The Cell-Cell Adhesion Molecule Carcinoembryonic Antigen-Related Cellular Adhesion Molecule 1 Inhibits IL-2 Production and Proliferation in Human T Cells by Association with Src Homology Protein-1 and Down-Regulates IL-2 Receptor¹

Carcinoembryonic Ag-related cellular adhesion molecule 1 (CEACAM1)³ or CD66a is a type I membrane protein and cell-cell adhesion molecule expressed in epithelial and a variety of hemopoietic cells (1). CEACAM1 is expressed as four major isoforms, including three or four Ig-like ectodomains with short (e.g., CEACAM1-4S) or long (e.g., CEACAM1-4L) cytoplasmic domains (2). The distal N-terminal domain, an IgV-like domain, mediates its cell-cell adhesion properties (3). The 73-aa-long cytoplasmic domain contains two immunoreceptor tyrosine-based inhibitory motifs (ITIM) that have been shown to mediate inhibitory functions in epithelial cells (4). The long cytoplasmic domain has been shown to associate with actin and signal via the rho family of GTPases (5). Src family kinases have been reported to phosphorylate the Tyr residues in the ITIM (6, 7). The inhibitory activity of the long cytoplasmic domain is most likely mediated by the binding of Src homology protein-1/2 (SHP-1/2) via its Src homology 2 domains to the phosphorylated ITIM (4, 8). A critical Ser residue (S-503 in rat CEACAM1-4L) has also been identified in the long cytoplasmic domain that controls signaling via the insulin receptor in rat hepatocytes (9, 10). The 14-aa-short cytoplasmic domain lacks Tyr residues, but can be phosphorylated by protein kinase C on Ser and Thr residues (11) and binds calmodulin (CaM) (11) and both actin and tropomyosin (12). Recently, we have shown that CEACAM1-4S reverts breast cancer epithelial cells to a normal morphology and mediates lumen formation in a three-dimensional culture (13).

CEACAM1 functions as the cell surface receptor for mouse hepatitis virus in mouse (14) and a bacterial receptor for opacity-associated (Opa)-positive neisseriae strains (15) and for Haemophilus influenzae (16) in humans. CEACAM1 is up-regulated in activated mouse and human T cells (17, 18, 19), suggesting a role in immunity. However, these studies have provided conflicting data as to its role as either an activating or inhibitory molecule. The binding of Neisseria to CEACAM1 on activated human T cells was shown to arrest further activation and proliferation of T cells, providing a case for inhibition (20). However, in intestinal T cells, a case has been made for both activation (21) and inhibition (22), and decidual lymphocytes that express CEACAM1 are inhibited by extravillous trophoblast cells that also express CEACAM1, perhaps leading to protection of embryonic tissue from immune attack (23). These observations, plus the unexplained role of a homophilic cell-cell adhesion in activated T cells, prompted us to further investigate the activity of CEACAM1 and its long and short cytoplasmic domain isoforms in model systems. We demonstrate that the time course of expression of CEACAM1 vs IL-2Rα in PBMCs follows reciprocal kinetics, that the -4L, but not the -4S isoform inhibits IL-2 production in transfected Jurkat cells, and that the -4L, but not the -4S isoform inhibits proliferation in transfected Kit-225 cells.

Human PBMC were purified by Ficoll/Hypaque gradient centrifugation and filtration through glass wool from whole blood or discard filters used to collect platelets in the blood donor center. Purified PBMCs were stimulated once with 5 μg/ml PHA plus 50 U of IL-2 (Roche, Basel, Switzerland). Cells were split every 2–3 days and supplemented with 100 U/ml IL-2.

Jurkat cells were from American Type Culture Collection (Manassas, VA), and Kit-225 cells were provided by J. Siegel (Center for Biologics Evaluation and Research, Food and Drug Administration). Stable cell lines were obtained by transfection with human CEACAM1-4L or -4S (12, 13). Expression levels of the two isoforms were equivalent, as measured by FACS and Western blot analysis (data not shown). A CEACAM1-4L/enhanced green fluorscent protein (eGFP) fusion protein was constructed by inserting eGFP after the C terminus of CEACAM1-4L with an intervening Gly₄ linker and stable tranfectants established. Expression levels were determined by FACS (see Fig. 5).

Anti-CD3ε was from BD Biosciences (San Diego, CA); rabbit anti-ZAP-70, anti-SHP-1, anti-SHP-2, anti-Jak 3, anti-IL-2R, and PY20 were from Santa Cruz Biotechnology (Santa Cruz, CA); 4G10 was from Upstate Biotechnology (Lake Placid, NY); anti-IL-2Rα was from BioSource International (Camarillo, CA); and anti-IL-2Rβ and anti-IL-2Rγ were from BD PharMinigen (San Diego, CA). Anti-CEACAM1 T84.1 (24) and rabbit Ab 22-9 to the long cytoplasmic domain (12) have been previously described. Anti-CEACAM1 4/3/17 was a gift from F. Grunert (VacGene, Freiburg, Germany) (25, 26). Confocal microscopy was performed on a Zeiss (Oberkochen, Germany) model 310.

CEACAM1 and IL-2Rα on PBMCs were stained with 1 μg/ml anti-CEACAM1 T84.1 or anti-IL-2Rα, followed by FITC-conjugated goat anti-mouse IgG (mIgG). Kit-225 cells were stained with FITC anti-IL-2Rα (2 μg/ml) and PE anti-IL-2Rγ (2 μg/ml). CEACAM1 and IL-2β were stained with T84.1 (1 μg/ml) or anti-IL-2Rβ (2 μg/ml), followed by Alexa 488 goat anti-mIgG.

PBMCs were grown in 96-well plates (2.5 × 10⁴cells/200 μL). Cells were treated with anti-CD28 (2.5 μg/ml), anti-CEACAM1 (T84.1 or 4/3/17, 2.5 μg/ml) ± anti-CD3ε, plus mIgG to provide equivalent Ab concentrations. After 3 days, 0.5 μCi of [³H]thymidine was added for 5 h and cells were harvested on glass fiber filters (Wallac (Gaithersburg, MD) Cell Harvester), and cpm incorporation was determined (Wallac 1450 Microbeta Trilux Counter). Proliferation of transfected Jurkat cells was assayed similarly with the addition of rabbit anti-mIgG (10 μg/ml) to induce cross-linking. Kit-225 cells were IL-2 depleted for 72 h, plated in 96-well plates (2 × 10⁴/150 μL), incubated with increasing amounts of IL-2, and, after 24 h, 0.5 μCi of [³H]thymidine was added for 24 h.

Activated PBMCs (2 × 10⁵ cells/200 μL) were incubated with biotinylated anti-CD3ε plus Texas Red-avidin (1 μg/ml, 4°C, 1 h), then with Alexa 488 T84.1 (40 min), and transferred to 37°C for 30 min to induce capping. For ZAP-70, cells were permeabilized with PBS, 0.01% saponin, 0.25% gelatin, and 0.1% Nonidet P-40 for 20 min at room temperature and stained with rabbit anti-ZAP-70 (1 μg/ml). Jurkat cells were incubated with Alexa 488 T84.1 and goat anti-mIgG (4°C,1 h) followed by rhodamine-cholera toxin B. Colocalization of CEACAM1-4L/eGFP with SHP-1 or SHP-2 was performed by permeabilizing as above and staining with mouse anti-SHP-1 or anti-SHP-2 and Texas Red goat anti-mIgG. Similarly, permeabilized cells were stained with 4G10 and Texas Red goat anti-mIgG. Kit-225 cells were treated with anti-IL-2Rβ or anti-IL-2Rγ (1 μg/ml, 4°C, 1 h), followed by staining with Alexa 488 T84.1 and Texas Red-avidin (1 μg/ml, 40 min).

PBMCs were stimulated with anti-CD3ε (0.1 μg/ml) plus anti-CD28 and/or anti-CEACAM1 4/3/17 or T84.1 (0.25 or 2.5 μg/ml), plus make-up mIgG to provide equivalent total concentrations. After 3 days, aliquots were assayed for IL-2 activity using the HT-2 indicator cell assay (detected with Alamar Blue; Trek, Medina, NY). Jurkat transfectants in 96-well plates (1 × 10⁵cells/200 μL) were treated with anti-CD3ε (1 μg/ml), anti-CD28 (2.5 μg/ml), and/or anti-CEACAM1 (T84.1, CEA66, or 4/3/17; 2.5 μg/ml), plus make-up mIgG, as before, and rabbit anti-mIgG (10 μg/ml) to induce cross-linking. After 2 days, aliquots were assayed for IL-2 release.

PBMCs grown in 100 U/ml IL-2 were treated for 10 min with 2 nM calyculin A and 40 pM cypermetherin, or PMA (20 ng/ml), or pervanadate (1/100 dilution of 10 mM Na₃VO₄ plus 1 M H₂O₂), and then with T84.1 (2 μg/ml, 2 min). Cells were lysed (50 mM Tris-HCl, pH 7.6, 150 mM NaCl₂, 2 mM EDTA, 1 mM PMSF, 1 μg/ml leupeptin, 1 μg/ml aprotinin, 1% Nonidet P-40, 1 mM Na₃VO₄), precleared with protein A-agarose plus rabbit IgG, and treated with rabbit anti-CEACAM1 22-9. Immunoprecipitated proteins were resolved by SDS gel electrophoresis and Western blotted with appropriate Abs. Kit-225 cells were IL-2 depleted for 72 h, treated with IL-2 (500 U/ml, 30 min), lysed, precleared with protein G-agarose, and incubated with PY20-agarose. Immunoprecipitated proteins were resolved by SDS gel electrophoresis and Western blotted with appropriate Abs.

Freshly purified PBMCs were activated with PHA and IL-2, and the kinetics of induction of CEACAM1 and IL-2Rα was determined (Fig. 1 A). Both receptors were barely detectable at the start of the treatment, with IL-2Rα increasing more rapidly than CEACAM1 through the first 3 days of treatment. IL-2Rα expression peaked at 3 days, at which point CEACAM1 expression reached less than 20% maximal induction. As CEACAM1 expression climbed from days 3 to 7–9 by 5-fold, IL-2Rα expression decreased by over 4-fold. Thus, the two receptors exhibit reciprocal kinetics, suggesting a possible causal relationship. Because IL-2R controls the response to IL-2, we investigated the IL-2 response in more detail.

PBMCs were stimulated with anti-CD3ε and/or anti-CD28 Abs in the presence or absence of anti-CEACAM1 Abs, and IL-2 production was measured (Fig. 1,B). The results show that either anti-CEACAM1 Ab T84.1 or 4/3/17 inhibit IL-2 production below maximal levels obtained with mIgG control-treated cells when stimulated with either of two levels of anti-CD3ε Ab. As expected, combined treatment with anti-CD3ε and anti-CD28 gave maximal IL-2 production, while addition of anti-CEACAM1 Ab 4/3/17 inhibited stimulation by 26%. In the case of stimulation with 0.50 μg/ml CD3ε, both T84.1 and 4/3/17 inhibited IL-2 production by 50%. The inhibition results are remarkable in that only low levels of cell surface CEACAM1 are found on fresh PBMCs (see Fig. 1 A).

PBMCs were treated as above, and proliferation assays were performed (Fig. 1 C). Combined treatment with anti-CD3ε and anti-CD28 produced the highest level of proliferation, which was inhibited 33% by ligation of CEACAM1 with T84.1. However, the effect of T84.1 treatment was minimal (∼10% inhibition) for cells treated with anti-CD3ε alone. Thus, CEACAM1 inhibition is most pronounced in maximally activated T cells.

If CEACAM1 exerts its inhibitory activity on the TCR signal transduction pathway that depends on multiple ITAM phosphorylations, then tyrosine phosphorylation of ITIMs in CEACAM1-L isoforms may oppose TCR signal transduction. To explore this possibility, a PBMC cell line was established and CEACAM1 ligated with anti-CEACAM1 Ab T84.1, followed by immunoprecipitation and Western blotting (Fig. 2). The cell line had equivalent expression levels of CEACAM1 as measured by both FACS and Western blotting (data not shown). When analyzed by PCR using isoform-specific primers (27), the cell line expressed equivalent levels of CEACAM1-4L and -3L, a minor amount of -4S, and no detectable -3S (data not shown). Controls included treatment with pervanadate, an inhibitor of tyrosine phosphatases, and PMA, known to stimulate critical protein kinase C in T cells, and immunoprecipitation with irrelevant rabbit IgG (data not shown). Because previous studies on rat CEACAM1 had indicated that Ser⁵⁰³ mediated its physiological effects (8, 10), we also included treatment with the pSer/pThr-phosphatase inhibitors cypermetherin and calyculin A. The results show that, except for pervanadate treatment, CEACAM1 is only tyrosine phosphorylated when CEACAM1 is ligated. Neither treatment with PMA nor the pSer/pThr phosphatase inhibitors alone was able to induce tyrosine phosphorylation on CEACAM1. The results with pervanadate treatment suggest that transient, reversible phosphorylation occurs in the absence of CEACAM1 ligation. Both PMA and pSer/pThr phosphatase inhibitor treatment affect the level of pTyr observed when CEACAM1 is ligated, raising the possibility that pSer plays a major role in controlling pTyr on CEACAM1. It should be noted that CEACAM1 occurs as two major ECD isoforms in these immunoprecipitates (IPs) (CEACAM1-4L and CEACAM1-3L), and that both isoforms contain ITIMs and are tyrosine phosphorylated. In addition, an 80-kDa phosphoprotein was coimmunoprecipitated with CEACAM1. Interestingly, a CEACAM1-associated phosphoprotein of the same size was previously observed to correlate with the growth-inhibitory activity of rat CEACAM1 in epithelial cells (28), but its identity is unknown. Tyrosine phosphorylation of this band is strongly induced in the absence of CEACAM1 ligation plus pervanadate, but reduced when cells are treated with T84.1 plus pervanadate. Thus, CEACAM1 may interact with at least one other protein that is tyrosine phosphorylated.

Because the above results demonstrated that CEACAM1 is phosphorylated on its ITIMs in activated PBMCs, we then determined whether SHP-1 was coimmunoprecipitated with CEACAM1. Indeed, SHP-1 was found to coimmunoprecipitate with CEACAM1 on activated PBMCs both before and after CEACAM1 ligation; however, there was a 2- to 3-fold increase in the SHP-1 band after CEACAM1 ligation (Fig. 2). We also Western blotted for ZAP-70, a major constituent of the activated TCR complex, and CaM, which we (12) and Obrink and coworkers (29) have previously shown to associate with the cytoplasmic domain of CEACAM1. Low levels of ZAP-70 were found associated with CEACAM1 before and after ligation with T84.1 with a 2-fold increase after ligation. The association of CEACAM1 with both SHP-1 and ZAP-70 is of special significance given that SHP-1-deficient mice have hyperactive TCR responses (30, 31) and ZAP-70 has been shown to be a target of SHP-1 (32, 33). The results for CaM were even more dramatic, with a 5-fold increase of binding after CEACAM1 ligation. The increased association with CaM after CEACAM1 ligation correlates with our previous finding that CaM inhibits the association of CEACAM1 with actin and tropomyosin (12). Because T cell activation involves Vav and cytoskeletal reorganization (34), CEACAM1 may play a role in this pathway too. Indeed, Vav is also coimmunoprecipitated with CEACAM1 (data not shown).

To probe the role of CEACAM1 cytoplasmic domain isoforms, CEACAM1-transfected Jurkat cells were used as a model system. Cell lines were selected by FACS for production of equivalent cell surface levels of CEACAM1 isoforms and shown to have equivalent amounts of CEACAM1 protein by Western blotting (data not shown). Jurkat cells produce, but do not respond to IL-2, and do not normally express CEACAM1 unless treated with PMA or combined treatment with anti-CD3ε/CD28 Abs for >2 days (data not shown). In vector-transfected controls, IL-2 production is minimal for mIg-, anti-CD3ε-, and anti-CD28-treated cells. However, combined treatment with anti-CD3ε and anti-CD28 results in a substantial increase in IL-2 production (Fig. 3,A). Addition of anti-CEACAM1 Ab T84.1 to the vector control cells results in a small, but statistically insignificant reduction of IL-2. When Jurkat/CEACAM1-4L cells were treated with anti-CD3ε and anti-CD28 in the presence or absence of anti-CEACAM1 Ab, a 62% reduction in IL-2 production was observed (Fig. 3 A). The results with CEACAM1-4S-transfected cells showed no or minimal inhibition, demonstrating that the -4S isoform is not inhibitory in this assay.

When control vector-transfected Jurkat cells were treated with anti-CD3ε or anti-CD28 Abs, cell proliferation as measured by [³H]thymidine incorporation was in the range of 17–21,000 cpm (Fig. 3 B) with either no change or a slight increase observed for CEACAM1-4L-transfected cells when treated with three different anti-CEACAM Abs. However, when CEACAM1-4S-transfected cells were treated with anti-CEACAM1 Abs plus either anti-CD3ε or anti-CD28 Abs, a reproducible inhibition of cell proliferation was observed (10–15%). When vector-transfected or CEACAM1-transfected cells were treated with both anti-CD3ε and anti-CD28 in the absence of anti-CEACAM1 Abs, cell proliferation levels were equivalent to those found in the groups above. However, when these groups (both anti-CD3ε and anti-CD28) were treated in the presence of anti-CEACAM1 Abs, the vector control production was uniformly lowered compared with single treatments with anti-CD3ε or anti-CD28 alone. This reduction in baseline cell proliferation is most likely due to the aforementioned de novo production of CEACAM1. Thus, we compared the added effects of CEACAM1 isoforms introduced by transfection vs the vector controls. All three anti-CEACAM1 Abs reduced cell proliferation for both isoforms, 22–50% inhibition for CEACAM1-4L, and 43–54% inhibition for CEACAM1-4S. These results suggest that in this assay, CEACAM1-4S is a more potent inhibitor than CEACAM1-4L.

CEACAM1 may mediate its inhibitory activity by associating with the CD3 signaling complex. This is an attractive possibility because phosphorylation of its ITIMs would allow recruitment of the inhibitory phosphatases SHP-1/2, which, in turn, could counteract phosphorylation of the ITAMs on the CD3 complex or prevent docking of downstream effectors such as ZAP-70. Indeed, when a PBMC cell line was treated with biotinylated anti-CD3ε and avidin to induce capping of CD3, and counterstained with anti-CEACAM1 Ab T84.1, cocapping was demonstrated (Fig. 4, a–f). Capping of CD3 in cholesterol-rich rafts is a hallmark of TCR engagement and T cell activation (35). The fact that CEACAM1 cocaps with CD3 suggests that it too is recruited to, or constitutively resides, in the same lipid compartment (see later).

The expression pattern of a CEACAM1-4L/eGFP fusion protein, like the parent CEACAM1-4L transfectants, is a beadlike pattern across the membrane, suggesting that the protein resides in discrete membrane domains (data not shown). CD3ε has an identical staining pattern across the cell membrane (data not shown). When CD3 is capped with biotinylated anti-CD3ε plus avidin, a large portion of the CEACAM1-4L/eGFP fusion protein is cocapped (Fig. 4, g–i). Because the cocapping results are similar to those obtained with PBMCs, we conclude that CEACAM1-4L faithfully associates with the TCR signaling complex in the Jurkat model system. We therefore, proceeded to determine whether CEACAM1-4L/eGFP also colocalized with SHP-1.

The CEACAM1-4L/eGFP-transfected Jurkat cells were treated with pervanadate, permeabilized, and stained for SHP-1, SHP-2, and pTyr (4G10). The results show good colocalization of CEACAM1-4L/eGFP with SHP-1, but not with SHP-2 (Fig. 4, j–r). The pattern shows intense staining between the cells, as expected for a homophilic cell-cell adhesion molecule. A similar staining pattern is observed for pTyr (4G10), suggesting that proteins recruited to the cell-cell membrane interface are highly phosphorylated on tyrosine. Based on the coimmunoprecipitation results obtained with PBMCs, it is likely that one of these proteins is CEACAM1. When the cells are not pretreated with pervanadate, little or no pTyr is detected with 4G10 and there is no colocalization of CEACAM1-4L/eGFP with SHP-1 or SHP-2 (data not shown).

The beaded expression pattern of CEACAM1-4L in transfected Jurkat cells suggested that it was localized to specific membrane microdomains. Given its association with cross-linked CD3 (Fig. 4, g–i), and the fact that cross-linked CD3 is recruited to lipid rafts (35), we speculated that CEACAM1 is constitutively found in lipid rafts. Using GM1 as a marker for cholesterol-rich lipid rafts (36), we performed cocapping studies on CEACAM1 and GM1 in CEACAM1-4L/Jurkat cells. Fig. 4, s–u, illustrates several cells caught in early and late stages of capping, showing that CEACAM1-4L is associated with the GM1-positive microdomains at all stages of the capping process. Similar results were obtained for CEACAM1-4L/eGFP-transfected cells and CEACAM1-4S-transfected cells (data not shown). We conclude that CEACAM1 expression is localized to lipid rafts and is in position to interact with CD3 and other signal transduction complexes that are recruited to lipid rafts.

To explore the mechanism responsible for our initial observation regarding the reciprocal kinetics of IL-2R and CEACAM1 expression on activated PBMCs, we transfected Kit-225 cells, a human T cell line that requires IL-2 for proliferation, with CEACAM1-4L and -4S. Initially, we demonstrated that these cells have no cell surface expression of CEACAM1 by FACS analysis, nor detectable levels of CEACAM1 mRNA by PCR analysis (data not shown). In contrast, these cells expressed high levels of all three subunits of the IL-2R before transfection with CEACAM1 isoforms. After transfection, six cell lines were derived and analyzed for the expression levels of CEACAM1 isoforms and IL-2R subunits (Fig. 5). One line was a vector-transfected control, two lines had high expression of the -4L isoform (lines 4L-1 and 4L-2), one line had low expression of the -4L isoform (4L-3), and two lines had high expression of the -4S isoform (lines 4S-1 and 4S-2). For each line, the expression level of IL-2Rα was unchanged vs the vector-transfected controls (Fig. 5,B). However, the expression levels of both IL-2Rβ and IL-2Rγ subunits were reduced by 50% in both CEACAM1-4L and -4S transfectants in which high levels of CEACAM1 were expressed (Fig. 5, C and D). In the case of the low expressing line for CEACAM1-4L, the IL-2Rβ and γ subunits were reduced less.

To test the hypothesis that CEACAM1 isoforms affect the IL-2 response in these cells, proliferation was measured for each of the transfectants after prior IL-2 depletion (Fig. 6,A). The dose-response curve shows a 50% maximal response for the vector controls and CEACAM1-4S transfectants at 5.0 U/ml IL-2, while that of the CEAMCAM1-4L transfectants is 0.8 U/ml. This >6-fold reduction in the 50% maximal response level for the CEACAM1-4L transfectants does not appear to depend on the absolute levels on the IL-2R subunits, because the reduction in levels was similar for the CEACAM1-4S transfectants (Fig. 6 A). Thus, we conclude that transfection with either isoform reduces the absolute levels of IL-2Rβ and γ subunits, but only in the case of the CEACAM1-4L transfectants do we observe a shift in the 50% maximal proliferative response to IL-2.

We next asked whether the inhibition of the proliferation response of the CEACAM1-4L transfectants was due to a direct association of this isoform with IL-2R. As shown in Fig. 6 B, CEACAM1-4L colocalizes with both the IL-2Rβ and γ subunits, and as shown for other cells, is found in discrete membrane domains and at cell-cell boundaries when two cells are found together. In one report, the localization of IL-2Rα, but not IL-2Rβ and γ to membrane rafts has been reported (37), but in another report all three subunits were localized to lipid rafts (38). In addition, one report showed that Abs to GM1, which is localized to lipid rafts, inhibited IL-2-mediated proliferation (37). Based on our study, we predict that IL-2R clustering will lead to recruitment of the β and γ subunits into lipid raft domains, because CEACAM1 appears to reside within these microdomains, and the GM1 inhibition data are consistent with this conclusion.

To further demonstrate that CEACAM1 isoforms may associate with IL-2R receptor subunits, we performed immunoprecipitation studies (Fig. 7, A and B). When cell lysates were immunoprecipitated with anti-CEACAM1 Ab T84.1, both β and γ subunits of IL-2R were detected. When cell lysates were immunoprecipitated with either anti-IL-2Rβ or γ Abs and IPs probed with T84.1, both isoforms of CEACAM1 were detected, depending on the transfectant analyzed. However, in both experiments, the relative amounts of coimmunoprecipitated proteins were low compared with total protein detected in equivalent amounts of lysates.

Because only the CEACAM1-4L isoform was able to inhibit IL-2-dependent proliferation in Kit-225-transfected cells, we decided to further probe its effect on Jak3, the immediate downstream mediator of signal transduction of IL-2R. Furthermore, a study by Leon et al. (39) had shown that SHP-1 expression had a profound effect on Jak3 signaling linked to the IL-2R. Thus, we reasoned that recruitment of SHP-1 to CEACAM1-4L phosphorylated on its ITIM could play a similar inhibitory role in the IL-2R, as was shown for TCR ligation in either activated PBMCs or CEACAM1-4L-transfected Jurkat cells. When phosphotyrosine-containing proteins were immunoprecipitated from cells either treated or not with IL-2 and the separated proteins probed for the presence of Jak3, the level of phosphorylated Jak3 was reduced by 2.5- to 3.0-fold in two lines of the CEACAM1-4L transfectants vs the vector control (Fig. 7, C and D). Thus, transfection with CEACAM1-4L can dramatically reduce the ability of IL-2R to signal via its most immediate associated kinase, Jak3. We conclude that even though the absolute levels of CEACAM1-4L immunoprecipitated with IL-2R are low (Fig. 7, A and B), they are sufficient to affect the levels of phosphorylated Jak3.

Our data lend further strength to the accumulating evidence that the role of CEACAM1 in activated T cells is inhibitory, rather than activating. Although this result is not surprising for CEACAM1-4L, which contains two ITIMs, it had not been formally proved by convincing biochemical data. We show in this study that activated PBMCs express the ITIM-containing isoforms of CEACAM1 that are both tyrosine phosphorylated and associate with their cognate inhibitory Src homology 2-containing phosphatase, SHP-1. However, in order for this association to be inhibitory, the complex must, in turn, associate with the TCR complex, the signal transduction pathway that leads to both IL-2 production and proliferation. In this respect, we have shown by cocapping experiments that CEACAM1 does indeed associate with the TCR complex, and that small amounts of ZAP-70 are coimmunoprecipitated with CEACAM1 from activated PBMCs. These results agree with studies by Mustelin and coworkers (33), who showed that SHP-1, but not SHP-2, can dephosphorylate ZAP-70 and inhibit T cell activation. In other studies, these investigators showed that SHP1 remains in the cytosol and requires activation for its inhibitory activity (32). We show in this study that translocation of SHP-1, but not SHP-2, to the plasma membrane depends on the expression of CEACAM1-4L. Thus, activation of SHP-1 may involve translocation to an appropriate ITIM-containing molecule that is phosphorylated and present in the plasma membrane.

At least three other studies also conclude that CEACAM1 expression leads to an inhibitory T cell response. In the case of human decidual lymphocytes that highly express CEACAM1, an inhibitory response was observed for both T and NKT cells (23). These authors speculated that the CEACAM1-CEACAM1 interactions between the lymphocytes and the extravillous trophoblasts (EVT) were responsible for the inhibitory activity. In the case of murine splenic T cells, CEACAM1 expression required activation with anti-CD3ε and anti-CD28 Abs, and ligation of CEACAM1 resulted in inhibition of both an allogenic MLR and a delayed-type hypersensitivity response (22). In the case of Neisseria Opa interactions with CEACAM1 on activated human T cells, the result is also inhibitory (20). Thus, it seems likely that CEACAM1 inhibition is tied to TCR signaling, and according to our data, especially when both the CD3 and CD28 pathways are engaged. The inhibition results in both decreased production of IL-2 and proliferation.

The inverse kinetics associated with CEACAM1 and IL-2R expression in activated PBMCs suggests that regulation of IL-2R levels also contributes to CEACAM1 inhibition. Indeed, we show that CEACAM1 directly associates with IL-2R and is responsible for its down-regulation. Because CEACAM1 is a cell-cell adhesion molecule, we speculate that when activated T cells proliferate and express maximal amounts of CEACAM1 in a confined space such as a lymph node, cell-cell contact increases, leading to inhibition of proliferation. Similarly, when activated T cells expressing CEACAM1 reach tissues like intestinal epithelium and EVT rich in CEACAM1, T cell inhibition is a likely outcome. In the case of the intestinal epithelium, this outcome sets an upper threshold to further activation in which foreign Ags abound. In the case of EVT, the inhibition protects embryonic tissue from maternal T cell attack. In the case of cancer cells, as shown in a recent study by Markel et al. (40), melanoma cells that express CECAM1 are protected from attack by activated T cells. However, in all of this work, no attempt was made to distinguish between the effects of the ITIM- vs the non-ITIM-containing isoforms of CEACAM1. In our model system in which CEACAM1-4L/eGFP was introduced into Jurkat cells, the -4L isoform accumulated at cell-cell boundaries, and when treated briefly with pervanadate to mimic activation, SHP-1 and phosphotyrosine-positive receptors (including CEACAM1) were also recruited to the cell-cell boundaries. This is further evidence that tyrosine phosphorylation and recruitment of SHP-1 are involved in the inhibitory activity of CEACAM1-4L, and suggests that the inhibitory activity is generated at the cell-cell contact sites where CEACAM1-4L accumulates. Thus, it makes sense that CEACAM1-4L exerts its inhibitory effect late in T cell activation when cell density is at a maximum.

Our studies demonstrate that the two cytoplasmic domain isoforms of CEACAM1 confer different phenotypes to T cells. In the model Jurkat system, CEACAM1-4L inhibited IL-2 production more effectively than CEACAM1-4S, while the opposite was true for cell proliferation. Thus, one may predict that the absolute levels of the two isoforms in activated T cells may fine tune the inhibitory response or allow different signaling pathways to operate under different conditions. The role of the two isoforms in controlling the expression of IL-2R subunits in Kit-225 cells is equivalent in that the IL-2Rβ and γ subunits, but not the α subunit, are strongly down-regulated. However, the IL-2 dose-response curves for proliferation are very different for the two isoforms, with only the -4L isoform conferring strong inhibition. Thus, the -4L isoform is directly tied to both the TCR and IL-2R in terms of T cell proliferation, while the -4S isoform only inhibits IL-2-independent proliferation.

The fact that CEACAM1 expression is associated with the down-regulation of IL-2Rα in PBMCs and not with IL-2Rα, but rather with the β and γ, subunits in Kit-225 cells, is intriguing. We conclude that the functional IL-2R is differentially regulated in Kit-225 cells vs PBMCs. Nonetheless, the end result is the same: less functional IL-2R. Despite the strong down-regulation of functional IL-2R, both PBMCs and Kit-225 cells still respond to IL-2, but as we show in this study, the CEACAM1-4L isoform is further capable of inhibiting Jak3 kinase phosphorylation, the critical downstream mediator of the IL-2R response. Thus, CEACAM1-4L is capable of both lowering the cell surface levels of functional IL-2R and its ability to activate T cells. This outcome is likely to have significant in vivo effects on the ability of T cells to respond to activating signals. The fact that CEACAM1-4S can lower cell surface levels of functional IL-2R without affecting signal transduction, as measured in the proliferation assay, suggests that this isoform can modulate T cells in more subtle ways, for example making them dependent on higher levels of IL-2 for continued activation.

Immunoprecipitation data support the idea that CEACAM1-4L can directly associate with the IL-2R and inhibit Jak3 phosphorylation presumably via the SHP-1-CEACAM1-4L association. In agreement with this idea, increased SHP-1 expression has been shown to inhibit IL-2-mediated cell proliferation in HUT-78 cells that otherwise exhibit decreased SHP-1 expression and increased Jak3 phosphorylation (39). Also, Waldmann and coworkers (41) found that SHP-1 is recruited to IL-2R and inhibits Jak1 and Jak3 phosphorylation in NIH 3T3 cells transfected with all three subunits of the IL-2R.

In summary, the mechanism of inhibition of IL-2 production and cell proliferation in T cells by the CEACAM1-4L isoform clearly involves phosphorylation on its ITIMs and recruitment of SHP-1 at the level of the TCR, especially during coligation with CD28. When either CEACAM1 isoform is expressed in an IL-2-dependent cell line, the signaling subunits of the IL-2R are strongly down-regulated, and, in the case of CEACAM1-4L, result in decreased cell proliferation. The -4S isoform can down-regulate IL-2R expression, but does not inhibit IL-2 production in an IL-2-independent cell line, nor does it inhibit proliferation in an IL-2-dependent cell line. However, it does inhibit cell proliferation in the IL-2-independent cell line, suggesting that it has inhibitory activity in some signaling pathways, but not in others.