Mild Recessive Mutations in Six Fraser Syndrome–Related Genes Cause Isolated Congenital Anomalies of the Kidney and Urinary Tract

Abstract

Congenital anomalies of the kidney and urinary tract (CAKUT) account for approximately 40% of children with ESRD in the United States. Hitherto, mutations in 23 genes have been described as causing autosomal dominant isolated CAKUT in humans. However, >90% of cases of isolated CAKUT still remain without a molecular diagnosis. Here, we hypothesized that genes mutated in recessive mouse models with the specific CAKUT phenotype of unilateral renal agenesis may also be mutated in humans with isolated CAKUT. We applied next-generation sequencing technology for targeted exon sequencing of 12 recessive murine candidate genes in 574 individuals with isolated CAKUT from 590 families. In 15 of 590 families, we identified recessive mutations in the genes FRAS1, FREM2, GRIP1, FREM1, ITGA8, and GREM1, all of which function in the interaction of the ureteric bud and the metanephric mesenchyme. We show that isolated CAKUT may be caused partially by mutations in recessive genes. Our results also indicate that biallelic missense mutations in the Fraser/MOTA/BNAR spectrum genes cause isolated CAKUT, whereas truncating mutations are found in the multiorgan form of Fraser syndrome. The newly identified recessive biallelic mutations in these six genes represent the molecular cause of isolated CAKUT in 2.5% of the 590 affected families in this study.

Congenital anomalies of the kidney and urinary tract (CAKUT) are one of the most frequent congenital abnormalities in humans, taking a high toll on affected individuals, their families, and health care. The estimated occurrence of CAKUT is 3–6 per 1000 live births.¹ They represent the most frequent cause of CKD and ESRD in children in the United States, accounting for >40% of all cases.² CAKUT comprise a broad spectrum of structural malformations and functional anomalies, including unilateral renal agenesis, renal hypodysplasia, ureteropelvic junction obstruction, and vesicoureteral reflux. The clinically distinct CAKUT phenotypes have in common a disturbed embryonic codevelopment of tissues derived from the ureteric bud and the metanephric mesenchyme.³ Although >200 different forms of syndromic CAKUT have been described,⁴ isolated CAKUT account for the majority of cases.^5,6 Twenty-three autosomal dominant genes have been identified to cause isolated CAKUT, with TNXB, WNT4, and DSTYK being the most recent ones.^7–10 According to the Mouse Genome Informatics database (http://www.informatics.jax.org), 1768 monogenic mouse models for CAKUT have been described, many of which are recessive and do not have a human disease correlate.

In this study, we hypothesized that targeted genes in recessive mouse models with CAKUT may also be mutated in humans with isolated CAKUT. Candidate genes were selected based on recessive murine models for CAKUT with the distinct CAKUT phenotype of unilateral renal agenesis in homozygous null animals. We reckoned that mouse models with unilateral renal agenesis represented the most promising CAKUT candidate genes, because unilateral renal agenesis is a severe and specific CAKUT phenotype. To test this hypothesis, we performed barcoded next-generation sequencing (NGS)–based exon sequencing of these candidate genes as previously described by our group.^11,12

We analyzed the coding sequences of 12 recessive murine candidate genes (Supplemental Table 1) in 672 individuals from 590 families with isolated CAKUT (Supplemental Table 2). Mutations in 17 known CAKUT-causing genes were excluded before this study (see Concise Methods). For individuals with renal hypodysplasia (n=101), we also excluded the presence of an HNF1B deletion by quantitative PCR.

In 15 of 672 individuals from 590 families (2.5%) with isolated CAKUT, we identified 21 different mutations in six different recessive genes (Table 1). These genes were FRAS1, FREM2, GRIP1, FREM1, ITGA8, and GREM1. The most frequently mutated gene in this study was FRAS1, accounting for six unrelated individuals with isolated CAKUT. In the six murine “single kidney” genes BAG6, CTNNBIP1, DACT1, ILK, LIN7C, and LRP4, there were no biallelic mutations present in humans with isolated CAKUT. We considered recessive alleles as likely disease causing if they either were protein-truncating (n=2) or missense mutations affecting an evolutionary conserved amino acid residue and with a minor allele frequency of <1% in 13,000 control chromosomes of the National Heart Lung and Blood Institute Exome Sequencing Project (n=17). All detected alleles were confirmed by Sanger sequencing in genomic DNA of the affected individuals.

Table 1.

Recessive mutations in 6 murine candidate genes detected in 15 unrelated individuals with isolated CAKUT

Gene	Family -Individual	Sex	Geographic Origin	Renal Phenotype	Extrarenal Phenotype	Nucleotide Change (Zygosity)	Exon	Amino Acid Change	Segregation/State of Alleles	Mm	Gg	Xt	Dr	EVS	SIFT	MT	PP2
FRAS1	A1250-21	Male	Kuwait	Hypospadias	None	c.4579C>T (H)	34	R1527W	Hom	R	R	P	R	24 of 12,416	Deleterious	Disease causing	0.724
FRAS1	A1402-21	Male	Arabia	Right duplex	None	c.4579C>T (h)	34	R1527W	transa	R	R	P	R	24 of 12,416	Deleterious	Disease causing	0.724
						c.9364C>T (h)	62	R3122W	transa	R	R	R	R	0 of 13,000	Deleterious	Disease causing	1
FRAS1	A2381-22b	Female	Germany	Left agenesis	None	c.7867C>T (h)	55	R2623*	Paternal inheritance					0 of 13,000	N/A	N/A	N/A
						c.9821G>A (h)	64	R3274H	Maternal inheritance	R	R	R	R	0 of 13,000	Deleterious	Disease causing	0.999
FRAS1	A3455-21	Male	Macedonia	Right agenesis	Anal atresia	c.4579C>T (h)	34	R1527W	Maternal inheritance	R	R	P	R	24 of 12,416	Deleterious	Disease causing	0.724
						c.7622A>G (h)	53	N2541S	Paternal inheritance	N	N	N	N	28 of 12,274	Deleterious	Disease causing	0.997
FRAS1	A3975-21	Male	Germany	Right ectopic medullary cystic kidney disease	None	c.4159_4161 delinsTTA (h)	31	A1387L	trans in 1000Gc	A	A	A	A	13 of 12,487	N/A	N/A	N/A
						c.9806G>A (h)	64	R3269Q	trans in 1000Gc	R	R	R	R	88 of 12,570	Deleterious	Disease causing	0.999
FRAS1	A4175-21	Male	UK	VUR	None	c.9806G>A (H)	64	R3269Q	Hom	R	R	R	R	88 of 12,570	Deleterious	Disease causing	0.999
						c.9959C>T (h)	64	S3320F	n/a	S	S	S	S	0 of 13,000	Deleterious	Disease causing	0.999
FREM2	A1023-21d	Male	India	Right agenesis	None	c.4031G>A (h)	1	R1344H	trans in 1000Gc	R	R	R	R	25 of 12,981	Deleterious	Disease causing	0.085
						c.7535G>A (h)	15	R2512H	trans in 1000Gc	R	R	R	R	10 of 12,996	Deleterious	Disease causing	0.929
						c.7013C>T (h)	12	T2338I	n/a	T	T	T	T	0 of 13,000	Deleterious	Disease causing	0.999
FREM2	A1232-21	Male	Kuwait	PUV, right VUR, left UPJO, CKD	None	c.649C>T (h)	1	R217C	transe	R	R	R	E	0 of 13,000	Deleterious	Disease causing	0.836
						c.4031G>A (h)	1	R1344H	transe	R	R	R	R	25 of 12,981	Deleterious	Disease causing	0.085
FREM2	A1548-21	Male	India	B UPJO	CM	c.685C>T (h)	1	R229C	transe	R	R	R	R	0 of 13,000	Deleterious	Disease causing	0.95
						c.4820A>G (h)	1	D1607G	transe	D	D	D	D	0 of 13,000	Deleterious	Disease causing	0.983
FREM2	A358-21	Female	Kuwait	B VUR, CKD	None	c.7211T>C (H)	13	I2404T	Hom	I	V	V	V	0 of 13,000	Deleterious	Disease causing	0.075
GRIP1	A3390-21	Female	Macedonia	Right duplex	None	c.1846G>A (h)	16	G616R	Paternal inheritance	G	G	G	G	0 of 13,000	Deleterious	Disease causing	0.606
						c.2750G>T (h)	22	R917L	Maternal inheritance	R	R	R	R	0 of 13,000	Tolerated	Disease causing	0.087
FREM1	A3369-21	Male	Macedonia	B VUR III°	B SD II/III	c.4879G>T (H)	28	A1627S	Hom	A	A	A	A	1 of 12,307	Deleterious	Disease causing	0.893
FREM1	A688-21	Male	Macedonia	Right VUR V°, right RHD	None	c.4879G>T (H)	28	A1627S	Hom	A	A	A	A	1 of 12,307	Deleterious	Disease causing	0.893
ITGA8	A876-21	Female	Macedonia	Left duplex, left VUR III°	None	c.2982+2T>C (H)	28	—	Hom					0 of 13,000	N/A	N/A	N/A
GREM1	A3573-21	Female	Macedonia	Left agenesis	None	c.103C>G (H)	2c	P35A	Paternal inheritance, maternal inheritance	P	P	P	P	34 of 12,952	Tolerated	Disease causing	0.015

Mm, mus musculus; Gg, gallus gallus; Xt, xenopus tropicalis; Dr, danio rerio; EVS, allele frequency in EVS database; SIFT, sorting intolerant from tolerant prediction class; MT, mutation taster prediction class; PP2, PolyPhen2 humvar score; FRAS1 (NM_025074.6); H, homozygous; Hom, homozygous; h, heterozygous; trans, alleles are “in-trans,” meaning they are on different chromosomes; Agenesis, renal agenesis; N/A, not applicable; trans in 1000G, both alleles are present in the 1000G cohort but never in the same individual (Supplemental Table 6); VUR, vesicoureteral reflux graded in Roman numerals I–V according to the International Classification of VUR; FREM2 (NM_207361.4); PUV, posterior urethral valves; UPJO, ureteropelvic junction obstruction; CM, cardiomyopathy of unknown etiology; GRIP1 (NM_021150.3); duplex, duplex collecting system; FREM1 (NM_144966.5); B, bilateral; SD II/III, syndactyly toes II/III; RHD, renal hypodysplasia; ITGA8 (NM_003638.1); GREM1 (NM_013372.6).

^aNo parental DNA available. Alleles are most likely in trans because of uneven distribution (24 of 12,416 versus 0 of 13,000) of counts in the National Heart Lung and Blood Institute Exome Sequencing Project cohort of 6500 control individuals (http://evs.gs.washington.edu/EVS/).

^bA2381-22 was published in a previous study of our group reporting identical mutations.²⁵ A2381-21 had bilateral renal agenesis with absent urinary bladder (termination of pregnancy).

^cNo parental DNA available. Alleles are most likely in trans because individuals of the 1000 Genomes Project show absence of linkage disequilibrium (Supplemental Table 6).

^dA1023-21 was published in a previous study of our group for the heterozygous FREM2 mutation p.T2338I.²⁵

^eNo parental DNA available. Compound heterozygous state was confirmed by subcloning of genomic DNA and Sanger sequencing of mono-molecular clones.

In the genes FRAS1, FREM2, and GRIP1, which cause Fraser syndrome if mutated (Online Mendelian Inheritance in Man [OMIM] 219000), we detected recessive missense mutations in 11 unrelated individuals with isolated CAKUT (n=6, 4, and 1, respectively) (Table 1). Mutations in FRAS1, FREM2, or GRIP1 were detected in individuals with different isolated CAKUT phenotypes (Table 1). Ten of 11 individuals had no extrarenal manifestations characteristic for Fraser syndrome (e.g., cryptophthalmos, syndactyly, or genital malformations). One individual, A3455-21, had CAKUT and anal atresia, which is a diagnostic criterion of Fraser syndrome (Table 1). However, this was not sufficient to make the clinical diagnosis.¹³ Interestingly, we here discovered recessive missense mutations in FRAS1, FREM2, and GRIP1 as novel causes of isolated CAKUT, whereas biallelic truncating mutations are known to cause Fraser syndrome with multiorgan involvement (Supplemental Table 3).

In addition to the Fraser syndrome genes, we detected recessive mutations in the gene FREM1 that cause the closely Fraser-related phenotypes Manitoba Oculotrichoanal (MOTA) syndrome¹⁴ (OMIM 248450) or Bifid Nose with or without Anorectal and Renal Anomalies (BNAR) syndrome¹⁵ (OMIM 608980) in an allele-dependent manner. In this study, we identified two unrelated individuals with isolated CAKUT with the same homozygous mutation in FREM1 (FREM1 p.A1627S) (Table 1). Individual A688-21 had isolated CAKUT, whereas individual A3369-21 had CAKUT with bilateral syndactyly of toes II-III, which is a feature of MOTA syndrome. However, MOTA/BNAR-characteristic facial dysmorphism was absent from both individuals. Similarly to the detected mutations in the Fraser genes, our data support that biallelic missense mutations in FREM1 cause isolated CAKUT, whereas truncating mutations are only reported in individuals with the severe multiorgan phenotype of MOTA/BNAR syndrome (Supplemental Table 3).

In addition to Fraser spectrum genes, we detected homozygous recessive mutations in ITGA8 and GREM1, which have not previously been implicated in human CAKUT. In individual A876-21 with a left duplicated collecting system and left high-grade vesicoureteral reflux, we detected a homozygous obligatory splice-site mutation in ITGA8 (Table 1). The nucleotide change ITGA8 c.2982+2T>C is predicted to abrogate the splice donor site of exon 28 (5′ splicing site) by five publically available splice-site prediction algorithms (SpliceSiteFinder-like, MaxEntScan, NNSPLICE, GeneSplicer, and Human Splicing Finder).

In individual A3573-21, we detected a homozygous missense mutation in the BMP4-antagonist GREM1 (Table 1). GREM1 p.P35A segregated from the heterozygous parents to the affected child with unilateral renal agenesis. The unaffected sibling A3579-22 was heterozygous for the mutation (Table 1).

In this study, we identified mutations in six recessive murine “single kidney” candidate genes in 672 individuals from 590 families as novel single-gene causes of isolated CAKUT. These six recessive genes account for mutations in 2.5% of 590 families with isolated CAKUT. Together with heterozygous mutations in 17 known autosomal dominant CAKUT-causing genes, which account for another 6.3%,¹⁶ we are now able to molecularly “solve” almost 10% of cases with isolated CAKUT in our cohort.

We detected recessive biallelic missense mutations in the four Fraser spectrum genes in individuals with isolated CAKUT. The proteins encoded by the Fraser genes FRAS1, FREM2, and FREM1 form a tertiary protein complex lining the extracellular epithelial-mesenchymal interface.^17,18 Loss of function of any of these proteins disrupts the Fraser-protein complex and leads to the severe multiorgan developmental phenotype of Fraser syndrome in humans and mice (Supplemental Table 1).^14,19–21 In this study, 11 of 13 individuals with recessive mutation in the Fraser genes had no extrarenal phenotype. We observed almost exclusively biallelic missense mutations in individuals with isolated CAKUT, whereas individuals with Fraser syndrome generally are reported to have protein-truncating alleles (Supplemental Table 3). This may represent an example of allelism in a recessive disease, which has been described and discussed more extensively in nephronophthisis, a rare form of cystic kidney disease.²² Hence, we propose the hypothesis that the detected missense mutations causing isolated CAKUT represent hypomorphic alleles, which can be compensated for in the development of extrarenal tissues and consequently cause the mildest phenotype in the Fraser/MOTA/BNAR/CAKUT spectrum.²³

Phenotypic heterogeneity has been reported in Fraser/MOTA syndrome, even within affected families.²⁴ Similarly, the unrelated individuals A688-21 and A3369-21 have the same homozygous mutation (Table 1); however, only individual A3369-21 has syndactyly of toe II/III, which represents a “mild-end” feature of the Fraser spectrum.

In a previous study, we hypothesized that heterozygous mutations in FRAS1 and FREM2 may cause autosomal dominant isolated CAKUT in humans.²⁵ However, this hypothesis is not supported by the data of the current more extensive study and insight from growing public variant databases (Supplemental Table 4).

FRAS1 has been shown to be expressed in glomerular podocytes and podocyte precursors.²⁶ In this study, we did not observe nephrotic syndrome in any of the participants with mutations in the Fraser genes. This is in line with individuals with Fraser syndrome.¹³ Hence, we conclude that FRAS1 is not essential for glomerulogenesis.

Besides mutations in the Fraser genes, we identified two individuals with homozygous mutations in ITGA8 and GREM1, which previously have not been implicated in human CAKUT. Interestingly, all of the six genes found mutated in this study functionally converge on two signaling pathways essential for kidney development: Integrin-α8 interacts indirectly with the Fraser-protein complex, with this interaction being essential for GDNF expression in the metanephric mesenchyme.^17,27 GDNF-RET signaling at the ureteric bud/metanephric mesenchyme interface is counterbalanced by BMP4 signaling. GREM1 is an antagonist of BMP4, a known CAKUT-causing gene, and thus acts synergistically with the Fraser-protein complex/integrin-α8/GDNF axis.²⁸

On the basis of the well defined mouse phenotype of unilateral renal agenesis, we identified six recessive monogenic causes of isolated CAKUT in humans, two of them as novel CAKUT genes. We thereby identified the first recessive isolated CAKUT genes in humans. The six genes found to be mutated encode for proteins that functionally converge on the GDNF-RET/BMP signaling pathways at the interface of the ureteric bud and the metanephric mesenchyme. We showed genetic evidence that mild mutation in the Fraser spectrum genes (FRAS1, FREM2, GRIP1, and FREM1) causes isolated CAKUT as the mildest phenotype of the clinical spectrum. These six genes contribute to 2.5% of all cases of isolated CAKUT; thus, up to 20% of individuals with CAKUT could now be molecularly “solved” by exon sequencing in 29 genes and copy number variation analysis.^9,29–31 Rapidly developing sequencing technology will continue to facilitate identification of rare single-gene causes of human CAKUT and enables us to conduct large-scale mutation analyses providing families with a genetic diagnosis in an increasing number of cases.

Concise Methods

Human Participants

After informed consent, we obtained clinical data, blood samples, and pedigrees from individuals with CAKUT. Approval for research on humans was obtained from the University of Michigan Institutional Review Board. Mutations in the following genes known to be mutated in isolated CAKUT in humans were excluded before this study: BMP4, BMP7, CDC5L, CHD1L, EYA1, GATA3, HNF1B, PAX2, RET, ROBO2, SALL1, SIX1, SIX2, SIX5, SOX17, UMOD, and UPK3A. HNF1B deletions were excluded by quantitative PCR in individuals with the CAKUT phenotype of renal hypodysplasia.

Candidate Gene Selection

Candidate genes were selected from the Mouse Genome Informatics database based on the presence of the CAKUT phenotype of unilateral renal agenesis (phenotype ID MP: 0003604) in homozygotes. Eighty-one genotypes involving 39 genes matched this criterion. Twenty-eight genes were excluded because they either were known to be associated with isolated CAKUT (n=5), or were reported in mouse models with conditional/complex knock-out alleles, or were not purely autosomal recessive. FREM2 was included in this study because three of four Fraser spectrum genes (FRAS1, GRIP1, and FREM1) met the inclusion criteria. The selected candidate genes were BAG6, CTNNBIP1, DACT1, FRAS1, FREM1, FREM2, GREM1, GRIP1, ILK, ITGA8, LIN7C, and LRP4 (Supplemental Table 1).

Targeted Exon Sequencing

For 12 genes, 355 target-specific primer pairs were designed covering 273 coding exons (Supplemental Table 5). The amplicon size ranged from 150 to 290 bp. Targeted amplification and NGS was done as described previously by our group^11,12 with the following alterations: 12 primer pairs were multiplexed in 48 pools to allow for amplification of 576 PCR products per sample simultaneously using the Fluidigm 48.48 Access Array IFC System. NGS was carried out using an Illumina HiSeq2000 instrument (1×150 bp single reads) or Illumina MiSeq V2 instrument (2×250 bp paired reads). Our results showed that 326 amplicons (91.8%) had a coverage >100× and 11 amplicons (3.0%) were covered <10× (Supplemental Table 3). All identified mutations were confirmed by Sanger sequencing in genomic DNA. Segregation analysis was performed if parental DNA was available. Alternatively, fragments of genomic DNA harboring two alleles were TA cloned into pGEM Easy-vector (Promega). Six clones were Sanger sequenced in order to confirm the compound heterozygous state.

Bioinformatics

NGS data alignment and variant detection were performed using CLC Genomics Workbench 4.9 software. Variant filtering was carried out as published previously by our group.^11,12 Homozygous or ≥2 heterozygous alleles in one gene were considered for subsequent confirmation by Sanger sequencing (Supplemental Table 5).

Disclosures

None.