Distinct Spatio‐Temporal Extracellular Matrix Accumulation within Demyelinated Spinal Cord Lesions in Theiler's Murine Encephalomyelitis

Abstract

The accumulation of extracellular matrix (ECM) and glial scar formation are considered important factors for the failure of regeneration in central nervous system (CNS) injury and multiple sclerosis. Theiler's murine encephalomyelitis (TME) as a model of multiple sclerosis served to evaluate the spatio‐temporal course of ECM alterations in demyelinating conditions. Microarray analysis revealed only mildly upregulated gene expression of ECM molecules, their biosynthesis pathways and pro‐fibrotic factors, while upregulation of matrix remodeling enzymes was more prominent. Immunohistochemistry demonstrated progressive accumulation of chondroitin sulfate proteoglycans, glycoproteins and collagens within demyelinated TME lesions, paralleling the development of astrogliosis. Deposition of collagen IV, laminin, perlecan and tenascin‐C started 28 days postinfection (dpi), collagen I, decorin, entactin and neurocan accumulated from 56 dpi on, and fibronectin from 98 dpi on. The basement membrane (BM) molecules collagen IV, entactin, fibronectin, laminin and perlecan showed perivascular and parenchymal deposition, while the non‐BM components collagen I, decorin, neurocan and tenascin‐C only accumulated in a nonvascular pattern in demyelinated areas. Contrary, phosphacan expression progressively decreased during TME. The immunoreactivity of aggrecan and brevican remained unchanged. The spatio‐temporal association of matrix accumulation with astrogliosis suggests a mainly astrocytic origin of ECM deposits, which in turn may contribute to remyelination failure in TME.

INTRODUCTION

The infection of susceptible mouse strains with low neurovirulent strains of Theiler's murine encephalomyelitis virus (TMEV) leads to demyelinating lymphohistioplasmacytic leukomyelitis, which pathogenetically is in part attributed to a delayed‐type hypersensitivity reaction 9, 62. Chronic demyelinated lesions of Theiler's murine encephalomyelitis (TME) are characterized by mild inflammation and an increased number of activated astrocytes whereas remyelination processes are insufficient. These findings closely resemble the morphologic changes in other demyelinating diseases such as experimental allergic encephalomyelitis (EAE) in rats and mice as well as chronic progressive multiple sclerosis (MS) in humans 44, 50, 53. Failure of successful remyelination in MS, considered to be a main factor for the progressive clinical course, is largely based on the lack of mature myelinating oligodendrocytes within the MS plaques (60). This parallels recent findings of impaired oligodendroglial differentiation in chronic TME lesions (53). Intralesional accumulation of the extracellular matrix molecule hyaluronan was shown to inhibit oligodendrocyte maturation in MS whereas the respective mechanism in TME remains unclear so far 3, 53. However, in general, various components of the extracellular matrix (ECM) have a manifold impact on central nervous system (CNS) regeneration as demonstrated in different spontaneous and experimental conditions. Activation of resident glial and meningeal cells leads to upregulation of their matrix secretion and subsequent intralesional accumulation of ECM, resulting in the so‐called glial scar, which is considered to hamper regeneration substantially (8). Accumulating scar matrix is biochemically dominated by chondroitin sulfate proteoglycans (CSPGs) such as aggrecan, brevican, decorin, neurocan and phosphacan, as well as collagen I and the glycoprotein tenascin‐C. According to their physiological tissue distribution, these aforementioned molecules can be grouped as parenchymal or non‐basement membrane (BM) ECM molecules. Additionally, explicit BM components including collagen IV, entactin, fibronectin and perlecan are involved in the formation of a glial scar, where they are secreted by both endothelial cells and perivascular astrocytes. As these BM molecules contribute to blood–brain barrier (BBB) integrity and permeability, changes in their molecular composition or expression level influence the infiltration of inflammatory cells 37, 55. Rodent models of spinal cord injury served to discover a distinct spatio‐temporal course of ECM synthesis and deposition following tissue damage. They thereby allowed to specify the impact of particular molecules to the overall properties of the glial scar and to modify scar formation with regard to future therapeutic approaches 17, 18, 20, 22, 43, 46. Similarly in demyelinated MS plaques, different molecular patterns of ECM deposition were observed despite upregulation of important matrix remodeling enzymes such as matrix metalloproteinases (MMPs) 16, 25, 30, 40, 42, 48, 58. Thus, a crucial influence of ECM alterations on disease development must be assumed and dealt with when developing new therapeutic approaches. However, a detailed decoding of temporal or spatial processes during glial scar formation remains difficult in MS because of the spontaneous and unpredictable nature of this disease and the shortage of samples from early lesions. In contrast, TME features a progressive demyelination with intralesionally increased numbers of activated putatively ECM secreting astrocytes, concurrent elevation of MMP levels and only insufficient remyelination. Thus, TME represents a very suitable model to elucidate spatio‐temporal aspects of glial scar formation in chronic inflammatory and demyelinating CNS lesions. Considering recently published data suggesting a contribution of the BBB to TME pathology (32), as well as the morphological features previously mentioned, we hypothesize that chronic demyelinating TME is accompanied by profound alterations in spinal cord ECM and that these in turn contribute to a failure of regeneration. Therefore, the aims of the present study were twofold: (i) to characterize the composition and distribution of spinal cord ECM during TMEV infection; and (ii) to identify underlying molecular mechanisms and transcriptional changes.

MATERIALS AND METHODS

Animal experiments

Female SJL/JHanHsd mice (Harlan Winkelmann, Borchen, Germany) were used in this study. At the age of 5 weeks, the animals were inoculated into the right cerebral hemisphere with 1.63 × 10⁶ plaque‐forming units/mouse of the BeAn strain of TMEV as described (53). Mock‐infected mice (placebos) served as a control for all experiments. Groups of five to six TMEV‐infected and mock‐infected mice were killed 1 h (0 days) and 7, 14, 28, 42, 56, 98 and 196 days post infection (dpi) by administering an overdose of medetomidine and ketamine (53). The vertebral column was immediately removed and dissected transversally. Segments of the cervical, thoracic and lumbar vertebral cord were fixed in 10% formalin for 24 h, decalcified in 10% ethylenediaminetetraacetic acid disodium solution (EDTA) for 48 h, and embedded in paraffin wax. The remaining spinal cord was removed via complete laminectomy, immediately embedded in Tissue‐Tec® O.C.T.™ (Sakura, Zoeterwoude, The Netherlands), snap frozen and stored at −80°C (53). Frozen and formalin‐fixed tissue from 0, 7, 14, 28, 56, 98 and 196 dpi was used for histochemical and immunohistochemical examinations. For RNA isolation and microarray hybridization, frozen tissue from 14, 42, 98 and 196 dpi was used as described (54). All animal experiments were approved by the local authorities (Bezirksregierung Hannover, Germany, permission number 509c‐42502‐02/589, and Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit, Oldenburg, Germany, permission number 33‐42502‐05/963).

Histochemical examination and light microscopic evaluation

Serial transversal sections of formalin‐fixed, paraffin‐embedded thoracic spinal cord were stained with hematoxylin and eosin as well as luxol fast blue‐cresyl violet (LFB). Furthermore, histochemical stainings such as azan stain (27) and a modified picrosirius red (PSR) stain 10, 15 were applied to detect extracellular substances. For the latter, briefly, sections were deparaffinized and rehydrated. After dipping the sections in 3% acetic acid, the specimens were stained in 1% alcian blue solution pH 2.5 (1 g alcian blue 8GX in 100 mL 3% acetic acid supplemented by 1 crystal of thymol) for 30 minutes. Thereafter, sections were dipped repeatedly in 70% ethanol, and were then stained in an aldehyde fuchsin solution for 45 minutes. Following dipping in 70% ethanol and rinsing in running tap water, the specimens were stained in Weigert's hematoxylin, consisting of ferric chloride solution and 1% hematoxylin in absolute alcohol in equal amounts, for 10 minutes. After rinsing again in running tap water, sections were stained in 0.1% PSR solution (0.1 g sirius red F3B combined with 100 mL saturated aqueous picric acid) for 30 minutes. Finally, slides were dipped in 0.01% hydrochloric acid for 30 s, followed by dehydration and mounting. This staining procedure resulted in a purple color of sulfated mucosubstances like proteoglycans and glycoproteins, a blue labeling of carboxylated mucosubstances like hyaluronan, and a red decoration of collagens. The latter additionally exhibited yellow to red birefringency under polarized light (35).

For characterization of TMEV‐induced lesions, hematoxylin and eosin as well as LFB‐stained sections were evaluated by assessing the degree of meningitis, myelitis and demyelination semi‐quantitatively as described (52). The extent of histochemical staining reaction was determined morphometrically. Slides were digitalized with a color video camera (Color View II, 3.3 megapixel CCD; Soft Imaging System, Münster, Germany) mounted on an Axiophot microscope (Zeiss, Oberkochen, Germany) with 2.5× objective. The positive area for each staining method was measured after manually outlining the total white matter area using the analySIS 3.1 software (Soft Imaging System). Results were given as percentage calculated as the positively labeled area divided by the total white matter area.

RNA isolation and microarray hybridization

RNA was isolated from snap‐frozen spinal cord tissue using the RNeasy Mini Kit (Qiagen, Hilden, Germany), amplified and labeled with the MessageAmp II‐Biotin Enhanced Kit (Ambion, Austin, TX, USA), and microarray hybridization was performed using the Affymetrix mouse genome 430 2.0 array as described previously (54). MIME compliant data sets have been published in the ArrayExpress database (http://www.ebi.ac.uk/microarray‐as/ae/). Background adjustment and normalization of the microarray data were performed employing the RMA algorithm 5, 54.

Selection of ECM‐associated genes

Further analysis focused on a list of candidate genes that are considered to be associated with synthesis and degradation of ECM or fibrosis. The complete list comprised 350 probe sets representing 170 genes (Supporting Information Table S1) with 67 genes directly coding for ECM molecules. Sixty‐six enzymes belonging to the synthesis or degradation cascades of ECM molecules were extracted from the KEGG PATHWAY database (21) and published literature (34). These comprised 18 genes involved in chondroitin sulfate biosynthesis, 14 genes for heparan sulfate biosynthesis, 14 genes for keratan sulfate biosynthesis, 3 genes for hyaluronan synthases and 17 genes coding for enzymes involved in glycosaminoglycan degradation. Additionally, the list contained 24 genes of ECM modifying enzymes comprising 20 MMPs and four tissue inhibitors of metalloproteinases (TIMPs). Finally, 13 genes coding for pro‐ and anti‐fibrotic factors with known importance in CNS tissue were enclosed 36, 38, 47, 61.

Immunohistochemical examination

In order to validate the obtained microarray results and to specify the spatio‐temporal course of ECM deposition, immunohistochemistry was performed. The antibodies used for immunohistochemistry are listed in Table 1. Detection of glial fibrillary acidic protein (GFAP), aggrecan, collagen I and IV, decorin, fibronectin and laminin was conducted on formalin‐fixed, paraffin‐embedded sections employing an avidin‐biotin‐peroxidase complex method as described 52, 53. Briefly, sections were dewaxed and rehydrated through graded alcohols. Endogenous peroxidase was quenched with 0.5% hydrogen peroxide in methanol. For detection of collagen I and collagen IV, sections were heated in 10 mM sodium citrate buffer pH 6.0 for 15 minutes in a microwave oven (800 W). For demonstration of aggrecan, an enzymatic pretreatment was performed by using chondroitinase ABC from Proteus vulgaris (Sigma‐Aldrich Chemie GmbH, Taufkirchen, Germany; 0.5 U/mL in 0.05 M TRIS buffer pH 8.0) for 90 minutes at 37°C (7). Sections were incubated in goat or rabbit normal serum diluted 1:5 in phosphate‐buffered saline (PBS) for 20 minutes to block nonspecific binding sites before incubation with the primary antibody at 4°C overnight. Secondary antibodies were diluted 1:200 in PBS and applied for 30 minutes, unless stated otherwise in Table 1. Finally, sections were incubated with the avidin‐biotin‐peroxidase‐complex (ABC; Vector Laboratories, Inc., Burlingame, CA, USA) for 30 minutes. Positive antigen‐antibody reactions were visualized by incubation in 3,3‐diaminobenzidine‐tetrahydrochloride (DAB) and 0.03% hydrogen peroxide in PBS for 10 minutes followed by slight counterstaining with Mayer's hemalum.

Table 1.

Source and immunogen of primary antibodies, pretreatment, blocking serum and secondary antibodies used for immunohistochemical detection of GFAP, extracellular matrix molecules and von Willebrand factor. Abbreviations: Dako ARK = Dako Animal Research Kit™ Peroxidase (DakoCytomation GmbH, Hamburg, Germany); GFAP = glial fibrillary acidic protein; GaR‐b = goat anti‐rabbit immunoglobulins biotinylated (Vector Laboratories, Inc., Burlingame, CA, USA); NGS = normal goat serum; NRS = normal rabbit serum; RaG‐PO = rabbit anti‐goat immunoglobulins peroxidase‐conjugated (DakoCytomation GmbH); RaR‐b = rabbit anti‐rat immunoglobulins biotinylated (Vector Laboratories, Inc.); PBS = phosphate‐buffered saline.

Primary antibody	Source of primary antibody; catalog no.	Immunogen	Tissue fixation	Pretreatment*	Blocking serum†	Dilution of primary antibody§	Secondary antibody‡
Polyclonal rabbit anti‐GFAP	Dako‐Cytomation GmbH, Hamburg, Germany; # Z0334	GFAP isolated from bovine spinal cord	Formalin	None	NGS	1:2000	GaR‐b
Polyclonal rabbit anti‐aggrecan	Chemicon Europe Ltd., Hampshire, UK; # AB1031	GST fusion protein containing AA 1177–1326 of mouse aggrecan	Formalin	Chondroitinase digestion	NGS	1:400	GaR‐b
Monoclonal mouse anti‐brevican, clone 2	BD Biosciences, Heidelberg, Germany; # 610894	Amino acids 232–394 of rat brevican	Formalin	Citrate buffer, microwave heating	None	1:200	Dako ARK
Polyclonal rabbit anti‐murine collagen I	Abcam Ltd., Cambridge, UK; # ab21286	Collagen I extracted and purified from mouse skin	Formalin	Citrate buffer, microwave heating	NGS	1:300	GaR‐b
Polyclonal rabbit anti‐murine collagen IV	Abcam Ltd.; # ab19808	Collagen IV extracted and purified from mouse tumor tissues	Formalin	Citrate buffer, microwave heating	NGS	1:200	GaR‐b
Polyclonal goat anti‐murine decorin	R&D Systems GmbH, Wiesbaden, Germany; # AF1060	NS0‐derived recombinant mouse decorin	Formalin	None	NRS	1:200	RaG‐PO¶
Monoclonal rat anti‐entactin, clone ELM1	Neomarkers, Fremont, CA, USA; # RT‐797‐P0	Murine EHS laminin preparation	Frozen	None	NRS	1:2000	RaR‐b
Polyclonal rabbit anti‐human fibronectin	Sigma‐Aldrich Chemie GmbH, Taufkirchen, Germany; # F3648	Purified human fibronectin	Formalin	None	NGS	1:1000	GaR‐b
Polyclonal rabbit anti‐laminin	Sigma‐Aldrich Chemie GmbH; # L9393	Laminin purified from basement membrane of EHS mouse sarcoma	Formalin	None	NGS	1:100	GaR‐b
Monoclonal mouse anti‐neurocan; clone 650.24	Chemicon Europe Ltd.; # MAB5212	Not specified	Formalin	None	None	1:50	Dako ARK
Monoclonal rat anti‐heparan sulphate proteoglycan large (perlecan), clone A7L6	Abcam Ltd.; # ab2501	High molecular mass material derived from EHS mouse sarcoma matrix	Frozen	None	NRS	1:1000	RaR‐b
Monoclonal mouse anti‐phosphacan, clone 122.2	Chemicon Europe Ltd.; # MAB5210	Not specified	Formalin	None	None	1:40	Dako ARK
Monoclonal rat anti‐tenascin C, clone MTn‐12	Abcam Ltd.; # ab6346	Purified full‐length murine tenascin‐C	Frozen	None	NRS	1:8000	RaR‐b
Polyclonal rabbit anti‐von Willebrand factor	Dako‐Cytomation GmbH; # A0082	Von Willebrand factor isolated from human plasma	Formalin	Citrate buffer, microwave heating	NGS	1:500	GaR‐b

^* Performance of pretreatment, see Materials and Methods.

^† Diluted 1:5 in PBS.

^‡ Diluted 1:200 in PBS, incubation for 30 minutes at room temperature, unless otherwise stated.

^§ Diluted in PBS containing 1% bovine serum albumin.

^¶ Diluted 1:100 in PBS, incubation for 45 minutes at room temperature.

Frozen sections were used for demonstration of entactin, perlecan and tenascin‐C. Slides were hydrated in PBS for 5 minutes and afterwards endogenous peroxidase was blocked with 0.5% hydrogen peroxide in PBS. The following steps were performed as described above.

Brevican, neurocan and phosphacan were detected employing the Dako Animal Research Kit™ Peroxidase (DakoCytomation GmbH, Hamburg, Germany) according to the manufacturer's recommendations (53).

As negative controls, primary antibodies were replaced by control serum or isotype‐specific antibodies directed against irrelevant antigens, respectively. Quantitative evaluation of ECM molecules was performed morphometrically as previously described for histochemically stained sections. The density of GFAP‐positive cells/mm² was manually counted in the white matter using a calibrated 10‐fold reticle eyepiece and a 40‐fold objective (Olympus Europe, Hamburg, Germany). In every spinal cord section, randomly selected areas with a mean total area of 0.15 mm² were evaluated for normal appearing white matter (NAWM) and lesioned white matter, respectively.

Immunohistochemical double‐labeling

In order to clarify the spatial relationship of BM molecule deposits and blood vessels, a double immunostaining for collagen IV and the endothelial cell marker von Willebrand factor (formerly factor VIII‐related antigen) was performed on selected formalin‐fixed, paraffin‐embedded sections from TMEV‐infected animals. Briefly, sections were dewaxed and rehydrated through graded alcohols, and endogenous peroxidase was quenched with 0.5% hydrogen peroxide in methanol. Then, sections were heated in 10 mM sodium citrate buffer pH 6.0 for 15 minutes in a microwave oven (800 W). After incubation in goat normal serum diluted 1:5 in PBS for 20 minutes to block nonspecific binding sites, the sections were incubated with anti‐von Willebrand factor immunoglobulins at 4°C overnight. The secondary antibody according to Table 1 was diluted 1:200 in PBS and applied for 30 minutes. Sections were incubated with ABC for 30 minutes, and positive antigen–antibody reactions were visualized by incubation in DAB and 0.03% hydrogen peroxide in PBS for 10 minutes. Afterwards, anti‐collagen IV immunoglobulins were applied at a dilution of 1:200 for 1 h at room temperature, followed by incubation with the secondary antibody and ABC as previously described. The collagen IV immunostaining was visualized by using HistoGreen (Linaris GmbH, Wertheim‐Bettingen, Germany).

Statistical analysis

Statistical analysis was performed using SPSS for Windows version 13.0 (SPSS Inc., Chicago, IL, USA). Regarding the microarray data, the gene expression values of each individual animal were used. Histochemical and immunohistochemical data were analyzed using the percentage values of the morphometric evaluation and the density values of cell counting. To all data, a time point specific Mann–Whitney U‐test was applied for detection of differences between mock‐ and TMEV‐infected mice. For comparison of cell density values in NAWM and lesions of TMEV‐infected mice, a Wilcoxon test was employed. Statistical significance was designated as P < 0.05 in all tests performed.

RESULTS

Morphology of TMEV‐induced spinal cord lesions

TMEV‐infected mice showed lymphohistioplasmacytic meningitis with multifocal to coalescing perivascular infiltrates from 7 dpi to 196 dpi, with highest severity at 28 dpi and 98 dpi. In addition, a multifocal to coalescing predominantly perivascular lymphohistioplasmacytic infiltration in spinal cord white matter was observed between 7 dpi and 196 dpi, reaching a peak at 98 dpi (Figure 1A,E). As the first sign of demyelination, scattered foci of myelinophagia were detected as early as 14 dpi. LFB revealed a progressive loss of myelin during TME lesion development between 28 dpi and 196 dpi. Demyelinated lesions were mainly present in the ventral and lateral funiculi of the spinal cord, and were not detected in any of the mock‐infected animals (Figure 1B,F).

Figure 1.

Histochemical findings after Theiler's murine encephalomyelitis virus (TMEV) infection. Serial sections of thoracic spinal cord of TMEV‐infected mice from 56 days postinfection (dpi; A–D) and 98 dpi (E–H) representing hematoxylin and eosin stain (A, E), luxol fast blue‐cresyl violet (LFB) stain (B, F), azan stain (C, G) and picrosirius red (PSR) stain (D, H). (A) At 56 dpi, moderate lymphohistiocytic meningitis and leukomyelitis are present with vacuolation of the white matter including loss of eosinophilia (asterisk), and myelinophagia. (B) LFB stain reveals moderate demyelination of the lesion (asterisk). (C) In the center of the lesion, there is a minimal accumulation of extracellular azan‐positive matrix (arrowhead). (D) PSR stain reveals a diffuse bluish to greenish extracellular labeling at the periphery (asterisk) and a red filamentous staining of extracellular matrix in the center of the lesion (arrowhead) at 56 dpi. (E) At 98 dpi, the white matter lesions exhibit only mild mononuclear inflammation but increased cellularity indicating gliosis (asterisk) with peripheral dilatation of myelin sheaths. (F) LFB stain demonstrates an almost complete loss of myelin in the white matter lesion (asterisk) with sparse preserved myelin at the periphery (arrowhead). (G) There is a dense meshwork of azan‐positive extacellular matrix deposits throughout the lesion. (H) Paralleling the azan labeling, PSR stain demonstrates large amounts of accumulating extracellular matrix. Scale bars = 100 µm.

GFAP immunohistochemistry for detection of astrocytes revealed an unchanged number of astrocytes and their processes in the gray and white matter of the spinal cord in mock‐infected mice. At 28 days after TMEV infection, intralesional astrocytes appeared activated, characterized by elongation and thickening of their processes. At subsequent time points, the number of GFAP‐positive astrocytes and the amount of their processes continuously increased (Figure 2A,E,I). Thus, there was a significantly higher density of GFAP‐positive cells within TMEV lesions at 98 dpi and 196 dpi compared with NAWM and mock‐infected mice (Figure 3).

Figure 2.

Immunohistochemical findings after Theiler's murine encephalomyelitis virus (TMEV) infection. Serial sections of thoracic spinal cord of TMEV‐infected mice from 56 days postinfection (dpi; A–D, same lesion as in Figure 1A–D), 98 dpi (E–H, same lesion as in Figure 1E–H; N–P, different animal) and 196 dpi (I–M), showing immunoreactivity for glial fibrillary acidic protein (GFAP; A, E, I), collagen IV (B, F), laminin (C, G), collagen I (D, H), fibronectin (J), decorin (K), neurocan (L), phosphacan (M), perlecan (N), entactin (O) and tenascin‐C (P). (A) At 56 dpi, intralesional GFAP‐positive cells show a reactive phenotype with thickened processes (asterisk). (B) Mild deposits of nonvessel‐associated collagen IV (arrowhead) are present in the demyelinated white matter lesion. (C) Laminin immunoreactivity is detected in minimal amounts (arrowhead) at 56 dpi. (D) Mild collagen I accumulation occurs in a lattice‐like pattern (arrowhead) in the center of the lesion. (E) At 98 dpi, an increased number of GFAP‐positive astrocytes is present in the demyelinated white matter lesion (asterisk). (F) Marked collagen IV deposits are detected in this lesion in a diffuse parenchymal distribution pattern at 98 dpi. (G) At 98 dpi, laminin is prominently accumulated throughout the demyelinated area. (H) A comparable, diffuse collagen I immunoreactivity is present intralesionally. (I) The white matter lesions contain a prominently increased number of GFAP‐positive cells with thickened processes at 196 dpi (asterisk). (J) Fibronectin immunoreactivity occurs diffusely throughout the lesion (asterisk) at 96 dpi. (K) A dense meshwork of decorin deposits is present intralesionally at 196 dpi. (L) Mild intralesional neurocan deposition occurs in a circular pattern around capillaries (arrowhead) at 196 dpi. (M) The prominent phosphacan expression in normal white matter is markedly reduced in a demyelinated lesion (asterisk) at 196 dpi. (N) At 98 dpi, prominent parenchymal and vessel‐associated perlecan deposits are present in a demyelinated white matter lesion. (O) Entactin is markedly accumulated throughout the same lesion with and without blood vessel association. (P) A lattice‐like pattern of tenascin‐C immunoreactivity is detected in the demyelinated area. Immunohistology, ABC method, DAB chromogen, slight counterstain with Mayer's hemalum. Nomarski differential interference contrast. Scale bars = 100 µm.

Figure 3.

Density of astrocytes in the thoracic spinal cord of mice after Theiler's murine encephalomyelitis virus (TMEV) infection and mock infection, shown as glial fibrillary acidic protein (GFAP)‐positive cells per square millimeter of normal appearing white matter (NAWM; mock‐ and TMEV‐infected mice) and demyelinated white matter areas (lesion; TMEV‐infected mice). Box and whisker plots show median and quartiles with outliers and extreme values depicted as circles and filled circles, respectively. Following TMEV infection, the density of intralesional GFAP‐positive cells (astrocytes) was significantly increased at 98 dpi and 196 dpi compared with NAWM of TMEV‐ and mock‐infected mice. Significant differences between mock‐ and TMEV‐infected mice as revealed by a Mann–Whitney U test, and between NAWM and lesion in TMEV‐infected mice as detected by a Wilcoxon test are marked as follows: *P < 0.05.

Progressive deposition of ECM in TME lesions

Histochemical stains served to estimate the amount and distribution of ECM in normal and TME‐affected spinal cord. Within the spinal cord of mock‐infected mice, positive staining obtained by azan and PSR stain was restricted to meninges and blood vessel walls, and did not show any age‐related modifications until 196 dpi (Table 2). During the course of TME, prominent changes of the spinal cord ECM were observed, generally restricted to demyelinated white matter areas. Using azan stain, minimal amounts of ECM deposition were detected within demyelinated areas at 28 dpi following TMEV infection, characterized by a fine filamentous reaction pattern in the lesion center. With lesion progression, azan‐positive ECM accumulated intralesionally in a dense lattice‐like pattern (Figure 1C,G), frequently extending to the edge of the demyelinated area and resulting in a median of 1.47% positive area of total white matter area at 196 dpi, which differed significantly from 0.004% to 0.03% measured in mock‐infected animals (details in Table 2). With the modified PSR stain, two different staining patterns and colors were observed. As early as 28 dpi, a diffuse light bluish to greenish signal occurred within demyelinated lesions, occupying a median of 0.46% of total white matter area and indicating the presence of nonsulfated mucosubstances. At subsequent time points, this bluish reaction was only present at the lesions' margins or completely disappeared (Figure 1D). An additional red reticular staining indicative of collagens, proteoglycans and glycoproteins first appeared at 56 dpi in the lesions' center (Figure 1D,H). Progressive expansion of this red signal resulted in a median area percentage of 2.56% at 196 dpi with maximal values up to 3.7% of total white matter area (Table 2). Yellow to red birefringence of the red‐labeled structures using polarized light documented the contribution of collagen fibers to the accumulating matrix.

Table 2.

Morphometrically determined area of extracellular matrix molecule deposition in the thoracic spinal cord of mock‐ and TMEV‐infected mice. Abbreviations: PSR = picrosirius red stain; Mock = mock‐infected mice; TMEV = Theiler's murine encephalomyelitis virus‐infected mice; dpi = days post infection.

Histochemical stain or immunohistochemically detected molecule	Positive area per total white matter area [%]
	0 dpi			7 dpi
	Mock	TMEV	P‐value	Mock	TMEV	P‐value
Azan	0.032 (0.010;0.091)†	0.010 (0.003;0.015)	0.082	0.013 (0.005;0.034)	0.013 (0.006;0.022)	1.000
PSR, blue color	0.000 (0.000;0.002)	0.013 (0.000;0.099)	0.082	0.002 (0.000;0.017)	0.017 (0.005;0.026)	0.167
PSR, red color	0.025 (0.000;0.084)	0.000 (0.000;0.021)	0.126	0.011 (0.020;0.019)	0.003 (0.000;0.009)	0.167
Aggrecan	0.004 (0.000;0.011)	0.007 (0.000;0.122)	0.931	0.017 (0.000;0.026)	0.023 (0.007;0.066)	0.548
Brevican	0.009 (0.000;0.033)	0.008 (0.000;0.021)	1.000	0.014 (0.005;0.023)	0.011 (0.010;0.014)	0.381
Collagen I	0.191 (0.046;0.589)	0.100 (0.025;1.430)	0.792	0.175 (0.102;0.252)	0.060 (0.053;1.474)	0.548
Collagen IV	1.990 (0.736;3.282)	1.710 (1.307;4.073)	0.762	1.249 (0.559;2.890)	1.154 (0.691;1.132)	0.714
Decorin	0.167 (0.097;0.527)	0.145 (0.041;0.585)	0.792	0.236 (0.102;0.407)	0.127 (0.055;0.312)	0.381
Entactin	2.484 (1.236;3.101)	2.212 (1.537;2.628)	0.699	2.460 (1.699;3.718)	2.030 (1.552;3.213)	0.413
Fibronectin	0.023 (0.013;0.054)	0.078 (0.000;0.146)	0.662	0.085 (0.000;0.216)	0.028 (0.022;0.068)	0.381
Laminin	0.053 (0.019;0.177)	0.054 (0.009;0.119)	1.000	0.149 (0.031;0.208)	0.152 (0.011;0.169)	0.548
Neurocan	0.102 (0.031;0.270)	0.188 (0.103;0.297)	0.082	0.066 (0.018;0.157)	0.126 (0.101;0.155)	0.262
Perlecan	0.422 (0.205;0.881)	0.537 (0.229;1.115)	0.485	0.577 (0.225;0.751)	0.390 (0.214;0.778)	0.686
Phosphacan	99.972 (99.869;99.992)	99.884 (99.814;99.994)	0.429	99.962 (99.834;99.966)	99.953 (99.862;99.984)	1.000
Tenascin‐C	0.366 (0.170;0.463)	0.336 (0.091;0.551)	1.000	0.374 (0.211;0.471)	0.320 (0.054;0.488)	0.905

Histochemical stain or immunohistochemically detected molecule	Positive area per total white matter area [%]
	14 dpi			28 dpi
	Mock	TMEV	P‐value	Mock	TMEV	P‐value
Azan	0.004 (0.003;0.032)	0.015 (0.000;0.038)	0.662	0.015 (0.004;0.024)	0.040 (0.000;0.093)	0.429
PSR, blue color	0.008 (0.000;0.045)	0.006 (0.003;0.011)	0.548	0.011 (0.000;0.082)	0.463 (0.003;0.656)	0.052
PSR, red color	0.013 (0.000;0.062)	0.000 (0.000;0.017)	0.222	0.033 (0.004;0.050)	0.016 (0.000;0.094)	0.662
Aggrecan	0.021 (0.000;0.057)	0.036 (0.000;0.089)	0.662	0.013 (0.000;0.034)	0.005 (0.000;0.025)	0.537
Brevican	0.008 (0.000;0.023)	0.000 (0.000;0.004)	0.171	0.009 (0.000;0.029)	0.005 (0.000;0.015)	0.429
Collagen I	0.510 (0.071;0.894)	0.289 (0.022;6.405)	0.841	0.162 (0.096;0.337)	0.201 (0.176;0.266)	0.171
Collagen IV	1.147 (0.278;2.634)	2.976 (1.398;5.525)	0.114	1.127 (0.263;2.308)	3.485 (2.010;5.245)	0.009*
Decorin	0.363 (0.032;0.466)	0.393 (0.071;0.487)	0.931	0.094 (0.034;0.109)	0.188 (0.040;0.219)	0.247
Entactin	2.684 (1.897;3.338)	2.531 (1.408;3.422)	0.818	2.905 (1.564;3.444)	2.471 (1.294;2.913)	0.180
Fibronectin	0.032 (0.003;0.110)	0.026 (0.000;0.175)	1.000	0.447 (0.000;0.088)	0.027 (0.008;0.244)	0.792
Laminin	0.050 (0.017;0.200)	0.111 (0.006;0.288)	0.690	0.044 (0.016;0.124)	0.169 (0.006;0.543)	0.329
Neurocan	0.091 (0.017;0.175)	0.059 (0.033;0.122)	0.662	0.119 (0.062;0.153)	0.085 (0.058;0.159)	0.792
Perlecan	0.390 (0.103;0.685)	0.286 (0.081;0.975)	0.931	0.439 (0.216;0.786)	0.461 (0.224;2.411)	0.699
Phosphacan	99.960 (99.921;99.996)	99.982 (99.865;99.998)	1.000	99.891 (99.670;99.977)	99.540 (99.191;99.971)	0.177
Tenascin‐C	0.211 (0.112;0.670)	0.279 (0.195;0.618)	0.476	0.353 (0.231;0.466)	1.709 (0.032;4.392)	0.257

Histochemical stain or immunohistochemically detected molecule	Positive area per total white matter area [%]
	56 dpi			98 dpi
	Mock	TMEV	P‐value	Mock	TMEV	P‐value
Azan	0.018 (0.003;0.024)	0.090 (0.009;0.420)	0.030*	0.010 (0.003;0.030)	0.474 (0.003;3.550)	0.083
PSR, blue color	0.003 (0.000;0.005)	0.030 (0.000;1.049)	0.247	0.004 (0.000;0.013)	0.207 (0.000;0.493)	0.257
PSR, red color	0.010 (0.002;0.044)	0.101 (0.000;1.362)	0.177	0.014 (0.000;0.027)	0.497 (0.021;1.312)	0.019*
Aggrecan	0.002 (0.000;0.005)	0.006 (0.000;0.039)	0.352	0.002 (0.000;0.031)	0.003 (0.000;0.042)	0.931
Brevican	0.021 (0.000;0.031)	0.004 (0.000;0.015)	0.171	0.011 (0.000;0.032)	0.013 (0.000;0.038)	0.792
Collagen I	0.098 (0.050;0.815)	0.354 (0.013;0.637)	0.730	0.036 (0.011;0.064)	1.953 (0.138;13.755)	0.010*
Collagen IV	0.857 (0.043;2.085)	3.196 (2.073;7.421)	0.009*	0.509 (0.090;1.708)	8.710 (3.539;15.868)	0.010*
Decorin	0.080 (0.039;0.100)	0.073 (0.030;0.118)	0.762	0.041 (0.009;0.183)	0.212 (0.046;5.173)	0.052
Entactin	2.438 (2.063;3.173)	2.475 (1.699;5.222)	0.699	2.356 (0.922;2.874)	3.149 (1.986;5.711)	0.052
Fibronectin	0.010 (0.003;0.048)	0.005 (0.000;0.089)	0.792	0.018 (0.000;0.035)	0.430 (0.045;2.388)	0.004*
Laminin	0.019 (0.013;0.085)	0.046 (0.007;1.248)	0.352	0.059 (0.005;0.198)	1.189 (0.089;11.777)	0.017*
Neurocan	0.053 (0.025;0.109)	0.075 (0.036;0.150)	0.662	0.051 (0.017;0.243)	0.146 (0.059;0.436)	0.329
Perlecan	0.369 (0.105;0.853)	0.627 (0.369;5.084)	0.240	0.304 (0.021;0.539)	4.409 (0.307;5.262)	0.017*
Phosphacan	99.928 (99.878;99.989)	99.071 (88.127;99.982)	0.126	99.888 (99.854;99.980)	91.831 (89.665;96.469)	0.004*
Tenascin‐C	0.350 (0.074;0.538)	1.680 (0.299;4.033)	0.026*	0.215 (0.208;0.619)	1.887 (0.247;3.707)	0.032*

Histochemical stain or immunohistochemically detected molecule	Positive area per total white matter area [%]
	196 dpi
	Mock	TMEV	P‐value
Azan	0.012 (0.004;0.022)	1.474 (0.011;2.714)	0.030*
PSR, blue color	0.016 (0.000;0.022)	0.033 (0.004;0.885)	0.310
PSR, red color	0.005 (0.000;0.021)	2.561 (0.030;3.792)	0.008*
Aggrecan	0.008 (0.002;0.022)	0.012 (0.000;0.022)	0.610
Brevican	0.013 (0.000;0.035)	0.006 (0.003;0.012)	0.352
Collagen I	0.015 (0.000;0.051)	1.447 (0.108;7.642)	0.010*
Collagen IV	1.009 (0.133;1.638)	7.458 (1.495;8.584)	0.038*
Decorin	0.086 (0.020;0.356)	1.152 (0.020;7.004)	0.167
Entactin	2.515 (1.745;2.879)	5.202 (3.046;8.751)	0.004*
Fibronectin	0.002 (0.000;0.034)	0.393 (0.007;6.377)	0.038*
Laminin	0.085 (0.026;0.140)	3.050 (0.094;10.859)	0.017*
Neurocan	0.072 (0.048;0.088)	0.156 (0.057;1.194)	0.310
Perlecan	0.507 (0.143;0.870)	5.018 (2.908;5.711)	0.008*
Phosphacan	99.968 (99.908;99.990)	97.669 (87.386;99.877)	0.029*
Tenascin‐C	0.295 (0.076;0.453)	1.693 (0.067;6.600)	0.177

Data are given as percentage of positively labeled area per total white matter area.

^* Significant difference between TMEV‐infected and mock‐infected mice with P < 0.05 as detected by the Mann–Whitney U‐test.

^† Median (Minimum;Maximum).

Transcriptional changes of ECM‐related genes in TMEV‐infected mice

These histochemical results, in particular the diverse staining patterns in the PSR stain, strongly suggested a contribution of various molecules to ECM accumulation. In order to detect candidate molecules and to elucidate underlying molecular mechanism, microarray analysis of transcriptional changes within the spinal cord tissue was performed. Previous analyses of differential gene expression employing the spline‐based method embedded in the extraction of differential gene expression (EDGE) have revealed a total of 1210 genes differentially expressed at a false discovery rate of 1% in the study animals 23, 54. In the present study, a set of genes associated with ECM metabolism was manually selected for detailed analysis, including 20 genes already discovered by EDGE. Thus, a total of 350 probe sets was statistically analyzed to detect time point specific differences in gene expression between mock‐infected and TMEV‐infected mice (Supporting Information Table S1). Out of this list, 140 probe sets coding for 106 genes showed an altered gene expression at one to four of the investigation time points (Table 3, Supporting Information Table S2). In the group of ECM molecules, 31 of 40 differentially expressed genes exhibited mild upregulation with fold changes between 1.1 and 2.0 in the chronic phase of the disease. These included biglycan (gene symbol Bgn), chains of fibrillar and basement membrane collagens (eg, Col1a1, Col1a2, Col3a1, Col4a1), versican (Cspg2), neurocan (Cspg3), decorin (Dcn), fibronectin (Fn1), laminin chains (e.g. Lama3, Lama4, Lama5, Lamb1‐1, Lamb2, Lamc1, Lamc2), entactin (Nid2) and tenascin‐C (Tnc). In contrast, there was a minimal downregulation with fold changes between −1.1 and −1.4 of hyaluronan and proteoglycan link protein (Hapln1), aggrecan (Agc1) and phosphacan (Ptprz1; Table 3). Thirty‐one of the 49 enzymes involved in the biosynthesis of hyaluronan and chondroitin, keratan and heparan sulfates showed transcriptional changes that were generally only minimal and occurred particularly in the late phase of the disease. Enzymes of chondroitin sulfate biosynthesis tended to be downregulated, while there was an alternating expression profile of enzymes along the pathway of heparan sulfate biosynthesis (Supporting Information Table S2). Genes of glycosaminoglycan degradation enzymes were consistently upregulated from 42 dpi onwards including the maximal fold change among all selected genes by the ECM‐cleaving enzyme heparanase (Hpse), which showed an 8.53‐fold and 10.22‐fold upregulation at 98 dpi and 196 dpi, respectively (Table 3). Out of the MMP/TIMP group, further matrix modulating enzymes with robust fold changes in the chronic disease phase included MMP12 (Mmp12), being upregulated 5.26‐fold and 6.89‐fold at 98 dpi and 196 dpi, respectively, as well as MMP2 (Mmp2) and MMP‐19 (Mmp19). Several other MMP genes were minimally downregulated. The MMP inhibitors TIMP1 (Timp1) and TIMP2 (Timp2) showed upregulations of about twofold to 4.5‐fold between 42 dpi and 196 dpi (Table 3). Differentially expressed genes of pro‐fibrotic factors included transforming growth factor beta (TGFβ; Tgfb1, Tgfb3, Tgfbi) and one of its receptors, TGFβ receptor I (Tgfbr1), as well as the CC‐chemokine CCL2 (Ccl2) and its receptor CCR2 (Ccr2), which were upregulated about twofold from 42 dpi to 196 dpi. Two cytokines with anti‐fibrotic properties, interferon gamma (IFNγ; Ifng) and interleukin‐10 (IL‐10; Il10), were upregulated in a similar pattern (Supporting Information Table S2).

Table 3.

Probe set IDs of the Affymetrix mouse genome 430 2.0 array, official gene symbols and titles, and relative fold changes with respective P‐values for probe sets related to extracellular matrix molecules, glycosaminoglycan degradation and matrix metalloproteinases/tissue inhibitors of metalloproteinases (MMPs/TIMPs), and differentially expressed between TMEV‐infected and mock‐infected mice in the time course of TME. Abbreviations: dpi = days post‐infection; aka = also known as; MMP = matrix metalloproteinase; TIMP = tissue inhibitors of metalloproteinase; TMEV = Theiler's murine encephalomyelitis virus; TME = Theiler's murine encephalomyelitis.

Gene symbol	Probe set ID	Gene title	14 dpi		42 dpi		98 dpi		196 dpi
			Fold change	P‐value	Fold change	P‐value	Fold change	P‐value	Fold change	P‐value
Extracellular matrix molecules
Agc1	1449827_at	Aggrecan 1	−1.09*	0.041	−1.13*	0.041	−1.21*	0.004	−1.16	0.065
Bgn	1437889_x_at	Biglycan	1.19	0.394	1.14	0.093	1.07	0.537	1.56*	0.002
Bgn	1416405_at	Biglycan	1.22	0.394	1.31*	0.015	1.21	0.126	1.88*	0.002
Bgn	1448323_a_at	Biglycan	1.26	0.180	1.28*	0.015	1.19	0.177	1.84*	0.002
Col1a1	1423669_at	Procollagen type I alpha 1	2.05	0.240	1.07	0.180	−1.18	0.662	1.50*	0.004
Col1a2	1450857_a_at	Procollagen type I alpha 2	1.58	0.180	1.20	0.240	−1.15	0.537	1.75*	0.002
Col1a2	1423110_at	Procollagen type I alpha 2	1.51	0.394	−1.10	0.937	−1.28	0.429	1.60*	0.004
Col3a1	1427883_a_at	Procollagen type III alpha 1	1.28	0.485	1.29	0.093	1.03	0.931	2.35*	0.009
Col4a1	1452035_at	Procollagen type IV alpha 1	1.09*	0.041	1.11	0.180	−1.03	0.662	1.19	0.132
Col5a1	1416740_at	Procollagen type V alpha 1	1.33	0.240	1.10	0.240	1.03	0.429	1.34*	0.004
Col5a1	1434479_at	Procollagen type V alpha 1	1.10	0.485	−1.11	0.180	−1.12*	0.017	1.05	0.485
Col5a2	1422437_at	Procollagen type V alpha 2	1.21	0.394	1.10	0.394	−1.04	0.662	1.41*	0.026
Col6a1	1448590_at	Procollagen type VI alpha 1	1.31	0.394	1.11	0.394	−1.07	0.429	1.62*	0.002
Col6a2	1452250_a_at	Procollagen type VI alpha 2	1.23	0.132	1.03	0.589	−1.14	0.247	1.38*	0.015
Col6a2	1426947_x_at	Procollagen type VI alpha 2	1.11	0.240	1.13	0.180	−1.10	0.329	1.19*	0.015
Col6a3 /// LOC674521	1424131_at	Procollagen type VI alpha 3 /// similar to alpha 3 type VI collagen isoform 1 precursor	1.25	0.394	1.13	0.589	1.25*	0.009	1.36*	0.015
Col8a1	1418440_at	Procollagen type VIII alpha 1	1.07	0.240	−1.06	0.589	1.07	0.662	1.23*	0.041
Col8a2	1434667_at	Procollagen type VIII alpha 2	−1.00	0.937	−1.05	0.937	−1.14	0.126	1.35*	0.009
Col9a3	1460734_at	Procollagen type IX alpha 3	1.07	0.180	−1.10	0.310	1.23*	0.004	1.06	0.394
Col13a1	1422866_at	Procollagen type XIII alpha 1	1.05	0.818	1.08	0.485	−1.14*	0.004	1.09	0.310
Col14a1	1427168_a_at	Procollagen type XIV alpha 1	−1.05	0.937	−1.15	0.699	−1.26	0.247	1.48*	0.041
Col15a1	1448755_at	Procollagen type XV	−1.01	0.937	1.11	0.394	−1.11*	0.030	1.14	0.132
Col16a1	1427986_a_at	Procollagen type XVI alpha 1	1.29*	0.009	1.41*	0.004	1.45*	0.004	1.44*	0.009
Col18a1	1426955_at	Procollagen type XVIII alpha 1	1.13	0.065	1.06	0.310	1.04	0.537	1.29*	0.015
Col19a1	1421698_a_at	Procollagen type XIX alpha 1	1.21	0.132	−1.14	0.394	−1.21	0.537	−1.30*	0.009
Col20a1	1425234_at	Collagen type XX alpha 1	1.16*	0.026	1.22*	0.015	1.65*	0.004	1.44*	0.015
Col20a1	1430855_at	Collagen type XX alpha 1	1.05	0.394	1.31*	0.009	1.28	0.052	1.02	0.589
Col22a1	1453084_s_at	Collagen type XXII alpha 1	1.10	0.818	1.05	0.699	−1.15*	0.030	1.01	0.589
Col27a1	1429549_at	Procollagen type XXVII alpha 1	1.22*	0.026	1.16	0.699	−1.03	0.792	1.10	0.180
Cspg2	1427256_at	Chondroitin sulfate proteoglycan 2 (aka neurocan)	1.04	0.818	1.08	0.240	1.49*	0.004	1.40*	0.026
Cspg2	1427257_at	Chondroitin sulfate proteoglycan 2 (aka neurocan)	−1.21*	0.041	1.06	0.699	1.26*	0.004	1.02	0.699
Cspg3	1423604_at	Chondroitin sulfate proteoglycan 3 (aka neurocan)	1.01	0.818	1.16*	0.009	1.11*	0.030	−1.02	0.485
Cspg3	1434679_at	Chondroitin sulfate proteoglycan 3 (aka neurocan)	1.11*	0.002	1.37*	0.002	1.59*	0.004	1.48*	0.002
Dcn	1449368_at	Decorin	1.29	0.394	1.25	0.310	−1.34	0.247	1.94*	0.002
Eln	1420854_at	Elastin	−1.03	0.589	−1.04	0.394	−1.06	0.429	−1.19*	0.009
Fn1	1426642_at	Fibronectin 1	1.37	0.180	1.48*	0.009	1.32*	0.030	1.61*	0.004
Hapln1	1438020_at	Hyaluronan and proteoglycan link protein 1	−1.34*	0.015	−1.19*	0.009	−1.19	0.177	−1.26*	0.009
Lama2	1426285_at	Laminin alpha 2	−1.06	0.699	−1.04	0.937	−1.20*	0.009	1.14	0.180
Lama3	1444860_at	Laminin alpha 3	−1.04	0.485	−1.03	0.818	1.04	0.329	−1.17*	0.026
Lama3 /// LOC669814	1427512_a_at	Laminin alpha 3 /// similar to Laminin alpha‐3 chain precursor (Nicein alpha subunit)	1.01	0.937	−1.01	0.937	1.01	1.000	1.40*	0.026
Lama4	1424807_at	Laminin alpha 4	1.01	0.699	1.03	0.699	1.02	0.429	1.41*	0.004
Lama5 /// LOC671909	1427009_at	Laminin alpha 5 /// similar to Laminin alpha‐5 chain precursor	−1.06	0.240	1.02	0.589	−1.12	0.177	1.12*	0.026
Lamb1‐1	1424114_s_at	Laminin B1 subunit 1	1.01	0.937	−1.05	0.589	−1.14	0.082	1.34*	0.026
Lamb2	1416513_at	Laminin beta 2	1.03	0.699	−1.01	0.485	−1.08	0.537	1.32*	0.004
Lamc1	1423885_at	Laminin gamma 1	1.11	0.065	1.04	0.818	1.12	0.126	1.26*	0.009
Lamc2	1421279_at	Laminin gamma 2	1.09	0.240	1.17*	0.026	1.07	0.247	1.01	0.937
Nid2	1423516_a_at	Nidogen 2 (aka entactin)	1.02	0.310	1.30*	0.041	1.21	0.052	1.49*	0.026
Ptprz1	1452284_at	Protein tyrosine phosphatase, receptor type Z, polypeptide 1 (aka phosphacan)	−1.04	0.310	−1.12*	0.015	−1.08	0.052	−1.11*	0.026
Tnc	1416342_at	Tenascin C	1.33	0.093	1.38*	0.004	1.33*	0.017	1.26	0.394
Enzymes of glycosaminoglycan degradation
Arsb	1429190_at	Arylsulfatase B	−1.05	0.310	1.18*	0.041	1.24	0.537	1.41	0.093
Galns	1448744_at	Galactosamine (N‐acetyl)‐6‐sulfate sulfatase	−1.04	0.818	1.34*	0.002	1.50*	0.004	1.58*	0.002
Gns	1433488_x_at	Glucosamine (N‐acetyl)‐6‐sulfatase	1.18*	0.009	1.62*	0.002	2.09*	0.004	1.81*	0.002
Gns	1433546_at	Glucosamine (N‐acetyl)‐6‐sulfatase	1.19*	0.002	1.57*	0.002	1.89*	0.004	1.81*	0.002
Gns	1446861_at	Glucosamine (N‐acetyl)‐6‐sulfatase	1.28*	0.015	−1.00	0.937	1.24	0.082	1.05	0.699
Gusb	1430332_a_at	Glucuronidase, beta	−1.26	0.132	1.05	0.699	1.13	0.662	1.45*	0.009
Gusb	1448124_at	Glucuronidase, beta	1.19*	0.041	2.15*	0.002	3.10*	0.004	3.30*	0.002
Hexa	1449024_a_at	Hexosaminidase A	1.13	0.065	1.53*	0.002	1.98*	0.004	2.04*	0.002
Hexb	1437874_s_at	Hexosaminidase B	1.18	0.093	1.64*	0.002	1.96*	0.004	1.75*	0.002
Hexb	1460180_at	Hexosaminidase B	1.34*	0.004	1.96*	0.002	2.39*	0.004	2.24*	0.002
Hpse	1433930_at	Heparanase	1.08	0.394	2.80*	0.002	8.53*	0.004	10.22*	0.002
Hyal1 /// Nat6	1449954_at	Hyaluronoglucosaminidase 1 /// N‐acetyltransferase 6	−1.03	0.589	−1.03	0.589	1.02	0.792	1.26*	0.002
Ids	1427813_at	Iduronate 2‐sulfatase	−1.03	0.485	−1.11*	0.041	1.01	0.792	1.01	0.937
Ids	1455481_at	Iduronate 2‐sulfatase	−1.01	0.818	−1.08*	0.026	−1.15*	0.017	−1.10*	0.002
Idua	1418793_at	Iduronidase alpha‐L‐	1.03	0.394	1.11*	0.026	1.19	0.052	1.06	0.132
Idua	1439251_at	Iduronidase alpha‐L‐	1.00	0.818	1.23*	0.002	1.16	0.052	1.07	0.310
Naglu	1417706_at	Alpha‐N‐acetylglucosaminidase (Sanfilippo disease IIIB)	1.05	0.818	1.91*	0.002	2.92*	0.004	3.18*	0.002
Matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs)
Mmp2	1416136_at	Matrix metallopeptidase 2	−1.03	0.937	1.00	0.818	1.08	0.429	1.61*	0.002
Mmp2	1439364_a_at	Matrix metallopeptidase 2	1.04	0.937	1.13	0.240	1.13	0.429	1.64*	0.002
Mmp3	1418945_at	Matrix metallopeptidase 3	1.07	0.394	1.02	0.818	1.22*	0.017	1.28	0.310
Mmp9	1448291_at	Matrix metallopeptidase 9	−1.15	0.310	−1.28*	0.004	−1.14	0.177	−1.02	0.937
Mmp9	1416298_at	Matrix metallopeptidase 9	−1.06	0.589	−1.12*	0.041	−1.11	0.429	−1.08	0.310
Mmp12	1449153_at	Matrix metallopeptidase 12	1.02	0.589	1.47*	0.026	5.26*	0.004	6.89*	0.002
Mmp13	1417256_at	Matrix metallopeptidase 13	1.33	0.589	1.02	0.240	1.11	0.792	1.21*	0.015
Mmp14	1448383_at	Matrix metallopeptidase 14 (membrane‐inserted)	1.09	0.485	1.37*	0.041	1.08	0.537	1.41	0.240
Mmp14	1440920_at	Matrix metallopeptidase 14 (membrane‐inserted)	1.07	0.310	1.22*	0.026	−1.00	1.000	1.16	0.132
Mmp16	1440161_at	Matrix metallopeptidase 16	−1.05	0.394	1.13*	0.026	−1.16	0.537	−1.09	0.132
Mmp17	1448598_at	Matrix metallopeptidase 17	−1.02	0.589	−1.18*	0.004	−1.11	0.052	−1.17*	0.041
Mmp19	1421977_at	Matrix metallopeptidase 19	−1.02	0.937	1.01	0.937	1.47*	0.004	1.49*	0.002
Mmp23	1417281_a_at	Matrix metallopeptidase 23	1.09	0.394	1.03	0.240	−1.19*	0.030	1.16*	0.041
Mmp24	1417678_at	Matrix metallopeptidase 24	−1.10	0.240	−1.12*	0.041	−1.07	0.429	−1.05	0.699
Mmp24	1441456_at	Matrix metallopeptidase 24	−1.09	0.240	−1.15*	0.009	−1.07	0.329	−1.08*	0.041
Mmp24	1456376_at	Matrix metallopeptidase 24	−1.06*	0.041	−1.08	0.093	−1.18*	0.004	−1.09*	0.041
Mmp28	1447544_at	Matrix metallopeptidase 28 (epilysin)	−1.08	0.310	−1.04	0.589	−1.18*	0.017	1.08	0.310
Timp1	1460227_at	Tissue inhibitor of metalloproteinase 1	1.24	0.240	3.18*	0.002	4.56*	0.004	4.00*	0.002
Timp2	1460287_at	Tissue inhibitor of metalloproteinase 2	1.06	0.240	1.93*	0.002	2.67*	0.004	2.65*	0.002
Timp4	1450974_at	Tissue inhibitor of metalloproteinase 4	−1.15	0.310	−1.24*	0.015	−1.25*	0.017	−1.16	0.065
Timp4	1423405_at	Tissue inhibitor of metalloproteinase 4	−1.11	0.485	−1.20	0.093	−1.19*	0.017	−1.22*	0.026

^* Significant difference between TMEV‐infected and mock‐infected mice with P < 0.05 as detected by the Mann–Whitney U‐test.

Distinct spatio‐temporal morphological pattern of various ECM molecules in TME lesions

These microarray results led to the assumption that transcriptional upregulations and altered composition of ECM molecules contributed to the prominent, histochemically demonstrated matrix deposition in TME lesions. For detailed localization and phenotyping of the ECM, the following molecules were chosen for immunostaining: aggrecan, brevican, decorin, neurocan, phosphacan, perlecan (all proteoglycans), entactin, fibronectin, laminin, tenascin‐C (all glycoproteins), collagen I and IV (Table 1). This selection particularly considers the known involvement of these substances in glial scar formation.

Within the spinal cord of mock‐infected mice, the immunoreactivity pattern of all applied antibodies remained unchanged throughout the observation period until 196 dpi (Table 2). In detail, immunolabeling for collagen I, decorin and fibronectin was restricted to meningeal connective tissue. Collagen IV, entactin and laminin immunoreactivity were also present in meninges and additionally in blood vessel walls. Immunostaining for perlecan and tenascin‐C revealed a weak labeling of meningeal structures and a diffuse staining of the spinal cord gray and white matter. Neurocan immunolabeling was restricted to the spinal nerves and was completely absent in the gray and white matter of the spinal cord. In contrast, diffuse phosphacan expression was prominent in spinal cord gray and white matter. Aggrecan and brevican immunoreactivity was generally weak and restricted to the gray matter, where both molecules showed a perineuronal enhancement of immunolabeling.

In TMEV‐infected animals, all basement membrane components investigated as well as several non‐BM molecules contributed to ECM accumulation, which in general was restricted to lesioned areas.

Regarding the BM molecules, collagen IV and laminin started to accumulate intralesionally as early as 28 days after TMEV infection, displaying a filamentous extracellular pattern in addition to the regular vascular reaction (Figure 2B,C,F,G). Until 196 dpi, the positive immunoreaction of both molecules progressively expanded, resulting in a lattice‐like pattern throughout the lesioned area, with frequent perivascular enhancement and occupying up to 13.76% (collagen IV) and 11.78% (laminin) of the total white matter area (details in Table 2).

Immunohistochemical double labeling demonstrated that intralesional deposition of collagen IV occurred predominantly without spatial association to von Willebrand factor‐positive vascular structures in spinal cord lesions 98 dpi and 196 dpi (Figure 4).

Figure 4.

Immunohistochemical findings after Theiler's murine encephalomyelitis virus (TMEV) infection. Section of the thoracic spinal cord of a TMEV‐infected mouse from 196 days postinfection, showing immunoreactivity for von Willebrand factor (brown signal) and collagen IV (bluish to greenish signal). In a demyelinated white matter lesion, prominent collagen IV deposits are present predominantly without (black arrowheads) and occasionally with (white arrowhead) association to von Willebrand factor‐positive blood vessels (arrows). Immunohistology, ABC method, DAB chromogen (von Willebrand factor) and HistoGreen chromogen (collagen IV). Nomarski differential interference contrast. Scale bar = 40 µm.

Similarly, vascular perlecan immunoreactivity, which regularly was very weak, started to increase at 28 dpi within demyelinated areas. While the perlecan‐positive area continuously expanded until 198 dpi, an additional parenchymal staining appeared besides the vascular accentuation (Figure 2N). An increase in entactin immunoreactivity was observed at 56 dpi, displaying a partly granular, partly meshwork‐like pattern. While the median of the entactin‐positive white matter area increased up to 5.2% at 196 dpi, the immunolabeling frequently showed a distinct spatial relation to vascular structures (Figure 2O). Extracellular fibronectin immunoreactivity was observed only at 98 dpi and more prominently at 196 dpi in several demyelinated lesions, exhibiting a granular to laminar pattern (Figure 2J). Simultaneously, several intralesional macrophages and gitter cells contained fibronectin within their cytoplasm.

The first non‐BM molecule that started to accumulate within demyelinated areas was tenascin‐C, displaying a granular to dendritic pattern at 28 dpi. During disease progression, tenascin‐C continuously accumulated, reaching 1.89% and 1.69% area percentage at 98 dpi and 196 dpi, respectively (Figure 2P). An intralesional deposition of collagen I, decorin and neurocan was detected at 56 dpi and further increased until day 196 pi. The collagen I signal displayed a granular to filamentous pattern within demyelinated lesions (Figure 2D,H). This immunolabeling continued to expand, hence developing a lattice‐like and rarely an extended laminar pattern and reaching median area percentages of 1.96% and 1.48% at 98 dpi and 196 dpi, respectively. Mild decorin and neurocan immunolabeling was observed within a demyelinated area in a single animal 56 days after TMEV infection, whereas several mice showed positive signals at 98 dpi and 196 dpi, respectively. Simultaneously, the size of immunopositive areas increased, with decorin displaying a dendritic pattern and neurocan exhibiting a multifocal semi‐circular to circular labeling most likely attributed to capillary blood vessels (Figure 2K,L). However, statistic analysis failed to demonstrate a significant difference to the mock‐infected group for both molecules. In contrast to all other ECM molecules investigated, phosphacan expression showed a progressive reduction following TMEV infection. Loss of phosphacan immunoreactivity occurred as early as 28 dpi multifocally within demyelinated lesions and progressed until 196 dpi (Figure 2M). Aggrecan and brevican did not show any changes in their expression pattern during TME lesion development.

DISCUSSION

In this study, microarray analysis as well as histochemistry and immunohistochemistry were used to assess the contribution of ECM alterations and to identify their molecular mechanisms in demyelinating TME, a virus‐induced experimental model of MS.

Within demyelinated TME lesions, prominent ECM accumulation occurred in the lesion center, starting as early as 28 dpi and continuously progressed until the end of the observation period at 196 dpi. With this time course, matrix alterations closely followed the process of demyelination and paralleled the development of astrogliosis 52, 53. First clues about the molecular composition of the accumulating ECM were obtained by histochemical stains, which revealed a contribution of collagens, glycoproteins and proteoglycans (4). Additionally, a transient bluish to greenish signal in the PSR stain with an exceptional perilesional distribution indicated the presence of hyaluronan (4). Similarly, hyaluronan has been demonstrated to accumulate within demyelinated lesions of MS patients and rats with EAE. This was attributed to elevated expression of the hyaluronan receptor CD44 by glial cells, which appears to be a common effect in demyelinating conditions (2). In MS and EAE, this hyaluronan accumulation persists and interferes with remyelination by inhibiting oligodendrocyte progenitor cell maturation (3). Similar mechanisms may play a role in TME, where an insufficient differentiation of oligodendrocyte progenitor cells results in failure of complete remyelination (53).

Using immunohistochemistry, various ECM components were shown to contribute to intralesional matrix alterations. Progressively accumulating molecules included collagen I and collagen IV, the CSPGs decorin and neurocan, and the glycoproteins laminin, entactin, tenascin‐C and fibronectin, though with distinct differences concerning their spatial distribution. Deposition of collagen I, decorin, fibronectin, neurocan and tenascin‐C was found diffusely distributed throughout demyelinated areas in TME. Strikingly, also the BM components collagen IV, laminin, entactin and perlecan accumulated in the demyelinated white matter parenchyma without association to blood vessels. Similar observations have been made in MS, where networks of collagen IV, laminin and perlecan occur diffusely within plaques (56). In contrast, the diffuse deposition of non‐BM components was not observed in MS, where collagen I, decorin and fibronectin accumulate only in the perivascular space, and neurocan and tenascin‐C are even reduced in plaque centers 16, 30, 40, 55. An accumulation of the BM molecule entactin as detected in TME has not been described in primary demyelinating conditions before but occurs in experimental spinal cord injury (17). The amount and distribution of brevican and aggrecan remained unchanged in spinal cord lesions in TME, although upregulated expression and accumulation of these molecules were observed in rat EAE and human MS plaques, respectively 19, 20, 40, 46. These divergent reactions following CNS lesions probably rely on species‐specific differences, as brevican and aggrecan physiologically occur in spinal cord white matter in rats and humans but were restricted to the gray matter in mice 20, 40.

In general, the different spatial patterns of matrix molecule deposition in TME lesions raise the question of the cellular source of the accumulating matrix. As the observed ECM deposition parallels progressive astrogliosis both spatially and temporally, a substantial contribution of reactive astrocytes has to be assumed. This is in accordance with results from CNS injury models, where activated astrocytes were proven to represent the major source of CSPGs and glycoproteins, collectively forming the so‐called glial scar 14, 20, 28, 46. An astrocytic origin of ECM is further supported by findings in demyelinated MS lesions, where matrix deposits are closely associated with areas of astrogliosis, whereas chronic plaque centers with marked cell loss and tissue destruction exhibit prominent loss of ECM 16, 40, 56. Moreover, reactive astrocytes are able to secrete BM molecules such as collagen IV and laminin both in BM structures and in parenchymal patterns 17, 24. Schwann cells represent another potential source of BM structures, hence the occurrence of Schwann cell‐type remyelination in TME suggests a contribution of this cell population to parenchymal, nonvessel‐associated ECM deposition in the present study 1, 53. In experimental rodent spinal cord injury, meningeal fibroblasts are also substantially involved in intralesional matrix deposition including immediate upregulation of tenascin‐C, brevican and neurocan, and delayed secretion of collagens and laminin 14, 17, 20, 24, 46, 59. However, analog processes are considered negligible in TME as there is no primary meningeal tissue disruption and thus no considerable trigger for fibroblast proliferation and invasion into spinal cord tissue.

Regarding the functional impact of a glial scar, a certain degree of reactive astrogliosis has been proven crucial following CNS injury in order to re‐establish the BBB and prevent ongoing tissue disruption and influx of inflammatory cells 13, 49. However, overwhelming astrogliosis and accompanying accumulation of CSPGs, tenascin‐C and various BM components are generally considered as inhibitory environment for regenerative processes, particularly axonal sprouting 8, 11, 14, 17, 41, 43, 58. Furthermore, the glial scar considerably impairs remyelination, as ECM deposits inhibit maturation and differentiation of oligodendrocyte precursors 3, 53. This strongly suggests that the failure of remyelination in TME is partly attributable to the dense ECM accumulation within the lesions, which probably inhibits regenerative attempts of resident and progenitor cells.

The CNS‐specific proteoglycan phosphacan was the only ECM component that was progressively reduced in demyelinated TME lesions, accompanied by slight downregulation of the respective gene (Ptprz1). Phosphacan immunoreactivity was similarly decreased in the early phase of experimental spinal cord injury and in the cell‐poor center of chronic MS plaques; however, in association with astrogliosis, phosphacan amount was restored or increased in these entities 20, 40, 46. These differences may in part depend on the impossibility to differentiate immunohistochemically between extracellular, putatively accumulating phosphacan and cell membrane‐bound phosphacan that is probably diminished because of cell loss.

A second aspect in matrix accumulation in TME lesions was the successive temporal occurrence of individual ECM molecules during the disease course. The amount of collagen IV, laminin, perlecan and tenascin‐C started to increase from 28 dpi on, whereas accumulation of collagen I, decorin, entactin and neurocan was not observed until 56 dpi, or occurred only 98 dpi, as in the case of fibronectin. Interestingly, this temporal course was only in part paralleled by transcriptional changes of the respective genes. Accumulation of several molecules, including collagens I and IV, laminin, decorin and perlecan, was morphologically demonstrable from the early phase or midterm of the disease onwards, although transcriptional upregulation of their respective genes were detected only transiently (Col4a1), considerably delayed in the disease course (Col1a1, Col1a2, Col3a1; Dcn; Lama3, Lama4, Lama5, Lamb1‐1, Lamb2, Lamc1, Lamc2) or not at all (Hspg2). In contrast, progressive tenascin‐C accumulation was closely paralleled by upregulated gene expression (Tnc), and increased transcription of their respective genes preceded and accompanied the intralesional deposition of entactin (Nid2), neurocan (Cspg3) and fibronectin (Fn1). Matrix accumulations also exhibit diverse patterns in different types of MS plaques and have been considered as an important factor for regeneration failure, but a distinct timeline of ECM deposition is not definable because of the spontaneous character of this disease 16, 30, 40, 56, 58. Therefore, the precisely regulated temporal course of ECM production and deposition in TME further enhances the value of this model to examine the impact of single ECM molecules upon the pathogenesis and to develop preventive and therapeutic strategies in demyelinating diseases such as MS.

Interestingly, the present study revealed a lack of complete parallelism between morphological matrix accumulation and transcriptional changes of related genes. There are several possible explanations for this seeming incongruity, one of which is the different observation period covered by the two methods. Immunohistochemistry shows the summation of altered protein production, whereas microarray analysis reveals a snapshot‐like picture of one particular transcriptional state. Therefore, periods of increased gene expression between the different evaluation time points remain undetected, or minimal transcriptional alterations may be below the detection level, albeit sufficient for long‐term ECM accumulation. Finally, lesion areas in TME are relatively small compared with the total area of spinal cord transverse sections, which are used for microarray analysis, so that the transcriptional profile of the lesion may be superimposed by that of the surrounding intact tissue.

In order to detect superior regulating mechanisms of ECM alterations, several pro‐fibrotic factors such as TGFβ and IL‐13 and chemokines of the CC‐ and CXC‐family were included in the microarray analysis (61). In TME, the gene expression of TGFβ1 (Tgfb1) and its receptor TGFβ receptor I (Tgfbr1) was mildly upregulated in the chronic phase of the disease, probably because of microglia activation and infiltration of macrophages (38). This may have promoted the astrocytic secretion of CSPGs and collagens 24, 38. In contrast, a TGFβ1 upregulation correlating with an increased laminin content of blood vessel walls was shown in the acute phase of TME by others (51). However, the highly virulent GDVII strain of TMEV was used in this aforementioned study, thus virus‐dependent effects may account for differences in TGFβ1 expression profiles. The pro‐fibrotic effect of CCL2 and CCR2, which showed mild upregulation in the late phase of TME, is well known from pulmonary fibrosis, and an increased expression by reactive astrocytes has also been shown in MS lesions 31, 39, 47. However, their role in ECM modulation during CNS diseases is somewhat ambiguous as CCL2 signaling may result in MMP synthesis and thus contribute to BBB breakdown and inflammatory influx 6, 47. As monocytes represent another main source of CCL2, the observed upregulation in TME may at least partly be derived from the inflammatory infiltrate in the affected spinal cord (30).

With regard to the biosynthesis pathways of proteoglycans and hyaluronan, there was not a uniform pattern of enzyme regulation, with only minimal fold changes as well as alternating up‐ and downregulation within one pathway. As these findings may be based on periodic gene transcription or intrinsic feedback mechanisms, a partial contribution of synthesis pathway regulation in the development of ECM accumulation cannot be ruled out. In contrast, enzymes of proteoglycan cleavage showed obvious transcriptional activation in TME, with heparanase showing the strongest upregulation. As this enzyme specifically cleaves heparan sulfate proteoglycans, which are integral components of BMs, it disturbs BBB integrity and thus favors inflammatory influx 12, 33. Moreover, heparanase overexpression coincides with astrocyte activation and migration, which suggests a contributing role of this enzyme for the development of astrogliosis in TME (45). Transcriptional changes were also detected in the MMP/TIMP group of matrix modulating enzymes, in particular represented by prominent upregulation of MMP12 as well as TIMP1 and TIMP2 in the late phase of TMEV infection. This confirms previous polymerase chain reaction (PCR) results in TME, where increased MMP12 levels have been considered as a trigger of ongoing demyelination (52). The mildly increased expression of MMP2 and MMP19 parallels recent findings in MS lesions, where these enzymes are involved in BBB leakage and ECM remodeling 30, 57. MMP9 expression levels were mildly decreased during TMEV infection, which parallels previous PCR results but is in contrast to MS and other demyelinating diseases such as canine distemper 26, 29, 52. Conclusively, the overall tendency of upregulated matrix degradation did not prevent net accumulation of ECM within lesions. Putative mechanisms include inhomogeneous distribution of digesting enzymes in the TME‐affected spinal cord, particularly lack of matrix‐degrading enzymes in demyelinated areas, upregulation of TIMPs as well as altered enzymatic activities. Additionally, disturbed aggregation of matrix molecule subunits (eg, laminin chains) or post‐translational modifications of the ECM, which are not reflected by transcriptional changes, might withdraw them from regular degradation.

CONCLUSION

TMEV‐induced demyelination in the murine spinal cord is accompanied by progressive astrogliosis and prominent intralesional ECM accumulation resulting in a glial scar, which is considered to impair regenerative processes. Contributing molecular mechanisms include induction of pro‐fibrotic factors as well as alterations in matrix‐modulating pathways. The experimental and well‐reproducible character renders chronic demyelinating TME a highly valuable model for further studies on glial scar formation in demyelinating diseases. Moreover, the distinct spatio‐temporal pattern of matrix deposition allows to analyze the impact of individual ECM components to this condition and to modulate them subsequently for therapeutic strategies. Finally, morphological approaches may occasionally be superior to molecular methods in order to detect mild but slowly progressive alterations.

Supporting information

Table S1. Probe set IDs of the Affymetrix mouse genome 430 2.0 array, official gene symbols and titles, and relative fold changes with respective p‐values of all 350 probe sets manually selected for further analysis.

Table S2. Probe set IDs of the Affymetrix mouse genome 430 2.0 array, official gene symbols and titles, and relative fold changes with respective P‐values for probe sets related to chondroitin sulfate biosynthesis, heparan sulfate biosynthesis, keratan sulfate biosynthesis and fibrosis, and differentially expressed between TMEV‐infected and mock‐infected mice in the time course of TME.

Supporting info item

Supporting info item