CD100 is a human 150-kDa homodimer expressed at the surface of most hemopoietic cells, and its gene belongs to the Ig and semaphorin gene families. Semaphorin genes encode soluble and membrane-bound proteins, most of which have been shown to act as chemorepellents on growth cone guidance. CD100 is discrete, as it is a transmembrane leukocyte surface molecule that can also exist in a soluble form. While our previous studies using mAbs suggested that the transmembrane form of CD100 plays a role in lymphocyte activation, no function was shown for its soluble form. Here, we investigated the effect of soluble CD100 in a cell migration assay; both CD100 spontaneously shed from a stable transfectant and soluble recombinant CD100 inhibited spontaneous and chemokine-induced migration of human monocytes. Interestingly, only the dimeric form of CD100 exerted an effect. Moreover, soluble CD100 inhibited migration of cells from monocytic and B cell lineages. A similar inhibitory effect on migration was observed with H-SemaIII, but not H-SemaIV, semaphorins. In addition, both CD100 and H-SemaIII were recognized by two CD100 mAbs in an ELISA, and one of these mAb abolished the inhibitory effect of each of these semaphorins. We also provide evidence that CD100 and H-SemaIII act through the same receptor on immune cells, which is not neuropilin-1. Furthermore, we describe a function on immune cells for H-SemaIII, a semaphorin to date only studied in the nervous system.
CD100 was initially identified using a pair of mAbs produced in our laboratory (1, 2) as a homodimeric human Ag expressed on most hemopoietic cells, except on immature CD34+ cells, erythrocytes, and eosinophiles (3, 4). Our previous studies showed that CD100 was an activation molecule on lymphocytes. Actually, triggering CD100 with mAb through distinct epitopes led to different signals of costimulation in response to CD3 or CD2 activation of peripheral blood T lymphocytes (2, 5). We also showed that CD100 associates with CD45, a key molecule in T lymphocyte activation. This association has functional consequences, as CD100 mAb increase T cell homotypic adhesion induced by a CD45 mAb (6). We also provided evidence for the interaction of CD100 cytoplasmic domain with a serine-threonine kinase activity in T and NK cells (7, 8). All these results indicated that CD100 plays a role in T lymphocyte activation (for review, see Ref. 9). More recently, we described the association of CD100 with tyrosine phosphatase activities in B lymphocytes (10). Furthermore, CD100 stably transfected cells induce homotypic adhesion of B cells and increase their survival, thus providing evidence for the existence of a counter-receptor for CD100 on B lymphocytes (11).
Molecular characterization of human cDNA for CD100 revealed that it belongs to both the semaphorin and the Ig gene families (11). The Ig domain is known to play a role in cell-cell recognition. The semaphorins are defined by their semaphorin domain, which is 500 aa long, comprising 14–16 conserved cysteins. In the CNS secreted (12, 13, 14) and some transmembrane (15, 16) semaphorins are repulsive molecules. Up until now, functions are unclear for the many other described semaphorins (for review, see Ref. 17). Hence, CD100, which was reported at the same time as the first semaphorin was cloned in the nervous system (13), is still the unique immune semaphorin whose membrane functions have been described. In addition, as reported in the accompanying paper (18), it can be spontaneously shed from the cell surface of T lymphocytes, thus producing a soluble form.
In this report we focused on the function of soluble CD100. To this aim, we tested the effect of either CD100 shed from the surface of a stable transfectant or a chimeric protein containing the extracellular domain of CD100 on the migration of immune cells of different lineages. Interestingly, soluble dimeric CD100 inhibited the spontaneous migration of monocytic and B lymphoid cells, but not that of a T cell clone. Moreover, when comparing with the activities of other semaphorins, H-SemaIII had a similar effect, whereas H-SemaIV did not modify cell motility. The results indicate that CD100 and H-SemaIII act through a receptor different from neuropilin-1, which was shown to be a receptor for H-SemaIII in the nervous system (19, 20). Thus, our data provide evidence for a function of two soluble semaphorins in the immune system. The only previous report of an immune effect concerned the poxvirus semaphorin, which was shown to interact with immune cells via VESPR (21). Moreover, it shows a new example of interaction between immune and neural systems as molecules functionally described to date in the nervous system, such as H-SemaIII, can act on cells of the immune system.
Materials and Methods
Human cell lines, U937, Jurkat, Eskol (provided by Dr. E. F. Srour, University of Indiana, Indianapolis, IN), and B12 thymic T cell clone were grown in RPMI medium supplemented with 2 mM glutamine, 1 mM pyruvate, penicillin-streptomycin, and 10% heat-inactivated FCS. Stably CD100- or mock-transfected 3T3 fibroblasts (provided by G. Freeman, Dana-Farber Cancer Institute, Department of Medicine, Harvard Medical School, Boston, MA) were grown in DMEM supplemented with 2 mM glutamine, 1 mM HEPES, 10% FCS, gentamicin, and hygromycin B and were maintained to subconfluent density.
PBMC were obtained from healthy volunteers, and monocytes were purified using a monocyte isolation kit (Miltenyi Biotech, Germany). Briefly, PBMC isolated by density gradient separation (Lymphoprep, Nycomed, Oslo, Norway) were labeled with a mixture of hapten-conjugated mAbs. After washings, cells were incubated with anti-hapten microbeads and separated on a MACS column. Monocytes were collected from the effluent, and purity, as evaluated by CD14+ staining, was >90% (data not shown).
Semaphorin cDNAs were all introduced in expression plasmids that have previously been described for H-SemaIII-Myc and H-SemaIV-Myc (22, 23, 24). We used a plasmid containing alkaline phosphatase-CD100 extracellular part-Myc tag (25). The product of this construction is referred to as CD100-alkaline phosphatase (AP)4 in the text. We expressed full-length CD100 in COS cells using CD100 in pCDM8 vector (provided by G. Freeman) or CD100 mutated on Cys674 (18).
COS7 cells were cultured at a subconfluence density and transfected using Lipofectamine (Life Technologies) for 5 h. After 3 days of culture supernatants were collected and stored at 4°C. Alkaline phosphatase activity present in the supernatants was assayed as previously described (26).
A sandwich ELISA was developed to detect soluble semaphorins in supernatants. The anti-CD100 mAb used were BB18 and BD16 obtained and purified from our laboratory (1, 2). Briefly, the first mAb was coated in 100 μl of PBS overnight at 4°C in a 96-well plate (Maxisorp, Nunc). All steps were performed at 4°C. Wells were saturated with 1% BSA in PBS for 4 h, and supernatants were then added for 2 h. After extensive washing with PBS/1% BSA, 100 μl of the second biotinylated mAb (diluted in PBS/1% BSA) were added. After washings and incubation with streptavidin coupled to HRP, revelation was performed using ABTS (Roche, Mannheim, Germany) diluted in ABTS buffer. After 1-h exposure in the dark at room temperature, absorbance was measured at 405 nm using a Plate reader (Packard, Downers Grove, IL). To quantify CD100 contained in the supernatants used in the migration assay, we used purified CD100 to perform a dose-response curve. CD100 was purified from supernatant of COS cells expressing CD100-AP. Briefly, CD100 was immunoprecipitated on cyanogen bromide-Sepharose beads cross-linked to BB18 mAb. Then, CD100 was eluted in HCl-glycine, pH 2.8, and reprecipitated with beads coated with anti-Myc mAb (Tebu). CD100 was finally eluted in HCl-glycine, pH 2.8, submitted to dialysis, and concentrated (Centricon, Millipore, Bedford, MA). The concentration was then estimated on a silver-stained gel by comparison with known quantities of BSA.
Assays were performed with the Transwell system (Costar, Cambridge, MA) comprising two chambers (27, 28). Filters were not pretreated, but supernatants and medium used for the assay contained 10% FCS. In the upper chambers, 105 cells diluted in supernatant obtained from transiently or stably transfected cells under a final volume of 100 μl were added. Controls were either supernatant of COS cells transfected with a plasmid encoding for the alkaline phosphatase alone for experiments performed with COS supernatants or supernatant of mock-transfected 3T3 cells for experiments performed with 3T3 CD100 supernatant. Five-micron inserts were used for monocytes and Eskol cells, and 8-μm inserts were used for U937. After a 6-h (for monocytes) or an 18-h (for other cells) incubation at 37°C, cells in suspension in the upper and lower chambers were enumerated through microscopy for each well of the duplicata. For some experiments supernatants were concentrated using Microcon 50 (Millipore). In some assays the supernatant containing semaphorin was added only in the lower chamber or in both chambers, providing the same results as with supernatant added only in the top chamber. Some experiments were also performed after preincubation of the cells for 1 h with supernatant before washing and migration assay only with medium; the same results were obtained. A chemokine active on monocytes, monocyte chemoattractant protein-3 (MCP-3; Tebu), was added in the lower well at 50 ng/ml in some experiments. Finally, the effect of BD16 mAb or control isotype-matched BB27 (anti-CD101) mAb, added in the upper chamber for the entire duration of migration at 10 μg/ml of purified Ab, was also studied. Immunodepletion of CD100-AP supernatant was performed using BD16 or BB27 purified mAb. Briefly, four cycles of depletion were performed using 5 μg of purified mAb incubated for 1 h with supernatant followed by immunoprecipitation with protein A-Sepharose.
Cell surface staining
Cells (2 × 105) were washed in PBS and incubated with anti-neuropilin-1 (29, 30), provided by M. Tessier-Lavigne and Z. He, or anti-neuropilin-2 (unpublished observations) polyclonal rabbit Abs diluted in PBS for 15 min. Negative controls were rabbit polyclonal Abs directed to mouse Igs or anti-Shp1 (Santa Cruz Biotechnology, Santa Cruz, CA). Cells were washed and incubated with FITC-conjugated goat anti-rabbit polyclonal Ab, then analyzed using an XL flow cytometer (Coulter, Miami, FL).
RT-PCR was performed on U937 cells or on human fibroblasts obtained from bone marrow as previously described (31, 32). First-strand cDNA was reverse transcribed in a volume of 30 μl. One microgram of total RNA was annealed to 300 ng of desoxyhexanucleosides (pd(N)6, Pharmacia, Uppsala, Sweden) and incubated for 1 h at 37°C with Superscript II (Life Technologies) in buffer supplied by the manufacturer. One microliter of RT product was used as template in a 50-μl PCR reaction containing 100 ng of each oligonucleotide and 2.5 U of AmpliTaq Gold (Perkin-Elmer, Norwalk, CT). Reactions were performed in a 2400 Gene Amp PCR System (Perkin-Elmer) under the following conditions: 10 min at 94°C, 30–35 cycles (10 s at 94°C, 30 s at annealing temperature, and 30 s at 72°C), and finally 7 min at 72°C.
Primer sequences are the following: human neuropilin-1: forward primer, 5′-ACG ATG AAT GTG GCG ATA CT-3′; reverse primer, 5′-AGT GCA TTC AAG GCT GTT GG-3′ (35 cycles, annealing at 50°C); and ribosomal RNA S14: forward primer, 5′-GGC AGA CCG AGA TGA ATC CTC A-3′; reverse primer, 5′-CAG GTC CAG GGG TCT TGG TCC-3′ (30 cycles, annealing at 64°C).
Soluble CD100 induces inhibition of monocyte migration
Using a Transwell migration assay, we investigated the effects of two types of soluble CD100 on the migration capacity of monocytes isolated from normal peripheral blood. First, as CD100 can be spontaneously shed from the cell surface through the activity of a metalloprotease that generates a 240-kDa homodimeric soluble form (6, 18), we investigated the effect of this soluble CD100 on spontaneous cell migration. Soluble CD100 was harvested from supernatant of Jurkat cells or of 3T3 cells stably transfected with CD100. In control experiments with supernatant of mock-transfected 3T3 cells, approximately 25% of the monocytes spontaneously migrated after 6 h. This spontaneous migration was reduced by 48% in the presence of Jurkat supernatant and by 54% when monocytes were incubated with CD100-transfected cell supernatant (Fig. 1 A).
Second, we examined the effect of CD100 generated from a construction containing the extracellular part of CD100 in fusion with AP. After transient transfection in COS cells, we obtained a supernatant containing CD100-AP as verified by measuring AP activity (not shown). As shown on Fig. 1 B, a similar decrease in monocyte migration (∼40%) was obtained with recombinant CD100 compared with migration obtained with supernatant of AP-transfected COS cells.
Therefore, soluble CD100, regardless of its origin, was able to induce inhibition of spontaneous monocyte migration. Then we investigated the requirement of dimerization for CD100 function. To that purpose, we expressed either wild-type full-length CD100 or full-length CD100 mutated on the membrane-proximal cysteine residue in COS cells. We collected supernatants and tested them in a migration assay. We had shown that only monomeric and no dimeric soluble CD100 was produced with the mutated construction (18). Interestingly, no inhibition of the spontaneous migration of monocytes was observed with this supernatant (Fig. 1 C). Thus, it appears that only dimeric soluble CD100 is active on cell migration.
We determined the quantity of CD100 that can trigger an inhibitory effect on monocyte migration. We measured the quantity of CD100 in the supernatant by ELISA with a standard curve performed with purified CD100. We found that the supernatants contained 300–500 ng/ml of soluble CD100. Then we used different dilutions in the migration assay. As shown in Fig. 1 D, a plateau was reached for a concentration of 40 ng/ml, and an inhibitory effect was observed up to 5 ng/ml of CD100 in the COS supernatant.
To strengthen the idea that CD100 can inhibit the migration of cells, we induced a stronger migration of monocytes by adding the chemokine MCP-3 in the lower chamber of the migration system. The number of migrated cells was increased twice. Interestingly, we observed that CD100 also inhibited this chemokine-induced migration of monocytes (Fig. 1 E).
CD100 inhibits the migration of U937 and Eskol cells, but not that of a T cell clone
The effects of soluble CD100 obtained by transfection of COS cells with recombinant CD100 were studied on various cell types. Similar migration experiments were performed with B and T cells. A negative effect on cell migration (∼20% inhibition) was evident on Eskol cells, a pre-plasma B cell line (not shown), indicating that the inhibition of cell migration by soluble CD100 was not restricted to cells of the monocytic lineage. In contrast, when we tested the effect of CD100 on a T cell clone, B12, we did not observe any significant inhibition of spontaneous migration (not shown), indicating that CD100 does not act on all cell types.
Soluble CD100 present in the supernatant of COS cells reduced spontaneous cell migration in the monocytic cell line U937 by 30% (as shown in Fig. 1 F for a representative experiment) to 50%.
To confirm that the effect was due to soluble CD100, immunodepletion of CD100-containing supernatant was conducted using a CD100 mAb, BD16, or a control mAb, BB27 (6). After immunodepleting CD100, the inhibition of U937 cell migration was no longer observed (Fig. 1 G), whereas control depletion did not modify the inhibition induced by CD100 supernatant.
H-SemaIII, but not H-SemaIV, also inhibits migration of monocytic cells
We compared the effects of soluble CD100 to other human semaphorins. First, we performed transient transfection of COS cells with plasmids containing H-SemaIII or H-SemaIV and collected the supernatants. Using the same migration assay as that described above, with both U937 and monocytes, H-SemaIII induced a similar level of inhibition of cell migration as soluble CD100 (Fig. 2). In contrast, H-SemaIV had no effect on either cell type. These results show that H-SemaIII, a semaphorin originally isolated from nervous tissue, could act, similarly to CD100, on immune cell migration.
CD100 and H-SemaIII, but not H-SemaIV, are recognized by CD100 mAb in an ELISA
Thus, two human semaphorins exerted similar inhibitory effects on migration, whereas a third had no effect. It is interesting that H-SemaIII and H-SemaIV are more homologous to each other in the semaphorin domain than they are to CD100. Therefore, we wanted to determine whether recognition by anti-CD100 mAb would reflect the functional homologies detected between CD100 and H-SemaIII. To this aim, we tested the recognition of these molecules by mAb to CD100, which recognize conformational epitopes (nonblotting mAb). We developed an ELISA using a sandwich with CD100 mAb against distinct epitopes of the molecule: BB18 and BD16, produced in our laboratory.
In these assays the most efficient combination of mAb to recognize CD100 was with BD16 coated onto the dishes and biotinylated BB18 (BB18b) used after incubation with the supernatant containing semaphorins (Table I). Interestingly, H-SemaIII was also strongly recognized in this ELISA with the combination BD16 and BB18b, but was weakly recognized with the combination BB18 and BD16b. Our results suggest that the mAb recognizing soluble CD100 and H-SemaIII with the highest affinity is BD16. H-SemaIV did not give any significant signal in this assay, like other non-human semaphorins tested (M-SemaE, M-SemaH, and R-SemaW; data not shown).
|.||CD100 .||H-SemaIII .||H-SemaIV .|
|.||CD100 .||H-SemaIII .||H-SemaIV .|
Results are expressed as absorbance values at 405 nm for one representative of three experiments giving similar results. b, Biotinylated.
The CD100 mAb, BD16, abolishes the inhibition of migration induced by either CD100 or H-SemaIII
As BD16 recognizes both CD100 and H-SemaIII, we tested the effect of BD16 on the inhibitory activity of semaphorins on U937 migration. U937 cells were selected for these experiments because they are weakly positive for cell surface CD100 staining, thus limiting a possible effect of the CD100 mAb on cell surface CD100 (data not shown). The mAb to CD100 had no effect on spontaneous migration of cells obtained with the control supernatant. As shown in Fig. 3, when semaphorin and BD16 were both added to the upper chamber during the entire duration of the experiment, BD16 mAb abolished the inhibition of migration induced by CD100 or H-SemaIII, whereas incubation with BB27, anti-CD101 control mAb, had no effect. The inhibition was <5% in the presence of BD16 mAb. This indicated that the epitope recognized by BD16 mAb on CD100 and H-SemaIII prevents the fixation of each of the two semaphorins on their receptor.
The receptor for soluble CD100 and H-SemaIII on immune cell migration is not neuropilin-1 or neuropilin-2
We tested the expression of the known receptor for H-SemaIII, neuropilin-1, or for H-SemaIV, neuropilin-2, in the cells sensitive to soluble CD100 and H-SemaIII. Using flow cytometry with polyclonal Abs against neuropilin-1 and neuropilin-2 or appropriate controls, we found that neither U937 nor Eskol cells expressed these receptors on their surface (not shown). Furthermore, when RT-PCR were performed on U937 mRNA using NP-1-specific primers, this gene was not expressed (Fig. 4), although using S14 primers (positive control) gave the predicted product.
As cyclic nucleotides had previously been shown to reverse repulsion into attraction induced by H-SemaIII in the nervous system (33), we tested their effect on inhibition of cell migration. It appeared that cyclic nucleotides did not modify the inhibition induced by H-SemaIII or CD100 (not shown), thus suggesting that H-SemaIII uses a different pathway in immune cells.
To know whether soluble CD100 and H-SemaIII could have a synergistic effect on cell migration, we tested a combination of CD100 and H-SemaIII. To that aim, we mixed supernatants containing CD100 and H-SemaIII in a 1:1 proportion and compared the effect of this mixed supernatant to that of supernatants containing either CD100 or H-SemaIII. Different dilutions of supernatants were tested, and as shown in Table II, no synergistic effect with both semaphorins was observed. This lack of synergy in addition to the effect of BD16 strengthen the possibility of a unique receptor for soluble CD100 and H-SemaIII on immune cells.
|Dilution Factora .||CD100 .||H-SemaIII .||CD100 + H-SemaIIIb .|
|(± 1,700)||(± 1,600)||(± 300)|
|(± 1,100)||(± 2,500)||(± 2,400)|
|(± 1,300)||(± 2,500)||(± 3,600)|
|Dilution Factora .||CD100 .||H-SemaIII .||CD100 + H-SemaIIIb .|
|(± 1,700)||(± 1,600)||(± 300)|
|(± 1,100)||(± 2,500)||(± 2,400)|
|(± 1,300)||(± 2,500)||(± 3,600)|
Supernatants were tested at different dilutions in control-AP supernatant. Results are expressed as number of migrated cells as a function of the dilution factor for each supernatant. Data are means (± SD) of results obtained in three independent experiments. With the control-AP supernatant, 34,300 (± 3,500) cells have migrated.
This mixed supernatant was obtained by mixing CD100 and H-SemaIII supernatants in 1:1 proportion.
Here we show that the soluble form of the human leukocyte cell surface differentiation Ag CD100 inhibits the spontaneous or chemokine-induced migration of cells from the monocytic or B lymphoid lineages. The inhibitory activity was observed using either the supernatant of a human T cell line or of CD100 transfectants containing the spontaneously released CD100 or the supernatant of COS cells transfected with a cDNA construct expressing only the extracellular domain of CD100. Moreover, we showed that only dimeric CD100 exerts an inhibitory effect on cell migration. Another member of the semaphorin gene family, H-SemaIII, isolated from the nervous system, triggers a similar effect on immune cell migration. This was at variance to H-SemaIV, a member more homologous to H-SemaIII in the semaphorin gene family also isolated from the nervous system, which produced no effect on cell migration. Of special interest, using mAb to two discrete epitopes of CD100, we also show in an ELISA that only CD100 and H-SemaIII are recognized, while H-SemaIV is not. Furthermore, we provide evidence that CD100 and H-SemaIII must act through the same receptor, as the effect of both semaphorins is abolished by BD16 mAb, while no synergy between the two molecules is observed. Moreover, this receptor is different from neuropilin-1, the previously reported receptor for H-SemaIII in the nervous system.
Our results show a regulatory effect on immune cell migration of two products of the semaphorin gene family isolated from the nervous system or from the hemopoietic system. This inhibition of cell migration corresponds to a true difference in cell motility. Indeed, we could rule out that this effect corresponds to cell death or apoptosis, as the total numbers of cells recovered from both chambers in the assays were identical after migration under the various conditions. In addition, no effect of semaphorins could be observed on cell cycle, as measured by thymidine incorporation (data not shown). We could as well eliminate an effect on cell adhesion that would prevent cells from migrating across the membrane due to cell aggregation. Indeed, no effect on cell adhesion was observed; we did not test the CD100 effect on the interaction between T cells and monocytes increased by attractin (34), but we could not observe the increase in CD54 expression reported after interaction of viral semaphorins with their receptors (21). All these data show that CD100 and H-SemaIII act directly on cell motility. The migration assay we used, which takes place under gravity, does not allow us to look for a repulsive effect of the semaphorins that should occur against gravity. Rather, this assay detects an inhibition of cell motility, or cell paralysis, following interaction of the semaphorin(s) with its receptor at the cell surface level. This is confirmed by the fact that preincubation of the cells with the semaphorin, followed by removal of excess semaphorin from the medium produced a similar inhibition as the presence of the semaphorin in either the lower or upper chamber of the migration system. Nevertheless, we do not rule out that these semaphorins can also induce repulsion.
Our data provide the first evidence for a role for soluble CD100. CD100 is a discrete semaphorin, as it exists in both a transmembrane and a soluble form. As described in the companion paper (18), the soluble form can be spontaneously generated from the cell surface CD100 in different kind of cells. In addition, the release of CD100 can be increased by T cell activation or by interaction with appropriate mAb, which can act as an agonist (6), mimicking the interaction of surface CD100 with one of its receptors. Among the large family of semaphorins, secreted semaphorins were described according to their ability to regulate axonal growth cone guidance (12). For instance, growth cones collapse when chick sensory axons encounter collapsin-1. Many other semaphorins have been described, but their functions are still unclear. In particular, no function for immune transmembrane semaphorin has yet been reported. Only surface CD100 was shown to be directly involved in T lymphocyte activation through utilization of anti-CD100 mAb, which also revealed association of CD100 with serine kinases and protein tyrosine phosphatases, including CD45 (6, 7). We had shown that CD100-transfected cells increased B cell homotypic aggregation (11), thus raising the possibility of a role for surface CD100 on cells bearing a counter-receptor. However, these previous data did not rule out the possible effect of soluble CD100. Indeed, we further observed that CD100 could spontaneously be released from the cell surface of these transfectants.
When we tested the activities of several other semaphorins, only H-SemaIII induced the same effect as CD100, whereas H-SemaIV had no effect. Strikingly, H-SemaIII was also the only other semaphorin recognized by CD100 mAb in an ELISA, although H-SemaIII and CD100 are not the more homologous in their semaphorin domains. This raises the possibility of a similar conformational domain in H-SemaIII and CD100 that would be essential for activity on immune cells. As BD16 anti-CD100 mAb abolishes the inhibition of cell migration induced by the two semaphorins, this conformational region would be close or similar to the region recognized by BD16 mAb on these two molecules.
We then investigated the receptor for these semaphorins. As neuropilin-1 had been shown to be a receptor for H-SemaIII in the nervous system (19, 20) and had also been involved in the inhibition of endothelial and neural crest cell motility by H-SemaIII (35, 36), we first looked for neuropilins Interestingly, neither neuropilin-1 nor neuropilin-2, its homologue shown to interact with H-SemaIV (23), was expressed in cells sensitive to CD100 or H-SemaIII using flow cytometry or RT-PCR. CD100 and H-SemaIII may thus act through another receptor on immune cells. Plexins, which were first identified as neural cell adhesion molecules (37, 38), were recently shown to be receptors or components of a receptor complex for several semaphorins in different species. In humans VESPR binds viral semaphorins (21) and in Drosophila, d-PlexA was clearly shown to be the receptor for transmembrane d-SemaI (39). Thus, the receptor for CD100 and H-SemaIII on immune cells might belong to this wide family of receptors, especially since recent data established that a plexin can bind transmembrane CD100 (25). Moreover, receptors for semaphorins could consist of multimeric complexes, as an association between neuropilin-1 and plexins was found (25, 40).
Downstream from the receptor, it seems likely that the cytoskeleton is involved in the inhibition of cell migration. We are investigating the effect of semaphorins on actin cytoskeleton and the involvement of small GTPases from the Rho/Rac/CDC42 family (41), as they were shown to be involved in the growth cone collapse induced by collapsin-1 (42, 43). Interestingly, as the cytoskeleton is known to participate in the first stages of Ag uptake by Ag-processing cells, it seems important to investigate the roles of CD100 and H-SemaIII in Ag processing.
When we used our ELISA with anti-CD100 mAb on physiological fluids, we found some semaphorins in human serum and plasma (results not shown). Although we cannot exclude that semaphorins other than CD100 and H-SemaIII are detected in our assay, these results strengthen the physiological relevance of the effects of CD100 and H-SemaIII on immune cell migration described in vitro. Moreover, soluble semaphorins existing in the whole body would be able to regulate cell circulation at different stages of development. We do not know yet the origin of circulating semaphorins, but we can hypothesize that they play a role in different compartments of the organism. First, they could be involved in the regulation of lymphoid organ colonization by hemopoietic precursors. Second, the role of semaphorins in the development of tumors should be mentioned. In the initial phase they could prevent APCs to infiltrate tumors or to migrate to the locoregional lymph node, while at a later stage they could prevent the metastasis of the malignant cells. Human SemaE is involved in non-multiple drug resistance of human cancers in vitro and in vivo (44). Finally, the semaphorins might control the transmigration of leukocytes from blood to inflammation sites. Chemokines were shown to attract leukocytes, and semaphorins might antagonize the effect of chemokines. This effect could be essential for the arrest of immune response to restore a basal state of transmigration. An interesting issue to consider is the role of semaphorins in the blood-brain barrier. It was recently shown that chemokines, as MCP-1, are produced in the brain and promote leukocyte transmigration (45). We are investigating whether semaphorins exist in the cerebrospinal fluid in the normal state, contributing to an equilibrium between blood and brain fluid that protects the brain from leukocyte invasion.
Note added in proof.
We are grateful to Dr. Gordon J. Freeman for giving us CD100 transfectant, and to Dr. Georges Bismuth for helpful discussions.
This work was supported by Institut National de la Santé et de la Recherche Médicale, APEX INSERM (to A.C.), Sidaction and Association pour la Recherche sur le Cancer.
An alternative nomenclature for the semaphorin gene family has recently been proposed (Cell 97:551), which is not used in this report.
Abbreviations used in this paper: AP, alkaline phosphatase; MCP-3, monocyte chemoattractant protein-3.