Who leads Skydell's training?

Dr. Felix Lurye and Lisa Speer, architects of the Florida Bill.

How is SM-C used in aesthetics?

SM-C is a secretome topical for collagen stimulation with 900B particles.

How is cell viability verified?

Via COA detailing 88.5–95% viability and zero bacterial load.

How do I store biologics?

SM-C at -20°C (90 days); Stem Cells at -80°C (2 years).

Skydell Article Detail

Download PDF

Subjects

Abstract

Endonuclease G (EndoG) is an evolutionarily conserved enzyme that cleaves the Mixed Lineage Leukemia breakpoint cluster region (MLLbcr) under sublethal chemotherapeutic treatment conditions, causing leukemogenic chromosomal rearrangements. While endogenous inhibitors (EndoGI) control EndoG in lower organisms, no such EndoGI has been identified in mammalian cells. Due to the structural similarity of EndoGI from Drosophila melanogaster to the C-terminus (Ct) of human Ku80, we perform immunoprecipitation, surface plasmon resonance analysis and 3D molecular modeling, revealing binding of human EndoG to Ku80-Ct putatively between amino acid 110–184. Docking modeling predicts EndoGI-like peptides clustering around residues 686-707 of Ku80. Our experimental studies provide evidence that Ku80-Ct and 28-mer peptide Ku3 reduce MLLbcr breakage after doxorubicin treatment independently of DNA-PK activity. Proximity ligation and single molecule tracking studies show that Ku3 antagonizes Ku80-EndoG association and modulates chromatin-binding of EndoG. Such MLLbcr protection blocks EndoG´s pro-tumorigenic functions without limiting cytotoxicity, pursued for co-treatments that reduce secondary leukemia, a severe side effect of chemotherapy.

Introduction

Patients with therapy-related acute myeloid leukemia (AML) or, more rarely, acute lymphoblastic leukemia (ALL) represent a subset of patients who have received cytotoxic therapies, including chemotherapy and/or radiotherapy for treatment of a prior malignancy. Therapy-related AML or ALL also is referred to as secondary leukemia and characterized by poor prognosis¹. Chromosomal rearrangements in a 0.4 kb hotspot of clustered breaks, crossing the intron 11-exon 12 boundary in the MLL gene (MLLbcr), are the most frequently observed molecular changes in secondary leukemia and in de novo infant leukemia^1,2,3. Compilation of breakpoint distribution data from patients together with experimental results have connected clustered breakage at the MLLbcr with sensitivity to various cytotoxic drugs, such as topoisomerase inhibitors. Additional cell context-dependent features destabilizing the MLLbcr range from chromosome loop anchoring by CTCF and Cohesin, open chromatin (DNAse I hypersensitivity), formation of a complex non-B DNA secondary structure predisposing genomic regions to breakage, RNA Polymerase II binding, transcription and R-loop formation to replication stress as well as apoptotic cleavage^1,3,4,5,6.

Endonuclease G (EndoG) has been identified as a key molecule in these leukemia-causing processes^2,7,8. This ancient nuclease was originally reported to reside in the intermembrane space of mitochondria, translocating to the nucleus at the onset of apoptosis, where it fragments the genome independently of caspases. However, apoptosis-unrelated functions during stress responses were suggested to exist before pro-apoptotic specialization during evolution, such as the involvement of EndoG in DNA recombination⁹. Meanwhile, it has become clear that genome cleavage by apoptotic nucleases can also proceed sublethally and, in turn, trigger cancer-causing genome alterations^2,8,10. After replication stress, which can be induced by chemotherapy, sublethal breakage occurs. The breakage is followed by rearrangements at the MLLbcr. EndoG, which can process single- and double-stranded DNA as monomeric 3´- 5´ exo- and dimeric endonuclease, respectively¹¹, was identified as a trigger for these processes, acting downstream of DNA damage tolerance and base excision repair (BER) components at the MLLbcr^6,7.

Therefore, inhibitors of EndoG are promising candidates to protect normal cells when DNA-damaging chemotherapeutics are used to kill cancer cells. So far, natural EndoG inhibitors have not been described in mammals. In D. melanogaster, EndoGI forms a complex with EndoG as demonstrated by the solved 3D crystallographic structure¹². Loll and colleagues¹² reported that EndoGI Domain 1 from Drosophila is structurally most closely related to the C-terminal domain of Ku80 (Ku80-Ct, aa 566–732)¹³. Searching aa sequences related to Domain 1 reveals similarities with Ku C-terminal domain-containing proteins. Therefore, we wanted to test the hypothesis whether endogenous human EndoGI could be related to human Ku80-Ct and would protect from MLL rearrangements.

Ku80 is well-known to form a ring-shaped heterodimer with Ku70, which recognizes DNA ends with high affinity^14,15. The Ku70/80 (Ku) heterodimer recruits the DNA-dependent protein kinase catalytic subunit (DNA-PKcs) to form the active holoenzyme DNA-PK at DNA double-strand breaks (DSBs), subsequently phosphorylating itself and downstream DNA damage response factors¹⁶. DNA-PK tethers DNA ends and assembles the XRCC4-DNA ligase IV (LIG4) ligation complex, executing non-homologous end joining (NHEJ), the major repair pathway for two-ended DSBs. Of note, Ku also has the capability to bind and translocate along DNA structures other than blunt DNA ends, namely, cruciform structures or regressed replication forks¹⁷. Forming the so-called tool belt around DNA, Ku also interacts with multiple DNA repair, chromatin and transcription factors other than the core NHEJ components^16,18. It was therefore not surprising to see accumulating evidence for non-canonical Ku functions, namely in controlling BER during NHEJ to prevent formation of additional breaks¹⁹, in regulating gene expression through interaction with the BER and redox factor Apurinic/Apyrimidinic Endonuclease 1 (APE1)/Redox Effector Factor-1 (REF-1)²⁰, in rRNA processing and hematopoiesis²¹, in cytoplasmic DNA sensing, which potentiates T cell activation²², or in promoting fork reversal and chemoresistance²³. Along this line, Ku was classically viewed to protect DSBs from unscheduled exonucleolytic processing, but more recent data from Schizosaccharomyces pombe uncovered a role in fine-tuning end-resection and recombination-mediated replication restart upon binding to the single DNA end of reversed forks^24,25.

Here, we test the hypothesis that the Ku80-Ct resembles an endogenous, human EndoGI under conditions of treatment with the anthracycline doxorubicin, a clinically highly relevant genotoxic drug acting through DNA intercalation, adduct formation and generation of reactive oxygens^26,27,28. Anthracyclines are widely used for the treatment of many tumors and have been associated with MLL rearrangements in secondary leukemia¹. We utilize pre-established MLLbcr reporter cells²⁹ and biochemical assays to monitor dynamic changes at the MLLbcr regarding EndoG binding, breakage and rearrangements³⁰. Supporting our concept, we reveal that Ku80-Ct shows protective effects on the MLLbcr integrity with respect to all three outcomes, while opposing effects are noticeable for full-length Ku80. These activities do not dampen the cytotoxic effect of doxorubicin and can be separated from canonical Ku80 functions by mutant design and repair measurements of targeted DSBs. Using biomolecular simulations and experimental testing of in silico designed candidates, we identify peptides mimicking the effects of Ku80-Ct at the MLLbcr. We use proximity ligation assay, single molecule tracking and clustering analysis of the EndoG-Ku3 binding simulations to show that the top hit peptide interferes with EndoG-Ku80 complex formation and modulates chromatin binding of EndoG dimers. A shared motif of the candidate peptides plays an important role in this modulation. Based on these discoveries, we suggest that a peptide derived from Ku80-Ct can recapitulate the MLLbcr protecting effect of Ku80-Ct, highlighting the potential of peptide-based protective drugs as part of cytotoxic cancer treatment regimens to prevent secondary leukemias.

Results

Ku80-Ct mitigates MLLbcr rearrangements but not the repair component

We have previously shown that replication stress-inducing agents such as the chemotherapeutic drug doxorubicin induce EndoG binding to the MLLbcr, its cleavage and genomic rearrangements^6,7,30. Aiming at the identification of a specific inhibitor of such tumorigenic processes, we sought to determine whether the C-terminal domain of Ku80, showing structural similarities to EndoGI from lower organisms¹², affects such therapy-induced rearrangements. Supporting our concept, we observed pull-down of EndoG when immunoprecipitating endogenous Ku80 (Supplementary Fig. 1a). In surface plasmon resonance measurements (SPR) with a Ku80 fragment containing the C-terminal amino acids (aa) 384–732, we observed a strong binding affinity to EndoG with an equilibrium dissociation constant K_D of 72.7 ± 13.2 nM determined from single-cycle triplicate measurements (Fig. 1a, b). Therefore, we ectopically expressed the structurally defined C-terminal domain between aa 592 to 709 of Ku80 (Ku80-Ct) fused to the nuclear localization signal (NLS) (Fig. 1c). Importantly, this Ku80 fragment lacks the core of wild-type Ku80 (Ku80-wt), known to bind DNA ends and to form heterodimers with Ku70^31,32, as well as the major DNA-PKcs interaction site at the C-terminal end of Ku80-wt^13,33. Additionally, we created the full-length Ku80 variant with aa exchanges E720A and E721A, selectively abrogating binding of DNA-PKcs³³. To quantify MLLbcr rearrangements, we measured recombination at the MLLbcr in an EGFP-based reporter integrated into the genome of K562 leukemia cells following a 4 h treatment with doxorubicin (Fig. 1d)^29,30. Ku80-wt and Ku80-Ct proteins were detectable in these reporter cells (Fig. 1e). Recombination measurements revealed similar frequencies in cells with and without ectopic expression of Ku80-wt. In contrast, the recombination frequency was halved after expression of Ku80-Ct (Fig. 1f). Recombination reduction by Ku80-Ct was neither accompanied by viability changes, estimated by SSC/FSC-gating during FACS analyses (Supplementary Fig. 1b), nor by changes in the cell cycle distribution or the percentage of apoptotic cells according to DNA content analysis (Supplementary Fig. 1c).

Fig. 1: Ku80-Ct binds EndoG and mitigates doxorubicin-induced rearrangements adjacent to the MLLbcr. — **Fig. 1: Ku80-Ct binds EndoG and mitigates doxorubicin-induced rearrangements adjacent to the *MLL*bcr.**

To understand whether Ku80-Ct affects the DNA exchange processes underlying MLLbcr rearrangements in doxorubicin-treated reporter cells⁶, we performed DSB repair experiments using the same reporter construct but replacing the MLLbcr spacer between the mutated EGFP genes by a hygromycin resistance cassette that fails to stimulate recombination in response to genotoxic treatment²⁹ (Supplementary Fig. 2a). Using this construct, homologous DSB repair can be triggered by targeted cleavage upon expression of the meganuclease I-SceI instead of chemotherapeutic treatment³⁴. We found that in K562, neither co-expressed Ku80-wt nor Ku80-Ct altered DSB-induced recombination frequencies significantly when compared to the control (Supplementary Fig. 2b). To further check a possible effect on the balance between DSB repair pathways, we measured NHEJ using an EGFP-based reporter with I-SceI-recognition sites (Supplementary Fig. 2a, c). However, no Ku80-wt or Ku80-Ct dependent effect on NHEJ was found (Supplementary Fig. 2c). To account for the fact that repair of DSBs generated by targeted cleavage might be different from repair of doxorubicin-induced breaks, we measured MLLbcr rearrangements in K562 reporter cells replacing doxorubicin by etoposide, also known to cleave the MLLbcr⁷. Though recombination frequencies were stimulated by etoposide- versus mock-treatment, Ku80-Ct had no effect on recombination (Fig. 1g) nor did it affect viabilities during etoposide treatment, regardless of the length of treatment (4 h vs. 72 h; Supplementary Fig. 1d). As compared to the mode-of-action of doxorubicin, which relies on DNA adduct formation and DNA intercalation and thereby induction of R-loops, replication stress and only after fork collapse DSBs, etoposide directly generates DSBs through topoisomerase II-poisoning³⁵. We conclude that Ku80-Ct mitigates MLLbcr rearrangements, which cannot be explained by effects on DSB repair, cell cycle changes or cell viabilities.

Ku80-Ct prevents accumulation of EndoG in chromatin and binding to the MLLbcr, protecting the integrity of the therapy-related breakpoint cluster

Rearrangements at fragile sites in the genome can be triggered by various stimuli such as secondary structures in the DNA, replication-transcription conflicts or DNA lesions, most notably breaks¹. We have previously identified EndoG as the nuclease causing MLLbcr breakage during replication stress in different hematopoietic and tumor cell types^7,29. Prior analysis had also shown that in response to doxorubicin treatment, EndoG forms a few but distinct nuclear foci and binds to the MLLbcr³⁰. We therefore hypothesized that Ku80-Ct could protect the MLLbcr from such EndoG action. To test this hypothesis, we performed immunofluorescence microscopy on doxorubicin-treated HeLa cells, which have previously been shown to serve as a representative cell model for the processes leading to MLLbcr rearrangements^6,7,30. Ectopic expression of Ku80-wt doubled the average number of nuclear EndoG foci in doxorubicin-treated cells (Fig. 2a, b), indicating enhanced chromatin association. Strikingly, Ku80-Ct abrogated foci formation (Fig. 2a, b). For comparison, mock-treated cells did not show statistically significant differences in EndoG foci numbers. We also performed microscopic analysis of nuclear Ku80 intensities under these conditions, which indicated lower nuclear Ku80 signals in cells ectopically expressing Ku80-wt both after doxorubicin- and mock-treatment (Supplementary Fig. 3a). To understand whether Ku80-Ct modulates binding of EndoG to the MLLbcr, we performed chromatin immunoprecipitation (ChIP) experiments post doxorubicin treatment. Consistent with EndoG foci analysis, expression of Ku80-wt caused the most pronounced binding of EndoG to the MLLbcr region as judged from PCR analyses of the EndoG pull-downs. Accordingly, PCR amplification of genomic DNA adjacent to the MLLbcr region was less prominent in the EndoG-specific precipitates after Ku80-Ct expression or in controls (Fig. 2c). Of note, this decrease was detectable even though PCR analysis of input genomic DNA revealed the highest integrity of the MLLbcr region in cells expressing Ku80-Ct, showing 1.5-fold and 2.0-fold stronger MLLbcr-specific band intensities than in Ku80-wt expressing and control cells, respectively (Fig. 2d). These results suggested that Ku80-wt promotes recognition of the MLLbcr by EndoG, while Ku80-Ct protects this genomic region.

Fig. 2: Ku80-Ct interferes with nuclear foci formation of EndoG, MLLbcr binding and breakage. — **Fig. 2: Ku80-Ct interferes with nuclear foci formation of EndoG, *MLL*bcr binding and breakage.**

To more directly detect DNA damage at the MLLbcr, we also performed ChIP experiments using antibodies directed against histone H2AX phosphorylated on Ser139 (γH2AX), a well-established marker for various DNA lesions ranging from DSBs to sites of replication stress. As compared to control and Ku80-Ct-positive cells, cells ectopically expressing Ku80-wt showed enhanced PCR amplification of the MLLbcr region in γH2AX pull-downs (Fig. 2c). To understand whether such differences were related to overall changes in DNA repair, we evaluated γH2AX signals in the whole nucleus. As expected, doxorubicin treatment augmented the percentage of cells with elevated γH2AX signals, which was also seen in Ku80-Ct-positive cells (Supplementary Fig. 3b). We observed neither differences in the cell cycle distribution nor the percentage of apoptotic cells after Ku80-wt or Ku80-Ct expression, excluding related changes as possible confounding factors (Supplementary Fig. 3c). Even after prolonged (24 h) treatment with doxorubicin similar viabilities were found (Supplementary Fig. 3d). Collectively, our findings suggest that contrary to full-length Ku80-wt, which promotes MLLbcr association of EndoG in the chromatin and local DNA damage signaling without destabilizing the MLLbcr region, Ku80-Ct prevents chromatin association of EndoG and protects the integrity of the MLLbcr after treatment with the chemotherapeutic doxorubicin.

Structure-based discovery of peptide inhibitors of human EndoG

We aimed to identify Ku80-based inhibitory peptides that target human EndoG in a similar way as the EndoG/EndoGI of Drosophila melanogaster¹² (Fig. 3a, top). First, we built a homology model of human EndoG (see the Methods section for details). Next, we generated 5050 Ku80-based peptides with a length of 20 or more residues, from the aa sequence of the C-terminal region of Ku80 (PDB ID: 1Q2Z³⁶). We screened these peptides for their interaction with human EndoG, using the tool PPI-Detect³⁷, a sequence-based predictor of protein–protein and protein–peptide interaction likelihood. Peptide sequences accepted as interacting were those with a score higher than the score of the Ku80-Ct sequence. On this basis, we identified 356 peptide candidates with lengths between 20 and 60 residues. The most conserved region among these peptides corresponds to residues 687-693 of Ku80 (sequence: TKEEASG) (Fig. 3a, bottom). We then carried out peptide–protein docking calculations on the human EndoG model using the subset of the 137 peptides with length ≤ 30 residues and the best interaction scores from PPI-Detect. We performed the docking calculations using CABS-Dock³⁸. The docking analysis allowed us to identify a binding site of the peptides on the protein corresponding to residues 172–252 of EndoG. To further evaluate and filter the peptides, we ranked them according to the extent of their interactions with EndoG. The selection criterion was that more than 50% of the peptide length should be in contact with EndoG, thus leading to 13 candidate peptides (Table 1 and Fig. 3b). The contacts were determined using a cutoff distance of 10 Å between the alpha carbon atoms of EndoG and the peptide. Two of these candidates correspond to the N-terminal part of Ku80-Ct, while the remaining eleven peptides belong to the C-terminal region. Noteworthy, the motif EEA is present in all the selected candidates at least once, which suggests that these residues may be critical for binding to EndoG. To further refine the list of peptide candidates, we calculated their water solubility with the Innovagen (https://pepcalc.com/peptide-solubility-calculator.php) and PROSO II³⁹ tools, resulting in four peptides predicted to be soluble in water (Table 1).

**Fig. 3: Structure-based discovery of the Ku3 peptide.**

Table 1 List of selected Ku peptide candidates targeting EndoG and their properties

Full size table

Ku80-Ct-derived peptide Ku3 reduces MLLbcr rearrangements and breakage

Furthermore, we assessed whether the four EndoG inhibitory peptides identified in silico limit rearrangements at the MLLbcr. Two peptides (Ku2 and Ku4, Table 1) were derived from the N-terminal part of Ku80-Ct and another two from the C-terminal region (Ku1, Ku3) (Fig. 4a and Table 1). For recombination measurements, MLLbcr reporter cells derived from K562 were nucleofected with increasing amounts of each peptide, treated with doxorubicin to promote MLLbcr breakage³⁰ and thereafter released for rearrangements (Fig. 4b and Supplementary Fig. 4). Among the four peptides, we found that Ku3 downregulates recombination by ≤ 35% in a range from 50 to 175 μg peptide, while Ku1 showed such an effect only with 300 μg peptide (Fig. 4b and Supplementary Fig. 4a, b). Neither Ku2 nor Ku4 caused a recombination decrease, but elevated frequencies associated with data scattering at peptide levels of ≥ 300 μg (Supplementary Fig. 4c, d). Such aberrantly high frequencies were also noticed in water controls with Ku2 and Ku4 and accompanied by reduced viabilities (Supplementary Fig. 4c, d, lower panels), suggesting unspecific toxicities. Contrarily, after the introduction of Ku1 or Ku3, we did not observe viability changes after doxorubicin or mock-treatment (Supplementary Fig. 4a, b, lower panel). We confirmed that the scrambled version of Ku3 had no effect on recombination in the K562 reporter cells demonstrating sequence specificity (Supplementary Fig. 4e). Furthermore, we also measured decreases of the recombination frequencies without toxic effects when introducing Ku3 peptide at 25 to 75 μg into MLLbcr reporter cells derived from HeLa (Fig. 4c and Supplementary Fig. 4f). Efficient nuclear uptake of the peptide was visualized by use of chemically synthesized Ku3 N-terminally labeled with TAMRA dye (Fig. 4d). Collectively, peptides were found that inhibit MLLbcr rearrangements without mitigating the cytotoxic effects of the chemotherapeutic agent doxorubicin. Active peptides Ku3 and Ku1 were derived from the C-terminal extreme of Ku80-Ct, where the majority of peptide candidates clustered, and show a sequence overlap of 18 aa (ITKEEASGSSVTAEEAKK) comprising the motif EEA twice, which was present in all the identified candidates at least once (Fig. 4a and Table 1).

Fig. 4: Ku80-Ct-derived peptide Ku3 partially mitigates MLLbcr rearrangements. — **Fig. 4: Ku80-Ct-derived peptide Ku3 partially mitigates *MLL*bcr rearrangements.**

We next investigated the integrity of the MLLbcr region by genomic PCR analysis in the presence of peptide Ku3 (Fig. 5a). Doxorubicin treatment of the HeLa reporter cells caused a significant decrease (26%) of the MLLbcr-specific PCR band intensity (Fig. 5b). Nucleofection of the cells with 75 μg of Ku3 alleviated the doxorubicin-mediated destabilizing effect on the MLLbcr without inducing a change in mock-treated cells (Fig. 5b). Protection of the MLLbcr region in such chemotherapeutically treated cells was revealed by a 45% more pronounced PCR band intensity, matching the 1.5-fold increase in cells expressing Ku80-Ct as compared to Ku80-wt (Fig. 2d). Next, we performed PCR analysis in doxorubicin-treated K562 reporter cells, where Ku3 was found to more robustly downregulate MLLbcr rearrangements (Fig. 4b). To discriminate between different endogenous and reporter-based MLLbcr loci, we PCR amplified within the genomic MLLbcr locus (MLLbcr-1), adjacent to this genomic locus (MLLbcr-2) or across the boundary of the HR-EGFP cassette and the MLLbcr spacer within the recombination reporter (EGFP/MLLbcr-1), relative to stable control regions in the genome under conditions of chemotherapeutic treatment (GAPDH, RARαbcr, Intron20 in MLL) (Fig. 5a)^6,7,30. On average, we observed a 1.5-fold increase of the MLLbcr band intensities (Fig. 5b, c and Supplementary Fig. 5a), which reflected endogenous and reporter-based MLLbcr integrities. We noticed less pronounced protection when monitoring the MLLbcr core (MLLbcr-1), likely due to stable secondary structure formation¹. Importantly, we found comparable MLLbcr protection by Ku3 with and without the DNA-PK inhibitor (DNA-PKi) NU7441 (Supplementary Fig. 5a). When testing the effect of Ku3 on MLLbcr rearrangements in the absence and presence of DNA-PKi, we found significant (p < 0.05) and trendwise (p < 0.10), respectively, mitigation of MLLbcr rearrangements by Ku3. However, homologous DSB repair was augmented by DNA-PKi, likely due to a pathway shift, as seen after expression of I-SceI (Supplementary Fig. 5b, right panel). Such a pathway shift might introduce a confounding factor for the detection of the Ku3 effect. To circumvent a pathway shift during DSB repair, we performed measurements of NHEJ using a specific NHEJ reporter showing that Ku3, like Ku80-Ct, did not affect NHEJ frequencies (Supplementary Figs. 2c, 5c). Together with earlier screening data demonstrating that knockdown of Ku80 and Ku70, like EndoG, reduces MLLbcr rearrangements by 28-49%, while we saw a decrease by only 11% after knockdown of DNA-PKcs⁶, our data strengthen the idea of an NHEJ- and at least partially DNA-PKcs-independent effect of Ku3 on MLLbcr rearrangements. These data show that the 28 aa peptide Ku3 derived from the C-terminal end of Ku80-Ct is sufficient to partially mitigate chemotherapy-related destabilizing effects on the MLLbcr region.

Fig. 5: Ku3 interferes with MLLbcr cleavage. — **Fig. 5: Ku3 interferes with *MLL*bcr cleavage.**

Peptide Ku3 sequence-specifically interferes with Ku80 and EndoG complex formation in the nucleus

The identification of Ku80-Ct-derived peptide candidates was based on higher interaction scores with respect to the Ku80-Ct sequence, followed by further selection according to the docking simulations on the human EndoG model (Fig. 3). Consequently, we investigated whether the exogenous peptide Ku3 interferes with complex formation between endogenous Ku80 and EndoG. Focal accumulation of EndoG in the chromatin is rare even after doxorubicin treatment³⁰, which is in line with limited DNA strand breakage by EndoG in non-apoptotic cells⁸. Therefore, we visualized such Ku80-EndoG association by the sensitive in situ proximity ligation assay (PLA). As expected, only a few PLA foci indicating Ku80-EndoG complex formation were detected in HeLa reporter cells. Notably, PLA signals rose two- to three-fold after treatment with doxorubicin (Fig. 6a). When comparing cells nucleofected with 100 μg Ku3 with the mock controls, we observed a significant reduction of PLA foci numbers mediated by Ku3 (Fig. 6b). Such downregulation was not observed with 100 μg Ku1, the peptide sharing 18 aa with Ku3 but showing an anti-recombinogenic effect only at a higher level (Supplementary Fig. 4b). Direct comparison of cells nucleofected with Ku3 and with the scrambled version of Ku3 validated the inhibitory effect of Ku3, excluding a mere charge effect.

**Fig. 6: Ku3 sequence-specifically interferes with the formation of complexes between Ku80 and EndoG.**

To gain further understanding of the interaction between EndoG and Ku3, we performed extensive molecular dynamics simulations, positioning the peptide far away from EndoG, for unbiased binding. Clustering analysis of the simulations showed that Ku3 binds at an EndoG region similar to the docking’s predictions (Clusters 1 and 5, Supplementary Fig. 6a). In addition, Ku3 can bind regions close to the Mg²⁺ cation in each EndoG monomer (Clusters 2 and 3, Supplementary Fig. 6b). Interestingly, these regions are also similar to the analogous binding interfaces in the Drosophila’s EndoG-EndoGI complex¹² (Fig. 3a, top), which is also similar to the EndoG-Ku80 complex. Binding at these two regions could explain the inhibitory effects of Ku3 on both the activity of EndoG and binding to Ku80. The analysis of the simulations also showed that EndoG experiences major conformational fluctuations, especially upon Ku3 binding and as reflected by changes in the inter-chain angle (θ) (Supplementary Fig. 6c). The analysis of the relationship between the distance (d) of EndoG and Ku3 center of masses (COM) and θ, shows that upon Ku3 binding (d <≈ 45 Å), EndoG undergoes more frequent and wider fluctuations of the inter-chain angle (58° <θ < 100°), than when there is no binding (d > 50 Å, 65° <θ < 75°). This conformational flexibility could also influence the activity of EndoG in vivo.

Next, to further assess the impact of the peptide’s sequence on binding to human EndoG, we generated 100000 Ku3-derivatives with a maximum of 10 mutations with respect to Ku3’s sequence, using PPI-Detect. From this library, 18960 peptides showed an interaction score with EndoG equal or larger than that of Ku3. We then used these derivatives to generate a multiple sequence alignment (MSA) and identify conserved motifs determining peptide binding to EndoG. With this approach, we identified the motif EIVVQDGITLxxxxExEAxSGSSVxxExKK (The x denotes either gaps or non-conserved positions in the MSA), containing the submotif ExEAxSGSS, which we predicted to be important for peptide binding to EndoG with high specificity (Supplementary Fig. 7). This motif features two connected fragments: (1) a fragment containing two negatively charged residues followed by at least one small and hydrophobic residue [ExEAx], and (2) a fragment containing at least four small and polar residues [xSGSS]. With this result, we postulated that the fragment EEASGSS is the minimum active motif.

To test this hypothesis, we exchanged two conserved and differently charged amino acids derived from the 18 residues overlap between Ku3 and Ku1, namely E14 and K27 in Ku3 corresponding to E689 and K702 in Ku80 (Table 1 and Fig. 6c). Interestingly, both Ku3(E689A) (Mut1) and Ku3(K702A) (Mut2) not only abrogated reduction of PLA foci numbers by Ku3 but in fact elevated these Ku80-EndoG specific signals. The residues Ku3_E14 (Ku80_E689) and Ku3_K27 (Ku80_K702) establish electrostatic interactions with EndoG_{R100, R204, K214} and EndoG_{E79, E112} respectively. Moreover, Ku3_D6 (Ku80_D681), which is missing in Ku1, establishes conserved electrostatic interactions with Mg²⁺ and EndoG_R184 (Supplementary Fig. 8). Relative Ku3-specific effects on Ku80-EndoG complex formation were similarly observed in untreated cells (Supplementary Fig. 9a–c). Further supporting the concept of replication-associated action of Ku80-EndoG complexes on the MLLbcr and related sequences under sublethal conditions^6,7_, Ku80-EndoG PLA foci numbers rose 1.5-fold after aphidicolin treatment, inducing replication stress (Supplementary Fig. 9d). Our results thus indicate that Ku3 interferes with the formation of the complex of Ku80 with EndoG in the chromatin in a sequence-specific manner.

Ku3 variants differentially affect EndoG binding to chromatin

Single-molecule tracking (SMT) provides direct insight into the behavior of proteins and enzymes in the environment of a living cell. In particular, SMT reports on kinetic parameters such as the mobility of molecules and their bound fractions and binding times to cellular structures. For example, the binding time is expected to differ between nonspecific, transient binding events and specific, productive interactions.

Using live-cell SMT, we aimed at quantifying the chromatin-bound fraction and chromatin residence time of EndoG compared to inactive control mutants and determine the changes in the presence of Ku80-Ct and Ku80 derived peptides. We therefore fused a HaloTag C-terminally to wild-type EndoG (EndoG-HT), the catalytically dead mutant EndoGΔcat (Supplementary Methods), which potentially exhibits longer residence times due to missing catalysis, and a monomeric mutant EndoG P199E (mEndoG)¹¹, which is expected to exhibit shorter residence times due to a lower affinity, and created stable HeLa cell lines by lentiviral transduction (“Methods”). We confirmed proper localization of the three EndoG-HT variants by visualization with HaloTag-SiR-ligand. They mainly localized to the mitochondria/cytoplasm and only sparsely in the nucleus (Supplementary Fig. 10a), in accord with the location of endogenously expressed EndoG⁶. We further quantified the expression of the EndoG-HT variants by western blotting, which revealed elevated expression levels compared to endogenous EndoG by factors of 4.5 for EndoG-HT, 3.4 for EndoGΔcat-HT, and 1.3 for mEndoG-HT (Supplementary Fig. 10b, c).

We first tested whether elevated expression levels of the Halo-tagged EndoG variants lead to binding saturation. Therefore, we imaged individual molecules labeled with SiR dye coupled to HaloTag ligand (HTL) at 19.3 Hz and detected and tracked single molecules using the software TrackIt⁴⁰. We plotted the number of bound molecules tracked in at least two consecutive frames over the overall number of detected molecules (Supplementary Fig. 10d). A saturation of binding sites would be indicated if the curve approached a horizontal asymptote. However, we observed a close to linear increase, ruling out binding artifacts due to overexpression.

Next, we determined the overall bound fraction of EndoG-HT variants by imaging individual molecules labeled with HTL-PA-JF646 at 85 Hz (Fig. 7a)⁴¹. The histogram of frame-to-frame jump distances obtained from these tracks revealed an increase of short jumps for mEndoG compared to EndoG wt, while short jumps decreased for EndoGΔcat, indicating a lower bound fraction (Fig. 7b and Supplementary Fig. 11). Indeed, detailed analysis of the jump distance distributions using a model of three components of Brownian diffusion⁴⁰ confirmed a higher chromatin-bound fraction of mEndoG compared to EndoG wt, and a lower bound fraction for EndoGΔcat (Supplementary Fig. 12).

**Fig. 7: SMT: peptides differentially affect EndoG binding to chromatin.**

Third, we set out to quantify the residence time of the chromatin-bound fraction of EndoG variants. This analysis required monitoring chromatin-bound molecules over extended time periods. We therefore illuminated the cells with a time-lapse scheme including dark times of variable duration (Fig. 7c, d). The distribution of fluorescence survival times assembled from bound molecules indicated a faster decay and thus shorter residence time for mEndoG compared to EndoG wt, while EndoGΔcat showed on trend, a slower decay, indicating longer residence times (Fig. 7e). To obtain the residence times from the distribution of fluorescence survival times, we performed an inverse Laplace transformation, using the GRID method⁴². GRID yields a spectrum of dissociation rates that a molecular species experiences in the cell, which can be converted to residence times by inversion. The dissociation rate spectra of all EndoG variants consisted of several rate clusters (Fig. 7f and Supplementary Fig. 13). As expected from the distribution of fluorescence survival times, the rate cluster corresponding to the longest residence time was shifted to larger dissociation rates and thus shorter residence time for mEndoG compared to EndoG wt (Fig. 7f, g and Supplementary Table 2). Due to a lower affinity of the monomer to chromatin¹¹, shorter residence times are expected for monomeric compared to dimeric binding. In contrast, the residence time of EndoGΔcat did not markedly differ from the residence time of EndoG wt (Fig. 7g).

By combining the amplitude of the rate cluster corresponding to the longest binding times with the overall bound fraction, we obtained the long-bound fraction of EndoG variants (Fig. 7h). For mEndoG, the long-bound fraction was considerably higher than for EndoG wt, indicating a higher binding frequency of mEndoG. For EndoGΔcat, the long-bound fraction was comparable to the fraction of EndoG wt. Overall, our live-cell SMT experiments demonstrated that our assay is sensitive to differences in chromatin-bound fractions and chromatin residence times of EndoG, mEndoG and EndoGΔcat.

Next, we compared the chromatin-bound fraction and chromatin residence time of EndoG wt in the presence of Ku80-Ct. We therefore repeated our single-molecule measurements and analysis after transient transfection of HeLa cells stably expressing EndoG-HT with a plasmid expressing GFP-Ku80-Ct (EndoG + Ku80-Ct) and a GFP expression plasmid as control (EndoG transf. ctrl) (Supplementary Methods). Selecting for successfully transfected GFP-positive cells, we found similar residence times of EndoG in the presence and absence of Ku80 Ct (Fig. 7g and Supplementary Fig. 13), suggesting that, once bound, EndoG is not affected by the C-terminal domain of Ku80. However, the fraction of long EndoG binding events was reduced in the presence of Ku80-Ct compared to the transfection control (Fig. 7h and Supplementary Figs. 11, 12). This result corroborates our findings from immunofluorescence microscopic experiments, which revealed reduced chromatin binding of EndoG in the presence of Ku80-Ct.

Finally, we compared the chromatin-bound fraction and chromatin residence time of EndoG wt in the presence of the Ku80-Ct-derived peptides Ku3 and Mut1, and a Scrambled control. We again repeated our single-molecule measurements and analysis, now after nucleofection of TAMRA-labeled peptides (Supplementary Methods). Our experiments revealed comparable chromatin residence times for EndoG wt in the presence of Ku3 and Scrambled, while the residence time dropped in the presence of Mut1 (Fig. 7g and Supplementary Fig. 13). Similarly, the long-bound fraction of EndoG wt was comparable in presence of Ku3 and Scrambled. In presence of Mut1, the long-bound fraction increased (Fig. 7h and Supplementary Figs. 11 and 12). Remarkably, while Ku3 did not alter the chromatin binding properties of EndoG wt, the behavior of EndoG wt in the presence of Mut1 mirrored the behavior of EndoG wt compared to mEndoG. Since our clustering analysis of the EndoG-Ku3 binding simulations (Supplementary Fig. 6) showed that Ku3 is theoretically able to contact several interfaces of EndoG and might increase the interchain angle of the EndoG dimer, it is tempting to speculate that Ku3 affects the activity of EndoG through modification of the dynamics of the EndoG dimer.

Discussion

The ancient nuclease EndoG needs to be well-controlled to protect cellular RNA and DNA from endonucleolytic attack and degradation, as indicated by the co-existence of naturally occurring endogenous inhibitory proteins in several lower organisms^43,44. Such EndoG activity can cause rearrangements at the MLLbcr in human hematopoietic stem and progenitor cells, which are known to induce secondary leukemia in response to chemotherapeutic treatment^1,6,7. Here, we show that the 3D structure of the C-terminal domain from the human DNA repair protein Ku80 closely resembles the EndoGI from Drosophila melanogaster. Our results from sequence alignments and molecular modeling reveal the binding interface of human EndoG between aa 110–184 with human Ku80-Ct. Peptides of 20–30 residues derived from Ku80 Ct were identified as candidates with larger predicted binding affinity than Ku80-Ct, clustering around residues 686–707 and to a lesser extent also around residues 599–624 within full-length Ku80. Our experiments validate these structure-based predictions, demonstrating that peptide Ku3, derived from the extreme C-terminal cluster in Ku80-Ct, reduces (i) Ku80-EndoG complex formation in the nuclear chromatin, (ii) breakage of the chromosomal MLLbcr region and (iii) rearrangements adjacent to the MLLbcr after treatment with the chemotherapeutic agent doxorubicin. Such a protective effect was found to be specific for the peptide sequence, it neither exhibited toxic effects nor inhibited the cytotoxic effect of the chemotherapeutic agent, presenting Ku3 as a highly interesting lead compound for further research aiming at preventing the development of secondary leukemia during cancer therapy.

This study also uncovers that Ku80-Ct resembles an EndoGI-like protein fragment, which is based on primary and 3D structure comparisons and is supported by Ku80-Ct´s antagonistic effect on treatment-induced MLLbcr destabilization, previously established to depend on EndoG^6,7. The part of Ku80 corresponding to Ku80-Ct has been thought to serve as an extended, flexible arm that connects DNA-PKcs to DNA-bound Ku70-Ku80, thereby assembling the protein complex initiating NHEJ^45,46. Our results provide evidence for the interference of Ku80-Ct with MLLbcr rearrangements independently of DSB repair, alluding to another role of Ku80 exerted via this domain. Thus, DSB repair triggered by targeted cleavage through I-SceI was not affected by Ku80-Ct, neither when engaging a reporter construct for NHEJ nor for homologous DSB repair without the therapy-responsive spacer MLLbcr. Moreover, Ku80-Ct, like Ku3, is devoid of the major binding sites for DNA, Ku70 and DNA-PKcs^{31,32,46,47,48}. Finally, peptide Ku3 does not influence NHEJ but interferes with MLLbcr breakage independently of DNA-PK activity.

The ring-shaped Ku70-Ku80 complex serves as a scaffold for recruitment of repair factors, whereby the disordered Ku80 C-terminus has previously been considered suitable for interactions with other proteins¹⁶. The C-terminus of Ku80 was also reported to stimulate autophosphorylation of DNA-PKcs at specific sites, activating processing of DNA ends via the DNA-PKcs complex partner Artemis⁴⁹. Involvements of Artemis and of DNA-PK activity in MLLbcr rearrangements were excluded in our earlier study⁷. However, siRNA screening in this previous work revealed involvement of Ku80 in MLLbcr rearrangements together with EndoG⁶. Moreover, data from several studies suggest an overlapping set of proteins in the interactomes of EndoG and Ku80, including the base excision repair (BER) enzymes FEN1 and APE1^6,20,50,51. APE1 promotes MLLbcr binding and cleavage by EndoG in response to replication stress, as demonstrated in our previous work⁶. Of note, the Ku heterodimer was reported to possess an intrinsic 5′ deoxyribosephosphate (5′dRP) and apurinic/apyrimidinic (AP) lyase activity, excising abasic sites such as generated by glycosylases from ends, analogous to that exhibited by DNA polymerase ß during BER¹⁸. EndoG and Ku80 are also both involved in class switch recombination (CSR) triggered by the DNA cytidine deaminase AID and downstream BER, preferentially at ssDNA targets of R-loops in G-rich DNA^52,53. The G-rich MLLbcr has the propensity to form stable secondary structures and RNA-DNA hybrids, i.e., represents a potential off-target for such BER-coupled recombination processes^1,6. Ku80, its binding partner Ku70 and EndoG are known to recognize G4 and other non-B-DNA secondary structures in the genome, such as cruciforms or hairpins^17,54,55,56. Doxorubicin exacerbates the fragility of the MLLbcr through the generation of reactive oxygen species leading to clustered oxidative damage in such stable secondary DNA structures. As a DNA intercalator, it interferes with various DNA metabolic processes entailing transcription-replication conflicts and DSBs^1,6,26,27,28. Our model implies Ku complex-mediated recruitment of EndoG to specific sites in the chromatin, where the Ku complex and EndoG synergistically recognize secondary structures such as predicted for the MLLbcr¹ and augment resolution of these replication barriers by EndoG.

In the light of these observations, our data suggest that Ku80 and EndoG might get trapped at the MLLbcr, possibly while cooperating during processing of abasic sites and/or when Ku modulates the order and extent of end resection^{6,25,57,58,59}. We find that ectopic expression of Ku80-wt promotes specific binding of EndoG to the MLLbcr and local accumulation of γH2AX (Fig. 2), suggesting that Ku80 aids recruitment of nucleolytically active EndoG. Along this line, we previously showed that depletion of endogenous Ku80 reduces MLLbcr rearrangements⁶. Overexpression of exogenous Ku80-wt did not further impact on the integrity of the MLLbcr region (Figs. 1f, 2d), possibly reflecting a saturating effect of breaks in the MLLbcr or regarding the balance between Ku80 and EndoG¹⁷. What could be the biological meaning of such a sensitive and cooperative mechanism amplifying DNA damage selectively at fragile sites such as the MLLbcr? Distinct from EndoG´s canonical role in DNA fragmentation during the early stages of apoptosis, limited strand breakage by EndoG has been recognized as a source of mutagenesis and oncogenic transformation^8,10. Similarly, the pro-apoptotic DNAse Caspase 3 was proposed to play a role in mutagenesis after minor mitochondrial membrane permeabilization and sublethal Caspase 3 activation^2,8,10. Considering recent work by Sørensen and colleagues on Caspase 3⁶⁰ nonlethal self-inflicted DNA strand breaks by such nucleases, including EndoG, may serve as early sensors to activate DNA damage response checkpoints and differentiation programs⁶¹.

In this scenario, exogenous Ku80-Ct was expected to exert a dominant-negative effect on the crosstalk between endogenous Ku80 and EndoG, which was supported by the observed loss of EndoG foci in immunofluorescence microscopy (Fig. 2b), a reduced fraction of EndoG stably bound to the chromatin in SMT analysis (Fig. 7), as well as diminished breakage and rearrangements at the MLLbcr (Figs. 1f, 2d). Importantly, Ku80-Ct did not alter doxorubicin-induced toxicities, as deduced from FACS, DNA content, MTT and pan-nuclear γH2AX analyses (Supplementary Figs. 1–3). Taken together, Ku80-Ct possesses the predicted features of a human EndoGI antagonizing the MLLbcr destabilizing effect of EndoG during chemotherapeutic treatment, but without affecting cell killing.

In mitochondria, EndoG levels are high (Supplementary Fig. 10a) and an alternate, C-terminally truncated form of Ku80 is found⁶². However, a Ku80-dependent control mechanism of EndoG may not be necessary in mitochondria, because EndoG resides in the intermembrane space, i.e., is spatially separated from the mitochondrial genome, at least in unstressed cells⁶³. Due to the increased oxidative environment, doxorubicin-induced DNA damage is also elevated in mitochondria²⁶, and our recent works revealed a role of EndoG in nucleolytic removal of irreparably oxidatively damaged copies^64,65. However, multiple remaining copies of the mitochondrial genome and compensatory replication of these intact copies ensure mitochondrial genome homeostasis^64,65. Altogether, while the nucleolytic activity of human EndoG in the nucleus must be tightly controlled to maintain genomic integrity, proposedly through interactions with the EndoGI-like Ku80 C-terminal domain, such a control mechanism might be obsolete in mitochondria.

The peptide Ku3, identified in silico, functionally mimicked EndoGI features of Ku80-Ct (Figs. 3–5 and Supplementary Fig. 6). The predicted EndoG/Ku3 binding interface provides a molecular rationale for the effect of Ku3 as a competitive inhibitor of EndoG-Ku80 interactions. This competitive mode-of-action is experimentally supported by the inhibitory effect of Ku3 on Ku80-EndoG complex formation (Fig. 6). Moreover, loss of such inhibition was detectable for Ku3-derivatives mutated at positions E14 (E689 in Ku80) or K27 (K702 in Ku80) of the conserved motif identified in-silico. The analysis of extensive molecular dynamics simulations delivers a better understanding of the binding and dynamics of Ku3 with EndoG in solution. These simulations confirmed the binding mode predicted by docking calculations and revealed other binding regions at EndoG that further supported the competitive mode-of-action. Experimental support for Ku3 interactions with EndoG that cannot fully be explained by competition with Ku80 might also be extracted from our experimental observations with mutated peptides. Thus, interactions with mutated Ku3 derivatives not only lost the inhibitory effect on the Ku80-EndoG interaction but also even promoted complex formation (Fig. 6 and Supplementary Fig. 9). Importantly, our simulations showed that Ku3 also interacts with the EndoG dimer through coordination of the Mg²⁺ cation and interactions with its adjacent residues. Additionally, binding of Ku3 modifies the dynamics of the EndoG dimer, with an increase of the inter-chain angle, which may impact EndoG activity. Our work hints at the possibility that EndoG is susceptible to an allosteric regulation that may affect DNA binding directly or indirectly through conformational changes. SMT experiments indicated reduced affinity of EndoG chromatin interactions in the presence of mutated versus wild-type Ku3. We hypothesize that peptide Mut1 has lost interactions competing with EndoG-Ku80 complex formation but retains the capacity to promote structural changes of EndoG. This information can be exploited for the design of specific EndoG inhibitors in the future. Furthermore, Ku80-Ct interacts with EndoG through multiple regions along its sequence clustering around residues 686–707 and to a lesser extent also around residues 599–624 (Fig. 4a). Future investigations might therefore explore the possibility of combining peptides through a dual spacer to optimize binding affinities and specificities.

In summary, our work describes a peptide inhibitor of EndoG that selectively blocks its pro-tumorigenic functions but not its pro-apoptotic ones, unlike the previously published chemical inhibitors non-selectively inhibiting EndoG⁶⁶. Nonspecific effects on EndoG´s nuclease activity may prevent the cytotoxic effect of EndoG during chemotherapy, which undesirably mediates resistance of tumor cells. Targeted interference, which selectively blocks only certain molecular interactions, is difficult to achieve with small molecules⁶⁷. Here, we employed computational modeling to identify peptides that are highly specific for the molecular target EndoG and therefore protective against MLLbcr rearrangements during replication stress. Preventive combination therapy with such peptides or derivatives may allow the reduction of a severe side effect of chemo- and radiotherapies, namely the development of secondary tumors, especially therapy-related ALL, therapy-related myelodysplastic syndrome or therapy-related AML.

Methods

Molecular modeling

Homology modeling of human EndoG

The Swiss-Model server⁶⁸ was used to generate a homology model of the human EndoG (Uniprot ID: Q14249) based on 6 templates: 4QN0_A, 3ISM_A, 3ISM_B, 3S5B_A, 3S5B_B, and 5GKC_A. From these templates, nine homology models were built, and the best one, based on the QMEAN quality assessment⁶⁹ was selected.

In silico screening of Ku80-based peptide candidates

We used the sequence of the C-terminal domain of Ku80 (Ku80-Ct, PDB ID: 1Q2Z³⁶) to generate a library of 5050 Ku80-Ct-derived peptides with more than 20 residues. We screened these peptides for their interaction with human EndoG with PPI-Detect³⁷. PPI-Detect is based on a support-vector machine model that predicts the interaction likelihood between sequences. We selected the peptides with an interaction score higher than the full-length Ku80, resulting in 356 peptides with sizes between 20–60 residues.

Molecular docking

From the set of 356 peptides identified with higher interaction scores than the full Ku80, we extracted all the candidates with 30 or fewer residues. Such selection reduced the set of candidates to 137 peptides. All these peptides were docked with the EndoG model using the program CABS-Dock³⁸. We performed 10 replicas of the algorithm, extracting the 100 top-scored binding poses from each replica. The obtained 1000 structures were clustered into 10 clusters using the k-medoids algorithm. For each peptide, the lowest energy and the most populated clusters were identified and extracted as the most suitable binding poses. To further filter the ranked docked candidates, we computed the number of residues of the peptides interacting with EndoG and re-ranked according to the fraction of their length that it is in contact with EndoG in the most stable complex and/or the most populated one. A threshold of 50% of the peptide length was used to extract the peptides with the largest binding interface with EndoG. This analysis resulted in 13 peptides with a large contact area with the targeted binding site on EndoG (Table 1) that were selected for experimental testing.

Identification of Ku3-derived minimal binding motif

Starting from the Ku3 sequence, 100000 derivatives, each one with a maximum of 10 modifications, were screened using PPI-Detect³⁷. From them, 18960 peptides showed an interaction score with EndoG equal to or larger than that of Ku3. These peptides were selected to build a multiple sequence alignment (MSA) to identify conserved motifs that can be responsible for the observed binding. The programs MAFFT (v7)⁷⁰, MaxAlign⁷¹, and EMBOSScons⁷² were used to align, refine and extract conserved motifs, respectively. Consensus positions were accepted if they have positive matches (plurality) of at least half of the entries in the MSA. A WebLogo plot^73,74 was used to visualize the frequency of appearance of each residue per peptide position.

Molecular dynamics simulations

Based on our docking model of EndoG/Ku3, three configurations were prepared where the peptide was separated from the EndoG dimer at distances between their centers of masses (COM) of 61.75 Å, 64.35 Å, and 50.21 Å. We thus ensured that the side chains were sufficiently separated to avoid any intermolecular interaction in the starting structures. CHARMM-GUI⁷⁵ was employed to prepare the systems to run with GROMACS⁷⁶ using the CHARMM36m force field⁷⁷. EndoG and Ku3 were inserted in a cubic simulation box with 15 Å of padding. The box was solvated with the TIP3P water model⁷⁸ and NaCl at 150 mM. After minimization and equilibration, three replicas of 1 μs each per starting configuration (9 replicas in total, each1 μs) of production runs were performed, with a time step of 2 fs. Neighbor searching was performed every 20 steps. The PME algorithm⁷⁹ was used for treating electrostatic interactions with a cutoff of 1.2 nm. A single cutoff of 1.225 nm was used for Van der Waals interactions. Temperature coupling was done with the Nose-Hoover⁸⁰ algorithm. Pressure coupling was done with the Parrinello-Rahman⁸¹ algorithm. The analyses were performed using the last 700 ns of each replica (totaling 6.3 μs) using VMD⁸², CONAN⁸³, and GROMACS’s tools. Conformational clustering of the Ku3 peptide was performed with the VMD plug-in (https://github.com/luisico/clustering), employing an RMSD cutoff of 20 Å for the backbone atoms of the peptides. The distance between the molecules (d) was defined as the distance between their centers of masses (COM). The EndoG interchain angle (θ) was defined by the Cα atoms of EndoG-A_V95, EndoG-A_K214, and EndoG-B_V95 residues. The frequency of the d/θ pairs and the density plot were calculated with 2D binning of 151 × 151 points with OriginPro (v2022b) with the default settings. For the contact analysis with CONAN, a distance cutoff of 5 Å between the residue’s heavy atoms and a lifetime of ≥ 50 % was employed.

Cell culture and treatments

K562 cells (Cellosaurus RRID:CVCL_0004, provided by Heinrich-Pette-Institute, Hamburg, Germany) and the reporter cell line derivative K562MLL, stably transfected with the DNA recombination substrate pHR-EGFP/3’EGFP-MLLbcr.fwd²⁹, were cultured in RPMI 1640 medium (Gibco/Thermo Fisher Scientific, Waltham, Massachusetts, USA) supplemented with 10% fetal bovine serum (Pan-Biotech GmbH, Aidenbach, Germany) and 1% L-glutamine (Gibco/Thermo Fisher Scientific). HeLa cells (Cellosaurus RRID:CVCL_0030, provided by Heinrich-Pette-Institute, Hamburg, Germany) and the reporter cell line derivative HeLaMLL⁷ were cultured in DMEM (Gibco/Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (Biochrom, Berlin, Germany) and 1% L-glutamine (Gibco/Thermo Fisher Scientific). Culturing was performed in a humidified incubator at 37 °C with 5% CO₂. All cell lines were tested for mycoplasma contamination by PCR and confirmed to be negative. All human cell lines were authenticated by Microsynth AG, Balgach, Switzerland, using STR profiling within the last three years.

Twenty-four hours after seeding in a 6-well suspension cell culture plate (SARSTEDT AG & Co. KG, Nümbrecht, Germany), 24 h after plasmid electroporation and 24 h after peptide nucleofection, respectively, K562 (reporter) cells were treated for 4 h with either H₂O or 2.0 µM doxorubicin (University Hospital Ulm) in H₂O or with either DMSO or 10 µM etoposide (Sigma-Aldrich/Merck, St. Louis, Missouri, USA). Twenty-four hours after seeding in a 6-well adherent cell culture plate (SARSTEDT AG & Co. KG), 24 h after plasmid transfection and 24 h after peptide nucleofection, respectively, HeLa (reporter) cells were treated with H₂O or 0.5 µM doxorubicin in H₂O for 4 h, except for viability analysis 24 h after doxorubicin treatment.

For FACS-based recombination measurements, reporter cells were washed in Phosphate Buffered Saline (PBS) and re-cultivated in fresh medium for 72 h before harvest. For immunofluorescence microscopy cells were fixed, for ChIP or genomic DNA isolation cells were harvested immediately after treatment. For co-treatment with doxorubicin and DNA-PK inhibitor, HeLa reporter cells were treated with either H₂O or 1 µM NU7441 (STEMCELL Technologies, Vancouver, Canada) for 1 h before being exposed to H₂O or 0.5 µM doxorubicin in H₂O for 4 h.

Plasmids, electroporation and transfection

For analysis of Ku80-wt, Ku80-Ct or Ku80(E720A, E721A) the following expression plasmids with a pCMV6-AN-DDK backbone were generated (OriGene, Rockville, Maryland, USA): pCMV6-AN-DDK-Ku80 expressing Ku80-wt (SC112985 ORF, OriGene) with N-terminal Flag-tag and pCMV6-AN-DDK-Ku80 C-terminus expressing Ku80-Ct through fusion of the Ku80-NLS motif P562-K568 (encoded by 5´-CCTACAGCTAAAAAATTAAAG-3´).

For homologous DSB repair and NHEJ measurements, we engaged reporter plasmids p5xtrshSV40HygbPuroCMV-HR and EJ5SceGFP, respectively, together with pCMV-I-SceI for targeted cleavage by the meganuclease I-SceI^34,84. Plasmid p5bPuroCMV-wtEGFP, generated for EGFP expression in the reporter plasmid backbone, served to determine transfection efficiencies.

To introduce protein expression plasmids into K562 (reporter) cells, these suspension culture cells were split, cultured for 24 h, washed in RPMI 1640 medium, no glutamine, no phenol red (Gibco/Thermo Fisher Scientific) and resuspended in 400 µl of the same medium. Then, a total amount of 50 µg plasmid for expression of one of the Ku80 variants was added, and the mixture transferred into a Gene Pulser^®/Micropulser^TM electroporation cuvette (Bio-Rad Laboratories, Hercules, California, USA), followed by electroporation at 200 V and 1050 µF with a Gene Pulser Xcell^TM electroporation system (Bio-Rad Laboratories) and recultivation in fresh medium. For homologous DSB repair or NHEJ measurements, K562 cells were electroporated with 40 μg of the Ku80 variant expression plasmid, 10 μg of p5xtrshSV40HygbPuroCMV-H or EJ5SceGFP reporter plasmid, 10 μg of pCMV-I-SceI and 10 μg of filler plasmid pBS (pBlueScriptII KS (+/-) from Stratagene, Heidelberg, Germany), the latter of which was replaced by p5bPuroCMV-wtEGFP in transfection controls. For homologous DSB repair measurements, K562 cells were recultivated for 48 h, for protein isolation and Western blotting for 24 h.

HeLa (reporter) cells were transfected using FuGENE®HD (Promega, Fitchburg, Madison, USA) 24 h after seeding, whereby a mixture of 1.940 ml OptiMEM (Thermo Fisher Scientific), 20 µg plasmid DNA and 40 µl FuGENE^® HD solution were dripped on the cells, followed by 6 h incubation, a medium change and recultivation for 16 h.

Western blot

Cell lysates were prepared from scraped HeLa or centrifuged K562 cells by incubation in lysis buffer (50 mM Tris-HCl, Tris-Base, pH 7.4, 150 mM NaCl, 2 mM EGTA, 2 mM EDTA, 25 mM Sodium fluoride, 25 mM ß-Glycerol phosphate, 0.1 mM Sodium vanadate, 0.2% Triton X-100 0.3% Nonidet P40, protease inhibitor cocktail tablet, PIC, Sigma-Aldrich/Merck St. Louis, Missouri, USA) for 30 min on ice. After clearing the suspension by centrifugation (15 min, 1380 x g, 4 °C), the supernatant was subjected to determination of protein concentrations by use of the BCA Protein Assay Kit (Thermo Fisher Scientific) and to denaturing SDS-polyacrylamide gel electrophoresis, Western transfer and immunodetection of transferred proteins, as described before⁷. Primary antibodies for immunodetection were: anti-Ku80 (rabbit polyclonal, H-300, sc-9034, Santa Cruz Biotechnology Inc., 1/500), anti-Flag M2 (mouse monoclonal, F1804, Sigma-Aldrich/Merck, St. Louis, Missouri, USA, 1/1000), anti-Flag M2-Peroxidase (HRP, mouse monoclonal, A8592, Sigma-Aldrich/Merck, St. Louis, Missouri, USA, 1/1000), anti-EndoG (Mouse monoclonal, B-2, sc-365359, Santa Cruz Biotechnology Inc., 1/500), anti-γH2AX Ser139 (Mouse, monoclonal, Clone JBW 301, 05-636, Merck Millipore, Burlington, Massachusetts, USA, 1/1000), anti-DNA-PKcs phospho S2056 (Rabbit, polyclonal, ab18192, Abcam, 1/1000), anti-DNA-PKcs (mouse monoclonal, ab1832-500, Abcam, 1/200), anti-Vinculin (mouse monoclonal, V9131, Sigma-Aldrich, or mouse, monoclonal, sc73614, Santa Cruz Biotechnology Inc., 1/500) and anti-Tubulin (mouse monoclonal, ab7291-100, Abcam, 1/5000). Secondary antibodies were: horseradish peroxidase (HRP)-conjugated goat-anti-mouse (610-1319, Rockland, Pennsylvania, USA, 1/10000) and goat-anti-rabbit (611-1322, Rockland, Pennsylvania, USA, 1/10000). Western blot signals were visualized by Super Signal West Dura Extended Duration Substrate (Thermo Fisher Scientific) and a ChemiDocTM MP Imaging System (Bio-Rad Laboratories, Hercules, California, USA). Quantification of band intensities was performed with Image Lab 5.2.1 (Bio-Rad Laboratories) and corrected with the values of the corresponding anti-Tubulin loading controls.

FACS-based recombination, cell cycle and viability analysis

Determination of treatment-induced intrachromosomal recombination frequencies as well as extrachromosomal homologous DSB repair and NHEJ measurements upon expression of meganuclease I-SceI were performed as described previously^6,30. Briefly, fractions of EGFP-positive cells among 0.5-1.0 million viable cells were measured flow cytometrically by use of a FACSCalibur™ and BD CellQuest Pro 5.2.1 (BD Biosciences, San Jose, California, USA) or CytoFLEX B3-R1-V0 flow cytometer with APD detectors and GFP-oD1 bandpass filter and CytExpert, Version 2.5.0.77 (Beckmann Colter, Brea, CA, USA), as indicated. Hereby, viable cells were selected by the SSC/FSC gate and green-fluorescent cells therein detected in the diagonal autofluorescence-FL2/EGFP-FL1 dot plot gate (Supplementary Figs. 1, 2). To correct for interexperimental variations, mean recombination frequencies in appropriate controls were set to 100% in each experiment, and relative percentages calculated for each single value.

DNA content analysis was performed as before, applying the DNA intercalating agent propidium iodide (PI)^6,30. Briefly, cells were washed with PBS, ice-cold fixative (80% ethanol/100% acetone 1:1) added slowly and cells incubated for ≥ 15 min at − 20 °C, washed first in PBS/fixative (1:1) and then in PBS. In the dark, cells were resuspended in 200 μl PI solution (1/20 PI stock solution from Sigma-Aldrich/Merck; 1:400 50 μg/ml RNaseA from OriGene in PBS), incubated at 37 °C for 30 min and stored at RT until FACS analysis (FACSCalibur™, BD Biosciences) to quantify the distribution of viable cells in different cell cycle phases and the percentage of apoptotic cells with sub-G1 DNA content.

Viabilities of HeLa cells treated with doxorubicin were assessed by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide; Sigma-Aldrich/Merck, St. Louis, Missouri, USA) assay as before²⁹. Optical densities were measured with a Tecan Sunrise Photometer (Tecan, Crailsheim, Germany) at 570 nm.

Genomic polymerase chain reaction (PCR)

Genomic DNA was isolated using the QIAamp DNA Mini Kit (Qiagen, Hilden, North Rhine-Westphalia, Germany) according to the manufacturer’s instructions. PCR was performed with 20 ng DNA in a total reaction volume of 30 µl including 1.5 mM MgCl₂ for MLLbcr-2 and 1.0 mM MgCl₂ for MLL (Intron20 in MLL) primer sets when using my-Budget PCR kit (Bio-Budget, Krefeld, Germany) or with 20 ng DNA in a total reaction volume of 50 µl when using Platinum™ II Hot-Start PCR Master Mixes (2X) (Invitrogen/Thermo Fisher Scientific, Waltham, Massachusetts, USA) in a FlexCycler (Analytik Jena, Jena, Germany). PCR products were separated by agarose gel electrophoresis using SeaKem® LE Agarose (Lonza, Basel, Switzerland, 2.5%) or UltraPure™ Agarose-1000 (Invitrogen/Thermo Fisher Scientific, 3%) and visualized by SYBR™ Safe DNA Gel Stain (Invitrogen/Thermo Fisher Scientific). To ensure the reliability of the PCR conditions, two technical controls were included in each experiment. First, to verify that the selected cycle number was within the linear range, a control sample containing 50% of the DNA of the untreated sample was included. Second, a DNA-free control was included to check for potential contamination, i.e., a control sample containing only the PCR master mix without DNA. The primers (Biomers.net, Ulm, Germany) amplifying MLLbcr-specific and control fragments had the following sequences (Fig. 5a): MLLbcr-1 (351 bp fragment at the intron11-exon12 boundary surrounding the major therapy-related hotspot MLLbcr), FWD: 5´-GGCTCACAACAGACTTGGCAAT-3´, REV: 5´-TATTTCCCCCACCCCACTCC-3´; MLLbcr-2 (345 bp fragment, 0.8 kb downstream of the therapy-related hotspot MLLbcr within the MLL breakpoint region), FWD: 5´-TTGGGTGTAATCAGTTGCCTATT-3´, REV: 5´-GGGTGATAGCTGTTTCGGCA-3´; EGFP/MLLbcr (261 bp fragment of the MLLbcr-reporter across the boundary of the HR-EGFP cassette and the MLLbcr spacer), FWD: 5´-ACCACTACCAGCAGAACACC-3´, REV: 5´-ATACGAAACAGTTGTAAGTATCGTC-3´; MLL (236 bp fragment of the genomic MLL intron 20 used as a control in genomic PCR experiments): FWD: 5´-GCAACACAGGGCCCTAGTTAAT-3´, REV: 5´-AGGCAAATCAGTCACCTTTTAATCA-3´; RARα (205 bp fragment of the RARαbcr used as a control in PCR), FWD: 5´-TCCCTCCTCTTCAAGCGTTA-3´, REV: 5´-CCCATCACTCCTCTGGACTC-3´; GAPDH (294 bp fragment of the GAPDH gene promotor, used as a control in PCR), FWD: 5´-CCCAACTTTCCCGCCTCTC-3´, REV: 5´-CAGCCGCCTGGTTCAACTG -3´; The band intensities of MLLbcr-specific amplifications (MLLbcr-1, MLLbcr-2 and/or EGFP/MLLbcr) were corrected using band intensities of control PCR reactions (ctrl: GAPDH, RARα and/or MLL) from the same DNA sample and then normalized to bands from appropriate internal controls run in each experiment and set to 100%, such as obtained from samples treated with doxorubicin but without peptide. Images of agarose gels were recorded by ChemiDoc^TM MP Imaging System (Bio-Rad Laboratories, CA, USA) and band intensities quantified in the linear range using Image Lab 5.2.1 to 6.0.1 (Bio-Rad Laboratories).

Chromatin immunoprecipitation (ChIP)

Chromatin immunoprecipitation was performed as detailed previously^6,30. All steps and buffers used until the washing step of the immunoprecipitate were done on ice or at 4 °C. First, HeLa cells were fixed with 1% formaldehyde in medium for 10 min followed by 5 min quenching with 0.12 M glycine in medium on a shaker. Thereafter, cells were washed two times with ice-cold PBS, scratched off with cell scraping solution (PBS with 5 µl/ml phenylmethylsulfonyl fluoride, PMSF, Sigma-Aldrich/Merck), centrifuged and resuspended in 1 ml cell scraping solution and centrifuged. For DNA and protein extraction, the cells were resuspended in 1 ml ice cold cell lysis buffer (5.0 mM PIPES, pH 8.0; 167 mM NaCl; 1.2 mM EDTA; 0.01% SDS; 1.1% Triton X-100) with 5 µl PIC and 5 µl PMSF. After 10 min of incubation, the cells were dounced, centrifuged and broken cells dissolved in 1 ml ice cold nuclei lysis buffer (50 mM Tris-HCl, pH 8.0; 10 mM EDTA; 1.0% SDS) with 5 µl PIC and 5 µl PMSF and transferred into a polystyrol tube. After 10 min of incubation, the cells were sonified 5×15 s with 1 min breaks. Thereafter, 50 µl were taken for the input sample, and the remaining suspension frozen in liquid nitrogen and stored at − 80 °C. The input sample was mixed with 150 µl diethyl pyrocarbonate (DEPC) water and 10 µl 5 M NaCl and incubated overnight at 65 °C, then incubated with 0.5 µl RNase A for 15 min at 37 °C and with 5 µl proteinase K (Sigma-Aldrich/Merck) for 90 min at 42 °C. DNA was obtained by Phenol:Chloroform:Isoamyl alcohol phase separation and ethanol precipitation and dissolved in 30 µl TE (10 mM Tris-HCl, pH 8.0; 1 mM EDTA).

Chromatin samples were thawed on ice and centrifuged (15 min, 13800 x g). Equal amounts of DNA were used for each immunoprecipitation (5-20 µg). Sample volumes were adjusted with dilution buffer (16.7 mM Tris-HCl, pH 8.0; 167 mM NaCl; 1.2 mM EDTA; 0.01% SDS; 1.1% Triton X-100), and 1.5 µl 100 mM PMSF, 1.5 µl PIC and 50 µl protein G-Agarose Salmon Sperm DNA added (protein G-Agarose washed twice with TE, 1 g/ml BSA, 0.1% sodium azide, 0.4 mg/ml Salmon Sperm DNA Beads). Precleaning was done for 2 h on a rotator, antibody added, rotated overnight, 50 µl protein G-Agarose-Salmon Sperm DNA added and rotated for 2 h. Beads were centrifuged, washed by rotation with washing buffer for 10 min, then three times with high-salt wash buffer (50 mM HEPES, pH 7.9; 500 mM NaCl; 1.0 mM EDTA; 0.1% SDS; 1.0% Triton X-100; 0.1% deoxycholic acid sodium) and two times with TE. Beads were resuspended in 300 µl elution buffer (50 mM Tris-HCl, pH 8.0; 10 mM EDTA; 1.0% SDS) and incubated with 20 µl proteinase K (Sigma-Aldrich/Merck) for 2 h at 55 °C and overnight at 65 °C. Beads were removed by centrifugation (5 min, 16000 x g) and the supernatant subjected to Phenol:Chloroform:Isoamyl alcohol separation. After ethanol precipitation, the DNA was dissolved in 30 µl TE for PCR. For immunoprecipitation, the following antibodies were used: anti-γH2AX Ser139 (Mouse, monoclonal, Clone JBW 301, 05-636, Merck Millipore), anti-EndoG (Mouse monoclonal, B-2, sc-365359, Santa Cruz Biotechnology, Inc., CA, USA) and IgG (Mouse, sc-2025, Santa Cruz Biotechnology Inc.). Details on immunoprecipitation and Surface plasmon resonance are provided in the Supplementary Methods.

Immunofluorescence microscopy and Proximity-Ligation-Assay (PLA)

Immunofluorescence microscopy was performed as before^6,30, engaging a BZ-9000 Keyence analyzer and a Nikon 100 × 1.45 oil objective (Keyence, Neu-Isenburg, Germany). The BZ-II Analyzer software was used for quantitative analysis of microscopic images, which were processed by use of ImageJ. Exposure times as well as intensity and minimal focus size thresholds were maintained throughout the whole experimental set.

HeLa cells were cultivated on cover slips in 24-well plates for immunofluorescence microscopy and on Falcon® CultureSlides (Corning Inc., Corning, New York, USA) for PLA, washed three times with PBS for 5 min each, followed by a 1 min pre-extraction step with 0.2% Triton (Sigma-Aldrich/Merck). They were then washed again with PBS for 5 min and fixed with 3.7% formaldehyde (VWR Chemicals BDH) for 10 min.

For immunofluorescence microscopy, the cells were then PBS-washed three times, permeabilized for 12 min with 0.5% 100x Triton/PBS and PBS-washed three times. Blocking was performed with 5% goat serum/PBS for 1 h at RT, followed by a PBS-wash and the primary antibody, diluted in 5% goat serum/PBS, added for 1 h at 37 °C in a humidity chamber. The secondary antibody, diluted in 5% goat serum/PBS, was added after three PBS-washes for 1 h at 37 °C in a humidity chamber in the dark. After three PBS-washes, the nucleus was stained and slides mounted with VectaShield Mounting Medium for Fluorescence containing DAPI (Vector Laboratories, Burlingame, California, USA). The following primary antibodies were used: anti-γH2AX Ser139 (mouse, monoclonal, clone JBW 301, 05-636, Merck Millipore, 1/1000), anti-EndoG (mouse monoclonal, B-2, sc-365359, Santa Cruz Biotechnology Inc., 1/1000) and anti-Ku80 (rabbit monoclonal, S.669.4, Invitrogen/Thermo Fisher Scientific, Waltham, Massachusetts, USA, 1/400). The following primary antibodies were used: goat-anti-mouse Alexa 555 (A21424) or Alexa 488 (A32723), goat-anti-rabbit Alexa 555 (A21428) or Alexa 488 (A11008) from (Invitrogen/Thermo Fisher Scientific, 1/1000).

PLA staining was performed according to the manufacturer’s instructions using Duolink® In Situ Detection Reagent Green, Duolink® In Situ PLA Probe Anti-Rabbit PLUS, Duolink® In Situ PLA Probe Anti-Mouse MINUS, and Duolink® In Situ Mounting Medium with DAPI (Sigma-Aldrich/Merck). The primary antibodies (1/300) used were directed against Ku80 (rabbit monoclonal, S.669.4, Invitrogen/Thermo Fisher Scientific), Ku70 (mouse monoclonal, S5C11, sc-56130, Santa Cruz Biotechnology Inc.) and EndoG (mouse monoclonal, B-2, sc-365359, Santa Cruz Biotechnology Inc.). For positive controls, Ku80 and Ku70 were included. Negative controls were performed by including Ku80 or EndoG alone or by omitting antibodies entirely. Images were captured using a Zeiss fluorescence microscope Axio Observer 3/5/7 KMAT (Carl Zeiss AG, Oberkochen, Germany) equipped with appropriate filters for Duolink® In Situ Detection Reagent Green (FITC) and DAPI signals. Image analysis was performed using ZEN Blue software 3.1 (Carl Zeiss AG) with consistent settings maintained across all samples to ensure reproducibility. PLA imaging in Supplementary Fig. 9d was performed using Duolink In Situ Orange Starter Kit Mouse/Rabbit (DUO92102; Sigma) on a BZ-9000 fluorescent microscope and image analysis with BZ-II Analyzer 2.1 software (Keyence, Osaka, Japan).

Peptide synthesis and nucleofection

The peptides were synthesized and purified by the Core Facility Functional Peptidomics at Ulm University, Ulm Peptide Pharmaceuticals (U-PEP). The peptides were synthesized via solid-phase peptide synthesis (SPPS) as before³⁰. Peptides were purified via HPLC-1100 and characterized by mass spectrometry. The lyophilized peptides were stored at 4 °C and after solvation in Aqua ad iniectabilia (B.Braun, Melsungen, Deutschland) at − 20 °C. The peptides are represented by the following aa sequences (see also Table 1): Ku1: ITKEEASGSSVTAEEAKKFLAG; Ku2: RVLVKQKKASFEEASNQLINHIEQFL; Ku3: EIVVQDGITLITKEEASGSSVTAEEAKK; Ku4: ASFEEASNQLINHIEQFLDTNETPYFMKS; Scrambled: VDSTTIEKSIAEVAEEQESILKKVAGGT; Mut1: EIVVQDGITLITKAEASGSSVTAEEAKK; Mut2: EIVVQDGITLITKEEASGSSVTAEEAAK.

For transfection of cells via Amaxa^® nucleofector, called nucleofection, the cells were washed with PBS, re-suspended (1 × 10⁶ for K562 and 5 × 10⁵ for HeLa) in 100 µl Amaxa Cell Line Nucleofector^TM Kit V (Lonza, Basel, Switzerland), mixed with 10 µl H₂O as well as peptide and transferred to the cuvettes. The Nucleofector^TM II programs used were X-01 for K562 and I-13 for HeLa cells. After nucleofection K562 cells were transferred into 2 ml medium, HeLa cells distributed on three 6-well plates.

Statistical information

Statistical analyses were performed to evaluate the effects of Ku80-Ct and derived peptides under multiple conditions compared to control (ctrl) groups. Unpaired, nonparametric data were analyzed using the two-tailed Mann-Whitney U test following the Kruskal-Wallis H test. Paired, nonparametric data were analyzed using the Wilcoxon signed-rank test following the Friedman test. Statistical analyses were conducted using GraphPad Prism 8-10 software (GraphPad Software, San Diego, CA, USA). P-values < 0.1 are provided, whereby P-values < 0.0001 are indicated in the figures by use of asterisks. Data are presented as mean + SD/SEM.

Single-molecule measurements (SMT)

All methods for SMT, including cell culture, stable cell line generation, cell transduction, transfection, nucleofection and Western blotting, as well as SMT measurements, can be found in the Supplementary Methods.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All data of this work will be made available upon request without restrictions. Material requests should be addressed to the corresponding author, Lisa Wiesmüller, for questions related to the biomedical part to the corresponding author J. Christof M. Gebhardt, for questions related to SMT and to the modeling part to the corresponding author Elsa Sanchez-Garcia. Source data are provided in this paper.

Code availability

SMT experiments were analyzed using the TrackIt⁴⁰ software code for single-molecule tracking analysis available at https://gitlab.com/GebhardtLab/TrackIt⁸⁵.

References

Gole, B. & Wiesmüller, L. Leukemogenic rearrangements at the mixed lineage leukemia gene (MLL)-multiple rather than a single mechanism. Front. Cell Dev. Biol. 3, 149178 (2015).
Article Google Scholar
Hawkins, C. J. & Miles, M. A. Mutagenic consequences of sublethal cell death signaling. Int. J. Mol. Sci. 22, 6144 (2021).
Article CAS PubMed PubMed Central Google Scholar
Meyer, C. et al. The KMT2A recombinome of acute leukemias in 2023. Leukemia 37, 988–1005 (2023).
Article CAS PubMed PubMed Central Google Scholar
Bacolla, A., Tainer, J. A., Vasquez, K. M. & Cooper, D. N. Translocation and deletion breakpoints in cancer genomes are associated with potential non-B DNA-forming sequences. Nucleic Acids Res. 44, 5673–5688 (2016).
Article CAS PubMed PubMed Central Google Scholar
Canela, A. et al. Genome organization drives chromosome fragility. Cell 170, 507–521 (2017).
Article CAS PubMed PubMed Central Google Scholar
Gole, B., Mian, E., Rall, M. & Wiesmüller, L. Base excision repair proteins couple activation-induced cytidine deaminase and endonuclease G during replication stress-induced MLL destabilization. Leukemia 32, 159–167 (2018).
Article CAS PubMed Google Scholar
Gole, B., Baumann, C., Mian, E., Ireno, C. I. & Wiesmüller, L. Endonuclease G initiates DNA rearrangements at the MLL breakpoint cluster upon replication stress. Oncogene 34, 3391–3401 (2015).
Article CAS PubMed Google Scholar
Liu, X. et al. Redefining the roles of apoptotic factors in carcinogenesis.Mol. Cell. Oncol. 3, https://doi.org/10.1080/23723556.2015.1054550 (2016).
Galluzzi, L. et al. No death without life: vital functions of apoptotic effectors. Cell Death Differ. 15, 1113–1123 (2008).
Article CAS PubMed PubMed Central Google Scholar
Ichim, G. et al. Limited mitochondrial permeabilization causes DNA damage and genomic instability in the absence of cell death. Mol. Cell 57, 860–872 (2015).
Article CAS PubMed PubMed Central Google Scholar
Lin, J. L. J., Wu, C. C., Yang, W. Z. & Yuan, H. S. Crystal structure of endonuclease G in complex with DNA reveals how it nonspecifically degrades DNA as a homodimer. Nucleic Acids Res. 44, 10480–10490 (2016).
CAS PubMed PubMed Central Google Scholar
Loll, B., Gebhardt, M., Wahle, E. & Meinhart, A. Crystal structure of the EndoG/EndoGI complex: mechanism of EndoG inhibition. Nucleic Acids Res. 37, 7312–7320 (2009).
Article CAS PubMed PubMed Central Google Scholar
Zhang, Z. et al. Solution structure of the C-terminal domain of Ku80 suggests important sites for protein-protein interactions. Structure 12, 495–502 (2004).
Article CAS PubMed Google Scholar
Blier, P. R., Griffith, A. J., Craft, J. & Hardin, J. A. Binding of Ku protein to DNA. Measurement of affinity for ends and demonstration of binding to nicks. J. Biol. Chem. 268, 7594–7601 (1993).
Article CAS PubMed Google Scholar
Walker, J. R., Corpina, R. A. & Goldberg, J. Structure of the Ku heterodimer bound to DNA and its implications for double-strand break repair. Nature 412, 607–614 (2001).
Article ADS CAS PubMed Google Scholar
Wang, C. & Lees-Miller, S. P. Detection and repair of ionizing radiation-induced dna double strand breaks: New developments in nonhomologous end joining. Int. J. Radiat. Oncol. Biol. Phys. 86, 440–449 (2013).
Article CAS PubMed PubMed Central Google Scholar
Postow, L. Destroying the ring: Freeing DNA from Ku with ubiquitin. FEBS Lett. 585, 2876–2882 (2011).
Article CAS PubMed PubMed Central Google Scholar
Grundy, G. J., Moulding, H. A., Caldecott, K. W. & Rulten, S. L. One ring to bring them all–the role of Ku in mammalian non-homologous end joining. DNA Repair 17, 30–38 (2014).
Article CAS PubMed Google Scholar
Hegde, M. L. et al. Scaffold attachment factor A (SAF-A) and Ku temporally regulate repair of radiation-induced clustered genome lesions. Oncotarget 7, 54430–54444 (2016).
Article PubMed PubMed Central Google Scholar
Chung, U. et al. The interaction between Ku antigen and REF1 protein mediates negative gene regulation by extracellular calcium. J. Biol. Chem. 271, 8593–8598 (1996).
Article CAS PubMed Google Scholar
Shao, Z. et al. DNA-PKcs has KU-dependent function in rRNA processing and haematopoiesis. Nature 579, 291–296 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Wang, Y. et al. Early developing B cells undergo negative selection by central nervous system-specific antigens in the meninges. Immunity 54, 2784–2794 (2021).
Article CAS PubMed Google Scholar
Dibitetto, D. et al. DNA-PKcs promotes fork reversal and chemoresistance. Mol. Cell 82, 3932–3942 (2022).
Article CAS PubMed PubMed Central Google Scholar
Bétermier, M., Bertrand, P. & Lopez, B. S. Is non-homologous end-joining really an inherently error-prone process? PLOS Genet. 10, e1004086 (2014).
Article PubMed PubMed Central Google Scholar
Teixeira-Silva, A. et al. The end-joining factor Ku acts in the end-resection of double strand break-free arrested replication forks. Nat. Commun. 8, 1–14 (2017).
Article CAS Google Scholar
Cappetta, D. et al. Oxidative Stress and Cellular Response to Doxorubicin: A Common Factor in the Complex Milieu of Anthracycline Cardiotoxicity. Oxid. Med. Cell. Longev. 2017, 1521020 (2017).
Article PubMed PubMed Central Google Scholar
Huelsenbeck, S. C. et al. Rac1 protein signaling is required for DNA damage response stimulated by topoisomerase II poisons. J. Biol. Chem. 287, 38590–38599 (2012).
Article CAS PubMed PubMed Central Google Scholar
Scaglioni, L., Mondelli, R., Artali, R., Sirtori, F. R. & Mazzini, S. Nemorubicin and doxorubicin bind the G-quadruplex sequences of the human telomeres and of the c-MYC promoter element Pu22. Biochim. Biophys. Acta 1860, 1129–1138 (2016).
Article CAS PubMed Google Scholar
Ireno, I. C., Baumann, C., Stöber, R., Hengstler, J. G. & Wiesmüller, L. Fluorescence-based recombination assay for sensitive and specific detection of genotoxic carcinogens in human cells. Arch. Toxicol. 88, 1141–1159 (2014).
Article CAS PubMed Google Scholar
Eberle, J. et al. A Fibrinogen Alpha Fragment Mitigates Chemotherapy-Induced MLL Rearrangements. Front. Oncol. 11, 689063 (2021).
Article CAS PubMed PubMed Central Google Scholar
Osipovich, O., Durum, S. K. & Muegge, K. Defining the minimal domain of Ku80 for interaction with Ku70. J. Biol. Chem. 272, 27259–27265 (1997).
Article CAS PubMed Google Scholar
Osipovich, O., Duhe, R. J., Hasty, P., Durum, S. K. & Muegge, K. Defining functional domains of Ku80: DNA end binding and survival after radiation. Biochem. Biophys. Res. Commun. 261, 802–807 (1999).
Article ADS CAS PubMed Google Scholar
Falck, J., Coates, J. & Jackson, S. P. Conserved modes of recruitment of ATM, ATR and DNA-PKcs to sites of DNA damage. Nature 434, 605–611 (2005).
Article ADS CAS PubMed Google Scholar
Akyüz, N. et al. DNA substrate dependence of p53-mediated regulation of double-strand break repair. Mol. Cell. Biol. 22, 6306–6317 (2002).
Article PubMed PubMed Central Google Scholar
Yang, F., Teves, S. S., Kemp, C. J. & Henikoff, S. Doxorubicin, DNA torsion, and chromatin dynamics. Biochim. Biophys. Acta Rev. Cancer 1845, 84–89 (2014).
Article CAS Google Scholar
Harris, R. et al. The 3D solution structure of the C-terminal region of Ku86 (Ku86CTR). J. Mol. Biol. 335, 573–582 (2004).
Article CAS PubMed Google Scholar
Romero-Molina, S., Ruiz-Blanco, Y. B., Harms, M., Münch, J. & Sanchez-Garcia, E. PPI-Detect: A support vector machine model for sequence-based prediction of protein–protein interactions. J. Comput. Chem. 40, 1233–1242 (2019).
Article ADS CAS PubMed Google Scholar
Kurcinski, M. et al. CABS-dock standalone: a toolbox for flexible protein–peptide docking. Bioinformatics 35, 4170–4172 (2019).
Article CAS PubMed PubMed Central Google Scholar
Smialowski, P., Doose, G., Torkler, P., Kaufmann, S. & Frishman, D. PROSO II - A new method for protein solubility prediction. FEBS J. 279, 2192–2200 (2012).
Article CAS PubMed Google Scholar
Kuhn, T., Hettich, J., Davtyan, R. & Gebhardt, J. C. M. Single molecule tracking and analysis framework including theory-predicted parameter settings. Sci. Rep. 11, 1–12 (2021).
Article Google Scholar
Huynh, D. et al. Effective in vivo binding energy landscape illustrates kinetic stability of RBPJ-DNA binding. Nat. Commun. 16, 1–15 (2025).
Article ADS Google Scholar
Reisser, M. et al. Inferring quantity and qualities of superimposed reaction rates from single molecule survival time distributions. Sci. Rep. 10, 1–13 (2020).
Article ADS Google Scholar
Ghosh, M., Meiss, G., Pingoud, A. M., London, R. E. & Pedersen, L. C. The nuclease A-inhibitor complex is characterized by a novel metal ion bridge. J. Biol. Chem. 282, 5682–5690 (2007).
Article CAS PubMed Google Scholar
Temme, C. et al. The Drosophila melanogaster gene cg4930 encodes a high affinity inhibitor for endonuclease G. J. Biol. Chem. 284, 8337–8348 (2009).
Article CAS PubMed PubMed Central Google Scholar
Fell, V. L. & Schild-Poulter, C. The Ku heterodimer: function in DNA repair and beyond. Mutat. Res. Rev. Mutat. Res. 763, 15–29 (2015).
Article CAS PubMed Google Scholar
Radhakrishnan, S. K. & Lees-Miller, S. P. DNA requirements for interaction of the C-terminal region of Ku80 with the DNA-dependent protein kinase catalytic subunit (DNA-PKcs). DNA Repair 57, 17–28 (2017).
Article CAS PubMed PubMed Central Google Scholar
Gell, D. & Jackson, S. P. Mapping of protein-protein interactions within the DNA-dependent protein kinase complex. Nucleic Acids Res. 27, 3494–3502 (1999).
Article CAS PubMed PubMed Central Google Scholar
Sibanda, B. L., Chirgadze, D. Y. & Blundell, T. L. Crystal structure of DNA-PKcs reveals a large open-ring cradle comprised of HEAT repeats. Nature 463, 118–121 (2010).
Article ADS CAS PubMed Google Scholar
Weterings, E. et al. The Ku80 carboxy terminus stimulates joining and artemis-mediated processing of DNA ends. Mol. Cell. Biol. 29, 1134–1142 (2009).
Article CAS PubMed Google Scholar
Kalinowska, M., Garncarz, W., Pietrowska, M., Garrard, W. T. & Widlak, P. Regulation of the human apoptotic DNase/RNase endonuclease G: involvement of Hsp70 and ATP. Apoptosis 10, 821–830 (2005).
Article CAS PubMed Google Scholar
Abbasi, S. & Schild-Poulter, C. Mapping the Ku Interactome Using Proximity-Dependent Biotin Identification in Human Cells. J. Proteome Res. 18, 1064–1077 (2019).
Article CAS PubMed Google Scholar
Chaudhuri, J. & Alt, F. W. Class-switch recombination: interplay of transcription, DNA deamination and DNA repair. Nat. Rev. Immunol. 4, 541–552 (2004).
Article CAS PubMed Google Scholar
Zan, H. et al. Endonuclease G plays a role in immunoglobulin class switch DNA recombination by introducing double-strand breaks in switch regions. Mol. Immunol. 48, 610–622 (2011).
Article CAS PubMed Google Scholar
Zhang, T., Zhang, H., Wang, Y. & McGown, L. B. Capture and identification of proteins that bind to a GGA-rich sequence from the ERBB2 gene promoter region. Anal. Bioanal. Chem. 404, 1867–1876 (2012).
Article CAS PubMed Google Scholar
Zanden, C. M. V., Czarny, R. S., Ho, E. N., Robertson, A. B. & Ho, P. S. Structural adaptation of vertebrate endonuclease G for 5-hydroxymethylcytosine recognition and function. Nucleic Acids Res. 48, 3962–3974 (2020).
Article Google Scholar
Dahal, S. et al. Unleashing a novel function of Endonuclease G in mitochondrial genome instability. Elife 11, https://doi.org/10.7554/eLife.69916 (2022).
Bowater, R. & Doherty, A. J. Making ends meet: repairing breaks in bacterial DNA by non-homologous end-joining. PLoS Genet. 2, 93–99 (2006).
Article CAS Google Scholar
Frit, P., Barboule, N., Yuan, Y., Gomez, D. & Calsou, P. Alternative end-joining pathway(s): bricolage at DNA breaks. DNA Repair 17, 81–97 (2014).
Article CAS PubMed Google Scholar
Bossaert, M. et al. Identification of the main barriers to Ku accumulation in chromatin. Cell Rep. 43, https://doi.org/10.1016/j.celrep.2024.114538 (2024).
Larsen, B. D. et al. Cancer cells use self-inflicted DNA breaks to evade growth limits imposed by genotoxic stress. Science 376, 476–483 (2022).
Article ADS CAS PubMed Google Scholar
Benada, J., Alsowaida, D., Megeney, L. A. & Sørensen, C. S. Self-inflicted DNA breaks in cell differentiation and cancer. Trends Cell Biol. 33, 850–859 (2023).
Article CAS PubMed Google Scholar
Coffey, G. & Campbell, C. An alternate form of Ku80 is required for DNA end-binding activity in mammalian mitochondria. Nucleic Acids Res. 28, 3793–3800 (2000).
Article CAS PubMed PubMed Central Google Scholar
Ohsato, T. et al. Mammalian mitochondrial endonuclease G. Digestion of R-loops and localization in intermembrane space. Eur. J. Biochem. 269, 5765–5770 (2002).
Article CAS PubMed Google Scholar
Wiehe, R. S. et al. Endonuclease G promotes mitochondrial genome cleavage and replication. Oncotarget 9, 18309–18326 (2018).
Article PubMed PubMed Central Google Scholar
Schreier, H. K., Wiehe, R. S., Ricchetti, M. & Wiesmüller, L. Polymerase ζ Is Involved in mitochondrial DNA maintenance processes in concert with APE1 activity. Genes 13, https://doi.org/10.3390/genes13050879 (2022).
Jang, D. S. et al. Novel cytoprotective inhibitors for apoptotic endonuclease G. DNA Cell Biol. 34, 92–100 (2015).
Article CAS PubMed Google Scholar
Nhàn, N. T. T., Yamada, T. & Yamada, K. H. Peptide-Based Agents for Cancer Treatment: Current Applications and Future Directions. Int. J. Mol. Sci. 24, (2023).
Schwede, T., Kopp, J., Guex, N. & Peitsch, M. C. SWISS-MODEL: an automated protein homology-modeling server. Nucleic Acids Res. 31, 3381–3385 (2003).
Article CAS PubMed PubMed Central Google Scholar
Benkert, P., Biasini, M. & Schwede, T. Toward the estimation of the absolute quality of individual protein structure models. Bioinformatics 27, 343–350 (2011).
Article CAS PubMed Google Scholar
Katoh, K. & Standley, D. M. MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability. Mol. Biol. Evol. 30, 772–780 (2013).
Article CAS PubMed PubMed Central Google Scholar
Gouveia-Oliveira, R., Sackett, P. W. & Pedersen, A. G. MaxAlign: Maximizing usable data in an alignment. BMC Bioinform. 8, 1–8 (2007).
Article Google Scholar
Rice, P., Longden, L. & Bleasby, A. EMBOSS: The European molecular biology open software suite. Trends Genet. 16, 276–277 (2000).
Article CAS PubMed Google Scholar
Crooks, G. E., Hon, G., Chandonia, J. M. & Brenner, S. E. WebLogo: a sequence logo generator. Genome Res. 14, 1188–1190 (2004).
Article CAS PubMed PubMed Central Google Scholar
Schneider, T. D. & Stephens, R. M. Sequence logos: a new way to display consensus sequences. Nucleic Acids Res. 18, 6097–6100 (1990).
Article CAS PubMed PubMed Central Google Scholar
Jo, S., Kim, T., Iyer, V. G. & Im, W. CHARMM-GUI: A web-based graphical user interface for CHARMM. J. Comput. Chem. 29, 1859–1865 (2008).
Article ADS CAS PubMed Google Scholar
Abraham, M. J. et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1–2, 19–25 (2015).
Article ADS Google Scholar
Huang, J. et al. CHARMM36m: an improved force field for folded and intrinsically disordered proteins. Nat. Methods 14, 71–73 (2016).
Article PubMed PubMed Central Google Scholar
Mark, P. & Nilsson, L. Structure and dynamics of the TIP3P, SPC, and SPC/E water models at 298. K. J. Phys. Chem. A 105, 9954–9960 (2001).
Article CAS Google Scholar
Darden, T., York, D. & Pedersen, L. Particle mesh Ewald: An N⋅log(N) method for Ewald sums in large systems. J. Chem. Phys. 98, 10089–10092 (1993).
Article ADS CAS Google Scholar
Nosé, S. A molecular dynamics method for simulations in the canonical ensemble. Mol. Phys. 52, 255–268 (1984).
Article ADS Google Scholar
Parrinello, M. & Rahman, A. Polymorphic transitions in single crystals: A new molecular dynamics method. J. Appl. Phys. 52, 7182–7190 (1981).
Article ADS CAS Google Scholar
Humphrey, W., Dalke, A. & Schulten, K. V. M. D. Visual molecular dynamics. J. Mol. Graph. 14, 33–38 (1996).
Article CAS PubMed Google Scholar
Mercadante, D., Gräter, F. & Daday, C. CONAN: A tool to decode dynamical information from molecular interaction maps. Biophys. J. 114, 1267–1273 (2018).
Article ADS CAS PubMed PubMed Central Google Scholar
Bennardo, N., Cheng, A., Huang, N. & Stark, J. M. Alternative-NHEJ is a mechanistically distinct pathway of mammalian chromosome break repair. PLoS Genet. 4, e1000110 (2008).
Article PubMed PubMed Central Google Scholar
Takeda, A. A. S., De Barros, A. C., Chang, C. W., Kobe, B. & Fontes, M. R. M. Structural basis of importin-α-mediated nuclear transport for Ku70 and Ku80. J. Mol. Biol. 412, 226–234 (2011).
Article CAS PubMed Google Scholar
Kuhn,T. “TrackIt_v1.5”, Scientific Reports. Zenodo. https://doi.org/10.5281/zenodo.7092296 (2022).

Download references

Acknowledgements

We are grateful to Claus Storgaard Sørensen, Biotech Research and Innovation Center, University of Copenhagen, Denmark, Aswin Mangerich, Nutritional Toxicology, Institute of Nutritional Science, University of Potsdam, Germany, as well as Frank Kirchhoff and Jan Münch, Institute of Molecular Virology, Ulm University, Germany, for enlightening discussions. We would like to express our sincere gratitude to Dierk Niessing, Institute of Pharmaceutical Biotechnology, Ulm University, Germany, for generously providing access to his laboratory equipment and for his valuable consultations in the field of SPR analysis. We thank Jeremy M. Stark, Duarte Cancer Center, CA, for sharing the EJ5SceGFP construct for NHEJ measurements. We thank Ali Omar for his contributions to cloning the monomeric EndoG. SiR HaloTag ligand was kindly provided by Kai Johnsson, Max-Planck-Institute for Medical Research, Heidelberg. We thank the Core Facility FACS of Ulm University for their help with cell sorting, with special thanks to Dr. Simona Ursu, Dr. Sarah Warth, and Daniela Froelich. This project was financially supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation): CRC1279 (no. 316249678), project B05 to L.W. plus J.C.M.G. and project A06 to E.S.-G., CRC1506 (no. 450627322), project B03 to L.W. plus M. R.-S. and project C05 to J.C.M.G., CRC1430 (no. 424228829) project B04 to E.S.-G., Research Grant no. 468578170 to J.C.M.G. as well as under Germany’s Excellence Strategy - EXC 2033 – 390677874 – RESOLV to E.S.-G. J.E. was and A.S. is a member of the International Graduate School of Molecular Medicine of the Medical Faculty of Ulm University. E.S.-G was also supported by the DFG.

Funding

Open Access funding enabled and organized by Projekt DEAL.

Author information

Mara Hofmann & J. Christof M. Gebhardt
Present address: Department of Physics, Institute of Experimental Physics, Ulm University, Ulm, Germany
Boris Gole
Present address: Centre for human molecular genetics and pharmacogenomics, Faculty of Medicine, University of Maribor, SI, Maribor, Slovenia
Jessica Angulo-Capel
Present address: Institut d′Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), 149-153, Barcelona, Spain
These authors contributed equally: Julia Eberle, Ahmed Salem, Mara Hofmann, Anja Reisser, Yasser B. Ruiz-Blanco.

Authors and Affiliations

Department of Obstetrics and Gynecology, Ulm University, Ulm, Germany
Julia Eberle, Ahmed Salem, Boris Gole, Melanie Rall-Scharpf & Lisa Wiesmüller
Department of Physics, Institute of Biophysics, Ulm University, Ulm, Germany
Mara Hofmann, Anja Reisser, Jessica Angulo-Capel & J. Christof M. Gebhardt
Department of Biology, Computational Biochemistry, University of Duisburg-Essen, Essen, Germany
Yasser B. Ruiz-Blanco & Elsa Sanchez-Garcia
Department of Biochemical and Chemical Engineering, Chair of Computational Bioengineering, Technical University Dortmund, Dortmund, Germany
Yasser Almeida-Hernandez & Elsa Sanchez-Garcia
Institute of Pharmaceutical Biotechnology, Ulm University, James-Franck-Ring N27, Ulm, Germany
Thomas Monecke

Authors

Julia Eberle
View author publications
Search author on:PubMed Google Scholar
Ahmed Salem
View author publications
Search author on:PubMed Google Scholar
Mara Hofmann
View author publications
Search author on:PubMed Google Scholar
Anja Reisser
View author publications
Search author on:PubMed Google Scholar
Yasser B. Ruiz-Blanco
View author publications
Search author on:PubMed Google Scholar
Yasser Almeida-Hernandez
View author publications
Search author on:PubMed Google Scholar
Boris Gole
View author publications
Search author on:PubMed Google Scholar
Melanie Rall-Scharpf
View author publications
Search author on:PubMed Google Scholar
Jessica Angulo-Capel
View author publications
Search author on:PubMed Google Scholar
Thomas Monecke
View author publications
Search author on:PubMed Google Scholar
Elsa Sanchez-Garcia
View author publications
Search author on:PubMed Google Scholar
J. Christof M. Gebhardt
View author publications
Search author on:PubMed Google Scholar
Lisa Wiesmüller
View author publications
Search author on:PubMed Google Scholar

Contributions

L.W., B.G., and J.C.M.G. conceived the study. J.E., A.S., B.G., M.H., and A.R. performed experiments and processed all experimental data. Y.B.R.B. and E.S.G. discovered Ku3 via computational studies. Y.B.R.B. and Y.A.H performed the in silico modeling studies under the supervision of E.S.G; Y.B.R.B., Y.A.H., and E.S.G. designed the computational studies, analyzed and interpreted the results. Y.A.H. and E.S.G. wrote the computational part of the manuscript. J.A.-C. aided in SMT experiments. T.M. designed and T.M. and M.H. performed SPR analysis. L.W., B.G., J.C.M.G., J.E., A.S., M.H., and A.R. jointly designed the experiments, evaluated and interpreted experimental data. M.R.-S. supported experimental studies. L.W., J.C.M.G. and E.S.G. designed the manuscript flow and wrote the initial draft of the manuscript. All authors helped writing the manuscript.

Corresponding authors

Correspondence to Elsa Sanchez-Garcia, J. Christof M. Gebhardt or Lisa Wiesmüller.

Ethics declarations

Competing interests

The authors Lisa Wiesmüller, Boris Gole, Julia Eberle (née Wille), J. Christof M. Gebhardt, Anja Reisser (née Palmer), Yasser Bruno Ruiz-Blanco, Elsa Sanchez-Garcia are the inventors of the patent with application number PCT/EP2023/051816; applicant: Ulm University, Germany; status of application: Filed international on January 25, 2023; specific aspect of manuscript covered in patent application: Identification of peptide derived from human Ku80 for use in the prevention of genomic rearrangements causing therapy-related leukemia. The remaining authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Peter McHugh and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information (download PDF )

Description of Additional Supplementary Files (download PDF )

Supplementary Data 1 (download TXT )

Supplementary Data 2 (download TXT )

Supplementary Data 3 (download TXT )

Supplementary Data 4 (download TXT )

Supplementary Data 5 (download TXT )

Supplementary Data 6 (download TXT )

Supplementary Data 7 (download TXT )

Supplementary Data 8 (download TXT )

Supplementary Data 9 (download TXT )

Supplementary Data 10 (download TXT )

Supplementary Data 11 (download TXT )

Supplementary Data 12 (download TXT )

Reporting Summary (download PDF )

Transparent Peer Review file (download PDF )

Source data

Source Data (download ZIP )

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Eberle, J., Salem, A., Hofmann, M. et al. Discovery of an Endonuclease G-inhibitory Ku80-peptide protecting against leukemogenic rearrangements at the MLL breakpoint cluster. Nat Commun 17, 3562 (2026). https://doi.org/10.1038/s41467-026-72034-2

Download citation

Received: 25 March 2025
Accepted: 30 March 2026
Published: 17 April 2026
Version of record: 17 April 2026
DOI: https://doi.org/10.1038/s41467-026-72034-2

Discovery of an Endonuclease G-inhibitory Ku80-peptide protecting against leukemogenic rearrangements at the MLL breakpoint cluster