DGKE Variants Cause a Glomerular Microangiopathy That Mimics Membranoproliferative GN

The term membranoproliferative glomerulonephritis (MPGN) defines a heterogeneous group of kidney diseases that frequently lead to kidney failure.¹ The distinction of MPGN in three subtypes, based on different pathologic pictures, was recently replaced by a classification that accounts for the pathogenetic mechanisms underlying the diverse observed glomerular lesions. According to the new classification, MPGN is considered to fall in two etiological categories, in which either deposition of Ig or uncontrolled Ig-independent complement activation induces glomerular proliferative/exudative processes that result in mesangial expansion, hyperlobulated glomeruli, and, in the reparative phase, splitting (double contour or “tram track” appearance) of the glomerular basement membrane, thereby causing injury of podocytes, glomerular capillaries, and mesangial cells.² Familial occurrence has been described for both primary/idiopathic forms of MPGN^3–6 and for rare forms of thrombotic microangiopathy,^7–10 but little is known about the underlying genetic etiology. We performed homozygosity mapping and whole exome sequencing in a large index family from Turkey with four siblings affected with autosomal recessive disease with histologic signs of MPGN accompanied by prominent endothelial distress to identify genetic variants causative of this admixed phenotype. Participants with clinical and histological diagnosis of MPGN were ascertained and enrolled in the study after obtaining informed consent, in accordance with human participant research protocols approved by the Hacettepe University in Ankara (TBK08/1-57) and by the University of Texas Southwestern Medical Center in Dallas (112006-011).

The two healthy parents as well as two unaffected and three of the four affected individuals of UT-062 (Figure 1 A and C, Figure 2, and Table 1; see the Supplemental Material for a detailed description of the clinical history) were genotyped using the 250K Affymetrix single nucleotide polymorphism array. We identified three uninterrupted homozygous haplotypes that overlapped in all three affected individuals but not in the healthy siblings (Supplemental Figure 1A). Copy number variation analysis (Supplemental Figure 1B) excluded the presence of homozygous deletions within the identified homozygous regions in the affected siblings. The overall length of these regions was 12.3 Mb and included 1206 exons from 253 genes (Supplemental Table 1). Given the high number of exons in the regions, we performed whole exome capture followed by massive parallel sequencing on the DNA of two of the affected siblings (UT-062 V-2 and UT-062 V-4) to discover putative deleterious genetic variants (coverage data are reported in Supplemental Table 2). We identified a total of 30 coding single nucleotide variations within the three shared homozygous intervals, three of which were not annotated in the dbSNP database (build 131). Two of these variants were predicted to result in nonsynonymous changes in two residues of the gene G protein–regulated inducer of neurite outgrowth (GPRIN1) (p.Glu233Val and p.Gly236Val) that are poorly conserved in the evolution (Figure 1B and Supplemental Figure 2). Most importantly, this gene is exclusively expressed in neurons and no transcripts have been detected in the kidney,¹¹ excluding a causative role for the disease in this family. Furthermore, six loss-of-function variants (two frame shift and four missense mutations predicted to be deleterious by SIFT [sorting tolerant from intolerant] analysis¹²) have been found in GPRIN1 in 628 individuals sequenced in the 1000 Genomes Project¹³ and in the 6503 samples of the National Heart, Lung, and Blood Institute Exome Sequencing (NES) Project,¹⁴ indicating that this gene is weakly subject to evolutionary pressure. The third identified single nucleotide variation was a transition from C to T (c.127C>T) in the first coding exon of DGKE (NM_003647.2), a gene that encodes the ε isoform of diacylglycerol kinase,¹⁵ and creates a stop codon at glutamine 43 (p.Gln43X) that results in a predicted peptide missing all the functional domains (Figure 1C). The mutation was confirmed by Sanger sequencing and segregated in the pedigree with an autosomal recessive pattern. Because we performed homozygosity mapping,¹⁶ only missense variants within the homozygous identical by descent regions were tested for segregation. To support the causative role of DGKE mutations, we used exon sequencing to screen a worldwide cohort of 142 unrelated individuals diagnosed with MPGN. We found a homozygous deletion, c.610delA (p.Thr204GlnfsX6), in two affected siblings from an unrelated consanguineous Turkish family (HU-314) (Figure 1, A and C and Figure 2). The mutation causes a shift in the mRNA reading frame and is predicted to result in the creation of a stop codon at position 210 of the aberrant transcript. We reproduced the two mutations by site-directed mutagenesis in DGKE cDNAs cloned in myc-tagging expression vectors. After overexpression in HEK293T cells, our Western blot analysis detected a truncated protein of the predicted size obtained from the mutant c.610delA but not from the c.127C>T mutant (Figure 3A), suggesting that the truncated peptide is rapidly degraded. A third mutation (c.889-2A>G) that abolishes the obligatory acceptor splice site of intron 5 was found in the affected siblings of a family (HU-500) of Lebanese origin from Australia (Figure 1, A and C and Figure 2). We sequenced the DGKE cDNA of one of the affected siblings (V-1) and of the carrier healthy siblings (V-4) and found that intron 6 was retained in the presence of the homozygous change but not in the wild type (Figure 1D, Supplemental Figure 3), confirming that the mutation results in an abnormal transcript predicted to create a premature stop codon (amino acid 350, p.Trp350X). Interestingly, we could not amplify the mutated allele from the heterozygote carrier (Figure 1D), which suggested that the number of mutated transcripts is negligible due to non-sense–mediated decay. This hypothesis was confirmed by quantitative real-time PCR (Figure 1E). The mutations c.127C>T and c.889-2A>G were not detected in 332 chromosomes from 166 healthy participants from the same ethnicity, in the 628 individuals of the 1000 Genomes Project¹³ and in the 6503 samples of the NES Project.¹⁴ A heterozygous c.610delA change has recently been reported in one individual in the 1000 Genomes Project (added as rs147972030 to dbSNP134) but is not present in the NES cohort. Based on these two databases and on our controls, the minor allele frequency of this change is of 0.0002 (2 of 7297), indicating that this is an extremely rare variant, as is expected for mutations causing rare recessive diseases.¹⁷

Diacylglycerol kinases (DGKs) are intracellular lipid kinases that are devoted to phosphorylate diacylglycerol (DAG) to phosphatidic acid¹⁸ (Supplemental Figure 4). We determined by immunofluorescence confocal microscopy that Dgkε is expressed in adult mouse and rat kidneys and colocalizes with the podocyte marker WT1¹⁹ but not with the endothelial marker CD31 (Supplemental Figure 5A). Saturation of the antibody with molar excess of an unrelated Dgkε peptide proved the specificity of the immune reaction (Supplemental Figure 5B). We also detected DGKε by Western blot in human immortalized podocytes and rat kidneys and testes (Supplemental Figure 5, C and D), where the enzyme was previously reported to be expressed.¹⁸ DGKε is the smallest of the known mammalian DGKs and lacks extra-enzymatic regulatory domains,²⁰ a characteristic that suggests that this isoform is constitutively active. Another distinctive feature of DGKε is its marked substrate selectivity for DAG that is acylated with arachidonic acid at the sn-2 position (sn-2-arachidonoil-DAG). The selectivity for arachidonoil-DAG results in the enrichment with arachidonic acid of the phospholipids that participate to the phosphatidylinositol (PI) cycle and this effect is considered to be the main function of the enzyme.²¹ TRPC6 is a calcium-permeable cation channel expressed in the foot processes of podocytes, and is known to be directly activated by DAG through a protein kinase C–independent mechanism.²² To demonstrate that lack of DGKε affects DAG production in podocytes, we tested the effect of DGKE mutations on intracellular DAG concentration, taking advantage of the fact that currents through TRPC6 channels are directly regulated by DAG levels and that, as a consequence, their registration can be used as readout of intracellular DAG abundance.²³ We measured TRPC6 activity by ruptured whole-cell patch-clamp electrophysiological analysis in HEK293T cells cotransfected with TRPC6, M3 muscarinic receptor, and full-length or c.610delA mutant human DGKE. No detectable peptide was obtained from the c.127C>T mutant (Figure 3A). TRPC6 currents activated by the M3 receptor agonist carbachol were significantly reduced in cells transfected with wild-type DGKε (Figure 3, B–D), supporting the evidence that the enzyme metabolizes and decreases cellular DAG content. Consistent with the finding that the c.610delA mutation is loss of function, TRPC6 currents in cells expressing the mutant were not different from mock-transfected cells. Because HEK293 cells constitutively express DGKε, we also measured the activity of TRPC6 by electrophysiological analysis after suppressing DGKE expression with targeting small interfering RNAs (siRNAs) or with nontargeting siRNAs as a control (Figure 3E). In accordance with the previous experiment, TRPC6 currents were increased when DGKε expression was minimal (Figure 3, F–H), indicating that DGKε is required to maintain adequate cellular levels of DAG.

In this study, we have identified mutations in the podocyte expressed gene DGKE as the cause of a glomerular microangiopathy with histologic signs of both MPGN and endothelial distress in three unrelated families, by combining whole exome sequencing and homozygosity mapping.¹⁶ Homozygosity mapping is based on the remarkable finding that, for an affected individual born from consanguineous parents, the probability of carrying an autosomal recessive deleterious mutation in a genomic segment that is homozygous by descent is nearly equal to one,¹⁶ which confers a very high statistical power to this technique. We have shown that DGKE is expressed in podocytes and controls the intracellular concentration of DAG, a component of the PI cycle that participates in multiple cellular functions and in the lipid-mediated intracellular signaling.²⁴ Further experiments will be needed to verify if perturbation of this pathway might also affect the signaling between podocytes and endothelial cells, which would explain the vascular lesions observed in our patients. Three of the patients described in this work responded to immunosuppressive therapy (steroids and cyclosporine A or cyclophosphamide) with partial remission, alike a patient affected with diffuse mesangial sclerosis carrying a loss-of-function mutation in the gene PLCE1, which encodes for another enzyme of the PI cycle, (Supplemental Figure 4),²⁵ suggesting that the components of this metabolic and signaling pathway are of particular importance in the physiology of podocytes.