*Article* **Colistin and Carbapenem-Resistant** *Acinetobacter baumannii* **Aci46 in Thailand: Genome Analysis and Antibiotic Resistance Profiling**

**Nalumon Thadtapong <sup>1</sup> , Soraya Chaturongakul 2,3, Sunhapas Soodvilai <sup>1</sup> and Padungsri Dubbs 2,3,\***


**Abstract:** Resistance to the last-line antibiotics against invasive Gram-negative bacterial infection is a rising concern in public health. Multidrug resistant (MDR) *Acinetobacter baumannii* Aci46 can resist colistin and carbapenems with a minimum inhibitory concentration of 512 µg/mL as determined by microdilution method and shows no zone of inhibition by disk diffusion method. These phenotypic characteristics prompted us to further investigate the genotypic characteristics of Aci46. Next generation sequencing was applied in this study to obtain whole genome data. We determined that Aci46 belongs to Pasture ST2 and is phylogenetically clustered with international clone (IC) II as the predominant strain in Thailand. Interestingly, Aci46 is identical to Oxford ST1962 that previously has never been isolated in Thailand. Two plasmids were identified (pAci46a and pAci46b), neither of which harbors any antibiotic resistance genes but pAci46a carries a conjugational system (type 4 secretion system or T4SS). Comparative genomics with other polymyxin and carbapenemresistant *A. baumannii* strains (AC30 and R14) identified shared features such as CzcCBA, encoding a cobalt/zinc/cadmium efflux RND transporter, as well as a drug transporter with a possible role in colistin and/or carbapenem resistance in *A. baumannii*. Single nucleotide polymorphism (SNP) analyses against MDR ACICU strain showed three novel mutations i.e., Glu229Asp, Pro200Leu, and Ala138Thr, in the polymyxin resistance component, PmrB. Overall, this study focused on Aci46 whole genome data analysis, its correlation with antibiotic resistance phenotypes, and the presence of potential virulence associated factors.

**Keywords:** *Acinetobacter baumannii*; colistin; carbapenems; multidrug resistant; WGS

## **1. Introduction**

*Acinetobacter baumannii* is an opportunistic pathogenic bacterium that causes nosocomial infections in immunocompromised patients, especially patients treated in the intensive care unit (ICU) [1,2]. *A. baumannii* infections usually occur following: trauma, surgery, catheterization, or endotracheal intubation [3]. Moreover, this bacterium is well known for its multidrug resistant (MDR) characteristics, defined as resistance to at least one agent in three or more antibiotic categories [4], and as a nosocomial ESKAPE pathogen, a group including: *Enterococcus faecium*, *Staphylococcus aureus*, *Klebsiella pneumoniae*, *A. baumannii*, *Pseudomonas aeruginosa*, and *Enterobacter* species [5]. *A. baumannii* can resist almost all available antibiotics and it is possible for a strain to be pan drug resistant (PDR), which is defined as resistant to all agents in all antibiotic categories including last-resort antibiotics (carbapenems and polymyxins) [4].

Carbapenem-resistant *A. baumannii* or CRAB is considered by WHO (World Health Organization) as one of the leading threats to global human healthcare [6]. During the

**Citation:** Thadtapong, N.; Chaturongakul, S.; Soodvilai, S.; Dubbs, P. Colistin and Carbapenem-Resistant *Acinetobacter baumannii* Aci46 in Thailand: Genome Analysis and Antibiotic Resistance Profiling. *Antibiotics* **2021**, *10*, 1054. https://doi.org/10.3390/ antibiotics10091054

Academic Editor: Teresa V. Nogueira

Received: 29 July 2021 Accepted: 27 August 2021 Published: 30 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

COVID-19 pandemic, CRAB infections have increased among COVID-19 patients who are subjected to long-term stays in the ICU [7,8]. The incidence of *A. baumannii* infection in Thailand is widely distributed in all regions of the country [9]. Clinical isolates of *A. baumannii* accounted for 15–16% of hospital-acquired bacteremia and CRAB comprises 70–88% of total *A. baumannii* clinical isolates [10]. Presence of the carbapenemase encoding genes, *blaOXA-23* or *blaOXA-51*, in combination with insertion sequence elements is frequently found in CRAB [11].

Unfortunately, colistin-resistant *A. baumannii* strains that are either MDR or PDR have been reported worldwide [12,13]. Polymyxins, including polymyxin B and colistin, are alternative last-line drugs used against CRAB [14]. Cationic polymyxin molecules target the polyanionic portions of the outer membrane of the bacterial envelope, specifically lipid A in lipopolysaccharide (LPS) [15]. Polymyxins bind lipid A and disrupt the outer membrane, causing cytoplasm leakage [16]. In Thailand, colistin-resistant *A. baumannii* isolates are found in 35–44% of pneumonia patients [17,18]. Among the colistin-resistant *A. baumannii* strains, four known mechanisms have been identified: (i) modification of lipid A, (ii) loss of LPS, (iii) disruption of outer membrane asymmetry, and (iv) efflux pumps [19]. Modification of lipid A involves increased addition of phosphoethanolamine (PEtN) to lipid A resulting from a mutation in the PmrAB two-component system. Increased expression of *pmrC*, which encodes lipid A phosphoethanolamine transferase, enhances addition of PEtN to either the 40 -phosphate or 10 -phosphate of lipid A [20]. Inactivation or complete loss of LPS occurs as a result of mutations in LPS biosynthesis genes such as: *lpxA*, *lpxC*, and *lpxD* [21]. Mutations in *vacJ*, encoding an outer membrane lipoprotein, and *pldA*, encoding an outer membrane phospholipase, result in accumulation of phospholipid, disrupting LPS organization, membrane asymmetry, and colistin binding [22]. Efflux pumps have also been shown to play important roles in the osmotic stress response and colistin resistance, specifically the AdeRS two-component system, which regulates expression of the AdeABC [23] and EmrAB efflux pumps [24].

Our previous report characterized MDR *A. baumannii* Aci46 that was isolated from a pus sample from a Thai patient at Ramathibodi hospital [25]. This strain was reported as a CRAB, i.e., resistant to imipenem. In the current study, we further explored the Aci46 resistance profile against other antibiotics including the alternative last-line drug, colistin. Since the genotypic characteristics of the MDR Aci46 were still unknown, genome studies were applied to identify antibiotic resistance genes, sequence type, and international clonal (IC) group of Aci46. In addition, the Aci46 genome was compared with other MDR *A. baumannii* and drug sensitive *A. baumannii* to characterize the unique gene features of the MDR (or CRAB) Aci46.

#### **2. Results and Discussion**

#### *2.1. Antibiotic Resistance Phenotypes of Aci46*

A previous report has shown that Aci46 is susceptible to amikacin and resistant to cefoperazone-sulbactam, ceftazidime, ciprofloxacin, and imipenem [25]. To expand the antibiotic profile of Aci46, twenty drugs from eight classes (i.e., aminoglycosides, beta-lactams (and beta-lactam combined), carbapenems, quinolones, folate pathway blocks, phenicol, tetracycline, and colistin) were used in disk diffusion and microdilution assays. We found that Aci46 was resistant to all twenty drugs (Table 1, Figure S1 and Supplement Table S1) with a minimum inhibitory concentration (MIC) for colistin of 512 µg/mL (Figure 1). We have demonstrated that Aci46 is an XDR (extensively drug resistant) strain according to the definition described by Magiorakos et al. (non-susceptible to at least one agent in all but two antimicrobial categories specified for *Acinetobacter* spp.) [4].


**Table 1.** Antibiotic susceptibility profiling of *A. baumannii* Aci46 by disk diffusion method.

*Antibiotics* **2021**, *10*, 3 of 17

\* The ranges of inhibition zones were calculated from three individual replicates and "0" means no inhibition zone. \* The ranges of inhibition zones were calculated from three individual replicates and "0" means no inhibition zone.

**Figure 1.** Minimum inhibitory concentration (MIC) determination by microdilution assay using MTT staining. Aci46 viable cells and positive control are shown in purple while dead cells **Figure 1.** Minimum inhibitory concentration (MIC) determination by microdilution assay using MTT staining. Aci46 viable cells and positive control are shown in purple while dead cells or negative control are shown in yellow. The experiments were tested with three biological replicates.

#### *2.2. Whole Genome Sequencing Data*

To further investigate the genetic makeup of XDR Aci46, the chromosome and plasmids of Aci46 were subjected to next-generation sequencing. The summarized genome data is shown in Table 2. The Aci46 genome size is 3,887,827 bp with a GC content of 38.87%. The number of predicted protein coding sequences, rRNA genes, and tRNA genes were 3754, 3, and 63, respectively. We also identified two plasmids from the whole genome data, namely pAci46a and pAci46b. The size of pAci46a was 70,873 bp with a GC content of 33.39% while pAci46b was 8808 bp with a GC content of 34.31%. The number of predicted protein coding sequences for pAci46a and pAci46b were 102 and 11, respectively. No rRNA or tRNA genes were present in either case.


**Table 2.** Standard data of Aci46 genome features.

The microbial taxonomy of Aci46 was confirmed as *A. baumannii* at 100% identity based on variation of 54 genes encoding ribosomal protein subunits. Typing of Aci46 was classified by multi-locus sequence typing (MLST) using Oxford and Pasture schemes. The sequence type (ST) of Aci46 was ST1962 (*gltA*-1, *gyrB*-3, *gdhB*-189, *recA*-2, *cpn60*-2, *gpi*-140, *rpoD*-3) based on the Oxford scheme [26], while it belonged to ST2 (*cpn60*-2, *fusA*-2, *gltA*-2, *pyrG*-2, *recA*-2, *rplB*-2, *rpoB*-2) based on the Pasture scheme [27]. ST2 based on Pasture scheme is a predominant ST of CRAB found in Thailand and Southeast Asia [10,28]. However, ST1962 based on the Oxford scheme has never been reported in Thailand. ST1962 has been reported in the USA for only one strain (PubMLST database, to be published). From our previous report, we knew that Aci46 harbored class 1 integrase [25]. However, the class 1 integron can be transferred across two *A. baumannii* IC groups, IC I and IC II [29]. Phylogenetic analysis of the Aci46 genome compared with ten genomes of *A. baumannii* from three different IC's, with the *A. baylyi* ADP1 genome as an outgroup to root the tree, revealed that Aci46 is more closely related to *A. baumannii* MDR-ZJ06 and belongs to IC II (Figure 2).

**Figure 2.** Phylogenetic analysis of *A. baumannii* Aci46 and ten sequences of *A. baumannii* genomes for identification of international clonal group (IC). The phylogenetic tree was constructed by RAxML version 8.2.11 using 100 single-copy genes with bootstrap values set to 100 replicates. Selected strains belonging to IC I, IC II, and IC III are labeled. The accession numbers of the strains are as follows: AYE (CU459141.1), AB0057 (CP001182.2), AB307-0294 (CP001172), AC30 (ALXD00000000), ACICU (CP031380.1), MDR-ZJ06 (CP001937.2), R14 (PUDB01000000), ATCC17978 (CP000521.1), SDF (CU468230.2), and *A. baylyi* ADP1 (CR543861). **Figure 2.** Phylogenetic analysis of *A. baumannii* Aci46 and ten sequences of *A. baumannii* genomes for identification of international clonal group (IC). The phylogenetic tree was constructed by RAxML version 8.2.11 using 100 single-copy genes with bootstrap values set to 100 replicates. Selected strains belonging to IC I, IC II, and IC III are labeled. The accession numbers of the strains are as follows: AYE (CU459141.1), AB0057 (CP001182.2), AB307-0294 (CP001172), AC30 (ALXD00000000), ACICU (CP031380.1), MDR-ZJ06 (CP001937.2), R14 (PUDB01000000), ATCC17978 (CP000521.1), SDF (CU468230.2), and *A. baylyi* ADP1 (CR543861).

#### *2.3. Antibiotic Resistance Gene, Efflux Pump, and Virulence Gene Predictions 2.3. Antibiotic Resistance Gene, Efflux Pump, and Virulence Gene Predictions*

Based on the phenotypic characteristics of the antibiotic resistance profile, a search forthe presence of antibiotic resistance genes and genes encoding efflux pumps associated with the XDR phenotype in Aci46 was undertaken. ResFinder, CARD, and NDARO databases were used to predict the antibiotic resistance genes and efflux pumps present in Aci46. We found that Aci46 harbored sixteen resistance genes against eight classes of drugs (i.e., aminoglycosides, beta-lactams/carbapenems, beta-lactams/cephalosporins, colistin, fluoroquinolones, macrolides, tetracycline, and sulfonamide) and twenty-two genes belonging to five classes of drug transporters (i.e., RND (resistance-nodulation-division) efflux systems, MFS (major facilitator superfamily) family transporter, ABC (ATPbinding cassette) transporter, MATE (multidrug and toxic compound extrusion) family transporter, and SMR (small multidrug resistance)) (Table 3). The genes, *blaOXA-23*, *blaOXA-66* or *blaOXA-51*-like, and *oprD* genes are present in Aci46 and they are known confer carbapenem-resistance [11]. These data correlate with presence of *blaOXA-23* and *blaOXA-51* in other CRAB isolates found in Thailand [30]. Moreover, class 1 and class 2 integrase genes and *blaOXA-23* are often found in XDR *A. baumannii* [31,32]. Based on the phenotypic characteristics of the antibiotic resistance profile, a search for the presence of antibiotic resistance genes and genes encoding efflux pumps associated with the XDR phenotype in Aci46 was undertaken. ResFinder, CARD, and NDARO databases were used to predict the antibiotic resistance genes and efflux pumps present in Aci46. We found that Aci46 harbored sixteen resistance genes against eight classes of drugs (i.e., aminoglycosides, beta-lactams/carbapenems, beta-lactams/cephalosporins, colistin, fluoroquinolones, macrolides, tetracycline, and sulfonamide) and twenty-two genes belonging to five classes of drug transporters (i.e., RND (resistance-nodulationdivision) efflux systems, MFS (major facilitator superfamily) family transporter, ABC (ATP-binding cassette) transporter, MATE (multidrug and toxic compound extrusion) family transporter, and SMR (small multidrug resistance)) (Table 3). The genes, *blaOXA-23*, *blaOXA-66* or *blaOXA-51*-like, and *oprD* genes are present in Aci46 and they are known confer carbapenem-resistance [11]. These data correlate with presence of *blaOXA-23* and *blaOXA-51* in other CRAB isolates found in Thailand [30]. Moreover, class 1 and class 2 integrase genes and *blaOXA-23* are often found in XDR *A. baumannii* [31,32].

With regard to colistin resistance, we identified *lpxA* and *lpxC* in Aci46, these genes might play a role in loss of LPS and leading to colistin resistance [21]. In addition, we found four genes encoding efflux pumps (i.e., *adeR*, *adeS*, *emrA*, and *emrB*) and two genes with roles in lipid modification (i.e., *pmrA* and *pmrB*). AdeRS is a two-component system that regulates the expression of the AdeABC efflux pump, which is an RND efflux system [23]. The EmrAB efflux system belongs to the MFS family of transporters [24]. The PmrAB two-component system regulates PmrC expression. PmrC adds PEtN to lipid A [33]. In summary, the genotypic characteristics of Aci46 suggest that Aci46 resists colistin by way of lipid A modification, LPS loss, and AdeABC-mediated efflux. With regard to colistin resistance, we identified *lpxA* and *lpxC* in Aci46, these genes might play a role in loss of LPS and leading to colistin resistance [21]. In addition, we found four genes encoding efflux pumps (i.e., *adeR*, *adeS*, *emrA*, and *emrB*) and two genes with roles in lipid modification (i.e., *pmrA* and *pmrB*). AdeRS is a two-component system that regulates the expression of the AdeABC efflux pump, which is an RND efflux system [23]. The EmrAB efflux system belongs to the MFS family of transporters [24]. The PmrAB two-component system regulates PmrC expression. PmrC adds PEtN to lipid A [33]. In summary, the genotypic characteristics of Aci46 suggest that Aci46 resists colistin by way of lipid A modification, LPS loss, and AdeABC-mediated efflux.


#### **Table 3.** List of predicted antibiotic resistance and drug transporter genes.

Several virulence factors of *A. baumannii* have been identified by genome-based analysis [34]. The outer membrane protein, OmpA, which functions as a porin, is a key factor in virulence where it plays particular roles in cell invasion, development of cytotoxicity, and

apoptosis [35]. Capsular polysaccharides and LPS are also virulence factors and contribute to serum resistance, biofilm formation, and escape from the host immune response [36]. *A. baumannii* uses combined strategies, namely, bacterial fitness and pathogenicity, to cause disease in humans [35]. PAI (pathogenicity islands), such as prophages and secretion systems, have also been implicated in virulence and pathogenicity [37,38]. In Aci46, we have identified pathogenicity islands comprising four prophages, one T4SS (type four secretion system), one T6SS (type six secretion system), and one ICE (integrative and conjugation element) (Table 4 and Supplement Table S2). No antibiotic resistance genes were found on the plasmids and prophages. Genotypic characteristics underlying the antibiotic resistance profile of Aci46 are found on its chromosome, and not on plasmids or other mobile genetic elements. T4SS is located in plasmid pAci46a, similar to pAC30c in *A. baumannii* AC30 and pAC29b in *A. baumannii* AC29 [2]. Generally, T4SS plays a role in the transfer antibiotic resistance genes via horizontal gene transfer [38]. Based on comparative genome analysis, T4SS loci are found in clinical isolates associated with hospital outbreaks [39]. The function of T4SS is still unclear in *A. baumannii*, but it might be implicated in pathogenicity or host–pathogen interaction [38,40]. Thus, pAci46a might play a role in pathogenesis instead of drug resistance. Moreover, we found *attL* (gtaataacaaagcaatcccgcagggttgcgacaaatagccctctaaatcgctctaattgcccctagattcaatttta) and *attR* (gtaataacaaagcaatcccgcagggttgcgacaaatagccctctaaatcgctctaattgcccctagattcaatttta) sites on pAci46a (or ICE region). It is possible that pAci46a could be a conjugative plasmid responsible for plasmid mobilization, similar to pAC30c and pAC29b [2]. T6SS injects toxic effectors into other bacteria in the same niche; therefore, it is an important factor for competitive killing and host colonization [34]. The plasmid, pAci46b, carries eleven genes encoding: one outer membrane receptor protein, one replication protein, and nine hypothetical proteins. The functions of pAci46b are still unclear.


## *2.4. Comparative Pangenomic Analysis against Other A. baumannii Strains*

In order to further explore the novel gene(s) that might be involved in the colistin and carbapenem-resistant phenotypes in Aci46, pangenome analysis was performed (Figure 3). Pangenome (all genes from all genomes) and core genes (present in all genomes) (Supplement Table S7) comprise 4605 and 2754 genes, respectively. We found that the number of strain-specific genes for Aci46, ACICU, ATCC17978, AC30, and R14 are 45, 193, 382, 91, and 145, respectively (Supplement Table S3). MDR (Aci46, ACICU, AC30, and R14) (Supplement Table S4), polymyxin and carbapenem resistance (renamed PCRAB) (Aci46, AC30, and R14), and colistin and carbapenem resistance (renamed CCRAB) (Aci46 and R14) groups contained 504, 34, and 8 accessory genes (i.e., genes present in specific genomes), respectively. We hypothesized that the accessory genes in PCRAB and/or CCRAB might be involved in resistance to polymyxins (polymyxin B and colistin). The thirty-four PCRAB specific genes encode: amidase, carbapenemase BlaOXA-23, Czc-CBA cobalt/zinc/cadmium efflux RND transporter, twenty-three hypothetical proteins, lysophospholipid, nickel-cobalt-cadmium resistance protein, oxidoreductase, transcriptional regulator, and DNA-methyltransferase subunit M (Supplement Table S5). The eight CCRAB specific genes encode: one primosomal protein I and seven hypothetical pro-

teins (Supplement Table S6). One of the PCRAB specific genes, *blaOXA-23*, is frequently reported in CRAB [41,42]. Interestingly, three genes (*czcA*, *czcB*, and *czcC*) encode Czc-CBA cobalt/zinc/cadmium efflux RND transporters in PCRAB specific genes. CzcCBA cobalt/zinc/cadmium efflux RND transporters have functions in exporting cations (Co2+ , Zn2+, and Cd2+) from the cytoplasm and confer heavy metal resistance [43,44], and are reported to be associated with the XDR phenotype in *A. baumannii* [45]. This efflux system might function in colistin resistance in PCRAB, including Aci46, by exporting colistin and polymyxin (cationic molecules). In the case of CCRAB specific genes, their functions are unknown. *Antibiotics* **2021**, *10*, 9 of 17 Zn2+ , and Cd2+) from the cytoplasm and confer heavy metal resistance [43,44], and are reported to be associated with the XDR phenotype in *A. baumannii* [45]. This efflux system might function in colistin resistance in PCRAB, including Aci46, by exporting colistin and polymyxin (cationic molecules). In the case of CCRAB specific genes, their functions are unknown.

**Figure 3.** Venn diagram of comparative pangenome analysis among *A. baumannii* strains. The yellow, blue, green, white, and orange represent Aci46, ACICU, ATCC17978, AC30, and R14, respectively, with the number of total genes and the character of drug resistance in each genome (shown in green). The overlapped areas show genes encoding shared protein families among strains. The number of genes in core genes (Aci46, ACICU, ATCC17978, AC30, and R14), MDR (Aci46, ACICU, AC30, and R14), PCRAB (Aci46, AC30, and R14), and CCRAB (Aci46 and R14) the specific groups are labeled in white, yellow, red, and blue, respectively. CRAB: carbapenem-resistant *A. baumannii*, CCRAB: colistin and carbapenem-resistant *A. baumannii*, PCRAB: polymyxins and carbapenem-resistant *A. baumannii*. **Figure 3.** Venn diagram of comparative pangenome analysis among *A. baumannii* strains. The yellow, blue, green, white, and orange represent Aci46, ACICU, ATCC17978, AC30, and R14, respectively, with the number of total genes and the character of drug resistance in each genome (shown in green). The overlapped areas show genes encoding shared protein families among strains. The number of genes in core genes (Aci46, ACICU, ATCC17978, AC30, and R14), MDR (Aci46, ACICU, AC30, and R14), PCRAB (Aci46, AC30, and R14), and CCRAB (Aci46 and R14) the specific groups are labeled in white, yellow, red, and blue, respectively. CRAB: carbapenem-resistant *A. baumannii*, CCRAB: colistin and carbapenem-resistant *A. baumannii*, PCRAB: polymyxins and carbapenem-resistant *A. baumannii*.

#### *2.5. Pairwise SNP Analysis*

*2.5. Pairwise SNP Analysis* Although we identified genes that were specific to PCRAB that encoded CzcCBA efflux pumps, known MDR genes were among the core genes (Supplement Table S7). Thus, in order to identify resistance-associated mutations in MDR and CCRAB strains, non-synonymous SNPs between Aci46 and ATCC17978 and between Aci46 and ACICU were identified. SNPs within known MDR genes from the core genes are listed in Table 5. Twenty genes show SNPs in Aci46 vs. ATCC17978 i.e., seven antibiotic resistance genes (*blaADC-25*, *blaOXA-66*, *lpxA*, *lpxC*, *pmrB*, *gyrA*, and *gyrB*) and thirteen drug transporter genes (*adeA*, *adeB*, *adeF*, *adeG*, *adeH*, *adeJ*, *adeR*, *adeS*, *opmH*, *emrB*, *mdfA*, *macB*, and *abeS*). The deduced amino acid sequences of these genes among the MDR strains were similar. For example, the deduced amino acid sequences of *blaOXA-66* or *blaOXA-51*-like genes in Aci46, ACICU, AC30, and R14 showed conserved amino acids at Val36, Lys107, and Asn225 while ATCC17978 contained Glu36, Gln107, and Asp225 (Figure 4). In 2015, the Although we identified genes that were specific to PCRAB that encoded CzcCBA efflux pumps, known MDR genes were among the core genes (Supplement Table S7). Thus, in order to identify resistance-associated mutations in MDR and CCRAB strains, non-synonymous SNPs between Aci46 and ATCC17978 and between Aci46 and ACICU were identified. SNPs within known MDR genes from the core genes are listed in Table 5. Twenty genes show SNPs in Aci46 vs. ATCC17978 i.e., seven antibiotic resistance genes (*blaADC-25*, *blaOXA-66*, *lpxA*, *lpxC*, *pmrB*, *gyrA*, and *gyrB*) and thirteen drug transporter genes (*adeA*, *adeB*, *adeF*, *adeG*, *adeH*, *adeJ*, *adeR*, *adeS*, *opmH*, *emrB*, *mdfA*, *macB*, and *abeS*). The deduced amino acid sequences of these genes among the MDR strains were similar. For example, the deduced amino acid sequences of *blaOXA-66* or *blaOXA-51*-like genes in Aci46, ACICU, AC30, and R14 showed conserved amino acids at Val36, Lys107, and Asn225 while ATCC17978 contained Glu36, Gln107, and Asp225 (Figure 4). In 2015, the Trp22Met mutation of *blaOXA-51* was linked with carbapenem resistance function in

Trp22Met mutation of *blaOXA-51* was linked with carbapenem resistance function in *A. baumannii* [46]. This result suggested that amino acid sequences of antibiotic resistant and

and might be linked to drug resistance level. For colistin resistance, we found three genes

Colistin

*A. baumannii* [46]. This result suggested that amino acid sequences of antibiotic resistant and drug transporter proteins in MDR strains could be different from drug sensitive strains and might be linked to drug resistance level. For colistin resistance, we found three genes (*blaADC-25*, *pmrB*, and *gyrB*) that were mutated in Aci46 vs. ACICU. Of these only *pmrB* is related to colistin resistance. From comparisons of Aci46 vs. ACICU and Aci46 vs. ATCC17978, mutations in PmrB were detected in three positions: Ala138Thr, Pro200Leu, and Glu229Asp (Table 5). Known PmrB mutations that confer colistin resistance are Leu9-Gly12 deletion, Ala22Val, Ile232Thr, and Gln270Pro [47,48]. Hence, Ala138Thr, Pro200Leu, and Glu229Asp mutations in Aci46 PmrB might be novel mutations involved in colistin resistance. *Antibiotics* **2021**, *10*, 10 of 17 (*blaADC-25*, *pmrB*, and *gyrB*) that were mutated in Aci46 vs. ACICU. Of these only *pmrB* is related to colistin resistance. From comparisons of Aci46 vs. ACICU and Aci46 vs. ATCC17978, mutations in PmrB were detected in three positions: Ala138Thr, Pro200Leu, and Glu229Asp (Table 5). Known PmrB mutations that confer colistin resistance are Leu9- Gly12 deletion, Ala22Val, Ile232Thr, and Gln270Pro [47,48]. Hence, Ala138Thr, Pro200Leu, and Glu229Asp mutations in Aci46 PmrB might be novel mutations involved in colistin resistance.

**Figure 4.** Comparisons of deduced amino acid sequences of *blaOXA-66* from five strains of *A. baumannii* by multiple sequence alignment. **Figure 4.** Comparisons of deduced amino acid sequences of *blaOXA-66* from five strains of *A. baumannii* by multiple sequence alignment.


**Table 5.** List of non-synonymous SNPs in antibiotic resistance genes and efflux pumps from core gene group. **Table 5.** List of non-synonymous SNPs in antibiotic resistance genes and efflux pumps from core gene group.

*pmrB* 412G > A Ala138Thr 412G > A Ala138Thr

599C > T Pro200Leu 599C > T Glu229Asp

859A > G Asn287Asp

*lpxC* 358T > C Cys120Arg

1331C > T Ala444Val


**Table 5.** *Cont.*

#### **3. Materials and Methods**

#### *3.1. Bacterial Strains*

*A. baumannii* Aci46 was isolated from a male Thai patient treated at Ramathibodi hospital, Thailand [25]. This strain was isolated from a pus sample, identified by routine biochemical test, and confirmed by *blaOXA51*-like gene detection [25,49]. Aci46 was cultured on MHA (Mueller Hinton Agar, BD Difco, Eysins, Switzerland) and incubated at 37 ◦C for overnight.

#### *3.2. Antibiotic Susceptibility Testing by Disk Diffusion and Microdilution*

Antibiotic susceptibility was determined by disk diffusion method for 19 drugs (gentamicin, kanamycin, streptomycin, cephalothin, cefoxitin, cefotaxime, ceftazidime, ceftriaxone, ampicillin-clavulanic acid, imipenem, meropenem, ciprofloxacin, nalidixic acid, norfloxacin, trimethoprim, trimethoprim-sulfamethoxazole, ampicillin, chloramphenicol, and tetracycline) (Oxoid, Thermo Fisher Scientific, Waltham, MA, USA) and microdilution method for colistin. Aci46 was streaked on MHA and incubated overnight. Colonies were picked and resuspended in normal saline solution at 1 <sup>×</sup> <sup>10</sup><sup>8</sup> cfu/mL (OD<sup>600</sup> = 0.08–0.12 or 0.5 McFarland). The cell suspension was spread on MHA using a cotton swab. Antibiotic disks were placed on the agar surface. After incubation at 37 ◦C for 20–24 h., the zones of inhibition were measured and the results were interpreted following the CLSI (Clinical

and Laboratory Standard Institute) guideline [50]. For the microdilution method, cell suspensions of Aci46 were diluted in CAMHB (Cation-Adjusted Mueller Hinton Broth, BD Difco, Eysins, Switzerland) to 1 <sup>×</sup> <sup>10</sup><sup>6</sup> cfu/mL. Two-fold serial dilutions of colistin were prepared (1–1024 <sup>µ</sup>g/mL) and mixed with 5 <sup>×</sup> <sup>10</sup><sup>5</sup> cfu/mL of Aci46 in 200 µL total volume. After incubation at 37 ◦C for 20–24 h., the minimum inhibitory concentration was observed and cell viability was measured by MTT-based staining [51]. Ten µL of 5 mg/mL MTT (3- (4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (Invitrogen, Life Technologies, Carlsbad, CA, USA) in phosphate buffer saline was added in 100 µL of cell culture and incubated at 37 ◦C, 200 rpm, 1 h in the dark. One hundred µL of 10% SDS (sodium dodecyl sulfate, Merck, Darmstadt, Germany) and 50% DMSO (dimethyl sulfoxide, Sigma-Aldrich, St. Louis, MO, USA) was added and continually incubated at 37 ◦C, 200 rpm, 2 h in the dark. The absorbance of formazan dissolution was detected at 570 nm using a microplate reader (Azure Ao Absorbance Microplate Reader, Azure Biosystems, Dublin, CA, USA). Relative optical density at 570 nm was calculated by dividing OD<sup>570</sup> of drug-containing wells with the OD<sup>570</sup> of drug-free wells [52]. The cut-off for no detection was the relative OD<sup>570</sup> of 0.1. Viable cells under MTT staining can also be observed by the naked eye i.e., color change from yellow to purple. Colistin resistance was determined by CLSI guideline (MIC ≥ 4 µg/mL; resistant) [50]. *Escherichia coli* ATCC25922 was selected to be a control strain for disk diffusion and microdilution methods.

#### *3.3. Genomic DNA Extraction and Whole Genome Sequencing*

Whole genomic DNA of Aci46 was extracted using a modified Marmur procedure [53]. Briefly, Aci46 cells were harvested from 3 mL of cell culture in CAMHB and resuspended in EDTA-saline (0.01 M EDTA and 0.15 M NaCl, pH 8.0). Thirty µL of 110 mg/mL lysozyme and 10 µL of 20 mg/mL RNase A were added and incubated at 37 ◦C for 2 h. After incubation, 80 µL of 20% SDS and 10 µL of 5 mg/mL proteinase K were added and incubated again at 65 ◦C, 30 min. Then, 5 M NaCl was added at 0.5 volume followed by phenol-chloroform extraction. The upper liquid phase was transferred to a new 1.5 mL microcentrifuge tube. A 0.25 volume of 5 M NaCl and a 0.1 volume of 3 M sodium acetate were added and mixed well. Ice-cold absolute ethanol was added at 2 volumes and inverted gently. The DNA pellet was hooked and transferred into a new 1.5 mL microcentrifuge tube, air dried, and resuspended in DNase-RNase-free water. Quality and quantity of DNA were measured by UV spectrophotometer (OD260/OD<sup>280</sup> and OD260/OD<sup>230</sup> ratio) (DeNovix DS-11 FX+ spectrophotometer, DeNovix, Wilmington, DE, USA), Qubit dsDNA BR assay kit (Invitrogen, Life Technologies, Carlsbad, CA, USA), and 1% agarose gel electrophoresis (Bio-rad, Hercules, CA, USA). One hundred ng of extracted DNA was used for library preparation using TruSeq Nano DNA Kit (Illumina, San Diego, CA, USA) followed by pair-end sequencing on Illumina HiSeq platform (Illumina, San Diego, CA, USA).

#### *3.4. Genome Assembly, Annotation, and Pathogenicity Island Prediction*

Raw sequence data of Aci46 were trimmed by Trim Galore version 0.6.3 [54] and the quality was checked using FastQC version 0.11.8 [55]. Trimmed reads were assembled using SPAdes version 3.12.0 [56], corrected assembly error by Pilon version 1.23 [57], and calculated genome coverage by SAMTools version 1.3 [58] in PATRIC (Pathosystems Resource Integration Center) version 3.6.9 [59]. The quality of de novo assembled contigs was assessed by QUAST version 5.0.2 [60] and visualized using Bandage version 0.8.1 [61]. Coding sequences and functional genes were annotated using RASTtk (Rapid Annotation using Subsystem Technology toolkit) [62]. Antibiotic resistance genes were predicted using ResFinder version 4.1 [63], CARD (Comprehensive Antibiotic Resistance Database) [64], and NDARO (National Database of Antibiotic Resistant Organisms) [65] databases. Pathogenicity islands (type 4 secretion system and type 6 secretion system) and prophages were predicted using VRprofile version 2.0 [66] and PHASTER (PHAge Search Tool Enhanced Release) [67], respectively. In plasmid analysis, trimmed reads were used for searching and assembling plasmid sequences using plasmidSPAdes version 3.12.0 [68]

in PATRIC version 3.6.9 server [59]. Quality control, annotation, and pathogenicity island predictions of plasmids were assessed using the same tools as with genomic analysis.

#### *3.5. MLST and Phylogenetic Analysis*

Identification of *A. baumannii* Aci46 was confirmed by rMLST (ribosomal multilocus sequence typing) [69] in PubMLST server [70]. The ST (sequence typing) of Aci46 was identified by MLST (multilocus sequence typing) according to the Pasture scheme (based on seven housekeeping genes *cpn60*, *gdhB*, *gltA*, *gpi*, *gyrB*, *recA*, and *rpoD*) [27] and Oxford scheme (based on seven housekeeping genes *cpn60*, *fusA*, *gltA*, *pyrG*, *recA*, *rplB*, and *rpoB*) [26] in PubMLST server [70]. The IC (international clonal) group of Aci46 was determined by phylogenetic relationship analysis using RAxML version 8.2.11 [71] based on 100 single-copy genes selected by PATRIC PGFams program [72] in PATRIC version 3.6.9 server [59]. The phylogenetic tree was visualized by FigTree [73]. Genomes of *A. baumannii* from IC I, IC II, IC III groups were used to identify the IC of Aci46. IC I group included *A. baumannii* strain AYE (CU459141.1) [74], AB0057 (CP001182.2) [75], and AB307- 0294 (CP001172) [75]. IC II group included *A. baumannii* strain AC30 (ALXD00000000) [2], ACICU (CP031380.1) [76], MDR-ZJ06 (CP001937.2) [77], and R14 (PUDB01000000) [78]. IC III included *A. baumannii* strain ATCC17978 (CP000521.1) [79] and SDF (CU468230.2) [74]. The outgroup strain was *A. baylyi* ADP1 (CR543861) [80].

## *3.6. Comparative Pangenome and Pairwise SNP Analysis*

Genomic data from five strains comprised of colistin and carbapenem-resistant Aci46, polymyxin B and carbapenem-resistant AC30 (ALXD00000000) [2], colistin and carbapenemresistant R14 (PUDB01000000) [78], wild-type or drug sensitive ATCC17978 (CP000521.1) [79], and colistin-sensitive and carbapenem-resistant ACICU (CP031380.1) [76] were compared using the Protein Family Sorter with PATRIC genus-specific families (PLfams) strategy in PATRIC server [59]. In pairwise SNP analysis, genome data of Aci46 was aligned with the ATCC17978 genome (CP000521.1) and ACICU (CP031380.1) using BWA-mem aligner [81] and SNP calling by FreeBayes [82]. The deduced amino acid sequences of *blaOXA-66* from five genomes were compared by Clustal Omega in EMBL-EBI [83].

#### **4. Conclusions**

In summary, this study reported the genome data of colistin and carbapenem-resistant *A. baumannii* Aci46, which was isolated from a patient in a Thai hospital. The MLST genotype of Aci46 is Pasture ST2 which is a predominant ST found in Thailand and Oxford ST1962 which has never been reported in Thailand. The predicted antibiotic resistance genes (for example, *blaOXA-23*, *blaOXA-66*, and *blaADC-25*) are on the chromosome, not plasmids. Based on pangenome analysis, we found that the CzcCBA cobalt/zinc/cadmium efflux RND transporter might be involved in conferring resistance to colistin and/or carbapenem. From SNP analysis, we identified three points of non-synonymous mutations in *pmrB* (412G > A, 599C > T, and 687A > C) that change amino acid sequences. These amino acid changes, specifically Glu229Asp, Pro200Leu, and Ala138Thr may confer colistin resistance in MDR *A. baumannii* strains.

**Supplementary Materials:** Supplementary data are available online at https://www.mdpi.com/ article/10.3390/antibiotics10091054/s1, Supplement Figure S1: The relative optical density at 570 nm for determination of minimum inhibitory concentration (MIC) by microdilution assay using MTT staining, Supplement Table S1: Guideline for interpretation of disk diffusion results, Supplement Table S2: Details of pathogenicity islands, Supplement Table S3: List of strain-specific genes, Supplement Table S4: List of multidrug resistant-specific genes (Aci46-ACICU-AC30-R14), Supplement Table S5: List of polymyxins and carbapenem-resistant-specific genes (Aci46-AC30-R14), Supplement Table S6: List of colistin and carbapenem-resistant-specific genes (Aci46-R14), Supplement Table S7: List of core genes (present in all genomes).

**Author Contributions:** N.T., S.C., S.S. and P.D. conceptualized the study; N.T. designed the research, tested antibiotic resistant profiling, analyzed genomic sequences, and wrote the paper; P.D. collected bacterial samples and identified *A. baumannii* Aci46. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research project was partially supported by Postdoctoral fellowship award from Mahidol University, grant number MU-PD\_2020\_9.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study, because the isolate used in this study was obtained from a collection of isolates that has already been published.

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in a previous study.

**Data Availability Statement:** The whole genome and plasmid sequences of *A. baumannii* Aci46 have been deposited at DDBJ/ENA/GenBank under the BioProject ID PRJNA739068.

**Acknowledgments:** The research project was partially supported by Postdoctoral fellowship award from Mahidol University, grant number MU-PD\_2020\_9. We thank James M. Dubbs for editing the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Role of** *dipA* **and** *pilD* **in** *Francisella tularensis* **Susceptibility to Resazurin**

**Kendall Souder <sup>1</sup> , Emma J. Beatty <sup>1</sup> , Siena C. McGovern <sup>1</sup> , Michael Whaby <sup>1</sup> , Emily Young <sup>1</sup> , Jacob Pancake <sup>1</sup> , Daron Weekley <sup>1</sup> , Justin Rice <sup>1</sup> , Donald A. Primerano <sup>2</sup> , James Denvir <sup>2</sup> , Joseph Horzempa <sup>1</sup> and Deanna M. Schmitt 1,\***


**Abstract:** The phenoxazine dye resazurin exhibits bactericidal activity against the Gram-negative pathogens *Francisella tularensis* and *Neisseria gonorrhoeae*. One resazurin derivative, resorufin pentyl ether, significantly reduces vaginal colonization by *Neisseria gonorrhoeae* in a mouse model of infection. The narrow spectrum of bacteria susceptible to resazurin and its derivatives suggests these compounds have a novel mode of action. To identify potential targets of resazurin and mechanisms of resistance, we isolated mutants of *F. tularensis* subsp. *holarctica* live vaccine strain (LVS) exhibiting reduced susceptibility to resazurin and performed whole genome sequencing. The genes *pilD* (FTL\_0959) and *dipA* (FTL\_1306) were mutated in half of the 46 resazurin-resistant (RZR) strains sequenced. Complementation of select RZR LVS isolates with wild-type *dipA* or *pilD* partially restored sensitivity to resazurin. To further characterize the role of *dipA* and *pilD* in resazurin susceptibility, a *dipA* deletion mutant, ∆*dipA*, and *pilD* disruption mutant, FTL\_0959d, were generated. Both mutants were less sensitive to killing by resazurin compared to wild-type LVS with phenotypes similar to the spontaneous resazurin-resistant mutants. This study identified a novel role for two genes *dipA* and *pilD* in *F. tularensis* susceptibility to resazurin.

**Keywords:** *Francisella tularensis*; resazurin; DipA; PilD; tularemia; antimicrobial; antibiotic; resistance

## **1. Introduction**

The Centers for Disease Control and Prevention (CDC) estimates that there are nearly three million new cases of antibiotic resistant bacterial infections annually, resulting in 35,000 deaths and billions of dollars in health care costs in the United States [1]. The rise in antibiotic resistance has been attributed to over-prescription and improper use of commonplace antibiotics [2]. Despite the clinical threat posed by increasing antibiotic resistance, the development of new antibiotics has significantly slowed due to decreased profitability [3]. While there are over 40 antimicrobial agents in clinical trials, most belong to existing classes of antibiotics such as beta lactams and beta-lactamase inhibitors, tetracyclines, and aminoglycosides [4]. Each of these classes of antibiotics targets similar bacterial processes including cell wall synthesis and protein synthesis. The development of bacterial resistance against these new drugs in the future is likely since there is already selective pressure from current antibiotics. Therefore, new antibiotic targets should be explored to minimize the emergence of resistance and provide effective alternatives.

Phenoxazine-based compounds have been shown to have antimicrobial activity [5,6]. Actinomycin D, the most well-known phenoxazine-containing antibiotic, suppresses the growth of various Gram-positive and Gram-negative bacteria as well as *Mycobacterium tu-*

**Citation:** Souder, K.; Beatty, E.J.; McGovern, S.C.; Whaby, M.; Young, E.; Pancake, J.; Weekley, D.; Rice, J.; Primerano, D.A.; Denvir, J.; et al. Role of *dipA* and *pilD* in *Francisella tularensis* Susceptibility to Resazurin. *Antibiotics* **2021**, *10*, 992. https:// doi.org/10.3390/antibiotics10080992

Academic Editor: Teresa V. Nogueira

Received: 22 July 2021 Accepted: 14 August 2021 Published: 17 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

*berculosis* [7]. Other actinomycins, neo-actinomycins A and B, exhibit moderate antibacterial activity against methicillin-resistant *Staphylococcus aureus* (MRSA) and vancomycinresistant *Enterococcus* (VRE) strains [8]. We previously showed the phenoxazine dye resazurin inhibits growth of the Gram-negative human pathogens *Neisseria gonorrhoeae* and *Francisella tularensis* [9,10]. Resorufin pentyl ether, an analog of resazurin, significantly reduces vaginal colonization by *N. gonorrhoeae* in a mouse model of infection [10]. However, the mechanism by which resazurin kills *F. tularensis* and *N. gonorrhoeae* has not been defined. Based on its structural similarity to other phenoxazine compounds, resazurin may function as a DNA intercalating agent [5]. Unlike other phenoxazine compounds, resazurin is non-toxic to eukaryotic cells, and its bactericidal activity is limited to only two Gram-negative bacterial species suggesting a novel mechanism of action [5,6,9].

In this study, we sought to identify *F. tularensis* genetic determinants involved in susceptibility to resazurin in hopes of elucidating the mode of action of this compound.

#### **2. Results**

#### *2.1. Isolation of Resazurin-Resistant F. tularensis LVS Mutants*

A common strategy bacteria employ to develop antimicrobial resistance is by mutating gene(s) associated with the mechanism of action of the antibiotic. Therefore, to identify potential targets of resazurin, we selected for mutants of *F. tularensis* LVS that were capable of growing in the presence of 22 µg/mL of resazurin. This concentration of resazurin is twenty-fold higher than the previously determined MIC of resazurin for *F. tularensis* LVS [9]. The genomes of forty-six spontaneous resazurin-resistant (RZR) mutants were sequenced and compared to wild-type LVS to identify genetic variants. Nonsynonymous mutations were identified in ten different protein-coding *F. tularensis* LVS genes with approximately 50% of the isolates possessing mutations in FTL\_1306 (*dipA*) and FTL\_0959 (*pilD*) (Table 1). The same *pilD* variant was observed in twenty-two of the RZR LVS isolates involving a single base pair substitution leading to a premature stop codon at position 187 of the protein sequence (Table 2). Fifteen different coding mutations were characterized in *dipA* that included single base pair substitutions, insertions, and deletions with many resulting in nonsense mutations (Table 2). Of note, one RZR LVS isolate, RZR46, contained a deletion of a single thymine residue at Position 2 in *dipA* resulting in loss of the start codon (Table 2). Western blot analysis confirmed this mutation abolished expression of DipA protein in RZR46 (data not shown). These data suggest the genes *dipA* and *pilD* play a role in *F. tularensis* LVS susceptibility to resazurin.


**Table 1.** Protein-coding genes containing nonsynonymous mutations in 46 resazurin-resistant *F. tularensis* isolates sequenced.


**Table 2.** Description of *dipA* and *pilD* mutations found in sequenced resazurin-resistant (Rzr) *F. tularensis* LVS isolates.

\* indicates stop codon.

#### *2.2. Characterizing the Role of pilD and dipA in F. tularensis Susceptibility to Resazurin*

Almost all of the RZR LVS strains isolated possessed nonsynonymous mutations in multiple genes with some exceptions. The sole nonsynonymous mutation in RZR47 is in *pilD* while RZR46 contains a single nucleotide deletion resulting in loss in expression of the DipA protein (Table 2; data not shown). These findings suggest that loss of either *pilD* or *dipA* function may confer resazurin resistance with limited contributions/effects from other genes. To confirm the mutations in *dipA* and/or *pilD* were contributing to the reduced susceptibility of RZR strains to resazurin, we individually cloned wild-type copies of *dipA* and *pilD* into the *Francisella* vector pABST which contains the robust *groE* promotor of *F. tularensis* [11] and then introduced these new plasmids, pDipA and pPilD, into RZR46 and RZR47, respectively. Then, we tested the sensitivity of these strains to resazurin. We hypothesized that restoring expression of DipA and/or PilD in these RZR strains would alter their resistance phenotype. RZR46 and RZR47 containing pABST alone served as controls. The resazurin MICs for RZR46/pABST and RZR47/pABST were two-fold higher than the MIC for wild-type LVS (Table 3). Introduction of the pABST vector into the RZR strains did not alter their susceptibility to resazurin (data not shown). Complementation of RZR46 with *dipA* restored resazurin sensitivity back to wild-type with both RZR46/pDipA and LVS having MICs of 5.5 µg/mL (Table 3). In contrast, RZR47 containing the *pilD* expression construct has the same resazurin MIC as the vector control (Table 3). These data confirm a role for *dipA*, but not *pilD*, in *F. tularensis* LVS susceptibility to resazurin.

**Table 3.** Resazurin MIC for *dipA* and *pilD* complemented *F. tularensis* resazurin-resistant strains.


<sup>1</sup> MICs are mode values from at least six independent determinations.

To individually investigate the role of *dipA* and *pilD* in resazurin susceptibility in a wild-type background, we generated a *pilD* disruption mutant (FTL\_0959d) and a *dipA*

deletion mutant (∆*dipA*) in *F. tularensis* LVS. The MICs of both FTL\_0959d and ∆*dipA* were two-fold higher than that of wild-type LVS and comparable to the MICs determined for the spontaneous resazurin-resistant mutants RZR46 and RZR47 (Tables 3 and 4). We next sought to evaluate the bactericidal activity of resazurin against FTL\_0959d and ∆*dipA* over time. Treatment with resazurin resulted in a significant reduction in viable bacteria after 24 h compared to untreated controls for all strains tested (Figure 1). However, significantly more FTL\_0959d and ∆*dipA* bacteria were recovered 24 h post resazurin treatment compared to wild-type LVS (Figure 1). These data suggest that neither DipA nor PilD are targets of resazurin, but might be involved in the uptake and metabolism of this compound by *F. tularensis* LVS. deletion mutant (ΔdipA) in *F. tularensis* LVS. The MICs of both FTL\_0959d and Δ*dipA* were two‐fold higher than that of wild‐type LVS and comparable to the MICs determined for the spontaneous resazurin‐resistant mutants RZR46 and RZR47 (Tables 3 and 4). We next sought to evaluate the bactericidal activity of resazurin against FTL\_0959d and Δ*dipA* over time. Treatment with resazurin resulted in a significant reduction in viable bacteria after 24 h compared to untreated controls for all strains tested (Figure 1). However, sig‐ nificantly more FTL\_0959d and Δ*dipA* bacteria were recovered 24 h post resazurin treat‐ ment compared to wild‐type LVS (Figure 1). These data suggest that neither DipA nor PilD are targets of resazurin, but might be involved in the uptake and metabolism of this compound by *F. tularensis* LVS.

To individually investigate the role of *dipA* and *pilD* in resazurin susceptibility in a wild‐type background*,* we generated a *pilD* disruption mutant (FTL\_0959d) and a *dipA*


**Table 4.** MIC of resazurin for *dipA* and *pilD F. tularensis* LVS mutants. **Table 4.** MIC of resazurin for *dipA* and *pilD F. tularensis* LVS mutants.

*Antibiotics* **2021**, *10*, x FOR PEER REVIEW 4 of 10

1MICs are mode values from at least six independent determinations.

<sup>1</sup> MICs are mode values from at least three independent determinations.

**Figure 1.** Reduced killing of FTL\_0959d and Δ*dipA* by resazurin compared to wild‐type LVS. Bacte‐ ria were cultivated in tryptic soy broth supplemented with 0.1% cysteine HCl (TSBc) in the presence or absence of resazurin (Rz, 1.375 μg/mL) for 24 h. Cultures were then diluted and plated to deter‐ mine the number of viable *F. tularensis* LVS bacteria 24 h post inoculation. Data shown are mean ± SEM from three individual experiments. The limit of detection was 100 CFU per ml. Statistically significant differences in growth post‐inoculation were determined by two‐way ANOVA followed by Tukey's multiple comparisons test (\*\*\*, *p* < 0.001 comparing +Rz to −Rz for each strain; \$\$\$, *p* < 0.001 comparing FTL\_0959d and ΔdipA to LVS 24 h. post Rz treatment). **Figure 1.** Reduced killing of FTL\_0959d and ∆*dipA* by resazurin compared to wild-type LVS. Bacteria were cultivated in tryptic soy broth supplemented with 0.1% cysteine HCl (TSBc) in the presence or absence of resazurin (Rz, 1.375 µg/mL) for 24 h. Cultures were then diluted and plated to determine the number of viable *F. tularensis* LVS bacteria 24 h post inoculation. Data shown are mean ± SEM from three individual experiments. The limit of detection was 100 CFU per ml. Statistically significant differences in growth post-inoculation were determined by two-way ANOVA followed by Tukey's multiple comparisons test (\*\*\*, *p* < 0.001 comparing +Rz to −Rz for each strain; \$\$\$, *p* < 0.001 comparing FTL\_0959d and ∆*dipA* to LVS 24 h. post Rz treatment).

#### **3. Discussion 3. Discussion**

The clinical threat posed by antibiotic resistant bacterial infections highlights the need to identify and apply novel antimicrobial therapies [2]. Resazurin exhibits antibac‐ terial activity against *F. tularensis* and *N. gonorrhoeae*, but its mode of action has not been defined [9,10]. Here, we screened for resazurin‐resistant mutants of *F. tularensis* LVS to identify genetic determinants of resazurin susceptibility. Whole genome sequencing of RZR isolates identified nonsynonymous mutations in ten different protein‐coding genes. None of the genes identified in this screen were previously described to play a role in *F.* The clinical threat posed by antibiotic resistant bacterial infections highlights the need to identify and apply novel antimicrobial therapies [2]. Resazurin exhibits antibacterial activity against *F. tularensis* and *N. gonorrhoeae*, but its mode of action has not been defined [9,10]. Here, we screened for resazurin-resistant mutants of *F. tularensis* LVS to identify genetic determinants of resazurin susceptibility. Whole genome sequencing of RZR isolates identified nonsynonymous mutations in ten different protein-coding genes. None of the genes identified in this screen were previously described to play a role in *F. tularensis* resistance to other clinically relevant antibiotics such as aminoglycosides, tetracyclines and fluoroquinolones [12,13]. Therefore, these data suggest resazurin has a unique mechanism of action.

In particular, two genes, *dipA* and *pilD,* were mutated in the majority of the RZR LVS clones isolated and sequenced. Complementation of RZR46 with *dipA* restored sensitivity to resazurin comparable to wild-type LVS while expression of *pilD* in RZR47 did not.

However, both a laboratory generated *pilD* disruption mutant and *dipA* deletion mutant had higher MICs compared to wild-type LVS similar to the spontaneous RZR mutants. The failure to restore resazurin sensitivity in RZR47 upon complementation with wildtype *pilD* may be due to a dominant negative effect of the original *pilD* mutation or another intergenic/undetected variant in RZR47 may be contributing to the resazurin resistance. Regardless, less killing of FTL\_0959d and ∆*dipA* was observed over 24 h following resazurin treatment compared to wild-type LVS confirming a role for both *dipA* and *pilD* in resazurin susceptibility.

DipA is a 353 amino acid protein predicted to form a surface complex with other outer membrane proteins to facilitate translocation of virulence factors that interact with host cells [14]. Most of the *dipA* mutations characterized in this study occurred within the Sel1-like repeat domains (amino acids 95–170 and 192–263) which are known to be required for the biological function of DipA [14]. These structural domains provide a scaffold for protein–protein interactions [15]. DipA has been shown to associate with FopA, a *F. tularensis* outer membrane protein with noted homology to the OmpA porin [14]. OmpA has been shown to play a role in antibiotic resistance in other Gram-negative pathogens including *Escherichia coli* and *Acinetobacter baumannii* [16,17]. Given the correlation between DipA mutations and resazurin resistance in this study, it is possible that DipA plays a role in the stability or function of FopA allowing for uptake of resazurin. Without functional DipA, less resazurin may be taken up by *F. tularensis* LVS resulting in increased resistance to resazurin. Interestingly, none of the resazurin resistant strains isolated and sequenced in this study contained mutations in the LVS gene encoding FopA. The role of DipA and FopA in uptake of resazurin by *F. tularensis* is currently being investigated.

The PilD protein was also shown to contribute to *F. tularensis* susceptibility to resazurin. PilD is an inner membrane peptidase responsible for processing major and minor (pseudo) pilins involved in the formation of Type IV pili and assembly of a Type II protein secretion system [18,19]. In *P. aeruginosa*, PilD has also been shown to play a role in the export of select enzymes including alkaline phosphatase, phospholipase C, elastase, and exotoxin A [20]. Although the exact function of PilD in *F. tularensis* has not been fully investigated and defined, it is possible PilD may play a role in the processing and/or export of a protein that is a target of resazurin. Mutation of *pilD* would thus affect processing or export of the antibiotic target rendering *F. tularensis* less susceptible to resazurin. The exact role PilD plays in *F. tularensis* susceptibility to resazurin is under further investigation.

As *F. tularensis* is not the only bacterium susceptible to resazurin, we wondered whether *pilD* (FTL\_0959) and *dipA* (FTL\_1306) homologs were present in *N. gonorrhoeae* and played similar roles as described here. A BLAST search [21] against the National Center for Biotechnology Information database of non-redundant protein sequences identified an A24 family peptidase (WP\_003689814.1) and a SEL1-like repeat protein (MBS5742101.1) in *Neisseria* that were only 43% and 26% identical to *F. tularensis* PilD and DipA, respectively. The low homology between these proteins suggests that PilD and DipA have unique roles in *F. tularensis* susceptibility to resazurin. It also remains unknown whether DipA and PilD function in the same pathway to mediate *F. tularensis* killing by resazurin. The proposed function of DipA in transport and PilD in pilin processing suggests that they function separately from one another, but the relationship between DipA and PilD in mediating susceptibility to resazurin is currently being explored. Overall, mutations in *pilD* and *dipA* failed to confer complete resistance to resazurin. The growth of both FTL\_0959d and ∆*dipA* was still inhibited by resazurin and the MICs for these mutants were only two-fold higher than wild-type LVS. An alternative method to identify potential antibiotic targets from the one used in this study is high-throughput transposon sequencing (Tn-Seq) [22–24]. In this approach, a library of LVS transposon mutants would be generated and screened for resazurin resistance [25]. Then, next-generation sequencing of the transposon insertion would be performed and the insertion sites would be mapped to the LVS genome. Further understanding of resazurin's mode of action and resistance mechanisms will help guide the

development of resazurin derivatives with more potent activity as well as new therapeutic strategies to combat *Francisella* infections.

#### **4. Materials and Methods**

#### *4.1. Bacterial Strains and Growth Conditions*

Bacterial strains used in this study are listed in Table 5. For cultivation of *F. tularensis* LVS strains, frozen stock cultures were plated on chocolate II agar and incubated at 37 ◦C with 5% CO<sup>2</sup> for 2–4 days. These bacteria were then used to inoculate TSBc (trypticase soy broth (BD Biosciences) containing 0.1% L-cysteine hydrochloride monohydrate (Fisher)). *Escherichia coli* 5-α (New England Biolabs) bacteria cultivated on LB agar were used to inoculate LB broth. All broth cultures were incubated at 37 ◦C with agitation. When required, kanamycin (35 µg/mL for *E. coli*; 10 µg/mL for *F. tularensis*) and hygromycin (250 µg/mL) were supplemented into the media.

**Table 5.** Description of strains, plasmids, and primers used in this study.


<sup>1</sup> NEB—New England Biotechnologies. <sup>2</sup> IDT—Integrated DNA Technologies.

#### *4.2. Selection of Resazurin-Resistant (RZR) LVS Mutants and Whole Genome Sequencing*

Mutants resistant to resazurin were obtained by plating suspensions containing approximately 5 <sup>×</sup> <sup>10</sup><sup>8</sup> CFU of *F. tularensis* LVS bacteria on chocolate II agar supplemented with 11 µg/mL resazurin sodium salt (Acros Organics, dissolved in water) [9]. This concentration of resazurin is ten-fold higher than the previously determined minimal inhibitory concentration (MIC) of resazurin [9]. Colonies that formed on plates containing 11 µg/mL resazurin were then replica-plated onto 22 µg/mL resazurin. Forty-six resazurin-resistant (RZR) LVS clones that grew in the presence of 22 µg/mL resazurin were selected for further

analysis. Genomic DNA was isolated using the Bacterial Genomic DNA Isolation Kit (Norgen Biotek Corp., Thorold, ON, Canada) per the manufacturer's instructions. Sequencing libraries were prepared using Nextera XT DNA Library Preparation Kits (Illumina, San Diego, CA, USA) and individual libraries were indexed with barcodes using Nextera XT v2 Index Kits (Illumina, San Diego, CA, USA) in the Marshall University (MU) Genomics Core Facility. Library quality and size distribution was assessed on an Agilent Bioanalyzer equipped with High Sensitivity DNA Chips. Libraries were quantitated by Qubit fluorimetry. High throughput sequencing (2 × 100 bp paired-end) was performed on an Illumina HiSeq1500 in Rapid Run mode in the Genomics Core. This approach allowed for approximately 150× coverage of each individual bacterial genome.

Sequencing reads were trimmed to remove Illumina adapters and low quality base calls using Trimmomatic version 0.36 [27]. Trimmed reads were aligned to the *Francisella tularensis* reference genome (NCBI accession number AM23362.1) using BWA version 0.7.12 [28] and sorted and indexed using picard tools version 2.9.4. Variant calling was performed following the GATK version 3.8.0 pipeline [29]. Briefly, duplicates were marked with the MarkDuplicates tool and per-sample variants identified using HaplotypeCaller. Variants were then called across the entire sample set using the GenotypeGVCFs tool with the sample-ploidy option set to one. Variant calls with quality score below 20 were filtered out using the VariantFiltration tool. Variants were exported to tab-delimited format using VariantsToTable. After the conclusion of these GATK pipeline steps, variants were annotated to determine their effects on the protein sequence using snpEff version 4.3q [30].

#### *4.3. Construction of dipA and pilD Complemented Strains and Mutants*

Primers and plasmids used in this study are listed in Table 5. To complement RZR LVS mutants with either wild-type *dipA* or *pilD*, *F. tularensis* LVS chromosomal DNA extracted from stationary-phase broth cultures using the Bacterial Genomic DNA Isolation Kit (Norgen Biotek Corp.) served as a template for amplification of *dipA* (FTL\_1306) and *pilD* (FTL\_0959). Primer pairs dipA\_F/dipA\_R and pilD\_F/pilD\_R were used to PCRamplify *dipA* and *pilD*, respectively. The dipA amplicon was cloned into the EcoRI-NheI site of pABST [24] to generate the plasmid, pDipA. The pilD amplicon was cloned into the EcoRI-NdeI site of pABST to create the plasmid, pPilD. Both pDipA and pPilD were introduced into select *F. tularensis* LVS resazurin-resistant mutant strains by electroporation and selection on chocolate II agar supplemented with 10 µg/mL kanamycin.

The *pilD* disruption mutant (FTL\_0959d) was generated using an unstable, integrating suicide vector, pJH1, as previously reported [31,32]. An internal 470 base pair region of FTL\_0959 (*pilD*) was amplified by PCR using the primers F\_0959d and R\_0959d. This amplicon was digested with BamHI and SphI and then cloned into pJH1 that had been digested with these same enzymes, which produced pJH1-0959d. This disruption construct was mobilized into *F. tularensis* LVS by triparental mating [31]. Hygromycin-resistant colonies were screened by PCR using the primers pJH1-conf and 1F\_pilD which are specific for the vector-portion of pJH1-FTL\_0959 and pilD respectively, to confirm disruption of the *pilD* gene.

A *F. tularensis* LVS *dipA* (FTL\_1306) deletion mutant was generated using pJH1 and the I-SceI endonuclease as described previously [26]. To begin, the *dipA* deletion construct pJH1-∆dipA was generated using splicing by overlap-extension PCR. Regions (~500 bp) upstream and downstream of the *dipA* sequence targeted for deletion were amplified by PCR using the primer pairs 1F\_dipA with 2R\_dipA and 3F\_dipA with 4R\_dipA, respectively. The resulting amplicons contained regions of overlap and served as template DNA for a second PCR reaction using primers 1F\_dipA and 4R\_dipA. The resulting 1176 bp fragment was digested with PstI and SphI and then ligated into pJH1, which had been digested with the same enzymes, to produce pJH1-∆dipA. This vector was then transferred into *F. tularensis* LVS by tri-parental mating [31]. Merodiploid strains were recovered and transformed with pGUTS by electroporation to allow for expression of I-SceI which causes a double-stranded break forcing recombination and allelic replacement [26]. Colonies resis-

tant to kanamycin were screened by PCR for deletion of *dipA* using primers 1F\_dipA and 4R\_dipA. To cure pGUTS, the *F. tularensis* ∆*dipA* strains were repeatedly sub-cultured in TSBc, diluted, and plated to a density of 100 to 300 colony forming units CFU per chocolate II agar plate. Plates were incubated for at least 3 days at 37 ◦C, 5% CO<sup>2</sup> and colonies that formed were replica plated onto chocolate II agar supplemented with and without 10 µg/mL kanamycin to select against colonies harboring pGUTS. Those colonies sensitive to kanamycin were isolated and again tested for sensitivity to this antibiotic. The resulting LVS strain was named ∆*dipA*. Western blot analysis using an anti-DipA antibody [14] confirmed loss in expression of the DipA protein in LVS ∆*dipA* (data not shown).

#### *4.4. Agar Dilution Susceptibility Testing*

Minimum inhibitory concentrations (MICs) were determined by the agar dilution method using chocolate II agar according to CLSI guidelines [33]. Briefly, *F. tularensis* LVS bacteria were cultivated overnight in TSBc, diluted to an OD600 of 0.3 (approximately <sup>5</sup> <sup>×</sup> <sup>10</sup><sup>8</sup> CFU/mL), and then diluted to 1 <sup>×</sup> <sup>10</sup><sup>6</sup> CFU/mL. Ten microliters of this suspension (1 <sup>×</sup> <sup>10</sup><sup>4</sup> CFU) were plated onto chocolate II agar containing a series of two-fold dilutions of resazurin. Following incubation (37 ◦C with 5% CO<sup>2</sup> for 48–72 h), the MIC reported for each *F. tularensis* strain was the lowest concentration of resazurin that completely inhibited the growth of bacteria on the agar plate. No strain differed by more than one dilution in 3–5 tests.

#### *4.5. Time Kill Assays*

*F. tularensis* LVS overnight broth cultures were used to inoculate TSBc containing 1.375 µg/mL resazurin. Immediately following inoculation and 24 h later, cultures were serially diluted and plated onto chocolate II agar. Plates were incubated at 37 ◦C, 5% CO<sup>2</sup> and individual colonies were enumerated. The limit of detection was 100 CFU/mL.

#### *4.6. Statistical Analyses*

Data were analyzed for significant differences using GraphPad Prism software (Graph-Pad Software Inc., La Jolla, CA, USA). The statistical tests are indicated in the figure legend.

**Author Contributions:** Conceptualization, D.M.S.; methodology, J.H., D.A.P., J.D. and D.M.S.; resources, K.S., E.J.B., S.C.M., M.W., E.Y., J.P., D.W., J.H., D.A.P., J.D. and D.M.S.; data curation, D.A.P., J.D. and D.M.S.; writing—original draft preparation, K.S. and D.M.S.; writing—review and editing, K.S., E.J.B., S.C.M., M.W., E.Y., J.P., D.W., J.R., J.H., D.A.P., J.D. and D.M.S.; visualization, K.S., E.J.B., S.C.M., M.W., J.P., D.W., J.R. and D.M.S.; supervision, D.M.S.; project administration, D.M.S.; funding acquisition, D.A.P., J.H. and D.M.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by a pilot grant from the West Virginia IDeA Network of Biomedical Research Excellence (WV-INBRE) program which is supported by a grant from the National Institute of General Medical Sciences of the National Institutes of Health (P20GM103434). Whole genome sequencing was carried out by the Marshall University Genomics Core Facility, which is supported by the WV-INBRE grant (P20GM103434), the COBRE ACCORD grant (1P20GM121299) and the West Virginia Clinical and Translational Science Institute (WV-CTSI) grant (2U54GM104942). The Research Resource ID (RRID) citation for the MU Genomics Core is RRID:SCR\_018885. In addition, a portion of this work was funded by the National Heart Lung and Blood Institute of the National Institutes of Health (1R15HL147135).

**Data Availability Statement:** All sequencing data have been submitted to the Sequence Read Archive at the National Center for Biotechnology Information (NCBI) and are accessible via project number PRJNA748943.

**Acknowledgments:** The authors thank Karen Elkins for providing LVS and Jean Celli for the rat anti-FTT0369c antibody to detect expression of DipA.

**Conflicts of Interest:** The authors acknowledge that Joseph Horzempa has been awarded a patent (US20150258103A1) on the antimicrobial activity of the compounds used in this manuscript. However, the authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

## **References**

