T PILLONEL¹, C BERTELLI¹, J SCHRENZEL ², G GREUB<sup>1

1</sup>Center for Research on Intracellular Bacteria, Institute of Microbiology, University Hospital Center and University of Lausanne, Switzerland, ²Genomic Research Laboratory, Service of Infectious Diseases, Geneva University Hospitals, Geneva, Switzerland, ³Service of Infectious Diseases, University Hospital Center, Lausanne, Switzerland

Variations of S. aureus virulence are attributed i) to the presence/absence of virulence factors (VFs) encoded on mobile genetic elements (MGE), but also ii) to changes in VFs encoding genes part of the core genome as well as iii) to changes in the expression of any of these VFs. Recently, a case of severe septic shock with major exfoliative rash and multi-organ dysfunction was due to a community-acquired S. aureus. This strain was isolated in pure culture and its genome was sequenced to investigate the genetic basis of its high virulence. Multilocus sequence typing analysis identified this isolate as Sequence Type 8. Two publicly available VFs databases were used to annotate the virulome: VirulenceFinder 1.5 and the Virulence Factor Database (VFDB). The presence/absence of known VF was investigated using tBLASTn against the genome contigs. In addition, the genome was independently sequenced and analysed by the company Biomérieux. Those 3 approaches allowed the identification of 146 putative VFs. Surprisingly, only 38 were predicted both by Biomérieux and the 2 databases. These 38 proteins included protein A, fibronectin binding proteins, fibrinogen binding proteins, multiple enterotoxins, serine proteases as well as leukotoxins LukD and LukE. Contrarily to Biomérieux, both VFDB and VirulenceFinder had a hit for eta, the exfoliative toxin A, which was of particular interest for this genome due to the clinical picture with exfoliation. In depth analyses suggested that it was a false positive. Biomérieux classified the Clp protease and the HtrA serine protease as VFs, which was not the case of the two databases. Altogether, those results highlight the lack of consistency of VF databases that reflects the difficulty of database curation. The various definitions of VFs add another level of complexity in identifying which gene to include in a virulence database. Moreover, VFs present in all S. aureus strains (e.g. protein A) are generally not the focus of virulence analyses, which rather search for VFs part of the mobilome. These variable factors might explain very unusual severe clinical presentations such as the lethal case reported here.