How to drain without lymphatics? Dendritic cells migrate from the cerebrospinal fluid to the B-cell follicles of cervical lymph nodes

Under normal conditions, the transport of immune cells from blood to the central nervous system (CNS) is restricted by 2 physical barriers: the blood-brain barrier formed by CNS parenchymal microvessels and the blood cerebrospinal fluid (CSF) barrier formed by the choroid plexuses. Also, the circulation of immune cells from brain to lymphoid organs is hampered by the lack of CNS-draining lymphatic vessels. Nevertheless, immune responses may develop in the CNS, and cervical lymph nodes are considered as major sites of antigen presentation during neuroinflammatory diseases.^1,2  Indeed, antigens are drained from the CNS to cervical lymph nodes along the axons of craniofacial peripheral nerves.^3,4  Also, it was reported that dendritic cells (DCs) are able to migrate out of the CNS and, in turn, to elicit a CNS-targeted immune response.^5,6  However, it is not clear whether DCs circulating out of the CNS actually migrate from brain parenchyma or from the CSF compartment. This point is of importance because DCs are absent from normal CNS parenchyma,⁷  but they can be detected in CSF and in compartments associated with CSF circulation or production, including meninges and choroid plexuses.^8-10  Moreover, under neuroinflammatory conditions, DCs accumulate in the CSF^11,12  as well as in perivascular spaces,^13,14  anatomic compartments draining into the CSF. These findings, along with others, suggest that the CSF may be a major transport route for DCs circulating in the CNS and migrating either from CSF to CNS parenchyma or from CSF to the lymphoid organs.^11,12,15,16 

In the present study, we tracked bone marrow-derived myeloid DCs injected stereotaxically into the CSF or brain parenchyma of rats under normal conditions.

Animal care and procedures were conducted according to the guidelines approved by the French Ethical Committee (decree 87-848) and the European Community directive 86-609-EEC and meet the Neuroscience Society guidelines. The study protocol was approved by the ethical committee of Faculté de Médecine Laennec, Lyon, France. Eight- to 10-week-old female Sprague Dawley rats were obtained from Harlan (Gannat, France).

Murine GM-CSF, human Flt3-L, murine IL-4, and human TGF-β were obtained from PeproTech (Tebu). Mouse monoclonal antibodies recognizing rat MHC class II molecules (OX6 antibody), CD11b/CD11c (OX42 antibody), αE2 integrin or CD103 (OX62 antibody), CD80 (B7-1, clone 3H5), CD86 (B7-2, clone 24F) or CD54 (ICAM-1, clone 1A29) were purchased from Becton Dickinson Biosciences (Meylan, France). Mouse monoclonal antibody recognizing CD11c (clone 8A2) was purchased from Serotec (Oxford, United Kingdom). For flow cytometry experiments, FITC-labeled rat-adsorbed goat anti-mouse antibody (Serotec) was used as a secondary antibody and mouse anti-human CD3 antibody (Beckman Coulter, Marseille, France) was used as a control primary antibody. For immunocytochemistry, a fluorescein-conjugated goat anti-mouse antibody (Alexa Fluor 488; Molecular Probes, Leiden, The Netherlands) was used as a secondary antibody.

Female Sprague Dawley, Dark Agouti, or Lewis rats were killed, and bone marrow was flushed from femurs and tibias using 10 mL DMEM in a 10-mL syringe with a 26-gauge needle. Bone marrow cells were then resuspended and passed trough a cell strainer (70-μm pore). After 1 wash in phosphate-buffered saline (PBS), cells were resuspended in 10% DMSO, 20% FCS and stored in liquid nitrogen until use. When needed, frozen vials of cells (10-15 × 10⁶/vial) were thawed, cells were washed once in DMEM (Gibco) then cultured in 25 cm² culture plates at a density of 10⁶ cells/mL in IMDM (Gibco, Karlsruhe, Germany) supplemented with 15% FCS (Fetal Clone II; Perbio Science, Bonn, Germany) and antibiotics (penicillin/streptomycin; Invitrogen, Cergy Pontoise, France). Myeloid rat dendritic cells were then generated as previously described,^17,18  with slight modifications. Briefly, bone marrow cultures were grown for 7 days at 37°C in 5% CO₂ in the presence of murine GM-CSF (10 ng/mL) and human Flt-3 ligand (10 ng/mL). By the end of this period, clusters of nonadherent cells had formed that were removed, dispersed, and replated in fresh media consisting of DMEM supplemented with 10% FCS (BioWest, Nuaillé, France), penicillin/streptomycin (Invitrogen), murine GM-CSF (10 ng/mL), and murine IL-4 (10 ng/mL). After 3 days, large numbers of free-floating cells harboring irregular cell surfaces could be observed, as well as a small population of plastic-adherent macrophages and stromal cells. In some experiments, 1-μm diameter fluorescent microspheres (Molecular Probes) at a dilution of 0.01% solid were added to the culture medium for an additional 24-hour period. Following this culture procedure, free-floating cells were harvested, washed once in PBS, and used for cytologic examination, fluorescence-activated cell sorting (FACS) analysis, or in vivo experiments.

After fixing in acetone, cells were rinsed 3 times in PBS then incubated for 30 minutes at room temperature with a blocking solution containing 4% bovine serum albumin and 10% normal goat serum. Cells were then incubated overnight at 4°C with mouse monoclonal antibody OX42, OX6, anti-CD80, anti-CD86, or anti-CD54 diluted 1:100 to 1:400 in blocking solution. After several washes in PBS, cells were incubated for 50 minutes in blocking solution containing a fluorescein-conjugated goat anti-mouse antibody (dilution 1:100) then rinsed in PBS and mounted using an aqueous preparation (Fluoroprep; BioMérieux, Marcy l'Etoile, France). In experiments in which fluorescent microspheres had been added to the culture medium, harvested cells were fixed in 4% paraformaldehyde before being cytospun and processed for immunocytologic analysis as described above in this paragraph. In these cases, confocal laser scanning microscopy (LSM META Zeiss; Carl Zeiss, Jena, Germany) was performed to discriminate between internalization of particles and attachment to the cell membrane. Otherwise, images were recorded and analyzed by a computer-assisted system consisting of a specific image analysis software (analySIS auto; Soft Imaging System, Münster, Germany).

In each experiment, 0.2 × 10⁶ to 0.5 × 10⁶ cells were incubated on ice for 30 minutes with mouse OX6 antibody (anti-MHC class II molecules), OX62 antibody (anti-integrin αE2 or CD103), anti-CD11c, anti-CD80, anti-CD86, or control mouse anti-human CD3 antibody diluted 1:50 in PBS containing 2% FCS. After one wash in PBS-2%FCS, cells were then stained with FITC-conjugated rat-adsorbed goat anti-mouse IgG, washed twice in PBS, measured in a EPICS XL flow cytometer (Beckman Coulter), using CellQuest software (Becton Dickinson) for analysis.

Bone marrow-derived DCs were fixed for 30 minutes in 2% glutaraldehyde-0.1 M NaCacodylate pH 7.4. They were then washed 3 times in 0.1 M Nacacodylate/sucrose, pH 7.4, for 15 minutes and fixed afterward with 1% OsO₄-0.15 M NaCacodylate pH 7.4 for 30 minutes. After dehydration in an ascending gradient of ethanol, 5 minutes for each step, 30%, 50%, 70%, and 95%, impregnation steps and embedding were performed in Epon, finally polymerized at 60°C for 48 hours. Sixty to 80-nm sections were obtained using an ultramicrotome RMC-MTX (Research. Manufacturing Company, Tucson, AZ), and contrasted with uranyl acetate and lead citrate. Observations were made on a JEOL 1200EX transmission electron microscope (Jeol, Tokyo, Japan) equipped with a MegaView II high resolution transmission electron microscope (TEM) camera and an Analysis Soft Imaging system (Eloïse SARL, Roissy, France).

For T-cell preparations, cell suspensions obtained from the cervical lymph nodes of Lewis rats were passed through a cell strainer (70-μm pore) and negative magnetic selection (Milteny Biotec, Paris, France) was performed using OX6 (anti-MHC class II molecules), OX33 (CD45RA expressed on B cells), or OX42 (anti-CD11b/c antibody) primary antibodies. Bone marrow-derived DCs, obtained from Sprague Dawley rats as described in “Generation of rat bone marrow-derived DCs,” were pulsed for 24 hours with LPS (100 ng/mL; Sigma-Aldrich, Deisenhofen, Germany) then cocultured in graded doses with 2 × 10⁵ T cells in a 96-well round-bottom plate. Cells were plated in triplicate in a total volume of 200 μL/well with DC/T cell ratios from 1:20 to 1:160. After 72-hour culture, cells were pulsed with 2 μCi/well (0.074 MBq/well) ³H-thymidine (Amersham Biosciences, Uppsala, Sweden) for 18 hours then harvested on fiberglass fibers. Incorporated thymidine was quantified in a direct beta counter (Matrix 96; Packard, Groningen, The Netherlands), and results were expressed as the mean counts per minute (cpm) ± SD of triplicate cultures.

For injection experiments, DCs were labeled using CFSE (carboxyfluorescein diacetate succinimidyl ester; Molecular Probes) or fluorescent microspheres. Briefly, for CFSE labeling, cells were washed once in PBS and incubated for 5 minutes at 37°C in 1 μM CFSE. Then, 250 μL FCS was added, and cells were further incubated for 5 minutes at 37°C before being washed in PBS and resuspended at a dilution of 3 × 10⁴ cells/μL in phenol red-free DMEM. Alternatively, cells were incubated for 24 hours with fluorescent microspheres as described in “Generation of rat bone marrow-derived DCs,” and then rinsed twice in PBS and resuspended at a dilution of 3 × 10⁴ cells/μL in phenol red-free DMEM.

Stereotaxic injections of labeled DCs were performed in 21 female Sprague-Dawley normal rats. All injections were performed using phenol red-free DMEM as a vehicle. For intra-CSF injections in normal rats, 3 × 10⁵ DCs loaded with fluorescent microspheres (n = 5) or labeled with CFSE (n = 6) were diluted in 10 μL vehicle and injected in the left lateral ventricle. Briefly, each rat was deeply anesthetized by pentobarbital injection and placed in a stereotaxic frame, and its head was tilted slightly by raising the tooth bar to 5 mm. Solution (10 μL) was then slowly injected in the left lateral ventricle (stereotaxic coordinates: 1.4 mm lateral to the bregma and 4.5 mm down from the surface of the skull) over a period of 3 minutes, using a Hamilton syringe. The animal remained in the stereotaxic frame with the needle in place for 1 minute thereafter, and the needle was then slowly removed over a period of 2 minutes. Alternatively, DCs loaded with fluorescent microspheres (n = 3) or labeled with CFSE (n = 5) were injected into the corpus callosum (stereotaxic coordinates: 1.4 mm lateral to the bregma and 3.7 mm down from the surface of the skull), otherwise following the same protocol. Following the intra-CSF injections of DCs, rats were killed on day 1 (n = 2), 3 (n = 5), or 8 (n = 4) after injection. Following the intraparenchymal injections of DCs, rats were killed on day 3 (n = 4) or 8 (n = 4) after injection. Control experiments were performed in which rats received an intra-CSF injection of vehicle alone (n = 2).

On day 1, 3, or 8 after stereotaxic injection of DCs, animals were anesthetized by halothane inhalation and killed by intracardiac perfusion with 250 mL 4% paraformaldehyde in 100 mM pH 7.4 phosphate buffer. Then brains, cervical lymph nodes, and axillary lymph nodes were dissected out, immersed overnight in fixative at 4°C, and kept in PBS containing 30% sucrose at 4°C until use. When needed, tissues were then frozen in dry ice, and the blocks were embedded in polyethylene glycol and cut in 14-μm thick sections with a cryostat.

Rat myeloid DCs were generated, as described in “Materials and methods,” by the sequential treatment of whole bone marrow cultures with Flt-3 ligand + GM-CSF for 7 days, then GM-CSF + IL-4 for 3 days. Free-floating cells harvested after 10 days of culture displayed a round irregular morphology (Figure 1A) or harbored dendrites of various lengths (Figure 1E). Immunocytoflurorescence analysis showed that all cells exhibited strong MHC class II staining (Figures 1B,F) with 50% to 60% of them showing MHC class II⁺ dendrites (Figure 1F). Analysis by transmission electron microscopy allowed us to distinguish round irregular cells with a rich endosomal compartment (Figure 1C) from process-bearing cells with a less-developed endosomal compartment (Figure 1G). When performing a phagocytic assay with fluorescent microspheres, we found that (1) greater than 95% of cells ingested substantial amounts of microspheres while expressing MHC class II molecules (Figures 1D,H) and (2) cells showing the more immature DC morphology (ie, a round irregular cell body without dendrites) displayed the highest phagocytic activity (Figure 1D). Finally, by FACS analysis, we observed that cells uniformly expressed the rat dendritic cell markers CD11c and OX62,^17,19,20  as well as OX42 (CD11b/c) and low-to-intermediate levels of MHC class II molecules (OX6), CD80, CD86, and CD54 (Figure 1I; data not shown). Altogether, these results indicate that cells, generated under our experimental protocol, consisted of immature myeloid DCs. Accordingly, when cells were stimulated with LPS for 24 hours, they all acquired the morphologic, phenotypical, and functional features of fully mature DCs (Figure 1J and Figure S1; see the Supplemental Figure link at the top of the online article, at the Blood website). In particular, LPS-stimulated DCs exhibited dendrites bearing MHC class II molecules (Figure S1), and the levels of membranous OX6 and CD86 antigens, as assessed by FACS analysis, were dramatically up-regulated when compared with unstimulated DCs (Figures 1I,J). Similar findings were observed when analyzing CD80 and CD54 antigens by immunocytofluorescence (data not shown). Finally, confirming their mature phenotype, LPS-stimulated DCs elicited a dose-dependent allogeneic T-cell response (Figure S1).

In a first set of experiments, DCs were incubated with fluorescent microspheres to track them. Normal rats were then injected with labeled DCs (3 × 10⁵ cells) within the left lateral ventricle (n = 4), that is, in the CSF compartment or within the corpus callosum (in brain parenchyma) (n = 4). Animals were killed for CNS histologic analysis on day 1, 3, or 8 after injection. At all time points studied, DCs injected in the left lateral ventricle could be found in the CSF compartment. In particular, groups of labeled cells were frequently found on the apical surface of the choroid plexus homolateral to the injection (Figure 2A-B). Also, numerous cells were detected bilaterally in the recesses of the fourth ventricle, which communicate with the subarachnoid spaces, and in the brain cortical meninges (Figure 2C-D; data not shown). Altogether, these observations suggest that cells had passively followed the CSF flow through the third and fourth ventricle before getting access to the outer surface of the brain, where CSF circulates in the subarachnoid spaces.

Interestingly, some DCs injected in the left lateral ventricle were found in parenchymal locations, including the subventricular zone homolateral to the intraventricular injection and, in particular, the germinal zone, where neural stem cells reside (Figure 2E-F).²¹  Occasional cells were also detected in the brain parenchyma adjacent to the third ventricle as well as in frontal cortical locations, several millimeters away from the site of injection (Figure 2G; data not shown). Labeled cells located within brain parenchyma were detected starting day 3 or 8 after injection, indicating they had first circulated in CSF before infiltrating CNS parenchyma. Thus, our results show that intra-CSF-injected DCs loaded with microspheres are actually able to infiltrate brain parenchyma.

In contrast to the migratory behavior of DCs injected into the left lateral ventricle we observed that DCs injected into the corpus callosum remained mostly confined around the injection site or migrated only a short distance along the adjacent white matter tracts (Figure 2H). To ensure that phagocytized microspheres did not alter the migratory behavior of DCs, a second set of experiments was performed in which CFSE-labeled DCs were injected into the left lateral ventricle (n = 6) or corpus callosum (n = 5) of normal rats (Figure 3). In these experiments, analysis of CNS obtained on day 3 (n = 5) or 8 (n = 6) after injection gave similar results to those observed after injections of DCs loaded with microspheres (Figures 2 and 3). In particular, intra-CSF-injected DCs could be found in the subventricular zone homolateral to the injection site (Figure 3A-B), in the meninges (Figure 3C-D), the brain parenchyma adjacent to the third ventricle (Figure 3E-F) and occasionally in the frontal cortex (data not shown). Also, CFSE-labeled DCs injected within corpus callosum migrated little from their injection site (Figure 3G-H).

Histologic examination was performed on cervical lymph nodes and axillary lymph nodes obtained from animals killed on day 3 (n = 5) or 8 (n = 6) after injections of CFSE-labeled DCs into the left lateral ventricle or corpus callosum (Figure 4). Following injections in the left lateral ventricle, numerous CFSE-labeled cells were found in cervical lymph nodes (Figure 4A-B), whereas only rare labeled cells were detected in axillary lymph nodes (Figure 4C-D). Interestingly, the vast majority of CFSE-labeled cells were detected within B-cell follicles (Figure 4A, left) and, to a lesser extent, in the medulla (Figure 4B, right) of cervical lymph nodes. In accordance with the observed poor mobility of DCs injected within corpus callosum (Figures 2 and 3), no or only occasional labeled cells could be detected in cervical or axillary lymph nodes, following injections of DCs in corpus callosum (Figure 4E-F). To confirm the preferential distribution of intra-CSF-injected DCs within B-cell follicles of cervical lymph nodes, immunohistofluorescence experiments were performed using antibodies directed against CD3, OX33 antigen (CD45RA expressed on B cells), or OX6 antigen (MHC class II molecules). We observed that CFSE-labeled DCs were mainly localized in B-cell follicles as compared with T-cell-rich areas (Figure 5A-B). Moreover, colocalization of CFSE labeling with OX6 staining showed that intra-CSF-injected DCs had maintained the expression of MHC class II molecules in B-cell follicles (Figure 5C).

The prototypic migratory pattern of tissue-resident DCs was initially established by studying dermal DCs and epidermal-residing Langerhans cells.^22-24  In this paradigm, DCs circulating in the interstitial fluid of the skin are drained by lymphatic vessels and reach first the outer surface of lymph nodes before gaining access to T-cell-rich areas. In the skin, as in many other tissues, migration of immature DCs is accompanied by a process of maturation, allowing the acquisition of costimulatory molecules and of chemotactic receptors such as CCR7.^25,26  By the end of such a process, mature DCs localize within T-cell-rich areas of draining lymph nodes and direct the antigen-specific proliferation of T cells that, in turn, may amplify B-cell responses through the release of TH2-type cytokines. Our results suggest that such a functional scheme does not apply to the CNS. The fact that, within cervical lymph nodes, intra-CSF-injected DCs preferentially target the B-cell follicles suggests that, under neuroinflammation, specific mechanisms direct the migration of DCs to this location. Interestingly, the presence of dendritic cells in the germinal centers of lymphoid organs has been previously acknowledged.^27,28  However, the mechanisms driving DC migration to germinal centers are unclear. In this regard, it should be noted that a subset of blood-circulating myeloid DCs has been shown to specifically target splenic B-cell follicles and to stimulate B-cell proliferation.²⁹  On the basis of this finding, one may thus hypothesize that the transport of DCs from the CSF to the B-cell follicles of cervical lymph nodes might occur through blood. Further studies are required to clarify this point. In addition, one has to consider that migration of DCs from the CNS to cervical lymph nodes may be partly conditioned by the maturation state of the DCs. In this case further studies are needed to compare the migratory behavior of mature versus immature DCs when injected into the CNS.

As exogenously delivered DCs may not behave as endogenous antigen-presenting cells (APCs), we performed preliminary experiments in which fluorescent microspheres were injected into the CSF of rats with experimental allergic encephalomyelitis. Results from these preliminary studies suggest that intra-CSF-injected microspheres accumulate in the meninges and in the cervical lymph nodes, where again B-cell zones are targeted. This result further suggests that under inflammatory conditions APCs circulating within the CSF express a peculiar migratory behavior because they seem to target the B-cell follicles of cervical lymph nodes.

Besides information on the migration of DCs from brain to lymphoid organs, our data bring new insights into the traffic of CSF-circulating DCs within the CNS. Thus, the migration of DCs in the CSF compartment reproduces some aspects of the physiologic CSF circulation, from lateral ventricles to the brain cortical meninges.³⁰  However, not all intra-CSF-injected DCs follow the CSF flow as some of them either adhere to the apical surface of the choroid plexuses or infiltrate the germinal zone, where neural stem cells reside. The expression of adhesion molecules by choroid plexus epithelial cells has been reported and may participate in the physical interactions between choroid plexuses and the injected DCs.^31,32  Similarly, the presence of infiltrating DCs within the germinal zone of normal rats suggests that chemotactic factors locally synthesized in the subventricular zone might direct DC migration through the ependymal layer.

CNS-resident cells, including astrocytes and microglia, are thought to shape neuroimmune interactions under normal or inflamed conditions. Our data indicate that, owing to their unique migratory behavior, CSF-circulating DCs may also play a major role in CNS immune responses. In this view, deciphering the molecular mechanisms supporting the targeting of CSF-circulating DCs to B-cell follicles of cervical lymph nodes may allow new therapeutic targets for neuroinflammatory diseases to be identified.

We thank Gaëlle Cavillon for technical assistance in flow cytometry studies and Christine Servet-Delprat and Jean-François Ghersi-Egea for their helpful comments. We also thank Dr Simone Peyrol and the CeCIL (Centre Commun d'Imagerie Laennec) for the use of their electron microscopy facilities. We thank Patricia Hulmes (“allenglish,” Lyon) for critical reading of the manuscript.