Interfering with DNA repair pathways to enhance CRISPR-Cas9-mediated homology-directed repair in a chelicerate genetic model Authors: Sander De Rouck, Wannes Dermauw, Thomas Van Leeuwen Published: 2026 ABSTRACT SummaryThe two-spotted spider mite, Tetranychus urticae, is a major pest and an emerging genetic model. Recent CRISPR-Cas9 advances, especially the SYNCAS method for maternal delivery of Cas9 ribonucleoproteins, have enabled precise genome editing in this and other difficult-to-transform arthropods. Yet SYNCAS-mediated knockins vary in efficiency, possibly due to competition between DNA repair pathways, whose mechanisms in T. urticae and other chelicerates remain unknown. Here, we provide the first functional analysis of double-strand break repair in a chelicerate. Loss of DNA polymerase theta (Polθ) redirects repair almost entirely toward homology-directed repair, whereas absence of Ligase IV has no detectable impact. Using a reporter assay targeting phytoene desaturase, we demonstrate that Polθ-deficient strains enhance incorporation of repair templates, even when mutations are distant from the cut site. Also, insertion of larger fragments is improved. Finally, disrupting Polθ imposes only a modest fitness cost, highlighting its value for future genome engineering in this species.Keywords: arthropoda, Acari, Chelicerata, CRISPR, SYNCAS, double-strand break repair, NHEJ, MMEJ, polymerase theta, ligase IV Introduction Double-strand breaks (DSBs) are among the most severe forms of DNA damage, capable of compromising genomic integrity and threatening cell viability if not accurately repaired. These breaks can arise endogenously during cellular processes, such as DNA replication and meiotic recombination, or exogenously in response to environmental factors like ionizing radiation, chemotherapeutic agents, and oxidative stress.1 In addition, DSBs can be introduced deliberately via gene editing tools such as CRISPR-Cas systems with the aim to alter the gene in a particular way.2,3 There are four main repair pathways that mammalian cells use to address DSB damage in DNA: non-homologous end joining (NHEJ), theta-mediated end joining (TMEJ), single-strand annealing (SSA), and homology-directed repair (HDR) which are depicted in Figure 1. The first, NHEJ, is an error-prone mechanism that directly rejoins the broken DNA ends with minimal end processing, often leading to small insertions or deletions.1 The initiation of NHEJ occurs with the binding of the ring-shaped Ku70/Ku80 protein heterodimer to the ends of the DSB. This binding serves a dual purpose by safeguarding the DNA ends from further resection and facilitating the recruitment of DNA-dependent protein kinase (DNA-PKcs).4,5 Subsequently, DNA-PKcs orchestrates the assembly of the LigIV-XRCC4-XLF complex (consisting of Ligase IV, X-ray cross complementing group 4, and XRCC4-like factor) to ligate the DNA ends.6,7 In cases where direct ligation of the DSB ends is not feasible, supplementary proteins, including Artemis nuclease, polynucleotide kinase 3′ phosphatase (PNKP), and DNA polymerases such as Polμ and Polλ, become imperative for end processing.8,9,10 Schematic overview of the four major pathways to repair DNA double-strand breaks in mammals: NHEJ, TMEJ, SSA, and HDR A major difference between these pathways lies in how much DNA end resection they require. Although the Artemis-DNA-PKcs complex can trim DNA ends slightly, NHEJ mainly protects ends through Ku70-Ku80 binding. DNA-PKcs recruits the LigIV-XRCC4-XLF complex to ligate DNA ends. When direct ligation is not possible, end processing requires additional factors such as Artemis and polymerases μ and λ. In contrast, CtIP and the MRN complex perform extensive resection, creating single-stranded DNA overhangs needed for TMEJ, SSA and HDR. In TMEJ, PARP1 recruits Polθ which aligns microhomologies and synthesizes DNA. Nicks are then sealed by ligase activity. Further resection by BLM and EXO1 allows RPA to bind ssDNA, supporting SSA and HDR. RAD51 promotes strand invasion by forming nucleoprotein filaments on single-stranded DNA, while RAD52 assists by mediating the annealing of complementary strands. Figure based on Chang et al. (2017).1 The second pathway, TMEJ (also known as alternative non-homologous end joining or Alt-NHEJ), is a distinct end-joining repair pathway that can introduce nucleotide changes during the repair process. TMEJ starts with short-range resection of the DSB.11 Similar to NHEJ, this pathway operates without the need for a template to facilitate the repair of damaged DNA.12 Instead, the realignment of DSB ends is achieved through the use of short microhomologous sequences (or microhomologies), typically spanning 2 to 6 base pairs13 (although lengths can vary between organisms12), proximal to the damaged DNA ends.14 Short-range end resection promotes TMEJ, while a more extensive resection tends to favor SSA and HDR.15 In most metazoans, the initiation of end resection is orchestrated by the damage-sensing MRN complex (MRE11-RAD50-NBS1) in conjunction with its interacting partner CtIP (C-terminal binding protein interacting protein).10,16,17,18 The replication protein A (RPA), a ssDNA-binding protein, assumes a regulatory role by inhibiting TMEJ through prevention of annealing. Conversely, poly ADP-ribose polymerase 1 (PARP1) acts as a facilitator and kinetic promoter of TMEJ by helping to tether DNA fragments and promote the annealing reaction,19,20 though it is not an essential requirement for the pathway.21 DNA polymerase theta (Polθ, also referred to as POLQ in H. sapiens or mus308 in Drosophila) in turn, has the capability to displace RPA from ssDNA, thereby facilitating TMEJ. In mammalian cells, Polθ stabilizes annealed overhangs and might fill resulting gaps through template-directed DNA synthesis. However, it has been hypothesized that Polθ has low processivity and fidelity and thus only initiates synthesis from annealed microhomologies, after which it is replaced by the more processive and accurate Polδ.22 Finally, remaining nicks are sealed by either DNA Ligase I or III.23 SSA and TMEJ share several aspects; they both rely on 3′ ssDNA overhangs to facilitate annealing.12,24 Contrary to TMEJ, SSA depends on extensive end resection by the 5′ to 3′ exonuclease activities of either EXO1 or Bloom helicase (BLM)-DNA2, producing prolonged single-stranded DNA overhangs exceeding 1,000 nucleotides (nt).25,26 RPA then coats the resulting ssDNA overhangs followed by the binding of RAD52 which mediates annealing of homologous sequences.27,28 It is not known how remaining gaps are filled and ligated. The outcome of SSA repair frequently manifests as extensive deletions flanked by homologous sequences. These deletions arise from the homologous pairing of distantly located loci, followed by the trimming of non-homologous 3′ flaps prior to the ligation step.24 Of particular note, helicase POLQ-like (HELQ, also referred to as HEL308 in H. sapiens or mus301/spindle-C in Drosophila) is a DNA helicase homologous to the N-terminal region of Polθ and is present in most eukaryotic organisms except plants. HELQ was recently shown to be essential for SSA and for forms of microhomology-mediated end joining (MMEJ) other than TMEJ.29,30,31 The remaining pathway, HDR, depends on a homologous DNA template for accurate repair and is required for precise genome modifications such as targeted knockin (KI) of foreign DNA or specific mutations, in contrast to the other pathways which often result in knockout (KO) mutations.10,32 Like SSA, HDR requires long-range end resection followed by RPA coating.33 Subsequently, Rad51 recombinase replaces RPA, forming long helical filaments on the ssDNA that inhibit Rad52-mediated SSA.34,35,36 Finally, ssDNA is paired with the homologous repair template and the 3′ invading end of the ssDNA strand is used to prime DNA synthesis, allowing to complete the DSB repair.1,10,37 While the core machinery of the DSB repair pathways is believed to be highly conserved across eukaryotes, functional studies in arthropods (particularly insects) reveal deviations from mammalian models both in pathway-component presence/absence and in the balance among repair outcomes. A comparative analysis among Aedes aegypti, Drosophila melanogaster, and Homo sapiens showed that core DSB repair factors such as the MRN complex, Ku70/Ku80, and RAD51 are retained in A. aegypti and D. melanogaster. However, several key regulators present in mammals such as BRCA1, TP53BP1, and PALB2, which normally promote HDR, are absent in these species.38 This could indicate that although the basic pathways for DSB repair are conserved in arthropods, the regulation of choice between NHEJ and HDR diverges significantly from that of mammals, potentially explaining the strong bias toward end joining, which is sometimes observed in genome-editing experiments in insects.39,40,41,42 One strategy to shift this balance is to inactivate or inhibit components from the end-joining pathways. For example, KO of NHEJ factors, including Ku70, Ku80, LigIV, XRCC4, and XLF in a Bombyx mori cell line, substantially increased HDR-mediated repair.42 Likewise, in Drosophila, loss of LigIV shifted DSB repair almost entirely toward HDR when donor DNA template was supplied.40 However, the role of LigIV in regulating repair outcomes in Drosophila appears to be contested as the effect on DSB repair seems to be highly variable in ligIV KO lines.43,44 Beyond factors involved in NHEJ, targeting proteins involved in TMEJ can also improve HDR, as observed in human cell lines.45,46,47 While substantial progress has been made in elucidating DNA repair pathways in various model systems (yeast, mammals, and insects), studies examining these pathways and their hierarchy in chelicerate species are still lacking. This work focuses on the DNA repair mechanisms in the two-spotted spider mite, Tetranychus urticae, a major agricultural pest notorious for its broad host range and rapid pesticide resistance development.48,49 The extreme polyphagy of T. urticae has been extensively characterized, with the species being capable of feeding on over 1000 different plant species.50 Host-transfer studies have shown that shifts to a different or chemically treated host elicit transcriptional reprogramming, particularly within detoxification and primary metabolism gene families, underscoring a high degree of transcriptional plasticity.51,52,53,54 In addition to being a considerable pest, T. urticae is also an emerging genetic model for Chelicerata55,56,57 due to its well-annotated genome,58 the ease of maintaining and crossing laboratory populations,55 the ability to generate near-isogenic lines through haplodiploid reproduction59,60 as well as its short generation time and large population sizes that facilitate population-genetic studies.51 Genetic mapping approaches such as bulked segregant analysis (BSA) have further advanced the utility of T. urticae as a model system.61 Another important factor contributing to the model status of T. urticae is the availability of reverse genetics tools such as RNAi62,63,64 and CRISPR-Cas9.65 This last tool only became routinely applicable thanks to the recent development of a method termed SYNCAS, which enables efficient delivery of Cas9 ribonucleoprotein complexes into eggs, relying on maternal injection of Cas9 RNPs combined with BAPC and saponins, which achieves high KO efficiencies in the offspring of injected females.66 This method also proved to be effective to knock out genes in other hard-to-transform arthropods such as Frankliniella occidentalis, Bemisia tabaci, Amblyseius swirskii, and Phytoseiulus persimilis.66,67 In addition, SYNCAS facilitated the generation of mutants in Orius strigicollis.68 Although the introduction of single-nucleotide polymorphisms (SNPs) has been successful in T. urticae using SYNCAS, obtained efficiencies appear to vary between studies.66,69,70 Here, we identified orthologs of the canonical DSB repair pathways in T. urticae and generated mutant spider mite lines in which one or multiple pathways were disrupted to study their effects on repair outcomes. We further investigated the potential and limitations of HDR-mediated KIs using SYNCAS, focusing on the influence of distance from the cleavage site and insert size. Our results indicate that end joining in T. urticae is primarily mediated via TMEJ, as mutants in this pathway repaired DSBs almost exclusively via HDR. Moreover, TMEJ mutants demonstrated the potential to improve gene-editing efficiencies, highlighting their utility for future genome engineering applications in this species. Results Identification of putative orthologs of human DSB repair proteins in T. urticae, and analysis of ligIV and polθ genes across Acari Reciprocal BLASTp and tBLASTn analyses were used to identify putative T. urticae orthologs of human and Drosophila proteins involved in DNA repair pathways. Similar to D. melanogaster and A. aegypti,38,71 many key elements from the human pathways could not be identified in this spider mite species using reciprocal BLAST (Table 1). Especially for the NHEJ pathway many proteins appear to be absent or could not be identified, including DNA-PKcs, Artemis, polymerases λ and μ, XLF, XRCC4, PAXX, APLF, and APTX. Also for the other pathways, several constituents are missing, of which the most remarkable are NBS1 and CtIP given their crucial role in end resection in humans (Figure 1).17,73,74,75 Based on experiments conducted in other species,40,42,46,76 ligIV and polθ appeared as interesting candidate genes to create KO lines impaired in the NHEJ and TMEJ pathways, respectively. To reinforce identification and establish true orthology, we next mined a selection of arthropod genomes for the presence of ligIV and polθ and performed a phylogenetic analysis. Table 1Overview of human and Drosophila proteins involved in double-stranded break repair, and their putative orthologs in T. urticaeHuman geneNHEJTMEJHDRSSAD. melanogasteraT. urticaebKu70X–––FBgn0011774tetur03g09560Ku80X–––FBgn0041627tetur08g05290DNA-PKcsX–––––ArtemisX–––––Polymerase λX–––––Polymerase μX–––––XLFX–––FBgn0029825–XRCC4X–––FBgn0069301–PAXXX–––––Ligase IVX–––FBgn0030506tetur26g00400APLFX–––FBgn0026737–PNKPX–––FBgn0037578tetur13g03440APTXX–––FBgn0038704–HELQ–––XFBgn0002899LOC107369727cMRE11–XXXFBgn0020270tetur09g02240RAD50–XXXFBgn0034728tetur08g00920NBS1–XXXFBgn0261530–CtIP–XXXFBgn0036991–PARP1–X––FBgn0010247tetur04g08160Polymerase θ–X––FBgn0002905LOC107369726cXRCC1–X––FBgn0026751tetur18g01930Ligase III–X––FBgn0286075–Ligase I–X––FBgn0262619tetur02g07210BRCA1––XX––BRCA2––XXFBgn0050169–Exo1––XXFBgn0015553tetur06g02100DNA2––XXFBgn0288690tetur06g00670BLM––XXFBgn0002906tetur13g00890WRNX–XXFBgn0038608–RPA (sub1)––XXFBgn0010173tetur02g11760RPA (sub2)––XXFBgn0288834tetur01g13250RPA (sub3)––XXFBgn0288833tetur13g01180Rad51––X–FBgn0003479tetur02g05590Rad54––X–FBgn0002989tetur04g07950Polymerase δ (sub1)––X–FBgn0263600tetur07g01510Polymerase δ (sub2)––X–FBgn0027903tetur09g04010Polymerase δ (sub3)––X–FBgn0283467–Polymerase δ (sub4)––X–––RAD52––XX––ERCC1–––XFBgn0028434tetur01g09360ERCC4/XPF–––XFBgn0002707tetur08g07640MSH2–––XFBgn0015546tetur32g02030SLX1–––XFBgn0037263–SLX4–––XFBgn0002909–Pathways in which a corresponing protein is involved are marked with an X. Data for D. melanogaster were recovered from previous literature.10,38,71,72 Horizontal lines indicate genes for which no candidate orthologs were found. In bold are the genes that were further subject of this study.ahttps://flybase.org/.bhttps://bioinformatics.psb.ugent.be/orcae/overview/Tetur or chttps://www.ncbi.nlm.nih.gov/gene/. Overview of human and Drosophila proteins involved in double-stranded break repair, and their putative orthologs in T. urticae Pathways in which a corresponing protein is involved are marked with an X. Data for D. melanogaster were recovered from previous literature.10,38,71,72 Horizontal lines indicate genes for which no candidate orthologs were found. In bold are the genes that were further subject of this study. https://flybase.org/. https://bioinformatics.psb.ugent.be/orcae/overview/Tetur or chttps://www.ncbi.nlm.nih.gov/gene/. The phylogenetic tree showed discordance with the known Chelicerata phylogeny77,78 as LigIV of ticks, a spider, and a scorpion clustered (albeit with low bootstrap support) more closely with those of vertebrates than with those of other Chelicerata (Acariformes and Parasitiformes) (Figure S1; File S1). Further, LigIV was not conserved within mites and ticks (Acari) and could only be identified in a few species of all Acariformes (Trombidiformes, Sarcoptiformes, and Eriophyoidea) examined while present in all Parasitiformes. More specifically, within the Acariformes, LigIV could only be identified in species of the Bryobia and Tetranychus genus, two other trombidiform species, and two sarcoptiform species. In contrast to LigIV, Polθ was conserved across Acari but, similar to LigIV, Polθ of Parasitiformes, a spider and a scorpion clustered more closely with those of vertebrates (Figure S1). Interestingly, a gene, hereafter referred to as helq, was located in close proximity to T. urticae polθ, with conserved synteny observed in the polθ region of T. urticae, T. truncatus, and P. citri (helq flanked by integrin α-PS2 and hay, and polθ located between hay and scalloped; Figure S2; File S1). helq encodes a protein that is highly similar to Polθ (65%–80% identity within Tetranychinae) but nearly half its size (950–1,000 amino acids compared with 1700–1800 amino acids within Tetranychinae) and lacks the C-terminal DNA polymerase domain of Polθ (File S1). This gene is therefore likely homologous to HELQ in H. sapiens or mus301/spindle-C in D. melanogaster.79,80 Unlike in Tetranychinae, however, this gene is not located near polθ in either D. melanogaster or H. sapiens.81,82 Lastly, an InterProScan search revealed that T. urticae LigIV contained the following domains: DNA ligase, ATP-dependent, N-terminal domain (IPR036599), DNA ligase 4, adenylation domain (IPR044125), nucleic acid-binding, OB-fold (IPR012340), and BRCA1 C-terminus (BRCT) domain superfamily (IPR036420) (Figure 2). Both Polθ and helq of T. urticae have two P-loop-containing nucleoside triphosphate hydrolase domains (IPR027417), a winged helix-like DNA-binding domain (IPR036388), and a POLQ-like, helical domain (IPR048960), while only Polθ had a DNA-directed DNA polymerase, family A, palm domain (IPR001098) at the C-terminal end (Figure 2). Schematic representation of the resulting LigIV and Polθ proteins in the generated KO mutant lines DBD: DNA binding domain, AD: adenylation domain, OBD: oligonucleotide/oligosaccharide binding (OB)-fold domain, BRCT: BRCA1 C terminus, HD: helicase domain, and PD: polymerase domain. Numbers under domains indicate their approximate range in the amino acid sequence based on InterProScan domain predictions (ebi.ac.uk/interpro). Numbers in bold indicate total length of the protein. CRISPR-Cas9 KO of ligIV and polθ To study the DNA repair pathways in T. urticae and ultimately obtain a mite strain in which HDR is predominant, various mutant mite lines were generated using CRISPR-Cas9 in which NHEJ, TMEJ, or both pathways were disrupted. This was done via the SYNCAS method where virgin adult females were injected with CRISPR-Cas9 mix resulting in potential G0 male mutants. For ligIV, 6 out of 23 G0 males sent for sequencing contained mutations while 8 out of 88 and 3 out of 29 males had editing events in polθ in the helicase and polymerase domain respectively. Events displaying frameshift mutations were picked for further crossing to establish stable KO lines. Resulting mite lines LigIV-1 and LigIV-2 were deficient in NHEJ as the ligIV gene (ORCAE: tetur26g00400) was knocked out while lines Polθ-P1, Polθ-P2, Polθ-P3 and Polθ-HP all contained KO mutations in the polθ gene (NCBI: LOC107369726), impairing TMEJ (Figure 2, File S2). However, Polθ-HP carried a mutation early in the gene that disrupted both the helicase and polymerase domain of the Polθ enzyme, whereas the other lines lacked only the polymerase domain due to frameshift mutations located further downstream in the gene sequence (Figure 2, File S2). It is also important to note that the used sgRNAs were specific for polθ (0 predicted off-target effects according to the CRISPOR design tool) and do not target the helq gene described above. By crossing LigIV-1 with Polθ-P1 and LigIV-2 with Polθ-P2 (Figure S4), two additional lines were generated in which both NHEJ and TMEJ were disrupted. In mutant lines with an intact helicase domain, the expression of polθ may be strongly reduced due to nonsense-mediated mRNA decay (NMD). This is a quality control mechanism present in eukaryotes that degrades mRNA transcripts containing premature stop codons. Expression of polθ in mutant lines Polθ-P1 and Polθ-P2 was confirmed via qPCR, though the expression was 1.5-fold lower compared to the wild-type (WT) strain (Figure S5). Thus, by assuming that the helicase is still correctly folded in these mutants, the impact of a polymerase-deficient Polθ on DNA repair can also be investigated. Effect of disrupted DSB pathways on DNA repair To evaluate the effects of the disrupted pathways on DNA repair in spider mites, the phytoene desaturase (pd) gene was targeted with CRISPR-Cas9 using the SYNCAS method while providing a single-stranded oligodeoxynucleotide (ssODN) repair template containing a stop codon. Upon cleavage of pd, the resulting DSB could be repaired either via one of the end-joining pathways (NHEJ/TMEJ), leading to various mutations, or via HDR utilizing the ssODN, resulting in the incorporation of the stop codon. Both repair outcomes result in loss of red pigment in the eyes and legs (henceforth referred to as “albinism”), a visually distinguishable phenotype.65,83 Consequently, the sgRNA target site of the pd gene of all albino offspring were sequenced (369-bp fragment) to identify the specific mutation types that occurred (Figure 3, see also Figure S6 for a graphic representation of the experimental setup). This experimental design enables the quantification of both DSB induction rate (% albino offspring) and HDR efficiency (% albinos with the specific stop codon incorporated). Hence, HDR enhancement was assessed across the various mutant lines (Figure 4; Table S1). Concept of the screening method used to assess DNA repair outcome (A) T. urticae males displaying the WT (left) and albino phenotype (right) used as marker in the experiment. Image taken with a VHX-970F Digital microscope (Keyence) and VH-Z50 lens. Scale bar, 100 μm. (B) Alignment of the sgRNA target-site region in pd comparing the WT sequence with the supplied ssODN repair template and a subsample of albinos that were analyzed. Sequences that contained the stop codon (TAA) were counted as homology-directed repair (HDR) events while the others were counted as end-joining (EJ) events. Stacked bar plot displaying the effect of DNA repair KO lines on editing outcomes with 3 replicates per mite line The y axis represents the percentage of the total offspring with a certain type of mutation. “WT” represents the percentage of the offspring that still contained red pigment in their body conform the WT phenotype, all others displayed the albino phenotype. For some albino samples, no PCR product could be obtained, so the type of mutation could not be determined. Hence, these were categorized as “unknown.” Samples that were successfully sequenced were categorized according to the type of mutation: HDR if the STOP codon was present or EJ otherwise. Top and bottom charts display the offspring resulting from eggs oviposited 0–18 h and 18–44 h post injection respectively. Letters above graphs represent significant differences (α < 0.05) between strains comparing the amount of HDR events occurring in the total offspring. Data used to generate plots can be found in Table S1. When comparing the EJ and HDR events occurring within the albino offspring, DNA repair exhibited a significant (α < 0.05) shift toward HDR in all lines where polθ was disrupted, with nearly 100% of albino individuals carrying the stop codon. This is in stark contrast with the WT and ligIV KO lines where approximately 50% of repair events involved HDR, which clearly demonstrates the crucial role of TMEJ in end-joining repair of Cas9-induced DSBs in oocytes. Since no significant differences were observed between polθ KO lines (Polθ-P1, Polθ-P2, and Polθ-HP) and double KO lines (LP-1 and LP-2) in terms of HDR:EJ ratios, it is likely that the role of NHEJ in DSB repair is very minimal or perhaps even absent in addressing Cas9-induced breaks in the female germline (Figure 4). In terms of HDR occurring within the total offspring, again no significant difference was found for the ligIV KO lines compared to the WT, while significantly higher HDR frequencies were found for the polθ KO lines with the strongest effect seen for Polθ-P1 and Polθ-P2 (Figure 4, File S3). Finally, microhomologies flanking the deletions identified in albino males were analyzed (Table 2, File S6). Microhomology lengths from 1 to 4 nt were detected at the deletion junctions. In addition, an analysis of mutants carrying insertions at the target site revealed templated insertions in several samples (Figure S3, File S6), a characteristic signature of TMEJ. Table 2Observed microhomology length directly flanking deletions in albino males from various mite linesMicrohomology length (nt)Line01234WT368–4LigIV-1259–1LigIV-2711612Polθ-P21––––LP-1––1––LP-2–1––– Observed microhomology length directly flanking deletions in albino males from various mite lines Evaluating KI performance in the polθ KO lines Given the high HDR efficiency observed in the Polθ-P1 and Polθ-P2 strains, we performed additional experiments comparing one of these strains to the original WT strain. The first parameter that was assessed was the distance from the Cas9 cleavage site where the stop codon could still be introduced. This is an important feature for precise gene editing as a suitable protospacer sequence is not always available near the mutation site of interest. Injection experiments were conducted as described earlier, but the ssODN template contained a stop codon insert at 0, 15, or 30 nt away from the cleavage site (Figure 5, File S4). At these distances, stop codons could still be successfully introduced in both lines; however, a reduction in KI frequencies could be observed between a stop codon at 0 nt and a stop codon at 15 or 30 nt distance. In addition, the KI frequency obtained using the Polθ-P1 line was always equally high or higher compared to the WT strain (Figure 5; Table S2). An important note regarding the HDR rates in the albino offspring is that at 30 nt from the cut site, the KI ratio appears to be much lower compared to the other templates. This is not because end joining occurred but because these albinos did not incorporate the stop codon. The “TTT” motif present in the repair template to disrupt the PAM sequence was successfully incorporated in these mutants (indicating HDR) and also disrupted pd resulting in albinism (Figure 5). Effect of the insert distance from the CRISPR-Cas9 cut site on gene editing outcome From above to below: percentage of the total offspring resulting from injected females that displayed albinism; percentage of the sequenced albino offspring that successfully incorporated the stop codon from the respective ssODN repair template (alignment at the bottom); and percentage of the total offspring that contained the stop codon. x axis indicates the distance from the cleavage site and time interval of oviposition by females after injection with CRISPR-Cas9 components. Black bars indicate mean efficiency. Letters in graphs represent significant differences (α < 0.05) between strains and treatments within each time interval (without accent: 0–18 h; with accent in bold: 18–44 h). Data used to generate plots can be found in Table S2. A second interesting parameter that was assessed was the sequence size that could be inserted at the Cas9 cleavage site. Three different fragment sizes (20, 38, and 100 bp) were attempted to be knocked in at the DSB (File S4). Similar to the distance from the cut site, the increasing size of the insert negatively affected the observed KI frequencies with KI mutants being recovered for the 20 and 38 bp but not for the 100 bp fragment (Figure 6; Table S3). Again, the Polθ-P1 line outperformed the WT strain in terms of overall KI efficiency. Interestingly, no albino mutants could be found in the Polθ-P1 in the treatment using the 100 bp insert. Effect of insert size at the CRISPR-Cas9 cut site on gene-editing outcomes (A) Graphical representation of the experimental concept. (B) From above to below: percentage of the total offspring resulting from injected females that displayed albinism, percentage of the sequenced albino offspring that successfully incorporated the complete insert from the respective ssODN repair template (File S4), and percentage of the total offspring that contained the stop codon. x axis indicates the insert size and time interval of oviposition by females after injection with CRISPR-Cas9 components. Black bars indicate mean efficiency. Letters in graphs represent significant differences (α < 0.05) between strains within each time interval. Data used to generate plots can be found in Table S3. Fitness cost The Polθ-P1 strain appears to have potential to facilitate the generation of KI mutants making future gene editing in spider mites easier. However, the disruption of biological pathways can have adverse effects resulting in a fitness cost. To assess whether the polθ KO strains suffered from impaired fitness, we independently crossed the three mutant lines to the parental lines and allowed the alleles to segregate and recombine in a large population. In this fitness experiment, the relative allele frequencies were monitored for a total of 30 weeks (Figure 7). These allele frequencies were converted to allele counts assuming 400 alleles per sample (200 diploid females), and KO vs. WT counts were analyzed using a binomial generalized linear model with a logit link. The model included “week” as a continuous predictor and “population” as a fixed categorical effect. It was clear that “week” had a strong negative effect on the log-odds of the KO allele (estimate = −0.024 ± 0.002, z = −10.47, p < 2 × 10−16), indicating a highly significant temporal decrease in KO allele frequency. Populations already differed in allele frequencies at the first sampling time point (week 2), with Polθ-P2 and Polθ-P3 starting lower than Polθ-P1, but all populations exhibited a similar downward trend over time. The slope of −0.024 corresponds to an approximately 2.4% decrease in the odds of carrying the KO allele per week, indicating that a fitness cost is associated with the KO of polθ across the three biological replicates. In each cage, the spider mite population was kept large during the whole experiment to avoid bottlenecks with 6–12 highly infested bean plants being present. Hence, population sizes should be in the range of thousands of mites over the course of the experiment,51,84 and genetic drift is expected to be negligible.85,86 Additionally, KO allele frequencies remained in an intermediate range throughout the experiment (i.e., no extremes where allele frequencies neared 0 or 1), meaning the logit-transformed frequency changes follow the deterministic selection model closely. Under these conditions, the slope of the logistic decline in allele frequency should approximate the selection coefficient, s, acting against the KO allele.87,88 Fitness cost evaluation of polθ KO (A) Graphic representation of the crosses made to generate the initial mixed population to start the fitness experiment. (B) Evolution of the polθ KO allele frequency in sampled females over time based on relative peak height in Sanger chromatograms (Table S4, File S5) where the first three divergent peaks were used for calculations. Data are represented as mean ± SEM based on these three peaks and represent the technical variance. The three different KO lines are biological replicates. Discussion CRISPR-based genome editing in T. urticae has only recently become readily feasible through the SYNCAS method. Although targeted mutagenesis can now be achieved efficiently, the efficiency of introducing specific mutations via HDR has been variable.66,69,70 Improving precise gene editing requires a deeper understanding of how DSBs are repaired in this species, as DNA repair pathway usage determines whether intended edits are introduced by HDR or random mutations are introduced by error-prone end joining. This study therefore aimed to characterize the architecture of DSB repair pathways in T. urticae and to determine how the absence of specific pathways shapes gene editing outcomes. Genomic analyses revealed that putative orthologs of several core components of the classical NHEJ pathway, including DNA-PKcs, XRCC4 and XLF, could not be identified in T. urticae, while additional factors shared across TMEJ, SSA, and HDR—such as NBS1 and CtIP—also appear absent. Phylogenetic analysis of LigIV and Polθ further demonstrated that LigIV is poorly conserved across mites, mirroring the genomic absence of other NHEJ factors in T. urticae, while Polθ is broadly retained across Acari. While this suggests reduced end-joining machinery in mites, it is premature to conclude that NHEJ is completely redundant. In other invertebrates, which DNA repair pathway is used depends heavily on the specific tissue and the type of DNA break. For example, in Caenorhabditis elegans, somatic cells use classical NHEJ for certain repairs, while germ cells rely almost entirely on TMEJ. Especially when CRISPR/Cas9 was used, NHEJ played no detectable role.89 Therefore, our experiments can only give insight into DNA repair occurring in the oocyte/early embryo. These experiments demonstrated that disruption of polθ generated a near-complete shift toward HDR, indicating that DSB repair in T. urticae eggs depends heavily on TMEJ and that TMEJ and HDR compete directly. In contrast, ligIV KO mites showed minimal to no change in repair outcomes, which is in line with the observations made for germ line cells in C. elegans.89 Also in other organisms, such as the zebrafish Danio rerio and the mosquito Aedes aegypti, TMEJ seems to dominate DSB repair during the embryonic stage.90,91 Interestingly, even in one of the double KO lines where both NHEJ and TMEJ are disrupted, a single repair event still displayed a deleterious mutation distinct from the expected event upon precise HDR. This may indicate that additional repair pathways, such as SSA, are active in T. urticae. Consistent with this hypothesis, T. urticae carries a homolog (helq) of H. sapiens HELQ, known to be essential for SSA.29,30 However, the SSA pathway typically produces large deletions flanked by homologies,24,92 which was not observed here. Another possibility is that incomplete HDR events, where the repair template is only partially used, could generate aberrant junctions resembling mutagenic end joining, as observed in the zebrafish, Danio rerio.93 Additionally, for 21 out of 794 albino mites no PCR product could be obtained to determine the type of mutation. It cannot be excluded that these PCRs failed due to the presence of large deletions (potentially due to SSA repair) which resulted in the loss of one or both primer binding sites. Further analysis of the mutations occurring in albinos with deletions revealed that microhomologies of 1 to 4 nt were present in both the WT and ligIV KO lines with microhomologies of 1 and 2 nt being most frequent. This is in line with observations made for C. elegans where it was shown that microhomologies of a single nucleotide are already sufficient to achieve TMEJ.94 In some cases, no microhomology was found directly flanking the deletion. This might be explained by Polθ tolerating mismatches in the microhomologies, as recent works suggest,13 masking the microhomology used. In addition to the use of microhomologies, another typical feature of TMEJ is the occurrence of templated insertions, caused by aborted cycles of microhomology pairing and synthesis driven by Polθ. When an initial attempt to align and stabilize the broken single-stranded ends fails Polθ dissociates, template-switches, and initiates a new round of priming and elongation using adjacent sequences before final ligation occurs. The original genomic template for these insertions is often found proximal to the break site.95,96 No significant difference was observed in HDR:EJ ratios between the mutant line where the complete coding sequence of polθ was disrupted (Polθ-HP) and the lines where the sequence encoding the helicase domain was still intact (Polθ-P1 and Polθ-P2). This result is not unexpected given the role of the polymerase domain in TMEJ. After DNA ends are resected and short microhomologies anneal, the intermediate is unstable, containing gaps, mismatches, and overhangs. The polymerase domain initiates the gap filling from the annealed microhomologies, stabilizing the joint and committing the lesion to the TMEJ pathway. The helicase domain removes ssDNA-binding proteins such as RPA and promotes annealing of short microhomologies, stabilizing the initial end-synapsis intermediate, but is not known to be involved in strand synthesis. Without the polymerase domain to initiate the fill-in synthesis, these activities alone are insufficient to generate a product that can be ligated, and TMEJ fails.97,98,99,100 However, we did not validate whether the helicase domain is correctly folded and thus functional in Polθ-P1 and Polθ-P2 as truncation of multi-domain proteins can perturb folding, oligomerization or stability.101,102,103 Consequently, no strong conclusions can be drawn regarding the helicase function of Polθ in the Polθ-P1 and Polθ-P2 lines based on the data presented in this work. In terms of HDR occurring within the total offspring, the polθ KO lines displayed significantly higher HDR frequencies compared to the WT line (Figure 4, File S3). In addition, HDR frequencies were much higher in Polθ-P1 and Polθ-P2 compared to Polθ-HP which could hint at the helicase domain playing a role in HDR. If this was the case, however, one would also expect higher HDR frequencies in LP-1 and LP-2 as these lines contain the same KO alleles. The large variation between the various polθ KO lines could also be the result of genetic variation between these lines as these lines were established by backcrossing to fix the KO allele. Hence, segregating alleles from the WT GSS population that potentially influence gene-editing outcomes (e.g., factors that influence Cas9 uptake in the eggs) might have become fixed in some of the KO lines while they were lost in other lines. However, a more plausible theory is that this is simply experimental variation as only three replicated injections were performed for each line. Experiments probing the allowed distance between the Cas9 cleavage site and the location of interest to introduce a specific mutation revealed that increasing distance negatively impacted KI efficiencies. This result was in line with the expectations as similar observations have been made in other organisms where efficient ssODN-mediated editing is constrained to regions in close proximity to the DSB. For example, in C. elegans, Paix and colleagues showed that ssODN-mediated edits are most efficient when located within ∼10 nt of the Cas9 cut site, with efficiencies dropping sharply as distance increases.104 However, the reduced efficiency depends largely on the orientation of the ssODN being used, as edits in the PAM-distal region are more effectively obtained using the targeting strand (complementary to the gRNA) while edits in the PAM-proximal region are best performed using the non-targeting strand (same as gRNA sequence) as demonstrated in C. elegans and human cells.104,105 Here, edits were made only in the PAM-distal region using the targeting strand as ssODN. Additional mechanistic insights from mammalian systems support this trend. Harmsen et al. demonstrated that DNA mismatch repair (MMR) activity can suppress incorporation of ssODN-mediated substitutions, particularly when the intended edit is further from the DSB.106 This suggests that efficient base-pair replacement relies on a combination of proximity to the cut site and evasion of mismatch recognition pathways. Complementary evidence from Dekker et al. (2011) showed that transient suppression of the MMR protein MLH1 significantly increased single-base substitution efficiencies using ssODNs in mouse embryonic stem cells.107 These findings indicate that ssODN-mediated editing efficiency is modulated by both physical distance from the DSB and cellular DNA repair pathways, highlighting the importance of careful ssODN design and, in some contexts, transient modulation of repair proteins to maximize editing outcomes. Studying the effect of MMR on gene editing in T. urticae might further boost the utility of the CRISPR-Cas tools in this species. Our results also indicate a clear negative correlation between insert size and HDR efficiency in T. urticae. While short insertions of 20 and 38 bp could still be introduced with relatively high efficiencies in the polθ KO line, attempts with a 100 bp insert failed to generate any recovered KIs. Given the robustness of the method in all other experiments, it is unlikely that this is simply a failure of the CRISPR approach in all three replicates. More likely would be that offspring where Cas9 successfully cut the target-site were unable to repair the DSB and died during development. This outcome highlights an unexpectedly sharp KI limitation in this species, as a 100 bp insert is not exceptionally large compared to other organisms where considerably larger insertions have been achieved using ssDNA as repair template. For instance, in zebrafish, ssDNA donors enabled successful insertion of a 750 bp fragment108 and mouse embryos, ssDNA-mediated KIs of fragments exceeding 1 kb have been obtained.109 Even in C. elegans, successful insertion of a 180 bp fragment was reported,110 demonstrating that the HDR capacity of T. urticae is unusually constrained when ssDNA templates are used. However, the screening depth in our experiment may have been too limited to draw definitive conclusions, as only 360 males were analyzed for the 100 bp insert. It remains possible that successful KI mutants could be recovered when injecting a substantially larger number of females and screening thousands of offspring. Also the use of dsDNA templates instead of ssDNA or different homology-arm lengths could improve KI success of larger constructs.111,112,113 Furthermore, it is important to note that in contrast to many model organisms where embryos are injected, allowing delivery of large DNA constructs, the SYNCAS method is based on maternal injections. Hence, delivery of large DNA templates to the eggs can be an additional bottleneck factor, limiting KI of larger constructs. HDR efficiency is also influenced by the cell cycle, as the repair machinery is predominantly active during the S and G2 phases, when sister chromatids or homologous templates are available.114,115,116 DSBs arising in G1 or early S phase are less efficiently repaired by HDR and are more likely to be resolved through end-joining pathways. In T. urticae, the precise cell cycle stage of oocytes or eggs at the time of Cas9-induced DSB is unknown. Larger inserts require more extensive DNA synthesis, and if HDR is initiated late in S/G2, there may be insufficient time or cellular resources to complete the repair before the cells enter mitosis, leading to failed integration. Regardless of the reason why, the insertion of large DNA fragments using ssDNA as a repair template might be a challenging strategy. Hence, future work that would attempt KI of transgenes in spider mites could also focus on alternatives such as the use of linear dsDNA donors which can serve as templates for HDR and for integration through end-joining pathways. This could allow incorporation of donor DNA even in cells where HDR is limited or incomplete. In various organisms, dsDNA donors have successfully facilitated the integration of relatively large fragments, demonstrating that precise KIs can be achieved when the donor is compatible with multiple repair routes.117,118,119,120,121 By exploiting both HDR and end-joining mechanisms, dsDNA donors may therefore increase the likelihood of recovering transgene insertions in T. urticae, overcoming the size limitations observed with ssDNA templates. Careful design of the donor, including appropriate homology arms and protection against degradation or multimerization, will likely be essential to maximize integration efficiency.104,113,122,123 However, as stated earlier the delivery itself of large constructs to the eggs via maternal injection might be the first hurdle to overcome. In the last experiment, it was observed that loss of polθ imposed a measurable and significant, though relatively minor, fitness cost under laboratory conditions (i.e., low UV conditions). This was done by monitoring the allele frequencies over time, a strong method for estimating fitness because it integrates all components of fitness (viability, development, reproduction, sex-specific effects, etc.) over the full life cycle and under realistic competitive conditions.124,125,126,127 Our setup started from controlled crosses to generate a heterozygous F1 population which was backcrossed to the KO line to establish the F2 starting population. This design allows recombination to occur in F1 females, thereby homogenizing the genetic background around the KO locus. An alternative approach would have been to start the population by simply mixing an equal number of females from the KO and WT line. However, the starting allele frequency may then be subjugated to genetic drift when only a few males disproportionately sire offspring and there can be selection occurring on the mite lines as a whole rather than on the KO allele itself. Starting with controlled crosses and obligate heterozygotes enforces a balanced initial allele frequency, reduces founder effects, and ensures that both alleles enter the experiment on comparable genetic footing. As a result, a decline in KO allele frequency in subsequent generations can be attributed more confidently to a fitness cost associated with the KO allele itself, rather than to unrelated fitness differences linked to the KO line’s genetic background (which is a haplotype selected from the WT strain). Within these populations established by direct crosses where both KO and WT alleles segregated, the KO allele frequency declined slowly during the sampling period of 30 weeks, indicating that although disruption of polθ is not lethal, it reduces competitive fitness. Deleterious effects resulting from the loss of Polθ has been reported in other organisms. In mice, increased chromosomal instability and sensitivity to DNA-damaging agents was reported128,129 while zebrafish embryos were not viable after DSB induction.130 Studies on D. melanogaster suggest that the absence of Polθ does not result in immediate lethality but does impair genomic stability and fitness under stress. For example, null mutants are viable, but they display increased sensitivity to ionizing radiation and chemical mutagens, consistent with a role in repair of damaged chromosomes.131 Moreover, these mutants exhibit elevated rates of mitotic crossovers when DNA breaks are induced, implying that Polθ normally suppresses recombination events that can destabilize chromosomes.131 Furthermore, it was shown that Polθ is particularly critical when other repair pathways are compromised as flies lacking both Polθ and the Holliday junction resolvase Slx4 suffer from synthetic lethality and high levels of cell death.131 Follow-up work could assess the fitness cost of polθ KO in T. urticae under DNA-damaging conditions as well as study the effect on mutation rates compared to the WT strain. The presence of this fitness cost is an important consideration for future work using polθ KO lines as it might confound phenotypic analysis. Conclusion In conclusion, the CRISPR-Cas9 experiments that were performed in this study using the SYNCAS approach demonstrate that DSBs introduced in T. urticae are primarily repaired via TMEJ. Furthermore, it was shown that a polθ KO line, even though DNA sizes that can be introduced still appear to be limited, can facilitate the generation of KI mutants via ssODNs; thus, providing a valuable tool for future genetic studies in this species. Limitations of the study An important limitation of this study is the tissue-specific nature of the used SYNCAS method, which relies on maternal injections to target oocytes and early eggs. Consequently, the observed dominance of TMEJ repair is strictly representative of germline and early embryonic environments, meaning the redundancy or necessity of NHEJ in somatic tissues remains undetermined. Furthermore, our conclusion that larger fragments (e.g., 100 bp) fail to achieve KI status is bounded by our screening depth of 360 male offspring, leaving open the possibility that larger inserts might integrate at very low frequencies detectable only through massive screening. In addition, while no phenotypic differences were observed between the full KO (Polθ-HP) and polymerase-deficient lines (Polθ-P1,2), the independent functional role of the N-terminal helicase domain remains unresolved because the stability, folding, or oligomerization of these truncated fragments was not explicitly validated. Finally, the different KO lines might also differ in other alleles due to the genetic variation present in the parental GSS strain. These genetic differences might influence experimental outcomes and be partly responsible for the phenotypic variations observed between the various KO lines. Resource availability Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Thomas Van Leeuwen (thomas.vanleeuwen@ugent.be). Materials availability Mite lines generated in this study can be made available on request but may require a payment and/or a completed materials transfer agreement. Data and code availability •All data supporting this work can be found in the supplemental information of this article and on Figshare (https://doi.org/10.6084/m9.figshare.32475891). All data supporting this work can be found in the supplemental information of this article and on Figshare (https://doi.org/10.6084/m9.figshare.32475891). •This paper does not report original code. This paper does not report original code. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request Acknowledgments This work was supported by the Horizon 2020 Research and Innovation Program (grant 101136611-NEXTGENBIOPEST) and Research Foundation Flanders (FWO) (grants G017923N and G0A1525N). S.D.R. is a post-doctoral fellow of FWO (grant 1237426N). Author contributions Conceptualization, S.D.R. and T.V.L.; conducting experiments, S.D.R.; data curation, S.D.R. and W.D.; formal analysis, S.D.R. and W.D.; funding acquisition, S.D.R. and T.V.L.; investigation, S.D.R.; methodology, S.D.R., W.D., and T.V.L.; visualization, S.D.R. and W.D.; writing—original draft, S.D.R.; writing—review and editing, S.D.R., W.D., and T.V.L. Declaration of interests The authors declare no competing interests. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work the authors used Gemini in order to improve writing style. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERChemicals, peptides, and recombinant proteinsAlt-R® S.p. Cas9 nuclease V3Integrated DNA Technologies10007808BAPtofect® (BAPC)Phoreus Biotech/Quillaja saponinSigma-AldrichS4521GoTaq® G2 DNA polymerasePromegaM7845GoTaq® qPCR master mixPromegaA6001Proteinase KSigma-AldrichP6556Allura red ACSigma-Aldrich458848AgaroseThermoFisher16500500Critical commercial assaysRNeasy plus kitQiagen73404Maxima first strand cDNA synthesis kitThermo Fisher ScientificK1641DNeasy blood & tissue kitQiagen69504Deposited dataOriginal dataThis paperhttps://doi.org/10.6084/m9.figshare.32475891Bryobia sp. de novo assembly sequenceNCBI sequence read archiveAccession: SRR15410508Tetranychus urticae genome databaseORCAEhttps://bioinformatics.psb.ugent.be/orcae/overview/TeturDrosophila melanogaster genomic resourcesFlyBasehttps://flybase.orgExperimental models: Organisms/strainsTetranychus urticae: GSS strainProf. Thomas van leeuwen labStumpf et al.132OligonucleotidesSingle guide RNAs (File S4)Integrated DNA technologies/ssODN repair templates (File S4)Integrated DNA technologies/Genotyping and RT-qPCR primers (File S4)Integrated DNA technologies/Software and algorithmsCRISPORConcordet and Haeussler133https://crispor.gi.ucsc.edu/CLC genomics workbench (Version 24)QIAGENMAFFTKatoh et al. 2002https://mafft.cbrc.jp/alignment/software/IQ-TREENguyen et al.134https://iqtree.github.io/ModelFinderKalyaanamoorthy et al.135https://iqtree.github.io/InterProScanJones et al.136https://www.ebi.ac.uk/interpro/search/sequence/Primer-BLASTNCBIhttps://www.ncbi.nlm.nih.gov/tools/primer-blast/qbase+BiogazelleBioEdit software (Version 7.2.5)Hall137https://bioedit.software.informer.com/Python (Version 3.12)Python software foundationhttps://www.python.org/OtherNanoject III MicroinjectorDrummond Scientific/3–000-203-G/X Glass capillariesDrummond Scientific/P1000 Micropipette pullerSutter instruments/BV-10 Micropipette bevelerSutter instruments/M205FCA MicroscopeLeica/VHX-970F Digital microscopeKeyence/VH-Z50 LensKeyence/DeNovix® DS-11 FX SpectrophotometerDeNovix/Mx3005P qPCR MachineAgilent technologies/ Experimental model and study participant details All experiments were performed using the GSS strain of T. urticae which is described in detail by Stumpf et al.132 Mites were maintained on Phaseolus vulgaris cv. Prelude in climate chambers at 26 °C, 60% relative humidity (RH) and 16:8 light:dark photoperiod. In order to have mites of similar age in the experiments, adult females were placed on a detached bean leaf on wet cotton and allowed to lay eggs for 24 h after which they were removed. After 8 days, deutonymph females were transferred to a new bean leaf without males to ensure no fertilization could occur and resulting offspring would be haploid males. These females remained on the leaf until the injection experiments were performed (3–4 days in adult stage). After experiments, mites were kept in the incubators under the same condition. Method details Screening for putative orthologues of DSB repair proteins in T. urticae Putative T. urticae orthologues of human and Drosophila melanogaster proteins involved in DNA repair pathways were identified using reciprocal BLASTp searches (or tBLASTn if no hit was found) in ORCAE (https://bioinformatics.psb.ugent.be/orcae/overview/Tetur). Protein sequences of known DNA repair factors from humans (retrieved from Swiss-Prot) and D. melanogaster (retrieved from FlyBase) were queried against the T. urticae protein database (Tetur_PEP_LAST) using BLASTp and reference genome (T.urticae_LONDON_GENOME) using tBLASTn with default parameters. Candidate T. urticae proteins were subsequently used as queries in reciprocal BLASTp searches against the human or D. melanogaster reference proteomes on NCBI (refseq_select). Proteins were considered putative orthologues when the reciprocal BLASTp returned the original query protein as the top hit. Identification of T. urticae ligIV and polθ and distribution of ligIV and polθ among Acari species D. melanogaster LigIV (FBgn0030506) and Polθ (FBgn0002905, also known as Mus308) protein sequences were first used as a query in a BLASTp search against the T. urticae prediction proteome in NCBI (T. urticae Annotation Release 101). The most highly ranked T. urticae hit for each query was used in a reciprocal BLASTp search against the D. melanogaster predicted proteome to confirm the identification of putative T. urticae LigIV and Polθ orthologues. Subsequently, T. urticae LigIV and Polθ were used in a reciprocal BLASTp search against predicted proteomes of following Acari groups/(super)families/orders: Ixodida (ticks, Rhipicephalus microplus, R. sanguineus, and Ixodes scapularis), Mesostigmata (Varroa destructor and Galendromus occidentalis), Astigmata (Dermatophagoides pteronyssinus), Oribatida (Archegozetus longisetosus and Oppia nitens), Eriophyoidea (Aculops lycopersici and Fragariocoptes setiger), Parasitengona (Leptotrombidum deliense and Dinothrombium tinctorum), Eupodinae (Halotydeus destructor), Eleutherengona (Heterostigmata (Pyemotes zhonghuajia and Polyphagotarsonemus latus) and Tetranychoidea (Brevipalpus yothersi, Panonychus citri, Amphitetranychus viennensis, Tetranychus truncatus and T. urticae strains London and ipm)). A tBLASTn search using T. urticae LigIV as query was performed on the genomes of these species, to avoid missing unannotated ligIV. In addition to the above Acari species, we also mined the genome of Bryobia sp, another member of the Tetranychoidea, using a tBLASTN search and T. urticae LigIV and Polθ as query. The Bryobia sp. genome was assembled de novo using CLC Genomics Workbench version 24, with default settings and the Sequence Read Archive accession SRR15410508. Acari LigIV and Polθ proteins were complemented with those of other Chelicerata (Centruroides sculpturatus and Parasteatoda tepidariorum), Insecta (Bombyx mori, Tribolium castaneum), Crustacea (Daphnia pulex), Vertebrata (Homo sapiens, Danio rerio, and Xenopus laevis), and a fungus (Neurospora crassa). For all species, except Bryobia sp., accessions and/or annotations/genomes used in BLASTp/tBLASTn are shown in File S1. LigIV and Polθ proteins were aligned using MAFFT with the FFT-NS-I and E-INS-I option, respectively and the resulting LigIV and Polθ alignments were used in a maximum likelihood phylogenetic analysis with IQ-TREE134 using default settings, the LG + F + I + G4 model (identified to be the best-fit model by ModelFinder135) and with 1000 ultrafast bootstraps. Resulting trees were rooted with LigIV and Polθ of Neurospora crassa, respectively. Lastly, InterProScan136,138 was used to determine InterPro domains in T. urticae LigIV and Polθ, while the R-package Gviz139 was used to visualize the genome regions of polθ and helq in Tetranychinae species. Generating KO lines KO lines in ligIV and polθ were created using CRISPR-Cas9 via the SYNCAS method.66 Briefly, for each target-site 200 adult virgin females were immobilized on Allura red stained 2% agarose gel with their dorsal side in contact with the agar. Females were then injected ventrally in the body cavity between the third pair of legs with 3 nL of CRISPR-Cas9 solution containing 15 μg/μL Cas9, 6 μg/μL sgRNA (specific for 1 target), 1 μg/μL BAPC and 0.1 μg/μL Quillaja saponins. Spacer sequences (File S4) were designed using CRISPOR133 to minimize potential off-target cleavage. Recombinant Streptococcus pyogenes Cas9 protein (Alt-R S.p. Cas9 Nuclease V3) and sgRNAs were purchased from Integrated DNA Technologies. BAPC (BAPtofect) and Quillaja saponin (Cat# S4521) were ordered from Phoreus Biotech and Sigma-Aldrich respectively. All injections were performed using a Nanoject III microinjector (Drummond Scientific) and needles made from 3 to 000-203-G/X Glass Capillaries (Drummond Scientific). Needles were pulled using a P-1000 Micropipette Puller (Sutter Instruments) with following settings: “heat: 500, pull: 60, velocity: 70, delay: 200, pressure: 500, Ramp: 490” and were sharpened with a BV-10 Micropipette Beveler (Sutter Instruments) at a 15° angle. Injections were carried out under an M205FCA microscope (Leica). After injecting, females were allowed to lay eggs for a total of 48 h. From the eggs deposited 24–48 h post injection, resulting adult males were mated with unfertilized females. After two days, DNA was extracted from the males by crushing individuals in 10 μL STE buffer (100 mM NaCl, 10 mM Tris-HCl, 1 mM EDTA, pH 8) and 1 μL proteinase K (10 mg/mL). This crude extract was incubated at 37 °C for 30 min followed by proteinase K deactivation at 98°C for 10 min. A PCR was performed on these raw DNA extracts using the GoTaq G2 DNA Polymerase and a primer pair specific for the target-site which were designed with Primer-BLAST (File S4). Resulting PCR product was evaluated by running it on 2% agarose gel and was sent to LGC Genomics for Sanger sequencing. Once males harboring KO mutations were identified, their heterozygous daughters were used for subsequent inbreeding to establish stable KO lines. Hence, two lines were obtained for ligIV (LigIV-1 and LigIV-2) and four lines for polθ (Polθ-P1, Polθ-P2, Polθ-P3 and Polθ-HP). Sequencing results are available in File S2. Two double KO lines (LP-1 and LP-2) were created by crossing the LigIV lines with the Polθ lines (LigIV-1 x Polθ-P1 and LigIV-2 x Polθ-P2) as depicted in Figure S5. Evaluating knock in frequencies in KO lines To determine the time intervals for ovulation post injection during in these experiments, 300 virgin females were injected with CRISPR-Cas9 mix targeting the pd gene which upon KO results in albinism (Figure 3B). Females were then allowed to oviposit for following intervals post injection: 0–18 h, 18–21 h, 21–25 h, 25–41 h, 41–44 h and 44–48 h. This was repeated three times and based on these results (Table S5) it was decided to let the females oviposit from 0 to 18 h and 18–44 h post injection for all KI experiments. In a first experiment, the various ligIV and polθ KO lines were injected with a CRISPR-Cas9 mix targeting pd which also contained a ssODN template that was designed to introduce a stop codon at the Cas9 cleavage site (Figure 3, File S4). The resulting CRISPR-Cas9 mix consisted of 18.8 μg/μL Cas9, 7.5 μg/μL sgRNA, 5.2 μg/μL ssODN, 5.2 μg/μL BAPC and 0.1 μg/μL Quillaja saponin. For each KO line, injections were performed in three independent replicates with 100 virgin females being injected per replicate. Resulting male offspring were scored for albinism after which albinos were Sanger sequenced to analyze the type of mutation that occurred at the Cas9 cleavage site in pd (random insertions and deletions vs. introduction of the stop codon). Hence, KO and KI frequencies could be compared between the various lines which were calculated as follows: For samples displaying KO mutations, indels were analyzed and microhomology regions were determined, which is summarized in File S6. In a next experiment, the KI frequencies were evaluated in Polθ-P1 and compared to the WT GSS strain using different ssODN repair templates which varied in insert size and distance from the Cas9 cleavage site but still resulted in the disruption of pd upon integration (File S4). Again, three replicates of 100 virgin females were injected per line after which albino offspring were analyzed. Determining polθ expression RNA was extracted with the RNeasy Plus kit (Qiagen) from the WT GSS strain as well as Polθ-P1 and Polθ-P2 using 100 adult females for each line. Concentrations of RNA samples were determined using DeNovix DS-11 FX spectrophotometer (DeNovix, USA) and quality was assessed by running them on a 1% agarose gel. Using the Maxima First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA), 1 μg of RNA was converted to cDNA after which RT-qPCR reactions were conducted with the GoTaq qPCR Master Mix (Promega, USA) and qPCR primers (File S4) in a Mx3005P qPCR machine (Agilent Technologies, Belgium). Cycle conditions were 95 °C for 2 min followed by 40 cycles of 95 °C for 15 s, 55 °C for 30 s, and 60 °C for 30 s, followed by a melting curve analysis step. Melting curves and no-template controls (NTC) were used to confirm, respectively, the specificity of amplification and absence of contamination. The program qbase+ (Biogazelle, Belgium) was used for the analysis of raw quantification cycle (Cq) values, which were all first normalized against the housekeeping genes ubiquitin C (tetur03g06910) and GAPDH (tetur25g00250). Two technical replicates were used for each sample, and mean log2FC values relative to reference genes were assessed. Fitness experiment To assess whether the disruption of polθ is linked to a fitness cost, three populations were generated by crossing 50 virgin WT females with 50 males of either Polθ-P1, Polθ-P2 or Polθ-P3. The resulting heterozygous F1 females were backcrossed to males from their respective mutant parental line (Figure 7). Three days after setting up the couples, 100 F1 females were allowed to oviposit over a time span of 24h for each line and these synchronized F2 eggs were then used to inoculate a single bean plant. The population was expanded by adding extra bean plants over time (upto 12 plants in total) and the population was kept large with a minimum of 6 highly infested plants to avoid population bottlenecks. Fitness was evaluated by monitoring the frequency of the KO alleles in the populations. Two weeks after the inoculation of bean plants sampling of mites began with a regular interval of two weeks for a total of twenty weeks followed by one more sampling 10 weeks later. For each sample four leaves were taken from four randomly picked plants and 50 females were collected per leaf, totaling 200 mites per sample. DNA was extracted from each sample using a DNeasy Blood & Tissue Kit (Qiagen, Germany) following the manufacturer’s instructions. PCR was performed using the primer pair “polymerase θ (polymerase domain)” (File S4) and resulting product was sent for Sanger sequencing. BioEdit software – version 7.2.5137 – was used to analyze Sanger chromatograms and measure peak heights of the WT and KO allele. Relative allele frequency was calculated by taking the average of the first three nucleotide positions where the WT and KO allele differed (File S5): Quantification and statistical analysis For the experiment comparing the various DNA repair KO lines, two ligIV KO lines and four polθ KO lines originating from different mutant G0 males were created to obtain biological replicates. For each obtained mutant line, three independent injections were performed. Also, for the other gene editing experiments, three independent injections were performed for each treatment group within each mite line. For the fitness experiment of polθ, three independent cage experiments were set up using Polθ-P1, Polθ-P2 and Polθ-P3 as biological replicates. Statistical analyses of gene editing outcomes were conducted in R (version 4.5.2). For each dataset and timepoint interval (0–18 h and 18–44 h), contingency tables were constructed by aggregating raw counts across replicates within each strain. Pairwise 2 × 2 comparisons were performed using Fisher’s exact tests to evaluate differences between individual groups. Raw p-values were adjusted for multiple testing using the Benjamini–Hochberg procedure to control the false discovery rate (FDR at α = 0.05). All analyses were carried out separately for each dataset (Tables S1–S3), and the results were exported to Excel (File S3). For the fitness cost experiment, statistical analyses were performed in R. Allele frequencies were first converted to allele counts assuming 400 alleles per sample (corresponding to 200 diploid females used for DNA extraction). Temporal changes in KO allele frequency were analyzed using a binomial generalized linear model (GLM) with a logit link, where KO vs. WT allele counts were used as the binomial outcome. The model included week as a continuous predictor and population as a fixed categorical effect to account for initial differences among cages. Model coefficients, standard errors, z-values, and p-values were obtained from the GLM summary output. Footnotes Supplemental information can be found online at https://doi.org/10.1016/j.isci.2026.117402. Contributor Information Sander De Rouck, Email: sander.derouck@ugent.be. Thomas Van Leeuwen, Email: thomas.vanleeuwen@ugent.be. Supplemental information Document S1. Figures S1–S6 and Tables S1–S5 and File S1–S6 References Associated Data Supplementary Materials Document S1. Figures S1–S6 and Tables S1–S5 and File S1–S6 Data Availability Statement •All data supporting this work can be found in the supplemental information of this article and on Figshare (https://doi.org/10.6084/m9.figshare.32475891). All data supporting this work can be found in the supplemental information of this article and on Figshare (https://doi.org/10.6084/m9.figshare.32475891). •This paper does not report original code. This paper does not report original code. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request