A Salmonella type III effector, PipA, works in a different manner than the PipA family effectors GogA and GtgA

Nuclear factor-kappa B (NF-κB) plays a critical role in the host defense against microbial pathogens. Many pathogens modulate NF-κB signaling to establish infection in their host. Salmonella enterica serovar Typhimurium (S. Typhimurium) possesses two type III secretion systems (T3SS-1 and T3SS-2) and directly injects many effector proteins into host cells. It has been reported that some effectors block NF-κB signaling, but the molecular mechanism of the inactivation of NF-κB signaling in S. Typhimurium is poorly understood. Here, we identified seven type III effectors—GogA, GtgA, PipA, SseK1, SseK2, SseK3, and SteE—that inhibited NF-κB activation in HeLa cells stimulated with TNF-α. We also determined that only GogA and GtgA are involved in regulation of the activation of NF-κB in HeLa cells infected with S. Typhimurium. GogA, GtgA, and PipA are highly homologous to one another and have the consensus zinc metalloprotease HEXXH motif. Our experiments demonstrated that GogA, GtgA, and PipA each directly cleaved NF-κB p65, whereas GogA and GtgA, but not PipA, inhibited the NF-κB activation in HeLa cells infected with S. Typhimurium. Further, expressions of the gogA or gtgA gene were induced under the SPI-1-and SPI-2-inducing conditions, but expression of the pipA gene was induced only under the SPI-2-inducing condition. We also showed that PipA was secreted into RAW264.7 cells through T3SS-2. Finally, we indicated that PipA elicits bacterial dissemination in the systemic stage of infection of S. Typhimurium via a T3SS-1-independent mechanism. Collectively, our results suggest that PipA, GogA and GtgA contribute to S. Typhimurium pathogenesis in different ways.


Introduction
Nuclear factor-kappa B (NF-κB) is a transcriptional factor that controls cellular processes such as proliferation, differentiation and death, as well as immune responses [1]. In the classical NF-κB signaling pathway, stimulated pattern recognition receptors (PRRs) or tumor necrosis factor (TNF) superfamily receptors lead to the initiation of this signaling pathway, which results in the induction of the degradation of IκBα (nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha) through its phosphorylation or ubiquitination by the lymph nodes (MLNs), or spleen were homogenized in phosphate-buffered saline (PBS), and 10-fold serial dilutions were plated on LB agar plates containing the appropriate antibiotic for determination of the bacterial count. For histopathology, cecum samples were fixed with 10% buffered-formalin (Mildform 10N; Wako, Osaka, Japan), embedded in paraffin according to standard procedures, sectioned at 5 μm, and stained with hematoxylin and eosin as described previously [19]. The pathologic changes were scored by a slight modification of the blinded examination described by Barthel et al. [20]: namely, the submucosal edema, polymorphonuclear leukocyte (PMN) infiltration, goblet cell numbers, and epithelial damage were examined for a total score of 0-13. More than 3 scores are considered as a sign of inflammation. The scores for PMN infiltration were modified as follows: 0 = <50 PMN/high power field; 1 = 50-70 PMN/high power field; 2 = 71-110 PMN/high power field; 3 = 111-150 PMN/high power field; 4 = >150 PMN/high power field.
RNA was extracted from the mouse cecum with the use of an RNeasy Mini Kit (Qiagen, Hilden, Germany) or an RNA Basic Kit (Nippon Genetics, Tokyo). cDNA was synthesized using TaqMan Reverse Transcription Reagent (Thermo Fisher, San Jose, CA). Real-time polymerase chain reaction (PCR) was performed with the primer pairs described previously [19], using a KAPA SYBR FAST qPCR Master Mix Kit (Kapa Biosystems, Woburn, MA) and analyzed using the comparative Ct method (Applied Biosystems, Foster City, CA). The levels of mRNA expression of target genes were normalized by the levels of Gapdh mRNA.
The competitive index (CI) assay has been described previously [21]. For intraperitoneal infection, mice were inoculated with 1 × 10 4 CFU of S. Typhimurium. For intragastric infection, mice were inoculated with 5 × 10 7 CFU of bacteria.

Bacterial strains and culture conditions
The bacterial strains used are listed in S1 Table. The S. Typhimurium wild-type strains used were ATCC 14028 and its spontaneous nalidixic acid-resistant derivative strain SH100 [22]. E. coli strain DH5α was used as the host for the construction of plasmids. Unless otherwise indicated, bacteria were grown in LB broth or on LB agar. Antibiotics were added to the media at the following concentrations: ampicillin (100 μg/ml), chloramphenicol (25 μg/ml), nalidixic acid (50 μg/ml), or kanamycin (25 μg/ml). Overexpression of the gene from a tac promoter was induced with Isopropyl β-D-1-thiogalactopyranoside (IPTG; 1 μM).

Construction of mutant strains and plasmids
The bacterial plasmids and primers used are listed in S2 and S3 Tables, respectively. For the construction of S. Typhimurium TH1624 (T1), an invA::pEP185.2 [23] was transduced from IR715 into SH100 using a P22 phage. Gene deletions of S. Typhimurium were constructed by the PCR-based λ Red recombination system [24]. For construction of the double-gene deletion in S. Typhimurium, TH1821 (ΔgogAΔpipA::Km), a ΔpipA::Km mutation from strain TH1772, was introduced by P22 phage, after elimination of the chloramphenicol (Cm) cassette from strain TH1671 (ΔgogA::Cm) using pCP20. Using  gogA, gtgA, and pipA were amplified by PCR and cloned into pTAKN-2 (BioDynamics Laboratory, Tokyo). The PCR product for gtgA was digested with BamHI to distinguish between gogA and gtgA. After TA cloning, these genes were subcloned into a plasmid pFLAG-CTC (Sigma, St. Louis, MO) for the expression of FLAG-fusion proteins in E. coli and S. Typhimurium, a plasmid pMW118 (Nippon Gene, Tokyo, Japan) for a complementation assay or a plasmid pEGFP-C1 (Clontech, Mountain View, CA) for expression of EGFP-fusion proteins in mammalian cell lines . Point mutations (on GogA, GtgA, or PipA) were constructed with overlapping  primers, TH816 and TH817, TH765 and TH766, or TH846 and TH847, respectively, using a  QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA). These mutations were verified by DNA sequencing. pEGFP-NleC [25] and pCMV-FLAG-p65 were provided by Drs. Toru Tobe (Osaka University) and Hiroshi Ashida (Tokyo Medical and Dental University), respectively.

Transfection
EGFP plasmids or pCMV-FLAG-p65 were transfected into HeLa or HEK293T cells with the reagent PEI-MAX (Polysciences, Warrington, PA) and plasmid in a ratio of 3:2, according to the manufacturer's protocol.

NF-κB reporter assay
The NF-κB reporter assay was described previously [19]. The NF-κB reporter plasmid pGL4.32 containing an NF-κB response element that drives the transcription of the luciferase reporter gene luc2P, the luciferase control plasmid pGL4.74 (Promega, Madison, WI), and the pEGFP-effector plasmid were co-transfected into HeLa cells using Fugene HD (Promega), according to the manufacturer's protocol. Forty-eight hrs later, the cells were stimulated with 10 ng/ml TNF-α (Peprotech, Rocky Hill, NJ) for 30 min. NF-κB activity was calculated using a Dual-Glo Assay System (Promega) according to the manufacturer's instructions.
For the NF-κB reporter assay in HeLa cells infected with S. Typhimurium, pGL4.32 and pGL4.74 were transfected into HeLa cells which were seeded at 2.5 × 10 4 cells/well in 24-well plates for 24 hrs. Forty-eight hrs after the transfection, the cells were infected with S. Typhimurium, which was incubated in LB containing 0.3M NaCl to induce T3SS-1 [26], and the gentamycin killing assay was performed as described previously [19]. After 4 or 20 hrs of infection, the cells were washed three times with PBS and lysed with 1 × Passive Lysis Buffer (Promega). NF-κB activity was calculated as described above.

Immunofluorescence staining
HeLa cells were seeded at 1.0 × 10 4 cells on 13-mm cover glasses in a 24-well plate 24 hrs prior to the transfection. The cells were transfected with pEGFP-effector and stimulated with TNF-α as described above. Cells were fixed with 4% formaldehyde in PBS and permeabilized with 0.1% Triton X-100. The cells were incubated with anti-p65 primary antibody (#8242; Cell Signaling Technology, Beverly, MA) and probed with Alexa Fluor 594 goat anti-rabbit IgG (Molecular Probes, Eugene, OR). Cover glasses were mounted with Vectashield medium with the stain DAPI (Vector Laboratories, Burlingame, CA), and samples were visualized using a fluorescence microscope (VertA1; Zeiss, Jena, Germany).

Cell fractionation
HeLa cells were seeded in 6-well plates at 2.0 × 10 5 cells/well 24 hrs prior to the transfection or the infection. The transfection of pEGFP-effector into the cells, stimulation of the cells with TNF-α, and infection with S. Typhimurium were described above. After the cells were detached with TrypLE Select (Gibco, Grand Island, NY), the cytoplasmic or the nuclear fraction was isolated as described previously [27].

Gene expression analysis in S. Typhimurium by qPCR
Analysis of bacterial gene expression by qPCR has been described previously [21]. S. Typhimurium were grown in LB containing 0.3M NaCl [26] or low phosphate and magnesium-containing medium (LPM, pH 5.8) [28] for induction of T3SS-1 or T3SS-2, respectively. RNA extraction and qPCR were performed as described above. The data were analyzed using the comparative Ct method (Applied Biosystems). Transcription of the target gene was normalized to the levels of gyrA mRNA.

CyaA translocation assay
The CyaA translocation assay was performed as described previously [19]. The secretion of PipA from S. Typhimurium into RAW264.7 cells was measured using a cAMP enzyme immunoassay (EIA) kit (Cayman Chemical, Ann Arbor, MI). After 20 hrs of infection, the cells were washed three times with PBS and lysed using the standard protocol recommended by the manufacturer. A cAMP EIA was performed using an iMark microplate reader (BIO-RAD, Hercules, CA).

LDH assay
Cytotoxicity was assessed by LDH release into the culture medium using a CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega, Madison, WI). After 20 hrs of infection, the supernatants were collected from infected cells and the OD 490 was measured using an iMark microplate reader. The relative LDH release was calculated using a CytoTox 96 kit according to the manufacturer's protocol.

Statistical analyses
Statistical tests were performed with Prism 8 software (GraphPad, La Jolla, CA). One-way ANOVA was used to analyze the relative NF-κB activity in HeLa cells transfected with the pEGFP-effector plasmid. The relative NF-κB activity in HeLa cells infected with S. Typhimurium, the gene expression, the bacterial number, and the cytokine expression were analyzed by Student's t-test. For the analysis of bacterial numbers, the bacterial numbers (CFU/ml) were transformed logarithmically prior to the statistical analysis. Histopathological scores were compared by a nonparametric Mann-Whitney U-test. The competitive index (CI) was analyzed using a one sample t-test. Probability (p) values < 0.05 were considered significant.

Identification of Salmonella type III effectors that interfered with NF-κB activation
For the identification of type III effectors that inactivate the NF-κB pathway, each of the known type III effectors in S. Typhimurium was cloned into one copy of the mammalian expression plasmid pEGFP-C1, and the resulting plasmids were transfected into HeLa cells together with both the NF-κB reporter firefly reporter plasmid (pGL4.32) and the control Renilla luciferase plasmid (pGL4.74). After transfection, the HeLa cells were stimulated with TNF-α, and we measured the NF-κB activity by performing a luciferase reporter assay. We confirmed the expression of each GFP-fusion protein in HeLa cells by immunoblotting using anti-GFP antibody or immunofluorescent microscopy (S1 Fig). As shown in earlier reports, transfection with NleB1, which is a type III effector in enteropathogenic or enterohemorrhagic E. coli (EPEC or EHEC), led to a decrease in the NF-κB activity compared to transfection of cells with an empty vector [29][30][31]. In contrast, a Salmonella effector, SpvC, was shown in our previous study to be a phosphothreonine lyase that does not affect NF-κB signaling [29][30][31]. As with the NleB1 transfection, the relative NF-κB activity of HeLa cells transfected with seven plasmids was significantly reduced compared to that in the cells transfected with SpvC (Fig 1). Those seven plasmids contained the gogA, gtgA, pipA, sseK1, sseK2, sseK3, or steE gene.

Salmonella-infected HeLa cells
To study the effects of these effectors on the NF-κB pathway during S. Typhimurium infection, we constructed an S. Typhimurium strain lacking all seven of the identified genes (gogA, gtgA, pipA, sseK1, sseK2, sseK3 and steE) and infected HeLa cells transfected with both pGL4.32 and

Fig 1. A comprehensive analysis of the Salmonella type III effectors that inhibit NF-κB activation.
HeLa cells were transfected with the indicated pEGFP-effector fusion plasmid together with pGL4.32 and pGL4.74. After 48 hrs, the cells were stimulated with TNF-α (10 ng/ml) and further cultured for 30 min, and the NF-κB activity was measured. The relative NF-κB activity is the value of the NF-κB activity of HeLa cells transfected with the pEGFP-effector fusion plasmid relative to that of the HeLa cells transfected with the empty vector, which was taken as 100. One-way ANOVA analysis (Dunnett's multiple comparison test against the corresponding HeLa cells transfected with pEGFP-SpvC) was performed for statistical analysis. Individual data are shown as a scatter plot, and bars are the mean. Asterisks indicate statistically significant differences (p < 0.05). The effectors that inhibit NF-κB activation are shown in bold.
https://doi.org/10.1371/journal.pone.0248975.g001 pGL4.74. At 4 or 20 hrs after infection, we measured the luciferase levels and calculated the relative NF-κB activity. This strain exhibited greater NF-κB activity than the wild-type S. Typhimurium at 4 hrs after infection, but the same level as the wild-type at 20 hrs after infection ( Fig  2A). This indicated that at least one of these genes is related to the inhibition of NF-κB activation in the S. Typhimurium-infected cells at 4 hrs after infection.
Next, we examined which of these seven effectors is required for suppression of NF-κB activation. GogA, GtgA and PipA belong to the PipA family of effectors, which interfere with NF-κB signaling by cleaving p65 [13,14]. SseK1, SseK2, and SseK3 are highly related effector proteins, and they share a high degree of homology with NleB1, an EHEC or EPEC T3SS effector that functions as an N-acetylglucosamine (GlcNAc) transferase to modify death domain proteins and inhibit NF-κB activation [32][33][34]. SteE is an effector through T3SS-1 and T3SS-2 [35]; Recent reports have shown that SteE (also known as SarA) promotes secretion of IL-10 from infected B cells [36] and induces anti-inflammatory macrophage polarization [37]. However, it is not clear whether these phenotypes are related to the inhibition of NF-κB activation.
GtgA, GogA, and PipA proteins have high homology to one another. These proteins belong to the peptidase M85 Pfam family (PF13678). Members of this family have a consensus zinc metalloprotease HExxH motif. GtgA and GogA possess the HEVVH amino acid (aa) sequences between aa 182 and 186, and PipA has the motif in aa residues 180-184. Several groups have reported that NleC, a type III effector from EPEC or EHEC, contains an HExxH motif and inhibits NF-κB signaling by a proteolytic cleavage of p65 [25,[38][39][40][41][42]. It has also been shown that GtgA, GogA, and PipA collectively function as zinc metalloproteases and target the NF-κB p65 [13,14]. However, the roles of individual effectors in S. Typhimurium pathogenesis are not yet clear.
Next, to reveal the role of these effectors in the HeLa cells infected with S. Typhimurium, we constructed ΔgogAΔgtgA strains complemented with a plasmid expressing flag-tagged gogA, gtgA, or pipA. The expressions of each gene from the plasmid were driven by the tac promoter with its inducer, isopropyl β-D-1-thiogalactopyranoside (IPTG) and were monitored by immunoblotting using anti-FLAG antibody (S2 Fig). The introduction of a plasmid expressing gogA, gtgA, or pipA into the ΔgogAΔgtgA strain resulted in a decrease in the activation of NF-κB compared to the introduction of an empty plasmid, pMW118 ( Fig 2D). However, the phenotype of ΔgogAΔgtgA could not be restored by induction of the plasmid pGogA H182Y , pGtgA H182Y , or pPipA H180Y ; these plasmids have a histidine-to-tyrosine point mutation at the first histidine in the consensus zinc metalloprotease motif on GogA, GtgA, or PipA, respectively ( Fig 2D). Collectively, these data suggest that GogA and GtgA, but not PipA, attenuate the NF-κB response in the S. Typhimurium-infected HeLa cells, and that PipA can dampen the NF-κB activation only when overexpressed in the ΔgogAΔgtgA strain.

PLOS ONE
The role of the PipA family of effectors in Salmonella infection

GogA and GtgA cleave p65 in S. Typhimurium-infected HeLa cells
We next determined whether NF-κB p65 is cleaved in HeLa cells infected with the S. Typhimurium wild-type or mutant strains. We observed cleaved p65 in HeLa cells infected with the wild-type and all three of the single mutants, but not in HeLa cells infected with the ΔgogAΔgt-gAΔpipA::Km mutant at 2 hrs after infection (Fig 3A). In HeLa cells infected with double mutants, ΔgogAΔpipA::Km and ΔgtgAΔpipA::Km could cleave p65, but ΔgogAΔgtgA could not ( Fig 3B). However, cleaved p65 was not detected in HeLa cells infected with the wild-type or any of the mutants at 16 hrs after infection (S3 Fig). These findings indicate that GogA and GtgA act in a redundant manner to digest the p65 protein in the early stage of infection with S. Typhimurium. To confirm this finding, we infected HeLa cells with ΔgogAΔgtgA carrying a plasmid cloned from a wild-type or a mutant type of gogA, gtgA, or pipA. As expected, the phenotype of ΔgogAΔgtgA was complemented by pGogA, pGtgA, or pPipA, but not by pGo-gA H182Y , pGtgA H182Y , or PipA H180Y (Fig 3C). These results are consistent with the finding that GogA and GtgA, and even PipA, block NF-κB signaling in HeLa cells infected with S. Typhimurium, when at least one of these effectors is expressed in the ΔgogAΔgtgA strain.

PipA family effectors, GogA, GtgA, and PipA, possess the zinc metalloprotease activity
It has been reported that GogA, GtgA, and PipA inhibit NF-κB signaling by the cleavage of p65 [14,43]. However, our present results demonstrated that GogA and GtgA, but not PipA, blocked the NF-κB activation in the S. Typhimurium-infected cells. To confirm that GogA, GtgA, and/or PipA act as zinc metalloproteases, we constructed the mammalian expression plasmids pEGFP-GtgA, pEGFP-GogA, and pEGFP-PipA, which have a histidine-to-tyrosine point mutation at the first histidine on GogA, GtgA and PipA, respectively (pEGFP-Gt-gA H182Y , pEGFP-GogA H182Y and pEGFP-PipA H180Y ), and we transfected the plasmids into HEK293T cells. The p65 signal was reduced in the cells that expressed wild-type GogA, GtgA or PipA, but not in the cells that expressed the point mutants (S4 Fig). To facilitate the enzymatic activity of these proteins, we determined the amounts of cytoplasmic p65 in HEK293T cells transiently expressing the wild-type or point mutants of GogA, GtgA, or PipA by using an anti-p65 antibody. Full and cleaved p65 were observed in the cells expressing the wild-type GogA, GtgA, or PipA, as well as those expressing NleC, but cleaved p65 was not detected in the cells transfected with the point mutants (S5 Fig). Taken together, these data indicate that GogA, GtgA, and PipA exhibit zinc metalloprotease activity, which is consistent with the findings of previous studies.

PipA functions as a T3SS-2 effector
As shown in Figs 2 and 3, the phenotype of a ΔgogAΔgtgA strain was complemented by overexpression of gogA, gtgA, or pipA. We therefore reasoned that PipA may not be expressed in HeLa cells infected with S. Typhimurium. Originally, PipA was identified as a virulence factor which is required for Salmonella enteropathogenicity [44]. It has also been reported that PipA is secreted under T3SS-1-inducing conditions [35]. These reports suggest that PipA is a T3SS-1 effector. However, PipA is located in an operon with a gene encoding the T3SS-2 effector PipB (S6 Fig) [44,45]. Moreover, it has been indicated that PipA is up-regulated in macrophages at the late stage of infection [46]. Hence, we hypothesized that PipA functions as a T3SS-2 effector. Previous studies have shown that the expressions of T3SS-1 and its effectors are induced in LB medium containing 0.3M NaCl, while the expressions of T3SS-2 and its effectors are induced in low phosphate and magnesiumcontaining medium (LPM; pH 5.8) [26,28]. To determine whether the pipA gene was expressed under the T3SS-1-inducingand/or T3SS-2-inducing conditions, the mRNA levels of the gogA, gtgA (these levels were indistinguishable and are given as gogA/gtgA hereinafter), and pipA genes were assessed by qRT-PCR. Under the T3SS-1-inducing conditions, a SPI-1 positive regulator, invF, and gogA/gtgA were induced, but a SPI-2 positive regulator, ssrB, and pipA were not (Fig 4A). On the other hand, under the T3SS-2-inducing conditions, ssrB, gogA/gtgA, and pipA were induced ( Fig 4B).
Next, to confirm that PipA is secreted through T3SS-2, we constructed a gene fusion protein between pipA and a calmodulin-dependent adenylate cyclase (cyaA), and performed a CyaA translocation assay. The PipA-CyaA fusion protein was then produced in the S. Typhimurium wild-type or in a mutant strain lacking the invA (T3SS-1-deficient mutant, T1) or invA and ssaV genes (T3SS-1 and T3SS-2-deficient mutant, T1T2). To confirm the function of T3SS-2 in the strains tested, we used SseJ, a component of the T3SS-2 translocon, fused to CyaA (SseJ-CyaA) as a positive control and cultivated each strain under T3SS-2-inducing conditions. To confirm the T3SS-2-dependent translocation of PipA into host cells, cells of a mouse macrophage-like cell line, RAW264.7 (RAW), were infected with the S. Typhimurium wild-type or the mutant strain T1 or T1T2. After 20 hrs of infection, a higher level of translocation of PipA-CyaA was induced in the wild-type and T1-deficient strains than in either the T1-deficient or T2-deficient strain (Fig 4C). These data suggest that PipA can be behave as a T3SS-2 effector. To further characterize the function of PipA as a T3SS-2 effector, we analyzed the contribution of PipA to intracellular survival and induction of cell death in macrophage-infected S. Typhimurium. The invA::pEP185.1 ΔpipA::Km strain resulted in the same level of intracellular survival and induction of cell death in RAW cells as the T1 strain (Fig 4D and 4E).

PLOS ONE
The role of the PipA family of effectors in Salmonella infection

PipA contributes to systemic dissemination of S. Typhimurium in mice
The S. Typhimurium ΔgogAΔgtgAΔpipA mutant elicits severe inflammation in the cecum of Slc11a1 (previously Nramp1) +/+ mice compared to Slc11a1 −/− mice at 4 days after infection [14]. We therefore assessed the inflammation in the cecum of Slc11a1 +/+ mice (CBA) infected with the S. Typhimurium wild-type, or the ΔgogAΔgtgA or ΔgogAΔgtgAΔpipA::Km mutant. No difference was observed in the number of bacteria in the cecum, MLNs, or spleen, or in the histopathological score in the cecum between the S. Typhimurium wild-type, ΔgogAΔgtgA, and ΔgogAΔgtgAΔpipA::Km strains at 4 days after infection (S7 Fig). In the cecum of CBA mice infected with the S. Typhimurium wild-type strain at 4 days after infection, inflammation was not sufficiently induced (S7 Fig). We thus repeated the experiments at 7 days after infection, and we found that ΔgogAΔgtgA tended to have higher colonization in the colon contents than the wild-type (Fig 5A). However, there were no significant differences in the number of CFUs in the colon contents, MLNs, or spleen among mice infected with the S. Typhimurium wildtype, ΔgogAΔgtgA, and ΔgogAΔgtgAΔpipA::Km strains (Fig 5A). Moreover, at 7 days after infection, the ΔgogAΔgtgA or ΔgogAΔgtgAΔpipA::Km strains tended to induce a greater level of inflammation in the cecum of mice compared to the wild-type strain, but no differences in the histopathological score or inflammatory cytokines in the cecum were detected among mice infected with the three strains (Fig 5B and 5C).
The survival of BALB/c (Slc11a1 -/-) mice orally infected with the S. Typhimurium ΔpipA mutant was significantly prolonged compared to that of mice infected with the wild-type [45]. As shown in Fig 4, PipA functions as a T3SS-2 effector. Therefore, we hypothesized that the zinc protease activity of PipA secreted from a T3SS-2 was related to systemic infection of S. Typhimurium in mice. To examine this hypothesis, we first analyzed the virulence of the T3SS-1-and pipA-deficient mutant using a competitive index (CI) assay. C57BL/6 mice were intragastrically inoculated with invA::pEP185.1 (T1) and invA::pEP185.1 ΔpipA::Km (T1 PipA). The average of the CIs of colon contents from these mice was not significantly different from 1 (logCI = 0) (Fig 6A). On the other hand, the average of the CIs of the mesenteric lymph nodes (MLNs) and spleens of the mice were significantly lower than 1 (Fig 6A). However, the average of the CIs of the spleens from mice intraperitoneally infected with these S. Typhimurium strains was not significantly different from 1 (Fig 6B). These data show that T1 PipA reduces the systemic dissemination of bacteria in a mouse typhoid model.
Finally, we wanted to know whether the zinc metalloprotease activity of PipA contributes to systemic infection in mice. However, the S. Typhimurium ΔgogAΔgtgA strain carrying either pMW-PipA or pMW-PipA H180Y , shown in Figs 2 and 3, was not well suited for studies of the role of the enzymatic activity of PipA, because plasmid instability was predicted during infection of mice [23]. Therefore, to study the role of PipA in vivo, we constructed S. Typhimurium strains carrying the wild-type pipA gene or pipA H180Y gene, which expresses the inactivated form of zinc metalloprotease, inserted in the phoN gene on the genome of S. Typhimurium (phoN::pipA or phoN::pipA H180Y ). Successful complementation was achieved for T1 PipA by the expression of the wild-type pipA gene, but not the pipA H180Y mutant ( Fig  6C). These results suggest that the zinc protease activity of PipA through T3SS-2 is conducive to dissemination of S. Typhimurium to systemic sites of infection.

Discussion
Here, we report that Salmonella effector proteins, i.e., GogA, GtgA, PipA, SseK1, SseK2, SseK3, and SteE, dampen NF-κB signaling. Although all of these effectors except SteE have previously been shown to inhibit NF-κB activation, our present findings demonstrate that only GogA and GtgA are related to the blocking of the NF-κB signaling in HeLa cells infected with S. Typhimurium. In Fig 2C, the NF-κB activity of HeLa cells infected with the ΔgogA strain was higher than that of HeLa cells infected with the wild-type strain but lower than that of HeLa cells infected with the ΔgogAΔgtgA or ΔgogAΔgtgAΔpipA strain. Further, the ΔgogA strain, as well as the ΔgtgA and ΔpipA strains, was capable of cleaving NF-κB p65 (Fig 3A). These findings could indicate that not only the ΔgtgA and ΔpipA strains, but also the ΔgogA strain dampens the NF-κB signaling, and that GogA and GtgA play a redundant role in cleaving p65.

PLOS ONE
The role of the PipA family of effectors in Salmonella infection NF-κB p65 was cleaved in HeLa cells infected with the S. Typhimurium wild-type or ΔssaV::Cm (T2) strain, but not in those infected with the ΔgogAΔgtgA or ΔinvA::pEP185.2 (T1) strain (S8 Fig). This result indicates that this phenotype observed in HeLa cells at 4 hrs post-infection is dependent on T3SS-1. In contrast, SseK1, SseK2, SseK3, and/or SteE may function at a late stage of infection, since these effectors were translocated through T3SS-2 or both T3SS-1 and T3SS-2 [32,33,35]. Several studies have reported that other effectors of T3SS-2 or of both T3SS-1 and T3SS-2-i.e., AvrA, GogB, SpvD, SseL, SspH1, and SteA-inhibit NF-κB activation [8][9][10][11][12]47]. Interestingly, as shown in Fig 2A, the level of NF-κB activity was low In HeLa cells infected with the S. Typhimurium wild-type at 20 hrs after infection, and similar to the levels at 4 hrs after infection. This indicates that the NF-κB signaling is suppressed in HeLa cells during the late stage of infection. However, the level of NF-κB activity In HeLa cells infected with the ΔgogAΔgtgAΔpipAΔsseK1ΔsseK2ΔsseK3ΔsteE strain after 20 hrs of infection was similar to that in the HeLa cells infected with the wild-type. Moreover, cleaved p65 was not detected in HeLa cells infected with the wild-type after 16 hrs of infection (S3 Fig). elHeThese data suggest that other T3SS-2 effectors inhibit the NF-κB signaling pathway in a way that is different from cleaving p65. Further studies will be needed to understand the role of suppression of NF-κB activation in the late stage of S. Typhimurium-infection. PipA, GogA, and GtgA were reported to cleave NF-κB p65 and inhibit the NF-κB response in epithelial cells and fibroblasts [13,14]. Our present findings showed that PipA does not contribute to the dampening of the NF-κB activation in Salmonella-infected HeLa cells. However, the ΔgogAΔgtgA (pPipA) strain, which overexpresses PipA from the plasmid, dampens the NF-κB activation. Cleaved p65 was also observed in HeLa cells transfected with PipA. These facts suggest that the PipA protein can cleave p65 but does not appear to do so at the times measured in HeLa cells during S. Typhimurium infection. PipA, GogA and GtgA are classified as members of the PipA family because of the high similarity among their amino acid sequences, but there are some differences between these three proteins. First, the amino acid sequence of PipA is homologous to those of GtgA and GogA, but the similarity of the amino acid sequence between PipA and GtgA or between PipA and GogA is slightly lower than that between GtgA and GogA (S6 Fig). The promoter region of the pipA gene and five N-terminal amino acids of the PipA protein are also different from those of the gtgA or gogA gene and the GtgA or GogA protein, respectively (S6 Fig). These findings may indicate that the patterns of protein expression and secretion for PipA differ from those for GogA and GtgA. Consistent with these findings, we here showed that PipA was not expressed under the SPI-1-inducing conditions.
The gogA and gtgA genes but not the pipA gene are located on Gifsy-1 or Gifsy-2 prophages integrated in the genome of S. Typhimurium. Another T3SS-1 and T3SS-2 effector, GogB, which suppresses NF-κB signaling, is also located on the Gifsy-1 prophage [12]. SteE, which was identified in the present study, is also encoded on this prophage. Interestingly, the cleavage of p65 was not observed in HeLa cells infected with S. Enteritidis, which causes a food-borne disease (as does S. Typhimurium) (S8 Fig). In the genome of the highly virulent strain S. Typhimurium UK-1, which was isolated from an infected horse, one more copy of metalloprotease (STMUK_2657) is located on the Gifsy-1 prophage compared to gogA (STMUK_2648), gtgA (STMUK_0993), and pipA (STMUK_1056) [48]. STMUK_2657 is a homolog of NleC, which is a type III effector in EHEC or EPEC and interferes with NF-κB activation by cleaving p65 [49]. Thus, the inactivation of the NF-κB signaling pathway by these effectors may be important for the virulence of S. Typhimurium.
PipA belongs to Salmonella pathogenicity island 5 (SPI-5), which is conserved in all Salmonella serovars [44]. In bovine ligated ileal loops infected with S. Dublin, PipA is related to the induction of intestinal secretory and inflammatory responses [44]. PipA also contributed to intestinal colonization and systemic dissemination in 1-day-old chickens [50]. These findings suggest a possible role of PipA for virulence in not only the serovar Typhimurium, but also other serovars. In this study, we indicated that the zinc metalloprotease activity of PipA plays an important role for the systemic infection of S. Typhimurium, but were not be able to identify the target for PipA. In EPEC or EHEC, a type III effector, NleD, was identified as a zinc metalloprotease and was reported to cleave JNK and p38 MAPKs, but not NF-κB [42]. This might indicate that PipA digests another target in Salmonella infection. Further studies are needed to identify substrates for the PipA zinc metalloprotease and to determine the role of PipA in Salmonella pathogenesis. Sun et al. showed that deletions of the gogA, gtgA, and pipA genes increase intestinal inflammation in C57/BL6 Slc11a1 (previously Nramp1) +/+ mice, but not wild-type C57/BL6 mice (Slc11a1 −/− ) [14]. Slc11a1 encodes the divalent metal iron transporter that confers resistance to S. Typhimurium infection [51]. Some groups have shown that S. Typhimurium leads to inflammation in CBA mice (Slc11a1 +/+ ), approximately 10 days after infection [52,53]. In the present study, therefore, we used CBA mice, but we did not observe any difference in the histopathological score or the production of inflammatory cytokines between the mice infected with the S. Typhimurium wild-type strain and those infected with the ΔgogAΔgtgAΔpipA mutant. We noticed that the number of bacteria in colon contents of CBA mice infected with the wild-type strain SH100 (10 4 CFU/g) at 7 days after infection was lower than that in the feces of mice infected with strain IR715 (10 8 CFU/g), which is a nalidixic acid-resistant strain of ATCC 14028 used as a parental strain in this study, during 4 to 14 days after infection in a report by Rivera-Chaves et al. [52]. This finding likely provides at least one of the reasons why the wild-type did not elicit inflammation in the cecum of CBA mice in our present experiments. Even among the CBA mice infected with the S. Typhimurium wild-type strain at 13 days after infection, severe colitis in the cecum was not found by gross pathology (S9A Fig). We also obtained CBA mice from another breeder, Japan Charles River, but we did not observe severe inflammation in the cecum of mice infected with S. Typhimurium at 13 days after infection (S9B Fig). C57BL/6 and BALB/c mice are highly resistant to the intestinal inflammation induced by S. Typhimurium. Since severe colitis is elicited in mice pretreated with streptomycin and in germ-free mice infected with S. Typhimurium, the microbiota contribute to the resistance to S. Typhimurium colonization in these mice [20,54]. Our present findings suggest that the CBA mice used in this study may have somehow acquired colonization resistance to S. Typhimurium during housing before we obtained them.
In summary, our results demonstrated that the Salmonella effectors GogA and GtgA, as well as PipA, have metalloprotease activity, but the metalloprotease activity of PipA did not function in the HeLa cells infected with S. Typhimurium at an early stage of infection. It appeared that PipA was not expressed in the HeLa cells at an early stage of infection, because the expression of PipA was induced under T3SS-2-inducing conditions, but not under T3SS-1-inducing conditions. We also showed that PipA is translocated into RAW cells through T3SS-2, and that the enzymatic activity of PipA is responsible for systemic infection of S. Typhimurium. Further studies are needed to characterize the individual effectors identified in this study and thereby clarify the role played by the blocking of NF-κB signaling in S. Typhimurium infection.