JOURNAL OF CELLULAR PHYSIOLOGY 206:147–159 (2006)
Characterization of Liver Function in
ZOE¨D. BURKE, CHIA-NING SHEN, KATE L. RALPHS, AND DAVID TOSH*
Centre for Regenerative Medicine, Department of Biology and Biochemistry,
University of Bath, Claverton Down, Bath, United Kingdom
cell line AR42J-B13 (B13) to hepatocytes based on the expression of liver proteins. We have extended our original observations to
determine: (1) the effects of Dex on pancreatic gene expression; (2) the time course of expression of liver enriched transcription
factors during conversion from pancreatic to hepatic phenotype; (3) the functional potential of transdifferentiated hepatocytes;
(4) the proliferative capacity of transdifferentiated hepatocytes; and (5) whether ectopic expression of transcription factors can
induce the hepatic phenotype in pancreatic B13 cells. The results were as follows. The B13 cell markers amylase, synaptophysin,
C/EBPb and C/EBPa were induced first, followed by HNF4a and then RXRa. Using RT-PCR analysis and immunolocalisation
studies, we detected hepatic markers (e.g., apolipoprotein B) in Dex-treated cells. In transdifferentiated hepatocytes albumin was
secreted, insulin stimulated lipid deposition and ciprofibrate enhanced the expression of catalase. Proliferation of trans-
cyclin D and phosphohistone H3 with liver proteins. Lastly, ectopic expression of C/EBPa or C/EBPb in AR42J-B13 cells was
sufficient to induce transdifferentiation, based on nuclear localization of HNF4a and induction of UDP-glucuronosyltransferase
molecular and cellular events that occur during transdifferentiation. J. Cell. Physiol. 206: 147–159, 2006. ? 2005 Wiley-Liss, Inc.
Transdifferentiation is the conversion of one cell type
to another and belongs to a wider class of cell-type
switches termed metaplasias that include conversion
Shen et al., 2004). Documented examples of transdiffer-
entiation include the appearance of hepatocytes in the
several in vitro models for the transdifferentiation of
pancreatic cells to hepatocytes (Shen et al., 2000; Tosh
et al., 2002) and the reverse, liver to pancreas switch
(Horb et al., 2003). Two models have been exploited for
the conversion of pancreatic cells to hepatocytes, one
based on the pancreatic cell line AR42J (Shen et al.,
2000) and the other based on the mouse embryonic
pancreas (Shen et al., 2003).
The AR42J rat cell line is an amphicrine cell line that
synthesizes and secretes the digestive enzyme amylase,
expresses the neuronal intermediate cytoskeletal ele-
ved in exocytosis of neurotransmitters (Christophe,
1994). AR42J possess many of the features typical of
on the observation that several different cell types can
be induced. For example, AR42J cells can convert
to pancreatic b-cells when cultured with b-cellulin
(Mashima et al., 1996a). A subclone of AR42J, referred
to as AR42J-B13 (B13 cells) has been isolated that is
more susceptible to conversion to b-cells when cultured
with HGF (Mashima et al., 1996b). Alternatively,
addition of glucagon-like peptide-1 can convert AR42J
cells to glucagon-secreting a-cells or insulin-producing
b-cells (Zhou et al., 1999). Enhancement of the exocrine
phenotype is observed in AR42J cells cultured with low
hepatocytes by extended culture with 1 mM dexametha-
sone (Dex) (Shen et al., 2000).
The liver has a multitude of functions including the
synthesis and secretion of serum proteins (Morgan and
Peters, 1971), the regulation of carbohydrate metabo-
lism (Mithieux, 1997) and the control of cholesterol
homeostasis (Hylemon et al., 2001). Hepatocytes also
mediate detoxification through activation of the Phase I
and II enzymatic pathways (Coughtrie et al., 1988;
Tephly and Burchell, 1990). These differentiated prop-
erties of hepatocytes are mediated and maintained
through the co-ordination of a battery of transcription
converging on gene promoters to achieve highly liver-
specific gene expression (Costa et al., 2003; Watt et al.,
2003). Two members of the liver-enriched transcription
factory family (C/EBPa and C/EBP b) have well-
established roles in the regulation of hepatocyte differ-
entiation (Diehl et al., 1994), regeneration (Greenbaum
et al., 1998), glucose homeostasis (Kimura et al., 1998),
and Phase I metabolism (Luc et al., 1996). HNF4a is
required to drive the differentiation of hepatoblasts to
hepatocytes and is important in hepatic morphogenesis
? 2005 WILEY-LISS, INC.
Contract grant sponsor: Biotechnology and Biological Sciences
Research Council, the Medical Research Council and the Well-
Chia-Ning Shen’s present address is Stem Cell Program,
Genomics Research Centre, Academica Sinica, Taipei, Taiwan,
*Correspondence to: David Tosh, Centre for Regenerative Medi-
cine, Department of Biology and Biochemistry, University of
Bath, Claverton Down, Bath BA2 7AY, UK.
Received 18 October 2004; Accepted 13 April 2005
while RXRa is involved in the regulation of bile acid,
steroid cholesterol, fatty acid, and xenobiotic metabo-
lism (Wan et al., 2000).
It is not known to what extent the B13 subclone
expresses synaptophysin or neurofilament and whether
the expression of these markers changes after Dex-
treatment. The first part of the investigation was there-
fore to distinguish in the B13 subclone, the two
simultaneous classes of Dex effect: (1) elevation of
activity of some pancreatic genes and (2) transdiffer-
entiation to a liver phenotype using time course studies
to document changes in the expression of both pancrea-
tic and hepatic markers. We also wanted to determine
whether transdifferentiated hepatocytes were able to
perform more active indicators of hepatocyte function
(e.g., albumin secretion). The second part of the
investigation was to determine the proliferative capa-
city of transdifferentiated hepatocytes under different
culture conditions. Here we propose that transdiffer-
entiated hepatocytes may provide a useful tool for
studying hepatocyte function.
MATERIALS AND METHODS
Dexamethasone, ciprofibrate, and phenobarbital were
obtained from Sigma Chemical Co (St. Louis, MO). Recombi-
nant human oncostatin M was from R&D System Inc.
(Minneapolis, MN). Oil Red O was purchased from Lamb
(London, UK). All other reagents were obtained as described
previously (Shen et al., 2000; Tosh et al., 2002).
Cell lines and culture conditions
B13 cells were grown as described (Shen et al., 2000). Dex
was added as a solution in ethanol to a final concentration of
and HGF were added as a solution in phosphate-buffered
saline (containing 0.1% bovine serum albumin) at a final
concentration of 10 ng/ml together with 1 mM Dex. Non-
essential amino acids (GibcoBRL, Invitrogen Corp) were
diluted 1/100 in culture medium containing 1 mM Dex with or
without 10 ng/ml HGF. Where indicated, phenobarbital and
ciprofibrate were added to a final concentration of 10 and
200 mM respectively.
Lipid accumulation in transdifferentiated hepatocytes
B13 cells were initially cultured in medium containing 1 mM
Dex for 7 days and then treated with Dex in the absence or
presenceof170nMInsulin. After 3days, cells were fixedin4%
paraformaldehyde and stained for oil red O. The staining was
performed as described previously (Novikoff et al., 1980) and
the average number of lipid droplets counted in 50 cells.
Measurement of albumin secretion
B13 cells were cultured for 14 days with 1 mM Dex. The
medium was removed, cells were washed twice with PBS, and
then incubated in 0.8 ml serum-free DMEM for 2 days. At
the end of the incubation period, the medium was used for
measuring albumin secretion. Albumin secretion was mea-
sured with a rat albumin ELISA Kit according to the
manufacturer’s instructions (Bethyl,Texas). For themeasure-
ment of total protein content, cells were washed twice with
PBS, lysed with buffer containing 20 mM HEPES (pH 7.6),
150 mM NaCl, 1 mM EDTA, 2 mM dithiothreitol (DTT), 1%
Triton X-100, and 10 mg/ml protease inhibitors (leupeptin,
aprotinin, and pepstatin). Protein concentration of whole-cell
extracts was determined using the Bio-Rad protein assay
reagent. Statistical analyses was performed on triplicate
samples using the Student paired t-test in Excel. Significance
was set at P<0.05.
Immunofluorescence analysis and antisera
B13 cells were immunostained as described in Shen et al.
(2000) with the following modifications. For synaptophysin
and neurofilament staining, the cells were fixed in acetone:-
methanol for 5 min at ?208C, rinsed with PBS, and incubated
in 2% blocking buffer. The antibodies were diluted and ob-
tained as follows: Rabbit anti-catalase (1/100, Rockland Inc.,
Gilbertsville, PA), rabbit anti-transferrin (TFN) (1/100), and
rabbit anti-albumin (ALB) (1/200) both from Sigma Chemical
Co, mouse anti-apolipoprotein B (ApoB) (1/100, Chemicon
International, Temecula, CA), sheep polyclonal anti-UDP-
glucuronosyltransferase (UGT) (1/300, Cypex Ltd, Dundee,
UK), mouseanti-CYP3A1 (1/300) (agift fromProfessor Roland
Wolf, University of Dundee, UK), rabbit anti-CYP2E1 (1/200)
(a gift from Dr Matthew Wright, University of Aberdeen, UK),
sheep anti-glucose-6-phosphatase (G6Pase) (1/300, a kind gift
from Professor Ann Burchell, University of Dundee, UK), goat
anti-HNF4a, rabbit anti-RXRa, and mouse anti-C/EBPb (1/
100) and rabbit anti-C/EBPa (1/100) were from Santa Cruz,
rabbit anti-C/EBPa (1/100) (a generous gift from Professor
100, BD Biosciences, Pharmingen), rabbit anti-phosphohis-
tone H3 and rabbit anti-Cyclin D (1/100, Upstate Lake Placid,
NY). Mouse anti-neurofilament (1/50, Sigma Chemical Co),
Anti-mouse, anti-sheep, and anti-rabbit fluorescein isothio-
cyanate conjugated IgG antibodies (1/100) were from Vector
Laboratories Inc. (Burlingame, CA).
Fluorescent specimens were viewed using a Zeiss confocal
microscope (LSM510) or a Leica DMRB microscope. Confocal
images were processed with Photoshop. For cell counting,
images were collected under the same conditions from at least
five different regions of treated cells for three individual
experiments. Positive cells were counted manually and the
total numbers for each specimen were between 200 and 300.
Values are the mean?SD for the selected regions. Statistical
analysis was performed for differences between treatments by
the Student’s t-test.
Transient transfection of AR42J-B13 cells
CMV-C/EBPa and CMV-C/EBPb plasmids were transiently
transfected into B13 cells using GeneJuice transfection
reagent (Novagen) with 2 mg DNA per dish, according to the
manufacturer’s instructions. Cells were incubated with trans-
fection complexes at 378C 5% CO2for 24 h and cultured for
5 days post-transfection.
RT-PCR was performed as described (Shen et al., 2003).
Total RNA was extracted from cultures using the TRI reagent
(Sigma, Poole, UK) and treated with RQ-1 DNase (Promega,
Southampton, UK) to remove contaminating genomic DNA.
First strand complementary DNA was synthesized using
SuperScript II reverse transcriptase (Invitrogen). The PCR
reactions were carried out under the following conditions:
denaturation at 948C for 1 min, annealing at 558C for 1 min,
and extension at 728C for 1 min. The number of cycles was 31.
Primers were obtained from Invitrogen, and primer sequences
used are as follows: For C/EBPb (sense primer: 50ACAAGCT-
GAGCGACGAGTAC 30, antisense primer: 50ACAGCTGCTC-
CACCTTCTTC 30), C/EBPa (sense primer: 50GTTCCA-
GATCGCACACTGCG 30, antisense primer: 50TGACCAAG-
GAGCTCTCAGGC 30),CYP3A1 (sense primer 50GGAAATTC-
GATGTGGAGTGC 30, antisense primer: 50AGGTTTGCC-
TTTCTCTTGGC 30), CYP2B1/2 (sense primer 50CGCATGGA-
GAAGGAGAAGTC 30, antisense primer: 50CCTCAGTGTT-
CTTGGGAAGC 30), CYP7A1 (sense primer 50CCTCCT-
GAGAGCGTGT 30), Albumin (sense primer 50GGTCA-
GAACCTCATTGTARRR 30, antisense 50ATTCACACTCTC-
GATTCCTTGGTGT 30, antisense primer 50ATGGAAGAGA-
GGCTGGGATT 30), b-Actin (sense primer 50TCCGTAAAG-
ACCTCTATGCC 30, antisense primer 50AAAGCCCATGC-
lysis buffer (20 mM HEPES (pH 7.6), 150 mM NaCl, 1 mM
BURKE ET AL.
EDTA, 2 mM DTT, 1% Triton X-100) containing a 1:100
dilution of protease inhibitor cocktail (Sigma). Cells were
centrifuged at 13,000 rpm and the supernatant was kept for
use. Protein concentration of cell extracts was determined
denatured by mixing with 2? SDS–PAGE loading buffer
(0.125M Tris-HCl (pH 6.8), 4% (w/v) SDS, 20% (v/v) glycerol,
0.2M DTT, 0.02% (w/v) bromophenol blue) and incubated at
1008C for 5 min. Unless otherwise stated, 5 mg of protein was
separated on a 5, 7.5, or 10% CriterionTMpre-cast Tris–HCl
polyacrylamide gel (Bio-Rad) and transferred on to BioTra-
ceNT1nitrocellulose membrane (Pall Corporation, Pensacola,
FL). The membrane was blocked using 5% (w/v) Marvel1in
0.1% (v/v) PBS-Tween and subsequently probed. Antibodies
were obtained and diluted as follows: anti-synaptophysin and
anti-transferrin both at 1:2,000, anti-neurofilament and anti-
amylase were used at 1:1,000 and 1:2,000 respectively, anti-
catalase 1:4,000, anti-UGT 1:10,000, anti-Cyp2E1 1:5,000.
Sources of antibodies were as described in the Immunofluor-
used as a loading control. Secondary antibodies were obtained
and diluted as follows: peroxidase labeled anti-rabbit or anti-
mousewere both 1:4,000(Amersham Biosciences, Bucks, UK),
rabbit polyclonal to sheep IgG (HRP) 1:4,000 (Abcam, Cam-
bridge, UK).The signal wasdetected with the ECLTMWestern
blotting analysis system (Amersham) and developed on
Characterization of AR42J-B13 cells
Short-term culture of AR42J cells with Dex elevates
expression of amylase but this response is quite distinct
from the longer-term effect of Dex on hepatic transdif-
ferentiation. The object of this part of the study was to
distinguish the two simultaneous classes of Dex effect
on the pancreatic phenotype and the switch from a
course of the Dex effect on the exocrine marker amylase
andfurther extend the resultsto showthat it is part of a
change in theexpression of abattery ofgenes (including
In response to Dex, there is an initial enhancement of
amylase expression which peaks around 3–5 days and
then gradually decreases in some cells thereafter
(Fig. 1A). At the end of the 2 week culture period,
amylase was either absent or weakly positive in a few
cells. We also determined the expression of neurofila-
ment and synaptophysin in B13 cells cultured in the
absence and presence of Dex. Both synaptophysin and
neurofilament were expressed in control B13 cells.
Following Dex-treatment, synaptophysin appears to
have a similar expression profile to amylase: with
highest levels detected at day 5, after which expression
declines considerably. The characteristic filamentous
control B13 cells, decreased to undetectable levels by
day 14 in culture upon treatment with Dex (Fig. 1A).
Cellsthatexpressed amylasedid nottendtoexpress the
hepatic protein albumin (Fig. 1B).
Changes in protein levels were also determined by
Western blotting (Fig. 1C), confirming an initial
increase in expression of both amylase and synaptophy-
sin and a gradual decrease to undetectable levels of
neurofilament as seen by immunostaining.
Induction of the hepatic phenotype in
transdifferentiated liver cells
and the functional capacity of transdifferentiated
hepatocytes, we examined the expression of several
markers that represent a range of hepatic functions:
transferrin (TFN) and albumin (serum protein produc-
tion), G6Pase (gluconeogenesis), ApoB (cholesterol
homeostasis), CYP2E1 and UDP-glucuronsyltransfer-
ase (UGT) (Phase I and II xenobiotic metabolism
respectively) over 14 days of Dex treatment (Fig. 2A).
The liver markers transferrin, G6Pase, UGT, CYP2E1,
and ApoB were expressed in B13 cells after 3 days of
Dex-treatment while albumin expression was not
observed until day 9. In TD hepatocytes, ApoB first
appeared as very punctate staining following Dex
treatment and was more intense in enlarged, flattened
cells (Fig. 2). UGT and CYP2E1 were localized to the
nuclear envelope and endoplasmic reticulum and the
intensity of staining of both increased during the period
of the liver markers increased over the 14 day period of
number of cells present at day 5 and day 14 of Dex-
ALB 0.5% (5d) and 14% (14d); ApoB 59% (5d) and 94%
(14d); CYP2E1 21% (5d) and 63% (14d) and UGT 36%
(5d) and 65% (14d). Furthermore, we confirmed some of
these changes by Western blotting analysis for the
selected markers TFN, ALB, and CYP2E (Fig. 2B).
Interestingly, Western blotting detected a slight decr-
result may indicate that these cells are secreting trans-
ferrin, a change not detectable by immunostaining.
In addition to immunostaining and western blotting,
we also performed RT-PCR analysis of albumin, carba-
moylphosphate synthetase (CPS), and several cyto-
chrome P450 family members, CYP3A1, CYP2B1, and
CYP7A1 (Fig. 3A,B). Transcripts for each of these
markers were detected, the first to be induced following
Dex treatment were CPS (Fig. 3A) and CYP3A1 at day
CYP2B1 was induced (Fig. 3B). Interestingly, albumin
transcripts were detected as early as 3 days post
induction, in contrast to the late (day 9) appearance of
the protein observed by immunostaining (Fig. 2). This
would suggest that post-transcriptional mechanisms
are in operation to delay the synthesis of albumin in
early stage TD hepatocytes. Cytochrome reductase was
also present in control B13 cells contrary to the findings
of Marek et al. (2003).
Albumin secretion and lipid
accumulation in TD hepatocytes
TD hepatocytes upregulate albumin at the level of
protein synthesis and gene expression (Figs. 2 and 3A),
we previously had no indication as to whether transdif-
et al., 2000). Using an ELISA-based assay we show for
the first time that TD hepatocytes do indeed secrete
albumin into the medium (Fig. 3C).
It is well known that insulin stimulates the synthesis
of lipid in the liver (Saltiel and Kahn, 2001). In order to
accumulate lipid, 7-day Dex-treated AR42J-B13 cells
were cultured in medium containing 1 mM Dex with or
without 170 nM insulin (Fig. 4). After 3 days, the cells
are visible and in 10-day Dex-treated cells on average of
3 droplets per cell were observed. In contrast, there was
a significantly higher number of prominent lipid
droplets found in cells treated with Dex and insulin,
TRANSDIFFERENTIATED HEPATOCYTES FROM PANCREATIC CELLS
insulin is able to stimulate the synthesis and accumula-
tion of lipids in TD hepatocytes.
are responsive to xenobiotics
In the present study, we unambiguously demonstrate
the expression and significant induction of the perox-
isomal enzyme catalase by ciprofibrate using indirect
immunofluorescence detection. We were interested to
investigate whether the TD hepatocytes respond appro-
priately to the xenobiotics ciprofibrate and pheno-
barbital. Previously we showed catalase activity in TD
hepatocytes based on a histochemical technique (Tosh
et al., 2002). Using an antibody raised against catalase,
we found that the enzyme was not expressed in control
pancreatic cells but was expressed in TD hepatocytes
synaptophysin in control and Dex-treated B13 cells. A: Culture with
Dex caused an initial increase in amylase and synaptophysin
expression with a peak at 3–5 days followed by a gradual decrease.
Neurofilament gradually declined over the time-course of Dex
treatment. Scale bars¼10 mM. B: Dual immunostaining for amylase
(red) and albumin (green) in B13 cells following 14 days treatment
Time course of expression of amylase, neurofilament, and
with Dex. C: Western blotting analysis showing profile of expression of
amylase (57 kDa), synaptophysin (34 and 37 kDa), and neurofilament
(200 kDa) in B13 cells over 14 days treatment with Dex. In order to
detect the presence of amylase in control B13 samples 15 mg of protein
was required compared to only 5 mg for Dex-treated samples. Tubulin
loading control also shown (50 kDa).
A: Time course of induction of transferrin (TFN), glucose-6-phospha-
tase (G6Pase), albumin, ApoB, CYP2E1, and UGT in B13 cells at 3, 5,
9, and 14 days post induction with 1 mM Dex. Untreated B13 cells are
also shown. All proteins were induced and gradually expression
increased as more B13 cells transdifferentiated to hepatocytes. Scale
Differentiated properties of transdifferentiated hepatocytes. bars¼10 mM. B: Western blotting analysis demonstrating the
increase levels of TFN (77 kDa), ALB (66 kDa), and CYP2E1
(56 kDa) at 1, 3, 5, 9, and 14 days post induction with 1 mM Dex.
TFN, ALB, and CYP2E1 expression in adult liver and a tubulin
loading control are also shown.
BURKE ET AL.
TRANSDIFFERENTIATED HEPATOCYTES FROM PANCREATIC CELLS
TD hepatocytes with ciprofibrate both in the absence
and presence of Dex. In the presence of the peroxisomal
proliferator ciprofibrate, catalase expression was en-
hanced even in the absence of Dex with approximately
20%–25% of cells staining positive for the enzyme
(Fig. 5A,B). Secondly, the combined treatment with
Dexandciprofibrate further enhancedtheexpressionof
the enzyme resulting in a large proportion (68%) of TD
cells expressing the enzyme as demonstrated by cell
counting and Western blotting analysis (Fig. 5A,B,E).
The Phase II enzyme testosterone/4-nitrophenol UGT
was not detected in control AR42J-B13 cells (Fig. 2), but
as we have previously indicated (Tosh et al., 2002), is
induced in TD hepatocytes following treatment with
1 mM Dex (Figs. 2 and 5C). Approximately 10% of cells
are positive for UGT following Dex treatment for
14 days. Addition of 10 mM phenobarbital in the absence
of Dex to TD hepatocytes dramatically increased the
expression of UGT in a small number of cells but
the overall number of positive cells was not altered
in the combined presence of Dex and phenobarbital
(Fig. 5C,D). Hepatocytes upregulate expression of UGT
in response to phenobarbital suggesting that the
response of TD hepatocytes is similar to that of normal
hepatocytes (Ritter et al., 1999).
Induction of liver enriched
Many liver-enriched transcription factors function in
a co-ordinated fashion to maintain the differentiated
state of hepatocytes (Costa et al., 2003). Previously, we
demonstrated that TD hepatocytes treated for 7 days in
the presence of Dex co-expressed G6Pase and the
transcription factor C/EBPb (Shen et al., 2000). To
extend this earlier observation, that induction of trans-
cription factors is coincident with the appearance of the
hepatic phenotype following Dex treatment of B13 cells,
we carried out a time course of immunostaining for a
number of additional transcription factors, HNF4a,
RXRa, C/EBPa, as well as C/EBPb alone or in combina-
tion with the functional hepatic marker UGT (Fig. 6).
The first factors to be detected 3 days after induction
were C/EBPb, exhibiting strong nuclear staining in a
small number of cells, and C/EBPa that showed weak
nuclear staining in more mature cells. In addition to
being expressed in the liver, HNF4a is also detected in
the pancreas and in pancreatic B13 cells it appears
cell. However, as previously demonstrated (Shen et al.,
2000), after 5 days of culture with Dex we observed
strong expression of HNF4a in the nuclei of transdiffer-
entiated B13 cells. The nuclear receptor RXRa showed
cells after 9 days of Dex treatment. Dual immunostain-
ing of TD hepatocytes with C/EBPb and UGT demon-
strate the coincident induction of both transcription
factor and functional marker following treatment with
Dex (results not shown). Remarkably, RT-PCR analysis
as 8 h following Dex treatment, while C/EBPa tran-
scripts were induced more weakly between 3 and 6 days
We previously described that ectopic expression of C/
B13 cells to hepatocytes—based on the loss of amylase
expression and induction of transferrin and transthyr-
etin (Shen et al., 2000). We have now extended these
studies to determine whether the C/EBPa isoform can
mediate transdifferentiation of B13 cells to hepatocytes
and also whether C/EBPb can induce additional liver
markers. Cells transfected with either C/EBPa or C/
EBPb were co-stained for UGT and HNF4a (Fig. 7A).
Following transfection with either of the C/EBP iso-
forms, UGT and HNF4a were detected. Double immu-
nostaining for C/EBPb and UGT demonstrates the
metabolizing enzymes in transdifferentiated hepatocytes. A: RT-PCR
analysis of carbamoylphosphate synthetase (CPS), albumin (Alb), and
C/EBPb and C/EBPa in control and Dex-treated B13 cells. B: RT-PCR
analysis of CYP Reductase, CYP3A1, CYP2B1, CYP7A1, and b-actin
transcripts 0, 3, 6, and 14 and 28 days post culture with 1 mM Dex. All
transcripts were induced following induction with Dex. C: Albumin
secretion in control and 14 day-Dex-treated B13 cells. Statistical
analyses was performed on triplicate samples using the Student
paired t-test in Excel. Significance was set at P<0.01.
Analysis of albumin expression and secretion and Phase I
BURKE ET AL.
with C/EBPb (Fig. 7B). However, due to cross reactivity
between antibodies, we were unable to carry out double
staining for UGT with C/EBPa and HNF4a. RXRa was
not detected in the cell nucleus transfected with either
construct (data not shown) indicating that either
additional factors are required to induce the transloca-
tion of this transcription factor or the timescale of the
transfection experiment may not have been sufficient to
observe the relatively late activation of RXRa which
appears 9 days following Dex treatment (Fig. 6). These
results demonstrate that the whole process of transdif-
ferentiation is relatively slow. It would be of interest to
or whether the cells have to get all the way through to a
mature hepatic phenotype before the process becomes
Effects of HGF on transdifferentiated hepatocytes
Homozygous inactivation of HGF, or its receptor c-
met, yields an embryonic lethal phenotype (Schmidt
et al., 1995). This is due to hypoproliferation and
apoptosis of the hepatic parenchyma implicating HGF
HGF on transdifferentiated hepatocytes, we used the
combined treatment of 1 mM Dex and 10 ng/ml HGF for
5 days, after which a significant reduction in amylase
cells were found to express G6Pase compared to Dex
treatment alone (Fig. 8A,B). In accordance with the
increase in the number of cells expressing G6Pase, the
number of cells expressing C/EBPb and nuclear HNF4a
also increased (Fig. 8A). Given that B13 cells cultured
with HGF alone do not express G6Pase, C/EBPb, or
either serves to enhance the hepatic phenotype induced
by Dex or is increasing the number of cells expressing
the hepatic phenotype. The glucocorticoid receptor
antagonist RU 486 prevented the conversion from pan-
creatic to hepatic phenotype since there was no expres-
sion of G6Pase, C/EBPb, or nuclear HNF4 in B13 cells
treated with the inhibitor (Fig. 8A).
Induction of hepatocyte proliferation
Hepatocytes rarely replicate under normal circum-
stances and the same is true of TD hepatocytes (Kurash
et al., 2004). However, we have now discovered a way to
induce cell division at least in a proportion of TD
hepatocytes. TD hepatocytes are able to enter active
phases of the cell cycle following culture with HGF or
Non-essential amino acids (NEAA). HGF is a potent
hepatic mitogen during liver regeneration (Masumoto
and Nakumura, 1996) and NEAA have been shown to
promote proliferation of hepatocytes in culture through
presence and absence of HGF and/or NEAA for 7 days
the mitotic marker phosphohistone H3 (Hendzel et al.,
1997). The three markers exhibit distinct staining
patterns during different phases of the cell cycle
(Hendzel et al., 1997; Prosperi et al., 1997). Under
control culture conditions, the staining pattern of the
stages of the cell cycle (i.e. G1, G2, and M phase) and
In the presence of Dex alone, 33% of cells express UGT
and 21% costain for UGT and cyclin D (Fig. 9, white
arrowheads) indicating that some of these cells are at
Dex for 10 days (C, D). Control B13 cells and transdifferentiated hepatocytes were then exposed to insulin
for 3 days (B, D). Lipid deposition was determined using oil red O staining. Positive lipid droplets are
indicated by red deposits.
Lipid deposition in transdifferentiated hepatocytes. B13 cells were treated without (A, B) or with
TRANSDIFFERENTIATED HEPATOCYTES FROM PANCREATIC CELLS
BURKE ET AL.
least capable of entering the G1phase of the cell cycle.
Cellular localization of cyclin D does not appear to be
influenced by the presence of HGF, whereas cells
expressing nuclear cyclin D were more evident in the
presence of NEAA. When cultured in the combined
presence of either Dex and NEAA or Dex and HGF, the
number of UGT positive cells increases to 48% and 55%,
respectively. The percentage of cells coexpressing UGT
and cyclin D in the presence of Dex/NEAA was 25% and
in the presence of both Dex and HGF, 40%. This
observation suggests that HGF may influence the G1
phase of the cell cycle. Interestingly, HGF has been
shown to upregulate the expression of cyclin D in
primary rat hepatocytes (Moriuchi et al., 2001) and
can also promote the transition of cells from G1to S
phase (Cho andKim, 2003). Inthe combined presence of
Dex, HGF, and NEAA 45% of the total number of cells
expressed UGT and 22% of cells exhibited weak
transdifferentiated hepatocytes. A: Peroxisomal proliferation in TD
hepatocytes is induced by ciprofibrate. Catalase staining in control
B13 cells (negative), transdifferentiated hepatocytes (induced by
treatment with 1 mM Dex for 21 days), 1 mM Dex for 14 days followed
by 7 days with the combined treatment of 1 mM Dex and 200 mM
ciprofibrate (DexþCipro) and 1 mM Dex for 14 days followed by 7 days
treatment with 200 mM ciprofibrate alone (Cipro). Catalase is localized
to the peroxisomes (as indicated by the punctate staining). Ciprofi-
brate enhances the levels of catalase even after removal of Dex.
B: Percentage of cells staining positive for catalase and P values from
Student’s t-test values are indicated as compared to Dex treatment
alone. C: UGT staining in TD cells induced with: 1 mM Dex for 14 days
(Dex), 1 mM Dex for 7 days followed by 7 days with 10 mM
Induction of catalase and UDP-glucuronosyltransferase in Phenobarbital alone (Pheno) and 1 mM Dex treated for 7days followed
by 7 days with Dex and 10 mM Phenobarbital (DexþPheno).
D: Phenobarbital enhances expression of Phase II metabolic marker
UGT in TD hepatocytes. Percentage of cells staining positive for UGT
and P values from Student0s t-test values are indicated as compared to
Dex treatment alone. Extended culture of 21 days for ciprofibrate
studies (compared to 14 days for phenobarbital studies) are required
due to the longer induction time of catalase. E: Western blotting
analysis of catalase protein levels in transdifferentiated hepatocytes
induced by treatment with 1 mM Dex for 19 days (Dex), 1 mM Dex for
12 days followed by 7 days with the combined treatment of 1 mM Dex,
and 200 mM ciprofibrate (DexþCipro) and 1 mM Dex for 12 days
followed by 7 days treatment with 200 mM ciprofibrate alone (Cipro).
EBPb, RXRa, and HNF4a expression was determined in B13 cells at 3, 5, 9, and 14 days after culture with
1 mM Dex. Control B13 cells are also shown. All transcription factors were induced following treatment
Time course of induction of transcription factors in transdifferentiated hepatocytes. C/EBPa, C/
TRANSDIFFERENTIATED HEPATOCYTES FROM PANCREATIC CELLS
presence of Dex alone or in combination with NEAA
combined presence of Dex and HGF or Dex, HGF and
the number of cells expressing the liver marker TFN
from 43% (Dex alone) to 61% and the percentage of cells
co-expressing TFN and cyclin B1 also increased (6.5%
compared with 47%), indicating that HGF may also
influence the entry of TD hepatocytes into the G2phase
of the cell cycle. Finally, cells strongly expressing the
mitotic marker phosphohistone H3 were occasionally
identified under all the treatments tested, but no
on the expression of the phosphohistone H3 were
observed by immunostaining alone (Fig. 9). Cell count-
ing revealed an increase in the number of cells co-
to 47% when cultured in the presence of Dex, HGF, and
NEAA, suggesting that proliferation of TD hepatocytes
through the entire cell cycle may be enhanced by
addition of all three factors.
Although there are examples of transdifferentiation
in the literature, such as the appearance of pancreatic
exocrine tissue in the human liver (Wolf et al., 1990) or
the conversion of adult exocrine cells to hepatocytes
(Lardon et al., 2004), the phenomenon remains con-
have utilized the B13 model of hepatic transdifferentia-
tion of the pancreas, to address: the phenotype of the
starting and finishing cells, the functional integrity and
proliferative capacity of the TD hepatocytes and the
potential to induce the hepatic phenotype in pancreatic
cells by ectopic expression of transcription factors.
The cellular origin of the AR42J cell line remains
unknown (Christophe, 1994). From the present investi-
gation we can conclude that the B13 subclone char-
acteristically expresses amylase, neurofilament, and
ferentiation of B13 cells to hepatocytes. A: B13 cells were transfected
with C/EBPa or C/EBPb expression constructs and cultured for 5 days.
Transfected cells wereimmunostained forthe presenceC/EBPa and C/
EBPb and the induction of UGT (white arrows) and HNF4a (white
Ectopic expression of C/EBPa and C/EBPb induces transdif-
arrowheads). Transfection of both C/EBPa and C/EBPb were inde-
pendently capable of mediating transdifferentiation and the induction
of UGT and HNF4a expression. B: Double immunostaining demon-
strating the coexpression of C/EBPb and UGT in cells transfected with
BURKE ET AL.
cultured for 5 days with either 1 mM dexamethasone, 1 mM dexamethasone and 100 pM HGF or 1 mM
dexamethasone and 2.5 mM RU486. Cells were stained with anti-G6Pase, anti-C/EBPb, and anti-HNF-4a
antibodies. B: Percentage of cells staining positive for G6Pase in Dex, Dex/HGF, and Dex/HGF/RU486
treated cells. Total number of cells counted is >500. Error bars are standard deviations.
HGF enhances the transdifferentiation of pancreatic B13 cells to hepatocytes. A: B13 cells were
TRANSDIFFERENTIATED HEPATOCYTES FROM PANCREATIC CELLS
synaptophysin (Fig. 1). AR42J cells were previously
considered to be exocrine in nature (due largely to their
ability to synthesize and secrete digestive enzymes) but
since they express neuroendocrine properties (neurofi-
lament and synaptophysin), and because of their ability
to convert to different pancreatic and hepatic cell types,
they should be considered as a pancreatic progenitor
cell line. AR42J cells are therefore similar to normal
exocrine cells, which can de-differentiate and act as
multipotent cells (Bouwens, 1998, 2004). Further stud-
ies are currently aimed at determining the expression
pattern of other pancreatic progenitor markers in the
B13 cell line before and after Dex-treatment.
Studying hepatocyte function in-vitro has been ham-
pered by the gradual loss of functional capacity of
primary cultures (Padgham et al., 1993). Transdiffer-
entiated hepatocytes from B13 cells may offer an
alternative to the use of hepatocytes isolated from
rodents. Based on immunlocalisation, RT-PCR, and
functional studies, we show that TD hepatocytes are
able to express: the plasma proteins albumin and
transferrin, the gluconeogenic enzyme G6Pase, a pro-
tein constituent of lipoproteins ApoB and enzymes
characteristic of hepatic biotransformation, CYP2E1,
and CYP3A1. TD hepatocytes not only express enzymes
for Phase I and II detoxification but also respond
appropriately to the presence of xenobiotics by upregu-
lating expression of UGT and catalase enzymes. These
properties of TD hepatocytes confirm their potential for
use indetermining the molecularbasis ofthedetoxifica-
The CCAAT/enhancing binding protein (C/EBP)
family of transcription factors have been characterized
as key regulators for the development and function of
the myeloid, adipogenic, and hepatic systems (Tanaka
et al., 1997). Our previous observations implicating the
transcription factor C/EBPb as a key player underlying
the molecular basis of transdifferentiation are sup-
ported in the present study by the rapid induction of C/
EBPb (based on RT-PCR analysis) following Dex
treatment suggesting it is one of the earliest markers
for transdifferentiation. While C/EBPb RNA was
detectable only hours after the start of Dex treatment,
C/EBPa, HNF4a, and RXRa appeared later suggesting
there is a coordinated change in the expression of liver
enriched transcription factors. It is possible that these
factors are themselves induced by C/EBPb to maintain
from the laboratory of Zaret and colleagues who have
shown that the C/EBP site on the albumin enhancer is
footprinted in foetal liver (Bossard et al., 1997). We also
found that ectopic expression in B13 cells of another
member of the C/EBP family C/EBPa, is able to induce
the typical expression pattern of UGT (nuclear envel-
ope/endoplasmic reticulum) and the nuclear transloca-
tion HNF4a. It is not possible to show an effect of C/
EBPa transfection on amylase expression because both
antibodies are raised in rabbit. Interestingly, liver
enriched transcription factors such as C/EBPb and
HNF4a are normally restricted to hepatocytes. How-
ever, both C/EBPb and HNF4a are expressed in
proliferating oval cells following liver damage and it is
believed that HNF4a may mediate regeneration of the
liver parenchyma by regulating the commitment of a
proportion of oval cells to become hepatocytes (Nagy
et al., 1994). The C/EBP and HNF4a transcription
factors are also prime candidates for reprogramming
embryonic and adult stem cells to a hepatic phenotype.
HGF can induce B13 cells to transdifferentiate into
insulin-producing cells, but it can also reduce the effect
of Dex on amylase secretion (Mashima et al., 1996b).
Here we demonstrate an additional role for HGF. We
found that HGF promotes the hepatic phenotype based
on increased G6Pase, C/EBPb and HNF4a induction
following combined Dex and HGF treatment. Further-
these cells towards active proliferation. Generally
considered quiescent cells, hepatocytes have a very long
life span (estimated to be around 200–400 days in vivo)
with low turnover rates (Magami et al., 2002). Our
previous studies have suggested that under normal
conditions TD hepatocytes do not proliferate (Kurash
et al., 2004). However, we have identified culture
taining to identify co-expression of UGT in combination with cyclin D
or phosphohistone H3 in untreated B13 cells, and B13 cells treated for
7 days with 1 mM Dex in the absence (Dex) or presence of NEAA
(DexþNEAA) or 7 days with 1 mM Dexþ10 ng/ml HGF in the absence
(DexþHGF) or presence of NEAA (Dex, HGFþNEAA). Perinuclear
Proliferation of transdifferentiated hepatocytes. Immunos-
and nuclear staining of cyclin D are indicated by white arrowheads
and white arrows respectively. Centromeric (blue arrows), granular
(blue arrowheads) and maximal mitotic staining (yellow arrow) for
phosphohistone H3 is indicated. These data demonstrate the potential
proliferative capacity of TD hepatocytes in the presence of Dex in
combination with NEAA and/or HGF.
BURKE ET AL.
conditions (HGF and NEAA in combination with Dex)
that are able to induce proliferation of some TD
hepatocytes. The ability to expand TD cell populations
will be useful for future studies on the analysis of the
hepatic phenotype. In summary, the B13 cell model
represents a research tool for determining the factors
that regulate liver gene expression and hepatic trans-
differentiation of the pancreas.
We gratefully acknowledge the generous gift of
antibodies from Professor Ann Burchell, University of
Dundee (Glucose-6-Phosphatase), Professor Roland
Wolf, University of Dundee (CYP3A1), Professor Chiao
Shih, University of Texas (C/EBPa) and Dr. Matthew
Wright, University of Aberdeen, (CYP2E1). We thank
critical reading of the manuscript.
Bossard P, McPherson CE, Zaret KS. 1997. In vivo footprinting with limiting
amounts of embryo tissues: A role for C/EBP beta in early hepatic development.
Bouwens L. 1998. Transdifferentiation versus stem cell hypothesis for the
regeneration of islet beta-cells in the pancreas. Microsc Res Tech 15:332–336.
Bouwens L. 2004. Islet morphogenesis and stem cell markers. Cell Biochem
Biophys 40(Suppl 3):81–88.
ing protein and cell replication via PI3-kinase pathway. Hepatol 37:686–695.
Christophe J. 1994. Pancreatic tumoral cell-line AR42J—an amphicrine model.
Am J Physiol 266:G963–G971.
Costa RH, Kalinichenko VV, Holterman AX, Wang X. 2003. Transcription factors in
liver development, differentiation, and regeneration. Hepatology 38:1331–1347.
Coughtrie MWH, Burchell B, Leaky JEA, Hume R. 1988. The inadequacy of
perinatal glucuronidation: Immunoblot analysis of the developmental expres-
sion of individual UDP-glucuronosyltransferase isozymes in rat and human
liver microsomes. Mol Pharmacol 34:729–735.
Diehl AM, Michaelson P, Yang SQ. 1994. Selective induction of CAATT enhancer
binding isoforms occurs during rat liver development. Gastroenterology
Greenbaum LE, Li W, Cressman DE, Peng Y, Ciliberto G, Poli V, Taub R. 1998.
CAATT enhancer-binding protein beta is required for normal hepatocyte
proliferation in mice after partial hepatectomy. J Clin Invest 102:996–1007.
Hendzel MJ, Wei Y, Mancini MA, Van Hooser A, Ranalli T, Brinkley BR, Bazett-
Jones DP, Allis CD. 1997. Mitosis-specific phosphorylation of histone H3
initiates primarily within pericentromeric heterochromatin during G2 and
spreads in an ordered fashion coincident with mitotic chromosome condensa-
tion. Chromosoma 106:348–360.
Horb ME, Shen C-N, Tosh D, Slack JMW. 2003. Experimental conversion of liver
to pancreas. Curr Biol 13:105–115.
Hylemon PB, Pandak WM, Vlahcevic ZR. 2001. Regulation of hepatic cholesterol
homeostasis. In: Arias IM, Boyer JL, Chisari FV, Fausto N, Schachter D,
Shafritz DA, editors. The liver: Biology and pathobiology. Philadelphia:
Lippincott, Williams and Wilkins.
Kimura T, Christoffels VM, Chowdhury S, Iwase K, Matsuzaki H, Mori M,
Lamers WH, Darlington GJ, Takiguchi M. 1998. Hypoglycaemia-associated
hyperammonemia caused by impaired expression of ornithine cycle enzymes
genes in C/EBPa knockout mice. J Biol Chem 273:27505–27510.
Kurash JK, Shen C-N, Tosh D. 2004. Induction and regulation of acute phase
proteins in transdifferentiated hepatocytes. Exp Cell Res 292:342–358.
Lardon J, Dr Breuck S, Rooman I, Van Lommel L, Kruhoffer M, Orntoft T, Schuit
F, Bouwens L. 2004. Plasticity in the adult rat pancreas: Transdifferentiation
of exocrine to hepatocyte-like cells in primary culture. Hepatology 2004;39:
Logsdon CD, Moessner J, William JA, Goldfine ID. 1985. Glucocorticoids increase
amylase mRNA levels, secretory organelles, and secretion in pancreatic acinar
AR42J cells. J Cell Biol 100:1200–1208.
LucPV,Adesnik M,Ganguly S, ShawPM. 1996.Transcriptionalregulation of the
CYP2B1 and CYP2B2 genes by C/EBP-related proteins. Biochem Pharmacol
Magami Y, Azuma T, Inokuchi H, Kokuno S, Moriyasu F, Kawai K, Hattori T.
2002.Cell proliferationand renewalof normal hepatocytes andbile duct cellsin
adult mouse liver. Liver 22:419–425.
Marek CJ, Cameron GA, Elrick LJ, Hawksworth GM, Wright MC. 2003.
Generation of hepatocytes expressing functional cytochromes P450 from a
pancreatic progenitor line in vitro. Biochem J 370:763–769.
Mashima H, Ohnishi H, Wakabayashi K, Mine T, Miyagawa J, Hanafusa T, Seno
M, YamadaH, Kojima I. 1996a. Betacellulinand Activin A coordinately convert
amylase-secreting pancreatic AR42J cells into insulin-secreting cells. J Clin
Mashima H, Shibata H, Mine T, Kojima I. 1996b. Formation of insulin-producing
cells from pancreatic acinar AR42J cells by hepatocyte growth factor.
Masumoto K, Nakumura T. 1996. Emerging multipotent aspects of hepatocyte
growth factor. J Biochem (Tokyo) 119(4):591–600.
Mithieux G. 1997. New knowledge regarding glucose-6 phosphatase gene and
protein and their roles in the regulation of glucose metabolism. Eur J
Morgan EH, Peters T Jr. 1971. The biosynthesis of rat serum albumin. J Biol
Moriuchi A, Hirono S, Ido A, Ochiai T, Nakama T, Uto H, Hori T, Hayashi K,
Tsubouchi H. 2001. Additive and inhibitory effects of simultaneous treatment
with growth factors on DNA synthesis through MAPK pathway and G1 cyclins
in rat hepatocytes. Biochem Biophys Res Commun 280:368–373.
Nagy P, Bisgaard HC, Thorgeirsson SS. 1994. Expression of hepatic transcription
factors during liver development and oval cell regeneration. J Cell Biol 126:
Nelsen CJ, Rickheim DG, Tucker MM, McKenzie TJ, Hansen LK, Pestell RG,
Albrecht JH. 2003. Amino acids regulate hepatocyte proliferation through
modulation of cyclin D1 expression. J Biol Chem 278:25853–25858.
NovikoffAB, NovikoffPM,Rosen OM, Rubin CS. 1980.Organelle relationshipsin
cultured 3T3-L1 preadipocytes. J Cell Biol 87:180–196.
Padgham CR, Boyle CR, Wang X-U CR, Raleigh SM, Wright MC, Paine AJ. 1993.
Alteration of transcription factor mRNA during isolation and culture of rat
hepatocytes suggests the activation of a proliferative mode underlies the
dedifferentiation. Biochem Biophys Res Commun 197:599–605.
Parviz F, Matullo C, Garrison WD, Adamson JW, Ning G, Kaestner KH, Rossi
JM, Zaret KS, Duncan SA. 2003. Hepatocyte nuclear factor 4 alpha controls the
development of a hepatic epithelium and liver morphogenesis. Nat Genet 34:
Prosperi E, Stivala LA, Scovassi AI, Bianchi L. 1997. Cyclins: Relevance of
subcellular localisation in cell cycle control. Eur J Histochem 41:161–168.
Rao MS, Reddy JK. 1995. Hepatic transdifferentiation in the pancreas. Semin
Cell Biol 6:151–156.
Rao MS, Subbarao V, Reddy JK. 1986. Induction of hepatocytes in the pancreas of
copper-depleted rats following copper repletion. Cell Differentiation 18:109–
Ritter JK, Kessler FK, Thompson MT, Grove AD, Auyeung DJ, Fisher RA. 1999.
Expression and inducibility of the human bilirubin UDP-glucuronosyltransfer-
ase UGT1A1 in liver and cultured primary hepatocytes: Evidence for both
genetic and environmental influences. Hepatology 30:476–484.
Saltiel AR, Kahn CR. 2001. Insulin signalling and the regulation of glucose and
lipid metabolism. Nature 414:799–806.
Schmidt C, Bladt F, Goedecke S. 1995. Scatter factor/hepatocyte growth factor is
essential for liver development. Nature 373:699–702.
Shen C-N, Slack JMW, Tosh D. 2000. Molecular basis of transdifferentiation of
pancreas to liver. Nat Cell Biol 2:879–887.
Shen C-N, Seckl JR, Slack JMW, Tosh D. 2003. Glucocorticoids suppress beta-cell
development and induce hepatic metaplasia in embryonic pancreas. Biochem J
Shen C-N, Burke ZD, Tosh D. 2004. Transdifferentiation, metaplasia, and tissue
regeneration. Organogenesis 1:36–44.
Tanaka T, Yoshida N, Kishimoto T, Akira S. 1997. Defective adipocyte
differentiation in mice lacking the C/EBPb and/or the C/EBPd gene. EMBO J
Tephly TR, Burchell B. 1990. UDP-glucuronosyltransferases—a family of
detoxifying enzymes. Trends Pharmacol Sci 11 :276–279.
Tosh D, Slack JMW. 2002. How cells change their phenotype. Nat Rev Mol Cell
Tosh D, Shen C-N, Slack JMW. 2002. Differentiated properties of hepatocytes
induced from pancreatic cells. Hepatology 36:534–543.
Wan YJ, An D, Cai Y, Repa JJ, Hung-Po Chen T, Flores M, Postic C, Magnuson
MA, Chen J, Chien KR, French S, Mangelsdorf DJ, Sucov HM. 2000.
Hepatocyte-specific mutation establishes retinoid X receptor alpha as a
heterodimeric intergrator of multiple physiological processes in the liver. Mol
Cell Biol 20:4436–4444.
Watt AJ, Garrison WD, Duncan SA. 2003. HNF4: A central regulator of
hepatocyte differentiation and function. Hepatology 37:1249–1253.
Wells WA. 2002. Is transdifferentiation in trouble? J Cell Biol 157:15–18.
Wolf HK, Burchette JL, Garcia JA, Michalopoulos G. 1990. Exocrine pancreatic
tissue in human liver: A metaplastic process? Amer J Surg 14:590–595.
Zhou J, Wang X, Pineyro MA, Egan JM. 1999. Glucagon-like peptide 1 and
exendin-4 convert pancreatic AR42J cells into glucagon- and insulin-producing
cells. Diabetes 48:2358–2366.
TRANSDIFFERENTIATED HEPATOCYTES FROM PANCREATIC CELLS