Prevalence of Shiga Toxin-Producing Escherichia coli O157 in Wild Scottish Deer with High Human Pathogenic Potential.

Shiga toxin-producing E. coli (STEC) infections associated with wildlife are increasing globally, highlighting many 'spillover' species as important reservoirs for these zoonotic pathogens. A human outbreak of STEC serogroup O157 in 2015 in Scotland, associated with the consumption of venison meat products, highlighted several knowledge gaps, including the prevalence of STEC O157 in Scottish wild deer and the potential risk to humans from wild deer isolates. In this study, we undertook a nationwide survey of wild deer in Scotland and determined that the prevalence of STEC O157 in wild deer is low 0.28% (95% confidence interval = 0.06-0.80). Despite the low prevalence of STEC O157 in Scottish wild deer, identified isolates were present in deer faeces at high levels (>104 colony forming units/g faeces) and had high human pathogenic potential based on whole genome sequencing and virulence gene profiling. A retrospective epidemiological investigation also identified one wild deer isolate from this study as a possible source of a Scottish human outbreak in 2017. These results emphasise the importance of food hygiene practices during the processing of wild deer carcasses for human consumption.

Analysis of Escherichia coli O157 strains in cattle and humans between Scotland and England & Wales: implications for human health.

For the last two decades, the human infection frequency of Escherichia coli O157 (O157) in Scotland has been 2.5-fold higher than in England and Wales. Results from national cattle surveys conducted in Scotland and England and Wales in 2014/2015 were combined with data on reported human clinical cases from the same time frame to determine if strain differences in national populations of O157 in cattle could be associated with higher human infection rates in Scotland. Shiga toxin subtype (Stx) and phage type (PT) were examined within and between host (cattle vs human) and nation (Scotland vs England and Wales). For a subset of the strains, whole genome sequencing (WGS) provided further insights into geographical and host association. All three major O157 lineages (I, II, I/II) and most sub-lineages (Ia, Ib, Ic, IIa, IIb, IIc) were represented in cattle and humans in both nations. While the relative contribution of different reservoir hosts to human infection is unknown, WGS analysis indicated that the majority of O157 diversity in human cases was captured by isolates from cattle. Despite comparable cattle O157 prevalence between nations, strain types were localized. PT21/28 (sub-lineage Ic, Stx2a+) was significantly more prevalent in Scottish cattle [odds ratio (OR) 8.7 (2.3-33.7; P<0.001] and humans [OR 2.2 (1.5-3.2); P<0.001]. In England and Wales, cattle had a significantly higher association with sub-lineage IIa strains [PT54, Stx2c; OR 5.6 (1.27-33.3); P=0.011] while humans were significantly more closely associated with sub-lineage IIb [PT8, Stx1 and Stx2c; OR 29 (4.9-1161); P<0.001]. Therefore, cattle farms in Scotland were more likely to harbour Stx2a+O157 strains compared to farms in E and W (P<0.001). There was evidence of limited cattle strain migration between nations and clinical isolates from one nation were more similar to cattle isolates from the same nation, with sub-lineage Ic (mainly PT21/28) exhibiting clear national association and evidence of local transmission in Scotland. While we propose the higher rate of O157 clinical cases in Scotland, compared to England and Wales, is a consequence of the nationally higher level of Stx2a+O157 strains in Scottish cattle, we discuss the multiple additional factors that may also contribute to the different infection rates between these nations.

Genome structural variation in Escherichia coli O157:H7.

The human zoonotic pathogen Escherichia coli O157:H7 is defined by its extensive prophage repertoire including those that encode Shiga toxin, the factor responsible for inducing life-threatening pathology in humans. As well as introducing genes that can contribute to the virulence of a strain, prophage can enable the generation of large-chromosomal rearrangements (LCRs) by homologous recombination. This work examines the types and frequencies of LCRs across the major lineages of the O157:H7 serotype. We demonstrate that LCRs are a major source of genomic variation across all lineages of E. coli O157:H7 and by using both optical mapping and Oxford Nanopore long-read sequencing prove that LCRs are generated in laboratory cultures started from a single colony and that these variants can be recovered from colonized cattle. LCRs are biased towards the terminus region of the genome and are bounded by specific prophages that share large regions of sequence homology associated with the recombinational activity. RNA transcriptional profiling and phenotyping of specific structural variants indicated that important virulence phenotypes such as Shiga-toxin production, type-3 secretion and motility can be affected by LCRs. In summary, E. coli O157:H7 has acquired multiple prophage regions over time that act to continually produce structural variants of the genome. These findings raise important questions about the significance of this prophage-mediated genome contingency to enhance adaptability between environments.

Predicting Host Association for Shiga Toxin-Producing E. coli Serogroups by Machine Learning.

Escherichia coli is a species of bacteria that can be present in a wide variety of mammalian hosts and potentially soil environments. E. coli has an open genome and can show considerable diversity in gene content between isolates. It is a reasonable assumption that gene content reflects evolution of strains in particular host environments and therefore can be used to predict the host most likely to be the source of an isolate. An extrapolation of this argument is that strains may also have gene content that favors success in multiple hosts and so the possibility of successful transmission from one host to another, for example, from cattle to human, can also be predicted based on gene content. In this methods chapter, we consider the issue of Shiga toxin (Stx)-producing E. coli (STEC) strains that are present in ruminants as the main host reservoir and for which we know that a subset causes life-threatening infections in humans. We show how the genome sequences of E. coli isolated from both cattle and humans can be used to build a classifier to predict human and cattle host association and how this can be applied to score key STEC serotypes known to be associated with human infection. With the example dataset used, serogroups O157, O26, and O111 show the highest, and O103 and O145 the lowest, predictions for human association. The long-term ambition is to combine such machine learning predictions with phylogeny to predict the zoonotic threat of an isolate based on its whole genome sequence (WGS).

A guide to machine learning for bacterial host attribution using genome sequence data.

With the ever-expanding number of available sequences from bacterial genomes, and the expectation that this data type will be the primary one generated from both diagnostic and research laboratories for the foreseeable future, then there is both an opportunity and a need to evaluate how effectively computational approaches can be used within bacterial genomics to predict and understand complex phenotypes, such as pathogenic potential and host source. This article applied various quantitative methods such as diversity indexes, pangenome-wide association studies (GWAS) and dimensionality reduction techniques to better understand the data and then compared how well unsupervised and supervised machine learning (ML) methods could predict the source host of the isolates. The study uses the example of the pangenomes of 1203 Salmonella enterica serovar Typhimurium isolates in order to predict 'host of isolation' using these different methods. The article is aimed as a review of recent applications of ML in infection biology, but also, by working through this specific dataset, it allows discussion of the advantages and drawbacks of the different techniques. As with all such sub-population studies, the biological relevance will be dependent on the quality and diversity of the input data. Given this major caveat, we show that supervised ML has the potential to add real value to interpretation of bacterial genomic data, as it can provide probabilistic outcomes for important phenotypes, something that is very difficult to achieve with the other methods.

Whole Genome Sequence Analysis Reveals Lower Diversity and Frequency of Acquired Antimicrobial Resistance (AMR) Genes in E. coli From Dairy Herds Compared With Human Isolates From the Same Region of Central Zambia.

Antibiotic treatment of sick dairy cattle is critical for the sustainability of this production system which is vital for food security and societal prosperity in many low and middle-income countries. Given the increasingly high levels of antibiotic resistance worldwide and the challenge this presents for the treatment of bacterial infections, the rational use of antibiotics in humans and animals has been emphatically recommended in the spirit of a "One Health" approach. The aim of this study was to characterize antimicrobial resistance (AMR) genes and their frequencies from whole genome sequences of Escherichia coli isolated from both dairy cattle and human patients in central Zambia. Whole genome sequences of E. coli isolates from dairy cattle (n = 224) and from patients at a local hospital (n = 73) were compared for the presence of acquired AMR genes. In addition we analyzed the publicly available genomes of 317 human E. coli isolates from over the wider African continent. Both acquired antibiotic resistance genes and phylogroups were identified from de novo assemblies and SNP based phylogenetic analyses were used to visualize the distribution of resistance genes in E. coli isolates from the two hosts. Greater acquired AMR gene diversity was detected in human compared to bovine E. coli isolates across multiple classes of antibiotics with particular resistance genes for extended-spectrum beta lactamases (ESBL), quinolones, macrolides and fosfomycin only detected in E. coli genomes of human origin. The striking difference was that the Zambian or wider African human isolates were significantly more likely to possess multiple acquired AMR genes compared to the Zambian dairy cattle isolates. The median number of resistance genes in the Zambian cattle cohort was 0 (0-1 interquartile range), while in the Zambian human and wider African cohorts the medians and interquartile ranges were 6 (4-9) and 6 (0-8), respectively. The lower frequency and reduced diversity of acquired AMR genes in the dairy cattle isolates is concordant with relatively limited antibiotic use that we have documented in this region, especially among smallholder farmers. The relatively distinct resistant profiles in the two host populations also indicates limited sharing of strains or genes.

Patchy promiscuity: machine learning applied to predict the host specificity of Salmonella enterica and Escherichia coli.

Salmonella enterica and Escherichia coli are bacterial species that colonize different animal hosts with sub-types that can cause life-threatening infections in humans. Source attribution of zoonoses is an important goal for infection control as is identification of isolates in reservoir hosts that represent a threat to human health. In this study, host specificity and zoonotic potential were predicted using machine learning in which Support Vector Machine (SVM) classifiers were built based on predicted proteins from whole genome sequences. Analysis of over 1000 S.enterica genomes allowed the correct prediction (67 -90 % accuracy) of the source host for S. Typhimurium isolates and the same classifier could then differentiate the source host for alternative serovars such as S. Dublin. A key finding from both phylogeny and SVM methods was that the majority of isolates were assigned to host-specific sub-clusters and had high host-specific SVM scores. Moreover, only a minor subset of isolates had high probability scores for multiple hosts, indicating generalists with genetic content that may facilitate transition between hosts. The same approach correctly identified human versus bovine E. coli isolates (83 % accuracy) and the potential of the classifier to predict a zoonotic threat was demonstrated using E. coli O157. This research indicates marked host restriction for both S. enterica and E. coli, with only limited isolate subsets exhibiting host promiscuity by gene content. Machine learning can be successfully applied to interrogate source attribution of bacterial isolates and has the capacity to predict zoonotic potential.

Convergence of plasmid architectures drives emergence of multi-drug resistance in a clonally diverse Escherichia coli population from a veterinary clinical care setting.

The purpose of this study was to determine the plasmid architecture and context of resistance genes in multi-drug resistant (MDR) Escherichia coli strains isolated from urinary tract infections in dogs. Illumina and single-molecule real-time (SMRT) sequencing were applied to assemble the complete genomes of E. coli strains associated with clinical urinary tract infections, which were either phenotypically MDR or drug susceptible. This revealed that multiple distinct families of plasmids were associated with building an MDR phenotype. Plasmid-mediated AmpC (CMY-2) beta-lactamase resistance was associated with a clonal group of IncI1 plasmids that has remained stable in isolates collected up to a decade apart. Other plasmids, in particular those with an IncF replicon type, contained other resistance gene markers, so that the emergence of these MDR strains was driven by the accumulation of multiple plasmids, up to 5 replicons in specific cases. This study indicates that vulnerable patients, often with complex clinical histories provide a setting leading to the emergence of MDR E. coli strains in clonally distinct commensal backgrounds. While it is known that horizontally-transferred resistance supplements uropathogenic strains of E. coli such as ST131, our study demonstrates that the selection of an MDR phenotype in commensal E. coli strains can result in opportunistic infections in vulnerable patient populations. These strains provide a reservoir for the onward transfer of resistance alleles into more typically pathogenic strains and provide opportunities for the coalition of resistance and virulence determinants on plasmids as evidenced by the IncF replicons characterised in this study.

Support vector machine applied to predict the zoonotic potential of E. coli O157 cattle isolates.

Sequence analyses of pathogen genomes facilitate the tracking of disease outbreaks and allow relationships between strains to be reconstructed and virulence factors to be identified. However, these methods are generally used after an outbreak has happened. Here, we show that support vector machine analysis of bovine E. coli O157 isolate sequences can be applied to predict their zoonotic potential, identifying cattle strains more likely to be a serious threat to human health. Notably, only a minor subset (less than 10%) of bovine E. coli O157 isolates analyzed in our datasets were predicted to have the potential to cause human disease; this is despite the fact that the majority are within previously defined pathogenic lineages I or I/II and encode key virulence factors. The predictive capacity was retained when tested across datasets. The major differences between human and bovine E. coli O157 isolates were due to the relative abundances of hundreds of predicted prophage proteins. This finding has profound implications for public health management of disease because interventions in cattle, such a vaccination, can be targeted at herds carrying strains of high zoonotic potential. Machine-learning approaches should be applied broadly to further our understanding of pathogen biology.

Phylogenomic approaches to determine the zoonotic potential of Shiga toxin-producing Escherichia coli (STEC) isolated from Zambian dairy cattle.

This study assessed the prevalence and zoonotic potential of Shiga toxin-producing Escherichia coli (STEC) sampled from 104 dairy units in the central region of Zambia and compared these with isolates from patients presenting with diarrhoea in the same region. A subset of 297 E. coli strains were sequenced allowing in silico analyses of phylo- and sero-groups. The majority of the bovine strains clustered in the B1 'commensal' phylogroup (67%) and included a diverse array of serogroups. 11% (41/371) of the isolates from Zambian dairy cattle contained Shiga toxin genes (stx) while none (0/73) of the human isolates were positive. While the toxicity of a subset of these isolates was demonstrated, none of the randomly selected STEC belonged to key serogroups associated with human disease and none encoded a type 3 secretion system synonymous with typical enterohaemorrhagic strains. Positive selection for E. coli O157:H7 across the farms identified only one positive isolate again indicating this serotype is rare in these animals. In summary, while Stx-encoding E. coli strains are common in this dairy population, the majority of these strains are unlikely to cause disease in humans. However, the threat remains of the emergence of strains virulent to humans from this reservoir.