Previously, we reported on the molecular epidemiology and drug resistant features of A. baumannii isolates collected from hospitals in Pakistan during 2013–2015. No colistin resistance was found among those isolates (Karah et al., 2020). In the follow up, here we report the current status of colistin resistance in isolates collected from five hospitals in Pakistan during 2019–2020. For this study, a collection of 150 clinical isolates was obtained. Out of the 150 clinical isolates, 28 (19.4%) were from the respiratory tract infections, 56 (38.8%) from the blood or central venous catheter, 23 (16.3%) from the wound infections and 36 (25%) from the urinary tract infections. The isolates showed resistance rates of 100% (150/150) to ampicillin/sulbactam, piperacillin/tazobactam, cefepime, cefotaxime, ceftazidime, ceftriaxone, gentamicin, tobramycin, ciprofloxacin, levofloxacin, and trimethoprim/sulfamethoxazole. In addition, 84% (126/150) of the isolates were resistant to imipenem, 9,3% (14/150) to minocycline and doxycycline, and 7,3% (11/150) to colistin and polymyxin B (Table 1). Except for the findings on colistin resistance, data were in line with previous studies describing extensive occurrence of multidrug-resistant strains of A. baumannii in Pakistan (Begum et al., 2013; Khalid et al., 2020). More alarmingly, there were two isolates resistant to all tested 2^nd line and 3^rd line antibiotics including colistin, moxifloxacin and doxycycline. Both of these isolates (2204 and 2208) were obtained from tracheal secretion of persons admitted to two different hospitals in 2020 and belong to OXA-66 type of strains with OXA-51 beta lactamase (Table 2). The MIC of colistin varied from strain to strain in colistin resistant isolate with a range between 8 ug/ml to 32 ug/ml. Nine out of eleven colistin resistant isolates were hetero resistant to all tested antibiotics except tetracycline drugs doxycycline and minocycline whereas the remaining two isolates were resistant to all tested antibiotics.

In order to monitor the potential effect of colistin resistance on the fitness and virulence of A. baumannii, a colistin resistant derivative of the well characterized colistin sensitive strain A. baumannii17978 was obtained by challenging it with sub optimal concentration of 0,25 ug/ml colistin by passing the bacteria through three consecutive overnight subcultures. The resulting colistin resistant variant of A. baumannii 17978, hereafter referred to as CR-1, exhibited a MIC value of 16 ug/ml for colistin. Whole genome sequencing of the isolate CR-1 was performed to assess adaptive response to colistin at genetic level. Mutations in one of the genes involved in encoding LPS biosynthetic machinery lpxA, lpxC or lpxD or mutations in pmrAB operon involved in the addition of phospho ethanolamine on LPS have been known to cause colistin resistance. However, no mutation was detected in these genes in the CR-1 isolate (Whole genome sequence Accession no. JANIHK000000000). Hence, the colistin resistance acquired by CR-1 appeared independent of mutations in these genes. In depth analysis of the whole genome sequence of CR-1 identified four single base pair mutations in loci AIS_1585, AIS_2445, AIS_3424 and AIS_2454 in the genome of CR-1 (Table 3). None of these point mutations have been shown to be associated with colistin resistance previously. Interestingly, the transposons mutants of these four genes along with several other mutants were shown to be associated with a salt induced colistin tolerance phenotype (Hood et al., 2013) suggesting the relevance of these single nucleotide mutations in colistin resistance acquired by CR-1. These mutations appear to be associated with colistin resistance phenotype. However to find role of these mutations in colistin resistance further investigations are required.

The CR-1 isolate was then tested for LPS biogenesis. Silver staining of LPS preparations resolved in acrylamide gel identified that CR-1 lacks a distinct band of lipopolysaccharide rather than an overall decrease in LPS content (Figure 1A). It is known that A. baumannii acquires colistin resistance at the cost of virulence loss due to mutations in lpxA, lpxC, lpxD or pmrAB (Da Silva and Domingues, 2017). However, these genes were intact in CR-1. Therefore, the CR-1 isolate was then tested for virulence phenotypes and capability to induce a pre-inflammatory immune response in epithelial cells. Lung epithelial cell line A549 and rectal carcinoma epithelial cells SW480 were infected with CR-1 and wild type A. baumannii 17978. The adherence capability of CR-1 with both cell lines was significantly decreased (Figures 1B,C). Consistently, the pro-inflammatory response induced by CR-1 in SW480 cells, measured as IL-6 and IL-8 levels, was significantly decreased in case of the CR-1 isolate as compare to wild type (Figures 1D,E). Since colistin resistance in CR-1 was acquired under laboratory conditions, to validate these findings, naturally colistin resistant clinical isolate AB-4 was also tested for interaction with epithelial cells in comparison to colistin sensitive hyper virulent strain A. baumannii AB5075. The cell adherence and cytokine induction capabilities of AB-4 were significantly decreased as compare to A. baumannii AB5075 (Figures 1B–D). The identity and molecular typing of AB-4 was determined by whole genome sequencing. The multi locus sequence (MLS) typing shows that the isolate belongs to international clone II and MLS type 2 (Supplementary Table S2). AB-4 is found to be equipped with beta lactamases OXA-66 and blaADC25 of A. baumannii. The broad range beta lactamase encoding genes bla-PER-1 and blaOXA-23 were also detected in the genome of AB-4. In addition, streptogramin B and macrolide resistance genes msr(E) and mph (E) were detected. The isolate was found to harbour aminoglycoside resistance genes aph (3”)-Ib, aph (3”)-VIb, aph (3”)-Id, aph (3”)-Via, aph (3”)-Ia and armA and sulfonamide resistance gene sul-2 (Supplementary Table S3).

These in vitro findings triggered us to test CR-1 for virulence potential and inflammatory response associated with colistin resistance in a mouse infection model. For that, BALB/c mice were intraperitoneally infected with wild type A. baumannii 17978 and its colistin resistant mutant strain CR-1. The hyper virulent strain AB5075 was used as a positive control to establish virulence phenotypes. The mice infected with A. baumannii AB5075 and A. baumannii 17978 exhibited a strong inflammatory response with respectively 4-fold and 2-fold increase in neutrophil count in the blood after 24 h of the post infection (Figure 2A). The colistin resistant mutant strain CR-1 also induced an IL-8 and neutrophil increase upon intraperitoneal infection. However, there was a significant decrease in the number of colonized bacteria into lungs and liver of the mice infected with colistin resistant isolates as compare to colistin sensitive isolates (Figures 2B,C). There was no significant difference detected in the colonization of CR-1 bacteria into spleen (Figure 2D). Altogether these findings suggest that the colistin resistant strain CR-1, although it harbours intact lpxA ⁺ , lpxC ⁺ , lpxD ⁺ and pmrA ⁺ B ⁺ loci although it exhibited a significant reduced virulence in the mouse model.

Nine out of 11 colistin resistant isolates were detected susceptible to tetracycline drugs, doxycycline and minocycline (Table 2). In order to determine potency of doxycycline against colistin resistant isolates, time dependent bactericidal and bacteriostatic effects of doxycycline in the presence and absence of colistin were investigated in a representative isolate. The colistin resistant but doxycycline susceptible strain AB-4 was selected as a representative strain for this purpose.

In time-kill curve studies, doxycycline showed a 4-log₁₀ reduction in CFU/ml at 4xMIC concentration in comparison to growth control after 2 h of incubation and 3-log₁₀ reduction at 2xMIC after 18 h of incubation against AB-04 and AB17978 (Figures 3A,B). There were no viable bacteria detected with the treatment of 4xMIC after 8 h of incubation. However, there was 4-log₁₀ reduction of CFU with 2xMIC treatment and 2-log₁₀ reduction with MIC with a ½ MIC treatment of doxycycline.

The MIC for colistin of AB-4 was 16 μg/ml. In the time-kill curve studies, colistin showed a 2-log₁₀ reduction in CFU/ml at 4xMIC concentration in comparison to growth control after 2 h of incubation and a 1-log₁₀ reduction at 2xMIC after 18 h of incubation (Figure 3C). Colistin showed bactericidal effect towards strain AB-4 at a concentration twice of MIC at second hour of incubation. A bactericidal effect was also noted at the second hour of incubation at a concentration of four times of their MICs whereas no bactericidal effect was found at their MIC during 24 h of incubation.

Subsequently, the time kill assay was performed to assess the effect of colistin in the presence of doxycycline with strain AB-4 at ½ MICs of colistin and doxycycline (Figure 3D). No bactericidal effect was observed with ½ MICs of doxycycline or colistin alone in first hour of incubation. However, treatment with both doxycycline and colistin at ½ MICs each significantly reduced the CFU, 3-5 log₁₀ at all tested time points. This finding suggests that a synergetic effect of antimicrobial activity of colistin and doxycycline exists in colistin resistant isolate AB-4.

Considering very limited, treatment options against colistin resistant isolates (Table 1), we were interested to explore other alternative treatment strategies. In this regard, extracts of the traditional medicinal plant Saussurea lappa have been shown to possess antimicrobial activity against Gram-negative and Gram-positive and anti-inflammatory activity in tracheal inflammation (Hasson et al., 2013; Rao Vadaparthi et al., 2015; Pyun et al., 2018). In vitro antimicrobial activity of various fractions of the herb Saussurea lappa was tested against colistin resistant A. baumannii isolates. However, a remarkable antimicrobial activity of S. lappa against these isolates was not observed in vitro at the concentration of less than up to the concentration of 10 mg/ml. In order to assess its anti-inflammatory role, a water extract of S. lappa was tested in the mouse model of systemic infection against the extreme drug resistant isolate A. bauamnnii AB-4 upon intraperitoneal infection and with the hyper virulent strain A. baumannii AB5075. The mice were treated with the water-soluble fraction from S. lappa. For that, a dose consisting of 10 mg of S. lappa powder was orally administrated to each mouse after 3 h of intraperitoneal infection. The inflammation induced by A. baumannii strains was measured as IL-8 levels and total neutrophil count in the blood of infected mice after 20 h of infection. There was a statistically significant reduction in IL-8 levels and total neutrophil count in the blood of the infected mice treated with S. lappa or tetracycline as compare to uninfected group (Figure 4A). This finding suggests that the compound from S. lappa modulates the inflammatory response induced by A. baumannii infection. Also, there was a significant decrease in numbers of viable bacteria colonized into liver and lungs of infected mice treated with S. lappa extract or tetracycline as compare to the untreated group (Figures 4B–D). Altogether, these findings suggest that the water-soluble fraction of S. lappa possess some immunomodulatory compound that may be used to treat A. baumannii infection in mice.

A. baumannii AB5075 displays a recognized capability of switching into an avirulent subpopulation of bacterial cells that appears as translucent colonies on agar plates (Figure 5A) (Tipton et al., 2015; Chin et al., 2018; Ahmad et al., 2019). The findings of the immunomodulatory role of Saussurea lappa extracts towards A. baumanii infection inspired us to investigate whether such effects would be limited to a certain subpopulation of A. baumannii AB5075. The effect of the S. lappa extract was therefore tested upon infecting mice with opaque and translucent colony variants of A. baumannii AB5075. As reported previously, the opaque variant was more virulent as compare to translucent counterpart in the mouse. The administration of S. lappa extract to mice groups infected with opaque and translucent variants exerted similar effect in lowering cytokine levels, white blood cell count neutrophils and colonization of bacteria into organs (Figures 5B–E).

Clinical isolates of A. baumannii (n = 150) were collected from clinical samples of patients visited or hospitalized in Jinnah hospital, Ittefaq Hospital Lahore, Services Institute of Medical Sciences, King Edward Medical University Laboratory and Mayo Hospital Lahore during 2019–2020. The strains were identified by their morphological and biochemical characteristics using API 20-NE (Bio Merieux, France) following manufacturer`s instructions. A. baumanniis 17978 was used for isolation of a colistin resistant derivative CR-1. The hyper virulent strain AB5075 and its opaque and translucent colony variants were used for infection tests (Ahmad et al., 2019). Escherichia coli ATCC 25922 was used as a control strain in MIC value tests. Salmonella typhimurium 14028 was used as positive control to induce proinflammatory response from epithelial cells. The strains were stored at minus 80°C in 10% glycerol prepared in LB broth.

Antimicrobial susceptibility testing was performed by the standard Kirby-Bauer disk diffusion method using cation adjusted Mueller-Hinton agar (MHA) (Oxoid, United Kingdom), according to Clinical Laboratory Standards Institute (CLSI) guidelines (Clinical and Laboratory Standards Institute (CLSI), 2019) Antibiotic discs (Oxoid, United Kingdom) used were piperacillin (100 µg), ampicillin-sulbactam (10 µg/10 µg), piperacillin-tazobactam (100 µg/10 µg), ticarcillin-clavulanic acid (75µ/10 µg), ceftazidime (30 µg), ceftriaxone (30 µg), cefepime (30 µg), imipenem (10 µg), gentamicin (10 µg), doxycycline (30 µg) and ciprofloxacin (5 µg).

The minimal inhibitory concentration of colistin was determined by micro broth dilution method. The base material colistin (Glaxosmith Kline pharmaceuticals) was prepared in water and tetracycline in 70% ethanol and stored at minus 20°C. MICs of all strains for each antibiotic were determined by standard microbroth dilution method. Bacterial inoculum equivalent to 0.5 McFarland (5 × 10⁸) was prepared and diluted 1:10 to achieve the final inoculum of 5 × 10⁷ CFU/ml. Concentration range of the antibiotics to be tested was; 0.125–256 µg in MHA using 96 well plates. The plates were incubated for 24 h at 37°C and the lowest concentration at which bacterial growth was completely inhibited was noted and declared as MIC. Escherichia coli ATCC 25922 was used as a control strain. MIC results were read and interpreted according to the CLSI 2019 breakpoints for A. baumannii.

The colistin resistant isolates were typed using single-locus molecular schemes based on the allelic identity of the A. baumannii-intrinsic bla _OXA-51-like gene (Pournaras et al., 2014). PCR amplification of bla _OXA-51-like was performed using in-house designed primers (Supplementary Table S1) and followed by Sanger sequencing of the amplicons. The thermal cycling program used for PCR consisted of: initial denaturation; 94°C for 5 min: 30 cycles of amplification: denaturation at 95°C for 20 s; annealing at 56°C for 20 s, and extension at 72°C for 1 min and 30 s: Final extension; 72°C for 8 min. This approach was able to detect the occurrence of insertion sequence (IS) elements, such as ISAba1, in the bordering regions of bla _OXA-51-like.

The whole genome sequencing of colistin resistant clinical isolate AB-4 and colistin resistant variant of A. baumannii 17978 CR-1 was performed using the MiSeq Desktop Sequencer and MiSeq Reagent Kit v3 (Illumina, San Diego, CA, United States). DNA preparation, library construction, and genome sequencing were done according to the manufacturer’s instructions. Sequence data were assembled and analyzed using the CLC genomics workbench (v7.0.4; CLC bio, Aarhus, Denmark). The whole genome sequence of A. baumannii ATCC 17978 (DDBJ/EMBL/GenBank database accession number CP000521) was used as a reference sequence.

The MLST web-based search engine, hosted by the Center for Genomic Epidemiology in Denmark (http://www.genomicepidemiology.org/), was used to assign the isolate into STs according to the Institute Pasteur’s MLST scheme (http://www.pasteur.fr/mlst) (Larsen et al., 2012). The occurrence of acquired antimicrobial resistance genes was detected using the Res Finder service, also hosted by the Center for Genomic Epidemiology in Denmark [24]. The occurrence of resistance genes was verified, and genetic surroundings were annotated based on the yields of nucleotide similarities obtained using the Basic Local Alignment Search Tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi) against the “Nucleotide collection (nr/nt)” and/or “Whole-genome shotgun contigs (wgs)” databases (Zhang et al., 2000).

Draft genome sequences of the isolates were deposited in the DDBJ/EMBL/GenBank database under the BioProject accession number: PRJNA862891. The isolates described in this paper are AB-4 with accession number JANIEU000000000 and CR-1 with accession number JANIHK000000000.

The colistin resistant strain A. baumannii AB-4 and colistin sensitive strain A. baumannii 17978 were selected for time kill assay. Bacterial suspensions equivalent to 1.0 MF (3 × 10⁸ CFU/ml) was prepared for each isolate to be tested by inoculating four to 5 colonies into 5 ml of CAMHB and incubating for 6 h. One ml of 1.0 MF suspension was diluted in 4 ml of sterile saline to achieve 6 × 10⁷ CFU/ml (1:5 dilution). The same was further diluted (1:100) by dispensing 0.1 ml of 1:5 diluted suspension in 10 ml of saline. The colony count was 6 × 10⁵ CFU/ml. This was used as a starting inoculum in this method.

MICs were determined by micro broth dilution method found to be 0,125 ug/ml in AB17978 and 0,5 ug/ml in AB-4. The following concentrations were tested along with growth and sterility control tubes; MIC, two times the MIC, four times the MIC, and one half the MIC.

The antibiotic agent was dispensed in a sterile screw capped glass tube containing 9.9 ml of CAMHB as 0.1 ml of 100-fold concentrated antibiotic agent. The tubes were incubated at 37°C, and test samples were taken at 0, 2, 4, 6, 18, 24 h after inoculation, for colony count. Serial dilution of contents were prepared and 0.1 ml aliquot of each dilution was plated on Tryptic Soya agar and incubated overnight at 37°C. Antibiotic concentration was determined that resulted in 3-log₁₀ reduction in CFU/ml as compared to the growth control curve. The time was determined that was required to achieve 3-log₁₀ reduction.

Lung epithelial cell line A549 and human rectal carcinoma epithelial cell line SW480 were tested for interaction with A. baumannii. Cells were cultured to confluence in 24-well cell culture treated plates in RPMI 1640 cell culture medium supplemented with fetal bovine serum at 37°C with a continuous supply of 5% CO₂. After a change of the medium, confluent layers cells grown in wells of 24 well plates were infected with respective bacterial strains at MOI of 100. For cell infection experiments, bacterial strains were grown to the logarithmic growth phase in LB broth at 37°C. Upon infection, the plate was incubated at 37°C with 5% CO₂ for 4 h. The eukaryotic cell monolayer was washed three times with PBS and subsequently to be treated with 1 ml of 5% trypsin. After dissociation of cells attached to the surface, harvested sample was diluted in PBS. The 2^nd dilution of every well was processed for enumeration of CFU by point inoculation following overnight incubation on LB agar plates. The CFU numbers of each strain were recorded. Bacterial adherence was determined as the percentage of bacterial cells associated with eukaryotic cells relative to the original inoculums. Each experiment was performed as five biological replicates and three technical replicates.

A-549 and SW-480 cells were cultured in 24-well plates in RPMI-1640 medium supplemented with 5% fetal bovine serum (FBS) at 37°C with a continuous supply of 5% CO₂. After a change of the medium, confluent layers cells were infected with respective bacterial strains at MOI of 100. For cell infection experiments, bacterial strains were grown into the logarithmic growth phase in LB broth at 37°C. Then the bacteria were harvested by centrifugation at 5000 rpm for 10 min and resuspended in RPMI. After infection, cells were incubated at 37°C with 5% CO₂ for 4 h. The suspension was, centrifuged after 4 h. The supernatant was analysed for production of IL-8 and IL-6 using ELISA according to manufacturer’s instructions (Bio Assay Technology Laboratory). Salmonella typhimurium 14028 was used a positive control to induce IL-8 from epithelial cells.

The root of S. lappa weighing 200 g was washed with water and kept in incubator at 37°C until get dried. Upon drying, roots were crushed into powder form and dissolved in autoclaved distilled water. The mixture was incubated overnight at room temperature followed by gentle centrifugation. The supernatant was collected in sterile tubes. The contents were dried in rotary evaporator. Left over powder was dissolved in water at the concentration of 100 mg/ml. The MIC of this water extract was tested against colistin resistant isolates described in Table 2 using the method described above for colistin.

BALB/c mice were used in the study. All animals were maintained and treated in accordance with the recommended rules of the WMA Helsinki declaration and the permit granted by University of Health Sciences, Lahore, Pakistan, ethical and research committee wide letter number UHS/REG-19/ERC/1236.

For infection experiments, bacterial strains were grown on LB agar plats overnight at 37°C. From the overnight culture, the isolates were sub-cultured into LB broth until the logarithmic phase. An inoculum consisting of 10⁷ cells of the fresh culture in log phase were used to infect animals. Adult mice were inoculated with A. baumannii isolates by intraperitoneal injection. When required, animals were feed with 10 mg of S. lappa extract or 100 ug of tetracycline drug. Animals were maintained under standard laboratory conditions with free access to food and water throughout the experimental period. Mice were sacrificed 20 h post inoculation (hpi) after anesthesia. For neutrophil count, 500 micro liters of the blood were withdrawn through a heart puncture and stored in vials containing 50 ul of 0.5 M EDTA. The relevant organs (such as lungs, liver and spleens were aseptically removed and used for quantitative bacteriology. Organs were homogenized in sterile saline using aerosol-proof homogenizers. Aliquots (100 μl) of 10-fold serial dilutions of the homogenates were cultured on LB agar plates to quantify the number of viable A. baumannii organisms in the respective organs. Total white blood cells and neutrophil count was measured by hematology analzer (Beckman Coulter DxH 900) and cytokine levels were measured by ELISA according to manufacturer’s instructions (Bio Assay Technology Laboratory).

Graph Pad Prism seven was used to construct graphs and perform statistical analysis of results. The type of statistical analysis preformed on each set of data is indicated in the respective figure legend.