X. Peng et al.
The ICLs formed from 2 are stable at high temperatures in
phosphate buffer (908C, pH 7.2, 30 min) and partially stable
in piperidine (1m; 908C, 30 min; Figure S4, lanes 3 and 4, in
the Supporting Information). DNA cleavage bands were ob-
served with all dGs and dAs when the purified ICLs were
treated with piperidine (1m; Figure S4, lanes 4, 11, 18, and
25, in the Supporting Information). Weak cleavage bands
were observed with dCs flanked by dAs and/or dGs. These
results indicated that the alkylations occurred at N7 of dG
and dA, and partially with dC. The stable ICLs observed
after piperidine treatment suggested that N7 is not the only
alkylation site (Figure S4, lanes 4, 11, 18, and 25, in the Sup-
porting Information). It is highly likely that the alkylation
took place at exocyclic amines. Monoalkylation of all gua-
nine and adenine units was also observed with single-strand-
ed DNA 4a or 4b (Figure S4, lanes 2, 9, 16, and 23, in the
Supporting Information). In a control experiment, oligonu-
cleotide 4 was treated with 10 mm, 100 mm, and 1.0 mm H2O2
alone; weak cleavage bands were observed at the purine
and pyrimidine sites (Figure S5 in the Supporting Informa-
tion). This indicated that H2O2 alone induced trace amounts
of hydroxyl radical upon heating, which led to the equal
cleavages with purine and pyrimidine nucleotides.[20]
In an effort to identify the interstrand cross-linking site of
compound 2, we used hydroxyl radical cleavage analysis of
the purified cross-linking product.[20] However, a completely
cleaved ladder similar to the hydroxyl radical control experi-
ment was observed with stronger cleavage bands at the posi-
tions of dGs and dAs (Figure S4, lanes 5, 12, 19, and 26, in
the Supporting Information). This further confirmed that
the interstrand cross-linking should occur at these positions.
However, the exact interstrand cross-linking site and pattern
are still unknown. This is caused by three factors: 1) high re-
activity of the quinone methide generated from 2 upon
H2O2 activation; 2) the complexity of the reaction products,
which include the interstrand cross-linking product, intra-
strand cross-linking and monoalkylated products, and ad-
ducts formed with dG, dA, and dC; 3) the cross-linking and
alkylated products are labile in piperidine.
To understand the difference for ICL formation between
compounds 2 and 1 or 3, their reaction mechanism was in-
vestigated by NMR spectroscopy analysis of the monomer
reactions. The treatment of 1–3 with H2O2 was performed in
deuterated potassium phosphate buffer (10 mm; pH 8.0) as
well as in DMSO/D2O. When phosphate buffer was used as
a solvent, the reaction was too fast to observe any inter-
mediates. Compounds 1–3 were completely consumed
within 5 min and converted to the phenol derivatives 6, 8,
and 10 (Figures S6B, S7B, and S9B in the Supporting Infor-
mation). The formation of 6, 8, and 10 was confirmed by
1H,13C NMR, and HRMS-ESI analysis of the isolated mate-
rials (Figures S21–S27 in the Supporting Information).
Scheme 2. Tandem QM generation and ICL formation induced by 2 upon
H2O2 activation.
phenol derivative 6. Analysis 3 h after addition of H2O2 re-
vealed that compound 2 was completely consumed (Fig-
ure 3C), as evidenced by the absence of diagnostic resonan-
ces at d=7.99 ppm (singlet) and d=4.86 ppm (singlet) cor-
responding to the vinyl and methylene protons of 2, respec-
tively. The resonances corresponding to 2 were replaced by
When the treatment of 2 with H2O2 was carried out in
a mixture of DMSO/D2O, we were able to observe all inter-
mediates formed (Scheme 2 and Figure 3A–D). The integral
ꢀ
ꢀ
ꢀ
ꢀ
change of protons C3 H, C6 H, C2’ H and C5’ H charac-
terized the kinetic transformation of compound 2 into the
3852
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 3850 – 3854