Escape from antimicrobial CRISPR-Cas9 in E. coli ST131 depends on the genetic context of the target gene

11 Escherichia coli (E. coli) is a common bacterium in the human gut and an important cause of 12 intestinal and extraintestinal infections. Some E. coli sequence types (ST) are associated with 13 high pathogenicity. The Extraintestinal Pathogenic E. coli (ExPEC) ST131 is a globally distributed 14 multidrug-resistant human pathogen associated with urinary tract and bloodstream infections. 15 Antibiotic-resistant infections often lead to antibiotic treatment failure, underscoring the need of 16 developing alternative treatments. The highly selective antimicrobial potential of CRISPR -Cas9 17 has been demonstrated in a range of model organisms. However, the effectiveness of CRISPR -18 Cas9 in combating ST131 -associated infections and the consequences of CRISPR -Cas9 19 treatment, such as the emergence of escapers, remains unclear. 20 Here, we investigated the antimicrobial activity of CRISPR-Cas9 against ST131 and assessed the 21 frequency and genetic basis of escape. We conjugatively delivered CRISPR -Cas9 to ST131 22 isolates which carried cefotaxime-resistance-encoding target gene blaCTX-M-15 in the chromosome 23 and characterized escape subpopulations. Two main types of escapers emerged: blaCTX-M-15-24 positive escapers carried dysfunctional CRISPR-Cas9 systems and arose at a ~10-5 frequency. 25 Instead, blaCTX-M-15-negative escapers presented chromosomal deletions involving blaCTX-M-15 loss. 26 The frequency of blaCTX-M-15 loss depended on the blaCTX-M-15 genetic context. Specifically, blaCTX-M-27 15-negative escapers emerged at low frequency (~10 -5) in isolates where blaCTX-M-15 was located 28 downstream of insertion sequence (IS) ISEcp1, while escapers emerged with high frequency (~10-29 3) in isolates where blaCTX-M-15 was flanked by IS26. This work emphasizes how the genetic context 30 of target genes can drive the outcome of CRISPR-Cas9 tools, where the presence of IS 26 may 31 drive increased frequencies of escape. 32 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 2 IMPORTANCE 33 In the past decade CRISPR-Cas9 has emerged as a n efficient antimicrobial tool capable of 34 selective elimination of targeted bacteria. Even though it has been well described that bacteria 35 can evolve to escape targeting by CRISPR-Cas9, the mechanisms of bacterial escape and their 36 consequences remain largely elusive. In this study, we demonstrate the antimicrobial efficacy of 37 CRISPR-Cas9 against natural isolates of Escherichia coli ST131, a clinically relevant pathogen, 38 and elucidate the mechanism of escape from antimicrobial activity. We identify two distinct 39 mechanisms of escape, which involve either dysfunctional CRISPR -Cas9 activity, or loss of the 40 target gene (blaCTX-M-15), with the latter occurring at frequencies that depend on the genetic context 41 of the target gene. These findings provide important insights into the frequency and mechanisms 42 of bacterial escape from CRISPR -Cas9-based antimicrobials and offer a foundation for the 43 development of more effective treatments. 44 Word count abstract: 247 45 Word count text: 6060 46 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 3

47 Antimicrobial resistance (AMR) is a global health threat. By 2050, up to 8.22 million annual deaths 48 are predicted to be associated with AMR (1). The misuse and overuse of antibiotics has drastically 49 accelerated the emergence of AMR (2), which contributes to antibiotic treatment failure (3). In 50 this context, Escherichia coli is one of the most important pathogens, accounting for the highest 51 number of AMR-attributable deaths in 2019 (4). Therefore, it is recognized by the World Health 52 Organization (WHO) as a priority pathogen for which urgent development of new antimicrobials 53 is needed (5). 54 Extraintestinal Pathogenic E. coli (ExPEC) strains are among the most common Gram -negative 55 pathogens in humans (6). They are associated with numerous types of infections, including 56 urinary tract infections (UTIs), which can develop into bloodstream infections (7–9). Since 2000, 57 Sequence Type (ST) ST131 is the most common pandemic lineage in the clinic (8). The ST131 58 lineage represents multidrug-resistant pathogen s frequently associated with extended-59 spectrum beta-lactamases (ESBL), aminoglycoside and fluoroquinolone resistance, and several 60 virulence factors (7,9–11). CTX-M enzymes are among the mo st prevalent type of ESBL (12,13), 61 due to their global dissemination (12,14). The genes encoding these enzymes, blaCTX-M genes, are 62 found in both chromosomes and plasmids (14,15), often as part of highly mobile genetic 63 structures, surrounded by insertion sequences (IS), transposons and integrons (14). These 64 mobile genetic elements ( MGEs) not only contribute to AMR mobilisation but can also act as 65 promoter sequences, regulating expression levels of surrounding genes (14,16,17). Of these, IS 66 are the smallest self -mobilizing units (0.7-2.5 kB) that typically code for a single transposase 67 (Tnp), which allows their mobilisation (18,19). Several families of ISs, notably IS Ecp1 and IS26, 68 are widely distributed across E. coli ST131 genomes and are frequently located within or flanking 69 CTX-M gene structures (20–22). While ISEcp1 is commonly found as a single copy upstream of 70 the AMR gene (14), IS26 is often found in multiple copies of the same orientation that flank the 71 AMR gene, in what is known as a pseudo-compound transposon (PCT) (23,24). 72 ST131-associated infections are difficult to treat. In fact, higher antibiotic treatment failure rates 73 have been reported compared to non-ST131 infections (25,26). In an urgent need to find 74 alternative treatments, CRISPR-Cas9 can be used as a promising novel antimicrobial (27,28). In 75 natural populations, CRISPR -Cas systems act as a prokaryotic immune system against MGE 76 infections. S ince its discovery CRISPR -Cas9 in particular has been developed as a highly 77 sequence-specificity tool that can recognize a short (20 bp) specific DNA sequence and 78 subsequently cleave it (29). The tool combines a single-guide RNA (sgRNA) with a Cas9 nuclease; 79 the sgRNA recognises the target sequence while the Cas9 nuclease executes target cleavage by 80 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 4 introducing a blunt-ended double-strand DNA break (DSB) into the target genome (30). DSBs are 81 strong genotoxic lesions that lead to cell death (31). This highly specific antimicrobial effect of 82 CRISPR-Cas9 has been studied extensively over the past ten years (32–36). However, the 83 consequences of CRISPR -Cas9 treatment failur e and the long -term effects of its use remain 84 largely unexplored. 85 Here, we assessed the antimicrobial potential of CRISPR-Cas9 targeting blaCTX-M-15, the most 86 reported ESBL gene among E. coli ST131 (13). We used a modified version of the broad-host range 87 conjugative plasmid pKJK5::csg (37) to deliver CRISPR-Cas9, and programmed CRISPR-Cas9 to 88 target four different E. coli ST131 isolates derived from human stool samples (38), all of them 89 carrying a chromosomal copy of blaCTX-M-15. While strong antimicrobial activity was observed, we 90 found escapers of CRISPR -Cas9 targeting for all isolates. We observed two main types of 91 escapers, either presenting dysfunctional CRISPR -Cas9, or chromosomal rearrangements 92 leading to loss of the target gene. Interestingly, for the second type of escapers, we found a direct 93 impact of the blaCTX-M-15 genetic context (either surrounded by IS Ecp1 or IS 26) on the escape 94 frequencies, emphasizing the importance of the genetic context of a target gene on escape from 95 CRISPR-Cas9. 96 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 5

97 CRISPR-Cas9 targeting of a chromosomally encoded blaCTX-M-15 gene in human-associated E. 98 coli ST131 isolates causes high levels of sequence-specific killing 99 We used the broad-host range conjugative plasmid pKJK5 as a delivery vector of a CRISPR-Cas9 100 cassette (csgc: cas9, sgRNA, gfp and catB), either targeting blaCTX-M-15 (pKJK5::csgc[blaCTX-M-15]) or a 101 non-targeting control (pKJK5::csgc[NT]). We first verified successful delivery of pKJK5::csgc into 102 a recipient E. coli DH5α lacking blaCTX-M-15, and found no significant differences (p = 0.68) between 103 treatment and control (Figure 1 ), showing that the delivery efficiency of pKJK5::csgc is 104 independent of the sgRNA target and that the system can be acquired in the absence of a CRISPR-105 Cas9 target. 106 Next, we assessed CRISPR-Cas9 targeting activity in four E. coli ST131 isolates (Ecp1-I, Ecp1-II, 107 26-I and 26 -II); which all carry a single chromosomally encoded blaCTX-M-15 copy. Completed 108 genome sequences were generated for all four isolates to enable us to have an unambiguous 109 assessment of blaCTX-M-15 location and copy number. Across isolates, pKJK5::csgc[NT] achieved 110 significantly higher conjugation efficiencies (ranging from 15 ± 8% to 58 ± 11%)) than 111 pKJK5::csgc[blaCTX-M-15] (ranging from 0.002 ± 0.0007% to 0.062 ± 0.03%; p<0.001) (Figure 1). This 112 difference was attributed to the non-viability of transconjugants when CRISPR-Cas9 targets the 113 chromosomally encoded blaCTX-M-15. 114 115 Figure 1 : Conjugation efficiencies for either pKJK5::csgc[NT] (purple) or pKJK5::csgc[ blaCTX-M-15] ( white) across 116 different recipients : either E. coli DH5α, lacking blaCTX-M-15, or the respective E. coli ST131 isolates, all with a 117 chromosomally encoded blaCTX-M-15. Conjugation efficiencies were calculated dividing the number of transconjugants 118 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 6 by the number of recipients. The c onjugation efficiencies from the five replicates are represented as circles and the 119 colour represents the replicate number. The distinct colours assigned to each replicate in the pKJK5::csgc[blaCTX-M-15] 120 treatment facilitates tracking the replicate origin of the escapers throughout the manuscript. The means of the 121 conjugation efficiencies are represented as triangles ± standard deviation (bars). p-values <0.001 (***), non-significant 122 (ns). 123 While strong CRISPR-Cas9 antimicrobial activity was demonstrated , all isolates contained 124 subpopulations of E. coli ST131 transconjugants (escapers) able to survive acquisition of 125 pKJK5::csgc[blaCTX-M-15]. 126 Overall escape frequencies varied across isolates ( (4.53 ± 5.30) x 10 -5 to (4.23 ± 1.95) x 10 -3) 127 (Figure 2A). To better understand the genetic basis of escape and their relative contribution to 128 the overall escape frequencies, we determined cefotaxime resistance phenotypes for each of the 129 escapers, since resistance to cefotaxime is known to be conferred by CTX-M enzymes (15,39). 130 This showed that both cefotaxime-resistant and cefotaxime-sensitive escapers were found 131 across isolates (Supplementary figure 3), revealing that E. coli ST131 could escape from 132 CRISPR-Cas9 while either maintaining or losing the target gene. 133 134 Figure 2: Mean escape frequencies from the CRISPR -Cas9 treatment for the four E. coli ST131 isolates. A) Overall 135 escape frequencies independent of the cefotaxime resistance phenotype. These frequencies were calculated as the 136 relative conjugation efficiency between the conjugation efficiency of pKJK5::csgc[blaCTX-M-15] for each replicate and the 137 average conjugation efficiency of pKJK5::csgc[NT]. B-C) Cefotaxime-resistant (B) and -sensitive (C) escape 138 frequencies. The escape frequencies for each phenotype were determined as the relative conjugation efficiency 139 between the cefotaxime-resistant or cefotaxime-sensitive conjugation efficiencies of pKJK5::csgc[ blaCTX-M-15] for each 140 replicate and the average conjugation efficiency of pKJK5::csgc[NT]. The escape frequencies from the five replicates 141 are represented as circles and the colour represents the replicate number. The mean escape frequencies, which 142 include replicates with no escapers for a specific phenotype, are represented as triangles ± standard deviation (bars). 143 Replicate 4 from isolate Ecp1-II was excluded from cefotaxime-resistant and cefotaxime-sensitive escape frequencies 144 as only one escaper could be directly recovered from the filter mating assay. p-values <0.001 (***). 145 Cefotaxime-resistant escapers have dysfunctional CRISPR-Cas9 systems. 146 *** *** *** .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 7 Cefotaxime-resistant escape frequencies were generally low ( ~10-5) but showed significant 147 variation across isolates (Figure 2B; p<0.001). Post-hoc testing revealed this was due to isolate 148 26-I having a significantly lower escape frequency than all others (p<0.001). Nevertheless, the 149 escape frequencies for isolates 26 -I and 26 -II should be interpreted cautiously, as only few 150 escapers were detected, and their presence was inconsistent across replica tes (Figure 2B and 151 Supplementary figure 3). Therefore, in-depth analysis of cefotaxime -resistant escapers was 152 only performed for isolates Ecp1-I and Ecp 1-II. First, we sequenced the blaCTX-M-15 gene to 153 genetically confirm the cefotaxime -resistant phenotypes and found no mutations in the target 154 site or PAM. We then hypothesized that escapers likely survived CRISPR-Cas9 targeting through 155 acquiring mutations in cas9 and/or sgRNA (csgc) on the pKJK5::csgc[blaCTX-M-15]. To address this, 156 we assessed CRISPR-Cas9 integrity using PCR . All transconjugants from the pKJK5::csgc[NT] 157 control showed expected blaCTX-M-15 and csgc amplicons, but deletions, duplications, and partial 158 or complete lack of amplification within cas9 and/or sgRNA were found for 48% (Ecp1-I) and 64% 159 (Ecp1-II) of the escapers (Supplementary figure 4A). Furthermore, one Ecp1-II escaper exhibited 160 an IS150 transposition into cas9. While IS150 is present in both the donor (K-12 MG1655) and the 161 recipient (ST131 isolate Ecp1 -II) genome, the presence of a single nucleotide polymorphism 162 unique for IS150 in the K -12 genome suggested that the transposition into pKJK5::csg[ blaCTX-M-15] 163 occurred in the donor prior to its delivery to Ecp1 -II. The different mutations were distributed 164 across the cas9 and sgRNA sequence (Supplementary figure 4B-C), with no obvious patterns 165 revealed except for a hotspot in the α-helical lobe found in escapers from isolate Ecp1 -I 166 (Supplementary figure 4B) (40). 167 Those escapers that did not exhibit identifiable csgc mutations were subjected to a phenotypic 168 CRISPR-Cas9 functionality assay. We mated escapers with a donor bacterium bearing either 169 targeted plasmid pCTX15 (carrying the wildtype blaCTX-M-15) or untargeted plasmid pCTRL 170 (Supplementary figure 5A ) and calculated their relative conjugation efficiencies 171 (pCTRL/pCTX15). Cefotaxime -sensitive escapers lacking blaCTX-M-15 were used as potential 172 positive CRISPR-Cas9 controls. For both Ecp1-I and Ecp1-II most cefotaxime-sensitive escapers 173 demonstrated protection against pCTX15 ( Supplementary figure 5B-C). In contrast, no 174 protection was observed for cefotaxime-resistant escape rs (Supplementary figure 5B-C), 175 indicating dysfunctional CRISPR -Cas9 activity. Only one cefotaxime -resistant escaper (5.10 176 Ecp1-II) showed a relative conjugation efficiency for all the replicates compatible with functional 177 CRISPR-Cas9 activity (Supplementary figure 5C), which could indicate coexistence of the 178 wildtype blaCTX-M-15 and a functional pKJK5::csgc[blaCTX-M-15]. This escaper showed no mutations in 179 blaCTX-M-15, and short-read WGS did not reveal relevant mutations in pKJK5::csgc[blaCTX-M-15] that 180 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 8 could interfere with CRISPR -Cas9 activity . Therefore, coexistence could be attributed to an 181 unknown mechanism. 182 To summarize, we found that cefotaxime-resistant escape frequencies were ~10-5 across isolates 183 and all the escapers carried blaCTX-M-15 with no mutations in the protospacer or PAM sequence . 184 Using genotypic and phenotypic screening, we attributed cell survival to dysfunctional CRISPR-185 Cas9 systems. 186 Cefotaxime-sensitive escapers lost blaCTX-M-15 in a manner dependent on its genetic 187 context. 188 To understand the differential cefotaxime-sensitive escape frequencies found across isolates 189 (Figure 2 C), w e sought to understand the genetic context of blaCTX-M-15. Long-read genome 190 analyses of the ancestral untreated isolates revealed three different genetic contexts. In isolates 191 Ecp1-I and Ecp1-II, blaCTX-M-15 is found downstream of an insertion sequence (IS) ISEcp1, which 192 disrupts lacY in the lac operon (Figure 3A). In contrast, isolates 26-I and 26-II carry blaCTX-M-15 193 flanked by two IS 26 copies of the same orientation , in a pseudo-compound transposon (PCT) 194 (18,23). This blaCTX-M-15 PCT is found surrounded by either two or three additional IS26 copies 195 respectively, creating two slightly distinct overall IS26-contained sequences. These sequences 196 share the same genome location in both isolates , disrupting a gene of unknown function 197 (EC958_2451) (Figure 3B-C). Significant differences were found between cefotaxime -sensitive 198 escape frequencies of the three genetic contexts (p -values for each combination < 0.001) . 199 Isolates with an ISEcp1 genetic context showed significantly lower cefotaxime-sensitive escape 200 frequencies than isolate 26-I (4-copy IS26 context), which in turn had a significantly lower escape 201 frequency than isolate 26 -II (5-copy IS26 context). Overall, this indicates a significantly higher 202 escape frequency through loss of blaCTX-M-15 in isolates with an IS26 genetic context compared to 203 ISEcp1. Consequently, different approaches were employed to characterize the escapers based 204 on their genetic context. 205 206 207 208 209 210 211 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 9 212 Figure 3: Genetic context of blaCTX-M-15 (orange arrow) for the four E. coli ST131 isolates, ORFs and gene distances not 213 drawn to scale. A) Isolate Ecp1-I and Ecp1-II share the same ISEcp1 genetic context disrupting a lacY gene. The lacZ 214 gene is found 1.2 kB upstream of the ISEcp1. ISEcp1 is represented as a yellow arrow box. B-C) IS26 genetic contexts 215 for isolates 26-I (B) and 26-II (C), with a blaCTX-M-15 pseudo-compound transposon (PCT). For both isolates, the respective 216 overall IS26-contained sequences are found disrupting the gene with unknow function ( EC958_2451), shown in the 217 figure as 2451. The two Tn3, shown as pink boxes, represent a single Tn3 split into two truncated ORFs, likely due to 218 IS26 disruption that led to an inversion of one of the parts . IS26 is represented as a green arrow box. 2453 and 2456 219 represent the genes EC958_2453 and EC958_2456. 220 An ISEcp1 genetic context is associated with large-scale genomic deletions alongside blaCTX-221 M-15 loss. 222 For cefotaxime-sensitive escapers from Ecp1-I (n=19) and Ecp1-II (n=19) (Supplementary figure 223 3), PCR and subsequent Sanger sequencing of the amplicons revealed complete loss of ISEcp1 224 + blaCTX-M-15 for most of the escapers (n=36) and partial deletions including the protospacer 225 sequence (the 20 bp targeted by pKJK5::csgc[blaCTX-M-15]) for the remaining escapers (n=2) 226 (Supplementary figure 6A). 227 To better characterize the dimensions of the ISEcp1 + blaCTX-M-15 deletions, a β-galactosidase 228 assay was performed, taking advantage of the proximal lacZ gene (Figure 3A). This enzymatic 229 assay allows phenotypic screening of blue or white colonies based on presence or absence, 230 respectively, of β-galactosidase, the enzyme encoded by lacZ. A negative phenotype (white) 231 indicated deletions of IS Ecp1 + blaCTX-M-15 together with lacZ (Supplementary figure 6B). In 232 contrast, a positive phenotype (blue) delimited the start of the deletion within 1.2 kB (somewhere 233 in between lacZ and ISEcp1) (Supplementary figure 6C). Both phenotypes were observed across 234 escapers, with β-galactosidase-negative escapers being more abundant (n=30) (Supplementary 235 C) Isolate 26-II blaCTX-M-15 PCT A) Isolates Ecp1-I and Ecp1-II lacZ lacY ISEcp1 blaCTX-M-15 IS26 aac(6’)-Ib blaOXA-1 catB IS26 blaCTX-M-15 Tn3 IS26 Tn3 IS26 2451 B) Isolate 26-I IS26 aac(6’)-Ib blaOXA-1 catB IS26 yokD tmrB ISKpn11 IS26 Tn3 IS26 Tn3 blaCTX-M-15 IS26 2451 2453 2456 blaCTX-M-15 PCT .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 10 figure 6D). Short-read WGS was performed on a representative escaper from each phenotype 236 and isolate to confirm their genotypic basis. 237 Additionally, all cefotaxime-sensitive escapers from Ecp1 -I and Ecp1 -II were subjected to both 238 the csgc genotypic and phenotypic CRISPR-Cas9 functionality assay previously described. All 239 escapers showed csgc amplicons of expected lengths. In the phenotypic assay most showed 240 functional CRISPR -Cas9 systems (n=32). Nevertheless, a small subset showed dysfunctional 241 CRISPR-Cas9 activity (n=6), highlighting that both the loss of blaCTX-M-15 and the presence of a 242 dysfunctional system can co-occur (Supplementary figure 5B-C). 243 An IS 26 genetic context is associated with small-scale genomic deletions of blaCTX-M-15 244 facilitated by the presence of two flanking IS26 copies. 245 For isolates 26-I and 26-II, both with an IS26 genetic context, 26-II was chosen as a representative 246 for further study due to its higher IS26 load (Figure 3 B-C). First, hybrid WGS was performed for a 247 subset of escapers (n=10) which revealed either the deletion of the blaCTX-M-15 PCT (n=8) or larger 248 deletions including downstream chromosomal sequences (n=2). All the deletions led to a single 249 remaining IS26 chromosomal copy from the two original ones. 250 We used these representative data to inform PCRs assaying the deletion of the blaCTX-M-15 PCT 251 (Supplementary figure 7A) in all the cefotaxime-sensitive escapers (n=98). The ancestral isolate 252 and the cefotaxime-resistant escapers (used as positive controls ) showed the expected 4.9 kB 253 amplicon covering the blaCTX-M-15 PCT alongside a smaller amplicon, likely a PCR artefact resulting 254 from recombination between the multiple copies of IS26. Crucially, cefotaxime-sensitive 255 escapers (n=94) lacked the 4.9 kB amplicon and only displayed a 972 bp amplicon, which 256 corresponded to a single IS26 copy in agreement with the deletion of the blaCTX-M-15 PCT observed 257 in the WGS (Supplementary figure 7B). Additionally, 3 escapers, including the two with larger 258 deletions characterized by WGS, showed no amplicon, agree ing with the presence of larger 259 deletions thus avoiding primer binding (Supplementary figure 7A and C-D). Finally, one escaper 260 showed a <4.9 kb amplicon, likely explained by a partial deletion within blaCTX-M-15, thereby 261 conferring loss of the resistant phenotype and escape of CRISPR-Cas9 targeting. 262 To summarize, the emergence of cefotaxime-sensitive escapers was significantly impacted by 263 the genetic context of blaCTX-M-15. While all cefotaxime-sensitive escapers exhibited chromosomal 264 rearrangements involving the loss of blaCTX-M-15, an ISEcp1 genetic context was associated with 265 lower escape frequencies and large-scale deletions. In contrast, an IS26 genetic context revealed 266 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 11 higher escape fre quencies, primarily driven by small-scale deletions (blaCTX-M-15 PCT) occurring 267 between the two flanking IS26 elements. 268

269 Here, we demonstrated the sequence-specific antimicrobial activity of CRISPR-Cas9 by targeting 270 the chromosomally encoded ESBL gene blaCTX-M-15 in human-associated isolates of the E. coli 271 lineage ST131. The delivery of a blaCTX-M-15 targeting CRISPR-Cas9 cassette using the broad host-272 range conjugative plasmid pKJK5::csgc achieved a significant reduction (from 236-fold to 22.000-273 fold, depending on the isolate ) in conjugation efficiency compared to a non -targeting control, 274 explained by the non -viability of E. coli ST131 transconjugants . This demonstrates the 275 antimicrobial use of CRISPR -Cas9 against E. coli ST131 isolates , which can be found in 276 antimicrobial resistant UTIs, for which non -antibiotic treatments are urgently needed (41). 277 Promisingly, the use of pKJK5::csgc is compatible with probiotics, which are already prophylactic 278 and therapeutic treatment options for recurrent UTIs (42–44). However, we observed widespread 279 escape of CRISPR-Cas9 targeting across isolates and identified both escapers that retained the 280 blaCTX-M-15 gene and carried dysfunctional CRISPR-Cas9, and escapers that lost blaCTX-M-15 through 281 chromosomal rearrangements. Crucially, for the escapers with blaCTX-M-15 deletions, we found 282 significantly different escape frequencies depending on the genetic context of the target gene, 283 with increased escape frequencies (up to 813-fold) when blaCTX-M-15 is found surrounded by two 284 copies of IS26, as part of an IS 26 pseudo-compound transposon (PCT), compared to when it is 285 found downstream of an ISEcp1. 286 Cefotaxime-resistant (blaCTX-M-15-positive) escapers retained an intact version of blaCTX-M-15, 287 suggesting that CRISPR-Cas9 escape through mutations in the protospacer (the 20 bp targeted 288 by pKJK5::csgc[blaCTX-M-15]) or protospacer adjacent motif (PAM) sequence is rare in these isolates. 289 As mutations in the target site were previously re ported for other genes (36,45–49), we 290 hypothesize that the absence of such escapers in our setup may be attributed to mutations in 291 blaCTX-M-15 occurring below the detection limit of the assay (~10-6). Instead, all cefotaxime-resistant 292 escapers presented dysfunctional CRISPR -Cas9 systems. In line with previous literature 293 (35,45,46,48,50,51), w e found deletions, duplications and insertion s across cas9 and sgRNA. 294 Interestingly, we only found one escaper with an IS transposition into cas9, even though this is a 295 commonly reported event (36,46,47,50,52). To minimize escapers with dysfunctional CRISPR -296 Cas9 systems, several solutions have been proposed, including the use of multi-target arrays 297 (35,36,48), the manipulation of the CRISPR-Cas9 plasmid copy number (50), coupling CRISPR-298 Cas antimicrobials with CRISPR-regulated toxin-antitoxin systems (ATTACK) (53) or the use of 299 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 12 non-DNA based delivery strategies such as nano -sized CRISPR complexes (54), especially as 300 plasmid-based delivery relies on often heterogenous transcriptional activity of recipients (55). 301 However, the optimal solution to this problem has yet to be established. 302 In contrast, cefotaxime-sensitive escapers revealed blaCTX-M-15 deletions, and this loss of the target 303 gene occurred in significantly higher frequencies (up to 813-fold) for isolates with an IS26 genetic 304 context compared to an IS Ecp1 genetic context. The observed blaCTX-M-15 deletions were likely 305 driven by homologous recombination (HR) , which could either occur in the pre-existing 306 population leading to positively selected genotypes upon CRISPR-Cas9 exposure, or as a DNA-307 damage repair mechanism after CRISPR-Cas9 targeting and a subsequent triggering of the SOS 308 response (33,36,56). Escapers with s uccessful DNA-damage repair could be associated with 309 weaker CRISPR-Cas9 activity (33), poor sgRNA folding (57) or expression (58) and/or variability in 310 the host DNA damage tolerance and responses (59). Deletions of blaCTX-M-15 compatible with HR 311 were observed across isolates. For isolates Ecp1-I and Ecp1 -II, sequencing of few cefotaxime-312 sensitive escapers revealed deletions with border homology (9 -11 bp), likely indicating HR. In 313 isolate 26 -II, the most common deletion observed among cefotaxime -sensitive escapers (a 314 deletion of the blaCTX-M-15 PCT leaving a single IS 26 copy in the chromosome ; Supplementary 315 figure 7B), is consistent with the product of HR between the two directly oriented IS26 copies 316 (60–62) (Supplementary figure 8A-B). Recombination be fore CRISPR -Cas9 cleavag e would 317 release a circular molecule carrying blaCTX-M-15, known as a translocatable unit (TU) 318 (18,24,60,63,64), (Supplementary figure 8A) which is expected to be readily lost in absence of 319 self-replicative features (60) or through active targeting by CRISPR-Cas9. CRISPR-Cas9 cleavage 320 of intermediate circular molecules from similar excised MGEs has been recently reported (65). 321 Overall, the higher escape frequencies observed in an IS26 genetic context could be attributed to 322 the homology found between the several IS26 copies (Figure 3 B-C), which may facilitate HR. In 323 fact, HR between homologous IS elements is a common driver of chromosomal rearrangements 324 in E. coli (66). 325 Alternatively, blaCTX-M-15 loss could be driven by an incomplete IS-mediated blaCTX-M-15 mobilization. 326 Evidence of this was found in escapers from isolate 26-II with larger blaCTX-M-15 deletions lacking 327 homology in the borders (Supplementary figure 7C-D). These deletions likely resulted from a 328 Tnp26-dependent intramolecular copy-in mobilisation (24,67), in where the downstream genes 329 EC958_2453 and EC958_2456 would have been used as respective intramolecular target s to 330 mediate chromosomal excision , releasing a TU carrying the enclosed sequence (60) 331 (Supplementary figure 8CI-II), which would be readily lost or targeted by CRISPR -Cas9. This 332 mobilization pattern could also contribute to escapers with a blaCTX-M-15 PCT deletion, if using the 333 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 13 sequence found downstream of the PCT as an intramolecular target site (Supplementary figure 334 8CIII). In contrast , we could not find evidence of blaCTX-M-15 deletions mediated by IS Ecp1 335 mobilisation, as ISEcp1 mobilises together with downstream genes (18,68–71). Therefore, this 336 would only be possible in lacZ-positive cefotaxime-sensitive escapers , which are a minority 337 (Supplementary figure 6D). Sequencing of two lacZ-positive escapers revealed blaCTX-M-15 338 deletions incompatible with ISEcp1 mobilisation, suggesting that this is not a common escape 339 mechanism. 340 To summarize, even though our data does not allow us to discern the exact escape mechanism 341 that leads to chromosomal rearrangements, deletions resulting from both HR and incomplete IS-342 mediated blaCTX-M-15 mobilization are not mutually exclusive, and it is likely that the total escape 343 population is a combinatorial result of these events, each occurring at different frequencies. 344 Escapers with chromosomal rearrangements involving the deletion of the target gene have been 345 reported for other genes and bacterial species (33,36,49,57,72–74), including deletions with 346 border homology, suggesting HR (33,75,76). Additionally, some studies have shown a reduction 347 in escape frequencies when knocking out (33) or inhibiting (57) expression of RecA, an important 348 protein involved in HR (77) and found in all the ST131 isolates used in this work . Interestingly, 349 studies where the target gene is found within a chromosomally integrated MGE reported escapers 350 in which the entire MGE was deleted, similar to the deletions of the blaCTX-M-15 PCT reported here. 351 This occurred when the target genes were in a genomic island flanked by two directly oriented 352 IS1193 copies (73), in several pathogenicity islands (49,78) and in prophages (78). 353 The presence of escapers with chromosomal rearrangements after a CRISPR-Cas9 treatment 354 might be especially relevant when occurring at high frequenc ies, as for isolate 26-II (~10-3). The 355 emergence of this type of escapers could likely be reduced by choosing target genes independent 356 from chromosomally integrated MGEs and with genetic contexts that present limited or no 357 homology. In this study , escapers with blaCTX-M-15 loss were resensitized to cefotaxime, which 358 could be a beneficial outcome for antibiotic therapy. However, this type of chromosomal 359 rearrangements can also involve loss of other genes, leading to phenotypic changes (73,76). 360 Furthermore, d eletions resulting from successful DNA repair after CRISPR -Cas9 cleavage 361 through HR could be accompanied by mobilisation of other HR-mediated MGEs (18). This might 362 be especially relevant when using CRISPR-Cas9 against AMR-carrying bacteria, which often carry 363 MGEs (18). While we did not directly observe this, such far-reaching genomic consequences of 364 CRISPR-Cas9 targeting should also be kept in mind when using CRISPR-Cas9 as an editing tool, 365 which often relies on HR (31,58). 366 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 14 Altogether, this study showed the antimicrobial activity of CRISPR-Cas9 against the 367 chromosomally encoded AMR gene blaCTX-M-15 from human-associated E. coli ST131 isolates and 368 identified escapers resulting from either dysfunctional CRISPR -Cas9 or with chromosomal 369 rearrangements that led to the deletion of blaCTX-M-15. Our work showed the impact that the genetic 370 context of the target gene has on escape frequencies, where an association with IS26 led to a high 371 frequency of blaCTX-M-15 loss. This research underlines the importance of understanding the genetic 372 environment to be able to predict the treatment outcome of CRISPR -Cas9 antimicrobials and 373 CRISPR-Cas9 gene editing. Finally, we also want to highlight the potential use of CRISPR-Cas9 as 374 a tool to better characterize IS mobilization patterns by studying escapers arising from the 375 targeting of the IS elements or cargo genes. 376

377 Growth conditions, buffers and media 378 Lysogeny broth (LB), either liquid or mixed with agar, was used as growth media. Antibiotics were 379 used to ensure plasmid maintenance, select for chromosomal markers or perform phenotypic 380 assays at: chloramphenicol (Cm) 25 μg mL-1, gentamicin (Gm), kanamycin (Km) and streptomycin 381 (Sm) all 50 μg mL-1, and cefotaxime (CTX) 5 μg mL-1. Glycerol stocks were prepared at 20% (w/v) 382 and frozen at -70 oC. Sterile 0.9 % (w/v) NaCl was used as a buffer as indicated. Unless otherwise 383 specified, all kits and reagents were used following manufacturer’s instructions and incubations 384 were performed overnight (O/N) at 37 oC, 180 rpm. 385 Bacterial strains 386 We selected the E. coli ST131 isolates used in this study (Ecp1-I, Ecp1-II, 26-I and 26-II) based on 387 the genetic context of blaCTX-M-15 (genomes deposited on Gen bank under BioProject 388 PRJNA1281408). All the ST131 isolates belong to clade C, the most relevant in the clinic (79). 389 Isolates Ecp1-I, Ecp1-II and 26-I were chromosomally tagged with an aacC1 gene, conferring Gm 390 resistance, through electroporation of a Tn5 transposon plasmid pBAMD1-6 (80), followed by PCR 391 screening to confirm absence of residual plasmid as well as competition experiments against the 392 respective ancestral isolates to verify that tagging did not incur a fitness cost. In brief, the 393 modified strains were grown together with their wildtype variants, and colony counts revealed 394 similar growth. Isolate 26-II was not genetically modified due to intrinsic Gm resistance. E. coli K-395 12 MG1655::mCherry (81) was used as CRISPR -Cas9 donor strain because it represses 396 expression of gfp from pKJK5::csgc, which allows us to verify successful plasmid conjugation to 397 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 15 the ST131 isolates (presence of transconjugants) by checking gfp expression. Further bacteria 398 and plasmid information can be found in Supplementary table 1 and 2, respectively. 399 Development of the CRISPR-Cas9 system 400 Two CRISPR-Cas9 cassettes were designed with distinct sgRNA targets. The [ blaCTX-M-15] sgRNA 401 targets [CGCGTGATACCACTTCACCT]. This sequence is found within blaCTX-M-15 and followed by a 402 protospacer adjacent motif (PAM) in the E. coli ST131 reference genome (82) and the genome of 403 the four E. coli ST131 isolates used in the study. Furthermore, the sequence showed low 404 predicted CRISPR-off target activity in both CRISPOR (83) and Cas -OFFinder (84). The non -405 targeting [NT] control sgRNA targets random nucleotide sequence [GGTAAGACCATTAGAAGTAG], 406 20 bp which we confirmed to be absent from all E. coli ST131 isolates. These cassettes were 407 generated and transferred into pKJK5, resulting in pKJK5::csgc[ blaCTX-M-15] or [NT], following 408 protocols adopted from (37) (full details in Supplementary method 1 and Supplementary table 409 3). 410 Filter mating conjugation E. coli K-12 MG1655 (donor) - E. coli ST131 (recipient) 411 Single colonies of donors (MG1655 pKJK5::csgc[ blaCTX-M-15] or [NT]) and recipients ( E. coli 412 DH5α::SmR or the respective E. coli ST131 isolates) were grown O/N in 5 mL LB. Donors were 413 supplemented with Cm to avoid segregational loss of pKJK5. Cells were washed twice in 5 mL 414 NaCl, followed by OD600 adjustment to 0.5 – 0.6. Recipients were diluted 100-fold in NaCl. Filter 415 mating was performed in a Millipore 1225 Sampling Manifold using a sterile Whatman Cyclopore 416 Clear 0.2 µm 25mm polycarbonate membrane on top of a sterile Whatman glass microfiber filter, 417 binder free, grade GF/C, 25mm. The vacuum manifold was sterilised using ethanol and UV before 418 and between batches. Filters were washed by pumping through 2 mL of NaCl. Straight after, 1 mL 419 of NaCl, 1 mL of donor and 1 mL of a 100 -fold diluted recipient were pumped through. Five 420 biological replicates were performed for each donor -recipient combination. Additionally, 421 controls with donor -only, recipient -only and NaCl -only were performed and yielded re sults 422 consistent with expectations, supporting the validity of the experiment. The polycarbonate filters 423 were placed onto 10 % LB -agar plates and incubated for 48 h at 37 oC in the absence of 424 antibiotics. Filters were recovered in 3 mL NaCl and vortexed. From the cell suspension, cells 425 were recovered in LB, after which differential selective plating on LB agar was used to quantify 426 the proportion of 1) donors (no antibiotic selection; donors identified by assessing mCherry 427 expression using a stereo fluorescent lamp (Nightsea), 2) recipients (Gm) and 3) transconjugants 428 (Gm + Cm). Glycerol stocks were made with the remaining cell suspensions. 429 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 16 For each replicate, conjugation efficiency was calculated by dividing transconjugant 430 concentrations (CFU/mL) by recipient concentrations (CFU/mL) and means were determined for 431 each isolate and treatment. For each replicate from the pKJK5::csgc[blaCTX-M-15] treatment, escape 432 frequencies were calculated as a relative conjugation efficiency by dividing the conjugation 433 efficiency of pKJK5::csgc[blaCTX-M-15] for each replicate with the average conjugation efficiency of 434 pKJK5::csgc[NT]. The mean escape frequency per isolate was calculated as the average of all five 435 biological replicates. 436 Recovery of E. coli ST131 escapers ( E. coli ST131 pKJK5::csgc[blaCTX-M-15]) and phenotypic 437 analysis 438 E. coli ST131 pKJK5::csgc[blaCTX-M-15] escapers were recovered on Gm + Cm plates and the 439 presence of blaCTX-M-15 was verified based on their cefotaxime resistance profile . To do so, 440 individual clones were replated onto LB agar containing (i) Cm + CTX and (ii) Cm to assess the 441 cefotaxime phenotype while ensuring CRISPR -Cas9 plasmid maintenance. Additionally, we 442 generated glycerol stocks of escapers after O/N incubation in LB + Cm. The same procedure was 443 used to select for 100 transconjugants from the non -targeting treatment ( E. coli ST131 444 pKJK5::csgc[NT]). A visual representation of the filter mating assay and the phenotypic screen of 445 the escapers can be found in Supplementary figure 1. 446 To calculate cefotaxime-resistant and cefotaxime-sensitive escape frequencies, the cefotaxime 447 resistance profiles were assessed for all escapers recovered directly from the filter mating assay 448 for isolates Ecp1 -I, Ecp1 -II and 26 -I. For isolate 26 -II, due to the larger size of the escape 449 population recovered, ten escapers per replicate were randomly selected for resistance profiling. 450 The proportions of each cefotaxime phenotype were then used to estimate cefotaxime-resistant 451 and cefotaxime-sensitive transconjugants concentrations (CFU/mL) per replicate. Conjugation 452 efficiencies for both phenotypes were calculated by dividing the respective transconjugant 453 concentrations by the recipient concentrations for each replicate of the pKJK5::csgc[ blaCTX-M-15] 454 treatment. Finally, cefotaxime -resistant and cefotaxime -sensitive escape frequencies were 455 calculated as a relative conjugation efficiency by dividing the cefotaxime-resistant or cefotaxime-456 sensitive conjugation efficiencies of pKJK5::csgc[ blaCTX-M-15] for each replicate with the average 457 conjugation efficiency of pKJK5::csgc[NT]. The mean escape frequency per phenotype and 458 isolate was calculated as the average of all biological replicates, including those where no 459 cefotaxime-resistant or cefotaxime -sensitive escapers were found. Replicate 4 from isolate 460 Ecp1-II was excluded from the calculations of cefotaxime -resistant and cefotaxime -sensitive 461 escape frequencies as only one escaper was directly recovered from the filter mating assay. 462 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 17 Genotypic analysis of blaCTX-M-15 463 PCR was used to study blaCTX-M-15 presence in escapers. To obtain enough DNA template material, 464 frozen glycerol stocks of escapers were scraped with a pipette tip and the tip was swirled in 10 μL 465 water to suspend attached material. The suspension was then heated to 95 oC for 15 minutes and 466 centrifugated at 3.500 rpm for 5 minutes. For cefotaxime-resistant and cefotaxime -sensitive 467 escapers from isolates Ecp1-I and Ecp1 -II, blaCTX-M-15 presence was studied using primers with 468 binding sites within the gene ( Supplementary figure 2A). Furthermore, for the cefotaxime-469 sensitive escapers primers designed to amplify IS Ecp1 + blaCTX-M-15 were also used 470 (Supplementary figure 2A). Phusion High-Fidelity polymerase (Thermo Scientific) was used in 471 both PCRs. Additionally, ExoSAP -cleaned (NEB) PCR amplicons were Sanger sequenced. In 472 cefotaxime-sensitive escapers from isolate 26 -II, the deletion of the blaCTX-M-15 PCT was studied 473 using primers annealing outside the PCT ( Supplementary figure 2B) with 1x VeriFi Hot Start 474 Polymerase (PCR Biosystems). Across PCRs, when no amplicon was found, 16S PCR or 475 amplification of other genes were used to verify template presence. Moreover, the respective 476 ancestral isolates and transconjugants from the non -targeting control ( pKJK5::csgc[NT]) were 477 used as positive controls for the presence of blaCTX-M-15. All primers can be found in 478 Supplementary table 3. 479 Genotypic analysis of CRISPR-Cas9 integrity 480 Cefotaxime-resistant and cefotaxime-sensitive escapers from isolates Ecp1 -I and Ecp1 -II were 481 subjected to a genotypic CRISPR -Cas9 integrity assay to understand whether mutations had 482 occurred in cas9 and/or the sgRNA. Four sets of primers generating overlapping amplicons were 483 used, including a region upstream of cas9 and downstream of the sgRNA. Additionally, a fifth set 484 was used to specifically amplify the sgRNA (Supplementary figure 2C). PCRs were performed 485 using 2x PCRBIO Taq Mix Red (PCR Biosystems). Amplicons with unexpected lengths (i.e. 486 amplicons with an estimated length >60 bp different from the expected amplicon length) were 487 purified from the agarose gel using the Monarch® DNA Gel Extraction Kit Protocol (NEB) and 488 Sanger sequenced. Transconjugants from the pKJK5::csgc[NT] control and the ancestral isolates 489 were used as controls, and 16S amplification was used to verify template presence. DNA 490 template for the PCRs was obtained as described above. All primers can be found in 491 Supplementary table 3. 492 Phenotypic CRISPR-Cas9 functionality assay 493 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 18 To check for the presence of dysfunctional CRISPR -Cas9 activity in cefotaxime -resistant 494 escapers from isolates Ecp1-I and Ecp1-II that showed expected lengths of csgc amplicons and, 495 in cefotaxime-sensitive escapers, a phenotypic CRISPR-Cas9 functionality assay was performed. 496 For this assay, liquid conjugations were performed with donors harbouring either plasmid 497 pCTX15, which we engineered to carry blaCTX-M-15, or plasmid pCTRL, which we engineered to carry 498 a mutated version of blaCTX-M-15 that allows escape from CRISPR -Cas9 targeting through a silent 499 point mutation of the PAM sequence. Specific details about plasmid engineering and the assay 500 can be found in Supplementary method 1 and 2. 501 Phenotypic β-galactosidase assay 502 To characterize the IS Ecp1 blaCTX-M-15 deletions found in cefotaxime -sensitive escapers from 503 isolates Ecp1-I and Ecp1-II, a phenotypic β-galactosidase assay was performed that makes use 504 of the proximity of blaCTX-M-15 to the lacZ gene. Individual escapers were plated onto 0.2 μg mL-1 X-505 Gal LB-agar plates. Colonies were screened for white (β-galactosidase / lacZ-negative) or blue (β-506 galactosidase / lacZ-positive) phenotypes. Cefotaxime-resistant escapers, transconjugants from 507 the pKJK5::csgc[NT] control and the ancestral isolates were used as positive controls. 508 Sequencing 509 Whole genome sequencing was performed to obtain genomic details of the four E. coli ST131 510 ancestral isolates and several escapers from the CRISPR -Cas9 treatment. For the ancestral 511 isolates, DNA extraction was performed from LB + CTX 3 μg mL-1 ON cultures using FastDNA Spin 512 Kit (MP Biomedicals). The extracted DNA was treated with 2 μL of RNAseA 20 mg mL -1 for 10 513 minutes at 37 oC and purified with SPRISelect beads (Beckman Coulter). Sequencing libraries 514 were prepared with 2 μg of purified DNA using the ONT Ligation Sequencing Kit (Oxford Nanopore 515 Technologies) and long-read sequencing was performed using PromethION (Exeter Sequencing 516 Facility). Genomes were assembled using Unicycler (version 0.5.0) (85), with automated 517 annotations generated using RASTtk (86). Short-read WGS (MicrobesNG, Birmingham) was 518 performed for one β-galactosidase-positive and one β-galactosidase-negative cefotaxime -519 sensitive escaper from isolates Ecp1 -I and Ecp1 -II. Similarly, short -read WGS (MicrobesNG, 520 Birmingham) was also performed for escaper 5.10 from isolate Ecp1 -II. Hybrid WGS 521 (MicrobesNG, Birmingham) was performed for ten cefotaxime -sensitive escapers from isolate 522 26-II (two per each filter mating replicate) and for one transconjugant from the pKJK5::csgc[NT] 523 control. Genome assembly was performed using Flye (87) and the respective deletions were 524 characterized. Benchling was used for DNA visualization (88). 525 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 19 Data analysis 526 Data processing, visualisation and statistical analyses were performed in R version 4.4.2 using R 527 studio 2023.9.1.494 (89). Statistical modelling was performed using package lme4 (version 1.1-528 37) (90). Data processing and plotting were performed using packages tidyverse (version 2.0.0) 529 (91), ggplot2 (version 3.5.1) (92), gggenes (version 0.5.1) (93), patchwork (version 1.3.1) (94) and 530 grid (version 4.4.2) (89). Significance of fixed effects was determined through comparing nested 531 models using chi -squared tests ( α = 0.05), beginning with a global model that included all 532 biologically relevant fixed effects and interaction terms. Interaction term statistical significance 533 was always tested first, and where interactions were statistically significant, all constituent fixed 534 effects were retained in the model. R package DHARMa (version 0.4.6) (95) was used to diagnose 535 potential model issues. Specifically, we checked if residuals deviation were deviated from the 536 expected distribution, if the model had over or under dispersion, if the datasets contained highly 537 influential outliers, and if the residuals were heteroscedastic. Post-hoc analyses were performed 538 using emmeans (version 1.10.2) (96). 539 Conjugation efficiency of CRISPR -Cas9 and the cefotaxime -resistant and cefotaxime-sensitive 540 escape frequencies were modelled using binomial family generalised linear models (GLMs) and 541 a logit link function, with the number of colonies counted for the recipients and transconjugants 542 as the weights value. Due to large differences in the count of transconjugants between 543 treatments, concentrations of recipients and transconjugants (CFU/mL) were logarithmically 544 transformed (when the dataset included zeros, log10+1 was used). 545 Our global model of conjugation efficiency of CRISPR -Cas9 included the predictor variables 546 plasmid identity, isolate identity, and an interaction between plasmid and isolate identity. For the 547 cefotaxime-resistant and cefotaxime -sensitive escape frequency analyses we modelled the 548 predictor variable genetic context of the blaCTX-M-15. For the cefotaxime -sensitive escape 549 frequencies, model validation revealed slight deviation in the first quartile of the residuals vs 550 predicted plot . However, this model remained the most appropriate, and the large observed 551 effect sizes mean it is improbable we are drawing erroneous conclusions from this analysis. For 552 the cefotaxime-resistant escape frequencies, model validation revealed some influential outliers 553 (replicates with no escapers) and model optimization was performed removing outliers. 554 For the phenotypic CRISPR-Cas9 functionality assay in E. coli K-12 MG1655 carrying either 555 pKJK5::csgc[NT] or pKJK5::csgc[ blaCTX-M-15] the relative conjugation efficiencies (pCTRL/pCTX15) 556 were modelled using a Gamma family GLM with a log link function and with CRISPR -Cas9 557 plasmid as a predictor variable. For the phenotypic CRISPR-Cas9 functionality assay with 558 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 20 escapers, the data were individually modelled for each isolate. For both datasets, pCTRL/pCTX15 559 was modelled using a Gamma family GLM with a log link function, with cefotaxime phenotype as 560 a predictor variable. Model validation revealed some heteroscedasticity for isolate Ecp1-II, which 561 could be attributed to the presence of few cefotaxime -sensitive escapers with dysfunctional 562 CRISPR-Cas9 systems. Nevertheless, the Gamma GLM remained the most appropriate for our 563 data. 564 DATA AVAILABILITY 565 The genome of the E. coli ST131 isolates (Ecp1 -I, Ecp1 -II, 26 -I, 26 -II) were deposited in 566 PRJNA1281408. The assembled genomes of escapers with blaCTX-M-15 deletions from isolate s 567 Ecp1-I, Ecp1-II and 26-II and for escaper 5.10 Ecp1-II can be found in __to be determined__. The 568 data analysis can be found in GitHub __to be determined__. All the raw data and the full escape 569 characterization can be found in the raw data excel file in __to be determined__ . Additional 570 experimental details, methods, tables and figures can be found in the online version of this 571 article. 572

AND FUNDING 573 S.V .H. gratefully acknowledges funding from the Biotechnology and Biological Sciences Research 574 council (BBSRC; BB/R010781/1), the Lister Institute for Preventative Medicine, and the Joint 575 Programme Initiative AntiMicrobial Resistance (JPI -AMR) ‘Harissa’ call (MISTAR; 576 MR/W031191/1)). D.S was supported in part by grant MR/N0137941/1 for the GW4 BIOMED MRC 577 DTP , awarded to the Universities of Bath, Bristol, Cardiff and Exeter from the Medical Research 578 Council (MRC)/UKRI. Genome sequencing of the ST131 isolates was supported by funding from 579 the Natural Environment Research Council awarded to AFCL (award number NE/R013748/1) and 580 utilised equipment funded by the UK Medical Research Council (MRC) Clinical Research 581 Infrastructure Initiative (award number MR/M008924/1) . For the purpose of open access, the 582 author has applied a ‘Creative Commons Attribution (CC BY) licence to any Author Accepted 583 Manuscript version arising from this submission. 584 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 21

585 1. Naghavi M, Vollset SE, Ikuta KS, Swetschinski LR, Gray AP , Wool EE, et al. Global burden of 586 bacterial antimicrobial resistance 1990 –2021: a systematic analysis with forecasts to 587 2050. The Lancet. 2024 Sep 28;404(10459):1199–226. 588 2. Dadgostar P . Antimicrobial Resistance: Implications and Costs. Infect Drug Resist. 2019 589 Dec 20;Volume 12:3903–10. 590 3. Haney EF , Hancock REW. Addressing Antibiotic Failure —Beyond Genetically Encoded 591 Antimicrobial Resistance. Front Drug Discov. 2022;2:892975. 592 4. Murray CJ, Ikuta KS, Sharara F , Swetschinski L, Robles Aguilar G, Gray A, et al. Global 593 burden of bacterial antimicrobial resistance in 2019: a systematic analysis. Lancet. 594 2022;399(10325):629–55. 595 5. World Health Organization. Prioritization of pathogens to guide discovery, research and 596 development of new antibiotics for drug -resistant bacterial infections, including 597 tuberculosis. 2017. 598 6. Poolman JT, Wacker M. Extraintestinal Pathogenic Escherichia coli, a Common Human 599 Pathogen: Challenges for Vaccine Development and Progress in the Field. J Infect Dis. 600 2016 Jan 1;213(1):6–13. 601 7. Daga AP , Koga VL, Soncini JGM, de Matos CM, Perugini MRE, Pelisson M, et al. Escherichia 602 coli Bloodstream Infections in Patients at a University Hospital: Virulence Factors and 603 Clinical Characteristics. Front Cell Infect Microbiol. 2019 Jun 6;9:191. 604 8. Manges AR, Geum HM, Guo A, Edens TJ, Fibke CD, Pitout JDD. Global Extraintestinal 605 Pathogenic Escherichia coli (ExPEC) Lineages. Clin Microbiol Rev. 2019 Jun 606 12;32(3):e00135-18. 607 9. Rogers BA, Sidjabat HE, Paterson DL. Escherichia coli O25b -ST131: a pandemic, 608 multiresistant, community-associated strain. J Antimicrob Chemother. 2011 Jan;66(1):1–609 14. 610 10. Johnson JR, Johnston B, Clabots C, Kuskowski MA, Castanheira M. Escherichia coli 611 sequence type ST131 as the major cause of serious multidrug -resistant E. coli infections 612 in the United States. Clin Infect Dis. 2010 Aug 1;51(3):286–94. 613 11. Pitout JDD, Finn TJ. The evolutionary puzzle of Escherichia coli ST131. Infect Genet Evol. 614 2020 Jul;81:104265. 615 12. Cantón R, Coque TM. The CTX -M beta-lactamase pandemic. Curr Opin Microbiol. 2006 616 Oct;9(5):466–75. 617 13. Nicolas-Chanoine MH, Bertrand X, Madec JY . Escherichia coli ST131, an intriguing clonal 618 group. Clin Microbiol Rev. 2014 Jul;27(3):543–74. 619 14. Cantón R, González -Alba JM, Galán JC. CTX-M Enzymes: Origin and Diffusion. Front 620 Microbiol. 2012 Apr 2;3:110. 621 15. Rossolini GM, D’Andrea MM, Mugnaioli C. The spread of CTX-M-type extended-spectrum 622 beta-lactamases. Clin Microbiol Infect. 2008 Jan 1;14(Suppl 1):33–41. 623 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 22 16. Siguier P , Gourbeyre E, Chandler M. Bacterial insertion sequences: their genomic impact 624 and diversity. FEMS Microbiol Rev. 2014 Sep;38(5):865–91. 625 17. Zhao WH, Hu ZQ. Epidemiology and genetics of CTX-M extended-spectrum β-lactamases 626 in Gram-negative bacteria. Crit Rev Microbiol. 2013 Feb;39(1):79–101. 627 18. Partridge SR, Kwong SM, Firth N, Jensen SO. Mobile Genetic Elements Associated with 628 Antimicrobial Resistance. Clin Microbiol Rev. 2018 Aug 1;31(4):e00088-17. 629 19. Mahillon J, Chandler M. Insertion sequences. Microbiol Mol Biol Rev. 1998 Sep;62(3):725–630 74. 631 20. Pitout JDD, Chen L. The Significance of Epidemic Plasmids in the Success of Multidrug -632 Resistant Drug Pandemic Extraintestinal Pathogenic Escherichia coli. Infect Dis Ther. 2023 633 Apr;12(4):1029–41. 634 21. Woodford N, Carattoli A, Karisik E, Underwood A, Ellington MJ, Livermore DM. Complete 635 nucleotide sequences of plasmids pEK204, pEK499, and pEK516, encoding CTX -M 636 enzymes in three major Escherichia coli lineages from the United Kingdom, all belonging 637 to the international O25:H4 -ST131 clone. Antimicrob Agents Chemother. 2009 638 Oct;53(10):4472–82. 639 22. Pajand O, Rahimi H, Darabi N, Roudi S, Ghassemi K, Aarestrup FM, et al. Arrangements of 640 Mobile Genetic Elements among Virotype E Subpopulation of Escherichia coli Sequence 641 Type 131 Strains with High Antimicrobial Resistance and Virulence Gene Content. 642 mSphere. 2021 Aug 25;6(4):e0055021. 643 23. Varani A, He S, Siguier P , Ross K, Chandler M. The IS6 family, a clinically important group 644 of insertion sequences including IS26. Mob DNA. 2021 Mar;23;12(1):11. 645 24. Harmer CJ, Hall RM. IS26 and the IS26 family: versatile resistance gene movers and 646 genome reorganizers. Microbiol Mol Biol Rev. 2024 Jun 27;88(2):e0011922. 647 25. Can F , Azap OK, Seref C, Ispir P , Arslan H, Ergonul O. Emerging Escherichia coli 648 O25b/ST131 clone predicts treatment failure in urinary tract infections. Clin Infect Dis. 649 2015 Feb 15;60(4):523–7. 650 26. Syre H, Hetland MAK, Bernhoff E, Bollestad M, Grude N, Simonsen GS, et al. Microbial risk 651 factors for treatment failure of pivmecillinam in community -acquired urinary tract 652 infections caused by ESBL-producing Escherichia coli. APMIS. 2020 Mar;128(3):232–41. 653 27. Pursey E, Sünderhauf D, Gaze WH, Westra ER, van Houte S. CRISPR -Cas antimicrobials: 654 Challenges and future prospects. PLoS Pathog. 2018 Jun 1;14(6):e1006990. 655 28. Mayorga-Ramos A, Zúñiga -Miranda J, Carrera-Pacheco SE, Barba -Ostria C, Guamán LP . 656 CRISPR-Cas-Based Antimicrobials: Design, Challenges, and Bacterial Mechanisms of 657 Resistance. ACS Infect Dis. 2023 Jul 14;9(7):1283–302. 658 29. Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, Charpentier E. A programmable dual-659 RNA-guided DNA endonuclease in adaptive bacterial immunity. Science (1979). 2012 Aug 660 17;337(6096):816–21. 661 30. Jiang F , Doudna JA. CRISPR-Cas9 Structures and Mechanisms. Annu Rev Biophys. 2017 662 May 22;46:505–29. 663 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 23 31. Xue C, Greene EC. DNA Repair Pathway Choices in CRISPR -Cas9-Mediated Genome 664 Editing. Trends Genet. 2021 Jul;37(7):639–56. 665 32. Citorik RJ, Mimee M, Lu TK. Sequence -specific antimicrobials using efficiently delivered 666 RNA-guided nucleases. Nat Biotechnol. 2014 Nov;32(11):1141–5. 667 33. Cui L, Bikard D. Consequences of Cas9 cleavage in the chromosome of Escherichia coli. 668 Nucleic Acids Res. 2016 May 19;44(9):4243–51. 669 34. Bikard D, Euler CW, Jiang W, Nussenzweig PM, Goldberg GW, Duportet X, et al. Exploiting 670 CRISPR-Cas nucleases to produce sequence -specific antimicrobials. Nat Biotechnol. 671 2014 Nov;32(11):1146–50. 672 35. Uribe R V ., Rathmer C, Jahn LJ, Ellabaan MMH, Li SS, Sommer MOA. Bacterial resistance 673 to CRISPR-Cas antimicrobials. Sci Rep. 2021 Dec 1;11(1):17267. 674 36. Reuter A, Hilpert C, Dedieu -Berne A, Lematre S, Gueguen E, Launay G, et al. Targeted -675 antibacterial-plasmids (TAPs) combining conjugation and CRISPR/Cas systems achieve 676 strain-specific antibacterial activity. Nucleic Acids Res. 2021 Apr 6;49(6):3584–98. 677 37. Sünderhauf D, Klümper U, Pursey E, Westra ER, Gaze WH, van Houte S. Removal of AMR 678 plasmids using a mobile, broad host -range CRISPR -Cas9 delivery tool. Microbiology 679 (Reading). 2023 May;169(5):001334. 680 38. Leonard AFC, Zhang L, Balfour AJ, Garside R, Hawkey PM, Murray AK, et al. Exposure to and 681 colonisation by antibiotic -resistant E. coli in UK coastal water users: Environmental 682 surveillance, exposure assessment, and epidemiological study (Beach Bum Survey). 683 Environ Int. 2018 May;114:326–33. 684 39. Paterson DL, Bonomo RA. Extended -Spectrum β-Lactamases: a Clinical Update. Clin 685 Microbiol Rev. 2005 Oct;18(4):657–86. 686 40. Jinek M, Jiang F , Taylor DW, Sternberg SH, Kaya E, Ma E, et al. Structures of Cas9 687 Endonucleases Reveal RNA -Mediated Conformational Activation. Science (1979). 2014 688 Mar 14;343(6176):1247997–1247997. 689 41. Sihra N, Goodman A, Zakri R, Sahai A, Malde S. Nonantibiotic prevention and management 690 of recurrent urinary tract infection. Nat Rev Urol. 2018 Dec;15(12):750–76. 691 42. Durrani B, Mohammad A, Ljubetic BM, Dobberfuhl AD. The Potential Role of Persister Cells 692 in Urinary Tract Infections. Curr Urol Rep. 2023 Nov;24(11):541–51. 693 43. Reid G, Bruce AW. Probiotics to prevent urinary tract infections: the rationale and 694 evidence. World J Urol. 2006 Feb;24(1):28–32. 695 44. Gupta V , Nag D, Garg P . Recurrent Urinary Tract Infections in Women: How Promising is the 696 Use of Probiotics? Indian J Med Microbiol. 2017 Jul 1;35(3):347–54. 697 45. Jiang W, Bikard D, Cox D, Zhang F , Marraffini LA. RNA-guided editing of bacterial genomes 698 using CRISPR-Cas systems. Nat Biotechnol. 2013 Jan 29;31(3):233–9. 699 46. Neil K, Allard N, Roy P , Grenier F , Menendez A, Burrus V , et al. High‐efficiency delivery of 700 CRISPR‐Cas9 by engineered probiotics enables precise microbiome editing. Mol Syst Biol. 701 2021 Oct;17(10):e10335. 702 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 24 47. Hamilton TA, Pellegrino GM, Therrien JA, Ham DT, Bartlett PC, Karas BJ, et al. Efficient inter-703 species conjugative transfer of a CRISPR nuclease for targeted bacterial killing. Nat 704 Commun. 2019 Oct 4;10(1):4544. 705 48. Gomaa AA, Klumpe HE, Luo ML, Selle K, Barrangou R, Beisel CL. Programmable removal 706 of bacterial strains by use of genome - targeting CRISPR -cas systems. mBio. 2014 Jan 707 28;5(1):e00928-13. 708 49. Vercoe RB, Chang JT, Dy RL, Taylor C, Gristwood T, Clulow JS, et al. Cytotoxic 709 Chromosomal Targeting by CRISPR/Cas Systems Can Reshape Bacterial Genomes and 710 Expel or Remodel Pathogenicity Islands. PLoS Genet. 2013 Apr;9(4):e1003454. 711 50. Li Q, Sun M, Lv L, Zuo Y , Zhang S, Zhang Y , et al. Improving the Editing Efficiency of CRISPR-712 Cas9 by Reducing the Generation of Escapers Based on the Surviving Mechanism. ACS 713 Synth Biol. 2023 Mar 17;12(3):672–80. 714 51. Aparicio T, de Lorenzo V , Martínez-García E. CRISPR/Cas9-Based Counterselection Boosts 715 Recombineering Efficiency in Pseudomonas putida. Biotechnol J. 2018 716 May;13(5):e1700161. 717 52. Sheng Y , Wang H, Ou Y , Wu Y , Ding W, Tao M, et al. Insertion sequence transposition 718 inactivates CRISPR-Cas immunity. Nat Commun. 2023 Jul 20;14(1):4366. 719 53. Wang R, Shu X, Zhao H, Xue Q, Liu C, Wu A, et al. Associate toxin -antitoxin with CRISPR-720 Cas to kill multidrug-resistant pathogens. Nat Commun. 2023 Apr 12;14(1):2078. 721 54. Kang YK, Kwon K, Ryu JS, Lee HN, Park C, Chung HJ. Nonviral Genome Editing Based on a 722 Polymer-Derivatized CRISPR Nanocomplex for Targeting Bacterial Pathogens and 723 Antibiotic Resistance. Bioconjug Chem. 2017 Apr 19;28(4):957–67. 724 55. Cyriaque V , Ibarra-Chávez R, Kuchina A, Seelig G, Nesme J, Madsen JS. Single-cell RNA 725 sequencing reveals plasmid constrains bacterial population heterogeneity and identifies 726 a non-conjugating subpopulation. Nat Commun. 2024;15:5853. 727 56. Moreb EA, Hoover B, Yaseen A, Valyasevi N, Roecker Z, Menacho-Melgar R, et al. Managing 728 the SOS Response for Enhanced CRISPR -Cas-Based Recombineering in E. coli through 729 Transient Inhibition of Host RecA Activity. ACS Synth Biol. 2017 Dec 15;6(12):2209–18. 730 57. Vialetto E, Miele S, Goren MG, Yu J, Yu Y , Collias D, et al. Systematic interrogation of CRISPR 731 antimicrobials in Klebsiella pneumoniae reveals nuclease-, guide- and strain-dependent 732 features influencing antimicrobial activity. Nucleic Acids Res. 2024 Jun 10;52(10):6079 –733 91. 734 58. Collias D, Vialetto E, Yu J, Co K, Almási É d. H, Rüttiger AS, et al. Systematically attenuating 735 DNA targeting enables CRISPR -driven editing in bacteria. Nat Commun. 2023 Feb 736 8;14(1):680. 737 59. Moreb EA, Hoover B, Yaseen A, Valyasevi N, Roecker Z, Menacho-Melgar R, et al. Managing 738 the SOS Response for Enhanced CRISPR -Cas-Based Recombineering in E. coli through 739 Transient Inhibition of Host RecA Activity. ACS Synth Biol. 2017 Dec 15;6(12):2209–18. 740 60. Harmer CJ, Hall RM. IS26 -Mediated Formation of Transposons Carrying Antibiotic 741 Resistance Genes. mSphere. 2016 Apr 6;1(2):e00038-16. 742 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 25 61. Harmer CJ, Hall RM. IS26 -Mediated Precise Excision of the IS26 -aphA1a Translocatable 743 Unit. mBio. 2015 Dec 8;6(6):e01866-15. 744 62. Aihara M, Gotoh Y , Shirahama S, Matsushima Y , Uchiumi T, Kang D, et al. Generation and 745 maintenance of the circularized multimeric IS26-associated translocatable unit encoding 746 multidrug resistance. Commun Biol. 2024;7:597. 747 63. Harmer CJ, Moran RA, Hall RM. Movement of IS26-associated antibiotic resistance genes 748 occurs via a translocatable unit that includes a single IS26 and preferentially inserts 749 adjacent to another IS26. mBio. 2014 Oct 7;5(5):e01801-14. 750 64. Harmer CJ, Moran RA, Hall RM. Movement of IS26-Associated Antibiotic Resistance Genes 751 Occurs via a Translocatable Unit That Includes a Single IS26 and Preferentially Inserts 752 Adjacent to Another IS26. mBio. 2014 Oct 7;5(5). 753 65. Wang P , Du X, Zhao Y , Wang W, Cai T, Tang K, et al. Combining CRISPR/Cas9 and natural 754 excision for the precise and complete removal of mobile genetic elements in bacteria. 755 Appl Environ Microbiol. 2024 Apr 1;90(4):e0009524. 756 66. Raeside C, Gaffé J, Deatherage DE, Tenaillon O, Briska AM, Ptashkin RN, et al. Large 757 chromosomal rearrangements during a long-term evolution experiment with Escherichia 758 coli. mBio. 2014 Sep 9;5(5):e01377-14. 759 67. He S, Hickman AB, Varani AM, Siguier P , Chandler M, Dekker JP , et al. Insertion Sequence 760 IS26 Reorganizes Plasmids in Clinically Isolated Multidrug -Resistant Bacteria by 761 Replicative Transposition. mBio. 2015 Jun 9;6(3):e00762. 762 68. Poirel L, Decousser JW, Nordmann P . Insertion Sequence ISEcp1B Is Involved in Expression 763 and Mobilization of a blaCTX-M β-Lactamase Gene. Antimicrob Agents Chemother. 2003 764 Sep;47(9):2938–45. 765 69. Zong Z, Partridge SR, Iredell JR. ISEcp1 -Mediated Transposition and Homologous 766 Recombination Can Explain the Context of blaCTX -M-62 Linked to qnrB2. Antimicrob 767 Agents Chemother. 2010 Jul;54(7):3039–42. 768 70. Poirel L, Lartigue MF , Decousser JW, Nordmann P . ISEcp1B -Mediated Transposition of 769 blaCTX-M in Escherichia coli. Antimicrob Agents Chemother. 2005 Jan;49(1):447–50. 770 71. Wachino JI, Yamane K, Kimura K, Shibata N, Suzuki S, Ike Y , et al. Mode of Transposition 771 and Expression of 16S rRNA Methyltransferase Gene rmtC Accompanied by ISEcp1. 772 Antimicrob Agents Chemother. 2006 Sep;50(9):3212–5. 773 72. Qi LS, Larson MH, Gilbert LA, Doudna JA, Weissman JS, Arkin AP , et al. Repurposing 774 CRISPR as an RNA -Guided Platform for Sequence -Specific Control of Gene Expression. 775 Cell. 2013 Feb 28;152(5):1173–83. 776 73. Selle K, Klaenhammer TR, Barrangou R. CRISPR -based screening of genomic island 777 excision events in bacteria. Proc Natl Acad Sci U S A. 2015 Jun 30;112(26):8076–81. 778 74. Guan J, Wang W, Sun B. Chromosomal Targeting by the Type III-A CRISPR-Cas System Can 779 Reshape Genomes in Staphylococcus aureus. mSphere. 2017 Nov 15;2(6):e00403-17. 780 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 26 75. Lam KN, Spanogiannopoulos P , Soto -Perez P , Alexander M, Nalley MJ, Bisanz JE, et al. 781 Phage-delivered CRISPR-Cas9 for strain-specific depletion and genomic deletions in the 782 gut microbiome. Cell Rep. 2021 Nov 2;37(5):109930. 783 76. Cañez C, Selle K, Goh YJ, Barrangou R. Outcomes and characterization of chromosomal 784 self-targeting by native CRISPR -Cas systems in Streptococcus thermophilus. FEMS 785 Microbiol Lett. 2019 May 1;366(9):fnz105. 786 77. Hu J, Crickard JB. All who wander are not lost: the search for homology during homologous 787 recombination. Biochem Soc Trans. 2024 Feb 28;52(1):367–77. 788 78. Wang P , Du X, Zhao Y , Wang W, Cai T, Tang K, et al. Combining CRISPR/Cas9 and natural 789 excision for the precise and complete removal of mobile genetic elements in bacteria. 790 Appl Environ Microbiol. 2024 Apr 17;90(4):e0009524. 791 79. Kocsis B, Gulyás D, Szabó D. Emergence and Dissemination of Extraintestinal Pathogenic 792 High-Risk International Clones of Escherichia coli. Life (Basel). 2022 Dec 10;12(12):2077. 793 80. Martínez-García E, Aparicio T, de Lorenzo V , Nikel PI. New transposon tools tailored for 794 metabolic engineering of Gram-negative microbial cell factories. Front Bioeng Biotechnol. 795 2014 Oct 28;2:46. 796 81. Klümper U, Riber L, Dechesne A, Sannazzarro A, Hansen LH, Sørensen SJ, et al. Broad host 797 range plasmids can invade an unexpectedly diverse fraction of a soil bacterial community. 798 ISME J. 2015;9(4):934–45. 799 82. Forde BM, Ben Zakour NL, Stanton -Cook M, Phan MD, Totsika M, Peters KM, et al. The 800 Complete Genome Sequence of Escherichia coli EC958: A High Quality Reference 801 Sequence for the Globally Disseminated Multidrug Resistant E. coli O25b:H4 -ST131 802 Clone. PLoS One. 2014 Aug 15;9(8):e104400. 803 83. Concordet JP , Haeussler M. CRISPOR: intuitive guide selection for CRISPR/Cas9 genome 804 editing experiments and screens. Nucleic Acids Res. 2018 Jul 2;46(W1):W242–5. 805 84. Bae S, Park J, Kim JS. Cas -OFFinder: a fast and versatile algorithm that searches for 806 potential off-target sites of Cas9 RNA -guided endonucleases. Bioinformatics. 2014 May 807 15;30(10):1473–5. 808 85. Wick RR, Judd LM, Gorrie CL, Holt KE. Unicycler: Resolving bacterial genome assemblies 809 from short and long sequencing reads. PLoS Comput Biol. 2017 Jun 1;13(6):e1005595. 810 86. Brettin T, Davis JJ, Disz T, Edwards RA, Gerdes S, Olsen GJ, et al. RASTtk : A modular and 811 extensible implementation of the RAST algorithm for building custom annotation pipelines 812 and annotating batches of genomes. Sci Rep. 2015 Feb 10;5:8365. 813 87. Kolmogorov M, Yuan J, Lin Y , Pevzner PA. Assembly of long, error-prone reads using repeat 814 graphs. Nat Biotechnol. 2019 May;37(5):540–6. 815 88. [Biology Software]. Benchling [Internet]. 2024. Available from: https://benchling.com 816 89. Posit team. RStudio: Integrated Development Environment for R [Internet]. Boston, MA: 817 Posit Software, PBC; 2023. Available from: http://www.posit.co/ 818 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint 27 90. Bates D, Mächler M, Bolker BM, Walker SC. Fitting Linear Mixed -Effects Models Using 819 lme4. J Stat Softw [Internet]. 2015 Oct 7;67(1):1 –48. Available from: 820 https://www.jstatsoft.org/article/view/v067i01 821 91. Wickham H, Averick M, Bryan J, Chang W, D’ L, Mcgowan A, et al. Welcome to the 822 Tidyverse. J Open Source Softw [Internet]. 2019 Nov 21;4(43):1686. Available from: 823 https://joss.theoj.org/papers/10.21105/joss.01686 824 92. Hadley Wickham. ggplot2: Elegant Graphics for Data Analysis [Internet]. Springer -Verlag 825 New York; 2016. Available from: https://ggplot2.tidyverse.org 826 93. David Wilkins. gggenes: Draw Gene Arrow Maps in “ggplot2” [Internet]. 2020. Available 827 from: https://CRAN.R-project.org/package=gggenes 828 94. Thomas Lin Pedersen. patchwork: The Composer of Plots [Internet]. 2025. Available from: 829 https://patchwork.data-imaginist.com 830 95. Hartig F . DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression 831 Models. CRAN: Contributed Packages. 2016. 832 96. Lenth R V . emmeans: Estimated Marginal Means, aka Least -Squares Means. CRAN: 833 Contributed Packages. 2017. 834 835 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 6, 2025. ; https://doi.org/10.1101/2025.08.04.668416doi: bioRxiv preprint