Population and pan-genomic analyses of Staphylococcus pseudintermedius identify geographic distinctions in accessory gene content and novel loci associated with AMR

Abstract

Staphylococcus pseudintermedius is a common representative of the normal skin microbiota of dogs and cats but is also a causative agent of a variety of infections. Although primarily a canine/feline bacterium, recent studies suggest an expanded host range including humans. This paper details population genomic analyses of the largest yet assembled and sequenced collection of S. pseudintermedius isolates from across the USA and Canada and assesses these isolates within a larger global population genetic context. We then employ a pan-genome-wide association study analysis of over 1,700 S. pseudintermedius isolates from sick dogs and cats, covering the period 2017-2020, correlating loci at a genome-wide level, with in vitro susceptibility data for 23 different antibiotics. We find no evidence from either core genome phylogenies or accessory genome content for separate lineages colonizing cats or dogs. Some core genome geographic clustering was evident on a global scale, and accessory gene content was noticeably different between various regions, some of which could be linked to known antimicrobial resistance (AMR) loci for certain classes of antibiotics (e.g., aminoglycosides). Analysis of genes correlated with AMR was divided into different categories, depending on whether they were known resistance mechanisms, on a plasmid, or a putatively novel resistance mechanism on the chromosome. We discuss several novel chromosomal candidates for follow-up laboratory experimentation, including, for example, a bacteriocin (subtilosin), for which the same protein from Bacillus subtilis has been shown to be active against Staphylococcus aureus infections, and for which the operon, present in closely related Staphylococcus species, is absent in S. aureus.IMPORTANCEStaphylococcus pseudintermedius is an important causative agent of a variety of canine and feline infections, with recent studies suggesting an expanded host range, including humans. This paper presents global population genomic data and analysis of the largest set yet sequenced for this organism, covering the USA and Canada as well as more globally. It also presents analysis of in vitro antibiotic susceptibility testing results for the North American (NA) isolates, as well as genetic analysis for the global set. We conduct a pan-genome-wide association study analysis of over 1,700 S. pseudintermedius isolates from sick dogs and cats from NA to correlate loci at a genome-wide level with the in vitro susceptibility data for 23 different antibiotics. We discuss several chromosomal loci arising from this analysis for follow-up laboratory experimentation. This study should provide insight regarding the development of novel molecular treatments for an organism of both veterinary and, increasingly, human medical concern.

Keywords: Staphylococcus pseudintermedius; antibiotic resistance.

Fig 1

Staphylococcus pseudintermedius pan-genome curves. (A) FDA-USDA data set from North America. (B) Global data set inclusive of FDA-USDA.

Fig 2

Staphylococcus pseudintermedius pan-genome statistics. (A) FDA-USDA North American data set. (B) Global data set inclusive of FDA-USDA.

Fig 3

Core-genome phylogeny of Staphylococcus pseudintermedius isolates from the FDA-USDA North American data set, derived from cat and dog hosts inferred with IQ-TREE v.2.0.3, showing absence of host-specific lineages.

Fig 4

PCA of accessory gene content of Staphylococcus pseudintermedius isolates from different sample sets and perspectives. (A) Cat and dog FDA-USDA isolates, showing no clear distinction in gene content between hosts. (B) FDA-USDA isolates from different geographic regions showing no difference in gene content between areas. (C) Global comparative set showing different gene content between several regions, such as North America, E. Asia, and E. Europe. North America: USA, Canada; Oceania: New Zealand, Australia; W Europe: Spain, Netherlands, Germany, France, Italy, Ireland, United Kingdom, Switzerland, Belgium; E Asia: China, South Korea, Hong Kong, Japan; SE Asia: Thailand; E Europe: Slovenia, Hungary, Poland, Russia, Czech Republic; S Asia: Sri Lanka, India; N Europe: Norway, Sweden, Denmark; South America: Argentina, Puerto Rico, Brazil, Grenada; Africa: Botswana; Middle East: Israel, Turkey.

Fig 5

Staphylococcus pseudintermedius MLST results. (A) North American FDA-USDA set of isolates. (B) Global set of isolates. For both sample sets, the number of typable isolates was far less than the total number of isolates because of unique MLST allelic combinations not found on the pubMLST database. See Fig. 4 legend for countries included for each region. The map was plotted with the online tool https://www.mapchart.net/.

Fig 6

Bayesian analysis of population structure with hierarchical clustering (hierBAPS) of Staphylococcus pseudintermedius genome sequence data. (A) North American FDA-USDA isolates. (B) Global set of isolates. See the Fig. 4 legend for countries included for each region.

Fig 7

Trends in Staphylococcus pseudintermedius lab-tested antibiotic resistance for the period 2017–2020. The total number of samples included in each antibiotic trend analysis is indicated under each antibiotic, and standard error bars are marked on each point.

Fig 8

Percentage of lab-tested Staphylococcus pseudintermedius multidrug-resistant FDA-USDA isolates in different North American regions; 3 to 5 = resistant to three to five different classes of antibiotics; >5 = more than five classes of antibiotics. For category >5 antibiotics: chi-squared = 37.506, df = 4, P-value = 1.417e − 07 and multiple comparisons: South was significantly different from Canada (P = 0.00068) and from Midwest (P = 0.00049). Midwest was significantly different from Canada (P = 0.03709), and West was significantly different from South (P = 0.04095). All other comparisons were non-significant. For category 3–5 antibiotics: chi-squared = 7.0415, df = 4, P-value = 0.1337 (not significant).

Fig 9

Percentage of multidrug resistance loci in Staphylococcus pseudintermedius genomes. (A) North American FDA-USDA isolates. (B) Global set of isolates. For NA >5 antibiotics: chi-squared = 20.615, df = 4, P-value = 0.0003775 and multiple comparisons: Canada was significantly different from Northeast (P = 0.0051) and from South (P = 0.0067). For NA 3–5 antibiotics: chi-squared = 11.494, df = 4, P-value = 0.02154 and multiple comparisons: South was significantly different from Midwest (P = 0.042). Statistics are not provided for the global set because many regions had few samples.

Fig 10

Percentage of Staphylococcus pseudintermedius genomes, from the global set of isolates, with known AMR loci to various antibiotic classes.

Fig 11

Pan-GWAS (Scoary) results for tested antibiotics. The numbers above the pie diagrams are the numbers of genes judged to be significantly correlated with resistance for that antibiotic, and their proportions are broken down into the four indicated components in these pie diagrams. AMR + plasmid = known AMR locus carried on a plasmid; plasmid + not AMR = gene carried on a plasmid that is not a known AMR locus; AMR + not plasmid = known AMR locus on the chromosome; not_AMR + not_plasmid = not a known AMR locus carried on the chromosome.