Porphyromonas gingivalis Evasion of Autophagy and Intracellular Killing by Human Myeloid Dendritic Cells Involves DC-SIGN-TLR2 Crosstalk

Signaling via pattern recognition receptors (PRRs) expressed on professional antigen presenting cells, such as dendritic cells (DCs), is crucial to the fate of engulfed microbes. Among the many PRRs expressed by DCs are Toll-like receptors (TLRs) and C-type lectins such as DC-SIGN. DC-SIGN is targeted by several major human pathogens for immune-evasion, although its role in intracellular routing of pathogens to autophagosomes is poorly understood. Here we examined the role of DC-SIGN and TLRs in evasion of autophagy and survival of Porphyromonas gingivalis in human monocyte-derived DCs (MoDCs). We employed a panel of P. gingivalis isogenic fimbriae deficient strains with defined defects in Mfa-1 fimbriae, a DC-SIGN ligand, and FimA fimbriae, a TLR2 agonist. Our results show that DC-SIGN dependent uptake of Mfa1+P. gingivalis strains by MoDCs resulted in lower intracellular killing and higher intracellular content of P. gingivalis. Moreover, Mfa1+P. gingivalis was mostly contained within single membrane vesicles, where it survived intracellularly. Survival was decreased by activation of TLR2 and/or autophagy. Mfa1+P. gingivalis strain did not induce significant levels of Rab5, LC3-II, and LAMP1. In contrast, P. gingivalis uptake through a DC-SIGN independent manner was associated with early endosomal routing through Rab5, increased LC3-II and LAMP-1, as well as the formation of double membrane intracellular phagophores, a characteristic feature of autophagy. These results suggest that selective engagement of DC-SIGN by Mfa-1+P. gingivalis promotes evasion of antibacterial autophagy and lysosome fusion, resulting in intracellular persistence in myeloid DCs; however TLR2 activation can overcome autophagy evasion and pathogen persistence in DCs.


Introduction
Antimicrobial autophagy or xenophagy plays an important role in controlling bacterial infection and promoting innate immunity. Recent evidence has revealed critical roles for autophagy in the ability of immune cells to recognize and selectively target microbes for elimination. [1][2][3][4] [1]. Dendritic cells (DCs) are innate immune cells that serve as a bridge to the adaptive immune response. DCs capture a wide variety of microbes in the peripheral tissues for which they are equipped with broad spectrum of pattern recognition receptors (PRRs). The major classes of PRRs expressed by DCs include Toll-like receptors (TLRs), NOD-like family receptors, CARD helicases and C-type lectin receptors [5,6]. Many of the PRRs come equipped with unique phagocytic machinery evolved for efficient antigen processing and presentation [7][8][9]. Phagocytosis relies on a network of endocytic vesicles such as early endosomes and/or autophagosomes, which fuse with lysosomes for degradation [10]. Intracellular vesicle maturation does not necessarily proceed through similar steps in different phagocytic cells [11]. Moreover, different phagosomal maturation pathways have been reported in the same cell type [11,12]. These different pathways are dictated primarily by the initial recognition step by PRRs and by the cargo contained in the vesicle [13]. Hence the immune cell type, the PRRs engaged and the properties of the microbe seem to be crucial for microbial clearance by autophagy.
DC-SIGN (DC specific ICAM-3 grabbing non-integrin) is a C-type lectin receptor involved in pathogen uptake, signaling and antigen presentation in DCs [14][15][16]. For uptake DC-SIGN contains internalizing motifs in its cytoplasmic tail [17]. Interestingly, DC-SIGN has been implicated in immune suppression and regulation in certain contexts [17][18][19]. Most notably DC-SIGN is targeted for immune escape by several pathogens such as HIV, hepatitis C virus, herpesvirus 8 (HHV-8), Mycobacterium tuberculosis, Helicobacter pylori and Streptococcus pneumonia [17,[20][21][22]. Recently, we reported that DC-SIGN engagement by the minor fimbriae (Mfa1) of Porphyromonas gingivalis yields weak DC maturation and an immunosuppressive cytokine profile. In the absence of Mfa1, P. gingivalis yields a very different DC response with high levels of IL-23 and IL-6 as well as induction of a Th1/Th17 type response [14,23]. Furthermore, this study demonstrated that the anaerobe P. gingivalis survives within DCs in an aerobic atmosphere, while it dies rapidly in the absence of DCs [14]. Early study of the relationship of fimbrial strain differences to alveolar bone loss, showed that this Mfa1 + Pg strain (DPG3) induced higher bone loss than Pg381 strain in a periodontitis mouse model [24]. The destruction induced by Mfa1 + Pg was similar to wild type strain P. gingivalis ATCC 53977 that has been reported to be invasive in the abscess model [25].
P. gingivalis expresses a number of virulence factors that bind to and signal through PRRs. The adhesion proteins, known as fimbriae, on P. gingivalis signal through PRRs, and facilitate invasion of host cells. P. gingivalis expresses both minor (MFa1) and major fimbriae (FimA) which are highly regulated depending on growth conditions [26,27]. We have previously shown that expression of Mfa1 is involved in targeting DC-SIGN while other studies have shown expression of FimA targets a non-DC-SIGN route, mostly through TLR2 [28,29]. The engagement of DC-SIGN and TLRs activates distinct signaling pathways [6,30] and we propose that differential signaling through these distinct PRRs results in differential intracellular routing and processing of P. gingivalis within DCs. TLRs are essential for phagosome maturation and subsequent bacterial clearance [31,32]. TLR signaling is also involved in the maturation of autophagosomes [33]. The ability of P. gingivalis to manipulate DC-SIGN and TLR signaling through differential fimbrial expression [26,34], could have profound effects on bacterial survival [26]. However, the role of P. gingivalis major and minor fimbriae in DC-SIGN-TLR2 crosstalk and its influence on survival of P. gingivalis within DCs has not been examined.
In the present study, a combination of approaches was used to address the role of DC-SIGN and TLRs in intracellular routing and survival of P. gingivalis, including blocking PRRs and autophagy, siRNA gene silencing and activation of autophagy in monocyte derived DCs (MoDCs), To address the role of fimbriae in this regard we utilized defined bacterial mutants, that solely express minor fimbriae (Mfa1 + Pg), major fimbriae (FimA + Pg) or are deficient in both fimbriae (MFB) [35] (Table 1). Our results indicate that engagement of DC-SIGN by MFA-1 allows P. gingivalis to evade autophagy and lysosome fusion, resulting in pathogen persistence and survival within DCs. In contrast, activation of autophagy or of TLR2 by P. gingivalis expressing FimA results in autophagy mediated killing of this pathogen within DCs. Collectively, our studies reveal a novel mechanism that enables this pathogen to evade host

Results
High intracellular content of Mfa1 + Pg within human MoDCs At 2, 12 and 24h after bacterial co-culture with MoDCs, the MoDCs were imaged for intracellular P. gingivalis by epifluorescence microscopy and transmission electron microscopy (TEM). All P. gingivalis strains except the double fimbriae negative P. gingivalis strain MFB (Table 1) were taken up by MoDCs. There were marked differences in the P. gingivalis content of MoDCs at 2, 12 and 24 hours, particularly when comparing double fimbriae positive strain Pg381 to Mfa1 + Pg (Fig. 1A). We observed a higher number of Mfa1 + Pg within MoDCs (Fig. 1A) (S1 Fig.) This difference was most apparent after 24 hours, with large numbers of intra-and extra-cellular bacteria present. In contrast, MoDCs infected with Pg381 showed minimal bacterial content after 24 hours. Survival of intracellular bacteria was then assessed quantitatively by lysing MoDCs and growing bacteria in broth cultures and on anaerobic blood agar plates. Mfa1 + Pg was recovered at higher numbers from MoDCs lysates in broth and on blood agar compared to Pg381 Fig. 1B). No significant difference was detected in the growth or death patterns of all strains in the media under anaerobic conditions in the absence of DCs (Fig. 1C).

Infection of MoDCs with Mfa1 + Pg increases the expression of DC-SIGN in MoDCs
To determine whether expression of DC-SIGN [14] was altered by P. gingivalis infection, MoDCs were infected with all the strains at different multiplicities of infection (MOIs) and gene expression of DC-SIGN was quantified at 2, 6, 12 and 24 hours ( Fig. 2A) (S2 Fig.). At 12 hours, a distinct pattern of DC-SIGN expression was detected in MoDCs infected with Mfa1 + Pg compared to Pg381 and FimA + Pg. Infection with Mfa1 + Pg up-regulated DC-SIGN mRNA at 1 and 10 MOIs in a dose dependent manner (p<0.01) ( Fig. 2A) (Table 2). In contrast, we observed decreased expression of DC-SIGN when MoDCs were incubated with Pg381 (MOI-10) down-regulated DC-SIGN mRNA expression significantly (p<0.05) at 12 hours (-4.55 fold). Fold regulations were calculated relative to un-infected MoDCs ( Fig. 2A) (Table 2). We also examined DC-SIGN expression on MoDCs by immunoelectron microscopy (Fig. 2B) and flow cytometry (Fig. 2C). The results confirm a difference in DC-SIGN expression in MoDCs as a function of P. gingivalis strain. Mfa1 + Pg induced higher positive immunolabeling for DC-SIGN in MoDCs relative to Pg381 (Fig. 2B). These results correlated well with our initial results from the mRNA analysis. DC-SIGN was detected on the membrane but also in the cytoplasm of MoDCs infected with Mfa1 + Pg. The presence of cytoplasmic staining is consistent with previous evidence for Mfa1 + Pg localization to DC-SIGN positive intracellular compartments [14] and the possibility of receptor recycling to the cell membrane after the phagocytic process. The cells infected with Pg381 showed minimal staining for DC-SIGN at the cell membrane and no cytoplasmic staining was detected at any of the time points (Fig. 2B). Stably transfected DC-SIGN positive and negative Raji cells served as positive and negative controls respectively for DC-SIGN expression by immuno-electron microscopy (S3 Fig.).
We quantitatively assessed the increase of DC-SIGN in MoDCs infected with MFa1 + Pg by flow cytometry analysis (Fig. 2C) (S4 Fig.). We also monitored the expression of other C-type lectins and TLRs on the infected MoDCs by flow cytometry. Although P. gingivalis up-  the expression of TLR2 and CXCR4, we did not observe strain specific differences in expression of these receptors (S4 Fig.). Moreover, there were no changes in the expression of DCIR and mannose receptor (MMR) upon P. gingivalis infection. We did observe increased expression of Dectin receptors, but only on Mfa1 + Pg infected MoDCs (S4 Fig.).

Difference in early but not late endosomal routing of Mfa1 + Pg in MoDCs
To analyze the phagosomal machinery involved in uptake and routing of different P. gingivalis strains by MoDCs, we assessed levels of the GTPase Rab family of proteins at early (2 hours) through late (24hours) stages of infection. Furthermore, we monitored co localization of the P. gingivalis strains with Rab proteins at these time points. Although both Pg381 and Mfa1 + Pg were taken up at 2 hours, only Pg381 was associated with Rab5 at significantly higher levels as compared to Mfa1 + Pg (Table 3) (S5 Fig. A and B). Association of Pg381 with Rab5 within MoDCs was detected up to 12 hours. After this point, detectable bacteria and Rab5 signals significantly decreased at 24 hours. The Rab5 signal was weak at all time points in MoDCs infected with Mfa1 + Pg. In addition, Mfa1 + Pg were more apparent than within MoDCs at 24 hours (S5 Fig. A and B).
MoDCs generally showed weak staining for Rab7 following infection either with Pg381 or Mfa1 + Pg up to 24 hours (S5 Fig. C and D). Since Rab7 was not detected during intracellular processing of any of the P. gingivalis strains examined, we investigated whether anti-bacterial autophagy may be involved in killing of Pg381 but not Mfa1 + Pg.

Induction of autophagy by Rapamycin inhibits the survival of Mfa1 + Pg within MoDCs
To investigate the role of autophagy as a putative survival mechanism utilized by Mfa1 + Pg, the viability of P. gingivalis strains in MoDCs was monitored after induction of autophagy by the mTOR inhibitor, Rapamycin. Initial studies established that the viability of MoDCs and of P. gingivalis alone were not Rapamycin-sensitive at the concentrations used. Entry into MoDCs resulted in a significant increase in survival of Mfa1 + Pg at 24 hours (p<0.001) (Fig. 3A). Rapamycin treatment of Mfa1 + Pg-infected MoDCs (Mfa1 + Pg+Rapa) significantly decreased P. gingivalis survival by~48% (p <0.001) (Fig. 3A). Increased autophagy induction was confirmed by immuno-labeling of LC3-II in MoDCs treated with Rapamycin for 11 hours (1 hour after P. gingivalis infections). Rapamycin treatment increased the LC3-II signal in cells infected with all fimbriated strains as well as in un-infected (Cont.) ( Fig. 3B and C).
To determine the kinetics of survival of P. gingivalis within MoDCs, as well as the involvement of autophagy, intracellular bacteria were monitored after 6, 12, 24 and 48 hours of incubation with MoDCs with or without Rapamycin (Fig. 4). The levels of P. gingivalis Mfa1 + Pg within MoDCs (Mfa1 + Pg+DC) were the highest at all time points except at 6 hours when we observed similar levels to that observed with Pg381+DCs (Fig. 4A). Levels of Pg381 and FimA + Pg within MoDCs (Pg381+DCs and FimA + Pg+DCs) were nonetheless significantly higher than bacteria without MoDCs, until 12 hours, at which point we observed a significant decrease in survival of both strains at 24 and 48 hours. Moreover, the highest level of Mfa1 + Pg was observed at 24 hours, with the numbers of P. gingivalis increasing within MoDCs at 6, 12 and 24 hours (Fig. 4A). Activation of autophagy with Rapamycin significantly inhibited Mfa1 + Pg survival at 6, 12 and 24 hours within MoDCs (Mfa1 + Pg+DCs+Rapa) (Fig. 4B). However, the numbers of Mfa1+Pg detected within MoDCs treated with Rapamycin (Mfa1+Pg +DCs+Rapa) were still higher than bacteria alone (Mfa1 + Pg) at 12 and 24 hours (Fig. 4B). For  Pg381, rapamycin also significantly decreased intracellular survival, with no significant differences detected relative to bacteria alone except at 12 hours (Fig. 4C). FimA + Pg exhibited significant intracellular survival (FimA + Pg+DCs) only at 12 hours, which was significantly inhibited with rapamycin treatment (Fig. 4D). The analysis of the data using three-factor repeated measures ANOVA showed that both time and intracellular environment were significant factors (P<0.05) in P. gingivalis survival, with rapamycin significantly impairing P. gingivalis survival with MoDCs.
Low autophagosomal marker LC3-II in MoDCs following uptake of Mfa1 + Pg Due the marked difference in Rab5 induction in MoDCs infected with Pg381 and Mfa1 + Pg, and the important role of Rab5 in regulation of subsequent autophagy [36,37] we tracked the LC3-II signal in MoDCs infected with labeled P. gingivalis strains between 2 to 24 hours (S6 Fig. A and B). LC3-II is the active form of cytosolic LC3 that associates with the autophagosome until cargo degradation [38]. Pg381 infection resulted in significant increases of LC3-II  Fig. A and B). Hence, further analysis of the uptake of different P. gingivalis strains by MoDCs and the expression of LC3-II were carried out at the 12 hr time point ( Fig. 5A and B). Infection with Pg381 and FimA + Pg, but not Mfa1 + Pg increased LC3-II levels in MoDCs during the first 12 hours. (Fig. 5A) (S6 Fig. A and B). Although there was generally a low level of co-localization between all P. gingivalis strains with LC3-II, strains Pg381 and FimA + Pg showed higher Pearson's correlation than Mfa1 + Pg with LC3-II. Moreover, quantification of the LC3-II signals in infected MoDCs revealed significant increases in cells infected with Pg381 and FimA + Pg (P<0.001 for both strains compared to Mfa1 + Pg). Cells infected with Mfa1 + Pg, in contrast, showed decreased levels of LC3-II compared to un-infected cells (Cont.) (Fig. 5B). Uptake of Pg381 and all mutant strains except MFB was confirmed by intensity quantification of CFSE (Fig. 5C). The highest uptake was detected in cells infected with Mfa1 + Pg yet these cells had the lowest level of LC3-II signal ( Fig. 5B and 5C). Infection of MoDCs by Mfa1 + Pg has previously been shown to depend on engagement of DC-SIGN [14]. We confirmed this result by blocking DC-SIGN with HIV gp120, prior to infection ( Fig. 6A and 6B). Blocking DC-SIGN increased the LC3-II signal in MoDCs prior to addition of Mfa1 + Pg (Fig. 6B). Moreover, blocking DC-SIGN restored the basal expression of LC3-II in MoDCs (Fig. 6B). To confirm the HIV gp120 blocking experiments, we additionally knocked down DC-SIGN using siRNA. DC-SIGN knockdown inhibited uptake of Mfa1 + Pg but not Pg381 ( Fig. 6C and 6D) and restored LC3-II signals in MoDCs ( Fig. 6C and 6D). Furthermore, DC-SIGN knockdown significantly decreased survival of Mfa1 + Pg in MoDCs   6E). A scrambled sequence control did not inhibit uptake or effect LC3-II signal. To confirm the contribution of actin-mediated endocytic trafficking in LC3-II induction, MoDCs were treated with cytochalasin-D (CytD) prior to infection. CytD significantly inhibited intracellular localization of both Pg381 and Mfa1 + Pg by MoDCs and restored the LC3-II signals to the basal level in MoDCs (S7 Fig. C and D).
We further confirmed LC3-II conversion in MoDCs by Western blot analysis (Fig. 7). Pg381 and FimA + Pg increased LC3-II expression in MoDCs. In contrast, Mfa1 + Pg-infected MoDCs showed no significant difference in LC3-II compared to the uninfected control (Cont.) or MFB treated MoDCs ( Fig. 7A and 7B). To determine if increased LC3-II signal in MoDCs infected with Pg381 was indeed due to increased induction, as opposed to accumulation from lack of autophagosomal-lysosome fusion, the latter was inhibited by Bafilomycin in the flux test as reported [39]. Dose response of Bafilomycin was confirmed in MoDCs, with the highest LC3-II accumulation observed between 3-4nM (S7 Fig. A and B). Bafilomycin treatment further increased LC3-II by Pg381, indicating an increase in autophagy by Pg381 rather than a block in autophagosome-lysosomal fusion ( Fig. 7C and 7D).
II was detected in red-fluorescent dye and the bacterial strains were pre-labeled with green CFSE. E) The figure shows the CFU counts of the P. gingivalis strains with MoDCs that lack DC-SIGN (si-DC-SIGN). The analysis of readings used One-way ANOVA analysis of different groups and Tukey's test for multiple comparisons (* p<0.001). doi:10.1371/journal.ppat.1004647.g006

TLR signaling increases LC3-II and inhibits the survival of Mfa1 + Pg within human MoDCs
To establish the role of TLR signaling in autophagy induction leading to DC maturation and intracellular killing of P. gingivalis, we activated TLR pathways using specific agonists for TLR1/2 (Pam3csk4) and TLR4 (E. coli LPS). The results demonstrate that TLR1/2 activation was highly potent in stimulating CD83 expression and intracellular LC3-II within MoDCs infected with P. gingivalis (Fig. 8A and 8B). In addition, we observed a significant reduction in the Mfa1 + Pg counts within MoDCs treated with Pam3csk4 after 24 hours of infection (Fig. 8C).

DC-SIGN targeting Mfa1 + Pg routed preferentially to single membrane intracellular structures within MoDCs
The initial interaction of P. gingivalis with the outer membrane of MoDCs was visualized by scanning electron microscopy (SEM) after 2 hours of infection ( Fig. 9A and 9B). Both strains (Pg381 and Mfa1 + Pg) were able to engage the MoDC surface at 2 hours ( Fig. 9A and 9B). To directly visualize formation of double membrane autophagosomes, TEM analysis was performed in MoDCs infected with all fimbriated strains. Tracking of Pg381 and FimA + Pg within MoDCs after 12 hours of infections demonstrated that the majority of bacteria were contained in double membrane structures. In contrast, greater numbers of Mfa1 + Pg were consistently detected within single membrane vesicles in the cytoplasm of MoDCs (Fig. 9C). Quantification of the double membrane structures that contained bacteria in were carried out in three randomly selected EM grids of each sample. The ratio of Mfa1 + Pg trapped in double membrane relative to the total intracellular bacteria was significantly lower than Pg381 and FimA + Pg (Fig. 9D).

Low level of lysosomal marker (LAMP1) in MoDCs after DC-SIGN mediated uptake of P. gingivalis
Due to the important role of lysosome fusion in autophagosome maturation [40], CFSE-labeled P. gingivalis and LAMP1+ lysosomes were tracked in MoDCs by epifluorescence microscopy and P.gingivalis-LAMP1 co-localization was quantified between 2 and 24 hours. The uptake of all three fimbriated P. gingivalis strains by MoDCs at 2, 12 and 24 hours relative to uninfected controls was confirmed. (Fig. 10A and 10B) (S8 Fig.). At 24 hours, Pg381 was mostly undetectable compared to Mfa1 + Pg within MoDCs (S8 Fig.). Uptake of Pg381 was associated with increased LAMP1 at 2 hours, while minimal LAMP1 was detected in cells infected with Mfa1 + Pg at all time points (S8 Fig.). The highest LAMP1 signal in Pg381 and FimA + Pg infected MoDCs was observed at 12 hours, yet there was no increase in LAMP1 signal within Mfa1 + Pg infected MoDCs (Fig. 10A). Quantification of LAMP1 intensities within Pg381 and FimA + Pg -infected MoDCs was higher than Mfa1 + Pg-infected and un-infected MoDCs (p<0.001) (Fig. 10B). In addition, co-localization was higher in Pg381 and FimA + Pg with LAMP1 at 12 hours (Fig. 10C) (Table 4 and 5).

Discussion
Our results indicate that canonical autophagosomal and lysosomal clearance of P. gingivalis within DCs is dependent on initial routing to early endosomes, followed by autophagosomal and lysosomal routing. We show that P. gingivalis is able to survive within DCs by subversion of this canonical pathway via its Mfa-1 fimbriae. DCs are unlike macrophages and neutrophils in that the lysosomal machinery is specialized for efficient epitope preservation, rather total degradation [41]. Nevertheless, DCs must have mechanisms to control and inhibit the growth of intracellular pathogens; otherwise DCs could serve as a significant niche for pathogen dissemination to distant organs as recently suggested [42]. Autophagy has been widely recognized as an antibacterial lysosomal mechanism with an immune regulatory component [2,4,43][1]. Although the configuration of autophagy apparently serves DCs well in most situations, very little is known about its role in elimination of specific intracellular pathogens. More specifically, it is unclear how engagement of different PRRs on human myeloid DCs influences the induction of and maturation of antimicrobial autophagosomes.
The minor Mfa1 fimbriae are 67-kDa glycoproteins that target DC-SIGN for entry into MoDCs (and Raji cells) [14]. As this strain does not express FimA, a TLR2 agonist [28] expressed by all the other fimbriated strains, the present work represents a unique opportunity to study the role of PRRs on DCs in the context of antimicrobial autophagy. DC-SIGN is a type II transmembrane receptor that recognizes a wide range of pathogens through internal mannose branched structures and terminal di-mannose n-oligosaccharides [44]. Three internalization motifs (di-leucine motif, tri-acidic cluster, ITAM motif) on its cytoplasmic tail facilitate pathogen uptake [45]. Despite the diverse immune functions and regulations mediated by DC-SIGN [15,23,46,47], the immune escape mechanisms associated with DC-SIGN engagement by pathogens remain largely "unexplained" [17,18,20,21].
Our results are consistent with a role for DC-SIGN in routing P. gingivalis into distinct intracellular vesicles that escape early autophagosomal recognition and subsequent vesicle maturation and degradation. This DC-SIGN dependent re-routing seems to begin at the stage of early phagosomal formation, whereas DC-SIGN independent routing 'traps' P. gingivalis within Rab5 rich vesicles immediately after uptake. Rab5 has been suggested as an early stage initiator of autophagy and facilitator of subsequent lysosomal fusion [37] [40]. In addition, Rab5 is reported to enhance autophagy by inhibition of mTORC1 (mTOR complex). The results of the current study suggest that when TLR signaling is weak (e.g. in the absence of FimA on the Mfa1 + Pg strain) engagement of DC-SIGN dominates the response. This was clarified by adding exogenous TLR1/2 agonist, which inhibited Mfa1 + Pg survival in parallel with higher LC3-II signals and stronger DC maturation. We surmise that lacking a strong signal for autophagy activation such as a TLR2 agonist, pathogens internalized by DC-SIGN are preferably routed to non-autophagosomal, non-lysosomal compartments where they survive in DCs.
One of the striking observations of the present work was the preferential routing of Mfa1 + Pg to single-membrane intracellular vesicles, while the other strains were predominantly contained within characteristic double membrane phagophores. Classically, autophagy has been identified by the formation of double membrane vesicles that ultimately fuse with the lysosome for degradation of its intracellular components [48]. However, recently an alternative pathway known as non-canonical autophagy or LC3 associated phagocytosis (LAP) has been described [49], which involves the formation of single membrane vesicles. Here, DC-SIGN engagement appears to facilitate containment within single membrane vesicles; thus allowing P. gingivalis to evading lysosomal fusion. Some reports suggest that certain pathogens use (single membrane) autophagosomes for replication by blocking autophagosomal maturation [2,50,51]. Whether P. gingivalis utilizes DC-SIGN to enter a non-canonical pathway of autophagy for bacterial survival in DCs remains to be determined. One of the earliest studies describing P. gingivalis evasion of the conventional endocytic pathway and rerouting to (canonical) autophagosomes was in human aortic endothelial cells in [50]. Endothelial cells have been reported to express DC-SIGN [52] but this was not addressed in the previous study [50]. DCs, in contrast to endothelial cells, are highly migratory professional antigen presenting cells. Most reports of autophagy as an anti-microbial mechanism are from studies of immune cells [1,2,50,51,53]. In the present study of myeloid DCs, P. gingivalis appears to use an alternative tactic to evade autophagic capture through early engagement with DC-SIGN receptor and through non-canonical autophagy.
Analyses of intracellular survival of the different P. gingivalis strains suggest that DCs can use autophagy and lysosomal fusion to clear this intracellular pathogen. Pg381, which expresses both Mfa1 and FimA, engages both DC-SIGN and TLR2. It appears that DC-SIGN Mfa1 + Pg escaped these autophagic (double-membrane) vesicles and enclosed within single membrane structures or freely occupy the cytoplasm (Green arrows). D) The ratio of bacteria included in the double membrane were compared to total number of bacteria within MoDCs and plotted as percentage. Counting of the bacteria included in single or doubled membrane vesicles after 12 hours of infection. Each strain was counted in three randomly selected grids for each sample. The analysis of the bacterial counts used Kruskal-Wallis test of different groups and Dunn's test for multiple comparisons (*p<0.01).
doi:10.1371/journal.ppat.1004647.g009  routing is overruled by a strong TLR signal since Pg381 activates the autophagic process and is subsequently cleared, although this strain still survives longer inside MoDCs than outside in an aerobic atmosphere. Indeed, TLR signaling is a well-accepted inducer of autophagy as well as of phagosomal maturation [31][32][33]. However, the majority of autophagy studies that addressed TLRs were conducted in macrophages and neutrophils, both of which are naturally equipped with strong intracellular killing arsenal and higher expression of TLRs.
DC-SIGN expression is considered restricted to immature DCs with few exceptions in endothelium and specific macrophage subpopulations [54,55]. Our current data shows that DC-SIGN engagement may cause a positive feedback loop that increases the receptor expression resulting in subsequent bacterial uptake and survival. Increases in DC-SIGN+DCs in tissues of patients with periodontitis has previously been reported [56]. This could further enhance the pathogenicity of P. gingivalis and other DC-SIGN targeting pathogens. In the case of P. gingivalis, its ability to regulate fimbrial expression in different environmental cues such as pH, temperature and hemin content [26,27,34] may aid in its pathogenicity during chronic inflammation by regulation of DC-SIGN engagement (and expression). Current efforts in our laboratory are directed to identifying the levels of expression of Mfa1 and FimA in clinical samples of disease vs. healthy patients.
Phagosomes are known to have a high degree heterogeneity and individuality depending on host cell types, microbe captured and pattern recognition receptors engaged [11]. When Staphylococcus aureus and Salmonella typhimurium engage TLRs, for example, they are delivered to lysosomes at an inducible rate manifest by increased clearance and phagolysosomal fusion. However, slower phagolysosme maturation results from engagement with members of C-type lectin family and scavenger receptors [57]. DCs infected through DC-SIGN maintain an immature state and are more resistant to apoptosis [58]. Typically, immature DCs have a short life and active apoptosis is initiated shortly after maturation to avoid immune overstimulation [59,60]. Dominant DC-SIGN engagement alters such homeostatic balance and hinders the intracellular resistance to such infection. The DC-SIGN route may be a hallmark of chronic inflammation in response to low grade infection as it provides a protective niche for microbial persistence within the host. It is also influential in depolarizing the immune effector response [14]. Several lines of evidence have emerged linking autophagy and chronic inflammatory diseases [1,61,62]. In immature primary DCs, autophagy induction by NOD2 was essential for routing bacteria to lysosome and MHC presentation [61]. Early recognition of the microbe, here by DC-SIGN, could be crucial in driving 'normal' versus 'up-normal' [1] autophagy and affect the fate of inflammatory process in chronic periodontitis. The relevance of this work from a clinical standpoint comes from evidence that Mfa1+ P. gingivalis strains infect myeloid DCs in oral mucosal tissues and in blood of humans with periodontitis, wherein it is disseminated to distant sites of angiogenesis [42]. We conclude that this intracellular pathogen can survive within DCs by evasion of autophagy through coordinated regulation of Mfa1 and FimA expression. This may also facilitate its dissemination [42]. Potential therapeutic tactics for resolving chronic inflammatory diseases by forced autophagy to activate a strong immune response is also suggested by these studies.

MoDCs treatment with HIV gp120 and Cytochalasin D
MoDCs were pre-incubated with HIV-1 gp120 Chiang Mai (CM) envelope protein (GP120) for 30 min at 37°C. GP120 protein was obtained through the National Institutes of Health AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health (Cat. no. 2968) For actin polymerization inhibition, MoDCs were treated with cytochalasin D at 0.5 μM, the minimal concentration needed to arrest cytoskeletal rearrangements in Raji cells [14]. Cells were then washed 2 times with PBS and co-cultured with CFSE-stained P. gingivalis for 2, 12 and 24 hours at 37°C. Cells were fixed with 1% paraformaldehyde and prepared for immunofluorescence staining and epifluorescence microscopy.

Induction of autophagy
MoDCs were incubated with Rapamycin (Cat. no. Tlrl-rap, InvivoGen, San Diego, CA) at 200nM one hour after P. gingivalis infections. Induction of autophagy was confirmed by fluorescence staining of LC3-II within MoDCs after 2 and 6 hours.

DC infection with Porphyromonas gingivalis
Bacteria suspensions were washed five times in PBS and re-suspended for spectrophotometer reading at OD 660 nm of 0.11, which previously determined to be equal to 5 x 10 7 CFU [64]. For bacterial CFSE staining, the suspension were washed (3 times) and re-suspended in 5μM of CFSE in PBS. The bacteria were incubated for 30 min at 37°C in the dark [14,65]. MoDCs were pulsed with Pg381, Mfa1 + Pg, FimA + Pg and MFB at 0.1, 1 and 10 MOI and incubated with the MoDCs for 2, 6, 12 and 24 hours and each experimental condition were performed in triplicate.

Intracellular content of P. gingivalis strains within MoDCs
After 24 hours of MoDCs infection with P. gingivalis strains, cells were washed three times in PBS and re-suspended in sterile water on ice for 20 min to lyse the cells. Lysates were resuspended in anaerobe broth for 3 days. After broth incubation, bacterial suspensions were washed three times in PBS and re-suspended for spectrophotometer reading at OD 660 in triplicate. Viable counts (CFU) were calculated based on a plate count serial dilution versus OD readings. For confirming the identity of the P. gingivalis (black pigmented Gram negative coccobacilli) suspensions were cultured on 5% blood agar plates in triplicate under anaerobic conditions (10% H2, 5% CO2 in nitrogen). Plates were incubated in anaerobic conditions at 35°C for 14 days until black colonies were detected and select colonies gram-stained.

Survival assay of P. gingivalis in human MoDCs
After MoDCs were infected with the P. gingivalis strains for 2, 6, 12, 24 and 48 hours, cells were washed three times in PBS and re-suspended in sterile water on ice for 20 min to lyse the cells. Bacterial suspensions were washed three times in PBS and re-suspended for spectrophotometer reading at OD 660 in triplicate. Corresponding CFU counts were calculated based on a linear regression of plate count in serial dilution versus OD readings. Black colonies were confirmed in blood agar plate under anaerobic conditions (10% H2, 5% CO2 in nitrogen).

Transmission electron microscopy (TEM)
After MoDCs fixation, the procedures were carried out at the Electron Microscopy and Histology Core, Department of Cellular Biology and Anatomy, Georgia Regents University. The cells were fixed in 2% glutaraldehyde in 0.1 M sodium cacodylate (NaCac) buffer, pH 7.4, postfixed in 2% osmium tetroxide in 0.1 M NaCac, stained en bloc with 2% uranyl acetate, dehydrated with a graded ethanol series and embedded in Epon-Araldite resin. Thin sections were cut with a diamond knife and stained with uranyl acetate and lead citrate. Cells were observed in transmission electron microscope (JEM 1230-JEOL USA Inc.) at 110 kV and imaged with a CCD camera and first light digital camera controller (Gatan Inc.).

Immunogold transmission electron microscopy (IEM)
The procedures were carried out at the Electron Microscopy and Histology Core, Department of Cellular Biology and Anatomy, Georgia Regents University. Staining for DC-SIGN using anti-CD209 (mouse monoclonal, R&D Systems, # MAB161), was carried out to identify the P. gingivalis-containing vesicles. After DCs were pulsed with different P. gingivalis strains for the 2, 12 and 24 hours, cells were centrifuged into a pellet. Cells were be fixed in 4% formaldehyde 0.2% glutaraldehyde in 0.1 M sodium cacodylate (NaCac) buffer, pH 7.4, dehydrated with a graded ethanol series through 95% and embedded in LR White resin. Thin sections were cut with a diamond knife on a Leica EM UC6 ultramicrotome (Leica Microsystems Inc.) and collected on nickel grids. Sections were incubated in blocking buffer (5% BSA, 3% normal serum, 0.05% Tween-20 in Tris-buffered saline, pH 7.4) at room temperature in a humid chamber for 2 hours and with primary antibody diluted in blocking buffer overnight at 4°C. Grids were washed and incubated with gold-labeled secondary antibody for 2 hours at room temperature then washed and stained with 2% alcoholic uranyl acetate and 0.08% alkaline bismuth subnitrate. Cells were observed in a JEM 1230 transmission electron microscope (JEOL USA Inc.) at 110 kV and were imaged with an UltraScan 4000 CCD camera & First Light Digital Camera Controller (Gatan Inc.)

Quantitative real time PCR
For RNA isolation, direct lysis of the cell suspensions were achieved by RNeasy kit (Cat. no. 74104, Qiagen) by adding 300 μl of Qiagen's buffer RLT per sample. The lysates were collected and pipetted directly into the Qiashredder spin column. Ethanol (70%) was added and then samples were transferred to RNeasy spin columns. The samples were washed with buffer RW1, RBE, then the RNA samples were collected and stored at -80°C. RNA quantity and integrity were tested and only ratios of absorbance at 260 and 280 nm of 1.8-2.0, were included in the study. One-step qrt-PCR were performed using Express qPCR SuperMix (Cat. no. A10312, Invitrogen). Pre-formulated individual TaqMan gene expression primers (Applied Biosystems) were used for DC-SIGN mRNA detection (Hs.01588349_m1). For qrt-PCR reactions, 5μl of the RNA sample, 25μl PCR master mix (2x) and 2.5μl TaqMan gene expression assay were used per reaction. All PCRs were performed in triplicate and were carried out on a real-time PCR, StepOne (Applied Biosystems). For calculations and statistical analysis, fold changes were calculated using (2 -ΔΔCT ) method in the experimental samples [66]. Statistical analysis for gene expression was performed using the one sample t-test, which estimates the calculated difference (in fold-regulation) between experimental and control samples. A p value of <0.05 is the cut-off for significant differences. . All markers were measured against isotype controls. The markers were measured as MFI using the Accuri C6 Flow Cytometry system.

LC3-II immuno-fluorescence and epifluorescence microscopy
MoDCs were infected with P. gingivalis prelabeled with carboxyfluorescein succinimidyl ester (CSFE) fluorescence. Cells were fixed with 1% paraformaldehyde, washed with PBS twice and permeabalized with 0.5% saponin. MoDCs were incubated with LC3-II antibody (ab51520) for 2 hours and then washed with PBS. Pellets were res-suspended in cytospin fluid (Cat. no. 6768315, Shandon) centrifuged at 400 rpm for 4 minutes. Slides were mounted with anti-fade reagent (Invitrogen, P36931) and dried for microscopic analysis. Microscopic images were obtained with epifluorescence microscope (Nikon E600) then analyzed by image enhanced fluorescence microscopy aided by deconvolution analysis. Quantifications of the fluorescent intensities and co-localization within infected cells were done by NIS-Elements BR and AR software. Three randomly selected regions of interest were selected for each field to quantify fluorescence dye intensities.

Western blot analysis
Cells were centrifuged and washed twice with PBS. After washing, cells were lysed by addition of cell lysis buffer (Cell signaling Cat. no. 9803S) and incubated for 20 minutes on ice. Samples were centrifuged and the supernatant was collected and stored at -80°C. Proteins were denatured at 70°C for 10 minutes immediately prior to loading. For immunoblotting, 50 μg of total cellular protein per lane were separated by blot 4% to 12% Bis Tris Plus gradient gel and transferred to PVDF (polyvinylidine difluoride) membranes using iBlotting dry transfer system (Lifetechnologies, Cat. no. IB1001). The membranes were incubated with primary antibody LC3B (Abcam, Cat. No. ab48394) or GAPDH (Meridian life science, Cat. No. H86504M) and secondary antibody peroxidase-conjugated goat anti-rabbit or goat anti-mouse IgGs in iBind solution for 2.5 hours (iBind western system, life technologies). The specific protein signals were visualized using chemiluminescent peroxidase substrate and exposing the membranes to the high performance chemiluminescene film for detection. Protein loading was verified by detection of GAPDH using mouse anti-GAPDH monoclonal antibody.

Detection of LAMP1, Rab5 and 7 via transduction of fluorescent protein chimera by baculovirus transgenes
Monocytes were transduced with CellLight BacMam 2.0 (lifetechnologies) to visualize lysosomal marker LAMP1 (C10504), early endosome Rab5 (C10587) and late endosome Rab7 (C10589). Transduction was performed simultaneously with differentiation of MoDCs at the 5th day. Cells were transfected with 30 PPC (particle per cell) for 24 hours. At the 6th day MoDCs were harvested and immature phenotype (CD1c + DC-SIGN + CD83 -CD14 -) by flow cytometry. Quantifications of the fluorescent intensities and co-localization within infected cells were done by NIS-Elements BR and AR software.

Ethical aspects
These studies were determined by the Human Assurance Committee at Georgia Regents University to be human subject exempt, due to the use of anonymized peripheral blood samples for monocytes.