Four-domain gut metagenomics reveals archaeal-centered cross-kingdom remodeling across coronary artery disease and acute myocardial infarction

Figures

• 

  Figure 1. Study design and prespecified analytic workflow Overview of the two-stage archaeal-centered study design. In the discovery cohort, stool shotgun metagenomes from patients with acute myocardial infarction (AMI) and healthy controls were profiled at the genus level across four microbial domains: archaea, bacteria, fungi, and viruses. This cohort was used for prespecified archaeal-centered cross-domain screening, AMI-healthy-control differential-correlation analysis, and paired stool-plasma microbiome-metabolome coupling. The extension cohort of angiographically normal controls (CON), severe coronary artery disease (CAD), and AMI was used for prespecified within-state screening, CAD-AMI transition testing, and three-state trajectory summaries.

• 

  Figure 2. Descriptive domain-specific α- and β-diversity analyses (A) Discovery cohort: descriptive α-diversity and β-diversity analyses across the archaeome, bacteriome, mycobiome, and virome in AMI and healthy controls. (B) Extension cohort: corresponding descriptive diversity analyses across CON, CAD, and AMI. PCoA, principal coordinates analysis.

• 

  Figure 3. Prespecified archaeal-centered cross-domain correlation screening (A) Archaeal genera involved in at least one FDR-significant archaeal-bacterial (A-B), archaeal-fungal (A-F), or archaeal-viral (A-V) co-variation edge in AMI or healthy controls; dot color denotes Spearman ρ on CLR-transformed abundances. (B) Edge counts by domain pair and sign in AMI and healthy controls. (C) Mean absolute correlation strength across significant edges. FDR, false-discovery rate; CLR, centered log-ratio.

• 

  Figure 4. Prespecified module construction and descriptive stability assessment (A) Module size versus eigengene variance explained by the first principal component (PC1). (B) Module size versus bootstrap stability, quantified by mean Jaccard similarity across resamples. (C) Summary of module number, represented genera, module size, eigengene variance explained, and stability metrics. Module construction was prespecified; stability metrics were descriptive. B, F, and V denote bacterial, fungal, and viral modules, respectively; numeric suffixes indicate within-domain module identifiers only. Detailed module composition, eigengene variance explained, stability metrics, and member genera are provided in Table S1.

• 

  Figure 5. Prespecified AMI-healthy-control differential-correlation analysis (A) Volcano plot of differential correlations across tested cross-domain edges; Δρ = ρ_AMI − ρ_control, where “control” denotes healthy controls. (B) Domain-pair-stratified volcano plots. (C) Enrichment of rewired edges by domain pair relative to all tested edges. (D) Group-specific correlations for all tested edges. Rewired edges were defined as |Δρ| ≥ 0.30 and BH-FDR q_Δ ≤ 0.10. BH-FDR, Benjamini-Hochberg FDR.

• 

  Figure 6. Prespecified extension-cohort CON-CAD-AMI analysis (A) Within-state archaeal-centered screening in CON, CAD, and AMI. (B) CAD-AMI differential-correlation testing for A-B and A-V edges. (C) State-wise correlations for CAD-AMI transition edges. (D) Descriptive heatmap of severity-like A-B trajectories, defined as monotonic change in |ρ| across CON-CAD-AMI without sign reversal; color encodes Spearman ρ. CON denotes angiographically normal controls; A-B, archaeal-bacterial; A-V, archaeal-viral co-variation.

• 

  Figure 7. Descriptive class-level summary of AMI-healthy-control differential plasma metabolite features Numbers of significant plasma liquid chromatography-mass spectrometry (LC-MS) features are shown by Class I annotation and direction of change.

• 

  Figure 8. Exploratory post hoc archaeal taxa-AMI severity associations Candidate archaeal genera were related to AMI severity markers across two AMI sets. (A) Methanobrevibacter versus peak cardiac troponin I (cTnI). (B) Methanosphaera versus left ventricular ejection fraction (LVEF). (C) Methanoregula versus LVEF. Analyses were exploratory post hoc clinical anchoring analyses.