Distribution of virulence-associated genes and antimicrobial susceptibility in clinical Acinetobacter baumannii isolates

Acinetobacter baumannii is undoubtedly one of the most clinically significant pathogens. The multidrug resistance and virulence potential of A. baumannii are responsible for hospital-acquired nosocomial infections. Unlike numerous investigations on the drug-resistant epidemiology of A. baumanni, virulence molecular epidemiology is less studied. Here, we collected 88 A. baumannii clinical isolates, tested their antimicrobial susceptibility to 10 commonly used antibiotics and analyzed the distribution of 9 selected virulence-associated genes, aims to investigate the primary characteristics of the virulence-associated genes that exist in clinically multidrug resistant (MDR) and non-MDR isolates of A. baumannii. The MIC results showed the resistance rates of ciprofloxacin (68.2%, 60/88), gentamicin (67.0%, 59/88), amikacin (58.0%, 51/88), tobramycin (58.0%, 51/88), doxycycline (67.0%, 59/88), meropenem (54.5%, 48/88) and imipenem (65.9%, 58/88) were all above 50%, except for levofloxacin (34.1%, 30/88), minocycline (1.1%, 1/88) and polymyxin B (0%, 0/88). The Pulsed field gel electrophoresis (PFGE) analysis revealed that the resistance rate of MDR A. baumannii isolates in the Epidemic group was predominant (79.5%, 44/58), but in the Sporadic group was only 6.7% (2/30). Further investigation on the distribution of virulence genes showed the virulence genes bap (95.5%), surA1 (92.0%), BasD (92.0%), paaE (88.6%), pld (87.5%), BauA (62.5%), omp33-36 (59.1%) and pglC (53.4%) were accounted for high proportion, except for traT (0%). Overall, our results revealed that MDR isolates predominated in the Epidemic A. baumannii isolates, and contained a very high proportion of virulence genes, which may lead to high risk, high pathogenicity and high treatment challenge.


INTRODUCTION
Acinetobacter spp. were glucose-non-fermentative, non-motile, non-fastidious, catalase-positive, oxidativenegative, and aerobic Gram-negative coccobacilli. Acinetobacter baumannii is the most clinically significant Acinetobacter species associated with hospital-acquired infections worldwide [1]. Although A. baumannii used to be considered as a low virulence pathogen, pneumonia has been the main manifestation of nosocomial infections caused by this pathogen, resulting in a significant impact on the mortality rate of patients [2]. This microorganism has also been identified as the etiological agent responsible for a wide range of other infections, including septicemia, meningitis, and more recently, severe and deadly cases of necrotizing fasciitis [3]. Research Paper mediate resistance to Gram-negative bacterial killing by serum [19].
Our previous surveys revealed that the antibiotic resistance among A. baumannii clinical isolates is very serious. [20,21]. And antimicrobial-resistant A. baumannii has very high potential to spread among ill patients in ICU (intensive care units), so it is crucial to adopt early recognition and timely implementation of appropriate infection control measures in preventing outbreaks. Owing to the lack of new antimicrobials in the pipeline for problematic MDR organisms, understanding the role and mechanism of virulence factors in A. baumannii pathogenesis constitute novel therapeutic targets for rational drug design. We conducted a prospective exploratory study by investigating the prevalence of A. baumannii within a general hospital in western China, then analyzing the distribution of 9 virulence genes, and trying to explore the primary characteristics of the virulence genes within clinically MDR and non-MDR isolates of A. baumannii.

Genotyping of A. baumannii clinical isolates
To determine the spread of A. baumannii within the hospital setting, all isolates were genotyped using PFGE (Pulsed Field Gel Electrophoresis). PFGE results were www.oncotarget.com analyzed by BioNumerics and pulsotype designation was based on isolates showing ≥80% relatedness. PFGE fingerprinting analysis revealed that 58 A. baumannii isolates (all ≥80% relatedness) were Epidemic group and the remaining 31 A. baumannii isolates (all <80% relatedness) were Sporadic group (Supplementary Figure 1).
The Epidemic group were classified into 13 major clusters from A to M, in which the Larger Clusters (cluster A, D, H and M, N ≥5) contains a total of 44.3%. Cluster A (19.3%, 17/88) was most frequently detected in the clinical isolates and was responsible for an epidemic between January and February in 2014, and in the Respiratory department (N = 9), ICU (N = 5), Internal Medicine Ward (N = 1), General surgery ward (N = 1) and Cardiovascular department (N = 1). Cluster D (12.5%, 11/88) was second most frequent that detected in the clinical isolates and was closely correlated in August (N  Table 1). Although a small proportion of these Larger Clusters were discretely distributed at different wards, most of the strains were found within were mainly concentrated in respiratory and ICU wards (Figure 1c).

Antimicrobial susceptibility of A. baumannii clinical isolates
To understand the drug-resistance of 88 A. baumannii clinical isolates, antimicrobial susceptibility testing was performed. The results showed that the resistance rates to levofloxacin, ciprofloxacin, gentamicin, amikacin, tobramycin, minocycline, doxycycline, meropenem, imipenem and polymyxin B accounted for 34 Table 1). The resistance rates of ciprofloxacin, gentamicin, amikacin, tobramycin, doxycycline, meropenem and imipenem were all above 50%, while those of levofloxacin, polymyxin B and minocycline were only 34.1%, 0% and 1.1%, respectively. Most importantly, in the Larger Clusters (cluster A, D, H and M), the rates of resistance to some antibiotics were almost over 80% (Table 1). From these results we can reasonably infer that minocycline and polymyxin B may be used as empiric treatment agents for A. baumannnii in China. Interestingly, the resistance rate of all the antibiotics in the Epidemic group was significantly higher than that of the Sporadic group, except for minocycline and polymyxin B (P value <0.05).
For cross-resistance profile in these isolates, we found that the prevalence of MDR A. baumannii isolates was 52.3% (46/88) in the present study. Combined with the genotype data of these 88 A. baumannii isolates, the resistance rate of MDR A. baumannii in the Sporadic group was only 6.7% (2/30) (P value <0.01, Table 1). On the contrary, MDR A. baumannii isolates almost predominated in the Epidemic group, and A. baumannii isolates in cluster A (94.1%, 16/17) and cluster H (100%, 6/6) were almost all MDR. Above all, the high proportion of MDR A. baumannii in the Epidemic group indicated that MDR A. baumannii is more likely to spread and cause infection in the hospital.
Associated with the drug-resistance of 88 A. baumannii clinical isolates, the results revealed that the distribution of each virulence associated gene has some unique characteristics. The rate of surA1 among aminoglycosides (tobramycin and amikacin), doxycycline and carbapenems (imipenem and meropenem) resistant isolates was significantly higher than that in homologous sensitive isolates. The rate of BauA among the vast majority of antibiotics resistant (levofloxacin, ciprofloxacin, gentamicin, tobramycin, amikacin, doxycycline, imipenem and meropenem) isolates was significantly higher than that of homologous sensitive isolates. The rate of pld among doxycycline, aminoglycosides and carbapenem resistant isolates was significantly higher than that of homologous sensitive isolates. The rate of paaE among ciprofloxacin, gentamicin, tobramycin, amikacin, doxycycline, imipenem and meropenem resistant isolates was significantly higher than that of homologous sensitive isolates. But what's interesting is that the rate of omp33-36 among levofloxacin sensitive isolates was significantly higher than that of levofloxacin resistance isolates. In summary, these relationships are statistically significant (P value <0.05) ( Table 4), and the antibiotic-resistant A. baumannii carries more virulence genes. Further analysis in cross-resistance profile found that, the carriage ratio of surA1, BauA, pld and paaE in MDR A. baumannii isolates (100.0%, 78.3%, 95.7%, 97.8%) was significantly higher (P value <0.05) than in non-MDR A. baumannii isolates (83.3%, 45.2%, 78.6%, 78.6%), and the carriage ratio of BasD in Epidemic isolates (94.5%) was significantly higher (P value <0.05) than in Sporadic isolates (86.9%) ( Table 3).

DISCUSSION
Acinetobacter spp. was initially considered as an opportunistic human pathogen of low virulence with minimal significance. However, during the last two decades, the increasing ubiquity and intensity of mechanical ventilation, central venous and urinary catheterization, and antibacterial therapy have caused a surge in the frequency and severity of Acinetobacter infections [22][23][24]. A. baumannii is endowed with an open pan genome, which means it can easily acquire new function. The persistence of A. baumannii in the hospital setting, together with the strong selection pressure imposed by the use of antimicrobials in the    Table 2. clinical practice, has promoted the evolution of clinical A. baumannii towards drug resistance isolates and the formation of some virulence factors which facilitate the survival of A. baumannii in the hospital [25]. Drug-resistant A. baumannii is more likely to survive in hospital environment, causing an outbreak in the hospital [26,27]. The treatment of A. baumannii infections is becoming a serious problem because of its strong antibiotic resistance [28]. Especially, surveillance by the CHINET project demonstrated that resistance of A. baumannii strains to carbapenems increased from 31 to 66.7% between 2005 and 2014 [29]. The current data showed that the resistance rate of imipenem (65.9%, Table 1) was higher than our previous report (60.9%) [10]. Although the resistance rate is lower than in Turkey (94.9%) [30] and Iran (82%) [31], it is still in a rising tendency. This suggests that carbapenems can't be habitually used in A. baumannii infectious treatment.
Here, we found that the MDR A. baumannii in the Epidemic group (44/58, 75.9%) was higher than that of Sporadic group (2/30, 6.7%) (P value < 0.05, Table 1). In addition, the Larger Clusters A, D, H and M in the Epidemic group are all substantially composed of MDR A. baumannii, and showed resistance to most of the antibiotics. This result may due to the clinical abuse of antibiotics, and resistant A. baumanii was screened to survive, and then gradually developed into MDR A. baumannii. So, their persistent selection and ongoing transmission in hospital may result in epidemic outbreak [32,33].
Traditional thinking was that antibiotic resistance caused a metabolic cost to the bacterium and hence the "anti-virulence" factor [34]. However, the White paper reported that a lung and blood clinical isolate (A. baumannii HUMC1) is not only an extreme drugresistant isolate (almost resistant to all antibiotics, except colistin), but also a hyper-virulent isolate [35]. Meanwhile, some researchers have also suggested that MDR A. baumannii can survive proliferation even after the host treated with antibiotics, and then producing a variety of metabolites that upset the physiological balance of the host as a virulence effect [13,36]. Accordingly, the clinical outcome of A. baumannii infections is jointly influenced by antimicrobial resistance and virulence factor.
The infection symptoms caused by A. baumannii are closely associated with its virulence factor. But the virulence genes distribution in clinical A. baumannii were rarely reported, except for bap, omp33-36 and traT [37,38]. This work examined the distribution of 9 virulence genes in clinically collected A. baumannii isolates. Our results showed that the existence rate of each virulence gene in the A. baumannii isolates was more than 50%, except traT (Table 3). Furthermore, the virulence genes carriage rate of A. baumannii in MDR was significantly higher than that of non-MDR A. baumannii (P< 0.05, Table  3), such as surA1, BauA, pld and paaE. This suggests that the drug resistant A. baumannii isolates appear to have greater potentials for toxicity.
Biofilm is an important virulence factor of A. baumannii. Loehfelm et al. suggested the development and  [14,64] traT F: GGTGTGGTGCGATGAGCACAG R: CACGGTTCAGCCATCCCTGAG Serum-resistance-associated [65] thickness of the mature biofilm structure and intercellular adhesion were closely associated with Bap family proteins (coded gene bap) [39]. The bap genes were detected with very high frequency in clinical A. baumannii from Iran (92%) [40], Republic of Korea (100%) [37] and USA (84%) [41]. This work also showed that A. baumannii isolates frequently possess bap genes (95.5%). These results indicate that most A. baumannii may have the biofilm formation ability. Baumannii acinetobactin utilization (BauA) is an outer membrane protein, acting out the siderophore-ferric complex receptor [42]. Whole protein includes 2 domains comprising a cork domain at N terminal of the protein and a transmembrane barrel at the C terminal [43]. This work showed the rate of virulence gene BauA was significantly higher in drug-resistant A. baumannii (P<0.05, Table 4), and A. baumannii isolates with BauA gene seems to show a wide range of drug resistance. So, we speculate that the transmembrane barrel may be associated with drug resistance in A. baumannii.
A. baumannii phenylacetate degradation pathway was a hybrid pathway [12], and a key step in this pathway is performed by the paaE gene [44]. This study indicates that the rates of paaE were higher in the ciprofloxacin, tobramycin, amikacin, doxycycline or imipenem resistant A. baumannii isolates than in the corresponding sensitive isolates (P< 0.05, Table 4). We speculate these antibiotics may be metabolized as substrates during the metabolism of phenylacetic.
Some recent studies have reported that the virulence gene omp33-36 contributes to resistance to carbapenems in A. baumannii [45][46][47]. But in our work, there was no significant difference in the frequency of omp33-36 gene carried between carbapenems-resistant and -sensitive A. baumannii (Table 4). In contrast, our work showed that A. baumannii, which was found to be levofloxacin-sensitive, appears to carry more virulence gene omp33-36, and only a few A. baumannii isolates developed resistance to levofloxacin. We think levofloxacin is an effective drug for the treatment of A. baumannii infection.
A recent study reported that traT (80%) was found high frequency in CR-A. baumannii isolates in Iran [38]. However, this study showed that none of the 88 A. baumannii clinical isolates carried the traT gene, which was similar to the report of Brasil, 2004 [48]. This indicates that traT gene frequently existing in CR-A. baumannii is not a universal phenomenon, for the moment only finding in Iran.
In summary, our findings showed that the prevalence of MDR A. baumannii isolates collected from different wards was 52.3%, and the MDR A. baumannii in the Epidemic group was significant higher than that of Sporadic group. Virulence associated genes detected in this study accounted for a high proportion among the 88 A. baumannii clinical isolates, especially in the MDR A. baumannii isolates, which means high risk, high pathogenicity and high treatment challenge. At the same time, our work indicates that some potential links between virulence genes and drug resistance. This work strongly suggests that hospitals need to strengthen infection control to prevent A. baumannii from causing an outbreak of infection.

Bacterial isolates and growth conditions
The 88 nonrecurring clinical isolates of A. baumannii used in this study were collected from patients with nosocomial infections during the period from April 2013 to October 2015, at the First Affiliated Teaching Hospitals of Chengdu Medical College in Sichuan province, China. All isolates were collected as part of routine care, data were collected, identified and handled anonymously. The strains were stored in glycerol at -80°C and bacteria were grown on MacConkey Agar (MCA plates) (Oxoid, England) before tested. Isolates were identified by VITEK 2 (bioMe'rieux, Marcy-l'E'toile, France) automated microbiology system. Besides, all isolates were further verified by detecting the existence of the rpoB and sequencing the 16S-23S rRNA gene spacer region [10].

Pulsed-field gel electrophoresis testing
All the 88 A. baumannii isolates were genotyped by using Pulsed-Field Gel Electrophoresis (PFGE) analysis as previously described [49]. A. baumannii isolates were grown on Mueller-Hinton agar medium plates overnight at 37 °C and bacterial genomic DNAs were prepared. After the genomic DNA was digested with ApaI (TaKaRa, Dalian, China), the DNA fragments were separated by electrophoresis in 1% gold agarose (Lonza) in 0.5 × Tris -borate -EDTA buffer with a CHEF apparatus (CHEF Mapper XA, Bio-Rad, USA). The conditions included 14°C and 6V/cm with alternating pulses at a 120° angle with a 5-35s pulse time gradient for a total of 22 h. The same method was also used to extract the Salmonella enterica serotype Braenderup H9812 genomic DNAs, which was used as the size marker after being digested by XbaI (TaKaRa, Dalian, China) [50]. The similarity examination of DNA patterns analyzed with BioNumerics 7.0 (Applied Math, USA) was based on Tenover criteria and Dice coefficient, and 80% relatedness was used as the threshold to distinguish the pulsotype [51].

Antimicrobial susceptibility testing
The 88 A. baumannii strains were tested for their susceptibilities to 10 antibiotics: levofloxacin, ciprofloxacin, gentamicin, amikacin, tobramycin, minocycline, doxycycline, meropenem, imipenem and polymyxin B. The minimal inhibitory concentrations (MICs) were determined by agar dilution method and interpreted as previously described in the guidelines according to the 2017 Clinical and Laboratory Standards www.oncotarget.com Institute (CLSI) [52]. Escherichia coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853 were both used as quality control strains. The MICs value used to indicate antibiotic resistance are ≥4 μg/ml for ciprofloxacin and polymyxin B in accordance with CLSI, ≥8 μg/ml for levofloxacin, meropenem and imipenem, ≥16 μg/ml for gentamicin, tobramycin, minocycline and doxycycline, and ≥64 μg/ml for amikacin.

Definition of multidrug resistance (MDR)
Base on researchers' definition of MDR among A. baumannii in many published works [53], we defined MDR A. baumannii as containing resistance to any combination of 3 or more of the 4 drug classes commonly used to treat Gram negative infections: fluoroquinolones, aminoglycosides, carbapenems and tetracyclines. Resistance to an antibiotic class was defined as resistance to all drugs representative of that class in this panel.
PCR amplification and DNA sequencing for virulence genes detecting 9 virulence genes as listed in Table 2 were detected by PCR to investigate the distribution in the 88 A. baumannii strains. The primers used were also listed in Table 2 and PCR was carried out according to references in the same table. The PCR products were sequenced by TSINGKE (Chengdu, China) and then confirmed by sequence alignment to the corresponding genes of A. baumannii with NCBI Nucleotide Blast (http://blast.ncbi. nlm.nih.gov/Blast.cgi)

Author contributions
C.L and Y.C have contributed equally to this work. R.W and X.J conceived and designed the study. C.L, Y.C, Y.X, Y.L, L.W, Z.M and S.L performed the experiments. C.L, Y.C, R.W and X.J analyzed the data and wrote the manuscript. All authors reviewed and approved the final version of the manuscript.

CONFLICTS OF INTEREST
The authors declare that there are no competing interests regarding the publication of this paper.