Attenuation of the Phosphatidylinositol 3-Kinase/Akt Signaling Pathway by Porphyromonas gingivalis Gingipains RgpA, RgpB, and Kgp*

Background: The significance of gingipains on the PI3K/Akt signaling pathway is poorly understood. Results: The inactivation of PI3K and Akt was induced by P. gingivalis, which was associated with gingipains. Conclusion: Gingipains are critical for suppressing the PI3K/Akt signaling pathway via extracellular proteolysis independent of P. gingivalis invasion. Significance: This study shows the first evidence that the gingipains negatively regulate the PI3K/Akt signaling pathway. Porphyromonas gingivalis is a major pathogen of periodontal diseases, including periodontitis. We have investigated the effect of P. gingivalis infection on the PI3K/Akt (protein kinase B) signaling pathway in gingival epithelial cells. Here, we found that live P. gingivalis, but not heat-killed P. gingivalis, reduced Akt phosphorylation at both Thr-308 and Ser-473, which implies a decrease in Akt activity. Actually, PI3K, which is upstream of Akt, was also inactivated by P. gingivalis. Furthermore, glycogen synthase kinase 3α/β, mammalian target of rapamycin, and Bad, which are downstream proteins in the PI3K/Akt cascade, were also dephosphorylated, a phenomenon consistent with Akt inactivation by P. gingivalis. However, these events did not require direct interaction between bacteria and host cells and were independent of P. gingivalis invasion into the cells. The use of gingipain-specific inhibitors and a gingipain-deficient P. gingivalis mutant KDP136 revealed that the gingipains and their protease activities were essential for the inactivation of PI3K and Akt. The associations between the PI3K regulatory subunit p85α and membrane proteins were disrupted by wild-type P. gingivalis. Moreover, PDK1 translocation to the plasma membrane was reduced by wild-type P. gingivalis, but not KDP136, indicating little production of phosphatidylinositol 3,4,5-triphosphate by PI3K. Therefore, it is likely that PI3K failed to transmit homeostatic extracellular stimuli to intracellular signaling pathways by gingipains. Taken together, our findings indicate that P. gingivalis attenuates the PI3K/Akt signaling pathway via the proteolytic effects of gingipains, resulting in the dysregulation of PI3K/Akt-dependent cellular functions and the destruction of epithelial barriers.

(mTOR) complex 2 to become fully activated (21). Active Akt is a central node in cell signaling downstream of PI3K by various cellular stimuli, and it modulates the substrates and downstream molecules Bad, caspase-9, glycogen synthase kinase 3 (GSK3) ␣/␤, mTOR, and forkhead transcription factors (22,23). However, the dysregulation of Akt activity underlies the pathophysiological properties of a variety of diseases, including type 2 diabetes and cancer (22,24,25). The PI3K/Akt signaling pathway has also been implicated in bacterial infectious diseases. Interestingly, bacteria and their virulence factors have been shown to regulate this pathway positively or negatively during infection (26 -29). P. gingivalis activates the PI3K/Akt signaling pathway, which is linked to cell survival (30,31) and immune responses (32). Among several known virulence factors of P. gingivalis, both LPS and fimbriae regulate immune responses through PI3K/Akt activation in monocytes/macrophages (5,(33)(34)(35) and human gingival fibroblasts (36). Gingipains have the potential to control both cell survival and apoptosis via the PI3K/Akt pathway (37). However, recent reports by other research groups have shown that P. gingivalis infection in the liver inactivates Akt and alters hepatic glycogen synthesis, leading to potential progression of diabetes (38), and that P. gingivalis LPS also represses mucin synthesis and Akt activation (39).
In light of these observations, the PI3K/Akt signaling pathway can be concluded to have pivotal roles in infectious diseases; however, the significance of this pathway has not been sufficiently clarified in P. gingivalis-mediated periodontal diseases. Here, we investigated the effect and role of P. gingivalis infection and gingipains on the PI3K/Akt signaling pathway in human gingival epithelial cells. We found that P. gingivalis infection caused the attenuation of the PI3K/Akt pathway, which was strongly associated with the activities of the gingipains, but P. gingivalis invasion was not required for the attenuation. Our studies revealed a unique function of gingipains as potent negative effectors of the PI3K/Akt signaling pathway.

EXPERIMENTAL PROCEDURES
Cells, Antibodies, and Inhibitors-The human gingival epithelial Ca9-22 cell line was obtained from the Culture Collection of Health Science Research Resources Bank, Japan Health Sciences Foundation, and the cells were grown in MEM␣ (Wako) containing 10% FCS. Human primary gingival epithelial cells (HGEP; HGEPp.05. lot ES1208166) were obtained from CELLnTec and maintained in CnT-24 culture medium (CELLnTec). Cells were cultured at 37°C in a humidified atmosphere of 5% CO 2 in air. The antibodies against GAPDH (no. sc-25778) and Bad (no. sc-8044) were purchased from Santa Cruz Biotechnology. Caveolin-1 (no. 610406) and ␤-catenin (no. 610153) were purchased from BD Biosciences. The tag antibodies against HA (no. MMS-101R) and DYKDDDDK (no. 018-22381) were purchased from Covance and Wako, respectively. PI3K p85 (no. 06-195) was purchased from Millipore. All phosphorylation-specific antibodies and their nonphosphorylated specific controls used in this study were purchased from Cell Signaling Technology, except for the antibodies described above. Cytochalasin D (CytD) (037-17561) and methyl-␤-cyclodextrin (MCD) (320 -84252) were obtained from Wako.  and , which are inhibitors of Rgp and Kgp, respectively, were purchased from the Peptide Institute.
Infection of Human Host Cells with P. gingivalis-P. gingivalis wild-type strain ATCC3327 (WT) and the gingipains-deficient mutant strain KDP136 (rgpA, rgpB, and kgp) (40) were used in this study. Before challenging the host cells, P. gingivalis strains were cultured on TSA/BHI agar plates containing hemin, menandione, and L-cysteine (41) under anaerobic conditions in an atmosphere of 10% CO 2 , 10% H 2 , and 80% N 2 at 36°C for 24 -48 h. P. gingivalis was incubated with the host cells at a multiplicity of infection (MOI) of 100 for the indicated times.
Western Blotting-Ca9-22 and HGEP cells were infected with P. gingivalis at an MOI of 100 at 37°C for the indicated times. The infected cells were lysed, and the cell lysates were run in SDS-PAGE using the indicated gel concentrations and then transferred to PVDF membranes for Western blotting. The target proteins were probed with the primary antibodies listed above, and subsequently goat anti-rabbit (1:1000) or goat antimouse HRP-conjugated (1:1000) secondary antibodies (Dako) were used for detection. The proteins were detected using ECL Western blotting detection reagents (GE Healthcare), according to the manufacturer's instructions and scanned using an ImageQuant LAS 4000 mini. The band densities were quantified using ImageJ software.
Immunofluorescence Analysis-Ca9-22 and HGEP cells, seeded at 3.0 ϫ 10 5 cells per well in 6-well plates were incubated in serumfree medium for 24 h. The cells were infected with or without P. gingivalis at an MOI of 100 at 37°C for 2 h. After incubation, the cells were fixed at room temperature for 10 min in PBS containing 4% paraformaldehyde. The fixed cells were treated for 5 min with 0.1% Triton X-100 for membrane permeabilization and blocked with 4% BSA in PBS for 30 min. The permeabilized cells were incubated with Ser(P)-473 Akt (1:50) and then incubated with anti-rabbit antibodies conjugated with Alexa Fluor 488 (1:500, Invitrogen) as secondary antibody in TBS containing 1% BSA and 0.5 g/ml DAPI for 1 h. The samples were analyzed using a Cellomics ArrayScan VTI HCS reader (Thermo Scientific) to quantify relative fluorescence intensities for the phospho-Akt positive cells versus control cells.
Cell Culture Insert System and Stimulation by P. gingivalis Culture Supernatant (Transwell Assay)-Ca9-22 cells (1.0 ϫ 10 5 cells/well in 24-well plates) were incubated with serum-free medium for 24 h. After incubation, the cells were treated by challenging them with P. gingivalis at an MOI of 300 through cell culture inserts for 24-well plates (BD Falcon), or the cells were incubated with 300 l of P. gingivalis culture supernatant for the indicated times, which was prepared by incubation of P. gingivalis (4.0 ϫ 10 7 bacteria/tube) in 400 l of MEM␣ under anaerobic conditions for 24 h; the suspension was then filtered through a 0.45-m PTFE filter (Sartorius Minisart SRP Syringe Filters). The cells treated under each condition were lysed and analyzed with SDS-PAGE and Western blotting using the antibodies for Akt (1:1000) and Ser(P)-473 Akt (1:1000).
Kinase Assay for PI3K and Akt-The activities of PI3K and Akt were measured using the PI3K Activity ELISA Pico Kit (Echelon Biosciences) and a modification of the nonradioactive Akt kinase assay kit (Cell Signaling Technology), respectively, according to the manufacturer's instructions. In brief, Ca9-22 and HGEP cells were infected with or without P. gingivalis at an MOI of 100 at 37°C for 4 h. The cells were then lysed in a buffer containing 1 mM leupeptin, 10 mM N-␣-tosyl-L-lysyl-chloromethyl-ketone, 1 mM KYT-1, and 1 mM KYT-36. For the determination of PI3K activity, PI3K was isolated by immunoprecipitation using an anti-PI3K p85 regulatory subunit antibody, and the activity of the PI3K p110 catalytic subunit that coimmunoprecipitated with PI3K p85 was assessed by measurement of PI(3,4,5)P3 generated from a PI(4,5)P2 substrate solution by ELISA. With regard to Akt activity, the immobilized Akt antibody slurry was used to immunoprecipitate Akt from cell lysates via gentle agitation for 4 h at 4°C followed by spinning at 10,000 ϫ g in a centrifuge. The pellets were resuspended in 40 l of kinase buffer containing 200 mM ATP and 1 g of GSK3 fusion protein and incubated for 30 min at 30°C to phosphorylate GSK3. The phosphorylation of GSK3 fusion proteins was evaluated by Western blotting using phospho-GSK3␣/␤ (Ser-21/9) antibodies (1:1000).
P. gingivalis Invasion Assay-Ca9-22 cells were treated with or without CytD (37°C for 60 min) or MCD (37°C for 60 min) at the indicated concentrations. After treatment with inhibitors, the cells were infected with P. gingivalis at an MOI of 100 at 37°C for 2 h under an atmosphere of 5% CO 2 . After infection, external nonadherent bacteria were removed by washing with PBS at least twice, and external adherent bacteria were killed with 200 g/ml metronidazole and 300 g/ml gentamycin in MEM␣ for 1 h. The cells treated with the two antibiotics were washed with PBS three times, and lysed by adding 1 ml of sterile water and incubated at 37°C. The lysed extracts were diluted to 1:100 and 1:1000, and the diluted samples including internal surviving bacteria were seeded and cultured anaerobically on blood agar plates for 7-10 days. The colonies of surviving P. gingivalis were counted and averaged from each of the diluted samples and quantified as the ratio of colony-forming units of P. gingivalis in each condition.
Expression of Bad, Akt, and PI3K p110␣-Ca9-22 cells were transiently transfected with pEF4C-His empty vector, pEF4C-His-human Bad, p3ϫFLAG-CMV, p3ϫFLAG-CMV-wild-type Akt (WT-Akt), or p3ϫFLAG-CMV-myr-Akt (constitutively active form of Akt) using Lipofectamine 2000 (Invitrogen). In brief, following incubation for 5 h, the medium was replaced, and fresh medium containing 10% FCS was added, followed by 24 h of incubation. The levels of proteins expressed from each gene were determined by Western blotting using the specific antibodies described above. The transfected cells were infected with P. gingivalis at 37°C for the indicated times and then were lysed. The lysates were analyzed with SDS-PAGE and Western blotting with the indicated antibodies as shown in each figure. Alternatively, the constitutively active form of PI3K p110␣ (myr-p110␣) tagged with HA was expressed by infecting Ca9-22 cells with retroviral vectors encoding these proteins. Briefly, the retroviral vectors pBABE-puro and pBABE-puro-PI3K myr-p110␣ (Addgene plasmid 12522) were co-transfected into human embryonic kidney 293T cells (20 ϫ 10 5 cells) by using the retrovirus packaging kit Ampho (Takara), according to the manufacturer's instructions. The retroviral vectors produced from 293T cells were harvested 48 h after transfection and filtered through 0.45-m pore size filters (Sartorius) to eliminate any remaining 293T cells. On the day prior to infection, Ca9-22 cells (1 ϫ 10 5 cells) were seeded in six-well plates. The viral supernatant was added at a ratio of 1:1 to the culture medium in the presence of 8 g/ml Polybrene, and the cells were spun at 1300 rpm for 120 min to increase the infection rate. The infected cells were incubated with the retroviruses for 24 h. A second infection was performed following the same protocol on the next day. After an additional 24 h of recovery in normal medium, the cells were passaged and selected using 1 g/ml puromycin.
Pulldown Assay for Biotin-labeled Membrane Proteins-Proteins on the plasma membrane of Ca9-22 cells (30 ϫ 10 5 cells) were biotin-labeled in PBS containing 0.5 mM CaC1 2 and 0.5 mM MgC1 2 (PBS-CaMg), pH 8.6, at 4°C for 30 min using an ECL protein biotinylation module (GE Healthcare), according to the manufacturer's instructions. After incubation, the biotin-labeled cells were solubilized with lysis buffer (50 mM Tris-HCl, pH 7.5, containing 150 mM NaCl, 20% glycerol, 1% Triton X-100, 1 mM Na 3 VO 4 , 50 mM NaF, 1 g/ml leupeptin, 1 g/ml aprotinin, and 1 mM PMSF) and incubated on ice for 30 min. The lysates were centrifuged at 20,000 ϫ g for 20 min, and the supernatants were transferred to new tubes on ice. The supernatants were then mixed with streptavidin-Sepharose beads (GE Healthcare) and incubated for at least 1 h. The streptavidin-Sepharose beads were collected by centrifugation and washed with lysis buffer three times. The precipitants were analyzed with SDS-PAGE and Western blotting using the indicated primary antibodies.
Extraction of Membrane Fraction-Ca9-22 cells were infected with P. gingivalis WT or KDP136 for 4 h and subsequently harvested with a cell scraper, washed with cold PBS, and utilized for preparation of membrane fractions using a subcellular protein fractionation kit for cultured cells (Thermo Scientific) according to the manufacturer's instructions. Briefly, the cells were collected and lysed, and the insoluble fraction was treated with membrane extraction buffer. The membrane fraction prepared using ME buffer contains the plasma membrane, mitochondria, and endoplasmic reticulum/Golgi membrane components but not nuclear membrane. The membrane fraction was analyzed using SDS-PAGE and Western blotting to detect the indicated proteins by specific antibodies.
Statistical Analysis-To establish the significance of the results, the Student's t test was used for numerical data. Fisher's exact test or 2 test was used for categorical data as appropriate. A p value of less than 0.05 was considered statistically significant.

P. gingivalis Dephosphorylates Akt at Thr-308 and Ser-473-
The PI3K/Akt signaling pathway is activated in response to insulin and growth factors (e.g. EGF), and this subsequently affects its downstream substrates, which regulate physiological functions such as survival/apoptosis and metabolism. To determine whether P. gingivalis has an influence on the PI3K/Akt signaling pathway, Ca9-22 cells were incubated with P. gingivalis. We unexpectedly observed that live P. gingivalis, but not heat-killed P. gingivalis, induced dephosphorylation of Akt at Thr-308 and Ser-473 in an MOI-dependent and infection time-dependent manner (Fig. 1, A and  B). The expression levels of total Akt remained constant over all time points, indicating that the decrease in Akt phosphorylation was due not to changes in the expression levels of Akt but to changes in its phosphorylation levels induced by P. gingivalis infection. Consistent with this result, immunofluorescence analysis showed that Akt dephosphorylation at Ser-473 occurred in P. gingivalis-infected cells, but not in control cells (Fig. 1C).
To further characterize the potential mechanisms of P. gigivalisinduced Akt dephosphorylation, P. gingivalis infection and the addition of culture supernatant from P. gingivalis was performed in cells in a Transwell assay to assess the requirement for direct interaction between bacteria and host cells. As shown in Fig. 1 (D and E), Akt dephosphorylation induced by P. gingivalis did not require a direct interaction of the epithelial cells with bacteria; P. gingivalis cell-free culture supernatant was sufficient for Akt dephosphorylation. These results in Fig. 1 indicate that P. gingivalis-induced Akt dephosphorylation was caused by P. gingivalis-secreted proteins.

P. gingivalis Induces Akt Inactivation and Subsequent
Dephosphorylation of Downstream Proteins of Akt-To confirm the kinase activity of Akt, the dephosphorylation of which was induced by P. gingivalis, we used an in vitro assay to evaluate the Akt protein and the phosphorylation of Akt substrates in the PI3K/Akt signaling pathway. Akt protein was precipitated with anti-Akt antibodies from cell extracts of P. gingivalis-infected cells and then incubated with a fusion protein containing the region corresponding to Akt-targeted residues surrounding GSK3␣/␤ (Ser-21/9). Akt precipitated from noninfected cells phosphorylated the GSK3 fusion protein, whereas Akt from the infected cells did not ( Fig. 2A). In addition, the experiment using anti-Ser(P)/Thr(P) Akt substrate antibody revealed that the phosphorylation levels of a number of Akt substrates were directly affected by P. gingivalis infection within 4 h (Fig. 2B). These affected Akt substrates may include GSK3␣/␤, Bad, and caspase-9. To clarify this point, we examined the influence of P. gingivalis on the phosphorylation levels of GSK3␣/␤, mTOR,  FEBRUARY 20, 2015 • VOLUME 290 • NUMBER 8

Gingipains Negatively Regulate PI3K/Akt Signaling Pathway
and Bad. As shown in Fig. 2C, the phosphorylation levels of GSK3␣/␤ at Ser-21/9, which is a direct substrate of Akt, and mTOR at Ser-2448, which is a downstream protein of Akt, decreased until 8 h postinfection. This result is consistent with Akt dephosphorylation and subsequent inactivation of its kinase activity (Fig. 2C). The Ser-136 residue of Bad, a proapoptotic factor and direct Akt substrate, was also dephosphorylated by P. gingivalis in Ca9-22 cells (Fig. 2D). These results indicate that P. gingivalis-induced Akt dephosphorylation at Thr-308 and Ser-473 results in the inactivation of Akt kinase activity in P. gingivalis-infected Ca9-22 gingival epithelial cells.
P. gingivalis-induced Akt Inactivation Is Independent of P. gingivalis Invasion-P. gingivalis is capable of invading gingival epithelial cells and fibroblasts (13,42,43), and the virulence factors produced by P. gingivalis including outer membrane vesicles, which enter into host cells (44), and SerB, which is a serine phosphatase and contributes to P. gingivalis invasion (45), also interacts with host proteins in gingival epithelial cells. Lipid rafts are also involved in P. gingivalis invasion (42,46). To investigate whether Akt dephosphorylation is caused by invading P. gingivalis or endocytosed virulence factors, we employed CytD and MCD. CytD is an inhibitor of actin polymerization, which is mainly involved in macropinocytosis, whereas MCD disrupts the integrity of cholesterol-enriched microdomains such as lipid rafts and caveolae, resulting in inhibition of lipid raft-and caveolae-dependent endocytosis. Following treatment with each inhibitor, Ca9-22 cells were infected with P. gingivalis at an MOI of 100, and internalized P. gingivalis was estimated by the number of colony-forming units produced from waterlysed Ca9-22 cells. Invasion of P. gingivalis into Ca9-22 cells was decreased to ϳ20% of control by treatment of CytD and MCD (Fig. 3A). However, these inhibitors did not prevent Akt dephosphorylation by P. gingivalis under the same conditions (Fig. 3B). These data indicate that invading P. gingivalis cells or endocytosed virulence factors from P. gingivalis are not associated with Akt dephosphorylation by P. gingivalis.
Gingipains Are Involved in the Inhibition of Akt Signaling by P. gingivalis-P. gingivalis-induced Akt inactivation was independent of P. gingivalis invasion as shown in Fig. 3. We next  suspected that gingipains, which are cysteine proteases, could be responsible for dephosphorylation of Akt because they are well known as major virulence factors of P. gingivalis. To demonstrate this possibility, we inhibited the proteolytic activity of gingipains using the gingipain-specific inhibitors KYT-1 and KYT-36. We observed that neither KYT-1 nor KYT-36 alone was sufficient to inhibit P. gingivalis-induced Akt dephosphorylation in Ca9-22 cells, but the combination of KYT-1 and KYT-36 did inhibit P. gingivalis-induced Akt dephosphorylation (Fig. 4A). To ascertain whether gingipains are associated with P. gingivalis-induced Akt dephosphorylation, we compared Akt dephosphorylation with both wild-type P. gingivalis (WT-Pg) and gingipain-deficient KDP136. Although WT-Pg decreased the phosphorylation levels of Akt as heretofore described, KDP136 did alter its phosphorylation for up to 240 min (Fig. 4B). As for the Akt downstream proteins GSK3␣/␤ and mTOR, although both of them were also dephosphorylated by WT-Pg, infection with KDP136 and treatment with gingipain-specific inhibitors showed no significant differences in their phosphorylation (Fig. 4C). These results indicate that gingipains play a crucial role in the dephosphorylation of Akt, GSK3␣/␤, and mTOR in P. gingivalis infection and that their proteolytic activity is indispensably involved in this event.
Gingipains Cleave Membrane Proteins Coupled to PI3K p85␣-Based on the results of Figs. 3 and 4, we hypothesized that gingipains act from the outside, not inside, of epithelial cells and cleave membrane proteins, resulting in inactivation of PI3K and dephosphorylation of Akt. To confirm this hypothesis, total  FEBRUARY 20, 2015 • VOLUME 290 • NUMBER 8

JOURNAL OF BIOLOGICAL CHEMISTRY 5195
membrane proteins of Ca9-22 cells were labeled with biotin and precipitated with streptavidin-conjugated Sepharose beads in an attempt to detect PI3K regulatory subunit p85␣ bound to membrane proteins. As shown in Fig. 5A, the biotinylated membrane proteins were precipitated from uninfected Ca9-22 cells. In cells infected with P. gingivalis, PI3K p85␣ did not coprecipitate from Ca9-22 cells infected for 4 h with WT-Pg, as compared with control and KDP136-infected cells at all time points (Fig. 5B). In contrast, ␤-catenin and integrin-linked kinase (47), which bind to E-cadherin and integrins, respectively, were unaffected under all conditions (Fig. 5B). In Fig. 5C, we used CytD to examine the possibility of an intracellular action by P. gingivalis invasion on the disruption of p85␣-membrane protein associations. CytD did not inhibit the disruption induced by WT-Pg. These results suggest that gingipains from WT-Pg disrupt the interaction between membrane proteins and PI3K p85␣ by their extracellular action.
Inactivation of PI3K Is Caused by Gingipains-Based on the results shown in Fig. 5, we hypothesized that PI3K activity might be affected by gingipains. We challenged Ca9-22 cells with WT-Pg and KDP136, and immunoprecipitates of PI3K under each condition were used for assessing PI3K activity. WT-Pg, but not control and KDP136, decreased the activity of PI3K by ϳ70% (Fig. 6A). Similarly, PI3K activity was decreased by 60 -70% in the presence of CytD during WT-Pg infection compared with control (Fig. 6B). In Fig. 6 (A and B), regulatory subunit p85␣ and catalytic subunit p110␣ of PI3K were detected in equal amounts in all conditions. These data indicate that PI3K activity is affected by WT-Pg infection, and gingipains are closely associated with the decrease in PI3K activity via an extracellular interaction.

Gingipains Alter the Localization of PDK1 and Akt at the
Plasma Membrane-We demonstrated a gingipain-dependent decrease in PI3K activity in Fig. 6. Next, we examined whether gingipains affect the localization of PDK1 and Akt at the plasma membrane, where they bind PI(3,4,5)P3 through their pleckstrin homology domain. As shown in Fig. 7A, 4 h after infection there was no significant change in total protein level in whole cell lysates under the indicated conditions (lower panels). WT-Pg remarkably reduced the distribution of PDK1 in the membrane fraction compared with control and KDP136-infected cells (upper panels in Fig. 7A). Interestingly, we also found that the distribution of p85␣, p110␣, and Akt increased in the membrane fraction extracted from WT-Pg-infected cells compared with noninfected and KDP136-infected cells (upper panels in Fig. 7A). In the cells treated with CytD, WT-Pg decreased the level of PDK1 and increased the level of Akt in the membrane fraction to similarly those of untreated cells infected by WT-Pg (upper panels in Fig. 7B). The total protein expression level was not changed (lower panels in Fig. 7, A and B). Caveolin-1 also served as a protein loading control in both Fig.  7 (A and B). Here we demonstrate that gingipain-induced PI3K inactivation changed the localization of PDK1 and Akt at the plasma membrane, suggesting that the changes are strongly implicated in a decrease in Akt phosphorylation because of the low level of PDK1 localization and subsequent blockade of the Akt signaling cascade.
Effect of Constitutively Active PI3K p110␣ and Akt on the Localization of PDK1 and Akt, and Akt Signaling in P. gingivalis-infected Cells-To support the above results, we tested the effect of expression of constitutively active forms of PI3K p110␣ and Akt on the localization of PDK1 and Akt and the phosphor- ylation of GSK3␣/␤ and mTOR by P. gingivalis infection. For this purpose, we utilized genes encoding a myristoylation signal to produce constitutively activated proteins independently of all upstream signaling events. We showed that the level of PDK1 and Akt in the membrane fraction was unchanged from 0 to 4 h post infection in the cells expressing myr-p110␣, whereas the localization of PDK1 was decreased, and that of Akt was increased by P. gingivalis infection in the cells transfected with empty vector (Fig. 8A). In myr-p110␣-expressing cells, the phosphorylation levels of Akt, GSK3␣/␤, and mTOR were sustained during infection compared with empty vector-transfected control cells (Fig. 8B). The expression of WT-Akt or myr-Akt also blocked the dephosphorylation of GSK3␣/␤ and mTOR by P. gingivalis infection (Fig. 8C). Based on these observations, we support that gingipains reduce the kinase activity of PI3K and affect the translocation of PDK1 and Akt at the plasma membrane and the phosphorylation level of the downstream proteins on the PI3K/Akt signaling pathway by extracellular interaction between gingipains and Ca9-22 cell surface during P. gingivalis infection.
Dysregulation of the PI3K/Akt Signaling Pathway by Gingipains in Human Primary Gingival Epithelial Cells-All of the experiments described above were performed with a cultured cancer cell line. We next investigated whether these signaling events also occur after P. gingivalis infection in HGEP cells. Western blotting, immunofluorescence analysis, and Akt kinase assay revealed that Akt was dephosphorylated and inactivated by infection with WT-Pg, but not KDP136, in HGEP cells (Fig. 9, A-C). We also examined the changes in phosphorylation levels of GSK3, mTOR, and Bad by P. gingivalis infection in HGEP cells. Infection with WT-Pg, but not KDP136, decreased the phosphorylation of GSK3␣/␤, mTOR, and Bad in   FEBRUARY 20, 2015 • VOLUME 290 • NUMBER 8

JOURNAL OF BIOLOGICAL CHEMISTRY 5197
HGEP cells (Fig. 9D). Next, we demonstrated that PI3K activity was decreased by ϳ70% by gingipains during P. gingivalis infection (Fig. 9E). To evaluate PI (3,4,5)P3 production, PDK1 translocation to the plasma membrane was examined, and these results indicate that there was a negligible amount of PDK1 in the membrane fraction in WT-Pg-infected cells compared with control and KDP136-infected cells (Fig. 9F). Taken together, these data show that alteration of the PI3K/Akt signaling pathway by gingipains occurs in HGEP cells in a manner similar to that in Ca9-22 cells.

DISCUSSION
In this study, we reveal a novel mechanism of attenuation of the PI3K/Akt signaling pathway by gingipains, the cysteine proteases of P. gingivalis. The proteolytic effects of gingipains play an essential role in the dysfunction of PI3K and changes in the Akt signaling pathway during P. gingivalis infection, independent of invasion (Fig. 10). The results elucidate a new role for gingipains in disturbing cell functions regulated by the PI3K/ Akt signaling pathway, including cell survival and growth, apoptosis, endocytosis, and metabolism.
Here, we demonstrate that P. gingivalis attenuates PI3K and Akt and decreases the phosphorylation levels of the Akt downstream proteins GSK3, mTOR, and Bad, in gingival epithelial cells. However, other groups have reported the activation of PI3K and Akt by P. gingivalis infection (30,31,43,48). These apparently contradictory results may be due to differences in cell type and experimental conditions such as infection time, strains, and culture conditions of P. gingivalis. For example, Chang et al. (43) have shown that high glucose treatment of P. gingivalis increases expression of fimA mRNA. Moreover, phenotypic differences might exist in the gingival epithelial cells used here, such as in the expression level and pattern of receptors and adaptor proteins in the PI3K/Akt signaling pathway, compared with the cells used in other studies. In addition, it is plausible that these cells may have originated from a different area of gingival tissue and/or from individuals of various health conditions and ages.
We demonstrate the possibility that inactivation of PI3K and Akt by P. gingivalis is not required for its direct interaction with cell surface (Fig. 1, D and E). Furthermore, our experiments using CytD and MCD show that this inactivation is independent of P. gingivalis invasion into gingival epithelial cells and intracellular interaction with virulence factors such as gingipains, SerB (45), and outer membrane vesicles (49) produced by P. gingivalis (Figs. 3 and 6). We also clarified the inactivation of Akt and the alteration of phosphorylation levels of its substrate proteins including Bad, GSK3, and mTOR (Figs. 2 and 8). Inhibitors of gingipains and infection with the gingipain-null mutant KDP136 revealed that gingipains are closely associated with P. gingivalis infection-mediated Akt dephosphorylation (Fig. 4). Thus, the alteration of the PI3K/Akt signaling pathway by gingipains from P. gingivalis have several effects on PI3K/ Akt-dependent cellular functions as discussed below. First, the dephosphorylation of Bad resulting from gingipain-induced Akt inactivation, which leads to increased activity of Bad, may contribute to apoptosis during P. gingivalis infection, as suggested by a previous report (37). GSK3 is known to regulate glycogen synthesis in response to insulin. Recently, Ishikawa et al. (38) indicated that P. gingivalis has an effect on glycogen synthesis in liver via the Akt/GSK3␤ signaling pathway. Our results support that gingipains are implicated in the attenuation of the PI3K/Akt/GSK3␤ pathway and decrease in glycogen synthesis and glucose metabolism. It may be worth noting in the current study, however, that all experiments were performed using gingival epithelial cells and not hepatocytes, which are highly specialized for the biosynthesis of glycogen. We also show that mTOR is dephosphorylated by gingipains. The PI3K/ Akt/mTOR signaling pathway has a role in endocytosis (50, 51) and internalization of bacteria such as Salmonella typhimurium, Helicobacter pylori, Streptococcus pneumoniae, Staphylococcus aureus (52)(53)(54)(55), and autophagy (56 -58). Some reports have shown a causal relationship between the Akt/mTOR signaling pathway and P. gingivalis and gingipains (59,60), suggesting that gingipains might cause a decrease in mTOR function via attenuation of the PI3K/Akt signaling pathway.
Here we describe a novel mechanism by which gingipainmediated inactivation of PI3K is associated with the disruption of complex formation between PI3K p85␣ and membrane proteins by wild-type P. gingivalis infection, but not the gingipaindeficient mutant KDP136, leading to alteration of the localization of PDK1 and Akt and decrease in the phosphorylation and activity of Akt. At first, we were wondering whether the amount of PI (3,4,5)P3 was reduced by gingipain-induced PI3K inactivation, leading to decrease in PDK1 localization at the plasma membrane. However, Akt was apparently increased by P. gingivalis infection. Therefore, we consider that there is little change FIGURE 9. Gingipains attenuate the PI3K/Akt signaling pathway in HGEP cells infected with P. gingivalis. HGEP cells were infected with or without P. gingivalis WT or KDP136 at an MOI of 100 for 4 h (A and C-F) or 2 h (B). After infection, the cells were lysed or fixed, and then the samples were analyzed by SDS-PAGE and Western blotting (A and D), immunostaining for phosphorylated Akt (B), in vitro kinase assay for Akt (C) and PI3K (E), and membrane extraction for translocation of PDK1 (F) as described under "Experimental Procedures." The antibodies were total protein-and phospho-specific as indicated. These results were independently demonstrated three times. Statistical significance: *, p Ͻ 0.05. cont., control; IP, immunoprecipitation.
in the amount of PI (3,4,5)P3 at the plasma membrane during infection, although the turnover and newly synthesis of PI(3,4,5)P3 might be decreased by gingipain-mediated PI3K inactivation. Furthermore, it is likely that residual levels of PI (3,4,5)P3 are sufficient to interact with Akt, which is mainly nonphosphorylated form and/or the form after working in cytosol. Consequently, we suspect the possibility that Akt occupies the residual PI(3,4,5)P3, and Akt accumulated at the plasma membrane is ready and waiting for the phosphorylation by PDK1. This mean brings us to the point that the accumulation of Akt prevents PDK1 translocation to the plasma membrane via the pleckstrin homology domain of PDK1. Actually, based on the data in Figs. 7 and 8, the level of PDK1 at the plasma membrane is sufficiently low to phosphorylate Akt at Thr-308 in WT-Pg-infected cells compared with in control cells and in KDP136-infected cells, which fail to normally transmit the signals from PDK1 to Akt. We evidently show that gingipains reduce the PI3K activity, PDK1 translocation at the plasma membrane, and the phosphorylation level of Akt and the downstream proteins on Akt signaling. These observations are supported by expression of constitutively active forms of PI3K p110␣ and Akt (Fig. 8).
According to the results in Fig. 5B, ␤-catenin and integrinlinked kinase were detected, and E-cadherin and integrin ␤4 remain unaffected by gingipains within 4 h of infection. Although EGFR and integrin ␤1 are cleaved, we confirmed that EGFR does not associate with PI3K p85␣ in the absence of serum and growth factors (data not shown). We suggest that gingipains cleave one or more membrane proteins that bind PI3K p85␣, resulting in reduction in both PI3K activity and phosphorylation of Akt. We are currently attempting to characterize the as yet unidentified membrane proteins that bind to PI3K p85␣. The PI3K/Akt signaling pathway is intrinsically regulated by PTEN (phosphatase and tensin homologue deleted on chromosome 10) and protein serine/threonine phosphatases, such as protein phosphatases 2A, 2B, and 1 (28,61,62). However, our results using siRNA for PTEN and protein phosphatase inhibitors (data not shown) suggest virtually no association between PI3K/Akt and these phosphatases. Moreover, there was little change in the interaction between Akt and C-terminal modulator protein by P. gingivalis infection (data not shown) (63,64). It is therefore likely that gingipain-induced attenuation of PI3K and Akt is mediated by interference with homeostatic PI3K signaling via cleavage of membrane proteins in complexes with PI3K p85␣.
In conclusion, our data provide strong evidence that gingipains are a critical virulence factor for disrupting cellular functions regulated by the PI3K/Akt signaling pathway in gingival epithelial cells and almost certainly cause the damage to the gingival epithelial barrier. These effects promote the invasion and colonization of gingival tissue by P. gingivalis and can lead to long term infection. Future investigations will clarify additional functions of gingipains and the significance of the PI3K/Akt signaling pathway in P. gingivalis infection.