ABSTRACT
A series of different frameshift mutations of a firefly luciferase reporter
plasmid was created so that no activity was obtained when they were transfected
into mammalian cells. Co-transfection of these constructs with short fragments of the original
sequence resulted in luciferase activity in different cell lines (A-549, NIH 3T3 and Jurkat). The level of this activity was dependent on the
length of the fragment, regardless of cell line examined. Two different
transfection techniques (lipofection and adenovirus-enhanced gene transfer) gave similar results. It was shown by polymerase
chain reaction that expression of detectable luciferase required recombination
of the transfected molecules. Cells with defined defects in DNA repair pathways
were examined for their ability to perform this extrachromosomal recombination.
Cells lacking normal Ku p80, (ADP-ribosyl)transferase, MLH1 or XP-C were all capable of restoring expression to the frameshifted
constructs. Given the pivotal roles of the above molecules in the pathways of
DNA repair, it seems that this recombination derives from a different activity.
DNA introduced into mammalian somatic cells is acted on by the cellular DNA
repair and recombination machinery. Transfected substrates have been used to
examine general recombination in a variety of cell types, with selectable
marker genes (
1
-
9
) or reporter constructs such as
lacZ
(
10
,
11
) for quantitation. Mismatch repair has been investigated productively using non-chromosomal DNA (
12
). Since the luciferase reporter gene can be analysed rapidly and is quantitated
easily over a wide range (
13
), it seemed to be a useful tool for examining extrachromosomal recombination (ECR) in a non-selective assay. We developed a transient transfection assay using short
fragments of the luciferase gene to correct non-expressing frameshifts which could be performed in a wide range of cell
types. Parameters important in homologous recombination (HR), such as length of
the homologous fragment (
14
,
15
) or size of the frameshift mutation, could be tested using this assay.
Since ECR has been used in DNA damage-sensitive cell lines to examine the relationship of DNA repair and
recombination (
3
,
4
,
9
,
11
), this assay was performed in cell lines mutant in various DNA repair genes, so
that the nucleotide excision, mismatch repair and double-strand break pathways of DNA repair (reviewed in
16
,
17
) could each be examined for their involvement. The CHO-derived line xrs-6, defective in Ku p80, which binds DNA ends and is associated with
DNA-dependent protein kinase (
18
,
19
), was assayed, as were cells from mice lacking ADPRT
(
20
), an abundant nuclear protein believed to be involved in DNA recombination (
21
). Other lines investigated were HCT116, a cell line defective in
MLH1
(involved in mismatch repair;
22
,
23
) and GM02249d, derived from an XP-C patient (
24
,
25
). An AT line (GM01525e) was also examined, as it had been reported that very
high recombination levels could be seen in AT cells, though in an
intrachromosomal context; ECR rates were similar to those observed in control
lines (
10
,
11
). Since all the mutant cells were as capable of carrying out ECR, as measured
with the assay we describe, as their respective control lines, we conclude that
it is not dependent on any single one of the above enzymes.
All media and calf sera were from Gibco BRL (Gaithersburg, MD) and were
supplemented with 2 mM L-glutamine and antibiotics. NIH 3T3 (mouse fibroblast) and A-549 [human lung carcinoma, American Type Culture Collection (ATCC)
CCL 185] were maintained as monolayers in Dulbecco's modified Eagle's medium
(DMEM) containing 10% foetal calf serum (FCS). Primary mouse fibroblasts from
both ADPRT knockout mice and wild-type animals were obtained from Dr Z.-Q.Wang (IMP, Vienna, Austria) and were grown in DMEM, 10% FCS with
the addition of 5 * 10
-5
M [beta]-mercaptoethanol. CHO-K1 and xrs-6 lines were obtained from the European Collection of
Animal Cell Cultures (Porton Down, UK) and were plated in Ham's F-12 nutrient mix supplemented with 10% FCS. HCT116 lines were obtained from
Dr A.Umar (National Institute of Environmental Health Science, NC) and were
kept in 1:1 DMEM/Ham's F-12 medium with 10% FCS; HCT116 stably transfected with chromosome 3.6 was
kept under selection with 400 [mu]g/ml (active weight) G-418 (Sigma, St Louis, MO). Jurkat E6-1 (human acute T cell leukaemia, ATCC TIB 152) was passaged in
RPMI 1640 medium supplemented with 10% FCS. EBV-transformed B lymphoblastoid lines GM02249d (XP-C), GM01525e (AT) and GM00558b were obtained from Coriell Cell
Repositories (Camden, NJ) and were grown in RPMI 1640, 15% FCS.
Plasmid pCMV-L (Fig.
1
), consisting of the
Photinus pyralis
luciferase gene under control of the CMV promoter, has been described (
26
). This plasmid was linearised at a unique
Nar
I site in the luciferase sequence and either treated with S1 nuclease for
different lengths of time and then filled in using T4 polymerase or simply
filled-in using the polymerase before religation. This provided plasmids pNXS-1, pNXS-2 and pNXX, as shown in Table
1
. Fragments of pCMV-L containing the
Nar
I site (shown in Fig.
1
) were subcloned into the appropriate sites of pUC19 using standard techniques (
27
). The sequences of all constructs used for this work were confirmed by
restriction digestion and DNA sequencing.
For all transfections, 3.7 [mu]g frameshifted luciferase reporter plasmid was used as the target. The mass
of the fragment DNA used was dependent on its length, as the same molar ratios
of target to fragment (1:2, though for the GM lines and primary fibroblasts a
ratio of 1:4 was chosen) were used for comparative experiments. Vector DNA
(pUC19 or pSP65) was used to ensure that the same absolute amount of DNA was
used in each set of experiments; this amount was 6.6 [mu]g for all but the GM lines and the primary fibroblasts, for which it was 9.4
[mu]g. Plasmid DNAs were linearised with
Xmn
I prior to transfection and were heat denatured for 10 min at 95oC just before mixing with the unheated repair fragments, which were then
added to the transfection complexes as described below. This heat denaturation
did not have a significant effect on expression, though linearisation of the
substrates was necessary for good recombination. Luciferase assays were
performed as described (
13
,
28
). Cells were harvested either 24 or 40 h after transfection and were washed
once with phosphate-buffered saline and then harvested in 0.1% Triton X-100, 250 mM Tris-HCl, pH 7.3. Luciferase activity was measured from an aliquot
of the supernatant using a Clinilumat LB9502 instrument (Berthold, Bad Wildbad,
Germany). Protein content was measured by Bradford dye binding using the BioRad
protein assay (BioRad, Hercules, CA); this was performed routinely to avoid
intra-assay variation.
Adherent cells (NIH 3T3, A-549, HCT116, CHO-K1, xrs-6 and primary mouse fibroblasts, passages 5 and 6) were plated
overnight in the appropriate medium at a density of 10
5
/well of a 6-well dish (Nunc, Roskilde, Denmark). Jurkat and GM cells were plated at 10
6
/well of a 24-well dish (Nunc) 2-4 h before transfection. Transfection conditions appropriate for
each of the cell lines used were defined using control reporter constructs.
LipofectAMINEtm was purchased from Gibco BRL. For lipofection of NIH 3T3, A-549, HCT116, CHO-K1, xrs-6 and primary mouse fibroblast cells, DNA was diluted
to 50 [mu]l with HEPES-buffered saline (HBS; 20 mM HEPES, pH 7.3, 150 mM NaCl) and added to
LipofectAMINE[ordf] in a final volume of 200 [mu]l, made up with DMEM without serum. The ratio used was 6.6 [mu]g DNA:12 [mu]g lipid. Complexes were allowed to form over 45 min, after
which time they were added to cells in a final volume of 1 ml, made up with
serum-free DMEM. After 3 h, the medium was replaced with that appropriate for
the cell line in question, without selective agents.
Biotinylated adenovirus stock dl1014 (E4 defective;
29
) was prepared and inactivated as previously described (
30
-
32
). Streptavidin-polylysine (StrApLys) and human transferrin-polylysine (TfpLys) conjugates were prepared according to
published methods (
28
,
32
) The coupling of OKT3, an [alpha]-CD3 antibody for DNA transfer into T lymphocytes, to polylysine ([alpha]CD3-pLys) has been described recently (
33
). Transferrinfection of A-549 cells proceeded as follows. An aliquot of 10
9
biotinylated virus particles (10
4
/cell) was incubated at room temperature with 0.5 [mu]g StrApLys for 30 min in a volume of 150 [mu]l HBS, after which the DNA was added in a volume of 100 [mu]l. After 30 min, 2.5 [mu]g TfpLys was added, to a final volume of 300 [mu]l. Transfection complexes were allowed to form for 30 min,
then were added to the cells in a final volume of 1 ml, made up with the
appropriate medium. Jurkat cells were transfected in essentially the same
manner, save that 1.0 [mu]g StrApLys was used and the TfpLys was replaced with 4 [mu]g [alpha]CD3-pLys. For transfection of the GM lines, volumes were as
above, but 5 * 10
9
particles (5 * 10
3
/cell) were used to transfect 9.4 [mu]g substrate DNA, along with 2 [mu]g StrApLys and 2.5 [mu]g TfpLys/7.5 [mu]g pLys in place of the TfpLys alone. Transfection complexes were
left on the cells for 3 h, after which time the medium was replaced.
DNA was phenol/chloroform extracted from cell lysates of NIH 3T3 cells
transiently transfected with pNXS-2 and NF3 correcting fragment. This lysate was positive for luciferase.
After ethanol precipitation into Tris-EDTA buffer, oligonucleotide sequences appropriate for discrimination
between frameshifted and repaired plasmid and repair fragment were used to
amplify the DNA, as shown diagrammatically in Figure
4
A. These were (1) 5'-GGCAGTACATCAATGGG-3', lying outside the
Sac
I-
Sph
I fragment and (2) 5'-GATAGAATGGCGCCGGGC-3', the 3'-end of which lay in the deleted sequence.
A mixture of pNXS-2 and NF3 was used as negative control and pCMV-L was used as a positive control for PCR (
34
). To ensure that sufficient target was present for amplification, the primers
(3) 5'-GCTTGGGAATTCCTTTGTGTTACA-3' and (4) 5'-GAAGAGAGTTTTCACTGC-3' were used to amplify a
region of the luciferase gene present in all samples.
Somatic cells are capable of performing homologous recombination on transfected
substrates. The luciferase reporter gene appeared to be a promising candidate
for studying such recombination. Different plasmids containing short
frameshifting deletions in the reporter gene (as shown in Table
1
) were transfected into A-549 and NIH 3T3 cells. As shown in Figure
2
, no luciferase activity was detected after transfection. However, expression
could be restored to all the constructs tested in both cell lines by co-transfecting a small, non-expressing fragment of the original sequence (Fig.
2
). Efficiency appeared to be independent of the size of the frameshift, within
the range tested here, though the ease with which the cell line could be
transfected was important.
In order to examine whether mutations and defects in the pathways of DNA repair
have any effects in this ECR assay, we examined primary ADPRT
-/-
mouse fibroblasts, the X-ray-sensitive CHO line xrs-6, defective in Ku p80, the hMLH1-defective line HCT116, the XP-C line GM02249d and the AT line GM01525e. pNXS1
and the NF3 fragment were transfected into these mutant cell lines and, where
possible, into similar, non-mutant cell lines. These lines were CHO-K1, the parental line for xrs-6 (
3
,
4
,
18
,
19
), HCT116 stably cloned with human chromosome 3.6, which complements its
mismatch repair mutation in an
in vitro
assay
(
36
), wild-type mouse cells for the ADPRT knockouts and a line from an apparently
normal donor, GM00558b, for the lymphoblastoid cells. All the mutant cell lines
could effectively correct the frameshifts, as comparisons of xrs-6 and CHO-K1, of HCT116 and HCT116 + chromosome 3.6 and of ADPRT
-/-
fibroblasts with wild-type cells showed no significant differences in expression (Fig.
5
). Expression was restored in the XP-C line GM2249d and in the AT line GM01525e as efficiently as in the
control lymphoblastoid cell line GM00558b. Dependence of restored luciferase
expression on the size of the correcting fragment, as had been observed for NIH
3T3, A-549 and Jurkat cells (Fig.
3
), was observed in all these lines (data not shown).
Our results show that recombination between a frameshifted luciferase reporter
plasmid and short DNA fragments containing the original sequence occurs
efficiently in mammalian somatic cells and results in restoration of luciferase
expression. Using the transfection systems available in our laboratory, such
recombination could be observed in a number of cell lines, including some
normally refractory to transfection (Jurkat and lymphoblastoid cells). This
restoration of expression can be observed with frameshifts of different sizes
and is dependent on the length of the DNA fragment used to achieve this
restored expression. The strong length dependence observed here compares
closely with results on HR using selective markers (
14
,
15
), though the lengths of the constructs are rather shorter due to the limit
imposed by the luciferase gene.
This assay is not selective, so that individual clones could not be examined for
the structure of the molecules involved in restored reporter expression.
However, PCR analysis showed that recombination of the substrates is necessary
for the generation of restored reporter gene expression (Fig.
4
). After luciferase expression in NIH 3T3 cells was regenerated with these DNAs,
a product could be amplified with the discriminatory primers, indicating that
they were now on the same, recombined molecule responsible for expression of
the reporter gene. Where the 5' end of one primer lay in the 7 bp deletion of pNXS-2 and the other primer lay outside the repair fragment on the
luciferase sequence, an
in vitro
mixture of substrates gave no amplification, indicating that expression is due
to replacement of the frameshift by the correct sequence.
We then investigated what activities were responsible for this. It has been
observed that double-strand breaks are recombinogenic (
6
,
35
,
37
). The introduction of the substrate DNAs described here might be seen as the
introduction of a large number of double-strand breaks. Recent work has shown that Ku protein binds to DNA breaks
and that its deficiency, as in xrs-6 cells, leads to defects in double-strand break repair and V(D)J recombination (
18
,
19
,
38
,
39
). The absence of any significant difference between control (CHO-K1) and xrs-6 cells indicated that Ku can be omitted from this system without
hampering ECR, which had not been wholly clear from previous evidence (
3
,
4
). Another cellular component which binds to DNA breaks is ADPRT (
21
,
40
). The availability of mouse cells lacking this enzyme (
20
) made possible an examination of whether there was a crucial role for this
enzyme in the recognition of the substrates used in this assay. As shown in
Figure
5
, this was not the case; the kinetics of restoration of expression were the same
in the ADPRT
-/-
cells and the controls (data not shown). The recognition of the substrates here
as double-strand breaks, if involved in this correction, is not reliant on Ku or
ADPRT.
If the restoration of expression to the frameshifts used in our system generates
substrates analogous to the insertion/deletion mutants used for the examination
of mismatch repair mutants, restoration of luciferase expression should be
hindered in cells defective in mismatch repair. The DNA repair systems involved
in mismatch correction, which have been linked to HNPCC in humans, are also
capable of repairing small insertion/deletion mismatches (
16
,
23
). These mismatches consist of small ( <= 5 bp) bulges in DNA, due to a region of unpaired bases in the duplex, and
can lead to large variations in the structure of the DNA (
41
); p53 is implicated in their recognition (
42
). The absence of the mismatch repair protein hMLH1 from extracts of HCT116
cells has been shown to prevent their repair of DNAs containing small ( <= 5 bp) bulges (
23
). However, efficient expression of luciferase was detected in these cells,
indicating correction of the frameshift and that the substrates described need
not necessarily be repaired by the mismatch repair system. The lack of effect
of differing frameshifts, as shown in Figure
2
, is in keeping with this finding, and shows that ECR is essentially independent
of mismatch repair, though this is not the case in chromosomal HR (
43
).
The lesions recognised by the nucleotide excision repair system (pyrimidine
dimers and chemical adducts) are rather different from those provided in our
assay (
24
). The ability of XP-C to recombine transfected DNA was not impeded by the mutation which
contributes to their DNA damage sensitivity, though this might have been
expected in XP-C cells, which have a repair activity on transcriptionally active DNA (
24
,
25
,
44
). This suggests that this pathway is not involved in restoration of expression
by ECR. A similar conclusion may also be drawn for the mutation in AT, which
disease involves problems in DNA metabolism.
The provision of so many DNA ends to the cells in our assay may overwhelm the
specific systems of DNA repair, so that even if the mutations have an effect,
they may not be seen. Despite this caveat, these results point toward the
existence of a non-specific system acting to recombine transfected DNA, the action of which
is not dependent on the molecules whose mutants were tested here.
The provision of cells by Drs P.Jeggo, A.Umar and Z.-Q.Wang is gratefully acknowledged. We would like to thank Dr Z.-Q.Wang for his critical reading of the manuscript and Drs Matt
Cotten and Walter Schmidt for many helpful discussions and ideas. The skilful
assistance of Ivan Botto in DNA sequencing and Sissy Aigner in oligonucleotide
synthesis has been invaluable throughout this project.
REFERENCES
Return