In this study, we performed whole genome sequencing of bloodstream isolates submitted to the Nigerian surveillance system. Bloodstream isolates collected between 2014 and 2020 included Staphylococcus aureus, Escherichia coli, Klebsiella pneumoniae, Acinetobacter baumanii, Salmonella, Pseudomonas aeruginosa, Enterobacter spp. as the most common genera (Figure 1a). Of 2360 isolates received, 63 were initially sent as presumptive Enterobacter, and 98 (4.2%) were eventually identified as Enterobacter species. Currently, there are 45 (22 named and 21 without assigned names) species of Enterobacter [5], [31], [34], seven were identified in this study (Figure 1b). Among the 22 named Enterobacter spp., 7 belong to the Enterobacter cloacae complex, and they include E. cloacae, E. hormaechei, E. mori, E. asburiae, E. ludwigii, E. nimipressuralis and E. kobei. All but E. mori, ludwigii, and nimipressuralis were identified in this study. We also identified E. roggenkampii, bugandensis, and cancerogenous, which are not members of the cloacae complex (Figure 1a).

Enterobacter hormaechei is frequently encountered in clinical specimens and is commonly considered a nosocomial pathogen [34] but there are only a few reports of bloodstream Enterobacter hormaechei infections from Africa. Duru et al., 2020 [12] reported the identification of Enterobacter spp. from blood samples, but the taxonomic resolution was limited to the genus level. In contrast to our study, which identified E. hormaechei as the most common Enterobacter species, a previous study conducted more than a decade ago [14] in Benin City, Nigeria, identified E. sakazakii and E. aerogenes as the most prevalent Enterobacter species from clinical samples, which included blood and did not report Enterobacter hormaechei. In the Sands et al.(2021) study, Enterobacter were prominent in Nigeria. The conspicuous dearth of any previous report on E. hormaechei from other clinical samples in Nigeria is likely due to the misidentification of Enterobacter species using biochemical identification methods.

Identifying the Enterobacter genus is often challenging [34]. The genus is often inaccurately identified by clinical laboratories using biochemical and other phenotype-based tests, VITEK2, and MALDI-TOF mass spectrometry [6, 35]. Resource-limited settings may face heightened challenges with identifying bacterial pathogens and therefore supportive reference laboratory services are critical [8, 16, 36]. In this study, four Enterobacter species were submitted as Gram-positive strains and three of those 4 belonged to the most common species- Enterobacter hormaechei. Isolates were submitted after identification by tube- or strip biochemical test-using labs as Enterobacteriaceae (n=11) or Enterobacter cloacae (n=5) or were misidentified by sentinels as Acinetobacter baumannii (n=3), Pseudomonas (n=2), Klebsiella pneumoniae (n=14), Escherichia coli (n=6), Halovenus (n=1), Klebsiella aerogenes (n=2)). At the reference laboratory, VITEK2 lacked the resolution to delineate these species as Enterobacter hormaechei, although it did place most of them in the cloacae complex. Only WGS identified these organisms accurately at the species and sub-species level. We show that E. hormaechei was commonly isolated in all the participating in this study’s hospitals (except BUTH from which overall, only a few isolates were obtained) (Figure 1b, 1c). The within-species misclassifications have little consequence for patient management but prevent early identification of clusters, which is important for infection prevention and control. Our results, and the important sub-specific nuances we found, emphasize the need to integrate WGS into routine clinical diagnosis of infectious diseases. Also, there is a need to update the VITEK2 and MALDI-TOF databases to improve the accuracy of speciation.

Analysis of the hsp60 gene (a housekeeping gene) has conventionally been used to sub-classify the Enterobacter cloacae complex into 13 genetic clusters (Hoffman clusters I - XII and an unstable sequence crowd xiii) [37]. A whole genome analysis study (1,997 Enterobacter genomes) updated the taxonomy of the Enterobacter genus [11]. ANI thresholds between 94 and 96.5% and 97-98% for subspecies have good correlations with current species designations [38]. Average Nucleotide Identity (ANI) was used for classifying strains in the Enterobacter cloacae complex into 22 clades (A-V) which correspond to Hoffman clusters (I-XII) [39] (Figure 1b; Figure 4b). Making connections across the Enterobacter literature is challenging because of the different schemes, to which multi-locus sequence typing, offering much finer sub-classification, has been added. WGS approaches make it possible to easily classify strains according to all schemes, and therefore be able to compare disparate datasets. This made it possible for us to collate information from the Sands et al study, which was performed in different locations in Nigeria on an overlapping timescale. Like Sands et al [24], we found MLST most helpful.

Our isolates belonged to 22 different sequence types including novel sequence type (now assigned ST1995, ST1996, ST1997, and ST1998), which, with the 12 STs reported by Sands et al (of which 2 STs were seen in both studies) shows that the E. hormaechei population in Nigerian hospitals is considerably diverse. E. hormaechei strains were multidrug resistant with the detection of aminoglycoside, cephalosporin, chloramphenicol, macrolide, colistin, and carbapenem resistance genes. An ST148 and ST109-bla_NDM (Figure 3a) appear to represent outbreaks, comprised of isolates with SNP distances of ≤1 and ≤1 respectively. ST148 Enterobacter strains have been known to cause outbreaks in the past and have been identified among species isolated at hospitals. A study in Canada that investigated Carbapenemase-Producing Enterobacterales transmission clusters at a hospital system identified bla_VIM-1-positive ST148 strains harbouring plasmid replicons IncR, HI2, and HI2A [40]. The ST148 strains in this study did not carry any carbapenemase gene but were multidrug-resistant (Figure 3a) and were phenotypically resistant to imipenem and meropenem with MICs of ≥8 and ≥16 (Table S5). This phenotype may be a result of overexpression of the chromosomal ampC gene alongside alteration in outer membrane transcriptome balance which is known to proffer other phenotypes such as carbapenem resistance [41]. OXA-48-like-producing ST109 E. cloacae was implicated alongside 22 Klebsiella pneumoniae and 3 Escherichia coli in outbreaks of OXA-48-like-producing Enterobacteriaceae in Czech hospitals in 2015. The ST109 E. cloacae strain harbored, in addition to the bla_OXA-48 gene, bla_CTX-M-15, bla_OXA-1, and bla_TEM-1 [42]. The ST109 strains in this study carried only the core bla_ACT-45 and bla_NDM-1 beta-lactamase genes (Figure 3). E. cloacae was identified in all the hospitals excluding UITH. A total of 26 isolates belonged to 10 different sequence types with the ST850 being most prevalent with SNP differences of ≤2. Nigeria was the only country from which Sands et al.,2021 recovered Enterobacter spp. from every sentinel, similar to our study. The ST859 E. cloacae strains from our study are not very distantly related to the ST850 E. cloacae genomes from Sands et al., with an SNP distance range of between 70 and 89. Altogether, they identified 3 ST850 Enterobacter cloacae among antimicrobial-resistant Gram-negative bacteria that cause neonatal sepsis in seven low- and middle-income countries, and all were reported from Nigeria. Unlike our study, Sands et al sentinels were in northern Nigeria. Thus, while our ST850 isolates appear to be focused at one sentinel, this clade may be circulating widely in Nigeria. E. cloacae isolates in this study were resistant to aminoglycosides, cephalosporins, and colistin. The presence of the mobile colistin resistance gene- mcr-10.1 in two E. cloacae strains in this study is very worrisome as colistin, which is difficult to access in Nigeria, is one of the last available antibiotics used in the treatment of carbapenem-resistant infections and mcr genes are easily transmitted. Although there are few reports of colistin-resistant Enterobacter in Africa, colistin-resistant E. cloacae was recently identified in Sierra Leone. The strain belonged to ST850 and was resistant to cefazolin, gentamicin, and trimethoprim [43]. ST850 strains from this study were colistin-sensitive, and the colistin-resistant E. cloacae strains belonged to ST1525 and ST760. They also possessed bla_CTX-M-15, bla_TEM1, bla_CMH, qnrS, qnrB, tetD, aph(6)-id, aadA1, dfrA, catA2, sul2 genes which confer resistance to beta-lactams, quinolone, tetracycline, trimethoprim, chloramphenicol, aminoglycoside, and sulfonamides.

The importance of Enterobacter species as bloodstream pathogens in Nigeria has heretofore been overlooked because precise identification of this genus is a difficulty for clinical laboratories due to limited biochemical capacity and the complicated taxonomy of the Genus. In this study, WGS allowed for the correct delineation of members of this genus, which uncovered Enterobacter hormaechei as the predominant species. It also allowed for retrospective identification of earlier missed outbreaks. In this study, the retrospective identification of potential outbreaks and the detection of genes conferring resistance to last-line drugs, carbapenems, and colistin is worrying.

The detection of E. hormaechei as the most prevalent Enterobacter bloodstream isolates and identification of key E. cloacae lineages in this study uncovers the need to strengthen clinical laboratory identification and for continued surveillance of this genus. This study emphasizes the importance of WGS in bacteriology, but it also demonstrates that concentrating WGS resources at the reference laboratory is a barrier to identifying lineages and clusters that are important at the patient care level, something that needs to be addressed in our setting in the future.