Triostin C

The crystal structures of three quinoxaline antibiotics-echinomycin 2QN, triostin C and the C222(1) form of triostin A--have been determined, and the structure of the P2(1)2(1)2(1) form of triostin A has been re-refined against our previously reported data. The molecular conformations are compared with those deduced from NMR data and those reported for two complexes of triostin A with oligonucleotides. Although the depsipeptide ring conformations are basically similar, the effective twofold molecular symmetry is violated by the folding of one of the quinoxaline chromophores in echinomycin 2QN and by a rotation of one of the ester planes with the formation of an intramolecular hydrogen bond in triostin C. In the oligonucleotide complexes of triostin A the chirality of the disulfide bridge is inverted. The alanine NH groups are involved in intermolecular hydrogen bonds in all four structures, and (except in echinomycin 2QN) the stacking of the chromophores in the crystal emulates the intercalation involved in DNA complex formation. In echinomycin 2QN, the antibiotic molecules are hydrogen bonded to form a helix along the crystallographic 6(5) screw axes, with a channel of disordered solvent running through the middle of the helix. Crystal data: (1), echinomycin 2QN, C53H66N10O12S2.2.5(C3H6O).2.5(H2O), M(r) = 1289.5, hexagonal, P6(5), a = b = 22.196(15), c = 24.64 (2) A, V = 10,513 (13) A3, Z = 6, Dx = 1.222 Mg m-3, lambda (Cu K alpha) = 1.5418 A, mu = 1.275 mm-1, T = 193 K, R = 9.0% for 4828 I > 2 sigma (I) and 11.8% for all 7102 unique reflections; (2), triostin C, C54H70N12O12S2.0.67(CHCl3).0.67(H2O), M(r) = 1234.2, orthorhombic, P2(1)2(1)2(1), a = 16.054 (8), b = 17.128 (9), c = 22.706 (12) A, V = 6244 (6) A3, Z = 4, Dx = 1.313 Mg m-3, lambda (Mo K alpha) = 0.71073 A, mu = 0.239 mm-1, T = 188 K, R = 7.7% for 4678 I > 2 sigma (I) and 14.0% for all 7260 unique reflections; (3), triostin A, C50H62N12O12S2.2(C7H14O2), M(r) = 1347.6, orthorhombic, P2(1)2(1)2(1), a = 20.94 (2), b = 18.53 (2), c = 18.80 (2) A, V = 7292 (13) A3, Z = 4, Dx = 1.228 Mg m-3, lambda (Cu K alpha) = 1.5418 A, mu = 1.245 mm-1, T = 293 K, R = 6.8% for 2116 I > 2 sigma (I) and 9.3% for all 2928 unique reflections; (4), triostin A, C50H62N12O12S2.HCl.2(C3H7NO), M(r) = 1269.9, monoclinic, C222(1), a = 10.622 (10), b = 17.035 (17), c = 35.21 (3) A, V = 6371 (10) A3, Z = 4, Dx = 1.324 Mg m-3, lambda (Mo K alpha) = 0.71073 A, mu = 0.199 mm-1, T = 153 K, R = 7.5% for 2164 I > 2 sigma (I) and 13.2% for all 3402 unique reflections. Extensive use was made of restraints on the geometrical and displacement parameters in the successful anisotropic refinement of these structures against weak data.

The natural compounds triostin and thiocoraline are potent antitumor agents that act as DNA bisintercalators. From a pharmaceutical point of view, these compounds are highly attractive although they present a low pharmacokinetic profile, in part due to their low solubility. Synthetically, they represent a tour de force because no robust strategies have been developed to access a broad range of these bicyclic (depsi)peptides in a straightforward manner. Here we describe solid-phase strategies to synthesize new bisintercalators, such as thiocoraline-triostin hybrids, as well as analogues bearing soluble tags. Orthogonal protection schemes (up to five from: Fmoc, Boc Alloc, pNZ, o-NBS, and Troc), together with the right concourse of the coupling reagents (HOSu, HOBt, HOAt, Oxyma, EDC, DIPCDI, PyAOP, PyBOP, HATU, COMU), were crucial to establish the synthetic plan. In vitro studies and structure-activity relationships have been shown trends in the structure-activity relationship that will facilitate the design of new bisintercalators.

The story of echinomycin is of an antibiotic whose anti-cancer activity was rediscovered thanks to scientific investigation of its mode of action at the molecular level. It was the first DNA bis-intercalator identified (in 1974). Molecular models for echinomycin and its congeners are now well-founded on crystallographic data. These are beginning to throw light on significant variations in conformation which affect the ability of the antibiotics to recognise specific nucleotide sequences in DNA. Kinetic and other physical experiments have revealed much about the stability and selectivity of antibiotic-DNA complexes: stacking forces are important, as are hydrogen bonding interactions. Echinomycin preferentially recognises sites in DNA containing the CpG step, but other sites of lower occupancy exist and the antibiotic can migrate between them. Currently a good deal of attention is being paid to the suggestion that echinomycin might change Watson-Crick base pairs to a Hoogsteen form.