K-mer analysis of blaACT genes for fast and reliable delineation of members of the Enterobacter cloacae complex

International audience; Objectives: On the last decade, ECC members has been defined by Hoffmann and Roggenkamp in 13 genetic clusters (I to XIII) based most exhaustively on hsp60 sequencing. More recently, Chavda et al. in 2016 proposed a new phylogenic classification by using both average nucleotide identity and a single-nucleotide polymorphism based phylogeny. The aim of this study is to improve the determination method based on hsp60.Methods: All the available ECC genomes (1.324) in the patric database v3.5.27 have been downloaded. The correct specie identification and the absence of contaminant specie were checked using the rMLST data base provided by PubMLST. The hsp60, rpoB and ampC genes were extract form all the genomes and clustered using cd-hit with an 99% identical cut off. A distance matrix between all the genomes based on the Kmer content was determined using Mash software and a neighbor joining tree (NJT) was constructed using the neighbor software available in the PHYLIP package. The tree and the molecular typing of all the strains were represented using itol.Results: Using Mash allowed to obtain a distance matrix of more than a thousand genomes in less than one hour. Clusters recovered in the NJT were in ad equation with the hsp60 clustering method for C11, C2, C5, C7. Nevertheless, genomes which belonged to the same clusters such as C9 or C4 appeared to be very divergent to each other. Furthermore, genomes which belonged to C13 seemed to be miss-classified C6. As the consequence, the hsp60 method could be not enough discriminant and precise to allow a correct clusterization and can lead to typing errors.The diversity of the rpoB gene sequences didn’t explained the phylogenetic clusters observed. Especially, it appeared that, C8 and C6 clusters shared the same rpoB sequences, as the consequence of what, rpoB sequences can’t be used as a clustering method. Interestingly, the diversity of the blaACT sequences improved the discrimination between the isolates and avoided mis-clusterization.Conclusion: The findings in this study provide an additional approach to identify in a more precise way and easily the cluster to which belongs the isolate of ECC.