Download Small polydispersed circular DNA (spcDNA) in human cells

Oncogene (1997) 14, 977 ± 985
 1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00
Small polydispersed circular DNA (spcDNA) in human cells: association
with genomic instability
S Cohen1, A Regev and S Lavi
Department of Cell Research and Immunology, The George S Wise Faculty of Life Sciences, Tel-Aviv University, Tel-Aviv,
Israel, 69978
Small Polydispersed Circular DNA (spcDNA) was
suggested to be associated with genetically unstable
cells. However, until now, qualitative and quantitative
research has been limited due to the lack of ecient
methods for detection and analysis. We developed a twodimensional (2-D) neutral-neutral gel electrophoresis
assay for the identi®cation, characterisation and quantitation of spcDNA. Using this method, we established the
relation of spcDNA to genetic and induced genomic
instability in human cells, both in vitro and in vivo.
Enhanced amounts of spcDNA were found in genetically
unstable cells and tissues. spcDNA was detected in a
tumor cell-line (HeLa) and in tumor tissue (colon
carcinoma) as well as in ®broblasts derived from patients
su€ering from the genomic instability disease, Fanconi's
Anemia. We failed to detect spcDNA in the genetically
stable normal human ®broblasts. However, following
treatment with the initiating carcinogen MNNG, an
induction of spcDNA was observed. The level of spcDNA
was quanti®ed according to mitochondrial DNA
(mtDNA) standards. In light of these ®ndings, we discuss
the possible role of spcDNA as a marker and an
enhancer of genomic instability.
Keywords: genomic instability; spcDNA; two dimensional electrophoresis gene ampli®cation; cancer;
mutator phenotype
Introduction
Genomic instability, the state of constant change in the
genome, is a multifactorial phenomenon, encompassing
various dynamic structural processes. These include:
translocation, aneuploidy, recombination, deletion,
gene ampli®cation and the appearance of small
polydispersed circular DNA (spcDNA). Such phenomena are a major feature of genomes in tumor cells,
while they seldom occur in normal mammalian cells
(Fiedler and Hart, 1982; Nowell, 1976; Tlsty, 1990;
Tlsty et al., 1989).
Thus, it is accepted that the tumor cell genome is
intrinsically unstable relative to its normal counterpart.
Furthermore, it was proposed that genomic instability
plays a signi®cant role in the neoplastic process,
preceding tumorigenesis in vivo (Hartwell, 1992). This
happens in tumor-prone genetic diseases, such as
Correspondence: S Lavi
1
Current address: Institut Jaques Monod, Molecular Embryology, 2,
Place Jussieu-Tour 43 75251 Paris, Cedex 05, France
Received 8 August 1996; revised 28 October 1996; accepted 28
October 1996
Fanconi's Anemia (FA), Xeroderma Pigmentosum
(XP) and Ataxia Talangiectasia (AT), where defective
DNA repair leads to genomic instability, preceding
cancer (for a review see Hartwell, 1992). Genomic
instability was also shown to be induced by cancer
initiating agents, such as chemical and physical
carcinogens (Kleinberger et al., 1986; Lavi, 1981;
Tlsty, 1982).
These ®ndings were established by measures of
gross chromosomal changes: ampli®cation or recombination (Tlsty et al., 1989; Wright et al., 1990). In
fact, gene ampli®cation is used as an indicator of total
genomic instability (Tlsty et al., 1989; Tlsty, 1990), as
it has been proposed that a common mechanistic
defect underlies ampli®cation and other abnormalities
of unstable genomes, such as translocation, inversion
and deletion (for a review see Stark et al., 1989).
Unfortunately, gene ampli®cation is a problematic
indicator of genomic instability, since it is a late event
in the neoplastic process, and is limited to tumor cells,
whereas genomic instability in general may precede
transformation. Furthermore, detection of gene
ampli®cation is restricted to speci®c genes and often
requires selection. Therefore, it would be desirable to
establish a better indicator of genomic instability and
to investigate its generation and relation to malignant
transformation and cancer.
Small polydispersed circular DNA is another
phenomenon associated with genomic instability. These
small sized extrachromosomal circles are very common
in a large variety of eukaryotic cells. They are present in
many normal tissues and in every examined cell line (for
a review see Gaubatz, 1990), among them human cells,
monkey BSC-1 cells, Chinese hamster ovary (CHO)
cells, mouse cells and Drosophila melanogaster cells. It is
well accepted that they are derived from the chromosome and it has been suggested that these molecules
might interact with each other (for a review see
Gaubatz, 1990). Their size ranges mostly between a
few hundred base pairs to a few kilobases. Cloning
experiments show a wide variable representation of
repetitive sequences as well as a sporadic appearance of
unique chromosomal sequences (e.g. Assum, 1993; for
reviews see Gaubatz, 1990; Rush and Misra, 1985;
Yamagishi, 1986).
Several features of spcDNA relate it to genomic
instability, suggesting it as a possible marker, an
alternative to gene ampli®cation. The diversity in the
size, amount, sequence content and organisation of
spcDNA implies the existence of di€erent modes of its
formation. Various mechanisms (the same as those
proposed for genomic instability in general and for
gene ampli®cation in particular) were suggested to be
involved in its generation and propagation. These
spcDNA and genomic instability in human cells
S Cohen et al
➝ 1st
➝
include chromosomal recombination and rearrangement (Fujimoto and Yamagishi, 1987; Jones and
Potter, 1985; Riabowol et al., 1985; Sunnerhagen et
al., 1986), replicative mechanisms (such as arrested
aberrant replication or autonomous replication, Jones
and Potter, 1985; Kunisada et al., 1985; Rush and
Misra, 1985; Sunnerhagen et al., 1989; Ruiz et al.,
1989; Schimke, 1988; Wahl, 1989) or transposition
events (for a review see Rio, 1990). Similarly to gene
ampli®cation, spcDNA amounts in rodent cells were
enhanced in response to cancer initiating agents: we
showed that the level of spcDNA is elevated in CHO
cells in response to MNNG treatment of the cells
(Cohen and Lavi, 1996). Increased quantities of
spcDNA were also detected following treatments of
cultured mouse cells with the drug cycloheximide or
with agents arresting DNA replication, such as
hydroxyurea (HU) or 7,1-dimethylbenzanthracene
(DMBA) plus naladixic acid (Sunnerhagen et al.,
1986, 1989). Carcinogens were also found to enhance
SV40 ampli®cation in CO60 cells (Lavi, 1982; Lavi
and Etkin, 1981) involving the formation of circular
molecules (Cohen and Lavi, 1996), and were reported
to induce cellular gene ampli®cation in di€erent cell
lines (Kleinberger et al., 1986; Schimke, 1988).
spcDNA amounts are enhanced in several conditions
associated with genomic instability, such as angiofibromas derived from Tuberous Sclerosis patients
(Neidlinger et al., 1988) and cells from Fanconi's
Anemia patients (Motejlek et al., 1993). Since
spcDNA is present in both normal and tumor cells
(in contrast to gene ampli®cation), it may serve as a
wider measure of genomic instability, independently of
immortalisation and transformation events. The
speci®c spcDNA pro®le, regarding features such as
amount, size and sequence content may provide
further useful data.
Such a measure requires an accurate reliable
quanti®cation method. Until now, the research of
spcDNA was primarily based on two approaches: one
is the electron-microscope (EM) size analysis of the
molecules and the second is cloning and sequencing.
Both have only limited quantitation capability and are
hampered by technical diculties. EM analysis can
only give information about the size of circular
molecules and a limited estimation of their amount
per cell (Kunisada and Yamagishi, 1983; Motejlek et
al., 1993). For the analysis of sequence content,
cloning experiments are used. To this end, large
amounts of cells have to be grown (4109), from
which extrachromosomal DNA is prepared and the
circular supercoiled DNA fraction is isolated by
successive CsCl-EtBr density gradients. This procedure is inecient and linear DNA contaminants are
always present. In addition, the cloning procedure can
easily cause artifacts such as underrepresentation of
the population or disruption of original sequences.
We have recently presented a new technique for
size, sequence and quantity analysis of spcDNA. This
technique is based on two-dimensional (2D) agarose
gel analysis of low-molecular-weight (LMW) DNA
(Brewer and Fangman, 1987). We have previously
shown that the heterogeneous population of circular
molecules generates a typical arc pattern, when
migrating on a neutral-neutral 2D agarose gel
(Cohen and Lavi, 1996). This pattern can be
2nd
1.
2.
3.
4.
0
0
0 0
0 0
978
~
➝
—
—
—
—
Relaxed circles
Supercoiled converted to relaxed circles
Supercoiled circles
Linear DNA
Figure 1 Neutral-neutral 2D gel analysis of CO60 DNA. SV40
transformed Chinese hamster (CO60) cells were treated with
10 mg/ml MNNG and low molecular weight (Hirt supernatant)
DNA was prepared from the cells 96 h post treatment. 12 mg of
DNA were separated on a neutral-neutral 2D gel. Following
blotting, the blot was hybridised to an SV40 probe and exposed
to Kodak sensitive AR ®lm for 24 h (upper panel). A schematic
representation of the resulting typical arc pattern and arc identity
as established by electron microscopy (Cohen and Lavi, 1996) is
given below. The dots appearing on arcs 1 ± 3 (upper panel)
represent the predominant size class of ampli®ed SV40 genomes
visualised following blotting and hybridisation to an
appropriate probe, and is illustrated in Figure 1.
Brie¯y, each arc represents a population of molecules
of the same structure and of a variety of molecular
masses. Originally, the sample contained three major
populations of DNA molecules: linear, supercoiled
circular and relaxed circular. Each of these populations generates its own arc (arcs no. 4, 3 and 1,
respectively). The fourth arc (arc no. 2) is comprised
of supercoiled molecules, which were relaxed by
nicking during the preparation of the second
dimension. Such molecules migrated as supercoiled
circles in the ®rst dimension, and as relaxed circles in
the second one, generating this additional arc.
Electron microscopy was used for structural identification of the molecules that comprise each arc (Cohen
and Lavi, 1996).
In this paper, we study spcDNA in normal and
tumor human cells, using the 2D gel analysis, in order
to understand the relation between spcDNA and the
state of genomic instability. spcDNA molecules were
detected in human tumor cells (HeLa) and tumor
tissue (colon carcinoma) and in ®broblasts from
patients su€ering from Fanconi's Anemia, a disease
spcDNA and genomic instability in human cells
S Cohen et al
979
associated with genomic instability. spcDNA was not
detected in the genetically stable normal human skin
®broblasts; however, in response to MNNG treatment,
spcDNA was induced. We also present here for the
®rst time the possibility to quantify the amount of
spcDNA using mitochondrial DNA, thus enhancing
the power of the 2D gel analysis from a qualitative to
a quantitative assay, suitable for the establishment of
spcDNA as a possible marker and enhancer of
genomic instability.
Results
We wished to use the neutral-neutral 2D gel
electrophoresis (Brewer and Fangman, 1987) to
detect spcDNA in various human cells and to
investigate its association with genomic instability.
Similarly to our previous experiments (Cohen and
Lavi, 1996), we analysed LMW DNA extracted from
the examined cells on the 2D gels. Following
blotting, the DNA was hybridised to a human Cot1 probe which represents the highly repetitive
genomic DNA (Britten and Kohme, 1968). In these
experiments we visualise the molecules that contain
the repetitive sequences and thus probably represent
the total genomic DNA.
As we showed before (Figure 1), circular DNA
molecules create a typical reproducible arc pattern on
neutral-neutral 2D gels. To obtain such a reference
pattern, DNA from MNNG-treated CO60 cells was
either mixed with the tested human DNA sample or
was separated simultaneously in the same electrophoresis tank as the examined cellular DNA. This DNA
contains a heterogeneous population of ampli®ed
circular SV40 molecules (Cohen and Lavi, 1996).
Following blotting and hybridisation to an SV40
probe, this DNA served as a reference for the
migration position of the circular DNA arcs. When
human DNA was mixed with the CO60 DNA, the
blot was hybridised twice, once to the cellular
repetitive DNA probe (Cot-1), and afterwards to the
SV40 probe. The MNNG-treated CO60 DNA and the
human Cot-1 probe do not cross-hybridise (Cohen
and Lavi, unpublished results).
Circular DNA in human tumor cells (HeLa) and tumor
tissue (colon carcinoma) is detected by 2D gel analysis
Previously it was reported that DNA from HeLa cells
contained circular DNA molecules (Jones and Potter,
1985; Kunisada and Yamagishi, 1984, 1987, Misra et
al., 1987, 1989). We asked whether this DNA could
also be identi®ed using our new 2D gel assay.
2D gel analysis of extrachromosomal DNA
extracted from HeLa cells was performed, followed
by blotting. Hybridisation to human Cot-1 DNA
showed arcs of circular DNA (Figure 2A). The arc's
identity was con®rmed by comparison to an internal
CO60 DNA control which was hybridised to a SV40
probe (data not shown). The faint arc seen below the
massive arc of the linear DNA represents single
stranded DNA as was determined by the migration
position of a denatured l linear size marker (data not
shown). This single stranded DNA appears from time
to time upon analysis of DNA preparations from
Figure 2 Neutral-neutral 2D gel analysis of low molecular
weight DNA from human tumor cells (HeLa) and human tumor
tissue (colon carcinoma). (A) Low molecular weight (Hirt
supernatant) DNA was prepared from HeLa cells and 20 mg
were analysed on a 2D gel. Following blotting, the gel was
hybridised to a human Cot-1 probe and exposed to Kodak
sensitive AR ®lm for 24 h. The arrow points the arc of relaxed
circles (arc 1) as determined by comparison to the migration
position of ampli®ed SV40 DNA from MNNG-treated CO60 cells
(data not shown). (B) Low molecular weight DNA was prepared
from a human colon carcinoma and 80 mg were analysed by
neutral-neutral gel electrophoresis. The blot was hybridised to a
human Cot-1 probe and an arc of relaxed circles was observed
(arrow) beside a massive arc of linear DNA. Low molecular
weight DNA from MNNG treated CO60 cells was mixed with the
human DNA sample. Following removal of the Cot-1 probe and
rehybridisation to an SV40 probe, the reference pattern (as that in
Figure 1B) was obtained and was used to identify the arcs (data
not shown)
di€erent cells (see also Figure 3A, B and Figure 4A)
and it is not related to a speci®c line or to speci®c
growth conditions.
We now proceeded to examine whether circular
DNA molecules could also be detected in vivo, in
human tumor tissue. Low molecular weight DNA was
extracted from human colon carcinoma, and analysed
by the 2D gel. Following hybridisation to the human
Cot-1 probe, an arc of open circles appeared clearly as
well as the heavy arc of the linear molecules (Figure
2B). Similar results were obtained upon analysis of
DNA from several other human colon carcinomas
(data not shown).
We conclude that, similarly to rodent cells, circular
molecules can be detected easily in cultured human
tumor cells, suggesting that they are quite abundant in
these cells. Furthermore, signi®cant amounts of
spcDNA were also detected in LMW DNA from
tumor tissue samples of colon carcinoma, thus
extending the in vitro ®nding to in vivo conditions.
spcDNA and genomic instability in human cells
S Cohen et al
980
spcDNA is detected in cells from patients su€ering from
a genetic disease associated with genomic instabilityFanconi's Anemia and not in normal human ®broblasts
Genomic instability characterises tumor cells. However,
it often precedes tumorigenesis in vivo. This happens,
for example in the tumor-prone recessive genetic
disease, Fanconi's Anemia (FA). DNA from skin
®broblasts of FA patients was previously reported to
contain circular molecules, and in a semi-quantitative
assay, it was estimated that the amount of these
molecules is elevated in comparison to their level in
cells of healthy donors (Motejlek et al., 1993). To
further investigate this ®nding, low-molecular-weight
DNA from skin ®broblasts of FA patients (K ± 277)
and from normal ®broblasts (F ± 1748) was analysed by
the 2D gel, as shown in Figure 3. Upon hybridisation
to a human Cot-1 DNA probe, an arc pattern of
circular DNA molecules was observed in DNA from
FA patients (Figure 3B), but not in normal ®broblasts
(Figure 3A). The identity of the arcs was again
con®rmed by comparison to the SV40 pattern of the
MNNG-treated CO60 DNA that was mixed with the
human DNA samples (data not shown). spcDNA was
also detected upon examination of DNA taken from
passages 6, 7 and 11 of three additional FA patients
(data not shown).
To normalise the signals that were obtained in Figure
3A and B, we reprobed the blots with a human
Figure 3 Neutral-neutral 2D gel analysis of low molecular
weight DNA from skin ®broblasts of a Fanconi's Anemia (FA)
patient and of a healthy donor. Skin ®broblasts from a healthy
donor (F ± 1748, A and C) and from a FA patient (K ± 277 cells, B
and D) were propagated in culture for several passages. Low
molecular weight DNA (Hirt supernatant, 30 mg from F ± 1748
and 9 mg from K ± 277) was analysed by 2D gel electrophoresis
and blotted and hybridised to a human Cot-1 probe (A and B).
The blots were exposed for 72 h to Fuji RX ®lm. After removal
of the probe the blot was rehybridised to a human mtDNA probe
(C and D) and exposed to Fuji RX ®lm for 24 h. The arrow (B)
points the arc of relaxed circles as was veri®ed by comparison to
the hybridisation pattern of MNNG-treated CO60 DNA (which
was mixed with human DNA before the 2D assay) to an SV40
probe (data not shown). `Cir.' indicates the unique migration
pattern of the circular mtDNA complexes
mitochondrial DNA (mtDNA) probe (Figure 3C and
D). mtDNA is puri®ed with the LMW DNA, and
therefore can serve as a constant parameter for the total
amount and integrity of circular DNA that was loaded
onto the gel. As shown in Figure 3C and D (marked as
`Cir.') mtDNA migrates on the 2D gels in a typical
pattern that can be easily distinguished from the linear
DNA and from the arcs of the circular smaller
molecules. Quantitative analysis by PhosphorImager
of the mtDNA signals showed that mtDNA quantity in
the normal ®broblasts' DNA sample (Figure 3C) was
slightly higher than twice that in the FA ®broblasts
sample (Figure 3D).
This con®rms that circular DNA molecules are
abundant in the genetically unstable ®broblasts from
the Fanconi's Anemia patient, in comparison to normal
skin ®broblasts. Since spcDNA could not be detected by
the 2D gel assay in the normal skin ®broblasts we
decided to further investigate the presence of spcDNA in
the normal, genetically stable, human ®broblasts, and
whether it could be induced by carcinogens.
spcDNA could not be detected in normal human skin
®broblasts, but is induced upon their treatment with
carcinogen
It was previously reported that the lowest amounts of
spcDNA were obtained in normal human skin
®broblasts (Motejlek et al., 1993 ; Kunisada et al.,
1985). LMW DNA from skin ®broblasts of a healthy
donor was analysed on a 2D gel (Figure 4A). As seen
in the previous sample of normal ®broblasts DNA
(Figure 3A) an arc of circular DNA could not be
detected even after longer exposures (48 h in PhosphorImager) and DNA from another healthy donor
behaved similarly (data not shown). These results
demonstrate that under our assay conditions, circular
DNA could not be detected in the normal human skin
®broblasts and that if such molecules do exist in
normal cells, then their amount is very low.
We have previously shown, using the 2D gels, that
MNNG treatment induced the formation of circular
SV40 molecules in CO60 cells (Cohen and Lavi, 1996).
Moreover, we have demonstrated the enhancement of
circular molecules in CHO cells in response to the same
MNNG treatment (Cohen and Lavi, 1996). Similarly,
exposure to drugs arresting DNA replication such as
Hydroxyurea and DMBA, and cycloheximide, which
indirectly blocks replication, were reported to induce
spcDNA in mouse 3T6 cells (Sunnerhagen et al., 1986,
1989). Thus, in rodent cells, which are notorious for
their genomic instability, spcDNA is induced following
carcinogen treatment. The e€ect of carcinogens on
spcDNA in the genetically stable human cells has not
been examined before, but we speculated that
carcinogens such as MNNG could induce spcDNA in
normal human ®broblasts as well.
Normal skin ®broblasts (F ± 1748), were treated with
MNNG. Hirt supernatant DNA was prepared from
control cells (Figure 4A) and from MNNG-treated
(5 mg/ml) cells (Figure 4B) 48 h post treatment, and
was analysed on the 2D gels.
Figure 4B shows that in response to MNNG
treatment, an arc of relaxed circles is clearly
observed. Hence, the exposure to MNNG induced
the formation of circular DNA molecules in normal
spcDNA and genomic instability in human cells
S Cohen et al
981
Figure 4 MNNG-treatment of normal skin ®broblasts induced the formation of circular DNA. Normal skin ®broblasts, taken
from a healthy donor (F ± 1748 cells) were propagated in culture for 10 passages and then treated with MNNG (5 mg/ml). Low
molecular weight DNA (Hirt supernatant) was prepared from the treated cells 48 h post treatment, and from untreated control cells.
10 mg and 8 mg of DNA (from untreated and treated cells, respectively) were separated on a neutral-neutral 2D gel. Following
blotting, the DNA was hybridised to a human Cot-1 probe (A and B) and exposed to Kodak sensitive AR ®lm for 48 h. After
removal of the probe the blot was rehybridised to a human mtDNA probe (C and D) and exposed to Kodak sensitive AR ®lm for 3
days. A and C: DNA from the untreated F ± 1748 cells; B and D: DNA from MNNG-treated F ± 1748 cells. The arrows point the
arcs representing relaxed circles (corresponding to arc no. 1 of ampli®ed SV40 DNA). LMW DNA from MNNG-treated CO60 cells
was mixed with the human DNA sample and hybridisation of the blot to an SV40 probe served as reference for the migration
position of the circular DNA arcs (data not shown)
human skin ®broblasts. This experiment demonstrates
that circular molecules can be induced in normal
genetically stable human cells in response to MNNGtreatment.
To normalise the signals that were obtained in
Figure 4A and B, we used our novel quanti®cation
method based on mitochondrial DNA (Figure 4C and
D). After hybridisation to a human mtDNA probe, we
quantitated the mtDNA signals (marked as `Cir.' in
Figure 4C and D) by PhosphorImager. The level of
mitochondrial DNA in the control sample (Figure 4C)
was twice that in the treated-cell sample (Figure 4D).
Therefore, the induction of the circular DNA by
MNNG treatment, is even stronger than demonstrated
in Figure 4A and B. Note, that Mitochondrial DNA
quantities were reported to remain unchanged in
carcinogen treated cells (Sunnerhagen et al., 1989), so
it can serve as a reliable standard when comparing
control and treated samples.
Discussion
Genomic instability is composed of various dynamic
changes in the genome. Accurate measurement of all
alterations has proven to be dicult, and thus, speci®c
features often serve as a marker of this general
phenomenon. Gene ampli®cation has been widely
used as such a marker. However, although genomic
instability is a characteristic of both non-transformed
and malignant cells, gene ampli®cation is limited only
to malignant cells (Tlsty, 1990). Thus, an alternative
marker, of a wider scope, is called for.
spcDNA was previously suggested to be related to
genomic instability. Since it is also attributed to
normal cells, and its pro®le (amount, sequence
content) varies between genetically stable and unstable cells it was proposed as an alternative marker.
However, technical diculties in qualitative and
quantitative analysis of spcDNA have hampered the
establishment of its relation to genomic instability and
its subsequent utilisation as a marker of this
phenomenon.
We have recently suggested that 2D gel electrophoresis can be utilised for the research of spcDNA. We
showed that spcDNA is easily detected in CHO cells
following separation of LMW DNA on 2D gels and
hybridisation to hamster Cot-1 probe (Cohen and Lavi,
1996). We now demonstrate the use of the 2D gels to
identify spcDNA in human cells. Our experiments
revealed that spcDNA molecules containing highly
repetitive sequences were frequently found in either
intrinsically unstable cells (transformed, tumor, FA) or
in cells subjected to an external carcinogen treatment.
Based on previous reports on spcDNA (Gaubatz,
1990), we chose to use human Cot-1 DNA as a probe for
the identi®cation of circles containing highly repetitive
sequences. Thus, we believe that the total genomic
sequences are represented. However, more speci®c
probes, such as families of mid-repetitive sequences, or
spcDNA and genomic instability in human cells
S Cohen et al
982
gene families, might display sharper di€erences between
control and treated cells. Further hybridisation analyses
will illuminate the features of the spcDNA phenomenon
in aspects of both size and sequence content and may
contribute to the understanding of the mechanisms
involved in the formation of spcDNA.
spcDNA in human tumor cells and tumor tissue
We demonstrated that spcDNA is found in HeLa cells
(Figure 2A) as well as human colon carcinoma
(Figure 2B). This preliminary result of spcDNA
detection in colon tumor is intriguing but little is
known about the abundance of circular molecules in
other tumor cells and it will be important to compare
spcDNA from a tumor and a corresponding healthy
tissue of the same patient. Furthermore, if spcDNA is
involved in the initiation step of carcinogenesis, then
comparison to a tissue sample from a healthy donor
will also be required, as phenotypically healthy tissues
of the patient may already be initiated and contain
elevated levels of spcDNA. It is also tempting to
study di€erent tumors and di€erent stages of
malignancy and to determine whether spcDNA could
be a marker for tumor progression. Note that
spcDNA has also been found in normal human
tissue and cells (Hollis and Hindley, 1986; S Cohen,
unpublished results). The amount of spcDNA may
di€er between di€erent tissues, a question which
should also be addressed. Furthermore, aged cells
have been previously found to contain elevated levels
of spcDNA, both in vivo and in vitro (Flores et al.,
1988; Gaubatz and Flores, 1990; Kunisada et al.,
1985). Thus, spcDNA emerges as an attribute of both
normal and pathological tissues and cells and might
be associated with the normal phenomenon of
genomic `plasticity' as well as the pathological
genomic instability. Quantitative analysis is required
in order to compare this property between normal,
aged and tumor tissues.
Correlation between spcDNA and unstable genomes
We have shown that spcDNA is detected in primary
cultures of skin ®broblasts from a Fanconi's Anemia
(FA) patient (Figure 3), which is a recessive genetic
disease that exhibits high levels of genomic instability
(Friedberg et al., 1995; Motejlek et al., 1993). On the
other hand, spcDNA was below detection in cells from
a healthy donor (Figure 4A). These results support the
idea that elevated levels of spcDNA characterise
genetically unstable cells and that normal cells contain
relatively low levels of spcDNA. They are also
consistent with the previous estimation that the
number of circles in normal human ®broblasts was
the lowest amongst all the examined cells: 10 to 60
copies per cell (Kunisada et al., 1985). Furthermore,
they agree with a previous observation made using a
semi-quantitative electron microscope analysis, that
cells from FA patients contain 85-fold more spcDNA
in average, in comparison to cells from healthy donors
(Motejlek et al., 1993).
We conclude that spcDNA is pronounced in
genetically unstable cells. This instability could be
either related to the transformation and the malignant
processes, as was found in tumor cells, or genetically
inherited as in FA cells. In both cases, this is an
intrinsic feature of the cells which maintain high steady
state levels of circular DNA molecules.
Exposure of normal human ®broblasts to carcinogen
results in an induction of spcDNA
Looking at the levels of spcDNA as a marker for
genomic instability, we investigated if growth conditions or cancer initiating agents could induce genomic
instability and a€ect the amount of circular molecules.
Treatment with MNNG and other carcinogens was
shown to enhance spcDNA in rodent cells (Cohen and
Lavi, 1996; Sunnerhagen et al., 1989). However, rodent
cells are known to be inherently unstable relative to
human cells, which were shown to be non-responsive
and refractory to the induction of genomic instability
phenomena, such as gene ampli®cation (Tlsty, 1990).
Since the e€ect of carcinogens on spcDNA in normal
human cells has not been tested before, we chose to
examine the amount of spcDNA in primary human
skin ®broblasts following MNNG treatment. We
showed that an induction of spcDNA is observed
48 h after treatment (compare Figure 4A and B) and
con®rmed this result by normalising the circular DNA
arc signal according to the mtDNA signal (Figure 4C
and D).
We show here for the ®rst time that exposure of
normal human cells to an external drug results in an
induction of spcDNA. We suggest that this induction
may re¯ect a temporary destabilization of the genome
independent of endogenous genetic changes (such as
the loss of functions responsible for maintenance of
genomic integrity). We speculate that the level of these
circles may decline in time as a result of passive
stochastic processes as well as active cellular mechanisms controlling genomic integrity. Alternatively, these
molecules could be involved with the very early steps
of malignancy and their presence might re¯ect the
cell's initiated state, as they might interact with the
chromosome or with each other, enhancing genomic
instability. In addition, circles containing sequences
conferring a selective advantage to the cells, are
expected to be maintained, similarly to gene
ampli®cation events and to the survival of new
mutations.
We have also observed that the amount of spcDNA
increases as the cells reach con¯uence (S Cohen,
unpublished results): Con¯uent normal ®broblasts
(harvested 120 h after seeding) contain detectable
spcDNA levels, in contrast to cells harvested 72 h
post seeding (Figures 3A and 4A). These ®ndings are in
agreement with previous experiments (DeLap et al.,
1978; Smith and Vinograd, 1972) which indicate that
elevated levels of spcDNA were found in con¯uent
cells in comparison with log-phase counterparts.
Similarly, enhanced levels of spcDNA were detected
in in vitro and in vivo senescent cells (Kunisada et al.,
1985; Yamagishi et al., 1985). We suggest that the
enhanced levels of spcDNA in these cases may be the
result of aberrant DNA replication in arrested cells,
similarly to spcDNA induction by replication arresting
drugs (Cohen and Lavi, 1996).
Further studies with additional cells and drugs are
required in order to determine if spcDNA induction by
carcinogens could be used as an assay for the detection
spcDNA and genomic instability in human cells
S Cohen et al
of exposure of cells to environmental stress. This would
have wide implications concerning the determination of
parameters for monitoring exposure to, as well as
identi®cation and de®nition of drugs and carcinogens.
Quanti®cation of spcDNA using mitochondrial DNA as
a normalizing standard
A normalising standard is a prerequisite for any
quantitative analysis of the spcDNA population. Such
normalisation can in principal be based on optical
density measurement of the total DNA amount loaded
on the 2D gel. However, the ratio between circular and
linear extrachromosomal DNA yields may change
between di€erent preparations of Hirt extracts: For
example, carcinogen treatment may in¯uence the
amount of extrachromosomal linear DNA due to
breakage events. Furthermore, when using DNA
samples prepared from tissues, the number of cells
cannot be evaluated. Thus, normalisation according to
non-circular standards, such as total or linear DNA in
the sample, is insucient and misleading.
We utilised mitochondrial DNA as an alternative
mode of normalisation. Since the circular mtDNA is
puri®ed along with the other LMW DNA it can serve
as an intrinsic standard for both the amount and
integrity of circular DNA loaded onto the gel.
Moreover, the fact that the amount of mtDNA
remains constant upon carcinogen treatment (Sunnerhagen et al., 1989) renders it particularly reliable as a
normalising standard in assays involving carcinogens
and other types of environmental stress. Thus, a
quantitative research of spcDNA using the neutralneutral 2D gel electrophoresis can be performed.
We also show in this work for the ®rst time the
pro®le of mitochondrial DNA molecules on 2D gels.
The approximately 16 Kbp mtDNA molecules of
normal human ®broblasts exhibit a unique migration
pattern, as displayed by the hybridization to mtDNA
speci®c probes (Figures 3C, D and 4C, D). Similar
patterns were observed with DNA from other
mammalian cell lines (Chinese hamster cell lines,
CHE and CHO) and from human tissues (S Cohen,
unpublished results).
This typical pattern can be easily distinguished from
the linear DNA and from the arcs of the circular
smaller molecules and it consists of a linear band in the
size of 16 Kbp and a non-linear smear representing the
circular structures. This smear was not completely
analysed but the horizontal lines between the spots
seem to represent molecules with the same structure
and the same mass (in the second dimension) that
migrated to di€erent positions in the ®rst dimension.
We speculate that these may consist of complexed
circular structures (such as concatamers). It was
reported that large multimer complexes of mtDNA
were puri®ed from HeLa cells along with the two
circular forms and with the linear mtDNA (Higuchi
and Linn, 1995). Changing the conditions of the ®rst
dimension (such as addition of EtBr) or treatment with
topoisomerase to separate the concatamers could be
helpful for further analysis of these structures.
This migration pattern may also serve for the
analysis of mitochondrial DNA per se. Changes in
the integrity of mtDNA as a result of aging or
oxidative stress have been extensively studied and
evidence indicate that mtDNA deletions and mutations may be important in age-related degenerative
diseases (Goldstein and Shmookler-Reis, 1984; Katsumata et al., 1984; Kowald and Kirkwood, 1993; Lezza
et al., 1994). Thus, analysis of structural and size
changes in mtDNA by 2D gel analysis may contribute
to the understanding of mitochondrial DNA-associated
pathologies.
Functional aspects of spcDNA
spcDNA may also play an active role in the initiation
and/or enhancement of genomic instability. spcDNA
may arise by the same mechanistic principles as gene
ampli®cation (Cohen and Lavi, 1996), thus leading to
alterations in chromosome structure and in gene
expression
patterns.
Autonomously
replicating
spcDNA or recombination events (either within the
spcDNA population or with chromosomes) could
contribute to additional genomic changes. Autonomous replication of spcDNA may require these
molecules to contain a functional origin. However,
there is some evidence that any human sequence
included in a large enough viral circle (which is
replication defective itself) could autonomously replicate (Krysan et al., 1989, 1991). Recombination events
with chromosomes during the generation and the
propagation of spcDNA could also enhance genomic
instability by clastogenic e€ects. Thus, following
initiation, the carcinogen-induced spcDNA may serve
as a mutator, promoting further chromosomal
abnormalities, and extending the e€ect of the transient
exposure to a carcinogen.
spcDNA could be the result of normal cellular
mechanisms and processes associated with genomic
plasticity. Such mechanisms may have been retained in
evolution to enable ecient responses to selective
environmental pressures, as well as to the changing
requirements of di€erent tissues and cells in the
developing organism. This population of molecules
may be either functionally important in itself or a byproduct of other processes (such as the V-D-J
recombination in lymphocytes, Fujimoto et al., 1985,
1987; Okazaki et al., 1987).
Quantitative measurements by our novel 2D gel
system of the spcDNA population in general and of
molecules bearing speci®c sequences in particular, can
serve as a tool in the diagnosis and analysis of genomic
plasticity and instability in cells and tissues.
Materials and methods
Tissue culture
Cells and Cell lines
Name
Description
Reference
Primary cells
F-1748
Normal human skin fibroblasts
K-277
Human skin fibroblasts from a
Fraconi's Anemia patient
Cell lines
CO60
SV40 transformed Chinese
Lavi, 1981
hamster embryo cells
HeLa
human cervical carcinoma cells
ATCC
(American Type
Culture Centre)
983
spcDNA and genomic instability in human cells
S Cohen et al
984
All cells were propagated in monolayer cultures in
Dulbecco modi®ed Eagle medium (Gibco Laboratories,
Grand Island, NY) supplemented with 10% fetal calf serum
(Biolabs, Jerusalem, Israel).
MNNG treatment
CO60 cells were treated with MNNG 24 h after being
seeded as was previously described (Berko-Flint et al.,
1990). Log phase cells were plated at a density of 56106/
14 cm-diameter plate. After 24 h, the cells were treated
with 10 mg/ml MNNG (N-methyl-N'-nitro-N-nitrosoguanidine; Aldrich), which had been freshly dissolved in
dimethyl sulfoxide (Sigma), and added to the growth
medium. One hour after the treatment, the medium was
replaced with fresh medium and the cells were allowed to
grow in carcinogen-free medium until they were harvested
(48 ± 96 h post treatment).
Human ®broblasts (F ± 1748) were treated with 5 mg/ml
MNNG. This concentration was determined as follows: F ±
1748 cells, seeded in a 24-well plate, were treated with a series
of MNNG concentrations ranging from 1 mg/ml to 10 mg/ml
(in duplicates) and the toxic e€ect of the treatment on the
cells during 5 days was estimated. Treatment with 10 mg/ml
MNNG was highly toxic 24 h post treatment, however upon
treatment with 5 mg/ml most of the cells were viable though
changes were observed in their shape (i.e., some of the treated
cells lost their regular elongated shape) demonstrating the
e€ect of treatment. In lower concentrations no toxicity was
observed and the structural changes were reduced as well
(data not shown). Therefore, we chose the treatment with
5 mg/ml MNNG for further experiments.
Low molecular weight DNA preparation
Low-molecular-weight cellular DNA was prepared according to the procedure previously described by Hirt (1967).
When tissue was used (instead of cells) frozen tissue
samples were ®rst ground under liquid nitrogen, by mortar
and pestle. The ®ne powder was dissolved in Hirt lysis
bu€er (Hirt, 1967), and subsequent steps were performed
as before. The amount of DNA was estimated by
spectrophotometer (Gilford) at 260 nm.
1 V/cm for 18 h and the second dimension, in 0.8%
agarose 5 V/cm for 3.5 h.
Blotting and hybridisation conditions
Southern blot analyses were carried out using Hybond N +
nylon membrane (Amersham, Amersham, UK) according
to the procedure previously described by Church and
Gilbert (1984). Brie¯y, the gels were rinsed in 0.2 N HCl
for 15 min followed by washing twice with 0.4 N NaOH
which was also used as a transfer bu€er for blotting. The
hybridizations were carried out in 658C in 0.5 N sodium
phosphate (pH 7.2), 7% SDS and 1 mM EDTA. After
hybridisation, blots were washed twice at 658C in 26 SSC,
0.1% SDS. Rehybridization was performed after stripping
with 0.4 N NaOH for 20 min followed by washing in
50 mM phosphate bu€er pH 6.5.
Radiolabeled probes
Supercoiled SV40 genomes puri®ed by a CsCl gradient
were used as a probe for cellular SV40 DNA.
Human Cot-1 DNA (BRL) served as a probe for the
detection of cellular circular DNA molecules.
Human mitochondrial DNA probe was prepared by PCR
generating a 533 bp fragment (nucleotides 3304 ± 3836) that
was previously reported to remain unchanged following
mitochondrial deletion during aging and other stresses (Lee
et al., 1994). The oligonucleotides are 5'AACATACCCATGGCCAACCT-3'
and
5'-GGCAGGAGTAATCAGAGGTG-3' (designated L1 and H1 in Lee et al., 1994).
All probes were labeled with [a-32P]dCTP (Redivue,
Amersham, Amersham, UK), using the Random-Prime
Labeling Kit (Boehringer Mannheim) according to the
manufacturer's instructions.
PhosphorImager analysis
Quantitative analyses of blots of two dimensional gels were
performed using the Fuji BAS1000 PhosphorImager and
the Tina 2.07 program (Dinko&Renium, Bet-Nekofa,
Israel).
Neutral-neutral two-dimensional (2D) gel
Separation of DNA on the neutral-neutral 2D gel was
performed according to the procedure described by Brewer
and Fangman (1987). Brie¯y, DNA is ®rst separated on
0.4% agarose gel at low voltage in TBE, in the absence of
EtBr. Then, the gel is rinsed with TBE containing 0.3 mg/
ml EtBr. The lane of choice is cut under u.v. illumination
and placed in a clean gel support at 908 to the direction of
the ®rst electrophoresis. The lane is cast with more dense
agarose (usually 0.8%) that contains 0.3 mg/ml EtBr. The
second dimension is run in cold room in TBE containing
0.3 mg/ml EtBr, at 5 V/cm. Minor modi®cations of the
electrophoresis parameters were performed in our experiments: The ®rst dimension was run in 0.4% agarose
Acknowledgements
We are grateful to D Abeliovich of the Department of
Human Genetics, Hebrew University Medical School,
Jerusalem, for providing human primary cultures (F ±
1748 and K ± 277 cells), and to D Hassin, Hillel Ya€e
hospital, Hadera, Israel for providing the colon carcinoma
samples. We thank S Karbi and K Frist for their excellent
technical assistance and V Vexler, A Shuv and M Semo for
artwork. SC acknowledges support from the Charles Clore
Foundation doctoral fellowships program. This work was
supported by grants from the Ela Kodesz Institute for
Research on Cancer Development and Prevention, and by
the Israel Science Foundation to S..L.
References
Assum G, Fink T, Steinbeisser T and Fisel KJ. (1993). Hum.
Genet., 91, 489 ± 495.
Berko-Flint Y, Karby S, Hassin D and Lavi S. (1990). Mol.
Cell. Biol., 10, 75 ± 83.
Brewer BJ and Fangman WL. (1987). Cell, 51, 463 ± 471.
Britten RJ and Kohne DE. (1968). Science, 161, 529 ± 540
Church GM and Gilbert W. (1984). Proc. Natl. Acad. Sci.
USA, 81, 1991 ± 1995.
Cohen S and Lavi S. (1996). Mol. Cell. Biol., 16, 2002 ± 2014.
DeLap RJ, Rush MG, Zouzias D and Kahn S. (1978).
Plasmid, 1, 508 ± 521.
Fidler IJ and Hart IR. (1982). Oncodev. Biol. Med., 4, 161 ±
176.
Flores SC, Sunnerhagen P, Moore TK and Gaubatz JW.
(1988). Nucleic Acids Res., 16, 3889 ± 3893.
Friedberg EC, Walker GC and Siede W. (1995). DNA repair
and mutagenesis, American Society for Microbiology
Press: Washington DC.
spcDNA and genomic instability in human cells
S Cohen et al
Fujimoto S, Tsuda T, Toda M and Yamagishi H. (1985).
Proc. Natl. Acad. Sci. USA, 82, 2072 ± 2076.
Fujimoto S and Yamagishi H. (1987). Nature (London), 327,
242 ± 243.
Gaubatz JW. (1990). Mutat. Res., 237, 271 ± 292.
Gaubatz JW and Flores SC. (1990). Mutat. Res., 237, 29 ± 36.
Goldstein S and Shmookler-Reis RJ. (1984). Mol. Cell.
Biochem., 64, 15 ± 30.
Hartwell L. (1992). Cell, 71, 543 ± 546.
Higuchi Y and Linn S. (1995). J. Biol. Chem., 270, 7950 ±
7956.
Hirt B. (1967). J. Mol. Biol., 26, 365 ± 369.
Hollis M and Hindley J. (1986). Gene, 46, 153 ± 160.
Jones RS and Potter SS. (1985). Nucleic Acids Res., 13,
1027 ± 1042.
Katsumata K, Hayakawa M, Tanaka M, Sugiyama S and
Ozawa T. (1994). Biochem. Biophys. Res. Commun., 202,
102 ± 110.
Kleinberger T, Etkin S and Lavi S. (1986). Mol. Cell. Biol., 6,
1958 ± 1964.
Kowald A and Kirkwood TB. (1993). Mutat. Res., 295, 93 ±
103.
Krysan PJ and Calos MP. (1991). Mol. Cell. Biol., 11, 1464 ±
1472.
Krysan PJ, Haase SB and Calos MP. (1989). Mol. Cell. Biol.,
9, 1026 ± 1033.
Kunisada T and Yamagishi H. (1983). Plasmid, 9, 8 ± 16.
Kunisada T and Yamagishi H. (1984). Gene, 31, 213 ± 223.
Kunisada T and Yamagishi H. (1987). J. Mol. Biol., 198,
557 ± 565.
Kunisada T, Yamagishi H, Ogita Z, Kirakawa T and Mitsui
Y. (1985). Mech. Ageing. Dev., 29, 89 ± 99.
Lavi S. (1981). Proc. Natl. Acad. Sci. USA, 78, 6144 ± 6148.
Lavi S. (1982). Gene Ampli®cation. Schimke RT. (ed). Cold
Spring Harbor Laboratory Press: New York, pp. 225 ±
230.
Lavi S and Etkin S. (1981). Carcinogenesis, 2, 417 ± 423.
Lee H-C, Pang C-Y, Hsu H-S and Wei Y-H. (1994). Biochim.
Biophys. ACTA, 1226, 37 ± 43.
Lezza AM, Bo€oli D, Scacco S, Cantatore P and Gadaleta
MN. (1994). Biochem. Biophys. Res. Commun., 205, 772 ±
779.
Misra R, Matera AG, Schmid CW and Rush MG. (1989).
Nucleic Acids Res., 17, 8327 ± 8341.
Misra R, Shih A, Rush M, Wong E and Schmid CW. (1987).
J. Mol. Biol., 196, 233 ± 243.
Motejlek K, Schindler D, Assum G and Krone W. (1993).
Mutat. Res., 293, 205 ± 214.
Neidlinger C, Assum G, Krone W, Dietrich C, Hochsattel R
and Klotz G. (1988). Hum. Genet., 79, 286 ± 288.
Nowell PC. (1976). Science, 194, 23 ± 28.
Okazaki K, Davis DD and Sakano H. (1987). Cell, 49, 477 ±
485.
Riabowol K, Shmookler-Reis RJ and Goldstein S. (1985).
Nucleic Acids Res., 3, 5563 ± 5584.
Rio DC. (1990). Annu. Rev. Genet., 24, 543 ± 578.
Ruiz JC, Choi KH, von-Ho€ DD, Robinson IB and Wahl
GM. (1989). Mol. Cell Biol., 9, 109 ± 115.
Rush MG and Misra R. (1985). Plasmid, 14, 177 ± 191.
Schimke RT. (1988). J. Biol. Chem., 263, 5989 ± 5992.
Smith CA and Vinograd J. (1972). J. Mol. Biol., 69, 163 ±
178.
Stark GR, Debatisse M, Giulotto E and Wahl GM. (1989).
Cell, 57, 901 ± 913.
Sunnerhagen P, Sjoberg RM and Bjursell G. (1989). Somat.
Cell. Mol. Genet., 15, 61 ± 70.
Sunnerhagen P, Sjoberg RM, Karlsson AL, Lundh L and
Bjursell G. (1986). Nucleic Acids Res., 14, 7823 ± 7838.
Tlsty TD. (1990). Proc. Natl. Acad. Sci. USA, 87, 3132 ±
3136.
Tlsty TD, Brown PC, Johnston R and Schimke RT. (1982).
Gene Ampli®cation. Schimke RT (ed). Cold Spring Harbor
Laboratory Press: New York, pp. 231 ± 238.
Tlsty TD, Margolin BH and Lum K. (1989). Proc. Natl.
Acad. Sci. USA, 86, 9441 ± 9445.
Wahl GM. (1989). Cancer Res., 49, 1333 ± 1340.
Wright JA, Smith HS, Watt FM, Hancock MC, Hudson DL
and Stark GR. (1990). Proc. Natl. Acad. Sci. USA, 87,
1791 ± 1795.
Yamagishi H. (1986). Bioessays, 4, 218 ± 221.
Yamagishi H, Kunisada T and Takeda T. (1985). Mech.
Ageing Dev., 29, 101 ± 103.
985