Meningeal B-cell follicles in secondary progressive multiple sclerosis associate with early onset of disease and severe cortical pathology

Abstract

Intrathecal antibody production is a hallmark of multiple sclerosis and humoral immunity is thought to play an important role in the inflammatory response and development of demyelinated lesions. The presence of lymphoid follicle-like structures in the cerebral meninges of some multiple sclerosis patients indicates that B-cell maturation can be sustained locally within the CNS and contribute to the establishment of a compartmentalized humoral immune response. In this study we examined the distribution of ectopic B-cell follicles in multiple sclerosis cases with primary and secondary progressive clinical courses to determine their association with clinical and neuropathological features. A detailed immunohistochemical and morphometric analysis was performed on post-mortem brain tissue samples from 29 secondary progressive (SP) and 7 primary progressive (PP) multiple sclerosis cases. B-cell follicles were detected in the meninges entering the cerebral sulci of 41.4% of the SPMS cases, but not in PPMS cases. The SPMS cases with follicles significantly differed from those without with respect to a younger age at multiple sclerosis onset, irreversible disability and death and more pronounced demyelination, microglia activation and loss of neurites in the cerebral cortex. Cortical demyelination in these SPMS cases was also more severe than in PPMS cases. Notably, all meningeal B-cell follicles were found adjacent to large subpial cortical lesions, suggesting that soluble factors diffusing from these structures have a pathogenic role. These data support an immunopathogenetic mechanism whereby B-cell follicles developing in the multiple sclerosis meninges exacerbate the detrimental effects of humoral immunity with a subsequent major impact on the integrity of the cortical structures.

Introduction

Multiple sclerosis is a CNS-specific, putatively autoimmune disease that causes inflammation in the brain and spinal cord and results in demyelination, axonal damage and neuronal loss. As in other chronic inflammatory diseases, the CNS of patients with multiple sclerosis shows infiltration of activated T cells and macrophages, dendritic cells, B cells and plasma cells. This implies potential roles for both cellular and humoral immune responses and the engagement of different immunopathological effector mechanisms in CNS tissue destruction. While the association of multiple sclerosis with certain MHC class II genes and an extensive literature in experimental models of multiple sclerosis favour a T-cell-mediated pathogenesis, an important role for humoral immunity in the development of multiple sclerosis lesions is also recognized (Sospedra and Martin, 2005; Owens et al., 2006; Frohman et al., 2006). Elevated CNS levels of immunoglobulins (Ig) and the presence of oligoclonal IgG in the CSF are the most consistent immunological abnormalities in multiple sclerosis (Tourtellotte et al., 1984; Archelos et al., 2000). Such intrathecal immunoglobulin synthesis is thought to be sustained by long-lived plasma cells recruited to or differentiating within the CNS (Prineas and Wright, 1978). Neuropathological studies indicate that antibody-mediated demyelination is one of the predominant pathogenetic mechanisms involved in white matter lesion formation in a substantial proportion of multiple sclerosis patients (Lucchinetti et al., 2000). Capping of surface IgG on microglia/macrophages engaged in myelin breakdown and co-deposition of IgG and activated complement fragments or complexes at the borders of active multiple sclerosis lesions strongly implicate antibodies as effectors of demyelination (Prineas and Graham, 1981; Storch et al., 1988). Despite detection of antibodies recognizing myelin and neuronal antigens in serum, CSF and demyelinating lesions of multiple sclerosis patients, it is still unclear whether such autoantibodies have a pathogenic role (Sospedra and Martin, 2005; Owens et al., 2006). The recent finding that CSF oligoclonal IgG bind Epstein–Barr virus proteins indicates that the compartmentalized B-cell response in multiple sclerosis could be sustained by a viral infection (Rand et al., 2000; Cepok et al., 2005).

Recently, molecular analyses of the Ig variable gene region of B cells and plasma cells isolated from the demyelinated lesions or CSF of multiple sclerosis patients have provided evidence that the intrathecal humoral immune response in multiple sclerosis is antigen-driven and that dominant B-cell clonotypes persist over time in the CNS compartment (Qin et al., 1998; Baranzini et al., 1999; Colombo et al., 2000; Owens et al., 2003; Qin et al., 2003; Colombo et al., 2003). Remarkably, it was found that a complete recapitulation of B-cell differentiation resembling a germinal centre reaction occurs in the CSF of multiple sclerosis patients (Corcione et al., 2004). By studying the cellular composition and organization of the inflammatory infiltrates in post-mortem brain tissue from a limited number of multiple sclerosis cases, we have shown that in secondary progressive multiple sclerosis (SPMS) the inflamed cerebral meninges contain structures that are strikingly similar to secondary B-cell follicles containing germinal centres (Serafini et al., 2004). The ectopic follicles found in SPMS comprised proliferating B cells, plasma cells, T cells and a network of follicular dendritic cells (FDC), which are essential for B-cell maturation due to their ability to retain antigens on their membrane and to stimulate B-cell proliferation and survival (Park and Choi, 2005). These findings indicate that the inflamed CNS is able to sustain B-cell responses through an inherent predisposition of the meningeal compartment to favour the organization of ectopic lymphoid tissue. In other organ-specific autoimmune diseases, such as Hashimoto's thyroiditis, myasthenia gravis and Sjogren's syndrome, ectopic germinal centres are a source of disease-relevant antibodies, which suggests that such abnormal formations may contribute to the pathogenic process through local amplification of the autoimmune response (Aloisi and Pujol-Borell, 2006).

Based on the assumption that a better knowledge of the events associated with ectopic follicle formation in multiple sclerosis may shed light on the still elusive immune-mediated mechanisms underlying CNS tissue destruction, we have begun to ascertain the pathological relevance of such abnormal structures by analysing autopsy brain tissue from a larger sample of multiple sclerosis cases with progressive disease courses, and to search for any relationship between presence of follicles, clinical course and neuropathological features.

Material and methods

Demographic and clinical data

This study was performed on autopsy brain tissue from 29 cases with SPMS and 7 with primary progressive (PP) multiple sclerosis. All tissues were obtained from the UK Multiple Sclerosis Tissue Bank at Imperial College, except for MSG1 that was provided by the Institute of Pathological Anatomy, U.C.S.C. Policlinico A. Gemelli, Rome, Italy. The multiple sclerosis cases selected for this study had a range of ages at death (35–81 years), ages at multiple sclerosis onset (10–56 years) and disease duration (6–52 years), reflecting the variability of the multiple sclerosis population. The demographic and clinical data for each multiple sclerosis group and the 36 individual cases are shown in Tables 1 and 2, respectively. This study includes three SPMS (MS79, MS80 and MS85) and two PPMS (MS83 and MSG1) cases that were already examined in Serafini et al. (2004). The UK Multiple Sclerosis Tissue Bank also provided control post-mortem brain tissues from three patients without evidence of neurological disease or neuropathological alterations: C4, a male patient (57-year-old) who died from bronchial cancer; C20, a female patient (82-year-old) who died from cardiac failure and C25, a male patient (35-year-old) who died with tongue carcinoma.

All post-mortem tissues were obtained via a UK prospective donor scheme with full ethical approval (MREC/02/2/39) except for case MSG1 where separate ethical approval was obtained from the Ethics Committee of the Istituto Superiore di Sanità, Rome. Confirmation of the diagnosis of multiple sclerosis for each case was provided by Dr F. Roncaroli (Consultant Neuropathologist, Department of Neuropathology, Imperial College London). A summary of the clinical history for each case was prepared by a clinical neurologist with an interest in multiple sclerosis (R.N.). Details of the history included date of symptom onset, date of diagnosis, number, date and character of relapses, date of onset of progression, date of wheelchair dependence (EDSS 7 equivalent) and date of death (EDSS 10 equivalent). Details of any co-morbidities were noted as well as treatments both for multiple sclerosis and other co-morbidities (Table 2).

Tissue and lesion classification

For each multiple sclerosis case, eight cerebral tissue blocks (4 cm³), including cortical and periventricular areas with white matter (WM) demyelination and an intact meningeal compartment were examined. In order to sample as widely as possible from the cerebrum and to avoid sample bias, tissue blocks from frontal, temporal, parietal, occipital and central (insula) lobes were selected from the digital images of coronal brain slices acquired at the time of dissection at the UK Multiple Sclerosis Tissue Bank. Other CNS regions (mamillary bodies, basal ganglia, pons, medulla oblongata, cerebellum and spinal cord) were also analysed in 19 SPMS and 5 PPMS cases. The post-mortem delay ranged between 5 and 44 h (median time, 16 h). Four tissue blocks from MSG1 were formalin fixed and paraffin embedded. Tissue blocks from the remaining cases (eight blocks per case) were snap frozen in isopentane on dry ice (four blocks) or fixed in 4% paraformaldehyde (mean fixation time = 17 ± 6 days), cryoprotected in a 30% sucrose, frozen in cooled isopentane and stored at −75°C (four blocks).

Inflammatory cell infiltrates were studied using haematoxylin–eosin staining. The degree of WM inflammation was evaluated manually by counting the number of DAPI-positive nuclei in four randomly selected perivascular infiltrates for each multiple sclerosis case (Table 2). The extent of demyelination and degree of lesion activity were evaluated by combining Luxol Fast Blue (LFB) histochemical staining with major histocompatibility complex (MHC) class II molecule immunostaining (see later). Because grey matter lesions (GML) are difficult to identify in sections stained with LFB, we also used myelin basic protein (MBP) and myelin oligodendrocyte glycoprotein (MOG) immunohistochemistry (Table 3). Since the extent of GM demyelination appeared almost identical when probed with anti-MBP and anti-MOG antibodies (data not shown), MOG immunostaining was used to quantify the areas of both cortical and WM demyelination. The inflammatory activity of GML and WML was defined by the pattern of MHC class II staining: active lesions contained numerous MHC class II+ cells in the lesion core and at the lesion border; chronic active lesions had a border of MHC class II+ cells and a lower number in the core; chronic inactive lesions had a very low MHC class II+ cell density throughout the lesion.

Immunohistochemistry

Air dried, acetone fixed (+4°C), 10 μm-thick cryosections, cut from the 4% paraformaldehyde fixed and snap frozen tissue blocks, were rehydrated with PBS and immunostained with the monoclonal or polyclonal antibodies listed in Table 3. Post-fixed sections were subjected to an antigen retrieval procedure, either microwave treatment in citrate buffer 10 mM (pH 6.0) or incubation in target retrieval solution (Dako; 90°C) or permeabilization with cold methanol (10 min at –20°C). Sections were then incubated for 30 min with 0.1% H₂O₂ in PBS to eliminate endogenous peroxidase activity, for 1 h with 10% of normal sera, and overnight at 4°C with the primary antibodies diluted in PBS containing 0.2% Triton 100-X and 1% BSA or 1% normal sera. Binding of biotinylated secondary antibodies (Jackson Immunoresearch Laboratories) was visualized with the avidin–biotin horseradish peroxidase complex (ABC Vectastain Elite kit; Vector Laboratories, Burlingame, CA) followed by 3,3′-diaminobenzidine (DAB) (Sigma Chemical Co., St Louis, MO) as substrate. All sections were counterstained with haematoxylin, sealed with Depex Polystyrene (DPX) and viewed with a Nikon E1000M microscope. Images were captured with a QICAM digital camera (QImaging Inc.), and analysed using Image Pro Plus software (Media Cybernetics Inc.). Negative controls included IgG isotype controls or preimmune serum, or omission of the primary antibody.

Immunofluorescence

Cryosections of multiple sclerosis and control brain tissues, prepared as earlier, were also stained using a double immunofluorescence technique employing anti-MBP polyclonal antibody in combination with anti-MHC class II, SMI-32 or RT97 monoclonal antibodies, and anti-CD20 monoclonal antibody in combination with anti-CXCL13 or Ki67 polyclonal antibodies (Table 3). After an initial post-fixation in cold acetone and blockade with 5–10% normal sera in PBS, sections were incubated overnight at 4°C with the primary antibodies (listed in Table 3), diluted in PBS containing 0.2% Triton X-100 and 1% normal sera. Sections were then washed, treated with Cy3-, rhodamine- or fluorescein-conjugated anti-rabbit or anti-goat IgG and fluorescein- or rhodamine-conjugated anti-mouse IgG in PBS containing 1% normal sera for 1 h, washed again in PBS and then in distilled water. Some sections were counterstained with 4′, 6-diamidino-2-phenylindole (DAPI, Sigma) for the localization of cell nuclei. Finally, sections were coverslipped with aqueous mounting medium Vectashield (Vector Laboratories). For RT97 immunostaining, a streptavidin-conjugated Alexafluor-546 amplification (Invitrogen) was performed after incubation with the biotinylated anti-mouse Ig secondary antibody. Single immunostaining with a fluorescein-labelled anti-Ig-A, -G, -M polyclonal antibody was also performed. For negative controls, the primary antibodies were replaced with preimmune serum or IgG isotype controls. Slides were viewed under epi-fluorescence with the Nikon E1000M microscope and images captured as described.

Quantitative analysis of demyelinated lesion size and neurite loss

To determine the number and size of WML and GML, one section from each paraformaldehyde fixed tissue block (four blocks per case) was immunostained for MOG. Low magnification images (four frames per section) were acquired at 0.5 × magnification. The areas of demyelination and the total area of WM and GM were manually outlined for each section and then automatically analysed using the Image Pro Plus software. GML were characterized into type I (grey matter/white matter border), type II (intracortical) or type III (subpial) according to the criteria of Peterson et al. (2001).

Neurite density was analysed in chronic active type III GML and equivalent areas of subpial normal appearing grey matter (NAGM). Two serial sections for each tissue block from 12 B-cell follicle-positive SPMS, 12 B-cell follicle-negative SPMS (both groups containing chronic active GML and NAGM) and 3 control cases were immunostained with anti-MBP polyclonal antibody and anti-MHC class II monoclonal antibody (for classification of the lesion activity) and with anti-MBP polyclonal antibody and RT97 monoclonal antibody recognizing the phosphorylated 200 kDa neurofilament protein. For each multiple sclerosis case, we analysed one subpial (type III) chronic active cortical lesion and the NAGM distant (≥1 cm) from the lesion. Our analysis was limited to layers II and III of the cerebral cortex as these were always fully demyelinated and contained activated microglia in the chronic active GML group. For each lesion and NAGM area, four images were acquired at 400× magnification and the number of RT97+ neurites was manually counted by two investigators blinded to the case number (R.M. and O.H., inter-individual variability <2%). Total neurite density (per 0.1 mm²) was calculated from four counting squares (0.0036 mm²) located at the corners of each image to ensure that no RT97+ structures were counted twice.

Statistical analysis

Data are expressed as median and half the interquartile range (IQR/2), as a measure of variability, or as mean ± standard error of the mean (SEM). Comparisons between groups of multiple sclerosis patients were carried out by the Mann–Whitney U-test for continuous variables and by the chi-square test or Fisher's exact probability test for categorical variables. Comparisons within each multiple sclerosis subgroup for continuous variables were carried out by the Wilcoxon test for repeated measures. Spearman's rank correlation coefficients were calculated to investigate the linear relationship between pairs of variables. Survival curves for age at which the patient became a wheelchair user were established by the Kaplan–Meier method and comparisons between SPMS subgroups were assessed by the generalized Wilcoxon test. Bonferroni's correction has been adopted to control for multiple comparisons.

Results

Ectopic B-cell follicles are detected in SPMS, but not in PPMS

In agreement with previous findings (Serafini et al., 2004), prominent inflammatory cell infiltrates comprising CD3+ T cells, CD20+ B cells, CD138+ or Ig+ plasmablasts/plasma cells, CD68+ macrophages and large perivascular B-cell aggregates were detected in the cerebral leptomeninges of 12 out of 29 cases with SPMS, hereafter denoted F+ SPMS (Table 2). In all F+ SPMS cases, the meningeal B-cell aggregates were characterized as ectopic follicles with germinal centres based on the presence of a reticulum of CD35+ and CXCL13+ cells (bona fide stromal cells and FDC), proliferating Ki67+ B cells and Ig+ plasmablasts/plasma cells (Fig. 1A–D). Only in one case (MS85), did we observe one out of two meningeal follicles that lacked CD138+ plasma cells and a well-developed FDC network, indicating incomplete germinal centre formation. Meningeal follicles were present in 48 out of 96 cerebral tissue blocks examined, with a mean of 6.4 ± 3.1 follicles (range 2–14) found in sections from the eight blocks studied for each F+ SPMS case (Table 2). In most F+ SPMS cases, numerous B cells and plasma cells were also observed in the perivascular cuffs of WML (Fig. 1E). The ectopic follicles were mainly found in the frontal, temporal and parietal lobes and in the cingulate gyrus. We found only one tissue block containing a meningeal B-cell follicle in the occipital lobe and one in the insula. Although most of the selected tissue blocks (71%) contained the external surface of the brain, the follicles were always found along, in the depth and, more rarely, at the entrance of the cerebral sulci, in particular in the central and lateral sulci (Fig. 1H–J). Only sparse inflammatory infiltrates were detected in the meninges covering the external surface of the brain.

In the remaining 17 SPMS cases that lacked ectopic follicles (denominated F− SPMS), meningeal immune cell infiltration was absent (6 cases) or much less prominent (11 cases) than in the F+ SPMS group (Table 2). Only sparse B cells and plasma cells were detected in the moderately inflamed meninges of the F− SPMS cases, particularly at the tip of the cerebral sulci (Fig. 1F). The perivascular cuffs of WML were also less prominent in the F− SPMS group than in the F+ SPMS group and contained only occasional B cells and plasma cells (data not shown). Counts of DAPI-stained nuclei in four randomly selected WM perivascular cuffs for each case revealed that the number of perivascular cells in the F− SPMS group was significantly lower than in the F+ SPMS group [median values (range), 34 (29–50) versus 94 (65–103), respectively; P < 0.0001].

Ectopic follicles were not detected in the seven cases with PPMS, of which only three showed moderate meningeal inflammation with scattered B cells and plasma cells (Fig. 1G and Table 2). In PPMS cases, the size of the inflammatory cell infiltrates in WML was significantly lower than that observed in F+ SPMS cases [median value (range) of DAPI-stained nuclei counted in four WM perivascular cuffs = 29 (19–34); P = 0.0007]. Only a few B cells and plasma cells were detected in the perivascular cuffs of WML in the PPMS cases (data not shown).

To investigate whether ectopic follicles could form in CNS areas away from the cerebral cortex, we also analysed tissue blocks containing mamillary bodies, dentate nucleus, pons, medulla oblongata, cerebellum and spinal cord from 9 F+ SPMS, 10 F− SPMS and 5 PPMS cases. Using both haematoxylin/eosin staining and CD20 immunostaining, no B-cell follicles were found in any of these areas, with the exception of one F+ SPMS case (MS 121) which had prominent meningeal inflammation and one B-cell follicle in the brainstem (data not shown).

Earlier age at clinical onset and at death in the SPMS cases with ectopic B-cell follicles

When comparing the 12 cases in the F+ SPMS group with the 17 cases in the F− SPMS group, we found that the two groups differed with regard to the age at clinical onset, which was significantly earlier for F+ than F− SPMS cases; [median value = 23.5 (4.7) versus 34 (5.7) years; P = 0.0019] (Fig. 2A). Interestingly, the proportion of F+ SPMS cases progressively decreased from 83% (5/6) to 46% (6/13) to 10% (1/10) in the SPMS population studied here developing disease in the second, third and fourth decade of life, respectively (chi-square test for trend, P = 0.0035), supporting a link between early disease onset and the formation of ectopic B-cell follicles in the late relapsing or progressive phase of the disease. In the SPMS patient population examined here the female : male ratio was 1.42. However, a greater female preponderance was seen in the F+ SPMS group when compared to the F− SPMS group (female : male ratio = 3.0 versus 0.88; Fisher exact probability test, P = 0.2510). Although this difference did not reach statistical significance due to the small sample size, it shows a trend towards a higher probability for women to develop ectopic B-cell follicles. Age at death was also significantly lower in the F+ than in the F− SPMS group [median value = 42 (5.6) versus 55 (9.0) years; P = 0.0003] (Fig. 2B). The median age at death of males was 40 (1.5) years in the F+ SPMS group and 53 (7.0) years in the F− SPMS group and that of females was 44 (5.5) years in the F+ SPMS group and 62 (14.8) years in the F− SPMS group, indicating that in both genders the F+ SPMS cases died earlier than the F− SPMS cases. It should be noted that the main causes of death in both the F− SPMS and F+ SPMS groups were directly related to MS. Only 3 out of 12 cases in the F+ SPMS group and 5 out of 17 cases in the F− SPMS group received treatment with immunosuppressive agents or interferon-β for a short period, indicating that the therapeutic regimen is very unlikely to account for the different clinical features in the F− and F+ SPMS subgroups (Table 2).

Because an assessment of Kurtzke Expanded Disability Status scale (EDSS) scores was not consistently present in the clinical notes of the multiple sclerosis cases examined, we used the age at which the patient became a wheelchair user (approximately equivalent to EDSS 7) as an indication of chronic motor dysfunction. We observed that the F+ SPMS cases (12 out of 12) became wheelchair bound at a significantly earlier age than the F− SPMS cases (15 out of 17) [median age = 33 (0.49) versus 47 (8.2) years, respectively; P = 0.0011] (Fig. 3A), which is consistent with the earlier age at disease onset.

Although the F− and F+ SPMS groups did not differ significantly with respect to number of relapses during the first 3 years after multiple sclerosis onset (mean ± SEM, 2.3 ± 0.5 versus 3.1 ± 0.6), a negative correlation between number of relapses in the first 3 years of the disease and the age at death was found in the F− SPMS group (r = –0.54; P = 0.0267; not significant after Bonferroni's correction), but not in the F+ SPMS group (r = −0.11; P > 0.05) (Fig. 3B). These findings suggest that disease progression in the latter group might be influenced by local inflammatory events that are not associated with high relapse frequency in the early phase of disease, a known predictor of more severe disease.

Due to the relatively small number of cases examined, no major conclusions could be made concerning the clinical features of PPMS cases as compared to those of F− and F+ SPMS cases. However, it is interesting to note that PPMS cases were more similar to F− SPMS than to F+ SPMS cases with respect to age at disease onset and at death (Fig. 2A and B).

More severe grey matter pathology in the SPMS cases with ectopic B-cell follicles

We next examined whether the F− and F+ SPMS groups differed with respect to the neuropathological pattern. Because of the proximity of meningeal follicles to the cerebral cortical grey matter, we paid particular attention to the characterization of GML and NAGM.

GM and WM demyelination

Using MOG immunostaining to evaluate the number and extent of GML, we found that both the number of GML and the percentage of the GM that had been demyelinated were significantly higher in the F+ SPMS group (3.4- and 5.7-fold, respectively) than in the F− SPMS group (B). When compared to PPMS cases, the F+ SPMS and F− SPMS cases had a significantly higher and lower cortical involvement, respectively (Fig. 4A and B). Conversely, neither the number of WML nor the area of demyelinated WM differed significantly between the F− SPMS and F+ SPMS groups. In the PPMS group, the area of demyelinated WM was similar to that in the F+ and F− SPMS groups, but the number of WML was significantly lower compared to the F− SPMS group (Fig. 4A and B).

Localization of GML

Following a descriptive system used in previous studies (Peterson et al., 2001; Bö et al., 2003) GML were classified as type I (leucocortical), type II (intracortical) and type III (subpial) lesions. In the F+ SPMS group, type-III lesions predominated (accounting for 70% of total GML), whereas in the F− SPMS group GML had no preferential distribution (Fig. 4C). The F+ SPMS cases had 5.3-fold more type III lesions, but only 2.1- and 1.4-fold more type I and type II lesions, respectively, than the F− SPMS cases. Remarkably, all ectopic follicles were found adjacent to subpial type-III lesions, suggesting a causal relationship between follicle formation and cortical damage (Fig. 5A and B). In most of the F+ SPMS cases, subpial demyelination extended over large cortical areas (Fig. 5C), whereas in the F− SPMS cases it was more limited and mainly localized in the depth of the cerebral sulci (Fig. 5D). In the PPMS group, type I and type III lesions tended to predominate over type II lesions (Fig. 4C). However, the number of type III lesions was significantly lower in the PPMS cases than in the F+ SPMS cases [median (range), 1.0 (0–3) versus 5.5 (3–12) lesions/case, respectively; P = 0.0006], indicating that the prominent inflammatory process localized in the meninges of F+ SPMS cases has a key role in the development of subpial demyelination.

Lesion inflammatory activity

In agreement with previous studies (Peterson et al., 2001; Bö et al., 2003), no perivascular immune infiltrates were observed in purely cortical GML (type-II and type-III lesions) of all the multiple sclerosis groups examined, and MHC class II immunostaining, a measure of lesion activity, was restricted to activated microglia. Conversely, leucocortical type-I lesions consistently displayed inflammatory infiltrates, although at lower levels than WML (data not shown). In the F+ SPMS group, the number of active, chronic active and chronic inactive GML was 12.7-, 3.3- and 3.4-fold higher than in the F− SPMS group, respectively (Fig. 6A). The F+ SPMS group also had a significantly higher number of chronic active GML than the PPMS group, whereas the F− SPMS group had less active GML than the PPMS group (Fig. 6A). The number of active, chronic active and chronic inactive WML did not differ significantly among the different multiple sclerosis groups (Fig. 6B).

In active and chronic active GML, MHC class II+ microglia displayed for the most part a radial ramified morphology with thick and shortened processes (Fig. 7). Consistent with the larger cortical demyelination observed in F+ SPMS cases, microglial activation was more prominent in F+ SPMS cases (Fig. 7A and C) than in F− SPMS (Fig. 7B and D) and PPMS cases (not shown). MHC class II+ amoeboid macrophage-like cells were rarely seen in active type-III lesions in the F+ SPMS group and more frequently in type-I lesions (data not shown). Remarkably, in the F+ SPMS group diffuse microglial activation was also detected in the NAGM, not necessarily adjacent to the demyelinated areas (Fig. 7E). Conversely, microglial activation was rarely observed in the NAGM of F− SPMS (Fig. 7F) and PPMS cases (data not shown). In the WM, microglial activation was lower in PPMS cases compared to SPMS cases, whereas no differences were evident between the F− and F+ SPMS groups (data not shown).

Neuronal damage in type-III lesions and NAGM of F+ and F− SPMS cases

Because all ectopic follicles in F+ SPMS cases were found adjacent to extensive subpial type-III demyelinated lesions, we investigated whether neuronal damage in type-III lesions and subpial NAGM differed between the F− and F+ SPMS cases. Immunostaining for SMI32, a marker of dephosphorylated neurofilaments, revealed the presence of diffuse axonal injury in most type-III cortical lesions examined and in the peri-plaque regions in both the F+ and F− SPMS groups, also in the early stages of demyelination (data not shown). To more precisely quantitate the extent of neuronal loss, double immunofluorescence with anti-MBP polyclonal antibody and RT97 monoclonal antibody, which labels both phosphorylated and de-phosphorylated neurofilament protein-containing neurites (predominantly axons), was performed in 12 F− SPMS, 12 F+ SPMS and 3 control cases (Fig. 8A). We observed that the density of RT97-positive neurites in NAGM and chronic active type-III lesions of F+ and F− SPMS cases was considerably lower than in control GM (Fig. 8A–D). Within both the F− SPMS and F+ SPMS groups, the numerical density of RT97+ neurites in type-III lesions was significantly lower than in the NAGM (Fig. 8A). Moreover, the reduction in the number of RT97+ neurites within type-III chronic active lesions was significantly more pronounced in the F+ SPMS cases than in the F− SPMS cases (Fig. 8A). A trend towards a greater reduction of RT97 immunostaining was also observed in the NAGM of F+ SPMS cases as compared to F− SPMS cases (Fig. 8A). Although decreased RT97 immunostaining does not necessarily reflect neuritic loss but may result from downregulation of neurofilament expression, these findings indicate that such a reduction was significantly more prominent in the cerebral cortex of the F+ SPMS group.

Discussion

The presence of ectopic B-cell follicles in the cerebral meninges of a substantial proportion of the multiple sclerosis cases analysed in the present study is an important finding that represents a step forward, both in our understanding of the underlying pathogenetic mechanisms and the design of targeted immunomodulatory therapies. By analysing post-mortem brain specimens from a larger sample of multiple sclerosis cases, we have confirmed and extended our previous finding (Serafini et al., 2004) to show that formation of ectopic follicles in the cerebral meninges accompanies disease progression in a proportion of SPMS cases in which the initial RR phase is followed by a progressive phase, but not in PPMS cases in which the progressive phase is present from the onset of the disease. Most importantly, by grouping the SPMS cases according to the presence or absence of meningeal follicles we show that, compared to the F− SPMS group, the F+ SPMS group is characterized by an earlier age at multiple sclerosis onset, irreversible disability and death, and by a more severe cortical pathology. Because the SPMS cases with age at death ≤50 years are over-represented in the multiple sclerosis population examined here (51%) when compared to the general multiple sclerosis population (estimated to be 13% from the UK Multiple Sclerosis Tissue Bank data collection), the real frequency of follicles in the multiple sclerosis population with an initial relapsing–remitting onset may be lower than that observed in this study (41%). However, it is clear that only rarely did we observe the presence of follicles in multiple sclerosis cases with an age of death over 50 years.

The younger age at onset in the F+ SPMS group compared to the F− SPMS group, and the finding that the proportion of F+ SPMS cases progressively decreases as the age at clinical onset increases, support the idea that a more aggressive inflammatory process favours the establishment of a permissive environment for ectopic follicle formation. The present data are in line with the results obtained in other organ-specific autoimmune diseases. In myasthenia gravis, ectopic thymic follicles form mainly in patients with early-onset disease (Roxanis et al., 2002), while in rheumatoid arthritis ectopic follicles are found in the synovial tissues with the highest degree of inflammation (Magalhães et al., 2002), indicating that formation of ectopic lymphoid tissue requires strong immune activation. Because relapsing–remitting multiple sclerosis cases were not included in our analysis and only one relapsing–remitting multiple sclerosis case lacking follicles was examined in a previous study (Serafini et al., 2004), it is not known whether ectopic follicles form before or after the transition to the SPMS phase. In a chronic relapsing model of murine EAE, meningeal follicles were detected during the relapsing phase and increased in number and size during disease progression (Magliozzi et al., 2004; Columba-Cabezas et al., 2006), indicating that repeated CNS inflammatory events are required to induce formation of ectopic follicles.

Remarkably, the large difference in the age at which the patients became wheelchair dependent and in the age at death between the F+ and F− SPMS groups indicates that formation of ectopic follicles is associated with a more severe disease course. Most importantly, females in the F+ SPMS group died nearly 20 years earlier than those in the F− SPMS group. Due to their potential relevance for multiple sclerosis prognosis and therapy, these findings need to be confirmed in a larger data set. Another finding that indirectly supports an association between ectopic follicle formation and rapid disease progression is the negative correlation between number of relapses during the first 3 years of disease and age at death in the F− SPMS group, but not in the F+ SPMS group. We hypothesize that while a high frequency of intermittent acute relapses contributes to the accumulation of disability in the F− SPMS group, the same progressive phase could be reached in the F+ SPMS group as a consequence of the persistent inflammatory milieu associated with meningeal follicles.

Failure to detect ectopic follicles in the cerebral meninges of a subgroup of SPMS cases and of PPMS cases could have several explanations. The first and more obvious one is that ectopic follicles have been missed due to the small number of brain tissue blocks analysed. However, we think that this is unlikely because ectopic follicle formation was generally accompanied by prominent meningeal inflammation and B-cell/plasma-cell infiltration in the WM, two features that were not observed in the F− SPMS and PPMS cases analysed in this study. A second explanation is that ectopic follicles might have developed transiently during a more active phase of disease and had disappeared by the time of death. This scenario would be consistent with the disappearance of the antigenic stimulus triggering the formation of meningeal follicles, possibly an infectious agent, whose persistence would instead be essential for the maintenance of the germinal centre reaction in the F+ SPMS cases. A third explanation is that in some cases of multiple sclerosis ectopic follicles might develop at sites different from the cerebral meninges. To date, we have failed to detect ectopic follicles at the inferior surface of the brain, around the lateral ventricles (caudate nucleus and subependymal WM of the cerebral hemispheres) and in the cerebellum, brainstem and spinal cord. However, we do not rule out that ectopic follicles in multiple sclerosis might localize in still unidentified brain regions close to the third and fourth ventricles, as shown in the EAE-affected CNS (Magliozzi et al., 2004; Columba-Cabezas et al., 2006). If we postulate that ectopic follicles are where the intrathecal humoral immune response, which is a hallmark of multiple sclerosis, is sustained, their formation in the cerebral meninges could be viewed as the extreme manifestation of a common pathological mechanism that spreads to the cortical surface only in a subset of patients in which a more severe inflammatory process has been induced.

Our findings do not allow us to determine whether follicle formation is causative or is the consequence of a more severe disease process. However, the observation that, when compared to the F− SPMS cases, the F+ SPMS cases showed a more extensive subpial demyelination, an increased number of active cortical lesions, and a more pronounced microglia activation and neurite loss within the areas of subpial demyelination and in the surrounding NAGM, indicates that ectopic follicles, or the inflammatory milieu favouring their formation, are involved in the exacerbation of cortical damage. In contrast to WML and leucocortical type-I lesions, purely cortical lesions (type-II and type-III) are devoid of inflammatory cell infiltrates and foamy macrophages, and show sparse deposition of Ig and complement activation products (Bö et al., 2003; Brink et al., 2005), indicating that the mechanisms mediating WM and GM damage might differ quantitatively and/or qualitatively. The proximity of ectopic follicles to large subpial demyelinated lesions, a consistent finding in all the F+ SPMS cases examined in this study, strongly supports the idea that these abnormal structures have a direct role in cortical injury, most likely by releasing soluble factors that diffuse into the subarachnoid space and through the pial membrane (Rennels et al., 1985). These may include pathogenic antibodies, pro-inflammatory cytokines and/or proteolytic enzymes that directly cause, or amplify, tissue damage. In contrast, no major differences in the number, extension and activity of WML were detected between F− and F+ SPMS cases, which indicates that the inflammatory process localized in the cerebral meninges mainly affects the adjacent cortical GM. These findings are in agreement with recent studies showing that there is no correlation between the extent of GM demyelination and focal or diffuse WM demyelination in multiple sclerosis (Kutzelnigg et al., 2005; Bö et al., 2007). Our observation that ectopic B-cell follicles localize in the cerebral sulci and do not develop at the external brain surfaces suggests that these anatomical sites represent ‘hot spots’ for accumulation of, or accessibility to, the putative triggering antigens and/or for B-cell migration and activation.

Previous neuropathological studies have shown the presence of demyelination, axonal damage and neuronal loss, in the cortical and deep GM of multiple sclerosis patients (Brownell and Hughes, 1962; Lumdsen et al., 1970; Kidd et al., 1999; Peterson et al., 2001; Bö et al, 2003). GML may represent a substantial proportion of the total lesions in the multiple sclerosis brain and are more prominent in SPMS and PPMS than in relapsing–remitting multiple sclerosis (Kutzelnigg et al., 2005). Our finding that the numerical density of neurofilament protein⁺ structures in subpial cortical lesions was significantly lower in F+ than in F− SPMS cases, despite an earlier age at death, indicates that the presence of ectopic B-cell follicles accelerates neuronal degeneration and/or dysfunction. Axon loss in WML has been linked with inflammatory episodes occurring during relapses, in addition to a longer term slower loss that is thought to predominate during the progressive phase (Trapp et al., 1998; Kornek et al., 2000). Because neuronal loss and brain atrophy are the most significant magnetic resonance imaging variables in determining the final disability in multiple sclerosis patients (Filippi and Rocca, 2005; Tedeschi et al., 2005), the present findings suggest that the extensive neurite loss associated with meningeal inflammation and ectopic follicles could be a major factor determining the earlier progression to irreversible disability and the more limited survival of F+ SPMS cases as compared to F− SPMS cases.

The finding that the F+ SPMS cases also showed a higher number of cortical demyelinated lesions with a predominance of subpial lesions and more chronic active GML compared to the PPMS cases examined in this study further supports an important role for meningeal inflammation accompanied by formation of ectopic follicles in exacerbating cortical damage. Although extensive subpial cortical demyelination was previously described in a limited number of PPMS and SPMS cases (Bö et al., 2003), to date no studies have directly compared the distribution of cortical lesions in different multiple sclerosis groups. Kutzelnigg et al. (2005) reported similar cortical demyelination in the forebrain of PPMS and SPMS cases, but did not determine whether the extent of subpial demyelination was also comparable. Although these authors showed that median values for lesioned cortical area were similar in the two groups, the highest percentage of demyelinated cortical area was found in SPMS (68.63 versus 38.68% in PPMS) (Table 1 in Kutzelnigg et al., 2005). This indicates that at least a subset of SPMS cases develops more extensive cortical damage than PPMS cases, which is in keeping with our observations in the F+ SPMS subgroup.

Formation of ectopic lymphoid tissue has been observed in several chronic inflammatory diseases with an autoimmune or infectious aetiology and is thought to be induced by the persistent inflammatory milieu of the target tissue (Aloisi and Pujol-Borrell, 2006). In multiple sclerosis, an immune response directed against CNS antigens (Owens et al., 2006) or infectious agents (Gilden, 2005) is thought to arise in peripheral lymphoid organs leading to the intracerebral migration of memory and effector lymphocytes. It can be envisaged that if antigen-specific, CNS-infiltrating B cells find a favourable environment in the inflamed meninges that allows them to initiate a germinal centre reaction and to undergo proliferation, somatic hypermutation, selection of high-affinity clones and differentiation into antibody-producing plasma cells, all downstream events mediated by the humoral immune response would be amplified locally, resulting in exacerbation of the destructive inflammatory process (Uccelli et al., 2005). As discussed earlier, several issues remain to be solved, in particular whether ectopic follicles develop in the multiple sclerosis brain already during early disease stages and localize at sites other than the cerebral meninges, and whether they are stable or transient structures whose formation is linked to environmental triggers.

In conclusion, this study is the first one to show an association between ectopic lymphoid tissue formation, clinical course and extent of tissue destruction in the target organ during a chronic inflammatory CNS disease. The identification of the antigenic stimuli driving ectopic follicle formation in multiple sclerosis and of the factors mediating the extensive cortical damage associated with meningeal follicles must be considered as a primary goal of future multiple sclerosis research and could shed light into the still elusive immune mechanisms that mediate CNS tissue injury. The present findings also suggest that prevention or eradication of lymphoid microenvironments nested within the CNS should be identified as an important goal for therapeutic intervention in multiple sclerosis patients.

Acknowledgements

All tissue samples were supplied by the UK Multiple Sclerosis Tissue Bank (www.ukmstissuebank.imperial.ac.uk), funded by the Multiple Sclerosis Society of Great Britain and Northern Ireland (registered charity 207495). The authors would like to thank members of the UK multiple sclerosis Tissue Bank Team (S. Gentleman, M. Graeber, F. Roncaroli, S. Fordham, I. Ghebrenegus and N. Patel) for assistance in the collection and characterization of the material used in this study and Ms Estella Sansonetti for graphical work. This work was supported by the Multiple Sclerosis Society of Great Britain and Northern Ireland (grant No 747/02), a fellowship from the Italian Multiple Sclerosis Foundation to R.M. Programme of Collaboration between Istituto Superiore di Sanità and National Institutes of Health, and 6th Framework Program of the European Union NeuroproMiSe LSHM-CT-2005-01863 to F.A.

References

Abbreviations:

Abbreviations:
 
• FDC

  follicular dendritic cell

 
• GML

  grey matter lesion

 
• LFB

  Luxol Fast Blue

 
• MHC

  major histocompatibility complex

 
• MBP

  myelin basic protein

 
• MOG

  myelin oligodendrocyte glycoprotein

 
• NAGM

  normal appearing grey matter

 
• Ig

  immunoglobulins

 
• PPMS

  primary progressive multiple sclerosis

 
• SPMS

  secondary progressive multiple sclerosis

 
• WML

  white matter lesion

Author notes