Abstract
For the archaeon Methanococcus maripaludis, a fully sequenced and annotated model species of hydrogenotrophic methanogen, we report validation of quantitative protein level expression ratios on a proteome-wide basis. Using an approach based on quantitative multidimensional capillary HPLC and quadrupole ion trap mass spectrometry, coverage of gene expression approached that currently achievable with transcription microarrays. Comprehensive mass spectrometry-based proteomics and spotted cDNA arrays were used to compare global protein and mRNA levels in a wild-type (S2) and mutant strain (S40) of M. maripaludis. Using linear regression with 652 expression ratios generated by both the proteomic and microarray methods, a product moment correlation coefficient of 0.24 was observed. The correlation improved to 0.61 if only genes showing significant expression changes were included. A novel two-stage method of outlier detection was used for the protein measurements when Dixon's Q-test by itself failed to give satisfactory results. The log(2) transformations of the number of peptides or isotopic peptide pairs associated with each ORF, divided by the predicted molecular weight, were found to have moderately positive correlations with two bioinformatic predictors of gene expression based on codon bias. We detected peptides derived from 939 proteins or 55% of the genome coding capacity. Of these, 60 were overexpressed, and 34 were underexpressed in the mutant. Of the 1722 ORFs encoded in the genome, 1597 or 93% were probed by cDNA arrays. Of these, 50 were more highly expressed, and 45 showed lower expression levels in the mutant relative to the wild type. 15 ORFs were shown to be overexpressed by both methods, and two ORFs were shown to be overexpressed by proteomics and underexpressed by microarray.