Selectable Markers for Use in Genetic Manipulation of Extensively Drug-Resistant (XDR) Acinetobacter baumannii HUMC1

Multidrug-resistant (MDR), extensively drug-resistant (XDR), and pan-drug-resistant (PDR) strains of Acinetobacter baumannii have frequently been characterized. The ability of A. baumannii to develop resistance to antibiotics is a key reason this organism has been difficult to study using genetic and molecular biology approaches. Here we report selectable markers that are not only useful but necessary for the selection of drug-resistant transformants in the setting of drug-resistant backgrounds. Use of these selectable markers can be applied to a variety of genetic and molecular techniques such as mutagenesis and transformation. These selectable markers will help promote genetic and molecular biology studies of otherwise onerous drug-resistant strains, while avoiding the generation of pathogenic organisms that are resistant to clinically relevant antibiotics.

I nfections due to Acinetobacter baumannii have been identified by the Infectious Diseases Society of America (IDSA) and the Centers for Disease Control and Prevention (CDC) as a significant public health concern (1, 2). Of particular concern regarding A. baumannii is the exceptionally high frequency of extensively drug-resistant (XDR) strains (2)(3)(4)(5)(6). New prophylactic and therapeutic strategies are needed to combat such strains. The key to development of such novel approaches is a better understanding of pathogenesis of these infections (2,4,5,(7)(8)(9).
The prevailing dogma espouses that a fitness cost is always associated with the acquisition of antibiotic resistance (10)(11)(12). On the contrary, recent reports suggest that a fitness advantage exists for some specific antibiotic resistance mutations in Salmonella enterica serotype Typhi, Pseudomonas aeruginosa, and A. baumannii (13,14). Thus, given the remarkable rise in frequency of XDR A. baumannii clinical strains, the use of an XDR strain is needed to best model clinically relevant infection dynamics in pathogenesis studies.
Unfortunately, our understanding of A. baumannii pathogenesis has been greatly hampered by a lack of available genetic manipulation techniques for highly resistant and clinically relevant strains (15)(16)(17). Advances in microbial genetics have provided tools such as transposon and site-directed mutagenesis that have rapidly improved our ability to study and manipulate organisms of interest (18)(19)(20)(21)(22).
However, such techniques require the use of a selectable marker to allow outgrowth of a desired mutant (23)(24)(25)(26)(27). Selectable markers take advantage of antibiotic resistance cassettes to allow for selection of mutants when grown under antibiotic selective pressure (28). The conundrum is that XDR A. baumannii strains are already resistant to commonly used selectable markers, precluding effective selection of such strains with most traditionally used selectable markers (17,(28)(29)(30)(31)(32). Thus, optimization of selectable markers is critical for the fundamental advancement of molecular biology research with XDR strains.
We have previously published that HUMC1, an XDR A. baumannii clinical blood and lung isolate resistant to all clinically reported antibiotics except colistin, is hypervirulent in murine models of infection (15,16,33,34). Given its virulence and near-pan-drugresistant status, intentional induction of colistin resistance in this strain, for example by inserting the MCR gene, would raise ethical concerns. Thus, while HUMC1 is a very useful model strain for studying pathogenesis, its intrinsic antibiotic resistance has made genetic manipulation challenging. To identify suitable selectable markers for such a resistant strain, we screened 23 compounds that constitute a broad array of antibiotics spanning multiple drug classes. Despite its intrinsic antibiotic resistance, we successfully identified selectable markers that are effective in vitro against HUMC1. Last, we show that supraphysiological levels of a drug, irrelevant to clinical use but achievable in vitro for selection of transformants, can overcome innate drug resistance displayed by an XDR strain.

RESULTS
MIC testing. Based on results generated in the clinical microbiology laboratory at the hospital at which HUMC1 was isolated, A. baumannii HUMC1 was resistant to all clinical antibiotics except for colistin (Table 1). However, we noted that the tetracycline MIC of 12.5 g/ml, while clinically defined as resistant due to an inability to achieve drug levels this high in vivo, was well within the range of concentrations achievable in vitro to enable selection of more-resistant clones. Furthermore, when we tested the related antibiotic doxycycline, we found a lower MIC (Table 1). Finally, two antibacterial agents that are not used clinically, puromycin and zeocin, also had activity against HUMC1 (Table 1).
Tetracycline resistance. Tetracycline resistance is conferred by the tetA gene from pBR322 and commonly found on many plasmids used for molecular biology. The fact that doxycycline retained activity against the strain despite tetracycline resistance suggested that the resistance observed was not due to the tetA gene. We confirmed that tetracycline resistance in the HUMC1 isolate was not due to the presence of the tetA gene. A BLAST search for tetA against the HUMC1 genome did not return any hits, and PCR for tetA using purified HUMC1 genomic DNA (gDNA) was negative as well. Colonies were successfully isolated by plating on agar plates supplemented with 50, 75, or 100 g/ml of tetracycline, and no growth was observed for the nontransformed HUMC1 control, indicating the ability of the tetA gene to be used as a selectable marker in HUMC1, despite clinically defined tetracycline resistance.
The purified pABBR_GFP plasmid was transformed into HUMC1 isolate, and transformants were selected by plating on tryptic soy agar (TSA) plate with 100 g/ml of tetracycline. Expression of green fluorescent protein (GFP) was confirmed in transformed HUMC1 with nontransformed HUMC1 as a negative control using a fluorescence microscope (Fig. 1).
Zeocin resistance. Zeocin is an antibiotic that is not used clinically. Resistance to zeocin is conferred by the Sh ble gene. Unfortunately, plasmids that contain the Sh ble gene with an Acinetobacter origin of replication are not readily available, so we developed pMSG360Zeo_AB and pCR-Blunt II-TOPO_AB (Fig. 2). Successful transformants were selected by plating on low-salt Luria-Bertani broth (LB) agar supplemented with 250 g/ml zeocin. The presence of the plasmid was further verified in the transformants by PCR.
In order to demonstrate efficacy of zeocin selection to maintain HUMC1 transformants, the Sh ble resistance gene from pCR-Blunt II-TOPO was cloned into pMU125 to form pMU125_ZeoR and was transformed into HUMC1 isolate. Transformants were selected for on low-salt LB agar supplemented with 250 g/ml zeocin. Successful transformation of pMU125_ZeoR into HUMC1 was confirmed by fluorescence microscopy ( Fig. 1).
Puromycin resistance. Resistance to puromycin is conferred by the pac gene encoding puromycin N-acetyltransferase (PAC). The pac open reading frame was cloned from pBacPuroR-NeoR by PCR and inserted into pABBR_MCS by the Gibson assembly method to form pABBR_PuroR (Table 2 and Fig. 2). The plasmid was sequenced, and it was confirmed that the pac open reading frame was in frame with the promoter and no mutations were present. The plasmid allowed for selection of puromycin-resistant colonies in Escherichia coli, but transformation of the plasmid was unable to confer puromycin resistance in A. baumannii. We attempted a second plasmid construct design in which the pac resistance cassette from pBacPuroR-NeoR vector was left intact. pBacPuroR_AB was formed by cloning the A. baumannii ori region from pABBR_MCS and assembling it into a linearized pBacPuroR-NeoR using the Gibson assembly method ( Table 2 and Fig. 2). Sanger sequencing was done to confirm the proper assembly of the construct. This second construct version allowed for selection of E. coli transformants, but once again we were unable to select A. baumannii transformants.

DISCUSSION
Standard clinical definitions and classifications of drug sensitivity for microbes are based on achievable levels of antibiotics in the body (35). However, these definitions can be unnecessarily conservative when considering in vitro use as a selectable marker for genetic manipulation. It is possible to achieve significantly higher drug concentrations in vitro than in vivo (plasma, serum, bone, tissue, etc.). Here we have demonstrated that concentrations of tetracycline unachievable in vivo can be easily used in vitro for selection of "highly drug-resistant" mutants in a clinically drug-resistant strain. Furthermore, we found that the XDR strain was susceptible to several selectable markers that are not used as clinical antibiotics and also to a drug (doxycycline) that is used clinically but was not reported by the clinical microbiology laboratory. Thus, we emphasize the need to conduct systematic screens of potential selectable markers not limited by presumptions based on resistance profiles reported clinically.
Previous efforts have attempted to introduce and optimize standard genetic and molecular biology techniques in A. baumannii such as transformation, gene knockout, and transposon libraries (27,32,36). However, there are still relatively few molecular tools that have been validated for use in A. baumannii compared to other bacterial species such as E. coli, Bacillus subtilis, Staphylococcus aureus, and Mycobacterium tuberculosis. For example, there were no Acinetobacter plasmids available through  Addgene.org (a nonprofit plasmid repository) (Cambridge, MA) at the time of this publication. Validation and standardization of these basic tools will benefit the research community in general and make Acinetobacter research more accessible. We attempted to develop our constructs conferring resistance in one of two ways. First, the coding sequence (CDS) region of the antibiotic resistance gene (tetA, Sh ble, or pac) was cloned in frame with the bla promoter which is recognized by the highly conserved sigma-70 (rpoD) "constitutive housekeeping" promoter. The sigma-70 sigma factor is highly conserved in E. coli and A. baumannii so it was reasonable to hypothesize that A. baumannii transcription machinery would successfully recognize the bla promoter and express the transgene in a manner similar to that in E. coli. Sequencing of the assembled construct confirmed that the pac gene had replaced the bla open reading frame (ORF) in frame with the promoter. As that method did not work, we next tried to leave the promoter region of the pac gene intact and instead add the A. baumannii ori sequence to the pBacPuroR_NeoR plasmid. The promoter sequence differed from the bla promoter that was present in pABBR_MCS so it was reasonable that a change in the promoter sequence would improve gene expression; however, this approach was also unsuccessful.
Thus, we were unable to develop a functional puromycin selectable marker in A. baumannii despite the functional activity displayed by E. coli transformants. This difficulty could be due to the use of genetic elements that have not been optimized for expression in A. baumannii such as the promoter elements and codon sequence. While we were unable to express a functional pac gene in Acinetobacter, successful expression may be possible with a different promoter or codon optimized sequence. Additionally, the robustness of the antibiotic resistance conferred by the Sh ble and tetA genes used in the plasmids could be improved with similar promoter and codon optimization considerations.
We also observed that A. baumannii ATCC 17978 and HUMC1 were susceptible to drugs that are not used clinically, including puromycin and zeocin. This is most likely due to lack of exposure to these antimicrobial agents so selective pressure has not promoted mutants with resistance to these drugs. Recent publications have shown that other nonclinically relevant antimicrobials, such as tellurite, can be used for in vitro selection schemes (30,37). Further effort to characterize selection systems, for drug resistance strains in particular, for basic science purposes continues to be of value.
A national surveillance study of U.S. intensive care units found that 50% of clinical isolates of A. baumannii were carbapenem-resistant, XDR strains (38). Further research is needed to better understand the basic physiology and host-pathogen interactions of the most difficult-to-treat and most lethal drug-resistant strains. Molecular tools such as selectable markers are needed to facilitate basic genetic studies and engender further research of these intractable strains. Our results enable transformation of antibioticresistant strains of A. baumannii by identifying alternative selectable markers and establishing effective constructs that are potentially useful in spite of an XDR phenotype.

MATERIALS AND METHODS
Bacterial strains. E. coli DH5␣, A. baumannii HUMC1 (15,16,33,34), and A. baumannii ATCC 17978 were cultured using aseptic technique. Single colonies were first streaked out on tryptic soy agar (TSA) from frozen glycerol stocks. Single colonies were picked and used to inoculate overnight broth cultures in tryptic soy broth (TSB).
Overnight cultures of the bacteria (A. baumannii HUMC1 or ATCC 17978) grown in TSB were diluted 1:100 into Mueller-Hinton II (MH2) broth and subcultured in a shaking incubator at 200 rpm and 37°C until the optical density at 600 nm (OD 600 ) reached 0.5. Bacteria were diluted to a working concentration of 1 ϫ 10 6 CFU/ml. The bacterial density was confirmed by plating serial dilutions on TSA and counting CFU.
MIC assays were conducted in standard, sterile, round-bottom (U-shaped), 96-well plates. Drug dilutions were done by serial twofold dilutions across plate columns. Wells of bacteria and media alone were included as positive and negative controls, respectively. One hundred microliters of 1 ϫ 10 6 CFU/ml bacterial culture was added to each one of the requisite wells. The plates were incubated for 24 h at 37°C.
Twenty microliters of 0.1% resazurin dye was added to each well, and metabolism of the dye was measured after 1 h.
Transformation. (i) Acinetobacter baumannii. A. baumannii cells were made electrocompetent according to published protocols (36). Briefly, 500 l of an overnight culture was used to inoculate 50 ml of TSB medium, and the subculture was incubated until it reached an OD 600 of 0.5. The cells were pelleted by centrifugation (10 min at 10,000 ϫ g) and washed five times with 1 ml of 10% glycerol. The cells were separated into 100-l aliquots and stored at Ϫ80°C for later use as we have previously described (33).
(ii) Escherichia coli. Chemically competent or electrocompetent E. coli DH5␣ cells were used for the transformations. E. coli DH5␣ competent cells were made using the Mix & Go E. coli transformation kit per the manufacturer's suggested protocol (catalog no. T3001; Zymo Research). Briefly, the DNA was incubated with competent cells on ice for 1 h prior to plating on TSA supplemented with 10 g/ml tetracycline, 25 g/ml zeocin, 125 g/ml puromycin, 50 g/ml kanamycin, or 100 g/ml ampicillin. The concentration of antibiotics used for selection of E. coli was chosen according to the manufacturer's directions. Electrocompetent E. coli DH5␣ cells were prepared using the same methods as described above for A. baumannii.
Construct assembly. The constructs were assembled using the Gibson assembly method (Fig. 2) (20). Overlap sequences for the vector and insert were determined using the NEBuilder assembly tool (New England BioLabs). Vector backbones were prepared by PCR amplification of plasmid DNA or by restriction enzyme digestion. Assembly of the parts to create the final constructs was accomplished using the NEBuilder HiFi assembly master mix per the manufacturer's protocol. Briefly, the corresponding linearized vector (100 ng) and insert were added in a 1:2 molar ratio of vector to insert. The linear fragments were incubated with 10 l of enzyme master mix at 50°C for 15 min. Two microliters of the assembly product was then used for bacterial transformation.
Preparation of the vector backbone and insert sequence for each plasmid were done as follows. For pBacPuroR_AB, the pBacPuroR-NeoR vector backbone was linearized by PCR, and the A. baumannii ori insert sequence was amplified by PCR from pABBR_MCS. For pABBR_PuroR, the pABBR vector backbone was linearized by PCR, and the puromycin resistance cassette insert sequence was amplified by PCR from pBacPuroR-NeoR. For pCR-Blunt II_AB, the pCR-Blunt II_AB vector backbone was linearized by PCR, and the A. baumannii ori insert sequence was amplified by PCR from pABBR_MCS. For pMSG360zeo_AB, the pMSG360zeo vector backbone was linearized by PCR, and A. baumannii ori insert sequence was amplified by PCR from pABBR_MCS. For pABBR_GFP, the linearized vector backbone was prepared by digestion with SapI, and the gfp insert sequence was amplified by PCR from pMU125. For pMU125_ZeoR, the linearized vector backbone was prepared by digestion with SapI, and the zeocin resistance cassette insert was amplified by PCR from pCR-Blunt II-TOPO.