Spinal poly-GA inclusions in a C9orf72 mouse model trigger motor deficits and inflammation without neuron loss

Abstract

Translation of the expanded (ggggcc)_n repeat in C9orf72 patients with amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) causes abundant poly-GA inclusions. To elucidate their role in pathogenesis, we generated transgenic mice expressing codon-modified (GA)₁₄₉ conjugated with cyan fluorescent protein (CFP). Transgenic mice progressively developed poly-GA inclusions predominantly in motoneurons and interneurons of the spinal cord and brain stem and in deep cerebellar nuclei. Poly-GA co-aggregated with p62, Rad23b and the newly identified Mlf2, in both mouse and patient samples. Consistent with the expression pattern, 4-month-old transgenic mice showed abnormal gait and progressive balance impairment, but showed normal hippocampus-dependent learning and memory. Apart from microglia activation we detected phosphorylated TDP-43 but no neuronal loss. Thus, poly-GA triggers behavioral deficits through inflammation and protein sequestration that likely contribute to the prodromal symptoms and disease progression of C9orf72 patients.

Keywords: ALS; C9orf72; FTD; FTLD; MND; Mouse model; Neurodegeneration; Neurological disorder.

Fig. 1

Expression and distribution pattern of poly-GA aggregates in GA-CFP mice. a Schematic diagram of the construct containing the murine Thy1 promoter driving expression of a synthetic gene encoding (GA)₁₄₉ with its endogenous C-terminal tail fused to CFP. (GA)₁₄₉-CFP is replacing the endogenous coding region. b Distribution of GA aggregates show many inclusions in the spinal cord and brainstem and no aggregates in cortical regions, hippocampus or molecular and granular layer of the cerebellum. BO olfactory bulb, BS brainstem, CA3 cornu ammonis fields 3, CBLgl cerebellar granular cell layer, CBLml cerebellar molecular cell layer, CBLncl lateral cerebellar nuclei, DG dentate gyrus, SCAl anterior horn of lumbar spinal cord, SCAt anterior horn of thoracic spinal cord, SCPl posterior horn of lumbar spinal cord, SCPt posterior horn of thoracic spinal corn. Scale bars represent 20 µm. c Increasing number and accumulation of aggregates in spinal cord (SC; upper row) and brainstem (BS; lower row) of 1-, 3- and 6-month-old GA-CFP mice detected by immunohistochemical staining using GA-CT antibody. Scale bar represents 20 µm. Quantitative immunoassay of RIPA-insoluble poly-GA in the spinal cord (d) and brainstem (e) of 1–6-month-old GA-CFP mice (n = 3 mice per time-point; measured in duplicates) shows increasing amounts of poly-GA in a time dependent manner. AU arbitrary unit, data are shown as mean, minimum and maximum

Fig. 2

GA-CFP mice develop p62, Rad23b and Mlf2 pathology similar to human C9orf72 mutation carriers. a Immunohistochemistry shows p62, Rad23b and Mlf2 aggregates in the spinal cord (SC) of 6-month-old GA-CFP mice but not of wildtype mice. b Immunofluorescence stainings show p62, Rad23b and Mlf2 positive inclusions that co-localize with poly-GA in the spinal cord of 6-month-old GA-CFP mice. c Immunohistochemistry detects specific Mlf2 aggregates in the frontal cortex (FCtx) and dentate gyrus (DG) of C9orf72 ALS/FTLD patients. d Double immunofluorescence reveals colocalization of Mlf2 aggregates with poly-GA and phosphorylated TDP-43 inclusions in C9orf72 patients. Scale bars represent 20 µm

Fig. 3

GA-CFP mice show no evidence for neuronal loss but increased TDP-43 phosphorylation. a Double immunofluorescence of 6-month-old GA-CFP spinal cord tissue (SC) shows poly-GA inclusions exclusively in NeuN-positive cells. Scale bar represents 20 µm. b, c Nissl staining and NeuN immunohistochemistry of 6-month-old GA-CFP and wildtype spinal cords. Scale bar represents 100 µm. Quantitative analysis of NeuN-positive neurons shows no significant difference between wildtype and GA-CFP mice (n _GA-CFP/wt = 3). d Immunostaining of poly-GA aggregates in choline acetyltransferase (ChAT)-positive motoneurons. Scale bar represents 20 µm. e–g Immunohistochemistry and quantitative analysis of ChAT-positive motoneurons of 6-month-old mice in the anterior horn of the spinal cord revealed no statistically significant differences in neuron count (n _{GA-CFP/wt mice} = 4) and size (n _{GA-CFP motoneurons} = 228; n _{wt motoneurons} = 195). Neurons were counted as described in the “Statistics” section. Scale bar represents 100 µm. h Immunoassay for phosphorylated TDP-43 in sarkosyl (1%)-soluble or urea (7M)-soluble spinal cord fractions from 6-month-old GA-CFP or wildtype (wt) mice. n _(wt) = 12; n _(GA-CFP) = 8. Unpaired t test (two-tailed; sarkosyl t = 0.3034, df = 18; urea t = 4.172, df = 18). Data are shown as box plot with whiskers at the 1st and 99th percentile. ***p < 0.001, ns not significant

Fig. 4

Activation of phagocytic microglia in GA-CFP mice. a, b, e Immunohistochemistry of 1- and 6-month-old mice shows microglia activation detected by the markers CD68 and Iba1 in the spinal cord (SC) of 6-month-old GA-CFP mice compared to wildtype mice in the anterior horn of the spinal cord. No clear difference was observed for the astrocyte marker GFAP. Scale bars represent 100 µm. c, d, f Quantitative RT-PCR shows increased mRNA expression of CD68 and Iba1 but not GFAP in 6-month-old GA-CFP mice compared to 1-month-old GA-CFP mice and controls. Expression was normalized to GAPDH and β-actin using the ∆∆Ct method. n _(wt) = 3; n _(GA-CFP) = 3; Statistical analyses were performed by an unpaired t test (two-tailed; CD68_1-month t = 3.079, df = 4; Iba1_1-month t = 2.385, df = 4; GFAP_1-month t = 0.4147, df = 4; CD68_6-months t = 4.805, df = 4; Iba1_​6-months t = 6.399, df = 4; GFAP_​6-months t = 1.771, df = 4) and data are shown as mean ± SEM; *p < 0.05; **p < 0.01, ns not significant

Fig. 5

GA-CFP mice are less active and develop balancing deficits. a–e Neurological and behavioral analysis of GA-CFP and wildtype (wt) littermates at 3–4 months (a–c) and 14 months of age (d, e). a Open field analysis at 3 months. Automated analysis of the time spent in the center, the rearing activity (total count) and the total distance traveled within 20 min for the different genotypes and genders. b Beam ladder with irregular step distance. Automated analysis of average time needed to cross the ladder and the number of hind paw slips at the age of 4 months. c Gait analysis according to the SHIRPA protocol. Fraction of mice showing hindlimb clenching and abnormal gait is shown (at the age of 3.5 months). n _{(GA-CFP male)} = 16; n _{(GA-CFP female)} = 15; n _{(wt male)} = 14; n _{(wt female)} = 15 for all tests at 3 months of age. d, e Repetition of the beam ladder and SHIRPA analysis of GA-CFP and wildtype (wt) littermates at 14 months of age. n _{(GA-CFP male)} = 15; n _{(GA-CFP female)} = 15; n _{(wt male)} = 13; n _{(wt female)} = 13. f, g Longitudinal analysis using a balance beam. GA-CFP males and females take more time to cross the beam and fall more often than wildtype mice starting at 4 months. n _{(GA-CFP male)} = 4; n _{(GA-CFP female)} = 4; n _{(wt male)} = 6; n _{(wt female)} = 6. Statistical analysis of open field and beam ladder assay was performed by a two-way ANOVA between sex and genotype followed by Bonferroni post hoc test. Statistical analysis of the beam walk was performed by a two-way ANOVA between time and genotype for each sex followed by Bonferroni post hoc test. Asterisks for open field analysis and beam ladder traversing time depict significance of genotype effects [open field F (1, 56) = 7.579; beam ladder traversing time F (1, 52) = 10.2]. Asterisks for beam ladder hind paw trials, SHIRPA and beam walk depict significance of genotype and sex-dependent effects (Bonferroni). Statistical analysis of hind paw clenching and gait was performed by a Chi-square test. Data are shown as box plot with whiskers at the 1st and 99th percentille (a–e, g) or as mean ± SEM (f); *p < 0.05; **p < 0.01; ***p < 0.001