TDP-43 Promotes Neurodegeneration by Impairing Chromatin Remodeling

Abstract

Regulation of chromatin structure is critical for brain development and function. However, the involvement of chromatin dynamics in neurodegeneration is less well understood. Here we find, launching from Drosophila models of amyotrophic lateral sclerosis and frontotemporal dementia, that TDP-43 impairs the induction of multiple key stress genes required to protect from disease by reducing the recruitment of the chromatin remodeler Chd1 to chromatin. Chd1 depletion robustly enhances TDP-43-mediated neurodegeneration and promotes the formation of stress granules. Conversely, upregulation of Chd1 restores nucleosomal dynamics, promotes normal induction of protective stress genes, and rescues stress sensitivity of TDP-43-expressing animals. TDP-43-mediated impairments are conserved in mammalian cells, and, importantly, the human ortholog CHD2 physically interacts with TDP-43 and is strikingly reduced in level in temporal cortex of human patient tissue. These findings indicate that TDP-43-mediated neurodegeneration causes impaired chromatin dynamics that prevents appropriate expression of protective genes through compromised function of the chromatin remodeler Chd1/CHD2. Enhancing chromatin dynamics may be a treatment approach to amyotrophic lateral scleorosis (ALS)/frontotemporal dementia (FTD).

Keywords: CHD2; Chd1; Drosophila; H3K4me3; TDP-43; amyotrophic lateral sclerosis; chromatin remodeling; epigenetics; frontotemporal dementia; heat shock proteins.

Figure 1. H3K4me3-related chromatin factors modify TDP-43 toxicity

(A) Heat map of TDP-43 modifiers, ranked by their impact on the external eye of YH3-Gal4>UAS-TDP-43 flies. * denotes modifiers with functions related to H3K4me3. (B) Chd1 RNAi enhances and lid RNAi suppresses TDP-43-mediated neurodegeneration. Top panel: external eye images. Bottom panel: internal retinal structure. Arrow indicates retinal width. (C) Quantification of intact retinal tissue. Control n=5, TDP-43 n=14, lid n=6, Chd1 n=8. F(3, 19) = 111.8, p<0.0001 one-way ANOVA. n.s. p>0.05, **p<0.01, ***p<0.001 Tukey posthoc test. (D) Life span analysis of flies expressing TDP-43 and indicated RNAi with the neuronal driver elav3A-Gal4. Life span carried out at 29°C. TDP-43 expression shortens life span compared to flies expressing only control RNAi. Chd1 RNAi further shortens life span while lid RNAi prolongs life span. n=200 flies per group. p<0.001 logrank test. (E) Knockdown of Chd1 or lid in the nervous system without TDP-43 expression has minimal effects on life span. Conditions as in D. (F) lid and Set1 robustly modulate, but TDP-43 does not affect, brain H3K4me3 levels. F(3, 20) = 26.91) p<0.0001 one-way ANOVA. n.s. p>0.05, **p<0.01, ***p<0.001 Tukey posthoc test. Graphs denote mean ± s.e.m. Full genotypes are detailed in Table S2. See also Figure S1 and Table S1.

Figure 2. Chd1 depletion promotes stress granule formation

(A) Chd1 knockdown efficiency in S2R+ cells. (B) Experimental flow. S2R+ cells were treated with dsRNA for 4d, stressed with 0.2mM sodium arsenite (SA) for 2 hr, fixed, immunostained and imaged. For automated image analysis, images were deconvolved, max projected and automated detection of cells and stress granules was performed with CellProfiler. (C) Chd1 knockdown facilitates stress granule formation upon arsenite stress, but does not induce stress granules on its own. Red FMR1, Blue DAPI. Scale bars, 7.5 μm. (D) Chd1 knockdown enhances stress granule number per cell, stress granule size and the percent of cells that stain positive for stress granules. Control dsRNA n = 183 cells, Chd1 dsRNA n = 193 cells, 14 images per condition. **p<0.01, *p<0.05, Mann-Whitney test. SG, stress granule. See also Figure S3

Figure 3. Chd1 modulates stress response impairments caused by TDP-43

(A) Stress sensitivity assay. Flies were heat shocked 2 hrs at 40°C and allowed to recover over-night (O/N), after which the percentage of dead animals was scored. (B) Chd1-deficient flies are heat-stress sensitive. Heteroallelic Chd1 null flies (Chd1[4]/Chd1[5]), ubiquitous Chd1 knockdown flies, and controls were heat-shocked at 40°C for 2 hr. Wild-type n=6, Chd1 null n=6, control RNAi n=5, Chd1 RNAi n=5 biological replicates. **p<0.01, ***p<0.001 two tailed t-test. (C) Ubiquitous expression of TDP-43 in adult flies induces heat-stress hypersensitivity. Chd1 RNAi increases and overexpression of Chd1 reduces mortality of TDP-43 animals following stress. Control n=13, TDP-43 n= 23, TDP-43+Chd1 RNAi n=11, TDP-43+Chd1-OE n=13 biological replicates. F(3, 56) = 229.1, p<0.001 one-way ANOVA. **p<0.01 ***p<0.001 Tukey posthoc test. (D) Ubiquitous expression of wild type or R406W Tau does not promote heat stress-sensitivity. n=3 biological replicates. n.s p>0.05 two-tailed t-test. (E) Western immunoblot analysis of TDP-43 and Tau expression in whole flies. (F) Without additional stress, TDP-43 induces chaperone gene expression. qRT-PCR analysis of mRNA levels in whole flies. Following 10 min heat stress at 35°C, TDP-43 impairs the heat-shock response. mRNA levels shown as fold induction from no stress, normalized to RpS20. *p<0.05, **p<0.01, ***p<0.001, two tailed t-test. n=5 biological replicates. (G) TDP-43 impairs and Chd1 rescues heat shock gene expression. Flies were heat shocked for 30 min at 35°C. Hsp mRNA levels shown as fold induction from no stress, normalized to RpS20. Hsp40 F(2,11)=25.55, p<0.001, Hsp68 F(2,12)=23.9, p<0.001, Hsp70 loci F(2,12)=25, p<0.00, Hsp83 F(2,12)=21.35, p<0.001. *p<0.05, **p<0.01, ***p<0.001 One way ANOVA followed by Tukey’s post-hoc test. (H) TDP-43 impairs heat shock gene induction in mammalian cells. TetON-GFP and TetON-GFP-TDP-43 cells were induced with doxycycline for 6d after which heat shock was applied for 1 hr at 46°C, and RNA was extracted for qRT-PCR analysis at 6 and 8 hr of recovery. *p<0.05, ***p<0.001 two tailed t-test. n=6 biological replicates. Expression data are relative to the control non-heat shock samples (defined as 1) to allow direct comparison between control and TDP-43 samples between the different stress conditions. All graphs denote mean ± s.e.m. Full genotypes are detailed in Table S2. See also Figure S2.

Figure 4. TDP-43 impairs nucleosomal dynamics following stress

(A) MNase protection assay experimental design. (B) Control (non-heat shock), and heat shock (10 min at 35°C) flies were fixed and used for MNase protection. Results of real-time qPCR with primers spanning Hsp70 loci are shown as digested/undigested (total) chromatin. 2-way ANOVA revealed significant effect of group F(5, 388) = 572.6, p<0.001. *p<0.05, ***p<0.001 Tukey’s post-hoc test. n=6 biological replicates. Error bars show s.e.m. (C) Histone H3 ChIP experimental design. (D) Control (non-heat shock), and heat shock (10 min at 35°C) flies were fixed and used for total histone H3 ChIP. Results of real-time qPCR with primers spanning Hsp70 loci are shown as IP/input ratio. 2-way ANOVA revealed significant effect of group F(5, 231) = 289.0, p<0.001, ***p<0.001, Tukey’s post-hoc test. n=6 biological replicates. Error bars show s.e.m. Chd1-OE, Chd1 over expression. For (B) and (D) shaded areas show range of nucleosome-occupied gene body without stress, for comparison with heat shock graphs. (E) Flies expressing tagged Chd1-HA, with or without TDP-43, were subjected to heat shock at 35°C or control temperatures for 10 min, fixed and used for ChIP analysis. n=6 biological replicates. Results are shown as IP/input ratio. Hsp70(−190) F(3,20)= 13.35, p<0.001. Hsp70(−140) F(3,20)= 6.79, p<0.01. Hsp70(−25) F(3,20)= 45.18, p<0.001. Hsp70(+247) F(3,20)= 57.51, p<0.001. Hsp70(+280) F(3,20)= 63.97, p<0.001. Hsp70(583) F(3,20)= 53.1, p<0.001. Hsp70(872) F(3,20)= 14.89, p<0.001. Hsp70(1275) F(3,20)= 44.84, p<0.001. One way ANOVA followed by Tukey posthoc test. *p<0.05, **p<0.01, ***p<0.001, n.s. not significant. Error bars show s.e.m. X-axis labels describe location of reverse primer relative to the transcription start site. Full genotypes detailed in Table S2. See also Figures S4 and S5.

Figure 5. TDP-43 physically interacts with Drosophila Chd1 and human CHD2

(A) Human CHD1 and CHD2 are orthologues of Drosophila Chd1. (B) Adult Drosophila soluble and chromatin fractions were used for IP. IP samples were analyzed using western immunoblot with an anti-HA antibody to detect Chd1-HA. IgG IP serves as control. β-tubulin western immunoblot shows no significant interaction with TDP-43. (C, D) Human HEK293 cells were fractionated to soluble and chromatin fractions, and IP was performed with anti-TDP-43 antibodies (C), anti-CHD2 antibodies (D) or rabbit IgG as negative control. Note selective interactions of CHD2 and TDP-43 primarily in the soluble fraction. Full genotypes detailed in Table S2. See also Figure S5.

Figure 6. Abnormal TDP-43 accumulations promote loss of CHD2 in cultured cells and human FTD cortex

(A) CHD2 immunostaining in TetOn-GFP and TetOn-GFP-TDP-43 cells induced with doxycycline for 6d. (B) Quantification of (A), ***p<0.001 two tailed t-test, n=412 GFP cells, n=332 GFP-TDP-43 cells. CHD2 intensity is decreased. (C) Reduced CHD2 intensity in cells bearing abnormal cytoplasmic TDP-43 in TetOn-GFP-TDP-43 cells. (D) Quantification of (C), ***p<0.001 two tailed t-test, n=333 cells with nuclear TDP-43 localization, n=54 cells with cytoplasmic TDP-43 localization. (E) Double immunolabeling of TDP-43 (Green) and CHD2 (Red) reveals reduced levels of CHD2 in the FTD temporal cortex. (F) Quantification of CHD2 intensity in human post-mortem FTD temporal cortex. n=6 controls, n=8 FTD, p<0.001, two tailed t-test. See also Table S4. (G) Quantification of CHD2 intensity in FTD temporal cortex cells with nuclear or cytoplasmic TDP-43. ***p<0.001 two tailed t-test. (H) Model of TDP-43-mediated impairments in chromatin remodeling. See text for details. Scale bars 10 μm. Arrowheads in graphs show mean values. See also Figures S5 and S6 and Table S4.