In the present study we followed up the results of a recent GWAS, which identified the cubilin gene CUBN
as a locus for albuminuria 
. As albuminuria is an established risk factor for progressive renal function loss, the GWAS findings raised the hypothesis that genetic variation in the CUBN
locus could be associated with progressive renal function loss and finally end stage renal disease. To test this hypothesis we studied the cited top SNP as well as a tagSNP in CUBN
in relation to final renal clinical outcomes, namely ESRD in native kidneys and GF in the transplanted kidney.
In a case-control design we studied rs7918972 and rs1801239 genotypes in ESRD patients versus kidney donors. The MAF for rs7918972 was significantly higher in ESRD patients as compared to kidney donors, imposing a 39% increased risk for ESRD per copy of the minor allele. Follow-up data after transplantation showed direction-consistent trend for an association between donor kidney rs7918972 and development of GF in recipients. Thus, the SNP in CUBN locus was associated with susceptibility to develop ESRD in two settings, namely ESRD in native kidneys and GF in transplanted kidneys.
Transplantation represents a unique setting, also from genetic point view: an organ with its own genotype functions in an organism with another genotype. We tested both donor and recipient genotype for association with the renal outcome to investigate whether it is the kidney genotype that determines its own fate or it is the recipient genotype that influences function and survival of the transplanted organ. This unique design is useful for genetic research in nephrology as it enables discrimination between the renal and extra-renal mechanisms 
In our study, it was donor rather than recipient CUBN genotype that was associated with GF, suggesting involvement of local, intra-renal pathways in processes of transplanted kidney survival which are independent of systemic influences.
Albuminuria is known as a predictor of cardiovascular and non-cardiovascular mortality 
. However, in our study CUBN
genotypes did not associate with cerebro- or cardiovascular accident as a cause of death in donors and cardiovascular and all-cause mortality after transplantation in recipients.
As no albuminuria data were available and urinary albumin levels are known to correlate with total protein, we tested association of the CUBN
SNPs with 24-h total urinary protein excretion as a surrogate phenotype. Interestingly, we found donor
rs7918972 to be associated with elevated proteinuria levels cross-sectionally at 1 year after transplantation. This is consistent with our results of association with the outcome, and also in line with the results of a recent study which revealed, using exome sequencing, a deleterious mutation in CUBN
in a family of proteinuric patients, thus confirming the CUBN
gene involvement in proteinuria 
In the original GWAS 
SNP rs1801239, associated with elevated urinary albumine-to-creatinine ratio and microalbuminuria. However, this SNP was not associated with CKD or estimated GFR. In agreement with this, our case-control study showed no association between this SNP and ESRD. Also, rs1801239 was not associated with GF in our longitudinal study Instead, it was the other CUBN
polymorphism, the tagSNP rs7918972, that was associated with ESRD in our study.
locus is characterized by a high variability, with both common and rare mutations. Mutations in the CUBN
locus are known to be the cause of Imerslund-Gräsbeck syndrome (OMIM #261100, Finnish type) which is a rare (the estimated prevalence is <6
1,000,000) autosomal recessive disorder characterized by vitamin B12 deficiency commonly resulting in megaloblastic anemia, and also neurological damage and mild proteinuria 
. However, we did not aim to address previously clinically-associated Mendelian mutations in the CUBN
in our study. We aimed to investigate whether common
variation, as opposed to rare mutations in Imerslund-Gräsbeck syndrome, in the CUBN
associates with kidney disease. In the same time, we targeted a lower part of the common variability range, with MAFs between 10 and 15%, aiming to reveal allegedly stronger genetic effects. We selected two SNPs in the CUBN
locus for the present study: first, the one previously published to be associated with albuminuria levels in the general population, i.e. the missense variant (Ile2984Val) rs1801239, and second the tagSNP in the CUBN
intron, rs7918972. The latter is in high linkage disequilibrium (r2
0.831) with another missense variant rs1801232 (Asn3552Lys) in CUBN
, which might be responsible for the biological impact of the polymorphism on the protein level. The minor allele of rs1801232 leads to an asparagine-to-lysine amino acid substitution in the C-terminal CUB27 domain of cubilin. Despite the amino acids differ in chemical properties (isoelectric points: Asn 5.4, Lys 9.8), the substitution was predicted to be benign by bioinformatics algorithms. However, the mutation is close to sites of N-glycosylation (amino acid 3533) and di-sulfide bond (between amino acids 3564 and 3586) and therefore might potentially interfere with secondary protein structure and, consequently, function. The Imerslund-Gräsbeck syndrome mutations, for which functionality was proven, affect the IF-cobalamin-binding region in the CUB8 domain of cubilin (rs121434430 Pro1297Leu), CUB6 domain (CUBN
IVS6 C-G in-frame insertion) or CUB23 domain (CUBN
IVS23 G-T transversion at the conserved donor splice site of exon 23). The SNPs that we studied were spatially distant from these variants and located to the CUB22 domain (rs1801239) and CUB27 domain (rs1801232 tagged by rs7918972).
The rs7918972 CUBN
SNP, associated with ESRD and GF in our study, is localized in high proximity to the neighboring gene, RSU1
. Although the nine SNPs tagged by rs7918972 are all located in the CUBN
locus, linkage disequilibrium with and involvement of the RSU1
is theoretically possible and cannot be entirely ruled out (Suppl. Fig. S1
). The RSU1
gene encodes Ras suppressor protein 1, which participates in the Ras signal transduction pathway, growth inhibition and nerve-growth factor induced differentiation processes. Its mRNA is expressed in the kidney (according to the NCBI GEO profiles), in a low-to-moderate quantity (51 transcripts per million, according to the NCBI EST profiles). However, functional proof is beyond the scope of the present study, and further research will be needed to discriminate between the effects of these neighboring genes.
Interestingly, we found an interaction between the two SNPs studied. According to our data, the rs7918972 minor allele requires a copy of the rs1801239 minor allele to express its risk phenotype, whereas the minor allele of rs1801239 displays protective effect in the absence of rs7918972 minor allele. This pattern was observed in the case-control study and warrants further investigation to determine whether found statistical interaction has biological implications.
Our study was conducted in kidney transplant recipients, and thus reflects a population that developed ESRD in their native kidneys and was eligible to receive a transplantation. As such, renal transplant recipients represent a relatively healthy subset of the ESRD population, a selection bias inherent to any study in renal transplantation. This should be considered a limitation to our study. In chronic renal disease, both in native kidney disease and in transplantation, mortality is high, and for analyses on ESRD the competing risks of mortality, in particular cardiovascular, are therefore relevant to consider. In the current population, mortality with functioning graft was 15.8 percent, and no association with either of the CUBN SNPs was observed.
Our longitudinal study of graft failure may have been underpowered to detect a significant SNP effect. Insufficient power might thus be an explanation of the fact that convincing statistical significance was not reached in the graft survival analysis for association with rs7918972. Studies in larger populations are warranted to confirm an association between the CUBN SNP and GF.
Our study confirms association of the CUBN with renal phenotypes of progressive renal function loss and urine protein loss. We first identified CUBN SNP rs7918972 as a novel genetic variant of susceptibility for ESRD in a case-control design. In a separate proof-of-principle longitudinal study, which served as an internal replication, we reproduced the association. Thus, rs7918972 was associated with susceptibility to develop progressive renal function loss in two settings, namely ESRD in native kidneys and GF in transplanted kidneys. It was kidney genotype that associated with increased risk, supporting impact of intra-renal pathways on organ damage. Our study set-up – analyzing both donor and recipient genotypes – provides a powerful design for hypothesis-driven studies on risk loci for renal damage enabling differentiation between local, intra-renal, and systemic, extra-renal, influences.