ã
Oncogene (2000) 19, 5054 ± 5062
2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00
www.nature.com/onc
Dual functions of E2F-1 in a transgenic mouse model of liver carcinogenesis
Elizabeth A Conner1, Eric R Lemmer1, Masako Omori1,2, Peter J Wirth1,3, Valentina M Factor1
and Snorri S Thorgeirsson*,1
1
Laboratory of Experimental Carcinogenesis, Division of Basic Sciences, National Cancer Institute, National Institutes of Health,
Bethesda, Maryland, MD 20892, USA
Deregulation of E2F transcriptional control has been
implicated in oncogenic transformation. Consistent with
this idea, we recently demonstrated that during
hepatocarcinogenesis in c-myc/TGFa double transgenic
mice, there is increased expression of E2F-1 and E2F-2,
as well as induction of putative E2F target genes.
Therefore, we generated transgenic mice expressing E2F1 under the control of the albumin enhancer/promoter to
test the hypothesis that E2F family members may
contribute to liver tumor development. Overexpression
of E2F-1 resulted in mild but persistent increases in cell
proliferation and death during postnatal liver growth,
and no increases in hepatic regenerative growth in
response to partial hepatectomy. Nevertheless, from 2
months postnatally E2F-1 transgenic mice exhibited
prominent hepatic histological abnormalities including
preneoplastic foci adjacent to portal tracts and pericentral large cell dysplasia. From 6 to 8 months onward,
there was an abrupt increase in the number of neoplastic
nodules (`adenomas') with 100% incidence by 10 months.
Some adenomas showed evidence of malignant transformation, and two of six mice killed at 12 months showed
trabecular hepatocellular carcinoma. Endogenous c-myc
was up-regulated in the early stages of E2F-1
hepatocarcinogenesis, whereas p53 was overexpressed in
the tumors, suggesting that both E2F-1-mediated
proliferation and apoptosis are operative but at di€erent
stages of hepatocarcinogenesis. In conclusion, E2F-1
overexpression in the liver causes dysplasia and tumors
and suggests a cooperation between E2F-1 and c-myc
oncogenes during liver oncogenesis. Oncogene (2000) 19,
5054 ± 5062.
Keywords: hepatocellular adenoma; hepatocellular carcinoma; c-myc; p53; mouse
Introduction
Cancer is increasingly viewed as a cell cycle disease and
most, if not all, tumors appear to have su€ered one or
more defects that derail the cell cycle machinery
(Bartek et al., 1999). A key target of carcinogens is
the retinoblastoma (Rb) pathway, whose role appears
*Correspondence: SS Thorgeirsson, National Cancer Institute, NIH,
37 Convent Drive MSC4255, Building 37, Room 3C28, Bethesda,
MD 20892-4255 USA
Current addresses: 2The First Department of Internal Medicine,
Okayama University Medical School, Okayama, Japan; 3Grants
Review Branch, Division of Extramural Activities, National Cancer
Institute, Bethesda, MD 20892, USA
Received 22 June 2000; revised 4 August 2000; accepted 9 August
2000
to be to guard and trigger the commitment in late G1
to initiate DNA replication and complete a round of
cell cycle division (Sherr, 1996; Strauss et al., 1995;
Bartek et al., 1997). The Rb protein has been shown to
bind and regulate a large number of cellular proteins,
including members in the E2F family (Macleod, 1999;
Dyson, 1998). To date there are six members of the
E2F family, E2F-1 through E2F-6 and two heterodimeric protein partners of the E2Fs, DP-1 and DP-2
(Macleod, 1999; Dyson, 1998). Binding of pRb family
members to the E2F transcription factors results in
transcriptional repression of E2F regulated genes
(Dynlacht, 1997). pRb binds preferentially to E2F-1,
-2, -3 and to a lesser degree E2F-4; whereas, p107 and
p130 bind preferentially to E2F-2 and E2F-5. E2F-6,
the newest member in the E2F family, lacks a pRb
binding domain and, therefore, does not bind to any
members of the pRb family (Cartwright et al., 1998).
Phosphorylation of the pocket proteins that occurs late
in G1 by cyclin-dependent kinases frees E2F, allowing
it to transactivate genes that regulate DNA replication
and cell cycle progression (Yamasaki, 1999).
It has been suggested that deregulation of E2F
transcriptional activity by Rb inactivation contributes
to the development of cancer (Weinberg, 1995; Nevins,
1992; Hunter and Pines, 1994). This hypothesis is
supported by the ®nding that several members of the
E2F gene family, in particular E2F-1, regulate the
expression of genes critical for cell proliferation
(Johnson et al., 1993, 1994; Qin et al., 1994). Although
E2F-1 promotes cell transformation in vitro and tumor
formation in E2F-1 transgenic mice (Johnson et al.,
1994; Singh et al., 1994; Pierce et al., 1999),
homozygous deletion of E2F-1 also results in an
increased incidence of cancer (Yamasaki et al., 1996;
Field et al., 1996). This dual role of E2F-1, which
appears to have both oncogenic and tumor suppressive
properties (Weinberg, 1996; Macleod and Jacks, 1999),
may re¯ect its ability to cause S phase entry and
apoptosis respectively (Shan and Lee, 1994; Johnson,
2000).
We have previously generated c-myc and c-myc/
TGF-a double transgenic mice which are prone to liver
cancer (Santoni-Rugiu et al., 1996). Both transgenic
lines exhibited an elevation in hepatocyte proliferation
and developed HCC, albeit with di€erent latency.
Interestingly, among the elements of cell cycle deregulation observed during c-myc-associated hepatocarcinogenesis was the abnormal increase in E2F activity.
As assessed by immunoprecipitation, pRb-free E2F-1 ±
DP1 and E2F-2 ± DP-2 complexes were abundantly
present in both c-myc and c-myc/TGF-a transgenic
livers and allowed transcriptional activation of E2Ftarget genes required for cell cycle progression,
Dual functions of E2F-1 in liver carcinogenesis
EA Conner et al
5055
including cyclin A and cdc2. We thus hypothesized that
E2F proteins (or up-regulation of E2F-1, possibly as a
direct e€ect of c-myc or via phosphorylation of pRb),
might contribute to liver oncogenesis by promoting
uncontrolled cell proliferation. To test this hypothesis,
we have generated the albumin-directed E2F-1 transgenic mouse and examined the e€ects of E2F-1 on liver
growth, regeneration and carcinogenesis. Our studies
suggest that E2F-1 when overexpressed in the liver
possesses both oncogenic and tumor-suppressive properties. The oncogenic activity of E2F-1 was mostly
evident during the preneoplastic stage and was
mediated by an early up-regulation of hepatocyte
proliferation resulting in rapid growth of multiple
focal lesions. However, slow but persistent escalation
of apoptotic death in association with an increase in
p53 transcriptional activity may convert E2F-1 from
positive to negative regulator of tumor growth and
delay malignant conversion. Moreover, a striking upregulation of c-myc expression in E2F-1 livers suggests
that cooperation between E2F-1 and c-myc oncogenes
is highly signi®cant and plays an important role in liver
oncogenesis.
Results
Generation of transgenic mice overexpressing E2F-1
under control of the albumin enhancer promoter
The E2F-1 transgene was made by subcloning the
human E2F-1 cDNA into a vector containing the
albumin enhancer promoter (Pinkert et al., 1987) and
the 3' ¯anking region of the human growth hormone
gene, which harbors the 3' untranslated region of exon
5 and the polyadenylation signal (Figure 1a). The
albumin enhancer promoter has been previously shown
to direct expression to the liver (Murakami et al.,
1993). Five E2F-1 transgenic lines were developed of
which two (lines 8 and 13) were chosen for further
characterization. High levels of the expected transcript
size (*1.9 kb) of the E2F-1 transgene were found in
both lines (Figure 1b) and correlated with high levels
of E2F-1 protein. Immunohistochemical analysis
revealed nuclear localization and homogeneous staining of E2F-1 protein in hepatocytes (data not shown).
DP1 co-immunoprecipitates with E2F-1 in transgenic
liver homogenates. Although E2F can bind DNA and
transactivate genes as a homodimer, the transcriptional
activity is maximized to optimal levels when E2F forms
a heterodimeric complex with the polypeptide DP1 (La
Thangue, 1994). To address this issue in our transgenic
mouse model, we analysed whether DP1 co-immunoprecipitated with E2F-1. Figure 1c shows high levels of
E2F-1 immunoprecipitates from liver homogenates
from lines 8 and 13. Stripping and reprobing the
identical membrane with an antibody to DP1 revealed
that E2F-1 co-immunoprecipitated with DP1, suggesting that E2F-1 is transcriptionally active (Figure 1c).
Effect of E2F-1 overexpression on cellular proliferation
During the ®rst 12 months, the E2F-1 transgenic
mouse lines showed only slight increases in liver weight
relative to total body weight in comparison to agematched controls (data not shown). The liver to body
Figure 1 Generation of E2F-1 transgenic lines. (a) Schematic
representation of the 4.2 kb Sac ± SacI vector-free DNA fragment
that was injected into nuclei of one-cell mouse embryo obtained
from mating of hybrid (c57BL/6J6CBA/J). Alb, albumin; Pr,
promoter; hGH, 3' ¯anking region of the human growth hormone
gene. (b) Northern blot showing E2F-1 transgene expression in
two E2F-1 transgenic lines in comparison to WT. Poly(A)+RNA
was extracted from two E2F-1 transgenic and WT livers and
subjected to Northern blot analysis using the human E2F-1
cDNA as probe. (c) Immunoprecipitation showing induction of
E2F-1/DP-1 complexes in E2F-1 transgenic livers. E2F-1 was
immunoprecipitated from liver sample homogenates from each
transgenic line (8 and 13), WT, and HeLa nuclear extract as
control. Each experiment was repeated at least twice and a
representative blot is shown above (b and c)
weight ratio for E2F-1 transgenic mice ranged from 5.0
to 6.3% as compared to 4.2 to 5.1% in controls. The
highest values in the E2F-1 transgenic mice were
reached after 10 months, which corresponds to the
appearance of neoplastic lesions and represents a 23%
increase over age-matched controls.
Evaluation of proliferative indices in E2F-1 transgenic livers showed an increase in cell proliferation as
measured by BrdU incorporation at each time point up
to 12 months (Figure 2). In 1-month-old livers, the
proliferation index was increased *1.4-fold over agematched WT livers. Subsequently, the proliferation
index decreased at 2 months and maintained thereafter
at values * twofold lower than those recorded at 1
month of age and * sixfold over age matched WT
livers.
To test whether E2F-1 overexpression a€ects the
regenerative capability of transgenic hepatocytes, PH
was performed in young adult (7 week) transgenic mice
(Figure 3a). The overall rate of BrdU incorporation
and time course of liver regeneration following PH
were comparable in E2F-1 transgenic and WT
hepatocytes. The percentage of hepatocytes showing
BrdU incorporation at 36 h was 22.0+2.0% in the
E2F-1 mice as compared to 18.7+3.3% in WT mice.
Oncogene
Dual functions of E2F-1 in liver carcinogenesis
EA Conner et al
5056
By 96 h E2F-1 mice regenerated 70% of their prehepatectomy mass which was similar to WT (Figure
3b).
Figure 2 Proliferation index in nontumorous tissues of E2F-1
transgenic mice. Transgenic line [8] and WT mice were injected
with BrdU and sacri®ced, and livers examined for BrdU
incorporation by immunohistochemistry as described in the
Materials and methods. Results are presented as the percentage
of hepatocytes showing BrdU staining at each time point
indicated. Each value represents the mean+s.e. of 5 ± 10 mice.
*P50.05 when compared with age-matched WT mice by
Student's t-test
Figure 3 Kinetics of DNA synthesis (a) and recovery of liver
weight (b) after PH in 7-week-old WT and E2F-1 transgenic
livers. Each data point represents the mean+s.e. of 3 ± 5 mice
Tumor development in E2F-1 transgenic mice
E2F-1 transgenic mice uniformly developed pericentral
dysplasia and foci adjacent to portal tracts followed by
an abrupt appearance of adenomas with subsequent
malignant conversion. The incidence and time course
of the appearance of the major hepatic lesions in line
eight animals are summarized in Figure 4 and in Table
1. Line 13 animals developed identical pathological
lesions in the liver, but there was a less rapid increase
in the incidence of dysplasia and the appearance of
tumors was delayed (data not shown). Large cell
dysplasia was ®rst noted around the central hepatic
veins between 2 ± 4 months, and by 10 months 100% of
the livers showed dysplastic changes. The cytological
features of dysplasia included cellular and nuclear
pleomorphism and hypertrophy, nuclear hyperchromasia, multiple prominent nucleoli, and eosinophilic
nuclear `pseudoinclusions' (Figure 5a). Mitoses were
infrequent and there was a paucity of apoptotic bodies
in dysplastic areas, suggesting low turnover of
dysplastic hepatocytes. Furthermore, the dysplastic
changes remained con®ned to the pericentral regions
of the liver lobule, and the severity of dysplasia
appeared not to progress beyond the cytological
features described above.
A prominent feature of E2F-1 associated hepatocarcinogenesis was the early development of focal
lesions (`foci') which was coincident with the development of the dysplastic changes and reached 100%
incidence by 10 months of age (Figure 4 and Table 1).
Although the number of preneoplastic foci was highly
variable, morphologically these lesions were uniform
and always appeared in close proximity to the portal
tracts. The majority of the foci were composed of
small round cells, with clear-cell phenotype but
eosinophilic, mixed and basophilic foci were also seen.
Figure 4 Incidence and time course of hepatic lesions in E2F-1
transgenic mice. Values are a percentage of animals a€ected at
each time point
Table 1 Number (%) of E2F-1 transgenic mice with liver lesions
Age
(months)
0±2
2±4
4±6
6±8
8 ± 10
10 ± 12
12+
Oncogene
Number (%) of mice with liver lesions
Number
of mice
Dysplasia
Foci
14
22
10
13
11
6
6
0 (0)
13 (59)
8 (80)
11 (85)
11 (100)
6 (100)
6 (100)
0 (0)
12 (55)
7 (70)
10 (77)
10 (91)
6 (100)
6 (100)
Adenoma
0
0
1
2
8
6
6
(0)
(0)
(10)
(15)
(73)
(100)
(100)
Carcinoma
0
0
0
0
0
2
1
(0)
(0)
(0)
(0)
(0)
(33)
(17)
Dual functions of E2F-1 in liver carcinogenesis
EA Conner et al
5057
Figure 5 Development of hepatic tumors in mice overexpressing E2F-1 in the liver. (a) Liver section from E2F-1 transgenic mouse
at 7 months showing evidence of liver cell dysplasia around a central vein. The centrally located dysplastic hepatocytes show cellular
and nuclear pleomorphism and hypertrophy, as compared with surrounding hepatocytes at the periphery of the photomicrograph.
An eosinophilic `pseudoinclusion' is seen within a large dysplastic nucleus (arrow), which also contains coarse staining chromatin.
(b) Lower power view of the liver from the same animal showing a `focus' composed of small cells directly adjacent to a portal tract
on the left, and large dysplastic cells surrounding a central vein on the right. (c) Macroscopic liver specimen from E2F-1 transgenic
mouse at 1 year showing multiple fungating tumors protruding from the surface of the median lobe (arrows) and subcapsular tumor
in the left lobe (arrowhead). (d) Liver section from E2F-1 transgenic mouse at 9 months showing the presence of tumor cells inside
the lumen of a central vein (arrows). These cells are not covered by endothelium, indicating early central vein invasion by tumor.
The surrounding liver shows evidence of severe macrovescicular steatosis. (e) Liver section from E2F-1 transgenic mouse at 10
months showing the presence of a nodule composed of larger eosinophilic staining cells (arrowheads) within a neoplastic clear cell/
eosinophilic adenoma (`nodule-in-nodule'). Most of the photomicrograph is taken up by the small cell nodule, but a small part of
the border between the nodule and surrounding liver is seen at the right upper corner (arrows). (f) Liver section from E2F-1
transgenic mouse at 1 year showing a typical trabecular hepatocellular carcinoma with cords of cells of varying thickness. Frequent
mitotic ®gures in tumor cells are indicated by the arrows. CV=central vein; PT=portal tract. Magni®cation 2006 (a,d,f) and 1006
(b and e)
A characteristic feature of the early stages of E2F-1
hepatocarcinogenesis was the strict zonal localization
of foci (periportal) and dysplastic lesions (pericentral),
and this zonal localization of lesions persisted
throughout the study period (Figure 5b). Morphologically, these preneoplastic foci were most similar to
Oncogene
Dual functions of E2F-1 in liver carcinogenesis
EA Conner et al
5058
those seen after chemical carcinogenesis (Solt and
Farber, 1976).
Neoplastic nodules were ®rst noted in the livers of
line 8 E2F-1 transgenic mice between 4 ± 6 months
(Figure 4 and Table 1). There was a sudden increase in
the incidence of nodules at 10 months, and by 12
months 100% of the animals had nodules. Macroscopically, the livers contained multiple nodules of
relatively small size, often involving several/all of the
lobes, (Figure 5c). Microscopic examination showed
that these hepatic nodules, which caused compression
of the surrounding liver parenchyma, were made up of
eosinophilic, clear, basophilic, or (rarely) `ground glass'
hepatocytes. These nodules had the characteristics of
hepatocellular adenoma (Frith and Ward, 1979). In
four of eight animals killed at 10 months, hepatic
histopathology showed features of malignant transformation in a few of the adenomas, including high
mitotic indices, blood vessel invasion (Figure 5d), and
central collections of deeply basophilic cells with large
nuclei giving a `nodule-in-nodule' appearance (Figure
5e). Although the diagnosis of unequivocal HCC in
mice is dicult (Frith and Ward, 1979), typical
trabecular tumors were found in the livers from two
of six animals killed at 12 months (Table 1, Figure 5f),
and in both cases the tumors were well-di€erentiated.
Macrovesicular hepatic steatosis was ®rst noted in
some E2F-1 transgenic livers at 6 ± 8 months, and by
10 ± 12 months 60% of animals had developed
prominent fatty change (Figure 5d). In some instances,
the fatty degeneration appeared to involve predominantly the mid-zonal regions of the liver lobule,
whereas in other cases there was extensive steatosis
throughout the entire lobule. Hepatic steatosis has
been noted in several transgenic mouse models of liver
carcinogenesis (Santoni-Rugiu et al., 1996; Moriya et
al., 1998), and its signi®cance is unknown.
Figure 6 Proliferation and cell death indices in E2F-1 hepatocarcinogenesis pathway. Proliferating cells visualized by BrdU
labeling. Apoptotic indices were scored from parallel sections
stained with H&E. Both indices are expressed as the percentage of
hepatocytes showing BrdU staining or apoptotic morphology at
each histological stage indicated. Values (mean+s.e.) for each
histological stage, except HCC, represent 10 ± 30 lesions from at
least ®ve di€erent mice. Values for HCC represent results from
two tumor samples from two di€erent animals. *P50.05 when
compared with proliferation frequency in dysplastic stages or
apoptotic frequency in adenomas by the Student's t-test
Mitotic and apoptotic indices in E2F-1 hepatic
oncogenesis
The ability of tumor cell populations to expand in
number is determined not only by net cell proliferation
but also by the rate of cell attrition (i.e. apoptosis)
(Hanahan and Weinberg, 2000). Cell proliferation as
measured by BrdU incorporation increased dramatically (sevenfold) from 1.4% in the early dysplastic
stages to nearly 10% in the foci and was maintained in
both adenomas and carcinomas (Figure 6). Apoptotis,
on the other hand, increased progressively, from the
early dysplastic lesions to carcinoma (Figure 6). At all
stages of tumorigenesis, the level of proliferation
surpassed the level of apoptotic cell death. However,
the proliferation : apoptosis ratio decreased from 0.25
in foci to 0.15 in adenomas to 0.09 in HCC.
Patterns of gene expression
The slow, but persistent escalation of apoptotic cell
death prompted us to determine involvement of p53 in
this model. E2F-1 has already been implicated to
induce apoptosis in a p53-dependent manner (Wang et
al., 2000; Pierce et al., 1998a) in other E2F-1 transgenic
mouse models. Figure 7 shows that during E2F-1
induced hepatocarcinogenesis high expression of p53
mRNA in the early stages (1 ± 5 months) decreases and
Oncogene
Figure 7 Northern blot analysis of temporal gene expression in
normal wild-type and E2F-1 transgenic livers. Poly(A)+ was
obtained from WT and E2F-1 livers as reported in the `Materials
and methods'. At each time point, 5 mg of Poly(A)+ isolated from
one mouse was subjected to Northern blot analysis. Membranes
were probed sequentially with E2F-1, c-myc, p53, p21, p16 cdc2,
and cyclin D1. Mouse ribosomal protein L7 was used for
normalization. Preneoplastic, samples collected before the appearance of tumors; Neoplastic, samples collected during the
neoplastic phase from peritumorous and tumorous tissues. The
experiment was repeated several times using di€erent animal
preparations. A representative membrane is shown
Dual functions of E2F-1 in liver carcinogenesis
EA Conner et al
rises again in the tumors in comparison to nontumorous tissues. Importantly, E2F-1 tumors also displayed
a similar pattern of up-regulation of p-53-inducible p21
mRNA, indicating that p53 was transcriptionally
active. Another negative regulator of the cell cycle,
p16, was also found gradually upregulated in the
tumors. Levels of c-myc were also examined since
aberrant expression of the nuclear oncogene has been
implicated in the development of a wide variety of both
experimental and naturally occurring tumors, including
HCC (DePhino, 1991; Nagy, 1988; Santoni-Rugiu,
1996). The steady state levels of c-myc expression were
considerably higher, although variable, in transgenic
than in age-matched WT livers starting from preneoplastic stages to tumors (Figure 7). We also
observed the overexpression of two E2F target genes,
cyclin D1 and cdc2, each playing a role in cell cycle
progression. Cyclin D1 mRNA was induced in all preneoplastic stages in comparison to WT although
expression gradually declined with only sporadic
expression in the tumors. Cdc2, on the other hand,
was upregulated at 1 month of age and in some of the
tumors. The constitutive expression of E2F-1 transgene
was maintained with minor variations both in the livers
and tumors throughout life span (Figure 7).
Discussion
We report here that overexpression of E2F-1 in the
liver leads to the development of hepatocellular
adenoma and, ultimately, HCC in transgenic mice.
This e€ect was found in two examined E2F-1
transgenic mouse lines (lines 8 and 13) and was
independent of site of transgene integration (EA
Conner, unpublished results). Both lines developed
the same phenotype with similar kinetics of stepwise
progression from preneoplastic to neoplastic lesions.
We present evidence that in this in vivo model of liver
carcinogenesis E2F-1 possesses both oncogenic and
tumor-suppressive properties. The oncogenic property,
mediated by moderate but persistent elevation of cell
proliferation during early stages of hepatocarcinogenesis, and the tumor-suppressive property, mediated by
a steady-state increase in the apoptotic rate, conceivably during tumor progression, delayed malignant
conversion.
The comparison of di€erent E2F-1 transgenic mouse
models shows that the interplay between growthpromoting and apoptotic activity of E2F-1 is tissuespeci®c and de®nes the impact of this transcription
factor on tumorigenesis (Table 2). Ultimate hepatic
tumor development would suggest that even a
relatively small (2 ± 6-fold) but steady rise in cell
proliferation driven by E2F-1 transgene could bring
about an increase in tumorigenicity in normally
mitotically silent liver. Although the unbalanced
Table 2
Target tissue
Megakaryocytes
Testis
Skin
Liver
mitogenic stimulus of E2F-1 triggers apoptosis, in
particular at the later stages of liver tumor development, the level of proliferation always surpassed the
level of apoptotic death. So it can be assumed that in
our model continual state of growth stimulation
overrides the apoptotic activity of E2F-1 and thereby
contributes to hepatocarcinogenesis. In this regard it is
worth noting that overexpression of E2F-1 in megakaryocytes induced not only hyperproliferation but
signi®cant apoptosis (Table 2). In this model system
increased E2F-1 activity blocked terminal di€erentiation but had no e€ect on tumorigenesis (Guy et al.,
1996). In another system of high cell turnover,
ubiquitous overexpression of E2F-1 promoted predominantly apoptosis resulting in testicular atrophy and
reduced fertility (Holmberg et al., 1998).
It is not clear how to reconcile multiple hepatic
tumor development with a steady up-regulation of
apoptosis in E2F-1 transgenic mice. Remarkably, in
E2F-1 hepatic tumors, which were mostly adenomas,
p53 expression was increased coincident with a rise in
apoptotic activity. Furthermore, the levels of p21, a
known target of p53, and p16, a negative regulator
of the cell cycle, were notably up-regulated in E2F-1
tumors. These results suggesting that the E2F-1mediated p53-dependent apoptosis may still be
operative and may account for the long latency,
relative small size and small proportion of precursor
lesions undergoing malignant change. Consistent with
this idea, K5 E2F-1 transgenic mice (Table 2), in
which E2F-1 deregulated expression promoted hyperproliferation, hyperplasia, p53-dependent apoptosis
and delayed sporadic tumor formation (Pierce et al.,
1998a,b), when subjected to the two-stage chemical
carcinogenesis treatment were resistant to skin tumor
development, presumably through an increase in p53dependent apoptotic cell death (Pierce et al., 1999).
Taken together these results imply that E2F-1 exerts
its tumor-suppressive e€ect during the promotion
stage of tumor development. Similar conclusions were
made in the E2F-1 knockout study which proposed
that the ability of E2F-1 to induce apoptosis
underlies its tumor suppressive properties (Field et
al., 1996).
Evidence is mounting that acquired resistance
toward apoptosis is a hallmark of most if not all types
of cancer (Hanahan and Weinberg, 2000). Inappropriately increased E2F or c-myc signals in vivo may select
for cells that have inactivated the ARF ± Mdm2 ± p53
pathway (Scherr and Weber, 2000; Pierce et al., 1998b).
More recently, it has been shown that p53 may
increase the expression of molecules which block the
function of inhibitors of apoptosis proteins (Goyal et
al., 2000), providing an additional mechanism for p53
escape from apoptotic cell death. At present it is
unclear whether selection for an inactivated ARF ±
Mdm2 ± p53 pathway or another mechanism is opera-
5059
Comparison of E2F-1 transgenic models
Proliferation
Apoptosis
Tumors?
++
7
++
+
++
++
++
+
No
No
Yes
Yes
References
Guy et al., 1996
Holmberg et al., 1998
Pierce et al., 1998a,b
Oncogene
Dual functions of E2F-1 in liver carcinogenesis
EA Conner et al
5060
tive in E2F-1 hepatocarcinogenesis. Further study of
E2F-1 tumors, particularly carcinomas, is required.
One intriguing ®nding in the E2F-1 transgenic mice
was a strong induction of endogenous c-myc which
occurred early and was maintained in all stages of E2F-1
hepatocarcinogenesis. A potential role of E2F-1 in
mediating biological activities of c-myc in vivo has been
recently reported (Rounbehier et al., 2000). Previously,
we found a similar accumulation of E2F-1 proteins
during c-myc associated liver tumor development
(Santoni-Rugiu et al., 1998). Although we do not know
whether this parallel increase in the expression of E2F-1/
c-myc is a direct or indirect e€ect of each oncogene, the
ability of E2F-1 to regulate levels of c-myc and vice versa
points to the importance of their cooperation in
transformation. Our recent data in E2F-1/c-myc double
transgenic mice revealed a 28% increase in liver/body
weight ratio and widespread dysplasia in comparison to
mono-transgenic E2F-1 by 3 months of age (EA Conner,
unpublished results). These initial results have been
providing more direct evidence for cooperation between
c-myc and E2F-1 in the liver oncogenesis. Interestingly,
cyclin D1, a known E2F target gene and contributor to
tumor development (Scherr, 1994) has been shown to
have myc binding sites (Daksis et al., 1994). Thus, the
observed increases in cyclin D1 expression suggest that
cyclin D1 may also play a role in the persistent elevation
of cell proliferation during the early stages of E2F-1 liver
oncogenesis. These ®ndings are in accordance with
Pierce et al. (1998b) who have shown that the K5 E2F1/cyclin D1 double transgenic mice have much more
severe phenotype with hyperproliferation and low
viability than either single transgenic mouse model.
It is also noteworthy that over-expression of E2F-1
did not provide any growth advantage during
compensatory regeneration after surgical removal of
two-thirds of the liver. The rate of DNA synthesis as
well as liver weight gain were comparable in E2F-1 and
wild type (WT) mice. Similarly, Lukas et al. (1999)
reported that E2F-1 does not in¯uence the regenerative
capacity of the liver. In two strains of mice nullizygous
for E2F-1 in the liver, E2F-1 de®ciency resulted in only
minor changes in gene expression and timing of liver
regeneration. The authors concluded that E2F-1
activity is not required for proliferation and suggest
that other E2F members, in particular E2F-4, are
critical for liver cell growth (Lukas et al., 1999).
However, in mice which overexpress E2F-4 in the skin,
as compared to the previously generated K5 E2F-1
mice, E2F-4 overexpression had a similar e€ect on
proliferation but was a poor inducer of apoptosis
(Wang et al., 2000).
In conclusion, E2F-1 overexpression in the liver
provides further support for the dual function of E2F-1
as both an oncogene and tumor suppressor. One future
application of this model will be to validate the
potential collaboration between E2F-1 and c-myc in
liver oncogenesis.
Materials and methods
Recombinant E2F-1 construct
The human E2F-1 cDNA clone, PSG5-E2F-1 (a gift from Dr
W Kaelin, Dana-Farber Cancer Institute) was digested with
Oncogene
EcoRI. The 1.3 kb E2F-1 fragment containing the full-length
coding sequence was inserted into an EcoRV site located
downstream of a 5' ¯anking region of the mouse albumin
promoter (Pinkert et al., 1987). A 0.6 kb fragment of the 3'
¯anking region of the human growth hormone (hGH) gene,
harboring the 3' untranslated region of exon 5 and the
polyadenylation signal, was inserted at the KpnI site of the
2335A-1 to obtain the ®nal construct (Figure 1a).
Generation of E2F-1 transgenic mice
A 4.2 kb vector-free fragment was obtained by digestion of
Rin-hE2F-1 with SacI, puri®ed and microinjected into onecell embryos obtained from mating hybridization (C57BL/
6J6CBA/J) F1 mice. Transgene screening was performed by
polymerase chain reaction (PCR) and Southern blot analysis
of tail DNA. The resulting founder mice were bred to
homozygosity in the B66CBA F1 background.
Gross and histopathological analyses
Between six and fourteen (median=8) mice were sacri®ced by
cervical dislocation at monthly intervals over a 12-month
period. Body weights were recorded and livers were obtained
by autopsy, weighed, and examined for macroscopic tumors,
which were recorded and measured. For routine light
microscopy, slices of liver 4 ± 5 mm in thickness were taken
from each lobe and from tumors 53 mm. These were ®xed in
10% neutral bu€ered formalin for 24 h, embedded in paran
wax, sectioned at 4 mM, and stained with hematoxylin and
eosin (H&E), according to standard methods. The liver slides
were reviewed in a blinded fashion by one of the investigators
(ER Lemmer), and histopathological diagnoses were based
upon criteria described by Frith and Ward (1979).
Cellular proliferation and apoptosis
Cell proliferation was measured by BrdU incorporation.
Mice were given intraperitoneal injections of BrdU (150 mg/
kg) 1 h before their sacri®ce. Liver samples were collected,
®xed in 70% alcohol : formalin (10 : 1) and paran embedded. Tissue sections were then subjected to immunohistochemistry using BrdU antibody (Becton Dickinson, San
Jose, CA, USA). For determining the labeling index,
hepatocytes were examined and the number of unstained
and stained cells was determined. At least 1000 cells were
counted per mouse. The apoptotic index was scored on
H&E-stained livers from at least ®ve animals per time point.
Brie¯y, 1000 hepatocytes/mouse from nontumorous areas
were randomly evaluated with a phase contrast microscope
(Nikon Microphot FXA) by an observer blinded to the
experiment (E.R.L.) The indices were represented at a
percentage (mean+s.e.) of the total cells counted. The
morphological criteria used to recognize apoptotic cells have
been described previously (Santoni-Rugiu et al., 1998), and
included the following: (a) shrunken cells, often with an
empty space between neighboring cells; (b) strongly and
homogenously eosinophilic cytoplasm; (c) condensation of
chromatin into dense particles; (d) nuclear fragmentation in
apoptotic bodies; and (e) isolated distribution of the
apoptotic bodies. At least four criteria were required to
classify a cell as apoptotic.
Regenerative liver growth
Male mice at 7 weeks of age were subjected to a standard
70% partial hepatectomy (PH) under metofane anesthesia
using the technique described by Higgins and Anderson
(1931). To monitor the kinetics of DNA synthesis, the
animals were given intraperitoneal injections of BrdU
(150 mg/kg) 1 h before their sacri®ce. Mice were killed 1 ±
96 h after PH in groups of three to ®ve.
Dual functions of E2F-1 in liver carcinogenesis
EA Conner et al
Immunohistochemistry
After deparanization and rehydration of alcohol/formalin®xed sections, endogenous peroxidase activity was inhibited
by incubation in 0.01 mol/L periodic acid in phosphatebu€ered saline (10 min), followed by treatment with 0.1 mol/
L sodium borohydride in phosphate-bu€ered saline (10 min).
DNA was denatured in 3N HCl for 15 min. Anti-BrdU
antibody (Becton Dickinson Co., San JoseÂ, CA, USA) at
1 : 200 dilution and avidin ± biotin ± peroxidase (Vectastain
ABC Kit, Vector Laboratories, Burlingame, CA) method
were used to stain BrdU-positive cells. For E2F-1 immunostaining, formalin- ®xed sections were microwaved in 10 mM
sodium citrate for 10 min and preincubated for 30 min in
blocking bu€er (Boehringer Mannheim, Indianapolis, IN,
USA) containing 1% BSA, 1% mouse serum, and 1.5%
normal goat serum. Three micrograms of mouse monoclonal
E2F-1 antibody (C-20, Upstate Biotechnology, Lake Placid,
New York, USA) was added to the slides in the same bu€er
and incubated overnight at 48C. Its binding was revealed by
using Vectastain ABC Elite kit and VIP peroxidase substrate
(Vector Laboratories, Burlingame, CA, USA) as chromogens,
according to the manufacturer's instructions (Santoni-Rugiu
et al., 1998).
Northern blot analysis
Poly(A)+ RNA was obtained from WT and transgenic livers
by oligo(dT)-cellulose chromatography, as described earlier
(Nagy et al., 1996). Five mg of mRNA from each sample were
electrophoresed in a 1% agarose gel/2.2 M formaldehyde gel,
transferred onto nylon membrane, and hybridized to
[32P]dCTP-labeled cDNA probes in QuikHyb Hybridization
Solution (Stratagene, LaJolla, CA). The probes used were
p53 (rat cDNA, 1.3 kb), E2F-1 (human cDNA, 1.3 kb); cmyc (mouse cDNA, 1.8 kb); p16 (mouse cDNA 1.1 kb), a
generous gift of Dr CJ Scherr; cyclin D1 (mouse cDNA
0.9 kb); p21 (mouse cDNA 0.85 kb). A 0.35 kb PCRgenerated fragment of the mouse ribosomal protein L7
cDNA served as a loading control. After washing o€ nonspeci®c binding membranes were exposed to PhosphorImager
for quantifying mRNA expression (Molecular Dynamics,
Sunnyvale CA).
Immunoprecipitation studies
Two hundred milligrams of liver sample was homogenized in
ice-cold lysis bu€er containing 50 mM HEPES, pH 7.3,
250 mM NaCl, 0.1% NP-40, 5 mM EDTA, 50 mM sodium
¯uoride and 0.1 mM sodium vanadate, 1 mM phenylemethylsulfonyl ¯uoride, 10 mg/ml apropotinin and 10 mg/ml leupeptin. After 30 min incubation on ice, homogenates were
centrifuged at 14 000 r.p.m. for 20 min at 48C. The supernatant was removed and centrifuged again at 40 000 r.p.m.
for 10 min at 48C. Protein concentrations were determined by
BioRad Protein Assay kit (BioRad, Hercules, CA, USA).
Immunoprecipitations were carried out overnight at 48C on
a rotating platform incubating in 1 ml of lysis, 1 mg of
protein with 4 mg of monoclonal antibody against mouse
E2F-1 (Upstate Biotechnology, Inc., Lake Placid, NY, USA).
Immune complexes were recovered by absorption to 50 ml of
Gamma Bind G Sepharose on a rotating platform for 1 h at
48C. After washing the beads 5 times with HNTG bu€er
[20 mM HEPES (pH 7.5), 150 mM NaCl, 0.1% Triton X-100,
and 10% glycerol], the pellets were resuspended in sample
bu€er [100 mM Tris-HCL (pH 6.8), 5% SDS, 5% glycerol,
0.005% bromophenol Blue, and 5% 2B-mercaptoethanol],
boiled 5 min, and centrifuged to separate the immunoprecipitated proteins from the beads.
5061
Western blot analysis
Twenty ml of each sample of immunoprecipitate or 40 ml of
whole cell lysate were separated by 10% SDS ± PAGE and
electroblotted to nitrocellulose. The immunoprecipitated
proteins were detected by a 2 h incubation at room
temperature in TBS-T containing 2% milk with antibodies
for E2F-1 (#05-379, Upstate Biotechnology, NY, USA), DP1 (#sc-610, Santa Cruz, CA, USA) or Rb (#14001A,
PharMingen, CA, USA). Immunoreactions were visualized
using ECL detection system (Amersham Corp., Arlington
Heights, IL, USA).
Acknowledgments
We thank Dr William Kaelin, Dana-Farber Cancer
Institute, for providing the human E2F-1 cDNa clone;
Dr Nicholas C Popescu and coworkers (Molecular Cytogenetics Section, Laboratory of Experimental Carcinogenesis, NCI) for performing FISH analyses; Dr Peter Nagy
(Laboratory of Experimental Carcinogenesis, NCI) for his
help and advice during the course of these studies; Nancy
D Sanderson for her help in generating the E2F-1 mice;
and Ms Tyjen Tsai for excellent technical assistance.
References
Bartek J, Bartkova J and Lukas J. (1997). Exp. Cell Res.,
237, 1 ± 6.
Bartek J, Lukas J and Bartkova J. (1999). J. Pathol., 187,
95 ± 99.
Cartwright P, Muller H, Wagener C, Holm K and Helin K.
(1998). Oncogene, 17, 611 ± 623.
Cohen C and Berson SD. (1986). Cancer, 57, 1535 ± 1538.
Daksis JI, Lu RY, Facchini LM, Marhin WW and Penn LJZ.
(1994). Oncogene, 9, 3635 ± 3645.
DePhino RA, Schreiber-Agus N and Alt FW. (1991). Adv.
Cancer Res., 57, 1 ± 46.
Dynlacht BD. (1997). Nature, 389, 149 ± 152.
Dyson N. (1998). Genes Dev., 12, 2245 ± 2262.
Factor VM, Jensen MR and Thorgeirsson SS. (1997).
Hepatology, 26, 1434 ± 1443.
Field SJ, Tsai FY, Kuo F, Zubiaga AM, Kaelin WGJ,
Livingston DM, Orkin SH and Greenberg ME. (1996).
Cell, 85, 549 ± 561.
Frith CH and Ward JM. (1979). J. Environ. Pathol. Toxicol.,
3, 329 ± 351.
Goyal L, McCall K, Agapite J, Hartwig E and Stellar H.
(2000). EMBO J., 19, 589 ± 597.
Guy CT, Zhou W, Kaufman S and Robinson MO. (1996).
Mol. Cell. Biol., 16, 685 ± 693.
Hanahan D and Weinberg RA. (2000). Cell, 100, 57 ± 70.
Higgins GM and Anderson RM. (1931). Arch. Pathol., 12,
186 ± 202.
Holmberg C, Helin K, Sehested M and Karlstrom O. (1998).
Oncogene, 17, 143 ± 155.
Hunter T and Pines J. (1994). Cell, 79, 573 ± 582.
Johnson DG. (2000). Mol. Carcinog., 27, 151 ± 157.
Johnson DG, Cress WD, Jakoi L and Nevins JR. (1994).
Proc. Natl. Acad. Sci. USA, 91, 12823 ± 12827.
Johnson DG, Schwarz JK, Cress WD and Nevins JR. (1993).
Nature, 365, 349 ± 352.
La Thangue NB. (1994). Trends. Biochem. Sci., 19, 108 ± 114.
Lukas ER, Bartley SM, Graveel CR, Diaz ZM, Dyson N,
Harlow E, Yamasaki L and Farnham PJ. (1999). Mol.
Carcinog., 25, 295 ± 303.
Macleod K. (1999). Curr. Opin. Genet. Dev., 9, 31 ± 39.
Oncogene
Dual functions of E2F-1 in liver carcinogenesis
EA Conner et al
5062
Macleod KF and Jacks T. (1999). J. Pathol., 187, 43 ± 60.
Moriya K, Fujie H, Shintani Y, Yotsuyanagi H, Tsutsumi T,
Ishibashi K, Matsuura Y, Kimura S, Miyamura T and
Koike K. (1998). Nature Med., 4, 1065 ± 1067.
Murakami H, Sanderson ND, Nagy P, Marino PA, Merlino
G and Thorgeirsson SS. (1993). Cancer Res., 53, 1719 ±
1723.
Nagy P, Bisgaard HC, Santoni-Rugiu E and Thorgeirsson
SS. (1996). Hepatology, 23, 71 ± 79.
Nevins JR. (1992). Science, 258, 424 ± 429.
Pierce AM, Gimenez-Conti IB, Schneider-Broussard R,
Martinez LA, Conti CJ and Johnson DG. (1998a). Proc.
Natl. Acad. Sci. USA, 95, 8858 ± 8863.
Pierce AM, Fisher SM, Cont CJ and Johnson DG. (1998b).
Oncogene, 16, 1267 ± 1276.
Pierce AM, Schneider-Broussard R, Gimenez-Conti IB,
Russell JL, Conti CJ and Johnson DG. (1999). Mol. Cell
Biol., 19, 6408 ± 6414.
Pinkert CA, Ornitz DM, Brinster RL and Palmiter RD.
(1987). Genes Dev., 1, 268 ± 276.
Qin XQ, Livingston DM, Kaelin WGJ and Adams PD.
(1994). Proc. Natl. Acad. Sci. USA, 91, 10918 ± 10922.
Rounbehier R, Schneider-Broussard R, Conti CJ and
Johnson DJ. (2000). Proc. American Assoc. Cancer
Research, 41, 394.
Santoni-Rugiu E, Jensen MR and Thorgeirsson SS. (1998).
Cancer Res., 58, 123 ± 134.
Santoni-Rugiu E, Nagy P, Jensen MR, Factor VM and
Thorgeirsson SS. (1996). Am. J. Pathol., 149, 407 ± 428.
Shan B and Lee WH. (1994). Mol. Cell Biol., 14, 8166 ± 8173.
Scherr CJ. (1994). Cell, 79, 551 ± 555.
Sherr CJ. (1996). Science, 274, 1672 ± 1677.
Sherr CJ and Weber JD. (2000). Curr. Opin. Genet. Dev., 10,
94 ± 99.
Singh P, Wong SH and Hong W. (1994). EMBO J., 13,
3329 ± 3338.
Solt D and Farber E. (1976). Nature, 263, 701 ± 703.
Strauss M, Lukas J and Bartek J. (1995). Nature Med., 1,
1245 ± 1246.
Wang D, Russell JL and Johnson DG. (2000). Mol. Cell.
Biol., 20, 3417 ± 3424.
Weinberg RA. (1995). Cell, 81, 323 ± 330.
Weinberg RA. (1996). Cell, 85, 457 ± 459.
Xu G, Livingston DM and Krek W. (1995). Proc. Natl. Acad.
Sci. USA, 92, 1357 ± 1361.
Yamasaki L. (1999). Biochim. Biophys. Acta, 1423, M9 ± 15.
Yamasaki L, Jacks T, Bronson R, Goillot E, Harlow E and
Dyson NJ. (1996). Cell, 85, 537 ± 548.
This paper is a `United States Government Work' paper as de®ned by the US Copyright Act.
Oncogene