-
PDF
- Split View
-
Views
-
Cite
Cite
Eun-Jeong Yoon, Hyun Soo Kim, Heungjeong Woo, You Jeong Choi, Dongju Won, Jong Rak Choi, Young Ah Kim, Seok Hoon Jeong, Trajectory of genetic alterations associated with colistin resistance in Acinetobacter baumannii during an in-hospital outbreak of infection, Journal of Antimicrobial Chemotherapy, Volume 77, Issue 1, January 2022, Pages 69–73, https://doi.org/10.1093/jac/dkab363
- Share Icon Share
Abstract
As carbapenem-resistant Acinetobacter baumannii is dominant in clinical settings, the old polymyxin antibiotic colistin has been revived as a therapeutic option. The development of colistin resistance during treatment is becoming a growing concern.
To access low- to mid-level colistin-resistant A. baumannii blood isolates recovered from an outbreak in a tertiary care hospital from a national antimicrobial surveillance study.
The entire bacterial genome was sequenced through long-read sequencing methodology. Quantitative RT–PCR was carried out to determine the level of gene expression. Relative growth rates were determined to estimate fitness costs of each isolate caused by the genetic alterations.
The A. baumannii isolates belonged to global clone 2 harbouring two intrinsic phosphoethanolamine transferases. Cumulative alterations continuing the colistin resistance were observed. PmrC overproduction caused by the PmrBA226T alteration was identified in A. baumannii isolates with low-level colistin resistance and an additional PmrCR109H substitution led to mid-level colistin resistance. Truncation of the PmrC enzyme by insertion of ISAba59 was compensated by ISAba10-mediated overproduction of EptA and, in the last isolate, the complete PmrAB two-component regulatory system was eliminated to restore the biological cost of the bacterial host.
During the in-hospital outbreak, a trajectory of genetic modification in colistin-resistant A. baumannii isolates was observed for survival in the harsh conditions imposed by life-threatening drugs with the clear purpose of maintaining drug resistance above a certain level with a reasonable fitness cost.
Introduction
Acinetobacter baumannii infection is a menace in clinical settings due to its remarkable ability to acquire resistance to various antimicrobials.1 To cope with the rampant spread of MDR A. baumannii, the old drug colistin, which had been withdrawn due to serious nephrotoxicity and neurotoxicity, was repermitted restrictively in the 1990s.2 The negatively charged outer membrane of A. baumannii, which is composed of anionic LPS/lipooligosaccharide (LOS) molecules, is targeted by the cationic cyclic peptide colistin.3 In A. baumannii, colistin resistance is mediated by reducing the net anionic charge of the bacterial surface through enzymatic modification, either by blocking a phosphate moiety on lipid A using phosphoethanolamine (PEtN)/galactosamine residues4,5 or by hyperacylation of lipid A,6 and complete loss of LPS/LOS by disrupting the lipid A biosynthetic pathway, particularly LpxACD.7 The intrinsic PEtN transferase (PEtNT) PmrC in A. baumannii is regulated by the PmrAB two-component regulatory system,4 and the amino acid substitutions, occurring either in the regulatory system or in the PmrC enzyme, are associated with an increasing level of pmrC expression, resulting in colistin resistance.8 The PmrC homologue EptA is intrinsic in global clone 2 (GC2) and the activated eptA confers colistin resistance by acquiring an IS-mediated strong promoter.8,9 Plasmid-mediated PEtNT mobile colistin resistance (MCR) has rarely been identified in A. baumannii clinical strains.10
Materials and methods
Colistin-resistant A. baumannii blood isolates were recovered within 1.5 months from six patients (Table S1, available as Supplementary data at JAC Online) hospitalized in a 700 bed general hospital in South Korea through the Korea Global Antimicrobial Resistance Surveillance System.11 The species was identified by MALDI-TOF MS and the blaOXA-51-like gene by PCR. This study was approved by the Institutional Review Board of the hospital (HDT 2020-04-001). Other experimental methods used in the study are included in the Supplementary Materials and methods.
Results
Outbreak of colistin-resistant A. baumannii
Among the six patients involved in the outbreak, five were initially hospitalized in an ICU and patient P07 was hospitalized in a general ward (GW) (Figure S1, Table S1). ICU patient P06 was transferred to a GW and the blood isolate G20AB006 was recovered on the ninth day of GW hospitalization. P06 and P07 received colistin for treatment before identification of the colistin-resistant A. baumannii blood isolates; P09 and P11 were treated with colistin afterwards; and P05 and P10 were never treated with colistin during their hospitalization.
At least one additional A. baumannii isolate was recovered from non-blood specimens from each patient, exhibiting similar antibiograms (Figure S1). Unfortunately, the non-blood isolates were never stored after routine examination, and further examination was not allowed.
Genetic background
All six A. baumannii blood isolates were identified as ST191 belonging to GC2. Since an O-antigen ligase, which is involved in the ligation of the O-antigen onto the lipid A core, was identified in all the genomes, the A. baumannii isolates in the study were expected to produce LPS rather than LOS.12
The six isolates were XDR, being resistant to the antimicrobials tested except tigecycline (Table S2). Fluoroquinolone resistance was conferred by GyrAS83L/ParCS80L mutations and β-lactam resistance was conferred by intrinsic blaADC-25 and blaOXA-66 with ISAba1-mediated blaADC-25 overexpression,13 together with two copies of blaOXA-23 within transposon Tn2009 in the chromosome. There were eight acquired genes for antimicrobial resistance, seven of which were identified in the A. baumannii genomic resistance island 2-0b (AbGRI2-0b):14 a class 1 integron with cassette arrays of aacA4-catB8-aadA1 followed by sul1, armA flanked by ISEc29 and ISEc28 elements, and adjoining msr(E)/mph(E).
Mechanism of colistin resistance
The four entirely sequenced genomes had an intact lipid A biosynthesis system7 and lipid A acylation-associated lpxLAb and lpxMAb, as the genome of the colistin-susceptible SSMA17 isolate had. None of the known mcr alleles was identified in the genomes.
The isolates possessed a pair of intrinsic PEtNT genes, eptA and pmrC, as in ordinary GC2 A. baumannii isolates.15 PmrBA226T alteration caused by the pmrBG676A mutation was identified in the genomes of both G20AB005 and G20AB006 isolates, and the accompanying colistin MICs were 2 and 4 mg/L (Figure 1c). G20AB010 had an additional PmrCR109H by the pmrCG326A mutation and the resulting colistin MIC was 16 mg/L. In isolates G20AB007, G20AB009 and G20AB011, an ISAba59 element was inserted within pmrC, leading to a disruption of its expression (Figure 2, Figure S2). Finally, in G20AB011, loss from the 3′ region of pmrC to the 5′ region of dnaJ, including the entire pmrAB, was identified (Figure 2).

Level of expression of PEtN transferase-encoding genes, relative growth rates and the genetic alterations associated with the colistin resistance. (a) The pmrC and eptA gene expression levels of each colistin-resistant A. baumannii blood isolate. The quantified level of gene expression, determined by quantitative RT–PCR, is presented relative to that in the colistin-susceptible A. baumannii SSMA17 (CP020581). The pmrC gene was not present in the G20AB007, G20AB009 and G20AB011 isolates. The expression level was measured in triplicate samples for two biological replicates, and two independent experiments were performed. Error bars indicate the standard error determined by a two-tailed Student’s t-test with reference to the values for SSMA17. **P < 0.01; *P < 0.05. (b) Relative growth rates of the A. baumannii isolates. Growth rates at the early exponential phase of the six colistin-resistant A. baumannii blood isolates were determined, and the relative values were calculated. The growth rates were measured for triplicate samples for two biological replicates, and two independent experiments were performed. Error bars indicate the SD determined by a two-tailed Student’s t-test with reference to the values for G20AB005. **P < 0.01. MH, Mueller–Hinton. (c) Colistin MICs of each strain and aa alterations associated with the colistin resistance. Of note, the colistin MIC for the drug-susceptible A. baumannii SSMA17 was 0.5 mg/L. CDS, coding sequence; Δ, truncated; O/P, overproduced.

Schematic representation of the genetic environment of the pmrCAB operon (a) and the eptA gene (b) in the chromosome of low- to mid-level colistin-resistant A. baumannii blood isolates. Open arrows correspond to ORFs, blue arrows indicate the genes encoding PEtN transferases, and those in orange indicate IS elements. The yellow asterisks indicate genes with alterations, namely pmrBG676A and pmrCG326A. Red- and blue-shaded areas represent regions of >99% nucleotide sequence identity in the same and opposite directions, respectively. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
The ISAba59 element was a novel IS. The 1039 bp IS, belonging to the IS903 group of the IS5 family, was composed of one ORF encoding a 310 aa DDE-type transferase and 17/18 bp imperfect inverted repeats. The ISAba59 integration presented 9 bp DRs of CTTTTTACC. ISAba12 is the nearest known IS, sharing 88% nt identity for the entire 1039 bp coverage.
The three isolates devoid of pmrC, i.e. G20AB007, G20AB009 and G20AB011, had an ISAba10 element insertion upstream from eptA. The EptA overproduction could functionally replace the PmrC (Figure 2). These isolates had colistin MICs between 8 and 32 mg/L. In the G20AB009 and G20AB011 isolates, doubling of the ISAba10 elements in the opposite direction was detected at 2191 bp upstream from the primary site, and it resulted in truncation of the TonB-dependent siderophore receptor (TBDR) (Figure 2).
Overproduction of PEtNT
The PmrBA226T alteration doubled the level of pmrC expression and the colistin MICs were 2–4 mg/L (Figure 1a and c). A 2-fold increase in pmrC expression was observed in the isolate with PmrBA226T and a 2-fold increase in eptA expression was found in the isolate having promoter replacement by the ISAba10 insertion upstream from the gene. Resulting colistin MICs by any mutation ranged from 8 to 32 mg/L.
The colistin-resistant isolates had the characteristic spectra of PEtN-modified lipid A at m/z 2033, shifted from the normal lipid A at m/z 1910. However, the intensity was not clearly associated with the level of colistin resistance.
Relative growth rates (rGRs) of the colistin-resistant A. baumannii isolates
The rGRs were deduced from the growth rates relative to that of the primary low-level colistin-resistant G20AB005 isolate (Figure 1b). The rGR of G20AB006, with an identical PmrBA226T alteration to G20AB005, was 0.96 ± 0.01 (P = 0.0030), and that of G20AB010, with PmrBA226T and PmrCR109H, was 0.97 ± 0.04 (P = 0.1167). Both isolates with eptA overexpression and truncated PmrC enzyme, G20AB007 and G20AB009, presented rGRs of 0.99 ± 0.03 and G20AB011, with the ISAba10-associated TBDR truncation, exhibited an rGR of 1.02 ± 0.02.
Addition of 1 mM FeCl3 further decreased the rGR of G20AB006 to 0.93 ± 0.08 and similarly lowered the rGRs of G20AB010 and G20AB007 to 0.92 ± 0.11 and 0.94 ± 0.08, respectively, regardless of the alteration or truncation of PmrC. The excess iron increased the rGR of G20AB009, with the truncated TBDR, to 1.11 ± 0.32, whereas it decreased the rGR of G20AB011, devoid of the TBDR and the entire PmrCAB, to 0.96 ± 0.10.
Discussion
The A. baumannii GC2 is a notorious MDR clone and the exemplary ST191 is common in clinical settings worldwide.16 The outbreak-associated isolates in this study were genetically close to the dominant ST191 clone in 2013. In this study, we observed how the dual PEtNT in the GC2 helped to achieve colistin resistance in a competent manner for survival. The PmrC-homologous EptA enzyme devoid of the regulatory system is overproduced through replacement of the gene promoter, mostly by insertion of ISAba1 and ISAba125.8,9 The ISAba10 insertion upstream from the eptA gene was a novel finding (Figure S3), and it is noteworthy that the ISAba10 element was first identified in South Korea.17 Overproduction of the EptA enzyme was likely more efficient than that of the PmrC enzyme: 16- to 64-fold elevated colistin MICs by 2× expression of eptA and 4- to 8-fold augmented colistin MICs by 2× expression of pmrC. A similar observation was reported by Gerson et al.18
In the study by Snitkin et al.,19 the initial isolate from a patient without colistin-treatment history was susceptible to colistin and the resistant bacterial isolate was selected through the colistin treatment. Right after the treatment ended, a drug-susceptible isolate was recovered. However, in this outbreak, the A. baumannii isolates were already resistant to colistin, even before the colistin usage, indicating in-hospital selection of the colistin-resistant bacteria. The low-level colistin-resistant isolate with PmrBA226T tended to favour the emergence of additional mutations to confer mid-level resistance to colistin. Sudden loss of PmrC by ISAba59 interruption was compensated by EptA overproduction by acquiring an ISAba10-associated promoter. Subsequent duplication of the ISAba10 element, resulting in TBDR disruption, led to a beneficial effect in the environment of excess iron20 and eliminating the PmrAB two-component regulatory system returned the growth rate to its original condition. Taken together, the colistin resistance and the bacterial physiology are systematic for better fitness of the bacterial host.
We demonstrated bacterial evolution in an in-hospital outbreak. Favoured by the dual intrinsic PEtNTs in the A. baumannii GC2 isolates, a certain level of colistin resistance was retained and, with the collaboration of IS elements, the physiology of the bacterial host was systematically controlled.
Funding
The research was supported by a fund (2020E540600) from the Research Program of Korean Disease Control and Prevention Agency. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Transparency declarations
None to declare.
Supplementary data
Supplementary Materials and methods, Tables S1 to S3 and Figures S1 to S3 are available as Supplementary data at JAC Online.
References
Author notes
Present address: Division of Antimicrobial Resistance Research, National Institute of Infectious Diseases, National Institute of Health, Korea Disease Control and Prevention Agency, Cheongju-si, South Korea.
These authors contributed equally as first authors.
§Eun-Jeong Yoon and Hyun Soo Kim contributed equally as last authors.