Reversal of neurological deficits by painless nerve growth factor in a mouse model of Rett syndrome

Abstract

Rett syndrome is a rare genetic neurodevelopmental disease, affecting 1 in over 10 000 females born worldwide, caused by de novo mutations in the X-chromosome-located methyl-CpG-binding protein 2 (MeCP2) gene. Despite the great effort put forth by the scientific community, a therapy for this devastating disease is still needed. Here, we tested the therapeutic effects of a painless mutein of the nerve growth factor (NGF), called human NGF painless (hNGFp), via a non-invasive intranasal delivery in female MeCP2+/- mice. Of note, previous work had demonstrated a broad biodistribution of hNGFp in the mouse brain by the nasal delivery route. We report that (i) the long-term lifelong treatment of MeCP2+/- mice with hNGFp, starting at 2 months of age, increased the chance of survival while also greatly improving behavioural parameters. Furthermore, when we assessed the phenotypic changes brought forth by (ii) a short-term 1-month-long hNGFp-treatment, starting at 3 months of age (right after the initial presentation of symptoms), we observed the rescue of a well known neuronal target population of NGF, cholinergic neurons in the medial septum. Moreover, we reveal a deficit in microglial morphology in MeCP2+/- mice, completely reversed in treated animals. This effect on microglia is in line with reports showing microglia to be a TrkA-dependent non-neuronal target cell population of NGF in the brain. To understand the immunomodulatory activity of hNGFp, we analysed the cytokine profile after hNGFp treatment in MeCP2+/- mice, to discover that the treatment recovered the altered expression of key neuroimmune-communication molecules, such as fractalkine. The overall conclusion is that hNGFp delivered intranasally can ameliorate symptoms in the MeCP2+/- model of Rett syndrome, by exerting strong neuroprotection with a dual mechanism of action: directly on target neurons and indirectly via microglia.

Keywords: immunomodulation; neurodevelopmental disorders; neurotrophic factors; neurotrophin-based therapy.

Figure 1

Intranasal delivery of hNGFp ameliorates behavioural deficits and increases lifespan in MeCP2^+/− mice. (A) Timeline: mice started treatment at 2 months of age and underwent weekly phenotyping until sacrifice. Motor performance was evaluated at P150. (B) Representative images of hindlimb clasping in vehicle (PBS) and hNGFp-treated MeCP2^+/− animals. (C–E) Plots indicating the (C) clasping, (D) gait and (E) breathing score measured once per week from the beginning of the treatment up until the humane end point [Clasping: two-way repeated measures ANOVA and Sidak’s multiple comparisons test; interaction: F(17,476) = 3.327, P < 0.0001; Gait: two-way repeated measures ANOVA and Sidak’s multiple comparisons test; interaction: F(17,476) = 4.842, P < 0.0001; mean ± SEM; Breathing: two-way repeated measures ANOVA and Sidak’s multiple comparisons test; interaction: F(17,476) = 4.405, P < 0.0001); orange box indicates significance in the multiple comparison test]. (F) Aggregated score calculated at Day 119 after the beginning of the treatment (unpaired two-tailed t-test; P = 0.0019; mean ± SEM and individual values). (G and H) Representative images and quantification of the score during a beam walk test [two-way ANOVA; Interaction: F(1,49) = 4.867; P = 0.0321; Tukey’s multiple comparisons test: WT + PBS versus MeCP2^+/− + PBS, P = 0.0035; MeCP2^+/− + PBS versus MeCP2^+/− + hNGFp; P = 0.0142; mean ± SEM and individual values] (n = 13–15 per group). (I) Kaplan-Meier survival analysis. MeCP2^+/− + PBS versus MeCP2^+/− + hNGFp; P = 0.01, Gehan-Breslow-Wilcoxon test. hNGFp = human nerve growth factor painless.

Figure 2

Effect of a 1-month hNGFp treatment. (A) Timeline of the experiment. (B–E) Scatter plot with bars of the behavioural score recorded the day before sacrificing and collecting the tissues of PBS- and hNGFp-treated MeCP2^+/− animals. (B) Clasping: two-tailed t-test; P = 0.0282. (C) Gait: two-tailed t-test; P = 0.0087. (D and E) Immobility and tremor: two-tailed t-test. Mean ± SEM and individual values are reported in all scatter plots with bar. (F) Latency to fall expressed in minutes in a rotarod test [two-way ANOVA; column effect (genotype): P = 0.001; Interaction: P = 0.2; Tukey’s multiple comparisons test: WT + PBS versus MeCP2^+/− + PBS: P = 0.0085; MeCP2^+/− + PBS versus WT + hNGFp: P = 0.0179]. (G) Body weight of all groups recorded right before sacrifice [two-way ANOVA; Genotype: F(1,50) = 52.35, P < 0.0001; mean ± SEM]. (H) Brain weight was recorded for all animals at the time of death [two-way ANOVA: Interaction: F(1,50) = 3.257; P = 0.0772; Tukey’s multiple comparisons test: WT + PBS versus MeCP2^+/− + PBS: P < 0.0001; WT + PBS versus MeCP2^+/− + hNGFp: P = 0.0004; MeCP2^+/− + PBS versus WT + hNGFp: P < 0.0001; WT + hNGFp versus MeCP2^+/− + hNGFp: P < 0.0001] (n = 12–14 per group). hNGFp = human nerve growth factor painless; WT = wild-type.

Figure 3

The density of ChAT⁺ neurons in the medial septum is rescued by hNGFp treatment. (A) Representative figures of DAPI, ChAT and NeuN fluorescent staining in the medial septum of WT + PBS, WT + hNGFp, MeCP2^+/− + PBS and MeCP2^+/− + hNGFp animals and their magnifications (scale bars are indicated in the panels). (B–D) Bar plots indicating the density of (B) ChAT⁺ cells, (C) NeuN⁺ cells and (D) DAPI (two-way ANOVA; Tukey’s multiple comparisons test for all plots) (n = 5 animals per group). (B) Density of ChAT⁺ cells [Interaction: F(1,16) = 2.461: P = 0.1363; effect of treatment: F(1,16) = 6.992: P = 0.0177; effect of genotype: F(1,16) = 36.06: P < 0.0001; WT + PBS versus MeCP2^+/− + PBS: P = 0.0003; MeCP2^+/− + PBS versus MeCP2^+/− + hNGFp: P = 0.0398; MeCP2^+/− + PBS versus WT + hNGFp: P < 0.0001]. (C) Density of NeuN⁺ cells [Interaction: F(1,16) = 1.391: P = 0.2555; effect of treatment: F(1,16) = 0.5170: P = 0.48; effect of genotype: F(1,16) = 0.4681: P = 0.5036]. (D) Density of DAPI [Interaction: F(1,16) = 0.6765: P = 0.4229; effect of treatment: F(1,16) = 1.691: P = 0.2119; effect of genotype: F(1,16) = 0.3096: P = 0.5856]. hNGFp = human nerve growth factor painless; WT = wild-type.

Figure 4

Human NGF painless rescues cortical microglia morphology. (A) Representative images of Iba1 immunofluorescence and IMARIS 3D reconstruction for each of the analysed groups. Scale bar = 20μm. (B) Scatter plot with bar representing the density of microglial cells in the cortex of all cohorts of animals. No significant difference was detected [two-way ANOVA, F(1,12) = 0.1165, P = 0.7388; n = 5 animals per group]. (C–F) Results from IMARIS 3D reconstruction (n = 3–5 cells per group; n = 8 animals per group). (C) 3D Sholl analysis [three-way ANOVA: Interaction: Genotype × Treatment: F(1,2128) = 54.54, P < 0.0001; Radius × Genotype Treatment, F(75,2128) = 1.377, P = 0.0191]. (D–F) Scatter plots with bars representing filament length, number of branching points and terminal points, respectively (two-way ANOVA, Tukey’s multiple comparisons test for all plots). (D) Filament length: F(1,27) = 5.237, P = 0.0302; WT + PBS versus MeCP2^+/− + PBS: P < 0.0001; WT + PBS versus MeCP2^+/− + hNGFp: P = 0.22; MeCP2^+/− + PBS versus MeCP2^+/− + hNGFp: P = 0.0072; MeCP2^+/− + PBS versus WT + hNGFp: P < 0.0001. (E) Number of branching points: F(1,27) = 3.973, P = 0,0.0564; WT + PBS versus MeCP2^+/− + PBS: P = 0.0011; WT + PBS versus MeCP2^+/− + hNGFp: P = 0.44; MeCP2^+/− + PBS versus MeCP2^+/− + hNGFp: P = 0.0370; MeCP2^+/− + PBS versus WT + hNGFp: P = 0.0007. (F) Number of terminal points: F(1,27) = 3.511, P = 0.0718; WT + PBS versus MeCP2^+/− + PBS: P = 0.0008; WT + PBS versus MeCP2^+/− + hNGFp: P = 0.34; MeCP2^+/− + PBS versus MeCP2^+/− + hNGFp: P = 0.046; MeCP2^+/− + PBS versus WT + hNGFp: P = 0.0005. Mean ± SEM and individual values are reported in all scatter plots with bar. hNGFp = human nerve growth factor painless; WT = wild-type.

Figure 5

Microglial morphological parameters are highly predictive of behavioural output. (A) Heat map representing values from the correlation matrix (Pearson correlation) of all analysed parameters (bottom triangle), and the significance of each pairwise comparison (top triangle: blue box indicates P < 0.05 after Holm-Sidak multiple testing correction) (all groups were included in the analysis, n = 8 animals per group). (B and C) Detail on clasping and gait correlation with filament length Pearson coefficient and corrected P-values, respectively are reported in the figure. hNGFp = human nerve growth factor painless; WT = wild-type.

Figure 6

Human NGF painless modulates the expression levels of disease-relevant cytokines in MeCP2^+/− mice. (A) Heat map showing the protein levels of cytokines in WT + PBS, WT + hNGFp, MeCP2^+/− + PBS and MeCP2^+/− + hNGFp cohorts of mice, quantified using the RayBio® C-Series mouse inflammation antibody array. (B) PCA analysis showing the clustering of the different treatments. (C–K) Histogram plots showing the protein levels of selected cytokines from the array and the statistical analysis from the two-way ANOVA and Sidak’s multiple comparisons test. (C) Fractalkine: Interaction: F(1,12) = 6.155, P = 0.0289; effect of Genotype: F(1,12) = 8.145, P = 0.0145; WT + PBS versus MeCP2^+/− + PBS: P = 0.0159; MeCP2^+/− + PBS versus MeCP2^+/− + hNGFp: P = 0.0805. (D) IL-12p70: effect of Treatment: F(1,12) = 6.915, P = 0.022; effect of Genotype: F(1,12) = 18.11, P = 0.001; WT + PBS versus MeCP2^+/− + PBS: P = 0.0819; MeCP2^+/− + PBS versus WT + hNGFp: P = 0.0023; WT + hNGFp versus MeCP2^+/− + hNGFp: P = 0.0493. (E) CD30L: effect of Genotype: F(1,12) = 5.619, P = 0.0354. (F) Eotaxin-2: effect of Genotype: F(1,12) = 5.639, P = 0.0351. (G) BLC: effect of Treatment: F(1,12) = 5.154, P = 0.0424. (H) IL-1a: effect of Treatment: F(1,12) = 5.104, P = 0.0433. (I) IL-13: effect of Treatment: F(1,12) = 4.962, P = 0.0458. (J) SDF1: effect of Treatment: F(1,12) = 4.160, P = 0.064. (n = 4 per group, mean ± SEM and individual values are reported in all scatter plots with bar). hNGFp = human nerve growth factor painless; WT = wild-type.

Figure 7

Human NGF painless rescues microglia neuroprotective activity towards spines. (A) Representative images of Homer1bc staining for postsynaptic puncta in cortical neurons cocultured with microglia. The experimental conditions are: neurons only (n only), neurons + microglia WT (n + mWT), neurons + microglia from MeCP2^−/− mice (n + mMECP2^−/−); cultures are either treated with 100 ng/ml of hNGFp or vehicle (NT = DMEMF12) for 3 h. Scale bar = 10 µm. (B) Violin plots of all quantified puncta densities in all conditions [three replicates, number of dendrites = 120 per group; two-way ANOVA; Interaction: F(2,624) = 11.48; P < 0.0001; Tukey’s multiple comparisons test: neurons only + NT versus neurons + microglia WT + NT: P = 0,0158; neurons + microglia MeCP2^−/− + NT versus neurons + microglia MeCP2^−/− + hNGFp: P < 0.0001; neurons + microglia WT + NT versus neurons + microglia MeCP2^−/− + NT: P = 0.0069]. hNGFp = human nerve growth factor painless; NT = non-treated/vehicle; WT = wild-type.