The Spermatogenic Ig Superfamily/Synaptic Cell Adhesion Molecule Mast-Cell Adhesion Molecule Promotes Interaction with Nerves¹

It has long been demonstrated that mast cells are closely apposed to nerves in a variety of tissues, including skin, intestine, and dura mater (1, 2, 3, 4, 5). Some mast cells actually form membrane-membrane contact with nerve cells in vivo (1, 2). In addition, nerves in association with mast cells often contain neuropeptides, such as substance P (SP)⁴ and calcitonin gene-related peptide (1, 6). Because nerves release these neuropeptides on stimulation, and mast cells express receptors for many neuropeptides (5, 7, 8, 9, 10), nerve activation results in mast cell activation, i.e., degranulation or secretion of mediators (11).

A variety of molecules synthesized by and released from mast cells can, in turn, influence neuronal activity (12, 13). Tryptase directly activates proteinase-activated receptors expressing on neurons (14). Products of arachidonic acid metabolism, such as cysteinyl leukotrienes and PGs, influence the local environment involving nerves (15). Cytokines including TNF-α and growth factors such as nerve growth factor (NGF) cause changes in local nerves so as to lower their threshold to activation (16, 17). These nerve-mast cell effects are postulated to participate in the promotion and regulation of many inflammatory diseases, such as multiple sclerosis, interstitial cystitis, and irritable bowel syndrome (18, 19, 20). However, little is understood about the molecules that connect mast cells with nerves, except that we recently showed a possible involvement of N-cadherin in nerve-mast cell interaction in vitro (21).

In the processes of different researches, Biederer et al. (22) and we (23) isolated the same molecule, named synaptic cell adhesion molecule (SynCAM) or spermatogenic Ig superfamily (SgIGSF). Structurally, SgIGSF/SynCAM has an extracellular domain composed of three Ig-like loops with significant homology to neural cell adhesion molecules-1 and -2 (24) and has a motif sequence in its intracellular domain that putatively connects to the actin cytoskeleton (25).

Biederer et al. (22) clearly demonstrated the function of SgIGSF/SynCAM in neurons. They showed that SgIGSF/SynCAM is localized preferentially on both sides of most synapses in the brain and functions as a homophilic adhesion molecule that spans the synaptic cleft (22). Moreover, they demonstrated that synaptic differentiation was induced even in nonneuronal cells at the contact site with neuronal cells when nonneuronal cells were transfected with SgIGSF/SynCAM and glutamate receptor cDNAs (22).

In contrast, we found that the microphthalmia transcription factor (MITF), a member of the basic-helix-loop-helix-leucine zipper family (26, 27), was essential for the expression of SgIGSF/SynCAM in bone marrow-derived mast cells (BMMC) (23). Because the tg mutant allele is practically a null mutation of the MITF gene (28), C57BL/6 (B6)-tg/tg BMMC do not express SgIGSF/SynCAM, whereas B6 wild-type (+/+) BMMC express it abundantly (23). IC-2 cells, an IL 3-dependent mast cell line established from BMMC of DBA/2 mice (29), were also lacking SgIGSF/SynCAM expression, although whether this defect was attributable to particular abnormalities of MITF in these cells has not been determined (30).

Recent reports by others (31) and ourselves (23, 32) revealed that SgIGSF/SynCAM binds in two ways, homophilic and heterophilic. Thus, there is the possibility that this adhesion molecule may mediate attachment between nerves and mast cells through its homophilic binding. Moreover, considering that the adhesion molecule drives synapse assembly in both neuronal and nonneuronal cells (22), SgIGSF/SynCAM-mediated attachment may result in major enhancement and more efficient communication between nerves and mast cells. In the present study we examined these possibilities in an in vitro model composed of cocultures of neurite-sprouting murine superior cervical ganglia (SCG) neurons with BMMC or IC-2 cells. Mast cells required SgIGSF/SynCAM to attach to SCG neurites and respond to neurite activation efficiently.

WBB6F₁ (F₁)-tg/tg mice were selected by their white coat color and were maintained in our laboratory (23). CBA^+/+, F₁^+/+, F₁-Sl/Sl^d, and WB-W/W mice were purchased from Japan SLC. SgIGSF/SynCAM-transgenic mice were generated previously (33) and were maintained by consecutive backcrosses to B6-tg/tg or B6-tg/+ mice. White offspring with high levels of SgIGSF/SynCAM expression as revealed by Western blot analyses with the tail tip lysates were judged as transgenic tg/tg mice. All animal experiments were performed with approval of the Osaka University Medical School committee.

Establishment and transfection of BMMC were performed as previously described (23). NIH-3T3 and IC-2 cells were maintained as described previously (23, 30). The transfectant clones of IC-2 cells were established previously (30).

Scorpion venom (SV; Leiurus quinquestriatus herbaeus), which induces depolarization in nerve cells by modifying the Na⁺ channel gating mechanism (34), was purchased from Sigma-Aldrich. CP-99,994-1 is an upgrade of the nonpeptide neurokinin-1 (NK-1) receptor antagonist CP-96,345, which has been shown to block the effects of SP (35). This compound was a gift from Pfizer.

Abs against the extracellular domain of SgIGSF/SynCAM (SgIGSF/SynCAM-ED) were generated as follows. We previously constructed a pEFBosFc plasmid vector that expressed SgIGSF/SynCAM-ED as a protein fused with the human IgG1 Fc fragment (32). By PCR using a pair of primers (forward, 5′-ACGCGTCGACGGCAGGTGCCCGACATGGCGAGTGCT-3′; reverse, 5′-GAAGATCTTACTTACTTTGGGGCCCCTGGAACAGAACTTCCAGCGTACCGTATACATACAGCAT-3′), a PreScission site (LEVLFQGP; Amersham Biosciences) sequence was inserted into the fusion site. The soluble Fc fragment-fused protein was purified as described previously (32). According to the manufacturer’s instructions, the protein was digested with the PreScission protease (Amersham Biosciences), and SgIGSF/SynCAM-ED was isolated. Immunization of chickens with SgIGSF/SynCAM-ED and purification of mAbs were performed at Medical and Biological Laboratories. Detailed information about the procedures is available on request from the company. mAbs (chicken IgY) were prepared at concentrations ranging between 0.1 and 0.2 mg/ml in PBS. The blocking activity of the Abs against SgIGSF/SynCAM homophilic binding was evaluated by cell aggregation assays, as shown in Fig. 1, and the 9D2 clone was used as a blocking Ab in the present study.

A rabbit polyclonal Ab against the C terminus of SgIGSF/SynCAM was generated according to the method described by Wakayama et al. (24) and was used for Western blot and immunocytochemical analyses. The Abs specific for stem cell factor (SCF; rabbit polyclonal) and GAPDH (V-18) were purchased from Chemicon International and Santa Cruz Biotechnology, respectively. Anti-c-Kit receptor tyrosine kinase (KIT) rabbit polyclonal Ab (DakoCytomation) was used for Western blot analyses, and anti-KIT Armenian hamster mAb (H2C7) (36), a gift from Dr. T. Hirata (National Institute of Genetics, Mishima, Japan) was used for immunocytochemistry. Normal chicken IgY (U04) was purchased from R&D Systems. Secondary Abs were all obtained from Medical and Biological Laboratories, except for Cy3-labeled anti-Armenian hamster IgG Ab (Jackson ImmunoResearch Laboratories.

Cells and mouse tissues were lysed in a buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% Triton X-100, and 1 mM PMSF. The following procedures were described previously (23). After stripping, the blots were probed with the anti-GAPDH Ab.

Primary culture of SCG neurons was established following a published protocol (37, 38, 39). Briefly, SCG neurons were plated onto Matrigel (BD Biosciences)-coated, 35-mm diameter culture dishes at a density of 0.5–1 × 10⁴ neurons/dish and were grown in F-12 medium (Invitrogen Life Technologies) containing 0.2 mM l-glutamine, 100 ng/ml NGF (Upstate Biotechnology), and 2 μM cytosine-β-d-arabinofuranoside (Sigma-Aldrich).

On day 5, the cultures of SCG neurons were washed three times and then filled with the medium containing 1 × 10⁴ BMMC or IC-2 cells. In some experiments, before the coculture was started, BMMC or IC-2 cells were incubated with Abs (chicken IgY or rabbit IgG; 0.14–14 μg/ml) for 10 min. After 3 h of coculture, the dishes were washed with warmed (37°C) α-MEM to remove nonadherent BMMC and IC-2 cells and were then observed through a ×20 objective of a differential interference contrast (DIC) light microscope (LSM410; Zeiss). This low power field contained five to 15 neurons (neurons were counted by their cell bodies). Images of 20 randomly selected fields were captured for each dish, and the number of BMMC or IC-2 cells attaching to one neuron was calculated by dividing the total number of BMMC or IC-2 cells remaining in the coculture by the total number of neurons. More than six dishes were prepared per group, and the mean ± SE were calculated. The results were reproduced twice by independent experiments. In some cases the cocultures were stained with Alcian Blue and Nuclear Fast Red to identify BMMC.

Cells were immunostained according to procedures described previously (23). Cells were visualized using a confocal laser scanning microscope (LSM510; Zeiss).

As described previously (39, 40), the calcium fluorophore fluo-3-acetoxymethyl ester (Molecular Probes) was used to assess Ca²⁺ mobilization as an indicator. Briefly, after overnight coculture, the cells were incubated in culture medium containing 1 μM fluo-3-acetoxymethyl ester for 30 min, followed by three rinses in HEPES buffer. Under observation through a confocal laser scanning microscope (LSM-410; Zeiss), SV was added to the coculture at a concentration of 1 μg/ml, and fluo-3 fluorescence (i.e., Ca²⁺ mobilization) was detected every 5 s using excitation and emission wavelengths of 488 and >505 nm, respectively. Images were captured and analyzed with the Scenic Pro M7 computer analysis system (Siemens) as described previously (39, 40).

Additionally two experiments were performed. In experiment A, the cocultures were developed in the presence of Abs (chicken IgY; 0.14–14 μg/ml). In experiment B, before the addition of SV, the cocultures were incubated with CP-99,994-1 at concentrations raging between 0.1 and 100 ng/ml for 30 min. The following procedures were the same as described above. When the fluo-3 fluorescence intensity in IC-2 cells increased by >25 arbitrary units after neurite activation evoked by SV, the IC-2 cells were considered to be responding. At least three coculture dishes were prepared per group, and data were obtained from >50 neurite-IC-2 cell units. All results were reproduced twice by independent experiments.

The procedure for tissue preparation was similar to that described by Hashimoto et al. (41). An H-600A electron microscope (Hitachi) was operated at 75 kV.

After the example of our previous studies (37, 38, 39), we cocultured BMMC and IC-2 cells with SCG neurons from CBA^+/+ mice. Because we planned to coculture BMMC with SCG neurons derived from F₁-Sl/Sl^d mice in a later experiment, we used mainly BMMC derived from F₁ mice. Firstly we examined the expression levels of SgIGSF/SynCAM in BMMC, IC-2 cells, and CBA^+/+ SCG. F₁^+/+ BMMC expressed easily detectable levels of SgIGSF/SynCAM as a protein of ∼110 kDa, whereas neither F₁-tg/tg BMMC nor IC-2 cells expressed it (Fig. 2). In CBA^+/+ SCG, SgIGSF/SynCAM was expressed >5-fold as abundantly as F₁^+/+ BMMC as a protein of ∼90 kDa (Fig. 2). The difference in the mobility size of SgIGSF/SynCAM between BMMC and SCG was probably due to cell type-specific glycosylation (23, 24, 25).

When CBA^+/+ SCG neurons were cultured in the presence of NGF, they sprouted numerous neurites within 1 wk. On the fifth day, F₁^+/+ BMMC were seeded onto the culture of SCG neurons. After 3 h, floating BMMC were removed by changing the medium several times. A considerable number of BMMC remained in the dishes; BMMC were identified by their small size and round shape. This identification was confirmed by staining with Alcian Blue and Nuclear Fast Red (data not shown). All BMMC were associated with SCG neurites and were absent from the area where SCG neurites had not developed.

The coculture of F₁^+/+ BMMC and CBA^+/+ SCG neurons after removal of nonadherent cells were stained with the Abs against SgIGSF/SynCAM and KIT. Representative low and high power field photomicrographs of the staining are shown in Fig. 3, A–C and D–H, respectively. The anti-SgIGSF/SynCAM Ab clearly stained almost all SCG neurites along their whole length (Fig. 3, A and E). Enhancement of the signals was detected in the area where F₁^+/+ BMMC attached to SCG neurites. Double staining with the anti-KIT Ab revealed that the enhanced SgIGSF/SynCAM signals were colocalized with KIT, but, in addition, extended beyond KIT signals toward the neurite side (Fig. 3, F and G). Because the anti-KIT Ab stained the peripheral margin of BMMC, but not the neurites (Fig. 3,F), the double-staining merged image indicated that SgIGSF/SynCAM signals were enhanced not only in F₁^+/+ BMMC, but also in SCG neurites at the site where both attached. Stacking the confocal plane images, SgIGSF/SynCAM was found to be concentrated on the cell membrane of BMMC in contact with SCG neurites (Fig. 3 H). These results suggested that SgIGSF/SynCAM mediated the attachment of BMMC to SCG neurites through its homophilic binding.

To examine whether SgIGSF/SynCAM on mast cells is required for the attachment, F₁^+/+ BMMC, F₁-tg/tg BMMC, or IC-2 cells were cocultured with SCG neurons. After 3 h, we counted the number of BMMC or IC-2 cells that attached to SCG neurites sprouting from one neuron. The numbers of F₁-tg/tg BMMC and IC-2 cells that attached to SCG neurites were 1/3rd and 1/10th as large as the number of F₁^+/+ BMMC that attached to SCG neurites, respectively (Table I).

We further performed two experiments, in which we examined 1) whether ectopic expression of SgIGSF/SynCAM improved the poor attachment of tg/tg BMMC and IC-2 cells, and 2) an Ab blocking the homophilic binding of SgIGSF/SynCAM inhibited the attachment of F₁^+/+ BMMC to SCG neurites. First, we transfected F₁-tg/tg BMMC and IC-2 cells with either a retrovirus vector containing SgIGSF/SynCAM cDNA (F₁-tg/tg BMMC^{SgIGSF/SynCAM} and IC-2^{SgIGSF/SynCAM}) or an empty vector (F₁-tg/tg BMMC^vector and IC-2^vector). In addition, we established BMMC from two tg/tg mice carrying a SgIGSF/SynCAM transgene (N₄A-tg/tg-1 and -2), which were obtained in the fourth generation of backcrosses of the transgenic founder (33) with B6-tg/+ mice. Transfection with SgIGSF/SynCAM cDNA, but not with an empty vector, resulted in its overexpression in F₁-tg/tg BMMC and IC-2 cells, whereas N₄A-tg/tg-1 and -2 BMMC expressed normal level of SgIGSF/SynCAM (Fig. 3). We examined the attachment of these cells to SCG neurites. The three types of tg/tg BMMC that expressed ectopic SgIGSF/SynCAM showed normal attachment to neurites, whereas vector transfection alone did not influence the attachment of F₁-tg/tg BMMC to neurites (Table I). IC-2 cells expressing ectopic SgIGSF/SynCAM exhibited an attachment level very comparable with that of F₁^+/+ BMMC (Table I).

Second, we generated an mAb against the extracellular domain of SgIGSF/SynCAM, named 9D2, which inhibited the homophilic binding of SgIGSF/SynCAM (see Fig. 1). We cocultured F₁^+/+ BMMC or IC-2^{SgIGSF/SynCAM} cells with SCG neurites in the presence of 9D2 at concentrations ranging from 0.14 to 14 μg/ml. 9D2 inhibited the attachment of F₁^+/+ BMMC and IC-2^{SgIGSF/SynCAM} cells to SCG neurites in a dose-dependent manner (Table II). At its maximal concentration (14 μg/ml), the Ab reduced the attachment levels of F₁^+/+ BMMC and IC-2^{SgIGSF/SynCAM} cells to those of F₁-tg/tg BMMC and intact IC-2 cells, respectively (Table II). In contrast, the anti-SgIGSF/SynCAM C terminus Ab and control chicken IgY did not influence the attachment of F₁^+/+ BMMC and IC-2^{SgIGSF/SynCAM} cells even at a concentration of 14 μg/ml. The results from these two experiments consistently indicated that SgIGSF/SynCAM on mast cells largely mediated the attachment of F₁^+/+ BMMC and IC-2^{SgIGSF/SynCAM} cells to SCG neuritis.

There still existed the possibility that KIT and its ligand SCF might be involved in the attachment of BMMC to SCG neurites. The following evidence suggested this. 1) KIT and SCF play important roles in attachment of BMMC to NIH-3T3 fibroblasts (42). 2) Some types of neurons express either KIT or SCF (43, 44), and mast cells express KIT and SCF (23, 45, 46). In fact, SCF expression was detectable in CBA^+/+ SCG cells as easily as in NIH-3T3 fibroblasts, although KIT expression in these two types of cells was below the limit of detection (Fig. 2). F₁^+/+ BMMC expressed both SCF and KIT (Fig. 2). 3) The tg/tg BMMC express KIT, but at a significantly reduced level (47).

To examine whether SCF expressed on SCG neurons might be involved in the attachment of BMMC to SCG neurites, we used F₁-Sl/Sl^d mice, which do not express SCF on the cell surface (48). In addition, we examined whether KIT expressed on SCG neurons had any role in the attachment by using WB-W/W mice, which lack cell surface expression of KIT (49). As a control, strain-matched wild-type (F₁^+/+ and WB^+/+) mice were used. We established a culture of SCG neurons derived from each of these four types of mice, then seeded F₁^+/+ or F₁-tg/tg BMMC onto it. The number of F₁^+/+ BMMC that attached to the neurites from F₁-Sl/Sl^d and WB-W/W mice was as large as the number of F₁^+/+ BMMC that attached to neurites from the corresponding wild-type mice (Table III). F₁-tg/tg BMMC showed poor attachment in every set of cocultures; the number of F₁-tg/tg BMMC that attached to the neurites was one-third or one-fourth the number of F₁^+/+ BMMC that attached (Table III).

KIT involvement was also examined by experiments using IC-2 cells. IC-2 cells expressing ectopic KIT (IC-2^KIT) were as poor in the attachment to SCG^+/+ neurites as intact IC-2 cells (Table I). Moreover, ectopic expression of KIT along with SgIGSF/SynCAM (IC-2^{SgIGSF/SynCAM + KIT}) did not generate any additional effects on the attachment of IC-2 cells than ectopic expression of SgIGSF/SynCAM alone (IC-2^{SgIGSF/SynCAM}; Table I). Based on these results, we concluded that neither SCF nor KIT was involved in the attachment of BMMC and IC-2 cells to SCG neurites.

We previously observed that SCG neurites communicate with RBL and BMMC by releasing SP in the coculture (39, 40). We examined the possible involvement of SgIGSF/SynCAM in such communication. The cocultures of IC-2 cells, intact and transfected, with SCG neurites were preloaded with the calcium fluorophore fluo-3, then stimulated with SV, a depolarizer of nerve cells (34). A representative result of the fluo-3 fluorescence trace in the coculture of IC-2^{SgIGSF/SynCAM} cells is shown in Fig. 4. Immediately after the addition of SV, the neurons became activated; the fluorescence intensity sharply increased in neuron cell bodies and neurites within a few seconds, reached a maximum in 30 s, and declined gradually toward the end point of observation (Fig. 4). After neuron activation, some IC-2 cells attendant to the neurites became activated, as recognized by a transient marked increase (>25 arbitrary units) in fluorescence intensity (Fig. 4,B). IC-2 cells did not directly mobilize Ca²⁺ in response to SV (1 μg/ml) in the absence of neurites (data not shown). The time lag between neurite activation and IC-2 cell activation was ∼18 s, and the magnitude of the transient increase in fluorescence intensity of IC-2 cells was ∼50–60 arbitrary units (Table IV). These parameters concerning neurite and IC-2 cell activations were comparable among the cocultures of intact IC-2 cells and those of three types of transfected IC-2 cells (Table IV). However, there was a significant difference in the cell proportion responding to neurite activation; only one-quarter of intact IC-2 and IC-2^KIT cells responded to neurite activation, whereas more than half of the IC-2^{SgIGSF/SynCAM} and IC-2^{SgIGSF/SynCAM + KIT} cells responded (Table IV).

IC-2^{SgIGSF/SynCAM} cells were cocultured with SCG neurites in the presence of 9D2, and then SV was added to the coculture. The proportion of IC-2^{SgIGSF/SynCAM} cells responding to neurite activation was decreased in a 9D2 dose-dependent manner and was the same as that of IC-2 cells responding in the absence of any Abs when 9D2 was present at a concentration of 14 μg/ml (Fig. 5 A). These results suggest that SgIGSF/SynCAM-mediated binding promoted SCG neurite-to-IC-2 cell communication.

To examine whether SP mediated the SCG neurite-to-IC-2 cell communication, we cocultured IC-2 cells, intact or SgIGSF/SynCAM-transfected, with SCG neurites in the presence of CP-99,994-1, an NK-1 receptor antagonist (35), and then added SV to the coculture. The presence of the antagonist altered neither the number of IC-2 cells attaching to the neurites nor the Ca²⁺ mobilization curves of neurites or IC-2 cells (data not shown). In contrast, CP-99,994-1 decreased the proportion of IC-2 cells responding to SV-evoked neurite activation in a dose-dependent manner (Fig. 5,B). When neurite-to-intact IC-2 cell communication was compared with neurite-to-IC-2^{SgIGSF/SynCAM} communication, there was a significant difference in the susceptibility to the antagonistic effect of CP-99,994-1. CP-99,994-1 required only a concentration of 1 ng/ml to reduce the proportion of intact IC-2 cells responding by 75%, whereas it required a concentration as high as 100 ng/ml to produce the similar effect on IC-2^{SgIGSF/SynCAM} cells (Fig. 5 B).

After overnight coculture, SCG neurites and attendant IC-2 cells were observed through an electron microscope. IC-2 cells were identifiable by their cell surface microprocesses and a few electron-dense granules (41). We observed several sets of IC-2 cells and SCG neurites attaching to each other and obtained essentially similar findings regardless of the type of IC-2 cells, intact IC-2 or IC-2^{SgIGSF/SynCAM} cells. At a point of contact between an IC-2 cell and a neurite, the plasma membranes of both cells were apposed and ran parallel to each other with a narrow space, but did not exhibit typical characteristics of synapses (Fig. 6).

We have examined whether SgIGSF/SynCAM was involved in the attachment and communication between mast cells and neurites by coculturing SCG neurons and mast cells with or without SgIGSF/SynCAM. Consistent with our previous immunohistochemical experiments in which we detected SgIGSF/SynCAM along peripheral nerve fibers in mouse and human lung tissues (50), cultured SCG neurites expressed SgIGSF/SynCAM along their entire length (Fig. 3). Thus, once F₁^+/+ BMMC came in touch with SCG neurites, SgIGSF/SynCAM expressed on the plasma membranes of both cells was likely to bind homophilically and promote the attachment between the cells. The following results supported this idea. 1) SgIGSF/SynCAM was localized intensively at the contact site between F₁^+/+ BMMC and SCG neurites. 2) Mast cells lacking SgIGSF/SynCAM, i.e., F₁-tg/tg BMMC and IC-2 cells, attached poorly to SCG neurites, and ectopic expression of this adhesion molecule significantly improved their attachment. 3) The attachment levels of F₁^+/+ BMMC and IC-2^{SgIGSF/SynCAM} cells were reduced to those of F₁-tg/tg BMMC and intact IC-2 cells, respectively, when the blocking Ab 9D2 was present in the coculture. Nonetheless, these results do not completely exclude the possible presence of a heterophilic binding ligand(s) on SCG neurites. Although the ligand(s) on neurites should be determined definitely, mast cells appeared to use SgIGSF/SynCAM to attach to neurites.

SCF-KIT interaction did not play any role in BMMC attachment to neurites, contrasting with the fact that it plays a significant role in BMMC attachment to fibroblasts (42). Recently we (30) have shown that normal KIT expression by mast cells is a prerequisite for SgIGSF/SynCAM to function as an adhesion molecule in BMMC attachment to fibroblasts. The necessity of KIT for mast cell attachment may differ depending upon the binding manner of SgIGSF/SynCAM, because mast cell attachment to neurites and fibroblasts is postulated to occur via its homophilic (present study) and heterophilic (23) binding, respectively. In this respect, it is worth noting that L1, a member of the Ig superfamily of neural cell adhesion molecules, binds homophilically and heterophilically and produces opposite effects on L1-mediated neurite growth, stimulation, and inhibition, respectively (51). This indicates that extracellular binding events affect intracellular signaling. Partner molecules associated with the cytoplasmic domain of SgIGSF/SynCAM are thus likely to be different between fibroblasts and neurites, which may account for the differences we observed.

The number of F₁-tg/tg BMMC and IC-2 cells that attached to SCG neurites was appreciable, but was as small as the number of F₁^+/+ BMMC attaching to SCG neurites in the presence of 9D2. This suggests the presence of another molecule(s) in addition to SgIGSF/SynCAM that mediates the attachment between mast cells and neurites. N-cadherin could be considered a major candidate molecule to account for the residual binding observed that could not be ascribed to SgIGSF/SynCAM, because it is expressed by mast cells as well as nerves (21, 23) and binds homophilically (52). Even if this speculation were true, the present results indicate that SgIGSF/SynCAM was the major adhesion molecule that mediated in vitro attachment between mast cells and SCG neurites.

Ectopic expression of SgIGSF/SynCAM doubled the proportion of IC-2 cells responding to SV-evoked neurite activation without changing the other parameters of the response. This indicates that SgIGSF/SynCAM does not affect the nature of the response of mast cells, but does affect the susceptibility of individual mast cells to neurite activation. In addition, the NK-1 receptor antagonist CP-99,994-1 blocked the communication between SCG neurites and intact IC-2 cells at much lower concentrations than the communication between SCG neurite and IC-2^{SgIGSF/SynCAM} cells (Fig. 5 B). However, no plasma membrane ultrastructures characteristic of synapses were observed at the points of contact between IC-2^{SgIGSF/SynCAM} cells and neurites. We previously obtained similar findings in rat intestine, where unmyelinated nerves closely appose to mucosal mast cells, but synapse-like specialized junctions are lacking between them (2, 37). It does not seem that synaptic structures develop in the apposed plasma membranes of nerves or mast cells. Nonetheless, SgIGSF/SynCAM can induce and localize synapse assembly in neurons and even in nonneuronal cells to form functional synapses (22). Thus, this adhesion molecule is likely not only to function as simple glue in nerve-mast cell interaction, but also to promote the development of a microenvironment in which mast cells have an enhanced susceptibility to nerve activation. A speculation could be drawn as a simple explanation of the present results. SgIGSF/SynCAM increases the number of NK-1 receptors on mast cells or the amount of SP released from the neurite. Although this speculation needs future elucidation, SgIGSF/SynCAM on mast cells appears to contribute to both in vitro attachment and communication with nerves.

Mast cells are not only major effectors in allergic reactions, but recent studies demonstrated that they are also involved in a variety of noninfectious inflammatory diseases, such as multiple sclerosis, migraines, atopic dermatitis, interstitial cystitis, and irritable bowel syndrome (20). As the condition of the diseases gets worse, nerve-mast cell interaction is proposed to grow stronger. Clinically, psychological stress often worsens the symptoms. In the intestinal mucosa of irritable bowel syndrome, the number of activated mast cells in close proximity to nerves is reported to correlate with the severity and frequency of abdominal pain and/or discomfort (18). Provided that SgIGSF/SynCAM also promotes nerve-mast cell interaction in vivo, it may play a pivotal role in the pathophysiology of these inflammatory diseases.

In conclusion, the present study not only showed that SgIGSF/SynCAM on mast cells predominantly mediated the in vitro attachment to SCG neurites, but suggested that this also promoted the functional communication between the two. Additional characterization of SgIGSF/SynCAM as an adhesion molecule in nerve-mast cell interaction will provide us with a deeper insight into the molecular basis underlying the linkage between nervous and immune systems.

We thank M. Kohara, K. Hashimoto, and T. Sawamura (Osaka University) and S. Tanioka (Kobe University) for excellent technical assistance.

The authors have no financial conflict of interest.