Oral Pathobiont Activates Anti-Apoptotic Pathway, Promoting both Immune Suppression and Oncogenic Cell Proliferation

Chronic periodontitis (CP) is a microbial dysbiotic disease linked to increased risk of oral squamous cell carcinomas (OSCCs). To address the underlying mechanisms, mouse and human cell infection models and human biopsy samples were employed. We show that the ‘keystone’ pathogen Porphyromonas gingivalis, disrupts immune surveillance by generating myeloid-derived dendritic suppressor cells (MDDSCs) from monocytes. MDDSCs inhibit CTLs and induce FOXP3 + Tregs through an anti-apoptotic pathway. This pathway, involving pAKT1, pFOXO1, FOXP3, IDO1 and BIM, is activated in humans with CP and in mice orally infected with Mfa1 expressing P. gingivalis strains. Mechanistically, activation of this pathway, demonstrating FOXP3 as a direct FOXO1-target gene, was demonstrated by ChIP-assay in human CP gingiva. Expression of oncogenic but not tumor suppressor markers is consistent with tumor cell proliferation demonstrated in OSCC-P. gingivalis cocultures. Importantly, FimA + P. gingivalis strain MFI invades OSCCs, inducing inflammatory/angiogenic/oncogenic proteins stimulating OSCCs proliferation through CXCR4. Inhibition of CXCR4 abolished Pg-MFI-induced OSCCs proliferation and reduced expression of oncogenic proteins SDF-1/CXCR4, plus pAKT1-pFOXO1. Conclusively, P. gingivalis, through Mfa1 and FimA fimbriae, promotes immunosuppression and oncogenic cell proliferation, respectively, through a two-hit receptor-ligand process involving DC-SIGN+hi/CXCR4+hi, activating a pAKT+hipFOXO1+hiBIM−lowFOXP3+hi and IDO+hi- driven pathway, likely to impact the prognosis of oral cancers in patients with periodontitis.

postmenopausal women 9 . Chronic inflammation of the periodontal tissues is mainly caused by a complex interaction of oral pathogens in the dental biofilm. Porphyromonas gingivalis (Pg) 10 is a highly influential member of the oral biofilm, causing a mucosal dysbiosis and disrupting immune homeostasis 11 through unclear mechanisms. Although, chronic inflammation is generally regarded as a causal factor in carcinogenesis 12 , the mechanistic role of CP and its dysbiotic microflora 13,14 in cancer risk are only now beginning to be explored 15 . Dendritic cells (DCs) are professional antigen presenting cells that bridge the gap between innate and adaptive immunity 16 . On the contrary, DCs can also play a role in inducing immune suppression under specific circumstances. From this perspective, DCs promote tolerance rather than immunity 17 . The functional diversity and plasticity of DCs has stimulated much interest in their use for immunotherapy for cancers, autoimmune and infectious diseases [18][19][20] . Distinct from DCs are myeloid-derived suppressor cells (MDSCs) with at least six subsets discovered so far. MDSCs in melanoma patients are Stat3 high and overexpress CD80, CD83 and CD209 or Dendritic Cell-Specific Intercellular adhesion molecule-3-Grabbing Non-integrin (DC-SIGN) 21 . MDSC induction of regulatory T cells (T regs ) contributes to tumor progression, failure of DC-based immunotherapy 22 and murine oral cancers 23 . A main factor in T reg induction by DCs and possibly MDSCs 24 is their production of immunoregulatory Indoleamine 2, 3-dioxygenase (IDO) enzyme. IDO is critical for T reg function and maintenance 25,26 promoting a tolerogenic state and inducing direct T cell suppression and enhancement of local T regs 27 . T regs control periodontitis in mice 28 , but are also reported to promote disease progression and pathogen immune escape 29,30 .
The induction of apoptosis in host cells and in their prompt removal by phagocytes is important for maintaining tissue homeostasis. This is particularly important for the development of an immunogenic response and antitumor immunity. Inhibition of apoptosis is mediated by downstream effectors of serine-threonine protein kinase-(Akt1) and Bcl-2 family members 31 . This anti-apoptotic pathway is activated in DCs by many factors including ligation of DC-SIGN 32 . DC-SIGN is a key uptake signaling receptor and immune-modulator for many pathogens, including Mycobacterium tuberculosis, Helicobacter pylori 33 , and as we have shown, fimbriated (Mfa1+) P. gingivalis strains 34 . Oral colonization with P. gingivalis leads to invasion of DCs and of their myeloid progenitors through the action of DC-SIGN ligand Mfa1 and TLR2/C-X-C chemokine receptor type 4 (CXCR4) ligand FimA. The chemokine stromal cell-derived factor 1 (SDF-1) and CXCR4 are significant markers of poor prognosis in many hematological malignancies 35 . Once phosphorylation occurs through Akt-1, the Forkhead box class-O (FOXO) proteins migrate from the nucleus and remain transcriptionally inactive resulting in their degradation or sequestration 35,36 . Since genes encoding pro-apoptotic molecules particularly Bcl-2 member Bim 37 are activated by FOXO members, its inactivation by DC-SIGN ligation can disrupt immune homeostasis. Deletion of FOXO1 36,38 reduces DC functions and enhances susceptibility to periodontitis in a mouse model 39 . It was described that FOXO1 silencing enhances cell proliferation and decreases apoptosis of papillary thyroid carcinoma cells via Akt-FOXO1 signaling 40 . However, the roles of phospho-Akt1 (pAKT1) in direct regulation of FOXO1 in CP or oral squamous carcinoma cells have not been described. Recently, we reported that human monocyte-derived DCs (MoDCs) exposed to P. gingivalis promote FOXO1 gene expression 41 . However, the mechanistic role of FOXO1/pFOXO1 in regulating myeloid cell plasticity and immune homeostasis in response to this pathogen is unknown.
We show here through a combination of human, mouse and in-vitro studies how the dysbiotic pathogen P. gingivalis disrupts immune surveillance in periodontitis. Mfa1-fimbriae expressing P. gingivalis strains invade monocytes and promote differentiation to apoptosis resistant IDO-competent MDDSCs. These MDDSCs induce immune tolerance through increased FOXP3 + T reg responses. Moreover, our data show in inflamed periodontal tissues that FOXP3 is a direct target of pFOXO1, which is regulated by pAKT1. Combined with our evidence for direct induction of OSCCs proliferation by P. gingivalis, we conclude that blocking this pathway is a promising interventional approach to restore immune-surveillance and reduce oncogenic cell proliferation in dysbiotic conditions such as periodontitis.
P. gingivalis expressing Mfa1 (DPG3) directs induction of anti-apoptotic and pAKT1-pFOXO1 mediated oncogenic signaling pathway in MDDSCs through DC-SIGN. We next examined by immunoblot, protein levels of pAKT and immaturity DC marker DC-SIGN in MDDSCs induced by DPG3 relative to Pg381 and MFI (totally lacks Mfa1) or uninfected MoDCs (Fig. 1A,B). DPG3 induces more phosphorylation of AKT1, relative to FimA expressing strains Pg381 and MFI (Fig. 1A,D). We posited that this could be promoting, not only survival of DPG3 through evasion of autophagy via DC-SIGN routing 44 , but also anti-apoptotic signaling. Since BIM, a pro-apoptotic gene controlled by FOXO1, promotes DC apoptosis 37 , we assessed relative p/tFOXO1 induction by DPG3 and Pg381 in MDDSCs by immunoblot ( Fig. 1A and E). Interestingly, the most anti-apoptotic inducing strain DPG3, considerably induced pFOXO1 whereas BIM protein expression decreased compared with control MoDCs (Figs 1A,F and S2C). Then, to address the role of DC-SIGN in activation of this signaling pathway, HIV-gp120 was used to block DC-SIGN as we have reported 34,46 . By immunoblot we detected increased DC-SIGN receptor expression by DCs induced by DPG3 relative to Pg381 and MFI or uninfected control (Fig. 1A,C), whereas blocking DC-SIGN with HIV-gp120 prior to infection reduced DPG3 (Mfa1) mediated-induction (Fig. 1B,C). However, it is noteworthy that DC-SIGN expression was not altered by gp120 alone in control MoDCs (Fig. 1B) and was thus a consequence of blocking P. gingivalis entry and/or activation of the DC-SIGN signalosome 34,46 . We should emphasize that DPG3 stimulation led to AKT serine473 (Ser473) phosphorylation which regulated FOXO1 threonine24 (Thr24) phosphorylation and expression in the DCs (Fig. 1A). Activation of AKT activity by Ser473 phosphorylation of its expression promotes cell survival through FOXO1 Thr24 phosphorylation. To verify the role of AKT in this pathway, DCs were co-treated with gp120, which impaired DC-SIGN-mediated survival signaling (Fig. 1B,D) and pFOXO1 (Fig. 1B,E). We also found that gp120 abolished this pathway in P. gingivalis381-induced MDDSCs by significantly decreased expression of DC-SIGN, pAKT1 and pFOXO1 functional proteins ( Fig. 1C-E), whereas the apoptotic protein BIM was increased (Fig. 1B,F). These data thus indicate a role of DC-SIGN in the DPG3-induced pAKT1-pFOXO1 mediated apoptosis resistance and immunosuppressive signaling pathway in MDDSCs.
Short-term murine oral infection with P. gingivalis fimbriae mutants activate immunosuppressive genes in blood and splenocytes. We next tested the ability of oral infection with P. gingivalis to induce this immunosuppressive pathway in gingival tissue, blood and secondary lymphoid organ, spleen of mice. Gene expression profiles of isolated blood ( Fig. 2A) and splenocytes ( Fig. 2B) of BL6 mice orally infected with P. gingivalis or its fimbriae deficient strains show distinct in-vivo responses depending on fimbria expression (Fig. S3). The strongest immunosuppressive responses were induced by Mfa1 + strain DPG3 after 12 hours of oral infection, including upregulation of Foxo1, Cire/Cd-209a, Cd40, Cd80 and Stat3 in blood (Figs 2A and S3B) and splenocytes (Fig. 2B), whereas Ido1 and Foxp3 were only induced in blood, but Bim, Foxo3 and Cd33 were downregulated in both. Serum IgG responses to P. gingivalis381 were also dampened by oral infection with DPG3 (Fig. 2C). These results emphasize the key role of P. gingivalis and its Mfa1 fimbriae type in early immunosuppressive responses.
Oncogene Akt1 induction and immunosuppressive state in DPG3-infected mice. We further hypothesized that oncogene Akt1 47 was activated locally in mice orally infected with Mfa1 + P. gingivalis. Immunostaining (Figs 2D-left panel, E and S4A) shows pAkt1 localized mostly at the level of sulcular epithelium and subjacent connective tissue, consistent with gingival inflammation, which may lead to angiogenesis and oncogenesis. Also, increased pFoxo1 ( Fig. 2D-middle panel: phosphorylation and nuclear exclusion of pFoxo1 protein indicated by arrow marks, 2F and S4B-left panel) expression was noted in DPG3 and Pg381 infected mouse splenocytes, compared to CMC treated. It is interesting to note that Foxp3 an important immunosuppressive factor in regulatory T cells (T regs ), is significantly increased in the gingiva of P. gingivalis and DPG3 infected mice (Fig. 2D,G and S4B-right panel), and in spleen of DPG3 and Pg381infected mice by immunoblot compared to uninfected CMC control. These data suggest that the fimbriae expression pattern of P. gingivalis strain type may play a key role in oncogenic progression and immunosuppression.
Distinct and direct induction of proliferation of oral cancer cell lines by P. gingivalis. We further hypothesized that activation of this pAKT-pFOXO1 signaling pathway by P. gingivalis would promote direct  FimA-expressing P. gingivalis promotes OSCC proliferation via CXCR4. As FimA targets CXCR4 on macrophages 49 , and is a co-receptor for the entry of specific strains of human immunodeficiency virus type I, we treated HN6 cells with CXCR4 inhibitor AMD3100, prior to co-culture with FimA + MFI. The results demonstrate a marked inhibition of HN6 proliferation (Fig. 3D) and comparable with control. In contrast blocking DC-SIGN mediated uptake of DPG3 using HIV-gp120 did not ablate proliferation (Fig. 3D). These results elucidate the critical role of CXCR4 and thus possibly FimA, in promoting OSCC cell growth. FimA activates CXCR4-mediated cellular signaling in OSCCs. Next, we tested whether the HN6 cells express DC-SIGN, since blocking DC-SIGN mediated uptake of DPG3 by gp120 was not able to block the proliferation of OSCCs. Immunoblot analysis confirmed that HN6 cells constitutively express CXCR4 (Fig. 3E), but not DC-SIGN (Fig. 3F); moreover, CXCR4, but not DC-SIGN is upregulated by P. gingivalis381 (Fig. 3F). Thus, differential receptor expression helps to explain the distinct roles of Mfa1 and FimA fimbriae in MDDSCs versus epithelial cells. Nonetheless, CXCR4 upregulation was also accompanied by increased phosphorylation of AKT1 (Ser473), pAKT1 accumulation in the cytoplasm, and translocation into the nucleus, resulting in activation of pAKT1-regulated transcription, as evidenced by increased expression and its Thr24 phosphorylation of transcription factors FOXO1 (Fig. 3E). The CXCR4 inhibitor AMD3100 reduced expression of oncogenic proteins SDF-1/CXCR4 expression, as well as pAKT1 and pFOXO1 activation (Fig. 3E), but it is interesting to note that the CXCR4 expression is comparable with the uninfected control. Moreover, control inhibitors of pAKT1 and pFOXO1, MK-2206 2HCL (MK) and AS184285 (AS), respectively, blocked activation of respective mediators (Fig. 3E). It is noteworthy that DPG3 also induces expression of CXCR4 in HN6 (Fig. 3E), however, the DC-SIGN expression was not detectable by immunoblot after exposure to Pg381and their mutant strains (Fig. 3F), thus, this data suggests that in addition to the receptor function of CXCR4 protein, its role as a metastasis mediator is only activated upon ligation by the cognate ligand, in this case, FimA. Although CXCR4 expression is higher in both DPG3 and MFI stimulated HN6 cells, DPG3 lacks FimA for ligation of CXCR4 leading to proliferation of OSCCs. Over all, these results suggest that CXCR4, provided it is ligated by FimA, is required for activation of the same pathway in OSCCs as in MDDSCs, namely, pAKT1-pFOXO1 pathway involved in immunosuppression and oncogenesis.

Systemic immunosuppressive and oncogenic state in CP patients.
To determine if this immunosuppressive and oncogenic pathway is active in humans with chronic periodontitis (CP), blood and gingival biopsy samples were obtained from CP and healthy control subjects. RNA of enriched DCs was analyzed for differentially expressed genes in CP patients normalized to healthy controls. The genes were cross-referenced with gene database from RNAseq of in-vitro MoDC infection model 41 . Of major significance were genes specifically induced in MoDCs in-vitro by DPG3, namely, CD274, IDO1, FOXO1, AKT1, CD209 (DC-SIGN), CXCR4, IL-10, FOXP3, CCL2 (MCP-1), STAT3 and STAT5B. Many of these same genes, particularly involved in immunosuppression (IDO1, IL-10, CD274, ARG1 and CD80), oncogenesis (CXCR4, SDF-1, AKT1 and STAT3) were highly induced in ex-vivo isolated blood panDCs of CP patients (Fig. 4A), whereas SOCS3 and SOCS1 were downregulated, which could render DCs more tolerogenic. Of vital note is the gene for IDO1, which was upregulated 5-fold relative to controls, consistent with our previous in-vitro P. gingivalis infection model 43 . Histopathological analysis of gingival tissues from representative CP subject, compared with healthy control, shows chronic inflammatory infiltrates consisting of lymphoid and myeloid cells in the CP connective tissues (Fig. 4B). Preferential induction of dysbiotic transcription factor pAKT1 in CP gingival tissues by myeloid dendritic cell subpopulation, as confirmed by immunofluorescence analysis (Figs 4C and S6) showing pAKT co-localization with pFOXO1 and DC-SIGN (DC's marker) as well as the immunosuppressive T reg protein FOXP3 co-localized with DC-SIGN, compared with healthy control, supporting involvement of pAKT1 in dysbiosis. Interestingly, immunoblot shows increased pAKT1, pFOXO1, FOXP3, IDO1 and decreased BIM proteins (involved in immunosuppressive and oncogenic pathway) in CP gingival tissues compared with healthy controls (Fig. 4D). These data support an immunosuppressive state in severe CP patients prone to the oncogenic process.

P. gingivalis and its Mfa1 fimbrial genotype increased in enriched blood DCs from CP patients.
To identify the presence of P. gingivalis and its relative Mfa1 + fimbrial expression in CP, we analyzed mRNA from enriched blood panDCs of CP patients and healthy controls. The healthy control DCs contained no detectable P. gingivalis 16 s rDNA, so we focused on CP samples. The results (Fig. 4E) show higher mRNA-expression of Mfa1, relative to the other fimbriae type, FimA, inside blood DCs. When considered in the context of immune homeostatic functions disrupted by DPG3 in-vitro (Fig. 1), in-vivo (Fig. 2) and in CP samples (Fig. 4), this supports Mfa1 as an important dysbiotic factor that promotes a systemic immunosuppressive and anti-apoptotic state. This result also suggests that Pg381 may adapt physiologically and genetically to the in vivo environment of CP by upregulating immunosuppressive genes.

Role of FOXO1, FOXP3 and BIM target genes in FOXO1 and AKT1 signaling in CP patients.
To confirm the mechanistic role of AKT1 or FOXO1 in regulation of FOXO1, FOXP3 and BIM transcription in CP tissues in situ, we searched for evolutionarily conserved FOXO-binding sites in human FOXO1, FOXP3 and BIM loci. Using the rVISTA tool, we found putative FOXO1 and FOXP3 -binding sites in previously defined conserved noncoding sequences of FOXO1, FOXP3, and BIM promoter region (Table S4). We reiterate the observation of dramatically higher levels of pFOXO1 in DPG3 infection induced MDDSCs compared to control monocytes ( Fig. 1) suggesting that FOXO1 expression is negatively regulated by PI3K-AKT signaling pathway. To further corroborate FOXO1 as a target for AKT1 in cells from gingival tissues of CP, we examined FOXO1 as well as FOXP3 and BIM expression and their nuclear localization using antibodies to FOXO1 (anti-FOXO1) or anti-AKT1. Chromatin Immunoprecipitation (ChIP) assays demonstrated genomic fragments containing the proximal FOXO1 (Fig. 4F) promoter regions in the DNA upstream of the translation start site selectively enriched with the pAKT1 antibody, whereas the BIM (Fig. 4H) was significantly decreased in gingival tissues. We next tested if FOXO1 regulates the expression of FOXP3 and BIM in cells from gingival tissues, selectively enriched and repressed for genomic fragments containing the region in FOXP3 (Fig. 4G) and BIM (Fig. 4H) promoters, respectively. This finding describes a critical role for an evolutionarily conserved AKT1 and FOXO1 binding elements in control of FOXO1, FOXP3 and BIM promoter activity and supports their role as target genes for immunosuppression and dysbiosis in CP.

Oncogenic profile confirmed in oral tissues from P. gingivalis infected CP patients. Since
MDDSCs provide anti-apoptotic, immunosuppressive functions in-vitro and in-vivo; we further analyzed the genes and proteins related to oral cancer in P. gingivalis infected CP patient tissues by qPCR and immunoblot, respectively. The results delineate increased expression of angiogenic and oncogenic markers CXCR4, SDF-1, and decreased expression of tumor suppressor genes, IRF1, LDOC1, and FOXO3 in CP patient tissues (Fig. 4A). A surprise consequence was the similar responses observed in CP patient samples as MDDSC phenotypic markers (CD16 −low , CD11b −low and CD33 −low ) (Fig. 4A) was also consistent with the results obtained in-vitro (Fig. S1A,B). Interestingly, immunoblot analysis shows increased expression of oncogenic markers CXCR4 and SDF-1 in oral tissues of CP compared with healthy controls (Fig. 4D), thus these data demonstrate that the pathobiont P. gingivalis activates anti-apoptotic pathway promoting immune suppression and possibly, oncogenesis.

Discussion
We report here two distinct, though linked mechanisms activated by the oral pathobiont P. gingivalis that may possibly promote oral and other cancers: (a) immunosuppression through induction of MDDSCs, predominantly dependent on the DC-SIGN targeting Mfa1 fimbriae and (b) oncogenic response, predominantly through the CXCR4-targeting FimA fimbriae. Several recent reviews discuss the various cancers associated with CP, including gastrointestinal, pancreatobiliary tract 50 22 . A recent review highlights the importance of phenotypic and transcriptional-profiling to redefine the DC lineage 62 . As blood panDCs of CP patients contained P. gingivalis Mfa-1, transcriptional profiling of panDCs was performed and cross-referenced with transcripts induced by infection of human monocytes and mice with Mfa1 + P. gingivalis. Most notable were of genes involved in DC differentiation, immune regulation and regulation of apoptosis, markedly FOXO1. Examination of gingival tissues of CP patients by functional genomics ChIP assay specifically links FOXO1 proteins with the DNA element of FOXP3, indicating that FOXP3 is a direct FOXO1 target gene in CP tissues. This is consistent with the demonstrated ability of MDDSCs to induce T reg . The progression of CP in humans is related to over-expression of FOXP3 and IL-17 63 and with uptake of P. gingivalis by blood myeloid-derived DCs in-vivo 15 . In addition, the transcription factor STAT5B, also activated by Mfa1 + P. gingivalis strain DPG3, binds to regulatory elements in the FOXP3 locus and may regulate FOXP3 expression. Thus, FOXO1 proteins appear to be primary regulators of MDDSC plasticity and T reg cell lineage and are induced by P. gingivalis infection. The production of IDO1 in chronically inflamed sites, as in CP 64 , can block Th1 and Th17 effector differentiation, T cell apoptosis and promote T regs . Immature or regulatory DCs can produce IDO 65 capable of biasing the immune system towards tumor growth support. High rates of necrotic and apoptotic cell death have been observed in IDO-KO mice after P. gingivalis LPS challenge compared to WT mice 26 . We showed that IDO was involved in both anti-apoptotic and immune suppressive response of MDDSCs. In addition, we have also shown increased expression of immunosuppressive cytokines IL-10, TGFβ 66 , immunosuppressive molecule PD-L1 67,68 and Survivin (BIRC5) 69 , which influence tumor progression in many types of cancers. Strikingly, we found that in CP patients, there was increased expression of immunosuppressive, angiogenic and oncogenic markers, and decreased expression of tumor suppressors in the gingiva at mRNA and protein level. Another interesting finding of this study was increased expression of Mfa1 at mRNA level in CP patient blood panDCs, which might be one of the main roles of Pg381 dissemination, wherein the bacterium adapts physiologically or genetically to their local environment.
Our results show a key role for DC-SIGN in P. gingivalis Mfa1-mediated activation of STAT3, which phosphorylates AKT1 (pAKT1) for full activation. Phospho-AKT1 suppresses BIM, promoting apoptosis resistance and monocyte differentiation into MDDSCs. The expression of active forms of FOXO family members in dividing cells promote cell cycle arrest at the G1/S boundary, which is a significant cellular mechanism by which FOXO promotes tumor suppression 70,71 . Phospho-AKT1 in turn phosphorylates and inactivates FOXO1 which further suppress BIM, a key FOXO target gene that mediates the ability of FOXO to induce cell death 72 . It is interesting to note that AKT-FOXO1 regulates the BIM and transcriptionally inactivates FOXO1 (pFOXO1) leaving no opportunity to bind with BIM to subsequently mediate apoptosis upon the persistent stimulation of Pg strains. Phosphorylated FOXO1 also regulates FOXP3 expression through its feedback regulatory loop mechanism as per the consistent and continuous stimulation of Pg strains in MDDSCs, to promote apoptosis resistance and immunosuppression. This pathway (DC-SIGN-pAKT1-pFOXO1-BIRC5-STAT3-IDO1-FOXP3) culminates in in gingival tissue of CP compared with healthy control. (E) Mfa1 and FimA mRNA in ex-vivo isolated panDCs from CP patients and Healthy control and normalized with the internal control. (F-H) Regulation of FOXO1, FOXP3, BIM transcription by FOXO1A or pAKT1. Immunoprecipitation of chromatin from the human FOXO1 (F), FOXP3 (G) and BIM (H) locus in mixed immune cell population from gingival tissues of CP and healthy control with anti-pAKT1 or anti-FOXO1A, followed by qPCR analysis of immunoprecipitants; results are presented relative to input by immunoprecipitation with isotype-matched control antibody. Data are representative of six independent experiments. *P ≤ 0.05, **P ≤ 0.01. MDDSC differentiation, T regs activation and IDO-mediated immunosuppression, which may regulate cellular processes leading to immune escape. We have also elucidated here an oncogenic mechanism whereby P. gingivalis activates OSCCs and have identified the molecular target. We have shown increased expression of metastasis markers CXCR4, SDF-1 73,74 and decreased tumor suppressor molecules LDOC1 and FOXO3, induction of oncogenes and promotion of tumor growth. Immunoblot analysis further confirmed increased expression of CXCR4 and SDF-1 in oral tissues of CP compared with healthy controls. Further, strain MFI (FimA + Mfa1 − Pgingivalis) stimulates significant proliferation of CXCR4-expressing OSC (HN6, HN12) cells, but not those of noncancerous cells (oral keratinocytes, ARPE and MEF), which indicate that the interactions between FimA and CXCR4 play a primary role in promoting OSCCs tumor growth through the pAKT1-pFOXO1 dependent pathway(s). In addition, the available evidence indicates that the broad somatic deletion of FOXO1/3/4 in mice results in thymic lymphomas and hemangiomas 75 , specifying that the FOXO family functions as a tumor suppressor in mice 70 . Interestingly, we found increased expression of pFOXO1 (transcriptionally inactive) and decreased expression of FOXO3 in in-vitro, in-vivo and ex-vivo samples, thus supporting the findings that FOXO family members can promote tumor progression.
In conclusion, our study supports a novel apoptosis resistant pathway (Fig. 5) that promotes both immunosuppression and oncogenesis, with the common factor being infection with the oral pathobiont P. gingivalis. This pathway, as activated in CP patients, and in mice orally infected with P. gingivalis is a common pathway involved in oncogenesis. Thus, our results collectively support a two-hit process, involving Mfa1-fimbriae-DC-SIGN mediated evasion of immune response, and FimA-CXCR4 mediated inflammation and cancer development. Manipulation of this pathway may be therapeutically beneficial for patients in the management of dysbiosis-related immunological disorders as in chronic periodontitis and in preventing cancer development.
Oral infection with Pg381 and fimbriae mutants in acute and chronic murine model of CP. All mice had C57BL/6 backgrounds were purchased from Jackson (Bar Harbor, ME). To determine the immunobiological consequences of oral infection with Pg381 and fimbriae mutants (DPG3, MFI and MFB) in-vivo (for both acute & chronic), we orally-infected B6 mice (n = 36) with these strains and performed gene expression profiles of isolated blood and splenocytes. Mice were fed a normal chow diet. Six-week old male mice were treated with a 10-day regimen of oral antibiotics (sulfamethoxazole/trimethoprim suspension 48 mg/mL) to allow for P. gingivalis colonization. Mice were challenged by oral application of vehicle (2% carboxymethylcellulose-CMC in PBS) or the P. gingivalis strains (1 × 10 9 Colony Forming Units -CFU) at the buccal surface of the maxillary vestibule 5 times a week for 3 weeks as described 77 . The isogenic fimbriae deficient mutants (Table S1) were maintained under proper antibiotic presence to maintain mutant traits as described earlier 44,76,78,79 . Liquid cultures were maintained until mid to late log phase. Bacteria were washed/re-suspended in PBS. 1:10 dilution of the culture and matched to an optical density of 1.0 at 660 nm (O.D. 660 = 1 × 10 9 CFU/ml). 2% CMC was added slowly to the bacterial suspension while vortexing to avoid clumping. Purity of the cultures was verified by Gram stain before introducing bacteria into animals. Mice were then sacrificed at 1, 12 and 24 hours, and then blood and spleens were collected for RNA isolation (Qiagen). After 4 weeks of chronic infection, mice were sacrificed and mandibles (with attached gingiva) were harvested and fixed in 4% PFA. After 3 days, samples were decalcified with EDTA 14% solution for 3 weeks, then embedded in paraffin and processed for H&E and immunostaining.
Antibody Analysis. We next determined differences in anti-P. gingivalis antibody titers in BL6 mice (n = 12) after oral infection using standard enzyme-linked immunosorbent assay (ELISA) protocol. Briefly, diluted mice sera (1:100) were reacted with the bacterial antigen for 2 hours, then washed and labeled with a secondary antibody goat anti-mouse IgG conjugated to alkaline phosphatase (1:5000). Assays were developed with p-nitrophenolphosphate and reactions were terminated by the addition of 3 M NaOH and the optical density (OD) was measured at OD 405 nm using a BioRad Microplate Reader. Mice were monitored twice a week for clinical symptoms and general health status.
Cell Proliferation and Inhibitory Assay. Cell proliferation assay was performed as described 48 with a few modifications, briefly, human head and neck oral squamous cell carcinoma (OSCCs) derived cell lines, HN6 and HN12 (a gift from Dr. Yeudall AW, Department of Oral Biology, AU, Augusta, GA, USA) were cultured as described 80 and seeded in 6-well plates at 1-5 × 10 5 cells per well in the growth medium. Cells were uninfected or incubated with DPG3, MFI, FadA + and LPS (100 ng/ml) (as positive controls) or inhibitors added at the concentration of AMD3100 (50 nM), MK-2206 (2 µM), AS1842856 (50 nM), gp120 (6 µg/ml) prior to co-culture with P. gingivalis at MOI = 1. . CP subjects with moderate to severe disease, determined by exhibition of the following: probing depth of >4 mm, attachment loss of >3 mm, bleeding on probing, and alveolar bone crest >3 mm from the cementoenamel junction. Gingival tissue biopsy and blood samples from CP patients (n = 24) and healthy controls (n = 25) were collected. We have previously reported our biopsy technique to obtain oral lymphoid foci from the interdental papilla without removing buccal and lingual epithelium at the deepest bleeding site 83,84 . These tissues were cryosectioned in OCT for H&E and immunofluorescence analysis by confocal microscopy. Blood was also collected for transcriptome and DC subset analyses as described above.

TaqMan-Array and SYBR-Green qPCR Assays. Analysis of gene expression in MDDSCs induced by
Pg381, DPG3, MFI, and MFB, MoDCs and monocytes was performed using RT-PCR. Total RNA was isolated using RNeasy kit from three groups such as (1) In-vitro control and P. gingivalis strains at 1 MOI infected monocytes were cultured for 6 hours, and day 6 MoDCs; (2) In-vivo mice blood and spleens at 1, 12 and 24 hours; and (3) Ex-vivo human blood panDCs isolated from healthy control and chronic periodontitis patients. Human blood DCs were negatively selected from PBMCs using panDC pre-enrichment kit (Robosep, Stem cell technologies). The isolation was performed two times to remove non-DCs. The purity of enriched DCs was 70-80%.
Complementary DNA was synthesized from 1.1 μg RNA through a reverse-transcription reaction (Applied Biosystems). Real-time quantitative PCR (RT-qPCR) was performed on TaqMan array fast plates as well as reconfirmed with SYBR green methods for selected genes (PCR primers are listed in Tables S2 & S3). The calculations for fold regulation used the 2 −ΔΔCt method as described in our previous report 41 . 18SrRNA, HPRT1 and GAPDH were used as internal controls.
Immunofluorescence Assay (IFA). IFA was performed for paraffin embedded tissues. Briefly, the paraffin embedded tissues from mice (CMC, Pg381 and DPG3 oral infected CP model) and human (healthy and CP patients) were rehydrated and antigen retrieved. After blocking with permeabilization buffer and 2% BSA, slides were then incubated with primary antibodies followed by respective secondary antibodies. Slides were mounted with prolong diamond anti-fading mounting medium (Invitrogen). Images were acquired with a Zeiss LSM510 Inverted Meta scanning confocal microscope. For quantitative analysis, four fields were selected randomly and total cells in the field were manually counted based on their expression of the target proteins.
Immunoblot and Receptors Blockade Assays. Immature MoDC were infected with WT-Pg381, DPG3 and MFI for 6 hours at 1 MOI. DC-SIGN was blocked with GP120 (6ug/ml) for 30 minutes before infection. Likewise, OSCCs were infected with Pg381, DPG3, MFI, FadA + at MOI = 1 and LPS (100 ng/ml) (both FadA + and LPS used as positive controls). CXCR4, pAKT, pFOXO1 and DC-SIGN were blocked with their respective antagonist (inhibitors) at the concentration of AMD3100 (50 nM), MK-2206 (2 µM), AS1842856 (50 nM), and gp120 (6 µg/ml). The concentrations of inhibitors were selected based on the IC 50 value as per the manufacturer's instruction. Whole-cell lysates were prepared from both MoDCs and OSCCs cultures by centrifuging and washing the pellet with ice-cold PBS. The gingival tissue used for this study was obtained from chronic periodontitis (CP) subjects (generalized severe) and healthy adult controls as previously described 83 . Proteins from human gingival tissues, mouse spleen and the cells were extracted by adding ice-cold Radioimmunoprecipitation assay (RIPA) buffer (Abcam) supplemented by protease inhibitor cocktail (Cell signaling). The cell lysates were centrifuged at 12,000 × g for 10 min at 4 °C. Samples were normalized to the amount of total protein in the supernatant using Pierce ™ BCA Protein Assay Kit (ThermoFisher). Protein aliquots (30 μg) were separated by size on 4-15% Mini-PROTEAN TGX stain-free precast gels (Bio-Rad Laboratories, Inc.) and transferred to Immobilon (PVDF) membrane (EMD Millipore). Nonspecific binding sites were blocked by incubation in 1 × TBS-T (0.2 m Tris, 0.14 m NaCl, 0.1% Tween 20) containing 5% non-fat milk for 1 hour at room temperature, followed by incubation overnight at 4 °C with 1:1000 dilution of primary antibodies (Akt, p-Akt, FOXO1, p-FOXO1, DC-SIGN (CD209), and GAPDH) in 1 × TBS-T containing 5% non-fat milk. After membranes were washed three times in 1 × TBS-T (10 min each at room temperature), horseradish peroxidase-conjugated secondary antibody was added at a 1:2000 dilution and incubated for 1 hour at room temperature. After three more washes with 1 × TBS-T, the immunoreactive peptide was detected by Western Lightning ECL Pro Chemiluminescent reagent (PerkinElmer, Inc) and imaged using ChemiDoc ™ MP Imaging System (Laboratories, Inc.). Data were analyzed by performing one-way ANOVA with Tukey's post hoc using Prism GraphPad.
Chromatin Immunoprecipitation (ChIP). ChIP analyses were done as described 36 . Briefly, once the unicellular suspension was obtained from the gingival tissues of CP patients and healthy subjects, cells (5 × 10 6 ) were fixed for 10 min at 25 °C with 10% formaldehyde. After incubation, glycine was added to a final concentration of 0.125 M to 'quench' the formaldehyde. Cells were pelleted, washed twice with ice-cold PBS and lysed. The lysates were pelleted, re-suspended and sonicated to reduce DNA length to 300-500 base pairs. Chromatin prepared (#D5210, Zymo Research Corporation, CA) from CP patients and healthy control gingival tissue and cells were pre-cleared for 1 hour with protein Ultralink A/G agarose beads (Pierce, #88803) and then was incubated overnight with 2 µg ChIP grade anti-AKT1 (MA5-14898; ThermoFisher), or anti-FOXO1A (ab39670; Abcam), or control rabbit immunoglobulin (2729; Cell Signaling), and one without antibody as the input. Particle Concentrator (#123.21D, Invitrogen), washed, and eluted in 100 µl of 0.1 M NaHCO3 and 1% SDS elution buffer. The eluted and input chromatins were incubated at 65 °C (for 3 hours) to reverse the crosslinking. After digestion with proteinase K, the ChIP and input DNA were purified (using Chip DNA Clean & Concentrator Kit, #D5205 Zymo Research) and analyzed by quantitative PCR using site-specific primers binding of AKT1 with the co-factors to the FOXO1 and FOXO1 to the FOXP3 and BIM locus (Table S4).
Ethical approval and informed consent. The Human Assurance Committee (HAC) at the Augusta University approved this protocol involving human cells as human subject exempt, due to the use of anonymized peripheral blood samples for monocytes. The study protocol and informed written consent form obtained from all subjects were reviewed and approved by the School of Dentistry Ethics Committee, University of São Paulo, Brazil (658.998/CEP), and the Augusta University IRB (831204-3). All procedures involving mice were reviewed and approved by the Augusta University Institutional Animal Care and Use Committee (IACUC ID: # 2013-0586)/ethics committee and were performed in accordance with the approved guidelines and regulations.

Data Availability
The datasets generated and analyzed during the current study are available from the corresponding authors on reasonable request.