Binding of Anthracyclines Within DNA–Topoisomerase ΙΙ Cleavage Complexes: Thermodynamics and Interactions

Abstract

All-atom molecular dynamics simulations of anthracycline–DNA–topoisomerase II (TOPII) cleavage complexes are undertaken in an aqueous environment, with particular emphasis on their thermodynamic description, using the double-decoupling method. Both the TOPIIα and TOPIIβ isoforms are considered herein, with doxorubicin and berubicin, two TOPII poisons, as the anthracyclines of interest. From a thermodynamic point of view, we find that all complexes investigated exhibit more negative standard binding Gibbs energies than berubicin–DNA intercalation complexes, showing that the presence of TOPIIα or TOPIIβ further stabilizes the complexes. The berubicin–DNA–TOPIIα complex exhibits the most negative standard binding Gibbs energy. Therefore, berubicin shows a preference for TOPIIα over TOPIIβ; the former is highly expressed in proliferating cells, including cancer cells, whereas the latter is expressed ubiquitously. From a nanoscopic point of view, our rigorous in silico approach sheds light on the interactions of the two anthracyclines with both DNA and TOPII and on how they contribute to the hindering of religation. Moreover, this study provides a detailed comparison between doxorubicin and berubicin in terms of relevant properties, such as contact numbers, hydrogen bonds, selected distances, and solvent-accessible surface areas.

1. Introduction

Anthracyclines constitute a class of anticancer drugs employed for the treatment of various types of cancer, such as leukemias and breast cancer. Two of the prototype molecules of this class, which are widely used in clinical practice, are doxorubicin and daunorubicin. Another widely used anthracycline is epirubicin. In general, anthracyclines cannot be administered for malignant tumors in the brain, since they poorly cross the blood–brain barrier (BBB) , and thus cannot reach therapeutic concentrations in the central nervous system. Berubicin, which is a novel anthracycline in clinical trials, crosses the BBB − and is being tested against glioblastoma multiforme, , one of the most aggressive brain cancers and a major challenge in oncology. In the clinical trials, berubicin was compared to lomustine, a chemotherapy medication used for the treatment of brain tumors among other cancers. It is concluded from the primary analysis that berubicin shows clinically similar outcomes (overall survival) in comparison with lomustine. In addition, the safety profile of berubicin continues to be favorable, including the absence of cardiotoxicity, which is a severe side effect of anthracyclines in clinical use (e.g., doxorubicin). ,,

Anthracyclines intercalate between DNA base pairs with their planar tricyclic systems. In addition, the minor groove participates in the further stabilization of the noncovalent anthracycline–DNA complexes, as reported in the literature. The planar tricyclic system of anthracyclines, together with a nonplanar ring, forms a tetracyclic chromophore that constitutes the structural backbone of these drugs. Furthermore, a side chain and a sugar moiety are covalently attached to the tetracyclic chromophore. Berubicin differs from doxorubicin only in its sugar moiety, which is O-benzylated. Anthracycline’s intercalation prevents the normal action of topoisomerase II (TOPII), which is an enzyme playing a significant role in the replication and mitosis processes. Two isoforms of TOPII are topoisomerase IIα (TOPIIα) and topoisomerase IIβ (TOPIIβ); the former is highly expressed in proliferating cells, including cancer cells, whereas the latter is expressed ubiquitously. − The aforementioned elevated expression makes TOPIIα a primary target of major anticancer drugs, ,− including anthracyclines such as doxorubicin. However, anthracyclines generally target TOPIIβ as well, resulting in cardiotoxicity, among other side effects. ,− It is also noted that anthracyclines are regarded as topoisomerase poisons rather than topoisomerase inhibitors, since they stabilize the DNA–topoisomerase cleavage complex.

In this article, berubicin and doxorubicin are investigated in silico within DNA–TOPIIα and DNA–TOPIIβ cleavage complexes in an aqueous environment at an appropriate salt concentration. Since only cleavage complexes are studied herein, the term “cleavage” is mostly omitted unless required to avoid confusion. As far as berubicin is concerned, this study is a critical continuation of the research presented in refs and , which investigated berubicin–DNA complexes with a special focus on the thermodynamics and kinetics. To the best of our knowledge, no detailed computational studies focusing on berubicin exist other than the aforementioned ones and the present work. Our findings in refs and show that berubicin, when noncovalently complexed with double-stranded DNA, exhibits more favorable binding Gibbs energies and longer deintercalation times in comparison to doxorubicin and other anthracyclines (e.g., daunorubicin). Deintercalation takes place through three distinct stages in which the transition from the intercalated to the reshuffling state is the rate-limiting step of the mechanism and the complementary role of the minor groove is highlighted. In refs and , thermodynamic properties were calculated with two different rigorous computational methods: the double-decoupling method − (DDM) and well-tempered metadynamics , (WT-MetaD). The latter method, by invoking transition state theory, allows the calculation of significant kinetic properties as well, such as rate constants and deintercalation times. , DDM, which is based on a properly designed molecular thermodynamic cycle, is a method of high computational cost due to the large number of molecular simulations needed. However, DDM is considered one of the most detailed methods to calculate absolute binding free energies for complex biomolecular systems, such as those examined herein. − In addition, it is more rigorous than other widely employed methods, such as molecular mechanics Poisson-Boltzmann surface-area (MM-PBSA) and molecular mechanics generalized-Born surface-area (MM-GBSA).

The in silico approach presented herein is based on atomistic molecular dynamics (MD) simulations on the time scale of hundreds of nanoseconds. MD is a powerful computational method for tracking the time evolution of chemical and biomolecular systems; , it may be envisioned as a computational microscope shedding light on the structure and dynamics of the simulated systems. The MD simulations conducted enable a comprehensive analysis of berubicin and doxorubicin within DNA–topoisomerase complexes, probing various properties of these ternary systems mainly from a nanoscopic point of view. We also discuss our findings herein with respect to the analysis presented in refs , ; the present comparative study investigates the selected anthracyclines as intercalators, as well as the role of the minor groove as an additional stabilizing factor, in the presence and absence of TOPIIα and TOPIIβ. Furthermore, the results are analyzed in light of published studies concerning drugs bound within DNA–topoisomerase complexes. In this article, we also pay particular attention to the rigorous thermodynamic description of the simulated systems using DDM. In general, the thermodynamic analysis is carried out in the same fashion as in ref . To be more specific, standard binding Gibbs energies for both anthracyclines within the DNA–topoisomerase complexes are calculated, as well as their decomposition into Lennard-Jones and electrostatic contributions. Emphasis here is placed on structure and thermodynamics. A study of binding and unbinding kinetics would be highly informative, but lies outside the scope of the present article.

This article is structured as follows. Section provides a detailed description of the models and methods employed for the in silico study of the DNA–TOPII and anthracycline–DNA–TOPII complexes under consideration herein. In Section , we present the results of the molecular simulations and discuss them in detail, focusing particularly on the thermodynamic description of the examined systems. The last section recaps the primary conclusions derived from the presented research. In addition, further details and results are available in the Supporting Information.

2. Models and Methods

2.1. System Preparation

Both berubicin and doxorubicin are solely considered in their protonated state since these chemical species are predominant at a pH of 7.2–7.6. The molecular structures of these two anthracyclines are depicted in Figure , in which specific groups of atoms used in our analysis are highlighted. Table presents all systems investigated in this article, showing that the study encompasses both binary and ternary systems in which berubicin and doxorubicin are noncovalently bound within the DNA–TOPII covalent cleavage complexes. The structures of DNA–TOPIIα and DNA–TOPIIβ are obtained from the Protein Data Bank under the codes 5GWK and 3QX3, respectively, and the DNA sequence is 5′-d­(AGCCGAGC|TGCA|GCTCGGCT)-3′, where | stands for the cleavage sites. It should be pointed out that each DNA strand contains only one cleavage site. These structures contain the cleavage core of the topoisomerase isoforms, which is the target domain for anthracyclines. It is noteworthy that from a structural point of view, TOPII is a homodimer consisting of two identical protomers, within each of which a binding site for the anthracycline is formed. Etoposide, a topoisomerase poison used as a chemotherapy agent, is bound within the aforementioned DNA sequence. The advantage of using ternary complexes as a starting point in this in silico study is that there are two pre-formed binding sites near the cleavage nicks. As a consequence, we can directly dock any inhibitor molecules after removing the two bound etoposide molecules, as described below in this section. This approach has also been adopted by other published studies concerning inhibitor–DNA–TOPII complexes, including those with bound anthracyclines. , It should be noted that the two binding sites are considered equivalent; this is primarily attributed to the same nucleotide sequence of each binding site, as well as the adjacent amino acids of TOPIIα and TOPIIβ. In addition, the two binding sites act in a nonconcerted fashion with respect to the cleavage and religation reactions. , In the downloaded structures from the Protein Data Bank, there are missing residues, located far from the cleavage sites, which generate unphysical N- and C-termini. These artificial termini are properly capped with acetyl (ACE)/N-methylamide (NME) using Maestro Schrödinger and the tLEaP program of AMBER14. Regarding the initial position of berubicin and doxorubicin molecules within the DNA–TOPII complexes, we carry out docking simulations which are fully described in Section S1 of the Supporting Information; both anthracyclines are placed at the two binding sites where the etoposide molecules were located.

1.

Molecular structures of (a) berubicin and (b) doxorubicin.

1. Systems Examined in This Study .

(1) DNA–TOPIIα

(2) DNA–TOPIIβ

(3) berubicin–DNA–TOPIIα

(4) doxorubicin–DNA–TOPIIα

(5) berubicin–DNA–TOPIIβ

(6) doxorubicin–DNA–TOPIIβ

^aTwo anthracycline molecules are intercalated in each ternary complex.

2.2. Molecular Dynamics Simulations in Aqueous Solutions

All complexes under study are solvated with approximately 75,000–80,000 water molecules using the tLeap utility of the AMBER14 package. As far as the selected force fields are concerned, topoisomerases, DNA and anthracyclines are described by the AMBER ff14SB, AMBER parmBSC1 and the generalized AMBER force field (GAFF), respectively, while the transferable intermolecular potential three-point (TIP3P) model is employed for the water molecules. Atomic partial charges and topology files are derived via the antechamber and tLeap modules of AMBER14. More specifically, the atomic partial charges for both berubicin and doxorubicin are derived using the Austin model 1bond charge correction (AM1-BCC) method. The parameters for the ions are obtained from the 12-6-4 set of Li and Merz. In addition to the magnesium ions present in the downloaded structures (crystallographic Mg²⁺), extra sodium and chloride ions are added to the water phase to electrostatically neutralize the system, as well as to set the salt concentration to 0.01 M. Initially, the systems are energy-minimized employing the conjugate-gradient method, first with positional restraints on the atoms of TOPII, DNA and anthracyclines using a harmonic constant of 10 kcal/mol/Ų, and subsequently without any positional restraint applied. Then, the energy-minimized structures are subjected to a three-step heating procedure using MD in the NVT ensemble to gradually reach a temperature of 310 K, starting from 0 K. The Langevin thermostat is employed with a collision frequency of 1.0 ps^–1. All heating steps are conducted for 400 ps each with positional restraints applied to the same atoms as in the energy minimization procedure. More specifically, the harmonic constants are set to 10, 5, and 2 kcal/mol/Ų for the first, second and third heating steps, respectively. The final equilibration step is an MD simulation in the NPT ensemble for 20 ns, performed without any positional restraint. The protocol described above is applied using the AMBER14 package; however, all production runs are carried out using the GROMACS package (version 2024.0), due to its superior computational performance on the multi-node CPU-based HPC infrastructure available to us. To this end, all topology and coordinate files are converted from AMBER to GROMACS format via the acpype script. Each production MD simulation is performed in the NPT ensemble for 0.6 μs. The adequacy of the simulation time is justified by the evolution of the root-mean-square deviations (RMSDs) of DNA–TOPII non-hydrogen atoms, as detailed below. The Nosé–Hoover thermostat , and the Parrinello–Rahman barostat are employed to maintain the temperature at 310 K and the pressure at 1 bar, respectively. The integration time step of the equations of motion is set to 2.0 fs and all bonds involving hydrogens are constrained to their equilibrium length using the LINCS algorithm. Concerning the long-range electrostatics, they are treated with the particle mesh Ewald method, while the Lennard-Jones interactions are calculated applying a cutoff radius of 1.2 nm. The latter interactions are smoothly switched to zero from 1.0 nm to the above-mentioned cutoff distance using the force-switch option of GROMACS. Finally, VMD software is employed for the molecular graphics presented in this article and the Supporting Information.

2.3. Analysis of the Systems

The analysis of the molecular simulations presented in Section is based on several properties, such as RMSDs, solvent accessible surface areas (SASAs) and the number of hydrogen bonds formed between selected atoms or groups of atoms. The GROMACS package contains computational tools for the calculation of the aforementioned properties (hbond, sasa , and rms utilities). In general, the properties involving the binding sites are averaged over the two existing in each ternary system, unless otherwise stated. The hydrogen-bond analysis is based only on standard geometric criteria: the distance between the donor and acceptor must be less than 0.35 nm and the angle among donor, hydrogen and acceptor must be less than 30°. As for thermodynamics, the analysis is based solely on the DDM method that is implemented for the calculation of the binding free energies, which are Gibbs energies, since all molecular simulations are conducted in the NPT ensemble. Due to the presence of two binding sites in the DNA–TOPII complexes examined, the supramolecular reactions shown in eqs and take place

$${\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}_{(\mathrm{a}\mathrm{q})}+\mathrm{D}\mathrm{N}\mathrm{A}-{\mathrm{T}\mathrm{O}\mathrm{P}\mathrm{I}\mathrm{I}}_{(\mathrm{a}\mathrm{q})}\rightleftharpoons \mathrm{a}\mathrm{n}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}(1)-\mathrm{D}\mathrm{N}\mathrm{A}-{\mathrm{T}\mathrm{O}\mathrm{P}\mathrm{I}\mathrm{I}}_{(\mathrm{a}\mathrm{q})}$$ 1

$${\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}_{(\mathrm{a}\mathrm{q})}+\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}(1)-\mathrm{D}\mathrm{N}\mathrm{A}-{\mathrm{T}\mathrm{O}\mathrm{P}\mathrm{I}\mathrm{I}}_{(\mathrm{a}\mathrm{q})}\rightleftharpoons \mathrm{a}\mathrm{n}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}(2)-\mathrm{D}\mathrm{N}\mathrm{A}-{\mathrm{T}\mathrm{O}\mathrm{P}\mathrm{I}\mathrm{I}}_{(\mathrm{a}\mathrm{q})}$$ 2

Regarding our thermodynamic analysis, only the standard binding Gibbs energy of the reaction (ΔG _bind ) of eq is estimated using DDM; owing to the high computational cost of this method we cannot study both reactions from a thermodynamic point of view. Full details about DDM and how it is properly implemented may be found elsewhere. This article provides only an overview of DDM, covering the main points. The two-leg thermodynamic cycle employed for these thermodynamic calculations is depicted in Figure . Within the DDM framework, the thermodynamic cycle is mathematically described by eq

$$\mathrm{\Delta }{G}_{\mathrm{b}\mathrm{i}\mathrm{n}\mathrm{d}}^{\mathrm{o}}\equiv \mathrm{\Delta }{G}_{2\to 1}=G_1-G_2=-(\mathrm{\Delta }{G}_{1\to 3}+\mathrm{\Delta }{G}_{3\to 4}+\mathrm{\Delta }{G}_{4\to 5}+\mathrm{\Delta }{G}_{5\to 2})$$ 3

2.

Molecular thermodynamic cycle used to calculate binding Gibbs energies through the DDM method. The DNA–TOPII cleavage complex is visualized with ribbons, while berubicin is shown with all its atoms. When the drug is fully colored, all interactions between berubicin and DNA–TOPII are turned on, whereas when it is black-colored, the non-bonded interactions are turned off. The yellow spring indicates the harmonic restraint applied between berubicin and DNA–TOPII. States (1) and (2) are the fully coupled bound and unbound end-states, respectively. The transition between them corresponds to the binding Gibbs energy. In the context of DDM, three auxiliary alchemical states are defined. State (3) is the bound state in which a harmonic restraint is applied between the DNA–TOPII cleavage complex and berubicin. State (4) is the bound state in which the restraint is applied and berubicin is decoupled. Finally, state (5) is the unbound state in which berubicin is still decoupled.

The notation used is that of ref . $\mathrm{\Delta }{G}_{i\to j}=G_j-G_i$ stands for the Gibbs energy change for the transition from state (i) to state (j). $\mathrm{\Delta }{G}_{1\to 3}$ is the Gibbs energy change for turning on the harmonic distance restraint between the anthracycline molecule and the DNA within the bound ternary system (anthracycline–DNA–TOPII) in an aqueous environment. The next term ( $\mathrm{\Delta }{G}_{3\to 4}$ ) represents the Gibbs energy change of switching off the restrained drug from its surroundings by deactivating both the Lennard-Jones and electrostatic interactions of the drug with all other molecules. Each decoupling process is governed by a coupling parameter, λ. When the interactions are completely turned on, λ equals zero, while the interactions are fully turned off when λ is set to unity. Intermediate values of λ correspond to gradually reduced interactions. The third term ( $\mathrm{\Delta }{G}_{4\to 5}$ ) concerns switching off the harmonic distance restraint, with the interactions of the anthracyclines being still turned off. Finally, the last term in eq , $\mathrm{\Delta }{G}_{5\to 2}$ , is the Gibbs energy change of coupling the unbound drug to the water phase by turning on its nonbonded interactions. The states involved are schematically depicted and explained in Figure . It is noted that $\mathrm{\Delta }{G}_{5\to 2}$ and $\mathrm{\Delta }{G}_{3\to 4}$ are calculated by employing the Bennett acceptance ratio (BAR). In the case of charged ligands, such as those used in this study, the standard binding Gibbs energy calculated by eq is corrected according to eq

$$\mathrm{\Delta }{G}_{\mathrm{b}\mathrm{i}\mathrm{n}\mathrm{d},\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}}^{◦}=\mathrm{\Delta }{G}_{\mathrm{b}\mathrm{i}\mathrm{n}\mathrm{d}}^{◦}+\mathrm{\Delta }\mathrm{\Delta }{G}_{\mathrm{b}\mathrm{i}\mathrm{n}\mathrm{d}}^{\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}}$$ 4

This correction term addresses electrostatic artifacts originating from the periodic boundary conditions used in the simulated systems. More details about eq are provided elsewhere. All standard binding Gibbs energies discussed and reported hereafter correspond to ΔG _bind,corr and are simply referred to as ΔG _bind . More details about the implementation of DDM in anthracycline–DNA complexes are given in the Supporting Information 2 of ref (Section S2 therein). Before closing this section, it should be stated that the binding Gibbs energy of the doxorubicin–DNA–TOPIIβ complex is not calculated, due to the substantial computational cost associated with the implementation of DDM, and thus only three binding Gibbs energies are reported in Section .

3. Results and Discussion

Prior to analyzing and discussing the results, we validate that all the trajectories of the production MD simulations are stable and have converged. To this end, the RMSDs of DNA and TOPII non-hydrogen atoms of all systems are calculated with respect to the corresponding configurations downloaded from the Protein Data Bank. The RMSDs are presented in Figure , where all curves reach a plateau in the region of 0.3–0.4 nm. In the analysis below, we place particular emphasis on DNA–TOPIIα complexes, since these are the targets for the anthracyclines. As a consequence, some tables and figures concerning the anthracycline–DNA–TOPIIβ complexes are provided in the Supporting Information and are discussed herein.

3.

RMSDs of DNA–TOPII non-hydrogen atoms in all systems under consideration as a function of the simulation time.

3.1. Anthracycline–DNA Intercalative Mode of Interaction

As discussed above, anthracyclines are classified as DNA intercalators and therefore the detailed investigation of intercalation is necessary. In Figures and S1, representative configurations of the DNA sequence are depicted for all complexes, as generated from a cluster analysis of the corresponding MD trajectories. In all anthracycline–DNA–TOPII complexes, the two intercalation sites are sufficiently widened to accommodate the drug molecules, whereas no such expansion is observed between the corresponding successive nucleotides in anthracycline-free DNA–TOPII complexes, as shown in Figures and S1. In each ternary complex, the anthraquinone groups of berubicin and doxorubicin are perpendicularly inserted between the successive base pairs constituting the two intercalation sites. To be more specific, the anthraquinone groups protrude toward the neighboring major groove of the DNA sequence, whereas the sugar moieties lie toward the neighboring minor groove. The geometry of the intercalated anthracycline–DNA complexes in an aqueous environment is generally in agreement with that detailed in refs and , in which TOPII is absent and only berubicin is considered. Moreover, such a geometry for anthracycline–DNA–TOPII complexes is reported elsewhere. For a quantitative comparison between DNA and DNA–TOPIIα complexes, all with anthracyclines intercalated within them, we conduct an analysis based on selected properties presented in Table . For completeness, these properties are provided for DNA–TOPIIβ in the Supporting Information (Table S1). The helical rise, which is an important structural property of DNA, approximately doubles at the intercalation base pairs with respect to the anthracycline-free DNA–TOPII complexes. This increase stabilizes the DNA–TOPII cleavage complexes, preventing religation, as discussed later in detail. In the present study, an analysis based on contact numbers is also carried out. More precisely, a contact between two groups of atoms (e.g., molecule, residue) is formed when the distance between an atom belonging to one group and an atom belonging to the other group is less than 0.3 nm. As seen in Tables and S1, the contact number between anthracycline and DNA, CN_DNA–anthra, does not differ in berubicin–DNA and berubicin–DNA–TOPII complexes, indicating that this property is not affected by the presence of TOPII. However, the contact number is markedly smaller for doxorubicin–DNA–TOPII complexes. Berubicin forms more contacts with DNA than doxorubicin due to the additional stabilization through the minor groove that is attributed to the hydrophobic substitution in the sugar moiety, as reported in refs and . In general, anthracyclines form a hydrogen bond network when bound within DNA, as also observed for berubicin herein and in refs and (and references therein). The last row in Tables and S1 presents the average number of hydrogen bonds, HB_DNA‑anthra, formed between anthracycline and DNA in the complexes under consideration, as well as in the TOPII-free berubicin–DNA complexes investigated in refs and . In the case of DNA–TOPIIα complexes, both berubicin and doxorubicin form the same number of hydrogen bonds, while HB_DNA‑anthra slightly drops in the doxorubicin–DNA–TOPIIβ complex. Moreover, the number of hydrogen bonds does not change between the TOPII-free DNA and DNA–TOPII complexes for berubicin, further confirming that the intercalation is successful in the binary as well as in the ternary complexes.

4.

Representative configurations of DNA for the following systems are shown: (a) DNA–TOPIIα, (b) berubicin–DNA–TOPIIα, and (c) doxorubicin–DNA–TOPIIα. TOPIIα and the water molecules are omitted for clarity, whereas the anthracycline molecules are rendered in green. Note the nicks, indicated by red circles, in the DNA configurations due to the covalent linkage of DNA with TOPIIα.

2. Selected Properties Concerning the Anthracycline–DNA and anthracycline–DNA–Topiiα Complexes in an Aqueous Environment.

	berubicin–DNA	berubicin–DNA–TOPIIα	doxorubicin–DNA–TOPIIα
helical rise [nm]	0.73 ± 0.01	0.64 ± 0.03	0.62 ± 0.05
CNDNA‑anthra	50 ± 5	49 ± 3	38 ± 1
HBDNA‑anthra	1.2 ± 0.3	1.2 ± 0.1	1.2 ± 0.2

^aThe values are averaged over the two intercalation complexes studied in refs and .

^bHelical rise is defined as the distance between the two consecutive base pairs constituting the intercalation site along the helical axis of the DNA sequence, calculated via the 3DNA 2.0 web server. , For comparison, the helical rise of the anthracycline-free DNA–TOPIIα is (0.31 ± 0.01) nm.

3.2. Identification of Active Site Residues and Indications towards Selectivity

Intercalation alone is not sufficient for the poisoning of TOPII; − the drugs also interact directly with the nuclear enzyme. , It is known that, in general, drug–protein interactions contribute crucially to molecular recognition and the stability of the binding configuration, thereby governing the underlying molecular mechanism. To effectively investigate the drug–protein interactions, a standard practice is to focus on those residues constituting the binding pocket, instead of the whole protein. − This approach is adopted in this section for the study of the TOPII-containing complexes.

In Figures and S2, the main domains of the DNA binding and cleavage core (DBCC) , are visualized using the berubicin–DNA–TOPII complexes as representative examples. The topoisomerase-primase (TOPRIM) and winged-helix (WHD) domains, along with DNA, form the binding sites for the anthracyclines, as reported in the related literature concerning intercalating TOPII poisons in general. From a biological point of view, this is why the most commonly observed mutations in both TOPII isoforms are found in the aforementioned domains. , In more detail, the TOPRIM domain forms most of the contacts with berubicin and doxorubicin, while the WHD domain contributes to a lesser extent, as shown below. Near the binding sites, the TOPRIM domain forms contacts with DNA through the minor groove, whereas the WHD domain forms contacts with DNA through the major groove. It is noted that, as reported in Section , the nonintercalated parts of berubicin and doxorubicin mostly lie toward the minor groove and consequently interact with the TOPRIM domain. We provide representative configurations of all the ternary cleavage complexes in Figures and S3–S5. The active site residues, which form the contact network, are clustered into three groups according to their average contact numbers with the anthracyclines during the simulations. The classification criterion is as follows: primary, secondary, and tertiary active site residues have contact numbers above 1.0, between 1.0 and 0.1, and below 0.1, respectively; graphical information about them is found in Figures and S3–S5. Average contact numbers between active site residues and the anthracycline molecules, CN_res‑anthra, are summarized in Table ; it is noted that only primary active site residues are considered due to their higher interaction strength. These results confirm that the anthracyclines mainly interact with the TOPRIM domain, as mentioned at the beginning of the paragraph, and secondarily with WHD.

5.

DBCC of TOPIIα in the case of berubicin–DNA–TOPIIα complex. The water molecules are omitted for clarity. Protein chains are represented as ribbons and the residue numbering corresponds to that of the downloaded files from the Protein Data Bank. The two berubicin molecules are depicted in black. (a) Overview of the DBCC domains and (b) zoomed-in view of the DBCC domains along with the DNA sequence (yellow color) near the binding sites.

6.

Representative configuration of the berubicin–DNA–TOPIIα complex with water molecules omitted for clarity. TOPIIα, DNA and berubicin are depicted in red, blue and black, respectively. (a) Overview of the complex including both binding sites. (b) Zoomed-in view of a single binding site with the active site residues being highlighted. The primary, secondary and tertiary contacts are depicted in red, orange and yellow spheres, respectively. The amino acid residues participating in the primary and secondary contacts are reported on the right.

3. Average Contact Numbers of the Primary Active Site Residues with the Bound Anthracyclines.

CNres‑anthra	berubicin–DNA–TOPIIα	doxorubicin–DNA–TOPIIα	berubicin–DNA–TOPIIβ	doxorubicin–DNA–TOPIIβ
ARG487(α)/ARG503(β)	10.4 ± 0.9	10.0 ± 1.0	12.1 ± 1.0	8.4 ± 1.1
GLU461(α)/GLU477(β)	2.9 ± 0.5	1.9 ± 1.3	2.1 ± 0.5	1.4 ± 0.5
GLY462(α)/GLY478(β)	2.0 ± 0.3	0.6 ± 0.6	1.2 ± 0.4	0.3 ± 0.1
LYS440(α)/LYS456(β)	1.7 ± 0.2	0.8 ± 0.2	1.0 ± 0.5	<0.1
ASP463(α)/ASP479(β)	2.6 ± 0.4	1.5 ± 1.5	0.9 ± 0.4	<0.1
GLY488(α)/GLY504(β)	3.5 ± 0.6	1.4 ± 0.4	1.2 ± 0.2	2.0 ± 0.4
MET762(α)/GLN778(β)	3.3 ± 0.2	0.4 ± 0.1	1.5 ± 0.4	1.9 ± 0.6
GLU506(α)/GLU522(β)	0.7 ± 0.3	4.2 ± 0.2	<0.1	2.0 ± 0.7
ALA505(α)/ALA521(β)	0.6 ± 0.4	0.1 ± 0.1	0.1 ± 0.1	<0.1
MET765(α)/MET781(β)	<0.1	<0.1	<0.1	1.1 ± 0.1
MET766(α)/MET782(β)	0.7 ± 0.1	<0.1	0.2 ± 0.1	1.1 ± 0.1

^aThe two residues in each entry of the first column are selected such that the residue in TOPIIα is the corresponding residue in TOPIIβ.

Key properties concerning the anthracycline–TOPII interactions in the bound state are presented in Table . Comparing the two anthracyclines under consideration, we conclude from the table in question that berubicin forms more contacts with both the TOPII isoforms than doxorubicin, a finding that is fully consistent with their sizes. To be more specific, the substitution in the sugar moiety of berubicin not only enables the drug to form more contacts with the same active site residues of TOPII, but also extends its contact network as seen on the right of Figures and S3–S5. From the contacts mentioned above, only a small fraction leads to the formation of hydrogen bonds, as evidenced by HB_{TOPII‑anthra} in Table . Among all the active site residues, ARG487­(α) and ARG503­(β) are particularly noteworthy because they play a crucial role in anchoring berubicin and doxorubicin within the binding sites, forming the most contacts with the drug molecules in all ternary complexes (the indices α and β in the parentheses correspond to residues belonging to TOPIIα and TOPIIβ, respectively). The significance of these two residues in stabilizing DNA–TOPII complexes in which TOPII poisons are bound is also highlighted elsewhere. ,− The active site residues other than ARG487­(α) and ARG503­(β) play a supporting role in further stabilizing the ternary complexes through their contacts with the two anthracyclines. As for the hydrogen bond network with the active site residues for both drug molecules, it is dynamic in that the residues participating in it are not constant during the simulations. − The most frequently formed hydrogen bonds of TOPII isoforms with berubicin and doxorubicin involve GLU461­(α)/GLU477­(β) and GLU506­(α)/GLU522­(β). GLU461­(α)/GLU477­(β) have also been identified within the hydrogen bond network of doxorubicin in the literature. , In general, the identified contact network in Table is in accordance with several articles regarding docking simulations of doxorubicin–DNA–TOPII complexes; − the key residues identified are ASP463, ARG487, GLY488, GLY462, GLU461 and MET762 for doxorubicin–DNA–TOPIIα, and GLN778, ARG503, ASP479, GLY478, MET782, MET781 and GLU477 for doxorubicin–DNA–TOPIIβ. It is widely accepted that the contact networks provide insight into the selectivity of TOPII poisons for TOPIIα over TOPIIβ, ,,,,, since their strength determines the thermodynamic stability of the ternary complexes. Consistent with this, we observe from Tables and that the contact networks in TOPIIα complexes for both anthracyclines are denser, indicating a preference for this TOPII isoform. The difference in the contact numbers between the two TOPII isoforms is greater in the case of berubicin, a finding consistent with a better selectivity of this anthracycline for TOPIIα compared to that of doxorubicin. This is related to the more hydrophobic nature of the binding sites of TOPIIα in comparison with those of TOPIIβ, as stated elsewhere. ,,,, Indeed, the substitution in the sugar moiety renders berubicin more hydrophobic than doxorubicin. The shielding of the two drugs from adjacent water molecules, due to their binding within TOPIIα and TOPIIβ, is quantified by the SASA of the anthracyclines (SASA_anthra), which are provided in the last row of Table . The SASA of berubicin exhibits a lower value in the case of TOPIIα, fully consistent with a better fit inside this ternary complex, whereas the SASA of doxorubicin is very similar for both isoforms, within statistical error.

4. Key Properties Concerning the Anthracycline–TOPII Interactions in the Bound State.

	berubicin–DNA–TOPIIα	doxorubicin–DNA–TOPIIα	berubicin–DNA–TOPIIβ	doxorubicin–DNA–TOPIIβ
CNTOPII‑anthra (total)	29.1 ± 1.5	20.9 ± 1.3	21.5 ± 2.5	18.2 ± 1.2
HBTOPII‑anthra	1.0 ± 0.1	1.8 ± 0.3	0.7 ± 0.1	0.9 ± 0.2
SASAanthra [nm2]	2.04 ± 0.04	1.88 ± 0.07	2.75 ± 0.14	1.90 ± 0.06

3.3. Structural Changes upon Binding

TOPII poisons, such as anthracyclines, act by trapping DNA–TOPII cleavage complexes at an intermediate stage, preventing the religation of the cleaved DNA ends. ,, Two coexisting factors govern the stabilization of the cleavage complexes. The first is of thermodynamic origin, whereby anthracyclines energetically stabilize the cleavage complexes; more details about this are provided in the next subsection. According to the second factor, TOPII poisons geometrically hinder religation. First of all, they keep the cleaved DNA ends separated from each other, ,,,,, since intercalation widens the gap between them. In addition to this, TOPII poisons interfere with the interactions between DNA and TOPII, which facilitate the alignment of cleaved ends toward religation. ,,,− In general, upon binding, TOPII poisons induce significant conformational and structural changes in TOPII, thereby hindering religation. ,, It is also noteworthy that the aforementioned conformational changes are observable in MD simulations on the time scale of hundreds of nanoseconds. ,

In the previous subsection, we have shown that the cleaved ends are separated enough to prevent DNA religation for both anthracyclines. These two drugs also induce conformational changes in the structure of TOPII. To structurally investigate this effect, we focus on the active site residues of TOPRIM and WHD within a protomer, given that they predominantly interact with the two anthracyclines. Graphical representations of these domains are provided in Figures and S2. More precisely, we calculate the average distance between the COMs of the primary active site residues in TOPRIM and WHD within a protomer, and the results are gathered in Table . It is concluded that both berubicin and doxorubicin induce a modest increase in the distance between these domains compared to the corresponding binary complexes, thus contributing to the hindering of DNA religation. It is noted that the interactions between the TOPRIM and WHD domains play a significant role in DNA religation. , The increase in distance between the TOPRIM and WHD domains not only results in weakened interactions between these domains but also attenuates the DNA–TOPII interactions, as discussed below. Another conclusion drawn from Table is that the reported distances are smaller in the binary DNA–TOPIIα cleavage complex, indicating more compact binding sites in this complex. This compactness is additionally verified in the literature, as well as by the larger contact numbers and more extensive hydrogen bond networks between the two anthracyclines and TOPIIα observed here (see Table ). It is generally recognized that TOPII catalyzes DNA religation by orienting the cleaved DNA ends via DNA–protein interactions. ,,,, Thus, a reduction in these interactions contributes to the hindering of religation. To study this structurally, we calculate the contact numbers between DNA and TOPII (CN_DNA–TOPII) and the SASA of DNA (SASA_DNA), both as functions of the base-pair sequence; these two properties are presented in Figures and S6. The separation of the TOPRIM and WHD domains within each protomer, due to the interference of the drugs in the DNA–TOPII interface, reduces CN_DNA–TOPII and increases the SASA of the DNA base pairs in the cleavage region (see the region between the vertical black lines in Figures and S6). The weakening of interactions between DNA and TOPII is primarily localized in the cleavage core, where DNA–TOPII interactions govern the religation reactions. ,,− For a better understanding of structural changes induced by the anthracyclines, four additional properties are provided in Table . To be more specific, the hydrogen bonds between DNA and TOPII, as well as the contacts between them, are significantly fewer in the ternary complexes than in the corresponding binary DNA–TOPII complexes. The structural alterations within the cleavage complexes due to the bound drugs render DNA less shielded by TOPII. In addition, as seen from the last row in Table , the SASA of DNA increases by an average of 6.35 nm² in the ternary complexes when the bound drugs are ignored in the calculations. The creation of free space induced by TOPII poisons is also reported in other studies. ,, It is pointed out that the SASAs of DNA (third row in Table ) are comparable in the corresponding binary and ternary complexes. This implies that the anthracyclines occupy the free space generated by the displacement of the TOPII residues, indicating that nearest-neighbor effects take place upon intercalation, as anticipated for intercalating drugs.

5. Average Distances between the COMs of WHD and TOPRIM Domains Within Each Protomer in [nm] .

berubicin–DNA–TOPIIα	doxorubicin–DNA–TOPIIα	berubicin–DNA–TOPIIβ	doxorubicin–DNA–TOPIIβ
1.97 ± 0.02	2.03 ± 0.04	2.22 ± 0.03	2.05 ± 0.01

^aOnly the primary active site residues are considered in the calculation of the COM of each domain.

^bThe corresponding distances are (1.79 ± 0.02) nm and (1.97 ± 0.01) nm for the DNA–TOPIIα and DNA–TOPIIβ complexes, respectively.

7.

(a) Contact numbers between DNA and TOPIIα and (b) SASA of DNA base-pairs, both as a function of the base-pair sequence. The black vertical lines delineate the cleavage region. Note that the drugs are ignored in the calculation of the presented SASA in the ternary complexes.

6. Properties of DNA in the Binary and Ternary Complexes under Investigation .

	berubicin–DNA–TOPIIα	doxorubicin–DNA–TOPIIα	berubicin–DNA–TOPIIβ	doxorubicin–DNA–TOPIIβ
HBDNA–TOPII	53.1 ± 1.1 (57.4 ± 1.4)	48.2 ± 1.1 (57.4 ± 1.4)	38.2 ± 1.8 (51.0 ± 0.4)	44.8 ± 1.1 (51.0 ± 0.4)
CNDNA–TOPII	586 ± 3 (644 ± 15)	574 ± 5 (644 ± 15)	409 ± 7 (632 ± 5)	522 ± 10 (632 ± 5)
SASADNA [nm2]	35.6 ± 0.4 (36.1 ± 0.2)	36.8 ± 0.4 (36.1 ± 0.2)	40.8 ± 0.4 (40.7 ± 0.2)	38.04 ± 0.5 (40.7 ± 0.2)
SASADNA [nm2]	42.0 ± 0.6 (36.1 ± 0.2)	43.8 ± 0.4 (36.1 ± 0.2)	48.0 ± 0.1 (40.7 ± 0.2)	45.2 ± 0.8 (40.7 ± 0.2)

^aThe values for the corresponding binary complexes are provided in parentheses.

^bThe drugs are ignored in the calculation of this property.

3.4. Thermodynamic Analysis of the Anthracycline–DNA–TOPII Complexes

To demonstrate the convergence and reliability of DDM calculations, we provide several diagnostic plots in Section S2 of the Supporting Information, along with a brief discussion of them. The thermodynamic analysis presented, which focuses primarily on standard binding Gibbs energies, is necessary to thoroughly investigate the stabilization of the cleavage complexes, a crucial factor for anthracyclines to exert their anticancer action. It is parenthetically pointed out that an investigation of the effect of DNA length on the calculated binding Gibbs energies is outside the scope of this article. Before starting the analysis, it should be mentioned that, to the best of our knowledge, there are no published standard binding Gibbs energies or binding equilibrium constants for the systems under consideration.

The decomposition of the standard binding Gibbs energies into individual terms, according to the thermodynamic cycle described by eq , provides molecular-level insight into the binding mechanism and the underlying driving forces. Our thermodynamic analysis is summarized in Table , as well as in Figures and . First, we note that the standard binding Gibbs energies of all the ternary complexes are negative, and notably more negative than those of the corresponding berubicin–DNA intercalation complexes obtained from refs and . Thus, from a thermodynamic point of view, the two anthracyclines interact more strongly when TOPII is present. This finding is fully consistent with the fact that anthracyclines target DNA–TOPII cleavage complexes. In the case of berubicin, ΔG _bind is more negative when bound to the DNA–TOPIIα complex; berubicin shows a significant preference for TOPIIα over TOPIIβ, stabilizing the complex in question more effectively. This finding provides direct evidence that berubicin is a strong and selective TOPII poison and thermodynamically more effective than doxorubicin. For a deeper understanding of the binding mechanism, we examine the Lennard-Jones and electrostatic contributions to the binding free energies. To be more specific, the thermodynamic term $-(\mathrm{\Delta }{G}_{5\to 2}+\mathrm{\Delta }{G}_{3\to 4})$ is decomposed into Lennard-Jones (ΔG _LJ) and electrostatic (ΔG _el) interactions in the same fashion as in ref . It is apparent from Table that ΔG _LJ dominates in all ternary complexes; the Lennard-Jones interactions reflect the nonpolar and hydrophobic interactions of anthracyclines with DNA and TOPII. Secondarily, the electrostatic interactions act favorably in the binding of the anthracyclines. It is noteworthy that the same finding regarding Lennard-Jones and electrostatic interactions is also observed in the berubicin–DNA complexes of ref . To gain more insight into the interactions of the two anthracyclines with DNA–TOPII in each complex, the potential energies between them are decomposed into Lennard-Jones and electrostatic contributions, and the results are presented in Figure . In addition, Figure shows the decomposition of the potential energy between the anthracycline and the DNA–TOPII complex into anthracycline–DNA and anthracycline-TOPII contributions. It is noted that the two aforementioned figures are obtained from unbiased MD simulations only. Figure also confirms that the Lennard-Jones interactions are the main contributions to the binding of anthracyclines, and shows that both anthracyclines interact more strongly with the DNA–TOPIIα complex. Figure provides further insight into the interactions of berubicin and doxorubicin within the DNA–TOPII complexes. More precisely, the anthracycline–DNA interactions mainly stabilize the drugs in the binding sites due to intercalation. It is also bibliographically known that doxorubicin is stabilized within DNA–TOPII complexes via stacking and hydrophobic interactions with DNA when intercalated. ,,, Moreover, stacking and hydrophobic interactions are observed between doxorubicin and TOPII. ,, The two figures in question show that berubicin interacts more strongly within the DNA–TOPII complexes than doxorubicin does; this is due to the Lennard-Jones interactions presented in Figure , as well as the interactions with DNA, as seen in Figure . This finding aligns with ref , in which it is demonstrated that berubicin forms more stable intercalation complexes compared with doxorubicin. More variable factors in such complexes are the electrostatic interactions of the drug molecule with DNA and TOPII, which partially account for the polar interactions and hydrogen bonding and result in further stabilization of the complexes. ,

7. Standard Binding Gibbs Energies and DDM Contributions in [kJ/mol].

ternary complex	ΔG LJ	ΔG el	$$-\mathrm{\Delta }{G}_{1\to 3}-\mathrm{\Delta }{G}_{4\to 5}$$	ΔG bind
berubicin–DNA–TOPIIα	–83.4 ± 1.6	–45.3 ± 1.1	11.6 ± 0.1	–117.1 ± 1.9
berubicin–DNA–TOPIIβ	–74.4 ± 1.9	–19.8 ± 1.0	–3.7 ± 0.1	–97.9 ± 2.1
doxorubicin–DNA–TOPIIα	–69.8 ± 1.0	–21.0 ± 0.7	11.5 ± 0.1	–79.3 ± 1.2

^aΔΔG _bind is included in this term.

8.

Decomposition of the potential energy between the anthracycline and the DNA–TOPII complex into Lennard-Jones and electrostatic contributions (BER and DOX stand for berubicin and doxorubicin, respectively).

9.

Decomposition of the potential energy between the anthracycline and the DNA–TOPII complex into anthracycline–DNA and anthracycline-TOPII contributions (BER and DOX stand for berubicin and doxorubicin, respectively).

4. Conclusions

The presented thermodynamic analysis, based on a rigorous method, shows that the berubicin–DNA–TOPIIα complex is the most favorable, indicating a clear preference of berubicin for TOPIIα over TOPIIβ. This preference is a significant and highly desirable finding for an anthracycline used in cancer chemotherapy. In addition, it is found that all complexes investigated exhibit more negative standard binding Gibbs energies than berubicin–DNA intercalation complexes, , showing that the presence of TOPIIα or TOPIIβ, covalently bonded with double-stranded DNA via the cleavage reaction, leads to more stable complexes. The intercalated anthracyclines are further stabilized due to their interactions with specific residues of TOPII in the neighborhood of the cleavage region, as well as the minor groove of the DNA sequence. Both berubicin and doxorubicin form hydrogen bond networks with DNA and the two TOPII isoforms, whereas the Lennard-Jones contributions to the standard binding Gibbs energies predominate over the electrostatic contributions. Our analysis, as described in Section , suggests that berubicin forms more stable complexes than doxorubicin when it is bound in DNA–TOPIIα complexes; however, for a complete comparison of these two anthracyclines, a kinetic analysis of the examined complexes is needed. This will be the topic of a subsequent study. Structurally, DNA intercalation increases the separation between the cleaved base pairs, thereby hindering religation.

From a molecular point of view, the understanding of TOPII poisoning by doxorubicin and berubicin, as detailed in this study, may aid in the rational design of new anthracycline derivatives. The greater hydrophobicity of berubicin with respect to doxorubicin, due to the additional benzyl group in the sugar moiety, leads to the formation of more stable complexes from a thermodynamic point of view. The chemical modification of the sugar moiety in berubicin positively affects the interactions with both the minor groove of DNA and the residues of TOPII, especially TOPIIα, whose binding pocket is more hydrophobic than that of TOPIIβ. Note that according to the literature, modifications in the sugar moiety of anthracyclines have a direct impact on their interactions with TOPII and DNA, ,,− and thus more effective anthracycline derivatives may be designed by rationally engineering this moiety, provided that the intercalation is not stereochemically impeded. The enhancement of intercalation into DNA may be also achieved by strengthening the electrostatic interactions with the negatively charged DNA backbone; this is also shown for the berubicin–DNA intercalation complexes of ref .

Topology, coordinate, and molecular dynamics parameter files for the production molecular dynamics simulations are provided at 10.5281/zenodo.20556054.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jcim.6c01823.

• The Supporting Information comprises technical details and information about molecular docking simulations (Section S1), additional details about the implementation of DDM (Section S2), as well as representative configurations of DNA in the binary and ternary complexes containing TOPIIβ; selected properties concerning the anthracycline–DNA and anthracycline–DNA–TOPIIβ complexes; a detailed graphical description of DBCC of TOPIIβ; a representative configuration of the doxorubicin–DNA–TOPIIα complex; a representative configuration of the berubicin–DNA–TOPIIβ complex; a representative configuration of the doxorubicin–DNA–TOPIIβ complex; and DNA–TOPIIβ contact numbers and SASA of DNA as functions of base-pair sequence in binary and ternary complexes containing TOPIIβ (PDF)

All authors made significant contributions to the manuscript and approved the final version.

The open access publishing of this article is financially supported by HEAL-Link.

The authors declare no competing financial interest.