Oncogene (2000) 19, 3623 ± 3631
ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00
www.nature.com/onc
Aberrant cell cycle progression contributes to the early-stage accelerated
carcinogenesis in transgenic epidermis expressing the dominant negative
TGFbRII
Cindy Go1, Wei He2, Ling Zhong3, Ping Li3, Juan Huang3, Bill R Brinkley3 and
Xiao-Jing Wang*,2,3
1
Department of Otolaryngology, Baylor College of Medicine, Houston, Texas, TX 77030, USA; 2Department of Dermatology,
Baylor College of Medicine, Houston, Texas, TX 77030, USA; 3Department of Molecular and Cellular Biology, Baylor College of
Medicine, Houston, Texas, TX 77030, USA
Mutations in the transforming growth factor b type II
receptor (TGFbRII) have been found in various
malignant tumors, suggesting that loss of TGFb
signaling plays a causal role in late-stage cancer
development. To test whether loss of TGFbRII is
involved in early-stage carcinogenesis, we have generated
transgenic mice expressing a dominant negative
TGFbRII (DbRII) in the epidermis. These mice
exhibited an increased susceptibility to chemical carcinogenesis protocols at both early and late stages. In the
current study, parameters for cell cycle progression and
chromosome instability were analysed in DbRII tumors.
DbRII papillomas showed an increased S phase in ¯ow
cytometry. Bromodeoxyuridine (BrdU) labeling and
mitotic indices in DbRII papillomas also showed a
threefold increase compared to papillomas developing in
non-transgenic mice. When papillomas further progressed to squamous cell carcinomas (SCC), both control
and DbRII SCC showed similar BrdU labeling indices
and percentages of S phase cells. However, DbRII SCC
cells showed a sixfold increase in the G2/M population.
Mitotic indices in DbRII SCC also showed a threefold
increase compared to non-transgenic SCC. Consistent
with a perturbed cell cycle, DbRII papillomas and SCC
showed reduced expression of the TGFb target genes p15
(INK4b), p21 (WAF-1) and p27 (Kip1), inhibitors of
cyclin-dependent kinases (cdks). However, most DbRII
papilloma cells exhibited normal centrosome numbers,
and DbRII SCC exhibited a similar extent of centrosome
abnormalities compared to control SCC (35 ± 40% cells).
Most of DbRII SCC exhibited diploid chromosome
pro®les. These data indicate that inactivation of
TGFbRII accelerates skin tumorigenesis at early stages
by the acceleration of loss of cell cycle control, but not
by increased chromosome instability. Oncogene (2000)
19, 3623 ± 3631.
Keywords: skin carcinogenesis; transforming growth
factor b receptor; transgenic; cell cycle progression;
cdk inhibitors, centrosome
Introduction
The transforming growth factor b (TGFb) is a potent
growth inhibitor in epithelial tissues (for reviews, see
*Correspondence: X-J Wang
Received 7 December 1999; revised 11 May 2000; accepted 23 May
2000
Filmus and Kerbel (1993); Fynan and Reiss (1993);
Satterwhite and Moses (1994)). It is believed that TGFb
acts as a tumor suppressor, since malignant epithelial
tumors develop mechanisms of escaping from TGFbinduced growth inhibition (Filmus and Kerbel, 1993;
Fynan and Reiss, 1993; Satterwhite et al., 1994). The
functions of TGFb are mediated by the tetraheteromers of the type I and II TGFb receptors
(TGFbRI and TGFbRII, (Derynck, 1994)). Upon
ligand binding, the intrinsic serine/threonine kinase of
TGFbRII phosphorylates TGFbRI to activate its
serine/threonine kinase activity (Wrana et al., 1994;
Chen and Weinberg, 1995). This activated receptor
complex transduces TGFb signals (Feng et al., 1995).
Deletion of the serine/threonine kinase domain of
TGFbRII produces a dominant negative form (DbRII),
which is able to block the growth inhibitory function
of TGFb in vitro (Chen et al., 1993; Wieser et al., 1993)
and in transgenic animals (Wang et al., 1997b;
Bottinger et al., 1997; Gorska et al., 1998), suggesting
that TGFbRII is required to mediate the growth
inhibitory e€ect of TGFb. Mutations in TGFbRII
were initially detected in TGFb-resistant cancer cell
lines (Markowitz et al., 1995; Myero€ et al., 1995;
Carcamo et al., 1995). Similar mutations have then
been reported in primary cancers of colon (Myero€ et
al., 1995; Takenoshita et al., 1997; Iacopetta et al.,
1998; Lu et al., 1998), head and neck (Lu et al., 1995,
1998; Wang et al., 1997a), ampulla (Imai et al., 1998),
and pancreas (Venkatasubbarao et al., 1998). These
data suggest that inactivation of TGFbRII might be
one mechanism by which epithelial tumor cells escape
from TGFb-induced growth inhibition and progress to
malignancy. To test this hypothesis in the skin, we
have generated transgenic mice expressing DbRII in the
epidermis, utilizing a truncated mouse loricrin vector
(ML.DbRII). ML.DbRII mice exhibit a marked
hyperplasia/hyperkeratosis at birth, suggesting that
the DbRII can block TGFb mediated growth inhibition
in the epidermis (Wang et al., 1997b). This is further
supported by the fact that primary keratinocytes
isolated from the epidermis of ML.DbRII transgenic
mice were resistant to exogenous TGFb1 induced
growth inhibition (Wang et al., 1997b). When
ML.DbRII mice were subjected to a two-stage chemical
carcinogenesis protocol, they exhibited an increased
susceptibility compared to non-transgenic mice, with
greatly accelerated benign papilloma formation, malignant conversion and metastasis (Go et al., 1999). In
addition, topical application of the tumor promoter
TPA to ML.DbRII mice elicited papillomas which had
Aberrant cell cycle progression in tumors expressing DbRII
C Go et al
3624
a low rate of c-rasHa mutations (Go et al., 1999),
suggesting that DbRII can play an initiating role in
chemical carcinogenesis. These data indicate that
blocking TGFb signaling plays a role in accelerating
carcinogenesis at both early and late stages. An
analysis of the mechanisms by which DbRII accelerates
carcinogenesis at later stages revealed that ML.DbRII
tumors exhibited increased angiogenesis, which was
consistent with the rapid progression of these tumors
to metastasis (Go et al., 1999). Here we provide in vivo
evidence that aberrant cell cycle progression contributes signi®cantly to accelerated skin carcinogenesis at
early stages.
Results
ML.DbRII tumor cells exhibit increased cell cycle
progression
Tumors in ML.DbRII and control mice were induced by
dimethylbenz[a]anthracene
(DMBA)/12-O-tetradecanoylphorbol-13-acetate (TPA), as previously described
(Go et al., 1999). Papillomas and early-stage SCC (15 ±
25 weeks and 30 ± 40 weeks after DMBA initiation.
respectively) were biopsied. To determine if DbRII
tumors possess aberrant cell cycle progression, ¯ow
cytometry was utilized to identify changes in DNA
content in tumor cells from ML.DbRII and control mice.
Cell populations in normal keratinocytes from neonatal
mice consisted of 88.7+2.0% in the G1 phase,
6.3+0.8% in the S phase, and 4.9+1.8% in the G2/M
phase (Table 1). Consistent with our previous report that
DbRII epidermis exhibited 2.5-fold increase in BrdU
labeling compared to non-transgenic epidermis (Wang et
al., 1997b), DbRII epidermis showed *twofold increase
in the S phase population (14.1+1.5%, Table 1). In
chemically-induced papillomas arising from non-transgenic mice, the S population is 6.3+3.4% (Figure 1,
Table 1) similar to that in neonatal epidermis, which is
*threefold higher than that in the adult epidermis
(Bertsch and Marks, 1982). DbRII papillomas exhibited
*twofold higher S phase (11.7+2.9%, Figure 1, Table
1). In chemically-induced SCC of non-transgenic mice,
the S population increased to 15.8+5.8% Table 1). SCC
in DbRII mice displayed an S population (10.9+4.2%)
similar to control SCC, but a sixfold increase in the G2/
M population (23.3+19%) compared to that of nontransgenic SCC (4.1%, Table 1). Thus, the G1 population in DbRII SCC was reduced to 59.4+18% (Figure 1,
Table 1). Out of seven ML.DbRII SCC analysed, only
one contained tetraploid cells (Figure 1, Table 1).
DbRII papillomas exhibit higher Bromodeoxyuridine
(BrdU) labeling
To further con®rm changes in DNA synthesis in DbRII
tumors, in vivo BrdU labeling was performed on
chemically-induced papillomas and SCC. Since the
BrdU index measures the S phase cells among cells
with proliferative potential (i.e., basal cells in normal
epidermis) instead of the entire tumor cell population,
this method is more sensitive than ¯ow cytometry in
determining the changes in DNA synthesis. The same
size of tumors (5 mm in diameter for papillomas and
SCC) from each group were excised 1 h after BrdU
injection. The labeling index of DbRII papillomas was
61+18 cells/mm basement membrane, (n=5, Figure 2),
threefold higher than that of control papillomas
(20+3.5 cells/basement membrane, n=6, P50.0001,
Figure 2). When papillomas progressed to SCC, BrdU
labeling indices were increased to 35+11 cells/mm
basement membrane for control SCC (Figure 2);
however, BrdU labeling in DbRII SCC (40+9.8 cells/
basement membrane, Figure 2) did not show a further
increase in comparison with either DbRII papillomas
or control SCC. These results are consistent with the
¯ow cytometry pro®les of these tumors. We also
examined the rate of apoptosis in both DbRII and
non-transgenic tumors, but it was not altered (data not
shown).
DbRII tumors exhibit higher mitotic indices
The G2/M phase in ¯ow cytometry does not
distinguish between the cells which are arrested in
Table 1 DNA content of tumor cells determined by ¯ow cytometry
Cell source (ear tag number)
NL epidermis (8)
DbRII epidermis (7)
NL papillomas (6)
DbRII papillomas (6)
NL, SCC:
5029-15
5029-16
5094-9
c8210
Mean+s.d.
DbRII, SCC:
c4731
c6015
c4412
c2847
c2849
c2850
c2859
Mean+s.d.
% G1
%S
% G2/M
88.7+2.0
81.1+1.6*
85.6+5.8
80.0+4.2
6.3+0.8
14.1+1.5*
6.3+3.4
11.7+2.9*
4.9+1.8
4.8+0.6
8.2+3.0
8.3+3.3
77.4
77.6
81.5
83.8
80.1+2.7
22.6
20.2
11.9
8.5
15.8+5.8
0
2.2
6.6
7.7
4.1+3.0
69.2
73.1
73.5
76.4
36.8
29.0
57.7
59.4+18*
8.5
8.6
15.5
6.8
18.2
6.3
12.5
10.9+4.2
22.3
18.3
11
16.8
0
64.7
29.8
23.3+19*
% G1 (T)
% S (T)
% G2/M(T)
30.2
10.7
4.1
(T): Tetraploid cells; NL: non-transgenic controls. Epidermal cells were prepared from neonatal NL and DbRII mice. *P50.05 in comparison
with that of NL in each pair
Oncogene
Aberrant cell cycle progression in tumors expressing DbRII
C Go et al
3625
Figure 1 DNA content pro®les of DMBA/TPA-induced tumor cells. Pap: papillomas: NL: Non-transgenic control. DbRII
papilloma exhibited an increased S phase compared to a NL papilloma. DbRII SCC cells (c2850) exhibited an increased S phase and
a huge G2/M peak. DbRII SCC cells (c2849) exhibited tetraploid populations (peaks at 100 and 200 channels). Hatched area: S
phase
Figure 2 BrdU labeling in DbRII and control tumors. DbRII papillomas showed about a threefold increase in BrdU labeling of
nuclei (yellow or green dots) over that of a non-transgenic control mouse (NL). DbRII SCC exhibited a similar BrdU labeling rate
compared to non-transgenic (NL) SCC. The red color is the K14 counterstain which highlights the epithelial component of tumors.
Bar represents 100 mm
G2 and the cells which have progressed to the
mitotic phase. In addition, although lack of an
obvious histogram of a tetraploid cell population,
the G2/M phase often appears at the same location
as the tetraploid G1 phase. To further determine
whether DbRII tumors exhibit increased mitosis as
well as that they are truly diploid cells, histology and
mitotic indices of tumors were further analysed.
Oncogene
Aberrant cell cycle progression in tumors expressing DbRII
C Go et al
3626
Although DbRII tumors progress to metastasis as
early as tumors harboring a gain-of-function p53
mutant (p53m, (Wang et al., 1998a,b)), histologically
they di€er from p53m tumors. DbRII SCC were welldi€erentiated and usually contained a single nucleus
(Figure 2, (Go et al., 1999)), whereas p53m tumors
were poorly di€erentiated and exhibited multiple
nuclei (tetra- or multi-ploidity) (Wang et al.,
1998a,b). However, DbRII tumors exhibited increased
mitotic indices compared to non-transgenic controls
(Figure 3). The mitotic index in DbRII papillomas
averaged 2.6+0.7 (n=10). which was threefold higher
than papillomas in the control group (0.9+0.3,
n=10, P50.0001). When papillomas further progressed to SCC, the mitotic index in the nontransgenic group increased to 1.24+0.1 (n=10,
P50.05 compared to non-transgenic papillomas).
However, di€erent from BrdU labeling indices that
both control and DbRII SCC exhibited a similar
extent increase, the mitotic index in DbRII SCC was
three-fold higher (4.6+1.5, n=10) than control SCC
(P50.0001 compared to non-transgenic SCC). This
result indicates that some DbRII tumor cells also
escaped from G2 arrest and entered the mitotic
phase.
DbRII tumor did not exhibit a significant increase in
centrosome abnormalities
Although ¯ow cytometry pro®les did not show obvious
aneuploid cell populations in the majority of
ML.DbRII tumors, it is still possible that tumors
contain a small portion of aneuploid cells or aneuploid
cells with a chromosome mode close to 40 (Aldaz et al.,
1987; Conti et al., 1986). Furthermore, ¯ow cytometry
pro®les cannot exclude chromosome instabilities resulting from unequal segregation, breaks and rearrangement of chromosomes. Therefore, we examined
centrosomes in tumors, a more sensitive marker of
chromosome instability (Brinkley, 1985). Because
centrosomes duplicate only once during the cell cycle,
excessive duplication, as indicated by more than two
centrosomes in the cytoplasm, would result in unequal
segregation and breaks of chromosomes (Brinkley,
1985). In non-transgenic papillomas examined, 97%
of tumor cells contained a normal number of
centrosomes (1 or 2), and only 3% of tumor cells
exhibited three centrosomes/cell (Figure 4). DbRII
papillomas showed an increase in centrosome abnormality, with 12% of tumor cells containing three
centrosomes/cell (Figure 4, Table 2). However, 88%
Figure 3 Increased mitotic indices in DbRII tumors. Arrows point out typical and atypical mitotic cells. DbRII papilloma (Pap)
and SCC exhibited increased mitotic cells in comparison with non-transgenic papilloma (NL Pap) and SCC, respectively. Bar
represents 25 mm
Oncogene
Aberrant cell cycle progression in tumors expressing DbRII
C Go et al
3627
Figure 4 Centrosome staining. Arrows point out examples of one or two centrosomes in a non-transgenic papilloma (NL Pap); a
cluster of three centrosomes in one cell of a DbRII papilloma. NL SCC, and DbRII SCC; and multiple centrosomes (43) in the
DbRII papilloma c2849. DbRII SCC c2849 exhibited a centrosome staining pattern similar to DbRII papilloma c2849. Bar
represents 50 mm
Table 2
Centrosome numbers in ML.DbRII and control tumors
Sources of tumors
No. of centrosomes/cell
(% of total cell population)
1
2
53 (abnormal)
Control papillomas (4)
ML.DbRII papillomas (4)
ML.DbRII papilloma c2849
Control SCC (4)
ML.DbRII SCC (4)
ML.DbRII SCC c2849
67
46
17
30
21
42
37
63
36
35
38
64
3
12
47
35
41
50
containing four or more centrosomes were rare. Again,
the DbRII c2849 SCC that exhibited a tetraploid cell
population (Figure 1, Table 1), exhibited abnormal
centrosome numbers in 50% tumor cells, with most
cells containing four or more centrosomes (Figure 4,
Table 2). These data indicate that although both DbRII
and control SCC exhibit chromosome instability,
DbRII SCC did not exhibit a signi®cant increase in
chromosome instability.
Centrosome numbers were counted from 200 cells in each tumor and
averaged from four tumors in each group. ML.DbRII c2849
papilloma and SCC were counted separately
DbRII tumors exhibit reduced p15 and p21 expression,
and loss of p27 protein
of these tumor cells still exhibited normal (1 or 2)
centrosome numbers, and none of the cells contained
54 centrosomes/cell. This result is in sharp contrast to
the previous reports on centrosome numbers in
papillomas containing the p53m transgene, in which
30% of tumor cells showed three centrosomes/cell and
45% contained 54 centrosomes/cell (Wang et al.,
1998a,b). Interestingly, one DbRII papilloma biopsied
from transgenic mouse c2849, showed signi®cant
increase in abnormal centrosome numbers. The
remaining cells of this tumor, which were not observed
grossly, later developed to an SCC and exhibited
aneuploid cell populations (Figure 1, Table 1). In this
case, 47% of the tumor cells showed abnormal
centrosomes, with most of these cells containing more
than four centrosomes (Figure 4, Table 2). When
papillomas progressed to SCC, both control and DbRII
tumors exhibited increased abnormal centrosome
numbers with a similar rate (41% in DbRII SCC and
35% in control SCC, Figure 4, Table 2). Most of these
abnormal cells contained three centrosomes, and cells
It has been reported that TGFb induces growth
inhibition via the elevation of the cdk inhibitors p15
(INK4b) (Hannon and Beach, 1994) and p21 (WAF-1)
(Datto et al., 1995) at the transcription level, and p27
(Kip1) at the post-translation level (Polyak et al.,
1994). To determine whether the DbRII accelerates
carcinogenesis by blocking TGFb-induced elevation of
these cdk inhibitors, RNase protection assay (RPA)
was performed to determine expression levels of p15
(INK4b) and p21 (WAF-1) transcripts in chemicallyinduced tumors of both DbRII and control mice.
Expression of p15 was reduced by 2 ± 3-fold in DbRII
papillomas and 3 ± 4-fold in SCC compared to nontransgenic papillomas and SCC, respectively (Figure
5a). In addition, expression of p16 (INK4a), a family
member of p15, did not show changes in tumors (data
not shown). Expression of p21 in DbRII papillomas
was also reduced by twofold in comparison with nontransgenic papillomas (Figure 5a). While expression of
p21 in DbRII SCC was at a level similar to that in
ML.DbRII papillomas, some of the non-transgenic
SCC also began to show a reduction in p21 expression
(Figure 5a). Western analysis was performed to
Oncogene
Aberrant cell cycle progression in tumors expressing DbRII
C Go et al
3628
Figure 5 (a) Changes in expression of p15 and p21 in DbRII
tumors. NL: non-transgenic control; Pap: papillomas. RNase
protection assays were performed using speci®c riboprobes as
labeled in each panel. Note that DbRII papillomas and SCC
exhibited lower p15 expression levels in comparison with NL
papillomas and SCC, respectively. DbRII papillomas also showed
lower p21 expression compared to NL papillomas. Expression of
p21 in DbRII SCC showed a slightly lower level than DbRII
papillomas, and two of NL SCC also showed decrease in p21
expression compared to NL papillomas. (b) Western analysis of
p27. Note the complete loss of the p27 protein in DbRII
papillomas and SCC as well as in NL SCC
determine p27 protein levels in control and transgenic
tumors. Expression of p27 in DbRII papillomas was
undetectable, whereas papillomas from non-transgenic
mice still retained p27 expression (Figure 5b). Both
DbRII SCC and non-transgenic SCC have exhibited a
complete loss of p27 protein (Figure 5b).
Discussion
The detection of mutations in TGFbRII in human
cancer suggests that inhibition of TGFb signaling may
be an important mechanism in malignant progression
of epithelial tumors. To test this in an in vivo model,
we previously performed chemical carcinogenesis
experiments on ML.DbRII transgenic mice and showed
that blocking TGFb signaling plays a role in
accelerating skin carcinogenesis at both early and late
stages (Go et al., 1999). In the present study we have
focused on the analysis of mechanisms of DbRIIinvolved in the earlier stages of carcinogenesis.
Acceleration of aberrant cell cycle progression appears to
be an important mechanism whereby DbRII accelerates
tumorigenesis
The role of TGFb in cell cycle control has been
suggested to be essential to its tumor suppressive e€ect
(Satterwhite and Moses, 1994). In vitro studies have
demonstrated that TGFb arrests cell cycle progression
in late G1, via the activation of several cyclindependent kinase (cdk) inhibitors (Polyak, 1996;
Satterwhite and Moses, 1994). To determine whether
blocking TGFbRII signaling results in premature cell
cycle progression in vivo, ¯ow cytometry was used to
analyse cell cycle populations of DbRII epidermis and
tumor cells. DbRII epidermis and papillomas showed a
*twofold increase in S phase compared to those of
non-transgenic epidermis and papillomas, respectively.
In SCC, DbRII tumors exhibited a lower G1
population and a sixfold increase in the G2/M
population compared to non-transgenic tumors (Figure
Oncogene
1, Table 1). Consistent with the ¯ow cytometry data,
BrdU labeling was increased *twofold in DbRII
epidermis (Wang et al., 1997b) and *threefold
papillomas (Figure 2). In addition, DbRII tumors also
exhibited increased mitotic indices (Figure 3). These
results indicate that more cells in DbRII tumors have
escaped from G1 arrest and have entered into the G2/
M phase. Although the detailed mechanisms remain to
be determined, the failure in G1 arrest in DbRII
tumors, can be attributed at least in part to the earlier
reduction of p15, p21 and p27 in DbRII papillomas
(Figure 5). All three of these cdk inhibitors are
implicated by in vitro studies to be downstream targets
of TGFb, and mediate TGFb-induced growth inhibition (Hannon and Beach, 1994; Datto et al., 1995;
Polyak et al., 1994). Interestingly, Nagase et al. (1999)
have recently identi®ed two skin tumor resistant alleles
Skts2 and Skts10 which lie near the loci of p21 and
p27, respectively. Our current data further suggest a
tumor suppressive role of these cdk inhibitors in skin
carcinogenesis. Consistent with the report of Nagase et
al. (1999), we observed a loss of expression of p21 and
p27 when non-transgenic papillomas converted to
malignant SCC (Figure 5). At this stage, tumors begin
to lose TGFbR and the signaling Smads (our
unpublished data). Therefore, the loss of endogenous
TGFb signaling at this stage is coincident with the loss
of these two cdk inhibitors. Collectively, our data
provide in vivo evidence that loss of TGFb signaling
results in deregulation of these cdk inhibitors. Therefore, in future studies, it would be of interest to
determine which speci®c cdks are responsible for
DbRII-induced acceleration of cell cycle progression.
The candidates of these cyclin-cdk complexes include
cyclin D-cdk4/6 (Hannon et al., 1994; Datto et al.,
1995; Polyak et al., 1994), cyclinA-cdk2 (Geng and
Weinberg, 1993) and cyclin E-cdk2 (Geng and
Weinberg, 1993; Nagahara et al., 1999). In addition,
it would be worthwhile to examine other upstream
modulators of cdks which are suggested to be TGFb
targets, e.g., Rb (Murphy et al., 1991), myc (Murphy et
al., 1991), and cdc25 (Iavarone and Massague, 1997).
Genetic instability in DbRII tumors may be a consequence
of the accelerated aberrant cell cycle progression
The propensity of DbRII papillomas to undergo a
rapid malignant conversion, suggests a potential role of
DbRII in genomic instability. Genomic instability
occurs in two di€erent forms (Cahill et al., 1998):
microsatellite instability which is associated with an
increased mutation rate but maintains diploidy of
tumor cells; and chromosome instability which induces
aneuploidy of tumor cells. Mutations in TGFbRII
have been shown to occur as a result of microsatellite
instability (Markowitz et al., 1995; Myero€ et al.,
1995). Since chromosome abnormalities are a hallmark
of malignant progression in chemically-induced skin
tumors (Aldaz et al., 1987; Conti et al., 1986), we were
interested in determining whether DbRII tumors
possessed increased chromosome instability. The
DNA content and centrosome numbers of DbRII
tumors were examined in benign papillomas and early
stage SCC. Six of the seven DbRII SCC analysed
showed diploid DNA content, and only one tumor
exhibited a typical aneuploid pro®le (Table 1). In
Aberrant cell cycle progression in tumors expressing DbRII
C Go et al
contrast to our current results, the study by Conti et al.
(1986) showed chromosome banding identi®ed aneuploid cells after 30 or 40 weeks of promotion with a
chromosome mode of 41 ± 47. These aneuploid cells
also formed the stem cell lines of these tumors after
selection in culture (Conti et al., 1986). In our current
study, tumor samples were subjected to ¯ow cytometry
without selection in culture. Therefore, tumors which
contained a small portion of aneuploid cells or
aneuploid cells with a chromosome mode similar to
that in the above study would not be detected by ¯ow
cytometry. We then examined the number of centrosomes in tumor cells, since this is a more sensitive
marker for chromosome instability. Interestingly, most
DbRII papillomas showed normal centrosome numbers
in the majority of tumor cells. Only one DbRII
papilloma that showed aberrant centrosome numbers,
its remaining biopsy developed aneuploid cells in SCC.
This result is in sharp contrast to tumors harboring the
p53m transgene, which possess a high proportion of
cells with multiple centrosome numbers in benign
papillomas (Wang et al., 1998a,b) and aneuploid cells
with a chromosome mode average of 80 to 130 in SCC
(Li et al., 1997, and our unpublished data). Taken
together, results from both of these studies suggest that
centrosome numbers in benign papillomas may be an
indicator to predict whether tumors will develop
chromosome instability. Consistent with previous
reports that malignant skin tumors develop chromosome instabilities (Conti, et al., 1986), both DbRII and
control groups SCC exhibited abnormal centrosome
numbers in 35 ± 40% of cells, although the overall
DNA content pro®les of these tumors still showed
diploid chromosome pro®les. However, DbRII did not
cause a further increase in chromosome instability
compared to non-transgenic tumors. Our current study
is in contrast to a recent report by Glick et al. (1999),
which showed that v-rasHa transduced keratinocytes
which were co-transfected with a DbRII construct
rapidly developed aneuploidy. Therefore, it appears
that the mechanisms involved in regulating genomic
stability in vivo are more complicated, and other tumor
suppressive mechanisms may be activated to compensate for loss of TGFb signaling in DbRII tumors. In
addition, the lack of increase in aneuploidy in DbRII
tumors suggests that DbRII may not directly perturb
G2 or mitotic spindle checkpoints, which is also
re¯ected by the signi®cant increase in the G2/M phase
in DbRII tumors. However, since DbRII tumor cells
have escaped from the DNA damage checkpoint (G1),
accumulation of such damage may eventually increase
chromosome instability. Taken collectively, our current
in vivo data may link DbRII-induced cell cycle
abnormalities and genetic instability together, since
mutations resulting from either subtle chromosomal
damage or microsatellite instability would be further
augmented by premature cell cycle progression.
It is noteworthy that the tumorigenesis kinetics in
ML.DbRII mice is very similar to that of transgenic
mice expressing p53m in the epidermis (Wang et al.,
1998a,b; Go et al., 1999), however the actual molecular
events leading to tumor progression were quite distinct
between these two models. First, in contrast to p53m
tumors that exhibited signi®cant chromosome instability (Wang et al., 1998a,b), ML.DbRII tumor cells
showed a large G2/M population, but rarely aneuploid
cells. Therefore, perturbation of cell cycle control
appears to be a primary mechanism whereby DbRII
accelerates tumorigenesis; whereas chromosome instability may be a direct e€ect of p53m. Secondly,
increased angiogenesis in ML.DbRII tumors correlated
with their earlier metastatic invasion (Go et al., 1999);
whereas p53m tumors did not exhibit an increase in
angiogenesis (Wang et al., 1998b). These results
highlight the importance of using transgenic models
to investigate di€erent molecular mechanisms leading
to distinct alterations in cancer development.
3629
Materials and methods
Chemical carcinogenesis protocols
Ten-week-old ML.DbRII heterozygous mice and their nontransgenic littermates were divided into groups (30 mice in
each group) for topical treatment with DMBA/TPA or
vehicle (acetone) alone. The back skin of each mouse was
carefully shaved 2 days before the treatment. Mice were
treated ®rst with a sub-carcinogenic dose of DMBA (50 mg in
acetone), followed by 20 weeks of TPA, 5 mg (in acetone)/
week, starting at 1 week after DMBA initiation. Tumors
biopsied in this study were removed at 15 ± 25 weeks and 30 ±
40 weeks after DMBA initiation for papillomas and SCC,
respectively.
Tissue histology
Biopsied tumors were ®xed in 10% neutral-bu€ered formalin
at 48C overnight, embedded in paran, sectioned to 6 mm
thickness, and stained with hematoxylin and eosin (H & E).
Flow cytometry of tumor cells
Epidermal keratinocytes from neonatal non-transgenic control and DbRII mice were prepared as previously described
(Wang et al., 1997b). Tumor-bearing mice were sacri®ced and
tumors were excised, cut to 1 ± 2 mm pieces, stirred in
DMEM medium (Biowittaker, MA, USA) containing
0.35% collagenase (GIBCO) at 378C for 30 min. The
supernatant of the collagenase crude preparation, containing
blood, necrotic cells, ®broblasts, and squames was discarded.
Tissue fragments were then stirred twice in 1% trypsin-PBS
(GIBCO) at 378C for 30 min. Tumor cells were collected by
®ltering each trypsinized crude preparation through a 50 mm
nylon mesh, and centrifugation at 200 g for 4 min. Each
trypsinized cell fraction was split into two parts: one was
cultured in 50% ®broblast conditioned medium (Greenhalgh
et al., 1989) with 0.05 mM Ca2+, another was washed with
PBS, ®xed in cold methanol for 10 min, and stored in PBS at
48C for ¯ow cytometry. After con®rming by cell culture that
cells in the trypsinized fraction were tumor cells originating
from keratinocytes, 26106 methanol ®xed cells were stained
with propidium iodide (50 mg/ml) in the presence of 100 mg/
ml RNase A at 378C for 30 min, and analysed by ¯ow
cytometry. The samples were run on a Coulter Electronics
Pro®le II. The propidium iodide was excited with a 488 nm
laser line at 15 mW power. The ¯uorescent emission was
collected in a PMT masked with a 575 nm band pass ®lter.
All samples were gated on Forward Angle Light Scatter and
a mitotic discriminator which consists of the integrated signal
plotted against the peak signal. The resultant events are
displayed on a linear scale so that G2 is in twice the channel
of G0/G1. Normal keratinocytes, isolated from neonatal
epidermis, are used to determine peaks of G0/G1 and G2/M
phase in each assay. Aneuploid tumors met the following
criteria: two individual or asymmetric G0/G1 peaks with two
well-de®ned G2/M peaks. 10 000 events were collected to
Oncogene
Aberrant cell cycle progression in tumors expressing DbRII
C Go et al
3630
provide adequate statistical validity. Histograms were
analysed using the ModFit LT software (Verity Software
House, Topsham, ME, USA).
scanning microscope (Molecule Dynamics Inc.). Four to ®ve
tumors in each group were examined, and 200 cells in each
tumor were counted for centrosome numbers.
BrdU uptake and staining
RNase protection assays
Tumor-bearing mice were injected (i.p.) with BrdU 125 mg/
kg in 0.9% NaCl. The same size of DMBA/TPA-induced
tumors from ML.DbRII and non-transgenic control mice
were biopsied 1 h after the injection, ®xed and processed as
previously described (Greenhalgh et al., 1996). Sections were
incubated with FITC-conjugated monoclonal antibody to
BrdU (Becton-Dickinson) and a guinea-pig antiserum to
mouse keratin 14 (K14), which reacts with the epithelial
component of papillomas. K14 was visualized by biotinylated
anti-guinea-pig IgG (Vector) and Streptavidin-Texas Red
(GIBCO). BrdU labeled cells were counted under 106ocular
and 206objective lenses. Consecutive ®elds were counted
throughout the entire tissue section. The labeling index was
expressed as the mean of BrdU-positive cells/mm basement
membrane+s.d.
Total RNA was isolated from chemically-induced tumors
with RNAzol B (Tel-Test, Inc.) as previously described
(Wang et al., 1998b). RNase protection assays (RPA) were
performed using the RPA II kit (Ambion, TX, USA) and 32Plabeled riboprobes. Riboprobe speci®c for p21 (WAF-1) was
generated as previously described (Wang et al., 1998a). The
SmaI/SacII fragment of murine p15 (INK4b) cDNA (kindly
provided by Dr Charles J Sherr) was subcloned into pGem 5
(Promega), and linearized with ApaI as the template of the
p15 riboprobe. The XhoI fragment of murine p16 (INK4a)
cDNA (provided by Dr Charles J Sherr) was subcloned into
the SalI site of pGem 5, and linearized at the internal AsuII
site of p16, to produce a 400 bp fragment of the p16
riboprobe. The intensity of protected bands was determined
by densitometric scanning of X-ray ®lms.
Apoptosis assays
Western analysis for p27
Apoptosis was evaluated using a Deoxynucleotidetransferase
Uridine End Labeling (TUNEL) assay with the TACS kits
from Trevigen Inc. (Gaithersburg, MD, USA). Tumors were
®xed in 10% neutral bu€ered formalin, sectioned, and
visualized by diaminobenzidine.
Total proteins of tumors were extracted as described (Wang
et al., 1997b). An equal amount of protein from each sample
was separated by electrophoresis on a 12.5% SDS ± PAGE
gel and transferred onto a nitrocellulose membrane. The blot
was blocked by 5% non-fat milk in T-TBS (0.1% Tween20TBS), at room temperature for 1 h. The goat anti-mouse p27
polyclonal antibody (Santa Cruz) was applied to the blot
with 0.4 mg/ml in the blocking bu€er, at 48C for overnight.
The blot was washed twice with T-TBS for 10 min, and
incubated with 1 : 500 peroxidase conjugated rabbit anti-goat
IgG (Sigma) at room temperature for 1 h. After washing as
described above, the ECL detection system (Amersham) was
used to visualize p27. To con®rm equal loading, the blot was
stripped with ImmunoPure IgG Elution Bu€er (Pierce,
Rockford, IL, USA) and re-screened with a sheep anti-mouse
K14 antibody.
Mitotic Index (MI)
Criteria and the counting method for mitotic cells were used
as described by van Diest et al. (1992). Brie¯y, 106ocular
and 406objective lenses of a light microscope were used
(high power ®elds, HPFs). In the counting area mitotic
®gures were counted in 10 consecutive HPFs starting at the
spot within the counting area with the highest density of
mitotic ®gures, avoiding ®elds with necrosis or in¯ammation
and ®elds containing less than 50% tumor cells. The average
number of mitotic cells/HPFs from 10 HPFs is determined as
MI. The MI for each group (10 tumors) was averaged from
individual MI.
Centrosome localization
Frozen tumors were sectioned to 30 mm thickness and ®xed
with 3.7% formaldehyde after 2 min treatment with 0.5%
Triton X-100. Samples were then incubated with human
autoantiserum PCM-1, that recognizes centrosomes in
mammalian cells (a gift from Dr Ron Balczon), and rabbit
anti-keratin K14, which reacts with the epithelial component
of tumors, followed by incubation with FITC-conjugated
goat anti-human IgG and Texas Red conjugated goat antirabbit IgG. The sections were then visualized using a laser
Acknowledgments
The authors would like to give special thanks to Dr Dennis
R Roop for his critical comments on the manuscript. We
would also like to thank Dr Charles J Sherr (St Jude
Children Hospital, Memphis, TN, USA) for providing
mouse p15 and p16 cDNA clones, and Dr Ron Balczon
(University of South Alabama, Mobile, AL, USA) for
providing human antiserum for PCM-1. Mrs Mattie
Brooks provided excellent technical assistance. This work
was supported by a Career Development Award from the
Dermatology Foundation and NIH grant CA79998 to X-J
Wang and NIH grant CA64255 to BR Brinkley.
References
Aldaz CM, Conti CJ, Klein-Szanto AJ and Slaga TJ. (1987).
Proc. Natl. Sci. USA, 84, 2029 ± 2032.
Bertsch S and Marks F. (1982). Cell Tissue Kinet., 15, 81 ± 87.
Bottinger EP, Jakubczak JL, Roberts IS, Mumy M,
Hemmati P, Bagnall K, Merlino G and Wake®eld LM.
(1997). EMBO J., 16, 2621 ± 2633.
Brinkley BR. (1985). Ann. Rev. Cell. Biol., 1, 145 ± 172.
Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JV,
Markowitz SD, Kinzler KW and Vogelstein B. (1998).
Nature, 392, 300 ± 303.
Carcamo J, Zentella A and Massague J. (1995). Mol. Cell.
Biol., 15, 1573 ± 1581.
Oncogene
Chen F and Weinberg RA. (1995). Proc. Natl. Acad. Sci.
USA, 92, 1565 ± 1569.
Chen RH, Ebner R and Derynck R. (1993). Science, 260,
1335 ± 1338.
Conti CJ, Aldaz CM, O'Connell J, Klein-Szanto AJ and
Slaga TJ. (1986). Carcinogenesis, 7, 1845 ± 1848.
Datto MB, Li Y, Panus JF, Howe DJ, Xiong Y and Wang
XF. (1995). Proc. Natl. Acad. Sci. USA, 92, 5545 ± 5549.
Derynck R. (1994). Trends Biochem. Sci., 19, 548 ± 553.
Feng XH, Filvaro€ EH and Derynck R. (1995). J. Biol.
Chem., 270, 24237 ± 24245.
Aberrant cell cycle progression in tumors expressing DbRII
C Go et al
Filmus J and Kerbel RS. (1993). Curr. Opin. Oncol., 5, 123 ±
129.
Fynan TM and Reiss M. (1993). Crit. Rev. Oncog., 4, 493 ±
540.
Geng Y and Weinberg RA. (1993). Proc. Natl. Acad. Sci.
USA, 90, 10315 ± 10319.
Glick A, Popescu N, Alexander V, Ueno H, Bottinger E and
Yuspa SH. (1999). Proc. Natl. Acad. Sci. USA, 96, 14949 ±
14954.
Go C, Li P and Wang XJ. (1999). Cancer Res., 59, 2861 ±
2868.
Gorska AE, Joseph H, Derynck R, Moses HL and Serra R.
(1998). Cell Growth Di€er, 9, 229 ± 238.
Greenhalgh DA, Welty DJ, Strickland JE and Yuspa SH.
(1989). Mol. Carcinog., 2, 199 ± 207.
Hannon GJ and Beach D. (1994). Nature, 371, 257 ± 261.
Iacopetta BJ, Welch J, Soong R, House AK, Zhou XP and
Hamelin R. (1998). J. Pathol., 184, 390 ± 395.
Iavarone A and Massague J. (1997). Nature, 387, 417 ± 422.
Imai Y, Tsurutani N, Oda H, Inoue T and Ishikawa T.
(1998). Int. J. Cancer, 76, 407 ± 411.
Li B, Rosen JM, McMenamin-Balano J, Muller WJ and
Perkins AS. (1997). Mol. Cell. Biol., 17, 3155 ± 3163.
Lu SL, Akiyama Y, Nagasaki H, Saitoh K and Yuasa Y.
(1995). Biochem. Biophys. Res. Commun., 216, 452 ± 457.
Lu SL, Kawabata M, Imamura T, Akiyama Y, Nomizu T,
Miyazono K and Yuasa Y. (1998). Nat. Genet., 19, 17 ± 18.
Markowitz S, Wang J, Myero€ L, Parsons R, Sun L,
Lutterbaugh J, Fan RS, Zborowska E, Kinzler KW and
Vogelstein B. (1995). Science, 268, 1336 ± 1338.
Murphy CS, Pietenpol JA, Munger K, Howley PM and
Moses HL. (1991). Cold Spring Harb. Symp. Quant. Biol.,
56, 129 ± 135.
Myero€ LL, Parsons R, Kim SJ, Hedrick L, Cho KR, Orth
K, Mathis M, Kinzler KW, Lutterbaugh J and Park K.
(1995). Cancer Res., 55, 5545 ± 5547.
Nagahara H, Ezhevsky SA, Vocero-Akbani AM, Kaldis P,
Solomon MJ and Dowdy SF. (1999). Proc. Natl. Acad.
Sci. USA, 96, 14961 ± 14966.
Nagase H, Mao JH and Balmain A. (1999). Proc. Natl. Acad.
Sci. USA, 96, 15032 ± 15037.
Polyak K. (1996). Biochim. Biophys. Acta., 1242, 185 ± 199.
Polyak K, Kato JY, Solomon MJ, Sherr CJ, Massague J,
Roberts JM and Ko€ A. (1994). Genes Dev., 8, 9 ± 22.
Satterwhite DJ and Moses HL. (1994). Invasion Metastasis,
14, 309 ± 318.
Takenoshita S, Tani M, Nagashima M, Hagiwara K, Bennett
WP, Yokota J and Harris CC. (1997). Oncogene, 14,
1255 ± 1258.
van Diest PJ, Baak JP, Matze-Cok P, Wisse-Brekelmans EC,
van Galen CM, Kurver PH, Bellot SM, Fijnheer J, van
Gorp LH and Kwee WS. (1992). Hum. Pathol., 23, 603 ±
607.
Venkatasubbarao K, Ahmed MM, Swiderski C, Harp C, Lee
EY, McGrath P, Mohiuddin M, Strodel W and Freeman
JW. (1998). Genes Chromosomes Cancer, 22, 138 ± 144.
Wang D, Song H, Evans JA, Lang JC, Schuller DE and
Weghorst CM. (1997a). Carcinogenesis, 18, 2285 ± 2290.
Wang XJ, Greenhalgh DA, Bickenbach JR, Jiang A,
Bundman DS, Krieg T, Derynck R and Roop DR.
(1997b). Proc. Natl. Acad. Sci. USA, 94, 2386 ± 2391.
Wang XJ, Greenhalgh DA, Jiang AB, He D, Zhong L,
Medina D, Brinkley BR and Roop DR. (1998a).
Oncogene, 17, 35 ± 45.
Wang XJ, Greenhalgh DA, Jiang A, He D, Zhong L, Medina
D, Brinkley BR and Roop DR. (1998b). Mol. Carcinog.,
23, 185 ± 192.
Wieser R, Attisano L, Wrana JL and Massague J. (1993).
Mol. Cell Biol., 13, 7239 ± 7247.
Wrana JL, Attisano L, Wieser R, Ventura F and Massague J.
(1994). Nature, 370, 341 ± 347.
3631
Oncogene