Characterization of the diversity of type IV secretion system-encoding plasmids in Acinetobacter

ABSTRACT

The multi-drug resistant pathogen Acinetobacter baumannii has gained global attention as an important clinical challenge. Owing to its ability to survive on surfaces, its capacity for horizontal gene transfer, and its resistance to front-line antibiotics, A. baumannii has established itself as a successful pathogen. Bacterial conjugation is a central mechanism for pathogen evolution. The epidemic multidrug-resistant A. baumannii ACICU harbours a plasmid encoding a Type IV Secretion System (T4SS) with homology to the E. coli F-plasmid, and plasmids with homologous gene clusters have been identified in several A. baumannii sequence types. However the genetic and host strain diversity, global distribution, and functional ability of this group of plasmids is not fully understood. Using systematic analysis, we show that pACICU2 belongs to a group of almost 120 T4SS-encoding plasmids within four different species of Acinetobacter and one strain of Klebsiella pneumoniae from human and environmental origin, and globally distributed across 20 countries spanning 4 continents. Genetic diversity was observed both outside and within the T4SS-encoding cluster, and 47% of plasmids harboured resistance determinants, with two plasmids harbouring eleven. Conjugation studies with an extensively drug-resistant (XDR) strain showed that the XDR plasmid could be successfully transferred to a more divergent A. baumanii, and transconjugants exhibited the resistance phenotype of the plasmid. Collectively, this demonstrates that these T4SS-encoding plasmids are globally distributed and more widespread among Acinetobacter than previously thought, and that they represent an important potential reservoir for future clinical concern.

KEYWORDS:

• Acinetobacter baumannii
• bacterial pathogen
• secretion system
• horizontal gene transfer
• antibiotic resistance
• carbapenem resistance

Introduction

The multi-drug resistant (MDR) Gram-negative bacterium Acinetobacter baumannii has become a pathogen of serious concern for both healthcare and research communities, responsible for a diverse array of opportunistic infections including respiratory tract, bloodstream, soft tissue, and wound infections [1,2]. Hospital-acquired A. baumannii poses a particular concern for patients in intensive care units (ICUs) and burn wards [3], where it is estimated to account for over 20% of bacterial infections in ICUs worldwide [4]. Incidences of community-acquired infection have also increased [1,5], with potential for severe disease and mortality rates as high as 64% [2,6]. Hypervirulent and antibiotic-resistant community-acquired strains have posed additional causes for concern [7,8].

A. baumannii belongs to a group of clinically problematic bacteria termed ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) pathogens. Commonly resistant to front-line antibiotics including carbapenems [1,2], A. baumannii has been recognized by both the World Health Organization (WHO) and Centers for Disease Control and Prevention (CDC) as a pathogen in urgent need of attention by the research community. A. baumannii is resistant to desiccation, and can survive on surfaces for over a month [3]. Surviving hospital grade disinfectants, it can be spread to patients on ventilators [9], accounting for up to 50% of ventilator-associated bacterial pneumonia [10]. A. baumannii also has a remarkable propensity to uptake exogenous DNA that includes antibiotic resistance genes, contributing to MDR and XDR phenotypes [3].

A. baumannii is a member of the Acinetobacter calcoaceticus-baumannii (ACB) complex, a group of important clinical pathogens [11,12]. The ACB complex includes A. calcoaceticus, A. baumannii, A. nosocomialis, A. pittii, A. seifertii, and A. dijkshoorniae [11,12]. Other species of clinical importance include A. haemolyticus, A. junni, A. johnsonii, A. lwoffi, A. ursingii, and A. schindleri [13,14]. Outside human-associated settings, Acinetobacter species, including A. baumannii, reside in diverse ecological niches, including soil, aquatic, and plant-associated environments, where it forms symbiotic relationships with animals, plants, and insects [13,15–20].

Bacterial conjugation represents a central mechanism of genomic plasticity [21]. An important form of horizontal gene transfer, conjugation mediates the contact-dependent transfer of DNA from a donor to recipient strain by means of a multi-subunit apparatus [21]. Environmentally and clinically, conjugation is a ubiquitous process that drives bacterial adaptation and evolution [21]. From a clinical perspective, conjugation can lead to the spread of genes that promote pathogen survival, antibiotic resistance, virulence, and biofilm formation [21]. Therefore, insight into the factors that promote conjugation can not only deepen our understanding of human pathogen evolution, but also of evolution in nature. The Type IV Secretion System (T4SS) represents an important apparatus to mediate symbiotic relationships [21–24]. Often encoded in gene clusters, this multi-subunit complex can mediate the transfer of DNA, protein or both [23–25]. In the context of conjugation, T4SSs penetrate both the inner and outer bacterial membranes, creating a translocon for DNA transfer [23,24].

While well-studied in model organisms such as E. coli [26,27], T4SS-mediated conjugation is less understood in Acinetobacter [28]. Genes encoding a T4SS apparatus with sequence similarity to the E. coli F-plasmid T4SS [29] were previously identified on seven plasmids in clinically important A. baumannii, including the epidemic MDR isolate A. baumannii ACICU [29,30], belonging to the global clone II (GC2) lineage[30]. These strains were classified as sequence type 2 (ST-2) using the Pasteur multilocus sequence type (MLST) scheme[30]. A recent analysis of A. baumannii plasmids identified related plasmids in several other sequence types [32], suggesting that these plasmids might be more widespread than previously anticipated. Here we examined the genetic diversity and geographical distribution of this group of T4SS-encoding plasmids, and investigated the potential for conjugative ability of an XDR member of this family.

Methods

Bacterial strains and culture conditions

A. baumannii ATCC 17978 was obtained from ATCC, and A. baumannii AB5075-UW was obtained from BEI Resources. Harbouring T4SS-encoding plasmid p1AB5075 [33], A. baumannii AB5075-UW served as a donor strain for conjugation. A. baumannii ATCC 17978 harbouring pMQ715 [34], provided as a kind gift from Dr. Robert Shanks, represented a tetracycline-resistant recipient for conjugation [35,36]. Strains were cultured in LB Broth (Miller) (BioShop), unless otherwise specified. Unless otherwise described, media was supplemented with the following antibiotics when appropriate: kanamycin (50μg/mL) (BioShop), ciprofloxacin (8μg/mL) (Sigma), and tetracycline (10μg/mL) (BioShop).

Identification of plasmids harbouring T4SS genes and MLST analysis

T4SS-encoding plasmids were identified using blastn [37]. The NCBI nucleotide collection was searched using the T4SS gene cluster from pACICU2 (Accession CP000865.1) [29] as the query and default blastn parameters. Duplicate plasmids sequences were excluded.

To assess the diversity of A. baumannii strains, MLST sequence types were identified using the previously established approach [38]. Where sufficient genome sequencing data was available, genome accession numbers were submitted to PubMLST (https://pubmlst.org/) as a query for A. baumannii typing using the MLST (Pasteur) scheme.

Pangenome phylogenetic analyses

All the genomic FASTA sequences were downloaded from GenBank using ape package version 5.7 [39] in R programming language version 4.3.2. To analyze the relationship between bacterial genomes, the genome was first annotated using Prokka software version 1.14.6 [40] with the argument “--kingdom Bacteria” in Python Conda environment (version 23.1.0). The pangenome alignment was performed using Roary version 3.13.0 with the arguments “-e –n –v –a” [41]. The phylogenetic tree was constructed using FastTree version 2.1.11-2 [42] with the default parameters. Annotating the plasmid genes, Prokka (version 1.14.6) was used. The pangenome alignment and phylogeny tree construction were performed using PEPPAN 1.0.6 using the default parameters [43]. The Newick tree format files generated from these analyses were visualized using the MEGA (version 11) [44] tool. All the code for the analysis is available at https://github.com/merlab/Acinetobacter_Phylo. Phylogenetic trees were visualized using iTOL [45].

Replicase analysis

To identify candidate replicase genes in T4SS-encoding plasmids, plasmid accession numbers were submitted for blastn analysis [37] against a previously established list of reference Acinetobacter replicases [32]. Open reading frames were curated manually using a coverage threshold of 80% [46], and for plasmids with unannotated sequences, candidate open reading frames were assessed using SnapGene (www.snapgene.com).

Gegenees plasmid analysis

Plasmid DNA sequences were aligned using Gegenees software [47]. Genbank sequences of full-length plasmids were submitted for analysis using recommended parameters [47] for shorter sequences (blastn, fragment size 200, step size 100), with the exception of two plasmids that generated errors that could not be resolved with assistance from the programme developers [47]. The resulting similarity matrix was plotted as a heat map, with plasmids denoted as Group 1 demonstrating <50% coverage to plasmids in Group 2, and plasmids within each group demonstrating >70% coverage relative to the reference plasmids pACICU2 and pABTJ1, respectively. For analysis of the core T4SS-encoding region, the associated core gene regions, spanning from the first gene to the last, were submitted using the above-described parameters.

Mauve plasmid analysis

The bioinformatics software Mauve [48] was used for comparative genetic analyses. Alignments were performed using progressiveMauve (https://darlinglab.org/mauve), using recommended default parameters and default HOXD scoring matrix. The sequence from pACICU2 was used as a reference. The recommended parameter seed families was chosen, because it improves anchoring sensitivity in regions below 70% identity [48].

T4ss gene cluster analysis

The previous study identified a cluster of 20 genes (traT, traD, traI, finO, traM, traA, traL, traE, traK, traB, trbG, traV, traC, traW, traU, trbC, traN, traF, traH and traG) across the seven A. baumannii plasmids (pAB_CC, pACICU2, ABKp1, pABTJ1, p1BJAB07104, p2BJAB0868 and p2ABTCDC071529) using E. coli T4SS proteins encoded on its F-plasmid as a comparison [26,29]. Internal assessment could not verify the same genes previously reported as finO, traA and traM [29], presumably due to increased genetic divergence among the plasmid sequences and/or differences in algorithm parameters. As the three-dimensional structure of several TraA homologs has been solved, the software Phyre2 was used to identify the Acinetobacter determinant. As such, finO and traM were excluded from this analysis.

To identify individual candidate T4SS-encoding genes, blast analyses were performed [37]. Protein sequences from plasmid p2ABTCDC0715 (Accession CP002524.1) were used as queries, as it harboured the most candidate T4SS-encoding genes in the original study [29]. For the remaining protein, TraE, the sequence from pABTJ1 (Accession CP003501.1, locus ABTJ_p0047) was used as a query [29]. More recently, the gene encoding traL is believed to lie directly upstream traE [32], consistent with the original annotation of p2ABTCDC0715 [49]. As such, locus ABTW07_2p012 was used for the TraL query. As many of the 119 plasmids identified contained unannotated and untranslated sequences, Tblastn was used to maintain a systematic analysis across all plasmids. Tblastn, like blastp, assesses conservation at the amino acid level. Results were curated using a coverage threshold of 80% [46]. For hits from unannotated sequences, results were curated for candidate open reading frames using SnapGene.

Identification of antibiotic resistance determinants

Candidate antibiotic resistance determinants were identified using Comprehensive Antibiotic Resistance Database (CARD) [50] and ResFinder 4.1 [51]. For CARD v3.1.0 (https://card.mcmaster.ca/), plasmid accession numbers were submitted through Resistance Gene Identifier, and resistance determinants were identified using RGI Criteria thresholds Perfect and Strict. For ResFinder 4.1 (https://cge.cbs.dtu.dk/services/ResFinder/), each plasmid sequence was submitted to the server as a fasta file, and resistance determinants were identified using an identity threshold of 90% and a minimal length of 60% [52,53]. Open reading frames were curated manually using a coverage threshold of 80% [46].

Assessment of mobile genetic elements

Candidate mobile genetic elements were identified using the MobileElementFinder [53] and ISFinder [54]. For MobileElementFinder (https://cge.cbs.dtu.dk/services/MobileElementFinder/), plasmid sequences were submitted for analysis using .fasta format. For ISFinder (https://isfinder.biotoul.fr/), plasmid nucleotide sequences were submitted for analysis with BLASTn, using default parameters.

Conjugation assays

Conjugation assays were performed as previously described [36]. Briefly, overnight cultures of donor and recipient were cultured overnight at 37°C and 250 rpm in LB supplemented with the appropriate antibiotic (kanamycin (50μg/mL) and ciprofloxacin (8μg/mL) for donor, tetracycline (10μg/mL) for recipient). Cells were washed twice in LB, and resuspended to an OD₆₀₀ of 1.0. Donor and recipient were mixed at a 1:1 ratio, and plated on LB-agar for overnight growth at 37°C. Cells were resuspended in LB and plated onto LB-agar supplemented with kanamycin and tetracycline. Candidate transconjugant colonies were assessed by replica plating onto LB-agar with and without each antibiotic. Conjugation frequencies were calculated as the number of transconjugants per recipient [55,56].

Characterization of transconjugants

Transconjugants were assessed for both genetic background and plasmid identity. For genetic background, profiling of MLST genes was performed using established primers for housekeeping genes rplB and fusA [38]. For plasmid assessment, diagnostic PCRs were performed. T4SS cluster primers (5’-AAAGGCTCAGATCCAGAACAAGTA-3’) and (5’-TGCAAAAGCTAAAATACCATCAAC-3’) amplified the region between traW, traU, and trbC. Antibiotic resistance region primers (5’-CGTGCTCGGTCTGTCTTGTGTTTC-3’) and (5’-GCCTGCGCTCAAACGGACATT-3’) amplified the region between cmlA1, aadA2, and aph(3”)−1b. Following PCR amplification, products were confirmed by sequencing (The Centre for Applied Genomics (TCAG) Facility, the Hospital for Sick Children).

Antibiotic inactivation assays

Antibiotic inactivation assays were performed on β-lactam resistant strains using a previously established approach [57]. Liquid cultures of LB supplemented with antibiotic (cefotaxime (20μg/mL)) were inoculated with the resistant strain of interest, and incubated overnight at 37°C and 250 rpm. Supernatants were used in disk diffusion assays, and putative inactivating strains were identified by the absence of a zone of inhibition. The following strain was used as a susceptible test organism: Micrococcus luteus GDW1580. The direct colony suspension method was used to prepare inocula to the 0.5 McFarland standard, and supernatants were spotted on sterile filter disks. M. luteus was grown on tryptone soya agar (BD), and plates were incubated at 30°C for 2 days.

Minimum inhibitory concentration (MIC) assays

Antibiotic susceptibility assays were performed in 96-well microtiter plates using the broth microdilution method using CSLI guidelines [58]. Cation-adjusted Mueller-Hinton Broth (BD) was used, and plates were incubated stationary at 37°C.

Results

Geographical distribution and genetic diversity of strains harbouring T4SS-encoding plasmids

To assess the genetic diversity and geographical distribution of the T4SS-encoding plasmids [29], a blast analysis was performed against the NCBI nucleotide collection using the T4SS gene cluster from pACICU2. In total, 118 plasmids were identified in four different species of Acinetobacter associated with human infection (Supplementary Table 1). The vast majority of strains were isolated from humans, although several were derived from Ovis aries (sheep), Sus scrofa domesticus (pig), and turkey hosts (Supplementary Table 1), suggesting that the plasmids are not restricted to domestic niches or human hosts. Almost half of the strains were noted to be from infections, and almost half were classified nosocomial. Strains included members of clinically important A. baumannii lineages GC1 and GC2, as well as nosocomial outbreak strains. More than half of strains were confirmed to be resistant to one or more antibiotic, including carbapenems, polymyxins, MDR or XDR strains.

With regard to global distribution, the T4SS-encoding strains were isolated from 20 different countries across 4 continents (Figure 1A-B, Supplementary Table 1). More than half of strains were isolated in Asia (62%), consistent with the original subset of plasmids [29]. Additionally, 19% of strains originated in North America, 7% were from Europe, and 4% were from Australia.

Figure 1. Global Distribution and Genetic Diversity of Strains Harbouring T4SS-encoding Plasmids. Strains that have no documented information were excluded. (A) Global distribution of strains harbouring T4SS-encoding plasmids. Pegs are placed within the country of origin for each strain, where colour represents MLST sequence type (Pasteur) for A. baumannii or species of Acinetobacter. (B) Pie chart indicating the country of origin of isolates. (C) Pie chart denoting the phylogenetic classification of strains. (D) Year of isolation of strains, plotted as a 5-year interval bar graph.

With respect to phylogenetic distribution, the plasmids were observed in a broader subset of host strains than anticipated. A. baumannii strains harboured 90% of plasmids [29], consistent with previous research [29,32]. However the host range spanned a broader range of A. baumannii sequence types than previously reported. Plasmids were identified in thirteen sequence types: 1, 2, 10, 20, 25, 79, 195, 345, 412, 422, 622, 1544, and 2252 (Figure 1C, Supplementary Figure 1, Supplementary Table 1). Ten percent of plasmids were identified in more divergent Acinetobacter strains. Eight percent were identified in A. seifertii, and one plasmid was from A. haemolyticus and A. sp. FDAARGOS_493 (Figure 1C, Supplementary Figure 1, Supplementary Table 1). Of special note, an additional plasmid was identified in Klebsiella pneumoniae (Figure 1C, Supplementary Figure 1, Supplementary Table 1). The majority of strains were isolated within the past two decades (Figure 1D, Supplementary Table 1). Collectively, this demonstrates that these T4SS-encoding plasmids are harboured in more genetically diverse host strains than previously anticipated.

Genetic diversity of T4SS-encoding plasmids

The genetic diversity of the T4SS-encoding plasmids was subsequently investigated (Figure 2A). The Acinetobacter plasmids spanned a wider size range than reported [29,32], from 54.6–166.7 kb, with a mean of 77.5 kb (Figure 2B, Supplementary Table 1). All quantifiable plasmids had a low GC-content (mean 33.8%) (Supplementary Table 1), consistent with that of Acinetobacter [59] and this subset of plasmids [29].

Figure 2. Genetic Diversity of T4SS-encoding Plasmids. (A) Schematic approach for the systematic study of T4SS-encoding plasmids. (B) Plasmid size, plotted as a bar graph in 5 kb intervals. (C and D) Heat maps of T4SS-encoding plasmids. The colours are plotted from green (high similarity) to red (low similarity), with the associated numbers reflecting the percentage similarity. Plasmids are designated as Group 1 (C) or Group 2 (D).

To quantify genetic diversity among the plasmids, the phylogenomic analysis software Gegenees [47] was used to generate a distance matrix based on sequence similarity at the DNA level. This phylogenomic analysis tool demonstrated that the plasmids lied in one of two groups. The larger group represented 93% of the plasmids, with sequence similarities of 53–100% relative to the smaller plasmid of each pairwise comparison (designated as Group 1 in Figure 2C). This included the originally described plasmids pACICU2, p2ABTCDC0715, pAB_CC, and ABKp1 [29]. This group of plasmids demonstrated >80% coverage relative to pACICU2, with 94% of plasmids having >90% coverage. The second group of plasmids represented the remaining 7% of plasmids, with 98–100% sequence similarity relative to the smaller plasmid of each pairwise comparison (Figure 2D, designated as Group 2). This included originally described plasmids pABTJ1, p2BJAB0868, and p1BJAB07104 [29]. This group of plasmids demonstrated >71% coverage relative to pABTJ1, with all but two plasmids having >90% coverage. The distribution of plasmids into two groups is consistent with the original study [29], and with pangenome phylogenetic analysis (Supplementary Figure 2).

The plasmids in Group 1 were harboured in diverse host strains, including all of the A. baumannii sequence types identified in this study, more divergent Acinetobacter (A. seifertii, A. haemolyticus, A. sp. FDAARGOS_493), and Klebsiella pneumoniae (Figure 1C). Interestingly, the latter plasmid was highly similar to two A. baumannii plasmids, pSSA12_1 and ABKp1 (Figure 2C). Recently a new classification system for A. baumannii plasmids was proposed [32]. Group 1 contained plasmids classified as lineage 1 (LN_1), consistent with this study [32]. Also consistent with this study, Group 1 plasmids harboured either a group GR6 replicase gene (93% of plasmids) or a GR6 replicase pseudogene or gene remnant (Supplementary Table 1) [32]. Several plasmids were found to harbour an additional replicase gene, either a second GR6 or GR32 (Supplementary Table 1).

Conversely, the plasmids in Group 2 were derived only from A. baumannii ST-2 strains (Figure 2C), although the sequence types of several strains have yet to be determined. Group 2 contained plasmids that belonged to plasmid lineage 5 (LN_5) [32]. Consistent with the literature, these plasmids harboured either a group GR25 replicase gene (88% of Group 2 plasmids) or a GR25 replicase gene remnant (Supplementary Table 1) [32].

To further examine plasmid divergence, comparative genetic analyses were performed with the software Mauve [48] (Figure 3, Supplementary Figures 3–10). Mauve performs multiple whole genome or plasmid alignments and identifies conserved genetic regions as locally collinear blocks (LCBs) [48], allowing for visualization of regions of divergence such as insertions or deletions. Across the plasmids, diverse genetic variations were observed, including the acquisition of genetic material both outside and within the T4SS-encoding gene cluster (Figure 3, blue and magenta boxes, respectively, Supplementary Figures 3–10). In addition to mobile genetic elements, these regions harboured diverse genes that included putative universal stress proteins, copper homeostasis gene clusters, membrane transporters, and antibiotic resistance determinants. Of interest to this research was the XDR strain A. baumannii AB5075-UW. Belonging to the GC1 lineage (ST-1), this strain was shown to be highly virulent in both Galleria mellonella and mouse pulmonary models of infection, and represents a model strain for pathogenesis studies [60]. The T4SS-encoding plasmid p1AB5075 was observed to harbour a large genetic region (∼17 kb) within the T4SS-encoding gene cluster not present in pACICU2 (Figure 3). This region was previously shown to contain mobile genetic elements flanking antibiotic resistance determinants to five different antibiotic classes [33]. Collectively, this suggests extensive evolution within this large group of T4SS-encoding plasmids.

Figure 3. Genetic Divergence Among T4SS-encoding Plasmids. Multiple whole plasmid alignments, where locally collinear blocks between plasmids are designated by Mauve in the same colour. (A) Alignment of plasmids with higher levels of sequence similarity to pACICU2. (B) Alignment of plasmids with lower levels of sequence similarity to pACICU2. The T4SS-encoding gene cluster in pACICU2 is indicated by a purple box. Selected regions of genetic rearrangement outside and inside the T4SS-encoding cluster are denoted by blue and magenta boxes, respectively. The plasmid size and MLST sequence type (Pasteur) for A. baumannii host strains are shown. TBD denotes to be determined.

Comparative analysis of T4SS-encoding region

To identify candidate T4SS-encoding genes in each of the 119 plasmids, blast analyses were performed using previously established T4SS-encoding proteins [29] as queries. The majority of hits demonstrated high levels of sequence identity at the amino acid level to the query sequences (>98% with coverage levels of >95%) (Supplementary Table 3). In 77% of plasmids, all eighteen T4SS-encoding genes queried were identified (Figure 4A). These plasmids were distributed among most of the A. baumannii sequence types identified, in addition to A. seifertii, A. haemolyticus, and Acinetobacter sp. FDAARGOS_493 (Figure 4B).

Figure 4. Diversity of T4SS Gene Cluster Organization in Acinetobacter Plasmids. (A) Number of intact T4SS-encoding genes identified, plotted as a bar graph. A total of 17 genes were assessed. (B) Pie chart denoting the phylogenetic classification of strains harbouring plasmids with all seventeen T4SS-encoding genes. (C) A topological representation of the T4SS transfer apparatus in Acinetobacter, based on the predicted structure [26]. (D) Gene organization in Acinetobacter T4SS-encoding regions. The previously established cluster in ABKp1 is shown as a comparison. Linear gene maps show all open reading frames within the genetic regions, with the corresponding size of each region shown. T4SS genes are colour coded based on (C), and all other open reading frames are shown in white. Asterisks (*) indicate candidate pseudogenes, and h indicates an additional more divergent homolog.

To assess genetic diversity within the core T4SS-encoding region, distance matrices based on sequence similarity [47] were constructed for Group 1 and Group 2 plasmids (Supplementary Figure 11A and 11B, respectively). Both Group 1 and 2 T4SS-encoding regions demonstrated high overall levels of similarity. Group 1 T4SS-encoding region sequence similarities ranged from 63–100% relative to the smaller plasmid of each pairwise comparison, with 96% of these pairwise comparisons having similarities of 80% or higher (Supplementary Figure 11A). Group 2 T4SS-encoding region sequence similarities ranged from 99–100% relative to the smaller plasmid of each pairwise comparison (Supplementary Figure 11B).

To visualize T4SS-encoding gene regions identified, linear gene maps were constructed. The overall relative organization of T4SS-encoding genes was conserved across the majority of the 119 plasmids (Figure 4C-D, Supplementary Figures 12–22). Several plasmids harboured regions that may reflect genetic rearrangements from an ancestral plasmid. For example, pAC30c harboured a region where traT is oriented in the opposite direction relative to the rest of the plasmids (Figure 4D), suggesting a possible inversion. Similarly, pCMCVTAb2-Ab66 contains two genes (traE pseudogene and traL) that are oppositely oriented (Supplementary Figure 21). Other plasmids contained T4SS-encoding regions that were larger than most plasmids, consistent with our observation of candidate insertions within the T4SS-encoding clusters (Figure 3). A number of plasmids harboured additional regions within the plasmid with T4SS-encoding genes. For example, pARLG_6295_1 contained two additional regions with T4SS-encoding genes (Figure 4D). Collectively, this demonstrates that while the T4SS-encoding genes are highly conserved, genetic diversity within this region exists.

Antibiotic resistance determinants harboured by T4SS-encoding plasmids

Acinetobacter has established itself as an emerging pathogen of high clinical importance in large part due to its capacity for horizontal gene transfer, leading to the acquisition of resistance to front-line antibiotics [3,9]. The original study of this group of T4SS-encoding plasmids noted the absence of detectable resistance determinants [29]. However given that this group of plasmids is considered to be putatively self-transmissible, larger, more widely distributed among Acinetobacter and more genetically diverse than previously appreciated, an analysis of antibiotic resistance determinants harboured on these plasmids is of importance. A systematic analysis of the 119 plasmids performed using Comprehensive Antibiotic Resistance Database (CARD) [50] and ResFinder 4.1 [51] resulted in the detection of an abundance of diverse resistance genes. Of the 119 plasmids, 47% were observed to have at least one determinant (Supplementary Table 3). Of these plasmids, 70% were observed to have one determinant, with 30% having multiple determinants (Figure 5A). Fourteen plasmids harboured two determinants, one contained six determinants, and two harboured eleven determinants (p1AB5075 and pAb8098). Of importance, the plasmids with resistance genes were not restricted to A. baumannii, as 70% of A. seifertii plasmids harboured a resistance determinant (bla_OXA-82) (Supplementary Table 3). Across the 119 plasmids, 95 resistance genes were detected, consisting of sixteen different determinants (Figure 5B) and spanning 6 different target drug classes (aminoglycosides, β-lactams, sulfonamides, diaminopyrimidines, phenicols, and antiseptics) (Figure 5C). Aminoglycoside resistance determinants included aminoglycoside phosphotransferases (aph(3’)-Via, aph(3'’)-Ib, aph(6)-Id and aph(3’)-VI), nucleotidyltransferases (ant(2'’)-Ia and aadA2 (ant(3'’) family)), and an acetyltransferase (aac(6’)-Ib10) (Figure 5B). A number of β-lactamase determinants were also frequently detected (bla_OXA-23, bla_OXA-82, bla_OXA-164, bla_GES-11, and bla_GES-14) (Figure 5B), which confer resistance or reduced susceptibility to carbapenems. The sulfonamide resistance determinant sul1, encoding a resistant variant of dihydropteroate synthase, was detected on three plasmids. Other resistance determinants included the diaminopyrimidine resistance gene dfrA7, and the phenicol and quaternary ammonium compound efflux pumps, cmlA1 and qacEΔ1, respectively.

Figure 5. Antibiotic Resistance Determinants in T4SS-encoding Plasmids. (A.) Pie chart denoting the frequency of resistance gene identification. (B.) Pie chart indicating the distribution of antibiotic resistance determinants among the plasmids. (C) Pie chart reflecting the antibiotic resistance determinants from (B) in the context of targeted drug class. (D) Pie chart reflecting the antibiotic resistance determinants from (B) in the context of resistance mechanism.

The drug classes most commonly associated with the resistance genes were aminoglycosides (46% of determinants) and β-lactams (42% of determinants) (Figure 5C), with the most common genes being aph(3’)-VIa (34% of determinants) and bla_OXA-23 (31% of determinants) (Figure 5B). Nine different resistance gene profiles were observed, with the most common being aph(3’)-VIa (29% of plasmids with determinants), bla_OXA-23 (27% of plasmids with determinants), aph(3’)-VIa and bla_OXA-23 (27% of plasmids with determinants), and bla_OXA-82 (13% of plasmids with determinants). With regard to resistance gene mode of action, enzymatic antibiotic inactivation was the most prevalent strategy encountered (Figure 5D), representing 88% of genes. Antibiotic target replacement and efflux were also observed, representing 6% and 5% of genes, respectively.

Across the 56 plasmids harbouring resistance determinants, the majority of resistance genes were detected at a conserved location on the plasmid (Figure 6, Supplementary Table 3). This included the A. seifertii plasmids carrying bla_OXA-82 (Supplementary Table 3). Conversely, the bla_OXA-23 determinant, identified on 29 plasmids, was identified at several different sites (Figure 6, Supplementary Table 3). This highlights the importance of genome plasticity to A. baumannii’s evolution as a drug-resistant pathogen of concern.

Figure 6. Antibiotic Resistance Genes in T4SS-encoding Plasmids. Multiple whole plasmid alignments of selected plasmids, where locally collinear blocks between plasmids are designated by Mauve in the same colour. The T4SS-encoding gene cluster in pACICU2 is indicated by a purple box. Antibiotic resistance determinants are designated by stars at the associated location, where the colour corresponds to the legend below.

Horizontal gene transfer is often mediated by mobile genetic elements such as insertion sequences and transposons. To provide additional insight, an analysis for candidate mobile genetic elements was performed with MobileElementFinder and ISFinder [53,54]. Mobile genetic elements were detected in 80% of the T4SS-encoding plasmids (Supplementary Table 4). With respect to the plasmids with resistance determinants, > 80% harboured mobile genetic elements in proximity of the resistance determinants (Supplementary Tables 3–4). The majority of mobile genetic elements were detected within 1.6 kb of the resistance determinant.

Transfer of XDR T4SS-encoding plasmid between A. baumannii strains

Bacterial T4SSs are a functionally diverse class of secretion apparatus that mediates the translocation of genetic material and/or secreted proteins that can facilitate bacterial fitness and survival, or contribute to symbiotic relationships [22]. From the perspective of pathogenic bacteria and conjugation, T4SSs can mediate the transfer of genes that encode antibiotic resistance, virulence factors, and beneficial metabolic functions [22]. The T4SS-encoding region of this large family of plasmids exhibits sequence similarity to the F-plasmid conjugative T4SS in E. coli [25,29], however the functional ability remains poorly studied. As such, we investigated the potential for plasmid transfer by conjugation.

Given that multi-drug resistance and plasmid-mediated transfer of antibiotic resistance genes represent important clinical challenges for A. baumannii [21], A. baumannii AB5075-UW was selected for further study. To investigate the capacity for conjugative transfer of XDR plasmid p1AB5075 to A. baumannii of a more divergent sequence type, A. baumannii ATCC 17978 (ST-437) was selected as a recipient strain. In three independent assays, candidate transconjugants were observed, with a mean conjugation frequency of 9.4 × 10⁻⁴ (Supplementary Table 5). Of the candidate transconjugants screened, 100% exhibited a transconjugant resistance profile (Figure 7A, Supplementary Table 5). To probe for the presence of p1AB5075, diagnostic PCRs were performed for both the T4SS-encoding gene cluster and antibiotic resistance cluster. Successful amplification was observed for transconjugants and the donor strain (Figure 7B-C, Supplementary Figures 23–25). To verify the genetic background of transconjugants, a previously established approach [38] demonstrated that housekeeping genes rplB and fusA regions were identical to that of the recipient strain (Figure 7D-E). p1AB5075 harbours the gene bla_GES-11, encoding a β-lactamase resistance enzyme that hydrolyzes the β-lactam ring structure of carbapenems and other β-lactam antibiotics, resulting in functional inactivation [33,60]. Antibiotic inactivation assays [57] with β-lactam resistant strains showed that transconjugant culture media grown in the presence of cefotaxime did not contain active drug (Figure 7F), consistent with resistance by antibiotic inactivation. Finally, minimum inhibitory concentration (MIC) assays performed on transconjugants showed antibiotic sensitivity levels of kanamycin, tobramycin and cefotaxime comparable to that of the donor strain (Table 1), suggestive of plasmid transfer. Collectively, this suggests that T4SS-encoding plasmid p1AB5075 is capable of transmission to a more divergent A. baumannii strain.

Figure 7. Conjugative Transfer of the XDR Plasmid p1AB5075 to A. baumannii ATCC 17978. (A) Resistance profiling of transconjugants. Kan, kanamycin, Cipro, ciprofloxacin, Tet, tetracycline. (B and C) Diagnostic PCR analysis of T4SS-encoding gene cluster (B) and antibiotic resistance gene cluster (C) in p1AB5075. (D and E) Host strain genetic profiling. Sequencing of housekeeping genes rplB (D) and fusA (E). (F) Antibiotic disk diffusion assays. Supernatants from antibiotic-supplemented cultures were assessed for the loss of activity against the indicator organism M. luteus.

Table 1. Minimum Inhibitory Concentrations (MICs) of Antibiotics Against A. baumannii Strains.

	MIC (μg/mL)
Strain	Donor	Recipient	Transconjugant #1	Transconjugant #2
Kanamycin	>256	8	>256	>256
Tobramycin	64	1	32	32
Cefotaxime	>256	16	>256	>256
Ciprofloxacin	64	≤0.5	≤0.5	≤0.5

Discussion

Bacterial conjugation represents an important driver of genomic evolution, responsible for the transfer of genes encoding diverse cellular processes [21]. In Gram-negative bacteria, the T4SS serves as an instrumental apparatus to mediate this transfer [21–24]. A. baumannii is recognized by both the WHO and CDC as an emerging pathogen in urgent need of attention by the research community. Of special concern, this pathogen has evolved a remarkable ability to incorporate exogenous DNA into its genome that includes antibiotic resistance determinants, resulting in MDR and XDR strains [3]. Here we present a systematic study to examine the genetic diversity of a group of T4SS-encoding Acinetobacter plasmids [29]. We describe a group of 119 plasmids spanning A. baumannii strains from thirteen sequence types, and non-A. baumannii strains including A. seifertii, A. haemolyticus, and Klebsiella pneumoniae. A. seifertii and A. haemolyticus are also considered to be emerging pathogens of clinical importance [61,62], and carbapenem-resistant and/or MDR strains have been identified both clinically and environmentally [15,61,63–66]. A. seifertii is associated with serious human infections including bloodstream infections [64,65]. A. haemolyticus is commonly nosocomial, and is associated with infections including endocarditis and respiratory infections [66,67].

The widespread of nature of the Acinetobacter genus in the environment, including A. baumannii, is well recognized [13,15,16]. Diverse Acinetobacter species, including A. baumannii, have been isolated from an array of environmental settings such as soil and water ecosystems, plant-associated niches, and animals, including agricultural settings [13,15–20]. In this study, several T4SS-encoding plasmids were identified in environmental settings (sheep, pig and turkey hosts), including an A. baumannii plasmid that harboured β-lactam and aminoglycoside resistance genes bla_OXA-23 and aph(3')-Via, respectively. This highlights the potential importance of this environmental reservoir of the antibiotic resistome.

Outside the Acinetobacter genus, this work identified the plasmid pE16KP0301-4 from Klebsiella pneumoniae. Genetically, this plasmid clustered with Group 1 (Figure 2C) and demonstrated high levels of sequence similarity to two other A. baumannii plasmids, with all 3 strains isolated in South Korea. MLST analysis indicated that this strain is in fact Klebsiella pneumoniae, rather than a misclassification. Given that plasmid replicases with 100% identity to A. baumannii LN_1 replicases have been identified in Klebsiella pneumoniae [32], this collectively suggests that the host range of this group of plasmids can extend beyond the Acinetobacter genus to other pathogens of concern.

From a geographical perspective, the group of 119 T4SS-encoding plasmids were globally distributed across 20 countries spanning 4 continents, with more than half of strains isolated from Asia and 19% from North America. As a future perspective, it would be interesting to examine the level of overrepresentation, if any, in the countries within these regions of the world.

Group 1 plasmids represented the largest subset of T4SS-encoding plasmids (108 plasmids in total). Genetic diversity was observed, with comparative analyses suggesting candidate insertions and deletions (Figure 2C and 3, Supplementary Figures 1–8). Genes of diverse putative function were harboured in these candidate insertion regions, including putative universal stress proteins, copper homeostasis gene clusters, and antibiotic resistance determinants. For example, pAba10042 harbours large regions not observed in comparison plasmid pACICU2 (Figure 3). A candidate acquisition within the T4SS gene cluster includes the aminoglycoside resistance gene aph(3’)-Via (Supplemental Table 3), and a region outside the cluster includes a putative universal stress protein with homology to A. baumannii universal stress protein A (UspA) [68]. UspA was previously shown to contribute to adapting to H₂O₂-associated oxidative stress response, acidic pH stress, and pathogenesis [68]. The plasmid pS21-2 harbours several candidate acquisitions (Figure 3). A large region outside the T4SS-encoding cluster includes putative copper resistance genes copC, copper-translocating ATPase, copS, copR, copA, copB, cusB, cusA, and cusF. Copper resistance determinants facilitate in mitigating the toxic effects of high environmental copper concentrations, and homologous genes in A. baumannii have been shown to contribute to virulence [69]. Collectively, this suggests that this subset of putative conjugative plasmids may confer additional functions to transconjugants which could be beneficial during infection.

Interestingly, based on sequence similarity, Group 1 plasmids often most closely resembled plasmids from differing A. baumannii host sequence types. For example, several ST-622 plasmids showed highest levels of similarity to ST-2 plasmids, whereas other ST-622 plasmids most closely resembled ST-1 plasmids (Figure 2C). Similarly, while A.seifertii represents a more divergent Acinetobacter, these plasmids more closely resembled A. baumannii ST-2 and ST-20 plasmids (Figure 2C). This is suggestive of many incidences of plasmid transfer over the course of evolution.

From the perspective of T4SS-encoding gene clusters, the majority of plasmids harboured all eighteen genes queried (Figure 4A), with high levels of sequence identity to the reference. While the overall T4SS-encoding gene organization was conserved across the majority of plasmids, evidence of genetic divergence was observed (Figure 4D). In addition to candidate gene rearrangements and insertions within the T4SS-encoding cluster, additional regions with T4SS-encoding genes were observed in a number of plasmids outside the core T4SS gene cluster. Plasmid pARLG_6295_1 presents an interesting example of genetic divergence, with two additional regions with T4SS-encoding genes outside the core (Figure 4D). In total, pARLG_6295_1 harbours two copies of each T4SS-encoding gene, and three copies of traL. In the future, it would be interesting to investigate whether the additional genes could have evolved differential regulation or differing functional abilities.

Acinetobacter’s notoriety as an emerging pathogen of high clinical importance is largely attributed to its propensity to acquire exogenous genetic material, resulting in resistance to front-line antimicrobials such as carbapenems and the evolution of MDR and XDR lineages [3,9]. Given that this large group of Acinetobacter plasmids has the potential for self-transmissibility, the presence of resistance determinants to clinically important antibiotics is of concern. Systematic analyses indicated that almost half of T4SS-encoding plasmids harboured at least one resistance determinant, with 30% of these plasmids having multiple determinants (Supplemental Table 3). Of special concern was high frequencies of resistance genes to β-lactams and aminoglycosides, given its growing concern in the healthcare community. As β-lactams and aminoglycosides represent clinically important treatments for Acinetobacter infections, the associated resistance determinants will be important to monitor for in the future. The diversity of resistance genes was also higher than anticipated, with sixteen different determinants spanning six drug classes (Figure 5B-C). Finally, the observation of plasmids with six or more resistance determinants was of special concern, given the detection of all T4SS-encoding genes (Supplemental Table 2). Should these strains prove to be conjugation competent, variants of these plasmids would be valuable to monitor for as more strain sequencing data accumulates in the future.

Conjugation is a main mode of horizontal gene transfer between bacteria, largely contributing to genome diversification and evolution [70]. From the perspective of pathogen evolution, it represents an important mechanism of spread for antibiotic resistance genes and virulence factors [70]. Recently, a new classification system for A. baumannii plasmids defined T4SS-encoding plasmids from this work as belonging to two lineages of A. baumannii plasmids, LN_1 and LN_5 [32], consistent with our observations. This study also highlighted another lineage of plasmids (LN_18) with two members [32], one of which has been shown to be self-transmissible, capable of transferring the carbapenem resistance gene blaN_DM-1 to diverse Acinetobacter [56]. Importantly, this classification describes LN_1 as the largest A. baumannii plasmid lineage, therefore this group of plasmids will be important to study in the future. Given that the T4SS-encoding plasmids described in this work remain poorly understood from the perspective of transmission, we examined the ability of the XDR p1AB5075 to transfer to a more divergent A. baumannii strain. Our findings of plasmid transfer from a ST-1 strain to ST-437 strain suggest a self-transmissible plasmid, and transfer of resistance to cefotaxime, kanamycin, and tobramycin highlight the potential future clinical importance. Future studies with a T4SS knockout plasmid variant will be valuable to further support the role of the T4SS-encoding gene cluster.

This study demonstrates that this group of T4SS-encoding plasmids is larger and more widespread within pathogenic Acinetobacter than previously anticipated. This work also investigated transmissibility, demonstrating that an XDR plasmid can transfer among A. baumannii and confer resistance to transconjugants. As very little is known about Acinetobacter pathogenesis, these insights could shed important light into how these pathogens have become clinically problematic. As incidences of resistance to all front-line antibiotics are rapidly increasing in A. baumannii, including clinical resistance to the last resort antibiotic colistin [71], new therapeutic approaches are required combat MDR globally. The T4SS apparatus in Acinetobacter and other Gram-negative pathogens represents an attractive candidate target. Unlike traditional antibiotics that function by inhibiting bacterial growth, secretion system inhibitors do not select for mutations causing resistance [72]. Given that the WHO cites antibiotic resistance as one of the greatest global threats to human health, future investigation of T4SS inhibitors may represent a promising tool to allow the research community to stay one step ahead of antibiotic-resistant pathogens like Acinetobacter.

Acknowledgements

We thank members of the D’Costa Lab for helpful discussions and the Common Equipment and Technical Service (CETS) at the University of Ottawa for technical assistance with instrumentation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Operating Grants from the Canadian Institutes of Health Research (CIHR) (PJ4-175369 and PJT-178191) to V.M.D.