Defective metabolic programming impairs early neuronal morphogenesis in neural cultures and an organoid model of Leigh syndrome

Abstract

Leigh syndrome (LS) is a severe manifestation of mitochondrial disease in children and is currently incurable. The lack of effective models hampers our understanding of the mechanisms underlying the neuronal pathology of LS. Using patient-derived induced pluripotent stem cells and CRISPR/Cas9 engineering, we developed a human model of LS caused by mutations in the complex IV assembly gene SURF1. Single-cell RNA-sequencing and multi-omics analysis revealed compromised neuronal morphogenesis in mutant neural cultures and brain organoids. The defects emerged at the level of neural progenitor cells (NPCs), which retained a glycolytic proliferative state that failed to instruct neuronal morphogenesis. LS NPCs carrying mutations in the complex I gene NDUFS4 recapitulated morphogenesis defects. SURF1 gene augmentation and PGC1A induction via bezafibrate treatment supported the metabolic programming of LS NPCs, leading to restored neuronal morphogenesis. Our findings provide mechanistic insights and suggest potential interventional strategies for a rare mitochondrial disease.

Fig. 1. Generation of an iPSC-based model of LS due to SURF1 mutations.

a Mitochondrial respiratory chain and putative SURF1 function in CIV assembly. MM mitochondrial membrane, IMM inner mitochondrial membrane, IMS intermembrane space. b Electropherograms showing 5′ > 3′ sequences of SURF1 in iPSCs from patients S1 and S2 carrying the mutations c.530T > G p.(V177G) and c.769G > A p.(G257R), respectively. Orange stars: mutation site. c Genome editing approach to correct mutation c.769G > A in patient line S2, and to introduce the same mutation in control line C1. Orange stars: mutation site; blue stars: correction site; black underlines: artificially introduced silent mutations; gRNA: guide RNA; HDR: homology direct repair. d Overview of iPSC lines used in this study. e Immunoblot of MT-CO2 (26 kDa) in NPCs derived from hESCs (H1), CTL (CTL_NoMut: C2, C1; SURF1_NoMut: S2_Corr1, S2_Corr2), and SURF1 (SURF1_Mut: S1, S2) (n = 2 independent experiments). f One dimensional (1D) blue-native gel electrophoresis (BNGE) analysis of mitochondrial complexes showing lack of incorporation of MT-CO2 into CIV in NPCs from CTL (CTL_NoMut: C1) and SURF1 (CTL_Mut: C1_Mut1). The structure of the complex II (CII) (visualized by SDHA antibody) was not affected (n = 2 independent experiments). g Two-dimensional (2D) BNGE showing mitochondrial complex assembly in NPCs from CTL (CTL_NoMut: C1) and SURF1 (CTL_Mut: C1_Mut1). Red arrows: structurally impaired CIV migrating at lower molecular weight (detected by MT-CO1 and COX4I1 antibodies); red arrowhead: accumulation of COX assembly intermediates (detected by MT-CO1 antibody); open red arrowheads: loss of III2 + IV supercomplex (detected by COX4I1 and UQCRC2 antibodies). We used CIII (detected by UQCRC2 antibody) and CII (detected by SDHB antibody) to align CTL blot with SURF1 blot (n = 2 independent experiments). h, i Quantification and representative images of COX activity intensity in CTL NPCs (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1, S2_Corr2) and SURF1 NPCs (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) (mean ± s.e.m.; each dot represents a biological replicate; n = 10 biological replicates per line over three independent experiments; ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). Scale bar: 500 nm.

Fig. 2. SURF1 mutations impair neuronal maturation in 2D and 3D cultures.

a 2D differentiation into mature differentiated neurons (DNs). We analyzed DNs at 4 weeks (4w) and 8 weeks (8w) starting from NPCs. b HCA quantification of TUJ1 + neurons in 4w and 8w DNs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1, S2_Corr2) and SURF1 (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) (mean ± s.e.m.; each dot represents a biological replicate; n = 20 biological replicates per line over three independent experiments; **p = 0.0013, ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). c, d Bioenergetics of 4w DNs and 8w DNs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1, S2_Corr2) and SURF1 (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) (mean ± s.e.m.; each dot represents a biological replicate; n = 20 biological replicates per line over three independent experiments; ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). e, f Sodium (below x axis) and potassium (above x axis) currents in 4w and 8w DNs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1, S2_Corr2) and SURF1 (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) (mean ± s.d.; n = 40 individual cells per line over three independent experiments; ***p < 0.001, ****p < 0.0001 CTL vs. SURF1; one-way ANOVA followed by Bonferroni multiple comparison test). g Electrophysiology traces in current-clamp recordings for 8w DNs from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2). Above: spiking activity; below: spontaneous postsynaptic activity; stars: glutamatergic postsynaptic currents. h 3D differentiation into cerebral organoids. We counted days of differentiation starting from embryoid bodies (EBs). i Cerebral organoids from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) at D40 and D90. Reproduced in CTL (CTL_NoMut: C1, C2) and SURF1 (SURF1_Mut: S1; CTL_Mut: C1_Mut1) (3–8 organoids per line per experiment, n = 3 independent experiments). Scale bars: 100 µm.

Fig. 3. Single-cell transcriptomics highlights an imbalance between proliferation and maturation in SURF1-mutant neural cultures and brain organoids.

a Uniform manifold approximation and projection (UMAP) plot showing the distribution of 4w DNs from CTL (CTL_NoMut: C1; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) (n = 2 independent experiments). b UMAP plot of 4w DNs from CTL (CTL_NoMut: C1; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) showing seven cellular clusters (resolution = 1) (n = 2 independent experiments). c Dot plot highlighting the expression of representative genes across clusters of DNs. Sizes of each dot reflect percentage of cells in the cluster where the gene is detected; colors reflect average expression level within each cluster (yellow: low expression; red: high expression). d Cell cycle distribution in 4w DNs from CTL (CTL_NoMut: C1; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) (n = 2 independent experiments). e UMAP plot of 4w DNs from CTL (CTL_NoMut: C1; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) showing cell cycle distribution (n = 2 independent experiments). f, g UMAP plots depicting signature distribution of cells expressing proliferative genes TOΠ2A and MΨX in 4w DNs from CTL (CTL_NoMut: C1; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) (n = 2 independent experiments). h UMAP plot showing distribution of D90 organoids from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) (3–8 organoids per line per experiment, n = 2 independent experiments). i UMAP plot of D90 organoids from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) showing 13 cellular clusters (resolution = 0.6) (3–8 organoids per line per experiment, n = 2 independent experiments). j Dot plot expression of representative genes across clusters of 90D organoids. k Cell cycle distribution in 90D organoids from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) (3–8 organoids per line per experiment, n = 2 independent experiments). l UMAP plot of 90D organoids from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) showing cell cycle distribution (38 organoids per line per experiment, n = 2 independent experiments). m, n UMAP plots depicting signature distribution of cells expressing proliferative genes TOΠ2A and MΨX in 90D organoids from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) (3–8 organoids per line per experiment, n = 2 independent experiments).

Fig. 4. Multi-omics analysis of SURF1 neural cultures identifies dysregulation of morphogens, metabolism/proliferation, and neuronal wiring.

a Schematics of multi-omics of 8w DNs from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2). b Multi-omics integration performed with xMWAS identified three communities. Genes are indicated by squares, proteins by circles, and metabolites by triangles. c MA plot showing log fold changes (LFC) of differentially expressed genes in 8w DNs from SURF1 (SURF1_Mut: S2) compared to CTL (SURF1_NoMut: S2_Corr1). Highlighted genes are shown using colors of their respective community (n = 3 independent experiments). d Volcano plot depicting differentially expressed proteins in 8w DNs from SURF1 (SURF1_Mut: S2) compared to CTL (SURF1_NoMut: S2_Corr1), after Benjamini–Hochberg correction for multiple testing. Highlighted proteins are shown using colors of their respective community (n = 3 independent experiments). e Volcano plot depicting differentially detected metabolites in 8w DNs from SURF1 (SURF1_Mut: S2) compared to CTL (SURF1_NoMut: S2_Corr1) after Benjamini–Hochberg correction for multiple testing (n = 3 independent experiments). 2FDP: D-fructose 2,6-bisphosphate; 2PG: 2-phosphoglyceric acid; GSSG: oxidized-glutathione; GSH: reduced-glutathione; NADP: nicotinamide adenine dinucleotide phosphate; NAA: N-acetyl aspartate; NAG: N-acetyl aspartyl glutamate.

Fig. 5. LS-associated defects in neuronal morphogenesis emerge at the level of NPCs.

a–c HCA masks and neuronal morphogenesis of 4w and 8w DNs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S1, S2: CTL_Mut: C1_Mut1) (mean ± s.e.m.; n = 20 biological replicates (dots) per line over three independent experiments; ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). Scale bar: 50 µm. d–h Bioenergetic profile and lactate release in NPCs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) (mean ± s.e.m.; n = 20 biological replicates (dots) per line over three independent experiments; ***p < 0.001, ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). i–k Bioenergetic profile and lactate release in fibroblasts from CTL (CTL_NoMut: C2, C3) and SURF1 (SURF1_Mut: S1, S2) (mean ± s.e.m.; n = 15 biological replicates (dots) per line (i, j) and n = 4 biological replicates (dots) per line (k) over two independent experiments; n.s. = not significant, **p < 0.01 CTL vs. SURF1; two-sided Mann–Whitney U test). l, m QRT-PCR in NPCs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) (normalized to AXTB; mean ± s.e.m.; n = 10 replicates per line over two independent experiments; ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). n, o 24 h proliferation and mtDNA quantification (normalized over genomic DNA, gDNA) in NPCs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) (mean ± s.e.m.; n = 20 biological replicates (dots) per line (n) and n = 4 biological replicates (dots) per line (o) over three independent experiments; ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). p–r HCA masks and neuronal morphogenesis in NPCs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S1, S2) (mean ± s.e.m.; n = 20 biological replicates (dots) per line over three independent experiments; ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). Scale bar: 50 µm. s–u Mitochondrial membrane potential (MMP) quantification and morphogenesis in NPCs from CTL (CTL_NoMut: C2, C3) and NDUFS4 (NDUFS4_Mut: NDU_1, NDU_2) (mean ± s.e.m.; n = 4 biological replicates (dots) per line over three independent experiments; ****p < 0.0001 CTL vs. NDUFS4; two-sided Mann–Whitney U test).

Fig. 6. Aberrant cytoarchitecture and neural progenitor organization in SURF1 brain organoids.

a MA plot showing LFC of differentially expressed genes in D90 brain organoids from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) (3–8 organoids per line per experiment, n = 3 independent experiments). Highlighted genes in community colors (orange: signaling and morphogens; blue: metabolism and cell cycle; green: axon guidance and synapsis). b, c Cerebral organoids from CTL (SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2) at D40 and D90. White arrows indicate SOX2-positive neuroepithelium surrounding embryonic-like ventricles; white arrowhead indicates abnormal structure of neuroepithelium layering. Reproduced in CTL (CTL_NoMut: C1, C2) and SURF1 (SURF1_Mut: S1; CTL_Mut: C1_Mut1) (3–8 organoids per line per experiment, n = 3 independent experiments). Scale bar: 100 µm. d Representative images and quantification of the size of D90 organoids from CTL (CTL_NoMut: C1, C2; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2; CTL_Mut: C1_Mut1) (mean ± s.e.m.; 20–30 organoids measured per line per experiment; n = 5 independent experiments; ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test). e–g Bioenergetic profile of D40 organoids from (CTL_NoMut: C1, C2; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2, C1_Mut1; CTL_Mut: C1_Mut1) (mean ± s.e.m.; 5–15 organoids per line per experiment; n = 2 independent experiments; ****p < 0.0001 CTL vs. SURF1; two-sided Mann–Whitney U test).

Fig. 7. SURF1 gene augmentation and bezafibrate treatment ameliorate bioenergetics and morphogenesis in patient neural cells.

a Bioenergetics of NPCs (SURF1_Mut: S1, S2) with mCitrine or WT-SURF1 (mean ± s.e.m.; n = 20 biological replicates (dots) per line over three independent experiments; ***p < 0.001; two-sided Mann–Whitney U test). b, c Morphogenesis and lactate in 4w DNs (SURF1_Mut: S1, S2) with mCitrine or WT-SURF1 (mean ± s.e.m.; n = 3 biological replicates (dots) per line over three independent experiments; **p = 0.0022, ****p < 0.001; two-sided Mann–Whitney U test). d Bioenergetics of NPCs (SURF1_Mut: S1, S2) with mCitrine or WT-SURF1 (mean ± s.e.m.; n = 5 biological replicates (dots) per line over two independent experiments; ****p < 0.0001; two-sided Mann–Whitney U test). e, f Morphogenesis of 4w DNs (SURF1_Mut: S1, S2) with mCitrine or WT-SURF1 (mean ± s.e.m.; n = 10 biological replicates (dots) per line over two independent experiments; *p = 0.0174, **p = 0.0061; two-sided Mann–Whitney U test). g–i PGC1A immunoblot and densitometry of NPCs from CTL (CTL_NoMut: C1, C2; SURF1_NoMut: S2_Corr2) and SURF1 (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) UT or with 400 µM BZ (mean ± s.e.m.; n = 2 independent experiments; *p < 0.05 SURF1-UT vs. CTL-UT, n.s  not significant CTL BZ vs. CTL-UT; **p < 0.01 SURF1 BZ vs. SURF1-UT; two-sided Mann–Whitney U test). j mtDNA quantification in NPCs from CTL (CTL_NoMut: C1, C2, C3; SURF1_NoMut: S2_Corr1) and SURF1 (SURF1_Mut: S2; CTL_Mut: C1_Mut1) UT or with 400 µM BZ (mean ± s.e.m.; n = 5 biological replicates (dots) per line over two independent experiments; n.s. = not significant CTL BZ vs. CTL-UT; *p < 0.05 SURF1 BZ vs. SURF1-UT; two-sided Mann–Whitney U test). k–m QRT-PCR (normalized to AXTB) and 24 h proliferation in NPCs (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) UT or with BZ (mean ± s.e.m.; n = 10 replicates (dots) per line over two independent experiments; *p < 0.05, ****p < 0.0001; two-sided Mann–Whitney U test). n–r Bioenergetics and morphogenesis of NPCs (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) UT or with BZ (mean ± s.e.m.; n = 15 biological replicates (dots) over three independent experiments; *p = 0.038, **p = 0.0069, ****p < 0.0001; two-sided Mann–Whitney U test). s, t Bioenergetics and morphogenesis of NPCs (SURF1_Mut: S1, S2; CTL_Mut: C1_Mut1) with mCitrine or PGC1A (mean ± s.e.m.; n = 10 biological replicates (dots) per line over two independent experiments; *p < 0.05; two-sided Mann–Whitney U test). Horizontal blue lines in all panels: average values CTL (CTL_NoMut: C1; SURF1_NoMut: S2_Corr1).