Primer acto

To our knowledge, the current study is the first to quantify human meningeal LTα, CXCL13 and CCL21 gene expression. LTα gene expression is upregulated in SPMS meninges relative to control cases, suggesting that LTα may be an important contributor to meningeal inflammation in SPMS, and the underlying GM pathology with which meningeal inflammation is associated. Dysregulated, co-stimulation independent LTα secretion following CD3 stimulation of CD8+ T cells from SPMS compared to RRMS patients and control cases has been demonstrated, and it is suggested that alterations in LTα secretion may be a factor in conversion to progressive disease (Buckle et al., 2003). Unfortunately the sample size of 14 SPMS patients studied by these authors was not large enough to determine if patients could be divided into higher and lower LTα-secreting groups so it is unknown whether differences in LTα-secretion might exist between F+ and F-SPMS, but the increased secretion observed in SPMS compared to control cases supports our finding of increased SPMS meningeal LTα expression in the current study.

Interestingly, we found no significant difference between LTα mRNA expression in F+SPMS compared to F-SPMS meninges. This was surprising, given the known role of LTα in TLO neogenesis in other studies and diseases (Kratz et al., 1996; Sacca et al., 1998; Calmon- Hamaty et al., 2011), and suggests perhaps that other chemokines may be required in concert with LTα to induce TLO formation in F+SPMS cases. However, LTα IHC was only observed in F+SPMS cases and appeared to be expressed by endothelial cells lining parenchymal vessels, in addition to cell-specific and more diffuse staining within areas of meningeal inflammation. This pattern of LTα IHC contrasts with that of a previous study, where LTα and CXCL13 were both observed to be expressed on the outer layers of post-capillary venules in a similar pattern to extracellular matrix expression, but not in endothelial cells (Corcione et al., 2004). It is possible that the increased tissue preservation of paraffin-embedded tissue compared to the snap frozen tissue used in the current study allowed more accurate observation of IHC localisation (Corcione et al., 2004). These authors detected LTα and CXCL13 in ~50% of vessel walls in chronic active lesions, and expression was frequently but not exclusively associated with demyelination or inflammatory infiltrates. F+SPMS is characterised by increased GM pathology and meningeal inflammation compared to F-SPMS, so an increase in the number of meningeal LTα-secreting CD8+ T cells in F+SPMS might be expected to lead to an increase in CSF concentrations of LTα in F+ compared to F-SPMS. Chapter 3 – Expression of lymphoid cytokines and chemokines in SPMS 103

This postulated increase in LTα CSF concentration in the F+SPMS cohort may not be reflected by the meningeal LTα gene expression observed in the current study, as meningeal tissue was dissected from areas overlying gyri and not exclusively from deep sulci in which substantial meningeal inflammation and TLOs are most often found in F+SPMS, due to the difficulty of accurate dissection of sufficient meningeal tissue from deep sulci.

While LTα concentration in CSF samples analysed in the current study was below the limit of detection by conventional ELISAs, it was detected at sub-picogram/ml concentrations in the CSF from two particularly inflammatory F+SPMS cases using an ECL assay with greater sensitivity. This suggests that global CNS levels of LTα may be higher in F+SPMS compared to F-SPMS, but also that levels are either generally low or LTα degradation occurred post- mortem. No other studies have been reported that investigate LTα in post-mortem CSF, and only one study to our knowledge has investigated secreted LTα concentration in CSF from 26 MS patients (Corcione et al., 2004). The mean concentration reported by these authors was approximately 10pg/ml but no details of patients’ MS stage or disease course was given, so it is unknown whether this level is representative of SPMS cases.

Quantification of LTα mRNA expression by CSF cells may also be used to give an indication of cytokine production in the CSF. Mononuclear cells expressing LTα mRNA were significantly enriched in the CSF compared to peripheral blood in MS patients, and compared to CSF in patients with other inflammatory neurological diseases (OIND) (Matusevicius et al., 1996). In addition, MBP-reactive LTα mRNA expressing cells were increased in MS patient CSF compared to OIND patients, particularly during relapse, suggestive of a pathogenic role (Matusevicius et al., 1996). Although mRNA expression data must be interpreted with caution, as post-transcriptional regulation of LTα mRNA translation or regulation of LTα secretion may cause discrepancies between mRNA expression and secretion, a positive correlation has been shown between TNF mRNA expression and secretion, suggesting that indications of increased LTα production in MS CSF provided by mRNA studies are likely to be correct (Rieckmann et al., 1995). Given these results, CSF concentrations of LTα in MS patients is expected to be increased compared to controls, and the paucity of data on LTα CSF concentration may be due to lack of investigation, or a relatively short half-life of LTα.

To estimate LTα degradation in post-mortem CSF and determine if LTα concentrations in CSF from living SPMS patients may be considerably higher than those measured in the current study, we compared our data to that obtained for TNF, which is similar in structure to LTα and Chapter 3 – Expression of lymphoid cytokines and chemokines in SPMS 104

estimated to have a similar instability index using the ExPASy online tool. The serum half-life of TNF in mice is less than 20 minutes (Ferraiolo et al., 1988), and while TNF concentrations of up to 70pg/ml have been reported in CSF from progressive MS patients, the highest concentrations of TNF in post-mortem CSF were approximately 4pg/ml in F+SPMS and only 0.25pg/ml in F-SPMS (Gardner et al., 2013; Rossi et al., 2013). LTα may be expressed at lower concentrations than TNF since it was only detected at sub-picogram concentrations in F+SPMS cases in the current study, but together these data also suggest that the variable post-mortem delays of up to 24 hours are likely to have resulted in significant LTα protein degradation and contributed to the very low concentrations.

Global increases of LTα throughout the CSF may contribute to CNS inflammation in SPMS through effects on BBB activation, increased expression of adhesion molecules and recruitment of immune cells to the CNS, since these effects have been observed in vitro (Cuff et al., 1998) and in vivo (Cuff et al., 1999). In addition, LTα may contribute to the general inflammatory milieu that is postulated to contribute to subpial demyelination since LTα has cytolytic functions and is toxic to oligodendrocytes (Selmaj et al., 1991c). However, significant increases in global CSF concentration, if present, are unlikely to be necessary for TLO formation. Since TLOs are often associated with deep sulci, it is postulated that meningeal inflammation within the relatively sheltered confines of a sulcus may lead to a local increase in CSF concentration of cytokines and cytotoxic factors, and result in damage of the adjacent GM, which may not be reflected by overall CSF concentrations (Serafini et al., 2007; Magliozzi et al., 2010; Howell et al., 2011).

LTα and TNF promote FDC differentiation and maintain DC homeostasis (Matsumoto et al., 1997), and these cell types are crucial for TLO maintenance by secreting homeostatic chemokines (including CXCL13 and CCL21), organising of TLO structure and cell positioning (Geurtsvankessel et al., 2009). FDCs within GCs are also required for B cell maturation, due to their role in antigen presentation, promotion of B cell proliferation and prevention of B cell apoptosis (Van Nierop and De Groot, 2002; Aloisi and Pujol-Borrell, 2006). The source of the FDCs observed in meningeal TLOs is unknown, but fibroblast-like synoviocytes have been suggested to undergo differentiation to become fully functional FDCs during rheumatoid arthritis (RA) (Lindhout et al., 1999; Kain and Owens, 2013). The chronic presence of LTα and/or TNF is suggested to drive local fibroblasts into converting to an FDC phenotype in RA and may occur in chronically inflamed meninges in F+SPMS (Van Nierop and De Groot, 2002; Gardner et al., 2013). B cells expressing LTα are capable of inducing FDC clusters and GC Chapter 3 – Expression of lymphoid cytokines and chemokines in SPMS 105

formation in the absence of T cells, and are suggested to induce TLO formation during inflammation (Fu et al., 1998). In addition, macrophages have been shown to express high levels of LTα and induce formation of TLOs, and stromal cell expression of lymphoid chemokines via TNFR1 signalling in vivo (Guedj et al., 2013). We therefore suggest that substantial meningeal inflammation containing B, T cells and macrophages expressing LTα may reach a critical mass in sulci in F+SPMS, and result in FDC differentiation, clustering and expression of lymphoid chemokines including CXCL13 and CCL21, which may maintain the inflammatory response and result in TLO formation. In areas of diffuse inflammation, and in F- SPMS, we suggest that the local concentrations of LTα, TNF and other pro-inflammatory cytokines may not reach a critical threshold required to drive TLO formation.

Although there was no significant difference in meningeal expression of LTα between F-SPMS and F+SPMS cases in the current study, it is possible that a difference in the expression of other pro-inflammatory cytokines or receptors between these groups potentiates the effects of LTα and TNF in F+SPMS. IFNγ and TNF meningeal gene expression is significantly upregulated in F+SPMS compared to controls, while no significant difference in IFNγ gene expression was observed in F-SPMS (Gardner et al., 2013). TNF was also increased in the CSF in F+SPMS compared to F-SPMS and controls. The expression of TNFR1, through which LTα mediates inflammation (Sacca et al., 1998), is influenced by both TNF and IFNγ in vitro. Microglia, oligodendrocytes and astrocytes constitutively express TNFR1 mRNA, and expression increases following addition of IFNγ (Dopp et al., 1997). Interestingly, addition of TNF only increased TNFR1 expression in oligodendrocytes, suggesting that increased levels of TNF and IFNγ in F+SPMS compared to F-SPMS and controls, may influence receptor expression on glial cells and therefore determine the effect of meningeal inflammation, resulting in pathological differences between F-SPMS and F+SPMS.