Prevalence and Genetic Characterization of Carbapenem- and Polymyxin-Resistant Acinetobacter baumannii Isolated from a Tertiary Hospital in Terengganu, Malaysia

Nosocomial infection caused by Acinetobacter baumannii is of great concern due to its increasing resistance to most antimicrobials. In this study, 54 nonrepeat isolates of A. baumannii from the main tertiary hospital in Terengganu, Malaysia, were analyzed for their antibiograms and genotypes. Out of the 54 isolates, 39 (72.2%) were multidrug resistant (MDR) and resistant to carbapenems whereas 14 (25.9%) were categorized as extensive drug resistant (XDR) with additional resistance to polymyxin B, the drug of “last resort.” Pulsed-field gel electrophoresis analyses showed that the polymyxin-resistant isolates were genetically diverse while the carbapenem-resistant isolates were clonally related. The 14 XDR isolates were further investigated for mutations in genes known to mediate polymyxin resistance, namely, pmrCAB, and the lipopolysaccharide biosynthesis genes, lpxA, lpxC, lpxD, and lpsB. All 14 isolates had a P102H mutation in pmrA with no mutation detected in pmrC and pmrB. No mutation was detected in lpxA but each polymyxin-resistant isolate had 2–4 amino acid substitutions in lpxD and 1-2 substitutions in lpxC. Eight resistant isolates also displayed a unique H181Y mutation in lpsB. The extent of polymyxin resistance is of concern and the novel mutations discovered here warrant further investigations.


Introduction
Acinetobacter baumannii is a Gram-negative bacterium increas-ing-ly found in the hospital environment due to its ability to survive for long periods on inanimate objects [1,2]. Nosocomial A. baumannii isolates are mostly multidrug resistant (MDR) (i.e., resistant towards more than three classes of antibiotics) [3]. However, extensive drug resistant (XDR) isolates, which are resistant to all but one or two classes of antibiotics, and even pandrug resistant (PDR) isolates that are resistant to all classes of antibiotics, are developing at an alarming rate [4]. Since A. baumannii has the ability to colonize both viable and damaged tissues and is also resistant towards nearly all antimicrobials, it has become a cause of great concern. Carbapenems are among the very few antibiotics left that can be used for the treatment of A. baumannii infections. Nevertheless, the efficacy of carbapenems is increasingly compromised by the rapid emergence of carbapenem-hydrolysing -lactamase enzymes. Several types of class D -lactamases including OXA-23, OXA-24, OXA-58, and intrinsic OXA-51-like enzymes are known to be important contributors to carbapenem resistance [5].
The rapid development of carbapenem-resistant MDR A. baumannii has led to the use of polymyxins (in particular polymyxin B and colistin or polymyxin E) as the drug of "last resort" [6]. Polymyxins are cyclic, positively charged peptide antibiotics capable of posing antimicrobial activities to a broad variety of Gram-negative pathogens, including A. baumannii, due to their interaction with the lipid A moiety of lipopolysaccharide (LPS). This leads to the disorganization 2 ISRN Microbiology and disruption of the outer membrane integrity, causing cytoplasmic leakage [7]. Unfortunately, the intensive use of the polymyxins in recent years has led to the emergence of polymyxin heteroresistant and resistant A. baumannii, as high as 40.7% reported in Spain [8] and 30.6% in Korea [9].
The basis of polymyxin resistance in A. baumannii has only recently been investigated and several mechanisms have been proposed. Several genetic loci have been implicated in the resistance towards polymyxins in Acinetobacter, namely, the pmrCAB operon [10,11] and the lpxA, lpxC, lpxD [12,13], and lpsB genes, that are involved in LPS biosynthesis [14]. Resistance can arise through mutations in the twocomponent system PmrAB, in which the downstream target PmrC catalyzes the addition of phosphoethanolamine to the lipid A component of LPS [10,11,15]. This modification reduces the net negative charge of the outer membrane thus reducing the affinity of polymyxins for the target. Park et al. (2011) [16] also showed that increased expression of pmrAB could be essential for colistin resistance in A. baumannii. Mutations or insertions in the genes encoding the lipid A biosynthesis machinery, namely, the lpxA, lpxC, or lpxD genes, also mediate polymyxin resistance by abolishing the production of LPS, thereby eliminating the target of polymyxins [12,13,17]. Hood et al. (2013) [14] recently reported the identification of over 20 genes that are implicated in inducible colistin tolerance in A. baumannii, with majority of the genes involved in processes that protect the bacterium from osmotic stress. One of the genes identified, lpsB, encodes a glycosyltransferase involved in LPS biosynthesis. Polymyxin resistance in A. baumannii thus developed through intrinsic mechanisms that respond and adapt to polymyxin treatment and not through acquisition of horizontally transferred resistance genes [12,13].
To our knowledge, there has been no published data on polymyxin-resistant isolates of A. baumannii from Malaysia. In this study, we determined the resistance profiles of A. baumannii isolated from Hospital Sultanah Nur Zahirah (HSNZ), the main tertiary hospital located in the state of Terengganu in the east coast of Peninsular Malaysia. The genetic diversity of the A. baumannii isolates was determined by pulsed-field gel electrophoresis (PFGE). This study also aims to determine the presence and distribution of genes encoding the class D carbapenemase OXA-51 , OXA-23 , OXA-24 , and OXA-58 , as well as the class B carbapenemase (or metallo--lactamase) genes VIM and IMP in the clinical isolates of A. baumannii. Besides that, we also aim to investigate the type of mutations that occur within the pmrCAB, lpxA, lpxC, lpxD, and lpsB genes that may confer resistance to polymyxin. Findings from this research project will help to determine the antimicrobial resistance, distribution of carbapenem and polymyxin resistance determinants, and the clonality of the A. baumannii isolates from HSNZ, thereby providing valuable information regarding this important nosocomial pathogen in this region.

Bacterial Isolates: Identification and
Genotyping. Fiftyfour nonrepeat clinical isolates of A. baumannii were collected throughout 2011 from sporadic cases of infection in Hospital Sultanah Nur Zahirah (HSNZ). The isolates were isolated and initially identified by standard biochemical methods at the clinical microbiology laboratory. Single colonies of A. baumannii were then cultured into nutrient broth and incubated at 37 ∘ C for 24 hours. Confirmation of the 54 isolates was done by 16S rDNA amplification and sequencing. Genomic DNA was extracted from the bacterial isolates using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA) and subjected to PCR using 16S rDNA universal primers [18] (Table 1). Strain typing of A. baumannii was carried out by pulsed-field gel electrophoresis (PFGE), as previously described [19]. Briefly, chromosomal DNA for PFGE analysis was prepared in agarose gel blocks, digested with restriction endonuclease ApaI (Promega, Madison, WI, USA). Restriction fragments were separated by using the CHEF Mapper (Bio-Rad, Hercules, CA, USA) with 0.5 × TBE buffer for 26 h at 14 ∘ C with pulse times of 2-40 s. XbaI-digested Salmonella enterica ser. Braenderup H9812 was used as the molecular size standard. The PFGE profiles were analyzed using BioNumerics version 6.0 software (Applied Maths, Kortrijk, Belgium). The unweighted pair group method with averages (UPGMA), with a position tolerance for comparison of 1.5%, was used to produce the dendrogram.
Bacterial cultures of A. baumannii adjusted to 0.5 McFarland standard were used to determine the minimal inhibitory concentration (MIC) to imipenem and meropenem using M.I.C. Evaluator (M.I.C.E.) strips (Oxoid Ltd., Basingstoke, UK). Inoculated media were incubated at 35 ± 2 ∘ C for 16-20 h. The MIC values of the respective inhibition ellipses that interact with the strips were taken and recorded. Data were analyzed according to the guidelines given for M.I.C.E. strips and the MIC was determined.
MIC values for polymyxin B were determined by the agar dilution method [21]. Agar dilution plates were prepared by using Mueller-Hinton Agar added with concentrations of 2, 4, 8, 16, 32, 64, and 128 g/mL polymyxin B sulfate (Sigma-Aldrich, St. Louis, MO, USA). A multipoint inoculator was used to deliver bacterial suspensions adjusted to 0.5 McFarland standard onto the agar surface and the agar subsequently incubated at 35 ± 2 ∘ C for 16-20 h. The MIC value for polymyxin B was determined as the lowest 4 ISRN Microbiology concentration of the antibiotic in which the bacteria were susceptible. The standard breakpoint for resistance according to the CLSI guideline [20] was ≥ 4 g/mL.

Detection of Carbapenem-and Polymyxin-Resistant
Determinants, DNA Sequencing, and Sequence Analysis. The PCR primers and conditions used to detect the presence of several carbapenem resistance genes such as OXA-23 , OXA-51 , OXA-24 , OXA-58 , IMP , and VIM are indicated in Table 1. The validity of the amplicons was determined by DNA sequencing. Briefly, the amplified products were purified by using the Wizard SV Gel and PCR Clean-Up System (Promega, Madison, WI, USA) and then submitted to a commercial facility (First Base Sdn. Bhd., Malaysia) for sequencing.
Genetic determinants that have been implicated in conferring polymyxin resistance in A. baumannii, namely, the pmrCAB, lpxA, lpxC, lpxD, and lpsB genes, were PCRamplified using specific primers and conditions as shown in Table 1. Purified PCR products were then sequenced at the commercial facility (First Base Sdn. Bhd., Malaysia).

Antibiotic Susceptibility Profiles of A. baumannii .
DNA sequence analyses of the 16S rRNA amplicons showed ≥99% sequence identity with A. baumannii 16S rRNA genes validating that all 54 isolates used in this study were A. baumannii.

Detection of Class D and Class B Carbapenemase Genes.
PCR was used to detect the presence of the class D carbapenemase genes OXA-51 , OXA-23 , OXA-24 , and OXA-58 . All 54 A. baumannii isolates yielded OXA-51 amplicons of the expected size of 353 bp. Sequence analyses of the amplified products confirmed their identity as OXA-51 or OXA-51-like gene. Forty-one (75.9%) of the 54 isolates yielded OXA-23 amplicons of the expected size of 501 bp, and sequencing of the amplified products validated the identity as  . None of the isolates harbor the OXA-24 or the OXA-58 genes. Similarly, no class B metallo--lactamase (MBL) genes, IMP or VIM , were found in all the isolates tested.

Sequence Analyses of Polymyxin Resistance Determinants.
All 14 polymyxin B resistant isolates were each subjected to PCR amplifications that target the pmrCAB, lpxA, lpxC, lpxD, and lpsB genes and the resulting amplicons were sequenced. The three isolates (AC6, AC11, and AC19) that were susceptible to polymyxin B were used as controls.
In this study, we found as many as 11 different point mutations within the pmrB gene when compared to the reference isolates ATCC 17978 and ATCC 19606 but these mutations were similarly found in the polymyxin-sensitive isolates. Thus, when the pmrB sequences of 14 polymyxinresistant isolates were compared with the 3 polymyxinsensitive isolates, no mutations within the pmrB gene could be found. Similarly, when comparing the sequences of the response regulator gene pmrA, 3 different point mutations (R149K, T153I, and Q165P) were observed but these mutations were commonly found in both sensitive and resistant isolates. However, all 14 resistant isolates had a P102H mutation within pmrA that was absent in the sensitive isolates ( Table 2). When comparing the sequences for the pmrC gene that encodes lipid A phosphoethanolamine transferase, up to 8 point mutations were found in comparison with the reference isolates. However no difference was found amongst the polymyxin-sensitive and polymyxin-resistant isolates.
The lpxA, lpxC, and lpxD genes encode the first three enzymes in the lipid A biosynthesis pathway. No mutation was found in lpxA. In contrast, lpxD showed up to 4 amino acid mutations in each resistant isolate while lpxC had 1-2 amino acid mutations ( Table 2). The 14 resistant isolates had the K141R, S158R, or both mutations in lpxC which encodes the enzyme involved in the second step of the lipid A biosynthesis pathway. The lpxD gene showed the most number of mutations variation among the resistant isolates with a total of 13 different amino acid substitutions. Each resistant isolate had between 2 and 4 amino acid substitutions in lpxD and most of the mutations occurred between amino acid residues 150 and 190 of LpxD. Further comparison of the lpxD sequence was carried out with known polymyxinsensitive isolates in the database and results of the multiple sequence alignment (data not shown) indicated that the amino acid substitutions found in our resistant isolates are unique.
sequences among the 14 resistant isolates, the 3 sensitive isolates, and the 2 ATCC reference isolates initially showed that each resistant isolate contained as many as 4 amino acid substitutions in lpsB. However, when the lpsB sequence from other polymyxin-sensitive isolates in the database was taken into account, the majority of these amino acid substitutions were found to be common among the other polymyxinsensitive isolates. The only unique mutation was H181Y and this substitution was found in 8 of the 14 resistant isolates ( Table 2). The remaining 6 resistant isolates did not harbor any unique mutation within lpsB.

Genotyping of Acinetobacter baumannii Isolates by Pulsed-Field Gel Electrophoresis. PFGE analysis of the 54
ApaI-digested A. baumannii gave 15 reproducible profiles (pulsotypes) with a Dice coefficient, F, ranging from 0.90 to 1.00. Cluster analysis of the ApaI pulsotypes grouped the 54 A. baumannii isolates into 9 clusters with a cutoff point at 80% similarity ( Figure 1). Carbapenem-resistant isolates were grouped into four main clusters, A-D, while carbapenemsusceptible isolates were grouped into two clusters, E and F. Cluster A is the largest cluster with 21 carbapenemresistant isolates which were closely related (81.6% similarity), with 4 band differences at most (F value = 0.90-1.00). Four of the isolates within the A cluster were also polymyxin resistant and XDR. In addition, all the carbapenem-resistant isolates in clusters A-D were all positive for OXA-51-like and OXA-23 , whereas the susceptible isolates in clusters E and F were negative for OXA-23 . However, there was no clustering among pulsotypes of the polymyxin-resistant, XDR isolates. In fact, isolates with identical PFGE patterns would differ in their resistance to polymyxin B. For the three isolates in cluster C which shared identical PFGE patterns, two of these (AC17 and AC18) were resistant to polymyxin B but the third isolate, AC19, was polymyxin sensitive. Both AC17 and AC18 shared identical mutations within the polymyxin-resistance determinants (i.e., E50D and V141I in lpxD, K141R in lpxC, and H181Y in lpsB). Likewise, AC29 and AC30 within cluster A shared an identical PFGE pattern but AC29 was sensitive to polymyxin B whereas AC30 was resistant.

Discussions
This study showed that the HSNZ isolates of A. baumannii were highly resistant to most antibiotics. Out of 54 isolates, 47 (87%) were resistant to tetracycline, which was the highest resistance among the 12 antibiotics tested. The percentages of resistance for carbapenems were also high with 77.8% for meropenem and 74.1% for imipenem. Although the percentage of carbapenem resistance was lower when compared to the rates reported from another medical facility in Kuala Lumpur (96.5% for imipenem and 98.2% for meropenem [19]), the finding that more than 50% of the isolates from HSNZ were carbapenem resistant is certainly a cause for concern as carbapenems have been the drugs of choice for Acinetobacter infections for over a decade [23,24].
were positive for the OXA-23 gene. Of these 41 isolates, 37 were resistant to both meropenem and imipenem.
The second and third subgroups of carbapenem hydrolyzing -lactamases were the OXA-24-like enzymes and OXA-58-like enzymes [23] encoded by  genes, respectively. In this study, none of the 54 isolates were found to harbor the OXA-24 and OXA-58 genes, similar to recent findings from Thailand [27]. Likewise, none of the isolates carried the metallo--lactamase genes, IMP and VIM , both of which have not been reported in A. baumannii in this region. However, reports from Korea [28,29] showed that these two genes were implicated in carbapenem resistance.
The recent emergence of polymyxin-resistant and heteroresistant A. baumannii [6] has necessitated the need to understand the mechanism of the development of polymyxin resistance in this pathogen. Since this older drug has recently returned in use and has been noted as the drug of last resort for Acinetobacter infections [6], we investigated the antibiotic resistance of the HSNZ isolates towards polymyxin B. It was found that 14 of the 54 isolates (or 25.9%) were resistant to polymyxin B, of which 4 had MIC values for polymyxin B >128 g/mL. The mechanism for polymyxin resistance in A. baumannii has only recently been investigated and the main mechanism appeared to be either covalent modification of the lipid A portion of LPS [11,15] or disruption of LPS biosynthesis [13,16]. By modification and/or mutations in the amino acid sequences, negative charges on the outer membrane can be reduced, leading to reduction in the affinity of the positively charged polymyxin component, hence giving rise to polymyxin resistance [10,11,15].
It is interesting that the 14 polymyxin-resistant isolates displayed an identical P102H mutation within the pmrA gene and this mutation was identical to that previously reported by Adams et al. (2009) [10] in a colistin-resistant derivative of A. baumannii AB0057. Overexpression of pmrAB and mutations within these genes, especially pmrB, were reported to contribute to polymyxin resistance [10,11,16]. As expression levels of pmrAB were not determined in this study and no mutation within pmrB was found, we are not certain if the P102H mutation in pmrA is a contributing factor to the polymyxin resistance in the 14 XDR isolates. However, analysis of the LPS biosynthesis genes identified novel mutations especially in the lpxD, lpxC, and lpsB genes. All 14 XDR isolates had mutations in both lpxD and lpxC genes whereas 8 out of the 14 isolates showed an identical H181Y substitution in lpsB. It does appear that disruption in the LPS biosynthesis genes could be associated with polymyxin resistance in the A. baumannii isolates studied. The effects of these mutations on the LPS of these isolates need to be further investigated as they differ from the mutations previously reported [30]. Furthermore the crystal structure of the A. baumannii LpxD protein was recently elucidated [31] and one of the mutations (G186S) lies within the proposed active site of the enzyme. Nevertheless, it should be noted that in a recent study [14], screening of transposon mutant libraries led to the identification of more than 20 genes that may be involved in inducible colistin resistance in A. baumannii. Most of these genes converge on pathways involved in osmotolerance, cell envelope biosynthesis along with protein folding [14]. The role that these factors may play in the development of polymyxin resistance among the 14 XDR isolates would also need to be investigated.
PFGE results indicated that the polymyxin-resistant isolates were genetically diverse. This is in contrast to the carbapenem-resistant isolates which were clonally related. The carbapenem-resistant cluster A was the largest cluster inferring that this is the major clonal group of A. baumannii existing within the hospital in 2011. Four of the 21 isolates within the A cluster were polymyxin resistant and XDR with three of these isolates having MIC values for polymyxin B > 128 g/mL. Similar results were obtained whereby polymyxin-resistant isolates were found in clusters E, B, and C although the MIC values for these isolates were ≤ 32 g/mL. The clustering of carbapenem-resistant isolates has indeed been previously reported [32,33]. Carbapenem resistance is due mainly to the acquisition of carbapenemases such as OXA-23 which are encoded on plasmids and other mobile elements. Isolates that acquire such genes will then undergo clonal expansion [34,35]. In contrast, polymyxin resistance has been shown to develop through mutations in certain determinants in response to selection pressure [10][11][12][13]17]. Thus these isolates were likely to develop randomly in the presence of polymyxins. Since polymyxin resistance is due to alterations or the complete absence of the outer membrane of LPS, it may incur a fitness cost to the bacterium. Therefore these isolates may not persist in the absence of polymyxin which could explain the random, nonclustering nature of the polymyxin-resistant isolates. This was indeed reported in a recent paper [36] where colistin-susceptible isolates outcompeted resistant isolates upon withdrawal of colistin in three out of four cases. However, in one case, an isolate with a pmrB L271R mutation appeared to be fit enough to be transmitted in the absence of colistin and not be outcompeted by susceptible isolates [36]. Colistin-resistant A. baumannii has also been reported to have reduced virulence in mice [37] although a more recent report indicated contrasting results whereby colistin resistance may not necessarily intrinsically affect virulence [38]. This is the first report of polymyxin-resistant and XDR A. baumannii from Malaysia. The emergence of XDR and polymyxin-resistant A. baumannii isolates from the main tertiary hospital in Terengganu, Malaysia, is a cause of concern. Stricter enforcement of antibiotic usage and an antibiotic stewardship program are clearly needed in the hospital to prevent the spread of XDR A. baumannii.

Conclusions
Multidrug resistant A. baumannii was observed in 39 of 54 isolates (72.2%) that were obtained from the main tertiary hospital in Terengganu, Malaysia, throughout 2011. High incidence of resistance to carbapenems (about 75%) was noted. Of concern, 14 of the 54 isolates (25.9%) were characterized as extensive drug resistant and were resistant to polymyxin B, considered the drug of "last resort. " Pulsed-field gel electrophoresis indicated that the polymyxin-resistant isolates were genetically diverse. A P102H mutation was 8 ISRN Microbiology found in the pmrA gene in all polymyxin-resistant isolates. In addition, novel mutations within the LPS biosynthesis genes lpxC, lpxD, and lpsB were detected in the polymyxinresistant isolates. Whether these mutations contribute to the development of polymyxin resistance in A. baumannii requires further investigations.