Images were captured with a Zeiss AxioObserver inverted widefield microscope (Facility for Imaging by Light Microscopy, Imperial College London) equipped with a motorized stage and a 20 × objective (Zeiss Plan Apochromat, NA 0.8, WD 0.55 mm). The whole slice area was imaged using the multi-position acquisition function of Zeiss Zen software (LED excitation wavelengths 365 nm, 470 nm, 555 nm, and 625 nm). The acquisition focal plane corresponded to the image maximal sharpness (best focus) at five different areas of the brain slice. Images were analyzed with FIJI (ImageJ) software [
34,
35]. The four channels were separated and scaled. NeuN-, Iba1-, and GFAP-positive staining was quantified in the contralateral primary motor/association cortex (M1/MPtA), and bilaterally in the retrosplenial cortex (RSC), barrel field of somatosensory cortex (S1BF), amygdala, ventromedial hypothalamus and hippocampal CA1, CA2, CA3 and DG subregions by observers blinded to the experimental groups. Due to tissue damage within the contusion and disruption of cortical layers, it was not possible to quantify neurons in the ipsilateral motor/association cortex. For the neuronal counting in the cortical regions, we used rectangular regions of interest of width 200 µm spanning cortical layers 1 to 6, a circle of diameter 600 µm in the amygdala, an oval (520 × 670 µm) in the hypothalamus; in the hippocampus we used the following rectangular regions of interest: CA1 (300 × 30 µm), CA2 (200 × 70 µm), CA3 (250 × 50 µm), DG (two 200 × 50 µm on top; one 200 × 55 µm on bottom). Neurons were manually counted using FIJI (ImageJ), in two slices per brain and the mean density calculated. For the quantification of microglia and astrocytes we, used circular regions of interest in the left M1/MPtA (1300 µm diameter), S1BF (1300 µm diameter), amygdala (600 µm diameter), the contusional cortex (700 µm diameter); oval regions in the RSC (1300 × 600 µm) and ventromedial hypothalamus (561 × 636 µm). Due to tissue damage in the contusion it was not always possible to position the contusional cortex ROI in exactly the same anatomical area; the ROI was always within the contusion, in either: M1/MPtA (6/6 sham; 1/4 TBI control; 2/5 TBI xenon), edge of S1BF (1/5 TBI xenon) or edge of RSC (3/4 TBI control; 2/5 TBI xenon); in all cases there was no overlap with the ROIs for S1BF or RSC; in 2 of the TBI control group it was not possible to find a non-overlapping ROI within the contusion and these were excluded. In the hippocampus, outlines of total CA1, CA2, CA3 and DG regions were drawn for each slice using the ImageJ line tool; in the corpus callosum, the outline of the central area was drawn for each slice. To quantify the number of Iba1-positive cells; the background was subtracted (Gaussian blur function), the image was binarized and particles with an area of 40 µm
2 or larger were counted using automated or manual counting. We classified microglia based on their morphology; resting microglia have smaller rounder soma with high ramification, while activated microglia assume hypertrophic or bushy phenotypes with a larger more amorphous soma with less ramification [
36,
37]. We used a quantitative method using the roundness and size of the cell soma in order to classify all the microglia in each ROI as resting or active, with smaller round cells (area < 100 µm
2; roundness > 0.5) classified as low activity or resting, and larger irregular cells (area > 100 µm
2; roundness < 0.5) classified as active microglia [
38]. To quantify reactive astrogliosis, we measured the area of GFAP positive staining within the regions of interest; the GFAP images were binarized after thresholding and the percentage of GFAP stained area within the regions of interest was measured. In all of the immunohistological measurements, TBI control and TBI xenon groups were compared with the sham group that had been treated identically to the TBI groups but without impact, in order to ensure that any effects are independent of the drugs administered or surgery. One of the immunohistology slides from the xenon group was damaged and could not be imaged. Due to tissue damage or imperfections such as folds, it was not possible to make neuronal count measurements in every ROI (
eg right RSC, left & right hypothalamus) from every animal (individual points are shown on the graphs).