Download Hepatocyte differentiation initiates during

Development 113, 217-225 (1991)
Printed in Great Britain © The Company of Biologists Limited 1991
217
Hepatocyte differentiation initiates during endodermai-mesenchymal
interactions prior to liver formation
STEPHANIE CASCIO and KENNETH S. ZARET*
Section of Biochemistry, Brown University, Box G, Providence, Rl 02912, USA
* Corresponding author
Summary
Previous studies with embryonic tissue explants showed
that cellular interactions with mesenchyme are required
for endodermal cells to differentiate into hepatocytes.
However, these studies assayed hepatocyte characteristics that were evident after days of culture, leaving
open the question of whether the primary inductive
interactions initiated hepatocyte differentiation, or
whether subsequent steps, such as may occur during cell
aggregation to form the liver, were necessary. Using the
technique of in situ hybridization, we find that serum
albumin mRNA, a liver-specific gene product, is first
detected in hepatic precursor cells of the endoderm as
early as 9.5 days of mouse embryo development, a full
day prior to cell aggregation and liver formation. The
endodermal cells express albumin mRNA upon migration into strands of connective tissue matrix within
mesenchyme. Thus, the onset of differentiation of the
endoderm is coincident with its interaction with mesenchyme. Early albumin transcripts are initiated at the
same site of the albumin promoter as in adult
hepatocytes, suggesting that at least a subset of the
transcription factors that control albumin transcription
in the adult may be involved in executing the early steps
of hepatic determination. We also observe a sharp
increase in albumin mRNA levels shortly after the
definitive formation of the liver, apparently reflecting
cell interactions that enhance hepatocyte differentiation.
Hepatocyte differentiation is therefore similar in several
respects to pancreatic exocrine cell development, and
may represent a general pattern for gut-derived tissues.
For both cell types, early interactions with mesenchyme
are coincident with the initial expression of differentiated gene products at a low level in proliferating
endoderm, and the initial pattern of expression is
amplified upon organ formation.
Introduction
lead to differentiation. In this paper, we define the
developmental time and place of transcriptional activation of the liver-specific serum albumin gene. Our
findings characterize the cell environment that influences the initial stages of hepatocyte differentiation.
In mammals, the liver is formed from the developing
foregut. Between 8.5 and 9.5 d of mouse gestation, a
region of the foregut epithelium thickens to form the
liver diverticulum. Over the next two days, endodermal
cells from the liver diverticulum proliferate and migrate
into the surrounding septum transversum, a region of
loose mesenchyme. By day 10.5, the liver bud is
recognizable microscopically (Theiler, 1989). By 12.5 d
of gestation, the liver is a relatively large, differentiated
organ composed of various cell types, of which up to
60% are hematopoietic cells (Silini et al. 1967; Paul
et al. 1969). Hematopoietic cells migrate to the embryonic liver from the yolk sac and are not derived from the
hepatic cell lineage (Johnson and Moore, 1975).
Elegant tissue culture and transplantation experiments have shown that endodermal cells require two
The development of the vertebrate liver is controlled by
a specific sequence of cellular interactions that have
been defined experimentally (LeDouarin, 1975), and
that serve as a paradigm for understanding fundamental
mechanisms of organogenesis. Our goal is to understand how cellular interactions in liver development
lead to the selective transcription of genes that are
characteristic of hepatocytes, thereby controlling the
differentiated state of the cell. Studies of cell-specific
gene transcription have provided a detailed account of
many cis- and trans-acting regulatory factors required
for gene expression in hepatocytes (Johnson, 1990) and
other adult vertebrate cell types (Maniatis et al. 1987;
Mitchell and Tjian, 1989). Little is known, however,
about how specific genes are transcriptionally activated
in a subset of cell types, or more generally, how distinct
mammalian cell types are specified in development. To
address these problems, it is important to know when a
cell begins to differentiate and what cell interactions
Key words: endoderm differentiation, hepatocyte, albumin,
transcription, liver, mouse.
218
S. Cascio and K. S. Zaret
inductive interactions to form cells that express
hepatocyte morphology (LeDouarin, 1968, 1975). In
avian embryos, prehepatic endoderm is induced first by
cardiac mesoderm, giving rise to proliferating endodermal cells. Next, these endodermal cells interact with
mesenchyme of the septum transversum and subsequently differentiate into hepatocytes (LeDouarin,
1968). Houssaint (1980) has observed similar inductive
requirements for mammalian liver development.
In each of these explant and transplantation studies,
hepatocyte differentiation was denned by the acquisition of a particular cell morphology or by the
accumulation of glycogen granules; these phenotypes
become apparent after several days of culture in vitro or
in a transplant. By this time, the hepatic precursor cells
aggregate in vitro or have assembled to form the liver
organ in vivo. Thus, it was not clear whether the initial
expression of differentiated hepatic functions occurs
during the inductive endodermal-mesenchymal interactions, or only after subsequent cellular interactions
that take place upon the formation of the liver
structure.
In this study, we performed an in situ analysis of
mRNA expression to attempt to identify earlier stages
of hepatocyte differentiation, and to reveal the cellular
environment necessary to activate liver-specific gene
transcription. Hepatocyte-specific mRNAs have been
detected in the livers of 13 d embryos, including those
for albumin, alpha-fetoprotein (AFP), urea cycle
enzymes (Morris etal. 1989), transthyretin, the transcription factor C/EBP (Kuo etal. 1990), and apolipoprotein-Al (Meehan etal. 1984), but the expression of
liver-specific mRNAs has not been reported at earlier
developmental stages. In this paper, we focus on the
serum albumin gene, because previous studies had
shown that by day 15 of mouse development the
albumin and AFP genes are transcriptionally very
active (Tilghman and Belayew, 1982; Panduro et al.
1987). We anticipated that the high level of albumin
gene transcription in hepatocytes, together with its lack
of transcription in other cell types (Powell et al. 1984;
Liu et al. 1988) would facilitate the detection of early
albumin transcripts during liver development.
20 min reflxation in 4% paraformaldehyde in PBS at 23°C.
After digestion with 1/igml"1 proteinase K for 30 min at
37°C, sections were treated with 0.25% acetic anhydride for
10 min at 23°C. A 0.6 kb Hindlll mouse albumin cDNA
sequence, from the coding region of the message (Kioussis
etal. 1981), was cloned into the pGEM-3 vector (Promega
Biotec) and used to generate RNA probes containing 3%UTP, at a specific activity of 10 7 -10 8 ctsmin~'/ig"'. Each
slide was hybridized with lC^ctsmin"1 of probe for 12-16 h at
50°C. The slides were washed as described by Wilkinson et al.
(1987), including the two high-stringency washes at 65°C in a
solution of 50 % formamide, 2xSSC, and 1 mM dithiothreitol.
The slides were dried, dipped in Kodak NTB-2 photographic
emulsion, and exposed for 6 days to 6 months at 4°C. The
slides were developed, stained with hematoxylin and eosin,
and viewed with bright-field and dark-field optics under a
Nikon microscope.
RNA isolation from embryos
Embryos were dissected from extra-embryonic membranes
and rinsed in phosphate-buffered saline. Tissue samples, as
whole embryos (days 8.5-9.5), midsections (9.5-10.5d), or
livers (12.5-13.5 d) were isolated and placed into 500 (i\ of a
solution containing 5 M guanidinium thiocyanate, 50 mM
sodium citrate, pH7, and 0.5% sodium /V-lauryl sarcosinate
(Chirgwin et al. 1979). Tissues were disrupted by pipetting the
solution gently for 1-3 min. The volumes were adjusted to
840 /A with more guanidinium solution and layered onto 560 /A
of a solution containing cesium chloride (lgmF 1 ) in 25 mM
sodium citrate, pH5. The samples were spun at 55 000 revs
min"1 for 3h in a TLS55 rotor in a TL-100 microultracentrifuge (Beckman Instruments). The RNA pellets
were suspended in water and precipitated with sodium acetate
and ethanol; the RNA pellets were washed with 70 % ethanol,
resuspended in water, and quantitated by UV absorbance.
For comparative purposes, 10 ^g of total RNA was obtained
from about 2.5 whole embryos at 9.0d, 1.4 midsections of
embryos at 10.5 d, 0.72 livers from embryos at 12.5 d, and 0.19
livers of embryos at 13.5 d.
Primer extension analysis of RNA
RNAs were hybridized to a 32P-end-labeled oligonucleotide
and analyzed by primer extension with reverse transcriptase
(McKnight and Kingsbury, 1982), as described by Liu etal.
(1988). Aliquots (0.8 fig) of the same RNAs were electrophoresed through 1 % agarose minigels, stained with ethidium
bromide, and photographed under ultraviolet light.
Collagen counterstaining of mesenchyme
Materials and methods
In situ hybridization of mouse embryos
All experiments used C3HeB/FeJ mice obtained from
Jackson Laboratory, Bar Harbor, ME. We count noon of the
day on which the vaginal plug is found as 0.5 d gestation.
Embryonic structures and stages were confirmed by comparison with mouse embryo anatomies described by Rugh (1968)
and Theiler (1989).
In situ hybridization was performed essentially as described
by Hogan etal. (1986) and Wilkinson etal. (1987). Carefully
staged mid-gestation mouse embryos were dissected free of
extra-embryonic membranes (10.5-15.5 d) or left within
decidua (8.5-9.5 d) and fixed in 4% paraformaldehyde in
PBS for 30-60 min on ice, dehydrated, and embedded in
paraffin. 5fim sections were collected on organosilanated
slides, deparaffinized, and prepared for hybridization by
Paraffin sections were deparaffinized, hydrated in distilled
water, stained in Van Gieson's solution (Mallory, 1961)
according to the manufacturer (Poly Scientific, NY), and
viewed with bright-field optics. The staining characteristics of
various tissue components are described in the legend to
Fig. 4.
Results
We used in situ hybridization to determine when serum
albumin mRNA is first expressed in the mouse embryo.
As seen in Figs 1 and 2, antisense RNA probes from an
albumin cDNA hybridized specifically to the 12.5 d
(Fig. IB, Fig. 2D) and 15.5 d (Fig. 2F) fetal mouse
livers, while albumin sense strand probes did not
hybridize to any tissue (Fig. ID). These findings are in
Hepatic differentiation in mouse embryos
219
Fig. 1. Antisense albumin RNA probes hybridize specifically to mouse embryonic liver tissue. Bright-field (A,C,E) and
dark-field (B,D,F) views at 40x original magnification of 12.5d embryo sagittal sections probed with an albumin antisense
strand probe (A,B), an albumin sense strand probe (C,D), and a /3-tubulin antisense strand probe (E,F). Panel D was
overexposed to show the positions of tissues. Abbreviations: L, liver; H, heart; He, head; I, intestine; Lu, lung; S,
stomach; Sp, spleen; Ve, vertebrae.
agreement with previous studies of mRNA isolated
from embryonic liver samples (Morris etal. 1989). fitubulin mRNA was available for hybridization in all
tissues (Fig. 1E,F), demonstrating that under our
conditions all tissues retained hybridizable RNA.
Albumin mRNA distribution in embryonic livers
Hybridization to the 12.5 and 15.5 d livers was patchy,
reflecting the very large proportion of hematopoietic
cells in the liver at this time. By contrast, almost all cells
within the liver bud of 10.5 d embryos hybridized
220
S. Cascio and K. S. Zaret
v-
^
• • » * •
•
•
;
'
•
•
*
•
-
• •
•
•
.
.
v---.-^.
"
sk
te
-
• • : » !
Fig. 2. Increasing albumin gene expression during liver organogenesis. All sections were probed with an albumin antisense
strand probe, and are oriented here in the reverse fashion from the sections in Fig. 1. Bright-field (A,C,E) and dark-field
(B,D,F) views at lOOx original magnification of embryos at 10.5 d gestation (A,B; six week exposure), 12.5 d gestation
(C,D, six day exposure), and 15.5d gestation (E,F, six day exposure). Abbreviations: L, liver; H, heart.
uniformly to the albumin antisense probe (Fig. 2A,B).
At this stage, the differentiating hepatocytes stained
more darkly with eosin than the cells of the surrounding
connective tissue (Fig. 2A). No hematopoietic cells
were evident in the 10.5 d liver bud; the differentiating
hepatocytes at this stage appeared homogeneous in
both size and staining properties (as seen in higher
magnification views not shown).
Albumin mRNA in migrating hepatic endoderm cells
In 9.5 d (20-26 somite) and earlier embryos, we found it
necessary to probe serial sections over a 200-300 ^m
sectioning distance to identify unambiguously hepatic
precursor cells. Both sagittal and transverse sections
were studied. We examined five 8.5d embryos (5-8
somites) and seven 9.0 d embryos (12-20 somites)
thoroughly by serial section, but were unable to detect
Hepatic differentiation in mouse embryos
221
Fig. 3. Albumin gene transcription in hepatocyte precursors of 9.5 d mouse embryos. All sections were probed with an
albumin antisense RNA probe. Three different embryos are shown under bright-field (A,D,G) and dark-field (B.E.H)
optics at lOOx original magnification, and under bright-field at 400x original magnification (C,F,I). The regions shown at
400x are outlined in the lOOx panels. Panels A, B, and C show a transverse section, 6 month exposure; D, E, and F are a
transverse section, 6 week exposure; G, H, and I are an oblique section, 9 week exposure. Arrowheads point to
hybridization grains indicating the presence of albumin transcripts. Individual hybridization grains are evident in the 400X
panels. Abbreviations: E, hepatic endodermal cells migrating from the foregut; F, position of the foregut at the liver
diverticulum; S, mesenchyme of the septum transversum; A, dorsal aorta; V, posterior cardinal vein; P, pleural cavity; R,
nucleated erythrocytes in vessels; N, neural tube.
specific hybridization to the albumin probe (data not
shown). The earliest point in mouse development at
which we could detect albumin mRNA by in situ
hybridization was 9.5d of gestation (Fig. 3B,E,H).
Exposure times of 6 weeks to 6 months revealed
albumin mRNA in clusters of cells arising from the
foregut epithelium and in cords of cells beginning to
invade the septum transversum (Fig. 3A,D,G). Hema-
222
S. Cascio and K. S. Zaret
topoietic cells appeared refractile but did not contain
the hybridization grains that were evident over albumin-positive cells. High-magnification studies of the
embryo sections demonstrated that only the proliferating hepatic precursors expressed albumin mRNA
(Fig. 3C,F,I).
The albumin mRNA-positive cells at day 9.5 were in
tight clusters as they began to invade the sparsely
distributed cells of the surrounding mesenchyme of the
septum transversum. Because tissue explant studies
showed that the migrating endoderm must interact with
mesenchyme for the former to differentiate into
hepatocytes (LeDouarin, 1968; Houssaint, 1980), we
investigated whether at 9.5 d the albumin-positive
endodermal cells were interspersed with mesenchyme.
An alternative possibility was that the endodermal cells
were pushing back a layer of mesenchyme; this occurs
initially as the endoderm thickens and hepatic precursors first begin to proliferate (Rugh, 1968; Theiler,
1989).
The septum transversum mesenchyme consists of
loose connective tissue containing collagen; we therefore used Van Gieson's method to stain collagen in
tissue sections from 9.5d embryos (Mallory, 1961).
.Under high magnification, it was clear that the hepatic
precursors were loosely organized and contacted
collagen fibrils and lighter-staining cells of the mesenchyme of the septum transversum (see arrowheads,
Fig. 4). Although these studies could not distinguish
whether the collagen was produced by the mesenchyme
or by the endodermal cells, or both, the abundance of
the fibrils is similar in both the loose mesenchyme of the
septum transversum and within the region invaded by
the endoderm. In sum, it is clear that the initial
expression of albumin mRNA occurs during the
invasion of the septum transversum, when the hepatic
precursor cells clearly contact cells and perhaps
collagen of the mesenchyme.
Dramatic increase in albumin mRNA upon liver
formation
We analyzed albumin mRNA in mid-gestation embryos
using the primer extension method (Fig. 5), to quantitate albumin mRNA and to determine the start site of
transcription over development. RNA was isolated
from whole embryos at day 9 (free of extra-embryonic
membranes such as yolk sac), from midsections of
embryos at 10.5 d, from fetal livers at day 12.5 and 13.5,
and from livers of adult mice. Primer extension was
performed with an end-labeled oligonucleotide complementary to an internal sequence of the albumin
mRNA, which generates an extension product of 110
nucleotides from the adult message (Liu etal. 1988).
Albumin mRNA was not detectable by primer extension in whole embryos at day 9.0 (Fig. 5) or in 9.5 d
embryo midsections (data not shown). However, 20,ug
of RNA from 10.5 d embryo midsections contained a
level of albumin mRNA that was approximately
equivalent to that in 10 ng of adult liver RNA, and 5 f.ig
of 12.5 d liver RNA contained an albumin mRNA level
approximately equivalent to that in 300 ng adult liver
A
Embryo (d)
9.0 10.5 12.513.5 Liver
Adult Liver
150 45
1 2 3
B
15 4.5
a
a
3a
4
5
6
7
b
8
b
9
45 150
ng RNA
10 11
5 6 8 910
Fig. 5. Albumin mRNA is initiated at the adult
transcription start site in early mouse embryos.
(A) Albumin mRNAs from embryos at 9.0 to 13.5d
gestation were analyzed by primer extension and
quantitated by comparison to a dilution series of adult liver
RNA. RNA was extracted from whole embryos at 9.0d
(lane 5), midsections of embryos at 10.5 d (lanes 6 and 7),
and livers of embryos at 12.5 and 13.5 d (lanes 8 and 9,
respectively). The hybridization reactions included 20 [ig of
9.0 and 10.5d embryo RNA, or 5^g of 12.5 and 13.5 d
embryo RNA, except for 10.5-3a, which contained 60/ig of
10.5 d midsection RNA. The different amounts of liver
RNAs (shown in nanograms, lanes 1-4 and 10, 11) were
mixed with 5 f.ig of E. coli tRNA prior to hybridization.
The primer-extension products were analyzed on 6 %
polyacrylamide gels and exposed to X-ray film.
(B) Samples of the same RNAs used in the lanes shown in
panel A were analyzed by electrophoresis on an agarose
gel, stained with ethidium bromide and photographed
under UV light to demonstrate the integrity of the RNA;
the ribosomal RNA bands are visible.
RNA (Fig. 5 and Table 1). When caluclated on a per
cell basis, there is an approximately 15- to 20-fold
increase in albumin mRNA upon formation of the liver
organ by day 12.5 (Table 1). Given that the level of
albumin mRNA in 10.5 d embryos was only several fold
higher than our lower limit of detection (Fig. 5), and
Fig. 4. Hepatic precursors at 9.5 d are in direct contact with connective tissue matrix and cells of the mesenchyme.
Transverse sections of 9.5 d embryos were stained by Van Geison's method, which renders collagen fibers red, nuclei
brown, and cytoplasm golden yellow. Arrowheads indicate stained collagen fibers in contact with hepatic precursors
migrating from the liver diverticulum of the foregut. E, hepatic endodermal cells; F, position of the foregut at the liver
diverticulum; S, mesenchyme of the septum transversum.
Hepatic differentiation in mouse embryos
Table 1. Albumin mRNA levels in mouse liver
development
Embryonic stage
Percent of hepatocyte mass
per total tissue mass
Albumin mRNA molecules
per hepatocytet
10.5 d
12.5 d
13.5 d
Adult
3-6%*
40 %t
40 %f
90%
200
3000
6500
20000
* As estimated from our embryo midsections and by Theiler,
1989.
t As determined by Paul et al. 1969, accounting for the large
proportion of hematopoietic cells in the embryo liver.
$ Values for embryonic stages are based upon the data in Fig. 5
and are relative to calculated values of albumin mRNA molecules
(2200 bp) per liver cell for both adult mouse (Tilghman and
Beleyew, 1982) and rat (Nahon et al. 1982), which were shown to
contain 0.8/(g albumin mRNA per mg total liver tissue RNA.
the long exposures necessary to visualize hybridization
to albumin transcripts in the 9.5 d embryo sections, it
was not surprising that we could not detect primer
extension signals in the 9.5 d RNA sample. In all cases,
albumin mRNA from embryonic tissues had the same
transcription start site as albumin mRNA isolated from
adult hepatocytes (Fig. 5). We conclude that the
albumin promoter initiates transcription at the same
site from the onset of differentiation through adult life.
Discussion
Our findings show that hepatic endoderm begins to
differentiate at the time of interaction with mesenchyme, as evidenced by the transcription of a hepatocyte gene product, serum albumin. The fact that the
albumin gene is transcribed several hundred to a
thousand fold more frequently in adult hepatocytes
than in other cell types (Liu et al. 1988), and that no in
situ hybridization signal was detected in other cell types
of the embryo (Figs 1 and 2), indicates the extreme
specificity with which this gene serves as a marker for
hepatocytes. Thus, hepatocyte differentiation begins
prior to or during the induction of endoderm at day 9.5
of mouse embryogenesis; this stage is significantly
earlier than had been identified by other studies of liver
development, and occurs prior to cell assembly into a
liver bud (see the Introduction). At this stage of our
analysis, we cannot distinguish whether interaction with
cardiac mesoderm or with mesenchyme of the septum
transversum is the primary inducer of hepatic endoderm differentiation. However, as discussed further
below, our findings demonstrate that the general
sequence of events for hepatic differentiation is similar
to that for pancreatic exocrine cell differentiation
(Rutter et al. 1978), which begins very shortly after an
inductive interaction between endoderm and mesenchyme.
It is interesting to contrast our findings with what is
known about the maintenance of adult hepatocyte
differentiation. In adult hepatocytes, the transcription
223
rates of many liver-specific genes, including albumin,
are exquisitely sensitive to tissue organization; liverspecific gene transcription is selectively lost shortly after
hepatocytes are isolated and cultured in vitro (Clayton
et al. 1985; Ben Ze'ev et al. 1988), whereas liver tissue
slices in vitro can maintain their transcription rates for
some time (Clayton et al. 1985). A moderate level of
albumin transcription can be retained when isolated
hepatocytes are cultured using a hormonally defined,
serum-free medium (Jefferson et al. 1985) or an
extracellular matrix substratum (Caron, 1990), indicating that liver gene transcription is dependent upon
external signals such as hormones, extracellular matrix,
and appropriate cell-cell contacts (Fraslin et al. 1985;
Ben-Ze'ev et al. 1988). Even though albumin transcription is activated in endodermal cells that clearly lack the
precise cell environment of the adult liver, we suggest
that some of the aforementioned modulatory signals
used in the adult may play roles in hepatic endoderm
differentiation in the embryo. For example, a liverspecific enhancer sequence (Pinkert et al. 1987)
upstream of the albumin gene has recently been shown
to be activated by a collagen extracellular matrix in a
transfected adult hepatocyte cell line (Liu etal. 1991;
DiPersio etal. 1991). Conceivably, in embryogenesis
the collagenous environment of the mesenchyme of the
septum transversum, or such components of cardiac
mesoderm, could play a direct role in transcriptionally
activating the albumin gene.
In many cases, such as in muscle, differentiated gene
expression occurs when the cells stop dividing. We
showed that the transcription of the albumin gene
begins as endodermal cells enter a stage of rapid
proliferation. Similarly, the initial expression of pancreatic exocrine cell enzymes parallels the onset of
endodermal cell precursor proliferation from the
embryonic gut (Rutter etal. 1968). Interestingly, the
mouse dwarf mutation disrupts the gene encoding the
transcription factor Pit-1 (GHF1) (Li et al. 1990), which
normally activates the growth hormone gene in
pituitary somatotrophic cells (Dolle etal. 1990). In
addition to causing the loss of expression of growth
hormone and other products of the anterior pituitary
gland, the dwarf mutation also causes hypoplasia of the
somatotrophic cells (Smith and MacDwell, 1931). Thus,
there could be a general link between the onset of
proliferation and the initial phase of differentiation for
a number of cell types.
Given that the first detectable embryonic albumin
transcripts are initiated by the same promoter as are
adult transcripts, it seems likely that at least a subset of
the transcription factors that maintain the activity of the
albumin promoter in the adult (Cereghini etal. 1987;
Lichtsteiner etal. 1987; Babiss etal. (1987); review by
Johnson, 1990) are switched on at the initial time of
hepatic specification in the early embryo. Future
experiments to define such embryonic regulatory
factors must therefore investigate the hepatic endodermal cells prior to and during their proliferative stage.
The long exposure times needed for detecting
albumin mRNA in 9.5 d embryos (Fig. 3) suggest that
224
S. Cascio and K. S. Zaret
hepatocyte precursors at this point have accumulated
little albumin message. Indeed, an earlier study of
albumin protein expression in mouse development,
using indirect immunofluorescence, reported that albumin antigen was first detected at the liver bud stage
between days 10.5 and 11.5 (Shiojiri, 1984). Conceivably, albumin gene transcription could commence upon
interaction of the endoderm with cardiac mesoderm,
but transcripts are not detectable until the endoderm
migrates into the mesenchyme. We observed a dramatic
increase in albumin mRNA accumulation between the
time the cells first assembled to form the liver bud
(10.5 d) and the time the liver was clearly evident
(12.5 d). We estimate that about 3-6% of our 10.5 d
embryo midsections were hepatic cells, while about
40% of the 12.5 d liver is composed of hepatocytes
(Paul etal. 1969). Considering these points, and the
relative amounts of embryo RNA assayed in each
primer extension reaction (and normalized to ribosomal
RNAs; Fig. 5), we find an approximately 15- to 20-fold
increase in albumin mRNA between days 10.5 and 12.5
of embryogenesis (see Table 1). This level of induction
agrees qualitatively with what we saw using in situ
hybridization of identically processed sections of 10.5
and 12.5 d embryos (e.g. Fig. 2; see exposure times in
Figure legend). It seems likely that some of the cell-cell
and cell-matrix interactions that play a role in
maintaining hepatocyte differentiation in the adult
(discussed above) may be used to increase the initial
level of liver gene expression during organ formation in
the embryo.
As seen here for a liver-specific gene, cell assembly
into an organ serves to amplify a transcriptional state
established prior to organ formation. Similarly, as
pancreatic endoderm precursors proliferate and invade
the mesenchyme, they initially express low levels of
exocrine cell enzymes. After forming tissue structures
characteristic of the pancreas, the cells begin to produce
much higher amounts of pancreas-specific proteins
(Rutter etal. 1978). Thus, for both hepatic and
pancreatic induction, there is a two-stage increase in the
production of cell-specific products; this scheme could
be general for other cell types derived from gut
endoderm. We suggest that the initial induction of such
endodermal derivatives by mesenchyme must be
accompanied by, and is perhaps executed by, the
expression of particular transcriptional regulatory factors that cause the initial expression of cell-specific
genes.
We thank Nancy Thompson and Joan Lemire for initial
advice on the in situ hybridization protocol. John Coleman,
Gary Wessel, Yehudit Bergman, Patrice Milos, Mike DiPersio, and David Jackson provided valuable comments on the
manuscript. The research was supported by a grant to K.S.Z.
from the National Institutes of Health (GM36477).
References
BABISS, L. E., HERBST, R. S., BENNETT, A. L. AND DARNELL, J.
E. JR (1987). Factors that interact with the rat albumin
promoter are present both in hepatocytes and other cell types.
Genes Dev. 1, 256-267.
BEN ZE'EV, A., ROBINSON, G. S., BUCHER, N. L. R. AND FARMER,
S. R. (1988). Cell-cell and cell-matrix interactions differentially
regulate the expression of hepatic and cytoskeletal genes in
primary cultures of rat hepatocytes. Proc. natn. Acad. Sci.
U.S.A. 85, 2161-2165.
CARON, J. M. (1990). Induction of albumin gene transcription in
hepatocytes by extracellular matrix proteins. Molec. cell. Biol.
10, 1239-1243.
CEREGHINI, S., RAYMONDJEAN, M., CARRANCA, A. G., HERBOMEL,
P. AND YANIV, M. (1987). Factors involved in control of tissuespecific expression of albumin gene. Cell 50, 627-638.
CHIRGWIN, J. M., PRZBVLA, A. E., MACDONALD, R. J. AND
RUTTER, W. J. (1979). Isolation of biologically active ribonucleic
acid from sources enriched in ribonuclease. Biochemistry 18,
5294-5299.
CLAYTON, D . F., HARRELSON, A. L. AND DARNELL, J. E. JR (1985).
Dependence of liver-specific transcription on tissue organization.
Molec. cell. Biol. 5, 2623-2632.
DIPERSIO, C. M., JACKSON, D. A. AND ZARET, K. S. (1991). The
extracellular matrix coordinately modulates liver transcription
factors and hepatocyte morphology. Molec. cell. Biol. (in press).
DOLLE, P., CASTRILLO, J.-L., THEILL, L. E., DEERINCK, T.,
WLLISMAN, M. AND KARIN, M. (1990). Expression of GHF-1
protein in mouse pituitaries correlates both temporally and
spatially with the onset of growth hormone gene activity. Cell
60, 809-820.
FRASLIN, J.-M., KNEIP, B . , VAULONT, S., GLAISE, D . , MUNNICH,
A. AND GUGUEN-GUILLOUZE, C. (1985). Dependence of
hepatocyte-specific gene expression on cell-cell interaction in
primary culture. EMBO J. 4, 2487-2491.
HOGAN, B., CONSTATINI, F. AND LACY, E. (1986). Manipulating the
Mouse Embryo. Cold Spring Harbor Laboratory.
HOUSSAINT, E. (1980). Differentiation of the mouse hepatic
primordium. I. An analysis of tissue interactions in hepatocyte
differentiation. Cell Differ. 9, 269-279.
JEFFERSON, D , . M., CLAYTON, D. F., DARNELL, J. E. JR AND REID,
L. M. (1985). Posttranscriptional modulation of gene expression
in cultured rat hepatocytes. Molec. cell. Biol. 4, 1929-1934.
JOHNSON, P. F. (1990). Transcriptional activators in hepatocytes.
Cell Growth Differ. 1, 47-52.
JOHNSON, G. R. AND MOORE, M. A. S. (1975). Role of stem cell
migration in initiation of mouse foetal liver haemopoiesis.
Nature 258, 726-728.
KIOUSSIS, D . , EIFERMAN, F., VAN DE R U N , P., GORIN, M. B . ,
INGRAM, R. S. AND TILGHMAN, S. M. (1981). The evolution of a-
fetoprotein and albumin. The structures of the cr-fetoprotein and
albumin genes in the mouse. / . biol. Chem. 256, 1960-1967.
Kuo, C. F., XANTHOPOULOS, K. G. AND DARNELL, J. E. JR (1990).
Fetal and adult localization of C/EBP: evidence for
combinatorial action of transcription factors in cell specific gene
expression. Development 109, 473-481.
LEDOUARIN, N. (1968). Synthese du glycogene dans les
hepatocytes en voie de differentiation: role des mesenchymes
homologue et heterologues. Devi Biol. 17, 101-114.
LEDOUARIN, N. M. (1975). An experimental analysis of liver
development. Med. Biol. 53, 427-455.
Li, S., CRENSHAW, E . B . Ill, RAWSON, E. J., SIMMONS, D. M.,
SWANSON, L. W. AND ROSENFELD, M. G. (1990). Dwarf locus
mutants lacking three pituitary cell types result from mutations
in the POU-domain gene pit-1. Nature 347, 528-532.
LlCHTSTEINER, S., WUARIN, J. AND SCHIBLER, U. (1987). The
interplay of DNA-binding proteins on the promoter of the
mouse albumin gene. Cell 51, 963-973.
Liu, J.-K., BERGMAN, Y. AND ZARET, K. S. (1988). The mouse
albumin promoter and a distal upstream site are simultaneously
DNase I hypersensitive in liver chromatin and bind similar liver
abundant factors in vitro. Genes Dev. 2, 528-541.
Liu, J.-K., DIPERSIO, C. M. AND ZARET, K. S. (1991) Extracellular
signals that regulate liver transcription factors during hepatic
differentiation in vitro. Molec. cell. Biol. 11, 773-784.
MALLORY, F. B. (1961). Pathologic Technique. Hafner Publ. Co.,
p. 152.
Hepatic differentiation in mouse embryos
MANIATIS, T . , GOODBOURN, S. AND FlSCHER, J. A . (1987).
Regulation of inducible and tissue-specific gene expression.
Science 236, 1237-1245.
MCKNIGHT, S. L. AND KINGSBURY, R. (1982). Transcriptional
control signals of a eukaryotic protein coding gene. Science 217,
316-324.
MEEHAN, R. R., BARLOW, D. P., HILL, R. E., HOGAN, B. L. M.
AND HASTIE, N. D. (1984). Pattern of serum protein gene
expression in mouse visceral yolk sac and fetal liver. EMBO J.
3, 1881-1885.
MITCHELL, P. J. AND TJIAN, R. (1989). Transcriptional regulation
in mammalian cells by sequence-specific DNA binding proteins.
Science 245, 371-378.
MORRIS, S. M. JR, KEPKA, D. M., SWEENEY, W. E. JR AND AVNER,
E. D . (1989). Abundance of mRNAs encoding urea cycle
enzymes in fetal and neonatal mouse liver. Arch. Biochem.
Biophys. 269, 175-180.
NAHON, J. L., GAL, A., FRAIN, M., SELL, S. AND SALA-TREPAT, J.
M. (1982). No evidence for post-transcriptional control of
albumin and a--fetoprotein gene expression in developing rat
liver and neoplasia. Nucl. Acids Res. 10, 1895-1911.
PANDURO, A., SHALABY, F. AND SHAFRITZ, D. A. (1987). Changing
patterns of transcriptional and post-transcriptional control of
liver-specific gene expression during rat development. Genes
Dev. 1, 1172-1182.
PAUL, J., CONKIE, D. AND FRESHNEY, R. I. (1969). Erythropoietic
cell population changes during the hepatic phase of
erythropoiesis in the foetal mouse. Cell Tissue Kinet. 2,
283-294.
PINKERT, C. A., ORNITZ, D . M., BRINSTER, R. L. AND PALMITER,
R. D. (1987). An albumin enhancer located 10kb upstream
functions along with its promoter to direct efficient, liver-specific
expression in transgenic mice. Genes Dev. 1, 268-276.
225
transcriptional control of specific messenger RNAs in adult and
embryonic liver. J. molec. Biol. 179, 21-35.
RUGH, R. (1968). The Mouse: its Reproduction and Development.
Burgess Publishing Company, Minneapolis, MN.
RUTTER, W. J., CLARK, W. R., KEMP, J. D . , BRADSHAW, W. S.,
SANDERS, T. G. AND BALL, W. D. (1968). In
Epithelial-Mesnechymal Interactions (Ed. by R. Gleischmajer
and R. E. Billingham.) Williams and Wilkins, Baltimore. Pp.
114-131.
RUTTER, W. J., PICTED, R. L., HARDING, J. D . , CHIRGWIN, J. M.,
MACDONALD, R. J. AND PRZYBYLA, A. E. (1978). An analysis of
pancreatic development: Role of mesenchymal factor and other
extracellular factors. In Molecular Control of Proliferation and
Differentiation, 35th Symp. Soc. Dev. Biol. (ed. J.
Papaconstantinou and W. Rutter) Academic Press, pp. 205-227.
SHIOJIRI, N. (1984). Analysis of differentiation of hepatocytes and
bile duct cells in developing mouse liver by albumin
immunofluorescence. Dev. Growth and Differ. 26, 555-561.
SILINI, G., Pozzi, L. V. AND PONS, S. (1967). Studies on the
haemopoietic stem cells of mouse foetal liver. J. Embryol. exp.
Morph. 17, 303-318.
SMITH, P. E. AND MACDWELL, E. D . (1931). The differential effect
of hereditary mouse dwarfism on the anterior-pituitary hormone.
Anat. Rec. 50, 85-93.
THEILER, K. (1989). The House Mouse. Springer-Verlag, New
York.
TILGHMAN, S. M. AND BELAYEW, A. (1982). Transcriptional control
of the murine albumin/cv-fetoprotein locus during development.
Proc. natn. Acad. Sci. U.S.A. 79, 5254-5257.
WILKINSON, D. J., BAILES, J. A. AND MCMAHON, A. P. (1987).
Expression of the proto-oncogene int-1 is restricted to specific
neural cells in the developing mouse embryo. Cell 50, 79-88.
POWELL, D. J., FRIEDMAN, J. M., OULETTE, A. J., KRAUTER, K. S.
AND DARNELL, J. E. JR (1984). Transcriptional and post-
{Accepted 27 May 1991)