Scarless Removal of Large Resistance Island AbaR Results in Antibiotic Susceptibility and Increased Natural Transformability in Acinetobacter baumannii

Curing AbaR in A. baumannii using a scarless genome editing strategy.To cure A. baumannii of the AbaR island while restoring the original insertion site, we devised a scarless genome editing strategy. This method, depicted in Fig. 1, allows deletion, insertion, or single-base mutation. In contrast to other methods, it does not require cloning or prior genetic engineering of the target strain (17–19). Rather, it relies on the use of chimeric PCR products and takes advantage of natural transformation, a phenotypic trait exhibited by most isolates of A. baumannii. The method is based on a two-step selection (Fig. 1): first the introduction of a selectable/counterselectable sacB_aacC4 cassette at the loci of interest (Fig. 1B), and then the replacement of the cassette by the desired sequence (Fig. 1C). First, a chimeric PCR product is assembled, consisting of the sacB_aacC4 cassette bearing a selection marker (aacC4) conferring resistance to apramycin, flanked by 2-kbp-long sequences identical to the target region (Fig. 1B). This chimeric PCR product is introduced by natural transformation, and A. baumannii transformants are selected using apramycin, an antibiotic to which most clinical isolates are sensitive. A second PCR product is assembled, still carrying the homologous flanking regions but whose sequence is designed to produce deletion, insertion, or single-base mutation (deletion in Fig. 1C). Genetic replacement and the loss of the previously inserted sacB gene is then selected using sucrose resistance selection. We applied this method to delete AbaR11 in strain AB5075. The strain AB5075 is an A. baumannii ST1 strain isolated from a bone infection in 2008 with multidrug resistance (aminoglycosides, fluoroquinolones, and β-lactams, including carbapenems) (20). This strain bears a 19.7-kbp-long AbaR11 inserted in the comM gene with 19 predicted genes corresponding to five modules found in other AbaRs (Fig. 2A) (4, 8, 21). In a previous study, we identified the strain AB5075 as naturally transformable (15). Using natural transformation of PCR chimeric products, the AbaR island was replaced upon genetic recombination by the sacB_aacC4 cassette (Fig. 2B). Then, the strain bearing the sacB_aacC4 cassette in the comM gene was subjected to transformation with an assembly PCR encompassing the insertion region cured of AbaR, thereby carrying an intact comM gene. Then we performed genetic analysis of sucrose-resistant and apramycin-susceptible clones to select an AbaR-cured strain (named AB5075-R) with a restored comM gene presumably encoding a full-length ComM protein (Fig. 2C).

• Open in new tab
• Download powerpoint

FIG 1

PCR-based genome editing of transformable A. baumannii. Schematic representations of steps for gene inactivation (dark blue gene) using natural transformation. (A) Annealing sites for primers of PCR assemblies on their respective templates: either A. baumannii genomic DNA for amplifications of 2 kbp upstream and downstream of the deleted region (Δ) or the laboratory plasmid pMHL-2 that bears the counterselectable cassette (light blue) with a sacB gene conferring sucrose sensitivity and the aacC4 gene encoding resistance to apramycin. Note that primers P2, P3, and P7 share annealing sequences with primers P5, P6, and P8, respectively. (B) Chimeric PCR product (orange) obtained after two-step PCR assembly of products P1 + P2 with P5 + P6 and P3 + P4. This chimeric product bears the sacB_aacC4 cassette flanked by two 2-kbp-long regions of homology with the recipient cell’s genome (gray areas). These homologous regions allow the recombination of the chimeric PCR product into the recipient cell’s genome, resulting, upon apramycin selection of the transformants, in the intermediary genotype represented in panel C. (C) Chimeric PCR product obtained after two-step assembly PCR of products P1 + P8 with P7 + P4 bearing the deleted gene flanked by homologous regions (gray areas), allowing recombination and replacement of the counterselectable cassette in the intermediary recipient genome. (D) The transformants are selected based on their sucrose resistance resulting in the final scarless edited genome.

• Open in new tab
• Download powerpoint

FIG 2

Genetic engineering of AbaR curing in A. baumannii. (A) Schematic representation of AbaR11 inserted in the comM gene in A. baumannii AB5075 genome. All the sequences and genes in gray color belong to the wild-type AB5075 chromosome, the interrupted comM gene is, however, highlighted in dark blue. Transposons Tn6019 and Tn6018 constituting AbaR11 are indicated as in reference 4. For illustration purposes, the scale of the AbaR island is not respected. (B) Intermediary genetic construct with the counterselectable cassette replacing the AbaR11 island. (C) Final genetic structure of the locus after restoring the comM gene integrity.

Resistance profiles associated with AbaR curing in two clinically relevant multidrug-resistant strains.We then sought to investigate the role of AbaR island in the multidrug resistance profile of strain AB5075. This AbaR contains 19 open reading frames (ORFs) with, besides transposon-related proteins, seven ORFs predicted to confer resistance to metals, three of unknown function, and the sup gene encoding a putative sulfate permease present in most AbaRs (Fig. 2A). Antibiotic susceptibilities were thus compared between the AB5075 parental strain and its AbaR-cured derivative, denoted AB5075-T (Table 1). As anticipated, the resistance profiles to 14 antibiotics belonging to five antibiotic classes were identical between the wild type and its AbaR-cured derivative. We then investigated the contribution to antibiotic resistance of a larger AbaR found only in another A. baumannii ST1 strain, the clinical strain AYE resistant to aminoglycosides, β-lactams (excluding imipenem, piperacillin-tazobactam, and ticarcillin-clavulanate), fluoroquinolones, and tetracycline and was of intermediate susceptibility to rifampicin (5, 22). This strain carries the largest AbaR described so far, AbaR1, at 86 kbp long, containing 25 genes putatively involved in resistance to antibiotics of several classes. By exploiting natural transformation of the strain AYE (15), we cured this strain of its AbaR1 island and compared the resistance profiles to 21 antibiotics of the parental strain to that of its AbaR-cured AYE-T derivative (Table 2). In contrast to strain AB5075, curing the AbaR island from the strain AYE resulted in restored susceptibility to aminoglycosides, tetracycline, and sulfonamide. The resistance armamentarium conferred by the 86-kbp-long AbaR1 may be attributed to specific genes that are listed in Table 2 according to previous analyses (4, 5, 23). Restored resistances to sulfonamides and tetracycline are consistent with the previous report of AbaR27 deletion in strain A424 leading to the loss of two copies of sul1 genes and one tetA gene (8).

In the AYE-T strain, partially decreased resistance to ticarcillin, ceftazidime, cefepime, and aztreonam was also observed. Only a partial loss of resistance to β-lactams upon AbaR1 curing is likely explained by the additional resistance mechanisms the AYE strain possesses, such as the overexpression of the bla_ADC-11 gene encoding a narrow-spectrum chromosomal cephalosporinase (22, 24). In AYE-T strain ceftazidime, Etest values decreased from 12 μg/ml to 3 μg/ml in medium supplemented with cloxacillin (250 μg/ml), the latter known to inhibit the cephalosporinase, whereas no significant differences were observed for cefepime, which is not a substrate for ADC-11 (25). Other chromosomal resistance determinants most likely contribute to the reduced susceptibility to cefotaxime, for instance, the predicted new class A β-lactamase on AYE chromosome (24). Despite the four putative chloramphenicol resistance genes on AbaR1, their deletion did not modify the high intrinsic resistance of A. baumannii to this antibiotic mediated by efflux (26). Deletion of AbaR1 resulted in a minor change of rifampicin susceptibility.

Contrarily to the previous report on AbaR27 deletion, we did not observe for both strains, AB5075 and AYE, an increased resistance to fluoroquinolone of the AbaR-cured strains (8).

Beyond the role of AbaR in GC1 resistance, this method should allow investigation of other genomic islands found in A. baumannii as long as strains are transformable. For instance, as strains of GC2 are seemingly transformable, scarless removal of the AbGRI1 resistance island found in this clone is anticipated to be feasible (27, 28).

Curing of AbaR and repair of the comM gene result in increased transformation ability.AbaR islands are inserted in the comM gene (4). In the nonpathogenic species A. baylyi, a mutation of the latter gene is associated with reduced transformability (9). We therefore pursued this by investigating the role of comM in A. baumannii as well as the physiological consequence of AbaR-island disruption of this gene. To do so, we performed transformation assays comparing the levels of transformation frequency between the wild-type and AbaR-cured strains using an rpoB mutation (I581F) conferring resistance to rifampicin as a genetic marker (Fig. 3). For both clinical strains, curing AbaR and thus restoring the comM gene increased significantly the ability to acquire the genetic marker, with a mean increase of approximately 1,000-fold in strain AB5075-T (Fig. 3A) and of approximately 300-fold in strain AYE-T (Fig. 3B). With one bacterium of 100 that performed natural transformation, strain AB5075-T presented the highest transformation ability (Fig. 3A). Using the same genome editing strategy as depicted in Fig. 1, we generated markerless deletions of the comM gene (ΔcomM) in both A. baumannii clinical strains. Consistently, deletion of the comM gene in both strains resulted in transformation levels comparable to levels observed for the wild-type strains bearing a genuine insertion in the comM gene by the AbaR island (Fig. 3A and B). Thus, prior to acquisition of AbaR, both strains were transformable at higher levels, and the subsequent disruption generated transformation defects that are consistent with previous observations made in other species or genera (9, 10). The data show that the AbaR-inactivated comM gene, once cured of AbaR, is still functional. This suggests that it has not been affected by mutational drifts, confirming that acquisition of AbaR occurred relatively recently (4).

• Open in new tab
• Download powerpoint

FIG 3

Transformation frequencies of A. baumannii strains AB5075 (A) and AYE (B) of a 3.8-kbp-long PCR product carrying a mutation in the rpoB gene conferring resistance to rifampicin (final concentration of 50 ng/μl, 125 ng per sample). For both strains, either the wild type or their “T” (ΔAbaR/repaired comM) or ΔcomM derivatives were tested. At least seven independent transformation assays were performed on at least two separate occasions represented by separate dots and corresponding boxplots. All results were above the limit of detection (10⁻⁸). Pairwise comparisons using the nonparametric Mann-Whitney-Wilcoxon test (two-tailed) gave P values of at least <0.005 between transformation frequencies obtained for the wild-type/ΔcomM strains and the T derivatives.

The fact that acquisition of the AbaR element in the comM gene alters the natural transformability of A. baumannii and its capacity for further acquisitions by natural transformation is intriguing. Interestingly, acquisition of AbaR and consequent comM interruption did not hinder GC1 subsequent acquisition of different capsule gene clusters or gyrA or parC alleles leading to fluoroquinolone resistance (29). The significance of lower transformability is not known, but theoretical work suggests that intermediate transformation rates allow bacteria to maximize fitness in fluctuating stressful environments such as repeated antibiotic stress (30).