Identification of new protein coding sequences and signal peptidase cleavage sites of Helicobacter pylori strain 26695 by proteogenomics

Correct annotation of protein coding genes is the basis of conventional data analysis in proteomic studies. Nevertheless, most protein sequence databases almost exclusively rely on gene finding software and inevitably also miss protein annotations or possess errors. Proteogenomics tries to overcome these issues by matching MS data directly against a genome sequence database. Here we report an in-depth proteogenomics study of Helicobacter pylori strain 26695. MS data was searched against a combined database of the NCBI annotations and a six-frame translation of the genome. Database searches with Mascot and X!Tandem revealed 1115 proteins identified by at least two peptides with a peptide false discovery rate below 1%. This represents 71% of the predicted proteome. So far this is the most extensive proteome study of H. pylori. Our proteogenomics approach unambiguously identified four previously missed annotations and furthermore allowed us to correct sequences of six annotated proteins. Since secreted proteins are often involved in pathogenic processes we further investigated signal peptidase cleavage sites. By applying a database search that accommodates the identification of semi-specific cleaved peptides, 63 previously unknown signal peptides were detected. The motif LXA showed to be the predominant recognition sequence for signal peptidases. BIOLOGICAL SIGNIFICANCE: Even so de novo sequencing algorithms were significantly optimized within the last years the method still has its limitations in terms of speed, quality and completeness. Hence, the result of each standard proteomic study highly relies on correct annotation of protein coding genes which is the basis of conventional data analysis. Due to the amazing technical improvements genome sequencing is well established and easily accessible for the science community and is no longer a bottleneck in research. However, the annotation of protein coding sequences in genomic data is usually based on gene finding software. These tools are limited in their prediction accuracy. It is typically problematic to determine exact gene boundaries. Furthermore it is even harder to additionally correctly predict signal peptides. In conclusion this can result in protein databases with erroneous, incomplete and even missing entries. Nevertheless proteogenomic evaluations are still rare. In this study we investigated the proteome of the human pathogen Helicobacter pylori (HP) - a human pathogen which infected about 50% of the world's population and is responsible for many gastric diseases like gastric and duodenal ulcers as well as gastric cancer. Using GeLC-MS and 2D-LC-MS and applied multiple proteases. Thus we were able to highly reliable identify 1115 proteins (FDR