Escherichia coli (E. coli) is both, a facultative commensal of the gut microbiota and an important bacterial cause of human as well as animal diseases. Due to the high genetic plasticity, gene transfer allows those bacteria to colonize different sites in the host acting as pathogens, being responsible for a wide range of infections [1, 67]. For toxins alone, variants have been described with respect to functional changes [44, 70, 132]. Based on of phylogenetic differences, the hypothesis of this thesis is that allelic variants of toxins differ in their biological effect with regard to their toxicity. Therefore, in silico analyses were performed to unravel sequence differences for all known toxins encoded in several E. coli and Shigella genomes. The gene sequence research of all currently published toxins until June 2016 revealed a total set of 39 toxin genes being present in the genome of E. coli and Shigella spp. To test the hypothesis, the selected reference toxin sequences were used as templates to screen the NCBI database as well as an internal database using the BLAST algorithm. The internal database consists of 423 whole genome sequences of the “ECore” dataset [140] and 69 Shiga-Toxin-producing E. coli whole genome sequences [31]. To identify genetic variations of each toxin gene, the software Geneious was used, providing alignments and phylogenetic trees. After identification of allelic variations, the gene sequences were translated into their protein sequences. Finally, the resulting secondary and three-dimensional protein structures were predicted in order to receive indications of potentially functional changes. Detected allelic protein structures were modelled via the open source platform I-Tasser and visualised by the python based software PyMol. The results show that genetic variations of all 39 toxin genes are available in both databases. The effector protein EspH and the metalloprotease YghJ were analyzed for the first time by their three-dimensional structures, with regard to possible functional changes for their resulting protein variants. Different variants of the Subtilase Cytotoxin SubAB were predicted and analyzed with regard to a different pathogenic potential and toxicity as was shown for the different Shiga-Toxin variants. Furthermore, the three-dimensional protein variants of the serine protease autotransporter of Enterobacteriaceae (SPATEs), indicate that the SPATE members are already allelic variations by name and split into two known functionally different classes [132]. The inclusion of the Shiga-Toxin variants [44, 111] revealed the same results, indicating that the chosen bioinformatics methods are sound and comparable. All the other identified allelic variations of toxins in the genome of E. coli and Shigella spp. due to in silico analyses had no impact on the resulting three-dimensional protein models, indicating a conserved function as in the case of EspF. This thesis gives a first overview about the occurrence and impact of toxin variants in the genome of E. coli and Shigella spp. on the basis of bioinformatics analysis. The “toxome” allows the screening of all currently known toxin genes and its variants in the whole genome sequence of the mentioned bacteria with a BLASTn algorithm.
