Dynamic Changes of Microglia/Macrophage M1 and M2 Polarization in Theiler's Murine Encephalomyelitis

Abstract

Microglia and macrophages play a central role for demyelination in Theiler's murine encephalomyelitis (TME) virus infection, a commonly used infectious model for chronic‐progressive multiple sclerosis. In order to determine the dynamic changes of microglia/macrophage polarization in TME, the spinal cord of Swiss Jim Lambert (SJL) mice was investigated by gene expression profiling and immunofluorescence. Virus persistence and demyelinating leukomyelitis were confirmed by immunohistochemistry and histology. Electron microscopy revealed continuous myelin loss together with abortive myelin repair during the late chronic infection phase indicative of incomplete remyelination. A total of 59 genes out of 151 M1‐ and M2‐related genes were differentially expressed in TME virus‐infected mice over the study period. The onset of virus‐induced demyelination was associated with a dominating M1 polarization, while mounting M2 polarization of macrophages/microglia together with sustained prominent M1‐related gene expression was present during the chronic‐progressive phase. Molecular results were confirmed by immunofluorescence, showing an increased spinal cord accumulation of CD16/32⁺ M1‐, arginase‐1⁺ M2‐ and Ym1⁺ M2‐type cells associated with progressive demyelination. The present study provides a comprehensive database of M1‐/M2‐related gene expression involved in the initiation and progression of demyelination supporting the hypothesis that perpetuating interaction between virus and macrophages/microglia induces a vicious circle with persistent inflammation and impaired myelin repair in TME.

Introduction

Multiple sclerosis (MS), one of the most frequent central nervous system (CNS) diseases in young adults, is a chronic demyelinating disease of unknown etiology and possibly multifactorial causes 13. Based on the generation of myelin‐specific immune responses, MS is regarded as an autoimmune disease 4, 66 presumably triggered by virus infections 32, 62. Because of clinical and pathological similarities, Theiler's murine encephalomyelitis (TME) represents a commonly used infectious animal model for the chronic‐progressive form of human MS 14, 58, 61, 72. Following intracerebral infection with a low virulent BeAn strain of TME virus (TMEV), susceptible mouse strains develop persistent CNS infection with immune‐mediated spinal cord demyelination and remyelination failure 25, 28, 40, 45, 54, 59, 84, 85, 88.

Microglia and CNS‐infiltrating macrophages play a central role in the pathogenesis of TMEV‐induced demyelination. They represent target cells for viral persistence during the chronic disease phase 39, 74 and contribute to myelin damage by the release of myelinotoxic factors (bystander demyelination), delayed‐type hypersensitivity reaction and induction of myelin‐specific autoimmunity 47, 55. Similarly, microglia induce myelin damage also in autoimmune and toxic rodent models for MS, such as experimental autoimmune encephalomyelitis (EAE) and cuprizone‐induced demyelination, respectively 46, 78, 91. The current concept of microglia/macrophages plasticity describes different cell populations with distinct and even opposing functions. For instance, M1‐type microglia/macrophages promote inflammation, which leads to protective immunity against pathogens but if uncontrolled also to immune‐mediated tissue damage by the release of pro‐inflammatory cytokines, reactive oxygen species and nitric oxide 67, 69. In contrast, M2‐type cells exhibit neuroprotective properties usually during advanced disease stages due to phagocytosis of debris, promoting tissue repair and termination of neuroinflammation by downregulating M1 and Th1 immune responses 42.

So far, only few reports mention the polarizing effects of TMEV upon microglia in vitro 23. Moreover, M1‐ and M2‐type cells represent merely two extremes of the polarization continuum and macrophages/microglia with an intermediate activation status can be observed inter alia in demyelinating MS lesions 90, demonstrating the need for quantitative analyses of M1‐/M2‐related factors in myelin disorders. Thus, the aims of the present study were to (i) select candidate genes involved in macrophage/microglia polarization by DNA microarray analyses and (ii) to compare their transcription levels to get insights in M1/M2 balances during the initiation and progression of virus‐induced demyelination. In addition, (iii) dynamic changes of M1‐ and M2‐type cells in TME were determined with the aid of immunofluorescence.

Materials and Methods

Experimental design

Five‐week‐old female SJL/J mice (Harlan, Borchen, Germany) were inoculated into the right cerebral hemisphere with 1.63 × 10⁶ plaque‐forming units/mouse of the BeAn strain of TMEV in 20 μL Dulbecco's modified Eagle medium (PAA Laboratories, Cölbe, Germany) with 2% fetal calf serum and 50 μg/kg gentamicin. Mock‐infected animals received 20 μL of the vehicle only. Inoculation was carried under general anesthesia with medetomidine (0.5 mg/kg, Domitor; Pfizer, Karlsruhe, Germany) and ketamine (100 mg/kg, Ketamine 10%; WDT eG, Garbsen, Germany). All experiments were performed in groups of six TMEV‐ and three to six mock‐infected mice, euthanized 14, 42, 98 and 196 days postinfection (dpi). For histology, immunohistochemistry and special stains, thoracic spinal cord segments were removed immediately after death and fixed in 10% formalin for 24 h, decalcified in disodium‐ethylenediaminetetraacetate for 48 h and subsequently embedded in paraffin wax. For microarray analysis and immunofluorescence, spinal cords were immediately removed, snap‐frozen in liquid nitrogen and stored at −80°C 27, 64, 88.

The animal experiments were approved and authorized by the local authorities [Niedersächsisches Landesamt für Verbraucherschutz‐und Lebensmittelsicherheit (LAVES), Oldenburg, Germany; permission number: 33.9.42502‐04/07/1331, 509c‐42502‐02/589 and 33‐42502‐05/963].

Histology

Leukomyelitis was evaluated on hematoxylin and eosin (HE)‐stained transversal sections using a semi‐quantitative scoring system based on the degree of perivascular infiltrates: 0 = no changes, 1 = scattered perivascular infiltrates, 2 = two to three layers of perivascular inflammatory cells, 3 = more than three layers of perivascular inflammatory cells as described previously 22. For the evaluation of myelin loss, serial sections of spinal cord were stained with Luxol fast blue‐cresyl violet (LFB–CV) and the degree of demyelination was semi‐quantitatively evaluated as follows: 0 = no change, 1 = 25%, 2 = 25%–50% and 3 = 50%–100% of the white matter affected 22. The scoring was performed separately on all four quarters of spinal cord transversal sections. For each animal the arithmetic average of leukomyelitis and myelin loss was calculated. Histological data used for the present study were generated in our previous studies 87, 88.

Immunohistochemistry

Immunohistochemistry was performed using a polyclonal rabbit anti‐TMEV capsid protein VP1‐specific antibody as described before 38. Briefly, for blocking of the endogenous peroxidase, formalin‐fixed, paraffin‐embedded tissue sections were treated with 0.5% H₂O₂ diluted in methanol for 30 minutes at room temperature. Subsequently, slides were incubated with the primary antibody at a dilution of 1:2000 for 16 h at 4°C. Goat‐anti‐rabbit IgG diluted 1:200 (BA9200, H + L, Vector Laboratories, Burlingame, CA, USA) was used as a secondary antibody for 1 h at room temperature. Sections used as negative controls were incubated with rabbit normal serum at a dilution of 1:2000 (Sigma‐Aldrich Chemie GmbH, Taufkirchen, Germany). Slides were subsequently incubated with the peroxidase‐conjugated avidin–biotin complex (ABC method, PK‐6000, Vector Laboratories) for 30 minutes at room temperature. After the positive antigen–antibody reaction visualization by incubation with 3.3‐diaminobenzidine‐tetrachloride in 0.1 M imidazole, sections were counterstained with Mayer's hematoxylin.

Immunofluorescence

Methanol‐fixed frozen sections of the thoracic spinal cord were rinsed in 0.1% Triton X‐100 (Sigma‐Aldrich) in phosphate‐buffered saline (PBS) for 30 minutes. Nonspecific binding was blocked with 20% goat or horse serum, respectively, diluted in PBS/0.1% Triton X‐100/1% bovine serum albumin for 30 minutes. After washing with 0.1% Triton X‐100 in PBS, slides were incubated with primary CD68– (monoclonal rat anti‐mouse antibody, Ab53444, clone FA‐11, Abcam Ltd., Cambridge, United Kingdom; dilution 1:200) and CD107b– (monoclonal rat anti‐mouse antibody MCA2293, clone M3/84, AbD Serotec, Duesseldorf, Germany; dilution 1:200) for the detection of macrophages/microglia. For visualization of M1‐type macrophages/microglia, a CD16/32‐specific antibody (monoclonal rat anti‐mouse, 553141, clone 2.4G2, BD Pharmingen, Heidelberg, Germany; dilution 1:25) and for M2‐type cells an arginase‐1‐specific antibody (polyclonal goat anti‐human, SC‐18351, Santa Cruz Biotechnology, Dallas, USA; dilution 1:50) and a Ym1‐specific antibody (polyclonal rabbit anti‐mouse antibody, ab93034, Abcam Ltd.; dilution 1:100) were used. Slides were incubated for 1 h followed by washing in PBS/0.1% Triton X‐100. As a negative control, slides were incubated with goat, rat or rabbit serum in the same concentration as the primary antibodies. Subsequently slides were incubated with secondary DyLight 488‐conjugated donkey anti‐goat (Jackson ImmunoResearch Laboratories, Dianova, Bar Harbor, USA; dilution 1:200), Cy2‐conjugated goat anti‐rabbit IgG antibodies (Jackson ImmunoResearch Laboratories, Dianova; 1:200) and Cy3‐conjugated goat anti‐rat (Jackson ImmunoResearch Laboratories, Dianova; dilution 1:200), respectively, for 1 h at room temperature and afterwards washed in PBS.

For double staining, slides were simultaneously incubated for 90 minutes with the CD107b‐specific monoclonal antibody (see above) and primary antibodies directed against CCL5 (polyclonal rabbit anti‐human antibody; ABIN674949, antibodies‐online GmbH dilution 1:10), CXCL10 (polyclonal rabbit anti‐mouse antibody; ABIN687442, antibodies‐online GmbH dilution 1:10), interferon (IFN)‐γ (polyclonal rabbit anti‐human antibody, ABIN669141, antibodies‐online GmbH; dilution 1:20), and TMEV [polyclonal rabbit anti‐TMEV BeAn VP1 antibody; dilution 1:2000 38 ], respectively. Cy2‐conjugated goat anti‐rat (Jackson ImmunoResearch Laboratories, Dianova; dilution 1:200) and Cy3‐conjugated goat anti‐rabbit (Jackson ImmunoResearch Laboratories, Dianova; dilution 1:200) secondary antibodies were simultaneously used to visualize the respective antigens (see above). Nuclear counterstaining was performed with 0.01% bisbenzimide (H33258, Sigma‐Aldrich) for 10 minutes and sections were mounted with Dako fluorescent mounting medium (Dako Diagnostika).

Statistical analyses

For noncategory data obtained by histology, immunohistochemistry and immunofluorescence, a Mann–Whitney U‐test was performed. A P‐value of less than 0.05 was considered as statistically significant.

Electron microscopy

Electron microscopy was performed as described previously 37, 89. Spinal cord samples were fixated with 2.5% glutaraldehyde and incubated overnight at 4°C. Post‐fixation was performed in 1% aqueous osmium tetroxide, and after five washes in cacodylate buffer (5 minutes each), samples were dehydrated through series of graded alcohols and embedded in Epon 812 medium (Serva, Heidelberg, Germany). Semi‐thin sections were cut on a microtome (Ultracut Reichert‐Jung, Leica Microsystems, Wetzlar, Germany) and stained with uranyl citrate for 15 minutes. After eight washing steps, samples were incubated with lead citrate for 7 minutes. Ultra‐thin sections were cut with a diamond knife (Diatome, Hatfield, USA) and transferred to copper grids. For descriptive ultrastructural analyses of white matter changes, 100 axons and their myelin sheaths per animal were examined for the presence of degenerative changes (myelin sheath vacuolization, myelin loss) and regeneration (oligodendrocyte‐type remyelination, Schwann cell‐type remyelination) by a transmission electron microscope (EM 10C, Zeiss, Oberkochen, Germany). For quantification of phagocytic activity (gitter cell morphology, presence of myelin fragments and/or apoptotic bodies within the cytoplasm), a total of 100 macrophages/microglia were investigated.

Microarray analyses

RNA was isolated from frozen spinal cord samples using the RNeasy Mini Kit (Qiagen, Hilden, Germany), amplified and labeled using the Message Amp II‐Biotin Enhanced Kit (Ambion, Austin, TX, USA) and hybridized to GeneChip mouse genome 430 2.0 arrays (Affymetrix, Santa Clara, CA, USA) as described 88. Six biological replicates were used per group and time point except for five TMEV‐infected mice at 98 dpi. Background adjustment and quantile normalization was performed using RMAExpress 6. MIAME compliant data set are deposited in the ArrayExpress database (E‐MEXP‐1717; http://www.ebi.ac.uk/arrayexpress).

Selection of M1‐ and M2‐associated genes

For molecular characterization of macrophage/microglia polarization, a data set of genes differentially expressed in the spinal cord of TMEV‐infected SJL mice obtained in our previous global gene expression analysis was used 88. The present analyses focused on a list of genes associated with M1 or M2 polarization of microglia/macrophages (Supplemental Table S1) according to peer‐reviewed publications 15, 17, 34, 53. The fold change was calculated as the ratio of the inverse‐transformed arithmetic means of the log2‐transformed expression values of TMEV‐infected vs. mock‐infected mice. Downregulations are shown as negative reciprocal values. Independent pair‐wise Mann–Whitney U‐tests (IBM SPSS Statistics version 20; IBM Corporation, Armonk, NY, USA) comparing TMEV‐ and mock‐infected mice were calculated followed by adaption of the P‐values according to the method described by Storey and Tibshirani using QVALUE 1.0 82. Significantly differentially expressed genes between TMEV‐ and mock‐infected mice were selected employing a q‐value ≤0.05 cutoff combined with a ≥2.0 or ≤–2.0 fold change filter. The relative percentage of differentially expressed M1 vs. M2 marker genes was compared for each time point employing Fisher's exact tests (P‐value ≤0.05).

Results

Histological scoring of spinal cord inflammation and demyelination

Histological data used for the present study were generated in our previous studies 87, 88. Examination of the HE‐stained spinal cord sections revealed a mononuclear inflammation (leukomyelitis) within the white matter of TMEV‐infected mice beginning at 14 dpi. The inflammatory changes increased towards 98 dpi and were significantly increased compared with mock‐infected control animals at all investigated time points: 14 dpi (P = 0.011), 42 dpi (P = 0.002), 98 dpi (P = 0.013) and 196 dpi (P = 0.002; Figures 1 and 2). The amount of demyelination increased until 196 dpi (Figures 1 and 2). At three investigated time points (42, 98 and 196 dpi), demyelination in the spinal cord of TMEV‐infected SJL mice was significantly increased compared with mock‐infected control mice (P = 0.002, P = 0.007, P = 0.002) as determined by the myelin stain LFB–CV (Figures 1 and 2).

Figure 1.

Histological lesions in the spinal cord of  T heiler's murine encephalomyelitis virus‐infected mice. A. Lymphocytic meningomyelitis (arrows) and B. Mild vacuolization of the spinal cord white matter in an infected animal at 42 days postinfection. C. Prominent infiltration of macrophages/microglia in the spinal cord and lymphocytic meningitis (arrow) at 196 days postinfection. D. Demyelination of the spinal cord white matter (asterisks) at 196 days postinfection. E. Higher magnification of (C) showing activated macrophages/microglia with a foamy cytoplasm (gitter cells). F. Note accumulation of myelin debris within the cytoplasm of macrophages/microglia, indicative of myelinophagia. GM = gray matter; bars = 200 μm (A,B), 100 μm (C,D) and 30 μm (E,F); hematoxylin‐eosin stain (A,C,E), luxol fast blue stain (B,D,F).

Figure 2.

Scoring of demyelinating leukomyelitis in  T heiler's murine encephalomyelitis virus‐infected mice. A. Histology reveals inflammatory responses in the spinal cord (leukomyelitis) at all investigated time points. B. Detection of demyelination in the spinal cord white matter at 42, 98 and 196 days postinfection. dpi = days postinfection; mock = mock‐infected control mice; TMEV = Theiler's murine encephalomyelitis virus‐infected mice; * = significant difference (P ≤ 0.05, Mann–Whitney U‐test). Box and whisker plots display median and quartiles with maximum and minimum values.

Quantification of virus load in the spinal cord and virus detection in CD107b⁺ microglia/macrophages

Immunohistochemistry for the detection of virus protein in the spinal cord of TMEV‐infected mice revealed infection at all investigated time points (14, 42, 98 and 196 dpi). Although at 14 dpi infected cells were found scattered in the grey and white matter, with the onset of demyelination at 42 dpi positive cells were located predominantly in the lesions of the ventral spinal cord white matter (Supplemental Figure S1). No positive signals were observed in mock‐infected control mice.

Immunofluorescence double staining was performed to demonstrate macrophage/microglia infection. Results showed that at 14 dpi 25%–50%, at 42 dpi 40%–57%, at 98 dpi 38%–67% and at 196 dpi 33%–58% of TMEV‐infected cells represent CD107b⁺ macrophages/microglia (Supplemental Figure S2).

Characterization of myelin alterations and regeneration by electron microscopy

Descriptive ultrastructural analyses revealed subtle myelin changes before the onset of overt demyelination at 14 dpi in an average of 0.3% of investigated axons, characterized by vacuolization of myelin sheaths. At 42 dpi 2.2% of axons showed myelin sheath vacuolization and 5.8% of axons showed a complete loss of myelin (Figure 3). At 98 dpi an average of 2.8% of vacuolated myelin sheaths were observed and 8.4% of axons were totally denuded in demyelinated foci. At 196 dpi 5.0% of axons within white matter lesions showed a complete loss of myelin sheath, while 2.5% of axons showed oligodendrocyte‐type remyelination and 0.7% Schwann cell‐type remyelination (Figure 3; Supplemental Table S2), indicative of beginning but abortive myelin repair 89. Remyelination by Schwann cells was characterized by the presence of oval to signet ring‐shaped cells in close proximity to axons, ensheathing axons with myelin on a one‐to‐one basis (Figure 3; 20, 94).

Figure 3.

Ultrastructural analyses of the spinal cord white matter of  T heiler's murine encephalomyelitis virus‐infected mice by transmission electron microscopy. A. Macrophages/microglia containing phagocytized myelin fragments (white asterisks) at 42 days postinfection, characteristic of myelinophagia (M = nucleus of a macrophage/microglial cell; magnification 13 300×). B. Demyelinated axons (black asterisks) lacking myelin sheaths and focal myelin vacuolization (arrow) in an infected mouse at 96 days postinfection. For comparison, myelinated axons with intact myelin sheaths are labeled with triangles (magnification 6600×). C. Oligodendrocyte in proximity to multiple remyelinated axons with thin myelin sheaths (black asterisks) during late chronic infection phase (196 days postinfection) indicative of oligodendrocyte‐mediated remyelination. Normally myelinated axons are labeled with triangles (O = nucleus of an oligodendrocyte; magnification 5300×). D. Schwann cell remyelination in a demyelinated area at 196 days postinfection characterized by comparatively thick newly formed myelin sheaths (arrows) and one Schwann cell per axon relationship (S = nucleus of Schwann cell; magnification 6650×).

Phagocytosis of myelin fragments associated with denuded axons, representing a hallmark of active demyelination, was observed starting 42 dpi. At this time point, an average of 40.2% of microglia/macrophages displayed gitter cell morphology with phagocytized myelin in the cytoplasm (myelinophages; Figure 3). At 98 and 196 dpi, 50.1% and 51.5% of investigated macrophages/microglia represent myelinophages. In addition, phagocytized apoptotic bodies were present in an average of 9.3% of macrophages/microglia at 42 dpi, followed by a decline at 98 (0.7%) and 196 dpi (0.5%; Supplemental Table S2).

Quantification of M1‐ and M2‐related gene expression by DNA microarray analyses

In order to get insights into polarization related to microglia/macrophages, DNA microarray analyses of spinal cord tissue have been performed. A total of 151 genes related to macrophages/microglia polarization were extracted from peer‐reviewed publications, of which 72 and 66 were unequivocally assigned as M1 and M2 marker genes, respectively. Thirteen genes were assigned to both polarization types (Supplemental Table S1).

A total of 59 genes (39.1%) were differentially expressed in TMEV‐infected mice over the study period (Figure 4; Supplemental Table S3). Most strikingly, although the number of differentially expressed genes increased over the study period for both phenotypes, comparison of the relative proportion of differentially expressed M1 vs. M2 marker genes revealed a significantly higher percentage of differentially expressed M1 marker genes at 14 (P = 0.035) and 42 dpi (P = 0.016). In addition, a statistical tendency (P = 0.078) of increased M1‐associated genes was observed at 98 dpi, whereas a comparable proportion of M1 and M2 marker genes was detected at later time points (Figure 4).

Figure 4.

Expression profile of  M 1‐ and  M 2‐related genes in the spinal cord during the course of  T heiler's murine encephalomyelitis. A. Heat map displays fold changes indicated by a color scale ranging from –4 (relative low expression) in green to 4 (relative high expression) in red. Fifty‐nine out of 151 selected genes are differentially expressed in infected mice. B. Comparison of the relative proportion (percentage) of differentially expressed M1 vs. M2 marker genes employing the Fisher's exact test revealed a significant dominance (* = P ≤ 0.05) of M1‐related genes at 14 and 42 days postinfection (dpi). A statistical tendency (P = 0.078) of an increased M1‐associated gene expression is observed at 98 dpi, whereas comparable proportions of M1 and M2 marker genes are detected at 196 dpi.

According to the function, differentially expressed genes were assigned to seven pathways, including chemotaxis (group I; 15 genes); phagocytosis, antigen processing and presentation (group II; 16 genes); cytokine and growth factor signaling (group III; 12 genes); toll‐like receptor signaling (group IV; two genes); apoptosis (group V; four genes); extracellular matrix interaction and cell adhesion (group VI; five genes); and miscellaneous genes not related to a specific pathway (group VII; five genes; supplemental Table S3). In group I, 53.3% of genes (8/15 genes) were upregulated on 14 dpi, while at subsequent time points nearly all genes were significantly upregulated. In group II and III 62.5% of genes (10/16 genes) and 50.0% of genes (6/12), respectively, were upregulated at 14 dpi followed by an upregulation of nearly all genes at 42, 98 and 196 dpi in both groups. Tlr1 (group IV) was significantly transcribed at 42, 98 and 196, whereas expression of Tlr2 was observed during the entire observation period. Seventy‐five percent of apoptosis‐related genes (3/4 genes; group V) were significantly upregulated in infected mice at 14 dpi and 100% at subsequent time points. Although at 14 dpi 40.0% of genes (2/5 genes), all genes (100%) were upregulated at 42, 98 and 196 dpi. Miscellaneous genes not assigned to a specific pathway (group VII) included Atf3, Arg1, Cepba, Chi3l3 and Hexb. No genes were differentially expressed at 14 dpi. Atf3, Arg1 and Cebpa were significantly increased at 42, 98 and 196 dpi, whereas the M2 marker Chi3l3 (aka Ym1) was only transcribed during the late chronic phase at 196 dpi (Supplemental Table S3).

Temporal changes of macrophages/microglia subsets and verification of DNA microarray results by immunofluorescence

Immunofluorescence was used to confirm the results obtained by gene expression profiling. The number of microglia/macrophages increased over time in the spinal cord of infected mice with highest numbers of CD107b⁺ and CD68⁺ microglia/macrophages in the late stages of the disease. CD16/32⁺ M1‐ and also arginase‐1⁺ M2‐type cells were significantly increased compared with non‐infected animals at 42, 98 and 196 dpi (Figures 5 and 6). Interestingly, a significant increase of Ym1⁺ M2‐type cells was found only at 98 and 196 dpi (Figures 5 and 6), suggestive of late M2 polarization during the chronic demyelinating phase. To further substantiate this, statistical analyses between the early (42 dpi) and the late demyelinating phase (196 dpi) have been performed. Results revealed a significant time‐dependent increase of CD107⁺ macrophages/microglia (P = 0.006), arginase‐1⁺ M2‐type cells (P = 0.042) and Ym1⁺ M2‐type cells (P = 0.009), whereas no significant temporal differences were observed for CD68⁺ macrophages/microglia and CD16/32⁺ M1‐type cells (data not shown). Accordingly, the ratio of arginase‐1⁺ M2‐type cells to CD16/32⁺ M1‐type cells (P = 0.044) and the ratio of Ym1⁺ M2‐type cells to CD16/32⁺ M1‐type cells (P = 0.008) significantly increased over time (Figure 5), characteristic of mounting M2 responses during disease progression.

Figure 5.

Quantification of different macrophage/microglia subsets in the spinal cord of  T heiler's murine encephalomyelitis virus‐infected mice by immunofluorescence. Significant increase of (A) CD68⁺ cells, (B) CD107b⁺ cells, (C) CD16/CD32⁺ cells and (D) arginase‐1⁺ cells at 42, 98 and 196 days postinfection (dpi) and of (E) Ym1⁺ cells at 98 and 196 dpi in infected mice compared with mock‐infected control mice. TMEV = Theiler's murine encephalomyelitis virus‐infected mice; mock = mock‐infected control mice; * = significant difference (P ≤ 0.05, Mann–Whitney U‐test). Box and whisker plots display median and quartiles with maximum and minimum values. (F) Significantly elevated ratios of arginase‐1⁺ M2‐type cells to CD16/32⁺ M1‐type cells (arginase‐1:CD16/32) and Ym1⁺ M2‐type cells to CD16/32⁺ M1‐type cells (Ym1:CD16/32) at 196 dpi compared with 42 dpi. Columns display median with maximum and minimum values. * = significant difference (P ≤ 0.05, Mann–Whitney U‐test).

Figure 6.

Detection of different macrophage/microglia subsets in the spinal cord of  T heiler's murine encephalomyelitis virus‐infected mice by immunofluorescence. Accumulation of (A) CD107b⁺ cells, (B) CD68⁺ cells, (C) arginase‐1 (Arg‐1)⁺ cells, (D) CD16/32⁺ cells and (E) Ym1⁺ cells in the spinal cord white matter at 196 days postinfection. Insets show higher magnifications of labeled cells. BIS = bisbenzimide (blue nuclear counterstain).

Employing the Spearman's rank correlation coefficient, the amounts of all investigated macrophage/microglia proteins (CD68, CD107b, CD16/32, arginase‐1, Ym1) were significantly, positively correlated with the expression level of the respective genes (Table 1).

Table 1.

Correlation between data obtained by gene expression analyses and immunofluorescence

Gene expression	Immunofluorescence
	CD107b	CD68	CD16/32	Arginase‐1	Ym1
Arginase‐1	0.756*	0.686*	0.650*	0.630*	0.662*
Cd68	0.735*	0.751*	0.757*	0.708*	0.717*
Cd32b	0.722*	0.854*	0.760*	0.624*	0.720*
Cd16	0.723*	0.852*	0.804*	0.589*	0.691*
Cd107b	0.629*	0.629*	0.636*	0.476*	0.595*
Ym1	0.320	0.348	0.321	0.449*	0.563*

Spearman's rank correlation coefficient was used to correlate absolute numbers (positive cells/spinal cord) of CD107b⁺, CD68⁺, CD16/32⁺, aginase‐1⁺ and Ym1⁺ cells with the respective mRNA level measured by microarray analysis in the spinal cord of Theiler's murine encephalomyelitis virus‐infected mice. Significant differences of the correlation coefficient from zero are marked as follows: * = P ≤ 0.01.

In order to demonstrate the expression of M1‐related chemokines (CCL5 and CXCL10) and IFN‐γ in macrophages/microglia during the early (42 dpi) and late demyelinating phase (196 dpi), immunofluorescence double staining has been performed. Results revealed that both chemokines are preferentially expressed by macrophages/microglia in spinal cord white matter lesions, since at 42 dpi 91%–100% and at 196 dpi 80%–92% of CCL5‐positive cells represent CD107⁺ cells (Supplemental Figures S3 and S4). Similarly, co‐localization with CD107b was observed in 70%–100% at 42 dpi and in 80%–94% of cells expressing CXCL10 at 196 dpi (Supplemental Figures S3 and S4). At 42 dpi 52%–80% and at 196 dpi 72%–93% of IFN‐γ‐positive cells are co‐labeled with CD107b (Supplemental Figures S3 and S4), showing that in addition to CNS‐infiltrating lymphocytes 63 also macrophages/microglia contribute to IFN‐γ production in demyelinating lesions of TMEV‐infected mice.

Discussion

The present study provides a comprehensive database of M1‐/M2‐related genes expressed during the initiation and progression of TME. Although most molecules are produced also by other resident CNS cells and recruited lymphocytes 11, 26, all selected genes can be transcribed by macrophages/microglia and are involved in their polarization, respectively 15, 17, 34, 53. Results revealed an imbalance of M1/M2 responses during the onset of virus‐induced demyelination, characterized by the dominance of CD16/32⁺ M1‐type cells and disproportionally elevated M1‐related gene expression in the spinal cord of infected mice. With disease progression an accumulation of arginase‐1⁺ and Ym1⁺ potentially neuroprotective M2‐type cells 12, 60, 77 together with mounting transcription of M2‐related genes was found. However, sustained prominent M1 responses emphasize the importance of innate immunity for immunopathology and progressive myelin loss in demyelinating disorders as discussed for MS 51, 90.

Differentially expressed M1‐related genes at 14 dpi in the spinal cord of TMEV‐infected mice predominately consist of factors, such as chemokines, involved in the CNS recruitment of macrophages, T cells and B cells (Table S3, group I). Simultaneously, migration of CD68⁺ antigen presenting cells and activation of genes related to innate and adaptive immunity within the CNS‐draining cervical lymph node has been observed in TMEV‐infected mice during the acute phase of the disease in our previous study 64. Under neuroinflammatory conditions, chemokines and their receptors are produced by different cell types, such as microglia, astrocytes, neurons and infiltrating leukocytes 83. In the present study, double labeling revealed that macrophages/microglia are a major source of CCL5 (aka RANTES) and CXCL10 (aka IP‐10) within demyelinating lesions, which are preferentially expressed by M1‐type cells 50. In TME, both chemokines have been shown to critically control leukocyte CNS influx and antiviral immune responses, respectively 56, 73, 75. In addition, CCL5 and CXCL10 are upregulated in the EAE model and the cerebrospinal fluid of MS patients during demyelinating events, demonstrating their functional role in immune‐mediated damage 18, 49, 79. Interestingly Ym1, detected by microarray analysis and immunofluorescence, also displays chemotactic activity and has been demonstrated to promote Th2 cytokine expression 9, 92, which might reduce Th1‐mediated immunopathology but probably also protective antiviral immunity in TMEV‐infected susceptible mice strains.

M1 responses are a hallmark of early innate immunity following viral infection mediated by the interaction between microglial toll‐like receptors (Table S3, group IV) and cellular compounds (damage‐associated molecular pattern) and pathogen‐associated molecular pattern, respectively 34, 35. However, besides their pivotal role for antiviral immunity, microglia have been demonstrated to induce also myelin‐specific adaptive Th1 responses in TMEV‐infected mice 65. Similarly, M1‐polarized cells foster immunopathology in primary autoimmune CNS disorders 57 and the drug Fasudil ameliorates the clinical severity of EAE by shifting macrophages/microglia from M1 to a protective M2 phenotype 46. In addition, selective inhibition of M1‐type microglia by minocycline treatment reduces neurodegeneration as demonstrated in mouse models for amyotrophic lateral sclerosis 36. Similar to TME, experimental spinal cord injury in mice leads to microglial polarization into a pro‐inflammatory and neurotoxic M1 phenotype, which might function as an early trigger of degeneration and immunological events at later disease stages 34. Excessive microglial responses can be observed also in human and canine spinal cord trauma, which leads to potentially destructive effects by the release of pro‐inflammatory cytokines, proteolytic molecules and reactive oxygen species 2, 3, 19, 52, 70, 80, 81. Taken together, an imbalance toward M1 dominance represents a potential prerequisite for lesion initiation in TME as currently discussed for MS 21. Similar to findings in the present study, early innate immune responses with activated pro‐inflammatory microglia can be detected in pre‐demyelinating and early demyelinating MS lesions, which are supposed to induce myelin damage and immunopathology 21, 51.

In the present study, the onset of demyelination and phagocytosis of myelin and apoptotic cells is accompanied by an upregulation of genes involved in antigen processing, presentation and T‐cell stimulation (Table S3, group II). The functional relevance of phagocytic macrophages/microglia for the pathogenesis of CNS damage is discussed controversially. On the one hand, phagocytosis of myelin debris enhances CNS regeneration following traumatic injury 93. Moreover, ingestion of myelin induces a foamy appearance and anti‐inflammatory function of cultured human macrophages and myelinophages within MS lesions acquire M2 phenotype, which are supposed to contribute to resolution of inflammation and tissue repair 8. In addition, phagocytosis of apoptotic cells by cultured rodent microglia leads to diminished pro‐inflammatory cytokine production with a reduced ability to activate T cells 48. On the other hand, incorporation of myelin and cellular debris by microglia is able to enhance their antigen presenting and myelin‐specific T‐cell stimulatory capacity in vitro 5, 10. Furthermore, isolated rat microglia exposed to myelin have been described to develop a neurotoxic phenotype with an increased inducible nitric oxide synthase, tumor necrosis factor‐α and glutamate expression 67.

Microarray analysis revealed the transcription of several genes participating in the interferon pathway predominately during the demyelinating phase (Table S3, group III). In TME, microglia/macrophages activated by virus or IFN‐γ enhance immune‐mediated tissue damage by presenting viral antigens and endogenous myelin epitopes to CD4⁺ T cells, which induces delayed‐type hypersensitivity and autoimmunity, respectively 7, 16, 33, 68. Moreover, beside its protective antiviral function, IFN‐γ increases the migration of macrophages and microglial activation, which induces myelinotoxic substances and free radicals causing progressive myelin loss (bystander demyelination) in TME 45, 58, 86, 87. IFN‐γ is the main cytokine associated with M1 activation of microglia and macrophages 34, 69 and has been shown in TMEV‐infected mice to be produce by CD4⁺ and CD8⁺ T cells 63. Noteworthy, demonstration of the cytokine in CD107b⁺ cells in the present study is indicative also of an autocrine regulation of M1 polarization as demonstrated in endotoxin‐stimulated macrophages 76.

Despite mounting M2 polarization and the expression of regeneration promoting factors, such as insulin‐like growth factor‐1 (igf1) and transforming growth factor‐β (Tgfb1) 24, 41, 91, CNS recovery remains abortive and only insufficient remyelination attempts by oligodendrocytes and Schwann cells were found in the spinal cord during the late chronic TME phase. Similar to the present observation, macrophages/microglia with both M1 and M2 properties can be found in active demyelinating MS brain lesions 90. Recent studies have demonstrated that the switch of M1‐ into M2‐type cells is required for efficient oligodendrocyte differentiation and myelin repair following toxin‐induced demyelination in rodents and that M2‐conditioned media drive oligodendrocyte maturation in vitro 60. In addition, M2‐type macrophages/microglia protect from EAE through deactivation of encephalitogenic Th1 and Th17 cells 71. Consequently, continuous M1 polarization observed until the late chronic phase (196 dpi) in TMEV‐infected mice has the potential to antagonize neuroprotective effects of M2 microglia/macrophages. In agreement with previous reports 39, 95, CD107⁺ microglia/macrophages represent a target for virus infection in the present study. Because TMEV has been demonstrated to preferentially infect activated myeloid cells with M1 characteristics, such as CD16/32 and IFN‐γ expression, in vitro 29, 30, it is also tempting to speculate that prolonged M1 polarization contributes to viral persistence in susceptible mouse strains by providing permissive target cells for TMEV. In addition, genes have been identified by the present microarray analysis that might be involved in disturbed viral elimination by influencing the interferon pathway (Table S3, group III). For instance, OASL1, a recently defined type I interferon negative regulator and translation inhibitor of IRF7, is differentially upregulated in TMEV‐infected mice. OASL1 causes T‐cell suppression in persistent lymphocytic choriomeningitis virus infection of mice and is regarded as a new target for preventing chronic infectious diseases 43, 44. In agreement with this idea, subpopulations of CNS‐infiltrating macrophages have been demonstrated to reduce protective antiviral immunity by inducing T‐cell exhaustion that leads to virus persistence in TMEV‐infected mice 31. Besides this, M2‐polarized cells have the ability to reduce antiviral immunity as described for human cytomegalovirus infection 1.

In conclusion, the perpetuating interaction between virus and macrophages/microglia induces a vicious circle with continuous inflammation and impaired myelin repair in the spinal cord of TMEV‐infected mice. The present findings support the hypothesis of a dual function of either polarized cells with promoting effects upon antiviral immunity and immunopathology, respectively, in TME. Hence, in contrast to the therapeutic effect of M2 dominance in primary autoimmune diseases, such as EAE, only a well‐orchestrated and timely balanced polarization of macrophages/microglia might have the ability to prevent virus persistence and reduce myelin loss in this infectious MS model.

Supporting information

Figure S1. Detection of Theiler's murine encephalomyelitis virus in the murine spinal cord by immunohistochemistry. (A) Quantification of infected cells at different time points. TMEV = Theiler's murine encephalomyelitis virus‐infected mice; mock = mock‐infected control mice; dpi = days postinfection; * = significant difference (P ≤ 0.05, Mann–Whitney U‐test). Box and whisker plots display median and quartiles with maximum and minimum values. (B) Note virus‐specific labeling (brownish signal) in the spinal cord white matter of an infected mouse at 98 dpi. Scale bar = 200 μm; insert; scale bar = 50 μm.

Figure S2. Phenotyping of Theiler's murine encephalomyelitis virus‐infected cells (TMEV) by immunofluorescence double staining. (A) Percentage of infected cells representing CD107b⁺ macrophages/microglia at different days postinfection (dpi). Columns display median with maximum and minimum values. (B) Co‐localization of TMEV (green) and CD107b (red) in an inflammatory spinal cord lesion at 98 dpi. Double‐stained cells exhibit a yellow color. Nuclei are stained with bisbenzimide (blue). Scale bars = 50 μm.

Figure S3. Detection of chemokines and interferon‐γ (IFN‐γ) in CD107b⁺ macrophages/microglia by immunofluorescence double staining. Percentage of (A) CCL5‐, (B) CXCL10‐ and (C) IFN‐γ‐positive cells co‐expressing CD107b at 42 and 196 days postinfection (dpi). Columns display median with maximum and minimum values.

Figure S4. Detection of chemokines and interferon‐γ (IFN‐γ) in CD107b⁺ macrophages/microglia in the spinal cord by immunofluorescence double staining. Top row: co‐localization of CCL5 (green) and CD107b (red) in an infected mouse at 196 days postinfection (dpi). Middle row: co‐localization of CXCL10 (green) and CD107b (red) in an infected mouse at 196 dpi. Bottom row: co‐localization of IFN‐γ (green) and CD107b (red) in an infected mouse at 42 dpi. Double‐stained cells exhibit a yellow color (right column). Nuclei are stained with bisbenzimide (blue). Scale bars = 50 μm.

Table S1. List of selected M1‐ and M2‐related genes.

Table S2. Ultrastructural changes in the spinal cord of Theiler's murine encephalomyelitis virus‐infected SJL mice.

Table S3. Differentially expressed M1‐ and M2‐related genes in the spinal cord of Theiler's murine encephalomyelitis virus‐infected mice‐group I: chemotaxis, group II (phagocytosis, antigen processing and presentation), group III: cytokine signaling and growth factors, group IV: toll‐like receptor signaling, group V: apoptosis‐related genes, group VI: extracellular matrix receptor interaction and cell adhesion molecules, group VII: miscellaneous genes.