ZMYND11 functions in bimodal regulation of latent genes and brain-like splicing to safeguard corticogenesis

Abstract

Despite the numerous pathogenic variants linked to neurodevelopmental disorders (NDDs) including autism (ASD) and intellectual disability, our understanding of the underlying mechanisms caused by risk genes remain unclear. Here, we show that mutations in ZMYND11, a newly implicated risk gene, impair human cortical progenitor and neuron production. ZMYND11, known for its tumor suppressor function, encodes a histone-reader that recognizes sites of transcriptional elongation and acts as a co-repressor. ZMYND11-deficient cortical neural stem cells upregulate inappropriate developmental pathways, leading to disrupted neurogenesis. In addition to its role on chromatin, ZMYND11 regulates a brain-specific RNA isoform switch involving the splicing regulator RBFOX2. Similar defects are observed in other chromatin-related ASD risk genes, some of which are partially rescued by enhancing ZMYND11 function. These findings uncover convergent pathways linking chromatin regulation and splicing to human brain development and advance our understanding of how genetic risk contributes to NDD.

Fig. 1. Engineered human stem cell model of ZMYND11 NDD shows decreased production of intermediate progenitors (IPCs).

a Schematic outline of 3D cortical organoid differentiation protocol. LSBX: LDN193189 (BMPi), SB431542 (TGF-βi), XAV939 (WNTi). BAG: BDNF, Ascorbic Acid, GDNF. b–c Immunostaining for SOX2 (magenta), FOXG1 (red) and TBR2 (green) on cortical organoid section at day 30 ((b), n = 3). Zoomed-in view (c). Scale bars, 200 μm in (b), 100 μm in (c). d Immunostaining for FOXG1 (red) and PAX6 (green) on monolayer culture at day 18 (n = 3). Scale bars, 50 μm. e Flow cytometry analysis for TBR2 and FOXG1 in monolayer culture at day 16 with quantification relative to WT (mean ± SEM, n = 5 for +/+ and −/−, n = 9 for +/−). f Quantification of flow cytometry for TBR2 and FOXG1 in monolayer cultures at day 16 from Class I mutants (n = 3 for ASH1L, DEAF1, CUL3, ASXL3; n = 6 for RELN, KDM5B) and Class II mutants (n = 3 for CHD8, DYRK1A, KMT2A, SUV420H1) relative to isogenic control (UMOD edited lines, n = 15, mean ± SEM). g Schematic of doxycycline-inducible FLAG-ZMYND11 overexpression in ZMYND11 KO during cortical differentiation (200 ng/ml, white: d0-d16, light gray: d0–d7, dark gray: d8–d16). h Flow cytometry analysis for TBR2 and FOXG1 in monolayer cultures ±dox at day 16 with quantification relative to -dox (mean ± SEM, n = 4). i Flow cytometry analysis for TBR1 on FLAG- and FLAG+ populations at day 16 with quantification relative to the FLAG- population (mean ± SEM, n = 4). Red dotted line marks average WT levels. Statistics in (e, h): one-way ANOVA followed by Tukey’s test. Statistics in (i): unpaired two-tailed t test (two groups). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. P values in (e): +/− vs +/+, P < 0.0001; −/− vs +/+, P < 0.0001; −/− vs +/−, P = 0.0123. P values in (h): d0-d16 dox vs no dox, P < 0.0001; d0-d7 dox vs no dox, P = 0.0003; d8-d16 dox vs no dox, P = 0.0023. P values in (i): FLAG+ vs FLAG-, P < 0.0001. Source data are provided as a Source Data file.

Fig. 2. Transcriptomic analysis reveals elevated levels of BMP and WNT signaling pathways in ZMYND11 deficient NSCs, with BMP inhibition partially rescuing IPC production.

a PCA of RNA-seq on hESCs and NSCs (hESCs, n = 2 replicates for WT and −/−, n = 2 clones for +/−; NSCs, n = 4 for +/+ and −/−, n = 8 for +/−). b Overlap of differentially expressed genes in ZMYND11-deficient NSCs (P value < 1e-100). c Gene ontology analysis on upregulated genes in ZMYND11-deficient NSCs. Gray dots: P.adjust > 0.1. d Western blot on p-SMAD1/5/8 and SMAD1 on NSCs (from left to right: WT + / + , HET2 + /−, HET4 + /-, KO1 −/−) with quantification (mean ± SEM, n = 3 for +/+ and −/−, n = 6 for +/−). e Neuron induction schematic and immunostaining for MAP2 (green), SOX2 (red) (n = 3). Scale bars, 10 μm. f Flow cytometry analysis for CD133 after neuron induction at d25 with quantification (mean ± SEM, n = 4 for +/+ and −/−, n = 8 for +/−). g High-dose BMP inhibition strategy (LDN193189: 500 nM) and immunostaining for MAP2 (green), TBR1 (red) after neuron induction (n = 3). Scale bars, 10 μm. h Flow cytometry analysis for TBR2 and FOXG1 on monolayer culture after standard differentiation vs. high-dose BMP inhibition at day 16 with quantification relative to WT in standard differentiation (mean ± SEM, n = 3 for +/+ and −/−, n = 5 for +/− for standard differentiation; n = 6 for +/+ and −/−, n = 11 for +/− for high-dose BMP inhibition). Statistics in (d, f): one-way ANOVA followed by Tukey’s test. Statistics in (b): hypergeometric test. Statistics in (c): over-representation analysis with P.adjust values by Benjamini-Hochberg method. Statistics in (h): one-way ANOVA followed by Sidak’s test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. P values in (d): +/− vs +/+, P = 0.0164; −/− vs +/+, P < 0.0001; −/− vs +/−, P = 0.0015. P values in (f): +/− vs +/+, P < 0.0001; −/− vs +/+, P = 0.0006; −/− vs +/−, P = 0.0018. P values in (h): for standard vs high BMPi, +/+ vs +/+, P = 0.9991; +/− vs +/−, P < 0.0001; −/− vs −/−, P = 0.0485; among high BMPi, +/− vs +/+, P < 0.0001; −/− vs +/+, P < 0.0001; −/− vs +/−, P = 0.0075. Source data are provided as a Source Data file.

Fig. 3. ZMYND11 directly binds and transcriptionally represses BMP and WNT signaling-related genes.

a Normalized ZMYND11 binding and H3K36me3 enrichment in WT cortical NSCs from transcription start sites (TSS) to transcription end sites (TES) ± 5 kb (n = 2). b Correlation of ZMYND11 binding with H3K36me3 enrichment in WT NSCs. c Genomic features distribution, number of called peaks and bound genes of ZMYND11 binding in WT NSCs. d Overlap of ZMYND11 bound genes with H3K36me3 enriched genes (P value < 1e-100). e k-means clustering of ZMYND11 bound genes (High = 71, Mid = 1393, Low = 5676), corresponding H3K36me3 and RNA-seq expression changes. BMP and WNT signaling pathways are highlighted in red. f Representative track (PAX6) of ZMYND11 and H3K36me3 representing ZMYND11 High binding cluster (red mark). g Gene ontology analysis of 579 ZMYND11 targets upregulated in KO. Gray dots: P.adjust > 0.1. h Representative track (BMP7) of ZMYND11 and H3K36me3 representing ZMYND11 Low binding cluster (red mark). Statistics in (b): Spearman’s correlation analysis. Statistics in (d): hypergeometric test. Statistics in (g): over-representation analysis with P.adjust values by Benjamini-Hochberg method. Source data are provided as a Source Data file.

Fig. 4. ZMYND11 promotes a brain-specific mRNA isoform switch.

a Number of differential alternative splicing (DAS) events in ZMYND11 −/− NSCs (rMATS-HISAT2). A3SS, A5SS: Alternative 3’/5’ splice site. SE: Spliced exon. RI: Retained intron. MXE: Mutually exclusive exons. b PCA of exon usage (Percent Spliced In, PSI) (n = 4 for +/+ and −/−, n = 8 for +/−). c Overlap of DAS events predicted by HISAT2-rMATS, STAR-rMATS and AltAnalyze. d Schematic outlining the extraction of tissue-specific PSI for high-confidence DAS events. e Heatmap of tissue-specific PSI for high-confidence DAS events. Dividing line separates inclusion/exclusion events in non-brain tissues. KO NSC splicing was mapped on right bar (red/blue: inclusion/exclusion events; numbers show matched events). f Scatter plot of PSI differences between WT vs. KO and brain vs. non-brain tissues. Red/black marks significant (P < 0.05) or non-significant events, respectively. Closed/open circles show matched vs. unmatched KO and non-brain tissue trends. g PSI fold changes relative to WT for high-confidence DAS events (n = 4 for +/+ and −/−, n = 8 for +/−). Boxplot shows median, quartiles and whiskers (min/max) values. h, i. Representative high-confidence DAS events such as CTTN exon 11 (h) and SYNGAP1 exon 14 (i) with Sashimi plots (top), PSI changes (middle, mean ± SEM, n = 4 for +/+ and −/−, n = 8 for +/−) and semi-quantitative PCR (bottom, from left to right: WT + / + , HET2 + /-, HET4 + /-, KO1 −/−). Numbers in Sashimi plots indicate average PSI values. B: Brain. NB: Non-Brain. Statistics in (f): linear regression with correlation coefficient calculated using the Pearson method. Statistics in (g–i): one-way ANOVA followed by Tukey’s test for three groups, unpaired two-tailed t-test for two groups. ns P  >  0.05, *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. P values in (g): P < 0.0001. P values in (h): +/− vs +/+, P = 0.2034; −/− vs +/+, P < 0.0001; −/− vs +/−, P = 0.0003; NB vs B, P = 0.0044. P values in (i): +/− vs +/+, P = 0.0035; −/− vs +/+, P < 0.0001; −/− vs +/−, P = 0.0014; NB vs B, P < 0.0001. Source data are provided as a Source Data file.

Fig. 5. Non-brain-like isoforms alter migration and proliferation of cortical NSCs.

a Schematic outlining functional assays after specific isoform knockdown in long-term neural stem cells (LTNSCs). b Migration assays (0 h, 24 h, 48 h) on LTNSCs with different isoform knockdown and quantification of scratch coverage compared with shNT (n = 48 areas across 3 independent experiments). Dashed lines mark migration. Red marks significantly changed shRNA. Boxplot shows median, quartiles, and whiskers (min/max). NT non targeting. c BrdU-PI assays on LTNSCs with MEAF6 long (brain) or short (non-brain) isoform knockdown with quantification (mean ± SEM, n = 3). d Schematic outlining splicing factor screening and semi-quantitative PCR on CTTN exon 11 and MEAF6 exon 6 (n = 2). Red rectangles mark ZMYND11 shRNA, blue rectangles mark significantly changed shRNA. Percent non-brain isoform values shown below the gels. e Heatmap of non-brain isoform percent values (row Z-score transformation) for (d) using selected high-confidence DAS events matching non-brain tissues and ZMYND11 KO. APP events are combined due to proximity. Statistics in (b): one-way ANOVA followed by Dunnett’s test. Statistics in (c): one-way ANOVA followed by Tukey’s test. ns P  >  0.05, *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. P values in (b): 24 h: shL1CAM_e3-L vs shNT, P < 0.0001; shL1CAM_e3-S vs shNT, P = 0.9996; shL1CAM_e28-L vs shNT, P = 0.0087; shL1CAM_e28-S vs shNT, P = 0.3781; shCTTN-L vs shNT, P = 0.2084; shCTTN-S vs shNT, P = 0.9971; shMEAF6-L vs shNT, P = 0.9786; shMEAF6-S vs shNT, P < 0.0001; 48 h: shL1CAM_e3-L vs shNT, P < 0.0001; shL1CAM_e3-S vs shNT, P = 0.9996; shL1CAM_e28-L vs shNT, P = 0.0151; shL1CAM_e28-S vs shNT, P = 0.4839; shCTTN-L vs shNT, P = 0.6008; shCTTN-S vs shNT, P = 0.9888; shMEAF6-L vs shNT, P = 0.9575; shMEAF6-S vs shNT, P < 0.0001. P values in (c): shMEAF6-L vs shNT for G1%, P = 0.1074; shMEAF6-S vs shNT for G1%, P = 0.0012; shMEAF6-L vs shMEAF6-S for G1%, P = 0.0002; shMEAF6-L vs shNT for S%, P = 0.0543; shMEAF6-S vs shNT for S%, P = 0.0002; shMEAF6-L vs shMEAF6-S for S%, P < 0.0001; shMEAF6-L vs shNT for G2%, P = 0.3799; shMEAF6-S vs shNT for G2%, P = 0.0055; shMEAF6-L vs shMEAF6-S for G2%, P = 0.0015. Source data are provided as a Source Data file.

Fig. 6. Several ASD risk factors share impaired IPC production and non-brain-like alternative isoform switches, which can be improved by ZMYND11.

a Schematic outlining Class I mutant lines and cortical differentiation. b PCA of RNA-seq on Class I mutant and isogenic control cortical NSCs (n = 11 for Class I mutants, n = 4 for isogenic controls). c Gene ontology analysis on upregulated genes in Class I mutant NSCs using biological processes. d Overlap of DAS events prediction from HISAT2-rMATS, STAR-rMATS and AltAnalyze. e Scatter plot of PSI difference between isogenic control and Class I mutants (x axis) & brain tissues and non-brain tissues (y axis). f Sashimi plots for MEAF6 exon 6 and L1CAM exon 3 from high-confidence DAS events in Class I mutant NSCs. Numbers in Sashimi plots indicate average PSI values. g Semiquantitative PCR of L1CAM exon 3 on both Class I mutant and Class II mutant NSCs and quantification (mean ± SEM, n = 3). h Schematic outlining ZMYND11 overexpression strategy during Class I mutant differentiation. i Flow cytometry analysis for FLAG expression in dox-treated Class I mutant monolayer culture at day 16 (left) with quantification (right, 200 ng/ml, mean ± SEM, n = 4). j Quantification of flow cytometry analysis for TBR2 (left) and TBR1 (right) on FLAG+ populations at day 16 relative to FLAG- population (mean ± SEM, n = 4). Statistics in (c): over-representation analysis with P.adjust values by Benjamini-Hochberg method. Statistics in (e): linear regression with correlation coefficient calculated using Pearson method. Statistics in (j): unpaired two-tailed t test. ns P  >  0.05, *P < 0.05; **P < 0.01. P values in (j): TBR2 + % for ASH1L, P = 0.0042; for DEAF1, P = 0.2700; for ASXL3, P = 0.0456; for CUL3, P = 0.0998; for KDM5B, P = 0.0094. TBR1 + % for ASH1L, P = 0.0303; for DEAF1, P = 0.1519; for ASXL3, P = 0.0013; for CUL3, P = 0.1848; for KDM5B, P = 0.0338. Source data are provided as a Source Data file.