Note: Descriptions are shown in the official language in which they were submitted.
211 0~:13
Alterations in the normal gene expression profile of a cell are thought to be early events in
oncogenic transformation (28). A large number of oncogenes are transcription factors (8) while
many oncogenes that are not transcription factors are involved in signal transduction pathways that
trigger activation of transcription factors such as the activation of Jun by the RAS signalling
pathway (6). What are the downstream effectors of Jun that might play a critical role in cellular
transformation? Potential candidates are genes that are regulated by AP- 1 but most probably only a
subset of these genes will play an important role in cellular transformation.
We have recently characterized the DNA methyltransferase 5' region and demonstrated that
it contains at least two functional AP-l sites (34) and that the promoter of the DNA MeTase could
be dramatically transactivated by Fos, Jun or Ras (Rouleau et al., unpublished results). The DNA
MeTase gene encodes an activity that is responsible for methylating cytosine residues in the
dinucleotide sequence CpG (5,31). A h~llm~rk of DNA methylation is the fact that 80% of the
CpG sites are methylated and that these sites are distributed in a nonrandom manner to generate a
pattern of methylation that is site-, tissue- and gene-specific (31,32, 49). Methylation patterns are
formed during development: establishment and maintenance of the appropriate pattern of
methylation is critical for development (22) and for defining the differentiation state of a cell
(18,33,46). The pattern of methylation is maintained by the DNA MeTase at the time of
replication (41) and the level of DNA MeTase activity and gene expression is regulated with the
growth state of different primary (41) and immortal cell lines (45). This regulated expression of
DNA MeTase has been suggested to be critical for preserving the pattern of methylation (41, 44).
We reasoned that an activity that has a widespread impact on the genome such as DNA
MeTase would play a role in cellular transformation. This hypothesis is supported by many lines of
evidence that have demonstrated aberrations in the pattern of methylation in transformed cells.
While many reports show hypomethylation of both total genomic DNA (15) as well as individual
genes in cancer cells (14), other reports have indicated that hypermethylation is an important
characteristic of cancer cells (12). First, large regions of t'ne genome such as CpG rich islands (1)
or regions in chromosomes 17p and 3p that are reduced to homozygosity in lung and colon cancer
211d213
respectively are consistently hypermethylated (9,25). Second, the 5' region of the retinoblastoma
(Rb) and Wilms Tumor (WT~ genes is methylated in a subset of tumors, and it has been suggested
that inactivation of these genes in the respective tumors resulted from methylation rather than a
mutation (29). Third, the short arm of chromosome 11 in certain neoplastic cells is regionally
hypermethylated (9). Several tumor suppressor genes are thought to be clustered in that area (35).
If the level of DNA MeTase activity is critical for maintaining the pattern of methylation as has been
suggested before (41,44), one possible explanation for this observed hypermethylation is the fact
that DNA MeTase is dramatically induced in many tumor cells well beyond the change in the rate of
DNA synthesis (12,19). The fact that the DNA MeTase promoter is activated by the Ras-AP-l
sign~lling pathway is consistent with the hypothesis that elevation of DNA MeTase activity and
resulting hypermethylation in cancer is an effect of activation of the Ras-Jun signalling pathway
(Rouleau et al., unpublished).
The lines of evidence that link cancer and hypermethylation are however still circumstantial.
The critical question that remains to be answered is whether these changes in DNA methylation
play a causal role in carcinogenesis. To address this question, we have chosen the adrenocortical
carcinoma cell line Yl as a model system. Yl cells are derived from a naturally occurring
adrenocortical tumor in LAFl mice and they bear a 30-40 fold amplification of the Ras
protooncogene (39). We have recently shown that the transformed state of these cells could be
reversed by forced expression of hGAPl20 suggesting that the Ras signalling pathway controls the
tumorigenicity of Yl cells (Macleod et al., unpublished). If hypermethylation is critical for
oncogenesis, then the transformed state of a cell should be reversed by partial inhibition of DNA
methylation. We have previously demonstrated that forced expression of an "antisense" mRNA to
the most 5' 600 bp of the DNA MeTase message (pZaM) can induce limited DNA demethylation
in 10 T 1/2 cells (46). To directly test the hypothesis that the tumorigenicity of Yl cells is
controlled by the DNA MeTase, we transfected either pZocM or a pZEM control into Yl cells. We
demonstrate that limited demethylation of Yl DNA results in reversal of the tumorigenic phenotype
suggesting that DNA MeTase plays a critical role in tumorigenesis.
2110213
Material and Methods
Cell Culture and DNA mediated gene transfer
Y1 cells were m~int~ined as monolayers in F-10 medium which was supplemented with
7.25% heat inactivated horse serum and 2.5% heat inactivated fetal calf serum (Immunocorp,
Montreal) (48). All other media and reagents for cell culture were obtained from GIBCO-BRL.
Y1 cells (lX106) were plated on a 150 mm dish (Nunc) 15 hours before transfection. The pZaM
expression vector encoding the 5' of the murine DNA Methyl Transferase cDNA (10ug) was
cointroduced into Y1 cells with 1 ~lg of pUCSVneo as a selectable marker by DNA mediated gene
transfer using the calcium phosphate protocol (2). Selection was initiated 48 hours after
transfection by adding 0.25 mg/ml G418 (GIBCO-BRL) to the medium. G418 resistant cells were
cloned in selective medium. For analysis of growth in soft agar, lX103 cells were seeded in
triplicate onto 30 mm dishes (Falcon) with 4 ml of F-10 medium containing 7.5% horse serum,
2.5% FCS, 0.25 mg/ml G418 (for transfectants) and 0.33% agar solution at 37C (16). Cells
were fed with 2 ml of medium plus G418 every two days. Growth was scored as colonies
containing >10 cells, 21 days after plating.
DNA and RNA analyses
Preparation of genomic DNA and total cellular RNA, labelling (using the random primer
labelling kit from Boehringer Mannheim), blotting RNA on to Hybond-N+(Amersham), and all
other standard molecular biology manipulations were performed according to Ausubel et al. (2).
MspI and HpaII restriction enzymes ( Boehringer Mannheim ) were added to DNA at a
concentration of 2.5 units/ug for 8 h at 37C. Radionucleotides (3000mCi/mmol) were purchased
from Amersham.
Nearest neighbour analysis
Two llg of DNA were incubated at 37C for 15 minutes with 0.1 unit of DNAase, 2.5 !al
of 32P-a-dGTP (3000 Ci/mmol from Amersham) and 2 units of Kornberg DNA polymerase
211~213
-- 4
(Boehringer) were then added and the reaction was incubated for an additional 25 minutes at 30C.
50 ~11 of water were added and the nonincorporated nucleotides were removed by spinning through
a microcon column (Amicon) at maximum speed for 30 seconds. The labelled DNA (20 ,ul) was
digested with 70 ,ug of micrococal nuclease (Pharmacia) in the manufacturer's recommended buffer
for 10 hours at 37C. Equal amounts of radioactivity were loaded on TLC phosphocellulose
plates (Merck) and the 3' mononucleotides were separated by chromatography in one dimension
(iso-butyric acid: H20: NH40H in the ratio 66:33:1). The chromatograms were exposed to XAR
film (Eastman-Kodak) and the spots corresponding to cytosine and S-methylcytosine were scraped
and counted in a ,~-scintillation counter.
Generation of p53 and retinoblastoma (RB) probes by PCR
Oligoprimers for the S'region of the mouse pS3 gene were selected from the published
genomic sequence (Accession number: X01235) (17) using the Primer selecting program (PC
Gene). The S' primer corresponding to bases 154-172: 5'TLC GAA TCG GTT TLC ACCC 3' and
the 3' primer corresponding to bases 472-492 5' GGA GGA TGA GGG CCT GAA TGC 3' were
added to an amplification recation n~ix ~Ul~ containing 100 ng of mouse DNA (from C2C12 cells)
using the incubation conditions recommended by the manufacturer (Amersham Hot tub) ( 1.5 mM
MgC12) and the DNA was amplified for 40 cycles of 2 minutes at 95C, 2 minutes at 55C and
0.5 minutes at 72C . The reaction products were separated on a low-melt agarose gel (BRL) and
the band corresponding to the expected size was excised and extracted according to standard
protocols (2).
Since the genomic sequence of the mouse RB gene was unavailable through Genbank we reverse
transcribed the retinoblastoma mRNA from 0.5 ~g of total mouse RNA (from C2C12 cells) using
random oligonucleotide primers (Boehringer) with Superscript reverse transcriptase (BRL) under
conditions recommended by the manufacturer. The RB sequence was amplified from the reverse
transcribed cDNA using oligonucleotides corresponding to bases 2-628 of the published cDNA
(4). The oligoprimers used were 5' GGA CTG GGG TGA GGA CGG 3' (1-18) and 5' TTT CAG
2~1V213
TAG ATA ACG CAC TGC TGG 3' (620-610). The amplification conditions were as described
above.
Tumorigenicity assays
LAF-l mice (6-8 week old males) were injected subcutaneously (in the flank area) with 106
cells. Mice were monitored for the presence of tumors by daily palpitation. Mice bearing tumors
of greater than 1 cm in diameter were sacrificed by asphyxiation with CO2, tumors were removed
by dissection and homogenized in guanidium isothiocyanate. Mice that were tumor free were kept
for ninety days and then sacrificed. RNA was prepared from the tumors by CsCl2 density gradient
centrifugation as described (2).
Electron microscopy
Cells were fixed in glutaraldehyde (2.5%) in cacodylate buffer (O.lM) for one hour and
further fixed in 1% osmium tetroxide. The samples were dehydrated in ascending alcohol
concentrations and propylene oxide followed by embedding in Epon. Semithin sections (l~m)
were cut from blocks with an ultramicrotome, counterstained with uranil acetate and lead citrate.
Samples were analyzed using a Philips 410 electron microscope (26).
Results
Expression of antisense to the DNA Methyltransferase gene in Yl cells results inlimited DNA demethylation.
To directly inhibit DNA methylation in Yl cells, we introduced either the DNA MeTase
antisense expression construct pZaM or a pZEM control vector (46) into Yl cells by DNA-
mediated gene transfer as described in Material and Methods. G418-resistant colonies were
isolated and propagated for both constructs. To confirm that the transfectants bear the introduced
construct, we prepared DNA from the transfectants and subjected it to digestion by either MspI or
HpaII, Southern blot analysis and hybridization with a 32p labelled 0.6 kb DNA MeTase cDNA
~llV~13
-- 6 --
fragment (Fig. lA). The results presented in Fig. lB demonstrate that the three pZaM
transfectants contained significant levels of the DNA MeTase cDNA sequence while the control
transfectants were clean. To test whether the pZaM constructs is expressed in the transfectants
and whether the metallothionein promoter is functional in these cells, we cultured the transfectants
with 50~1M of ZnS04, prepared RNA at different time points, subjected it to Northern blot
analysis and hybridization with the 32p labelled MET 0.6 probe. As observed in Fig. lC the
transfectants 7 and 9 express substantial amounts of the MET 0.6 cDNA (~1.3 kb chimeric
mRNA) even before induction with ZnS04. The ZnS04 induction increases the relative intensity
of a 1.3 kb RNA hybridizing to MET 0.6 suggesting that ZnS04 induces the initiation of
transcription from a discrete site but not the total expression of the antisense message (resulting in a
smear in the non induced RNA samples). We therefore did not induce our transfectants with
ZnS04 in the experiments described in this paper.
To determine whether expression of antisense RNA to the DNA MeTase gene leads to a
general reduction in the level of methylation of the genome, we resorted to "nearest neighbour"
analysis using [a-32P]-dGTP as previously described (33). This assay enables one to determine
the percentage of methylated cytosines residing in the dinucleotide sequence CpG (33).
Transfectants and control DNAs were nicked with DNAaseI, nick translated with a single
nucleotide [a-32P]-dGTP using DNA polymerase I and the labelled DNA was digested to 3'
mononucleotide phosphates with micrococal nuclease which cleaves DNA 3' to the introduced a-
32p. The [a-32P] labelled S' neighbours of dGMP were separated by chromatography on a TLC
plate, the resulting spots for dCMP and dCmetMP were scraped and counted by liquid scintillation.
The results of a triplicate experiment presented in Fig. 2a (sample autoradiogram) and b (graphic
representation) suggest that a limited but significant reduction in the total level of DNA methylation
(12 % for transfectant number 4 and 22% for 7) occurred in transfectants expressing the pZaM
constr~ict when compared to the control line pZEM.
2 1 ~
Demethylation of specific genes in Y1 pZo~M transfectants.
To further verify that expression of pZaM results in demethylation and to determine whether
specific genes were demethylated, we resorted to a HpaII/MspI restriction enzyme analysis
followed by Southern blotting and hybridization with specific gene probes. HpaII cleaves the
sequence CCGG, a subset of the CpG dinucleotide sequences, only when the site is unmethylated
while MspI will cleave the same sequence irrespective of its state of methylation. By comparing
the pattern of HpaII cleavage of specific genes in cells expressing pZocM with that of the parental
Yl or cells harboring only the vector, we determined whether the genes are demethylated in the
antisense transfectants. We first analyzed the state of methylation of the steroid 21-hydroxylase
gene C21 (42,43). This gene is specifically expressed and hypomethylated in the adrenal cortex
but is inactivated and hypermethylated in Yl cells (42,43). We have previously suggested that
hypermethylation of C21 in Yl cell is part of the transformation program that includes the shut
down of certain differentiated functions (43). DNA prepared from Yl, pZo~m (4,11) and pZEM
(4) transfectants was subjected to either MspI or HpaII digestion, Southern blot analysis and
hybridization with a 0.36 kb Xba-BamHI fragment containing the enhancer and promoter regions
of the C21 gene (see references 42 and 43 for physical map of the probe). This probe should
detect 0.36 kb and 0.16 kb HpaIl fragments when the promoter region is fully demethylated
(42,43). The promoter and enhancer region is heavily methylated in Yl cells and the pZEM
transfectants as indicated by the presence of the higher molecular weight partial HpaII fragments at
3.8 and 2 kb and the absence of any lower molecular weight fragments (Fig. 2c). In contrast, the
Yl pZccM transfectants bear a partially demethylated C21 5' region as indicated by the relative
diminution of the 3.8 and 2 kb fragments and the appearance of the fully demethylated faint bands
at 0.36 kb as well as as the fact that HpaII cleavage yields partial fragments at 0.56 and ~1 kb
indicating partial hypomethylation of sites upstream and downstream to the enhancer region (Fig.
2c). To determine whether hypomethylation was limited to the enhancer region or is it spread
throughout the C21 gene locus we performed a similar HpaII digestion and Southern blot transfer
on different preparations of DNA extracted from Yl cells, a control pZEM (4) transfectant and
-
211~13
three pZo!M antisense transfectants (Fig. 2d) and hybridized the filter with a 3.8 kb BamHI
fragment containing the body of the C21 gene and 3' sequences (see reference 42 and 43 for
physical map). Full demethylation of this region should yield a doublet at ~1 kb, a 0.8 kb
fragment and a 0.4 kb fragment as well as a number of low molecular weight fragments at 0.1-
0.2 kb. As observed in Fig. 3c the C21 locus is heavily methylated in Yl cells as well as the
control transfectant as indicated by the high molecular weight fragments above 23 kb. Only a faint
band is present in the expected 1 kb molecular weight range as well as a partial at 1.9 kb (Fig. 2c).
The DNA extracted from the antisense transfectants exhibits a relative diminution of the high
molecular weight fragments and relative intensification of the partial fragment at 1.9 kb as well as
the appearance of new partial fragments in the lower molecular weight range between 1 and 0.4 kb
indicating partial hypomethylation at large number of HpaII sites contained in the 3' region of the
C21 gene (42,43). The pattern of demethylation, indicated by the large number of partial HpaII
fragments (Fig. 2c), is compatible with a general partial hypomethylation rather than a specific loss
of methylation in a distinct region of the C21 gene.
To determine whether demethylation is limited to genes that are potentially expressible in
Yl cells such as the adrenal cortex-specific C21 gene (43) or if the demethylation is widely spread
in the genome, we have analyzed other genes such as the muscle specific MyoD gene as well as the
hippocampus specific 5HTlA receptor gene; both genes were hypomethylated (data not shown).
Another class of genes that might have undergone a specific hypomethylation includes the tumor
suppressor genes. We determined the state of methylation of two genes from this class, pS3 and
retinoblastoma (RB) which are both tumor suppressor genes involved in cell cycle regulation .
Loss of either one of these gene products has been shown to lead to deregulation of the cell cycle
and neoplasia (4,10). Using a probe to a 300 bp sequence from the 5' region of the mouse RB
cDNA we were able to determine the lèvel of methylation of this gene in Yl cells transfected with a
control vector as well as the pZaM transfectants (Fig. 2e). Cleavage of this region with HpaII
yields 0.6 kb and 0.1 kb fragments (Fig. 2e). The RB locus is heavily methylated in the control
cells as indicated by hybridization of the probe to high molecular weight fragments. This locus is
partially hypomethylated in the pZocM transfectants as indicated by the relative diminution in the
2~1~213
g
intensity of the high molecular weight fragments, the appearance of numerous partial fragments
between 23 and 0.6 kb and the appearance of the demethylated fragments at 0.6 kb and ~0.1 kb.
The p53 locus was studied using a 0.3 kb fragment from the 5' region 300 bp upstream to
the initiation site as a probe (Fig. 2f). Cleavage of the pS3 loci (two p53 genes are present in the
mouse genome ) with MspI yields fragments in the 4.4, 2.5, 0,56 and 0.44 kb molecular weight
range (Fig. 2f, first lane). Cleavage of the control Yl pZEM transfectants shows that only the sites
flanking the 0.56 kb fragments are demethylated in Yl cells. The rest of the locus is heavily
methylated as indicated by the intensity of the signal at the >4.4 kb range (Fig. 2f, lanes 2-4). In
comparison to the control transfectants the pS3 gene is partially hypomethylated in Yl cells
e~ essi,lg an antisense message to the DNA MeTase as implied by the relative reduction in the
intensity of the high molecular weight fragments above 4.4 kb and appearance of the 4.4 kb HpaII
fragment, the partially cleaved HpaII fragment at 4 kb, the faint partial fragment around 3.5 kb and
the faint fragment at 2.5 kb (Fig. 2f last three lanes). These results further substantiate the
conclusion that expression of an antisense to the DNA MeTase results in a genome wide par~ial
hypomethylation. Neither of the genes studied demonstrates a distinct selectivity in demethylation.
Morphological transformation and loss of anchorage independent growth of Y1
cells expressing antisense to the DNA MeTase.
To determine whether demethylation induced by the DNA MeTase antisense constructresults in a change in the growth properties of cancer cells we compared the growth and
morphological characteristics of the pZ~M transfectants versus the controls. To compare the
growth curve of pZaM transfectants and controls, we plated 5x104 Yl pZEM and pZo~M
transfectants (4 and 7) cells in triplicate, the cells were harvested and counted at the indicated time
points (Fig. 6A). The results of this experiment show that the antisense transfectants reach
saturation density at lower concentrations than the control cells suggesting that the transfectants
have reacquired "contact inhibition" which is one of the traits lost in cancer cells. The
211~13
~ 10 -
morphological properties of the Yl pZaM transfectants further support this conclusion (Fig.3A).
While control Yl and Yl pZEM cells exhibit limited contact inhibition and form multilayer foci,
Yl pZaM transfectants exhibit a more rounded and distinct morphology and grow exclusively in
monolayers (Fig.3A).
To determine whether the expression of antisense to the DNA MeTase results in reversal of
the tumorigenic potential we determined the ability of the transfectants to grow in an anchorage
independent fashion which is considered an indicator of tumorigenicity (16). The Yl pZaM
transfectants demonstrate an almost complete loss of ability to form colonies in soft agar, moreover
the colonies that do form contain only a few cells as demonstrated (Fig.3B). Growth on soft agar
was quantified by visual examination and presented graphically in Fig.3C. These experiments
demonstrate that inhibition of DNA methylation by expression of an antisense message to the DNA
MeTase leads to loss of tumorigenicity in vitro.
Yl cells expressing antisense to the DNA MeTase exhibit decreased
tumorigenicity in vivo
To determine whether demethylation can result in inhibition of tumorigenesis we injected
lX106 cells for each of the Yl pZaM, Yl and Yl pZEM transfectants subcutaneously into the
syngeneic mouse strain LAF-l. The presence of tumors was determined by palpitation. While all
the animals injected with Yl cells formed tumors two to three weeks post injection, the rate of
tumor formation in the animals injected with the pZaM transfectants was significantly lower (Fig.
4A; p>O.005).
Many lines of evidence suggèst that angiogenic potential and metastatic potential of cell
lines are directly related (23). The tumors that do arise from the pZaM transfectants exhibit very
limited neovascularization (Fig. 4B) while tumors that formed in the animals that were injected
with Yl cells or control transfectants were highly vascularized (Fig. 4B). This difference in
neovascularization is indicated by the pale color of the homogenates of tumors removed from
21~21~
animals injected with Yl pZocM transfectants cells versus the very dark homogenates of tumors
arising from control lines (Yl and YlpZEM ) ( Fig.4B).
One possible explanation for the fact that a small number of tumors did form in animals
injected with the pZaM transfectants is that they are derived from revertants that lost expression of
the antisense to the DNA MeTase under the selective pressure in vivo. To test this hypothesis we
isolated RNA from a tumor arising from the YlpZaM transfectant and compared the level of
expression of the 0.6 kb antisense message with the one observed for the transfectant line in vitro.
The isolated RNAs were subjected to Northern blot analysis and hybridization with a 32p labelled
MET 0.6 fragment. The filter was stripped of Its radioactivity and was rehybridized with a 32P
labelled oligonucleotide probe for 18S rRNA (Fig. 5A) as previously described (43). The
autoradiograms were scanned and the level of expression of MET 0.6 was determined relative to
the signal obtained with the 1 8S probe (Fig. SB). The expression of the antisense message is
significantly reduced in the tumors supporting the hypothesis that expression of an antisense
message to the DNA MeTase is incompatible with tumorigenesis.
Expression of pZaM in Yl cells leads to an induction of an apoptotic death
program upon serum deprivation.
Tumor cells exhibit limited dependence on serum and are usually capable of serumindependent growth (3). Factors present in the serum are essential for the survival of many
nontumorigenic cells. Several lines of evidence have recently suggested that the enhanced
survivability of tumorigenic cells is associated with inhibition of programmed cell death. For
example, the oncogene bc1-2 is not a stimulator of cell proliferation but rather causes inhibition of
apoptosis (40). The tumo~ suppressor pS3 can induce apoptosis in a human colon tumor derived
line (37) and certain chemotherapeutic agents have been shown to induce apoptosis in cancer cells
(11). As observation of the pZo~M transfectants indicated that they expressed enhanced
dependence on serum and limited .s~lrvivability under serum deprived conditions, we detennined
whether demethylation can indllce an apoptotic program in Yl cells. We reasoned that as factors in
1 3
the serum are known to act as survival factors for cells, an apoptotic program could be activated
only when these factors are remove. To test whether pZaM transfectants undergo programmed
cell death under serum deprived condition, we studied the effects of serum starvation on these
transfectants. pZaM transfectants and control Yl pZEM transfectants ( 3xlO5 per well) were
plated in low serum medium (1% horse serum) in six well plates, harvested every 24 hours and
tested for viability by trypan blue staining (Fig. 6B). While the control cells exhibited almost
100% viability up to 72 hours after transfer into serum deprived medium, the Yl pZaM cells
showed up to 75 % loss of viability at 48 hours (Fig. 6B).
The rapid onset of death in Yl pZaM clones under serum deprived conditions suggests
that an active process may be involved. Several observable changes distinguish apoptosis from
necrosis: apoptosis is an active process requiring de novo protein synthesis (38); apoptosis is
associated with death of isolated cells, unlike necrosis where patches or areas of tissue die; cells
dying by apoptosis do not elicit an immune response; and the most diagnostic feature of apoptosis
is the pattern of degradation of the DNA from apoptotic cells (13,24, 30, 36, 47). DNA from cells
dying by apoptosis generally exhibit a characteristic ladder when analyzed by gel electrophoresis
because Ca2+/Mg2+ dependent endonucleases cleave the DNA at internucleosomal regions (13).
Although the appearance of the 180 bp internucleosomal ladder is a diagnostic feature of apoptotic
death (13), other morphological changes such as chromatin condensation, cellular fragmentation,
and formation of apoptotic bodies are generally considered to be earlier events in the apoptotic
process and therefore also serve as useful markers (47). To test whether the serum deprived Yl
pZaM cells were dying as a result of an activated apoptotic death program, cells were plated in
starvation medium ( 1% horse serum) and harvested at 24 hour intervals. Total cellular DNA was
isolated from the cells and was subjected to electrophoresis on a 1.5% agarose gel followed by
transfer to nylon membrane and hybridization with random labeled Yl genomic DNA. After 48
hours in serum starved conditions, pZ(xM transfectants exhibit the characteristic 180 bp
internucleosomal DNA ladder while the control pZEM transfectants show no apoptosis at this time
1 3
-- 13 --
point (fig. 6C).
To determine whether cells expressing antisense to the DNA MeTase exhibit early
morphological markers of apoptosis, cells were serum starved for 24 hours (2% horse serum),
harvested and analyzed by electron microscopy. Fig.6D shows the electron micrographs of control
Y1 pZEM and Yl pZaM transfectants at various magnifications (I-V). The control cells have a
fine uniform nuclear membrane whereas the pZaM cells exhibit the cardinal markers of apoptosis
(47): condensation of chromatin and its margination at the nuclear periphery (panels I and II),
chromatin condensation (panel II), nuclear fragmentation (panel III), formation of apoptotic
bodies (panel V) and cellular fragmentation (panel IV). This set of experiments suggests that one
possible mechanism through which demethylation can inhibit tumorigenesis is by activating
programmed cell death.
.
Dlscusslon:
This paper tests the hypothesis that DNA hypermethylation plays a causal role intumorigenesis by expressing an antisense message to the DNA MeTase in an adrenocortical
carcinoma cell line. Expression of an antisense DNA MeTase (Fig. 1) leads to: (i) a general
reduction in the methylation content of the genome (Fig.2a), (ii) demethylation of regions
aberrantly methylated in this cell line such as the adrenal specific 21-hydroxylase gene (Fig. 2c,d)
as well as tumor suppressor loci (Fig.2e,f.), (iii) morphological changes indicative of inhibition
of the transformed phenotype, (iv) inhibition of tumorigenesis in vitro as determined by soft agar
assays (Fig. 3), (v) inhibition of tumorigenesis in syngeneic mice in vivo as well as a loss of
angiogenic function (Fig. 4) and (vi) to the induction of an apoptotic death program upon serum
deprivation (Fig. 6). These experiments strongly support the hypothesis that hypermethylation
plays a critical role in maintenance of the transformed state. Our data can explain previous
observations demonstrating an increase in DNA MeTase activity ~12,19) in cancer cells by
suggesting that this increase is critical for the transformed state. The fact that the RAS signalling
pathw~y has been shown to induce the activity of the DNA MeTase promoter provides us with a
mechanism to explain this increase in the DNA methylation capacity of cancer cells. It stands to
2 ~ 2 1 3
- 14 --
reason therefore that the DNA MeTase is an impol Lant effector of the RAS signalling pathway.
What is the possible mechanism by which hypermethylation can cause cellular
transformation? One plausible explanation that is supported by extensive data is that
hypermethylation results in inactivation of tumor suppressor genes. This hypothesis is supported
by the fact that CpG islands are methylated in cancer cells, in vitro methylation of CpG islands
leads to inactivation of transcription (29), and the methylation of tumor suppressor genes RB and
Wilms Tumor is associated with tumorigenesis. Hypermethylation might also explain the
inactivation of dirr~l~nLiated functions in tumor cells such as the heavy methylation and inhibition
of C21 expression in Yl cells (43). One obvious question is how does the general elevation in
DNA MeTase activity result in a methylation of a specific subset of sites in the genome? There
must be factors that interact with these regions and target them for methylation as has been
suggested before (42,44). The identification of these factors will be of critical significance in
understanding tumorigenesis. Similarly, the general hypomethylation induced by the antisense
message (Fig. 2) is most probably targeted at specific sites in the genome which are highly
sensitive to changes in DNA MeTase activity. However, there is no evidence that tumor
suppressor genes are the critical targets for hypermethylation in cancer cells or that tumor
suppressor genes are selectively demethylated in the DNA MeTase antisense transfectants. While
we demonstrate demethylation of pS3 and RB in the pZocM transfectants (Fig. 2 e,f), the level of
demethylation is not remarkably different than that observed for other sequences (Fig. 2c,d) or the
general genomic demethylation. Our unpublished data do not suggest any induction in the level of
expression of these genes in the pZ~cM transfectants.
Induction of apoptosis might be another possible explanation for inhibition of
tumorigenesis by demethylation. Several studies have suggested that cell death is triggered by an
endonuclease activity (13~. DNA hypermethylation might be inhibiting one of the genes involved
in induction of the death program. Alternatively, hypermethylation might result in modification of
the recognition sites for this putative endonuclease. While inhibition of gene expression by
methylation is the best analyzed function of DNA methylation, one should bear in mind that any
function of the genome might be modified by methylation. Sites that are especially sensitive to
2110~1~
changes in methylation might be controlling DNA functions such as repair, replication and
susceptibility to death-program related endon~ e~es The identification of these ç~n(licl~te sites
is ess~.nti~l to the underst~n(ling of the role that DNA methylation might play in the generation of
cancer.
One question that remains to be answered is how to explain the contradiction between the
fact that DNA MeTase is ovel~Apressed in cancer cells and the observed hypomethylation of the
genome of many cancer cells (14,15)? We would like to suggest that hypomethylation is a
secondary event, hypermethylation being the primary event that induces changes in the chrolllali-
structure of many sites and results in hypomethylation. Certain sites such as a CpG-island rich
area on chromosome 17p, which is reduced to homozygosity in lung and colon cancer, as well as
regions of chromosome 3p, that are con~i~t~ntly reduced to homozygosity in lung cancer, remain
hypermethylated (9,25). The maintenance of methylation at these sites is dependent on high levels
of DNA MeTase activity.
While additional experiments will be required to address these questions, this paper
demonstrates that inhibition of DNA methylation leads to a reversal of the Ll~nsrolllled state and
provides solid evidence that DNA methylation plays a critical role in cellular ll~ ,ro.lllalion.
In accordance with the present invention, antisense oligonucleotides to DNA MeTase are
disclosed as examples which are given to illustrate the invention rather than to limit its scope.
In accordance with the present invention, the inhibition of DNA MeTase may be carried
out using anti-DNA MeTase or antisense oligonucleotides to DNA MeTase. Such an inhibition
would be useful for the tre~tment of ~ e~es, such as cancer, sickle cell anemia and tl~ sçmi~
While the invention has been described in connection with specific embodiments thereof, it
will be understood that it is capable of further modifications and this application is intended to
cover any variations, uses, or adaptations of the invention following, in general, the principles of
the invention and incl~ ing such departures from the present disclosure as come within known or
customary practice within the art to which the invention pertains and as may be applied to the
essential features hereinbefore set forth, and as follows in the scope of the appended claims.
2 1 3
- 16 -
FIGURE LEGENDS
Fig.1. Expression of pZaM in Y1 adrenocortical cells.
A. The physical maps of the pl~mi~ls pZEM and pZaM were described in reference (46)
and are illustrated. The metallothionine (MT) promoter (shaded box), the human growth hormone
3' region (HGH) (open bar), and the MeTase cDNA sequences (hatched) are indicated.
B. To verify the presence of the transfected plasmid in the transfectants, genomic DNA was
subjected to Southern blot analysis and hybridization to the 0.6 kb MET region probe.
C. Positive clones expressed the expected 1.3 kb chimeric mRNA as determined by Northern blot
analysis . Total RNA (5ug) prepared from the three pZaM lines (7 and 9) and from the pZEM
transfectants was subjected to Northern blot analysis and hybridization with the 0.6 Kb DNA
MeTase cDNA probe (described in the physical map).
Fig.2. State of methylation of total genomic DNA and specific genes in YlpZaM
transfectants.
Nearest neighbour analysis of 2 ,ug DNA extracted from pZaM transfectant (4) and a
pZEM control was performed in triplicates as described in Material and Methods. a. An
autoradiogram of a representative TLC plate. The standard lane is of hemimethylated M13 DNA
synthesized in vitro. b. The spots corresponding to C and 5-methyl C were scraped and counted
in a liquid ~ scintillation counter. The values represent the means + SEM.
c. State of methylat;on of the C21 5'region. Genomic DNA (lOug) was extracted from
the transfected lines and subjected to digestion with either MspI(M) or HpaII(H), Southern blot
transfer and hybridization with a 32p labeled DNA probe (0.36 bp XbaI-BamHI probe, ) to the 5'
region of the mouse C21 gene (see reference 43 for description). The open arrows indicate
expected HpaII fragments resulting from demethylation of the different sites in the pZaM
211~13
,
transfectants. Complete digestion of the region will yield 0.36 and 0.16 kb fragments. MspI
cleavage of these sites is partial because they are nested within a Haem site which is usually
resistant to MspI cleavage when the internal cytosine is methylated (20).
d. State of methylation of the C21 gene. Genomic DNA (lOug) extracted from the
indicated lines was subjected to MspI/HpaII Southern blot transfer and hybridization with a 32p
labeled DNA probe for the mouse C21 gene (3.8 kb BamHI fragment) (42). The open arrows
indicate the position of demethylated fragments.
e. State of methylation of the retinoblastoma (RB) gene. A Southern blot MspI/HpaII
analysis of DNA extracted from the indicated lines and hybridization with a 32p labeled PCR
fragment correspoding to bases 2-628 of the RB cDNA. The open arrows indicates the position of
hypomethylated fragments.
f. State of methylation of the p53 gene. Genomic DNA (lOug) from the indicated lines
was isolated and treated as above, subjected to Southern blot transfer and hybridization with a
32p labeled PCR fragment corresponding to the 5' region of the pS3 gene (154-472 from the
sequence published in reference ). The open arrow indicates the position of the demethylated
fragments in the pZaM transfectants.
Fig.3. Morphological transformation and reduced anchorage independent growth
of Y1 cells transfected with pZaM.
A. Phase contrast microscopy at X200 magnification of living cultures of Yl clonal transfectants
with pZaM and Yl controls.
~110~13
-- 18 --
B. Yl pZEM cells ( clones 4 and 7 ) and Yl pZaM transfectants (clones 4, 7 and 9 ) were plated
in triplicates at lX103 cells per well and grown in soft agar for 21 days. The pictures are phase
contrast microscopy at X10 of representative regions after 21 days in culture.
C. Anchorage independent growth assay: Yl pZEM (clone 4 and 7 ) and Yl pZaM transfectants
( clone 4, 7 and 9 ) were plated in triplicate in soft agar and colonies were counted visually after 21
days of growth. The data points represent an average of three determinations.
Fig. 4. In vivo tumorigenicity of YlpZaM transfectants.
(A) Two control lines (Yl and pZEM4) and three YlpZaM transfectants (4,7 and 9) were tested
for their ability to form tumors in LAF-l mice. Tumor formation was assessed by palpitation two
months after injection. The number of mice forming tumors as well as the level of
neovascularization in the tumors is tabulated . The statistical significance'of the difference between
the control an antisense transfectants was determined using a X2 test; p>0.005.
(B) Tumors isolated from animals injected with the indicated cell lines were homogenized and
photographed. Control lines (Yl and pZEM 4 ) produce a dark homogenate due to extensive
vascularization while pZaM transfectants produce pale homogenates due to lack of
neovasclll~li7~tion .
Fig. 5. Loss of antisense expression in tumors derived from YlpZaM
transfectants.
a. RNA (lOug) isolated from the indicated tumors was subjected to Northern blot analysis and
hybridization with the 0.6 Kb MET cDNA probe (see physical map). Expression of the 1.3 Kb
antisense message is seen only in the original cell line pZaM 7 (dark arrow), and is undetectable in
tumors arising from pZaM 7 or Yl cell lines. The filter was stripped of radioactivity and
2 ~ 3
-- 19 --
rehybridized with a 32P labeled oligonucleotide corresponding to 18s rRNA (43).
b. The MET and 18S autoradiograms were scanned using the Scanmaster 3+ Densitometer
(Scanalytics). The relative expression of the antisense was determined by normalizing the signal
obtained with the MET probe relative to the 18S signal. The expression of antisense in the tumor
is below the background signal observed in nontransfected Yl cells.
Fig.6. Survival and apoptosis of YlpZaM cells.
(A) Density restricted growth assay. Yl pZaM and control pZEM transfectants were
seeded at the density of 2x104 per well onto 6 well plates in triplicates and the number. The cells
were harvested at the indicated time points and counted. The results represent an average of three
different wells + SEM.
(B) YlpZaM transfectants were seeded at the density of 2x104 per well onto 6 well plates in
triplicates in regular medium. The medium was replaced after 24 hours with serum-deprived
medium (1% horse serum). The cells were harvested at the indicated time points and the
percentage of viable cellls was determined using trypan blue staining.
(C) The indicated transfectants were plated in 1% serum containing medium and harvested after 1
and 2 days. Total cellular DNA was isolatea, separated by agarose gel electrophoresis, transfered
to nitocellulose membrane and probed with 32p labeled Yl genomic DNA. A 130bp
internucleosomal ladder characteristic to cells dying via apoptosis can be seen in the YlpZaM
transfectants only.
(D) Y1 transfectants were grown in 1% serum medium for 24 hours, fixed and analyzed by
electron microscopy for early signs of apoptotic death, ( I-V ) are various sections (the
magnification is indicated) of Yl pZaM transfectants, pZEM 4 are the control transfectants.
2 1 3
-- 20 --
Fig. 7. illustrates the Regulation mech~ni~m of DNA MeTase promoter which dele~ es
DNA methylation patterns and cellular transformation.
211~213
-- 21 --
References:
1. Antequara, F., J. Boyes, and A. Bird. 1990. High levels of de novo methylation and
altered chromatin structure at CpG islands in cell lines. Cell 62: 503-514.
2. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. A. Smith, J. G.
Seidman, and K. Struhl. 1988. Current Protocols in Molecular Biology. Wiley and Sons,
New York.
3. Barns, D., and G. Sato. 1980. Serum-free cell culture: A unifying approach. Cell 22:649-
655.
4. Bernards, R., G. Schakleford, M. Gerber, J. Horowitz, S. Friend, M. Schartl,
E. Bogenm~nn, J. Rapaport, T. McGee, T. Dryja, and R.A. Weinberg. 1989.
Structure and expression of the murine retinoblastoma gene and characterization of its encoded
protein. Proc. Natl. Acad. Sci. USA. 86: 6474-6478.
5. Bestor, T., A. Laudano, R. Mattaliano, and V. Ingram. 1988. Cloning and
sequencing of a cDNA encoding DNA methyltransferase of mouse cells. J. Mol. Biol. 203: 971-
983.
6. Binetruy, B., T. Smeal, and M. Karin. 1991. Ha-RAS augments C-Jun activity and
stimulates phosphorylation of its activation domain. Nature 351: 122-127.
7. Bird, A. P. 1986. CpG-rich islands and the function of DNA methylation. Nature 321: 209-
213.
8. Bourne, H. R., and H. E. Varmus. 1992. Oncogenes and cell proliferation. Current
~l t~213
Opinion in Genetics and Development. 2: 1-3.
9. de Bustros, A., B. D. Nelkin, A., Silverman, G. Ehrlich, Poiesz, and S. B.
Baylin. 1988. The short arm of chromosome 11 is a "hot spot" for hypermethylation in human
neoplasia. Proc. Natl. Acad Sci. USA . 85: 5693-5697.
10. Donehoweer, L.A., M. Harvey, B. Slagle, M.J. Mcarthur, C.A. Montgomery,
J.S. Butel, and A. Bradley. 1992. Mice deficient for p53 are developmentally normal but
susceptible to spontaneous tumors. Nature 356: 215-221.
11. Collins, M. K. L., J. Marvel, P. Malde, and A. Lopez-Rivas. 1992. Interleukin
3 protects murine bone marrow cells from apoptosis induced by DNA damaging agents. J. Exp.
Med. 176: 1043-1091.
12. El-Deiry, W.S., B. D. Nelkin, P. Celano, R. W. Chiu-Yen, J. P. Falco, S.
R. Hamilton, and S. B. Baylin. 1990. High expression of the DNA Methyltransferase gene
characterizes human neoplastic cells and progression stages of colon cancer. Proc. Natl. Acad Sci.
USA 88: 3470-3474.
13. Ellis, R.E., J. Yuan, and Horvitz H.R. 1991. Mechanisms and functions of cell
death. Annu. Rev. Cell Biol. 7: 663-698.
14. Feinberg, A. P., and B. Vogelstein. 1983. Hypomethylation distinguishes genes of
some human cancers from their normal counterparts. Nature 301: 89-92.
15. Feinberg, A. P., C. W. Gehrke, K. C. Kuo, and M. Ehrlich. 1988. Reduced
genomic content of 5-methyl cytosine in human colonic neoplasia. Cancer Res. 48: 1159-1161.
16. Freedman, V. H., and S. Shin. 1974. Cellular tumorigenicity in nude mice: Correlation
1 3
-- 23 --
with cell growth in semi-solid medium. Cell 3: 355-359.
17. Zakut-Houri, R., M. Oren, B. Bienz, V. Lavie, S. Hazum, and D. Givol.
1983. A single gene and a pseudogene for the cellular tumor antigen p53. Nature 306: 594-597.
1~. Jones, P. A. 1985. Altering gene expression with 5-azacytidine. Cell 40: 485-486.
19. Kautiainen T. L., and P. Jones. 1986. DNA methyltransferase levels in tumorigenic
and nontumorigenic cells in culture. J. Biol. Chem. 261: 1594-1598.
20. Keshet, E., and H.Cedar. 1983. Effect of CpG methylation on MspI. Nucl.Acids.Res.
11: 3571-3580.
21. Kimura, K.I., and Truman, J. W. 1990. Postmetamorphic cell death in the nervous
and muscular systems of Drosophila melanogaster. J. Neuroscience 10: 403-411.
22. Li, E., T. H. Bestor, and R. Jaenisch. 1992. Targeted mutation of the DNA
methyltransferase gene results in embryonic lethality. Cell 69: 915-926.
23. Liotta, L.A., P. S. Steeg, and W. G. Steter-Stevenson. 1991. Cancer metastasis
and angiogenesis: An imbalance of positive and negative regulators. Cell 64: 327-336
24. Macdonald, H.R., and R. K. Lees. 1990. Programmed death of autoreactive
thymocytes. Nature 343: 642-643.
25. Makos, M., B. D. Nelkin, M. I. Lerman, F. Latif, B. Zbar, and S. B. Baylin.
1992. Distinct hyperrnethylation patterns occur at altered chromosome loci in human lung and
colon cancer. Proc. Natl. Acad Sci. USA 89: 1929- 1933.
21 ~13
- 24 -
26. Maysinger, D., P. Piccardo, J. Flipovic-Grcic and A. C. Cuello. 1993.
Microencapsulation of genetically engineered fibroblats secreting nerve growth factor. Neurochem
Intl. 23: 123-129.
27. McDonnell, T.J., and S.J. Korsmeyer. 1991. Progression from lymphoid hyperplasia
to high-grade malignant lymphoma in mice transgenic for the t(l4; 18). Nature 349: 254-256.
28. Nowell, P. C.1986. Mechanisms of tumor progression. Cancer Res. 46: 2203-2207.
29. Ohtani-Fujita, N., T. Fujita, A. Aoike, N.E. Osifchin, P.D.,Robbins, and T.
Sakai, 1993. CpG methylation inactivates the promoter activity of the human retinoblastoma
tumor-supressor gene. Oncogene 8: 1063-1067.
30. Raff, M. C. 1992. Social controls on cell survival and cell death. Nature 356: 397-400.
31. Razin, A., and Riggs, A. D. 19~0. DNA methylation and gene function. Science 210:
604-610.
32. Razin, A., and M. Szyf. 1984. DNA methylation patterns. Biochim. Biophys. Acta
782: 331-342.
33. Razin, A., E. Feldmesser, T. Kafri, and M. Szyf. 1985. Cell specific DNA
methylation paKerns and a nucleosome locking model for their function, p. 239. IN Razin, A.,and
G. L.Cantoni. (Ed) Biochemistry and Biology of DNA methylation, Allan R. Liss, Inc. N.Y.
34. Rouleau, J., G. Tanigawa, and M. Szyf. 1992. The mouse DNA Methyltransferase5' region; a unique housekeeping gene promoter. J. Biol. Chem. 267: 7368-7377.
35. Saxon, P.J., E.S. Srivatsan, and E.G. Stanbridge. 1986. Introduction of human
D~ 3
-- 25 --
chromosome 11 via microcell transfer controls tumorigenic expression of HeLa cells. EMBO J.,
5, 3461-3466
36. Schwartz, L.M., S. W. Smith, M. E. Jones, and B.A. Osborne. 1993. Do all
programmed cell death occur via apoptosis? Proc. Natl. Acad. Sci. 90: 980-984.
37. Shaw, P., S. Bovey, R. Tardy, and B. Sohli. 1992. Induction of apoptosis by wild
type p53 in human colon tumor derived cell line. Proc. Natl. Acad. Sci. 89: 4495-4499.
38. Schwartz, L.M., L.Kosz, and B.K. Kay. 1992. Gene activation is required for
developmentally programmed cell death. Proc. Natl. Acad. Sci. USA .1992.87: 6594-6598.
39. Schwab, M., K. Alitalo, H. E. Varmus, J. M. Bishop and D. George. 1983.
A cellular oncogene (c-Ki-ras) is amplified, overexpressed, and located within karyotyplc
abnormalities in mouse adrenocortical tumour cells. Nature 303: 497-501.
40.. Strasser, A., A. W. Harris, and S. Cory. 1990. Novel primitive lymphoid tumors
induced in transgenic mice by cooperation between myc and bc1-2 Nature. 348: 331-333.
41. Szyf, M., F. Kaplan, V. Mann, H. Giloh, E. Kedar, and A. Razin. 1985. Cell
cycle dependent regulation of eukaryotic DNA methylase level. J. Biol. Chem. 260: 8653-8656.
42. Szyf, M., B. P. Schimmer, and J. G. Seidman. 1989. Nucleotide sequence-
specific de novo methylation in a somatic murine cell line. Proc. Natl. Acad. Sci. USA 86: 6853-
6857.
43. Szyf, M., D. S. Milstone, B. P. Schimmer, K. L. Parker, and J. G. Seidman.
1990.- Cis Modification of the steroid 21-hydroxylase gene prevents its expression in the Yl
mouse adrenocortical cell line. Molecular Endocrinology 4: 1144-1152.
~10213
-- 26 --
44. Szyf, M. 1991. DNA methylation patterns: An additional level of information? Biochem.
Cell Biol. 64: 764-769.
45. Szyf, M., V. Bozovic, and G. Tanigawa. 1991. Growth regulation of mouse DNA
methyltransferase gene expression. J. Bol. Chem. 266: 10027-10030.
46. Szyf, M., J. Rouleau, J. Theberge, and V. Bozovic. 1992. Induction of myogenic
differentiation by an expression vector encoding the DNA methyltransferase cDNA sequence in the
antisense orientation. J. Biol. Chem. 267: 12831-12836.
47. Wyllie, A.H., G.J. Beatie, and A.D. Hargreaves. 1981.Chromatin changes in
apoptosis. Histochem. J.,13,681-692.
48. Yasumura, Y., V. Buonsassisi, and G. Sato. 1966. Clonal analysis of differentiated
function in animal cell cultures. I. Possible correlated maintenance of dirrel~n~iated function and
the diploid karyotype. Cancer Res.26: 529-535.
49. Yisraeli, J., and M. Szyf. 1985. Gene methylation patterns and expression, p. 353-
378. IN. Razin, A., H. Cedar, and A. G. Riggs (Ed), DNA methylation: Biochemistry and
Biological significance. Springer-Verlag, New York.