Proteomic analysis of cellular soluble proteins from human bronchial smooth muscle cells by combining nondenaturing micro 2DE and quantitative LC-MS/MS. 2. Similarity search between protein maps for the analysis of protein complexes

1DE-MS Profiling for Proteoform-Correlated Proteomic Analysis, by Combining SDS-PAGE, Whole-Gel Slicing, Quantitative LC-MS/MS, and Reconstruction of Gel Distributions of Several Thousands of Proteins

SDS-PAGE has often been used in proteomic analysis, but generally for sample prefractionation although the technique separates proteins by molecular masses (M_ws) and the information would contribute to proteoform-level analysis. Here, we report a method that combines SDS-PAGE, whole-gel slicing, and quantitative LC-MS/MS for establishing gel distributions of several thousand proteins in a proteome. A previously obtained data set on rat cerebral cortex with cerebral ischemia-reperfusion injury¹ was analyzed, and the gel distributions of 5906 proteins were reconstructed. These distributions, referred to as 1DE-MS profiles, revealed that about 30% of the proteins had more than one proteoform detected in the gels. The profiles were categorized into six types by distribution (narrow, dispersed, or broad) and relative deviations between the abundance-peak apparent M_ws and calculated M_ws. Only 56% of the proteins showed narrow distributions and matched M_ws, while the others had rather complex profiles. Bioinformatic analysis on example profiles showed the resolved proteoforms involved alternative splicing, proteolytic processing, glycosylation and ubiquitination, fragmentation, and probably transmembrane structures. Profile-based differential analysis revealed that many of the disease-caused changes were proteoform dependent. This work provided a proteome-scale view of protein distributions in SDS-PAGE gels, and the method would be useful to obtain proteoform-correlated information for in-depth proteomics.

Visualized heterooligomeric subunit structures of 817 human cellular proteins by correlating native protein 2D maps with protein interaction databases

Previously, we have reported on the reconstruction of 4323 native protein 2D maps of human bronchial smooth muscle cells (HBSMC) by combining nondenaturing 2DE, grid gel-cutting, and quantitative LC-MS/MS. In this work, we report on the visualization of the heterooligomeric subunit structures of HBSMC proteins by correlating the native protein 2D maps with the information in protein interaction databases. Image analysis of the 2D maps was employed as the first approach. Each of the native protein 2D maps of 2327 proteins, which had three or more detected squares in the native protein 2D maps, was compared with the 2D maps of the remaining 2326 proteins scoring the degree of overlap in an area around the quantity peak and the protein partner which showed the best score was decided. The protein pairs were examined on their reported interactions referring to protein interaction databases. In order to consider the cases where a protein has multiple functions and the heterooligomeric subunit structures might not be detected from the image analysis, prior database search was employed as the second approach. Each of the 1689 HBSMC proteins, which had five or more detected squares in the native protein 2D maps, was examined on its interactor proteins described in the databases, then the native 2D map was compared with the maps of the interactor proteins to find the overlap which reasonably supported the interaction. Summarizing these examinations, 215 heterooligomeric subunit structures of 817 human cellular proteins could be visualized on the native protein 2D maps.

Simultaneous speculation of 401 monomeric or homo-oligomeric subunit structures of human cellular proteins, mining the information in 1901 native 2D protein maps reconstructed from one nondenaturing 2DE gel

Subunit structures of proteins are essential for their properties and functions, but there is a lack of method for global detection of the status of proteins being monomers or homo-oligomers. In this work, we report on a new method to simultaneously speculate hundreds of monomeric or homo-oligomeric subunit structures of cellular proteins, based on in-depth analysis of native 2D protein maps. Previously we have reported on the analysis of soluble proteins of human bronchial muscle cells (HBSMC) by combining nondenaturing 2DE, grid gel-cutting and quantitative LC-MS/MS. Totally 4323 proteins were detected and for each protein the quantity distribution on the gel was reconstructed as a native 2D map. In this work, this large dataset of maps were further mined with bioinformatic analysis. The native 2D maps of 1901 HBSMC proteins that were detected in at least five out of the grid-cut 972 gel squares were examined and 658 proteins that showed one major quantity-peak distribution were subjected to further analysis. After excluding those that mainly formed hetero-oligomeric structures, the monomeric or homo-oligomeric subunit structures of 505 proteins were speculated. The quotient of the apparent molecular mass of the quantity-peak position on the native 2D map divided by the theoretical molecular mass was calculated for each protein, to speculate the number of monomers which constituted its subunit structure. The suggested composition was then compared with the "Subunit structure" record of the protein in UniProtKB. When the database record included possible interactions with other proteins, their native 2D maps were extracted from the native map dataset, presented together and compared to confirm the prominent subunit structure. With this new approach, the monomeric or homo-oligomeric subunit structures of 401 proteins were speculated. Among them, 162 proteins had the speculated subunit structures coinciding with their database records, and 91 proteins with matched database records as being monomers or homo-oligomers but mismatched at the numbers of the composing monomers. For 148 proteins that did not have database record, their subunit structures were newly speculated. We expect this method, combining nondenaturing 2DE separation with in-depth proteomic and bioinformatic analysis, would suggest a means to achieve large-scale information on monomeric and homo-oligomeric subunit structures of cellular proteins.

Proteomic analysis of cellular soluble proteins from human bronchial smooth muscle cells by combining nondenaturing micro 2DE and quantitative LC-MS/MS. 1. Preparation of more than 4000 native protein maps

Soluble proteins of human bronchial smooth muscle cells (HBSMC) were separated by nondenaturing micro 2DE and a 30 mm × 40 mm area of the CBB-stained slab gel (1.0 mm thick) was cut into 1.1 mm × 1.1 mm squares, then the proteins in the 972 gel pieces (squares) were applied to quantitative LC-MS/MS. Grid-cutting of the gel was employed to; (i) ensure the total analysis of the proteins in the area, (ii) standardize the conditions of analysis by LC-MS/MS, (iii) reconstruct the protein distribution patterns from the quantity data [1]. Totally 4323 proteins were identified in successfully analyzed 967 squares and the quantity distribution of each was reconstructed as a color density pattern (a native protein map). The quantity of the proteins distributed from 3.6% to 1 × 10(-5) % of the total protein quantity in the grid area. Each protein map was characterized by several features, including the position of quantity peak square, number of detected squares, and degree of concentration (focused or dispersed). About 4% of the proteins were detected in 100 or more squares, suggesting that they might be ubiquitous and interacting with other proteins. In contrast, many proteins showed more concentrated quantity distribution and the quantity peak positions of 565 proteins with a defined degree of concentration were summarized into a quantity peak map. These results for the first time visualized the distribution patterns of cellular proteins on a nondenaturing 2D gel.