Oncogene (2006) 25, 6336–6344
& 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00
www.nature.com/onc
ONCOGENOMICS
Nuclear–mitochondrial genomic profiling reveals a pattern of evolution
in epithelial ovarian tumor stem cells
AA Wani1, N Sharma1, YS Shouche and SA Bapat
ONCOGENOMICS
National Centre for Cell Science, Ganeshkhind, Pune, Maharashtra, India
Analyses of genome orthologs in cancer on the background of tumor heterogeneity, coupled with the recent
identification that the tumor propagating capacity resides
within a very small fraction of cells (the tumor stem
cells-TSCs), has not been achieved. Here, we describe a
strategy to explore genetic drift in the mitochondrial
genome accompanying varying stem cell dynamics in
epithelial ovarian cancer. A major and novel outcome is
the identification of a specific mutant mitochondrial DNA
profile associated with the TSC lineage that is drastically
different from the germ line profile. This profile, however,
is often camouflaged in the primary tumor, and sometimes
may not be detected even after metastases, questioning the
validity of whole tumor profiling towards determining
individual prognosis. Continuing mutagenesis in subsets
with a mutant mitochondrial genome could result in
transformation through a cooperative effect with nuclear
genes – a representative example in our study is a tumor
suppressor gene viz. cAMP responsive element binding
binding protein. This specific profile could be a critical
predisposing step undertaken by a normal stem cell
to overcome a tightly regulated mutation rate and
DNA repair in its evolution towards tumorigenesis. Our
findings suggest that varying stem cell dynamics and
mutagenesis define TSC progression that may clinically
translate into increasing tumor aggression with serious
implications for prognosis.
Oncogene (2006) 25, 6336–6344. doi:10.1038/sj.onc.1209649;
published online 29 May 2006
Keywords: ovarian cancer; tumor stem cells; mtDNA;
CREBBP; mutational profiling
Introduction
Division and differentiation of a small number of stem
cells in healthy tissues ensures a continuous turnover of
cells and optimal organ functioning (Michor et al.,
2004). Genetic heterogeneity exists even at a state of
Correspondence: Dr SA Bapat, Lab 4, National Centre for Cell
Science, NCCS Complex, Pune University Campus, Ganeshkhind,
Pune, Maharashtra 411007, India.
E-mail: sabapat@nccs.res.in
1
These authors contributed equally to this work.
Received 7 February 2006; revised 16 March 2006; accepted 16 March
2006; published online 29 May 2006
homeostasis within an organ, and is attributed to the
clonal evolution of normal stem cell lineages (Shin et al.,
2004, Calabrese et al., 2004). Recent evidence indicates
that tumorigenesis is an aberrant process initiated by a
rare population of transformed stem cells termed as
tumor stem cells (TSCs) that maintains/reacquires the
capacity for indefinite proliferation along with tumorforming capabilities (Bonnet and Dick, 1997). Current
reports of the putative isolation of TSCs (Al-Hajj et al.,
2003, Singh et al., 2003) are based on the differential
expression of histologic or surface markers and selfrenewal mechanisms during tumor maintenance and
progression. Although their critical importance is
realized, much remains to be learned about the genetic
mechanisms leading to the emergence of a TSC from
a normal stem cell, yet accounting for tumor heterogeneity (Florian et al., 2006).
On this background, we asked whether it would be
possible to (i) trace the genetic heterogeneity between
various stem cell lineages within a tumor, and (ii),
understand the molecular differences between normal
and tumor cells that could possibly define specific events
within a stem cell lineage that places it on a trajectory
towards tumorigenesis. Specifically, we postulated
that if the heterogeneity within tumors were to be
a cumulative effect of genetic drift and amplification
of specific stem cell lineages, one would need to define a
mechanism for tracing all these in order to resolve their
effects at the molecular and cellular level. Mitochondrial
genome (mtDNA) analyses, currently applied to study
evolutionary history, population migration, forensic
medicine (Pakendorf and Stoneking, 2005) and human
disease (O’Brien et al., 2005), have been recently
applied to model stem cell turnover rates and clonal
evolution in normal tissues; hence was thought to be an
ideal tool for this study. Among the nuclear genes,
the cyclic AMP response element-binding protein
(CREBBP) is involved in multiple cellular processes,
functions as a transcriptional cofactor and is also
a histone acetyltransferase (HAT) (Petrij et al., 1995).
Germline mutations in CREBBP result in Rubinstein–
Taybi syndrome (Kitabayashi et al., 2001), that is
characterized by an increased predisposition to cancer;
further validated by the observation that CREBBP þ /
mice express an increased frequency of hematopoietic
malignancies (Kung et al., 1999). Several truncating
mutations have been identified in CREBBP in breast,
colon and pancreatic cancer cell lines and primary
Nuclear–mitochondrial genomic profiling
AA Wani et al
6337
tumors, leading to loss or mutation of one allele
(Costanzo et al., 2002).
In the present study, we report the mutational
profiling of the mitochondrial genome and nuclear
CREBBP gene that led to the identification of a unique
mitochondrial–nuclear profile defining the TSC population within a tumor.
Results
Epithelial ovarian cancer is associated with a high
incidence of mitochondrial mutations of which three
show a statistically significant association
The study was initiated with amplification and sequencing of the entire mitochondrial genome from 12 primary
epithelial ovarian tumor samples (Supplementary
Table 1), ensuring identification of all occurring
nucleotide changes. Further, large polymerase chain
reaction (PCR) products excluded the possibility that
nuclear pseudogenes could complicate the analysis
(Parfait et al., 1998). The analyses of these tumors led
to the identification of a total of 170 nucleotide
variations (Supplementary Table 2) that showed interesting distribution patterns. Their distribution was
throughout the mitochondrial genome, viz. in the noncoding (control) as well as protein-coding regions; in the
latter case, both synonymous (silent) as well as nonsynonymous changes were evident. Of these variations,
30 were entirely novel. Very strikingly, the mitochondrial sequences in this study (tumor samples as well
as controls) with the Indian population revealed a
unique profile of eight sequence variants, viz. A73G,
A263G, A1438G, A2706G, A4769G, C7028T, A8860G
and A15326G appeared at high frequencies in all
samples and could be of evolutionary significance.
Although the individual polymorphisms within this
profile are described elsewhere (MITOMAP: A Human
Mitochondrial Genome Database & Human Mitochondrial Genome Database), based on their high recurrence, we derived that the profile could be specific to the
Indian population.
Subjecting the mutational data to a two-tailed w2
analyses indicated 64 of the total mitochondrial
nucleotide variations to be of moderate significance
(Supplementary Table 3), of which, three mutations, viz.
A10398G, 523insCA and 523delA (Table 1) expressed
Table 1 Putative mtDNA mutation candidates in ovarian cancer
mtDNA mutation
A10398G
523ins CA
523delA
Present (healthy)
Present (Cancer)
Absent (healthy)
Absent (Cancer)
w2
Proportion present (healthy)
Proportion present (cancer)
P-value
27
8
79
4
8.768218
0.254717
0.666667
0.003
4
3
102
9
8.70365
0.037736
0.25
0.003
8
4
98
8
7.84613
0.075472
0.333333
0.005
Abbreviation: mtDNA, mitochondrial DNA.
high degree of correlation with the primary tumors
(n ¼ 12) than the control samples (n ¼ 106) (Palanichamy et al., 2004), thereby qualifying as putative ovarian
TSC markers.
Analyses of single-cell clones reveal the true nature
of mitochondrial profiles within a tumor
The above results suggested that a tumor may contain a
multitude of stem cell lineages; yet, whether the
dominant expression in the tumor represents the TSC
population is still uncertain. To resolve this, our further
studies were with a unique model developed by us
earlier (Bapat et al., 2005 – described in Materials and
methods). This mtDNA analyses, indeed, provided
a magnified insight into the system. The primary
tumor cells and ascites-derived cells expressed identical
mtDNA profiles (Table 2), consisting of 37 variations
distributed throughout the genome in the protein coding
as well as non-coding genes; with synonymous as well as
non-synonymous changes being identified. Six of these
changes were found to be novel. The same mutational
profile was also associated with 14 (of 19) of the
immortalized clones as well as the unsorted M1 and M2
cultures (Figure 1a). We further conducted a chase–back
study towards tracing the origin of these variations.
Normal tissue sample of the same patient was unavailable; hence, a peripheral blood sample from the son was
obtained – valid as mtDNA is almost exclusively
maternally inherited (Hayashida et al., 2005). Comparison of profiles revealed only one mtDNA sequence
variation between the son and the tumor cells of the
patient (310 insC instead of 310 insCC – Table 2). This
suggests that all the variations detected in these samples
could be germ line polymorphisms. A marked feature
in the immortalized cells from this profile was the
acquisition of two mutations (G7393A in CoxI gene:
amino acid (a.a.) change from glycine to glutamic acid
and C16147T in D-loop region) that were absent in the
primary cells. These variations are indicative of mtDNA
mutagenesis resulting from adaptation in culture, also
known to be associated with chromosomal instability
and altered in vitro ploidy levels (Schmid et al., 2004).
Identification of a specific mitochondrial mutation profile
associated with the tumor stem cell population
A major finding of this study is the identification
of a variant mtDNA profile in a small group of five
clones (viz. A2, A3, A4, B2 and C4; Figure 1a). This
group includes the two tumorigenic clones A2 and A4,
indicating that the profile defines the TSC lineage as a
distinct identity among others within the tumor. The
mutant profile is characterized by 12 distinct sequence
variations and seven common ones within the germ line
profile (Figure 1b). A majority of nucleotide variations
in this profile were in the non-coding regions (D-loop:8;
16sRNA:1; tRNA cysteine:1), the one in the ATPase
synthase gene (C8410T) was silent (no change in a.a.),
whereas that in the cytochrome b gene (T14766C) was
non-synonymous (isoleucine to threonine). Acquisition
of the mutant profile was accompanied by reversal of 32
Oncogene
Nuclear–mitochondrial genomic profiling
AA Wani et al
6338
Table 2 mtDNA variations in OT29 germline profile
S. No
Position
Gene
a.a. change
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
T152C
A263G
A1438G
A2706G
A4769G
A8860G
A15326G
A73G
310 (Ins-CC)a
G329A
T489C
573 (Ins-C)
G3483A
C4020 T
T4216C
C5354T
C7028T
T7278G
T7678C
A8701G
A9446G
T9540C
G9932A
A10398G
C10400T
T10873C
A11101G
G11719A
C12705T
A13651G
T14783C
G15043A
G15301A
T15313C
C16111T
C16223T
T16311C
D-loop
D-loop
12S RNA
16S
ND2
ATP Synthase 6
Cytochrome b
D-loop
D-loop
D-loop
D-loop
D-loop
ND1
ND1
ND1
ND2
COXI
COXI
COXII
ATP Synthase 6
COXIII
COXIII
COXIII
ND3
ND3
ND4
ND4
ND4
ND5
ND5
Cytochrome b
Cytochrome b
Cytochrome b
Cytochrome b
D-loop
D-loop
D-loop
Non-coding
Non-coding
Non-coding
Non-coding
Silent
Threonine–Alanine
Threonine–Alanine
Non-coding
Non-coding
Non-coding
Non-coding
Non-coding
Silent
Silent
Tyrosine–Histidine
Silent
Silent
Phenylalanine–Valine
Silent
Threonine–Alanine
Silent
Silent
Silent
Threonine–Alanine
Silent
Silent
Silent
Silent
Silent
Threonine–Alanine
Silent
Silent
Silent
Silent
Non-coding
Non-coding
Non-coding
Interspecies conservation (%)
NA
NA
NA
NA
NA
50
37.50
NA
NA
NA
NA
NA
NA
NA
25
NA
NA
75
NA
62.50
NA
NA
NA
62.50
NA
NA
NA
NA
NA
50
NA
NA
NA
NA
NA
NA
NA
Novel
No
No
No
No
No
No
No
No
Yes
No
No
Yes
No
Yes
No
Yes
No
Yes
No
No
Yes
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Abbreviations: a.a., amino acid; NA, not applicable. a-310 ins-C - present in the son.
variations expressed in the germ line profile (to
reported sequences in the CRS – Cambridge Reference
Sequence). The heteroplasmic state of some of these
variations (data not shown) suggests that acquisition of
the mutant profile and loss of some of the wild-type
polymorphisms is a continuous ongoing process that
ultimately culminates in fixing of mutations to a
homoplasmic state (Coller et al., 2005).
Sublineage demarcation within the germ line and mutant
mitochondrial profiles
Within the wild-type/germ line profile clones, further
lineage demarcation was evident. Six clones (E1, F3, F4,
G5, G7 and H1) express the germ line profile that
may be surmised to define at least one stem cell lineage,
six others (A5, B7, E2, F1, F2 and F5) show a variance
through acquisition of an additional mutation –
G6888R in the CoxI gene resulting in a termination
codon. The heteroplasmic expression of this mutation
probably ensures clone survival; it can be speculated
that further mutation fixing to a homoplasmic state
could be lethal and result in extinction of the clone. The
remaining two wild-type clones (D4 and H4) express
Oncogene
another heteroplasmic mutation in the tRNA proline
gene – G15959R, defining yet another lineage/subline
emerging from the common, wild-type profile.
Within the mutant profile lineage, demarcation into
three different sublineages is also possible despite their
similar mitochondrial profiles. A2 is functionally distinct
owing to its tumorigenic potential. A4 cells acquired an
additional heteroplasmic, non-synonymous mutation
(C12700T) in the ND5 gene: leucine to isoleucine, that
coincided with acquisition of tumorigenecity, whereas
the remaining three clones (A3, B2 and C4) retained the
profile with no change either in mtDNA sequence or
tumorigenecity. An observation with the heteroplasmic
C12700T mutation coinciding with acquisition of
tumorigenecity was that it was not fixed to homoplasmy
even after propagating the tumor sequentially in nude
mice for two further generations.
cAMP responsive element binding binding protein exon
mutations in the mutant mitochondrial DNA profile
From the above analyses in the mutant mtDNA profile
clones, it became quite evident that acquisition of a
particular mutant mitochondrial profile, although
Nuclear–mitochondrial genomic profiling
AA Wani et al
6339
Figure 1 (a) Schematic representation of tumor evolution in OT29 samples. The primary and metastatic ascites both express wild-type
mitochondrial DNA profiles. The 19 immortalized single-cell clones isolated from a primary ovarian tumor separate into two major
groups: Group 1 consists of five clones, of which one (A2) was tumorogenic (Lineage 1a); another (A4) underwent a spontaneous
transformation and acquired one heteroplasmic mutation (12700C>T) in ND5 gene (Lineage 1b) whereas the other remained
untransformed (A3, B2 and C4 – Lineage 1c). Group2: too diverged into three lineages – Lineage 2a consisted of six clones (wild-type
profile); Lineage 2b expressed one additional heteroplasmic mutation (G6888A) in the CoxI gene and Lineage 2c another
heteroplasmic transition (G15959A) in the tRNA proline gene (see Supplementary Figure 1). (b) Distribution of (12 þ 7) mutant
sequence variations marked on a representative linearized mitochondrial genome; *Sequence variations causing non-synonymous
amino-acid changes in coding genes.
highly implicative of a signatorial association with the
TSCs, cannot be the sole determinant of tumorigenecity.
It has been earlier suggested that complementation with
specific mutations in nuclear genes (Singh et al., 2005) or
epigenetic regulation (Feinberg et al., 2006) could be the
next step required to propel a stem cell towards a
tumorigenic program. To resolve this, we carried out
mutational analyses CREBBP in the five clones expressing the mutant mtDNA profile, the primary tumor and
ascites-derived cells and one representative clone expressing the germ line mtDNA profile, viz. H4. This led
to the detection of 23 novel mutations in the CREBBP
exon sequences, none of which were common to all the
samples, although some overlap was evident within
smaller groups (Figure 2a). Significantly, no exon
mutations were evident in the H4 genome that retained
its wild-type CREBBP profile, whereas both the primary
samples (tumor and ascites-derived cells) expressed three
identical mutations. Within the five mtDNA mutant
profile clones, four mutations in the nuclear hormone
receptor domain (NHRD) of the gene, eight in the
CREB binding domain (KIX), three in exons 12–17
(which have no known functional domains), three in the
bromodomain and two in the HAT domain, were
identified (Figure 2a). The tumorigenic clones A2 and
A4-T showed one identical non-synonymous mutation in the HAT domain (asparagine to isoleucine); an
identical non-synonymous mutation (aspartate to
asparagine) was also detected in this domain in the
non-tumorigenic clones B2 and C4. These mutations could be significant as the HAT domain mediates
a key acetylation function of the protein and triggers
off several downstream signal transduction pathways
(Goodman and Smolik, 2000). In this particular
Oncogene
Nuclear–mitochondrial genomic profiling
AA Wani et al
6340
Figure 2 Representation of genetic alterations in the nuclear cAMP responsive element binding binding protein (CREBBP) gene.
(a) Mutations in the coding region of the CREBBP gene. Functional domains are represented as boxes and include: NTAD
(NH2-terminal transactivation domain), KIX (CREB-binding domain). BRD (Bromodomain), HAT (histone acetyltrasferase domain)
and CTAD (COOH terminal transactivation domain), A2, A3, A4-T, B2 and C4 are the mutant clones; H4 is a representative of the
mitochondrial wild-type clone; PT is the primary tumor and Asc is the ascites-derived tumor cells. (b) Mutations in the non-coding
region of the CREBBP gene. Introns are represented in boxes: various mutations are indicated by different icons those common to all
samples are indicated as ., a putative hotspot common to the mutant clones is represented as *.
instance, the presence of a specific mutation coincidentally in the two tumorigenic/mtDNA mutant group
clones is indicative of an altered protein with implied
association of a predisposition or definitive role in
tumorigenity. Mutations in the bromodomain were also
expressed exclusively by the two tumorigenic clones.
This conserved domain has a specific role during histone
acetylation (Deng et al., 2003), binds specifically to p53
(Dhalluin et al., 1999) and is responsible for p53
acetylation-dependent co-activator recruitment during
signaling (Grossman, 2001). Clones A3 (non-tumorigenic) and A4-T (tumorigenic) expressed two identical
and one differential mutation each in the NHRD
domain; one differential mutation was seen in each
of the A2 (tumorigenic) and B2 (non-tumorigenic)
clones in the KIX domain (KIX is the region where
CREB binds and mediates the genomic effects of cAMP
(Murata et al., 2001). The mutations in A2 and A4-T in
HAT and bromodomains are suggestive of being
responsible for the tumorigenic functions of these
clones; the differential mutations between these clones
in the NHRD and KIX domain could contribute to the
qualitative differences during tumor formation and
progression from these clones in animal models (A4-T
has been shown earlier to be a more aggressive clone
than the A2 (Bapat et al., 2005).
Oncogene
cAMP responsive element binding binding protein intron/
exon boundary mutation analyses
As in case with exon mutations, both the primary
samples showed identical intron-exon boundary (I/Eb)
mutation profiles, whereas three nucleotide variations
were found to be common to all the samples (Figure 2b),
and may be considered as polymorphisms. Each singlecell clone expressed a distinct profile with some
mutations in common with other clones – in various
permutations and combinations. The primary tumor
and tumor-derived ascites samples expressed one variant
in the donor site of intron 9 whereas H4 genome
expressed a variant in the acceptor splice site of intron
18 (Table 3). These variations may not be significant in
view of tumorigenecity as the cells were not seen to be
tumorigenic despite the presence of the mutations.
The most striking intronic mutation was expressed in
the five mutant clones within a polyA tract (A13)
located at the tail end of intron 18 (IVS18). A successive
deletion was expressed at this site in the five clones,
with B2 (non-tumorogenic) having a single adenine
deletion (delA), C4 and A3 (non-tumorogenic) showing
an AA deletion and the two tumorogenic clones
showing a triple AAA deletion (Figure 3a, Table 3).
These successive mutations, due to an attribute of
being a part of the normal acceptor splice site, could
Nuclear–mitochondrial genomic profiling
AA Wani et al
6341
Table 3 CREBBP Splice Site mutations
Intron
Mutation
Part of splice site
Donor/acceptor splice site
Clones in which present
IVS9
IVS13
IVS15
IVS18
IVS18
IVS18
IVS18
IVS18
IVS22
IVS22
IVS25
IVS93A>T
IVS13+10A>T
IVS15141G>A
IVS183A>T
IVS182A>T
IVS1813DELA
IVS1813DELAA
IVS1813DELAAA
IVS22+117C>G
IVS22+120T>G
T>G
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Donor splice site
Acceptor splice site
4 bp from Acceptor splice Site
Acceptor splice site
Acceptor splice site
Acceptor splice site
Acceptor splice site
Acceptor splice site
Acceptor splice site(2 bp downstream)
Acceptor splice site(5 bp downstream)
Donor splice site(1 bp upstream )
Primary
A4
Primary
A4
H4
B2
A3,C4
A2,A4
Primary
Primary
C4
tumor and ascites cells
tumor and ascites cells
tumor and ascites cells
tumor and ascites cells
Abbreviation: CREBBP, cAMP responsive element binding binding protein.
Figure 3 (a) Electropheretograms depicting a putative hotspot at the intron 17/exon18 boundary in the mutant clones.
(b) Phylogenetic tree obtained using PHYLIP version 3.61 software, showing the relationship among the mutant clones viz A2, A3,
A4, B2, C4 with reference CREBBP sequence obtained from NCBI. Two different methods (KITSCH and UPGMA) were used
independently on a distance matrix calculated by DNA-DIST and yielded the same pattern of clustering in both cases. Bootstrap
values are placed above branches.
have some significance in splicing; alternatively, they
may serve as a definite marker for delineating the
mutant clones within the tumor or cause dosage
effects of the gene. The tumorigenic clone A4 expressed
two more splice site variants in the acceptor site of
introns 13 and 18, respectively. The latter may
complement the earlier mutation (del AAA in IVS18),
contributing further to the differential tumorigenecity between the A2 and A4-T clones. Similar
kind of intronic mutations causing splicing defects has
been reported earlier in several cancers; a comparable
study in leukemia associates such mutations with
altered levels of HAT activity of CREBBP (Shigeno
et al., 2004).
Phylogenetic divergence within the mutant clones
The above analyses cumulatively indicate that after the
mutant mitochondrial profile group diverges out from a
germ line mtDNA profile, each individual clone within
the group has the potential to become a TSC. To
understand why only a few clones of the mutant group
acquire this capability, we performed a phylogenetic
analysis to resolve the distribution coexistence within
the different clones (Figure 3b). The B2 clone (nontumorigenic) clustered with the reference CREBBP gene
at 100% homology; A3 clone branched out from this
cluster with 64% homology. A2 and A4-T (tumorigenic)
further branched out and expressed 53% similarity
with the (B2, reference sequence and A3) cluster,
but shared 100% homology with each other. C4
(non-tumorigenic) stood out as an outgroup in the tree.
Thus, one can predict that A2 and A4 are highly similar,
align on the same branch and are tumorigenic. On the
other hand, B2 and A3 have only a partial degree of
homology with the tumorigenic clones and are closer
to the reference CREBBP sequence. Lastly, C4 shows
Oncogene
Nuclear–mitochondrial genomic profiling
AA Wani et al
6342
minimum homology and is least related with any other
clone in the group yet has significantly evolved away
from the reference sequence.
Loeb, 2005). Based on our findings, the likely cascade
of events could be
(i)
Discussion
Most tumors, at the stage of detection, are a heterogeneous mixture of several subclones that makes it
difficult to establish the order of the genetic insults (van
Tilborg et al., 2000). The approach in our study aimed at
resolving such issues relating to the evolution of TSCs.
The three mtDNA mutations identified in the primary
samples were suggested to be highly significant
in epithelial ovarian cancer; yet were actually absent
in the TSC clones – although one (A10398G) was
expressed in the normal germ line profile of the patient.
Thereby, although TSCs have a distinct genetic identity
within the gross tumor as do a multitude of other nontransformed stem cell lineages, the overall dominant
expression would be germ line (at initial stages
of tumorigenesis) and mixed mutant-germ line at later
stages (due to overlap of expressions of dominating
clones in the tumor; Sidransky et al., 1992). This
suggests that genetic profiling of entire tumors could
often be a confusing exercise and possibly lead to errors
in tumor classification (Alonso et al., 2005).
A further finding is that, although TSCs have an
evasive yet distinctly identifiable mtDNA profile,
acquisition of such a mutant genome cannot be the sole
determinant of tumorigenecity. Different nuclear backgrounds and an identical mitochondrial genome in
clones with varying levels of tumorigenecity is a distinct
indication that mitochondrial mutations do not initiate
the tumorigenic process. However, previous observations with cancer cells expressing altered metabolic
patterns indicate the possibility that some of the
variations in the specific mutant mtDNA profile may
alter mitochondrial energy conversion efficiencies in
those particular lineages by increasing the expression of
genes involved in glycolysis and reducing Kreb’s cycle
activity (Costanzo et al., 2002). Negative clone selection
also occurs; some of the germ line profile lineages
may thus be eliminated during tumor progression, for
example, the lineage that accumulates a mutation
leading to a termination codon in the CoxI gene. Thus,
concurrent to the positive selection, a continuous
turnover of stem cell populations through neutral
evolution of untransformed stem cell lineages during
tumor progression can occur.
What exactly defines the events within a lineage on a
trajectory towards tumorigenesis? Our study clearly
shows that TSCs have a distinctive mitochondrial–
nuclear signature within a tumor, and gives a rare
glimpse of changing mutational patterns accompanying
varying stem cell dynamics and turnover within the
organ during tumor evolution. It has been realized
that the classically defined ‘two hits’ (first germ line
and second somatic) produce only a benign precursor
lesion and that additional events are necessary
for transformation (Knudson, 1996; Beckman and
Oncogene
Alterations in the mitochondrial genome: at
present, it is not clear whether this altered pattern
gives a selective advantage, or only leads to
increased reactive oxygen species production.
Further, the emergence of a mutant mitochondrial
profile is either as a consequence of increased
mutagenesis or differential segregation of mutant
mtDNA copies in the cell. Acquisition of the
mutant profile however, although not capable of
mediating transformation, may represent a mutator
phenotype that could in addition to the biochemical
effects, accelerate the rate of mutations in the
mutant clones, whereas some of the others lead to
negative clonal selection.
(ii) The second target is the nuclear genome wherein
neutral evolution and mutagenesis in various genes
(exons as well as introns) leads to the currently
prevalent caveat of tumor suppressor inactivation
and oncogene activation leading to the emergence
of a TSC.
(iii) Further evolution may not be entirely neutral, but
could be determined by extrinsic factors. Chemotherapy can eliminate the initial TSC population
along with a majority of the tumor cells. However,
the genetic variation at the nuclear level within the
mutant group clones may ensure the survival of
some of these populations that have a predilection
towards transformation. Such a situation could lead
to tumor recurrence, which at the clinical level is
often more aggressive than the primary disease.
Our findings will thus be critical in addressing several
themes such as expression analyses of tumors, surface
antigens that may be used for developing targeted
antibodies and drug discovery towards the development
of chemotherapy for eliminating the entire mutant
group including the TSCs. This would lead to better
optimization of therapeutic regimes for ovarian cancer.
Further developments on the identification of continuing mutagenesis in subsets with the mutant phenotype
could also provide a mechanism for monitoring minimal
residual disease.
Materials and methods
Primary tissue samples and cell cultures
The present study is approved by the Bio-Ethics Committee of
NCCS; informed consent was obtained from all patients.
Frozen samples of primary ovarian serous adenocarcinomas
(OT6, OT13, OT14, OT17, OT19), papillary adenocarcinoma
(OT9) and cystadenomas (OT10, OT22) and three samples of
tumor cells from the ascites of patients with metastatic serous
adenocarcinomas (OASC1, OASC2 and OASC3) were retrieved from our tissue bank and used for DNA analysis.
Thirty normal ethnic controls belonging to the same haplogroup as the patients and 75 downloaded sequences covering
almost all known Indian haplogroups (Rieder et al., 1998)
were also used for comparison (data not shown).
Nuclear–mitochondrial genomic profiling
AA Wani et al
6343
In the second part of the study, we used an in vitro model
developed earlier by us that comprises 19 single-cell clones
isolated from the ascites of an ovarian serous adenocarcinoma
patient (Bapat et al., 2005). Of these, one clone was
tumorigenic (A2); another one (A4) underwent a spontaneous
transformation in vitro, whereas the remaining 17 clones were
non-tumorigenic. In addition, following samples from the
same patient were also used
sequences obtained were applied for constructing phylogenetic
trees based on evolutionary distances using the neighbor
joining method implemented through NEIGHBOUR (DNADIST) from the PHYLIP version 3.61 package (Andrews et al.,
1999). A total of 1000 bootstrap value replicates resembling
data sets were generated using the programs SEQBOOT to
build a consensus tree. Tree files were viewed in the
PHYLODRAW program (Felsenstein, 1993).
(i) Paired sample of cryopreserved primary tumor and cells
isolated from ascites,
(ii) two immortalized, unsorted (heterogenous), non-tumorigenic populations designated as M1 and M2 that were
derived from the ascites cultures and
(iii) a control sample that comprised of primary MNCs isolated
from a periferal blood sample of the son of the patient.
Statistical analysis
Two-tailed w2 analysis was carried out for determining the
significance of mtDNA mutations in the primary tumor
samples (n ¼ 12) in comparison with controls (n ¼ 106).
The a-value for the test was 0.005.
Polymerase chain reaction amplification, DNA sequencing and
mutation analysis
Genomic DNA was extracted using DNeasy tissue kit (Qiagen,
Germany). The entire mitochondrial genome was amplified in
24 separate reactions using overlapping primer pairs as
described earlier (Rieder et al., 1998). The use of large PCR
products excluded the possibility that nuclear pseudogenes
will complicate the analysis (Choi et al., 2000). The CREBBP
gene was amplified as above, using 26 sets of primers designed
at the exon–intron boundaries from the five clones expressing
a mutant profile, one representative clone of the germ
line profile, viz. H4 and the primary cells derived from the
tumor and the ascites. For more details see Supplementary
Information.
Mitochondrial gene and a.a. maps for the mtDNA-encoded
proteins used in the analysis were from the MITOMAP:
A Human Mitochondrial Genome Database; with additional
data for the mutation and polymorphism analysis from the
Human Mitochondrial Genome Database (mtDB) website
maintained by Uppsala University. Comparison of mutations
found in the different segmentation categories was with the use
of Bayesian confidence intervals (CIs) and non-informative
priors. Confidence intervals including zero indicate the
difference not to be statistically significant. CREBBP variant
Abbreviations
CREBBP, cAMP responsive element binding binding protein;
TSC, tumor stem cells; mtDNA, mitochondrial DNA; ROS,
reactive oxygen species; AMP, adenosine mono-phosphate;
CRS, Cambridge Reference Sequence; NHRD, nuclear
hormone receptor domain; KIX, CREB binding domain;
HAT, histone acetyltransferase.
Acknowledgements
We thank Dr GC Mishra, Director, National Center for Cell
Science (Pune, India) for encouragement and support. This
work is funded by the Department of Biotechnology (DBT).
Ms N Sharma receives a research fellowship from the Council
of Scientific and Industrial Research (CSIR). We also thank
Dr CB Koppikar (Jehangir Hospital, Pune, India) and
Dr Sanjay Gupte (Gupte Hospital, Pune) for providing the
tissue and tumor samples, Mr Sarang Satoor for the backup in
DNA sequencing and Mr AM Mali for excellent technical
assistance. Statistical analysis was carried out under the
kind guidance of Dr AP Gore (Department of Statistics, Pune
University, India).
References
Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison
SJ, Clarke MF. (2003). Proc Natl Acad Sci USA 100:
3983–3988.
Alonso A, Alves C, Suarez-Mier MP, Albarran C, Pereira L,
Fernandez de Simon L et al. (2005). J Clin Pathol 58: 83–86.
Andrews RM, Kubacka I, Chinnery PF, Lightowlers RN,
Turnbull DM, Howell N. (1999). Nat Genet 23: 147.
Bapat SA, Mali AM, Koppikar CB, Kurrey NK. (2005).
Cancer Res 65: 3025–3029.
Beckman RA, Loeb LA. (2005). Genetics 171: 2123–2131.
Bonnet D, Dick JE. (1997). Nat Med 3: 730–737.
Calabrese P, Tavare S, Shibata D. (2004). Am J Pathol 164:
1337–1346.
Choi J, Jung H, Kim H, Cho H. (2000). Bioinformatics 16:
1056–1058.
Coller HA, Khrapko K, Herrero-Jimenez P, Vatland JA,
Li-Sucholeiki XC, Thilly WG. (2005). Mutat Res 578:
256–271.
Costanzo A, Merlo P, Pediconi N, Fulco M, Sartorelli V, Cole
PA et al. (2002). Mol Cell 9: 175–186.
Deng Z, Chen C, Chamberlin M, Lu F, Blobel GA, Speicher D
et al. (2003). Mol Cell Biol 23: 2633–2644.
Dhalluin C, Carlson JE, Zeng L, He C, Aggarwal AK, Zhou
MM. (1999). Nature 399: 491–496.
Feinberg AP, Ohlsson R, Henikoff S. (2006). Nat Rev Genet 7:
21–33.
Felsenstein J. (1993). Phylip – Phylogeny Inference Package.
Version 3.5. University of Washington: Seattle, WA.
Florian S, Sonneck K, Hauswirth AW, Krauth MT,
Schernthaner GH, Sperr WR et al. (2006). Leukemia
Lymphoma 47: 207–222.
Goodman RH, Smolik S. (2000). Genes Dev 14: 1553–1577.
Grossman SR. (2001). Eur J Biochem 268: 2773–2778.
Hayashida K, Omagari K, Masuda J, Hazama H, Kadokawa
Y, Ohba K et al. (2005). Cell Biol Int 29: 472–481.
Human Mitochondrial Genome Database http://www.
genpat.uu.se/mtDB.
Kitabayashi Y, Aikawa LA, Nguyen A, Ohki YM. (2001).
EMBO J 20: 7184–7196.
Knudson AG. (1996). J Cancer Res Clin Oncol 122: 135–140.
Kung AL, Rebel VI, Bronson RT, Ch’ng LE, Sieff CA,
Livingston DM et al. (1999). Genes Dev 14: 272–277.
Michor F, Iwasa Y, Nowak MA. (2004). Nat Rev Cancer 4:
197–205.
Oncogene
Nuclear–mitochondrial genomic profiling
AA Wani et al
6344
MITOMAP (2005). A Human Mitochondrial Genome Database
http://www.mitomap.org.
Murata T, Kurokawa R, Krones A, Tatsumi K, Ishii M, Taki
T et al. (2001). Hum Mol Genet 10: 1071–1076.
O’Brien TW, O’Brien BJ, Norman RA. (2005). Gene 354:
147–151.
Pakendorf B, Stoneking M. (2005). Ann Rev Genomics Hum
Genet 6: 165–183.
Palanichamy MG, Sun C, Agrawal S, Bandelt HJ, Kong QP,
Khan F et al. (2004). Am J Hum Genet 75: 966–978.
Parfait B, Rustin P, Munnich A, Rotig A. (1998). Biochem
Biophys Res Commun 247: 57–59.
Petrij F, Giles RH, Dauwerse HG, Saris JJ, Hennekam RC,
Masuno M et al. (1995). Nature 376: 348–351.
Rieder MJ, Taylor SL, Tobe VO, Nickerson DA. (1998).
Nucleic Acids Res 2: 967–973.
Schmid M, Steinlein C, Haaf T. (2004). Cytogenet Genome Res
104: 277–282.
Shigeno K, Yoshida H, Pana L, Luo JM, Fujisawa S, Naito K
et al. (2004). Cancer Lett 213: 11–20.
Shin MG, Kajigaya S, McCoy Jr JP, Levin BC, Young N.
(2004). Blood 103: 553–561.
Sidransky D, Mikkelsen T, Schwechheimer K, Rosenblum
ML, Cavanee W, Vogelstein B. (1992). Nature 355:
846–847.
Singh KK, Kulawiec M, Still I, Desouki MM, Geradts J,
Matsui S. (2005). Gene 354: 140–146.
Singh SK, Clarke ID, Terasaki M, Bonn VE, Hawkins C,
Squire J et al. (2003). Cancer Res 63: 5821–5828.
van Tilborg AAG, de Vries A, de Bont M, Groenfeld LE, van
der Kwast TH, Zwarthoff EC. (2000). Hum Mol Genet 9:
2973–2980.
Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).
Oncogene