TDP-43 proteinopathy in ALS is triggered by loss of ASRGL1 and associated with HML-2 expression

Abstract

TAR DNA-binding protein 43 (TDP-43) proteinopathy in brain cells is the hallmark of amyotrophic lateral sclerosis (ALS) but its cause remains elusive. Asparaginase-like-1 protein (ASRGL1) cleaves isoaspartates, which alter protein folding and susceptibility to proteolysis. ASRGL1 gene harbors a copy of the human endogenous retrovirus HML-2, whose overexpression contributes to ALS pathogenesis. Here we show that ASRGL1 expression was diminished in ALS brain samples by RNA sequencing, immunohistochemistry, and western blotting. TDP-43 and ASRGL1 colocalized in neurons but, in the absence of ASRGL1, TDP-43 aggregated in the cytoplasm. TDP-43 was found to be prone to isoaspartate formation and a substrate for ASRGL1. ASRGL1 silencing triggered accumulation of misfolded, fragmented, phosphorylated and mislocalized TDP-43 in cultured neurons and motor cortex of female mice. Overexpression of ASRGL1 restored neuronal viability. Overexpression of HML-2 led to ASRGL1 silencing. Loss of ASRGL1 leading to TDP-43 aggregation may be a critical mechanism in ALS pathophysiology.

Fig. 1. Decreased levels of ASRGL1 in ALS patients associate with TDP-43 proteinopathy.

a Volcano plot showing the differential expression analysis (DEA) of RNA sequencing data from brain samples of ALS individuals (n = 323) and normal controls (n = 68). Blue dots: genes upregulated in ALS; red dots: genes downregulated in ALS; gray dots: genes with no statistically significant differences (Differential expression analysis (DEA)). Differentially expressed genes (DEGs) were counted as genes with a False discovery rate (FDR) adjusted p value < 0.05). b, c, e ASRGL1 immunostaining of brain samples. Motor cortex samples (Broadman area 6) of ALS individuals (n = 4), multiple sclerosis individuals (MS) (n = 4), Alzheimer’s disease (AD) (n = 4), and normal controls (NC) (n = 4) were immunostained for ASRGL1. Visual cortex samples (Broadman area 17) from the same ALS individuals were also stained for ASRGL1 and consecutive slides of those samples were stained for TDP-43. ASRGL1+ cells were quantified by automated microscopy and presence of cytoplasmic TDP-43 was analyzed manually in a blinded manner. b Percentage of cells positive for ASRGL1 in FFPE motor cortex samples (BA6), (one-way ANOVA (Mean ± SEM)). c Percentage of ASRGL1+ cells in BA6 and BA17 of the same ALS individuals (two-sided unpaired T-test; Mean ± SEM). d Correlation of the percentage of ASRGL1+ cells and the percentage of cells with cytoplasmic TDP-43 in ALS samples (red dots: BA6; black dots: BA17) (Pearson’s r test; Middle line represents best fit line and error lines represent 95% confidence bands of the best fit line). e Representative images of ASRGL1 immunostaining in brain. f, g Pre-motor cortex samples (BA 6) (20 ALS and 20 normal controls) were analyzed by western blotting to measure ASRGL1 protein (Atlas antibody 029725). f Representative image of a western blot for ASRGL1 and the loading control vinculin. g Levels of ASRGL1 protein (ratio ASRGL1/Vinculin) in ALS brains analyzed by Mann–Whitney test (Median (IQR). h ASRGL1 RNA levels were analyzed in brain samples (BA6) from ALS patients (n = 19) and normal controls (n = 16) by qPCR (Mann–Whitney test (Median (IQR)). i ASRGL1 RNA levels were analyzed in differentiated motor neurons derived from iPSCs lines from ALS individuals (n = 6) and normal controls (n = 6), by qPCR (Mann–Whitney test (Median (IQR)). All pairwise comparisons in the Figure were performed with two-sided tests. Source data are provided as a Source Data file.

Fig. 2. Association of ASRGL1 loss and TDP-43 proteinopathy in brain samples of ALS patients.

a–f FFPE brain cortex samples from 3 pre-selected ALS individuals with very low levels of ASRGL1 and 3 normal controls were stained by immunofluorescence with ASRGL1 and TDP-43 antibodies. a Representative immunofluorescence images of brain cortex showing ASRGL1 and TDP-43 in an ALS individual and a normal control. b Percentage of neurons positive for ASRGL1 in normal controls and ALS brain samples. c Percentage of neurons showing cytoplasmic TDP-43. d Percentage of glial cells positive for ASRGL1 in controls and ALS patients. e Percentage of glial cells showing cytoplasmic TDP-43. Data in (b–e) analyzed by two-sided Unpaired T-test and represented as mean ± SEM. f Amplified image of brain cells of the same cortical brain sample of an ALS patient showing nuclear TDP-43 staining associated with perinuclear staining of ASRGL1 (upper panels) and loss of ASRGL1 associated with cytoplasmic TDP-43 (lower panels). Source data are provided as a Source Data file.

Fig. 3. TDP-43 is predicted to form isoaspartates through asparagine deamidation and TDP-43 interacts with ASRGL1.

a Spontaneous formation of isoaspartate by asparagine deamidation or aspartate isomerization. b Prediction of TDP-43 asparagine deamidation under physiological conditions using the NGOME algorithm. Top panel: Predicted deamidation half-lives (time point at which the residue is deaminated in 50%) considering sequence propensities (purple line) and sequence propensities + structural protection (green line). Orange circles highlight actual asparagine residues; the lines are the predictions for all positions of the query sequence. Middle panel: Structure protection factor (PF) as a function of residue number. Bottom panel: Percentage of deamidation predicted for asparagine residues after 2 days. c Example of 3D modeling of a TDP-43 peptide with isoaspartate substitution in a “hot spot” predicted by NGOME-Lite, showing the variation in the Gibbs free energy values, indicative of thermodynamic instability. d Diagram explaining Proximity ligation assay (PLA). e, f PLA was performed in HEK 293 cells transfected with ASRGL1 and TDP-43 plasmids and incubated with antibodies: no primary, only TDP-43, only ASRGL1, TDP-43 + IgG and ASRGL1 + TDP-43. e Representative PLA images of HEK 293 cells transfected with ASRGL1 and TDP-43 plasmids and incubated with ASRGL1 + TDP-43 antibodies and controls. f Comparison of percentage of PLA mean fluorescence intensity (PLA MFI) in relation with “No primary” control. (Brown-Forsythe ANOVA test/Bonferroni correction; Mean ± SEM; number of experimental replicates = 3). g Interaction of ASRGL-1 with TDP43 (recombinant proteins) using thermal shift assays. Tm value was calculated from the derivative of the thermal melt curves (Sigmoidal, Four-parameter logistic curve (4PL); Mean (95% CI); number of experimental replicates = 3). h, i PLA was performed in brain samples of ALS individuals (n = 3) and normal controls (n = 3) with TDP-43 and ASRGL1 antibodies. Negative controls were performed on a normal control brain since the expression of ASRGL1 is higher in normal controls. h Representative images of PLA in a normal control and an ALS patient brain stained with TDP-43 + ASRGL1 antibodies and the negative controls. i PLA MFI comparison (one-way ANOVA with Bonferroni correction (Mean ± SEM). All pairwise comparisons in the Figure were performed with two-sided tests. Source data are provided as a Source Data file.

Fig. 4. ASRGL1 silencing causes TDP-43 proteinopathy and neuronal death.

a, b IPSc-derived human neurons (normal donor) line stably expressing tdTomato protein and (c, d) Mouse neurons derived from a commercial neural stem cell line stably expressing GFP were transfected with ASRGL1 shRNAs or scrambled shRNAs. Fluorescent neurons were counted automatically after 48 h. a, c Representative fluorescence microscopy images. b, d Comparison of cellular viability (Percentage of counted cells/field in relation to scrambled shRNAs) (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 3). e, f Misfolded TDP-43 was analyzed in IPSc derived (normal donor) neuronal cultures co-transfected with an intrabody against misfolded TDP-43, full-length TDP-43, and ASRGL1 shRNAs or scrambled shRNAs, by co-immunoprecipitation and western blotting. Proteasome inhibitor MG132 was used as positive control. e Image of a TDP-43 blot. Due to the low levels of misfolded protein, blots were exposed longer. f Comparison of misfolded TDP-43 levels (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 7). g, h Neuronal cultures were transfected with TDP-43 and ASRGL1 shRNAs or scrambled shRNAs and analyzed by western blotting. g Image of a TDP-43 C-terminal fragments (CFT) blot. Due to the low levels of CFTs, blots were exposed longer. h Comparison of CTF levels (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 7). i–o Neuronal cultures were transfected with ASRGL1 shRNAs or scrambled shRNAs and analyzed by western blotting. i Image of a phosphorylated TDP-43 blot. j Comparison of phosphorylated TDP-43 levels (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 4). k Comparison of the percentage of neurons with cytoplasmic TDP-43 (two-sided unpaired T-test; Mean ± SEM. l Confocal microscope images of nuclear TDP-43 (arrows) and TDP-43 cytoplasmic inclusions (arrow heads) in neurons. m RNA levels of UNC13A cryptic exon, by qPCR. n Image of a UNC13A protein blot. o Comparison of UNC13A protein levels in percentage relative to scrambled shRNA transfected cells (Mann–Whitney test; Mean ± SEM; number of experimental replicates = 4). All pairwise comparisons in the Figure were performed with two-sided tests. Source data are provided as a Source Data file.

Fig. 5. ASRGL1 silencing impairs protein degradation.

a, b Neuronal cultures were transfected with ASRGL1 shRNAs or scrambled shRNAs. a Representative image of a western blot for ubiquitinated proteins. b Comparison of ubiquitinated proteins levels in percentage relative to the scrambled shRNA transfected cells (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 3)). c, d After transfection with scrambled shRNAs or ASRGL1 shRNAs, neuronal cultures were treated with cycloheximide (CHX) for 6 h. c Representative blot showing the levels of TDP-43 before and after 6 h of CHX. d Percentage of variation in TDP-43 levels after 6 h of CHX (number of experimental replicates = 3; Unpaired T-test; Mean ± SEM). e, f Neuronal cultures were transfected with ASRGL1 shRNAs or scrambled shRNAs. e Western blot of spliced XBP1 (sXBP1), unspliced (uXBP1) and vinculin in ASRGL1-silenced and control neurons. f Comparison of sXBP1 levels in percentage relative to the scrambled shRNA transfected cells (Mann–Whitney test; Median (IQR); number of experimental replicates = 4). All pairwise comparisons in the Figure were performed with two-sided tests. Source data are provided as a Source Data file.

Fig. 6. ASRGL1 rescues TDP-43-proteinopathy and associated neurotoxicity.

a, b iPSCs-derived human neuronal cultures from a normal donor were co-transfected with TDP-43, with shRNAs against ASRGL1 or scrambled shRNAs and with increasing concentrations of an ASRGL1-encoding plasmid or with pcDNA as a control. The number of cells presenting cytoplasmic TDP-43 was analyzed by immunofluorescence followed by confocal microscopy. a Percentage of neurons presenting cytoplasmic TDP-43 (one-way ANOVA with Bonferroni correction; Mean ± SEM; number of experimental replicates = 4). b Percentage of neuronal viability in relation to the control (pcDNA), as measured by flour spectrometry after applying Alamarblue (Thermo Fisher) (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 6). c, d IPSc-derived human neuronal cultures from a normal donor were co-transfected with wild type or isoaspartate containing C-terminal TDP-43 peptides, and with an ASRGL1 plasmid or an empty vector (pcDNA). c Percentage of neuronal viability in relation to the control (neurons transfected with pcDNA) (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 6). d Neurite length expressed as a percentage of the control (neurons transfected with pcDNA) (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 6). All pairwise comparisons in the Figure were performed with two-sided tests. Source data are provided as a Source Data file.

Fig. 7. ASRGL1 silencing triggers death of motor neurons and TDP-43 proteinopathy in mice.

Adult female C57BL/6 mice received a stereotaxic injection to deliver 2 µl of viral solution (10e13 AAV9 viral particles/ml) in the motor cortex. Half of the mice (n = 5) received particles with a construct encoding 4 scrambled shRNAs and the other half (n = 5) received particles with a construct encoding 4 shRNAs against ASRGL1 (sequences in Supplementary Table 5). One month after the injection, brain sections were immunostained and analyzed by confocal microscopy. a Schematic representation of the injection site in the motor cortex (coordinates: 1.5 mm lateral, 1.1 mm anterior, 1.6 mm ventral) and the shRNA construct delivered within the AAV9 viral particles. b Representative images of brains sections from an ASRGL1-silenced mouse and a control stained for NeuN and CTIP2. c Comparison of NeuN+ cells (neurons) in the motor cortex of ASRGL1-silenced mice and controls (Unpaired T-test; Mean ± SEM). d Comparison of CTIP2+ cells (motor neurons) in the motor cortex of ASRGL1-silenced mice and controls (Mann–Whitney; Median (IQR)). e Brain section of an ASRGL1-silenced mouse showing cytoplasmic staining of TDP-43 (arrow heads), compared to a control mouse showing nuclear staining of TDP-43. f Percentage of cells showing cytoplasmic TDP-43 in ASRGL1-silenced mice and controls (Unpaired T-test; Mean ± SEM). All pairwise comparisons in the Figure were performed with two-sided tests. Source data are provided as a Source Data file.

Fig. 8. ASRGL1 is downregulated by HML-2 and rescues HML-2-driven neurotoxicity.

a Diagram showing the location of an HML-2 intronic copy in ASRGL1 gene (Ensembl.org). b–d iPSCs-derived human neuronal cultures from a normal donor were transfected with an HML-2 plasmid or with an empty vector (pcDNA). b ASRGL1 RNA levels quantified by real-time qPCR (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 3). c Western blot for ASRGL1 and vinculin (loading control). d Comparison of ASRGL1 protein levels expressed as a percentage of the control (pcDNA) in neuronal cultures transfected with pcDNA (1 µg) or HML-2 (1 µg), by western blotting (number of experimental replicates = 4; Mann–Whitney test; Median (IQR)). e, f IPSc-derived human neuronal cultures from a normal donor were transfected with a CRISPR/dCAS9 construct with guide-RNAs (gRNAs) targeting the LTR to activate endogenous expression of HML-2 or without gRNAs. e ASRGL1 RNA levels, by qPCR (Unpaired T-test; Mean ± SEM). f Comparison of ASRGL1 protein levels expressed as a percentage of the control (no guide RNAs), by western blotting (number of experimental replicates = 4; Mann–Whitney test; Median (IQR)). g, h iPSCs-derived neural stem cells were treated with increasing concentrations of zidovudine (AZT). g Western blot for ASRGL1 and vinculin. h ASRGL1 protein levels expressed as a percentage of the control (no zidovudine) (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 3). i, j IPSc-derived human neuronal cultures from a normal donor stably expressing td-Tomato protein were transfected with a pcDNA plasmid, co-transfected with pcDNA and HML-2 or co-transfected with HML-2 and a plasmid encoding ASRGL1, and neuronal viability was measured by automated fluorescence microscopy. i Representative images of fluorescence microscopy. j Percentage of neuronal viability relative to the control (pcDNA transfected cells) (Brown-Forsythe ANOVA test with Bonferroni correction; Mean ± SEM; number of experimental replicates = 3). k, l RNA and proteins were extracted from brain cortex samples of ALS patients (n = 20) and normal controls (n = 19); the levels of HML-2 env RNA were analyzed by qPCR and the levels of ASRGL1 protein were analyzed by western blotting. k HML-2 RNA levels in ALS brains and normal controls, by qPCR (Mann–Whitney test; Median (IQR)). l Correlation between HML-2 RNA and ASRGL1 protein levels (ratio of ASRGL1 to Vinculin, as measured by western blotting) (Pearson’s r test; red dots: ALS; black dots: controls). The levels of ASRGL1 (ratio of ASRGL1 to Vinculin) in ALS and control brains are shown in Fig. 1f–h. All pairwise comparisons in the Figure were performed with two-sided tests. Source data are provided as a Source Data file.

Fig. 9. Schematic representation of potential processes triggered by loss of ASRGL1 resulting in TDP-43 proteinopathy in ALS.

(1) TDP-43 is involved in RNA transcription, splicing, stability, transport, and translation. (2) It is predominantly found in the cell nucleus but to perform its functions, TDP-43 is shuttled in its native form from the nucleus to the cytoplasm. (3) Isoaspartates (I) are formed spontaneously and non-enzymatically by deamidation of asparagine residues (N) of TDP-43. They introduce a kink in the peptide chain, favoring misfolding. Misfolded and fragmented TDP-43 are ubiquitinated for proteasome degradation. (4) Since proteases do not recognize isoaspartates residues, they must be cleaved from the peptide chain by ASRGL1 in the perinuclear region for the misfolded/fragmented protein to be (5) degraded by the proteasome. (6) The HML-2 copy is in the opposite strand of the ASRGL1 gene thus, the RNA of HML-2 is complementary to the RNA of ASRGL1 (HML-2 cDNA could also base pair with the ASRGL1 RNA). (7) HML-2 sequences could cause an antisense silencing on ASRGL1, interfering with its splicing, translation and/or activating the RNA-induced silencing complex. (8) If ASRGL1 is inactivated or depleted, isoaspartates residues cannot get cleaved and the aberrant TDP-43 forms (hyper-phosphorylated, misfolded and fragmented peptides) do not get degraded, (9) leading to a toxic aggregation in the cytoplasm. (Black arrows indicate physiological processes and red arrows indicate pathological processes).