Fas-associated factor 1 mediates NADPH oxidase-induced reactive oxygen species production and proinflammatory responses in macrophages against Listeria infection

Fas-associated factor 1 is a death-promoting protein that induces apoptosis by interacting with the Fas receptor. Until now, FAF1 was reported to interact potentially with diverse proteins and to function as a negative and/or positive regulator of several cellular possesses. However, the role of FAF1 in defense against bacterial infection remains unclear. Here, we show that FAF1 plays a pivotal role in activating NADPH oxidase in macrophages during Listeria monocytogenes infection. Upon infection by L. monocytogenes, FAF1 interacts with p67phox (an activator of the NADPH oxidase complex), thereby facilitating its stabilization and increasing the activity of NADPH oxidase. Consequently, knockdown or ectopic expression of FAF1 had a marked effect on production of ROS, proinflammatory cytokines, and antibacterial activity, in macrophages upon stimulation of TLR2 or after infection with L. monocytogenes. Consistent with this, FAF1gt/gt mice, which are knocked down in FAF1, showed weaker inflammatory responses than wild-type mice; these weaker responses led to increased replication of L. monocytogenes. Collectively, these findings suggest that FAF1 positively regulates NADPH oxidase-mediated ROS production and antibacterial defenses.


Introduction
To better understand the biological role of FAF1, we examined the relationship between NADPH oxidase and FAF1 in host defense against bacterial infection. We show that FAF1 is a crucial positive regulator of the phagocytic NADPH oxidase complex, which promotes ROS production by macrophages in response to L. monocytogenes infection. FAF1 controls phagocytic NADPH oxidase-mediated inflammatory responses upon L. monocytogenes infection by interacting with p67phox, thereby inhibiting bacterial growth.

Knockdown of FAF1 weakens antibacterial immune responses both in vivo and in vitro
Based on a previous study of the role of FAF1 in antiviral responses against infection by RNA virus [27], we asked whether FAF1 is also involved in responses to bacterial infection. To examine in vivo host responses to infection by L. monocytogenes, FAF1 +/+ and FAF1 gt/gt mice were infected intraperitoneally with L. monocytogenes (5 × 10 5 CFU per mouse) and serum cytokine levels, bacterial load, and proinflammatory gene expression in the spleen and liver were measured at 24 h post-infection (hpi). The bacterial load in the spleen and liver of FAF1 gt/gt mice was approximately 10-fold and 3-fold higher, respectively, than that in FAF1 +/+ mice (Fig 1, panel A). Serum cytokine levels were also reduced markedly in FAF1 gt/gt mice ( Fig  1, panel B). Expression of proinflammatory genes in the spleen and liver of FAF1 gt/gt mice was also lower than that in FAF1 +/+ mice (Fig 1, panels C-D). To determine whether these effects were mediated by peritoneal macrophages in response to L. monocytogenes infection, FAF1 +/+ and FAF1 gt/gt mice were infected with L. monocytogenes intraperitoneally and peritoneal macrophages (PMs) were isolated at 24 hpi. Expression of proinflammatory cytokines and chemokines by these cells was measured by quantitative real-time PCR. Expression of mRNA encoding IL-6, CXCL10, and RANTES was significantly lower in PMs from FAF1 gt/gt mice than in those from FAF1 +/+ mice (S1 Fig). These data suggest that FAF1 plays an important role in host defense against L. monocytogenes infection in vivo. Next, to determine whether FAF1 is involved in inflammatory responses, we examined induction of FAF1 in macrophages exposed to L. monocytogenes. Importantly, mRNA expression of FAF1 was induced with 2-3 folds at 15 or 60 m post-infection (mpi) with L. monocytogenes in Raw264.7 cells, suggesting that FAF1 responds to bacterial infection at early time (S2 Fig, panel A). Also, FAF1 responded to high MOI of bacterial infection in mouse bone marrow-derived macrophages (BMDMs) and Raw264.7 cells (S2 Fig, panels B-C). At early time points after L. monocytogenes infection, host defense is regulated by secretion of several cytokines and chemokines, including IL-6, IL-12, and RANTES [30,31]. Therefore, we examined the effect of FAF1 on proinflammatory cytokine secretion by BMDMs and resident PMs against L. monocytogenes infection or TLR2 ligands. First, expression of FAF1 was confirmed by immunoblotting of extracts from BMDMs or PMs isolated from FAF1 +/+ and FAF1 gt/gt mice (S3 Fig, panels A-B). Next, cells were infected with L. monocytogenes or treated with zymosan or bacterial lipoprotein (BLP). The supernatants were harvested at 12 or 24 hpi to measure IL-6 and IL-12 by ELISA. As results, FAF1 knockdown reduced cytokine production by both BMDMs (Fig 1, panels E-F) and PMs (S3 Fig, panels C-D). Next, we generated control and FAF1-knockdown murine macrophage cells (Raw264.7) using lentiviruses harboring non-specific or FAF1-specific small hairpin RNA (shRNA). Expression of FAF1 was then determined by immunoblot analysis (S4 Fig, panel A). Next, cells were infected with L. monocytogenes or treated with zymosan or BLP and supernatants harvested at 12 or 24 hpi to measure cytokine secretion. Consistent with the results obtained from primary cells, shRNA-mediated knockdown of FAF1 led to a marked reduction in secretion of proinflammatory cytokines IL-6 and IL-12 upon stimulation via TLR2 (Fig 1, panels G-H). Additionally, production of chemokines regulated upon activation (i.e., normal T cell expressed and secreted (RANTES) and monocyte chemoattractant protein (MCP)-1) was lower in FAF1-knockdown cells than in control cells (S4 Fig, panels B-C). Collectively, these results suggest that FAF1 expression has a marked effect on host defense responses against L. monocytogenes. and FAF1 gt/gt mice were infected intraperitoneally with L. monocytogenes (5.0 × 10 5 CFU/mouse). The bacterial load in spleens and livers from infected mice (A), serum cytokine levels (B), mRNA levels for cytokines in spleens (C) and livers (D) were measured at 24hr post-infection (n = 9-13 per group). � P < 0.05, �� P <0.01 (Mann-Whitney test). Data are the mean ± SEM of individual values pooled from two independent experiments. (E and F) FAF1 +/+ and FAF1 gt/gt BMDMs were infected with L. monocytogenes (MOI = 1) for 2hrs or treated with BLP (100ng/ ml) or zymosan (100μg/ml) for the indicated times and the supernatants were analyzed for levels of IL-6 (E), IL-12 (F). (G and H) Control Raw264.7 cells (shRNA-NS) and FAF1 knockdown Raw264.7 cells (shRNA-FAF1) were infected with L. monocytogenes (MOI = 1) for 2hrs or treated with BLP (100ng/ml) or zymosan (100μg/ml) for the indicated times and the supernatants were analyzed for levels of IL-6 (G), IL-12 (H). Data are representative of three (E-H) independent experiments. Error bars, mean ± SD. � P < 0.05, �� P < 0.01 (Student's t-test), compared to FAF1 +/+ .

Knockdown of FAF1 attenuates proinflammatory responses during L. monocytogenes infection to macrophages
To determine whether FAF1 activates proinflammatory signaling pathways, we performed immunoblot analysis to examine expression of activated forms of molecules related to the NF-κB and MAPK signaling pathways in BMDMs isolated from FAF1 +/+ and FAF1 gt/gt mice infected with L. monocytogenes. The result showed that knockdown of FAF1 leads to a marked reduction in phosphorylation of p65 (NF-κB), IκBα, and SAPK/JNK, but had no effect on activation of p38 MAPK or Erk1/2 (Fig 2, panel A). In agreement with this, knockdown of FAF1 in Raw264.7 cells suppressed activation of p65 (NF-κB), IκBα, and SAPK/JNK, but not p38 or Erk1/2 MAPK (Fig 2, panel B). Furthermore, to measure expression of proinflammatory genes, BMDMs and resident PMs isolated from FAF1 +/+ and FAF1 gt/gt mice were infected with L. monocytogenes and subjected to real-time PCR at 12 hpi to detect Il6, Nos2 (iNOS), Ptgs2 (COX-2), Cxcl10, and RANTES. The expression of these genes in FAF1-knockdown BMDMs cells were lower than those in wild-type cells (Fig 2,

Interaction between FAF1 and p67phox leads to enhanced ROS production and bacterial clearance upon infection by L. monocytogenes in macrophages
To identify the target protein of FAF1, large-scale cultured HEK293 cells were used for immunoprecipitation with an anti-FAF1 antibody, followed by mass spectrometry analysis. The result identified NADPH oxidase activator 1 (NoxA1) (S6 Fig). NoxA1 regulates activation of Nox1, which can generate ROS and is expressed at high levels in colon cancer cells [3]. As a homolog of NoxA1, p67phox acts mainly as an activator of Nox2 in phagocytes. Moreover, there is a high degree of domain homology between NoxA1 and p67phox, although the proteins show only 28% amino acid identity [3,32,33]. Based on these reports, we asked whether FAF1 interacts with p67phox to regulate ROS production in phagocytes. Mock-or L. monocytogenes-infected Raw264.7 cells were harvested at various time points and cell lysates were immunoprecipitated with an anti-FAF1 antibody, followed by immunoblotting with antibodies against components of the phagocytic NADPH oxidase complex (Fig 3, panel A). The result indicated that FAF1 transiently interacts with p67phox, p47phox, and p40phox at 30 mpi with L. monocytogenes (Fig 3, panel A). To test whether FAF1 expression affects those interaction, efficiency of FAF1-specific siRNA was priorly determined in Raw264.7 cells or BMDMs for further experiments (S7 Fig, panels A-B). FAF1 knockdown-BMDMs showed a weak interaction between both molecules as well as lower expression of p67phox compared with control BMDMs upon L. monocytogenes infection (S8 Fig, panel A). Similar result was obtained following siRNA-mediated knockdown of FAF1 in Raw264.7 (S8 Fig, panel B). Moreover, confocal microscopy analysis exhibited that FAF1 is translocated to phagosomal membranes upon zymosan treatment in BMDMs where it co-localizes with p67phox (Fig 3, panel B). Additionally, immunoprecipitation with an anti-V5 antibody using FAF1-overexpressing Raw264.7 cells showed strong interaction between ectopic FAF1 and endogenous p67phox without stimulation, suggesting that FAF1 might present high affinity to p67phox (Fig 3, panel C). Next, we performed CFU assay to examine the growth rate of intracellular L. monocytogenes in BMDMs following siRNA-mediated knockdown of FAF1. Knockdown of FAF1 showed a significant increase of bacterial growth compared to control BMDMs (Fig 3, panel D). This result encouraged us to verify that reduced ROS production and inflammation in FAF1-knockdown cells might result in a more favorable environment for L. monocytogenes replication. To investigate whether FAF1 affects ROS production upon L. monocytogenes infection, we used fluorescence absorbance to measure ROS production following siRNA-mediated knockdown of FAF1 in BMDMs. As expected, H 2 O 2 and O 2 -produced by FAF1-knockdown BMDMs were significantly lower than those produced by control BMDMs in response to L. monocytogenes infection (Fig 3, panel E). Similar results were obtained from Raw264.7 cells following siRNAmediated knockdown of FAF1 (Fig 3, panel F). Moreover, knockdown of FAF1 exhibited lower ROS production upon stimulation with zymosan in BMDMs, Raw264.7, and PMs compared to control cells. (S9 Fig, panels A-C). ROS acts as a signal transduction mediator in response to diverse stimuli [5]. To determine further whether proinflammatory cytokine production in response to L. monocytogenes infection or TLR2 signaling correlated directly with ROS generation in the presence or absence of FAF1, we measured NO and IL-6 levels in and FAF1 gt/gt BMDMs were infected with L. monocytogenes (MOI = 1) for the indicated times, and subjected to immunoblotting with phosphorylated and total forms of p65, IκBα, JNK, p38, p42/p44 MAPK, FAF1 and β-actin (left). Band intensity of each blot was shown in graphs (Right). (B) Control Raw264.7 cells (shRNA-NS) and FAF1 knockdown Raw264.7 cells (shRNA-FAF1) were infected with L. monocytogenes (MOI = 1) for the indicated times, and then subjected to immunoblotting with the phosphorylated and total forms of p65, IκBα, JNK, p38, p42/p44 MAPK, FAF1 or β-actin (left). Band intensity of each blot was shown in graphs (Right). (C) FAF1 +/+ and FAF1 gt/gt BMDMs were infected with L. monocytogenes (MOI = 3) for 2hrs. At 12 hpi, cells were harvested and analyzed by quantitative real-time PCR analysis for IL-6, iNOS, COX-2, CXCL10, and RANTES genes. (D) Control Raw264.7 cells (shRNA-NS) and FAF1 knockdown Raw264.7 cells (shRNA-FAF1) were infected with L. monocytogenes (MOI = 1) for 2hrs. At 12 hpi, cells were harvested and analyzed by quantitative real-time PCR analysis for IL-6, iNOS, COX-2, CXCL10, and RANTES genes. Data are representative of two (A, B) or three (C, D) independent experiments. Error bars, mean ± SD. � P < 0.05, �� P < 0.01 (Student's t-test), compared to control (FAF1 +/+ or shRNA-NS).   monocytogenes-infected Raw264.7 cells were harvested at indicated times and cell lysates were used for immunoprecipitation (IP) with an anti-FAF1 antibody, followed by immunoblotting with anti-p67phox, p47phox, p40phox, and FAF1 antibodies. (B) BMDMs were treated with mock or zymosan (100μg/ml) for 30min, followed by confocal microscopy with an anti-p67phox antibody and an anti-FAF1 antibody. Scale bars represent 10μm. (C) Raw264.7 cells containing empty vector or V5-tagged FAF1 were used for IP with an anti-V5 antibody, followed by immunoblotting with anti-p67phox, p47phox antibodies. (D) At 36hr post-transfection with siRNA-control or siRNA-FAF1, BMDMs were infected with L. monocytogenes (MOI = 0.1) for 1hr. Bacteria-containing medium was replaced with fresh medium containing gentamycin. Then, cells were washed with PBS and lysed by Triton X-100 at the indicated time, followed by CFU titration for counting intracellular bacteria. (E-F) At 36hr post-transfection with siRNA-control or siRNA-FAF1, BMDMs (E) or Raw264.7 cells (F) were infected with L. monocytogenes, then used for measuring H 2 O 2 or O 2 production. (G) FAF1 +/+ and FAF1 gt/gt BMDMs were infected with L. monocytogenes (MOI = 1) for 2hrs or stimulated with zymosan w/ or w/o NAC or DPI (20μM). At 18hr, the supernatants were analyzed for NO production or IL-6 secretion. (H) Control Raw264.7 cells (shRNA-NS) and FAF1 knockdown Raw264.7 cells (shRNA-FAF1) were infected with L. monocytogenes (MOI = 1) for 2hrs or stimulated with zymosan w/ or w/o NAC or DPI (20μM). At 18hr, the supernatants were analyzed for NO production or IL-6 secretion. (I) At 36hr post-transfection with siRNA-control or siRNA-FAF1, BMDMs were infected with L. monocytogenes (MOI = 0.1) for 1hr. Bacteria-containing medium was replaced with fresh medium containing gentamycin w/ or w/o DPI or NAC. Then, cells were washed with PBS and lysed by Triton X-100 at the indicated time, followed by CFU titration for counting intracellular bacteria. Data are representative of two (A, C, D, G, H, I) or three (B, E, F) independent experiments. Error bars, mean ± SD. � P < 0.05, �� P < 0.01 (Student's t-test), compared to FAF1 +/+ or control (si-control or shRNA-NS). N.S., no significant; N.D., no detection. Secretion of IL-6 by these cells also fell markedly, regardless of FAF1 expression. In other words, impaired ROS production resulted in no significant difference in cytokine secretion by wild-type and FAF1-knockdown macrophages. Knockdown of FAF1 attenuates IκBα degradation and NF-κB activation, but not phosphorylation of IKKs, suggesting that FAF1 indirectly regulates the inflammatory responses via ROS production on TLR2 signaling (S10 Fig). It was also supported by evidences that FAF1 augments inflammatory responses depending on NADPH oxidase complex (S11 Fig, panels A-B). Likewise, knockdown of FAF1 led to increased bacterial growth than control in BMDMs. However, there was no differences under DPI or NAC treatment, which suggests that FAF1 inhibits bacterial growth by mediating ROS production (Fig 3, panel I). Taken together, these findings demonstrate that FAF1 enhances inflammatory responses and intracellular bacterial clearance via ROS generation by interacting with p67phox upon infection by L. monocytogenes.

FAF1 directly interacts with p67phox via amino acids 330-489
FAF1 contains three well characterized domains, the FID, DEDID, and C-terminal domains [21,24,29]. To define which domain of FAF1 interacts with p67phox, we generated GSTtagged domain constructs and performed GST pull-down assays in HEK293T cells. p67phox bound strongly to both the DEDID and C-terminal domains of FAF1 (Fig 4, panel A). The region of FAF1 responsible for interaction with p67phox was narrowed down to amino acids 330-489 (Fig 4, panel B). In the reverse experiment, GST-tagged domain constructs of p67phox were generated, and GST pull-down assays were performed in HEK293T cells to determine which domain of p67phox interacts with FAF1. FAF1 bound to the 4X tetratricopeptide repeat (TPR) domain of p67phox (Fig 4, panel C). A diagram of the domains mediating binding between FAF1 and p67phox is shown in panel D (Fig 4, panel D). Next, a mutant construct of FAF1 with a deletion in amino acids 330-489 was generated (Δ330-489). This mutant showed weak binding to endogenous p67phox and p47phox compared with wild-type FAF1 by immunoprecipitation with an anti-V5 antibody in Raw264.7 cells (Fig 4, panel E). Furthermore, ectopically expressed p67phox co-localized with wild-type FAF1 but not FAF1 Δ330-489 in HEK293T cells (Fig 4, panel F). Taken together, these data indicate that amino acids 330-489 of FAF1 comprise the critical region that mediates interaction with the 4X TPR domain of p67phox.

FAF1 potentiates ROS-mediated inflammatory responses in a p67phox binding-dependent manner
As described above, we identified the binding site through which FAF1 interacts with p67phox and found that a deletion mutant of this region (FAF1 Δ330-489) was unable to bind to p67phox. To determine whether FAF1 Δ330-489 lost the ability to promote inflammatory responses due to the weak interaction with p67phox, we generated Raw264.7 cells stably overexpressing a control vector, wild-type FAF1, or FAF1 Δ330-489. The expression level of FAF1 in each stable cell line was determined by western blotting (S12 Fig, panel A). The cells were then stimulated with TLR2 ligands, including L. monocytogenes, zymosan, and BLP, and levels of proinflammatory cytokines and chemokines in the supernatants were measured. As expected, cells expressing FAF1 Δ330-489 showed reduced secretion of proinflammatory cytokines and chemokines in response to TLR2 signaling compared with cells expressing wild-type FAF1 (Fig 5, panels A-B and S12 Fig, panels B-C). We next evaluated the activation of molecules involved in NF-κB and MAPK signaling in L. monocytogenes-infected Raw264.7 cells. Cells expressing wild-type FAF1 showed enhanced activation of the NF-κB and SAPK/JNK pathways upon L. monocytogenes infection, as previously observed (Fig 2, panels A-B), while activation of these signaling molecules was not altered in cells overexpressing FAF1 Δ330-489 compared with control cells (Fig 5, panel C). In addition, IL-6, O 2 , and NO production were impaired in cells overexpressing FAF1 Δ330-489 compared with cells overexpressing wildtype FAF1. Cytokine secretion was not induced in the presence of ROS inhibitors, with no significant difference between individual cell lines (Fig 5, panels D-E). While the growth of intracellular bacteria in cells overexpressing wild-type FAF1 was lower than in control cells, cells overexpressing FAF1 Δ330-489 exhibited a similar level of bacterial growth as control cells (Fig 5, panel F). These findings indicate that the effects of FAF1 on inflammatory responses, HEK293T cells were transfected with mammalian GST or GST-tagged FAF1 domain constructs together with V5-tagged p67phox, and at 48hrs, cell lysates were used for GST pull-down assay, followed by immunoblotting with an anti-V5 antibody. (B) HEK293T cells transfected with GST-FAF1 narrow-down domain constructs and V5-tagged p67phox were used for GST pull-down assay. (C) HEK293T cells were transfected with mammalian GST or GST-tagged p67phox domain constructs together with V5-tagged FAF1, and at 48hrs, cell lysates were used for GST pulldown assay, followed by immunoblotting with anti-V5 antibody. (D) Binding mapping. Schematic diagram of FAF1 and p67phox. (E) Raw264.7 cells containing empty vector or V5-tagged FAF1 wild-type or FAF1 Δ330-489 were used for IP with anti-V5 antibody, followed by immunoblotting with anti-p67phox and p47phox antibodies. (F) HEK293T cells were transfected with V5-tagged FAF1 wild-type or V5-tagged FAF1 Δ330-489 together with GST-p67phox, followed by confocal microscopy assay with anti-V5 (green) and GST (red) antibodies. Data are representative of two independent experiments. and intracellular bacterial clearance through ROS generation are dependent on its interaction with p67phox.
https://doi.org/10.1371/journal.ppat.1008004.g005 FAF1 positively regulates NADPH oxidase-mediated ROS production were infected with L. monocytogenes, and a chemiluminescence assay was performed to measure NADPH oxidase activity. The result showed that overexpression of wild-type FAF1 augments the phagocytic NADPH oxidase activity, whereas overexpression of FAF1 Δ330-489 doesn't (Fig 6, panel A). This finding suggested that amino acids 330-489 is the critical region for the interaction with p67phox, which results in increased ROS production in response to L. monocytogenes infection. We also found that expression of FAF1 leads to higher and persistent expression of p67phox in a binding-specific manner (Fig 6, panel B). Overexpression of FAF1 induces a bit more mRNA expression of p67phox and p47phox without stimulation (S13 Fig). Additionally, Raw264.7 cells overexpressing control vector, wild-type FAF1 were treated with mock or zymosan/cycloheximide and used for immunoblot analysis for expression levels of p67phox and p47phox over time. This result showed that overexpression of FAF1 increases the stability of p67phox and p47phox upon zymosan treatment (Fig 6, panel C). The intensity of the protein bands on the blot shown in Fig 6, panel D is quantitated (Fig 6, panel D). Furthermore, to check the localization of p67phox around phagosomes depending on expression level of FAF1, BMDMs were treated with zymosan particles for 30 min following siRNA-mediated knockdown of FAF1, then followed by confocal microscopy using anti-p67phox antibody ( Fig  6, panel E). As correlated with prior data, knockdown of FAF1 exhibited considerable decrease of p67phox localized to phagosomal membranes compared to control, which supports evidently that FAF1 potentiates the stability of p67phox in phagosomes. Collectively, these results suggest that FAF1 in macrophages effectively augments p67phox stability in response to infection with L. monocytogenes, resulting in increased phagocytosis-mediated NADPH oxidase activity.

Discussion
FAF1 (Fas-associated factor 1), a member of the Fas death-inducing signal complex, modulates a variety of biological processes by interacting with diverse molecules [24][25][26]. However, the role of FAF1 in host defense against bacterial infection remains unclear. Here, we report that FAF1 is a positive regulator that increases activity of the phagocytic NADPH oxidase complex, resulting in production of ROS and in activation of NF-κB signaling, inflammatory responses, and antibacterial activity upon L. monocytogenes infection. First, FAF1 gt/gt mice exhibited reduced serum cytokine levels, reduced inflammatory gene expression, and increased bacterial burden during L. monocytogenes infection. Second, primary macrophages (BMDMs and PMs) isolated from FAF1 gt/gt mice showed decreased ROS production and inflammatory responses as well as bacterial clearance than macrophages from FAF1 +/+ mice upon L. monocytogenes infection or TLR2 stimulation. Consistent with these data, knockdown of FAF1 in Raw264.7 cells also showed significantly reduced ROS production, NF-κB activation and inflammatory responses, upon L. monocytogenes infection. Third, FAF1 transiently interacted strongly with the p67phox-p47phox-p40phox complex at early time points after L. monocytogenes infection in macrophages, and FAF1 region comprising amino acids 330-489 was responsible for interaction with the TPR domain of p67phox. Finally, interaction between FAF1 and p67phox stabilized p67phox and increased activity of phagocytic NADPH oxidase upon L. monocytogenes infection. Collectively, these findings strongly suggest that FAF1 plays a crucial role in promoting antibacterial responses by interacting with p67phox in macrophages during intracellular microbial infection.
NADPH oxidase and dual oxidase induce ROS production by various cells and tissues in response to growth factors, cytokines, and pathogen-mediated signals [1]. Among these, the phagocyte NADPH oxidase is a multi-component complex in which the membrane glycoprotein gp91phox (known as NOX2) is tightly associated with p22phox; this complex is activated via association with cytosolic regulatory proteins such as p67phox, p47phox, p40phox, and the small GTPase Rac, resulting in ROS generation [1,3,4]. Based on homology to gp91phox (Nox2), the Nox family of NADPH enzymes comprises seven members: Nox1 through Nox5, plus Duox1 and Duox2 [35]. All gp91phox-related enzymes (except gp91phox (Nox2)) belonging to the Nox family are non-phagocytic enzymes expressed in epithelial or endothelial cells within diverse tissues and organs [1]. ROS generated by these NADPH oxidases take part in biological processes such as cell signaling, hormone biosynthesis, and host innate immune responses [5,36]. In particular, ROS play essential roles in phagocyte-mediated defense against bacterial infection; ROS kill engulfed pathogens directly, or indirectly by activating intracellular signaling pathways related to inflammatory responses, which then protect the host. ROS are necessary to eliminate intracellular bacteria such as mycobacteria, Listeria, and Salmonella, efficiently [37].
Recognition of pathogens by TLRs is the first line of host innate defense; indeed, interaction between TLRs and NADPH oxidase in phagocytes has been well studied. For example, Yang et.al., report that Nox2 is essential for TLR2-dependent inflammatory responses and for FAF1 positively regulates NADPH oxidase-mediated ROS production intracellular control during mycobacterial infection. They showed that Nox2 and TLR2 interact directly during mycobacterial infection [13,14]. Moreover, the interplay between TLR4 and NADPH oxidases such as Nox4 or Nox1 was studied in non-phagocytic cells [16,38]. However, lipopolysaccharide (LPS; a TLR4 agonist) also activates NADPH oxidase in phagocytes indirectly by increasing association of gp91phox (Nox2) with regulatory proteins in plasma membrane [39]. Here, we demonstrated that FAF1 is a crucial regulator of the phagocytic NADPH oxidase (Nox2) complex required for ROS production by macrophages in response to L. monocytogenes infection as well as TLR2 stimulation. We also examined whether FAF1 regulates inflammatory responses in macrophages stimulated by TLR4. As results, TLR4-mediated cytokine secretion by BMDMs isolated from FAF1 gt/gt mice was lower than that by cells from FAF1 +/+ mice (S14 Fig, panels A-B). These results suggest that FAF1 also activates signaling pathways related to TLR4 stimulation, as proposed by DeLeo et.al. However, further studies demonstrating a detailed mechanism of how FAF1 controls TLR4 signaling pathway are needed.
Nox2 in resting phagocytes is inactive; however, it is activated by phagocytosis of invading bacteria, leading to ROS production and their subsequent effects on host defense (e.g., killing bacteria and regulating intracellular signaling). Once bacteria are recognized by host TLRs, NADPH oxidase Nox2 (gp91phox) heterodimerizes with p22phox at the phagocyte membrane and is rapidly activated by cytosolic regulatory proteins [1,3,4]. These p22phox and cytosolic regulatory proteins (p67phox, p47phox, p40phox, and the small GTPase Rac) are indispensable for regulation of NADPH oxidase; indeed, increasing evidence suggests that it is important to control NADPH oxidase subunits to ensure appropriate ROS generation. For example, RUBICON interacts with p22phox, thereby increasing ROS production in response to infection by Gram-positive bacteria [12]. Moreover, Nox-dependent ROS production occurs in Parkinson's disease (autosomal recessive, early onset) 7 (Park7)-p47phox [40]. Our findings suggest that FAF1 is a key molecule that regulates activation of NADPH oxidase via strong interaction with p67phox.
As noted above, mass spectrometry analysis identified the NoxA1 protein as an interacting protein of FAF1. NoxA1 is expressed at high levels by colon epithelial cells and is responsible for activation of Nox1 (and thereby for subsequent ROS production) [1,3,33]. p67phox is a homolog of NoxA1 expressed by phagocytes. Thus, we hypothesized that FAF1 interacts with p67phox to regulate ROS generation in macrophages. In this study, we found that FAF1 interacts with p67phox at an early time point after L. monocytogenes infection. Moreover, amino acids 330-489 of FAF1 are required for interaction with p67phox. Conversely, FAF1 interacted with the TPR domain (amino acids 1-155) of p67phox. This domain, which comprises four 34 amino acid-long TPR motifs, is involved in a variety of protein-protein interactions [4,32,41,42]. In macrophage cell lines stably overexpressing wild-type FAF1 or FAF1 Δ330-489, we found that overexpression of wild-type FAF1 but not FAF1 Δ330-489 increased NADPH oxidase activity, ROS production, proinflammatory cytokine production, and antimicrobial activity. Ultimately, interaction between FAF1 and p67phox facilitated stabilization of p67phox and increased activity of NADPH oxidase upon L. monocytogenes infection. These results suggest that regulation of phagocytic NADPH oxidase by FAF1 is dependent on binding to p67phox. Consequently, we demonstrated that FAF1 positively regulates the NADPH oxidase 2 complex via stabilization of p67phox. FAF1 interacts potentially with many different proteins and functions as a negative and/or positive regulator in a variety of biological possesses [25,26,43]. Previous studies report that FAF1 homologs suppress antibacterial immunity in Drosophila and Locusta migratoria [44,45]. In addition, Park et al. showed that FAF1 suppresses IKK activation and nuclear translocation of NF-κB in fibroblasts [29,46]. However, we clearly demonstrate that FAF1 acts as a positive regulator of antibacterial responses by regulating ROS production. In particular, we identified the physiological role of FAF1 in defense responses against L. monocytogenes infection in FAF1 +/+ and FAF1 gt/gt mice. FAF1 gt/gt mice attenuates bacterial clearance due to reduced inflammatory responses. We also have focused on the role of FAF1 in phagocytes such as macrophages in vitro. Raw264.7 cells in which FAF1 was knocked down, as well as BMDMs and resident PMs isolated from FAF1 gt/gt mice, showed lower ROS production, proinflammatory responses, and bacterial killing activity than FAF1 +/+ mice upon L. monocytogenes infection. In addition, activation of molecules involved in NF-κB signaling was markedly reduced in FAF1-knockdown cells when exposed to L. monocytogenes. However, the upstream molecules responsible for FAF1 activation upon L. monocytogenes infection or TLR2 stimulation remain unclear. To assess this, further studies are necessary to examine the detailed molecular mechanisms (e.g., phosphorylation of FAF1) that operate in macrophages under infectious conditions.
In summary, we demonstrated that FAF1 is a critical positive regulator of the phagocytic NADPH oxidase (Nox2) complex responsible for ROS generation by phagocytes upon L. monocytogenes infection or TLR2 stimulation. FAF1 interacts directly with p67phox and stabilizes p67phox, thereby triggering NADPH oxidase-mediated ROS production, release of proinflammatory cytokines, and bacterial clearance in response to infection by L. monocytogenes. Taken together, the results suggest a plausible mechanism involving interaction between p67phox and FAF1 and increases our understanding of molecules that control ROS signaling and antibacterial defense responses against L. monocytogenes infection or TLR2 stimulation.

Ethics statement
All animal experiments were managed in strict accordance with the Guide for the Care and

Bacterial strain
L.monocytogenes (KVCC-BA0000087) was from Korean Veterinary Culture Collection (KVCC), and grown at 37˚C in Brain Heart Infusion (BHI) broth medium (BD 237500). Log phase bacteria (O.D. value, 0.6-0.8) were used for all assay. Cultures were aliquoted and stored at -80˚C. To determine bacterial titer (Colony-forming unit, CFU), bacteria thawed was diluted 10-fold. Each diluent was plated on BHI agar (BD 241830) and incubated at 37˚C for one day.

Mice and in vivo experiments
FAF1 +/+ and FAF1 gt/gt mice on a C57BL/6 background were kindly provided by Dr. Eunhee Kim (Chungnam National University, Daejeon, Korea). A hypomorphic allele, designated FAF1 gt/gt , was generated by a gene-trap insertion in intron 8 [47]. All mice were bred in pathogen-free condition. Offspring were genotyped by PCR as previously described [27].
Sex-matched mice (six-week-old) were intraperitoneally infected with L. monocytogenes (5 × 10 5 CFU/mouse). Liver, spleen, peritoneal macrophage and serum were collected to determine the bacterial load, cytokines or chemokines levels or mRNA levels of cytokines or chemokines as described below.

Construction of expression plasmids
FAF1 plasmid was kindly provided by Dr. Eunhee Kim (Chungnam National University, Daejeon, Korea). To generate plasmid constructs with GST expressing vector, FAF1 and its mutants or p67phox and its mutants were amplified by PCR and inserted into pEBG vector. For construction of V5-tagged expression plasmid, FAF1, FAF1 Δ330-489 and p67phox were amplified by PCR and inserted into pIRES-V5.

Immunoprecipitation and GST pull-down assay
For immunoprecipitation, Raw264.7 cells or BMDMs were lysed with RIPA buffer containing protease inhibitor cocktail (Roche). Lysates were incubated with a primary antibody overnight at 4˚C, followed by incubation with protein A/G agarose (Santa Cruz Biotechnology) for 3hrs at 4˚C. Then immunoprecipitates were washed with 1% nonidet P-40 for 3 times. For GSTpull down assay, HEK293T cells were transiently transfected with the indicated plasmids, and at 36hr post-transfection, were lysed with RIPA buffer containing protease inhibitor cocktail (Roche). Lysates were pre-cleared by sepharose 6B (GE Healthcare) for 3hrs at 4˚C, followed by incubation with glutathione (GSH) sepharose 4B (GE Healthcare) overnight at 4˚C. Precipitated beads were washed with 1% nonidet P-40 for 3 times.

Immunoblotting
Whole cell lysates (WCLs) were prepared using Radio-immunoprecipitation assay (RIPA) lysis buffer (50mM Tris-HCl, 150mM NaCl, 0.5% sodium deoxycholate, 1% IGEPAL, 1mM NaF, 1mM Na 3 VO 4 , proteinase inhibitor cocktail). Samples were separated by SDS-PAGE and transferred onto a PVDF membrane (Bio-Rad) using Trans-Blot semi dry transfer cell (Bio-Rad). Membranes were blocked for 1hr in 5% bovine-serum albumin with TBST or 5% skim milk within PBST, followed by incubation with primary antibody overnight at 4˚C. Next day, membranes were washed with TBST or PBST, and incubated at room temperature with HRPconjugated secondary antibody. Then, membranes were washed 3 times with TBST or PBST, followed by developing with Western blotting detection reagents (GE healthcare, ECL select Western Blotting Detection Reagent).

Mass spectrometry analysis
HEK293 cells were lysed with RIPA buffer containing protease inhibitor cocktail (Roche). Lysates were incubated with anti-FAF1 antibody overnight at 4˚C, followed by incubation with protein A/G agarose (Santa Cruz Biotechnology) for 3hrs at 4˚C. Then immunoprecipitates were washed with 1% nonidet P-40 for 3 times and separated by PAGE gels, followed by Coomassie blue staining. Protein bands were excised from the gel and identified by Q-TOF mass spectrometer.

ELISA
ELISA was performed to detect the secreted cytokines and chemokines in sera or cell culture supernatants. Mouse IL-6 (BD biosciences, 555240) and mouse IL-12 (BD biosciences, 555165), RANTES (Invitrogen), MCP-1 (Invitrogen) were used for analysis according to the manufacturer's protocols.

Quantitative real time PCR
Total RNA was isolated from cells and murine tissues using the RNeasy Mini Kit (Qiagen). cDNA synthesis was performed using reverse transcriptase (TOYOBO), cDNA was quantified by real-time polymerase chain reaction (PCR) using QuantiTect SYBR Green PCR kit

Confocal microscopy
BMDMs or HEK293T cells were seeded into eight-chamber slides, followed by treatment or transfection desired. The cells were fixed in 4% paraformaldehyde at RT for 20min, then permeabilized by incubation for 20min with 100% Methanol at -20˚C. Then, the fixed cells were incubated with 2% FBS for 1hr to block non-specific binding of antibodies. The appropriate primary antibodies were incubated overnight at 4˚C. The cells were washed three times with PBST, and incubated with the appropriate secondary antibodies (Invitrogen) for 1hr at RT without light exposure. Then, the cells were washed three times with PBST, and stained with DAPI (ratio, 1:100,000), washed with PBS, and mounted in mounting solution (VECTOR). Images were acquired under a Nikon laser scanning confocal microscope (C2plus) and analyzed using NIS-Elements software.

Quantification of intracellular bacteria growth
BMDMs or Raw264.7 cells were infected with L. monocytogenes (MOI = 0.1) for 1hr and washed 3 times with sterile PBS, followed by incubation with DMEM containing gentamycin (100μg/ml) for the indicated times. Finally, cells were harvested and lysed with 0.1% Triton X-100 (Sigma) to release the intracellular bacteria, and cell lysates were then diluted 10-fold in BHI broth (BD). Each sample were plated on BHI agar (BD) and incubated at 37˚C for one day. Colony-forming unit (CFU) was utilized to ensure quantification of intracellular bacteria.

Graphing and statistics
Prism 6 software (GraphPad Software) was used for charts and statistical analyses. The significance of results was analyzed by an unpaired two-tailed Student's t-test and Mann-Whitney test with a cutoff P value of 0.05. Error bars and P-values are indicated in the figure legends. BMDMs were isolated from FAF1 +/+ and FAF1 gt/gt mice, followed by IB with anti-FAF1 or β-actin antibody. (B) Resident PMs were isolated from FAF1 +/+ and FAF1 gt/gt mice, followed by IB with anti-FAF1 or β-actin antibody. (C and D) FAF1 +/+ and FAF1 gt/gt PMs were stimulated with L. monocytogenes (MOI = 1) or zymosan (100μg/ml) or BLP (100ng/ml). At the indicated time, supernatants were harvested and measured for IL-6 (C) or IL-12 (D) secretion by ELISA. Data are representative of two (C, D) independent experiments. Error bars, mean ± SD. � P < 0.05, �� P < 0.01 (Student's t-test). transfection with a control siRNA or p47phox-specific siRNA (200nM), Raw264.7 cells containing vector or V5-tagged FAF1 were infected with L. monocytogenes for the indicated times, followed by immunoblotting with anti-p-p65, anti-p65, anti-p-IκBα, anti-IκBα, anti-p47phox, anti-V5, or β-actin antibody. (B) At 36hr post-transfection with a control siRNA or p47phoxspecific siRNA (200nM), Raw264.7 cells containing vector or V5-tagged FAF1 were infected with L. monocytogenes for 12hrs. The supernatants were used for measuring IL-6 cytokine production using ELISA. Data are representative of two (A) or three (B) independent experiments. Error bars, mean ± SD. � P < 0.05, �� P < 0.01 (Student's t-test).