Neisseria gonorrhoeae (GC) is a human pathogen that adheres to and invades genital surfaces. Although pili are required for the initial adherence, the interaction of GC with epithelial cells is also promoted by a family of outer membrane proteins, the opacity (Opa) proteins such as OpaA protein from strain MS11. Studies have demonstrated that the interaction of the OpaA GC with epithelial cells involves binding to heparan sulfate attached to syndecan receptors. However, other Opa proteins interact with CEA gene family member 1 (CGM1) or biliary glycoprotein (BGP), members of the CD66 antigen family. In this study, we demonstrate that, in addition, the 180-kD carcinoembryonic antigen (CEA) is a receptor for Opa proteins. This conclusion was based on the following observations. First, transfected HeLa cells expressing CEA (HeLaCEA) and the CEA-expressing colon cancer cell line (LS 174T) bound and subsequently engulfed the Opa+ bacteria. These interactions were inhibited by anti-CEA antibody, but could not be inhibited by addition of heparin. Furthermore, OpaI E. coli directly bound purified CEA. We also compared the adherence and invasion by Opa+ bacteria of CD66 transfected HeLa cells: HeLa-BGPa, HeLa-CGM6, HeLa-NCA, HeLa-CGM1a, HeLa-CEA, and HeLa-Neo serving as negative control. Using OpaI as the prototype, the relative ability of the transfected HeLa cell lines to support adherence was (CEA = BGPa >CGM1a >NCA >>CGM6 = Neo). The ability to mediate invasion of the transfectant cells was (CGM1a >CEA >BGPa >NCA >CGM6 = Neo). Among the Opa proteins tested, OpaC proved to be bifunctional, able to mediate adherence to both syndecan receptors and to CD66 antigens.
Neisseria gonorrhoeae (GC)1 cause uncomplicated gonorrhea, pelvic inflammatory disease, disseminated gonococcal infection, and ophthalmia neonatorum. These diseases result from the ability of GC to adhere to and penetrate mucosal epithelial cells (1, 2). Pili are the components promoting the initial adherence and infection by GC (3–5).
Pathobiological studies of gonococcal infection are hampered by the highly restricted species specificity of gonorrheal disease so that there is no good animal model available currently. The most informative model has been in vitro infection of human fallopian tube organ culture (FTOC; 6, 7). Initial attachment of GC to FTOC mucosal columnar epithelial cells is mediated by pili. About 18 h after infection, the attached organisms are endocytosed by the nonciliated cell. Phagocytosed GC appear to be rapidly transcytosed through the epithelial cells and enter the submucosal space at the base of the cells (8). Expression of a family of outer membrane proteins, the phase-variable opacity (Opa) proteins, by GC enhances their ability to adhere and to invade the epithelial cells in FTOC (9). Moreover, Opa expressing Escherichia coli can attach and invade human fallopian tube epithelium at low levels (10). These results indicate that Opa proteins alone may be sufficient to promote adherence to and invasion of the human cells. Studies with infected human volunteers have demonstrated a strong selection for Opa+ gonococci among the reisolates suggesting that Opa proteins play a role in gonococcal pathogenesis (11, 12).
Identification of the receptors involved in the interaction with host cells is a crucial step in understanding disease causation by bacteria. Some components on the mammalian cell surface have been shown to act as receptors for pathogenic microorganisms. One of the best defined receptor– ligand relationships is the interaction of Yersinia pseudotuberculosis invasin with β1 chain integrins on the mammalian cell surface (13). E-cadherin has recently been demonstrated as the receptor for internalin, a surface protein for entry of Listeria monocytogenes into epithelial cells (14). In GC, a variety of components participate in the interaction with the eukaryotic cells. These include pili (3), LPS (15), a 36-kD glycolipid-binding adhesin (16), interaction with carbohydrate structures on epithelial cells (17), and the Opa proteins (18). In gonococcal strain MS11, the Opa protein family consists of 11 unlinked opa genes whose sequences are known (19). One distinct Opa protein, the OpaA, has been correlated with adherence and subsequent internalization of GC by Chang conjunctival cells (20–22). The other members of Opa family are able to stimulate PMN adherence and phagocytosis (22, 23). It has been demonstrated that the interaction of the OpaA GC with epithelial cells involves binding to heparan sulfate syndecan receptors that are situated on the cell surface (24, 25). Moreover, recent studies indicated that gene family member (CGM)1 (26) and biliary glycoprotein (BGP) (27), members of the carcinoembryonic antigen (CEA) or CD66 family, serve as receptors for Opa+ bacteria for promoting adherence and internalization. In the instance of BGP or CD66a, it has been shown that the portion of the molecule that interacted with the Opa protein is the NH2-terminal domain that is homologous to the IgG variable domain (IgV-like; 28). This is supported by the ability of CGM1 to interact with Opa proteins since its surface-exposed portion consists solely of one IgV-like domain (26).
However, in addition to CD66a (BGP) and CD66d (CGM1), the CEA family also includes CD66b (CGM6), CD66c (NCA), and CD66e, the classical tumor associated CEA. We have investigated the activity of the different CD66 antigens with Opa proteins by expressing these five members of the CD66 family in HeLa cells and demonstrated that the transfectants differ in their ability to support adherence and internalization. Finally, OpaC protein from MS11 demonstrated a dual function; it interacted with both heparan sulfate syndecan receptors and members of the CEA family.
Materials And Methods
Bacterial Strains, Monoclonal Antibodies, and Cell Lines.
GC strain MS11 was cultured and maintained as previously described (29). Only pilus− GC and LOSb (lacto-N-neotetraose) phenotype were used (30). Recombinant opa genes from GC MS11 were constructed and expressed in E. coli HB101 as described previously (23). The designations of Opa proteins of both GC and E. coli are based on papers of Swanson et al. (11) and Belland et al. (23). E. coli HB101 containing the vector pGEM-3Z is designated as pGEM. E. coli HB101 expressing OpaA, OpaB, OpaC, OpaH, and OpaI genes are designated as pEXA, pEXB, pEXC, pEXH, and pEXI, respectively. Bacterial suspensions were from Luria-Bertani plates containing 50 μg/ml carbenicillin after growth for 16–20 h at 37°C. COL-1 mAb, specific for CGM1 and CEA only, was donated by Z. Shi (Zymed Laboratories Inc., San Francisco, CA) and mAb IB4 reactive with CD18 was provided by S. Wright (Merck Inc., Rahway, NJ).
The human colon adenocarcinoma cell line LS 174T was purchased from American Type Culture Collection (Rockville, MD). HeLa-CEA, HeLa-CGM1a, HeLa-BGPa, HeLa-NCA, and HeLa-CGM6 cells were constructed by transfecting HeLa cells with CEA, CGM1a, BGPa, NCA, and CGM6 cDNAs, and selected for surface antigen expression (31). HeLa-Neo cells are HeLa cells that were transfected with neomycin-resistance gene only (31).
Wild-type Chinese hamster ovary (CHO)-K1 and isogenic mutants 745 and 677 that have specific defects in proteoglycan biosynthesis were provided by J.D. Esko (University of Alabama, Birmingham, AL) (32–34). Mutant CHO cell 745 lacks xylosyltransferase and expresses no heparan sulfate and chondroitin sulfate, while mutant 677 which has defective N-acetylglucosaminyl and glucoronosyltransferases, produces no heparan sulfate, but 2–3 times increased chondroitin sulfate.
Adherence and Internalization Assays.
All cell lines were cultured in RPMI 1640 medium (GIBCO BRL, Gaithersburg, MD) with 10% FCS (Hyclone Labs., Logan, UT). For adherence assays, cells were grown to confluence (1–2 × 105 cells/well) in 24-well culture plates (Falcon, Lincoln Park, NJ), and washed twice with serum-free RPMI. E. coli were suspended in RPMI at OD540 of 0.04, and 0.5 ml of the bacterial suspensions was added to each well. The plates were incubated at 37°C with 5% CO2 for 4 h. The incubation period was limited to 2 h when mAb were included to avoid their inactivation. Experiments were terminated by washing three times with 1 ml of serum-free RPMI. Adherent bacteria were counted by suspending the cells in PBS containing 0.5% saponin (Calbiochem Corp., La Jolla, CA) and plating dilutions on Luria-Bertani agar medium containing 50 μg/ml of carbenicillin or on GC plates. The level of adherence of GC and E. coli to cells was calculated by determining the CFU associated with the host cell monolayers. Internalization assays were done in a similar fashion to adherence assays, but after the period of bacterial interaction to cell lines, the monolayers were washed twice and then incubated for 90 min with 1.5 ml of RPMI supplemented with 100 μg/ml gentamicin (GIBCO BRL). For adherence inhibition assays, the bacteria were added as a suspension in RPMI containing heparin at 30 μg/ml or 25 μg/ml mAb. The experiments were performed in duplicate or triplicate.
Binding of CEA to OpaI E. coli.
pGEM or pEXI were suspended in 1 ml of serum-free RPMI at OD540 of 0.8. 4 μg of purified CEA antigen (Calbiochem Corp.) was added to each bacterial suspension and incubated at room temperature for 60 min with gentle shaking. The bacteria were pelleted and washed once with RPMI and subjected to SDS-PAGE. The CEA bound to the bacteria was detected by Western blotting with COL-1 mAb after electrophoretic transfer to Immobilon-P membrane (Millipore Corp., Bedford, MA) (35). The bound COL-1 mAb was detected with HRP-conjugated protein A by chemiluminescence (Amersham Life Science, Arlington Heights, IL).
Adherence and Invasion of Opa+ Bacteria into HeLa-CD66 Transfectants.
The HeLa-Neo, HeLa-BGPa (CD66a), HeLaCGM6 (CD66b), HeLa-NCA (CD66c), HeLa-CGM1a (CD66d), and HeLa-CEA (CD66e) were examined for their ability to support adherence and invasion of OpaI bacteria. The internalization of Opa+ bacteria into HeLa-CD66 transfectants was measured by gentamicin killing assay and the results were confirmed by electron microscopy. pEXI adhered poorly to HeLa-Neo and CGM6, but attached very well to HeLa-CEA and HeLa-BGP (Fig. 1 A). The adherence of pEXI to HeLa-CGM1a and HeLa-NCA was intermediate.
The ability of these cell lines to promote invasion showed a different pattern. Although the adherence of pEXI to HeLa-BGPa was high, invasion was limited (Fig. 1,B). Electron microscopy demonstrated that HeLa-CEA cells were also able to engulf pEXI (Fig. 1,B) and OpaI GC (Fig. 2). However, compared to HeLa-CGM1a (Fig. 1,B, 2E), HeLaCEA cells demonstrated lower ability to internalize pEXI or OpaI GC (Figs. 1,B and 2 B). Virtually all HeLaCGM1a cells contained large numbers of GC, but a lower degree of cellular invasion by OpaI bacteria was observed with HeLa-CEA cells. HeLa-NCA and HeLa-CGM6 were hardly invaded at all.
OpaI E. coli Bind the CEA Antigen.
The HeLa-CEA cells supported the highest level of adherence with pEXI (Fig. 1,A), indicating that the CEA protein is a receptor for Opa protein. To demonstrate a direct binding of Opa+ bacteria with CEA, purified CEA was mixed with equal amounts of Opa negative pGEM or positive pEXI bacteria. The bacteria–CEA mixture was incubated for 1 h, pelleted, and solubilized. The amount of CEA antigen bound to the bacteria was detected with COL-1 antibody by Western blotting. As seen in Fig. 3, pEXI bound much more CEA than pGEM.
Multiple Opa Proteins Interact with HeLa-CEA Cells, and Are Inhibited by Anti-CEA Antibody.
Since strain MS11 can express several Opa proteins, we examined whether other Opa proteins possess the same function, as illustrated in Fig. 4,A. HeLa-CEA cells also bound OpaA, OpaB, OpaC, and OpaH E. coli with similar efficiency, but only OpaA-mediated adherence was strongly inhibited by soluble heparin. The adherence mediated by OpaI and OpaC was inhibited by anti-CEA antibody (COL-1), but not by IB4 directed to CD18 antigen (Fig. 4 B). These results demonstrate that CEA antigen serves as receptor for several Opa proteins.
OpaI Bacteria Bind HeLa-CEA and CEA-expressing Colon Cancer Cells in a Heparin-independent Manner.
To further substantiate that the binding of Opa+ GC with CEA is due to ligand–receptor interaction, we compared the well-characterized CEA-expressing colon cancer cell line LS 174T with HeLa-CEA in adhesion assays. As seen in Fig. 5 both HeLa-CEA and LS 174T bound the OpaI GC in a heparinindependent fashion, in sharp contrast to the OpaA-mediated interaction. The colon cancer cell line supported adherence about threefold less efficiently than HeLa-CEA, but the overall pattern was identical.
OpaC Bacteria Interact with Both Heparan Sulfate and Members of the CEA Family.
Previously published studies demonstrated that OpaA-mediated adherence to Chang conjunctival and CHO cells is dependent on binding to heparan sulfate syndecan receptors (24, 25). These studies also showed that OpaC had this activity, and that OpaC GC were able to bind tritiated heparin. However, the data shown in Fig. 4, A and B indicated that pEXC was able to bind the HeLaCEA, and that this binding was inhibited by treatment with CD66 antibody, suggesting that OpaC was able to mediate both adherence to syndecans and CD66 antigens. To provide further evidence that OpaC-mediated adherence to CHO epithelial cells is heparan sulfate dependent, we examined the adherence to heparan sulfate-deficient CHO mutants 745 and 677. Compared to wild-type CHO-K1, OpaC GC adhered to mutant 745 and 677 at a very low level (Fig. 6,A). Further proof for reactivity with CD66 antigens is provided in Fig. 6 B showing that pEXC, but not pEXA, is internalized by HeLa-CGM1a.
GC isolated either from the male urethra or from the cervix of infected females are most often Opa+ GC (36–38). Swanson et al. infected male volunteers intraurethrally with an Opa− strain of GC, and found that isolates recovered from the infection were predominantly Opa+ (11). Similar findings were also obtained with different challenge strains in further volunteer studies (12). Furthermore, high levels of Opa-specific antibody after gonococcal infection have been reported (39). Taken together, these data strongly suggest in vivo expression of Opa proteins plays an important role in gonococcal pathogenesis.
The present studies demonstrate that one of the mechanisms for Opa protein-mediated adherence was binding to CEA and CEA-related proteins. To provide this evidence, we chose to use a highly defined system, namely the use of E. coli expressing recombinant Opa proteins and a single cell line, HeLa transfected with various CD66 antigens. This strategy circumvents the unavoidable tendency of GC to alter their Opa phenotype by antigenic variation and allows certainty that the observed effects are strictly attributable to the expression of specific Opa proteins. To confirm that the results were reflective of gonococcal pathogenesis, we included experiments using Opa+ GC and also extended them to a well-characterized CEA-expressing colon cancer cell line, LS 174T. The CEA HeLa cell transfectants (HeLa-CEA) and the CEA-expressing colon cancer cell line bound and engulfed Opa+ bacteria and this interaction was inhibited by anti-CEA mAb. In addition, OpaI E. coli bound purified CEA antigen more effectively than Opa− E. coli.
CEA, a 180-kD tumor-associated cell-surface glycoprotein, was first described in 1965 (40). Later, it was found that antibodies to CEA reacted at low level with normal cells. Also, a number of closely related, cross-reacting antigens were identified that were initially termed nonspecific cross-reacting antigens (NCA; 41–43). The gene encoding CEA is located on human chromosome 19 (44), and is a member of a family of >20 expressible closely related genes (45) that belong to the Ig gene superfamily (46). The human CEA family consists of CEA, NCA, BGP (47), CGM1 including splicing variants CGM1a, CGM1b, and CGM1c (31, 48), CGM2, CGM6, CGM7, and pregnancyspecific glycoproteins (49–51). Although the in vivo function of CEA has not been determined, evidence shows that CEA functions as an intercellular adhesion molecule (52– 54). The binding of CEA to E. coli has been reported (55– 57), and CEA may also play a role as an accessory molecule in binding tumor cells to collagen type I (58).
We investigated the relative ability of the individual CD66 antigens to mediate adherence and internalization of Opa+ bacteria using HeLa cell transfectant cell lines. HeLa- CEA and HeLa-BGPa showed the highest level of adherence to OpaI, but BGP internalized the bacteria poorly. On the other hand, HeLa-CGM1a and HeLa-NCA showed intermediate levels of adherence, and only CGM1a promoted aggressive phagocytosis of OpaI E. coli (26) and GC (Fig. 2,E). HeLa-CGM6 (CD66b) could neither bind or engulf Opa+ bacteria (Fig. 1 A), although CGM6 is highly expressed on this HeLa cell line (59). Since CEA is not expressed in neutrophils (60), the most likely candidate for promoting phagocytosis of Opa+ bacteria is CGM1a. These results indicated that CD66 family showed variable ability in promoting adherence and internalization of Opa+ bacteria although they share homology in their NH2 termini (60).
The mechanism of CGM1a-promoted internalization may be that in CGM1a there is a cytoplasmic domain containing the YLYL motif, termed the immunoreceptor tyrosine activation motif, which is found in several receptor subunits that associate with cytoplasmic tyrosine kinases (61, 62). Activation of this motif leads to actin polymerization and phagocytosis (for review see reference 63). BGPa, which does not promote internalization of bacteria, does not contain immunoreceptor tyrosine activation motif in its cytoplasmic domain, but rather, two copies of immunoreceptor tyrosine inhibitory motif that are separated by 26 amino acids (47). This motif has been found in a number of immunoreceptors that mediate inhibitory effects (64, 65). However, CEA found on LS 174T cells is anchored to the membrane through a glycosyl-phosphatidylinositol anchor (66, 67). The GPI-anchored proteins mediate many functions in cell growth and cell–cell communication, and can initiate signal transduction (68–70). Thus, it is possible that the phagocytosis of Opa+ bacteria by means of CGM1a in neutrophils and CEA in epithelial cells uses the two distinct signaling mechanisms.
Among the Opa proteins tested, only OpaA showed no reactivity with CEA or CGM1a. OpaA has previously been shown to be reactive with heparan sulfate borne by syndecans and to bind radioactive heparin (24, 25). These studies showed that OpaC also had this specificity. Thus, the question of whether OpaC protein can interact with both heparan sulfate syndecan receptors and members of the CEA family arose. The data shown in Fig. 6,A demonstrates that OpaC GC bound to normal CHO cells, but failed to bind to CHO mutant cells defective in heparan sulfate synthesis, which was further proof for this specificity. On the other hand, OpaC promoted heparin-independent adherence to HeLa-CEA cells (Fig. 4,A), and induced the phagocytosis of OpaC E. coli into HeLa-CGM1a (Fig. 6 B). Thus, MS11 OpaC protein appears to target the two distinct receptors.
Gonorrhea is a genital disease that affects exclusively human beings, and adherence and invasion of epithelial cells is one of the important steps in this disease, as well as bacterial infections in general (71). However, the factors determining the efficient adherence and internalization of GC with or into specific cells are still poorly understood and are very dependent on the specific tissue culture cell line used (72). Identification of CEA and BGP as a receptor for Opa proteins is an important step in our understanding of GC pathogenesis. Normal adult human colon epithelium is known to produce small amounts of CEA-like antigens (73). GC, when isolated from rectal infections, express Opa proteins (38). It has also been shown that BGP is expressed on normal epithelia located in many organs including the uterus, cervix, prostate, and colon (74). Thus, it is likely that most of the tissues susceptible to GC infection express one or another member of the CD66 antigen family.
We thank Dr. Asesh Banerjee, Dr. Vijay Pancholi, and Mr. James Parker for useful suggestions and editorial comments on the manuscript.
This work was supported by Public Health Service grants AI 10615 and AI 26558.
1Abbreviations used in this paper: BGP, biliary glycoprotein; CEA, carcinoembryonic antigen; CGM, CEA gene family member; CHO, Chinese hamster ovary; FTOC, fallopian tube organ culture; GC, Neisseria gonorrhoeae; NCA, nonspecific cross-reacting antigens; Opa, opacity.
Address correspondence to Tie Chen, Laboratory of Bacterial Pathogenesis and Immunology, The Rockefeller University, 1230 York Ave., New York, NY 10021-6399.