TDP-43 regulates LC3ylation in neural tissue through ATG4B cryptic splicing inhibition

Abstract

Amyotrophic lateral sclerosis (ALS) is an adult-onset motor neuron disease with a mean survival time of three years. The 97% of the cases have TDP-43 nuclear depletion and cytoplasmic aggregation in motor neurons. TDP-43 prevents non-conserved cryptic exon splicing in certain genes, maintaining transcript stability, including ATG4B, which is crucial for autophagosome maturation and Microtubule-associated proteins 1A/1B light chain 3B (LC3B) homeostasis. In ALS mice (G93A), Atg4b depletion worsens survival rates and autophagy function. For the first time, we observed an elevation of LC3ylation in the CNS of both ALS patients and atg4b^-/- mouse spinal cords. Furthermore, LC3ylation modulates the distribution of ATG3 across membrane compartments. Antisense oligonucleotides (ASOs) targeting cryptic exon restore ATG4B mRNA in TARDBP knockdown cells. We further developed multi-target ASOs targeting TDP-43 binding sequences for a broader effect. Importantly, our ASO based in peptide-PMO conjugates show brain distribution post-IV administration, offering a non-invasive ASO-based treatment avenue for neurodegenerative diseases.

Keywords: ALS; Antisense oligonucleotides; Autophagy; Digital PCR; Post-translational modification.

Fig. 1

ATG4B function and expression is compromised in ALS. a ATG4B immunohistochemistry of human spinal cords demonstrates the enrichment of ATG4B in motor neurons. Left panel shows a higher magnification of a motor neuron in the anterior horn of the lumbar spinal cord of a healthy individual and an ALS patient. Right panel shows a lumbar spinal cord sections from anterior horn from CTL (n = 3) and ALS individuals (n = 3). Arrows point to motor neuron cells. Scale bars indicate 20 µm in length for high- (black labelled scale bar) and low- magnification (red labelled scale bar). b ATG4B protein expression analysis in human spinal cord lysates from healthy controls (n = 5) and ALS individuals (n = 9) does not show statistically significant differences. c Membrane-rich fractions of human spinal cords where this protein is active exhibit decreased levels in ALS (n = 5) compared with CTL (n = 4). Further, this membrane-rich fraction shows a low molecular weight protein, potentially derived from cryptic exon translation, indicated with an arrow. d ATG4B mRNA is downregulated in polysomes of TARDBP KD HeLa cells and more than 90% of the remaining ATG4B in mRNA contains cryptic exons in polysomes (active ribosomes) in TARDBP KD cells. Data from three experiments. e In line with decreased functionality of ATG4B, one of its substrates, LC3B, shows increased amount in high molecular weight conjugates (60 to 120 kDa) respect to free LC3B in ALS (n = 9) compared with CTL (n = 5) in human spinal cords. f Contrary to adult individuals, LC3ylation pattern is not influenced by Atg4b expression during development of mouse spinal cord. However, the loss of Atg4b in motor neurons triggers an accumulation of the autophagy substrate Sqstm1. HB9 is a motor neuron marker. GFAP is an astrocyte marker. Data from three experiment. g shRNA TARDBP transduction of human iPSC motor neurons efficiently reduces TARDBP mRNA, triggering higher levels of cryptic exon splicing in ATG4B transcript (n = 4) compared with scrambled transduction (n = 4). h Human iPSC motor neuron TARDBP KD (n = 4) express lower levels of ATG4B protein compared with SCR (n = 4). Data shown in graphs are mean values ± SD, *, **, *** and **** indicate, respectively, p < 0.05, p < 0.01, p < 0.001, p < 0.0001 for parametric unpaired two-sided Student’s t-test. Kolmogorov–Smirnov (distance) normality test was assessed. KD knockdown. CTL control samples. SCR scrambled control transduction. PSI percentage spliced-in. MTN Motor neuron

Fig. 2

ATG4B cryptic exon is expressed in blood cells but does not discriminate ALS cases. a dPCR results indicate a good partitioning in frontal cortex and WBC for both total ATG4B expression and cryptic exon of this transcript. ATG4B cryptic exon expression is higher in ALS frontal cortex (n = 1) than in control sample (n = 1). b In WBC, ATG4B expression (left panel), ATG4B cryptic exon expression (middle panel) and ATG4B cryptic exon PSI (right panel) are not different between CTL (n = 12) and ALS cases (n = 12 Data shown in bar graphs are mean values ± SD. PSI (Percentage Spliced-In). WBC (Whole Blood Cells). Mann Whitney test assessed for statistical analysis

Fig. 3

Atg4b expression is required for motor neuron survival in the murine model of motor neuron disease hSOD1-G93A mice. a Double transgenesis experiments show that G93A atg4b^−/− female mice (n = 20) have a shorter survival than G93A ones (n = 25). This effect is even greater for G93A atg4b^−/− male (n = 15) compared with G93A ones (n = 21). Kaplan-Meyer survival analyses. b Disease course as evidenced by weight loss related atrophy of both female (left panel) and male (right panel) is influenced by genotype. Data from females: WT (n = 13), atg4b^−/− (n = 12), G93A (n = 25), G93A atg4b^−/− (n = 20). Data from males: WT (n = 27), atg4b^−/− (n = 20), G93A (n = 18), G93A atg4b^−/− (n = 15). Two-Way ANOVA test; p-value for genotype (fixed effect type III). c atg4b^−/− genotype is associated to a higher Lc3b conjugated proteins, independently of G93A expression in LSC. Data from WT (n = 6), atg4b^−/− (n = 6), G93A (n = 8), G93A atg4b^−/− (n = 8). Bar graphs are mean values ± SD. *, and ** indicate, respectively, p < 0.05, p < 0.01 in One-way ANOVA test and Tukey corrected multiple comparisons. On the other hand, Atg4b deletion exacerbates autophagy impairment in LSC from G93A transgenic mice demonstrated by densitometric analyses of Sqstm1 western blot. Data from WT (n = 10), atg4b^−/− (n = 6), G93A (n = 21), G93A atg4b^−/− (n = 13). Bar graphs are mean values ± SD. *, **, and **** indicate, respectively, p < 0.05, p < 0.01, p < 0.0001 in Two-way ANOVA test and Tukey corrected multiple comparisons. d LC3ylation is also regulated by Atg4b in other central nervous system structures including cerebrum, cerebellum, brainstem and with less intensity in peripheral nervous tissue (sciatic). LC3ylation of peripheral organs such as heart, liver, and lung are not altered by Atg4b deletion. Data from three atg4b^−/− and three WT mice. Sciatic Nv indicates Sciatic nerve

Fig. 4

Atg4b is required for Lc3b lipidation in fibroblasts under autophagy inhibition. a LC3ylation is not dependent of Atg4b in mouse fibroblast. b Lc3b-II accumulation from autophagy inhibition after CQ treatment is not achieved in atg4b^−/− and G93A atg4b^−/− genetic background. Data from NT situation: WT (n = 4), G93A (n = 4), atg4b^−/− (n = 4) and G93A atg4b^−/− (n = 4); data from CQ treatment: WT (n = 4), G93A (n = 4), atg4b^−/− (n = 3) and G93A atg4b^−/− (n = 3); data from HBSS treatment: WT (n = 4), G93A (n = 4), atg4b^−/− (n = 3) and G93A atg4b^−/− (n = 4). c Sqstm1 is accumulated in G93A atg4b^−/− under autophagy induction by nutrient starvation with HBSS incubation. Data from NT situation: WT (n = 4), G93A (n = 4), atg4b^−/− (n = 4) and G93A atg4b^−/− (n = 4); data from CQ treatment: WT (n = 3), G93A (n = 3), atg4b^−/− (n = 3) and G93A atg4b^−/− (n = 3); data from HBSS treatment: WT (n = 3), G93A (n = 3), atg4b^−/− (n = 3) and G93A atg4b^−/− (n = 4) Bar graphs are mean values ± SD. * and ** indicate, respectively p < 0.05, and p < 0.01 in. Two-way ANOVA test and Tukey corrected multiple comparisons. NT (not treated), CQ (Chloroquine), HBSS (Hanks’ Balanced Salt Solution)

Fig. 5

ATG4B regulates ATG3 LC3ylation and its distribution across cell membranes. a SC detergent-resistant membranous fractions from atg4b^−/− mice contain specific high molecular weight bands of Lc3b and Atg3 in different compartments compatible with endosomes and MAMs. Data from a pool of two WT and two atg4b^−/− mice. b Immunoprecipitation experiment using spinal cord lysates from mouse confirms Atg3 as a target of LC3ylation. Data from one WT and one atg4b^−/− mice. c LC3ylated Atg3 (60 KDa) is accumulated in spinal cord from atg4b^−/− mouse and Atg3 (38 KDa) only increases in G93A compares with WT. Data from WT (n = 6), atg4b^−/− (n = 5), G93A (n = 7), G93A atg4b^−/− (n = 9). Two-way ANOVA test and Tukey corrected multiple comparisons. d TDP-43 regulates ATG3 LC3ylation by controlling ATG4B levels in HeLa cells. Data from NT (n = 4), pATG4B (n = 3), TARDBP KD (n = 5), TARDBP KD + pATG4B (n = 3). Two-way ANOVA test and Tukey corrected multiple comparisons. e LC3ylated ATG3 is increased in spinal cord homogenates from human ALS (n = 8) compared with CTL (n = 4). Two-way ANOVA test and Tukey corrected multiple comparisons. f Lc3b has many interactors in the spinal cord from atg4b^−/− (n = 2) compared with WT (n = 2). Lc3b co-IP proteins participate in translation, proteostasis and lipid metabolism, among others. Arrows indicate anti-Lc3b and anti-Atg3 blot indicates specific atg4b^−/− 60 KDa band. Bar graphs are mean values ± SD (± SEM in f. *, **, *** and **** indicate, respectively, p < 0.05, p < 0.01, p < 0.001, p < 0.0001 in Two-way ANOVA test and Tukey corrected multiple comparisons. pATG4B indicates plasmid for ATG4B overexpression. Red arrows indicate 60 KDa LC3ylated ATG3. Black arrows indicate 38 KDa unmodified ATG3. Yellow asterisk indicates 95 KDa LC3ylated Atg3 in the IP experiment. Red asterisk indicates 60 KDa LC3ylated Atg3 in the IP experiment

Fig. 6

ASOs targeting cryptic exon splicing rescue ATG4B expression. a In TARDBP KD HeLa cells, the use of a specific antisense oligonucleotide against ATG4B cryptic exons (pip8b2-ATG4B) employing PMO chemistry prevents ATG4B cryptic exon splicing and restores mRNA expression. Data from NT (n = 9), TARDBP KD (n = 9), TARDBP KD + pip8b2-ATG4B 10 µM (n = 6), TARDBP KD + pip8b2-ATG4B 20 µM (n = 3). Graphs are mean values ± SD. *, *** and **** indicate, respectively, p < 0.05, p < 0.001, p < 0.0001 in Two-way ANOVA test and Tukey corrected multiple comparisons. Two-way ANOVA test and Bonferroni corrected multiple comparisons. b Furthermore, this antisense oligonucleotide reaches brain tissue after IV injection (n = 3). Interestingly, though at 2 days signs of renal toxicity –evidenced by high Kim-1 levels- were present, this was normalized 7 days after the injection (n = 5). * represents p < 0.05 in One-way ANOVA Kruskal–Wallis test and Dunn’s corrected multiple comparisons for biodistribution analysis. In the case of Kim-1 levels, *** indicates p < 0.001 individual comparison in 24 h after Two-way ANOVA test and Bonferroni's multiple comparisons test. Graphs are mean values ± SD. *, and *** indicates, respectively, p < 0.05, p < 0.001. (c) The rescue effect of LNA mixmer ASOs is evident at lower concentrations (nM scale) targeting 3’ splice junction (3 J) and TDP-43 biding sequence (5) of ATG4B cryptic exon reducing cryptic exon splicing. ASOs do not alter TARDBP mRNA expression. Data from NT (n = 3), TARDBP KD + SCR (n = 3), TARDBP KD + 3 J (n = 3), TARDBP KD + 5 (n = 3) and TARDBP KD + 5U (n = 3). Graphs are mean values ± SD. ** and **** indicate, respectively, p < 0.01, p < 0.0001 in Two-way ANOVA test for ASO treatment effect. Red range indicates expression levels in TARDBP KD + SCR. Gray range indicates expression levels in NT

Fig. 7

ASO composed by repetitive CA sequence can restore the correct splicing of TDP-43 mini-gene reporter in TARDBP KD cells. a A dual reporter mini-gene assay of TDP-43 splicing function. b ASO CA can partially rescue the control situation in TARDBP KD HeLa cells. Data from 24 h post-transfection: NT (n = 204 cells), TARDBP KD + SCR 400 nM (n = 114 cells), TARDBP KD + ASO CA 200 nM (n = 116 cells), TARDBP KD + ASO CA 400 nM (n = 156 cells). Data from 48 h post-transfection: NT (n = 254 cells), TARDBP KD + SCR 400 nM (n = 147 cells), TARDBP KD + ASO CA 200 nM (n = 235 cells), TARDBP KD + ASO CA 400 nM (n = 196 cells). Graphs are mean values ± SD. ** indicates a minimum p < 0.01 in Two-way ANOVA test and Tukey corrected multiple comparisons for all combination in each post-transfection time. (c) Multitarget ASOs does not rescue cryptic exon splicing of selected genes. Data from NT (n = 3), TARDBP KD + SCR 100 nM (n = 3), TARDBP KD + ASO CA 100 nM (n = 3), TARDBP KD + ASO CATA 100 nM (n = 3), TARDBP KD + ASO ACNN (n = 3). Bar graphs are mean values ± SD. **, *** and **** indicate, respectively, p < 0.01, p < 0.001, p < 0.0001 in Two-way ANOVA test and Tukey corrected multiple comparisons