ABSTRACT
Antimicrobial resistance (AMR) is a global problem but information about the prevalence and mechanisms of resistance in sub-Saharan Africa are lacking. We determined the percentage of drug resistant isolates and resistance mechanisms in 307 Gram negative isolates randomly collected from south western Nigeria. Susceptibility testing revealed 78.1%, 92.2% and 52.6% of all isolates were resistant to fluoroquinolones, third generation cephalosporins and carbapenems respectively. There were more resistant isolates from the stools of uninfected patients than from specimens of patients with symptoms of infections. Only a small proportion of E. coli (10%) and Klebsiella (7%) isolates produced a carbapenemase. Whole genome sequencing of selected isolates identified the presence of globally disseminated clones. This depicts a crisis for the use of first line therapy in Nigerian patients, it is likely that Nigeria is playing a significant role in the spread of AMR due to her high population and mobility across the globe.
INTRODUCTION
Nigeria and other sub-Saharan African countries face an increasing number of healthcare associated infections caused by multi-drug resistant (MDR) Gram-negative bacteria[1, 2]. Pathogenic species have evolved resistance to multiple antimicrobial agents including the mainstays of treatment[3, 4]. This is of particular concern as there are few new antibiotics in development with activity against Gram-negative bacteria[5]. Whilst Gram-negative bacteria are often intrinsically more resistant to many antibiotics than Gram-positive species, drug resistance to the most clinically important antibiotics is largely mediated by genes which are transmitted on plasmids that can readily spread through bacterial populations[6]. Species belonging to the Enterobacteriaceae family are the most commonly isolated MDR bacteria causing healthcare associated infections globally including in sub-Saharan Africa[7]. These bacteria include extended-spectrum β-lactamase (ESBL) producing K. pneumoniae and E. coli which are associated with both hospital and community infections with very high mortality rates[8. 9].
Carbapenems have become a mainstay of therapy for the treatment of ESBL producing Gram-negative bacteria. This has led to an increase in carbapenem use for treatment of serious infections[10]. As a result, there has been a selective pressure for carbapenem resistance and carbapenem resistant strains have spread globally [11](11). Worryingly, carbapenem-resistant bacteria are often resistant to other classes of antibiotics including aminoglycosides, fluoroquinolones and other ß-lactams, with often colistin and tigecycline as the only effective drugs. Resistance to both these antibiotics can also easily evolve making them unreliable as ‘last resort’ therapies[12, 13].
Mechanisms of carbapenem resistance include production of carbapenemase enzymes and/or repression of porins to limit entry of these drugs into the bacterial cell. Carbapenemases belong to different enzyme families including the metallo-carbapenemases (including the NDM and VIM enzymes) or non-metallo-types (including the KPC and OXA-48 enzymes)[14]. Some of these enzymes are associated with a particular species, for example the ‘Klebsiella Pneumoniae Carbapenemase’ or ‘KPC’ enzymes are typically found in K. pneumoniae species[15]. Other enzymes, such as NDM are found in many species [16].
Nigeria is often referred to as the “Giant of Africa”, owing to its large population and economy with approximately 182 million inhabitants, by far the largest in Africa. The Nigerian population is highly mobile and over 70% of Nigerians are under the age of 50. The large size and high level of mobility of this population makes import and export of antibiotic resistant bacteria a real concern for both Nigeria, but also the wider global community. Gram-negative bacteria cause a significant number of infections in Nigerian hospitals and represent the majority of isolates from both wounds and urinary tract infections; these form the largest group of clinical specimens received in Nigerian clinical microbiology laboratories[1]. Carbapenem-resistant Gram-negative bacteria have become prevalent in many parts of the world including Nigeria and sub Saharan Africa. However, to our knowledge there are few data and no organized antimicrobial resistance (AMR) surveillance networks for Africa. Recently, we showed that carbapenem resistant bacteria are present in Nigeria. The details of the strains with this phenotype and the mechanisms of resistance have not been studied in detail [3, 4](3,4). The potential role of the Nigerian population in the global spread of antibiotic resistance is great but the local situation is not understood. We determined a retrospective analysis of percentage resistance and resistance mechanisms in clinical and commensal isolates of Gram negative bacteria.
METHODS
Sample sites and bacterial isolates
The majority of the Nigerian population is found in the southwest of the country; this is also where major transportation hubs are located. Gram negative bacterial isolates for this study were recovered from patients admitted to three large teaching hospitals located in three states of south western Nigeria from a range of clinical specimens with invasive and colonized infections (Table 1, and Figure 1). The Olabisi Onabanjo University Teaching Hospital (OOUTH) is a 185-bed tertiary health care center and major referral center for Ogun State. The University College hospital (UCH) is in Ibadan in Oyo state and has 850 beds. The Obafemi Awolowo University Teaching Hospitals Complex (OAUTHC) is a teaching hospital with 535 beds and is in Osun state.
In addition, isolates from stool samples sent for routine examination from patients but without infection were also collected. Isolates were non-duplicate and unbiased (i.e. no selective criteria beyond being a Gram negative bacterial species were applied) and were randomly collected from the hospital laboratories in 2011 and 2013. A total of 306 isolates were retained and no information about the antibiotic susceptibility of any isolates was used as an inclusion criterion. The isolates comprised E. coli, Klebsiella spp, Pseudomonas aeruginosa, Proteus spp and others (Serratia and Citrobacter spp,). Species assignments were confirmed for all isolates using standard biochemical tests and API 20E strips (BioMérieux, Basingstoke, UK) for Enterobacteriaceae.
Antibiotic susceptibility testing
Susceptibility of all isolates to a panel of antibiotic classes commonly used in these hospitals such as fluoroquinolones, third generation cephalosporins and carbapenem were determined by the disk diffusion method on Mueller–Hinton agar using antibiotic disks from Oxoid Ltd. (Basingstoke, UK) according to the recommendations of EUCAST and interpreted according to EUCAST breakpoints[17]. All susceptibility testing experiments included the control organisms E. coli NCTC 10418 and P. aeruginosa NCTC 10662.
Identification of carbapenemase production
The Enterobacteriaceae isolates were tested for production of a carbapenemase using the disc based ‘Carbapenemase detection set’ from Mast Group (Bootle, UK) and interpreted using the ‘carbapenemase detection set calculator’ tool as per the manufacturers guidelines.
Identification of known carbapenemase genes
PCR and sequencing were used to identify genes encoding various known beta-lactamases (including carbapenemases, NDM, VIM, KPC and GES). Primers used are shown in Table S1 having previously extracted DNA by crude boiling method.
RAPD PCR
A random amplified polymorphic DNA typing approach was used for each species as a rapid and inexpensive way to assess the diversity of strains within each population. Primers and conditions are given in Table S1.
Whole genome sequencing and bioinformatics
To characterise the strain types, plasmid content and nature of drug resistance genes present in the collection, 10 isolates (due to paucity of fund) were selected for whole genome sequencing based on their antimicrobial profiling, carbapenemase production, genotypes, clinical specimen and source and species. DNA was extracted with the QIAamp DNA Mini Kit according to manufacturer instruction. Paired-end Illumina sequencing was used to generate high-quality 250 bp reads. Assembly used Velvet [18] and contigs were re-ordered against relevant reference genomes using MAUVE[19]. Assemblies were annotated using RAST (http://rast.nmpdr.org/rast.cgi). Assemblies were used to search for plasmid content and to determine MLST types using the ‘PlasmidFinder’ and ‘MLST’ tools hosted at the Centre for Genomic Epidemiology (https://cge.cbs.dtu.dk/services/PlasmidFinder and http://cge.cbs.dtu.dk/services/MLST). The ‘Comprehensive Antibiotic Resistance Database’, CARD was searched to identify antibiotic resistance genes[20]. Specific assembly of plasmids was attempted using ‘plasmidSPAdes’ and plasmid content identified by plasmid network reconstruction using ‘PLACNET’ (http://castillo.dicom.unican.es/request/)[21]. When necessary, reads were mapped against assemblies using Bowtie [22] and visualized in Artemis[23].
RESULTS
Antimicrobial resistance in the 307 Nigerian isolates
The percentage of the entire collection of 306 isolates that were resistant to fluoroquinolones, third generation cephalosporins and carbapenems being 78.1%, 92.2% and 52.3%, respectively (Tables 2 and 3). This pattern of high numbers of clinical isolates being resistant to these classes of drug was very similar in all three study sites (fluoroquinolone resistance was seen in 75 - 83% of isolates, cephalosporin resistance in 90 - 100% of isolates and carbapenem resistance in 50 - 55% of isolates). Of concern was the observation that the percentage of isolates from stool of uninfected patients, i.e. those being carried as commensal bacteria that were resistant to third generation cephalosporins (100%) and carbapenems (69.1%) was higher than in isolates from patients being treated for an infection. This suggests that extremely high numbers of antibiotic resistant bacteria are prevalent in the community and that multidrug resistance is not restricted to isolates found in the hospital environment.
When stratified by species, E. coli were most commonly resistant to third generation cephalosporins (93.7%) and carbapenems (59.4%), followed by Pseudomonas species (where 91.7% of isolates were resistant to third generation cephalosporins and 51.7% of isolates were resistant to carbapenems) (Tables 2, 3 and 4). Of the 307 isolates, no species had 50% or more of isolates which were sensitive to all three classes of antibiotic. The isolates of Proteus (n=24) were least likely to be carbapenem resistant; nonetheless 48% of isolates were resistant to this class of drug.
The isolates in this study were collected in two different years, the percent of fluoroquinolone resistant isolates in 2011 and 2013 was very similar. However, between the two years the percentage of isolates resistant to cephalosporins fell from 97% to 87% and the percentage of carbapenem resistant isolates fell from 59% to 43% (Table 1).
Typing of isolates
RAPD PCR was used to type 54 of 306 isolates representing 18 each of E. coli, K. pneumoniae and P. aeruginosa. A wide variety of strains were present for each species with only a small number of repeated patterns observed e.g. Figure 2 where 13 RAPD patterns were identified from 18 isolates of E. coli demonstrating a lack of dominance by specific clones.
Carbapenemase production and identification of carbapenemase genes
As there was a high prevalence of carbapenem resistance in the isolates, the E. coli and Klebsiella strains were tested for production of carbapenemases using the ‘carbapenemase detection set’ from Mast Group (Bootle, UK). Only 6% of all isolates produced a carbapenemase (7% of Klebsiella and 10% of E. coli). Specific carbapenemase genes were amplified by PCR and genes verified by DNA sequencing alleles for all 306 isolates. In agreement with the phenotypic testing, only 19 of the 306 isolates (6.2%) carried a known carbapenemase gene. The PCR revealed the presence of variants of VIM (n=9), GES (n=10) and NDM (n=2) families. These genes were detected in K. pneumoniae (n=6), E. coli (n=4) and P. aeruginosa (n=9) isolates. Two isolates carried two carbapenemase genes (NDM and VIM and GES and VIM). The KPC, IMP or OXA-48 genes were not detected in any isolate.
As 51.3% (157 isolates) of the 306 isolates were resistant to carbapenems but most did not appear to produce a known carbapenemase the presence of other known resistance mechanisms was investigated. The CTX-M genes have been shown to be very common and important in Gram negative isolates among other extended-spectrum β-lactamases around the world, we therefore used primers specific for each of the CTX-M sub-groups were used to detect these genes. Of the 218 E. coli and K. pneumoniae isolates, 79.4% (173) contained a CTX-M allele; DNA sequencing of a random selection of 40 isolates revealed all to be CTX-M-15. None of these isolates demonstrated de-repression of efflux but all showed either complete loss or reduced production of outer membrane porins (data not shown).
Characterisation of antibiotic resistance mechanisms and strain types in representative isolates
To investigate the molecular basis of drug resistance ten isolates were chosen for whole genome sequencing. These included representative isolates of the most common species and resistance phenotypes present in the collection. The ten isolates included two K. pneumoniae, two E. coli, three P. aeruginosa, two P. mirabilis and one P. rettgeri isolate (Table 4). The choice of isolate was informed by susceptibility testing, year of isolation, site of isolation (both geographical and specimen type) and by results of random amplified polymorphic DNA (RAPD) typing. The sequencing identified some globally established strain types in circulation in Nigeria, notably K. pneumoniae ST11 and P. aeruginosa ST224.
Both the K. pneumoniae isolates belonged to ST11 and were from urine of different patients in 2011 from UCH. Both genome assemblies were essentially identical and both carried NDM-1 and CTX-M-15 (Table 4). In addition, both isolates also carried OXA-1 and SHV-11. Interestingly, there was direct evidence for the mobility of the NDM-1 gene in this strain. Whilst the gene encoding NDM-1 was detected by PCR using a boiled colony preparation as a template in both isolates (U52 and U37), in both genome assemblies created after sequencing isolated DNA preparations, it was only initially seen in the genome assembly for U52. Analysis of the genetic location of this gene showed it to be present in a context like that seen by others previously, in an operon with bleMBL and associated with trpF, dsbC and cutA (Figure 3). In isolate U37 this region was absent and no sequence reads mapped against the U52 reference (Figure S1) demonstrating likely mobility of this whole region as has been suggested previously [24]. Both IncFIB and FII plasmid replicons were present in both strains supporting a plasmidic context for blaNDM-1 (Figure 3). In addition to the beta-lactamase genes, both isolates also carried trimethoprim (dfrA12), macrolide (mphA) aminoglycoside (rmtF) chloramphenicol (cat) and sulphonamide (sul1) resistance genes. Consistent with fluoroquinolone resistance, mutations in gyrA were seen,
One of the P. aeruginosa belonged to ST244 and carried the mutant PDC-1 AmpC enzyme as well as genes that contribute to resistance to chloramphenicol (cmx, catB7), aminoglycosides (aph(3”)-I1, aph(6)-Id) and fosfomycin (fosA). The other two isolates were both members of ST233 and both carried PDC-3. These latter two isolates also carried VIM-2 and OXA-33, were of the same MLST type and both isolated from OOUTH although isolated two years apart. Reads from the F46 strain carrying VIM-2 were assembled using both Velvet and SPAdes (using the plasmidSPAdes option); both resulted in assemblies with the VIM-2 gene present on a contig of ~7000bp. When this sequence was compared with known sequences in Genbank using the BLAST algorithm a perfect match for an integron carrying VIM-2 was found (accession number KT768111.1). Figure 3 shows a plasmid network reconstruction and the genetic context of the VIM-2 genes in these two isolates.
The two E. coli strains sequenced belonged to ST226 and ST156. Neither carried known carbapenemase genes although both had multiple mutations within ampD suggesting de-repression of the chromosomal ampC gene. Both strains also carried TEM-1 and various other mobile resistance genes including genes conferring aminoglycoside resistance (aph(6)-Id, aph(3”)-I1). An IncFII plasmid replicon was present in isolate S46 (the ST226 isolate).
Two P. mirabilis strains were sequenced, isolate F10 carried two CMY genes; CMY-41 reported once previously in a Citrobacter freundii isolated from food in Egypt [25] and CMY-31 previously reported in Klebsiella and Salmonella[26. 27]. A Q1 plasmid replicon was present in F10. This isolate also carried two separate aminoglycoside resistance genes (aadA5, aph(3”)-I1), as well as chloramphenicol (catI), sulphonamide (sul1) and plasmidic quinolone resistance genes (qnrA1). P. mirabilis isolate F56 was found to carry a novel CMY enzyme with a single substitution (of glutamic acid for aspartic acid at codon 144) distinguishing this protein from CMY-48 isolated from C. freundii. Isolate F56 also carried a chloramphenicol acetyltansferase gene (catI) and three aminoglycoside resistance genes (aadA5, aph(3”)-I1 and aph(6)-Id).
The Providencia isolate (S39) sequenced carried an SRT-2 AmpC beta-lactamase variant; this has previously been described in Serratia marcescens [28]. No other beta-lactamase genes or plasmid replicons were detected in this isolate.
DISCUSSION
This study suggests that there is a very high prevalence of antibiotic resistance in Nigerian isolates of Gram-negative bacteria to three key classes of antibiotic. A high frequency of resistance to fluoroquinolones and cephalosporins have been seen in other areas of the world increasing the reliance on carbapenems for the treatment of infections caused by Gram negative bacteria. In this study over 50% of the Nigerian isolates in our collection were carbapenem-resistant; empiric use of these antibiotics for the treatment of serious infections is unlikely to be effective. Resistant isolates appear to be widely spread in the community and were not restricted to hospital patients. Isolates from stools of healthy individuals were more likely to be resistant to all three classes of antibiotic tested than those from clinical samples suggesting that the wider Nigerian population commonly carry resistant isolates including carbapenem resistant isolates at a high frequency. From our data, resistance to major antibiotics would appear to be the norm in Gram negative bacteria carried in the Nigerian population.
Characterisation of the mechanisms of carbapenem resistance in our collection of isolates showed that some well-known and globally disseminated carbapenemase genes are in circulation within Nigeria. These included NDM, VIM and GES enzymes. However, less than 10% of the isolates in the study carried a known carbapenemase (according to both molecular and phenotypic testing) and none carried KPC or OXA family carbapenemases. A recent report from Edo state, (in south Nigeria, further east from the locations in this study) has reported the existence of OXA family carbapenemases of OXA-48 and OXA-181 and NDM-1[29]. Carbapenem antibiotics are available in Nigeria but have historically not been widely used in hospital medicine as they have not been part of the common antibiotic formulary. In most Nigerian hospitals third generation cephalosporins, aminoglycosides and fluoroquinolones are the most prescribed antibiotics. There was essentially pan-resistance to the cephalosporins and fluoroquinolones in the isolates. The high level of phenotypic resistance to carbapenems in this collection could be caused by the carriage of currently carbapenemases that were not detectable by the methods used. However, we hypothesize that the very high frequency of carriage of ESBLs (~80% of isolates of Enterobacteriaceae contained CTX-M variants) and AmpC variants in combination with porin loss (in Pseudomonas isolates) selected by prolonged and heavy cephalosporin usage are the cause of carbapenem resistance in these isolates. A recent study described Enterobacteriaceae isolates from the USA which were carbapenem resistant without carriage of known carbapenemases) [30](30).
This study covered the South West of Nigeria, where the population density of the country is highest with approximately 50 million people. The study area included major population centers close to other major cities with a diverse population in terms of culture, race, religion and social standing. The major international transportation hubs of Nigeria are also in the South West of the country and over 15,000 international flights leave annually to over 30 different countries and over 8 million passengers fly through Nigeria annually [31]. International destinations are varied with Europe and the Middle East being most common followed by destinations in Asia with a smaller number of flights departing to North and South America [31].
Whilst local antibiotic use is likely to have made an impact on the incidence of antibiotic resistance in the collection of isolates, globally disseminated strain types and resistance genes were identified. This is highly relevant given the mobility of the Nigerian population and the implications for this mobility in global transfer of strains and genes. For example, a recent case report documents a Canadian visitor who suffered a lower leg fracture requiring surgical repair in Nigeria and was repatriated two months later with a wound infected with Klebsiella, Pseudomonas and E. coli isolates all carrying carbapenemases[32].
K. pneumoniae strains belonging to ST11 were detected; these have been associated with the carriage of CTX-M-15 and KPC enzymes, mainly in China. ST11 is also a single locus variant from ST258 which has been associated with the international dissemination of KPC enzymes on the pKPQIL plasmid[15]. ST258 isolates have also been recently associated with NDM carriage in India, Sweden and the United Kingdom[33]. In this study, E. coli ST226 was recovered from an uninfected patient; this strain type has been found circulating in highly resistant diarrhoeagenic E. coli in China. The other E. coli ST identified, ST156, has previously been found in Bangladesh[34], and in NDM-1 carrying clinical isolates of E. coli from the UK[35]. P. aeruginosa clone ST233 has been described as a dominant international ‘high-risk clone’ amongst metallo-β-lactamase-producing P. aeruginosa and two VIM-2 positive isolates were found in patients in this study. Isolates of this sequence type have been seen in the UK[16], and have also been reported previously as carrying VIM-2 or IMP-1 in Norway (in an isolate thought to be imported from Ghana)[36], Japan[37], and South Africa[38]. The other P. aeruginosa sequence type identified in this study (ST244) is a globally disseminated P. aeruginosa clone identified in several countries, including Poland[39], Brazil[40], Spain[41], South Korea[42], the Czech Republic[43], Greece[44], Russia[45], China [46] and Tanzania[47]. ST244 isolates have been found to carry various carbapenemases including IMP and VIM enzymes as well as extended-spectrum ß-lactamases, such as PER-1 and VEB-1[39, 48].
This study demonstrates that antibiotic resistance in Gram-negative bacteria in Nigeria is common place and compromises the effectiveness of the mainstays of broad spectrum empirical therapy. Perhaps most worryingly, this does not appear to be a problem restricted to hospital patients with resistance rates in commensal isolates being carried commensally equally high. The establishment of a reservoir of resistant strains and resistance genes has occurred in Nigeria and this reservoir is likely to be mobilised globally. Our data underpin the urgent requirements for enhanced surveillance of drug-resistance in sub-Saharan Africa and the need for interventions to minimise the selection and transmission of antibiotic resistant Gram-negative bacteria.
Funding
DO received support as a Newton International fellowship from the Royal Society.
Biographical Sketch
Dr David Ogbolu was born in Ibadan, Nigeria, Africa’s second largest city. After graduating and working as a Biomedical Scientist in 1997 with a specialism in Medical Microbiology he completed a masters degree and subsequently PhD studying mechanisms of antibiotic resistance in Nigerian bacteria. In 2011, David was awarded a Newton Fellowship from the Royal Society to continue his studies and further collaborations with colleagues in the UK. David is now a Senior Lecturer at Ladoke Akintola University of Technology.
Acknowledgements
We thank the staff within the laboratories of the study site hospitals for help in collection of the isolates used in this study.