Human C-terminal CUBN variants associate with chronic proteinuria and normal renal function

Abstract

BACKGROUNDProteinuria is considered an unfavorable clinical condition that accelerates renal and cardiovascular disease. However, it is not clear whether all forms of proteinuria are damaging. Mutations in CUBN cause Imerslund-Gräsbeck syndrome (IGS), which is characterized by intestinal malabsorption of vitamin B12 and in some cases proteinuria. CUBN encodes for cubilin, an intestinal and proximal tubular uptake receptor containing 27 CUB domains for ligand binding.METHODSWe used next-generation sequencing for renal disease genes to genotype cohorts of patients with suspected hereditary renal disease and chronic proteinuria. CUBN variants were analyzed using bioinformatics, structural modeling, and epidemiological methods.RESULTSWe identified 39 patients, in whom biallelic pathogenic variants in the CUBN gene were associated with chronic isolated proteinuria and early childhood onset. Since the proteinuria in these patients had a high proportion of albuminuria, glomerular diseases such as steroid-resistant nephrotic syndrome or Alport syndrome were often the primary clinical diagnosis, motivating renal biopsies and the use of proteinuria-lowering treatments. However, renal function was normal in all cases. By contrast, we did not found any biallelic CUBN variants in proteinuric patients with reduced renal function or focal segmental glomerulosclerosis. Unlike the more N-terminal IGS mutations, 37 of the 41 proteinuria-associated CUBN variants led to modifications or truncations after the vitamin B12-binding domain. Finally, we show that 4 C-terminal CUBN variants are associated with albuminuria and slightly increased GFR in meta-analyses of large population-based cohorts.CONCLUSIONCollectively, our data suggest an important role for the C-terminal half of cubilin in renal albumin reabsorption. Albuminuria due to reduced cubilin function could be an unexpectedly common benign condition in humans that may not require any proteinuria-lowering treatment or renal biopsy.FUNDINGATIP-Avenir program, Fondation Bettencourt-Schueller (Liliane Bettencourt Chair of Developmental Biology), Agence Nationale de la Recherche (ANR) Investissements d'avenir program (ANR-10-IAHU-01) and NEPHROFLY (ANR-14-ACHN-0013, to MS), Steno Collaborative Grant 2018 (NNF18OC0052457, to TSA and MS), Heisenberg Professorship of the German Research Foundation (KO 3598/5-1, to AK), Deutsche Forschungsgemeinschaft (DFG) Collaborative Research Centre (SFB) KIDGEM 1140 (project 246781735, to CB), and Federal Ministry of Education and Research (BMB) (01GM1515C, to CB).

Keywords: Chronic kidney disease; Genetic diseases; Genetics; Nephrology.

Figure 1. Flow chart for cohort genotyping.

The 3 different cohorts consisted of individuals with suspected genetic diseases (see Methods for more details). Altogether, 39 individuals with biallelic filtered CUBN variants were identified. In the chronic PU cohort, 11 individuals with biallelic filtered CUBN variants were identified, plus 1 additional individual carrying 1 biallelic filtered CUBN variant and the GWAS variant p.N2157D (see Supplemental Table 7). HGMD, Human Gene Mutation Database; PT, proximal tubule.

Figure 2. Frequency, density, and position of CUBN variants along the cubilin protein.

Cubilin protein structure summarized with the 8 EGF-like (light blue) and 27 CUB domains (darker blue). The red dots correspond to the theoretical Ca²⁺-binding sites, and the arrows on top indicate the variants found in the 3 cohorts of this study (the green arrows represent the filtered CUBN variants, and the red arrows indicate the 4 GWAS missense variants A1690V, N2157D, A2914V, and I2984V). Gaussian kernel density curves (with a 0.2 width parameter) as well as the position of HGMD variants (red), the variants from this study (green), and PTVs from the in-house (yellow) and gnomAD (blue) databases are shown as curves and vertical lines, respectively. The minor allele frequencies (MAFs) of the HGMD and gnomAD variants are shown as red and blue dots, respectively.

Figure 3. Structural modeling of CUBN missense variants.

(A) Location of the cubilin T55M missense variant in RCSB entry 6GJE. The variant is located in the hydrophobic core of the cubilin β-helix that interacts with amnionless. (B) Location of the cubilin N1303H missense variant in RCSB entry 3KQ4. The variant is located in CUB domain 8, close to the interface with IF/B12. Red spheres represent Ca²⁺ ions binding to the CUB domains. (C) Location of the cubilin missense plus the 4 GWAS variants (in red) in in silico structural models of CUB domains 11–27. Red spheres represent Ca²⁺ ions predicted to bind CUB domains 11, 13, and 26. The variants S1947Y and D3492Y are notably close to these Ca²⁺-binding motifs. (D) Example of 1 CUB domain with β-sheets (β2–10) and loops (L2–9).

Figure 4. Clinical and biological profiles of patients carrying biallelic filtered CUBN variants.

(A) Proportions of carriers of biallelic filtered CUBN variants in the chronic PU cohort (n = 107), when considering only patients with normal renal function (eGFR >60 mL/min per 1.73 m², n = 39) or only patients with ESRD (eGFR <15 mL/min per 1.73 m², n = 30). (B–D) Patients with biallelic filtered CUBN variants from the 3 cohorts (genetic kidney disease cohorts I and II, chronic PU cohort) were merged into the group of patients with biallelic filtered CUBN variants. (B) The UAPR is plotted for patients with (n = 27) and without (n = 9) biallelic filtered CUBN variants. As a general rule, glomerular proteinuria is characterized by a UAPR above 50%, whereas tubular proteinuria was below 50% (7, 43). (C) Age of first manifestation of proteinuria for patients with (n = 38) and without (n = 88) biallelic filtered CUBN variants. Data represent mean values ± 95% CIs. ***P < 0.0001, by t test. (D) Renal function of patients with biallelic filtered CUBN variants (n = 35) was found to be declining according to normal aging, whereas in patients without biallelic filtered CUBN variants (n = 79), the decline was more rapid. Renal function (eGFR) was calculated using the Schwartz formula for children and the CKD-EPI formula for adults. The blue and red dashed lines represent the logarithmic trend curves for both groups.