Multiclonal human origin and global expansion of an endemic bacterial pathogen of livestock

Abstract

Most new pathogens of humans and animals arise via switching events from distinct host species. However, our understanding of the evolutionary and ecological drivers of successful host adaptation, expansion, and dissemination are limited. Staphylococcus aureus is a major bacterial pathogen of humans and a leading cause of mastitis in dairy cows worldwide. Here we trace the evolutionary history of bovine S. aureus using a global dataset of 10,254 S. aureus genomes including 1,896 bovine isolates from 32 countries in 6 continents. We identified 7 major contemporary endemic clones of S. aureus causing bovine mastitis around the world and traced them back to 4 independent host-jump events from humans that occurred up to 2,500 y ago. Individual clones emerged and underwent clonal expansion from the mid-19th to late 20th century coinciding with the commercialization and industrialization of dairy farming, and older lineages have become globally distributed via established cattle trade links. Importantly, we identified lineage-dependent differences in the frequency of host transmission events between humans and cows in both directions revealing high risk clones threatening veterinary and human health. Finally, pangenome network analysis revealed that some bovine S. aureus lineages contained distinct sets of bovine-associated genes, consistent with multiple trajectories to host adaptation via gene acquisition. Taken together, we have dissected the evolutionary history of a major endemic pathogen of livestock providing a comprehensive temporal, geographic, and gene-level perspective of its remarkable success.

Keywords: Staphylococcus aureus; agriculture; host adaptation; phylodynamics; population genomics.

Fig. 1.

Global distribution of bovine S. aureus isolates examined in the study according to CC. The map shows the sampling locations of the 1,647 bovine isolates with available data among the 1,896 total bovine isolates collected. The blue color scale filling each country represents the number of bovine S. aureus genomes included in the analyses (log scale). Pie charts indicate the distribution of these genomes per CC in each of the continents represented.

Fig. 2.

Contemporary bovine S. aureus originated from human-to-bovine host switches that occurred in the last 2,500 y. Time-scaled tree of the global, bovine-enriched S. aureus dataset (n = 3,915) with host ancestral reconstruction. The maximum-likelihood tree was generated with IQ-Tree2, dated using BactDating, and host ancestral states were inferred using SIMMAP. Branches are colored according to the host yielding the highest probability. Main CCs are collapsed in triangles and colored according to their reconstructed original host. Nodes supported by 100% bootstrap are indicated by an asterisk. Key nodes are annotated with their predicted most recent common ancestors (in years before present). Arrows point to the main evolutionary jumps into bovine population.

Fig. 3.

Phylogeographic analysis of the bovine host-restricted S. aureus CC151 based on core and accessory genome. Left: Bayesian time-stamped tree from a core genome alignment (1,645,057 bp, of which 14,665 bp were variable sites) of CC151 genome sequences. Branches colored according to the reconstructed locations in the discrete trait analysis. Inset: graphic summary of migrations between countries, in which the thickness of arrows is proportional to the number of migration events inferred. Right: Network or accessory genome of the same CC151 sequences (based on 483 accessory genes, defined as genes in more than 1 genome, and not in all genomes). Node colors correspond to sampling locations in both the tree and the network.

Fig. 4.

Lineage-dependent differences in the frequency of host transitions reveal high-risk S. aureus clones. (A) CC-specific analysis of ancestral host reconstruction. Analyses for each CC include a time-stamped phylogenetic tree generated using the MASCOT model in BEAST2 to reconstruct ancestral host state; and a network representing the accessory genome of the same samples. Each node represents a genome, and edges are weighted as the similarity between two genomes. The number of S. aureus genomes used for each analysis was CCC151: 246 bovine; CC97: 175 bovine, 35 human, 8 ovine; CC188: 86 bovine, 63 human; CC1: 67 bovine, 121 human; CC425: 65 bovine, 15 human; CC133: 98 bovine, 19 ovine; CC130: 23 bovine, 86 human, 25 ovine (SI Appendix, Fig. S1 and Dataset S1). (B) Percentage of S. aureus transitions between bovine and human host species over the total of transitions per tree in the BEAST2 tree posterior distribution. CC1 and CC188 are CCs with human origins, CC97 and CC425 with bovine origins, CC130 with ovine origin. CC133 not shown as no human isolates included in the analyses.

Fig. 5.

Routes of international dissemination for bovine S. aureus revealed by ancestral trait analysis. This analysis was performed on the bovine-enriched S. aureus dataset (n = 3,915) which included 1,614 bovine assemblies with known sampling time and locations. (A) Schematic worldwide map showing inferred location transitions within the bovine population. Nodes represent each location (individual countries or groups of countries) included in the analysis. Arrows represent estimated, directional migrations (i.e., location state changes) between locations. Arrows width and color are proportional to the estimated amount of migrations. (B) Correlation between age (in years before present) of the CCs of ancestral bovine and human origins and the Simpson’s D diversity index calculated from the number of countries from which they were identified in our dataset, showing a statistically significant positive correlation.

Fig. 6.

Network analysis of the accessory genome reveals host species and CC-dependent gene combinations. Pairwise similarity network of 4,841 S. aureus isolates based on shared accessory genes. For host species association see SI Appendix, Fig S1). Each node represents a genome, and edges are weighted as the similarity between two genomes, with lineage-dependent accessory genes removed. Similarity was calculated using the Jaccard Index, and the network visualized within Graphia using a force-directed 2D layout. A k-NN (k = 10) algorithm was applied to reduce edge density, and components of less than 50 nodes removed. The network is colored by the host species, and important CCs are highlighted, colored by the primary host species of the CC or subcluster (dotted gray lines).

Fig. 7.

Bovine-specific S. aureus genes are associated with phage, SaPIs, and genomic islands. Bovine-associated genes were mapped to the bovine reference genome RF122 to model genomic location. The central density plot represents the distribution of all bovine-associated genes identified in this study. Circle dot plots represent the location of bovine associated genes as identified in, from inner-most to outward, CC1 (green), CC97 (pink), CC188 (purple) and the full dataset (gray; *the bovine exclusive lineages CC126 and CC151 core genes are included here). Known genomic features of interest are displayed as blocks on the outermost plot.