Sleep fragmentation engages stress-responsive circuitry, enhances inflammation and compromises hippocampal function following traumatic brain injury

Abstract

Traumatic brain injury (TBI) impairs the ability to restore homeostasis in response to stress, indicating hypothalamic-pituitary-adrenal (HPA)-axis dysfunction. Many stressors result in sleep disturbances, thus mechanical sleep fragmentation (SF) provides a physiologically relevant approach to study the effects of stress after injury. We hypothesize SF stress engages the dysregulated HPA-axis after TBI to exacerbate post-injury neuroinflammation and compromise recovery. To test this, male and female mice were given moderate lateral fluid percussion TBI or sham-injury and left undisturbed or exposed to daily, transient SF for 7- or 30-days post-injury (DPI). Post-TBI SF increases cortical expression of interferon- and stress-associated genes characterized by inhibition of the upstream regulator NR3C1 that encodes glucocorticoid receptor (GR). Moreover, post-TBI SF increases neuronal activity in the hippocampus, a key intersection of the stress-immune axes. By 30 DPI, TBI SF enhances cortical microgliosis and increases expression of pro-inflammatory glial signaling genes characterized by persistent inhibition of the NR3C1 upstream regulator. Within the hippocampus, post-TBI SF exaggerates microgliosis and decreases CA1 neuronal activity. Downstream of the hippocampus, post-injury SF suppresses neuronal activity in the hypothalamic paraventricular nucleus indicating decreased HPA-axis reactivity. Direct application of GR agonist, dexamethasone, to the CA1 at 30 DPI increases GR activity in TBI animals, but not sham animals, indicating differential GR-mediated hippocampal action. Electrophysiological assessment revealed TBI and SF induces deficits in Schaffer collateral long-term potentiation associated with impaired acquisition of trace fear conditioning, reflecting dorsal hippocampal-dependent cognitive deficits. Together these data demonstrate that post-injury SF engages the dysfunctional post-injury HPA-axis, enhances inflammation, and compromises hippocampal function. Therefore, external stressors that disrupt sleep have an integral role in mediating outcome after brain injury.

Keywords: HPA-axis; Neuroinflammation; Schaffer collateral; Sleep fragmentation; Stress; Traumatic brain injury; fear conditioning; glucocorticoid receptor; hippocampus.

Figure 1.. TBI increases cortical microgliosis 7 DPI.

(A) 8–10-week-old male and female C57BL/6 mice received sham or TBI injury and assigned to remain undisturbed or receive 4h mechanical SF for 7 days. (B) TBI significantly increased righting time compared to sham (main effect TBI, p < 0.01). (C) All groups lost weight following sham or TBI but recovered to baseline with no effect of TBI or SF (main effect DPI, p < 0.01). (D) Representative images of Iba-1 labelling of lateral (primary somatosensory) and medial (retrosplenial) ipsilateral cortex to indicate microglial/macrophage responses to TBI and SF. (E, F) TBI increased percent area of Iba-1 labelling both lateral and medial to lesion regardless of SF (main effect TBI, p < 0.01). Righting times and weight changes, n = 12–14/group. Immunolabelling, n = 7–8/group; scale bar indicates 200 μm; individual values represent averages of 2–3 images per region per animal; error bars indicate standard error of the mean (SEM); * = p < 0.05, ** = p < 0.01.

Figure 2.. Post-TBI SF enhances TBI-induced cortical inflammation through increased interferon- and stress-associated genes 7 DPI.

(A) Heatmap of standardized Z-score with genes significantly altered in the ipsilateral cortex by SF compared to CON in both sham and TBI groups. (B) Heatmap of standardized Z-score with cortical genes uniquely altered by Sham SF compared to Sham CON. (C) Heatmap of standardized Z-score with cortical genes uniquely altered by TBI SF compared to TBI CON. (D) Summary of genes changed with SF compared to CON in Sham and TBI groups (red = increased, blue = decreased). (E) Heatmap of standardized Z-score with cortical genes significantly altered by both TBI CON and TBI SF compared to Sham CON. (F) Heatmap of standardized Z-score with cortical genes uniquely altered by TBI CON compared to Sham CON. (G) Heatmap with standardized Z-score of cortical genes uniquely altered by TBI SF compared to Sham CON. (H) Summary of cortical gene expression changes with TBI compared to Sham CON in CON and SF groups (red = increased, blue = decreased). (I) Ingenuity Pathway Analysis (IPA) of top upstream regulators inhibited (Activation Z-score ≤ −2) and activated (Activation Z-score ≥ 2) with Sham SF, TBI CON, and TBI SF compared to Sham CON. n = 6/group Sham CON, TBI CON, TBI SF, n = 5/group Sham SF.

Figure 3.. TBI increases hippocampal microgliosis, but post-TBI SF uniquely increases CA1 neuronal activity 7 DPI.

(A) Representative images of Iba-1 and ΔFosB labelling to indicate microgliosis and neuronal activity, respectively, in ipsilateral CA1. (B) TBI increased Iba-1 percent area labelling regardless of SF (main effect TBI, p < 0.01). (C) TBI SF animals had the highest number of ΔFosB⁺ cells in the ipsilateral CA1 compared to all other groups (Tukey multiple comparisons, p < 0.05 to Sham CON and Sham SF, p < 0.01 to TBI CON). (D) Representative images of Iba-1 and ΔFosB labelling in ipsilateral CA3. (E) TBI increased Iba-1 percent area of the ipsilateral CA3 (main effect TBI, p < 0.01). (F) There were no differences in CA3 neuronal activity indicated by number of ΔFosB⁺ cells. (G) Representative images of ΔFosB labelling in the PVN. (H) Sham SF animals had a higher mean ΔFosB⁺ cells in the PVN than Sham CON (main effect SF, p < 0.05), indicating increased PVN neuronal activity. n = 7–8/group; scale bar indicates 200 μm; individual values represent averages of 2 images per region per animal; error bars indicate SEM; * = p < 0.05, ** = p < 0.01.

Figure 4.. Post-TBI SF causes persistent microgliosis associated with increased cortical expression of glial pro-inflammatory genes 30 DPI.

(A) 8–10-week-old male and female C57BL/6 mice received sham or TBI injury and assigned to remain undisturbed or receive 4h mechanical SF for 30 days. (B) TBI significantly increased righting time compared to sham (main effect TBI, p < 0.01). (C) All groups lost weight following sham or TBI, however all groups quickly recovered to baseline (main effect DPI, p <¸0.01) with animals exposed to SF having slightly increased weight gain compared to animals left undisturbed (DPI by SF, p < 0.05). (D) Representative images of Iba-1 labelling of lateral (primary somatosensory) and medial (retrosplenial) ipsilateral cortex. (E) Post-TBI SF increased percent area of Iba-1 lateral to the lesion area compared to all other groups (Tukey multiple comparisons, p < 0.01 to Sham CON and Sham SF, p < 0.05 to TBI CON). (F) TBI increased percent area of Iba-1 labelling medial to lesion area regardless of SF (main effect TBI, p < 0.01). (G) Heatmap of standardized Z-score with genes uniquely altered in ipsilateral cortex by Sham SF compared to Sham CON. (H) Heatmap of standardized Z-score with cortical genes unique altered by TBI SF compared to TBI CON. (I) Summary of cortical gene expression changes with SF compared to CON in Sham and TBI groups (red = increased, blue = decreased). There was no overall SF effect. (J) Heatmap of standardized Z-score with cortical genes significantly altered by TBI CON and TBI SF compared to Sham CON. (K) Heatmap of standardized Z-score with cortical genes uniquely altered by TBI CON compared to Sham CON. (L) Heatmap of standardized Z-score with genes uniquely altered by TBI SF compared to Sham CON. (M) Summary of cortical gene expression changes with TBI compared to Sham CON in CON and SF groups (red = increased, blue = decreased). (N) IPA of top upstream regulators inhibited (Activation Z-score ≤ −2) and activated (Activation Z-score ≥ 2) with Sham SF, TBI CON, and TBI SF compared to Sham CON. Righting times and weight changes, n = 12–14/group. Immunolabeling, n = 6–7/group; scale bar indicates 200 μm; individual values represent averages of 2–3 images per region per animal; error bars indicate SEM; * = p < 0.05, ** = p < 0.01. NanoString n = 7/group Sham CON, n = 5/group TBI CON, n = 6/group Sham SF and TBI SF.

Figure 5.. Post-TBI SF enhances ipsilateral CA1 microglia reactivity and imbalanced neuronal activity 30 DPI.

(A) Representative images of Iba-1 and ΔFosB labelling in ipsilateral CA1. (B) Post-TBI SF increased Iba-1 percent area in the ipsilateral CA1 compared to the Sham SF group (Tukey multiple comparisons, p < 0.05). (C) Sham SF animals had the highest number of ΔFosB⁺ cells in the ipsilateral CA1 compared to all other groups (Tukey multiple comparisons, p < 0.01 to Sham CON and TBI CON, p < 0.05 to TBI SF). (D) Representative images of Iba-1 and ΔFosB labelling in ipsilateral CA3. (E) There was a main effect of Injury by Sex (p < 0.05), but no significant differences between groups (Tukey multiple comparisons, p > 0.05) in iba-1 percent area of the ipsilateral CA3. (F) SF and TBI increased ΔFosB⁺ cells in a sex-dependent manner in the ipsilateral CA3 (main effect TBI and SF, p < 0.01; main effect sex, p < 0.05) with TBI SF animals of both sexes having the highest mean number of ΔFosB⁺ cells, indicating the highest neuronal activity. Immunolabeling, n = 6–7/group; scale bar indicates 200 μm; individual values represent averages of 2 images per region per animal; error bars indicate SEM; * = p < 0.05, ** = p < 0.01.

Figure 6.. Altered BNST and PVN neuronal activity in TBI SF mice associates with enhanced DEX-induced GR activity in the CA1 30 DPI.

(A) Representative images of ΔFosB labelling in the posterior BNST. (B) TBI increased ΔFosB⁺ cells in the BNST (main effect TBI, p < 0.01) with male TBI SF animals having the highest number of ΔFosB⁺ cells in the ipsilateral posterior BNST compared to all over male groups and both female Sham CON and Sham SF groups (Tukey multiple comparisons, p < 0.05). (C) Representative images of PVN ΔFosB labelling. (D) Sham SF animals had the highest neuronal activity of the PVN, indicated by ΔFosB⁺ cells, compared to all other groups (Tukey multiple comparisons, p < 0.01). (E) A separate cohort of 8–10-week-old male and female C57BL/6 mice received sham or TBI injury and assigned to remain undisturbed or receive 4h SF for 30 days. On 30 DPI, animals were given Veh or DEX injection to the ipsilateral dorsal CA1 and then sacrificed 30 minutes post-injection to determine pGR activity. (F) Representative images of pGR labelling in the dorsal ipsilateral CA1 at site of injection. Dashed white lines indicate quantified area of CA1. Inset is a representative cell with pGR labelling. (G) DEX increased pGR mean fluorescent intensity (MFI) only in TBI animals with a trending (TBI by DEX, p < 0.05; Bonferroni multiple comparisons, p = 0.06) increase with post-TBI SF compared to Sham SF groups. (A-D) Immunolabeling, n = 6–7/group; scale bar indicates 200 μm; individual values represent averages of 2 images per region per animal; error bars indicate SEM; * = p < 0.05, ** = p < 0.01. (E-G) Immunolabeling, n = 5–6/group; scale bar indicates 200 μm; individual values represent averages of 2 images per region per animal; error bars indicate SEM; * = p < 0.05, ** = p < 0.01.

Figure 7.. Injury and SF induce Schaffer collateral deficits 30 DPI.

(A) 8–10-week-old male and female C57BL/6 mice received sham or TBI injury and assigned to remain undisturbed or receive 4h SF for 30 days. Electrophysiological assessment of ipsilateral Schaffer collateral was performed 31 DPI. (B) PPR showed no difference between groups. (C) TBI SF decreased I/O (stimulus by TBI effect, p < 0.05), indicating dysfunction of the Schaffer collateral CA1 with TBI SF animals have the lowest mean I/O. (D) Summary of fEPSP relative to baseline recordings before and after tetanic stimulation of the ipsilateral Schaffer collateral. TBI and SF decreased fEPSP responses following tetanic stimulation (time by TBI; time by SF, p < 0.01). TBI SF animals had the lowest initiation of LTP that persistently remained low. Electrophysiological assessment, n = 6/group Sham CON, Sham SF, TBI CON, n = 5/group TBI SF; individual values represent averages of 2–5 slices per animal; error bars indicate SEM; * = p < 0.05, ** = p < 0.01.

Figure 8.. Post-TBI SF causes trace fear condition acquisition deficits consistent with compromised dorsal hippocampus function.

(A) A separate cohort of 8–10-week-old male and female C57BL/6 mice received sham or TBI injury and assigned to remain undisturbed or receive 4h SF for 30 days. Behavioral testing was conducted throughout the 30 days of SF: open field 7 DPI, Y-maze 25 DPI, and trace fear conditioning 26–28 DPI. (B) TBI nor SF influence time spent in center of open field, (C) but animals exposed to SF regardless of injury had decreased rearing episodes (Main effect SF, p < 0.05). TBI and SF did not influence number of arm entries in Y-maze spatial learning task (D) not percent of spontaneous alternations (E). Trace fear conditioning determine dorsal hippocampus-dependent learning and memory behavior. (F) Trace fear conditioning protocol consisted of a two-minute habituation, 20s tone, 18s delay followed by 2s shock (termed trace phase), and two minutes inter-trial interval (ITI). (G) Post-TBI SF caused unique deficits in trace fear conditioning acquisition (TBI by SF by phase, p < 0.01) with significantly less freezing behavior compared to both Sham SF (*) and TBI CON (#) during tone3 and trace3 phases (Bonferroni pairwise comparisons, p < 0.05). The TBI SF group also had less freezing than Sham SF in ITI2, ITI3, and tone4 phases of acquisition and less freezing than TBI CON during the trace5 phase (Bonferroni pairwise comparisons, p < 0.05). (H) Freezing behavior specifically during the tone phase is an indicator of short-term fear memory. TBI SF specifically decreased freezing behavior compared to both Sham SF and TBI CON groups during tone 3 and the Sham SF group (TBI by SF by phase, p < 0.05; Bonferroni pairwise comparisons p < 0.05). This deficit was unique to learning in acquisition phase as no significant deficits occurred in context-dependent (I) and cued-dependent (J) freezing behavior. n = 10–12/group; error bars indicate SEM; * = p < 0.05, ** = p < 0.01.