Atlas of glomerular disease-specific genetic effects on blood transcriptome

Technical Summary: Atlas of Glomerular Disease-Specific Genetic Effects on Blood Transcriptome

Problem Statement
Primary glomerulonephropathies (GN), including IgA nephropathy (IgAN), IgA vasculitis (IgAV), focal segmental glomerulosclerosis (FSGS), membranous nephropathy (MN), and minimal change disease (MCD), represent a major cause of kidney failure worldwide. While genome-wide association studies (GWAS) have identified over 30 risk loci for IgAN and several for other GN types, translating these statistical associations into molecular mechanisms remains challenging. A critical limitation is the lack of adequately powered, genotype-informed regulatory maps derived from large, well-phenotyped cohorts of GN patients. Existing resources, such as GTEx, rely on healthy tissue and may not capture disease-activated regulatory programs. Furthermore, the specific genetic effects on the blood transcriptome (gene expression, splicing, and RNA editing) within the context of distinct GN subtypes are largely undefined.

Methodology
The study leveraged the CureGN study, a prospective cohort of patients with biopsy-diagnosed primary GN. The authors performed:

1. Whole Genome Sequencing (WGS): Conducted at 30x depth on 4,307 individuals with primary GN and 3,845 healthy controls of diverse ancestries.
2. Bulk Blood Transcriptome Sequencing: Performed on 1,822 GN participants with matched WGS data (447 FSGS, 403 IgAN, 123 IgAV, 408 MCD, and 441 MN).
3. Genetic Association Analyses:
   • Common Variants: Genome-wide single-variant association tests stratified by ancestry and meta-analyzed.
   • Rare Variants: Gene-based and gene-set collapsing analyses (Burden, SKAT-O, ACAT) and sliding window scans for rare deleterious variants.
   • Polygenic Risk Scores (PRS): Validation of existing PRS models against healthy and disease controls, and correlation with clinical phenotypes.
4. Quantitative Trait Locus (QTL) Mapping:
   • Context-Specific Mapping: Cis-expression QTLs (eQTLs), splicing QTLs (sQTLs), and double-strand RNA-editing QTLs (edQTLs) were mapped for each GN subtype separately.
   • Covariates: Models controlled for age, sex, genetic ancestry, latent technical factors (PEER), and estimated blood cell-type fractions (via computational deconvolution of bulk RNA-seq).
   • Interaction QTLs (ieQTLs): Mapped interactions between genotypes and age, sex, estimated glomerular filtration rate (eGFR), and proteinuria.
   • Allele-Specific Expression (ASE): Used as an orthogonal validation method for cis-regulatory effects.
5. Integration and Colocalization:
   • Colocalization: Used COLOC to determine if GWAS signals and QTL signals share causal variants (PP4 > 0.8) or distinct variants (PP3 > 0.8).
   • Cross-Disease Mapping: Performed QTL mapping across all 1,822 participants with adjustment for GN type to identify shared regulatory architecture.

Key Results

• QTL Discovery: The study identified 16,068 eGenes, 4,644 sGenes, and 4,611 edQTLs (at FDR < 0.05) in at least one GN type. Approximately 5–10% of these signals were unique to a specific GN type, while ~90% were shared between at least two conditions.
• Context Specificity: Colocalization analysis revealed that ~80% of shared eGenes between traits were driven by the same causal variant, whereas ~2% had distinct causal variants, indicating context-specific regulatory mechanisms for the same gene (e.g., RUNX3 in FSGS vs. IgAN).
• Novelty vs. GTEx: Cross-phenotype QTL mapping uncovered 6,466 eGenes, 2,705 sGenes, and 5,321 edQTLs not previously detected in GTEx blood tissue. These novel discoveries were significantly enriched in immune-related pathways (e.g., viral infection, cytokine signaling, neutrophil degranulation).
• Clinical and Environmental Interactions:
  • Age: Identified 232 age-dependent eQTLs. For example, the genetic effect of RBL2 expression was attenuated with age, while RGMB genetic effects were enhanced.
  • Disease Severity: Identified 57 eQTLs interacting with eGFR and 13 with proteinuria. Notably, proteinuria attenuated the genetic regulation of LDLRAP1, potentially linking genetic risk to hyperlipidemia in nephrotic syndrome.
• GWAS Integration: Colocalization prioritized candidate causal genes for 13 GWAS loci. Notable findings include:
  • TNFSF13 (IgAN): Strong colocalization (PP4=1.00) between GWAS, eQTL, sQTL, and protein QTL (pQTL) signals. The risk allele is associated with increased expression of the full-length transcript and elevated soluble APRIL levels.
  • CARD9 and ITGAX (IgAN): Colocalization with splicing events, suggesting risk alleles alter splicing patterns (e.g., truncation of the CARD domain).
  • FCAR (IgAV): Colocalization with an RNA editing event in the 3' UTR.
• Rare Variants: While no new genome-wide significant rare variant associations were found, rare variant collapsing identified significant associations between FSGS and gene sets related to "minimal change glomerulonephritis" and "foamy urine."

Significance and Claims
The paper claims to provide the first comprehensive, disease-context-specific atlas of genetic effects on the blood transcriptome for primary GN. By coupling WGS with bulk blood RNA-seq in a large, diverse cohort, the study generates a resource that:

1. Expands Regulatory Maps: Greatly expands the catalog of genetic variants associated with transcriptomic effects in blood, specifically identifying thousands of QTLs absent from healthy tissue references like GTEx.
2. Refines GWAS Interpretation: Demonstrates that integrating GWAS with disease-context-specific QTLs improves the biological interpretability of established loci and prioritizes candidate molecular mediators (e.g., TNFSF13, NFKB1, CARD9).
3. Highlights Disease Context: Underscores that while the majority of regulatory architecture is shared across GN types, a measurable fraction of genetic effects is phenotype-specific or driven by distinct causal variants, reflecting the unique immune dysregulation of each disease.
4. Dynamic Regulation: Provides a framework for understanding how inherited variation influences immune pathways dynamically across the lifespan and in response to disease severity (eGFR, proteinuria).

The authors conclude that this resource offers critical tools for functional annotation and improved interpretability of novel GWAS loci as sample sizes for GN continue to increase. They acknowledge limitations, including the use of bulk RNA-seq (necessitating deconvolution) and the need for larger samples to fully characterize rare variant contributions.