Divergent mitochondrial and fibrotic signatures in atrial fibrillation: insights from organoids and GEO data

Figures

• 

  Fig. 1. Schematic overview of selection and processing of human AF datasets from the GEO database. Step 1: Relevant human transcriptomic datasets were identified through systematic searches of the NCBI Gene Expression Omnibus (GEO) database using AF-related keywords. Step 2: Four complementary datasets representing distinct biological contexts were selected: GSE128188 (human atrial appendage tissue), GSE222793 (iPSC derived atrial cardiomyocytes), GSE138252 (human cardiomyocytes), and GSE239321 (engineered atrial cardiomyocyte fibroblast coculture system). Step 3: Data preprocessing and analysis included normalization (e.g., RMA/quantile normalization), principal component analysis (PCA) to assess sample clustering, and differential expression analysis to identify significantly altered genes between conditions. Step 4: Functional interpretation of differentially expressed genes was performed using gene set enrichment analysis (GSEA), Gene Ontology (GO) annotation (molecular function, biological process, and cellular component), and pathway analysis to identify key biological mechanisms associated with AF

• 

  Fig. 2. Cross-dataset transcriptomic analysis across the atrial fibrillation spectrum. A Principal component analysis (PCA) of normalized gene expression across four GEO datasets representing complementary AF biological contexts. Each point represents an individual sample colored by experimental group. GSE128188: atrial fibrillation (AF, red) vs. sinus rhythm (SR, blue) in left atrial appendage; GSE138252: AF with heart failure (AF + HF, red) vs. non-failing controls (NF, blue); GSE222793: adipogenic medium (red) vs. basal medium (blue) in iPSC-derived atrial cardiomyocytes; GSE239321: atrial fibroblasts (red) vs. ventricular fibroblasts (blue). Clear separation between groups confirms dataset quality and biological signal. B Volcano plots displaying differentially expressed genes across four GEO datasets spanning the AF spectrum. The x-axis shows log₂ fold change and the y-axis shows − log₁₀ (adjusted p value). Blue: upregulated; red: downregulated (|log₂FC|> 1, BH-adjusted p < 0.05). C Normalized enrichment score (NES) heatmap of mitochondrial (top) and fibrosis-associated (bottom) pathways across four GEO datasets (GSE222793, GSE239321, GSE128188, and GSE138252), ordered from…

• 

  Fig. 3. Contractile dysfunction in human atrial organoids induced by fast pacing and Angiotensin II. Functional assessment of human cardiac organoids exposed to atrial fibrillation (AF)-like conditions. A Schematic representation of the workflow for generating human atrial organoids from hiPSCs. “Background” section (Days 0–6): hiPSC culture. Human induced pluripotent stem cell (hiPSC) colonies were seeded onto Matrigel-coated plates and maintained under standard culture conditions. Medium was changed daily to preserve pluripotency and colony integrity. “Methods” section (Days 7–15 +): Atrial organoid formation. hiPSC colonies were dissociated into single cells or small clusters and allowed to form aggregates. Cell aggregates were subsequently embedded in Matrigel droplets to promote three-dimensional organization and organoid development. This process enables structural assembly and maturation toward atrial-like tissue. B Schematic contraction analysis workflow. Microscope videos of beating organoids were converted into pseudo-color heatmaps representing contraction intensity and subsequently quantified for contractile measurements. C Brightfield images show gross morphology of control…

• 

  Fig. 4. Transcriptomic and experimental validation of mitochondrial gene expression across the AF spectrum and organoid models. Mitochondrial subunit and regulator expression in AF GEO datasets and human atrial organoids. A Heatmap showing log₂ fold change of mitochondrial genes across four AF-related GEO datasets (GSE222793, GSE239321, GSE128188, and GSE138252). Genes include Complex I–IV subunits and mitochondrial biogenesis/dynamics regulators. Blue: downregulation; red: upregulation. (B–E) qPCR analysis of mitochondrial genes in FP (purple) and AngII-treated (green) human cardiac organoids, compared to control (gray). B Complex I genes (NDUFA9, NDUFB8, NDUFS1, NDUFV1), C Complex II genes (SDHA, SDHB, SDHC, SDHD), D Complex III/IV genes (UQCRC1, UQCRC2, UQCRFS1, ATP5A1, ATP5B, COX1), and E biogenesis and mitochondrial dynamic regulators (PGC1α, NRF1, MFN1, OPA1, TFAM, TFB1M, TFB2M, FIS1). All data are normalized to GAPDH and presented as mean ± SD. p < 0.05, p < 0.01, p < 0.001

• 

  Fig. 5. Protein-level suppression of mitochondrial markers in atrial organoids exposed to fast pacing or Angiotensin II. Mitochondrial protein expression is downregulated in human atrial organoids exposed to FP or AngII. A Western blot analysis of PGC1α and Cytochrome C in control, FP, and AngII-treated groups. GAPDH used as loading control. B Quantification of Western blot band intensity normalized to GAPDH. AngII-treated organoids show stronger suppression of both mitochondrial markers. C Immunofluorescent staining of organoid cross-sections. Top panels: co-staining for PGC1α (green), F-actin (Phalloidin, red), and nuclei (DAPI, blue). Bottom panels: co-staining for Cytochrome C (green), Phalloidin, and DAPI. Merged images show cytoskeletal disorganization and spatial loss of mitochondrial protein signals, particularly in AngII-exposed organoids. Scale bars = 100 μm

• 

  Fig. 6. Fibrotic gene activation in human AF datasets and organoid models reflects divergent remodeling mechanisms. Upregulation of fibrosis-related genes in human atrial fibrillation (AF) datasets and organoid models. A Heatmap of fibrosis and ECM remodeling genes across GEO datasets (GSE222793, GSE239321, GSE128188, GSE138252), including AGTR1, TGFβ1, SMAD2/3, CTGF, MMP2/9, COL1A1, and TIMP2. Red: upregulated; blue: downregulated. B qPCR analysis of fibrosis pathway genes in fast-paced (FP, purple) and Angiotensin II (AngII, green) organoids, showing increased expression of AGTR1, TGFβ1, SMAD2/3, and CTGF. C ECM remodeling genes (MMP2, MMP9, COL1A1, TIMP2) are significantly upregulated in AngII-treated organoids compared to controls, with milder effects in FP. Data shown as mean ± SD; p < 0.05, p < 0.01, p < 0.001

• 

  Fig. 7. Fibrotic protein expression and tissue remodeling in atrial organoids under FP and AngII stress. Validation of fibrosis in atrial organoids using fibrotic protein markers. A Western blot analysis of Col1a1 and αSMA in control (C), FP, and AngII-treated organoids. GAPDH used as loading control. B Quantification of band intensities normalized to GAPDH. Both FP and AngII treatments increase Col1a1 and SMA protein expression, with AngII inducing a stronger response. C Immunofluorescent staining of organoid cross-sections for Col1a1 (top) and SMA (bottom), co-labeled with Phalloidin (F-actin, red) and DAPI (nuclei, blue). Merged images reveal widespread matrix remodeling and myofibroblast marker expression in FP and AngII groups. AngII-treated organoids exhibit diffuse, dense fibrosis consistent with contractile impairment. Scale bars = 100 μm

• 

  Fig. 8. Targeted directional concordance between atrial organoid remodeling models and AF-spectrum transcriptomic datasets. Quantitative comparison of FP and AngII organoid responses with four GEO datasets representing complementary AF-spectrum contexts. A Directional concordance of FP organoids against each GEO dataset, calculated separately for mitochondrial (n = 22, blue) and fibrosis-related (n = 9, orange) gene panels. FP shows the highest concordance with GSE138252 (AF with heart failure), with 80% mitochondrial and 83% fibrosis concordance. B Directional concordance of AngII organoids using the same approach. AngII shows the highest mitochondrial concordance with GSE222793 (cardiometabolic atrial remodeling; 65%) and the highest fibrosis concordance with GSE138252 (67%). Dashed horizontal line indicates the 50% reference level (chance)