Fetal bovine serum (FBS) was purchased from BioWhittaker Inc. (Walkersville, MO, USA). The following antibodies were purchased: NOX2, Rac1, and ANXA2 antibodies (BD Biosciences, Franklin Lakes, NJ, USA); NCF1 antibody (LifeSpan Biosciences, Seattle, WA, USA); p-JNK, JNK, PKCα^{Ser 657}, p-NF-κBp65, NF-κBp65, IL-1β p17, caspase-1 p10, and β-actin antibodies (Santa Cruz Biotechnology, Paso Robles, CA, USA); Horseradish peroxidase (HRP)-conjugated goat anti-rabbit and goat anti-mouse IgG antibodies (Jackson Immunoresearch, West Grove, PA, USA). 2′, 7′-dichlorofluorescein diacetate (CM-H₂DCFDA) was obtained from Invitrogen (Carlsbad, CA, USA). Methyl-β-cyclodextrin (MβCD), lipopolysaccharide (LPS), adenosine 5′-triphosphate disodium salt hydrate (ATP), 3-methyladenine, bafilomycin A1, N-acetyl-l-cysteine (NAC), rotenone, and 5-azacytidine were purchased from Sigma Chemical Company (St. Louis, MO, USA). The concentrations of all of pharmacological inhibitors listed did not show any significant cytotoxic effects by themselves as confirmed by FACS analysis in each experiment. All other reagents were of the highest purity commercially available and were used as received.

RAW 264.7 murine macrophage and Caco-2 human intestinal epithelial cells were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured at 37°C in 5% CO₂ in RPMI-1640 and DMEM containing 10% FBS and antibiotics. These cell lines have previously been used to evaluate the function of virulence factors of V. vulnificus in regulation of pro-inflammatory process and cytotoxic effects (Lee et al., 2015a; Song et al., 2016).

Cells (1 × 10⁶ cells/mL) grown to confluence in 6-well plates were incubated in serum/antibiotic-free media for 24 h. The bacterial suspensions in serum/antibiotic-free media were added to the cells at a multiplicity of infection (MOI) of 1, 5, and 10, after which the infected cells were incubated for 1 h. For IL-1β ELISA, co-culture supernatants were obtained and harvested by centrifugation at 4,000 × g for 10 min at 4°C. This was repeated once more with the resulting supernatant. The clarified supernatants were filtered through a 0.22 μM pore membrane to ensure the removal of any remaining bacterial cells and debris. The IL-1β concentration in the culture medium was quantified by an enzyme-linked immunosorbent assay (ELISA) (R&D systems, Minneapolis, MN, USA) according to the manufacturer's instructions.

All V. vulnificus strains (MO6-24/O; WT, MO6-24/O ΔrtxA; RTX mut, MO6-24/O vvpE; VvpE mut, MO6-24/O ΔvvhBA; VvhA mut, and MO6-24/O ΔvvpM; VvpM mut) are isogenic and naturally resistant to polymyxin B (Table 1). Unless otherwise noted, V. vulnificus strains were grown in Luria Bertani (LB) medium supplemented with 2.0% (wt/vol) NaCl (LBS) at 30°C. All media components were purchased from Difco (Difco Laboratories Inc., Detroit, MI). V. vulnificus were grown to mid-log phase (A₆₀₀ = 0.500) corresponding to 2 × 10⁸ CFU/mL and centrifuged at 6,000 × g for 5 min. The pellet was washed with phosphate buffered saline (PBS) and adjusted to desired colony-forming unit (CFU)/mL based on the A₆₀₀ determined using a UV–VIS spectrophotometer (UV-1800, Shimadzu, Japan) to estimate culture density.

Various deficient mutants in VvpM, RTX, and VvpE genes were generated by the methods as described previously (Jeong et al., 2000; Lee et al., 2007). For generation of VvhA mutant (Milton et al., 1996), the vvhBA operon was inactivated in vitro by deletion of the open reading frame (ORF) of vvhBA (998-bp of 1,935-bp) using the PCR-mediated linker-scanning mutation method as described previously (Jang et al., 2016). Briefly, pairs of primers VVHBA01-F and -R (for amplification of the 5′ amplicon) or VVHBA02-F and -R (for amplification of the 3′ amplicon) were designed and used (Table S1). The vvhBA operon with 998-bp deletion was amplified by PCR using the mixture of both amplicons as the template and VVHBA01-F and VVHBA02-R as primers. The resulting ΔvvhBA was ligated into SpeI-SphI-digested pDM4 (Milton et al., 1996) to generate pKK1407. Escherichia coli S17-1 λpir, tra strain (Simon et al., 1983) containing pKK1407 was used as a conjugal donor to either V. vulnificus MO6-24/O to generate the VvhA mutant KK1408 (Table 1).

To complement the vvpM mutation, an ORF of vvpM was amplified from the genomic DNA of V. vulnificus MO6-24/O by PCR with the primer pair VVPM001F and VVPM001R (Table S1) and then digested with BamHI and SacI. The amplified vvpM ORF was subcloned into the broad-host-range vector pRK415 (Keen et al., 1988) linearized with the same enzymes (Table 1) to result in pKK1535. E. coli S17-1 λ pir, tra strain (Simon et al., 1983) containing pKK1535 was used as a conjugal donor to VvpM mutant. The plasmid pKK1535 was delivered into the VvpM mutant by conjugation as described previously (Jang et al., 2016). Recombinant (r) VvpM protein (Lee M. A. et al., 2014; Lee et al., 2015a) was kindly provided by Prof. Kyu-Ho Lee (Sogang University, Korea). We have further checked the LPS contamination in rVvpM protein by using endotoxin quantitation kit (Pierce® LAL Chromogenic Endotoxin Quantitation Kit, Thermo Fisher Scientific Inc. Waltham, MA, USA). The level of endotoxin in 100 pg/mL of rVvpM was less than 0.003 EU. Thus, we suggest that rVvpM is a purified recombinant protein containing very low level of endotoxin and suitable for our experiments in this study.

Total RNA was extracted using the RNeasy Plus Mini Kit (Quiagen, Valencia, CA, USA). Reverse transcription (RT) was carried out with 3 μg of RNA using a Maxime RT premix kit (iNtRON Biotechnology, Sungnam, Korea). β-Actin was used as an endogenous control. The cDNA (5 μL) for Annexins family were amplified.

The real-time quantifications of pro-inflammatory cytokines and autophagy related genes were performed using a Rotor-Gene 6000 real-time thermal cycling system (Corbett Research, New South Wales, Australia) with a QuantiMix SYBR Kit (PhileKorea Technology, Daejeon, Korea) according to the manufacturer's instructions. β-Actin was used as an endogenous control.

Cells were synchronized in the G₀/G₁ phase by culture in serum-free media for 24 h before incubation of rVvpM The necrotic cell death was detected with an Annexin V and PI staining kit (BD Biosciences, Franklin Lakes, NJ) according to the manufacturer's instructions. Briefly, the cells were detached with 0.05% trypsin/EDTA and 1 × 10⁵ cells were resuspended with Annexin V binding buffer. And then the cells were stained with Annexin V (25 μg/ml) and PI (125 ng/mL), and incubated for 15 min at room temperature in the dark. The sample was read by flow cytometry and analyzed using CXP software (Beckman Coulter, Brea, CA). Samples were gated to exclude debris (forward light scatter [FSC] area vs. side scatter-area), and then any cell doublets were excluded using FSC-area vs. FSC-width analysis.

Cell death was also detected with an Apoptosis/Necrosis Detection kit (Abcame Cambridge, MA, USA) according to the manufacturer's instructions. Briefly, the cells were treated with Apopxin Green indicator as a phosphatidylserine marker and 7-aminoactinomycin D (AAD), and were then incubated in the dark for 60 min at room temperature. After the cells were rinsed with ice-cold PBS, the level of cell death was examined using a luminometer (Victor3; Perkin-Elmer, Waltham, MA) and quantified by measuring absorbance at excitation and emission wavelengths of 490 and 525 nm for detection of Apopxin Green Indicator or at excitation and emission wavelengths of 490 and 650 nm for detection of 7-AAD.

Cells were grown until 75% of the surface of the plate and transfected for 36 h with ON-TARGETplus siRNAs mixed by 4 different siRNAs specific for ANXA2, cav-1, PKCα, JNK, NF-κBp65, Atg5, and NLRP3 (GE Dharmacon, Lafayette, CO, USA) or non-targeting (nt) siRNA as a negative control (GE Dharmacon, Lafayette, CO, USA) with HiPerFect Transfection Reagent (QIAGEN, Valencia, CA, USA) according to the manufacturer's instructions. The siRNA efficacy for ANXA2, cav-1, PKCα, JNK, NF-κBp65, Atg5, and NLRP3 determined by Western blot was 71, 69, 76, 75, 69, 71, and 61%, respectively (Figure S1).

Cells were washed twice with ice-cold PBS, scraped into 2 mL of 500 mM sodium carbonate (pH 11.0), transferred to a plastic tube, and homogenized with a Sonicator 250 apparatus (Branson Ultrasonic, Danbury, CT) using three 20-sec bursts. The homogenate was adjusted to 45% sucrose by the addition of 2 ml 90% sucrose prepared in 2-(N-morpholino) ethanesulfonic acid (MES)-buffered solution consisting of 25 mM MES-buffer solution (pH 6.5) and 0.15 M NaCl and placed at the bottom of an ultracentrifuge tube. A 5–35% discontinuous sucrose gradient was formed (4 ml each of 5 and 35% sucrose, both in MES-buffer solution containing 250 mM sodium carbonate) and centrifuged at 40,000 × g for 20 h in a Beckman SW41 Rotor (Beckman Coulter, Fullerton, CA). Eight fractions were collected and analyzed by 12% SDS-PAGE.

Cells were fixed in 4% paraformaldehyde in PBS for 10 min at room temperature, permeabilized in 0.1% Triton X-100 in PBS for 5 min, and blocked in PBS containing 5% (v/v) normal goat serum (NGS) for 30 min at room temperature. Samples were then stained with primary antibody for overnight at 4°C. Following three washes with PBS, the samples were incubated with Alexa 488-conjugated goat anti-rabbit/mouse IgM (Invitrogen Co., Carlsbad, CA, USA), and counterstained with PI in PBS containing 5% (v/v) NGS for 2 h. After washing with PBS, samples were mounted on slides and visualized with an Olympus FluoView™ 300 confocal microscope with 400x objective. The expressions of LC3 in immunofluorescence image was quantified by using Image J software (NIH, Bethesda, MD), which measures the stained area per microscopic filed with consistent threshold. On the other hand, the cell numbers showing the membrane translocation of PKCα and the nuclear translocation of p-NF-κB were directly counted per random microscopic filed and converted the numbers to a percentage by multiplying by 100. Ten random fields per coverslip were counted.

2′, 7′-dichlorofluorescein diacetate (CM-H₂DCFDA) and MitoSOX™ Red Mitochondrial Superoxide Indicator (Thermo Fisher Scientific Inc.Waltham, MA, USA) were used to detect the intracellular and mitochondrial ROS production, respectively. To quantify the intracellular ROS levels, the cells treated with 10 mM CM-H₂DCFDA were rinsed twice with ice-cold PBS and then scraped. A 100 μL cell suspension was loaded into a 96-well plate and examined using a luminometer (Victor3; Perkin-Elmer, MA, USA) and a fluorescent plate reader at excitation and emission wavelengths of 485 and 535 nm, respectively.

Western blotting was performed as previously described with minor modifications (Lee S. J. et al., 2014). Cells were harvested, washed twice with PBS, and lysed with buffer (20 mM Tris [pH 7.5], 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 1 mg/mL aprotinin, and 1 mM phenylmethylsulfonylfluoride [PMSF]) for 30 min on ice. The lysates were then cleared by centrifugation (22,250 × g at 4°C for 30 min). Equal amounts of protein (20 μg) were resolved by 10~15% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene fluoride (PVDF) membranes. The membranes were washed with TBST solution (10 mM Tris-HCl [pH 7.6], 150 mM NaCl, and 0.05% Tween-20), blocked with 5% skim milk for 1 h, and incubated with appropriate primary antibody at 4°C for overnight. The membrane was then washed and incubated with a horseradish peroxidase-conjugated secondary antibody for 2 h. The bands were visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech Inc., Buckinghamshire, UK) and detected by using the Bio-rad ChemiDoc™ XRS+ System with images manipulated using the Image Lab™ software #170-8265 (Bio-Rad, Hercules, CA, USA). The subcellular fractionation method for the isolation of membrane and cytosolic proteins were previously reported (Cox and Emili, 2006).

The DNA binding activity of NF-κBp65 was determined by a NF-κBP65 transcription assay kit (Cayman chemical, Ann Arbor, MI) according to the manufacturer's instructions. Briefly, double-stranded DNA (dsDNA) sequence containing NF-κB p65 response element was incubated with 2 mg/ml nuclear extracts. DNA-bound NF-κBp65 complex was captured by an immobilized antibody specific for NF-κBp65 and detected by addition of specific secondary antibody conjugated to HRP. The absorption was measured at 450 nm.

Interaction of either Rac1 with ANXA2, NCF1, and caveolin-1 or ASC with NLRP3 was analyzed by immunoprecipitation and Western blotting. Cells were lysed with lysis buffer (1% Triton X-100 in 50 mM Tris–HCl pH 7.4 containing 150 mM NaCl, 5 mM EDTA, 2 mM Na₃VO₄, 2.5 mM Na₄PO₇, 100 mM NaF, 200 mM microcystin lysine–arginine, and protease inhibitors). Cell lysates (400 μg) were mixed with 10 μg of each antibodies. The samples were incubated for 4 h, mixed with Protein A/G PLUS-agarose immunoprecipitation reagent (Pierce, Rockford, IL, USA) and then incubated for an additional 12 h. The beads were washed four times, and the bound proteins were released from the beads by boiling in SDS-PAGE sample buffer for 5 min.

ChIP was performed using the EZ-ChIP kit (EMD Millipore, Billerica, MA, USA) according to the manufacturer's protocol. Briefly, cells were treated with 1% formaldehyde for 15 min to cross-link proteins to DNA, lysed, and then sonicated. The lysate was incubated with primary antibodies overnight at 4°C. The immunocomplex was purified by incubation with 60 μL of protein G-agarose beads for 1 h and eluted for DNA purification. Quantitative real-time PCR was performed with primers for the IL-1β promoter flanking the putative NF-κBp65 binding sites. The following primer sequences of IL-1β promoter were used (sence and antisence respectively): −237 ~ −56 bp, 5′- TCCACCACGATGA CACACTT -3′ and 5′- GCTGTGAAATTTTCCCTTGG -3′. Anti-RNA polymerase II and normal mouse IgG were used as the positive and negative control for immunoprecipitation, respectively. The human IL-1β promoter sequence was found using the Eukaryotic Promoter Database. The putative binding sites were predicted using Alggen Promo software, version 3.0.2 (Messeguer et al., 2002; Farre et al., 2003).

Genomic DNA from RAW264.7 cells was prepared with the QIAamp DNA Mini Kit (Qiagen, Valencia, CA, USA). The extracted DNA was treated with sodium bisulfite using EzWay™ DNA Methylation Detection Kit according to the manufacturer's instructions (KOMABIOTECH, Seoul, Korea). The methylation status of IL-1β gene was determined by methyl-specific PCR (MSP) analysis. We conducted MSP of IL-1β gene promoter containing the −328, −294, and −277 CpG sites due to their close distance to putative NF-κB binding sites. The following primer sequences were used (sense and antisense respectively): Methylated, 5′- TTTTAGTTTAAGTATAAGGAGGCGA-3′ and 5′-ACAC ATTCGCAAATATATCATCGTA-3′; Unmethylated, 5′-TTTTAGTTTAAGTATAAGGAGGTGA-3′ and 5′-ACACATTCACAAATATATCA TCATA -3′.

Cells treated with rVvpM were fixed in Karnovsky's solution and 2% OsO₄ in 0.1 M cacodylate at 4°C for 2 h, respectively. The cells were rinsed with distilled water briefly and stained with 0.5% uranyl acetate en bloc solution for overnight. After dehydration in graded ethanol series, samples were infiltrated with Spurr's resin. The samples were imaged by JEM1010 (JEOL, Tokyo, Japan) transmission electron microscope operating at 80 kV.

The mouse Inflammasomes RT2 Profiler PCR Array (Qiagen, Valencia, CA, USA) was used to analyze gene expression of inflammasome components and signaling pathways in a cells treated with rVvpM for 24 h according to the manufacturer's instructions. In this array a set of optimized primer assays allows the detection of mRNA transcripts of 84 genes as well as five housekeeping genes in a 96-well plate by real-time PCR. Arrays were performed a minimum of three times for each treatment. Functional gene grouping profiles were provided as available online at http://www.sabiosciences.com/rt_pcr_product/HTML/PAMM-097Z.html#howitwork. PCR array data were analyzed using the GeneGlobe Data Analysis Center on QIAGEN's website at http://www.qiagen.com/kr/shop/genes-and-pathways/data-analysis-center-overview-page/.

All animal procedures were performed following the National Institutes of Health Guidelines for the Humane Treatment of Animals, with approval from the Institutional Animal Care and Use Committee of Seoul National University (SNU-140108-4). Seven-week-old mice (n = 10) were received i.g. inoculation of boiled V. vulnificus (Cont.), V. vulnificus (WT), VvpM mutant, and VvpM complement at 1.3 × 10⁹ CFU/mL for 16 h and sacrificed. Half of ileum tissue was collected, washed, and homogenized. The homogenates of each organ were serially diluted and spread on LB agar containing polymyxin B (100 U/mL). CFUs were normalized to grams of intestinal tissues (CFU/g) to represent superficial bacterial counts. The other half were embedded in O.C.T. compound and stored at −70°C. Samples were then cut into 6-μm-thick frozen sections.

Frozen sections of tissue samples were subjected to hematoxylin and eosin (H&E) staining for histological examinations. Histological parameters were determined in a blinded fashion by two experienced gastrointestinal pathologists as previously described (Clark et al., 2006). Briefly, scores were assigned as follows: 0 = no damage (normal); 1 = slight submucosal and/or lamina propria separation (mild); 2 = moderate separation of the submucosa and/or lamina propria and/or edema in the submucosa and muscular layers (moderate); 3 = severe separation of the submucosa and/or lamina propria and/or severe edema in the submucosa and muscular layers with regional villous sloughing (severe); or 4 = loss of villi and necrosis (necrosis). Intermediate scores of × 0.0 or × 0.5 were also used to more precisely assign the degree of intestinal damage.

To determine the functional role of VvpM we performed further experiment by using the ileal-ligated mouse model. Before surgery, mice were fasted for 24 h and anesthetized by intraperitoneal injection of a 2:1 mixture of Zoletil™ (20 mg/kg, Virbac Laboratories, Carros, France) and Xylazine HCl (10 mg/kg, Rompun®, Bayer, Germany). While maintaining the body temperature at 37°C using a heating pad, a small abdominal incision was made and a loop of middle ileum of intestine was isolated by silk suture (2–3 cm in length). The closed ileal loop was instilled with 100 μl of phosphate-buffered saline (PBS) containing WT, vvpM mutant, and vvpM complement at 1.3 × 10⁹ CFU/mL for 2 h. After putting the ileal loop back into the peritoneal cavity, the cavity was closed with suture. At 2 h after the inoculation of both strains, the mice were sacrificed and the intestinal loops were removed for real-time PCR and the hematoxylin and eosin (H&E) staining.

Results are expressed as means ± standard errors (S.E.). All experiments were analyzed by ANOVA, followed in some cases by a comparison of treatment means with a control using the Bonferroni-Dunn test. Differences were considered statistically significant at P < 0.05.

V. vulnificus (WT) secretes functional virulence factors, such as multifunctional autoprocessing repeats in the toxin (RTX) (Lee et al., 2007; Jeong and Satchell, 2012), elastase (VvpE) (Lee et al., 2015a,b, 2016), cytolytic pore-forming hemolysin (VvhA) (Jeong and Satchell, 2012; Song et al., 2016), and zinc-metalloprotease (VvpM) (Lee M. A. et al., 2014; Lee et al., 2015a). RAW 264.7 murine macrophage cells were exposed to V. vulnificus (WT) and various mutants deficient in RTX, VvpE, VvhA, and VvpM genes. A robust increase in the IL-1β mRNA expression level were observed 3 h after incubation with 1 MOI of WT (Figure 1A). WT also stimulated the expression of both pro-IL-1β and cleaved IL-1β (Figure 1B), and the secretion of IL-1β into the cell media (Figure 1C). Interestingly, LPS as a positive control has similar ability with WT in the induction of IL-1β expression and the release. However, the level of IL-1β was far less evident in cells infected with the VvpM mutant as compared to cells infected with other WT mutants. Among the various pro-inflammatory cytokines, the expression of IL-1β appeared to be highly susceptible to VvpM mutant infection (Figure 1D). To determine the direct effect of VvpM, cells were exposed to various concentrations (0~200 pg/mL) of the recombinant protein (r) VvpM purified by V. vulnificus for 24 h. A significant increase in IL-1β expression was observed after 3 h of incubation with 100 pg/mL of rVvpM (Figure 1E). rVvpM preferentially increased the level of IL-1β mRNA among the mRNA amplicons of pro-inflammatory cytokines (Figure 1F). Interestingly, an increase in the active form of IL-1β (17 kDa) for 12 h (Figure 1G) and its secretion into the cell media for 24 h (Figure 1H) were also observed in a time-dependent manner after treatment with rVvpM in both RAW 264.7 cells and Caco-2 human intestinal epithelial cells. We determined the functional role of rVvpM to induce cell death by means of flow cytometric analyses over a period of 24 h (Figure 1I). rVvpM significantly induced the necrosis (an 8.5 ± 0.3-fold increase compared to a vehicle) of the cells, whereas for apoptotic cell death, a marginal effect was noted (a 1.3 ± 0.2-fold increase compared to the vehicle). We further determined the functional role of rVvpM to induce cell death by staining with annexin V for apoptotic cells, PI for necrotic cells, and CytoCalcein for live cells. Consistent with our data for flow cytometric analyses, we found that rVvpM is essential for triggering the necrotic cell death rather than the apoptosis (Figure S2). We further confirmed the necrosis-promoting effect of rVvpM using another reagent that monitors necrotic cells with 7-aminoactinomycin D. As shown in Figure S3A, we found that rVvpM shows greater stimulatory potency on necrotic cell death than on apoptotic cell death, confirming that rVvpM is essential for triggering necrotic macrophage death rather than apoptosis. In addition, the levels of IL-1β and necrotic cell death remained unchanged after treatments with identical amounts of trypsin, boiled rVvpM, or boiled WT for 12 h (Figure S3B), suggesting that the functional roles of rVvpM during the processes of IL-1β production and necrotic cell death differ from the outcome of the general proteolysis of surface proteins or contamination by the bacterial product. Thus, these results indicate that rVvpM is responsible for triggering necrotic macrophage death during V. vulnificus infection and that it evokes a pro-inflammatory response via mainly IL-1β production.

It is well established that annexin A2 (ANXA2) plays a critical role as a host mediator of bacterial pathogens and is essential in the pathogenesis of Enteropathogenic E. coli infections (Zobiack et al., 2002). We have previously reported that exotoxins produced by V. vulnificus contributes to their pathogenesis in multiple ways by interacting with intestinal proteins, including ANXA2 and ANXA2. However, our previous data have revealed that ANXA2, but not ANXA4 plays a critical role as a membrane mediator of V. vulnificus and that it has unique biological properties responsible for host cell death coupled with an inflammatory response (Lee et al., 2015a). We initially determined the existence of ANXA isotypes in RAW264.7 cells. ANXAs are expressed in the order of ANXA2, ANXA3, ANXA5>ANXA1, ANXA4, ANXA6 (Figure S4). The expressions of ANXA7-13 were not detected in these cell lines. Interestingly, the knockdown of ANXA2 resulted in a reduction of necrotic cell death (Figure 2A) and IL-1β production (Figure 2B), as elicited by rVvpM. In attempting to define the membrane location of functional ANXA2, our data revealed that the lipid raft markers caveolin-1 and flotillin-2 were found in mainly fraction 4, whereas ANXA2 was highly enriched both in the lipid raft parts (fraction 4) and the non-lipid raft parts (fractions 6–8) (Figure 2C). Moreover, the subunits of the NADPH oxidase (NOX) enzymes, NOX2 (gp91^phox) and NCF1 (p47^phox), and a cytosolic component of the NOX family, Rac1, were enriched in fraction 4. Given that caveolin-1 is constitutively expressed in lipid rafts to maintain cell homeostasis forming caveolin-1-mediated caveoli (Li et al., 2007), it is possible that individual lipid rafts attached with caveolin-1 are present in fraction 4. These individual lipid rafts were known to be dynamic microdomains and carry several membrane-bound or attached proteins or enzymes such as G-proteins, protein kinases, and NOX enzymes (Li et al., 2007). Interestingly, an increase in the level of ANXA2 together with caveolin-1, but not flotillin-2, appeared in fraction 3 after incubation with 100 pg/mL of rVvpM compared to the vehicle alone, suggesting that rVvpM enhances the recruitment of ANXA2 into a caveolin-1-enriched component. In addition, a treatment with rVvpM resulted in the translocations of NOX2, NCF1, and Rac1 into fraction 3 including the caveolin-1-enriched component. This indicate that rVvpM stimulates the movement and clustering of individual lipid rafts without flotillin-1 to form a number of lipid rafts macrodomains or platforms with aggregation or recruitment of receptors, NOX enzymes, and other proteins such as Rac1 at fraction 3. Regarding the exception of flotillin-1 move in in the cells activated by rVvpM, it reported that flotillin-1 is independently regulated and does not strictly correlate with the expression patterns of caveolin family members (Volonte et al., 1999). Indeed, it reported that the distribution patterns of flotillin 1 did not change upon cholesterol removal (Kaakinen et al., 2012). Therefore, our result in the present study indicate that rVvpM initiates the organization of rafts and the spatial distribution of caveolin which are crucially dependent on lipid raft clustering in fraction 3. The effect of rVvpM on the membrane location of ANXA2 was further visualized by staining the ANXA2 and the lipid raft marker molecule cholera toxin subunit B (CTB). As shown in Figure 2D, rVvpM significantly increased the co-localization of CTB with ANXA2. In addition to ANXA2, CTB significantly co-localized with caveolin-1 and Rac1 in the cells treated with rVvpM. Because the above approaches are qualitative at best, we also attempted to quantify the results by means of the co-immunoprecipitation of Rac1 with proteins related to the lipid rafts in the presence of rVvpM. It was noted that Rac1 co-immunoprecipitated with ANXA2, caveolin-1, and NCF1, and importantly, that these interactions were enhanced by the rVvpM treatment (Figure 2E). In good agreement with the above results, the knockdown of caveolin-1 significantly abrogated rVvpM-induced necrosis (Figure 2F) and the production of IL-1β (Figure 2G). These findings indicate that rVvpM facilitates the lipid-raft-mediated clustering of ANXA2 and NOX enzymes to trigger IL-1β production coupled with necrotic macrophage death.

We subsequently determined whether rVvpM regulates the production of ROS with regard to the functional role of the lipid raft-mediated clustering of NOX enzymes. We found that rVvpM significantly increased ROS production between 10 and 60 min after incubation with 100 pg/mL of rVvpM compared to the vehicle alone (Figure 3A). The ROS production induced by rVvpM at 30 min was significantly inhibited by the knockdown of caveolin-1 and ANXA2 and by a treatment with the positive control N-acetylcysteine (NAC) as an antioxidant (Figure 3B). Despite the significant role of rVvpM in promoting autophagy induction (Lee M. A. et al., 2015), the autophagy inhibitor 3-methyladenine (3-MA) did not affect rVvpM-induced ROS production, suggesting that the signaling events related to ROS production are unrelated to the characteristics of autophagy. Moreover, we found that mitochondrial complex I inhibitor, rotenone (0.1 μM) has a very weak inhibitory effect on ROS production induced by rVvpM for 30 min. Consistently, rVvpM did not show any significant effect on mitochondiral ROS production (Figure S5). These results indicate that epithelial ROS generated by clustering of NADPH oxidase enzymes within lipid rafts facilitate the production of ROS, which results in a prominent amplification of the transmembrane signal at short time period. Importantly, a prominent increase in ROS production was also observed in cells infected with WT (Figure 3C). In contrast, while mutants deficient in VvpM and RTX significantly decreased the ability of WT to induce ROS production, macrophages infected with VvpE and VvhA mutants revealed only a modest defect. Moreover, the instances of necrotic cell death (Figure 3D) and IL-1β production (Figure 3E) induced by rVvpM were significantly blocked by a pre-treatment with NAC. These results indicate that ROS production is required for VvpM to promote IL-1β production coupled with necrotic macrophage death. ROS produced by the clustering of NADPH oxidase enzymes results in a prominent amplification of the transmembrane signal through the activation of PKC (Wu et al., 2008; Zhang et al., 2010). rVvpM significantly induced pan-PKC phosphorylation between 30 and 60 min (Figure 3F). In an experiment to identify the specific PKC isotypes, the translocation of PKCα, but not PKCδ or PKCζ, from the cytosol to the membrane compartment was observed after cells were treated with rVvpM for 30 min (Figure 3G). Importantly, the increase in PKCα phosphorylation was significantly blocked by a pretreatment with a ROS scavenger, NAC, but not by 3-MA (Figure 3H), indicating that PKCα phosphorylation is a downstream factor of ROS production induced by rVvpM acting on the membrane lipid raft, which is not affected by autophagy. The membrane expression of PKCα phosphorylation was further confirmed by immunofluorescence staining in rVvpM-treated RAW264.7 cells (Figure 3I). Not surprisingly, while WT significantly increases the level of PKCα phosphorylation, the VvpM mutant showed the most similar level to that noted in control mice compared to the other mutant forms (Figure 3J), indicating that PKCα phosphorylation mediated by ROS is a more efficient infectious mechanism of VvpM as compared to RTX. It was noted that rVvpM induces phosphorylation of PKCα at Ser 657 site responsible for PKCα autophosphorylation that leads to stabilization and maturation of the PKCα. In addition, rVvpM significantly stimulated calcium influx, which was enhanced by A23187 as a positive control (Figure S6A). Interestingly, the silencing of PKCα by PKCα siRNA significantly blocked the necrotic cell death (Figure 3K) and IL-1β production (Figure 3L) induced by rVvpM. Given that previous result showing the PKC activation also able to stimulate the ROS production via phosphorylation of NOX subunits (Cosentino-Gomes et al., 2012), we have further examined whether PKCα regulates ROS production in RAW264.7 cells treated with rVvpM. However, the knockdown of PKCα failed to regulate ROS production induced by rVvpM (Figure S6B), PKCα is downstream factor of ROS during necrotic macrophage death induced by rVvpM. Together, these results indicate that rVvpM coupling with ROS production uniquely stimulates atypical PKC activation to promote IL-1β production and necrotic macrophage death.

Many bacterial stimuli regulate the MAPKs and NF-κB pathways, which are interesting candidates as downstream mediators of ROS and PKC. rVvpM increased the phosphorylation of JNK between 15 and 60 min but does not affect the phosphorylation of ERK or p38 MAPK (Figure 4A), and its effect at 30 min is inhibited by the silencing of PKCα, but not by 3-MA (Figure 4B). In addition, the silencing of JNK by siRNA has a significant inhibitory effect on the necrosis (Figure 4C) and production of IL-1β (Figure 4D) in RAW264.7 cells. rVvpM also induced NF-κBp65 phosphorylation between 60 and 120 min (Figure 4E). Treatment with JNK siRNA but not 3-MA significantly blocked rVvpM-induced NF-κBp65 activation (Figure 4F). The increased accumulation of NF-κBp65 phosphorylation in the nucleus was further confirmed by immunofluorescence staining and counter-labeling with propidium iodide (PI) (Figure 4G). In agreement with this, the necrosis (Figure 4H) and production of IL-1β (Figure 4I) induced by rVvpM were significantly blocked by the knockdown of NF-κBp65. The effect of rVvpM on the activation of JNK and NF-κBp65 was further confirmed in these cells following infection with WT and various deficiency mutants (Figure 4J). WT significantly induced the activation of JNK and NF-κBp65, whereas for the VvpM mutant, only a marginal effect was noted, and the cells appeared to be highly susceptible to JNK among the various mutants tested here, although all four deficiency mutants caused significant defects in the ability of WT to activate NF-κBp65. Consistently, our results have revealed that all mutations have inhibitory effects on transcription activity of NF-κB induced by WT in RAW 264.7 and Caco-2 cells (Figures S7A,B). The data above provide important evidence that JNK-mediated NF-κBp65 phosphorylation is required for the regulation of IL-1β production and necrotic cell death as evoked by rVvpM. In addition, the phosphorylation of JNK and NF-κB induced by rVvpM was significantly inhibited by a treatment with a ROS scavenger N-acetylcysteine (NAC) (Figure S7C), suggesting that ROS induced by rVvpM plays key role in activation of PKC, JNK, and NF-κB in the promotion of necrotic macrophage death

To determine the role of NF-κB in the transcriptional regulation of IL-1β mRNA expression, we conducted chromatin immunoprecipitation (ChIP) assays followed by qRT-PCR in cells treated with rVvpM. With ALGGEN PROMO (Messeguer et al., 2002; Farre et al., 2003), we found that the IL-1β promoter located between the transcription start site and the point–500 bp upstream contains two putative NF-κB binding sites and six methylation sites. Test results also found a primer that includes two putative NF-κB binding sites at the region proximal −56 bp to −237 bp of the start site of the IL-1β promoter (Figure 5A). Figure 5B shows that our primer sets result in an amplicon from the anti-phospho-NF-κBp65 immunoprecipitates and, importantly, that the interaction between NF-κB and the IL-1β promoter was enhanced by a rVvpM treatment for 60 min. Interestingly, however, the level of NF-κBp65 binding to the IL-1β promoter was significantly inhibited by the silencing of ANXA2 and JNK (Figure 5B) but not by treatments with 3-MA or NLRP3 siRNA, which is a critical regulator of IL-1β production (Lamkanfi et al., 2011; Guo et al., 2015; Figure 5C). The ability of rVvpM to induce interaction between NF-κB and the IL-1β promoter is consistent with the data from a real-time PCR analysis (Figure 5D). These results indicate that rVvpM acting on the lipid raft-mediated clustering of ANXA2 transcriptionally regulates the NF-κBp65 binding to the IL-1β promoter via JNK activation. Subsequently, we postulated that rVvpM may affect methylation, which negatively regulates the interaction between NF-κB and the IL-1β promoter. We attempted to analyze the methylation status of the IL-1β gene promoter containing the −328, −294, and −277 CpG sites due to their close proximity to putative NF-κB binding sites. As shown in Figure 5E, a primer pair including these CpG sites that specifically amplified either the methylated or unmethylated form of the IL-1β promoter produced 203 bp methylated bands from vehicle-treated RAW264.7 cells, indicating that the IL-1β promoter was methylated in these alleles. However, treatment with 100 pg/mL of rVvpM for 60 min markedly inhibited the level of IL-1β promoter methylation, which was significantly blocked by the knockdown of ANXA2 and JNK and by a pre-treatment with NAC. As expected, treatment with 3-MA and NLRP3 siRNA failed to regulate the methylation status of the IL-1β gene promoter induced by rVvpM. These results were consistent with the data from a real-time PCR analysis quantified by the relative value of CpG methylation compared to an unmethylated form in the IL-1β promoter (Figure 5F). Interestingly, the levels of necrotic cell death (Figure 5G) and the production of IL-1β (Figure 5H) were markedly enhanced by a pre-treatment with the DNA methylation inhibitor 5-azacytidine (5-aza). These results suggest that the hypomethylation of the IL-1β promoter is able to affect the high level of the interaction between NF-κB and the IL-1β promoter during necrotic macrophage death as evoked by rVvpM.

Despite the independent roles of autophagy and inflammasome in regulating the ROS-PKCα-JNK-NF-κBp65 pathway through the lipid-raft-mediated clustering of ANXA2, we revisited the issue of whether rVvpM has the ability to regulate the autophagic process and inflammasome formation due to the fundamental parts of these processes play role in regulating the production of IL-1β during bacterial infection. Importantly, while the expressions of LC3-II and Beclin-1 were increased by a treatment with rVvpM in a time-dependent manner, the level of p62 was attenuated (Figure 6A), indicating that rVvpM has the ability to regulate the autophagic process. We also confirmed through confocal microscopy that rVvpM increases autophagic vesicle formation, during which a significant increase in LC3 puncta positive cells was observed (Figure 6B). Comparing the functional roles of VvpM with those of other virulence factors which evoke the autophagic responses, we found that WT markedly induced the expression levels of LC3-II and Beclin-1 while inhibiting that of p62 (Figure 6C). In contrast, cells infected with the VvpM mutant among all cells infected with the mutants tested here exhibited the lowest LC3-II level similar to that of the control, suggesting that VvpM is a major virulence factor responsible for autophagy induction. Moreover, the functional effects of the VvpM mutant in LC3 activation were significantly restored by the complementation of the VvpM mutant with a functional VvpM gene (Figure S8). A unique feature of autophagosomes characterized by the peculiar double-membraned vesicle was also observed in cells treated with rVvpM by transmission electron microscopy (TEM), where rVvpM caused significant abnormalities such as swollen mitochondria while also widening the endoplasmic reticulum (ER) (Figure 6D). To confirm the enhanced autophagic flux, we used bafilomycin A1 (BafA1) which is a known inhibitor of V-ATPase (vacuolar-type H⁺ ATPase) that prevents the maturation of autophagic vacuoles. rVvpM significantly enhanced the accumulation of LC3-II in cells pretreated with BafA1 (Figure S9), suggesting that the lysosomal degradation pathway was not inhibited by rVvpM. Importantly, the autophagy inhibitor 3-MA significantly blocked necrotic cell death (Figure 6E) and IL-1β production (Figure 6F) as induced by rVvpM. This indicates that rVvpM differentially regulates IL-1β production through the induction of autophagic macrophage death. It was unclear as to whether the activation of two different signaling pathways for the regulation of IL-1β production reflects the selective interaction between VvpM and the lipid raft. We focused on the fact that ANXA2 was also highly enriched in the non-lipid raft parts of the cell membrane. Notably, the level of LC3-II evoked by rVvpM was markedly inhibited by the knockdown of ANXA2 with ANXA2 siRNA (Figure 6G). However, the knockdown of caveolin-1 (Figure 6H) and treatment with the lipid raft sequester MβCD failed to regulate the level of LC3-II in cells treated with rVvpM (Figure S10). These results indicate the ANXA2 in the non-lipid raft parts is functionally relevant for autophagy induction and that it distinctively regulates necrotic macrophage death and the pro-inflammatory response as a host mediator of rVvpM. To gain insight into how ANXA2 influences the autophagic process via the non-lipid raft parts, we further determined the effect of rVvpM on the expression of mRNA for induction, expansion/closure, and fusion/degradation in an autophagy system (Figure 6I). The prominent expressions of Atg5 and Atg16L1 for autophagy expansion/closure were observed following an rVvpM treatment with a trend toward increased mRNA expressions in an overall autophagy system. Intriguingly, the knockdown of Atg5 in RAW264.7 cells reduced the levels of necrotic cell death (Figure 6J) and IL-1β production (Figure 6K), suggesting the involvement of Atg5 in IL-1β production coupled with autophagic macrophage death elicited by rVvpM.

Given that RTX and VvhA produced by V. vulnificus are also known to cause the death of phagocytic cells via NLRP3-dependent caspase-1 activation in vitro and in vivo (Toma et al., 2010; Lo et al., 2011), it is important to clarify whether rVvpM regulates the expression of genes associated with the assembly, activation, and down-stream signaling of inflammasomes, which are linked to autophagy (Yuk and Jo, 2013). Our results from a RT² Profiler PCR array showed that expression changes were detected in 13 of the 84 genes studied here after a 24 h rVvpM treatment (Figures 7A,B). Expression levels were increased in 10 genes (>two-fold, as shown by the green bar graph) but decreased in the other three genes (>two-fold, as shown by the blue bar graph). The remaining 61 genes showed no significant change in comparison with a vehicle treatment. The 10 genes showing increased expression levels were NLRP3 (NM_145827), Casp1 (caspase-1, NM_009807), Tnf (tumor necrosis factor, NM_013693), Ccl7 (chemokine ligand 7, NM_013654), Ikbkb (IKK-β, NM_010546), IL-1β, Map3k7 (TGF-β activated kinase 1, NM_172688), Mapk8 (JNK1, NM_016700), Mapk9 (JNK2, NM_016961), and Traf6 (Tnf receptor-associated factor 6, NM_009424), while those showing decreased expression levels were Ctsb (cathepsin B, NM_007798), Ccl12 (chemokine ligand 12, NM_011331), and Cxcl3 (chemokine (C-X-C motif) ligand 3, NM_203320). Among these differentially regulated genes, we identified two genes (IL-1β and NLRP3) which were highly expressed in cells treated with rVvpM compared to a treatment with a vehicle (>10-fold). Importantly, the necrotic cell death (Figure 7C) as well as the production of IL-1β (Figure 7D) induced by rVvpM were significantly abrogated by NLRP3 knockdown. These results indicate VvpM is another virulence factor of V. vulnificus to regulate NLRP3 inflammasome responsible for IL-1β production coupled with necrotic macrophage death. Consistently, our results have revealed that rVvpM has ability to induce speck formation of pyrin domain of the adaptor protein, ASC (apoptosis-associated speck-like protein containing a CARD) (Figure S11). We also found that the ASC was co-immunoprecipitated with NLRP3, and importantly, that the interaction between ASC and NLRP3 was enhanced by a treatment with rVvpM (Figure 7E). In contrast, these interactions induced by rVvpM were significantly blocked by a treatment with the autophagy inhibitor 3-MA and knockdown of Atg5 expression, suggesting that the formation of NLRP3 inflammasomes induced by rVvpM is an autophagy-dependent process. An increase in the active form of caspase-1 was also observed after 6 h of incubation with 100 pg/mL of rVvpM (Figure 7F). However, the activation of caspase-1 induced by rVvpM was markedly inhibited by transfection with siRNAs for NLRP3 (Figure 7G). Importantly, WT significantly induced the activation of caspase-1 and IL-1β maturation, whereas the VvpM mutant showed a marginal effect and appeared to be highly susceptible among the various mutants (Figure 7H). Overall, these novel findings demonstrate that VvpM facilitates the formation of NLRP3 inflammasomes via autophagy activation to regulate caspase-1-mediated IL-1β production.

To determine whether the functional role of VvpM in promoting IL-1β production coupled with macrophage cell death contributes to the growth of V. vulnificus in the gut, we further determined the level of bacterial colonization in mice inoculated intragastrically with a control, WT, and with the VvpM mutant and VvpM complementation (comp) at 1.3 × 10⁹ CFU/mL for 16 h (Figure 8A). WT colonization in the ileum was increased by 4.6 ± 1.1 (× 10⁶ CFU/g tissue) at 16 h. However, when mice were inoculated with the VvpM mutant, the levels were diminished by 3.3 ± 0.6 (× 10⁶ CFU/g tissue) compared to those associated with the WT. In contrast, complementation of the VvpM mutant with a functional VvpM gene completely reversed the effect of WT on ileal colonization. It was noted that all of the mice given oral administrations of WT had survived at 18 h post-injection (data not shown). WT also induced severe inflammation of the intestine, where it caused shortened villi heights accompanied by increased numbers of inflammatory cells at 16 h infection, resulting in an increased histopathological damage score compared to that in the control mice (Figure 8B). However, the VvpM mutant significantly failed to regulate intestinal villi structures and the related inflammation. In contrast, complementation of the VvpM mutant with a functional VvpM gene restored the effect of the VvpM mutant on the histopathological damage score. These results indicate that VvpM contributes to V. vulnificus colonization, facilitating pro-inflammatory responses in vivo.

To eliminate the possibility that the lower inflammatory responses evoked by the VvpM mutant can simply be explained by the presence of less bacteria, we undertook an additional experiment using the ileal-ligated mouse model in which a closed ileal loop was instilled with 100 μl of PBS containing WT, the VvpM mutant, or VvpM complementation at 1.3 × 10⁹ CFU/mL for 2 h. We anticipated that 2 h was the minimum duration for enterotoxicity and pro-inflammatory responses based on our recent report (Kim et al., 2013). There were no statistical differences between the groups treated with WT and the VvpM mutant in terms of the bacterial growth (data not shown). However, the injection of mice with the VvpM mutant at a time of 2 h failed to elevate the histopathological damage score (Figure 8C) or the expression levels of pro-inflammatory cytokines (Figure 8D) as caused by WT infection. The results in Figure 8E revealed that inoculation with WT for 2 h significantly increased the expressions of LC3-II, Beclin 1, Caspase-1 and mature IL-1β as well as the phosphorylation of JNK and NF-κB in mice, which were prevented by infection with the VvpM mutant. Notably, the functional effects of the VvpM mutant were significantly restored by the complementation of the VvpM mutant with a functional VvpM gene (Figure 8E). Thus, these results suggest that VvpM is a major virulence factor of V. vulnificus responsible for IL-1β production coupled with necrotic cell death.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.