Epstein-Barr Virus–induced Molecule 1 Ligand Chemokine Is Expressed by Dendritic Cells in Lymphoid Tissues and Strongly Attracts Naive T Cells and Activated B Cells

BLAST searches of the National Center for Biological Information expressed sequence tag (EST) database using human ELC as a query retrieved contiguous mouse ESTs for ELC. IMAGE Consortium (LLNL) cDNA clones 1088818 and 77575 were obtained from Genome Systems, Inc. (St. Louis, MO) as EcoR1-Not1 inserts in the pT7T3-Pac vector, and sequenced.

For Northern blot analysis, 10–15 μg of total RNA from mouse tissues or purified cells was subjected to gel electrophoresis, transferred to Hybond N+ membranes (Amersham Corp., Arlington Heights, IL), and probed using randomly primed ³²P-labeled mouse ELC EST 1088818, which spans bases 1–755 of the ELC cDNA. To control for loading and RNA integrity, membranes were stripped and reprobed with a mouse elongation factor (EF)-1α probe. For in situ hybridizations, frozen sections (5 μm) from C57BL/6 mice were treated as described (26). In brief, sections were fixed in 4% paraformaldehyde, washed in PBS, prehybridized for 1–3 h, and hybridized overnight at 60°C with sense or antisense digoxigenin-labeled riboprobes in hybridization solution. After washing at high stringency, sections were incubated with sheep antidigoxigenin antibody (Boehringer Mannheim Biochemicals, Indianapolis, IN) followed by alkaline phosphatase–coupled donkey anti– sheep antibody (Jackson Immunoresearch Labs, West Grove, PA) and developed with nitroblue tetrazolium/5-bromo-4-chloro-3-indoyl phosphate (NBT/BCIP). Immunohistochemistry with anti-CD3, anti-B220, and anti-CD11c (PharMingen, San Diego, CA), anti-DEC205 (kindly provided by R. Steinman, Rockefeller University, NY), and MOMA-1 (kindly provided by G. Kraal, Free University, Amsterdam, The Netherlands) antibodies was as described (37). Reverse transcription (RT)-PCR analysis of RNA from purified cell populations was with an Advantage^TM RT-for-PCR kit (Clontech, Palo Alto, CA) using a mixture of primers specific for ELC (5′ primer, ggtgctaatgatgcggaagac; and 3′ primer, agacacagggctccttctggt) and hypoxanthine-guanine phosphoribosyl transferase (5′ primer, cctgctggattacatcaaagcactg; and 3′ primer, tccaacacttcgtggggtcct).

Sequence encoding amino acids 26–108, which are predicted to encompass mature mouse ELC based on comparison with the signal peptide cleavage site of human ELC (32), was isolated by PCR and subcloned into Nde1-Xho1 sites of the pET23b vector (Novagen, Inc., Madison, WI) in-frame with the COOH-terminal HIS₆-tag. TAP302 cells (kindly provided by Tracy Handel, University of California, Berkeley, CA) were transformed with the vector, the HIS-tagged chemokine was purified using NiNTA agarose (QIAGEN, Inc., Chatsworth, CA) according to the manufacturer's instructions, and elution was with 250 mM imidazole (Fisher Scientific Co., Pittsburgh, PA). Further purification was achieved using a C-18 reverse-phase HPLC column (Vydac, Hesperia, CA) and elution with an acetonitrile gradient. SDS-PAGE of this preparation showed a single band of the expected molecular mass for ELC (∼9 kD) that represented >90% of the total protein. Protein concentration was measured using a protein assay (Bio-Rad Laboratories, Hercules, CA).

B cells were purified by staining non-B cells with anti–CD43-biotin (PharMingen) and depletion using streptavidin-MACS^® beads and a MACS^® (Miltenyi Biotec, Inc., Auburn, CA). T cells were purified by staining B cells, macrophages, and granulocytes with biotinylated antibodies to B220, Mac-1, and Gran-1 (PharMingen) and depletion by MACS^®, except for the RT-PCR analysis where T cells were isolated by positive selection for CD3 (Caltag Laboratories, Inc., South San Francisco, CA) by MACS^®. DCs used as a source of RNA for Northern blotting were purified as described (38). In brief, spleen or LN suspensions were prepared in RPMI containing 10% FCS, 10 mM Hepes, and antibiotics and adjusted with sodium chloride to mouse osmolarity (39) by gently pressing through a 70-μM mesh cell strainer (Fisher Scientific Co.). The suspension was layered on 2 ml of metrizamide (14.5% by weight in the same medium) solution (Accurate Chemical and Science Corp., Westbury, NY) and spun at 600 g for 10 min, and the cells at and above the interface were isolated. Flow cytometry confirmed that ∼70% of the cells were CD11c+ DCs. For RT-PCR analysis, in addition to isolation from cell suspensions as above, DCs were purified from the tissue fragments that did not go into suspension by digesting in collagenase D (Boehringer Mannheim Biochemicals) at 1.6 mg/ml and bovine pancreas DNAse (Sigma Chemical Co.) at 100 μg/ml for 30 min at room temperature. EDTA was then added to a concentration of 10 mM for 5 min (40), and the digested preparation was mashed through a cell strainer. Cells were then stained with antibodies to B220, TcR, and CD11c (PharMingen) and purified to >93% CD11c+B220−TcR− by sorting on a FACStar plus^® (Becton Dickinson, San Jose, CA).

Cell suspensions were prepared from spleen, LNs, thymus, or bone marrow of C57BL/6 mice as described (8). In the case of spleen and bone marrow, red blood cells were lysed using Tris-ammonium chloride. Chemotaxis assays were performed as described (23) using 10⁶ total cells per 5-μM transwell (Corning Costar Corp., Acton, MA). Cells that migrated to the lower chamber were enumerated by collecting events for a fixed time (60 s) on a FACScan^®. By counting a 1:5 dilution of input cells in the same way, the absolute number of cells that transmigrated could be determined. To determine which subsets of cells were migrating, a fraction of the cells that migrated to the lower chamber was stained and analyzed by flow cytometry. Cells from resting and activated spleen were stained with B220-FITC (clone 6B2), CD4-PE and CD8-PE (Caltag Laboratories, Inc.), and L-selectin–biotin/streptavidin cychrome (PharMingen). Migrating thymocytes were stained with CD8-FITC, CD4-PE (Caltag Laboratories, Inc.), and CD3-biotin (PharMingen), and bone marrow cells were stained with IgD^b-FITC (PharMingen), B220-PE, and IgM-biotin (Caltag Laboratories, Inc.). Granulocytes were identified by their characteristic large side scatter. For macrophage chemotaxis assays, spleen cells were first depleted of B220+, CD4+, and CD8+ cells by MACS^® (Miltenyi Biotec, Inc.), and migrating cells were identified with Mac-1–PE (Caltag Laboratories, Inc.). In some experiments, cells were preactivated by incubation with phorbol-12,13-dibutyrate (PDBu, 100 ng/ml; Calbiochem Corp., San Diego, CA), 17 μg/ml polyclonal anti–mouse IgM (Jackson ImmunoResearch Labs), or 20 μg/ml LPS. Synthetic human SDF1α (N33A) synthesized by chemical ligation (a gift from M. Siani, Gryphon Sciences, South San Francisco, CA) was used as a control. This SDF1α preparation maximally attracts lymphocytes, macrophages, and granulocytes at 100–500 ng/ml (26) and therefore was used at 300 ng/ml in these studies.

Mouse EBI-1 (41) was isolated by RT-PCR using a 3′ primer downstream of the stop codon and a 5′ primer starting at amino acid 26 of the full-length protein, truncating the hydrophobic NH₂-terminal (possible leader) peptide. Both primers contained Sal-1 sites, and the PCR product was cloned into a modified form of pREP4 (Invitrogen Corp., San Diego, CA) that contains an NH₂-terminal prolactin leader sequence and FLAG epitope followed by an in-frame Sal-1 site (42) (provided by S. Coughlin, University of California, San Francisco, San Francisco, CA). Human embryonic kidney 293 cells were transiently transfected with this construct using LipoTAXI according to the manufacturer's instructions (Stratagene Inc., La Jolla, CA). EBI-1 surface expression 2 d after transfection was confirmed by flow cytometry with anti-FLAG antibody M2 (VWR Scientific Products, West Chester, PA). Calcium mobilization studies were performed on Indo-1–loaded cells as described (43) using a spectrometer (model 4500; Hitachi Ltd., Tokyo, Japan). Intracellular calcium concentrations were calculated using the Hitachi 4500 intracellular cation measurement program.

To permit the expression pattern and chemotactic activity of mouse ELC to be studied, we searched for ESTs that may encode mouse ELC using human ELC as the query sequence (32). Seven mouse ESTs were identified in the GenBank EST database (EST nos. 1088818, 906750, 885418, 1183472, 775758, 807269, and 822508) that could be aligned into a 755-bp contig encoding a predicted protein of 108 amino acids. The high nucleotide sequence identity (84%) and amino acid identity (78%) of this sequence with human ELC (Fig. 1) strongly suggested that we had identified mouse ELC. The proteins with greatest homology to both mouse and human ELC are mouse and human SLC (Fig. 1). A major difference between ELC and SLC is the extended cysteine-containing COOH-terminal region unique to SLC. When this region is excluded from the comparison, ELC and SLC are 33% identical. The next closest match to ELC is MIP3α (33), with 23% identity. The similarity between mouse and human ELC and SLC, together with the observation that the human genes colocalize on chromosome 9 (16, 32), is consistent with the possibility that these molecules arose from a gene-duplication event, and suggests they may have related functions.

Northern blotting analysis revealed that mouse ELC was constitutively expressed at high levels in both mesenteric and peripheral LNs (Fig. 2). Significant expression was also identified in spleen and Peyer's patches, and low expression was seen in the thymus (Fig. 2). No expression was detected in bone marrow or in several nonlymphoid tissues (Fig. 2). To determine whether ELC was produced by a restricted set of cells within lymphoid tissues, which might suggest a role in controlling cell positioning, in situ hybridization analysis was performed (Fig. 3). Strikingly, the ELC probe hybridized strongly to a subset of cells within the T cell zone (or periarteriolar lymphatic sheath) of spleen but not to cells in the B cell follicles or the red pulp (Fig. 3,A). Within LNs, ELC-expressing cells were distributed evenly throughout the T zone (or paracortex) and were absent from follicles (Fig. 3,B). Similarly, in Peyer's patches, a hybridization signal was detected in the T cell region (the interfollicular zone) but not within follicles or in the subepithelial dome region (Fig. 3,C). No hybridization signal to T zones of secondary lymphoid tissues was seen with a control (sense) probe (Fig. 3 I).

The distribution of ELC-expressing cells in spleen, LNs, and Peyer's patches appeared similar to the distribution of T zone DCs in these tissues (3). To further investigate this similarity, spleen sections serial to those used for hybridization were stained to detect T cells (Fig 3. D, red) or with antibodies to CD11c (44) to detect DCs (Fig. 3,E, red). These sections were also stained with antibody to the marginal metallophilic macrophage marker, MOMA-1 (45), to outline the white pulp cords (Fig. 3, D and E, brown). The distribution of ELC-positive cells (Fig. 3,A) was very similar to the distribution of CD11c+ DCs located in the T zone (Fig. 3, A, D, and E). Interestingly, although clusters of CD11c+ DCs were also located in the bridging channels between T zone and red pulp (Fig. 3,E) as observed previously (3, 44), little ELC hybridization signal was detected in these regions (Fig. 3, A and E). Staining of a further serial section with the DC marker DEC205 that is expressed most highly on DCs in the T zone (3, 46) revealed a close concordance between the ELC hybridization pattern and the distribution of DEC205-expressing cells (Fig. 3, A and F). Combined staining for the fixation-resistant marker B220 and for ELC confirmed that the ELC-expressing cells were distributed across the T cell zone but did not extend into the B cell area of the splenic white pulp (Fig. 3,G). Close examination of the ELC hybridization signal in the T zone showed that many of the cells had a striking dendritic morphology (Fig. 3 H).

To directly test whether DCs were a source of ELC, Northern blot analysis was performed with RNA from partially purified spleen and LN DCs (Fig. 4,A). ELC expression was detected in the DC preparations, whereas purified B cells, T cells, and macrophages did not express detectable amounts of ELC (Fig. 4,A). To rule out the possibility that the Northern blot signal derived from a small population of contaminating cells, CD11c+ splenic DCs were FACS^® sorted, and ELC expression was measured by RT-PCR. In agreement with the Northern blot result, ELC expression was detected in the purified DCs but was undetectable in purified T cells, B cells, or macrophages (Fig. 4,B). It was anticipated that if DCs were the major ELC-expressing cells in the spleen, the PCR signal in the sorted DCs would be increased compared with the unsorted spleen preparation, and this is what was observed (Fig. 4 B). Taking the findings from the in situ hybridization, Northern blot, and RT-PCR analysis together, we conclude that DCs present within the T zone are the major source of ELC in secondary lymphoid tissues.

To characterize the chemotactic properties of mouse ELC, a COOH-terminal HIS-tagged version was expressed in Escherichia coli and purified to >90% by affinity- and high-pressure liquid chromatography. In migration assays with mouse spleen cells, a strong response was identified in CD4 and CD8 T cells (Fig. 5,A). Interestingly, CD4 cells were reproducibly found to be more sensitive to low ELC concentrations than CD8 cells (Fig. 5,A). B cells showed a weaker chemotactic response to ELC than both CD4 and CD8 T cells (Fig. 5,A). Strikingly, CD4 T cells of a naive (L-selectin^hi) phenotype showed a stronger migratory response than CD4 T cells of a postactivation or memory (L-selectin^lo) phenotype (Fig. 5 B). Lymphocytes from blood and LNs showed a similar response as cells from the spleen (data not shown).

The ELC receptor was identified originally as a molecule upregulated in EBV-transformed B cells (34). Other studies indicated that CCR7 is also upregulated by treatment of PBLs with phorbol ester, PHA, or anti-CD3 (35, 36). In preliminary experiments with cells cultured overnight in phorbol ester, we observed an increase in the magnitude of the B cell response to ELC (data not shown). This experiment encouraged us to test whether more physiological stimuli also increased B cell responsiveness. This was of particular interest because after engagement by antigen in vivo, B cells migrate rapidly into the outer T zone (7, 8). Cells activated through the B cell receptor by pretreatment with anti-IgM for 4–6 h showed a 1.5-fold increase in the magnitude of their response to ELC (Fig. 5, C and D). Similar findings were made using purified B cells (Fig. 5,E), confirming that this effect was B cell intrinsic. Activation of B cells in vivo by LPS treatment also promotes migration of some cells into the T zone (47), and LPS activation was also found to enhance the B cell chemotactic response to ELC (Fig. 5,E). The increased ELC responsiveness of activated B cells did not simply reflect an overall increase in B cell motility, as there was little difference in the frequency of activated and resting cells that migrated in the absence of chemokine (Fig. 5, C–E). Furthermore, the activated cells did not demonstrate increased responsiveness to SDF1α (Fig. 5, C–E), indicating that the enhanced response was ELC specific.

Since ELC expression was also detected in the thymus (Fig. 2), the chemotactic response of thymocytes was measured (Fig. 6). Mature single positive thymocytes responded to ELC similarly to mature T cells in the periphery (Fig. 6,A), whereas immature (CD4−CD8− double negative) thymocytes and most CD4/CD8 double positive thymocytes did not show a measurable response (Fig. 6,B). Interestingly, however, when the responding cells were stained with anti-CD3 antibodies, an enrichment of CD3-bright CD4/CD8 double positive cells was detected in the population that migrated in response to ELC (Fig. 6,B), indicating that the most mature fraction of double positive cells had acquired responsiveness to ELC. We also measured the ELC responsiveness of immature B lineage cells in the bone marrow. B220+IgM−IgD− pre-B cells and B220+IgM+IgD− immature B cells showed a similar dose-dependent migratory response towards ELC as mature B cells (Fig. 6,C), indicating that the cells acquire responsiveness to this chemokine early in development. Consistent with the failure to detect CCR7 expression in myeloid cells (34, 41), bone marrow granulocytes (Fig. 6,D) and spleen monocytes/macrophages (Fig. 6 E) did not respond to ELC.

The migration response of lymphocytes to mouse ELC was chemotactic since the cells failed to migrate when incubated in ELC in the absence of a gradient (Fig. 5,F), and the pertussis toxin sensitivity of the response (Fig. 5,F) indicated the expected involvement of a Gi-coupled receptor. To test whether the mouse homologue of human ELC we had identified was a ligand for mouse CCR7/EBI-1, 293 cells were transiently transfected with a FLAG epitope– tagged form of the receptor and assayed for mouse ELC responsiveness. CCR7-expressing cells (Fig. 7,A) showed a rapid rise in intracellular calcium in response to ELC but not to two other CC chemokines, MIP1α and MCP1 (Fig 7,B). The response of the transfected cells was dose sensitive (Fig. 7,C), and exposure to high dose ELC fully desensitized the cells to further ELC exposure (Fig. 7,B). ELC did not stimulate a calcium flux in 293 cells transiently transfected with BLR1, showing that the response of CCR7-transfected cells to ELC was specific (Fig. 7 D). These results establish that mouse ELC can stimulate cells through mouse CCR7/EBI-1.

In this report, we have identified mouse ELC and demonstrated that this CCR7 ligand is highly expressed by interdigitating DCs in the T zone of secondary lymphoid tissues. ELC strongly attracts resting T cells and more weakly attracts B cells. However, acute activation increases B cell responsiveness to ELC. These findings suggest two major roles for ELC: (a) to attract naive T cells into the T zone of lymphoid tissues and promote T cell–DC encounters, and (b) to attract antigen-binding B cells into the T zone of lymphoid tissues.

T and B lymphocytes enter LNs and Peyer's patches by crossing HEV located predominantly in the T zone (6, 48). Lymphocyte attachment to HEV involves chemokine- induced activation of integrin-mediated adhesion (49). The physiologically relevant chemokine on HEV that mediates integrin activation has not been defined, but recent studies demonstrated that SLC, SDF, and ELC could each promote integrin-mediated arrest of human lymphocytes in in vitro rolling chambers (17, 50). The high expression of SLC by HEV cells (17) together with our failure to detect ELC expression by HEV (Fig. 3) suggests that SLC may play the more prominent role in triggering lymphocyte adhesion to HEV. Whether SDF is expressed by HEV has not yet been well characterized, and its importance at this site remains to be established. Once lymphocytes have attached to HEV, they then migrate rapidly into the T zone. Our finding that ELC is highly expressed by T zone DCs and that ELC attracts naive lymphocytes suggests that this chemokine plays an important role in attracting cells away from HEV into the T zone and into contact with DCs. Interestingly, ELC and SLC have a strikingly similar chemotaxis profile, since both molecules attract naive T cells more efficiently than memory cells, and attract B cells more weakly than T cells. In this regard, it is interesting to note that SLC is the closest homologue of ELC (Fig. 1) and that both chemokines colocalize on chromosome 9 (16, 32), whereas most other CC chemokines cluster on chromosome 17 (11). We are currently investigating the possibility that SLC is also a ligand for CCR7/EBI-1. Since SLC is expressed at highest levels by HEV and at lower levels by stromal cells in the T zone (17), this chemokine may be inefficient at attracting cells into the T zone once they have crossed HEV. The existence of DC-derived ELC gradients may facilitate their movement away from the HEV and in amongst the T zone DCs. A further chemokine, PARC/ DCCK1, that is expressed by DCs in germinal centers and possibly also T zone DCs, and that can attract naive T cells (24, 25), may work with ELC to promote naive T cell–DC encounters. However, in contrast to the concentrated expression of ELC in secondary lymphoid tissues in humans (32) and mice (Figs. 2 and 3), PARC/DCCK1 is expressed at highest levels in the lung (25), suggesting distinct roles for the two molecules.

Lymphocytes enter the spleen by a different route to LNs or Peyer's patches, being released in the marginal sinus that surrounds the white pulp, or at terminal arterioles in the red pulp, and then migrating into the T zone (5, 52). However, as with entry into LNs, entry into the lymphoid region of the spleen is pertussis toxin sensitive (53, 54), implicating the involvement of chemokines. The expression of ELC in DCs within the splenic T zone indicates that ELC is likely to play a similar role in spleen and LNs, attracting lymphocytes into the T zone and promoting T cell encounter with DCs. Interestingly, despite the absence of HEV in the spleen, SLC expression is also detected in stromal cells in the splenic T zone (17), suggesting that these two chemokines may again act in concert to attract cells into the T zone.

The responsiveness of single positive thymocytes to ELC is likely to be important in ensuring that when these cells leave the thymus, they can enter into the T zone of secondary lymphoid tissues and interact with DCs. Whether ELC plays a role in thymocyte movement within the thymus is less clear. The expression level of ELC in mouse thymus was low, and we were unable to obtain a reproducible in situ hybridization signal (data not shown). However, the acquisition of ELC responsiveness by mouse thymocytes at the CD3^hi, CD4/CD8 double positive stage suggests ELC may have a role in cell positioning within the mouse thymus. In this regard, ELC may function with several other chemokines identified recently in this tissue (20, 55–59), to help compartmentalize the cells appropriately for each developmental/selection event. Interestingly, ELC appears to be expressed at much higher levels in human thymus (32, 33), suggesting a more prominent role for ELC in human T cell development.

The weak but significant responsiveness of resting B cells to ELC is consistent with the original finding of EBI-1/ CCR7 expression in B lymphocytes (34). Previous studies have shown SLC also weakly attracts resting B cells (17), and it seems likely that ELC and SLC work together to promote migration of B cells into the T zone after they have crossed HEV or entered the marginal sinus of the spleen. Once in the T zone, the B cells may then enter the gradient of BLC that attracts the cells into follicles (26). Interestingly, immature B cells which do not express BLR1 (60, 61) are able to migrate into the T zone of the spleen but fail to enter follicles (62, 63). Positioning within the T zone may play an important role in the maturation or selection of these cells (63). Since immature B cells in the bone marrow are responsive to ELC (Fig. 6 C), it is reasonable to propose that once these cells leave the marrow and enter peripheral lymphoid tissues (especially the spleen), they are able to migrate into the T zone along a gradient of ELC.

Upon binding antigen, the migration of mature recirculating B cells is altered dramatically. Rather than moving into follicles, the cells accumulate within 6–8 h in the outer T zone (7, 37). This repositioning is likely to be critical for antigen-specific B and T cells to find each other. One mechanism by which antigen-engaged B cells might relocate to the outer T zone is through increased responsiveness to T zone–expressed chemokines. Our finding that stimulating cells through the antigen receptor increased the B cell response to ELC raises the possibility that ELC may participate in recruiting antigen-engaged B cells to this site.

In summary, our findings suggest a central role for ELC in secondary lymphoid tissue function. The high expression by T zone DCs and the strong chemotactic activity for naive T cells suggest that ELC attracts T cells into the T zone and promotes T cell–DC encounters. The weaker responsiveness of both immature and mature B cells supports a role for ELC in attracting these cells into the T zone as the necessary first step before they migrate into follicles, whereas the enhanced response of activated B cells implicates ELC as an important cue that promotes localization of antigen-engaged cells in the T cell zone. Understanding the factors that regulate ELC expression by DCs and the signals that control ELC responsiveness of lymphocytes is likely to teach us much about how secondary lymphoid tissues function in immunity.

We thank Elijah Wasson and Dana Avrahami-Tzfati for help in preparation of recombinant ELC, Mike Siani for the generous gift of SDF1α, Chris Turck for HPLC, and Art Weiss for use of the Hitachi 4500 spectrometer. We also thank M. Dee Gunn, Rich Locksley, and members of the laboratory for comments on the manuscript. J.G. Cyster is a Pew Scholar in the Biomedical Sciences.

This work was supported in part by a grant from the National Institutes of Health (AI40098).

BLC

B lymphocyte chemoattractant

BLR

Burkitt's lymphoma receptor

CCR

CC chemokine receptor

DC

dendritic cell

DCCK

dendritic cell–derived CC chemokine

EBI-1

EBV-induced molecule 1

ELC

EBI-1 ligand chemokine

EF

elongation factor

EST

expressed sequence tag

HEV

high endothelial venules

MCP

monocyte chemoattractant protein

MIP

macrophage-inflammatory protein

PARC

pulmonary and activation–regulated chemokine

RT

reverse transcription

SDF

stromal cell–derived factor

SLC

secondary lymphoid tissue chemokine