Stress vulnerability shapes disruption of motor cortical neuroplasticity

Abstract

Chronic stress is a major cause of neuropsychiatric conditions such as depression. Stress vulnerability varies individually in mice and humans, measured by behavioral changes. In contrast to affective symptoms, motor retardation as a consequence of stress is not well understood. We repeatedly imaged dendritic spines of the motor cortex in Thy1-GFP M mice before and after chronic social defeat stress. Susceptible and resilient phenotypes were discriminated by symptom load and their motor learning abilities were assessed by a gross and fine motor task. Stress phenotypes presented individual short- and long-term changes in the hypothalamic-pituitary-adrenal axis as well as distinct patterns of altered motor learning. Importantly, stress was generally accompanied by a marked reduction of spine density in the motor cortex and spine dynamics depended on the stress phenotype. We found astrogliosis and altered microglia morphology along with increased microglia-neuron interaction in the motor cortex of susceptible mice. In cerebrospinal fluid, proteomic fingerprints link the behavioral changes and structural alterations in the brain to neurodegenerative disorders and dysregulated synaptic homeostasis. Our work emphasizes the importance of synaptic integrity and the risk of neurodegeneration within depression as a threat to brain health.

Fig. 1. CSDS phenotypes defined by symptom load have distinct alterations in motor learning.

a Experimental timeline (day 0 defined by the last day of CSDS = first day of behavioral testing). b Schematic depiction of the CSDS paradigm and control conditions. c Behavioral testing showed reduced nest building (U = 95, P < 0.0001, Mann-Whitney U test) and social interaction (t₄₇ = 2.399, P = 0.021, student’s t-test) but no change in sucrose consumption (U = 264, P = 0.493, Mann-Whitney U test) in the CSDS group. Individual test results (red dashed lines: cutoff as described in methods) were used for classification as resilient or susceptible phenotype based on symptom load with increased occurrence of the susceptible type after CSDS (P = 0.033, Fisher’s exact test); ctrl n = 23, CSDS n = 26 mice. d Susceptible mice failed whereas resilient mice excelled on the accelerating rotarod compared to controls (maximum time: F_2,687 = 22.03, P < 0.0001; learning speed [LS50]: F_2,687 = 10.08, P < 0.0001; one-way ANOVA with Dunett’s post-hoc test). Performance during the first trial did not differ between the three groups (F_2,687 = 1.574, P = 0.208; one-way ANOVA); ctrl n = 20, resilient n = 12, susceptible n = 14 mice. e Learning the fine motor task of skilled forelimb reaching over 5 days was impaired in stressed mice (time F_2.209,70.70 = 3.606, P = 0.028; stress F_2,32 = 5.211, P = 0.011; interaction F_8,128 = 0.762, P = 0.637, RM ANOVA with Dunett’s post-hoc test); ctrl n = 15, resilient n = 7, susceptible n = 9 mice (group size reduced by task specific exclusions, see methods for details). f Susceptible and resilient stress phenotypes were persistent ~3 weeks after CSDS with only susceptible mice versus controls showing reduced social interaction (H₂ = 10.22, P = 0.006, Kruskal-Wallis test with Dunn’s post-hoc test) and nest building (F_2,43 = 4.852, P = 0.013, one-way ANOVA with Dunett’s post-hoc test). Sucrose consumption did not significantly differ between stressed and control mice as observed before (H₂ = 3.844, P = 0.146, Kruskal-Wallis test); ctrl n = 20, resilient n = 12, susceptible n = 14 mice. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Results are shown as mean ± SEM.

Fig. 2. Multimodal HPA axis response after CSDS discriminates stress phenotypes.

a Plasma corticosterone (CORT) levels were increased 24 h post CSDS (F_2,42 = 5.954, P = 0.005, one-way ANOVA with Holm-Sidak’s post-hoc test); ctrl n = 19, resilient n = 12, susceptible n = 14. b Post-stress plasma CORT changed significantly relative to baseline (dashed line) as tested by one sample t-tests in controls (t₁₃ = 2.365, P = 0.034) and susceptible mice (t₈ = 3.063, P = 0.016) but not in resilient ones (t₉ = 0.843, P = 0.421); ctrl n = 14, resilient n = 10, susceptible n = 9. c Fecal CORT reflects cumulative release 24 h after CSDS and was significantly different between stressed phenotypes and stressed vs. control mice (F_2,41 = 7.897, P = 0.001, one-way ANOVA with Holm-Sidak’s post-hoc test); ctrl n = 20, resilient n = 12, susceptible n = 12 mice. d Adrenal gland weight measured 36 days post CSDS was significantly higher in the susceptible mice compared to controls and resilient mice (F_2,41 = 4.580, P = 0.016, one-way ANOVA with Holm-Sidak’s post-hoc test); ctrl n = 19, resilient n = 12, susceptible n = 14 mice. *P < 0.05, **P < 0.01, ***P < 0.001. Results are shown as mean ± SEM.

Fig. 3. Response and recovery of motor cortical spine dynamics after CSDS.

a Development of spine density from day 2 until 17 post CSDS (time F_3.155,163.3 = 3.425, P = 0.017, stress F_2,64 = 8.324, P = 0.0006, Interaction F_8,207 = 4.211, P = 0.0001, RM ANOVA mixed model with Dunett’s post-hoc test). b, c Change in spine loss but not spine gain drives the change in spine density on day 2 (loss: F_2,64 = 6.784, P = 0.002, one-way ANOVA with Dunett’s post-hoc test, gain: F_2,64 = 0.767, P = 0.487, one-way ANOVA). d Net gain or loss between imaging sessions is differently influenced by stress over time (time F_2.303,109.8 = 2.221, P ≥ 0.05, stress F_2,51 = 0.347, P ≥ 0.05, interaction F_6,143 = 3.164, P ≤ 0.01, RM ANOVA mixed model). e Example time lapse images of dendritic spines at baseline (day −10), day 2 and 11 post CSDS (green arrowhead = new spine, red arrowhead = lost spine). Scale bar: 5 µm). f, g Survival fraction of spines formed after the different motor learning tasks (rotarod [ART]: F_2,48 = 4.918, P = 0.011, reaching: F_2,31 = 4.477, P = 0.020, both one-way ANOVA with Dunett’s post-hoc test). a–d, g, h: no. of ROIs/mice from day −10 to 17: ctrl 26/19 to 23/12, resilient 22/10 to 10/5, susceptible 16/10 to 13/8, see also Supplementary Table 1). *P < 0.05, **P < 0.01, ***P < 0.001. Results are shown as mean ± SEM.

Fig. 4. Motor cortical glia cells show long-term response to CSDS.

a–c Layer- and phenotype-specific reactivity of astrocytes (Layer I–III: stress F_2,220 = 6.500, P = 0.002, reactivity score F_3,220 = 14.30, P < 0.0001, interaction F_6,220 = 1.140, P = 0.340, layer V: stress F_2,220 = 0.999, P = 0.370, reactivity score F_3,220 = 36.78, P < 0.0001, interaction F_6,220 = 0.509, P = 0.801, two-way ANOVA with Holm-Sidak’s post-hoc test when applicable). No. of ROIs/mice: ctrl n = 26/13, resilient n = 18/9, susceptible n = 14/7. d–h Morphological changes of Iba1+ microglia cells in layer I–III are limited to the susceptible group (ramification index: F_2,121 = 3.148, P = 0.047, tree length F_2,121 = 3.576, P = 0.031, spanned area F_2,121 = 3.459, P = 0.035, total area F_2,121 = 4.061, P = 0.020, one-way ANOVA with Holm-Sidak’s post-hoc test). No. of cells/mice: ctrl n = 48/11, resilient n = 45/9, susceptible n = 31/6. i, j Microglia-dendrite colocalization was significantly increased in susceptible mice compared to controls but also markedly increased compared to resilient mice (H₂ = 14.69, P = 0.0006, Kruskal-Wallis test with Dunn’s post-hoc test). No. of dendrites/mice: ctrl n = 43/11, resilient n = 47/9, susceptible n = 36/6). Scale bar: 200 µm in a, 20 µm in h, j. *P < 0.05, ***P < 0.001. Results are shown as mean ± SEM.

Fig. 5. Proteins linked to neurodegeneration and synapses are regulated long-term in CSF after CSDS.

a Workflow for identification of regulated proteins from the CSF sample set (ctrl n = 17, resilient n = 11, susceptible n = 10 mice). b Significant KEGG pathway results for neurodegenerative disorders and their regulated protein groups (gene names shown). c Regulated synaptic protein groups by SynGo database annotation (gene names shown). d Interaction network of the regulated candidates from the neurodegenerative and synaptic annotations.