[CANCER RESEARCH 60, 6510 – 6518, November 15, 2000]
Clonal Selection versus Genetic Instability as the Driving Force in
Neoplastic Transformation1
Ming Chow2 and Harry Rubin3
Department of Molecular and Cell Biology and Virus Laboratory, Life Sciences Addition, University of California, Berkeley, California 94720-3200
ABSTRACT
Recent clonal studies of spontaneous neoplastic transformation in cell
culture indicate that it develops at confluence in a small minority of
individual clonal populations before it does in the uncloned parental
culture. Either preferential selection of spontaneous variants or genetic
destabilization in clones can be inferred to explain the result. In the
present experiments, using a subline of NIH 3T3 cells that is relatively
refractory to transformation, we demonstrate unequivocally that transformed foci appear under selective conditions in some clones long before
there is any sign of neoplastic change in the polyclonal culture from which
they were derived. Because the transformed cells that appear in the
susceptible clones are not inhibited in the size or number of foci formed on
a confluent background of the uncloned parental population, the genetic
events underlying transformation must occur much less frequently in the
latter. This disparity can be accounted for by the much larger number of
selectable cells in the susceptible clones at confluence than in the parental
culture, where such cells are a minority. The preferential transformation
exhibited by experimental isolation and expansion of susceptible clones
accords with evidence from various sources that neoplastic transformation in culture is a multistep process dependent primarily on selection of
spontaneously occurring genetic variants. There is no necessity to posit a
significant role for genetic destabilization in neoplastic transformation.
These considerations bolster computer models of human cancer that
implicate selective expansion of rogue clones rather than genetic instability as the driving force in the origin of most tumors. Both the genetics of
the selected clone and the epigenetics of the selective environment would
then contribute to tumor development.
INTRODUCTION
It is well established that most human cancers are of monoclonal
origin (1– 4). Given the nature of neoplastic growth, selection is an
intrinsic factor in the expansive growth of the progeny of a single cell
that produces a tumor. In recent years there has been an emphasis on
genetic instability as the driving force in the development of cancer
(5). Much of this emphasis can be attributed to the multistep progression of cancer and the conventional estimates of normal somatic
mutation rates that are far too low to account for its incidence (6). It
was suggested that a mutator phenotype arises in which the rate of
mutation is greatly increased. Mutations in mismatch repair genes do
in fact raise the rate of mutation, particularly in simple sequence
repeats, and underlie hereditary nonpolyposis colorectal cancer (7).
Defects in mismatch repair do not, however, precede the APC4
mutations which provide the selective growth advantage that initiates
most cases of the much more prevalent sporadic type of colorectal
Received 3/15/00; accepted 9/12/00.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance with
18 U.S.C. Section 1734 solely to indicate this fact.
1
This work was supported by grants from the Council for Tobacco Research and the
Elsasser Family Fund.
2
Present address: Gene Logic, Inc., 2001 Center Street, Berkeley, CA 94704.
3
To whom requests for reprints should be addressed, at Department of Molecular and
Cell Biology and Virus Laboratory Life Sciences Addition, University of California,
Berkeley, CA 94720-3200. Phone: (510) 642-6617; Fax: (510) 643-6791; E-mail:
[email protected].
4
The abbreviations used are: APC, adenomatous polyposis coli; LDP, low density
passage; MCDB 402, molecular, cellular, and developmental biology medium 402; CS,
calf serum; PD/D, population doublings per day.
cancer (8, 9). Computer modeling indicates that selection is the
driving force in the initiation of most human cancers and genetic
instability is a late event in their development (10, 11). On this basis,
mutation that confers selective growth advantage would be considered
the critical event in the origin of cancer.
Although the studies of the genetics of human cancer have highlighted
the role of selection, they provide little information on the dynamics of
the process. However, studies of spontaneous neoplastic transformation
in cell culture provide considerable insight on the subject. The first
indication that selection plays a role in spontaneous transformation came
from studies of the adaptation of mouse embryo fibroblasts to cell culture,
which revealed that there were large increases in saturation density after
multiple passages of cells at high density (12). Because the multiplication
of normal cells is inhibited by contact at high population density, it
appeared that there was selection of variants that overcome that inhibition, a characteristic of neoplastic cells. The role of selection was supported by the finding that mouse embryo fibroblasts passaged at high
density gained the capacity to produce sarcomas in syngeneic mice within
3 months, whereas those passaged at low density did not become tumorigenic through the entire 9 months of the study (13). Additional support
for selection as the driving force in spontaneous transformation came
from findings that it was produced in established lines of mouse fibroblasts by procedures that selectively inhibit the multiplication of normal
cells, such as suspension in soft agar (14), long-term contact inhibition in
confluent cultures (15), or reduced growth rate in either low serum
concentration (16) or in fetal serum with suboptimal growth-supporting
activity (15, 17).
All of the foregoing studies used fibroblasts, but most human
cancers are of epithelial origin (18). It is therefore of particular
significance for human cancer that selective conditions were found to
promote transformation in epithelial cells. Diploid rat liver epithelial
cells kept in a stationary state at confluence for 3 weeks between
every monthly passage gained the capacity in only 20 cell divisions to
produce liver carcinomas in syngeneic rats, whereas those passaged
weekly before they became confluent did not produce tumors until
they had undergone about 6 times as many divisions (19). A significant sidelight occurred when the liver epithelial cells were treated
with a mutagenic carcinogen before being cultured under selective
and nonselective conditions. The treated cells became tumorigenic
under both regimes at about the same disparate times as they had in
spontaneous transformation, indicating that selection was the dominant force in transforming carcinogen-treated cells as well. Indeed, the
carcinogen seemed to add little to the spontaneous variation that
provided the altered cells for selection. There was no correlation
between aneuploidy and tumorigenicity in the spontaneously transformed populations: populations of aneuploid cells often were nontumorigenic. However, there was a correlation in the carcinogen-treated
cells, indicating that the carcinogen treatment produced a specific type
of aneuploidy that was associated with carcinogenesis.
The question of selection versus genetic instability arose again in
spontaneous transformation of the NIH 3T3 line of mouse fibroblasts.
Cells from the original transformation-sensitive stock of NIH 3T3
cells gave rise to clones among which there was a small minority that
started to produce a few dense transformed foci at confluence sooner
than the polyclonal population from which the clones were derived
6510
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
CLONAL SELECTION IN NEOPLASTIC TRANSFORMATION
(20). When six of the clones were cocultured before the first round of
confluence, they produced many fewer dense foci in subsequent
rounds than expected from averaging the number that occurred in
parallel cultures of the individual clones (21). It was proposed that the
individual clones were genetically unstable, and intimate contact with
other clones stabilized them. We pursued this proposal further in the
present study using clones of a subline of NIH 3T3 cells that is
relatively refractory to spontaneous transformation. We report that a
minority of the clones produced well-defined transformed foci well
before there was any sign of transformation in the uncloned parental
culture. A quantitative analysis of the results shows that the disparity
can be fully accounted for by the much smaller number of transformable cells from the susceptible clones in the polyclonal parental
population at confluence than there are in each susceptible clone at
confluence. Simple numerical accounting thus eliminates the need for
the arbitrary assumption of genetic destabilization by clonal isolation
to explain preferential transformation in the minority of clones that
were susceptible.
This conclusion, of course, concurs with William of Ockham’s
maxim, “Multiplicity ought not to be posited without necessity,”
otherwise known as Occam’s Razor.
assay of the first set and after each of three additional assays for groups 2 and 5.
Growth rates were initiated with the cells from the confluent assays culture by
passaging them once at low density (5 ⫻ 103 cells/dish) for 2 days to recover from
contact inhibition, and passaging them again at low density into 4 culture dishes.
Two of the cultures were counted at 1 day and another two at 4 days. These
measurements of growth rate were done in parallel with similar ones of cells from
the regular LDPs that had never experienced crowding. Growth rates were calculated as PD/D. The relative growth rate of cells recovered from the confluent assay
cultures was determined by dividing its PD/D by that of the control LDP cells. All
measurements of growth rate of the clones were accompanied by the same
measurements of the uncloned culture.
The best description of their morphology comes from viewing photographs
of selected clones and the uncloned culture. Those confluent assay cultures that
were not trypsinized for passage or growth rate determinations were fixed with
Bouin’s reagent and stained with 4% Giemsa buffered at pH 7.0 to determine
the extent of transformed focus formation. As is documented with photographs
(e.g., Fig. 4), the uncloned culture produced no discernible foci in the first set
of serial assays, but many of the clones did. However, the foci produced by the
A⬘ clones did not reach the large size and high cell density of those produced
by the transformation-sensitive SA⬘ cells in our previous studies (20, 23, 25).
The scale used to characterize the degree of morphological transformation is
shown in Fig. 6. Where the transformation was particularly strong, it resulted
in a marked increase in saturation density of the assay culture.
MATERIALS AND METHODS
RESULTS
The A⬘ cells used here are a transformation-resistant subline (20) of the
Growth Rates of Clones in First Four LDPs. Fig. 1 shows the
original NIH 3T3 line of mouse embryo cells (22). They were derived by 146 growth rates of the cells in every clone during the first four LDPs
LDPs from the primal stock of NIH 3T3 cells. The original transformation- arranged according to group number. The clones of group 1 were
sensitive stock of NIH 3T3 cells (which we designated SA⬘) as well as its
relatively uniform with a group-averaged growth rate of 1.85 ⫾ 0.08
clones readily underwent progressive neoplastic transformation in serial
PD/D (the “⫾” part is “the average of the absolute deviations of data
rounds of incubation at confluence in a standard assay for transformed foci (15,
20). By contrast, the A⬘ subline was difficult to transform under these condi- points from their mean”). Four of the group 2 clones multiplied
relatively rapidly (averaged 1.86 ⫾ 0.14 PD/D), but two of them (6 h
tions (23) and in the present experiments failed to exhibit unequivocal focus
formation through many serial rounds of the standard assay. Cells were averaged 1.27 ⫾ 0.07 PD/D, and 8 h averaged 1.57 ⫾ 0.03 PD/D)
maintained by LDPs of 5 ⫻ 103 cells in 21-cm2 plastic culture dishes. The were distinctly slower than the rest. The uncloned A⬘ culture was kept
growth medium was MCDB 402 (24), containing 10% CS vol/vol. A primary with group 2 and had a consistently high growth rate (averaged
(1°) assay for transformed focus production was done by seeding 105 cells in 1.90 ⫾ 0.09 PD/D). There was a wide spread in the growth rates of
MCDB 402 with 2% CS, and incubating the cultures for 14 days with medium groups 3, 4, and 5 (1.64 ⫾ 0.17 PD/D, 1.74 ⫾ 0.14 PD/D, and
changes every 3– 4 days. The cells grew to confluence in 4 –5 days, and 1.49 ⫾ 0.27 PD/D, respectively) with group 5 exhibiting the slowest
remained at confluence with no further increase in cell number between 7 days
growers. The rank order of growth rates of individual clones in each
and the end of the assay at 14 days. Under this selective condition, progresgroup was mostly consistent throughout the four passages, indicating
sively transforming and fully transformed cells continued to multiply to form
foci of overlapping cells interspersed among the surrounding monolayered that they represented heritable differences between the clones.
Reduction in Growth Rates of Clones Subcultured after Serial
population. Secondary (2°) assays were duplicates of the 1° assay using sister
Assays
at Confluence. Table 1 shows the average growth rates for each
cultures of the latter, and further serial assays up to the sixth (6°) assay were
done in the same way using cells from the termination of the previous assay. group of clones as well as the uncloned culture for the LDPs and for cells
In some of the later serial assays, when heavy transformation was expected, recovered from the first two serial assays. Experimental cells of all groups
104 or 103 cells were also diluted with 105 nontransformed uncloned A⬘ cells showed a decrease in growth rate after the assays at confluence despite
from the LDPs that had never been at confluence. Some assays were deliber- the fact that they were given a 2-day passage at low density to recover
ately prolonged by extending the incubation period to 28 days. The cell count from the constraint of confluence before passage for growth rate deterat the end of each serial assay represented the saturation density of the culture.
mination. The extent of reduction after the 1° assay varied from 18 to
The cells were cloned by seeding an average of one cell/well in 96-well
25% for the clones, and 15% for the uncloned culture. The relative
plates (Falcon 3072) with growth medium MCDB 402 plus 10% CS vol/vol.
The approximate number of cells/well was monitored under the microscope reduction was apparently less after the 2° assay. However, a close
and recorded at 2, 4, 5, and 6 days. Wells containing more than one colony inspection of the data shows there was an anomalous drop in the growth
were excluded. Clones were designated by the coordinate position of the well rate of the LDP controls of some clones rather than a rise in the growth
they occupied. Twenty-seven clones were harvested by trypsinization at 5, 6, rates of their experimental counterparts. The results were consistent with
and 7 days and divided into five groups, as shown in Figs. 1-3, depending on
the idea that the prolonged period of constraint at confluence inflicts
their initial growth rates as estimated from the cell number in the wells. The heritable damage on the cells (23, 26).
clones were expanded in 21-cm2 dishes and subcultured five times in LDPs of
Fig. 2 shows the relative growth rates of groups 2 and 5 through the
5 ⫻ 103 cells/dish every 4 days. The uncloned cells were maintained in parallel
5° assay at confluence. With the exception of clone 8a of group 2 and
with the clones by the same LDPs. The growth rate of each clone and of the
clones 8b and 5c of group 5, all of the clones and the uncloned culture
uncloned culture was estimated by the cell yield at each passage. At the fifth
exhibited reduced growth rates after assay at confluence. However,
passage, an aliquot of the cells was used to initiate the first set of six serial
assays at confluence for transformation, and the remaining cells were used to the heritable damage implied by these results showed no correlation
continue the LDPs. At the 13th LDP, a second set of four serial transformation with the extent of transformation described below, particularly in the
case of the uncloned culture that showed no evidence whatever of
assays was initiated.
The growth rate of every clone was determined after the first and second serial transformation in the first five assays at confluence.
6511
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
CLONAL SELECTION IN NEOPLASTIC TRANSFORMATION
Fig. 1. Growth rates of clones and uncloned culture in first four LDPs. The growth rates of every clone in the five groups were estimated in each of the first four LDPs by dividing
the yield of 4 days by the number of cells seeded and converting the value into PD/D. Parallel determinations were made in LDPs of the uncloned A⬘ culture. The results are shown
by group, and the A⬘ cells are recorded with group 2 in a broken line. Bars, variation between two independent measurements at each point.
Saturation Densities of Clones in Serial Assays at Confluence. Fig. 3 shows the saturation densities after each of the six
assays of the first set. The uncloned culture (dotted line in group 2)
had a higher saturation density (6 ⫻ 105) than any clone in the 1°
assay, but it dropped to half that value in the 2° assay and remained
at that low level for all of the subsequent assays. It produced no
foci up to the 6° assay, when some very faint foci appeared (Fig.
4). Clone 1a of group 1 showed a steady increase in saturation
density between assays 3° and 6° with a final value above 106
cells/culture in the final assay. This was consistent with the appearance of many small foci in clone 1a beginning in the 3° assay.
Clone 10c of group 2 exhibited a marked increase in saturation
density between the 5° and 6° assay, when it went from producing
a small number to a large number of moderately dense foci. Clone
9g of group 2 also increased significantly in saturation density in
the 6° assay as well as in the number of moderately dense foci.
Clone 6h displayed a more modest increase in saturation density in
the 6° assay and had a large number of foci, but they were lighter
than those of 10c and 9g. In group 3, clone 2c increased steadily in
saturation density between assays 3° and 6° just as clone 1a of
group 1 had. This was in keeping with a large increase in the
number of moderately dense foci. The only clone in groups 4 and
5 to show a significant increase in saturation density was clone
11a. It had already produced a few light foci in the 1° assay. These
became denser and larger in the 3° assay and increased in number
in the 4°, 5°, and 6° assays, which had increases in saturation
density. Fig. 3, bottom right, also shows the average and SE of
saturation densities for all of the clones that began to exceed that
of the uncloned culture in the 4° assay. Whereas the saturation
density measurements of the late assays showed prominent increases for five of the clones with no increase for the uncloned
culture, it was evident by inspection that many clones were producing visible foci that did not result in saturation density increases. An elementary way to document this, which captures the
essence of the phenomenon that more sophisticated measurements
cannot, was through photographs of the culture dishes. These
included cultures from the first set of six serial assays and cultures
from a second set of four serial assays as described below.
6512
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
CLONAL SELECTION IN NEOPLASTIC TRANSFORMATION
Table 1 Reduction in growth rates (PD/D) of clones and their uncloned parental
culture after recovery from confluent assays
Cells were passaged at low density after the 1° and 2° assays at confluence, incubated
for 2 days to recover from confluence, then passaged again at low density to determine
growth rates (Column III). Cells from LDPs of each clone and from the uncloned parental
culture were used as controls (Column II). The average PD/D and standard deviations for
each group of clones are shown for the 1° and 2° assays of the first set, and the ratio of
the post-assay growth rate to LDP growth rate is in the last column.
Column I
Column II
Column III
Clonal group
Cells from LDPs
Cells from confluent
culture
Column III
(PD/D)
Assay no. (PD/D)
Column II
1.77 ⫾ 0.19
1.60 ⫾ 0.10
1.72 ⫾ 0.10
1.61 ⫾ 0.12
1.54 ⫾ 0.17
1.42 ⫾ 0.21
1.62 ⫾ 0.16
1.52 ⫾ 0.11
1.52 ⫾ 0.20
1.44 ⫾ 0.17
1.93
1.71
(1°) 1.33 ⫾ 0.18
(2°) 1.41 ⫾ 0.17
(1°) 1.36 ⫾ 0.13
(2°) 1.37 ⫾ 0.10
(1°) 1.17 ⫾ 0.32
(2°) 1.20 ⫾ 0.26
(1°) 1.35 ⫾ 0.12
(2°) 1.40 ⫾ 0.13
(1°) 1.24 ⫾ 0.24
(2°) 1.13 ⫾ 0.31
(1°) 1.63
(2°) 1.56
0.75
0.89
0.79
0.80
0.76
0.84
0.83
0.92
0.82
0.78
0.84
0.91
1
2
3
4
5
A⬘
Visual Demonstration of Preferential Progressive Transformation in Clones. Fig. 4A illustrates progressive transformation in 2
clones (1a and 2e) and its absence in the uncloned A⬘ parental cells
spanning six serial assays of the first series. Clone 1a was negative in
the 1° assay, and began showing light foci in the 3° assay (not shown);
the light foci became confluent with one another in the 4° assay, and
progressed to small, dense foci on a confluent background of light foci
in the 6° assay. Clone 2e was also negative in the 1° assay but began
to produce light foci in the 2° assay (not shown) that progressed to
discrete, moderately dense foci in the 4° assay and to dense foci on a
confluent background of lighter foci in the 6° assay. The background
of light foci was not as thick as that of clone 1a in the 6° assay, which
was reflected in the higher saturation density of the latter (Fig. 3). The
uncloned A⬘ cells remained completely negative through the 5° assay
but exhibited scattered, small, light foci in the 6° assay. The results
indicate that the diverse cells of the uncloned parental culture, which
must have contained many cells similar to those of clones 1a and 2e,
underwent much less transformation than those clones.
We entertained the possibility that some transformation was occurring in the uncloned culture but was not expressed because of growthsuppressing interactions among the diverse cells of this polyclonal
mixture that did not occur in the more homogeneous clones. To
investigate this possibility, the individual clones and the uncloned
parental cells were diluted for the 6° assay, and 104 cells were seeded
together with a 10-fold excess of 105 A⬘ LDP cells to form a confluent
nontransformed background for the display of foci from the cloned
cells. Representative results are illustrated in Fig. 4B, in which clones
2e and 10c produce distinct transformed foci; but the 6° assay of A⬘
cells shows no evidence of focus formation on a background of A⬘
LDP cells. The 10c clone had produced small numbers of light foci
with 105 cells in the 4° and 5° assay, but did not exhibit a significant
increase in saturation density until the 6° assay, in which even 104
cells produced a large number of foci. A culture of the LDP A⬘ cells
that formed the background for the 6° assay cells is also shown. It is
evident that foci appeared in the 6° assay of the clones on the same
background on which the 6° assay of the uncloned population failed
to produce foci. This provides unequivocal evidence that the events
that underlie transformation simply had not occurred in the uncloned
culture to nearly the same extent as they had in some of its derivative
clones.
We then examined the effect of extending the period of confluence
to determine whether foci would develop in cultures that were negative in the standard 14-day assay period and also to reveal expansion
or progression of foci that were already apparent at 14 days. The
extended assay was done with cells from the second set of serial
assays and representative results can be seen in Fig. 5. The foci
present in clone 2e at 14 days increased in size at 28 days, indicating
the selective growth advantage of cells in the foci. Clone 4b exhibited
no foci at 14 days, but a few moderately dense foci and many small,
light ones could be seen at 28 days, indicating either slow growth of
the transformed cells in this clone or delayed onset of transformation.
The uncloned A⬘ cells failed to display any foci at either 14 days or
28 days. The clone 2c cultures at the bottom of Fig. 5 were from a later
assay of the second set and showed development from light to dense
foci between 14 and 28 days. The results show the diversity of
dynamics of focus formation among the clones and reinforce the
evidence of greatly reduced transformation in the uncloned population
relative to some of its clones.
Overall Trends in Transformation among All of the Clones in
the First Two Sets of Serial Assays. The degree of transformation of
every clone and of the uncloned population was classified in each
assay of the first two series according to the scale in Fig. 6. The scale
of transformation is based on the seeding of 105 cells in each assay but
was elaborated in the later assays by seeding 104 cells with an excess
(105) of nontransformed cells from the LDPs of the uncloned A⬘ cells.
Fig. 2. Relative growth rates of clones subcultured after serial assays at confluence.
After each serial assay of the first set, some cells from the clones of groups 2 and 5 were
passaged once at low density for 2 days to allow recovery from the inhibitory effects of
confluence, then passaged again to determine their growth rates over a 4-day period at low
density. The growth rates of the uncloned A⬘ cells were determined in the same way after
parallel assays. At the same time, the growth rates of the clones and uncloned cells from
the LDPs that had never been confluent were determined. The ratios of the growth rates
of the post-assay cells and the LDPs are shown. The uncloned A⬘ cells are represented in
a broken line. Bars, total variation made up of individual variations in the cell counts of
experimental and control cultures.
6513
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
CLONAL SELECTION IN NEOPLASTIC TRANSFORMATION
Fig. 3. Saturation densities of clones in serial assays at confluence. The number of cells were counted at the end (14 –15 days) of each of six serial assays at confluence of the first
set for every clone and the uncloned A⬘ cells. The saturation densities are separated by group and the A⬘ cells are recorded with group 2. The bottom graph on the right shows the
averages of the clones from each assay and the values for the uncloned A⬘ cells. Broken line in group 2 and bottom graph, uncloned A⬘ cells.
In those cases which included assays for 104 cells, the classification in
the relatively strong (⫹⫹) and moderate (⫹) categories for 105 cells
was reinforced by the appearance of discrete foci in the assay of 104
cells, which were not seen in the weak (⫾) category. The graphs in
Fig. 7 show an increase in the number of transformed clones in each
category with successive serial assays. About half of the clones had
moderate-to-relatively strong transformation in the last two assays of
the first series, whereas the uncloned A⬘ cells were negative through
assay 5° and only weakly transformed in assay 6°.
The proportion of clones in all categories of transformation was
higher in all of the second set of assays than in comparable assays of
the first set, indicating some progression during the eight additional
LDPs between the beginning of the first and second sets. The uncloned culture was negative in assays 1° and 2° of the second series,
but exhibited variable low-level transformation in assays 3° and 4°.
Hence, some transformation appeared in the polyclonal parental culture of the second set of assays at an earlier point than it did in the first
set and was associated with earlier transformation among the clones.
However, the parental transformation was at a much lower level than
might be expected from the fact that one-third of the clones already
exhibited relatively strong transformation in the 3° assay, and there
would have been many thousands of such clones in the assay of 105
cells of the parental culture. The apparent paucity of well-defined foci
in the parental culture will be analyzed below in terms of the minority
fraction of productive cells as compared with the number in some
isolated clones. In this regard, it is notable that most of the clones that
displayed unequivocal transformation in the first set of assays did so
in the second set, indicating that transformability is a heritable property of the clones, although subject to population drift in multiple
LDPs (25).
DISCUSSION
Our previous studies, which showed a higher rate of confluencemediated transformation in recently cloned cell populations than in
their polyclonal parental population (20, 21), left unresolved the
6514
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
CLONAL SELECTION IN NEOPLASTIC TRANSFORMATION
culture that underlie the transformation seen in cultures of the isolated
clones. This conclusion is augmented by the finding of large increases
in saturation density of a number of clones with successive rounds of
confluence when there was no such increase in the polyclonal popu-
Fig. 5. Effect on focus formation of extending the assay at confluence to 28 days. The
top three pairs are from the 2° assay of the second set; the bottom pair is from the 4° assay
of the second set. Some cultures were fixed and stained at the standard assay time of 14
days and others at 28 days, as indicated.
Fig. 4. Preferential progressive transformation in clones and their capacity to produce
transformed foci on a confluent background of uncloned A⬘ cells. A, 105 cells of all of the
clones and of the uncloned A⬘ culture were seeded in six serial assays at confluence of the
first set. Assays 1°, 4°, and 6° of clones 1a, 2e, and the uncloned cells are shown. B, 104
cells from the 6° assay of all of the clones and of the uncloned A⬘ culture were mixed with
105 cells of the standard LDP of the A⬘ cells and seeded for assays at confluence. The
assay of clones 2e and 10c and the uncloned A⬘ cells with a background of LDP A⬘ cells
are shown, as is the assay of 105 LDP cells by themselves at the lower right.
possibility that the mixed population simply suppressed the proliferation of transformed cells rather than somehow avoiding the events
that underlie transformation. Here we showed that the transformed
cells from about one-fifth of the clones produced characteristic multilayered foci against a confluent background of the nontransformed
polyclonal parental population. (See Figs. 4B and 6.) Hence, in the
regular seeding of 105 cells of the uncloned parental population in the
assay for focus formation, there would be at least 2 ⫻ 104 cells
capable of undergoing and clearly exhibiting transformation if they
were cultured as isolated clones, yet no clearly identified transformed
foci were detected in six serial assays at confluence. The absence of
such foci in the polyclonal parental population indicates that some
feature of the clonal mixture reduced the number of genetic changes/
Fig. 6. Scale of transformation of clones and the uncloned A⬘ culture from the 6° assay
of the first set. The assays were initiated with 105 cells in the left column, and 104 cells
mixed with 105 cells of LDP uncloned A⬘ cells in the right column. This was the first assay
in which the uncloned A⬘ cells exhibited any sign of transformation, although it was in the
(⫾) category. Only the ⫹ and ⫹⫹ categories produced distinct foci in the mixture of 104
cells with 105 LDP uncloned A⬘ cells.
6515
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
CLONAL SELECTION IN NEOPLASTIC TRANSFORMATION
Fig. 7. The dynamics of transformation in serial assays of
the first two sets. The extent of transformation was visually
evaluated according to Fig. 7 for all 27 clones and for the
uncloned A⬘ culture in each of six serial assays of the first set
and four serial assays of the second set. The number of clones
are shown in the following combined degrees of transformation: 䡺, ⫹⫹; ‚, ⫹ and ⫹⫹; E, ⫾, ⫹, and ⫹⫹. The degree
of transformation of the uncloned A⬘ cultures for each serial
assay is shown at the top of each graph. The graph at the left
is of the six serial assays of the first set. The graph at the right
is of the four serial assays of the second set.
lation (Fig. 3). The failure of focus formation in the polyclonal
population persisted even when the incubation period of the assay was
increased from 14 to 28 days, whereas the foci initiated in the clonal
cultures were increasing in size and number.
In isolating clones and expanding them to large populations in the
selective assay at confluence for transformed foci, we increase the
number of cells originating from a single cell in comparison with the
number of clonally related cells in a polyclonal population. That is,
the ratio of cell numbers in any particular clone at saturation density
to the number of phenotypically similar cells in the polyclonal population is inversely proportional to the fraction of those clones in the
polyclonal population. If we take as a class of clones, for example,
those which eventually produce well-defined ⫹⫹ transformed foci
(Fig. 6), an isolated clone of that class will reach confluence with a
much larger number of potential focus formers than the total number
of similar cells in the polyclonal population, in inverse proportion to
the fraction of ⫹⫹ focus-forming clones among all clones in the latter
population. The smaller the fraction, the larger is the differential.
There were 5 of 27 clones that consistently produced ⫹⫹ foci in the
last three assays of the first set, or roughly one-fifth the total number
of clones. Therefore, the total number of ⫹⫹ focus producers in the
polyclonal population at confluence will be about one-fifth of the
number in each of the isolated ⫹⫹ clones at confluence. The saturation density of each of the cloned and uncloned cultures remains
roughly constant between 2 and 4 ⫻ 105 cells during the first few
assays (Fig. 3). Because the production of foci is dependent on the
number of focus-forming cells at confluence, the isolated clones with
the potential for producing ⫹⫹ foci would be expected to display
such foci in a much earlier assay than the polyclonal population.
Well-defined focus production first appears in two clones in the 3°
assay of the first set, and reaches a plateau of five consistent focus
producers in the 4° assay. The failure of such foci to appear in the
polyclonal population in any of the six assays of the first set (nor in
any assay of the second set) follows from the 5-fold reduction in
number of potential ⫹⫹ cells in that population compared with the
number in each isolated clone. According to this crude estimate, it
would take at least 15 serial assays of the polyclonal population to
produce foci of the ⫹⫹ type. A similar calculation can be made for
the combined number of ⫹ and ⫹⫹ focus-producing clones, which
constitute about half of the total number of clones, e.g., if there had
been a large number of polyclonal cultures, and the number of serial
assays extended beyond six, a fraction of them might be expected to
exhibit foci by the 7° assay. If we include the somewhat marginal and
sometimes transient ⫾ focus-forming clones in the total, which
reaches about half its maximum at about the 3° assay, it is not
surprising that some weak foci appear in the polyclonal population by
the 6° assay. Similar analysis can be applied to the second set of
assays in Fig. 7. Therefore, the relative ease with which the clonal
populations are transformed compared with the polyclonal parental
population can be explained simply in terms of clonal diversity in the
latter.
In earlier papers using the original transformation-sensitive line that
produced large, dense foci (20, 21), and particularly where six isolated
clones were compared with a mixture of the six clones, we found that
some of the isolated clones progressed to dense focus formation
before any were seen in the mixture (21). This was interpreted by
assuming that isolated clones were genetically unstable at confluence,
and that they were stabilized by metabolic cooperation with one
another in the mixture, thereby preventing early steps in progression.
The simpler explanation is that only three of the six clones progressed
to dense focus formation in the 2° assay and did so at different rates.
According to the inverse proportionality between a cloned and uncloned population in the number of potential focus-formers established above, there would be a 2-fold reduction of potential focusformers in the mixture with a proportionate delay in focus formation
to assays beyond those done in the experiment.
Because the argument of clonal diversity fully accounts for the
paucity of foci in the polyclonal parental culture, the previous hypothesis of clonal destabilization is unnecessary. In fact, there is no
independent verification that clonal isolation results in an increase in
the rate of transformation-related variation. This possibility had been
considered (20, 21) because transformation was frequently associated
with a heritable reduction in the growth rate of cells at low population
density (26 –28), which could be ascribed to genetic damage with
accompanying mutations. However, selection of progressively transformed cells by LDP in low CS concentration (16) or fetal bovine
serum (15, 17) does not result in a reduction in growth rate. A
complementary negative finding is that the heritable reduction in
growth rate induced by methotrexate treatment of NIH 3T3 cells does
not produce transformation (29). Perhaps more cogently, treatment of
diploid liver epithelial cells with a mutagenic carcinogen only marginally increased their spontaneous rate of transformation under selective conditions and had no significant effect on the rate of transformation under nonselective conditions (19). Whereas we cannot
exclude a role for genetic destabilization in transformation for all
conditions, there is no reason to postulate such a role in the present
case because the preferential transformation of minority clones can be
fully accounted for by the selection and expansion of susceptible cells.
Additional evidence against the destabilizing effect of clonal isolation
is the existence of some clones even in the transformation-sensitive
subline of NIH 3T3 cells that fail to transform in many repeated
rounds of selection (25) as well as the many that show no transformation or only marginal effects in the present experiments (Fig. 7).
The conditions in culture for selective growth of cells in various
6516
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
CLONAL SELECTION IN NEOPLASTIC TRANSFORMATION
stages of transformation are those which preferentially inhibit multiplication of nontransformed cells. We achieved artificial selection in
the present paper by isolating and expanding clones, a few of which
were particularly susceptible to confluence-mediated transformation.
The contact inhibition of confluent cultures was also used to drive
selection of spontaneous transformants in a line of diploid liver
epithelial cells (19). Because all epithelial cells in the organism are in
direct contact with their homotypic neighbors, confluence per se does
not act as a selective force for carcinoma development in vivo. In the
case of colorectal cancer in humans, biallelic mutations at the APC
locus apparently confer selective advantage on stem cells in the
colonic crypt, which allows them to continue to multiply above the
bottom two-thirds of the crypt and thereby to form dysplastic adenomas (7, 30). The particular mutations that underlie selection in culture
are unknown. The differences in morphology of transformed foci (20,
31) and the number of steps in transformation (32) indicate that many
genetic loci are involved in the process. The effect of these mutations
in culture is formally the same as that of APC in the colon in that both
endow selective advantage to the cells under conditions that regulate
the multiplication of normal cells. Rapid exponential multiplication of
cultured cells minimizes or even selects against transformation (17),
perhaps because transformed cells tend to multiply, when unconstrained, at a lower rate than nontransformed cells (26 –28). The only
parallel in vivo might be the relatively unconstrained cell growth of
early embryogenesis during which tumor development is minimized.
Another form of selection for transformation in culture is to lower
the concentration of growth factors while maintaining exponential
growth at low population density (16, 17). The corollary in vivo would
be a long-term reduction in the concentration of a hormone that
stimulates cell multiplication. Accordingly it has been suggested that
the diminution of testosterone production that occurs in aging men
selects for testosterone-independent cells and contributes to the steep
increase in prostate cancer beyond the age of 50 (33). It has also been
pointed out that deprivation of estrogenic hormones, which produces
regression of spontaneous mammary tumors in mice, eventually results in progression of the residual neoplasia to a hormone-independent state (34).
Selection also plays a role in chemical carcinogenesis of experimental animals. Farber has proposed the resistant hepatocyte model of
liver carcinogenesis based on the findings that the cells in early stages
of tumor development multiply in the presence of a carcinogen which
is inhibitory to the growth of the normal hepatocytes (35). There are
a number of cases in rodents in which carcinogens apparently select
for the growth of cells with preexisting mutations in ras genes (36,
37). Long-term repeated applications of nonmutagenic promoters
alone give rise to a few liver tumors (38, 39) which apparently result
from selection of preexisting enzyme-altered cells that have a defect
in growth control (39, 40). The compensatory hyperplasia that is an
early step in experimental carcinogenesis of the skin (41) and colon
(42) may also favor clonal expansion of cells that are defective in
growth control and thereby promote the accumulation of mutations
associated with progression.
If selection is to play the dominant role in carcinogenesis, there
must, of course, be sufficient rogue clones from which to select. It was
recently pointed out that conventional estimates of mutation frequencies in normal tissues may be far too low (43). For example, 1 of 4,000
normal kidney cells in aging men was found to have a mutation at the
X-linked HPRT locus (44), which encodes a much smaller gene than
many tumor suppressor genes (43). Thus biallelic mutations of tumor
suppressor genes may not be uncommon in normal tissues; mutations
in micro- and minisatellites, some of which occur within functional
genes, occur at much higher rates. There are enough cell divisions of
normal intestinal stem cells and growing tumors to account for the
11,000 mutations/cell reported in colorectal polyps and carcinomas
(45) without inferring an increased mutation rate.5 Simpson (43)
suggests that the aging human body accumulates enough mutations to
account for multistep carcinogenesis by selection of preexisting mutations. This would be especially true if selective conditions for clonal
expansion were heightened with age. One indication that this is the
case is that rat hepatocarcinoma cells are much more tumorigenic
when inoculated into the liver of old rats than into that of young rats
(46). Another indication of tissue-wide changes with age is the slowdown in multiplication rates of intestinal mucosa accompanied by a
large increase in heterogeneity among the crypt cells in the onset of
DNA synthesis (47, 48).
In fact, the combination of accumulated mutations and disturbances
of the regulatory environment associated with aging raises the question of why cancer is not more common than it is. It may be that the
normal architecture of a tissue maintains a regulatory environment
that keeps rogue clones behaving normally or disposes of them by
apoptosis. This would be in keeping with the first principle of Elsasser’s theory of organisms, which postulates that “there can be regularity in the large where there is heterogeneity in the small: ‘order above
heterogeneity’” (49). This elementary principle has been expressed as
macrodeterminism in developmental biology (50) and as holistic
control of neuronal excitation in the brain (51). A classical example is
the teratocarcinoma of mice that is produced when young mouse
embryos are grafted into the testes of adult mice (52): inoculation of
core cells of the teratocarcinomas into blastocysts results in their
integration as normal elements in chimeric tissues of many types,
including reproductively functional sperm (53). The primary principle
of ordered heterogeneity subsumes this behavior and several other
genetic and epigenetic aspects of the cancer problem (54). It therefore
provides a theoretical foundation to account for the long-term stability
of multicellular structures despite the accumulation of many somatic
mutations (55–57).
ACKNOWLEDGMENTS
We thank John Cairns, Morgan Harris, Kenneth Kinzler, George Klein,
Lawrence Loeb, Darryl Shibata, David Sidransky, and Ian Tomlinson for their
helpful comments on the work and we thank Dorothy M. Rubin for preparing
the manuscript.
REFERENCES
1. Fialkow, P. J. Clonal origin of human tumors. Biochim. Biophys. Acta, 458: 283–321,
1976.
2. Fearon, E. R., Hamilton, S. R., and Vogelstein, B. Clonal analysis of human colorectal
tumors. Science (Washington DC), 238: 193–197, 1987.
3. Sidransky, D., Frost, P., Von Eschenbach, A., Oyasu, R., Preisinger, A. C., and
Vogelstein, B. Clonal origin of bladder cancer. N. Engl. J. Med., 326: 737–740, 1992.
4. Bedi, G. C., Westra, W. H., Gabrielson, E., Koch, W., and Sidransky, D. Multiple
head and neck tumors: evidence for a common clonal origin. Cancer Res., 56:
2484 –2487, 1996.
5. Lengauer, C., Kinzler, K. W., and Vogelstein, B. Genetic instabilities in human
cancer. Nature (Lond.), 396: 643– 649, 1998.
6. Loeb, L. A. Mutator phenotype may be required for multistage carcinogenesis. Cancer
Res., 51: 3075–3079, 1991.
7. Kinzler, K. W., and Vogelstein, B. Lessons from hereditary colorectal cancer. Cell,
87: 159 –170, 1996.
8. Young, J., Leggett, B., Gustafson, C., Ward, M., Searle, J., Thomas, L., Buttenshaw,
R., and Chenevix-Trench, G. Genomic instability occurs in colorectal carcinomas but
not in adenomas. Hum. Mutat., 2: 351–354, 1993.
9. Homfray, T. F. R., Cottrell, S. E., Ilyas, M., Rowan, A., Talbot, I. C., Bodmer, W. F.,
and Tomlinson, I. P. M. Defects in mismatch repair occur after APC mutations in the
pathogenesis of sporadic colorectal tumours. Hum. Mutat., 11: 114 –120, 1998.
10. Tomlinson, I. P. M., Novelli, M. R., and Bodmer, W. F. The mutation rate and cancer.
Proc. Natl. Acad. Sci. USA, 93: 14800 –14803, 1996.
11. Tomlinson, I. P. M., and Bodmer, W. F. Selection, the mutation rate and cancer:
Ensuring that the tail does not wag the dog. Nat. Med., 5: 11–12, 1999.
5
I. P. M. Tomlinson, personal communication.
6517
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
CLONAL SELECTION IN NEOPLASTIC TRANSFORMATION
12. Todaro, G. J., and Green, H. Quantitative studies of the growth of mouse embryo cells
in culture and their development into established lines. J. Cell Biol., 17: 299 –313,
1963.
13. Aaronson, S. A., and Todaro, G. J. Basis for the acquisition of malignant potential by
mouse cells cultivated in vitro. Science (Washington DC), 162: 1024 –1026, 1968.
14. Rubin, H. Uniqueness of each spontaneous transformant from a clone of Balb 3T3
cells. Cancer Res., 48: 2512–2518, 1988.
15. Rubin, H., and Xu, K. Evidence for the progressive and adaptive nature of spontaneous transformation in the NIH 3T3 cell line. Proc. Natl. Acad. Sci. USA, 86:
1860 –1864, 1989.
16. Rubin, A. L., Yao, A., and Rubin, H. Relation of spontaneous transformation in cell
culture to adaptive growth and clonal heterogeneity. Proc. Natl. Acad. Sci. USA, 87:
482– 486, 1990.
17. Rubin, H. Adaptive evolution of degrees and kinds of neoplastic transformation in cell
culture. Proc. Natl. Acad. Sci. USA, 89: 977–981, 1992.
18. Rodriguez, E., Sreekanitaiah, C., and Chaganti, R. S. K. Genetic changes in epithelial
solid neoplasia. Cancer Res., 54: 3398 –3406, 1994.
19. Lee, L. W., Tsao, M-S., Grisham, J. W., and Smith, G. J. Emergence of neoplastic
transformants spontaneously or after exposure to N-methyl-N⬘-nitro-N-nitrosoguanidine in populations of rat liver cells cultured under selective and nonselective
conditions. Am. J. Pathol., 135: 63–71, 1989.
20. Chow, M., and Rubin, H. The cellular ecology of progressive neoplastic transformation: a clonal analysis. Proc. Natl. Acad. Sci. USA, 96: 2093–2098, 1999.
21. Chow, M., and Rubin, H. Coculturing diverse clonal populations prevents the earlystage neoplastic progression that occurs in the separate clones. Proc. Natl. Acad. Sci.
USA, 97: 174 –178, 2000.
22. Jainchill, J. L., Aaronson, S. A., and Todaro, G. J. Murine sarcoma and leukemia
viruses: assay using clonal lines of contact-inhibited mouse cells. J. Virol., 4:
549 –553, 1969.
23. Chow, M., and Rubin, H. Relation of the slow growth phenotype and neoplastic
transformation: possible significance for human cancer. In Vitro Cell. Dev. Biol.
Anim., 35: 449 – 458, 1999.
24. Shipley, G. D., and Ham, R. G. Improved medium and culture conditions for clonal
growth with minimum serum protein and for enhanced serum-free survival of Swiss
3T3 cells. In Vitro Cell. Dev. Biol. Anim., 17: 656 – 670, 1981.
25. Chow, M., and Rubin, H. Clonal dynamics of progressive neoplastic transformation.
Proc. Natl. Acad. Sci. USA, 96: 6976 – 6981, 1999.
26. Rubin, H., Yao, A., and Chow, M. Heritable, population-wide damage to cells as the
driving force of neoplastic transformation. Proc. Natl. Acad. Sci. USA, 92: 4843–
4847, 1995.
27. Reznikoff, C. A., Bertram, J. S., Brankow, D. W., and Heidelberger, C. Quantitative
and qualitative studies of chemical transformation of cloned C3H mouse embryo cells
sensitive to postconfluence inhibition of cell division. Cancer Res., 33: 3239 –3249,
1973.
28. Smith, G. J., Bell, W. N., and Grisham, J. W. Clonal analysis of the expression of
multiple transformation phenotypes and tumorigenicity by morphologically transformed 10T1/2 cells. Cancer Res., 53: 500 –508, 1993.
29. Chow, M., Koo, J. J., Ng, P., and Rubin, H. Random, population-wide genetic damage
induced in replicating cells treated with chemotherapeutic drugs. Mutat. Res., 413:
251–264, 1998.
30. Lamlum, H., Papadopolou, A., Ilyas, M., Rowan, A., Gillet, C., Hanby, A., Talbot, I.,
Bodmer, W., and Tomlinson, I. APC mutations are sufficient for the growth of early
colorectal adenomas. Proc. Natl. Acad. Sci. USA, 97: 2225–2228, 2000.
31. Rubin, H. Experimental control of neoplastic progression in cell populations: Foulds’
rules revisited. Proc. Natl. Acad. Sci. USA, 91: 6619 – 6623, 1994.
32. Rubin, H. Incipient and overt stages of neoplastic transformation. Proc. Natl. Acad.
Sci. USA, 91: 12076 –12080, 1994.
33. Prehn, R. T. On the prevention and therapy of prostate cancer by androgen administration. Cancer Res., 59: 4161– 4164, 1999.
34. Foulds, L. Neoplastic Development, Vol. 1. New York: Academic Press, 1969.
35. Farber, E., and Rubin, H. Cellular adaptation in the origin and development of cancer.
Cancer Res., 51: 2751–2761, 1991.
36. Brookes, P. Chemical carcinogens and ras gene activation. Mol. Carcinog., 2:
305–307, 1989.
37. Cha, R. S., Thilly, W. G., and Zarbl, H. N-nitroso-N-methylurea-induced rat mammary tumors arise from cells with preexisting oncogenic Hras 1 gene mutations. Proc.
Natl. Acad. Sci. USA, 91: 3749 –3753, 1994.
38. Rossi, L., Ravera, M., Repetti, G., and Santi, L. Long-term administration of DDT or
phenobarbital-Na in Wistar rats. Int. J. Cancer, 19: 179 –1977, 1977.
39. Schulte-Hermann, R., Timmermann-Trosiener, I., and Schuppler, J. Promotion of
spontaneous preneoplastic cells in rat liver as a possible explanation of tumor
production by nonmutagenic compounds. Cancer Res., 43: 839 – 844, 1993.
40. Ogawa, K., Onoé, T., and Takeuchi, M. Spontaneous occurrence of ␥-glutamyl
transpeptidase-positive hepatcytic foci in 105-week-old and 72-week-old Fischer 344
male rats. J. Natl. Cancer Inst., 67: 407– 412, 1981.
41. Cramer, W. The early stages of carcinogenesis by 20-methylcholanthrene in the skin
of the mouse. II. Microscopic tissue changes. J. Natl. Cancer Inst., 2: 379 – 402, 1942.
42. Chang, W. Histogenesis of DMH-induced colon neoplasms in the mouse. J. Natl.
Cancer Inst., 60: 1405–1418, 1978.
43. Simpson, A. J. G. A natural somatic mutation frequency and human carcinogenesis.
Adv. Cancer Res., 71: 209 –240, 1997.
44. Martin, G. M., Ogburn, C. E., Colgin, L. M., Gown, A. M., Edland, S. D., and
Monnat, R. J. Somatic mutations are frequent and increase with age in human kidney
epithelial cells. Hum. Mol. Genet., 5: 215–221, 1996.
45. Stoler, D. L., Chen, N., Basik, M., Kahlenberg, M. S., Rodriguez-Bigas, M. A.,
Petrelli, N. J., and Anderson, G. R. The onset and extent of genomic instability in
sporadic colorectal tumor progression. Proc. Natl. Acad. Sci. USA, 96: 15121–15126,
1999.
46. McCullough, K. D., Coleman, W. R., Smith, G. J., and Grisham, J. W. Age-dependent
induction of hepatic tumor regression by the tissue microenvironment after transplantation of neoplastically transformed rat liver epithelial cells into the liver. Cancer
Res., 57: 1807–1813, 1997.
47. Lesher, S., Fry, R. J. M., and Kohn, H. I. Age and the generation time of the mouse
duodenal epithelial cell. Exp. Cell Res., 24: 334 –343, 1961.
48. Fry, R. J. M., Tyler, S. A., and Lesher, S. In: P. J. Lindop and G. A. Sacher (eds.),
Relationships between Age and Variability, pp. 43–55. Semmering, Austria: Taylor &
Francis, Ltd., 1966.
49. Elsasser, W. M. Reflections on a Theory of Organisms. Baltimore: Johns Hopkins
University Press, 1998.
50. Weiss, P. The Science of Life. Mt. Kisco, New York: Futura Publishing Co., 1973.
51. Sperry, R. W. A modified concept of consciousness. Psychol. Rev., 76: 532–536,
1969.
52. Stevens, L. The development of transplantable teratocarcinomas from intratesticular
grafts of pre- and post-implantation mouse embryos. Dev. Biol., 21: 364 –382, 1970.
53. Mintz, B., and Illmensee, K. Normal genetically mosaic mice produced from malignant teratocarcinoma cells. Proc. Natl. Acad. Sci. USA, 72: 3585–3589, 1975.
54. Rubin, H. Cancer as a dynamic developmental disorder. Cancer Res., 45: 2935–2942,
1985.
55. Parsons, R., Li, G-M., Longley, M., Modrich, P., Liu, B., Berk, T., Hamilton, S. R.,
Kinzler, K. W., and Vogelstein, B. Mismatch repair deficiency in phenotypically
normal human cells. Science (Washington DC), 268: 738 –740, 1995.
56. Vijg, J., Dollé, M. E., Martus, H-J., and Boerrigter, M. E. Transgenic mouse models
for studying mutations in vivo: applications in aging research. Mech. Ageing Dev.,
99: 257–271, 1997.
57. Larson, P. S., de las Morenas, A., Cupples, L. A., Huang, K., and Rosenberg, C. L.
Genetically abnormal clones in histologically normal breast tissue. Am. J. Pathol.,
152: 1591–1598, 1998.
6518
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.
Clonal Selection versus Genetic Instability as the Driving Force
in Neoplastic Transformation
Ming Chow and Harry Rubin
Cancer Res 2000;60:6510-6518.
Updated version
Cited articles
Citing articles
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/60/22/6510
This article cites 50 articles, 32 of which you can access for free at:
http://cancerres.aacrjournals.org/content/60/22/6510.full#ref-list-1
This article has been cited by 10 HighWire-hosted articles. Access the articles at:
http://cancerres.aacrjournals.org/content/60/22/6510.full#related-urls
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 2000 American Association for Cancer
Research.