Macrophage Subset Sensitivity to Endotoxin Tolerisation by Porphyromonas gingivalis

Macrophages (MΦs) determine oral mucosal responses; mediating tolerance to commensal microbes and food whilst maintaining the capacity to activate immune defences to pathogens. MΦ responses are determined by both differentiation and activation stimuli, giving rise to two distinct subsets; pro-inflammatory M1- and anti-inflammatory/regulatory M2- MΦs. M2-like subsets predominate tolerance induction whereas M1 MΦs predominate in inflammatory pathologies, mediating destructive inflammatory mechanisms, such as those in chronic P.gingivalis (PG) periodontal infection. MΦ responses can be suppressed to benefit either the host or the pathogen. Chronic stimulation by bacterial pathogen associated molecular patterns (PAMPs), such as LPS, is well established to induce tolerance. The aim of this study was to investigate the susceptibility of MΦ subsets to suppression by P. gingivalis. CD14hi and CD14lo M1- and M2-like MΦs were generated in vitro from the THP-1 monocyte cell line by differentiation with PMA and vitamin D3, respectively. MΦ subsets were pre-treated with heat-killed PG (HKPG) and PG-LPS prior to stimulation by bacterial PAMPs. Modulation of inflammation was measured by TNFα, IL-1β, IL-6, IL-10 ELISA and NFκB activation by reporter gene assay. HKPG and PG-LPS differentially suppress PAMP-induced TNFα, IL-6 and IL-10 but fail to suppress IL-1β expression in M1 and M2 MΦs. In addition, P.gingivalis suppressed NFκB activation in CD14lo and CD14hi M2 regulatory MΦs and CD14lo M1 MΦs whereas CD14hi M1 pro-inflammatory MΦs were refractory to suppression. In conclusion, P.gingivalis selectively tolerises regulatory M2 MΦs with little effect on pro-inflammatory CD14hi M1 MΦs; differential suppression facilitating immunopathology at the expense of immunity.


Introduction
Chronic periodontitis (CP) is a persistent inflammatory condition of the periodontal tissues resulting in destruction of the periodontium which, if left untreated, could result in tooth loss. CP results as a consequence of the host inflammatory response to persistent microbial challenge represented by a dysbiotic biofilm in which Porphyromonas gingivalis (PG) is an important member [1][2][3]. PG is an intracellular oral mucosal pathogen which evades recognition and uptake by neutrophils, infecting oral epithelial cells, fibroblasts and underlying dendritic cells and macrophages (MWs) [4][5][6]. Clearance of such intracellular pathogens would necessitate cell mediated immunity, involving Th 1 subset cells. Porphyromonas gingivalis LPS (PG-LPS) however, predominantly induces Th 2 -mediated humoral responses to extracellular pathogens; hence immune-deviation towards a non-clearing response is integral to pathogen persistence [7]. PG-LPS also possesses low endotoxin activity and targets TLR2, at the expense of the traditional LPS receptor, TLR4, although P. gingivalis strains exhibit differential structural LPS formats to and, as a consequence, differential utilisation of both TLR2 and TLR4 [8]. Thus, PG subverts both adaptive and innate immune function to survive in oral mucosal tissue.
Immune subversion can be achieved by both immunomodulatory and immunosuppressive mechanisms. PG-LPS is able to induce endotoxin tolerance (ET) in MWs; ET was first characterised by LPS pre-exposure rendering innate immune cells refractory to subsequent endotoxin challenge, reviewed in [9]. ET would appear to be both beneficial and harmful to host and pathogen alike; suppressing harmful over-exuberant tissue-destructive pro-inflammatory responses, manifestation of sepsis, autoimmunity and cancer in the host [10], whereas, simultaneously, suppresses protective inflammatory responses mounted against the oral pathogen. Oral mucosal MWs are important to ET; their differentiation and activation status determining whether the mucosal environment is beneficial to the host tissue or pathogen. PG modulates host cell function in order to facilitate its own survival [11,12]. Upon LPS recognition, this pathogen induces an inflammatory response modulated by a wide range of inflammatory molecules. Of interest however, is that PG only weakly induces inflammatory cytokines, favouring an insufficient clearing response, bacterial proliferation and persistence. The cytokine production in response to this expanded bacterial number contributes to localised tissue destruction characteristic of chronic periodontitis [13][14][15].
MWs express both TLR2 and TLR4; responses to their respective PAMPs, lipopeptides and LPS are optimised by association with the co-receptor molecule, CD14, driving potent inflammatory responses characterised by high levels of the proinflammatory cytokines, TNFa, IL-1b, IL-6, IL-8. Indeed, CD14 gene polymorphisms are associated with inflammatory periodontal disease, where CD14 hi expression is indicative of higher levels of inflammation [22]. CD14 expression is partially predictive of mucosal MW effector phenotype: CD14 lo MWs produce antiinflammatory/regulatory cytokines (TGFb and IL-10) and low levels of pro-inflammatory cytokines [23]. As such, mucosal MWs, existing in a non-pathogenic and homeostatic state, resemble the M2 MW phenotype. CD14 hi MWs, on the other hand, produce high levels of pro-inflammatory-and low levels of regulatorycytokines: resembling M1 MWs, readily activated by PAMPs which, if uncontrolled, drive chronic inflammatory pathology. Thus, mucosal MW effector phenotype (inflammatory vs regulatory) may be controlled by regulation of TLR and CD14 expression. Of significance to control of effector phenotype is the observation that gingipains, released from outer membrane vesicles of P.gingivalis, have been described to cleave CD14 from the membrane surface [24]. Such a mechanism can suppress MW inflammatory responses (LPS hypo-responsiveness) and represents another tolerogenic response associated with ET.
The relevance of ET in the pathology of CP is the subject of intense research efforts. ET may benefit both the host and pathogen; tolerance would normally be viewed as beneficial in the context of a destructive inflammatory pathology, whereas in the case of PG, ET may favour pathogen persistence. PG-LPS is predominantly recognised by TLR2, instead of TLR4. In CP, both TLR2 + and TLR4 + monocytes are recruited into the gingival lamina propria whereas, concurrently, in diseased human CP gingiva, mucosal tissue was generally tolerised where TLR2, TLR4, TLR5 and MD-2 expression was down-regulated. Functional studies substantiated these results, PG-LPS pretreatment of monocytes suppressed subsequent PG-LPS stimulation of both pro-inflammatory (TNFa, IL-1b, IL-6, IL-8) and antiinflammatory cytokines (IL-10) [25]. The aim of this study was thus two-fold: to investigate whether Porphyromonas gingivalis differentially modulates cytokine production in the pro-inflammatory M1-like MW subset in comparison to the anti-inflammatory/ regulatory M2-like subset and to expand on current understanding of P. gingivalis-induced endotoxin tolerance in the context of these functionally disparate MW subsets, relevant to mucosal MW effector function.

Bacteria and pathogen associated molecular patterns (PAMPs)
Bacterial products were obtained from Autogen Bioclear, Calne, UK. P. gingivalis strain ATCC 33277 was originally isolated from human gingival sulcus and obtained from the American Type Culture Collection. Due to the ability of P.gingivalis to induce inflammatory factors via membrane receptors and to invade mucosal cells by phagocytosis, the effects of Porphyromonas gingivalis lipopolysaccharide (PG-LPS) were compared to those obtained for whole bacterial cells, heat-killed Porphyromonas gingivalis, HKPG (in the absence of any secreted bacterial products). PG-LPS was extracted by successive enzymatic hydrolysis and purification by Phenol-TEA-DOC protocol, described in [27]. HKPG were prepared by heating a bacterial suspension of P.gingivalis to 120uC for 30 minutes followed by several washes in endotoxin-free water. Peptidoglycan (PGN) was purchased from Sigma-Aldrich, Poole, Dorset, UK.
Activation of monocyte and macrophage cytokine production THP-1, THP-1(CD14 lo ) and THP-1(CD14 hi )-derived M1-and M2-like MWs were stimulated by the bacterial pathogen associated molecular patterns (PAMPs); 100 ng/ml PG-LPS, 1610 7 cells/ml HKPG and 10 mg/ml of the TLR2-ligand, lipoteichoic acid (LTA) (Autogen Bioclear, Calne, UK) and cultured for 18 hours (determined as optimal PAMP concentration and time period for expression of all the pro-inflammatory cytokines TNFa, IL-1b and IL-6, data not shown). Supernatants were then harvested and either used immediately for colorimetric analysis of NFkB activity or alternatively, stored at 220uC until required for cytokine assay by sandwich ELISA.
Tolerisation by pre-incubation with Porphyromonas gingivalis PAMPs, LTA and PGN THP-1, THP-1(CD14 lo ) and THP-1(CD14 hi )-derived M1-and M2-like MWs were pre-treated for 24 hours with either 100 ng/ml PG-LPS, 1610 7 cells/ml HKPG, 10 mg/ml LTA or 10 mg/ml PGN (determined as the optimal concentration and time duration for tolerisation, data not shown) or R10 medium alone (tolerisation negative control). Pre-stimulus culture medium was carefully removed, after which MWs were washed in fresh R10 prior to stimulation for a further 18 hours at 37uC/5% CO 2 . MWs were either stimulated with 100 ng/ml PG-LPS, 1610 7 cells/ml HKPG, 10mg/ml LTA or R10 medium alone (stimulation negative control). After this stimulation period, supernatants were harvested and either used immediately for colorimetric analysis of NFkB activity or stored at 220uC until required for cytokine assay by sandwich ELISA. To demonstrate a physiologically-relevant tolerisation; after stimulation or tolerisation protocols, MW viability was routinely checked by either MTT assay or trypan blue exclusion. No significant reductions in viability were observed for PAMPs used in this study, viability was routinely .85%.

Cytokine measurement
Cytokines; TNFa, IL-1b, IL-6 and IL-10 were analysed by sandwich ELISA using capture and detection antibodies commer- cially available from R&D Systems UK Ltd., Abingdon and BD-Pharmingen, Oxford, UK. Protocols were followed according to manufacturer's instructions and compared to standard curves, between the range of 7 to 5,000 pg/ml, using the international standards available from NIBSC, Potter's Bar, UK. Colorimetric development was measured spectrophotometrically by an OPTI-Max tuneable microplate reader at 450 nm and analysed by Softmax Pro version 2.4.1 software (Molecular Devices Corp., Sunnyvale, CA, USA).

NFkB activity measurement
NFkB activity was measured using a colorimetric reporter gene assay for secreted embryonic alkaline phosphatase (SEAP) associated with the stably-transfected reporter gene cell lines, THP-1Blue (CD14 lo ) and THP-1Blue-CD14 (CD14 hi ). Briefly, at conclusion of the experiment, conditioned medium was harvested and incubated with Quantiblue colorimetric reagent (Autogen Bioclear, Calne, UK) for 30 minutes at 37uC/5% CO 2 . Colorimetric development was then measured spectrophotometrically by an OPTIMax tuneable microplate reader at 620 nm and analysed by Softmax Pro version 2.4.1 software. The resulting colour development was directly proportional to the reporter gene SEAP expression and hence NFkB activity.

Statistical analysis
Measure of statistical significance was analysed using a balanced analysis of variance (General Linear Model, Minitab version 16) followed by a multiple comparison test (LSD, least significant difference). Significance was set at p,0.05 (*p,0.05, **p,0.01 and ***p,0.001).

PG-LPS and HKPG induce separate pro-inflammatory cytokine profiles in M1 and M2 MWs
Upon stimulation M1 and M2 MW subsets produce different cytokine profiles; M1 MWs exhibit a predominantly pro-inflammatory cytokine profile whereas M2 MWs express a more antiinflammatory or regulatory profile. This experiment was undertaken to establish whether M1 and M2 MWs responded similarly to challenge with the oral pathogen, P.gingivalis. Indeed, PG induced distinct cytokine profiles in M1 and M2 MWs. Stimulation of these MW subsets was comparable, however, when stimulated by either HKPG or PG-LPS: PG-LPS induced M1 expression of the pro-inflammatory cytokines TNFa, IL-1b and IL-6 at a ratio of 99:2:1. On the other hand, PG-LPS induced a TNFa: IL-1b: IL-6 ratio in M2 MWs of 4:2:1, where the cytokine expression between these two MW subsets was significant to p = 0.0098 for TNFa, p = 0.046 for IL-1b and p = 0.062 for IL-6 ( Figure 1a, b & c). A similar cytokine profile was observed when M1 and M2 MWs were stimulated by HKPG. HKPG induced an M1 expression profile of 249:8:1 and 10:12:1 in M2 MWs, where the cytokine expression between these two MW subsets was significant to p = 0.0008 for TNFa, p = 0.044 for IL-1b and p = 0.033 for IL-6 ( Figure 1d, e & f). As a consequence of heterogeneity of CD14 expression on M1-and M2-like macrophages, IL-10 secretion was not routinely detectable above the lower limit of detection of the IL-10 ELISA, and as such was not presented in this figure. IL-10 secretion however, was detectable when examining a more homogenous CD14 hi and CD14 lo macrophage population (refer to later stable transfectant data figures and tables).
In addition, these THP-1 derived macrophage subsets both display a differential response towards the enteropathic E. coli K12 LPS and the oral pathogenic P. gingivalis LPS. In agreement with other studies [8], PG-LPS exhibits low endotoxin activity when compared with the same concentration of K12-LPS. In the case of M1 and M2 MWs, endotoxin activity was determined by the strength of induction of TNFa secretion; PG-LPS resulted in 13% TNFa induction in M1s compared to K12-LPS whereas in the case of M2 MWs, PG-LPS resulted in 25% TNFa induction. This may be consistent with PG-LPS utilisation of TLR2, as TNFa induction was closer in amplitude to that of the TLR2 agonist, LTA (see table 1).

P. gingivalis differentially suppresses M1 and M2 MW proinflammatory cytokines
Macrophage challenge with Porphyromonas gingivalis (PG-LPS and HKPG) differentially suppresses MW subset cytokine production upon stimulation with the same pre-treatment challenges. Pretreatment of M1 pro-inflammatory MWs fails to suppress TNFa, IL-1b and IL-6 when later challenged by PG-LPS and HKPG (see figure 2a, b & c). M2-like MWs, on the other hand, were sensitive to tolerance induction. PG-LPS pre-treatment strongly suppressed M2 production of TNFa, upon stimulation with either PG-LPS (reduced by 94%, p = 0.0383) or HKPG (reduced by 66%, p = 0.0032) (See fig. 2d). Pre-treatment with HKPG partially suppressed TNFa production stimulated by HKPG (reduced by 9%, p = 0.258) but clearly suppressed PG-LPS induced TNFa (reduced by 92%, p = 0.0433) (see fig. 2d). In addition to PG-LPS tolerising TNFa production to PG-LPS stimulation and HKPG tolerising HKPG stimulation, these data also demonstrate a level of cross-tolerisation between HKPG and PG-LPS with respect to TNFa production by M2 MWs. M2 production of IL-1b, however, failed to show any significant suppression in response to both pretreatment and stimulation by either HKPG or PG-LPS ( fig. 2e). IL-6 production, on the other hand, was partially suppressed, dependent on pre-stimulation and challenge stimulus. Pretreatment with HKPG partially suppressed IL-6 production stimulated by HKPG (reduced by 57%, p = 0.0067) but clearly suppressed PG-LPS induced IL-6 (reduced by 79%, p = 0.0078) (see fig. 2f). Pre-treatment with PG-LPS failed to suppress IL-6 production stimulated by HKPG, but clearly suppressed PG-LPS induced IL-6 (by 48%, p = 0.0013) ( fig. 2f).

PG-LPS and HKPG induction of pro-inflammatory cytokine profiles in M1 and M2 MWs is CD14-dependent
In the homeostatic, regulatory mucosal environment, mucosal MWs exhibit an M2-like phenotype characterised by a regulatory cytokine profile and the absence of surface markers such as CD14 and CD89. The inflammatory environment results in recruitment  PG-LPS and HKPG induce differential NFkB activation amplitudes in CD14 hi /CD14 lo M1 and M2 MWs MW production of the pro-inflammatory cytokines TNFa, IL-1b and IL-6 has been described to be dependent on the transcription factor, NFkB. The previous section demonstrated the ability of PG-LPS and HKPG to induce these cytokines in a subset-specific manner; considering NFkB -dependence of these cytokines, it was essential to investigate whether P. gingivalis also induced activation of this signalling component. Indeed, M1 and M2 MW activation of NFkB was found to be determined by both differentiation and CD14 expression. In line with the cytokine expression data previously, CD14 lo and CD14 hi MWs demonstrated differential NFkB activity responses when stimulated by HKPG and PG-LPS. In the case of the pro-inflammatory M1-like MWs, M1 CD14 lo expressed lower NFkB activation than M1 CD14 hi MWs (lower than CD14 hi by 75% and 62% for HKPG (p = 0.0013) and PG-LPS (p = 0.0033), respectively) ( Figure 4a). The opposite trend is observed for M2-like MWs: M2 CD14 lo expressed higher NFkB activation than M2 CD14 hi MWs (higher than CD14 hi by 117% and 96% for HKPG (p = 0.014) and PG-LPS (p = 0.0053), respectively) ( Figure 4b). This differential profile of NFkB activation, parallels that observed for P. gingivalis induction of TNFa by all of the CD14 hi/lo M1/M2 MW subsets, suggestive of a direct link between NFkB activation and MW production of these pro-inflammatory cytokines. P. gingivalis differentially suppresses CD14 hi/lo M1 and M2 MW NFkB activity, TNFa and the anti-inflammatory cytokine, IL-10 Previous data in this manuscript have demonstrated that PG-LPS and HKPG activation of NFkB and induction of the NFkBdependent pro-inflammatory cytokines are differentially regulated in M1 and M2 MW subsets and amplitudes dependent on CD14 expression. Preliminary investigation of PG-induced tolerance/ suppression demonstrated that M2 MWs were sensitive to suppression whereas M1 MWs were refractory. These THP-1derived MW subsets are heterogenous with respect to their CD14 expression; mucosal MWs however, demonstrate distinct CD14 profiles where tolerogenic/homeostatic mucosal MWs are CD14 lo and are analogous to an M2 phenotype whereas inflammatory invasive MWs are CD14 hi and resemble the pro-inflammatory M1 subset [23,28,29]. As a consequence of this, the ability of PG to induce tolerance/suppression in both CD14 lo and CD14 hi MW subsets, in the context of pro-inflammatory TNFa production and NFkB activation, was investigated.
M2 MWs were observed to be sensitive to tolerisation and crosstolerisation by both PG-LPS and HKPG with respect to NFkB activation. CD14 lo and CD14 hi M2 MW NFkB activation were totally suppressed to unstimulated control levels, upon pretreatment with these PG PAMPs (figure 5c and 5d). The proinflammatory MW subset however, was differentially sensitive to tolerance induction by PG. The CD14 hi M1 phenotype of MW, (representative of invasive, recruited pro-inflammatory MWs) was refractory to tolerance induction by both PG-LPS and HKPG (figure 5b) whereas CD14 lo M1 MWs were sensitive to pretreatment suppression. PG-LPS stimulation control levels of NFkB activation were suppressed by 60% and 48% upon pre-treatment with PG-LPS and HKPG, respectively, whereas HKPG stimulation control was suppressed by 66% and 78%, respectively (refer to figure 5a).
The induction of TNFa production by these MW subsets displayed the same tolerance sensitivity profile as presented with NFkB activation (refer to figure 6). CD14 lo and CD14 hi M1 & M2 MWs exhibited different sensitivities to PG PAMP tolerisation and cross-tolerisation. In general, PG-LPS and HKPG-stimulation of TNFa production was suppressed upon pre-treatment with both the same PAMP (PG-LPS pre-treatment followed by PG-LPS stimulation and HKPG pre-treatment followed by HKPG stimulation) and the alternative PAMP (PG-LPS pre-treat, HKPG stimulus and HKPG pre-treat, PG-LPS stimulus). This suppression or tolerisation was clearly evident in both CD14 hi and CD14 lo M2 MWs (figure 6c, 6d) and less so in the case of CD14 lo M1 MWs where PG-LPS stimulation control levels were suppressed by 72% and 74% upon pre-treatment with PG-LPS and HKPG, respectively, whereas HKPG stimulation control was suppressed by 66% and 75%, respectively (figure 6a). In contrast, HKPG and PG-LPS failed to suppress HKPG and PG-LPS-stimulated TNFa production by the pro-inflammatory CD14 hi M1 MW subset (figure 6b).
In addition, the anti-inflammatory cytokine, IL-10, also demonstrated a distinct tolerisation profile in response to HKPG and PG-LPS. Both of these P. gingivalis products exhibited both homo-and hetero-tolerisation of IL-10 secretion. Suppression of IL-10 was clearly demonstrated for both CD14 lo and CD14 hi M2 MWs ( figure 7c and d) where, irrespective of pre-treatment and stimulus combination, P. gingivalis suppressed CD14 lo M2 IL-10 by 70 to 85% and CD14 hi M2 MWs by 34 to 78%. Interestingly, this pattern of tolerisation was extended to the pro-inflammatory CD14 hi M1 MWs (figure 7b), where PG-LPS and HKPG suppressed IL-10 production between 68 to 76%, and less so the degree of suppression in the CD14 lo M1subset 23 to 44% suppression of P. gingivalis stimulus ( figure 7a).
This tolerisation-sensitivity profile of these distinct MW subsets was reproduced when investigating other NFkB -dependent proinflammatory cytokines such as IL-1b and IL-6. Table 3 highlights the ability of PG-LPS and HKPG as well as the TLR2 PAMP, LTA, to tolerise and cross-tolerise these pro-inflammatory cytokines (TNFa, IL-1b and IL-6). What is evident from this table is that PG-LPS, HKPG and LTA-induced cytokines are sensitive to suppression by pre-treatment with PG-LPS, HKPG and LTA: as with figures 5 and 6, the CD14 hi M1 MW subset was found to be refractory to tolerance induction when compared to the other subsets and that there was a preferential cytokine sensitivity to suppression where, in general, TNFa was the most sensitive and IL-1b the least sensitive to suppression (refer to table 3).

Peptidoglycan differentially cross-tolerises P. gingivalisstimulated macrophage subsets
Cross-tolerisation has been described between different microbial species, their PAMPs and the corresponding PRRs, which may have a role to play in the inflammatory process of CP, which, in addition to P. gingivalis, is generally driven by a collection of oral pathogens. In addition to the suggestion of cross-tolerisation exhibited between HKPG, PG-LPS and LTA and the differing suppression observed between these PAMPs in the previous table, it was desirable to investigate this process with respect to the bacterial cell wall PAMP, peptidoglycan (PGN). In contrast to PG-LPS, HKPG and LTA tolerisation, PGN exhibits a different pattern of macrophage tolerisation. PGN tolerisation, in general, resulted in a higher level of suppression of the pro-inflammatory cytokines (TNFa, IL-1b, IL-6) and the anti-inflammatory cytokine, IL-10, in all the CD14 hi/lo M1/M2 subsets when compared to P.gingivalis and LTA tolerisation. The most striking result however, was the observation that CD14 lo M1 MWs were refractory (3% and 0% suppression) to tolerisation of IL-6 response, whereas CD14 hi M1 and M2 MWs exhibited a high level of suppression.
Finally, PGN-induced suppression of NFkB activity was weakest in both CD14 hi M1 and M2 macrophages (refer to table 4).

Discussion
This investigation has resulted in several conclusions being drawn with respect to MW responses to the oral pathogen, Porphyromonas gingivalis. Firstly, the PAMP-induced profile of proinflammatory cytokine production is dependent on both the route Figure 3. PG-LPS and HKPG induction of M1 and M2 MW pro-inflammatory cytokines are CD14-dependent. THP-1-derived CD14-highand CD14-low-expressing (CD14 hi and CD14 lo ) M1 and M2 MWs were generated by differentiating CD14 + and CD14 2 stable transfectant THP-1-blue monocytes with either 25 ng/ml phorbol 12-myristate 13-acetate (PMA) for 3 days or 10 nM 1,25-(OH) 2 vitamin D 3 for 7 days, respectively. CD14 hi / CD14 lo M1 (bold) and M2 (shaded) MW subsets were stimulated with either 100ng/ml PG-LPS (a, b, c & d) or 1610 7 cells/ml HKPG (e, f, g & h). Cytokine production is expressed as the mean 6 SD in pg/ml for TNFa (a & e), IL-1b (b & f), IL-6 (c & g) and IL-10 (d & h). Data displayed represents triplicate samples for n = 3 replicate experiments. Significant differences in cytokine production between activated CD14 hi and CD14 lo MWs are indicated as *p,0.05, **p,0.01, ***P,0.001 and ns, not significant. doi:10.1371/journal.pone.0067955.g003  of MW differentiation and the level of expression of the coreceptor, CD14. In general, M1-like MWs were characterised as TNFa hi , IL-1b lo , IL-6 lo whereas M2-like MWs were TNFa lo , IL-1b hi and IL-6 hi . With respect to the induction of the antiinflammatory cytokine, IL-10, CD14 hi/lo M1 macrophages exhibited low-level expression of IL-10 whereas higher expression was restricted to the CD14 lo M2 subset. Secondly, these MWs displayed differential sensitivities to tolerance induction by both P. gingivalis-derived bacterial PAMPs and the TLR2 ligand, LTA ie. direct homo-and cross-/hetero-tolerance. P. gingivalis induced suppression of inflammatory cytokines in the CD14 lo/hi M2-and CD14 lo M1-like subsets, whereas, the pro-inflammatory CD14 hi M1-like subset was refractory to tolerance induction. Finally, this MW pro-inflammatory cytokine tolerisation profile appeared to be linked to sensitivity to suppression of the pro-inflammatory transcription factor, NFkB. Irrespective of stimulation, the M1 and M2 subsets displayed differing cytokine effector profiles: M1 MWs exhibited a proinflammatory phenotype (TNFa hi , IL-1b lo , IL-6 lo , IL-10 lo ) whereas M2 MWs were less inflammatory and tending to antiinflammatory/regulatory when compared to M1s (TNFa lo , IL-1b hi , IL-6 hi , IL-10 + ). In line with characteristic mucosal MW phenotypes, CD14 expression determined M1 and M2 cytokine amplitudes and NFkB activation resulting from P. gingivalis stimulation. CD14 hi M1 MWs (representative of recruited, pro-inflammatory pathological MWs) was described as TNFa hi , NFkB hi whereas the CD14 lo M1subset was TNFa lo , NFkB lo. On the other hand, CD14 lo M2s (representative of regulatory, antiinflammatory mucosal MWs) were TNFa lo , NFkB med and CD14 hi M2s were TNFa lo , NFkB lo . Contrary to our understanding of proinflammatory and anti-inflammatory MW subsets, M2-like MWs produce higher levels of both IL-6 and IL-1b in response to PG-LPS and HKPG. These two cytokines, although thought of as proinflammatory, exhibit clear anti-inflammatory properties. IL-6 exerts its anti-inflammatory effects through induction of SOCS proteins and STAT-3 activation [30] and reviewed in [31]. Indeed, SOCS-3 is associated with M1 classical MW polarisation and is suppressive to anti-inflammatory signal and expression of IL-6 and IL-10. Conversely, SOCS-3 expression knockdown favours M2 polarisation [32]. Thus, the reciprocal relationship between SOCS-3 and STAT-3 would appear to regulate pro-or anti-inflammatory effect of IL-6 and the polarisation of MWs between M1 and M2 effector subsets. IL-1b, on the other hand, may mediate anti-inflammatory responses via its ability to induce IL-10 expression [33]; indeed, results from this study are suggestive of a positive correlation between IL-1b and IL-10, as these cytokines are produced strongest by the CD14 lo antiinflammatory/regulatory M2 macrophages. In addition, IL-1b secretion has been demonstrated to be negatively associated with the pro-inflammatory IKKb-dependent NFkB pathway [34]; suggestive of a non-pro-inflammatory role for IL-1b and the complex wiring of the NFkB pathway in determining cell effector phenotype. Modulation of effector phenotype would thus play an important role in determining whether responses initiated in the oral mucosa are pro-inflammatory, destructive or anti-inflammatory, tolerogenic. Specific modulation of such subsets would directly affect pathogenic mechanisms associated with pathogens infecting the oral mucosa.
Mucosal MWs are considered to exist in discrete functional subsets, governed by the environment that exists in the mucosal tissue itself. In homeostatic conditions, mucosal MWs fail to express CD14 and express a functional phenotype resembling the regulatory, anti-inflammatory M2 subset [23,28,29]. Upon mucosal dysfunction, barrier breakdown and inflammatory pathological conditions, these tolerogenic MWs change their effector phenotype to a predominantly pro-inflammatory M1-like subset. Manipulation of MW effector phenotype via controlling monocyte/MW infiltration into the mucosa, plasticity between M1 and M2 subsets, or indeed specific MW subset tolerance induction would be of great benefit for future therapeutic management of such inflammatory pathologies as chronic periodontitis. In the context of mucosal MWs, whether CD14 expression is integral to tolerance induction or is just reflective of a tolerisable sensitive subset is not proven. CD14 is known to be co-expressed with both TLR2 and TLR4, both of which can be utilised by P.gingivalis. PG-LPS is generally recognised as transducing its signal through TLR2. Data presented in this study suggested that P. gingivalis and PAMPs derived from other microbes which signal through different PRRs, induce cross-tolerance, whereby peptidoglycan (which signals through NOD2) differentially tolerised both PG-LPS and HKPG (TLR2)-induced macrophage cytokines. In line with other published studies, it is probable that PAMPs such as E. coli-K12 LPS (gram-negative bacterial PAMP signalling through TLR4) are able to differentially suppress M1 and M2 responses to gram-positive bacteria and signals transduced through both TLR2 (homo-tolerance) and non-TLR2 (hetero-or cross-tolerance) PRRs such as TLR4 and NOD2 [35][36][37].
The fact that CD14 hi M1 pro-inflammatory MWs were refractory to tolerance-induction by HKPG and PG-LPS suggested that MW tolerance sensitivity was, in part, dependent on CD14. This was observed for both pro-inflammatory cytokine production (TNFa, IL-1b and IL-6) and NFkB, whereas IL-10 was suppressed and CD14 hi M2 (d) MW subsets were pre-stimulated (tolerised) with either 100 ng/ml PG-LPS (unshaded) or 1610 7 cells/ml HKPG (shaded) for 24 hours prior to stimulation with PG-LPS or HKPG and incubated for a further 18 hours (untolerised controls indicated in bold). Pro-inflammatory TNFa cytokine production is expressed in pg/ml as the mean 6 SD for the CD14 hi/lo M1 and M2 MW subsets. Data displayed represents triplicate samples for n = 3 replicate experiments. Significant effects compared to the un-tolerised stimulus control (bold) for each MW subset are indicated as ***p,0.001 and ns, not significant. doi:10.1371/journal.pone.0067955.g006 in these macrophages; suggesting that tolerance-induction was only partially dependent on NFkB activity and CD14 expression. Several mechanisms have been described which are involved in ET of NFkB -dependent readouts. These include the upregulation of endogenous suppressors of NFkB activation such as SIGIRR, ST2, A20, Myd88 s and IRAK-M [9,38]. NFkB is also important with respect to MW subset polarisation; IkBa overexpression resulted in M2 polarisation [19], whereas IKKb deletion favoured M1 MWs [20]. Thus, M1 subset polarisation is dependent on the classical p65/p50 NFkB heterodimer and M2 polarisation was found to be dependent on the alternative NFkB p50/p50 homodimer [21]. Manipulation of such classical and alternative NFkB pathways is likely to have a dramatic influence on MW plasticity, hence determining immune response as either pro-inflammatory/immune activatory or anti-inflammatory/tolerogenic.
Stimulation and pre-stimulation protocols investigate tolerisation by PG-LPS and HKPG, allowing the study of signals via PRRs but does not consider soluble/secreted immunomodulatory components produced by live bacteria. P.gingivalis secretes gingipains which are involved in endotoxin tolerance by cleavage of CD14 from the cell surface, leading to LPS hypo-responsiveness [24], either through CD14 absence from the LPS-binding receptor complex or through secreted CD14 competing for the LPS/LBP complex, hence antagonising the LPS-TLR signal, reviewed in [9]. It is probable that these gingipains may also induce ET through the shedding of PRRs such as TLR4, TLR2 and TLR5. In addition, this induction of ET may also be mediated via a protease-mediated shedding of both membrane-bound TNFa and its receptor, TNF-Rp75 [39]; hence suppression of TNFamediated inflammatory responses.
Chronic periodontitis is not just driven by P. gingivalis alone. To appreciate all the underlying pathological mechanisms, the complex interactions between the host immune factors and the microbial ecosystem of oral commensal and pathogenic bacteria requires investigation. Indeed, CP is characterised by bacterial plaque formation; it is likely these complex bacterial biofilms play a significant role in protecting pathogens from host immune responses either as a consequence of inaccessibility to damaging immune responses or through the regulation/deviation of these defences. One such intriguing pathogen response to host immunity by PG biofilms was found to be via the degredation of both pro- Figure 7. P. gingivalis differentially suppresses CD14 hi/lo M1 and M2 MW IL-10 production. CD14 lo M1 (a) CD14 hi M1 (b), CD14 lo M2 (c) and CD14 hi M2 (d) MW subsets were pre-stimulated (tolerised) with either 100 ng/ml PG-LPS (unshaded) or 1610 7 cells/ml HKPG (shaded) for 24 hours prior to stimulation with PG-LPS or HKPG and incubated for a further 18 hours (untolerised controls indicated in bold). Anti-inflammatory IL-10 cytokine production is expressed in pg/ml as the mean 6 SD for the CD14 hi/lo M1 and M2 MW subsets. Data displayed represents triplicate samples for n = 3 replicate experiments. Significant effects compared to the un-tolerised stimulus controls (bold) for each MW subset are are indicated as*p,0.05, **p,0.01, ***p,0.001 and ns, not significant. doi:10.1371/journal.pone.0067955.g007 inflammatory (IL-1b, IL-6) and anti-inflammatory (IL-1Ra) cytokines [40]. Such an immuno-suppressive mechanism was again indicative of microbial protease activity.
The significance of these data in the context of CP is difficult to interpret. In general, any mechanism, which induces tolerance is likely to be beneficial to chronic pathologies that result from overexuberant immune responses. These data clearly demonstrate a role for PG in tolerance induction of M2 and CD14 lo M1 MWinflammatory mediators, whereas no suppression was observed with inflammatory CD14 hi M1 MWs. This suggested some beneficial effect to the oral pathogen by failing to suppress the pro-inflammatory macrophage. Indeed, early P. gingivalis infection events were found to be anti-inflammatory or tolerant, enabling the pathogen to expand its numbers. This population expansion leads to an increase in inflammatory mechanisms, resulting in tissue destruction, lesions and a reduction in bacterial numbers. As a consequence of this cycling in pathogen numbers; it is likely that this relapsing/remitting chronic inflammatory disease is characterised by immunopathological mechanisms constantly switching between inflammation and tolerance/regulation. Maintenance of this chronic cycling between ET and destructive inflammation over a long period is detrimental to the host; long-term ET rendering the host more susceptible to infection (immunocompromised) and long-term inflammatory responses resulting in host tissue destruction without pathogen clearance.
In conclusion, this investigation has further characterised M1and M2-like MW subsets with respect to pro-inflammatory cytokine profile upon stimulation with P. gingivalis PAMPs. It demonstrates a dichotomy in cytokine secretion where M1 MWs are indeed the predominant pro-inflammatory cell. This effector response was further elucidated in the context of subsets relevant to mucosal MWs where, in response to P. gingivalis, the CD14 lo M2 subset, representative of regulatory, anti-inflammatory cells, was indeed a low-level producer of TNFa and IL-10 + whereas CD14 hi M1 MWs, representative of infiltrated pro-inflammatory pathological cells, were predominantly pro-inflammatory and strongly produced TNFa. This cytokine profile is likely to be as a consequence of NFkB activation, as NFkB activation profile for these MW subsets, closely paralleled the cytokine response. In addition, upon investigation of sensitivity of these subsets to tolerisation, it was observed that the subset least sensitive to P. gingivalis-induced suppression of pro-inflammatory cytokines and NFkB activation was the inflammatory pathology-related subset, CD14 hi M1. This would suggest that such mechanisms of ET may be beneficial for survival and immunopathological mechanisms driven by the pathogen. To conclude, any future manipulation of MW subset suppression can only realistically be employed upon a full understanding of the immunopathological mechanisms behind such relapsing/remitting diseases as CP and by considering; who is tolerance induction of benefit to….host or pathogen?
Author Contributions