Removal of AMR plasmids using a mobile, broad host-range CRISPR-Cas9 delivery tool

Abstract

Antimicrobial resistance (AMR) genes are widely disseminated on plasmids. Therefore, interventions aimed at blocking plasmid uptake and transfer may curb the spread of AMR. Previous studies have used CRISPR-Cas-based technology to remove plasmids encoding AMR genes from target bacteria, using either phage- or plasmid-based delivery vehicles that typically have narrow host ranges. To make this technology feasible for removal of AMR plasmids from multiple members of complex microbial communities, an efficient, broad host-range delivery vehicle is needed. We engineered the broad host-range IncP1-plasmid pKJK5 to encode cas9 programmed to target an AMR gene. We demonstrate that the resulting plasmid pKJK5::csg has the ability to block the uptake of AMR plasmids and to remove resident plasmids from Escherichia coli. Furthermore, due to its broad host range, pKJK5::csg successfully blocked AMR plasmid uptake in a range of environmental, pig- and human-associated coliform isolates, as well as in isolates of two species of Pseudomonas. This study firmly establishes pKJK5::csg as a promising broad host-range CRISPR-Cas9 delivery tool for AMR plasmid removal, which has the potential to be applied in complex microbial communities to remove AMR genes from a broad range of bacterial species.

Keywords: AMR gene removal; CRISPR-Cas plasmids; antibiotic resensitization; antimicrobial resistance; broad host-range plasmids; plasmid curing.

Fig. 1.. Insilico CRISPR-Cas9 cassette construction and pKJK5 recombineering. (a) Source of genes included in cassette (cas9, sgRNA, GFPmut3b) and alterations undertaken. See (c) for sgRNA details. Gene length is indicated in base pairs (bp). (b) Final cassette layout. Gene lengths are to scale; spacings, restriction sites, promoters, terminators and ribosome binding sites are not. (c) sgRNA region in detail. Highlighted in red: nucleotide mutations introduced in upper stem region to form SacI restriction site. The region to be exchanged for N20 specificity exchange is indicated with blue crossover lines. (d) Homologous recombineering allowed insertion of the CRISPR-Cas9 cassette into dfrA, disrupting this gene in pKJK5’s accessory gene load. See the Methods section for details.

Fig. 2.. pKJK5::csg as a barrier to plasmid acquisition. Transformation efficiency of E. coli DH5α+pKJK5::csg[aacC1]/[nt] with pHERD20T (untargeted plasmid) or pHERD30T (targeted plasmid). [aacC1] transformation with pHERD30T did not yield any transformants, data points are displayed as ½ of the limit of detection. Grey box, data points underneath the limit of detection. n, 6, diamonds, mean, circles, individual data points.

Fig. 3.. pKJK5::csg can conjugatively remove resident target plasmids. Means (diamonds) and standard deviation (lines) of various colony counts. (a) Proportions of recipients with various plasmid content. Colony counts on plates selecting for recipients+target plasmid, recipients+pKJK5, or recipients+both plasmids divided by colony counts on plates selecting for recipients only, giving proportions of recipients with various plasmid contents; n=6. The dotted line indicates 100 %. *These treatments are significantly different as analysed by fitting a binomial GLM followed by Tukey’s post-hoc test, P<0.05. (b) Cell density on various selective plates. Colony counts on all different selective plates given in colony-forming units per ml of culture (c.f.u. ml⁻¹); n=6. LB, LB agar without selection. K, kanamycin, selects for recipients. G, gentamicin, selects for pHERD30T. T, tetracycline, selects for pKJK5::csg. A generalized linear model and Tukey’s post-hoc test revealed that numbers of colonies are different between treatments for all selective media; P<0.001; F=273.9; d.f.=11 and 60; R²=0.98. See the Methods section for details.

Fig. 4.. pKJK5::csg prevents transformation of various isolates with a targeted plasmid. Transformation efficiency of various isolates carrying pKJK5::csg[aacC1] or pKJK5::csg[nt] with target plasmid pHERD30T. Diamonds and lines indicate mean±standard deviation, points indicate individual replicates; n=3–6. Shaded areas indicate the limit of detection; counts of 0 were manually set to ½ of the limit of detection. bhiF2, C743E1, TV1-2, 6TB-1, coliform pig faeces, environmental and human isolates. PA14, Pseudomonas aeruginosa PA14. SBW25, Pseudomonas fluorescens SBW25. Transformation efficiency of aacC1 and nt treatments are significantly different for all strains; P<0.001 as assessed by Tukey’s HSD after fitting an inverse Gaussian GLM; F=347.6; df=11 and 48; P<2.2×10⁻¹⁶; adjusted R²=0.9858.