RBM3 mediates structural plasticity and protective effects of cooling in neurodegeneration

Abstract

In the healthy adult brain synapses are continuously remodelled through a process of elimination and formation known as structural plasticity. Reduction in synapse number is a consistent early feature of neurodegenerative diseases, suggesting deficient compensatory mechanisms. Although much is known about toxic processes leading to synaptic dysfunction and loss in these disorders, how synaptic regeneration is affected is unknown. In hibernating mammals, cooling induces loss of synaptic contacts, which are reformed on rewarming, a form of structural plasticity. We have found that similar changes occur in artificially cooled laboratory rodents. Cooling and hibernation also induce a number of cold-shock proteins in the brain, including the RNA binding protein, RBM3 (ref. 6). The relationship of such proteins to structural plasticity is unknown. Here we show that synapse regeneration is impaired in mouse models of neurodegenerative disease, in association with the failure to induce RBM3. In both prion-infected and 5XFAD (Alzheimer-type) mice, the capacity to regenerate synapses after cooling declined in parallel with the loss of induction of RBM3. Enhanced expression of RBM3 in the hippocampus prevented this deficit and restored the capacity for synapse reassembly after cooling. RBM3 overexpression, achieved either by boosting endogenous levels through hypothermia before the loss of the RBM3 response or by lentiviral delivery, resulted in sustained synaptic protection in 5XFAD mice and throughout the course of prion disease, preventing behavioural deficits and neuronal loss and significantly prolonging survival. In contrast, knockdown of RBM3 exacerbated synapse loss in both models and accelerated disease and prevented the neuroprotective effects of cooling. Thus, deficient synapse regeneration, mediated at least in part by failure of the RBM3 stress response, contributes to synapse loss throughout the course of neurodegenerative disease. The data support enhancing cold-shock pathways as potential protective therapies in neurodegenerative disorders.

Extended Data Figure 1. Stereological assessment of volume and synapse size to validate 2D assumption-based approaches for counting synapse density

a) CA1 volume and synapse mean length and area in the stratum radiatum remain essentially unchanged on cooling and rewarming in wild type mice. Volume was measured using disector principle and synapse mean length and area determined in the same sections, as described, n= as reported for Figure 1a. Representative electron micrographs (pseudo-coloured for ease of synapse identification) for data not shown in Figure 1b and 1c, from prion-infected mice at 4 and 6 w.p.i. b) and for 5x FAD mice at 2 and 3 months c) before cooling (black framed images) and cooled (blue framed images). d) Schematic showing lost capacity for structural plasticity precedes synapse loss and neuronal loss in both mouse models. Scale bar = 1μm. All data in bar charts are mean ± s.e.m. Student’s t-test, two tailed. Non-significant p values.

Extended Data Figure 2. Synaptic protein levels during cooling-rewarming in prion and 5×FAD mice

a) Levels of pre-(SNAP25, VAMP2) and post-(PSD95, NR1) synaptic proteins do not change before (black bars) and after cooling to 16-18°C (blue bars) in prion-infected mice at 4 and 6 w.p.i, and b) 5×FAD mice at 2 and 3 months. Representative western blots are shown for 3 mice per temperature and time point. Bar graphs show quantification of synaptic protein levels relative to GAPDH. All data represent means ± s.e.m. (n = 3-11 mice per time point). Student’s t-test, two tailed. Non-significant p values.

Extended Data Figure 3. Cooling does not induce changes in PrP^Sc or Aβ levels

a) Levels of total PrP (upper blot) and PrP^Sc (lower blot) do not change notably before (white line), during (blue line) or after (red line) cooling to 16-18°C in prion-infected mice. PrP^Sc is detected after digestion with proteinase K. Levels are undetectable by western bloting at 6 w.p.i., as expected. b) Cooling does not change levels of Aβ oligomers in 5×FAD mice, arrow indicates Aβ monomers (Lane 1 - synthetic Aβ oligomers, last lane - one year old 5×FAD control (C+)). Representative western blots are shown for 3 mice per temperature and time point. Data in bar charts are mean ± s.e.m. Student’s t-test, two tailed. Non-significant p values.

Extended Data Figure 4. Cooling induces sustained increase in RBM3 levels but not in CIRP

Levels of CIRP do not change after cooling in a) prion-infected at 4 and 6 w.p.i or b) in 5x FAD mice at 2 and 3 months. Representative western blots are shown for 3 mice per temperature and time point. Bar graphs show quantification of CIRP levels relative to GAPDH. All data represent means ± s.e.m (n = 6-9 mice per time point). Student’s t-test, two tailed. Non-significant p values. c) Increased levels of RBM3 are sustained for at least 72hrs after cooling in wild type mice. Bar graph shows quantification of RMB3 against GAPDH in control (white bar), cooled (blue bar), and 12, 48 and 72 hrs recovery after cooling (red bars). All data represent means ± s.e.m, (n= 3-6 mice per time points, *p < 0.05, Mann Whitney U test, two tailed).

Extended Data Figure 5. Levels of RBM3 remain elevated after early cooling

to 16-18°C in prion-infected mice (magenta boxes) compared to control prion-infected mice. These levels remained high up to six weeks later and declined at 12 w.p.i. Representative western blots are shown for 3 mice per time point.

Extended Data Figure 6. Exploration time in exposure phase of novel object testing is normal in all groups; RBM3 knockdown abolishes improved memory after cooling

a) exploratory behaviour measured in seconds is not different in mice with early cooling from prion diseased mice and is not affected by the duration of disease (n as reported in figure 3d); b) lentivirally mediated RNAi of RBM3 eliminates the protective effect of cooling on novel object memory impairment in prion disease (dark green bar); c) but does not affect exploratory behaviour in training phase. Data analysed using one way ANOVA, Brown-Forsythe test with Tukey’s post hoc analysis for multiple comparisons (n = 11-16 mice per time point, **p < 0.01).

Extended Data Figure 7. Induction of hypothermia at time point when RBM3 induction fails is not neuroprotective

Cooling at 5 and 6 w.p.i., when synaptic plasticity and RBM3 induction fails (see Figure 1 and 2, main text), does not increase survival in prion infected mice. Kaplan-Meier survival plots for prion-infected mice (black line, no cooling; n = 10; orange line, mice cooled at 5 and 6 w.p.i., n = 16). Student’s t-test, two tailed. Non-significant p values

Extended Data Figure 8. PrP^Sc levels remain unchanged in prion with over-expression of RMB3

In prion-infected mice total PrP and PrP^Sc levels do not alter a) after early cooling to 16-18°C (magenta boxes), or b) following treatment with LV-RBM3 (dark green) and LV-shRNA-RBM3 (pale green). PrP and PrP^Sc levels tested in 9 w.p.i and terminal mice. PrP^Sc is detected after digestion with proteinase K. Representative western blots are shown for 3 mice per temperature and time point, C shows uninfected control mouse.

Extended Data Figure 9. Mild hypothermia also extends survival in prion-infected mice

Kaplan Meier plot showing that cooling to 26°C at an early stage also significantly lengthens survival (n = 27 v n = 16 mice); **p <0.01, Student’s t-test, two tailed.

Extended Data Figure 10. RNAi of RBM3 down-regulation accelerates impaired structural synaptic plasticity in the 5×FAD mouse model, and also reduced synapse number and function in wild type mice

Impaired structural synaptic plasticity after cooling occurs in shRNA-RBM3 treated 5×FAD mice at 3 months. Representative electron micrographs are shown and are pseudo-coloured as in main text figures. Quantification shows significant reduction in synapse number by RNAi of RBM3. (n = 82-93 images from 3 mice per time point, Student’s t-test, two tailed) b) RBM3 knockdown reduces synapse number and novel object memory in wild type mice. (n =93 images from 3 mice per time point, Student’s t-test, two tailed ***p <0.0001; n.o.r. n=11 mice LV-shRNA-control and 10 mice LV-shRNA-RBM3, Mann Whitney U test, * p <0.05). Scale bar = 1μm

Figure 1. The capacity for synaptic regeneration is lost early in neurodegenerative disease

a) synapse numbers decline on cooling and recover on rewarming in wild type mice, counted in both 3D and 2D. Representative electron micrographs (pseudo-coloured for ease of synapse identification) and bar charts showing quantitation are shown for each experiment (n = 4 animals at 18°C and n = 2 at 37°C; 192 images from 2 mice per condition for 3D analyses; 93 images from 3 animals per condition, for 2D analyses). A typical tripartite synapse is shown at higher magnification. b) The same response is seen in prion diseased mice at 4 and 5 w.p.i. but this fails at 6 w.p.i (arrow), and in c) 5×FAD mice, where it fails at 3 months (arrow). ***p <0.0001, **p <0.01; *p <0.05; n.s. non significant. Student’s t-test; two tailed. All data in bar charts are mean ± s.e.m. Scale bar = 1μm. Source data for all figures can be found in the Supplementary tables.

Figure 2. Failure to induce RBM3 parallels lost capacity for synaptic recovery in neurodegenerative disease models

a) Cooling induces increased RBM3 levels in hippocampi of wild type mice. The response fails b) at 6 w.p.i. in prion-infected mice and c) at 3 months in 5×FAD mice (arrows). Representative western blots are shown. Bar graphs show quantification of RBM3 levels relative to GAPDH, (n = 6-11 mice per time point; all experiments in triplicate) **p < 0.01, *p <0.05, Mann Whitney U test in a and c, Student’s t-test in b. All data are mean ± s.e.m.

Figure 3. Early cooling induces RBM3 over-expression and is neuroprotective in prion-infected mice

a) Cooling at 3 and 4 w.p.i. resulted in sustained high levels of RBM3 in hippocampus for several weeks (n= 3-8 mice per time point) causing b) marked protection of synapse number at 7, 8 and 9 w.p.i. (62 images from 2 mice per time point). Scale bar = 1μm. c) Early cooling maintained synaptic transmission (n= 4-8 cells from 2 mice per time point; representative raw traces of evoked EPSCs are shown) and d) prevented decline in burrowing behaviour and loss of novel object recognition memory, expressed as ratio of exploratory preference (n ≥10 mice per group) in contrast to un-cooled mice. e) Haematoxylin and eosin stained sections show striking reduction in hippocampal spongiosis and protection of CA1 neurons (bar chart) in cooled mice that is abolished by RBM3 knockdown (n = 4-6 animals per treatment, except LV-shRNA-RBM3: n = 2). Scale bar = 50μm. One-way ANOVA, Brown-Forsythe test with Tukey’s post hoc analysis for multiple comparisons was used in d and e. f) Early cooling significantly prolonged survival, but this was abolished by knockdown of RBM3 (n = 31 cooled mice; n = 17 not cooled; n = 10 cooled + RNAi of RBM3). Mann-Whitney U test. *p < 0.05; **p< 0.01; ***p < 0.001. Student’s t-test two tails was used unless otherwise stated. All data in bar charts are mean ± s.e.m.

Figure 4. Lentivirally mediated over-expression of RBM3 restores structural synaptic plasticity and is neuroprotective in neurodegenerative disease

a) LV-RBM3 produces high levels of expression in hippocampus; LV-shRNA-RBM3 reduces endogenous levels. Representative western blot and bar graph, where n = 6 mice per treatment, repeated in triplicate. b) LV-RBM3 rescues the deficit in structural plasticity in prion-infected and 5×FAD mice at 6 w.p.i. and 3 months of age respectively (93 images from 3 mice per time point) and c) produced lasting protection of synapse number in prion-infected mice at 7 and 9 w.p.i., while LV-shRNA-RBM3 accelerated synapse loss (62 images from 2 mice per time point). d) LV-RBM3 resulted in significant recovery in synaptic transmission (left panel, inset: representative raw traces of evoked EPSCs are shown; n= 6 cells from 2 mice per time point) and e) protected against behavioural decline, while RNAi of RBM3 accelerated memory loss and burrowing deficits (n ≥10 mice per group). RBM3 over-expression: f) increases global protein synthesis rates in hippocampal slices of prion-infected mice at 9 w.p.i. (n = 4-6 mice), g) is neuroprotective (n= 3-6 mice) and h) significantly lengthened survival (n = 20 mice), while knockdown accelerated disease (n = 14 mice), compared to LV-control treated mice (n = 31 mice), Mann Whitney U test. p * <0.05 p ** <0.01 *** p < 0.001. Student’s t-test, two tails except in e,f and g which used one-way ANOVA, Brown Forsythe test with Tukey’s post hoc analysis for multiple comparisons. All data represent means ± s.e.m. Scale bar = 1μm in b and c and 50μm in g.