|Home | About | Journals | Submit | Contact Us | Français|
The oral cavity is home to unique resident microbial communities whose interactions with host immunity are less frequently studied than those of the intestinal microbiome. We examined the stimulatory capacity and the interactions of two oral bacteria, Porphyromonas gingivalis (P. gingivalis) and Fusobacterium nucleatum (F. nucleatum), on Dendritic Cell (DC) activation, comparing them to the effects of the well-studied intestinal microbe Escherichia coli (E. coli). Unlike F. nucleatum and E. coli, P. gingivalis failed to activate DCs, and in fact silenced DC responses induced by F. nucleatum or E. coli. We identified a variant strain of P. gingivalis (W50) that lacked this immunomodulatory activity. Using biochemical approaches and whole genome sequencing to compare the two substrains, we found a point mutation in the hagA gene. This protein is though to be involved in the alteration of the PorSS/gingipain pathway, which regulates protein secretion into the extracellular environment. A proteomic comparison of the secreted products of the two substrains revealed enzymatic differences corresponding to this phenotype. We found that P. gingivalis secretes gingipain(s) that inactivate several key proinflammatory mediators made by DCs and/or T cells, but spare Interleukin-1 (IL-1) and GM-CSF, which can cause capillary leaks that serve as a source of the heme that P. gingivalis requires for its survival, and GM-CSF, which can cause epithelial-cell growth. Taken together, our results suggest that P. gingivalis has evolved potent mechanisms to modulate its virulence factors and dampen the innate immune response by selectively inactivating most proinflammatory cytokines.
Humans harbor diverse microbial communities in various tissues, and these microorganisms can dramatically alter the local and systemic immune functions. The bi-directional interactions between the microbiome and the immune system have been extensively studied in the gut and skin, but relatively little is known about such events in the oral cavity [1–3], which is exposed to an extraordinarily wide variety of external agents. In addition to encounters with food, air and fluids from other individuals, the microbiome of the oral cavity includes over 700 species of microbes, some of which are thought to be involved in pathological conditions ranging from caries to cancer [4–7]. Among the oral microbiota, Porphyromonas gingivalis (P. gingivalis) is relatively well studied because of its association with chronic inflammation of gum tissue (periodontitis) [8, 9]. Although P. gingivalis, by itself is not sufficient to cause gingivitis or bone loss , it has been classified as “keystone pathogen”  based on the findings that it seems to be the essential component of the microbial community that causes oral disease .
An intriguing possibility is that P. gingivalis sets the stage for the entire cascades of disease by altering the local immune microenvironment, thus allowing pathobionts to subsequently colonize or overgrow . There are supportive evidence for this notion that has come from studies on P. gingivalis interactions with cells of the immune system. It had been suggested that P. gingivalis Lipopolysaccharide (LPS) skews the immune responses away from an inflammatory TH1-type response and towards a semi-TH2-like response [13, 14]. There are also data showing that P. gingivalis can directly act on complement system through its arginine-specific gingipains to block antimicrobial response [10, 15–17]. LPS is not the only molecule in a bacterium’s arsenal. We therefore asked if intact P. gingivalis might have even stronger effects on the activation of Dendritic Cells (DCs), one of the earliest immune cells to be activated in an immune response.
Because P. gingivalis has been shown to need companion or “bridge” organisms , we also used the oral commensal Gram-negative anaerobe, F. nucleatum that co-aggregates with, and promotes the colonization of periodontitis-associated bacteria such as P. gingivalis [18, 19]. Although F. nucleatum primarily colonizes the human oral cavity, it has been associated with diseases of various other organs, including periodontal, lung, abdominal and gynecological infections [20–23]. We postulated that comparing the two oral microbes for their direct impact on DCs, alone and in combination, might offer some insight into how individual components of the oral microbiota might contribute to control DC activation and its downstream consequences on T cell effector functions.
Using phenotypic and functional assays, we showed that P. gingivalis induced very low levels of inflammatory gene expression in DCs, in contrast to F. nucleatum and E. coli, which were potent activators. In addition, P. gingivalis inactivated a broad assortment of human and mouse cytokines and chemokines that are normally produced by activated DCs, thus further dampening the inflammatory cascade. We compared a mutant strain of P. gingivalis (W50) to the standard strain obtained from American Type Culture Collection (W50-ATCC). We discovered that a mutant strain has markedly stronger immunosuppressive properties compared to W50-ATCC, and was able to degrade a wide variety of mouse and human cytokines and chemokines.
Taken together, our data suggest that P. gingivalis has evolved the ability to operate somewhat under the immunological radar. It is a very weak activator of DCs. It also dampens DC activation by other bacteria, and it backs that up with the ability to degrade almost all of the cytokines and chemokines released by activated DCs and T cells. Such immunomodulatory mechanisms may underlie P. gingivalis ability to operate as a “keystone pathogen” in the oral cavity.
The authors have declared no conflicts of interest.
Adult 8-14-wk-old male and female B10.A RAG2-/-, mice and 5C.C7 RAG2-/- TCR transgenic mice [containing T cells specific for peptide 88–103 of moth cytochrome c (MCC)] were purchased from the NIAID/Taconic Farms, Inc. exchange. Mice were housed in the NIH animal facilities (an AAALAC accredited facility) in a specific pathogen-free barrier colony that is certified by the United States Office of Laboratory Animal Welfare (OLAW) under the guidance of the Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals. All studies were carried out in accordance with protocols LIG-32 and LCMI-14E approved (5/01/2013) by the Institutional Animal Care and Use Committee (IACUC) of the NIAID, NIH. All experimental protocols involving mice in this paper were specifically stated and approved by the IACUC in LIG-32 or LCMI-14E. The mice euthanasia protocol approved under this protocol involves CO2 asphyxiation followed by cervical dislocation. In general mice are placed in a 20% Co2 displacement unit until they are unconscious and then the Co2 is increased to 100%, followed by cervical dislocation as a secondary method.
Bacterial Lipopolysaccharide (LPS) was derived from E. coli Serotype 0127:B8 (Sigma Aldrich). All recombinant cytokines were obtained from Peprotech Inc. Cells were cultured in complete medium [(Iscove’s Modified Dulbecco’s Medium (IMDM: GibcoBRL)] supplemented with 10% heat-inactivated Fetal Bovine Serum (FBS: tested to be free of endotoxin, mycoplasma, virus, and bacteriophage; GibcoBRL), L-glutamine, 55 μM 2ß-Mercaptoethanol (GibcoBRL), penicillin, streptomycin, and gentamicin (Biosource). The antibiotics prevented any bacterial growth in our cultures, even when we added large numbers of living bacteria.
Cytokines in the culture supernatant (CSN) were quantitated using SearchLight Multiplex cytokine array (Aushon Biosystems, Billerica, MA). IL-12 heterodimer (IL-12p75) was specifically detected by using the ELISA kit Quantikine (R&D Systems).
Briefly, bone marrow cells were flushed out of the femurs and tibias of 8-12-wk-old mice into complete medium and pipetted vigorously to make a single cell suspension, then passed through a cell strainer (70-μm Nylon mesh; BD Falcon TM). Erythrocytes were lysed using ACK lysis buffer (Biosource) and cells washed 2x with complete medium and cultured at 1x106 cells in 2ml/well in a 24-well plate with complete medium supplemented with 30U/ml GM-CSF and 60U/ml IL-4 for 6 days as described previously .
Spleens from B10.A/SgSnAi TCR–Cyt 5C.C7-1 RAG-2-/- mice were dissociated into homogenous single cell suspensions and passed through a cell strainer (70-μm Nylon mesh) from BD FalconTM. Erythrocytes were lysed using ACK lysis buffer (Biosource). Splenocytes were washed 3x in cold complete medium and adjusted to 1x106/ml, then cultured in complete medium plus 1 μM of the Moth Cytochrome C (MCC) 88–103 peptide at 37°C with 5% CO2. Fresh medium was added after 3 days of culture. After 5 days, the cultured cells were harvested, washed 2x with medium, and re-cultured in the absence of peptide with 10–20 U/ml of recombinant mouse (rm) IL-2. The medium was replenished every 5–7 days with fresh medium plus rmIL-2. For T-DC coculture experiments, these CD4+ T cells were purified with a CD4+ T cell isolation kit containing anti-CD8α, CD11b, CD45, DX5, and Ter-119 mAbs for depletion (Miltenyi Biotech.) with the addition of anti-CD11c and anti-class-II (Miltenyi Biotech). We used 3–4 LS columns in tandem instead of using LD columns to deplete the cells. These T cells, which were >95% CD4+, were used at various days of culture as a source of in vitro primed/antigen-activated T cells.
1x105 bead depleted (CD4+ T cell isolation kit, Miltenyi Biotech) cultured 5C.C7 CD4+ T cells were incubated with 2x104 BMDCs/well from B10.A RAG-/- mice ±0.1μM moth cytochrome c (MCC) peptide 88–103 in triplicate in 96-well u-bottom plates in a final volume of 200 μl/well for 48 h. The supernatants were removed and kept frozen at –30°C before measuring the cytokines by ELISA.
In coculture experiments, 5x107 bacteria were added to the 24-well plate containing BMDCs for 16 h. For bacterial supernatant (SN) preparation, ~1011 bacteria were harvested in the exponential stage of growth, washed twice with PBS, and incubated at 37°C in 5 ml PBS for 1 h. The cells were then spun at 4500 rpm for 15 min and the SN filtered through a 0.22-μm syringe filter before being used (bacteria SN). In order to treat cytokines, 200 μl of the bacteria SN or ~109 whole bacteria were added to 1 ml of complete medium (containing 10% FCS) spiked with various cytokines and kept at 37°C for 24 h before quantitating residual cytokines with cytokine array.
All steps were done on ice or in 4°C centrifuges to prevent the sudden upregulation of MHC class II caused by handling . Plates containing DC cultures were cooled for 20 min. The cells were then gently harvested, spun, counted and distributed into 96 well plates at about ~106 per well. We added 10 μl “ultrablock” (a mixture of 1:1:1 mouse, rat and goat serum containing 10μg/ml of the monoclonal anti-FcR antibody 2.4G2) for 10 min before adding a master-mix of a combination of antibodies anti-mouse CD80, CD40, CD11c and MHC class II (BD Pharmingen). Each antibody was titrated for optimal signal-to-background before use. After 20 min, the cells were washed and resuspended in PBS containing the viability dye, 7AAD plus 1% FCS, for 5 min, then centrifuged and resuspended in FACS buffer for analysis.
P. gingivalis used in this study 53978 (W50), BAA-308 (W83), ATCC33277, and F. nucleatum, were obtained from American Type Culture Collection (ATCC). P. gingivalis W50-NIDCR was a generous gift from Dr. P. Kolenbrander in the National Institute of Dental and Craniofacial Research (NIDCR). All oral bacteria were grown anaerobically (GasPak EZ; BD, Sparks, MD) in BactoTM Brain Heart Infusion (BD, Sparks, MD or Anaerobe Systems Morgan Hill, CA), supplemented with 5 mg/L hemin and 1 mg/L vitamin K. E. coli were grown in Luria Bertani Broth (K.D Medical Columbia, MD). Bacteria were harvested in the exponential stage of their growth and washed 2x with PBS before being used at an O.D. of 600 nm.
P. gingivalis was harvested in the exponential stage of growth and immediately incubated in a 65°C water bath for 1 h and washed 2x with PBS before coculturing with BMDCs.
To assess the gingipain activity of P. gingivalis strains, 20 μl of bacterial supernatant was mixed with Nα-Benzoyl-DL-arginine β-naphthylamide hydrochloride (BANA) substrate and 100 μl of 100 mM Tris-HCl buffer pH 7.4 (with 2mM dithiothreitol), and incubated at 37°C for 2 h. Optical densities at 410 and 600 nm were measured every 5 min. using a PowerWave HT (Biotek) spectrophotometer. Assays were performed three times in triplicate.
2.5 ml of cell-free bacterial SN were concentrated (10x) using Amicon Ultra-4 (10,000 MWCO) (Millipore) by spinning at 2500xg for 15 min at 4°C. The concentrated SN was separated by SDS-PAGE for one-dimensional gel and visualized by Coomassie or silver staining and two-dimensional gel was analyzed by a service contract (Applied Biomics, Hayward CA) as previously described . Given the robustness of the biological/enzymatic assays we observed, the proteomic analyses were performed once each.
BMDCs were cultured in a 24-well plate for 6 h with 108 bacteria/well and RNA was isolated from the cells by RNeasy kit (Qiagen). RNA was treated to remove DNA contamination (Invitrogen) and the cDNA was synthesized using random primers, and reverse transcribed using Superscript III (Invitrogen). Real-time PCR analyses were performed for expression of IL-1 α, IL-1β, TNF α, IL-6, IL-12p40, IL-10, IL-23p19, IL-12p35, and using Taqman gene expression primers and probes mix (Applied Biosystems). The mRNA expression levels were calculated as dCt = Target-Actin B/nono, where Actin B/nono was the endogenous control.
Peptides were identified from mass spectroscopy (MS) data using SEQUEST algorithms and a minimum of two peptides. Peptide thresholds of 95% were used for identification as previously described .
All statistical analyses were performed using Prism (GraphPad Software).
It is well established that DCs respond to various bacterial products from gut bacteria such as E. coli by upregulating cell surface costimulatory molecules and rapidly secreting cytokines and chemokines. To compare the ability of oral bacteria to elicit these functional responses by DCs, we used bone marrow-derived DCs (DCs) obtained from Recombinant Activating Gene-2 deficient mice (RAG2-/-), which lack T and B cells, to eliminate the possibility of B and/or T cell contamination (which might alter the function of the DCs). We stimulated the DCs with P. gingivalis and F. nucleatum, two different oral bacteria species that have been shown to co-aggregate in vitro [18, 27], and with E. coli and assessed their activated status 16 hours later. Fig 1A shows that nearly 100% of the DCs stimulated with E. coli upregulated the co-stimulatory molecule B7.1, and nearly 90% upregulated B7.2, CD40 and the antigen-presenting molecule, MHC class II. The oral anaerobe F. nucleatum was slightly less active, while P. gingivalis stimulated only about 12–15% of the DCs over the background.
We saw dramatic differences when we analyzed the secretion of cytokines by the activated DCs. Fig 1B shows that P. gingivalis failed to induce any of these cytokines from DCs, while both E. coli and F. nucleatum were potent inducers. In some cases, the cytokine levels in P. gingivalis-stimulated cultures were even lower than those secreted by the untreated controls.
To begin to uncover the mechanism by which P. gingivalis blocks cytokines, we started by determining whether it can block transcription. We used qRT-PCR to analyze the amount of cytokine mRNA in DCs after 6 hrs of stimulation with various bacteria. As shown in Fig 1C, both E. coli and F. nucleatum induced greater than 100-fold changes in the mRNA expression of various cytokines in DCs. In contrast, P. gingivalis induced only very low levels of mRNA for IL-1 α, IL-1 β and TNF α, and no mRNA for the other cytokines. Thus, P. gingivalis is a very poor activator of DCs.
To determine whether P. gingivalis was merely a weak activator of DCs or whether it might actually be suppressive, we asked whether P. gingivalis could modulate cytokine secretion by DCs activated by E. coli or F. nucleatum. We stimulated DCs with E. coli or F. nucleatum, plus IFNγ(strong stimulus for IL-12 production) , either alone or in combination with 5x107 P. gingivalis bacteria and found that the addition of P. gingivalis severely impaired the levels of IL-12 produced by the DCs (Fig 2A). The bacterium did not simply kill the DCs, as they remained viable (excluded Trypan blue) by stimulating T helper cells and receive help (S1 Fig). In addition, because it had been shown previously that the LPS of P. gingivalis does not stimulate strong inflammatory responses from DCs , we asked if the suppressive activity of P. gingivalis was due to its unique LPS  by asking if it was sensitive to heat. Fig 2B shows that the inhibitory activity was markedly reduced in DC cultures given heat-killed P. gingivalis. Thus, in addition to having a unique LPS, P. gingivalis also expresses a suppressive activity that is heat-labile, suggesting that the suppressive activity is protein.
To identify the genetic loci associated with P. gingivalis suppressive function, we examined several strains of P. gingivalis to determine whether there might be genetic variation among them that would allow us to locate relevant gene(s). Thus far, we had used a W50 strain [a gift of Dr. P. Kolenbrander, National Institute of Dental and Craniofacial Research, (NIDCR)], referred to hereafter as W50-NIDCR. To test whether the suppressive activity was unique to this strain, we compared it to three other strains of P. gingivalis (W83, W50, and 33277) from the American Type Culture Collection (ATCC). We stimulated DCs overnight with E. coli LPS plus IFNγ , in the absence or presence of each P. gingivalis strain, and measured IL-12 protein in the CSN.
Fig 2C shows that W50 from ATCC (hereafter refer to as W50-ATCC) was considerably less potent than W50-NIDCR at reducing IL-12 production (Fig 2C), while strain W83 was only slightly less inhibitory than W50-NIDCR. Fig 2D shows that, when the two different W50 P. gingivalis were titrated, W50-NIDCR was almost 100 times more potent than W50-ATCC at inhibiting IL-12 production.
The evolutionary adaptation of a microbe to its host often involves multiple genetic modifications. To determine the mechanism by which W50-NIDCR was able to potently modify its virulence factors, we took a genomic approach to compare the four P. gingivalis strains and as a control, we compared them to E coli.
First, we sequenced the whole genomes of W50-NIDCR, W50-ATCC and ATCC 33277, and assembled contigs using the reference genome from the previously sequenced W83 strain (Fig 3A). Across the entire genome of the W50-NIDCR, we were able to identify unique single nucleotide polymorphisms (SNPs) in 20 different loci by BRESEQ analysis. Two of these loci were the 16S and 23S rRNA, while SNPs in 4 of the other loci fell into non-coding or intergenic regions. Thirteen of the remaining 14 loci included a high concentration of SNPs in multiple predicted protein coding regions with yet unknown function.
The 14th was a SNP difference in the coding region of the hagA gene, which encodes hemagglutinin A, a 280-kDa protein encoded by an 8 Kb gene containing multiple large direct repeat units [30, 31]. A single A→T transversion in this locus in W50-NIDCR resulted in a codon change (AAA to TAA) yielding a premature stop codon after amino acid 75 in the open reading frame of the HagA gene (Fig 3B).
HagA is an outer membrane protein that has cysteine proteinase adhesin domains involved in adherence to host cells and to other bacteria [32, 33]. It is also a major virulence determinant in P. gingivalis . Moreover, during periodontal disease, antibodies targeted the protein products of HagA [35–37]. In addition to its role in adhesion, there is some evidence that HagA may have a regulatory effect on the processing and/or secretion of proteins . We therefore used a proteomic approach to look for proteins that might be differentially secreted by W50-NIDCR and W50-ATCC.
We compared cell-free SNs from each strain, resolved the proteins on one-dimensional SDS-PAGE and revealed by silver staining (Fig 4A). This preliminary analysis suggested that there were indeed unique differences in proteins of different molecular weights secreted by each strain. To further identify these proteins, we concentrated the SNs and visualized the resolved protein bands by Coomassie blue staining (Fig 4B). We excised total of 13 spots (Coomassie gel) from W50-ATCC and W50-NIDCR combined and processed them for identification by liquid chromatography-tandem mass spectrometry (LC-MS). However, we focused only on those proteins that unique to W50-NIDCR in comparison to W50-ATCC, which are marked 7–13 (Fig 4B) while the complete comparison is listed in S2 Fig. These bands contained peptides from both arginine- and lysine-specific gingipains (Table 1), a complex set of “trypsin-like” cysteine and arginine proteases that contribute to the degradation of various host proteins [39, 40]. One of these proteins (immunoreactive 61 kDa antigen PG91) was found in multiple bands (bands 7–13) with no known function. The reason for the presence of this protein in multiple bands is indicative of either posttranslational modification of PG91 or that it might be a target of other active proteolytic enzymes in W50-NIDCR (Table 1). The other non-gingipain protein identified in the W50-NIDCR supernatant was Omp28—an outer membrane protein expressed by a wide variety of Gram-negative bacteria, including P. gingivalis isolates . Many outer membrane proteins are part of the secretory system (small size vesicles 20–500 nm), which are the means by which bacteria disseminate their virulence factors into the extracellular milieu to modulate host immune responses , and have been shown to contain gingipain and other proteolytic enzymes .
We decided to further characterize the proteome of the SN obtained from the two W50 strains at higher resolution using two-dimensional difference in gel-electrophoresis (2D DIGE). Therefore, we analyzed the two W50 SNs by labeling them green for W50-ATCC and red for W50-NIDCR (Fig 4C). The protein spots that showed differential abundance between W50-ATCC and W50-NIDCR were excised and subjected to LC-MS analysis. Table 2 lists a number of distinct proteins (13 spots containing at least 3 peptides. comprising 35 proteins) that were highly enriched in W50-NIDCR (orange and red spots) compared to W50-ATCC (green). Like the 1D analysis, the 2D gel data showed that the supernatant of W50-NIDCR showed more gingipain when compared to the supernatant of W50-ATCC, and that many of the 35 proteins identified from the SN were either gingipain or involved in the processing of gingipain (Table 2). This analysis also revealed an over-expression of several proteases that are associated with the secretory system (PorSS/gingipain pathway). These include multiple gingipains (Rgp and Kgp), PrtH domain proteases, various hemagglutinins, and the majority of these proteins contained C-terminal domains (CTDs) that are usually processed by PorSS/gingipain pathways.
To directly test the levels of gingipain in the supernatants of W50-NIDCR, W50-ATCC, (and F. nucleatum as a control) we used an enzyme assay specific for Arg-gingipains (Rgp)  and found that the SN from W50-NIDCR contained significantly more gingipain than the supernatants of W50-ATCC and F. nucleatum (Fig 5). Importantly, various P. gingivalis strains possess membrane-associated gingipain activities on the cell surface, and secretion of gingipain from P. gingivalis has not been widely reported yet  except in one strain (HG66) [45, 46]. Collectively, these data suggest that the mutation in W50-NIDCR results in an alteration in PorSS/gingipain secretory pathway and increased secretion or translocation of gingipain from the bacterial outer membrane and its accumulation in the extracellular environment. Alternatively, it is also plausible that this mutation affects gingipain surface attachment. These would need future studies to clarify
The data so far suggested that the lack of cytokine production we had seen at the mRNA level might not be the only mechanism by which P. gingivalis suppresses immunity, but that P. gingivalis seems to actively secrete proteases that act directly on the cytokines produced by activated immune cells. To test this, we obtained a large set of recombinant mouse cytokines, diluted them to known concentration, pooled them and spiked the sample with either cell-free culture supernatants or intact bacterium from F. nucleatum, W50-ATCC and W50-NIDCR, incubated them for 22 hours at 37°C, and measured the remaining cytokine levels. Fig 6 show that the supernatants from W50-NIDCR destroyed all of the tested cytokines except IL-12p40, GM-CSF and IL-1 α. The intact bacteria were even more potent, destroying both IL-12p40 and GM-CSF while still having little effect on IL-1 α.
Since P. gingivalis is a human pathogen that does not normally infect mice, we asked if the destructive activity on mouse cytokines was an evolutionary useless feature or whether it would also hold for molecules involved in human immune responses. We tested a wide array of human cytokines and chemokines and found that all of them were sensitive to the action of the bacterial supernatant except IL-1 α, VEGF and GM-CSF (Fig 7). Collectively, these data demonstrate that inactivation of human cytokines are similar to mouse in which the supernatant from W50-NIDCR was more effective than that of W50-ATCC in inactivation of various human cytokines and chemokines.
The goal of this study was to examine the broad influence of oral bacteria on DC activation and cytokine production. In this study, we used two oral bacteria, P. gingivalis and F. nucleatum, and compared their properties to those of the well-studied bacterium E. coli. Using assays that directly measure DC functional responses, a series of genomic and proteomic analyses, a gingipain-specific enzyme assay and direct tests on mouse and human cytokines, we found that P. gingivalis dampens innate immune responses. First, P. gingivalis is not a strong activator of DCs. DCs cultured in the presence of P. gingivalis hardly upregulate costimulatory or antigen-presenting MHC molecules. They secreted only small amounts of inflammatory cytokines and did not contain high levels of mRNA for those cytokines in the presence of P. gingivalis. Second, P. gingivalis extinguishes DC activation induced by other bacteria. Because when DCs are activated by E. coli or F. nucleatum (a commensal oral bacterium) in the presence of IFNγ, they generally secrete copious amounts of inflammatory cytokines such as IL-12. However, if P. gingivalis was added to such cultures, the cytokine secretion induced by E. coli or F. nucleatum was suppressed. Because P. gingivalis secretes proteases such as the gingipains, that degrades many of the mouse and human cytokines that we had tested, this will ensure a non-response even if other local oral bacteria (such as F. nucleatum) induce immune responses. In fact, neighboring bacterial populations seem to benefit from the presence of P. gingivalis, as they tend to be increased in number (especially Staphylococcal, Streptococcal and Propionibacterial species), showing that the presence of P. gingivalis is not only a benefit to itself but also to other local bacteria .
Intriguingly, we found that P. gingivalis’s secreted products do not affect human IL-1 and GM-CSF (Fig 7A). In contrast, P. gingivalis had strong effects on mouse IL-1 β and GM-CSF (Fig 6). Whereas mouse IL-1 β contains an arginine (a target of gingipain) at position 126 (which would be exposed at the end of the first beta strand in the crystal structure of mouse IL-1 β(PDB id 8I1B), position 126 in the human IL-1 β is changed to threonine. Similarly, there are regions of mouse GM-CSF that contain arginine or lysine residues (also targets of gingipain) that have different amino acids in the respective position of the sequence in the human molecule; and these regions in the structure of mouse GM-CSF are on loops that can be easily targeted by the enzyme. There are several ways to interpret the resistance of the human molecules to the bacterial proteases. One possibility is that the human molecules may have evolved to resist degradation by gingipains as a mechanism to enhance inflammation to remove the bacterium. Another possibility is that the bacterium may have taken an advantage of changes in IL-1 to benefit its survival. Like many pathogenic bacteria, P. gingivalis require iron for its growth and virulence [47, 48], and this is accomplished by hydrolyzing hemoglobin [49, 50]. At high doses, IL-1 α causes capillary leakage [51, 52], which benefits the bacterium that relies on heme for its survival. Indeed, there are data suggesting that P. gingivalis induces high levels of IL-1 from human gingival epithelial cells . Moreover, GM-CSF may also serve the bacterium, as it inhibits terminal differentiation of epithelial and endothelial cells and induces their growth [54, 55] thus, potentially providing a larger subgingival niche for the bacterium to survive. Combined, our data suggest that P. gingivalis evades/inactivate those molecules that will lead to its clearance and at the same time selectively spares those molecules that enhance its survival.
Finally, different studies on P. gingivalis have given different results. For example, it has been reported by Banchereau’s group, that the gums of people with gingivitis contained large numbers of immature/resting Langerhans cells , supporting our view that the bacterium is not an strong stimulator of DCs. In contrast, there are studies that have shown that very high doses of P. gingivalis LPS (doses that would have been lethal, had the LPS come from E coli), does stimulate DCs, but seems to prime T cells for the subsequent production of IL-5 rather than IFNγ [13, 14, 56]. In general, bacteria have more weapons that just the LPS. It has been shown that P. gingivalis (inside the cells) secretes a phosphatase (SerB) that prevents translocation of NF-kB to the nucleus (thus hindering cytokine secretion from gingival epithelial cells), several proteases that degrade complement components, and “capturing” ligands that bind complement antagonists . Nevertheless, it has also been shown that in spite of the SerB-mediated suppression, gingival epithelial cells secrete inflammatory cytokines . Our data suggest (perhaps not surprisingly) that the differences found from study to study might be due to the differences in the substrains of P. gingivalis that were used. The W50-NIDCR seems to be particularly effective at evading/dampening the secretion of innate and adaptive immune mediators.
Admittedly, this study has only begun to unravel what seems to be potentially deep nexus of interactions between W50 and DCs. While some of our observations suggest a linear correlation (eg. HagA and degradative ability of P. gingivalis supernatant), it is important to note that the biology of host-microbe interactions are complex. For instance, the composition of the P. gingivalis supernatant itself can be altered by vesiculation resulting from mutations in HagA secretory pathways. Indeed, as a reviewer has correctly pointed out, the mutation in the CTD region of HagA that we observed could also alter its localization. This can potentially short-circuit the ability of the PorSS pathway of CTD protein secretion, which will result in altered localization of gingipain proteinases and adhesins on the surface of bacterial cells. Future studies using more or less suppressive strain(s) isolated from patients may reveal which of these particular attributes correlate best with disease severity.
IL-12 can be elicited from resting or LPS-activated DCs by coculturing them with antigen-activated T cells . To test whether DCs that have been cultured with P. gingivalis are viable or dead, we tested their ability to produce IL-12 when subsequently cocultured with activated 5C.C7 T cells [specific for Moth Cytochrome c (MCC)]. We also tested whether the presence of the bacteria in the DC/T-cell culture would inhibit secretion of IL-12. (A) DCs from B10.A-Rag2-/- mice were pre-activated with F. nucleatum, P. gingivalis or nothing (medium) for 20 h, then washed and cocultured at 2x104 cells/well with 5x105 cell/well antigen-activated 5C.C7 T cells in the presence of 0.1 μM MCC peptide 83–103 in a 96-well plate at 37°C for 48 h. CSN were analyzed for the presence of IL-12 using specific ELISA. (B) Same as (A) except 5x107 F. nucleatum or P. gingivalis were added to the coculture of DCs with T cells. These data are expressed as the mean ± SD of triplicate wells representative of two independent experiments.
We thank Drs. Susan Pierce and Thomas van Dyke for comments and critical review of the manuscript; Paul Kolenbrnader for discussion and providing P. gingivalis W50-NIDCR; Ethan Tyler and Alan Hoofring for illustrations.
The Intramural Research Program at the National Institutes of Health, National Institute of Allergy and Infectious Diseases supported this work. N.J.S. is funded by the NIAID extramural grants R21AI126184 and R01AI110719. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
All relevant data are within the paper.