Functional expression of the lymphoid chemokines CCL19 (ELC) and CCL 21 (SLC) at the blood-brain barrier suggests their involvement in G-protein-dependent lymphocyte recruitment into the central nervous system during experimental autoimmune encephalomyelitis

Authors

  • Carsten Alt,

    1. Max-Planck Institute for Physiological and Clinical Research, W. G. Kerckhoff-Institute, Department of Vascular Cell Biology, Bad Nauheim, Germany
    Search for more papers by this author
  • Melanie Laschinger,

    1. Max-Planck Institute for Physiological and Clinical Research, W. G. Kerckhoff-Institute, Department of Vascular Cell Biology, Bad Nauheim, Germany
    Search for more papers by this author
  • Britta Engelhardt

    Corresponding author
    1. Max-Planck Institute for Physiological and Clinical Research, W. G. Kerckhoff-Institute, Department of Vascular Cell Biology, Bad Nauheim, Germany
    2. Max-Planck-Institute for Vascular Biology, Münster, Germany
    • Max-Planck-Institute for Vascular Biology, c/o ZMBE, Institute for Cell Biology, von-Esmarch-Str. 56, D-48149 Münster, Germany Fax: +49-251-83-58616
    Search for more papers by this author

Abstract

Migration of autoaggressive T cells across the blood-brain barrier (BBB) is critically involved in the initiation of experimental autoimmune encephalomyelitis (EAE), an animal model for multiple sclerosis. The direct involvement of chemokines in this process was suggested by our recent observation that G-protein-mediated signaling is required to promote adhesion strengthening of encephalitogenic T cells on BBB endothelium in vivo. To search for chemokines present at the BBB, we performed in situ hybridizations and immunohistochemistry and found expression of the lymphoid chemokines CCL19/ELC and CCL21/SLC in venules surrounded by inflammatory cells. Their expression was paralleled by the presence of their common receptor CCR7 in inflammatory cells in brain and spinal cord sections of mice afflicted with EAE. Encephalitogenic T cells showed surface expression of CCR7 and the alternative receptor for CCL21, CXCR3. They specifically chemotaxed towards both CCL19 or CCL21 in a concentration dependent and pertussis toxin-sensitive manner comparable to naive lymphocytes in vitro. Binding assays on frozen sections of EAE brains demonstrated a functional involvement of CCL19 and CCL21 in adhesion strengthening of encephalitogenic T lymphocytes to inflamed venules in the brain. Taken together our data suggest that the lymphoid chemokines CCL19 and CCL21 besides regulating lymphocyte homing to secondary lymphoid tissue are involved in T lymphocyte migration into the immunoprivileged central nervous system during immunosurveillance and chronic inflammation.

Abbreviations:
BBB:

Blood-brain barrier

MS:

Multiple sclerosis

PTX:

Pertussis toxin

tEAE:

Transferred EAE

1 Introduction

The central nervous system (CNS) is considered an immunoprivileged site, where the endothelial blood-brain barrier (BBB) tightly controls lymphocyte entry into the CNS. Under physiological conditions lymphocyte traffic into the CNS is low, whereas during inflammatory diseases of the CNS such as multiple sclerosis (MS) or in the animal model experimental autoimmune encephalomyelitis (EAE)large numbers of circulating lymphocytes readily gain access to the CNS. EAE is induced by auto-aggressive T lymphoblasts, which migrate into the CNS, causing the molecular events leading to edema,inflammation and demyelination 1. Thus, the interaction of circulating encephalitogenic T cells with the healthy BBB endothelium is a critical step in the initiation of EAE. Later during ongoing disease the continuous interaction of circulating leukocytes with the inflamed BBB maintains the inflammatory lesion during chronic autoimmune disease.

In general, lymphocyte recruitment across the vascular wall involves lymphocyte interaction with vascular endothelium under flow. At the molecular level this process is mediated by the sequential interaction of cell adhesion molecules such as the selectins, the integrins and members of the Ig-superfamily, and signals provided by chemokines on the endothelial surface, which bind to their specific receptors on the leukocyte surface 2. Chemokines comprise a family of small chemoattractant proteins that bind to G-protein-coupled receptors expressed on target cells. Chemokine receptor signaling leads to an increase in integrin avidity on the leukocyte surface, allowing the leukocyte to firmly bind to the endothelium under flow and subsequently to follow chemokine concentration gradients across the vascular wall into selected tissues. Chemokines can either be classified into four major groups, namely CXC, CC, C or CX3C, according to a cysteine-containing sequence motif near the N terminus 3 or into two groups, i.e. lymphoid or inflammatory chemokines, on the basis of the conditions and location of their production. Lymphoid (constitutive or homeostatic) chemokines are involved in maintaining physiological trafficking and positioning of cells of the adoptive immune system during hematopoiesis and immunosurveillance, and the inflammatory (or induced) chemokines, the expression of which is induced during inflammation, mediate the recruitment of inflammatory cells into these sites 4.

Using intravital fluorescence videomicroscopy we were recently able to assess the interaction of encephalitogenic T cell blasts with the white matter BBB-endothelium in SJL/N mice in vivo 5. We found that interaction of encephalitogenic T cell blasts with the spinal cord white matter microvasculature is unique as, in absence of rolling, α4–integrin mediates the G-protein-independent prompt arrest (capture) followed by the G-protein-dependent adhesion strengthening of T cell blasts to microvascular VCAM-1. Inevitably, chemokines involved in the integrin-mediated arrest of autoaggressive T lymphoblasts at the BBB must be expressed and/ or at least present at the level of the BBB endothelium. To identify such chemokines, we performed in situ hybridizations and immunohistochemistry for a large panel of chemokines on frozen sections of brain and spinal cord of mice afflicted with EAE and healthy littermates as a control. To our surprise we detected constitutive expression of the lymphoid chemokine EBV-induced molecule 1 ligand chemokine (ELC)/CCL19 in a subpopulation of CNS venules and induced expression of the secondary lymphoid chemokine (SLC)/CCL21 in inflamed CNS venules during EAE. Expression of their common receptor CCR7 was detected in a subpopulation of cells present in the perivascular inflammatory cuffs. Encephalitogenic T cells showed surface expression of CCR7 and the alternative receptor for CCL21, CXCR3. They specifically chemotaxed towards both CCL19 or CCL21 in a concentration-dependent and pertussis toxin (PTX)-sensitive manner, comparably to naive lymphocytes in vitro. Additionally, in frozen section assays functional deletion of CCR7 and CXCR3 or blocking CCL19 and CCL21 reduced binding of encephalitogenic T lymphocytes to inflamed venules in the brain. Taken together, our data suggest that the lymphoid chemokines CCL19 and CCL21 are produced by brain vascular endothelial cells, and involved in α4-integrin-mediated adhesion strengthening of autoaggressive T lymphocytes and, later during disease, possibly of circulating inflammatory cells to the BBB endothelium, leading to the maintenance of inflammation during chronic autoimmune disease.

2 Results

2.1 Pretreatment of encephalitogenic T cells with PTX delays the onset of EAE

Based on our recent finding that in vivoi-protein-dependent α4–integrin/VCAM-1-mediated binding of autoaggressive T cells to the spinal cord microvasculature is required for adhesion strengthening, we asked whether Gαi-protein mediated signaling in encephalitogenic T cell blasts is required for their ability to transfer EAE. Therefore, Gαi-protein signaling was inhibited by in vitro treatment of encephalitogenic T lymphoblasts with PTX or its enzymatically inactive mutant MTX prior to their transfer into syngeneic recipients. Whereas mice injected with untreated or MTX-pretreated encephalitogenic T lymphoblasts developed clinical EAE starting at days 8.6±0.5 and 8.4±1.1 after transfer of T cells, respectively, mice injected with PTX-pretreated T lymphoblasts showed a significantly delayed onset of transferred EAE (tEAE) (day 13.0±1.2; Fig. 1). Thus, intact Gαi-protein-mediated signaling in autoaggressive T cells is required for their successful entry into the CNS and the development of EAE, suggesting the involvement of chemokines in the recruitment process.

Figure 1.

Pretreatment of encephalitogenic T cells with PTX delays the development of clinical tEAE. SJL.PLP7 were freshly stimulated with PLP peptide amino acids 139–151 in vitro and treated with PTX or the enzymatically inactive mutant MTX. Clinical disease scores were recorded daily for 15 days. Curves show mean ± SD for five animals per group. The data shown for freshly stimulated and PTX-treated T cells are representative for five independent experiments. The data shown for MTX are representative for three independent experiments.

2.2 Lymphoid chemokines CCL19 and CCL21 are expressed at the BBB

To search for chemokines expressed at the level of the BBB, we performed in situ hybridizations for a large panel of chemokines on brain and spinal cord sections of SJL/N mice, afflicted with EAE and of healthy SJL/N mice as a control (Table 1, 2). We did not detect expression of chemokine mRNA, with the exception of CCL19/ELC and CXCL12/SDF-1, in brain and spinal cord sections of healthy SJL/N mice using in situ hybridization. In contrast to CXCL12, which was detected within the choroid plexus, in astrocytes or microglial cells scattered throughout the CNS parenchyma and in some rare perivascular cells (data not shown), expression of CCL19 was localized to cerebral venules as shown by a hybridization signal exactly resembling expression of the endothelial cell-specific VEGF receptor 2 (Fig. 2).

During EAE, expression of CCL19 was strongly up-regulated and found in almost all venules surrounded by perivascular cuffs (Fig. 2 and Table 2). Additionally, expression of the lymphoid chemokine CCL21/SLC was found to be induced in almost all venules surrounded by inflammatory cuffs in brain and spinal cord sections of mice afflicted with active EAE (Fig. 3). Expression of the common receptor for both chemokines, CCR7, could be detected in a subpopulation of cells present within inflammatory cuffs (Fig. 3). As a control for CNS expression of CCL19 and CCL21, we performed in situ hybridization for both chemokines on frozen sections of lymph nodes, where their expression has been described before in detail 6, 7. We detected CCL19 expression in stromal cells located in the T cell zones of peripheral lymph nodes derived from healthy SJL/N mice (Fig. 3). Specific expression of CCL21 mRNA could be detected in the high endothelial venules (HEV) at levels comparable to those found in inflamed venules in the CNS of mice afflicted with EAE (Fig. 3). Furthermore, CCR7 expression was observed in a number of parenchymal cells within the lymph nodes (Fig. 3).

No other chemokine investigated here was found to be induced in BBB endothelial cells during EAE (Table 2). Their expression was observed within the brain or spinal cord parenchyma, either in cells localized within inflammatory cuffs or alternatively in astrocytes or microglial cells in no spatial relation to the distribution of cellular infiltrates (Table 2). The CNS expression pattern of CXCL12/SDF-1 during EAE was indistinguishable from its expression pattern in the healthy CNS. Finally, expression of XCL1/lymphotactin was not detected in brain or spinal cords of mice afflicted with EAE.

Thus, in contrast to the expression of the inflammatory chemokines, which was only observed in cells within the CNS parenchyma, we detected constitutive expression of the lymphoid chemokine CCL19 and inducible expression of lymphoid chemokine CCL21 in CNS venules. The presence of CCR7 receptor expression in infiltrating cells during EAE suggests a possible involvement of these chemokine/receptor pairs in the recruitment of inflammatory cells across the BBB.

Table 1. In situ hybridization probesa)
ChemokineClone IDFragment
  1. a) Listed are the cDNA clones with their corresponding IMAGE consortium clone ID and the restriction endonucleases used to obtain fragments for subcloning.

  2. b) Clones were kindly provided by H. Augustin.

CXCL9IMAGE : 2135986Xho1
CXCL10b)EcoRI/Xba1
CXCL12b)EcoRI/Xba1
CCL2b)BamHI/EcoRl
CCL3IMAGE : 533862EcoRI/Xho1
CCL4IMAGE : 621095EcoRI/Not1
CCL5IMAGE : 832342EcoRI/Not1
CCL6IMAGE : 1195057EcoRI/Not1
CCL19IMAGE : 832043EcoRI/Not1
CCL21IMAGE : 389013EcoRI/Pst1
CCL22IMAGE : 619257EcoO109I/Spe1
XCL1IMAGE : 576815EcoRI/Not1
Figure 2.

 CCL19 is expressed at the BBB in the healthy CNS. In situ hybridization analysis of CCL19 mRNA expression in the healthy brain of an SJL/N mouse is shown. A hybridization signal for CCL19 can be localized to venules of the CNS of healthy SJL/N mice with the CCL19 antisense probe, whereas no hybridization signal is seen with the CCL19 sense control probe. The endothelial cell specific hybridization signal for VEGF-receptor-2 mRNA is shown for comparison. Sections are counterstained with toluidine blue. Bar=20 μm.

Figure 3.

CCL19 and CCL21 are expressed at the inflamed BBB. In situ hybridization analysis for CCL19, CCL21 and their common receptor CCR7 in the brain of an SJL/N mouse afflicted with EAE (day 14, clinical score 2/3) and in a peripheral lymph node derived from a healthy SJL/N mouse is shown. A specific hybridization signal for CCL19 and CCL 21 mRNA can be localized to inflamed vessels surrounded by inflammatory cells, whereas a hybridization signal for CCR7 mRNA can be localized to individual cells present within the inflammatory cuff. Control in situ hybridization for both lymphoid chemokines and their receptor on peripheral lymph nodes demonstrates their known expression within the lymph node parenchyma (CCL19, CCR7) and in high endothelial venules (CCL21). Sections are counterstained with toluidine blue. Bar=20 μm.

Table 2. Semiquantitative analysis of the expression of chemokine mRNA during EAEa)
 brainspinal cord
 inflammatory cellsendothelial cellsinflammatory cellsendothelial cells
  1. a) Semiquantitative analysis was performed by counting the total number of inflammatory cuffs per section in relation to the number of inflammatory cuffs with specific in situ hybridization signals. Results were divided into three groups: ++, >50% of the infiltrates show positive cells; +, <50% of the infiltrates show positive signals; –, no detectable hybridization signal.

  2. b) Expression was detected in plexus choroideus and in brain parenchymal cells not spatially correlated to inflammatory cuffs.

CXCL9 (MIG)++++
CXCL10 (IP-10)++++
CXCL12 (SDF-1)b)++
CCL2 (MCP-1)+++
CCL3 (MIP-1α)+++
CCL4 (MIP-1β)+++
CCL5 (RANTES)+++
CCL6 (C 10)++++
CCL19 (ELC)++++
CCL21 (SLC)++++
CCL22 (MDC)++
XCL1 (Lymphotactin)

2.3 CCL19 and CCL21 are present in inflammatory cuffs in the CNS of mice afflicted with EAE

To localize CCL19 and CCL21 protein in the CNS of SJL/N mice we performed immunohistochemistry on brain sections of mice afflicted with EAE and healthy littermates. In concordance with the mRNA expression pattern observed in the CNS of healthy SJL/N mice specific immunostaining for CCL19 was detected on a subpopulation of venules, whereas CCL21 was not detected in the healthy CNS (Fig. 4). During EAE, immunostaining for both CCL19 and CCL21 was observed within inflammatory cuffs (Fig. 4). Staining was mostly diffuse but strictly restricted to inflammatory cuffs, suggesting that both chemokines might be released by the BBB endothelial cells into the perivascular space or might be additionally detected bound to their specific receptors on infiltrating cells.

Figure 4.

Detection of CCL19 and CCL21 on the protein level in healthy and EAE brain. In healthy brain, immunostaining for CCL19 but not CCL21 can be demonstrated on a subpopulation of venules throughout the brain. In EAE, immunostaining for both chemokines can be detected within inflammatory cuffs. Identical staining with several control sera showed no background staining (data not shown). Immunoperoxidase staining with hematoxylin counterstain. Bar=20 μm.

2.4 CCR7 and CXCR3 are present on encephalitogenic T lymphocytes

In addition to its known receptor CCR7, mouse CCL21 has also been shown to bind and signal via the CXC receptor CXCR3 8. Using immunohistochemistry on cytospin preparations of encephalitogenic T lymphoblasts, we investigated the presence of CCR7 and CXCR3, and detected specific staining for both CCR7 and CXCR3 (Fig. 5). This suggests that CCL19 and CCL21 can bind to their specific receptors on encephalitogenic T lymphoblasts.

Figure 5.

CCR7 and CXCR3 are present on encephalitogenic T cell blasts. Immunostaining for CCR7 and CXCR3 on cytospin preparations of SJL.PLP6 are shown. After fixation with acetone T cells were stained with purified goat anti CCR7 and purified goat anti-mouse CXCR3. The apparent intracellular staining for CCR7 is partially explained by the fact that the antibody is directed against the intracellular C terminus. Staining for CD45 is shown as positive control, and staining with irrelevant goat IgG as a negative control. Both chemokine receptors can be detected on encephalitogenic T lymphoblasts. Immunoperoxidase staining with hematoxylin counterstain. Bar=20 μm.

2.5 Encephalitogenic T lymphocytes specifically chemotax towards CCL19 and CCL21

Next we asked whether encephalitogenic T cells can specifically chemotax towards CCL19 and CCL21 in a two-chamber system in vitro. Already in the absence of either chemokine encephalitogenic T lymphocytes spontaneously migrated across fibronectin-coated filters into the lower chamber (Fig. 6). Addition of increasing concentrations of either CCL19 or CCL21 to the lower chambers significantly increased the number of migrating T lymphocytes. Migrating was concentration dependent, with CCL19 having an effect at concentrations as low as 1 ng/ml and CCL21 at 10 ng/ml. Interestingly, only migration towards CCL19, but not towards CCL21, reached a plateau level. The simultaneous presence of both chemokines into the lower chamber did not increase the migration efficacy (data not shown). CCL19 or CCL21 induced chemotaxis of T lymphoblasts was comparable to that observed for lymphocytes freshly isolated from lymph nodes (Fig. 6). Migration was specifically induced by the investigated chemokines as pretreatment of either chemokine with its respective blocking antibody reduced the migration rate of autoaggressive T lymphocytes to background levels (Fig. 6). Furthermore, increased migration towards both chemokines required intact Gαi-protein signaling of the T lymphocytes as pretreatment with PTX but not MTX reduced migration to background levels irrespective of the chemokine concentration (Fig. 6). Thus, in vitro CCL19 and CCL21 specifically induce G-protein-dependent chemotaxis of encephalitogenic T cells.

Figure 6.

Encephalitogenic T cells specifically chemotax towards CCL19 and CCL21. (A) Migration of SJL.PLP6 across fibronectin-coated Transwell filters is shown. SJL.PLP6 migrated in a concentration-dependent manner towards CCL19 or CCL21 (triangles). Migration could be specifically reduced to spontaneous migration levels by pretreatment of the chemokine with its respective blocking antibody (bars; 50 μg/ml). Pretreatment of SJL.PLP6 with PTX (stars) but not MTX (diamonds) abrogated the directed migration to the level of spontaneous migration. (B) Comparison of chemotaxis of encephalitogenic T cells (triangles) and lymphocytes (circles) freshly isolated from lymph nodes is shown. Encephalitogenic T cells demonstrate increased migration towards CCL19 and CCL21 at lower concentrations compared to increased migration of lymph node lymphocytes.

2.6 Functional deletion of CCR7 and CXCR3 or blocking CCL19 and CCL21 reduces binding of encephalitogenic T lymphocytes to inflamed brain venules in vitro

To obtain more direct proof for the functional involvement of CCL19 and CCL21 in adhesion strengthening of encephalitogenic T lymphocytes to CNS venules, we performed Stamper-Woodruff-binding assays on unfixed frozen sections of EAE brains 9. Encephalitogenic T lymphocytes selectively bound to inflamed venules exposed in frozen sections of EAE brains (Fig. 7). Pretreatment of T lymphocytes with PTX or specific functional deletion of CCR7 and CXCR3 by desensitization with CCL19 and CCL21 significantly reduced the number of T lymphocytes firmly adhering to inflamed venules in EAE brains (Fig. 7). Alternatively, pretreatment of frozen sections with blocking antibodies directed against CCL19 and CCL21 also reduced the number of T lymphocytes firmly adhering to cerebral vessels of EAE brains. Thus, functional binding of CCL19 and CCL21 to their respective receptor(s) is involved in inducing adhesion strengthening of encephalitogenic T lymphocytes to inflamed venules in EAE brains.

Figure 7.

CCL19 and CCL21 are involved in adhesion strengthening of encephalitogenic T cells to inflamed vessels in EAE brain. Adhesion of the encephalitogenic T cells to inflamed venules exposed in serial frozen sections of EAE brains is shown for each condition. Pretreatment of T lymphocytes with PTX significantly reduced the number of firmly adhering T cells (a: control, 26 bound T cells; b: PTX-pretreated T cells, 3 bound T cells) as did desensitization of T lymphocytes with CCL19 and CCL21 (c: control, 15 bound T cells; d: desensitized T cells, 8 bound T cells). Also blocking CCL19 and CCL21 interfered with adhesion strengthening of T cells (e: control, 23 bound T cells; f: in presence of blocking antibodies, 8 bound T cells). Toluidine blue counterstain. Bar=50 μm.

3 Discussion

The crucial role of chemokines in the successful recruitment of encephalitogenic T cells across the BBB is suggested by the requirement for signaling via G-protein-coupled, PTX-sensitive receptors by encephalitogenic T cells to firmly arrest on BBB endothelium in vivo5 and, as shown here, to successfully transfer EAE. Searching for candidate chemokines present at the BBB, we found expression of a large panel of inflammatory chemokines in the CNS parenchyma during EAE, but only the lymphoid chemokines CCL19 and CCL21 were expressed at the BBB endothelium in vivo. CCR7, the common receptor for CCL19 and CCL21 and the alternative CCL21 receptor, CXCR3, were found on encephalitogenic T lymphoblasts that specifically chemotaxed towards both chemokines in a concentration-dependent manner comparable to naive lymphocytes in vitro. Furthermore, binding assays on frozen sections of EAE brains showed a functional involvement of CCL19 and CCL21 in adhesion strengthening of encephalitogenic T lymphocytes to inflamed venules in the brain. Thus, the lymphoid chemokines CCL19 and CCL21 are functionally expressed at the BBB and could, therefore, be critically involved in the initiation and chronic maintenance of CNS inflammation during EAE.

Chemokines have been previously implicated in the pathogenesis of EAE, as CNS expression of high levels of a diversity of inflammatory chemokines like CCL2/ MCP-1, CCL3/MIP-1α, CCL5/RANTES and CXCL10/ CRG-2/IP-10 was observed in tight temporal correlation to disease onset, and in tight spatial correlation to the appearance of inflammatory infiltrates around CNS venules (reviewed in 10). In accordance with our in situ hybridization studies, previous studies demonstrated that the major producing source for CXCL10/IP-10 and CCL2/MCP-1 was astrocytes 11 and for CCL3/MIP-1α, CCL4/MIP-1β, CCL5/RANTES and CCL6/C10 was both astrocytes and mononuclear cells within perivascular cuffs 12, 13. In addition, in this study we found similar expression patterns for CXCL9/MIG and CCL22/MDC. It is important to note that the major cellular sources for inflammatory chemokines detected during EAE were produced behind the BBB.

Functional evidence for the involvement of CNS chemokines in the pathogenesis of EAE was derived from in vivo neutralization studies where blocking antibodies directed against CCL3/MIP-1α or CCL2/MCP-1 inhibited onset of EAE or ameliorated chronic EAE, respectively 14. Additionally, antibodies against CXCL10/IP-10 were shown to interfere with the development of EAE transferred with encephalitogenic T cells 15. These data are complemented by investigations performed in mice deficient for certain chemokines or chemokine receptors. Mice deficient in CCL3/MIP-1α or its receptor CCR5 develop EAE indistinguishable from wild-type mice 16, whereas mice deficient for CCR1 show a milder disease compared to controls 17. In addition, mice deficient for CCL2/MCP-1 or its receptor CCR2 completely fail to develop EAE after immunization with autoantigens 1820. Most reports suggested that chemokines induced in EAE mediate the migration of circulating inflammatory cells across the BBB. Interestingly though, encephalitogenic T cells lacking CCR2 are still able to induce EAE in syngeneic wild-type recipients demonstrating that CCR2 is not required by encephalitogenic T cells for entry into the CNS 19. Rather, these data suggest that CCL2/MCP-1 induced in the CNS during EAE and its receptor CCR2 are important in eliciting effector functions of inflammatory cells within the CNS necessary for the manifestation of EAE.

In contrast to microvessels elsewhere, the highly specialized endothelium of the BBB forms a physical barrier between blood and CNS, where paracellular or transcellular diffusion of chemokinesacross the BBB is hindered by the complex tight junctions between the BBB endothelial cells and the lack of pinocytotic activity of the BBB endothelium, respectively. Therefore, the obvious question, how chemokines produced within the CNS parenchyma could possibly attract lymphocytes in the blood stream from which they are separated by the BBB, arises. In vivo transport of chemokinesfrom the abluminal to the luminal surface of endothelial cells has so far only been observed in dermal microvessels 21 and in HEV of peripheral lymph nodes 22.Although in vitro the inflammatory chemokines CCL2/MCP-1 and CCL3/MIP-1α can bind to the abluminal side of freshly isolated human brain microvessels 23, 24, it remains to be shown whether these chemokines can be transported across the BBB in vivo and induce G-protein-dependent leukocyte arrest to the luminal surface of the BBB endothelium.

Therefore, we hypothesized that chemokines mediating the G-protein-dependent arrest of encephalitogenic T cells to the BBB in vivo are most probably produced by the BBB endothelium itself. In search for chemokines expressed by BBB endothelium we performed in situ hybridizations on brain and spinal cord sections of SJL/N mice afflicted with EAE and of healthy littermates. Out of 12 chemokines investigated none of the inflammatory chemokines was found to be induced at the BBB during EAE. In contrast, we detected constitutive expression of the lymphoid chemokine CCL19/ELC in CNS venules of healthy animals and its up-regulation in inflamed venules in the CNS of mice afflicted with EAE. Additionally, expression of the lymphoid chemokine CCL21/SLC, which was not expressed at the healthy BBB, was induced in inflamed CNS venules during EAE.

CCL19 and CCL21 are structurally related chemokines referred to as lymphoid chemokines, because they are constitutively expressed within secondary lymphoid tissue (summarized in 4, 6), where they regulate the homing of T and B cells and dendritic cells into lymph nodes and the formation of secondary lymphatic tissue by acting via their common receptor CCR7, as demonstrated in CCR7-deficient mice 25 and a mutant mouse, the plt (paucity of lymph node T cells) mouse, lacking detectable levels of both CCL19 and CCL21 expression 26, 27. This is further supported by the observation that CCR7 and its ligands CCL19 and CCL21 regulate lymphocyte homing from blood into lymphoid tissues by triggering their integrin-mediated arrest on the HEV in vivo28.

Outside of lymphoid tissue constitutive expression has only been observed for CCL21, but not for CCL19 in lymphatic endothelium. Inducible expression of CCL19 and CCL21 outside of secondary lymphatic tissue has been observed during rheumatoid arthritis 29 and experimental autoimmune diabetes 30 in spatial and temporal correlation to the formation of tertiary lymphoid tissue in the respective target organs. In EAE, development of ectopic lymphatic tissue within the CNS has not been observed. However, induction of the cytokines lymphotoxin-α and TNF-α, which in lymphoid tissue have been shown to maintain the constitutive expression of CCL19 and CCL21 31, are induced in the CNS during EAE 32, 33. Therefore, these cytokines could be responsible for the expression of both chemokines at the BBB.

EAE can only be transferred by freshly activated, but not by resting, autoantigen-specific CD4+ T cells, usually from the Th1 subset, and it was suggested that this is due to their ability to access the CNS 34, 35. Expression of chemokine receptors on the surface of T cells has been reported to be regulated such that, upon activation, T cells lose lymphoid-organ homing receptors like L-selectin and CCR7 and gain expression of other subsets of adhesion molecules or chemokine receptors, enabling efficient immunosurveillance of non-lymphoidtissues for injury and infection 2. In human peripheral blood activated/memory T cells have been divided into two populations based on their CCR7 expression. Whereas CCR7-negative T cells were defined as effector/memory T cells producing large amounts of inflammatory cytokines, CCR7-positive T cells lacking profound cytokine production showed surface expression of L-selectin and were, therefore, referred to as "central memory" T cells, which could possibly account for those activated/memory T cells able to migrate to lymph nodes 36. At the time point of their injection, encephalitogenic T cells are maximally activated effector/memory T cells, producing high amounts of pro-inflammatory cytokines and, despite their staining positively for CCR7, lack surface expression of L-selectin at that stage 37. Thus, it seems that in vitro cultured encephalitogenic T cells can not be grouped into any of these previously described phenotypes.

The chemokines or chemokine receptors involved in the homing of effector/memory T lymphocytes into non-lymphoid tissue during immunosurveillance have not been well characterized. At a first glance it seems surprising that encephalitogenic T cells should use CCR7 and the lymphoid chemokines CCL19 and CCL21 expressed at the BBB to specifically migrate into the CNS. However, recently we demonstrated that T lymphoblasts require high-affinity α4-integrin to be captured on VCAM-1 expressed by CNS microvessels in vivo under flow 5. Whereas α4-integrin-mediated capture of T cells was independent of G-protein signaling, the subsequent α4-integrin-mediated adhesion strengthening required G-protein-mediated signals in situ. As binding assays on frozen sections of EAE brains provided more proof for a functional involvement of CCL19 and CCL21 in adhesion strengthening of encephalitogenic T lymphocytes to inflamed venules in the brain in vitro, it is tempting to speculate that captured T cells can bind CCL19 or CCL21, which via CCR7 trigger an increase in α4-integrin avidity leading to the adhesion strengthening of encephalitogenic T cells on VCAM-1 in CNS venules and their transmigration as observed by us previously in vivo5. In contrast, naive T lymphocytes would never "see" the lymphoidchemokines present on BBB endothelium, because they lack high-affinity α4-integrin required to slow down in CNS venules. The lack of any known L-selectin ligands present on BBB endothelium hinders naive T cells to tether and roll via L-selectin on the BBB endothelium, a prerequisite in HEV for CCR7 engagement of CCL19 and CCL21. Vice versa, encephalitogenic T cells lacking L-selectin would rather home to the immunoprivileged CNS than to peripheral lymph nodes which lack luminal expression of VCAM-1 on their HEV. However, one can imagine that later during chronic inflammatory disease L-selectin ligands might be induced on the inflamed BBB, allowing then the migration of lymphocytes, which usually home to secondary lymphoid tissues, into the CNS, thus maintaining chronic CNS inflammation.

In summary, our findings suggest that the lymphoid chemokines CCL19 and CCL21 due to their expression at the BBB might, besides regulating lymphocyte homing to secondary lymphoid tissue, be involved in the migration of activated effector T lymphocytes across the healthy BBB into the immunoprivileged CNS and in the recruitment of mononuclear cells across the inflamed BBB during chronic autoimmune disease.

4 Materials and methods

4.1 Antibodies and reagents

Goat-anti-murine CCL19 (AF880), goat-anti-murine CCL21 (AF457) and normal goat IgG (AB108C) were all obtained from R&D Systems (Wiesbaden, Germany). Goat-anti-CCR7 (A19) and goat-anti-mouse CXCR3 (Y-16) were obtained from Santa Cruz (Heidelberg, Germany). Biotin-SP-conjugated donkey-anti-goat IgG was obtained from Jackson ImmunoResearch/Dianova (Hamburg, Germany). The hybridomas M1/9 producing rat-anti-murine CD45 and 9B5 producing rat-anti-human CD44 (used as control) were purchased from the American Type Culture Collection (Rockville, MD). Culture supernatants were used for histochemistry. PTX was obtained from Sigma (Deisenhofen, Germany). Mutant PTX (MTX; PT9 K/129G) lacking the enzymatic activity of PTX was used as a control. MTX was kindly provided by Dr. R. Rappuoli (Chiron, Siena, Italy; 38).

4.2 T lymphocyte lines and induction of EAE

Establishment and culture of the CD4+ MHC class II-restricted proteolipid protein (PLP)-specific T cell lines derived from SJL/N mice and induction of active EAE (aEAE) and T cell tEAE have all been described in great detail before 37, 39, 40. Gi-protein mediated signals in encephalitogenic T lymphoblasts were blockedby incubation in vitro with PTX or MTX (100 ng/ml) for 2 h at 37°C. Unbound PTX or MTX was thoroughly removed by washing the T lymphoblasts twice before injection. PTX did not influence the surface phenotype of encephalitogenic T cells as determined by FACS analysis (data not shown). Clinical disease was checked daily and scored as followed: 0.5 (limp tail), 1 (hind leg weakness), 2 (hind leg paralysis), 3 (hind leg paraparesis and incontinence).

4.3 In situhybridization

In situ hybridization was essentially performed as described 41. Briefly, single-stranded 35S-labeled antisense or sense RNA probes were generated by in vitro transcription using T3 or T7 RNA polymerases as described by the manufacturer (Stratagene, La Jolla, CA). Probes for in situ hybridization were prepared from IMAGE Consortium (LLNL) cDNA clones 42; respective IMAGE Consortium clone IDs and restriction enzymes are shown in Table 1. Full-length cDNA clones of CXCL10, CXCL12 and CCL2 were kindly provided by H. Augustin (Freiburg, Germany). All cDNA clones were subcloned into pBluescript II KS+ (Stratagene, La Jolla, CA). A cDNA clone for VEGF-R2 (flk) was kindly provided by G. Breier (Bad Nauheim, Germany) 43. Expression of each chemokine mRNA was investigated in sections of brain and spinal cord derived from least two different SJL/N mice afflicted withEAE (days 14–17 post immunization, clinical score 2–3) and from at least two different healthy age-matched control mice. Specificity of the hybridization signals was verified using sense probes on serial control sections, which never showed any hybridization signal. As a positive control, tissue in which expression of the particular chemokine had been previously demonstrated was included. After hybridization, slides were coated with photographic emulsion (Kodak NTB-2) and exposed for 4 weeks. After fixation, sections were counterstained with toluidine blue, dehydrated, mounted and analyzed with dark and bright field microscopy.

4.4 Immunohistochemistry

Immunohistochemistry was performed on frozen brain sections exactly as described 40 using a three-step immunoperoxidase technique for goat IgG (Vector, Linaris Biologische Produkte, Wertheim, Germany). Each batch of anti-chemokine antibodies needed to be carefully tested for optimal concentration in immunostaining and antibodies were used from 10 to 20 μg/ml.

4.5 Cytospin

Cytospins of T lymphoblasts were performed in a Cytospin 3 (Thermo Shandon, PA). PLP-specific T lymphoblasts (2×105) were suspended in 50 μl RPMI 1640/20%FCS, placed in one cytocentrifuge assembly funnel and centrifuged at 700 rpm for 5 min exactly, according to the manufacturers protocol. Slides were air dried over night, fixed with acetone and stained according to theimmunohistochemistry protocol as described above.

4.6 Chemotaxis assay

Chemotaxis assays were performed slightly modified according to 44. Freshly isolated lymphocytes and encephalitogenic T cells were suspended in migration assay medium (DMEM 4,500 mg/l glucose, 5% calf serum, 2 mM L-glutamine, 25 mM Hepes) at 106 cells/ml. Transwell culture inserts (diameter 6.5 mm; pore size 5 μm; Costar, Cambridge, MA) were coated with 10 μg/ml fibronectin (Roche, Basel, Switzerland) for 1 h in 7% CO2 at 37°C and dried for 2 h at room temperature. Chemokines were diluted in migration assay medium and added in triplicates to the lower chamber in a final volume of 600 μl. Cells (105) were added to the upper chamber in a volume of 100 μl and chemotaxis assays were run for 2 h in 7% CO2 at 37°C. Cells were analyzed under the microscope, collected from the lower chamber and cell numbers were determined using a CASY TT cell counter (Schärfe System, Reutlingen, Germany).

4.7 Frozen section adhesion assay

Frozen section assays using a modified version of the original Stamper-Woodruff protocol 45 were performed as described by us before 9. Briefly, brain sections from mice afflicted with EAE were freshly cut, air dried and incubated with 1×106 encephalitogenic T lymphocytes per section for 30 min at 20°C under shear. Desensitization of CCR7 and CXCR3 on encephalitogenic T cells was achieved by preincubation of T cells with a cocktail of CCL19 and CCL21 (10 μg/ml each) for 30 min at 37°C followed by extensive washing as described46. Brain sections were preincubated with a cocktail of blocking antibodies against CCL19 and CCL21 (50 μg/ml each) or normal goat IgG (100 μg/ml) as a control. Sections were analyzed by counting the number of T lymphocytes specifically bound to cerebral vessels. In total 96 vessels in 20 brain sections in four individual assays using two different encephalitogenic T cell lines were analyzed.

4.8 Statistics

Quantitative data are given as mean values ± SD and were calculated by unpaired Student's t-test using INSTAT software for Macintosh. Results with p<0.05 were considered significant.

Acknowledgements

Expert technical assistance of Martina Schulz, Irene Küchenmeister, Veronika Schmidt and Monika Rieschel is gratefully acknowledged. We thank Friedemann Kiefer for discussion of the manuscript. Special thanks go to Dietmar Vestweber for discussion of the manuscript and his continuous support. Part of this work has been funded by Astra-Zeneca, Sweden.

Footnotes

  1. 1

    WILEY-VCH

  2. 2

    WILEY-VCH

  3. 3

    WILEY-VCH

  4. 4

    WILEY-VCH

  5. 5

    WILEY-VCH

  6. 6

    WILEY-VCH

  7. 7

    WILEY-VCH

Ancillary