Juvenile Traumatic Brain Injury Model

Controlled cortical impact was induced in rats as previously described.¹⁴ Briefly, rats were anesthetized with isoflurane (Webster Veterinary Supply, Sterling, MA, USA) and given a 5-mm diameter craniotomy over the right fronto-parietal cortex (1 mm posterior, 2 mm lateral from Bregma) and CCI was delivered to jTBI animals using a 3-mm impactor at a 20° angle to cortex, 1.5 mm depth, 200 milliseconds impact duration, 6 m/s velocity. Body temperature was maintained at 37°C during surgery. A subcutaneous buprenorphine injection was administered for pain relief (0.01 mg per mL per kg at 1 hour and 24 hours after surgery) and animals were returned to their dams. All sham animals underwent the same procedures as jTBI animals, except for the CCI.

Water Maze Design and Testing

Water maze testing has been previously described for this model.¹⁴ Briefly, testing occurred at 30 and 60 dpi over a 3-day paradigm, with 5 blocks of 2 trials each (10 total trials) including cued testing (visible platform), spatial learning (hidden platform), and probe trials for spatial memory (no platform). Swim strategy analysis¹⁶ revealed eight different swim patterns grouped into three general learning strategies (Table 1; Figure 5A): Spatial strategy (spatial direct, spatial indirect, focal correct), Systematic strategy (scanning, random, focal incorrect), or Looping strategy (chaining, peripheral looping).

Brain Tissue Processing and Immunohistochemistry

At 60 dpi, rats were anesthetized with a combination of Ketamine (Ketaject 100 mg/mL, Phoenix, St Joseph, MO, USA) and Xylazine (AnaSed 100 mg/mL, Lloyd Laboratories, Shenandoah, IA, USA) at the appropriate dose/body weight, and then transcardially perfused with 4% paraformaldehyde. The brains were excised and cryoprotected in 30% sucrose solution for 48 hours, and frozen on dry ice. Coronal cryostat free floating sections (50 μm) were cut and collected as serial sections spaced 1.2 mm apart, then processed for standard immunohistochemistry experiments as previously described.¹⁴

Immunoglobulin G Extravasation Staining for Blood–Brain Barrier

Sections were rinsed in phosphate-buffered saline (PBS) and blocked for 1 hour in 1% bovine serum albumin (BSA) (Sigma-Aldrich, St Louis, MO, USA) made in PBS, pH=7.4 (Fisher Scientific, Pittsburg, PA, USA) before incubation for 2 hours at room temperature with Alexa-Fluor-800 biotin-conjugated affinity purified goat anti-rat immunoglobulin G (IgG) (1:500, Rockland Immunochemicals, Gilbertsville, PA, USA) in PBS containing 0.25% Triton X-100 and 0.25% BSA (Sigma-Aldrich) made in PBS, pH=7.4. After washing, sections were scanned on an Odyssey infrared scanner to quantify fluorescence in the cortex and striatum.

Immunolabeling of Blood–Brain Barrier Proteins

For P-gp staining, sections were pretreated for antigen retrieval using 33% acetic acid+66% EtOH solution for 10 minutes at 20°C (Fisher Scientific). For claudin 5, glial fibrillary acidic protein (GFAP), P-gp, LRP1, and RAGE, sections were blocked for 1 hour in 1% BSA in PBS before overnight primary antibody incubation at 4°C. All antibody incubations were in 0.25% BSA with 0.25% Triton X-100 made in PBS, pH=7.4. For immunolabeling, we used mouse anti-claudin 5 (1:500, Life Technologies: Invitrogen, Grand Island, NY, USA), mouse anti-P-gp (1:100, Abcam, Cambridge, MA, USA and Calbiochem, EMD Chemicals, Merck KGaA, Darmstadt, Germany), mouse anti-LRP1 (1:1,000, Calbiochem, EMD Chemicals, Merck KGaA), rabbit anti-RAGE (1:500, Abcam, Cambridge, MA, USA) and chicken anti-GFAP (1:1,000, Millipore, Temecula, CA, USA). After PBS rinses, we incubated in secondary antibody for 2 hours at room temperature at 1:1,000 as appropriate for each primary antibody: goat anti-mouse secondary antibody coupled with Alexa-Fluor-488 or with Alexa-Fluor-594, goat anti-chicken coupled with Alexa-Fluor-568, and for infrared analysis either goat anti-mouse secondary antibody coupled with Alexa-Fluor-800 or goat anti-rabbit secondary antibody coupled with Alexa-Fluor-680 (all secondaries from Invitrogen, Grand Island, NY, USA). After washes in PBS, sections for classical immunofluorescence were mounted on glass slides and coverslipped with vectashield antifading medium containing DAPI (Vector, Vector Laboratories, Burlingame, CA, USA). Sections for infrared analysis were mounted on glass slides and air-dried. Control sections for all studies in which primary or secondary antibodies were omitted resulted in negative staining (not shown).

Immunolabeling for Aβ

Sections for amyloid analysis were pretreated for 4 minutes in 88% formic acid at room temperature and all other immunostaining procedures were identical to the procedure described above. We used a monoclonal antibody raised in mouse, against rodent Aβ at the N-terminal amino acids 1 to 16 (1:1,000, from M Paul Murphy) and visualized staining with goat anti-mouse secondary antibody Alexa-Fluor-488 (1:1,000; Life Technologies: Invitrogen, Grand Island, NY, USA). Adjacent sections were coincubated with the mouse rodent-Aβ antibody and either the C-terminal antibody against Aβ1-42 (rabbit, 1:1,000, Covance, Emeryville, CA, USA), or the microglial marker ionized calcium binding adaptor molecule 1 (IBA1) (rabbit, 1:1,000, Wako Chemicals, Richmond, VA, USA). To control for nonspecific binding of mouse antibodies on rat tissue (Figure 3F), sections received only the secondary antibody, goat anti-mouse Alexa-Fluor-594 (staining was negative), followed by a full protocol with both the rodent-Aβ antibody and goat anti-mouse Alexa-Fluor-488. Sections were also stained with the 6E10 anti-human-Aβ1-16 antibody (mouse, 1:5,000, Covance) or using protocols excluding primary or secondary antibodies, resulting in negative staining (not shown).

To determine any fibrillar β-pleated sheet structures, sections were double stained with a rodent-Aβ (see above) followed by Thioflavin S staining. Briefly, tissue was immersed for 5 minutes in double-distilled (dd) H20, followed by a 5-minute incubation in 1% Thioflavin S (Sigma-Aldrich) made in ddH20, and subsequent differentiation using consecutive 5 minutes washes in 70% EtOH, 95% EtOH, and ddH20 (Fisher Scientific). Slides were then coverslipped as described above.

Quantification of Immunohistochemistry

Coronal sections immunostained with infrared secondary antibodies (IgG and RAGE) were scanned using the same parameters for sham and jTBI at a 21-μm resolution (Li-Cor Odyssey Bio-Science, Lincoln, NE, USA) and identical regions-of-interest (ROIs) were drawn using Li-Cor Odyssey software.¹⁴ For IgG, average values per animal were obtained from ROIs in the perilesional parietal cortex and striatum in the hemisphere ipsilateral to the lesion, and analogous areas were evaluated in the hemisphere contralateral to the lesion or craniotomy, for a total of three to five serial coronal sections as available (n=4 sham, n=4 jTBI). For RAGE, average values per animal were obtained from ROIs in the parietal and temporal cortex in hemispheres both ipsilateral and contralateral to the lesion or craniotomy from four coronal sections at bregma levels −1.6, −2.8, −4.0, and −5.2 mm (n=3 sham, n=3 jTBI).

For quantification of classical immunolabeling of claudin 5, P-gp, LRP1, and Aβ, images were evaluated and collected using an epifluorescent microscope (Olympus, BX41, Center Valley, PA, USA). The threshold and morphological user-defined parameters were selected to maximize visualization of positive staining in the ROIs for each protein staining pattern. These parameters were kept consistent for all animals during image acquisition. For claudin 5, P-gp, and LRP1 a series of images were taken in the parietal and temporal cortex, both above and below the rhinal fissure on each hemisphere ipsilateral and contralateral to the lesion or craniotomy using a × 20 objective (422 μm × 338 μm). For claudin 5, images of large intraparenchymal vessels (30 to 100 μm diameter) and a separate analysis on microvessels (10 to 20 μm diameter) were acquired from two serial coronal sections at bregma level −2.8 and −4.0 mm (n=4 sham, n=4 jTBI; 8 images/animal for large vessels and 8 images/animal for small vessels). For P-gp, images were acquired from four serial coronal slices at bregma level −2.8, −4.0, −5.2, and −6.4 mm (n=6 sham, n=6 jTBI; 24 images per animal). For LRP1, images were acquired from two serial coronal slices at bregma level −4.0 and −5.2 mm (n=3 sham, n=3 jTBI; 8 images per animal). Individual images were analyzed with MorphoPro software (Explora-Nova, La Rochelle, France), using the following procedure: (1) Top Hat morphologic filter to outline vascular staining, (2) user-defined thresholding value applied to each image, and (3) calculation of area of staining from background for each protein of interest (claudin 5, P-gp, or LRP1). Final values are represented as a percentage of the sham group.

For quantification of rodent-Aβ immunoreactivity, we used eight whole serial coronal slices per animal spaced 1.2 mm apart, from bregma levels +3.2 to −5.2 mm as shown in Figure 4 (n=8 sham, n=6 jTBI; 112 total individual values). Mercator software (Explora-Nova) was used to automatically calculate the area (μm²) of positive staining that was normalized to the total area (μm²) of each respective coronal slice and presented as % of Aβ load.

Brain Tissue Processing for Western Blotting

A Protein FFPE extraction kit (Qiagen, Hilden, Germany) was used to process perfused brain slices for Western blotting. Parietal and temporal cortical tissue above the rhinal fissure was excised from three coronal sections at bregma levels −1.6, 2.8, and −4.0 mm, that were adjacent to slices of interest used for immunohistochemistry. Briefly, tissue was homogenized and processed according to the kit instructions, and then samples were assayed for total protein concentration by bicinchoninic assay (Pierce Biotechnology, Rockford, IL, USA). The human Aβ1-40 and Aβ1-42 peptides (Biopeptide, San Diego, CA, USA) were prepared using a 1-mg sample that was reconstituted to obtain 231 μmol/L Aβ1-40 and Aβ1-42 peptides in NaOH (Sigma-Aldrich). Peptides were then incubated overnight at 37°C and aliquoted for storage at −80°C, then thawed on ice when ready for use. For gel electrophoresis, all samples were prepared with loading sample buffer and reducing agent (Invitrogen, Carlsbad, CA, USA) for a total of 40 μg of rat protein, 4 μg of Aβ1-40 peptide, and 0.5 μg of Aβ1-42 peptide, loaded on a 4% to 12% sodium dodecyl sulfate polyacrylamide gel (Nupage, Invitrogen, Carlsbad, CA, USA). Proteins were transferred to a polyvinylidene fluoride membrane (Perkin-Elmer, Rodgau,Germany), blocked for 1 hour in Odyssey blocking buffer (Li-Cor Bio-Science), and incubated with the same monoclonal antibody against rodent-Aβ used in immunohistochemistry (1:2,000; from Dr M Paul Murphy, University of Kentucky) or a monoclonal antibody anti-human-Aβ 6E10 (1:750; Covance). P-glycoprotein and claudin 5 antibodies were used as in the immunostaining, namely mouse anti-P-gp (1:100, EMD Chemicals, Merck KGaA) and mouse anti-claudin 5 (1:200, Life Technologies: Invitrogen, Grand Island, NY, USA), and these blots were incubated with Biotin-SP-conjugated goat anti-mouse (1:5,000; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 1.5 hours at room temperature, followed by PBS washes. The blots were then incubated for 1 hour at room temperature with Streptavidin-conjugated-IRDye 800 (1:10,000; Li-Cor Bio-Science), before scanning. Each Western blot was coincubated with a rabbit polyclonal antibody against β-actin (1:1,250; Sigma-Aldrich) in Odyssey blocking buffer (Li-Cor Bio-Science) overnight at 4°C. After PBS washes, proteins were visualized using a coincubation with goat anti-mouse secondary antibody coupled with Alexa-Fluor-800 (1:10,000; Rockland Immunochemicals) and goat anti-rabbit secondary antibody coupled with Alexa-Fluor-680 (1:10,000; Life Technologies: Molecular Probes, Grand Island, NY, USA) for 2 hours at room temperature. After PBS washes, the polyvinylidene fluoride membrane was visualized using the Li-Cor infrared scanner.

Widespread Aβ Accumulation

Using a rodent-Aβ antibody specific to the N-terminus and a formic acid pretreatment, we detected a diffuse pattern of immunoreactivity that is both extracellular (Figures 3A and 3C–3H) and perivascular (arrowheads in Figures 3C and 3E–3H). Examples of positive rodent-Aβ immunolabeling are shown at a distance from the injury, such as the frontal (Figures 3C and 3H), temporal (Figure 3A), and parietal cortex (Figure 3D). Positive staining was also detected in the entorhinal cortex, striatum, and thalamus (Figure 4A), with no staining using the secondary antibody alone (Figure 3A).

Open in a separate window

Figure 4

Anterior to posterior patterns of rodent-Aβ distribution. (A) Representative coronal sections with outlines of rodent-Aβ positive staining in juvenile traumatic brain injury (jTBI) are shown at bregma +2.0 mm, −0.4, −1.4, and −5.2 mm. The jTBI lesion cavity is apparent in the parietal cortex at −1.4 mm and still visible at bregma −0.4 and −3.8 mm to a lesser extent (black arrowheads). (B) All positive rodent-Aβ immunoreactivity was quantified to obtain %Aβ load relative to total brain area at 8 bregma levels, with significant changes along the anterior-to-posterior axis (P<0.042), but no significant changes between groups at any single level. (C) Summation of total brain Aβ load was higher in jTBI versus sham (P<0.073), with jTBI animals showing high staining variability common in models of natural Aβ accumulation (values are represented as mean±s.e.m.; Aβ, β-amyloid).

Western blot showed that the rodent-Aβ antibody was specific for rat tissue but not for human Aβ1-40 and Aβ1-42 peptides, while the anti-human 6E10 antibody detected bands for human Aβ peptides, but no signal for rat tissue (Figure 3B). The 6E10 antibody detected monomeric Aβ species (4 kDa) in both human Aβ1-40 and Aβ1-42 peptide preparations, and several Aβ aggregates that are more abundant in the Aβ1-42 preparation (Figure 3B). In the rat tissue, the rodent-Aβ antibody reveals an APP (β-amyloid precursor protein) signal near the expected region ∼110 kDa and several additional bands, possibly APP fragments or Aβ aggregates of various sizes (Figure 3B).

Rodent-Aβ immunoreactivity was associated with microglial cells, stained with IBA1. Microglia exhibited an elongated soma, little somatic cytoplasm, and multiple thinner processes, as shown in an example from the jTBI parietal cortex (Figure 3D). No activated microglia were detected at this time point (data not shown). Rodent-Aβ immunoreactivity colocalized with C-terminal-Aβ1-42 staining in several brain regions, such as the frontal cortex (Figures 3E–3H). Thioflavin S staining (data not shown) is negative despite positive rodent-Aβ immunoreactivity, suggesting a lack of fibrillar β-pleated sheet structures. Overall, these data are consistent with an early Aβ accumulation pattern having a diffuse morphology and containing Aβ1-42 species.

Total rodent-Aβ immunoreactivity was quantified using the Mercator program and expressed as % Aβ load relative to total coronal brain area at each bregma level (Figures 4A–4C). Representative jTBI animals are shown with the lesion location (black arrowheads in Figure 4A) and outlines of positive staining at individual bregma levels (Figure 4A). By 60 dpi, the cortical jTBI lesion cavity is apparent from bregma −1.0 mm up to −4.0 mm (not shown) and higher rodent-Aβ immunoreactivity was observed in more anterior and posterior levels from the lesion site (Figure 3B). Repeated measures ANOVA showed significant changes across all coronal sections regardless of group (Figure 4B, P<0.042), with moderate changes across bregma levels in jTBI group alone (P<0.057), but no changes observed in sham alone (P>0.1). Similarly, the jTBI group showed a trend for higher staining in total brain compared with sham (Figure 4C, P<0.073). These data suggest that early brain injury may exacerbate a process of endogenous rodent-Aβ accumulation throughout vulnerable regions over time.

Physiopathological Changes in the Blood–Brain Barrier

We explored BBB phenotypes at a delayed time point after juvenile injury as a unique platform for addressing whether jTBI disrupts the neurovascular unit and contributes to brain pathophysiology. Within the first days after injury, we observe high IgG staining.¹ Blood–brain barrier disruption is repaired by 2 months in our jTBI model, as evidenced by the absence of IgG extravasation and increases in tight-junction protein claudin 5 (Figure 1). Other studies also report that improved BBB integrity coincides with altered expression of tight-junction proteins at later time points. Similar to our model, claudin 5 levels are upregulated after evans blue and IgG leakage subsides at 1 and 2 weeks after rat cortical injury.¹⁷ Claudin 5 is also increased, along with resolution of evans blue staining, after 4 and 8 weeks in mice exposed to an infectious agent targeting the BBB.¹⁸ Thus, altered levels of claudin 5 long after jTBI suggest delayed modifications in BBB structural properties. Furthermore, our differential observations on large and small vessels (Figure 1; Supplementary Figure 1) indicated a more complex phenotypic response of the endothelium, which may be related to location along the vascular tree.

We investigated the triad of endothelial cell proteins (P-gp, LRP1, and RAGE) previously shown as potential mediators of Aβ trafficking and other proteins across the BBB (Figure 6).^{5, 19} P-glycoprotein is well known as an efficient gatekeeper on the luminal side and often pharmaceutically by-passed to allow efficient drug delivery. Its expression is influenced by several distinct molecular pathways and its putative role in disease and injury is emerging.^{20, 21} We observed a significant decrease in P-gp in cortical vessels following jTBI by immunostaining and protein blot analyses. Similar to our findings, brain P-gp protein level and function decreased after irradiation of normal rats,²² in aged 3-year-old rats,⁶ and in sporadic AD.²¹ Additionally, P-gp expression and function were impaired during multiple sclerosis pathology and its dysfunction coincides with lymphocyte infiltration and inflammation in experimental allergic encephalomyelitis.²⁰

Our observed P-gp decreases may be one result of long-lasting neuroinflammation, contributing to an increase in rodent-Aβ immunoreactivity. Importantly, P-gp knockout mice exhibited decreased LRP1 in brain capillaries without changes in RAGE proteins⁷ and aged 3-year-old rats showed decreased vascular levels of P-gp and LRP1 associated with accumulation of Aβ.⁶ We did not detect significant changes in LRP1 between sham and jTBI, but LRP1 was positive in blood vessels and neurons within the temporal cortex of jTBI animals. Our findings are in line with several model systems and age groups, which emphasize the putative role of the BBB in chronic brain pathophysiology after acute injury. Our observed P-gp decrease may have a complex relationship to accumulation of rodent-Aβ in multiple brain areas. Frontal and temporal lobes are vulnerable to neuropathology after TBI.¹⁰ In our model, these same regions exhibited reduced P-gp (Figure 1) and increased rodent-Aβ immunoreactivity in the brain parenchyma and vessel walls (Figures 3 and ​and4).4). Clinically, our findings are similar to low endothelial P-gp that correlated with increased parenchymal Aβ1-40 or Aβ1-42 plaques in the medial temporal lobe of aged nondemented humans.²³ Following peripheral Aβ1-42 injections in adult mice, brain endothelial cells show downregulated expression of RAGE, LRP1, and P-gp at the mRNA level; however, P-gp immunostaining remained unchanged.²⁴ Collectively, these findings highlight complex interactions between Aβ species and the BBB transporters that may be influenced by injury type and study time point.

Aβ Accumulation and Blood–Brain Barrier Changes After Juvenile Traumatic Brain Injury

The important consequences of cerebrovascular dysfunction and the role of the BBB neuropathological protein trafficking are gaining momentum. In normal rodents, classical AD-like pathology with fibrillar amyloidosis is rare. This may be due to Aβ sequence differences of three amino acids at the N-terminal.²⁵ However, we used an N-terminal specific antibody for rodent Aβ,²⁶ and endogenous accumulation has been reported long after a brain insult in normal experimental injury models. For example, increased APP and Aβ immunostaining was observed in thalamic nuclei for up to 9 months after stroke in normal rats.²⁷ In normal adult rats, intranasal nerve growth factor administration resolved increases of Aβ1-42 for up to 2 weeks post-TBI.²⁸ In a rodent model of aging, senescence accelerated mice (SAMP8) showed increased BBB disruption at 12 months and exhibited vascular Aβ compared with their littermate controls.²⁹ These data indicate the close relationship of BBB disruption and brain pathology in aging and postinjury.

Blood–brain barrier phenotypic changes at 2 months occurred in parallel with higher rodent-Aβ immunolabeling in the parenchyma and vasculature of jTBI animals (Figures 3 and ​and4).4). Notably, we observed diffuse Aβ deposition in regions remote to the original site of impact, especially in the superficial cortical layers. Additionally, diffuse Aβ in superficial cortical layers is the primary conformation present in other models of natural Aβ accumulation, such as aged beagles.³⁰ Diffuse Aβ is also more common in clinical cases with mild cognitive impairment³¹ and is becoming recognized as an important contributor to disease progression by the National Institute of Aging-Alzheimer's Disease.³² Aβ load, evaluated by immunohistochemical staining similar to our methodology, was the best predictor for dementia in very old individuals.³³ Interestingly, we observed sparse rodent-Aβ immunoreactivity in some shams (n=3 of 8 total) and only modest effects of increased rodent-Aβ immunolabeling throughout the brain of jTBI animals (n=6 total). In a similar TBI model, sham animals with a craniotomy exhibited brain tissue modifications and deficits in learning and memory postsurgery.³⁴ Therefore, we expect that inclusion of additional animals or a more appropriate naive control group may improve our data variability and improve statistical power.

Parallel Changes in Vascular Phenotype and Behavioral Dysfunction

In our injury model, we observed changes in behavioral outcome by 2 months after injury,¹⁴ which indicate that the brain may be undergoing compensatory mechanisms. Here, we observed strategic water maze impairment without overt memory deficits, which may reflect underlying brain-repair mechanisms occurring after injury. Interestingly, clinical defects of initial strategy formulation affected performance on a task of complex visuospatial executive function in individuals with amnestic mild cognitive impairment who are more likely to develop AD.³⁵ In a CCI study in adult mice, sham and TBI animals exhibited similar water maze outcomes after 2 months, but TBI animals exhibited thigmotaxic strategy impairment with increased percent time swimming along maze walls, similar to our looping category.³⁶ In comparison with adult models, juvenile rats may be especially vulnerable due to a lack of cognitive reserve. While the neural correlates are unclear, clinically it is thought that having increased cognitive reserve from education and other life experiences may be protective against brain aging and disease.³⁷ Thus, brain injury at a developmental stage may influence underlying mechanisms and reduce cognitive reserve. Our observations of delayed impairments in jTBI suggest a parallel with acceleration of brain aging and disease due to trauma.

The authors thank Dr André Obenaus (Loma Linda University) for project assistance and Monica Rubalcava for a portion of imaging performed at the Loma Linda University School of Medicine Advanced Imaging and Microscopy Core (LLUSM AIM) supported by NSF Grant, MRI-DBI 0923559 (to SM Wilson).