Interfacial inhibitors

Abstract

Targeting macromolecular interface is a general mechanism by which natural products inactivate macromolecular complexes by stabilizing normally transient intermediates. Demonstrating interfacial inhibition mechanism ultimately relies on the resolution of drug-macromolecule structures. This review focuses on medicinal drugs that trap protein–DNA complexes by binding at protein–DNA interfaces. It provides proof-of-concept and detailed structural and mechanistic examples for topoisomerase inhibitors and HIV integrase inhibitors. Additional examples of recent interfacial inhibitors for protein–DNA interfaces are provided, as well as prospects for targeting previously ‘undruggable’ targets including transcription, replication and chromatin remodeling complexes. References and discussion are included for interfacial inhibitors of protein–protein interfaces.

Because biological systems consist of macromolecular ensembles that need to move with respect to each other to perform their enzymatic or structural functions, and because such reactions create a spectrum of molecular interfaces between macromolecules, it is understandable how small molecules that bind to such interfaces can interfere with the function of the macromolecular complexes (Fig. 1). Hence, the complexity of macromolecular complexes and their dynamic behavior creates unique opportunities to develop and discover small molecules that bind at such interfaces with high selectivity.

Figure 1.

Schematic illustration of interfacial inhibition. The normal cycle of enzyme mechanical interconversion from ‘closed’ state to ‘open’ is interrupted by use of the interfacial inhibitors. The cycle is stalled at the ‘open’ state by a ‘wedge’-like inhibitor (green triangle) analogous to inhibition of Top2 catalyzed DNA relegation step by etoposide. The stabilization of the ‘closed’ state is achieved by use of a ‘lock’-like inhibitors (yellow), similar to dexrazoxane trapping of the closed state of Top2 ATP binding subunits.

The interfacial inhibitor concept arose from the observation that topoisomerase inhibitors, which are widely used as anticancer drugs, produce topoisomerase-linked DNA breaks that correspond to normally transient catalytic intermediates of the topoisomerase reactions (see below). The hypothesis that topoisomerase-linked DNA breaks were generated by the binding of the topoisomerase inhibitors at the interface of the broken DNA and the enzyme was proposed in the 90s^1–3 but remained unproven for 10 years until the co-crystal structure determination of the natural product derivative topotecan bound to the topoisomerase I (Top1) cleavage complex (Top1cc)⁴ and more recently of etoposide bound to the topoisomerase II cleavage complex (Top2cc)⁵ (see below and Figs. 2 and 3). Independently, the co-crystal structure of the fungal macrolide brefeldin A bound at the interface of the small GTP-binding protein Arf and its guanine-nucleotide-exchange factor Sec7⁶ provided proof of principle that the interfacial concept is not limited to protein–nucleic acid interfaces but also applies to protein–protein interfaces.⁷ The term ‘interfacial inhibitor’⁷ was coined to describe this previously unanticipated mode of inhibition. Since the concept was first proposed, the number of examples of natural product that act as interfacial inhibitors has grown steadily, encompassing cell surface receptors (exemplified by the nicotinic receptor inhibitors), signal transduction molecules (exem plified by the mTOR inhibitors), scaffolding macromolecular complexes (exemplified by tubulin inhibitors), and protein-DNA complexes (exemplified by topoisomerase, polymerase and ribosome inhibitors).^8,9 Recently purely synthetic drugs were found to act as interfacial inhibitors of HIV integrase (see below), thereby opening perspectives for a large number of targets and medicinal chemists. The present review expands our previous reviews on this topic.^7–9

Figure 2.

Interfacial inhibition of Top1 by topotecan. (A) Structural alignment of human topoisomerase type I (Top1) in absence of inhibitor (blue cartoon, PDB ID: 1K4S) and bound (light pink cartoon, PDB ID: 1K4T) to topotecan (green spheres) shows change in DNA position relative to Top1. Structural alignment and cartoon representation of structures were obtained with PyMOL. (B) Schematic representation of the topotecan binding mode to Top1–DNA complex. The flanking DNA base pairs TA (blue solid line rectangles) and GC (red dash line rectangles) form π–π stacking interactions with polycyclic core of topotecan, the key residues R364, D533 and N722 form direct and water mediated hydrogen bond interactions with topotecan.

Figure 3.

Interfacial inhibition of Top2 by etoposide and dexrazoxane. (A) structure of etoposide-hTop2–DNA ternary complex (PDB ID: 3QX3). Etoposide (green spheres) molecules were found in both cleavage sites of the dimeric hTop2β dsDNA covalent complex. (B) Structure of dexrazoxane (green spheres) bound at the interface of ATP (yellow spheres) binding subunits (light pink and blue) of TOP2.

To illustrate the points described above, the next sections describe specific examples of protein nucleic acid complexes (topoisomerase and HIV integrase inhibitors) and one recent example of interfacial protein-protein inhibitor for the viral cofactor STING.

Topoisomerases are ubiquitous enzymes that alter DNA topology by relieving supercoiling-associated tension in double stranded DNA. Topoisomerases perform their function by transiently cutting one strand (type I topoisomerases including Top1, Top1mt, Top3α and Top3β) or both DNA strands (type II topoisomerases including Top2α and Top2β in humans, and bacterial gyrase and Topo IV).^13,14 The ubiquitous presence and crucial biological roles of topoisomerases explain their prevalence as therapeutic target. Pharmaceutical and medicinal chemistry research efforts have yielded a collection of important antibiotics (e.g., oxolinic and nalidixic acid analogues and quinolone derivatives) and anticancer agents (e.g., irinotecan, topotecan, etoposide, doxorubicin, aclarubicin, dexrazoxane and mitoxantrone).¹⁵ A dozen of topoisomerase inhibitors are in common clinical practice worldwide for treating bacterial infections and a broad range of cancers. Structural and biochemical studies have unambiguously demonstrated that topoisomerase inhibitors are interfacial inhibitors.^4,5,16–18 Their binding at the subunit or protein–DNA interfaces interferes with the topoisomerase catalytic cycle by preventing the rapid conversion of enzyme-DNA configurations required for the topoisomerase reactions (Fig. 1).^{13–15,19,20}

Topoisomerase-mediated DNA supercoil removal is a three-step process. First, the topoisomerases cut one (type I) or two (type II) strands of double-stranded DNA, forming a covalent binary topoisomerase–DNA complex, which is referred to as a cleavage complex (Top1cc for Top1 and Top2cc for Top2). For Top1 (type IB), the subsequent step of supercoil removal occurs via rotation when the clamp-like structure of the topoisomerase encapsulating the DNA allows the rotation of the free DNA end around the intact strand.^21,22 For type IA (Top3α and Top3β in humans and Topo I in bacteria) and type II topoisomerases (Top2α and Top2β in humans or gyrase and Topo IV in bacteria), supercoils are removed through strand passage and the overall process is orchestrated by subunit rearrangements within homodimers (Top2) or tetramers (gyrase and Topo IV). Type IB topoisomerases (Top1 and Top1mt) change the DNA linking number in steps of one as the DNA swivels (rotates) around the enzyme covalently bound to DNA. By contrast, topoisomerization reactions for type IA and type II topoisomerases proceed by strand passage (gate mechanism) for each catalytic cycle, changing the linking number in steps of 1 (Type IA enzymes) or 2 (type II enzymes). In the final step of the reaction, topoisomerases (and their covalent bond to DNA) are released by nucleophilic attack of the free DNA end at the break site toward the topoisomerase-DNA covalently bound catalytic tyrosine.^13,14

Human Top1 (type IB) is an essential enzyme targeted by the two FDA-approved anticancer drugs topotecan and irinotecan. Together structural, biochemical, and mutation analyses²³ revealed the critical importance of drug interaction with both the cleaved DNA and the enzyme, highlighting its interfacial nature. When bound to the Top1–DNA complex, camptothecin or its synthetic analogues intercalate into the DNA at the site of Top1-induced nick, and at the same time form a network of direct and water-mediated hydrogen bonds.^4,24 It was shown that DNA sequence preference differs across chemical classes of Top1 inhibitors demonstrating the impact of π–π stacking interactions between drug and DNA nucleobases on the binding and selectivity of inhibitors.²⁵ Additionally, it was observed that Top1 mutations at residues R364, D533 and N722 confer resistance towards camptothecin-based inhibitors.²³ These residues were later found to play key roles in drug binding to Top1–DNA complex (Fig. 2B).²⁴

Upon inhibitor binding, the ends of the cut DNA strand are spatially separated to a distance of ~11 Å, preventing their religation. Remarkably, the binding site of camptothecins (PDB IDs: 1K4T, 1T8I), of the non-camptothecin indenoisoquinolines in clinical development (PDB IDs: 1SC7, 1TL8) and of indolocarbazoles (PDB ID: 1SEU) determined via crystallographic studies could not be observed in the cleavage complex alone (PDB ID: 1K4S).^4,16 The binding site of Top1 inhibitors and its interfacial nature became apparent only when the structure of the Top1-DNA complex trapped by inhibitors was crystallographically resolved. Figure 2 shows structural alignment of the binary Top1–DNA cleavage complex and the Top1–DNA-topotecan ternary complex. Such alignment reveals the required alteration of the structure of the complex in order to form the binding site that accommodates topotecan. The DNA double helix on the 5′-end of the cut strand adopts a new position relative to the binary complex (Fig. 2). In addition, the position of some Top1 loops is altered, but the overall fold of the enzyme as determined by X-ray crystallography remains the same.

Top2 inhibitor-stabilized cleavage complexes also involve the trapping of type II topoisomerases: topoisomerase IV from Streptococcus pneumonia and human Top2β.^5,26,27 In the case of Top2β, each monomer of the Top2 dimer accommodates one inhibitor molecules of the Top2–DNA cleavage complex (Top2cc) (Fig. 3).⁵ Similar mode of binding is observed in the case of the topoisomerase IV complex with the formation of a binding pocket at the protein–DNA interface when the binary system is subjected to treatment with a drug.²⁷

Another mode of interfacial inhibition of Top2 heterodimers is exemplified by dexrazoxane (ICRF-187), which binds at the interface of an antiparallel dimer of the Top2 ATP binding domain (Fig. 3B).²⁸ Such mode of inhibition stabilizes the closed conformation of the ATP binding domains, ultimately preventing the strand passage step from taking place.

While etoposide binds at the Top2-DNA interface by intercalating in the cleavage site of the Top2cc, acting as a wedge preventing the ends of the DNA from coming together and religating, dexrazoxane acts as a molecular lock stabilizing ATP binding at the Top2 dimer interface.^5,28 In both instances, the Top2–DNA–drug complexes are mechanically locked and unable to perform the intended transformation due to their inability to change conformation. Additionally, the formation of new binding pockets upon inhibitor binding presents a challenge and new insights into possible ways of targeting such macromolecular complexes for therapeutic purposes are warranted. The challenge primarily concerns rational drug development and understanding the mechanism of inhibition. Due to significant conformational differences between binary Top2–DNA and ternary drug–Top2–DNA complexes, it would be difficult to predict the exact mode of drug binding. Hence, in absence of any structural information about the ternary complex, only hypothesis can be made with regards to mechanism of inhibition and ways to improve potencies through medicinal chemistry efforts.

HIV integrase strand transfer inhibitors (INSTIs) have transformed the standard of care for HIV patients. They are currently three FDA-approved INSTIs: raltegravir, elvitegravir and the best in class dolutegravir.²⁹ Integrase is required for HIV replication because it is responsible for inserting the viral DNA in the host chromosome.³⁰ The INSTIs prevent integrase from performing its catalytic reaction by blocking the enzyme after it completes its initial cleavage step, which is referred to as 3′-processing, that activates the proviral DNA before integrase can catalyze the second transesterification step, referred to as strand transfer. To block strand transfer, INSTIs bind at the interface of the integrase polypeptide, the viral DNA and the two catalytic divalent metals (Fig. 4A). Analysis of the molecular interactions reveals that the molecule sits on the two catalytic magnesium ions, which are coordinated by the three catalytic acid residues from the DDE motif D64, D116 and E152 of integrase.³¹ As shown for dolutegravir (but also true for raltegravir and elvitegravir, the drug coordinates the divalent cations via its three co-planar oxygen atoms (Fig. 4B). The difluoro phenyl ring of dolutegravir occupies a pocket previously occupied by the displaced terminal adenine of the viral DNA and stacks against the penultimate cytosine (C-2). Dolutegravir is also stabilized by two van der Waals interactions with residues P145 and T143 of the integrase protein (Fig. 4B).

Figure 4.

Interfacial inhibition of HIV-1 integrase by dolutegravir. (A) Co-crystal structure of dolutegravir in prototype foamy virus (PFV) integrase, which represents a surrogate for HIV-1 integrase (PDB ID: 3S3M). Dolutegravir (green spheres) binds at the interface of the catalytic site of a monomer of full-length PFV integrase (light pink surface representation), DNA (blue cartoon) and two catalytic magnesium ions (red spheres). For clarity, the PVF integrase representation excludes the first 95 amino acid residues. (B) Schematic representation of the dolutegravir binding mode to PFV integrase-DNA–magnesium complex. The structure of dolutegravir is shown in green. Magnesium ions are highlighted in red. Protein residues are represented in light pink and labeled following the HIV-1 integrase sequence numbering. The penultimate cytosine DNA nucleotide is represented in blue. Dashed arcs represent van der Waals interactions.

To illustrate that the interfacial inhibition mechanism also applies to protein–protein interfaces, we will now discuss the example of STING, which is encoded by the TMEM173 gene. STING is a homodimer signaling protein responsible for the activation of the interferon (IFN) pathway.³² It is activated by the binding of the second messenger cyclic di-GMP produced by GMP-AMP synthetase (cGAS). It was recently shown that the antiviral drug DMXAA inhibits STING by the binding of two drug molecules at the interface of the two STING monomers (Fig. 5A). The interfacial inhibition at the protein–protein interface is exemplified by the polar interaction of each DMXAA molecule with residues T262&T266 from one monomer and with R237 from the other monomer, respectively (Fig. 5B). Each DMXAA molecule is also stabilized by a polar interaction via a water molecule with residue Y239 from each monomer. The two DMXAA molecules bind a pocket and are stabilized by van der Waals interactions with protein residues S161, I164, L169 and I234 from both monomers.³²

Figure 5.

Interfacial inhibition of STING by the antiviral DMXAA. (A) Co-crystal structure of DMXAA with STING homodimer (PDB ID: 4LOL). (A) STING monomers are in light blue and light pink surface representation. Two molecules of DMXAA (green spheres) are bound at the protein–protein interface of the homodimeric complex. (B) Schematic representation of the DMXAA binding mode to STING. DMXAA molecules are shown in green. Protein residues are represented in light blue or light pink, according to their origin from one or the other monomer. Dashed lines represent polar interactions and dashed arcs represent van der Waals interactions.

Interfacial inhibitors open new avenues for drug design, and for elucidating the molecular mechanism of action of medicinal compounds. The term interfacial inhibition highlights a specific character of drug/macromolecular complex interactions, i.e. binding of a drug at a macromolecular interface restricting subunit movements within the macromolecular complex. It could be seen in numerous examples that a number of known inhibitors, while binding at the macromolecular complex subunit interfaces, also fit into conventional enzyme inhibitor classification as competitive, non-competitive or uncompetitive. Also, the interfacial inhibitor targets expand beyond enzymes (such as topoisomerases or HIV integrase described here) to structural macromolecules. In the case of enzymes, interfacial inhibitors can bind at the ligand site and therefore act as competitive inhibitors (e.g., graniseton in the case of the 5-HT3 receptor).¹⁰ They can also affect enzyme active sites or cofactor sites by allosteric (propagated) mechanisms, or block enzymatic activity by trapping an intermediate configuration of the macromolecular complex (uncompetitive inhibitors such as some topoisomerase and HIV integrase inhibitors described here). In the case of structural proteins, the interfacial binding distorts the macromolecular complex polymers, thereby affecting cellular structure and function. An example of such inhibition is tubulin polymerization inhibitors (colchicine and vinblastine), which bend microtubules out of shape.¹¹ The increasing realization that molecules interact with each other as large protein complexes in a highly dynamic manner (such as chromatin remodeling complexes that adjust chromatin to its metabolic and structural states)¹² makes it likely that the ‘undruggable targets’ will be within reach once we can recapitulate those complexes biochemically and structurally Table 1.

Table 1.

Examples of protein–nucleic acid interfacial inhibitors

Drug	Origin	Substrate	PDB	Refs.
Camptothecin	Plant extract	Topoisomerase I–DNA complex	1T8I	4,16,24,33
Topotecan	Synthetic		1K4T
Indenoisoquinoline	Synthetic		1TL8, 1SC7
Indolocarbazole	Bacterial toxin		1SEU
Etoposide	Synthetic	Topoisomerase II–DNA complex	3QX3	5,26,34,35
Amsacrine	Synthetic		4GOU
Mitoxantrone	Synthetic		4GOV
Ametantrone	Synthetic		4GOW
Quinolones	Synthetic	Topoisomerase IV–DNA complex	3FOE	17,36
Quinolones	Synthetic	DNA Gyrase–DNA complex	2XCT	18,37,38
GSK299423	Synthetic		2XCR
AM8191	Synthetic		4PLB
GSK966587	Synthetic		4BUL
4Q5V	Micro organism	DNA polymerases	4Q5V	39,40
Alpha-amanitin	Fungus	RNA polymerase II	2VUM, 3CQZ	41–43
Fusidic acid	Fungus	Bacterial ribosome	4V9L, 4WQF	9,44–49
Thiostrepton	Bacterial		3CF5
Raltegravir	Synthetic	HIV-1 integrase–DNA complexes	3OYA	31,50
Elvitegravir	Synthetic		3L2U
Dolutegravir	Synthetic		3S3M