On the structural basis and design guidelines for type II topoisomerase-targeting anticancer drugs

Type II topoisomerases (Top2s) alter DNA topology via the formation of an enzyme–DNA adduct termed cleavage complex, which harbors a transient double-strand break in one DNA to allow the passage of another. Agents targeting human Top2s are clinically active anticancer drugs whose trapping of Top2-mediated DNA breakage effectively induces genome fragmentation and cell death. To understand the structural basis of this drug action, we previously determined the structure of human Top2 β-isoform forming a cleavage complex with the drug etoposide and DNA, and described the insertion of drug into DNA cleavage site and drug-induced decoupling of catalytic groups. By developing a post-crystallization drug replacement procedure that simplifies structural characterization of drug-stabilized cleavage complexes, we have extended the analysis toward other structurally distinct drugs, m-AMSA and mitoxantrone. Besides the expected drug intercalation, a switch in ribose puckering in the 3′-nucleotide of the cleavage site was robustly observed in the new structures, representing a new mechanism for trapping the Top2 cleavage complex. Analysis of drug-binding modes and the conformational landscapes of the drug-binding pockets provide rationalization of the drugs’ structural-activity relationships and explain why Top2 mutants exhibit differential effects toward each drug. Drug design guidelines were proposed to facilitate the development of isoform-specific Top2-targeting anticancer agents.


Preparation of the hTop2 core -DNA-doxorubicin ternary complex crystal by post-crystallization drug replacement
To obtain the hTop2 core cleavage complex stabilized by doxorubicin, etoposide was soaked out by transferring the hTop2 core -DNA-etoposide crystals (1) into a substitute mother liquor containing 30% MPD for 16 hours.
Doxorubicin was then introduced by adding 1 mM drug (in DMSO) to the drop containing drug-free crystals. After 16 hours of soaking, crystals were looped and flash-frozen in liquid nitrogen for data collection.

Structure determination
In each deposited structure, missing residues are as follows: For the parameters listed in table 1, all non-glycine residues were included for the Ramachandran analysis.
Residues which fells in the disallowed region of the Ramachandran plot include: S794 and G868 of chain A and V852 of both chain in hTop2 core -DNA-m-AMSA structure; D1101 of chain A and V852 of both chains in hTop2 core -DNA-mitoxantrone structure; D540 of chain A and D1101 of chain B in hTop2 core -DNA-ametantrone structure; D1101 of both chains in hTop2 core -DNA structure. Although are outliers of Ramachandran plot, these residues fit very well to the corresponding electron density.

Doxorubicin may stabilize the Top2 cleavage complex in a new quaternary conformation.
Doxorubicin and other structurally related anthracycline derivatives represent a group of drugs commonly used to treat various types of cancers, including leukemias, lymphomas, and breast, uterine, ovarian, and lung cancer (2). To understand how doxorubicin interacts and stabilizes Top2cc, we introduced doxorubicin into the drug-free hTop2 core -DNA binary complex crystals and the structure was determined at 2.7 Å resolution (Table S1).
Surprisingly, this structure revealed a new drug binding mode distinct from those observed in the etoposide-, m-AMSA-and mitoxantrone-bound structures. Rather than binding to both cleavage sites, a single doxorubicin molecule binds asymmetrically within the four-base stagger between the two cleavage sites ( Figure S6), which disrupts the usually highly preserved two-fold symmetry associated with Top2 structures. The aglycone moiety of doxorubicin intercalates between the +1/+4 and +2/+3 base pairs and adopts a skewed orientation relative to the two phosphoribosyl backbones while the amino sugar and hydroxymethyl ketone moieties rest in the minor groove.
Because there are very few protein-mediated contacts ( Figure S6C), the drug's conformation is stabilized mainly by its interactions with DNA. Notably, instead of being sandwiched between two Watson-Crick base pairs, doxorubicin stacks against a Watson-Crick base pair (+2/+3) on one side and a reverse Hoogsteen base pair (+1/+4) on the other.
A large repositioning of the +4 adenine base accompanied by a change in ribose ring-puckering likely favors the formation of this non-canonical base pair.
The preference for having the amino sugar and hydroxymethyl ketone groups located in the minor groove, as seen in the structures of doxorubicin-DNA binary complexes (3), may explain why doxorubicin does not bind at the cleavage site where the minor groove is approached by Top2. In the absence of additional quaternary structure changes, the minor groove binding pocket is not spacious enough to accommodate the two appended drug moieties.
The protrusion of the amino sugar moiety toward the distal cleavage site prevents the binding of a second doxorubicin molecule within the four-base stagger because the simultaneous presence of two drug molecules at both symmetry-related base pair steps would cause steric clashes between the two amino sugar groups.
Previous studies have revealed a strong Top2 DNA cleavage preference for the −1 nucleotide to be an adenine in the present of doxorubicin, which indicates that doxorubicin would also target the cleavage site for function (4). In addition, as the bound doxorubicin is mainly stabilized by its interactions with DNA with very few protein-mediated contacts, the new binding mode fails to explain the reported SARs of anthracyclines (2). Thus, we need to be extra cautious about the pharmacological relevance of this structure, and we suspected that the potential "off-site" binding of doxorubicin may be due to steric repulsion between the drug's bulky minor groove-protruding moieties and the flanking residues, considering that the quaternary conformation of Top2 is constrained by the crystal packing.
Therefore, despite the usefulness of the soaking method in illustrating the binding modes of m-AMSA and mitoxantrone, this technique may not be applicable to compounds such as doxorubicin, due to their bulky groups facing DNA minor groove that can't be accommodated by the protein-embraced DNA cleavage site. And we suspect that doxorubicin would trap the cleavage complex in a more opened conformation due to a wider separation of catalytic groups in the presence of this relatively bulkier drug. With this potential limitation in mind, we suggest that the reported soaking procedure for drug exchange would be most suitable for those Top2-targeting agents whose binding (at the cleavage site) can be accommodated by local side-chain rearrangements. Figure S1. The electron density maps of the bound drugs in the hTop2 core -DNA-m-AMSA and     Ihkl is the measured intensity for any given reflection. c R cryst = (Σ||F o | -k|F c ||)/(Σ|F o |). R free = R cryst for a randomly selected subset (5%) of the data that were not used for minimization of the crystallographic residual. d Categories were defined by PHENIX (5). All non-glycine residues are included for this anlaysis. Although D1101 of both chains in the structure fell in the disallowed region of the Ramachandran plot, these residues fit very well to the corresponding electron density.