the hepatic phenotype. (A) Quantitative real-time RT-PCR analyses for ALB, ADH1B, G6PC, GLUL, and AFP gene expression of human liver cells treated with Ad-LIP, Ad-LAP, and/or Ad-PDX-1. The results are presented as the relative levels of the mean ± standard deviation versus control liver cells (n ≥ 7 in 4 different experiments; *P < 0.05 for all). (B) Western blot analyses for PDX-1, C/EBPβ (LIP and LAP), AFP, and AAT proteins in control untreated cells (lane 1) and cells treated with Ad-PDX-1 (lane 2), Ad-LIP (lane 3), and Ad-LAP (lane 4). Mitogen-activated protein kinase served as the protein load control. Representative results (n = 5).

To examine whether CEBPβ and the restoration of adult hepatic markers prevent the activation of the pancreatic lineage, we ectopically expressed LAP in a primary culture of adult human liver cells in combination with PDX-1. LAP prevented the effects of PDX-1 on both the hepatic dedifferentiation and the activation of the pancreatic lineage in liver cells (Figs. 5A and 6). These data suggest an inhibitory role of C/EBPβ activity on PDX-1–induced activation of the pancreatic lineage and function in the liver.

Figure 6.

Hepatic dedifferentiation is obligatory for the activation of the pancreatic lineage. Quantitative real-time RT-PCR analyses for (A) the pancreatic hormone, (C) neuroendocrine secretory granule protein, and (D) endogenous pancreatic transcription factor gene expression. The results are presented as the relative levels of the mean ± standard deviation versus untreated control liver cells. (B) C-peptide secretion at KRB media containing 17.5 mM glucose in response to Ad-PDX-1 and/or Ad-LAP treatments (n ≥ 7 in 4 different experiments; *P < 0.05).

Diminished C/EBPβ Activity Per Se Is Insufficient for Inducing Mature Hepatic Pancreatic Phenotype in the Liver.

We further examined whether decreased C/EBPβ activity per se is sufficient for inducing the pancreatic phenotype and function in the liver. A PDX-independent decrease in the hepatic markers was induced with the ectopic expression of LIP. The inhibition of C/EBPβ activity by ectopic LIP expression modestly activated the expression of the INS gene in adult human liver cells (Fig. 7A). The activation of INS gene expression by LIP was substantially lower than that induced by PDX-1; however, combined PDX-1 and LIP expression resulted in a marked stimulation of the expression of the INS gene. The pronounced effect of LIP on INS gene expression is in agreement with the documented inhibitory effect of C/EBPβ on the activation of INS gene expression in β-cells.23 Unlike ectopic PDX-1 expression,4LIP expression alone did not induce differentiated characteristics specific to β-cells, such as glucose-stimulated C-peptide secretion (Fig. 7B) or the expression of pancreatic transcription factors that characterizes mature pancreatic endocrine differentiation. LIP induced the expression of NGN3 alone (Fig. 7C). NGN3 expression marks immature endocrine progenitor cells in the pancreas and is not expressed in adult pancreatic islets.24 Interestingly, the coexpression of PDX-1 and LIP resulted in a specific increase in the expression pancreatic specific transcription factors, including the expression of the endogenous PDX-1 gene (Fig. 7D).

Figure 7.

Hepatic dedifferentiation promotes the PDX-1 effect on INS gene expression and C-peptide secretion but is insufficient to induce a mature pancreatic phenotype in liver. (A) Quantitative real-time RT-PCR analyses for the pancreatic hormone gene expression. The results are presented as the relative levels of the mean ± standard deviation versus untreated control liver cells. (B) C-peptide secretion at KRB media containing 17.5 mM glucose in response to Ad-PDX-1 and/or Ad-LIP treatments (n ≥ 7 in 4 different experiments; *P < 0.05, **P < 0.01). (C) RT-PCR analyses for the pancreatic transcription factors in adult control liver cells (lane 1), in Ad-PDX-1–treated liver cells (lane 2), and in Ad-LIP–treated liver cells (lane 3). Representative results of ethidium bromide staining of agarose-separated polymerase chain reaction products (n = 8). (D) Quantitative real-time RT-PCR analyses for endogenous PDX-1, ISL-1, and NGN3 gene expression. The results are presented as the relative levels of the mean ± standard deviation versus untreated control liver cells (n ≥ 8 in 4 different experiments; *P < 0.05, **P < 0.01 between Ad-PDX-1 and combined treatment with both Ad-PDX-1 and Ad-LIP).

These data suggest that the loss of hepatic differentiation per se, mediated by a decrease in C/EBPβ activity, may be sufficient only for inducing an immature pancreatic phenotype in liver cells and further strengthen the notion that the pancreatic lineage could be a default developmental option of the hepatic program.25

Discussion

This study suggests a role for hepatic dedifferentiation in the activation of the pancreatic lineage in the adult liver and identifies PDX-1 as a hepatic dedifferentiation factor.

PDX-1, previously shown to be essential for controlling both pancreas organogenesis and mature β-cell function,26, 27 can, upon ectopic expression in adult liver cells, repress adult hepatic markers and activate AFP expression without inducing cell proliferation (Figs. 1–3 and the supplementary material). The effects of PDX-1 on the hepatic state of differentiation and function are accompanied by and appear to be mediated in part by its capacity to suppress the expression of the transcription factor C/EBPβ. Hepatic dedifferentiation and C/EBPβ repression were induced by PDX-1, but not by several other pancreatic transcription factors analyzed. However, the capacity of NKX 6.1 and NEUROD1 to activate the pancreatic lineage in the liver was augmented upon C/EBPβ repression by LIP, and this further strengthened the notion that hepatic dedifferentiation is necessary for the activation of the alternate pancreatic repertoire in the liver (Supplementary Fig. 2). Considering this along with its capacity to activate the pancreatic repertoire and function in the liver,1–11 we suggest that PDX-1 may possess the unique capacity to induce liver-to-pancreas transdifferentiation.

The effects of PDX-1 on the activation of the pancreatic lineage were temporally and spatially distinct from those on hepatic dedifferentiation. Although irreversibly activating the pancreatic lineage in predisposed liver cells,1, 2, 4PDX-1 transiently suppresses adult hepatic markers in each liver cell (Figs. 2 and 3).

PDX-1–induced C/EBPβ suppression in the liver may have a wide physiological impact on this organ's differentiation and function as the C/EBP family of proteins plays important roles in liver development in the embryo and in normal hepatic function in adults.28, 29 Indeed, several publications have demonstrated that sustained high levels of ectopic PDX-1 expression in the liver in vivo cause hepatic dysmorphogenesis and ablated function.6, 7 However, these undesirable effects of prolonged PDX-1 expression on hepatic function do not invalidate its use in generating pancreatic function in the liver. Rather, they emphasize the importance of limiting both the duration and extent of PDX1 expression and its location to those cells capable of activating the pancreatic lineage.1, 2, 4

Previous studies have demonstrated that a single allele of either C/EBPβ or C/EBPα is sufficient for normal liver development.28, 29 In contrast, our data indicate that a 50% decrease in C/EBPβ levels induces a profound decline in the adult hepatic repertoire. There are several possible explanations for this apparent discrepancy; it is anticipated that the actual C/EBPβ repression in each PDX-1–positive liver cell is more substantial than that demonstrated at messenger RNA and protein levels (Figs. 1B,C and 4) because the transgene is expressed in only 50%-70% of the cells and does not affect the remaining 30%-50% of the cells in the culture.4 Moreover, in our experimental system, C/EBPα cannot compensate for the reduced expression of C/EBPβ because of its initial low levels (Supplementary Fig. 1). Therefore, cells expressing high levels of PDX-1 may completely lack both C/EBPα and C/EBPβ and therefore display dramatic reductions in hepatic markers.

Although it has been previously suggested that C/EBPβ–PDX-1 interplay may affect the activation of insulin gene expression in pancreatic tissues,5, 23, 30 our data raise the possibility that these proteins have a much wider reciprocal impact on the cellular phenotype. Although PDX-1 inhibits hepatic differentiation through C/EBPβ suppression, C/EBPβ not only inhibits insulin gene expression but also restricts pancreatic differentiation in the liver. Ectopic C/EBPβ (LAP) expression inhibits the PDX-1 effects on the activation of pancreatic hormone, pancreatic adult marker, and pancreatic transcription factor expression (Fig. 6). Thus, PDX-1 and C/EBPβ control the pancreatic and liver function in part through their reciprocal function. These data further suggest a role of these transcription factors in the separation between the hepatic and pancreatic repertoires and functions. Further studies are needed to determine whether PDX-1 and C/EBPβ directly interact with each other.

PDX-1 did not affect the expression of other liver-enriched nuclear factors, such as hepatocyte nuclear factor 1α (HNF1α), hepatocyte nuclear factor 4 (HNF4), hepatocyte nuclear factor 6 (HNF6), and forkhead box A2 (FOXA2). These factors are essential for both liver and pancreas development and function and, therefore, may not possess proactive roles in the interconversion between the two tissues. The distinct effects of these factors in the liver versus the pancreas could be manifested at posttranscriptional levels. Indeed, the expression of these factors remained unaltered also in the opposite process of activating the hepatic lineage in the pancreatic cell line AR42J-B13. However, it was noticed that hepatocyte nuclear factor 4α may increase CEBPβ activity once translocated into the nucleus upon dexamethasone treatment.30

Our data suggest a fundamental role for transcription factors in continuously controlling mature tissue differentiation and proper function in mammals. C/EBPβ and PDX-1 may play important roles in the developmental and functional separation of the liver from the pancreas, both demonstrating dominant roles in marking these tissues spatially, not only during embryonic development but also through adult life.16, 31 Furthermore, our study suggests that the developmental decisions made during embryonic organogenesis in mammals could be altered in differentiated cells through ectopic expression of dedifferentiation-inducing factors. The capacity of dominant transcription factors in instructing adult tissues to assume new fates may have important implications for regenerative medicine.

Acknowledgements

We thank M. D. Walker for fruitful discussions and for a critical review of the manuscript. C. V. E. Wright is acknowledged for the generous supply of anti–PDX-1 antibodies. M. D. Walker (Weizmann Institute), C. B. Newgard (Duke University), M. S. German and R. Gasa (University of California at San Francisco), and H. Sakaue and M. Kasuga (Kobe University, Japan) are acknowledged for their generous contributions of recombinant adenoviruses. We thank B. A. Sela (Institute for Chemical Pathology) and A. Schlosberg (Pathology Institute, Sheba Medical Center) for their fruitful help.

Ancillary

Advertisement