ABSTRACT
The antitumor agent
cis
-diamminedichloroplatinum(II) (cisplatin) introduces cytotoxic DNA damage
predominantly in the form of intrastrand crosslinks between adjacent purines.
Binding assays using a series of duplex oligonucleotides containing a single 1,2 diguanyl intrastrand crosslink indicate that human cell extracts contain factors that
preferentially recognise this type of damage when the complementary strand
contains T opposite the 3
'
, and C opposite the 5
'
guanine in the crosslink. Under the conditions of the band-shift assay used, little binding is observed if the positions of the T and
C are reversed in the complementary strand. Similarly, duplexes containing CC
or TT opposite the crosslink are recognised relatively poorly. The binding
activity is absent from extracts of the colorectal carcinoma cell lines LoVo
and DLD-1 in which the hMutS
[alpha]
mismatch recognition complex is inactivated by mutation. Extensively purified
human hMutS
[alpha]
exhibits the same substrate preference and binds to the mismatched platinated
DNA at least as well as to an identical unplatinated duplex containing a single
G
.
T mismatch. It is likely, therefore, that human mismatch repair may be triggered
by 1,2 diguanyl intrastrand crosslinks that have undergone replicative bypass.
Cisplatin is used in the treatment of several types of cancer and is
particularly effective against testicular tumors (
1
). In common with many chemotherapeutic agents, the clinical effectiveness of
cisplatin is limited by the emergence of drug-resistance (
2
). Cisplatin adducts in DNA are repaired by nucleotide excision repair (NER) (
3
,
4
) and cells deficient in this repair pathway, such as those derived from
xeroderma pigmentosum patients, are hypersensitive to cisplatin. Increased
excision of cisplatin adducts has been observed in some resistant cell lines
derived in the laboratory (
5
). Other resistant variants can tolerate higher levels of cisplatin-induced DNA damage without detrimental effects on survival (
6
). 1,2 dipurinyl intrastrand crosslinks comprise >= 80% of cisplatin adducts (
7
). In comparison to the less abundant 1,3 diguanyl crosslinks, the abundant 1,2
adducts, are rather poor substrates for removal by NER (
4
,
8
). Their relatively long persistence in DNA suggests that many 1,2 diguanyl
crosslinks may undergo replicative bypass.
In addition to the protection conferred by NER, cells can sometimes acquire
cisplatin resistance through an increased ability to tolerate unexcised
cisplatin lesions in their DNA. This tolerance arises through loss of a DNA
mismatch repair pathway. Cisplatin-resistant variants of a human ovarian carcinoma cell line exhibit
microsatellite instability (
9
) and are mismatch repair deficient owing to defective expression of the hMLH1
mismatch repair protein (
10
). In addition, some human colorectal carcinoma cell lines deficient in hMLH1 or hMSH2 are more resistant to cisplatin than
sublines in which the mismatch repair defect has been complemented by
chromosome transfer (
11
). Thus, in contrast to NER which promotes cell survival, a functional DNA
mismatch repair pathway appears to contribute to cisplatin cytotoxicity.
Mismatch repair corrects single base mispairs and the looped intermediates,
typically one to three bases, that arise by slippage during replication of
repeated DNA sequences (for review see
12
). Reversal of these replication errors prevents the increased spontaneous
mutation rates and microsatellite instability that characterise mismatch repair
defective cells. In normal human cells, DNA mismatches are recognised by one of
two known mismatch recognition complexes. The best characterised of these,
hMutS[alpha], is a heterodimer of hMSH2 and hMSH6 (also known as GTBP) (
13
,
14
). hMutS[alpha] binds to single base mispairs, loops of one base and, to a lesser
extent, loops of two bases. A second heterodimer, hMutS[beta], although able to recognise two adjacent unpaired bases, prefers larger
looped structures and binds more effectively to loops of three or four bases.
hMutS[beta] is a heterodimer in which hMSH2 is partnered by hMSH3 (
15
).
Acquired drug resistance by loss of mismatch repair is a feature of N-methyl-N-nitrosourea (MNU) and N-methyl-N'-nitro-N-nitrosoguanidine tolerant
cells. These compounds are analogs of the methylating agents used in clinical
practice. Mismatch repair interacts with one particular methylated base, O
6
-methylguanine (O
6
-meGua) among the numerous DNA methylation products (for review see 16). In particular, binding to O
6
-meGua-containing base pairs by a mismatch recognition complex in cell extracts has
been demonstrated (
17
). These data indicated that O
6
-meGua:T base pairs are recognised quite well by a mismatch recognition
activity [which is now known to be hMutS[alpha] (
18
)] and somewhat better than O
6
-meGua:C base pairs. This recognition is thought to provoke incomplete, and
therefore lethal, repair attempts at the incorrigible O
6
-meGua-containing base pairs. To investigate whether mismatch binding
activities can recognise a common cisplatin DNA adduct, we have used synthetic
oligonucleotides containing a single 1,2 diguanyl cisplatin intrastrand
crosslink. By annealing this oligonucleotide to different complementary
strands, we have investigated binding to sequences that might arise during
replicative bypass of this type of DNA damage. DNA in which the 1,2 diguanyl
crosslink is paired to two complementary cytosines is recognised poorly by
hMutS[alpha]. In contrast, hMutS[alpha] binds preferentially to these intrastrand crosslinks if thymine
is positioned opposite the 3' guanine and cytosine opposite the 5' guanine of the crosslink. Thus, crosslinks that have undergone
promutagenic replication are likely to be recognised by this mismatch binding
complex.
Biochemicals were obtained from Sigma except where stated otherwise. Unmodified
oligonucleotides were synthesized on an Applied Biosystems 380B DNA
synthesizer. Oligonucleotides containing a single 1,2 diguanyl cisplatin
crosslink (top strand shown below), prepared as described in Szymkowski
et al.
(
7
) were a kind gift from Dr John Essigmann, MIT. Platinated strands were end-labeled with T4 DNA polynucleotide kinase (New England BioLabs) and
annealed to 5-fold excess of non-labeled bottom strands. The sequence of the duplex substrates is as
follows:
Abbreviation
5'-TCTTCTTCTA
3'-AGAAGAAGATCCGGAAGAAGAAGA-5'
5'-TCTTCTTCTA
3'-AGAAGAAGATTCGGAAGAAGAAGA-5'
5'-TCTTCTTCTA
3'-AGAAGAAGATCTGGAAGAAGAAGA-5'
5'-TCTTCTTCTA
3'-AGAAGAAGATTTGGAAGAAGAAGA-5'
The cross linked guanines are shown underlined.
Band shift assays were performed as previously reported (
19
). Briefly, cell extract (15-20 [mu]g) was precincubated at room temperature with 40 fmol of matched non-radioactive 34mer in 20 [mu]l reaction buffer comprising 25 mM Hepes
.
KOH, pH 8.0, 0.5 mM EDTA, 0.1 mM ZnCl
2
, 10% glycerol, 50 [mu]g poly(dI
.
dC)poly(dI
.
dC). After 5 min, the reactions were supplemented with
32
P labelled substrate (20 fmol), and incubation continued for a further 20 min.
Aliquots of 10 [mu]l, supplemented with bromophenol blue, were analyzed by electrophoresis on
6% polyacrylamide gels. Reaction products were detected by autoradiography. When non-radioactive competitor oligonucleotides were included, they were present during the preincubation and subsequent
incubation. In experiments to assess efficiency of binding to different
radioactive substrates, equal amounts of radioactivity were used.
The thymidine kinase-deficient subline of the Raji Burkitt's lymphoma was maintained in spinner
culture in RPMI medium containing 5% fetal calf serum (Life Technologies,
Inc.). Exponentially growing cells were harvested by centrifugation. The human
colorectal adenocarcinoma lines LoVo and DLD-1 were grown respectively in Ham's F12 medium or RPMI supplemented by 10%
fetal calf serum. LoVo and DLD-1 cells were detached from the flasks by trypsin-free cell dissociation solution (Sigma) and harvested by
centrifugation. Cell extracts for binding were prepared from fresh or frozen (-80oC) cells as described previously (
19
).
All steps were performed at 0-4oC. The purification was carried out essentially as reported by
Drummond
et al
. (
13
) omitting the final MonoQ step. Extracts were prepared by homogenizing ~5 * 10
9
cells in Buffer A (25 mM Hepes
.
KOH, pH 8.0, 1 mM EDTA, 2 mM [beta]-mercaptoethanol, 0.5 mM spermidine, 0.1 mM spermine). The material which precipitated between 5 and 65% saturated (NH
4
)
2
SO
4
was collected by centrifugation, dissolved in Buffer A and dialysed against
Buffer A for 5 h. After centrifugation (3000
g
) for 10 min to remove precipitated material, the sample was applied to a single-stranded DNA cellulose column (1.8 cm
2
* 5 cm) equilibrated with Buffer A containing 0.1 M NaCl. The material
that passed through the column was reloaded, and the column was washed at a rate of 1.5 ml/min with Buffer A containing 0.2 M NaCl and 2.5 mM MgCl
2
. Protein eluting in a subsequent wash with Buffer A containing 0.2 M NaCl, 2.5
mM MgCl
2
, and 1 mM ATP was retained and loaded onto a Q Sepharose column (0.7 cm
2
* 1 cm, Pharmacia) which was prepared according to the manufacturer's
instructions and equilibrated in Buffer A containing 0.2 M NaCl. After washing
with 2 ml of the same buffer, hMutS[alpha] was obtained by elution with 2 ml of Buffer A containing 0.65 M NaCl.
The fraction was concentrated 10-fold by Microcon 30 (Amicon, Inc., MA, USA) and the concentration of NaCl
reduced to 0.2 M. Small aliquots were snap frozen and stored at -70oC.
Extracts of the Burkitt's lymphoma cell line Raji selectively recognise duplex
oligonucleotides containing a single G
.
T mispair. The mismatch-specific complex formed with the standard 34mer heteroduplex is shown arrowed in Figure
1
a. When 24mer duplexes containing the 1,2 diguanyl cisplatin crosslink were used
as substrates, a band at the position of the G
.
T complex was seen with the Pt-GG.CT substrate in which the complementary strand contained CT opposite
the cross-link (lane 5 from left). This band was not observed with any of the other
three platinated substrates, Pt-GG, Pt-GG.TC and Pt-GG.TT or with the non-platinated matched DNA. A non-specific complex was formed with all the
substrates tested, including the G
.
T mismatch and the perfectly matched 24mer duplex (lane 2). A minor band, which migrated between the G
.
T complex and the non-specific complex, was observed. This minor band was formed with the Pt-GG substrate (lane 3). It was also present with the Pt-GG.CT duplex (lane 5) and binding to this platinated substrate
characteristically produced two bands (see below). Thus, under the conditions
of these experiments, platinated DNA with CT positioned opposite a 1,2 diguanyl
cross-link is recognised more efficiently than similar molecules with other
combinations of pyrimidines in the complementary strand. A faint complex that
migrated about half-way down the gel was also observed with Pt-GG (lane 3). This is a minor activity under our experimental
conditions and may reflect binding by other recognition factors that interact
with cisplatin modified DNA such as RPA (
20
) or one of the previously identified proteins containing the HMG box motif (
21
).
In contrast to wild-type Raji cell extracts, neither extracts of hMSH2-defective LoVo cells nor hMSH6(GTBP)-defective DLD-1 cells bound detectably to Pt-GG.CT. Figure
2
shows that whereas Raij cell extracts recognised G
.
T mispairs, Pt-GG.CT and A
.
C mispairs, LoVo and DLD-1 extracts recognised only A
.
C mispairs. The faint band that is apparent with LoVo and DLD-1 extracts may represent some residual non-specific binding to the platinated oligonucleotide and is unrelated
to G
.
T mismatch binding. The absence of G
.
T mismatch binding by LoVo and DLD-1 extracts confirms previous observations (
14
,
22
,
23
). The poor recognition of the platinated Pt-GG.CT duplex by these extracts is consistent with the involvement of the hMutS[alpha] mismatch recognition complex in binding DNA duplexes containing 1,2 diguanyl crosslinks of this type. The A
.
C mismatch binding activity is known to be independent of both hMSH2 and
hMSH6(GTBP) (
24
) and serves as an internal control for the cell extracts.
hMutS[alpha] was purified extensively from Raji cell extracts by adsorption to single-stranded DNA cellulose and selective elution with ATP (
13
). Two prominent proteins of approximate M
r
= 100 000 and 160 000 together with a number of minor products were detected by Coomassie
staining of an SDS polyacrylamide gel of the purified material (data not shown). The sizes of the major components
are compatible with hMSH2 and hMSH6(GTBP) and we estimate that the hMutS[alpha] preparation was >50% pure. We compared G
.
T mispair and cisplatin adduct recognition by this purified preparation. Non-platinated duplexes containing different complementary strands which introduced G
.
T mispairs were compared to their, otherwise identical, platinated counterparts.
One or two G
.
T mispairs in the non-platinated duplexes stimulated binding to approximately similar extents
(Fig.
3
). In contrast, with platinated substrates which were identical except for the
single 1,2 diguanyl crosslink, only Pt-GG.CT was recognised to a detectable extent. This selectivity reflects the
preference exhibited by unfractionated cell extracts and confirms that only one
of the platinated substrates is recognised to any significant degree. This
recognition is most probably by the hMutS[alpha] mismatch binding complex.
The relative affinity of the purified hMutS[alpha] complex for a G
.
T mispair and 1,2 diguanyl cisplatin cross-link was investigated (Fig.
4
). Substrates that were radioactively labelled to comparable specific activities were mixed with increasing amounts of purified hMutS[alpha]. The partially purified hMutS[alpha] mismatch binding complex recognises this particular platinated
substrate at least as well as a single G
.
T mismatch. Binding to the platinated Pt-GG.CT duplex was detectable at an estimated hMutS[alpha]:DNA ratio of ~1. Binding to the G
.
T mispair in an otherwise identical substrate which did not contain a crosslink
was easily detectable at a hMutS[alpha]:DNA ratio of ~3.5. The Pt-GG substrate in which the complementary strand contained CC
opposite the crosslink, was not detectably bound by ratios of hMutS[alpha]:DNA up to 7. Recognition of the platinated substrate always generated
two resolvable bands in contrast to the single complex observed with a G
.
T mispair. The reason for this behaviour is not clear at present although it may
reflect different extents of hMutS[alpha] loading onto the DNA.
Cells can acquire resistance to drugs, such as the methylating agents MNU by
becoming tolerant to the presence of O
6
-meGua in DNA (
16
). It is generally considered that O
6
-meGua-containing base pairs provoke unsuccessful attempts at mismatch repair that
result in cell death. Methylation tolerance arises as a direct consequence of
the loss of the mismatch repair pathway (
16
). Defects in the mismatch repair proteins hMLH1, hMSH2 and hMSH6(GTBP) have
been identified among tolerant cell lines (for review see
25
) and base pairs involving O
6
-meGua are recognised by mismatch binding factors in cell extracts (
17
) and by a purified hMutS[alpha] complex (
18
).
Cell lines selected for resistance to cisplatin have similar mismatch repair
defects (
9
,
10
). This observation implies that one of the products of DNA platination might
provoke mismatch repair attempts analogous to those at O
6
-meGua. Recognition by mismatch binding complexes is a prerequisite for
such attempts. The experiments reported here indicate that, among four
potential substrates for recognition by mismatch binding activities, a duplex molecule containing a single 1,2 diguanyl cisplatin cross-link in which the complementary strand contains T opposite the 3', and C opposite the 5' crosslinked guanine is highly preferred. A 1,2 diguanyl
crosslink paired to two cytosines was recognised less favorably in our experiments. This is in agreement with other studies which indicate that purified hMutS[alpha] binds to this substrate with about an order of magnitude lower affinity
than to a single G
.
T mispair (
18
). The same substrate is recognised, although rather poorly, by hMSH2 acting
alone (
26
). Platinated DNA is not
per se
a good substrate for this mismatch recognition complex. This is consistent with
previous approaches in which the use of platinated DNA as a probe for possible
recognition factors of cisplatin DNA damage identified a member of the HMG
group of proteins (
21
) and RPA (
20
) but not mismatch binding proteins.
Mismatch repair is a post-replicative correction pathway and a current model for the emergence of
methylation tolerance, and the related cross-tolerance to 6-thioguanine (
27
), invokes replication of adducted bases as a key step (
16
). Our observation of a more favorable interaction of hMutS[alpha] with duplex DNA containing a single 1,2 diguanyl cisplatin crosslink
paired to CT in the complementary strand implies that mismatch correction
attempts might be more likely following replication of cisplatin-adducted DNA. This replication would be potentially mutagenic in that
insertion of thymine opposite the first guanine of the crosslink would be
followed by extension of the daughter strand by incorporation of cytosine.
Replication bypass of cisplatin-DNA lesions has been observed in cell extracts (
28
) and has been inferred in intact cells (
29
). Some cisplatin-resistant cell lines are apparently more able to perform replicative bypass of platinum-DNA adducts than their sensitive parent cells (
30
). The mechanism of this type of trans-lesion DNA replication is unknown. In general, however, bypass of 1,2
diguanyl cisplatin crosslinks has been considered inefficient and only DNA polymerase [beta] appeared capable of significant DNA synthesis opposite adducts of this type (
31
). While the presence of active mismatch repair might explain the poor
translesion synthesis in cells or by replication extracts, the apparent failure
of purified replicative DNA polymerases [delta] and/or [epsilon] to bypass 1,2 diguanyl adducts is more problematical. Recent
evidence indicates, however, that the replicative polymerases are indeed able
to bypass these adducts in structures that resemble replication forks (
32
). Analysis of the products of this type of bypass should define the true
probability of the type of miscoding we postulate.
In addition to potentially lethal intervention by mismatch correction, 1,2
diguanyl adducts that have undergone replication bypass may be recognised by
proteins of the NER pathway. Positioning a mispaired thymine opposite either, or both platinated guanines of a 1,2 crosslink markedly increases the susceptibility of the
crosslink to recognition and removal by NER (
33
). Post-replicative crosslink removal by NER might therefore exert a protective
effect which opposes the lethal processing by mismatch repair. This is
consistent with the relative sensitivity of xeroderma pigmentosum cells to
cisplatin. It seems likely that a cell's sensitivity to cisplatin treatment
will be, at least partly, determined by the relative efficiencies of the
mismatch correction and NER pathways.
The data presented here and in the accompanying paper (
33
) also have implications for the mechanisms of damage recognition by mismatch
repair and NER factors. All three unplatinated oligonucleotides which contained
one or two G
.
T mispairs were efficiently recognised by hMutS[alpha]. The structural alteration introduced by the addition of a 1,2 diguanyl
crosslink reduced recognition by the mismatch binding complex in two of the
three substrates, but apparently stimulated it in the third. A mispaired
thymine opposite either or both crosslinked guanines improves the probability
of recognition by NER factors (
33
). hMutS[alpha] and the NER damage recognition factors, such as XPA, in conjunction with
oligonucleotides containing defined 1,2 diguanyl cisplatin crosslinks might
provide a useful approach to defining the precise structural requirements for effective DNA recognition by mismatch repair and NER proteins.
We are particularly grateful to John Essigmann for making available the
platinated oligonucleotides and we would like to thank him, Rick Wood and
Jonathan Moggs for their comments on the manuscript. We also thank Jean-Sebastien Hoffmann for providing a preprint of his paper. The skilled
assistance of the Oligonucleotide Synthesis and Cell Production laboratories of
the ICRF, Clare Hall is also acknowledged. Y.M. was the recipient of an STA Fellowship and E. O'R. was supported by a Human Capital and Mobility Fellowship from the European Union.
*To whom correspondence should be addressed. Tel: +44 171 269 3870; Fax: +44 171
269 3801; Email: karran@icrf.icnet.uk
+
Present address: Division of Genetics and Mutagenesis, National Institute of
Health Sciences, Tokyo 158, Japan
REFERENCES
Return