Enhancement of Repeat-Mediated Deletion Rearrangement Induced by Particle Irradiation in a RecA-Dependent Manner in Escherichia coli

Simple Summary Repeat-mediated deletion (RMD) is generally induced by a DNA double-strand break (DSB), but the effect of DSB complexity on RMD initiation is unclear. In order to address this issue, this research firstly used an Escherichia coli reporter line in which amp restoration was controlled by lacI repeats. Then, particle irradiation was used to stochastically generate complex DSBs. The results clearly confirmed the enhancement of RMD rearrangement induced by proton and carbon irradiation in a dose- and LET-dependent manner. Meanwhile, RMD rearrangement was suppressed by intermolecular homology, in which the length of homology is more important than the composition of homology. Exogenous recombinase could significantly promote particle irradiation-induced RMD events. The RecA-dependent pathway was suggested to be involved in the enhancement of RMD under particle irradiation. These results could broaden our understanding of RMD generation and help assess the repair process of complex DSBs located around the repeat sequences. Abstract Repeat-mediated deletion (RMD) rearrangement is a major source of genome instability and can be deleterious to the organism, whereby the intervening sequence between two repeats is deleted along with one of the repeats. RMD rearrangement is likely induced by DNA double-strand breaks (DSBs); however, it is unclear how the complexity of DSBs influences RMD rearrangement. Here, a transgenic Escherichia coli strain K12 MG1655 with a lacI repeat-controlled amp activation was used while taking advantage of particle irradiation, such as proton and carbon irradiation, to generate different complexities of DSBs. Our research confirmed the enhancement of RMD under proton and carbon irradiation and revealed a positive correlation between RMD enhancement and LET. In addition, RMD enhancement could be suppressed by an intermolecular homologous sequence, which was regulated by its composition and length. Meanwhile, RMD enhancement was significantly stimulated by exogenous λ-Red recombinase. Further results investigating its mechanisms showed that the enhancement of RMD, induced by particle irradiation, occurred in a RecA-dependent manner. Our finding has a significant impact on the understanding of RMD rearrangement and provides some clues for elucidating the repair process and possible outcomes of complex DNA damage.


Introduction
Chromosome rearrangement is the main cause of genetic instability [1] and is more likely due to repetitive sequences, which are widespread in the genomes of prokaryotic and eukaryotic organisms.For a long time, genome repeat sequences have been considered as "junk DNA"; however, current evidence indicates that variations in repeats can alter the expression of genes, while changes in the number of repeats have been linked to certain human diseases [2].The nature of genomic variation caused by repeats depends on their relative orientation, whereby repetitive sequences are classified into two types: one is direct repeats, which are oriented in the same direction, while the other is inverted repeats, which are oriented opposite each other [3][4][5].Direct repeats result in deletions and duplications, whereas inverted repeats might cause inversions in the regions flanked by them [6,7].
In bacterial genomes, direct repeats are more common than inverted repeats, and the systematic investigation of direct sequence recombination has provided some insight into the molecular mechanisms involved in genomic rearrangement, whereby repeat-mediated deletions (RMDs) are a type of chromosomal rearrangement between two homologous sequences which causes the sequence between the repeats to be lost, along with one of the repeats [8].In Escherichia coli (E.coli), RMD rearrangement between direct repeats occurs via both RecA-dependent and RecA-independent mechanisms, depending on the size of the repeats and on the intervening sequences between the repeats [9,10].RecA-dependent recombination is the main mechanism that maintains genomic integrity, whereby the single-strand DNA (ssDNA) tracts that have formed at break sites are used to search for a homologous double-strand DNA to use as a template to repair the break [11,12], while RecAindependent sequence rearrangement contains three major mechanisms: simple-strand annealing, sister chromosome slipped misalignment, and single-strand annealing [13,14].RMD might be generally associated with DNA breaks [15].Each cell in a living organism is constantly exposed to various detrimental factors including physical, chemical, and biological factors which could cause dangerous damage to their genetic carrier DNA.Among those factors, radiation is a well-accepted DNA damage inducer; however, the phenomena and mechanisms of radiation-induced rearrangement of repetitive sequences have not been expansively investigated.Using the bacterium Deinococcus radiodurans, γ-irradiation induced a deletion of 438 bp direct repeats, which was suggested to be RecAindependent and in which single-strand annealing (SSA) was the driving force promoting the occurrence of DNA RMD events [14].Whereas recombination between chromosomal and plasmid DNA has been shown to be strictly dependent on the RecA and RecF proteins, it has also been proven that the proximity of DSBs relative to repeat sequences could affect deletion rearrangement.Moreover, in a recent study, RMD rearrangement was induced with two DSBs: the 5 DSB that was just downstream from the first repeat and the 3 DSB that was at a considerable distance upstream of the second repeat.These data confirmed that an increase in the 3 DSB/repeat distance from 3.3 kb to 28.4 kb only caused a modest decrease in rearrangement frequency, meaning that deletion can be induced by a chromosomal break that is located far away from a repeat [16,17].Therefore, this finding inspired us to posit that the features of DSBs could deeply influence this event.However, compared to the proximity of a DSB, the complexity of the DSB represents a more critical feature, thus elucidating that the complexity of DSBs in RMD rearrangement will better facilitate the understanding of the mechanism involved in repeat-related rearrangements.
DNA damage outcomes are determined by radiation quality, whereby low linear energy transfer (LET) irradiation, such as X-and γ-rays, generates low-complexity DNA damage, whereas high LET irradiation, such as protons and heavy ions, causes high-complexity DNA damage [18].Complex damage comprises at least two DNA lesions within one helical turn of DNA, while a clustered DSBs represents the most severe form of complex DNA damage and poses a serious challenge for DNA damage repair.Additionally, its consequences are more likely to cause gross genomic rearrangement.As a special type of rearrangement, the performance of RMD under high-LET irradiation attracts our interest.Hence, by utilizing particle irradiation, such as proton and carbon irradiation, different complexities of DSBs could be generated.Therefore, this work intends to investigate the mechanisms of DSB complexity and their impact on RMD rearrangement by using an E. coli reporter strain, which is well established for studying the performance of repeat sequences under high LET radiation circumstances.

Establishment of lacI Direct Repeat Reporter Strain and Mutant Strain of E. coli
All constructs, sgRNAs, primers, and N20 sequences followed by the PAM used in this study are presented in Table S1 in the Supplemental Materials.All plasmids listed in Supplementary Table S1 were constructed using the Golden Gate method [20].Plasmids and genomic DNA were extracted using the AxyPrep kit (Corning) according to the manufacturer's instructions.DNA fragments were PCR-amplified using KOD-Plus-Neo polymerase.Restriction endonucleases BsaI and T4 DNA ligase were purchased from Thermo Scientific.The primers listed in Supplementary Table S2 were designed using the web-based software J5 DeviceEditor (version number:1.0.3,URL: https://j5.jbei.org/,accessed on 23 October 2019) to create linear dsDNA products [20].
The reporter strain and deletion mutants (∆recA and ∆recB) were constructed using a two-plasmid-based CRISPR/Cas9 genome editing system, and the recombineering method was performed according to a previously published protocol [21].Briefly, for reporter strain construction, we inserted the sequences as the N20 sequences of guide RNA (gRNA) within the lacZ and mhpR genes, and as knock-in sequences in plasmid-encoded single-guide RNA (psgRNA) within the the amp and lacI genes, which resulted in the construction of plasmids of plasgRNA and pmlsgRNA, respectively.The MG1655 strain was first transfected with plasgRNA and pRedCas9, and 2 g/L arabinose was added to induce Cas9 expression at 30 • C.After the knock-in strains were screened by PCR amplification and DNA sequencing, the plasmids were removed through multiple-generation cultures at 37 • C. The resulting strain was further transfected with both pmlsgRNA and pRedCas9.After conducting similar induction and screening processes, a transgenic strain harboring the lacI:lacI:lacO::amp gene was established and named MG1655 TR .Similarly, deletion mutants (∆recA and ∆recB) were induced into MG1655 TR using prasgRNA with the RecA gene and prbsgRNA with the RecB gene as N20 sequences.We also constructed another series of psgRNA-derived plasmids, known as ph1sgRNA, ph2sgRNA, ph3sgRNA, and ph4sgRNA, which contained overlapping homologous segments with the lacI gene sequence.Here, the length of homology for ph1sgRNA was 213 bp and targeted the lacI gene from +114 to +326; the length of homology for ph2sgRNA was 159 bp and targeted the lacI gene from +512 to +670; the length of homology for ph3sgRNA was 268 bp and targeted the lacI gene from +945 to +1212; and the length of homology for ph4sgRNA was 1212 bp, targeting the lacI gene from +1 to +1212.The psgRNA plasmid was used as the blank control.

Particle Irradiation of E. coli Strains
The irradiation experiment was performed according to a previous study.In detail, a single colony of the aforementioned E. coli strains was added to 5 mL of LB medium in a sterile 50 mL centrifuge tube and incubated at 37 • C with shaking at 200 rpm for 6 h, or until an OD600 value of approximately 1 × 10 8 colony-forming units (CFUs)/mL was reached.Then, 1 mL of the culture was transferred to a 35 mm Petri dish (approximately 0.3 mm in depth) and subjected to proton or carbon irradiation.Proton irradiation was performed at the Beijing Tandem Accelerator Nuclear Physics National Laboratory, China, with an energy of 22 MeV and LET of 2.61 keV/µm.Carbon irradiation was performed at the Heavy Ion Research Facility in Lanzhou, China, using 80 MeV/nucleon with a LET of 31.46 keV/µm.

Measurement of E. coli Survival Rate in Response to Particle Irradiation
The irradiated bacteria (five independent samples for each irradiated treatment) were diluted and plated on LB plates, then incubated overnight at 37 • C. The number of colonies was counted to determine the CFUs.Survival rate was calculated by dividing the CFUs of the irradiated cultures by the CFUs of the culture without irradiation (0 Gy group or CT group).The final data represent the average of five independent experiments.

Detection of Ampicillin (Amp)-Resistant Clones
The irradiated bacteria were diluted and plated on control LB agar plates, with the selection plates containing 100 µg/mL of ampicillin (Amp), and then incubated overnight at 37 • C. CFUs were counted manually using a dissecting microscope (ES-18BZL, Motic, Xiamen, China).The Amp R clones were further verified by collecting clones grown on Amp selection plates, which were then singly cultured in an LB liquid medium containing 100 µg/mL ampicillin.The mutation rate was calculated by dividing the number of Amp R clones on the Amp selection plates by those on the control LB plates.The final data represent the average of five independent experiments.

Amplification and Sequence Analysis of lacI Gene
Verified Amp R clones were collected from five independent experiments for further PCR analysis; mutant clones were suspended in 100 µL H 2 O and used as the PCR template to amplify the lacI gene.In the amplification reaction (total volume: 25 µL), 5 µL of the template of mutant clones (100-500 ng DNA), 0.1 units Taq DNA polymerase mix, 200 µM dNTP, and 10 pmol of each primer were added and incubation was performed according to the manufacturer's instructions.The primers are listed in Supplementary Table S2.The PCR product for further DNA sequencing was obtained by Sangon Biotech Company (Shanghai, China).

Statistical Analysis
The results for E. coli survival rate and mutation rates of Amp R clones are presented as the mean ± standard deviation.Differences between the two groups were evaluated by Student's t-test.Differences in Amp R distribution were evaluated by the chi-square test using SPSS.Significance was considered at p < 0.05 and is indicated as * for p < 0.05 and ** for p < 0.01.

Enhancement of RMD Rearrangement of LacI Repeats by Particle Irradiation
To investigate RMD rearrangement under particle irradiation, E. coli K12 MG1655 (MG1655 TR ) was used.In this strain, an ampicillin-resistance-based reporter assay was developed.It uses two lacI direct repeats which are separated by a 405 bp spacer, located upstream of E. coli lac operon (lacO) site.This leads to the Amp gene being constitutively repressed by the regulatory lacI gene encoding the lacI repressor in combination with the lacO operator in the promoter.Therefore, the restoration of Amp resistance (Amp R ) results from a loss of function by the lacO or/and two lacI genes.Based on the integrity of the two lacI gene repeats, the restoration of Amp resistance (Amp R ) was classified into three types: type I implied that both lacI genes maintained their structural integrity together with synchronous mutations on the two lacI or/and on lacO; type II means that only one lacI remained due to RMD rearrangement and mutation on lacI or/and on lacO; type III was a large fragment deletion or complex structural variation that occurred on lacI, as schematically shown in Figure 1A.Subsequently, the MG1655 TR strain was subjected to different doses of proton and carbon irradiation, respectively.DSB complexity was determined by radiation doses and radiation quality.Survival rates were firstly determined and shown in Figure 1B, particle irradiation could significantly decrease the survival rate of E. coli in both dose-dependent and LET-dependent manners (in all cases, p < 0.01).Further, both particle irradiations led to an increased proportion of Amp R clones compared to the control group (0 Gy) (in all cases, p < 0.01), also in a dose-and LET-dependent manner, as shown in Figure 1C.Next, the restoration pattern of Amp R activation was explored.Firstly, 100 Amp R clones generated in the 0 Gy group during five independent experiments were analyzed.It was shown that 96% of the Amp R clones were classified as type I, 3% as type II, and 1% as type III.Therefore, under spontaneous DSB conditions, the activation of Amp R is mainly due to small point mutations in both lacI gene repeats or/and lacO.However, particle irradiation caused a significant reduction in the proportion of type I and an increase in the proportion of type II, from 3% (3/100, CT) to 46.6% (55/118, 50 Gy proton irradiation), 51.9% (79/152, 100 Gy proton irradiation), 57.7% (45/78, 200 Gy proton irradiation), 61.1% (115/188, 50 Gy carbon irradiation), 66% (138/209, 100 Gy carbon irradiation), and 70.2% (111/158, 200 Gy carbon irradiation).The results also showed that particle irradiation tended to enhance the RMD rearrangement of lacI repeats in a dose-and LET-dependent manner, as shown in Figure 1D.Meanwhile, the prevalence of type III increased synchronously, especially under carbon irradiation, from 1% (1/100, 0 Gy) to 6.9% (13/188, 50 Gy), 10% (21/209, 100 Gy), and 10.7% (17/158, 200 Gy).Considering the LET of proton and carbon irradiations, an increase in types II and III might be due to the presence of more complex DNA damage.
Here, the main classification standard for type II was the occurrence of RMD, with only one lacI remaining; however, the activation of the amp gene requires the loss of function of the remaining lacI or/and lacO, so the mutation spectrum of type II for 0 Gy and 100 Gy proton and carbon irradiation was further analyzed.The results showed that 66.7% (2/3) of Amp R clones in the 0 Gy group were lacO mutations and base pair substitutions, while the remaining 33.3% (1/3) were located in lacI following a base insertion, as shown in Table 1.Analysis of 100 Gy proton irradiation revealed that 77.6% (45/58) of the Amp R clones were lacO mutations, and the remaining 22.4% (13/58) were located in lacI.All the lacO mutations were base pair substitutions, while the lacI mutations were dominated by single insertions (61.5%, 8/13) followed by single base pair substitution mutations (38.5%, 5/13), as shown in Table 2.For 100 Gy carbon irradiation, 49.1% (52/106) of Amp R clones were lacO mutations and base pair substitutions, while the remaining 50.9% (54/106) were located in lacI.The spectrum was dominated by single base pair substitutions (57.4%, 31/54), followed by single insertions/deletions (22.2%, 12/54) and >4 bp large fragment deletions or insertions (20.4%, 11/54), as shown in Table 3.
Table 1.Type II mutation spectrum of lacO and lacI genes in Amp resistance (Amp R ) clones collected from 0 Gy group."→" represents substitution; "+" represents insertion.

Site Mutation
Site Mutation  decrease the survival rate of E. coli in both dose-dependent and LET-dependent manners (in all cases, p < 0.01).Further, both particle irradiations led to an increased proportion of Amp R clones compared to the control group (0 Gy) (in all cases, p < 0.01), also in a dose-and LETdependent manner, as shown in Figure 1C.Next, the restoration pattern of Amp R activation was explored.Firstly, 100 Amp R clones generated in the 0 Gy group during five independent experiments were analyzed.It was shown that 96% of the Amp R clones were classified as type I, 3% as type II, and 1% as type III.Therefore, under spontaneous DSB conditions, the activation of Amp R is mainly due to small point mutations in both lacI gene repeats or/and lacO.Further, we attempted to determine whether the mutation in the lacI gene was generated alongside the occurrence of RMD.Thus, the Amp R type I clones were further sequenced in an attempt to provide some clues for elucidating this issue.In type I Amp R clones in the 0 Gy group, only 3.2% (3/94) of mutations were located in both lacI repeats, whereas their number was 12.1% (8/66) in the 100 Gy proton irradiation group and 28% (14/50) in the 100 Gy carbon irradiation group, respectively, as shown in Supplementary Table S3.Interestingly, the mutations distributed in the two lacI genes were totally identical, Biology 2023, 12, 1406 8 of 16 which possibly suggested that multiple base damages occurred in the lacI repeats, alongside the small point mutation generated during the DNA damage repair process.When homologous recombination is used to repair DSBs located in the spacer sequence between lacI repeats or flanking sequences, the mutant lacI repeat might be used as the template, which leads to the mutation being copied into the later-repaired lacI repeat, which is to say that in type I, the lacI gene might undergo an intact recombination, which would provide some clues for the time course of mutation generation in lacI and RMD rearrangement of lacI, thus suggesting that mutation in the lacI gene occurs prior to RMD and the possibility that the mutation might be associated with HR repair was also excluded.

Suppressive Effect of Intermolecular Homology on RMD Frequency
In the previous results, the enhancement of RMD rearrangement induced by particle irradiation mainly occurred between two intramolecular lacI repeat sequences.Therefore, it raised a question about the effect of intermolecular homology on radiation-induced RMD.To investigate this issue, we artificially transfected four plasmids (ph1sgRNA, ph2sgRNA, ph3sgRNA, and ph4sgRNA) harboring various lengths and compositions of lacI homology into MG1655 TR system, respectively, in addition, psgRNA which harbored no lacI homology was set as blank control.The schematic is shown in Figure 2A.Then, these five plasmid-harboring systems were subjected to 100 Gy of either proton or carbon irradiation.The results showed that particle irradiation decreased cell survival rates, although no significant changes were observed in the presence of intermolecular homology (p > 0.05, compared to psgRNA, in all cases) (Figure 2B).However, the increased mutation rates induced by particle irradiation without homology (psgRNA) were significantly changed under the presence of intermolecular homology, as shown in Figure 2C.In the MG1655 TR (ph2sgRNA) strain, the mutation rate was similar to that of the MG1655 TR (psgRNA) strain either for proton or carbon irradiation (p > 0.05, in both cases, compared to psgRNA).However, the mutation rate for MG1655 TR (ph3sgRNA) was significantly increased under carbon irradiation (p < 0.01) and slightly increased under proton irradiation (p > 0.05).Conversely, both ph1sgRNA and ph4sgRNA significantly repressed the amp gene activation induced by particle irradiation (in all cases, p < 0.01), especially for ph4sgRNA, which contained the whole length of the lacI homology and an extremely low mutation rate was observed (Figure 2C).
Next, we analyzed the effect of intermolecular homology on RMD.The mutation types of Amp R clones generated from five cells harboring different plasmids under particle irradiation were molecularly analyzed.For ph1sgRNA, ph2sgRNA, and ph3sgRNA, RMD frequencies in the CT group were identical, and only slightly higher than psgRNA, but with no significant difference (in all cases, p > 0.05).However, for ph1sgRNA, ph2sgRNA, and ph3sgRNA, enhancement trends in RMD frequency induced by proton irradiation were slightly suppressed compared to psgRNA (p > 0.05, in all cases), whereas they were significantly suppressed under carbon irradiation compared to psgRNA (p < 0.01 for ph1sgRNA and ph2sgRNA, p < 0.05 for ph3sgRNA) (Figure 2D).Furthermore, for ph4sgRNA, which harbored the whole lacI homology, enhancement trends in RMD frequency induced by particle irradiation were totally suppressed-3.3%versus 50% for proton irradiation and 4.7% versus 63.3% for carbon irradiation (p < 0.01, in both cases).This observed suppressive effect might be initiated by intermolecular homology which could compete with intramolecular lacI repeats during DSB repair.Therefore, this competition can lead to a decrease in the frequency of RMD rearrangement between intramolecular lacI sequences.Considering the significant differences in composition between ph1sgRNA, ph2sgRNA, and ph3sgRNA, and the differences in length between ph1sgRNA and ph4sgRNA, the competitive effect was regulated by the compositions and lengths of the intermolecular homology.
observed suppressive effect might be initiated by intermolecular homology which could compete with intramolecular lacI repeats during DSB repair.Therefore, this competition can lead to a decrease in the frequency of RMD rearrangement between intramolecular lacI sequences.Considering the significant differences in composition between ph1sgRNA, ph2sgRNA, and ph3sgRNA, and the differences in length between ph1sgRNA and ph4sgRNA, the competitive effect was regulated by the compositions and lengths of the intermolecular homology.

Stimulative Effect of Exogenous Recombinase on RMD Initiated by Particle Irradiation
RMD rearrangement events often result from homologous recombination during the DSB repair process.Therefore, we further artificially increased cell recombination capacity to examine the effect of exogenous recombinase on RMD events induced by particle irradiation.The MG1655 TR strain was transfected with the pRedCas9 plasmid, which constitutively expressed λ-Red recombinase under arabinose induction (MG1655 TR Red + ).Conversely, the MG1655 TR strain was transfected with the pRedCas9 plasmid without arabinose induction (MG1655 TR Red − ) and used as the control.The schematic is shown in Figure 3A.Next, they were subjected to 100 Gy proton or carbon irradiation, respectively.The results showed that exogenous λ-Red recombinase expression (MG1655 TR Red + ) significantly improved the survival rates of E. coli cells under either proton or carbon irradiation compared to MG1655 TR Red − (p < 0.05 for proton irradiation, p < 0.01 for carbon irradiation), as shown in Figure 3B.Mutation rates with the expression of the recombinase were significantly decreased under both proton and carbon irradiation (in all cases, p < 0.01) (Figure 3C).We further collected Amp R clones to analyze the activation pattern of the amp gene, and the results showed that for the MG1655 TR Red − strains, 72 Amp R clones were analyzed in the CT group, and 94.4% of the Amp R clones were classified as type I, 4.2% as type II, and 1.4% as type III.In the proton irradiation group, 48.3% (29/60) were type I, 50% (30/60) were type II, and 1.7% (1/60) were type III.In the carbon irradiation group, the proportion of type I was 28.1% (16/57), 64.9% (37/57) were type II, and the rest (7%) were type III.However, for the MG1655 TR Red + strains, the CT group exhibited a slight increase in type II (5.8%, 7/120) and a decrease in type I (93.3%, 112/120), yet no significant difference was observed compared to CT group of MG1655 TR Red − strains.Particle irradiation, either by 100 Gy proton or carbon radiation, could notably increase the proportion of type II, from 50% to 78.1% in the proton irradiation group and from 64.9% to 78.9% in the carbon irradiation group (p < 0.01, in both cases).Thus, the participation of recombinase could significantly stimulate the enhancement of RMD frequency under particle irradiation.However, no significant difference was observed between proton and carbon radiation when analyzing the frequency of RMD rearrangement, thereby suggesting the existence of a threshold for RMD rearrangement.Nonetheless, carbon irradiation significantly promoted the occurrence of mutation type III (Figure 3D).These results suggest a stimulative effect of exogenous recombination, which is initiated by particle irradiation, on RMD rearrangement.

Role of the RecA-Dependent Pathway on Particle Irradiation-Induced RMD Events
In E. coli, RMD occurs via both RecA-dependent and RecA-independent mechanisms In the RecA-dependent pathway, DSB repair is initiated by the RecBCD enzyme; RecBCD resects DSB ends, degrading them into ssDNA, and loads the recombinase RecA to pro mote strand exchange [22].After confirming that exogenous λ-Red recombinase was in volved in RMD, we further determined the involvement of endogenous recombinase RecA

Role of the RecA-Dependent Pathway on Particle Irradiation-Induced RMD Events
In E. coli, RMD occurs via both RecA-dependent and RecA-independent mechanisms.In the RecA-dependent pathway, DSB repair is initiated by the RecBCD enzyme; RecBCD resects DSB ends, degrading them into ssDNA, and loads the recombinase RecA to promote strand exchange [22].After confirming that exogenous λ-Red recombinase was involved in RMD, we further determined the involvement of endogenous recombinase RecA in the induction of RMD following particle irradiation.Here, ∆recA and ∆recB mutants were adopted by knocking out the RecA or RecB genes.The results showed that ∆recA and ∆recB promoted a significant decrease in the survival rate of E. coli after proton or carbon irradiation, whereby only 7% and 6% of the cells could survive under 100 Gy proton irradiation for ∆recA and ∆recB, respectively, with survival rates of 5% and 4% for cells undergoing carbon irradiation (in all cases, p < 0.01, compared to the wild type), as shown in Figure 4A.Additionally, analysis of the mutation rates showed that mutation rates in ∆recA and ∆recB mutants were significantly reduced by one order of magnitude compared to that in wild-type (Figure 4B).Furthermore, the mutation rate of ∆recA mutants did not increase following particle irradiation; on the contrary, proton irradiation led to a reduction in mutation rate (p < 0.01, compared to CT).For the ∆recB mutant, the background mutation rate is as low as in ∆recA; however, the difference is that particle irradiation resulted in a significant LET-dependent increase compared to the control (p < 0.05 for proton irradiation, p < 0.01 for carbon irradiation) (Figure 4B).By combing the extremely low survival rate, it is possible that an overkill effect might play an important role in the low mutation rates observed in the ∆recA and ∆recB mutants.Next, we analyzed 60 Amp R clones which had been generated by spontaneous mutations in ∆recA and found that 96.6% of Amp R clones in the CT group were classified as type I, 1.7% as type II, and 1.7% as type III, which did not significantly differ from the wild-type CT group.For ∆recA, a total of 25 Amp R clones generated by 100 Gy proton irradiation and 25 Amp R clones generated by 100 Gy carbon irradiation showed a slight enhancement in type II compared to the CT group-from 1.7% to 4% for proton irradiation and also to 4% for carbon irradiation.However, this degree of enhancement was much lower than that in the wild-type group (from 3% to 51.9% for proton and to 66% for carbon irradiation in the wild-type group) and exhibited significant differences (in both cases, p < 0.01, compared to its corresponding wild-type group) (Figure 4C).These results suggested that enhancements in RMD rearrangement induced by particle irradiation are RecA-dependent.In addition, particle irradiation could cause an enhancement in RMD frequency in ∆recB either for proton irradiation or carbon irradiation (in both cases, p < 0.01, compared to the corresponding CT group), yet the degree of enhancement was much lower than that in the wild-type group (in both cases, p < 0.01, compared to its corresponding wild-type group), which further supported the role of RecA in enhancing RMD rearrangement.In addition, under the circumstance of RecA and RecB mutation, the frequency of type III induced by particle radiation significantly increased, which suggested that the absence of a recombination mechanism might lead to the activation of other repair pathways such as alternative end joining, which is more prone to structural mutations such as larger fragment deletion and translocation.group (in both cases, p < 0.01, compared to its corresponding wild-type group), which further supported the role of RecA in enhancing RMD rearrangement.In addition, under the circumstance of RecA and RecB mutation, the frequency of type III induced by particle radiation significantly increased, which suggested that the absence of a recombination mechanism might lead to the activation of other repair pathways such as alternative end joining, which is more prone to structural mutations such as larger fragment deletion and translocation.

Discussion
Particle irradiation can cause multiple types of DNA damage, among which DSB is the most serious form.However, cells have also evolved a variety of repair pathways to deal with DSBs, among which non-homologous end joining repair (NHEJ) and homologous recombination (HR) repair are the most important DSB repair pathways [23].NHEJ, which contains two subpathways (canonical NHEJ (c-NHEJ) and alternative end joining (a-EJ)), religates broken ends without requiring a template [24].Homologous recombination (HR), includes gene conversion, break-induced replication, and single-strand annealing, and uses a homologous donor for repair, which can potentially lead to unexpected rearrangements, such as RMD [25].RMD is mainly caused by DSBs; site-specific enzyme digestion meth-ods (such as IsceI endonuclease) or the CRISPR/Cas system are commonly used to create DSBs [14,16].The advantage of the aforementioned methods is that they can generate single or multiple DSBs at any specific site; in contrast, ionizing radiation (IR) inflicts a wide spectrum of DNA lesions, including DSBs, that are randomly dispersed throughout the genome and pose greater challenges to DNA damage repair [26][27][28][29].The most well-accepted characteristic quality parameter of IR is linear energy transfer (LET), with values varying from ~1 keV/µm (low-LET: X, γ-rays) to higher (~10 keV/µm for protons or 10-100 keV/µm for carbon ions) and very high values (>100 keV/µm for alpha particles or heavy-charged particles [30]).The signature of high-LET radiation comprises closely spaced DNA lesions that form a complex cluster of DNA damage, including strand breaks, oxidatively generated base damage, and abasic sites within one or two DNA helical turns [31,32].However, repairing this complex DNA damage has always been challenging and poorly understood.Therefore, this study focused on RMD rearrangement caused by complex DNA damage following particle irradiation with the aim of providing some clues for elucidating the repair process and possible outcomes of complex DNA damage.First, a reporter strain was used in which amp gene activation regulated by lacI repeat sequences.Activation of the amp gene required either lacO inactivation or the simultaneous inactivation of the two lacI genes.According to the integrity of the two lacI repeats, simultaneous inactivation of two lacI repeats was classified into three types.Here, the focus was on type II, in which RMD rearrangement occurred between the two lacI genes.Our results demonstrated that particle irradiation could significantly activate the amp gene, which was mainly due to an increase in type II, i.e., the enhancement of RMD.
In this study, the frequency of spontaneous RMD rates was at a relatively low level, approximately 10 −7 (Figure 1C,D), which was not consistent with previous works, which demonstrated a background RMD rate of 10 −5 -10 −4 [6].The special requirement of amp activation in the present experimental system might provide an explanation for this difference; the necessity of amp activation was not only due to the occurrence of RMD, but was also accompanied by the loss-of-function of residual lacI or lacO, which indicated that more than one DNA damage event would be generated at the lacI site and nearby if the amp gene was activated.Therefore, unlike other experimental systems, only the generation of RMD initiated by single DSB could be detected.Generally, IR induces its major lethal or mutagenic effects not solely by the presence of simple DNA damage, like single-strand breaks (SSBs), base lesions, and even a DSB, but primarily by the induction of clustered spatiotemporal damage to DNA, especially following high-LET radiation [30].This difference in RMD induction was considered to be the main advantage of our experimental system, which was suitable for better understanding RMD induced by complex DNA damage generated by high-LET particle irradiation.Meanwhile, the enhanced frequency of RMD rearrangement exhibited an LET-dependent manner (Figure 1D); the proportion and complexity of clustered DNA lesions increase with the increase in LET [32].Considering the difference in LET between proton and carbon radiation, high-LET carbon irradiation is more complex in inducing complex DNA damage forms which lead to a significant increase in type II and type III Amp R clones.From the mutation spectrum of type I and type II, especially the symmetrical distribution of point mutations that occur on the lacI gene, it can be inferred that some non-DSB damage accumulated in or around the repeat sequence, leaving mutations in the repeat sequence during the repair process.Then, in the later stage of using HR to repair DSB, these mutations would be left in the lacI gene after RMD or copied to another repeat sequence through gene conversion [33].Although most mutations in type II and type I were base pair substitutions (Tables 1-3), which seemingly contradicts the traditional concept that high-LET radiation can lead to complex mutation forms, such as larger fragment insertions, deletions or translocations, this is mainly due to our classification of Amp R activation types; complex translocation or structural variation were more likely to be generated in type III.
The rate of DNA recombination rearrangement between repeats is linearly dependent on substrate length [34].Long DNA repeats, of at least 100 nucletotides, can lead to intrachromosomal rearrangements by acting as substrates for the bacteria recombination machinery [35].In this study, the length of lacI was 1212 bp, which facilitated intermolecular recombination.However, the existence of the exotic plasmid-based homologous lacI sequence with different lengths and compositions exhibited various impacts on the frequency of RMD rearrangement between the two intramolecular lacI repeats.In terms of the aspect of the Amp R activation rate, a sequence that is homologous upstream of lacI has a stronger ability to suppress Amp R activation, whereas an homologous sequence downstream of lacI has a weaker ability to promote Amp R activation (ph1sgRNA versus ph2sgRNA versus ph3sgRNA); the longer the length of the homologous sequence, the stronger the ability to suppress Amp R activation (ph4sgRNA versus ph1sgRNA) (Figure 2C).Together, the analysis of the proportion of mutation types revealed that the enhancement of RMD was suppressed, especially for ph4sgRNA, which has a fragment homologous with the complete lacI sequence, indicating that the exogenous homologous fragment competitively provided more complete templates, which increased the opportunity to repair DNA damage in the genome using homologous fragments from the plasmids and reduced the probability of RMD occurring between repetitive fragments in the genome.
RMD involves homologous recombination (HR) between two repeat elements.The core step in HR is homology sampling, which is mediated by a conserved RecA-ssDNA nucleoprotein filament that registers an intact identical dsDNA molecule [1].Recombineering is a simple and efficient way to engineer DNA molecules in vivo by utilizing homologous recombination mechanisms.In Escherichia coli, the bacteriophage λ-Red homologous recombination system has been extensively used for genetic modification [36,37].In this study, the plasmid-based λ-Red recombineering system was adopted to create the reporter strains (MG1655 TR ) and also as an exogenous recombinase (MG1655 TR Red + ), in which the λ-Red recombinase gene was cloned into temperature-sensitive plasmids.Exogenous expression of λ-Red recombinase could improve the usage of homologous recombination repair, thereby leading to an increased survival rate, decreased mutation rate, and enhanced RMD rearrangement (Figure 3B).In contrast, by knocking out the RecA gene, the cell survival rate decreased significantly (Figure 4B) and the frequency of RMD was not enhanced under particle irradiation (Figure 4C) due to pronounced recombination and DNA repair deficiency, which strongly supports the conclusion that a RecA-dependent HR mechanism is involved in complex DSB repair.While knocking out the RecB gene could not suppress the enhancement of RMD induced by particle irradiation, the main reason might be that in E. coli cells deficient for RecBCD function (i.e., in RecB or/and recC null mutants), the RecF pathway may also take over recombination processes that involve dsDNA ends [38,39].

Conclusions
Overall, using a reporter line of lacI-repeat-controlled amp activation and taking advantage of complex DSB generation by particle irradiation, this study clearly demonstrated the influence of DSBs complexity on occurrence of RMD rearrangement and the regulation of intermolecular homology and exogenous recombinase.Meanwhile, the RecA-dependent pathway was suggested to be involved in the enhancement of RMD under particle irradiation.These results could broaden our understanding of RMD generation and help assess the repair process of complex DSBs located around repeat sequences.and P.B.; Supervision, L.S., P.B. and T.W.All authors have read and agreed to the published version of the manuscript.

Figure 1 .Figure 1 .
Figure 1.Particle irradiation enhanced the proportion of RMD rearrangement in Amp R activation clones in Escherichia coli.(A) Schematic diagram of the reporter system, in which the amp gene is activated through three types of patterns based on the integrity of two lacI repeats: type I is that both lacI genes maintain structural integrity together with synchronous point mutations on the two lacI or/and on lacO; type II is that only one lacI remained due to RMD rearrangement together with point mutation on lacI or/and on lacO; type III is large fragment deletion or complex structural variation occurring on lacI.Dashed box represents the loss of function of lacI or/and lacO.(B) Survival ratesFigure 1. Particle irradiation enhanced the proportion of RMD rearrangement in Amp R activation clones in Escherichia coli.(A) Schematic diagram of the reporter system, in which the amp gene is activated through three types of patterns based on the integrity of two lacI repeats: type I is that both lacI genes maintain structural integrity together with synchronous point mutations on the two lacI or/and on lacO; type II is that only one lacI remained due to RMD rearrangement together with point mutation on lacI or/and on lacO; type III is large fragment deletion or complex structural variation occurring on lacI.Dashed box represents the loss of function of lacI or/and lacO.(B) Survival rates of E. coli cells under particle irradiation (n = 5; **, p < 0.01).(C) Mutation rates of Amp resistance (Amp R ) induced by the indicated irradiation types (n = 5; **, p < 0.01).(D) Proportional distribution of the three types of Amp activation patterns in Amp R clones under particle irradiation, (**, p < 0.01, doses versus 0 Gy), 0 Gy (n = 100), 50 Gy proton (n = 118), 100 Gy proton (n = 152), 200 Gy proton (n = 78), 50 Gy carbon (n = 188), 100 Gy carbon (n = 209), 200 Gy carbon (n = 158).

Table 2 .
Type II mutation spectrum of lacO and lacI genes in Amp resistance (Amp R ) clones collected from 100 Gy proton irradiation group."→" represents substitution; "+" represents insertion.