The Critical Role of Residues 43R and 44Q of Cell Adhesion Molecules-1 in the Protection from Killing by Human NK Cells This information is current as of October 1, 2021. Gal Markel, Raizy Gruda, Hagit Achdout, Gil Katz, Morris Nechama, Richard S. Blumberg, Robert Kammerer, Wolfgang Zimmermann and Ofer Mandelboim J Immunol 2004; 173:3732-3739; ;

doi: 10.4049/jimmunol.173.6.3732 Downloaded from http://www.jimmunol.org/content/173/6/3732

References This article cites 29 articles, 16 of which you can access for free at:

http://www.jimmunol.org/content/173/6/3732.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

by guest on October 1, 2021 • Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2004 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

The Critical Role of Residues 43R and 44Qof Carcinoembryonic Antigen Cell Adhesion Molecules-1 in the Protection from Killing by Human NK Cells1

Gal Markel,* Raizy Gruda,* Hagit Achdout,* Gil Katz,* Morris Nechama,* Richard S. Blumberg,† Robert Kammerer,‡§ Wolfgang Zimmermann,‡ and Ofer Mandelboim2*

The multifunctional carcinoembryonic Ag cell adhesion molecule (CEACAM)1 has recently become the focus of intense immunological research. We have previously shown that the CEACAM1 homophilic interactions inhibit the killing activity of NK cells. This novel inhibitory mechanism plays a key role in melanoma immune evasion, inhibition of decidual immune response, and controlling NK autoreactivity in TAP2-deficient patients. These roles are mediated mainly by homophilic interactions, which are mediated through the N-domain of the CEACAM1. The N-domain of the various members of the CEACAM family shares a high degree of similarity. However, it is still unclear which of the CEACAM family members is able to interact with CEACAM1 and Downloaded from what are the amino acid residues that control this interaction. In this study we demonstrate that CEACAM1 interacts with CEACAM5, but not with CEACAM6. Importantly, we provide the molecular basis for CEACAM1 recognition of various CEACAM family members. Sequence alignment reveals a dichotomy among the CEACAM family members: both CEACAM1 and CEACAM5 contain the R and Q residues in positions 43 and 44, respectively, whereas CEACAM3 and CEACAM6 contain the S and L residues, respectively. Mutational analysis revealed that both 43R and 44Q residues are necessary for CEACAM1 interactions. Implications for differential expression of CEACAM family members in tumors are discussed. The Journal of http://www.jimmunol.org/ Immunology, 2004, 173: 3732–3739.

he human carcinoembryonic Ag (CEA)3 protein family and the presence or absence of various signal transduction motifs. encompasses several forms of with different bio- It is not surprising, therefore, that these proteins are actively in- T chemical features. These proteins are encoded by 29 volved in numerous physiological and pathological processes. tandemly arranged on 19q13.2 (1). All CEA CEACAM1 is a transmembrane protein that can be detected on family genes have been classified into two major subfamilies, the some immune cells as well as on epithelial cells (1). Many differ-

CEA cell adhesion molecule (CEACAM) and the pregnancy-spe- ent functions were attributed to the CEACAM1 protein. It was by guest on October 1, 2021 cific glycoprotein subgroups (1). The CEACAM proteins, which shown that the CEACAM1 protein exhibits antiproliferative prop- are part of the larger Ig superfamily, include CEACAM1, -3, -4, -5, erties in carcinomas of colon (3), prostate (4), as well as other -6, -7, and -8. They share a common basic structure of sequentially types of cancer. Additional data support the central involvement of ordered different Ig-like domain(s) and are able to interact with CEACAM1 in angiogenesis (5) and metastasis (6). CEACAM1 each other. For example, it was reported that various CEACAM also has a role in the modulation of innate and adaptive immune proteins, such as CEACAM1 or CEACAM5, exhibit both ho- responses. We have recently provided substantial evidence that mophilic and heterophilic interactions (2). CEACAM1 homophilic interactions inhibit NK-mediated killing The various CEACAM proteins have different biochemical fea- activity independently of MHC class I recognition (7). This novel tures, such as anchorage to cell surface (GPI-linked, transmem- mechanism plays a pivotal role in the inhibition of activated de- brane or secreted forms), length of cytoplasmic tail (long or short), cidual lymphocytes in vitro (8) and most likely also in vivo after CMV infections (8). The CEACAM1 homophilic interactions are probably important in some cases of metastatic melanoma, as in- *Lautenberg Center for General and Tumor Immunology, Hebrew University Hadas- sah Medical School, Jerusalem, Israel; †Gastroenterology Division, Brigham and creased CEACAM1 expression was observed on NK cells derived Women’s Hospital, Harvard Medical School, Boston, MA 02115; ‡Department of from some patients compared with healthy donors (7). Indeed, it Urology, Ludwig Maximilians University, Munich, Germany; and ¤Institute for Mo- lecular Immunology, GSF National Research Center for the Environment and Health, was recently reported that there is a clear association of Munich, Germany CEACAM1 expression on primary cutaneous melanoma lesions Received for publication April 23, 2004. Accepted for publication July 1, 2004. with the development of metastatic disease and poor survival (6). The costs of publication of this article were defrayed in part by the payment of page In addition, we have recently demonstrated the cardinal role of charges. This article must therefore be hereby marked advertisement in accordance CEACAM1-mediated inhibition in maintaining NK self-tolerance with 18 U.S.C. Section 1734 solely to indicate this fact. in TAP2-deficient patients (9). Additional reports indicate that 1 This work was supported by research grants from the Israel Science Foundation, the Israel Cancer Association (20042031-B), the Israel Cancer Research Foundation, and CEACAM1 engagement either by TCR cross-linking with mAb or the Cancer Research Institute. by Neisseria gonorrhoeae Opa proteins inhibits T cell activation 2 Address correspondence and reprint requests to Dr. Ofer Mandelboim, Lauten- and proliferation (10, 11). berg Center for General and Tumor Immunology, Hebrew University Hadassah The CEACAM1 protein interacts with other CEACAM protein Medical School, P.O. Box 12272, Jerusalem 91120, Israel. E-mail address: [email protected] family members, such as CEACAM1 itself and CEACAM5 (12). 3 Abbreviations used in this paper: CEA, carcinoembryonic Ag; m, murine. It was suggested that the binding site of human CEACAM1 is CEACAM, CEA cell adhesion molecule. located at the N-terminal Ig-V-type domain of the CEACAM1

Copyright © 2004 by The American Association of Immunologists, Inc. 0022-1767/04/$02.00 The Journal of Immunology 3733

protein, and in particular, amino acids 39V and 40D and the salt GAATA (including the EcoRI restriction site). Amplified fragments were bridge between 64R and 82D were predicted to play an important mixed and fused by an additional PCR that was performed with the 5Ј- Ј role in this binding (13). Most amino acid sequences of the N- HindIII primer and the 3 -EcoRI primer. The construct was cloned into pcDNA3 vector (Invitrogen Life Technologies, Carlsbad, CA) and perma- terminal domain of CEACAM1, -3, -5, and -6 are identical, and all nently transfected to 721.221 cells. of the predicted binding residues are conserved among the four For the generation of BW cells expressing the chimeric CEACAM1-␨ proteins. Thus, it is expected that all these proteins would interact protein, we used the same technique. The extracellular portion of the hu- with each other. However, there is still no conclusive evidence for man CEACAM1 protein was amplified by PCR using the 5Ј primer CCCAAGCTTGGGGCCGCCACCATGGGGCACCTCTCAGCC (in- heterophilic interaction between CEACAM1 and CEACAM6, for cluding HindIII restriction site) and the 3Ј primer GTAGCAGAGAG example. This is of particular importance, because in certain tu- GTGAGAGGCCATTTTCTTG (including the first nine nucleotides of mors the CEACAM1 protein is down-regulated, followed by up- the mouse ␨-chain transmembrane portion). The mouse ␨-chain was regulation of CEACAM6 protein expression (14, 15). amplified by PCR using the 5Ј primer CTCTCACCTCTCTGCTACT In this work, we demonstrate the inability of CEACAM1 to bind TGCTAGATGGA (including last nine nucleotides of human CEACAM1 extracellular portion) and the 3Ј primer GGAATTCCTTA CEACAM6. We also directly show that the presence of both res- GCGAGGGGCCAGGGTCTG (including EcoRI restriction site). The two 43 44 idues R and Q in the CEACAM1 is crucial for the homophilic amplified fragments were mixed, and PCR was performed with the 5Ј CEACAM1 interaction and that substitution of these residues with HindIII primer and the 3Ј EcoRI primer for generation of the CEACAM1-␨ the 43S and 44L residues that are present in CEACAM6 abolishes construct. The CEACAM1-␨ construct was cloned into pcDNA3 expres- sion vector (Invitrogen Life Technologies) and was stably transfected into the inhibitory effect. Importantly, the reciprocal substitution of 43S 44 43 44 BW cells. All transfectants were periodically monitored for expression by and L of CEACAM6 to the R and Q residues, respectively, staining with the appropriate mAb. results in the gain of inhibitory heterophilic interactions with the Downloaded from CEACAM1 protein. Thus, the dichotomy of CEACAM family Generation of 721.221 cells expressing mutated CEACAM1 or members by recognition of CEACAM1 is determined by the pres- CEACAM6 proteins ence of R and Q at positions 43 and 44. For generation of the mutated CEACAM proteins, we amplified by PCR two overlapping fragments of the . The upstream fragment was am- Materials and Methods plified by using a gene-specific5Ј-edge primer (including the HindIII re- Cells striction site) and an internal 3Ј primer bearing the mutation. The down- stream fragment was amplified using an internal 5Ј primer bearing the http://www.jimmunol.org/ The cell lines used in this work were the MHC class I-negative 721.221 Ј human cell line, the murine thymoma BW cell line that lacks expression of mutation and a gene-specific3-edge primer (including EcoRI restriction ␣ ␤ site). Next, both purified fragments were mixed together with the 5Ј-edge - and -chains of the TCR, and the NK tumor line YTS. Primary NK cells Ј were isolated from PBL using the human NK isolation and the au- primer and the 3 -edge primer to generate the mutated full-gene cDNA. All toMACS instrument (Miltenyi Biotec, Auburn, CA). For the enrichment of different mutants of the same CEACAM gene were generated using the CEACAM1-positive NK cells, isolated NK cells were further purified by same appropriate edge primers and different internal primers. The various depletion of CD16-positive NK cells, using the anti-CD16 mAb B73.1.1 cDNAs were then cloned into the pcDNA3 mammalian expression vector and the autoMACS instrument. NK cells were grown in culture as previ- and stably transfected into the .221 cell line. All transfectants were peri- ously described (16). CEACAM1-positive NK clones were identified by odically monitored for expression by staining with the appropriate mAb. For CEACAM1-RQ43,44SL, the 5Ј-CEACAM1 edge primer was flow cytometry using the anti-CEACAM1 mAb 5F4 and were tested for Ј inhibition in killing assays against .221/CEACAM1 cells. CCCAAGCTTGGGGCCGCCACCATGGGGCACCTCTCAGCC, the 3 - by guest on October 1, 2021 CEACAM1 edge primer was GGAATTCCTTACTGCTTTTTTACT Antibodies TCTGAATA, the 5Ј internal primer was GCCAACAGTCTAATTGTA GGA, and the 3Ј internal primer was TCCTACAATTAGACTGTTGCC. For The Abs used in this work were mAb Kat4c (DakoCytomation, Carpente- CEACAM1-R43A, the 5Ј internal primer was GATGGCAACGCTCAAAT ria, CA), directed against CEACAM1, -5, - 6, and -8; the anti-CD99 mAb TGTA, and the 3Ј internal primer was TACAATTTGAGCGTTGCCATC. 12E7; the rabbit polyclonal anti-CEACAM (DakoCytomation); and the For CEACAM1-Q44L, the 5Ј internal primer was ATGGCAACCGTCTA specific anti-CEACAM1 mAb 5F4 (10). Rabbit polyclonal Abs against ATTGTAG, and the 3Ј internal primer was CTACAATTAGACGGTTGC purified ubiquitin were used as the control. CAT. For CEACAM6-SL43,44RQ, the 5Ј-CEACAM6 edge primer was Ј Generation of CEACAM1-Ig fusion protein CCCAAGCTTGCCGCCACCATGGGACCCCCCTCAGCC, the 3 - CEACAM6 edge primer was GGAATTCCCTATATCAGAGCCAC The extracellular portion of the CEACAM1 protein was amplified by PCR CCTGG, the 5Ј internal primer was GGCAACCGTCAAATTGTAGGA, and using the following primers: the 5Ј primer CCCAAGCTTGGGGCCGC the 3Ј internal primer was TCCTACAATTTGACGGTTGCC. For CACCATGGGGCACCTCTCAGCC (including the HindIII restriction CEACAM6-S43R, the 5Ј internal primer was GATGGCAACCGTCTAAT site) and the 3Ј primer GCGGATCCCCAGGTGAGAGGC (including the TGTA, and the 3Ј internal primer was TACAATTAGACGGTTGC BamHI restriction site). A silent mutation, adenine885guanidine (no CATC. For CEACAM6-L44Q, the 5Ј internal primer was GATG change in glycine281) was performed by site-directed mutagenesis to can- GCAACAGTCAAATTGTA, and the 3Ј internal primer was cel the BamHI site in the amplified sequence. The production of the TACAATTTGACTGTTGCCATC. CEACAM1-Ig and CD99-Ig fusion proteins by COS-7 cells and purifica- tion on a protein G column were previously described (8, 17). The fusion Cytotoxicity assays proteins were periodically analyzed for degradation by SDS-PAGE. The cytotoxic activity of YTS and NK cells against various targets was Generation of transfectants assayed in 5-h [35S]Met release assays, as described previously (16). In The 721.221 cells expressing CEACAM1 and CEACAM6 proteins were experiments in which rabbit polyclonal Abs were included, the final con- ␮ generated as previously described (7). The CEACAM5 cDNA was sub- centration was 20 g/ml. In all cytotoxicity assays performed, spontaneous cloned into pcDNA3 vector. This construct was permanently transfected to release did not exceed 20% of maximal labeling. 721.221 cells. For the generation of 721.221 cells expressing the CEACAM6 protein fused to the tail of CEACAM1, we first amplified the Cross-linking of BW/CEACAM1 cells extracellular portion of the CEACAM6 without the GPI-anchoring se- quence using the 5Ј primer CCCAAGCTTGCCGCCACCATGGGAC BW/CEACAM1 cells (0.5 ϫ 105/well) were incubated with various CCCCCTCAGCC (including the HindIII restriction site) and the 3Ј primer amounts of Kat4c mAb on ice for1hin96-well, round-bottom microplates AATGGCCCCTCCAGAGACTGTGATCATCGT (including the first nine (Nalge Nunc, Rochester, NY). Treated BW/CEACAM1-␨ cells, present in nucleotides of the CEACAM1 transmembrane portion). The transmem- 200 ␮l of RPMI 1640 complete medium, were then cultured in 96-well, brane and tail of the CEACAM1 protein were amplified with the 5Ј primer flat-bottom microplates (Nalge Nunc) precoated with 1 ␮g/well sheep anti- GTCTCTGGAGGGGCCATTGCTGGCATTG (including the last nine nu- mouse IgG Abs (ICN Biomedicals, Costa Mesa, CA) for 24 h at 37¡C. cleotides of the CEACAM6 extracellular portion before the GPI anchor Supernatant was harvested, and the amount of murine IL-2 (mIL-2) was motif) and the 3Ј primer GGAATTCCTTACTGCTTTTTTACTTCT determined by ELISA. 3734 RESIDUES R43 AND Q44 DETERMINE RECOGNITION OF CEACAM1

Results cell system. BW cells were stably transfected with the extracellular CEACAM1 protein does not recognize CEACAM6 portion of the CEACAM1 fused to mouse ␨-chain (BW/ CEACAM1-␨) as described in Materials and Methods. The spe- Recent clinical studies demonstrated that surface expression of the cific functionality of BW/CEACAM1-␨ was assessed by cross- CEACAM1 protein on tumors is associated with poor prognosis in linking the CEACAM1 receptor using different amounts of melanoma and lung adenocarcinoma patients (6, 18). We have immobilized Kat4c mAb as described in Materials and Methods. shown that CEACAM1 homophilic interactions confer protection Engagement of the CEACAM1 protein elicited the synthesis and from human NK-mediated cytotoxicity (7). We have also demon- secretion of mIL-2 in a dose-dependent manner (Fig. 2A). Treat- strated that in some melanoma patients bearing CEACAM1-posi- ment with the control anti-CD99 12E7 gave no response (Fig. 2A). tive tumors, a dramatic increase in the proportion of CEACAM1- Next, BW parental cells and BW/CEACAM1-␨ were cocultured positive NK cells was observed (7). Because heterophilic with irradiated .221, .221/CEACAM1, or .221/CEACAM6 cells interactions of the CEACAM1 protein with the CEACAM6 pro- for 48 h. Significant amounts of mIL-2 were detected in the su- tein were reported previously (12), and because some CEACAM1- pernatant of BW/CEACAM1-␨ cells coincubated with .221/ positive tumors down-regulate the CEACAM1 protein expression CEACAM1 cells (Fig. 2B). In contrast, no mIL-2 was detected and instead replace it with CEACAM6 (14, 15), it was important when BW/CEACAM1-␨ cells were coincubated with .221 or .221/ to investigate whether CEACAM1 can interact with CEACAM6. CEACAM6 cells (Fig. 2B). No secretion of mIL-2 was observed 721.221 (.221) cells were transfected with the CEACAM1 when parental BW cells were used (Fig. 2B). These combined cDNA (.221/CEACAM1) and with the CEACAM6 cDNA (.221/ results clearly indicate that the CEACAM1 and CEACAM6 pro- CEACAM6) as described in Materials and Methods. The expres- teins do not bind or functionally interact. sion level was monitored with the Kat4c mAb (Fig. 1). For mea- Downloaded from suring direct binding of CEACAM1 to the transfected cells, the GPI anchorage of CEACAM6 is not responsible for the lack of extracellular portion of the CEACAM1 fused to the Fc portion of heterophilic interactions with the CEACAM1 human IgG1 (CEACAM1-Ig) was used in flow cytometry binding Several explanations may account for the lack of heterophilic in- assays. The production and purification of the CEACAM1-Ig fu- teractions between CEACAM1 and CEACAM6 proteins. sion protein were performed as described in Materials and Meth- CEACAM1 is a transmembrane protein, whereas CEACAM6 is ods. Homophilic binding of the CEACAM1-Ig fusion protein was http://www.jimmunol.org/ GPI-anchored to the cell membrane. It is possible that the GPI observed only to the .221/CEACAM1 cells (Fig. 1). In contrast, despite a slightly higher expression level of the CEACAM6 pro- tein (detected by Kat4c mAb, Fig. 1), CEACAM1-Ig did not bind to .221/CEACAM6 cells (Fig. 1). The control CD99-Ig fusion pro- tein did not stain any of the transfectants (data not shown). The potential heterophilic interactions between the CEACAM1 and CEACAM6 proteins were further investigated using the BW by guest on October 1, 2021

FIGURE 2. CEACAM1 and the CEACAM6 proteins do not function- ally interact. A, The amount of mIL-2 in culture supernatant of Kat4c- FIGURE 1. CEACAM1-Ig do not recognize the CEACAM6 protein. treated and control 12E7 BW/CEACAM1-␨ cells as measured by ELISA. Stable .221/CEACAM1 and .221/CEACAM6 were generated as described. The x-axis is the amount of immobilized mAb per reaction, and the y-axis The expression level was monitored with the Kat4c mAb (empty histo- is the optic density at a wavelength of 650 nm. This figure shows the mean grams). Binding of CEACAM1 was assessed with the CEACAM1-Ig fu- of three independent experiments. B, mIL-2 secretion by BW parental cells sion protein (empty histograms). The reagents used are indicated in each or by BW/CEACAM1-␨ cells coincubated for 48 h with irradiated .221, histogram. The background (shaded histograms) is the corresponding stain- .221/CEACAM1, or with .221/CEACAM6 cells. The y-axis is the optic ing of .221 parental cells. This figure shows one representative experiment density at a wavelength of 650 nm. The average of four independent ex- of 20 performed. periments is shown. The Journal of Immunology 3735 anchor of CEACAM6 and the absence of transmembrane and cy- tosolic portions weaken the interaction. Furthermore, it is possible that other transmembrane elements play a key role in the interac- tions. For example, a cysteine residue located in the transmem- brane domain of HLA-C was reported to be crucial for the inhi- bition mediated by an unknown inhibitory NK receptor (19). To test whether the GPI anchor of CEACAM6 protein is responsible for the lack of CEACAM1 binding, we generated a chimeric con- FIGURE 4. Sequence alignment of CEACAM family members. Letters struct comprised of the entire extracellular portion of CEACAM6 in bold indicate amino acid residue 1. Identical residues of the known fused to the transmembrane and tail portions of CEACAM1 motifs crucial for binding are underlined. Different residues in the binding (CCM6-TailCCM1). The .221 cells were stably transfected with motifs are highlighted with black (for RQ residues) or gray (for SL resi- dues) backgrounds. the CCM6-TailCCM1 construct (.221/CCM6-TailCCM1). The ex- pression level of the CCM6-TailCCM1 chimeric protein, detected by Kat4c mAb, was similar to that of the other .221/CEACAM stable transfectants (Figs. 1 and 3A). Importantly, no binding of N-domain (13). Sequence alignment of the N-domains of CEACAM1-Ig was observed to .221/CCM6-TailCCM1cells (Fig. CEACAM-related proteins, including CEACAM1, CEACAM3, 3A). In agreement with the binding results, the presence of mIL-2 CEACAM5, and CEACAM6, revealed exceptional homology was not detected in the supernatant of BW/CEACAM1 cells co- (Fig. 4). Within the CEACAM1 N-domain, several amino acid incubated with .221/CCM6-TailCCM1cells (Fig. 3B). These re- residues were suggested to be crucial for binding. These include sults indicate that the lack of heterophilic interactions of amino acids 39V and 40D (13) and the salt bridge between 64R and Downloaded from CEACAM1 and CEACAM6 is not due to the transmembrane and 82D (13). All of the above-reported amino acid residues are present cytosolic tail portions of the proteins. in the N-domain of CEACAM6 (Fig. 4), implying that they may not account for the lack of heterophilic interactions with the 43 44 Residues R and Q are critical for CEACAM1 binding CEACAM1 protein. CEACAM-related proteins share a common basic structure of sev- Taheri et al. (20) demonstrated that three short sequences lo- eral sequential Ig-like domains. The Ig-like domains serve as fun- cated in the N-domain of CEACAM5 are critical for CEACAM5 http://www.jimmunol.org/ damental building blocks of the various CEACAM-related pro- homophilic interactions. These short sequences include teins, and they differ only slightly from one protein to another. 30GYSWYK, 42NRQII, and 80QNDTG. Importantly, these three Importantly, the binding site of the CEACAM1 is located in the short sequences are present in the N-domain of the CEACAM1 protein. However, only the 30GYSWYK and 80QNDTG short se- quences are preserved in the N-domain of the CEACAM6 protein, whereas the 43R44Q residues are replaced with 43S44L residues within the 42NRQII sequence (Fig. 4). We have therefore gener- ated a mutated construct of the CEACAM6 gene that includes by guest on October 1, 2021 amino acids R and Q at positions 43 and 44 instead of S and L, respectively (CCM6-SL43,44RQ). In addition, we have generated the reciprocal mutation in CEACAM1 that includes amino acids S and L at positions 43 and 44 instead of R and Q, respectively (CCM1-RQ43,44SL). The .221 cells were stably transfected with the various constructs and tested for expression using Kat4c mAb (Fig. 5A). The expression levels of the mutated proteins were sim- ilar to those of CEACAM1 and CEACAM6 (Figs. 1 and 5A). Next, we tested .221/CCM6-SL43,44RQ and .221/CCM1- RQ43,44SL in flow cytometry binding assays using CEACAM1- Ig. Remarkably, substitution of 43R44Q with 43S44L in .221/ CCM1-RQ43,44SL abolished homophilic binding (Fig. 5A). This abolishment was probably not merely due to a steric disturbance of the CEACAM1 N-domain structure, because the reciprocal muta- tion, 43S44L with 43R44Q in .221/CCM6-SL43,44RQ, conferred strong binding of the CEACAM1-Ig fusion protein (Fig. 5A). The CD99-Ig fusion protein did not stain any of the transfectants (data not shown). Strikingly, recognition of the mutated CEACAM1 protein by the conformation-dependent anti-CEACAM1 mAb 5F4 (10, 13), was abolished, whereas specific staining of .221/CCM6- SL43,44RQ was observed (Fig. 5B). These results imply that the FIGURE 3. Substitution of the GPI link of CEACAM6 with the trans- 43R and 44Q residues are critically involved in conferring the ap- membrane and tail of CEACAM1 does not induce heterophilic binding. A, propriate conformation required for recognition by CEACAM1. Staining of .221/CCM6-TailCCM1 cells with Kat4c (empty histogram) The binding results were also confirmed by functional assays mAb or CEACAM1-Ig (empty histogram). The reagents used are indicated using the BW cell system. Significant amounts of mIL-2 were in each histogram. The background is the staining of .221 parental cells ␨ (shaded histogram). This figure shows one representative experiment of 10 detected only in supernatants of BW/CEACAM1- cells coincu- performed. B, mIL-2 secretion by BW/CEACAM1-␨ cells coincubated for bated with irradiated .221/CCM6-SL43,44RQ or .221/CEACAM1 48 h with irradiated .221 or with .221 transfectants. The y-axis indicates the cells (Fig. 5C). The stronger mIL-2 induction after incubation of optic density at a wavelength of 650 nm. The average of four independent BW/CEACAM1-␨ cells with the .221/CCM6-SL43,44RQ cells experiments is shown. compared with .221/CEACAM1 cells might be due to the higher 3736 RESIDUES R43 AND Q44 DETERMINE RECOGNITION OF CEACAM1 Downloaded from http://www.jimmunol.org/

FIGURE 5. Recognition of CEACAM1 is dependent on the presence of 43R44Q. A, Staining of .221/CCM1-RQ43,44SL or .221/CCM6-SL43,44RQ cells with Kat4c mAb or CEACAM1-Ig fusion protein as indicated in each histogram. The corresponding staining of .221 parental cells was used as background (sheded histograms). This figure shows one representative experiment of six performed. B, Staining of .221/CCM1-RQ43,44SL or .221/CCM6-SL43,44RQ cells with the conformation-dependent 5F4 mAb (thick lines). The corresponding staining of .221 parental cells was used as background (thin lines). This figure shows one representative experiment of six performed. C, mIL-2 secretion by BW/CEACAM1-␨ cells coincubated for 48 h with irradiated .221 or by guest on October 1, 2021 .221 transfectants. The y-axis indicates the optic density at a wavelength of 650 nm. The average of five independent experiments is shown. protein expression measured by Kat4c mAb (Figs. 1 and 5A). The resentative clone NK23 is presented in Fig. 6). As previously re- presence of mIL-2 could not be detected in the supernatants of ported (7Ð9), inhibition of NK killing was observed when .221/ BW/CEACAM1-␨ cells coincubated with irradiated .221/ CEACAM1 cells were used. This inhibition was the result of CEACAM6 cell or .221/CEACAM1-RQ43,44SL (Fig. 5C). CEACAM1 inhibition, because anti-CEACAM Abs abrogated this Mouse IL-2 was not detected when BW parental cells were used effect (Fig. 6). The lack of heterophilic interactions between (data not shown). Thus, residues 43R44Q are critical for the func- CEACAM1 and CEACAM6 was evident in the NK killing assays, tional CEACAM1 interactions. because .221/CEACAM6 and .221/CCM6-TailCCM1 cells were killed as efficiently as parental .221 cells (Fig. 6). In agreement 43 44 Residues R and Q are critical for CEACAM1-mediated with the above results (Fig. 5), no inhibition was observed when inhibition of NK cell cytotoxicity .221/CCM1-RQ43,44SL cells were used (Fig. 6). Remarkably, a We have previously reported that CEACAM1 plays a major role in strong inhibition of killing was observed when CCM6-SL43,44RQ regulation of NK cell cytotoxicity (7), inhibition of decidual im- cells were used as targets (Fig. 6). The inhibition was even stron- mune responses after activation (8), and conferring protection from ger than that observed with the homophilic CEACAM1 interac- NK autoreactivity in TAP2-deficient patients (9). To test whether tions, probably due to the higher CCM6-SL43,44RQ expression. residues 43R44Q would also be important in the inhibition of NK This inhibition was the result of heterophilic interactions with killing, we isolated NK cells from several healthy donors, depleted CEACAM1 protein on NK cells, because killing was restored the CD16-positive NK cells, cultured activated NK clones as de- when anti-CEACAM Abs were included in the assay (Fig. 6). The scribed in Materials and Methods, and stained them for control anti-ubiquitin had little or no effect when included in the CEACAM1 expression. assays (Fig. 6). CEACAM1-positive NK clones were assayed for cytotoxic ac- Specificity of CEACAM1 binding to CEACAM6 is controlled by tivity against .221 parental cells and various stable transfectants, 43 44 including 221/CEACAM1, .221/CEACAM6, .221/CCM6-Tail- the presence of both R and Q residues CCM1, .221/CCM1-RQ43,44SL, and .221/CCM6-SL43,44RQ To determine whether both residues are required for binding, we cells. NK cytotoxicity assays were performed with no Ab included, mutated the amino acid residues in positions 43 and 44 in in the presence of anti-CEACAM polyclonal Abs, or with the con- CEACAM1 (contains 43R44Q) and CEACAM6 (contains 43S44L). trol anti-ubiquitin polyclonal Abs. All NK clones efficiently killed Using site-directed mutagenesis in the CEACAM1, we changed parental .221 cells regardless of whether Abs were included (rep- the 43R residue to 43A (CCM1-R43A) and the 44Q residue to 44L The Journal of Immunology 3737

FIGURE 6. NK-mediated cytotoxicity. CEACAM1-positive NK clones were obtained as described in Materials and Methods. NK clones were tested in killing assays against the indicated cells in an E:T cell ratio of 2:1. When rabbit polyclonal Abs were included in the assays, the final concentration was 20 ␮g/ml. This figure shows the results of a representative NK clone.

43 43 (CCM1-Q44L). In CEACAM6, we changed the Sto R priate conformation for 5F4 (Fig. 7A). Compatible with the latter Downloaded from (CCM6-S43R) and 44Lto44Q (CCM6-L44Q). All mutants were observation, no binding of CEACAM1-Ig to .221/CCM6-L44Q generated as described in Materials and Methods and stably trans- could be detected (Fig. 7B). Point mutation in the 43R residue of fected into .221 cells. The expression level was monitored by CEACAM1 did not affect 5F4 mAb binding (Fig. 7A), suggesting Kat4c mAb, and conformation was monitored by 5F4 mAb (Fig. that by itself the 43R residue had no significant effect on confor- 7A). Importantly, substitution for 44Q in CEACAM1 protein by mation of 5F4-recognized epitope. Despite that, the CEACAM1-Ig 44

L in .221/CCM1-Q44L completely abrogated 5F4 binding, fusion protein did not recognize .221/CEACAM1-R43A cells (Fig. http://www.jimmunol.org/ whereas the Kat4c binding observed was similar to that of wild- 7B). Thus, elements of CEACAM1 other than the presence of the type CEACAM1 (Fig. 7A). This suggests that the 44Q residue is 5F4 epitope and the presence of the 44Q residue play a crucial role essential for maintaining appropriate conformation, which is cru- in CEACAM1 binding. In this regard, it should be noted that the cial for binding of 5F4 mAb. Indeed, this mutation also resulted in expression level of .221/CCM1-R43A obtained was lower than the lack of recognition by the CEACAM1-Ig (Fig. 7B). Similar results expression levels of the other transfectants (Fig. 7A), which might were obtained when both 44Q and 43R residues in CEACAM1 account for the lack of efficient binding of CEACAM1-Ig. There- were mutated (Fig. 5). Surprisingly, however, the reciprocal mu- fore, to test whether the 43R residue by itself can confer tant .221/CCM6-L44Q was not recognized by 5F4 mAb, indicat- CEACAM1 binding, we replaced the 43S of CEACAM6 with 43R. ing that it is not the only factor crucial for conferring the appro- .221/CCM6-S43R cells were not stained by either the 5F4 mAb by guest on October 1, 2021

FIGURE 7. Both the 43 and 44 residues of CEACAM1 are crucial for the interaction. A, Staining of .221 and various .221 stable transfectants with 5F4 u f Ј Ⅺ mAb ( ) or with Kat4c ( ). The staining of the secondary reagent FITC-conjugated goat anti-mouse F(ab )2 of each cell type was used as background ( ). The y-axis indicates the median fluorescence intensity (MFI). This figure shows one representative experiment of four performed. B, Staining of .221 and various .221 stable transfectants with CEACAM1-Ig fusion protein. The y-axis indicates the median fluorescence intensity (MFI). This figure shows one representative experiment of four performed. C, YTS cells expressing the CEACAM1 protein (YTS/CCM1) or mock-transfected (YTS/control) were tested in killing assays against .221 and .221 transfectants. The E:T cell ratio was 2:1. This figure shows the average of three independent experiments. D, Staining of .221/CEACAM5 cells with 5F4, Kat4c, or CEACAM1-Ig was performed as indicated in each histogram. The corresponding staining of .221 parental cells was used as background (thin lines). This figure shows one representative experiment of six performed. 3738 RESIDUES R43 AND Q44 DETERMINE RECOGNITION OF CEACAM1

(Fig. 7A) or the CEACAM1-Ig fusion protein (Fig. 7B). Gain-of- family. It was postulated previously that some proteins belonging binding of CEACAM1-Ig to CEACAM6 was evident only when to the CEACAM family, including CEACAM1 and CEACAM6, both 43S and 44L residues when replaced with 43R and 44Q, re- interact homophilically in head-to-head N-domain interactions (13, spectively (Fig. 5A). Thus, both 43R and 44Q residues are critical 27, 28). In addition, it was reported that the CEACAM5 protein for interaction with CEACAM1. interacts homophilically through head-to-tail interactions between The binding results were also confirmed in functional killing the N-domain of one molecule and the B3-domain of the other assays. To optimize the isolation of the experimental variables, we (22). Interestingly, the N-domains of all CEACAM-related pro- used the YTS NK tumor line. The NK tumor line YTS was either teins are exceptionally homologous, with ϳ90% homology (Fig. mock-transfected (YTS/control) or transfected with CEACAM1 4). Thus, it is expected that the various CEACAM-related proteins protein (YTS/CCM1) as previously described (7) and tested in would interact with each other. However, in this work we demon- killing assays against the various .221 transfectants. The function strate that the CEACAM1 protein does not interact with the of CEACAM1 protein in YTS/CCM1 cells was confirmed, be- CEACAM6 protein using both binding and various functional as- cause killing of .221/CEACAM1 cells was inhibited compared says. In addition, other reports stated that CEACAM1 protein does with killing by YTS/control cells, whereas .221/CEACAM6 and not interact with CEACAM3 proteins (27). In view of these ob- .221/CCM6-TailCCM1 cells were killed with similar efficiency servations and the exceptional homology between the N-domains (Fig. 7C). In agreement with the CEACAM1-Ig binding results, of the various CEACAM-related proteins, it is conceivable to as- the inhibition of YTS/CCM1 cells was abolished when the .221/ sume that there are probably only a few key amino acid residues CCM1-RQ43,44SL and .221/CCM1-Q44L transfectants were used that control the interactions of CEACAM1 protein. Indeed, we as targets (Fig. 7C), demonstrating the critical role of residue 44Q. show in this study that R and Q at positions 43 and 44 determine However, in agreement with our above observation, the presence CEACAM1 interactions. Downloaded from of 44Q only is not enough to confer inhibition, and only a mild Watt and colleagues (13) proposed that amino acids 39V and 40D inhibitory effect was observed when the .221/CCM1-R43A cells are critical for homophilic CEACAM1 interactions and also dem- were used (Fig. 7C). This result was also supported by the obser- onstrated that intact salt bridge between 64R and 82D enables bind- vation that inhibition of YTS/CCM1 cells by heterophilic interac- ing activity through general stabilization of the CEACAM1 pro- tions with CEACAM6 was observed only with the .221/CCM6- tein. However, these findings cannot explain the specificity of SL43,44RQ double mutation, whereas no inhibition was observed CEACAM1 heterophilic interactions, because all these amino acid http://www.jimmunol.org/ when .221/CCM6-S43R or .221/CCM6-L44Q cells were used residues are conserved in the N-domain of all CEACAM-related (Fig. 7C). Similar results were obtained with primary NK clones proteins (Fig. 4). (data not shown). In conclusion, both R and Q residues in positions Taheri and colleagues (20) showed that three short sequences 43 and 44, respectively, are required for functional interaction with located in the N-domain of the CEACAM5 protein are critical for CEACAM1. CEACAM5 homophilic interactions. These short sequences in- Supporting our hypothesis, it was previously reported that clude 30GYSWYK, 42NRQII, and 80QNDTG. We therefore spec- CEACAM1 can heterophilically interact with the CEACAM5 pro- ulated that similar sequences might be important for the tein (12). Indeed, the CEACAM5 protein is the only CEACAM CEACAM1 interactions. Alignment of the N-domains of the var- by guest on October 1, 2021 family member other than CEACAM1 that contains 43R44Q resi- ious CEACAM-related proteins revealed that the main difference dues (Fig. 4). We therefore next investigated the interactions be- in these three sequences among the various CEACAM proteins is tween CEACAM1 and CEACAM5. The expression level of .221/ located in the 42NRQII sequence (Fig. 4). The CEACAM family CEACAM5 transfectant was monitored with Kat4c and was can be divided into two main groups in this regard: CEACAM1 similar to that of the other CEACAM transfectants (Fig. 7D). The and CEACAM5 that contain 43R and 44Q residues, and .221/CEACAM5 cells were not stained by the anti-CEACAM1- CEACAM3 and CEACAM6 that contain 43S and 44L residues specific 5F4 mAb (Fig. 7D). As expected, efficient heterophilic (Fig. 4). A clear binding of the CEACAM1-Ig fusion protein to binding of the CEACAM1-Ig fusion protein to .221/CEACAM5 .221 cells expressing either CEACAM1 protein or CEACAM5 was observed (Fig. 7D). protein was observed (Figs. 1 and 7), whereas no binding could be detected to .221/CEACAM6 cells (Fig. 1). Furthermore, substitu- Discussion tion of 43R44Q residues in CEACAM1 protein with 43S44L resi- During the past decade many different functions were implicated dues resulted in complete abrogation of the homophilic interac- for the CEACAM1 protein. These include immune regulation (7Ð tions, which was also confirmed in various functional assays (Figs. 11), inhibition of cancer proliferation (3, 4), insulin clearance (21), 5Ð7). Remarkably, the reciprocal substitution of 43S44L residues of differentiation and arrangement of tissue three-dimensional struc- CEACAM6 with 43R44Q resulted in efficient recognition by the ture (22), intercellular adhesion (23), and bacterial protein binding CEACAM1 protein evident in both binding and functional assays (24). We have recently demonstrated that the homophilic (Figs. 5Ð7). Moreover, point mutations in positions 43 and 44 in CEACAM1 interactions are strong enough to deliver inhibitory CEACAM1 and CEACAM6 revealed that both 43R and 44Q res- signals and thereby decrease human NK-mediated cytotoxicity and idues are needed for proper binding and inhibition. This conclusion decidual lymphocyte functions (7Ð9). Most of the above activities is also supported by the observation that 5F4 mAb is unable to are mediated via homophilic and heterophilic interactions of the block the inhibition mediated by CEACAM1 (7, 8), and as shown CEACAM1 protein. The various CEACAM1-mediated activities in this study, 5F4 recognition is largely restricted to the 44Q res- are regulated by the expression of CEACAM1 protein on different idue (Fig. 7A). cells, the levels of CEACAM1 expression, and the expression of Proteins of the CEACAM family can be expressed on various potential ligands for CEACAM1. Therefore, elucidating the mo- tumors (1). CEACAM1 expression might be advantageous to the lecular basis of CEACAM1 binding might result in novel treat- tumor, for example by preventing NK-mediated killing (7). How- ments for either preventing or augmenting the above functions. ever, it can also be disadvantageous, because it reduces the pro- N-domains such as those derived from CD80 (25) and ICAM-3 liferation of tumor cells (3, 4). The binding of different CEACAM (26) have, in general, an important role in mediating protein in- members by CEACAM1, as shown in this study, is controlled by teractions between various molecules that belong to the Ig super- both residues 43R and 44Q. This enables fine-tuning of the immune The Journal of Immunology 3739 response vs tumor proliferation. One notable example of this is the 11. Boulton, I. C., and S. D. Gray-Owen. 2002. Neisserial binding to CEACAM1 ϩ chain of events following colorectal cell transformation. During arrests the activation and proliferation of CD4 T lymphocytes. Nat. Immunol. 3:229. the early phases of colorectal cancer development, the cancerous 12. Oikawa, S., M. Kuroki, Y. Matsuoka, G. Kosaki, and H. Nakazato. 1992. Ho- cells usually express CEACAM1 protein (14, 15) despite its po- motypic and heterotypic Caϩϩ-independent cell adhesion activities of biliary glycoprotein, a member of carcinoembryonic antigen family, expressed on CHO tential antiproliferative effects (3). This might enable the newly cell surface. Biochem. Biophys. Res. Commun. 186:881. developing tumor to evade NK-mediated attack at the early stages 13. Watt, S. M., A. M. Teixeira, G. Q. Zhou, R. Doyonnas, Y. Zhang, F. Grunert, at the expense of tumor growth. Later, when the immune response R. S. Blumberg, M. Kuroki, K. M. Skubitz, and P. A. Bates. 2001. Homophilic adhesion of human CEACAM1 involves N-terminal domain interactions: struc- is no longer dominated by NK cells, CEACAM1 protein expres- tural analysis of the binding site. Blood 98:1469. sion is down-regulated, and a concomitant up-regulation of 14. Thompson, J., M. Seitz, E. Chastre, M. Ditter, C. Aldrian, C. Gespach, and CEACAM6 protein occurs (14, 15). The mechanism of W. Zimmermann. 1997. Down-regulation of carcinoembryonic antigen family member 2 expression is an early event in colorectal tumorigenesis. Cancer Res. CEACAM1 down-regulation in colorectal cancer development is 57:1776. still unknown. However, it has been recently reported that 15. Neumaier, M., S. Paululat, A. Chan, P. Matthaes, and C. Wagener. 1993. Biliary CEACAM1 expression is repressed at the transcriptional level dur- glycoprotein, a potential human cell adhesion molecule, is down-regulated in colorectal carcinomas. Proc. Natl. Acad. Sci. USA 90:10744. ing tumorigenesis of prostate cancer by the transcription repressor 16. Mandelboim, O., H. T. Reyburn, M. Vales-Gomez, L. Pazmany, M. Colonna, Sp2 (29). Because the CEACAM6 protein contains 43S and 44L G. Borsellino, and J. L. Strominger. 1996. Protection from lysis by natural killer residues, interactions with CEACAM1 are abolished. At this stage, cells of group 1 and 2 specificity is mediated by residue 80 in human histocom- patibility leukocyte antigen C alleles and also occurs with empty major histo- rapid proliferation of the cancer cells occurs consequent to re- compatibility complex molecules. J. Exp. Med. 84:913. moval of the inhibitory effect exerted by the homophilic 17. Katz, G., G. Markel, S. Mizrahi, T. I. Arnon, and O. Mandelboim. 2001. Rec- CEACAM1 interactions. At the same time, the CEACAM6 protein ognition of human leukocyte antigen (HLA)-Cw4 but not -Cw6 by the natural

killer cell receptor KIR2DS4. J. Immunol. 166:7260. Downloaded from provides other adhesive properties (1, 27) that are also essential for 18. Laack, E., H. Nikbakht, A. Peters, C. Kugler, Y. Jasiewicz, L. Edler, J. Brummer, further tumor development. Full delineation of the biochemical U. Schumacher, and D. K. Hossfeld. 2002. Expression of CEACAM1 in adeno- carcinoma of the lung: a factor of independent prognostic significance. J. Clin. structural properties determining CEACAM1 binding and speci- Oncol. 20:4279. ficity might provide better tools for identifying other putative 19. Davis, D. M., O. Mandelboim, I. Luque, E. Baba, J. Boyson, and physiological ligands for CEACAM1. J. L. Strominger. 1999. The transmembrane sequence of human histocompatibil- ity leukocyte antigen (HLA)-C as a determinant in inhibition of a subset of nat- ural killer cells. J. Exp. Med. 189:1265. References 20. Taheri, M., U. Saragovi, A. Fuks, J. Makkerh, J. Mort, and C. P. Stanners. 2000. http://www.jimmunol.org/ Self recognition in the Ig superfamily: identification of precise subdomains in 1. Beauchemin, N., P. Draber, G. Dveksler, P. Gold, S. Gray-Owen, F. Grunert, carcinoembryonic antigen required for intercellular adhesion. J. Biol. Chem. 275: S. Hammarstrom, K. V. Holmes, A. Karlsson, M. Kuroki, et al. 1999. Redefined 26935. nomenclature for members of the carcinoembryonic antigen family. Exp. Cell Res. 252:243. 21. Poy, M. N., Y. Yang, K. Rezaei, M. A. Fernstrom, A. D. Lee, Y. Kido, 2. Benchimol, S., A. Fuks, S. Jothy, N. Beauchemin, K. Shirota, and C. P. Stanners. S. K. Erickson, and S. M. Najjar. 2002. CEACAM1 regulates insulin clearance in 1989. Carcinoembryonic antigen, a human tumor marker, functions as an inter- liver. Nat. Genet. 3:270. cellular adhesion molecule. Cell 57:327. 22. Huang, J., J. D. Hardy, Y. Sun, and J. E. Shively. 1999. Essential role of biliary 3. Beauchemin, N., T. Kunath, J. Robitaille, B. Chow, C. Turbide, E. Daniels, and glycoprotein (CD66a) in morphogenesis of the human mammary epithelial cell A. Veillette. 1997. Association of biliary glycoprotein with protein tyrosine phos- line MCF10F. J. Cell Sci. 112:4193. phatase SHP-1 in malignant colon epithelial cells. Oncogene 14:783. 23. Rojas, M., A. Fuks, and C. P. Stanners. 1990. Biliary glycoprotein, a member of the immunoglobulin supergene family, functions in vitro as a Ca2ϩ-dependent 4. Busch, C., T. A. Hanssen, C. Wagener, and B. Obrink. 2002. Down-regulation of by guest on October 1, 2021 CEACAM1 in human prostate cancer: correlation with loss of cell polarity, in- intercellular adhesion molecule. Cell Growth Differ. 1:527. creased proliferation rate, and Gleason grade 3 to 4 transition. Hum. Pathol. 24. Virji, M., S. M. Watt, S. Barker, K. Makepeace, and R. Doyonnas. 1996. The 33:290. N-domain of the human CD66a adhesion molecule is a target for Opa proteins of 5. Volpert, O., W. Luo, T. J. Liu, V. T. Estrera, C. Logothetis, and S. H. Lin. 2002. Neisseria meningitidis and Neisseria gonorrhoeae. Mol Microbiol. 22:929. Inhibition of prostate tumor angiogenesis by the tumor suppressor CEACAM1. 25. Schwartz, J. C., X. Zhang, A. A. Fedorov, S. G. Nathenson, and S. C. Almo. J. Biol. Chem. 277:35696. 2001. Structural basis for co-stimulation by the human CTLA-4/B7-2 complex. 6. Thies, A., I. Moll, J. Berger, C. Wagener, J. Brummer, H. J. Schulze, G. Brunner, Nature 410:604. and U. Schumacher. 2002. CEACAM1 expression in cutaneous malignant mel- 26. Holness, C. L., P. A. Bates, A. J. Little, C. D. Buckley, A. McDowall, D. Bossy, anoma predicts the development of metastatic disease. J. Clin. Oncol. 20:2530. N. Hogg, and D. L. Simmons. 1995. Analysis of the binding site on intercellular 7. Markel, G., N. Lieberman, G. Katz, T. I. Arnon, M. Lotem, O. Drize, adhesion molecule 3 for the leukocyte integrin lymphocyte function-associated R. S. Blumberg, E. Bar-Haim, R. Mader, L. Eisenbach, et al. 2002. CD66a in- antigen 1. J. Biol. Chem. 270:877. teractions between human melanoma and NK cells: a novel class I MHC-inde- 27. Kuroki, M., H. Abe, T. Imakiirei, S. Liao, H. Uchida, Y. Yamauchi, S. Oikawa, pendent inhibitory mechanism of cytotoxicity. J. Immunol. 168:2803. and M. Kuroki. 2001. Identification and comparison of residues critical for cell- 8. Markel, G., D. Wolf, J. Hanna, R. Gazit, D. Goldman-Wohl, Y. Lavy, S. Yagel, adhesion activities of two neutrophil CD66 antigens, CEACAM6 and and O. Mandelboim. 2002. Pivotal role of CEACAM1 protein in the inhibition of CEACAM8. J. Leukocyte Biol. 70:543. activated decidual lymphocyte functions. J. Clin. Invest. 110:943. 28. Oikawa, S., C. Inuzuka, M. Kuroki, F. Arakawa, Y. Matsuoka, G. Kosaki, and 9. Markel, G., H. Mussaffi, K. L. Ling, M. Salio, S. Gadola, G. Steuer, H. Blau, H. Nakazato. 1991. A specific heterotypic cell adhesion activity between mem- H. Achdout, M. De Miguel, T. Gonen-Gross, et al. 2003. The mechanisms con- bers of carcinoembryonic antigen family, W272 and NCA, is mediated by N- trolling NK cell autoreactivity in TAP2-deficient patients. Blood 103:1770. domains J. Biol. Chem. 266:7995. 10. Morales, V. M., A. Christ, S. M. Watt, H. S. Kim, K. W. Johnson, N. Utku, 29. Phan, D., C. J. Cheng, M. Galfione, F. Vakar-Lopez, J. Tunstead, A. M. Texieira, A. Mizoguchi, E. Mizoguchi, G. J. Russell, et al. 1999. Regu- N. E. Thompson, R. R. Burgess, S. M. Najjar, L. Y. Yu-Lee, and S. H. Lin. 2004. lation of human intestinal intraepithelial lymphocyte cytolytic function by biliary Identification of Sp2 as a transcriptional repressor of carcinoembryonic antigen- glycoprotein (CD66a). J. Immunol. 163:1363. related cell adhesion molecule 1 in tumorigenesis. Cancer Res. 64:3072.