(C) PLOS One This story was originally published by PLOS One and is unaltered. . . . . . . . . . . Dynamics of Salmonella enterica and antimicrobial resistance in the Brazilian poultry industry and global impacts on public health [1] ['Nabil-Fareed Alikhan', 'Quadram Institute Bioscience', 'Norwich', 'United Kingdom', 'Luisa Zanolli Moreno', 'Department Of Preventive Veterinary Medicine', 'Animal Health', 'School Of Veterinary Medicine', 'Animal Science', 'University Of São Paulo'] Date: 2022-08 To explore the link between poultry raised in Brazil and UK imported chicken, we also included 318 Salmonella genomes from poultry meat tested at the point of import into the UK from Brazil. These isolates were originally sequenced as part of imported food surveillance. Of the 318 genomes, 91% of the isolates were either Salmonella Heidelberg (n = 259) or Salmonella Minnesota (n = 26), whereas a single S. Enteritidis (ST11; n = 1) was detected. This high number of S. Heidelberg and S. Minnesota strains in poultry imported into the UK from Brazil is consistent with the results found in the newly sequenced genomes from poultry from Brazil and the previous reports cited above. Serovar and MLST prediction, and sequencing quality control metrics are available in S2 Table . A total of 183 S. enterica genomes, with varied genotypes and serovars, isolated in Brazil between 2012 and 2018 from thirteen states across the country were analysed. All isolates were obtained from broiler chicken samples sent to diagnostic laboratories for monitoring of Salmonella infection (see S1 Table ). DNA from these isolates was subsequently extracted and sequenced using the Illumina NextSeq 500 platform with paired end 150bp reads. Through a combination of de novo genome assembly tools and in silico typing tools (see Methods ), we determined that these isolates represented 36 different S. enterica serovars and 38 7-gene multi-locus sequence types (MLST). Salmonella Enteritidis (ST11; n = 4), which is the predominant cause of foodborne salmonellosis, accounted for only 2% of all isolates from Brazil. Similar to previous reports [ 4 , 5 , 10 , 11 ], Salmonella serovars Heidelberg (ST15) and Minnesota (ST548) were the most common with 37 and 28 genomes, respectively ( S2 Table ). Among the remaining isolates, serovars Schwarzengrund (ST96; n = 15), Senftenberg (multiple STs; n = 13), Mbandaka (ST413; n = 8), Anatum (ST64; n = 6), Braenderup (multiple STs; n = 5), Ouakam (ST1610; n = 5), Cerro (multiple STs; n = 4), Muenchen (ST112; n = 4) and Ohio (ST329; n = 4) were identified ( S2 Table ). Characterisation of all isolates, including the less frequent serovars (n<4) is available in Fig 1 and S2 Table . The time phylogenies’ results indicate that both monophyletic groups (Figs 2 and 3 ) represent recent expansions, likely aided by the introduction of the Salmonella Enteritidis vaccine into Brazil chicken production in 2003 [ 6 ]. The date for the S. Minnesota monophyletic group agrees with dates calculated previously, where the clade was designated “SM-PLII” and described as rapidly expanding in the beginning of the 2000s [ 24 ]. S. enterica control programmes and vaccines have been described as playing an important role in serovar shifts in poultry production since the 1930s [ 16 ]. To determine the validity of these date ranges, a separate model was run for both monophyletic groups where isolation dates were fixed to a single date and in both cases a measure of the deviance information criterion reported that the randomised models were worse, indicating the temporal signal for both monophyletic groups is significant. We also constructed separate time-dated phylogenies for the Salmonella Minnesota and Salmonella Heidelberg monophyletic groups to establish the possible date of emergence. Dating of the Salmonella Heidelberg monophyletic group was facilitated by additional older genomes (2010–2015) from imported poultry samples from Brazil into The Netherlands [ 4 ], suggested the Salmonella Heidelberg monophyletic group emerged between 1996 (CI95: 1968–2004) and 2004 (CI95: 1987–2006) ( S2 and S3 Figs). The former date represents the date of the split from the closest known outgroup and the latter date represents the time to the most recent common ancestor (tMRCA) of all Brazil and UK imported isolates. This time frame is consistent with the tMRCA of the Salmonella Minnesota monophyletic group, which was calculated as 2006 (CI95: 1986–2007) ( S4 and S5 Figs). Support for these dates can be seen in the root-to-tip charts and parameters illustrating the models’ convergence in S2 and S4 Figs. The effective sample size parameters for both models were appropriate (Salmonella Heidelberg: mu 244, sigma 288, alpha 244; Salmonella Minnesota: mu 394, sigma 501, alpha 274). (A) Maximum likelihood phylogenetic tree of Salmonella Minnesota genomes, including isolates from Brazil, imported chicken meat from Brazil, and representatives from the rest of the world. Nodes are coloured by country of origin as indicated in the key. An interactive version of this figure with supporting metadata is available at https://microreact.org/project/brch-3a (B) Maximum likelihood phylogenetic tree of genomes highlighted in lilac in part A. Tips were colour coded according to sample origin. Inner ring indicates the source niche. Outer ring indicates the collection year. An interactive version of this figure with supporting metadata is available at https://microreact.org/project/brch3b . Again, all genomes within this study that were obtained from Brazil, or known to be imported into the UK from Brazil via chicken meat, were within a single monophyletic group ( Fig 3 ). This clade also included genomes from other imported poultry derived isolates (from Brazil to Portugal) and UK clinical isolates. Two Salmonella Minnesota genomes provided by the Animal and Plant Health Agency (APHA) were found in the monophyletic group ( Fig 3A –lilac). Further investigation of the source of these isolates found that one (L01552-07) was a feed isolate from a UK feed mill, while the other (L01529-16) was part of an independent testing programme that did not sample UK chicken. The remaining seven Salmonella Minnesota sourced from APHA were found in two nodes elsewhere on the tree. Hence, as established with Salmonella Heidelberg, no Salmonella Minnesota directly attributed to UK domestic chicken was found within the monophyletic group. The analysis was repeated for the Salmonella Minnesota genomes using a Salmonella Minnesota genome as a reference. The genomes from isolates that were collected from Portugal were sourced from a previous study assessing the presence of S. enterica in imported chicken products from Brazil [ 5 ]. These genomes were first identified through including all publicly available Salmonella Minnesota genomes listed on EnteroBase ( S3 Table ). (A) Maximum likelihood phylogenetic tree of Salmonella Heidelberg genomes, including isolates from Brazil, imported chicken meat from Brazil, domestic UK poultry, and representatives from the rest of the world. Nodes are coloured by country of origin as indicated in the key. An interactive version of this figure with supporting metadata is available at https://microreact.org/project/brch-fig2a (B) Maximum likelihood phylogenetic tree of genomes highlighted in lilac in part A. Tips were colour coded according to sample origin. Inner ring indicates the source lab. Outer ring indicates the collection year. An interactive version of this figure with supporting metadata is available at https://microreact.org/project/brch-fig2b . By mapping sequenced reads against a Salmonella Heidelberg reference sequence CFSAN00324 (Accession GCA_000962725.1), we were able to construct a phylogenetic relationship between these genomes of different sources ( Fig 2 ). All Salmonella Heidelberg genomes within this study that were isolated from Brazil, or known to be imported into the UK from Brazil via chicken meat, were within a single monophyletic group ( Fig 2A ). This clade (lilac) also included six domestic (UK) clinical isolates. All genomes from domestic UK poultry were outside the monophyletic group ( Fig 2A ; red). Notably, the genomes from Brazil were distant from those from the rest of the world. The phylogenetic trees of Salmonella Heidelberg, represented as transmission networks ( S1A Fig ) derived from Fig 2A and location data, confirm the clustering of Brazilian isolates. UK (domestic poultry) and other contextual isolates were separate from this central loop. The transmission network ( S1B Fig ) using the isolates’ source shows interplay between human (clinical), poultry and food but with no certain source. Both Salmonella Heidelberg and Salmonella Minnesota serovars were the most common serovars in the 183 isolates sampled in Brazil and the 318 isolates obtained from imported Brazilian chicken in the UK, suggesting some selective advantage at the point of production ( S1 Table ; Country: “Brazil” & “Brazil (import)”). To explore this further, we constructed a representative dataset of global Salmonella Heidelberg and Salmonella Minnesota genomes. The Salmonella Heidelberg and Salmonella Minnesota datasets also included genomes from domestic UK poultry (30 and 9 respectively) to compare with imported Heidelberg and Minnesota ( S1 Table ; Source lab “APHA” [Animal and Plant Health Agency]). Additional genomes were selected to capture global genomic diversity through publicly available genomes on EnteroBase (see Methods and S3 Table ). In addition, for Salmonella Heidelberg almost all of the UK-derived genomes which cluster with the Brazilian chicken sequences can either be attributed to cases with a recent history of foreign travel or are from imported chicken meat products ( S3 Table ); only six isolates were defined as clinical UK domestic samples. These findings indicate that Salmonella from imported Brazilian poultry products have not caused a major disease burden to UK consumers. Between 2004–2018, there were between approximately 8,100–14,000 confirmed reports of human Salmonella infections per year in England and Wales, reported by the UK Health Security Agency (UKHSA). Given the volume of chicken meat imported into the UK from Brazil, the contribution to this burden of disease from imported Brazilian chicken was investigated. In Brazil, in the present study and others [ 6 , 7 ], a decrease in Salmonella Enteritidis followed by an increase in Salmonella Minnesota and Salmonella Heidelberg serovars in poultry was observed after 2003, the year of introduction of the Enteritidis vaccine. No such rise for these serovars was observed in the historical data provided by UKHSA Salmonella surveillance of clinical cases in the UK ( Fig 4 ), as would be expected if Brazilian imported chicken was a source of salmonellosis for humans in the UK. Indeed, there was a low and stable incidence of infections due to Salmonella Heidelberg and Salmonella Minnesota (≤0.5%) derived from humans between 2004 and 2019 ( Fig 4A ). This proportion is at least 20 times lower than infections due to Salmonella Enteritidis or Salmonella Typhimurium, which are the serovars accounting for the largest numbers of human cases ( Fig 4B ). Investigating AMR as a driver of the success of Salmonella Heidelberg and Salmonella Minnesota in Brazilian poultry The predominance of Salmonella Minnesota and Salmonella Heidelberg, two unrelated Salmonella serovars, in Brazilian poultry could be driven by their common resistance to beta-lactams, sulfamethoxazole and oxytetracycline drugs. This was clearly observed at the genotypic level (S2 Table). To investigate if the same genetic elements were disseminating the resistance between the two serovars, long read sequencing of nine isolates was performed to resolve the genomic context of genes conferring AMR (S1 Table). These nine isolates (five Heidelberg and four Minnesota) were selected from within the UK collection (UKHSA) of isolates from raw chicken meat and other food products imported from Brazil and clinical samples, all of which clustered in the phylogenetic trees with the isolates from poultry tested in Brazil. All nine isolates carried sul2 and tetA and seven carried bla CMY-2 . Most of the genes were carried on IncC plasmids, including IncC plasmids carrying sul2 and tetA (n = 4) and sul2 and tetA in combination with bla CMY-2 (n = 4). bla CMY-2 was also found independently on IncI1 (n = 3) plasmids (S4 Table). The genetic contexts of sul2, tetA and bla CMY-2 genes were investigated amongst IncC and IncI1 plasmids (S6A Fig). IncC plasmids had two gene configurations of sul2 and tetA which were found in both Salmonella Heidelberg and Salmonella Minnesota isolates; in both configurations, the AMR genes were accompanied by a Tn3 transposase, as previously reported [25,26]. In one of the configurations, found only in Salmonella Minnesota isolates, an IS91-like insertion sequence (IS) was additionally integrated between sul2 and tetA. In the second configuration, found only in Salmonella Heidelberg isolates, the sul2/tetA/Tn3 genetic block was in the reverse orientation to that in the first configuration (S6B Fig). In both IncC and IncI1 plasmids in both Salmonella Heidelberg and Salmonella Minnesota isolates, the bla CMY-2 gene was characterised by the upstream presence of ISEcp1 (S6C Fig), as previously reported [27,28], demonstrating the potential role of ISEcp1 in mobilising bla CMY-2 across serovars (S6C Fig). Looking in the wider set of 183 Brazilian poultry genomes, a combination of sul2, tetA and bla CMY-2 was present in 77.8% and 75% of Salmonella Heidelberg and Salmonella Minnesota genomes, respectively, identified in 12 states. Only 2.5% of other S. enterica serovars in this dataset harboured sul2 and tetA and none of them harboured bla CMY-2 (S2 Table). To have a comparison for the gene and plasmid distribution in Brazil, a set of Salmonella genomes derived from poultry in Colombia was examined [29,30]. This demonstrated that Brazilian and Colombian Salmonella Heidelberg isolates shared a similar association of sul2 and tetA with IncC and bla CMY-2 with IncI1 plasmids. However, only in Brazil was the association of sul2, tetA and bla CMY-2 with IncC plasmids observed, in both Salmonella Heidelberg and Salmonella Minnesota (S7 Fig). Phylogenetic analysis was performed for the IncC plasmids from Brazil and the UK. By mapping reads from Brazil isolates onto one IncC contiguous plasmid sequence recovered from the nine UK isolates, we found 67 out of 183 carry some version of the IncC plasmid, with clades largely specific with serovar (Fig 5A), suggesting the plasmids were introduced once into each serovar. Most Brazilian isolates were closely related to at least one of the nine isolates with long-read genome data but there were at least two clades that carry plasmid profiles not captured in the UK isolates. Similarly, a comparison for the IncI1 plasmid from Brazil and the UK showed 39 genomes out of 183 carried some version of the IncI1 plasmid, again with serovar specific clades. However, the genomes that harbour the IncI1 plasmid differ from those that harbour the IncC, and a different topology in the respective phylogenies was observed between the two types (Fig 5B). PPT PowerPoint slide PNG larger image TIFF original image Download: Fig 5. Phylogeny of sul2, tetA and/or bla CMY-2 carrying IncC plasmids in selected isolates. Selected isolates from imported Brazilian raw chicken and food products tested in the UK and chicken obtained in Brazil. (A) Maximum likelihood phylogenetic tree of 67 genomes identified with an IncC plasmid. Genomes were selected from Brazilian samples collected from poultry. Nodes are coloured by predicted serovar as indicated in the key. (B) Maximum likelihood phylogenetic tree of 39 genomes identified with an IncI1 plasmid. Genomes were selected from Brazilian samples collected from poultry. Nodes are coloured by predicted serovar as indicated in the key. https://doi.org/10.1371/journal.pgen.1010174.g005 We have so far described that the common element driving the prevalence of Salmonella Minnesota and Salmonella Heidelberg in Brazilian chicken production, both tested in the country of origin as well as the point of import into the UK, is the acquisition of AMR encoded by sul2, tetA and/or bla CMY-2 . These genes are generally encoded on variants of IncC plasmid or were split between IncC and IncI1 plasmid types. This was observed in isolates collected across 12 states of Brazil, with the highest occurrence in those with the largest production of chickens in the country (Paraná, Santa Catarina, São Paulo and Mato Grosso do Sol). Comparisons of the genetic structure of these plasmids that have these general features show differences even within a small subset of the isolates described here (S6 Fig), and in comparing IncC plasmid sequences from this subset to all Brazil strains in the present study, there are likely to be more plasmid variants that are not fully described (Fig 5). We note that two isolates carrying these plasmids were from Salmonella serovars other than Minnesota and Heidelberg (i.e., Worthington and Newport) indicating that these plasmids can be found in other Salmonella serovars, although rare in this dataset. Laboratory work would be required to determine the relative rates of transfer of these plasmids into diverse Salmonella serovars. In addition, nucleotide BLAST searches (last accessed: 2 February 2021) using the above mentioned long read sequences (UKHSA) of sul2 and tetA (n = 4), sul2, tetA and bla CMY-2 (n = 4) carrying IncC plasmids and bla CMY-2 carrying IncI1 plasmids (n = 3) were performed. The searches showed that for the sul2 and tetA carrying IncC plasmids identified within the nine UKHSA long-read sequenced plasmids, similar plasmids (BLAST query cover ≥99% and identity ≥99%) were present in public genomes of Salmonella Heidelberg and Salmonella Typhimurium and in one publicly available Salmonella Minnesota genome. For sul2, tetA and bla CMY-2 carrying IncC, a similar plasmid to that observed in the UKHSA long-read genomes was found in three Minnesota isolates. For the two bla CMY-2 carrying IncI1 plasmids within the nine long-read UKHSA genomes, similar plasmids were present in Salmonella Heidelberg, Salmonella Typhimurium, Salmonella Kentucky, Salmonella Ohio, Salmonella Anatum, Salmonella Derby, Salmonella Brandenburg and some Escherichia coli. The publicly available genomes of Salmonella Heidelberg, Salmonella Typhimurium and Salmonella Minnesota that were similar to sul2, tetA and/or bla CMY-2 carrying IncC plasmids found within the nine UKHSA long-read genomes, originated from Brazil, USA and Russia. According to the metadata available, the genomes from Brazil and the USA originated from chicken related samples (S4C Table). The public genomes that were similar to the identified bla CMY-2 carrying IncI1 plasmids originated from a wide range of countries including Canada, Denmark, Japan, Uruguay and the USA. Similarly, there was a wide range of sources of isolation including poultry (predominantly), swine, bovine and human related samples (S4C Table). In conclusion, as observed in other countries such as the United States, there have been important shifts in the Salmonella found in Brazilian poultry, where formerly Salmonella Enteritidis predominated. Here we suggest that the introduction of the Salmonella Enteritidis vaccine in poultry in Brazil led to serovar replacement, with Salmonella Minnesota and Salmonella Heidelberg becoming the dominant Salmonella in this setting. These two serovars are very different; Salmonella Heidelberg is a serogroup B Salmonella and Salmonella Minnesota, rarely observed outside of Brazilian poultry, is in serogroup L according to the Kauffman-White-Le Minor scheme [31]. Despite this difference, common carriage of the bla CMY-2 , sul2 and tetA genes, conferring resistance to beta-lactam, sulphonamide and tetracycline antimicrobials were detected, which were rarely seen in other Salmonella serovars co-circulating in Brazilian poultry production. What is not clear is why these very different serovars were able to acquire plasmids carrying these AMR genes and other serovars, co-circulating in Brazilian poultry and under the same selection pressures, were not; this suggests there are barriers to horizontal gene transfer of these plasmids. It is also possible that the selection for Salmonella Minnesota and Salmonella Heidelberg was aided by other genetic factors, such as other virulence genes. Salmonella serovars Heidelberg and Minnesota were less associated with clinical disease in animals and presented low association with human disease in the data described here. Subsequent to the emergence of Salmonella Heidelberg and Salmonella Minnesota in Brazilian poultry, including that imported to the UK and Europe, human infection involving these serovars remains low in the UK. Unfortunately, there were no official data on Salmonella from human infections in Brazil to assess whether or not these serovars have an impact on Brazilian public health. Nevertheless, their mobilisable AMR features have potential broad dissemination capacity, biologically and geographically. Therefore, besides reduction of AMU, study of the barriers / factors / intermediate steps influencing the transfer of AMR genes can help in designing effective intervention strategies to reduce antimicrobial resistance. These insights were revealed through the concurrent evaluation of Salmonella genomes from Brazilian poultry, imported Brazilian chicken meat, and domestic human salmonellosis cases, and reinforce the importance of a One Health approach, which includes genomic surveillance. [END] --- [1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1010174 Published and (C) by PLOS One Content appears here under this condition or license: Creative Commons - Attribution BY 4.0. via Magical.Fish Gopher News Feeds: gopher://magical.fish/1/feeds/news/plosone/