Syndecan-1, a Cell Surface Proteoglycan, Negatively Regulates Initial Leukocyte Recruitment to the Brain across the Choroid Plexus in Murine Experimental Autoimmune Encephalomyelitis

Cell surface heparan sulfate proteoglycans are gaining attention because of their large potential to bind a wide variety of bioactive molecules and thereby influence developmental processes but also pathological situations, including inflammation (1). As highly glycosylated, sulfated molecules and, hence, negatively charged molecules, they have a large capacity to bind to other charged molecules such as cytokines and chemokines (2) and have been implicated in the presentation of cytokines to cells (3–5) and in the activation of integrin receptors (6–8). Syndecans are the best studied class of cell surface heparan sulfate proteoglycans (9); they exist in four different forms, syndecan 1–4, all of which consist of a core protein containing an ectodomain (extracellular domain), which carries the glycosylation sites, a transmembrane domain and a cytoplasmic domain. The transmembrane and cytoplasmic domains are highly conserved among the four syndecans, whereas the ectodomain shows the greatest variability both in protein sequence and in glycosaminoglycan chains (9, 10). Furthermore, the ectodomain of all syndecans have been reported to be shed from the cell surface by matrix metalloproteinases (9, 11, 12), which can then have paracrine or autocrine effects or act as competitive inhibitors, introducing a further complicating mode by which they can alter cell functions.

Broadly speaking, the different syndecans have distinct distributions in vivo, with syndecan-1 occurring on mainly epithelial and plasma cells, syndecan-2 on endothelial cells, syndecan-3 on neural crest–derived cells, and syndecan-4 showing low-level ubiquitous expression (13). Although syndecan-1 knockout (Sdc-1^−/−) mice show exacerbated disease symptoms in several inflammatory models (14), precisely why this is the case remains unclear. Most studies report enhanced immune cell infiltration in Sdc-1^−/− mice, which has been correlated with enhanced extravasation from postcapillary venules because of effects ranging from modulation of integrin activity and interaction with cell adhesion molecules (14, 15) to modulation of cytokine and chemokine gradients (16). Yet, the expression of syndecan-1 on endothelium and immune cells in vivo is difficult to detect, with the notable exception of B lymphocytes, which may be partially because of its dynamic regulation (9, 17–19).

In this study, we investigate how syndecan-1 impacts on neuroinflammation using the murine model of the 35–55 peptide of myelin oligodendrocyte glycoprotein (MOG_35–55)–induced experimental autoimmune encephalomyelitis (EAE), a T lymphocyte–mediated inflammation where the steps in disease development have been well defined and can be readily followed: early recruitment of the disease-inducing encephalitogenic Th17 cells to the CNS occurs via the choroid plexus and follow a CCL20-dependent gradient, whereas subsequent recruitment of both encephalitogenic T cells and bystander cells, such as macrophages, which contribute to disease severity, occurs via postcapillary venules within the brain parenchyma (20). Following peak recruitment, disease symptoms normally wane, resulting in a recovery phase characterized by increasing numbers of CD25⁺Foxp3⁺ regulatory T cells (Tregs) and reduced numbers of effector Th17 and Th1 cells. Our data demonstrate that the main site of syndecan-1 expression in the brain is the choroid plexus epithelium where its expression correlates with that of CCL20 and where it negatively correlates with IL-6 expression, hence, impacting on the initial recruitment of the encephalitogenic T cells to the brain. We demonstrate that the choroid plexus epithelium is an important source of IL-6 at early EAE and that its expression is increased in the absence of syndecan-1. In the absence of syndecan-1, early leukocyte recruitment via the choroid plexus is enhanced, and CNS levels of IL-6 are elevated, which collectively results in higher numbers of the disease-inducing Th17 cells and potentially their prolonged survival, thereby contributing to enhanced disease severity. In addition, Sdc-1^−/− mice have intrinsically elevated plasma cell numbers and higher MOG-specific Ab levels during EAE, which we propose contributes to impaired recovery. Our data work identifies the choroid plexus epithelium as a novel source of IL-6 in EAE and demonstrates that its expression is negatively correlated with syndecan-1 expression at this site.

Syndecan-1 knockout (Sdc-1^−/−) mice have been previously described (21). Blimp-1-GFP⁺ mice (22) were bred with Sdc-1^−/− mice to generate Sdc-1^−/−/Blimp-1-GFP⁺ mice and heterozygous littermate controls. C57BL/6-Ly5.1 (CD45.1) mice were used in passive transfer and bone marrow chimera experiments. All experiments were conducted according to German Animal Welfare guidelines.

The Abs used in immunofluorescence staining and flow cytometry were as follows: pan-laminin (455) (23), laminin γ1 (24), CD45 (30G.12), CD45.2 (104; eBioscience), CD45.1 (A20; BD Pharmingen), CD11b/MAC-1 (M1/70; BD Pharmingen), CD11c (N418; eBioscience), CD4 (H129.19; BD Pharmingen), CD8 (53-6.7; eBioscience), B220 (RA3-6B2; BD Pharmingen), syndecan-1 (281-2; BD Pharmingen), CD19 (1D3; BD Pharmingen), CD25 (7D4; BD Pharmingen), Foxp3 (FJK-16S; eBioscience), IL-17 (TC11-18H10.1; BD Pharmingen), biotin-labeled IL-6 (MP5-32C11; eBioscience), IFN-γ (XMG1.2; BD Pharmingen), CD16/CD32 (2.4G2; BD Pharmingen), CCL20 (MIP-3α) (AB9829; Abcam), CD68 (FA-11; BioLegend), MHC class II (M5/114.15.2; eBioscience), metallophilic macrophages 1 (BMA Biomedicals) and S100A9 (25), and mouse endothelial cell Ag 32 (MECA32) (26).

EAE was induced using MOG_35–55 as described previously (27, 28). We define early EAE as day 10 after immunization, peak EAE as day 17 after immunization, and the recovery phase as >day 20 after immunization.

Ten days after MOG_35–55 immunization, draining lymph node (LN) cells were isolated and seeded at 1 × 10⁷ cells ml⁻¹ in IMDM (Life Technologies) plus 10% FCS. The cells were stimulated with 20 μg ml⁻¹ MOG_35–55 peptide for 2 d and then in the presence of 10 ng ml⁻¹ IL-2, 10 ng ml⁻¹ IL-23, 5 ng ml⁻¹ IL-1β, 10 ng/ml IL-6, and 10 mg/ml anti–IFN-γ and anti–IL-4 for 3 d. A total of 2 × 10⁷ cells/mouse were transferred i.v. Mice were then injected i.p. with 200 ng pertussis toxin on days 0 and 2 after transfer. The polymorphic lineage determinants (CD45.1 or CD45.2) were used for tracking donor versus host immune cells.

Recipient mice were lethally irradiated with a single dose of 11 Gy and reconstituted by i.v. injection of 10⁷ donor bone marrow cells. CD45.1 versus CD45.2 allelic markers were used to trace donor versus host cells. Animals were analyzed after 6–8 wk by flow cytometry for reconstitution of the hematopoietic system. Only mice with >95% donor cell engraftments were used in active EAE experiments.

Mice were perfused with PBS before spleens, LNs, and brains were harvested, and total cells were isolated by cell straining (70 μm for spleens and LNs, 100 μm for brains). Erythrocytes were lysed in spleen preparations by incubating the cells on ice for 3 min with lysing buffer (BD Pharm Lyse); leukocytes were isolated from blood using a Ficoll gradient (Cedarlane Laboratories); and brain homogenates were separated into neuronal and leukocyte populations by discontinuous density gradient centrifugation using isotonic Percoll (Amersham Biosciences). For intracellular cytokine staining, isolated leukocytes were stimulated with PMA (10 ng/ml)/ionomycin (1 μg/ml) (Sigma-Aldrich) in the presence of brefeldin A (Sigma-Aldrich) at 37°C for 6 h. The intracellular staining kit (eBioscience) was used to permeablize and fix the cells prior to staining for Foxp3 and intracellular cytokines. Flow cytometry analysis was performed using a FACSCalibur (BD Biosciences) with the Abs listed above.

Mice were intracardially perfused with 4% paraformaldehyde in PBS, and samples were dissected and fixed 1.5 h at 4°C in 4% paraformaldehyde. Tissues were immersed in 30% sucrose in PBS and subsequently frozen in Tissue-Tek (Sakura Fenetek). Eight-micrometer sections were blocked in 1% BSA in PBS and incubated with primary Ab overnight at 4°C and with secondary Ab for 2 h at room temperature. Because the heparan sulfate chains of syndecans may mask binding sites of primary Abs, unmasking techniques were performed prior to blocking of tissue samples. These included 0.1 M acetic acid for 5 min at room temperature or 1 U/ml heparitinase III (Sigma-Aldrich) in 50 mM Tris buffer (pH 7.5) for 10 min at room temperature. Secondary Abs included goat anti-rabbit and donkey anti-rat IgG conjugated with Alexa Fluor 488 or Cy3 (Molecular Probes). Sections were examined using a Zeiss AxioImager microscope equipped with epifluorescent optics and documented using a Hamamatsu ORCA ER camera. Images were analyzed using Volocity 6.0.1 software (ImproVision; PerkinElmer).

CD4⁺ T lymphocytes from LNs of MOG_35–55-immunized wild-type (WT) or Sdc-1^−/− mice were cultured at 37°C for 3 d with irradiated splenic dendritic cells (DCs) (30 Gy) (from nonimmunized mice) as APCs plus MOG_35–55, OVA fragment 329–339 (Ova_323–339) (Schafer-N), or with Ab to CD3 and CD28 (BD Pharmingen). T lymphocyte proliferation was determined by [³H]thymidine (Amersham Biosciences) incorporation over 12 h. WT or Sdc-1^−/− T lymphocytes were cocultured with WT or Sdc-1^−/− splenic DC in separate experiments. Similar experiments were performed with irradiated Sdc-1^−/− and WT B lymphocytes (5 Gy) as APCs.

To measure in vivo proliferation of MOG_35–55-specific T cells, CD45.1⁺ T cell blasts were transferred to CD45.2⁺ WT and Sdc-1^−/− mice, and BrdU incorporation into the CD45.1⁺CD4⁺ population was measured. BrdU was injected i.p. on days 21 and 23 after transfer. LN, spleen, blood, and CNS were removed 12 h after the last injection, and CD45.1⁺CD4⁺ T lymphocytes were isolated and analyzed for BrdU incorporation by flow cytometry.

For proliferation assays, LN cells or splenocytes (2.5 × 10⁵ cells per 0.2 ml) from EAE mice were cultured in 96-well plates in RPMI 1640 medium (Life Technologies) plus 10% FCS, 10 mM sodium pyruvate, 1% penicillin/streptomycin, and 0.05 mM 2-ME in the presence of different concentrations (0, 1.1, 3.3, 11, 33, and 100 μg/ml) of MOG_35–55 at 37°C in 5% CO₂ for 72 h. Cell proliferation was determined by [³H]thymidine incorporation over the last 12 h. For the cytokine responses, LN cells or splenocytes (1 × 10⁶ cells per 0.2 ml) from EAE mice were cultured as above in the presence or absence of 20 μg/ml MOG_35–55 at 37°C in 5% CO₂ for 72 h. Culture supernatants were harvested, and cytokines were analyzed using the cytometric bead assay. All proliferation and cytokine response measurements were performed in triplicate.

CNS samples from EAE WT and Sdc-1^−/− mice were frozen in liquid nitrogen and subsequently homogenized on ice in radioimmunoprecipitation assay buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% NaDoC, 0.1% SDS, 50 mM Tris HCl [pH 8], and proteinase inhibitor mixture (Sigma-Aldrich)). After centrifugation (12,000 × g for 10 min), supernatants were collected, and cytokine and chemokine concentrations were determined using the Chemomix kit (Bender MedSystems) and mouse FlowCytomix kits (eBioscience). CSF samples were directly analyzed for chemokine and cytokine levels using the same kits.

Because syndecan-1 (CD138) could not be used as a marker for plasma cells in the Sdc-1^−/− mice, they were crossed to the Blimp-1-GFP⁺ transgenic line to generate mice carrying GFP⁺ plasma cells. Blimp-1 is a transcription factor specific for plasma cells (22). Blood, spleen, LN, bone marrow, and CNS of naive and EAE-induced Sdc-1^−/−/Blimp-1-GFP⁺ mice were analyzed by flow cytometry for syndecan-1⁻GFP⁺.

Parallel analyses of serum titers for MOG_35–55-specific Abs was performed using SBA Clonotyping System-B6/C57J-HRP kit (Southern Biotechnology Associates) as described previously (29, 30). Briefly, 96-well ELISA plates (Nunc) were precoated overnight at 4°C with 10 μg/ml MOG_35–55 peptide. Plates were incubated with serum samples, and bound Ig was detected using HRP-conjugated goat anti-mouse Abs to different Igs. ABTS was used as a color substrate, and OD was measured at 405 nm.

WT and Sdc-1^−/− mice were sacrificed at different stages of EAE and 2–5 μl cerebrospinal fluid (CSF) per mouse was collected from the cisterna magna as described previously (31). Brains were then isolated and the positions of the choroid plexus in the fourth ventricle, and the lateral ventricles were determined under a dissecting microscope and isolated. The choroid plexus appeared as a vascularized “gelly” membrane floating in the liquid. Tissues were snap-frozen in liquid nitrogen, and tissues and CSF were stored at −80°C.

To investigate presence of syndecan-1 ectodomain in CSF, 2-μl CSF samples were dot-blotted onto a nitrocellulose membrane (Whatman). The membrane was blocked in TBS containing 1% FCS, washed in TBS containing 1% FCS and 0.01% Tween 20, and incubated with biotinylated anti–syndecan-1 ectodomain Ab (1:500; BD Pharmingen). Bound Abs were detected with HRP-conjugated secondary Abs and the ECL system (Pierce). A total of 2 μg purified recombinant syndecan-1 protein (R&D Systems) was used as positive control, and secondary Ab alone was used as a negative control.

To investigate whether syndecan-1 binds IL-6 or CCL20, 50 ng purified recombinant syndecan-1 was coated onto Nunc Maxisorb microtitre plates by overnight incubation at 4°C. Plates were wash with PBS, blocked with 1% BSA in PBS and incubated with different concentrations of IL-6 (0–10 μg/ml) or CCL20 (1–2 μg/ml). Bound cytokine or chemokine was detected with biotin-labeled anti–IL-6 or anti-CCL20, respectively, and HRP-conjugated secondary Abs. Tetramethylbenzidine was used as a color substrate; OD was measured at 450 nm.

Total RNA from the whole brains, choroid plexuses, LNs, or spleens were prepared using the RNeasy kit (Qiagen) and cDNAs were generated (Omniscript RT Kit; Qiagen). Quantitative PCR (q-PCR) was performed using Brilliant SYBR Green QPCR MasterMix, and the primers are listed below according to the manufacturer’s instructions: syndecan-1, 5′-GTG GCG GCA CTT CTG TCA TC-3′ (sense) and 5′-GCA CCT GTG GCT CCT TCG TC-3′ (antisense); syndecan-2, 5′-TGT GTC CGC AGA GAC GAG AA-3′ (sense) and 5′-GGA ATC AGT TGG GAT GTT GTC A-3′ (antisense); syndecan-3, 5′-ATA CTG GAG CGG AAG GAG GT-3′ (sense) and 5′-TTT CTG GTA CGT GAC GCT TG-3′ (antisense); syndecan-4, 5′-AAC CAC ATC CCT GAG AAT GC-3′ (sense) and 5′-AGG AAA ACG GCA AAG AGG AT-3′ (antisense); IL-6, 5′-AAC CAC ATC CCT GAG AAT GC-3′ (sense) and 5′-AGG AAA ACG GCA AAG AGG AT-3′ (antisense); CCL20, 5′-TCC AGA GCT ATT GTG GGT TTC A-3′ (sense) and 5′-GAG GAG GTT CAC AGC CCT TTT-3′ (antisense); and GAPDH, 5′-AGG TCG GTG TGA ACG GAT TTG-3′ (sense) and 5′-GGG GTC GTT GAT GGC AAC A-3′ (antisense).

mRNA expression was normalized to endogenous GAPDH expression in the same sample. Quantitative real-time PCR was performed with an Applied Biosystems PRISM 7300 Sequence Detection System by using the default thermal cycling conditions (10 min at 95°C and 40 cycles of 15 s at 95°C plus 1 min at 60°C). Relative quantitation was performed using the comparative cycle threshold method (32). Three to five biological replicates were used for each point investigated.

Quantitative data are expressed as means ± SEM. Significance of cell numbers was analyzed with an unpaired Student t test. A p value < 0.05 was considered significant.

Sdc-1^−/− mice showed enhanced EAE severity and prolonged recovery compared with WT littermates (Fig. 1A), which correlated with elevated numbers of the disease-inducing Th1 and Th17 cells and lower numbers of CD4⁺CD25⁺Foxp3⁺ Tregs at early (day 10) and peak (day 17) stages of the disease but not in the recovery phase (>day 20) (Fig. 1B, 1C). Demyelination, which is associated with disease severity, was analyzed by staining for myelin basic protein (MBP) together with CD45 to define sites of infiltrated leukocytes and pan-laminin or laminin γ1 chain to define cuff borders (23), revealing a similar loss of MBP staining at sites of leukocyte penetration of the CNS parenchyma in WT and Sdc-1^−/− mice (Fig. 1D). However, stereological analyses revealed larger cuff numbers in Sdc-1^−/− brains at peak and, to a lesser extent, recovery phases and hence more extensive demyelination (Fig. 1E), consistent with the more severe disease phenotype.

Flow cytometry of the CNS, spleen, LN, and blood for different immune cell populations at peak and recovery phases of the disease revealed that CD4⁺ T lymphocytes and CD11b⁺ macrophages were elevated in Sdc-1^−/− CNS only at peak disease (Supplemental Fig 1A, 1B). Numbers of T cells, macrophages, and B cells in the circulation (data not shown), LNs, and spleens were not altered in the Sdc-1^−/− mice compared with WT mice (Supplemental Fig. 1A, 1B), and there was no difference between naive WT and Sdc-1^−/− mice in CD4, CD8, B220, CD11b, and CD11c cell populations, as assessed by flow cytometry of the spleens and LNs (data not shown). Because myeloid cells in the CNS have been associated with enhanced disease severity, different myeloid populations were analyzed during the recovery phase, revealing a tendency for higher numbers of CD11b⁺MHCII^high, CD11b⁺CD68⁺, and CD11b⁺S100A9⁺ macrophages, which denotes activated macrophages in the CNS of Sdc-1^−/− mice, although these differences were not statistically significant (Supplemental Fig. 1C). metallophilic macrophages 1⁺CD11b⁺ macrophages that have been shown to negatively regulate Treg numbers in EAE (33) were not different in WT and Sdc-1^−/− mice (Supplemental Fig. 1C).

Passive transfer of Sdc-1^−/− T cells to WT recipients excluded the possibility that a T cell source of syndecan-1 contributed to this phenotype (Supplemental Fig. 2A, 2B). However, bone marrow chimeric animals—Sdc-1^−/− recipients carrying WT bone marrow (Fig. 2A) and WT recipients carrying Sdc-1^−/− bone marrow (Fig. 2B)—identified the involvement of both an immune cell and a CNS-resident cell source of syndecan-1. Adoptive transfer of WT CD45.1 T cells to the CD45.2 Sdc-1^−/− or WT recipients permitted analysis of whether in vivo proliferation of encephalitogenic T cells was promoted in the syndecan-1–negative environment, thereby accounting for the elevated CD4⁺ T cell numbers. These studies revealed the absence of any difference in the proliferation of WT or Sdc-1^−/− T cells in the WT or Sdc-1^−/− CNS (Supplemental Fig. 2C). Similarly, in vitro proliferation assays revealed the absence of differences between WT and Sdc-1^−/− T cell proliferation either in response to anti-CD3 (Supplemental Fig. 2D) or in response to MOG_35–55 or Ova_323–339 (Supplemental Fig. 2E, 2F), in the presence of Sdc-1^−/− or Sdc-1^+/+ APCs, which included DCs and B cells. MOG_35–55-specific proliferative and cytokine responses were measured from draining LNs, and splenocytes of bone marrow chimeric mice revealed an absence of differences in proliferation but elevated levels of IFN-γ, IL-2, IL-6, IL-17, and TNF-α both in mice lacking an immune cell or a resident cell source of syndecan-1 (Supplemental Fig. 2G, 2H), suggesting that syndecan-1 also has peripheral effects.

These results suggest that syndecan-1 contributes to regulation of T cell differentiation in the periphery and that the elevated Th1 and Th17 cells numbers in the Sdc-1^−/− CNS at early and peak disease were due to enhanced T cell infiltration into the CNS and/or their prolonged survival, rather than enhanced proliferation.

To precisely define whether immune cells and/or CNS-resident cells express syndecan-1 during EAE, immunofluorescence microscopy was performed on tissue sections, whereas isolated cells were examined by flow cytometry and q-PCR. Triple immunofluorescence staining for syndecan-1, pan-laminin to define the borders of blood vessels and the leptomeninges in the CNS (28), together with CD45 to define sites of leukocyte infiltration, revealed no detectable staining for syndecan-1 either on the endothelium, astrocytes, or immune cells in perivascular cuffs within the brain (Fig. 3A). As the heparan sulfate chains of syndecans may mask Ab binding sites in vivo, several unmasking techniques were performed, including heparitinase treatment to remove heparan sulfate; however, no staining of perivascular cuffs was detectable; rather, there was strong staining of the choroid plexus. Immunofluorescence staining with pan-laminin, to mark all basement membranes, together with MECA32 to mark the choroid plexus blood vessels (26), or syndecan-1, revealed strong staining of syndecan-1 on the basolateral surface of the epithelium of the choroid plexus and little or no staining of the choroid plexus blood vessels (Fig. 3B). Syndecan-1 Abs stained only B220⁺IgM^high plasma cells in the LN and spleen (data not shown). Flow cytometry was performed on immune cells from naive and EAE mice to investigate potentially low levels of syndecan-1 expression, revealing no detectable signal on T cells, macrophages, and monocytes and only a low-level signal on B220⁺ B cells (Supplemental Fig. 3A), as reported previously (17). Because endothelium has been reported to express syndecan-1 in other tissues, brain endothelial cells and intact choroid plexus were isolated, and syndecan-1 expression was investigated by q-PCR, revealing a >10-fold lower signal in brain endothelial cells than in choroid plexus samples (Supplemental Fig. 3B). Interestingly, several brain endothelial cell lines and passaged primary cells were found to have a significantly higher PCR signal than that measured in choroid plexus (data not shown), indicating that syndecan-1 expression on endothelium may be regulated by the in situ milieu. To ensure that loss of syndecan-1 did not result in upregulation of one of the other three syndecans, q-PCR was performed on whole brain, LN, and spleen extracts of peak EAE mice and on isolated choroid plexuses of WT and Sdc-1^−/− mice in the naive situation and at early and peak EAE, revealing the absence of upregulation of any of the other syndecans (Supplemental Fig. 3C).

Because the choroid plexus was identified as the major CNS resident source of syndecan-1 and because initial encephalitogenic T cell recruitment to the brain has been reported to occur at this site (20), we focused on this site. CCL20 has been shown to be required for Th17 cell recruitment across the choroid plexus (20, 34), and its expression has been reported to be influenced by various cytokines including IL-6 (35, 36), a known major exacerbating factor in EAE (37). We therefore examined syndecan-1, CCL20, and IL-6 expression at the choroid plexus in the naive brain and during the course of EAE.

In naive WT brains, immunofluorescence staining revealed strong expression of CCL20 both within choroid plexus epithelial cells and on their surface, which showed partial colocalization with syndecan-1 staining (Fig. 4A). q-PCR revealed that IL-6 mRNA is also expressed at the choroid plexus (Fig. 5A).

During the course of EAE in WT mice, immunofluorescence staining revealed a marked loss of syndecan-1 and CCL20 from the choroid plexus epithelial cells (Fig. 4B), yet q-PCR for syndecan-1 mRNA revealed upregulation at early and peak disease stages (Fig. 5B), which correlated with an increase in IL-6 mRNA (Fig. 5A) and corresponding elevated levels of IL-6 in CNS extracts (Fig. 5C). Because syndecans can be shed from the cell surface, we examined whether the syndecan-1 ectodomain was detectable in the CSF of naive and EAE mice. Because of the very small volume of CSF collectable per mouse, dot blots were performed, revealing the presence of syndecan-1 ectodomain at both early and peak stages of EAE but not in the CSF of naive mice (Fig. 5D). It was not possible to detect CCL20 in the CSF of naive or EAE mice either by dot blot or by ELISA, which may be due to the reagents available, whereas IL-6 was weakly detectable in the CSF using the cytometric bead assay only (data not shown). Taken together, the data suggest that syndecan-1 is normally shed from the choroid plexus epithelia during EAE into the CSF and that this correlates with loss of CCL20 from the epithelial cell surface and an associated upregulation of IL-6 expression. Using both direct and indirect ELISA and far Western techniques, we could not demonstrate a direct binding between syndecan-1 and CCL20, which again may be due to the reagents available. However, IL-6 showed a clear dose-dependent binding to syndecan-1 (Supplemental Fig. 4A).

Compared with WT mice, naive Sdc-1^−/− mice showed a markedly reduced CCL20 immunofluorescence signal on the choroid plexus epithelium, which was only slightly further reduced during the course of EAE (Fig. 4B). As in WT mice, it was not possible to detect CCL20 in the CSF of Sdc-1^−/− mice either by dot blots or by ELISA (data not shown). However, q-PCR revealed that CCL20 mRNA expression was comparable in WT and Sdc-1^−/− mice at both early and peak EAE (Fig. 5E), suggesting that it was normally produced but not retained on the epithelial cell surface. By contrast, IL-6 mRNA expression was significantly higher in the choroid plexus of Sdc-1^−/− mice at early and peak EAE compared with WT mice (Fig. 5A), which correlated with 4- to 40-fold higher levels of IL-6 in CNS extracts at early and peak EAE but not in the recovery phase (Fig. 5C). Because IL-6 can be produced by many tissues and its source in EAE remains ill-defined, q-PCR was performed to compare IL-6 mRNA levels in brain, spinal cord, and the choroid plexus of WT and Sdc-1^−/− mice, with or without EAE. We found that IL-6 mRNA levels were increased in brain, spinal cord, and choroid plexus during early and peak stages of EAE compared with naive situation (Supplemental Fig. 4B), but there were no differences between brains and spinal cords of WT and Sdc-1^−/− mice. Only in the choroid plexus were higher levels of IL-6 mRNA measured in Sdc-1^−/− mice compared with WT. This suggests that although IL-6 can be produced by other cells in the CNS, only the choroid plexus source is altered in the Sdc-1^−/− mice, thereby, accounting for the higher IL-6 protein levels measured in the Sdc-1^−/− brains.

Taken together, this suggests that shedding of syndecan-1 from choroid plexus epithelium at early and peak EAE correlates with elevated IL-6 secretion at the choroid plexus, contributing to elevated IL-6 protein level in the CNS. Because of the small amount of CSF that could be collected, we were not able to measure immune cell numbers in the CSF by flow cytometry. However, immunofluorescence staining of choroid plexus sections revealed higher numbers of CD45⁺ leukocytes in the CSF of Sdc-1^−/− compared with WT mice at the onset of EAE symptoms (days 9–10) (Fig. 6) that was no longer apparent at peak EAE (data not shown), suggesting enhanced initial recruitment of leukocytes via the choroid plexus in Sdc-1^−/− mice.

Because several chemokines that play a role in leukocyte recruitment have been shown to bind to syndecan-1 (14, 38) and because chemokines other than CCL20 may also be expressed at the choroid plexus, ELISA was used to measure levels of other chemokines that have been implicated in EAE, including CCL2, CCL3, CCL4, CCL5, CCL7, CXCL1, CXCL2, and CXCL10, in the CSF and also in the whole brain tissue. Several chemokines were elevated both in the CNS and in the CSF at early and peak EAE in Sdc-1^−/− compared with WT mice, with CCL2 and CXCL10 being particularly high in the CSF (Supplemental Fig. 4C, 4D), both of which carry heparan sulfate binding domains (39), raising the possibility that syndecan-1 control of chemokine gradients may not be restricted to CCL20.

Because only B cells show clear syndecan-1 expression and because syndecan-1 is used as a marker of plasma cells, which have been implicated in prolonged recovery in EAE (40), we examined plasma cell numbers in Sdc-1^−/− mice. Because we could no longer use syndecan-1 as a marker of plasma cells, we crossed the Sdc-1^−/− mice with the Blimp-1-GFP reporter mouse (41) and performed flow cytometry analyses of spleen, LN, bone marrow, and CNS of naive and EAE mice to determine plasma cell numbers. Fig. 7A shows that syndecan-1⁺CD45⁺ plasma cells are also Blimp-1-GFP⁺. Our data reveal that plasma cell numbers were elevated in the spleen, LN, and blood but not the bone marrow of both naive (data not shown) and EAE Sdc-1^−/− mice (Fig. 7B), which, in EAE mice, correlated with increased circulating levels of MOG_35–55-specific IgG (but not IgM levels) (Fig. 7C). However, we did not detect Blimp-1-GFP⁺ plasma cells in CNS of EAE WT and Sdc-1^−/− mice. Also, immunofluorescence microscopy performed at peak EAE and in recovery phases failed to reveal the presence of Blimp-1-GFP⁺ plasma cells in the brains of Sdc-1^−/− or WT mice or syndecan-1⁺B220⁺ plasma cells in the brains of WT mice (data not shown).

Our data reveal that syndecan-1 has effects at two distinct stages of EAE; one effect is at the initial recruitment of encephalitogenic T cells to the brain via the choroid plexus, which is due to syndecan-1 expression on choroid plexus epithelium, and the second effect is at the recovery phase, which is mainly because of syndecan-1 effects on immune cells. In both cases, syndecan-1 has immune-suppressive effects, and in its absence, disease severity is enhanced and the recovery phase is prolonged.

The choroid plexus is the first site of entry of the EAE-inducing Th17 cells into the brain (20, 34). It consists of villous structures that extend into the CSF-filled ventricular spaces that form a blood–CSF barrier at the level of tight junctions between epithelial cells of the choroid plexus. A CCL20 gradient across the choroid plexus epithelium has been previously shown to be required for the migration of the initial disease-inducing CCR6⁺ Th17 cells into the CSF from where they disseminate to the meningeal and perivascular spaces to further recruit leukocytes via the CNS parenchyma postcapillary venules (20, 34). We demonstrate strong expression of syndecan-1 on choroid plexus epithelium, which decreases with EAE development, because of its shedding into the CSF and correlates with reduced levels of CCL20 on the choroid epithelial cell surface. We could not detect direct binding of CCL20 to syndecan-1, despite previous reports on the interaction of this chemokine with glycosaminoglycans (42) and on the interactions of syndecan-1 with various chemokines, including RANTES/CCL5, eotaxin/CCL11, TARC/CCL17, MARC/CCL7, IL-8/CXCL8, and KC/CXCL1 (14, 38). In addition, we could not detect CCL20 in the CSF of EAE mice, but the absence of differences in q-PCR levels of CCL20 mRNA throughout EAE and between WT and Sdc-1^−/− choroid plexuses supports its continual expression and subsequent lack of retention at the choroid plexus epithelium surface, which we propose increases the steepness of gradient across the choroid plexus epithelium and promotes migration of the CCR6⁺ Th17 cells into the CSF. It is important to note that the detection of CCL20 in direct binding assays and in CSF is likely to be handicapped by the tools currently available for its detection. The increased levels of several other chemokines that carry heparan sulfate–binding motifs in the CSF of the Sdc-1^−/− mice, in particular CCL2 and CXCL10, raises the possibility that syndecan-1 modulation of chemokine gradients at the choroid plexus may not be restricted to CCL20. Importantly, the effects of syndecan-1 loss on CCL20 at the choroid plexus is independent of the reported CCR6 priming functions at other sites (43, 44) where syndecan-1 is not expressed, such as the LN and spleen.

In addition to the association between syndecan-1 and the CCL20 localization on choroid plexus epithelium, our data demonstrate that the absence of syndecan-1 associated with increased IL-6 secretion at the choroid plexus. Sdc-1^−/− mice have inherently higher CNS levels of IL-6 compared with WT mice, which is further exacerbated during EAE and which we show is associated with enhanced initial leukocyte recruitment into the CSF via the choroid plexus and increased numbers of Th1 and Th17 cells in the CNS. q-PCR of the spinal cord, brains, and choroid plexuses from naive and EAE WT and Sdc-1^−/− mice showed that IL-6 expression increases at all sites in EAE but only at the choroid plexus were higher levels of IL-6 mRNA measured in Sdc-1^−/− mice compared with WT. This suggests that although IL-6 can be produced by other cells in the CNS, only the choroid plexus epithelial cell source is abnormal in the Sdc-1^−/− mice, accounting for the higher IL-6 protein levels in the Sdc-1^−/− brains. Data from both WT and Sdc-1^−/− mice showed a correlation between loss of syndecan-1, in the case of WT mice by shedding into the CSF during EAE, and upregulation of IL-6 mRNA expression at the choroid plexus that was paralleled by elevated levels of IL-6 in the CNS, which we proposed may further contribute to enhanced survival of the encephalitogenic Th1 an Th17 cells in the CNS as previously shown by others (45).

Loss of syndecan-1 has been shown to be associated with elevated levels of tissue or serum levels of IL-6 in several inflammatory (14, 46) and tumor models (8, 47); however, the converse has also been reported (48, 49). In vitro silencing of syndecan-1 expression has been reported to both upregulate (47) and downregulate (48) IL-6 mRNA in endometrial stromal cells and epithelial cells, respectively. In mesothelioma cells, upregulation of syndecan-1 expression is associated with increased IL-6 expression (49), suggesting that the effects of syndecan-1 on IL-6 are likely to be influenced by factors such as cell activity or pathophysiological state. How such effects are mediated is not clear but may be due to the high growth factor binding potential of syndecan-1 (3), which subsequently impacts on IL-6 expression. In breast cancer cells, downregulation of syndecan-1 also results in reduced cellular responses to IL-6 (8), whereas IL-6 downregulates syndecan-1 in a variety of cell types, including cells of liver origin (50), osteosarcoma cell lines (51), and B lymphoid cells (52), suggesting an interdependency between syndecan-1 and IL-6 regulatory pathways.

IL-6 is known to specifically bind to heparan sulfate glycosaminoglycan chains (53, 54), and we show in this study that IL-6 binds to syndecan-1, which has been proposed to provide a mode of retaining IL-6 close to its site of secretion, thus favoring a paracrine mode of action (53). IL-6 can both inhibit or stimulate its own secretion depending on cell type; it is possible therefore that in the naive situation syndecan-1–bound IL-6 acts locally at the choroid plexus epithelium to inhibit its own secretion and that this inhibition is lifted when syndecan-1 is shed during EAE or is absent as in the Sdc-1^−/− mouse. This theory requires in vitro testing that, however, is likely to be difficult given the variability noted here in the expression of syndecan-1 in in vitro–cultured (endothelial) cells.

IL-6 plays several roles in inflammation (55), and its significance in EAE is reflected in the complete resistance to disease of mice lacking IL-6 (37, 56). However, the critical source of IL-6 in EAE has remained difficult to define. DCs have recently been shown to be an important source of IL-6 during EAE, but only transiently during the early T cell activation stages (57). Our data suggest that IL-6 expression at the choroid plexus occurs subsequent to T cell activation, principally at the stage of early recruitment of T cells via the choroid plexus and up to peak EAE. To our knowledge, this is the first description of IL-6 expression at the choroid plexus and, together with other published data (57), supports the concept that IL-6 is produced by different cell types at different stages of EAE.

Our studies revealed that the endothelial cells of perivascular cuffs in the brain did not express detectable amounts of syndecan-1. Even though the mRNA for syndecan-1 was detectable in freshly isolated brain endothelial cells, levels were >10-fold lower than in choroid plexus epithelium, and the protein was not detectable by immunofluorescence staining, regardless of unmasking techniques used. This is in contrast to others who have reported syndecan-1 expression on endothelium, albeit in tissues other than brain (6, 19, 58). Past studies have also focused on in vitro–cultured endothelial cell lines, which we found to have significantly higher levels of syndecan-1 mRNA than freshly isolated primary endothelial cells, suggesting that the in vivo milieu may control the syndecan-1 expression on endothelium and that this varies with tissue type. This is supported by reports that syndecan-1 can be induced or repressed in a transient manner during development and in a variety of pathological processes (9).

The bone marrow chimeric studies suggest a role for an immune cell source of syndecan-1 in the recovery phase of EAE; however, Th1, Th17, and Treg numbers were similar in the WT and Sdc-1^−/− mice during the recovery phase, and IL-6 CNS levels were also not significantly different. Rather, the slower recovery for EAE symptoms observed in the Sdc-1^−/− mice may be partially accounted for by the slightly elevated numbers of macrophages (albeit not significant). In addition, the low syndecan-1 expression detected on B cells prompted us to also invest a role for B cells. B cells have been proposed to have an Ag-presenting role in EAE (29); for this reason, we tested the ability of B cells from the Sdc-1^−/− mice to act as APCs, revealing no differences to WT B cells.

The prolonged recovery phase in Sdc-1^−/− mice correlated with higher numbers of MOG-specific plasma cells in the spleen, LNs, and circulation but not in the brain and with elevated serum titers of MOG_33–55 Ab, which has been previously reported to correlate with enhanced demyelination and slower EAE recovery (29, 40). How MOG specific Abs can prolong recovery in EAE is not clearly defined; in mice, inflammation in the CNS can result in demyelination in the absence of a MOG-specific B cells response, but, if present, Abs will enhance disease severity and demyelination. The mode by which the MOG-specific Abs act is likely to include direct binding to Ag and activation of complement cascades or FcR-mediated tissue destruction. In EAE, Abs have been shown to have both pathological and regulatory effects (59), with those recognizing conformational epitopes being pathogenic (60) and generally of the IgG class (61) as shown in this study in the Sdc-1^−/− mice. The fact that Sdc-1^−/− mice had inherently higher numbers of plasma cells indicate that syndecan-1 is not only a marker for plasma cells but is also a negative regulator of plasma cell formation, a novel finding that requires further investigation. That plasma cell numbers were elevated in the secondary lymphoid tissues but not the bone marrow of Sdc-1^−/− mice suggests that short-lived plasma cells that appear in the secondary lymphoid tissue shortly after Ag exposure rather than long-lived plasma cells that reside in the bone marrow are principally affected by loss of syndecan-1.

In summary, we demonstrate immunosuppressive effects for syndecan-1 in EAE and provide new information on molecular processes important for initial encephalitogenic T cell recruitment to the brain via the choroid plexus and for disease recovery. Our data support a role for syndean-1 in modulation of the immune response by contributing to cytokine and potentially also chemokine gradients, interestingly at the level of the choroid plexus epithelial barrier rather than endothelial barrier, as well as on plasma cell development.

We thank Stefan Liebner (University of Frankfurt, Frankfurt, Germany) for the gift of brain-derived endothelial cells; Johanna Breuer, Maik Worlitzer, Jens Swanborn, and Nicholas Schwab (University of Muenster) for teaching and helping us collect CSF; Stephen Nutt and David Tarlington (Walter and Eliza Hall Institute, Melbourne, VIC, Australia) for the generous gift of the Blimp-1-GFP mice; and Rupert Hallmann for critical reading of the manuscript. We also thank Merton Bernfield (Boston Children’s Hospital, Boston, MA) who provided Sdc-1^−/− mice.

The authors have no financial conflicts of interest.