Neurological and cognitive assessment
Traumatic brain injury affects motor and sensory performance.
We evaluated the performance of mice on the pole climbing test. At 48h, mice with mild and severe TBI performed significantly worse than sham-injured mice as indicated by the total time taken by the animals to descend the pole (mild:11.7 ± 0.6 and severe: 21.4 ± 3.6 vs. 6.4 ± 0.8 seconds, P < 0.001 and P < 0.0001 respectively). Less time is indicative of better motor coordination and balance. Both TBI groups showed recovered motor performance at 7- and 35-days post-injury (dpi) in comparison to their performance at 48h after injury (7d: 7.1 ± 0.4 and 8.7 ± 1, P < 0.05 and P < 0.01 respectively; 35d: 8.9 ± 2 and 8.1 ± 1, P < 0.01) (Figure 1, a-d).
Furthermore, we found group differences among the sham, mTBI, and sTBI groups in time to sense and remove the stimuli on the adhesive removal test (Figure 1, e-f). At 48h post-TBI, the mean time to sense and remove the tape was significantly higher in mild and severe-TBI mice than the sham group (mild:21.4 ± 3 and severe: 25.9 ± 4 vs. sham: 8.4 ± 1.5 seconds, P < 0.05 and P < 0.01 respectively). The sensory deficit persisted on day 7 in both injury groups (mild:17.7 ± 5.8, and severe: 20.9 ± 3.5 vs. sham: 8.2 ± 3.7 seconds, P < 0.01). However, 35 days post-TBI, only the group that received severe injury showed persistent sensory deficits compared to sham (15.4 ± 2.5 vs. 7.9 ± 3.5 seconds, P < 0.05).
Severe traumatic brain injury causes long-term cognitive deficits.
Spontaneous object recognition test
During the sample phase (learning trial), the time spent acquiring information about the objects, such as size and shape, was measured as exploration time. To exclude position preference, the mean time spent exploring each object in the 3 corners of the box within each group was compared (phase 1), and no significant differences were found (data not shown). When comparing the time spent exploring the different objects during this phase, the sham group spent more time exploring all the objects at 48 h compared to animals with mild and severe TBI: (sham total exploratory time 24.9 ± 3.8 vs. mild 7.4 ± 2.7 and severe 12.9 ± 2.5 seconds, P < 0.0001 and P < 0.01 respectively).
On the Novelty test (phase 2), one familiar object was replaced with a novel one. The exploration time of the remaining two familiar objects and the novel one was compared. At all time-points, the sham animals spent more time exploring the novel object than animals with mild and severe TBI (Figure 2, a, c and e).
The same novel object was then moved to a novel position on the novel location test (phase 3). At all time-points, the sham animals spent more time exploring the novel object/location than the animals with mild and severe TBI (Figure 2, b, d, and f). Animals with mild and severe TBI, at 7- and 35-days post-trauma, showed persistent impaired spontaneous object recognition compared to the sham animals.
The Morris water maze test
Both the sham and mTBI mice showed similar learning curves 31-35 days post-injury. However, as expected, a longer latency to the platform was found after a severe TBI compared to the Sham and mTBI on all days. The group with severe TBI showed little learning during the four trial days, indicating impaired spatial learning that persisted up to 35 days (Figure 3, a). When examining the percentage time spent in the target quadrant and the latency to correct quadrant (data not shown), the animals with mild injury were similar to the sham group, unlike the animals post severe TBI who remained significantly impaired. The same was found when we examined the percentage of time spent swimming in the correct quadrant on the probe trial (Figure 3, b) (day 5). The animals with mTBI performed like the sham and remembered where the platform was located, whereas the animals with a severe injury did not, indicating impaired spatial memory (Figure 3, c).
Microglia Isolation and typing
Conventional Flow cytometry analysis identified microglia and infiltrating macrophages.
To investigate the dynamics of microglial activation following TBI, we isolated microglia from different brain regions: hippocampus, thalamus, and cortex ipsilateral and contralateral to the injury site. Live cells were stained with anti-CD11b, CD45 and CX3CR1. Figure 4 a-b shows representative flow cytometry scatter plots that were used to measure the frequency of microglia in the different parts of the brain. Interestingly, starting at 48 hours post-TBI the frequency of isolated microglia (CX3CR1hi) increased in all regions studied in both mild and severe TBI groups. In the mild TBI group the increase in microglia was significant in the ipsilateral cortex at 48 hours, and in both ipsi- and contralateral cortices at 7 days. While in the severe TBI group, a significant increase was noted in all investigated regions at 48 hours. This increase in frequency remained significant in the cortical areas and hippocampi on days 7 and 35 (except for the ipsilateral cortex on day 35). The thalamus showed significantly increased microglia frequency at 48 hours bilaterally, but only the increase in the ipsilateral thalamus remained significant at day 7 and the contralateral thalamus on day 35. Furthermore, myeloid cells (CD45hi CX3CR1lo) likely representing infiltrating macrophages were detected at 48h in the ipsilateral side in the three investigated brain regions of the severe TBI group only. These cells were not detected at subacute and chronic time points. The increase of microglia numbers in both hemisphere of injured mice reflects a diffuse injury that extends beyond the injury site.
Microglia activation markers, mainly CD11b, CD206, TNFa and AIEI, were found upregulated at 48h post-injury. Even though the markers were increased in both hemispheres, the highest MFIs were noted on the ipsilateral side. As one would expect, microglia markers followed the same dynamics as observed for the cell frequencies in figure 4 c. almost all activation markers that were upregulated at the early time point return to baseline on day 7. However, on day 35 we observed a re-emergence of these markers in the brains of the severe TBI group. The increase of IAIE, which is considered a marker of activation in microglia 20, at this later time point suggests that microglia are chronically activated. The activation was noticeable at the level of the ipsilateral Hippocampus and the contralateral thalamus and cortex (figure 5). TNFa, the inflammatory cytokine, was found to follow the same pattern as Class II, suggesting that the chronically activated microglia are likely proinflammatory (Data not shown).
Non-linear Dimensionality Reduction Reveals Multiple Distinct Microglia Cell States
To investigate the phenotypic heterogeneity and dynamics of cells after traumatic brain injury, we used the combination of 8 markers projected into the two-dimensional space followed by cluster analysis. This method allowed the characterization of several subpopulations of cells.
The markers CD11b and TNFa contributed the most to the model as was shown by the sum of square error (Figure 6, a). We identified a population of infiltrating macrophages (CD45hi, CD11bhi, CX3CR1lo, CD68-, F4/80-, CD206hi, TNFa-, IA/IEhi) and three populations of microglia that were characterized by a relatively high expression of CX3CR1 and CD68. These cells could be divided into three clusters: microglia cluster 1 (CD45+, CD11bhi, CX3CR1++, CD68++, F4/80lo, CD206+, TNFa-, IA/IE+) microglia cluster 2 (CD45+, CD11blo, CX3CR1++, CD68++, F4/80lo, CD206+, TNFa+, IA/IE+), and highly activated microglia (CD45++, CD11bhi, CX3CR1hi, CD68hi, F4/80+, CD206 hi, TNFa++, IA/IE ++) (Figure 6, b).
Clusters 1 and 2 are similar for most markers, except for significantly higher CD11b and lower TNFa expression in cluster 1 compared to cluster 2. The third cluster, representing highly activated microglia, is characterized by increased expression of almost all the markers, especially TNFa and IAIE (Figure 6, c).
In the non-injured brains, the relative proportions of identified clusters were different between brain regions. The predominant population is from cluster 1, with a smaller percentage of cluster 2 cells. Interestingly, highly activated microglia are also found at a very low frequency throughout the different parts of the control brains (Figure 7). At 48 hours after TBI, we observed an increase in highly activated microglia (cluster 3) in the cortex of mild TBI animals and all brain regions of severe TBI animals. These cells resolved by day 7 in both TBI groups. Interestingly, cluster 3 cells reappeared on day 35 in the cortex and subcortical area of severe TBI mice. Furthermore, we have documented a switch in abundance between cluster 1 and cluster 2 at the injury site 48 hours dpi (Figure 7). Although cluster 2 decreases at the later time points in animals with severe TBI, these cells persist with a frequency significantly higher than in the sham group. Also, we noted an equivalent increase of cluster 2 in both the mild TBI and severe TBI 7 days post injury. However, the level of cluster 2 goes down in the mTBI group at 35 days but remains significantly higher in the severe TBI group compared to the rest of the groups. The increase in cluster 2 types of microglia and highly activated microglia at the acute time point suggests a proinflammatory environment early after the injury in both mild and severe TBI.
The highly activated microglia correlate with learning deficit.
To explore the association between the changes in microglial phenotype with spatial learning and memory, we performed the Morris water maze on day 31 to 34 followed by microglia isolation in 24 mice included in the four experimental groups, normal, sham, mild TBI and severe TBI. The aim here was to investigate microglial phenotype within the hippocampus and thalamus as they are critical in the formation and processing of spatial learning and memory 21. As expected, mice with severe TBI showed significant spatial learning and memory deficits (Figure 8, a). Furthermore, these mice showed increased highly activated microglia at 35 days post-injury (Figure 8, b). Using general linear regression, we found that the frequency of the highly activated microglia explained up to 47% of the variability observed in the cumulative latency to the platform, indicating a possible relationship between the existence of activated microglia and spatial learning/memory deficit chronically after the injury (Figure 8, c & d). Microglia clusters 1 and 2 did not contribute much to the model and showed no correlation with cumulative latency to platform.