3.1 Integration of fluorescence tags at conserved ARG regions.
In this study, we developed a new tool and used the sfGFP fluorescence tag to track the spread of ARG plasmid in bacterial communities. Unlike previous studies, where a fluorescence tag was either inserted in a resistance gene or integrated randomly on a plasmid21, 22, 23, our approach targets the conserved regions associated with different ARGs. These previous techniques were limited in their ability to quickly label various plasmids containing the same ARG, and they also posed a risk of causing unintended biological consequences due to the random nature of the insertions. To address these limitations, we developed a new fluorescence tagging system for plasmids that targets conserved regions associated with different ARGs. ARGs are often linked to different mobile genetic elements (MGEs), such as the, carbapenemase genes blaKPC, blaNDM and blaOXA−48, which are associated with Tn4401/NTMKPC24, Tn125 25and Tn199926, respectively. The mcr-1 is often found in complete or partial Tn6330 with a core mcr-1-pap2 cassette 27, 28. Our analysis of NCBI database genomes of blaKPC, blaNDM, blaOXA−48 and mcr-1 genes showed that many of these gene cassettes have conserved regions across different plasmid types (as shown in Fig. 1a). For instance, the mcr-1-pap2 cassette (approx. 2000 bp) in Tn6330 is highly conserved in mcr-1-containing IncX4, IncI2 and IncHI2 plasmids.
Our analysis also indicated that the regions surrounding the AMR genes were consistently transferred along with the AMR genes during horizontal gene transfer. This made these regions ideal targets for fluorescence-based tracking. Our new system targets the conserved regions associated with specific ARGs, such as blaKPC, blaNDM, blaOXA−48 and mcr-1. This allows us to label different clinically relevant plasmids containing the same ARGs, regardless of their plasmid background, using the same fluorescence tagging system. If the ARGs are located on the chromosome, our design can also insert the sfGFP gene into the chromosome.
Six clinical isolates containing different ARGs were selected for fluorescence tagging. These isolates were obtained from E. coli, S. Typhimurium and K. pneumoniae and carried mcr-1 (CSZ4, 14E1050 and LS3479), blaKPC−2 (50700), blaNDM−1 (HZE22), and blaOXA−48 (49210). The mcr-1 gene was located on plasmids pCSZ4 (IncX4), p14E1050 (IncI2), and pLS3479 (IncHI2), while blaNDM-1, blaKPC-2 and blaOXA-48 were located on plasmids pHZE22 (IncHI2), p50700 (IncF) and p49210 (IncL) respectively (Table S2 and Fig. 1b).
3.2 Two plasmid CRISPR/Cas9 system was used for fluorescence tagging.
A two plasmid CRISPR/Cas9 system (consisting of plasmid pCas9-oriT and pSGKP-gfp) was utilized for fluorescence tagging (Table S2, Fig. 2a). Plasmid pCas9-oriT was engineered by adding an oriT region to the thermosensitive plasmid pCas9-apr, which provids Cas9 and lambda Red proteins [29]. This plasmid enabled the transfer of Cas9 and lambda Red genes to clinical isolates through conjugation. The plasmid pSGKP-gfp expressed the sgRNA under the control of the J23119 promoter29, superfolder green fluorescent protein gene (sfGFP) under the control of the pTrc promoter, the sucrose-sensitive gene sacB 30 as well as the homologous DNA fragment targeting the AMR gene. 500bp upstream and downstream elements of the sfGFP insertion site were PCR amplified and ligated to pSGKP-gfp at the XbaI site. The sfGFP gene on the plasmid pSGKP-gfp is regulated by a modified pTrc promoter, which can be repressed by the LacIq protein (Fig. 2b). Four pSGKP-gfp plasmids (pSGKP-arr2-TKPC500GFP, pSGKP-arr2-TNDM500GFP, pSGKP-arr2-TOXA500GFP and pSGKP-Tmcr-500GFP), targeting blaKPC, blaNDM, blaOXA−48 and mcr-1 gene cassettes were constructed.
To tag sfGFP gene in the conserved downstream region of the ARGs, the pCasKP-oriT plasmid was transferred into the clinical isolates E. coli CSZ4, S. Typhimurium 14E1050, S. Typhimurium LS3479, K. pneumoniae 50700, E. coli HZE22 and K. pneumoniae 49210, respectively, through conjugation. Then, the plasmid pSGKP-gfp with different homologous template (e.g. pSGKP-arr2-TKPC500GFP, pSGKP-arr2-TNDM500GFP, pSGKP-arr2-TOXA500GFP and pSGKP-Tmcr-500GFP) was electrotransformed into the aforementioned strains harboring pCasKP-oriT. After induction with L-arabinose, the sfGFP fluorescent tag was successfully inserted into the region downstream of mcr-1, blaNDM−1, blaKPC−2, blaOXA−48 in the strains E. coli CSZ4, S. Typhimurium14E1050, S. Typhimurium LS3479, E. coli HZE22, K. pneumoniae 50700 and K. pneumoniae 49210, respectively (Fig. 2c). Further PCR screening and laser scanning confocal microscope (LSCM) showed that all these strains were successfully labelled with sfGFP at the targeted sites (Fig. 2d-e). The labeling efficiency ranged from 11.09% (pLS3479) to 100% (85.85% for p14E1050, and 100% for pCSZ4, pHZE22, p50700 and p49210), as shown in Fig. 2F.
3.3 Fluorescence labeling did not interfere normal functions of the bacterial strains.
To assess the effects of fluorescent labeling on host strains, we compared the minimum inhibitory concentration (MIC) values and growth curves of sfGFP-labeled strains to those of their parent strains. The results showed no differences in the MIC values of CST, MEM, or AMP in E. coli CSZ4, S. Typhimurium 14E1050, S. Typhimurium LS3479, E. coli HZE22, K. pneumoniae 50700, and K. pneumoniae 49210 between the sfGFP-labeled strains and their parent strains (Fig. 3a). Additionally, the growth kinetics of the sfGFP-labeled and parent strains were found to be the same (Fig. 3b and Fig S2). These findings suggested that the fluorescent labeling did not alter the antimicrobial susceptibility of the AMR genes or impose any significant fitness costs on the bacterial hosts.
We then examined whether the fluorescent tagging would impact the plasmid conjugation frequency. Our conjugation experiments using both broth and agar mating found no obvious differences in conjugation frequency between the sfGFP-labelled strains and their parent strains (Fig. 3c-d). We also investigated if the insertion with sfGFP segment would impact the transposition frequency. We cloned the Tn6330 harboring mcr-1 or mcr-1-sfGFP into the suicide plasmid pJS01 28, 31, producing the plasmid pJS05 (containing a 4673 bp Tn) and pJS06 (containing a 5582 bp Tn) (Fig. 3e-g), respectively. The transposition frequencies into E. coli MG1655 were 1.48 ± 0.66 × 10− 6 and 1.98 ± 0.21 × 10− 6 per transformed cell for pJS05 and pJS06, respectively. No significant transposition frequency differences were observed between pJS05 and pJS06 (Fig. 3h). These results indicated that the sfGFP gene labeling at the downstream conserved-region of ARGs did not impact antibiotic susceptibility, growth characteristics, conjugation frequency, or transposition frequency.
3.4 Monitoring ARGs transfer in complex microbiota.
To improve the detection of ARGs transfer from clinical isolated isolates to complex microbiota, we constructed a non-transferable ColE1 plasmid pMQ. The plasmid pMQ expresss mCherry fluorescence protein and lacIq, which inhibits the sfGFP promoter pTrc if both plasmids co-existed. The design is to introduce the pMQ into sfGFP tagged clinical isolates, and the presence of pMQ will inhibit the sfGFP expression, resulting in only red fluorescence due to mCherry gene in pMQ. When the sfGFP labeled ARG plasmid transfers to other recipients, the sfGFP promoter inhibition disappear, leading to green fluorescence emission (Fig. 4a). This system allowed easy distinction and quantification of transconjugant cells using flow cytometry or rapid detection via a confocal microscope.
We then experimentally evaluated the design. The plasmid pMQ was transformed into E. coli CSZ4-sfGFP, K. pneumoniae 49210-sfGFP, E. coli HZE22-sfGFP to track the transfer of IncX4-mcr-1, IncL-blaOXA−48 and IncHI2-blaNDM−1 plasmids, respectively. As shown in Fig. 4B, E. coli C600 was fluorescent negative, while E. coli CSZ4-sfGFP, K. pneumoniae 49210-sfGFP, E. coli HZE22-sfGFP showed green fluorescence. Upon introduction of pMQ into E. coli CSZ4-sfGFP, K. pneumoniae 49210-sfGFP or E. coli HZE22-sfGFP, only red fluorescence was produced. In the conjugation system, both red and green fluorescence were observed, representing donor pMQ/AMR plasmids-harboring strains and E. coli C600 transconjugants, respectively. The ratio of green to red was higher in IncX4-mcr-1, IncL-blaOXA−48 mating cultures (Fig. 4b), indicating a higher conjugation frequency of the two plasmids. To confirm the fluorescence detecting results, we randomly selected ~ 10 green and red colonies, conducted PCR and Sanger sequencing, and found that the green colonies were all from donors, while the green colonies were all from the E. coli EC600 transconjugants. These results indicated that our CRISPR-AMRtracker system successfully detected the AMR plasmid transfer in vitro.
The high conjugation frequency of the mcr-1-carrying IncX4 plasmid pCSZ4 (31229 bp, Fig S3) from E. coli CSZ4 to E. coli C600 strain prompted us to select it for further studies. To investigate the potential of IncX4-mcr-1 transfer in mixed bacterial populations, we firstly created a mixed recipient populations of four Gram-negative species: E. coli 25922, S. Typhimurium ATCC 14028, K. pneumoniae ATCC and A. baylyi ADP1. Using our above described pMQ/AMR-sfGFP system, we studied the IncX4-mcr-1 transfer in the mixed population using LSCM. Our results showed the fluorescence of both red and green cells in the conjugation cultures of E. coli 25922, S. Typhimurium ATCC 14028 and K. pneumoniae ATCC 700603 (Fig. 4c). To validate the fluorescence tracking results, we further conducted conjugation assay using single donor and recipient strains. The results showed that plasmid pCSZ4-sfGFP transferred successfully from donor to the recipient strains, including E. coli C600, S. Typhimurium ATCC 14028 and K. pneumoniae ATCC 700603, with frequencies of 1.12 ± 0.12 × 10− 1, 7.03 ± 3.07 × 10− 5 and 2.45 ± 0.01 × 10− 5, respectively (Fig. 4d), but not to A. baylyi ADP1. This is consistent with the mixed population plasmid transfer results, supporting the potential application of our system in monitoring plasmid-mediated ARG transfer in natural mixed bacterial populations, such as the fecal microbiota.
3.5 Plasmid pCSZ4-sfGFP transfer in natural fecal microbiota community.
We then used the IncX4-mcr-1 CRISPR-AMRtracker system to examine plasmid-mediated transfer of ARGs in the fecal microbiota community under both aerobic and anaerobic conditions (Fig. 5a). The conjugation frequency of plasmid pCSZ4-sfGFP was 1.04×10− 3 in the aerobic condition, which was about 5-fold lower under anaerobic conditions (2.00×10− 4) (Fig. 5b). Transconjugants generated under aerobic conditions were isolated via cell sorting (FACS) and characterized by 16S rRNA gene amplicon sequencing. The recipient sample yielded 156 OTUs at the genus level, compared to 20 OTUs in the sort transconjugants sample. The sequenced transconjugants belonged to seven different genera, including Escherichia, Shigella, Brevibacillus, Lactobacillus, Staphylococcus, Enterococcus and Bacillus (Fig. 5c). To validate these results, we conducted the fluorescence-based mating approach on five pure strains of different species, including B. subtilis 163, B. subtilis ATCC 6633, E. faecalis ATCC 29212, S. aureus ATCC 29213 and L. reuteri 03. Green fluorescent cells were observed in all mating mixtures, except for B. subtilis ATCC 6633 (Fig. 5d).