Effect of geometric isomerism in dinuclear platinum antitumour complexes on the
rate of formation and structure of intrastrand adducts with oligonucleotides
Effect of geometric isomerism in dinuclear platinum antitumour complexes on the rate of formation and structure of intrastrand adducts with oligonucleotides
Kirste J.
Mellish
,
Yun
Qu
,
Neel
Scarsdale
1
and
Nicholas
Farrell*
Virginia Commonwealth University, Department of Chemistry, 1001 West Main
Street,
Richmond
, VA 23284-2006,
USA
and
1
Department of Biochemistry and Molecular Biophysics, 1101 East Marshall Street,
Richmond
,
VA 23298-0614,
USA
Received October 31, 1996;
Revised and Accepted January 22, 1997
ABSTRACT
The dinuclear platinum complexes [{
trans
-PtCl (NH
3
)
2
}
2
[mu]
-{NH
2
(CH
2
)
n
NH
2
}](NO
3
)
2
[1,1/t,t (
n
= 4,6)] and [{
cis
-PtCl(NH
3
)
2
}
2
[mu]
-{NH
2
(CH
2
)
n
NH
2
}](NO
3
)
2
[1,1/c,c (
n
= 4,6)] exhibit antitumour activity comparable with cisplatin. 1,1/c,c
complexes do not form 1,2 GG intrastrand adducts, the major adduct of
cisplatin, with double-stranded DNA. This
1
H NMR spectroscopy study shows that, in the absence of a complementary strand,
1,1/c,c (
n
= 4,6) form a 1,2 GG (N7, N7) intrastrand adduct with r(GpG), d(GpG) and
d(TGGT). Initial binding to r(GpG) (and also reaction with GMP) at 37
o
C was slower for 1,1/c,c compared with 1,1/t,t, whereas the second binding step
(adduct closure) was faster for 1,1/c,c. However, the
1
H NMR spectra of the 1,1/c,c adducts at 37
o
C show two H8 signals, one of which is broad and becomes sharper on increasing
the temperature, indicating restricted rotation around the Pt-N7 bond. For the d(GpG)-1,1/c,c (
n
= 4) adduct, 2D NMR spectroscopy assigned the broad H8 signal to the 3
'
G, which has
syn
base orientation and 60% S-type/40% N-type sugar conformation. The 5
'
G has
anti
base orientation and S-type sugar conformation. Apart from the restricted rotation around the 3
'
G, the structure is similar to that of 1,2 GG intrastrand adducts of 1,1/t,t.
This steric hindrance may explain the inability of 1,1/c,c complexes to form
1,2 GG intrastrand adducts with sterically more demanding double-stranded DNA.
INTRODUCTION
The requirement for novel platinum-based anticancer agents that are active in tumours resistant to the clinically important anticancer
drug cisplatin (
cis
-[PtCl
2
(NH
3
)
2
]) led to the synthesis of dinuclear bisplatinum complexes consisting of two platinum coordination spheres linked by
a variable length diamine chain (
1
). One series of bisplatinum complexes is the bifunctional complexes of general formula [{PtCl(NH
3
)
2
}
2
[mu]-{NH
2
(CH
2
)
n
NH
2
}]
2+
, which contain two monofunctional coordination spheres with the single chloride
leaving group on each platinum either
cis
(1,1/c,c) or
trans
(1,1/t,t) to the straight chain aliphatic diamine linker (Fig.
1
).
Cisplatin exerts its cytotoxicity by binding covalently to DNA at the N7
position of guanine and, to a lesser extent, adenine bases to form bifunctional adducts. The 1,2-intrastrand adducts between adjacent guanines or adjacent adenine and guanine account for 50-65% and 20-30% of total platinum bound to DNA respectively; 1,3-intrastrand adducts between two guanines separated by
one base, 1,2-interstrand cross-links between two guanines on opposite strands and monofunctional
adducts account for the remainder (
2
,
3
). The 1,2-intrastrand adducts produce a rigid bend of 30-35o directed toward the major groove (
4
-
6
) and a localized unwinding of the double helix of 13o (
7
). The 1,3-intrastrand adduct produces a bend of similar magnitude (
5
) and unwinding of 23o (
7
) and also causes local denaturation of the double helix (
8
). The 1,2-interstrand cross-link produces a rigid bend of 20-45o directed toward the minor groove (
9
,
10
) and unwinding of 80-90o (
9
,
10
). Cisplatin-induced DNA damage causes inhibition of DNA replication (
11
) and transcription (
12
) and induction of programmed cell death (apoptosis) (
13
).
An important mechanism of both intrinsic (
14
,
15
) and acquired (
16
-
18
) tumour cell resistance to cisplatin is an increased ability to repair or
tolerate cisplatin-DNA adducts. The rationale behind the development of bisplatinum
complexes was that their novel structure would result in DNA adducts and
conformational distortions distinct from those observed with cisplatin and potentially provide a mechanism to overcome cisplatin resistance. This is in contrast to direct structural analogues of cisplatin. Carboplatin,
which differs from cisplatin only in the substitution of chloride leaving
groups for the 1,1-cyclobutanedicarboxylato ligand, forms identical DNA adducts to cisplatin
(
19
-
21
) and shows the same spectrum of antitumour activity (
22
). Compounds in which the amine ligands of cisplatin have been replaced with 1,2-diaminocyclohexane (
23
,
24
) or amine/cyclohexylamine (
25
,
26
) also form the same types of DNA adducts as cisplatin.
MATERIALS AND METHODS
Starting materials
GMP (guanosine 5'-monophosphate), r(GpG) [guanylyl (3' -> 5') guanosine], d(GpG) [2'-deoxyguanylyl (3' -> 5') 2'-deoxyguanosine]
and
N
-acetylcysteine were obtained from Sigma. d(TGGT) [thymidylyl (3' -> 5') 2'-deoxyguanylyl (3' -> 5') 2'-deoxyguanylyl (3' -> 5') thymidine] was synthesized by the cyanoethyl phosphoramidite method using an Expedite 8909 DNA synthesizer (Perseptive Biosystems) and reagents from Glen Research (Sterling
VA). The dinuclear platinum complexes [{
trans
-PtCl(NH
3
)
2
}
2
[mu]- {NH
2
(CH
2
)
n
NH
2
}](NO
3
)
2
[1,1/t,t (
n
= 4,6)] (
34
,
35
) and [{
cis
-PtCl(NH
3
)
2
}
2
[mu]-{NH
2
(CH
2
)
n
NH
2
}](NO
3
)
2
[1,1/c,c (
n
= 4,6)] (Schaaff,T.G., Qu,Y., Farrell,N. and Wysocki,V.H., unpublished results)
were synthesized as described previously.
1D
1
H NMR spectroscopy
The reactions of 1,1/c,c and 1,1/t,t complexes (
n
= 4,6) (3-5 mM in 99.996% D
2
O) with GMP (1:2) or r(GpG) (1:1) were monitored over 24 h at 37oC by
1
H NMR spectroscopy using a Varian Gemini 300 MHz spectrometer, typically using 80o pulses, a spectral width of 15 p.p.m., an acquisition time of 1.78 s, a relaxation delay of 0 s and 64 scans. Spectra were referenced to
internal 5,5-dimethylsilapentane sulfonate (DSS). Reactions were quantitated by
integration of H8 signals.
1
H NMR spectra were also measured at 37oC for the final products of the 1:1 reaction (24 h at 37oC) of 1,1/c,c and 1,1/t,t complexes (
n
= 4,6) (4-5 mM in D
2
O) with d(GpG) or d(TGGT). The effect of temperature on the
1
H NMR spectra of the reaction products of 1,1/c,c complexes (
n
= 4,6) with d(GpG) and d(TGGT) was investigated over the temperature range 12-62oC. The pH dependence of the H8 chemical shifts of representative
reaction products was assessed by adding trace amounts of NaOD (4% in D
2
O) or DCl (2% in D
2
O). The reaction at 37oC of the 1,1/c,c (
n
= 6)-d(GpG) and 1,1/t,t (
n
= 6)-d(GpG) final products with
N
-acetylcysteine (12 mM) was monitored over 5 days by
1
H NMR spectroscopy.
2D NMR spectroscopy
The 1,1/c,c (
n
= 4)-d(GpG) reaction product was lyophilized and then dissolved in 99.999% D
2
O for 2D NMR measurements.
1
H-
1
H dqf-COSY (double quantum filtered correlated spectroscopy) (
36
), TOCSY (total correlated spectroscopy) (
37
) and ROESY (rotating frame nuclear Overhauser effect spectroscopy) (
38
) and
31
P-
1
H HMQC (heteronuclear multiple quantum coherence correlated spectroscopy) (
39
) experiments were acquired on a Varian Unity Plus 500 MHz NMR spectrometer at 32oC. The spectral frequency was 500 MHz for
1
H and 202 MHz for
31
P. Pulsed field gradients were used for coherence selection in both the dqf-COSY and HMQC experiments. Quadrature detection was achieved in the
indirectly detected frequency dimension using the States method (
40
) in the dqf-COSY, TOCSY and ROESY experiments and using the sensitivity enhanced
method of Rance (
41
) in the HMQC experiment.
1
H NMR spectra were referenced to internal DSS and
31
P NMR spectra to external trimethyl phosphate (TMP; 1 M in D
2
O). The dqf-COSY experiment consisted of 512 complex t
1
points. Each t
1
point consisted of 1024 complex points over a 7489 Hz sweep width averaged for
64 transients with a 1.6 s repetition rate. The TOCSY experiment was acquired
with a 50 ms mixing time using an MLEV-17 pulse sequence with a B
1
field strength of 8.3 kHz for spin locking and consisted of 200 complex t
1
points. Each t
1
point consisted of 1024 complex points over a 7485 Hz sweep width averaged for
128 transients with a 1.6 s repetition rate. The ROESY experiment was acquired
with a 400 ms mixing time using a 3 kHz B
1
field for spin locking and consisted of 256 complex t
1
points. Each t
1
point consisted of 1024 complex points over a 7489 Hz sweep width averaged for 64 transients with a 1.6 s repetition rate. The
31
P-
1
H HMQC experiment consisted of 64 complex t
1
points over a 5000 Hz
31
P sweep width. Each t
1
point consisted of 1024 complex points over a 7486 Hz
1
H sweep width averaged for 2048 transients with a 1 s repetition rate. The delay
for the evolution of multiple quantum coherence was set to 20 ms to allow for
the detection of multibond
31
P,
1
H couplings. The data sets were processed for phase sensitive display with Felix
2.3 (Biosym/Molecular Simulations Inc.) on a Silicon Graphics Indigo II
workstation. For the dqf-COSY, TOCSY and ROESY experiments, sinebell weighting functions and zero
filling were used in both dimensions to yield a 2048 * 2048 data matrix. For the HMQC experiment, a combination of cosine bell
and kaiser weighting functions were used with zero filling in both dimensions
to yield a 2048 * 512 data matrix.
RESULTS AND DISCUSSION
In order to explain the differences in binding to DNA observed between 1,1/c,c
and 1,1/t,t complexes, firstly the kinetics of binding of 1,1/c,c and 1,1/t,t (
n
= 4,6) to GMP and r(GpG) were compared. Secondly, 1D and 2D
1
H NMR spectroscopy studies were carried out on the adducts of 1,1/c,c (
n
= 4,6) with d(GpG) and d(TGGT) to obtain information on their structure
relative to that previously described for the 1,1/t,t isomers (
33
,
42
)
Kinetics of reaction with GMP
The reaction between 1,1/c,c and 1,1/t,t (
n
= 4,6) complexes (5 mM) and GMP (10 mM) was followed at 37oC by
1
H NMR spectroscopy as shown in Figure
2
. For all four complexes the
1
H NMR spectra show the stepwise substitution of chloride by GMP with formation of the intermediate Cl-Pt-Pt-GMP ([{PtCl(NH
3
)
2
} [mu]-{NH
2
(CH
2
)
n
NH
2
}-{Pt(GMP)(NH
3
)
2
}]
+
, species I) and conversion to the final product GMP-Pt-Pt-GMP ([{Pt(GMP) (NH
3
)
2
}[mu]-{NH
2
(CH
2
)
n
NH
2
}-{Pt(GMP)(NH
3
)
2
}], species II) as observed previously for the 1,1/t,t (
n
= 4) case (
43
). The H8 and H1' chemical shifts of species I and II are given in Table
1
. For the
n
= 4 complexes a minor H8 peak (species III) was also observed at 8.84 p.p.m.
(1,1/c,c) or 8.86 p.p.m. (1,1/t,t) as the reaction proceeds; this may represent
a hydrolysed intermediate, i.e. GMP-Pt-Pt-H
2
O ([{Pt(H
2
O)-(NH
3
)
2
}[mu]-{NH
2
(CH
2
)
n
NH
2
}- {Pt(GMP)(NH
3
)
2
}]
2+
) (
43
).
Kinetics of reaction with r(GpG)
Due to the limited availability of d(GpG), r(GpG) was used for kinetic studies;
previous studies with cisplatin have shown that the nature of the reaction and
the final product is equivalent for r(GpG) and d(GpG) (
45
).
Figure
3
shows the time course of the reactions of 1,1/c,c and 1,1/t,t (
n
= 4,6) complexes (3 mM) with r(GpG) (1:1) at 37oC followed by
1
H NMR spectroscopy and quantitated by integration of H8 signals. The reaction proceeds via the formation of two intermediates
(i.e. platinum bound at either the 5' or 3' guanine) followed by binding of the second platinum at the second
guanine to give the final product, the 1,2-intrastrand adduct (
42
). The H8 and H1' chemical shifts of the final product and the half-times of the two reaction steps are shown in Table
1
.
.
Base and H1' chemical shifts of the adducts of d(GpG) and d(TGGT) with bisplatinum complexes (37oC, pH 6.0-6.7)
[delta] H8 (p.p.m.)
[delta] H6 (p.p.m.)
[delta] H1' (p.p.m.)
d(GpG)
8.00, 7.76
6.16, 6.02
d(GpG)-1,1/c,c (
n
= 4)
8.60, 8.52
a
6.35
d(GpG)-1,1/t,t (
n
= 4)
8.63, 8.61
6.41, 6.30
d(GpG)-1,1/c,c (
n
= 6)
8.61, 8.60
a
6.37
d(GpG)-1,1/t,t (
n
= 6)
8.57
6.39, 6.20
d(TGGT)
7.95, 7.85
7.53, 7.40
6.21, 6.11, 6.04, 5.89
d(TGGT)-1,1/c,c (
n
= 4)
8.64, 8.51
a
7.69, 7.66
6.4-6.2
d(TGGT)-1,1/t,t (
n
= 4)
b
8.70, 8.63
7.69, 7.56
6.32, 6.32, 6.06, 6.06
d(TGGT)-1,1/c,c (
n
= 6)
8.69, 8.56
a
7.70, 7.66
6.4-6.2
d(TGGT)-1,1/t,t (
n
= 6)
b
8.62, 8.60
7.71, 7.57
6.32, 6.27, 6.07, 6.07
a
Broad peak.
b
From Kaspàrkovà
et al
. (33).
General features of reaction products
The H8 and H1' chemical shifts of the products of the reaction of the bisplatinum complexes with GMP and r(GpG) are summarized in Table
1
. The base and H1' chemical shifts of the products of the reaction of bisplatinum complexes
with d(GpG) (5 mM) or d(TGGT) (4 mM) (1:1 ratio, 24 h at 37oC) are shown in Table
2
. As with r(GpG), the major product of the reactions between bisplatinum complexes and d(GpG) or d(TGGT) was the 1,2 intrastrand GG adduct. The small amount of side products could be due to dimer and polymer
formation, as discussed previously (
42
). Figure
4
shows the base and H1' region of the
1
H NMR spectra of the 1,1/c,c adducts in comparison with free d(GpG) and d(TGGT).
The H8 signals of all reaction products were shifted 0.5-0.9 p.p.m. downfield from the free oligonucleotides, which is indicative of platination at guanine N7 (
43
). This was confirmed by pH titration of the H8 chemical shifts (data not shown); no change was
observed in the chemical shift around pH 2-4, the region of N7 protonation.
.
1
H chemical shifts (p.p.m.) for the d(GpG)-1,1/c,c (
n
= 4) adduct (32oC, pH 6.5)
H8
H1'
H2'
H2''
H3'
H4'
H5', H5''
(CH
2
)
4
linker
d(GpG)-1,1/c,c (
n
= 4)
2.30, 1.46
5' G(1)
8.58
6.33
2.77
2.71
4.96
4.23
3.75, 3.66
3' G(2)
8.50
a
6.36
2.76
2.57
4.72
4.27
4.21, 4.10
a
Broad peak.
Structural analysis of the d(GpG)/1,1/c,c (
n
= 4) reaction product
The non-exchangeable protons of the d(GpG)-1,1/c,c (
n
= 4) adduct were assigned by 2D NMR spectroscopy. The protons of the two sugar
rings and the diamine chain were identified by
1
H-
1
H TOCSY and dqf-COSY, which show through-bond coupling. H2' and H2'' were distinguished by
1
H-
1
H ROESY, which shows through-space interactions (NOEs); the NOE between H1' and H2'' is more intense than the NOE between H1' and H2' (
48
). The H8 base protons were assigned to their respective sugar rings by ROESY.
The residues were identified as 5' or 3' by
31
P-
1
H HMQC, as only H3' of the 5' residue and H5'/H5'' of the 3' residue are coupled to
31
P.
Reaction of the d(GpG)-1,1/c,c (
n
= 6) and d(GpG)-1,1/t,t (
n
= 6) reaction products with
N
-acetylcysteine
The d(GpG)-1,1/c,c (
n
= 6) and d(GpG)-1,1/t,t (
n
= 6) adducts were reacted with an excess of
N
-acetylcysteine to investigate whether they are stable to attack by sulphur
nucleophiles. No change was observed in the
1
H NMR spectra of the adducts after up to 5 days incubation with
N
-acetylcysteine at 37oC (data not shown), indicating that no reaction occurs.
CONCLUSIONS
The rate of reaction with GMP and initial binding to r(GpG) is slower for
1,1/c,c complexes compared with 1,1/t,t complexes. 1,1/c,c complexes are able
to form 1,2 intrastrand adducts with r(GpG), d(GpG) and d(TGGT) and the second
binding step with r(GpG) (i.e. intrastrand adduct closure) is faster for
1,1/c,c complexes than for 1,1/t,t complexes. The 1,1/c,c (
n
= 4)-d(GpG) adduct shows similar structural features to the previously
described adducts of 1,1/t,t (
n
= 6) with d(GpG) (
42
) and d(TGGT) (
33
), i.e. the 3' G has
syn
base orientation and a higher percentage N-type sugar conformation than the 5' G, which has
anti
base orientation. This is distinct from 1,2 intrastrand adducts of cisplatin,
in which the base and sugar conformations are: 5' G,
anti
/N-type; 3' G,
anti
/S-type (
51
,
52
). However, unlike 1,1/t,t complexes, the 1,2 intrastrand adducts of 1,1/c,c
complexes with r(GpG), d(GpG) and d(TGGT) show restricted rotation around one
of the guanines; this was identified as the 3' G for the 1,1/c,c (
n
= 4)-d(GpG) adduct. The other 1,1/c,c adducts are likely to have a similar
structure. Therefore, the greater steric demand of double-stranded DNA is likely to favour the formation by 1,1/c,c complexes of
interstrand cross-links between two 5' guanines, rather than 1,2 intrastrand adducts. Alternatively, if
formed, the 1,2 intrastrand adduct may isomerize to an interstrand cross-link. Isomerization of the 1,2 intrastrand adduct of cisplatin (
53
) and the 1,3 intrastrand adduct of transplatin (
54
) to an interstrand cross-link has been observed previously. In both cases the Pt-N7 bond of the 3' G was ruptured. In cells, rearrangement of the 1,1/c,c 1,2 intrastrand adduct to a DNA-protein or DNA-glutathione cross-link is also possible, however, no
evidence for this was obtained as no reaction was observed between the d(GpG)- 1,1/c,c (
n
= 6) adduct and
N
-acetylcysteine, as is also the case for the d(GpG)-1,1/t,t (
n
= 6) adduct. Whether this would still be the case for a larger DNA sequence
remains to be examined. In summary, the presence of steric hindrance at the 3' G is the main difference between single-stranded 1,2 GG intrastrand adducts of 1,1/c,c and 1,1/t,t
complexes. This observation is consistent with the high levels of interstrand
cross-links and lack of 1,2 GG intrastrand adducts observed previously for 1,1/c,c complexes with double-stranded DNA (
28
).
ACKNOWLEDGEMENT
This work was supported by grant DHP-2E from the American Cancer Society.
2 Fichtinger-Schepman,A.M.J., Van Der Veer,J.L., Den Hartog,J.H.J., Lohman,P.H.M and Reedijk,J. (1985) Biochemistry, 24, 707-713.
3 Fichtinger-Schepman,A.M.J., Van Der Velde-Visser,S.D., Van Dijk-Knijenberg,H.C.M., Van Oosterom,A.T., Baan,R.A. and Berends,F. (1990) Cancer Res., 50, 7887-7894.
4 Rice,J.A., Crothers,D.M., Pinto,A.L. and Lippard,S.J. (1988) Proc. Natl. Acad. Sci. USA, 85, 4158-4161.MEDLINE Abstract
5 Bellon,S.F. and Lippard,S.J. (1990) Biophys. Chem., 35, 179-188.MEDLINE Abstract
46 Bloemink,M.J., Reedijk,J., Farrell,N., Qu,Y. and Stetsenko,A.I. (1992) J. Chem. Soc. Chem. Commun., 1002-1003.
47 Den Hartog,J.H.J., Altona,C., Van Den Marel,G.A. and Reedijk,J. (1985) Eur. J. Biochem., 147, 371-379.
48 Wijmenga,S.S., Mooren,M.M.W. and Hilbers,C.W. (1993) In Roberts,G.C.K. (ed.), NMR of Macromolecules-A Practical Approach. IRL Press, Oxford, UK, pp. 217-288.
49 Patel,D.J., Kozlowski,S.A., Nordheim,A. and Rich,A. (1982) Proc. Natl. Acad. Sci. USA, 79, 1413-1417.MEDLINE Abstract
50 Rinkel,L.J. and Altona,C. (1987) J. Biomol. Struct. Dynam., 4, 621-649.
51 Admiraal,G., Van Der Veer,J.L., De Graaf,R.A.G., Den Hartog,J.H.J. and Reedijk,J. (1987) J. Am. Chem. Soc., 109, 592-594.
52 Den Hartog,J.H.J., Altona,C., Chottard,J.-C., Girault,J.-P., Lallemand,J.-Y., De Leeuw,F.A.A.M., Marcelis,A.T.M. and Reedijk,J. (1982) Nucleic Acids Res., 10, 4715-4730.
53 Yang,D., Van Boom,S.S.G.E., Reedijk,J., Van Boom,J.H. and Wang,A.H.-J. (1995) Biochemistry, 34, 12912-12920.MEDLINE Abstract
