TDP1 (Tyrosyl-DNA Phosphodiesterase I)
The TDP1 enzyme activity was long anticipated within the DNA topoisomerase field, as these phosphotyrosyl linkages represent the chemical bond that covalently attaches DNA topoisomerases (TOPs) to the end of a cleaved DNA strand (DNA topoisomerases are reviewed by (Pommier et al. 2016)). In eukaryotic cells, the 5′phosphotyrosyl linkages are transiently formed during the catalytic cycle of type IA and type II topoisomerases, including TOP3 and TOP2 family members, respectively. Eukaryotes also contain a type IB enzyme (TOP1) that transiently forms a 3′phosphotyrosyl linkage (Pommier et al. 2016). However, these topoisomerase-DNA covalent complexes are extensively exploited for the treatment of cancer by inducing cytotoxicity via drug-mediated stabilization of enzyme-DNA covalent complexes. Chemotherapeutics such as topotecan and irinotecan irreversibly stabilize the TOP1-DNA covalent complexes, while agents such as etoposide and doxorubicin reversibly stabilize the TOP2-DNA covalent complexes (DNA topoisomerase targeting chemotherapeutic are reviewed in (Pommier 2013)). These enzyme-DNA reaction intermediates can also be trapped onto the DNA by single-strand nicks, abasic sites caused by reactive oxygen species, and other chemotherapeutics such as platinum agents (van Waardenburg et al. 2004; Pommier 2013). Although TDP1 is able to process 5′phosphotyrosyl linkages, mammalian cells contain a related enzyme, TDP2, that revealed a higher affinity for 5′phosphotyrosyl but is structurally and mechanistically unrelated to TDP1 (Pommier et al. 2014).
Tdp1 Catalytic Mechanism and Structural Analysis
Since TDP1 enzymes miss the aspartic acid residue from the PLD-conserved catalytic motif His-Lys-Asp-Asn (HxKx4Dx6N, x being any amino acid), it was placed within a newly created subclass for enzymes that contain two HxKx(n)N catalytic motifs (Interthal et al. 2001). Mutagenesis studies revealed that the lysine residue in the N-terminal located motif plays a role in positioning the non-bridging oxygen atoms of the phosphodiester, while the catalytic histidine of this first motif functions as a nucleophile that attacks the phosphodiester adduct linkage forming a phosphohistidyl bond (Interthal et al. 2001; Raymond et al. 2004; Comeaux and van Waardenburg 2014). Moreover, the lysine and asparagine residues of the second (C-terminally located) motif play a role in the electron relay that protonates the catalytic histidine within this second catalytic motif, determining its role as general acid/base (Interthal et al. 2001; Raymond et al. 2004). Additional mutagenesis studies with the yeast enzyme indicated a highly conserved two-step catalytic mechanism for TDP1 enzymes (Fig. 1b). In short, for the hydrolysis of a 3′phosphotyrosyl linkage, step 1 is the association or formation step, in which the N-terminal catalytic histidine (in human TDP1 His263 and in yeast TDP1 His182) acts as a nucleophile (Hisnuc) that attacks the 3′phosphotyrosyl linkage. This attack dissociates the tyrosyl (and by extension TOP1) from the 3′end of DNA and forms a TDP1-DNA covalent reaction intermediate. In the dissociation or resolution step (step 2), the C-terminal catalytic histidine (in human TDP1 His493 and in yeast TDP1 His432) acts as a general acid/base (Hisgab). This Hisgab activates a water molecule that facilitates hydrolysis of the 3′phosphohistidyl bond that resolves the TDP1-DNA covalent reaction intermediate. However, when TDP1 dissociates the DNA, it leaves behind a single-strand break with 3′phosphate and 5′hydroxyl groups. Subsequent action of polynucleotide kinase 3′phosphatase in mammalian cells and the 3′phosphatase Tpp1 and 5′kinase Trl1 in budding yeast will remove the 3′phosphate and add a phosphoryl group to the 5′end. This “swap” of chemical groups allows the DNA strand to be re-ligated by DNA ligase (Yang et al. 1996; Interthal et al. 2001; Raymond et al. 2004). An additional function of the Hisgab residue is to donate a proton to a strong nucleophilic-leaving group, such as the phenoxy anion of the leaving active site tyrosine of TOP1 (Raymond et al. 2004). This is probably to prevent reformation of the original substrate, a 3′phosphotyrosyl linkage. Thus, Tdp1 replaces the adduct by forming a second and resolvable adduct with the DNA (Fig. 1a).
The full-length TDP1 crystal structure is still unknown, as cellular proteolysis rapidly cleaves the N-terminal domain from the catalytic core domain, obstructing purification of the full-length enzyme. The N-terminal amino acid region of TDP1 is poorly conserved in size and sequence. Although human and yeast TDP1 shows similar biochemical activity with or without the N-terminal domain, this does not exclude a function of this domain in the cell. Indeed, various groups reported that the N-terminal region of human TDP1 is posttranslationally modified by SUMO (small ubiquitin-like modifier) and phosphorylation. In addition, the N-terminal amino acids facilitate protein-protein interactions with, among others, XRCC1 (scaffold protein involved in base excision repair pathway) and PARP1 [Poly(ADP-ribose) polymerase 1] (Comeaux and van Waardenburg 2014). For now, TDP1 N-terminal region remains structurally and functionally understudied in contrast to the TDP1 catalytic core domain that is thoroughly investigated.
The role of the Hisnuc within the first HxKx(n)N motif was confirmed by the TDP1-DNA-TOP1 peptide structure, which clearly revealed TDP1 interaction with a DNA-peptide adduct (Davies et al. 2003). An asymmetrically distributed positive charge within a cleft provides electrostatic interactions with the negatively charged phosphate backbone of the DNA (Fig. 2b yellow ellipse), which runs into the catalytic pocket. Adjacent of the cleft, the TDP1 surface provides ample space for the peptide or protein to dock (Fig. 2b, black cone). This is also the side where yeast and human TDP1 shows most of the topological and electrostatic differences (Fig. 2) (Davies et al. 2003; He et al. 2007; Comeaux and van Waardenburg 2014), suggesting that yeast and human TDP1 might interact with different bulky protein-DNA adducts. Curiously, the single-strand DNA-binding cleft is too narrow to fit a double-stranded DNA molecule; a substrate TDP1 is able to hydrolyze. It would be very interesting to discover how TDP1 processes and interacts with a double-stranded DNA molecule. Does TDP1 bind the DNA molecule as double stranded, or is TDP1 able to partly denature the double-stranded end of the DNA?
Structural data of TDP1 catalytic mutants is only available for the yeast enzymes, including the yeast analog of the human SCAN1-associated substitution Hisgab to arginine (in human H493R and in yeast H432R), the alternative Hisgab to asparagine (H493N in human, H432N in yeast) substitution, and the Hisnuc to alanine substitution (H182A) (He et al. 2007; Gajewski et al. 2012; Comeaux and van Waardenburg 2014). These catalytic mutants in yeast and human TDP1 induce different degrees of toxicity (Comeaux and van Waardenburg 2014). Comparing the catalytic core domain structures of the TDP1 mutants H432R, H432N, and H182A with wild type showed that the tertiary protein fold is practically unchanged with an α-carbon RMSD of 0.60 Å, 0.49 Å, and 0.72 Å, respectively (Gajewski et al. 2012). Conversely, the catalytic pocket revealed substitution-dependent topological changes, which were most significant for the SCAN1-associated HisgabArg substitution. The arginine side chain causes a shallower and more positively charged pocket (Gajewski et al. 2012). This change in pocket topology and electrostatic charge, together with the weaker acid/base characteristic of arginine, contribute to the longer half-life of the Tdp1HisgabArg-DNA covalent intermediate. Another interesting observation is that the toxicity of these TDP1 mutant enzymes is dependent on the presence of its N-terminal domain. It is unknown if this domain, which is not necessary for in vitro catalytic activity of the wild-type catalytic pocket, is needed (A) to process cellular DNA adducts, (B) for protein-protein interactions, (C) for cellular localization, or all of the above.
Tdp1 substrates formed by therapeutic and endogenous agentsa
3′phospho-glycolate, 3′abasic site
Tdp1 mutants (SCAN1-TDP1H493R)
Chain termination/nucleoside analogsb
Methyl methanesulfonate, temozolomide
Methylated (alkylated) bases
Reactive oxygen species
3′ and 5′abasic site, 3′ and 5′apurinic/apyrimidinic site, 3′phospho-glycolate
In addition, TDP1 has been shown to possess limited 3′exonuclease activity. TDP1 is able to excise a single RNA or DNA nucleotide from the 3′end of the DNA, but only nucleotides that contain a 3′hydroxyl group. This restriction prevents TDP1 from functioning as a general 3′exonuclease and indicates that the product of TDP1 activity, a 3′phosphoryl end, impedes TDP1-mediated degradation of the DNA strand. Moreover, the ability to remove an RNA nucleotide from the end of a DNA polymer suggests TDP1 may play a role in DNA replication. During replication, an estimate of 10,000 to 100,000 RNA nucleotides is incorporated into the DNA per round of replication of eukaryotic genomes (Comeaux and van Waardenburg 2014). Recent reports show that TDP1 might act as a very specific endonuclease while processing 5′ or 3′apurinic/apyrimidinic sites, suggesting TDP1 can cleave non-nucleotide DNA insertions that result in cleavage of the DNA phosphate backbone (Lebedeva et al. 2013).
Tdp1 and Human Physiology/Disease
Tdp1 is expressed ubiquitously within eukaryotes and, in the cell, can be found in the cytoplasm, the nucleus, and mitochondria (Das et al. 2010; Fam et al. 2013). The TDP1 amino acid sequence contains a nuclear localization signal (human TDP1 contains also an alternative nuclear localization signal), but no mitochondria localization signal has been identified. As such, the mitochondrial function of TDP1 is understudied. Interestingly, TDP1 is not essential for cell viability. Three independent groups reported that Tdp1 knockout mice are viable. However, these mice are smaller than their wild-type litter mates and show to be more sensitive to camptothecin treatment specifically proliferating intestinal cells and hematopoietic cells (Comeaux and van Waardenburg 2014).
Human TDP1 is expressed in neurons and most peripheral tissues displaying a punctate cytoplasmic and nucleus distribution, independent of sex and age (Fam et al. 2013). The high cytoplasmic level of TDP1 suggests a storage function for a ready response to the nucleus or mitochondria. Indeed, TDP1 is re-localized upon oxidative stress (Fam et al. 2013). How re-localization is achieved is not understood, but it is suggested that posttranslational modification of TDP1 with SUMO might signal re-localization to the nucleus. Moreover, it was shown that phosphorylation of human TDP1 at Ser81 stimulates TDP1-XRCC1 interaction (Comeaux and van Waardenburg 2014; Pommier et al. 2014). The lack of a mitochondrial localization signal impedes the study of TDP1 function in the mitochondria and nucleus and the potential interplay/communication between these two cellular compartments, with the cytoplasm as storage medium. This is assuming that de novo synthesis of TDP1 is not necessary for re-localization, which is unknown at the moment.
A catalytic mutant of human TDP1 containing an arginine substitution of the Hisgab residue (c.1478A>G missense mutation) is associated with the rare autosomal recessive neurodegenerative spinocerebellar ataxia with axonal neuropathy or SCAN1 (Takashima et al. 2002). Nine individuals with normal intelligence are affected within one extended Saudi Arabian family. SCAN1 has a late childhood (13–15 years) onset of progressive cerebellar ataxia and peripheral neuropathy. Patients are losing touch sensations, pain sensation, and vibration sense of their hands and legs (Fam et al. 1993; Takashima et al. 2002). Nerve conduction exams revealed decrease in amplitudes, a characteristic of axonal neuropathy, suggesting that the terminally differentiated postmitotic neurons from the cerebellum, dentate nuclei, and anterior spinal and dorsal root ganglia are affected (Fam et al. 1993). Moreover, the ataxia will expand from an ataxic gait to become wheelchair bound in early adulthood. Magnetic resonance imaging (MRI) of these patients revealed progressive cerebellar atrophy, of which the vermis region is most affected (Takashima et al. 2002). Non-neurological symptoms associated with SCAN1 are mild hypercholesterolemia and borderline hypoalbuminemia. Currently, there is no mouse model of this disease (knock-in of the mutated TDP1H493R allele). The Tdp1 knockout mouse does not exhibit a SCAN1 phenotype but develops age-dependent mild progressive cerebellar atrophy with hypoalbuminemia.
Biochemical characterization of the human TDP1H493R mutant showed that this mutant was not inactive as originally was proposed. Instead, this mutant enzyme exhibits a slower dissociation rate from its TDP1-DNA adduct. In other words, the TDP1H493R mutant enzyme is trapped longer onto the DNA and as such forms a potentially lethal DNA adduct. Moreover, the recessive phenotype of SCAN1 and the formation of a more stable enzyme-DNA complex revealed that TDP1 can act in trans to resolve the TDP1H493R-DNA adduct, which was biochemically confirmed (Comeaux and van Waardenburg 2014). The delayed dissociation is an accumulation of a weaker acid/base characteristic of the arginine side chain, the shallower and more positively charged catalytic pocket (Gajewski et al. 2012; Comeaux and van Waardenburg 2014). These topological changes will affect the penetration and/or the position of the water molecule that needs to be activated by the weaker general acid/base arginine to resolve the TDP1-DNA chemical linkage.
Intriguingly, the SCAN1 patients do not show to be more prone to the development of cancer or immunodeficiencies, as observed for other DNA repair-related mutations. Yet the question remains, why is only the cerebellum affected in these patients? This might be explained by the observation that during mice embryogenesis, the brain/neuronal region accumulates TOP1-DNA covalent complexes (Katyal et al. 2014). Endogenously generated DNA lesions such as abasic sites and single-strand nicked DNA can trap these TOP1-DNA complexes. Consequently, TDP1 and a noncanonical function of ATM were found to be essential for neuronal cell viability. Knockout mice missing either of these proteins are sick but survive, yet the double knockout mice, lacking both TDP1 and ATM proteins, show a synthetic lethal phenotype and die (Katyal et al. 2014). Although TDP1 is not essential for cell viability, in higher eukaryotes, TDP1 is a critical factor for normal neurological development. Thus, these results give us, for the first time, some insight in the etiology of SCAN1; accumulation of TOP1-DNA and TDP1H493R-DNA adducts affects the development and viability of the cerebellum more than other neuronal regions, resulting in progressive cerebellar atrophy and the onset of SCAN1. So, are the DNA repair pathways in the cerebellar differentially regulated compared to other neuronal tissues and other cells? Or does the localization of TDP1 to the mitochondria play a role in the onset of SCAN1, since mitochondria contain their own isoform of TOP1 (TOP1mt) (Pommier et al. 2016)?
From its discovery, TDP1 was associated with drug resistance, specifically those that target TOP1-DNA intermediates (Yang et al. 1996). This association grew stronger over time with the identification of new TDP1 substrates (Table 1) that are generated by chemotherapeutics (Comeaux and van Waardenburg 2014; Pommier et al. 2014). Additional reports show that TDP1 protein levels are elevated in tumor tissue and cancer-derived cell lines. Immunohistochemistry revealed that TDP1 protein levels are elevated in samples of 16 different tumor tissues. Subsequently, it was shown that RNAi-mediated TDP1 knockdown sensitizes colorectal cancer and rhabdomyosarcoma cell lines to irinotecan (a TOP1-DNA targeting prodrug). Moreover, in a subset of rhabdomyosarcoma cell lines, TDP1 knockdown was lethal. In other words, viability of these cells depends on TDP1 expression, while healthy cells do not. What is more, a yeast screen revealed that overexpression of TDP1 induced chromosome instability (CIN), which was verified in rhabdomyosarcoma cells (Duffy et al. 2016). Moreover, elevated expression of TDP1 in both yeast and human cell models increases cell sensitivity to a broad spectrum of DNA-damaging therapeutics (He et al. 2007; Comeaux and van Waardenburg 2014; Duffy et al. 2016).
Tdp1 as Therapeutic Target
We propose a second therapeutic approach that focuses on stabilizing the TDP1-DNA covalent complex, which is the center of the TDP1 catalytic cycle (Fig. 3). This approach is similar to how TOP1-targeting chemotherapeutics (topotecan and irinotecan) function by stabilizing the TOP1-DNA covalent complex (Pommier et al. 2016). The SCAN1 TDP1H493R mutant, whose toxicity is caused by the extended lifetime of its enzyme-DNA adduct, supports this approach. Moreover, additional substitutions of either catalytic histidine residue within yeast and human TDP1 result in a more toxic phenotype, related to the formation of the TDP1-DNA adduct (He et al. 2007; Gajewski et al. 2012; Comeaux and van Waardenburg 2014). Intriguingly, combining topotecan with the expression of these toxic TDP1 mutants in a human cell model shows a more than additive cytotoxicity. In other words, it increases the therapeutic range of topotecan, which would allow for a treatment dose reduction and as such reduce side effects. High-throughput screens and structure activity relationships are underway for this approach. These potential novel chemotherapeutics might be able to be used as a single treatment option or better combined with FDA-approved anticancer agents that induce TDP1 substrates or other synthetic lethal treatment options.
In the last 20 years, we gained knowledge of TDP1 biochemistry and cellular function, but it is unclear how TDP1 functions in the cell. It is still undetermined how TDP1 is recruited to its protein-DNA adducts and its mechanism of interacting and opening the TOP1-DNA adduct. Furthermore, the function of the N-terminal domain is understudied, although reported to be posttranslationally modified and to mediate protein-protein interactions. We do not know its function in cellular catalysis and cellular distribution. Besides that TDP1 can act in trans to process a TDP1-DNA adduct, we have no understanding of which other repair processes are able to resolve the TDP1-DNA lesion. These are some of the essential questions to be answered, as TDP1 is of therapeutic interest, and we need to gain knowledge of its cellular functions to anticipate off-target effects.
- Fam HK, Salih MAM, Takashima H, Boerkoel CF. Spinocerebellar ataxia with axonal neuropathy, autosomal recessive. In: Pagon RA, Adam MP, Bird TD, Dolan CR, Fong CT, Stephens K, editors. GeneReviews. Seattle: University of Washington; 1993.Google Scholar
- He X, van Waardenburg RC, Babaoglu K, Price AC, Nitiss KC, Nitiss JL, et al. Mutation of a conserved active site residue converts tyrosyl-DNA phosphodiesterase I into a DNA topoisomerase I-dependent poison. J Mol Biol. 2007;372:1070–81. doi:10.1016/j.jmb.2007.07.055.PubMedCrossRefPubMedCentralGoogle Scholar