A One Health Perspective on Salmonella enterica Serovar Infantis, an Emerging Human Multidrug-Resistant Pathogen

Salmonella enterica serovar Infantis presents an ever-increasing threat to public health because of its spread throughout many countries and association with high levels of antimicrobial resistance (AMR). We analyzed whole-genome sequences of 5,284 Salmonella Infantis strains from 74 countries, isolated during 1989–2020 from a wide variety of human, animal, and food sources, to compare genetic phylogeny, AMR determinants, and plasmid presence. The global Salmonella Infantis population structure diverged into 3 clusters: a North American cluster, a European cluster, and a global cluster. The levels of AMR varied by Salmonella Infantis cluster and by isolation source; 73% of poultry isolates were multidrug resistant, compared with 35% of human isolates. This finding correlated with the presence of the pESI megaplasmid; 71% of poultry isolates contained pESI, compared with 32% of human isolates. This study provides key information for public health teams engaged in reducing the spread of this pathogen.

classified as human.Isolates from chickens, ducks, turkeys and quail were categorised as poultry.All strains from poultry products, such as eggs and chicken meat, were grouped together.Isolates from all other animals were grouped into 'other animals' and non-poultry animal products and other food samples were grouped into 'food'.All samples from animal feed were grouped and farm swabs, soil, water, air and sewage were designated environmental.
Isolates missing source information to assign to these groups were classed as unknown.For the UKHSA isolates with travel history information, the continent of travel was used to designate the continent of isolate origin.The world map plot was generated with R (v.4.1.0)and the package ggplot2 (v.3.3.6)(2,3).

Whole Genome Sequencing
DNA from the NICD isolates was extracted using QIAamp DNA Mini Kit (Qiagen) and sequenced as described in Mattock et al., 2021 (4); libraries were prepared using either Illumina Nextera XT library preparation kits or using a custom library preparation method described in Rasheed et al., 2020 (5).Twelve NICD isolates not used in Mattock et al., 2021 are included here and a library was prepared using one of the aforementioned methods and sequenced on an Illumina NextSeq 500.
The DNA of the UKHSA samples was extracted using the Qiasymphony DSP DNA Midi Kit (Qiagen), following a protocol described in Nair et al., 2016 (6).The historical UKHSA samples, described in Lee et al., 2021, were also sequenced using either Illumina Nextera XT or the custom library preparation methods (7).Two additional historical UKHSA samples were included and sequenced using the custom library preparation method.The contemporary UKHSA samples were prepared using the Illumina Nextera XT protocol and sequenced on an Illumina HiSeq 2500 instrument.
DNA from the APHA isolates was extracted using the MagMAX CORE Nucleic Acid Purification Kit (ThermoFisher) with a KingFisher Flex Purification System (ThermoFisher).
The DNA was then sequenced using the custom library preparation method and an Illumina NextSeq 500.

Phylogenetic analysis
All the sequence data generated by UKHSA were trimmed using Trimmomatic (v0.27) with leading and trailing at <Q30 (8).All other sequence data were trimmed with Trimmomatic (v0.36) and the same options.Sequence type and eBG were determined for each isolate using Metric-Oriented Sequence Typer (MOST) (v.1.0)with a UKHSA Salmonella database (9,10).
The quality of borderline sequence typing was assessed using Tablet (11).
Using the Cloud Infrastructure for Big Data Microbial Bioinformatics (12), each sample was mapped and variant called against the eBG31 reference CP070301 (7), using Snippy (v.4.6.0) with the options minfrac 0.9 and 30 (13).Due to the large number of sequences included, it was not computationally possible to produce a phylogeny including all of the isolates.Therefore, the SNP distance between all of the consensus FASTAs produced by Snippy was calculated using snp-dists (v.0.7) (https://github.com/tseemann/snp-dists)with the -m option to output in the molten format.Using MCL (v14-137) and the abc option, Markov clustering was performed on the pairwise SNP distance matrix with the following SNP distance thresholds: 0-SNPs, 5-SNPs, 10-SNPs, 25-SNPs, 50-SNPs, 100-SNPs and 250-SNPs (10,14).This resulted in clusters where each cluster member was less than 'n' SNPs from another member.The smallest SNP threshold that it was possible to generate a phylogeny with was 25-SNPs; the first isolate in each 25-SNP cluster was chosen to be the representative for that cluster.
The treedater relaxed clock test was performed to determine support for using a strict clock; the strict clock was deemed the best approach (21).Treedater (v.0.5.0) was used with a strict clock and the dates ranges of the isolates in each 25-SNP cluster, excluding leaves with an unknown isolation date, to date the phylogeny.Phangorn (v.2.10.0) was used to identify the nodes where the clades diverged (22).