Inflammation and blood-brain barrier breach remote from the primary injury following neurotrauma

Adult female PVG hooded rats (160–180 g) were bred at the Animal Resources Centre (Murdoch, WA, Australia), housed under a standard 12-h light/dark cycle and fed standard rat chow and water ad libitum. Procedures adhered to the National Health and Medical Research Council Australian Code of Practice for the care and use of animals for scientific purposes and were approved by the University of Western Australia’s Animal Ethics Committee, approval numbers RA3/100/1201, RA3/100/673, and RA3/100/1485. The rats were anesthetized intraperitoneally with a combination of xylazine (Ilium Xylazil, Troy Laboratories, 10 mg/kg) and ketamine (Ketamil, Troy Laboratories, 50 mg/kg), and euthanized with an intraperitoneal injection of Euthal (pentobarbitone sodium 850 mg/kg; phenytoin sodium 125 mg/kg; Virbac).

The partial optic nerve transection procedure was similar to that described previously [25]. Briefly, anesthetized rats were shaved along the skull towards the right eye. Under aseptic conditions, the skin was incised and retracted, and the right optic nerve accessed by deflecting the Harderian lachrymal gland immediately behind the right eye. The nerve parenchyma was exposed about 1 mm behind the eye by making a slit in the dura mater with ophthalmic scissors. A controlled 200-μm cut was made in the dorsum of the right optic nerve using a diamond radial keratotomy knife (Geuder), with the depth determined by the protrusion of the blade beyond a surrounding guard. Care was taken not to stretch the optic nerve or damage major ophthalmic blood vessels. The deflected tissue was replaced, and the skin was sutured with 4/0 non-absorbable silk suture. The eyes were lubricated with Luxyal eye drops, and the rats were placed on a warming blanket for recovery with subcutaneous injections of analgesic (Carprieve, 5 mg kg⁻¹ in sterile water, Norbrook Australia, Pty. Ltd.; Victoria, Australia) and sterile PBS (1 ml). No postoperative infections were observed.

The right optic nerves from normal, age-matched, unoperated animals were used as controls, as sham-operated animals have been shown to be not different to normal in terms of a range of relevant cellular and oxidative stress outcomes [25]. Three cohorts of animals were used, with animals within each cohort being randomized into groups reflecting the number of days post-injury, for details see Table 1. In brief, cohort 1 was used for immunohistochemical studies assessing microglia/macrophages in the right optic nerves of animals at 3, 7, and 28 days post-injury and in uninjured controls. For cohort 1, three groups of uninjured controls collected at days 3, 7, and 28 were compared and found to be not different (ANOVA, p > 0.05); the 3-day controls were used for statistical comparisons. Cohort 2 was used for BBB analyses in the optic tracts and superior colliculus of animals at 3, 7, 14, and 28 days post-injury and in uninjured controls, assessing the presence of Evans blue dye. Once it was established that no Evans blue dye was visible in the optic tracts and superior colliculus, a subset of cohort 2 was randomly selected for immunohistochemical studies assessing microglia and macrophages as well as fibrinogen and caveolin-1 immunoreactivity with increases being surrogate markers for BBB breach in these brain regions (n = 5–6/group). Cohort 3 was used for cytokine analyses in the homogenized brains of animals at 1, 3, and 7 days post-injury and in uninjured controls.

Animals for BBB analyses (cohort 2) were administered a 2% (w/v) sterile solution of Evans blue (Sigma, E2129_10G) in PBS (2.8 ml/kg) via tail vein injection 1 h prior to euthanasia. Immediately following euthanasia, the animals for cohorts 1 and 2 (BBB and immunohistochemical assessments) were transcardially perfused-fixed with 0.9% (w/v) saline followed by 4% (w/v) paraformaldehyde (Sigma-Aldrich; St. Louis, Missouri, USA) in 0.1 M phosphate buffer, pH 7.2–7.4. The right optic nerves and brains were harvested and post-fixed in 4% paraformaldehyde for 1 h (nerves) or overnight (brains) and then cryoprotected in 15% sucrose in PBS and stored at 4 °C. The optic nerves were cryosectioned transversely (14 μm thickness), and the brains were cryosectioned coronally (20 μm thickness) and collected onto a series of slides so that each slide had a representative section from each area of interest. Sections for Evans blue analysis were collected onto Superfrost® Plus slides, air dried for an hour, and only briefly rinsed in PBS to reduce leakage of the dye from the sections before mounting using Fluoromount-G (Southern Biotech). The sections for immunohistochemistry were collected onto Superfrost® Plus slides and stored at − 80 °C.

Immunohistochemical analyses were conducted according to the previously described procedures [39]. Primary antibodies used for immunohistochemical assessments of the optic nerves detected inducible nitric oxide synthase (iNOS) (1:400, BD Biosciences, ab3523), Arginase-1 (Arg-1) (1:400, Santa Cruz Biotechnology, sc-18355), CD45 (1:200, BD Pharmingen, 555482), and CD11b (1:400, 553310, BD Pharmingen). Primary antibodies used for immunohistochemical assessments of the brains (optic tracts and superior colliculus) recognized CD45 (1:500, Abcam, ab10558), CD11b (1:300, Sapphire Bioscience, ab2910), ionized calcium-binding adapter molecule 1 (IBA1) (1:1000, Abcam, ab5076), CD68 (ED1) (1:1000, Millipore Corporation, MAB1435), Caveolin-1 (1:400, Sapphire Bioscience, Ab2910), and Fibrinogen (1:500, HiTech Group Australia, A008002). Secondary antibodies were species-specific AlexaFluor 488-, 555-, and 647-conjugated antibodies (1:500, Invitrogen Australia). All tissue sections for a particular immunohistochemical outcome were processed at the same time.

A single section was selected for optic nerves that showed the dorsal injury site and the ventral nerve. The fields of view imaged were located in the ventral portion of the right optic nerve of that section, directly below the dorsal primary injury site. The intensities of iNOS and Arg-1 immunoreactivity were assessed on the single central optical slice most in focus from the relevant 13 image z-stack. Mean fluorescence intensities were quantified on the whole image of ventral optic nerve and were normalized by subtracting the average of three randomly selected background regions of interest per animal. The numbers of CD45+ and CD11b+ cells were assessed within similar regions of interest. This non-stereological technique does not allow for an estimate of total cell numbers within the nerve but does allow for comparative assessment between time points after injury. Similarly for the brains, assessments of the numbers of CD45+, CD11b+, IBA1+, and ED1+ cells were conducted in single visual slices from images of single sections capturing either the lower, middle, or upper regions of both the ipsilateral and contralateral optic tracts and at the rostral end of the superior colliculus for each animal (Fig. 1). While sampling of more than one section may have been preferable, the brain regions of interest were often only clearly present in one section per slide. The optic tract is relatively small and was only apparent in one section per region of interest per slide along the rostrocaudal plane. Similarly, the optic nerve injury site, and therefore the ventral region directly under the injury, is only clearly apparent in one or two sections per slide. Not all markers were assessed for all regions at all time points; analyses were based on identifying the first time at which significant increases were observed and at times were constrained by the availability of sections containing the precise region of interest (e.g., superior colliculus at day 1). The numbers of positively stained cells within each field of view were expressed per unit area. Assessments of Evans blue fluorescence were conducted qualitatively, scanning multiple sections of the right optic nerves and regions throughout the brain including optic tracts, ventricles, choroid plexus, and the superior colliculus and applying a score to fluorescence; representative images are shown. Caveolin-1 immunoreactivity of endothelial cells lining the ventricles in the brain was quantified by measuring the length of immunoreactive cells expressed as a percentage of the total length of the lining of the ventricle that was visible in the single image used for analysis; the choice of section for imaging was based upon ventricle size and shape to ensure consistency of the region of analysis for all animals. Assessment of fibrinogen immunofluorescence was performed throughout the optic tracts, on single central visual slices from the relevant 13 image z-stack. Fibrinogen was quantified as mean fluorescent intensity and the area above the threshold as a percentage of the total area of interest. Threshold values were normalized to the background of each image by dividing the average of three random background measurements by the average background of a control image then multiplying the original threshold value by this transformation factor. Average intensities above the threshold were normalized by subtracting the background from the mean fluorescence intensity.

Rats from cohort 3 (n = 8 per group) were euthanized and transcardially perfused with ice-cold 0.9% saline, after which the brains were harvested, washed with ice-cold PBS, and homogenized in ice-cold RIPA buffer (Sigma Aldrich) supplemented with 1× protease inhibitor cocktail (Sigma Aldrich) (10 ml). Brain homogenates were incubated on ice for 20 min to aid protein solubilization, after which they were centrifuged at 12000g for 20 min at 4 °C, and the lysates were stored at − 80 °C until use. Total protein concentration of the brain lysates from each animal was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific Pty Ltd.) according to the manufacturer’s instructions.

Data are presented as mean ± standard error of mean. Statistical analyses of activated microglia and macrophage densities, Arg1 and iNOS immunointensities, and caveolin-1 and fibrinogen immunoreactivity were performed on SPSS (IBM) software, using one-way ANOVA or two-way ANOVAs to compare regions at a particular time point. For two-way ANOVAs, overall p, F, and df values are provided, with p, F, and df values provided for subsets where significance was reached. Tukey post hoc tests were used for optic nerve data and Sidak post hoc tests for optic tract data. If Levene’s test indicated inequality of variances, data were log or square root transformed for two-way ANOVAs or Games-Howell post hoc tests employed for one-way ANOVAs. Post hoc p values were provided where significance was reached using ANOVAs. Statistical analyses of cytokine data were performed on GraphPad Prism 6 software using one-way ANOVA to compare means of n = 8 animals at each time point, conducted individually for each cytokine, using Tukey post hoc tests.

Semi-quantitative assessment 3, 7, and 28 days after injury in the ventral portion of the right optic nerve directly below the dorsal primary injury site revealed significant increases in the immunointensity of both the pro-inflammatory biomarker inducible nitric oxide synthase (iNOS) (Fig. 2a, b) and anti-inflammatory biomarker arginase-1 (Arg-1) (Fig. 2c, d) at 7 days (p ≤ 0.05, F = 5.04, df = 38 and p ≤ 0.05, F = 7.41, df = 38, respectively). Immunointensity of iNOS returned to control levels 28 days after injury (p > 0.05), while arg-1 remained elevated (p ≤ 0.05). Contrary to the microglia/macrophage “polarization” phenotype previously described [40], we demonstrate colocalization of iNOS and Arg1 immunoreactivity within individual cells, evident 7 and 28 days after injury (Fig. 3).

The numbers of cells expressing CD11b (complement 3 receptor) a constitutive marker of leukocytes and CD45 which is expressed on all nucleated hematopoietic cells was assessed in the ventral portion of the right optic nerve. Quantification of CD45+ and CD11b+ microglia and macrophages demonstrated significant increases in the numbers of both CD45+ and CD11b+ cells 7 days after injury (p ≤ 0.05, F = 4.74, df = 23 and p ≤ 0.05, F = 9.15, df = 23, respectively) (Fig. 4a, c). The numbers of CD11b+ cells remained significantly higher than in uninjured controls at 28 days post-injury (p ≤ 0.05) (Fig. 4c, d), whereas CD45+ cells returned to levels not different from control at this time point (p > 0.05) (Fig. 4a, b).

Changes in densities of inflammatory cell populations were assessed in the lower, middle, and upper optic tracts (i.e., progressing from juxta-chiasmal = lower; mid-way between the chiasm and the superior colliculus = middle; pre-superior colliculus = upper) of the brain at 1, 3, 7, 14, and 28 days following partial injury to the optic nerve. Time points were chosen to complement the optic nerve analyses of the present study together with published work assessing microglia and macrophages in the partial optic nerve cut model [25, 39]. CD11b+ cells were assessed on both the left and right optic tracts, corresponding to the contralateral and ipsilateral sides to the injury, respectively. Two-way ANOVA revealed no significant changes in the densities of CD11b+ leukocytes in the lower tracts, compared to uninjured controls (p = 0.17, F = 1.63, df = 5) (Fig. 5a). There were changes in the number of CD11b+ cells/mm² in the middle (two-way ANOVA: p = 2 × 10⁻³, F = 4.53, df = 5; contralateral: p = 0.02, F = 3.09, df = 5; ipsilateral: p = 0.17, F = 1.69, df = 5) and upper tracts (two-way ANOVA: p = 1 × 10⁻⁴, F = 6.217, df = 5; contralateral: p = 9 × 10⁻³, F = 3.95, df = 5; ipsilateral: p = 0.051, F = 2.561, df = 5), with post hoc analysis revealing a significant increase relative to uninjured controls 7 days post-injury on the contralateral side (p = 0.03, middle tract; p = 0.01, upper tract), returning to control levels at day 14 (Fig. 5b, c).

Additional analyses were conducted focusing on the contralateral optic tract, to investigate first increases in further microglial/ macrophage subpopulations in a subset of sections of the optic tract at 1, 3, and 7 days following partial injury to the optic nerve. CD11b+ cells that also colocalized with IBA1 (a microglia/ macrophage marker which recognizes both resident and activated microglia/ macrophages) and ED1 (a marker of resident and infiltrating activated macrophages) were unchanged in the lower tract at the measured time points (p = 0.14, F = 2.29, df = 2) but significantly increased at 7 days compared to uninjured controls in the middle (p = 6 × 10⁻⁴, F = 13.54, df = 2) and upper tract (p = 0.03, F = 20.33, df = 2) (Fig. 6a, b), indicating that activation was enhanced in these regions. In contrast to outcomes within the optic nerve, there were no changes in the density of CD45⁺-positive cells in the contralateral optic tract. Quantification of CD45+ microglia and macrophages did not reveal any significant changes at 3 or 7 days following injury in any of the optic tract regions analyzed (p = 0.57, F = 0.59, df = 2 for lower; p = 0.41, F = 0.96, df = 2 for middle; and p = 0.60, F = 0.53, df = 2 for upper optic tract) (Fig. 7a, b).

Cells positive for the markers IBA1 and ED1 in isolation were also quantified in the contralateral optic tract at 3 and 7 days post-injury. Density of IBA1+ cells significantly increased from uninjured controls at 3 days post-injury, with a further increase at 7 days post-injury in the lower (p = 1 × 10⁻⁴, F = 20.08, df = 2; 3 days p = 0.01, 7 days p = 0.01) and middle (p = 7 × 10⁻⁶, F = 33.48, df = 2; 3 days p = 3 × 10⁻³, 7 days p = 4 × 10⁻³) regions of the optic tract. Increases in the upper tract only reached significance at 7 days (p = 3 × 10⁻⁵, F = 28.38, df = 2; 7 days p = 2 × 10⁻⁵) (Fig. 8a, b, e; high and low magnification representative images, respectively). The densities of ED1+ cells increased at 7 days following injury in the lower, middle, and upper tracts (p = 7 × 10⁻⁵, F = 23.33, df = 2; p = 0.00, F = 77.39, df = 2 ;and p = 0.00, F = 87.61, df = 2, respectively; 7 days p = 6 × 10⁻³, p = 1 × 10⁻⁴, p = 0.00, respectively) (Fig. 8c, d, e; high and low magnification representative images, respectively).

IBA1+ and ED1+ cells were also quantified in the superior colliculus at 1, 3, and 7 days post-injury, assessing both the contralateral and ipsilateral sides. IBA1+ cells did not significantly change relative to control uninjured nerve on either the contralateral or ipsilateral sides (p = 0.057, F = 4.21, df = 3; ipsilateral: p = 0.25, F = 1.46, df = 3) (Fig. 9a). However, post hoc tests revealed that ED1+ cells were significantly increased at day 7 on the contralateral side (p = 0.002) (contralateral: p = 2 × 10⁻⁴, F = 10.47, df = 3; ipsilateral: p = 0.50, F = 0.80, df = 3) (Fig. 9b).

Extravasation of Evans blue dye into the brains of rats following partial optic nerve injury was assessed qualitatively, being mindful of the caveats of this technique for assessing BBB integrity [41]. Despite our earlier observations of Evans blue dye in the brain at days 1 and 3 following this injury using multispectral imaging and quantitative assessment of extravasated Evans blue respectively [38], we did not observe Evans blue fluorescence at above background levels in cryosections of the brain viewed by confocal microscopy in the current study (Fig. 10a). The lack of Evans blue fluorescence in the brain sections in the current study may have been due to storage in sucrose and PBS prior to sectioning, and the single washing of the brain sections required to remove the OCT prior to section mounting. Previously, we observed Evans blue dye in the optic nerve at days 1 and 3 following injury [38]. In the current study, assessment of later time points indicated the presence of intense Evans blue staining at the optic nerve injury site and low-intensity diffuse staining ventral to the site of injury at 7 days post-injury (Fig. 10b). However, Evans blue fluorescence was not different to control at 28 days after partial transection (Fig. 10c, d), indicating closure of the BBB in the optic nerve at this time point after injury.

Caveolin-1 is a structural component of caveolae, mediates transcytosis through endothelial cells via interactions with claudin [42], and can be used as an indicator of BBB integrity [38]. Caveolin-1 immunoreactivity surrounding the third ventricle was assessed at 3, 7, 14, and 28 days following injury, to provide insight regarding the time of closure of the BBB following remote injury. The percentage of the third ventricle lined with caveolin-1+ cells was not different to control at 3, 7, and 28 days following partial optic nerve transection but was significantly reduced at 14 days (ANOVA: p = 5 × 10⁻⁴, F = 6.52, df = 4; 14 days p = 5 × 10⁻³) (Fig. 10e, f), indicating a transient compromise of caveolin-1-mediated junctions at this later time point.

Given the potential for washout of Evans blue dye from the tissue sections, immunoreactivity of fibrinogen was used as an additional measure of BBB integrity. Fibrinogen is a blood-derived protein whose accumulation in the brain indicates loss of BBB integrity [43]. The area of fibrinogen immunoreactivity, indicative of the size of the breach of the BBB, did not change with injury in the optic chiasm (two-way ANOVA: p = 0.24, F = 1.41, df = 4) (Fig. 11a). However, the intensity of staining in the chiasm, perhaps indicating a focal breach, did change with injury (contralateral: p = 0.03, F = 3.10, df = 4; ipsilateral: p = 2 × 10⁻³, F = 5.18, df = 4), with post hoc analysis revealing that increases in signal intensity relative to control were confined to the ipsilateral side at 7 days post-injury (p = 0.03) (Fig. 11b). Changes in the area of fibrinogen immunoreactivity in the lower tract were seen on both sides (contralateral: p = 0.02, F = 3.33, df = 4; ipsilateral: p = 0.03, F = 2.98, df = 4); however, post hoc tests only revealed differences between 7 and 28 days, rather than relative to control (contralateral: p = 0.03; ipsilateral: p = 0.04) (Fig. 11a). No differences in the intensity of fibrinogen immunoreactivity were observed in the lower tract (contralateral: p = 0.10, F = 2.09, df = 4; ipsilateral: p = 0.27, F = 1.35, df = 4). In the middle tract, changes in the area of fibrinogen immunoreactivity were confined to the contralateral side (contralateral: p = 3 × 10⁻⁴, F = 7.12, df = 4; ipsilateral: p = 0.06, F = 2.48, df = 4), with post hoc analysis revealing differences relative to control 7 days post-injury (p = 4 × 10⁻³). There were no significant differences in the intensity of fibrinogen immunoreactivity in the middle tract (contralateral: p = 0.24, F = 1.45, df = 4; ipsilateral: p = 0.06, F = 2.55, df = 4). Changes in the upper tract were similar to the middle tract but occurred earlier (contralateral: p = 0.001, F = 5.98, df = 4; ipsilateral: p = 0.69, F = 0.56, df = 4) with post hoc tests showing that area of the breach increased at day 3 on the contralateral side (p = 0.004) (Fig. 11a, c). No changes were observed in the intensity of fibrinogen immunoreactivity in the upper tract (contralateral: p = 0.086, F = 2.26, df = 4; ipsilateral: p = 0.10, F = 2.10, df = 4) (Fig. 11b). Finally, in the superior colliculus, no significant changes in fibrinogen immunoreactivity were observed for area (contralateral: p = 0.058, F = 2.95, df = 3; ipsilateral: p = 0.06, F = 2.82, df = 3) or intensity of immunoreactivity (contralateral: p = 0.32, F = 1.23, df = 3; ipsilateral: p = 0.16, F = 1.88, df = 3) (Fig. 11a, b).

The inflammatory response in the whole brain homogenates following partial injury to the optic nerve was determined by measuring changes in cytokine protein levels in homogenized rat brains at 1, 3, and 7 days post-injury compared to uninjured controls (Fig. 12). While it would perhaps have been preferable to assess the cytokines in the individual optic tract regions of interest, the volume of tissue required for these assays precluded these assessments. In these experiments, ten cytokines were measured: IL-6, IL-1β, TNFα, IL-2, IL-4, IL-10, IL-12, GMCSF, IL-1α, INFγ; however, differences between injured and control rat brains were only detected for three of them, TNFα, IL-2, and IL-10. The pro-inflammatory cytokine, IL-1β, and the anti-inflammatory cytokine, IL-4, remained unaltered in injured rat brains compared to controls, and five cytokines were below the detection limit of the multiplex assay, IL-6, IL-12, GMCSF, IL-1α, and INFγ. Tumor necrosis factor (TNFα) is a multifunctional cytokine most often referred to as a potent pro-inflammatory cytokine, produced by microglia and astrocytes. There was a significant increase in TNFα protein levels in rat brains in the early time point, 1 day post-injury (1.6-fold increase ± 0.14) compared to uninjured controls (p < 0.0001, F = 4.13, df = 3) (Fig. 12). At 3 days post-injury, the concentration of TNFα returned to levels that were not significantly different from uninjured controls (p > 0.05), and these levels were maintained at 7 days post-injury (Fig. 12). The pro-inflammatory cytokine IL-2 was significantly elevated in the rat brains at an early time point following injury, 1 day (1.4-fold increase, p < 0.05, F = 2.12, df = 3), remained elevated at 3 days post-injury (1.4-fold increase, p < 0.05, F = 2.12, df = 3), and decreased to levels not significantly different from controls at 7 days post-injury (Fig. 12). IL-10 is regarded primarily as an anti-inflammatory cytokine, having a potent inhibitory effect on the production of several pro-inflammatory mediators including TNFα and IL-1β. IL-10 levels were biphasic, being significantly elevated from control levels at both an early time point, 1 day (1.6-fold increase, p < 0.001, F = 13.28, df = 3), and a later time point, 7 days (1.4-fold increase, p < 0.05, F = 13.28, df = 3) post-injury, with a decline to control levels in between at 3 days post-injury (Fig. 12).