Phenotypic and Genotypic Characterization of Clinical Isolates Belonging to the Acinetobacter calcoaceticus-Acinetobacter baumannii (ACB) Complex Isolated From Animals Treated at a Veterinary Hospital in Switzerland

Objectives: We investigated a collection of strains belonging to the Acinetobacter calcoaceticus-Acinetobacter baumannii (ACB) complex obtained from a veterinary clinic with regard to their genetic relatedness, presence of antibiotic resistance genes and antimicrobial susceptibility profiles. Methods: Fifty-eight ACB-complex strains from animals treated at a veterinary clinic between 2006 and 2017, and seven strains collected from the hospital environment during 2012 were analyzed. Assignment to sequence types (ST) and international complexes (IC) was done by multilocus sequence typing (MLST) according to the Pasteur scheme. Genes encoding carbapenemases, aminoglycoside-modifying enzymes, macrolide-, quinolone- and co-trimoxazole resistance genes, the ISAba1 element, virulence associated intI1 genes and plasmid associated toxin-antitoxin markers were identified by microarray. Genes encoding blaOXA−51-like carbapenemases were amplified by PCR and sequenced. Susceptibility profiles were determined by disc diffusion or by broth microdilution. Results: Among 50 A. baumannii isolates from animals, two predominant clones were observed linked to CC1 (n = 27/54% of the isolates) and CC25 (n = 14/28%), respectively. Strains of IC I harbored blaOXA−69, aac(3′)-la, aadA1, sul1, intI1, and splA/T genes. Isolates belonging to CC25 possessed blaOXA−64. Six (12%) isolates belonging to CC2 and carrying blaOXA−66 were also noted. One isolate belonged to CC10 (blaOXA−68), one to CC149 (blaOXA−104), the remaining isolate was assigned to ST1220 and possessed blaOXA−116. Of six environmental A. baumannii, four (66.7%) belonged to CC25 (blaOXA−64), one (16.7%) to CC2 (blaOXA−66) and one to CC3 (blaOXA−71). Nine isolates (eight from animals and one environmental strain) were non-baumannii strains and did not harbor blaOXA−51-like genes. None of the isolates carried blaOXA−23, blaOXA−48, or blaOXA−58, and none were resistant to carbapenems. Conclusions: Clonal lineages of the veterinary A. baumannii isolates in our collection are identical to those globally emerging in humans but do not harbor blaOXA−23. A. baumannii CC25 may be specific for this particular veterinary clinic environment.


INTRODUCTION
The genus Acinetobacter is ubiquitous in diverse environments and as of today comprises 60 validly published species names (www.szu.cz/anemec/Classification.pdf). A list of Acinetobacter spp. also available at http://www.bacterio.net (1).
A. baumannii is the most frequent Acinetobacter species isolated from patients in intensive care units (ICUs) and is the causative agent of ventilator associated pneumonia, catheterrelated bloodstream infections, meningitis, and wound infection, often causing clonal outbreaks involving critically ill patients (8,9). By constrast, A. calcoaceticus, although found sometimes in clinical specimens, seems to be more environmental, has unknown clinical significance and is usually well-susceptible to antibiotics (10).
The diversity of strains in epidemiology studies of A. baumannii is frequently investigated by multilocus sequence typing (MLST) using either the Oxford or the Pasteur scheme (11,12). The majority of outbreak strains reported globally belong to IC I, IC II, and IC III, corresponding to clonal complexes (CC)1, CC2, and CC3 of the Pasteur scheme (9,(12)(13)(14).
Treatment of infections is frequently compromised due to the fact that ACB complex strains possess multiple intrinsic and acquired mechanisms that may result in antimicrobial resistance (10,15). Overexpression of intrinsic ß-lactamases and of multidrug resistance efflux pumps, loss of outer membrane proteins and mutations in the quinolone resistance-determining regions (QRDRs) of the gyrA and parC are commonly detected in ACB isolates (10). Hence, following CLSI guidelines, ACB isolates are considered intrinsically resistant to antibiotics such as aminopenicillins, aztreonam, ertapenem, trimethoprim, chloramphenicol, macrolides, and fosfomycin (16).
Importantly, in addition to the presence of chromosomally located bla OXA−51 -like genes encoding for naturally occurring carbapenemases, plasmid mediated carbapenem resistance genes including bla OXA−23 , bla OXA−48 , and bla OXA−58 have emerged globally in A. baumannii further restricting therapeutic options for treating infections in humans, with bla OXA−23 harboring A.
baumannii representing one of the most problematic hospitalacquired human pathogens (17).
Data on molecular characteristics and antimicrobial resistance mechanisms of Acinetobacter of veterinary origin are still scarce compared to those of isolates from humans. However, it has been shown that A. baumannii isolated from animals may share clonal lineages and possess identical transmissible antibiotic resistance genes to those from humans, suggesting common pathways and/or sources of infection (15,18). Furthermore, reports on the emergence of infections due to carbapenem resistant A. baumannii in hospitalized companion animals are of concern and emphasize the need for epidemiological studies and surveillance in order to maintain veterinary and public health (19)(20)(21)(22).
The present study was designed to characterize clinical isolates belonging to the ACB complex originating from companion animals and horses hospitalized during 2006-2017 at a university veterinary clinic in Switzerland by (i) determining the genetic relatedness using multilocus sequence typing, (ii) performing genetic profiling using a microarray-based assay, and (iii) assessing their antimicrobial susceptibility profiles.

Bacterial Isolates
Between 2006 and 2017, a total of 93 non-duplicate Acinetobacter spp. isolated from hospitalized animals (one strain per animal) were obtained. Only isolates with clinical significance were collected. In addition, strains taken from the hospital environment during 2012 (n = 7) were included in the study. Strains were identified to the level of the genus Acinetobacter using the VITEK R 2 Compact system (Biomérieux, Nürtingen, Germany).
Species identification was performed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS, Bruker Daltronics, Bremen, Germany) and by amplification and sequencing of the 350 bp highly variable zone 1 of the rpoB gene (23,24). Custom sequencing was done by Microsynth, Balgach, Switzerland.
In total, 65 ACB complex strains were identified, including 56 A. baumannii (50 isolates from animals, and six environmental strains), seven A. pittii (six animal, and one environmental isolate), and two animal A. calcoaceticus isolates.
In addition, strains collected from the hospital environment during 2012 (n = 7) were included in the study.
In accordance with local legislation, ethics approval was not required and no animal experiments were carried out for this study.

Multilocus Sequence Typing
Multilocus sequence typing was performed according to the scheme developed by the Pasteur Institute (12). This scheme involves PCR amplification and sequencing of internal fragments of seven housekeeping genes (fusA, gltA, pyrG, recA, cpn60, rpoB, and rplB). Primers and PCR conditions are listed at the A. baumannii MLST database website http://pubmlst.org/ abaumannii/. Sequencing of the amplification products was performed by Microsynth (Balgach, Switzerland). Sequences were uploaded to http://pubmlst.org/abaumannii/ to identify alleles and sequence types. The population structure of STs of the A. baumannii isolates was evaluated using the goeBURST software (http://www.phyloviz.net/goeburst/). CCs were defined as single-locus (SLVs) and double-locus variants (DLVs).

Identification of Antimicrobial Resistance Genotypes
DNA was purified using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany), according to manufacturer's protocol.
Isolates were genotyped using the oligonucleotide based microarray CarbDetect AS-2 Kit (Alere Technologies GmbH, Jena, Germany) to detect all currently known relevant carbapenemase genes, extended-spectrum ß-lactamase (ESBL) genes, aminoglycoside, macrolide, quinolone and co-trimoxazole resistance genes found in Enterobacteriaceae and Pseudomonadales (26). Additional markers included the ISAba1 element, integrase and transposase genes, and plasmid associated toxin-antitoxin (T/A) markers. An overview of the target genes and multiplex labeling, hybridization and data analysis has been described by Braun et al. (26). In brief, DNA was labeled internally with biotin-11-dUTP using a linear amplification protocol to generate single stranded (ss) DNA. Biotin labeled ssDNA was transferred and hybridized with DNA probes in oligonucleotide microarray strips. Hybridization was detected using streptavidin-horseradish peroxidase and a dye precipitation. The signals were detected using the platform ArrayMate Reader provided by Alere Technologies GmbH.
PCR and DNA sequencing analyses of bla OXA−51 -like genes in A. baumannii isolates was carried out using custom synthesized primers (Microsynth, Balgach, Switzerland) and conditions published previously (27). Nucleotide sequences were analyzed with the CLC Main Workbench 8.0.1 and the BLASTN program of NCBI (http://www.ncbi.nlm.nih.gov/blast/).

Phenotypical Characterization of Antibiotic Susceptibility
Antimicrobial susceptibility testing was carried out according to Clinical and Laboratory Standards Institute (CLSI) performance standards (16), using the disk-diffusion method and the antibiotics cefotaxime (CTX), cefepime (FEP), ciprofloxacin (CIP), sulfamethoxazole/trimethoprim (SXT), gentamicin (GM), and tetracycline (TE). Minimal inhibitory concentrations (MIC) of imipenem were determined using the E-Test R (bioMérieux, Marcy L'Etoile, France) according to manufacturer's protocol. Determination of the MIC of colistin was performed by broth microdilution according to the European Committee on Antimicrobial Susceptibility Testing EUCAST (eucast.org).
The degrees of antimicrobial resistance among the ACB complex isolates was defined in accordance to Falagas and Karageorgopoulos (30).
Multidrug resistance (MDR) was thereby defined as resistance to three or more classes of antimicrobial agents including β-lactams, fluoroquinolones, sulfonamides, aminoglycosides, tetracyclines, and polymyxins, and excluding those which cannot be regarded as potentially effective, i.e., to which ACB complex strains are intrinsically resistant: aminopenicillins, aztreonam, ertapenem, trimethoprim, chloramphenicol, macrolides and fosfomycin (16).

RESULTS
An overview of the isolates, their origins, molecular and phenotypic characteristics is shown in Figure 1.
Overall, 58 clinical isolates belonging to the ACB complex were collected from animals admitted to the veterinary hospital of Zürich, Switzerland between 2006 and 2017. The majority thereof (50/86.2%) were A. baumannii isolates collected from horses (n = 34), dogs (n = 6), cats (n = 5), rabbits (n = 2), and from one chicken, one donkey and one reptile, respectively. Six (10.3%) of the clinical isolates were A. pittii collected from cats (n = 2), chicken (n = 2), and from one Andean bear and one rodent, respectively. Two isolates (3.4%) were A. calcoaceticus from a cow and from a horse, respectively.
Environmental isolates collected from the premises of the veterinary hospital during 2012 included six A. baumannii and one A. pittii (Figure 1). Multilocus sequence typing revealed that the majority (27/48.2%) of the 56 A. baumannii isolates belonged to ST20 and its SLVs ST1, ST1217, and ST1219, as well as its DLV ST1214. eBurst analysis assigned these sequence types to clonal complex CC1 (Figure 2). Fourteen (25%) of the A. baumannii strains belonged to ST25 and its SLV ST1218 and were assigned to CC25. Of the remaining isolates, six (10.7%) belonged to ST2 (CC2), and one (1.8%) to ST23 (CC10) and one to ST46 (CC149). One isolate typed and ST1220 and was not assigned to any CC (Figure 2).
Microarray based genotyping revealed the presence of bla OXA−51 -like genes in all A. baumannii isolates. None of the A. pittii or A. calcoaceticus isolates tested positive for bla OXA−51 -like genes (Figure 1).
The ß-lactamase genes bla TEM and bla ACT were found in 44 and 2 of the isolates, respectively (Figure 1).
The ISAba1 element which accounts for enhanced expression of bla OXA−51 -like genes was detected adjacent to bla OXA−66 in two of the isolates (MAC5 and MAC22, Figure 1). In one isolate (MAC-52) the element was not associated with bla OXA−66 (Figure 1).
Genes associated with resistance to aminoglycosides, macrolides, sulphonamides and trimethoprim were detected as shown in Figure 1. The aac(3 ′ )-Ia, aadA1, aphA, and sul1 genes occurred predominantly in association with the presence of the class 1 integrase gene intI1 in A. baumannii belonging to CC1 and CC2, whereas the majority of strains that lacked the intI1 gene harbored strA and/or strB and sul2 (Figure 1).
The type II T/A genes splA and splT were identified in all A. baumannii belonging to CC1.
None of the isolates tested positive for mcr genes. The antimicrobial susceptibility profiles of the isolates are summarized in Figure 1. The majority (52/92.9%) of the 56 A. baumannii strains was resistant to three or more classes of antimicrobials and were MDR according to Falagas and Karageorgopoulos (30). By contrast, among the A. calcoaceticus and A. pittii strains, one isolate (MAC-70) was resistant to gentamicin, the rest remained susceptible to all tested antimicrobials, and none were MDR.

DISCUSSION
There is growing concern that multidrug resistant, bla OXA−23 harboring A. baumannii in hospitalized companion animals and horses may be emerging as a threat to veterinary and public health (15). However, information on A. baumannii in veterinary medicine is still limited and there is a lack of comparable data to strains isolated from humans (15,31). In this study, we provide a molecular and phenotypic analysis of strains belonging to the ACB complex isolated from diseased animals admitted to the veterinary hospital of the university of Zürich, Switzerland during 2006-2017. The main limitations of this study include its retrospective design and its restriction to a single center.
The two predominant lineages of A. baumannii comprised CC1, which is a globally distributed clade (9), and CC25, a lineage that has been responsible for epidemics in different European countries (32).
There are few reports on A. baumannii isolated from pets in Switzerland and overall, these isolates belonged primarily to CC1and CC2 (33,34). Likewise, Ewers et al. (19) observed a prevalence of 26% of CC2 among A. baumannii recovered from animals hospitalized in various veterinary clinics in Germany, which is remarkably higher than the prevalence of 12% observed in this study. By contrast, A. baumannii ST25 (CC25) was not described in either of these studies. Its abundance in the collection of ACB isolates from the veterinary hospital of Zürich suggests that this clinical setting may be likely support the spread of this particular clonal lineage. Moreover, of the six environmental A. baumannii recovered during 2012, four (66.7%) belonged to ST25 (CC25), suggesting the existence of an environmental reservoir of this ST in or outside of the hospital setting. Its prevalence in the hospital environment may also be due to the elevated resistance to desiccation and high biofilmforming capacity on abiotic surfaces, as demonstrated for this particular sequence type (35).
Notably, A. baumannii ST25 has recently been isolated from pets in France (20,21). In both studies, the isolates possessed bla OXA−23 and were resistant to carbapenems, whereas the isolates analyzed in the current report possessed intrinsic bla OXA−51 -like carbapenemases only. Interestingly, the isolates from France were detected in companion animals in the community. Compared to clinical settings, little is known about Acinetobacter carriage in animals beyond these settings, but several studies during the last decade have detected A. baumannii in dogs in the community (21,36), domestic birds (37), livestock (38) and other farmed animals such as mink (39). These reports indicate that community-acquired A. baumannii infections among animals may be increasing and that animals outside clinical settings may represent a reservoir for A. baumannii, including carbapenem resistant strains (15).
Overall, the bla OXA−51 alleles identified in the A. baumannii isolates correlated with their respective CCs, in accordance with previous observations for human isolates (27,40).
Two isolates possessed the ISAba1 element upstream of bla OXA−66 . As reported earlier, ISAba1 mediates overexpression of bla OXA−51 -like enzymes, resulting in resistance to carbapenems (41,42). However, there was no difference in the MICs of imipenem for these two isolates compared to those lacking the ISAba1 insertion, confirming recent observations that resistance to carbapenems is not guaranteed only by the presence of ISAba1, but depends on its orientation upstream of the bla gene (43,44).
Aminoglycoside resistance genes were distributed unevenly among the A. baumannii isolates. The occurrence of aadA1 and aphA in association with intI1 among CC1, CC2, and CC3 is supportive of previous observations (45). By contrast, these genes were not prevalent among the CC25 isolates, among which strA and strB genes predominated. This may indicate that interclonal horizontal gene transfer plays a minor role in the dissemination of aminoglycoside resistance in the isolates analyzed in this study.
Of the two colistin resistant isolates, one belonged to CC25, which also occurs in humans. Colistin resistance in human Acinetobacter isolates is a source of great concern, although to date, there have been no reports of mcr positive Acinetobacter spp. (46).
In general, there was a good correlation between the presence of resistance genes detected by microarray and the phenotype of the isolates. There was however, a discrepancy between results of genotypic and phenotypic testing for three of isolates containing sulfonamide resistance genes and 4 isolates harboring aminoglycoside resistance genes, where the presence of resistance genes did not correspond with phenotypic resistance. Conversely, there was a lack of bla ESBL genes such as bla GES , bla PER , or bla VEB that would explain the phenotypic resistance to cefotaxime observed in 10 isolates (10). On the other hand, ISAba1 and ISAba125 governed hyperexpression of the chromosomal bla ADC cephalosporinase also leads to resistance to 3rd generation cephalosporins in A. baumannii (10,47). Thus, further investigations targeting the genetic environment of bla ADC in the cefotaxime resistant isolates are warranted to explain their phenotype.
In this study, we observed the presence of type II T/A genes splA and splT in A. baumannii CC1. Type II T/A systems are usually plasmid encoded and mediate plasmid maintenance through the post-segregational killing of plasmidfree daughter cells (48). The splA and splT genes are so far unique to A. baumannii and are encoded on small, ca. ∼10 kb plasmids of the Rep-3 superfamily (49,50). There is a lack of knowledge regarding these plasmids, however, some harbor bla OXA−24 /bla OXA−40 or bla OXA−72 and are prevalent among carbapenem-resistant human clinical IC II isolates in Eastern Europe (49). The significance of the splA/T carrying plasmids identified among the CC1 isolates in this study remains to be investigated.
Finally, non-baumannii ACB complex species accounted for 13.8% of the animal, and 14.3% of the environmental isolates. These isolates were distinguished from A. baumannii strains by the lack of bla OXA−51 -like genes, the low prevalence of acquired antibiotic resistance genes and high rate of susceptibility to antimicrobial agents, in agreement with observations from human isolates (8,51).
In conclusion, this study provides a molecular and phenotypic analysis of ACB complex isolates obtained from animals admitted to a veterinary hospital in Switzerland during 2006-2017. Using established methods applied to isolates of human origin enabled the identification of clonal lineages and resistance determinants that occur globally among human isolates, including CC1 and CC25 A. baumannii. As opposed to A. baumannii CC1, CC25 isolates have infrequently been described in companion animals, but were prevalent among the isolates in this study. Contrasting to frequently occurring human clinical isolates worldwide, the veterinary ACB complex isolates in this study did not possess any known acquired carbapenemase genes. However, since A. baumannii, including CC25 isolates are emerging as bla OXA−23 carrying veterinary isolates in other countries, increased surveillance and targeted measures to prevent the dissemination of ACB complex strains are warranted.

AUTHOR CONTRIBUTIONS
RS designed the study. KZ, SP-S, and SS carried out the microbiological and molecular biological tests. SP-S, KZ, and MN-I analyzed and interpreted the data. MN-I drafted the manuscript. All authors read and approved the final manuscript.