Evaluation of proteome reference maps for cross-species identification of proteins by peptide mass fingerprinting

Abstract

We tested whether proteome reference maps established for one species can be used for cross-species protein identification by comparing two-dimensional protein gel patterns and protein identification data of two closely related bacterial strains and four plant species. First, proteome profiles of two strains of the fully sequenced bacterium Sinorhizobium meliloti were compared as an example of close relatedness, high reproducibility and sequence availability. Secondly, the proteome profiles of three legumes (Medicago truncatula, Melilotus alba and Trifolium subterraneum), and the nonlegume rice (Oryza sativa) were analysed to test cross-species similarities. In general, we found stronger similarities in gel patterns of the arrayed proteins between the two bacterial strains and between the plant species than could be expected from the sequence similarities. However, protein identity could not be concluded from their gel position, not even when comparing strains of the same species. Surprisingly, in the bacterial strains peptide mass fingerprinting was more reliable for species-specific protein identification than N-terminal sequencing. While peptide masses were found to be unreliable for cross-species protein identification, we present useful criteria to determine confident matching against species-specific expressed sequence tag databases. In conclusion, we present evidence that cautions the use of proteome reference maps and peptide mass fingerprinting for cross-species protein identification.

Proteomic Analysis of Soybean Root Hairs After Infection by Bradyrhizobium japonicum

“…For example, in the case of M. truncatula, which has a partially sequenced genome and a large collection of EST sequences, the positive identification or matching rate using PMF data reached 55% (Watson et al 2003). Even in a few cases, a certain degree of success in identifying putative candidates can be achieved by querying PMF data against other organisms (Mathesius et al 2002). Recently, Kwon and associates (2003) demonstrated that EST databases can be used reliably for high-throughput proteomic studies in an organism without genome information using MS/MS data.…”

The pitfalls of proteomics experiments without the correct use of bioinformatics tools

“…The NADH dehydrogenase subunit 5 (Fragment) of the butterfly, Parnassius ruckbeili ruckbeili (Deckert) (TrEMBL accession number: Q76JY5) has an important number of theoretical peptides with mass values )3000 Da. Concerning the pI and MW, both criteria will be relatively close to theoretical values for species with complete genome sequences in databases, and as far as cross-species identification is concerned, and in order to avoid false positive identification, a molecular mass variation of 630% and a pI variation of 62.0 are generally used [6,45,80,81,91,103,108].…”

“…In any given proteomics study, a key step during the research is the identification of proteins linked to the biological phenomenon under investigation, whatever this may be exactly [5,52,[80][81][82]. This imperative has stimulated the development of many new instruments [83][84][85][86][87][88][89][90] and techniques [90,91], especially protein databases [92] and bioinformatics tools [7,93,94].…”

Plant Protein Analysis

“…Unfortunately the data can be ambiguous and rely heavily on availability of a proteomic or genomic sequence or at least a substantial EST collection for the species being studied. Cross-species PMF studies from four plant species (Mathesius et al 2002) for instance concluded that PMF data are not particularly useful for inter-species protein identification except for the highly conserved proteins.…”