Biallelic null variants in PNPLA8 cause microcephaly by reducing the number of basal radial glia

Abstract

Patatin-like phospholipase domain-containing lipase 8 (PNPLA8), one of the calcium-independent phospholipase A2 enzymes, is involved in various physiological processes through the maintenance of membrane phospholipids. Biallelic variants in PNPLA8 have been associated with a range of paediatric neurodegenerative disorders. However, the phenotypic spectrum, genotype-phenotype correlations and the underlying mechanisms are poorly understood. Here, we newly identified 14 individuals from 12 unrelated families with biallelic ultra-rare variants in PNPLA8 presenting with a wide phenotypic spectrum of clinical features. Analysis of the clinical features of current and previously reported individuals (25 affected individuals across 20 families) showed that PNPLA8-related neurological diseases manifest as a continuum ranging from variable developmental and/or degenerative epileptic-dyskinetic encephalopathy to childhood-onset neurodegeneration. We found that complete loss of PNPLA8 was associated with the more profound end of the spectrum, with congenital microcephaly. Using cerebral organoids generated from human induced pluripotent stem cells, we found that loss of PNPLA8 led to developmental defects by reducing the number of basal radial glial cells and upper-layer neurons. Spatial transcriptomics revealed that loss of PNPLA8 altered the fate specification of apical radial glial cells, as reflected by the enrichment of gene sets related to the cell cycle, basal radial glial cells and neural differentiation. Neural progenitor cells lacking PNPLA8 showed a reduced amount of lysophosphatidic acid, lysophosphatidylethanolamine and phosphatidic acid. The reduced number of basal radial glial cells in patient-derived cerebral organoids was rescued, in part, by the addition of lysophosphatidic acid. Our data suggest that PNPLA8 is crucial to meet phospholipid synthetic needs and to produce abundant basal radial glial cells in human brain development.

Keywords: brain organoid; developmental encephalopathy; iPLA2γ; outer radial glia.

Figure 1

Identification of biallelic loss-of-function variants in PNPLA8 in patients with diverse neurological phenotypes. (A) Pedigrees, showing autosomal recessive inheritance with biallelic PNPLA8 variants in 12 unrelated families. The probands are indicated by arrows. Filled symbols denote affected individuals; squares represent males and circles females. Double lines indicate first-cousin status. Pathogenic or likely pathogenic variants in PNPLA8 are denoted as ‘mut’, and WT sequences in PNPLA8 are represented as ‘wt’. (B–D) Clinical photographs of Patients 1, 6 and 8. (E–G) Axial T₁-weighted (E) and sagittal T₂-weighted (F) MRI images and a brain CT image (G) of Patient 1. The arrow indicates pontocerebellar hypoplasia. (H–J) T₁-weighted images of MRI (H and I) and a brain CT image (J) of Patient 2. The arrow indicates pontocerebellar hypoplasia. (K and L) Brain CT images of Patient 5. The arrow indicates cerebellar atrophy. (M and N) Axial T₁-weighted (M) and sagittal T₂-weighted (N) MRI images of Patient 6. The arrow indicates pontocerebellar hypoplasia. (O and P) T₂-weighted MRI images of Patient 7. The arrow indicates pontocerebellar hypoplasia. (Q and R) Axial T₁-weighted (Q) and sagittal T₂-weighted (R) MRI images of Patient 8. The arrow indicates pontocerebellar hypoplasia. (S and T) Axial T₂-weighted (S) and sagittal T₁-weighted (T) MRI images of Patient 9. The arrow indicates cerebellar atrophy. (U) T₁-weighted MRI images of Patient 10, with very mild cerebellar atrophy (arrow). (V and W) T₂-weighted MRI images of Patient 11c, exhibiting cerebellar atrophy (arrow). (X and Y) T₂-weighted MRI images of Patient 12, exhibiting cerebellar atrophy (arrow). Each patient is labelled as P-1 (Patient 1), P-2 (Patient 2), etc.

Figure 2

PNPLA8 protein isoforms, with the position of the identified variants. Schematic representation of PNPLA8 protein isoforms attributable to alternative translation initiation sites (top) and PNPLA8 exon locations reflecting the canonical full-length transcript (NM_001256007.3, bottom). Novel disease-associated variants (red) and previously reported variants (black) are noted. The variants identified from non-neuronal cohorts are noted in blue. Each family is labelled as F-1 (Family 1), F-2 (Family 2), etc.

Figure 3

Loss of PNPLA8 reduces the size of subventricular zone-like regions and the number of upper-layer neurons in induced pluripotent stem cell-derived cerebral organoids. (A) Schematic representation of the designed sequence of single guide RNA (sgRNA) next to protospacer adjacent motif (PAM) for PNPLA8. (B) Sanger sequencing of the CRISPR/Cas9-mediated homozygous nucleotide insertion in induced pluripotent stem cell (iPSC) lines. Altered sequences are indicated by an arrow and are highlighted in yellow. (C) Immunoblotting analysis of PNPLA8 using wild-type (WT) and PNPLA8 knockout (KO) iPSC lines. β-Actin was used as an internal protein-loading control. The arrow indicates the estimated 77 kDa PNPLA8 band. The fold change of PNPLA8 levels relative to β-actin, as quantified from protein bands (WT, KO-1 and KO-2), was 1.05 ± 0.07, 0.07 ± 0.07 and 0.04 ± 0.01, respectively (WT versus KO-1, P < 0.0001; WT versus KO-2, P < 0.0001), based on data obtained from four independent technical experiments. The asterisk indicates a non-specific band. Full-length blots are shown in Supplementary Fig. 9C. (D) Schematic illustration of the generation of cerebral organoids from PNPLA8 KO and WT iPSC lines. KOSR and B21 indicate knockout serum replacement and Brew 21, respectively. (E) Bright-field microscopy images of cerebral organoids at different developmental time points. Scale bars = 500 μm. (F and G) Representative immunofluorescence images of neural stem and progenitor cell (NPC) marker, SOX2, basal intermediate progenitor cell (bIP) marker, TBR2, and nuclear marker, DAPI, at 8 and 12 weeks of culture. ventricular zone-like regions (VZ) and subventricular zone-like regions (SVZ) are highlighted according to the spatial distribution of NPCs. Scale bars = 200 μm. (H and I) Quantification of the surface area of the VZ (H) and SVZ (I) at 8 and 12 weeks of culture. Average values ± SEM from the number of organoids in three independent experiments (at least three organoids per experiment) are plotted: WT at week 8 (n = 13); WT at week 12 (n = 14); KO-1 at week 8 (n = 12); KO-1 at week 12 (n = 17); KO-2 at week 8 (n = 12); KO-2 at week 12 (n = 16). *P < 0.05, **P < 0.01; ns = not significant. One-way ANOVA followed by Dunnett’s multiple comparisons test for each time point. (J) Representative immunofluorescence images of CTIP2⁺ deep-layer neurons and SATB2⁺ upper-layer neurons at 12 weeks of culture. Scale bar = 50 μm. (K and L) Quantification of the SATB2⁺ cells (K) and CTIP2⁺ cells (L) in the cortical plate-like region (CP). Average values ± SEM from three independent experiments (at least three organoids per experiment) are plotted: WT (n = 9); KO-1 (n = 9); KO-2 (n = 9). **P < 0.01; ns = not significant. One-way ANOVA followed by Dunnett’s multiple comparisons test.

Figure 4

There is a reduced number of basal radial glial cells in the subventricular zone-like regions of PNPLA8 knockout cerebral organoids. (A) Representative immunofluorescence images of PAX6⁺HOPX⁺ basal radial glial cells (bRGCs) at 12 weeks of culture. Scale bar = 100 μm. (B) Quantification of PAX6⁺HOPX⁺ bRGCs in a 100-μm-wide field of subventricular zone-like regions (SVZ). SVZ is highlighted according to the spatial distribution of neural stem and progenitor cells (NPCs). Average values ± SEM from three independent experiments (at least three organoids per experiment) are plotted: WT (n = 10); KO-1 (n = 11); KO-2 (n = 10). *P < 0.05, **P < 0.01, ***P < 0.001. One-way ANOVA followed by Dunnett’s multiple comparisons test. (C) Representative immunofluorescence images of SOX2⁺ NPCs and TBR2⁺ basal intermediate progenitor cells (bIPs) at 12 weeks of culture. Scale bar = 100 μm. (D) Quantification of TBR2⁺ bIPs in a 100-μm-wide field of SVZ. SVZ is highlighted according to the spatial distribution of NPCs. Average values ± SEM from three independent experiments (three organoids per experiment) are plotted: WT (n = 9); KO-1 (n = 9); KO-2 (n = 9). ns = not significant. One-way ANOVA followed by Dunnett’s multiple comparisons test. (E) Schematic illustrations of the labelling protocols for estimation of the cell cycle. Cerebral organoids were treated with bromodeoxyuridine (BrdU) for 24 h to identify cells undergoing the S phase. (F) Representative immunofluorescence images of BrdU⁺ cells and Ki67⁺ cells at 12 weeks of culture. Scale bar = 50 μm. (G and H) Quantification of Ki67⁺ cells (G) and BrdU⁺ cells (H) in the VZ. Average values ± SEM from three independent experiments (three organoids per experiment) are plotted: WT (n = 9); KO-1 (n = 9); KO-2 (n = 9). **P < 0.01; ns = not significant. One-way ANOVA followed by Dunnett’s multiple comparisons test.

Figure 5

A reduced number of basal radial glial cells and upper-layer neurons in Patient 1 induced pluripotent stem cell-derived cerebral organoids. (A) Schematic view of generating induced pluripotent stem cells (iPSCs) from peripheral blood from Patient 1 and a parental control. (B) Bright-field microscopy images of cerebral organoids at different developmental time points. Scale bars = 500 μm. (C and G) Representative immunofluorescence images of basal radial glial cells (bRGCs) (C) and basal intermediate progenitor cells (bIPs) (G) at 8 weeks of culture. Scale bars = 100 μm. (D and H) Quantification of bRGCs (D) and bIPs (H) in a 100-μm wide field of subventricular zone-like regions (SVZ). The SVZ is highlighted according to the spatial distribution of neural progenitor cells. Average values ± SEM from three independent experiments (at least three organoids per experiment) are plotted: bRGCs of Patient 1 (n = 10); bRGCs of control (n = 10); bIPs of Patient 1 (n = 10); bIPs of control (n = 10). ****P < 0.0001; ns = not significant. Student’s unpaired t-test with Welch’s correction. (E) Representative immunofluorescence images of HOPX⁺ cells in the ventricular zone-like region (VZ) at 8 weeks of culture. Scale bar = 50 µm. (F) Quantification of HOPX⁺ cells in the VZ. Average values ± SEM from three independent experiments (at least three organoids per experiment) are plotted. Control (n = 10); Patient 1 (n = 10); **P < 0.01. Student’s unpaired t-test with Welch’s correction. (I) Representative immunofluorescence images of CTIP2⁺ deep-layer neurons and SATB2⁺ upper-layer neurons at 8 weeks of culture. Scale bars = 50 μm. (J and K) Quantification of the proportion of SATB2⁺ (J) and CTIP2⁺ (K) cells in the cortical plate-like region (CP). Average values ± SEM from three independent experiments (at least three organoids per experiment) are plotted: Control (n = 9); Patient 1 (n = 10); ***P < 0.001; ns = not significant. Student’s unpaired t-test with Welch’s correction.

Figure 6

Spatially resolved differential gene expression profiles in the ventricular zone-like region of cerebral organoids. (A) Schematic overview of spatial transcriptomic analysis using the photo-isolation chemistry technique. (B) Representative immunofluorescence images of cerebral organoids at 8 weeks of culture. The ventricular zone-like region (VZ) and subventricular zone-like regions (SVZ) are visualized with PAX6 and HOPX. The VZ was irradiated with ultraviolet (UV) light where PAX6⁺ radial glial cells were densely packed. Merged images of immunofluorescence and the UV-irradiated area are shown (right-most panels). We extracted RNA samples from four distinct lines of cerebral organoids (n = 4). Each RNA sample was extracted from at least four VZs from at least two organoids. (C) Principal component analysis for the expression profiles. Principal component (PC)1 explained 43.7% and PC2 explained 27.2% of the variation. (D) Gene Ontology (GO)-term networks generated by BiNGO enrichment analysis for differentially expressed genes of patient apical radial glial cells. Node size represents the GO hierarchy. Yellow nodes indicate significant enrichment levels with P < 0.05. (E) Gene set enrichment analysis results ranked by normalized enrichment score (NES) using Curated gene sets and Hallmark gene sets of MSigDB signatures. Negative NES, depleted signature in Patient 1; positive NES, enriched signature in Patient 1. NES cut-off >1.5 or <−1.5. False-discovery rate q-value cut-off <0.05. (F) Enrichment plot for ‘Neocortex basal radial glia-up’, showing the profile of the running enrichment score and positions of gene set members on the rank-ordered list.

Figure 7

Loss of PNPLA8 affects phospholipid metabolism in neural progenitor cells, leading to reduced abundance of basal radial glial cells. (A) Gene set enrichment analysis results in the ventricular zone-like region (VZ) ranked by normalized enrichment score (NES) using the Pathway Interaction Database gene sets of MSigDB signatures. Positive NES, enriched signature in Patient 1. NES cut-off >1.5. False-discovery rate q-value cut-off <0.05. (B) Enrichment plot showing ‘lysophospholipid pathway’. (C) Heat map of the marker genes from the ‘lysophospholipid pathway’ gene set in the comparison of Patient 1 apical radial glial cells (aRGCs; left columns) versus control aRGCs (right columns). Expression values are represented as colours and range from red (highest expression) to pink (moderate), light blue (low) and dark blue (lowest expression). The core enrichment genes, which contribute most to the enrichment result of the gene set, were determined by the running enrichment scores. (D) Induced pluripotent stem cell-derived neural progenitor cells (NPCs) and culture supernatants were subjected to lipidomic analysis to quantify phospholipids (PLs), lysophospholipids (LPLs) and free fatty acids (FFAs). (E) Individual lipid species detected in lipid extracts from wild-type (WT) NPCs or PNPLA8 knockout (KO) NPCs, grouped by class. The log₂ fold change values of liquid chromatography–mass spectrometry/mass spectrometry profiles are shown. Data from PNPLA8 KO-1 and KO-2 NPCs are combined as ‘KO’. Bars represent the mean values for individual lipid species ± SEM for four independent experiments. Each dot indicates a single measurement of lysophosphatidic acid (LPA; n = 83), lysophosphatidylethanolamine (LPE; n = 47), lysophosphatidylcholine (LPC; n = 24), lysophosphatidylinositol (LPI; n = 27), phosphatidic acid (PA; n = 104), phosphatidylcholine (PC; n = 104), phosphatidylethanolamine (PE; n = 101), phosphatidylinositol (PI; n = 92), phoshatidylglycerol (PG; n = 104), phosphatidylserine (PS; n = 89), cardiolipin (CL; n = 280) and free fatty acids (FFA; n = 64). (F) Representative immunofluorescence images of HOPX⁺PAX6⁺ basal radial glial cells at 8 weeks of culture. Scale bars = 50 μm. (G) Quantification of the proportion of HOPX⁺PAX6⁺ cells in the subventricular zone-like region (SVZ). Average values ± SEM from three independent experiments (at least three organoids per experiment) are plotted: Control − LPA (n = 9); Control + LPA (n = 9); Patient−LPA (n = 11); Patient + LPA (n = 11); *P < 0.05, **P < 0.01, ****P < 0.0001; ns = not significant. One-way ANOVA followed by Dunnett’s multiple comparisons test.