Topoisomerase Poisons Bound at a DNA Junction-like QuadruplexTopoisomerase Poisons Bound at a DNA Junction-like QuadruplexJames H. Thorpe, Jeanette R. Hobbs, Susana C. M. Teixeira and Christine J. CardinJames H. Thorpe, Jeanette R. Hobbs, Susana C. M. Teixeira and Christine J. Cardin

The University of Reading Chemistry DepartmentThe University of Reading Chemistry Department

IntroductionIntroduction

Topoisomerases are responsible for the interconversion of the topological states of DNA generating transient single or double strand breaks in the DNA phosphodiester backbone to allow passage of one (topo I) or two (topo II) DNA strands during replication, resulting in an associated relaxation of their supercoiled state. For this reason the topoisomerases have long been an attractive target in the fight against cancer.

A series of third generation anti-tumour agents have been developed in New Zealand to target malignant melanomas amongst other forms of tumour, with some inducing DNA cleavage in the presence of both topoisomerase I and II making them part of only a small range of drugs to exhibit such properties and a major step forward in the fight against cancer.

We have previously show that the related 9-amino - acridine carboxamides, which are known to inhibit only topoisomerase II at the point of DNA cleaving through catalytic inhibition, can intercalate into duplex DNA with the carboxamide side chain located in the major groove binding to the N7/O6 of the adjacent guanine.

Here we present the first examples of a range of cytotoxic agents bound at DNA junction-like quadruplex.

ExperimentalExperimental

Crystals of the duplex d[CG(5BrU)ACG]2 with 9-bromo-phenazine-1-carboxamide in the orthorhombic space group C222 were grown by vapour diffusion over a period of ~12 months. MAD data collection was carried out at the DESY synchrotron facility, where several wavelengths were measured to maximise the anomalous and dispersive signals. Data processing was carried out with mosflm and SCALA. Two heavy atom positions were determined by SOLVE and

phase calculations carried out with MLPHARE and DM. The initial model was built ab initio into the 2Å MAD map with the map fitting program XFIT from the XTALVIEW program suite. Crystallographic refinement was carried out using SHELX97 with SHELXPRO used for map production.

ResultsResults

The structure of the duplex d[CGTACG]2 bound to a range of topo poisons exhibits an unusual intercalation site and helical fraying (Figure 1) stabilised through disordered modes of drug binding and an array of cations.

The intercalation cavity is formed by the strand exchange of a cytosine base rotated to pair with a guanine of a symmetry related helix at ~90° and a distance of ~15Å generating a pseudo-Holliday type junction (Figure 2).

Of the cytotoxic agents shown to bind at this junction (Figure 4) the acridine based drugs have thus far only stabilised this strained system with the aid of a Mg2+ ion, bound at the centre of the cross-over, to the four cytosine phosphates, a feature which is thus far absent for the phenazines.

Two such cavities are linked through a quadruplex formed by the minor groove interactions of the N2/N3 guanine cavity sites (Figure 3) at an angle of ~40°, effectively

ConclusionsConclusions

DNA junctions have been known for some time to provide high-affinity binding sites for intercalators and as this work suggests may help stabilise the X-stacked forms through charge neutralisation from their cationic side chains, as does Mg2+. Resolution of these junctions can occur by resolvase enzymes but also some eukaryotic topo I enzymes. Human topo II has also been shown to preferentially bind to such sites. The stabilising effects exhibited by this class of compound therefore illustrates their potential to bind with DNA

Three views of the 2Å MAD map used to build the initial model. Green = 1 and yellow = 2.Figure 2.Figure 2. Two views of the four way helical junction and drug cavity formed by the DNA cross-

over and quadruplex. For clarity the bound drugs have been removed and cobalt positions are shown in purple.

doubling the size of the intercalation site and allowing for a large degree of disordered binding of the drug chromophore despite the carboxamide side chain anchoring to the N7/O6 guanine positions in the major groove.

The other end of the duplex exhibits a terminal base fraying in the presence of Co2+ ions linking symmetry related guanine bases (see MAD pictures) intertwined through the minor groove, yielding a quasi-continuous stack. A second hydrated cobalt ion is bound to the pre-ultimate guanine G8 and linked through water sites to its symmetry related partner at ~7.5Å.

N

X

YO NH

(CH2)zMe2N

(1) 9Br-phenazine (X=N, Y=Br, Z=2)(2) DACA (X=CH, Y=H, Z=2)(3) 9-aminoDACA (X=NH2, Y=H, Z=2)(4) DACA3 (X=CH, Y=H, Z=3)(5) Bis-DACA (X=9-aminooctylDACA, Y=H, Z=2) junctions as well as duplex DNA and and even

strand-nicked DNA (‘hemi-intercalated’) as in the cleavable complex, suggesting a structural basis for the dual poisoning of topoisomerase I and II by this family of drugs.

It must be noted however that we only obtained crystals in the presence of Co2+ ions, and although they do not directly influence the helix cross-over and intercalation site, they are essential for this junction formation.Further WorkFurther Work

This work suggests a possible explanation for the dual poisoning properties of certain anti-tumour agents, and as an extension of these studies, work looking at the formation of the cleavable complex between this family of cytotoxins and the topoisomerases has begun.Reference: Reference: Biochem., 49, 15055-15061, 2000

Acknowledgements: W.A.Denny (University of Auckland, NZ); P. Charlton (Xenova plc); DESY synchrotron and staff; A.K. Todd (Institute of Cancer Research, London); A. Adams (Trinity College, Dublin).

Table 1: Data Collection, phasing and reflection Statistics.Inflection Point Whiteline Max. Remote 1 Remote 2

BrUa Xtalb BrUa

(Å) 0.9208 0.9201 0.9192 0.9000 1.1000Resolution Range (Å) 26.6-2.0 26.6-2.0 26.6-2.0 26.6-2.0 25.9-2.25Unique Reflections 2215 2211 2218 2372 1454Completeness (%)c 98.7 (91.0) 98.9 (92.5) 98.9 (92.5) 99.5 (93.3) 89.7 (93.6)Multiplicity 3.4 3.4 3.4 3.4 2.4R-mergec,d 8.0 (30.8) 8.0 (30.8) 7.6 (32.5) 8.9 (36.3) 7.1 (25.6)R-merge (anom)c,e 7.2 (30.1) 7.2 (30.1) 6.9 (31.4) 7.3 (34.1) 6.9 (20.0)anom data comp (%)c,f 98.0 (87.8) 98.0 (87.4) 98.3 (89.0) 98.7 (94.3) 73.9 (78.3)R-Cullisg (acc/cent) 0.95/0.96 0.98/1.01 0.80/0.62 0.77/0.60Phasing Powerh (acc/cent) 0.48/0.46 0.28/0.28 1.32/1.50 1.60/1.71Reflections (acc/cent) 1748/451 1750/453 1755/457 1749/448 1171/269DNA atoms 229Water Molecules 33R-factori 18.58R-freej 27.62_____________________________________________________________________________________________________a Fluorescence spectrum measured from the powdered Br-U sample. b Fluorescence spectrum measured from the cryo-cooled crystal sample. c Values in parenthesescorrespond to the outermost resolution shell. dR-merge = hkl

I|It(hkl) – {I(hkl)}| / hkl

IIt(hkl), calculated for the whole data set. e R-merge (anom) = hkl

I|It(hkl) –

{I(hkl)}| / hkl

IIt(hkl). f Percentage of reflections with a bijvoet pair. g R-Cullis is the mean residual lack of closure error divided by the dispersive difference. h

Phasing power = rms (|FH|/E), where FH is the heavy atom amplitude and E is the residual lack of closure error. I R-factor = hkl||Fo(hkl) – k|Fc(hkl)||/hkl|Fo(hkl)|. j R-factor for 10% of the reflections not used in the refinement.

Figure 1Figure 1

Schematic view of the numbering and labelling scheme for the X-stacked junction and bound cobalt ions.

Figure 4Figure 4

The structural formulae of the tricyclic drug systems which have currently been shown to help stabilise this unusual X-stacked DNA junction.Figure 3.Figure 3. Two illustrations of the DNA quadruplex forming the large intercalation cavity. (a)

A stereoview of the cavity with drugs removed for clarity. (b) A 2Å sigma-A map showing the floor of the cavity and the minor groove interactions of the guanine N2/N3 positions.