Detection of the carbapenemase gene blaVIM-5 in members of the Pseudomonas putida group isolated from polluted Nigerian wetlands

There are increasing concerns about possible dissemination of clinically relevant antibiotic resistance genes, including genes encoding for carbapenemases in the environment. However, little is known about environmental distribution of antibiotic resistance in Africa. In this study, four polluted urban wetlands in Nigeria were investigated as potential reservoirs of carbapenem-resistant bacteria (CRB). CRB were isolated from the wetlands, characterized by Blue-Carba test, MIC determinations and whole genome sequencing (WGS). Nine of 65 bacterial isolates identified as members of the Pseudomonas putida group (P. plecoglossicida and P. guariconensis, respectively) harboured the metallo-beta-lactamase gene blaVIM-5. WGS revealed the blaVIM-5 in three novel Tn402-like class 1 integron structures containing the cassette arrays aadB|blaVIM-5|blaPSE-1, aadB|blaVIM-5|aadB|blaPSE-1, and blaVIM-5|aadB|tnpA|blaPSE-1|smr2|tnpA, respectively. Strains carrying the aadB|blaVIM-5|blaPSE-1 cassette also carried an identical integron without blaVIM-5. In addition, the strains harboured another Tn402-like class 1 integron carrying bcr2, several multidrug resistance efflux pumps, and at least one of ampC, aph(3”)-lb, aph(6)-ld, tetB, tetC, tetG, floR, and macAB. This is the first report of a carbapenemase gene in bacteria from environmental sources in Nigeria and the first report of blaVIM-5 in environmental bacteria isolates. This result underscores the role of the Nigerian environment as reservoir of bacteria carrying clinically relevant antibiotic resistance genes.

reported the detection of the carbapenemase gene bla  in clinical isolates of Proteus mirabilis and Pseudomonas aeruginosa from Nigeria, however, no information was provided about the genetic context of the genes. More importantly, very little is known about the genetic platforms linked to detected carbapenemase genes in Africa 28,29 . These are important omissions considering the widespread and uncontrolled use of beta-lactam antibiotics in human clinical therapy and food animal production in Nigeria 30 , and the ubiquitous release of untreated wastewater from several point and non-point sources into the aquatic ecosystem as a result of poor sanitation 31 . These predisposing factors for the proliferation of antibiotic resistance in environmental reservoirs make the investigation of Nigerian aquatic ecosystem as reservoir of carbapenemase-producing bacteria an urgent task.
The objective of this study was to investigate four polluted urban wetlands in Nigeria as potential reservoirs of bacteria harbouring genes encoding resistance to carbapenems. Isolated carbapenemase-producing bacteria species were subjected to WGS to understand the genetic platforms linked to detected carbapenemase genes. The results provide further insight into the global epidemiology of this important group of antibiotic resistance genes.

Results
CRB were recovered from urban Nigerian wetlands. Various colonies of bacteria showing resistance to meropenem were isolated on all agar types: Muller Hinton (MH), Eosine Methylene Blue (EMB), and Pseudomonas isolation (PI) agar used for bacteria isolation from all sediment samples analysed. A total of 65 isolates showing different colony morphologies were isolated from Awba (AW = 14) and Apete (AP = 22) wetlands in Ibadan, and from Abule Agege (AA = 14) and Ogbe Creek (OC = 15) in Lagos (Fig. 1). According to their partial 16S rRNA gene sequences (about 1,300 bp) the strains belonged to six different genera (relative frequencies in parenthesis), namely Pseudomonas (52.3%), Stenotrophomonas (15.3%), Cupriavidus (12.3%), Burkholderia (9.2%), Pandoraea (6.2%) and Ralstonia (4.6%). Carbapenemase activity was detected by the Blue-Carba test in 61 isolates (93.8%) with at least one representative of each genus among the carbapenemase-producing strains. The isolates negative in the Blue-Carba test belonged to the genera Cupriavidus (3 isolates from AP in Ibadan) and Ralstonia (1 isolate from OC). Members of the Enterobacteriaceae were not present among the cultured CRB as deduced by absence of lactose-fermenting colonies on EMB agar plates. However, other Enterobacteriaceae (E. coli, Enterobacter spp., Citrobacter spp.) with resistance to third generation cephalosporins, fluoroquinolones and sulphonamides were isolated from the same samples in an accompanying study (Adelowo et al., unpublished), which suggests that there were indeed only few, if any, cultivable carbapenem-resistant Enterobacteriaceae at the sites. bla VIM-5 was detected in isolates belonging to the Pseudomonas putida group. The metallobeta-lactamase (MBL) gene bla VIM-5 was the only carbapenemase gene detected of all carbapenemase genes (bla VIM , bla KPC/BIC , bla NDM , bla IMP , bla SPM , bla AIM , bla DIM ) screened for by PCR among the 61 carbapenemase-positive isolates. This gene was found in nine isolates (prevalence rate of 15%) from AW and AP (2 samples each). The gene shared 99% nucleotide sequence identity with bla VIM-5 on the class 1 Integron of Enterobacter cloacae SMART 28 (accession number LC169578). The search for the genetic basis of carbapenemase activity in the other 52 isolates is ongoing. In order to investigate the genetic background of the nine bla VIM-5 -positive strains, their genomes were sequenced. Genome characteristics as well as isolation dates and origins of the isolates are provided in Supplementary Table S1. The strains were identified via complete 16S rRNA gene sequence comparison and Multi Locus Sequence Analysis (MLSA) of 520 single copy core gene products as Pseudomonas plecoglossicida and Pseudomonas guariconensis (Fig. 2); members of the P. putida group 32,33 . Strains identified as P. plecoglossicida clustered into two groups (MR69, MR70, MR134 and MR135 from AW, and MR83, MR170 from AP). Whole genome-wide SNP counts ranged from 19 to 55 for the most closely related strains, showing that the strains are not identical (Fig. S1). bla VIM-5 was present on 3 different class 1 integrons structures in the isolates. Analysis of the genomes revealed the presence of bla VIM-5 on Tn402-like class 1 integrons with canonical 5′CS and 3′CS, defective transposition module, and three different gene cassette contents and arrangements (Fig. 3). All the bla VIM-5 -bearing integrons contained the aminoglycoside adenyltransferase gene aadB sharing 100% sequence identity with aadB from P. aeruginosa (accession number JF412714) 34 and the carbenicillin-hydrolysing class A beta-lactamase gene bla PSE-1 sharing 100% sequence identity with bla PSE-1 from P. aeruginosa RIVM-EMC2982 (accession number CP016955). In P. plecoglossicida strains MR69, MR70, MR134 and MR135 the integron harboured the cassettes aadB|bla VIM-5 |bla PSE-1 . In P. plecoglossicida strains MR83 and MR170 the integron contained a second aadB copy in the order aadB|bla VIM-5 |aadB|bla PSE-1 . In P. guariconensis strains MR119, MR144, and MR149 the integron carried two identical copies of tnpA, encoding for a transposase of the IS116/IS110/IS902 family with 99% sequence identity with transposase of P. aeruginosa (accession number EF614235) 35 , as well as smr2, encoding for a small multidrug resistance efflux protein sharing 75% sequence identity with smr2 from P. aeruginosa (accession number KC776907) in the order bla VIM-5 |aadB|tnpA|bla PSE-1 |smr2|tnpA. Based on sequence analysis of the integrase genes, the gene cassettes on the integrons were under the control of weak promoters PcW and a second inactive promoter P2. Weak promoters usually have strong integrase excision activity 36 .
In P. plecoglossicida strains MR69, MR70, MR134 and MR135, the bla VIM-5 -bearing integron was part of 491 kb-large chromosomal contigs. At both integron ends, defined by canonical 25 bp-long inverted repeats IRi and IRt, there were identical copies of an IS6 family transposase IS6100 followed by a second IRt copy. At the 5′-CS side of the integron there was another integron which lacked bla VIM-5 but was otherwise 100% identical with the bla VIM-5 -bearing one. At the Tn402-derivative side, the IS6-type transposase gene was adjacent to a 2.3 kb-long sequence with 100% nucleotide sequence identity to the adjacent region of class 1 integrons on the large plasmids pBM413 (from P. aeruginosa strain PA121617; accession number CP016215) and pSY153-MDR (from P. putida strain SY153, accession number KY883660). The 4 genes within these 2.3 kb region encode for hypothetical proteins conserved in Pseudomonas and some other Gammaproteobacteria. Following those genes was a 477 kb-large region wich had high synteny with a segment in the chromosome of P. putida KT2440 (accession number AE015451). For the other sequenced strains the genomic locus of the bla VIM-5 integrons could not be resolved and identical integrons without bla VIM-5 were not detected. However, none of the contigs appeared to be derived from plasmids. All sequenced strains also carried a Tn402-like class 1 integron with bcr2, encoding for bicyclomycin resistance protein as sole gene cassette.

Resistance phenotypes and additional resistance genes identified via WGS.
In addition to their various phylogenetic affiliations with the P. putida group, the reaction of the nine isolates to antibiotics used in determination of minimum inhibitory concentration (MIC) differed (    Ten additional antibiotic resistance genes were identified in the genomes of the strains conferring resistance to six classes of antibiotics ( Table 1). The aph(3′)-ib, aph(6′)-Id, macAB, tetC and ampC were common to all strains while the tetB, tetG,and floR were detected exclusively in P. guariconensis MR119, MR144 and MR149. A gyrA mutation (Thr83Ile) conferring fluoroquinolone resistance was present in all the P. plecoglossicida strains but not in the P. guariconensis isolates, consistent with the sensitivity of the latter strains to ciprofloxacin.

Discussion
This study reports the occurrence and genetic environment of bla VIM-5 in bacteria isolated from four Nigerian wetlands. Ultimately, this is no suprise given the widespread pollution of aquatic ecosystems in Nigeria. The detection of the bla VIM-5 gene in closely related members of the P. putida group suggests an establishment of these species as reservoir of bla VIM in the two wetland ecosystems. Members of the P. putida group are typically found in the environment but have also been reported as important opportunistic pathogens 37 and reservoir-hosts of metallo-beta-lactamases 38 .
Previously, reports of bla VIM-5 have only been limited to clinical bacterial isolates from the Asian and African continents being identified in clinical isolates of P. aeruginosa, K. pneumoniae, Serratia marcescens, Enterobacter cloacae and Proteus mirabilis from Turkey 27,39-41 , Thailand 27 , Nigeria 27 , India 42 and China (unpublished, GenBank Accession Number KM589496). This however, is the first report of bla VIM-5 in environmental bacterial isolates. There is no information available on the gene context of the bla VIM-5 detected in the Nigerian clinical isolates. However, much like we observed in this study, the bla VIM-5 in the clinical isolates reported in some of the aforementioned studies 39,41,42 and the unpublished Chinese isolate are located on class 1 integrons. Interestingly, in all the clinical isolates, the bla VIM-5 occurred as the first cassette on the integrons, while in six of our environmental isolates (MR69, MR70, MR83, MR134, MR135 and MR170), the gene occurred as the second cassette suggesting different evolutionary pathways for the bla VIM-5 -bearing class 1 integron in the clinical and environmental isolates.
The localisation of bla VIM-5 within three novel integron structures in two species of Pseudomonas recovered from two wetlands points to a high propensity for mobilisation of this gene to other hosts within and outside the wetland ecosystem. This propensity is underscored by the involvement of weak promoters PcW and inactive second promoters P2 in the control of this gene and the associated gene cassettes in all isolates. A previous study has reported an inverse correlation between promoter strength and excision activity 36 . Similar to this observation, the bla VIM-5 on the class 1 integron of E. cloacae SMART 28, a clinical isolate from Turkey is also under the control of a weak promoter PcW 41 . Further, the genome sequences of MR69, MR70, MR134 and MR135 revealed the presence of an additional integron which is otherwise 100% identical to the bla VIM-5 -bearing integron except for the absence of bla VIM-5 , suggesting that the acquisition of the bla VIM-5 cassette was a recent event. It also appears that the capturing of the bla VIM-5 cassette by the integrons of all sequenced strains occurred via three independent events judging from the localisation of the cassette in three novel integron structures.
Reports of carbapenemase genes in bacteria from environmental sources in Nigeria are not available for a comparative assessment of our results. However, the frequency of carbapenemase genes detection in this study is slightly higher than the prevalence rates of 5.5% and 10.2% reported in two of the hospital-based studies carried out in Nigeria 22,23 . Further, while we detected only bla VIM , these two Nigerian studies detected other carbapenemase genes in addition to bla VIM and the carriage of multiple carbapenemase genes among the investigated CRB. The three independent Nigerian studies are however similar in one respect: the low detection frequency of known carbapenemase genes among the CRB tested. This might be an indication of the presence of novel gene(s) or gene variants in these strains, or the mediation of carbapenem resistance by intrinsic carbapenemase genes [43][44][45] .
To conclude, we isolated bacteria carrying the MBL gene bla VIM-5 in members of the P. putida group in two polluted wetlands in southwestern Nigeria. The link between the study sites and the city's water supply system heightens the public health risk associated with this observation as this, together with the association of the gene with seemingly active mobile genetic elements, increases the chance of dissemination of the gene to the human population through the water-human route. There is therefore an urgent need for further studies aimed at assessing the actual distribution and persistence of CRB and corresponding carbapenemase genes in the Nigerian aquatic ecosystems.

Materials and Methods
Sampling sites and sample collection. Four wetlands in Ibadan and Lagos, located in south-western Nigeria (Fig. 1), were selected as sampling sites. In Ibadan, Awba (AW) wetland (07.4468°N, 03.8763°E) is within a university campus and receives untreated wastewater and sewage from at least five of the university's hostel facilities, the university's fish farm and the zoological garden. Apete (AP) wetland (07.4577°N, 03.8828°E) is located behind the campus of a polytechnic and receives domestic wastewater from a network of sources including hostel facilities of the polytechnic and seepages from a solid waste dump site. Water from AW drains into Awba Dam, which serves as the university's water reservoir. Awba Dam and AP drain into Eleyele Lake, which is the water source for about half of Ibadan's population of 3.5 million residents. In Lagos, Abule Agege (AA) (06.5145°N, 03.4002°E) is a large expanse of mangrove wetland located within a university campus. The wetland has a direct link to the Lagos Lagoon and is polluted from several non-point sources including untreated sewage and wastewater from a network of hostel facilities of the university and direct discharge of wastewater from freshwater aquaculture. Ogbe Creek (OC) (06.5135°N, 03.3937°E) runs through a mangrove swamp within the campus of the same university as AA. Its primary source of pollution is seepages from a solid waste dump site located upland about 300 meters from the creek. Triplicate sediment samples were collected monthly from October 2014 through January 2015 from the upper 1 cm portion of each wetland. The samples were pooled together and stored at −80 °C until processed for isolation of bacteria. Bacteria isolation. CRB were isolated from the sediment samples by selective enrichment on MH agar, EMB agar, and PI agar, all supplemented with 4 mg/L meropenem (Glentham Life Sciences, Corsham, UK) as described 12 , except that saline suspensions (25 g sediment in 225 mL 0.85% saline) of pooled triplicate samples were inoculated into tryptone soy broth (TSB) supplemented with meropenem (4 mg/L) and the enriched TSB cultures (200 µl each) were plated on selective EMB, PI and MH agar plates. Further, all incubations took place at 35 °C. All carbapenem-resistant isolates representing different colony morphotypes based on size, colour, surface texture and colony edge were selected from the plates, subcultured on selective MH plates and stored in 15% (w/v) glycerol at −80 °C.
Identification of carbapenem-resistant bacteria. Total genomic DNA was extracted (DNeasy blood and Tissue Kit, Qiagen) and used as template in PCR with universal primers 27F 46 and 1378R 47 targeting the bacterial 16 S rRNA gene. Amplicons were purified (GeneJet PCR Purification Kit, Thermo Fisher Scientific), sequenced from both ends (Macrogen Europe, Amsterdam) and obtained sequences were used in BLASTn searches (https://blast.ncbi.nlm.nih.gov/Blast) for phylogenetic assignment of the isolates.

WGS and analyses of blaVIM containing isolates.
Between 500-1000 ng genomic DNA of each isolate was sheared with a Covaris S220 sonication device (Covaris Inc.; Massachusetts, USA) with the following settings: 55 s, 175 W, 5% Duty factor, 200 cycles of burst, 55.5 μL input volume. Sequencing libraries were prepared by using the NEBNext ® Ultra ™ DNA Library Prep Kit for Illumina ® (New England Biolabs, Frankfurt, Germany) as per the manufacturer's instructions. The libraries were sequenced with an Illumina ® MiSeq machine using v3 chemistry and paired-end approaches with 76 cycles (strains MR69, MR70, MR83, MR119, MR144, MR170) and/or 301 cycles (strains MR69, MR83, MR119, MR134, MR135, MR149) per read. Raw sequences were subjected to adapter clipping and quality trimming using Trimmomatic 53 , and processed reads were assembled with SPAdes v3.6.2 54 . Assembly quality and taxonomic placement of the genome were assessed with CheckM v1.0.4 55 . Prokka v1.11 was used for automated annotations 56 . Antibiotic resistance gene markers were searched for manually and via the Comprehensive Antibiotic Resistance Database (CARD) 57 . In strains MR69, MR70, MR134 and MR135 the entire bla VIM -carrying integron as well as the linkages with the large chromosomal segment and the adjacent integron without bla VIM were Sanger sequenced. In the other strains PCR-mapping and Sanger sequencing were used for scaffolding of the gene cassettes.
Orthologs shared by the different isolates as well as selected references were determined using the bidirectional BLAST+ approach implemented in Proteinortho 5 58 . In order to determine the phylogenetic relationships between the isolates and references via Multi Locus Sequence Analysis (MLSA) with maximum resolution, the single copy core genome of all comparison genomes was determined, excluding all genes with duplicates or missing orthologs in any comparison genome. After alignment with MUSCLE 59 , the gene products were concatenated and un-alignable regions were filtered out using gblocks 60 . The remaining aligned conserved core genome product amino acid residues (187,324 positions) were subjected to phylogenetic clustering using the Neighbor Joining algorithm with 1000 bootstrap permutations. Whole genome SNP analysis was carried out in CLC Genomics Workbench 11.0.1.

Data Availability Statement
The 16S rRNA gene sequences of the isolated CRB are available in DDBJ/ENA/GenBank under accession numbers MG674312-MG674376. The WGS projects have been deposited at DDBJ/ENA/GenBank under the accession numbers PJCJ00000000 -PJCR00000000.