Snf2h-mediated chromatin organization and histone H1 dynamics govern cerebellar morphogenesis and neural maturation

Abstract

Chromatin compaction mediates progenitor to post-mitotic cell transitions and modulates gene expression programs, yet the mechanisms are poorly defined. Snf2h and Snf2l are ATP-dependent chromatin remodelling proteins that assemble, reposition and space nucleosomes, and are robustly expressed in the brain. Here we show that mice conditionally inactivated for Snf2h in neural progenitors have reduced levels of histone H1 and H2A variants that compromise chromatin fluidity and transcriptional programs within the developing cerebellum. Disorganized chromatin limits Purkinje and granule neuron progenitor expansion, resulting in abnormal post-natal foliation, while deregulated transcriptional programs contribute to altered neural maturation, motor dysfunction and death. However, mice survive to young adulthood, in part from Snf2l compensation that restores Engrailed-1 expression. Similarly, Purkinje-specific Snf2h ablation affects chromatin ultrastructure and dendritic arborization, but alters cognitive skills rather than motor control. Our studies reveal that Snf2h controls chromatin organization and histone H1 dynamics for the establishment of gene expression programs underlying cerebellar morphogenesis and neural maturation.

Figure 1. Snf2h and Snf2l are dynamically regulated in the developing cerebellum.

(a–d) In situ hybridization for Snf2l, Snf2h, En2 and En1 from wild-type (WT) embryonic (E) 14.5 sagittal sections adapted from Genepaint.org. Scale bar, 500 μm. (e) Boxed areas from a–d highlight the robust expression of Snf2h, En2 and En1 within the developing rhombic lip, while Snf2l is not detectable. A, anterior; P, posterior. Scale bar, 200 μm. (f) E17.5 WT sagittal cerebellar sections serially immunolabelled with Snf2h and Pan-Engrailed (En) antibodies. Brackets highlight regions of robust Snf2h and Engrailed immunoreactivity (−ir). Arrows denote robust Snf2h-ir within the EGL. DAPI labels all nuclei. Scale bar, 200 μm. (g) Confocal Z-stacks through the WT cerebellar vermis co-labelled with Snf2h (red) and Calbindin (green), a marker of the PC lineage at the indicated ages. Boxed areas are enlarged at bottom. Arrows denote Snf2h+ PCs. Asterisks denote Snf2h+ interneurons at P21. EGL, external granule layer; IGL, internal granule layer; PCL, Purkinje cell layer; o, outer; i, inner. Scale bars, 20 μm (top panels); 10 μm (bottom panels). (h) Confocal Z-stacks through the WT cerebellar vermis at post-natal day 3 (P3) co-labelled with Snf2h (red) and NeuroD1 (green), a marker of differentiated neurons. Arrows denote NeuroD1+, Snf2h+ PCs, as distinguished by their nuclear size and laminar position. Circles denote NeuroD1+, Snf2h+ granule cells (GCs) within the iEGL and the IGL. Scale bar, 20 μm. (i) Confocal Z-stacks through the P40 WT cerebellar vermis co-labelled with calbindin (green) and Snf2h (red, left panel); or Snf2l (red, right panel). Asterisks denote Snf2h+ interneurons. Scale bar, 5 μm. At least three mice from each genotype were used for evaluation. (j) Snf2h, Snf2l, En1 and En2 immunoblots of WT cerebellar extracts, except for E12 and E17 where hindbrain extracts were used. Actin served as loading control. (k) Plot of relative Snf2h, Snf2l, En1 or En2 expression during cerebellar development. The peak expression for each protein was normalized to 1, n=3.

Figure 2. Snf2h loss in cerebellar progenitors causes cerebellar ataxia or cognitive deficits when ablated in post-mitotic PCs.

(a) Amino-acid conservation of Snf2h exon 5 across species. The lysine residue within this ATP-binding motif and essential for catalytic activity is highlighted in red. Divergent amino acids are boxed grey. Xenopus, Xenopus laevis; Dros., Drosophila melanogaster; Sacc., Saccharomyces cerevisiae. (b) Schematic of the targeting strategy (LoxP sites, black arrowheads; Frt sites=white arrowheads) used to ablate Snf2h expression using the Nestin-Cre or PCP2-Cre drivers. (c) Plot of body weights from Snf2h cKO-Nes and control mice from P2 to P22. *P<0.05, one-way ANOVA, n=5–10. (d) Control (arrow) and Snf2h cKO-Nes (asterisk) littermates at P20. (e) Whole-mount images of Snf2h cKO-Nes and control cerebella at P40. (f) Snf2h immunoblot from Snf2h cKO-Nes and control cerebellar extracts at birth. Actin served as loading control. Values denote averaged densitometry, n=4. (g) Dowel, hanging wire and elevated platform tests reveal severe motor abnormalities in Snf2h cKO-Nes mice relative to control littermates at P20–P25. **P<0.01, one-way ANOVA, n=10–14. (h) Kaplan–Meier survival curves of Snf2h cKO-Nes, cDKO-Nes and control littermates. Snf2h cKO-Nes mice were not viable under standard laboratory conditions after ~P30–P45, while cDKO-Nes mice did not survive past birth, n=20–30. (i) Rotarod test: Snf2h cKO-PCP2 exhibit enhanced ability to stay on the rotarod after five sessions of training relative to controls. **P<0.01, one-way ANOVA, n=10–14. (j) Fear conditioning test: Snf2h cKO-PCP2 exhibit no differences in freezing response during training, but a decreased freezing response in context-dependent learning relative to controls. *P<0.05, one-way ANOVA, n=10–14. (k) Social interaction test: Snf2h cKO-PCP2 exhibit reduced interaction time with a stranger mouse in a controlled social environment relative to control littermates. *P<0.05, one-way ANOVA, n=10–14. Values are presented as the mean±s.e.m.

Figure 3. Snf2h loss results in intrinsic GNP cell death without altering Shh signalling.

(a) E17.5 sagittal cerebellar sections from Snf2h cKO-Nes and control littermates immunolabelled for phosphorylated Histone H3 (phospho-H3) or Ki67, and counterstained with the nuclear marker DAPI. Note a severe reduction of cells undergoing mitosis or S-phase throughout the mutant EGL (arrows). EGL, external granular layer. Scale bar, 100 μm. (b) Confocal Z-stacks through the E18.5 and P3 cerebellum from Snf2h cKO-Nes and control littermates immunolabelled for BrdU after a 90-min BrdU-pulse (arrows). Scale bars, 100 μm. (c) Quantification of BrdU+, Ki67+, phospho-H3+ or TUNEL+ cells throughout the mutant and control EGL at E17.5 and P0. **P<0.01, Student’s t-test, n=4. (d) EGL size at P7 and P14 in Snf2h cKO-Nes and control littermates. **P<0.01, student’s t-test. n.s., not significant, n=4. Values are presented as the mean±s.e.m. for c,d. (e) In situ hybridization through the cerebellar vermis from E18.5 Snf2h cKO-Nes and control littermates for Sonic Hedgehog (Shh), its receptor Patched-1, and their downstream target Gli-1. Note the spatiotemporal expression gradients (anterior high, posterior low) observed in both genotypes. Scale bar, 200 μm. Boxed areas are shown in rightmost panels to highlight robust mRNA levels in both genotypes. Scale bar, 50 μm. At least three mice from each genotype were used for evaluation.

Figure 4. Snf2h loss affects PC and GNP expansion resulting in cerebellar hypoplasia.

(a) Confocal Z-stacks from Snf2h cKO-Nes and control littermates that were BrdU-birthdated at E12.5, a time of PC birth, and co-labelled for BrdU (green) and Calbindin (red) at E17.5 (brackets). Note the reduction of BrdU+, Clabindin+PCs in mutant embryos (arrows). Scale bar, 50 μm. (b) Confocal Z-stacks from Snf2h cKO-Nes and control littermates that were BrdU-birthdated at E18.5, a time of robust GNP expansion, and co-labelled for BrdU (green) and Calbindin (red) at P7. DAPI (blue) stains all nuclei. Note the spatiotemporal (anterior low, posterior high) distribution of BrdU+ GCs throughout the internal granule layer in control cerebella that is altered in mutant brains. DCN, deep cerebellar nuclei. Scale bar, 100 μm. (c) Confocal Z-stacks through the cerebellar vermis from Snf2h cKO-Nes and control littermates that were BrdU-birthdated at E18.5, and labelled for BrdU (magenta); or co-labelled for Pax6 (red) and NeuN (green) at P7. DAPI (blue) stains all nuclei. Scale bar, 50 μm. At least three mice from each genotype were used for evaluation. (d) Quantification of double-labelled BrdU+ and Calbindin+ PCs at E17.5 (E12.5 BrdU birthdating); double-labelled BrdU+ and NeuN+ GCs at P7 (E18.5 BrdU-birthdating); or double-labelled BrdU+ and GFAP+ glial cells from mutant and control mice at P7 (E18.5 BrdU-birthdating). **P<0.01, Student’s t-test. n.s., not significant, n=3. Values are presented as the mean±s.e.m.

Figure 5. List of downregulated genes at P0 and corresponding gene expression changes at P10 from Snf2h cKO-Nes cerebella.

Cyan highlights TFs and yellow highlights cell adhesion molecules. Note that ~110 genes were deregulated at P0, while ~2,900 genes were deregulated by P10. Three microarrays per genotype were averaged from wild-type and mutant P0 and P10 cerebellar extracts. Underlined values denote P<0.05. Statistics was carried out using WEBARRAY online tool (http://www.webarraydb.org/webarray/index.html) that utilizes linear model statistical analysis (modified t-test).

Figure 6. Snf2h and Snf2l co-modulate the En1 locus.

(a) RT–qPCR analysis of selected genes from Snf2h cKO-Nes and control cerebella at P0 and P10. *P<0.05, Student’s t-test. n.s., not significant, n=6 from one of two independent experiments. (b) Immunoblots for Snf2h and Snf2l from Snf2h cKO-Nes and control cerebellar extracts at P9. Graph below depicts quantification for Snf2h and Snf2l, normalized to actin. **P<0.01, Student’s t-test, n=3. Values are presented as the mean±s.e.m. (c,d) Confocal Z-stacks of P7 cerebellum from Snf2h cKO-Nes and controls immunolabelled for (c) Snf2l. DCN, deep cerebellar nuclei. Scale bar, 200 μm; or (d) Pan-Engrailed (red) and counterstained with DAPI (green). Boxed areas are enlarged at bottom and denote similar En immunoreactivity in cells from both genotypes. Scale bars, 50 μm (top); 10 μm (bottom). (e) Confocal Z-stacks of P30 cerebellum from Snf2h cKO-PCP2 and controls immunolabelled with Pan-Engrailed (red) and Calbindin (green). Engrailed images in bottom panel are pseudocolored (silver) for contrast. Scale bar, 5 μm. Values (d,e) denote Engrailed immunopixels normalized to WT within cells with a nuclear size >10 μm² (circles). At least 50 PCs from three independent mice were quantified per genotype. (f) Kaplan–Meier curves of Snf2l KO, Snf2l−/+::Snf2h cKO-Nes and cDKO-Nes mice. Note that one Snf2l allele in a Snf2h-null background rescues the lethality of cDKO-Nes mice up to ~P25 (n=25). (g,h) Immunoblots for Snf2l, Snf2h, En2 and En1 from Snf2h cKO-Nes and control cerebellar extracts at P5 (g) and P20 (h). Actin served as loading control. Values denote averaged densitometry (n=4). (i) Top: schematic diagram of the mouse En1 locus indicating primer set locations. Yellow boxes, 5′ and 3′ untranslated regions (UTRs); striped yellow boxes, coding region; black line=non-coding region; TSS, transcription start site (arrow); kb, kilobases away from the TSS (+1). Bottom: ChIP-qPCR from WT cerebellar extracts for the En1 locus reveals Snf2h enrichment throughout the gene (R3, R2 and R1) at P7 and P21. Snf2l enrichment occurs only in the 5′-UTR (R2) at P21. *P<0.05, Student’s t-test, n=4 from one of three independent experiments. Values are presented as the mean±s.e.m. (j) Immunoblots for Snf2h, En2 and En1 from Snf2h^+/+::Snf2l^−/+ (WT); Snf2h^−/+::Snf2l^−/+ (Snf2h heterozygote); or Snf2h^+/+::Snf2l^−/y (Snf2l KO) cerebellar extracts at P9. Values denote averaged densitometry relative to WT levels (n=4).

Figure 7. Snf2h loss alters PC maturation.

(a) Snf2h (red) and Calbindin (green) co-labelling through the cerebellum from Snf2h cKO-Nes and control littermates at P7. Note the absence of Snf2h expression in mutant PCs. Scale bar, 20 μm. (b) Confocal Z-stacks through the cerebellar vermis from Snf2h cKO-Nes and control littermates at P7 immunolabelled for Calbindin (green). Roman numerals denote the corresponding lobules of the mammalian cerebellum. Scale bars, 100 μm (left panels); 20 μm (right panels). (c) Light microscopy Z-stacks of PCs through the cerebellar vermis stained with the Golgi–Cox method from Snf2h cKO-Nes mice and control littermates at P20. Scale bar, 10 μm. (d) Molecular layer size at P7 and P14 in Snf2h cKO-Nes and control littermates. (e) Quantification of pyknotic PC nuclei using toluidine blue reveals an ISWI-dependent modulation of PC survival in Snf2h cKO-Nes, Snf2l^−/+::Snf2h cKO-Nes and cDKO-Nes mice. (f,g) PC apical arbor dendritic length measurements from the indicated genotypes at P20 or P30, respectively. For panels (d–g) **P<0.01, Student’s t-test, n.s., not significant, n=6. Values are presented as the mean±s.e.m. (h) Left panels: confocal Z-stacks through the cerebellum from P30 Snf2h cKO-PCP2 mice and control littermates co-immunolabelled for Snf2h (red) and calbindin (green). Arrows and bars denote mutant PC arbors that do not reach the pial surface (ps) (bars, arrows). Middle panels: higher magnification images of PCs (arrowheads) denote Snf2h+ staining in control but not mutant PCs. Right panels: pseudocolored Calbindin-ir (silver) denotes the atrophied dendritic arbor in mutant PCs (bars, arrows). Scale bar, 20 μm (left and rightmost panels); 10 μm (middle panels). For a–c and h, at least three mice from each genotype were used for evaluation.

Figure 8. Snf2h is necessary for proper chromatin folding.

(a) Transmission electron microscopy (TEM) of GNPs and PCs through the cerebellar vermis from Snf2h cKO-Nes, cDKO-Nes mice and control mice at E18.5. Red circles denote nuclear invaginations. GNP, granule neuron progenitor; PC, Purkinje cell; Nu, nucleus; No, nucleolus. Scale bars, 2 μm. (b,c) TEM from Snf2h cKO-Nes and control mice through the cerebellum at P7 and P21. Note the abnormal morphology of mutant PCs and ‘cell ghosts’ in mutant cerebella. PC, Purkinje cell. Scale bars, 2 μm. (d,e) TEM through the cerebellar vermis from control and Snf2l−/+::Snf2h cKO-Nes mice at P7, revealing the aggravation of chromatin ultrastructure abnormalities within PCs upon additional removal of one Snf2l copy. Boxed area is enlarged in bottom rightmost panel. Scale bars, 10 μm (top panels); 2 μm (bottom panels). (f) TEM from P50 Snf2h cKO-PCP2 mice and control littermates through the cerebellar vermis. Higher magnification images are provided in descending panels. Note the loss of nucleolar, heterochromatin and euchromatin ultrastructure and increased electron density, evidenced as ‘chromatin clumps’ within mutant PC nuclei. Stars denote nucleolar dense fibrillar centers (dfc). gc, granular centre; fc, fibrillar centre; no, nucleolus; eu, euchromatin; hch, heterochromatin. Scale bars, 2 μm (top panels); 500 nm (middle and bottom panels). At least three mice from each genotype were used for evaluation.

Figure 9. Snf2h mediates chromatin transitions through linker histone H1 dynamics.

(a,b) Immunoblots of acid-extracted cerebellar histones from Snf2h cKO-Nes (Snf2h^−/−) and control littermates (Snf2h^−/+) for (a) histone H3 post-translational modifications at P9; or (b) histone variants H2A.Z and macro-H2A at P2 and P9. Values denote average densitometry relative to control samples (n=4). (c) Immunoblots of core histones from P2 and P9 cerebellar extracts from Snf2h cKO-Nes and control littermates. Values denote average densitometry relative to control samples (n=4). (d) Colloidal blue staining of isolated P2 and P9 histones from Snf2h cKO-Nes and control littermates to examine stoichiometry. (e) Mean normalized GFP-H1e FRAP curves of siScrambled (siScr); siSnf2h; siSnf2h+addback human (h) SNF2H; or siSnf2h+addback hSNF2L from transiently transfected mouse Neuro2A cells 48 h after treatment. A significant difference between recovery curves of siScrambled and siSnf2h is indicated. **P=0.008, n=20, Student’s t-test. Error bars were omitted for clarity. (f) FRAP images of GFP-H1e from siScr (top) or siSnf2h (bottom) Neuro2A treated cells at the indicated times. Scale bar, 1 μm. (g) Mobile fractions and t-half values for GFP-H1e FRAP experiments. **P=0.008, n=20, Student’s t-test. (h) Histone immunoblots of Neuro2A cells treated with siScr or siSnf2h for the indicated times. Snf2h KD is observed by 48 h. Actin served as loading control. Values denote average densitometry relative to siSnf2h 24 h treatment. (n=4). (i) Proposed model of Snf2h-dependent chromatin organization. Left, Snf2h interacts with the C-terminal tail of H2A to mediate histone H1 deposition and promote higher-order chromatin compaction and terminal differentiation. Right-top, Snf2h is required for normal progression through the cell cycle. Snf2h cKO-Nes mice have compromised expansion of the GNP and PC progenitor pools resulting in cerebellar atrophy. Right-bottom, after cell cycle exit, Snf2h- and Snf2l-dependent chromatin remodelling drives the establishment and maintenance of gene expression profiles. The co-regulation at the En1 locus is depicted as an example and this regulation promotes neural maturation. The embryonic removal of Snf2h (Nestin model) results in cerebellar hypoplasia and reduced dendritic arborization of PCs, causing severe ataxia and premature death. Similarly, Snf2h ablation in PCs (PCP2 model) also affects PC arborization, but conversely results in cognitive deficits rather than motor alterations.