blaVIM- and blaOXA-mediated carbapenem resistance among Acinetobacter baumannii and Pseudomonas aeruginosa isolates from the Mulago hospital intensive care unit in Kampala, Uganda

Background Between January 2015 and July 2017, we investigated the frequency of carbapenem resistant Acinetobacter baumannii (CRAB) and carbapenem resistant Pseudomonas aeruginosa (CRPA) at the Mulago Hospital intensive care unit (ICU) in Kampala, Uganda. Carbapenemase production and carbapenemase gene carriage among CRAB and CRPA were determined; mobility potential of carbapenemase genes via horizontal gene transfer processes was also studied. Methods Clinical specimens from 9269 patients were processed for isolation of CRAB and CRPA. Drug susceptibility testing was performed with the disk diffusion method. Carriage of carbapenemase genes and class 1 integrons was determined by PCR. Conjugation experiments that involved blaVIM positive CRAB/CRPA (donors) and sodium azide resistant Escherichia coli J53 (recipient) were performed. Results The 9269 specimens processed yielded 1077 and 488 isolates of Acinetobacter baumannii and Pseudomonas aeruginosa, respectively. Of these, 2.7% (29/1077) and 7.4% (36/488) were confirmed to be CRAB and CRPA respectively, but 46 were available for analysis (21 CRAB and 25 CRPA). Majority of specimens yielding CRAB and CRPA were from the ICU (78%) while 20 and 2% were from the ENT (Ear Nose & Throat) Department and the Burns Unit, respectively. Carbapenemase assays performed with the MHT assay showed that 40 and 33% of CRPA and CRAB isolates respectively, were carbapenemase producers. Also, 72 and 48% of CRPA and CRAB isolates respectively, were metallo-beta-lactamase producers. All the carbapenemase producing isolates were multidrug resistant but susceptible to colistin. blaVIM was the most prevalent carbapenemase gene, and it was detected in all CRAB and CRPA isolates while blaOXA-23 and blaOXA-24 were detected in 29 and 24% of CRAB isolates, respectively. Co-carriage of blaOXA-23 and blaOXA-24 occurred in 14% of CRAB isolates. Moreover, 63% of the study isolates carried class 1 integrons; of these 31% successfully transferred blaVIM to E. coli J53. Conclusions CRAB and CRPA prevalence at the Mulago Hospital ICU is relatively low but carbapenemase genes especially blaVIM and blaOXA-23 are prevalent among them. This requires strengthening of infection control practices to curb selection and transmission of these strains in the hospital.

The hospital environment, mostly the intensive care units (ICUs), is a hotspot of antimicrobial resistance and source of MDR bacterial infections [7], particularly infections involving Acinetobacter species and Pseudomonas aeruginosa. The aim of this study was to determine the prevalence of carbapanem resistant Acinetobacter baumannii (CRAB) and carbapenem resistant Pseudomonas aeruginosa (CRPA) in clinical specimens from the Mulago Hospital ICU. Also, carriage of carbapenemase encoding genes and class 1 integrons was investigated, as well as mobility potential of the genetic elements among CRPA and CRAB isolates.

Study design and setting
This was a cross sectional study conducted at Mulago National Referral Hospital in Kampala, Uganda. Culturing and other laboratory procedures were performed at the laboratories of the Departments of Medical Microbiology and Immunology & Molecular Biology, Makerere University College of Health Sciences.

Samples
The bacteria investigated were isolated from clinical specimens referred by physicians for diagnostic testing (mainly culturing and identification of bacterial pathogens) in the period between 2015 and 2017. During this period, the Clinical Microbiology Laboratory processed clinical specimens from a total of 9269 patients for isolation of CRAB and CRPA. The specimens included blood, tracheal aspirates, ear swabs, catheter tips, sputum, pus swabs, throat swabs etc. Majority (78%) of the specimens yielding CRAB and CRPA isolates were from the ICU while 20% were from the Ear Nose & Throat (ENT) department. The Burns unit contributed the remaining 2% of the isolates.

Culturing, identification and drug susceptibility testing
Culturing and identification of isolates to species level was performed as described previously [1]. Briefly, presumptive identification of the isolates was based on colony morphology, Gram staining characteristics and biochemical properties. Biochemical tests included triple sugar iron (TSI), Sulphur indole motility test (SIM), as well as citrate, urease and oxidase tests. Pseudomonas aeruginosa colonies were identified by their spreading pattern, serrated edges, fruity sweet-grape smell and bright green color. In addition, a negative TSI, positive catalase and oxidase tests and growth at 42°C were also used to identify Pseudomonas aeruginosa. Acinetobacter isolates were identified based on negative oxidase and glucose fermentation tests, negative catalase and motility tests and their inability to grow under anaerobic conditions. Further, to identify Acinetobacter baumannii, PCR-detection of the bla OXA-51 gene intrinsic to this species was performed [1,6,8].

Carbapenemase assays
Phenotypic screening for carbapenemase activity was done using the Modified Hodge Test (MHT). Additionally, combination disc tests, imipenem-EDTA and boronic acid inhibition tests, were used to screen for MBL and KPC production respectively, as described previously [1,9,11]. In the MHT assay, a 1:10 dilution of the indicator strain (Klebsiella pneumoniae ATCC 700603 or E. coli ATCC 25922) was made by diluting 0.5 ml of culture (0.5 McFarland) to 5 ml with sterile normal saline. Then, a lawn culture (1,10 dilution) was streaked onto a MHA plate using a sterile swab. A 10 μg meropenem disk was placed in the center of the test area on the MHA plate, and test isolates streaked in a straight line from the edge of the disk to the edge of the plate. Klebsiella pneumoniae ATCC BAA-1705 and Klebsiella pneumoniae ATCC BAA-1706 were used as positive and negative controls, respectively [1,[12][13][14]. Furthermore, isolates that were found to be non-susceptible to imipenem or meropenem were also screened for MBL activity. In the MBL assay, overnight liquid cultures of the test isolates (adjusted to a 0.5 McFarland) were streaked on a MHA plate. Two discs of imipenem (10 μg) were placed 15 mm apart (center-to-center); of the two discs, one was impregnated with 10 μl of 0.5 M EDTA to achieve a disc content of 1.5 mg EDTA. Following this, incubation was done at 37°C, overnight. An increase in inhibition zone diameter by ≥5 mm in the EDTA-supplemented disc was interpreted as positive for MBL production [1]. Isolates that were positive were also screened for KPC production. In this test, a 10 μg meropenem disc and another 10 μg meropenem disc containing 3-AminoPhenyl boronic Acid (3-APBA) (300 μg/ml) (Tokyo Chemical Co. Ltd., Japan) were placed 20 mm apart, center-tocenter. An increase in inhibition zone diameter by ≥5 mm around the combination disc compared to the meropenem disc alone was considered positive for KPC production [11].

Conjugation and identification of transconjugants
Conjugation experiments were performed as described previously [18,19] using CRAB/CRPA as donors and E. coli J53 (F− met pro Azi r ) which is resistant to sodium azide, as a recipient. Additional donor characteristics included co-carriage of class 1 integrons and bla VIM gene. CRBA/CRPA (donor) and E. coli J53 (recipient) were mixed in a ratio of 10:1 (donor:recipient) in Luria Bertani broth and the mixture incubated at 37°C overnight. To select for transconjugants, serial dilutions of the cultures were plated on MacConkey agar containing ceftazidime (4 μg/ml) and sodium azide (100 μg/ml), which medium is toxic to donor cells. Confirmation of transfer was determined by PCR as described above to detect bla VIM among transconjugants.

Quality control
Well-characterized carbapenemase producing clinical strains of Klebsiella pneumoniae (Nr.8) and Pseudomonas aeruginosa from our collection [1,9] were used as positive controls in molecular assays. For each gene that was PCR-amplified, PCR products were randomly selected and sequenced, and sequences confirmed through BLAST-searching at the National Centre for Biotechnology Information (https://blast.ncbi.nlm.nih. gov/Blast.cgi) as previously described [1]. 'No template' PCR controls and negative control PCRs with template DNA extracted from carbapenemase gene negative strains (K. pneumoniae DSMZ 9377, E. coli ATCC 25922 and P. aeruginosa ATCC 27853) were also included [1].

Results
A total of 9269 specimens from various wards at Mulago Hospital yielded 1077 and 488 isolates of Acinetobacter baumannii and Pseudomonas aeruginosa, respectively. Altogether, 2.7% (29/1077) and 7.4% (36/488) isolates were confirmed to be CRAB and CRPA respectively, Table 1. Acinetobacter species were detected in virtually all the tracheal aspirate samples and most of these samples yielded poly-microbial isolates that included Klebsiella pneumoniae and Escherichia coli.
Prevalence of carbapenemase genes and class 1 integrons bla VIM was the most prevalent carbapenemase gene and it was detected in all CRPA and CRAB isolates, Table 3. Of note, the other MBL encoding genes i.e. bla IMP , bla SPM, and bla NDM , were not detected; as well, carriage of the OXA-type carbapenemase genes was not detected in CRPA. Also, in agreement with activity assays, none of our isolates was positive for bla KPC carriage. We however found the OXA-type carbapenemase genes to be prevalent among CRAB i.e. 29% & 24% for bla OXA-23 & bla OXA-24 , respectively. Co-carriage of bla OXA-23 and bla OXA-24 was seen in 14% of the CRAB isolates. All CRAB isolates carried the bla OXA-51 gene while none carried bla OXA-58 . Class 1 integrons of sizes 500 bp to 1000 bp were detected in 63% (29/46) of the isolates: 64% (16/25) & 62% (13/21) CRPA & CRAB respectively, Table 3. Furthermore, the classification of Acinetobacter baumanii based on bla OXA-51 gene sequences [20,21] showed that 12, eight and one CRAB isolate in this study belonged to pan-European International Clones II, I & III, respectively.
All the bla OXA-23 -positive isolates were detected in samples from the ICU; these were tracheal aspirates and sputum samples -tracheal aspirates provided 83% of bla OXA-23 positive CRAB while sputum samples yielded 17%. Likewise, 80% of bla OXA-24 positive CRAB isolates were detected in tracheal aspirates from the ICU. All isolates with co-carriage of bla OXA-23 and bla OXA-24 genes were detected in tracheal aspirates from the ICU. Only 20% of the bla OXA-24 positive CRAB isolates were from the ENT ward.

Drug susceptibility profiles of CRAB and CRPA isolates
bla VIM -plus-bla OXA positive CRAB exhibited high levels of resistance to commonly used antibiotics especially ceftazidime (100%), gentamicin (88%), ciprofloxacin  (88%), cefepime (75%), piperacillin/tazobactam (63%), and amikacin (50%), Table 4. Most of the CRAB and CRPA isolates were MDR (resistance to three or more classes of antimicrobials) however, all were susceptible to colistin. Furthermore, the susceptibility profiles of isolates carrying a single OXA-type gene (either bla OXA-23 or bla OXA-24 ) were not different from profiles of isolates that carried two genes concurrently. Similarly, bla VIM -positive CRPA isolates showed high level resistance to gentamicin (100%), ceftazidime (89%), amikacin (72%), cefepime (72%), piperacillin/tazobactam (56%) and ciprofloxacin (50%); 94% of the isolates were MDR but susceptible to colistin. As well, all CRAB and CRPA isolates that were positive for both bla VIM and class 1 integrons were highly resistant to the tested antibiotics but the susceptibility profiles were not different from isolates that did not carry class 1 integrons, Table 5.

Transfer experiments
To demonstrate the potential for mobility of carbapenemase encoding genes in the hospital environment, conjugation experiments were successfully performed. Donors were CRPA and CRAB isolates that were positive for both bla VIM and class 1 integrons while the recipient was sodium azide resistant E. coli strain J53. PCR assays showed that 25 and 39% of CRPA and CRAB donors respectively, successfully transferred the bla VIM genetic element to E. coli, Additional file 1: Figure S1.

Discussion
Antimicrobial drug resistance (AMR) is a well-recognized global problem and high rates of resistance to first-line antibiotics among Gram-negative bacteria have been reported in Uganda [22]. As such, routine surveillance studies are necessary to monitor antibiotic resistance trends in low-income countries where antibiotic selection pressure is likely to be high.
This study has revealed that the prevalence of CRAB and CRPA in clinical specimens from Mulago Hospital remains relatively low and comparable to recently reported rates for carbapenem resistant Gram negative bacteria in Uganda [1,9,23]. Importantly, almost all the CRAB and CRPA isolates were from the ICU. Furthermore, bla VIM and bla OXA-23 were the most prevalent carbapenemase genes with the former being detected in all CRAB and CRPA isolates. As such, bla VIM and bla OXA-23 are the genes mediating carbapenem resistance in CRAB and CRPA in the Mulago Hospital ICU. Again, these findings are consistent with a recent study at Mulago Hospital and other settings in developing countries where similar studies have been done [1-3, 9, 24]. In addition to the bla VIM gene, we also detected bla OXA-23 , bla OXA-24 and bla OXA-51 genes in CRAB. Unlike bla OXA-23 -and bla OXA-24 -like genes, bla OXA-51 naturally occurs in Acinetobacter baumannii [1,6,8]. Furthermore, while in the previous study at Mulago [1] the detection of bla OXA-58 -like gene in CRAB was reported, it was not detected in this study. However, this observation is in agreement with studies by Minandri et al. 2011 [25] and Wu et al. 2015 [26] who discussed the transition from carriage of bla OXA-58 -like to bla OXA-23 -like and bla OXA-24 -like genes. Furthermore, through conjugation experiments we demonstrated the potential for horizontal gene transfer of the bla VIM gene. Transfer rates in our study were low but in agreement with another study [27] which reported similar transfer rates. We have therefore provided evidence suggestive of mobility for carbapenem resistance genes in the Mulago  Although CRAB and CRPA that simultaneously carried bla VIM and class 1 integron were highly resistant to antibiotics, the susceptibility profiles were not different from isolates that did not carry class 1 integrons. Hence, carriage of integrons did not affect the resistance profiles implying that integrons in these isolates either did not carry drug resistance genes or they carried cassettes encoding resistance phenotypes that were not screened for. Furthermore, the susceptibility profiles of CRAB carrying a single OXA-type gene (either bla OXA-23 or bla OXA-24 ) were not different from profiles of isolates that carried two genes concurrently. As OXA genes generally confer a β-lactam resistance phenotype, the additive effect might be better detected with minimum inhibition concentrations (MICs), not agar diffusion approaches we used.
Additionally, although the CRAB and CRPA isolates we investigated were not associated with outbreaks, their detection in clinical specimens is of public health significance. Indeed, we observed high resistance to common antibiotics but this has been reported before [1,2,26]. Of note, carriage of bla OXA-23 alone, or bla OXA-23 plus bla OXA-24 in bacteria has been associated with high resistance to carbapenems and to other classes of antimicrobials [25,26]. While bla VIM carriage in this setting has been reported before [3,28,29], a high prevalence of bla VIM is worrying as this genetic element renders carbapenems (imipenem & meropenem) ineffective [3,[29][30][31], narrowing treatment options to colistin & polymyxin B that have been associated with drastic side effects e.g. nephrotoxicity, neurotoxicity, etc. [3,[29][30][31]. As such, a narrow window remains before total antimicrobial resistance is seen in these bacterial isolates as they continue to be subjected to antibiotic pressure in the hospital [32][33][34][35]. Unlike in the previous study [1] we did not detect the three MBL genes bla SPM , bla NDM and bla IMP that have been associated with high resistance to carbapenems especially to imipenem [36,37]. Further, the high prevalence of bla VIM in CRAB and CRPA in this study is suggestive of the role of the accessory genome in evolution of carbapenem resistance in our setting. bla VIM is characterized as an acquired MBL gene translocated between bacteria via class 1 integrons, in horizontal gene transfer mechanisms mostly involving conjugative plasmids [28,[38][39][40].
There were certain limitations in this study. We screened isolates for colistin susceptibility using the agar diffusion method instead of the recommended broth microdilution or the Etest methods. Further studies should take this into consideration given that poor agar dilution characteristics can limit the predictive accuracy of the disk diffusion method.