Intercellular Transfer of Chromosomal Antimicrobial Resistance Genes between Acinetobacter baumannii Strains Mediated by Prophages

The spread of antimicrobial resistance genes (ARGs) among Gram-negative pathogens, including Acinetobacter baumannii, is primarily mediated by transferable plasmids; however, ARGs are frequently integrated into its chromosome. How ARG gets horizontally incorporated into the chromosome of A. baumannii, and whether it functions as a cause for further spread of ARG, remains unknown. Here, we demonstrated intercellular prophage-mediated transfer of chromosomal ARGs without direct cell-cell interaction in A. baumannii.

ABSTRACT

The spread of antimicrobial resistance genes (ARGs) among Gram-negative pathogens, including Acinetobacter baumannii, is primarily mediated by transferable plasmids; however, ARGs are frequently integrated into its chromosome. How ARG gets horizontally incorporated into the chromosome of A. baumannii, and whether it functions as a cause for further spread of ARG, remains unknown. Here, we demonstrated intercellular prophage-mediated transfer of chromosomal ARGs without direct cell-cell interaction in A. baumannii. We prepared ARG-harboring extracellular DNA (eDNA) components from the culture supernatant of a multidrug-resistant (MDR) A. baumannii NU-60 strain and exposed an antimicrobial-susceptible (AS) A. baumannii ATCC 17978 strain to the eDNA components. The antimicrobial-resistant (AR) A. baumannii ATCC 17978 derivatives appeared to acquire various ARGs, originating from dispersed loci of the MDR A. baumannii chromosome, along with their surrounding regions, by homologous recombination, with the ARGs including armA (aminoglycoside resistance), bla_TEM-1 (β-lactam resistance), tet(B) (tetracycline resistance), and gyrA-81L (nalidixic acid resistance) genes. Notably, the eDNAs conferring antimicrobial resistance were enveloped in specific capsid proteins consisting of phage particles, thereby protecting the eDNAs from detergent and DNase treatments. The phages containing ARGs were likely released into the extracellular space from MDR A. baumannii, thereby transducing ARGs into AS A. baumannii, resulting in the acquisition of AR properties by the recipient. We concluded that the generalized transduction, in which phages were capable of carrying random pieces of A. baumannii genomic DNAs, enabled efficacious intercellular transfer of chromosomal ARGs between A. baumannii strains without direct cell-cell interaction.

INTRODUCTION

Acinetobacter baumannii is an opportunistic Gram-negative pathogen that causes severe infections, such as ventilator-associated pneumonia and bacteremia, especially in hospitalized patients, and is frequently associated with nosocomial outbreaks (1, 2). A. baumannii is becoming a serious public health concern, because the pathogen has developed multidrug resistance properties against carbapenems, aminoglycosides, and fluoroquinolones, which has limited the selection of antimicrobial agents for treating A. baumannii infections in clinical settings (3). The worldwide spread of multidrug-resistant (MDR) A. baumannii is attributed to the clonal spread of specific genetic lineages, such as the international clone II (IC2) (4, 5).

Carbapenem resistance in MDR A. baumannii depends primarily on the production of carbapenemases, which can hydrolyze carbapenems (6, 7). One of the most common mechanisms of increased aminoglycoside resistance in MDR A. baumannii is the production of the 16S rRNA methyltransferase ArmA, which prevents aminoglycosides from binding to the 30S ribosomal subunit through postmodification of a specific residue within the aminoacyl-tRNA binding site (A site) of 16S rRNA (8, 9). These horizontally acquired antimicrobial resistance genes (ARGs) are located on plasmids and/or the chromosome of A. baumannii, as well as on replicons of other Gram-negative pathogens, including members of the family Enterobacteriaceae (10, 11).

Complete genome sequence data for several MDR A. baumannii clinical isolates have been deposited in databases, with the data revealing that MDR A. baumannii possibly harbors numerous ARGs on its chromosome rather than on plasmids, which is contrary to what is documented in the case of MDR Enterobacteriaceae (12–14). The naturally high competence of A. baumannii would accelerate the efficient uptake of ARGs through the type IV pili and the Com membrane channel, resulting in their accumulative integration into the chromosome (15–17); however, how exogenous bacterial DNA fragments carrying ARGs get transferred and integrated into other A. baumannii chromosomes remains largely unclear. In addition, whether and how the chromosomal ARGs act as a source for ARG transfer to other A. baumannii strains also remain unanswered.

Mechanisms that enable DNA transfer across bacterial cells, without cell-cell interaction, had been previously reported. Outer membrane vesicles (OMVs) associated with or including DNA represent one of the sources for delivering DNA fragments to other bacterial cells (18–21). Accordingly, ARGs, such as bla_NDM-1 and bla_OXA-23, on plasmids can be transferred to A. baumannii by OMVs (22, 23). Moreover, the free DNA fragments released from chromosomes into the environment through cell lysis or an active machinery, such as the type IV secretion system, might serve as a means for augmented DNA transfer to other bacterial cells (24–27), with the same scenario is applicable for transducing phages and gene transfer agents (28–30). Recently, Krahn et al. demonstrated the transfer of chromosomal bla_NDM-1 and suggested the involvement of prophages in the chromosome (14); however, involvement of prophages in transferring chromosomal ARGs to A. baumannii remains incompletely elucidated. Here, we assessed the phenomenon associated with the transfer of chromosomal ARGs, using A. baumannii strains as the model, and identified the mechanisms mediating ARG transfer. Our results provide novel insights into the mechanism involved in the transfer of chromosomal ARGs across A. baumannii cells in the absence of direct cell-cell interaction.

RESULTS

Transfer of ARGs without direct cell-cell interaction between A. baumannii strains.

We used the A. baumannii NU-60 strain belonging to IC2 (sequence type 208) as the parent strain providing extracellular DNA (eDNA) components for our evaluation of potential ARG transfer, since it carries diverse chromosomal ARGs, including armA (aminoglycoside resistance), bla_TEM-1 (β-lactam resistance), and tet(B) (tetracycline resistance) (Fig. 1).

FIG 1.

Circular genomic map of the A. baumannii NU-60 strain. The middle black line represents G+C content, followed by the A. baumannii NU-60 genome (black), antimicrobial resistance genes (red), capsid protein genes (green), and prophages (blue).

We first collected the eDNA-containing components from the supernatants of 24-h liquid cultures of A. baumannii NU-60 by ultracentrifugation, followed by their addition to antimicrobial-susceptible (AS) A. baumannii ATCC 17978 and Acinetobacter baylyi ADP1 on agar plates. The latter two strains were used as the recipient strains for ARG transfer assays, since these have been previously prepared for various transformation assays involving acceptance of exogenous DNAs (31, 32). Subsequently, antimicrobial-resistant (AR) A. baumannii ATCC 17978 derivatives, with the acquired ARGs, were selected by culturing on plates containing amikacin (AMK), carbenicillin (CAR), or tetracycline (TET) (Fig. 2A). While the AMK-, CAR-, and TET-resistant derivatives were obtained on plate, no AR derivative appeared when A. baylyi was exposed to the eDNA-containing components. Moreover, exposure to purified DNA, extracted from bacterial cells using commercially available kits, did not produce AR derivatives on any of the plates containing the antimicrobial agents.

FIG 2.

(A) Appearance of antimicrobial-resistant A. baumannii ATCC 17978 derivatives following exposure to eDNA collected from the parent A. baumannii NU-60 strain. AMK, amikacin; CAR, carbenicillin; TET, tetracycline. (B) Number of CFU of antimicrobial-resistant derivatives obtained after exposure to eDNA collected from culture supernatants over 48 h. The eDNA collected from 20 ml of culture supernatant at each time point was used for one assay (n = 3; means ± standard deviations [SD]). (C) Quantification of eDNA collected from 20 ml of culture supernatants (black bars) and optical density (O.D.) at 600 nm of bacterial cultures (black circles) (n = 3; means ± SD). N.D., not detected.

We next evaluated the relationship between the appearance of AR derivatives and duration of culture before collecting the supernatants (Fig. 2B). AR derivatives were obtained when we used eDNA-containing components collected from supernatants at the exponential growth phase (6 h), with the appearance of AR derivatives increasing upon using components derived from cultures up to the stationary growth phase (24 h); however, the number of AR derivatives decreased drastically when the source cultures reached the phase of decline (48 h). The amount of eDNAs collected from supernatants changed almost in parallel with changes in the number of AR derivatives obtained and with bacterial growth (Fig. 2C); a discrepancy was observed at the 48-h time point, when larger amounts of DNA were collected, but the frequency of occurrence of AR derivatives was quite low (discussed later).

AR derivatives could acquire donor-derived ARGs and showed resistance to the corresponding antimicrobial agents.

We obtained AMK-, CAR-, and TET-resistant derivatives by selecting on AMK-, CAR-, and TET-containing agar plates, respectively.

Each derivative showed resistance to the corresponding antimicrobial agent similar to that of the eDNA-providing parent A. baumannii NU-60 (see Fig. S1 in the supplemental material). Furthermore, we confirmed that the genes armA, bla_TEM-1, and tet(B) were carried by the AMK-, CAR-, and TET-resistant derivatives (ATCC 17978ΩarmA, ATCC 17978Ωbla_TEM-1, and ATCC 17978Ωtet(B), respectively). These results together indicated that the eDNA fragments harboring the ARGs derived from the chromosome of A. baumannii NU-60 (Fig. 1) were released into the supernatant during liquid cultivation and that the ARGs were integrated into the recipient A. baumannii ATCC 17978 strain, thereby providing AR properties to the recipient.

Transfer of donor chromosomal armA to the recipient A. baumannii strain.

The genetic regions surrounding armA in the A. baumannii NU-60 strain are shown in Fig. 3A. The 19-kb genetic region containing armA was flanked by two copies of IS6 elements harboring 14-bp inverted repeats (IRL and IRR) at the terminal ends and showed the formation of a composite transposable element (composite transposon) known as Tn1548 (33, 34). This composite transposon has been frequently found on the chromosomal/plasmid DNAs of A. baumannii strains isolated from geographically distinct areas (11, 35, 36). The surrounding genetic regions of Tn1548 in the NU-60 strain were almost identical to those of the ATCC 17978 strain, indicating that Tn1548 was horizontally inserted into the chromosome.

FIG 3.

(A) Schematic representation of the genetic environment of 16S rRNA methyltransferase gene armA and its surrounding regions in A. baumannii NU-60 (parent). Its corresponding region of A. baumannii ATCC 17978 (recipient) (GenBank accession no. CP000521) is also depicted. The identified recombination regions of the A. baumannii ATCC 17978ΩarmA (derivative) strain are shaded blue. Gray shades indicate the region showing genetic identity of >93%. Open reading frames (ORFs) derived from A. baumannii NU-60 and ATCC 17978 strains are depicted in light green and yellow, respectively. ARGs, transposase genes/insertion sequences, and hypothetical genes are shown in red, green, and gray, respectively. (B) Schematic representation of the genetic environment of bla_TEM-1 and its surrounding regions in A. baumannii NU-60 (parent) and A. baumannii ATCC 17978 strains. The identified recombination regions of the A. baumannii ATCC 17978Ωbla_TEM-1 (derivative) strain are shaded blue. Gray shades indicate the region showing genetic identity of >79%. ORFs derived from the A. baumannii NU-60 and ATCC 17978 strains are depicted in the same colors as those for panel A.

We determined the transferred genetic region containing armA in the A. baumannii ATCC 17978ΩarmA strain (Fig. 3A), finding that it had captured an approximately 33-kb-long DNA fragment containing Tn1548 of A. baumannii NU-60 (NU60_1322 to NU60_1286; blue-shaded region in Fig. 3A), in exchange for the original genome region (A1S_2198 to A1S_2208; 11-kb; blue-shaded region in Fig. 3A). The crossover points for the recombination were determined by aligning the A. baumannii NU-60, ATCC 17978ΩarmA, and ATCC 17978 sequences. The 33-kb-long DNA fragment underwent recombination with the terminal ends at the internal positions of the genes A1S_2198 and NU60_1322 as well as A1S_2208 and NU60_1286 (Fig. 3A and Fig. S2). The ATCC 17978ΩarmA strain likely emerged by capturing the DNA fragment of the NU-60 strain through homologous recombination, using the highly conserved genetic regions/genes in both the donor NU-60 and recipient ATCC 17978 strains.

Moreover, we investigated the transferred DNA fragment that included the armA gene of another amikacin-resistant derivative (A. baumannii ATCC 17978ΩarmA-2), revealing that this ATCC 17978ΩarmA-2 strain had captured an approximately 34-kb-long DNA fragment, similar to that of the A. baumannii ATCC 17978ΩarmA strain; however, the nucleotide positions at which the recombination occurred was different. The right terminal end of the transposed fragment was within the A1S_2202 gene (Fig. 3A), and the left terminal end was located at the upper 12 kb of the A1S_2198 gene (not shown). The transferred genetic regions, including armA, differed across the investigated AMK-resistant derivatives. Nevertheless, these results showed that chromosomal armA, together with its long surrounding genetic regions, was transferred across A. baumannii without direct cell-cell interaction.

Transfer of donor chromosomal bla_TEM-1, tet(B), and gyrA-81L to the recipient A. baumannii strain.

The other noninnate ARGs examined, including bla_TEM-1 and tet(B), were also located, similar to armA, on the chromosome of A. baumannii NU-60 (Fig. 1). We subsequently identified the genetic region containing bla_TEM-1 that was transferred to A. baumannii ATCC 17978 (Fig. 3B) and found that bla_TEM-1, together with aminoglycoside resistance genes (aph, aad, and aac), was encompassed by multiple copies of IS26, forming a 20-kb mobile element. A 29-kb DNA fragment including bla_TEM-1 was incorporated into the chromosome of A. baumannii ATCC 17978Ωbla_TEM-1, with recombination having occurred at the terminal ends within A1S_1089 and NU60_2568 genes (left end) and upstream of A1S_1130 and NU60_2534 (right end), resulting in the loss of a 50-kb DNA fragment. The transfer of tet(B) was confirmed by nucleotide sequencing of A. baumannii ATCC 17978Ωtet(B) (Fig. S3). Collectively, our analyses revealed that ARGs, which were dispersed on the chromosome of the parental NU-60 strain, were capable of being transferred to the recipient ATCC 17978 strain.

Additionally, we investigated whether chromosomal, innate, and essential genes involved in antimicrobial resistance, such as gyrA for quinolone resistance, were transferred through a transfer mode similar to that used for armA, bla_TEM-1, and tet(B). We obtained the A. baumannii ATCC 17978ΩgyrA-81L strain, showing reduced susceptibility to nalidixic acid (NA), by using the aforementioned assay. A. baumannii ATCC 17978ΩgyrA-81L emerged through the exchange of gyrA from A. baumannii ATCC 17978 (which encodes a Ser at position 81 in the quinolone resistance-determining region [QRDR]) with that of A. baumannii NU-60 (Leu81 in QRDR) (Table S1). The approximately 30-kb DNA fragment, including the 9.5-kb upstream and 18-kb downstream regions of gyrA, was exchanged through homologous recombination. These results indicated that A. baumannii strains successfully captured and exchanged DNA fragments, including the housekeeping genes essential for bacterial proliferation, such as gyrA.

Identification of eDNA used for ARG transfer.

The eDNA found in culture supernatants was predicted to be present in diverse forms: naked forms released from lysed bacterial cells, OMV surface-associated forms, OMV-included forms, and phage-included forms. We performed transmission electron microscopy (TEM) analyses of the pellets containing antimicrobial resistance-conferring components, obtained after ultracentrifugation, and revealed the presence of OMVs of ∼200-nm diameter (Fig. 4A). Since OMVs are capable of carrying cargo of diverse bacterial materials, including lipids, proteins, RNA, and DNA (37), we evaluated the involvement of OMV-associated/included DNA in ARG transfer. The eDNA-containing components, collected from the 24-h bacterial culture supernatant through ultracentrifugation, were subjected to density gradient ultracentrifugation using OptiPrep (10 to 45%). SDS-PAGE and Western blot analysis, performed using an antibody against the outer membrane protein A (OmpA), a primary protein component of OMVs, revealed its presence in all separated fractions (Fig. S4) (38, 39). Aliquots of the fractions obtained after ultracentrifugation were used in ARG transfer assays, revealing that exposure of A. baumannii ATCC 17978 to fractions 3 through 10 stably yielded AMK- and CAR-resistant A. baumannii ATCC 17978 derivatives (Fig. 4B). Agarose gel electrophoresis revealed the presence of condensed DNA in fractions 2 through 12 at approximately 23 kb, which probably corresponded to the measurable DNA shown in Fig. 2C, and was susceptible to digestion by DNase; however, DNase treatment did not disturb the emergence of AMK- and CAR-resistant derivatives (Fig. 4B). Since DNase can access both naked DNA and OMV surface-associated DNA, the obtained DNA that was readily digested by DNase was unlikely to be used for the ARG transfer phenomenon observed in this study. Triton X-100 treatment is capable of lysing OMVs (40, 41), resulting in the leakage or release of DNA contained within OMVs; previous studies had shown Triton X-100 treatment prevents the transfer of ARG-carrying plasmids, wrapped within OMVs, to AS A. baumannii cells (22, 23). In the present study, Triton X-100/DNase double treatment failed to interrupt the emergence of AR derivatives (Fig. 4B), suggesting that OMV-containing DNA does not primarily contribute to the ARG transfer observed under the experimental conditions employed in this study.

FIG 4.

(A) TEM imaging of pelleted components of A. baumannii NU-60 culture supernatant. Bar, 200 nm. The arrows indicate OMVs. (B) Agarose gel electrophoresis of DNA-containing samples separated using an OptiPrep density gradient (10 to 45%). Samples were treated with DNase or left untreated and then electrophoresed. The samples in each fraction were treated with DNase or Triton X-100/DNase and subsequently used for the antimicrobial resistance transfer assay, which was performed three times independently. +, samples that yielded both AMK- and CAR-resistant derivatives more than twice in three trials. The result of the antimicrobial resistance transfer assay was assigned as positive upon observation of a single antimicrobial-resistant colony.

Involvement of prophage-encoded capsid proteins in the transfer of chromosomal ARGs.

TEM examination of OptiPrep-separated fractions revealed the predominant presence of OMVs in each fraction, in addition to small amounts of phage-like particles, including long tails either with or without phage head-like components (Fig. S5). We hypothesized that these components can internally store ARG-carrying DNA of the host bacterial strain and deliver it to other bacterial cells. To verify this hypothesis, we purified the OptiPrep-fractionated samples using size exclusion chromatography. Fractions 8 through 10, obtained after OptiPrep gradient ultracentrifugation (Fig. 4B), which stably generated AMK- and CAR-resistant derivatives and included only small amounts of OMV-derived components (Fig. S4), were collected and subjected to size exclusion chromatography. We observed two major peaks (Fig. 5A), and AR derivatives emerged when the fractions associated with the first peak were used in the ARG transfer assay. TEM examination of the components in the fractions (Fig. 5A, fractions a, b, and c) revealed the presence of phage particles featuring a hexagonal head (15 to 60 nm) with or without a long tail (200 to 250 nm), as well as the presence of small OMVs (30 to 50 nm) (Fig. 5B). Notably, the OMVs were lysed upon Triton X-100 treatment, whereas the phage head-like particles with long tails retained their original shapes after detergent treatment (Fig. 5B). Moreover, the emergence of AMK-, CAR-, and TET-resistant derivatives occurred when the three fractions (Fig. 5A, fractions a, b, and c) were used in the ARG transfer assay after treatment with Triton X-100. These results indicated that Triton X-100-tolerant phage particles represented a candidate agent mediating the ARG transfer.

FIG 5.

(A) Size exclusion chromatography. AU, arbitrary units. (B) TEM imaging of fractions (a to c) separated by size exclusion chromatography as shown in panel A. Arrows indicate fraction components other than OMVs, including phage particles with or without long tails. The samples in each fraction were treated with Triton X-100 or left untreated. (C) SDS-PAGE and silver staining of samples (fractions a to c). Lane M, protein size marker. (D) Summary table of Mascot identification of proteins assigned IDs 01 to 03 in panel C.

To identify the proteins associated with the phage particles, we performed SDS-PAGE using the three fractions (Fig. 5A, fractions a, b, and c). Three major bands (identifiers [IDs] 01, 02, and 03), commonly observed in all three fractions (Fig. 5C), were subjected to nano-liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis, followed by Mascot database searches, in order to identify the composition of the proteins (Fig. 5D). Two of them (IDs 01 and 03) were predicted to be uncharacterized (hypothetical) proteins, whereas the third one (ID 02) was identified as a capsid protein. Therefore, we predicted that oligomerized capsid proteins could carry, as internal cargo, DNA fragments, including ARGs of the host A. baumannii strain, and that the presence of capsid proteins could contribute to diminishing the susceptibility of DNA to Triton X-100/DNase treatment. A. baumannii NU-60 harbored the genes for two highly similar capsid proteins, here named capsid 1 (detected using nano-LC-MS/MS and Mascot analyses) and capsid 2 (Fig. 1 and Fig. S6). To evaluate the role of these capsid proteins in ARG transfer, we constructed three deletion mutants of the capsid protein genes (Δcapsid1, Δcapsid2, and Δcapsid1-Δcapsid2 mutants) in the parent NU-60 strain and measured the efficiency of ARG transfer (Fig. 6A). AMK- and CAR-resistant derivatives were generated with 20- and >40-fold lower efficiencies, respectively, when we used ultracentrifuged components from the Δcapsid1 mutant strain supernatant than when we used the corresponding components from the NU-60 parent strain. Conversely, deletion of the capsid 2 gene did not substantially alter ARG transfer efficiency; however, deletion of both capsid 1 and 2 genes attenuated the emergence of AR derivatives (i.e., the emergence was below the detection limit of the assay). These results indicated that capsid 1 is critically involved in enabling ARG transfer and that capsid 2 moderately promotes transfer. When providing ultracentrifuged components of the NU-60 parent strain treated with mitomycin C, increased efficiency of ARG transfer was observed in a dose-dependent manner, implying that excessive phages, induced via mitomycin C treatment, relatively enhanced the chances of transducing ARGs (Fig. 6B).

FIG 6.

(A) Antimicrobial resistance transfer efficiencies calculated by counting colonies selected on AMK and CAR plates. Bars, means ± SD (n = 3). A sample corresponding to the supernatant of a 1.65-ml bacterial culture of A. baumannii NU-60 and Δcapsid1, Δcapsid2, and Δcapsid1-Δcapsid2 mutants was used for each antimicrobial resistance transfer assay. Statistical significance was determined by Welch’s t test relative to results obtained for the NU-60 strain. **, P < 0.01; #, limit of detection (no colony growth on plates containing antimicrobials); N.S., not significant. (B) Trend of antimicrobial resistance transfer efficiencies under mitomycin C treatment. A sample corresponding to the supernatant of an 8.25-ml culture of A. baumannii NU-60 was used for each antimicrobial resistance transfer assay. Bars, means ± SD (n = 3). (C) Antimicrobial resistance transfer efficiencies when A. baumannii ATCC 17978 (wild) and ΔrecA, ΔrecA/pKT230, and ΔrecA/pKT230-recA mutants were used as the recipient cells. Bars, means ± SD (n = 3). A sample corresponding to the supernatant of a 5.5-ml bacterial culture of A. baumannii NU-60 was used for each antimicrobial resistance transfer assay. Statistical significance was determined by Welch’s t test relative to results obtained for the wild strain. **, P < 0.01; *, P < 0.05; #, limit of detection (no colony growth on plates containing antimicrobials); N.S., not significant.

To identify prophages hidden in the chromosomal DNA of the NU-60 strain and encoding capsid 1 and 2 proteins, we used the PHAge Search Tool Enhanced Release (PHASTER) server. Analyses revealed the presence of four intact prophages in the chromosome, with the capsid 1 and 2 genes associated with prophage ϕNU60-I (50.0 kb) and ϕNU60-II (52.6 kb), respectively (Fig. 1 and Fig. S7). The genes coding for two of the proteins (IDs 01 and 03) (Fig. 5D) identified by LC-MS/MS analysis were located outside the prophage regions predicted by the PHASTER server; the role of these proteins remains unclear in this study.

Generalized transduction mediates chromosomal ARG transfer.

Illumina short reads prepared from the samples related to phage particles (sample 1, fractions 3 to 10 collected after OptiPrep gradient ultracentrifugation with Triton X-100/DNase treatment [Fig. 4B]; sample 2, fraction a obtained after size exclusion chromatography with Triton X-100/DNase treatment [Fig. 5A]) were mapped to the genome of the NU-60 strain (Fig. 1). The output reads derived from DNA included in the phage particles covered 44% (sample 1) and 37% (sample 2) of the length of the A. baumannii NU-60 genome (4,043,233 bp). These results suggested that the phage particles were capable of carrying random pieces of the host A. baumannii genome, except phage DNA, for assembly.

The results of quantitative PCR (qPCR) analyses for estimating the copy numbers of the capsid genes and ARGs (armA and bla_TEM-1) in the phages indicated that most of the phage cargo was comprised of portions of the genome required for assembling particles, such as capsid proteins, rather than non-phage-related genes, including ARGs, since the copy number of capsid 1 genes was approximately 10³- to 10⁴-fold larger than that of armA and bla_TEM-1 (Fig. S8).

Finally, we evaluated whether recA, which generally mediates homologous recombination events in bacterial cells, contributes to the acquisition of ARG-containing fragments conveyed by phages in the recipient A. baumannii strain. Disruption of recA avoided the emergence of AMK- and CAR-resistant derivatives completely, while complementation of recA successfully recovered the ability to acquire ARGs (Fig. 6C). These results together indicated that recA plays a central role in acquiring ARGs, conveyed by phages, through homologous recombination in the recipient A. baumannii cells.

Collectively, our results demonstrated that ARG transfer occurred due to the generalized transduction mechanism triggered by prophages hidden in the A. baumannii genome. In the majority of the produced phages, DNAs for their assembly are enveloped, whereas such phages occasionally mispack host bacterium-derived genomic DNA, including ARGs. Here, the mispacked DNA, including the ARGs in the phages, were used to transfer AR properties to the AS A. baumannii strains, resulting in emergence of new AR organisms (Fig. 7).

FIG 7.

Summary model of ARG transfer between A. baumannii via generalized transduction.

DISCUSSION

Horizontal transfer of ARGs in A. baumannii is recognized as being primarily mediated by plasmids. However, plasmid transfer representatively mediated by conjugation presents several limitations, such as the requirement for direct cell-cell interaction, which might unexpectedly trigger cell-cell killing through type VI secretion systems (42, 43). Therefore, bacteria might benefit by transferring ARGs without direct cell-cell interaction in order to avoid disadvantages associated with such interaction. Here, we elucidated the mechanism mediating this DNA transfer, in the absence of direct cell-cell interaction, to establish the transfer of chromosomal ARGs between A. baumannii strains.

Consequently, a small portion of phage-related DNA in culture supernatant was employed as the cause of ARG transfer. Since the involvement of prophages in gene transfer is well known in the ordinal bacterial community as generalized transduction, it was expected to be involved in the spread of ARG across recent clinically relevant pathogens. However, the evidence explaining its contribution to the ARG transfer phenomenon in recent clinically relevant Gram-negative pathogens, including A. baumannii, has not yet been revealed, despite such pathogenic bacteria having rapidly acquired a variety of ARGs and generalized transduction being identified as the principal mechanism for ARG transfer in the clinically relevant Gram-positive pathogen Staphylococcus aureus (44).

Here, we revealed that specific prophages hidden in the chromosome of MDR A. baumannii could mediate the transfer of a variety of chromosomal ARGs. ARG transfer was found to be mediated by a generalized transduction mechanism in which shared DNA (including the ARGs) possibly was packed in phage particles comprised of specific capsid proteins, was discharged along with the phages following phage-related lysis, and got transferred to the recipient A. baumannii strains (Fig. 7). DNA fragments transferred to the recipient cells then may be integrated into the corresponding regions of the chromosome through RecA-dependent homologous recombination (Fig. 7). This type of ARG transfer, ultimately completed through homologous recombination, is in line with the finding that ARG transfer occurred with other A. baumannii cells but not with A. baylyi cells, despite the latter exhibiting an alternative yet highly competent nature.

The transferred genetic regions, including ARGs, were all consistently around 30 kb in size, despite the DNA fragments including diverse ARGs. Therefore, we expected that the phages would also contain DNA of approximately aligned sizes, which is in agreement with the characteristics of typical transducing phages enveloping regular-sized DNA (45).

Our findings are in agreement with the previous report by Krahn et al., which showed horizontal transfer of chromosomal ARGs in A. baumannii strain R2090 (14); it demonstrated the transfer of bla_NDM-1 along with its 66-kb surrounding region located on the chromosome between A. baumannii strains and further suggested that prophage genomes located on the chromosome contributed to the transfer. The ϕR2090-I prophage region was similar to that of the ϕNU60-I prophage in the present study; proteins encoded by 25 of the 58 predicted genes of ϕNU60-I were identified in the ϕR2090-I prophage with at least 90% amino acid identities. Although the contribution of ϕR2090-I prophage to ARG transfer was not completely revealed, it was assumed to be involved in the bla_NDM-1 transfer. These findings suggested that prophages indeed play a crucial role in transferring chromosomal ARGs, along with long surrounding DNA regions, in A. baumannii.

Phages carrying ARGs have been identified in diverse environments, such as in urban wastewater, environmental water, and human fecal samples (46–49), and horizontal gene transfer of ARGs through phage transduction is regarded as a cause for the emergence and spread of AR bacteria in the environment. Our results showed that DNA packed in phages tolerated nuclease and detergent treatments and that this property might enable ARG retention for an extended period in various environments, thereby augmenting opportunities for their acceptance by other bacteria. Considering the advantage offered by the protection of ARGs in phage particles, phages could be expected to play a central role in spreading ARGs in the environment without direct cell-to-cell interaction. Therefore, ARG-carrying phage particles, produced by MDR bacteria such as A. baumannii in clinical settings, might represent a principal vehicle for ARG dissemination across the microbial environment. Our in vitro model for phage-mediated ARG transfer supports the concept of ubiquitous ARG circulation in the environment.

We used one representative A. baumannii clinical isolate to evaluate the phage-mediated intercellular transfer of chromosomal ARGs. Although the presence of an identical ϕNU60-I prophage (50 kb), carrying the capsid 1 gene found in the NU-60 genome, has not been admitted in any A. baumannii genome deposited in the database, the core capsid 1 protein gene involved in ARG transfer has been widely presented in A. baumannii strains isolated from geographically different areas, such as Japan (50), South Korea (51), China (52), Malaysia (53), and the United Kingdom (54). A similar phenomenon concerning the transfer of ARGs, using pelleted particles obtained through ultracentrifugation of culture supernatants, was identified when we used other A. baumannii clinical isolates (data not shown). However, a definitive correlation between the presence of capsid protein genes identified in this study and the phenomenon of ARG transfer was not observed. In certain strains harboring the capsid protein genes ARG transfer has not been demonstrated, whereas in other strains lacking the capsid protein genes ARG transfer did occur. Therefore, ARG transfer might be mediated by a variety of prophages hidden in the chromosome of a host A. baumannii strain. The extent to which these mechanisms are applicable for eDNA transfer might differ in each A. baumannii strain, potentially resulting in diverse eDNA transfer abilities. Future studies for investigating the relationship between capsid protein genes in prophages and ARG transfer phenomena, using other A. baumannii isolates, will address this question.

In conclusion, we described the efficacious phage-mediated horizontal transfer of chromosomal ARGs across A. baumannii strains. A. baumannii can release various forms of eDNA, some of which are readily degraded by DNase while others are protected from DNase digestion. In this study, we observed that DNA protected from DNase digestion by phages was transferred across A. baumannii strains (Fig. 7). Additionally, horizontal gene transfer is a major driving factor in the spread of ARGs among pathogenic bacteria, and our findings suggested that phages contribute to this phenomenon, likely in environments where direct cell-to-cell interaction occurs infrequently. Our study provides novel insights into the horizontal transfer of recent ARGs in the human opportunistic pathogen A. baumannii, with this finding offering a better understanding of A. baumannii evolution in terms of acquisition of MDR.

MATERIALS AND METHODS

Bacterial strains.

A. baumannii and A. baylyi strains used in this study were routinely cultured in Luria-Bertani (LB) broth, supplemented with appropriate antimicrobials or left unsupplemented, at 37°C.

NGS analyses.

DNA used for next-generation sequencing (NGS) analyses was extracted using a QIAamp DNA minikit (Qiagen, Hilden, Germany). To determine the complete genome sequence of A. baumannii strain NU-60, DNA was subjected to a PacBio RS2 sequencer with P6C4 chemistry (PacBio, Menlo Park, CA, USA). De novo assembly of the obtained reads (average length, 11 kb) was performed using Canu (55). Error correction of the assembled circular sequence was performed using Pilon (56) with Illumina short reads, which were obtained with a Nextera XT DNA library kit (Illumina, San Diego, CA, USA), MiSeq reagent kit, version 3 (600 cycles; Illumina), and a MiSeq platform (Illumina). Annotation was performed using the DFAST server (57).

Whole-genome sequencing analyses of A. baumannii ATCC 17978ΩarmA, Ωbla_TEM-1, Ωtet(B), and ΩgyrA-81L derivatives were performed using Illumina short reads and an assembler of the A5-miseq pipeline (58). Gap closing was performed with classical PCR and Sanger sequencing analyses. Illumina short reads were also mapped to the A. baumannii NU-60 genome sequence using CLC Genomic Workbench (Qiagen) with default parameters. The ResFinder (59) and PHASTER (60) servers were used to detect ARGs and prophages in the genome of A. baumannii. A circular genomic map of the A. baumannii NU-60 strain was rendered using BLAST Ring Image Generator (BRIG) (61).

PCR detection of ARGs.

Conventional PCR detection of armA, bla_TEM-1, and tet(B) in A. baumannii strains was performed using the primers listed in Table S2 in the supplemental material. The amplified products of armA, bla_TEM-1, and tet(B) were 315 bp, 824 bp, and 264 bp, respectively.

Introduction of ARGs into A. baumannii.

The A. baumannii strains used for ARG transfer were precultured in LB broth containing 25 μg/ml AMK, and aliquots of the cultures were inoculated into plain LB broth, followed by incubation at 37°C with shaking (120 rpm). Mitomycin C was added to a final concentration of 1, 10, 100, and 1,000 ng/ml, when necessary. The bacterial cells were removed by centrifugation at 6,000 × g for 15 min, and the supernatant was passed through a 0.22-μm polyvinylidene difluoride (PVDF) filter (Sartorius, Göttingen, Germany). The filtrate was ultracentrifuged at 150,000 × g for 90 min at 10°C in a Type 50.2 Ti rotor (Beckman Coulter, Brea, CA, USA) to pellet the particles in the supernatant. The pelleted particles were resuspended in 20 mM HEPES-NaOH buffer (pH 7.5) and passed through a 0.22-μm PVDF filter. An aliquot of each sample was spread on LB-agar plates to check for donor cell (A. baumannii NU-60 strain) contamination in the samples; it was confirmed that there was no contamination in any prepared sample.

A. baumannii strain ATCC 17978 was cultured in LB broth at 37°C with shaking until the optical density at 600 nm reached 0.5, after which 1 ml of the bacterial culture was pelleted by centrifugation, and cell pellets were resuspended in 50 μl of phosphate-buffered saline and spotted on LB-agar plates. After air drying the surface of the spotted bacterial suspensions, the samples collected by ultracentrifugation of culture supernatants were overlaid on the dried bacterial spots, and the plates were incubated at 30°C for 24 h. The bacterial cells that grew on the LB-agar plates were scraped into 1 ml of LB broth and suspended with vigorous shaking. Finally, 100 μl of the bacterial suspension was spread on LB-agar plates supplemented with AMK (25 μg/ml), TET (50 μg/ml), CAR (400 μg/ml), or NA (75 μg/ml), and the AR derivatives assumed to have acquired antimicrobial resistance were selected.

Antimicrobial susceptibility testing.

Antimicrobial susceptibility was tested according to the method provided by Clinical and Laboratory Standards Institute (62). Bacterial suspensions were adjusted to 10⁸ CFU/ml in Mueller-Hinton (MH) broth (Becton, Dickinson, Franklin Lakes, NJ, USA) and spread on MH-agar plates using a sterile cotton swab, followed by placement of an Etest strip (bioMérieux, Marcy-l'Étoile, France) on the plates and incubation at 35°C for 18 h.

DNA visualization and quantification.

DNA was visualized using agarose gel electrophoresis and ethidium bromide staining and quantified using a Qubit dsDNA HS assay kit and a Qubit 4 fluorometer (Thermo Fisher Scientific, Waltham, MA, USA).

TEM analysis.

Samples were pipetted onto Parafilm, and TEM grids were layered on the samples. After incubation for 3 min at room temperature, the grids were layered on drops of uranyl acetate solution. The samples were viewed using a JEM-1400EX transmission electron microscope (JEOL, Tokyo, Japan).

Purification of DNA involved in ARG transfer.

The precipitated particles collected after ultracentrifugation of bacterial culture supernatants were suspended in 20 mM HEPES buffer (pH 7.5) containing 45% (vol/vol) OptiPrep (Axis-Shields, Dundee, Scotland). The samples were placed at the bottom of ultracentrifugation tubes, layered with 35%, 30%, 25%, 20%, 15%, and 10% OptiPrep solution separately, and ultracentrifuged at 100,000 × g for 16 h at 10°C in an SW41 Ti rotor (Beckman Coulter). An aliquot was collected from the top of the supernatant and used for the ARG transfer experiment. The separated samples were treated with DNase I (TaKaRa, Shiga, Japan) and/or Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA) as required.

Samples obtained after OptiPrep gradient ultracentrifugation were buffer exchanged into 20 mM HEPES-NaOH (pH 7.5) containing 100 mM NaCl using Amicon Ultra-15 filter units (Millipore, Billerica, MA, USA). The samples were loaded onto a Superdex 200 Increase 10/300 GL column (GE Healthcare, Little Chalfont, UK) and eluted with 20 mM HEPES-NaOH (pH 7.5) containing 100 mM NaCl.

SDS-PAGE and Western blotting.

Samples were separated using SDS-PAGE and stained using Coomassie brilliant blue (Nacalai Tesque, Kyoto, Japan) or a Sliver Stain MS kit (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan). The separated proteins were transferred to nitrocellulose membranes, and the protein band corresponding to OmpA was detected using rabbit anti-OmpA primary antiserum (1:1,000; Eurofins Genomics, Tokyo, Japan), a horseradish peroxidase-conjugated secondary antibody (goat anti-rabbit IgG; 1:1,000; Medical & Biological Laboratories, Nagoya, Japan), and ImmunoStar Zeta (FUJIFILM Wako Pure Chemical Corporation). Protein bands were imaged using an Amersham Imager 600 system (GE Healthcare).

Identification of phage-associated proteins.

Silver-stained protein bands on SDS-PAGE gels were excised, trypsin digested, and subjected to nano-LC-MS/MS and Mascot search analyses by Japan Bio Services Co., Ltd. (Saitama, Japan).

Deletion and replacement of genes in A. baumannii.

Capsid genes were deleted in A. baumannii strain NU-60 according to a previous method (63). Briefly, Escherichia coli S17-1 λpir, carrying pMo130-TelR (sacB⁺, xylE⁺; Addgene, Watertown, MA, USA) embedded with an appropriate DNA fragment, was obtained by performing PCR with the primers listed in Table S2, after which the strain was conjugated with A. baumannii NU-60 on membrane filters placed on LB-agar plates and incubated for 24 h at 30°C. The membrane filter was washed with LB broth, and aliquots were placed on LB-agar plates containing chloramphenicol (30 μg/ml) and tellurite (30 μg/ml) for selecting A. baumannii conjugants featuring the first crossover replacement. Insertion of fragments derived from pMo130-TelR (sacB⁺, xylE⁺) was confirmed by spraying 0.45 M pyrocatechol as an indicator (pyrocatechol is converted to yellow-colored 2-hydroxymuconic semialdehyde by XylE). The conjugants featuring the first crossover replacement were passaged daily with LB broth containing 10% sucrose in order to select bacterial cells featuring a second crossover, which then resulted in the deletion of the targeted gene. Bacterial cells subsequently were grown on agar plates containing 10% sucrose, and the colonies that appeared were sprayed with 0.45 M pyrocatechol solution to confirm the loss of vector-derived fragments. The expected deletion of target genes was confirmed by performing PCR and sequencing analyses.

The A. baumannii ATCC 17978 strain, with recA gene replacement (A. baumannii ATCC 17978ΔrecA strain), was constructed according to the previous method (64). The upstream and downstream regions of recA were separately amplified using the primers shown in Table S2. The tet(B) gene of A. baumannii strain NU-60 was also amplified. Overlap PCR was performed with the above-described three PCR products to obtain tet(B) gene cassettes flanked by recA upper and lower genetic regions. The constructed tet(B) gene cassettes were electroporated into A. baumannii ATCC 17978. The transformants were selected on LB agar plates containing tetracycline. The designed gene replacement was confirmed by performing PCR and sequencing analyses. Complementation experiments were performed by introducing pKT230 plasmids carrying recA into the A. baumannii ATCC 17978ΔrecA strain.

qPCR.

qPCR analyses to estimate the copy numbers of armA, bla_TEM-1, and capsid 1 and 2 genes in purified DNA samples were performed using TB green premix EX Taq (TaKaRa) and a StepOnePlus system (Thermo Fisher Scientific). The primers used for qPCR are listed in Table S2; 0.4 ng of DNA was added to each reaction mixture, and qPCR was performed in a final volume of 20 μl.

Data availability.

The complete genome sequence of A. baumannii strain NU-60 was deposited in GenBank under accession no. AP019685.