Brain Advance Access originally published online on October 28, 2006
Brain 2006 129(12):3249-3269; doi:10.1093/brain/awl296
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The cellular inflammatory response in human spinal cords after injury
1 BioTherapeutics Research Group, Robarts Research Institute London, Ontario, Canada 2 Graduate Program in Neuroscience, London Health Sciences Centre and University of Western Ontario London, Ontario, Canada 3 Departments of Pathology and Clinical Neurological Sciences, London Health Sciences Centre and University of Western Ontario London, Ontario, Canada 4 The Miami Project to Cure Paralysis, The University of Miami Miller School of Medicine and Miami Veterans Affairs Medical Center Miami, FL, USA 5 Department of Pathology, The University of Miami Miller School of Medicine and Miami Veterans Affairs Medical Center Miami, FL, USA
Correspondence to: Dr Lynne C. Weaver, Spinal Cord Injury Laboratory, BioTherapeutics Research Group, Robarts Research Institute, 100 Perth Drive, PO Box 5015, London, Ontario, Canada N6A 5K8 E-mail: lcweaver{at}robarts.ca
 |
Summary |
|---|
 Top Summary IntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Spinal cord injury (SCI) provokes an inflammatory response that generates substantial secondary damage within the cord but also may contribute to its repair. Anti-inflammatory treatment of human SCI and its timing must be based on knowledge of the types of cells participating in the inflammatory response, the time after injury when they appear and then decrease in number, and the nature of their actions. Using post-mortem spinal cords, we evaluated the time course and distribution of pathological change, infiltrating neutrophils, monocytes/macrophages and lymphocytes, and microglial activation in injured spinal cords from patients who were ‘dead at the scene’ or who survived for intervals up to 1 year after SCI. SCI caused zones of pathological change, including areas of inflammation and necrosis in the acute cases, and cystic cavities with longer survival (Zone 1), mantles of less severe change, including axonal swellings, inflammation and Wallerian degeneration (Zone 2) and histologically intact areas (Zone 3). Zone 1 areas increased in size with time after injury whereas the overall injury (size of the Zones 1 and 2 combined) remained relatively constant from the time (1–3 days) when damage was first visible. The distribution of inflammatory cells correlated well with the location of Zone 1, and sometimes of Zone 2. Neutrophils, visualized by their expression of human neutrophil
-defensins (defensin), entered the spinal cord by haemorrhage or extravasation, were most numerous 1–3 days after SCI, and were detectable for up to 10 days after SCI. Significant numbers of activated CD68-immunoreactive ramified microglia and a few monocytes/macrophages were in injured tissue within 1–3 days of SCI. Activated microglia, a few monocytes/macrophages and numerous phagocytic macrophages were present for weeks to months after SCI. A few CD8+ lymphocytes were in the injured cords throughout the sampling intervals. Expression by the inflammatory cells of the oxidative enzymes myeloperoxidase (MPO) and nicotinamide adenine dinucleotide phosphate oxidase (gp91phox), and of the pro-inflammatory matrix metalloproteinase (MMP)-9, was analysed to determine their potential to cause oxidative and proteolytic damage. Oxidative activity, inferred from MPO and gp91phox immunoreactivity, was primarily associated with neutrophils and activated microglia. Phagocytic macrophages had weak or no expression of MPO or gp91phox. Only neutrophils expressed MMP-9. These data indicate that potentially destructive neutrophils and activated microglia, replete with oxidative and proteolytic enzymes, appear within the first few days of SCI, suggesting that anti-inflammatory ‘neuroprotective’ strategies should be directed at preventing early neutrophil influx and modifying microglial activation.
Key Words: inflammation; macrophage; neutrophil; oxidative activity; SCI
Abbreviations: ß-APP, ß-amyloid precursor protein; H & E, haematoxylin and eosin; H & E/LFB, combined H & E and Luxol fast blue; MMP, matrix metalloproteinase; MPO, myeloperoxidase; NADPH, nicotinamide adenine dinucleotide phosphate; SCI, spinal cord injury
Received January 9, 2006. Revised September 15, 2006. Accepted September 19, 2006.
|
Introduction |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Injury to the spinal cord provokes an inflammatory reaction that initially results in further tissue damage. Attenuation of the early inflammatory response to spinal cord injury (SCI) may therefore limit the extent of tissue injury and, accordingly, the consequent disability (Hall, 1992
; Bethea, 2000; Gris et al., 2004). Hence, the development of effective acute therapies for SCI in humans requires a precise knowledge of the time course of entry of various inflammatory cells into the injured human spinal cord and activation of resident spinal cord cells such as microglia, and of their many actions, destructive or beneficial, that affect neural tissue.
The mechanisms and time course of inflammation are well documented in animal models of SCI (Popovich et al., 1997; Schnell et al., 1999; Hausmann, 2003; Sroga et al., 2003). In rats, neutrophils appear at the primary lesion site 4–6 h after injury, peaking in number at 12–24 h and disappearing within 
5 days (Taoka et al., 1997; Carlson et al., 1998). Oxidative and proteolytic enzymes produced by neutrophils sterilize the damaged area and prepare it for subsequent ‘repair’, but overwhelming numbers of neutrophils can cause ‘by-stander’ tissue damage (Taoka et al., 1997). Macrophages in the injured spinal cord are derived from blood-borne monocytes and resident microglia (see Popovich et al., 1999). Blood-borne monocyte/macrophages infiltrate the lesion 2 days after SCI in rats, achieve their highest density at 5–7 days, and persist for weeks to months (Popovich et al., 1997; Carlson et al., 1998). Microglia become activated within minutes to hours after SCI and are transformed into macrophages (Popovich et al., 2002). Macrophages in the injured rodent cord contribute to by-stander damage by releasing pro-inflammatory cytokines, reactive oxygen species, nitric oxide and proteases (Popovich et al., 1999; Popovich et al., 2002). In contrast to these destructive effects, they also participate in the removal of injured tissue debris and the release of protective cytokines that promote neuronal regeneration, wound healing and tissue repair (Rabchevsky and Streit, 1998; Schwartz, 2003). In animals, T-lymphocytes enter the injured spinal cord at different times, depending upon species and strain of animal (Sroga et al., 2003). T-lymphocytes are responsible for cell-mediated adaptive immunity (Friese and Fugger, 2005; Jones et al., 2005). Whether T-lymphocytes cause secondary degeneration or mediate wound repair after SCI remains highly controversial (Hauben and Schwartz, 2003; Friese and Fugger, 2005; Jones et al., 2005; Crutcher et al., 2006).
The secondary spinal cord damage caused by neutrophils and macrophages in animal studies is caused in part by oxidative and proteolytic enzymes. Myeloperoxidase (MPO), a well-known oxidative enzyme, is expressed abundantly by neutrophils and other phagocytes (Taoka et al., 1997; Bao et al., 2004) and generates hypochlorous acid that kills pathogens, but also damages nearby tissue. Another damaging oxidative enzyme produced by inflammatory cells is nicotinamide adenine dinucleotide phosphate (NADPH) oxidase that generates the superoxide anion (Vaziri et al., 2004; Brandes and Kreuzer, 2005). The activity of this enzyme requires the catalytic subunit gp91phox on the cell membrane. gp91phox serves as an excellent marker of oxidative activity in the rat spinal cord (Bao et al., 2005). Inflammatory cells also release matrix metalloproteinases (MMPs), especially MMP-9, to permit them to penetrate the blood–CNS-barrier (Noble et al., 2002). MMP-9 is upregulated in leucocytes entering the spinal cord, directly facilitating their extravasation, and promoting the tissue damage that they cause (Mun-Bryce and Rosenberg, 1998; Wang et al., 2000; Noble et al., 2002).
In contrast to the detailed information available from experimental animals, only a few studies of the inflammatory response to human SCI have been conducted and the very acute period after injury has rarely been assessed (Yang et al., 2004; Chang, 2006). Accordingly, we used immunohistochemical methods to evaluate the cellular inflammatory response to human SCI by determining the time course and distribution of neutrophil, mononuclear phagocyte and T-lymphocyte infiltration, as well as, microglial activation in post-mortem spinal cords with injury-to-death (‘survival’) intervals ranging from minimal [i.e. victims who ‘die at the scene’ (DAS)] to 1 year. Furthermore, we have assessed the potential of these inflammatory cells to contribute to secondary damage by investigating their expression of markers of oxidative stress, including the oxidative enzymes MPO and the catalytic subunit (gp91phox) of NADPH oxidase and of the pro-inflammatory protease MMP-9.
|
Material and methods |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The Miami Project to Cure Paralysis provided spinal cords from 17 cases of SCI (2 females and 15 males with ages ranging from 6 to 88 years) and from 4 control cases (4 males, 21–86 years) who had sustained vertebral fractures and/or traumatic head injuries but had no structural evidence of SCI. The hospitals of the London Health Science Centre provided injured spinal cords from 11 cases (6 females, 5 males, 13–82 years). More than one spinal cord segment (i.e. cervical and thoracic) was used from each control case. Survival times after injury ranged from minimal (i.e. patients who died at the scene of the accident) to 1 year (Table 1). The circumstances of SCI, patient demographic information and details of the general and neuropathological post-mortem assessment for each case were obtained from autopsy reports. Patients with infectious and inflammatory complications not directly associated with trauma, such as sepsis, were excluded from the study because the effect of concurrent systemic inflammation on the intraspinal inflammatory response is unknown. The spinal cord injuries were classified on the basis of their histological appearance as ‘contusion/cyst’ (n = 17), ‘massive compression’ (n = 7) or ‘laceration’ (n = 4) (Bunge et al., 1993
; Shkrum and Ramsay, 2006). Contusional injuries were characterized by an intact pia and relative preservation of the anatomical relations of various elements of the spinal cord, and variable degrees of injury ranging from involvement of the entire cross-sectional area to large usually asymmetric areas of tissue damage. In older injuries, cyst-like areas were present. Massive compression injuries were characterized by disruption of the pia and severe distortion and disruption of spinal cord parenchyma. Laceration injuries, which by definition were perforating or penetrating injuries caused by weapons or projectiles, were associated with breaching of the pia and linear tearing of the cord tissue.
|
In all cases, tissue samples from the centre of SCI and at various distances above and below the injury were obtained. The data from tissue from the centre of the lesion were used to compare the inflammatory responses between cases whereas those from the remote, uninjured segments of the spinal cord served as within-case controls. Between-case comparison of the remote samples was not possible because, for different cases, the distance of these samples from the lesion centre was variable. All tissue samples had been removed within 24 h of death and fixed in neutral buffered formalin for several weeks. Blocks from the spinal cords were dehydrated, embedded in paraffin wax, cut into sets of 6 µm thick sections and placed on positively charged glass slides. One set of sections was stained with haematoxylin-eosin (H&E), and the remaining sets were used for immunohistochemistry. A subset of sections from four cords within each injury-to-death interval was stained with H & E/Luxol fast blue (H&E/LFB). A second subset of these sections was stained with Bielschowsky's silver stain.
Immunohistochemistry
The antibacterial protein family -defensins-1-3 (subsequently termed ‘defensin’) was used as a highly selective marker of neutrophils (Schluesener and Meyermann, 1995; Barnathan et al., 1997; Agerberth et al., 2000) and was labelled by anti-NCL-Defensin antibody (HNP 1–3) (1:1000, Novocastra Laboratories Ltd., Newcastle, UK). The remaining cells and enzymes were identified by the following antibodies: anti-ß-amyloid precursor protein (ß-APP) (1:4000, Chemicon International), labelling damaged axons; anti-PG-M1 (1:1000, Dako, Glostrup, Denmark), directed against CD68, a lysosomal protein expressed by phagocytic macrophages of microglial and monocytic origin (Greaves et al., 1998; van den Berg et al., 2001); anti-CD8 (1:100, Dako, Glostrup, Denmark), labelling cytotoxic T and natural killer cells; anti-CD4 (1:100, Novocastra Laboratories Ltd., Newcastle, UK), labelling helper/regulator T cells; anti-CD20cy (1:100, Dako, Glostrup, Denmark), labelling B cells; anti-MPO (1:10 000, Dako, Glostrup, Denmark) recognizing MPO; anti-gp91phox, labelling the active conformation of NADPH oxidase (1:500, Upstate Biotechnology, Lake Placid, NY, USA); and anti-MMP-9 (1:1000, Chemicon International, Temecula, CA, USA), labelling inactive and activated forms of MMP-9.
The sections used for immunohistochemistry were dried at 37°C overnight, deparaffinized in xylene, and rehydrated through a series of graded ethyl alcohols. Endogenous peroxidase was inactivated by treatment with 3% H2O2 in 100% methanol for 5 min. Following a 5 min wash in phosphate-buffered saline (PBS, pH 7.4), microwave heat-induced antigen retrieval in 10 mmol/l citrate buffer, pH 6.0, was carried out. Non-specific staining was blocked by incubating the sections in 10% normal horse serum for 1 h at room temperature. Afterwards, the sections were incubated with the various primary antibodies at room temperature for 48 h in 2% normal horse serum in a humidified chamber. Sections were then washed in PBS and incubated overnight in either a donkey anti-mouse secondary antibody conjugated to biotin or a donkey anti-rabbit secondary antibody conjugated to biotin (both at 1:1000, both from Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA) in 2% normal horse serum. Next, sections were washed in PBS and incubated with extravidin peroxidase (1:1500, Sigma, St Louis, MO, USA) in 2% normal horse serum for 4 h. Tissues were then incubated for 7 min with the chromogen diaminobenzidine (DAB) and glucose oxidase (both from Sigma, St Louis, MO, USA) for visualization of antibody binding. Following PBS washes, the sections were counterstained with Gill III haematoxylin, dehydrated through ascending ethyl alcohols, cleared in xylene, and cover-slipped with DPX mountant. Sections from surgical specimens of acutely inflamed appendices (in which numerous neutrophils and abundant lymphoid follicles are present) were exposed to each of the primary antibodies, to serve as positive controls for immunolabelling of neutrophils, monocytes/macrophages and lymphocytes. Negative controls included sections in which the primary antibody was omitted, and sections incubated with isotype-matched antibodies (1:100–1:10 000, IgG). These positive and negative controls were processed with every batch of immunohistochemical slides.
Quantification of the area of necrosis at the lesion epicentre
The cord injury was assessed microscopically, using brightfield optics, by examining one H&E- or H&E/LFB-stained section from the lesion centre of each case or from cervical, thoracic and lumbar sections from control cases. The entire cross-sectional area of the section was digitized using a Retiga 1300 videocamera (Quantitative Imaging Corporation, Burnaby, BC, Canada), and ImagePro Plus software (Media Cybernetics, Silver Spring, MD, USA) as described below. Using the quantification software, calibrated as described below, the area of necrosis (areas in which the normal staining intensity of the tissues and the capacity to distinguish histologically between various cytological elements were lost), in the short survival cases, and the cystic regions, in the longer survival cases, in each photomontage were selected by the experimenter. These areas will be defined as Zone 1 in Results. The cumulative area occupied by the pixels in the area selected was determined and expressed as a percentage of the ‘area of interest’ (AOI), which in this case, was the entire cross-sectional area of the spinal cord, excluding the pia.
Semi-quantitative assessment of overall injury
To assess the overall injury, each spinal cord cross-section was divided into eight distinct areas (left and right ventral horns and left and right dorsal, ventral and lateral funiculi—the posterior horns are too narrow to allow reliable assessment) and the overall injury in each of these areas was scored as: 0 = no injury; 1 = trace injury; 2 = slight injury (injury occupies one-quarter of the region); 3 = moderate injury (injury occupies half of the region); 4 = moderately severe injury (injury occupies three-quarters of the region); and 5 = severe injury (injury occupies all of the region). The presence of scant petechiae defined ‘trace’ injury whereas ‘injury’ in all other categories included, in various combinations, extensive bleeding, necrosis (identified by tissue pallor, loss of cellular architecture and cystic cavities), ‘oedema’ (characterized by tissue microvacuolation) and axonal swelling. The overall injury score is a summation of the pathological changes that will be defined as Zones 1 and 2 in Results. This assessment of injury did not show a predilection in terms of severity or location for any of the eight grey or white matter regions. Therefore the average score for the grey matter regions and for the white matter regions was calculated.
Quantification of neutrophil (defensin), microglia/macrophage (CD68) and MPO immunolabelling
A section of the spinal cord at the centre of the injury of each case was viewed through a x4 objective. Digital photographs of 20–24 individual fields of view, covering the entire cross-sectional area of the spinal cord, were recorded digitally using a Retiga 1300 videocamera (Quantitative Imaging Corporation, Burnaby, BC, Canada), and ImagePro Plus software (Media Cybernetics, Silver Spring, MD, USA). The individual images were then ‘stitched’ together electronically by ImagePro Plus (Fig. 1A–C). Afterwards, using the quantification software, the area represented by one pixel was calibrated in µm2 for the x4 objective and a colour-based threshold detection function was applied to each photomontage by the experimenter to select the brown intracellular DAB staining, excluding the blue haematoxylin-stained nuclei and any extracellular staining or background (Fig. 1B–E). The cumulative area occupied by these pixels was determined and expressed as a percentage of the entire cross-sectional area of the spinal cord, excluding the pia (AOI). This method was used to quantify separately the inflammatory response attributable to neutrophils or microglia/macrophages. The area of immunoreactivity assessed by this method gives an overall estimate of the ‘inflammatory response’ associated with a given cell type (Popovich et al., 1997; Saville et al., 2004). The extent of immunoreactivity reflects the number of cells involved in the inflammatory response and their size. Cell size is likely to be a significant part of the measured macrophage and microglia response, since activation of these cells causes them to enlarge, but, in the case of neutrophils, the area of immunoreactivity will be more closely correlated with the numbers of neutrophils because these cells shrink as they involute. We chose this method instead of cell counting because neutrophils, microglia and macrophages may be impossible to distinguish on cytological criteria when the entire cell (such as a ramified microglia) is not apparent in a section and or when cells cluster. The areas of defensin immunoreactivity in grey and white matter were compared to assess any differences between these areas in the neutrophil inflammatory response. In an attempt to confirm the relative specificity of defensin and MPO immunolabelling for neutrophils, areas of MPO immunolabelling were measured in four cases selected from the 1–3 day interval and in four cases from the 5–10 day interval and compared with corresponding areas occupied by defensin-ir cells (i.e. neutrophils) or CD68-ir cells (i.e. monocytes/macrophages; Table 3).
|
Quantification of CD8+cells
The relative abundance of CD8+cells in these cases could not be assessed by measuring areas of immunoreactivity due to the paucity of these cells in the spinal cords. Instead, they were counted in four randomly selected high power (x40) fields of view and an average number per field was generated for each case.
Morphological characterization of microglia and macrophages and expression of enzymes
Phagocytic microglia and monocyte/macrophages were immunoreactive for CD68. CD68-labelled microglia with discrete thin processes were termed ‘ramified microglia’ and were considered to be microglia in the earliest state of activation. ‘Activated microglia’ refers to those with hypertrophic cell bodies and shorter, thicker cell processes, indicating transformation into macrophages. Macrophages were defined as large rounded cells (>20 µm in diameter) with large clear cytoplasmic vacuoles and inclusions that indicated ongoing or previous phagocytic activity. Currently, no available techniques can distinguish between phagocytic macrophages derived from microglia or from haematogenous monocyte/macrophages.
The expression of MPO and gp91phox, markers of oxidative reactivity, and MMP-9, a pro-inflammatory protease, was examined at each of the time intervals after injury (intervals are defined in Results). We used cellular morphology and serial sections stained for defensin and CD68 in each case to identify the cell populations with immunoreactivity for MPO, gp91phox and MMP-9. Despite the clustering of cells, a sufficient number were separate enough to permit their morphology to be examined. When only a fraction of a population of a specific cell type expressed a given protein, the percentage of cells expressing it in five x40 fields was determined in each spinal cord. When ramified microglia were in a ‘resting state’, they did not express CD68 and were not readily visible. Therefore, we had no measure of the entire population of ramified microglia from which to determine the proportion expressing proteins such as MPO or gp91phox. We instead described the approximate numbers of MPO- or gp91phox-ir cells, with morphology of ramified microglia, found in x40 microscopic fields.
Statistical analysis
Data were subjected to parametric statistical analysis using completely randomized one-way ANOVA (Snedecor and Cochran, 1989). Fisher's LSD protected t-test was used to determine differences between mean values. The probability value required to attain statistical significance was P < 0.05. In one analysis a one-tailed Student's t-test was used to determine the difference between mean values (P < 0.05). Spearman's correlation analysis was used to determine the relationship between time after injury and the relative area of necrosis at the lesion epicentre (P < 0.05).
|
Results |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
These data are organized by inflammatory cell type and the cases are categorized into the following five groups: no SCI and 0–4 h, 1–3 days, 5–10 days, and weeks to months, up to 1 year after injury.
Zonal distribution of injury
SCI resulted in a zonal distribution of pathological changes at the lesion site in cords at intervals >4 h after injury (Fig. 2A). The relative sizes of these zones varied substantially from case to case. An area of intense tissue injury, either in the form of necrosis (in the shorter survival cases; Fig. 2B and C left) or as cystic change (in the longer surviving cases; Fig. 2D–F left), was often, but not always, centred on the grey matter (Zone 1, outlined in green on Fig. 2B–F left). An exception to this general pattern is the cord at 2 weeks that had selective necrosis of the white matter (Fig. 2D). The areas of necrosis and cystic cavities contained, or were surrounded by, inflammatory cells such as neutrophils, labelled by their expression of defensin, (Fig. 2B right) and macrophages, labelled by their expression of CD68, (Fig. 2C–F right). Areas of incomplete injury (Zone 2) were located around the perimeter of Zone 1 and in more widely but haphazardly distributed locations outside and mutually exclusive of Zone 1. The changes in Zone 2 included axonal injury, Wallerian degeneration and inflammatory cells (neutrophils and activated microglia) in the cases that survived 1–3 or 5–10 days after injury, and residual axonal degeneration, progressive gliosis and variable numbers of macrophages in cases that survived weeks to months after injury. Small regions of apparently intact grey matter and white matter (Zone 3) lay adjacent to areas containing Zone 2 changes. In the very acutely (0–4 h) injured spinal cords, the patterns of zone formation had not developed.
|
Histopathological characterization
In the uninjured control cases, no histopathological abnormalities were identified. Similarly, no histopathological abnormalities were identified in the DAS cases, with the exception of rare petechiae in the grey matter.
The spinal cords from cases that died 1–4 h after injury contained various degrees of tissue fragmentation, petechiae (particularly in the grey matter) (Fig. 2G), a haphazard, swirling malorientation of the fibre tracts in the funiculi (Fig. 2H versus I) and no abnormalities of ß-APP immunolabelling (data not shown). The true extent of injury or areas where necrosis would have appeared, had the individual survived, could not be predicted from the distribution of abnormalities in this tissue.
Malorientation of the fibres in the funiculi and fragmentation of the tissue were still evident 1–3 days after SCI (Fig. 2J). Petechiae were more prominent and often confluent. Necrosis (loss of staining intensity and cytological definition, as defined in Material and methods) was established usually in one confluent area, although some spinal cords had separate areas of necrosis. The necrotic areas tended to be haemorrhagic (Fig. 2J) and they were pronounced in, and sometimes limited to, grey matter (Fig. 2K). These areas of necrosis are defined as Zone 1 (Fig. 2A). Neutrophils were evident around the margins of the necrotic areas and, in one case, these inflammatory cells were associated with a distinct loosening of tissue texture (Fig. 2L). In an otherwise intact adjacent grey matter, changes in the anterior horn cells, included lysis of the Nissl substance, intensification of cytoplasmic eosinophilia, perineuronal vacuolation and karyorhexis or vesicular nuclear change, representing ischaemic neuronal necrosis, chromatolysis or both (Zone 2 change, see Fig. 2A). Focal white matter vacuolation and axonal swelling in various degrees were evident at the margins of the necrosis and scattered haphazardly through the white matter (Fig. 2M). Axonal swellings were incompletely filled with argyrophilic material (Fig. 2N), and solitary or discrete groups of axonal swellings were strongly ß-APP-immunopositive (Fig. 2O). These are examples of the types of pathological changes defined as Zone 2.
The distribution of the pathological changes at 5–10 days after injury resembled the distribution in cords at 1–3 days. However, at 5–10 days the necrotic areas contained abundant macrophages (Fig. 2C–F) and the adjacent, better preserved tissues (Zone 2), contained numerous and widely distributed axonal swellings with strong argyrophilia and a variable intensity of ß-APP-immunopositivity (data not shown). The intensity of myelin staining was proportional to the argyrophilic axonal labelling, and provided no qualitative evidence of demyelination (data not shown). However, demyelination was difficult to detect against the background of the varied axonal and myelin changes that were present in this time interval and, in the limited numbers of sections available for this study, demyelination could not be confirmed or excluded. The anterior horn cells adjacent to areas of necrosis resembled those in the 1–3 days time period (Fig. 2P).
In spinal cords at weeks to months after injury, areas of necrosis contained abundant macrophages or, in the longer survival cases, various degrees of cyst formation occurred (Fig. 2D–F). The distribution of these cysts was reminiscent of the distribution of necrosis in cases with shorter survival (1–10 days). In one case of extensive necrosis, a central area of ‘mummified’ necrosis (Zone 1) was surrounded by a rich mantle of macrophages in Zone 2 (Fig. 2E). The better preserved areas of the spinal cord (Zone 2) in the cases surviving for a few weeks contained macrophages, plump reactive gemistocytic astrocytes and numerous ß-APP-immunopositive and argyrophilic axonal swellings (not shown). Many of the axonal swellings in these cases were ß-APP-immunonegative. Fine gliovascular septa bridged the cystic spaces and a fine, dense network of fibrillary astrocytic processes surrounded the cystic areas in cases surviving for several months (data not shown). In one case with extensive spinal cord cystic change, the cystic area contained numerous peripheral nerve bundles and mild pial and perivascular fibrosis. Some gradation in intensity of myelin staining occurred at margins of the necrotic areas (Fig. 2Q), ranging from light, close to the lesion, to normal, approximately a millimetre away, but, in these areas, the numbers of argyrophilic axons were also proportionally graded with distance from the lesion (Fig. 2R).
The areas of necrosis and/or cysts (Zone 1) within lesion epicentres (expressed as % of total section area) were measured for each case (Fig. 2S) and plotted against the injury-to-death intervals for the control group (n = 3) and injury groups (n = 22). Analysis was performed on H & E or H & E/LFB-stained sections. The relative size of Zone 1 in the spinal cords increased significantly with time after injury (Spearman correlation, r = 0.7781, P < 0.0001).
The semi-quantitative assessment of overall tissue damage in Zones 1 and 2 combined (overall injury, see Material and methods) in grey and white matter, revealed no significant histopathological damage at 0–4 h after SCI when the data were compared with those of the uninjured cords (Fig. 2T). The severity of histologically visible overall injury increased significantly in the 1–3-day survival group, in comparison with the control and 0–4 h survival interval groups and then remained constant as survival times increased (Fig. 2T). Although the overall injury scores included areas of necrosis and cavity formation (Zone 1), this score also addressed pathological changes outside these areas (Zone 2). For this reason, and particularly because of the widespread bleeding and oedema in the cords at early time intervals, the overall injury scores were inflated at the early intervals (0–4 h and 1–3 days). As time progressed, the overall injury was more dominated by necrosis and cavity formation, and less by the changes in Zone 2, as haemorrhage and oedema resolved. Alternatively, but less likely, the increase in areas classified as Zone 1 may be spurious because, as cysts form, the areas that once were necrotic become more easily recognized. The changing contributions of Zones 1 and 2 summed to cause a relatively constant overall state of injury, in comparison to the increasing areas of necrosis and cavity formation.
With the exception of four cases with ‘laceration’ injuries, all cases were examples of ‘contusion/cyst’ or ‘massive compression’ injuries (Table 1). The necrosis/cavity formation and severity of SCI in the laceration cases resembled that of the compression injuries with the exception of the pial breaches, and, therefore, these cases were included in both analyses.
Neutrophils enter the lesion site within hours to a few days after injury
Neutrophils in blood vessels and in relatively intact tissue within the injured spinal cord had intracellular immunoreactivity for defensin. These cells also had striking extracellular halos of immunoreactive product, because defensin is a secreted protein (Schluesener and Meyermann, 1995; Barnathan et al., 1997). In the control cases and in uninjured cord segments of SCI, defensin-immunoreactive (-ir) neutrophils were all intravascular and had extracellular halos of defensin immunoproduct (Fig. 3A). At the earliest time interval after SCI (0–4 h), neutrophils were endovascular or were found in petechial haemorrhages in the injured tissue (Fig. 3B). Extravasation of neutrophils from small blood vessels was first observed at 3–4 h after injury. This change in location of the cells (i.e. in haemorrhages or in the process of migrating, or having migrated, through vessel walls) was not yet associated with an increase in ‘free’ extravascular neutrophils in the spinal cord parenchyma. At 1–3 days after SCI, neutrophil infiltration was striking and diffuse throughout grey and white matter, and numerous neutrophils were located in and around central grey haemorrhagic, necrotic areas (Fig. 3C, D and G). Neutrophils in these regions at 5–10 days after injury were smaller than the extravascular neutrophils at 1–3 days after SCI (Fig. 3E versus C), and their lobular nuclei were condensed, suggesting apoptosis. These nuclear changes rendered these cells much more difficult to identify in H&E-stained sections. At 1–5 days after SCI, in areas of necrosis and haemorrhage, the neutrophils had markedly reduced or absent extracellular halos of defensin (Fig. 3D and E), perhaps due to depletion of this secreted protein. Weeks to months after SCI, only a few extravascular neutrophils were scattered through the necrotic areas (Fig. 3F). As the injury-to-death interval lengthened, neutrophils were less likely to be within the tissue parenchyma and more likely to be within blood vessel lumina. Of the eight cases in the weeks-to-months category, only those with the longest survival (the case at 7 months and the case at 1 year) had neutrophils exclusively in blood vessel lumina, resembling the distribution of these cells in control cases (Fig. 3A).
|
Neutrophils were no more numerous in cases of laceration than in cases of contusion or massive compression with the same survival interval. Figure 3G (3 weeks) illustrates a spinal cord with a laceration injury that contains only a few neutrophils. Accordingly, data from these cords were included in the quantitative analysis of neutrophil infiltration.
Quantification and time course of neutrophil infiltration in SCI
Uninjured spinal cords and cords at 0–4 h after SCI had only a small amount of defensin immunoreactivity (0.060 ± 0.014 and 0.067 ± 0.028%, respectively; Fig. 3G and H). At 1–3 days following injury, a significant 15-fold increase (compared to the control cases and cases with 0–4 h survival) in the area of defensin immunoreactivity occurred (0.88 ± 0.23%). This increase waned to a still significant 3-fold increase at 5–10 days after SCI (0.17 ± 0.040%). The area of defensin immunoreactivity was not different from controls or cases at the 0–4 h interval at weeks to months after SCI (0.10 ± 0.019%). The defensin immunoreactivity (0.093%) in the case at one year after SCI was the same as that in the control cases. At no time was the neutrophil influx significantly different between grey matter and white matter.
Activation of microglia and infiltration of monocyte/macrophages at the lesion site within days of SCI
In control cases and within 0–4 h of SCI, scant to moderate numbers of CD68-ir microglia were evenly distributed throughout the grey and white matter (Fig. 4A and B). These microglia had particulate cytoplasmic immunolabelling and ramified morphology. The processes of the ramified microglia were more distinct in grey matter than in white matter (Fig. 4A and B), and microglia in white matter tended to be more prominent around blood vessels. In addition, CD68-ir perivascular macrophages were noted (Fig. 4B). At 1–3 days after SCI, microglia entered a transition phase in which their processes had become shorter and thicker and their cell bodies larger, due to activation (Fig. 4C). At this time, the number of activated microglia increased and they were arranged in a patchy distribution around the margins of lesions. However, in areas of preserved white matter, CD68 immunoreactivity appeared to be within ramified microglia, much like that in control spinal cords. At 1–3 days after SCI, small numbers of CD68-ir monocytes, identified by their characteristic nuclear morphology, were found adherent to the endothelium of blood vessels and in adjacent extravascular spaces. These monocytes were likely extravasating into the injury site. A few migrating monocytes were also found in the 5–10 day cases. At 5–10 days (Fig. 4D), in areas of necrosis and lining the inner edges of cavities, CD68-ir cells were numerous and usually rounded with variegated cytoplasmic immunolabelling. Their morphology defined them as phagocytic ‘foamy’ macrophages. Early activated microglia possessing shorter processes and/or rounded cell bodies, were identified in regions adjacent to areas of necrosis (Fig. 4E). However, once the foamy macrophage morphology was established, the microglial or haematogenous origin of the CD68-ir cells could not be determined. In one unique case of central grey necrosis (Fig. 4I: 5d and T8) the entire grey matter was composed of foamy CD68-ir phagocytic macrophages, whereas the white matter contained only ramified microglial with no evidence of such phagocytic macrophages or of microglial activation, and therefore was similar to that of control cords. CD68 immunoreactivity of the phagocytic macrophages varied in intensity: strongly labelled cells tended to be small, with less cytoplasm; weakly labelled cells were large, with foamy cytoplasm. A moderate to marked increase in the number of activated microglia occurred at the margins of SCI and around axonal swellings in grey and white matter. Weeks to 1 year after SCI, the lesion contained many foamy macrophages (Fig. 4F–H). CD68 immunoreactivity, when present, was in these macrophages (Fig. 4F and G) but, at these survival times, most of the cases contained numerous foamy macrophages (in H & E sections) that were not immunoreactive for CD68 (Fig. 4G and H, see inset in G and I ‘2 week C7’). However, in a subpopulation of cases, expression of CD68 could be very great at weeks after SCI as shown in Fig. 4F and I, ‘3 wk C7’). No differences in CD68 immunoreactivity were observed between the laceration, contusion/cyst and massive compression-type injuries at any time interval after SCI and all types were included in the quantitative analysis.
|
Quantitative analysis and time course of microglia/macrophages in the lesion site after SCI
The detectable ramified CD68-ir microglia in uninjured cords and at 0–4 h after SCI occupied a very small percentage of the area of the spinal cord (uninjured: 0.048 ± 0.012% and 0–4 h: 0.10 ± 0.043%, Fig. 4I and J). At 1–3 days following injury, the abundance of ramified microglia and the increased size of activated microglia, as they hypertrophied and became more rounded, contributed to the almost 6-fold significant increase in total area of immunoreactivity (0.28 ± 0.17%), compared to that of the control cords. At 5–10 days and weeks to months after injury, the area occupied by CD68-ir microglia/macrophages was still significantly greater (0.33 ± 0.18% and 0.36 ± 0.11%, respectively) than that in controls and in cords at 0–4 h after injury. The area of CD68-ir remained significantly increased in the weeks-months survival time interval, despite reduced expression of CD68 by the foamy macrophages, because of the abundance of these cells in the lesion (Fig. 4G and H). The weeks-to-months group also contained a subpopulation of cases that had very large areas of CD68 immunoreactivity (2.60 and 14.33%) due to intense CD68 expression in almost all of the macrophages. Data from these cases were not included in the histogram of Fig. 4J, as they were statistical outliers. The single case at 1 year after SCI had many foamy macrophages within the lesion centre but very little expression of CD68 (0.009%, Fig. 4H).
Lymphocytes infiltrate the injured spinal cord
All CD8+ cells were uniform in shape and size and had the characteristic rounded cytological appearance of lymphocytes. In control cases and in cases surviving for 0–4 h and 1–3 days after injury, rare, scattered, solitary CD8+ lymphocytes were located inside blood vessels, in perivascular spaces and in petechial haemorrhages (Fig. 5A and G). Areas of haemorrhage, necrosis and tissue fragmentation, at 1–3 and 5–10 days after injury, contained a few extravascular CD8+ lymphocytes (Fig. 5B and G). As the time after injury increased from weeks to months, CD8+ lymphocytes were encountered more often in some of the cases. In this subset of cases the numbers of CD8+ lymphocytes ranged from a few to groups of 10–20 cells per x40 field (Fig. 5G). These groups were found in perivascular spaces in regions of tissue damage, including areas at the margins of cystic cavities, and usually were randomly distributed among macrophages (Fig. 5C–E). CD4+ lymphocytes followed the same pattern of distribution as CD8+ lymphocytes but were fewer (compare Fig. 5E and F). We could not detect any CD20+ B lymphocytes in the injured spinal cord.
|
Expression of MPO in the lesion centre
In uninjured spinal cords, and at all time points assessed after injury, all neutrophils expressed MPO (Fig. 6A and B). Double labelling was not consistently successful in our study and instead, we used two methods to establish the cellular localization of MPO. First, by examining MPO-stained sections at high power (x40), all stained cells had morphology typical of neutrophils (i.e. round cells of 10–12 µm diameter with lobular nuclei; Table 2). Next, areas of MPO immunoreactivity in a subset of eight cases at the 1–3 and 5–10 day intervals were analysed and compared with areas of defensin and CD68 in serial sections. The areas of immunoreactivity for MPO and defensin were identical or very similar in the five cases in which high expression of defensin revealed a neutrophil influx (Table 3). Some ramified microglia (1–2 per x40 microscopic field) in the uninjured control cords or at 0–4 h after SCI also expressed MPO. At 1–3 and 5–10 days after SCI many ramified and activated microglia (up to 8 per x40 field), close to areas of haemorrhage and tissue damage, expressed MPO (Table 2). Weeks to months after injury, approximately one-quarter of the phagocytic ‘foamy’ macrophages expressed MPO and the intensity of expression was variable (Fig. 6C and D, Table 2). None of the cells identifiable by cytological criteria as ‘foamy’ macrophages or microglia expressed defensin. The comparison of areas of MPO, defensin and CD68 in three cases at 5–10 days after injury suggests that MPO in the cord could be attributed to both cell types at this time (Table 3).
|
|
|
Expression of gp91phox, a marker of oxidative reactivity, in the lesion centre
Immunoreactivity for gp91phox was associated with the plasma membrane of intravascular and perivascular neutrophils at 0–4 h after SCI, but neutrophils did not express this protein in control cases (Fig. 7A, Table 2). At 1–3 days following injury, the majority of neutrophils in and around areas of haemorrhage and necrosis were immunoreactive for gp91phox (Fig. 7B, Table 2). Some neutrophils displayed cytoplasmic as well as membranous labelling whereas others exhibited only membranous labelling. At 5–10 days after SCI, less than half of the neutrophils were gp91phox-ir (Fig. 7C, Table 2). By weeks to months, this proportion became even smaller. Throughout the time course after SCI, scattered endothelial and perivascular cells were immunoreactive for gp91phox (Fig. 7C).
|
In control cords and at all times after injury, ramified microglia near damaged tissue were clearly gp91phox-ir, but the distribution of such labelled cells within grey and white matter, and labelling intensity, varied among cases (Fig. 7A and D, Table 2). At 1–3 and 5–10 days after SCI, some ramified and activated microglia (1–2 per x40 field) in areas adjacent to tissue damage expressed gp91phox (Table 2) At 5–10 days and weeks to months after injury some phagocytic macrophages (2–4 per x40 field) were immunoreactive for gp91phox but the majority did not express this enzyme (Fig. 7C) (Table 2). The gp91phox immunoreactivity of macrophages was pronounced at the plasma membrane with lesser amounts in their cytoplasm.
MMP expression in the lesion centre
At 0–4 h after injury, neutrophils in blood vessels and in perivascular spaces were strongly immunoreactive for MMP-9 (Fig. 7E, Table 2) whereas no expression of this protein was evident in control cases. Almost all neutrophils, identified by their characteristic nuclear morphology, were strongly MMP-9-ir at 1–3 days after SCI (Fig. 7F, Table 2). At 5–10 days after injury, scattered MMP-9-ir neutrophils were present in necrotic areas but overall the proportion of these MMP-9-ir cells had declined (Fig. 7G, Table 2). The few neutrophils, at weeks to months after SCI, did not express MMP-9, like those in control cases (Table 2). In contrast, neither ramified nor activated microglia were MMP-9-ir at any time point assessed after injury. Phagocytic macrophages also did not express MMP-9 (data not shown).
Inflammation and histopathological appearance rostral and caudal to the lesion centre
The foregoing descriptions were concerned with the appearances of the SCI at the lesion centre. In most cases, tissue samples from segments near the lesion were available, but did not permit a systematic analysis of the progression of inflammation away from the lesion epicentre because, for different cases, the distance of these samples from the lesion centre or from its grossly evident margins was variable. A detailed analysis of changes in tissue adjacent to the injury in three cases is provided in Fig. 8 and its legend. In general, in the first week after injury, intravascular and perivascular neutrophils and activated microglia were observed in region 1–3 segments rostral and caudal to the lesion centre, but still in areas of grossly detectable injury (Fig. 8A and B). However, the intensity of inflammation was less than at the lesion centre. The changes in the prevalence and immunohistochemical appearances of neutrophils in tissue at the margins of the injury site followed the same time course as neutrophils in the depths of the lesion site. In and near damaged areas of these segments, CD68-ir activated microglia were prevalent in all cases later than 0–4 h after SCI. Ramified microglia were present in all groups examined. Foamy macrophages were rarely encountered outside the lesion, although abundant in the lesion centre and margins from 5 days to several months after injury (Fig. 8C). The lymphocyte distribution in grossly intact segments near the injury epicentre was not different from that in control spinal cords. The pattern of expression of MPO, gp91phox and MMP-9 by neutrophils, microglia and macrophages at the margins of the lesion was not different from that in the lesion centre although it was less abundant owing to the small number of inflammatory cells outside the lesion epicentre (Fig. 8B). The appearance of spinal cord sections obtained >8 segments away from the lesion resembled those in the uninjured controls (Fig. 8A).
|
In some compression-type SCI, pressure at the site of the SCI forces tissue to extrude above (and, less commonly, below) the injured area in a telescope-like fashion such that tissue damage ends in an adjacent segment, most commonly in the base of the posterior columns, in a small circular area (Ito et al., 1997
). These circular areas of necrosis were encountered in two cases (3 and 5 days post-injury), one of which is illustrated in Fig. 8B. Inflammatory cells in this tissue were neutrophils, activated microglia and macrophages, expressing defensin, CD68, MPO and gp91phox, like those in the ventral white matter area of the same section (Fig. 8B). Highly ramified and activated microglia were located outside and along the rim of the necrotic area (data not shown). As the inflammation waned with distance from the injury epicentre, the pathology in the spinal cords also diminished (Fig. 8C). Tissue at the margins of the lesion was fragmented and contains foamy macrophages and lymphocytes whereas, at short distances from the lesion margins, tissue was essentially normal with only occasional macrophages.
|
Discussion |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Our study has demonstrated that the first cells to participate in the inflammatory response to human SCI, namely neutrophils and microglia, can generate a variety of oxidative and proteolytic enzymes that have the capacity to cause secondary injury by enlarging the lesion and potentially worsening neurological dysfunction. Numerous macrophages are present in the injured cord and appear to be actively phagocytic for weeks to months after the injury. Although they can express the oxidative enzymes, they only do so weakly if at all. The early inflammatory response to SCI in humans is an important target for therapeutic intervention.
In the minutes to hours after traumatic SCI, ischaemia, oxidative damage, oedema and glutamate excitotoxicity all contribute to substantial secondary damage (Blight, 1985; Tator and Fehlings, 1991; Young, 1993; Kwon et al., 2004). The cellular inflammatory events that follow immediately then may play a major role in expansion of the lesion size. Indeed, although we found a correlation between the areas of necrosis and presence of inflammatory cells, these cells often extended into more intact tissue adjacent to the lesion centre. Their impact on this spared tissue is uncertain. The increase in Zone 1 observed in our study may suggest progressive injury of adjacent Zone 2 regions, perhaps related to inflammation, that may even account for some of the ‘secondary’ effects of SCI, especially worsening spasticity and pain.
The first haematogenous inflammatory cell to arrive at a site of injury is the neutrophil. In our study, extravascular neutrophils were in areas of haemorrhage as early as 4 h after injury