The duplex form of the self-complementary oligonucleotide, 5'd(A(1)T(2)G(3)G(4)G(5)T(6)A(7)C(8)C-(9)C(10)A(11)T(12)), was treated with cis-Pt(A2)Cl2 anticancer drugs(A2=(NH3)2 or en(ethylenediamine)). Previous NMR reports of the Pt(en) reaction revealed one hairpin-like product with platination at two of G(4), G(5), or A(7) but not at G(3). To resolve these issues, 3'-P-32-end-labeling of reaction mixtures was examined. Gel electrophoresis gave one pronounced product band (autoradiographic detection) with hairpin-like mobility for both drugs. However, in conflict with the NMR studies, this product has a G(3),G(4) intrastrand crosslink (DNA sequencing methods). Furthermore, with each drug, 5'-P-32-end-labeling and gel electrophoresis gave two significant comparable hairpinlike product bands: one for the G(3),G(4) crosslinked adduct identified as the exclusive product by 3'-P-32-end-labeling and the second for a G(4),G(5) crosslinked adduct consistent with NMR studies. To resolve these issues, the Pt(en) reaction was subjected to exhaustive 5'-end-labeling with nonradioactive ATP. The G(3),G(4) adduct identified as the exclusive product by 3'-P-32-end-labeling was found by UV-visualization to constitute only 4% of the product. The major product (96%) was the G(4),G(5) crosslinked adduct. From gel electrophoresis under denaturing conditions, the G(3),G(4) adduct exists in the denatured state to a much greater extent than the G(4),G(5) adduct. Clearly, during both types of enzymatic labeling, the minor product (probably as the denatured form) was labeled at much higher efficiency, suggesting that caution should be exercised in interpreting the increasingly widely used P-32/gel electrophoresis methods. In a GGG sequence, the N7 of the central residue, G(4), is the most nucleophilic site, and we found that the monofunctional complex, [chloro(diethylenetriamine)platinum(II)] chloride, preferentially attacks this site. A(7) binding was shown to be insignificant both by the diethyl pyrocarbonate reaction and by enzymatic digestion, which reveals only G,G crosslinking. These results suggest that an initial G(4) monoadduct forms a 1,2 crosslink to the 3'-end G(5) much more favorably than to the 5'-end G(3) and that 1,2 G,G crosslinking is much more favorable than 1,4 G,A crosslinking in either direction. Relatively stable hairpins such as the G(4),G(5) adduct described here could explain features of the P-31 NMR spectrum observed on treating polymeric DNA with Pt anticancer drugs and could stabilize cruciforms in palindromic regions. The latter possibility is discussed in the light of the recent discoveries on the structure-specific recognition protein and its partial sequence homology with the high mobility group protein-1, a species known to recognize cruciforms [Bruhn, S. L.; Pil, P. M.; Essigman, J. M.; Housman, D. E.; Lippard, S. J. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 2307. Pil, P. M.; Lippard, S. J. Science 1992, 256, 234].
