Temporal Genomic Phylogeny Reconstruction Indicates a Geospatial Transmission Path of Salmonella Cerro in the United States and a Clade-Specific Loss of Hydrogen Sulfide Production

Abstract

Salmonella Cerro has become one of the most prevalent Salmonella serotypes isolated from dairy cattle in several U.S. states, including New York where it represented 36% of all Salmonella isolates of bovine origin in 2015. This serotype is commonly isolated from dairy cattle with clinical signs of salmonellosis, including diarrhea and fever, although it has also been identified in herds without evidence of clinical disease or decreased production. To better understand the transmission patterns and drivers of its geographic spread, we have studied the genomic similarity and microevolution of S. Cerro isolates from the northeast U.S. and Texas. Eighty-three out of 86 isolates were confirmed as multilocus sequence type 367. We identified core genome SNPs in 57 upstate New York (NY), 2 Pennsylvania (PA), and 27 Texas S. Cerro isolates from dairy cattle, farm environments, raw milk, and one human clinical case and used them to construct a tip-dated phylogeny. S. Cerro isolates clustered in three distinct clades, including (i) clade I (n = 3; 2013) comprising isolates from northwest Texas (NTX), (ii) clade II (n = 14; 2009-2011, 2014) comprising isolates from NY, and (iii) clade III comprising isolates from NY, PA, and central Texas (CTX) in subclade IIIa (n = 45; 2008-2014), and only CTX isolates in subclade IIIb (n = 24; 2013). Temporal phylogenetic analysis estimated the divergence of these three clades from the most recent common ancestor in approximately 1980. The CTX clade IIIb was estimated to have evolved and diverged from the NY ancestor around 2004. Furthermore, gradual temporal loss of genes encoding a D-alanine transporter, involved in virulence, was observed. These genes were present in the isolates endemic to NTX clade I and were gradually lost in clades II and III. The virulence gene orgA, which is part of the Salmonella Pathogenicity Island 1, was lost in a subgroup of Texas isolates in clades I and IIIb. All S. Cerro isolates had an additional cytosine inserted in a cytosine-rich region of the virulence gene sopA, resulting in premature termination of translation likely responsible for loss of pathogenic capacity in humans. A group of closely related NY isolates was characterized by the loss of hydrogen sulfide production due to the truncation or complete loss of phsA. Our data suggest the ability of Salmonella to rapidly diverge and adapt to specific niches (e.g., bovine niche), and to modify virulence-related characteristics such as the ability to utilize tetrathionate as an alternative electron acceptor, which is commonly used to detect Salmonella. Overall, our results show that clinical outcome data and genetic data for S. Cerro isolates, such as truncations in virulence genes leading to novel pheno- and pathotypes, should be correlated to allow for accurate risk assessment.

Keywords: Salmonella enterica subsp. enterica serotype Cerro; WGS; dairy; emerging pathogen; epidemiology; hydrogen sulfide; virulence genes.

FIGURE 1

Tip-dated phylogeny of 86 Salmonella enterica subsp. enterica serotype Cerro isolates from NY (isolate names in bold print), PA (isolate names in italic print), CTX (isolate names in regular print) and NTX (isolate names in regular print), isolated from cattle (C; data on clinical symptoms unavailable), cattle with clinical symptoms (CC), cattle farm environment (CFE), produce farm environment (PFE), food (F) and human clinical case (HC). The unrooted tree was constructed based on high quality core genome single nucleotide polymorphisms (SNPs) in BEAST using general time-reversible (GTR) substitution model, strict molecular clock model and coalescent constant size population model, in four independent runs of 100,000,000 Monte Carlo Markov Chain (MCMC) simulations. The violet bars on the nodes represent credible intervals for time of taxa divergence. The cluster of isolates with abolished hydrogen sulfide production is marked with turquoise circles. The blue squares represent from 0 to 5 detected genes in D-alanine transporter-encoding cluster in following order from left to right: STM1633, ST1634, ST1635, STM1636, and STM1637. Purple hexagons represent isolates that carry orgA gene encoding oxygen-regulated invasion protein.

FIGURE 2

Accessory genome-based PCA clustering of 86 (A) and 84 (B) Salmonella Cerro isolates indicates distinct gene patterns of isolates from NTX, compared to isolates from upstate NY, PA, and CTX. PCA analysis was performed in R using prcomp function based on presence/absence data for 1293 genes that were part of an accessory genome of this isolate set. First three principal components were plotted in R using ggplot. The CTX and PA isolates are not visible in the figure, since the dots representing these isolates are covered with the dots representing NY isolates.