Feb 19, 1988 - Genetic Tagging of Tumor Cells with Retrovirus Vectors: Clonal. Analysis of Tumor Growth and Metastasis In Vivo. BOZENA KORCZAK, IAN B.
MOLECULAR AND CELLULAR BIOLOGY, Aug. 1988, p. 3143-3149 0270-7306/88/083143-07$02.00/0 Copyright © 1988, American Society for Microbiology
Vol. 8, No. 8
Genetic Tagging of Tumor Cells with Retrovirus Vectors: Clonal Analysis of Tumor Growth and Metastasis In Vivo BOZENA KORCZAK, IAN B. ROBSON, CATHY LAMARCHE, ALAN BERNSTEIN, AND ROBERT S. KERBEL* Divisions of Cancer and Cell Biology and Molecular and Developmental Biology, Mount Sinai Hospital Research Institute, 600 University Avenue, and Department of Medical Genetics, University of Toronto, Toronto, Ontario, Canada M5G IX5 Received 19 February 1988/Accepted 3 May 1988
Retrovirus vector infection was used to introduce large numbers of unique genetic markers into tumor cell populations for the purpose of analyzing comparative changes in the clonal composition of metastatic versus that of nonmetastatic tumors during their progressive growth in vivo. The cell lines used were SP1, a nonmetastatic, aneuploid mouse mammary adenocarcinoma, and SP1HU9L, a metastatic variant of SP1. Cells were infected with AeApMoTN, a replication-defective retrovirus vector which possesses the dominant selectable neo gene and crippled long terminal repeats. G418' colonies were obtained at a frequency of 4 x 10-3. Southern blot analysis of a number of clones provided evidence of random and heritable integration of one or two copies of the proviral DNA. Clonal evolution of primary tumor growth and the nature of lineage relationships among spontaneous metastases and primary tumors were analyzed by subcutaneously injecting 105 cells from a pooled mixture of 3.6 x 102 G418' SP1HU9L or 104 G418r SPl colonies into syngeneic CBA/ J mice. The most striking finding was the relative clonal homogeneity of advanced primary tumors; they invariably consisted of a small number (less than 10) of distinct clones despite the fact that hundreds or thousands of uniquely marked clones had been injected. In the case of the metastatic SPlHU9L cells, the nature of these "dominant" clones varied from one tumor to another. Analysis of a number of lung metastases revealed that a proportion of them were derived from dominant primary tumor clones and were composed of one, and sometimes two, distinct progenitors. In some animals, all the lung metastases were derived from a common progenitor clone, whereas in others, each metastatic nodule had a different progenitor. The results show the following. (i) Retrovirus vector infection can be used to introduce large numbers of unique and stable clonal markers into tumor cell populations. (ii) The progeny of a very limited number of clones dominate in advanced primary tumors. (iii) Mammary carcinoma metastases are of mono- or biclonal origin. The significance of the results is discussed.
There is now widespread acceptance that the growth of most solid tumors represents a complex microevolutionary process leading to the emergence and expansion of clonal subpopulations of metastatic tumor cells (10, 20, 21). Tumor progression appears to be driven primarily by two forces: genetic mutation and environmental selection pressures (21). Together, these forces lead to the development of a genotypically and phenotypically heterogeneous tumor cell mass, with the possible co-existence of multiple subpopulations of cells within a tumor. Analysis of such a dynamic evolutionary process involving the emergence and decline of successive new clonal subpopulations would be considerably aided by techniques in which individual cells can be tagged by a unique marker so that their fates can be traced. This has been attempted through the use of endogenous or radiation-induced cytogenetic (chromosomal) markers (12), drug resistance genetic markers (15, 23), or X-linked isoenzyme mosaicisms (22). Such markers have been exploited, for example, to demonstrate the probable monoclonal origin of most primary tumors, as well as artificial or spontaneous metastases in some tumor systems (12, 22). Unfortunately, their usefulness to study malignant tumor progression is limited by the tediousness and difficulty of the analyses involved and by the small number (usually two to five) of uniquely marked cell populations that can be effectively isolated or analyzed. Thus, *
Corresponding author. 3143
only a very minor fraction of the component subpopulations of a tumor can be marked by these methods. An approach which circumvents these problems is the use of transferred (or endogenous) DNA markers. Because insertions of the transferred DNA sequences occur at many possible sites in the genomes of recipient cells, transfection of foreign DNA can be used to generate large numbers of uniquely marked cell clones (13, 27, 28). Therefore, by pooling large numbers of individual transfectants a cell population with a large number of unique genetic markers can be obtained. We have used this approach, in previous studies, to genetically tag the nonmetastatic mouse mammary adenocarcinoma cell line SP1, in which one metastatic clone emerged probably because of the exposure to CaPO4. When injected subcutaneously into syngeneic mice, this cell population heterogenous in its metastatic properties led to the overgrowth of the primary tumors by the progeny of one particular clone in different animals. Furthermore, lung metastases which arose in these animals were derived from this same clone. These observations suggested that primary tumors can become overgrown in a nonrandom manner by genotypically distinct subpopulations of tumor cells having metastatic ability (C. Waghorne, M. Thomas, A. Lagarde, R. S. Kerbel, and M. L. Breitman, submitted for publication). These findings prompted us to evaluate clonal dominance characteristics of primary tumors, in which a large proportion of the tumor cells is homogenous in metastatic properties. Comparison of the fate of individual clones derived
MOL. CELL. BIOL.
KORCZAK ET AL.
3144
E co RV Saci Kpn I
Cbla
A lOObp
_-
B Pvull E co RV
E-co RV
Xbal
Seti
It
i
184 bp deletin tie
Smal I
179 bp deletion Ap 100bp
FIG. 1. Schematic diagram of the structure of retrovirus vector AeApMoTN (A) and its 3' LTR (B). (A) Boxes drawn with thick lines represent the 5' and 3' LTRs. EcoRV, Sacl, and KpnI sites are marked. The box drawn with thin lines, which is labeled tk, represents the 237-base-pair herpes simplex virus thymidine kinase gene promoter. The adjacent box, which is drawn with thin lines and labeled neo, represents the 1.5-kilobase-pair bacterial neo gene, which confers resistance to the antibiotic G418 on mammalian cells. The thin lines connecting the LTRs with the tk-neo fusion gene represent sequences derived from pFVXM (14), the Moloney murine leukemia virus vector backbone. (B) The 3' LTR of AeApMoTN has two deletions. Ae is a 184-base-pair deletion in the U3 region encompassing the retrovirus enhancer sequences (17) and including the PvuII and EcoRl sites. Ap is a 179-base-pair deletion encompassing the retrovirus promoter from the XbaI site to the SmaI site and including those sites and the SstI site.
from nonmetastatic or metastatic cell populations might provide insights into the mechanism which governs tumor progression and metastasis. We also wished to assess a new method of genetically tagging tumor cell populations that involves retrovirus vectors. The approach taken to tag tumor cells by retroviral-vector-mediated gene transfer involves the use of helper-virus-free retrovirus vectors which can infect animal cells in culture with very high efficiencies (up to 100%), resulting in the random integration of a limited number (usually one or two) of proviruses into the recipient cell genome. Such an approach has been used successfully to monitor individual hematopoietic stem cells as they repopulate the blood-forming system of genetically anemic (4) or lethally irradiated (16) mice. It has also been used to study cell lineages in the early mouse embryo and in the nervous system (for a review, see reference 24). Here, we describe the results of experiments in which we have molecularly analyzed the clonal composition of primary tumors as well as the lineage relationships of lung metastases to primary tumors obtained at late stages of tumor growth by similar methods. MATERIALS AND METHODS Cell lines and animals. The SP1 cell line originated as a spontaneous intraductal adenocarcinoma in an 18-month-old retired female breeder of the inbred mouse strain CBA/J. The characteristics of this tumor have been described elsewhere (3). After subcutaneous inoculation, it manifests a completely nonmetastatic phenotype (11, 13). The SP1HU9L subline, a tumorigenic and highly metastatic
variant of SP1, was derived by in vitro treatment of SP1 cells with hydroxyurea (11). Both cell lines were successfully infected with a retrovirus vector that contains the dominant selectable neo gene; G418-resistant (G418r) clones were subsequently isolated. A total of 105 cloned or uncloned (described below) G418r cells, designated SP1-neo or SP1HU9L-neo, were injected subcutaneously into CBA/J mice. The animals were sacrificed 6 to 8 weeks later, when they were moribund, and individual primary tumors (usually 2 to 3 cm in diameter) were excised and frozen in liquid nitrogen. Individual macroscopic lung metastases were also isolated and subcultured in tissue culture as separate cell lines. The mice used for these experiments were 8- to 12-week-old CBA/J females which were purchased from Jackson Laboratory, Bar Harbor, Maine, and maintained in the Animal Care Facility of the Mount Sinai Hospital Research Institute. SP1 mammary carcinoma cells were previously MAP tested for the presence of 13 different viruses and were found to be virus free (11). They were also tested periodically for mycoplasma, and only mycoplasma-free cells were used. Retrovirus vector. The defective Moloney murine leukemia virus shown in Fig. 1A was constructed as follows. The wild-type 3' long terminal repeat (LTR) was removed from the plasmid pFVXM (14) and replaced by a defective LTR, shown in Fig. 1B, derived from plasmid ApMLV-C/R/B (17), which has a 184-base-pair deletion encompassing the retroviral enhancer sequence (Ae) and an additional 174-bp deletion, from XbaI to SmaI, encompassing the retroviral promoter sequence (Ap). The 2.9-kilobase (kb) EcoRI/ClaI
VOL. 8, 1988
GENETIC TAGGING OF TUMOR CELLS WITH RETROVIRUS VECTORS
fragment of pFVXM, containing the 5' LTR, was ligated to the 1.8-kb ClaIlEcoRI fragment of ApMLV-C/R/B, containing the 3' LTR with both Ae and Ap deletions, and to EcoRI-digested pBR322. This construct was digested with BglI, and the 1.7-kb BamHI fragment containing the bacterial neo gene under the control of the herpes simplex virus
thymidine kinase gene promoter (D. Huszar, unpublished data) was inserted. This plasmid, called AeApMoTN, was transfected into psi2 helper cells (18) by calcium phosphate precipitation (31). Individual clones resistant to G418 (200 ,ug/ml) were isolated, and clones producing virus at the highest titer (approximately 105 CFU/ml) were used as a source of helper-free virus stock.
Retrovirus-mediated gene transfer. Cells were infected with helper-free stocks of the AeApMoTN retrovirus vector, described above, as follows. In brief, 60-mm dishes were seeded with 5 x 105 cells and, 24 h later, the medium was supplemented with 1 ml of a retrovirus suspension containing 105 viral particles and 4 ,ug of Polybrene. Cells were subcultured into G418-containing (400 ,wg/ml) selective media 48 h later, and a pooled population or individual clones of drug-resistant (i.e., G418') cells were harvested 14 to 18 days later. To determine whether the stocks of AeApMoTN were indeed helper virus free, cell culture fluids from SP1 cells infected with the vector were used to infect Rat-2 cells. We did not detect G418' Rat-2 cells, indicating that the infected cells did not produce infectious viral particles. Isolation of genomic DNA and Southern blot analysis. High-molecular-weight DNA was extracted from frozen primary tumors as previously described (31) and from cultured cells by methods described previously (2). DNA was digested with 4 to 6 units of restriction enzyme per ,ug of DNA and separated by electrophoresis. Southern transfer (25) to Bio Trace nitrocellulose was followed by prehybridization and hybridization by the method supplied by the manufacturer. DNA was hybridized to a randomly primed (8) 0.9-kb EcoRI-PstI neo fragment isolated from the plasmid pSV2neo (26). The labeled probe had a specific activity of 2 x 108 to 8 x 108 cpm/,g of DNA. RESULTS
Retrovirus vector AeApMoTN as a unique tumor cell marker. To ensure that no infectious vector could spread through our tumor population, we made use of a crippled vector that includes deletions in both promoter and enhancer sequences in the 3' LTR. Because of these deletions, this retrovirus vector, shown in Fig. 1A, is capable of only one round of infection. Virion RNA transcribed from this vector will contain only the defective U3 sequences (Fig. 1B) which will be copied into both the 5' and 3' LTRs of the provirus generated by infection with this virion. Thus, the provirus ultimately formed will lack enhancer and promoter elements in both 5' and 3' LTRs and will therefore be unable to direct expression of genes adjacent to the integration site and will not be expressed as virion RNA. There is theoretical difficulty with replication of this type of construct; initial strongstop DNA may have trouble "jumping" to the 3' R region or extending into U3 after the jump, since the 5' half of the 3' R has been deleted. Infection of target cells with these vectors, which we call clipped wing and others have called selfinactivating (32), results in a population of cells that contain integrated proviruses but no replicating vectors. Nonmetastatic SP1 cells and the metastatic variant SP1HU9L cells were infected with the retrovirus vector
A
1
2
3
4
B
5
6
1
2
C
3
4
1
2
3145
D
3
4
1
2
3
FIG. 2. Southern blot analysis of proviral DNA in mice (A to D) injected with individual SP1HU9L--neo clones. Genomic DNA (10 jig) of four SP1HU9L-neo clones (lane 1), cells growing in tissue culture for 6 to 8 weeks (lane 2), and metastases arising after injection with individual clones (lanes 3 to 6), was digested with BamHI, separated on a 0.6% agarose gel, transferred to Bio Trace nitrocellulose, and hybridized with the neo probe.
AeApMoTN described above. The efficiency of infection was found to be 4 x 10-3, about 2 orders of magnitude lower when compared with those of some nontumorigenic cell populations that have been used as recipients but nevertheless significantly higher in comparison with those of other conventional gene transfer techniques that involve plasmid vectors. We isolated and pooled 104 clonotypically marked SP1-neo cells and 3.6 x 102 SP1HU9L-neo cells. SP1 mammary adenocarcinoma cells are highly aneuploid; they have over 70 chromosomes, some with distinct translocations (11). Because of the possibility of neoplastic cell genomic instability (e.g., chromosome loss, rearrangements, duplications, etc.), it was necessary for our studies to establish first the stability of proviral sequences inserted into aneuploid SP1HU9L cells. Individual clones isolated from retrovirus vector-infected SPlHU9L-neo cells (Fig. 2A to D, lanes 1) were injected subcutaneously into CBAIJ mice. Southern analysis was performed on the same clones growing in tissue culture for the same length of time as in mice (Fig. 2A to D, lanes 2) and on distinct lung metastases (Fig. 2A, lanes 3 to 6; B, lanes 3 and 4; C, lanes 3 and 4; and D, lane 3). Genomic DNA was digested with BamHI, a restriction enzyme which does not cut within the provirus; thus, a unique fragment was generated for each insertion when probed with the radiolabeled neo probe. The size of this fragment reflects its integration site in the host genome and is specific for each infected cell because of the fact that viral integration is an apparently random process (1). Our results showed that cell-virus junctions were preserved during extended growth in tissue culture as well as in vivo. We also found that these clones retained their G418 resistance phenotype; this suggests that the neo gene did not undergo any obvious genetic or epigenetic changes during in vivo or in vitro growth. To verify further that the neo provirus in these cells had not sustained major rearrangements or deletions, DNA isolated from primary tumors which were injected with a polyclonal mixture (360 clones) of SPlHU9L-neo cells (Fig. 3) was digested with KpnI, a restriction enzyme which has two recognition sites within the provirus. Thus, a 2.3-kb neo-specific fragment should be detected in all cells. As shown in Fig. 3, a 2.3-kb band was present in DNA from the primary tumors, although it is possible that a small fraction (10% or less) of rearranged provirus was present that was not detected by the Southern blot technique. From these experiments, we concluded that the majority of proviral DNA was indeed inserted into SPlHU9L-neo cells in a stable, heritable manner. Analysis of selection and clonal evolution during tumor progression in vivo. (i) Selection events detected in nonmetastatic and metastatic primary tumors. The results described
3146
MOL. CELL. BIOL.
KORCZAK ET AL.
a
2.3kb-
#
A B C D E F G FIG. 3. Hybridization of the neo probe to DNA isolated from primary tumors. Genomic DNA was isolated from primary tumors after CBA/J mice had been injected with a polyclonal mixture of tagged SPlHU9L-neo cells. DNA (10 jig) was digested with KpnI, separated on a 0.6% agarose gel, transferred to Bio Trace nitrocellulose, and hybridized with the neo probe. The 2.3-kb fragment corresponds to the size of the KpnI band of zeApMoTN probed with labeled neo. Each lane contains DNA from individual animals A to G; the respective hybridization profiles after BamHI digestion are shown in Fig. 4.
above indicated that it would be possible to use retroviral infection as a method of generating large numbers of tumor cells with unique genetic markers. In an initial set of experiments, we analyzed the composition of primary tumors derived after injection with nonmetastatic SP1-neo cells, representing about 104 distinct clones (Fig. 4), and metastatic SPlHU9L-neo cells, representing 3.6 x 102 clones (Fig. 5). A total of four mice, injected with SP1-neo cells, were examined for the AeApMoTN provirus in their primary tumors by DNA digestion with BamHI to generate a unique cell-virus junction fragment. As shown in Fig. 4, the SP1-neo cell inoculum consisted of a complicated, diffuse pattern of faint hybridizing bands reflecting the large number of distinctly marked cells; in contrast, a small number of bands was detected in the DNA from primary tumors. Southern blot analysis of the SPlHU9L-neo cell inoculum and SP1HU9L-neo primary tumors revealed a similar pattern (Fig. 5A to F, lanes 1). Each primary tumor was ultimately dominated by only a few clones, but we cannot speculate on the exact number of these progenitors since the Southern kb 23.1
-
.4*
9.6-
6.4-
0
FIG. 4. Southern blot analysis of primary tumors from four animals injected with nonmetastatic SP1-neo. Genomic DNA (10 ,ug) from inoculum (mixed population of 104 SPl-neo clones) (lane i), noninfected SP1 cells (lane 0), and primary tumors derived from SP1-neo inoculum in four mice (lanes 1, 2, 3, and 4) was digested with BamHl, separated on a 0.6% agarose gel, transferred to Bio Trace nitrocellulose, and probed with the neo probe.
blot analysis of genomic DNA was not sensitive enough to detect cells which contribute less than 5 to 10% to the tumor mass. At the time of analysis, every SPlHU9L-neo tumor contained a different set of marked cell clones, but it is possible that some of these clones were present in more than one tumor. Nevertheless, these observations argue for a random selection process from a large subset of potentially tumorigenic cells. (ii) Selection from primary tumors to metastases. In order to approach the problem of origin of metastases and their relationship to primary tumors, we used the same type of analysis as described above for primary tumors. The results from seven mice injected with SPlHU9L-neo cells are shown in Fig. 5. Several conclusions can be drawn from the Southern blot analysis. First, we noted that only one to three junction fragments were present in cells derived from spontaneous lung metastases; this pattern of hybridization is clearly simpler than that seen with most of the primary tumors. This finding suggested progressive selection of tumor cell progenitors during development from primary tumor to metastasis. Second, the identical 11-kb fragment present in different nodules from animal A, 8-kb fragment in animal B, and 6.2-kb fragment in animal C after digestion with BamHI indicated that, within these individual animals, metastases arose from the same clonal progenitor. In contrast, Southern blot analysis of metastases from animals D and E showed that each nodule contained unique junction fragments; this suggests that these metastases were derived from different progenitors. Clonal origin of metastases. To determine whether metastases represented a clonal outgrowth from the primary tumor, DNA from metastatic nodules was digested with the restriction enzyme BamHI. In four of seven animals analyzed, only one band was detected (Fig. SB, C, D, and E) in the metastases. This finding is consistent with previous results demonstrating the presumed clonal origin of most metastases (12, 22, 23, 29). However, we also noticed that metastases Fl and Gl contained two and three strongly hybridizing bands, respectively, after BamHI digestion. We decided to examine whether this pattern was due to multiple integration events of vector sequences into the same cell or whether these nodules contained more than one clone. For this purpose, cells from nodules Fl and GI were cloned in culture and DNA from 10 individual clones was examined. DNA from nodule Gl contained fragments of 27, 13.5, and 3.8 kb, and in clones derived from this nodule we detected 27- and 13.5-kb fragments (lane a) or a 3.8-kb fragment (lane b) only. An interpretation consistent with this finding is that clone a has two integrated copies of AeApMoTN, whereas clone b has one; thus, the metastasis in animal G was probably derived from two different progenitor cells (i.e., was biclonal). In the case of nodule Fl, which contained 17and 9-kb fragments, all analyzed clones had the same hybridization pattern as shown in Fig. 5. We did not find any clones in vitro with the lower (9-kb) band which might represent a sampling artifact. From this analysis, we conclude that this nodule consists of the progeny of two different cells and is biclonal, as is nodule G1. We also noticed faint bands accompanying one prominent band in the cases of animals A and B. Such a difference in the intensity of hybridization indicates that these nodules were generated from two cells, one of which had overgrown the other. DISCUSSION In these studies, we have clonally analyzed the in vivo progression of a mouse mammary adenocarcinoma by genet-
VOL. 1988 TAGGING OF TUMOR CELLS WITH RETROVIRUS VECTORS VOL. 1988 ~~GENETIC 8, 8,
A
B
C
kb
E
D
3147
G
F
- -I II I -I
II
23.1-
0*
0 0
go so*0o
I
9.6-
64-
s.
.,.
I0
P
2
1
3
4
P
5
234A
1
P
P12
1233
P
12
P
1
P
a
1
b
a
Analysis of cell-virus junctions in malignant tumors derived in seven mice (A to G) after injection with SPlHU9L-neo. Genomic from inoculum (mixed population of 3.6 x 102 SPlHU9L-neo cells) (lane i), from uninfected SP1HU9L cells (lane 0), from primary tumors (lanes P) from individual animals A to G and from individual metastases derived from these tumors (lanes 1 to 5) and from clones from metastases (lanes a and b) was digested with BamHl, separated on a 0.6% agarose gel, transferred to Bio Trace nitrocellular, and probed with the neo probe. FIG.
5.
DNA (10
~Lg)
ically tagging
the
Thus,
our
vector.
tumor
cells
in
vitro
results extend the
with
use
a
retrovirus
of retroviruses
as
genetic markers to analyze cell lineage relationships and development during hematopoiesis, in the early mouse embryo, and in the nervous system (4, 5, 16; for a review, see reference 24). Our data indicate that the proviral DNA integration events are heritable and stabl e in the tumor cell population we analyzed, despite the inherent genetic instability of neoplastic cells (10, 20). Moreover, the retroviral construct we have used for these studies, AeApMoTN, is particularly appropriate to use as a genetic marker for tumor cells because of its crippled LTRs;' thus, it cannot activate host cell genes flanking the proviral integration site, nor can it be rescued by an infectious murine leukemia virus that might be present in many tumor cell populations. Analysis of tumors tagged with a retrovirus vector demonstrated an apparent and remarkable progression of primary tumors towards clonal homogeneity. Thus, tum-ors which developed from mixtures of large numbers of uniquely marked
SPl-neo
clones
or
SPlHU9L-neo
ultimately dominated by only months of growth in CBAIJ mice.
were
metastases
were
either clonal
or
a
metastatic clones
few clones after 2
Moreover,
biclonal.
the
lung
Miller and col-
leagues (19) also observed clonal dominance in primary tumors developed in animals injected with various ratios of two differentially drug-marked tumor cell lines. The basis for this type of phenomenon is unknown and is currently under investigation in our laboratory. Our results may have implications for theories concerning the clonal origin of neoplasms (6, 7, 9, 30). Studies using a variety of genetic markers have shown that most types of tumors are monoclonal at the time of analysis. From these observations, it has been assumed that this reflects their
origin from a single transformed cell (6, 7, 9, 30). However, in light of the studies presented here, it is also possible that tumors which originated from a large number of different progenitors can ultimately achieve clonal homogeneity during tumor development. It should be emphasized that in the results reported here a different set of clones appeared to dominate in every animal analyzed even though these ani-
mals
were
injected
with the
same
population of marked
cells.
This observation suggests that the apparent relative homo-
geneity
that
we
observed within
a
single animal
was
not due
to the dominant characteristic of some clones at the time of
injection but rather to the fact that that they mnay have acquired these properties during tumor development. We are currently examining the stability of these putative dominant clones after reinjection. We were also concerned that dominance by a single clone during tumor development in vivo may have been a consequence of the tagging technique using a
retrovirus vector. However, several considerations make
this
possibility extremely unlikely. First, the ability of
retrovirus vector used in these studies to activate
transcription of cellular capped by deletions in
or
the
enhance
(e.g., oncogenes) was handitranscription regulatory sequences. Second, the use of a helper-free, replicationdefective vector eliminated the possibility of virus spread and the appearance of new integration sites, which might change the scenario of tumor development. Indeed, all of the
analyzed
genes
retrovirat
only one or two integrated proprimary tumor analyzed conof clones argues against selection for
tumors contained
viruses. The fact that each
sisted of a
a
different set
provirus integration site. We
also
obtained
evidence
that
the
metastases
which
developed from SP1HU9L-neo tumors were derived from at most only two different progenitor cells. The biclonality of some metastatic lesions might reflect the fact that it is possible for metastatic cells to escape from the primary tumor and colonize the lungs as a heterogeneous clump of cells. The presence of two populations of cells in an individual metastasis raises questions about their interaction, their eventual collaboration during the process of metastasis, and the ability of each of these founder populations to metastasize on
its
own.
Examination of the clonal
cases
composition of different
metas-
from the
same
animals revealed that in
individual nodules
were
populated by
tases recovered
the
same
some
clone.
This observation is consistent with the idea that metastases may
derive
possible
from
other
metastases.
that the individual
However,
lung nodules
were
it
is
also
formed inde-
3148
KORCZAK ET AL.
pendently from descendants of the same clone that independently reached the lung. At present, we cannot distinguish between these possibilities. Comparison of the clonal composition of metastases with the composition of their respective parental tumors led to an unexpected observation. We noticed on several occasions that clones which were dominant in the primary tumors did not necessarily contribute to the formation of metastases, and vice versa; i.e., clones which populated metastases were sometimes undetectable at the primary site of inoculation. This raises the possibility that, within a metastatically homogeneous population of cells, dominant growth properties do not necessarily correlate with metastatic potential. It is important to stress that this conclusion may apply only to a situation in which virtually all cells in the initial inoculum have metastatic potential. Thus, quite different conclusions were drawn from experiments in which the initial tagged population of SP1 cells (in this case tagged by random plasmid DNA integrations) contained only one or a few metastatic clones (C. Waghorne et al., submitted); in those experiments, it was demonstrated that metastatic clones had a definite growth advantage over nonmetastatic cells within the primary tumor and eventually came to dominate the primary tumor mass. Hence, the present study demonstrates how different tumor subpopulation compositions can be obtained, depending on the complexity of the metastatic potential of the initial population of injected cells. Our results, showing clonal dominance of primary tumors, have an intriguing biological precedent in the studies of the clonal origins of cells within the hematopoietic system (4, 5, 16). Several groups have studied the clonal composition of various hematopoietic cell compartments (e.g., thymus, spleen, and marrow) after injection of large numbers of bone marrow stem cells uniquely tagged with a retrovirus vector into either genetically anemic (4) or lethally irradiated (16) mice. In both cases, evidence was obtained that a limited number of stem cells contributed to the hematopoietic system at any given time. Taken together, these results raise the possibility that clonal evolution and dominance during tumor growth may not necessarily require or always depend upon tumor cell genetic instability (20) but may instead be a more general manifestation of rapid and extensive cell growth and renewal, whether normal or neoplastic. In summary, our results demonstrate the considerable potential for analyzing cell lineage aspects of tumor growth, progression, and metastasis through the use of retrovirus vector-tagged tumor cell populations. The results presented here raise several questions about the mechanisms which underlie the process of selection during tumor growth and metastasis. Our knowledge of the different sequential events which accompany tumor development should now allow us to study in greater detail and with greater insight the properties of the selected clones and the external and internal factors which support their expansion. ACKNOWLEDGMENTS We thank Martin Breitman for critical review of the manuscript. Lynda Woodcock and Nicole Petrin provided excellent secretarial support. This work was supported by grants from the National Cancer Institute of Canada (to R.S.K. and A.B.) and by Public Health Service grant CA 41233 from the National Institutes of Health to R.S.K. R.S.K. is a Research Associate of the National Cancer Institute of Canada.
MOL. CELL. BIOL. LITERATURE CITED 1. Bernstein, A., S. Berger, D. Huszar, and J. Dick. 1985. Gene transfer with retrovirus vectors, p. 235-261. In J. K. Setlow and A. Hollender (ed.), Genetic engineering, vol. 7. Academic Press, Inc. (London), Ltd., London. 2. Breitman, M. L., L.-C. Tsui, M. Buchwald, and L. Siminovitch. 1982. Introduction and recovery of a selectable bacterial gene from the genome of mammalian cells. Mol. Cell. Biol. 2:966976. 3. Carlow, D. A., R. S. Kerbel, T. J. Feltis, and B. E. Elliott. 1985. Enhanced expression of class I major histocompatibility complex gene (Dk) products on immunogenic variants of a spontaneous murine carcinoma. J. Natl. Cancer Inst. 75:291-301. 4. Dick, J. E., M. C. Magli, D. Huszar, R. A. Phillips, and A. Bernstein. 1985. Introduction of a selectable gene into primitive stem cells capable of long-term reconstitution of the hemopoietic system of W/WV mice. Cell 42:71-79. 5. Dick, J. E., M. C. Magli, R. A. Phillips, and A. Bernstein. 1986. Genetic manipulation of hematopoietic stem cells with retrovirus vectors. Trends Genet. 2:165-170. 6. Esumi, M., T. Aritaka, M. Arii, K. Suzuki, K. Tanikawa, H. Mizuo, T. Mima, and T. Shikata. 1986. Clonal origin of human hepatoma determined by integration of hepatitis B virus DNA. Cancer Res. 46:5767-5771. 7. Fearon, E. R., S. R. Hamilton, and B. Vogelstein. 1987. Clonal analysis of human colorectal tumors. Science 238:193-196. 8. Feinberg, A. P., and B. Vogelstein. 1983. A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal. Biochem. 132:6-13. 9. Fialkow, S. P. J. 1976. Clonal origin of human tumors. Biochim. Biophys. Acta 458:283-321. 10. Foulds, L. 1954. Tumor progression. Cancer Res. 17:337-339. 11. Frost, P., R. S. Kerbel, B. Hunt, S. Man, and S. Pathak. 1987. Selection of metastatic variants with identifiable karyotypic changes from a nonmetastatic murine tumor after treatment with 2'-deoxy-5-azacytidine or hydroxyurea: implications for the mechanisms of tumor progression. Cancer Res. 47:26902695. 12. Hu, F., R.-Y. Wang, and T. C. Hsu. 1987. Clonal origin of metastasis in B16 murine melanoma: a cytogenetic study. J. Natl. Cancer Inst. 78:155-163. 13. Kerbel, R. S., C. Waghorne, S. Man, B. Elliott, and M. L. Breitman. 1987. Alteration of the tumorigenic and metastatic properties of neoplastic cells is associated with the process of calcium phosphate-mediated DNA transfection. Proc. Natl. Acad. Sci. USA 84:1263-1267. 14. Kriegler, M., C. F. Perez, C. Hardy, and M. Botchan. 1984. Transformation mediated by the SV40 T antigens: separation of the overlapping SV40 early genes with a retroviral vector. Cell 38:483-491. 15. Lagarde, A. E., T. P. Donaghue, J. W. Dennis, and R. S. Kerbel. 1983. Genotypic and phenotypic evolution of a murine tumor during its progression in vivo toward metastasis. J. Natl. Cancer Inst. 71:183-191. 16. Lemischka, I. R., D. H. Raulet, and R. C. Mulligan. 1986. Developmental potential and dynamic behavior of hematopoietic stem cells. Cell 45:917-927. 17. Linney, E., B. Davis, J. Overhauser, E. Chao, and H. Fan. 1984. Non-function of a Moloney murine leukaemia virus regulatory sequence in F9 embryonal carcinoma cells. Nature (London)
308:470-472. 18. Mann, R., R. C. Mulligan, and D. Baltimore. 1983. Construction of a retrovirus packaging mutant and its use to produce helperfree defective retrovirus. Cell 38:153-159. 19. Miller, B. E., F. R. Miller, D. J. Wilburn, and G. H. Heppner. 1987. Analysis of tumor cell composition in tumors composed of paired mixtures of mammary tumor cell lines. Br. J. Cancer 56: 561-569. 20. Nowell, P. C. 1976. The clonal evolution of tumor cell populations. Science 194:23-28. 21. Nowell, P. C. 1986. Mechanisms of tumor progression. Cancer Res. 46:2203-2207. 22. Ootsuyama, A., K. Tanaka, and H. Tanooka. 1987. Evidence by
VOL. 8, 1988
23.
24. 25.
26.
27.
GENETIC TAGGING OF TUMOR CELLS WITH RETROVIRUS VECTORS
cellular mosaicism for monoclonal metastasis of spontaneous mouse mammary tumors. J. Nati. Cancer Inst. 78:1223-1227. Poste, G., J. Tzeng, J. Doll, R. Greig, D. Rieman, and I. Zeidman. 1982. Evolution of tumor cell heterogeneity during progressive growth of individual lung metastases. Proc. Natl. Acad. Sci. USA 79:6574-6578. Price, J. 1987. Retroviruses and the study of cell lineage. Development 101:409-419. Southern, E. M. 1975. Detection of specific sequences among DNA fragments separated by gel electrophoresis. J. Mol. Biol. 98:503-517. Southern, P. J., and P. Berg. 1982. Transformation of mammalian cells to antibiotic resistance with a bacterial gene under control of the SV40 early region promoter. J. Mol. Appl. Genet. 1:327-341. Talmadge, C., Y. Tanio, A. Meeker, J. Talmadge, and B. Zbar. 1987. Tumor cells transfected with the neomycin resistance gene (neo) contain unique genetic markers useful for identification of tumor recurrence and metastasis. Invasion Metastasis 7: 197-207.
3149
28. Talmadge, J. E., S. R. Wolman, and I. J. Fidler. 1982. Evidence for the clonal origin of spontaneous metastases. Science 217: 361-363. 29. Talmadge, J. E., and B. Zbar. 1987. Clonality of pulmonary metastases from the bladder 6 subline of the B16 melanoma studied by Southern hybridization. J. Natl. Cancer Inst. 78:315320. 30. Vogelstein, B., E. R. Fearon, S. R. Hamilton, A. C. Preisinger, H. F. Willard, A. M. Michelson, A. D. Riggs, and S. H. Orkin. 1987. Clonal analysis using recombinant DNA probes from the X-chromosome. Cancer Res. 47:4806-4813. 31. Wigler, M., R. Sweet, G. K. Sim, B. Wold, A. Pellicer, E. Lacy, T. Maniatis, S. Silverstein, and R. Axel. 1979. Transformation of mammalian cells with genes from procaryotes and eucaryotes. Cell 16:777-785. 32. Yu, S. F., T. Rueden, P. W. Kantoff, C. Garber, M. Seiberg, U. Ruether, W. F. Anderson, E. F. Wagner, and E. Gilboa. 1986. Self-inactivating retroviral vectors designed for transfer of whole genes into mammalian cells. Proc. Natl. Acad. Sci. USA 83:3194-3198.
