Little is known about microglial activation and macrophage localization after diffuse brain injury (DBI). DBI-mediated perisomatic traumatic axonal injury (TAI) was recently identified within the neocortex, hippocampus, and thalamus, providing an opportunity to characterize immune cell responses within diffusely injured brain loci uncomplicated by contusion. By using moderate midline/central fluid percussion injury, microglial/macrophage responses were examined with antibodies targeting immune cell phenotypes and amyloid precursor protein, a marker of TAI. Parallel assessments of blood-brain barrier alterations were also performed. Within 6 to 48 hours postinjury, microglial activation within injured loci was observed, whereas microglia within non-TAI-containing regions maintained a resting phenotype. Microglial activation shared a spatiotemporal relationship with TAI though no clear interactions were observed. By 7 to 28 days postinjury, activated microglia contained myelin debris, yet revealed limited aggregation. Immunophenotypic macrophages were also localized to injured loci. Select macrophages approximated somatic membranes of perisomatically axotomized neurons with evidence of bouton disruption. No causality was established between blood-brain barrier alterations and these inflammatory responses. These findings indicate rapid, yet initially nonspecific, and persistent microglial/macrophage responses to DBI. DBI-mediated inflammatory responses suggest further expansion of traumatic brain injury histopathologic evaluations to identify neuroinflammation indicative of diffuse pathology.
Traumatic axonal injury
The immune response to traumatic brain injury (TBI) involves microglia as well as leukocytes recruited from the surrounding vasculature. To date, this immunologic response has been studied most extensively in the context of focal TBI, despite the fact that diffuse TBI remains the most significant component of the morbidity associated with traumatic injury. Focal injuries, often associated with blunt force trauma, are associated with cerebral contusions and hematomas, whereas diffuse injuries, a common result of motor vehicle accidents, depend on inertial forces and result in more subtle scattered microscopic pathology. To date, immune cell responses to experimental focal brain injury have been studied extensively (1-6). Additionally, limited studies have documented generalized immune cell responses to diffuse brain injury (DBI) (7, 8), yet both clinical and experimental studies have been complicated by focal brain contusion and related hemorrhagic components (9-11), allowing leukocytes to enter the injured brain parenchyma via overt disruption of the blood-brain barrier (BBB). This contusion-related BBB disruption also permitted the unregulated passage of blood-borne stimulatory molecules complicating the evaluation of immune cell activation.
Fortunately, in the experimental setting, animal models have been developed to minimize the potential for contusion and/or hemorrhage while maintaining many of the important features of diffuse TBI, including diffuse axonal injury and its attendant anterograde and retrograde consequences. Specifically, the use of midline/central fluid percussion injury (cFPI) in the rat allows for the generation of traumatic axonal injury (TAI), the experimental counterpart of diffuse axonal injury, characterized by the features of focal impairment of axonal transport, leading to progressive axonal swelling and secondary axotomy (12-15). Moderate cFPI can induce perisomatic TAI (i.e. within 40-60 μm of the sustaining soma), allowing for the critical assessment of both retrograde neuronal somatic responses to injury as well as specific anterograde/Wallerian responses (16, 17). Using this model, TAI has been observed scattered within various anatomical loci, including the mediodorsal neocortex, hippocampal dentate gyrus, and dorsolateral thalamus without any associated contusions and/or hemorrhagic-mediated tissue damage (16, 17). These factors provide the opportunity to explore microglial/macrophage responses within diffusely injured brain regions uncomplicated by focal brain pathology.
In the current study we used immunocytochemical light, electron, and confocal microscopic characterization of the immune responses to DBI in the absence of focal tissue damage. Microglia within brain loci revealing TAI responded rapidly, albeit in an initially nonspecific fashion, via changes in their cellular morphology. This response was followed by their persistent activation and phagocytic activity. In contrast, the microglia in non-TAI-containing regions maintained a resting phenotype. Immunophenotypic macrophages displayed similar spatiotemporal responses to DBI and its associated TAI. Select macrophages approximated scattered somata of traumatically injured neurons showing evidence of synaptic disruption and loss. Taken together, these findings suggest that neuroinflammatory responses to DBI with TAI may be distinct from those related to more focal TBI pathology. These findings have implications for postmortem human histopathologic evaluation of TBI and its characterization based upon specific microglial/macrophage responses.
Materials and Methods
Animal Preparation and Injury
To explore microglial/macrophage responses to DBI, rats were subjected to moderate cFPI consistent with methods described previously (16-18). Adult male Sprague-Dawley rats (375-400 gm) were anesthetized with 4% isoflurane in 70% N2O and 30% O2, intubated, and maintained on a ventilator with 1% to 2% isoflurane for injury preparation. Intubated animals were placed on a heating pad connected to a thermostat controlled by a rectal probe (Harvard Apparatus, Holliston, MA) to maintain 37°C body temperature. To prepare the animal for cFPI, the top portion of a Leur-Loc syringe hub of a 20-gauge needle, 2 fixation screws, and dental acrylic were fixed to a midline craniotomy in the skull over the intact dura and then connected to the injury device. Briefly, a 4.8-mm circular craniotomy along the sagittal suture midway between bregma and lambda was generated with care taken not to disrupt the underlying dura and superior sagittal sinus. The top portion of the Leur-Loc hub (Becton Dickinson, Franklin Lakes, NJ) was cut away from the 20-gauge needle, beveled, scored, and affixed over the craniotomy site using cyanoacrylate. After the integrity of the seal between the hub and the skull was confirmed, fixation screws were inserted into 1-mm holes drilled into the right frontal and occipital bones. Dental acrylic (Hygenic Corp., Akron, OH) was applied around the hub and over the screws and allowed to harden to provide stability during the injury induction. After the dental acrylic hardened, the skin was closed over the hub with sutures, topical lidocaine ointment was applied, and the animal was removed from anesthesia and monitored in a warmed cage until fully recovered (∼1 hour).
Before injury, each animal was again anesthetized with isoflurane. The incision was quickly opened and the male end of a spacing tube was inserted into the Leur-Loc hub. The female end of the spacer-hub assembly, filled with normal saline, was then inserted onto the male end of the fluid percussion device, ensuring that no air bubbles were introduced into the system. An ∼2.1 atm (range 1.9-2.3 atm) injury was administered, consistent with brain injury of moderate severity (16-18). The pressure pulse measured by the transducer was displayed on a storage oscilloscope (Tektronix 5111; Tektronix, Beaverton, OR), and the peak pressure was recorded. Injury preparation and induction were completed before the animal's recovery from anesthesia. After injury, the spacer-hub assembly was immediately removed en bloc, bleeding was controlled with Gelfoam (Pharmacia, Kalamazoo, MI), and the incision was closed with sutures. Animals were monitored for spontaneous respiration and, if necessary, ventilated with room air to ensure adequate postinjury oxygenation. Postinjury recovery times for the following reflexes were recorded: toe pinch, tail pinch, corneal blink, pinnal, and righting. After recovery of the righting reflex, animals were placed in a holding cage with a heating pad to ensure maintenance of normothermia and monitored until the appropriate perfusion time. For sham-injured control animals, the above steps were followed without injury induction. In that righting was the last reflex to recover, the time frame encompassing this parameter was used to ensure injury of comparable severity. All injured animals had righting reflex recovery times greater than 6 minutes but less than 8 minutes compared with less than 2 minutes for sham-injured animals, indicating injuries of comparable severity (data not shown). Experiments were conducted in accordance with National Institutes of Health and institutional guidelines concerning the care and use of laboratory animals (Institutional Animal Care and Use Committee).
Animals (n = 6/injury group, 2/sham injury group, and 2 naive) were killed at 6 and 24 hours as well as at 2, 7, 14, 21, and 28 days postinjury via an overdose of sodium pentobarbital intraperitoneally (150 mg/kg) and were perfused transcardially with 4% paraformaldehyde in Millonig's buffer. After perfusion, brains were removed and blocked in a coronal blocking device to include the thalamus with overlying neocortex and hippocampus (Fig. 1). Previous studies have identified TAI scattered throughout these regions,consistent with the occurrence of diffuse injury (16, 17). The tissue block was flat-mounted on a metal plate with cyanoacrylate, embedded in agar, and sectioned in 0.1 M phosphate buffer at a thickness of 40 μm using a Vibratome (Leica Microsystems, Bannockburn, IL). Serial coronal sections (n = 60 sections at 40 μm/section) were collected starting from 1,600 μm caudal to the anterior commissure. This sampling strategy allowed for a comprehensive examination of the neocortex, hippocampus, and thalamus (19). Systematic uniform sampling of coronal sections was used with every fifth section collected for a total of 12 sections per animal. Specific anatomical loci known to elicit DBI-mediated TAI, namely the mediodorsal neocortex, hippocampal dentate gyrus, and dorsolateral thalamus, were the focus of extensive bilateral observations, allowing for examination of 24 fields from each animal at each time point. Additional tissue was stored in Millonig's buffer in 12-well culture plates (Falcon, Newark, DE).
The low-magnification microphotograph outlines diffusely injured brain loci revealing significant traumatic axonal injury (TAI) uncomplicated by focal pathology in mediodorsal neocortex (A), hippocampal dentate gyrus (B), and dorsolateral thalamus (C). Activated microglia and macrophages were localized to these areas after diffuse brain injury. In contrast, asterisks delineate related regions also sampled in the current investigation that did not contain significant TAI or other forms of overt pathology.
Immunocytochemistry and Other Labeling Approaches for Confocal Microscopy
Double-labeling strategies were used to permit simultaneous visualization of microglia/macrophages and TAI, a key feature of diffuse TBI. To label microglia, tissue sections were incubated overnight with lectin (Alexa 488-conjugated isolectin B4 from Griffonia simplicifolia [5 μg/mL]; Molecular Probes, Eugene, OR) in 20% normal horse serum (NHS) in artificial cerebrospinal fluid (aCSF) (126 mM NaCl, 3 mM KCl, 1.25 mM NaH2PO4, 2 mM MgCl2, 2 mM CaCl2, and 10 mM glucose buffered with 26 mM NaHCO3) consistent with previously described methods (20-23). TAI was then visualized using an antibody to the amyloid precursor protein (APP) as described below. APP moves via anterograde transport and pools at sites of impaired axonal transport, thus serving as a marker of TAI (24). To ensure maintenance of lectin binding, all steps of the lectin-APP double-labeling procedure were conducted in aCSF.
To label macrophages, sections were first preincubated for 60 minutes in 10% NHS with 0.2% Triton X-100 in PBS and then incubated overnight with rat-specific macrophage (1:2000; mouse anti-CD68 [ED1]; Serotec, Oxford, England) primary antibody in 1% NHS in PBS/2% bovine serum albumin. After removal of the antibody solution, sections were rinsed and incubated for 2 hours in Alexa 488 goat anti-mouse IgG (1:1000; Molecular Probes) with 1% NHS in PBS/2% bovine serum albumin before processing for TAI. In addition to double-labeling strategies for either microglia or macrophages with TAI, additional tissue sections were subjected to lectin-CD68 dual-labeling to provide further assessment of microglial/macrophage phenotypes, given the transition of activated microglia into macrophages after injury. All steps were performed in aCSF, and Alexa 594 goat anti-mouse IgG (1:1000) secondary antibody was used to identify CD68 immunoreactivity.
For labeling TAI, sections were rinsed and incubated with 10% normal goat serum (NGS) in PBS (except for lectin-APP processing which used aCSF) for 45 minutes followed by an overnight incubation in C-terminus-specific APP primary antibody (1:1000; rabbit anti-C-APP; Zymed, San Francisco, CA) with 1% NGS in PBS. Sections were then rinsed and incubated for 2 hours with Alexa 594 goat anti-rabbit IgG (1:1000) in 1% NGS in PBS. Sections were rinsed 6 times in PBS for 5 minutes and twice in 0.1 M phosphate buffer for 10 min; following these rinses, sections were mounted on gelatin-coated slides. Sections were mounted with an anti-fade mounting medium (ProLong; Molecular Probes), cover-slipped, and sealed with nail polish. As an internal control, additional sections were processed as described above; however, primary antibodies were omitted from the procedure. Images were captured using a confocal microscope (Leica TCS-SP2 AOBS; Leica Microsystems) with appropriate excitation lasers, detectors, and analysis software.
Immunocytochemistry for Routine Light and Electron Microscopy
In these protocols, sections were processed with antibodies to either CD11b/c or CD68 alone or together with antibodies to APP to detect diffusely injured axons. CD11b/c antibody recognizes the microglia-specific complement type 3 receptor, thus serving as a complementary marker to lectin. In addition to these approaches, adjacent brain sections were processed with antibodies to endogenous rat albumin. This was done to assess the potential for injury-induced BBB disruption. Animals for light microscopy (LM) and electron microscopy (EM) evaluation (n = 3/injury group, 1/sham-injury group, and 1 naive) were perfused with 4% paraformaldehyde/0.1% glutaraldehyde in Millonig's buffer. Sections were then processed for CD11b/c, CD68, and albumin or CD11b/c-APP and CD68-APP dual immunoreactivity (24). Briefly, endogenous peroxidase activity was blocked using 0.3% H2O2 in PBS for 30 minutes followed by temperature-controlled modified microwave antigen retrieval (25) and preincubation in 10% NGS with 0.2% Triton X-100. Sections were incubated overnight with either CD11b/c (1:700, mouse anti-rat CD11b/c; BD Pharmingen, San Diego, CA) or CD68 (1:700) primary antibody. Sections were then incubated for 1 hour with biotinylated rat-absorbed goat anti-mouse IgG (1:200; Vector Laboratories, Burlingame, CA) secondary antibody. Select sections for dual-labeling were subjected to additional processing beginning with preincubation in 10% NGS followed by C-terminus-specific APP (1:1000) primary antibody. At this time, sections for single-label albumin (1:40,000; rabbit anti-rat albumin antiserum; Bethyl Laboratories, Montgomery, TX) primary antibodies were also processed in 1% NGS in PBS. Sections processed with APP or albumin antibodies were then incubated for 1 hour with biotinylated rat-absorbed goat anti-rabbit IgG (1:200 Vector). All sections were then visualized via incubation in avidin-horseradish peroxidase complex (Vectastain ABC Standard Elite Kit; Vector Laboratories) for 1 hour followed by 0.05% diaminobenzidine, 0.01% H2O2, and 0.3% imidazole in 0.1 M phosphate buffer for 10 to 20 min. The common chromogen diaminobenzidine was used for convenience in that structural correlates of BBB disruption, axonal injury, and microglial/macrophage change were so discreet as to create no problem with morphologic recognition. Sections processed for either CD11b/c, CD68, or albumin as well as select sections processed for CD11b/c-APP and CD68-APP were mounted on gelatin-coated slides, dehydrated, and cover-slipped for routine LM evaluation, whereas the remaining dual-labeled sections were subjected to continued processing for EM. As an internal control, additional sections were processed as described above; however, primary antibodies were omitted from the procedure. LM images were captured using an Eclipse 800 microscope (Nikon, Tokyo, Japan) fitted with a Spot-RT digital camera (Diagnostic Instruments, Sterling Heights, MI).
As noted, dual-labeled sections were processed for EM to ascertain ultrastructural detail relevant to those changes observed via confocal microscopy. After antibody processing, the tissue was osmicated in 1% OsO4 and then placed in graded alcohols and propylene oxide before placement in epoxy resin (Ted Pella, Redding, CA). Sections were then embedded between plastic slides (Thomas Scientific Co., Swedesboro, NJ), placed in a 55°C oven for 3 days, and then scanned to identify immunoreactive axonal swellings along with microglia/macrophages. Once identified, these sites were removed, mounted onto plastic studs, and thick sectioned to the depth of interest using an ultramicrotome (Ultracut R; Leica, Vienna, Austria). Semithin sections (1 μm) stained with 1% toluidine blue were again screened for evidence of microglial activation/macrophage localization with TAI. Serial thin sections (70 nm) were then cut, picked up onto Formvar-coated slotted grids, and stained in 5% uranyl acetate in 50% methanol for 2 minutes and 0.5% lead citrate for 1 minute. Images were captured using a JOEL 1230 electron microscope with a digital camera (Gatan Digital Micrograph, Pleasanton, CA).
Naive and Sham Injury-General Findings and Cellular Phenotypes
Macroscopically, sham-injured brains showed no evidence of compression, contusion, or tissue loss. Tissue sections from naive and sham-injured animals processed for APP and examined by light and confocal microscopy demonstrated limited background staining with only isolated immunoreactive somata. There was no evidence, however, of immunoreactive axons or swellings adjacent to these somata. Microglial markers (isolectin B4 and CD11b/c) demonstrated conventional resting morphology in that microglia exhibited a small cell body with highly ramified processes and were distributed evenly throughout the brain parenchyma (Fig. 2A-C). Sections processed for macrophage immunoreactivity (CD68) demonstrated a limited number of immunoreactive cells, primarily in perivascular locations (Fig. 3A-C). Sham-injured tissue sections processed for albumin revealed only limited immunoreactivity underlying the craniotomy site, together with minimal immunoreactivity within the subcortical white matter. Otherwise, albumin immunoreactivity was confined to the vascular lumen and walls, with no extension into the brain parenchyma (Fig. 4A).
Spatiotemporal evaluation of microglia activation within diffuse brain injury (DBI) loci. Using antibodies to CD11b/c that recognize the microglia-specific complement type 3 receptor, microglial morphology was visualized within loci known to elicit DBI-mediated traumatic axonal injury (TAI). (A-C) Note that at all matched postinjury time points, sham injury reveals microglia with resting morphology characterized by a small cell body with highly ramified processes. (D-F) Between 6 and 48 hours postinjury, most microglia demonstrate the initial signs of activation including increased cell body size, with a concomitant reduction in process ramification, leading to an increased immunoreactive ameboid-like morphology. However, others maintain a resting phenotype. Within DBI loci, activated morphology predominates and persists at 7 (G-I), 14 (J-L), and 28 (M-O) days postinjury. At all postinjury time points, microglial immunoreactivity within areas remote from sites containing TAI was similar to sham injury immunoreactivity (not shown). Scale bar = 100 μm.
Spatiotemporal evaluation of macrophage localization within diffuse brain injury (DBI) loci. Similar to microglial evaluation, antibodies to CD68 that recognize a macrophage-specific cell surface marker were used to visualize macrophages within loci known to elicit DBI-mediated traumatic axonal injury (TAI). (A-C) Sham injury at all matched postinjury time points reveals limited numbers of immunoreactive cells primarily in perivascular locales. Between 6 and 48 hours postinjury, parenchymal macrophage immunoreactivity can be observed within the neocortex (D) and hippocampus (E) with minimal thalamic (F) involvement. However, note that by 7 days postinjury, all 3 regions (G-I) demonstrate robust macrophage immunoreactivity that persists at 14 (J-L), and 28 (M-O) days postinjury. Similar to microglial immunoreactivity, at all postinjury time points macrophage immunoreactivity within areas remote from sites containing TAI was similar to sham injury immunoreactivity (not shown). Scale bar = 100 μm.
These low-magnification brain microphotographs illustrate the spatiotemporal course of albumin immunoreactivity after diffuse brain injury. Note that with sham injury, albumin immunoreactivity is confined to the vascular walls and lumen with limited immunoreactivity observed in the subcortical white matter. However, note that at 6 hours postinjury, diffuse albumin immunoreactivity can be observed throughout the interstices of the brain. By 24 hours postinjury, scattered immunoreactivity persists throughout the neocortex and hippocampus whereas the thalamus now reveals limited immunoreactivity. By 7 days postinjury, only minimal albumin immunoreactivity is present. Note that at all postinjury time points, no evidence of contusion, hemorrhage, or overt tissue necrosis can be seen.
Diffuse Injury-Light, Electron, and Confocal Microscopic Findings
Injured brains shared macroscopic features identical with those of sham-injured brains with the exception of limited subarachnoid hemorrhage found under the craniotomy site and limited petechial hemorrhages within the lateral aspect of the corpus callosum. There was no evidence of contusion, hemorrhage, or tissue disruption within regions demonstrating TAI.
APP immunoreactive axonal swellings, a signature of diffuse TBI, were visible at 6, 24, and 48 hours postinjury. Consistent with previous descriptions of experimental DBI in this animal model, axonal swellings were most prominent within the mediodorsal neocortex, hippocampal dentate gyrus, and dorsolateral thalamus with additional swellings found scattered in the corpus callosum, subcortical white matter, and brainstem (16, 17). Select swellings could be traced back to their somata of origin, whereas others were found in isolation. Focal axonal swellings demonstrated no distal immunoreactivity, consistent with axonal disconnection as previously characterized by Kelley et al (17). By 7 days postinjury, these immunoreactive axonal swellings were no longer visible.
Albumin Immunoreactivity as a Marker of BBB Alteration
At 6 hours postinjury, endogenous albumin immunoreactivity could be identified distributed diffusely throughout the interstices of the neocortex, hippocampus, and thalamus. This blood-brain disruption to the normally intravascularly confined serum albumin occurred without evidence of overt contusion, hemorrhage, or tissue disruption (Fig. 4B). By 24hours postinjury, scattered albumin immunoreactivity persisted within the interstices of the neocortex and hippocampus, whereas the thalamus now revealed limited immunoreactivity, again confined to the extracellular compartment (Fig. 4C). By 7 days postinjury, only minimal albumin immunoreactivity was evident with the tissue now appearing reminiscent of that obtained from sham-injured animals (Fig. 4D).
Microglia/Macrophage Responses to Diffuse Brain Injury
Six and 24 hour postinjury time points revealed findings comparable to observations at 2 days postinjury; however, these observations differed significantly from findings observed between 7 and 28 days postinjury. Accordingly, we report our observations grouped at the acute (6-24 h), postacute (2 d), and long-term (7-28 days) postinjury time points to illustrate the continuum of immune cell responsiveness after injury.
Acute Neuroinflammatory Responses
At 6 to 24 hours postinjury, LM and confocal evaluation revealed scattered immunoreactive microglia within the mediodorsal neocortex, hippocampal dentate gyrus, and dorsolateral thalamus that maintained highly ramified processes, consistent with a resting phenotype. Other scattered microglia demonstrated reduced immunoreactivity within their processes together with rounding of their cell bodies, all of which were suggestive of the initial stages of activation (Fig. 2D-F). Microglia found scattered among diffusely injured axons did not show any consistent spatial relationship to the axonal swellings, the proximal axonal shafts, or the downstream disconnected axonal segments that now revealed early anterograde change (Fig. 5). In contrast with those microglia found within these diffusely injured loci, microglia within the adjoining parenchyma wherein no axonal swellings could be found, maintained a resting morphology.
This confocal micrograph using composite image planes is from the dorsolateral thalamus at 24 hours postinjury after double-label processing for microglia and traumatic axonal injury (TAI). Microglia (isolectin B4: green, double arrowheads) and TAI (amyloid precursor protein: red, single arrowheads) were identified to assess potential interactions between inflammatory cells and TAI, a key characteristic of diffuse brain injury (DBI). Note that although microglial activation and DBI-mediated TAI pathogenesis share a spatiotemporal relationship, there is no evidence of direct microglial association with any component of injured axons at this postinjury time frame. Although activated microglia can be seen scattered among traumatically injured axons, no direct clustering around these fibers occurs. Scale bar = 20 μm; Note: Isolectin also stains blood vessels (BV).
At the EM level, microglia were identified via the presence of electron-dense CD11b/c reaction product that outlined their cell membranes. Resting and activated microglial ultrastructural features were similar, although the resting cells displayed highly ramified appendages, whereas activated cells revealed enlarged, amoeboid-like cell membranes. Activated microglia contained a prominent, round nucleus with heterogeneous chromatin clumps beneath the nuclear envelope. Their cytoplasm revealed granular endoplasmic reticulum with long, narrow cisternae extending through the cytoplasm. Mitochondria together with electron-dense laminar bodies as well as inclusion bodies characteristic of lipofuscin were also distributed throughout the cytoplasm (Fig. 6). Despite axonal injury, no specific microglia-axonal associations were discerned at the EM level within the above-described loci. Activated microglia never directly engaged either the axonal swellings or their proximal/distal axonal segments (Fig. 6).
Electron micrograph of the dorsolateral thalamus at 24 hours postinjury. Dual-label antibody processing for microglia (CD11b/c) and traumatic axonal injury (TAI) (amyloid precursor protein [APP]) was used, although both appear electron-dense at the electron microscopy level. Note that the presence of an electron-dense reaction product and distinct ultrastructural morphology allow for the identification of both the microglia and reactive axons. The activated microglial ultrastructure (single asterisk) demonstrates an ameboid-like membrane, ringed with an electron-dense CD11b/c reaction product. Also note the prominent round nucleus with heterogeneous chromatin clumping beneath the nuclear envelope as well as granular endoplasmic reticulum with long, narrow cisternae extending throughout the cytoplasm. Electron-dense laminar bodies as well as inclusion bodies characteristic of lipofuscin can also be seen distributed throughout the cytoplasm. TAI (double asterisk) is characterized by axonal swellings containing electron-dense APP reaction product, organelle pooling, neurofilament misalignment, and microtubule loss. Additionally, more advanced axonal pathology (arrow) is found within this field. In this image, the microglia approximates the axonal swelling yet does not directly engage it, confirming light microscopy and confocal impressions. Scale bar = 2 μm.
In the 6- to 24-hour time frame, immunophenotypic tissue macrophages could also be identified by LM and confocal microscopy. These macrophages were localized to the neocortical and hippocampal parenchyma at 6 hours postinjury without thalamic involvement (Fig. 3D, E). The adjoining, non-TAI containing brain regions revealed no parenchymal macrophage immunoreactivity. Within the neocortex, immunophenotypic macrophages displayed a graded parenchymal distribution; heavier macrophage concentrations occurred in the superficial cortex abutting the subarachnoid space, with reduced macrophage concentrations within the deeper cortical layers. In contrast, the hippocampal macrophage distribution was more heterogeneous in that cells were scattered evenly throughout the hippocampal region. Although diffusely distributed within these loci, immunophenotypic macrophages were not associated with any component of axonal injury. EM evaluation of the macrophage ultrastructure using CD68 electron-dense reaction product membrane deposition was confirmatory, revealing activated microglia with no axonal associations.
Tissue sections processed for lectin-CD68 dual-labeling revealed distinct inflammatory cell phenotypes within diffusely injured loci. Within the mediodorsal neocortex and hippocampal dentate gyrus, inflammatory cells were primarily either single-labeled with lectin or double-labeled with both markers (Fig. 7). Occasional single-labeled, CD68-positive cells were observed within the parenchyma immediately adjacent to perivascular locales. Cells within the dorsolateral thalamus were typically labeled with lectin only. These findings were consistent with observations of single-labeled tissue sections using CD11b/c and CD68 to examine microglia and immunophenotypic macrophages respectively.
These confocal micrographs using composite image planes are from the mediodorsal neocortex at 24 hours postinjury after dual-label processing for microglia and macrophages. (A) is lectin staining (green), (B) is CD68 immunoreactivity (red), and (C) is the overlap. Two distinct activated microglial phenotypes are observed: single-labeled, lectin-positive activated microglia (arrows) and dual-labeled activated microglia expressing the macrophage marker CD68 (arrowheads). Activated microglia expressing the macrophage marker comprise a significant number of the immunophenotypic macrophages found within diffuse brain injury (DBI) loci. In addition to activated microglia, resting microglia (double arrowheads) were also observed within the DBI loci. Scale bar = 40 μm; Note: Isolectin also stains blood vessels (BV).
Postacute Neuroinflammatory Responses
By 48 hours postinjury, microglial immunoreactivity within the above-described loci again revealed resting and activated phenotypes, whereas microglia within adjoining, non-TAI-containing parenchyma maintained a resting morphology. Within the brain sites revealing diffuse TAI, activated microglia now appeared more numerous than previously seen at the 6- to 24-hour time frame. Activated microglia demonstrated reduced branch ramification and more intense immunoreactive amoeboid-like morphology. Despite enhanced activation, no evidence of direct microglial engagement of the traumatically injured axons was seen. EM evaluation revealed activated microglia scattered near sites of axonal injury; however, despite exhaustive EM evaluation, no evidence of microglial projections or axonal engulfment was found.
Routine LM and confocal microscopy also revealed immunophenotypic macrophages that persisted within neocortical and hippocampal parenchyma, with scattered thalamic immunoreactivity now recognized by 48 hours postinjury (Fig. 3F). Although numerous immunophenotypic macrophages were found in isolation scattered throughout these brain loci, occasional macrophages could now be seen approximating the somata of perisomatically axotomized neurons, particularly those within the neocortex (Fig. 8). Here the immunophenotypic macrophages encompassed significant portions of the neuronal soma, with many adopting a rounded or semicircular shape to closely follow the somatic contour. EM evaluation confirmed the close adherence of the macrophages to the neuronal somatic membrane (Fig. 9). Despite this approximation of the somatic membrane, there was no evidence of somatic engulfment or phagocytic activity; however, this macrophage investment was associated with somatic presynaptic terminal disruption (Fig. 9A) and loss (Fig. 9B).
Confocal micrograph using composite image planes of the mediodorsal neocortex at 48 hours postinjury. Double-label antibody processing for macrophages (CD68: green) and traumatic axonal injury (TAI) (amyloid precursor protein: red) were used to examine potential immune cell-neuron interactions after diffuse brain injury (DBI). Note that whereas macrophages can be identified within the parenchyma in isolation (single arrow), select macrophages can also approximate somata linked to TAI (double arrowhead). Here the macrophage encompasses a significant portion of the somatic membrane by adopting a semicircular morphology. Although the semicircular macrophage appears to partially encircle the soma, no phagocytic activity is observed. Scale bar = 8 μm.
These electron micrographs of diffuse brain injury (DBI) loci at 48 hours postinjury were processed after dual-labeling of macrophages (CD68) and traumatic axonal injury (TAI) (amyloid precursor protein) to identify these components using methodology similar to the microglia-TAI evaluation. In these examples, macrophages approximate somata within DBI-mediated TAI loci. Although select macrophages (asterisk) maintain their conventional rounded morphology (A), other microglia adopt semicircular morphology to approximate a significant portion of the somatic membrane (B). Macrophage ultrastructure is similar to that of activated microglia, whereas neuronal somatic ultrastructure (Nu) reveals no significant pathologic change. Note that this macrophage investment is associated with apparent synaptic disruption (box with enlargement in [A]). Other examples reveal somatic membranes devoid of synapses (B). Scale bar = 1 μm; enlargement scale bar = 250 nm.
Long-Term Neuroinflammatory Responses
By 7 days postinjury, LM and confocal microscopy revealed enhanced microglial/macrophage immunoreactivity within the diffusely injured loci (Figs. 2G-I, 3G-I) with increased immunophenotypic macrophages recognized within the thalamus (Fig. 3I). Once again, only resting microglial morphology and sparse macrophage immunoreactivity could be observed within adjoining, non-TAI-containing regions. LM evaluation of TAI-containing loci revealed no evidence of widespread necrotic neuronal cell death or tissue necrosis similar to that typically described after focal brain pathology. Similarly, no related microglial/macrophage clustering was observed in relation to these diffusely injured loci. At 14 to 28 days postinjury, activated microglial/macrophage immunoreactivity persisted within the previously identified neocortical, hippocampal, and thalamic loci (Figs. 2J-O, 3J-O). Limited activated microglia/macrophages were now identified within tissue immediately adjacent to these loci, with more remote sites continuing to maintain resting microglial morphology and sparse macrophage immunoreactivity. In contrast, enhanced macrophage immunoreactivity was recognized along the dentate gyrus granule cell layer (Fig. 3N). Although not observed at earlier time points, limited activated microglial aggregations/clusters were now seen, particularly within the thalamus at 28 days postinjury (Fig. 10). However, these aggregations, previously described after focal pathology, were not widespread within these injured loci. Comparable to LM observations from earlier time points, no evidence of necrotic cell death or tissue necrosis was found.
This micrograph reveals an atypical finding of microglial aggregations within the dorsolateral thalamus at 28 days postinjury. Using antibodies to CD11b/c, the majority of activated microglia within diffuse brain injury (DBI) loci do not cluster (arrow). However, isolated findings of grouped activated microglia, as typically seen after postmortem examination of focal TBI tissue, can be observed (circle). Scale bar = 100 μm.
Although evaluations of immune cell-axonal interactions at earlier time points did not reveal specific spatial associations, EM evaluation of activated microglia/macrophages at all time points beyond 7 days postinjury revealed consistent immune cell interactions in relation to injured axons. These interactions included immune cell projections and/or cell body contact with damaged axons and/or their related debris, with phagocytosis as inferred from the presence of myelin debris within the cytoplasm of the immunoreactive cells (Fig. 11). Similar to LM findings in this time frame, necrotic cell death or tissue necrosis was not observed. Those macrophages that localized to neuronal somatic membranes again did not reveal evidence of somatic engulfment or phagocytic activity.
These electron micrographs of diffuse brain injury (DBI) loci at 7 to 28 days postinjury identified microglia using electron-dense CD11b/c reaction product deposition on the cell membrane. These micrographs reveal activated microglial recognition of axonal debris (arrows) via projection formation (A) and direct cell body contact (B). Phagocytosis is inferred from myelin debris observed within the immune cell cytoplasm (boxed area). Neither widespread immune cell clustering nor somatic engulfment, each suggestive of neuronal cell death, is observed at this time interval. Scale bar = 2 μm.
The inflammatory response to TBI represents a coordinated effort by resident microglia and peripheral blood leukocytes to protect the brain after trauma. Although this protective function is essential for preserving viable tissue and promoting recovery, recent evidence suggests that the neuroinflammatory response itself may also be responsible for the initiation of delayed/secondary injury cascades (26). Accordingly, a comprehensive examination of TBI-mediated neuroinflammation is essential for designing rational approaches to therapeutically modulate pathologic aspects of this response. With this rationale in mind, the current study provides a spatiotemporal characterization of microglia/macrophage interactions after DBI as assessed by light, confocal, and electron microscopy. Microglial activation and macrophage localization were evaluated within the diffusely injured brain, revealing TAI in the absence of focal contusion or hemorrhage.
As intrinsic primary immune effector cells of the brain parenchyma, microglia are the first line of defense after traumatic insult (27). Focal TBI often results in primary axotomy, together with other destructive and ischemic neuronal changes, whereas diffuse TBI results primarily in diffuse axonal injury, known as TAI in the experimental setting, that leads to delayed or secondary axotomy without concomitant tissue disruption. Previous studies documenting focal TBI-mediated microglial activation have demonstrated a robust response to primary axotomy generated via either ex vivo slice preparation (20-22), in vivo stab wounding (23), or in vivo medial forebrain bundle transection (28), all of which involve direct tissue transection. By comparison, the microglial response to diffuse TBI and its associated axotomy have received little attention, with only indirect studies of this issue in rodents and an incomplete characterization in humans wherein microglial activation was suggested to have no specific associations with injured axons (7, 8). In this context, the current article significantly extends and clarifies the limited literature existing on this important area of inquiry. In this study, microglial activation shared a spatiotemporal relationship with TAI within diffusely injured brain loci that included the mediodorsal neocortex, hippocampal dentate gyrus, and dorsolateral thalamus. However, the activated microglia did not show, via confocal microscopy or EM, any specific and/or consistent relationship to sites of axonal injury in the acute/postacute stages (i.e. 6-48 hours) after injury. The observed activation persisted until at least 28 days postinjury and, in this sense, was consistent with time courses of microglial activation after brain and spinal cord trauma (9,21,29,30). In the current study, the immune cell recognition of axonal debris at the EM level occurred only between 7 and 28 days postinjury. In contrast to this microglial activation, other microglia within adjoining, non-TAI-containing parenchyma maintained a resting morphology. These data suggest an early generalized microglial response within DBI loci followed by a delayed targeted response to the attendant axotomy, with the microglia in non-TAI-containing regions remaining quiescent.
This initial microglial activation without specific axonal targeting suggests that in DBI, microglia are activated within their local microenvironments yet remain stationary despite DBI-mediated axotomy. A recent study by Nimmerjahn et al (31) supports this observation via documentation that resting microglia actively survey their local microenvironments, sending projections to areas of local tissue damage associated with overt tissue disruption and BBB compromise, while their cell bodies remain stationary. With diffuse injury the lack of migration and/or projection formation within the acute/postacute time frames may reflect the lack of appropriate chemotactic stimuli that are typically generated after focal brain injury wherein overt BBB and/or tissue disruption occurs (6, 23). Further, moving on the premise that the unregulated release of intra-axonal contents into the surrounding parenchyma after primary axotomy serves as a stimulus for microglial responsiveness, the rapid axolemmal closure associated with DBI-mediated perisomatic TAI and its attendant axotomy (17) most likely does not provide the stimulus needed for chemotactic activation. In the current study, the subtle interstitial albumin immunoreactivity at 6 hours postinjury, combined with the absence of overt neuronal/tissue damage, supports a scenario in which blood-borne signaling molecule concentrations may be sufficient to achieve microglial activation yet insufficient to stimulate migration and/or projection formation, at least in the early stages after injury. Our finding of microglia within adjoining, non-TAI-containing regions that are exposed to albumin yet fail to demonstrate an activated phenotype also suggests a specific role for TAI pathogenesis in microglial activation although the precise mechanism(s) of activation remains unclear.
In addition to microglial activation within injured loci, immunophenotypic macrophages were also found within neocortical and hippocampal parenchyma at 6 hours postinjury, with their recognition in the thalamus by 48 h. In these loci, the macrophages persisted until at least 28 days postinjury. Csuka et al (8), using an impact-acceleration model of DBI that did not directly relate immune cell interactions with sites of axonal injury, described macrophage immunoreactivity within the meninges and perivascular compartments, suggesting minimal postinjury parenchymal distribution. In the current study, immunophenotypic macrophages were distributed within the parenchyma of the diffusely injured loci. For example, macrophages either dual-labeled for lectin-CD68 or single-labeled for CD68 were observed within superficial and deep layers of the neocortex, a finding consistent with previous observations of macrophage localization after lateral fluid percussion-induced mild brain injury (9). Dual-labeled lectin-CD68 immunoreactivity at relatively early postinjury time points demonstrates activated microglia that have adopted a macrophage phenotype (32), whereas the finding of single-labeled, CD68-positive macrophages primarily adjacent to perivascular locales suggests the potential for coordinated movement through an injury-induced, transiently perturbed BBB (33, 34) as suggested by postinjury interstitial albumin immunoreactivity (35-38).
Although numerous parenchymal immunophenotypic macrophages were observed in isolation, others approximated somata of perisomatically axotomized neurons, particularly within the neocortex by 48 hours postinjury. EM evaluation of macrophage-soma interactions revealed macrophages encompassing large portions of the somatic membrane, potentially participating in somatic bouton disruption and loss. This phenomenon was reminiscent of an inflammatory-mediated deafferentation of synaptic terminals after primary axotomy, which confers an adaptive advantage for potential synaptic reorganization, known as synaptic stripping (39). To date, however, only microglia have been implicated in separating boutons from neuronal cell bodies (40-42). Given the suggestion of somatic reorganization and repair after DBI-mediated axotomy (16), macrophage-somatic associations seem plausible; however, immunophenotypic macrophages, not microglia, were recognized approximating somatic membranes after DBI. This difference in immune cell specificity may reflect different neuronal mechanisms of immune cell signaling after trauma given the primary versus secondary nature of axotomy. Perhaps neuronal mechanisms stimulated by primary axotomy favor immediate microglial responses, whereas delayed mechanisms after secondary axotomy allow for differentiation of perineuronal microglia that approximate somata under nonpathologic conditions (43) to a macrophage phenotype.
After TBI, immune cell activation is driven primarily by neuronal degeneration (27, 44). However, the lack of DBI-mediated cell death in the cFPI model (16) suggests that ongoing microglial/macrophage activation (up to 28 days in the current communication) is supported by DBI-mediated secondary pathologies such as cellular membrane perturbation and/or Wallerian degeneration leading to synaptic disruption. Membrane perturbation may stimulate cytokine/chemokine release by injured neurons and/or surrounding glia (45, 46). Several lines of evidence supporting microglia-neuron cross-talk are provided by both in vivo and in vitro studies (47, 48). The release of proinflammatory cytokines, including interleukin-1, interleukin-6, and tumor necrosis factor-α, by activated microglia after trauma has been demonstrated (49-51), whereas injury-mediated neuronal release of ATP has been shown to induce microglial activation and chemotaxis (52, 53). The postinjury release of these molecules may then contribute to sustained immune cell activation long after the initial injury stimulus has dissipated (54-56). Similarly, Wallerian change and related deafferentation are known stimuli for immune cell activation. Given the rapid Wallerian degeneration that takes place after DBI-mediated perisomatic TAI (17), the presence of axonal breakdown products may also contribute to prolonged activation.
Lastly, the findings in the current study may have implications for neuropathologic evaluation of TBI. Microglial clustering in postmortem histologic neuropathology is currently used as a diagnostic indicator of TBI-related pathobiology that includes axotomy (57). These clusters or nodules have been documented with survival times of at least 7 weeks postinjury (58), with more recent evidence for clustering between 10 and 15 days postinjury (7, 59). In the current study, only isolated microglial aggregations occurred at 28 days postinjury. Further, the LM morphology of these aggregations was dissimilar from that of aggregations and clusters/nodules involved in microglial-mediated neuronophagia. Given the absence of focal pathology within DBI-mediated TAI-containing loci, previous descriptions of microglial clusters may be indicative of nonrecognized focal pathology, wherein necrotic cell death and/or hemorrhagic tissue damage triggers microglial recruitment via chemotactic signaling. Our findings suggest that histopathologic and forensic identifications of TBI that rely on surrogate markers of neuronal injury, such as microglial clustering, may overlook and thereby underestimate more subtle forms of diffuse pathology wherein neuronal injury does not elicit more focal neuroinflammatory responses.
In sum, the current study builds on previous studies to characterize DBI-mediated neuroinflammatory responses. An early, yet initially nonspecific, and persistent immune cell response within diffusely injured brain loci marked by TAI is observed with eventual immune cell recognition of TAI pathobiology. Importantly, these findings illustrate previously unrecognized posttraumatic cellular associations that suggest mechanisms of injury-induced immune cell-neuronal cross-talk. This study illustrates the complexity of neuroinflammatory responses to diffuse TBI and provides the impetus for future studies of cross-talk mechanisms and their relationship to potential neuronal recovery after trauma.
The authors thank Susan Walker, Lynn Davis, Judy Williamson, and Thomas Colburn for their excellent technical assistance as well as Dr. Scott Henderson for his help with confocal microscopy. Additionally, the authors acknowledge Drs. W. Shawn Carbonell and James Mandell from the University of Virginia Department of Neuropathology for their advice and technical guidance related to various microglial immunocytochemical procedures.
This work was supported by National Institutes of Health (NIH) grants NS045824, T32NS007288, and HD055813. Additional funding came from the Virginia Commonwealth University (VCU) School of Medicine MD/PhD Program (B.J.K.). Microscopy was performed at the VCU Department of Anatomy and Neurobiology Microscopy Facility, supported, in part, with funding from NIH National Institute of Neurological Diseases and Stroke Center Core Grant 5P30-NS047463.
. Is traumatic axonal injury (AI) associated with an early microglial activation? Application of a double-labeling technique for simultaneous detection of microglia and AI. Acta Neuropathol (Berl) 1999;97:491–94