Phylogenetic analysis was performed with the Maximum Likelihood method in SeaView software (Gouy et al., 2010) using the PhyML program with the following parameters: LG model, 8 rates of categories, tree searching operation best of NNI&SPR (Nearest Neighbor Interchange and Subtree Pruning and Regrafting) (Guindon and Gascuel, 2003). Tree node confidence is expressed as Alrt values (Approximate likelihood-ratio test) on a scale of 0–1, where 1 represents the maximum node confidence (Anisimova and Gascuel, 2006). Protein sequences were retrieved from the National Center for Biotechnology Information (NCBI) database (Benson et al., 2013) using the blastp algorithm (Altschul et al., 1990) with human TDP2 as the query sequence, followed by alignment of the full-length protein sequences using the Multiple Alignment using Fast Fourier Transform (MAFFT) algorithm (Katoh et al., 2002). Alignment quality was assessed using the Guidance2 server and the score was 0.767980 which corresponds to high alignment quality. For comparison, a score above 0.5 is considered sufficient to use the alignment for tree building using the maximum likelihood method (Penn et al., 2010). Syntenic analysis of the TDP2 gene was performed using Genomics, a browser for conserved synteny synchronized with genomes from the Ensembl database (Louis et al., 2013). Protein domain structures of human TDP2 and zebrafish Tdp2a and Tdp2b were visualized using IBS software (W. Liu et al., 2015).

The zebrafish (Danio rerio) AB strain was obtained from the European Zebrafish Resource Centre (EZRC, Karlsruhe, Germany) and was maintained at a temperature of 28°C under a 14-h light and 10-h dark cycle as previously described (Aleström et al., 2020). Tdp1 ^−/− mutants were created in our lab and characterized in Anticevic et al., 2023. Embryos were cultured in E3 media (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, and 0.33 mM MgSO₄) at 28°C until 2 days post-fertilization (dpf). Prior to experiments, embryos were manually dechorionated and, if required, treated with 10 mM formaldehyde for 30 min (FA, KEMIKA: 0633501) or with 50 μM etoposide (ETO, Thermo Scientific Chemicals: J63651) for 1 h at 28°C. All procedures followed ethical guidelines (EU Directive 86/609/EEC, Croatian Federal Act on Animal Protection) under project license HR-POK-023.

RNA isolation from adult zebrafish tissue samples weighing up to 50 mg was performed using the Monarch Total RNA Miniprep Kit (NEB, T2040L). The tissue samples were homogenized using an Ultra Turrax T25 homogenizer at medium intensity for 60 s (13,500 rpm) followed by 5 min incubation with proteinase K (20 mg/mL) at 55°C. After centrifugation at 13,000 g for 2 min at room temperature, the supernatant was separated, and RNA purification was carried out following the manufacturer’s instructions. Five embryos per condition at different developmental stages were collected, including 6 h post-fertilization (6 hpf) and 1, 2, 3, 4, and 5 days post-fertilization (dpf), respectively. Two-day-old morphant embryos were collected after tdp2a or tdp2b gene silencing to determine the tdp2 expression levels after the respective gene silencing. Pools of five embryos were collected at 2 dpf to determine the expression of the injected mRNAs: tdp2b, tdp2b ^D285A , tdp2a, and HsTDP2. For the extraction of RNA, samples were sonicated 3 × 5 s on ice, followed by proteinase K treatment according to manufacturer’s instructions using the Monarch Total RNA Miniprep Kit (NEB, T2040L). The isolated RNA was subsequently aliquoted and stored at −80°C. For reverse transcription, the ProtoScript II First Strand cDNA Synthesis Kit (NEB, E6560L) was used, following the manufacturer’s instructions. Total RNA from zebrafish tissues and embryos was added in a volume corresponding to 100–1000 ng of RNA, resulting in a concentration of 5–50 ng/μL of cDNA for subsequent expression analysis.

qPCR analysis was conducted using GoTaq qPCR mix (PROMEGA, A6001) (Table 1). The housekeeping gene atp50 (atp5po, ATP synthase peripheral stalk subunit OSCP, Gene ID: 335191) was used as a reference gene for normalization. Quantification was performed using the Qgene method (Simon, 2003), and gene expression levels were reported as Mean Normalized Expression (MNE). MNE was calculated based on the primer efficiencies (E) and mean Ct values for both the housekeeping gene (Ct (HKG)) and the target gene (Ct (gene)), using the equation: M N E =   E H K G C t H K G / E g e n e C t g e n e × 10 6 as previously described (Popovic et al., 2010; Lončar et al., 2016; Mihaljevic et al., 2016).

The antisense morpholino oligonucleotides targeting tdp2a and tdp2b were designed and ordered from Genetools LLC (Nasevicius and Ekker, 2000). In particular, the tdp2a morpholino targets exon 3–intron 3 boundary and the tdp2b morpholino targets the 5′UTR to prevent splicing and translation, respectively (Table 2). Morpholinos were diluted in a 0.015% Phenol-red/300 mM KCl solution to obtain injection mixes containing 500 μM tdp2aMO, 300 μM tdp2bMO, or 500µM + 300 μM tdp2a + tdp2bMO; 1nL injection mix was injected into zebrafish embryos between the one and the four-cell stage.

To verify the efficiency of the splice-blocking tdp2a morpholino, RNA was extracted from 2 dpf embryos, reverse-transcribed to cDNA as described above, and PCRs were performed on those cDNA samples to determine the effects of the morpholino on transcript splicing as previously described (Anticevic et al., 2023). The expected size of the amplicon on WT samples is 524 bp, using the primer pair shown in Table 3. To verify the efficiency of tdp2b morpholino, a functional assay was performed to quantify the enzymatic activity of Tdp2b to confirm gene silencing (Moulton, 2007).

To verify the specificity of the tdp2b morpholino, we performed rescue experiments in which we co-injected tdp2bMO and mRNA encoding full-length tdp2b coding sequence into one-cell stage embryos. Since the tdp2b morpholino targets the endogenous 5′UTR of tdp2b, it cannot bind to the mRNA rescue construct which has a different upstream sequence derived from the plasmid from which it was in vitro transcribed. DrTdp2b (NM001079703.1) coding sequence was amplified using Infusion primers (shown in Table 4) on cDNA derived from 6 hpf WT embryos. The PCR product was then cloned into the pCS2+HisMyc vector between the XhoI and XbaI restriction sites using the Infusion kit (Takara Bio USA, Inc.) (Rohr et al., 2006). The resulting plasmid was mutated using the primer shown in Table 4 to obtain a catalytically inactive Tdp2b^D285A. Zebrafish DrTdp2a (ENSDART00000102212.5) coding sequence was amplified using infusion primers (Table 4) from cDNA derived from the intestine of adult male zebrafish and cloned into the pCS2+HisMyc vector. Human TDP2 (NM_016614.3) was obtained from Genscript and cloned into the pCS2+HisMyc using infusion primers (Table 4). All plasmids were linearized using the NotI restriction enzyme and in vitro transcribed using the HiScribe SP6 RNA kit (NEB, #E2070) in conjunction with the ARCA kit (NEB, #S1411) to cap the resulting RNAs which were then purified using the Monarch RNA cleanup kit (NEB, #T2040) for subsequent injections. For injection experiments, 1 nL of a solution of mRNA alone (250 ng/ul) or mRNA (250 ng/ul) with tdp2bMO (300 μM) in 300 μM KCl was injected between the 1- and 4-cell stage.

The Cy5-labeled substrate oligomer (100 pmol) (Table 5) was mixed with a 20-bp complementary oligonucleotide containing a 5′overhang in a volume of 33.3 μL. The sample was denatured at 95°C for 5 min and then reannealed by gradually reducing the temperature at a rate of 2°C/s for 5 s, followed by 0.1°C/s for 600 s using the gradient PCR (T100 Thermal Cycler, Biorad). This process generated a 3 μM double-stranded substrate oligomer with a 5′overhang, which is a model substrate for TDP2 (Ledesma et al., 2009).

The Tdp2 activity assay was performed as previously described (Zagnoli-Vieira et al., 2018; Zaksauskaite et al., 2021) with few modifications. 2 dpf zebrafish embryos were deyolked in deyolking buffer (55 mM NaCl, 1.8 mM KCl and 1.25 mM NaHCO₃, pH 8.5) and washed twice with deyolking wash buffer (110 mM NaCl, 3.5 mM KCl, 2.7 mM CaCl2 and 10 mM Tris-HCl, pH 8.5). The deyolked embryos were transferred in a solution containing 40 mM Tris/HCl pH 7.5, 100 mM NaCl, 0.1% Tween-20, 1 mM DTT, 1 mM PMSF, and protease inhibitors (leupeptin, aprotinin, chymostatin, pepstatin at a concentration of 1 μg/mL) and sonicated for 30 s using a probe sonicator with 3 μm peak-to-peak amplitude. Following sonication, the lysate was incubated for 30 min on ice and then centrifuged at 10,000 g for 5 min at 4°C. The supernatant, which contained proteins, was collected, and the protein concentration was determined using the Bradford assay (Bradford, 1976). Subsequently, 10 μg of the protein solution was mixed with 1 × Tdp2 activity assay buffer (50 mM Tris/HCl pH 8.0, 10 mM MgCl₂, 80 mM KCl, 1 mM DTT, 0.01% Tween-20), the Cy5-labeled substrate oligomer (40 nM) and a competitor oligo (3 μM) (Table 5). The reaction was incubated for 1.5 h at 37°C, and stopped by the addition of 2x formamide loading buffer (80% (w/v) deionized formamide, 1 mg/mL xylene cyanole, 1 mg/mL bromophenol blue, and 10 mM EDTA (pH 8.0)) and boiled for 5 min at 95°C. The samples were then applied to a 20% polyacrylamide gel containing 8M urea, which had been pre-run for 1 h at 80 V. The gel was run for 2 h at 100 V to achieve optimal sample separation. The gel was visualized using the ChemiDoc MP imaging system (Bio-Rad, 1708280).

The TDP1 activity assay optimized for detecting Tdp1 activity in zebrafish embryos was performed as described previously (Anticevic et al., 2023). In brief, 600 ng of embryo lysates were incubated with a Tdp1 oligonucleotide substrate (Table 5), containing a tyrosine at the 3′ end of the DNA and Cy5 at the 5′end. Active Tdp1 removes the tyrosine from the 3′ end, causing a shift in the size of the substrate. 2 dpf tdp1 ^−/− mutants and WT embryos, with or without transient overexpression of tdp2a and tdp2b mRNA, were deyolked and homogenized for 10 s in 100 μL of lysis buffer (200 mM Hepes, 40 mM NaCl, 2 mM MgCl₂, 0.5% Triton X-100 with protease inhibitors), followed by incubation on ice for 30 min. Next, the supernatant protein solution (600 ng) was incubated with 2.5 µM labeled oligonucleotide substrate in assay activity buffer (25 mM Hepes (pH 8.0), 130 mM KCl, and 1 mM dithiothreitol (DTT)) in a final reaction mixture of 10 μL. The reaction proceeded at 37°C for 1 h, after which loading buffer was added, and the mixture was boiled at 95°C for 5 min. All samples were loaded onto a pre-run 20% homemade urea gel and run at a constant voltage (120 V) for 2 h. The resulting oligonucleotide products were visualized using the ChemiDoc MP Imaging System to detect Cy5 fluorescence.

RADAR (rapid approach to DNA adduct recovery) assay is a well-known method for DPC isolation from cell models (Kiianitsa and Maizels, 2013; 2020). In this study, similar to our previous work (Anticevic et al., 2023), we employed a modified assay optimized for DPC isolation from zebrafish embryos, enabling better reproducibility and increased sensitivity. 2 dpf embryos were used for total DPC detection as previously described (Anticevic et al., 2023) and 1 dpf embryos were used for detection of specific DPCs (Top2, Ku80 and H3). In brief, embryos were collected and lysed using pre-warmed lysis buffer (6 M guanidinium thiocyanate (GTC), 10 mM Tris-HCl (pH 6.0), 20 mM EDTA, 4% Triton X100, 1% N-lauroylsarcosine sodium, and 1% β-mercaptoethanol), followed by incubation at 50°C for 5 min. The DNA with crosslinked proteins was precipitated by adding an equal volume of 98% ethanol, followed by centrifugation at 10,000 rcf for 10 min at 4°C. The resulting pellet was washed four times with wash buffer (20 mM Tris-HCl (pH 7.4), 1 mM EDTA, 50 mM NaCl, 50% EtOH) and dissolved in 8 mM NaOH (1 mL). To quantify the DNA content in the DPC samples, a 25 μL aliquot of each sample was treated with proteinase K (20 mg/mL) and quantified using the Pico Green assay according to the manufacturer’s instructions (Invitrogen, P7581). The DPC samples were normalized to the sample with the lowest DNA content and treated with DNAse (Millipore, E1014) for 1 h at 37°C. Subsequently, the DPC samples were snap-frozen in liquid nitrogen and subjected to overnight lyophilization using a FreeZone 2.5 lyophilizer (Labconco, United States). The lyophilized samples were dissolved in 50 μL SDS loading buffer containing 4 M urea, 62.5 mM Tris-HCl (pH 6.8), 1 mM EDTA, and 2% SDS.

To detect total DNA-protein crosslinks (DPCs), 250 ng of DNA-normalized DPCs isolated from 2 dpf zebrafish embryos were separated via SDS-PAGE electrophoresis. The DPC samples were mixed with 5x Laemmli buffer and 5% β-mercaptoethanol before being applied to homemade gradient 5%–18% polyacrylamide gels. Following electrophoresis, the resolved DPCs were visualized using silver staining (Sigma Aldrich, PROTSIL1). To detect specific DPCs, total cellular DPCs were isolated from 1 dpf embryos and applied to dot blot analysis using the Bio-Dot^® Microfiltration System (BioRad, 1703938). For Top2-DPCs, 1000 ng of DNA-normalized DPCs were transferred to a nitrocellulose membrane (GE10600002 Amersham™ Protran^®) via vacuum aspiration. Similarly, 500 ng of DNA-normalized DPCs were used for Ku80 detection, and 250 ng for histone H3 detection. To visualize specific DPCs, the membrane was immunoblotted with anti-TOP2 (Abcam, ab52934, 1:1000), anti-Ku80 (Cell Signaling, #2753, 1:1000) or anti-histone H3 primary antibodies (Cell Signaling, #9715, 1:2000). After overnight incubation at +4°C with the appropriate primary antibody, visualization of specific signals was performed in the same manner as described in the Western blot analysis. To verify the DNA quantifications, dot blot analysis was performed using a nylon membrane (RPN303B, GE Healthcare). DNA (2 ng) was applied to the nylon membrane, and DNA was detected with α-dsDNA antibody (abcam ab27156, diluted 1:7000, incubated overnight at + 4°C) and the HRP-coupled anti-mouse secondary antibody (A9044, Sigma-Aldrich. 1: 10000) (1 h at room temperature).

Zebrafish embryo (2 dpf) lysates collected for Tdp2 activity assay were also analyzed by Western blotting to determine yH2AX levels. In all samples, SDS was added to a final concentration of 0.5%, followed by incubation on ice for 30 min. For Western blot analysis, 5 μg of total protein solutions were boiled for 5 min at 95°C with 5x Laemmli buffer (50 mM Tris-HCl, pH 6.8; 2% SDS; 10% w/v glycerol; 0.05% bromophenol blue; and 5% β-mercaptoethanol). Samples were separated on SDS-PAGE gradient gels (5%–18%) using the Mini-PROTEAN 3 Cell electrophoresis chamber (Biorad). The separated proteins were then transferred to a polyvinylidene difluoride membrane (PVDF, 03010040001, Roche) using the Mini Trans-Blot Cell transfer system (Biorad) via wet transfer at 100 V for 1h and 15 min (0.025% SDS). Blocking was performed using 5% low-fat milk (T145.1, Carl Roth) in TBST (10 mM Tris-HCl (pH 7.5), 15 mM NaCl, 0.02% Tween 20) with gentle rocking for 2 h at room temperature. The membranes were then washed and incubated overnight at 4°C with anti-yH2AX antibody (Abcam, ab81299, 1:2500) in 2.5% BSA TBST buffer or anti-Tubulin antibody (Santa Cruz, sc-134238, 1:7000) which was used as a loading control. The following day, membranes were washed three times for 5 min with TBST buffer and incubated for 1 h with a secondary antibody: goat anti-rabbit IgG-HRP (Sigma-Aldrich, a0545, 1:100,000) for yH2AX and goat anti-mouse IgG-HRP (SigmaAldrich, a9044, 1:100,000) for tubulin, while gently rocking at RT. The membranes were then washed three times for 15 min with TBST buffer and once with TBS (10 mM Tris-HCl (pH 7.5), 15 mM NaCl) buffer. Proteins were detected using ECL blotting substrate (1705061, Biorad) and visualized using the ChemiDoc™ XRS + System (Biorad). Protein size was estimated by use of protein marker (1610374, Biorad).

Phenotypes were observed and recorded at 2 dpf. Embryos were dechorionated manually and images were taken using a Samsung 13-megapixel camera with an f/1.9 aperture applied to the ocular of Motic SMZ-171 binocular.

Quantification of silver-stained gels, Dot blots, Western blots, and agarose gels (PCRs) for the evaluation of morpholino-mediated silencing efficiencies was conducted using the ImageJ software (Abràmoff et al., 2004). Graphical representation of the expression data and statistical analysis were conducted using the unpaired two-tailed Student’s t-test with GraphPad Prism 8 software. Statistical significance was considered when p < 0.05, indicating differences between two independent conditions. Each experiment was repeated three times, and the results are presented in each column as mean ± standard error (SEM).

Tyrosyl-DNA phosphodiesterase 2 (TDP2) is a highly conserved protein found in all domains of life, including bacteria, fungi, algae, plants, and animals (Figure 1A and Supplementary Figure S1). Yeasts lack the TDP2 protein (Figure 1A), as previously reported (Ledesma et al., 2009). Since TDP2 is an evolutionarily ancient protein, it has likely been lost in yeast lineages during evolution. We observed occasional TDP2 duplications over the course of evolution, specifically in some cyprinid species, including zebrafish and European carp (Cyprinus carpio) and within the invertebrate group which is phylogenetically closer to vertebrates (Figure 1A, in dark green), specifically in tunicates (Styela clava), echinoderms (Anneissia japonica) and cnidarians (Dendronephthya gigantea). Due to the teleost-specific whole genome duplication (WGD) event around 320 million years ago (Jatllon et al., 2004), zebrafish often have two paralogs corresponding to a single gene in other vertebrate species (Ravi and Venkatesh, 2008), as is the case for tdp2a and tdp2b, which are paralogs of human TDP2 (Figure 1A). Phylogenetic analysis showed that Tdp2b is closer to TDP2 in mammals and other vertebrates, whereas the Tdp2a cluster in teleost fish diverged from the main vertebrate cluster (Figure 1A, in light blue).

The domain structure is highly conserved between human and zebrafish TDP2 orthologs (Figure 1B) and consists of the N-terminal non-canonical UBA (ubiquitin-associated) domain and the C-terminal catalytic exonuclease/endonuclease/phosphodiesterase (EEP) domain with four conserved catalytic motifs (Figure 1B, in blue) and residues N120, E152, D262 and H351, which form the magnesium coordination site (Schellenberg et al., 2012; Shi et al., 2012). Zebrafish Tdp2b is more similar to human TDP2 than Tdp2a, which has a longer N-terminal part (Figure 1B) that is mostly unstructured (Schellenberg et al., 2012; Shi et al., 2012), and is overall a longer protein compared to Tdp2b and human TDP2 (Figure 1B). The human TDP2 gene is located on chromosome 6, whereas in zebrafish tdp2a is located on chromosome 16, and tdp2b on chromosome 19 (Figure 2A). The syntenic analysis showed that the gene environment is conserved between human and zebrafish Tdp2. A comparison of the genes surrounding TDP2 in humans and zebrafish showed that a gene cluster consisting of ACOT13, C6orf62, and GMNN is located upstream of human TDP2, which is also found in the vicinity of the zebrafish tdp2b gene (Figure 2A). In addition, the RIPOR2 and CARMIL genes are located upstream of zebrafish tdp2b (Figure 2A). On the other hand, downstream of human TDP2 is a gene cluster containing KIAA0319 and ALDH511, which is found upstream of zebrafish tdp2a (Figure 2A). Similarly, downstream of human TDP2, we found MRS2 and NRSN1, which are located further downstream of zebrafish tdp2a (Figure 2A). Interestingly, this small chromosomal region surrounding TDP2 shows gene duplication similar to the tdp2 gene. For example, the downstream gene NRSN1 has two orthologs: nrsn1 downstream of tdp2a and nrsn1l downstream of tdp2b (Figure 2A). The same is true for the downstream gene SOX4, which has two orthologs: sox4b downstream of tdp2a and sox4a downstream of tdp2b (Figure 2A). In summary, the gene environment of Tdp2 is partly conserved in humans and zebrafish.

In the first 4 hours of vertebrate embryonic development, rapid cell divisions occur without distinct G1 and G2 phases (Siefert et al., 2015). At this stage, maternally deposited mRNAs play a crucial role in early development, including the mRNAs of DNA repair genes (Zhang et al., 2014). Some DNA repair activity is present in zygotes and early-stage embryos, but their ability to recognize and respond to DNA damage is limited (Dey et al., 2023). After 6 h post-fertilization (hpf), maternal transcripts are mostly degraded, and embryonic transcription is fully active (Mathavan et al., 2005). We investigated the expression dynamic of both zebrafish tdp2 orthologs, tdp2a and tdp2b, in different embryonic stages ranging from 6 hpf to 5 days post-fertilization (dpf) (Figure 2B). Expression levels were measured using predetermined thresholds (Lončar et al., 2016; Anticevic et al., 2023), where high expression was considered when the normalized expression (MNE) was >60 × 10⁶ (Ct values <22), moderate when MNE is 2 × 10⁶–60 × 10⁶ (Ct = 23–26), and low when MNE is <2 × 10⁶ (Ct > 27). Both tdp2 orthologs showed high expression, but tdp2b exhibited 10 to 40 times higher expression compared to tdp2a throughout zebrafish development (Figure 2B). Interestingly, both genes exhibited similar expression patterns, with the highest expression observed at the 6 hpf stage, which gradually decreased and reached a stable expression level at 2 dpf (Figure 2B). Both genes were significantly more expressed in earlier developmental stages (6 hpf and 1 dpf) compared to the later stages (2 – 5 dpf): tdp2a three times more and tdp2b eight times more (Figure 2B).

Using the same gene expression quantification method, we found that tdp2a and tdp2b are both highly expressed in adult tissue including gonads, brain, kidney and intestine, while their expression is moderate in liver (Figures 2C, D). Notably, both genes showed highest expression in gonads with pronounced gender differences. Tdp2a is very highly expressed in testes, aprox. 50-fold more than in ovaries (p < 0.1) (Figure 2C). In contrast, tdp2b is very highly expressed in ovaries: 5 times higher than in testes (p < 0.1) (Figure 2D). Another difference in expression between the two orthologs was observed in the intestinal tissue where tdp2a is more highly expressed in both genders (10 times higher than tdp2b). In brain, both orthologs exhibited very high expression, followed by high expression in kidney and moderate expression in liver (Figures 2C, D).

To investigate the function of tdp2a and tdp2b in zebrafish embryos, we designed a splice-blocking morpholino to specifically silence tdp2a gene expression and a translation-blocking morpholino to specifically inactivate tdp2b. We determined the silencing efficiency of the tdp2a splice-blocking morpholino by performing PCRs on cDNA derived from 2 dpf embryos (Supplementary Figures S2A, B). No PCR amplification was observed in the tdp2a morphant samples, indicating 100% silencing efficiency of the tdp2a morpholino (Supplementary Figure S2B). The silencing of tdp2b was confirmed by measuring the reduction in Tdp2b enzymatic activity in five independent experiments (biological replicates) (Figures 3A, B; Supplementary Figures S3A, B, E).

To measure the enzymatic activity of Tdp2a and Tdp2b in zebrafish embryos, we performed a TDP2 activity assay (Zeng et al., 2012; Zagnoli-Vieira et al., 2018). In this assay, a Cy5-labelled oligonucleotide containing a 5′phosphotyrosyl moiety (5′-PY) was incubated with whole embryo lysates (Zaksauskaite et al., 2021) of WT, tdp2a or tdp2b morphants (Figures 3A, B and Supplementary Figures S3A, B, E). To generate a specific substrate for Tdp2, the labelled oligonucleotide was annealed with a complementary primer, resulting in a double-stranded substrate oligomer with a 5′overhang containing the tyrosine residue. In the presence of active Tdp2, the tyrosine residue is removed from the 5′end of the oligomer, resulting in a cleavage product seen as an additional band (p-oligo-Cy5, red arrow) (Figure 3A). In WT embryos, in agreement with previous findings (Zaksauskaite et al., 2021), we observed successful processing of the phosphotyrosyl moiety into a phosphate group (red arrow, Figure 3A). Interestingly, we also detected bands that were lower than the band with the phosphate group (oligo-Cy5), suggesting additional cleavage events (Figure 3A). When comparing Tdp2 activity in embryos, we did not detect significant changes in the specific band intensity between WT embryos and tdp2a-silenced embryos (Figure 3A). In contrast, incubation of the substrate with embryos in which tdp2b was silenced resulted in a notable 50% reduction in Tdp2 activity (Figures 3A, B and Supplementary Figures S3A, B, E). Similarly, simultaneous silencing of tdp2a and tdp2b in zebrafish embryos resulted in a reduction of Tdp2 activity compared to silencing of tdp2b alone (Figure 3A).

To test the specificity of silencing, we co-injected mRNA encoding full-length Tdp2b with tdp2bMO and observed a complete rescue of Tdp2 activity (Figures 3A, B and Supplementary Figures S3A, B, E). Overexpression of tdp2b not only restored the loss of Tdp2 activity but also increased the substrate processing by nearly two-fold, resulting in a more intense signal of the oligonucleotide lacking phosphotyrosyl (p-oligo-Cy5) (Figures 3A, B and Supplementary Figures S3A, B, E). As expected, overexpression of catalytically inactive Tdp2b with a mutation in the active site (D285A) could not rescue the activity and showed same activity levels as tdp2b morphants (Figure 3B, Supplementary Figures S3A, B). Overexpression of tdp2b in both WT and tdp2b-silenced embryos resulted in a similar level of Tdp2 activity (Supplementary Figures S3E). Altogether, these results suggest that the Tdp2b ortholog accounts for the overall Tdp2 activity in zebrafish embryos.

Tdp2a, tdp2b and tdp2a/b morphants showed no visible phenotype at 2 dpf (Figure 3D and Supplementary Figure S4). We conclude that transient impairment of Tdp2 function does not cause abnormalities during early embryonic development.

The Tdp2a ortholog is enzymatically active, but does not contribute to the overall Tdp2 activity in 2-days old zebrafish embryos (Figure 3A). The total Tdp2 activity measured after tdp2a silencing was the same as in WT embryos, and is a consequence of active Tdp2b in these samples (Figure 3A). However, overexpression of Tdp2a significantly increased processing of Tdp2 substrate in both WT and tdp2b-silenced embryos (Figure 3B, Supplementary Figures S3A, B), showing that the Tdp2a ortholog is enzymatically active. Overexpression of human TDP2 did not result in a significant increase in the substrate processing (Figure 3B, Supplementary Figures S3A, B) which could be a consequence of differences in codon usage between human and zebrafish (Nakamura et al., 2000; Plotkin and Kudla, 2011; Bazzini et al., 2016; Wu et al., 2019). Indeed, analysis of codon usage frequencies for the human TDP2 CDS showed that out of 57 codons in the HsTDP2 CDS, six are significantly underrepresented, and 3 are somewhat underrepresented in zebrafish when compared to human codon frequency (Nakamura et al., 2000).

In addition, we investigated whether Tdp2 zebrafish orthologs can cleave a Tdp1 substrate in zebrafish embryos, given that partial redundancy of Tdp1 and Tdp2 was previously observed (Shimizu et al., 2023). To this end, we used the Tdp1-deficient zebrafish line which we previously created and characterized (Anticevic et al., 2023) and Tdp1 activity assay (Yang et al., 1996). Overexpression of Tdp2a or Tdp2b in Tdp1-deficient and in WT embryos did not result in increased Tdp1 substrate cleavage (Figure 3C, Supplementary Figures S3C, D). All injected mRNAs were stable and highly expressed, as verified by qPCR (Supplementary Figure S2C). In addition, overexpression of different constructs did not affect the morphology and development of embryos (Supplementary Figure S4).

Total DPCs were isolated from wild-type (WT) and tdp2-silenced embryos at 2 days dpf post-fertilization and conclusions were derived from four biological replicates. To compare DPC levels in tdp2-deficient embryos with DPC levels induced by a strong, model inducer, formaldehyde (FA), WT embryos were treated with 10 mM formaldehyde (30 min, 28°C) in each experiment. We previously optimized FA exposure conditions to induce DPCs without effects on embryonic phenotypes (Anticevic et al., 2023). Silencing of tdp2b led to a 2.1-fold increase in DPC accumulation compared to WT embryos, whereas silencing of tdp2a did not significantly affect cellular DPC levels (Figure 4 and Supplementary Figures S5A, B). Simultaneous silencing of tdp2a and tdp2b resulted in the same DPC increase as silencing of tdp2b alone (2.2-fold increase) (Figures 4A, B, Supplementary Figures S5A, B), while FA caused a 1.8-fold increase (Figures 4A, B, Supplementary Figures S5A, C).

In a more detailed analysis of the accumulated DPCs, we categorized the DPCs into three subgroups based on their molecular weight as previously described (Anticevic et al., 2023): High Molecular Weight (HMW >151 kDa), Medium Molecular Weight (MMW, 41 kDa–150 kDa), and Low Molecular Weight (10–40 kDa) (Figure 4C). We are aware that this classification is not ideal, but it helps forming new hypotheses about the repair of DPCs by TDP2. For example, this classification showed that TDP2 deficiency causes the largest increase in protein crosslinks of medium molecular weight (MMW) (Figure 4C), suggesting that TDP2 might be involved in the repair of DPCs in that size range.

Silencing of tdp2a did not have significant effects on DPCs in any size range compared to WT embryos (Figure 4C, Supplementary Figures S5A, B). However, significant differences in the MMW and LMW DPC levels were observed in tdp2b-silenced embryos, which showed a 3-fold (p < 0.01) and 2.3-fold (p < 0.05) increase in MMW and LMW DPCs, respectively, compared to WT embryos (Figure 4C).

Silencing both zebrafish tdp2 orthologs resulted in DPC accumulation comparable to silencing of tdp2b alone, evidenced by a 3-fold (p < 0.01) increase in the MMW and a 2.3-fold (p < 0.05) increase in the LMW DPC range, respectively. Simultaneous silencing of tdp2a and tdp2b, as well as silencing of tdp2b alone, showed similar effects on MMW and LMW DPCs as formaldehyde (FA) treatment, resulting in a 2.2-fold (p < 0.001) and 2.3-fold (p < 0.01) increase, respectively, compared to WT levels. Silencing of tdp2b also led to an accumulation of HMW DPCs, showing a 2-fold change (p < 0.05), reflecting the same increase observed in tdp2a/b double morphants (2-fold, p < 0.05). This increase was again comparable to that observed when DPCs were induced by FA, resulting in a 1.4-fold increase (p < 0.001) (Figures 4A, C; Supplementary Figures S5A, B).

To confirm that the increase in DPC levels was specifically due to loss of Tdp2b, we measured DPC accumulation following Tdp2b overexpression in tdp2b-silenced embryos. Indeed, co-injection of Tdp2b mRNA with the tdp2b morpholino was able to partially reduce total DPC levels (from 2.1-fold to 1.5-fold); MMW and LMW DPC levels were reduced from 3-fold to 2-fold (p < 0.05) and from 2.3-fold to 1.8-fold, respectively, while HMW DPC levels were unaffected (Figure 4 and Supplementary Figures S5A–C).

We further analysed the DPC isolates in order to identify which crosslinked proteins were accumulated in zebrafish embryos as a consequence of Tdp2b silencing. The levels of a known substrate of Tdp2b, Top2, were increased by 1.7-fold after Tdp2b silencing, and the overexpression of Tdp2b in the tdp2b-silenced embryos resulted in the complete rescue of the observed Top2-DPC increase (Figure 5A and Supplementary Figure S5D). Other DPC substrates of Tdp2 are currently unknown. We decided to test if two other DPCs which are known to be one of the most abundant in the cells under physiological condition (i.e., endogenous DPCs), Ku80 and histone H3 (Kiianitsa and Maizels, 2020) are affected by Tdp2b silencing. We observed 1.9-fold increase in Ku80-DPCs as a consequence of tdp2b silencing and Tdp2b overexpression rescued the Ku80-DPC levels (Figure 5B and Supplementary Figure S5E). In contrast, histone H3-DPCs accumulated in tdp2b-silenced embryos, but overexpression of Tdp2b did not reduce the observed accumulation (Figure 5C and Supplementary Figure S5F).

We wanted to investigate whether TDP2 deficiency leads to an increase in double strand breaks (DSBs) as it has previously been shown in cell culture that the accumulation of DPCs leads to an increase in DSBs (Vaz et al., 2016). Although it remains unknown whether silencing of TDP2 in human cells under physiological conditions leads to DSB accumulation, several studies showed a significant increase in DSBs after treatment with etoposide in cells lacking functional TDP2 (Gómez-Herreros et al., 2013; Zagnoli-Vieira et al., 2018).

Phosphorylation of histone H2AX at serine 139 (γH2AX) is an early marker for DSBs and occurs upon recognition of DSBs by the DNA damage-dependent kinases ATM, ATR, and DNA-PK (Paull et al., 2000; Revet et al., 2011). To investigate whether tdp2 silencing leads to DSB formation at the organismal level, we quantified γH2AX levels after tdp2a and tdp2b silencing, while WT embryos exposed to the model DSB inducers formaldehyde (FA) (10 mM, 30 min, 28°C) and etoposide (ETO) (50 μM, 1 h, 28°C) were used as positive controls and reference points (Muslimović et al., 2009; Kumari et al., 2012). Silencing of tdp2a had no effect on DSB levels in 2 dpf embryos, while silencing of tdp2b caused a 1.8-fold increase in γH2AX levels, similar to the effect of simultaneous silencing of tdp2a and tdp2b (1.9-fold increase) (Figures 5D, E). In comparison, FA and ETO treatments of WT embryos caused a 2.5-fold and 2.1-fold increase in DSBs, respectively (Figures 5D, E).

Since overexpression of Tdp2b successfully rescued Tdp2b activity and total DPC accumulation in Tdp2b-deficient embryos (Figures 3A, B; Figure 4), we investigated whether it could also rescue the pronounced DSB accumulation induced by tdp2b silencing (Figures 5D, E and Supplementary Figure S3F). Indeed, Tdp2b overexpression reduced the elevated DSB levels in tdp2b-silenced embryos, back to WT levels (Figures 5D, E and Supplementary Figure S3F). Interestingly, overexpression of Tdp2b in WT embryos significantly further reduced γH2AX levels compared to WT (Supplementary Figure S3F).