MOLECULAR AND CELLULAR BIOLOGY, Sept. 2010, p. 4159–4174
Copyright © 2010, American Society for Microbiology. All Rights Reserved.
Vol. 30, No. 17
Opposing Roles of Dnmt1 in Early- and Late-Stage Murine Prostate Cancer?
Shannon R. Morey Kinney,1† Michael T. Moser,1Marien Pascual,2John M. Greally,2
Barbara A. Foster,1and Adam R. Karpf1*
Department of Pharmacology and Therapeutics, Roswell Park Cancer Institute, Buffalo, New York 14263,1and
Departments of Genetics (Computational Genetics) and Medicine, Albert Einstein College of Medicine,
1300 Morris Park Avenue, Bronx, New York 104612
Received 28 February 2010/Returned for modification 17 April 2010/Accepted 14 June 2010
Previous studies have shown that tumor progression in the transgenic adenocarcinoma of mouse
prostate (TRAMP) model is characterized by global DNA hypomethylation initiated during early-stage
disease and locus-specific DNA hypermethylation occurring predominantly in late-stage disease. Here, we
utilized Dnmt1 hypomorphic alleles to examine the role of Dnmt1 in normal prostate development and in
prostate cancer in TRAMP. Prostate tissue morphology and differentiation status was normal in Dnmt1
hypomorphic mice, despite global DNA hypomethylation. TRAMP; Dnmt1 hypomorphic mice also dis-
played global DNA hypomethylation, but were characterized by altered tumor phenotype. Specifically,
TRAMP; Dnmt1 hypomorphic mice exhibited slightly increased tumor incidence and significantly in-
creased pathological progression at early ages and, conversely, displayed slightly decreased tumor inci-
dence and significantly decreased pathological progression at advanced ages. Remarkably, hypomorphic
Dnmt1 expression abrogated local and distant site macrometastases. Thus, Dnmt1 has tumor suppressor
activity in early-stage prostate cancer, and oncogenic activity in late stage prostate cancer and metastasis.
Consistent with the biological phenotype, epigenomic studies revealed that TRAMP; Dnmt1 hypomorphic
mice show dramatically reduced CpG island and promoter DNA hypermethylation in late-stage primary
tumors compared to control mice. Taken together, the data reveal a crucial role for Dnmt1 in prostate
cancer and suggest that Dnmt1-targeted interventions may have utility specifically for advanced and/or
metastatic prostate cancer.
Changes in DNA methyltransferase (Dnmt) expression and
DNA methylation are observed in human prostate cancer (3,
38, 41). Of particular interest, genes with tumor suppressive
function become hypermethylated and silenced, which corre-
lates with the development of specific disease phenotypes (2, 3,
38). Although an association between prostate cancer and al-
terations in DNA methylation has been established, in vivo
models are required to determine whether these changes func-
tionally contribute to the disease. In this context, studies in
which pharmacological inhibitors of Dnmts were shown to
inhibit prostate cancer in murine models have proven infor-
mative (34, 56). However, it remains unknown whether genetic
disruption of epigenetic components, such as Dnmts, also im-
pacts prostate cancer development. This is a critical question
since the pharmacological inhibitors of Dnmts have pleiotropic
effects, including those unrelated to activation of methylation-
silenced genes (21, 23, 31). Moreover, no studies to date have
examined whether Dnmts or DNA methylation play roles in
normal prostate development; this information is vital to fully
understanding the effects that inhibiting DNA methylation
may have on prostate cancer.
Dnmt1 is a maintenance DNA methyltransferase that prop-
agates preexisting DNA methylation patterns in genomic DNA
(44). Dnmt1 also is involved in de novo DNA methylation in
cancer cells and interacts with other key epigenetic control
molecules, including histone-modifying enzymes (11, 19). Mu-
rine models have been used to investigate the in vivo functions
of Dnmt1. Complete genetic knockout of Dnmt1 is embryonic
lethal in mice (29). However, hypomorphic expression of
Dnmt1 allows murine development to proceed but causes
global DNA hypomethylation and impacts cancer development
and progression (7, 14, 28). Specifically, hypomorphic expres-
sion of Dnmt1 can lead to the development of lymphoma (14).
Furthermore, crossing Dnmt1 hypomorphic mice with murine
tumor models alters tumor progression, resulting in either in-
creased or decreased tumor development, depending on the
disease stage and tissue site (1, 7, 53). For example, reduced
expression of Dnmt1 dramatically decreases intestinal polyp
formation in ApcMin/?mice, either alone or in combination
with 5-aza-2?-deoxycytidine treatment (7, 27). However, it was
later noted that reduced expression of Dnmt1 has a dual effect
on intestinal cancer in ApcMin/?mice, in which the develop-
ment of early stage intestinal microadenomas is accelerated,
whereas the formation of adenomatous polyps is significantly
reduced (53). In addition, ApcMin/?Dnmt1 hypomorphic mice
develop liver cancer associated with the loss of heterozygosity
of Apc (53). Similarly, in Dnmt1 hypomorphic mice crossed to
Mlh1?/?mice, a dual effect was noted wherein mice developed
fewer intestinal cancers but displayed increased T- and B-cell
lymphomas (52). In addition, a recent study demonstrated that
hypomorphic Dnmt1 expression is associated with reduced
squamous cell carcinoma of the tongue and esophagus, result-
ing in decreased invasive cancer (1). Taken together, the data
* Corresponding author. Mailing address: Roswell Park Cancer In-
stitute, Elm and Carlton Streets, Buffalo, NY 14263. Phone: (716)
845-8305. Fax: (716) 845-8857. E-mail: adam.karpf@RoswellPark.org.
† Present address: New England Biolabs, 240 County Road, Ipswich,
?Published ahead of print on 28 June 2010.
suggest that Dnmt1 has diverse effects on cancer development,
which are dependent on tissue context and tumor stage.
TRAMP is a well-established transgenic prostate cancer
model driven by prostate-specific expression of the simian virus
40 (SV40) T/t oncogenes (16). TRAMP mice are characterized
by Dnmt mRNA and protein overexpression, altered DNA
methylation, and altered gene expression during prostate can-
cer development (2, 33, 35, 37). Of the three enzymatically
active Dnmts, Dnmt1 shows the greatest level of overexpres-
sion in TRAMP, and this correlates with Rb inactivation, a key
genetic event driving prostate cancer in the model (37). Most
critically, global DNA hypomethylation occurs during early
and late disease stages, while DNA hypermethylation occurs
primarily at late disease stages in TRAMP (35).
Here, we utilized Dnmt1 hypomorphic mice and the
TRAMP model to assess the role of DNA methylation in
both normal prostatic development and prostate cancer.
The Dnmt1 hypomorphic mouse model used involves two
different hypomorphic alleles (N and R), resulting in four
genotypes with progressively reduced DNA methylation
(Dnmt1?/?, Dnmt1R/?, Dnmt1N/?, and Dnmt1N/R) (7, 52).
The N allele consists of a PGK-Neo insertion that deletes a
portion of exon 4 of Dnmt1, resulting in severely reduced
Dnmt1 expression, while the R allele involves a lacO inser-
tion into intron 3 of Dnmt1, which partially reduces Dnmt1
expression (7, 52). Based on our previous work establishing
the timing of DNA hypomethylation and DNA hypermeth-
ylation in TRAMP, we hypothesized that hypomorphic
Dnmt1 expression in TRAMP may have tumor-promoting
effects at early disease stages and tumor-inhibitory effects at
later stages of prostate cancer progression. Our data are
consistent with this hypothesis and, more importantly, re-
veal a critical and unanticipated role for Dnmt1 in prostate
MATERIALS AND METHODS
Dnmt1 hypomorphic mice and TRAMP; Dnmt1 hypomorphic mice. Animal
studies were carried out under IACUC-approved protocols at Roswell Park
Cancer Institute. Dnmt1 hypomorphic mice were provided by Peter Laird (Uni-
versity of Southern California [USC] Norris Cancer Center) and have been
described previously (7, 52). C57BL/6 Dnmt1 hypomorphic mice carrying one
hypomorphic allele and one wild-type (WT) (?) allele (Dnmt1N/?or Dnmt1R/?)
were mated to produce offspring that are C57BL/6 Dnmt1?/?(WT), Dnmt1R/?,
Dnmt1N/?, or Dnmt1N/R. Body, prostate, and urogenital tract (UG) weights were
measured at euthanasia. To produce TRAMP; Dnmt1 hypomorphic mice that are
50:50 C57BL/6-FVB, C57BL/6 Dnmt1R/?males were backcrossed to FVB ho-
mozygous TRAMP females for four generations, resulting in 93.75% FVB ho-
mozygous TRAMP; Dnmt1R/?mice. The resulting mice were crossed with
C57BL/6 Dnmt1N/?mice to produce offspring that are 53.1:46.9 C57BL/6-FVB
TRAMP; Dnmt1 hypomorphic mice (of all four genotypes). Body, prostate, and
UG weights were measured at euthanasia. Genotyping for Dnmt1 and Tag alleles
was completed on tail snip DNA extracted by using a Puregene DNA extraction
kit (Gentra Systems, Minneapolis, MN), as described previously (7, 16). Primer
sequences are available from the authors upon request.
qRT-PCR. RNA samples were extracted and used for quantitative real-time
reverse transcription-PCR (qRT-PCR) analysis as described previously (35). The
following genes were analyzed: Dnmt1, Dnmt3a, Dnmt3b, and 18S rRNA. SYBR
green absolute quantification was used to determine mRNA copy number, fol-
lowing normalization to 18S rRNA. PCRs were performed in triplicate. Primers
were obtained from IDT (Coralville, IA) and were designed by using Primer3 or
were previously reported (28, 46, 48). Primer sequences are available from the
authors upon request.
Western blot analyses. Western blotting for Dnmt proteins was completed as
described previously (35, 37). Ponceau S total protein stain was used to confirm
equivalent protein loading.
Global DNA methylation analyses and mass array quantitative DNA methyl-
ation analysis (MAQMA). Quantification of 5-methyl-deoxycytidine (5mdC) lev-
els was determined by liquid chromatography-electrospray ionization quadrapole
mass spectrometry, as described previously (35, 50). Bisulfite pyrosequencing of
the murine B1 repetitive DNA element was done as described previously (35).
The mean methylation value of all sites was averaged for each sample. MAQMA
was used to determine the methylation status of Irx3, Cacna1a, Cdkn2a, and
Nrxn2 as described previously (2, 10, 36).
HELP. HpaII tiny fragment enrichment by ligation-mediated PCR (HELP)
analysis was performed as described previously (24, 39, 51). Six prostate samples
were analyzed: (i) Dnmt1?/?(24 weeks), (ii) Dnmt1N/R(24 weeks), (iii) TRAMP;
Dnmt1?/?(15 and 24 weeks), and (iv) TRAMP; Dnmt1N/R(15 and 24 weeks). To
obtain representative samples from each experimental group, Dnmt1?/?and
Dnmt1N/Rprostate samples were selected based on those that were closest to the
mean prostate weight and mean 5mdC level, for the corresponding genotype.
Similarly, TRAMP; Dnmt1?/?and TRAMP; Dnmt1N/Rprostate samples from
15-week-old mice were selected based on those that were closest to the median
prostate weight and the mean 5mdC level for each genotype. TRAMP; Dnmt1?/?
and TRAMP; Dnmt1N/Rprostate samples from 24-week-old mice were selected
based on those that were closest to the median disease score/median prostate
weight and the mean 5mdC level for each genotype. Sequence features, including
CpG islands and repetitive elements, were defined using the UCSC genome
browser annotations for the mm8 mouse genome assembly, the same database
used for designing the microarrays. The annotated sequences were then cross-
correlated with the HELP data. We focused on the annotated RefSeq genes,
defining promoters as the ?1-kb region flanking the transcription start site.
H&E and IHC staining. Five micron tissue sections were cut from paraffin-
embedded blocks and mounted on slides. Slides were deparaffinized with xylene,
rehydrated with alcohol, and equilibrated with Tris-phosphate buffer. Samples
were stained with hematoxylin and eosin (H&E) or incubated with SV40 Tag
antibody (1:400; catalog no. 554149; BD Pharmingen), AR (1:200; catalog no.
06-680; Millipore), p63 (1:100; catalog no. SC-8431; Santa Cruz), E-cadherin
(1:500; catalog no. 610181; BD Pharmingen), smooth muscle actin (1:60; catalog
no. V2258-0.2ML; Sigma-Aldrich), or Dnmt1 (1:50; catalog no. NB 100-264;
Novus Biologicals) for immunohistochemistry (IHC), according to standard
methods. Negative IHC controls utilized bovine serum albumin buffer alone,
without primary antibody, during the primary antibody staining step. Tissue
sections were scored for tumor grade using a compound Olympus XI-50 micro-
scope equipped with QCapture imaging software. Two independent slides of
each sample were analyzed.
Tumor incidence and tumor pathology measurements. Primary prostate tumor
and macrometastatic tumor incidence were determined at necropsy by using a
dissecting microscope. To measure prostate tumor stage, two slides of H&E-
stained tissue sections 50 ?m apart were analyzed for each mouse at 12, 15, and
24 weeks of age. The disease stage was assigned as described previously (20, 35).
For comparison of the pathological grades, a disease index was calculated from
the percentage of each pathological stage determined for the dorsal, lateral, and
ventral prostate lobes according to the following formula: disease index ? (%
normal ? 0) ? (% PIN ? 1) ? (% well-differentiated [WD] ? 2) ? (%
moderately differentiated [MD] ? 3) ? (% poorly differentiated [PD] ? 4). One
disease index value was calculated for each mouse based on the arithmetic
average of the three prostate lobes, after averaging for the two slides ana-
lyzed. IHC of the SV40 tag was used to assess micrometastatic tumor inci-
dence. Two slides (50 ?m apart) of Tag-stained lymph node, kidney, lung,
and liver tissue sections were analyzed for each mouse at 12, 15, and 24 weeks
Statistics. Statistical analyses were performed by using the GraphPad Prism v5
(GraphPad Software, San Diego, CA). To test for differences between two
groups with nonnormal distributions, we used a two-tailed nonparametric Mann-
Whitney test. To test for differences in tumor incidence, we used the Fisher exact
test. Disease index scores were compared by using the Mann-Whitney test. The
figure legends contain specific information regarding the statistical analyses used
for the indicated experiments.
Prostate development in Dnmt1 hypomorphic mice. Survival
defects in Dnmt1N/Rhypomorphic mice have not been previ-
ously reported (7, 52). However, in our experiments, only
4160 MOREY KINNEY ET AL.MOL. CELL. BIOL.
?20% of the expected amount of Dnmt1N/Rmice were gener-
ated, whereas Dnmt1R/?and Dnmt1N/?mice were generated
at expected levels (Table 1). These data suggest that Dnmt1N/R
mice are near a critical threshold of DNA methylation re-
quired to sustain mouse development. To determine whether
reduced Dnmt1 expression affects normal prostate develop-
ment, we examined prostate weight and glandular morphology
using H&E staining and the expression of several prostate
differentiation markers by using IHC.
Dnmt1N/Rmice showed decreased body weights at 15 and 24
weeks and reduced prostate weights at 15 weeks of age (Fig.
1A and B). However, following normalization to body weight,
the prostate weights of mice from all four genotypes were
similar (Fig. 1C). Thus, Dnmt1N/Rmice appear to have a gen-
eral survival and growth defect that does not specifically
affect the prostate. Consistent with our findings, other
strains of Dnmt1 hypomorphic mice also have reduced body
weight (1, 14).
Dnmt1 hypomorphic mouse prostates displayed a normal
glandular morphology and general appearance (Fig. 2A). In
addition, Dnmt1 hypomorphic mice showed typical patterns of
expression of several prostate differentiation markers, includ-
ing p63 (basal cells), smooth muscle actin, androgen receptor
(AR; luminal epithelial cells), and E-cadherin (luminal cells)
(Fig. 2B). Dnmt1N/Rmice showed a slight increase in cytosolic
AR staining, but nuclear AR staining was retained (Fig. 2B).
Male Dnmt1 hypomorphic mice of all genotypes were fertile,
suggesting that prostate function in these mice is normal (data
not shown). Taken together, the data suggest that the prostates
of surviving Dnmt1 hypomorphic mice develop normally.
Dnmt expression and DNA methylation in Dnmt1 hypomor-
phic mice. To define the Dnmt expression phenotype of Dnmt1
hypomorphic mice, we measured Dnmt1, Dnmt3a, and Dnmt3b
mRNA expression in the prostate and, as an additional control,
in liver tissue by qRT-PCR. We analyzed mRNA as Dnmt
protein expression is virtually undetectable in normal mouse
prostate (35, 37). Dnmt1 mRNA expression was reduced in
both the prostates and the livers of Dnmt1 hypomorphic mice,
with the greatest decline in Dnmt1N/Rmice (Fig. 3A and data
not shown). In contrast, Dnmt3a and Dnmt3b displayed vari-
able expression patterns, with a decrease in Dnmt3b expression
in the prostate of Dnmt1N/Rmice and upregulation of Dnmt3a
and Dnmt3b in the liver of Dnmt1N/Rmice at 24 weeks (Fig. 3A
and data not shown).
The reduced expression of Dnmt1 in the prostate of
Dnmt1N/Rmice suggests that DNA methylation may be altered
in these mice. To test this, we first measured global 5mdC
levels and methylation of the B1 repetitive element, as de-
scribed previously (35, 50). Both measures of global DNA
methylation were reduced in the prostates and livers of
Dnmt1N/Rmice but not in the other hypomorphic genotypes
(Fig. 3B and C and data not shown). To more comprehensively
evaluate DNA methylation in the Dnmt1N/Rmouse prostate,
we utilized the HpaII tiny fragment enrichment by ligation-
mediated PCR assay (HELP), an epigenomic approach (24).
We used HELP to compare DNA methylation profiles of pros-
tates from Dnmt1?/?and Dnmt1N/Rmice at 24 weeks of age.
The proportions of hypermethylated HELP fragments in
Dnmt1?/?and Dnmt1N/Rmice were 78.7 and 76.7%, respec-
tively, indicating a slight DNA hypomethylation effect (Fig.
3D). Consistent with this, a small loss in methylation occurred
at gene bodies and repetitive DNA elements (Fig. 3D). Sur-
prisingly, however, increased methylation was seen at promot-
ers and CpG islands (Fig. 3D). This unexpected increase at
normally hypomethylated regions suggests a compensatory
DNA hypermethylation response in the Dnmt1N/Rmouse pros-
tate, possibly involving Dnmt3 enzymes. Analysis of hypom-
ethylated genes identified by HELP in Dnmt1N/Rmice revealed
a number of genes previously found to be regulated by DNA
FIG. 1. Body and prostate weights of Dnmt1 hypomorphic mice.
(A) The body weights of 15- and 24-week-old mice of the indicated
genotypes were determined at euthanasia. (B) The prostate weights of
15- and 24-week-old mice of the indicated genotypes were determined
at euthanasia. (C) Prostate weight normalized to body weight. Error
bars indicate the standard error. Mann-Whitney test P values of sig-
nificant differences (P ? 0.05), compared to Dnmt1?/?mice, are
shown. In all panels, 8 to 15 mice were analyzed per group.
TABLE 1. Inheritance of Dnmt1 hypomorphic allelesa
GenotypeNo. of mice
aF1offspring from cross of Dnmt1R/?and Dnmt1N/?C57BL/6 mice.
VOL. 30, 2010Dnmt1 AND MURINE PROSTATE CANCER 4161
methylation, including Sohlh2, MyoG, PRAME, and p1a (17,
32, 47, 49) (Table 2). Some of these genes can be classified as
cancer-germ line antigens, a gene family methylated in most
normal tissues (6). In addition, HELP identified a number of
genes not previously known to be regulated by DNA meth-
ylation (Table 2). These data point to a complex DNA meth-
ylation phenotype in the Dnmt1N/Rmouse prostate, character-
ized by global and gene specific DNA hypomethylation but also
by DNA hypermethylation at specific genomic regions.
Generation of TRAMP; Dnmt1 hypomorphic mice and ana-
lysis of Dnmt expression. Dnmt1 hypomorphic mice are
C57BL/6 (7, 52). In TRAMP, 50:50 C57BL/6-FVB mice
present with tumors contained within the prostate (a finding
more reflective of the human disease), rather than spreading to
the seminal vesicles, which is observed in the pure C57BL/6 back-
ground (20). Thus, to generate TRAMP; Dnmt1 hypomorphic
mice in the 50:50 genetic background, we backcrossed Dnmt1
hypomorphic mice to homozygous FVB TRAMP mice (see Ma-
terials and Methods). To minimize potential differences in tumor
pathology due to strain effects, in all analyses we only compared
age-matched littermates of the four TRAMP; Dnmt1 hypomor-
phic genotypes: (i) TRAMP; Dnmt1?/?, (ii) TRAMP; Dnmt1R/?,
(iii) TRAMP; Dnmt1N/?, and (iv) TRAMP; Dnmt1N/R. We col-
lected samples from ?20 animals per genotype per time point.
Due to the limited size of the prostate, at early ages (12 and 15
weeks), prostate tissues from half of the mice were embedded for
histological analyses, and prostate tissues from the remaining
mice were frozen for molecular analyses. At 24 weeks, prostate
molecular studies; at this time point, individual tissues were usu-
ally large enough to use for both types of analyses.
To confirm specific knockdown of Dnmt1 in the target tis-
sue, we measured Dnmt1, Dnmt3a, and Dnmt3b mRNA and
protein expression in the prostates of TRAMP; Dnmt1 hypo-
morphic mice at 12, 15, and 24 weeks. Dnmt1 mRNA expres-
sion was significantly decreased in TRAMP; Dnmt1N/Rmice at
all three time points, whereas its levels were more variable in
the TRAMP; Dnmt1R/?and TRAMP; Dnmt1N/?mice (data not
shown). Dnmt3a and Dnmt3b mRNA expression was inconsis-
tent, with the only significant change an increase in Dnmt3a in
TRAMP; Dnmt1N/?mice at 24 weeks (data not shown). Similar
to Dnmt1 mRNA expression, Dnmt1 protein levels were de-
creased in TRAMP; Dnmt1 hypomorphic mice (Fig. 4A and B).
Although the most dramatic effect occurred in TRAMP;
Dnmt1N/Rmice, significant decreases were also observed in
TRAMP; Dnmt1R/?mice and TRAMP; Dnmt1N/?mice at a
minimum of one time point (Fig. 4B). IHC staining in prostate
tissue confirmed the reduction of Dnmt1 expression in
TRAMP; Dnmt1N/Rmice, especially in poorly differentiated
regions (data not shown). In contrast to Dnmt1, the Dnmt3a
and Dnmt3b protein levels did not show a consistent pattern of
FIG. 2. Prostate tissue architecture and differentiation marker expression in Dnmt1 hypomorphic mice. (A) H&E staining of lateral
prostates from 15-week-old mice of the indicated genotypes was performed as described in Materials and Methods. (B) IHC staining of
markers of prostate differentiation was performed on lateral prostates from 15-week-old mice of the indicated genotypes. The antibodies and
conditions are described in Materials and Methods, and the negative control is shown at the right. Staining is shown for p63 (basal cell
marker), smooth muscle actin (SMA; marker of the muscular layer surrounding glands), androgen receptor (AR; marker for luminal
epithelial cells, with predominantly nuclear staining), and E-cadherin (marker for luminal epithelial cells, with predominantly plasma
membrane staining). Scale bar, 100 ?m.
4162 MOREY KINNEY ET AL.MOL. CELL. BIOL.
altered expression in TRAMP; Dnmt1 hypomorphic mice (Fig.
4C and D).
Since Dnmt expression is aberrant in TRAMP tumors as a
function of the disease (35, 37), we additionally utilized liver as
a control tissue to measure Dnmt expression in TRAMP;
Dnmt1 hypomorphic mice. Similar to the effects observed in
the prostate, Dnmt1 mRNA expression was significantly de-
creased in the liver of TRAMP; Dnmt1 hypomorphic mice,
whereas there were no consistent changes in Dnmt3a or
Dnmt3b expression in this tissue (data not shown).
DNA methylation in TRAMP; Dnmt1 hypomorphic mice. To
assess DNA methylation in TRAMP; Dnmt1 hypomorphic
mice, we first measured global DNA methylation in prostate
tissues. 5mdC levels were significantly decreased in TRAMP;
Dnmt1N/Rmouse prostates at all ages and in TRAMP;
Dnmt1N/?mouse prostates at 15 and 24 weeks (Fig. 5A and C).
FIG. 3. Dnmt mRNA expression and global DNA methylation in the prostate of Dnmt1 hypomorphic mice. (A) Dnmt1 (left), Dnmt3a (center),
and Dnmt3b (right) mRNA expression in prostate tissues from mice of the indicated genotypes, at 15 and 24 weeks of age, were measured by
qRT-PCR as described in Materials and Methods. Three to six mice were analyzed per group, and the means and standard errors are plotted.
(B) 5mdC levels in the prostate tissues from mice of the indicated genotypes, at 15 and 24 weeks of age, were measured by liquid chromatography-
tandem spectrometry as described in Materials and Methods. Three to four mice were analyzed per group, and the means and standard errors are
plotted. (C) The B1 repetitive element methylation levels in prostate tissues from mice of the indicated genotypes, at 15 and 24 weeks of age, were
measured by sodium bisulfite pyrosequencing as described in Materials and Methods. Three to four mice were analyzed per group, and the means
and standard errors are plotted. (D) HELP analysis of 24-week-old Dnmt1?/?and Dnmt1N/Rmouse prostate. The total number of fragments
analyzed are indicated by a line and plotted on the right axis, and the proportion of methylated fragments are indicated by columns and plotted
on the left axis. HELP regions were subclassified as listed, as described in Materials and Methods.
VOL. 30, 2010 Dnmt1 AND MURINE PROSTATE CANCER4163
B1 methylation levels were unchanged in all genotypes at 12
weeks but were decreased in TRAMP; Dnmt1N/Rmice at 15
and 24 weeks and in TRAMP; Dnmt1N/?mice at 15 weeks (Fig.
5D and F). Similarly, in liver, B1 methylation was decreased in
TRAMP; Dnmt1N/Rmice at 15 and 24 weeks and in TRAMP;
Dnmt1N/?mice at 15 weeks (data not shown). These data
indicate that hypomorphic Dnmt1 expression in TRAMP
causes global DNA hypomethylation in both tumor and normal
tissues and that this effect is most dramatic in TRAMP;
We next examined whether hypomorphic Dnmt1 expression
alters locus-specific DNA hypermethylation in TRAMP tu-
mors. Our prior studies, using restriction landmark genomic
scanning (RLGS), identified loci that are hypermethylated in
late-stage TRAMP tumors (2, 35, 37). Here, we used mass
array quantitative methylation analysis (MAQMA) to analyze
the methylation status of four loci identified by RLGS in our
previous studies: Irx3, Cacna1a, Cdkn2a, and Nrxn2 (2, 35, 37).
Irx3 displays 5? region hypermethylation in TRAMP, whereas
the other three genes display downstream gene body hyper-
methylation, suggesting that the causes of these epigenetic
lesions may be distinct (2, 35, 37). MAQMA analysis of pros-
tate samples from 24-week-old mice revealed that, relative to
TRAMP; Dnmt1?/?mice, Irx3 was hypomethylated in TRAMP;
Dnmt1N/?and TRAMP; Dnmt1N/Rmouse prostates (Fig. 6A).
In contrast, the only significant methylation change at the
other three loci was a loss of Nrxn2 methylation in TRAMP;
Dnmt1R/?mice (Fig. 6B to D). These data suggest that Dnmt1
may contribute to promoter DNA hypermethylation in
TRAMP but may play a smaller role in downstream gene body
To more comprehensively examine the effect of hypomor-
phic Dnmt1 expression on DNA methylation in TRAMP, we
conducted HELP analysis on TRAMP; Dnmt1?/?mice and
TRAMP; Dnmt1N/Rmice at 15 and 24 weeks of age (24). Sam-
ple selection was made as described in Materials and Methods.
The experimental design allowed us to address how DNA
methylation patterns change during progression from early-
stage (15 weeks) to late-stage (24 weeks) prostate cancer, as
well as to define the role of Dnmt1 in these changes. Overall,
81.1 and 86.0% of HELP fragments were methylated in
TRAMP; Dnmt1?/?mice at 15 and 24 weeks, respectively,
indicating a high degree of methylation that was further in-
creased during tumor progression (Fig. 7A). In contrast,
TRAMP; Dnmt1N/Rmice showed a reduced level of methyl-
ation at 15 weeks (77.7%) that was only slightly increased at 24
weeks (78.5%) (Fig. 7A). The methylation level of TRAMP;
Dnmt1N/Rat 24 weeks was almost equivalent to Dnmt1N/Rmice
at this time point (78.7%; Fig. 3D). These data suggest that
hypomorphic Dnmt1 expression has a dramatic impact on
DNA methylation genome-wide during tumor progression in
To determine the regions of the genome that were affected
by Dnmt1 reduction, we further analyzed the HELP data to
assess DNA methylation at repetitive DNA elements, CpG
islands, gene bodies, and promoter regions (Fig. 7B to E). At
repetitive DNA elements and gene bodies, TRAMP; Dnmt1N/R
mice showed substantial DNA hypomethylation at 15 weeks,
which was further evident at 24 weeks (Fig. 7B and C). This is
consistent with the overall HELP fragment data (Fig. 7A) and
likely reflects the fact that these regions are the most abundant
class of HELP fragments (and genomic regions) analyzed. In
contrast, and notably, CpG islands and promoter regions
showed similar methylation in TRAMP; Dnmt1?/?mice and
TRAMP; Dnmt1N/Rmice at 15 weeks but highly divergent
methylation at 24 weeks (Fig. 7D and E). This effect was
TABLE 2. Characteristics of the top 21 loci hypermethylated in 24-week-old Dnmt1?/?mice relative to Dnmt1N/Rmice, identified by HELPa
Gene IDChromosomeProteinFull name
Location of hypermethylated HELP
PromoterGene body CpG island
Histidine ammonia lyase
Zinc finger protein
Somatostatin receptor 5
Synaptonemal protein complex protein 2
Spermatogenesis and oogenesis specific basic
Solute carrier family transporter
Preferentially expressed antigen in melanoma
Preferentially expressed antigen in melanoma
Preferentially expressed antigen in melanoma
Parahox cluster neighbor
ATP binding cassette transporter
Free fatty acid receptor 3
Free fatty acid receptor 3
Solute carrier family transporter
Tumor rejection antigen p1a
aThat is, hypermethylated HELP fragments with a ?1.6 log ratio (HpaII/MspI).
4164 MOREY KINNEY ET AL.MOL. CELL. BIOL.
FIG. 4. Dnmt protein expression in prostates from TRAMP; Dnmt1 hypomorphic mice. (A) Representative Western blots of Dnmt1, Dnmt3a, and
Dnmt3b in prostate tissues from 24-week-old TRAMP mice of the indicated genotypes. The arrow indicates the position of Dnmt3a, as determined by
Western analysis of Dnmt3a-null cell lines (data not shown). Ponceau S total protein staining was used to confirm equivalent protein input. (B) Dnmt1
protein expression in prostate tissues from TRAMP mice of the indicated genotypes and ages was determined by quantification of compiled Western blot
data, as described in Materials and Methods. The number of samples analyzed per group is indicated on the bars, and means and standard errors are
plotted. (C) Dnmt3a protein expression. (D) Dnmt3b protein expression. Mann-Whitney test P values of significant differences (P ? 0.05), compared to
TRAMP; Dnmt1?/?mice, are shown.
VOL. 30, 2010Dnmt1 AND MURINE PROSTATE CANCER4165
characterized by a robust increase in both CpG island and
promoter region methylation in TRAMP; Dnmt1?/?mice at 24
weeks compared to 15 weeks, with no corresponding increase
in TRAMP; Dnmt1N/Rmice (Fig. 7D and E). The dramatic
increase in CpG island methylation in TRAMP; Dnmt1?/?
mice at 24 weeks is consistent with earlier studies showing that
RLGS spot loss (which measures DNA methylation predomi-
nantly at CpG islands) is a late event during TRAMP tumor
progression (35). Analysis of the genes showing the greatest
degree of hypermethylation in TRAMP; Dnmt1?/?mice rela-
tive to TRAMP; Dnmt1N/Rmice at 24 weeks revealed that most
of the hypermethylated regions are in promoter CpG islands
(Table 3). With the exception of a small subset of these genes,
including Csnk1g2, Ints6, and En1, most of the identified genes
are not previously known to be regulated by DNA methylation
(25, 43, 45).
Primary tumor incidence and pathological stage in TRAMP;
Dnmt1 hypomorphic mice. The significantly altered DNA
methylation patterns in TRAMP, Dnmt1 hypomorphic mice
(particularly in TRAMP; Dnmt1N/Rmice) suggested that the
tumor phenotype in these mice may be affected. To address
this question, we analyzed several parameters in TRAMP;
Dnmt1 hypomorphic mice, including body, prostate, and UG
tract weight, primary tumor incidence, and tumor pathology.
Similar to Dnmt1N/Rmice, TRAMP; Dnmt1N/Rmice showed
significantly reduced body weights relative to control mice
(TRAMP; Dnmt1?/?) (data not shown). However, after normal-
ization to body weight, the TRAMP; Dnmt1 hypomorphic mice
did not show consistent changes in prostate weight compared to
FIG. 5. Global DNA methylation in prostate tissues of TRAMP; Dnmt1 hypomorphic mice. (A to C) 5mdC levels in prostate tissues from
TRAMP mice of the indicated genotypes and ages was measured as described in Materials and Methods. The number of samples analyzed
per group is indicated on the bars, and the mean and standard errors are plotted. (D to F) B1 repetitive element methylation levels in
prostate tissues from mice of the indicated genotypes and ages was measured as described in Materials and Methods. The number of samples
analyzed per group is indicated on the bars, and the means and standard errors are plotted. Mann-Whitney test P values of significant
differences (P ? 0.07), compared to TRAMP; Dnmt1?/?mice, are shown.
FIG. 6. Locus-specific DNA methylation in prostates from
TRAMP; Dnmt1 hypomorphic mice. MAQMA was used to deter-
mine locus-specific DNA methylation in prostate tissues from 24-
week-old TRAMP mice of the indicated genotypes, as described in
(B) Cacna1a gene body methylation; (C) Cdkn2a gene body meth-
ylation; (D) Nrxn2 gene body methylation. In each panel, the num-
ber of samples analyzed per group was (from left to right) 18, 12, 22,
and 16, and means and the standard errors are plotted. Mann-
Whitney test P values of significant differences (P ? 0.05), com-
pared to TRAMP; Dnmt1?/?mice, are shown.
4166MOREY KINNEY ET AL.MOL. CELL. BIOL.
the TRAMP; Dnmt1?/?mice (Fig. 8A). At early time points (12
and 15 weeks), TRAMP; Dnmt1N/?mice showed elevated pri-
mary tumor incidence compared to TRAMP; Dnmt1?/?mice
(Fig. 8B). Similarly, TRAMP; Dnmt1N/Rmice displayed an
elevated tumor incidence at 15 weeks (Fig. 8B). However,
these changes were not statistically significant (Table 4).
Despite the increased tumor incidence at early time points,
at 24 weeks of age, TRAMP; Dnmt1 hypomorphic mice
showed similar (i.e., TRAMP; Dnmt1N/?mice and TRAMP;
Dnmt1N/Rmice) or reduced (TRAMP; Dnmt1R/?) primary
tumor incidence compared to control mice (Fig. 8B and
Table 4). These data suggested that hypomorphic Dnmt1
expression may have opposing effects on prostate tumor
formation, with a promotion effect at early stages and a
suppressive effect at later stages of tumor progression.
To further define the effect of Dnmt1 reduction on prostate
tumor development in TRAMP, we determined the patholog-
ical stage of primary tumors by examining H&E-stained pros-
tate tissues. Tissue sections were scored for tumor stage (N,
normal; PIN, prostatic intraepithelial neoplasia; WD, well dif-
ferentiated; moderately differentiated; PD, poorly differenti-
ated), and the percentage of tissue in each stage was deter-
mined as described previously (20). This method allowed us to
calculate a disease index, which is based on the percentage of
each pathological stage determined for each prostatic lobe.
Disease index values represent the pathological stage averaged
across three prostate lobes (dorsal, lateral, and ventral lobes).
At 12 weeks of age, TRAMP; Dnmt1N/Rmice had a significantly
increased disease index value compared to TRAMP; Dnmt1?/?
mice (Fig. 9A). At 15 weeks of age, all TRAMP; Dnmt1 hypo-
FIG. 7. HELP analysis of TRAMP; Dnmt1?/?and TRAMP; Dnmt1N/Rmouse prostate at 15 and 24 weeks of age. Sample selection and HELP
assays were performed as described in Materials and Methods. In all panels, the total number of HELP fragments analyzed is shown at top, and
the proportion of fragments that are methylated is plotted. (A) All HELP fragments; (B) fragments in repetitive DNA elements; (C) fragments
in gene bodies; (D) fragments in CpG islands; (E) fragments in promoter regions. HELP regions were subclassified as described in Materials and
VOL. 30, 2010 Dnmt1 AND MURINE PROSTATE CANCER4167
morphic genotypes had increased disease index values com-
pared to TRAMP; Dnmt1?/?mice (Fig. 9B). In contrast, at 24
weeks of age both TRAMP; Dnmt1R/?mice and TRAMP;
Dnmt1N/Rmice had decreased disease index values compared
to TRAMP; Dnmt1?/?mice (Fig. 9C). These data suggest that
Dnmt1 reduction accelerates the early stages of prostate tumor
progression but inhibits the later stages of tumor progression
in TRAMP mice.
Metastatic tumor formation in TRAMP; Dnmt1 hypomor-
phic mice. One of the advantages of the TRAMP model is that
primary prostate cancer progresses to local and distant site
metastases, reminiscent of the human disease (15). This fact
allowed us to investigate the impact of hypomorphic Dnmt1
expression on metastatic tumor development in vivo, which, to
our knowledge, has not been investigated previously (1, 7, 9,
14, 52). Initially, we examined macrometastatic tumor growth
at necropsy by visual inspection of target tissues with a dissect-
ing microscope. A negligible level of metastases developed in
TRAMP mice at early time points (12 and 15 weeks), as ex-
pected (Fig. 8C and Table 4). However, at 24 weeks, 30% of
TRAMP; Dnmt1?/?mice developed metastatic tumors (Fig.
8C and Table 4). Strikingly, macrometastatic tumor develop-
ment was reduced in all three TRAMP; Dnmt1 hypomorphic
mouse genotypes, with a complete elimination of macrometas-
tases in TRAMP; Dnmt1N/Rmice (Fig. 8C and Table 4). Both
local and distant site macrometastases were reduced in Dnmt1
hypomorphic mice; these reductions are statistically significant
in TRAMP; Dnmt1N/Rmice (Table 4).
To determine the stage at which hypomorphic Dnmt1 ex-
pression impacts metastatic tumor development, we next as-
sessed micrometastatic lesions using IHC staining for Tag on
lymph node, liver, lung, and kidney tissues, the common sites
of metastases in TRAMP (18, 20). Consistent with the macro-
metastasis data, only a negligible level of micrometastatic tu-
mors was present at 12 and 15 weeks of age in all genotypes
(Table 5). However, at 24 weeks, 40 and 15% of TRAMP;
Dnmt1?/?mice developed local and distant micrometastatic
tumors, respectively (Table 5). Interestingly, the incidence of
local micrometastatic tumors at 24 weeks was similar in
TRAMP; Dnmt1?/?mice, TRAMP; Dnmt1N/?mice, and
mice but was reduced in TRAMP;
Dnmt1R/?mice (Table 5). In contrast, all strains of Dnmt1
hypomorphic mice showed reduced micrometastatic tumors at
distant sites; this effect was most dramatic in TRAMP;
Dnmt1N/Rmice, in which no lesions were observed (Table 5).
These findings were consistent with the complete absence of
distant macro-metastatic tumors in TRAMP; Dnmt1N/Rmice
(Table 4) and reveal that Dnmt1 plays a critical role in estab-
lishing distant site metastases in TRAMP.
Finally, to determine whether reduction in circulating an-
drogens in the TRAMP; Dnmt1N/Rmice accounted for the
observed inhibitory effect on metastatic tumor growth at 24
TABLE 3. Characteristics of the top 27 candidate loci hypermethylated in 24-week-old TRAMP; Dnmt1?/?mice relative to TRAMP;
Dnmt1N/R mice, identified by HELPa
Gene ID Chromosome Protein Full name
1-Acylglycerol-3-phosphate O-acyltransferase 3
Casein kinase 1, gamma 2
Cisplatin resistance related protein CRR9p
MYC binding protein 2 (aka PAM in human)
Integrator complex subunit 6 (DICE1, DDX26)
Low-density lipoprotein-related protein 12
SUMO/sentrin-specific peptidase 2
Zinc finger protein 295
Raftlin lipid raft linker 1
Adrenergic receptor, beta 1
Integrin-linked kinase-associated serine/threonine phosphatase 2C
Rtf1, Paf1/RNA polymerase II complex component, homolog (S. cerevisiae)
Regulating synaptic membrane exocytosis 4
Paired-like homeodomain transcription factor 2
Transcription factor AP-2, epsilon
Solute carrier family 6 (neurotransmitter transporter, GABA), member 11
DENN/MADD domain containing 5B
Potassium voltage-gated channel, shaker-related subfamily, member 7
A kinase (PRKA) anchor protein 13
Telomeric repeat binding factor 2
F-box protein 31
Zinc finger protein 703
Phosphate cytidylyltransferase 1, choline, beta isoform
aHypermethylated HELP fragments ? 3.75 log ratio (HpaII/MspI).
4168 MOREY KINNEY ET AL.MOL. CELL. BIOL.
weeks, we conducted two independent tests. First, we mea-
sured UG weight, which would be expected to show a dramatic
decrease in mice with reduced circulating androgens due to the
strict dependence of the rodent urogenital tract on testoster-
one for growth (26). As shown in Fig. 10A, TRAMP; Dnmt1?/?
mice and TRAMP; Dnmt1N/Rmice show very similar UG sizes,
after normalization to body weight, at all three time points (12,
15, and 24 weeks). As a second test, we measured nuclear AR
staining, since nuclear localization of AR is dependent on
androgens (4). We stained 17 TRAMP; Dnmt1?/?and 22
TRAMP; Dnmt1N/Rmouse prostate and seminal vesicle sec-
tions (two slides per animal) with AR antibody for IHC, as
described in Materials and Methods. The slides were deiden-
tified, and nuclear AR staining was scored (as yes or no [Y/N]).
The resulting data revealed no difference in nuclear AR stain-
ing between the two genotypes (data not shown; representative
AR IHC staining of seminal vesicles is shown in Fig. 10B).
Taken together, these data suggest that it is highly unlikely that
alterations in circulating androgens account for the dramatic
reduction of metastatic tumor growth observed in TRAMP;
Dnmt1 in prostate development and DNA methylation. In
wild-type mice, hypomorphic Dnmt1 expression did not alter
the general morphology of the prostate, nor did it appear to
alter its differentiation state. However, Dnmt1N/Rmice
(which showed the lowest level of Dnmt1 expression and
greatest degree of DNA hypomethylation) had a significant
FIG. 8. Prostate weight and tumor incidence in TRAMP; Dnmt1 hypomorphic mice. (A) Prostate weights (normalized to body weight) of
TRAMP mice of the indicated genotypes were determined at necropsy at the indicated ages. The number of samples analyzed per group is
indicated on the bars, and the means and standard errors are plotted. No significant differences (P ? 0.05; Mann-Whitney test) were observed.
(B) The primary tumor incidence of TRAMP mice of the indicated genotypes was determined at necropsy at the indicated ages. Each bar indicates
the mean of a sample group where the number of samples analyzed per group is the same as in panel A. No significant differences (P ? 0.05; Fisher
exact test) were observed. (C) The macrometastatic tumor incidence of TRAMP mice of the indicated genotypes was determined at necropsy at
the indicated ages. Each bar indicates the mean of a sample group where the number of samples analyzed per group is the same as in panel A.
Fisher exact test P values of significant differences (P ? 0.05), compared to TRAMP; Dnmt1?/?mice, are shown.
VOL. 30, 2010 Dnmt1 AND MURINE PROSTATE CANCER4169
survival defect, as evidenced by a 5-fold decrease from the
expected Mendelian ratio. Surviving Dnmt1N/Rmice also
had reduced body weights. These data provide evidence that
Dnmt1 is required for normal vertebrate development and
are in agreement with previous data showing that Dnmt1
knockout mice are embryonic lethal and that Dnmt1 knock-
down zebrafish show reduced survival (29, 42). Our data
suggest that the level of Dnmt1 expression in Dnmt1N/Rmice
is close to the threshold required for normal murine devel-
opment. Importantly, however, the defects observed in
Dnmt1N/Rmice did not appear to specifically impact the
development of the prostate. Since Dnmt3a and Dnmt3b
expression was retained in Dnmt1 hypomorphic mice, it is
reasonable to hypothesize that these enzymes may compen-
sate for Dnmt1 reduction in the surviving Dnmt1N/Rmice.
Most important for the present study, the apparently normal
development of the prostate in surviving Dnmt1 hypomor-
phic mice suggests that this genetic model is valid for as-
sessing the impact of reduced Dnmt1 expression on prostate
As expected, Dnmt1N/Rprostate and livers had reduced lev-
els of 5mdC and B1 repetitive element methylation, which are
measures of global DNA methylation status. In addition,
HELP analyses revealed slight genome-wide hypomethylation
in the Dnmt1N/Rprostate. However, HELP also revealed a
more complicated pattern of DNA methylation alterations in
the Dnmt1N/Rprostate. Although a number of genes were
identified that become hypomethylated in the Dnmt1N/Rpros-
tate, these events frequently occurred outside of CpG islands.
Moreover, unexpectedly, CpG island regions overall were
hypermethylated in the Dnmt1N/Rprostate compared to
Dnmt1?/?mice. It is possible that this effect reflects compen-
satory epigenetic mechanisms (possibly mediated by Dnmt3
enzymes) that act in response to global DNA hypomethylation.
Dual or combinatorial Dnmt disruption approaches in vivo
could be used to test this idea. Potentially, this type of feedback
response could explain the frequent coexistence of global
DNA hypomethylation and CpG island hypermethylation ob-
served in human cancer.
Dnmt1 and DNA methylation alterations during prostate
cancer progression. Tumor progression in TRAMP is charac-
terized by two major alterations in DNA methylation: (i) global
hypomethylation appearing at early stages that becomes more
pronounced at later stages and (ii) locus-specific CpG island
hypermethylation, which is chiefly observed in the late stages
of prostate cancer development (35). In TRAMP; Dnmt1N/R
mice, these two alterations were exacerbated or inhibited, re-
spectively. By 15 weeks of age, 5mdC and B1 methylation levels
in the prostates of TRAMP; Dnmt1 hypomorphic mice were
significantly reduced. In agreement with this, HELP analysis
revealed genome-wide DNA hypomethylation in TRAMP;
Dnmt1N/Rmice at 15 weeks. The hypomethylating effect was
seen specifically at repetitive DNA elements and gene bodies
at this time point but was not seen at the promoters or CpG
islands. A likely explanation for the lack of hypomethylation at
the CpG islands and promoters at 15 weeks in TRAMP;
Dnmt1N/Rmice is that these regions are largely hypomethyl-
ated at baseline and the aberrant hypermethylation of these
regions had not occurred to significant levels at this time. In
contrast, at 24 weeks, TRAMP; Dnmt1?/?mice showed dra-
matic increases in both CpG island and promoter hypermeth-
ylation. Remarkably, these changes appeared to be completely
abrogated in TRAMP; Dnmt1N/Rmice, supporting a major role
for Dnmt1 in CpG island and promoter region DNA hyper-
methylation during prostate cancer progression. The current
data set does not allow us to resolve whether Dnmt1 is involved
in de novo or maintenance methylation at these hypermethyl-
Our prior work using RLGS identified two general catego-
ries of DNA hypermethylation events in TRAMP: one in which
promoter methylation is correlated with gene repression and a
second in which gene body methylation is associated with in-
creased gene expression (2, 35, 37). Interestingly, in TRAMP;
Dnmt1 hypomorphic mice it appeared that promoter hyper-
methylation is inhibited (illustrated by Irx3), whereas down-
stream hypermethylation was inconsistently affected (as illus-
trated by Cacna1a, Cdkn2a, and Nrxn2). This suggests that
Dnmt1 may be primarily involved in initiating or maintaining
aberrant promoter hypermethylation, with a less important
role in catalyzing downstream gene DNA hypermethylation in
TRAMP. Further studies are necessary to determine whether
hypomorphic Dnmt1 expression alters the expression of hyper-
methylated gene targets in TRAMP.
Opposing roles for Dnmt1 in early- and late-stage primary
prostate cancer. The dual nature of the DNA methylation
changes observed in TRAMP revealed in our previous studies
(i.e., global hypomethylation appearing at early stages, CpG
island hypermethylation at late stages) led us to hypothesize
that hypomorphic Dnmt1 expression may accelerate early-
stage prostate tumor development and, conversely, inhibit late-
stage prostate cancer. Our data support this hypothesis. At 12
and 15 weeks of age, TRAMP; Dnmt1 hypomorphic mice
showed slightly increased primary prostate tumor incidence, as
well as significantly increased pathological stage (i.e., the dis-
TABLE 4. Primary and metastatic tumor incidence in TRAMP;
Dnmt1 hypomorphic micea
Time (in wks) and
aFisher exact test P values, compared to TRAMP; Dnmt1?/?mice, were
determined. “Local” refers to lymph node metastases; “distant” refers to liver,
lung, or kidney metastases. ?, significant difference (P ? .05). NA, not applicable.
n, number of mice.
4170 MOREY KINNEY ET AL.MOL. CELL. BIOL.
ease index score). Global DNA hypomethylation was also ob-
served in Dnmt1 hypomorphic mice at these time points. Thus,
reduced Dnmt1 expression and the associated reduction in
global DNA methylation appear to functionally alter the dis-
ease phenotype. Global DNA hypomethylation is associated
with genomic instability and oncogene expression, both of
which contribute to oncogenesis (9, 12, 22). Our data suggest
that one or both of these mechanisms may contribute to tu-
morigenesis in TRAMP.
In direct contrast to the tumor-promoting effect seen at early
ages in TRAMP; Dnmt1 hypomorphic mice, at a later time
point (24 weeks), these mice showed similar or slightly reduced
primary prostate tumor incidence, as well as significantly de-
creased pathological stage (i.e., the disease index). These ef-
fects coincided with dramatic reductions in locus-specific DNA
hypermethylation genome-wide, as revealed by HELP analy-
ses. These observations support the notion that aberrant locus-
specific DNA hypermethylation contributes to late stages of
primary tumor development in TRAMP.
It should be noted that some of the changes in tumor phe-
notype observed here could be related to either strain or
Dnmt1 allele-specific effects. Because the experimental mice
were not 100% FVB mice, this could affect the linkage for the
Dnmt1Rallele to the C57BL/6 strain, such that Dnmt1R/?and
Dnmt1N/Rmice may display phenotypic similarities that are
based on strain and not genotype. In fact, there were a few
instances where this appeared to occur, including the disease
index scores at 24 weeks of age. A previous report utilizing the
identical Dnmt1 hypomorphic model revealed an analogous
effect on tumor phenotype wherein Dnmt1R/?and Dnmt1N/R
FIG. 9. Prostate pathological stage and disease index in TRAMP; Dnmt1 hypomorphic mice. Pathological stage was determined for the dorsal,
lateral, and ventral prostate lobes (DLV) and averaged as described in Materials and Methods. (A) The proportion of the prostate classified as
normal (N), PIN, well-differentiated tumor (WD), moderately differentiated tumor (MD), and poorly differentiated tumor (PD) for mice of each
TRAMP genotype, at 12 weeks of age, is plotted on the left. A disease index score was calculated as described in Materials and Methods and is
plotted at right. The number of samples analyzed per group is indicated on the bars, and the means and standard errors are plotted. (B) Patho-
logical staging and disease index score as described in panel A for 15-week-old mice. (C) Pathological staging and disease index score as described
in panel A for 24-week-old mice. Mann-Whitney test P values of significant differences (P ? 0.09), compared to TRAMP; Dnmt1?/?mice, are
VOL. 30, 2010 Dnmt1 AND MURINE PROSTATE CANCER4171
mice in the Mlh1?/?background sometimes showed effects
distinct from that seen in Dnmt1N/?mice, despite the fact that
experimental mice had been backcrossed for at least 10 gen-
erations to reduce strain variability (52). It is also possible that
specific differences between the configuration of the Dnmt1 R
and N alleles could have distinct phenotypic effects (52). Nev-
ertheless, in almost all instances, TRAMP; Dnmt1N/Rmice,
which have the most robust loss of Dnmt1 expression and
DNA hypomethylation, also showed the most divergent mo-
lecular and biological phenotypes, confirming the validity of
the model system.
Dnmt1 promotes prostate cancer metastasis. The most
striking finding in the present study was the dramatic inhibition
of prostate tumor metastases observed in TRAMP; Dnmt1N/R
mice. While approximately one-third of control mice displayed
macrometastatic tumor growth at 24 weeks of age, no lesions
were observed in TRAMP; Dnmt1N/Rmice. Moreover, clear
reduction of macrometastatic tumors occurred in the other
hypomorphic TRAMP; Dnmt1 strains. The reduced level of
macrometastases corresponded to both local and distant site
tumors. In contrast to the effect on macrometastases, IHC
staining for Tag-positive cells (scored as micrometastases)
revealed similar levels of local (lymph node) involved mi-
crometastatic tumors in control and Dnmt1 hypomorphic
mice. These data suggest that the early stages of metastases,
e.g., invasion and colonization of the draining lymph nodes,
are not inhibited by Dnmt1 reduction. Rather, it is the
growth of these microscopic metastatic lesions at secondary
sites that appears to be impacted. In light of our data, it is
notable that numerous studies suggest that growth of mac-
roscopic foci at distant sites is the rate-limiting step in tumor
Importantly, TRAMP; Dnmt1 hypomorphic mice showed re-
duced levels of micrometastatic tumor growth at distant organs
(i.e., liver, lung, and kidney). This effect was most dramatic in
TRAMP; Dnmt1N/Rmice, in which no distant site micrometa-
static lesions were detected. Taken together, our data suggest
that Dnmt1 contributes to at least two different stages of pros-
tate metastasis: (i) the growth of already present micrometa-
static lesions in the lymph nodes and (ii) the colonization and
growth of metastases at distant organs. As with other solid
tumors, metastasis is the key event conferring poor prognosis
in human prostate cancer (13); thus, identification of factors
that contribute to this process, such as Dnmt1, is critically
In agreement with our findings, Day and coworkers have
shown that treatment of intact or castrated TRAMP mice
with the DNA methyltransferase inhibitor 5-aza-2?-deoxycy-
tidine (DAC) inhibits both primary tumor growth and the
development of lymph node macrometastases (34, 56).
While DAC has effects beyond inhibition of DNA methyl-
ation, the data suggest that inhibition of DNA hypermeth-
ylation mediated by Dnmt1 may be directly responsible for
the phenotypes observed in the current and prior studies.
The data showing that robust DNA hypermethylation occurs
in late-stage prostate cancer, castration-recurrent prostate
cancer, and metastatic prostate cancer in both mouse mod-
els and humans also support this idea (2, 35, 37, 38). More-
over, studies using in vitro cell models suggest that aberrant
DNA methylation mediated by Dnmt enzymes contributes
to the development of cellular phenotypes associated with
metastasis (5, 30, 40, 54, 55). It will be of particular impor-
tance to define the genes targeted by DNA hypermethyl-
ation that contribute to prostate cancer metastasis.
In summary, based on our earlier characterization of the
FIG. 10. Urogenital tract (UG) weight and androgen receptor
(AR) staining in TRAMP; Dnmt1?/?and TRAMP; Dnmt1N/Rmice.
(A) UG weight normalized to body weight. The number of samples
analyzed per group is indicated on the bars, and the means and
standard errors are plotted. No significant differences between the
two genotypes (P ? 0.05; Mann-Whitney test) were observed.
(B) Representative example of AR IHC staining in the seminal
vesicles of 24-week-old mice of the indicated genotypes. IHC was
performed as described in Materials and Methods, and the negative
control is shown at right. Scale bar, 100 ?m.
TABLE 5. Micrometastatic tumor incidence in TRAMP; Dnmt1
Time (in wks) and
aAs determined by IHC staining for large T antigen (Tag). “Local” refers to
lymph node tumors; “distant” refers to liver, lung, or kidney tumors. Fisher exact
test P values, compared to TRAMP; Dnmt1?/?mice, were determined. NA, not
applicable. n, number of mice.
4172 MOREY KINNEY ET AL.MOL. CELL. BIOL.
epigenetic changes during TRAMP tumorigenesis, we hy-
pothesized that Dnmt1 may play a dual role in prostate
caner progression characterized by tumor suppressor activ-
ity during early stages of the disease and oncogenic function
at late stages. Our findings from the TRAMP; Dnmt1 hypo-
morphic mouse model confirm this hypothesis and suggest
that Dnmt1 has opposing effects on early and late stage
prostate cancer. Importantly, the apparent tumor-promot-
ing effect of Dnmt reduction on early-stage lesions in Dnmt1
hypomorphic mice does not support the use of DNA hy-
pomethylating agents as chemopreventive approaches for
prostate cancer. However, the robust inhibitory effect of
Dnmt1 reduction on prostate tumor metastasis (and in par-
ticular the prominent reduction of distant-site metastasis),
which constitutes the clinically relevant human condition, is
striking. This outcome supports further investigation of
Dnmt1 inhibitors as therapeutic interventions for advanced
and metastatic prostate cancer.
This study was supported by NIH R21CA128062 (A.R.K.), Roswell
Park Alliance Foundation (A.R.K.), 5T32CA009072 (S.R.M.K.), DOD
PC060354 (S.R.M.K.), and NCI Center Grant (CA16056) (Roswell
Park Cancer Institute [RPCI]).
We thank Petra Link of the Karpf lab and Ellen Karasik and Bryan
Gillard of the RPCI Mouse Tumor Model Core for outstanding tech-
nical support. We thank Peter Laird (USC) for generously providing
Dnmt1 hypomorphic mice and for valuable advice and David Goodrich
(RPCI) for numerous helpful suggestions.
1. Baba, S., Y. Yamada, Y. Hatano, Y. Miyazaki, H. Mori, T. Shibata, and A.
Hara. 2009. Global DNA hypomethylation suppresses squamous carcinogen-
esis in the tongue and esophagus. Cancer Sci. 100:1186–1191.
2. Camoriano, M., S. R. Kinney, M. T. Moser, B. A. Foster, J. L. Mohler, D. L.
Trump, A. R. Karpf, and D. J. Smiraglia. 2008. Phenotype-specific CpG
island methylation events in a murine model of prostate cancer. Cancer Res.
3. Cooper, C. S., and C. S. Foster. 2009. Concepts of epigenetics in prostate
cancer development. Br. J. Cancer 100:240–245.
4. Dehm, S. M., and D. J. Tindall. 2007. Androgen receptor structural and
functional elements: role and regulation in prostate cancer. Mol. Endocrinol.
5. Deng, T., Y. Kuang, L. Wang, J. Li, Z. Wang, and J. Fei. 2009. An essential
role for DNA methyltransferase 3a in melanoma tumorigenesis. Biochem.
Biophys. Res. Commun. 387:611–616.
6. De Smet, C., C. Lurquin, B. Lethe, V. Martelange, and T. Boon. 1999. DNA
methylation is the primary silencing mechanism for a set of germ line- and
tumor-specific genes with a CpG-rich promoter. Mol. Cell. Biol. 19:7327–
7. Eads, C. A., A. E. Nickel, and P. W. Laird. 2002. Complete genetic suppres-
sion of polyp formation and reduction of CpG-island hypermethylation in
ApcMin/?Dnmt1-hypomorphic Mice. Cancer Res. 62:1296–1299.
8. Eccles, S. A., and D. R. Welch. 2007. Metastasis: recent discoveries and novel
treatment strategies. Lancet 369:1742–1757.
9. Eden, A., F. Gaudet, A. Waghmare, and R. Jaenisch. 2003. Chromosomal
instability and tumors promoted by DNA hypomethylation. Science 300:455.
10. Ehrich, M., M. R. Nelson, P. Stanssens, M. Zabeau, T. Liloglou, G. Xinari-
anos, C. R. Cantor, J. K. Field, and D. van den Boom. 2005. Quantitative
high-throughput analysis of DNA methylation patterns by base-specific
cleavage and mass spectrometry. Proc. Natl. Acad. Sci. U. S. A. 102:15785–
11. Esteve, P. O., H. G. Chin, A. Smallwood, G. R. Feehery, O. Gangisetty, A. R.
Karpf, M. F. Carey, and S. Pradhan. 2006. Direct interaction between
DNMT1 and G9a coordinates DNA and histone methylation during repli-
cation. Genes Dev. 20:3089–3103.
12. Feinberg, A. P., and B. Tycko. 2004. The history of cancer epigenetics. Nat.
Rev. Cancer 4:143–153.
13. Foley, C. L., and M. R. Feneley. 2009. The clinical significance and thera-
peutic implications of extraprostatic invasion. Surg. Oncol. 18:203–212.
14. Gaudet, F., J. G. Hodgson, A. Eden, L. Jackson-Grusby, J. Dausman, J. W.
Gray, H. Leonhardt, and R. Jaenisch. 2003. Induction of tumors in mice by
genomic hypomethylation. Science 300:489–492.
15. Gingrich, J. R., R. J. Barrios, R. A. Morton, B. F. Boyce, F. J. DeMayo, M. J.
Finegold, R. Angelopoulou, J. M. Rosen, and N. M. Greenberg. 1996. Met-
astatic prostate cancer in a transgenic mouse. Cancer Res. 56:4096–4102.
16. Greenberg, N. M., F. DeMayo, M. J. Finegold, D. Medina, W. D. Tilley, J. O.
Aspinall, G. R. Cunha, A. A. Donjacour, R. J. Matusik, and J. M. Rosen.
1995. Prostate cancer in a transgenic mouse. Proc. Natl. Acad. Sci. U. S. A.
17. Guo, Z. S., J. A. Hong, K. R. Irvine, G. A. Chen, P. J. Spiess, Y. Liu, G. Zeng,
J. R. Wunderlich, D. M. Nguyen, N. P. Restifo, and D. S. Schrump. 2006. De
novo induction of a cancer/testis antigen by 5-aza-2?-deoxycytidine augments
adoptive immunotherapy in a murine tumor model. Cancer Res. 66:1105–
18. Hurwitz, A. A., B. A. Foster, J. P. Allison, N. M. Greenberg, and E. D. Kwon.
2001. The TRAMP mouse as a model for prostate cancer. Curr. Protoc.
Immunol. Chapter 20:Unit 20.5.
19. Jair, K. W., K. E. Bachman, H. Suzuki, A. H. Ting, I. Rhee, R. W. Yen, S. B.
Baylin, and K. E. Schuebel. 2006. De novo CpG island methylation in human
cancer cells. Cancer Res. 66:682–692.
20. Kaplan-Lefko, P. J., T. M. Chen, M. M. Ittmann, R. J. Barrios, G. E. Ayala,
W. J. Huss, L. A. Maddison, B. A. Foster, and N. M. Greenberg. 2003.
Pathobiology of autochthonous prostate cancer in a pre-clinical transgenic
mouse model. Prostate 55:219–237.
21. Karpf, A. R., and D. A. Jones. 2002. Reactivating the expression of methyl-
ation silenced genes in human cancer. Oncogene 21:5496–5503.
22. Karpf, A. R., and S. Matsui. 2005. Genetic disruption of cytosine DNA
methyltransferase enzymes induces chromosomal instability in human cancer
cells. Cancer Res. 65:8635–8639.
23. Karpf, A. R., B. C. Moore, T. O. Ririe, and D. A. Jones. 2001. Activation of
the p53 DNA damage response pathway after inhibition of DNA methyl-
transferase by 5-aza-2?-deoxycytidine. Mol. Pharmacol. 59:751–757.
24. Khulan, B., R. F. Thompson, K. Ye, M. J. Fazzari, M. Suzuki, E. Stasiek,
M. E. Figueroa, J. L. Glass, Q. Chen, C. Montagna, E. Hatchwell, R. R.
Selzer, T. A. Richmond, R. D. Green, A. Melnick, and J. M. Greally. 2006.
Comparative isoschizomer profiling of cytosine methylation: the HELP as-
say. Genome Res. 16:1046–1055.
25. Kim, E. K., J. Y. Kang, Y. H. Rho, Y. S. Kim, D. S. Kim, and Y. S. Bae. 2009.
Silencing of the CKII? and CKII?? genes during cellular senescence is
mediated by DNA methylation. Gene 431:55–60.
26. Kincl, F. A., M. Maqueo, and R. I. Dorfman. 1965. Influence of various
steroids on testes and accessory sex organs in the rat. Acta Endocrinol.
27. Laird, P. W., L. Jackson-Grusby, A. Fazeli, S. L. Dickinson, W. E. Jung, E.
Li, R. A. Weinberg, and R. Jaenisch. 1995. Suppression of intestinal neopla-
sia by DNA hypomethylation. Cell 81:197–205.
28. La Salle, S., C. Mertineit, T. Taketo, P. B. Moens, T. H. Bestor, and J. M.
Trasler. 2004. Windows for sex-specific methylation marked by DNA meth-
yltransferase expression profiles in mouse germ cells. Dev. Biol. 268:403–415.
29. Li, E., T. H. Bestor, and R. Jaenisch. 1992. Targeted mutation of the DNA
methyltransferase gene results in embryonic lethality. Cell 69:915–926.
30. Lin, R. K., C. H. Hsu, and Y. C. Wang. 2007. Mithramycin A inhibits DNA
methyltransferase and metastasis potential of lung cancer cells. Anticancer
31. Link, P. A., M. R. Baer, S. R. James, D. A. Jones, and A. R. Karpf. 2008.
p53-inducible ribonucleotide reductase (p53R2/RRM2B) is a DNA hypo-
methylation-independent decitabine gene target that correlates with clinical
response in myelodysplastic syndrome/acute myelogenous leukemia. Cancer
32. Lucarelli, M., A. Fuso, R. Strom, and S. Scarpa. 2001. The dynamics of
myogenin site-specific demethylation is strongly correlated with its expres-
sion and with muscle differentiation. J. Biol. Chem. 276:7500–7506.
33. Mavis, C. K., S. R. Morey Kinney, B. A. Foster, and A. R. Karpf. 2009.
Expression level and DNA methylation status of glutathione S-transferase
genes in normal murine prostate and TRAMP tumors. Prostate 69:1312–
34. McCabe, M. T., J. A. Low, S. Daignault, M. J. Imperiale, K. J. Wojno, and
M. L. Day. 2006. Inhibition of DNA methyltransferase activity prevents
tumorigenesis in a mouse model of prostate cancer. Cancer Res. 66:385–392.
35. Morey Kinney, S. R., D. J. Smiraglia, S. R. James, M. T. Moser, B. A. Foster,
and A. R. Karpf. 2008. Stage-specific alterations of DNA methyltransferase
expression, DNA hypermethylation, and DNA hypomethylation during pros-
tate cancer progression in the transgenic adenocarcinoma of mouse prostate
model. Mol. Cancer Res. 6:1365–1374.
36. Morey Kinney, S. R., W. Zhang, M. Pascual, J. M. Greally, B. Gillard, E.
Karasik, B. A. Foster, and A. R. Karpf. 2009. Lack of evidence for green tea
polyphenols as DNA methylation inhibitors in murine prostate. Cancer Pre-
vention Res. 2:1065–1075.
37. Morey, S. R., D. J. Smiraglia, S. R. James, J. Yu, M. T. Moser, B. A. Foster,
and A. R. Karpf. 2006. DNA methylation pathway alterations in an autoch-
thonous murine model of prostate cancer. Cancer Res. 66:11659–11667.
38. Nelson, W. G., A. M. De Marzo, and S. Yegnasubramanian. 2009. Epigenetic
alterations in human prostate cancers. Endocrinology 150:3991–4002.
39. Oda, M., J. L. Glass, R. F. Thompson, Y. Mo, E. N. Olivier, M. E. Figueroa,
VOL. 30, 2010 Dnmt1 AND MURINE PROSTATE CANCER4173
R. R. Selzer, T. A. Richmond, X. Zhang, L. Dannenberg, R. D. Green, A. Download full-text
Melnick, E. Hatchwell, E. E. Bouhassira, A. Verma, M. Suzuki, and J. M.
Greally. 2009. High-resolution genome-wide cytosine methylation profiling
with simultaneous copy number analysis and optimization for limited cell
numbers. Nucleic Acids Res. 37:3829–3839.
40. Olsson, L., and J. Forchhammer. 1984. Induction of the metastatic pheno-
type in a mouse tumor model by 5-azacytidine, and characterization of an
antigen associated with metastatic activity. Proc. Natl. Acad. Sci. U. S. A.
41. Patra, S. K., A. Patra, and R. Dahiya. 2001. Histone deacetylase and DNA
methyltransferase in human prostate cancer. Biochem. Biophys. Res. Com-
42. Rai, K., L. D. Nadauld, S. Chidester, E. J. Manos, S. R. James, A. R. Karpf,
B. R. Cairns, and D. A. Jones. 2006. Zebra fish Dnmt1 and Suv39h1 regulate
organ-specific terminal differentiation during development. Mol. Cell. Biol.
43. Rauch, T., Z. Wang, X. Zhang, X. Zhong, X. Wu, S. K. Lau, K. H. Kernstine,
A. D. Riggs, and G. P. Pfeifer. 2007. Homeobox gene methylation in lung
cancer studied by genome-wide analysis with a microarray-based methylated
CpG island recovery assay. Proc. Natl. Acad. Sci. U. S. A. 104:5527–5532.
44. Robertson, K. D. 2002. DNA methylation and chromatin: unraveling the
tangled web. Oncogene 21:5361–5379.
45. Ropke, A., P. Buhtz, M. Bohm, J. Seger, I. Wieland, E. P. Allhoff, and P. F.
Wieacker. 2005. Promoter CpG hypermethylation and downregulation of
DICE1 expression in prostate cancer. Oncogene 24:6667–6675.
46. Rozen, S., and H. Skaletsky. 2000. Primer3 on the WWW for general users
and for biologist programmers. Methods Mol. Biol. 132:365–386.
47. Schenk, T., S. Stengel, S. Goellner, D. Steinbach, and H. P. Saluz. 2007.
Hypomethylation of PRAME is responsible for its aberrant overexpression
in human malignancies. Genes Chromosomes Cancer 46:796–804.
48. Schmittgen, T. D., and B. A. Zakrajsek. 2000. Effect of experimental treat-
ment on housekeeping gene expression: validation by real-time, quantitative
RT-PCR. J. Biochem. Biophys. Methods 46:69–81.
49. Shen, L., Y. Kondo, Y. Guo, J. Zhang, L. Zhang, S. Ahmed, J. Shu, X. Chen,
R. A. Waterland, and J. P. Issa. 2007. Genome-wide profiling of DNA
methylation reveals a class of normally methylated CpG island promoters.
PLoS Genet. 3:2023–2036.
50. Song, L., S. R. James, L. Kazim, and A. R. Karpf. 2005. Specific method for
the determination of genomic DNA methylation by liquid chromatography-
electrospray ionization tandem mass spectrometry. Anal. Chem. 77:504–510.
51. Thompson, R. F., M. Reimers, B. Khulan, M. Gissot, T. A. Richmond, Q.
Chen, X. Zheng, K. Kim, and J. M. Greally. 2008. An analytical pipeline for
genomic representations used for cytosine methylation studies. Bioinformat-
52. Trinh, B. N., T. I. Long, A. E. Nickel, D. Shibata, and P. W. Laird. 2002.
DNA methyltransferase deficiency modifies cancer susceptibility in mice
lacking DNA mismatch repair. Mol. Cell. Biol. 22:2906–2917.
53. Yamada, Y., L. Jackson-Grusby, H. Linhart, A. Meissner, A. Eden, H. Lin,
and R. Jaenisch. 2005. Opposing effects of DNA hypomethylation on intes-
tinal and liver carcinogenesis. Proc. Natl. Acad. Sci. U. S. A. 102:13580–
54. Yaqinuddin, A., S. A. Qureshi, R. Qazi, and F. Abbas. 2008. Down-regulation
of DNMT3b in PC3 cells effects locus-specific DNA methylation, and re-
presses cellular growth and migration. Cancer Cell Int. 8:13.
55. Yaqinuddin, A., S. A. Qureshi, R. Qazi, S. Farooq, and F. Abbas. 2009.
DNMT1 silencing affects locus specific DNA methylation and increases
prostate cancer derived PC3 cell invasiveness. J. Urol. 182:756–761.
56. Zorn, C. S., K. J. Wojno, M. T. McCabe, R. Kuefer, J. E. Gschwend, and
M. L. Day. 2007. 5-Aza-2?-deoxycytidine delays androgen-independent dis-
ease and improves survival in the transgenic adenocarcinoma of the mouse
prostate mouse model of prostate cancer. Clin. Cancer Res. 13:2136–2143.
4174MOREY KINNEY ET AL.MOL. CELL. BIOL.