Topoisomerases are ubiquitous enzymes that regulate the topology of DNA during replication, transcription, chromosome condensation, segregation, and other nucleic acid transactions (Chen et al., 2013; Pommier et al., 2016; McKie et al., 2021; Pommier et al., 2022). Topological changes in DNA are accomplished by type I topoisomerases, which act by generation of single-strand breaks, and type II enzymes, which accomplish topological transformations via double-strand breaks. There is a wealth of information concerning the biochemical properties of these enzymes, and recent reviews include mechanisms of type 1A enzymes that use a strand passage mechanism (Dekker et al., 2002; Dasgupta et al., 2020), type 1B topoisomerases (Champoux, 2001; Pommier et al., 2016; Seol and Neuman, 2016), and type II enzymes (Vos et al., 2011; Chen et al., 2013; Bush et al., 2015; Hauk and Berger, 2016). The DNA cleavage mechanisms of topoisomerases are summarized in Figure 1. All organisms encode both type 1 and type II topoisomerases, and eukaryotes use type 1A, type 1B, and type II topoisomerases. Human cells have six distinct topoisomerases TOP1, mitochondrial TOP1 (TOP1MT), TOP2α, TOP2β, TOP3α, and TOP3β (Champoux, 2001; Vos et al., 2011; Pommier et al., 2016; Pommier et al., 2022).

Human topoisomerases participate in a broad spectrum of cellular functions. Some of the functions require more than one topoisomerase to cooperate, while the others uniquely require specific topoisomerases. For example, both TOP1 and TOP2α are important for replication, and all the nuclear topoisomerases participate in transcription regulation. By contrast, there are also very specific roles that are unique to specific topoisomerases. For example, TOP1MT does not play roles in nuclear functions (Baechler et al., 2019), and TOP3β uniquely can act on RNA and DNA (Ahmad et al., 2017). Of particular interest is the action of TOP2α. This protein functions in a variety of processes including replication and transcription and also plays specific roles in meiosis (Zhang et al., 2014). Because only type II topoisomerases can catalyze catenation and decatenation of intact double-stranded DNA (Kreuzer and Cozzarelli, 1980; Sundin and Varshavsky, 1981), these enzymes are uniquely required to separate replicated chromosomes that have maintained catenation (Baxter, 2015). A variety of studies specifically showed that TOP2α but not TOP2β is the enzyme that plays this unique role (Linka et al., 2007; McClendon et al., 2008; Gonzalez et al., 2011). TOP2α also plays unique roles in both chromosome structure and chromosome condensation (Maeshima and Laemmli, 2003; Shintomi and Hirano, 2018; Paulson et al., 2021). The utilization of specific topoisomerases in various biological processes remains an area of active investigation, and post-translational modifications are likely to be critical for the recruitment and enzyme activity of specific required enzymes. The potential roles of SUMOylation in these processes are discussed in SUMOylation in Regulation of Topoisomerases.

The induction of DNA damage by topoisomerases arises from the intermediates of the topoisomerase reaction (Figure 1). While the enzyme normally catalyzes DNA rejoining after performing topological changes, a variety of conditions can prevent the rejoining. These resulting lesions, which consist of a topoisomerase covalently bound to DNA via their normal phosphotyrosyl intermediates, are referred to here as TOP-DPCs (topoisomerase DNA–protein crosslinks). The best studied examples of TOP-DPC induction are small molecules that are anti-cancer topoisomerase-targeting drugs including camptothecins for TOP1 and etoposide and doxorubicin for TOP2. These drugs prevent DNA rejoining by binding to the enzyme–DNA interface and result in their respective TOP-DPCs (Pommier, 2006; Nitiss, 2009b). There are a variety of other conditions that can lead to accumulation of TOP-DPCs. One prominent example is the presence of DNA damage such as oxidative lesions, abasic sites and breaks in DNA, or DNA damage caused by carcinogens. Eukaryotic topoisomerase I appears to be particularly prone to trapping by DNA damage, and abasic sites also efficiently trap TOP2 in vitro (reviewed by Nitiss et al. (2019)). In addition, mutations in topoisomerases can impair the enzyme’s ability to carry out re-ligation (Levin et al., 1993; Stantial et al., 2020; Boot et al., 2022).

The diverse ways in which topoisomerases can lead to DNA damage, including a variety of spontaneous mechanisms, suggest that cells have an equally diverse set of pathways for dealing with the damage. Repair of topoisomerase-induced damage has been the subject of many recent reviews (Sun et al., 2020b; Riccio et al., 2020; Zagnoli-Vieira and Caldecott, 2020). In brief, the logic of repair pathways includes detection of a topoisomerase that is trapped (as opposed to an enzyme that is proceeding through the catalytic cycle), removal of the bulk covalently bound protein by proteolysis, removal of the remaining portion of the protein by nucleolytic enzymes such as Tdp1 or non-specific nucleases such as Mre11, and repair of strand breaks by break repair pathways. This logic is not always followed; nucleolytic proteins may act without prior proteolysis (Schellenberg et al., 2017). These mechanisms provide many opportunities for SUMO participation and are discussed in SUMOylation in Repair of Topoisomerase-Mediated DNA Damage.

Small ubiquitin-like modifiers (SUMOs) are a class of ubiquitin-like proteins (UBLs) that modify cellular proteins via a three-tiered enzymatic cascade akin to ubiquitin modification (ubiquitylation). The SUMO proteins include a compact globular β-grasp fold and a carboxyl-terminal glycine residue whose pattern of conjugation to substrates also resembles ubiquitylation (Flotho and Melchior, 2013). SUMOylation modulates a broad spectrum of cellular activities ranging from subcellular transport, transcriptional regulation, cell cycle progression, apoptosis, and proteasomal degradation (Geiss-Friedlander and Melchior, 2007; Flotho and Melchior, 2013; Celen and Sahin, 2020).

The mammalian SUMO family includes four members: SUMO-1, SUMO-2, SUMO-3, and SUMO-4, among which SUMO-2 and -3 are nearly identical and share 97% protein sequence identity but only share 42–43% identity with SUMO-1. In general, most SUMOylatable proteins contain a consensus motif Ψ-K-x-D/E in which Ψ is a large hydrophobic residue, K (lysine) is used for SUMO conjugation, and x indicates any amino acid residue, followed by either a (D) aspartic acid or a (E) glutamic acid residue. SUMO-2 and -3 carry internal lysine residues that conform to the consensus motif and hence are capable of polymerizing via isopeptidyl linkage (Flotho and Melchior, 2013; Pichler et al., 2017). In contrast, SUMO-1 lacks the internal consensus motif and is therefore proposed to cap the SUMO-2/3 chain and terminate its elongation (Matic et al., 2008). SUMO-1, SUMO-2, and SUMO-3 are constitutively expressed in all tissues, whereas SUMO-4 expression is restricted to certain tissues (Baczyk et al., 2017) and its functions remain largely unknown.

SUMO acts by modifying target proteins via 1) activation by E1, 2) conjugation by E2, and 3) ligation by E3. SUMO-activating enzyme subunits 1 and 2 (SAE1 and 2) are the two E1 enzymes, and Ubc9 defines the only E2 enzyme. Unlike ubiquitin, whose attachment absolutely requires an E3 ligase, SUMO moieties can be conjugated onto a target protein by the SUMO E2 conjugating enzyme alone as long as the substrate bears a SUMO consensus motif. Although SUMO E3 enzymes (ligases) are dispensable for SUMOylation in many cases, they can recognize specific targets and enhance the efficiency of SUMOylation (Figure 2). For example, SUMO ligases can transfer SUMO proteins to non-consensus lysine residues that are not recognized by Ubc9 (Flotho and Melchior, 2013; Pichler et al., 2017).

The most-studied SUMO E3 ligases are a family of SP-RING–containing proteins termed the Siz and protein inhibitor of activated STAT (PIAS) proteins (Rytinki et al., 2009). The Siz/PIAS family comprises four members including Siz1, Siz2, methyl-methanesulfonate–sensitive protein 21 (Mms21), and molecular zipper protein 3 (Zip3) in the yeast S. cerevisiae and five members including PIAS1, PIAS3, PIASxα (PIAS2α), PIASxβ (PIAS2β), and PIASy (PIAS4) in humans. Akin to the RING-type ubiquitin ligases, the Siz/PIAS SUMO ligases bridge substrates and the E2∼SUMO thioester and facilitate SUMO transfer via their SP-RING domains (Hochstrasser, 2001). The nuclear pore complex protein Ran-binding protein 2 (RanBP2) is another type of SUMO E3 ligase. Unlike the Siz/PIAS ligases, RanBP2 bears catalytic domains that interact with E2 and stimulate E2’s SUMOylating activity toward substrates rather than connecting substrates and E2 (Pichler et al., 2004).

SUMOylation can also be deconjugated by a family of cysteine proteases, termed SUMO isopeptidases or SUMO-specific proteases. These SUMO proteases target SUMO-protein conjugates as well as poly-SUMO-2/3 chains by hydrolyzing their isopeptidyl bonds. Also, the SUMO genes encode SUMO precursors that need to expose their C-terminal di-glycine motifs for their adenylation and thioester bond formation by the E1 enzymes. Some of the SUMO proteases serve as processing factors for the C-terminal maturation of the SUMO precursors. The Ulp/SENP (ubiquitin-like protease/sentrin-specific protease) proteases constitute hitherto the largest family of SUMO proteases (Mukhopadhyay and Dasso, 2007; Shin et al., 2012; Nayak and Muller, 2014; Kunz et al., 2018).

SUMO modification is broadly understood to regulate protein localization and alter protein–protein interactions (Flotho and Melchior, 2013). It has been appreciated that SUMO modification can destabilize proteins, and this targeted degradation plays important roles in processes such as DNA repair and homologous recombination (Psakhye et al., 2019). One important class of proteins is the SUMO-targeted ubiquitin ligases (STUbLs). These proteins poly-ubiquitylate extensively SUMOylated proteins (Miteva et al., 2010) to induce their proteasomal degradation. The induction of instability by polySUMOylation is antagonized by SUMO proteases. These processes work together to fine-tune modulation of proteins in space and time.

Ubiquitin-mediated proteasomal degradation of the bulky protein component of TOP-DPCs is a pivotal step in the DPC repair (Desai et al., 1997; Desai et al., 2001; Mao et al., 2001; Desai et al., 2003; Xiao et al., 2003; Lin et al., 2008; Lin et al., 2009; Sun et al., 2020b), and SUMOylation has been established to act as a signal to the ubiquitin–proteasome pathway (UPP) under certain circumstances including DNA damage responses (Uzunova et al., 2007; Galanty et al., 2009; Miteva et al., 2010; Galanty et al., 2012; Yin et al., 2012; Vyas et al., 2013; Chang et al., 2021). The initial experiments in mammalian cells could demonstrate that SUMOylation of TOP-DPCs occurred but were unable to directly demonstrate that the observed SUMOylation was a repair reaction, i.e., SUMOylation led to proteasomal degradation of TOP-DPCs, and that the degradation enhanced cell survival. Early genetic experiments in S. cerevisiae demonstrated that reduced UBC9 activity conferred CPT sensitivity in yeast (Jacquiau et al., 2005). A more direct connection between SUMOylation and TOP-DPC repair was obtained in the yeast S. pombe. Two SUMO ligases, the SMC5/6 SUMO ligase Nse2 (Mms21 in S. cerevisiae) and the PIAS/Siz SUMO ligase Pli1 (Siz1 in S. cerevisiae), were found to SUMOylate TOP1-DPCs and recruit the SUMO-targeted ubiquitin ligase (STUbL) (Miteva et al., 2010; Chang et al., 2021) Slx8 to remove the DPCs (Heideker et al., 2011; Steinacher et al., 2013). Mutation in SUMO ligase/STUbL is lethal in combination with other DNA repair defects. Notably, the lethality is suppressed when Top1 is deleted (Chen et al., 2007; Heideker et al., 2011). Note that TOP1 is not essential for viability in either S. cerevisiae or S. pombe. These experiments can be simply interpreted as follows: SUMO ligases and STUbL are essential for repairing DNA damage arising from TOP1-DPCs that result from trapping of TOP1 by lesions in the DNA. These experiments highlight the importance of repair pathways for TOP-DPCs under normal growth conditions.

A recent set of experiments demonstrates more directly SUMO modification of topoisomerases leading to proteasomal degradation of both TOP1- and TOP2-DPCs (Sun et al., 2020a). TOP1- and TOP2-DPCs undergo rapid and sequential SUMO-2/3 and SUMO-1 modifications by PIAS4 via its DNA-binding SAP. RNF4, the human ortholog of Slx8 in S. pombe, targets the SUMOylated TOP-DPCs binding the SUMO polymers through its SIM (SUMO-interacting motif) domains and ubiquitylates the DPCs for their subsequent degradation by the proteasome (Sun et al., 2020a). This SUMO–ubiquitin pathway also operates in an analogous fashion in S. cerevisiae with Siz1 serving as the main SUMO ligase and Slx5–Slx8 ubiquitylating the topoisomerases in a SUMO-dependent manner. Notably, it was found that this SUMO-mediated ubiquitin–proteasome pathway (UPP) was not contingent on either replication or transcription and that inhibition of the pathway did not fully abolish the removal of TOP-DPCs, suggesting the presence of additional UPPs and proteolysis mechanisms recruited by DNA transactions. It has been well established that there are repair pathways for TOP-DPCs that depend on ongoing transcription (Lin et al., 2008) or replication (D'Arpa et al., 1990). While the UPP has been proposed to play a role upon collisions with TOP-DPCs and active DNA transaction machinery such as advancing replication forks or transcription bubbles (Xiao et al., 2003; Lin et al., 2009), the roles of SUMOylation in these detection processes remain to be elucidated.

How does RNF4 recognize TOP-DPCs and thereby trigger its degradation? Given the role of SUMOylation in subnuclear transport of TOP2 during cell division, one possibility is that SUMOylation relocates topoisomerases from chromatin to other nucleoplasmic compartments once topoisomerases reseal DNA, and failure of topoisomerases to release from DNA (persistent TOPcc and trapped closed clamp conformation) can be sensed by PIAS4, which is associated with topoisomerases even in the absence of inhibitors. In consequence, PIAS4 SUMOylates the trapped TOPcc for the UPP-mediated degradation. In agreement with this hypothesis, several earlier observations showed that SUMO-1 modifications of TOP1 induced by camptothecins led to re-localization of the protein (Mo et al., 2002; Rallabhandi et al., 2002; Christensen et al., 2004). These findings suggest that TOP1-DPCs may be directed by SUMOylation to other sites where the lesions may be (Nie et al., 2012) repaired or sequestered (Figure 4).

A recent study reported that Siz2, an analog of Siz1, SUMOylates the vicinity of DPCs (chromatin-bound proteins adjacent to DPCs) to recruit Wss1, a replication-associated metalloprotease carrying the SprT domain, that binds and digests DPCs via its SIMs (Serbyn et al., 2021). Another study in S. cerevisiae identified the AAA ATPase Cdc48 (also known as p97 or valosin-containing protein (VCP) in mammals) and its substrate-recruiting adaptor Ufd1/Doa1 as a binding partner of Wss1 (Balakirev et al., 2015). Wss1 was found to form a ternary complex with Cdc48 and Ufd1 to bind and process SUMOylated TOP1-DPCs. The Cdc48 complex is a hexameric unfoldase/segregase that unfolds and segregates protein substrates in an ATP-dependent manner (van den Boom and Meyer, 2018). Ufd1, the adaptor of the Cdc48 complex, specifically recognizes ubiquitylated proteins via its ubiquitin-binding domain (Bodnar et al., 2018). Ufd1 also bears SIMs, making the Cdc48 complex a selective receptor for proteins that are both SUMOylated and ubiquitylated (Nie et al., 2012). It is therefore conjectured that, upon TOP-DPC SUMOylation and ubiquitylation by the Siz1–Slx5/8 pathway, the Cdc48 complex binds the heterologous SUMO–ubiquitin chain and unfolds TOP-DPCs to facilitate their digestion by Wss1. In mammalian cells, SUMOylation was shown to channel DPC repair pathway(s) to SPRTN, the ortholog of Wss1, during DNA replication (Ruggiano et al., 2021). Unlike Wss1, SPRTN lacks SIM but is still able to interact with both ubiquitin and SUMO moieties conjugated to DPCs. This raises the possibility that the PIAS4–RNF4 pathway also recruits SPRTN for TOP-DPC proteolysis and that SPRTN binds the ubiquitin–SUMO hybrid conjugates using its ubiquitin-binding UBZ domain (Kuo et al., 2016; Liu et al., 2021). This model is further supported by a recent study in human cells, which identified TEX264 (testes-expressed 264), a gyrase inhibitory-like protein, as a novel p97 co-factor that recognizes SUMO-1–modified TOP1-DPCs with its putative SIMs and recruits p97 and SPRTN for the repair (Fielden et al., 2020).

ACRC (acidic repeat–containing protein), also known as GCNA (germ cell nuclear antigen-1), is a SPRTN-related protease metalloprotease that repairs DPCs in metazoans in parallel with SPRTN. ACRC is highly expressed in germ cells while maintaining low levels in somatic and most cancer cells. A recent study shows that formaldehyde-induced DPCs are rapidly modified SUMO polymers, which recruit ACRC to the DPC sites via its SIMs for proteolysis (Borgermann et al., 2019). ACRC/GCNA was later found by two concurrent studies to co-localize with TOP2 on the mitotic chromosome in C. elegans and suppress TOP2-dependent damage during meiosis in Drosophila (Bhargava et al., 2020; Dokshin et al., 2020). In human germ cell tumors, ACRC/GCNA removes TOP2-DPCs to prevent replication stress and copy number variation (Bhargava et al., 2020). It is yet to be determined whether SUMOylation is required for ACRC to act against TOP2-DPCs. Emerging evidence also suggests proteases FAM111A (mammal) and Ddi1 (yeast) as parallel pathways for TOP1-DPC repair during DNA synthesis (Kojima et al., 2020; Serbyn et al., 2020). Whether SUMOylation plays a regulatory role for these proteases warrants further investigations.

Proteolysis by the proteasome or other proteases cannot entirely remove topoisomerase adducts from DNA, since the protein phosphotyrosyl linkage would not be expected to be a protease substrate. Under most circumstances, proteolysis will leave a short peptide of undetermined length bound to DNA. Therefore, an additional step, complete removal of the peptide, will be needed. All eukaryotes have specialized enzymes for complete removal of the peptides, the tyrosyl phosphodiesterases (TDPs (Zagnoli-Vieira and Caldecott, 2020)) Notably, under many conditions, TDPs require at least partial proteolysis in order to efficiently remove TOP-DPCs (Debethune et al., 2002; Interthal et al., 2005; Gao et al., 2014). Both TDP1 and 2 appear to be modulated by SUMOylation through different mechanisms.

TDP1 was initially identified as an excision nuclease specialized for 3′ phosphotyrosyl adducts including TOP1-DPCs (Yang et al., 1996; Pouliot et al., 1999; Interthal et al., 2005). Subsequent studies demonstrated activity against 5′ phosphotyrosyl adducts, consistent with a role in the repair of both TOP1 and TOP2-DPCs (Nitiss et al., 2006; Murai et al., 2012). TDP1 was found to be modified by all three SUMO isozymes (primarily SUMO-1) at its N-terminal lysine 111 within its consensus motif in mammalian cells (Hudson et al., 2012). TDP1 SUMOylation facilitates its recruitment to TOP1-DPCs and does not appear to affect its cleavage activity (Hudson et al., 2012). The SUMO E3 ligases specific for TDP1 SUMOylation have not been identified, and it is possible that PIAS4 may act on TDP1 to coordinate ubiquitylation and TDP1-dependent hydrolysis for TOP1-DPC repair through SUMOylation.

Human TDP2 has been shown by proteomic studies to carry two non-canonical SUMO modification sites (Hendriks et al., 2014; Matlock et al., 2015), but their impacts on TDP2 remain unknown. A study reported that ZNF451 (ZATT), a zinc-finger–containing SUMO ligase, targets TOP2α- and β-DPCs for SUMO-2/3 modifications, which recruit TDP2 to target the TOP2–DNA junction without proteolysis (Schellenberg et al., 2017). It was shown that TDP2 binds SUMOylated TOP2-DPCs via its split-SIM motifs and releases the intact protein with formation of protein-free DSBs. However, it remains enigmatic when and how the cell employs ZATT to obviate the need for proteolysis. The SUMOylation sites in TOP2 for ZATT are yet to be discovered. Since ZATT-mediated SUMO-2/3 modification is proposed to alter the conformation of TOP2-DPCs and expose its tyrosine–DNA linkage, such modification likely takes place in the DNA binding and cleavage domain, bridging TDP2 and the exposed scissile bond.

Overall, SUMOylation plays key roles in repair of TOP-DPCs. SUMO ligase activity targeting the DPCs clearly contributes to proteolysis by proteasome and non-proteasome proteases. There are several key questions that will likely yield important insights. Given the large number of pathways by which TOP-DPCs can be processed, one of the major questions is pathway choice. It is noteworthy that we understand recognition that is independent of transcription and replication better than we understand pathways that are initiated upon inhibition of replication or transcription. The regulation of modification of topoisomerase-specific repair proteins such as TDPs as well as general repair nucleases such as the MRE11 complex and proteases such as SPRTN will lead to new insights for these questions.