Methodological Aspects of μLC-MS/MS for Wide-Scale Proteomic Analysis of Anthracycline-Induced Cardiomyopathy

Abstract

The efforts to utilize microflow liquid chromatography hyphenated to tandem mass spectrometry (μLC-MS/MS) for deep-scale proteomic analysis are still growing. In this work, two-dimensional LC separation and peptide derivatization by a tandem mass tag (TMT) were used to assess the capability of μLC-MS/MS to reveal protein changes associated with the severe chronic anthracycline cardiotoxicity phenotype in comparison with nanoflow liquid chromatography (nLC-MS/MS). The analysis of the control and anthracycline-treated rabbit myocardium by μLC-MS/MS and nLC-MS/MS allowed quantification of 3956 and 4549 proteins, respectively, with 84% of these proteins shared in both data sets. Both nLC-MS/MS and μLC-MS/MS revealed marked global proteome dysregulation in severe anthracycline cardiotoxicity, with a significant change in approximately 55% of all detected proteins. The μLC-MS/MS analysis allowed less compressed and more precise determination of the TMT channel ratio and correspondingly broader fold-change protein distribution than nLC-MS/MS. The total number of significantly changed proteins was higher in nLC-MS/MS (2498 vs 2183, 1900 proteins shared), whereas the opposite was true for a number of significantly changed proteins with a fold-change cutoff ≥ 2 (535 vs 820). The profound changes concerned mainly proteins of cardiomyocyte sarcomeres, costameres, intercalated discs, mitochondria, and extracellular matrix. In addition, distinct alterations in immune and defense response were found with a remarkable involvement of type I interferon signaling that has been recently hypothesized to be essential for anthracycline cardiotoxicity pathogenesis. Hence, μLC-MS/MS was found to be a sound alternative to nLC-MS/MS that can be useful for comprehensive mapping of global myocardial proteome alterations such as those associated with severe anthracycline cardiotoxicity.

Figure 1

Comparison of sample preparation and quantification strategy. (A) Percentages of peptides that were assigned to contain N-terminal carbamylation (black), lysine carbamylation (red), and pyroglutamic acid (blue). The samples were prepared with or without added urea under two different temperatures of disulfide bond reduction (60 °C vs 37 °C) and digestion (37 °C vs 25 °C). The bar plots illustrate the ratio (%) of TMT labeling efficiency for the two TMT to protein ratios (4:1 vs 8:1) with the total number of quantified proteins in the sample. The error bars represent the standard deviation (n = 3 replicates). (B) Accuracy of TMT-based and LFQ quantification was compared using S. pneumoniae (strain Rx1) proteins spiked into the protein matrix of mouse B lymphoblasts at six ratios. (C) Quantity of each proteome and its ratios for the samples from Figure 1B were injected sequentially in the LFQ approach and single injection of a multiplexed sample in the TMT approach. (D) Bar plots illustrating the number of quantified proteins and instances of missing values for TMT-based and LFQ quantification with and without the incorporation of the match-between-runs algorithm (MBR) within MaxQuant. (E) Comparison of the theoretical and analyzed ratios for TMT-based and LFQ quantification using either μLC or nLC. The discrepancy (%) between the theoretical and analyzed ratios for each technique is presented above the bars.

Figure 2

Morphological changes in the rabbit heart, schematic workflow and results of the comparison of μLC vs. nLC techniques utilized for wide-scale analysis of myocardial proteome changes. (A) Transverse section of a normal control (CTRL) heart compared with a heart suffering from advanced daunorubicin (DAU)-induced cardiomyopathy. In the DAU group, there is macroscopically seen significant dilatation of both ventricles. In the microscopic picture of the left ventricular myocardium, degenerative changes (myofibril disintegration to vacuolization of the cytoplasm and disintegration of the cell nucleus) predominate, leading to the death of groups of cardiomyocytes. The remnants of decayed cardiomyocytes are removed by macrophages (in the presence of mild mononuclear infiltrate), and they are replaced by connective tissue (blue color) forming a marked myocardial scarring. (B) Schematic workflow overview of the wide-scale analysis of myocardial proteome changes induced by severe ANT cardiomyopathy. (C) Number of quantified proteins in TMT-labeled and multiplexed myocardium samples determined using either μLC or nLC, following the previous high-pH fractionation. The Venn diagram illustrates the proportion of unique and overlapped proteins quantified by μLC (orange) and nLC (green) in myocardium samples after previous high-pH fractionation. (D) Violin plots representing log₂ fold-change between DAU-induced and control groups demonstrate a superior dynamic range for μLC (orange) in comparison to nLC (green). (E) Enhanced resolution of TMT ions in MS2 was substantiated by box plots of PIF (parent ion fraction) values for all PSMs identified by μLC (orange) and nLC (green).

Figure 3

Concordance rate between two data sets representing μLC or nLC techniques utilized for wide-scale analysis of myocardial proteome changes was evaluated. (A) Volcano plots demonstrating that both μLC and nLC techniques confirmed substantial DAU-induced proteome dysregulation. Significant proteins (Benjamini-Hochberg FDR < 1%, illustrated by a horizontal red line) are indicated by blue (downregulated) and red (upregulated) colors. The μLC technique was capable of quantifying a smaller number of proteins (3956 vs 4549) but demonstrated a higher percentage of significantly changed proteins (down- and upregulated) with a fold-change ≥ 2 (illustrated by vertical red lines) in comparison to the nLC technique (21% vs 12%). (B) Venn diagram depicting the number and overlap of significantly changed proteins quantified by μLC (orange) and nLC (green). (C) Scatter plot representing 2D enrichment analysis (Spearman correlation coefficient = 0.94, p < 0.001), (D) Scatter plot depicts correlation of significant protein changes determined by μLC and nLC (Spearman correlation coefficient = 0.97, p < 0.001). (E) Heat maps exhibiting the differentially expressed proteins between the DAU-treated and control groups, as analyzed by either μLC or nLC. Data were log₂ transformed and normalized by subtraction of median from each observed value prior to clustering analysis.

Figure 4

The figure exhibits the profile plots of protein level changes in Gene Ontology (GO) clusters, quantified by either μLC or nLC techniques. The clusters that exhibited a significant upregulation in the DAU group vs control group are represented by (A) GOBP–immune and defense response, (B) GOCC–extracellular matrix, and (C) GOCC–cell junction. The clusters that exhibited a significant downregulation are associated with (D) GOCC–NADH complex, (E) GOBP–ribonucleotide biosynthetic process, and (F) GOCC–mitochondrion. The mean value of the logarithms of the intensities within each cluster is indicated by a red thick line.