OUP user menu

Demyelination, Inflammation, and Neurodegeneration in Multiple Sclerosis Deep Gray Matter

Marco Vercellino MD, PhD, Silvia Masera MD, Marcella Lorenzatti BSc, Cecilia Condello MD, Aristide Merola MD, Alessandra Mattioda MD, Antonella Tribolo MD, Elisabetta Capello MD, Giovanni Luigi Mancardi MD, Roberto Mutani MD, Maria Teresa Giordana MD, PhD, Paola Cavalla MD
DOI: http://dx.doi.org/10.1097/NEN.0b013e3181a19a5a 489-502 First published online: 1 May 2009

Abstract

Gray matter (GM) lesions are recognized as important components of the pathology of multiple sclerosis (MS), and involvement of the deep gray matter (DGM) is suggested by magnetic resonance imaging. The aims of this study were to determine the frequency and distribution of lesions and characterize the inflammatory and neurodegenerative changes in DGM of MS patients. Histochemistry, immunohistochemistry, and morphometry were performed on whole coronal sections of 14 MS and 12 control (6 normal, 6 from amyotrophic lateral sclerosis patients) brains. Demyelinating lesions were frequent in MS DGM; most often in the thalamus and caudate, but they were also seen in the putamen, pallidum, claustrum, amygdala, hypothalamus, and substantia nigra. Most DGM lesions involved both GM and white matter. Inflammation in active DGM lesions was similar to that in lesions only in white matter but was less intense, and there was a preponderance of activated microglia, scarce myelin-laden macrophages, and a lesser extent of axonal damage. Neuronal loss was observed both in DGM lesions and nondemyelinated DGM with neuron atrophy in nondemyelinated DGM. In conclusion, demyelination and neurodegenerative changes are common in MS DGM and may contribute to clinical impairment. Inflammation in DGM lesions is intermediate between the destructive inflammation of white matter lesions and the minimal inflammation of cortical lesions. We hypothesize that alterations of glutamate reuptake mechanisms may contribute to these differences.

Key Words
  • Basal ganglia
  • Caudate
  • Excitotoxicity
  • Gray matter
  • Multiple sclerosis
  • Neuronal loss
  • Thalamus

Introduction

Involvement of the gray matter (GM) is now recognized to be an important pathological feature of multiple sclerosis (MS). Gray matter demyelinating lesions have been documented in the cerebral cortex (1-4), cerebellar cortex (5), spinal cord (6), and the hippocampus (3, 7, 8). Gray matter demyelination seems to be more frequent in progressive MS (4) and is likely an important determinant of disability and cognitive deficits. There are, however, no reliable clinical markers of GM demyelination (9, 10).

Neurodegenerative changes with loss of neurons and axonal transection have also been described in the GM in MS patients (1, 3, 6, 8), both in demyelinating lesions and in normal-appearing GM. The pathogenesis of GM lesions is not understood and may differ from that of white matter (WM) lesions because pure GM lesions lack many typical inflammatory features of WM lesions (1, 11-13). The pathogenesis of neurodegeneration in MS GM is also unclear.

No systematic neuropathologic studies have focused on deep gray matter (DGM) (i.e. caudate, thalamus, putamen, pallidum, hypothalamus, amygdala, and claustrum) in MS. Recent data from magnetic resonance imaging (MRI) studies suggest that pathological changes in the DGM, particularly in the caudate nucleus and in the thalamus, are frequent in MS patients. Caudate nucleus (14, 15) and thalamic (16-19) atrophy has also been observed, and lower N-acetyl-D- aspartate values have been described in the thalamus in MS patients (18-20). Moreover, atrophy of caudate, thalamus, and putamen seems to be present even in the earliest stages of the disease and may precede involvement of other GM structures (21, 22). Selective atrophy of the thalamus with sparing of cortical GM and other DGM nuclei has also been described in pediatric MS (16).

The aims of this study were to 1) evaluate the extent, distribution, and principal morphological features of demyelinating lesions in the DGM in MS patient autopsy specimens; 2) describe the major features of inflammation and gliosis in MS DGM lesions and normal-appearing DGM (NADGM) and to compare them with cortical GM and WM lesions; and 3) assess the presence and extent of neurodegenerative changes (i.e. neuronal loss, neuronal atrophy, axonal damage, and synaptic loss) in the DGM in MS patients.

Materials and Methods

This study was performed on formalin-fixed paraffin-embedded material from 14 autopsy MS brains (6 relapsing-remitting [RR] MS, 7 secondary progressive [SP] MS, and 1 hyperacute MS), 6 control brains of patients without brain diseases, and 6 brains from amyotrophic lateral sclerosis (ALS) patients. The MS cases represented all currently available MS brains in our archives for which whole coronal sections were present. The goal was to provide information on the frequency and variability of DGM involvement in an autopsy series of unselected MS cases. Material was obtained from the archives of the University of Turin and the University of Genoa. ALS brains were used as controls to MS brains, as no demyelination, inflammation or major neurodegenerative changes are known in ALS deep grey matter.

An entire coronal section at the level of the mamillary bodies was analyzed in each brain. Postmortem intervals were less than 36 hours in all disease and control cases. The clinical details of the MS cases are summarized in Table 1. The mean age at death of MS cases was 50.2 years (range, 27-66 years); mean duration of disease course was 15.1 years (range, 6 months-30 years). The mean age at death of control and ALS cases was 59.4 years (range, 52-68 years). The cases were determined to have had RR, SP, or acute MS courses based on retrospective analysis of hospital records.

View this table:
TABLE 1.

Histology and Immunohistochemistry

Each entire coronal section was divided into smaller blocks for processing. Consecutive 5-μm sections were obtained. Standard hematoxylin and eosin, Luxol fast blue, and Bielschowsky preparations were obtained for each section. Immunohistochemistry was performed using antibodies to the following molecules: myelin basic protein (MBP), HLA-DR, CD68, CD3, CD20, CD138, vascular cell adhesion molecule 1 (VCAM-1), fibrinogen, immunoglobulin G (IgG), β-amyloid precursor protein (APP), neurofilaments, myeloperoxidase (MPO), excitatory amino acid transporter 1 (EAAT1), excitatory amino acid transporter 2 (EAAT2), glial fibrillary acidic protein (GFAP), and synaptophysin (Table 2). After deparaffinization, sections were treated with 3% hydrogen peroxide for 10 minutes and then processed for antigen retrieval as appropriate for each antibody (Table 2). The sections were incubated with 10% normal serum for 30 minutes and later incubated overnight with the primary antibodies. After washing with Tris buffer solution, the sections were incubated at room temperature for 30 minutes with the Envision complex (DAKO, Glostrup, Denmark). Peroxidase labeling was visualized with 10% 3,3′-diaminobenzidine (DAB). Sections were counterstained with hematoxylin.

View this table:
TABLE 2.

Double immunostaining was performed when necessary to identify possible colocalization of antigens. Immunohistochemistry was performed as previously described. Peroxidase labeling for the first antibody was visualized with 10% DAB. Immunohistochemistry for the second antibody was performed after blocking with normal serum, and peroxidase labeling was visualized using a Vector VIP kit (Vector Laboratories, Burlingame, CA).

Image Acquisition

The sections were examined using either a Zeiss Axiophot microscope or a Zeiss Axio Imager.A1 microscope (Carl Zeiss MicroImaging, Göttingen, Germany). Images were acquired using either a Nikon Digital Sight DS-DM camera (Nikon Corporation, Tokyo, Japan) or a Zeiss Axiocam MRc5 camera (Carl Zeiss MicroImaging).

Evaluation of Demyelination and Inflammation

Demyelinating lesions were defined as sharply demarcated areas of complete loss of MBP immunostaining with relative sparing of axons as demonstrated by Bielschowsky impregnation and neurofilament immunostaining. The areas of demyelinated WM, demyelinated cortex, and demyelinated DGM in each MBP-stained section were measured on the digital image with the Eclipse.Net software, version 1.16.6 (Laboratory Imaging, Prague, Czech Republic). Values were compounded, and the percentages of demyelinated WM, demyelinated cortex, and demyelinated DGM were calculated for each case.

Axonal loss, deformation, and thickening in WM and GM demyelinating lesions were assessed on a visual basis in Bielschowsky and neurofilament-stained sections. Inflammation in demyelinating lesions was quantified according to Bø et al (1, 23) and Van der Valk and De Groot (24), assessing MBP and HLA-DR immunostaining. Lesions were classified as active (hypercellular), chronic active (hypercellular border, hypocellular center), or chronic inactive (hypocellular). The HLA-DR, fibrinogen, IgG, VCAM-1, EAAT1, EAAT2, and synaptophysin immunostaining, and the presence of myelin-laden macrophages (MBP/HLA-DR) in active lesions were evaluated on a semiquantitative basis by 2 independent observers who scored the intensity of staining as follows: − (no staining); + (weak staining/rare stained cells); ++ (moderate staining/moderate number of stained cells); and +++ (strong staining/high number of stained cells). The numbers of APP-positive axonal spheroids and CD3+ cells in active lesions were assessed in 100× microscopic fields (corresponding to 0.038 mm2) and expressed as number of elements per square millimeter. Correlations between the extent of WM, cortical, and DGM demyelination, and between the clinical parameters (age, disease duration) previously described were evaluated using Spearman rank correlation test. Differences in the extent of DMG demyelination between RR and SP MS cases were evaluated using the Mann-Whitney U test.

Assessment of Neuronal and Glial Cell Density

Neuronal and glial cell density in the demyelinated DGM was assessed only on chronic established lesions, that is, excluding active demyelinating lesions. These assessments were performed by counting the numbers of neurons and glial cells on digital images of 100× microscopic fields in Nissl-stained sections consecutive to the MBP-stained sections in the entire caudate nucleus and the thalamic mediodorsal (MD) nucleus on the coronal section in MS and control brains. The caudate and thalamic MD nuclei were chosen because they had the highest frequency of DGM lesions. The neuropathologist who performed the counting was blinded to demyelinated/nondemyelinated DGM status and to case and control status. Neurons were identified by morphological criteria and assessed only when the nucleolus was clearly visible. No distinction between different types of glial cells was made in the Nissl-stained sections.

Neuronal and glial cell densities were transformed with the Abercrombie correction (25) and expressed as number of cells per square millimeter. Values of neuronal density were compared between the demyelinated DGM, nondemyelinated DGM, and control DGM of the same nucleus using a Kruskal-Wallis test with post hoc comparison between groups (method after Conover) (BrightStat software, http://www.brightstat.com). Values of glial cells density were compared between demyelinated and nondemyelinated DGM of the same nucleus using the Mann-Whitney U test. Correlations between neuronal density in MS nondemyelinated DGM, clinical parameters (age,disease duration), and extent of WM demyelination were evaluated using Spearman rank correlation test. Differences inneuronal density in MS nondemyelinated DGM between RRand SP MS cases were evaluated using the Mann-Whitney U test.

Assessment of Neuronal Size and Shape

Assessment of neuronal size and shape was performed by 2 observers who were blinded to case/control status on digital images of randomly selected 100× microscopic fields in Nissl-stained sections using the Axiovision software (Carl Zeiss MicroImaging). Twenty fields were evaluated for each MS and control brain in the caudate and thalamic MD nuclei, and at least 25 neurons were sampled for each case. Neurons were identified and assessed as previously described. Area and perimeter of each neuron were measured on the digital image. Neuronal shape was evaluated using the f-circle formula, where f-circle is defined by 4 π × ([neuronal area]/[neuronal perimeter]2) (26). Increasing an f-circle value indicates that theobject shape is more circular and less elliptical. Mean neuronalarea and f-circle were compared between demyelinated DGM,nondemyelinated DGM, and control DGM of thesame nucleus using a Kruskal-Wallis test with post hoc comparison between groups. Correlations between mean neuronal area in MS nondemyelinated DGM, clinical parameters (age, disease duration), and extent of WM demyelination were evaluated using Spearman rank correlation test. Differences in the mean neuronal area in MS nondemyelinated DGM between RR and SP MS cases were evaluated using the Mann-Whitney U test.

Assessment of Synaptic Density

Synaptic density was evaluated in the caudate and thalamic MD nuclei of MS cases by measuring optical density of synaptophysin immunostaining according to previously published methods (27-29). Optical density of pixels was assessed on digital images of 40× microscopic fields using Scion Image software (version 4.0.2; National Institutes of Health). Vessels and neurons were manually excluded. To ensure that the synaptophysin labeling was not saturated, we constructed a DAB saturation curve in which the DAB reaction was stopped at different time points and the optical density was measured at each point. Saturation was observed at incubation times greater than 120 seconds. We, therefore, selected 80 seconds as an optimal DAB incubation time, and this was applied to all synaptophysin-stained sections. Optical density measured in each DGM lesion was compared with optical density measured in the adjacent nondemyelinated DGM in the same nucleus in the same section. All DGM lesions in the caudate and thalamic MD nuclei were evaluated, with the exception of active demyelinating lesions. Lesions that involved the whole DGM nucleus, for which the adjacentnondemyelinated DGM was not assessable, were also excluded. The optical density values between demyelinated DGM and adjacent nondemyelinated DGM in the same section were compared using a Mann-Whitney U test.

Results

Distribution and Major Pathological Features of DGM Lesions

Most DGM lesions were very difficult to detect in Luxol fast blue-stained sections, whereas they were easily recognizable with MBP immunostaining (Fig. 1). In each case, several demyelinating lesions were identified in the DGM nuclei in MBP-stained sections (Tables 3 and 4); the numbers ranged from 0 to 9 lesions per case. The extent of DGM demyelination was highly variable between cases, ranging from 0% to 25.5% of the total area of the DGM in the coronal sections (mean, 5.05%; median, 2.25%) (Table 4).

FIGURE 1.

Myelin basic protein immunohistochemistry of deep gray matter (DGM) multiple sclerosis lesion. (A) Pure DGM lesion, putamen. There is a strict respect for the GM/white matter (WM) border. Myelin basic protein immunostaining, original magnification: 2.5×. Scale bar = 400 μm. (B) Normal putamen (control). Myelin basic protein immunostaining, original magnification: 10×. Scale bar = 100 μm. (C) Hypercellular (active) mixed WM/DGM active lesion centered on a blood vessel that involves the caudate nucleus. Myelin basic protein immunostaining, original magnification: 2.5×. Scale bar = 400 μm. (D) Normal caudate nucleus (control). Myelin basic protein immunostaining, original magnification: 5×. Scale bar = 200 μm. (E) Deep GM lesion, caudate nucleus. There is demyelination of the GM, but the WM tracts crossing the caudate nucleus are spared. Myelin basic protein immunostaining, original magnification: 10×. Scale bar = 100 μm. (F) Normal caudate nucleus (control), showing a WM tract crossing the nucleus with myelin staining evident in the GM. Myelin basic protein immunostaining, original magnification: 10×. Scale bar = 100 μm. (G) Periventricular DGM lesion, thalamus. There is patchy myelin preservation. Myelin basic protein immunostaining, original magnification: 2.5×. Scale bar = 400 μm. (H) Deep GM lesion, thalamus. Myelin basic protein immunostaining, original magnification: 20×. Scale bar = 50 μm.

View this table:
TABLE 3.
View this table:
TABLE 4.

Demyelinating lesions were mostly detected in the periventricular areas of the caudate nucleus (Figs. 1C, E) and of the thalamus (medial and anterior thalamic nuclei) (Figs. 1G, H). In some cases, almost complete demyelination of the caudate nucleus in the coronal section was observed. No lesions were found in the ventral thalamic nuclei (Table 3).

Demyelinating lesions were also observed in the putamen (Fig. 1A), pallidum, hypothalamus, amygdala, claustrum, and substantia nigra (Table 3). Lesions in the putamen and pallidum were usually centered on blood vessels (Fig. 1A), whereas lesions in the caudate, thalamus, and hypothalamus commonly seemed to follow the ventricular surface (Figs. 1C, E, G). Despite their location close to the cerebrospinal fluid, the mamillary bodies were always devoid of demyelinating lesions.

Most lesions involving the DGM were mixed WM/GM lesions (62.9%); however, several pure DGM lesions were also observed. In many lesions involving the DGM, variable proportions of the lesion border showed a strict respect for the GM/WM boundary, stopping suddenly at the anatomical border with the WM and sometimes sparing the WM tracts crossing a GM nucleus (Fig. 1E). This pattern has been previously described in cortical and spinal GM lesions (2, 6) and was observed both in pure DGM lesions and in mixed WM/DGM lesions.

No correlation was found between the extent of DGM demyelination and the extent of WM demyelination, nor with clinical variables such as age, disease duration, or disease course. There was, however, a correlation between the extent of DGM demyelination and the extent of cortical demyelination (Spearman ρ = 0.765, p = 0.001) (Fig. 2).

FIGURE 2.

Scatter plot demonstrating the relationship between the area of deep gray matter (DGM) demyelination and the area demyelinated in the cortex (Spearman ρ = 0.765, p = 0.001).

The DGM lesions showed no overt alterations of tissue morphology, except for the complete loss of MBP immunostaining, at visual inspection. Gross tissue destruction and scarring were instead usually clearly evident in WM lesions; axon thickening, deformation, and loss were clearly apparent in Bielschowsky preparations and in sections stained for neurofilaments. Only minimal gliosis was present in DGM lesions, as also has been described in cortical GM lesions (13, 30). Gliosis indicated by increased numbers of GFAP-positive astrocytes was, however, evident in all chronic WM lesions (not shown).

WM and Cortical Lesions

In all cases, there were several partially confluent demyelinating lesions in the WM. Active WM lesions (hypercellular lesions with macrophages containing Luxol fast blue-positive and MBP-positive material) were identified in 6 cases. The extent of WM demyelination ranged from 2% to 75% of the total area of the WM in the coronal sections (mean, 14.54%; median, 7.25%). Several cortical lesions were observed in most MS brains, and in many cases, there was widespread subpial demyelination. The extent of cortical demyelination ranged from 0% to 49.5% of the total area of the cortex in the coronal section (mean, 9%; median, 2.75%).

Inflammation in Active Demyelinating DGM Lesions

Most DGM lesions were chronic inactive lesions. Three chronic active lesions (i.e. with increased density of HLA-DR-positive cells at the border) were detected. In these lesions, inflammatory cells were mainly HLA-DR-positive microglia with very few macrophages. There were 6 active DGM lesions, hypercellular with HLA-DR-positive cells throughout the lesion (Figs. 3A, 4A, B); all of these were mixed WM/GM lesions. The GM portions of active lesions showed overall less pronounced inflammatory features (i.e. HLA-DR immunoreactivity) compared with the WM parts of the same lesion. Most inflammatory cells in the GM part of the lesions were ramified HLA-DR-positive microglia, and less frequently macrophages, whereas macrophages were predominant in the WM portions. Activated microglia and macrophages in DGM lesions were more common at the lesion border and near inflamed vessels; HLA-DR-positive cells were often observed in close contact with neurons (Figs. 4A, B). Foamy macrophages laden with MBP-positive material were abundant in the WM part of the lesions, but only extremely few macrophages in the GM portions of the lesions contained MBP-positive material (Fig. 3B). Myeloperoxidase immunostaining was observed mainly on intravascular leukocytes; most macrophages and microglia in active DGM lesions did not show any appreciable MPO immunostaining (not shown).

FIGURE 3.

Immunohistochemistry of deep gray matter (DGM) and white matter (WM) multiple sclerosis lesions. (A) Border of an active mixed WM/DGM lesion in the caudate nucleus showing a high density of HLA-DR-positive microglia and macrophages. HLA-DR (purple)/myelin basic protein ([MBP] brown) double immunostaining, original magnification: 10×. Scale bar = 100 μm. (B) Myelin-laden macrophages in an active DGM lesion in the caudate nucleus. CD68 (purple)/MBP (brown) double immunostaining, original magnification: 100×. Scale bar = 10 μm. (C) Immunoglobulin G immunostaining in an active WM lesion. Immunoglobulin G immunostaining, original magnification: 10×. Scale bar = 100 μm. (D) Immunoglobulin G immunostaining in active DGM lesion in the caudate nucleus; IgG immunostaining is much more intense and diffuse in the WM part (lower left corner) than in the DGM part of this active mixed WM/DGM lesion. Immunoglobulin G immunostaining, original magnification: 10×. Scale bar = 100 μm. (E) Fibrinogen immunostaining in an active WM lesion. Fibrinogen immunostaining, original magnification: 10×. Scale bar = 100 μm. (F) Fibrinogen immunostaining in active DGM lesion in the caudate nucleus; the immunostaining is much more intense and diffuse in the WM part (lower left corner) than in the DGM part of this active mixed WM/DGM lesion. Fibrinogen immunostaining, original magnification: 10×. Scale bar = 100 μm. (G) Perivascular and parenchymal CD3+ T lymphocytes in active DGM lesion in the putamen (CD3 immunostaining, original magnification: 10×. Scale bar = 100 μm.

FIGURE 4.

Immunohistochemistry of deep gray matter (DGM) multiple sclerosis lesions. (A) HLA-DR+ cells closely apposed to neurons at the border of an active DGM lesion in the caudate nucleus. HLA-DR (purple)/myelin basic protein ([MBP] brown) double immunostaining, original magnification: 100×. Scale bar = 10 μm. (B) CD68+ cells closely apposed to neurons in an active DGM lesion in the claustrum (CD68 immunostaining, original magnification: 100×. Scale bar = 10 μm. (C) Immunoglobulin G-positive neurons in normal-appearing DGM (NADGM) close to an active DGM lesion in the putamen. Immunoglobulin G immunostaining, original magnification: 20×. Scale bar = 50 μm. (D) Activated HLA-DR-positive microglia in NADGM in the thalamus. HLA-DR immunostaining, original magnification: 20×. Scale bar = 50 μm. This pattern of microglia activation in the NADGM was diffusely present in different DGM structures. (E) β-Amyloid precursor protein (APP)-positive axonal spheroids in active WM lesion. β-Amyloid precursor protein immunostaining, original magnification: 40×. Scale bar = 25 μm. (F) The APP-positive axonal spheroids in active DGM lesion in the caudate nucleus; the density of APP-positive axonal spheroids was much higher in the WM part than in the DGM part of active mixed WM/DGM lesions. β-Amyloid precursor protein immunostaining, original magnification: 40×. Scale bar = 25 μm. (G) Reduction of excitatory amino acid transporter 1 (EAAT1) immunostaining in an active DGM lesion in the caudate nucleus. The reduction of EAAT1 immunostaining is greatest in the area with more intense inflammation (highest density of HLA-DR-positive microglia and macrophages). Excitatory amino acid transporter 1 (purple)/HLA-DR (brown) double immunostaining, original magnification: 10×. Scale bar = 100 μm. (H) Excitatory amino acid transporter 1 immunostaining in a normal caudate nucleus (control). Excitatory amino acid transporter 1 immunostaining, original magnification: 10×. Scale bar = 100 μm.

Several blood vessels in the WM part of the lesions showed conspicuous perivascular inflammatory infiltrates of CD3+ and CD20+ lymphocytes; scattered CD3+ lymphocytes were observed also in the parenchyma. Perivascular inflammatory infiltrates of CD3+ and CD20+ lymphocytes were present also in the GM part of the lesions but to a much lesser extent (Fig. 3G). Rare CD138+ plasma cells were observed in the perivascular inflammatory infiltrates both in WM and GM parts of the DGM lesions. Overall, the number of CD3+ lymphocytes in the GM part of the DGM lesions was lower than in the WM part of the lesions (mean density of CD3+ lymphocytes in the GM, 5.5/mm2; mean density of CD3+ lymphocytes in the WM, 88.3/mm2).

Endothelial cell expression of VCAM-1 was observed on most small vessels both in the WM and the GM parts of the lesions without appreciable differences. Perivascular fibrinogen immunostaining was observed in the GM part of the lesions near inflamed vessels, albeit to a lower extent than in the WM part of the lesions (Figs. 3C, D). Diffuse IgG immunostaining was observed in the GM part of the lesions, which was more pronounced near inflamed vessels but to a lesser extent than in the WM part of the lesions (Figs. 3E, F). Scattered neurons showing cytoplasmic IgG immunostaining were observed in the GM part of the lesions, not only within the lesions but also at some distance from lesion borders (Fig. 4C).

Rare APP-positive axonal spheroids were observed in the GM part of the lesions; they were far more abundant in the WM part of the lesions (mean density of APP-positive axonal spheroids in the GM, 12.8/mm2; mean density of APP-positive axonal spheroids in the WM, 166.4/mm2) (Figs. 4E, F). Scattered reactive GFAP-positive astrocytes were observed both in the WM and the GM parts of the lesions.

A comparison of inflammatory features between the GM and WM portions of active WM/DGM lesions is shown in Table 5. The semiquantitative evaluation showed a reduction of EAAT1 and EAAT2 immunostaining in active DGM lesions compared with nondemyelinated DGM, particularly in the border areas with the highest numbers of macrophages and activated microglia. A relative increase of EAAT1 and EAAT2 immunostaining was observed in the nondemyelinated DGM immediately surrounding active DGM lesions (Fig. 4G). No apparent reduction of synaptophysin immunostaining was detected in active DGM lesions compared with adjacent nondemyelinated GM.

View this table:
TABLE 5.

Inflammation in the NADGM

Distinctive inflammatory changes such as the widespread presence of HLA-DR-positive activated microglia and occasional perivascular lymphocyte infiltration were found in MS NADGM distant from focal demyelinating lesions (Fig. 4D). Diffuse inflammation in the NADGM seemed to be proportional, even if to a lower degree, to diffuse inflammation in theNAWM in the same case and was quite homogeneous among different GM nuclei in each case. No CD3+ or CD20+ lymphocytes were observed in the NADGM parenchyma, and no extravascular fibrinogen or IgG immunostaining was detected.

Immunohistochemical Findings in Normal Control and ALS Cases

No demyelinating lesions were found in the DGM in ALS/control cases (Figs. 1B, D, F). Extremely rare HLA-DR, MPO, and CD68+ cells were observed inside blood vessels. Immunoglobulin G and fibrinogen immunostaining was observed only inside blood vessels. No CD3, CD20, CD138, VCAM-1, APP, or GFAP immunostaining was found. The EAAT1 and EAAT2 diffusely stained astrocytes in ALS/control DGM (Fig. 4H).

Neuronal Loss

There was a reduction of neuronal density both in the caudate nucleus and in the thalamic MD nucleus, in the demyelinated DGM in comparison with nondemyelinated DGM (−35.5% caudate nucleus, p < 0.001; −20.3% thalamic MD nucleus, p < 0.01), and in nondemyelinated DGM in comparison with control DGM (−33.0% caudate nucleus, p < 0.001; −33.7% thalamic MD nucleus, p < 0.001) (Table 6). Nocorrelation was found between the neuronal density in MS nondemyelinated DGM and the extent of WM demyelination, age, duration of disease, or disease course.

View this table:
TABLE 6.

Glial Cell Loss

There was a significant reduction of glial cell density in the caudate nucleus and in the thalamic MD nucleus in demyelinated DGM compared with nondemyelinated DGM(−70.3% caudate nucleus and thalamic MD nucleus, p < 0.001) (Table 6).

Neuronal Size/Shape

There was variable reduction of the area of neurons in the caudate nucleus and in the thalamic MD nucleus in the demyelinated DGM compared with the nondemyelinated DGM (−9.8% caudate nucleus, p < 0.05; −13.8% thalamic MD nucleus, p < 0.01), and in nondemyelinated DGM compared with control DGM (−26.3% caudate nucleus, p< 0.001; −23.5% thalamic MD nucleus, p < 0.001) (Table 7). No correlation was found between the neuronal area and the extent of WM demyelination, age, or disease course. There was an inverse correlation between the neuronal area in MS nondemyelinated DGM and the disease duration (Spearman ρ = −0.719, p = 0.029).

View this table:
TABLE 7.

A lower f-circle score indicating a less spherical neuronal shape was observed both in the caudate nucleus and in the thalamic MD nucleus in the demyelinated DGM compared with nondemyelinated DGM (−16.6% caudate nucleus, p < 0.01; −13.1% thalamic MD nucleus, p < 0.01), and in nondemyelinated DGM compared with control DGM (−16.2% caudate nucleus, p < 0.01; −13.2% thalamic MD nucleus, p < 0.01) (Table 7).

Synaptic Density

No significant differences in optical density of synaptophysin immunostaining were observed between DGM lesions and adjacent nondemyelinated DGM in the caudate and thalamic MD nuclei in MS brains. This is similar to what has been described using similar approaches for GM lesions in the cerebral cortex (3), the cerebellar cortex (5), and in the hippocampus (8). By contrast, reduction of synaptic density in type I (leucocortical) cortical lesions has been described in another study using a different method (45).

Discussion

We investigated the main features of DGM pathology (i.e. demyelination, inflammation, and neuronal, axonal, and synaptic pathology) in an autopsy series of MS cases. Deep GM demyelination is frequent, especially in the caudate and in the medial and anterior thalamic nuclei. Inflammation in the DGM shares similar pathological features with inflammation in the WM, but is less intense, with a preponderance of activated microglia, scarce myelin-laden macrophages, and a lower degree of axonal injury and blood-brain barrier disruption. In addition, we found that neurons are lost not only in DGM lesions but also in the nondemyelinated DGM; there is also neuronal atrophy in MS DGM. We hypothesize that alterations of glutamate reuptake mechanisms might be particularly involved in the pathogenesis of DGM demyelination and of neuronal loss in MS DGM.

The distribution of DGM lesions is in accordance with the classic localization of MS lesions (i.e. mostly involving the periventricular areas and, thus, nuclei such as the caudate and the medial and anterior thalamic nuclei) and the areas close to veins and venules. The mamillary bodies are a notable exception because in our series, they were always devoid of demyelinating lesions despite their location close to the cerebrospinal fluid.

Most lesions involving the DGM were mixed WM/GM lesions; this differs from what is found in the MS cortex, in which mixed WM/GM lesions (i.e. leucocortical lesions, type I [2]) represent only 5% to 30% of lesions (1-3). The strict delimitation of many of the DGM lesions by the anatomical GM/WM boundaries was also described in cortical and spinalGM lesions (2, 6). The reverse (i.e. WM lesions that suddenly stop at the border with the GM) was never observed in our series, nor to our knowledge has it been described in other studies of MS GM pathology; this suggests that whereas WM lesions are able to cross the anatomical border with the GM, the reverse does not seem to occur. This peculiar lesion pattern could be related to different mechanisms of demyelination in the WM and the GM.

We documented axonal damage in active DGM lesions but to a much lesser extent than in active WM lesions. Overall, the extent of axonal loss and of tissue destruction, gliosis, and scarring seems to be far more severe in WM lesions than in DGM lesions, in which tissue architecture isusually well preserved. A similar difference was also observed in cortical lesions compared with WM lesions (1-3). The reasons for these differences in the extent of tissue damage are not clear but again point to different mechanisms of demyelination and tissue damage in GM and WM lesions.

The extent of inflammation in active DGM lesions seems to lie between the intense destructive inflammation of WM lesions and the minimal inflammation of pure cortical lesions (1, 11-13). This correlates with and might explain the greater degree of neuronal loss in DGM lesions compared with cortical lesions.

The reasons underlying the different phenotypes of inflammation between DGM and WM are not clear. The level of expression of VCAM-1 does not seem to correlate with or explain the difference in the extent of inflammation. One possible explanation might be the relatively lower levels of myelin antigens in the GM, thus eliciting lesser inflammatory responses; however, on this basis, a much lower frequency of GM lesions in MS would probably be expected, whereas on the contrary, GM lesions outnumber WM lesions in many MS cases (2-4). Alternatively, different effector mechanisms of inflammation might be involved in GM and WM. We observed a reduced expression of glutamate transporters EAAT1 and EAAT2 within active DGM lesions, particularly in areas with more intense inflammation; the expression was increased in the nondemyelinated DGM surrounding active DGM lesions. The EAAT1 and EAAT2 are the principal regulators of extracellular glutamate concentration (31); a reduced function of EAAT1 and EAAT2 can lead to toxic extracellular glutamate levels (32, 33). Several soluble mediators of inflammation are known to impair the expression and functioning of EAAT1 and EAAT2 (34-36); activated microglia and macrophages can be an additional source of glutamate production during inflammation (37, 38). The upregulation of EAAT1 and EAAT2 in the nondemyelinated DGM surrounding active DGM lesions may represent an attempt to buffer excess glutamate diffusing from the areas of active inflammation (39). Similar changes in the expression of glutamate transporters have been described in cortical lesions (30) and WM lesions (38), and suggest an alteration of glutamate homeostasis; resulting excitotoxic damage to neurons and oligodendrocytes may be a possible causal factor of demyelination and neuronal loss in MS DGM during inflammation. The consequences of altered glutamate homeostasis are presumably much greater in the DGM, which harbors a higher number of glutamatergic synapses and a higher content of glutamate than the WM.

Interactions between neurons and inflammatory cells might also contribute to the differences in the inflammatory features between DGM and WM. Inflammation might be downregulated in the GM, resulting in lower tissue damage, as neurons are known to inhibit microglia/macrophage activation and function, through expression of CD22 and CD200 (40-43), and to induce a switch of T cells to a regulatory phenotype (44). Further studies are warranted to investigate the features, regulation, and consequences of MS autoimmune inflammation in the unique milieu of the GM and the interplay between inflammatory cells and resident CNS cells.

Remarkably, mild diffuse inflammatory alterations are also frequently observed in the nondemyelinated DGM distant from focal demyelinating lesions. The features of inflammation in the nondemyelinated DGM are quite similar to those commonly found in MS NAWM, with diffuse microglial activation and occasional perivascular inflammatory infiltrates. Normal-appearing WM/normal-appearing GM diffuse inflammatory changes have been hypothesized to be the consequences of a spreading of the autoimmune inflammatory process outside focal lesions (4) but may also represent secondary responses to widespread anterograde and retrograde axonal degeneration.

In addition to demyelination, there is neuronal loss in MS DGM. Neuronal loss has also been described in other GM structures in MS brains including cerebral cortex (1, 2, 45), cerebellar cortex (5), and hippocampus (8). A correlation between atrophy and neuronal loss in the thalamic MD nucleus in MS brains was previously described in a postmortem MRI study (19) in which estimated neuronal loss was 35%-very similar to the value observed in this study. The extent of neuronal loss in MS DGM seems to be higher than in the MS cortex (1, 8, 45). This is consistent with the findings of recent MRI studies in which atrophy of the DGM preceded cortical atrophy in early MS (21, 22).

Neuronal loss in the demyelinated GM may relate to acute damage to neurons during inflammation or to loss of neurons in chronic lesions caused by lack of trophic support. The close association of HLA-DR-positive macrophages and microglia with neurons in active DGM lesions supports the hypothesis of acute neuronal damage during DGM demyelination. Myeloperoxidase expression by activated microglia has also been described to correlate with cortical demyelination in a recent study (46), but we did not observe MPO immunoreactivity in activated microglia or macrophages in active DGM lesions in this study. Immunoglobulin G-immunoreactive neurons were sometimes found in active DGM lesions. Whereas the presence of humoral responses targeting neuronal antigens cannot be completely excluded a priori, binding of IgG to dystrophic or damaged neurons might indicate neuronal damage rather than being causative of the damage.

Neuronal loss and atrophy in the caudate and thalamic MD nuclei nondemyelinated GM might be related to retrograde degeneration of neurons, the axons of which had been transected in WM lesions and/or to deafferentation of DGM neurons (47, 48). We found no correlation between neuronal loss or atrophy in the DGM and the extent of WM demyelination, but in view of the random distribution of WM lesions, the extent of WM demyelination in a coronal section may not adequately reflect the total WM lesion burden. Neuronal atrophy in MS has also been described in other GM structures such as the hippocampus (8) and the cerebral cortex (45).

The DGM nuclei are critical links of cortical-subcortical circuits controlling several neurological functions, and their injury or dysfunction may lead to motor, sensory, cerebellar, and cognitive deficits. Overt motor symptoms of DGM involvement (e.g. extrapyramidal signs and symptoms, dyskinetic movements) are infrequent in MS patients (49); however, DGM dysfunction may also lead to more subtle impairment in motor planning. The DGM nuclei also have a prominent role in cognition and behavior, and a correlation between thalamic atrophy and cognitive deficits in MS has been described by recent MRI studies (17, 50). Another MRI study described acorrelation between fatigue, a common symptom in MS patients, and decreased tissue perfusion in the DGM nuclei (51), but the clinical relevance of DGM lesions deserves further investigation. Similar to cortical demyelination (9, 10), DGM demyelination is probably not easy to detect with conventional MRI because tissue destruction is less pronounced and the contrast between NADGM and DGM lesions is low. Postmortem MRI studies evaluating the sensitivity of MRI for DGM demyelination are lacking at the present time.

In conclusion, involvement of the DGM in MS is frequent and may contribute to disability and cognitive deficits in MS patients. Demyelination, neuronal and axonal loss, and neuron atrophy are all common pathological features of DGM involvement in MS. The effector mechanisms of inflammation and tissue damage seem to differ between WM and DGM lesions, and further studies are needed to elucidate the mechanisms of inflammation in the GM and the complex interactions between neuroepithelial and immune cells.

Footnotes

  • This study was supported by Grant No. 2004.1424 from the Compagnia di San Paolo and by the Regione Piemonte.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
View Abstract