The Resistome, Mobilome, Virulome and Phylogenomics of Multidrug-Resistant Escherichia coli Clinical Isolates from Pretoria, South Africa

Antibiotic-resistant Escherichia coli is a common occurrence in food, clinical, community and environmental settings worldwide. The resistome, mobilome, virulome and phylogenomics of 20 multidrug resistant (MDR) clinical E. coli isolates collected in 2013 from Pretoria, South Africa, were characterised. The isolates were all extended-spectrum β-lactamase producers, harbouring CTX-M (n = 16; 80%), TEM-1B (n = 10; 50%) and OXA (n = 12, 60%) β-lactamases alongside genes mediating resistance to fluoroquinolones, aminoglycosides, tetracyclines etc. Most resistance determinants were found on contigs containing IncF plasmid replicons and bracketed by composite transposons (Tn3), diverse ISs and class 1 integrons (In13, In54, In369, and In467). Gene cassettes such as blaOXA, dfrA5-psp-aadA2-cmlA1a-aadA1-qac and estX3-psp-aadA2-cmlA1a-aadA1a-qac were encompassed by Tn3 and ISs; several isolates had same or highly similar genomic antibiotic resistance islands. ST131 (n = 10), ST617 (n = 2) and singletons of ST10, ST73, ST95, ST410, ST648, ST665, ST744 and ST998 clones were phylogenetically related to clinical (human and animal) strains from Egypt, Kenya, Niger, Nigeria, Tanzania, and UK. A rich repertoire of virulence genes, including iss, gad and iha were identified. MDR E. coli harbouring chromosomal and plasmid-borne resistance genes in same and multiple clones exist in South Africa, which is very worrying for clinical epidemiology and infectious diseases management.

identification and antimicrobial susceptibility testing. The isolates were isolated after growing them on blood agar and subsequently on Mueller-Hinton agar at 37 °C for 24 hours. They were then screened for ESBL production using cefoxitin, ceftazidime, and clavulanic acid antibiotic discs on Mueller-Hinton agar plates according to already reported protocols 6 . The species and antimicrobial sensitivity of the isolates were determined with the MicroScan WalkWay7465 (Beckman Coulter California USA) using antibiotic panels involving 32 antibiotics: penicillins, cephems, carbapenems, polymyxins, fluoroquinolones, aminoglycosides, tetracyclines, tigecycline, sulphamethoxazole-trimethoprim, nitrofurantoin and fosfomycin (Table S1). The MICs were interpreted according to the CLSI guidelines (CLSI M100 29 th Ed., 2019) 29 , except for antibiotics such as colistin and tigecycline for which EUCAST (2019) breakpoints were used due to the absence of CLSI breakpoints [30][31][32] . The identification of the species was confirmed by the NCBI's ANI (average nucleotide identity) database.
Analysis of whole genome sequence data. Whole-genome sequencing (WGS) was performed on the Ion torrent (Covaris, USA) and the Illumina Miseq (San Diego, USA) systems using already described methods 21,33,34 . Briefly, the genomic DNA of the isolates were extracted and sheared to 200-bp libraries; 280-bp (for Ion Proton) and 350 bp (for Illumina Miseq) fragments were selected using 2% agarose gels and Pippen prep (Sage Science, Beverly, MA, USA). Individual libraries were pooled and sequenced on the Ion Proton (ThermoFisher, Waltham, MA, USA) or Illumina Miseq (San Diego, USA). The generated raw reads were de novo assembled using the SPAdes assembler.
Assembled sequences were annotated using ResFinder (https://cge.cbs.dtu.dk/services/ResFinder/) at default threshold ID (90%) and minimum length (60%) values to identify resistance genes. MLST 2.0 (https://cge.cbs.dtu. dk/services/MLST/) was used to identify the sequence types of the isolates. The INTEGRALL database (http:// integrall.bio.ua.pt/) was used to identify integrons and gene cassettes within the genomic sequences. NCBI's PGAP 35 , ISFinder (https://isfinder.biotoul.fr/) and the RAST SEEDVIEWER (http://rast.nmpdr.org/seedviewer. cgi) were used to annotate and identify the insertion sequences (ISs) and transposons bracketing the resistance genes. PlasmidFinder 2.1 (https://cge.cbs.dtu.dk/services/PlasmidFinder/) and pMLST 2.0 (https://cge.cbs.dtu. dk/services/pMLST/) were used to identify the plasmid replicons and incompatibility groups on the various contigs. The sequences have been deposited at GenBank under the Bioproject PRJNA355910, with the individual accession numbers delineated in Table 1, S1 and S2. Mutations in gyrAB, parCE, mgrB, pmrAB, and phoPQ were manually curated from NCBI's BLAST by comparing the genomes of the isolates to wild type reference sequences 14 . phylogenomic analysis. Whole-genome sequences of E. coli isolates curated at the PATRIC website (https://www.patricbrc.org/), between 2013 and 2018, including South African isolates, were downloaded and used alongside this study's isolates for the whole-genome phylogeny analysis to ensure a current epidemiological and evolutionary analysis (Dataset 1). The phylogeny of the E. coli isolates was characterised using Parsnp (https://harvest.readthedocs.io/en/latest/content/parsnp.html) 36 and edited with Figtree (http://tree.bio.ed.ac. uk/software/figtree/). Isolates of the same clade are highlighted with the same colour whilst those of the same countries have the same label (strain name) colours. The source of the strains viz., animal, environment and human, are shown with distinct colours and annotations. BacWGSTdb was used to type and associate the isolates to international clones, their resistance genes and clinical data 37 . The resistome of strains of close phylogenetic www.nature.com/scientificreports www.nature.com/scientificreports/

Results
patient demographics and isolate characteristics. The 20 isolates were obtained from eight males and 12 females (Table 1) within the ages of 3 and 72, from mainly blood (n = 5) and urine (n = 11). The isolates were all obtained from Kalafong (n = 9) and Steve Biko/Tshwane Academic (n = 11) tertiary academic hospitals, all based in Pretoria, South Africa. Antibiotic susceptibility. All the isolates were resistant to the penicillins (amoxicillin and piperacillin), and 3 rd and 4 th generation cephalosporins, but were susceptible to amoxicillin-clavulanate/sulbactam, piperacillin-tazobactam, cefotaxime-clavulanate, ceftazidime-clavulanate, cephamycin (cefoxitin), and carbapenems. For non-β-lactam antibiotics, almost all isolates were resistant to gentamicin, tobramycin, ciprofloxacin, levofloxacin, nalidixic, tetracycline, and sulphamethoxazole-trimethoprim (SXT), but were susceptible to amikacin and norfloxacin (they were however resistant to norfloxacin according to EUCAST breakpoints) (Supplementary Table S1). Only eight isolates were resistant to minocycline whilst seven were resistant to chloramphenicol and four were resistant to colistin. Two isolates were categorically defined as resistant to tigecycline; however, the MICs (minimum inhibitory concentrations) of the remaining ten isolates (≤1 mg/L) are such that they could be either resistant (>1 mg/L) or intermediate resistant (0.5 mg/L). All but two of the isolates were susceptible to fosfomycin and nitrofurantoin, which are important urinary tract infection (UTI) antibiotics 23,38 . There was categorical agreement between the CLSI and EUCAST MIC breakpoint interpretations for all the isolates and all antibiotics except for norfloxacin (all isolates) and ceftazidime in only E003 (Supplementary Table S1).   15 and bla TEM-1B genes. The isolates harboured all three β-lactamase genes. The bla CMY-2 gene was identified in two isolates, E003 and E057 (Table 1), which were however susceptible to cefoxitin. In all these isolates, resistance to penicillins and cephems, except cefoxitin, was observed. However, the isolates became susceptible to the penicillins and cephalosporins in the presence of β-lactamase inhibitors (clavulanate, sulbactam and tazobactam). No carbapenemase genes were found, and carbapenem resistance was absent in all the strains (Table 1 and S1). www.nature.com/scientificreports www.nature.com/scientificreports/ We identified only two types of plasmid-mediated quinolone resistance (PMQR) genes in the isolates ( Table 1). The aac(6')Ib-cr gene was present in 12/20 (60%) isolates, whilst the OqxAB gene was present in only one isolate, E035. Qnr genes were absent. We identified mutations in the gyrA, gyrB, parC and parE quinolone resistance-determining region (QRDR) genes in all the isolates. gyrA had two mutations (A828S, D678E), gyrB had five (E219K, A618T, R206L, E185D and S492N), parC had seven (E62K, A620V, A192V, A471G, D475E and Q481H) and parE had three mutations (V136I, S458A, T172A) ( Table 2). One isolate (E053) had mutations in all the genes. The most frequent QRDR mutations were I529L in parE, S801, E84V, A192V, A471G, and Q841H in parC, S83L, D87N, and E678D in gyrA, as well as D185E and A618T in gyrB (Table 2). Sixteen (80%) isolates had mutations in all four QRDR genes. Except for E003, for which no PMQR gene was found, all the strains were resistant to at least three of the fluoroquinolones.
Ten tetracycline-resistant isolates had the tet(A) gene, and four isolates had the tet(B) gene. Among the two tetracycline-susceptible strains, E011 and E053, E053 harboured a tet(A) gene. The chloramphenicol acetylating transferase, cat, gene was detected in several isolates: 13/20 (65%) had catB3 and four had catA1. The cml and floR efflux genes were identified in four and one isolate respectively, albeit only seven isolates expressed chloramphenicol resistance (Table 1 & S1). The mph(A) macrolide phosphotransferase gene was found alongside β-lactamases, sul, cat, str and aminoglycoside modifying enzyme genes in five isolates, whilst the rifampicin ADP ribosylating transferase aar2 gene was only identified in one isolate. In one multidrug-resistant isolate, we detected the chloramphenicol, macrolide and rifampicin resistance genes, together with those conferring resistance to aminoglycosides, fluoroquinolones, β-lactamases, tetracycline and trimethoprim/sulfamethoxazole. Two isolates had increased colistin MICs, although only E035 had a chromosomal mutation (H6R) in the phoQ gene, but no mutation was detected in isolate E053. Other mutations were detected in pmrB (H2R, E123D, D283G and V351I) and pmrA (T31S, I128N and G144S) genes (Table 3). No plasmid-mediated colistin gene mutations were detected in either isolate ( Table 1). The molecular mechanisms underlying tigecycline resistance in the two isolates remains unknown; no tet(X) resistance genes were found in the genomes 31,39 . The resistome of all the isolates are found in Dataset 2.
As shown in Table 5, most of the resistance genes were bracketed by either class 1 integrons, ISs and transposons or by all three. Composite and Tn3 transposons as well as IS6 insertion sequences were most common, with ISEc9 being commonly found with bla CTX-M genes. The resistomes and mobilomes in these isolates were found to have between 98% and 100% sequence identity and length coverage with already deposited genome sequences at Genbank; the most common among these were E. xiangfangensis WCHEX045001 chromosome (CP043382.1) and E. coli GZ04-0086 plasmid pCTXM-GZ04 (CP042337.1) ( Table 5).
Sequence types and phylogenomics. The E. coli isolates were multiclonal, with ST131 (n = 10), ST617 (n = 2) and singletons of ST10, ST73, ST95, ST410, ST648, ST665, ST744 and ST998 being identified (Tables 1, 4 & 5). The two ST617 strains (E019 and E020) virtually have the same plasmid replicon types, resistome, virulome, integron types, genomic features and patient characteristics (66-year old female from Tshwane Academic hospital). A slightly similar observation was also made for isolates E005 and E009 (ST131). However, these patterns were not observed among the other strains of ST131 that had different patient demographics (different sexes, ages and hospitals) (Figs 1, 2) The phylogenomic analyses of the isolates showed that they were more closely aligned to strains from Tanzania and Egypt than to any other country; none of the isolates were phylogenetically related to any strain from South Africa. Specifically, E019 and E0120 (ST617), described above to have the same resistome, mobilome, virulome, genomic and demographic features, as well as K091 (ST998) and E035 (ST10), were found to be on the same clade/node whilst E005 and E009, found to also have very similar genomic characteristics, were distantly placed on different branches on the tree (Figs 1, 2; Dataset 1). E062, E056 and E058, as well as E005 and E011, all ST131, clustered together on one branch and clade. However, ST131 strains such as E063 and E060 were distant from other ST131 strains; other STs such as E003 (ST744), E057 (ST665) and K011 (ST410) were not closely related to any strains on the tree (Fig. 1A; Supplementary Dataset 1).
K075 was on the same clonal node as CFSAN061771 (ST1485) from Egypt and same clade/branch as CFSAN061765 (ST1722:bla CMY-2 , bla EC , bla OXA-244 ), also from Egypt. E040 and RDK06_554 (Tanzania: aph(3")-Ib, aph (6)  www.nature.com/scientificreports www.nature.com/scientificreports/ (Fig. 2), including E. coli 05:H4 strain ECO0291 (with E057) and E. coli 021:H52 strain ECO0336 (with E003), all of phylogroup A. Evidently, the STs and resistance genes between this study's isolates and those from Egypt and Tanzania were different. (Table 1) were recorded in all the isolates combined, with E053 (n = 11 virulence genes) and K075 (n = 12 virulence genes) having the most repertoire of virulence genes; K011, E019 and E020 had the least (n = 2 virulence genes). Virulome similarity could be seen between isolates belonging to the same clone than between those of different clones. E035 (ST10) and E040 (ST95) had unique set of virulence genes, whilst K075 and E053 had very diverse set of virulence genes. The commonest virulence genes among the strains were iss (n = 18 isolates), gad (n = 17 isolates) and iha (n = 12 isolates), with katP, cba, aaiC, ireA, pic, mcm, air, and eilA occurring in single isolates ( Fig. 3; Supplementary File 2). A specimen source-virulome association comparison was made ( Fig. 4; Supplementary file 2) and there was little evidence to suggest that strains from blood had more virulence genes than those from urine, albeit the strain with the most virulence genes was from blood. eilA, air, and lpfA were only found in a single strain (K075) from blood.

Discussion
In this study, 20 clinical E. coli isolates showed an extensive repertoire of resistance genes bracketed by composite Tn3 transposons, ISs and class 1 integrons on contigs containing mainly IncF plasmid replicons in multiclonal and same clone strains. The strains can be rightly defined as MDR strains due to their phenotypic resistance to penicillins, cephalosporins, aztreonam, fluoroquinolones, aminoglycoside, tetracycline and SXT. The ESBL phenotypes of the strains, as confirmed by the disc synergy test, was confirmed by the presence of ESBL genes (bla CTX-M , bla OXA , and bla TEM-1B ) and the susceptibility of the strains to the β-lactamase inhibitors viz., clavulanic acid, sulbactam and tazobactam, when combined with either the penicillins or cephalosporins. A major observation was the cefoxitin susceptibility of E003 and E057, both of which harboured bla CMY-2 within the same genetic context of IS1380 ISEc9:bla CMY-2 ::sugE (the reverse orientation was found in E003) that was of closest nucleotide identity with Salmonella Derby strain 116 plasmid (MK191846.1). Whereas the genetic context of these bla CMY-2 suggests that they might have been acquired horizontally, the host strains could not show phenotypic resistance to cefoxitin as expected of strains with acquired bla CMY-2 40,41 . Moreover, AmpC β-lactamases such as bla CMY-2 are not expected to be inhibited by β-lactamase inhibitors (clavulanate, sulbactam and tazobactam) as was observed in E003 and E057, which were susceptible to β-lactam/β-lactamase inhibitor combinations. These observations strongly suggest that the bla CMY-2 genes in these two isolates might not have been expressed or were not active in the isolates 40,41 . Besides these two isolates, other resistance discrepancies were observed between the phenomes and genomes of other isolates that harboured resistance genes but did not express resistance phenotypically. Examples include the susceptibility of E063, E035 and K091 strains to tobramycin although they had aadA, aph(3')-IIa, or strA/B genes. Other such discrepancies, already stated in the results above, can be found by studying Supplementary Tables 1 and 2.     www.nature.com/scientificreports www.nature.com/scientificreports/ The bla CTX-M-15 gene, surrounded by composite transposons mostly including ISEc9, was found in almost all the E. coli isolates, which is higher than that reported by two other studies from South Africa where 45% and 59% of isolates in Port Elizabeth and Cape Town, respectively, possessed this gene. Furthermore, the co-existence of bla OXA-10 and bla TEM genes is consistent with observations in local studies 17,42 as well as in studies from China 42,43 . Similarly, the co-occurrence of aac(6')Ib-cr and bla  are consistent with observations made in isolates from China and the USA 44,45 . ISEc9 and IncF plasmids have been implicated in the mobilization and dissemination of bla CTX-M-15 globally 20,21,45,46 ; the IncF plasmids mobilizing bla CTX-M-15 also co-harboured aac(6')Ib-cr, bla OXA-10 and bla TEM genes within E. coli ST131 46 . Our findings support this global data and shows that these resistance genes are both clonally and horizontally disseminated.

Sample
Unfortunately, the individual effects of the various mutations found in the QRDR of parCE and gyrAB as well as in mgrB, pmrAB, and phoPQ, in conferring resistance to fluoroquinolones and colistin could not ascertained in this study. This limitation makes it difficult to determine which resistance mechanism underlies the observed resistance, particularly as PMQR genes were also found in some of the isolates. The mutations observed in the parCE and gyrAB genes were not found in isolates that were reported from Durban, South Africa, except for R206L and E185D (in gyrB) and S458A in parE 14 . Similar studies in Portugal and India reported similar QRDR mutations 47,48 (Table 2). We did not find qnr genes in the isolates, although a similar work in Durban reported several qnr variants 14,49,50 . However, the presence of OqxAB efflux genes have been reported in bacterial isolates from South Africa 14 .
tet genes are commonly reported from South Africa, Africa and worldwide on chromosomes or plasmids alongside bla CTX-M-15, aac(6′)Ib-cr, bla OXA-10 and bla TEM 22,[51][52][53] ; specifically, the tet(A/B) genes in these isolates were mostly bracketed by Tn3 and composite transposons as well as by ISs. Despite chloramphenicol rarely being used to treat E. coli infections, several isolates contained the cat gene, indicating co-selection and/or transmission of chloramphenicol resistance genes by other antibiotics. Notably, all catB3 genes were found as catB3:bla OXA-1 :a ac(6')-Ib-cr5 within composite transposons, suggesting that the use of fluoroquinolones, aminoglycosides and β-lactams could co-select and drive the dissemination of this resistance gene even in the absence of phenicols. Moreover, cat genes have been shown to be co-transmitted on plasmids with aad and sul genes through horizontal transmission and not natural selection 54 . Notably, sul and aad genes were identified in 9/11 (82%) of the isolates in which the cat gene was also identified.
The sul and dfr gene cassettes identified in the isolates were previously reported in a study done in Enterobacteriaceae in Tunisia 55 . However, these genes have only been reported in Streptococcus pneumoniae isolates from South Africa 56 , whilst the rare sul3 gene was only recently reported in clinical isolates in Tanzania 57 . The genetic environment of sul1 and sul2 genes were mostly consistent, with sul1 being mostly associated with QacEΔ1 and aadA genes (aadA1/5:QacEΔ1:sul1) and sul2 being always found with aph genes (aph(6)-Id:aph(3′)-Ib:sul2) within composite transposons or ISs. The association of these genes on the same MGEs might explain the co-resistance to SXT, chloramphenicol and macrolides in these strains as these antibiotics are not prescribed for treatment of infections caused by E. coli in South Africa.
Although mph(A), which is responsible for macrolide resistance, is not clinically important in Enterobacteriaceae, they can be transferred to medically important Gram-positive bacteria for which macrolides are indicated 58 . The mph(A) gene were normally found alongside tetR and IS6 (IS6::tetR::mph(A)). The simultaneous presence of cat, mph(A) and floR genes in clinical E. coli isolates in South Africa has not been previously described, although similar findings were reported from Nigerian poultry and American calves 59,60 .
The frequency of class 1 integrons in these strains (95%) was much higher than isolates reported from Tunisia (64%), India (61%) and Korea (54%) [61][62][63][64] . The dominance of the dfrA17 and aadA5 cassettes, conferring resistance to trimethoprim and streptomycin, respectively, and their association with class 1 integrons has been described in several countries worldwide but not South Africa 26,62,65,66 . The isolates contained seven different cassette arrays, more than previously described from any single location 65 . We also identified the β-lactamase bla OXA-10 cassette in one isolate, which was previously described in a South China study 43 . We found no cassettes encoding bla CTX-M and bla TEM , confirming that these genes are rarely spread by integrons. The integrons carrying the dfrA5-psp-aadA2-cmlA1a-aadA1-qac (E040) and estX3-psp-aadA2-cmlA1a-aadA1a-qac (K075) cassette arrays www.nature.com/scientificreports www.nature.com/scientificreports/ are the first to be described in Africa. There is a close similarity between the arrays of these two integrons, although their host strains were of different STs, and their resistomes and mobilomes were different (Tables 1 and 5); further analysis would be required to clarify this similarity. African strains used in this study were basically related to strains from Tanzania and Egypt. Strains of same and different clones clustered together in many instances, with isolates of the same clone only clustering together in a few instances. (B) The strains clustered according to sequence types, with ST131 and ST617 strains being on the same branches. However, ST10 and ST998 were also found on the same branch, showing the higher resolution of whole-genome MLST over conventional MLST.
www.nature.com/scientificreports www.nature.com/scientificreports/ As shown in Table 5, all the class 1 integrons were bracketed by ISs and composite transposons that can mobilize these resistance genes from plasmids to chromosomes and vice versa. The synteny and localization of several resistance genes within these MGEs suggest the presence of resistance genomic islands within the genomes. However, the transferability of these genes and MGEs were not experimentally ascertained, although the horizontal transmission of these resistance genes through MGEs within and across species cannot be entirely ruled out. Further, the close sequence identity of the contigs bearing the resistance genes and MGEs with already known plasmids and chromosomes confirms the location of these contigs on either chromosomes or plasmids.
The resolving power of WGS over MLST (multi-locus sequence typing) is clearly observed in Figs 1 and 2 in that strains of the same STs were found on different branches and nodes. The demographics, virulome, resistome, mobilome and genomic features of E005 and E009 as well as of E019 and E020 suggest that they might have originated from the same patients. Although 10 isolates were of ST131, only three (E062, E056 and E058) and two (E095 and E011) groups were phylogenetically related on the same clade, with the others clustering with other strains of different STs. These differences were further seen between this study's isolates and those from Egypt and Tanzania, with which they closely clustered (Fig. 1). As seen in Fig. 2, they also varied in ST from those from the UK. These seeming discrepancies is due to the lower resolving power of MLST, which only uses seven house-keeping genes to type bacteria. www.nature.com/scientificreports www.nature.com/scientificreports/ In addition, the difference in resistance genes between the closely clustered strains from South Africa and Egypt and Tanzania, further shows that not all these resistance genes were chromosomal. This is because the phylogenetic tree was drawn with the core genomes of the individual isolates without their accessory genomes (plasmids) 37 . The absence of any close relationship between the isolates in this study and other South African strains demonstrates the absence of an intra-country dissemination of E. coli; however, further investigations are necessary to confirm this assertion. Interestingly, the isolates were closely related to strains from Egypt, Tanzania and UK, suggesting the possible exchange of people between South Africa and these countries. Therefore, it is necessary for public health officials to screen patients coming from other countries (for medical tourism) for resistance genes to reduce the exchange of resistance genes across borders 3 .
The diversity and multiplicity of virulence genes found in these isolates, that were mainly obtained from blood and urine, is quite concerning. This is more so as the isolates were also MDR. Evidently, the small sample size of strains made it impossible to obtain a better association between specimen source and the virulome as suggested by Irenge et al. 67 . However, it is worthy of consideration, that isolates from the urine would also need virulence genes to initiate infection; hence, it is not surprising that the virulome of urine and blood isolates were . Frequency distribution of virulence (virulome) genes found per Escherichia coli isolate. Several virulence genes were found in the isolates, ranging from two to 24. Some isolates had more virulence genes diversity than others, with some virulence genes being found in only an isolate from blood (K075). . Association between the virulome and specimen source of each Escherichia coli isolate. The isolate with the highest virulome composition and diversity was from blood (K075) followed by one from urine (E053). Thus, there is little to suggest that isolates from blood had more virulence genes than those from urine as shown in the chart.
www.nature.com/scientificreports www.nature.com/scientificreports/ comparable. Thus, it would be better to rather compare clinical with environmental strains in terms of virulence. The diversity and complexity of the virulome found in this study is quite comparable to that reported recently from the DRC 67 , although more virulence genes were reported in DRC than was observed herein.
The findings of this study present a worrying presence of a rich repertoire of resistance and virulence genes as well as MGEs in clonal and multiclonal E. coli strains within Pretoria. Although no carbapenemase, mcr and tet(X3/4) genes respectively mediating resistance to carbapenems, colistin and tigecycline were found, the chromosomally mediated colistin and tigecycline resistance in some of the strains is a cause for concern. We recommend additional molecular surveillance studies to provide statistically stronger data to inform pertinent interventions to contain these MDR strains from further dissemination. ethical approval. Ethical approval was provided by the Human Research Ethics Committee of the University of Witwatersrand (Ref M1710100). All protocols and consent forms were executed according to the agreed ethical approval terms and conditions. All clinical samples were obtained from a reference laboratory and not directly from patients, who agreed to our using their specimens for this research. The guidelines stated by the Declaration of Helsinki for involving human participants were followed in the study.