Properly folded and functional PorB from Neisseria gonorrhoeae inhibits dendritic cell stimulation of CD4+ T cell proliferation

Neisseria gonorrhoeae is an exclusive human pathogen that evades the host immune system through multiple mechanisms. We have shown that N. gonorrhoeae suppresses the capacity of antigen-presenting cells to induce CD4+ T cell proliferation. In this study, we sought to determine the gonococcal factors involved in this adaptive immune suppression. We show that suppression of the capacity of antigen-pulsed dendritic cells to induce T cell proliferation is recapitulated by administration of a high-molecular-weight fraction of conditioned medium from N. gonorrhoeae cultures, which includes outer membrane vesicles that are shed during growth of the bacteria. N. gonorrhoeae PorB is the most abundant protein in N. gonorrhoeae–derived vesicles, and treatment of dendritic cells with purified recombinant PorB inhibited the capacity of the cells to stimulate T cell proliferation. This immunosuppressive feature of purified PorB depended on proper folding of the protein. PorB from N. gonorrhoeae, as well as other Neisseria species and other Gram-negative bacterial species, are known to activate host Toll-like receptor 2 (TLR2) signaling. Published studies have demonstrated that purified Neisseria PorB forms proteinacious nanoparticles, termed proteosomes, when detergent micelles are removed. Unlike folded, detergent-solubilized PorB, PorB proteosomes stimulate immune responses. We now demonstrate that the formation of PorB proteosomes from structurally intact PorB eliminates the immunosuppressive property of the protein while enhancing TLR2 stimulation. These findings suggest that gonococcal PorB present in shed outer membrane vesicles plays a role in suppression of adaptive immune responses to this immune-evasive pathogen.

CD4 ϩ T cells (9,10), resulting in down-regulation of T cell proliferation in response to antigens (11). Studies of gonococcal infections in the female mouse infection model have shown that N. gonorrhoeae elicits T helper type 17 (Th17) responses through the induction of host transforming growth factor ␤. Further, the Th17 response drives induction of localized inflammation, including recruitment of host neutrophils (to which N. gonorrhoeae is relatively resistant) (12). Recent studies in a mouse model of gonorrhea have shown that N. gonorrhoeae induces production of interleukin 10 (IL-10) and regulatory T cells (Tr1) that suppress Th1-and Th2-dependent adaptive immune responses (13).
Our previous studies have shown that N. gonorrhoeae suppresses the capacity of dendritic cells, professional antigen-presenting cells that play a key role in promoting pathogen-specific adaptive immune responses, to stimulate antigen-specific T cell proliferation (14). N. gonorrhoeae causes this suppression in part by promoting secretion of inhibitory factors, such as IL-10, and expression of cell-autonomous factors, including up-regulation of programmed death ligand 1, but the key molecular component(s) of N. gonorrhoeae that engage in this process were not identified. In this study, we determined that gonococcus-conditioned medium carries factors that recapitulate the suppression of dendritic cell-induced T cell proliferation observed with whole bacteria. We show that properly folded, recombinant PorB, a major protein component found in conditioned medium, has similar suppressive properties. Surprisingly, prior studies of PorB from N. gonorrhoeae and other Neisseria species have demonstrated that the protein can act as an immune-stimulating adjunct. We further demonstrate that the stimulatory properties of PorB result from a loss of immunesuppressive activity as a consequence of the loss of the properly folded PorB protein structure that occurs when detergent is removed from this integral membrane protein. Taken together, our results suggest that, although the native PorB trimer from N. gonorrhoeae can stimulate signaling in some immune cells through activation of Toll-like receptor 2, PorB overcomes this stimulation by profoundly inhibiting dendritic cell-promoted T cell proliferation when presented to cells in its native, properly folded state.

N. gonorrhoeae-conditioned medium inhibits dendritic cell-induced, antigen-specific T cell proliferation
To determine whether N. gonorrhoeae releases factors responsible for inhibition of antigen-pulsed dendritic cellinduced T cell proliferation, conditioned medium from N. gonorrhoeae cultures (Ng-CM) or live N. gonorrhoeae was added to dendritic cells during ovalbumin (OVA) exposure for 24 h, and then the dendritic cells were washed and co-cultured with carboxyfluorescein succinimidyl ester (CFSE)-labeled T cells from OT-II mice. After 7 days, T cell proliferation in each coculture was evaluated by quantifying the dilution of CFSE fluorescence. Treatment of dendritic cells with either N. gonorrhoeae or Ng-CM resulted in a dose-dependent reduction of proliferation of co-cultured T cells compared with OVAtreated dendritic cells (Fig. 1, A-C). Ng-CM also inhibited the capacity of human monocyte-derived dendritic cells to induce proliferation of allogeneic T lymphocytes (Fig. 1, D and E), demonstrating that Ng-CM-mediated suppression occurs in both human and mouse dendritic cells.
To further characterize the components released from N. gonorrhoeae that are responsible for inhibition of T cell proliferation, Ng-CM was subjected to ultrafiltration through a 100-kDa cutoff filter, and the retentate was restored to its original volume by addition of Graver-Wade medium (GWM) (15). The filtrate and retentate fractions were then assessed for their capacity to inhibit dendritic cell-induced T cell proliferation. The 100-kDa retentate from Ng-CM demonstrated inhibition of T cell proliferation that was indistinguishable from unfiltered Ng-CM, whereas the 100-kDa filtrate from Ng-CM exhibited no detectable inhibition of T cell proliferation (Fig. 2, A and  B). Although N. gonorrhoeae does not secrete known protein exotoxins, N. gonorrhoeae and other pathogenic Neisseria species prolifically shed outer membrane vesicles (OMVs, also called membrane blebs) into the environment during growth (16,17). Ng-CM, Ng-CM 100-kDa filtrate, and Ng-CM 100-kDa retentate were analyzed by SDS-PAGE and silver staining, which revealed that the majority of proteins present in Ng-CM were retained by the 100-kDa filter (Fig. S1A). The finding that the majority of the proteins found in Ng-CM and the inhibitory factor are retained by a 100-kDa filter suggests that the immunosuppressive effect of Ng-CM is due to OMVs or proteins in the OMVs.

N. gonorrhoeae PorB 1B , but not LOS, inhibits OVA-pulsed dendritic cell-induced T cell proliferation
OMVs from N. gonorrhoeae contain large quantities of LOS as well as outer membrane proteins, primarily PorB (17,18). We therefore sought to determine whether LOS or gonococcal proteins abundant in OMVs present in Ng-CM were responsible for inducing inhibition of dendritic cell-directed, OVAinduced T cell proliferation. The amount of LOS in Ng-CM was estimated following SDS-PAGE and silver staining by comparing the staining intensities to a titration series of purified LOS (Fig. S1A). The quantity of Ng-CM that conferred maximal inhibition of dendritic cell-mediated T cell proliferation (25 l, Fig. 1C) was estimated to contain ϳ100 ng of LOS. When dendritic cells were treated with 100 ng of purified LOS in the presence of OVA prior to co-culture with CD4 ϩ T cells, no inhibition of T cell proliferation was observed (Fig. S1B). When 10-fold more purified LOS (1 g) was used, a slight inhibition of T cell proliferation was observed, but this trend did not meet statistical significance (Fig. S1C). These data indicate that purified LOS was unable to recapitulate the inhibition of dendritic cell-induced T cell proliferation that was observed with Ng-CM and suggest that outer membrane proteins may be responsible for the inhibition.
N. gonorrhoeae porin (PorB) is the most abundant gonococcal protein in both outer membranes and OMVs, constituting ϳ50% of the total protein content (19,20), and has been shown to have pleotropic effects on host cells through a variety of mechanisms (21)(22)(23). N. gonorrhoeae has two PorB serotypes, PorB 1A and PorB 1B ; the strain used in these experiments, FA1090, expresses the PorB 1B serotype. As anticipated, immu-

Immunosuppression by Neisseria gonorrhoeae PorB
noblot analysis demonstrated that PorB 1B was present in Ng-CM and that all detectable PorB 1B was retained by the 100-kDa cut-off filter (Fig. 3A). To examine the effects of PorB 1B on dendritic cell-mediated T cell proliferation, PorB 1B lacking its signal sequence and containing a N-terminal hexahistidine tag (HT) followed by a tobacco etch virus protease recognition site was expressed as inclusion bodies in E. coli and refolded into an active state in the presence of the detergent LDAO as described under "Experimental procedures." For some experiments, the hexahistidine tag was removed using tobacco etch virus prote- Immunosuppression by Neisseria gonorrhoeae PorB ase (Fig. 3B). Refolded HT-PorB 1B and PorB 1B in LDAO micelles were highly pure and eluted as homogenous single peaks in size exclusion chromatography. The estimated molecular mass of purified, refolded PorB 1B was ϳ250 kDa (Fig. 3C), consistent with two PorB trimers (ϳ114 kDa), the known tertiary structure for native N. gonorrhoeae PorB 1B , and an LDAO detergent micelle (ϳ25 kDa). Functional activity of refolded and purified PorB 1B was demonstrated by its capacity to promote size-dependent permeation of a series of saccharides following reconstitution into liposomes ( Fig. 3D) (24). The addition of purified PorB 1B (50 g/ml) to dendritic cells during OVA exposure led to profound inhibition of T cell proliferation in our co-culture system, recapitulating the effects observed with live N. gonorrhoeae treatment (Figs. 4, A and B). The addition of an equivalent volume of LDAO-containing buffer, the vehicle for PorB 1B , caused a small, reproducible decrease in dendritic cell-induced T cell proliferation, but this decrease was not statistically significant (Fig. S2). The PorB 1B preparations were found to have low levels of endotoxin contamination (between 1.5 and 6 endotoxin units/mg of protein), which, at the highest amount of PorB 1B added, corresponds to a maximum concentration of ϳ 2.5 ng of E. coli LPS/ml in the dendritic cell culture. To ensure that E. coli LPS contamination of the recombinant PorB 1B was not responsible for the PorB 1Bmediated inhibition of T cell proliferation observed, OVA-pulsed dendritic cells were exposed to E. coli LPS at concentrations between 1 and 100 ng/ml and assayed for the ability to stimulate OT-II T cell proliferation. E. coli LPS at these concentrations demonstrated no detectable effect on OVA-pulsed dendritic cell-induced T cell proliferation (Fig. S3).
Although these experiments demonstrated an inhibitory effect on dendritic cell-mediated T cell proliferation when recombinant PorB 1B was added to the culture, it did so at concentrations much higher than the levels of PorB 1B present in either the Ng-CM or live gonococci that produced an equivalent inhibitory effect (data not shown). However, because PorB 1B is an integral membrane protein, a substantial percentage of the PorB 1B is likely to aggregate or bind to the tissue culture plate when the protein is diluted from its detergentcontaining buffer into cell culture medium. Therefore, we sought to determine whether the amounts of dendritic cellassociated PorB 1B were similar between cells treated with recombinant purified PorB 1B and those treated with live N. gonorrhoeae by performing immunoblot analysis of the dendritic cells prior to their use in the co-culture T cell proliferation assay. The amount of PorB 1B associated with dendritic cells after treatment with 50 g/ml of purified PorB 1B was similar to that observed in cells treated with live N. gonorrhoeae at an m.o.i. of 10 ( Fig. 4C). We observed more cell-associated PorB 1B in cells treated with 5 g/ml of recombinant porin than with live N. gonorrhoeae at an m.o.i. of 1, even though the former caused little inhibition, whereas the latter decreases T cell proliferation by over 50%. These data indicate that N. gonorrhoeae likely produces additional inhibitory factors that complicate the comparison between live bacteria and a single purified protein. One such factor is the inhibitory interleukin IL-10, which has been shown to be secreted in response to live N. gonorrhoeae (10).
To further assess the loss of the suppressive effect of PorB 1B at concentrations lower than 50 g/ml, we tested the capacity of PorB 1B to inhibit dendritic cell-induced T cell proliferation at three different subcytolytic concentrations of LDAO detergent. We observed minor but consistent concentration-dependent decreases in proliferation with detergent alone (ϳ30% at the highest concentration of detergent). Although complete suppression of T cell proliferation was observed with 50 g/ml PorB 1B regardless of the detergent concentration, we also observed that, as the concentration of LDAO was increased, the inhibitory potency of PorB 1B increased (Fig. 4D). These data suggest that, when PorB 1B loses association with detergent micelles and presumably loses its native, membrane-associated trimeric structure, it also loses the capacity to inhibit dendritic cell-induced T cell proliferation.
To ensure that PorB 1B was not simply competing with OVA for proteolytic digestion or loading into antigen-presenting MHC molecules, we tested whether other proteins, either BSA or an isolated recombinant gonococcal protein (the transcriptional regulator GdhR) expressed in Escherichia coli, could inhibit dendritic cell-induced CD4 T cell proliferation. In contrast to recombinant PorB 1B , neither protein impacted the capacity of OVA-pulsed dendritic cells to induce OT-II T cell proliferation ( Fig. 4E) (25).

Immunosuppression by Neisseria gonorrhoeae PorB
Because N. gonorrhoeae is an exclusive human pathogen, we also assessed whether addition of PorB 1B would inhibit T cell proliferation in a model of human dendritic cell-induced T cell proliferation. As shown previously for treatment with live N. gonorrhoeae and with Ng-CM (Fig. 1, D and E), T cell proliferation induced by co-culture with allogeneic dendritic cells

Immunosuppression by Neisseria gonorrhoeae PorB
was inhibited when the dendritic cells were first treated with recombinant PorB 1B (Fig. 4, F and G). Moreover, neither BSA nor GdhR were capable of inhibiting T cell proliferation when added at the same concentration of purified PorB 1b .

PorB 1B suppression of dendritic cell-mediated T cell proliferation requires properly folded PorB 1B and is not mediated through TLR2 signaling
Purified Neisseria PorB proteins have been shown previously to have immunostimulatory activity in both cultured immune cells and when injected into whole animals (23, 26 -28). In these reports, the protein, which was either purified from N. gonorrhoeae or refolded from E. coli inclusion bodies, had the detergent used to solubilized this integral membrane protein removed through a process of ethanol precipitation and dialysis during the final steps of preparation. The resulting detergentfree protein aggregates have been termed PorB "proteosomes" (29). We sought to test whether the suppressive effect on dendritic cell-induced T cell proliferation we observed with PorB 1B treatment of dendritic cells was independent of the folded structure of PorB 1B . Unfolded PorB 1B was prepared by dialysis into PBS of urea-solubilized recombinant PorB 1B inclusion bodies, whereas PorB 1B proteosomes were prepared from our refolded, detergent-containing PorB 1B preparations by ethanol precipitation and dialysis using the methods of Massari et al. (29) (Fig. 5A). Dynamic light scattering performed on refolded and proteosome preparations of recombinant PorB 1B clearly demonstrated that refolded PorB 1B in LDAO micelles exists as a single homogenous species with an estimated Stokes radius of 10 nm, whereas proteosome PorB 1B preparations without detergent formed much larger particles with a Stokes radius of ϳ700 nm (Fig. 5B). Based on the crystal structure of Neisseria meningitidis PorB, the size estimate generated by dynamic light scattering for the refolded protein is consistent with two PorB 1B trimers sitting within an LDAO micelle, which is also consistent with estimates generated by retention time in size exclusion chromatography (Fig. 5C). Each of these protein preparations resulted in highly purified PorB 1B , as determined by SDS-PAGE (Fig. 5C). These preparations were then tested for their effects on OVA-pulsed dendritic cell stimulation of OT-II T cell proliferation. As we had observed previously, refolded PorB 1B demonstrated suppression of T cell proliferation, but neither the unfolded nor the proteosome preparation of PorB 1B suppressed T cell proliferation in this system (Fig. 5D). These data suggest that the suppressive properties of PorB are disrupted when the structure of the protein is not maintained.
Neisseria PorB proteins have been shown previously to activate host TLR2 (28,30). Structural and functional studies indicate that the seventh extracellular loop of PorB 1B appears to be the PorB 1B -derived ligand for TLR2 (31). Unfolded, refolded, and proteosomal HT-PorB 1B was applied to a reporter cell line expressing human TLR2. Each PorB 1B protein preparation activated TLR2 under these conditions. Refolded PorB 1B generated

Immunosuppression by Neisseria gonorrhoeae PorB
a maximal TLR2 activation of a roughly 1.5-fold increase over baseline reporter activity. In contrast, proteosome PorB 1B preparations induced a nearly 7-fold activation, and unfolded PorB 1B induced a nearly 15-fold activation (Fig. 6A). These data suggest that high levels of TLR2 activation previously reported for proteosome PorB 1B preparations from a number of Neisseria species could be the result of high-binding valency associated with an aggregated protein ligand, whereas actual native PorB 1B stimulation of TLR2 is much more modest. Because TLR2 activation by capsular polysaccharide is known to mediate the immunosuppressive effects of outer membrane vesicles from the gut commensal Bacteroides fragilis, we sought to determine whether TLR2 engagement by PorB 1B was activating tolerogenic signaling pathways in dendritic cells by utilizing dendritic cells generated from Tlr2 Ϫ/Ϫ mice (32). T cell proliferation following co-culture with Tlr2 Ϫ/Ϫ dendritic cells was indistinguishable from the proliferation observed when co-cultured with WT dendritic cells (Fig.  6B). Furthermore, N. gonorrhoeae or refolded PorB 1B treatment of dendritic cells prior to co-culture with T cells resulted in equivalent inhibition of T cell proliferation regardless of the genotype of the dendritic cell. Taken together, these data suggest that, although TLR2-activating capacity is present in isolated N. gonorrhoeae PorB regardless of the folded structure of the protein, the immunosuppressive effects of the protein, which requires functionally folded PorB, are dominant and overcome any TLR2-activating activity.
Inhibition of dendritic cell-induced T cell proliferation by N. gonorrhoeae PorB was observed even 3 days into the dendritic cell/T cell co-culture, suggesting that PorB acts directly on the dendritic cells rather than by inducing inhibitory T cell signaling, such as T regulatory cell differentiation, within the co-culture (Fig. S4). To test whether PorB treatment impacted the maturation of dendritic cells exposed to OVA prior to coculture with T cells, we examined the fraction of cells with high CD11c and MHC class II expression and found no significant differences between cells treated with refolded, unfolded, or proteosome PorB (Fig. 7, A and B). In contrast, we observed that treatment of dendritic cells with refolded PorB resulted in a significant decrease of the co-stimulatory dendritic cell surface protein CD86 compared with treatment with unfolded or proteosome PorB (Fig. 7C). Treatment of dendritic cells at the highest doses of refolded PorB (50 g/ml) also caused a significant decrease in surface expression of the costimulatory molecule CD40 compared with treatment with proteosome PorB (Fig. 7D). Taken together, these studies support the finding that properly folded N. gonorrhoeae PorB acts directly on dendritic cells to reduce their capacity to stimulate antigen-driven T cell proliferation.

Discussion
N. gonorrhoeae is a highly adapted human pathogen that is closely related to commensal Neisseria species. Like commensal Neisseria that live on human mucosal surfaces without eliciting protective immunologic responses, N. gonorrhoeae has the ability to escape the human immune response. Mechanisms involved in successful evasion of N. gonorrhoeae from human immune responses are multifactorial and complex. Recent findings have suggested that N. gonorrhoeae can actively suppress host immune responses through interaction with differ-

Immunosuppression by Neisseria gonorrhoeae PorB
ent types of immune cells. Antigen-presenting dendritic cells direct host immune responses toward either immunity or tolerance (33,34). Studies have shown that the human vaginal mucosa contains four major subsets of myeloid-derived dendritic cells, each displaying unique functions that direct different types of immune responses (35). Therefore, understanding how N. gonorrhoeae interacts with dendritic cells to produce an end response is an important step in devising strategies to overcome the immune evasion of this pathogen. Prolific blebbing of OMVs is a characteristic of Neisseria species; however, the consequences of this process are not fully understood (16,17). A recent study demonstrated that the polysaccharide capsule associated with OMVs from the commensal bacillus B. fragilis is capable of inducing a tolerogenic phenotype in dendritic cells (32). Our studies demonstrate that an OMV-containing fraction of conditioned medium from N. gonorrhoeae is also capable of inducing immune suppression in dendritic cells. Unlike B. fragilis, N. gonorrhoeae does not produce capsular polysaccharide, which led us to further investi-gate the key components of Ng-CM that harbor these immunosuppressive properties. Yu et al. (36) recently demonstrated that dendritic cells exposed to N. gonorrhoeae suppress CD4 ϩ T cell responses to HIV-derived antigens presented by those dendritic cells. This suppression appears to be mediated largely by activation of the immune-inhibitory CEACAM1 receptor on dendritic cells by N. gonorrhoeae Opa proteins, which have also been implicated in promoting production of the immune-suppressive cytokines IL-10 and transforming growth factor ␤, thereby inhibiting T cell proliferation (13). However, because N. gonorrhoeae Opa proteins bind only to human CEACAM-1 but not mouse CEACAM-1 (37), mechanisms exclusive of Opa-CEACAM-1 interactions must exist that are responsible for suppression of antigen-dependent T cell proliferation by mouse dendritic cells exposed to N. gonorrhoeae or Ng-CM. Although OMV-based vaccines have demonstrated that OMVs are capable of inducing immune responses to proteins and polysaccharides associated with OMVs, the relative antigenicity of OMV-associated proteins compared with those of the

Immunosuppression by Neisseria gonorrhoeae PorB
isolated proteins or proteins in association with bacterial pathogen-associated molecular patterns (PAMPs) that are also found in OMVs is not known. It is possible that physiologic levels of OMV shed by pathogenic and commensal bacteria actually suppress immune responses, as suggested by this study and studies published previously (32). In addition to PorB, LOS is another major component of OMVs shed by Neisseria species. LOS is a potent inducer of the inflammatory cytokine response, and N. meningitidis LOS has adjuvant properties that presumably are mediated by TLR4 activation by the lipid A moiety of LOS (38). The oligosaccharide component of gonococcal LOS can vary structurally because of phase-variable expression of enzymes in the synthetic pathway (39), and these LOS variants have been reported to alter the cytokine secretion profiles of dendritic cells because of their stimulation of different surface lectin receptors, with changes in the cytokine profile influencing CD4 ϩ T cell subtype differentiation (40). We found that, unlike N. gonorrhoeae cells or OMV-containing Ng-CM, purified gonococcal LOS was unable to elicit inhibition of T cell proliferation by OVAprimed dendritic cells.
PorB is the most abundant outer membrane protein in N. gonorrhoeae. Importantly, we found that purified recombinant PorB 1B has the capacity to inhibit dendritic cell-mediated CD4 ϩ T cell proliferation. Neisseria PorB proteins have been shown to activate host TLR2 responses (28,30,41), and purified PorBs from several commensal Neisseria species have been shown to stimulate immune responses when injected in mice and to induce dendritic cell maturation when added to cells in culture (26,28,42). The major difference between those published studies and this study is the preparation of PorB. All published studies demonstrating immune activation by PorB have been carried out with a PorB preparation in a detergentfree particle termed a proteosome (28,43). Because PorB is an integral membrane protein, it is unlikely that it maintains its native functional structure under these conditions. The purified recombinant PorB in the nontoxic detergent (LDAO) micelles we use in this study mimics native PorB both structurally and physiologically, as shown by its performance on gel filtration chromatography and its capacity to facilitate sugar permeation when reconstituted into liposomes (Fig. 4C). The potency of detergent-solubilized PorB 1B for inhibiting dendritic cell-mediated T cell proliferation increases as the amount of LDAO in the culture medium is increased, which, we suspect, suppresses the tendency of PorB to aggregate when LDAO micelles disperse upon dilution. Importantly, recombinant PorB 1B in the absence of refolding or refolded PorB 1B taken through the precipitation and detergent removal procedure used to generate PorB proteosomes in other published reports eliminates its immunosuppressive activity. These data strongly suggest that the native micelle/ membrane-associated structure of PorB is required for its inhibitory function.
Unfolded, refolded, and proteosome PorB 1B all have the capacity to stimulate TLR2, as has been reported previously, but at markedly different levels. Our data suggest that proteosome PorB 1B retains its TLR2-stimulating activity while simultaneously eliminating the immunosuppressive activity of PorB 1B , which would explain the reported immunologic adjuvant properties of Neisseria species PorB that populate the current biomedical literature. Interestingly, PorB proteosomes have been shown to induce much more robust anti-PorB antibody responses than an equivalent quantity of native gonococcal PorB administered in naturally derived outer membrane vesicles (43). Given that poor immune responses to N. gonorrhoeae in infected humans are well-documented, there is certainly indirect evidence that naturally occurring gonococcal PorB associated with the bacteria or OMVs does not support a robust immune response in the natural host.
Unlike B. fragilis OMV-associated capsular polysaccharide, which induces a tolerogenic dendritic cell phenotype through host TLR2 activation (32), N. gonorrhoeae cells and purified recombinant PorB 1B inhibited the capacity of dendritic cells derived from Tlr2 Ϫ/Ϫ or WT mice to induce T cell proliferation. These data demonstrate that properly folded PorB from N. gonorrhoeae present in whole bacteria or in OMVs shed from the bacteria are capable of exerting immunosuppressive effects on dendritic cells. The mechanism underlying this suppression has yet to be determined. Gonococcal PorB has been shown to traffic to host cell mitochondria. In some studies, this interaction has been reported to promote host cell apoptosis, whereas, in others, it has been reported to inhibit host cell apoptosis (21,44,45). Some gonococcal PorB proteins have been shown to interact with host cell surface receptors, including a scavenger receptor expressed in endothelial cells and Gp96, both of which have been implicated in immune cell signaling and immune tolerance (46 -48). These and potentially novel signaling pathways clearly need to be investigated further for their potential role in mediating immunosuppressive signaling downstream of gonococcal PorB.

Ethics statement
All experiments involving the use of mice were conducted in accordance with National Institutes of Health guidelines for the care and use of laboratory animals and were approved by the Institutional Animal Care and Use Committee at UNC. Human dendritic cells were obtained from subjects enrolled in a UNC Institutional Review Board-approved study (Study 05-2860) who were deidentified prior to use in this study. The UNC Office of Human Research Ethics reviewed the proposed use of deidentified human subject-derived dendritic cells and determined that the use described (Study 12-0024) does not constitute human subject research as defined under federal regulations (45 CFR 46.102 (d or f) and 21 CFR 56.102(c)(e)(l)) and does not require further IRB approval.

Preparation of N. gonorrhoeae and N. gonorrhoeae conditioned medium
Inocula of a predominantly Opa ϩ frozen stock of FA1090 of N. gonorrhoeae were prepared as described previously (14). To minimize phase or antigenic variation, passage of FA1090 and its derivatives was kept to a minimum. The bacterial density of each inoculum was estimated by measuring A 600 and confirmed by plating of serial dilutions.

Immunosuppression by Neisseria gonorrhoeae PorB
To generate Ng-CM, N. gonorrhoeae strain FA1090 was grown overnight from frozen stock streaked on GC medium broth (GCB) plates. The cells were swabbed from the plates, introduced into 10 ml of GWM at an A 600 of 0.2, and grown for 4.5 h in a shaking incubator at 37°C with 5% CO 2 . Bacteria were removed by centrifugation at 1200 ϫ g for 10 min, followed by filtration through a sterile 0.22-m filter. The Ng-CM was further fractionated for some experiments by ultrafiltration using a Centriprep device with a 100-kDa cutoff filter (EMD Millipore, Billerica, MA) according to the manufacturer's instructions.

Purification and functional activity of recombinant N. gonorrhoeae PorB 1B
Recombinant PorB 1B was produced in E. coli as inclusion bodies and refolded using a modification of the method described by Olesky et al. (24). Full details are provided in supporting Materials and Methods, but a brief description is as follows. PorB 1B expression was induced in BL21*(DE3) harboring pT7-HTb-por1090, which replaces the PorB signal sequence with a 28-amino acid extension (HT) containing a hexahistidine tag and a tobacco etch virus protease cleavage site at the N terminus of the protein (HT-PorB 1B ). HT-PorB 1Bcontaining inclusion bodies were isolated and solubilized in 8 M urea. Unfolded HT-PorB was prepared from urea-solubilized inclusion bodies by dialysis into PBS. Refolded HT-PorB 1B was prepared by slowly diluting urea-solubilized inclusion bodies into refolding buffer (200 mM 3-(cyclohexylamino)propanesulfonic acid, 400 mM NaCl, 50 mM Tris-HCl, 0.3% LDAO (pH 11)), and after stirring overnight, the solution's pH was lowered to 8.0 with concentrated HCl, and any precipitated protein was removed by centrifugation and filtration. The protein was purified by immobilized Ni 2ϩ chromatography. In most experiments, the protein used had the HT tag removed by digestion with tobacco etch virus protease, and the untagged, properly folded PorB 1B was isolated in the flow-through of a nickel-nitrilotriacetic acid column run in the presence of 15 mM imidazole, followed by size exclusion chromatography on a Sephacryl S-300 column in 1ϫ PBS, 0.1% LDAO (PBSL). Preliminary experiments showed no difference between PorB 1B and HT-PorB 1B for inhibition of dendritic cell-mediated T cell proliferation (Fig. S5); therefore, in later experiments, the HT tag was left attached to the protein. Proteosome HT-PorB 1B was prepared from refolded HT-PorB 1B by precipitation with ethanol, resuspending the precipitated protein in 8% ␤-octyl glucoside, and then dialyzing in PBS to remove the detergent. Purified PorB preparations were tested for the presence of LPS using a chromogenic Limulus amebocyte lysate endotoxin assay (Toxinsensor TM , Genscript) and human TLR4-expressing reporter cells (HEK-Blue TM hTLR4, Invivogen) according to the manufacturer's protocol.
Functional activity of refolded and purified PorB 1B trimers was demonstrated by the swelling assay originally described by Nikaido and Rosenberg (49). Permeation rates of arabinose (MW 150), glucose (MW 180), galactose (MW 180), GlcNAc (MW 221), sucrose (MW 342), and raffinose (MW 595) were quantified as a ratio of the rate of permeation of the indicated sugar to the permeation rate of arabinose. Stachyose (MW 660) was used to find the isoosmotic concentration of the liposome preparation as described previously (24,49).

Generation, infection, and stimulation of bone marrowderived dendritic cells
Bone marrow-derived dendritic cells (BMDCs) were prepared from 9-to 12-week-old C57BL/6 or C57BL/6 Tlr2 Ϫ/Ϫ mice (The Jackson Laboratory, Bar Harbor, ME) as described previously (14). After 7 days of growth and differentiation, BMDCs were washed, resuspended, and incubated with 100 g/ml OVA (Sigma-Aldrich, St. Louis, MO) and either N. gonorrhoeae FA1090 at the indicated m.o.i., GWM, Ng-CM, purified gonococcal LOS purified from FA1090, recombinant PorB 1B , or PBSL (vehicle). After incubation for 24 h, the cells were collected and washed for co-culture with T cells or downstream assays.

BMDC-T cell co-culture/proliferation assay
T cells were isolated from spleens and lymph nodes of OT-II mice (C57BL/6-Tg(TcraTcrb)425Cbn/J, The Jackson Laboratory) and labeled with CFSE (Life Technologies) as described previously (14). Labeled, enriched T cells (5 ϫ 10 5 cells/ml) were co-cultured with BMDCs at a density of 5 ϫ 10 4 cells/ml in 48-well plates or plates containing transwell inserts (Costar, Corning, NY). After 7 days in co-culture, T cell proliferation was assessed by measuring the dilution of CFSE fluorescence in CD4 ϩ ,TCRV␤5 ϩ lymphocytes using flow cytometry on a BD FACSCanto or a BD LSRII-SOS (BD Biosciences, Palo Alto, CA) as described previously (14). Data from 10,000 -100,000 events were acquired for each sample and saved as an FCS 3.0 file that was subsequently analyzed with FlowJo software (Tree Star, Ashland, OR).

Assessment of TLR2 activation
HEK-Blue TM hTLR2 reporter cells were acquired from Invivogen (San Diego, California). The cells were cultured in growth medium (Dulbecco's modified Eagle's medium (Gibco) containing 10% heat-inactivated fetal bovine serum, L-glutamine (2 mM), glucose (4.5 g/liter), penicillin/streptomycin (50 units/ml and 50 g/ml), and Normocin (100 g/ml)) and then maintained in selection medium (growth medium supplemented with 100 g/ml neocin). Cells were seeded in 96-well plates at a density of 5 ϫ 10 4 cells in 200 l per well, and after 24 h, they were exposed to vehicle, the indicated concentrations of PorB 1B preparations, or 30 ng/ml Pam3CSK4 (Invivogen). Secreted alkaline phosphatase activity was assessed after 24 h of incubation by adding 20 l of cell culture supernatant to 180 l of Quanti-blue TM reagent (Invivogen) in a new 96-well plate and measuring optical density at 625 nm after a 2 h incubation at 37°C.

Statistical analysis
Statistical analyses were performed using Prism 5 or 6 software (GraphPad, La Jolla, CA). Significance of the differences between multiple groups was assessed using one-way or twoway ANOVA with Tukey or Bonferroni as a post hoc test for multiple comparisons. In all cases, p Ͻ 0.05 was considered statistically significant.