A journey through the molecular networks of endomyocardial biopsies to explore ATTR and ALλ cardiac amyloidosis

Figures

• 

  Figure 1. Subject cohorts under investigation (A) Cohort 1 and cohort 2 gender distribution. The clinical history of the control group is reported through pie charts. (B and C) Cohort 1 age distribution, by gender and overall (Tukey-Kramer test, p ≤ 0.05; n = 12 per group); data are represented as a combination of mean, quantiles (25%–75%), and outliers. (D and E) Cohort 2 age distribution, by gender and overall (Student’s t test, p ≤ 0.05; ALλ, n = 19; ATTR, n = 20); data are represented as a combination of mean, quantiles (25%–75%), and outliers.

• 

  Figure 2. Cohort 1 proteome remodulation (A) Venn diagram of protein identified in control, ALλ, and ATTR (n = 12 per group). (B) Volcano plot of DAPs in ALλ (adj.P.Val ≤0.00001, −0.5 ≥log2(FC) ≥0.5). (C) Volcano plot of DAPs in ATTR (adj.P.Val ≤0.00001, −0.5 ≥log2(FC) ≥0.5). (D) Principal-component analysis performed using high-confidence DAPs (adj.P.Val ≤0.00001); PC1 explains 60.8% of the variance while PC2 explains 9.55%. (E) Hierarchical clustering (Ward method and Euclidean distance metric) performed using high-confidence DAPs (adj.P.Val ≤0.00001). (F) Biological processes (BPs) most enriched in ALλ and ATTR patients. (G) BPs most enriched in control group. Quantitative analysis was performed using limma R package; linear model was fitted to log2-transformed data, including age and gender as covariates. Functional enrichment analysis considered each subject, first individually and then aggregated into the corresponding group. Scale bars in circular layout graphs indicate the normalized number of proteins found per BP and per condition. Hierarchical clustering trees group BPs based on the number of proteins they share. BPs were enriched by FDR ≤0.05, while those differentially…

• 

  Figure 3. Cohort 2: ALλ vs. ATTR comparison (A) Pulmonary artery diastolic pressure (PADp, mmHg). (B) Mixed venous oxygen saturation (SVO2, %). (C) Right atrial pressure (RAp, mmHg). (D) Right ventricular end-diastolic pressure (RVEDp, mmHg). PADp, SVO2, RAp, and RVEDp differences were evaluated by Student’s t test; data are represented as a combination of mean, quantiles (25%–75%), and outliers. (E) Venn diagram of protein identified in ALλ and ATTR. (F) Volcano plot of DAPs in ALλ and ATTR (adj.P.Val ≤0.05, −0.5 ≥log2(FC) ≥0.5). Quantitative analysis was performed using limma R package; linear model was fitted to log2-transformed data, including age and gender as covariates. (G) Hierarchical clustering (Ward method and Euclidean distance metric; n DEPs = 55, adj.P.Val ≤0.05). Pink and red boxes include biological processes enriched from proteins most abundant in ALλ vs. ATTR, respectively (FDR ≤0.5); the most connected PPI components are shown.

• 

  Figure 4. Hubs/bottlenecks validation (A) Unweighted network topology. (B) Weighted network topology. Cohort1 unweighted PPI networks: control (5,684 nodes, 233,300 edges), ALλ (6,802 nodes, 279,547 edges), and ATTR (6,102 nodes, 243,658 edges); cohort 2 weighted PPI networks: ALλ (1,170 nodes and 26,191 edges) and ATTR (1,170 nodes and 26,191 edges). The statistical significance of the model’s robustness was assessed by comparing the average unweighted and weighted betweenness of the original networks with the betweenness distribution (violin plots) of the corresponding random networks (n = 1,000 per condition). These values consistently differ, indicating a shift away from random behavior and supporting the hubs/bottlenecks selection.

• 

  Figure 5. Unweighted and weighted network topological analysis (A) Most significant unweighted hubs/bottlenecks specifically identified in ALλ and ATTR patients or control participants, included in cohort 1; identification frequency (IF, n = 12), betweenness, centroid, and bridging centrality values are shown. (B) Protein-protein interaction (PPI) subnetworks, including ADGRL1, APPK1, and VANGL2 first neighbors, and the abundance level in control, ALλ, and ATTR, respectively. (C) Most significant weighted hubs/bottlenecks in ALλ (red) and ATTR (green) extracted by CoPPIs algorithm. The boxes on the right show the corresponding enriched reactome pathways (FDR≤0.05). Betweenness, bridging, and centroid centrality were calculated. Hubs/bottlenecks were extracted by considering the proteins whose centrality values were above the 75th quantile in ALλ (or ATTR), but not in the 75th quantile of ATTR (or ALλ).