A Recql5 mutant enables complex chromosomal engineering of mouse zygotes

Complex chromosomal rearrangements (CCRs) are often observed in clinical samples from patients with cancer and congenital diseases but are difficult to induce experimentally. For generating animal models, these CCRs must be induced as desired, otherwise they cause profound genome instability and/or result in cell death. Here, we report the first success in establishing animal models for CCRs. The disruption of Recql5, which degrades RAD51 during DNA repair, successfully induces CRISPR/Cas9-mediated CCRs, establishing a mouse model containing triple fusion genes and megabase-sized inversions. Notably, some of these structural features of individual chromosomal rearrangements use template switching and microhomology-mediated break-induced replication mechanisms and are reminiscent of the newly described phenomenon “chromoanasynthesis.” Whole-genome sequencing analysis revealed that the structural variants in these mice caused only target-specific rearrangements. Thus, these data show that Recql5-deficient mice would be a novel powerful tool for analyzing the pathogenesis of CCRs, particularly chromoanasynthesis, whose underlying mechanisms are poorly understood.


Introduction
Recent advances in bioinformatics technologies have led to the detection of complex chromosome rearrangements (CCRs) consisting of ≥ 3 chromosomal breaks in patients with cancer and congenital diseases 1 . These rearrangements caused by catastrophic cellular events can affect phenotype, thereby inducing a disease-promoting environment 2 . In particular, chromoanasynthesis, a recently discovered form of CCRs, is caused by erroneous DNA replication of a single chromosome through fork stalling and template switching (FoSTeS) and microhomology-mediated break-induced replication (MMBIR), which generate regions with complex rearrangements 3 . However, the pathogenic mechanisms underlying these diseases remain unclarified, and the establishment of appropriate animal models is essential for their elucidation. Although recent technologies, such as clustered regularly interspaced short palindromic repeats (CRISPR)-dependent base editing, prime editing, and DNA integration, have allowed for high-precision genome interrogation 4 , they have not yet been adapted to model CCRs in the germ line.
Here, we hypothesized that the efficient induction of CCRs could be achieved by manipulating the DNA repair pathway as accumulating evidence indicates that changes in DNA repair timing often accompany genomic rearrangements 2 . However, given the importance of DNA repair genes in genome maintenance, these strategies may have adverse consequences. Indeed, most genes involved in the DNA repair pathway are essential, and their homozygous disruption leads to embryonic lethality in mice 5 .
However, Recql5-deficient mice were reported to live to adulthood 6 . RecQ protein-like 5 (RECQL5) helicases can displace the DNA repair protein RAD51 from single-stranded (ss)DNA and disassemble nucleoprotein filaments, thereby suppressing homology-directed repair (HDR) 6 . Transient accumulation of the homologous recombination and RAD51 has been reported in Recql5-deficient cells, which could alter DNA repair pathways, thereby contributing to chromosomal rearrangements 6 .
Hence, we investigated whether CCRs might be induced in Recql5-deficient mice and succeeded in establishing CCRs model mice. Notably, a novel DNA repair system, FoSTeS/MMBIR, was involved in a CCR model mouse line.

Recql5 mutant enables CRISPR/Cas9-mediated CCRs in mouse zygotes
Using a previously developed in vivo electroporation technique, called improved genome editing via oviductal nucleic acid delivery (i-GONAD) 7 ( Figure S1), we first established a mouse strain with a deletion of the RAD51-binding domain in RECQL5 (Recql5 em1Cu ) ( Figure S2A, S2B). These Recql5 em1Cu/em1Cu mice were fertile and unexpectedly showed no overt signs of other diseases, such as tumorigenesis or inflammation ( Figure S2C, S2D); five male and four female mice were observed for more than 60 weeks.
Second, we sought to investigate whether the Recql5 mutant approach could be effectively utilized to generate inversion rearrangement mouse models. We recently succeeded in inducing a 7.67-Mb inversion in wild-type (WT) mice 8 ( Figure 1A). We induced this large inversion in Recql5 em1Cu/ em1Cu mice as well, with higher genome editing efficiency than that in WT mice ( Figure 1B, 1I). Homozygous inversion mice, In (15) #6 , were generated by breeding heterozygous males and females ( Figure 1C), and exhibited a white-spotted phenotype due to disrupted Adamts20 expression.
The human HMGA2-WIF1 fusion gene, which was generated by the inversion on chromosome 12, was shown to activate the Wnt/β-catenin pathway and has been found in salivary gland tumors and breast adenomyoepitheliomas 9 . Likewise, we efficiently modeled the HMGA2-WIF1 inversion on chromosome 10 in Recql5 em1Cu/ em1Cu mice ( Figure 1D, 1E, and 1I), and the resulting mice, named In(10) #2 , invariably harbored the Hmga2-Wif1 inversion. These mice expressed the Hmga2-Wif1 fusion gene ( Figure 1F, 1G), and displayed a recessive pygmy phenotype due to the Hmga2 mutation 10 ( Figure   1H). Taken together, these results indicate that the Recql5 mutant approach can efficiently generate inversion mouse models.
Next, we attempted to apply this technique to produce CCR model mice. We  11 . We designed gRNAs targeting these three genes and single-stranded oligodeoxynucleotides (ssODNs) that joined the chromosomal breakpoints, each of which had a sequence homologous to each junction point, such that two inversions were induced by the HDR process between the targeted regions and the homologous ssODNs. The 5′ and 3′ ends of the ssODNs were protected with two consecutive phosphorothioate-modified bases to improve the efficiency of HDR 12 . We then injected CRISPR/Cas9 ribonucleoproteins (RNPs) targeting these genes into pregnant females to generate chromosomal rearrangements 7 ( Figure S1). The generation of predicted chromosomal rearrangements was first confirmed by PCR of genomic DNA and then validated by sequencing the corresponding fusion transcript. Here, founder (F0) mice, in which a central breakpoint was detected, were defined as having induced CCRs and were used for subsequent analyses.
In the Recql5 em1Cu/ em1Cu strain, we obtained six F0 pups via cesarean section and found that four had chromosomal rearrangements in the target locus, yielding three viable F0 CCR mice ( Figure 2B, 2C: #1, #4a, #4b). By contrast, control WT (C57BL/6N) strains showed partial chromosomal rearrangements in three of the eight pups; however, we could not obtain the surviving founder, F0 ( Figure 2C). Imprecise repair of double-strand breaks (DSBs) has the potential to be highly deleterious, owing to genomic instability, including the formation of chromosomal rearrangements 13 .
Despite this seemingly difficult chromosomal rearrangement pattern, we confirmed the expression of the corresponding fusion transcript using reverse transcription (RT)-PCR and direct sequencing ( Figure 2B). Homozygous CCRs (10) Figure 2H).

Evaluation of genome-wide target specificity in the Recql5 mutant
We characterized the genomic structure of the mouse strains in detail using whole-genome sequencing (WGS). The CCRs(10) #4a strain showed multiple breakpoint 5 junctions and was classified as deletion (red), duplication (green), and inversion (teal and blue) based on paired-ends with read depth changes ( Figure 3A, 3B). Sequencing of the breakpoint revealed microhomology patterns and sister chromatid-containing templates, in which the added insertion was dependent on the Cas9 target site ( Figure   3C). Notably, microhomologies seemed to be used to switch the nearby template during DSB repair stalling and collapse-a process termed FoSTeS and MMBIR 14,15,16 . This structural feature of individual chromosomes is reminiscent of the newly described phenomenon chromoanasynthesis, which is observed in tumors, as well as in patients with congenital diseases 3 . In addition to FoSTeS/MMBIR and tandem inversions, a 601-kb deletion was identified in the CCRs (10)  Contrastingly, HDR appeared to repair two of the five breakpoints examined in the WT background, while the others were likely repaired by microhomology-mediated end joining or non-homologous end joining; however, no FoSTeS/MMBIR was observed (Table S1). A complex genome architecture confuses the DNA repair machinery and induces template-switching events driven by FoSTeS and MMBIR 3,17 . Thus, if the RAD51-filament is retained, loss of Recql5 function may promote DNA repair machinery confusion and dictate the choice between the FoSTeS/MMBIR pathways.
These candidate rearrangement breakpoints were identified from WGS data using the Manta structural variation detection algorithm; however, no target-independent complex rearrangements were identified by comparing the test genomes 18 ( Figure 3D, Figure S3D, S4D). Thus, these results establish the efficiency and specificity of chromosomal engineering using the proposed approach.

Recql5 mutant mediates a broad pattern of chromosomal rearrangements
To test the broad applicability of the Recql5 mutant approach, we generated CCRs on chromosome 2, which included topologically associating domains at the HoxD loci.
These loci are necessary for the development of the proximal part of the limb, including 6 the future arm and forearm 19 . To study the effects of the structural variants, we engineered CCRs comprising sequential inversions between Atf2 neighborhood (Atf2N) and Hoxd1N and between Hoxd1N and Nfe2l2N ( Figure 4A). In the Recql5 em1Cu/em1Cu strains, we screened F0 mice using PCR amplification and DNA sequencing of three junction points and detected chromosomal rearrangements in two of the four F0 mice ( Figure 4B, 4C). In one of the possible CCR mouse lines, named F0-#2, two HDR-repaired junction points were detected, whereas the other junction point was repaired with structural changes that could not be amplified by PCR ( Figure 4B). By contrast, the control C57BL/6N strain did not show any breakpoints in the five pups ( Figure 4C). Although the RNP-based CRISPR/Cas system is characterized by high-efficiency germline transmission 20 , F0-#2 did not transmit the targeted CCRs to their 34 offspring. In additional studies, we examined DNA extracted from the testes and semen of F0-#2 animals; however, the semen samples did not exhibit a PCR signal between Atf2N and Nfe2l2N ( Figure 4D). In our experience, F0-#2 was the first mouse line in which the mutation was not transmitted to the gametes using an RNP-based CRISPR/Cas system. Although we cannot accurately explain this transmission disturbance, asymmetric disjunction may result in a wide range of highly imbalanced gametes, many of which do not survive until the end of spermatogenesis 21  To the best of our knowledge, this is the first study to successfully introduce CCRs into mouse zygotes using CRISPR-based methods. This approach opens new avenues for investigating the role of chromosomal rearrangements in diseases and provides a powerful tool for engineering genetic modifications in various organisms.

Limitations of the study
Here, we describe a new genome editing method using a Recql5 mutant mouse and show that it can efficiently induce various types of chromosomal rearrangements, including CCRs and inversions, in mouse zygotes. Although this technique has considerable potential, it is also associated with unexpected DSB repair mechanisms such as FoSTeS/MMBIR. Thus, the structural features of induced rearrangements are 7 likely to depend on a variety of factors, such as features specific to the targeted genomic regions, the number of breakpoints, and the unique properties of mouse lines.
In FoSTeS and MMBIR models, the absence of a homologous template during HDR possibly results in the activation of microhomology pairing repair of broken ends, a more error-prone DNA repair. Thus, these factors need to be carefully considered during the application of such techniques to therapeutics.

Experimental model and study participant details
C57BL/6NCrSlc mice (Japan SLC, Shizuoka, Japan) were used in this study. The animals were housed at a constant temperature (22 ± 2 °C) and humidity (50 ± 10%), with a 12-h light/12-h dark cycle. All animal experiments were approved by the Institutional Animal Care and Use Committee of Chubu University (Permit Number #202110033) and were conducted in accordance with the institutional guidelines.

CRISPR RNP and ssODN preparation
The CRISPR guide RNAs were designed using CHOPCHOP 23 were incubated at 25 °C for 10 min to form the RNP complex. The ssODNs were manufactured by Eurofins Genomics (Tokyo, Japan) and were designed to join two DNA sequences so that the junction would be positioned at the center of the predicted cleavage sites, which were located within 3 bp of the PAM sequences (Table S3). The 5′ and 3′ ends of the ssODNs were protected with two consecutive phosphorothioate-modified bases (*) to improve the HDR efficiency 12 (Table S3).

Analysis of CRISPR/Cas9-engineered mice
To screen for CRISPR/Cas9-induced mutations, genomic DNA was isolated from the tails or ears of founder mice using lysis buffer [100 mM NaCl, 200 mM sucrose, 10 mM ethylenediaminetetraacetic acid, 300 mM Tris (pH 8.0), and 1% sodium dodecyl sulfate], and DNA was examined by PCR amplification. The obtained PCR products were purified using NucleoSpin Gel and PCR Cleanup kit (Takara Bio, Shiga, Japan) and sequenced directly or were cloned into the pTAC-1 vector (Biodynamics, Tokyo, Japan), and the sequences of individual clones were determined using Sanger sequencing (Eurofins Genomics). The PCR primers used for genotyping are listed in Table S4.

RT-PCR
RT-PCR was performed using total RNA. Total RNA was isolated from ear tissue using ISOSPIN Cell & Tissue RNA (Nippon Gene, Tokyo, Japan). Template cDNA was obtained using ReverTra Ace qPCR RT Master Mix (Toyobo, Osaka, Japan). The RT-PCR products were treated with NucleoSpin Gel and PCR Cleanup kit (Takara Bio) 1 3 and directly analyzed using Sanger sequencing (Eurofins Genomics). The primers used for RT-PCR are listed in Table S4.

Food intake analysis
Mice were housed individually, and food intake was measured in terms of grams of diet consumed per day.

Mating test
Upon sexual maturation, male mice were caged with two females for at least 8 weeks.
During the mating test, pups were counted for litter size measurements and their tails or ears were biopsied for genotyping.

Quantification and statistical analysis
The Student's t-test (two-tailed test) was used for body weight, food intake, and mating analyses. Data are presented as the mean ± standard deviation. For Mendelian genotype ratios of progeny obtained from sibling mating between CCRs mice, the chi-square test was performed using Excel version 16.36 (Microsoft, Redmond, WA, USA). Statistical comparisons were made using Tukey's honestly significant difference test. The threshold for statistical significance was P < 0.05.