Characterization of homologous DNA recombination activity in normal and immortal
mammalian cells
Characterization of homologous DNA recombination activity in normal and immortal mammalian cells
Bhaskar
Thyagarajan
,
Monica
McCormick-Graham
,
Daniel P.
Romero
and
Colin
Campbell*
Department of Pharmacology, University of Minnesota Medical School, 435 Delaware
Street, SE,
Minneapolis
, MN 55455,
USA
Received May 3, 1996;
Revised and Accepted August 16, 1996
ABSTRACT
Homologous DNA recombination levels were measured in normal and spontaneously
immortalized murine and human fibroblasts, and in a number of primate and
murine established fibroblast cell lines. Immortal cell lines and tumor-derived clones homologously recombined extrachromosomal plasmid substrates at frequencies approximately 100-fold higher than did normal cells. To further explore the mechanism responsible for this phenotype, homologous recombination frequency was measured using nuclear extracts derived
from normal and immortalized murine and human fibroblasts. Extracts prepared
from immortal cells catalyzed high levels of homologous recombination, whereas very little recombination activity was detected in extracts prepared from normal fibroblasts. Similarly, only extracts derived from immortal cells contained strand-transferase activity as measured by the recently described pairing-on-membrane assay. Mixing experiments indicated that a
recombination enhancing factor or factors present in immortal cells, rather
than a recombination inhibitor in normal cells, was responsible for the
enhanced homologous recombination activity observed using extracts derived from
the former.
INTRODUCTION
Genomic instability is a hallmark of oncogenically transformed cells (
1
,
2
). Deletions, duplications, translocations and loss of ploidy, including hyperdiploidy, are common cytogenetic anomalies seen in tumor cells. It has long been suspected that these, and other genetic
alterations are likely to play a fundamental role in the oncogenic process. For
the most part, however, it has been difficult to rigorously differentiate
amongst those genome rearrangements that play a causative role in cancer, and
those events that are an unrelated consequence of it. (Exceptions include the specific translocations associated with certain lymphoid cancers;
2
.) It therefore remains uncertain to what extent enhanced genetic instability
may predispose a cell to cancer.
Recently, a direct connection between genetic instability and cancer has been
detected. Mutations in a number of genes that are homologous to yeast and
bacterial DNA mismatch repair genes have been associated with hereditary
nonpolyposis colorectal cancer in humans (
3
-
7
). Individuals heterozygous for such mutations are also more susceptible to
cancers of the ovary, endometrium and stomach (
8
). Molecular analysis using PCR has determined that tumor cells from these
patients contain thousands of mutations, in the form of microsatellite
instabilities (
9
-
12
). Kolodner and Alani (
13
) provided evidence that these mutations are also seen in sporadic tumors,
suggesting that deficiencies in DNA mismatch repair may be a common aspect of
human carcinogenesis. It has also been shown that cells with mutations in these
genes display a mutator phenotype
in vitro
(
14
), and are profoundly incapable of repairing mismatched DNA (
15
,
16
).
It therefore seems likely that, as a result of deficient DNA mismatch repair,
misincorporation of nucleotides during DNA replication results in the creation
of a host of mutations that eventually leads to cellular transformation.
However, it is conceivable that a deficient mismatch repair mechanism could also predispose cells to cancer through an entirely distinct mechanism. Experiments
conducted in bacteria (
17
,
18
) and yeast (
19
,
20
) indicate that inactivation of cellular mismatch repair mechanisms results in
greatly enhanced levels of recombination between partially identical DNA sequences (homeologous recombination). A similar observation has very recently been made in mammalian cells. While a
lack of complete sequence identity between recombination substrates greatly
reduces the frequency of gene targeting in wild-type mouse ES cells, gene targeting frequency is not similarly reduced in
ES cells that lack a functional
Msh2
mismatch repair gene (
21
). These observations raise the possibility that some of the chromosomal rearrangements seen in cancer cells may result from aberrant homologous recombination.
Consistent with this idea, it has previously been shown that homologous intra-plasmid recombination occurs at higher levels in human tumor-derived cells and virally transformed clones, than in normal diploid fibroblasts (
22
). However, it remains unclear whether this elevated homologous recombination activity contributes to the oncogenic process, or is instead a nonspecific consequence of it. In
order to provide such information, it will be necessary to gain greater insight
into molecular mechanisms responsible for the observed differences in homologous recombination activity seen in normal and immortal cells.
Toward this goal, biochemical and genetic homologous recombination assays were
used to characterize homologous recombination activity in normal and
immortalized mammalian somatic cells. We have determined that immortal cells
possess homologous recombination levels that are nearly 100-fold higher than those observed in normal cells. We present evidence suggesting that it is the presence of a positively acting, homologous recombination enhancing activity within immortal cells that is responsible for this distinction. Consistent with this interpretation, we used a recently described strand-transferase assay (
23
,
24
) to identify putative strand-transferase proteins that are present in protein extracts derived from
immortalized cells but that appear to be either absent or inactive in similar
extracts made from normal cells.
MATERIALS AND METHODS
Cells
HT1080 (human spontaneous fibrosarcoma;
25
) and CV-1 (immortal African green monkey kidney;
26
) cells were obtained from the ATCC. Murine 3T3 and 3T6 cells (immortal cells
derived from murine embryo fibroblasts;
12
) were the kind gift of Dr Li-Na Wei, University of Minnesota. COS-1 (SV40 transformed CV-1 derived;
28
) cells were provided by Dr Ping Law, University of Minnesota. Normal human
fibroblasts were the kind gift of Dr Robert O'Dea, University of Minnesota.
Murine embryo fibroblasts were derived from a FVB-N strain mouse (Jackson Labs) as described by Green and Todaro (
27
). Murine embryo fibroblast (MEF) cells were used between passage numbers 1 and
8. Within this range of passage number, we detected no discernible difference
in their growth rate, morphology, or homologous recombination activity. Spontaneously immortalized MEF cells (referred to as MEF-Im) were obtained by serially passaging the MEF cells until senescence,
and subsequently isolating and expanding foci. In our hands this process
required at least 12 serial passages of 1 in 3.
Homologous recombination assays
To measure extrachromosomal inter-plasmid HR in primate cells, the plasmids pSV2neoDL and pSV2neoDR (a kind
gift of Dr Raju Kucherlapati, Albert Einstein College of Medicine, see Fig.
1
) were co-transfected into recipient cells along with pRSVedl884, a plasmid encoding
the SV40-derived large T antigen (
28
), using the calcium phosphate co-precipitation protocol (
29
). Ten micrograms of each plasmid was used per 10 cm dish, and in all cases, the plasmids were supercoiled. Construction of plasmids
pSV2neoDL [248 base pair deletion within the 5' portion of the neomycin phosphotransferase (neo) gene] and pSV2neoDR
(283 base pair deletion within the 3' portion of the neo gene), each of which harbors a non-reverting, inactivating deletion mutation within the coding region
of the neomycin phosphotranferase gene from Tn5, has been described (
30
).
Following transfection, cells were incubated in growth media for 48 h prior to
the recovery of plasmid DNA (
31
). This low molecular weight DNA was restriction digested with an excess of
Dpn
I. This material was then used to transform electro-competent
Escherichia coli
(strain DH10B,
rec
A deficient). A portion of the transformation mix was plated on ampicillin-containing plates, while the remainder was plated on kanamycin-containing plates. The HR frequency was determined by the ratio of
double resistant (ampicillin and kanamycin) colonies to ampicillin resistant
colonies. [The plasmid pRSVedl884 (
32
) lacks an SV40 origin of replication, and will not replicate under the
conditions used in our assay, while both the recombination substrate plasmids
contain an SV40 origin of replication, and will replicate. As a consequence of
this, only pSV2neo-derived plasmids will survive
Dpn
I digestion and be measured in the bacterial transformation assay.]
To measure HR in murine cells, the plasmids pSV2neoDLPy and pSV2neoDRPy were co-transfected into recipient murine cells, and further processed as described above. Both of these plasmids,
which replicate autonomously in murine cells were created by introducing the early region of polyoma virus into the
Bam
HI sites of pSV2neoDL and pSV2neoDR. Control experiments indicate that wild-type and deletion-bearing plasmids replicate at essentially identical rates in both
normal and immortal cells.
For transfection-based assays involving both murine and human cells, an important control involved directly co-transforming recombination deficient
E.coli
with 0.5 [mu]g of each of the recombination substrates (either pSV2neoDR and pSV2neoDL,
or pSV2neoDLPy and pSV2neoDRPy). A portion of these bacteria were then plated
on ampicillin-containing agar plates, and the majority plated on plates containing
ampicillin and kanamycin. Under these conditions, we observed a very low level
of HR activity, carried out by the
E.coli
. As shown in Table
1
, the level of this activity is significantly lower than that seen in any of the
transfection-based experiments.
.
Extra chromosomal HR is elevated in immortal cells
Cell line
Kan resistant colonies
Amp resistant colonies
HR frequency
a
No of experiments
b
HT1080
111
164 000
68
19
COS-1
2105
213 300
987
3
CV-1
8
10 600
75
2
3T6
3
5 600
54
6
3T3
33
18 600
177
11
MEF-Im
15
90 000
17
11
HDF-1
5
469 300
1.1
19
MEF
25
12 334 000
0.2
12
E.coli
c
13
101 500 000
0.01
4
a
HR frequency is expressed as the number of kanamycin-resistant colonies per 100 000 ampicillin-resistant colonies.
b
An experiment refers to an independent transfection. The total number of
kanamycin- and ampicillin-resistant colonies from each set of these transfections was summed
and used to calculate the average HR frequency.
c
0.5 [mu]g of the recombination substrate plasmids were co-transformed into
E.coli
and the number of ampicillin- and kanamycin-resistant colonies determined.
Cell-free HR was determined by incubating plasmids that harbored non-overlapping deletions within the neo gene separately or together with nuclear protein extracts prepared from the
respective cell types (
33
). In some experiments double-stranded plasmids pSV2neoDL and pSV2neoDR were utilized, while in the
others, a single-stranded M13mp18 derivative containing the neoDR allele (ssneoDR) was used
in conjunction with pSV2neoDL linearized with the restriction endonuclease
Sal
I (see Fig.
1
, and text). Following a 30 min incubation at 37oC as described (
33
), the plasmid DNA (500 ng of each substrate) was extracted with phenol:
chloroform (1:1), ethanol precipitated, and used to transform
rec
A deficient
E.coli
(strain DH10B). DNA from the separate plasmid incubations was pooled prior to
transformation. The number of kanamycin (recombinant) and ampicillin (non-recombinant) resistant colonies were counted. The homologous recombination
frequency obtained from separate plasmid incubations was calculated, and subtracted from that obtained from co-incubation experiments. We directly co-transformed un-treated DNA recombination substrates into
E.coli
, and measured the HR frequency. As indicated in Table
2
, we did not obtain any kanamycin-resistant colonies under these conditions. We also performed an additional
control in which the plasmid pSV2neo (which harbors intact neomycin- and ampicillin-resistance markers) was incubated with a nuclear protein extract, and subsequently transformed into recombination-deficient
E.coli
. In these experiments, the ratio of kanamycin-resistance to ampicillin- resistance was 1.04 +- 0.16. This result indicates that there is no inherent
difficulty in recovering kanamycin-resistant bacteria following incubation of plasmid DNA with a nuclear
protein extract.
RESULTS
Homologous DNA recombination (HR) activity is elevated in immortal mammalian
somatic cells
It has been previously shown that homologous intra-plasmid recombination levels in immortalized human cells are significantly greater than those seen in diploid fibroblasts (
22
). To confirm and extend this observation, and gain additional insight into the
mechanism through which this process occurs, we measured HR levels in immortal
and normal cells derived from human, non-human primate and murine sources.
A co-transfection assay was used to measure the frequency with which two
plasmid molecules, transiently present within the nuclei of cultured cells,
homologously recombine. This assay is depicted schematically in Figure
1
, and is described in detail in the Materials and Methods. Due to the limited
replicative life-span of normal fibroblasts in culture, it is not feasible to use other,
more traditional methods to measure HR frequency that involve selection for stably transfected drug-resistant clones.
Transfected plasmid molecules are capable of replication within mammalian cells
in the presence of the SV40 large T antigen. Plasmid DNA recovered from transfected cells is subsequently treated with the
restriction endonuclease
Dpn
I, which will digest only the parental, unreplicated plasmid DNA (originally
isolated from an
E.coli
host that methylates adenine residues located within the sequence GATC). This
treatment was done to ensure that all plasmid DNA analyzed further has
replicated within the transfected cells. We performed a number of control
experiments (not shown) in which the SV40 large T antigen-encoding plasmid pRSVedl884 was omitted from the transfection mixture.
When low-molecular weight DNA was recovered from these cells, treated with
Dpn
I, and used to transform
E.coli
, no ampicillin- or kanamycin-resistant colonies were observed. This result proves that
Dpn
I digestion eliminates plasmids that have not replicated.
We have incorporated plasmid replication into our transfection protocol for two
reasons. First, this intracellular amplification enhances the yield of
plasmids, providing greater numbers of both recombinant, and non-recombinant molecules for subsequent analysis. In light of previous
results indicating that HR levels in normal cells are quite low (
22
), obtaining greater numbers of plasmids to analyze would provide HR frequency
values of greater accuracy. Second, we were concerned that, if a substantial
proportion of the DNA used to transfect the mammalian cell lines never entered
the nucleus of the recipient cell line, it would be difficult to accurately
compare HR frequencies amongst the various cell lines tested. By digesting with
Dpn
I, we have been able to ensure that all of the molecules used to generate
bacterial colonies, in the final step of the assay, were originally within the
recipient mammalian cell line.
Since we have used a large amount of the SV40 large T antigen encoding plasmid
(pRSVedl884), it is likely that the majority of plasmid DNA within these
transfected cells will have undergone at least one round of replication. While
we are not controlling for transfection efficiency,
per se
, by digesting the low molecular weight DNA with
Dpn
I we are providing reasonable assurance that the HR frequencies from different
cell lines can be accurately compared. Previous investigations have determined
that plasmid replication does not influence the frequency of extra-chromosomal HR in mammalian somatic cells (
22
).
Molecular characterization of HR products
To confirm that homologous recombination had occurred following transfection, plasmid DNA prepared from kanamycin-resistant bacterial colonies was subjected to restriction digestion
analysis using the endonuclease
Pst
I. The plasmids pSV2neoDR and pSV2neoDL can be easily distinguished from full-length pSV2neo, following digestion with this restriction endonuclease
(see Fig.
2
, lanes 1-3). When a number of kanamycin resistant colonies were analyzed in this
way, they all contained an intact neomycin phosphotransferase gene
(representative examples are shown in Fig.
2
, lanes 4 and 5). In contrast to the recombinant products observed following
cell-free HR reactions (see below) all of the plasmids analyzed in this fashion
were monomeric.
It is important to note that while the levels of HR seen in the normal cells
(MEF and human diploid fibroblasts) are very low, they are at least 10-fold higher than levels that are observed when pSV2neoDR and pSV2neoDL are
co-transformed directly into DH10B bacteria (Table
1
). It is also noteworthy that the levels of HR activity in different
immortalized cell lines vary considerably, in agreement with previous reports (
22
,
36
).
HR levels are elevated in protein extracts prepared from immortal cells
The previous results provided evidence that both murine and human normal cells are significantly less capable of catalyzing HR between plasmids than their immortalized counterparts. However, these transfection-based results must be interpreted cautiously, since it is possible that
other factors such as differential plasmid transfection or replication within the various cells tested could influence the
results in an unpredictable fashion. With these concerns in mind, we pursued a
biochemical approach to confirm this observation. Nuclear protein extracts were
prepared from a number of normal and immortalized cell lines, to determine
whether the same pattern of recombination activity would be observed. The use
of this approach would overcome a number of theoretical objections based upon
the transfection-based assay (see Discussion), and provide more direct confirmation that
immortalized cells possess more HR activity than normal cells.
In the cell-free HR assay (
37
), two different kinds of DNA substrates were used. Experiments were conducted
using: (i) a single-stranded derivative of pSV2neoDR along with linearized, double-stranded pSV2neoDL; and (ii) circular, double-stranded pSV2neoDR and pSV2neoDL. Using the cell-free HR assay described in the Materials and Methods, we
determined that all of the immortalized cell lines efficiently catalyze HR
(Table
2
). This was true irrespective of which DNA substrates were used. As indicated in
Table
2
, the absolute value of the HR frequency measured using these two double-stranded circular plasmids was about 100-fold lower than the corresponding value obtained from the single-strand circular, double-strand linear substrate combination. (This observation, which has been made previously, see
reference
37
, is probably due to the relative inefficiency with which the linearized plasmid
pSV2neoDL transforms bacteria, thereby dramatically reducing the number of
ampicillin-resistant colonies recovered.) In contrast, inspection of the data in
Table
2
indicates that the levels of HR activity in normal cells was negligible when
compared with that seen in immortal cells. We tested both of the human dermal
fibroblasts strains identified above, and found that neither possessed
detectable HR activity. Although the parental murine embryo fibroblasts
possessed a low level of HR activity when assayed with the single-strand and linearized double-strand DNA substrates, this was not statistically greater than the
background levels of HR seen in control experiments in which the two
recombination substrates were separately incubated with extract, and then
subsequently combined and introduced into the
rec
A
E.coli
strain used in these experiments.
To gain insight into the nature of the cell-free recombination reaction we performed restriction analysis of plasmid DNA obtained from kanamycin-resistant colonies resulting from co-incubation of pSV2neoDL and pSV2neoDR with nuclear protein
extracts. In a majority of cases, the bacteria harbored two plasmids. One plasmid invariably harbored a full-length neomycin phosphotransferase gene conferring kanamycin resistance. The second
plasmid generally corresponded to one of the two non-recombinant (i.e. DL or DR) plasmids, although occasionally a plasmid contained a double
deletion-bearing neomycin phosphotransferase gene was recovered. The two plasmids
found in these bacteria were present as either monomers, or occasionally, as
dimers. It is unknown to what extent interconversion of these two forms can
occur within the recombination-deficient
E.coli
host. In any event, it is clear that the products recovered are those expected
from conservative recombination reactions (as opposed to what would result from
a nuclease-mediated annealing reaction), and strongly argue that the kanamycin-resistant bacterial colonies arise as a consequence of legitimate HR
events catalyzed by the nuclear protein extracts. These experiments indicate that nuclear extracts from immortalized and transformed cells possess much higher levels of HR activity when compared
with extracts from normal cells.
Biochemical characterization of cell-free HR activity
As outlined in Table
3
, we observed that heat treatment of the protein extract destroyed the
recombination activity, and that the activity was almost entirely dependent
upon added ATP. We also observed that the nuclear HR activity was barely
detected when the ATP analog ATP[[gamma]]S was present. In addition, the HR activity occurred in the apparent
absence of DNA synthesis, since there was no requirement for exogenous deoxynucleoside triphosphates (data not shown) and the DNA polymerase inhibitors aphidicolin and dideoxy
adenosine triphosphate did not diminish HR activity (Table
3
).
Nuclear extracts from normal cells do not inhibit HR activity in immortal cell
extracts
The presence of enhanced HR activities in immortalized cells may reflect the
acquisition of new enzymatic activity, implying the activation of normally
quiescent genes, or alternatively, the loss of recombination inhibitors that
are normally present. We reasoned that if an inhibitor of HR were present in
normal cells, mixed extracts from the two cells would show diminished levels of
HR, while if an enhancer of recombination were present in immortalized cells,
the mixed extract would have levels of HR activity nearly the same as the
immortalized cells. To distinguish between these two hypotheses, we conducted
the experiments outlined in Figure
3
. As Figure
3
indicates, addition of 5 [mu]g of protein derived from normal fibroblast nuclei did not decrease the
activity present in either 1 [mu]g or 3 [mu]g of nuclear protein derived from immortal cells (there is no inherent
activity seen with 5 [mu]g of nuclear protein from normal cells). Due to the relatively large
standard errors, it was not possible to determine whether there was in fact
some synergistic activation of the protein from the normal cells. It is
possible that an abundance of some activity present in the immortalized cells
can act in concert with other proteins constitutively present in the normal cells, and thus explain the very modest increase seen in the assays in which extracts from the two
cells were mixed.
Identification of strand-exchange proteins present in immortal but not normal cell extracts
To further confirm that increased levels of HR proteins are present within the
nuclei of immortal cells, we utilized the newly described pairing on membrane
(POM) assay (
23
,
24
). This assay measures the ability of proteins that have been
electrophoretically separated and transferred to nitrocellulose to catalyze DNA
strand-transfer between single-stranded circular DNA that has been previously immobilized upon the membrane, and radiolabelled
homologous double-stranded linear molecules that are present in solution. Nuclear protein
extracts from normal and immortal cells were separated by SDS-polyacrylamide gel electrophoresis (
35
), and transferred to a membrane that had been coated with single-stranded circular [Phi]X174 DNA. This membrane was incubated with linearized, double-stranded [Phi]X174 DNA that had been radiolabelled at its 3' ends, as described by Akhmedov
et al.
(
23
). A number of hybridization signals, indicating the presence of apparent strand-transferase proteins, were observed in extracts from immortal cells
(Figure
4
A, lanes 2, 4, 5, and 6), while essentially no radioactive products were seen in
extracts made from normal human and murine cells (lanes 1 and3, respectively). An identical SDS-polyacrylamide gel, stained with Coomassie blue is shown in Figure
4
B. Parallel experiments in which double stranded, radiolabelled M13 DNA was used
in place of [Phi]X174 (the two phage genomes share no significant sequence similarity) did
not produce radioactive signals in any of the protein samples tested (data not
shown). This latter observation, which was repeated three times, indicates that
the hybridization signals seen in Figure
4
require homology between the single- and double-stranded DNA molecules.
DISCUSSION
We have utilized three different recombination assays to show immortalized cells
are significantly more capable of catalyzing HR than are normal cells. These
observations were made in both mouse and human cells, and in spontaneously
immortalized and established cell lines. First, we used co-transfection of two plasmids to measure inter-plasmid extrachromosomal HR (Table
1
). Our data are in agreement with a previous report in which extrachromosomal
intra-plasmid recombination was shown to be elevated in virally transformed
human fibroblasts and a number of established cell lines (
22
). We have extended this observation to include spontaneously immortalized
murine and non-human primate cells. The extent to which immortalization enhances HR is
somewhat difficult to determine, since different immortalized cell lines differ
in their HR frequency (Table
1
; 22 and 36). Nevertheless, our data indicate that the level of enhancement can be quite significant, since immortalized MEF cells are 85-fold more recombinogenic than the cells from which they are derived. Similarly, the
fibrosarcoma-derived cell line HT1080 is 62-fold more recombinogenic than either of the normal human dermal
fibroblast strains tested. Experiments performed on HT1080 and normal human diploid fibroblast cells indicated that intra-molecular HR was also ~60-fold more frequent in the former, compared with the latter. We
also noted that the absolute levels of intra-molecular HR were approximately 10-fold higher than the levels of inter-molecular HR. These results are in qualitative agreement with a previous observation (
22
) that indicated immortalization resulted in an increase in intra-molecular plasmid HR frequency. (This observation also suggests a likely explanation for the absence of dimeric recombinant plasmids following co-transfection of pSV2neoDL and pSV2neoDR into mammalian somatic cells; these dimers share extensive homology which probably leads to intra-molecular recombination, generating monomer plasmids.)
The results presented in Table
1
were obtained by summing all of the kanamycin-resistant and ampicillin-resistant colonies obtained from a number of separate transfections.
We used this approach, rather than presenting the data as the mean value
obtained from each separate transfection because, in the case of the normal mammalian somatic cells, we frequently did not obtain any kanamycin-resistant colonies from a particular transfection. In the case of the human HT1080 cell line, it is possible to calculate
mean and standard error values by analyzing the results of a number of separate
experiments, since each transfection yielded kanamycin-resistant colonies. We have performed this analysis on two sub-clones of the HT1080 cell line, and determined that the HR frequency
is 81 +- 7, and 60 +- 13 kanamycin-resistant colonies per 10
5
ampicillin-resistant colonies, respectively. The former subclone of HT1080 was the one used in the
experiments shown in Table
1
-
3
. These results demonstrate that the HR frequency values we have measured are
highly reproducible. We have also performed this type analysis on the intra-plasmid HR experiments conducted with HDF and HT1080 cells. Here we have
determined that the intra-plasmid HR frequency of the former is 10 +- 1.6, while that of the latter is 628 +- 55 (in both cases, this value refers to the number of
kanamycin-resistant colonies per 10
5
ampicillin-resistant colonies, plus or minus the standard error of the mean.)
While all of the immortal cells possess HR levels significantly greater than
those seen in the normal human or mouse fibroblasts, we see no obvious relationship between HR activity and species of origin. In
addition, there does not appear to be a correlation between the amount of plasmid DNA that has replicated within a cell line (and is
resistant to
Dpn
I digestion) and HR activity. Murine 3T3 cells yielded relatively few ampicillin-resistant colonies (Table
1
), and had one of the higher HR activities, whereas MEF cells yielded many-fold more ampicillin-resistant colonies, and gave the lowest HR frequency (Table
1
). It is therefore unlikely that enhanced HR is the result of increased
concentrations of plasmid molecules within the cells. A recent paper has shown
that the search for homology does not represent a rate-limiting step in mammalian somatic cell HR (
38
). It is therefore not likely that the HR frequency data presented in Table
1
reflects differences in transfection efficiency amongst the various cell lines
tested.
There also does not appear to be a correlation between the frequency of cellular
HR and the cell cycle. For example, we measured the growth rate of both normal
and immortalized MEF cells. While the latter does grow slightly faster than the
former (less than two-fold difference in the double time, not shown), its HR frequency is about
85-fold greater than normal MEF cells. In contrast, the growth rate of CV-1 cells is intermediate between that of the normal and immortalized
MEF cells, yet, as Table
1
indicates, its HR frequency is even greater than that of the immortalized MEF
cells. It is therefore unlikely that the very great disparity between HR
frequency seen in normal and immortal clones may be explained on the basis of
differences in the cell division cycle of these respective cells.
A number of theoretical objections to the use of a transfection-based assay to compare HR frequencies amongst different cell lines can be
raised. Using this assay, it is not possible to ensure that additional factors,
such as the extent to which plasmid transfection efficiency, or replication are
not responsible for the different HR frequencies observed. For this reason, we
utilized two additional HR assays, in which any of the objections raised to the
transfection assay will not apply.
First a cell-free homologous recombination assay allowed us to independently confirm that immortalized cells contain significantly more HR activity than do normal cells (Table
2
). Since the levels of HR activity of both of the normal human cell strains used
in this assay were undetectable, it was not possible to measure the extent to
which immortalization enhanced HR levels. However, the HR frequency observed in
murine embryo fibroblasts (MEFs) that had spontaneously immortalized was
approximately 60-fold higher than that seen in the parental MEF cells. It is remarkable
that both the transfection-based assay and the nuclear extract-based assay indicate that immortalization stimulates HR levels to a
similar extent (62 to 85-fold, versus 60-fold, respectively). We performed a Spearman-rank correlation test, in which the HR data [cell lines COS-1, 3T3, MEF (immortalized), HT1080, MEF (normal) and
HDF] from these two different assays were compared. This analysis reveals a
correlation coefficient of 0.9, (
P
= 0.019), strongly suggesting that HR levels measured using the transfection-based assay are not biased by features such as differential transfection
efficiency, plasmid replication, which are not relevant to the cell-free HR assay.
The ability to measure HR using cell-free extracts allowed us to address a number of questions. The first is
whether the ability of immortalized cells to catalyze HR represents the
acquisition of a novel activity or, conversely, the loss of normal inhibitory
proteins. As presented in Figure
3
, our data suggest the former, since the addition of normal cell extract to
immortalized cell extract did not have a detectable inhibitory effect on the
level of activity observed when the immortal cell extract was tested alone.
This observation was made with two different concentrations of immortal cell
extract, well within the region where increasing protein yielded increasing
amounts of homologous recombinants (Table
2
). These data are therefore consistent with the interpretation that normal cells lack a factor or factors that, when present, greatly
stimulate HR. We acknowledge, however, that there may be other interpretations
of these data (see below).
Experiments using the pairing-on-membrane (POM;
23
,
24
) assay support the conclusion that immortal cells possess higher levels of HR than their normal counterparts. As shown in Figure
4
, high levels of active strand-transferase proteins are present in nuclear extracts of immortal, but not
normal, cells. Upon close inspection there is a small amount of strand-transferase activity in lane 3 of Figure
4
A, corresponding to the nuclear extract prepared from MEF cells. This
observation may reflect a low constitutive level of strand-transferase activity present in all the cells of the population, or it may
be the result of small numbers of immortalized MEF cells present within the
normal cell population. Since MEF cells are known to undergo frequent
spontaneous immortalization, we favor the latter hypothesis, particularly since
no such strand-transferase activity is detectable in the normal human cells (Fig.
4
A, lane 1), which are not susceptible to spontaneous immortalization.
We detect the presence of three to four strand-transferase proteins in extracts from immortal cells. It remains to be determined whether these
results indicate the coordinate expression of a number of distinct proteins.
Alternately, it is possible that these multiple signals represent post-translational modifications, or the proteolytic breakdown products of a
smaller number of precursor proteins.
The POM assay results provide additional support for the conclusion that the
enhanced level of HR in immortal cells is the result of the synthesis of novel
enzymes that are not expressed in normal cells. Since the presence of DTT and
SDS within the electrophoresis buffer disrupts protein-protein interactions, a
protein could inhibit strand-transferase activity only if it were of the same molecular weight as the
strand-transferase itself. Therefore, if one assumes that the strand-transferases observed in Figure
4
A are responsible for the HR activity measured by the experiments shown in Table
1
and
2
, two interpretations are possible. The first is that the strand-transferase (HR) protein is absent from normal cells, but present in immortal cells. Alternatively,
constitutively expressed HR proteins may be post-translationally modified in immortal cells through a process that does not occur in normal
cells. For example, it is possible that a protein kinase required to activate strand-transferases is present within immortalized cells, but absent in normal
cells. For this model to be true, it must be assumed that this kinase is either
absent or incapable of functioning when the immortal and normal cell extracts
are mixed.
We cannot differentiate between these two hypotheses at this time. Having
identified a number of strand-transferase proteins (Fig.
4
A), it may be possible to determine whether proteins of similar molecular weight
are differentially phosphorylated, or otherwise modified in normal versus
immortal cells. In addition, it may now be possible to purify these proteins
and clone the genes that encode them. We believe that this approach may
ultimately lead to the identification of novel genes that play an important
role in oncogenic transformation.
The major issue raised by this work concerns the role that HR plays in the oncogenic process. As was outlined in the introduction, a plausible, if speculative, argument can be made that HR plays a role in the
genomic rearrangements that appear to play a role in oncogenesis. Although
additional work will be required to establish a clear correlation between
genomic instability and elevated HR activity, our results do indicate that
elevated levels of HR are seen at early stages in the transformation process.
The MEFs used in this study were between passage numbers 1 and 8. There is no
relationship between passage number and HR frequency within this range of
passage number, however, as soon as we can identify and isolate spontaneously
immortalized murine embryo fibroblasts, they show elevated levels of HR activity, as well as
numerous strand-transferase proteins. (In addition, our results indicate that elevated HR
activity is a common, if not ubiquitous feature of immortalized cells.) Results from mixing
experiments, and from the POM assay both suggest that the enhanced levels of HR
seen in immortal cells are likely to result from the induction of recombination
activity, rather than the loss of recombination inhibitors. To conclusively rule out other interpretations, however, will require more detailed studies.
Based on the data presented here, it would be premature to conclude that altered
HR levels play a direct causative role in oncogenesis. However, it seems
reasonable to propose that elevated HR levels may help to confer a growth
advantage on mammalian somatic cells. As was outlined in the introduction,
mutations that confer a mutator phenotype on human cells are clearly cancer pre-disposing. Enhanced cellular HR may contribute to increased genomic instability, and thereby facilitate the genetic
events that lead to transformation. It is therefore likely that a greater
understanding of the molecular genetics of HR in mammalian somatic cells will
provide additional insight into the process of oncogenesis.
ACKNOWLEDGEMENTS
This work was supported in part by grants from the Minnesota Medical Foundation,
the H. Louise Ruddell Charitable Trust, the American Heart Association, MN
affiliate, and the Leukemia Task Force of Minnesota. M. M.-G. was supported through funds provided by the University of Minnesota
Graduate School. We are grateful to Dr Robert O'Dea for providing normal human
fibroblasts, and for helpful discussions. Drs P.-Y. Law and L.-N. Wei provided immortalized monkey, and murine cell lines,
respectively. We thank Drs Jennifer Cruise and Cecilia Warner for helpful
editorial suggestions.
REFERENCES
1 Boveri, T. (1929) In The Origin of Malignant Tumors, Williams and Wilkins, Baltimore.
2 Sandberg, A. A. 1990. In The Chromosomes in Human Cancer and Leukemia, Elsevier, New York.
3 Bronner, C. E., Baker, S. M., Morrison, P. T., Warren, G., Smith, L. G., Lescoe, M. K., Kane, M., Earabino, C., Lipford, J., Lindblom, A., Tannergard, P., Bollag, R. J., Godwin, A. R., Ward, D. C., D. C. Nordenskjold, M, Fishel, R., Kolodner, R. and Liskay, R. M. (1994) Nature 368, 258-261.
4 Fishel, R., Lescoe, M. K., Rao, R. R., Copeland, N. G., Jenkins, N. A., Garber, J., Kane, M., and Kolodner, R. (1993) Cell75, 1027-1038.MEDLINE Abstract
5 Leach, F. S., Nicolaides, N. C., Papadopoulos, N., Liu, B., Jen, J., Parsons, R., Peltomaki, P., Sistonen, P., Aaltonen, L. A., L. A. Nystrom-Lahti, M., Guan, X.-Y., Zhang, J., Meltzer, P. S., Yu, J.W., Kao, F.-T. D., Chen, J., Cerosaletti, K. M., Fournier, R. E. K., Todd, S., Lewis, T., Leach, R. J., Naylor, S. L., Weissenbach, J., Mecklin, J.-P., Jarvinen, H., Petersen, G. M., Hamilton, S. R., Green, J., Jass, J., Watson, P., Lynch, H. T., Trent, J. M., de la Chapelle, A., Kinzler, K. W., and Vogelstein, B. (1993) Cell75, 1215-1226.
6 Nicolaides, N. C., Papadopoulos, N, Liu, B., Wei, Y.-F., Carter, K. C., Ruben, S. M., Rosen, C. A., Haseltine, W. A., Fleischmann, R. D., Fraser, C. M., Adams, M. D., Venter, J. C., Dunlop, M. G., Hamilton, S. R., Petersen, G. M., de la Chapelle, A., Vogelstein, B., and Kinzler, K. W. (1994) Nature 371, 75-80.
7 Papadopoulos, N., Nicolaides, N. C., Wei, Y.-F., Ruben, S. M., Carter, K. C, Rosen, C. A., Haseltine, W. A., Fleischmann, R. D., Fraser, C. M., Adams, M. D., Venter, J. C., Hamilton, S. R., Petersen, G. M., Watson, P., Lynch, H. T., Peltomaki, P., Mecklin, J. P., de la Chapelle, A., Kinzler, K. W.,and Vogelstein, B. (1994) Science263, 1635-1639.
8 Lynch, H. T., Smyrk, T. C., Watson, P., Lanspa, S. J., Lynch, J. F., Cavalieri, R. J., and Boland, C. R. (1993) Gastroenterology104,1535-1549.
9 Aaltonen, L. A., Peltomaki, P., Leach, F. S., Sistonen, P., Pylkkanen, L., Mecklin, J.-P., Jarvinen, H., Powell, S. M., Jen, J., Hamilton, S. R., Petersen, G. M., Kinzler, K. W., Vogelstein, B., and de la Chapelle, A. (1993) Science260, 812-816.
10 Ionov, Y. M., Peinado, A. ,Malkhosyan, S., Shibata, D., and Perucho, M. (1993) Nature 363, 558-561.
11 Peinado M. A., Malkhosyan, S., Velazquez, A., and Perucho, M. (1992) Proc. Natl Acad. Sci. USA89, 10065-10069.
12 Thibideau, S. N., Bren, G., and Schaid, D. (1993) Science260, 816-819.
13 Kolodner, R. D., and Alani, E. (1994) Curr. Opin. Biotech.5, 585-594.
14 Bhattacharyya, N. P., Skandalis, A., Ganesh, A., Groden, J., and Meuth, M. (1994) Proc. Natl Acad. Sci. USA91, 6319-6323.
15 Parsons, R., Li, G.-M., Longley, M. J., Fang, W.-H., Papadopoulos, N., Jen, J., de la Chapelle, A., Kinzler, K. W., Vogelstein, B., and Modrich, P. (1993) Cell75,1227-1236.
16 Umar, A., Boyer, J. C., Thomas, D. C. Nguyen, D. C., Risinger, J. I., Boyd, J., Ionov, Y., Perucho, M.,and Kunkel, T. A. (1994) J. Biol. Chem.269, 14367-14370.MEDLINE Abstract
17 Petit, M. A., Dimpfl, J., Radman, M., and Echols, H. (1991) Genetics129, 327-332.
18 Rayssiguier, C., Thayler, D. S., and Radman, M. (1989) Nature 34, 396-401.
19 Alani, E., Reenan, R. A. G., and Kolodner, R. (1994) Genetics137, 19-39.MEDLINE Abstract
20 Selva, E. M., New, L., Crouse, G. F., and Lahue. R. S. (1995) Genetics139,1175-1188.
21 de Wind, N., Dekker, M., Berns, B., Radman, M., and te Riele, H. (1995) Cell 82, 321-330.MEDLINE Abstract
22 Finn, G. K., Kurz, B. W., R. Z. Cheng, R. Z., and Shmookler Reis. R. J. (1989) Mol. Cell. Biol.9, 4009-4017.
23 Akhmedov, A. T., Bertrand, P., Corteggiani, E., Lopez, B. S. (1995) Proc. Natl Acad. Sci. USA92, 1729-1733.
24 Bertrand, P., Corteggiani, E., Dutreix, M., Coppey, J., and Lopez, B. S. (1993) Nucleic Acids Res.21, 3653-3657.MEDLINE Abstract
25 Rasheed, S., Nelson-Rees, W. A., Toth, E. M., Arnstein, P., and Gardner M. B. (1974) Cancer33, 1027-1033.MEDLINE Abstract
26 Jensen, F. C., Girardi, A. J., Gilden, R. V., and Koprowski, H. (1964) Proc. Natl Acad. Sci. USA 52, 53-59.
27 Todaro, G. J., and Green, H. (1963) J. Cell Biol. 17, 299-313.
*To whom correspondence should be addressed at: Department of Pharmacology, 3-249 Millard Hall, University of Minnesota Medical School, 435 Delaware
Street, SE, Minneapolis, MN 55455, USA
