INTRODUCTION

The mucosal immune system of the gastrointestinal tract is distinct from the systemic immune system by the general immunologically suppressed tone in the gut and the induction of specific immune unresponsiveness after antigen challenge, termed oral tolerance. It is believed that CD8⁺ suppressor T cells primed in the gut may play an important role in oral tolerance (1-7). However, the mechanism underlying the generation of the CD8⁺ suppressor T cells has not been completely delineated. Previous studies in our lab as well as others have shown that intestinal epithelial cells are able to process and present antigens to T cells (8, 9). More interestingly, intestinal epithelial cells are able to selectively induce CD8⁺ suppressor T cell proliferation in an allogeneic mixed epithelial cell/T cell reaction (METR)¹ system (8), and conventional restriction elements do not appear to be involved, since neither antibodies against class I nor class II MHC block proliferation (10). Furthermore, the proliferation is not driven by soluble factors from the epithelial cells (8). One molecule that appears to be clearly involved in this interaction is CD8, since there is activation of CD8-associated p56^lck and antibody to CD8 blocks CD8⁺ suppressor T cell proliferation in this system (10).

We have hypothesized that the generation of CD8⁺ suppressor T cells is attributed to the expression of a unique CD8 binding ligand on the surface of intestinal epithelial cells (10). Therefore, a series of epithelial cell-specific monoclonal antibodies were generated and screened for their ability to inhibit CD8⁺ T cell proliferation in our culture system. Two of these mAbs, L12 and B9, were found to be potent inhibitors of this mixed epithelial cell/T cell reaction system but did not inhibit conventional MLRs (10, 11). The epithelial cell surface molecules recognized by these two mAbs might therefore be candidates for the ligand regulating the METR. Previous studies have documented phenotypic and functional differences between the molecules recognized by mAb L12 and mAb B9. These two mAbs have a different tissue distribution determined by immunohistochemical staining. Both antibodies can stain intestinal epithelium, but L12 stains epithelial cells from the airway as well, and B9 stains thymic epithelium and the syncytiotrophoblast in the placenta (11). As described here, a 180-kDa epithelial cell membrane glycoprotein (gp180) is recognized in Western blot by both mAb B9 and mAb L12, in either normal human colon epithelial cells or in a colon cancer cell line, T84. In this paper, we studied the biochemical and functional characteristics of gp180. Our data suggest that gp180 may be one of the epithelial cell membrane proteins mediating epithelial cell/T cell interactions.

EXPERIMENTAL PROCEDURES

Monoclonal Antibodies and Conjugation of mAb to Sepharose 4B Beads

Freshly isolated human intestinal epithelial cells were used to immunize Balb/c mice on days 0 and 21. Splenocytes were harvested 3 days later and fused with the nonsecreting myeloma line SP2/0 as described previously (11). Hybridomas were screened by 1) specificity (by staining) for intestinal epithelial cells but not T cells, B cells, or monocytes; 2) the ability of mAbs to inhibit CD8⁺ T cell proliferation in METR but not conventional MLRs; and 3) the ability to inhibit the activation of CD8-associated p56^lck in METR. Two mAbs were identified in this screening, B9 and L12 (both IgG1 isotype) (10, 11). Monoclonal antibodies were prepared and purified by recombinant protein G-Sepharose 4B (Pharmacia Biotech Inc.) affinity chromatography according to the procedure described by Yokoyama (12) and Andrew and Titus(13). Purity of the mAbs was determined by minigel. mAb B9 or L12 was conjugated to Sepharose 4B using a procedure supplied by Pharmacia.

Cell Lines

T84, DLD-1, CaCO₂, and HT29 are malignant intestinal epithelial cell lines obtained from the American Type Culture Collection (ATCC, Rockville, MD). OKT4, OKT8, and W6/32 are hybridomas obtained from the ATCC that are capable of producing monoclonal antibodies against human CD4, CD8, and class I MHC, respectively. 3G8 and 3G4 are murine T cell hybridoma transfectants that constitutively express human CD8 and CD4, respectively. Their functional properties have been described previously (14, 15). These two cell lines were kind gifts from Dr. Steven Burakoff (Dana-Farber, Boston, MA).

Isolation of Enterocytes

Enterocytes were isolated by a method described previously (8). Surgical specimens were obtained from the operating room. The specimens were then washed extensively with PBS containing 1% penicillin/streptomycin and 1% fungizone (Flow Laboratory, Inc., McLean, VA). The mucosa was stripped off from the submucosa, minced into small pieces, and placed in 1 mM dithiothreitol (Sigma) for 5 min at room temperature to remove mucus. The pieces were washed in PBS and incubated in dispase (3 mg/ml in RPMI 1640, Boehringer Mannheim GmbH, Germany) for 30 min at 37 °C, vortexing every 5 min. The tissue pieces were removed, and the cell suspension was then collected and centrifuged on a Percoll density gradient. Enterocytes were located at the 0-30% layer interface. Cells were washed 3 times with PBS and resuspended in culture medium (CM) (RPMI 1640 with 10% FCS, 50 units/ml penicillin, 50 µg/ml streptomycin, 2 mM glutamine, all from Life Technologies, Inc.). Preparations of purified enterocytes were >90% viable and free of macrophage and B cell contamination as determined by staining with anti-CD14 and anti-CD20 mAbs (Coulter Corp., Hialeah, FL) and were contaminated with between 2 and 4% intraepithelial lymphocytes (CD3⁺ cells).

Isolation of T Cells and Non-T Cells from Peripheral Blood

Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized venous blood collected from normal donors, as described previously (16), by Ficoll-Paque density gradient centrifugation. T cells and non-T cells were isolated from PBMCs by a rosetting method using neuraminidase-treated sheep red blood cells (16).

T cells were also enriched by a nylon wool column (Polysciences, Inc., Warrington, PA). PBMCs were adjusted to 75 × 10⁶/ml in RPMI 1640, 5% FCS, prewarmed at 37 °C in a humidified 5% CO₂ incubator. 1.5 ml of the PBMC suspension was loaded onto each prewarmed nylon wool column (each column contained 0.6 g of nylon wool, balanced with RPMI 1640, 5% FCS, preincubated at 37 °C for 45 min in a humidified 5% CO₂ incubator). The loaded column was then incubated at 37 °C for 45 min in a humidified 5% CO₂ incubator in an upright position. T cells were eluted by 18 ml of 37 °C RPMI 1640, 5% FCS and collected.

Cell Staining and Flow Cytometry

1-2 × 10⁵ isolated enterocytes were incubated with monoclonal antibodies on ice for 45 min. The cells were washed 3 times with PBS/BSA solution (1% BSA, 0.1% NaN₃, in PBS), and then resuspended in 50 µl of 1:50 diluted fluorescein isothiocyanate-conjugated F(ab)₂ goat anti-mouse IgG (Tago, Inc., Burlingame, CA) and incubated on ice for another 45 min. The cells were washed again 3 times and finally resuspended in 400 µl of PBS for flow cytometric analysis. Controls included IgG1 anti-DNP mAb (negative control) and W6/32 (anti-class I MHC, positive control).

Ten to twenty million isolated enterocytes, T84 cells harvested by non-enzyme cell dissociation solution (Sigma), or T84 cells in monolayer cultures were washed 3 times with PBS and treated with PIPLC (Sigma) at concentrations of 0.3-1 unit/ml RPMI at 37 °C for 45 min. At the end of this incubation, the cell-free supernatant was collected for further studies. The treated cells were stained by various monoclonal antibodies and analyzed by flow cytometry, to confirm the removal of GPI-anchored molecules.

Metabolic Labeling with [³⁵S]Methionine/Cysteine

Subconfluent intestinal epithelial cell line cultures in T75 tissue culture flasks were washed with PBS 3 times and starved for 4 h in methionine/cysteine-free RPMI 1640 containing 10% dialyzed FCS. The starved cells were cultured with trans-label [³⁵S]methionine/cysteine (ICN Biomedicals, Inc., Costa Mesa, CA) (1 mCi/20 million cells) in methionine/cysteine-free RPMI 1640, 10% dialyzed FCS for 18 h. Cells were then washed with PBS 3 times and lysed for immunoprecipitation studies.

Western Blot

Western blot analyses were performed as described previously (10). Briefly, cell lysates were resolved on 10% SDS-PAGE and transferred onto a nitrocellulose membrane at 15 V overnight in transfer buffer (20% methanol, 150 mM glycine, 25 mM Tris, pH 8.3). After transfer, the nitrocellulose sheet was blocked by 50 ml of 5% nonfat milk in PBS. The nitrocellulose sheet was washed once with PBS and incubated with primary antibody (2-10 µg/ml) in a 0.5% nonfat milk/PBS solution at 4 °C overnight. After washing 3-5 times with washing buffer (0.05% Tween 20 in PBS), horseradish peroxidase-conjugated goat anti-mouse IgG antibody (1-2 µg/ml) (Cappel-Organon Teknika Corp., Durham, NC) was added. This incubation was continued at room temperature for 1-2 h. The sheet was then washed 3-5 times with washing buffer and incubated with 12 ml of chemiluminescence reagent (Du Pont NEN) at room temperature for 1 min. XAR-5 films were exposed and developed.

Epithelial Cell Membrane Isolation

Cultured T84 cells were washed 3 times with PBS and harvested by non-enzyme cell dissociation solution (Sigma). The cells were resuspended in cell disruption buffer (1 mM PMSF, 5 mM iodoacetamide, 20 µg/ml aprotinin, 20 µg/ml leupeptin, in PBS, pH 7.4, all from Sigma) and sonicated to disrupt the cellular structure, followed by centrifugation at 500 × g for 10 min. The pellet containing nuclei was discarded, and the supernatant was subjected to ultra-centrifugation at 100,000 × g for 1 h at 4 °C. The supernatant, containing cytosol, was removed, and the membrane pellet was collected.

Immunoaffinity Purification of gp180

The supernatant generated by PIPLC treatment of T84 epithelial cell monolayer was obtained as described above. The PIPLC-generated supernatant from 80 × 10⁶ [³⁵S]methionine/cysteine-labeled T84 cells was passed through a preclearing column of Sepharose 4B beads (Pharmacia Biotech Inc.) to remove nonspecific binding proteins. The precleared supernatant was then subjected to mAb B9-Sepharose 4B affinity chromatography. The column was washed 5 times and eluted with 0.1 M glycine HCl buffer, pH 2.7. The eluate was neutralized immediately with 0.1 volume of 1 M Tris-HCl solution, pH 9.6, dialyzed exhaustively against PBS and concentrated to 80 µl. Part of the purified gp180 was examined for purity by SDS-PAGE and autoradiography, and the rest was stored at 70 °C for further studies.

Immunoprecipitation

The T84 cell pellet or cell membrane pellet (2 × 10⁶ cell equivalents) was lysed in 0.2 ml of lysis buffer (1% Nonidet P-40, 0.1% SDS, 150 mM NaCl, 20 mM Tris-HCl, 1 mM PMSF, 5 mM iodoacetamide, 20 µg/ml aprotinin, 20 µg/ml leupeptin, pH 7.4) for 2 h on ice, vortexing every 15 min. The lysate was precleared with 20 µl of Sepharose 4B for 1 h at 4 °C. The precleared lysate was mixed with monoclonal antibody B9-Sepharose 4B beads (20 µl) and incubated at 4 °C for 8 h or overnight. The suspension was then spun in a microcentrifuge, and the beads were washed 5 times with 1 ml of washing buffer (0.1% Triton X-100, 0.1% SDS, in PBS, pH 7.4). After the final wash, sample buffer was added to the beads. The sample was heated at 100 °C for 5 min and subjected to SDS-PAGE analysis.

Eighty million T84 cells were metabolically labeled with [³⁵S]methionine. gp180 was immunoprecipitated as described above and then eluted with elution buffer (50 mM diethylamine, 0.5% sodium deoxycholate, pH 11.5, Sigma). The purified gp180 was dialyzed against PBS and concentrated to 150 µl.

5.4 µl of digestion buffer (0.2 M Na₂HPO₄, 0.2 M NaH₂PO₄, 2 mM EDTA, 4 mM PMSF, 1% SDS, 100 mM -mercaptoethanol) was added to 16 µl of purified gp180, mixed, and boiled for 5 min. 1.4 µl of 20% Nonidet P-40 and 1 µl (0.25 unit) of N-glycanase (Genzyme Diagnostics, Cambridge, MA) were added to the mixture and incubated at 37 °C overnight. The control without enzyme was placed with all components and treated exactly the same except that the 1 µl of N-glycanase was replaced with 1 µl of PBS. The treated materials were analyzed by reducing SDS-PAGE, autoradiography, and Western blot (mAb B9 and/or L12).

1 µl (1 milliunit) of neuraminidase (Calbiochem), 5 µl of 0.5 M Na₂HPO₄, 5 µl of 0.5 M NaH₂PO₄ were added to 48 µl of purified gp180 and incubated at 37 °C for 1 h. 19 µl was sampled for SDS-PAGE, autoradiography, and Western blot analysis. The rest was saved for further study.

1 µl (1 milliunit) (Genzyme Diagnostics) of O-glycanase was added into 40 µl of neuraminidase-treated gp180. The enzyme treatment was performed at 37 °C for 4 h. At the end of the treatment, 20 µl of treated gp180 was used for analysis, and the rest was saved for further treatment.

Binding-Cross-linking-Western Blot

Ten million T84 cells were treated with PIPLC as described above, and the supernatant was collected. The PIPLC-generated supernatant was concentrated by centricon (Amicon) to 0.5 ml. Forty million T cells or non-T cells were washed 3 times with RPMI, and the cell pellets were resuspended in the concentrated PIPLC-generated supernatant with addition of 0.1% BSA into the mixture. After 15 min incubation at room temperature, 7 µl of a 15 mg/ml DTBP (dimethyl 3,3-dithiobispropionimidate, a bifunctional cross-linker; Pierce) solution was added into the mixture, and the incubation was continued for 30 min at room temperature. The cells were washed 5 times with 1% BSA in PBS, resuspended in 50 µl of PBS, and 17 µl of 4x reducing sample buffer was added. The samples were boiled for 5 min and spun for 10 min at 4 °C. The supernatant was collected and analyzed by reducing SDS-PAGE and mAb B9 Western blot.

Stimulation of T Cells

2 × 10⁶ T cells were stimulated with either OKT4 + affinity purified rabbit anti-mouse IgG (RAM, Cappel), OKT8 + RAM, 2 × 10⁶ intestinal epithelial cells, or purified gp180, for 1-10 min at 37 °C. The stimulation was stopped by addition of 1 ml of ice-cold stop buffer (100 µM Na₂VO₃ in PBS). The cells were centrifuged and the stop buffer removed. Two hundred µl of lysis buffer (1% Nonidet P-40, 100 µM Na₂VO₃, 1 mM PMSF, 5 mM iodoacetamide, 20 µg/ml aprotinin, 20 µg/ml leupeptin, 140 mM NaCl, 20 mM Tris-HCl, pH 7.4) was added into the cell pellet and kept on ice for 30 min, vortexing every 5-10 min. The cell lysate was centrifuged at 4 °C for 10 min in a microcentrifuge. The pellet was removed, and the supernatant was transferred to a clean tube for further studies. For the assays using epithelial cell lines as stimuli, a hypotonic lysis buffer was used to ensure that >80% of T cells were lysed, whereas >85% of epithelial cells were intact. The hypotonic lysis buffer contains 20% PBS, 80% deionized water, 100 µM Na₂VO₃, 1 mM PMSF, 5 mM iodoacetamide, 20 µg/ml aprotinin, 20 µg/ml leupeptin.

Detection of Tyrosine Kinase Activation in T Cells

An anti-phosphotyrosine Western blot was used to detect induced tyrosine kinase activity in T cells. Ten µl of 4x sample buffer was added to 30 µl of the T cell lysate (either stimulated or unstimulated). The lysate was boiled for 3 min and analyzed by 10% SDS-PAGE and Western blot using mAb 4G10 (Upstate Biotechnology Inc., Lake Placid, NY), using a technique described previously (10).

T cell lysate (either stimulated or unstimulated, 2 × 10⁶/200 µl) was precleared with 50 µl of 50% protein A-Sepharose 4B (PAS) at 4 °C for 1 h. 5 µl of antibody against p56^lck (Upstate Biotechnology Inc.) or p60^fyn (Upstate Biotechnology Inc.) and 50 µl of 50% PAS were added into the precleared lysate. The incubation was carried out at 4 °C for 2 h. The suspension was centrifuged and the supernatant removed. The PAS pellet was washed 3 times with wash buffer (1% Nonidet P-40, 150 mM NaCl, 0.1 mM Na₂VO₃, 25 mM Tris, pH 7.6) and once with kinase buffer (10 mM MnCl₂, 1 mM dithiothreitol, 20 mM HEPES, pH 7.2). The PAS pellet was resuspended in 30 µl of kinase buffer, and 10 µCi of [-³²P]ATP (Amersham) was added. The kinase reaction was performed at room temperature for 20 min and stopped by addition of 20 µl of 4x reducing sample buffer and boiled for 3 min. The reaction was analyzed by SDS-PAGE and autoradiography.

RESULTS

Freshly isolated normal human enterocytes and various human intestinal epithelial cell lines were stained by mAbs L12 and B9 and analyzed by flow cytometry. The greatest staining was seen in freshly isolated cells, the T84 and CaCO₂ cell lines. HT29 and DLD-1 stained with lesser intensity. Staining by B9 and L12 were comparable to each other in all cells and cell lines, as shown in Fig. 1A. Mean channel fluorescence is depicted since nearly 100% of the cells stain with these mAbs (Fig. 1B).

Membrane preparations from T84 epithelial cells were resolved on 7.5% SDS-PAGE under either reducing or nonreducing conditions and analyzed by Western blot, using L12 and B9 antibodies. As shown in Fig. 2, both antibodies specifically recognized a 180-kDa protein under both reducing and nonreducing conditions, and this protein band was not recognized by the isotype control, mouse IgG1. The same result was achieved when freshly isolated human enterocytes were used for these Western blots (data not shown). These data document that the 180-kDa protein is a membrane protein, since it was present in the membrane fractions of T84 cells (Fig. 2.) and was demonstrated by surface staining with mAbs B9 and L12 (Fig. 1B).

The 180-kDa protein, purified on a mAb B9 antibody affinity column from T84 cell lysates, was analyzed by SDS-PAGE under reducing conditions followed by Western blotting using mAb L12 B9 or an isotype control IgG1. The result is shown in Fig. 3. By Western blot, both antibodies recognized the 180-kDa epithelial cell membrane protein purified by the mAb B9 affinity column, although the intensity of the band is greater in the mAb B9 Western blot. These findings suggest that the molecule recognized by mAb B9 expresses the epitope recognized by mAb L12 either on the same molecule or on associated molecules.

The 180-kDa protein was purified by mAb B9 affinity column from ³⁵S-labeled T84 cells and was treated with N-glycanase, O-glycanase, and neuraminidase. The glycanase-treated 180-kDa protein was resolved on SDS-PAGE, transferred to nitrocellulose membrane sheets, and exposed on XAR-5 film. A significant molecular weight shift was only observed after N-glycanase treatment (Fig. 4A), resulting in more than a 50% reduction in the molecular mass of the protein from 180 to 76 kDa, indicating that the 180-kDa protein is heavily N-glycosylated. Due to the glycosylated nature of this protein, it was identified as gp180. Treatment with N- and O-glycanase together resulted in the resolution of hazier band of similar molecular mass, most likely reflecting the interaction of the three enzymes.

In the same experiment, the membranes were subjected to mAb L12 or B9 Western blot. As shown in Fig. 4B, gp180 treated by either O-glycanase, neuraminidase, or no enzyme control had no effect on the ability of either mAb L12 or B9 to recognize the 180-kDa protein. However, N-glycanase treatment not only resulted in a molecular weight shift as seen in Fig. 4A but also a loss of the epitope for both mAb L12 and B9 (i.e. the protein was no longer recognized by either of the antibodies by Western blot). This finding suggested that both mAbs recognize either a carbohydrate epitope or an epitope defined by both carbohydrate and protein.

Previous immunohistochemical studies have demonstrated that mAb L12 and B9 stain the apical side of the epithelium brighter than the basolateral side (8). Based on previous work (17) reporting that GPI-anchored proteins are typically apically sorted, the expression of gp180 was studied after treatment with PIPLC. First, surface staining by mAb L12 or B9 decreased significantly after T84 cells were treated by PIPLC with about a 50% reduction achieved (Fig. 5). PIPLC had no effect on the staining for MHC class I. Second, the supernatant generated from PIPLC treatment of T84 cells contained detectable gp180. This was demonstrated by mAb B9 Western blot analysis of supernatant generated by PIPLC treatment of T84 cells compared with that of control without enzyme (Fig. 6A). Furthermore, gp180 in PIPLC-generated T84 supernatant could be recovered by immunoaffinity purification using mAbs L12 or B9 but not a mouse IgG1 control antibody (Fig. 6B).

Soluble gp180 was generated from the supernatant of PIPLC-treated T84 cells. This supernatant was then incubated with T cells or non-T cells, followed by the addition of a mild cross-linker, DTBP. The cells were then washed, lysed, and analyzed by SDS-PAGE under reducing conditions that cleave the disulfide linkages formed by DTBP (i.e. releases the bound protein from its ligand), and subjected to a mAb B9 Western blot. The results (Fig. 7) demonstrate a 180-kDa band present only in T cells incubated with the PIPLC supernatant, not in non-T cells, suggesting that gp180 is capable of associating with these cells.

Previous studies from this lab have shown that p56^lck could be activated when T cells were co-incubated with either freshly isolated normal human intestinal epithelial cells or a malignant intestinal epithelial cell line, DLD-1 (10). These studies also suggested that T cell/epithelial cell interactions are mediated in part by CD8 molecules on T cells (10). We tried to determine whether CD8 molecules were involved in the binding of gp180 to T cells. PBT cells preincubated with either medium, OKT8 (mAb to CD8), or OKT4 (mAb to CD4) were co-incubated with DLD-1 cells, or DLD-1 cells pretreated with mAb B9, and/or L12 were co-incubated with PBT cells. The cell lysates were analyzed by SDS-PAGE followed by an anti-phosphotyrosine Western blot. As shown in Fig. 8A, mAb OKT8, mAb B9, or mAb L12 (although results with this mAb were more variable) resulted in significant inhibition of epithelial cell-induced tyrosine kinase activation (as evidenced by a significant reduction in protein tyrosine phosphorylation in the anti-phosphotyrosine Western blot) whereas anti-CD4 did not. Since our previous data (10) showed that activation of CD8-associated p56^lck, an Src-like tyrosine kinase, in T cells was a required early event in epithelial cell-induced CD8⁺ suppressor T cell activation and proliferation, our current result supports the possibility that gp180 and CD8 are the key elements involved in T cell-epithelial cell interactions in our co-culture system. To further confirm this, two murine T cell hybridoma cell lines, 3G4 and 3G8 (from Dr. Burakoff, Dana-Farber, Boston, MA) (14, 15), were used. Both 3G8 and 3G4 cells were generated from a single murine T cell hybridoma, transfected with either full-length human CD8 cDNA (3G8) or with full-length human CD4 cDNA (3G4). Murine p56^lck in these cells is capable of associating with the intracytoplasmic tails of human CD8 or CD4. When these cells were incubated with intestinal epithelial cells (10) or purified gp180 (Fig. 8B), an increase in protein tyrosine kinase activity (including a 56-kDa band) was observed only in 3G8 cells and not in 3G4 cells. Such a signaling event induced by gp180 was blocked by the anti-gp180 mAb B9 (data not shown). The 56-kDa protein appears to be p56^lck as evidenced by our previous studies (10) and by kinase assays immunoprecipitating p56^lck (Fig. 9A). These data support the concept that gp180 binds to T cells, most probably via CD8, activating p56^lck.

These findings were extended in studies utilizing peripheral blood T cells (PBT). Isolated gp180 was co-cultured with PBT cells for 2 min, followed by anti-lck or anti-fyn immunoprecipitation. Kinase activity was directly measured in an in vitro kinase assay. As seen in Fig. 9A, activation of lck was detected but activation of fyn (Fig. 9B), a tyrosine kinase able to associate with the TcR (18), was not. These data suggest that gp180 is capable of binding to CD8 and activating lck but does not appear to bind to the TcR, unlike the conventional CD8 ligand, class I MHC.

DISCUSSION

Consistent exposure to exogenous antigen has unique effects on an immune response. By definition then, mucosal immune responses should be distinct from those in antigen pristine environments, such as the systemic immune system. Recent studies from a number of laboratories support such a concept. Antigen administered orally most often results in the induction of systemic tolerance, and this phenomenon has been the basis for new therapeutic approaches in a number of autoimmune disorders (19-21). Oral tolerance may be mediated by several mechanisms (e.g. induction of anergy, activation of suppressor T cells, secretion of suppressive cytokines), but early studies supported a role for CD8⁺ suppressor T cells (2-7). That is, the transfer of CD8⁺ splenic T cells from an orally tolerized animal to a naive one would transfer antigen-specific nonresponsiveness (2-4). The mechanism whereby CD8⁺ T cells were activated in this model has not been clearly defined. Several laboratories have proposed a scenario where the manner by which antigen is handled in the intestine would dictate the type of immune response generated. Specifically, it has been suggested that the intestinal epithelial cell is a key regulator of mucosal immune responses, acting as an antigen-presenting cell (8, 9, 22). Normal intestinal epithelial cells constitutively express class II molecules (small intestine > large intestine) and are capable of processing and presenting exogenous antigens to primed T cells in mouse, rat, and man (8, 9, 23, 24). However, in contrast to conventional antigen-presenting cells, intestinal epithelial cells appear to selectively activate CD8⁺ suppressor T cells (8, 9, 23, 24). Conventional restriction elements, class I and class II, do not appear to regulate this event as blocking mAbs to such molecules fail to inhibit proliferation of the CD8⁺ T cells (8, 10). CD8 itself, however, does appear to be involved since antibodies to this surface molecule inhibit IEC-induced T cell proliferation (8, 10). Furthermore, the activation of CD8-associated p56^lck is a necessary but not sufficient event required to generate this response (10). These findings led to the suggestion that nonclassical class I molecules (class Ib), such as CD1d in man and TL in mouse, both expressed on intestinal epithelium, might regulate IEC-T cell interactions (25-28). Although antibodies to CD1d can inhibit proliferation in an IEC-T cell co-culture system (26), they fail to inhibit the activation of p56^lck (29), and there are no data to support the ability of CD1d to associate with CD8 itself.²

To define an epithelial cell-surface antigen capable of regulating the activation of CD8⁺ T cells, we generated a series of mAbs against normal epithelial cells and screened them both for their ability to inhibit CD8⁺ T cell proliferation in IEC-T cell co-cultures and for their epithelial specificity. Two mAbs, B9 and L12, were identified in the initial screen and were chosen for further characterization. Both mAbs were selective in their inhibition of CD8⁺ T cells, that is they failed to inhibit conventional MLR cultures or the proliferation of CD8⁺ T cells in response to mitogen. However, they did inhibit proliferation of CD8⁺ T cells activated in normal intestinal epithelial cell-peripheral blood T cell co-cultures as well as lamina propria T cell proliferation induced by IEC (30). Comparable to the results seen with PBT cells, the latter system (lamina propria lymphocyte) is not restricted by class I or class II MHC which again supports the existence of novel regulatory elements expressed on IEC (30).

Such regulatory elements might be present in other sites as well. The distribution of B9 and L12 in various tissues is of interest. mAb B9 stains epithelial cells in the intestine from the stomach to rectum with expression greater in the villus or surface epithelium than in the crypts (villus > crypt). It also stains thymic epithelium and syncytiotrophoblast cells at the maternal/fetal interface but does not stain keratinocytes, squamous epithelium of the esophagus or columnar airway epithelium. The presence of a molecule recognized by mAb B9 in the thymus and placenta may point to an important immunoregulatory function for gp180. However, while we have identified a 180-kDa band in placental lysates by mAb B9 Western blot,³ we have not identified the molecule in the thymus recognized by this mAb. mAb L12 also stains intestinal epithelium but is more uniform in distribution and can be expressed at greater density by immunohistochemical analysis. However, unlike mAb B9, it stains airway epithelium and fails to stain thymic epithelium. Thus these mAbs appear to be recognizing distinct structures. On a cellular level the staining is comparable with apical as well as basolateral staining by both mAbs. Clearly only the basolaterally expressed molecule would be capable of interacting with T cells within the epithelium (intraepithelial lymphocyte) or the lamina propria. Interestingly the majority of intraepithelial lymphocytes are CD8⁺ (31, 32), and to date no molecule has been identified that either recruits or maintains such CD8⁺ T cells in the epithelium. There is no obvious rationale for the apical expression of gp180, although the same holds true for apically expressed class II MHC molecules on these cells.

The concept that gp180 on intestinal epithelial cells can interact with CD8 on CD8⁺ T cells has been suggested by our previous studies (10) and is strengthened by our current data showing that purified gp180 can activate p56^lck tyrosine kinase in a murine T cell line transfected with human CD8 cDNA but not the same cell line transfected with human CD4 cDNA (Fig. 8B). Furthermore, we have recently documented that a CD8-Fc fusion protein was capable of binding to purified gp180 coated on enzyme-linked immunosorbent assay plates.³

Given the nature of the requirements for immune activation in the intestine, it makes sense that mucosal T cells would respond to a distinct set of restriction elements and accessory molecules. The question remains as to what gp180 truly represents, a novel restriction element or adhesion/accessory molecule. In the former case gp180 would act like class I, presenting peptide to the TcR as well as binding to and cross-linking CD8. In the latter case, gp180 would serve just to bind to CD8 resulting in activation. Evidence for the latter hypothetical model exists with the fact that while gp180 activates CD8-associated p56^lck, it is incapable of promoting T cell proliferation, even in the presence of interleukin-2 (data not shown), and it fails to activate the TcR-associated kinase fyn. Furthermore the structure of gp180, >50% N-linked sugars and preliminary sequence data documenting that gp180 is a novel molecule with some homology to adhesion molecules in the Ig supergene family, speaks more for an adhesion molecule than a class I-like molecule. That this molecule may associate with classical or nonclassical class I molecules is suggested by the finding that upon immunoprecipitation we can see an associated 45-kDa molecule co-precipitated (with mAb L12), and preliminary studies suggest an association of gp180 with CD1d (29). Clearly formal characterization awaits the cloning and expression of gp180. However, whatever the final outcome, the existence of such a molecule on intestinal epithelial cells speaks to novel mechanisms of immunoregulation in the gut. Understanding such mechanisms will allow for manipulation of mucosal responses to eventuate in positive (vaccination) as well as negative (tolerance induction) outcomes for orally administered antigen.