Mutations in PRDM15 Are a Novel Cause of Galloway-Mowat Syndrome

Abstract

Background 

Galloway-Mowat syndrome (GAMOS) is characterized by neurodevelopmental defects and a progressive nephropathy, which typically manifests as steroid-resistant nephrotic syndrome. The prognosis of GAMOS is poor, and the majority of children progress to renal failure. The discovery of monogenic causes of GAMOS has uncovered molecular pathways involved in the pathogenesis of disease.

Methods 

Homozygosity mapping, whole-exome sequencing, and linkage analysis were used to identify mutations in four families with a GAMOS-like phenotype, and high-throughput PCR technology was applied to 91 individuals with GAMOS and 816 individuals with isolated nephrotic syndrome. In vitro and in vivo studies determined the functional significance of the mutations identified.

Results 

Three biallelic variants of the transcriptional regulator PRDM15 were detected in six families with proteinuric kidney disease. Four families with a variant in the protein’s zinc-finger (ZNF) domain have additional GAMOS-like features, including brain anomalies, cardiac defects, and skeletal defects. All variants destabilize the PRDM15 protein, and the ZNF variant additionally interferes with transcriptional activation. Morpholino oligonucleotide-mediated knockdown of Prdm15 in Xenopus embryos disrupted pronephric development. Human wild-type PRDM15 RNA rescued the disruption, but the three PRDM15 variants did not. Finally, CRISPR-mediated knockout of PRDM15 in human podocytes led to dysregulation of several renal developmental genes.

Conclusions 

Variants in PRDM15 can cause either isolated nephrotic syndrome or a GAMOS-type syndrome on an allelic basis. PRDM15 regulates multiple developmental kidney genes, and is likely to play an essential role in renal development in humans.

Visual Abstract

Galloway-Mowat syndrome (GAMOS) is a rare, inherited developmental disorder that is characterized by microcephaly, developmental delay, and a renal glomerulopathy.¹ It is clinically and genetically heterogeneous, and although the hallmark renal manifestations are congenital or infantile nephrotic syndrome, isolated proteinuria and later-onset nephrotic syndrome have also been described.^2–4 There is no consistent histologic feature on renal biopsy, and light microscopy findings include diffuse mesangial sclerosis, FSGS, and minimal change. Ultrastructural evaluation often demonstrates podocyte effacement and abnormalities of the glomerular basement membrane.^5,6 Affected individuals invariably develop ESKD, and the prognosis of GAMOS is poor. In addition to the renal manifestations and microcephaly, clinical features can include structural brain abnormalities, facial dysmorphisms, skeletal anomalies, and hiatal hernia. To date, mutations in a number of different genes, including WDR73, WDR4, NUP107, and members of the evolutionarily conserved KEOPS complex, have been identified as causing GAMOS.^2–4,7,8

In 1998, Mildenberger et al.⁹ described a child with a GAMOS-like syndrome, with microcephaly, facial dysmorphisms, developmental delay, and heavy proteinuria, although without overt nephrotic syndrome. Subsequently, we identified three additional affected individuals from unrelated families with similar clinical manifestations. We performed whole-exome sequencing (WES) and linkage analysis, and identified a recurring variant in the transcriptional regulator PRDM15 as a novel cause of this GAMOS-like syndrome, which we term GAMOS type 9, Mildenberger type. We additionally identified different variants in PRDM15 in two unrelated individuals with isolated early-onset nephrotic syndrome, and we demonstrate that PRDM15 plays a critical role in renal development.

Methods

Research Participants

After obtaining informed consent, we collected blood samples and pedigrees from individuals with proteinuria or nephrotic syndrome, including families with extrarenal manifestations like microcephaly. Approval for human subjects research was obtained from Institutional Review Boards of the University of Michigan and Boston Children’s Hospital. The diagnosis of GAMOS and nephrotic syndrome were on the basis of published clinical criteria.^1,10 Clinical data were obtained using a standardized questionnaire (http://www.renalgenes.org).

Homozygosity Mapping and WES

WES was performed as previously described, using Agilent SureSelect human exome capture arrays (Thermo Fisher Scientific) with next-generation sequencing (NGS), on an Illumina platform.¹¹ Sequence reads were mapped against the human reference genome (National Center for Biotechnology Information build 37/hg19), using CLC Genomics Workbench (version 6.5.1) (CLC bio). For homozygosity mapping, downstream processing of aligned BAM files were done using Picard and samtools.¹² SNV calling was done using GATK5, and the generated VCF file was subsequently used in HomozygosityMapper. Genetic regions of homozygosity by descent were plotted across the genome as candidate regions for recessive genes, as previously described.¹² Mutation calling was performed in line with proposed guidelines by scientists with knowledge of clinical phenotypes, pedigree structure, and genetic mapping.¹³

High-Throughput Mutation Analysis by Array-Based Multiplex PCR and NGS

We utilized the 48.48 Access Array microfluidic technology (Fluidigm) to perform barcoded, multiplex PCR, as described previously.^8,14 A total of 91 individuals with nephrotic syndrome and brain anomalies, and 816 individuals with isolated nephrotic syndrome, were sequenced. Then, 2×250 bp paired-end sequencing was performed on an Illumina MiSeq instrument. Sequence alignment was conducted using CLC Genomics Workbench (CLC bio). Identified variants were confirmed by Sanger sequencing.

Thermal Stability Assay

The wild-type (WT), p.Met154Lys, and p.Glu190Lys PRDM15 SET domains were expressed as His-tag fusions in Escherichia coli BL21Star (DE3) from pProEXHTb vector. The proteins were first purified by affinity chromatography using Ni²⁺resin. After His-tag cleavage with tobacco etch virus protease, the proteins were further purified by a second Ni²⁺ column and size-exclusion chromatography. Thermal stability of the WT and p.Met154Lys SET domains was analyzed, measuring temperature-dependent differences in intrinsic tryptophan fluorescence with Prometheus NT.48 (Nanotemper), at concentrations of 0.5, 1, and 2 mg/ml, resulting in essentially identical melting curves.

Complementary DNA Cloning

Human PRDM15 full-length complementary DNA (cDNA) was subcloned by PCR from human full-length cDNA (GenBank accession number NM_001040424.2). cDNA was subcloned into the pENTR-D-TOPO vector (Thermo Fisher Scientific). Clones reflecting the variants identified in the affected individuals were introduced in the cDNA constructs using the Quick change II XL site-directed mutagenesis kit (Agilent Technologies), following the manufacturer’s instructions.

Luciferase Assay

A DNA portion (392 bp) of the Rspo1 promoter region containing the PRDM15 binding site was amplified by PCR, and cloned into the PGL4.23 firefly luciferase reporter vector (#E8411; Promega) using KpnI and HindIII restriction enzymes. HEK293 cells were cotransfected with the following combination of plasmids: empty vector, WT, or mutant PRDM15; the Rspo1 reporter plasmid; and a constitutive Renilla luciferase plasmid for normalization. Luciferase activity was measured 24–36 hours after transfection, using a dual-injector 96 Microplate Luminometer (#910376; VERITAS). Firefly luciferase readings were then normalized to those of the Renilla luciferase.

Xenopus Experiments

Xenopus laevis embryos were obtained, cultured, and staged according to standard protocols.^15,16 All procedures were performed in agreement with the German animal use and care law (Tierschutzgesetz), and were approved by the administration of the state of Baden-Württemberg (Regierungspräsidium Tübingen).

For generating a Xenopus prdm15 cDNA probe, a 1008 bp fragment was cloned into the pSC-B vector (Stratagene, La Jolla, California) with the following primers: forward, 5′-ttg​aac​cca​agt​tcc​tcc​ac-3′; reverse, 5′-gct​gtt​ggt​gga​gaa​gaa​gc-3′. Whole-mount in situ hybridization was performed according to standard protocols.¹⁷ Digoxigenin-labeled antisense RNA probes were generated by means of in vitro transcription using T7 or T3 RNA polymerases (Roche, Grenzach-Wyhlen, Germany). Xenopus embryos at stages 33 and 36 were collected, fixed using formaldehyde, and stained using BM Purple and NBT/BCIP (Roche).

To perform loss-of-function experiments, Prdm15 morpholino oligonucleotide (MO) (5′-TCA​TTC​ACA​CCT​GCT​CCT​CAA​TAG​C-3′) and control MO (5′-CCT​CTT​ACC​TCA​GTT​ACA​ATT​TAT​A-3′) were purchased from Gene Tools (Philomath, Oregon) and diluted in diethyl pyrocarbonate-treated water. We analyzed the translational blocking efficiency of the used MO by cloning the Prdm15 MO binding site of Xenopus and human PRDM15 (rescue construct) in front of, and in frame with, GFP in the pCS2⁺ vector, with the following primers: Xenopus_prdm15_MO_forward, 5′-GAT​CCG​CTA​TTG​AGG​AGC​AGG​TGT​GAA​TGA​TGG​GG-3′; Xenopus_prdm15_MO_reverse, 5′-AAT​TCC​CCA​TCA​TTC​ACA​CCT​GCT​CCT​CAA​TAG​CG-3′; human_PRDM15_MO_forward, 5′-GAT​CCT​CCG​CGG​CCG​CCC​CCT​TCA​CCA​TGG​CG-3′; human_PRDM15_MO_reverse, 5′-AAT​TCG​CCA​TGG​TGA​AGG​GGG​CGG​CCG​CGG​AG-3′.

Subsequently, 1 ng of the indicated RNA and 10 ng of Prdm15 MO or control MO were injected into two-cell stage embryos, and the translation of green fluorescent protein (GFP) was monitored at stage 15.

To target the pronephric tissue, 15 ng of Prdm15 MO was injected unilaterally into one vegetal-ventral blastomere of eight-cell-stage Xenopus embryos.¹⁸ The uninjected side served as an internal control. Successful and correct injections were controlled by the coinjection of 0.5 ng RNA coding for GFP, using a fluorescence microscope (M-VX, U-RFL-T; Olympus, Tokyo, Japan). For rescue experiments, 15 ng of Prdm15 was coinjected with 0.5 ng human full-length PRDM15 or human full-length PRDM15 with the specific point mutations identified in human patients (hPRDM15_M154K, hPRDM15_E190K, and hPRDM15_C844Y). Embryos were imaged using an SZX12 Olympus microscope. Measurement of the fxyd2 expression in the anterior pronephric area and the posterior tubular length were done using ImageJ (64-bit)/Fiji (National Institutes of Health).

CRISPR-Cas9-Mediated Genome Editing

CRISPR-Cas9 experiments were performed in immortalized human podocytes, which were a kind gift from Moin Saleem (University of Bristol, Bristol, United Kingdom), and were cultured as previously described.¹⁹ Cell lines were tested monthly for mycoplasma contamination.

CRISPR-Cas9-mediated PRDM15^−/− human podocyte cell lines were generated as previously described.²⁰ Briefly, guide RNAs were designed using the online CRISPR design tool CHOPCHOP (http://chopchop.cbu.uib.no/), and were cloned into the BbsI sites of the pSpCas9(BB)-2A-GFP (pX458) vector (Addgene plasmid #48138; gift from Feng Zhang). Constructs were then transfected into human immortalized podocytes using Lipofectamine 2000, according to the manufacturer’s instructions. A scramble guide RNA was used as a control. Clonal cell lines were isolated by dilution, and selected clones were analyzed by Sanger sequencing. The sequences of the oligonucleotides used to generate clonal cell lines and the sequences of the PCR primers used for Sanger sequencing are listed in Supplemental Table 1. PRDM15 depletion was further confirmed via Western blotting and indirect immunofluorescence (Supplemental Figure 1). Cell lines C7 (guide RNA targeting exon 4) and D12 (guide RNA targeting exon 1) were utilized for RNA-sequencing experiments.

Antibodies

For immunofluorescence and immunoblotting experiments, the following primary antibodies were used, respectively: mouse anti-PRDM15 (1:100, ab69206; Abcam) and rabbit anti-PRDM15 (gifted by the Guccione laboratory).²¹ Donkey anti-rabbit and anti-mouse Alexa Fluor 488 and Alexa Fluor 594 conjugated secondary antibodies, and 4',6-diamidino-2-phenylindole dihydrochloride, were obtained from Invitrogen. HRP-labeled secondary antibodies were purchased from Santa Cruz Biotechnology.

Immunofluorescence and Confocal Microscopy in Cell Lines

For immunostaining, human immortalized podocytes were seeded on cover slips, and grown at permissive temperature. Cells were fixed and permeabilized for 10 minutes, using 4% paraformaldehyde and 0.01% SDS. After blocking, cells were incubated overnight at 4°C. The cells were incubated in secondary antibodies for 60 minutes at room temperature, followed by 5 minutes staining with 1× 4',6-diamidino-2-phenylindole dihydrochloride/PBS. Confocal imaging was performed, using Leica SP5× system with an upright DM6000 microscope, and images were processed with the Leica AF software suite.

RNA Sequencing and Bioinformatics Analysis

Total RNA was isolated using the Qiagen RNeasy Mini Kit, according to the manufacturer’s instructions. All RNA samples were measured for quantity and quality. Library preparation and RNA sequencing for the CRISPR cell lines was performed by the Molecular Biology Core Facilities at the Dana-Farber Cancer Institute. Libraries were prepared using Roche Kapa mRNA HyperPrep sample preparation kits, from 100 ng of purified total RNA, according to the manufacturer’s protocol. The finished double-stranded DNA libraries were quantified by Qubit fluorometer, Agilent TapeStation 2200, and quantitative RT-PCR, using the Kapa Biosystems library quantification kit, according to manufacturer’s protocols. Uniquely indexed libraries were pooled in equimolar ratios and sequenced on an Illumina NextSeq500, with single-end 75 bp reads, by the Molecular Biology Core Facilities at the Dana-Farber Cancer Institute. Sequenced reads were mapped to the GRCh38 with National Center for Biotechnology Information Ensembl IDs, using STAR aligner.²² Analysis of differential expression was conducted using the edgeR package in R and a false discovery rate threshold of <0.05. edgeR was also used to generate multidimensional scaling plots, volcano plots, and TMM-normalized counts per million, which were utilized to generate heat maps.^23,24 Enriched Gene ontology terms were generated using the goseq package in R.²⁵ Raw data were submitted to the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/), under accession number GSE159978.

Quantitative Real-Time PCR

Total RNA was isolated using the Qiagen RNeasy Mini Kit, and 2 µg of RNA was reverse transcribed into cDNA using the iScript cDNA synthesis Kit (Bio-Rad), according to the manufacturer’s instructions. Real-time PCR was performed using TaqMan probes purchased from Thermo Fisher Scientific (Supplemental Table 1).

Statistical Analyses

Statistical analyses were performed using GraphPad Prism software (San Diego, CA). Significances were calculated using an unpaired t test. P values <0.05 were considered statistically significant and are indicated in the figures. For Xenopus data, GraphPad Prism 6 software was used to calculate P values by a nonparametric, one-tailed Mann–Whitney U rank-sum test. Statistical significance was indicated as follows: *P≤0.05; **P≤0.01; ***P≤0.001; ****P<0.001.

Results

Clinical Manifestations of Four Patients with a GAMOS-Like Syndrome

Four affected individuals from unrelated families of Arab origin (MIC, B44, B53, and B54) exhibited clinical findings consistent with a GAMOS-like syndrome, including microcephaly, abnormal cerebral gyration, developmental delay, and heavy proteinuria. The clinical characteristics of the index case, MIC-21, were described previously by Mildenberger et al., and therefore we refer to this as GAMOS type 9, Mildenberger type.⁹ The three affected individuals in families B44, B53, and B54 exhibited a strikingly similar multiorgan phenotype. All affected individuals from the four families had postaxial polydactyly and facial dysmorphisms, such as a narrow forehead or trigonocephaly. Three individuals had colobomas, and two were noted to have microcoria. Three of the four affected individuals had cardiac involvement (Figure 1, A–C, Table 1).

Figure 1.: Biallelic variants in PRDM15 can cause Galloway-Mowat syndrome and isolated nephrotic syndrome in humans. (A) Renal histology of individual MIC-21 shows diffuse mesangial sclerosis, with areas of glomerular scarring (black arrow) on light microscopy. Bars, 50 μm. (B) Brain magnetic resonance imaging for patient B53-21 at age 6 weeks. Left: axial T1-weighted image showing missing myelination of the posterior limb of the internal capsule and lateral thalamus, which should be visible by this age. Right: axial T2-weighted image showing hypoplasia and abnormal gyration of the left temporal lobe (orange arrowhead), and enlarged left ventricle. (C) Hand x-rays of individual MIC-21, demonstrating bilateral postaxial hexadactyly. (D) Exon structure of human PRDM15 cDNA and position of mutations. Positions of start codon (ATG) and stop codon (TGA) are indicated. Below is the protein domain structure of PRDM15. The SET (orange) and 17 zinc-finger (blue) domains are depicted in relation to encoding exon position. Sanger sequencing is provided for the PRDM15 variants identified in six unrelated families and for healthy controls. Red arrowheads denote altered nucleotides. (E) Multiple amino acid sequence alignment of PRDM15 throughout evolution, and multiple sequence alignment of human PRDM15 paralogs PRDM10, PRDM4, and PRDM1. (F) Comparison of PRDM15 with the crystal structure of the mouse PRDM9 (PDB-4C1Q). Met316 (red box) in PRDM9, corresponding to PRDM15 Met154, is part of the hydrophobic core of the protein and is in proximity to strand β10, which is directly involved in SAH binding. A substitution of Met154 to lysine is predicted to disrupt the tertiary structure of the SET domain. (G) Glu352 (red box) in PRDM9 corresponds to Glu190 in PRDM15, and forms several hydrogen bonds with Asn320 in strand β10 as well as Ser251 and Gly352 on a loop between β4 and β5 involved in SAH binding. The p.Glu190Lys mutation will likely destabilize this SAH binding region. (H) Cys780 (red box) in human PRDM9 corresponds to Cys844 of PRDM15. Cys844 constitutes one of the zinc (Zn) binding residues at the “knuckle” of C2-H2 Zn fingers (PDB-5EGB). Mutation of Cys844 to Tyr will disrupt Zn binding.

Table 1. - Genetic and clinical details for six unrelated individuals in whom three different PRDM15 mutations were identified

Genetic and Clinical Parameters	Individual A3530-21	Individual A3714-24	Individual MIC-21	Individual B44-21	Individual B53-21	Individual B54-23
cDNA change	c.461T>A	c.568G>A	c.2531G>A	c.2531G>A	c.2531G>A	c.2531G>A
Amino acid change	p.Met154Lys	p.Glu190Lys	p.Cys844Tyr	p.Cys844Tyr	p.Cys844Tyr	p.Cys844Tyr
Zygosity (segregation a )	Hom (ND)	Hom (het, het)	Hom (het, het)	Hom (het, het)	Hom (ND)	Hom (het, het)
PP2 b /SIFT c /MT d	0.942/del/DC	0.986/del/DC	1.000/del/DC	1.000/del/DC	1.000/del/DC	1.000/del/DC
gnomAD e	Not present	0/1/251,002	Not present	Not present	Not present	Not present
Ethnicity	Egyptian	Egyptian	Arab	Arab	Arab	Arab
Sex	Female	Female	Female	Male	Male	Male
Renal phenotype
Age at onset	1 wk	11 mo	1 wk	Before 2 mo	1 wk	2 wk
Proteinuria	Glomerular (congenital NS)	Glomerular (infantile NS)	Mixed glomerular and tubular	Glomerular	Mixed glomerular and tubular	Mixed glomerular and tubular
Other tubular involvement	None	None	Glucosuria, aminoaciduria	Aminoaciduria	Aminoaciduria	None
ESKD (age)	No	Yes (4 yr)	Yes (9 mo)	Yes (18 mo)	Yes (7 mo)	Yes (unknown)
Renal biopsy	ND	EM: GBM thickening	Light microscopy: DMS, tubular cysts (6 mo)	Light microscopy: diffuse glomerulosclerosis; EM: GBM thickening (postmortem)	ND	Light microscopy: MCD EM: abnormal GBM, podocyte effacement (4 wk)
Extrarenal manifestations
Central nervous system	None	None	Microcephaly, abnormal gyration, Dandy-Walker malformation	Microcephaly, abnormal gyration	Microcephaly, abnormal gyration, temporal lobe hypoplasia	HC not documented; simplified gyration, Dandy-Walker malformation
Facial dysmorphisms	None	None	Narrow forehead	Trigonocephaly	Narrow forehead	Trigonocephaly
Ocular	None	None	Microcoria, corneal clouding	Microcoria, retinal coloboma	Iris coloboma	Microphthalmia, chorioretinal coloboma
Cardiac	None	None	ASD	None	PFO, ASD	ASD
Skeletal	None	None	Polydactyly, nail hypoplasia	Polydactyly	Polydactyly, pes calcaneus	Polydactyly, pes calcaneus, dysplastic vertebrae

cDNA, complementary DNA; Hom, homozygous; ND, no data; het, heterozygous; del, deleterious; DC, disease causing; NS, nephrotic syndrome; EM, electron microscopy; GBM, glomerular basement membrane; DMS, diffuse mesangial sclerosis; MCD, minimal change disease; HC, head circumference; ASD, atrial septal defect; PFO, patent foramen ovale.

^aSegregation is listed as (maternal allele, paternal allele) when available. If parental DNA was not available, segregation is listed as “ND.”

^bPolyPhen-2 score, which predicts potential impact of an amino acid change on the structure and function of a protein. Ranges from 0 to 1.0; 0=benign, 1.0=probably damaging (http://genetics.bwh.harvard.edu/pph2).

^cSIFT, Sorting Intolerant From Tolerant, which predicts whether an amino acid change will affect protein function (http://sift.jcvi.org).

^dMutationTaster, prediction tool to determine deleteriousness of an amino acid substitution (http://www.mutationtaster.org).

^egnomAD, genome aggregation database, reported as the number of homozygotes/allele count/total allele number (http://gnomad.broadinstitute.org/).

All four affected individuals developed heavy proteinuria within the first 2 months of life, and progressed to ESKD by 1 year of age. Interestingly, in addition to glomerular proteinuria, all four individuals were noted to have tubular involvement, suggesting a more generalized renal developmental abnormality. Two individuals had associated hypoalbuminemia, but none developed edema. Three of the four individuals underwent renal biopsy, with variable light microscopy findings, including diffuse mesangial sclerosis, diffuse glomerulosclerosis, and minimal change (Figure 1A, Table 1). Two of the individuals had glomerular basement membrane abnormalities on electron microscopy. Ultrastructural evaluation was not done for the third individual. A summary of the clinical characteristics is provided in Table 1, with more detailed clinical information provided in Supplemental Table 2.

Recessive Variants in PRDM15 Detected in Patients with GAMOS-Mildenberger type

We performed WES and homozygosity mapping (Supplemental Figure 2) for affected individuals B44-21, B53-21, and B54-23.¹² Each exome was analyzed separately for biallelic variants within homozygous peaks, and our analysis process for individuals B44-21 and B53-21 is summarized in Supplemental Figure 3. We detected the same homozygous variant in PRDM15 (NM_001040424.2: c.2531G>A; p.Cys844Tyr) in individuals B44-21 and B53-21. No other variants in a shared gene were identified. Because of low coverage in the exome data for individual B54-23, no variants in PRDM15 were detected through WES. However, direct Sanger sequencing confirmed that individuals B54-23 and MIC-21 both also harbored this same PRDM15 c.2531G>A change (Figure 1D). This variant was absent in the large population database gnomAD (http://gnomad.broadinstitute.org) and in the Saudi Human Genome Program (SHGP) variant database (https://shgp.sa).^26–29

Assuming a common ancestor, we then applied genome-wide linkage analysis to three individuals, MIC-21, B44-21, and B54-23,³⁰ which revealed a common homozygous haplotype of 183 kb flanked by the SNPs rs13048349 and rs9979880 (chromosome 21: 41,774,299–41,957,463) encompassing the PRDM15 gene (Supplemental Figure 4A). This region also included the gene C2CD2 (Supplemental Figure 4B). Neither C2CD2 nor the genes RIPK4, ZBTB21, and UMODL1, which are flanking the shared ancestral haplotype, are known to cause a human phenotype that comprises features seen in the affected individuals in our cohort. Further, no suspected pathogenic variants in these genes were detected in the WES analysis for individuals B44-21, B53-21, and B54-23.

PRDM15 is a member of the PRDM family of zinc-finger proteins, which play important roles in transcriptional regulation and chromatin remodeling. It has been implicated in early embryo development, and knockout mice are embryonic lethal.²¹ Analysis of previously published, single-cell RNA-sequencing data demonstrate that although PRDM15 is expressed at low levels, it is predominantly found in the cap mesenchyme and podocytes in human fetal kidneys, and has peak expression levels at a gestational age of around 10 weeks. This is comparable with expression levels of other genes known to cause GAMOS (Supplemental Figure 5).^31–34 The Cys844 residue is a critical zinc-complexing cysteine residue within the protein’s 15th zinc-finger domain, and is highly conserved throughout evolution (Figure 1, D and E).

Variants in the PRDM15 SET Domain Can Cause Isolated Nephrotic Syndrome

We next queried WES data from approximately 1000 individuals with steroid-resistant nephrotic syndrome and identified two different homozygous missense variants in PRDM15 in two individuals from unrelated families, A3530-21 and A3714-24, with early-onset nephrotic syndrome (c.461T>A, p.Met154Lys; c.568G>A, p.Glu190Lys; Table 1) and in whom no other genetic causes of nephrotic syndrome were identified. Both variants are located within the protein’s SET domain, and are well-conserved evolutionarily and across other homologous PRDM proteins (Figure 1, D and E). Neither variant is present homozygously in gnomAD or in the SHGP variant database.^26–29 The affected individuals both presented with nephrotic syndrome early in life and demonstrated progressive renal insufficiency (Table 1). Neither demonstrated any extrarenal manifestations.

Pedigree structures for the four families with GAMOS-Mildenberger type and the two families with isolated nephrotic syndrome are depicted in Supplemental Figure 6. All affected individuals were from consanguineous unions. In the five families in whom WES data were available, we did not identify any pathogenic or likely pathogenic variants on the basis of American College of Medical Genetics guidelines in the 51 genes (Supplemental Table 3) previously reported to cause GAMOS or isolated nephrotic syndrome.³⁵

Finally, we performed high-throughput mutation analysis using multiplex PCR in 91 individuals with nephrotic syndrome and brain anomalies and 816 individuals with isolated nephrotic syndrome in whom WES data were not available. No additional biallelic variants in PRDM15 were identified. We additionally did not identify any biallelic variants in PRDM15 in 3029 exomes from individuals with congenital anomalies of the kidney and urinary tract, nephronophthisis, and their unaffected parents.

The Three PRDM15 Variants Disrupt the Protein’s Structural Stability

Because the protein structure of PRDM15 has not been resolved, we used the available x-ray crystal structures of human and mouse PRDM9, a homolog of PRDM15, to model the three human mutations identified in patients with GAMOS-Mildenberger type and isolated nephrotic syndrome (Figure 1, F–H, Supplemental Figure 7).^36,37 The amino acid residue Met154 in PRDM15 corresponds to Met316 in mouse PRDM9. It is part of the hydrophobic core of the protein, and is in proximity to the β10 strand within the SET domain, which is involved in S-adenosylhomocysteine (SAH) binding (Figure 1F). The p.Met154Lys substitution is predicted to alter the conformation of the SET domain, and thereby disrupt its ability to bind SAH.

The p.Glu190Lys change is also predicted to destabilize the SAH binding region. The PRDM15 Glu190 residue corresponds to Glu352 in mouse PRDM9, which forms several hydrogen bonds with residues on strands β10, β4, and β5, all of which are involved in SAH binding. Substitution of Glu190 in PRDM15 to a lysine residue will likely interfere with hydrogen bond formation (Figure 1G).

Finally, Cys844 in PRDM15 corresponds to Cys780 in human PRDM9, which is located within one of the protein’s zinc-finger domains, and represents one of the canonical zinc ligands. Mutation of this cysteine residue to tyrosine will disrupt its ability to complex zinc, and therefore destabilize the zinc-finger structure (Figure 1H).

Both SET Domain Variants Destabilize the PRDM15 Protein In Vitro

Because our in silico predictions suggest that the two PRDM15 SET domain variants disrupt the protein’s structural stability, we tested the thermal stability of the WT and two mutant PRDM15 proteins (p.Met154Lys and p.Glu190Lys). We overexpressed the SET domains of WT and mutant proteins in bacteria as His-tag fusions. Although the p.Met154Lys mutant could be purified using an Ni²⁺ resin (in a lower yield than WT), the p.Glu190Lys mutant protein was insoluble (Figure 2A). Furthermore, using a thermal stability assay measuring temperature-dependent differences in intrinsic tryptophan fluorescence, we demonstrated that the p.Met154Lys mutant was less stable when compared with WT (Figure 2B).

Figure 2.: The p.Met154Lys and p.Glu190Lys variants disrupt PRDM15 stability, whereas the p.Cys844Tyr mutation abrogates transcriptional activity. (A) Coomassie blue-stained gel showing purified WT and mutant (p.Met154Lys and p.Glu190Lys) PRDM15 recombinant SET domains expressed as His-tag fusions in E. coli. The p.Met154Lys mutant protein is not as strongly expressed when compared with WT. The p.Glu190Lys mutant was insoluble. (B) A thermal stability assay by tryptophan absorption was performed for the WT and p.Met154Lys mutant PRDM15 SET domains. The p.Met154Lys mutant was found to be significantly less stable when compared with WT. (C) Luciferase reporter assay demonstrating transcriptional activation after overexpression of WT PRDM15 and the two SET domain mutants, p.Met154Lys and p.Glu190Lys. This was significantly reduced with overexpression of the p.Cys844Tyr mutant protein.

The Zinc-Finger Variant, But Not the SET Domain Variants, Interfere with Transcriptional Activation

The role of PRDM15 as a transcriptional activator of the WNT modulator Rspo1 has been described previously.²¹ To assess whether the variants identified in our patients interfere with PRDM15 transcriptional activity, we performed luciferase assays in HEK293 cells. The promoter region of the Rspo1 gene containing the PRDM15 binding site was subcloned into a luciferase reporter vector, as previously described.²¹ HEK293 cells were cotransfected with empty vector, WT PRDM15, or mutant PRDM15, along with the Rspo1 luciferase reporter (firefly luciferase). Renilla luciferase was used as an internal control to normalize for transfection efficiency and signal intensity. Overexpression of the WT PRDM15 protein and the two SET domain mutants, p.Met154Lys and p.Glu190Lys, resulted in similar degrees of transcriptional activity. However, the p.Cys844Tyr mutant protein was unable to activate transcription (Figure 2C). This is consistent with our in silico data demonstrating that the p.Cys844Tyr mutation interferes with the protein’s ability to complex zinc, and thereby disrupts zinc-finger-DNA binding.

Prdm15 Is Required for Proper Pronephric Development in X. laevis

To analyze the potential role of prdm15 during early pronephric development in a model organism, we used the South African clawed frog X. laevis. Although Xenopus have a more primitive pronephric kidney compared with mammalian metanephric models, essential aspects of nephron development, such as patterning along the proximal-distal axis, are conserved.

We used whole-mount in situ hybridization to assess the specific expression pattern of prdm15 in the developing Xenopus pronephros at stages 33 and 36. Although the prdm15 sense probes gave no signal (data not shown), the antisense distribution was similar to the global pronephric-specific marker gene fxyd2 (γ-subunit of the Na⁺/K⁺-ATPase) (Figure 3, B–D).³⁸

Figure 3.: Prdm15 is required for glomerular and tubular development in X. laevis. Embryonic stages and scale bars are indicated in each panel and apply to all images included in the corresponding panel. (A) Illustration of the Xenopus pronephros (on the basis of Raciti et al. ⁵³ and Cizelsky et al. ⁵⁴) and its corresponding segment-specific marker genes. Overall: fxyd2; glomerulus: wt1; nephrostomes: lhx1; proximal tubule: foxc1 and slc5a1; intermediate tubule: slc12a1; intermediate, distal, and connecting tubule: clcnk; distal and connecting tubule: lhx1. (B) Whole-mount in situ hybridization (WMISH) against prdm15 in X. laevis at stage 33. Specific prdm15 expression was detected in the developing pronephric tissue (arrowhead). Left, lateral view with anterior toward the right; right, close up view of the pronephros. (C) prdm15 is expressed in the X. laevis pronephros at stage 36. Left, sagittal section of WMISH in X. laevis against prdm15 at stage 36 with anterior to the right; right, magnification of the proximal part of the pronephros. Arrowhead points to proximal tubule. (D) fxyd2 at stage 36 shows a specific expression in the pronephric tubule. Left, sagittal section of WMISH in X. laevis against fxyd2 with anterior to the right; right, magnification of the proximal part of the pronephros. Arrowhead points to proximal tubule. (E and F) Human PRDM15 rescues the MO-mediated knockdown of Prdm15. Injection of 15 ng Prdm15 MO leads to the reduction of pronephric-specific marker gene expression of wt1 (Wilms tumor suppressor gene-1) and fxyd2 (γ-subunit of the Na⁺/K⁺-ATPase) (arrowheads), whereas the injection of a control MO has no effect. Lateral views of injected sides with anterior to the right, transversal sections and quantitative representations are given. n, number of independent experiments; N, number of analyzed embryos in total. **P≤0.01. (G) Quantification of fxyd2 expression in the anterior pronephric area reveals a significant size reduction upon Prdm15 knockdown compared with control MO-injected embryos. Human PRDM15 rescues the renal phenotype. Left, lateral view with anterior toward the right (injected side); middle, appropriate expression area (red); right, quantitative representations. n, number of independent experiments. **P≤0.01; ****P<0.001. (H–L) Prdm15 loss-of-function results in reduced marker gene expression (arrowheads), whereas injection of a control MO has no effect. Lateral views of injected sides with anterior to the right, transversal sections and quantitative representations are given. n, number of independent experiments; N, number of analyzed embryos in total. *P≤0.05. clcnk, chloride channel protein; foxc1, forkhead box c1; lhx1, lim homeobox 1; slc5a1, solute carrier family 5 member 1; slc12a1, solute carrier family 12 member 1.

By utilizing the adequately controlled, MO-induced depletion approach, which allows reliable manipulation of protein abundance,³⁹ we examined the functional role of endogenous Prdm15 during Xenopus pronephric development. To test the efficiency of the Prdm15 MO, we cloned the binding site in front of, and in frame with, GFP. RNA coding for this construct was then injected together with either Prdm15 MO or control MO. Expression of GFP was reduced in Prdm15 MO-injected embryos compared with control MO-injected siblings (Supplemental Figure 8), indicating the functionality of the Prdm15 MO. However, the Prdm15 MO did not block translation of a construct in which the corresponding 5′UTR of human PRDM15 (Supplemental Figure 8) was cloned in front of GFP (Supplemental Figure 8), demonstrating that RNA coding for human PRDM15 is suitable for rescue experiments.

To investigate the functional relevance of Prdm15 in the embryonic kidney, we injected Prdm15 MO unilaterally in Xenopus embryos at the eight-cell stage, into one vegetal-ventral blastomere that will give rise to pronephric tissue. Prdm15 MO-injected embryos revealed a loss of the glomerulus-specific marker gene wt1 (Wilms tumor suppressor gene-1) and fxyd2 (Figure 3, E–F), as shown by whole-mount in situ hybridization. Quantification of fxyd2 expression in the anterior pronephric area confirms the significant size reduction of the pronephros at stage 36 upon Prdm15 knockdown (Figure 3G). Subsequently, our analyses showed that all regions of the pronephros were severely affected upon Prdm15 depletion, whereas a control MO injection had no effect on pronephros development (Figure 3, H–L). The coinjection of human full-length PRDM15 RNA partially rescued the expression of marker genes after Prdm15 depletion (Figure 3, E–G), indicating the specificity of the Prdm15 MO-induced phenotype and a conserved function of Prdm15 across species. In contrast, human PRDM15 RNAs reflecting the patient-specific point mutations (hPRDM15_M154K, hPRDM15_E190K, and hPRDM15_C8444Y) failed to rescue the Xenopus pronephros phenotype (Figure 4, A–C), demonstrating a disease relevance of these variants.

Figure 4.: The Prdm15 loss-of-function phenotype in X. laevis is rescued by human wild type, but not mutant, PRDM15. (A) Coinjection of human full-length PRDM15 with the variants identified in affected individuals (p.Met154Lys, p.Glu190Lys, and p.Cys844Tyr) fail to rescue the Prdm15 MO-mediated reduction of fxyd2 (whole pronephros) expression. Top, lateral view with anterior toward the right (injected side); bottom, appropriate expression area of fxyd2 (red). (B) Phenotypic quantification of fxyd2 expression comparing Prdm15 MO knockdown and coinjection with human WT and mutant PRDM15 RNA. n, number of independent experiments. N, number of analyzed embryos in total. **P≤0.01. (C) Measurement of the anterior pronephric area demonstrates a significant size reduction both upon Prdm15 knockdown and coinjection of human full-length PRDM15 with the variants identified in the affected individuals compared with WT PRDM15. n, number of independent experiments. *P≤0.05; ***P≤0.001. ns, not significant.

PRDM15 Depletion Disrupts Expression of Early Renal Developmental Genes

Because members of the PRDM family of zinc-finger proteins are involved in the regulation of transcription and chromatin structure, we next sought to determine the downstream targets of PRDM15 using RNA sequencing. We generated two CRISPR/Cas9-mediated PRDM15 knockout human podocyte cell lines (C7 and D12), using two different guide RNAs (Supplemental Figure 1). Cells transfected with a scramble guide RNA (A6) were used as controls. In the D12 cell line, there is a frameshift mutation, leading to early truncation of the PRDM15 protein. C7 harbors a splice site variant predicted to lead to aberrant splicing, and leads to knockdown of PRDM15 at the protein level, which was confirmed by Western blotting (Supplemental Figure 1C) and immunofluorescence (Supplemental Figure 1D). Baseline PRDM15 expression is low in the human podocyte cell line, but comparable with the expression levels of other genes known to cause GAMOS (Supplemental Table 4).

Total RNA from the control and two CRISPR/Cas9 knockout cells were submitted for RNA sequencing in triplicate. Principal component analysis revealed a clear separation of control (A6) from experiment (C7 and D12), and tight clustering of all triplicates (Figure 5A). Genes that were differentially regulated (adjusted P<0.05) are depicted in red on the volcano plots for C7 and D12 (Supplemental Figure 9). To exclude nonspecific transcriptional changes, we next identified shared genes that were differentially regulated in both PRDM15 knockout cell lines C7 and D12, with at least a two-times fold change, adjusted P<0.05, and average count of at least four. We found 71 genes that were upregulated and 80 genes that were downregulated upon PRDM15 loss (Figure 5B, Supplemental Table 5).

Gene ontology analysis of these 151 genes was performed, and 27 terms were found to reach statistical significance (Figure 5C). These were enriched for biologic processes involved in kidney development (blue) and more generally in cell proliferation and differentiation pertaining to embryonic development (green). This suggests that PRDM15 may play an important role in the embryonic processes governing renal development, and is consistent with the phenotype seen in our patients and the phenotype of Prdm15^−/− mice, which are embryonic lethal.²¹ Among the genes that are differentially regulated include several known to cause developmental kidney abnormalities when mutated in humans, including WT1, JAG1, and PAX2 (Figure 5D).^40–43 More recently, our laboratory identified mutations in TFCP2L1 as a novel cause of a renal tubulopathy, and Sema3a has been a reported cause of proteinuria in mouse models, although no human mutations have yet been identified.^44,45 The dysregulation of these disease-causing genes upon PRDM15 depletion was validated by quantitative PCR (Figure 5E).

Figure 5.: CRISPR/Cas9 knockout of PRDM15 results in dysregulation of multiple renal developmental genes. (A) Principal component analysis separates the A6 (scramble control) cells from CRISPR/Cas9 PRDM15 knockout human podocytes cell lines (C7 and D12). (B) Venn diagram depicting number of genes with P _Adj<0.05, average counts ≥4, and log₂ fold change ≤−1 (for downregulation, red) and log₂ fold change ≥1 (for upregulation, blue) in two PRDM15-depleted cell lines (C7 and D12). Eighty genes shared between C7 and D12 were downregulated, whereas 71 genes shared between C7 and D12 were upregulated. (C) Gene ontology analysis of the 151 differentially regulated genes that were shared between the C7 and D12 cell lines, when compared with the scramble control (A6). Gene ontology (GO) terms with a false discovery rate <0.05 are included. These clustered into three main groups: general cellular developmental processes (green), renal developmental pathways (blue), and genes involved in reproductive structure and placental development (red). (D) Heat map of the nine differentially regulated genes in GO:0072006 (nephron development), as well as two additional genes, TFCP2L1 and SEMA3A, both of which have been described to cause renal glomerular and/or tubular anomalies in mouse models. (E) Quantitative RT-PCR confirmation of the differential expression of WT1, SEMA3A, PAX2, and FOXC1 in PRDM15-depleted cell-lines, C7 and D12, displayed as dot plots with the center line representing the mean+SD. Expression levels were normalized to control cell line A6. P values were determined by a paired t test.

Discussion

Here, we describe mutations in the transcriptional regulator PRDM15 as a novel cause of both isolated nephrotic syndrome and GAMOS. We identified two variants in the SET domain, p.Met154Lys and p.Glu190Lys, in two individuals with early-onset steroid resistant nephrotic syndrome, and what we suspect to be a founder mutation, p.Cys844Tyr, in four individuals of Arab descent with features of GAMOS, including microcephaly, abnormalities on brain imaging, and a progressive proteinuric nephropathy. We further demonstrate that Prdm15 plays an important role for pronephros development in X. laevis, and that knockout of PRDM15 results in transcriptional dysregulation of several genes involved in renal glomerular and tubular development.

Our findings suggest that there is allelism to the PRDM15 variants, whereby the p.Cys844Tyr zinc-finger variant causes a multisystem developmental disorder, whereas the two variants in the SET domain lead to an isolated renal phenotype. The p.Cys844Tyr variant interferes with PRDM15 transcriptional activity, and therefore is likely to result in a complete null mutation, which may explain the more severe phenotype. Prdm15^−/− mice are embryonic lethal, and have been reported to demonstrate early postimplantation developmental defects, including brain and eye anomalies,^21,46 a finding consistent with the clinical features in the four individuals with GAMOS-Mildenberger type. However, the SET mutations do not alter PRDM15 transcriptional activity. Rather, our in vitro data demonstrate that they destabilize the SET domain. It is possible that the function of this domain is context-dependent, and thus destabilization leads primarily to an isolated renal phenotype. Future studies will be needed to better understand the mechanisms by which these allelic effects occur.

Prdm15 is required for pronephros development in X. laevis, as indicated by abnormalities in pronephros formation and in the downregulation of multiple pronephros markers upon morpholino knockdown, including wt1 and fxyd2. We additionally demonstrate, through RNA-sequencing experiments in human podocytes, that PRDM15 depletion leads to dysregulation of multiple genes that have been implicated in renal development, several of which, when mutated, can cause either nephrotic syndrome or renal malformations in humans. WT1, for example, is downregulated after PRDM15 knockout. It is a known regulator of podocyte development, and mutations in humans are an established cause of nephrotic syndrome, with invariable progression to ESKD.^40,47 In addition, mouse models have provided evidence that WT1-related glomerulopathy is mediated by activation of the Notch pathway.^48,49 Recently, Mzoughi et al.⁴⁶ demonstrated that PRDM15 is a key regulator of NOTCH signaling, and in our experiments, JAG1 is upregulated after PRDM15 depletion. Taken together, one potential mechanism by which PRDM15 mutations cause glomerular disease may be via downregulation of WT1 and subsequent dysregulation of the NOTCH signaling pathway.

Our studies also suggest that, in addition to its role in podocyte development, PRDM15 may also play a role in renal tubular development. This is supported by the finding that patients with the zinc-finger p.Cys844Tyr variant also have features consistent with a more generalized tubulopathy, including glycosuria and aminoaciduria. In addition, proximal, connective, and distal tubular markers are downregulated after Prdm15 knockdown in X. laevis, and multiple early markers of renal development are dysregulated in the two PRDM15 CRISPR/Cas9 cell lines. Among these, the transcription factor TFCP2L1 is of interest, because mutations in this gene have recently been reported to cause a novel renal tubulopathy in humans.⁴⁵Tfcp2l1-deficient mice have reduced maturation of the distal tubule and collecting duct, and also have early mortality, suspected to be secondary to renal failure.^50,51

A limitation of our study is that, because of the absence of available developmental podocyte cell lines and the early embryonic lethality of Prdm15^−/− mice, RNA-sequencing experiments were carried out using a mature podocyte cell model, and may not fully reflect the transcriptional programming that occurs during development. Future studies in organoid models or kidney-specific knockout animals will be important to more fully assess the role of PRDM15 during renal development. Furthermore, although our studies have focused on the renal manifestations of PRDM15-related disease, the syndromic features of the individuals with GAMOS-Mildenberger suggest that PRDM15 may play a broader role during embryogenesis, and, in particular, may play a role in brain, skeletal, and cardiac development. Indeed, it has been previously shown that loss of Prdm15 in mice can cause anterior/posterior patterning defects and brain malformations.⁴⁶ Further, PRDM15 has been demonstrated to be a key effector of NOTCH signaling, which has been implicated in cardiac development in humans and mice.⁵² Future studies will be needed to further delineate the full spectrum of the role of PRDM15 in development and disease.

Taken together, we demonstrate that mutations in PRDM15 are a novel cause of a progressive nephropathy, with clinical manifestations ranging from isolated nephrotic syndrome to a Galloway-Mowat type syndrome with mixed glomerular and tubular disease, microcephaly, and developmental delay. We demonstrate that PRDM15 depletion leads to abnormal pronephros development in Xenopus, and suggest that one pathway by which PRDM15 loss of function may cause kidney disease is via downregulation of WT1 and subsequent dysregulation of NOTCH signaling.

Disclosures

O. Gross reports consultancy agreements with Boehringer Ingelheim, Codon-X Therapeutics Inc., ONO Pharmaceutical Co. Ltd., Reata Pharmaceuticals, and Sanofi US Services Inc./Genzyme, Regulus Therapeutics; is a scientific advisor for, or holds membership with advisory committees of Boehringer Ingelheim, Reata Pharmaceuticals, and Sanofi US Services Inc./Genzyme, Regulus Therapeutics. F. Hildebrandt is a cofounder, scientific advisory committee member, and holds stock in Goldfinch-Bio. All remaining authors have nothing to disclose.

Funding

This research was supported by National Institutes of Health grant DK076683 (to F. Hildebrandt). M. Zenker was supported by the German Ministry of Education and Research (GeNeRARe, 01GM1519A). M.J. Schmeisser was supported by the Eliteprogramm of the Baden-Wuerttemberg Stiftung, the Care for Rare Foundation, the Eva Luise and Horst Köhler Foundation, the Else Kröner Fresenius Foundation (grant 2018_A78), the Volkswagen Foundation, and the German Ministry of Education and Research (GeNeRARe, 01GM1519A). E. Guccione was supported by NMRC/OFIRG/0032/2017 (funded by National Medical Research Council, NMRC, Singapore) and NRF-CRP17-2017-06 (funded by National Research Foundation, NRF, Singapore). N. Mann was supported by National Institutes of Health grant T32-DK007726-33 at Boston Children’s Hospital. V. Klämbt and T. Jobst-Schwan were supported by the German Research Foundation (VK 403877094 and Jo 1324/1-1). S. Shi and J. Kadlec were supported by the National Center for Scientific Research/French National Institute of Health and Medical Research ATIP-AVENIR start-up grant. A.C. Onuchic-Whitford is supported by the F32 Ruth L. Kirschstein National Research Service Award (DK122766). F. Buerger was supported by the German Research Foundation fellowship grant (404527522). T.M. Kitzler was supported by a postdoctoral fellowship award from the KRESCENT Program, a national kidney research training partnership of the Kidney Foundation of Canada, the Canadian Society of Nephrology, and the Canadian Institutes of Health Research. A.T. van der Ven was supported by a postdoctoral research fellowship from the German Research FoundationVE 969. A.J. Majmundar was supported by a National Institutes of Health training grant (T32DK-007726), the 2017 Postdoctoral Fellowship Grant from the Harvard Stem Cell Institute, and the American Society of Nephrology Lipps Research Program 2018 Polycystic Kidney Disease Foundation Jared J. Grantham Research Fellowship. We also thank the Yale Center for Mendelian Genomics for whole-exome sequencing analysis (grant U54HG006504). In addition, sequencing and analysis was also provided by the Broad Institute of MIT and Harvard Center for Mendelian Genomics, and was funded by the National Human Genome Research Institute, the National Eye Institute, and the National Heart, Lung and Blood Institute grant UM1 HG008900 (to Daniel MacArthur and Heidi Rehm).