Biallelic loss-of-function variants in PLD1 cause congenital right-sided cardiac valve defects and neonatal cardiomyopathy

Abstract

Congenital heart disease is the most common type of birth defect, accounting for one-third of all congenital anomalies. Using whole-exome sequencing of 2718 patients with congenital heart disease and a search in GeneMatcher, we identified 30 patients from 21 unrelated families of different ancestries with biallelic phospholipase D1 (PLD1) variants who presented predominantly with congenital cardiac valve defects. We also associated recessive PLD1 variants with isolated neonatal cardiomyopathy. Furthermore, we established that p.I668F is a founder variant among Ashkenazi Jews (allele frequency of ~2%) and describe the phenotypic spectrum of PLD1-associated congenital heart defects. PLD1 missense variants were overrepresented in regions of the protein critical for catalytic activity, and, correspondingly, we observed a strong reduction in enzymatic activity for most of the mutant proteins in an enzymatic assay. Finally, we demonstrate that PLD1 inhibition decreased endothelial-mesenchymal transition, an established pivotal early step in valvulogenesis. In conclusion, our study provides a more detailed understanding of disease mechanisms and phenotypic expression associated with PLD1 loss of function.

Trial registration: ClinicalTrials.gov NCT01196182.

Keywords: Cardiology; Cardiovascular disease; Genetic diseases; Genetics; Heart failure.

Figure 1. Recessive variants in PLD1 cause a spectrum of valvular congenital heart disease and neonatal cardiomyopathy.

(A) Pedigrees. Solid symbols indicate the affected individuals (black indicates individuals with a congenital heart defect; red indicates individuals with fetal/neonatal cardiomyopathy), and symbols with a slash through them indicate deceased individuals. A double line indicates consanguinity. Males are indicated by squares and females by circles. Solid black triangles with a slash through them indicate fetal death or termination of pregnancy. Gray triangles indicate a miscarriage. Dup, duplication; ex, exon; fs, frame shift; NA, not available; Ter, termination; wk, age in weeks at termination of the pregnancy. ^#Previously published family (3). Families H–M were identified through analysis of PCGC data, and pedigrees were not available (8, 9). (B) PLD1 domain structure and location of pathologic (bottom) and presumptive nonpathologic (top) missense variants. Black indicates an inactive allele (16) used as a baseline control. Gray dots, homozygous missense variants found in gnomAD individuals (11). Statistical comparison of loop variants in patients versus controls was performed using Fisher’s exact test (P = 0.017). (C) Macroscopic appearance of the heart of fetus II-5 from family A. Upper left panel: Position of the heart in the thorax. Note the extremely dilated right ventricle with a thin translucent wall. A sharp demarcation can be seen between the abnormal right ventricular myocardium and normal left ventricular myocardium. Upper right panel: Anterior view of a formalin-fixed heart including large vessels, showing the paper-thin right ventricular wall, which was partially collapsed because of tissue weakness. Bottom panels: Postnatal echocardiograms of child II-1 from family A displaying a thin-walled right ventricle with Ebstein’s anomaly of the tricuspid valve and tricuspid regurgitation (see also Supplemental Video 1). This child also had pulmonary atresia. Ao, ascending aorta; Diaph, thoracic diaphragm; LA, left atrium; LAA, left atrial appendage; LV, left ventricle; PT, pulmonary trunk; RAA, right atrial appendage; RV, right ventricle; TV, tricuspid valve; TR, tricuspid regurgitation.

Figure 2. Activity of pathogenic PLD1 variants and their placement on the 3D protein structure.

(A) Activity of missense PLD1 variants above the K898R negative control as normalized to the WT PLD1 positive control. Data points represent independent experiments performed in duplicate and averaged. Error bars represent the SEM. Magenta, N-terminus; blue, PH-domain; yellow, catalytic domain; green, C-terminus N-term, N-terminus; C-term, C-terminus. (B) Placement of missense variants on the PLD1 catalytic domain crystal structure. α Helices are shown in cyan, and β strands are shown in magenta. Spheres indicate pathogenic missense mutations: green spheres indicate active site variants; orange spheres indicate PI(4,5)P₂–binding (PIP2-binding) mutants; and blue spheres indicate structural mutants. Insets show close-ups of WT residue interactions potentially disrupted by pathogenic variants. Pathogenic residues are represented by yellow sticks; black dashed lines indicate hydrogen bonds; and green dashed lines indicate cation-pi interaction. See Supplemental Videos 7–24 for rotatable presentations and a depiction of the impact of mutations on local structure.

Figure 3. PLD1 is required for EndoMT in AVC explants in vitro.

The role of PLD1 in EndoMT was determined using a collagen gel assay by incubating Hamburger-Hamilton (HH) stage 16 chick AV canal cushion explants on collagen gels for 48 hours with chemical (A), small-molecule inhibitors (B and C) of PLD, or an siRNA (D). The number of cells undergoing EndoMT per explant was counted for each condition over 3 separate experiments. (A) HH16 AVC explants were incubated with either 0.6% normal butanol (1-BuOH), which inhibits PLD generation of phosphatidic acid, or tertiary butanol (3-BuOH), which does not inhibit phosphatidic acid production. n = 30 explants for each condition. (B) Representative photomicrographs of AVC explants on collagen gels incubated with 5 μM doses of small-molecule inhibitors of PLD1i-a (VU0359595), PLD2i (VU0285655-1), and PLD1/2i (VU0155056, which inhibits both PLD1 and PLD2), or with DMSO. (C) HH16 AVC explants (n = 30 for each condition) were incubated with small-molecule PLD inhibitors or DMSO. PLD1/2i (VU0155056) inhibits both PLD1 and PLD2 (*P = 2.9 × 10^–17 compared with control); PLD1i-B (VU0155069) selectively inhibits PLD1 (*P = 3.9 × 10^–20 compared with control); PLD2i (VU0285655-1, *P = 0.28 compared with control); and PLD1i-a (VU0359595, *P = 1.4 × 10^–21 compared with control). (D) HH16 AVC explants were incubated with siRNAs targeting TGFBR3 (*P = 2.5 × 10^-6 compared with control), 2 different regions of PLD1-A (*P = 5.7 × 10^–3 compared with control) and PLD1-B (*P = 0.011 compared with control), or a scrambled siRNA (control). Control explants (n = 19) were GC content–matched, randomized siRNA constructs with no homology to any known chick gene. TGFBR3 (n = 23) was an siRNA targeting a gene known to be required for EndoMT in vitro. PLD1-A (n = 26) and PLD1-B (n = 25) were independent constructs targeting different regions of PLD1. All P values were calculated using the Student’s t test, and Bonferroni’s correction was used to correct for multiple testing. Error bars represent the SEM.