MSACL 2022 Abstract

Introduction:
Mycobacterium species are notorious human pathogens and mycobacterial diseases are being increasingly reported worldwide. Our overarching goal is to narrow down the gap in diagnosing different species, subspecies, and strains of Mycobacterium to guide proper treatment. In this project, we address the limitation in the current diagnostic method such as species-specific PCR assays, targeted sequencing or MALDI-TOF-based protein profiling and commercial probes in the clinical laboratory which are finite to identifying only few species and can misidentify closely related taxa. To this end, we first analyzed peptides derived from secreted mycobacterial proteins which allowed the selection of unique peptides that able to identify the organisms at the species, subspecies and/or strain level. Finally, we developed a provisional pipeline, Peptide Taxonomy/ORganism Checker (PEP-TORCH) to identify specific Mycobacterial species at sub-strain level using untargeted proteomics data using mycobacterial cultures liquid media filtrates (MGIT).

Method:
MGIT samples were processed and tryptically digested prior to the analysis by mass-spectrometry. We have used Data Dependent Acquisition (DDA) mode to identify peptides in seventy MGIT samples with different Mycobacterium infections. Based on preliminary clinical information, the samples were divided into M. tuberculosis, M. avium, M. intracellualre, M. kansasi and M. abscessus. The list of peptides belonging to each species was searched against Unipept and analyzed to find unique or combination of peptides unique to species. The shortlisted peptides are cross-checked for their intensities in DDA mode. These are further taken forward to be targeted by parallel reaction monitoring (PRM) in MS and validate the set of peptides in particular species. Subsequently, for the pipeline development, the list of peptides from single patient sample was searched against Unipept using “pept2taxa” (API mode). Packages such as JsonLite, Tidyr, dplyr were used in R platform to generate the algorithm and the scripts prior to the extraction of unique and combined peptide markers for the identification of single species infection.

Outcomes:
Peptide markers have been utilized widely for differentiating bacterial species. To find the suitable peptide markers for each mycobacterium species, we did the respective taxonomy analysis for each peptide and summarized the detection frequency of each peptide. The taxonomy analysis of these tryptic peptides indicates that it is specific for M. tuberculosis complex, including M. tuberculosis, M. canettii and M. orygis, in which M. canettii and M. orygis are not common pathogens. In addition, for the analysis of M. abscessus (MAB), we identified 400 peptides that were analyzed together in Unipept. The Unipept result gave us in total of 703 Mycobacterium species along with 7 MAB subspecies and strains. The data was filtered based on >85% missing which yielded 23 species that were identified with the maximum number of peptides were considered for further analysis. Of 400 peptides, 76 peptides were identified that are unique to MABs and are not found in any other Mycobacterium species. This showed us promising results to consider them further to target for validation in MABs. In the aim of increasing good target peptides our second criteria were to identify peptides not unique as individual but unique as a set in MABs. Since PRM has more sensitivity to identify low abundant peptides which could be otherwise missed in some samples in DDA, a comprehensive list of the ~90 peptides are considered for PRM validation in MABs.

Subsequently, the pipeline, PEP-TORCH has been automated that could interrogate a list of tryptic peptides from one single patient to deliver the sub-species or strain level information extracting from the Unipept database. By far, the pipeline has been successful in identifying single species infection using unique and unique combination of peptide markers. To further demonstrate the effectiveness of the pipeline, in an ongoing process, this pipeline is being validated using 70 MGIT samples infected with different Mycobacterium species. The overall workflow of the PEP-TORCH depends on the tryptic peptides being existing as unique or existing in multiple but joint to unique species set. Deducing such information from Unipept, the pipeline first identifies species level. In the second loop, the subspecies and strains were searched to be either unique or joint to unique strain to finally identify the subspecies and strain levels. In the first step, tryptic peptides from one sample are searched against Unipept. In the database, we have information at the genus, species and subspecies levels which are unorganized. The first step of our pipeline is to sort and extract this information for the species/ strain level identification. As a proof of concept, one long Unipept Output from the tryptic peptide set from one patient sample was also manually searched for unique peptides and combined peptide unique to species.

Discussion:
Our analytical approach has several advantages including broad species coverage as well as high sensitivity and specificity when compared to the commonly used DNA probes or MALDI-TOF MS. To date, other pipeline can detect infection based on only unique peptides. On the contrary to other pipelines which can distinguish between species based on unique peptides, our pipeline also considers the non-unique but joint-unique to species, in case, the unique peptides are not identified in the tryptic peptide set.