TOP3A coupling with replication forks and repair of TOP3A cleavage complexes

ABSTRACT

Humans have two Type IA topoisomerases, topoisomerase IIIα (TOP3A) and topoisomerase IIIβ (TOP3B). In this review, we focus on the role of human TOP3A in DNA replication and highlight the recent progress made in understanding TOP3A in the context of replication. Like other topoisomerases, TOP3A acts by a reversible mechanism of cleavage and rejoining of DNA strands allowing changes in DNA topology. By cleaving and resealing single-stranded DNA, it generates TOP3A-linked single-strand breaks as TOP3A cleavage complexes (TOP3Accs) with a TOP3A molecule covalently bound to the 5´-end of the break. TOP3A is critical for both mitochondrial and for nuclear DNA replication. Here, we discuss the formation and repair of irreversible TOP3Accs, as their presence compromises genome integrity as they form TOP3A DNA-protein crosslinks (TOP3A-DPCs) associated with DNA breaks. We discuss the redundant pathways that repair TOP3A-DPCs, and how their defects are a source of DNA damage leading to neurological diseases and mitochondrial disorders.

1. Introduction

Topoisomerases are ubiquitous enzymes that adapt DNA topology to a wide range of fundamental metabolic processes both in the nuclear and mitochondrial genomes [1]. Humans possess two Type IA topoisomerases, topoisomerase IIIα and topoisomerase IIIβ, designated as TOP3A and TOP3B, respectively. TOP3 enzymes are present and among the most conserved proteins in all domains of life [2]. The universal presence of TOP3A in insects and vertebrates, as well as genetic and biochemical evidence, suggests that TOP3A is indispensable in fundamental processes involving DNA metabolism and folding, namely in the resolution of double Holliday junctions in the nucleus and the replication of mitochondrial DNA (mtDNA), thereby making TOP3A an essential gene in all living organisms [2].

A growing number of reported human patients with TOP3A mutations, missense and truncating variants present a combination of Bloom syndrome-like (BS) disorder (similar to pathogenic variants in BLM, a nuclear binding partner of TOP3A), and primary mitochondrial diseases including dilated cardiomyopathy, mtDNA depletion in muscles, progressive external ophthalmoplegia, axonal sensory-motor neuropathy and bilateral ptosis [3–8]. Based on recent cellular and biochemical evidence, the clinical outcome of pathological TOP3A variants depends on the degree of loss of TOP3A catalytic activity; with milder variants causing adult-onset mitochondrial disease and more severe variants causing a Bloom syndrome-like (BS) disorder with mitochondrial dysfunction in childhood [7]. TOP3A mutations are also found in patients with different cancer types including melanoma, hepatocellular carcinoma, stomach adenocarcinoma, colorectal adenocarcinoma, glioblastoma [9]. These mutations consist mostly of missense variants (16 missense, 6 truncating, and 1 splice variants among 23 mutations in all over the regions of TOP3A) [9].

Here we first discuss the two isoforms of human TOP3A, as they related to the subcellular localization of TOP3A both in the nucleus and mitochondria, and TOP3A fundamental functions in both cellular compartments. We highlight recent advances on the roles of TOP3A in nuclear and mitochondrial DNA replication. We summarize the tools to identify the formation of irreversible TOP3A cleavage complexes (TOP3Accs) and discuss the molecular mechanisms by which toxic TOP3Accs damage the genome and the multiple pathways that repair those genomic lesions to ensure genome integrity.

2. Two isoforms of human TOP3A

The TOP3A gene encodes two isoforms from two alternative translational initiation codons. Initiation at the first methionine (ATG) produces a 1,001 amino acids-residue protein, referred to as the long TOP3A isoform (1–1001). Translation initiation at the second methionine (ATG) yields a 976 amino acids-residue protein, referred to as the short TOP3A isoform (26–1001) (Figure 1a). The long isoform contains a mitochondrial targeting signal (MTS) at its amino terminus, followed by the common catalytic core region and a nuclear localization signal (NLS) at the carboxyl terminus of the protein. It is predominantly localized in mitochondria as well as in the nucleus (Figure 1b). The short isoform lacks the MTS at the amino terminus, and it is present exclusively in the nucleus [10]. Other metazoans including Drosophila also have the two TOP3A isoforms with dual subcellular localization (nuclear and mitochondrial) [11]. However, the regulatory mechanisms for the usage of second methionine (ATG) of TOP3A remain to be addressed.

Figure 1.

Isoforms and localization of human TOP3A.

(a) Domain organization of the human TOP3A isoforms 1. The catalytic tyrosine residue (Y362) and the arginine residue (R364) are indicated above the diagram in pink and orange, respectively. Human TOP3A has two, long and short, isoforms. The long and short isoforms are 1001 and 976 amino acids in length. The long isoform contains the mitochondrial targeting signal (MTS) at its N-terminus (1-25 aa) and its nuclear localization signal (NLS) at the C-terminus. The long isoform is localized in both the nucleus and mitochondria. The short isoform lacking the MTS but retaining the NLS is localized exclusively in mitochondria. ZNFs represents the zinc finger motifs of TOP3A. (b) Immunofluorescence microscopy showing the presence of TOP3A foci in both the nucleus and mitochondria in human cultured U2OS cells. TOMM20, a mitochondrial surface protein, is used as marker of mitochondrial network staining. Inset: zoomed view of indicated dash-lined area.

The type IA core region contains the four type IA topoisomerase domains along with the active site tyrosine at amino acid residue 362 in human (Y362), and the carboxyl terminal regions contain several potential zinc finger motifs (ZNFs) (Figure 1a) [1,12]. Using immunofluorescence microscopy in cultured human cells, we recently demonstrated that cells ectopically expressed FLAG-tagged full-length (1001aa) TOP3A localizes to both nuclei and mitochondria (Figure 1b) [13].

In the nucleus, human TOP3A forms and acts as a complex together with the RecQ-family Bloom helicase (BLM) and the scaffolding OB-fold RecQ-Mediated genome Instability proteins (RMI1 and RMI2), collectively called the BTRR “dissolvasome” complex [2]. The “dissolvasome” is required for the dissolution of Holliday junctions arising during DNA homologous recombination (HR) to prevent genetic crossovers [14]. This complex has also been shown to catalyze the resolution of a wide range of complex substrates, such as D-loops, late-replication intermediates, and catenated DNA. It has also been proposed to function as a repair complex in response to replication defects and to restart stalled replication forks [15,16]. Besides BLM, TOP3A also acts in coordination with other helicases such as FANCM at stalled replication forks and Plk1-interacting checkpoint helicase (PICH) during mitosis [17,18].

In mitochondria, human TOP3A acts without BLM and the RMI1/2 proteins, which are excluded from mitochondria. TOP3A has been shown to act as an essential decatenase resolving hemicatenane structures (mtDNA circles linked to each other by single-stranded links) arising at the end of mitochondrial DNA (mtDNA) replication when newly replicated mtDNA molecules remain intertwined at the mtDNA replication origin [4]. In addition, two recent studies revealed that TOP3A is required for the initiation and elongation steps of mtDNA replication, suggesting that TOP3A is the main replicative topoisomerase in mtDNA replication [19,20].

3. TOP3A and nuclear DNA replication

Recent evidence supports multiple roles of the “dissolvasome” (BTRR complex) at replication forks to maintain genome integrity by resolving precatenanes behind replication forks, and promoting efficient replication restart (Figure 2). The BLM dissolvasome acts in replication restart independently of its function in suppressing sister chromatid exchanges (SCEs), given that the most characteristic feature of Bloom syndrome (BS) cells is a high incidence of SCEs. Previous studies revealed that BS cells display a slower progression through S-phase and accumulate abnormal replication intermediates, and that BLM might help repair and restart stalled or damaged replication forks [21–24]. Cells from BS patients are hypersensitive to hydroxyurea (HU), a ribonucleotide reductase inhibitor, and camptothecin (CPT), a TOP1 poison, that cause replication fork stalling [25] and BLM localization to these stalled forks [26,27]. Mechanistically, BLM is phosphorylated in response to replication associated double-strand breaks produced by CPT and dissociates from TOP3A in a DNA replication-dependent manner [25]. BLM-deficient cells exhibit reduced replication fork velocity and increased fork pausing [28]. Moreover, the dissolvasome has been implicated in replication fork progression, as depletion of RMI1 leads to a reduction in replication fork rate and a failure to recover from replication fork arrest [29]. Therefore, it is plausible that the TOP3A topoisomerase component of the BLM dissolvasome complex is critical for normal DNA replication. However, whether TOP3A always requires its binding partners BLM and the RMI proteins for this function remains an open question.

Figure 2.

Substrates and sites of action of TOP3A in replication.

Green circle represents TOP3A and the newly synthesized replicating DNA are indicated in red lines in each panel. The boundaries of topologically constrained domains are schematized as black rectangles with anchor signs in panel a, b, and d. (a) During replication elongation, replication forks generate positive supercoiling (Sc+) ahead of the translocating replisome (blue triangle), which is removed by TOP1 and TOP2A (not shown). When the replisome swivels due to the twisting force, positive supercoiling diffuses and redistributed behind the replisome and generates precatenanes with single-stranded DNA segments that are sites of action of TOP3A 1, 31. (b) Proposed action of TOP3A on reversed fork structures resulting from stalling of replisomes. In this situation, fork reversal may happen and TOP3A can bind to template strands of nascent single-stranded DNA segments and may promote fork reversal 13. (c) TOP3A-mediated resolution of DNA hemicatenanes, which are late replication intermediates arising during replication. TOP3A resolves hemicatenanes by passing a single strand of DNA through a break made in another DNA strand (single-strand passage mechanism). TOP3A may perform this function with its partner proteins, BLM and RMI1 and 2, as the dissolvasome complex in nucleus 2. TOP3A alone can also resolve hemicatenanes structure during the last step of mitochondrial replication 4. (d) TOP3A is involved in the dissolution of “double Holliday junction”, an important intermediate of recombination during completion of replication.

A recent study in human cells shows that TOP3A is required for replication fork progression in nuclear genomes [13]. The functions of TOP3A at replication forks have been established at single molecule level by DNA combing assay, where TOP3A-depleted human cells show reduced replication velocity, increased origin firing and shorter inter-origin distances [13]. This suggests that TOP3A activity is indispensable to maintain normal DNA replication in human cells. In bacteria, the association of Topo III with replication fork has been revealed by its colocalization with the beta subunit of DNA Pol III holoenzyme using fluorescence microscopy [30]. Recently, TOP3A has been found at active replication foci in human cells, as TOP3A foci were found colocalized with CDC45, a component of the CMG helicase [13]. The physical presence of human TOP3A at DNA replication forks has also been observed by iPOND (isolation of protein on nascent DNA) assay [13]. In E. coli, Topo III also interacts physically and functionally with the DNaX core complex that tether two DNA polymerases [30]. The4 DnaX complex stimulates the ability of Topo III to unlink catenated and precatenated DNA rings. Yet, the existence and nature of interactions of human TOP3A with DNA polymerases for efficient nascent chain elongation remains to be addressed.

A major topological aspect associated with DNA replication, and which is relevant to the activity of TOP3A is its role in removing precatenanes as replication forks rotate around the DNA axis in response to excessive positive supercoiling ahead of replication forks (Figure 2a) [1,2,31]. Early biochemical studies with purified E. coli Topo III (TOP3A in human) showed that Topo III is capable in both decatenation of sister chromosomes for removing precatenanes and nascent chain elongation during DNA replication [32–34]. More recently, in vivo studies in E. coli confirmed that Topo III can remove precatenanes behind replication forks during DNA replication (Figure 2a) [30,35]. Thus, in addition to TOP1 and TOP2A, TOP3A contributes to the complete unlinking of the parental strands during replication (see Figure 1 in [1]). Replication elongation causes accumulation of positive supercoiling ahead of the replication fork and negative supercoiling behind it. The combined action of TOP1 and TOP2A can remove this positive supercoiling, whereas negative supercoiling can be removed by TOP1, TOP2A or TOP3A. In addition, in response to positive supercoiling, forks can rotate to diffuse the positive supercoiling that accumulates ahead of the forks. In this situation, precatenanes accumulate behind replication forks, linking newly replicated sister chromosomes together (Figure 2a).

TOP3A is incapable of cleaving double-stranded DNA (dsDNA). Even in dsDNA substrates, it needs a single-stranded segment to act. This explains why TOP3A can only relax hypernegatively supercoiled DNA, which tends to form single-stranded segments. Biochemical experiments also show strong decatenation activity on plasmids substrates with small gaps [36]. In E. coli, it has been suggested a long time ago that the only potential site of action for a type I topoisomerase to decatenate replicated sisters is at single-stranded gaps [37]. During DNA replication, gaps form in the nascent lagging strand during Okazaki fragment synthesis. In such case, it is plausible that the DNA structures containing the Okazaki fragments in the lagging strand are preferred regions for TOP3A to perform its decatenation reactions. The ssDNA template in this gap is coated with the single-stranded biding protein SSB, and E. coli Topo III is known to have a strong interaction with SSB [38,39]. Topo III binds to single-strand gaps on nascent duplexes near replication forks via its interaction with SSB [30]. Topo III-SSB interaction likely plays an important role in localizing Topo III to the replisome, although not solely. The localization of Topo III to the replication fork is consonant with its activity. Similarly, recent observations indicate that gaps can form in the nascent leading strand as well because of polymerase pausing and continued unwinding by DnaB [40]. Of note, the BTR dissolvasome complex has been shown to sense replication protein A (RPA)-coated single-stranded DNA, which may establish its role in restarting stalled replication forks [16].

Another DNA structure where TOP3A may function is “reversed fork” (Figure 2b). Replication fork could be reversed upon replication fork stalling. A recent study using human cells suggested the role of TOP3A in reversed fork regions, supported by the observations of the recruitment of TOP3A to stalled/reversed replication forks after HU treatment as evidenced by immunofluorescence microscopy [13]. This is plausible because the formation of single-stranded DNA (ssDNA) is associated with extensive fork reversal arising from 3’-end resection, which is detectable only by native BrdU labeling [41]. An induction of native BrdU foci after HU treatment is indicative of ssDNA within nascent DNA strands. In TOP3A-depleted cells, the dramatic reduction of HU-induced ssDNA foci suggests that TOP3A is required for the formation of ssDNA in newly replicated DNA of daughter molecules [13]. The localization of TOP3A in high density ssDNA regions in stalled/reversed replication fork suggests the importance of TOP3A in the formation of ssDNA in reversed fork regions during the processing of stalled replication fork [13]. However, whether the depletion of known critical factors involved in replication fork reversal, such as SMARCAL1 (SWI/SNF-related, matrix-associated, actin-dependent, regulator of chromatin, and subfamily A-like 1), ZRANB3 (zinc finger, RAN-binding domain containing 3) and HLTF impact TOP3A recruitment remains to be investigated.

The activities of TOP3s can also be relevant to other topological aspects associated with DNA replication. At late stages of replication (replication termination), hemicatenane structures (a late replicative intermediate), where two sister molecules are still connected by segments of unreplicated DNA, form (Figure 2c) [42]. Such structures typically form when replication forks converge. The accumulation of positive supercoils in the short unreplicated segment results in converging fork stalling and rotation. At this stage, these hemicatenane structures are inaccessible to TOP1 or TOP2, and TOP3A is the only topoisomerase capable of resolving the hemicatenanes (Figure 2c) [2]. Similar structures can also be generated from recombination intermediates “double Holliday junctions” (Figure 2d) [2,43]. Further studies are warranted to fully elucidate the range of activities of TOP3A in removing hemicatenanes during replication in human cells.

After the completion of replication, TOP3A has also been shown to generate positively supercoiled segments in replicated DNA in coordination with PICH helicase [18]. TOP2A mediates the decatenation of this supercoiled substrate to enable the segregation of sister chromatids during anaphase. Of note, the dissolvasome may induce torsional stress by introducing entanglements between sister chromatids [2]. In yeast, Rmi1 (RMI1 in human) has been proposed to play a role in the cohesion of sister chromatids, a function that is independent of other proteins in Cohesin complex [44,45].

4. TOP3A and mtDNA replication

Human mtDNA is a closed circular double-stranded DNA molecule of 16,569 base pairs in length [42]. The two strands of mtDNA differ in their base composition and are designated as heavy (H) and light (L) strand. A human cell contains several hundred to few thousand copies of mtDNA. Human mtDNA is replicated independently of the cell cycle within the mitochondrial matrix and segregated between mitochondria. mtDNA encodes 13 mRNA for the OXPHOS machinery, 22 tRNA and two rRNA for the translation of the proteins encoded by its genome. Human mtDNA contains two strand specific replication origins, termed O_H and O_L, oriented in opposite directions. O_H resides in the mtDNA noncoding region (NCR). O_L is located outside the NCR, ~11 kb downstream of O_H [42,46]. The exact mechanism of mtDNA replication is still not fully understood. Two debatable mechanisms of mtDNA replication have been proposed: i) the strand displacement model and ii) ribonucleotide incorporation throughout the lagging strand (RITOLS) model, termed as “bootlace” mechanism [47].

In mitochondria, the presence of two specific topoisomerases, TOP3A and TOP1mt, is well documented [10,48]. Both TOP3A and TOP1mt have been implicated in the mtDNA topology in human cells [19]. Genetic and biochemical evidence suggest a division of labor between TOP3A and TOP1mt with TOP1mt removing both positive and negative supercoiling and TOP3A removing negative supercoiling behind the fork during replication elongation.

The mitochondrial isoform of TOP3A has been well studied in Drosophila, where lack of this isoform leads to the loss of mtDNA, neurological diseases, infertility and aging [11,49]. Subsequent studies in human cells showed that TOP3A is required for proper segregation of mtDNA after completion of replication [4] (see Figure 2c). More recently in human cells after silencing the TOP3A gene and using mitochondria specific TOP3A constructs, TOP3A has been implicated in the initiation and elongation steps of the mitochondrial genome replication [20].

During the elongation phase of mtDNA replication, the accumulation of positive supercoiling ahead of the replisome may facilitate the rotation of the replisome, resulting in the formation of precatenanes (Figure 2a), which can be removed by TOP3A. Precatenanes removal by TOP3A can be promoted by ssDNA regions, either short or long, present in the lagging strand template. Type II topoisomerases (TOP2A or TOP2B) also remove the unresolved precatenanes after the completion of DNA replication. Additionally, there are contrasting reports about the presence and functions of TOP2A and TOP2B in mitochondria in human cells [4,19,50,51]. Indeed, recent studies using 2D agarose gel electrophoresis show the accumulation of mtDNA replication intermediates throughout the mitochondrial genome in TOP3A-depleted human cells, suggesting the role of TOP3A in mtDNA replication elongation [19,20].

Studies in mouse cells showed that the initiation of mtDNA replication from OriH can produce either 7S DNA (to form the D-loop) or initiate full-length mtDNA replication in strand displacement mtDNA replication [52]. Pulse labeling studies of replicating mtDNA have suggested that the synthesis of the D-loop removes negative supercoils to produce mtDNA in an open circular form [52,53]. Previously, TOP1mt has been shown to bind at the regions of D-loop, suggesting its role in controlling topology during mtDNA replication initiation [54]. A recent study has shown the loss of TOP3A reduces the extention of 7S DNA, suggesting the role of TOP3A in replication initiation [20]. According to the strand displacement model, mtDNA replication is asymmetric in nature [55,56]. Because the synthesis of L strand lingers behind that of the H strand, extensive single-stranded gap can be generated in the D-loop region. TOP3A performs strand passage in sites of these single-stranded gaps in the interlocked circular molecules to remove negative supercoiling.

Due to the asymmetric nature of mtDNA replication, the termination of replication for both strands has been suggested to take place in the mtDNA noncoding regulatory region where TOP3A has been implicated in the decatenation of mtDNA [4,57]. The loss of TOP3A activity results in the accumulation of catenated mtDNA replication termination intermediates, resembling hemicatenanes with the O_H regions implicated as a primary site of mtDNA replication termination [4]. O_H can also act as a replication termination region under physiological conditions [58]. However, whether these hemicatenane structure is formed physiologically in vivo, remains an open question.

The replication (and transcription) machineries of mtDNA are related to bacteriophage proteins [59]. The replisome in mitochondria includes the mitochondrial DNA polymerase DNA polymerase gamma (POLG), the replicative helicase TWINKLE and the RNA polymerase POLRMT, all of which show homology to proteins of the T-odd lineage of bacteriophages. Other essential proteins of the mtDNA replication machinery, such as mitochondrial single-stranded DNA-binding protein (mtSSB) and topoisomerases, are either of bacterial origin or shared with the nucleus [46]. A recent study using the BioID2 approach shows that the mitochondrial isoform of TOP3A localizes and interacts with the replication proteins POLG, TWINKLE, TFAM and MGME1 [20].

Mitochondrial DNA (mtDNA) is packaged into a highly ordered nucleoprotein complex structures, which form mitochondrial nucleoids. Typically, nucleoid contains a single mtDNA molecule, suggesting that each mtDNA acts as independent unit [60]. However human HeLa cells and cardiomyocytes contain nucleoids with multiple copies of mtDNA [61,62]. The accumulation of high levels of aberrant mtDNA replication intermediates explains the presence of multimeric mtDNAs in these cells. TFAM binds to single mtDNA molecule to form the mitochondrial nucleoid, which is the primary nucleoid protein involved in mtDNA packaging [63] in association with mitochondrial membranes [64,65]. During mtDNA replication, the attachment of mtDNA to the inner mitochondrial membrane (IMM) inhibits the formation of topological interlinks between daughter mtDNA molecules, as suggested by the localization of the mitochondrial replicative helicase TWINKLE to the IMM [66]. Importantly, since other members of the dissolvasome (BLM, RMI1, or RMI2) in nucleus do not appear to be localized in mitochondria [4], mitochondrial helicases (TWINKLE, RECQL4, SUV3, DNA2 and PIF1) [67] may work with TOP3A in mitochondria to regulate its activity.

5. Formation of TOP3A cleavage complexes (TOP3Accs)/TOP3A-DNA protein crosslinks (TOP3A-DPCs)

The catalytic reactions of all topoisomerases are transient and reversible. Topoisomerases cleave/break the DNA backbone, forming covalent linkages between their active site tyrosines and DNA, adjust the spatial structure of DNA, and rejoin the DNA cleavage/breaks [1]. Type I topoisomerases cut one strand while type II cut both strands of duplex DNA. Type IA and IB topoisomerases are covalently attached to the 5´- and 3´-phosphate of the cleaved DNA, respectively [1]. The key catalytic intermediates for all topoisomerase catalytic reactions are termed as the topoisomerase cleavage complexes (TOPccs). TOPccs are normally transient and self-reversible in nature. They are formed by the covalent attachment of topoisomerases’ catalytic tyrosine residue to the ends of broken DNA strand(s) [68,69].

TOPccs can be stabilized or trapped on DNA (persistent TOPccs), which are referred to as topoisomerase DNA-protein crosslinks (TOP-DPCs) due to abortive catalytic cycles of topoisomerases that damage the genome if left unrepaired [70]. TOP-DPCs are highly cytotoxic when they are generated by antibacterial and anticancer topoisomerase inhibitors as well as carcinogenic DNA alterations [1,70,71]. TOP-DPCs can also be formed endogenously (abasic sites, mismatches, alkylated bases, DNA breaks), leading to a failure in completion of the enzymatic reaction [1]. TOP3A cleaves one DNA strand generating a TOP3A-linked single strand breaks (TOP3Accs) with a TOP3A molecule covalently bound to the 5´-end of the break (TOP3A-DPCs) (Figure 3a). After passing another DNA strand through the break, TOP3A reseals the ends of the broken DNA and dissociates from DNA. Camptothecins selectively poison TOP1 as well as etoposide and its derivatives poison only TOP2 (both TOP2A and TOP2B) but not TOP1 enzymes. However, no discovery of selective small molecule topoisomerase poisons/inhibitors of type IA topoisomerases till now limit the study to uncover the repair pathways for TOP3Accs.

Figure 3.

Mutation sites of “self-trapping/toxic TOP3A” that forms TOP3A-DPCs.

(a) TOP3A-DPCs are formed at the 5’ ends of DNA strand (5’-terminal DPC). TOP3A binds single-stranded DNA. TOP3A cuts exposed single-stranded DNA regions of duplex DNA by covalently linking its active tyrosine residue (Y) to the 5’ end of the DNA backbone using Mg²⁺ as a cofactor to form TOP3A cleavage complexes (TOP3Accs). TOP3Accs can be converted into TOP3A-DPCs if they persist and fail to reseal themselves or if they are left unrepaired. (b) The catalytic core domain of the type IA topoisomerases is conserved in all organisms. The conserved arginine (R) amino acid residue, close to catalytic tyrosine residue in the catalytic core domain in type IA topoisomerases, is conserved from bacteria to human. Orange bars indicate arginine residues (at 321 in E. coli Topo I, 330 in E. coli Topo III, 335 in S. cerevisiae Top3, 338 in human TOP3B, and 364 in human TOP3A). Mutant (having mutations in arginine 321 site in E. coli Topo I) have had diminished resealing activity in vitro) 74. Mutation in this site by arginine substitution with tryptophan (R364W in human TOP3A) leads to the formation of persistent TOP3A-DPCs 13, comparable to E. coli Topo I and human TOP3B 77. (c) Structure of human TOP3A with the DUF and OB fold domain of RMI1 together (PDB: 4CGY, 4CHT). The indicated domains of TOP3A are colored according to the assignment in panel A. The location of the catalytic tyrosine (Y362) (pink), arginine residue (R364) (orange) and the cofactor Mg²⁺ (green are) are indicated. Zoom in view of showing catalytic tyrosine (Y362) (pink), and proximal arginine residue (R364) (orange)

The core domain of all type IA topoisomerases is conserved from prokaryotes to higher eukaryotes (Figure 3b) [2]. E. coli topoisomerase III (Topo III) is homologous to human TOP3A and TOP3B. Like human, the E. coli bacteria also have two subtypes of type IA topoisomerases, Topo I and Topo III (Figure 3b). Previous studies in bacteria showed that few individual mutations including R321W in the core catalytic domain of E. coli Topo I caused the accumulation of Topo I ccs, leading to cell death [72–74]. A subsequent study by our group reported that the corresponding mutation in human TOP3B (R338W), the other type IA topoisomerase, results in the formation of TOP3B-DPCs in human cells. Additionally, sequence alignment revealed that arginine residue at 364 amino acid position in human TOP3A corresponds to R321 and R338 in E. coli Topo I and human TOP3B, respectively (Figure 3b). By mutating the arginine residue to tryptophane (R364W), which is near catalytic tyrosine residue at 362 amino acid position (Y362) (Figure 3c) we recently showed expression of this TOP3A mutant (TOP3A-R364W) in human cells causes the entrapment/accumulation of TOP3A-DPCs [13]. Cells expressing this mutation displayed replication defects (reduced replication speed, increased asymmetric forks etc.), DNA damage, and genome instability [13]. As E. coli Topo I with mutation in arginine 321 residue has been proposed to have diminished rejoining activity [74], we hereby termed this humanTOP3A mutant as “self-trapping/toxic TOP3A”. In our recent study, we used this “self-trapping/toxic TOP3A” to demonstrate the presence of TOP3Accs by Rapid Approach to DNA Adduct Recovery (RADAR) assays to isolate DNA covalently bound protein adducts. We also used this R364W mutant as a genetic probe to investigate the cellular consequences and the repair pathways of TOP3A-DPCs in human cells [13]. TOP3A foci colocalized with replication foci in toxic TOP3A-expressing human cells, as observed by immunofluorescence microscopy after pre-extraction, confirms their attachment/entrapment in chromatin [13].

Because TOP3A is critical for both nuclear and mitochondrial DNA replication, TOP3A-DPCs are prominent in the replicative phase of the cell cycle. Indeed, our recent study shows that a large fraction of TOP3A-DPCs are formed during S-phase, a lesser extent in G2/M-phase, whereas TOP3A-DPCs are non-detectable in G1-phase, suggesting that TOP3A-DPCs formation are cell cycle regulated [13]. Consequently, toxic TOP3Accs primarily affect replicating cells, resulting a slower growth of toxic TOP3A- expressing human cells. Moreover, a marked decrease in TOP3A-DPCs level after DNA polymerase inhibition by aphidicolin, suggests the association of TOP3A with ongoing replication fork [13]. The detection of formation of mitochondrial TOP3A-DPCs needs to be examined.

6. Repair of TOP3Accs/TOP3A-DPCs

Multiple repair pathways with sequential steps repair TOP-DPCs [1]. DNA transaction machineries including DNA and RNA polymerases can serve to sense and initiate DNA repair [70,75]. Conflicts and collisions between TOP-DPCs and ongoing DNA transactions trigger signals that recruit repair machineries. Because TOP-DPCs are bulky, cells employ various molecular machineries for protein degradation to reduce the size of the protein component and debulk TOP-DPCs to allow the access of downstream repair machineries [1,70,75].

Proteolytic degradation of the protein component has emerged as a key event to repair diverse DPCs [76]. Debulking of TOP-DPC requires proteolytic processing of the topoisomerases covalently attached to DNA. This process can be mediated by the proteasome and/or DPC proteases (non-proteasomal). The ubiquitin-proteasome system degrades a variety of DPCs including TOP1-DPCs, TOP2-DPCs and TOP3B-DPCs [70,75,77,78]. SUMOylation is required for subsequent ubiquitination and proteasomal degradation of DNMT1 DPCs, TOP1-DPCs and TOP2-DPCs [78,79]. In particular, the SUMO ligase PIAS4-mediated SUMOylation and subsequent polyubiquitination by the SUMO-targeted ubiquitin ligase (STUbL) RNF4 induce proteasomal degradation of TOP1- and TOP2-DPCs [78]. Yet, our recent study on TOP3A-DPCs showed that SUMO-Ubiquitin cascade does not apply to TOP3A-DPCs repair as no SUMOylation of TOP3A-DPCs are observed, and the proteasome machinery is dispensable for TOP3A-DPCs repair [13].

Beyond the proteasomal degradation pathway, the DPC protease SPRTN (Wss1 in yeast) is a key protease in the degradation of DPCs including TOP1-DPCs and TOP2-DPCs [80–82]. SPRTN is a replication-associated protease and is a constitutive part of replisomes [81,83]. Additionally, a recent study proposed that SPRTN has been reported to act in replication-independent manner to repair post-replicative DNMT1-DPCs through SUMO-targeted ubiquitylation of DNMT1 adducts mechanism [84]. Reasonably, our recent study shows that SPRTN debulks and repairs TOP3A-DPCs (Figure 4) as TOP3A is associated with replication forks and TOP3A-DPCs are abundant in the replicative phase of cell cycle [13]. Other proteases: such as GCNA (also known as ACRC), Ddi1, and FAM111A have been reported to act in TOP1- and TOP2-DPCs repair [85–88]. However, the role of these proteases in TOP3A-DPCs repair remains to be investigated.

Figure 4.

Repair pathways of TOP3A-DPCs repair.

Overall scheme for the debulking and excision step for the repair of TOP3A-DPCs. The protease, Spartan (SPRTN) proteolyzes (debulk) TOP3A-DPCs. TOP3A-DPCs are polyubiquitylated but not SUMOylated, while other post-translational modifications (PTMs) of TOP3A-DPCs are yet to be discovered. Polyubiquitylation of TOP3A-DPCs may serve as a signaling mechanism for the recruitment of SPRTN at TOP3A-DPcs sites. Nucleic-acid excision pathways for TOP3A-DPCs include excision by the endonucleases MRE11 and CtIP, or excision by tyrosyl-DNA phosphodiesterase 2 (TDP2). TDP2-mediated excision of TOP3A-DPCs requires prior debulking of TOP3A-DPCs by SPRTN to provide access to the tyrosyl – DNA links. The non-proteolytic pathways of TOP3A-DPCs repair includes the MRE11 and CtIP endonuclease. PTMs of TOP3A-DPCs may not needed to instigate MRE11-mediated repair. Yet, other nucleases involved in TOP3A-DPCs repair are unknown and the requirement of PTMs of TOP3A-DPCs for their function remains to be explored. Phase of the cell cycle (S phase versus G2/M phase) are likely determinant of pathway choice for TOP3A-DPCs repair. Thus far, TOP3A-DPCs are repaired by at least two parallel pathways: SPRTN-TDP2 during S-phase and CtIP-MRE11 during G2/M. After debulking and excision step, protein free single-ended double-strand breaks (seDSBs) ultimately require homology-directed recombination (HDR) to ensure replication fork progression.

The debulking of TOP-DPCs appears to be a prerequisite step for the following excision so that excision repair enzymes can access and remove the DNA-linked topoisomerase tyrosyl remnant with nuclease and/or phosphodiesterase activities [70,75]. Cells employ two specialized tyrosyl-DNA phosphodiesterase enzymes (TDPs), TDP1 and TDP2, that directly hydrolyze these phosphotyrosyl bonds [89]. In human, TOP2-DPCs and TOP3B-TDPCs with 5’-phosphotyrosyl DNA linkage are hydrolyzed by TDP2 while TOP1-DPCs with 3’-phosphotyrosyl DNA linkage are excised primarily by TDP1 [43,77,89]. Consistent with these molecular activities, our recent study demonstrates the action of TDP2 in TOP3A-DPCs processing/repair as TOP3A-DPCs are 5’-phosphotyrosyl DPCs (Figure 4) [13]. We also found that the genetic relationship between SPRTN and TDP2 appears to be an epistatic in TOP3A-DPCs repair, implicating a sequential action of SPRTN and TDP2 with debulking of TOP3A-DPCs by SPRTN preceding the excision of the remaining TOP3A-DPCs peptides by TDP2 (Figure 4) [13].

Endonucleases can also resolve TOP-DPCs by cleaving the DNA strand(s) to which the topoisomerases are covalently attached [1]. For example, MRE11 has both as double-strand-specific 3’–5’ DNA exonuclease and single-strand DNA endonuclease activities to initiate DSB repair [90,91]. In human, MRE11 forms complexes with RAD50, and NBS1, collectively called MRN complex. In yeast, Mre11/Rad50/Xrs2 complex (MRN in humans) can excise TOP1-DPCs, TOP2-DPCs and TOP2-like-DPCs (SPO11-DPCs) [92–96]. In vertebrates, MRE11 appears to be essential in processing TOP2-DPCs [97,98]. Our recent study shows the accumulation of TOP3A-DPCs in MRE11-depleted cells, suggesting the involvement of MRE11 in TOP3A-DPCs repair (Figure 4) [13]. Previous studies have also implicated CtIP in TOP2-DPCs removal [99,100]. CtIP, a known interactor of MRN complex, stimulates the TOP2-DPCs removal as well as the endonuclease activity of MRE11 [92–94,101–103]. Regarding TOP3Accs, we found that CtIP depletion in human cells increases TOP3A-DPCs, suggesting the coordination of CtIP and MRE11 in TOP3A-DPCs removal [13]. In our recent study, an increased accumulation of TOP3A-DPCs after either ATM depletion or inhibition, and the epistasis of CtIP and ATM, suggests that ATM senses and regulates TOP3A-DPCs removal along with CtIP [104]. This finding is supported by an established connection between phosphorylated CtIP and MRE11 in DNA end-resection where CtIP phosphorylation is mediated by ATM [104]. Altogether these findings suggest that MRE11, CtIP, and ATM can act in concert for the removal of TOP3A-DPCs [13]. Moreover, the synergistic relationship between MRE11 and TDP2 on TOP3A-DPCs removal suggests the parallel action of TDP2 and MRE11 in TOP3A-DPCs repair (Figure 4) [13]. Future investigations will address whether other nucleases also play a role in repairing TOP3A-DPCs in human cells.

An important remaining as well as challenging question is how cells decide the repair pathway choice for TOP-DPCs. Cellular transactions such as replication, transcription and cell cycle stage are likely to be major determinants for TOP-DPC repair pathway choice [1]. In our study, in the case of TOP3A-DPC repair, at least cell cycle is one of the determining factors for pathway choice, with two parallel pathways: SPRTN-TDP2 acting in S-phase and ATM-CtIP-MRE11 axis working in G2/M-phase of cell cycle (Figure 4) [13]. Other determinants of TOP3A-DPC repair warrant further investigations.

Post-translational modifications, such as ubiquitylation and SUMOylation, of TOP-DPCs have emerged as a signaling mechanism for TOP-DPC repair [1,70,78]. A major signaling mechanism is SUMOylation followed by ubiquitylation that recruits debulking molecular machines, such as the proteasome, and facilitates TOP-DPC repair [78]. SUMOylation of TOP1-DPCs has been proposed as signaling mechanism for TOP1-DPCs repair by SPRTN [76]. In our recent study, while assessing the post-translational modifications of TOP3A-DPCs, we found that TOP3A-DPCs can be polyubiquitylated (Figure 4) [13]. However, no SUMOylated TOP3A-DPCs was detectable as mentioned above, consistent with the absence of SUMO sites for TOP3A by proteomics study [105]. The presence of a ubiquitin binding zinc finger domain (UBZ) in SPRTN [106] suggests that SPRTN may recognize and target ubiquitylated TO3A-DPCs (Figure 4). Indeed, our recent study shows that the UBZ domain-deficient SPRTN mutant fails to resolve TOP3A-DPCs [13]. However, whether the polyubiquitylation signal alone is sufficient to trigger the recruitment of SPRTN and additional factors is currently unknown (Figure 4).

Previously, we identified TRIM41 as E3 ligase for TOP3B-DPCs polyubiquitylation for TOP3B-DPCs repair [77]. However, our studies show that TRIM41 does not appear as an E3 ligase for TOP3A-DPC polyubiquitylation [13]. Further investigations are warranted to identify ubiquitin E3 ligase that signals for TOP3A-DPCs polyubiquitylation and their eventual repair. In addition, it should be noted that the SUMOylation of TOP2-DPCs by SUMO E3 ligase ZATT mediates chromatin remodeling for TDP2 to gain access to TOP2-DPC sites [107]. But this mechanism seems to be irrelevant for TOP3A-DPCs repair as no SUMOylation of TOP3A-DPCs is observed.

After the debulking and excision steps, protein free DNA breaks/termini (PDBs) are generated; which require the repair of the strand breaks. Several mechanisms have been implicated: single strand break repair (SSBR), non-homologous end joining (NHEJ), homologous recombination (HR), single-strand annealing (SSA) or microhomology end-joining (MMEJ) to ensure the restoration of DNA and productive transactions without genomic alterations [1,75]. The contribution of these repair mechanisms to seal TOP3A-DPCs induced breaks warrants future investigations.

7. Further perspectives

Although human TOP3A play essential roles in DNA replication of the nuclear and mitochondrial genomes, the detailed in vivo mechanisms of its functions are still poorly understood. It is also important to know whether TOP3A plays a role in other DNA transactions like transcription, translation and telomere maintenance. Recently, the depletion of TOP3A has been reported to reduce the expression of some mitochondrial transcripts, but not all [19,20]. Whether the effect of TOP3A on transcription is global and either direct or indirect (i.e. via replication) remains unknown. A detailed mapping of TOP3A binding genome wide is likely to help understand the molecular processes implicating TOP3A. In this context, our TOP3A “toxic mutant” (R364W) can be an important and useful tool.

TOP3A-DPCs may not pathological per se; cells may employ them for important physiological functions such as anchoring the DNA to nuclear and mitochondrial structures. Such TOP3A DPCs would have to be removed by various machinery to deal with endogenously produced TOP3A-DPCs and the defects in those machinery may have severe consequences. Therefore, it will be important to use relevant cell types and models in assessing TOP3A-DPCs induction by environmental factors and in studying repair mechanisms.

Given the importance of TOP3A in cellular replication, the discovery of small molecule inhibitors/poisons of TOP3A is likely to contribute to the study of the cellular function and regulation of TOP3A.

Funding Statement

L.K.S and Y.P are supported by the Center for Cancer Research, the Intramural Program of the National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892 [Grant Z01 BC Z01 BC 006161-17 and Grant Z01 BC 006150-19].

Disclosure statement

No potential conflict of interest was reported by the author(s).