Database search for identification of proteins and post-translational modifications

Identification of mammoth tissue proteins was performed using Proteinpilot 4.5 software (AB SCIEX). The raw MS spectrometry data was processed using Paragon^TM algorithm in the ProteinPilot software to search against UniProt Mammuthus primigenius database (134 entries) first and subsequently against UniProt mammalian database (1,375,125 entries), which contains 25,876 proteins of Loxodonta africana, 223 of Elephas maximus, and 134 of Mammuthus primigenius. Trypsin was selected as the digestion enzyme and methane-thiosulfonation as the cysteine alkylation method. All combinations (380 patterns) of one amino acid substitutions and all the described modifications (295) were taken into consideration. False discovery rate (FDR) analyses were performed using a reverse decoy database to reduce false positive hits. Proteins with FDR rates under 1% were considered as significantly identified proteins in this study. Protein-Pilot automatically clustered the identified proteins into groups that share common peptides so that the minimal set of justifiable identified proteins was listed. The protein within each group that can explain more spectral data with confidence is shown in the results as the primary protein of the group. Common potential contaminant proteins such as trypsin and keratins were rejected from the identified protein lists. For annotation analysis, each identified protein was replaced with the homologous human protein using Blast search against UniProt human protein database, and a set of the homologous human proteins was subjected to the database for Annotation, Visualization and Integrated Discovery (DAVID) analysis (http://david.abcc.ncifcrf.gov/).

Comprehensive identification of post-translationally modified peptides was also performed using Proteinpilot 4.5 software. To enhance confidence of the identification, database search was performed under more restricted conditions. Specifically, the raw MS spectrometry data was searched against UniProt database taxonomically restricted to L. Africana. Amino acid substitutions were not taken into consideration. Peptides identified with confidence lower than 95% were rejected. The other parameters or data processing were the same as those of the protein identification.