Concussion susceptibility is mediated by spreading depolarization-induced neurovascular dysfunction

Abstract

The mechanisms underlying the complications of mild traumatic brain injury, including post-concussion syndrome, post-impact catastrophic death, and delayed neurodegeneration remain poorly understood. This limited pathophysiological understanding has hindered the development of diagnostic and prognostic biomarkers and has prevented the advancement of treatments for the sequelae of mild traumatic brain injury. We aimed to characterize the early electrophysiological and neurovascular alterations following repetitive mild traumatic brain injury and sought to identify new targets for the diagnosis and treatment of individuals at risk of severe post-impact complications. We combined behavioural, electrophysiological, molecular, and neuroimaging techniques in a rodent model of repetitive mild traumatic brain injury. In humans, we used dynamic contrast-enhanced MRI to quantify blood-brain barrier dysfunction after exposure to sport-related concussive mild traumatic brain injury. Rats could clearly be classified based on their susceptibility to neurological complications, including life-threatening outcomes, following repetitive injury. Susceptible animals showed greater neurological complications and had higher levels of blood-brain barrier dysfunction, transforming growth factor β (TGFβ) signalling, and neuroinflammation compared to resilient animals. Cortical spreading depolarizations were the most common electrophysiological events immediately following mild traumatic brain injury and were associated with longer recovery from impact. Triggering cortical spreading depolarizations in mild traumatic brain injured rats (but not in controls) induced blood-brain barrier dysfunction. Treatment with a selective TGFβ receptor inhibitor prevented blood-brain barrier opening and reduced injury complications. Consistent with the rodent model, blood-brain barrier dysfunction was found in a subset of human athletes following concussive mild traumatic brain injury. We provide evidence that cortical spreading depolarization, blood-brain barrier dysfunction, and pro-inflammatory TGFβ signalling are associated with severe, potentially life-threatening outcomes following repetitive mild traumatic brain injury. Diagnostic-coupled targeting of TGFβ signalling may be a novel strategy in treating mild traumatic brain injury.

Keywords: biomarker; blood–brain barrier; concussion; dynamic contrast-enhanced MRI; repetitive mild traumatic brain injury.

Figure 1

Modelling repetitive mild TBI. (A) TBI administration using a weight-drop method produced substantial rotational motion upon impact. (B) Ten minutes after sustaining a single mild TBI, rats had a lower composite neurobehavioural score (open field, beam walk, inverted wire mesh tests), which returned to baseline 24 h post-injury [one-way ANOVA: F(2,262) = 11.53, P < 0.0001, Tukey-corrected two-tailed Student's t-tests: 10 min versus baseline: P < 0.0001, 10 min versus 24 h: P = 0.01, baseline versus 24 h: P = 0.13; baseline n = 92, 10 min n = 81, 24 h n = 92]. Sham controls did not differ in neurobehavioural scoring at 10 min or 24 h post-injury compared to baseline [one-way ANOVA: F(2,94) = 2.83, P = 0.07; baseline n = 34, 10 min n = 29, 24 h n = 34]. (C) Repetitive mild TBI animals (n = 111) showed a marked decline in neurobehavioural score upon repetitive impacts compared to controls (n = 37) after three or four impacts (Mann-Whitney tests with false discovery rate correction: baseline: P = 0.67, 24 h: P = 0.4, 48 h: P = 0.051, 72 h P = 0.01, 96 h P < 0.0001). (D) Immediate post-impact convulsions were observed in 16% (n = 5) of animals after a single injury and increased to 53% (n = 17) of animals after two impacts [chi-square test for trend: χ²(1,32) = 7.51, P = 0.006, n = 32]. Sham controls did not display convulsive movements. (E) Post-impact mortality increased with exposure to repetitive impacts, occurring in 2.7% (n = 3 of 111) of animals after the third impact and in 8% (n = 9 of 111) of animals after the fourth impact [chi-square test for trend: χ²(1,111) = 15.76, P < 0.0001, n = 111].

Figure 2

Heterogeneous outcomes following repetitive mild TBI. (A) The distribution of neurobehavioural scores did not differ between repetitive mild TBI and sham control animals at baseline (left) or after a single mild TBI (middle). After four impacts, neurobehavioural scores of repetitive mild TBI animals showed a bimodal (rather than a normal) distribution (Shapiro-Wilk test: P < 0.0001), with a trough value of 6 (right). (B) Left: Retrospective grouping of ‘susceptible’ and ‘resilient’ animals revealed neurobehavioural score decline in susceptible, but not resilient or sham control animals after three [72 h: one-way ANOVA: F(2,122) = 8.91, P = 0.0002; Tukey-corrected two-tailed Student's t-test: susceptible versus control: P = 0.0006, resilient versus control: P = 0.83, susceptible versus resilient: P = 0.006; control n = 37, susceptible n = 54, resilient n = 34] and four impacts [96 h: one-way ANOVA: F(2,113) = 18.47, P < 0.0001; Tukey-corrected two-tailed Student's t-test: susceptible versus control: P < 0.0001, resilient versus control: P = 0.21, susceptible versus resilient: P = 0.0004; control n = 37, susceptible n = 45, resilient n = 34]. Middle: Susceptible animals (but not resilient) showed a progressive increase in duration of post-impact convulsions between the first and second, and second and third impacts [susceptible: one-way ANOVA: F(3,65) = 3.64, P = 0.017, Tukey-corrected two-tailed Student's t-tests: TBI 1 versus 2: P = 0.01, TBI 2 versus 3: P = 0.002; susceptible n = 20, resilient n = 12]. Right: Susceptible animals showed a longer latency to regain the righting reflex compared to controls (n = 14) after one [one-way ANOVA: F(2,43) = 3.70, P = 0.03; Tukey-corrected two-tailed Student's t-test: susceptible versus control: P = 0.025; susceptible n = 20, resilient n = 12, control n = 14], two [one-way ANOVA: F(2,43) = 6.43, P = 0.004; Tukey-corrected two-tailed Student's t-test: susceptible versus control: P = 0.003], or three impacts [one-way ANOVA: F(2,41) = 9.23, P = 0.0005; Tukey-corrected two-tailed Student's t-test: susceptible versus control: P = 0.003]. Susceptible animals had a longer righting reflex latency than resilient animals after three impacts [one-way ANOVA: F(2,41) = 9.23, P = 0.0005; Tukey-corrected two-tailed Student's t-test: susceptible versus resilient: P = 0.002]. (C) The mean pressure delivered during each impact (detected with pressure sensitive film) did not differ between susceptible and resilient animals (unpaired Student's t-tests with false discovery rate correction: Impact 1: P = 0.7; susceptible n = 37, resilient n = 11; Impact 2: P = 0.7; susceptible n = 37, resilient n = 11; Impact 3: P = 0.5; susceptible n = 21, res n = 11; Impact 4: P = 0.2; susceptible n = 15, resilient n = 11; Impact 5: P = 0.7; susceptible n = 3, resilient n = 11). (D) Following administration of repetitive impacts, susceptible animals showed a reduction in daily weight gain [one-way ANOVA F(2,78) = 27.09, P < 0.0001, Tukey-corrected two-tailed Student's t-tests: susceptible versus resilient: P < 0.0001, susceptible versus control: P < 0.0001, resilient versus control: P = 0.085; susceptible n = 39, resilient n = 22, control n = 20] compared to sham control and resilient animals, which continued to gain weight during their adolescent phase of growth. (E) No evidence of contusions or haemorrhage was apparent in susceptible, resilient, or control animals after a single mild TBI. (F) Post-mortem brain extraction revealed frequent (71%, n = 12 of 17) subcutaneous haematoma but no skull fractures or intracerebral haemorrhage. Susceptible (55%, n = 6 of 11), but not resilient (n = 6), animals frequently showed evidence of epidural, subdural, or subarachnoid bleeding (white arrows).

Figure 3

CSDs are the earliest electrophysiological events following mild TBI and can induce BBBD. (A) ECoG recordings obtained immediately following mild TBI showed a large amplitude, long-duration, propagating slow potential change associated with depression of cortical activity in 53% (n = 37 out of 71) impacts, consistent with the occurrence of spreading depolarization (duration indicated with green bars) recorded in both left (black trace) and right (blue trace) hemispheres. Scale bar = 60 s. Trace units = mV. (B) Time to regain locomotor activity was longer when CSDs were recorded following mild TBI (mean 360 ± 26 s, n = 32 impacts) compared to impacts without CSDs (194 ± 24 s, n = 27) or after isoflurane anaesthesia without impact (67 ± 6 s, n = 30) [one-way ANOVA, F(2,86) = 15.34, P < 0.0001, Tukey-corrected two-tailed Student's t-tests CSD versus control: P < 0.0001; CSD versus no CSD: P < 0.0001; no CSD versus control: P = 0.0002]. (C) BBB permeability was assessed before and after triggered CSDs using fluorescent angiography and intravital microscopy. TBI animals showed evidence of increased BBB permeability to peripherally injected sodium fluorescein within 1 h following CSD. (D) BBB permeability increased (% change) in TBI animals following triggered CSDs (n = 15) but not in controls (n = 17; Mann–Whitney test, P = 0.0007).

Figure 4

Repetitive mild TBI leads to BBBD and neuroinflammation. (A) Slope values, reflecting a gradual change in brain signal following the injection of a non-BBB-permeable contrast agent, were calculated for each brain voxel across sequential scans. (B) A cumulative frequency histogram of slope values was constructed and showed a rightward shift for repetitive mild TBI animals 1 week (n = 19) post-impact compared to controls (n = 15) or 24 h (n = 4) post-impact. (C) Above-threshold permeability values are plotted on T₂-weighted MRI scans of control, susceptible, and resilient animals 1 week after exposure to repetitive mild TBI. (D) Susceptible animals had a higher percentage of brain volume with pathological voxels than sham controls injury. Resilient animals did not differ in BBBD extent from controls [one-way ANOVA F(2,31) = 4.62, P = 0.018, Tukey-corrected two-tailed Student's t-tests: susceptible versus control: P = 0.01, resilient versus control: P = 0.5; susceptible n = 12, resilient n = 7, control n = 15]. (E) Representative images of GFAP and Iba-1 immunofluorescent staining in post-mortem hippocampal tissue collected from control (n = 6), susceptible (n = 6), and resilient (n = 5) animals 1 week after repetitive mild TBI. Scale bar = 100 μm. (F) Quantification of total immunofluorescence intensity revealed higher GFAP and Iba-1 intensity in susceptible compared to resilient and control animals [GFAP: one-way ANOVA F(2,15) = 10.76, P = 0.013, Tukey-corrected two-tailed Student's t-tests: susceptible versus control: P = 0.002, susceptible versus resilient: P = 0.004, resilient versus control: P = 0.97; susceptible n = 6, resilient n = 6, control n = 6; Iba-1: one-way ANOVA F(2,15) = 6.16, P = 0.011, Tukey-corrected two-tailed Student's t-tests: susceptible versus control: P = 0.037, susceptible versus resilient: P = 0.014, resilient versus control: P = 0.87; susceptible n = 6, resilient n = 6, control n = 6]. (G) Quantification of IL-1β, IL-6, and TNFα mRNA showed elevated expression of IL-6 mRNA in susceptible animals compared to resilient and control animals [IL-6: one-way ANOVA F(2,20) = 5.25, P < 0.015, Tukey-corrected two-tailed Student's t-tests: susceptible versus resilient: P = 0.04, susceptible versus control: P = 0.02, resilient versus control: P = 0.98; susceptible n = 8, resilient n = 7, control n = 8]. No differences were found between IL-1β [one-way ANOVA F(2,20) = 0.11, P = 0.9] and TNFα [one-way ANOVA F(2,20) = 1.79, P = 0.2] expression levels. (H) HMGB1 protein expression was assessed by western blot, revealing greater levels in susceptible but not resilient animals compared to controls [one-way ANOVA F(2,18) = 5.86, P < 0.01, Tukey-corrected two-tailed Student's t-tests: susceptible versus control: P = 0.009, resilient versus control: P = 0.2, susceptible versus resilient: P = 0.2; susceptible n = 9, resilient n = 7, control n = 5].

Figure 5

Repetitive mild TBI susceptibility is associated with hippocampal TGFβ signalling. (A) TGFβ signalling activity was quantified via western blot by comparing the ratio of pSmad2 and Smad2 protein expression levels. Susceptible animals showed higher TGFβ signalling activity than resilient or control animals [one-way ANOVA F(2,18) = 20.98, P < 0.0001, Tukey-corrected two-tailed Student's t-tests: susceptible versus control: P = 0.0006, susceptible versus resiliant: P < 0.0001, resiliant versus control: P = 0.72; susceptible n = 9, resiliant n = 7, control n = 5]. (B) pSmad2:Smad2 levels were higher in repetitive mild TBI animals compared to controls but not in animals treated with the selective TGFβ receptor inhibitor, IPW-5371, indicating successful target engagement and reduced TGFβ signalling with IPW treatment [one-way ANOVA F(2,10) = 18.80, P = 0.0004, Tukey-corrected two-tailed Student's t-tests: vehicle versus control: P = 0.0003, IPW versus control: P = 0.15, IPW versus vehicle: P = 0.002; IPW n = 6, vehicle n = 3, control n = 4]. (C) IPW-treated animals have a lower percentage of brain volume with pathological voxels than vehicle controls (Mann–Whitney test: P = 0.03, IPW n = 9, control n = 12). (D) The duration of post-impact convulsions was shorter in animals treated with IPW following impacts two (after receiving the first dose of drug) to five compared to vehicle controls [Mann–Whitney test: P = 0.004; IPW n = 14, vehicle n = 8]. (E) IPW-treated animals had a shorter latency to regain the righting reflex compared to vehicle controls and did not differ from sham (anaesthesia-only) controls [Kruskal-Wallis test: F(2,106) = 21.97, P < 0.0001 with Dunn post hoc: control versus vehicle P < 0.0001, control versus IPW P = 0.1, vehicle versus IPW P = 0.01; control n = 56, IPW n = 30, vehicle n = 20]. (F) When tested between 1 and 2 months post-impact, vehicle TBI animals had lower neurobehavioural scores than isofluorane-only (sham) controls. Treatment with IPW resulted in higher neurobehavioural scores compared to vehicle-treated animals, and IPW animals did not differ from sham controls [one-way ANOVA F(2,51) = 7.10, P = 0.002, Tukey-corrected two-tailed Student's t-tests: vehicle versus control: P = 0.002, vehicle versus IPW: P = 0.03, IPW versus control: P = 0.18; IPW n = 30, vehicle n = 13, control n = 11].

Figure 6

BBBD in contact-sport athletes. (A) Assessment of BBB integrity by DCE-MRI in 26 contact-sport athletes with a history of concussion (mean number of reported concussions: 2.6, 95% CI: 1.7–3.5, min-max: 1–6) and 61 age-matched controls with no concussion history showed diffuse areas of increased BBB leakage in contact-sport athletes. (B) Regional analysis of brain areas with greater BBBD (>2 SD of controls) showed that concussed players deviated from a normal distribution (Shapiro-Wilk test: P = 0.004), with a subset of concussed players having more regions with significant BBBD. (C and D) The percentage of players or controls with above-threshold BBBD in various brain areas is shown. BBBD in concussed players, but not controls, was frequently localized to the right hemisphere, with players sharing common brain regions of BBBD. (E and G) Repeated DCE-MRI scans in two mixed martial arts athletes showed increased brain volume with BBBD 1 week after mild TBI compared to baseline. (F and G) Repeated DCE-MRI scans in four American football players showed a reduction in brain volume with BBBD 1 month after concussion compared to 1 week after injury in three players, with one player showing greater BBBD 1 month post-injury.