Human C-Terminal CUBN Variants Associate with Chronic Proteinuria and Normal Renal Function

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

Mathilda Bedin, … , Corinne Antignac, Matias Simons

J Clin Invest. 2019. https://doi.org/10.1172/JCI129937.

Clinical Medicine In-Press Preview Genetics Nephrology

Background: Proteinuria is considered as an unfavorable clinical condition that accelerates renal and cardiovascular disease. However, it is not clear if all forms of proteinuria are damaging. Mutations in CUBN cause Imerslund-Gräsbeck syndrome (IGS) featured 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.

Methods: We 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.

Results: We identified 39 patients, in whom biallelic pathogenic variants in theC UBN gene are associated with chronic isolated proteinuria with childhood onset. Since the proteinuria displayed 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 proteinuria-lowering treatments. Yet, renal function was normal in all cases. By contrast, we did not find any biallelic pathogenic CUBN variants in patients with reduced renal function or focal segmental glomerulosclerosis. Unlike the more N-terminal IGS mutations, 37 out of the […]

Find the latest version: https://jci.me/129937/pdf

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

2 function

4 Mathilda Bedin1, Olivia Boyer2,3, Aude Servais2,4, Yong Li5, , Laure Villoing-Gaudé1, Marie- 5 Josephe Tête2, Alexandra Cambier7, Julien Hogan7, Veronique Baudouin7, Saoussen Krid3, 6 Albert Bensman3, Florie Lammens8, Ferielle Louillet9, Bruno Ranchin10, Cecile Vigneau11, 7 Iseline Bouteau12, Corinne Isnard-Bagnis13, Christoph J. Mache14, Tobias Schäfer15, Lars 8 Pape16, Markus Gödel17, Tobias B. Huber17, Marcus Benz18, Günter Klaus19, Matthias 9 Hansen20, Kay Latta20, Olivier Gribouval2, Vincent Morinière21, Carole Tournant21, Maik 10 Grohmann22,23, Elisa Kuhn22, Timo Wagner22, Christine Bole-Feysot24, Fabienne Jabot- 11 Hanin24, Patrick Nitschké24, Tarunveer S. Ahluwalia25, Anna Köttgen5, Christian Brix Folsted 12 Andersen26, Carsten Bergmann22,23,27, Corinne Antignac2,21, Matias Simons1 13 14 1 Laboratory of Epithelial Biology and Disease, Imagine Institute, INSERM U1163, Université 15 de Paris, Paris, France 16 2 Laboratory of Hereditary Kidney Disease, Imagine Institute, INSERM U1163, Université de 17 Paris, Paris, France 18 3 Department of Pediatric Nephrology, Necker Hospital, Assistance Publique Hopitaux de 19 Paris (APHP), Paris, France 20 4 Department of Nephrology, Necker Hospital, Assistance Publique Hopitaux de Paris 21 (APHP), Paris, France 22 5 Institute of Genetic Epidemiology, Faculty of Medicine and Medical Center, University of 23 Freiburg, Freiburg, Germany 24 7 Department of Pediatric Nephrology and Transplantation, APHP, Robert-Debré Hospital, 25 Paris, France 26 8 Centre Hospitalier Régional Universitaire de Lille, Lille, France 27 9 Centre Hospitalier Universitaire de Rouen, Rouen, France 28 10 Department of Pediatric Nephrology, Hospices Civils de Lyon, Bron, France 29 11 Centre Hospitalier Universitaire de Rennes, INSERM, U1085 IRSET-9, Rennes, France 30 12 Centre Hospitalier Universitaire de Poitiers, Poitiers, France 31 13 Department of Nephrology, Pitié Salpetrière Hospital, Paris, France 32 14 Children’s Hospital, Medical University Graz, Austria 33 15 Renal Division, University Medical Center Freiburg, Freiburg, Germany 34 16 Department of Pediatric Kidney, Liver and Metabolic Disease, Hannover Medical School, 35 Hannover, Germany 36 17 III. Department of Medicine, University Medical Center Hamburg-Eppendorf, Hamburg, 37 Germany

1 18 Kindernephrologie Dachau, Dachau, Germany 2 19 Department of Child and Adolescent Medicine, University Medical Center Marburg- 3 Giessen, Marburg, Germany 4 20 KfH-Nierenzentrum für Kinder und Jugendliche und Clementine-Kinderhospital, Frankfurt, 5 Germany 6 21 Department of Genetics, Necker Hospital, Assistance Publique Hopitaux de Paris (APHP), 7 Paris, France 8 22 Center for Human Genetics, Bioscientia, Ingelheim 9 23 Center for Human Genetics, Mainz 10 24 Bioinformatic Platform, INSERM UMR 1163, Institut Imagine, 75015 Paris, 11 France. Bioinformatics Core Facility, Université de Paris - Structure Fédérative de Recherche 12 Necker, INSERM US24/CNRS UMS3633, Paris, France 13 25 Steno Diabetes Center Copenhagen, Gentofte, Denmark 14 26 Department of Biomedicine, Aarhus University, Aarhus, Denmark 15 27 Renal Division, Department of Medicine, University Hospital Freiburg, Freiburg, Germany 16

18 Correspondance should be addressed to: 19 Matias SIMONS, INSERM UMR1163, Laboratory of Epithelial Biology and Disease, Imagine 20 Institute, 24 Boulevard du Montparnasse, 75015, Paris, France 21 e-mail : [email protected] 22 Tel: +33 (0)1 42 75 4331 23 24 Running title: 25 CUBN variants and isolated proteinuria 26

27 Conflict of interest statement:

28 The authors have declared that no conflict of interest exists.

1 Abstract 2 Background 3 Proteinuria is considered as an unfavorable clinical condition that accelerates renal and 4 cardiovascular disease. However, it is not clear if all forms of proteinuria are damaging. 5 Mutations in CUBN cause Imerslund-Gräsbeck syndrome (IGS) featured by intestinal 6 malabsorption of vitamin B12 and in some cases proteinuria. CUBN encodes for cubilin, an 7 intestinal and proximal tubular uptake receptor containing 27 CUB domains for ligand 8 binding. 9 10 Methods 11 We used next-generation sequencing for renal disease genes to genotype cohorts of patients 12 with suspected hereditary renal disease and chronic proteinuria. CUBN variants were 13 analyzed using bioinformatics, structural modeling and epidemiological methods. 14 15 Results 16 We identified 39 patients, in whom biallelic pathogenic variants in the CUBN gene are 17 associated with chronic isolated proteinuria with an early childhood onset. Since the 18 proteinuria displayed a high proportion of albuminuria, glomerular diseases such as steroid- 19 resistant nephrotic syndrome or Alport syndrome were often the primary clinical diagnosis, 20 motivating renal biopsies and proteinuria-lowering treatments. Yet, renal function was normal 21 in all cases. By contrast, we did not find any biallelic pathogenic CUBN variants in proteinuric 22 patients with reduced renal function or focal segmental glomerulosclerosis. Unlike the more 23 N-terminal IGS mutations, 37 out of the 41 proteinuria-associated CUBN variants led to 24 modifications or truncations after the vitamin B12-binding domain. Finally, we show that four 25 C-terminal CUBN variants are associated with albuminuria and slightly increased GFR in 26 meta-analyses of large population-based cohorts. 27 28 Conclusions 29 Collectively, our data suggest an important role for the C-terminal half of cubilin in renal 30 albumin reabsorption. Albuminuria due to reduced cubilin function could be an unexpectedly 31 common benign condition in humans that may not require any proteinuria-lowering treatment 32 or renal biopsy. 33 34

1 Introduction

2 The loss of proteins into the urine (or proteinuria) is an important risk factor for renal and

3 cardiovascular disease. In particular, albuminuria is associated with an increased risk for

4 chronic and diabetic kidney disease (CKD and DKD, respectively), end-stage renal disease

5 (ESRD) and mortality (1-3). Although the reasons for this are not entirely clear, the overload

6 of the renal tubules with proteins and albumin-bound lipids has been proposed to be

7 damaging for tubular epithelial cells (4),(5). Anti-proteinuric therapy, for example through

8 angiotensin-converting-enzyme (ACE) or angiotensin II receptor (AT1) inhibition, is therefore

9 an important renoprotective therapy (6).

10 The main cause of proteinuria is the dysfunction of the glomerular filtration barrier, which

11 leads to symptoms like edema related to the massive albumin loss into the urine. Genetic

12 forms of glomerular proteinuria are steroid-resistant nephrotic syndrome (SRNS) or Alport

13 syndrome (AS). Another form of proteinuria is caused by defects in the proximal tubular

14 protein reabsorption. Tubular proteinuria is of smaller range than glomerular proteinuria

15 because this kind of proteinuria only affects the proteins that are filtered by the glomerulus.

16 Typically, these are proteins, such as β2-microglobulin, with a smaller size than albumin (7).

17 Albumin itself is filtered to a limited extent and usually accounts for less than half of the

18 urinary protein in tubular proteinuria (8). One example is Dent’s disease, in which mutations

19 in CLCN5 and OCRL1 lead to defective trafficking of the uptake receptor complex (9),

20 consisting of megalin (gene name: LRP2), cubilin (CUBN) and amnionless (AMN) (10, 11).

21 While mutations in LRP2 cause Donnai-Barrow syndrome, a multi-system developmental

22 disorder (12), CUBN and AMN mutations lead to IGS, featured by intestinal vitamin B12

23 malabsorption and in about half the cases proteinuria (13). The phenotypes reflect the

24 expression patterns of all three proteins: while megalin is expressed more broadly, the

25 expression of cubilin and amnionless is mostly limited to the small intestine and the kidney.

26 Within the kidney, recent single RNA-sequencing studies of mouse and human kidney have

27 suggested exclusive expression in the proximal tubular compartment for all three proteins

28 (14-16).

1 With regard to the protein structures, megalin and amnionless are type I transmembrane

2 proteins, whereas cubilin is a peripheral protein that requires amnionless for anchoring to the

3 membrane. Anchoring occurs via β-helix-β-helix association between amnionless and the N-

4 terminal hydrophobic stretches of three cubilin subunits (17). Each cubilin protomer has 8

5 EGF domains and 27 CUB domains, some of which are involved in Ca2+-dependent ligand

6 binding (17, 18). Most IGS mutations of CUBN are in the N-terminal half of cubilin, either

7 affecting the interactions with amnionless or the vitamin B12/intrinsic factor (IF)-binding CUB

8 domains 5-8 (CUB5-8; CUB stands for Complement C1r/C1s, Uegf (epidermal growth factor-

9 related sea urchin protein) and Bone morphogenic protein-1) (17). Interestingly, one

10 individual with a homozygous deletion of exon 53 harboring CUB20 was shown to have

11 proteinuria without vitamin B12 malabsorption (19). Several CUBN variants have also shown

12 strong associations with albuminuria in recent genome-wide association studies (GWAS)

13 (20-22), which is in agreement with the albuminuria observed in cubilin knockout mice (10).

14 Altogether, these findings suggest that cubilin could be necessary for preventing urinary

15 albumin loss in humans. However, it is unclear whether the albuminuria due to cubilin

16 dysfunction impairs renal function.

17 Here, we identify a large number of patients with isolated proteinuria associated with biallelic

18 variants in the CUBN gene in three different cohorts of individuals with suspected genetic

19 renal disease. Almost all variants are located after the vitamin B12/IF-binding domain,

20 suggesting that more C-terminal CUB domains are crucial for renal protein reabsorption.

21 Importantly, we show that cubilin deficiency leads to proteinuria with a high proportion of

22 urinary albumin without impairing renal filtration function. In addition, we find that four C-

23 terminal missense variants with strong albuminuria associations in GWAS are related to

24 slightly higher estimated glomerular filtration rate (eGFR) levels in a large meta-analysis of

25 individuals from the general population.

1 Results

2 Biallelic CUBN variants cause isolated proteinuria with normal renal function

3 Next-generation sequencing (NGS) data obtained using a panel containing 309 renal disease

4 genes (Renome panel) on a French cohort of 759 patients were retrospectively analyzed. All

5 cases had been referred to our reference center for genetic testing because of suspected

6 genetic kidney disease (Genetic kidney disease cohort I; Figure 1 and Supplemental Text).

7 We grouped the patients into a non-proteinuric and a proteinuric group, with the latter

8 consisting of suspected SRNS and AS (Supplemental Figure 1). While synonymous and non-

9 synonymous missense variants were equally distributed between the two groups, all protein-

10 truncating variants (PTVs), including splicing, frameshift and stop-gain variants, were

11 strongly enriched in the proteinuric group (Supplemental Figure 1A,B), confirming the

12 association of CUBN variants with proteinuria.

13 Based on the recessive nature of IGS, we asked whether there were any patients with

14 biallelic CUBN variants passing established filtering criteria for Mendelian disease in the

15 proteinuric group (see Methods for filtering criteria). We found 14 patients from 11 families of

16 European or African descent with biallelic likely pathogenic variants in the CUBN gene (Table

17 1; Supplemental Table 1 and 2). By contrast, the non-proteinuric group contained only two

18 cases with biallelic CUBN variants and a control group matched for ethnic groups did not

19 contain any cases (Supplemental Figure 1C). Interestingly, all 14 patients with biallelic CUBN

20 variants showed a very similar phenotype. While no signs of vitamin B12 deficiency, such as

21 megaloblastic anemia, could be detected, the patients shared a chronic proteinuria ranging

22 from 0.5 to 3g/24h. The average age of onset of the proteinuria discovery was 10.9 years

23 (Table 1; Supplemental Table 1). Although not measured in all cases, the proportion of

24 albumin in the urinary protein was higher than 50% and urinary α1- or β2-microglobulin was

25 mostly low or absent (Table 1; Supplemental Table 1) (7, 8), which is in contrast to other

26 forms of tubular proteinuria, including megalin deficiency (12).

1 In all cases, renal function was normal at an average age of 17 years, as measured by

2 serum creatinine levels and eGFR (Table 1; Supplemental Table 1). For this reason,

3 obtaining follow-up clinical information for these patients was sometimes difficult. Renal

4 biopsies had been performed in 9 cases, and in all of them lesions were minimal, unspecific

5 or not present (Table 1; Supplemental Tables 1 and 2). In 7 patients, treatments with ACE

6 inhibitors had already been started but without any lowering effects on the proteinuria. These

7 findings were confirmed in an independent German and Austrian cohort of 1350 suspected

8 SRNS and AS cases (Genetic kidney disease cohort II), in which we identified 13 additional

9 individuals from 12 families with biallelic CUBN variants (Table 1; Supplemental Tables 1, 2

10 and 3). Also here, the chronic subnephrotic proteinuria, featured by a high proportion of

11 urinary albumin, was combined with normal renal function in all patients (Table 1;

12 Supplemental Table 1).

14 Proteinuria-associated CUBN variants localize to C-terminal CUBN domains

15 Altogether, we found 30 novel variants in these two studies (Figure 2; Supplemental Figure

16 2) that not only associated with proteinuria as an established cubilin phenotype but also

17 passed rather stringent filtering criteria (see methods). While three variants (c.6125-2A>G,

18 p.R2030* and p.Y3018S) were found in both cohorts, all variants were found with

19 frequencies below 0.1% in our in-house genome database (mostly enriched for European or

20 African ancestries) or in public reference genome databases (e.g. gnomAD (23))

21 (Supplemental Table 4). In the crystal structure of cubilin (18) or in silico models of individual

22 CUB domains, all missense variants were predicted to have different detrimental effects on

23 cubilin function, ranging from effects on amnionless binding (p.T55M) to stability and ligand

24 binding of CUB domains (all other variants) (Figure 3A-D; Supplemental Table 5). Most

25 variants affected residues or led to truncations after CUB8 (aa: 1487-3618), which is in

26 contrast to the previously described IGS mutations that are all before CUB8 (aa: 66-1390)

27 (24) (Figure 2; Supplemental Figure 2). The only two variants before CUB8, p.T55M and

28 p.W1158*, were found to be in trans with variants after CUB8, consistent with the association

1 of C-terminal CUBN variants with isolated proteinuria. As most PTVs present in the general

2 population, such as our in-house genome database or gnomAD, lead to truncations after

3 CUB8 (Figure 2), it may even be concluded that the loss of C-terminal CUB domains is more

4 tolerated than the loss of those related to vitamin B12 absorption.

6 Biallelic CUBN variants are enriched in patients with normal renal function in a chronic

7 proteinuria replication cohort

8 To investigate whether CUBN variants that lead to proteinuria are always associated with

9 normal renal function, we next assembled an additional cohort of 107 patients with chronic

10 subnephrotic proteinuria (chronic PU cohort; more details in Supplementary Text). Similar to

11 the abovedescribed patients, all patients had been referred to our reference center because

12 of a suspected diagnosis SRNS or AS. However, prior efforts using Sanger sequencing or

13 smaller NGS panels than the Renome panel had failed to identify the molecular cause. In this

14 study population, 39 had normal renal function (serum creatinine <110 µmol/L in men or 90

15 µmol/L in women or eGFR > 60), whereas 30 patients had ESRD (<10ml/min GFR,

16 transplanted or on hemodialysis) (Table 1; Supplemental Table 1; Figure 4A). Sequencing

17 was performed with a smaller custom-made NGS panel enriched for genes important for

18 proximal tubule function that included for the first time in these patients the gene CUBN. The

19 sequencing revealed that 10.3% of the patients in this cohort have homozygous or two

20 heterozygous filtered variants in CUBN, translating into a mutation rate of 28.2% in

21 individuals with chronic proteinuria and normal renal function and 0% in patients with chronic

22 proteinuria and reduced renal function (Table 1; Supplemental Table 1; Figure 4A).

23 The 12 patients from this cohort had a very similar phenotype to the abovedescribed

24 patients, including the age of onset, the type and range of proteinuria, the lack of severe

25 lesions in renal biopsies and the lack of proteinuria lowering effects with ACE inhibitors

26 (Table 1; Supplemental Table 1; Figure 4B,C). Most importantly, renal function was also

27 normal, even for the oldest patient with the age of 66 years (Table 1; Supplemental Table 1).

28 Figure 4D depicts a normal age-dependent eGFR decline for the CUBN patients, while the

1 CUBN-negative cases all show a more rapid decline. Of note, we also identified two patients

2 with hemizygous CLCN5 and OCRL1 variants, respectively, responsible for Dent’s disease 1

3 and 2 (Supplemental Table 6). However, in these patients serum creatinine levels were

4 elevated, suggesting a fundamental difference between Dent’s disease and cubilin

5 deficiency. Except for one single heterozygous variant localized close to vitamin B12/IF-

6 binding region (p.N1303H; Figure 3B), all 14 likely functional CUBN variants from this cohort

7 were in the C-terminal CUB11-27 (aa:1928-3618; Figure 2). Also here, structural models

8 showed that all the variants could have effects on the folding and function of the CUB

9 domains (Figure 3C, Supplemental Table 5). Altogether, these data confirm that biallelic C-

10 terminal CUBN variants are associated with chronic proteinuria and normal renal function.

12 Four C-terminal missense variants associate with albuminuria and increased glomerular

13 filtration rate in population-based studies

14 To more generally evaluate the importance of CUBN variants for kidney function, we turned

15 to large population studies. We focused on four C-terminal variants (p.A1690V, p.N2157D,

16 p.A2914V and p.I2984V) that previously had shown strong associations with albuminuria in

17 GWAS (20-22, 25, 26). The frequencies of these GWAS variants were higher than the 0.1%

18 threshold, which is why they were not included in the initial analysis [in-house genome

19 database: f(A1690V)=0.00109, f(N2157D)=0.00667, f(A2914V)=0.00945, f(I2984V)=0.0932;

20 gnomAD: f(A1690V)=0.00173, f(N2157D)=0.00565, f(A2914V)=0.0122, f(I2984V)=0.0875

21 (Supplemental Table 3)]. However, one of them (p.N2157D) was identified in the chronic PU

22 cohort in trans with a low-frequency variant passing our filtering criteria (p.S1947Y) in a child

23 with chronic proteinuria and normal renal function (Supplemental Table 7). As homozygous

24 p.S1947Y has previously been shown to cause proteinuria in a child of similar age (27),

25 p.N2157D seems to be a variant affecting cubilin function. Furthermore, according to the

26 structural modeling, all four GWAS variants have the potential to disturb CUB domain

27 stability or ligand binding (Figure 3C; Supplemental Table 5).

1 To test whether the four GWAS variants affect eGFR, we performed a large meta-analysis of

2 population-based cohorts from the CKDGen Consortium, comprising between 331,340 and

3 597,710 individuals (28). In all four cases, we found in addition to the association with

4 albuminuria a modest but significant association with higher eGFR for the minor compared to

5 the major allele (p(A1690V)=0.02432; p(N2157D)=0.00854; p(A2914V)=0.0008926;

6 p(I2984V)=0.0005845; Table 2). This was also confirmed in a smaller, independent cohort

7 comprising 13,550 individuals with and without type 2 diabetes (Supplemental Tables 8 and

8 9). While p.N2157D was only rarely found in this cohort, both eGFR and albuminuria was

9 significantly increased for the other three variants (Supplemental Table 9). While all

10 associations were tested under an additive model, for the most common variant, p.I2984V,

11 also the recessive model could be performed, confirming that homozygotes indeed show the

12 strongest association with albuminuria (Supplemental Tables 10-12). When stratified for

13 diabetes status, the eGFR was significantly increased for all three variants in diabetic people

14 (p(A1690V)=0.04; p(A2914V)=0.03; p(I2984V)=0.02), whereas in the non-diabetic people

15 this was only the case for p.A2914V (p(A1690V)=0.34; p(A2914V)=0.01; p(I2984V)=0.16 ;

16 Supplemental Table 9). Together, these data provide strong support for the benign nature of

17 the albuminuria associated with C-terminal CUBN variants.

1 Discussion

2 Altogether, the combined analysis of the three different cohorts with suspected glomerular

3 disease identified 39 patients with isolated proteinuria and normal renal function due to

4 biallelic filtered CUBN variants. However, despite the early onset the proteinuria does not

5 seem to be associated with an unfavorable prognosis for kidney disease in our patients. We

6 further show that the high percentage of urinary albumin in these patients is often

7 misinterpreted as glomerular injury in the clinical setting, justifying renal biopsy and ACE or

8 AT1 inhibition as proteinuria-lowering treatment. Yet, we show that the proteinuria does not

9 seem to be associated with an unfavorable prognosis for kidney disease in our patients.

10 Apart from establishing a diagnosis in individuals with isolated subnephrotic proteinuria, the

11 detection of CUBN variants may therefore avoid inefficient therapies aimed at reducing

12 glomerular proteinuria.

13 Assuming that cubilin mostly functions in the proximal tubules (14-16), the proteinuria is

14 explained by the reduced protein reabsorption on the luminal surface of proximal tubular

15 cells. The reabsorption of albumin is particularly affected, supporting mouse studies that

16 cubilin could be the main albumin receptor in the proximal tubules (10). By contrast, megalin

17 deficiency as in Donnai-Barrow syndrome typically leads to lower urinary albumin-to-protein

18 ratios (7, 8, 12, 29). Moreover, renal function has been reported to be reduced in several

19 families (29-31), which is also the case in diseases like Dent’s disease, in which both

20 megalin and cubilin are reduced (32). Accordingly, LRP2 variants have been found to be

21 associated with reduced GFR but not with albuminuria (33), altogether suggesting

22 fundamental differences for the contribution of megalin and cubilin to renal health.

23 Another unexpected finding is the clear genotype-phenotype correlation associated with

24 CUBN variants (11, 34). While all the IGS mutations can exclusively be found before or

25 within the vitamin B12/IF-binding region (CUB5-8), proteinuria without vitamin B12

26 malabsorption is caused by variants located after this region. In contrast to previous in vitro-

27 studies (35), this means that the renal ligands, most notably albumin, should bind to more C-

28 terminal CUB domains. In particular, CUB13 and 26 that possess proteinuria-causing

1 variants in or close to Ca2+-binding motifs (p.S1947Y and p.D3492Y, respectively) are strong

2 candidate domains for albumin binding. Although this has not yet been studied

3 systematically, it seems that missense variants that only affect vitamin B12/IF-binding, such

4 as the Finnish variant p.P1297L, are not associated with proteinuria (36). The reverse

5 conclusion is therefore that IGS patients that have proteinuria should have mutations that

6 either affect general expression or the interaction with amnionless or lead to cubilin

7 truncation. According to the literature, such patients do not seem to present with renal

8 insufficiency (13, 37-39), which is in agreement with the isolated proteinuria cases described

9 here. Functional studies have also shown that vitamin B12 uptake is maintained when the

10 receptor is truncated after CUB8 (40), consistent with the presence of a putative intestinal

11 transcript truncated directly after CUB8 in the Genotype-Tissue Expression (GTEx) database

12 (41). Combined with our finding that premature truncation of cubilin is more likely to happen

13 after CUB8 in the normal population, it can thus be concluded that in humans vitamin B12

14 malabsorption is less tolerated than albuminuria.

15 Our findings were supported and extended to the general population by performing a meta-

16 analysis of over half a million participants. By evaluating four CUBN variants, previously

17 identified due to their strong association with albuminuria (20-22, 25, 26), we found a

18 moderate but significant association with higher eGFR. While potential correlations of nearby

19 SNPs cannot be excluded for variants identified in GWAS, support for functional effects of

20 these variants comes from our structural modelling and from the observation that compound

21 heterozygosity with a filtered CUBN variant can also lead to the combination of chronic

22 proteinuria and normal renal function. As evidence for positive selection of an European

23 haplotype containing one of the GWAS variants (p.I2984V) has already been reported (42), it

24 can be speculated that there may even be some sort of evolutionary advantage in reducing

25 proximal tubular uptake (43), for example in conditions where the tubules are overloaded

26 with proteins and lipids. Support for this view comes from the clinical observation that tubular

27 damage is key for the progression of DKD and primary FSGS (44, 45), from mouse models

28 of proteinuria and hyperlipidemia (46, 47), in which the reduction of tubular uptake was

1 shown to prevent early injury, or from mice lacking albumin, that become protected against

2 AS (48). However, as our population-based studies were cross-sectional, it would be

3 interesting to study any protective effects of cubilin deficiency on the progression of specific

4 glomerular diseases in a longitudinal manner. Moreover, given that the observed effects on

5 eGFR are mild and that eGFR itself reflects creatinine clearance with a margin of error,

6 future studies could benefit from more specific tubular damage markers, e.g. urinary

7 epidermal growth factor (49).

8 In sum, our study proposes a new paradigm for the non-detrimental effects of tubular

9 proteinuria, which contrasts with the general dogma that proteinuria is always damaging.

10 Although adverse long-term effects associated with CUBN deficiency cannot fully be

11 excluded, we recommend genetic testing for CUBN variants in individuals with chronic

12 subnephrotic proteinuria to avoid unnecessary further medical actions. Analogous to the

13 familial renal glucosuria caused by SGLT2 mutations (50), benign Mendelian traits such as

14 the one presented here may have the potential to define safe drug targets, especially if

15 protective effects for a specific disease can be demonstrated for the genetic variants (51).

1 Methods

3 DNA extraction and preparation for next-generation sequencing

4 Genomic DNA was extracted from blood samples and washed in Amicon columns (Merck

5 Millipore). DNA quality was evaluated by agarose gel electrophoresis. The concentration of

6 DNA and presence of impurities were calculated using Xpose scanner (Luescher). For

7 Genetic kidney disease cohort I, 759 patients with suspected genetic renal disease were

8 sequenced by the “Renome panel” containing 309 known renal disease genes. Diseases

9 belonging to the proteinuric group are Alport syndrome (AS) and steroid-resistant nephrotic

10 syndrome (SRNS). The non-proteinuric group consists of renal tubular dysgenesis (RTD),

11 renal hypodysplasia (CAKUT), tubulointerstitial nephritis (TIN), nephronophthisis (NPHP) and

12 polycystic kidney disease (PKD). The control group consisted of 694 individuals with

13 matched ethnic backgrounds (527 Caucasians and 157 Maghrebians). All identified CUBN

14 variants were verified by Sanger sequencing. Whenever DNA of the parents was available,

15 segregation was confirmed in the compound heterozygous cases (Supplemental Table 2).

16 For the Chronic PU cohort, 107 selected individuals exhibiting isolated chronic proteinuria

17 (between 0.5 and 3 g/d) but no genetic diagnosis despite the previous testing of several

18 SRNS and/or AS genes were sequenced using a custom-made next-generation sequencing

19 panel targeting seventeen genes known to have functions in the proximal tubule

20 (SureSelectXT, Agilent Technologies, France). High-throughput sequencing was carried out

21 using a MiSeq/HiSeq platform (Illumina, San Diego, CA). The selected patients were

22 heterogeneous in ages and were recruited through adult and pediatric nephrology

23 departments from all over France. All the clinical information was provided and collected by

24 clinicians prescribing the genetic testing, including familial information and pedigree.

25 For Genetic kidney disease cohort II, DNA was extracted from human blood samples from

26 a total of 1350 patients with suspected SRNS and AS. All exons and adjacent intronic

27 boundaries of up to 324 glomerular genes (depending on the version of the customized multi-

28 gene panel) known or hypothesized to cause SRNS, FSGS, Alport syndrome and differential

1 diagnoses were targeted by a custom SeqCap EZ choice sequence capture library (“GLOM

2 panel”; NimbleGen, Madison, Wisconsin, USA) and subsequently sequenced on an Illumina

3 MiSeq or HiSeq platform (2x150 PE) according to the manufacturer´s protocol. DNA samples

4 were analysed with an average coverage of 120-fold (MiSeq) or more than 200-fold (HiSeq),

5 respectively. Bioinformatic analysis was performed using the SeqPilot SeqNext moduleTM

6 (v3.5.2, JSI medical systems, Kippenheim, Germany) as well as an in-house bioinformatic

7 pipeline. For all approaches, potential pathogenic variants were confirmed by Sanger

8 sequencing. For compound heterozygous individuals, segregation was confirmed by

9 sequencing the parents. All the clinical information was provided and collected by clinicians

10 prescribing the genetic testing.

12 Variant filtering

13 In order to evaluate the likelihood of pathogenicity for the identified variants, we used the

14 guidelines of the American College of Medical Genetics (52). As a general approach, we

15 combined a defined frequency cutoff, bioinformatic damage prediction and structural

16 modeling for the filtering of variants. For the filtering of splicing variants, only variants in

17 coding regions or essential splice sites were considered in our study. All missense variants

18 were predicted to be damaging with at least two out of three damage prediction algorithms:

19 Mutation Taster (http://www.mutationtaster.org/), PolyPhen-2

20 (http://genetics.bwh.harvard.edu/php2/) and SIFT (http://sift.jcvi.org/). Only one variant,

21 p.N1303H, was predicted to be damaging only by one algorithm. However, structural

22 modeling showed that this variant is located directly in the CUB8-vitamin B12/IF interface

23 (Supplemental Table 5). All variants were either absent from reference populations (e.g.

24 gnomAD (23)) or rare with global allele frequencies below 0.001. Only p.S1947Y has

25 previously been reported in the Human Gene Mutation Database (24). In CUBN-positive

26 cases, no additional gene variant with pathogenic relevance for the disease phenotype was

27 present among the patients described in this manuscript.

1 Structural analysis of variants

2 Structural models of individual cubilin CUB domains were generated from previously

3 determined structures using the Phyre2 server (53). Figures were prepared with PyMOL.

5 Calculation of the estimated glomerular filtration rate and proteinuria range

6 The eGFR was calculated with Schwartz formula for pediatric patients (age < 18 years old)

7 and using CKD-EPI formula for adults (54-56). Proteinuria was measured as total amount of

8 protein in 24-hour urine collections. If only spot urine was available, urinary protein-to-

9 creatinine (mostly in mg/mmol) was used as an estimate for the 24-hour urine (57).

10 11 Statistics

12 Continuous values are here reported as means with ± SD (standard deviation). For these

13 values, 95% confidence intervals were calculated using the appropriate Student’s-t

14 probabilities. Dichotomous data are showed as percentages. We applied �2 or Fisher exact

15 tests to dichotomous data in order to compare differences between two groups. For

16 continuous data comparison we used an unpaired t test for the Gaussian sampled data. Two-

17 tailed P values < 0.05 were regarded as statistically significant. Statistical analyses were

18 performed using Rstudio and Prism 4 (GraphPad) software.

20 Lookup of genetic variants in the CKDgen Consortium

21 Four low-frequency or common missense variants known to be associated with albuminuria

22 from GWAS of population-based cohorts (rs141640975, rs144360241, rs45551835, and

23 rs1801239) (20-22, 25, 26) were evaluated for association with albuminuria and eGFR using

24 summary statistics from a large-scale meta-analysis of mostly population-based studies

25 within the CKDGen Consortium (28, 58). Alleles, effect direction, standard error, p-value as

26 well as sample size were extracted from genome-wide results using an additive model.

27 Sample size varied between 331,340 and 597,710 across variants. We used fixed effects in

28 the meta-analysis as previously described (28).

1 Lookup of genetic variants in a European exome-wide association study (ExWAS) with

2 and without diabetes

3 GWAS variants were evaluated for association with albuminuria and eGFR using summary

4 statistics (alleles, effect, standard error, p value and sample size) from an ExWAS discovery

5 meta-analysis comprising 5 studies (3 population based and 2 type 2 diabetes studies) from

6 Denmark (20) under an additive model and for I2984V (rs1801239) also with the recessive

7 model. Similarly, association summary for the three variants for eGFR and albuminuria was

8 also reported, based on stratification for diabetes status. Total sample size varied between

9 13,124 to 13,550 (3837 to 3990 individuals with diabetes and 9251 to 9449 individuals

10 without diabetes) across variants. We used inverse variance fixed effects in the meta-

11 analysis as previously described (20).

12 13

1 Author contributions

2 M.B. analyzed and compiled patient data with help from O.B., A.S, MJ.T.,C.A., M.S.

3 M.B.,L.VG.,M.S.,O.G.,C.BF.,V.M.,C.T.,C.A.,M.G.,E.K.,T.W.,C.B. performed genotyping and

4 analyzed the sequencing results.

5 M.B., F.JH., P.N. performed statistical analysis.

6 Y.L.,T.S.V.,A.K. performed population-based studies.

7 A.C.,J.H.,V.B.,S.K.,A.B.,F.L.,F.L.,B.R.,C.V.,I.B.,C.IB.,C.J.M.,T.S.,L.P.,M.G.,T.B.H.,M.B.,G.K.,

8 M.H.,K.L.,O.B.,A.S. recruited patients and gathered the clinical data for the study.

9 C.F.B.A performed the structural modeling.

10 M.S. conceived and supervised the project.

11 M.S. wrote the paper with help from M.B. and C.A.

12 All authors critically reviewed the paper.

1 Funding and Acknowledgements

2 The work has been supported by the ATIP-Avenir program, the Fondation Bettencourt-

3 Schueller (Liliane Bettencourt Chair of Developmental Biology) as well as State funding by

4 the Agence Nationale de la Recherche (ANR) under the “Investissements d’avenir” program

5 (ANR-10-IAHU-01) and the NEPHROFLY (ANR-14-ACHN-0013) grant to MS. TSA and MS

6 also received funding from the Novo Nordisk Foundation, Steno Collaborative Grant 2018

7 (NNF18OC0052457). TSA acknowledges all the study participants and collaborations with

8 the Novo Nordisk Foundation Center for Basic Metabolic Research, Section of Metabolic

9 Genetics, University of Copenhagen, Copenhagen, Denmark, and the Center for Clinical

10 Research and Prevention, Bispebjerg and Frederiksberg Hospital, Capital region,

11 Copenhagen, Denmark where the genotyping and phenotyping for the Danish studies were

12 performed. The summary statistics from the CKDGen Consortium are available

13 at http://ckdgen.imbi.uni-freiburg.de. We thank the CKDGen Consortium for their

14 collaboration in providing early access to data. The work of AK was supported by a

15 Heisenberg Professorship of the German Research Foundation (KO 3598/5-1). CB received

16 support from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)

17 Collaborative Research Centre (SFB) KIDGEM 1140 (project no. 246781735) and the

18 Federal Ministry of Education and Research (BMBF, 01GM1515C).

24 25 26 27 28 29 30 19

1 References 2 3 1. Astor BC, Matsushita K, Gansevoort RT, van der Velde M, Woodward M, Levey AS, 4 Jong PE, Coresh J, Chronic Kidney Disease Prognosis C, Astor BC, et al. Lower 5 estimated glomerular filtration rate and higher albuminuria are associated with 6 mortality and end-stage renal disease. A collaborative meta-analysis of kidney 7 disease population cohorts. Kidney Int. 2011;79(12):1331-40. 8 2. Gansevoort RT, Matsushita K, van der Velde M, Astor BC, Woodward M, Levey AS, 9 de Jong PE, Coresh J, and Chronic Kidney Disease Prognosis C. Lower estimated 10 GFR and higher albuminuria are associated with adverse kidney outcomes. A 11 collaborative meta-analysis of general and high-risk population cohorts. Kidney Int. 12 2011;80(1):93-104. 13 3. Ninomiya T, Perkovic V, de Galan BE, Zoungas S, Pillai A, Jardine M, Patel A, Cass 14 A, Neal B, Poulter N, et al. Albuminuria and kidney function independently predict 15 cardiovascular and renal outcomes in diabetes. J Am Soc Nephrol. 2009;20(8):1813- 16 21. 17 4. Abbate M, Zoja C, and Remuzzi G. How does proteinuria cause progressive renal 18 damage? J Am Soc Nephrol. 2006;17(11):2974-84. 19 5. Moorhead JF, Chan MK, El-Nahas M, and Varghese Z. Lipid nephrotoxicity in chronic 20 progressive glomerular and tubulo-interstitial disease. Lancet. 1982;2(8311):1309-11. 21 6. Brenner BM, Cooper ME, de Zeeuw D, Keane WF, Mitch WE, Parving HH, Remuzzi 22 G, Snapinn SM, Zhang Z, Shahinfar S, et al. Effects of losartan on renal and 23 cardiovascular outcomes in patients with type 2 diabetes and nephropathy. N Engl J 24 Med. 2001;345(12):861-9. 25 7. Norden AG, Scheinman SJ, Deschodt-Lanckman MM, Lapsley M, Nortier JL, Thakker 26 RV, Unwin RJ, and Wrong O. Tubular proteinuria defined by a study of Dent's 27 (CLCN5 mutation) and other tubular diseases. Kidney Int. 2000;57(1):240-9. 28 8. Smith ER, Cai MM, McMahon LP, Wright DA, and Holt SG. The value of 29 simultaneous measurements of urinary albumin and total protein in proteinuric 30 patients. Nephrol Dial Transplant. 2012;27(4):1534-41. 31 9. Christensen EI, Devuyst O, Dom G, Nielsen R, Van der Smissen P, Verroust P, 32 Leruth M, Guggino WB, and Courtoy PJ. Loss of chloride channel ClC-5 impairs 33 endocytosis by defective trafficking of megalin and cubilin in kidney proximal tubules. 34 Proc Natl Acad Sci U S A. 2003;100(14):8472-7. 35 10. Amsellem S, Gburek J, Hamard G, Nielsen R, Willnow TE, Devuyst O, Nexo E, 36 Verroust PJ, Christensen EI, and Kozyraki R. Cubilin is essential for albumin 37 reabsorption in the renal proximal tubule. J Am Soc Nephrol. 2010;21(11):1859-67. 38 11. Christensen EI, Nielsen R, and Birn H. From bowel to kidneys: the role of cubilin in 39 physiology and disease. Nephrol Dial Transplant. 2013;28(2):274-81. 40 12. Dachy A, Paquot F, Debray G, Bovy C, Christensen EI, Collard L, and Jouret F. In- 41 depth phenotyping of a Donnai-Barrow patient helps clarify proximal tubule 42 dysfunction. Pediatr Nephrol. 2015;30(6):1027-31. 43 13. Wahlstedt-Froberg V, Pettersson T, Aminoff M, Dugue B, and Grasbeck R. 44 Proteinuria in cubilin-deficient patients with selective vitamin B12 malabsorption. 45 Pediatr Nephrol. 2003;18(5):417-21. 46 14. Qiu C, Huang S, Park J, Park Y, Ko YA, Seasock MJ, Bryer JS, Xu XX, Song WC, 47 Palmer M, et al. Renal compartment-specific genetic variation analyses identify new 48 pathways in chronic kidney disease. Nat Med. 2018;24(11):1721-31. 49 15. Menon R, Otto EA, Kokoruda A, Zhou J, Zhang Z, Yoon E, Chen YC, Troyanskaya O, 50 Spence JR, Kretzler M, et al. Single-cell analysis of progenitor cell dynamics and 51 lineage specification in the human fetal kidney. Development. 2018;145(16). 52 16. Park J, Shrestha R, Qiu C, Kondo A, Huang S, Werth M, Li M, Barasch J, and 53 Susztak K. Single-cell transcriptomics of the mouse kidney reveals potential cellular 54 targets of kidney disease. Science. 2018;360(6390):758-63.

1 17. Larsen C, Etzerodt A, Madsen M, Skjodt K, Moestrup SK, and Andersen CBF. 2 Structural assembly of the megadalton-sized receptor for intestinal vitamin B12 3 uptake and kidney protein reabsorption. Nat Commun. 2018;9(1):5204. 4 18. Andersen CB, Madsen M, Storm T, Moestrup SK, and Andersen GR. Structural basis 5 for receptor recognition of vitamin-B(12)-intrinsic factor complexes. Nature. 6 2010;464(7287):445-8. 7 19. Ovunc B, Otto EA, Vega-Warner V, Saisawat P, Ashraf S, Ramaswami G, Fathy HM, 8 Schoeb D, Chernin G, Lyons RH, et al. Exome sequencing reveals cubilin mutation 9 as a single-gene cause of proteinuria. J Am Soc Nephrol. 2011;22(10):1815-20. 10 20. Ahluwalia TS, Schulz CA, Waage J, Skaaby T, Sandholm N, van Zuydam N, Charmet 11 R, Bork-Jensen J, Almgren P, Thuesen BH, et al. A novel rare CUBN variant and 12 three additional genes identified in Europeans with and without diabetes: results from 13 an exome-wide association study of albuminuria. Diabetologia. 2019;62(2):292-305. 14 21. Boger CA, Chen MH, Tin A, Olden M, Kottgen A, de Boer IH, Fuchsberger C, 15 O'Seaghdha CM, Pattaro C, Teumer A, et al. CUBN is a gene locus for albuminuria. J 16 Am Soc Nephrol. 2011;22(3):555-70. 17 22. Haas ME, Aragam KG, Emdin CA, Bick AG, International Consortium for Blood P, 18 Hemani G, Davey Smith G, and Kathiresan S. Genetic Association of Albuminuria 19 with Cardiometabolic Disease and Blood Pressure. Am J Hum Genet. 20 2018;103(4):461-73. 21 23. Lek M, Karczewski KJ, Minikel EV, Samocha KE, Banks E, Fennell T, O'Donnell- 22 Luria AH, Ware JS, Hill AJ, Cummings BB, et al. Analysis of protein-coding genetic 23 variation in 60,706 humans. Nature. 2016;536(7616):285-91. 24 24. Stenson PD, Mort M, Ball EV, Evans K, Hayden M, Heywood S, Hussain M, Phillips 25 AD, and Cooper DN. The Human Gene Mutation Database: towards a 26 comprehensive repository of inherited mutation data for medical research, genetic 27 diagnosis and next-generation sequencing studies. Hum Genet. 2017;136(6):665-77. 28 25. Zanetti D, Rao A, Gustafsson S, Assimes TL, Montgomery SB, and Ingelsson E. 29 Identification of 22 novel loci associated with urinary biomarkers of albumin, sodium, 30 and potassium excretion. Kidney Int. 2019. 31 26. Teumer A, Tin A, Sorice R, Gorski M, Yeo NC, Chu AY, Li M, Li Y, Mijatovic V, Ko 32 YA, et al. Genome-wide Association Studies Identify Genetic Loci Associated With 33 Albuminuria in Diabetes. Diabetes. 2016;65(3):803-17. 34 27. Bullich G, Domingo-Gallego A, Vargas I, Ruiz P, Lorente-Grandoso L, Furlano M, 35 Fraga G, Madrid A, Ariceta G, Borregan M, et al. A kidney-disease gene panel allows 36 a comprehensive genetic diagnosis of cystic and glomerular inherited kidney 37 diseases. Kidney Int. 2018;94(2):363-71. 38 28. Wuttke M, Li Y, Li M, Sieber KB, Feitosa MF, Gorski M, Tin A, Wang L, Chu AY, 39 Hoppmann A, et al. A catalog of genetic loci associated with kidney function from 40 analyses of a million individuals. Nat Genet. 2019;51(6):957-72. 41 29. Storm T, Tranebjaerg L, Frykholm C, Birn H, Verroust PJ, Neveus T, Sundelin B, 42 Hertz JM, Holmstrom G, Ericson K, et al. Renal phenotypic investigations of megalin- 43 deficient patients: novel insights into tubular proteinuria and albumin filtration. Nephrol 44 Dial Transplant. 2013;28(3):585-91. 45 30. Kumar G, Chaudhry M., Faris K., Al Masri O. Renal involvement in a child with 46 Donnai-Barrow syndrome. Asian J Pediatr Nephrol. 2018;1(93-5. 47 31. Pober BR, Longoni M, and Noonan KM. A review of Donnai-Barrow and facio-oculo- 48 acoustico-renal (DB/FOAR) syndrome: clinical features and differential diagnosis. 49 Birth Defects Res A Clin Mol Teratol. 2009;85(1):76-81. 50 32. Blanchard A, Curis E, Guyon-Roger T, Kahila D, Treard C, Baudouin V, Berard E, 51 Champion G, Cochat P, Dubourg J, et al. Observations of a large Dent disease 52 cohort. Kidney Int. 2016;90(2):430-9. 53 33. Chasman DI, Fuchsberger C, Pattaro C, Teumer A, Boger CA, Endlich K, Olden M, 54 Chen MH, Tin A, Taliun D, et al. Integration of genome-wide association studies with 55 biological knowledge identifies six novel genes related to kidney function. Hum Mol 56 Genet. 2012;21(24):5329-43.

1 34. Kozyraki R, and Cases O. Vitamin B12 absorption: mammalian physiology and 2 acquired and inherited disorders. Biochimie. 2013;95(5):1002-7. 3 35. Yammani RR, Seetharam S, and Seetharam B. Identification and characterization of 4 two distinct ligand binding regions of cubilin. J Biol Chem. 2001;276(48):44777-84. 5 36. Kristiansen M, Aminoff M, Jacobsen C, de La Chapelle A, Krahe R, Verroust PJ, and 6 Moestrup SK. Cubilin P1297L mutation associated with hereditary megaloblastic 7 anemia 1 causes impaired recognition of intrinsic factor-vitamin B(12) by cubilin. 8 Blood. 2000;96(2):405-9. 9 37. Broch H, Imerslund O, Monn E, Hovig T, and Seip M. Imerslund-Grasbeck anemia. A 10 long-term follow-up study. Acta Paediatr Scand. 1984;73(2):248-53. 11 38. Ercan Z, Demir ME, Ulas T, Ingec M, and Horoz M. A long-term follow-up of an 12 Imerslund-Grasbeck syndrome patient with proteinuria. Nefrologia. 2013;33(1):147-8. 13 39. Grasbeck R. Imerslund-Grasbeck syndrome (selective vitamin B(12) malabsorption 14 with proteinuria). Orphanet J Rare Dis. 2006;1(17. 15 40. Pedersen GA, Chakraborty S, Steinhauser AL, Traub LM, and Madsen M. AMN 16 directs endocytosis of the intrinsic factor-vitamin B(12) receptor cubam by engaging 17 ARH or Dab2. Traffic. 2010;11(5):706-20. 18 41. Consortium GT. Human genomics. The Genotype-Tissue Expression (GTEx) pilot 19 analysis: multitissue gene regulation in humans. Science. 2015;348(6235):648-60. 20 42. Tzur S, Wasser WG, Rosset S, and Skorecki K. Linkage disequilibrium analysis 21 reveals an albuminuria risk haplotype containing three missense mutations in the 22 cubilin gene with striking differences among European and African ancestry 23 populations. BMC Nephrol. 2012;13(142. 24 43. Simons M. The Benefits of Tubular Proteinuria: An Evolutionary Perspective. J Am 25 Soc Nephrol. 2018. 26 44. Bonventre JV. Can we target tubular damage to prevent renal function decline in 27 diabetes? Semin Nephrol. 2012;32(5):452-62. 28 45. Alexopoulos E, Stangou M, Papagianni A, Pantzaki A, and Papadimitriou M. Factors 29 influencing the course and the response to treatment in primary focal segmental 30 glomerulosclerosis. Nephrol Dial Transplant. 2000;15(9):1348-56. 31 46. Motoyoshi Y, Matsusaka T, Saito A, Pastan I, Willnow TE, Mizutani S, and Ichikawa I. 32 Megalin contributes to the early injury of proximal tubule cells during nonselective 33 proteinuria. Kidney Int. 2008;74(10):1262-9. 34 47. Kuwahara S, Hosojima M, Kaneko R, Aoki H, Nakano D, Sasagawa T, Kabasawa H, 35 Kaseda R, Yasukawa R, Ishikawa T, et al. Megalin-Mediated Tubuloglomerular 36 Alterations in High-Fat Diet-Induced Kidney Disease. J Am Soc Nephrol. 37 2016;27(7):1996-2008. 38 48. Jarad G, Knutsen RH, Mecham RP, and Miner JH. Albumin contributes to kidney 39 disease progression in Alport syndrome. Am J Physiol Renal Physiol. 40 2016;311(1):F120-30. 41 49. Azukaitis K, Ju W, Kirchner M, Nair V, Smith M, Fang Z, Thurn-Valsassina D, Bayazit 42 A, Niemirska A, Canpolat N, et al. Low levels of urinary epidermal growth factor 43 predict chronic kidney disease progression in children. Kidney Int. 2019;96(1):214-21. 44 50. Santer R, and Calado J. Familial renal glucosuria and SGLT2: from a mendelian trait 45 to a therapeutic target. Clin J Am Soc Nephrol. 2010;5(1):133-41. 46 51. Dewey FE, Gusarova V, Dunbar RL, O'Dushlaine C, Schurmann C, Gottesman O, 47 McCarthy S, Van Hout CV, Bruse S, Dansky HM, et al. Genetic and Pharmacologic 48 Inactivation of ANGPTL3 and Cardiovascular Disease. N Engl J Med. 49 2017;377(3):211-21. 50 52. Richards S, Aziz N, Bale S, Bick D, Das S, Gastier-Foster J, Grody WW, Hegde M, 51 Lyon E, Spector E, et al. Standards and guidelines for the interpretation of sequence 52 variants: a joint consensus recommendation of the American College of Medical 53 Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 54 2015;17(5):405-24. 55 53. Kelley LA, Mezulis S, Yates CM, Wass MN, and Sternberg MJ. The Phyre2 web 56 portal for protein modeling, prediction and analysis. Nat Protoc. 2015;10(6):845-58.

1 54. Selistre L, Rabilloud M, Cochat P, de Souza V, Iwaz J, Lemoine S, Beyerle F, Poli- 2 de-Figueiredo CE, and Dubourg L. Comparison of the Schwartz and CKD-EPI 3 Equations for Estimating Glomerular Filtration Rate in Children, Adolescents, and 4 Adults: A Retrospective Cross-Sectional Study. PLoS Med. 2016;13(3):e1001979. 5 55. Schwartz GJ, and Work DF. Measurement and estimation of GFR in children and 6 adolescents. Clin J Am Soc Nephrol. 2009;4(11):1832-43. 7 56. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, 3rd, Feldman HI, Kusek 8 JW, Eggers P, Van Lente F, Greene T, et al. A new equation to estimate glomerular 9 filtration rate. Ann Intern Med. 2009;150(9):604-12. 10 57. National Kidney F. K/DOQI clinical practice guidelines for chronic kidney disease: 11 evaluation, classification, and stratification. Am J Kidney Dis. 2002;39(2 Suppl 1):S1- 12 266. 13 58. Teumer A, Li Y, Ghasemi S, Prins BP, Wuttke M, Hermle T, Giri A, Sieber KB, Qiu C, 14 Kirsten H, et al. Genome-wide association meta-analyses and fine-mapping elucidate 15 pathways influencing albuminuria. Nat Commun. 2019;10(1):4130. 16

23 Table 1. Clinical and biological data of patients with or without biallelic pathogenic CUBN variants Genetic kidney Genetic kidney disease disease cohort I cohort II Chronic PU cohort Patients with Patients with Patients without Patients with biallelic filtered Clinical and biological data biallelic filtered All patients biallelic filtered biallelic filtered P value CUBN variants CUBN variants CUBN variants CUBN variants (95% IC or %) (95% IC or %) (95% IC or %) (95% IC or %) (95% IC or %)

N 14 13 107 12 95

10.87 ± 11.99 6.37 ± 5.12 24.31 ± 16.64 7.85 ± 3.11 26.37 ± 16.50 (a) Age at first manifestation (yr) (0–36.59) (0–17.43) (0–57.33) (1.01–14.69) (0–60.28) 0.004 [n=14] [n=13] [n=99] [n=11] [n=88]

0.69 ± 0.45 0.76 ± 0.68 1.55 ± 1.37 1.00 ± 0.61 1.68 ± 1.47 (a) Proteinuria (g/d)* (0–1.65) (0–2.26) (0–4.29) (0–2.35) (0–20.32) NS (0.1405) [n=14] [n=11] [n=57] [n=11] [n=46]

63.06 ± 8.99 61.29 ± 14.00 60.79 ± 12.06 61.58 ± 18.00 Urinary albumin/protein ratio 64.16 ± 11.76 (37.55–90.76) (a) (43.63–82.49) (19.73–100.00) (29.8–91.78) (15.22–100.00) NS (0.9320) (%) [n=13] [n=9] [n=14] [n=5] [n=9]

0 1 (25) 53 (75.71) 0 53 (82.81) (b) Biopsies showing FSGS 0.0001 [n=9] [n=4] [n=70] [n=6] [n=64]

Biopsies with no, unspecifc or 9 (100) 3 (75) 17 (24.29) 6 (100) 11 (17.2) (b) 0.0001 minimal lesions [n=9] [n=4] [n=70] [n=6] [n=64]

19.87 ± 15.94 10.48 ± 6.47 36.76 ± 19.84 26.0 ± 20.27 38.13 ± 19.49 (a) Age at last follow-up (yr) (0–54.06) (0–24.45) (0–76.18) (0–71.17) (0–77.58) NS (0.0682) [n=14] [n=13] [n=89] [n=10] [n=79]

114.26 ± 18.11 128.07 ± 20.06 64.66 ± 50.11 105.64 ± 14.63 59.47 ± 50.65 eGFR at last follow-up (a) (75.41–153.10) (83.91–172.23) (0–164.22) (73.04–138.25) (0–160.82) 0.0054 (mL/min/1,73m2) [n=14] [n=11] [n=89] [n=10] [n=79]

0 0 30 (29.13) 0 30 (32.97) (b) ESRD 0.0168 [n=14] [n=13] [n=103] [n=12] [n=91]

41.90 ± 19.79 41.9 ± 19.79 (a) Age of ESRD (yr) NA NA (1.49–82.31) NA (1.94–81.86) NA [n=30] [n=30]

(a) Two-sample t-test for significance between CUBN mutated and non -mutated patients from the chronic PU cohort (b) Fisher exact test for significance between CUBN mutated and non-mutated patients from the chronic PU cohort [n] indicates the total number of patients with the given information; NA, not available; yr, years; FSGS, focal segmental glomerulosclerosis; ACE, Angiotensin-conversion enzyme; ESRD, end-stage renal disease. * Proteinuria in g/24h were measured in most patients. For children and adults where only urinary protein to creatinine ratios in spot urine were reported, conversion into g/24h was performed (56). Patients not included here are patients where only albuminuria or protein dipstick was available or patients without further information on the amount of proteinuria.

Table 2. Albuminuria and eGFR meta-analysis of CKDGen cohorts for CUBN variants Variant Protein f Albuminuria eGFR RS ID Effect on SE P N Effect on SE P N log(UACR) log(eGFR)

rs14164097 A1690V 0,004 0,4258 0,020 3,52E 48159 0,0105 0,004 0,02432 56969 5 1 7 -94 8 6 6 rs14436024 N2157 0,008 0,1152 0,015 3,92E 46384 0,008 0,003 0,00854 33134 1 D 3 9 -13 2 1 0 rs45551835 A2914V 0,015 0,2008 0,008 2,77E 54547 0,0049 0,001 0,000892 34502 2 4 -126 9 5 6 8 rs1801239 I2984V 0,105 0.0615 0.003 4,77E 55767 0,0019 0,000 0,000584 59771 6 2 -81 2 6 5 0 Effect on estimated glomerular filtration (eGFR) in the CKDgen cohort of the four GWAS CUBN variants A1690V, N2157D, A2914V and I2984V. f is the frequency; Effect on log(UACR) and on log(eGFR) is the effect size or “beta” (positive value means positive directionality and therefore an increased value of albuminuria and eGFR); SE is the standard error; P is the P-value; and N is the sample size. The unit for UACR is mg/g and for eGFR is ml min-1 per 1.73 m2.

Figure 1. Flow chart for cohort genotyping. The three different cohorts consisted of individuals with suspected genetic diseases (Alport syndrome (AS), steroid-resistant nephrotic syndrome (SRNS), nephronophthisis (NPHP), congenital anomalies of the kidney and urinary tract (CAKUT), tubulointerstitial nephritis (TIN), polycystic kidney disease (PKD) and renal tubular dysgenesis (RTD)). Altogether, 39 individuals with biallelic filtered CUBN variants (biallelic) were identified. In the chronic PU cohort, 11 individuals with biallelic filtered CUBN variants were identified plus 1 additional individual carrying one biallelic filtered CUBN variant and the GWAS variant p.N2157D (see Supplemental Table 7).

Figure 2. Frequency, density and position of CUBN variants along the cubilin protein. Cubilin protein structure summarized with the 8 EGF-like (lighter blue) and 27 CUB domains (brighter blue); the red dots correspond to the theoretical Ca2+-binding sites; arrows on top indicate the variants found in the three cohorts of this study; the green arrows represent the filtered CUBN variants, and the red arrows indicate the four GWAS missense variants A1690V, N2157D, A2914V and I2984V. Gaussian kernel density curves (with 0.2 width parameter) as well as the position of HGMD variants (red), the variants from this study (green), PTVs from the in-house (yellow) and gnomAD (blue) are shown as curves as vertical lines, respectively. The minor allele frequencies (MAF) of 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 b-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 cubilin missense plus the four GWAS variants (in red) in in silico structural models of domains CUB 11 to 27. Red spheres represent Ca2+ ions predicted to bind domains CUB 11, 13 and 26. Variants S1947Y and D3492Y are notably close to these Ca2+-binding motifs. (D) Example of one CUB domain with b-sheets (b2-10) and loops (L2-9).

Figure 4. Clinical and biological profiles of patients carrying biallelic filtered CUBN variants. (A) Proportions of biallelic filtered CUBN variants carriers in the Chronic proteinuria (PU) cohort (N=107), when considering only patients with normal renal function (eGFR > 60 ml/min per 1.73m2, N=39) or only the patients in end-stage renal disease (ESRD; eGFR < 15 ml/min per 1.73m2, N=30). (B-D) Patients with biallelic filtered CUBN variants from the three cohorts (genetic kidney disease cohorts I and II, chronic PU cohort) are merged as “Patients with biallelic filtered CUBN variants” group. Urinary albumin to protein ratio (UAPR) is plotted in 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%, while tubular proteinuria is below 50 % (B) (7, 43). Age of first manifestation of proteinuria for patients with (n=38) and without (n=88) biallelic filtered CUBN. Data represent mean values ± 95% confidence intervals (C). t test: *p < 0.01; *p < 0.001; ***p < 0,0001; NS, not significant. Renal function of patients with biallelic filtered CUBN variants (n=35) is declining according to normal age decline, whereas in patients without biallelic filtered CUBN variants (n=79) the decline is more rapid (D). Renal function (eGFR, estimated glomerular filtration rate) was calculated using Schwartz formula in children and CKD-EPI formula for adults. The blue and red dotted lines represent the logarithmic trend curves for both groups.