Sarm1 deletion suppresses TDP-43-linked motor neuron degeneration and cortical spine loss

Abstract

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative condition that primarily affects the motor system and shares many features with frontotemporal dementia (FTD). Evidence suggests that ALS is a 'dying-back' disease, with peripheral denervation and axonal degeneration occurring before loss of motor neuron cell bodies. Distal to a nerve injury, a similar pattern of axonal degeneration can be seen, which is mediated by an active axon destruction mechanism called Wallerian degeneration. Sterile alpha and TIR motif-containing 1 (Sarm1) is a key gene in the Wallerian pathway and its deletion provides long-term protection against both Wallerian degeneration and Wallerian-like, non-injury induced axonopathy, a retrograde degenerative process that occurs in many neurodegenerative diseases where axonal transport is impaired. Here, we explored whether Sarm1 signalling could be a therapeutic target for ALS by deleting Sarm1 from a mouse model of ALS-FTD, a TDP-43^Q331K, YFP-H double transgenic mouse. Sarm1 deletion attenuated motor axon degeneration and neuromuscular junction denervation. Motor neuron cell bodies were also significantly protected. Deletion of Sarm1 also attenuated loss of layer V pyramidal neuronal dendritic spines in the primary motor cortex. Structural MRI identified the entorhinal cortex as the most significantly atrophic region, and histological studies confirmed a greater loss of neurons in the entorhinal cortex than in the motor cortex, suggesting a prominent FTD-like pattern of neurodegeneration in this transgenic mouse model. Despite the reduction in neuronal degeneration, Sarm1 deletion did not attenuate age-related behavioural deficits caused by TDP-43^Q331K. However, Sarm1 deletion was associated with a significant increase in the viability of male TDP-43^Q331K mice, suggesting a detrimental role of Wallerian-like pathways in the earliest stages of TDP-43^Q331K-mediated neurodegeneration. Collectively, these results indicate that anti-SARM1 strategies have therapeutic potential in ALS-FTD.

Keywords: Amyotrophic lateral sclerosis; Axonal protection; Dendritic spines; Sterile alpha and TIR motif-containing protein 1; TAR DNA-binding protein 43; Wallerian degeneration.

Fig. 1

Sarm1 deletion mitigates TDP43^Q331K-mediated motor neuron loss and axon degeneration. a. Breeding scheme. b. Nissl-stained lumbar motor neurons of 10-month-old mice. Representative images shown. Scale bar, 50 μm. c. Quantification of Nissl-stained lumbar motor neurons per section at spinal cord L3. (n = 4 NTG; n = 4 Q331K-Sarm1^+/−; n = 6 Q331K-Sarm1^−/−) ANOVA P = 0.0404. Pairwise comparisons: NTG vs. Q331K-Sarm1^+/−: * P = 0.0351; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: * P = 0.0392. d. Representative images showing intact YFP+ axons of L4 ventral roots from mice of three genotypes at different time points. Scale bar 100 μm. e. Quantification of fluorescent axons in wholemount YFP+ L4 ventral roots at different time points. ANOVA Interaction P = 0.0157. Pairwise comparisons: 5 weeks, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ** P = 0.0026; 3 months, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ** P = 0.0040; 10 months, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: * P = 0.0289. f. Percentage of YFP+ axons undergoing fragmentation in L4 ventral roots at different time points. ANOVA Genotype P < 0.0001. Pairwise comparisons: 5 weeks, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ** P = 0.0014; 3 months, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.1104; 10 months, NTG vs. Q331K-Sarm1^+/−: * P = 0.0464; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.7870. g. Fibre diameters measured at the thickest part of intact YFP+ axons in L4 ventral roots at different time points. ANOVA Interaction P = 0.0051. Pairwise comparisons: 5 weeks, NTG vs. Q331K-Sarm1^+/−: ns P = 0.1295; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.2198; 3 months, NTG vs. Q331K-Sarm1^+/−: ns P = 0.1567; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.1567; 10 months, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.1810. h. Distribution of diameters of YFP+ L4 motor axons at 10 months of age. ANOVA Interaction P = 0.0014. Pairwise comparisons: Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: * P = 0.0386. For (e-h) (n = 5 mice per genotype); ****P < 0.0001. For (c) one-way (e-h) two-way ANOVA followed by Holm-Sidak post-hoc test for pairwise comparisons. Error bars represent mean ± s.e.m

Fig. 2

Sarm1 deletion reduces TDP-43^Q331K-mediated NMJ degeneration. a. Representative images of immunofluorescent staining of NMJs in the gastrocnemius muscle from mice at different time points (Green = synaptophysin and β-III-tubulin co-stain, Red = α-bungarotoxin). Scale bar, 50 μm. (b-d) Percentages of fully innervated, partially innervated, and denervated NMJs in the gastrocnemius muscles of each genotype at different time points. b. At 5 weeks, ANOVA Interaction P < 0.0001. Pairwise comparisons: Fully innervated, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.0520; Partially innervated, NTG vs. Q331K-Sarm1^+/−: ** P = 0.0027; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.5852; Denervated, NTG vs. Q331K-Sarm1^+/−: ** P = 0.0024; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.1107. c. At 3 months: ANOVA Interaction P < 0.0001. Pairwise comparisons: Fully innervated, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ** P = 0.0032; Partially innervated, NTG vs. Q331K-Sarm1^+/−: *** P = 0.0004; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.5585; Denervated, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: * P = 0.0124. d. At 10 months: ANOVA Interaction P < 0.0001. Pairwise comparisons: Fully innervated, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.1949; Partially innervated, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.0711; Denervated, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ** P = 0.0030. For (b-d) (n = 5 mice per genotype); ****P < 0.0001; two-way ANOVA followed by Holm-Sidak post-hoc test for pairwise comparisons. Error bars represent mean ± s.e.m

Fig. 3

Neurodegeneration is more prominent in the entorhinal cortex than motor cortex of TDP-43^Q331K mice. a-b. Brain weights of mice at different time points. a. Male (n = 6–8 NTG; n = 4–6 Q331K- Sarm1^+/−; n = 6–10 Q331K-Sarm1^−/−). ANOVA genotype P < 0.0001. Pairwise comparisons: 5 weeks, NTG vs. Q331K-Sarm1^+/−: * P = 0.0160; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.5696; 3 months, NTG vs. Q331K-Sarm1^+/−: * P = 0.0458; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.9658; 10 months, NTG vs. Q331K-Sarm1^+/−: **** P < 0.0001; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.6290. b. Female (n = 5–8 NTG; n = 3–5 Q331K-Sarm1^+/−; n = 4–5 Q331K-Sarm1^−/−). ANOVA genotype P = 0.0006. Pairwise comparisons: 1 month, NTG vs. Q331K-Sarm1^+/−: ns P = 0.3873; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.3283; 4 months, NTG vs. Q331K-Sarm1^+/−: ns P = 0.4957; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.8347; 15 months, NTG vs. Q331K-Sarm1^+/−: * P = 0.0279; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.5565. c. Brain volumes of female mice at 10 months of age measured by ex vivo MRI (n = 11 NTG; n = 10 Q331K-Sarm1^+/−; n = 11 Q331K-Sarm1^−/−). ANOVA P < 0.0001. Pairwise comparisons: NTG vs. Q331K-Sarm1^+/−: *** P = 0.0001; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.6180. d-e. MRI study-specific template (d. coronal; e. transverse) with an overlay representing voxel-wise volume differences (%) between Q331K-Sarm1^+/− and NTG mice at 10 months of age. The colour of the overlay indicates the inter-group volume difference (warm and cool colours represent gain and loss of volume, respectively, ranging from − 25 to 25%), while the transparency indicates the statistical significance, ranging from FWE-corrected p value 0.5 (transparent) to 0 (opaque). Areas in which FWE-corrected p value < 0.05 are contoured in black. (Red arrow – entorhinal cortex; yellow arrow – cingulate cortex) f-g. Representative images of YFP+ neurons from the primary motor cortex (upper) and entorhinal cortex (lower) from mice 10 months of age. Scale bar 200 μm. f. Density of YFP+ neurons in both the primary motor and entorhinal cortex (n = 5 NTG; n = 4 Q331K-Sarm1^+/−; n = 5 Q331K-Sarm1^−/−). YFP+ neuron density motor cortex: ANOVA P = 0.2661. YFP+ neuron density entorhinal cortex: ANOVA P = 0.0013, Pairwise comparisons: NTG vs. Q331K-Sarm1^+/−: *** P = 0.0009; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.1079. (a-b) Two-way (c, g) one-way ANOVA followed by Holm-Sidak post-hoc test for pairwise comparisons; error bars represent mean ± s.e.m

Fig. 4

TDP-43^Q331K-mediated dendritic spine defects in the motor cortex are partially suppressed by Sarm1 deletion. a. Representative images of the Thy1-YFP motor cortex apical dendrites from Layer V cortical neurons (upper), and the corresponding Neurolucida tracing (middle; Red = thin spine, Pink = stubby spine, Blue = mushroom spine) from mice of each genotype at 10 months of age. Scale bar, 5 μm. Examples of different spine morphologies (lower) [28]. b. Density of apical dendritic spines per micrometre in motor cortex. ANOVA P = 0.0016. Pairwise comparisons: NTG vs. Q331K-Sarm1^+/−: ** P = 0.0014; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: * P = 0.0268. (c-e) Spines were classified either as mushroom, stubby or thin type according their morphologic features. Density of each type of apical dendritic spines per millimetre in motor cortex. c. Thin spine density: ANOVA P = 0.0034. Pairwise comparisons: NTG vs. Q331K-Sarm1^+/−: ** P = 0.0034; Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: * P = 0.0227. d. Stubby spine density E. Mushroom spine density. For (b-e) (n = 5 mice per genotype); one-way ANOVA followed by Holm-Sidak post-hoc test for pairwise comparisons; error bars represent mean ± s.e.m

Fig. 5

Sarm1 deletion attenuates the pre-weaning loss of male TDP-43^Q331K mice but does not influence age-related behavioural impairments. a. Ratios of mice genotyped at birth (all of which were successfully weaned) broken down by gender. Female (χ² = 0.289, d.f. = 3, P = 0.962), Male (χ² = 8.387, d.f. = 3, P = 0.039); Chi square test. b. Latency to fall of male transgenic mice on accelerating rotarod (n = 5–10 NTG; n = 4–8 Q331K-Sarm1^+/−; n = 6–13 Q331K-Sarm1^−/− mice per genotype). Fixed effects (Age x Genotype) P < 0.0001. Pairwise comparisons: Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.6873. c. Weights of male mice (n = 4–10 NTG; n = 4–8 Q331K-Sarm1^+/−; n = 13 Q331K-Sarm1^−/− mice per genotype). Fixed effects (Age x Genotype) P = 0.0113. Pairwise comparisons: Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.8984. d. Score of male mice on hindlimb clasping test (n = 5–10 NTG; n = 4–8 Q331K-Sarm1^+/−; n = 3–13 Q331K-Sarm1^−/− mice per genotype). Fixed effects (Age x Genotype) P = 0.0003. Pairwise comparisons: Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.0764. For (b-d) Mixed-effects analysis followed by Holm-Sidak post-hoc test for pairwise comparisons. e. Number of marbles buried at different ages (n = 7–12 NTG; n = 4–9 Q331K-Sarm1^+/−; n = 7–18 Q331K-Sarm1^−/− mice per genotype). ANOVA Genotype P < 0.0001. Pairwise comparisons: 3–4 months, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P = 0.4154; 5–7 months, Q331K-Sarm1^+/− vs. Q331K-Sarm1^−/−: ns P > 0.9999. Two-way ANOVA followed by Holm-Sidak post-hoc test for pairwise comparisons. For (b-e) ****P < 0.0001; error bars represent mean ± s.e.m