Comparative genome-scale modelling of Staphylococcus aureus strains identifies strain-specific metabolic capabilities linked to pathogenicity

Abstract

Staphylococcus aureus is a preeminent bacterial pathogen capable of colonizing diverse ecological niches within its human host. We describe here the pangenome of S. aureus based on analysis of genome sequences from 64 strains of S. aureus spanning a range of ecological niches, host types, and antibiotic resistance profiles. Based on this set, S. aureus is expected to have an open pangenome composed of 7,411 genes and a core genome composed of 1,441 genes. Metabolism was highly conserved in this core genome; however, differences were identified in amino acid and nucleotide biosynthesis pathways between the strains. Genome-scale models (GEMs) of metabolism were constructed for the 64 strains of S. aureus These GEMs enabled a systems approach to characterizing the core metabolic and panmetabolic capabilities of the S. aureus species. All models were predicted to be auxotrophic for the vitamins niacin (vitamin B3) and thiamin (vitamin B1), whereas strain-specific auxotrophies were predicted for riboflavin (vitamin B2), guanosine, leucine, methionine, and cysteine, among others. GEMs were used to systematically analyze growth capabilities in more than 300 different growth-supporting environments. The results identified metabolic capabilities linked to pathogenic traits and virulence acquisitions. Such traits can be used to differentiate strains responsible for mild vs. severe infections and preference for hosts (e.g., animals vs. humans). Genome-scale analysis of multiple strains of a species can thus be used to identify metabolic determinants of virulence and increase our understanding of why certain strains of this deadly pathogen have spread rapidly throughout the world.

Keywords: core genome; mathematical modeling; pangenome; pathogenicity; systems biology.

Fig. 1.

S. aureus dataset construction. (A) Phylogenic tree of 225 S. aureus genomes based on seven housekeeping genes (arcC, aroE, glpF, gmk, pta, tpi, and yqiL). A set of 64 strains (labeled) were selected from this set to create a heterogeneous dataset of S. aureus strains based on the evolutionary distance (tree topology), as well as (B) drug resistance (MRSA, MSSA, VRSA, and VISA), (C) host specificity (human vs. animal), and (D) virulence/environmental association (CA-MRSA, community-associated MRSA; HA-MRSA, healthcare-acquired MRSA; LA-MRSA, livestock-associated MRSA). Evolutionary distance is based on tree topology.

Fig. 2.

S. aureus pangenome statistics. (A) The S. aureus pangenome can be subdivided into three categories: (i) the core genome (the set of genes shared by all genomes), (ii) the accessory genome (the set of genes present in some but not all genomes), and (iii) the unique genome (genes that are unique to a single genome). The function of each gene in a group is classified using COGs. COG categories are as follows: For cellular processes and signaling, D is cell cycle control, cell division, and chromosome partitioning; M is cell wall/membrane/envelope biogenesis; N is cell motility; O is posttranslational modification, protein turnover, and chaperones; T is signal transduction mechanisms; U is intracellular trafficking, secretion, and vesicular transport; V is defense mechanisms; W is extracellular structures; Y is nuclear structure; and Z is cytoskeleton. For information storage and processing, A is RNA processing and modification; B is chromatin structure and dynamics; J is translation, ribosomal structure, and biogenesis; K is transcription; and L is replication, recombination, and repair. For metabolism, C is energy production and conversion; E is amino acid transport and metabolism; F is nucleotide transport and metabolism; G is carbohydrate transport and metabolism; H is coenzyme transport and metabolism; I is lipid transport and metabolism; P is inorganic ion transport and metabolism; and Q is secondary metabolites biosynthesis, transport, and catabolism. (B) Distribution of the genes in the pangenome for each examined S. aureus strain.

Fig. 3.

Pangenome, core, and novel genes of the 64 analyzed S. aureus strains. Pangenome features are as follows: The purple squares denote the number of novel genes discovered with the sequential addition of new genomes. The yellow dots denote the values of the core genes as genomes are added to the pangenome. The purple bars indicate the number of new genes added to the total pangenome size as new genomes are added. Each of the values represents the median from a distribution of randomly selected genomes at each genome addition. The purple line represents the number of new genes found for each genome addition. For comparison, the same trend for a closed genome is reported as a dashed line. The equations below the graph show parameters for fits to Heap’s law. Positive exponents indicate an open state and that the category is boundless so new genes are likely to be discovered continually as new genomes are sequenced.

Fig. S1.

Conservation of translation machinery in S. aureus strains. Genes involved in transcription and translation were selected from E. coli (Ec) and B. subtilis (Bs) to search for the presence of homologous proteins in 64 S. aureus strains. (A) The total number of genes in each category. The results were grouped into three panels: (B) conserved core genes involved in transcription and translation, (C) genes lost in some strains only (strains aligned vertically and genes horizontally), and (D) genes absent in all strains. The query gene acronyms correspond to gene names given in Dataset S1 and are ordered from top to bottom, according to the eight protein categories: ribosomal proteins, tRNA aminoacylation, rRNA modifications, tRNA modifications, ribosome assembly, transcription, translation, and RNA processing according to coding next to A.

Fig. S2.

Reconstruction of gene gain and loss for the selected 64 strains of S. aureus. (A) A phylogeny for the S. aureus species was constructed based on the 1,441 identified core genes. All core genes were aligned to reconstruct the phylogeny of the S. aureus species based on its core genome. Gene gain/loss was calculated based on extrapolation of a last common ancestor for (B) all genes and (C) the virulence factors specifically. Full-size images are available in Dataset S2.

Fig. 4.

Core metabolic and panmetabolic capabilities of the S. aureus species. The core metabolic and panmetabolic content was determined for genome-scale metabolic models (GEMs) of 64 unique S. aureus strains. (A) The core content, illustrated by the intersection of the Venn diagram, is shared with all strains. The pancontent consists of all content in any model and includes the core content. Note that the Venn diagram is not to scale and is simplified to only include the first 4 out of n = 64 strains. (B) Classification of reactions in the core reactomes and panreactomes by metabolic subsystem.

Fig. S3.

Experimental growth screens on chemically defined media. Four S. aureus strains were grown for 24 h in different media compositions. TSB media is a standard chemically undefined media for S. aureus growth. A minimal media (CDM-Min) was defined based on M9-glucose media with addition of vitamins and proline, serine, and leucine. This media did not support growth of any S. aureus strains. Additional amino acids were added to the CDM-Min media to test growth capabilities including arginine, threonine, phenylalanine, and valine. A CDM-max media consisted of all amino acids except tyrosine and cysteine.

Fig. 5.

S. aureus virulence factors and predicted metabolic capabilities. (A) Presence and absence of virulence factors and predicted metabolic capabilities across the 64 S. aureus strains examined in this study. Metabolic capabilities were predicted using the strain-specific metabolic models (dark purple, growth capability; light purple, nutrient required; yellow, no growth or nutrient is not required). The virulome consists of curated virulence factors known to be present in different strains of S. aureus. Orange indicates a factor is present, and yellow indicates a factor is absent. Full matrix with strains, predicted growth capabilities, and virulence factor is available in Dataset S1. Virulence factor and growth profiles can be used to classify strains. For example in B, a classification is constructed that separates human-associated S. aureus strains from livestock-associated strains using only the presence of two virulence factors and the ability to catabolize maltotriose. Abbreviations are as follows: staphylokinase precursor (sak), maltotriose (malttr), and IgG binding protein A precursor (spa).

Fig. S4.

Experimental confirmation of USA300’s ability to grow in the presence of spermidine. S. aureus strains USA300, Newman, and ATCC8325 were grown in TSB media supplemented with 6 mM spermidine. Only USA300 strains were able to survive. Our metabolic models indicate that this is due to the presence of the speG gene encoding spermidine acetyltransferase (see main text).