Neisseria meningitidis Adhesin NadA Targets β1 Integrins

Meningococci are facultative-pathogenic bacteria endowed with a set of adhesins allowing colonization of the human upper respiratory tract, leading to fulminant meningitis and septicemia. The Neisseria adhesin NadA was identified in about 50% of N. meningitidis isolates and is closely related to the Yersinia adhesin YadA, the prototype of the oligomeric coiled-coil adhesin (Oca) family. NadA is known to be involved in cell adhesion, invasion, and induction of proinflammatory cytokines. Because of the enormous diversity of neisserial cell adhesins the analysis of the specific contribution of NadA in meningococcal host interactions is limited. Therefore, we used a non-invasive Y. enterocolitica mutant as carrier to study the role of NadA in host cell interaction. NadA was shown to be efficiently produced and localized in its oligomeric form on the bacterial surface of Y. enterocolitica. Additionally, NadA mediated a β1 integrin-dependent adherence with subsequent internalization of yersiniae by a β1 integrin-positive cell line. Using recombinant NadA24–210 protein and human and murine β1 integrin-expressing cell lines we could demonstrate the role of the β1 integrin subunit as putative receptor for NadA. Subsequent inhibition assays revealed specific interaction of NadA24–210 with the human β1 integrin subunit. Cumulatively, these results indicate that Y. enterocolitica is a suitable toolbox system for analysis of the adhesive properties of NadA, revealing strong evidence that β1 integrins are important receptors for NadA. Thus, this study demonstrated for the first time a direct interaction between the Oca-family member NadA and human β1 integrins.

Neisseria meningitidis is a well-known Gram-negative diplococcus, which is able to colonize the nasopharynx of humans with relatively high frequency. Under certain conditions this pathogen translocates across the mucosal layer of the respiratory tract and causes invasive meningococcal disease (IMD) 2 comprising septicemic and/or fulminant meningitis. N. meningitidis is endowed with a broad repertoire of adhesions, which are believed to support colonization and eventually invasion of mucosal epithelial cells. The most extensively investigated adhesins are the type IV pili (Tfp) and the non-pilus adhesins: (i) opacity proteins Opa and Opc and (ii) the autotransporter proteins App (adhesion penetration protein), NhhA (Neisseria hia homolog) and NadA (Neisseria adhesin A) (1). The two last-mentioned adhesins are typical members of the oligomeric coiled-coil adhesin (Oca) family, also known as trimeric autotransporters or as type Vc secretion system, of which the prototype is the trimeric coiled-coil adhesin YadA of enteropathogenic yersiniae (2)(3)(4)(5). NadA is produced only by 50% of meningococcal isolates, in particular the nadA gene is obviously present in about 84% of isolates in hypervirulent lineages such as electrophoretic types ET-5, ET-15, and ET-37 (6,7). Interestingly, the nadA gene of ET-15 meningococci is frequently (68%) disrupted by an IS1301 insertion (8). The C-terminal NadA amino acid sequence is closely related to that of the Yersinia adhesin YadA, which has been shown to present a tripartite structured organization: the N-terminal globular head domain, the intermediate ␣-helical region capable of forming a homotrimeric coiled-coil stalk also called passenger domain and a highly conserved C-terminal anchor domain (four ␤-strands inserted into the outer membrane), which is responsible for translocation of the head/stalk region and trimerization of the adhesin (4).
Whereas YadA of enteropathogenic Yersinia species mediates binding to diverse ECM proteins (9 -12), epithelial cells, and neutrophils, NadA of N. meningitidis does not bind to ECM proteins but binds to a restricted number of cell types such as Chang cells, HEp-2 or human monocytes/macrophages but fails to bind to HUVEC endothelial cells or human endometrium cell line Hec-1B (13)(14)(15).
The large diversity of cell adhesins and the capability of the polysaccharide capsule of N. meningitidis to mask the function of non-pilus adhesins hampers the analysis of the role of individual adhesins for host cell interaction including identification of receptors and prevention of complement lysis. Unraveling the host cell receptor for NadA would be pivotal for a better understanding of the role of NadA in meningococcal pathogenesis, particularly also with respect to the lack of a conventional mouse infection model. This prompted us to develop a novel approach for studying the interaction of neisserial adhesins with host cells and their role in colonization and/or pathogen-* This work was supported by the Friedrich-Baur-Stiftung, the Munich Centre for Integrated Protein Science (CIPSM) and by the German Research Foundation (DFG) Grant SFB594. This work was also supported in part by the DFG Grant SFB479 (to O. K.). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1 and S2. 1  with anti-host effector functions, and (ii) the chromosomally encoded invasin (Inv) which is recognized by ␣5␤1 integrins, and (iii) the yersiniabactin system for ferric iron uptake, which is encoded by the High Pathogenicity Island (HPI) and is required for mouse virulence (16,17). By specific deletion of known virulence determinants Y. enterocolitica can be used as toolbox for studying pathogenicity factors in human serum, cell culture models, and experimentally infected mice. In this study we introduced the nadA gene into Y. enterocolitica to dissect putative virulence functions of NadA in regard of NadA host cell receptor interaction. For the first time we provide strong evidence that the Oca member NadA is directly recognized by ␤1 integrins and triggers an internalization signal.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-The bacterial strains used in this study are listed in Table 1. Y. enterocolitica strains were grown in Luria-Bertani (LB) or brain heart infusion (BHI) medium at 27°C. For induction of yadA expression, overnight cultures grown at 27°C were diluted 1:40 in RPMI 1640 cell culture medium (Invitrogen, Karlsruhe, Germany) and grown at 37°C for 5 h (4). Escherichia coli strains were cultivated at 37°C in LB medium. Neisseria species were plated on GC-agar supplemented with Vitox (Oxoid, Hampshire, UK) and grown at 37°C in 5% CO 2 .
Strain Construction-To express full-length nadA in Y. enterocolitica strain WA-314 ⌬yadA, the nadA gene (allele 1) encoding the mature NadA protein (lacking the leader peptide sequence amino acids 1-23) was amplified by polymerase chain reaction (PCR) from N. meningitidis strain MC58 (gene ID: 904134) using the oligonucleotide primers MC58 -1f (TAC TAG AGC TCG CCA CAA GCG ACG ACG ATG, SacI site) and N-1089r (TAC TAG AGC TCT TAC CAC TCG TAA TTG ACG C, SacI site) (bp position 1 refers to bp position 7 ϭ second ATG in the nadA gene ID: 904134) ( Table 2). The resulting DNA fragment was digested with SacI and cloned into pGPS-SS, resulting in pGPS-nadA. This plasmid was transformed into E. coli SM10 and subsequently transferred by conjugation into WA-314 ⌬yadA. The transconjugants were selected for integration of pGPS-nadA into pYV-A-0 resulting in WA-314 ⌬yadA:nadA construct. The pYVO8-nadA plasmid from WA-314 ⌬yadA:nadA and the pYVO8-SS plasmid from WA-314 ⌬yadA:SS were additionally transformed into WA-c ⌬inv resulting in WA-c ⌬inv(pYVO8-nadA) and WA-c ⌬inv(pYVO8-SS). For generation of nadA-or yadA-expressing Y. enterocolitica strains lacking the pYV plasmid and chromosomally encoded invasin, the pYV plasmid-cured and invasinnegative WA-c ⌬inv strain was used. To clone nadA and yadA genes carrying the yadA promoter and terminator region and the sequence encoding the YadA LP, nadA, and yadA were amplified by PCR from WA-314 ⌬yadA:nadA or WA-314 ⌬yadA:yadA, respectively, using the oligonucleotide primers A-144f (TTA ATC TAG ATA GTG CTG TTT TTT GCA TG, XbaI) and A-119r (AAT TGG ATC CAA CTG AAA CCA TGA TAA AAA GC, BamHI). After digestion DNA fragments were cloned into the plasmid pACYC184:virF (p) and transferred into WA-c ⌬inv resulting in WA-c ⌬inv(pnadA) and WA-c ⌬inv(pyadA). pACYC184:virF was generated by amplifying the Yersinia transcriptional activator gene virF from strain WA-314 with the oligonucleotide primers virF-151f (AAT AGC ATG CTT GCC AGT CAC CTA ATAC C, SphI) and virF-86r (AAT AGT CGA CTT GCT CAT CCC ATT GAA TC, SalI) digested with SphI and SalI and cloned into pACYC184 plasmid. Generation of Recombinant NadA 24 -210 Protein for Production of Rabbit Anti-NadA Serum-For production of recombinant NadA protein, the nadA gene (bp 69 -630) from N. meningitidis serogroup B strain MC58 was amplified by PCR using the oligonucleotide primers MC58 -69f (TAA TTA TCA TAT GGC CAC AAG CGA CGA CGA TG, NdeI) and MC58 -630r (ATT ATC TCG AGG GCC GTC TGT TTG GCT TC, XhoI). The DNA fragment was cloned into pET21 vector b ϩ (Merck, Darmstadt, Germany). After transferring the plasmid into E. coli BL21 (DE3), protein expression was induced at 37°C by addition of 1 mM IPTG at A 600 ϳ0.6 and subsequent incubation for additional 4 -5 h. The recombinant NadA 24 -210 protein was purified by affinity chromatography on Ni 2ϩ -conjugated chelating fast flow Sepharose 4B resin (GE Healthcare, Munich, Germany). 200 -500 g of purified recombinant NadA 24 -210 protein were additionally used to immunize rabbits for 91 days according to standard immunization protocol (Pineda Antikörper-Service, Berlin, Germany).
Infection of Cell Monolayers with Y. enterocolitica-For adhesion and invasion assays, 1 ϫ 10 5 Chang cells per well were seeded in 24-well tissue culture plates overnight and subse-quently infected with WA-c ⌬inv(pnadA), WA-c ⌬inv(pyadA), or WA-c ⌬inv(p) with a multiplicity of infection (moi) of 100 in DMEM, and incubated for 3 h at 37°C in 5% CO 2 . Non-adherent bacteria were removed by washing cells three times with PBS, and cells were lysed with 1% Triton X-100 in PBS. Serial dilutions of lysed cell supernatants were plated onto LB agar containing chloramphenicol (20 g/ml) for selection of yersiniae. Quantification of intracellular bacteria was performed by the gentamicin protection assay. For this Chang cells were infected as described above, incubated for 3 h at 37°C, nonadherent bacteria were removed, and cells were additionally incubated for 90 min in presence of 50 g/ml gentamicin at 37°C in 5% CO 2 . After washing the cell monolayer, intracellular bacteria were released by cell lysis with 1% Triton X-100 in PBS, and the lysates were plated on agar plates. Adherence assays with GE-11-␤1 and GE-11 cells were also performed as described for Chang cells, with the exception that cells were infected with a moi of 50 for 1 h at 37°C in 5% CO 2 . Statistical significance of at least three independent experiments performed in triplicates was determined by Student's t test.
Infection of Cell Monolayers with N. meningitidis-1 ϫ 10 5 GE-11-␤1 and GE-11 cells per well were seeded in 24-well tissue culture plates and grown to confluency overnight. N. meningitidis strains were grown on GC plates overnight at 37°C and 5% CO 2 . Neisseriae were scraped from plates, washed twice with PBS and resuspended in DMEM supplemented with 1% FCS. Afterward, a moi of 100 was adjusted in DMEM supplemented with 1% FCS and adherence was performed for 3 h at 37°C in 5% CO 2 . The number of cell-associated bacteria was determined after washing the cell monolayer three times followed by cell lysis of cells with 1% saponin. Serial dilutions of supernatants were plated on GC agar. The number of intracellular bacteria was quantified by gentamicin protection assay (100 g/ml gentamicin for 1 h at 37°C in 5% CO 2 ). Statistical significance of at least three independent experiments performed in triplicate was determined by Student's t test.

RESULTS
Expression of nadA in Y. enterocolitica-To study the functional role of NadA in Y. enterocolitica under YadA promoter conditions we fused the upstream and proximal portion of yadA comprising its promoter region and leader peptide (LP) encoding region with the nadA gene sequence encoding the mature NadA protein (Fig. 1). The nadA gene was cloned into the suicide plasmid pGPS-SS and integrated into the pYV-A-0 plasmid of Y. enterocolitica strain WA-314 ⌬yadA, via homologous recombination. Additionally, full-length nadA encoding  (4)). Numbers refer to amino acid residues of the protein.
the processed NadA together with the yadA promoter, LP and terminator regions was ligated into the plasmid pACY184:virF (p). The resulting plasmid pnadA was transferred into the pYVcured invasin-negative Y. enterocolitica strain WA-c ⌬inv resulting in WA-c ⌬inv(pnadA). Expression of nadA in both strains was confirmed by Western blot analysis. Outer membrane fractions of WA-c ⌬inv(pnadA) incubated at 100°C showed a high molecular mass protein at ϳ120 kDa corresponding to the oligomeric form of NadA (Fig. 2A, lane 1) and a low molecular mass band at ϳ35 kDa corresponding to the monomeric form of NadA (Fig. 2A, lane 1), which were absent in WA-c ⌬inv(p) (Fig. 2A, lane 2). Localization of NadA in the outer membrane could also be demonstrated with strain WA-314 ⌬yadA:nadA (Fig. 2B, lane 1), lacking in the WA-314 ⌬yadA control strain (Fig. 2B, lane 2). Oligomeric NadA produced by Y. enterocolitica had the same electrophoretic mobility in SDS-PAGE as NadA produced by unencapsulated N. meningitidis MC58 ⌬siaD strain (data not shown), indicating that NadA is likely not further post-translationally modified by N. meningitidis and Y. enterocolitica. Localization of NadA in the outer membrane could additionally be confirmed by immunofluorescence microscopy of unfixed yersiniae revealing NadA exposition on the surface of strain WA-314 ⌬yadA:nadA and WA-c ⌬inv(pnadA) (Fig. 3). These results demonstrate that full-length NadA produced by Y. enterocolitica forms heatstable oligomers and is efficiently transported across the outer membrane and exposed probably in its trimeric form on the surface of Y. enterocolitica similar to YadA.
nadA-expressing Yersiniae Do Not Interact with Extracellular Matrix (ECM) Proteins-Previously it has been demonstrated that NadA produced by N. meningitidis or E. coli does not bind to ECM proteins (13). Therefore we investigated the ability of full-length NadA produced on the surface of Y. enterocolitica to mediate interaction with immobilized collagen type I, fibronectin, and matrigel, respectively, using an ELISA technique (22). As expected NadA-positive yersiniae failed to bind any of the tested ECM proteins in contrast to YadA-positive yersiniae which are known to bind to different ECM proteins (11, 23) (supplemental Fig. S1).
nadA-expressing Yersiniae Mediate Adhesion to and Invasion into Chang Cells-Cell-association and internalization was analyzed for nadA-expressing inv-negative yersiniae and human Chang cells. Thus, Chang cell monolayers were infected with different derivatives of WA-c ⌬inv yersiniae to investigate the role of NadA in Y. enterocolitica cell monolayer interaction. Quantification of cell-associated bacteria revealed a ϳ300-fold higher adhesion capacity for YadA-positive yersiniae as well as an ϳ11-fold higher adhesion capacity for NadA-positive yersiniae to Chang cells, compared with the control strain WA-c ⌬inv(p) (yersiniae background control) (Fig. 4A). Using the gentamicin protection assay we also determined the number of internalized yersiniae. As shown in Fig. 4B YadA-positive (ϳ1000-fold) and NadA-positive yersiniae (ϳ15-fold) showed also significantly increased uptake into Chang cells compared with WA-c ⌬inv(p). These results demonstrate that both YadA and NadA mediate adherence and induce internalization of yersiniae into Chang cells with NadA being probably a weaker adhesin than YadA.
nadA-expressing Yersiniae Do Not Bind Soluble CEACAM-GFP Constructs-N. meningitidis and N. gonorrhoeae express members of the Opa protein family which facilitate interaction with several host cell types (24,25). Opa HS proteins mediate attachment and invasion into several epithelial cell lines via heparin sulfate proteoglycans, whereas Opa CEA proteins interact with host cell receptors of the CEACAM family (26). Therefore, we tested the ability of nadA-expressing yersiniae to bind to CEACAM 1, 3, 5, 6, and 8. Hence, yersiniae strains WA-c ⌬inv(pnadA), WA-c ⌬inv(pyadA), WA-c ⌬inv(p) (negative control), a non-opaque N. gonorrhoeae strain (Ngo Opa-) and an isogenic, Opa CEA -expressing N. gonorrhoeae strain (Ngo Opa CEA ; CEACAM-binding positive control) were incubated with recombinant GFP-tagged CEACAM1, CEACAM3, CEA, CEACAM6, or CEACAM8 extracellular, N-terminal domains. Binding of the fluorescent receptor domains to the microorganisms was analyzed by flow cytometry according to the protocol of Kuespert et al. (27). We detected no interaction of recom-  binant CEACAMs with nadA-or yadA-expressing yersiniae, nor with the non-opaque N. gonorrhoeae strain. In contrast, the Opa CEA -expressing N. gonorrhoeae strain showed marked interaction with CEACAM1, CEACAM3, CEA, and CEACAM6 (supplemental Fig. S2).

tation. As shown in
NadA 24 -210 Protein Directly Binds to the Human ␤1 Integrin Subunit-To further substantiate receptor-ligand interactions of NadA 24 -210 and ␤1 integrins we performed Far Western blotting. Thus, purified human ␣5␤1 integrin (prey protein) was separated by SDS-PAGE, transferred onto membranes, renatured, and incubated with NadA 24 -210 (bait protein). Interaction of NadA 24 -210 with ␣5 or ␤1 integrin was detected with rabbit anti-NadA serum and showed one single high molecular mass band at 120 kDa which represents the ␤1 integrin subunit (Fig. 7, lane 2). As Yersinia Invasin is known to bind directly to ␣5␤1 integrins (28), the Inv397 (O:8) protein was used as positive control and showed the same high molecular mass band at 120 kDa after detection with anti-Invasin serum (Fig. 7, lane 4). Comparison with a Far Western blot performed without NadA or Invasin bait protein, detected with an anti-␤1 integrin-specific antibody, revealed also one single band at 120 kDa representing the ␤1 integrin subunit (Fig. 7, lane 1). We therefore conclude that recombinant NadA 24 -210 protein directly interacts with the ␤1 integrin subunit of the ␣5␤1 integrin heterodimer.
nadA-expressing Yersiniae Directly Bind to Human ␣5␤1 Integrin-To further analyze whether the observed direct interaction between NadA and human ␤1 integrins can also be detected with oligomeric NadA expressed on the cell surface of yersiniae, we used Alexa488-labeled recombinant human ␣5␤1 integrin for cytometric binding studies. As shown in Fig. 8, the NadA-positive strain WA-c ⌬inv(pYV-nadA) demonstrated a significantly higher binding (ϳ2.15-fold) of Alexa488-labeled human ␣5␤1 integrin compared with the NadA-negative strain WA-c ⌬inv(pYV-SS), confirming the direct interaction between NadA localized on the bacterial surface of Y. enterocolitica and human ␣5␤1 integrins.
NadA-specific Cell Adhesion and Invasion Is Not Detectable in N. meningitidis-NadA has previously been described as an invasin, mediating invasion of nadA-expressing E. coli and meningococci into human Chang cells (13). To test whether ␤1 integrins are involved in mediating adhesion and invasion of N. meningitidis we used the unencapsulated N. meningitidis strain MC58 ⌬siaD and the isogenic nadA mutant MC58 ⌬siaD ⌬nadA for cellular infection of GE-11-␤1 and GE-11 cell monolayers for 3 h (moi 100). As shown in Fig. 9A, we found that the   nadA-positive strain MC58 ⌬siaD showed significantly higher numbers of cell-associated bacteria compared with the isogenic nadA mutant MC58 ⌬siaD ⌬nadA. Nevertheless, no significant difference for ␤1 integrin-positive GE-11-␤1 and ␤1 integrinnegative GE-11 cells could be observed for the strain MC58 ⌬siaD and MC58 ⌬siaD ⌬nadA, respectively. This demonstrates that nadA-expressing unencapsulated meningococci have higher binding capacity to GE-11 and GE-11-␤1 cells, but this effect seems not to be solely dependent on the presence of ␤1 integrins. We also investigated the role of NadA in N. meningitidis in cell invasion of GE-11-␤1 and GE-11 cells by using MC58 ⌬siaD and MC58 ⌬siaD ⌬nadA in gentamicin protection assays. Interestingly, both strains showed higher numbers of intracellular bacteria for GE-11 cells, compared with GE-11-␤1 cells (Fig. 9B). This indicates that probably not only ␤1 integrins are exploited by Neisseria but also other receptors, which might be present on GE-11 cells in higher amounts than on GE-11-␤1 cells. However, comparing MC58 ⌬siaD and MC58 ⌬siaD ⌬nadA, no significant contribution of NadA for cell entry either into GE-11 or into GE-11-␤1 cells could be detected. nadA-expressing Yersiniae Mediate Adhesion and Invasion into Human and Mouse ␤1 Integrin-expressing Cells-As N. meningitidis is endowed with diverse surface exposed bacterial adhesins probably masking a NadA-␤1 integrin-dependent colonization of human cells under applied in vitro conditions, we used the Yersinia model to test whether ␤1 integrins are involved in mediating adherence and entry of nadA-expressing  yersiniae. Therefore, we quantified the number of cell-associated and intracellular bacteria after infection of GE-11-␤1 and GE-11 cells. GE-11-␤1 and GE-11 cells were incubated with WA-c ⌬inv(pnadA), WA-c ⌬inv(p) or the invasin-positive WA-c strain (Inv/␤1 integrin-mediated invasion) (29, 30) for 1 h and an moi of 50 and analyzed for cell-association and intracellular bacteria. We found that the nadA-expressing strain WA-c ⌬inv(pnadA) showed a significantly higher number (ϳ2.6-fold) of cell-associated bacteria for ␤1 integrin-positive GE-11-␤1 cells than for ␤1 integrin-negative GE-11 cells. This result is in accordance with the result obtained for the invasinpositive WA-c strain showing also increased cell-association (ϳ3.5-fold) for GE-11-␤1 cells compared with GE-11 cells. The control strain WA-c ⌬inv(p) showed weak interaction with GE-11-␤1 and GE-11 cells compared with WA-c ⌬inv(pnadA) and WA-c (Fig. 10A). Concerning ␤1 integrin-mediated internalization of nadA-expressing yersiniae into GE-11-␤1 and GE-11 cells, gentamicin protection assays revealed that the number of intracellular WA-c ⌬inv(pnadA) bacteria was significantly higher (ϳ18-fold) for ␤1 integrin-positive cells than for ␤1-negative cells. This result resembles that of the invasin-positive strain WA-c, showing higher numbers of intracellular bacteria (ϳ28-fold) in presence of human ␤1 integrins in comparison to ␤1 integrin-negative cells (Fig. 10B). These results confirm that ␤1 integrins appear to function as receptors for NadA, supporting ␤1 integrin-dependent adhesion and internalization and emphasize the advantage of yersiniae expressing NadA as the only effective cell adhesin to demonstrate function and specificity of a neisserial adhesin.

DISCUSSION
Y. enterocolitica is a suitable bacterial pathogen to investigate fundamental aspects of virulence including bacterial adhesion, invasion, subversion of the innate immune defense, mechanisms of extracellular survival and multiplication in the mouse infection model (31)(32)(33). The major pathogenicity determinants (virulence plasmid pYV, invasin gene inv, HPI), their gene products and pathogenicity functions have been well characterized. Therefore, it is conceivable that pathogenicity factors of non-mouse virulent pathogens similar to Yersinia ones, can be studied in Yersinia by genetic replacement. In this study this approach has been applied to the neisserial adhesin NadA by replacement of the yadA gene by nadA. By fusing the coding sequence of mature NadA with the promoter region and the coding region of the N-terminal signal sequence of yadA we could demonstrate NadA production, secretion, insertion into the outer membrane and surface exposition of NadA by Y. enterocolitica. This is remarkable as the genus Neisseria belongs to the ␤ subdivision of Proteobacteria in contrast to Yersinia belonging to the ␥ subdivision and suggests a certain degree of functional autonomy of Oca family members. Previously it has been demonstrated that NadA forms heat-stable oligomers (6). This characteristic of Oca family members could also be demonstrated with yersiniae expressing nadA. A typical function of Yersinia YadA is binding to extracellular matrix (ECM) proteins. We could demonstrate that NadA produced by yersiniae does not contribute to binding to ECM proteins (matrigel, fibronectin, and collagen type I). However, we could demonstrate that NadA as well as YadA mediate adhesion to Chang cells and triggering of internalization by using a pYVand inv deleted Y. enterocolitica mutant. These latter results are in agreement with experiments using E. coli expressing nadA, as previously shown (13). Moraxella catarrhalis Oca family member UspA1 and Neisseria Opa proteins are known to be recognized by CEACAMs, which are expressed by diverse host cells (34). By using soluble recombinant GFP-tagged CEACAMs and Y. enterocolitica expressing yadA or nadA and N. gonorrhoeae as controls we could demonstrate by flow cytometry that neither NadA nor YadA are recognized by CEACAM 1, 3, 5, 6, or 8, respectively. Previously it has been demonstrated that NadA induces chemokine IL-8 production of diverse host cell types similar to Yersinia Inv which is recognized by ␤1 integrins (14,(35)(36)(37). This prompted us to check whether NadA could also interact with ␤1 integrins. Interaction of NadA and ␤1 integrins could be substantiated by using recombinant NadA protein covering the NadA binding domain (NadA 24 -210 ) in binding studies with epithelial-like GE-11 (human ␤1 integrin-negative and GE-11-␤1 (human ␤1 integrin-positive) or fibroblast-like 2-4 (mouse ␤1 integrin-negative) and 2-4-8 (mouse ␤1 integrin-positive) cells. We could clearly demonstrate by flow cytometry that NadA 24 -210 binds to human and mouse ␤1 integrins.
The specificity of NadA 24 -210 binding to human ␤1 integrins could further be corroborated by blocking experiments with anti-human ␤1 monoclonal antibodies. Additional blocking experiments with different anti-human ␣ monoclonal antibodies further confirmed interaction of NadA 24 -210 with the ␤1 integrin subunit, whereas the ␣ subunit seems not to be involved. Interaction of NadA and ␤1 integrins was further analyzed by Far Western blotting using recombinant NadA (NadA 24 -210 ) as "prey" and ␤1 integrin as "bait", revealing direct interaction of NadA with the ␤1 integrin subunit. Direct interaction of NadA and ␤1 integrins could additionally be verified for nadA-expressing, invasin-negative yersiniae, and Alexa488-labeled human ␣5␤1 integrin, revealing that native NadA localized on the bacterial surface is involved in binding to human ␤1 integrins. Additionally, we compared the interaction of pYV-negative Y. enterocolitica expressing nadA or inv in ␤1 integrin-specific invasion, respectively, with epithelial-like cells derived from ␤1 integrin-knock-out mouse embryonal (GE-11) cells and GE-11 cells transfected with human ␤1 integrins (GE-11-␤1 cells). Inv and NadA both significantly contributed to ␤1 integrin-mediated adhesion and internalization. These results were compared with non-encapsulated N. meningitidis MC58 ⌬siaD and a double mutant MC58 ⌬siaD ⌬nadA. Surprisingly, when the non-encapsulated isogenic pair MC58 ⌬siaD/MC58 ⌬siaD ⌬nadA was compared for cell invasion, we found a higher rate of neisserial invasion for ␤1 integrin-negative GE-11 cells which was independent of the presence of NadA, whereas for adhesion there was a weak significant effect in favor of NadA-␤1 interaction with GE-11-␤1 cells. This result shows that probably because of the presence of multiple adhesins of N. meningitidis the identification of neisserial adhesin-specific host receptors is severely restricted unless a heterologous welldefined bacterial host/carrier is used for expression of the respective adhesins. In conclusion, the NadA binding analysis using Y. enterocolitica (⌬inv mutant) as heterologous expression system for nadA and the recombinant NadA binding module structure in conjunction with ␤1 integrin Far Western blotting and defined isogenic pairs of ␤1 integrin-positive and ␤1 integrin-negative cell lines revealed for the first time that the NadA head domain interacts specifically with ␤1 integrins. The ␤1 integrin subunit might thus function as host cell receptor for N. meningitidis expressing nadA gene. Therefore NadA is the first adhesin of the Oca family which directly interacts with the ␤1 integrin subunit.
Bacterial adhesin-␤1 integrin interactions have been described for several pathogens colonizing and/or invading the mucosal epithelium of the gastrointestinal or the respiratory tract including Yersinia species and E. coli. Surface ␤1 integrin expressing cells such as M cells of the Peyer's patches (PP) and nasal-associated lymphoid tissue (NALT) are recognized by bacterial adhesins resulting in bacterial translocation across the mucosal layer and triggering the release of chemokines such as IL-8/CXCL8 (38 -41). Moreover, antigen-presenting dendritic cells (DCs), macrophages and neutrophils, which have been recruited to bacterial entry sides, also express ␣4␤1 and/or ␣5␤1 integrins. In analogy to invasin-expressing Yersinia it is not unlikely that N. meningitidis interacts through NadA with ␣␤1 integrins of M-cells of the NALT and with DCs, macrophages and neutrophils of the submucosa (14,24,36).