Endogenous TDP-43 mislocalization in a novel knock-in mouse model reveals DNA repair impairment, inflammation, and neuronal senescence

Abstract

TDP-43 mislocalization and aggregation are key pathological features of motor neuron diseases (MND) including amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). However, transgenic hTDP-43 WT or ΔNLS-overexpression animal models mainly capture late-stages TDP-43 proteinopathy, and do not provide a complete understanding of early motor neuron-specific pathology during pre-symptomatic phases. We have now addressed this shortcoming by generating a new endogenous knock-in (KI) mouse model using a combination of CRISPR/Cas9 and FLEX Cre-switch strategy for the conditional expression of a mislocalized Tdp-43ΔNLS variant of mouse Tdp-43. This variant is either expressed conditionally in whole mice or specifically in the motor neurons. The mice exhibit loss of nuclear Tdp-43 concomitant with its cytosolic accumulation and aggregation in targeted cells, leading to increased DNA double-strand breaks (DSBs), signs of inflammation and DNA damage-associated cellular senescence. Notably, unlike WT Tdp43 which functionally interacts with Xrcc4 and DNA Ligase 4, the key DSB repair proteins in the non-homologous end-joining (NHEJ) pathway, the Tdp-43ΔNLS mutant sequesters them into cytosolic aggregates, exacerbating neuronal damage in mice brain. The mutant mice also exhibit myogenic degeneration in limb muscles and distinct motor deficits, consistent with the characteristics of MND. Our findings reveal progressive degenerative mechanisms in motor neurons expressing endogenous Tdp-43ΔNLS mutant, independent of TDP-43 overexpression or other confounding etiological factors. Thus, this unique Tdp-43 KI mouse model, which displays key molecular and phenotypic features of Tdp-43 proteinopathy, offers a significant opportunity to further characterize the early-stage progression of MND and also opens avenues for developing DNA repair-targeted approaches for treating TDP-43 pathology-linked neurodegenerative diseases.

Keywords: DNA double-strand break; TDP-43; amyotrophic lateral sclerosis; inflammation; motor deficits; motor neuron; muscle atrophy; neurodegeneration; senescence.

Fig. 1.. Inactivation of the TDP-43 nuclear localization signal (NLS) induces genomic instability.

(a), Schematic presentation of NLS point mutations (K95A, K97A, and R98A) in TDP-43. (b and c), Immunofluorescence (IF) analysis of TDP-43 localization and DNA double-strand breaks (DSB) in doxycycline-inducible TDP-43 wild-type (WT) and mutationally inactivated NLS (mNLS) neuronal cells using anti-TDP-43 (upper panel) and anti-γH2AX antibodies (lower panel) (b). Nuclei were stained with DAPI and cytoskeleton with Alexa-Flour 568 Phalloidin. Scale bars, 10 μm. Quantitative analysis of γH2AX foci in the nucleus (n = 35 cells in each experiment) (c). Data were analyzed using a t-test from two independent experiments (N = 2); mean ± SD, **P < 0.01. (d and e), Neutral comet assay showing TDP-43 mNLS expression-induced DSB accumulation in neuronal cells compared to TDP-43 WT cells. Scale bars, 20 μm (d). Quantitation of comet tail moments for each experimental group (n = 50 cells); one-way ANOVA; **P < 0.01 (e). (f and g), doxycycline-induced WT or mNLS TDP-43 expressing neuronal cells were transfected with antisense RNA (siRNA) to the 3'UTR of TARDBP mRNA or control siRNA and harvested at 72 h post-transfection for immunoblotting (IB) analysis using indicated antibodies (f). Quantitation of IB band intensities is expressed in fold-change from two independent experiments (N = 2); one-way ANOVA; **P < 0.01 (g). (h) Cell viability was compared between the TDP-43 WT and mNLS expressing cells using the MTT assay upon exposing cells to the stress granule inducer sodium arsenite (NaAs) at 0.5 mM concentration for 30 min, followed by recovery at indicated time points. Data were analyzed by one-way ANOVA from two independent experiments (N = 2); mean ± SD, *P < 0.05.

Fig. 2.. Generation of Tdp-43 knock-in mouse model.

(a) Illustration of the target allele design for the murine Tardbp gene. (b) Genotyping PCR identifies a wild-type and a heterozygous littermate where the band size of 351 bp indicates the presence of the floxed target allele. (c) Schematic of the double-heterozygous Cre^+/−:Tdp-43ΔNLS^+/− strain generation. (d) Illustration of the Cre-mediated recombination of floxed WT and mutant Exon-3 deletion and re-orientation, respectively, resulting in the expression of mutant Tdp-43ΔNLS variant in the desired cell type in the central nervous system (CNS).

Fig. 3.. Neuronal Tdp-43ΔNLS expression induces Tdp-43 mislocalization and formation of pathological aggregates in the cytosol

(a-b) Immunohistochemistry (IHC) with anti-TDP-43 antibody in the motor cortex of ALS mice Ubc-Cre::Tdp-43ΔNLS (whole body hence after; upper panel) and Mnx1-Cre::Tdp-43ΔNLS (neuron-specific hence after; lower panel). Scale bars, 20 μm (a). (b) Quantitation of the number of cells with Tdp-43 mislocalization phenotype in sham versus whole body or neuron-specific mice cortical tissues (N = 5 mice per group). (c) Representative IF images with anti-phosphoTdp-43 (S409/410) antibody in the motor cortex of sham versus whole body or neuron-ALS mice brains. Nuclei were counterstained with DAPI. Scale bars, 10 μm; N = 5 mice per group. (d-e) Thioflavin-S staining images of the cortical tissue from sham and ALS mice brains. Nuclei were counterstained with DAPI. Scale bars, 10 μm (d). (e) Quantitation of fluorescence intensity (arbitrary unit, a.u.) of Thioflavin-S-positive aggregates (N = 30 cells from cortical tissues of 5 mice in each group). (f) Representative Congo Red staining images of the cortex from sham and ALS mice. Pink stain indicates amyloid plaques in the cytosol and inter-cellular spaces. Nuclei were counterstained with hematoxylin. Scale bars, 10 μm. Data are expressed as mean ± SD and a nonparametric Mann-Whitney rank test was used for inter-group comparisons. ****P < 0.0001.

Fig. 4.. Tdp-43 mislocalization induces accumulation of NeuN transcription factor in the cytosolic protein aggregates

(a) IHC staining with anti-TDP-43 and anti-NeuN antibodies in the motor cortex of ALS and sham mice. Scale bars, 10 μm. (b) IHC staining with anti-TDP-43 and anti-NeuN antibodies in the thoracic region of the spinal cord cortex of ALS and sham mice. Scale bars, 10 μm. Nuclei were counterstained with NeuroTrace Nissl staining (435/455 nm). White arrowheads in the Nissl-stained images indicate morphologically degenerating neurons in the motor cortex and ventral horn of the thoracic spinal cord.

Fig. 5.. Motor neuron-specific Tdp-43ΔNLS expression causes muscle atrophy and gait deficits in Tdp-43 mutant mice.

(a) Representative live-mice images show abnormal hindlimb reflexes in 12-month-old ALS-Tdp-43 mice but not in sham-control mice. (b) Histopathological analysis of sham and ALS mice soleus (I-II) and spinal muscle tissues (III-VI). Hematoxylin-Eosin (H&E) staining displays degenerating muscle disc ALS mice compared to the sham mice muscles (I-II). IHC staining with anti-phosphorylated TDP-43 (S409/410) revealed the presence of pathological pTDP-43 in the spinal skeletal muscle of ALS mice (III-IV) and soleus muscles (V-VI). Scale bars, 50 μm. Inset images show cytosolic pTDP-43 staining in muscle cells. (c) DigiGait analysis indicates reduced stride length of hindlimbs in ALS compared to that in sham mice. (d) Hindlimb paw area (cm²) measurements exhibited a decrease in paw area in the right hindlimb while an opposite effect was observed for the hind-left limb than that in sham mice. (e) Stance-to-swing ratios were higher for both hindlimbs in ALS mice than that in sham mice. (f) ALS mice presented reduced gait symmetry, overall, when compared to those in sham mice. (g) Rotarod testing showed a gradually decreasing trend in latency to fall for ALS mice compared to age-matched sham mice. Data are expressed as mean ± standard deviation (SD) and a nonparametric Mann-Whitney rank test was used for inter-group comparisons. **P < 0.01.

Fig. 6.. Tdp-43ΔNLS induces genome damage in the central nervous system (CNS).

(a-b) Immunoblotting (IB) of cortical brain extracts from sham (N = 3) and whole boy Tdp-43 KI mice (N = 6) using anti-γH2ax. Gapdh served as the loading control (a). (b) Quantitation of relative band intensity of the ALS mice group compared to that of the sham mice group. (c-d) IHC analysis of γH2ax expressions in the cortex of the whole body and neuron-specific Tdp-43ΔNLS expressing mice. Nuclei were counterstained with hematoxylin (c). Scale bars: 20 μm. (d) Quantitation of the number of cells with >5 foci of γH2ax in the nucleus. (e-f) TUNEL analysis to estimate the neuronal genome damage in the cortex and spinal cord of neuron-specific Tdp-43ΔNLS expressing mice (e). (f) Quantitation of the number of cells with TUNEL-positive nuclei. Scale bars: 20 μm. (g-h) Long-range PCR amplification (LA-PCR) of ~10 kb length exhibited reduced genomic DNA integrity in the cortex of ALS mice (N = 6) compared to that in sham mice (N = 3). A 200 bp short-amplification product was used as an internal control (g). (h) Quantitation of relative band intensity of each genomic target in the ALS group than that in the sham group. Data are expressed as mean ± SD and analyzed by two-tailed nonparametric Mann-Whitney U test; mean ± SD; *P < 0.05, **P < 0.01, ****P < 0.0001.

Fig. 7.. Tdp-43ΔNLS causes trapping of XRCC4 and Ligase 4 in the cytosol of neurons.

(a) Representative images of proximity ligation assay (PLA) between TDP-43 and DNA Ligase 4 (Lig4) in the cortex and hippocampus of sham (upper panel) and ALS (lower panel) mice. Cell bodies were counterstained with Alexa-Fluor 488-conjugated Nissl stain. PLA signals were visualized as red foci/puncta at 568 nm. Scale bars, 10 μm (b) Representative PLA (TDP-43 vs. XRCC4) images in the cortex and hippocampus of sham (upper panel) and ALS (lower panel) mice. Scale bars, 10 μm Cell bodies were counterstained with fluorescent-tagged Nissl stain. (c and d) Quantitation of PLA signal intensity from 12 fields (two fields per animal) in each group for TDP-43 vs Lig4 (c) and TDP-43 vs XRCC4 (d). Data are expressed as mean ± SD and analyzed by two-tailed paired t-tests; ****P < 0.0001.

Fig. 8.. Tdp-43ΔNLS mice display signs of neuro-inflammation in the CNS.

(a-b) Representative IF images showed enhanced expression of inflammatory marker Iba-1 in Tdp-43ΔNLS ALS mice’s cortex compared to sham mice (N = 6 /group) (a). Nuclei were counterstained with DAPI. Scale bars, 10 μm. (b) Quantitation of the signal intensity (a.u.). (c) Representative IF images displaying GFAP-positive astrocyte activation surrounding the cell with Tdp-43 aggregates (green; indicated with a white arrowhead) in the cortical region of ALS mice but not in sham mice. Nuclei were counterstained with DAPI. Scale bars, 10 μm. (d) Quantitation of relative mRNA levels (fold-change) of neuro-inflammatory markers Il-6 and Tnf-α in the cortical tissue of ALS and sham mice (N = 6). Gapdh served as the internal control. Data are expressed as mean ± SD and analyzed by two-tailed unpaired t-tests; ns = nonsignificant, ***P < 0.001.

Fig. 9.. Tdp-43ΔNLS mice exhibit neuronal senescence phenotype in the CNS.

(a-c) Representative IF with anti-γH2AX antibody combined with fluorescence-based senescence staining (CellEvent, Thermo) images showed higher populations of neurons dual-positive for γH2AX and senescence in the motor cortex (a) and spinal cord (thoracic) regions (b) of Tdp-43ΔNLS ALS mice than that in sham mice (N = 6 /group). Cell bodies were counterstained with Nissl. Scale bars, 40 μm (whole field); 10 μm (zoomed images). (c) Quantitation of the percent population of DNA damage-associated senescent cells. (d) Quantitation of relative mRNA levels (fold-change) of senescence-associated markers Edn1, p21, and Ankrd1 in the cortical tissue of ALS and sham mice (N = 6). Gapdh served as the internal control. Data are expressed as mean ± SD and analyzed by two-tailed unpaired t-tests; ns = nonsignificant, *P < 0.05, **P < 0.01, ***P < 0.001.