Granulocyte CEACAM3 Is a Phagocytic Receptor of the Innate Immune System that Mediates Recognition and Elimination of Human-specific Pathogens

The gram-negative pathogens Neisseria gonorrhoeae, N. meningitidis, Moraxella catarrhalis, and Haemophilus influenzae are highly adapted to the human as their sole natural host. They colonize mucosal surfaces in different parts of the body: the nasopharynx in the case of N. meningitidis, M. catarrhalis, and H. influenzae or the urogenital tract in the case of N. gonorrhoeae. Though colonization with these pathogens can develop into invasive and sometimes life-threatening diseases, colonization most often (N. meningitidis, M. catarrhalis, and H. influenzae) or in a large proportion of the cases (N. gonorrhoeae) remains without symptoms (1). In particular, in gonorrhea the often mild symptoms can resolve despite the lack of effective acquired immunity (1–3), suggesting that innate immune defenses are able to contain the pathogen.

The four pathogenic species share several common adaptations such as phase or antigenic variation of surface structures and expression of outer membrane proteins that mediate the binding to members of the carcinoembryonic antigen-related cell adhesion molecule (CEACAM) family on human cells (4–8). In the case of H. influenzae, CEACAM recognition is mediated by the P5 protein found on typeable and nontypeable strains (9), whereas UspA1 has been identified recently as the M. catarrhalis CEACAM-binding protein (7). CEACAM recognition by pathogenic Neisseriae is mediated by members of the phase-variable Opa protein family (Opa_CEA) (for review see reference 10).

Opa_CEA proteins from gonococci and meningococci can recognize four members of the human CEACAM family, namely CEACAM1, CEACAM3, CEA, and CEACAM6 (6, 11, 12). The complete spectra of CEACAMs recognized by the P5 protein of H. influenzae or the UspA1 protein of M. catarrhalis have not been analyzed in detail, but they seem to overlap the spectrum of Opa_CEA-recognized CEACAMs (7, 8). Except for CEACAM1, homologues of these human receptors have not been detected in rodents, limiting the in vivo analysis of CEACAM function.

In vitro experiments suggest that several CEACAMs contribute to cell–cell adhesion by homo- or heterotypic interaction and participate in signal transduction into the cell (13). With regard to pathogen-induced signaling events, Opa_CEA-mediated engagement of CEACAM1 expressed on activated CD4-positive T cells has been shown to cause an arrest of T cell proliferation in vitro (14). This immunosuppressive effect is presumably due to negative signals emanating from a functional immunoreceptor tyrosine-based inhibitory motif in the cytoplasmic tail of CEACAM1 (15). CEACAM1 together with CEACAM3 and CEACAM6 is expressed on granulocytes that internalize CEACAM-binding bacteria in the absence of opsonizing antibodies or complement factors (4, 5, 16, 17). This opsonin-independent phagocytosis is accompanied by increased activity of Src family protein tyrosine kinases (PTKs), stimulation of the small GTPase Rac, and prominent rearrangements of the actin cytoskeleton (17, 18), suggesting that human phagocytes can specifically detect and efficiently internalize bacteria that express CEACAM-binding adhesins. However, it is so far unclear which of the different CEACAM receptors expressed on granulocytes is responsible for the signaling events observed in response to CEACAM-binding pathogens and if these biochemical events are linked to the elimination of CEACAM-binding bacteria by human phagocytes.

In this study, we demonstrate that the granulocyte-specific orphan receptor CEACAM3 is linked to the stimulation of the small GTPase Rac in response to pathogenic, CEACAM-binding N. gonorrhoeae, M. catarrhalis, and H. influenzae. Stimulation of Rac is dependent on Src family kinase–mediated phosphorylation of the ITAM-like sequence in the cytoplasmic domain of CEACAM3 and leads to the efficient internalization of the pathogens. Since blockage of CEACAM3 or interference with Rac stimulation reduce phagocyte clearance of CEACAM-binding bacteria, our results suggest an important function for CEACAM3 in the control of human-specific pathogens by the innate immune system.

Opa_CEA-expressing (Opa₅₂), nonpiliated N. gonorrhoeae MS11-B2.1 (strain N309), nonopaque, piliated gonococci MS11, and commensal N. cinerea were provided by Thomas Meyer (Max-Planck-Institut für Infektionsbiologie, Berlin, Germany). Neisseria were grown on GC-agar (GIBCO BRL) supplemented with vitamins at 37°C, 5% CO₂ and subcultured daily. The unencapsulated variant of H. influenzae strain RD (obtained from A. Reidl, Zentrum für Infektionsforschung) and M. catarrhalis strain 11994 (obtained from DSMZ) were grown on BHI agar at 37°C, 5% CO₂. For certain assays, bacteria were labeled with 0.2 μg/ml 5-(6)-carboxyfluorescein-succinylester (FITC; Molecular Probes) in PBS for 15 min at 37°C in the dark and washed three times with PBS before use.

The human embryonic kidney cell line 293T (293 cells) was grown in DMEM/10% calf serum at 37°C, 5% CO₂. Cells were subcultured every 3–4 d. Prior to infection and Rac pull-down assays, cells were serum starved overnight in DMEM containing 0.5% calf serum. Primary human granulocytes were purified from freshly drawn blood as described (17). Viability of cells was determined before infection using trypan blue staining and in all cases was >90%.

Plasmids encoding cDNAs of human CEACAM1, CEACAM3, CEACAM3 Y241F, and CEACAM6 were provided by Wolfgang Zimmermann (Universitätsklinikum Grosshadern, München, Germany). CEACAM3 WT and CEACAM3 Y241F were amplified with primers CEA3HA-sense, 5′-GGGAAGCTTGCCATGGGGCCCCCCTCAGCCTCTCCCCAC-3′ and CEA3HA-anti, 5′-GGGGACGTCATAGGGATAAGAAGCCACTTCTGCTTTGTGGTC-CATCCG-3′ and subcloned into the HindIII-AatII–digested plasmid pBluescript FAK-HA (19) allowing an in-frame fusion of the COOH-terminal cytoplasmic domain of CEACAM3 with a double HA tag. CEACAM3 lacking the cytoplasmic domain (CEACAM3 ΔCT) was constructed accordingly using primers CEA3HA-sense and CEA3HA-deltaCTanti, 5′-GGGGACGTCATAGGGATATTTGGCAAGGAGCAGGAAACACACCA-GTGC-3′. Introduction of the Y230F point mutation into CEACAM3 WT and CEACAM3 Y241F was accomplished using the Quikchange mutagenesis kit (Stratagene), resulting in CEACAM3 Y230F and CEACAM3 Y230/241F. CEACAM3 constructs were verified by sequencing and subcloned via HindIII-NotI restriction sites into pcDNA3.1 (Invitrogen). Myc-tagged Rac N17 and Cdc42 N17 were provided by Alan Hall (University College, London, UK), VSV-tagged Rho N19 was provided by Thomas Adam (Universitätsklinikum Charité, Berlin, Germany), kinase-inactive c-Src (Src K297M) was provided by David Schlaepfer (The Scripps Research Institute, La Jolla, CA), and the GST-PAK CRIB domain (GST-CRIB) was provided by Gary Bokoch (The Scripps Research Institute, La Jolla, CA). GST-CRIB was expressed in Escherichia coli BL-21 and purified using GSTrap FF (Amersham Biosciences). TAT-RacN17, TAT Cdc42N17, and TAT-LacZ constructs were provided by Steve Dowdy (University of California at San Diego, La Jolla, CA), and the encoded proteins were expressed and purified as described (20).

Phalloidin-TRITC was from Molecular Probes. Rabbit antiserum against N. gonorrhoeae (AK92) was provided by Thomas Meyer, and mAbs against the Opa proteins (clone 4B12) were from Mark Achtman (Max-Planck-Institut für Infektionsbiologie, Berlin, Germany), against the VSV epitope (clone P5D4) were from Sigma-Aldrich, against the myc-epitope (clone 9E10) were from Santa Cruz Biotechnology, against Rac (clone 23A8) were from Upstate Biotechnology, and against endoglin (clone P3D1) were from the Developmental Studies Hybridoma Bank. mAb against the HA tag (clone 12CA5) and against Src (clone 2–17) were purified from hybridoma supernatants. mAb Kat4c recognizing CEACAM1, CEA, and CEACAM6 was from DAKO, mAb COL-1 reacting with CEACAM3 and CEA was purchased from Zymed Laboratories, mAbs 26/3/13, specific for CEA, 9A6, specific for CEACAM6, and D14HD11, crossreactive with all CEACAMs, were from Genovac. Secondary antibodies were from Jackson ImmunoResearch Laboratories. Prior to phagocytosis inhibition experiments, antibodies were cleared from sodium azide by 10 washes with PBS using Vivaspin concentrators 50 K (Vivascience).

Phagocytes were seeded at 5 × 10⁴ cells/well in 24-well plates on acid-washed glass coverslips and infected with 10 bacteria per cell. After 1 h, infected cells were fixed with 2% glutaraldehyde in PBS for at least 1 h at 4°C and postfixed with 1% osmium tetroxide in PBS for 1 h on ice. The samples were dehydrated in a graded series of ethanol, critical point dried from CO₂, sputter coated with 8 nm gold-palladium, and examined at 25 kV accelerating voltage in a Hitachi S-800 field emission scanning electron microscope.

293 cells were transfected by the standard calcium phosphate coprecipitation method using 5 μg of CEACAM constructs or empty control vector. For cotransfection, 5 μg of cotransfected constructs together with 3 μg of CEACAM constructs, respectively, were used, and in all samples total DNA was adjusted to 8 μg using the empty control vector. Cells were employed in infection experiments 48 h after the transfection. Cell lysis and Western blotting was performed as described (21).

Immunofluorescence labeling was performed as described (17). For differentiating between extra- and intracellular bacteria, cells were infected with FITC-labeled bacteria, and fixed samples were stained before permeabilization with polyclonal rabbit-anti–N. gonorrhoeae and goat-anti–rabbit-Cy5 in staining buffer (PBS, 5% FCS), resulting in FITC-labeled intracellular and FITC/Cy5-labeled extracellular bacteria. After permeabilization of the cells, phalloidin-TRITC was used to stain cellular actin.

For gentamicin protection assays, 6 × 10⁵ 293 cells in 24-well plates were infected with 20 bacteria/cell. After the infection, the medium was replaced with DMEM containing 50 μg/ml gentamicin. After 45 min of incubation, cells were lysed by the addition of 1% saponin in PBS for 20 min. Suitable dilutions were plated to determine the number of recovered viable bacteria.

Rac-GTP loading in transfected 293 cells was determined using a GST fusion protein encompassing the CRIB domain of PAK (22). Briefly, cells were lysed in Triton buffer (25 mM Hepes, pH 7.4, 1% Triton X-100, 150 mM NaCl, 20 mM MgCl₂, 10% glycerol, 10 mM sodium pyrophosphate, 100 mM NaF, 1 mM Na₃VO₄, and 10 μg/ml each of aprotinin, leupeptin, pefabloc, and pepstatin), and cleared lysates were incubated with 10 μg of GST-CRIB immobilized on glutathione-agarose beads for 45 min at 4°C. After three washes with Triton buffer, the precipitates were taken up in 2× SDS sample buffer and analyzed by Western blotting.

Phagocytosis was determined by FACS^® analysis according to Voyich et al. (23). Briefly, 10⁶ granulocytes were infected with 2 × 10⁷ FITC-labeled bacteria in 1 ml phagocytosis buffer (PB; 1× PBS, 10 mM glucose, 1% heat inactivated serum) for 15 min at 37°C. In inhibition experiments, 20 μg of the indicated, azide-free antibodies were added to the cells 1 min before the infection. Phagocytosis was stopped by addition of ice-cold PB, samples were washed, taken up in cold PBS, 2% FCS, 2 mg/ml Trypan blue, and analyzed on a FACSCalibur (Becton Dickinson). To obtain an estimate of the amount of phagocytosed bacteria, the percentage of FITC-positive granulocytes was multiplied by the mean fluorescence of these cells. For TAT-mediated protein transduction, granulocytes were incubated for 1 h at 37°C with the indicated amounts of purified TAT fusion proteins before infection and analyzed as above. In control experiments, granulocytes were incubated with FITC-labeled TAT-RacN17 and observed by confocal microscopy to confirm the intracellular localization of the protein (Fig. S1 C).

Assays were performed according to Virji et al. (24). Briefly, 2 × 10⁶ bacteria were added to 10⁶ PMN in 300 μl DMEM, 0.5% heat-inactivated FCS in 24-well plates. Before and after a 60-min incubation at 37°C, aliquots of the samples were collected and dilutions plated on GC agar plates to determine the number of viable bacteria. In inhibition experiments, 20 μg of the indicated, azide-free antibodies were added to the cells 1 min before the infection.

Fig. S1 demonstrates (A) quenching of the fluorescence emitted by FITC-labeled bacteria upon Trypan blue addition; (B) Expression and purification of TAT-fusion proteins; and (C) TAT-mediated protein transduction of human granulocytes. Fig. S2 shows (A) FACS^® analysis of CEACAM expression on primary human granulocytes; (B) Oxidative response of granulocytes upon infection with gonococci; (C and D) CEACAM expression by transiently transfected 293 cells analyzed by Western blotting (C) or FACS^® (D), respectively; and (E) adhesion of Opa_CEA-expressing N. gonorrhoeae to CEACAM-transfected 293 cells as observed by immunofluorescence microscopy. Figs. S1 and S2 are available.

Human granulocytes express three of the four CEACAM family members, namely CEACAM1, CEACAM3, and CEACAM6, that bind neisserial Opa_CEA proteins (Fig. S2 A). Accordingly, primary granulocytes can efficiently recognize and eliminate Opa_CEA N. gonorrhoeae in the absence of specific antibodies or complement (4, 16). Indeed, when Opa_CEA gonococci were incubated with granulocytes at a 2:1 ratio, only ∼40% of the initial bacterial inoculum could be recovered after 90 min (Fig. 1 A). Nonopaque organisms were not recognized or eliminated (Fig. 1 A). Contact of granulocytes with Opa_CEAN. gonorrhoeae stimulated the cells to produce reactive oxygen derivatives (Fig. S2 B) and to exhibit pronounced lamellipodia-like protrusions on the surface (Fig. 1, B and C). These membrane protrusions were involved in the engulfment of the bacteria (Fig. 1 C). In contrast, nonopaque gonococci only moderately stimulated the oxidative burst of the granulocytes (Fig. S2 B), and neither nonopaque gonococci nor nonpathogenic N. cinerea induced lamellipodia-like protrusions on the phagocyte surface, indicating that the observed responses were due to the Opa_CEA–CEACAM interaction (Fig. 1, D and E). These results suggested that granulocytes express CEACAM family member(s) with phagocytic properties. This receptor(s) seems to transduce signals that stimulate bactericidal mechanisms resulting in elimination of CEACAM-binding bacteria. However, the identity of the CEACAM family member(s) responsible for these processes was unknown.

To study the contribution of specific CEACAM molecules to the internalization of CEACAM-binding bacteria, we took advantage of 293 cells that do not express members of the CEACAM family (Fig. S2, C and D; reference 25). Transfection of 293 cells with CEACAM1, CEACAM3, CEACAM6, or the empty control vector (pcDNA) resulted in the surface expression of the respective receptor in ∼70–80% of the cell population (Fig. S2 D). CEACAM-transfected cells supported strong attachment of Opa_CEA N. gonorrhoeae (Fig. S2 E). In addition, CEACAM3 and CEACAM6 mediated internalization of Opa_CEA gonococci within 30–60 min after infection (Fig. 2 A). In contrast, internalization mediated by CEACAM1 was <15% compared with CEACAM3, and more than 100-fold lower numbers of bacteria were recovered from control-transfected cells (pcDNA) (Fig. 2 A). Importantly, the internalization process via CEACAM3 and CEACAM6 depended on the correct phenotype of the bacteria, since nonopaque, piliated gonococci were barely taken up by these receptors (Fig. 2 B). These results indicated that both CEACAM3 and CEACAM6, but not CEACAM1, are able to mediate efficient internalization of Opa_CEA gonococci when expressed in epithelial cells.

The opsonin-independent phagocytosis of Opa_CEA-expressing gonococci by human granulocytes is accompanied by cytoskeletal rearrangements leading to lamellipodia formation and the production of reactive oxygen metabolites (4, 11, 16), processes that are regulated by the small GTPase Rac (26). To analyze if CEACAM3- or CEACAM6-mediated phagocytosis might be connected to small GTPases of the Rho family, 293 cells were cotransfected with CEACAMs and with constructs encoding dominant-negative versions of Cdc42, Rac, and RhoA. Importantly, a strong decrease in CEACAM3-mediated uptake of opaque gonococci was observed upon coexpression of dominant-negative Rac (Rac N17; Fig. 3 A). In contrast, bacterial internalization via CEACAM6 was only marginally diminished by Rac N17, suggesting that CEACAM3 might be the granulocyte receptor that is connected to the small GTPase Rac. Interestingly, neither CEACAM3- nor CEACAM6-mediated internalization was compromised by coexpression of dominant-negative versions of the closely related small GTPases Cdc42hs (Fig. 3 A) or RhoA (unpublished data).

To investigate further if CEACAM3 is connected to Rac after N. gonorrhoeae infection, cells infected for 30 min were stained for gonococci and Rac. Importantly, cells transfected with CEACAM3 showed recruitment of Rac to the vicinity of attached Opa_CEA bacteria (Fig. 3 B). In contrast, the distribution of Rac did not change in cells transfected with CEACAM6, despite the attachment of multiple gonococci (Fig. 3 B).

Biochemically, Rac stimulation is reflected by its GTP-loading status. Upon infection with Opa_CEA gonococci, we observed a strong increase in GTP-Rac in CEACAM3-expressing 293 cells compared with uninfected cells (Fig. 4 A, top). This increase was already detectable 15 min after infection and reached maximum levels between 30 and 60 min paralleling the kinetics of efficient, CEACAM3-mediated internalization. In contrast, no increase in GTP-Rac was observed in control vector-transfected or in CEACAM6-expressing cells after infection (Fig. 4 A, top). An increase in Rac-GTP–loading was only detected in response to Opa_CEA-expressing bacteria but not upon infection with nonopaque gonococci or nonpathogenic N. cinerea (Fig. 4 B), demonstrating that it is the specific interaction between the bacterial Opa_CEA adhesin and host CEACAM3 that mediates stimulation of Rac. Importantly, the levels of Rac-GTP induced by the different bacterial strains paralleled the internalization of these bacteria via CEACAM3 (Fig. 4 C). Together, these data corroborate the view that CEACAMs might be differentially connected to intracellular signaling pathways and that in particular CEACAM3 might be responsible for the stimulation of the small GTPase Rac in response to Opa_CEA gonococci.

Since the possession of the cytoplasmic domain is a distinctive feature of CEACAM3 that is absent from GPI-anchored CEACAM6, we wondered whether residues in the cytoplasmic domain of CEACAM3 are responsible for the internalization and/or for the observed stimulation of Rac. Therefore, a panel of CEACAM3 mutants were generated containing a deletion of the cytoplasmic domain (ΔCT) or point mutations disrupting the ITAM-like sequence (Fig. 5 A) by substitutions of tyrosine for phenylalanine at either one or both positions (Y230F, Y241F, Y230/241F). In addition, a COOH-terminal HA tag was introduced. When 293 cells were transiently transfected with the respective constructs, equivalent levels of the proteins were expressed (Fig. 5 B). Importantly, uptake mediated by HA-tagged WT CEACAM3 was comparable to the internalization observed before with the untagged receptor, demonstrating that the addition of the HA tag did not alter the functionality of the protein (Fig. 5 C). However, deletion of the cytoplasmic domain of CEACAM3 (ΔCT) led to almost complete abolishment of internalization of Opa_CEA gonococci (Fig. 5 C). The ITAM-like sequence of CEACAM3 seemed to play a critical role, since substitution of either Y230 or Y241 resulted in ∼30% reduction in bacterial uptake. Simultaneous mutation of both tyrosine residues resulted in a >60% reduction of internalization (Fig. 5 C). Nevertheless, uptake mediated by CEACAM3 Y230/241F was consistently higher than internalization via CEACAM3 ΔCT, suggesting that additional residues within the cytoplasmic domain aside from the ITAM-like sequence can contribute to a low level of uptake. To analyze if the cytoplasmic domain is required for stimulation of Rac, pull-down analyses were performed. As observed before, infection of CEACAM3-expressing 293 cells with opaque gonococci resulted in enhanced GTP-loading of Rac (Fig. 5 D). In contrast, pcDNA, CEACAM3 ΔCT, or CEACAM3 Y230/241F-transfected cells did not respond to the bacterial challenge (Fig. 5 D) supporting the view that the ITAM-like sequence in the cytoplasmic domain of CEACAM3 links receptor engagement with enhanced GTP-loading of Rac.

Purified c-Src phosphorylates the CEACAM3 cytoplasmic domain in vitro (27), and Src PTKs are activated and phosphorylate CEACAM3 in granulocytes in response to opaque gonococci (17, 28). To analyze if Src PTKs play a role in CEACAM3-mediated bacterial internalization and Rac stimulation, a kinase-inactive version of c-Src (Src K297M) was overexpressed in 293 cells. Notably, cells coexpressing CEACAM3 and Src K297M showed a prominent decrease in uptake of Opa_CEA N. gonorrhoeae (Fig. 6 A). To investigate if Src PTK activity is upstream of Rac stimulation by CEACAM3, lysates were analyzed by pull-down assays with the GST-CRIB domain. Again, CEACAM3-transfected cells showed an increase in GTP-Rac upon infection with Opa_CEA pathogens (Fig. 6 B). However, in cells coexpressing CEACAM3 and Src K297M the increase in GTP-Rac was completely suppressed (Fig. 6 B). Together, these results suggest that CEACAM3-mediated stimulation of the small GTPase Rac depends on Src kinase–initiated signaling via the CEACAM3 cytoplasmic domain ITAM-like sequence.

To address the functional significance of CEACAM3-initiated processes in human granulocytes, we analyzed the opsonin-independent phagocytosis of bacteria by primary granulocytes. Therefore, phagocytes were incubated with FITC-labeled bacteria, and internalization of bacteria was determined by FACS^® analysis after quenching with trypan blue (23). Control experiments showed that addition of trypan blue reduced the fluorescence of extracellular bacteria by >90% (Fig. S1 A). As expected, nonopaque gonococci were only poorly phagocytosed (depending on the donor, 5–10% of the granulocyte population were FITC positive) and yielded a signal of intracellular fluorescence comparable to uninfected cells, whereas granulocytes readily internalized Opa_CEA-expressing bacteria (75–95% of the granulocyte population) (Fig. 7 A). However, in the presence of the CEACAM3-specific monoclonal antibody a decreased phagocytosis of Opa_CEA gonococci was observed, whereas addition of an isotype-matched control antibody against endoglin or addition of a CEACAM6-recognizing antibody only marginally interfered with internalization (Fig. 7 B). PP2, a specific pharmacological inhibitor of Src family PTKs, blocked the uptake of Opa_CEA gonococci to the same extent as CEACAM3-specific antibodies (Fig. 7 B). This observation is in agreement with the view that CEACAM3-mediated phagocytosis requires Src PTK activity. The reduced fluorescence signal from intracellular bacteria in the presence of the CEACAM3-specific antibodies was indeed due to decreased internalization of bacteria, as shown by differential staining of intra- and extracellular bacteria in infected samples (Fig. 7 C). In the absence of antibodies or in the presence of a CEACAM6-specific antibody, numerous Opa_CEA-expressing N. gonorrhoeae were located inside the primary granulocytes. However, a marked reduction of intracellular bacteria could be observed in the presence of the CEACAM3-specific antibody (Fig. 7 C), suggesting that CEACAM3 is the predominant CEACAM family member responsible for the opsonin-independent uptake of CEACAM-binding gonococci by human granulocytes.

If CEACAM3 is the major CEACAM family member contributing to the uptake of Opa_CEA gonococci, then interference with Rac stimulation should also affect the internalization of the bacteria. Therefore, granulocytes were pretreated with a purified TAT fusion of dominant-negative Rac (TAT-RacN17), β-galactosidase (TAT-LacZ), or dominant-negative Cdc42 (TAT-Cdc42N17) (Fig. S1 B) and then incubated with FITC-labeled gonococci. Importantly, TAT-RacN17 severely diminished the opsonin-independent phagocytosis of Opa_CEA N. gonorrhoeae by human granulocytes in a dose-dependent manner, whereas neither TAT-LacZ nor TAT-Cdc42N17 interfered with this process (Fig. 8 A). These results further support the hypothesis that CEACAM3 is the major granulocyte receptor of the CEACAM family mediating uptake of Opa_CEA gonococci.

To investigate if the efficient CEACAM3-mediated phagocytosis finally leads to the killing and elimination of the microorganisms, primary human granulocytes were incubated with Opa_CEA gonococci, and the amount of viable bacteria was determined after 60 min of coincubation. As observed before, ∼60% of the inoculated Opa_CEA-expressing, but not nonopaque gonococci, were eliminated by the phagocytes within 60 min (Fig. 8 B). However, when the CEACAM3-specific antibody was added before infection an improved survival of Opa_CEA gonococci could be observed, whereas addition of the isotype-matched control antibody or a CEACAM6-recognizing antibody only marginally altered the survival of the bacteria (Fig. 8 B). Together, these results suggest that CEACAM3-mediated Rac stimulation is a major pathway leading to the opsonin-independent uptake and elimination of CEACAM-binding gonococci by human granulocytes.

To address if CEACAM3 plays a significant role in the elimination and control of other CEACAM-binding pathogens, we investigated if M. catarrhalis or H. influenzae are recognized by CEACAM3 and if they trigger Rac-GTP loading. Indeed, in CEACAM3-expressing 293 cells a nonencapsulated variant of H. influenzae strain RD stimulated a strong increase in Rac-GTP upon infection, whereas no such increase was observed in cells transfected with the empty control vector (Fig. 9 A). Again, Rac-GTP loading in response to H. influenzae was correlated with enhanced CEACAM3-mediated internalization of these pathogens, albeit at lower levels than observed in response to Opa_CEA gonococci (Fig. 9 B). Moreover, a UspA1-positive M. catarrhalis strain enhanced Rac-GTP loading and was internalized by 293 cells in a CEACAM3-dependent manner (Fig. 9, C and D). Interference with CEACAM3, and to a lesser extent with CEACAM6, on primary human granulocytes decreased phagocytosis of H. influenzae, and uptake of M. catarrhalis was strongly diminished by CEACAM3-specific antibodies (Fig. 9 E), suggesting that CEACAM3 is involved in clearance of diverse microorganisms that target human CEACAMs. Together, these data support the view that CEACAM3 is a major granulocyte receptor responsible for Rac stimulation by an opsonin-independent phagocytosis of diverse CEACAM-binding, human-specific pathogens.

The human pathogens N. gonorrhoeae, N. meningitidis, M. catarrhalis, and H. influenzae avoid acquired immune defenses of their host by sophisticated mechanisms such as variation of surface structures or secretion of IgA-specific proteases. However, despite widespread colonization with some of these microorganisms, severe disease outcomes occur only in a small portion of the host population, suggesting that innate immune defenses are able to contain the bacteria in most cases. We demonstrate here that human granulocytes not only recognize CEACAM-binding variants of these pathogens but that bacterial engagement of a granulocyte-specific member of the CEACAM family, CEACAM3, initiates phagocyte effector mechanisms leading to the elimination of the microbes.

It has been observed previously that human granulocytes can recognize opaque variants of pathogenic Neisseriae in an opsonin-independent manner (29) and that this recognition depends on members of the CEACAM family (4, 5, 17). Overexpression of individual CEACAM family members in nonphagocytic cells has revealed that several CEACAMs can mediate uptake in such a cellular background (11, 12, 18). However, the role of individual receptors in the interaction of granulocytes with CEACAM-binding pathogens has not been analyzed. The data presented in this study provide for the first time biochemical and functional evidence that CEACAM3 is the predominant CEACAM family member on human granulocytes mediating elimination of CEACAM-binding bacteria. Since there is no endogenous ligand known for CEACAM3 (13), the recognition of pathogenic Haemophilus, Moraxella, and Neisseria species now assigns a function to this orphan receptor of the innate immune system.

The efficient CEACAM3-mediated clearance of bacterial pathogens seems to be closely linked to the ability of this receptor to stimulate the small GTPase Rac, a critical regulator of actin polymerization, phagocytosis, and the oxidative response of granulocytes (30). Though CEACAM6 can mediate internalization when overexpressed in epithelial cell lines (11, 12), bacterial engagement of CEACAM6 did not lead to Rac stimulation, and interference with CEACAM6 did not block the clearance of CEACAM-binding bacteria by human granulocytes. The difference between CEACAM3- and CEACAM6-mediated events is due to the presence of an ITAM-like sequence in the CEACAM3 cytoplasmic tail. ITAMs have been characterized in several receptors belonging to the Ig superfamily (31). ITAM function depends on the phosphorylation of key tyrosine residues that is often mediated by Src family protein tyrosine kinases (32). Interestingly, tyrosine residue Y241 within the CEACAM3 ITAM-like sequence is embedded in the context of YxxM, where a methionine residue is located at the +3 position with respect to the tyrosine (compared with YxxL/I in the canonical ITAM [31]). Such a YxxM sequence has been shown to serve as a preferred docking site for SH2 domains of the p85 subunit of phosphatidylinositol 3′-kinase (PI-3K) (33). Therefore, we would predict that phosphorylated Y241 connects phosphorylation of CEACAM3 to the reported recruitment of PI-3K (34). Since the products of PI-3K can activate several Rac GEFs, such as Vav and SWAP70 (35, 36), PI-3K binding is a potential molecular link between CEACAM3 ITAM phosphorylation and downstream Rac stimulation.

CEACAM3 does not engage in homo- or heterotypic interactions with other CEACAM family members, and endogenous ligands are unknown (37). In light of the results presented in this paper, CEACAM3 is best viewed as a specifically adapted single chain phagocytic receptor involved in the clearance of CEACAM-binding bacteria by human granulocytes. Most of the known CEACAM-binding bacterial species are human-specific pathogens. Likewise, the expression of CEACAM3 is also restricted to human granulocytes, and no homologues of CEACAM3 have been detected in rodents or nonhuman primates (38, 39), suggesting a striking coevolution of human innate immune defenses with a restricted set of specialized pathogens. It is tempting to speculate that gonorrhea-associated complications effecting reproduction such as early trimester abortion or infertility (40) might have provided a selective advantage to individuals able to counteract CEACAM-binding bacteria by innate immune defenses. On the other hand, pathogens might evolve variants of their CEACAM-binding adhesins that retain association with CEACAM family members expressed on epithelial cells (such as CEA or CEACAM1) but that loose recognition by CEACAM3. Such adhesins would allow attachment to and invasion of epithelial cells but would avoid elimination by granulocytes. Indeed, several meningococcal and gonococcal Opa_CEA proteins have been characterized that exhibit this property (11, 12, 41, 42). It will be interesting to investigate whether the ability to cause severe forms of disease is correlated with the expression of CEACAM-binding adhesins that are not recognized by CEACAM3.

Together, the results presented in this paper describe the orphan receptor CEACAM3 as a novel single-chain phagocytic receptor of the innate immune system. Since internalization via CEACAM3 is opsonin independent, this receptor should be of particular importance in the control of bacteria that do not induce protective acquired immune responses. Since multiple commensal and pathogenic bacteria found in association with human mucosal surfaces are exploiting CEACAM family members (43–45), CEACAM3-mediated, opsonin-independent phagocytosis by granulocytes might be a general defense mechanism evolved by the human innate immune system directed against CEACAM-binding microorganisms.

We are indebted to T.F. Meyer for the Neisseria strains used in this study. We thank W. Zimmermann for CEACAM cDNAs; J. Berger and S. Ackermann-Hauck for help with scanning electron microscopy and coloration; M. Achtman, S. Dowdy, E. Gulbins (Universität Essen, Essen, Germany), A. Reidl, and D. Schlaepfer for reagents; and J. Blaß and D. Deininger for expert technical assistance.

This study was supported by funds from the Bundesministerium für Bildung und Forschung (01 KI 8906/0), Deutsche Forschungsgemeinschaft (Ha2568/3-1), and Fonds der Chemischen Industrie to C.R. Hauck.