Population-scale genomics reveals divergent pathogenicity of variant classes across paralogous collagen IV genes

Problem Statement
Alport Syndrome (AS) is a hereditary kidney disorder caused by pathogenic variants in the paralogous genes COL4A3, COL4A4, and COL4A5, which encode the $\alpha3(IV)$, $\alpha4(IV)$, and $\alpha5(IV)$ chains of Type IV collagen. While the clinical severity of biallelic or X-linked variants is well-characterized, the pathogenicity of heterozygous autosomal variants in COL4A3 and COL4A4 remains less understood. Current clinical guidelines and variant classification frameworks often treat these two paralogous genes as functionally interchangeable, assuming equivalent pathogenicity for the same variant classes (e.g., glycine substitutions, truncating variants, and non-collagenous domain [NC1] missense variants). However, this assumption has not been rigorously tested at a population scale, potentially leading to inaccurate risk stratification and genetic counseling.

Methodology
The authors leveraged whole-genome sequencing (WGS) data from the UK Biobank (UKB; n ≈ 500,000) and replicated findings in the All of Us Research Program (n ≈ 414,000), restricting analyses to individuals of European ancestry. The study employed a multi-faceted approach:

1. Phenotype Definition: Cases were defined for glomerular haematuria, proteinuria (using ICD-10 codes and urinary albumin-to-creatinine ratios), and severe chronic kidney disease (CKD), with rigorous exclusion of non-glomerular causes.
2. Per-Variant Association: Fisher's Exact Test was used to assess associations between individual rare coding variants and renal phenotypes.
3. Gene-Based Collapsing and PheWAS: Rare predicted pathogenic variants were classified into three functional categories: Glycine substitutions (Gly), Truncating variants (Trunc), and NC1 domain missense variants. These were collapsed into binary carrier status variables for COL4A3 and COL4A4 separately. Phenome-wide association studies (PheWAS) were conducted using SAIGE to test associations against 1,363 ICD-10 phenotypes.
4. Effect Size Comparison: The authors formally tested for differences in genetic effect sizes ($\beta$) between COL4A3 and COL4A4 for each variant class using two-tailed Z-tests.
5. Genome-Wide Association Studies (GWAS): GWAS were performed to identify loci associated with haematuria, including a secondary analysis adjusting for a weighted burden score (WBS) of rare COL4A3/A4 variants to determine if the signal was driven by rare variants or independent common variants/modifiers.

Key Results

• Variant Associations: The study identified 64 COL4A3 and 92 COL4A4 rare variants significantly associated with haematuria or proteinuria. Glycine substitutions were the most frequent class across all phenotypes.
• Gene-Specific Divergence: A critical finding was the divergent pathogenicity between the two genes based on variant class:
  • Glycine Substitutions: Conferred similar risks in both COL4A3 and COL4A4.
  • Truncating and NC1 Missense Variants: These were strongly associated with haematuria and proteinuria in COL4A4 carriers. In contrast, COL4A3 carriers with truncating or NC1 variants showed substantially attenuated or absent associations, despite comparable carrier frequencies and predicted pathogenicity scores (CADD/REVEL).
  • This divergence persisted after excluding the British founder variant p.Ser969Ter in COL4A4 and was replicated in the All of Us cohort.
• Carrier Rates: COL4A4 truncating and NC1 missense carriers exhibited significantly higher case rates for proteinuria (52.7% and 37.2%, respectively) compared to COL4A3 carriers (17.1% and 12.1%).
• Genetic Architecture: The COL4A3/COL4A4 locus was identified as the dominant genetic determinant of haematuria. The association signal was entirely attributable to the aggregate effects of rare coding variants; adjusting for a weighted burden score abolished the locus signal, with no evidence of independent common variant or trans-acting modifier effects. The HLA-B and THSD4 loci were identified as independent signals associated with haematuria.

Significance and Claims
The paper claims to provide the first population-scale evidence that COL4A3 and COL4A4 are not functionally interchangeable regarding the pathogenicity of truncating and NC1 domain variants. The authors conclude that:

1. Gene Identity Matters: Risk stratification and variant interpretation for collagen IV nephropathies must consider gene identity, not just variant class. COL4A4 truncating and NC1 variants confer a substantially stronger pathogenic effect than their COL4A3 counterparts.
2. Clinical Implications: These findings challenge the assumption of equivalent pathogenicity across paralogous genes, suggesting that current guidelines may overestimate risk for COL4A3 truncating/NC1 carriers and underestimate it for COL4A4 carriers.
3. Broader Principle: The study demonstrates that paralogous genes encoding subunits of the same heteromeric complex can differ substantially in their tolerance to loss-of-function variation. This suggests a need for "gene-aware" rather than "variant-class-aware" frameworks in clinical genetics, a principle potentially applicable to other heteromultimer-encoding gene families.
4. Mechanistic Insight: The results suggest that the $\alpha4(IV)$ chain may play a disproportionately important role in chain selection or protomer stabilization compared to $\alpha3(IV)$, though the specific biochemical mechanisms remain to be elucidated.

The authors note limitations, including the restriction to European ancestry and the reliance on hospital-based ICD-10 codes which may underestimate the prevalence of asymptomatic microscopic haematuria. Consequently, the reported carrier case rates are interpreted as conservative estimates of penetrance.