Whole genome sequencing reveals high clonal diversity of Escherichia coli isolated from patients in a tertiary care hospital in Moshi, Tanzania

Escherichia coli are Gram-negative bacterial commensals that naturally inhabit the human gastrointestinal tract (GIT). Through horizontal transfer and other mechanisms, commensal E. coli regularly acquire virulence, pathogenicity and multi-drug resistance properties from pathogenic E. coli. Consequently, E. coli is an important causative agent for a range of nosocomial and opportunistic infections including neonatal meningitis, diarrhoea, septicaemia, urinary tract and wound infections [1‐5]. The global emergence and spread of multidrug resistant (MDR) E. coli in both community and as nosocomial infections, as well as in animals, warrants public health concerns [6‐9]. Furthermore, virulent E. coli strains share resistance, virulence and pathogenic factors with avirulent or less virulent strains, enabling them to cause overlapping pathogenesis beyond their classical capacities [10]. Several E. coli outbreaks leading to serious health, social and economic impacts have been reported in high income countries (HICs) including the Netherlands [11], the UK [12], Norway, and Georgia [10].

Africa is highly burdened by diarrhoea and urinary tract infections (UTIs) that are E. coli related. For instance, in rural Kenya it was reported that 64.5% of UTI or bacteriuria was E. coli related [13]. In north-western Tanzania, 41.2 and 70% of UTI or bacteriuria was E. coli related among under-fives and febrile children, respectively [14, 15]. Similarly, in north-eastern part of Tanzania, E. coli accounted for 56.1% of all UTI cases among children [5] and E. coli accounted for 79 and 75% of UTI in non-malaria febrile children and adults, respectively [16]. In Dar es Salaam, two previous reports showed that, E. coli accounted for 51.1% of UTI in a survey done in 2010 [17] whereas as high as 64.0% of hospital- and community-acquired UTI was accounted for by E. coli in 2004 [18].

Finally, we recently showed that in Moshi, north-eastern Tanzania, E. coli was one of the most common bacterial pathogen isolates from a range of clinical manifestations [1]. Advanced molecular diagnostics such as whole genome sequencing (WGS) have revealed the emergence of a fatal diarrhoea-causing E. coli strain that combines virulence factors (VFs) from two E. coli strains [4]. The VFs are important properties that do enable an infectious agent to effectively and efficiently establish itself on or within its host by enhancing its potential to cause harm or disease. Some of the factors include those for host cell adherence, immune evasion, toxins and protease production for disrupting host cell pathways and scavenging minerals like iron in order to increase their survival. Certain E. coli strains have VFs that been linked to a serious outbreaks in China in 1999, where it caused 177 deaths [19], and in Germany in 2011, where it claimed 54 lives [20].

Molecular typing studies to evaluate E. coli related infections in low and middle-income countries (LMICs) are rare and little information is available regarding the specific subtypes causing infections in Tanzania or whether the high prevalence is due to sporadic infectious or clonal transmission between patients. The purpose of this study was to determine the serotypes, antimicrobial resistance and virulence genes. Further, we carried out a phylogenetic tree reconstruction to determine relatedness of E. coli isolated from patients in a tertiary care hospital in Tanzania using the WGS-based diagnostic platform installed at KCMC hospital in Moshi, Tanzania.

The present study revealed a high diversity of E. coli strains circulating in KCMC hospital settings as measured by Multi-Locus Sequence Typing and Serotyping. However, strains belonging to O25:H4-ST131 and O89:H10-ST617 (ST10 clonal complex) were found to predominate. These findings are similar to the findings from other continents describing the spread and predominance of these endemic clones in health facilities [31‐34]. The observed clonal diversity in this hospital may suggest sporadic introductions of diverse strains into the hospital from the community. To explore whether or not similar STs were clonally related, SNP difference between isolates and phylogenetic analysis suggested the existence of multiple clones of E. coli in these settings. A similar clonal diversity was observed when STs from the present study were compared to similar STs of E. coli isolates from companion and other domesticated animals in Mwanza, north western Tanzania.

Overall, levels of antimicrobial resistance in E. coli isolates were observed to be high. Trimethoprim/sulphamethoxazole (co-trimoxazole) resistance was observed to be lower (70%) than the one (93%) found by Seni et al. [30], but in the present study on average 54–60% resistance to amoxicillin-clavulanate, ampicillin, ceftazidime, ceftriaxone, and ciprofloxacin was higher than that reported by Seni et al. [30]. Strong agreement (81–100%) between sequence and phenotype-based resistance to all drugs tested was observed. However, trimethoprim/sulphamethoxazole revealed moderate agreement (45.8%) between the two methods. The phenotypically determined resistance to trimethoprim/sulphamethoxazole was higher than sequence-based resistance. The plausible explanations for the observed difference could be that our analysis used only known resistance genes and did not include point mutations. Also, resistance in gram negative bacteria including E. coli is multifactorial and not all genes involved in resistance mechanisms have been uncovered.

Following analysis of resistance genes, over 70% of the E. coli strains carried genes encoding beta-lactamases, with bla_OXA-1 being predominant (50%) followed by bla_CTX-M-15 (44.7%) and bla_TEM-1B (31.6%). The prevalence of bla_CTX-M-15 found in the present study contrasts with Manyahi et al. [18] who found bla_CTX-M-15 as the most prevalent gene (90.6%) in a tertiary hospital in Dar es Salaam. Nonetheless, similar to the present study, O25:H4-ST131 was reported to be a major cause of MDR E. coli infections [35]. The proportion (70%) of E. coli strains carrying genes encoding beta-lactamases, correlates in this study with a high (50%) presence of aac(6’)Ib-cr encoding ciprofloxacin resistance enzymes. Although aac(6’)Ib-cr encodes low level ciprofloxacin resistance by itself (as well as aminoglycoside resistance), and usually requires additional mutations (e.g. in chromosomal gyrA or parC) to confer high level resistance, it is a threat to ciprofloxacin which in our settings is one of the most prescribed drugs. Other studies in the US [36, 37], Brazil [38] and Korea [34] have documented similar findings to the present study of the co-carriage of aac(6’)Ib-cr, ESBL genes and the existence of ciprofloxacin resistance in E. coli ST131. Further, an agreement between the presence or absence of aac(6’)Ib-cr and phenotypic resistance results in particular ciprofloxacin was noted by Madoshi et al. [39] who characterised E. coli isolates from healthy cattle and cattle attendants in Morogoro, Tanzania [39]. Among other fluoroquinolones, ciprofloxacin is one of the most prescribed antibiotics in Tanzania. Plausibly this explains the observed linkage between ciprofloxacin resistance and carriage of aac(6’)Ib-cr and CTX-M ESBL [31]. The linkage between quinolone resistance and sub-lineages of ST131 has been recently explored by Zakour et al. [40] whereby this work suggested that quinolone use is associated with the acquisition of virulence and fluoroquinolone resistance determinants and expansion of clade C2/H30-Rx of quinolone-resistant ST131. Also a gene (mph(A)) conferring resistance to macrolides including erythromycin and azithromycin was detected in a high proportion. The presence of the mph(A) in E. coli isolates of the current study was 50%, and thus, higher than the 13% that was found in E. coli from 5 countries from 4 continents by Nguyen et al. [41]. High exposures to erythromycin and azithromycin could be one possible reason leading to emergence of resistance to macrolides [42].

Furthermore, the present study noted a high proportion (73.7%) of dfrA genes encoding trimethoprim resistance enzymes similar to findings by Madoshi et al. [39] in Morogoro, Tanzania. Nonetheless, the present study proportion of dfrA genes was relatively higher than that (43%) found in E. coli isolates from healthy college students in Ghana and Nigeria in 2005 and 2009 [43]. The observed proportion difference could be explained by the fact that resistance in hospital-based studies is relatively higher than in community-based ones. Another reason could be that the present study was hospital-based whereas the West African study population were healthy individuals.

The present study also characterised VFs in all E. coli isolates. The VFs that facilitate E. coli and host cells or E. coli and E. coli adherence including long polar fimbriae (lpfA), adhesin (iha) and intimin (eae) were predominant among the circulating E. coli O25:H4-ST131 strains. These VFs have been identified as suggestive of virulent serotypes and may be used as reliable markers for the identification of pathogenic E. coli [44]. Another prevalent group of VFs this study noted were those responsible for E. coli immune evasion by increasing serum survival (iss). Although the predominance of iss in the present study was from E. coli isolated from hospitalised patients, in Mwanza north-western Tanzania, Msahana et al. [45] noted a similar pattern of iss in E. coli that were community-acquired. In this study we observed no clear correlation or pattern of VFs with the clinical findings or patient’s outcomes. However, colonisation events rather than infections were common as wound or pus swab and stool specimens constituted the majority of strains studied. The observed existence of multiple VFs further underlines the exceptional ability of E. coli in colonising or causing infections to a wide range of hosts and niches.

Potential clinical implications for the obtained results are that caution should be taken when interpreting and utilising microbiology results especially when E. coli is isolated in LMICs as most often E. coli is regarded of low threat. Contrary to that notion, these findings highlight that E. coli should not be regarded non-pathogenic until pathogenic and antimicrobial resistance determinants have been truly confirmed absent. Additionally, the WGS-based findings hint on the existence of nosocomial transmissions in the hospital thus prompting the formulation of pragmatic antimicrobial stewardships and infection prevention and control initiatives.