The CGM1a (CEACAM3/CD66d)-mediated phagocytic pathway of Neisseria gonorrhoeae expressing opacity proteins is also the pathway to cell death.

Phagocytosis of Opa+ Neisseria gonorrhoeae (gonococcus, GC) by neutrophils is in part dependent on the interaction of Opa proteins with CGM1a (CEACAM3/CD66d) antigens, a neutrophil-specific receptor. However, the signaling pathways leading to phagocytosis have not been characterized. Here we show that interaction of OpaI bacteria with neutrophils or CGM1a-transfected DT40 cells induces calcium flux, which correlates with phagocytosis of bacteria. We identified an immunoreceptor tyrosine-based activation motif (ITAM) in CGM1a, and showed that the ability of CGM1a to transduce signals and mediate phagocytosis was abolished by mutation of the ITAM tyrosines. We also demonstrated that CGM1a-ITAM-mediated bacterial phagocytosis is dependent on Syk and phospholipase C activity in DT40 cells. Unexpectedly, the activation of the CGM1a-ITAM phagocytic pathway by Opa+ GC results in induction of cell death.

Gonorrhea is one of the most frequently reported sexually transmitted diseases (1). Neisseria gonorrhoeae (gonococcus, GC), 1 the etiologic agent of gonorrhea, can adhere to and penetrate mucosal cells (2,3) and attain access to submucosal sites. Among the GC surface proteins that mediate this process is the opacity (Opa) protein family. This family consists of 11 unlinked opa genes, whose sequences are known (4). In addition, each Opa protein is able to switch its expression on and off, resulting in the unavoidable tendency of GC to alter their phenotypes by antigenic variation. Opa-expressing GC and Escherichia coli (Opa ϩ GC or E. coli) can attach to and invade human fallopian tube epithelium (5), indicating that Opa proteins alone are sufficient to promote adherence to and invasion of human cells. Furthermore human challenge experiments (6,7) strongly suggest that in vivo expression of Opa proteins plays an important role in gonococcal pathogenesis on the mucosal surface.
It is well recognized that Opa proteins have the ability to stimulate adherence and phagocytosis of the Opa ϩ bacteria by polymorphonuclear leukocytes (PMN). This interaction with PMN occurs in an opsonin-independent manner (8 -12). In addition, Opa ϩ bacteria stimulate a chemiluminescent response in human neutrophils as a result of oxidative burst activity. Furthermore, recent studies demonstrated that members of the carcinoembryonic antigen (CEA) or CD66 family serve as receptors mediating adherence and possibly phagocytosis of Opa ϩ bacteria (13)(14)(15)(16). These findings were mainly obtained with HeLa transfectants expressing individual CEA family members. However, neither the nature of the signals nor the pathways that are involved in phagocytosis of the bacteria are clear.
Various components of the B cell receptors (BCR), T cell receptors (TCR), and Fc receptors (FcR) contain common sequence motifs in their cytoplasmic tails, called the immunoreceptor tyrosine-based activation motif (ITAM; Ref. 21). The phosphorylated tyrosine residues within ITAMs can recruit protein-tyrosine kinases (PTK), e.g. Syk, whereupon these become activated. Recruitment of substrates such as phospholipase C-␥ (PLC-␥) by these kinases leads to stimulation of calcium flux from intracellular stores (22)(23)(24)(25). This calcium flux reflects an early event after activation of PTK.
In this study, we demonstrate that CGM1a bears a func-* This work was supported by Public Health Service Grants AI 47736 (to T. C.) and AI 26558 (to E. G.) and a grant from the Deutsche Forschungsgemeinschaft (to W. Z.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  tional ITAM in its cytoplasmic tail, which mediates phagocytosis of Opa ϩ bacteria followed by cell death.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Cell Culture, and Antibodies-GC strain MS11 was cultured and maintained as previously described (26). Only pilus Ϫ GC and LOS b (lacto-N-neotetraose) phenotype were used (27). Recombinant opa genes from GC MS11 were constructed and expressed in E. coli HB101 as described previously (28). The designations of Opa proteins of both GC and E. coli are based on Swanson et al. (6) and Belland et al. (28). The Opa ϩ bacteria used in this study are Opa Ϫ , OpaC, OpaD, OpaF, and OpaI GC or Opa Ϫ and OpaI E. coli.
Preparation of Neutrophils-Neutrophils from human blood were purified by centrifugation through Polymorphprep (Life Technologies, Inc.). The purified PMN were suspended either in RPMI or Hanks' balanced salt solution (HBSS; Cellgro, Herndon, VA) at a concentration of 2 ϫ 10 6 /ml.
Transfections-CGM1a cDNA and its mutants were cloned into pBEH expression vector, and NCA and BGPa were cloned into pRC/CMV vector. 10 g of expression plasmids were co-transfected with 1 g of pBabe-puro r vector (31) into 5 ϫ 10 6 DT40 cells by electroporation at 250 V and 960 F in 0.5 ml. Stable transfectants were selected in 0.5 g/ml puromycin 24 h after electroporation. The presence of CD66 antigens in individual clones was determined by flow cytometry analysis using FACScan (Becton Dickinson, Mountain View, CA) analysis with COL-1, YTH71.3., and secondary antibodies conjugated with FITC.
Calcium Measurements-DT40 cells or neutrophils (4 ϫ 10 6 ) were suspended in 3 ml of RPMI medium. 6 fl of 1 M Fura-2AM (Molecular Probe, Eugene, OR) dissolved in Me 2 SO were added to the suspension, which was then incubated in the dark at 37°C for 30 min with occasional shaking. Cells were washed twice with HBSS and resuspended in 3 ml of HBSS. 0.5 ml of cell suspension was mixed with 1 ml of phosphate-buffered saline containing 1 mM CaCl 2 and 1 mM MgCl 2 . For activation assays, 0.1 ml of the bacterial suspensions at a concentration of 4 ϫ 10 8 was added to the labeled cells. The cytosolic calcium concentration was determined with a fluorescence spectrophotometer (LB50B, Perkin Elmer, Foster City, California) at an excitation wavelength of 340 and 360 nm, and an emission wavelength of 510 nm. Calculation of the calcium concentration was performed using the FL WinLab software (Perkin Elmer). We should state here that we usually test three (at least two) different transfectants of each mutant to make sure that they behave equally. It should be noted that anti-receptor antibodies are often used to stimulate the activation of receptors such as BCR and TCR. However, none of the anti-CEA antibodies we tested (COL-1, B1.1, 4 -12-5, and YTH71.3) was able to stimulate calcium flux in DT40-CGM1a, even when cross-linked with secondary antibodies.
Phagocytosis Assays-DT40 cells transfected with CGM1a (DT40-CGM1a) and control DT40 cells were suspended in RPMI with 2% fetal calf serum at the concentration of 2-4 ϫ 10 5 /ml. 0.5 ml each of these cell suspensions was added to 24-well plates and after addition of 50 l of bacterial suspensions in RPMI at the concentration of 8 ϫ 10 7 , the cells were allowed to incubate for 3 h at 37°C in the presence of 5% CO 2 . Then gentamicin was added into each well to the final concentration of 100 g/ml and incubated for another 90 min. The cells were washed three times with RPMI and lysed in phosphate-buffered saline containing 0.5% saponin (CalBiochem Corp.), and dilutions were plated on LB-agar or CG-agar. The level of internalization of bacteria into cells was calculated by determining the colony-forming units (cfu) recovered from the DT40 cell lysates.
Detection of Cell Death-Annexin V-FITC binding to phosphatidylserine (PS) was used as the cell death assay according to the manufacturer's recommendations (PharMingen, San Diego, CA). Annexin V-FITC was used in conjunction with propidium iodide (PI) to distinguish apoptotic cells (annexin V-FITC positive, PI negative) from necrotic cells (annexin V-FITC positive, PI positive). Cells exhibited apoptotic cell-specific phosphatidylserine on their surface and, therefore, bound annexin V-FITC or took up PI, which cannot penetrate live cells. However, in our experiment, the cells that were either annexin V-FITC positive or PI positive were classified as dead cells (see Fig. 6). Neutrophils or DT40 cells (1 ϫ 10 5 ) were suspended in 0.1 ml of 2% fetal calf serum in RPMI containing 1.5 ϫ 10 6 bacteria (GC) and incubated at 37°C for 6 h with occasional shaking. 50 l of binding buffer (10 mM Hepes/NaOH, pH 7.4, 140 mM NaCl, and 2.5 mM CaCl 2 ), 7.5 l of Annexin V-FITC and 15 l of PI were added in each vial. The cell suspensions were gently mixed and incubated for 15 min at room temperature in the dark. Finally, after addition of 550 l of binding buffer, samples were analyzed immediately by flow cytometry.

Opa ϩ GC Stimulates Calcium Flux in Neutrophils and DT40-CGM1a but Not in DT40-BGPa or DT40-NCA-Studies
have indicated that Opa protein-mediated binding and phagocytosis of GC by PMN occurs through the CEA family members on PMN (13,14,16). Both BGPa and CGM1a contain possible ITAM-like motifs (Fig. 1A). NCA, which is membrane anchored by a glycosylphosphatidyl inositol (GPI) moiety, is expressed on granulocytes and has also been shown to bind to Opa proteins. To examine whether the interaction between GC and neutrophils results in any signaling in the PMNs, measurement of calcium flux was performed. OpaI GC were used to stimulate neutrophils. As shown in Fig. 2A, OpaI GC but not Opa Ϫ GC induced calcium flux in neutrophils, suggesting that at least one of the three CEA family members in question conveys these signals, possibly by an ITAM.
To determine which of the three CEA-related receptors is responsible for the calcium flux in neutrophils induced by Opa ϩ GC ( Fig. 2A), the DT40 B cell line was chosen to perform functional studies. There are several reasons why we chose this cell line. First, the ITAM, which could be responsible for the signaling through the CEA-related molecules, was originally identified in the Ig␣/Ig␤ heterodimer of the BCR (32,33), and DT40 B cells have proven to be suitable for studying the functionality of presumed ITAMs (34, 35) e.g. by induction of calcium flux as observed after BCR activation. Furthermore, human neutrophils are terminally differentiated cells and cannot be genetically manipulated. Manipulation of established myeloid lines of human origin such as HL60 cells cannot be performed readily either because endogenous CD66 antigen expression changes during induction of maturation by retinoic acid 4 . Finally, no Opa ϩ GC were found to interact exclusively with individual CEA antigens like CGM1a (36,37). In short, a cellular system that lacks endogenous human CEA antigens and can be manipulated, like the DT40 cell line, is needed to understand the behavior of CEA-related molecules in neutrophils. We believe that the DT40 cell line is the best model available.
CGM1a, BGPa (both represent splice variants that code for tyrosine-containing cytoplasmic tails) and NCA cDNAs were cloned into pBEH or pRc/CMV expression vectors, and the resulting plasmids were stably transfected into the DT40 B cell line, generating the DT40-CGM1a, DT40-BGPa, and DT40-NCA transfectants. All three cell lines expressed similar levels of the respective CEA-related antigens (Figs. 2B and 3). When OpaI GC were added to these three cell lines, calcium flux was only observed in the DT40-CGM1a transfectant (Fig. 2B). In contrast, Opa Ϫ GC could not stimulate any calcium flux. Likewise, no calcium flux was detected in untransfected DT40 cells treated with OpaI GC. These results show that only CGM1a is able to convey the calcium flux and suggest that it contains a functional ITAM.

Mutation of Tyrosine Residue Tyr-196 in CGM1a
Abolishes Its Signaling Capability-Phosphorylation of tyrosine residues within an ITAM is a key step for its activation. The cytoplasmic domain of CGM1a contains two tyrosine residues in its potential ITAM. To test whether these tyrosine residues were required for activation of the calcium flux, we substituted them with phenylalanine. Three different tyrosine mutants, CGM1a-Y196F, CGM1a-Y207F, and CGM1a-Y196F/Y207F (Fig. 1B), were constructed and the resulting expression plasmids were transfected into DT40 cells. Transfectants that expressed similar levels of the mutated molecules were selected for the calcium flux experiments (Fig. 3). As depicted in Fig. 3, mutation of tyrosine residue 196 (Y196F) or mutation of both tyrosines in the cytoplasmic tail of CGM1a (Y196F/Y207F) completely abolished its signaling ability as measured by calcium flux. On the other hand, OpaI GC were able to stimulate a low level of calcium flux in the Y207F CGM1a mutant. These results indicate that both tyrosines are necessary to constitute a fully functional ITAM with Tyr-196 being the essential one. This mutation analysis of the cytoplasmic tyrosines of CGM1a confirms that the motif surrounding these tyrosines (YX 2 LX 7 YX 2 M) is a functional ITAM.
The ITAM in CGM1a Mediates Phagocytosis of Opa ϩ Bacteria-CGM1a promotes strong phagocytosis of Opa ϩ bacteria. We wanted to know whether signaling through the ITAM in the cytoplasmic domain of CGM1a is involved in this process. First, using a colony-forming assay to quantify the number of internalized bacteria, we demonstrated that DT40-CGM1a cells are also capable of phagocytosing OpaI GC or E. coli but not Opa Ϫ GC or E. coli (Fig. 4A). This result was confirmed by electron microscopy analysis (Fig. 4, B and C). To elucidate the importance of the tyrosines in this process, the tyrosine mutants were tested in the phagocytosis assay. As in the calcium flux experiments, mutation of Tyr-196 or mutation of Tyr-196 and Tyr-207 in the ITAM severely impaired or abolished, respectively, the ability of CGM1a to confer phagocytosis of OpaI E. coli (Fig. 4D). However, some capacity to phagocytose OpaI E. coli was still preserved in DT40 cells expressing CGM1a-Y207F. From these results, we conclude that the ITAM in CGM1a mediates both calcium flux and the phagocytosis of Opa ϩ bacteria.
Syk and PLC-␥ Enzymes Participate in Phagocytosis of Opa ϩ Bacteria-We tested the ability of DT40-CGM1a cells lacking Syk and PLC-␥ activity to mediate calcium flux and phagocytosis, because the downstream pathway of ITAM activation involves Syk, whose activation in turn leads to PLC-␥ stimulation. DT40-Syk-CGM1a or DT40-PLC-␥-CGM1a cells are CGM1a-transfected DT40 cells in which the Syk or PLC-␥ genes have been inactivated. As depicted in Fig. 5A and B, DT40-Syk-CGM1a cells cannot be induced to generate a calcium flux or to phagocytose OpaI bacteria. However, DT40-PLC-␥-CGM1a cells can still phagocytose OpaI E. coli at low level although stimulation of calcium flux could not be detected (Fig. 5). This suggests that an increased intracellular Ca 2ϩ concentration is not essential for phagocytosis. These data demonstrate that both Syk and PLC participate in the path-ways leading to phagocytosis of Opa ϩ bacteria, but Syk is essential for this process.

Activation of the ITAM in CGM1a (CEACAM3/CD66d) Results in Cell Death in DT-40 Cells-Signal transduction via
BCR activation results very complicated consequences. However, one of them leads to apoptosis of B cells (38). We examined whether activation of CGM1a ITAM had a similar effect on DT-40 cells. As shown in Fig. 6, OpaI GC stimulated substantially more cell death in DT40-CGM1a as measured by both phosphatidylserine exposure and uptake of propidium iodine than Opa Ϫ GC did. Untransfected DT40 B cells, used as a control, could not be killed by OpaI or Opa Ϫ GC (Fig. 6). The observed cell death is dependent on ITAM activation because CGM1a-Y196F and CGM1a-Y196F/Y207F mutants (Fig. 1) lost the ability to promote cell death (Fig. 6). Some cell death was observed in DT40-CGM1a with Opa Ϫ GC, which might be because of low numbers of Opa ϩ GC in the Opa Ϫ GC suspension because GC are able to switch Opa gene expression on and off. Alternatively, CGM1a might be activated by other components on GC, which are still unidentified. It should be noted that the anti-CGM1a specific antibody COL-1, which was unable to induce calcium flux, could not stimulate the death DT40-CGM1a either (data not shown).
To further validate the effects of the CGM1a-ITAM activation pathway, DT40-Syk-CGM1a and DT40-PLC-␥-CGM1a cells were tested for cell death. As shown in Fig. 5, OpaI GC was not able to promote death of cells lacking Syk, but still killed a significant fraction of DT40-PLC-␥-CGM1a cells. The similarity between the patterns of bacterial phagocytosis and cell death suggests that these two processes share the same initiating events.

Killing of Neutrophils by GC Depends on Expression of Individual Opa
Proteins-We showed that activation of the phagocytic pathway via CGM1a led to cell death in DT40 cells. Thus, the question arises whether the CGM1a-ITAM pathway to cell death can also be observed in neutrophils, which are the only cells naturally expressing CGM1a. However, neutrophils cannot be genetically manipulated and so far none of the myeloid cell lines currently available fully resemble neutrophils and thus can convincingly replace PMNs. Furthermore, antibodies specific for CGM1a do not activate neutrophils.
Fortunately, recent data did show that various Opa proteins interacted differently with CGM1a and BGPa. For example, OpaC or OpaI interact with both CGM1a and BGPa, and OpaD or OpaF interact with BGPa only (36,37). Thus, the rational is that if the CGM1a-ITAM pathway, demonstrated in DT40 cells, also applies to neutrophils, only GC-expressing Opa proteins that interact with CGM1a (such as OpaC or OpaI) should stimulate the death of neutrophils, as opposed to OpaD or OpaF-expressing GC. Opa Ϫ , OpaC, D, F, and I GC were used to perform similar experiments with neutrophils as done with the B cell lines in Fig. 6. As shown in Fig. 7, OpaC and I GC killed many more neutrophils than Opa Ϫ and OpaD or OpaF GC did. It should be noted that GC expressing these different Opa proteins have the similar ability to associate with neutrophils (39). These results suggest that activation of the ITAM in CGM1a also stimulates cell death in neutrophils.

DISCUSSION
Internalization of microorganisms into either professional or non-professional phagocytic cells requires the interaction of phagocytosis-promoting receptors on the cell surface with ligands on the surface of the microorganisms. In the present study, we demonstrated that the neutrophil-specific receptor, CGM1a contains an ITAM, which is essential for CGM1amediated phagocytosis of Opa ϩ bacteria. This conclusion is further validated by showing that CGM1a-transfected, Syk-  Fig. 4 and cell death in Fig. 6 were followed, respectively. DT40-Syk-CGM1a lost all these abilities, whereas DT40-PLC-␥-CGM1a preserved some capacity to phagocytose bacteria. deleted DT40 cells are no longer able to promote effectively either calcium ion influx or phagocytosis of OpaI bacteria. Finally, we showed that CGM1a-mediated phagocytosis of opacity (Opa) proteins-expressing Neisseria gonorrhoeae leads to cell death.
Because CGM1a is only expressed in PMNs, it is a concern whether CGM1a-mediated signaling data obtained from the B cell line DT40 reflects its biological properties in neutrophils. ITAM-containing receptors have nearly exclusively been identified in cells from immune system. Despite the differences in the structures they recognize and the effector functions they carry out, BCRs, TCRs, and several FcRs utilize remarkably similar signal transduction components to initiate and propagate their signaling responses. They use components that contain distinct recognition (ITAM) and signal transduction subunits such as Syk, with which the ITAM in CGM1a may be also associated (Fig. 5). This type of receptors is crucial for transmission of activation signals in immune cells. Here, we show that the human CGM1a, although ectopically expressed in a chicken B cell line, exhibits the ITAM characteristics of BCRs, supporting the notion that ITAMs are highly conserved motifs, and that the CGM1a ITAM might play a similar role in neutrophils.
Opa ϩ Neisseria interact with BGP and CGM1 on neutrophils during in vitro infection, and both antigens promote phagocytosis of Opa ϩ bacteria in HeLa transfectants (13,14,16). Do CGM1a and BGPa work independently or do they collaborate during phagocytosis of GC by neutrophils? What kind of collaboration could be envisaged? BGPa can deliver inhibitory signals through its ITIM in the cytoplasmic tail to counteract ITAM-mediated positive signals, which requires SHP-1 and SHP-2. 2 The protein-tyrosine phosphatases SHP-1 and SHP-2, which are typically involved in negative signaling, bind to the ITIM-like motif in the cytoplasmic tail of BGPa (40,41). Indeed, it has recently been shown that interaction of Opa ϩ GC with BGPa antigens down-regulates the activity of SHP-1 (42). Therefore, BGPa may be co-ligated with CGM1a on the surface of neutrophils when they interact with Opa ϩ bacteria. Crosslinkage of the two molecules might regulate the phagocytosis process. Opa ϩ GC-stimulated signal transduction events in neutrophils might consequently be analogous to the positive (ITAM) versus negative (ITIM) signaling seen with lymphocyte receptors such as BCR and FcRIIB, respectively (43).
In addition, this study raises an interesting issue. This is the first report of phagocytosis of bacteria by B cells. Antigenimmunoreceptor interaction is an important step for the establishment of the immune response, because there is signaling from the receptor to activate different cellular functions. Internalization of antigen-bound receptors by B cells is critical for antigen processing and presentation to T lymphocytes. Here we show that DT40-CGM1a B cells were able to phagocytose Opa ϩ bacteria (either GC or E. coli) in a very efficient manner. Although the cell line we used was manipulated to express CGM1a antigen, our data, nevertheless, suggest that B cells possess the machinery to phagocytose whole bacteria. Demonstration of the ability of B cells to phagocytose bacteria is a new information with a potential.
What is the biological fate of Opa ϩ GC after phagocytosis by PMN? It is generally thought that the phagocytosed gonococci are killed. However, neutrophils are often packed with intact, seemingly viable, intracellular GC (44). Thus, the possibility exists that these Opa ϩ GC may be protected from bactericidal attack by the granulocytes and instead use these cells as an intracellular niche or as vehicles to reach the next host. Alternatively, they could simply kill the PMNs. In fact, it was recently found that the ITAM activation pathway also induces cell apoptosis in immune cells (23,38,(45)(46)(47)(48)(49). For example, activation of the ITAM in the BCR of DT40 B cells results in apoptosis (38,48). In this context it is interesting to note that OpaI and OpaC GC but not OpaD and OpaF GC stimulate the death of neutrophils (Fig. 7) although the level of interaction of OpaC, D, F, and I GC with neutrophils is similar (39). Perhaps, phagocytosis of Opa ϩ GC by neutrophils through CGM1a leads in part to apoptosis of these cells. In addition, compared with other cells from immune system, there is not much known of the mechanisms of cell death in PMNs, even though these cells are noted for their death fate. In humans, PMNs succumb to apoptosis within 72 h. Thus, finding that expression of ITAMcontaining CGM1a, a specific PMN antigen, mediated the death in B-cell and possibly in neutrophils is very significant. Whether BGPa mediated internalization of Opa ϩ GC in neutrophils could reduce apoptosis of neutrophils remains to be determined, because BGPa contains an ITIM to counteract the activated ITAM. 2 It may seem to a paradox that host cells phagocytosing microorganisms kill them and also may be killed by microorganisms. However, the two aspects might in reality represent two separate battlefields in the interaction between bacteria (GC) and host cells (PMNs). This phenomenon also occurs in interaction between other pathogenic bacteria and host cells. For example, whether the outcome of macrophages phagocytosing Shigella or Salmonella is the killing of the microorganisms or the death of the host cell (50,51) remains to be defined in vivo. The study of the interaction of Opa ϩ bacteria with host cells through CEA (CEACAM/CD66) antigens addresses two facets of phagocytosis of GC by PMNs, which are critical for the understanding of the pathogenesis of gonococcal infection.