Pericontusion Axon Sprouting Is Spatially and Temporally Consistent With a Growth-Permissive Environment after Traumatic Brain Injury

Experimental Groups

Groups of adult male, Sprague Dawley rats (250–275 g) were randomly assigned as follows: (i) immunohistochemistry (IHC) analysis at post-injury day 7, 14 and 28 (n = 6, 6, and 3, respectively, for each time-point for both controlled cortical impact injury [CCI] or sham injury groups); (ii) growth-associated protein 43 (GAP43) in situ hybridization at post-injury day 7 (n = 5 and 3 per CCI and sham injury group, respectively); and (iii) for BDNF Western blot analysis at post-injury days 7 (n = 3, 5, 5 per naïve, CCI and sham injury group, respectively). All procedures were approved by the UCLA Chancellor’s Committee for Animal Research.

Surgery and Controlled Cortical Impact Injury

A unilateral CCI to the left hemisphere model was used as previously characterized (27, 32, 33). Animals were anesthetized with isoflurane (3% in 100% O₂ [1.5–2.0 l/minute flow rate] for induction; 1.5–2% for maintenance) and received an application of ophthalmic ointment to both eyes when they were under a surgical level of anesthesia. During all surgical procedures, body temperature was maintained at 37 ± 0.5°C using a thermostatically controlled heating pad (Harvard Apparatus, Edenbridge, KY). After shaving and sterilizing the scalp, the anesthetized animal was placed into a stereotaxic frame, a midline incision was made over the skull, and the skin, fascia and temporal muscles were reflected bilaterally. Using a microscope and intermittent saline-cooled perfusion of the skull area, a 6-mm-diameter craniotomy was made over the left sensorimotor cortex, centered at Bregma and 3.5 mm lateral to midline. An electronically controlled pneumatic piston (Hydraulics Control, Inc, Emeryville, CA) mounted on a stereotaxic micro-manipulator (Kopf Instruments, Tujunga, CA) and angled at 22.5° to the vertical to allow the rounded 4-mm-diameter tip to make contact perpendicular to the surface of the brain, was used to create the CCI. The injury (20 psi; 2 mm dural compression) was induced by depression and rapid contraction of the pneumatic piston. Hemostat material (Surgicel^™, Johnson & Johnson, Langhorne, PA) was placed over the injury site before the bone flap was replaced and sealed in place with cyanoacrylate glue, used sparingly to prevent excess from running onto the hemostat. Sham injury control rats were subjected to all surgical procedures, but did not experience induction of the CCI. Scalp incisions were sutured closed and Bupivacaine (0.05–0.07 mg/kg, s.c.) was infiltrated into this wound margin. After removal from anesthesia rats were maintained in a warm recovery cage for ~30 minutes before being returned to home cages. Additional control rats that did not experience any anesthesia or surgical procedures were used in the Western blot experiment.

Immunohistochemistry

Following overdose with sodium pentobarbital (100 mg/kg, i.p.) rats with CCI or sham injury were transcardially-perfused with 0.1 M phosphate-buffered saline (PBS) at approximately mean arterial blood pressure followed by 4% paraformaldehyde in PBS. Following cryoprotection in 25% sucrose in PBS until the brain sank, the brain was frozen and 50-μm-thick coronal sections were cut. Standard free-floating, single label IHC was performed on one series of sections. In brief, antigen retrieval was carried out using citrate buffer (1 M, pH = 6.0, incubation overnight at 4°C) followed by two rounds of 10 minutes heating to 95°C and 10 minutes cooling. Following TRIS-buffered saline washes (TBS, 0.2M) single label IHC was performed for stereologic analysis of cell density. All sections were quenched in a solution of 10% methanol and 10% H₂O₂ in distilled water for 60 minutes before washing 3X in TBS. Sections were then blocked for 60 minutes in TXTBS (TBS containing 0.2% Triton X-100; Sigma, St. Louis, MO) with 10% normal horse serum (NHS; Vector Labs, Burlingame, CA), 1% bovine serum albumin (BSA, Vector Labs). Adjacent sections were then incubated over-night at room temperature (RT) on a shaker in either GAP43 primary antibody (1:3000; #MAB347, Chemicon/Millipore, Billerica, MA) in TBS containing 5% NHS, or the biotin-conjugated lectin Wisteria floribunda (WFA; 1:1000; #L-1766, Sigma) in TBS containing 2% BSA. After washing three times in TBS, sections were incubated with biotinylated, rat-adsorbed anti-mouse IgG (1:500; Vector) in TBS with 5% NHS for 1 hour followed by three washes in TBS. Signal amplification was achieved using the Vectastain elite ABC kit (Vector) for 30 minutes following the manufacturer’s instructions, followed by 3 washes in TBS. Antibody binding was then visualized with diaminobenzidine in distilled water containing 0.03% H₂O₂ and excess stain was removed by washing in TBS 3X. Sections were then mounted on gel-coated slides, dehydrated, and cover-slipped with DPX mounting medium.

For double label IHC, sections were processed for antigen retrieval as above and then blocked for 15 minutes each in streptavidin and then biotin solution (Vector) and in 10% NHS in PBS for 1 hour, followed by overnight incubation in either anti-GAP43 (1:1500) or WFA, as described above. Following rinsing in PBS, GAP43 sections were incubated in rat-adsorbed, biotinylated secondary antibody for 2 hours (1:500 dilution in 5% NHS, Vector), followed by 3 washes in PBS. Antigens were visualized with streptavidin-conjugated fluorescent Alexa-488 and -555 probes (Molecular Probes/Invitrogen, Carlsbad, CA, 1.25 μg/ml in PBS for 60 minutes). Double labeling was then achieved using another biotin blocking step followed by mouse-on-mouse blocking reagent when required (MKB-2213, Vector) and a serum blocking step as before, all of which facilitated the use of additional mouse-host primary antibodies on the same preparation for GAP43 IHC. Sections were incubated overnight at RT on a shaker in either mouse monoclonal antibodies: anti-NeuN, (1:5,000; #MAB377, Chemicon/Millipore), or poly-sialic acid neural cell adhesion molecule (PSA-NCAM, 1:300; #MAB5622, Millipore), or rabbit antibodies: glutathione S-Transferase pi (GSTpi, 1:500; #MSA-102, Stressgen, Ann Arbor, MI), anti-MAP2 (1:500; #AB5622, Millipore), fibronectin (1:400; #F3648, Sigma), or BDNF (1:50; #SC-546, Santa Cruz Biotechnology Inc, Santa Cruz, CA). Following 1-hour incubation in primary host-specific biotinylated secondary antibody, antigens were visualized with a different colored Alexa Fluoro-streptavidin conjugate as before. Finally, all sections were either stained with the nucleic acid marker TOPRO-3 (TPR3; 1:1000 in PBS, Invitrogen) or DAPI (Vectashield, Vector). Sections were mounted on slides and cover-slipped with anti-bleaching medium (Vector).

Multi-IHC-labeled sections were imaged by confocal microscopy (LSM Pascal System, Zeiss, Thornwood, NY) in single channel acquisition mode to prevent crosstalk and with fluorophore-appropriate laser excitation and filter-sets configured for optimal emission. Montage (X-Y) data were acquired with a 12-μm optical section thickness at x40 magnification over 4 to 8 fields of view in automatic mode. Z-axis data were acquired at x63 at 1 μm optical thickness. These data are shown to demonstrate that the positive staining is more widespread compared to that typically shown in a x40 or x63-objective field-of-view. Optimal primary and secondary antibody concentrations were obtained using standard dilutions as single labels and test sections were run without primary antibody to confirm the specificity of the labeling for each secondary antibody isotype. Double-label test sections were run exhaustively without one of the primary antibodies to confirm the specificity of antibody binding, especially when using 2 mouse primary antibodies. At least 3 injured brains were double-labeled (1 section every 600 μm) for each doublet of IHC markers to assess reproducibility and to summarize trends.

Cell Density Counts

GAP43+ cell profile counts were made on 3 to 5 single label DAB-stained IHC sections/brain that contained the injured/ipsilateral sensorimotor or parietal cortex using sections spaced at least 600 μm apart. The total number of GAP43+ profiles, defined as either a stained cell body or neurite/fibril-like segment, were counted in the entire ipsilateral grey and white matter encompassing the region from the midline to the rhinal fissure (Fig. 1A). Large, tangentially lying GAP43+ processes were counted once. The counts were expressed as average density of GAP43+ profiles in the area of evaluated tissue

Due to the high density of WFA stained cells in uninjured brain, counts of WFA+ cells were performed using the optical fractionator probe technique with StereoInvestigator software (MicroBrightfield, Williston, VT) on alternate sections adjacent to those used for the GAP43 analysis. To reduce effects of minor variation in location of contusion boundaries with respect to the cell counting area, the anatomical boundaries of the counting area at each time-point were first defined on 3 to 5 single label DAB-stained IHC sections/brain (600 μm apart) containing the injured sensorimotor or parietal cortex. This was achieved by determining the average distance between the midline and the most medial-lying hypointense (contused tissue) cortical grey matter region, and a tangential line from the rhinal fissure to the white matter and the most ventral-lying hypointense cortical grey matter. These medial and lateral distances were averaged for each brain and then for all injured brains at each time-point. The resulting medial and lateral average measurements/time-point were then used to draw the contour boundaries of the counting area on each of the 3 to 5 sections/brain for both injured and sham-injured groups. This resulted in a conservative lesion boundary in all injured brains that encompassed both edges of some contused tissue as well as normal-appearing tissue containing WFA+ cells. Counts were made at x200 magnification with a counting frame size of 550 μm² and between 15 to 25 sampling sites of 175 μm²/section. Only WFA+ cells with a characteristic ring-shaped, stained soma and well-stained soma-initial processes were counted. Small, round cells with intracellular staining but with no processes were not counted. Data are expressed as area density counts rather than estimated cell numbers.

In Situ Hybridization

Western Blots

On post-surgery day 7, rats were briefly anesthetized with isofluorane, decapitated and the brains quickly removed and placed on ice for dissection. Tissue blocks of ~3 to 4 mm³ were taken from the injury zone encompassing both the contusion core and bordering regions in the left injured cortex and from a similar region in the left/ipsilateral cortex of sham injury and naïve rats, as previously described (26). Tissue was stored at −80°C before being harvested by homogenization in lysis buffer containing Tris-HCl (0.05M, pH = 7.5), NaCl (0.15M), protease inhibitors (Complete,™, Roche, Basel, Switzerland) and centrifuged at 13,000 rpm for 30 minutes at 4°C. Protein concentration was measured using the RC-DC Protein Assay Kit (BioRad, Hercules, CA). Protein (25 μg) was mixed with 2× Laemmli buffer and boiled for 5 minutes and proteins were resolved on a 15% SDS-PAGE gel (BioRad) running at 33 mA for 60 minutes with a set of molecular weight standards –Precision Plus Protein Standards 10–250 kD (BioRad). Proteins were electro-transferred to polyvinylidene diflurodide membrane (BioRad, Hercules, CA) and blots were stained for total protein with Sypro Ruby (BioRad) and imaged under ultraviolet light (Fluoromax system, BioRad) for visual confirmation of equal lane loading. After washing in a solution of TRIS-buffered saline, Tween-20 (TTBS) and blocking in 5% milk for 1 hour, membranes were probed overnight with BDNF primary antibody (1:1000; #SC-546, Santa Cruz Biotechnology). Membranes were washed in TTBS for 3× 10 minutes and then incubated with anti-rabbit HRP-conjugated secondary antibody (1:10,000 in 1% milk) for 60 minutes. Following washing, antigen detection was achieved using an enhanced chemiluminescent substrate kit (SuperSignal West Femto Substrate, Thermo, Fisher Scientific, Waltham, MA) and images of the membranes were acquired using a camera system sensitive to chemiluminescence (Fluoromax system, BioRad). Blots were stripped and re-probed for β-actin (1:500; SC-1616, Santa Cruz Biotechnology) and re-imaged. The protein band corresponding to the mature form of BDNF at 14 kD was determined by reference to the molecular weight standards and quantified by integrated optical density measurement using ImageJ (36) and normalized to the corresponding integrated optical density. All gels were run with naïve control samples that were used to normalize the data for all samples from each gel to control for minor changes in experimental conditions.

Statistics

For all data the Kolmogorov-Smirnov with Lillie correction and Levene median tests were used to describe the distribution of the data and to determine the equality of the variances, respectively. Having passed these normality tests, group numerical data were expressed as the means ± standard error of the mean. For GAP43 IHC data, a one-way ANOVA was used to test for an overall difference between the groups and a post hoc test (Dunnett test) was used to determine individual group differences. GAP43 mRNA ROI data were not corrected for multiple comparisons since a priori evidence suggests that primary and secondary cortices differ in GAP43 mRNA expression (34) and can thus be treated as independent parameters. An initial 2-way ANOVA was used to test for overall group and time effects in the WFA data. Since there were no significant differences between sham groups the data were pooled and used in a 1-way ANOVA to test for group differences across time. Statistical significance (α) was set at p < 0.05 for all comparisons.

Pericontusional GAP43+ Neuronal Sprouting Is Maximal at 7 Days Post-injury

We first investigated the extent of cortical neuronal sprouting after TBI using GAP43 IHC. GAP43+ cellular staining in the hippocampus in both sham and injured rats was used as a positive control for antibody staining. At 7 days post-injury GAP43+ profiles were noted in most sections that contained contused parenchyma and were visible for several millimeters from the contusion edge (Fig. 1A). GAP43+ perikarya, often with a stained process (Figs. 1C, 2B), were distributed around the vertical edges of the contusion, underneath the contusion core within layer V and within the corpus callosum (Fig. 1A–C). GAP43+ perikarya and processes were seen in the gray matter on either side of the contusion core and often extended from the deeper to more superficial cortical layers (Fig. 1D), and there were growth cone-like GAP43+ blebs noted around the contusion edges (Fig. 1E). No GAP43+ processes were seen in the contralateral cortex or in any cortical region of the sham-injured brains. There were a few GAP43+ profiles in the caudate in a few brains, but no other brain region showed the marked changes seen in the cortex ipsilateral to the contusion.

Temporal analysis of GAP43+ process density showed that sprouting peaked at 7 to 14 days and that by 28 days, no GAP43+ staining was observed (Fig. 1F). The density of GAP43+ processes was higher at 7 than 14 days after injury, although this did not reach significance (p = 0.09). Since there are reports that GAP43 can be expressed by astrocytes (39, 40) or immature oligodendrocytes (41), we validated the phenotype of GAP43 expression using double-labeled IHC. No GAP43+ cells expressed the mature neuronal somatic marker NeuN, the somato-dendritic marker MAP2 (Fig. 2A–C), or the mature oligodendrocyte marker GSTpi (Fig. 2D).

Since protein synthesis may be reduced early after fluid percussion brain injury (42), we wanted to exclude the possibility that we had underestimated the sprouting response due to potential effects of trauma on protein synthesis after CCI. Therefore, we mapped cortical GAP43 mRNA using in situ hybridization with a riboprobe similar to that used previously to determine GAP43 distribution in the adult rodent brain (34). We performed the analysis at 7 days after injury, i.e. the peak time of the IHC response in ipsilateral sensorimotor and parietal cortex regions. Data showed the expected high hippocampal anti-sense probe binding in both injured and sham-injured rats (Fig. 3A, C) and no significant binding in any region with the sense probe (Fig. 3B, D). Cortical ROI analysis showed that GAP43 mRNA, expressed as the ipsilateral/contralateral ratio, was significantly increased above sham values in the cortex of injured rats, in both far lateral regions in all cortical layers and in medial superficial layers (Fig. 3A′ versus C′, E). The increase in GAP43 mRNA was more widespread compared to the IHC data, wherein increases were largely limited to pericontusional tissue, suggesting that translation into protein is affected by CCI.

Pericontusional Perineuronal Nets Are Reduced by Injury at 7 Days and Begin to Repopulate by 28 Days

Our previous work had shown qualitative decreases in perineuronal net (PNN)-associated CSPGs after injury (26) and CSPGs in their normal arrangement are likely to present a significant barrier to sprouting axons and/or new synaptic connections. Therefore, we used the lectin WFA to label the CSPG-rich PNNs of cortical cells through binding of the cell surface molecule N-acetylgalactosamine (43). WFA+ cells were almost completely absent from the pericontused cortical regions despite the presence of normal, intact tissue at 7 days; this absence occurred up to several mm from the contusion edge (Fig. 4A, B). Within the cystic contusion cavity there were numerous, highly intense WFA+ cells that appeared to be condensed dead cells, infiltrating glia or fibroblasts (Fig 5A). Density counts of the pericontused cortex revealed that there was a significant group but no time effect of injury on WFA+ cell density (Fig. 4D, p < 0.001), and that the decrease in cell density due to injury was significant at all times (p < 0.05). At 28 days some normal appearing, branched WFA+ cells were often noted within the pericontusional area (Fig 4E), but overall the cell numbers was still significantly depleted compared to sham injured animals (Fig 4F).

Sprouting Neurons Occur in Regions of Low Perineuronal Nets and Low Extracellular Chondroitin Sulphate Proteoglycans

We hypothesized that spontaneous neuronal sprouting would occur only in brain regions in which either PNN or CSPG or both were reduced after TBI. Data showing an absence of IHC double-labeling for sprouting neurons and PNNs appear to support this hypothesis (Fig. 5A, B), since GAP43+ neurons generally appeared only in pericontused regions that are deficient in WFA+ cells at 7 days post-injury, and no sprouting appeared contralateral within the normally distributed WFA+ cells (Fig. 5E). In some cases, pericontused GAP43+ neurons were partially WFA+ (Fig. 5B). Cortical GAP43+ fibers were noted in the gray matter in layers II to V (Fig 5C) and in the contusion cavity among highly intense, WFA+, dead or infiltrating cells (Fig. 5D), although none were co-labeled.

Since TBI results in a temporary reduction in extracellular CSPG proteins (26), we suspected that reduction in CSPGs contained within PNNs would occur in similar brain regions, i.e. that the injured brain environment is temporarily less inhibitory to sprouting than previously thought. Our current data appear to support this supposition for day 7 post-injury. In contralateral cortex, densely arranged WFA+ cells were associated with intense versican staining (Fig. 6A, B left, 6C), but in cortex ipsilateral to contusion there was a sharply demarcated zone where both the incidence of WFA+ cells and versican intensity were reduced (Fig. 6B, D).

We also looked at the spatial arrangement of sprouting neurons relative to regions of reduced extracellular CSPGs by staining for neurocan and aggrecan (Supplementary Fig. 1). As anticipated from results with versican, GAP43+ neurons were found in regions where extracellular neurocan was low, such as within some of the pericontused tissue regions compared to contralateral cortex (Supplementary Fig. 1A, top of image, vs. 1B) and compared to some pericontused regions that were hyperintense for extracellular neurocan, consistent with the development of a glial scar in this model (Supplementary Fig. 1A). Most GAP43+ perikarya were neurocan−, but some were neurocan+ (Supplementary Fig. 1A inset). Unlike in the contralateral cortex where only the outer rims of perikarya were positively stained for neurocan, however, staining was intracellular within perikarya of the pericontused cortex; this likely indicates that the neurons are actively synthesizing neurocan for extracellular deposition. GAP43+ neurons did not appear to be arranged in a similar fashion in sections stained for aggrecan, i.e. sprouting neurons were observed in regions containing aggrecan+ hypertrophied fibers, presumed to be glial cells, and the GAP43+ cells were often surrounded by an aggrecan+ PNN (Supplementary Fig. 1C, D).

Sprouting Neurons Occur Within Regions of Poly-Sialic Acid Neural Cell Adhesion Molecule+ Cells But Are Not Restricted to Regions Containing Positive Effectors of Sprouting

Double-staining was conducted for GAP43 and the growth-associated protein PSA-NCAM), a protein often associated with sprouting neurons but also with actively proliferating cells. There was robust upregulation of PSA-NCAM+ cells throughout the pericontusional cortex at 7 days (Fig. 7A), many of which appeared to be immature and/or glial cells, although none were seen contralaterally (Fig. 7B vs. D). GAP43+ perikarya and processes were associated with gray and white matter regions containing PSA-NCAM+ cells (Fig. 7A merge, 7C); however, on the cellular level there was only occasional overlap between these processes (Fig. 7B merge).

Fibronectin is a reparative molecule that also promotes cellular growth and has been reported to be enhanced after TBI (44). Therefore, we sought to determine whether it is enhanced in regions with sprouting neurons. By double-labeled IHC, fibronectin+ cells and blood vessels were limited to the most severely contused injury core regions, with only very minor incursions within the pericontused tissue (Supplementary Fig. 2A, B). GAP43+ neurons often overlapped with fibronectin+ regions at the contusion edge (Supplementary Fig. 2C), but they never co-labeled with fibronectin+ cellular structures; GAP43+ axonal profiles were often separated spatially from fibronectin+ regions (Supplementary Fig. 2C, E). Fibronectin immunostaining was never seen in other, more remote, ipsilateral regions or in the contralateral cortex.

Enhanced BDNF levels are associated with increased plasticity and promotion of a growth-permissive environment (45). We reasoned, therefore, that BDNF might be enhanced within pericontused regions containing sprouting neurons. By 7 days post-injury, the ipsilateral cortex remote from the contusion injury showed similar BDNF fluorescent intensity to the contralateral cortex (Fig. 8A, B); many GAP43+ neurons were intensely stained with BDNF (Fig. 8A₂). Within pericontused regions, often the site of a significant number of densely arranged GAP43+ processes; however, BDNF staining intensity was markedly reduced (Fig. 8A-insets, 8A merge and 8A₃). GAP43+/BDNF− fibers were seen a short distance from the contusion edge (Fig 8A₁). In parallel, Western blot analysis of tissue from around the contusion core showed an almost 2-fold reduction in BDNF protein compared to sham-injured levels (Fig. 8C; p = 0.06). This occurred despite the likely inclusion of normal tissue within the sample due to the limited ability to discriminate between injured and non-injured brain when sampling for Western blots.