Phage-Plasmids Spread Antibiotic Resistance Genes through Infection and Lysogenic Conversion

Abstract

Antibiotic resistance is rapidly spreading via the horizontal transfer of resistance genes in mobile genetic elements. While plasmids are key drivers of this process, few integrative phages encode antibiotic resistance genes. Here, we find that phage-plasmids, elements that are both phages and plasmids, often carry antibiotic resistance genes. We found 60 phage-plasmids with 184 antibiotic resistance genes, providing resistance for broad-spectrum-cephalosporins, carbapenems, aminoglycosides, fluoroquinolones, and colistin. These genes are in a few hot spots, seem to have been cotranslocated with transposable elements, and are often in class I integrons, which had not been previously found in phages. We tried to induce six phage-plasmids with resistance genes (including four with resistance integrons) and succeeded in five cases. Other phage-plasmids and integrative prophages were coinduced in these experiments. As a proof of concept, we focused on a P1-like element encoding an extended spectrum β-lactamase, bla_CTX-M-55. After induction, we confirmed that it is capable of infecting and converting four other E. coli strains. Its reinduction led to the further conversion of a sensitive strain, confirming that it is a fully functional phage. This study shows that phage-plasmids carry a large diversity of clinically relevant antibiotic resistance genes that they can transfer across bacteria. As plasmids, these elements seem plastic and capable of acquiring genes from other plasmids. As phages, they may provide novel paths of transfer for resistance genes because they can infect bacteria that are distant in time and space from the original host. As a matter of alarm, they may also mediate transfer to other types of phages. IMPORTANCE The dissemination of antimicrobial resistance is a major threat to global health. Here, we show that a group of temperate bacterial viruses (phages), termed phage-plasmids, commonly encode different and multiple types of resistance genes of high clinical importance, often in integrons. This is unexpected, as phages typically do not carry resistance genes and, hence, do not confer upon their hosts resistance via infection and genome integration. Our experiments with phage-plasmids isolated from clinical settings confirmed that they infect sensitive strains and render them antibiotic resistant. The spread of antibiotic resistance genes by phage-plasmids is worrisome because it dispenses cell-to-cell contact, which is necessary for canonical plasmid transfer (conjugation). Furthermore, their integrons become genetic platforms for the acquisition of novel resistance genes.

Keywords: antibiotic resistance; bacteriophages; integrons; phage genomics; plasmids; prophage induction.

FIG 1

Number of mobile genetic elements encoding ARGs. The values after the bars indicate the number of elements encoding ARGs over the total number of elements considered in the analysis.

FIG 2

Genetic environments of ARGs in P-Ps’ pangenome graphs. Nodes are gene families. Genes (including ARGs) were grouped into a gene family (default parameters of PPanGGOLiN [31]) if they had a similarity of >80% identity covering at least 80% of the sequence. Orange: persistent, green: shell. Edges are shown for adjacent genes within the gene families (genetic contiguity). Gene families with colored numbers (red, blue, violet, green) are direct neighbors of ARGs containing genetic elements (transposon, IS, integron, recombinase [separated by colors]). Black numbers are given for proximal gene families. ABg1: 1-3,6,8-9:hypothetical, 4:nucleoside, 5:ATPase AAA, 7:3′–5′ exonuclease pyrophospho-hydrolase. ABg2: 1,3:hypothetical, 2,7:ribonucleoside-diphosphate red. 4:toprim domain protein, 5:ATPase AAA, 7:3′–5′ exonuclease pyrophosphohydrolase. SSU5_pHCM2: 1:PhoH, 2-5, 7-11:hypothetical, 6:DNA ligase. P1: 1:SSB, 2-5,15,17,19-20,23-24,28: hypothetical, 6:cell division inhibitor (Icd-like), 7-11:tail fiber, 12:recombinase, 13-14:tail fiber assembly, 16:ResMod subM, 18:Ref family, 21:bleomycin hydrolase, 22:transglycosylase, 25:DNA repair, 26:Phd/YefM (T-A), 27:doc (T-A), 29:lysozyme, 30:head processing. pKpn: 1:transcriptional regulator, 2,7,9,12-15,17-19:hypothetical, 3:phoH, 4:porphyrin biosynthesis protein, 5-6: AAA family ATPase, 8: ribonucl.-diphosphate reductase subunit 10–11:tail fiber domain-containing protein, 16:HsdR. pSLy3: 1:DUF3927 family, 2:tellurite/colicin resistance, 3-7:hypothetical.

FIG 3

Integrons encoded in P-Ps. Genomic organization of integrons found in P-Ps, arranged by P-P groups and coaligned by the class I Intl1 integrase. The P-P type is highlighted by different colors. Gene-to-gene assignment is based on a BLASTP comparison when the alignment is at least 90% identical and covers at least 90% of the sequences. Blue gene arrows indicate the integrase genes, red arrows show the AMR genes (99% identity with 99% coverage to reference sequences), and green arrows represent the rest of the genes within the integrons. Numbers indicated above nonhomologous cassettes represent different types of ARGs: (1,14): ant(2′')-Ia; (2): aac(6′)-33; (3;13;19): qacEΔ1; (4): aac(3)-Ib; (5): dfrA15; (6): arr-2; (7): group II intron reverse transcriptase/maturase; (8): aac(6′)-Ib; (9): bla_CARB-2; (10;17;25): aadA2; (11): cmlA6; (12): catB11; (15;20): dfrA12; (16;21;24): DUF1010 protein; (18): aac(3)-VIa; (22): bla_GES-1; (23): arr-6.

FIG 4

Induction of P-Ps and prophages in CRE strains. (A) CRE strains with ARG-encoding P-Ps were induced by 5 μg/mL MMC. 4 h after induction, phage particles were purified, and chromosomal DNA (gDNA) was removed by DNase I digestion. The phage capsid was degraded by proteinase K, and the virion DNA was purified and sequenced. The obtained short reads were coassembled with long reads from the genomic sequencing experiment (see Materials and Methods). The assemblies were compared to P-P and phage genomes and subsequently assigned. The read mapping coverage was computed (by mapping the short reads from the MMC experiment on them). The reads that did not map to the assemblies were used to compute the background coverage caused by the undigested gDNA (by mapping to genomic contigs obtained via the long read assembly). (B) In the genome of the P1-like P-P of the E. coli 163A9, the CTX-M-55 gene is found next to two DDE transposases. The ARGs encoded in the SSU5-like P-Ps from the E. cloacae strains 169C2, 170E2, 171A5, and 211G7 are in a complex region containing transposases and integrons. Homology assignments between P-Ps were done when the sequence similarity was at least 80% identity and covered 80% of the sequence of the gene (retrieved from an all-versus-all BLASTP comparison). The similarity between P-Ps is indicated by the weighted gene repertoire relatedness (wGRR). (C) The average read coverage was obtained and calculated as described in panel A. All contigs (= ctg) larger than 10 kb are shown. The coverage (a.u.: arbitrary unit) is plotted on a logarithmic x axis for the P-P contigs (P1 = P1-like, SSU5 = SSU5-like, N15 = N15-like) (orange), the contigs assigned to prophages or virion loaded DNA (blue), and the average background coverage (the rest of the coverage) obtained after mapping the remaining reads to genomic contigs (gray) for each tested CRE strain.

FIG 5

Lysogenic conversion of different E. coli strains (55989 [2], CIP 105917 [3], CIP 53.126 [4], and CIP 76.24 [5]) by the P1-like P-P of strain 163A9 [1]. (A and B) After the infection and plating experiment, four tested E. coli strains acquired resistance to carbenicillin. Examples of colonies of strain 55989 with the P1-like P-P of 163A9 are shown on LB agar plates with 100 μg/mL carbenicillin (left in panel A). Prior to infection, E. coli 55989 cells do not have the resistance (control, right in panel A). The original host of the P1-like P-P and all four lysogens are resistant to 100 μg/mL ampicillin (B). (C and D) Their genomes were isolated and sequenced by short reads (Fig. S12). The read coverage for the P-P genome (C) and the chromosomes of the host strain E. coli 55989 (D) are shown. (E) Genome comparison of the P-P from 163A9 and P1, coaligned to the first gene of P1. The alignment is matched with the read coverage plot in panel C. The functions of the P1 genes were retrieved from Łobocka et al. (78).