Drugging Topoisomerases: Lessons and Challenges

1. Foreword

The first part of this assay summarizes the known mechanisms by which drugs target topoisomerases, complementing and updating more detailed reviews.^1–12 The relatively unknown mechanism of action of topoisomerase inhibitors can be traced to the complexity of the topic with 6 different genes in humans cells and bacteria, drugs acting as interfacial inhibitors that trap ternary complexes, and drug cytotoxic mechanisms mediated by the trapping of topoisomerases on DNA rather than by classical enzymatic inhibition. The second part of the review addresses the remaining challenges for the development and precise use of topoisomerase inhibitors for the treatment of cancers and infections.

2. DNA topoisomerases

Topoisomerases are universal and present in eukaryotes, archaebacteria and eubacteria.^13–19 Human cells encode six topoisomerases whereas bacteria generally contain only 4 topoisomerases and lack the type IB enzymes (Table 1 and Fig. 1).⁵ The ubiquity of topoisomerases stems from DNA’s double-helical (duplex) structure and length, which promote DNA entanglement in the compacted nucleus of eukaryotic cells or the nucleoid of bacteria. The opening of duplex DNA and separation of its two strands during transcription and replication generate supercoiling (torsional tension) on both sides of the open DNA segment. Excessive positive supercoiling tightens the DNA and prevents further strand separation thereby stalling the polymerases. Negative supercoiling behind the polymerases, on the other hand, tends to extend DNA strand separation and facilitates the formation of abnormal nucleic acid structures such as R-loops, which can stall RNA polymerase when the transcripts remain bound to the unwound DNA template. Negative supercoiling also promotes the formation of non-canonical DNA structures such as z-DNA, intramolecular hairpins and guanosine quartets (G4’s). Topoisomerases prevent the formation of such potentially deleterious structures by removing free supercoiling.

Topoisomerase remove supercoiling by different mechanisms. Type IB enzymes work by letting the broken strand rotate around the intact strand (Fig. 1B),^20–23 whereas type IA and type IIA enzymes work by passing one strand or one duplex, respectively, from the same DNA molecule through the single- or double-strand break generated by the topoisomerase in another duplex (Fig. 1 and Table 1).^15,24,25

Replication of circular DNA molecules and chromatin loops produces interlinked DNA products (catenanes)⁵ that need to be removed by the strand passing activities of topoisomerases. While type IIA enzymes (Fig. 1C and Table 1) act as full decatenases, passing one duplex through another, the strand passing activity of type IA enzymes is restricted to single-stranded DNA segments adjacent to duplex regions (Fig. 1A), which enables Top3 to pass a DNA single-strand and resolve hemicatenanes and double-holiday junctions following replication of supercoiled DNA.²⁶

Topoisomerases always break DNA by transesterification reactions using an active site tyrosine as the nucleophile that attacks the DNA phosphodiester backbone. Type IA and IIA enzymes break the DNA by attacking and bonding to the 5’-phosphate whereas type IB enzymes break DNA by covalent attachment to the 3’-phosphate (Table 1 and Figs. 1 and 2). The resulting 3’-hydroxyl ends in the case of type IA and IIA enzymes and 5’-hydroxyl ends in the case of the type IB enzymes reverse the phosphotyrosyl bonds, thereby enabling the release of the topoisomerase and religation of the DNA (Figs. 1 & 2). The nicking-closing activities of topoisomerases are remarkably fast (up to 6000 cycles per minute for Top1 and 250 for Top2);^6,23 yet the enzymes are susceptible to the drugs selectively when the DNA is in the cleaved state (Fig. 2).²³

3. Topoisomerase inhibitors and the interfacial inhibition principle

The molecular mechanism of action of topoisomerase inhibitors, i.e. their specific binding at the interface of topoisomerase-DNA complexes, led to the interfacial inhibition concept.^6,27,28 We proposed this hypothesis initially for Top2 inhibitors to explain the sequence selective trapping of Top2cc by different drugs; namely the preference for an adenine at position −1 in the case of doxorubicin and other anthracyclines (Fig. 3C),²⁹ and for cytosine −1 and adenine +1 for etoposide and m-AMSA, respectively.^30,31 Similarly for Top1, camptothecin preferentially traps a subset of Top1cc; those with a guanine +1.³² A unifying model emerged by which one drug molecule stacks against the base pairs flanking the topoisomerase-induced cleavage site (Fig. 2C & G).^29,30

To account for the stereospecificity of camptothecins (i.e. only the natural 20-S-isomer is active against Top1; see Fig. 3A) ³³ and for the high drug resistance of specific Top1 mutants,³⁴ we also proposed that the enzyme forms specific amino acid contacts with the drug as it stacks against the bases flanking the cleaved DNA. This led to the ternary complex hypothesis with the drug simultaneously interacting with the DNA and the enzyme (Fig. 2C & G).

It took over 10 years to confirm by x-ray crystallography the Top1-DNA-camptothecin model (Fig. 2D) with camptothecin stacking against the +1 guanine and forming a hydrogen bond network with the enzyme.^35–38 The confirmation of the Top2cc trapping interfacial model (Fig. 2C) for antibiotics and anticancer drugs was obtained more recently (Fig. 2G).^39–42 The interfacial inhibition principle extends beyond topoisomerase inhibitors. A large number of natural products act as interfacial inhibitors not only by binding at the interface of nucleic acids and proteins (α-amanitin, aminoglycoside antibiotics), but also at the interface of polypeptides that form multi-proteins complexes and move around each other to perform their biological function (vinblastine, colchicine, rapamycin, brefeldine A, benzodiazepines, anesthetics).⁶

4. Mechanism of action of topoisomerase inhibitors: trapping of topoisomerase-DNA complexes versus catalytic inhibition

Topoisomerase inhibitors are exquisitely selective and without ambiguity eligible as “targeted therapies”. Clinically relevant Top1 inhibitors (Fig. 3) do not affect Top2, and, conversely, Top2 inhibitors do not trap Top1 enzymes. Furthermore, the inhibitors of bacterial topoisomerases (gyrase and Topo IV) are inactive against host cell topoisomerases (Top2 and Top1), which accounts for their antibacterial potency without impact on the host genome.

The therapeutic mechanism of action of topoisomerase inhibitors revealed another new paradigm for drug action (in addition to interfacial inhibition detailed above): enzyme poisoning rather than catalytic inhibition drives drug activity. This concept first emerged for the antibacterial topoisomerase inhibitors and was demonstrated for the anticancer topoisomerase inhibitors soon after Top1 was discovered as the target of camptothecin.⁴³ Indeed, yeast cells lacking Top1 are immune to camptothecin.^44,45 Similarly, human cancer cells depleted for Top1 become resistant to camptothecin,⁴⁶ implying that Top1 is required for the cytotoxicity of camptothecin, whereas lack of Top1 catalytic activity (as in cells lacking Top1) is tolerated. Biochemical evidence for the requirement of Top1 for the cytotoxicity of camptothecins and non-camptothecin Top1 inhibitors is supported by the formation of Top1cc in cells treated with Top1 inhibitors.^46–49 Induction of Top1cc in biochemical system is actually routinely used to discover and evaluate Top1 inhibitors.^50,51

Genetic evidence for Top2 requirement for the anticancer activity of Top2 inhibitors (Table 1) or for the requirement of gyrase or/and topo IV for the antibiotics (Table 1) has been more difficult to obtain than for Top1 inhibitors because cells lacking type IIA topoisomerases are not viable. Nevertheless, several independent studies established that reducing Top2 (and Top1) levels in tumors minimizes drug activity.^{34,46,52–54} Conversely, the therapeutic activity of doxorubicin is correlated with Top2 overexpression in the case of amplification of the TOP2A locus together with the HER2 locus on chromosome 17 in a subset of breast cancers.⁵⁵ Biochemical evidence for the trapping of Top2-DNA complexes (Top2cc) by anticancer drugs is relatively straightforward and multiple assays can be used to detect the Top2cc not only with recombinant Top2⁵⁶ but also in cells.^57–59

Because the cytotoxic effect of topoisomerase inhibitors requires and is positively correlated with the levels and activity of topoisomerases, assays are being developed to measure the enzymes in patient samples to monitor drug anticancer activity (see Table 2 and section 8).^60,61

The enzyme poisoning mechanism of action first identified for topoisomerase inhibitors has recently been extended to poly(adenosine diphosphoribose) polymerase (PARP) inhibitors. Indeed, as in the case of Top1 inhibitors, knocking out PARP renders cells immune to PARP inhibitors, and treating cells with PARP inhibitors such as olaparib produces PARP1 and PARP2 DNA complexes.⁶² These findings imply that the remarkable activity of PARP inhibitors in breast and ovarian cancers and in Ewing sarcoma can be related, at least in part to the trapping of PARP-DNA complexes when the cancer cells are deficient in homologous recombination repair (BRCA and Fanconi anemia genetic deficiencies).⁶² The fact that three very important classes of anticancer drugs, the Top1, Top2 and PARP inhibitors act by poisoning protein-DNA complexes suggests the possibility that other DNA processing protein complexes, such as transcription factors including the Myc-Max heterodimer could be targeted by interfacial inhibitors. Interfacial inhibitors would lock them on the DNA and thereby initiate a cytotoxic cascade to kill cancer cells.

5. Anticancer Top1-targeted drugs

Camptothecin derivatives are the only FDA-approved Top1-targeted anticancer drugs (Fig. 3A). They are water-soluble semi-synthetic derivatives of the plant alkaloid, camptothecin. The potent anticancer activity of camptothecin was known for ≈ 20 years³³ before the identification of Top1 as its molecular target.^43,47

Topotecan (Hycamtin^®) is routinely prescribed for ovarian cancer, and recurrent small cell lung cancer (SCL).⁶³ Irinotecan (Camptosar^®, Campto^®) is widely used in gastrointestinal (colorectal and gastro esophageal) malignancies.⁶³ Topotecan and irinotecan are also used in primary brain malignancies (glioblastomas), sarcomas, and cancers of the cervix. Irinotecan is a prodrug, which is readily hydrolyzed to its active metabolite, SN-38 by carboxyl esterase (Fig. 3A).

Dose-limiting toxicities are myelosuppression for both topotecan and irinotecan and diarrhea for irinotecan. Bone marrow toxicity is common to all other classical cytotoxics and is probably related to the high proliferative index of bone marrow cells and to cell death priming.⁶⁴ Severe diarrheas are only observed with irinotecan and their mechanism is not fully understood. It has been related to the hepatic elimination of SN-38 and its glucuronated metabolite that produce high intestinal concentrations of SN-38.⁶³

In spite of their potent anticancer activity, all the camptothecins (Fig. 3A) suffer from well-defined limitations.³ In addition to their dose-limiting toxicity, which prevents the use of curative doses,⁶⁵ camptothecins are rapidly inactivated by E-ring opening (Fig. 3A). Indeed, the E-ring α-hydroxylactone spontaneously hydrolyzes within minutes at physiological pH to camptothecin carboxylate, which is sequestered by its high affinity binding to serum albumin. Irinotecan and topotecan are also substrates for the drug efflux transporters, especially ABCG2.⁶⁶ Finally, both irinotecan and topotecan are formulated for IV administration and oral formulations are not been pursued because of the intestinal toxicity of irinotecan.

To avoid the dose-limiting toxicity of camptothecins and their short half lives and to reduce normal tissue toxicity while increasing drug delivery to tumors, camptothecins have been conjugated to a macromolecular core as in etirinotecan pegols (NKTR-102). NK-102 is in advanced clinical development (Phase III) for ovarian, breast and colon cancers. The macromolecular conjugation allows slow drug release with lower peak concentration, extended half-life (up to 15 days), and enhanced tumor penetration through leaky tumor vasculature. Another comparable approach is liposomal formulation (see section 8.3).

The chemical instability of camptothecins has been impossible to overcome by simple semi-synthetic derivations of any of the camptothecins including topotecan and irinotecan derivatives.^3,33 The intact α-hydroxylactone E-ring is indeed critical for the binding of camptothecins to the Top1-DNA cleavage complex.^35–38

Two families of non-camptothecin Top1 inhibitors that overcome the E-ring instability of camptothecins are in clinical development (Fig. 3B).^4,67 The indenoisoquinolines were discovered by screening the NCI Developmental Program drug database for compounds producing cytotoxicity profiles highly correlated with camptothecin across the 60 diverse cancer cell lines of the NCI drug screen [the NCI-60⁶⁸].^3,69,70 Two derivatives are currently in clinical trials indotecan (LMP400) and indimitecan (LMP776). They are developed by the NCI and Purdue University, and licensed to Linus Oncology. In addition to their chemical stability, the indenoisoquinolines offer several advantages over the camptothecins. They target additional genomic sites, their cleavage complexes are markedly more persistent than for camptothecins,^51,71 they overcome multidrug resistance drug efflux pumps⁵¹ and they produce less bone marrow suppression at equal antitumor activity ⁷².

The other non-camptothecin in clinical trials is 8,9-dimethoxy-5-(2-N-methylaminoethyl)-2,3-methylenedioxy-5H-dibenzo[c,h][1,6]naphthyridin-6-one (Genz-644282; SAR402674) (Fig. 3B), was developed from a structure-activity relationship conducted around the dibenzo[c,h][1,6]naphthyridin-6-one compound family.⁸⁰ Genz-644282 has equivalent or superior activity in xenograft models compared with standard drugs and has a favorable cytotoxic profile in vitro in bone marrow and tumor cell colony forming assays.^81,82 The compound was licensed to Genzyme by Rutgers University and is now in the drug development portfolio of Sanofi (SAN).⁶⁷ Genz-644282 has comparable Top1-targeting activity as the indenoisoquinolines.⁷³ Both the indenoisoquinolines (LMP776 and LMP400, indimitecan and indotecan) and Genz-644282 appear active and relatively well tolerated in Phase I clinical trials. ⁸³

6. Anticancer Top2-targeted drugs

All the drugs shown in Figure 3 (panel C–F) inhibit Top2 by targeting Top2cc and inhibiting their religation,^1,5,7,11,74 most likely through interfacial inhibition (see section 3 and Fig. 2). They offer a broad spectrum of chemical diversity,¹¹ potency,^75,76 sequence selectivity,³¹ and ability to trap concerted Top2cc.^5,29,75–77 We refer to “concerted Top2cc” as those where both strands of the DNA are cleaved simultaneously with a canonical 4 base pair overhang stagger (Figs. 1 & 2).

The chemical diversity of Top2-targeted drugs was recently reviewed, and we invite the reader to consult the excellent overview of C. Bailly.¹¹ We will focus on the clinical use and chemical biology of prototypical Top2-targeted drugs and the drugs in clinical trials.

The anthracycline daunorubicin (daunomycin) was discovered in the 1950’s from Streptomyces soil bacteria as an extremely potent anticancer drug. It remains used today primarily for the treatment of acute leukemia.⁷⁸ Doxorubicin (adriamycin), another bacterial toxin, was discovered soon after daunorubicin and is more widely used.⁷⁸ It is active in first line therapy for breast cancers, bone and soft tissue sarcomas, bladder cancers, anaplastic thyroid cancer, Hodgkin’s and non-Hodgkin’s lymphomas and multiple myeloma.⁷⁹ Epirubicin (4’-epi-doxorubicin), an active isomer of doxorubicin (Fig. 3C) was developed later (FDA approval in 1999) to limit the side effects of doxorubicin, possibly due to its faster elimination. Epirubicin is used in breast, esophageal and gastric cancers.⁷⁹ The molecular pharmacology and mechanism of action of anthracyclines are complex. In addition to their anti-Top2 activity, anthracyclines are potent DNA intercalators and generate reactive oxygen intermediates. Their effects on Top2⁸⁰ first proposed by Kohn and coworkers⁸¹ was demonstrated well after their approval by the FDA as anticancer agents.

The Top2 inhibitory effect of anthracyclines exhibits notable peculiarities. First, because of the very effective DNA intercalation of anthracyclines, the trapping of Top2 cleavage complexes, which is achieved at submicromolar drug concentration decreases as drug concentration increases, resulting in a bell-shape concentration response with lack of trapping of Top2cc at or above 10 µM concentration.^56,82 Second, compared to etoposide, anthracyclines trap Top2cc with high selectivity at limited genomic sites with an adenine at the −1 position (see Fig. 2).²⁹ Third, most of the Top2cc are concerted and correspond to DNA double-strand breaks.⁷⁵ Finally, the reversal to Top2cc is slow upon drug removal, explaining the potent effects of the anthracyclines on Top2.

Besides bone marrow suppression, which is common to all topoisomerase-targeted anticancer drugs, anthracyclines are cardiotoxic. This dose-limiting and cumulative cardiotoxicity was until recently primarily attributed to the generation of reactive oxygen species.⁷⁸ However, a recent study has linked cardiotoxicity to top2β targeting in the nucleus and subsequent mitochondrial damage.⁸³ Although interesting, this possibility will probably require further studies to elucidate whether doxorubicin could damage mitochondria more directly by targeting mitochondrial Top2β.⁸⁴

Etoposide (VP-16; Vepesid^®) (Fig. 3D) is widely used in oncology for a broad range of solid tumors including small cell lung cancers, testicular and germ cell tumors, endocrine tumors, osteosarcomas and Ewing’s sarcomas, neuroblastomas and Kaposi sarcoma. Like doxorubicin, etoposide was developed clinically and approved by the FDA (in 1983) without knowing that Top2 was its molecular target. It is a semi-synthetic demethylepipodophyllotoxin derivative without activity on tubulin (by contrast to the podophyllotoxins). Etoposide stands apart from other Top2cc-targeted drugs for the following reasons. First, it is the most selective Top2cc-targeted drug currently in the clinic. As it does not act as a DNA intercalating agent, the Top2cc produced by etoposide form in a monotonic manner without decrease at high drug concentration.^85–87 Second, the Top2cc trapped by etoposide are frequently uncoupled (“non-concerted”) with the majority being single-strand breaks instead of the expected double-strand breaks during concerted inhibition^86,87 (see beginning of this section), suggesting that etoposide traps Top2 homodimers asymmetrically with a single drug molecule bound into one of the two breaks (see Fig. 2G but with a single drug molecule and only one of the homodimers trapped in the cleavage complex). Third, as mentioned in section 3, the base sequence preference of etoposide is determined by the presence of cytosine at position −1.³⁰ Fourth, etoposide produces a very high frequency of Top2cc compared to the intercalating Top2 inhibitors.^88,89 Fifth, the Top2cc trapped by etoposide are readily reversible upon drug wash out, which is different from the anthracyclines. Finally, etoposide traps both Top2α and β very effectively,^59,90 whereas doxorubicin tends to target more selectively cellular Top2α over Top2β.⁹¹ The trapping of Top2β has been related to the induction of secondary leukemia in patient previously treated with etoposide,⁹² and linked to Top2β-mediated DNA translocations (see section 8.2).^93,94

The anthracenedione mitoxantrone (Novantrone^®) (Fig. 3E) was developed as a synthetic analog of anthracyclines at the American Cyanamid Laboratories in the late 1970s⁷⁸ and approved by the FDA in 1996 for prostate cancer. Like anthracyclines, mitoxantrone is both a potent DNA intercalator and Top2cc poison. Its reduced potential to undergo redox reactions compared to doxorubicin may explain its reduced cardiotoxicity.⁷⁸ It is used in first line therapy for pediatric and adult acute leukemia⁷⁹ and second line therapy for breast, prostate cancers and hematological malignancies.⁷⁹ Mitoxantrone is also approved for worsening forms of multiple sclerosis since 2000. Mitoxantrone has to be used with caution because of its risks of cardiotoxicity and secondary leukemia in relationship with Top2β poisoning.⁹⁴

Four Top2-targeted drugs are presently in clinical development: F14512, vesaroxin, C-1311 and XK469.¹¹ F14512 is a demethylepipodophyllotoxin derivative with a spermine side chain (Fig. 3D) targeting cells overexpressing the polyamine transport system (PTS).¹¹ F14512 also binds Top2cc more persistently than etoposide probably because of the DNA binding of its spermine moiety. Voreloxin is an intercalative quinolone derivative in Phase II–III clinical development in combination with cytarabine for relapsing and refractory acute myeloblastic leukemia.⁹⁵ DNA intercalation is important for its activity. C-1311 (Symadex; Fig. 3F) is an iminodazoacridinone derivative with tight DNA interactions both by DNA intercalation and possibly alkylation (reviewed in ¹¹). The fourth Top2-targeted drug in clinical trials is the quinoxaline R(+)-XK469 (Fig. 3F), which has been reported to target Top2β specifically.⁹⁶ However, its mechanism of action is complex with reported inhibition of protein kinases such as MEK, ERK and cdk1 (see ¹¹).

ICRF-187 (Dexrazoxane) (Fig. 3G) differs from the other Top2-targeted drugs because it acts as a catalytic inhibitor rather than by trapping Top2cc.^1,5,12,97 It is not used as a cytotoxic anticancer agent but as a modulator of anthracyclines’ cardiotoxicity⁷⁸ and to treat extravasations resulting from intravenous anthracycline injections. Merbarone (Fig. 3G) is another catalytic inhibitor of Top2 useful for cellular and mechanistic studies.¹² Finally, recent studies suggested that sodium salicylate, aspirine’s active component can act as catalytic Top2 inhibitor.⁹⁸

7. Top2-targeted antibiotics

Bacterial type IIA topoisomerases (Fig. 1 and Table 1) [for details see ^5,9,99,100] have been the target of antibiotics since the discovery of the antibacterial activity of novobiocin, coumermycin and nalidixic acid in the 1960s. Quinolones target the GyrA subunit of gyrase and the ParC subunit of Topo IV⁵ by interfacial inhibition^39–41 (see section 3⁶). Since coumermycin was never developed for clinical use and novobiocin has been withdrawn from the market, there are presently no clinical antibiotic targeting the GyrB subunit of gyrase (Table 1).⁵.

Prokaryotic Top2s are excellent targets because: 1) they are essential in all bacteria; 2) cleavage complexes are bactericidal (not just bacteriostatic); 3) their targeting does not affect host human enzymes (their selectivity is at least three order of magnitude higher for prokaryotic over eukaryotic enzymes); 4) the high degree of homology between gyrase and Topo IV⁵ enables the targeting of both enzymes and therefore the killing of a broad spectrum of bacteria with a single drug.

Quinolones (Fig. 3H) are totally synthetic and are among the most successful antimicrobials both clinically and economically.⁹ The first quinolone was discovered 50 years ago as an impurity in a batch of chloroquine,¹⁰¹ 14 years before the identification gyrase as its molecular target.¹⁰² Several generations of fluoroquinolones have evolved since the 1960s to extend their activity from Gram-negative urinary infections to Gram-positive bacteria and to a broad range of infections including anaerobic infections and multidrug-resistant tuberculosis (Fig. 3H).⁹

8. Challenges and speculations

This last section is a selected list of questions, challenges and possible answers regarding topoisomerase biology and drug targeting. Topics are treated independently from each other and can be read one at a time. They are also outlined in Table 2.