MITEAba12, a Novel Mobile Miniature Inverted-Repeat Transposable Element Identified in Acinetobacter baumannii ATCC 17978 and Its Prevalence across the Moraxellaceae Family

One of the most important weapons in the armory of Acinetobacter is its impressive genetic plasticity, facilitating rapid genetic mutations and rearrangements as well as integration of foreign determinants carried by mobile genetic elements. Of these, IS are considered one of the key forces shaping bacterial genomes and ultimately evolution. We report the identification of a novel nonautonomous IS-derived element present in multiple bacterial species from the Moraxellaceae family and its recent translocation into the hns locus in the A. baumannii ATCC 17978 genome. The latter finding adds new knowledge to only a limited number of documented examples of MITEs in the literature and underscores the plastic nature of the hns locus in A. baumannii. MITEAba12, and its predicted parent(s), may be a source of substantial adaptive evolution within environmental and clinically relevant bacterial pathogens and, thus, have broad implications for niche-specific adaptation.

A cinetobacter baumannii has been classed as one of the most predominant pathogens responsible for multidrug-resistant (MDR) nosocomial infections worldwide (1). Aside from its notorious MDR phenotype, A. baumannii also displays a remarkable capacity to persist on a variety of inanimate surfaces for extended periods, providing a reservoir for infection and facilitating transmission throughout clinical settings (2,3). Significant work has been undertaken to identify and track the arsenal of genes that contribute to the impressive persistence and resistance strategies available to A. baumannii (4)(5)(6)(7)(8). This has identified a highly dynamic and plastic genome, dominated by numerous integration events as well as alterations in expression of intrinsic genes modulated through mutations and deletion and/or insertion of mobile genetic elements (MGEs) (9)(10)(11). MGEs are present in nearly all prokaryote genomes and constitute the "mobilome," a term which has gained significant traction in recent years, driven by the increase in infections caused by MDR isolates. The mobilome itself is comprised of a number of genetic entities, including plasmids, bacteriophages, gene cassettes in integrons, and transposable elements, all capable of capturing and disseminating genetic material across bacterial genomes via horizontal gene transfer (HGT) (12).
Of the above-mentioned entities, transposable elements are seen as a major contributor to niche-specific adaptive evolution. They are capable of moving from one position to another within a given genome and are often associated with the dissemination of antimicrobial resistance determinants (13)(14)(15). One of the simplest autonomous types of mobile elements is the insertion sequence (IS), consisting of a transposase gene(s) that is typically bordered by terminal inverted repeats (TIRs), designated left (IRL) and right (IRR) relative to the direction of the transposase gene. The TIRs contain multiple domains required for transposase binding, donor DNA cleavage, and strand transfer, supporting the integration of the elements into host DNA via replicative or nonreplicative mechanisms (16). As a consequence of insertion, short direct repeat sequences of the target DNA are often generated (target site duplications, or TSDs), which differ in length and degree of sequence specificity depending on the IS element being translocated (17). Movement of an IS to a new location within a genome offers a variety of possible integration sites. Although some IS display clear trends/preferences in target sites, the large majority of IS demonstrate low target specificity (18).
Small mobile elements can be further delineated based on their movement autonomy. A limited range of nonautonomous elements exist in bacteria, such as repetitive extragenic palindromic sequences, Tn3-derived inverted-repeat miniature elements (TIMEs), and miniature inverted-repeat transposable elements (MITEs) (19)(20)(21). Like eukaryotic MITEs (22), bacterial MITEs are small (ϳ50 to 600 bp) AT-rich sequences that have lost their cognate transposase gene and, thus, contain noncoding DNA that in most, but not all, cases are flanked by TIRs (23). Based on their origins, MITEs can be categorized as type I or type II and are generated by internal deletion of parent transposable elements or by random convergence of TIR sequences, respectively (24). Movement of these elements is thought to be mediated by transposases of a coresident parental element acting in trans. The site of integration and length of the TSDs of MITEs are generally identical or highly similar to that of the coresident IS parent (23). Since their identification in bacteria (25), a number of these elements have been documented from a diverse range of species, where many have significantly influenced the evolutionary tempo of their host genomes (20). These elements are often overlooked due to the absence of a recognizable coding sequence (CDS) and their tendency to reside in intergenic regions. Thus, they represent a largely unexplored field in microbial genomics.
Through characterization of a subset of morphologically distinct colonies isolated during desiccation stress analyses, we identified a novel MITE that transposed to a new location within the A. baumannii ATCC 17978 genome. Due to shared similarities in TIRs and TSD sequence length, the 113-bp sequence is predicted to have proliferated through the activity of the transposase encoded by resident ISAba12 elements present in ATCC 17978 and, thus, was named MITE Aba12 . The prevalence of this novel, nonautonomous MGE across all publicly available sequenced bacterial genomes was exam-amplifications across the hns loci of the hypermotile strains identified that all amplicons were larger than the WT control ( Fig. 2A). DNA sequencing of these products revealed insertion of ISAba12 in three cases, originating from each of the three different background strains, which were located in two previously identified integration sites (29) (Fig. 2B). In the remaining four strains, all based on the ΔygiW background, a shorter insertion in hns was detected, and sequencing of one example revealed a 113-bp element integrated into a novel site (Fig. 2B). To determine if the integrated element was stably inserted in hns of the ΔygiW strain, PCR screening after five consecutive passages in liquid culture from five biological replicates was undertaken. All samples maintained the element within hns (data not shown). To examine whether isolates with a disrupted hns, irrespective of the site/type of integration, still produced the distinctive hypermotile phenotype, their motility phenotypes were assessed and found to be comparable to that seen for the previously identified hns mutant derivative (27,30) (see Fig. S1 in the supplemental material). Complementation with a WT copy of hns (ACX60_16755) carried on the pWH1266 shuttle vector (27) restored all isolates to their parental nonmotile phenotype (Fig. S1).
Identification and characterization of a novel active MITE in A. baumannii ATCC 17978. To characterize this novel 113-bp element found in the A. baumannii ΔygiW strain, its DNA sequence and that of its insertion site in hns were analyzed. This revealed that the 113-bp element carried 16-bp imperfect TIR sequences (different in 1 nucle-otide) and an 81-bp core region and generated 9-bp TSDs on insertion into hns (Fig. 2). The element is AT rich (78%) and does not contain any known CDS (31). Taken together, these traits strongly suggested that this element is a MITE (23).
To identify the abundance of the MITE within the A. baumannii ATCC 17978 genome, the 113-bp MITE sequence in hns from the ΔygiW background was used as a query for BLASTN searches. Only one copy at the 3= end of the ACX60_04650 locus, encoding a hypothetical protein harboring a partial KAP-family NTPase motif (32), was identified, fusing this gene with 31 bases from the MITE (Fig. 2C) and generating a premature stop codon. Comparative analyses with A. baumannii D36 revealed that the protein was 398 amino acids shorter and is therefore likely to be nonfunctional (data not shown). Genes coding for KAP NTPases are known to be frequently disrupted, leading to pseudogene formation (33). PCR with primers specific for the ACX60_04650 location (see Table 4) identified that the MITE was maintained in this position in the ΔygiW Δhns::MITE strain. Consequently, there are two MITE copies in the ΔygiW background, one at the ACX60_04650 locus and an additional copy located in the hns gene, inferring duplication of the novel element (data not shown).
ISAba12 is the proposed autonomous parent of the novel MITE in A. baumannii ATCC 17978. To identify the potential parent element that may have aided in translocation of the MITE, IS present in the ATCC 17978 chromosome and pAB3 plasmid (GenBank accession numbers CP012004.1 and CP012005.1, respectively) were first identified from results generated by ISseeker (11). Subsequent manual examination of the length and sequence of their TIRs and TSDs revealed that ISAba12 provided the best match to those of the MITE. ISAba12 harbors a single open reading frame coding for a transposase, with a characteristic DDE catalytic motif, between its 16-bp TIRs and generates 9-bp TSDs upon insertion (28). Thus, the novel MITE was most likely translocated into hns by a coresident ISAba12 transposase present in ATCC 17978 and will be referred to as MITE Aba12 .
As MITE Aba12 does not contain a transposase gene, it is not possible to define IRL and IRR sequences relative to this gene. Two of the three copies of ISAba12 in ATCC 17978 (at loci ACX60_04795 and ACX60_18935) have identical TIRs that perfectly match one TIR of MITE Aba12 . However, the nonidentical TIRs of the third copy of ISAba12 (ACX60_12380) each perfectly match one TIR of MITE Aba12 , allowing IRL and IRR of MITE Aba12 to be designated relative to this IS. MITE Aba12 is present in a diverse range of species from the Moraxellaceae family. To identify whether MITE Aba12 is widespread or restricted to A. baumannii ATCC 17978, the sequence found within hns from the ΔygiW Δhns::MITE Aba12 strain was used as a query to search bacterial genomes present in publicly available databases (10 July 2018). Orthologs of MITE Aba12 were identified in both chromosomes and plasmids, with an additional 30 strains from the Acinetobacter genus and one from Moraxella osloensis harboring the element at various frequencies (Table 1). MITE Aba12 was found within a range of environmental Acinetobacter species, with the greatest number of copies identified (n ϭ 22) in A. baumannii DS002, isolated from soil in Anantapur, India, in 2005. A number of Acinetobacter strains isolated from patients and hospital sewage in multiple countries also carried copies of MITE Aba12 , inferring its presence and dissemination into clinically relevant isolates worldwide. Acinetobacter sp. strains ACNIH2, SWBY1, and WCHA45 and A. johnsonnii XBB1 possessed MITE Aba12 on the chromosome as well as in plasmids (Table 1). An additional number of plasmid sequences carrying MITE Aba12 were also identified ( Table 1), but their corresponding chromosome sequences are not available. Copies of MITE Aba12 identified in the A. baumannii PR07 genome (GenBank accession number CP012035.1) were not included in further analyses as the genome contained strings of undetermined bases and was not of a high enough quality.
Using ISseeker (11), it was found that approximately 18.5% of the 1,035 A. baumannii genomes examined harbored at least one copy of ISAba12, with an average of 5.6 copies per genome (data not shown). Using the ISfinder tool (28), four relatives of ISAba12 were identified: ISAba13, ISAlw1, ISAha1, and ISAha2 (Table 2). These elements are present at various frequencies in Acinetobacter genomes, and the transposases encoded within the elements share between 92 and 94% amino acid identity with the transposase in ISAba12. Importantly, they have the same perfect 16-bp TIR sequence as the majority of ISAba12 elements ( Table 2). This led us to investigate whether other characterized IS contain TIR sequences similar to those in MITE Aba12 and thereby could translocate the nonautonomous element. An additional nine IS elements were found to have TIR sequences similar or identical to those of MITE Aba12 (Table 2). These IS elements are of a length similar to that of ISAba12, ranging from 1,023 to 1,052 bp, with the majority also generating 9-bp TSDs ( Table 2). Comparisons of MITE Aba12 against the sequences of each IS listed in Table 2 revealed nucleotide identity was confined to only the TIR sequences; thus, the origins of MITE Aba12 from an IS could not be readily deduced (data not shown).
MITE Aba12 is a highly conserved mobile element with potential to affect expression of neighboring host genes. To examine sequence identity across all the identified MITE Aba12 copies, a multiple-sequence alignment using Clustal Omega (34) was performed. From the analysis of 90 MITE Aba12 copies it was found that 10% of  (Fig. 3). Thus, a total of 34 different MITE Aba12 sequences were identified, leading to the assignment of 10 subgroups. MITE Aba12 sequence arrangements that harbored two copies or more were The transposase of ISAba5 is thought to be inactive (28).

FIG 3
Nucleotide alignment of all MITE Aba12 elements identified in this study. The nucleotide sequence above the alignment (black box) denotes the consensus sequence, MITE Aba12 (c), derived using WebLogo software (35). MITE Aba12 sequences with nucleotide variations are displayed. Subgroup representatives are numbered and in boldface type with numbers in parentheses indicating the total number of MITE Aba12 copies with that sequence. A, T, G, and C nucleotides are denoted in blue, yellow, purple, and green boxes, respectively. Black lines and asterisks represent the terminal inverted repeats (IRL and IRR) and conserved bases, respectively. See Table S1 for a full list of MITE Aba12 elements included in each subgroup and Table 1 for strain accession numbers.
segregated into subgroups that were ordered from 1 to 10 based on the most to least abundant ( Fig. 3 and Table S1). Significant variation existed across the central region of the element, as only 10 bases were conserved across all 90 identified copies ( Fig. 3) (it is possible that some nucleotide differences identified across MITE Aba12 copies can be attributed to sequencing errors). The MITE Aba12 TIR sequences were the most conserved, as eight and nine of the 16-bp IRL and IRR sequences, respectively, were identical across all MITE Aba12 copies analyzed (Fig. 3). Overall, no preference in the orientation of MITE Aba12 in the genomes could be identified (data not shown). MITEs generally insert into AT-rich regions (23). Of the 90 identified MITE Aba12 copies, 36 from 13 different genomes had 9-bp TSDs. From the 22 MITE Aba12 copies in A. baumannii DS002, TSDs could be identified for 17, which could infer a burst of recent activity. Interestingly, all MITE Aba12 elements from subgroup 4 were found in the same site and flanked by identical TSDs of 8 rather than 9 bp (TTTTTGTT). These elements were on large plasmids (ϳ62 to 112.5 kb), and BLASTN analyses using the MITE Aba12 element together with ϳ700 bp left and right of the sequence from pNDM-32 of A. baumannii CHI-32 identified they were all located in an identical position, sharing 100% identity across this ϳ1.5-kb region (data not shown). Overall, MITE Aba12 appeared to favor insertion into sequences with an AT ratio of Ն55.5%, as demonstrated by the skewed distribution of the columns to the right (Fig. 4A), although no identifiable trends in nucleotide sequence arrangements could be identified (Fig. 4B).
To assess how MITE Aba12 could influence host gene expression, a consensus sequence, MITE Aba12 (c), was generated using WebLogo (35) from all 113-bp MITE Aba12 elements (n ϭ 81) (Fig. 3). At least two stop codons in all six reading frames can be identified after translation of the MITE Aba12 (c) DNA sequence. Start codons followed by three, seven, or eight amino acids at the terminal ends of the element were identified in three of the reading frames (data not shown). Depending on the integration site in a given genome, these characteristics could allow for fusion with neighboring CDSs. Mfold (36) predicted weak secondary structures with ΔG of Ϫ23.99 or Ϫ26.94 kcal/mol in the two orientations of MITE Aba12 (c) (IRL to IRR or IRR to IRL, respectively) (data not shown). Using the ARNold tool (37), no predicted Rho-independent transcriptional terminators were identified. The Softberry program BPROM predicted two outwardfacing promoter sequences based on the Ϫ35 and Ϫ10 Escherichia coli 70 promoter consensus sequences (Fig. S2). These sequences were also compared with the strong outward-facing promoter found in ISAba1 coupled with flanking sequence associated with overexpression of the bla ampC gene in A. baumannii CLA-1 (38) (Fig. S2B). To verify whether the two putative outward-facing promoters identified within MITE Aba12 (c) could drive the production of mRNA transcripts, three previously published A. baumannii ATCC 17978-derived RNA-sequencing transcriptomes (30,39) were aligned to the reference ATCC 17978 genome (GenBank accession number CP012004.1) using the Integrative Genomics Viewer program (40). Transcripts originating within the MITE Aba12 sequence that could be attributed to the P out IRR putative promoter were identified across all three transcriptomes. However, transcripts reading out from P out IRL were limited (data not shown). Thus, it appears in ATCC 17978 that the P out IRR putative promoter within MITE Aba12 (c) has the potential to influence host gene transcription.
The fusion of small mobile elements with neighboring genes can affect gene function and in some cases lead to improved host fitness or formation of new proteins (41,42). The exhaustive analysis conducted on MITE Aba12 in publicly available GenBank entries revealed that some insertions of MITE Aba12 interrupted host genes, and in some cases the encoded protein could be fused with up to 19 amino acids encoded by MITE Aba12 sequences (data not shown). MITE Aba12 elements located in pAbPK1a from A. baumannii AbPK1 and in the chromosomes of A. baumannii B8300 and Acinetobacter sp. strain ACNIH1_#2 could create fusions to the 5= end of adjacent genes. Each had incorporated nucleotides reading outwards from the TIR of MITE Aba12 to generate the first four amino acids (MQQS) of the neighboring CDS. These particular arrangements also placed the host gene in proximal distance to the P out IRR promoter sequences, and given its activity in ATCC 17978, the element could also influence the expression of fused genes.
MITE Aba12 in M. osloensis CCUG 350 is located within a novel composite transposon. As previously stated, M. osloensis CCUG 350 carries one copy of MITE Aba12 (Table 1). It lies within an 8.5-kb region absent from five closely related M. osloensis genome sequences (Fig. 5A shows the sequence alignment). IS were found at the terminal ends of the novel insert and shared highest identity with the IS1 family member ISAba3 (81% identity; E value, 5eϪ55) (28). Both terminal IS carried 24-bp TIR sequences (5=-GGTGGTGTTTCAAAAAGTATGCTG-3=), and TSDs of 8 bp were identified at each end of the 8.5-kb insert (Fig. 5B). These features make this sequence synonymous with a composite transposon (17) now named Tn6645. In M. osloensis CCUG 350, the MITE Aba12 element was located between the ISAba3 element and a gene of unknown function containing a DUF 2789 motif (E value, 4.2eϪ27) (32). Additionally, an ISAba11like element, an alkylsulfatase gene, a TetR-family transcriptional regulator gene, and a partial gene encoding a major facilitator superfamily transporter were identified within the composite transposon (Fig. 5). The insertion of ISAba3 truncated the 3= end of the transporter gene by 540 bp and therefore is likely nonfunctional (a pseudogene).
BLASTN searches were used to search for Tn6645 in other bacterial genomes, but no additional full-length copies were identified (data not shown). However, approximately 4.3 kb of the 8.5-kb sequence aligned (96% identity) to a region in the chromosome of Acinetobacter guillouiae NBRC 110550 (43) (Fig. 5B). This region harbored the alkylsulfatase, TetR-family regulator, and the truncated transporter genes and may represent a source for this portion of the Tn6645 cargo.

DISCUSSION
Since their identification in bacteria 30 years ago, MITEs have been reported in a multitude of species, displaying significant diversity in their nucleotide sequence and functional properties (44). In this study, a novel MITE was identified in environmental and clinical isolates of Acinetobacter species, including A. baumannii, one of the leading bacterial organisms threatening human health (2). This novel element lacked any CDS that could produce a functional transposase, inferring that like other MITEs, MITE Aba12 is activated in trans. Given the high similarity between the TIR sequences of MITE Aba12 and those of ISAba12 (Fig. 2D), we propose the transposase from ISAba12 elements were responsible for MITE Aba12 mobilization in the A. baumannii ΔygiW strain. Whether ISAba12, or another IS with a TIR similar to that of MITE Aba12 (Table 2), can mobilize MITE Aba12 will need to be experimentally examined.
With the addition of MITE Aba12 , the list of nonautonomous elements reported in Acinetobacter grows to three. Like most prokaryotic MITEs, the two previously characterized MITEs from Acinetobacter isolates are flanked by TIRs and generate TSDs upon insertion (45,46). Compared to MITE Aba12 , both elements have been identified only on plasmid sequences and are approximately four times larger in size (439 and 502 bp, respectively) (45,46). Identical copies of the MITE originally identified in Acinetobacter sp. strain NFM2 flank class 1 integrons carrying different resistance determinants in a number of Acinetobacter strains, forming a structure comparable to that of a composite transposon (45,(47)(48)(49)(50)(51). MITE-297 is found on the large conjugative plasmid pA297-3 present in the A. baumannii global clone 1 reference strain A297 (RUH875) (46). In pA297-3, two copies of MITE-297 flank a 76-kb region carrying numerous IS and a mer module which confers resistance to mercury (46). Interestingly, within pA297-3, MITE Aba12 is also present in the intergenic region between the merD and 5hydroxyisourate hydrolase precursor genes (data not shown). MITE Aba12 is also found in an identical position in the ϳ208-kb pD46-4 plasmid from A. baumannii D46 (52) and a 141-kb plasmid from Acinetobacter sp. strain DUT-2. Given the position of the element within these plasmids, we suggest that MITE Aba12 has travelled with this mer operon, which may partly explain its distribution throughout these bacterial genomes. Our analyses also identified a copy of MITE Aba12 flanked by two IS on the large nonconjugative plasmid pALWED2.1 from the Acinetobacter lwoffii strain ED45-23, isolated from uncontaminated Russian permafrost sediments dated to be 20,000 to 40,000 years old (53). To our knowledge, this is the most primitive strain that has been sequenced and shown to carry a copy of MITE Aba12 . Interestingly, heavy-metal resistance operons identified on pALWED2.1 share identity with sequences from two additional Acinetobacter strains that also carry copies of MITE Aba12 (53). Our data, which provide another example of MITE Aba12 hitchhiking alongside resistance genes, supports the idea that HGT has played an important role in the evolution of heavy-metal resistance to confer a selective advantage to the organism (53).
Bacterial MITEs can possess various motifs that affect their own regulation and/or modulate expression of other genes within the residing genome (54)(55)(56). Using the MITE Aba12 (c) consensus sequence identified as part of this study, putative outwardfacing E. coli 70 promoters could be identified in both orientations (see Fig. S2 in the supplemental material). ISAba1 is present in high copy numbers across a number of A. baumannii genomes and has been shown to have a significant impact on host gene expression and genome architecture (11). Additionally, ISAba1 is frequently implicated in increased antibiotic resistance, achieved by insertion upstream of resistance genes, namely, those encoding cephalosporinases or carbapenamases (38,57,58). Despite the putative promoter sequences not being maintained across all MITE Aba12 elements, the two subgroups which exhibited the greatest number of conserved arrangements (subgroups 1 and 2, with 27 and 17 elements, respectively) have promoter sequences that exactly match the MITE Aba12 (c) consensus (Fig. 3). Elements within these subgroups were derived from a variety of species from the Moraxellaceae family and isolated from geographically distinct locations ( Table 1), suggesting that a selective pressure to maintain these nucleotides exists.
Analysis of the 90 MITE Aba12 copies revealed that sequence conservation was mainly confined to their TIRs, and they only deviated in length from the MITE Aba12 (c) consensus by a maximum of two nucleotides (Fig. 3). This is in contrast to significant size differences seen between variants of other types of bacterial MITEs (55,59,60). However, the lack of significant divergence seen within MITE Aba12 copies indicates that the element was generated from a single event and dispersed through bacterial genomes via HGT.
Shared identity of ISAba12 and the additional 13 IS harboring TIRs similar to those of MITE Aba12 was restricted to the TIR sequences (Table 2). Nevertheless, this finding significantly broadens the range of potential parental IS that could be capable of translocating MITE Aba12 . However, further experimental evidence is required to confirm whether these IS can translocate MITE Aba12 .
The observation of a MITE translocation within a prokaryote genome in real time is generally considered to be a rare event, as only a few examples have been documented in the literature (44). Remarkably, four separate instances where MITE Aba12 underwent translocation into hns, all of which were observed within the A. baumannii ΔygiW ATCC 17978 background, were identified. YgiW is known as a stress-induced protein in many Gram-negative bacterial species (61)(62)(63)(64). For instance, in Salmonella enterica serovar Typhimurium, YgiW (renamed VisP, for virulence-induced stress protein) was shown to be critical in stress resistance in vitro and in virulence (64). Similar to that of S. enterica and other bacteria, the ygiW homolog found in A. baumannii also contains the characteristic bacterial oligonucleotide/oligosaccharide-binding fold domain (DUF388) (65) and is located immediately upstream of the qseBC TCS genes (data not shown). As transposition of IS is strongly controlled, most likely to reduce potential deleterious effects within the cell, we question whether the deletion of a protein involved in the stress response influenced the transposition and/or properties that regulate expression and subsequent movement of ISAba12/MITE Aba12 elements within the ATCC 17978 genome. Furthermore, as isolates displaying hypermotility were only identified once during desiccation analyses, we speculate that these events represent a transposition burst (66). This new phenomenon offers a substitute for the selfish DNA hypothesis, where these intermittent bursts of IS transposition can increase copy numbers and therefore assist in their maintenance within bacterial genomes.
H-NS is defined as a DNA architectural protein known to play multiple fundamental roles across a number of Gram-negative pathogens, including regulation of AT-rich horizontally acquired genes, many of which are involved in multiple stress responses (67,68). Two distinct locations for ISAba12 insertions in the hns locus were previously identified in A. baumannii (27,29). These were also the target sites for the ISAba12 insertions in this study, inferring these sequences are favored integration hotspots. MITE Aba12 inserted into a novel location within hns, 151 bp from the start codon and upstream of the characterized DNA-binding domain (27). Two additional examples of IS-mediated disruption of hns in A. baumannii have been recently identified (9,69). ISAba125 was shown to be responsible in both studies, integrating into the intergenic region downstream of hns (ACICU_00289). In one case, the last two amino acids of H-NS are altered and the protein extended for an additional three amino acids by the integration of complete and partial copies of the ISAba125 element (69). Collectively, these results infer that H-NS is a hot spot for disruption in A. baumannii, where a number of different integration sites have now been identified.
Transposable elements are a key driving force in the worrying increase in MDR isolates across many bacterial species, particularly within the Acinetobacter genus. Despite their small size, MITEs have been shown to be a significant contributor to genetic variation in a number of pathogens. In conclusion, this work has identified and characterized a new MGE, MITE Aba12 , and determined its prevalence across the Moraexellaceae family. This also led to the identification of a novel composite transposon in M. osloensis CCUG 350, Tn6645. Due to the relatively small number of MITE Aba12 copies identified in sequenced genomes, the element may be maintained neutrally or under tight regulatory control from a yet-to-be-identified host and/or environmental factor(s). The full effects of MITE Aba12 on genetic variation and, thus, evolution of bacterial genomes, in addition to transcriptional and translational influences, have yet to be experimentally examined, opening a new and exciting avenue of research. The overall findings of this study not only illustrate the fluidity of the Acinetobacter pangenome but also highlight the importance of mobile sequences as vehicles for niche-specific adaptive evolution in a number of clinically and environmentally relevant bacterial pathogens.

MATERIALS AND METHODS
Bacterial strains, plasmids, media, and growth conditions. A. baumannii ATCC 17978 (70) was obtained from the American Type Culture Collection (ATCC) and is designated the WT strain in all analyses. Bacterial strains and plasmids are summarized in Table 3, and primers are listed in Table 4. All bacterial strains used in the study were grown in LB broth or on LB agar plates and incubated under aerobic conditions overnight (ON) (16 to 20 h) at 37°C unless otherwise stated. Antibiotic concentrations used for selection purposes were 100 g/ml ampicillin and 25 g/ml erythromycin, unless otherwise stated, and were purchased from AMRESCO and Sigma-Aldrich, respectively.
Construction of deletion and complementation derivatives. A. baumannii ATCC 17978 qseBC (ACX60_06100/05) and ygiW (ACX60_06095) deletion strains were constructed using the RecET recombinase system (71), with modifications as outlined previously (39). Primers used to generate mutant strains are listed in Table 4. For complementation of insertionally inactivated hns genes identified in this study, a previously generated pWH1266 shuttle vector carrying a WT copy of hns amplified from A. baumannii ATCC 17978 chromosomal DNA (pWH0268) was used to transform appropriate A. baumannii strains as previously described (27).
Desiccation survival assay. Desiccation survival assays followed the method outlined previously (8), with modifications. Briefly, ON cultures were diluted 1:25 in fresh LB broth and grown to late log phase (OD 600 of 0.8 to 1.0). Cells were subsequently washed three times in sterile distilled water and diluted to an OD 600 of 0.1. A total of 300 l was pipetted into the center of individual wells of 6-well culture plates and placed in a laminar-flow hood ON at 25°C to dry. All plates were incubated at 21°C with a relative humidity of 30% Ϯ 2%, maintained by the addition of saturated CaCl 2 within sealed plastic boxes. Humidity and temperature were monitored over the 30-day time course using a thermohygrometer. CFU were assessed on days 0, 1, 3, 5, 7, 9, 15, 21, and 30. For viable cell quantification, desiccated cells were rehydrated in sterile phosphate-buffered saline, scraped from their respective wells, and serially diluted. Suspensions of diluted cells were plated on LB agar and incubated ON, and desiccation survival was calculated from the number of CFU/ml. Experiments were undertaken in two biological replicates from two independent experiments. Average CFU and standard errors of the means were calculated and graphed.
Gene cloning and DNA sequencing. The upstream intergenic and coding regions of hns from hypermotile variants obtained after desiccation stress experiments were PCR amplified using Velocity DNA polymerase (Bioline, Australia) with hns_F and hns_R (Table 4) by following the manufacturer's instructions. Adenosine treatment was undertaken on purified amplicons prior to T/A ligation with pGEMT Easy (Promega) and transformation into E. coli DH5␣ pir. Transformants were screened by PCR, restriction digestion, and DNA sequencing.
Stability of MITE Aba12 in hns. Five colonies were separately inoculated into 10 ml of LB broth and passaged over a 5-day period using a dilution of 1:10,000. From the fifth passage, a loop of confluent bacterial suspension was streaked onto LB agar and incubated ON. A total of three well-isolated colonies from each of the five biological replicates were randomly selected and PCR screened with hns_F and hns_R (Table 3) to identify maintenance of the MITE within the hns gene.
Motility assays. Motility assays for A. baumannii ATCC 17978 WT and mutant derivatives were undertaken as previously described (30). Briefly, a colony was harvested from an LB agar plate grown ON and used to inoculate the center of an LB agar (0.25%) plate. Motility was assessed by visual examination after ON incubation at 37°C. Experiments were performed in duplicate over at least three independent experiments. Images are average representations of results obtained.
Comparative genomics, alignments, and clustering. For generation of multiple DNA sequence alignments of all full-length MITE Aba12 copies identified, sequences were obtained from NCBI GenBank and used as input data using Clustal Omega with default parameter settings applied (https://www.ebi .ac.uk/Tools/msa/clustalo/) (34). Prior to alignment, copies of MITE Aba12 identified in the opposite orientation (IRR to IRL) were reverse complemented. In strains with multiple copies of MITE Aba12 , these were numbered (_#1, _#2, _3#, etc.) based on their order from NCBI BLASTN (2.8.0ϩ) outputs (72). Subgroups in the alignment were defined based on the presence of two or more identical MITE Aba12 sequence arrangements, numbered from 1 to 10 and ordered according to abundance. To generate the MITE Aba12 (c) consensus sequence, all elements of 113 bp in length were used as input data and visualized using WebLogo software (35) (36), and ARNold, a Rho-independent transcription terminator finding tool (http://rna.igmors.u-psud.fr/toolbox/arnold/) (37), respectively. Default settings were applied for all programs mentioned above. Characterization of MITE Aba12 TSDs. A total of 20 bp upstream and downstream of each MITE Aba12 element from BLASTN outputs were used to screen for the presence of TSDs. The AT ratio percentages were calculated based on the number of adenosine or thymidine nucleotides in each of the 9-bp integration sites, and these percentages were plotted against the number of copies harboring each ratio. To identify trends in MITE Aba12 integration sites, all identified TSD sequences were used as input data using WebLogo software (35) with default settings applied.