Common Variants in Mendelian Kidney Disease Genes and Their Association with Renal Function

Abstract

Many common genetic variants identified by genome-wide association studies for complex traits map to genes previously linked to rare inherited Mendelian disorders. A systematic analysis of common single-nucleotide polymorphisms (SNPs) in genes responsible for Mendelian diseases with kidney phenotypes has not been performed. We thus developed a comprehensive database of genes for Mendelian kidney conditions and evaluated the association between common genetic variants within these genes and kidney function in the general population. Using the Online Mendelian Inheritance in Man database, we identified 731 unique disease entries related to specific renal search terms and confirmed a kidney phenotype in 218 of these entries, corresponding to mutations in 258 genes. We interrogated common SNPs (minor allele frequency >5%) within these genes for association with the estimated GFR in 74,354 European-ancestry participants from the CKDGen Consortium. However, the top four candidate SNPs (rs6433115 at LRP2, rs1050700 at TSC1, rs249942 at PALB2, and rs9827843 at ROBO2) did not achieve significance in a stage 2 meta-analysis performed in 56,246 additional independent individuals, indicating that these common SNPs are not associated with estimated GFR. The effect of less common or rare variants in these genes on kidney function in the general population and disease-specific cohorts requires further research.

CKD affects approximately 10% of the general population in industrialized nations, and is significantly associated with cardiovascular morbidity and mortality.^1–4 Traditional risk factors for CKD, including diabetes and hypertension, fail to fully explain the increased risk of CKD,^5–9 suggesting other factors including a genetic component. Family studies indicate familial aggregation of CKD and ESRD risk.¹⁰ For example, family studies have shown that genetic factors account for 36%–75% of the variability in kidney function, with similar estimates for disease susceptibility and CKD progression.^10–14 Therefore, unraveling the genetic underpinnings of CKD bears the potential of discovering novel disease mechanisms as a basis for research into much needed therapeutic targets and strategies.

Genome-wide association studies (GWAS) recently identified several genomic loci associated with kidney traits.^15–21 The strongest of these associations is at the UMOD locus,^{15,17,18,20,22,23} a gene in which rare variants are known to cause autosomal-dominant kidney diseases with high risk for ESRD: MCKD2 (Online Mendelian Inheritance in Man [OMIM] database #603860), HNFJ1 (OMIM #162000), or GCKD (OMIM #609886). In addition, other kidney disease genes in which mutations follow Mendelian inheritance patterns were uncovered in GWAS of kidney function (SLC7A9, SLC34A1)¹⁷ and albuminuria (CUBN)²¹ in the general population. Similar examples exist for traits such as hypertension and dyslipidemia, in which common variants in genes causing inherited Mendelian diseases are identified in population-based GWAS.^24,25

These findings lead us to hypothesize that additional common variants in monogenic kidney disease genes^26,27 are associated with kidney function in the general population but have not yet been identified by GWAS efforts due to power limitations related to multiple testing for nearly 2.5 million SNPs. Thus, we aimed to (1) create a comprehensive, curated database of monogenic kidney disease genes, (2) analyze the association of common genetic variants in these candidate genes with serum creatinine-based estimated GFR (eGFR) and secondarily with CKD in the general population, and (3) examine SNPs beneath the genome-wide threshold that would have been overlooked in prior GWAS of eGFR and CKD.^15,17–19

Results

Query of Online OMIM

The OMIM database query yielded 731 entries after excluding entries of kidney expressed genes without kidney phenotype: 77 for the term glomerular, 114 for renal tubular, 147 for renal hypoplasia, 128 for renal dysplasia, 84 for renal agenesis, 89 for ESRD, and 92 for proteinuria (Figure 1). Of these, we excluded a total of 513 entries that were not unique, described genes causing renal malignancy, were without an identified gene or were with a nonautosomal gene, or if we could not confirm kidney anomaly or dysfunction on our manual search, leaving a total of 218 OMIM-based disease entries corresponding to a total of 258 unique genes (some syndromes had more than one associated gene). We assigned each OMIM entry to one of three broad categories that corresponded best to their underlying pathology: (1) glomerular/developmental (n=128), (2) tubular function (n=45), or (3) secondary kidney disease (n=45) (see Supplemental Tables 1–3 or Fox²⁸).

Figure 1.

Flow chart showing the process of the three-step OMIM database search.

Interrogation of CKDGen Stage 1 Data Set

We interrogated SNPs with a minor allele frequency (MAF) >5% within these genes in the stage 1 meta-analysis eGFR and CKD data sets of the CKDGen Consortium encompassing 74,354 individuals from 26 cohorts (Supplemental Table 4). A complete listing of all tested SNPs and results is also available online (see Pattaro et al.¹⁵ and Fox²⁸). Supplemental Table 5 shows the eGFR stage 1 meta-analysis results for the index SNPs of the 258 genes identified in the three-step OMIM search. Supplemental Tables 6 and 7 show P values for the association of these SNPs with CKD and serum cystatin C–based eGFR (eGFRcys), respectively. Of the 49 loci showing a significant association with eGFR and/or CKD using a gene-specific Bonferroni correction (Table 1), 8 independent SNPs had a P value <10⁻⁴ for the association with eGFR. Of these, four SNPs, rs12922822 in UMOD (P=1.20×10⁻²⁵) and rs12460876 in SLC7A9 (P=9.50×10⁻¹²), were previously identified in GWAS meta-analyses of eGFR or CKD,^17–19 and rs4073745 in NSD1 and rs894250 in SCARB2 showed linkage disequilibrium (LD; D’>0.2) with the GFR-associated SNPs at the UMOD and SHROOM3 loci. These SNPs were not further considered. The rs11789185 SNP in the ENG gene did not show a direction consistent association with cystatin C and was dropped from our replication pool. Thus, this resulted in a total of three eGFR-associated SNPs (rs6433115 in LRP2, rs9827843 in ROBO2, and rs1050700 in TSC1) and one CKD-associated SNP (rs249942 in PALB2) that moved forward to stage 2 meta-analysis (Table 2). Of these four SNPs, only rs6433115 at LRP2 met the more stringent experiment-wide significance criteria after adjustment for multiple correlated association tests in meta-analysis using the Conneely and Boehnke method (stage 1 meta-analysis P=3.49×10⁻⁶, PACT_{experiment-wide} = 0.05 [P value adjusted for correlated tests for each gene with additional adjustment for the number of genes tested in the experiment]) (Table 2).^29,30

Table 1.

SNPs significantly associated with eGFR or CKD in stage 1 meta analysis in the CKDGen Consortium

Gene Name	Syndrome	SNP ID	Coded Allele Frequency	Trait	P	Phenotype
ACTN4	FSGS 1, FSGS1	rs755690	0.44	eGFR	0.004	Nephrotic syndrome, ESRD
AGXT	Primary hyperoxaluria type I	rs4538195	0.06	CKD	0.01	Calcium oxalate accumulation and renal failure
AQP2	Nephrogenic diabetes insipidus, autosomal	rs296736	0.52	eGFR	0.004	Diabetes insipidus
BBS1	Bardet-Biedl syndrome (BBS)	rs1671062	0.40	eGFR	0.001	Developmental abnormalities with common renal failure
BMP4	Syndromic microphthalmia 6 (MCOPS6)	rs11623717	0.58	eGFR	0.001	Hypoplastic kidneys, renal failure possible
CA2	Autosomal recessive osteopetrosis 3, (OPTB3)	rs3758078	0.36	eGFR, CKD	0.01, 0.04	Isozymes of carbonic anhydrase, associated with RTA
CACNA1S	Hypokalemic periodic paralysis (HOKPP)	rs3850625	0.12	eGFR	0.0004	Renal phosphate wasting, associated with hypokalemia and episodic weakness
CASR	Familial hypocalciuric hypercalcemia type I, HHC1 acquired hypocalciuric hypercalcemia	rs7638770	0.25	eGFR, CKD	0.001, 0.0003	PTH-independent renal tubular defect in calcium reabsorption, associated with hypercalcemia and hypermagnesemia
CFI	Complement factor I deficiency GN with isolated C3 deposits and factor I deficiency	rs6815517	0.72	CKD	0.01	Deficiency of the C3 inactivator associated with GN and renal failure
ENG	Hereditary hemorrhagic telangiectasia Rendu-Osler-Weber (HHT)	rs11789185	0.10	eGFR	0.001	Vascular dysplasia, associated with rare hematuria
ERCC8	Cockayne syndrome type A (CSA)	rs158938	0.65	eGFR	0.002	Immune complex-mediated GN
FGF23	Hyperphosphatemic familial tumoral calcinosis (HFTC)	rs720333	0.85	eGFR, CKD	0.003, 0.003	Increase renal absorption of phosphate, associated with deposition of basic calcium phosphate crystals
G6PC	Glycogen storage disease I	rs2593595	0.82	CKD	0.01	Glycogen storage with renal failure
GIF	Intrinsic factor deficiency (IFD)	rs558660	0.18	eGFR	0.01	Cobalamin transport, associated with tubular proteinuria
GSS	Glutathione synthetase deficiency	rs2236270	0.39	eGFR	0.0002	Urinary excretion of 5-oxoproline, associated with metabolic acidosis, RTA
HOXD13	VACTERL association	rs847148	0.69	eGFR	0.001	Dysplasia, hydronephrosis with likely failure
ICK	Endocrine-cerebroosteodysplasia (ECO)	rs316144	0.43	eGFR	0.001	Cystically dilated renal tubules
INF2	FSGS 5 (FSGS5)	rs7140154	0.18	eGFR	0.01	FSGS, may lead to ESRD
JAG1	Alagille syndrome 1 (ALGS1)	rs6040050	0.29	eGFR	0.0004	Renal dysplasia, renal mesangiolipidosis, medullary cystic disease
KCNJ1	Bartter syndrome, antenatal type 2	rs2238009	0.14	eGFR	0.01	Potassium channel, associated with salt wasting, hypokalemic alkalosis, hypercalciuria, low BP, and nephrocalcinosis Potassium channel, associated with salt wasting, hypokalemic alkalosis, hypercalciuria, low BP, and nephrocalcinosis
KRAS	Cardiofaciocutaneous syndrome	rs7960917	0.78	eGFR	0.001	Prenatal hydronephrosis, reflux
LAMB2	Pierson syndrome	rs9880088	0.10	eGFR, CKD	0.0002, 0.02	Congenital nephrotic syndrome; early onset ESRD
LCAT	Lecithin: cholesterol acyltransferase deficiency (LCAT)	rs2292318	0.13	eGFR, CKD	0.01, 0.01	Proteinuria and renal failure
LRP2	Donnai-Barrow syndrome	rs6433115	0.79	eGFR	3.5×10−6	Proteinuria
MKKS	McKusick-Kaufman syndrome (MKKS)	rs6032878	0.89	eGFR	0.01	Reproductive system developmental abnormalities, with possible renal failure
MMACHC	Methylmalonic aciduria and homocystinuria, type cblC	rs12029322	0.22	eGFR	0.004	Thrombotic microangiopathic nephropathy, hematuria, proteinuria, and renal failure
NEU1	Neuramidase deficiency sialidosis type I	rs13118	0.93	eGFR	0.003	Lysosomal storage, sialyloligosacchariduria
NSD1	Sotos syndrome	rs4073745	0.69	eGFR, CKD	3.7×10−6, 0.002	Rare vesicoureteric reflux
PALB2	Fanconi anemia	rs249942	0.10	eGFR, CKD	0.002, 0.0002	CAKUT, renal malformations
PKD2	Autosomal dominant polycystic kidney disease (ADPKD)	rs2728111	0.77	eGFR, CKD	0.01, 0.01	Renal cysts, ESRD
PLCE1	Early onset nephrotic syndrome (NPHS3)	rs12258052	0.31	eGFR	0.001	FSGS, nephrotic proteinuria, and ESRD
PROKR2	Kallmann syndrome 3 (KAL3)	rs6053283	0.25	eGFR	0.01	Reproductive dysfunction, angiogenesis, renal agenesis
PSTPIP1	Pyogenic sterile arthritis, pyoderma gangrenosum, and acne	rs3936040	0.17	eGFR	0.01	Proteinuria, possibly immune mediated
PVRL1	Cleft palate ectodermal dysplasia syndrome (CLPED1); orofacial cleft 7 (OFC7)	rs7122134	0.60	eGFR	0.01	Renal dysplasia
RECQL4	Baller-Gerold syndrome (BGS)	rs10111332	0.52	eGFR	0.003	Renal dysplasia
RET	Ret protooncogene	rs1864405	0.25	eGFR	0.002	Renal agenesis, vesicoureteric reflux
ROBO2	Vesicoureteral reflux 2 (VUR2)	rs9827843	0.58	eGFR	0.001	CAKUT, reflux nephropathy
SCARB2	Action myoclonus-renal failure syndrome (AMRF)	rs894250	0.50	eGFR	0.001	FSGS, nephrotic syndrome, renal failure
SCNN1B	Liddle syndrome	rs2106374	0.81	eGFR	0.001	Hypertension, renal failure rare
SLC7A9	Cystinuria type A	rs12460876	0.61	eGFR, CKD	9.5×10−12, 0.004	Cystine precipitation, obstruction, and renal failure
SOX17	Vesicoureteral reflux 1 (VUR1)	rs16920355	0.16	CKD	0.004	CAKUT and ESRD
STRA6	Syndromic microphthalmia 9 (MCOPS9)	rs974456	0.23	eGFR	0.002	Horseshoe kidney, hydronephrosis, hypoplasia, can lead to ESRD
SUCLA2	Autosomal recessive mitochondrial DNA depletion syndrome, encephalopathic form with methylmalonic aciduria	rs6561429	0.91	CKD	0.003	Possibly mitochondrial DNA depletion, rare renal tubular dysfunction
TBX1	Di George syndrome (DGS)	rs4819843	0.19	CKD	0.01	Renal hypoplasia, renal insufficiency
TSC1	Tuberous sclerosis 1 (TSC1)	rs1050700	0.71	eGFR	0.00004	Renal cysts and tumors, without kidney failure
TTC8	Bardet-Biedl syndrome (BBS)	rs17700521	0.73	eGFR, CKD	0.001, 0.01	Developmental abnormalities with common renal failure
UMOD	Medullary cystic kidney disease 2 (MCKD2)	rs12922822	0.18	eGFR, CKD	1.2×10−25, 5.5×10−17	Medullary cysts with gouty arthritis
USF2	Upstream stimulatory factor 2	rs1882694	0.65	eGFR	0.002	Cystic renal dysplasia and hypodysplasia
WDPCP	Bardet-Biedl syndrome (BBS 1–15)	rs1850983	0.97	eGFR, CKD	0.001, 0.002	Developmental abnormalities with common renal failure

Results are from meta-analysis of 74,354 participants. Gene-specific significance thresholds were defined according to the Bonferroni method, as 0.05/ number of independent LD blocks within each gene. If a locus was significantly associated with both eGFR (<60 ml/min per 1.73 m²) and CKD, both P values are provided, respectively. RTA, renal tubular acidosis; PTH, parathyroid hormone; CAKUT, congenital anomalies of the kidney and urinary tract.

Table 2.

Results of stage 1 and stage 2 association meta-analysis

Trait	SNP ID	Locus	Effect Allele	Stage 1 Discovery Meta-Analysis			Stage 2 Replication Meta-Analysis		Stage 1 and 2 Combined Meta-Analysis
				β	P	PACT experiment-wide	β	P	β	P
eGFR	rs6433115	LRP2	T	−0.007	3.5×10−6	0.05	0.0005	0.42a	0.003	0.02a
eGFR	rs1050700	TSC1	T	−0.006	4.3×10−5	0.06	−0.002	0.14	−0.004	3×10−4
CKD	rs249942	PALB2	A	0.15	2.1×10−4	0.10	−0.007	0.66	0.017	0.30
eGFR	rs9827843	ROBO2	C	0.005	4.8×10−5	0.10	0.0005	0.36	0.003	1.4×10−3

Results are from stage 1 meta-analysis of 74,354 participants and stage 2 meta-analysis of 56,246 participants using a fixed-effects model except where indicated otherwise. Stage 1 and stage 1 and 2 combined meta-analysis P values were two sided and one-sided for stage 2 meta-analysis. GFR was estimated by serum creatinine.

^aEffect estimates and P values from a random-effects model due to significant heterogeneity in stage 2 meta-analysis.

Stage 2 Meta-Analyses

Stage 2 meta-analysis of the stage 1 meta-analysis significant locus (LRP2) and our three other top loci was performed in a total of 19 independent cohorts totaling 56,246 individuals (Supplemental Table 4). In stage 2 meta-analysis, rs6433115 in LRP2 showed evidence of heterogeneity (I²=54.6%, P=0.002). We thus tested for stage 2 meta-analysis significance using a random-effects model that is more robust in the setting of between-study heterogeneity, under which rs6433115 was not significantly associated with eGFR (one-sided P=0.42). Using a distinct pathway-based approach, a different SNP in LRP2, rs10490130, was recently found to be associated with eGFR.³¹ However, this locus was not in strong LD with the SNP identified in this project (rs6433115; R²=0.11, D’=0.75). The other three SNPs analyzed in stage 2 were also not significantly associated with eGFR (Table 2).

Discussion

We have curated a comprehensive list of Mendelian genes that were previously linked to kidney pathology, followed by the exploration of common SNPs in these genes for association with eGFR or CKD in the CKDGen meta-analysis of population-based genetic association studies. Overall, our results show that common genetic variants are not typically associated with eGFR in Mendelian kidney disease genes.

The SNP associations we present for our stage 1 meta-analysis are obtained from the results of the same GWAS meta-analysis data sets as those used in our previously published GWAS.¹⁵ The intent of our study was to focus systematically on SNPs in the curated list of candidate Mendelian genes, most of which were not previously reported in our prior work.¹⁵ Indeed, it is well established that current GWAS approaches miss many true associations due to limited power and the high penalty to be paid for an unbiased approach. We thus identified 45 additional SNPs showing a nominally significant association with eGFR or CKD in the stage 1 meta-analysis that did not meet our criteria for being considered for stage 2 meta-analysis. These SNPs and the curated list of Mendelian kidney disease genes and their measures of association with eGFR and CKD may represent a valuable resource for further research (Supplemental Tables 1–3, 5–7, and Fox²⁸).

Our findings suggest that targeting common variants within Mendelian kidney disease genes for associations with more subtle phenotype variation such as cross-sectional measures of eGFR in the general population may not easily identify new gene loci that have not already been identified by current large-scale GWAS.^15,17 However, we feel that further exploration of these candidate genes should be extended beyond this first step. Examples include conducting future analysis of less common variants (e.g., with MAF 1%–5%) in these gene regions using SNP data sets imputed to the 1000 Genomes reference panel,³² or using targeted sequencing or whole exome chip data as well as further replication efforts in disease-specific and prospective cohorts.

The strengths of this study include the manual curation of a kidney gene database followed by a systematic search, identification of potential candidate loci, and the large sample size used for variant discovery. Our study has some limitations. First, because we analyzed population-based cohorts, our findings are not generalizable to cohorts enriched for kidney disease. Second, despite significant effort to curate a comprehensive and extensive list of Mendelian genes affecting renal function, our query may not capture all discovered genes; moreover, not all of the genes are completely confirmed as causative for the noted phenotypes. Third, of all of the numerous loci with a gene-based significant association with kidney function in stage 1, we only followed up the four most significantly associated SNPs. Thus, we cannot exclude that true positive associations are among the SNPs not further pursued. Finally, because this was a cross-sectional study, we were not able to explore the association of our candidate SNPs with kidney disease progression.

In summary, we have manually curated the largest published list of Mendelian kidney disease genes. In these genes, we did not identify novel common gene variants that are robustly associated with renal function in the general population. Further research is needed to determine the role of less common variants in these genes in disease and population-based studies.

Concise Methods

Query of OMIM

Using the unspecific search term kidney yielded 3951 entries in an OMIM query (http://www.ncbi.nlm.nih.gov/omim) (Figure 1). To focus our search on inherited kidney disease leading to kidney malformations, dysfunction, and ESRD, we utilized the following three-step search procedure: (1) recording of all listings using the search terms glomerular, renal tubular, renal hypoplasia, renal dysplasia, renal agenesis, ESRD, and proteinuria, each entered separately; (2) exclusion of duplicate entries, genes causing renal malignancy, genes expressed in kidney without a clear kidney phenotype, nonautosomal genes, and entries without clearly identified genes; and (3) manual OMIM and PubMed searches of the remaining loci to confirm an identifiable kidney phenotype (defined as documented kidney anomalies and any evidence of glomerular or tubular dysfunction). These conditions were then classified into three categories: (1) developmental and/or glomerular origin, (2) tubular function related, and (3) secondary kidney disease (e.g., amyloid deposition).

Stage 1 Analysis—Interrogation of the CKDGen GWAS Meta-Analysis Data Set

For each gene in the curated gene list, we examined the stage 1 meta-analysis results for the association between eGFR and all common SNPs in the gene region. We restricted our analyses to SNPs with a MAF >5% due to power considerations and imputation quality consideration for SNPs with low MAF. We defined each gene region as 10 kb upstream and 5 kb downstream of the known transcription start and end sites of the gene, using build 36 of the reference genome as the reference.

Each of the participating cohorts of the CKDGen Consortium (Supplemental Table 1) performed GWAS of eGFR and CKD or eGFRcys using linear and logistic regression for continuous and dichotomous traits, respectively. An additive genetic effect model for the genotype dosages was used, adjusting for age and sex in all studies and for the study site and relatedness where applicable. Study-specific genotyping and HapMap-based imputation platforms were previously described.¹⁵ Study-specific GWAS data were subjected to inverse-variance weighted fixed-effects meta-analysis using METAL software (http://www.sph.umich.edu/csg/abecasis/Metal/).³² Genomic control was applied as previously described.¹⁵

Within each gene, the SNP with the lowest eGFR-associated P value was chosen as the index SNP representing that gene. Once these index SNPs were identified, we secondarily tested for the association between these SNPs and CKD (eGFR <60 ml/min per 1.73 m²) from the CKDGen meta-analysis.

Gene-specific significance thresholds were defined, according to the Bonferroni method, as 0.05/ (the number of independent tests within the gene). The denominator is given by the number of independent LD blocks, whereby independent blocks were identified by grouping SNPs in LD (r²>0.20, based on HapMap Phase 2 release 21) together. To reduce the likelihood of a false positive result, we used the Conneely and Boehnke method for adjustment of multiple correlated association tests in meta-analysis.^28,29 This is an alternative method to other multiple testing adjustment methods that do not properly account for correlation between SNPs. This method computes the PACT_{experiment-wide} for each gene locus. Because individual-level genotypes were not available, we used genotypes from HapMap Phase 2 CEU samples to estimate correlation matrices for SNPs. A PACT_{experiment-wide} <0.05 was defined as being significant experiment wide.

Stage 2 Analyses

SNPs were selected for stage 2 meta-analysis of their association with eGFR in independent cohorts if they met the following criteria: (1) they had an association P value <10⁻⁴ in the CKDGen discovery result for either eGFR or CKD; (2) the associations with eGFR, CKD, and eGFRcys were all direction consistent; and (3) they mapped into regions not previously associated with eGFR in GWAS. If no SNPs reached the defined P value threshold of <10⁻⁴ for eGFR or CKD, we selected the SNP with the lowest P value for the association.

SNPs thus identified were analyzed in stage 2 analysis in 19 independent cohorts (Supplemental Table 4) using inverse-variance weighted meta-analysis in METAL software.^15,33 Significance in stage 2 meta-analysis was defined as one-sided P<0.05. Information regarding de novo and in silico genotyping and imputation platforms for the stage 2 cohorts was previously described.¹⁵

Disclosures

None.