Gut Virome Characteristics and Network Alterations in IgA Nephropathy

Abstract

Introduction

Emerging evidence implicates gut microbiota dysbiosis in the pathogenesis of IgA nephropathy (IgAN), yet the contribution of the gut virome remains unexplored. This study aimed to characterize virome signatures and virus-microbiota interactions in IgAN.

Methods

We performed a rigorously matched case-control study including 32 patients with biopsy-proven IgAN and 32 healthy controls. Fecal viral-like particles and bacterial communities were profiled using metagenomic sequencing and full-length 16S ribosomal RNA (rRNA) sequencing. Statistical analysis included diversity, differential abundance, network analysis, and correlation with clinical indices.

Results

IgAN subjects displayed significant reductions in gut virome richness (severe IgAN vs. healthy controls, P = 0.03), with a lower relative abundance of Caudoviricetes in severe IgAN (P = 0.045) and enrichment of Tectiliviricetes in mild disease (P = 0.03). We identified 113 differentially abundant bacteriophage contigs (82 up, 31 down; false discovery rate < 0.05); key predicted hosts shifted toward Bacteroides, Clostridium, and Roseburia in IgAN, whereas Faecalibacterium and Alistipes prevailed in controls. Viral and bacterial alpha diversity correlated in healthy controls but not in IgAN (r = 0.38, P = 0.03 vs. r = 0.04, P = 0.81). IgAN virome encoded more glyco-modifying enzymes (P < 0.05), with strong correlations to estimated glomerular filtration rate (eGFR) (r = 0.65, P = 0.001). Viral and bacterial alpha diversity were significantly correlated with proteinuria and gross hematuria (r = 0.18–0.25, – < 0.05).

Conclusion

This study describes potential alterations in gut virome diversity, bacteriophage composition, bacteriome-virome relationships, and predicted functional profiles in IgAN, suggesting potential relevance of the gut virome to intestinal ecological alterations.

Graphical abstract

IgAN stands as the most prevalent primary glomerulonephritis worldwide and represents a significant global health burden.¹ Despite advances in supportive therapies and immunosuppressants, approximately 30% to 40% of patients progress to end-stage renal disease within 2 decades.² Recent large cohort studies reveal a median renal survival of only approximately 12 years postdiagnosis, underscoring the disease’s unpredictable clinical course and poor prognosis.³ This marked heterogeneity in clinical course and long-term outcomes underscores a critical gap in our understanding of the disease-driving mechanisms and highlights the urgent need to identify modifiable pathogenic pathways that can inform precision, etiology-based therapeutic strategies.

IgAN is currently understood as a multifactorial, immune-mediated glomerular disorder occurring in genetically susceptible individuals. A growing body of evidence emphasizes the pivotal role of the gut-kidney axis in IgAN pathogenesis.⁴ Specifically, gut microbiota dysbiosis characterized by the depletion of beneficial commensals such as Prevotella copri and Alistipes putredinis, alongside the expansion of potentially pathogenic taxa, including Escherichia, Shigella, and Akkermansia muciniphila disrupts mucosal immune homeostasis.5, 6, 7 This dysregulation fosters aberrant mucosal IgA responses, promoting the production of pathogenic immune complexes that deposit in the glomeruli, leading to renal injury.⁸ However, findings across studies have been heterogeneous and sometimes conflicting, suggesting that key components of the intestinal ecosystem may remain insufficiently characterized.

This inconsistency may partly reflect the intrinsic complexity of the intestinal microbiota, which comprises trillions of microorganisms spanning multiple kingdoms, including bacteria, archaea, fungi, protists, and viruses.⁹ Although the role of bacteria in health and disease has begun to be systematically characterized, other microbial constituents, particularly viruses, have remained largely neglected until recently.¹⁰ Notably, viruses are estimated to be an order of magnitude more abundant than prokaryotes in the gut and possess unique capacities to interact with bacterial hosts, through bacteriophages, and with human cells, through eukaryotic-targeting viruses.^11,12 Through these interactions, the gut virome plays a critical role in shaping microbial community structure and modulating intestinal immune homeostasis.

In recent years, the gut virome, comprising diverse bacteriophages and eukaryotic viruses, has emerged as a key regulator of microbial ecology and host immunity.^13,14 Studies in other autoimmune and inflammatory syndromes (e.g., systemic lupus erythematosus, inflammatory bowel disease [IBD], and diabetes) have demonstrated that virome perturbations can profoundly reshape bacterial communities, modulate immune responses, and influence disease trajectories.15, 16, 17 Notably, therapeutic viral manipulations, including fecal virome transplantation, have shown promise in modulating disease phenotypes in animal models and clinical settings.^18,19 While associations between specific viral infections such as cytomegalovirus, Epstein-Barr virus, hepatitis viruses, and more recently SARS-CoV-2, with the kidney pathology of IgAN have been reported,20, 21, 22 these observations largely reflect exogenous viral exposures. The composition, ecological organization, and functional relevance of the endogenous gut virome in IgAN remain almost entirely unexplored.

To date, the gut virome and its ecological relationship with the bacteriome across disease stages have not been systematically characterized in IgAN. Here, we employed an integrative approach combining fecal viral metagenomic sequencing, full-length 16S rRNA profiling, and detailed clinical phenotyping in a well-characterized IgAN cohort and matched healthy controls. We characterized virome diversity, virus-bacterium ecological networks, and predicted functional features, and examined their associations with disease severity and kidney function. By providing the first comprehensive overview of the gut virome landscape in IgAN, this study establishes a descriptive, hypothesis-generating framework for future investigations into microbiome–virome interactions along the gut-kidney axis.

Methods

Study Cohort and Sample Collection

A total of 64 participants, including 32 patients with biopsy-confirmed IgAN and 32 propensity score–matched healthy controls, were enrolled from Peking University First Hospital. Propensity score matching was performed using age, gender, and body mass index as covariates to minimize potential confounding and balance baseline characteristics between patients with IgAN and healthy controls. Patients with IgAN were further stratified by eGFR into mild (eGFR > 60 ml/min per 1.73 m²) and severe (eGFR ≤ 60 ml/min per 1.73 m²) subgroups. None of the participants had received antibiotics, probiotics, systemic corticosteroids, or other immunosuppressive agents within 8 weeks before sample acquisition. The exclusion criteria encompassed end-stage renal disease, secondary IgAN, IBD, type 2 diabetes mellitus, infectious diseases, and other autoimmune or inflammatory disorders. Fresh fecal samples were collected using standardized procedures and stored at −80 °C within 2 hours of collection for downstream virome and bacteriome analyses. Dietary intake was not systematically recorded or controlled at the time of fecal sample collection. The study was approved by the ethics committee of Peking University First Hospital (IRB No. 2021[073]), and written informed consent was obtained from all subjects in accordance with the Declaration of Helsinki and its amendments.

Purification, Extraction, Amplification and Sequencing of Viral-Like Particles

Viral-like particles were isolated from fecal samples, using a previously described protocol with minor modifications.²³ Briefly, 300 to 400 mg of stool was resuspended in saline-magnesium buffer, homogenized, and centrifuged to remove particulate debris. The supernatant was filtered through a 0.45 μm pore-size membrane to eliminate residual cells and large particles. Viral particles were further enriched by sequential enzymatic and chemical treatments to remove bacterial contamination and free nucleic acids. Viral DNA was then extracted using the Qiagen MinElute Virus Spin Kit and quantified using the Qubit dsDNA High-Sensitivity Assay. Sequencing libraries were constructed following standard Illumina protocols and sequenced on the Illumina NovaSeq 6000 platform to generate paired-end reads (2 × 150 bp) for downstream virome analyses. Further details are described in the Supplementary Methods.

DNA Extraction, Amplification, and Sequencing of 16S rRNA Full-Length

Total microbial genomic DNA was extracted from fecal samples using the ZymoBIOMICS DNA Miniprep Kit (Zymo Research, USA) according to the manufacturer’s instructions. Full-length bacterial 16S rRNA genes were amplified using universal primers 27Fand 1492R. Polymerase chain reaction amplification was performed using Q5 high-fidelity DNA polymerase (New England Biolabs). Amplicons were purified using AgencourtAMPure XP beads, quantified, and pooled at equimolar concentrations. Sequencing was conducted on the PacBio Sequel platform (Pacific Biosciences) using single-molecule real-time circular consensus sequencing mode. High-quality circular consensus sequencing reads were retained for downstream analyses. Further details are described in the Supplementary Methods.

Virome Data Analysis

Raw viral metagenomic reads were processed using a stringent and standardized pipeline. After quality control and adapter trimming, host-derived reads were removed by alignment to the human reference genome. Clean reads were de novo assembled, and short contigs were excluded. Viral contigs were identified using an integrated strategy combining DeepVirFinder, VirSorter2, and geNomad under stringent thresholds.24, 25, 26 Predicted viral sequences were dereplicated and clustered into nonredundant viral operational taxonomic units (vOTUs) using CheckV, which was applied to assess genome quality and remove host-derived contamination. Only medium- or high-quality vOTUs, or those containing viral hallmark genes, were retained. Taxonomic annotation was performed against the RefSeq viral database, and viral taxonomy was assigned according to National Center for Biotechnology Information classification.²⁷ Viral contig abundance was normalized to sequencing depth and expressed as relative abundance to account for differences in sequencing depth across samples.

Bacterial 16S rRNA Data Analysis

To ensure methodological consistency across samples, full-length 16S rRNA gene sequencing data were processed using the Lotus2 pipeline following the same procedures described above.^28,29 Briefly, raw PacBio reads were quality filtered to remove low-quality sequences, contaminants, and host-derived reads. High-quality sequences were clustered into operational taxonomic units or amplicon sequence variants at a 97% sequence similarity threshold using VSEARCH, and taxonomic assignment was performed against the Greengenes2 reference database with Lambda as the taxonomic aligner, as implemented in Lotus2. The resulting taxonomic profiles were used for downstream microbial community analyses.

Virus-Host Prediction

Putative virus–host associations for high-confidence vOTUs were inferred using an integrated strategy combining transfer RNA sequence matching and CRISPR spacer alignment, 2 complementary and widely used approaches for host prediction in viral metagenomic studies. Briefly, viral-encoded transfer RNAs and CRISPR spacer matches to prokaryotic genomes were identified under stringent criteria, and the resulting evidence was integrated to assign putative host taxonomic affiliations. Only high-confidence virus-host links supported by at least one of these approaches were retained for downstream analyses. Further details are described in the Supplementary Methods.

Functional Annotation and Enrichment Analyses

To characterize the functional and metabolic potential of the gut virome, functional annotation was performed on a nonredundant viral gene catalog derived from high-confidence vOTUs. Protein-coding genes were predicted from vOTU contigs using Prodigal in metagenomic mode. Functional annotation was conducted using homology-based approaches across multiple reference databases, including KEGG Orthology, Gene Ontology, and carbohydrate-active enzyme (CAZy) databases.

KEGG annotations were assigned using KofamScan, Gene Ontology terms were annotated using InterProScan, and carbohydrate-active enzymes were identified using the dbCAN2 pipeline under stringent filtering criteria. Carbohydrate-active enzyme–encoding genes detected within viral contigs were considered putative auxiliary metabolic genes. All annotated genes were derived from rigorously filtered, dereplicated, and host-decontaminated viral sequences, ensuring high confidence in their viral origin for downstream functional and enrichment analyses. Further details are described in the Supplementary Methods.

Statistics

All statistical analyses were performed in R software (version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria; https://www.r-project.org/). Alpha diversity metrics were calculated using the vegan package, with group comparisons conducted using t test or Wilcoxon rank-sum test, as appropriate. Beta diversity was assessed based on Bray–Curtis distances and visualized using principal coordinates analysis, with group-level differences evaluated using PERMANOVA (adonis2, 999 permutations). Differential abundance analyses of vOTUs and bacterial operational taxonomic units were performed using DESeq2, and functional enrichment analyses were conducted using clusterProfiler. Correlations between microbial or viral features, functional attributes, and clinical indices were assessed using Pearson’s or Spearman’s correlation, with multiple testing corrected using the Benjamini–Hochberg method. Procrustes analysis was used to assess concordance between bacterial and viral community structures. Cooccurrence networks were visualized using Cytoscape, and data visualization and regression analyses were performed using ggplot2 and ggpubr. For cross-kingdom correlation analysis, bacterial genera detected in the 16S rRNA sequencing dataset were filtered based on prevalence. Only genera present in ≥10% of stool samples were retained to ensure that correlations were calculated among commonly detected taxa. All statistical tests were 2-sided, with P ≤ 0.05 considered statistically significant.

Results

Clinical Cohort and Gut Virome Overview

A total of 64 fecal samples (IgAN = 32, healthy controls = 32) were analyzed. Age, gender, and body mass index were matched between groups; and clinical characteristics are summarized in Supplementary Table S1. Shotgun virome sequencing generated a median of 46,037,010 reads per sample (interquartile range: 42,613,173–50,049,573), which were clustered into 375,522 unique species-level vOTUs. This study focused exclusively on DNA viruses (bacteriophages), excluding eukaryotic and RNA viruses. The overall experimental design, integrating virome sequencing with 16S rRNA profiling for bacteriome-virome and clinical correlation analyses, is shown in Figure 1a.

Figure 1.

Overview of virome profiling workflow and structural features in IgAN. (a) Schematic representation of the study design and analytical workflow. Fecal samples were collected from HCs (n = 32) and patients with IgAN (n = 32), stratified by eGFR (IgAN-m: eGFR ≥ 60 ml/min per 1.73 m², n = 17; IgAN-s: eGFR < 60 ml/min per 1.73 m², n = 15). Viral-like particles were enriched and subjected to shotgun metagenomic sequencing, followed by quality control, de novo assembly (MEGAHIT), viral identification using an integrated approach (DeepVirFinder, VirSorter2, and geNomad), and clustering into viral OTUs at 95% nucleotide identity with CheckV-based quality control and decontamination. In parallel, full-length 16S rRNA sequencing was performed for bacterial profiling, including quality filtering, OTU/ASV clustering at 97% sequence similarity, and taxonomic annotation against reference databases. Identified viral and bacterial features were subsequently analyzed for diversity, differential abundance, functional potential, virus-bacterium interactions, and correlations with clinical parameters. (b) Stacked bar plots showing the relative abundances of major viral taxa at the family level in HCs and IgAN groups. (c) Stacked bar plots showing the relative abundances of major viral taxa at the family level among subgroups stratified by kidney function. (d) Stacked bar plots showing the relative abundances of dominant viral classes across groups. (e) Boxplots illustrating the relative abundances of representative viral taxa between HCs and IgAN groups. (f) Boxplots illustrating the relative abundances of representative viral taxa among kidney function-stratified subgroups. Statistical significance was assessed using the Wilcoxon rank-sum test. eGFR, estimated glomerular filtration rate; HC, healthy controls; IgAN, IgA nephropathy; OTU, operational taxonomic units.

To visualize group-level differences, taxonomic bar plots displaying the top 15 viral families were generated for healthy controls, patients with IgAN, and IgAN subgroups (Figure 1b and c). At the class level, the virome was overwhelmingly dominated by Caudoviricetes (> 90% of total abundance), whereas other classes (Megaviricetes, Malgrandaviricetes, Pokkesviricetes, Tectiliviricetes, and “Other”) were detected at lower levels (Figure 1d). No significant differences were observed between IgAN and healthy controls for Caudoviricetes (P = 0.26) or Megaviricetes (P = 0.19) (Figure 1e). Subgroup analysis revealed a significant depletion of Caudoviricetes in severe IgAN compared with mild IgAN (P = 0.045), suggesting loss of this class in advanced disease. Tectiliviricetes were significantly enriched in mild IgAN compared with healthy controls (P = 0.03), possibly reflecting early-stage alterations. Other minor classes (Malgrandaviricetes, Pokkesviricetes, and unclassified viruses) showed no significant differences across groups (Figure 1f).

Gut Virome Alterations and Differentially Abundant Bacteriophages in IgAN

To investigate virome differences between patients with IgAN and healthy controls, we first assessed viral community diversity. Alpha diversity metrics (ACE and Chao1) showed a decreasing trend in IgAN compared with healthy controls (ACE, P = 0.06; Chao1, P = 0.08; Figure 2b). Subgroup analysis showed significantly lower richness in severe IgAN than in healthy controls (ACE, P = 0.03; Chao1, P = 0.04; Figure 2c and d), whereas no significant difference was observed between mild IgAN and healthy controls. Principal coordinates analysis indicated no significant differences in beta diversity between healthy controls and patients with IgAN or among IgAN subgroups (Supplementary Figure S1).

Figure 2.

Viral diversity, taxonomic composition, and differentially abundant bacteriophages in IgAN. (a and b) Alpha diversity metrics (ACE and Chao1) comparing viral richness between HCs and patients with IgAN. (c and d) Subgroup analysis of alpha diversity metrics among HCs, mild IgAN, and severe IgAN groups. (e–h) Differential abundance and taxonomic profiles of bacteriophages in IgAN. (e) Volcano plot showing differentially abundant viral contigs between HCs and IgAN groups. Points above the horizontal dashed line indicate significant differences with false discovery rate adjusted P < 0.05. (f) Taxonomic differences in phages at the family level between patients with IgAN and HCs. (g and h) Differential abundance of bacteriophages grouped according to predicted bacterial host families (g) and predicted bacterial host genera (h) based on computational host prediction analysis. Blue points indicate viral taxa enriched in IgAN; red points indicate viral taxa depleted in HCs. Group differences in alpha diversity were assessed using the Wilcoxon rank-sum test. Differential abundance analysis of viral families was performed using DESeq2. HC, healthy controls; IgAN, IgA nephropathy.

We next examined taxonomic alterations in the gut virome. A total of 113 phage contigs were differentially abundant between IgAN and healthy controls, with 82 enriched and 31 depleted in IgAN (Figure 2e). Differentially represented taxa were dominated by Caudoviricetes, Mimiviridae, and unclassified phages (Figure 2f). Host assignment analysis indicated that phages predicted to infect Bacteroidaceae, Lachnospiraceae, and unclassified hosts were most prevalent (Figure 2g). At the genus level, 26 bacterial genera were predicted as putative phage hosts, with bacteriophages predicted to target Bacteroides and unclassified bacterial genera accounting for the largest proportion. Notably, bacteriophages predicted to use Clostridium, Bacteroides, Fervidibacillus, Longicatena, Pseudomonas, Mediterraneibacter, Pyrobaculum, Rhodococcus, Roseburia, and Thomasclavelia were exclusively enriched in IgAN; whereas bacteriophages predicted to use Veillonella, Segatella, Pusillibacter, Parabacteroides, Neisseria, Intestinimonas, Faecalibacterium, Coprococcus, Anaerostipes, Alistipes, and Agathobacter were uniquely enriched in healthy controls (Figure 2h).

Subgroup analysis further confirmed consistent patterns across disease severity strata. Several viral families and host-specific phages, including Caudoviricetes, Salasmaviridae, Mimiviridae, and unclassified taxa, showed reproducible alterations. Furthermore, bacteriophages predicted to target Bacteroides, Agathobacter, Bifidobacterium, Scatomonas, Parabacteroides, Odoribacter, Flavonifractor, and Clostridium as hosts were persistently detected across IgAN subgroups (Supplementary Figures S2, S3, and S4).

Altered Cross-Kingdom Diversity Relationships and Network Structure of the Gut Bacteriome and Virome in IgAN

We first assessed the cross-kingdom relationship between bacterial and viral communities at the level of alpha diversity. In healthy controls, bacterial and viral richness showed significant coupling, with positive correlations for the ACE index (r = 0.38, p = 0.033; Figure 3a) and the Chao1 index (r = 0.35, P = 0.047; Supplementary Figure S5A). In contrast, these associations were lost in IgAN (ACE, r = 0.04, P = 0.814, Figure 3b; Chao1, r = −0.04, P = 0.833; Supplementary Figure S5B). Procrustes analysis was performed between bacterial genus-level profiles and viral vOTUs. Despite reduced alpha diversity correlations, the 2 communities retained significant structural alignment (r = 0.55, P = 0.007; Figure 3c). Systematic pairwise comparisons of alpha diversity indices revealed broad and robust correlations in healthy controls, particularly for richness-based measures (observed richness, Chao1, ACE) (Figure 3d). In contrast, most correlations were absent in IgAN, with only a few indices showing weak residual associations (Figure 3e).

Figure 3.

Altered bacteriome-virome diversity relationships and network structure in IgAN. (a and b) Spearman correlations between bacterial and viral alpha diversity indices (ACE) in HCs and patients with IgAN; shaded regions indicate 95% confidence intervals. (c) Procrustes analysis between bacterial genus-level profiles and viral vOTUs assessing overall community-level concordance. Arrows connect matched bacterial and viral profiles from the same individual and point toward the bacterial ordination. Longer arrows indicate greater within-subject discordance between the 2 communities. Statistical significance was determined using Procrustes randomization tests. (d and e) Pairwise correlation matrices of bacterial and viral alpha diversity indices in HCs (d) and IgAN (e) groups. ∗∗Asterisks indicate levels of statistical significance (∗P < 0.05, ∗∗P < 0.01, ∗∗∗ P < 0.001). (f) Bubble heatmap showing significant bacterial-viral taxonomic correlations. (g) Cross-kingdom cooccurrence network based on significant bacterial-viral correlations. Nodes represent bacterial genera (circles) and viral families (squares), and edges represent significant associations (false discovery rate–adjusted P < 0.05). Edge colors indicate the direction of correlation (positive or negative), and node sizes are proportional to their degree. Spearman’s rank correlation was used for correlation analyses, and P values were adjusted using the false discovery rate method. Procrustes analyses were performed using Aitchison distance for both bacterial and viral datasets. HC, healthy controls; IgAN, IgA nephropathy.

To explore disease-relevant cross-kingdom associations, we performed downstream network analyses by integrating differentially abundant bacterial taxa identified by 16S rRNA sequencing and differentially abundant viral taxa at the taxonomic level (differential bacterial taxa are listed in Supplementary Table S2). The bacterial genera included in the analysis were confirmed to be present in the 16S rRNA sequencing dataset and were commonly detected in stool samples. For example, Ruthenibacterium, Prevotella, and Copromorpha were detected in 73.4%, 59.4%, and 56.3% of samples, respectively. A bubble heatmap illustrated distinct interaction patterns (Figure 3f). For example, Ruthenibacterium negatively correlated with Hafunaviridae, whereas Prevotella displayed positive correlations with Orthoherpesviridae, Fiersviridae, and Guelinviridae. Conversely, Copromorpha was negatively correlated with Potyviridae, Grimontviridae, Rudiviridae, and Pootjesviridae. Integration of these associations into a cross-kingdom network representation (Figure 3g) illustrated the overall organization of virus-bacteria correlations. Within this network, Ruthenibacterium occupied a central connecting position among multiple taxa, whereas several viral families, including Guelinviridae, Hafunaviridae, and Rudiviridae, exhibited high connectivity and were identified as network hubs based on their topological positions.

Predicted Functional Profiles of the IgAN-Associated Virome

To investigate potential functional alterations in the gut virome, we analyzed CAZy families encoded by predicted viral genes. Four major CAZy families—GH15, GT2, GT39, and GT41—were more frequently identified in the IgAN virome than in healthy controls (GH15, GT2, GT39, GT41; all P < 0.05,Figure 4a–d). These enzymes participate in carbohydrate metabolism, glycoside hydrolysis, and cell wall degradation, suggesting potential viral contributions to gut carbohydrate processing. Subgroup analysis revealed stage-specific alterations. Compared with healthy controls, GH15 and GT39 were more frequently detected in patients with mild IgAN (GH15, GT39, all P < 0.01; Supplementary Figure S6). To elucidate the broader functional landscape, we performed Gene Ontology and KEGG enrichment analyses based on predicted viral gene content. Gene Ontology enrichment highlighted functions related to organic substance metabolism, carbohydrate derivative metabolism, and catalytic activity, particularly hydrolases and isomerases, consistent with the CAZy profile (Figure 4e). KEGG pathway analysis further supported this metabolic pattern, revealing enrichment in pathways related to glutathione metabolism, peroxisome, and lipopolysaccharide biosynthesis (Figure 4f).

Figure 4.

Functional signatures of the gut virome in IgAN. (a–d) Boxplots showing the relative abundances of 4 viral carbohydrate-active enzyme (CAZyme) families, including (a) GH15, (b) GT2, (c) GT39, and (d) GT41, between HCs and patients with IgAN. (e) Gene Ontology enrichment analysis based on predicted viral gene annotations. (f) KEGG pathway enrichment analysis of predicted viral genes. Group differences were assessed using the Wilcoxon rank-sum test. HC, healthy controls; IgAN, IgA nephropathy.

Virome Profiles Correlate With Clinical Indices and Disease Severity in IgAN

At the global level, Mantel analysis revealed significant correlations between gut microbial community structures and host clinical parameters (Figure 5a). Specifically, viral alpha-diversity indices (ACE, Chao1, and Observed species) were significantly associated with 24-hour urinary protein (r = 0.23–0.24, P < 0.05), whereas bacterial alpha-diversity indices (ACE, Chao1, Shannon, and Pielou) showed significant correlations with gross hematuria (r = 0.18–0.25, P < 0.05). Notably, the composite carbohydrate-active enzyme index, including GH15, GT2, GT39, and GT41 families, exhibited a significant positive correlation with IgA levels. At the taxonomic level, correlation analyses were performed on bacterial genera and viral families that were previously identified as differentially abundant between patients with IgAN and healthy controls. These taxa showed distinct association patterns with clinical indicators (Figure 5b). Opportunistic bacterial genera such as Escherichia_710834 and Evtepia were positively correlated with IgA levels. Members of the Zobellviridae family were negatively correlated with eGFR but positively correlated with serum creatinine. Moreover, several viral taxa were correlated with host environmental variables, including body mass index, gender, and residential location, suggesting that virome composition may be shaped by individual host characteristics.

Figure 5.

Correlations between gut microbial features and clinical indices in IgAN. (a) Global correlation network integrating bacterial and viral alpha diversity indices, viral CAZyme families, and host clinical or environmental variables. The upper triangular matrix shows pairwise Pearson’s correlation coefficients among clinical parameters. The lower network illustrates Mantel test correlations linking microbial community structures and viral functional profiles, including the composite CAZyme index (CAZy_total) (comprising GH15, GT2, GT39, and GT41), with clinical indices. Color scale represents Pearson’s r values, and edge styles indicate Mantel test significance (P < 0.01, 0.01–0.05, ≥ 0.05). (b) Spearman’s correlation heatmap showing associations between differential bacterial genera, viral families, and clinical or pathological parameters. Colors indicate Spearman’s rho values; asterisks denote statistical significance (P < 0.05, ∗P < 0.01, ∗∗P < 0.001). ALB, albumin, BMI, body mass index; BUN, blood urea nitrogen; CAZyme, carbohydrate-active enzyme; eGFR, estimated glomerular filtration rate; Gd-IgA1, galactose-deficient IgA1; RASI, renin–angiotensin system inhibitor; SCr, serum creatinine; UA, uric acid; UTP, 24-hour urinary protein.

Discussion

To our knowledge, this study represents the first integrative analysis of the gut virome in IgAN, combining viral metagenomics, bacterial 16S rRNA profiling, and clinical association analyses. Our data suggest potential differences in viral diversity, compositional features of major viral families, and virus-bacteria association patterns in IgAN within this exploratory dataset. These virome features were associated with several clinical indicators of kidney function and disease severity within this dataset. Together, these observations provide initial insights into gut virome alterations in IgAN.

Caudoviricetes represented the most abundant viral class in the gut virome and showed significantly lower relative abundance in severe IgAN than in healthy individuals. In contrast, Tectiliviricetes displayed a relative enrichment in patients with mild IgAN, whereas Megaviricetes were additionally enriched in more advanced disease stages. These patterns may suggest disease stage–associated changes in gut virome composition, potentially reflecting altered ecological conditions in the intestinal microbial ecosystem under IgAN conditions. Notably, Caudoviricetes are among the best-characterized viral classes in current reference databases, whereas phages associated with gram-positive bacteria and Archaea remain less well-annotated.³⁰ Therefore, these observations should be interpreted with appropriate caution. Alterations in gut virome structure have been reported in other immune-mediated intestinal disorders, particularly IBD. Multiple metagenomic studies have described disease-associated shifts in dominant phage populations, including changes in tailed bacteriophages (Caudovirales) and a reduction in overall virome diversity.³¹ Notably, similar patterns have been observed across independent cohorts, including pediatric ileal Crohn’s disease and very early–onset IBD, suggesting that virome remodeling may represent a recurring feature of mucosal inflammatory conditions.^10,32 Although the specific taxonomic trajectories differ between IgAN and IBD, these parallels provide contextual support for the biological relevance of gut virome compositional changes in immune-mediated diseases.³³

We found that phages predicted to target Clostridium, Bacteroides, Roseburia, and several additional genera were exclusively enriched in IgAN, whereas phages predicted to infect Faecalibacterium, Coprococcus, and other genera were uniquely enriched in healthy controls. Notably, in our previous systematic review of the IgAN gut microbiota literature, more than half of published studies reported an increased abundance of Bacteroides in IgAN cohorts, supporting the robustness and cross-cohort reproducibility of this bacterial signature.⁶ Importantly, the accompanying shifts in predicted phage-host associations may suggest that bacteriophage-bacterium interactions represent an additional regulatory dimension for interpreting gut microbial dysbiosis in IgAN. Rather than simply mirroring changes in bacterial abundance, altered phage pressure on functionally distinct bacterial groups may influence bacterial ecological behavior and functional states, thereby potentially shaping the microbial landscape in which dysbiosis is established and maintained. From this interaction-centered perspective, the gut virome may extend current bacteria-focused models by providing a complementary layer of regulation relevant to mucosal immunity and gut-kidney axis communication.

Loss of bacteriome-virome diversity coupling in IgAN indicates altered ecological relationships between phages and their bacterial hosts. In healthy controls, bacterial and viral richness exhibited coordinated patterns, consistent with a relatively synchronized cross-kingdom microbial community. In contrast, this association was no longer observed in IgAN, suggesting disease-associated changes in the structural organization of the gut microbial ecosystem. Network-based analyses further supported this observation by revealing sparser and more centralized interaction patterns in IgAN. Within these inferred networks, Ruthenibacterium was positioned at key junctions linking multiple viral associations, whereas viral families such as Guelinviridae and Hafunaviridae showed a high degree of connectivity within the network structure. These patterns reflect topological features of the inferred bacterial-viral associations rather than broad host infectivity or direct functional dominance. Given the known host specificities of these viral families, including archaeal hosts for Guelinviridae and Clostridial hosts for Hafunaviridae, the observed network configurations likely reflect coordinated shifts in host-associated phage populations accompanying changes in the bacterial and archaeal communities. Together, these findings suggest that IgAN is associated with restructuring of cross-kingdom microbial community architecture, highlighting altered patterns of bacteriome-virome organization without implying direct causal or immune-mediated effects.³⁴

At the functional level, the gut virome in IgAN showed a higher relative abundance of predicted glycan-modifying enzymes, including GH15, GT2, GT39, and GT41. These enzyme families are involved in polysaccharide hydrolysis, glycosyl transfer, and glycan remodeling. The coordinated enrichment of these enzyme families suggests that the virome may actively shape the intestinal glycan microenvironment through functional interactions with bacterial communities, rather than acting solely as a passive genetic component. Functionally, GH15 is primarily associated with the hydrolysis of α-glycosidic linkages and participates in the turnover of complex carbohydrates within the mucosal layer³⁵; whereas GT2, GT39, and GT41 belong to glycosyl transferase families that mediate glycan elongation, branching, and structural modification, thereby influencing glycan complexity and spatial organization.³⁶ The simultaneous enrichment of glycan-degrading and glycan-synthesizing activities is consistent with coordinated glycan remodeling rather than unidirectional carbohydrate depletion, with the potential to result in sustained alterations of mucosal and bacterial surface glycan architecture. Given the central role of glycans in mucosal immune recognition, even modest changes in glycan composition or accessibility may influence antigen exposure, microbial adherence, and IgA binding properties.^36,37 Accordingly, enhanced virome-associated glycan-modifying capacity could indirectly modulate IgA-microbiota interactions by reshaping bacterial surface polysaccharides or host mucin-associated glycans, thereby altering the immunological interface at the intestinal barrier. Given that IgA production, glycosylation, and antigen specificity are tightly coupled to the mucosal glycan milieu, virome-associated perturbations of glycan structure may provide a biologically plausible link between viral functional remodeling and the dysregulated IgA responses characteristic of IgAN.^38,39

Several limitations of this study should be acknowledged. First, this investigation was designed as an exploratory, hypothesis-generating analysis, and the modest sample size limits statistical power and precludes definitive inference regarding disease-associated virome differences. Accordingly, the findings should be interpreted as descriptive rather than confirmatory. Second, the absence of an independent external validation cohort limits assessment of the generalizability of the observed virome features. Third, although cases and controls were matched for key demographic variables and strict exclusion criteria were applied, residual confounding from unmeasured factors, including diet, lifestyle, socioeconomic status, and prior medication exposure, cannot be fully excluded. In addition, current technical and reference limitations in gut virome research, including incomplete viral databases and constrained functional annotation, may lead to underestimation of virome diversity and limit the resolution of virus-host inference. Finally, the cross-sectional design precludes causal interpretation. Future studies incorporating larger cohorts, external validation, longitudinal sampling, as well as integrated multi-omics and immune profiling will be important to further clarify the biological and clinical relevance of gut virome alterations in IgAN.

Overall, this study provides initial observations on gut virome differences in IgAN and highlights potential directions for future investigations aimed at understanding the role of virome-bacteriome interactions in disease pathogenesis.

Disclosure

All the authors declared no competing interests.

Supplementary Material

Supplementary File (PDF)

Supplementary Methods. Figure S1. Principal coordinates analysis based on Bray-Curtis dissimilarity of gut virome profiles. Figure S2. Differential abundance and taxonomic profiles of bacteriophages in mild IgAN. Figure S3. Differential abundance and taxonomic profiles of bacteriophages in severe IgAN. Figure S4. Differential abundance and taxonomic profiles of bacteriophages in IgAN subgroups. Figure S5. Bacteriome-virome alpha diversity relationships in IgAN. Figure S6. Functional signatures of the gut virome in IgAN subgroups. Table S1 Clinical characteristics of the study cohort. Table S2 Differential gut bacterial genera between patients with IgA nephropathy and healthy controls identified by 16S rRNA sequencing. STROBE Checklist.