The Prevalence and Phenotype of Activated Microglia/Macrophages within the Spinal Cord of the Hyperostotic Mouse (twy/twy) Changes in Response to Chronic Progressive Spinal Cord Compression: Implications for Human Cervical Compressive Myelopathy

Background Cervical compressive myelopathy, e.g. due to spondylosis or ossification of the posterior longitudinal ligament is a common cause of spinal cord dysfunction. Although human pathological studies have reported neuronal loss and demyelination in the chronically compressed spinal cord, little is known about the mechanisms involved. In particular, the neuroinflammatory processes that are thought to underlie the condition are poorly understood. The present study assessed the localized prevalence of activated M1 and M2 microglia/macrophages in twy/twy mice that develop spontaneous cervical spinal cord compression, as a model of human disease. Methods Inflammatory cells and cytokines were assessed in compressed lesions of the spinal cords in 12-, 18- and 24-weeks old twy/twy mice by immunohistochemical, immunoblot and flow cytometric analysis. Computed tomography and standard histology confirmed a progressive spinal cord compression through the spontaneously development of an impinging calcified mass. Results The prevalence of CD11b-positive cells, in the compressed spinal cord increased over time with a concurrent decrease in neurons. The CD11b-positive cell population was initially formed of arginase-1- and CD206-positive M2 microglia/macrophages, which later shifted towards iNOS- and CD16/32-positive M1 microglia/macrophages. There was a transient increase in levels of T helper 2 (Th2) cytokines at 18 weeks, whereas levels of Th1 cytokines as well as brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF) and macrophage antigen (Mac) −2 progressively increased. Conclusions Spinal cord compression was associated with a temporal M2 microglia/macrophage response, which may act as a possible repair or neuroprotective mechanism. However, the persistence of the neural insult also associated with persistent expression of Th1 cytokines and increased prevalence of activated M1 microglia/macrophages, which may lead to neuronal loss and demyelination despite the presence of neurotrophic factors. This understanding of the aetiopathology of chronic spinal cord compression is of importance in the development of new treatment targets in human disease.


Introduction
Mechanical compression of the spinal cord can cause neural tissue damage, reduction of neuronal cell activity and protein synthesis, and neuronal cell death. Cervical compressive myelopathy is characterized by progressive stenosis of the cervical canal and compression of the spinal cord due to spondylosis, degener-ative disc disease, and ossification of the posterior longitudinal ligament (OPLL) [1][2][3][4]. Symptoms usually start to appear after middle age with slowly progressive clumsiness and paresthesia in the hands, gait disturbance, and signs of posterior and pyramidal column dysfunction; eventually leading to tetraplegia or tetraparesis. Human pathological studies have reported flattening of the anterior horns, loss of anterior horn cells, cavity formation, ascending demyelination in the posterior columns, descending demyelination in the lateral columns and proliferation of hyalinized small blood vessels [5][6][7]. However, since it is difficult to properly estimate and follow the progression of these changes in humans, considerable uncertainty exists regarding the biological and molecular mechanisms responsible for the demyelination that takes place in these disorders, and for the progressive loss of neurons and oligodendrocytes. Moreover, although pro-inflammatory cytokines and related immune effector molecules are considered to be produced after chronic, slow compression of the spinal cord as seen in cervical compressive myelopathy, and could result in induction of cell death through necrosis and apoptosis, their role in compression-induced damage remain questionable [8,9].
The inflammatory changes following spinal cord injury (SCI) are complex and involve the activation of resident microglia and recruitment of neutrophils, macrophages and lymphocytes into the lesion from the systemic circulation [10][11][12]. This leads to tissue damage, demyelination and neurological dysfunction [13], as well as apoptosis of neurons and oligodendrocytes [14,15]. Activated microglia and recruited macrophages (which are antigenically not distinguishable, henceforth these cell types have been referred to as microglia/macrophages) are implicated in neuroinflammation through the induction or modulation of a broad spectrum of cellular responses [10]. These cells are the primary source for proinflammatory cytokines within the spinal cord, and their presence and activity have the potential to act as markers of disease onset and prognosis of neurological outcome following SCI [13,16]. Interestingly such neuroinflammation, including the microglial response, has also been identified as a contributor to cell death in ischemic injury in the brain [17] and chronic neurodegenerative disorders [18]. These findings point to the potential commonality of mechanisms underlying cell damage and cell death in both acute neural injury and in slow-developing pathologies of neural systems, like those seen in Alzheimer's disease and Parkinson disease, even in the absence of prominent leukocyte infiltration [18].
Recent studies have demonstrated phenotypic changes in macrophages during the immunological and inflammatory responses to various conditions [19,20]. This divergence is referred to as macrophage polarization and has been reported in nonneural [21] and neural tissues [22,23], and also in both in vitro and in vivo experiments [24]. Approximately, two subtypes of macrophages have become of great interest in the field of spinal cord regeneration: classically activated macrophages (M1 phenotype) and alternatively activated macrophages (M2 phenotype) [25][26][27][28]. Whereas the M1 phenotype is the product of exposure to T helper 1 (Th1) cytokines, such as interferon gamma (IFN-c), tumor necrosis factor-alpha (TNF-a), and interleukin (IL)-6, the M2 phenotype is activated via T helper 2 (Th2) cytokines, such as IL-4, IL-10, and IL-13 [22,29,30]. While the M1 phenotype is known for their high expression of inflammatory cytokines and bactericidal activity, M2 phenotype exhibits enhanced phagocytic and anti-inflammatory properties; although at least three subsets of M2 macrophages have also been documented [21,25]; for such reasons, macrophages with the latter phenotype are considered to function in recovery of SCI [19,23,31]. Thus, modification of the SCI microenvironment to increase the number of M2 macrophages may promote neuroprotection. A similar possibility has also been recently attributed to microglia by showing that these cells can also be induced under certain conditions to both extremes of the M1 and M2 differentiation spectrum [25,32].
The study of the pathological mechanisms of spinal cord dysfunction related to cervical spondylotic myelopathy (CSM) and OPLL has been impaired in the past due to the lack of good in vivo models. However, this has changed with the recent characterization of the tip-toe Walking Yoshimura (twy/twy) mouse; an autosomal recessive mutant. The twy/twy mouse has a spontaneous mutation in the nucleotide pyrophosphatase (Npps) gene developing posterior calcification of the atlantoaxial membrane at the cervical (C) 1-C2 vertebral level. The defective vertebral column causes cervical spinal cord compression progressively over several months. Thus, our group has used the twy/twy mouse as a suitable model to investigate the effects of the chronic, slow compression of the spinal cord that is seen in CSM and OPLL [33][34][35][36].
Recent studies have suggested that neuronal and oligodendrocytic apoptosis through activation of the Fas death receptor pathway is a key event in the twy/twy mouse spinal cord [37] and have shown that neutralization of Fas ligand with a functionblocking antibody reduced neural inflammation at the lesion mediated by activated microglia and macrophages [7]. Another study from our laboratory reported that increased expression of TNF-a and TNF receptor 1 (TNFR1) released by the activated microglia/macrophages correlated with neuronal and oligodendrocytic apoptosis [8]. Based on the above findings, the present study was designed to provide insight on previously unexplored aspects of microglia/macrophage phenotypic changes induced by chronic, slow spinal cord compression seen in cervical compressive myelopathy. Specially, we investigated the expression and colocalization of markers of microglia/macrophages (both M1 and M2 phenotypes) as well as the levels of neuroinflammatory cytokines closely related to these cells, which could promote neurotoxicity or neuroprotection and lesion repair in the twy/twy spinal cord.

Materials and Methods
The Spinal Hyperostotic twy/twy and Control Mice and Spinal Cord Progressive Compression Evaluated by Computed Tomography (CT) The Ethics Review Committee for Animal Experimentation of University of Fukui approved the experimental protocol. Spinal hyperostotic twy/twy mice purchased from the Central Institute for Experimental Animals (Kawasaki, Japan), were used in all experiments (aged 12 weeks; n = 29, 18 weeks; n = 29, 24 weeks; n = 29) (Table S1). Homozygous twy/twy mice were maintained by brother-sister mating of heterozygous Institute of Cancer Research (ICR) mice (+/twy). ICR mice at the age of 12, 18, and 24 weeks were used as control animals (n = 10 in each age group). The disorder is inherited in an autosomal recessive manner and the homozygous hyperostotic mouse is identified by a characteristic tip-toe walking at 6 to 8 weeks of age, but no congenital neurological abnormalities are detected at that age. The twy/twy mouse exhibits spontaneous calcified deposits posteriorly at the C1-C2 vertebral level, producing a variable degree of compression of the spinal cord between C2 and C3 cord segments with a general ankylosis of joints. The calcified mass grows in size progressively with age particularly in the atlantoaxial membrane, causing profound motor paresis at the age of 18-24 weeks [9,33,37].
For hematoxylin and eosin (H&E) staining, the resected cervical spine of each twy/twy mouse of different ages (n = 5 for each time point) was fixed in buffered formaldehyde for 48 hours at 4uC. The sample was then decalcified for 2 weeks at 4uC in 0.5 M ethylenediaminetetraacetic acid (0.5 M Tris-HCl buffer) at pH 7.6 and then embedded in paraffin using standard procedures. Serial 4-mm-thick cryostat sagittal and axial sections were prepared.
In order to confirm differences in the severity of compression before H&E staining, flow cytometry, and immunoblot analysis, we measured the spinal canal area in the cervical spine of twy/twy mice on CT scans (GE Medical Systems, Milwaukee, WI) obtained under anesthesia with ravonal (Tiopental H , Mitsubishi Tanabe Pharma, Osaka, Japan), using Image J, the image analysis software of the National Institutes of Health (Bethesda, MD). The correlation between age and spinal canal area at the site of maximum compression at C1-C2 vertebral level was determined. We also compared the spinal canal areas at C1-C2 to that at thoracic (Th) 1 vertebral level. In the same way, we also compared the spinal cord areas at the site of maximum compression in H&E staining at C1-C2 to that at Th1 vertebral level using the color image analyzer (MacSCOPE, Minani, Fukui, Japan).

Immunohistochemistry
Deep anesthesia was induced in each group of mice (n = 5 twy/ twy mice and n = 2 control ICR mice for axial sections, and n = 2 twy/twy mice for sagittal sections in each time point) followed by transcardial perfusion and fixation with 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS); the spinal cords were dissected and post-fixed in the same fixative for a few hours. The tissue samples were immersed in 10% sucrose in 0.1 M PBS at 4uC for 24 hours, and 30% sucrose in 0.1 M PBS for 24 hours. Segments of the cervical spinal cord were embedded in optimal cutting temperature compound (Sakura Finetek, Torrance, CA) and cut on a cryostat into serial 10 mm-thick axial or sagittal frozen sections, which were serially mounted on glass slides and fixed with 2% paraformaldehyde in 0.1M PBS for 5 minutes, rinsed in PBS and stored at 280uC.
All images were obtained using a fluorescence microscope (Olympus AX80, Olympus Optical, Tokyo) or a confocal laser scanning microscope (model TCS SP2, Leica Instruments, Nussloch, Germany), where the 488-and 543-nm lines of the argon/helium-neon laser were used for fluorescence excitation.

Semi-quantitative Analysis of Stained Tissues
Changes in CD11b-(red), NeuN-(green) and CD4-(red) positive areas at 12, 18, and 24 weeks of age twy/twy mice, and control ICR mice were assessed by the following procedure: serial axial sections were divided into five groups (slide glass) by collecting every fifth section separately from the site of maximum compression (between the C2 and C3 dorsal roots) and half of the spinal cord on the compressed side was analyzed using grain counting with the light intensity automatically set by the color image analyzer (MacSCOPE).
The proportions of CD11b-positive cells double immunostained with iNOS, CD16/32, arginase-1 or CD206 in each groups were determined semi-quantitatively by the following procedure: the serial axial sections divided into five groups (slides) as mentioned above from the site of maximum compression (between the C2 and C3 dorsal roots) and the number of positive cells per crosssection in each fluorescence stain was determined automatically using grain counting based on light intensity by a color image analyzer (MacSCOPE). The light intensity and threshold values were maintained at constant levels when collecting digitized images in all analysis. We documented the extent to which the microglia/macrophages present within the spinal cord were polarized by the M1/M2 ratio, as determined by the number of CD11b cells that were also positive iNOS and CD16/32/the number of CD11b cells that were also positive for CD206 and arginase-1.

Flow Cytometry
Immediately after deep anesthesia, the mouse was perfused intracardially with 200 ml of ice-cold 0.1 M PBS, and the spinal cords were harvested (n = 3 for each time point). The cervical spinal cord around the maximally compressed site was surgically dissected and dissociated with collagenase, 175 U/ml (Sigma-Aldrich, St. Louis, MO) for 1 hour at 37uC. Cells were washed in Dulbecco's modified Eagle's Medium (Invitrogen Life Technologies, Carlsbad, CA) containing 10% fetal bovine serum and filtered through a 40 mM nylon cell strainer (BD Biosciences, San Jose, CA) under centrifugation to remove tissue debris and obtain a single-cell suspension, as described in detail previously [38].
From this point on, prior to every staining, a cell-count was performed in every sample to ensure a cell density of 1.0610 6 cells/100 mL. Cells were incubated for 1 hour on ice with the following fluorescent antibodies: allophycocyanin (APC) rat anti-CD45, 0.25 mg/ml (BioLegend, San Diego, CA); Pacific Blue rat anti-Ly-6G/Ly-6C, 1.0 mg/ml (equivalent to Gr-1; BioLegend) and PerCP-Cy 5.5 rat anti-CD11b, 0.25 mg/ml (BD Pharmigen, San Jose, CA). For intracellular staining [39], the cells were resuspended in Fixation buffer and treated with Permeabilization Buffer (both from Santa Cruz Biotechnology) followed by resuspension in ice-cold PBS and incubation for 1 hour with goat anti-arginase 1, 1:200 conjugated to fluorescein isothiocyanate (FITC),1:200 (Santa Cruz Biotechnology) and phycoerythrin (PE)conjugated rabbit anti-iNOS, 3.0 mg/ml (Abcam plc); or PE/Cy7 conjugated rat anti-CD16/32, 1.0 mg/ml (Biolegend) and FITC rat anti-CD206, 1.0 mg/ml (Biolegend). Samples with cells alone were used as negative controls to eliminate background autofluorescence, and samples where the cells had been incubated with a single-added antibody were used as positive controls to set up the cytometer alignment and to remove any spectral overlap.
Flow cytometry was performed immediately using a FACS Canto TM II (Becton Dickinson Biosciences, San Jose, CA). Forward scatter was set to further eliminate any cellular debris from analysis. In each test, a minimum of 250,000 cells were analyzed and the data were processed using BD FACSDiva software (Becton Dickinson Biosciences). The different cell populations present in the suspension were classified according to the combination of expressed antigens, as stated in previous reports, as follows: CD11b high /CD45 low /GR-1 negative represented resting microglia [39], and CD11b high /CD45 high /GR-1 negative represented activated microglia/macrophages [38]. At the previously described time points, CD11b high cells in the spinal cord were sub-fractioned into a CD45 low /GR-1 negative population, identifying them as resting microglia. In a similar fashion, CD11b high cells were sub-fractioned into a CD45 high /GR-1 negative population, which identified them as activated microglia/macrophages.
The phenotype of microglia/macrophage sub-populations was corroborated through their expression of iNOS or CD16/32 (proinflammatory M1 phenotypes) as well as arginase 1 and CD206 (anti-inflammatory M2 phenotypes).

Myeloperoxidase (MPO) Staining and Assay
The 3,39-diaminobenzidine (DAB) staining kit (Muto Pure Chemicals Co., Tokyo) was used in each mouse (n = 3 for each time point) for cytochemical staining of MPO according to the instructions supplied by the manufacturer. Briefly, the peroxidase reaction was developed with 0.05% 3,39-DAB in 50 mmol/L Tris-HCl (pH 7.6) and 0.03% H 2 O 2 for 1 to 2.5 min. The sections were counterstained with eosin, dehydrated, and mounted. Oxidized 3,39-DAB (a brown, highly insoluble indamine polymer) was visible under light microscopy.
MPO activity levels in harvested compressed spinal cord tissues (n = 3 for each time point) were measured with MPO assay kit (BioVision, Milpitas, CA) using a spectrophotometer at 412 nm. One unit of MPO activity was defined as the amount of enzyme degrading 1 mmol of 5-thio-2-nitrobenzoic acid (TNB) per minute at 25uC. MPO activities in the spinal cord tissues were calculated by using a standard curve generated with MPO and expressed in units per miligram weight of wet tissue.

Immunoblot Analysis
Immediately after deep anesthesia, the spinal cord of each mouse (n = 3 for each time point) around the maximally compressed site (between C2 and C3 dorsal roots) was carefully dissected en bloc from the cervical spine and stored immediately at 280uC in liquid nitrogen. Segments were centrifuged at 15,0006g for 30 seconds using a BioMasher Rapid Homogenization Kit (Funakoshi, Tokyo), then solubilized in RIPA lysis buffer 1X (Santa Cruz Biotechnology), homogenized and stored at 280uC. The protein concentration was determined in the obtained samples by a Lowry protein Assay using a DC protein assay kit (Bio-Rad Laboratories, Hercules, CA). Laemmli sodium dodecylsulfate buffer samples containing the protein mixtures were boiled and subjected to immunoblot analysis. Total protein (20 mg/lane) was separated on 12.5% SDS-PAGE and transferred onto polyvinylidene difluoride membrane (PE Applied Biosystems, Foster, CA) for 70 minutes using a semi-dry blot apparatus. The membrane was washed twice in PBS containing 0.05% Tween 20, blocked by 5% skimmed milk in PBS for 1 hour at room temperature, and then incubated with one of the following antibodies: rabbit anti-IFN-c, 0.2 mg/ml (Abcam plc), rabbit anti-TNF-a, 0. the membrane was immersed in ECL Advance Western Blot Detection kit (GE Healthcare, Buckinghamshire, UK) for 1 minute and then exposed to X-ray film for visualization of peroxidase activity and determination of the level of each specific protein. The band intensities were normalized to b-actin, 1:2000 (Abcam plc), and Kaleidoscope Prestained Standards (Bio-Rad Laboratories) was used as the molecular weight control.

Statistical Analysis
All values are expressed as mean6 standard deviation (SD). Differences between groups were examined for statistical significance using one-way factorial analysis of variance (ANOVA). Before the priori comparison, Kolmogorov-Smirnov test was used for verification of normality. A p value ,0.05 denoted the presence of a significant difference with Tukey's post hoc analysis. The above tests were conducted using SPSS software version 11.0 (SPSS, Chicago, IL). The area of the spinal cord that was positive for CD11b immunostained increased and that area which was positive for NeuN immunostained cells decreased with advancement of spinal cord compression. The CD11b-positive area increased according to the degree of spinal cord compression in both the gray and white matters, especially in the anterior horn and anterior column of the maximally compressed site compared with the rostral or caudal sites of the spinal cord ( Fig. 2A). On the other hand, the NeuN-positive area, mainly in the anterior horn, decreased accordingly to the degree of spinal cord compression; the difference between 18-and 24-week-old twy/twy mice was significant (Fig. 2). The results of ICR mice in each age group (12-, 18-, and 24-week-old) were the same as those from 12-weekold twy/twy mice (Fig. S1A).

Chronic Progressive Spinal Cord Compression Induces Changes in the Phenotype Markers of Microglia/ Macrophages
To evaluate the phenotype of microglia/macrophages, tissues were immunostained with iNOS and CD16/32 for the M1 phenotype, and arginase-1 and CD206 for the M2 phenotype, as well as the pan-specific marker CD11b for microglia/macrophages. Double-positive merged cells were found particularly in the anterior horn and anterior column. In control ICR mice (with the same results seen in each age group; Fig. S1B) and in 12-week twy/ twy mice, multiple cells co-expressing arginase-1 and CD11b were found; but no CD11b-positive cells co-expressing iNOS, CD16/32 or CD206 were identified (Fig. 3A). In 18-week twy/twy mice, while some CD11b-positive cells co-expressed iNOS or CD16/32, the number of those co-expressing arginase-1 or CD206 and CD11b remained higher (Fig. 3B). In 24-week twy/twy mice, while the number of CD11b-positive cells co-expressing iNOS or CD16/32 remained elevated, that of cells co-expressing CD11b and arginase-1 or CD206 persisted as the most abundant type (Fig. 3C). These differences between iNOS-CD16/32 and arginase-1-CD206 were statistically significant in 18-and 24week-old twy/twy mice (Fig. 3D). Figure 3E shows the antigen expression ratio of M1 phenotype (CD11b-positive cells coexpressing iNOS or CD16/32)/M2 phenotype (CD11b-positive cells co-expressing arginase-1 or CD206) using immunofluorescence staining. The percentage of microglia/macrophages that was of M2 phenotype was 82.0% in 18-week-old and 61.6% in 24week-old twy/twy mice with a concomitant increase in the percentage of microglia/macrophages that were of M1 phenotype.
To further determine the activity of CD11b-positive cells in the chronically compressed spinal cords, the profile of the CD11bpositive cells was analyzed by flow cytometry (Fig. 4). Of the 250,000 spinal cord cells, 10.460.7% (26,04161,794 cells) was CD11b positive . The immunoprofile of CD11b-positive cells shifted from CD11b high /CD45 low /GR-1 negative cells (resting microglia) to CD11b high /CD45 high /GR-1 negative cells (activated microglia/ macrophages) with the advancement of spinal cord compression (Figs. 4A-C). While almost all CD11b-positive cells were resting microglia in control ICR mice (with the same results seen in each age group; Fig. S2A) and in 12-week twy/twy mice, the population of activated microglia/macrophages increased significantly in the 18-and 24-week twy/twy mice (Figs. 4D and E).
We also determined the subtypes of resting microglia and activated microglia/macrophages in terms of M1/M2 phenotypic marker expression (Fig. 5). Of the CD11b high /CD45 low /GR-1 negative cells (resting microglia), 10

M2 Microglia/Macrophages are a Source of Neurotrophic Factors in the Chronically Compressed Spinal Cord
To evaluate the expression of neurotrophic factors and phagocytic activity in correlation with M1 and M2 phenotypes, double immunofluorescence staining for BDNF, NGF, and Mac-2 with either iNOS or arginase-1 was performed. Double-positive cells were found particularly in the anterior horn and anterior column. In 12-week twy/twy mice, some expression of BDNF, NGF and no expression of Mac-2 was found (Fig. 6A): those expressions were also noted in 18-week-old twy/twy mice (Fig. 6B) and their intensity reached peak levels in 24-week-old twy/twy mice  (Fig. 6C). Some neurotrophic factors and Mac-2 colocalized with arginase-1-or CD206-positive cells, whereas they did not with iNOS-or CD16/32-positive cells (Fig. 6). These results indicate that the M2 phenotype, but not the M1 phenotype, is a source of neurotrophic factors.

Helper T Cells, but not Neutrophils, Infiltrate the Chronically Compressed Spinal Cord
To evaluate the infiltration of neutrophils and helper T cells in the chronically compressed spinal cord, tissues were stained for MPO and CD4. No MPO-positive cells and very small amount of MPO activity were detected irrespective of the degree of spinal cord compression (Fig. S4), while the area of CD4-positive increased with advancement of spinal cord compression; especially in the gray matter of 18-week-old twy/twy mice (Figs. 7A-F). Few positive cells were seen in each age group of the control ICR mice and in 12-week-old twy/twy mice (Fig. S1C).

Increased Expression of Th2 Cytokines and Neurotrophic Factors in Chronically Compressed Spinal Cord
Western blotting was performed to evaluate the correlation between the severity of spinal cord compression and IFN-c, TNF-a, IL-6, IL-4, IL-10, IL-13, BDNF, NGF, and Mac-2 protein levels (Fig. 8). IFN-c (28 kDa band) was weakly expressed and its expression level did not change with age. The intensities of the bands of TNF-a (19 kDa band) and IL-6 (26 kDa band) increased with worsening of spinal cord compression (Fig. 8A). The intensities of the bands for IL-4 (18 kDa band), IL-10 (37 kDa band), and IL-13 (13 kDa band) reached peak levels in 18-week twy/twy mice, but somewhat decreased in 24-week twy/twy mice (Fig. 8B). The expression of these proteins in control ICR mice (with the same results seen in samples of different age groups; Fig. S2B) was the same as that seen in 12-week twy/twy mice. These results indicate increased Th1 cytokine expression reflects worsening of spinal cord compression, whereas increased Th2 cytokine expression

Discussion
We reported in the present study that chronic progressive compression of the cervical spinal cord due to a calcified lesion of the atlantoaxial membrane at C1-C2 in the twy/twy mouse caused neuronal loss, as indicated by NeuN-positive cells, an increase in CD11b-positive cells in both the grey matter and white matter, Figure 3. The prevalence of phenotypically activated microglia/macrophages in association with increased severity of spinal cord compression in twy/twy mice. Immunofluorescence staining for the expression of iNOS and CD16/32 (green) for classically activated microglia/ macrophages (M1 phenotype) and arginase-1 and CD206 (green) for alternatively activated microglia/macrophages (M2 phenotype) co-localized with CD11b (red) in the anterior column of 12-(A), 18-(B) and 24-week-old (C) twy/twy mice. The numbers of CD11b-, CD11b/iNOS-and CD11b/CD16/32positive cells (arrow heads) increased with the worsening of spinal cord compression. The CD11b/arginase-1-and CD11b/CD206-expressing cells (arrow heads) were the predominant population. The differences between iNOS, CD16/32 and arginase-1, CD206 were statistically significant in 18and 24-week-old twy/twy mice. In control ICR mice, the expression of these factors was same as in 12-week-old twy/twy mice (D). The M1/M2 antigen expression ratio was higher in 24-week twy/twy mice compared with younger mice and control mice (E). Scale bars = 50 mm (A-C). Data are mean6SD. *p,0.05 (n = 5 for each time point). A-C microphotographs were taken using confocal laser scanning microscope. Several groups have characterized the pathological features of chronic spinal cord compression of patients with CSM and OPLL [5][6][7] or of the animal model of spinal cord compression seen in the twy/twy mouse [8,9,[33][34][35][36][37]. Significant similarities in histopathological and pathophysiological changes have been described in chronic progressive spinal cord compression and traumatic SCI. These similarities include a decrease in the neuronal population with degeneration and demyelination followed by a cascade of ''secondary'' injury [2]. Activated microglia, derived from resting microglia, and recruited macrophages from the peripheral circulation are among the main effector cells of the inflammatory response that follows SCI, and are associated with the production of proinflammatory cytokines and related immune effector molecules that can induce both necrotic and programmed cell death, which correlate with neurological deficit [10,11,16]. Activation of microglia after SCI is evident on day one, and the number of activated microglia has been shown to increase during the first 7 days and then plateau, 2-4 weeks after injury; with small differences related to the animal strain being used as a model [13]. The present results demonstrated a decrease in the NeuN-positive area in the anterior horn opposite the posterior compression and posterior horn and an increase in the CD11b-positive area, especially in the anterior horn and anterior column; these changes were in proportion with the degree of spinal cord compression. In our previous studies, we observed a significant reduction in the CD11b high cells in the spinal cord were sub-fractioned into a CD45 low /GR-1 negative population, identifying them as resting microglia; or CD45 high /GR-1 negative population, which identified them as activated microglia/macrophages. The numbers of resting microglia (CD11b high /CD45 low /GR-1 negative cells) were significantly lower in 24-week-old mice (D), while the numbers of activated microglia/macrophages (CD11b high /CD45 high /GR-1 negative cells) were higher in 18-and 24-week-old twy/twy mice (E). The number of resting microglia and activated microglia/macrophages in control ICR mice was same as in 12-week-old twy/twy mice (D, E). Data are mean6SD. **p,0.01 (n = 3 for each time point). doi:10.1371/journal.pone.0064528.g004 number of remaining surviving neurons when the transverse remnant area of the spinal cord of the twy/twy mouse spinal cord decreased to #70% of the control (+/twy mice) [33], and further that the extent of demyelination and Wallerian degeneration in the white matter increased proportionately with the magnitude of spinal cord compression [35]. We also reported a significant increase in TUNEL-positive cells in the anterior horn and also glial cells in the white matter, in spinal cord segments located rostral and caudal to the segment with maximal compression [8]. It is well-known that activated microglia/macrophages secrete cytotoxic substances including various cytokines (e.g., TNF-a, IL-1 and -6), reactive free radicals, and nitric oxide. In addition, the major role of microglia at the lesion centre is probably rapid phagocytosis of debris [25,40]. Our results suggest that neuronal loss through the mechanisms of necrosis and apoptosis is associated with activation of microglia/macrophages, production of proinflammatory cytokines by these cells, and putative phagocytosis, which may also be associated with axonal demyelination in the twy/twy spinal cord. This chain of events is somewhat similar to the delayed apoptosis of oligodendrocytes following activation of microglia after SCI [41].
As part of the response to chronic and progressive spinal cord compression, the present results showed high numbers of CD11bpositive cells, representing resting and activated microglia, as well as recruited macrophages, at different stages of microglia/ macrophages activation and polarization [29]. The proportion of both M1 and M2 microglia/macrophages increased with the severity of spinal cord compression with abundant cells appearing as the M2 phenotype at 24 weeks in twy/twy mice. Furthermore, the results also showed increased M1/M2 antigen expression rate, a ratio known to correlate with CNS damage and repair [22]. M2 microglia/macrophages seem to promote neuroprotection, neurogenesis, and repair even in severely compressed spinal cord. On the other hand, further progression of the spinal cord compression may have resulted in the induction of CD11b -positive cells into the M1 phenotype, which is known for its deleterious effects [27]. We reported previously the presence of axonal and dendritic outgrowth, which were induced by overexpression of neurotrophic factors at the site of compression in the twy/twy mice spinal cord [36]. Immunostaining in the present study demonstrated high coexpression of neurotrophic factors in M2 phenotype macrophages as well as increased expression of Mac-2, which reflects enhanced phagocytic activity [40,42]. These changes possibly represent the spinal cord response to chronic progressive spinal cord compression in order to maintain neuronal function [28,36].
The results of flow cytometry provided further details on the phenotypes of the involved cells, where resting microglia mainly expressed M2 markers, as described previously by other groups [17,32]. Our data confirmed that resting microglia are also the source of activated microglia/macrophages, since the population of resident microglia decreased while that of the activated microglia/macrophages increased in response to severity of spinal mechanical compression. In addition, the phenotypic changes in activated microglia/macrophages showed dynamic behavior with important M2 phenotype populations in 18-and 24-week-old twy/ twy mice and further increase in the M1 phenotype at 24 weeks, probably reflecting worsening of chronic spinal cord compression. The results related to the serial changes in the expression levels of phenotypic markers in activated microglia deserve special attention. It has already been documented both in vitro and in vivo that resting microglia possesses the M2 phenotype and are the only neural cells that express arginase-1 [26,32]. However, after their activation, these cells express different markers in a time-related manner. For example, arginase-1 is a well-documented early marker expressed in those cells that undergo the alternative pathway of activation to acquire the M2 macrophage phenotype, whereas iNOS is a marker of cells activated through the classical pathway, changing into M1 activated macrophages [22]. As the activation process advances, the fully mature activated cells express CD16/32 and CD206, which are phenotypic hallmarks of M1 and M2 macrophages, respectively [17,28]. Our immunohistochemical and flow cytometric data showed a similar behavior where the expression of such phenotypic markers increased with time, representing completion of the macrophage activation process. CD206 is an endocytic receptor that recognizes glycoprotein for antigen processing and presentation and is widely used to identify the M2 phenotype [26]. Our results showed that the expression of iNOS and arginase-1 persisted and increased with time; possibly related to the persistence of the insult triggering microglial/macrophages activation, since these markers mainly reflect the early stages of activation [22].
Cells that were double immunopositive for M1 and M2 phenotypic markers were seen locally within the spinal cord, which is highly suggestive of microglia/macrophages undergoing phenotype change due to environmental influences created by the cervical compression. Polarization of the microglia/macrophages requires an active response in the form of inflammatory cytokines, other immune effector molecules and neurotrophic factors released from the cells in the injured tissue as well as from different cells recruited to the site of the primary insult [27]. The sources of these cytokines remain controversial. Astrocytes seem to be the main source of TNF-a since they are the most abundant cell type after SCI capable of producing this cytokine [43]. Previous studies indicated that the main source of IFN-c is blood-derived cells recruited to the site of injury (macrophages, natural killer cells, neutrophils and helper T cells) through the inflammatory process mediated by some ILs [44,45]. Some of these recruited cells can produce Th2 cytokines, but the main source of IL-4 and IL-13 seems to be microglia [46,47]. However, the pathophysiology of chronic spinal cord compression in the twy/twy mouse spinal cord could include a limited blood cell access into the injured spinal cord, and there is no evidence of the destruction of the blood spinal cord barrier [33,35,37], thus limiting the infiltration of helper T cells and neutrophils. Under such circumstances, the expression of IFN-c is expected to be weak and TNF-a becomes the most important Th1 cytokine responsible for macrophage activation. The preserved blood spinal cord barrier would also limit the supply of Th2 cytokines; leaving resting and activated microglia at the injury site as the main source of Th2 cytokines for alternative activation of microglia/macrophages [46,47]. While there is no evidence to our knowledge of disruption of the blood spinal cord barrier in the twy/twy mouse to date, future research may prove otherwise. For analysis of this aspect, special twy/twy background chimeric mice models would be required [42], which are not currently commercially available; therefore, the possibility of cytokines deriving from blood and having an important role in determining inflammatory and immunomodulatory activity within the spinal cord remains as a possible limitation of our animal model. Our immunoblot results provided insight on the polarization process; the expression of Th2 cytokines increased from 18 weeks in possible correlation with the enlarged activated microglia population. The transient increase of Th2 cytokines explains our findings of M2 phenotype being the predominant type of activated cell; which would also explain the enhanced expression of neurotrophic factors and Mac-2 [19,26,29]. The present results also demonstrated increased expression of Th1 cytokines, as reported previously by our group [8], together with the reported increase in IL-6 expression as part of the inflammatory process [42,48]. Thus, over-expression of TNF-a in the twy/twy mice may be the most important Th1 cytokine responsible for M1 macrophage activation [28].
Interestingly, the resultant increase in the neurotropic factors at the site of spinal cord compression did not reverse the pathological changes of increased expression of proinflammatory cytokines and apoptosis of neurons and glia. Accordingly, the only treatment available at present is the surgical removal of the cause of chronic spinal cord compression [1][2][3][4]. However, surgery is unlikely to reverse the symptoms in patients with advanced cervical compressive myelopathy. In such patients, the use of specific antibodies against Fas ligand to reduce apoptosis of neurons and oligodendrocytes [7,37], promotion of neuronal regeneration, through blockade of the inflammatory cascade, and induction of the M2 phenotype [48] are promising options for combined medical therapies in future management of this condition.
In conclusion, the present study demonstrated that chronic and progressive spinal compression induced over-expression of Th1 cytokines (TNF-a, IL-6) and increased the population of classically activated macrophages (M1 phenotype). These changes could be responsible, at least in part, for neuronal loss and may also induce the demyelination of axons found in CSM and OPLL; however, they could also induce alternative activation by Th2 cytokines (IL-4, -10, and -13) and increase M2 microglia/macrophages, which provide neuroprotection and enhanced phagocytic activity. This work provides the rationale for therapeutic targeting of alternative activation of microglia/macrophages in human CSM and OPLL. Figure S1 Assessment of aging effects in control ICR mice by immunofluorescent staining. (A) Immunofluorescence staining for the expression of CD11b (red) and NeuN (green) in 12-, 18-and 24-week-old ICR mice. (B) Immunofluorescence staining for the expression of iNOS and CD16/32 (green) identifying classically activated microglia/macrophages (M1 phenotype) and arginase-1 and CD206 (green) characterizing alternatively activated microglia/macrophages (M2 phenotype); co-localized with CD11b (red) in the anterior column of 12-, 18and 24-week-old ICR mice. (C) Immunostaining of infiltrating helper T cells in 12-, 18-and 24-week-old ICR mice. In these