Mass spectrometric proteome profiling using a deep spectral library reveals homogenization of right and left atrial proteomes in persistent atrial fibrillation patients

Figures

• 

  Figure 1. Experimental setup and workflow. The pipeline for generating an MS/MS reference spectral library including the isolation of selected anatomical regions from the donor hearts that were subjected to a DDA-MS (left). Right and left atrial appendage samples were collected from patients with SR and persAF and analyzed via DIA-MS (right). Application of the MS/MS spectral library enabled a robust analysis of the AF proteome. *Patients included in the cohort summarized in Supplementary material online, Table S2, **Patients included in the cohort summarized in Table S3. Numbers indicate a total number of patients. DDA, data-dependent acquisition; DIA, data-independent acquisition; MS, mass spectrometry; PCT, pressure cycling technology. The figure was created with BioRender.com.

• 

  Figure 2. Canonical AF markers in RA-persAF proteome. (A) Schematic representation of conditions and sample locations. Numbers indicate number of patients. (B) Differential expression analysis comparing RA proteome from SR and persAF. Colours are assigned for the cases with FDR < 0.05 and |log2(FC)| > 0.58, other points are shown in grey. Legends show the total number of DEPs across both groups. (C) Immunoblots (top) and quantification (bottom) validating key dysregulated proteins in RA-persAF identified by proteomic analysis, including CES2, MFG-E8, GSTZ1, SLC25A36, TSPAN9, CCDC80, GABARAPL1. Data was normalized to CSQ (calsequestrin). * P < 0.05, ** P < 0.01, *** P < 0.001 vs. RA-SR. (D) Scatter plot showing positively correlating log2FC between RA-persAF markers and mapped genes associated with AF (published NHGRI-EBI catalog of GWAS). Pink colour represents up-regulation and blue colour represents down-regulation in persAF relative to SR. Statistical significance was determined using GEE analysis with Benjamini–Hochberg correction. FC, fold change; RA-persAF, right atrium from patients with persistent AF. The figure was created with BioRender.com.

• 

  Figure 3. Protein association network and subcellular proteome model of RA-persAF. (A) Protein association network with respective pathway enrichment in RA-AF proteome. For the full-size version, see Supplementary material online, Figure S3. Proteins are shown in nodes, node size represents −log10 p value, colour represents log2 FC and interaction intensity reflects confidence level. For each cluster, KEGG pathway associations were retrieved. Statistical significance was determined using GEE analysis with Benjamini–Hochberg correction. ECM, extracellular matrix; FC, fold change. (B) Schematic representation of an average right atrial appendage cardiomyocyte illustrating the subcellular distribution of protein mass fractions per cell, estimated using proteomic ruler approach. Values are normalized to total cellular protein content. Left: RA-SR (control). Right: RA-persAF. In both panels, compartments are labelled with their absolute percentage contribution (%) to total protein mass per cell, enabling direct comparison between conditions. (C) Spearman's correlation scatter plot showing the relationship between MYH6, a driver of myofibrillar content change in persAF, and myofibrillar content…

• 

  Figure 4. Profiling atrial and ventricular signature in RA-persAF proteome. (A) Venn diagram illustrating the overlap between RA-persAF markers (n = 875) and robust atrial (n = 53) and ventricular (n = 50) markers derived from two independent, publicly available cardiac proteomics datasets.38,44 (B) Heatmaps showing the log2FC of the overlapping atrial38,44 (left) and ventricular38,44 (right) markers across RA-persAF/RA-SR. (C) Comparison of NPPA (natriuretic peptide A) and NPPB (natriuretic peptide B) in RA-SR and RA-persAF, displayed as log2LFQ abundance (proteomics dataset). *P < 0.05, **P < 0.01 vs. RA-SR. (D) Immunoblots (left) and quantification (right) of NPPA and NPPB in RA-SR and RA-persAF, normalized to CSQ (calsequestrin). *P < 0.05, **P < 0.01 vs. RA-SR. (E) Comparison of atrial natriuretic peptide (NT-proANP) and brain natriuretic peptide (NT-proBNP) blood plasma levels in patients with SR (n = 23) and persAF (n = 17). *P < 0.05, ***P < 0.001 vs. SR. (F) Heatmap generated based on the log2FC of DEPs in RA-persAF/RA-SR (P < 0.05), LV/LV-HF and LV-HF/LV (P < 0.05). (G) Spearman's correlation scatter plot showing the distribution of DEPs from (E) based on their log2FC in…

• 

  Figure 5. Comparative analysis of previously defined AF-associated remodelling features in RA- and LA-persAF. (A) Schematic representation of conditions and sample locations. Numbers indicate number of patients. (B) Quantitative comparison of representative fibrosis (top), myolysis (middle) and failing ventricle-like (bottom) proteins, displayed as log2LFQ abundance (proteomics dataset). Each dot represents a technical replicate/patient. (C) Representative Masson-Goldner trichrome-stained sections of the RA in SR (RA-SR), RA in persAF (RA-persAF), and LA in persAF (LA-persAF) patients (left) and Mean ± SEM of fibrotic area as a percentage relative to the total tissue section area (right). Scale bar: 50 µm. Each dot represents an individual patient. Statistical significance was assessed using the GEE analysis for proteomics data and Kruskal–Wallis test for histology data. * P < 0.05. Numbers indicate the number of patients. LFQ, label-free quantification.

• 

  Figure 6. Comparative analysis of the global proteomic profiles of the RA- and LA-persAF. (A) PCA of RA and LA in SR (top) and persAF (bottom). Shown are the first two principal components. (B) Volcano plots showing the distribution of DEPs in RA vs. LA in SR (top) and persAF (bottom). Colours are assigned for the cases with FDR < 0.05 and |log2(FC)| > 0.58, other points are shown in grey. Legends show the total number of DEPs across both groups. (C) Chi-square analysis with Yates’ correction comparing the fraction of significant proteins to all identified proteins between RA vs. LA in SR and persAF. *** P < 0.001. (D) Heatmap depicting the RA-specific (left) and LA-specific (right) top 5 GO BPs in SR and their respective significance in persAF. (E) Immunoblots (upper left) and quantification (upper right) of RA protein markers RGS6 (regulator of G protein signaling) and CACNA2D2 (calcium voltage-gated channel auxiliary subunit α2δ2) in LA and RA in SR and persAF, normalized to CSQ (calsequestrin). Immunoblots (lower left) and quantification (lower right) of LA protein markers ACSS1 (acetyl-coenzyme A synthetase short-chain family member 1) and MYOT (myotilin) in LA and RA in SR and…

• 

  Figure 7. The role of the right atrium in enhanced proteome uniformity across both atria in persAF. (A) Mirror plots showing RA-SR markers and LA-SR markers [FDR < 0.05, |log2(FC)| > 0.58] and their abundance in RA-persAF relative to RA-SR. The left panel shows markers that contribute to increased atrial proteome symmetry in persAF, while the right panel shows preserved markers in RA-persAF that do not contribute to proteome symmetry. All data shown as log2FC. (B) Overlap between LA-SR (relative to RA-SR) and RA-persAF (relative to RA-SR) [P < 0.05, |log2(FC)| > 0]. The overlapping proteins were subjected to GO analysis and displayed in the lower panel together with the underlying proteins. The top terms consisting of more than 3 proteins are displayed. Statistical significance was determined using GEE analysis with Benjamini–Hochberg correction. FC, fold change; GO, gene ontology; LA, left atrium; RA, right atrium; RA-SR, right atrium from patients with sinus rhythm; RA-persAF, right atrium from patients with persistent AF.