Characteristic Metabolic Changes in Skeletal Muscle Due to Vibrio vulnificus Infection in a Wound Infection Model

ABSTRACT Vibrio vulnificus is a bacterium that inhabits warm seawater or brackish water environments and causes foodborne diseases and wound infections. In severe cases, V. vulnificus invades the skeletal muscle tissue, where bacterial proliferation leads to septicemia and necrotizing fasciitis with high mortality. Despite this characteristic, information on metabolic changes in tissue infected with V. vulnificus is not available. Here, we elucidated the metabolic changes in V. vulnificus-infected mouse skeletal muscle using capillary electrophoresis time-of-flight mass spectrometry (CE-TOFMS). Metabolome analysis revealed changes in muscle catabolites and energy metabolites during V. vulnificus infection. In particular, succinic acid accumulated but fumaric acid decreased in the infected muscle. However, the virulence factor deletion mutant revealed that changes in metabolites and bacterial proliferation were abolished in skeletal muscle infected with a multifunctional-autoprocessing repeats-in-toxin (MARTX) mutant. On the other hand, mice that were immunosuppressed via cyclophosphamide (CPA) treatment exhibited a similar level of bacterial counts and metabolites between the wild type and MARTX mutant. Therefore, our data indicate that V. vulnificus induces metabolic changes in mouse skeletal muscle and proliferates by using the MARTX toxin to evade the host immune system. This study indicates a new correlation between V. vulnificus infections and metabolic changes that lead to severe reactions or damage to host skeletal muscle. IMPORTANCE V. vulnificus causes necrotizing skin and soft tissue infections (NSSTIs) in severe cases, with high mortality and sign of rapid deterioration. Despite the severity of the infection, the dysfunction of the host metabolism in skeletal muscle triggered by V. vulnificus is poorly understood. In this study, by using a mouse wound infection model, we revealed characteristic changes in muscle catabolism and energy metabolism in skeletal muscle associated with bacterial proliferation in the infected tissues. Understanding such metabolic changes in V. vulnificus-infected tissue may provide crucial information to identify the mechanism via which V. vulnificus induces severe infections. Moreover, our metabolite data may be useful for the recognition, identification, or detection of V. vulnificus infections in clinical studies.

seawater (4,5). In severe cases, the invasion of bacteria into skeletal muscle is commonly found not only in wound infection but also in oral infection (6,7). Moreover, patients with underlying diseases, such as chronic liver disease, diabetes, hemochromatosis, immunocompromised states, or serum iron-elevating conditions, may develop a severe infection that leads to septicemia and necrotizing manifestation (3,8). V. vulnificus-induced acute septicemia entails a high mortality rate, exceeding 50% within 24 to 72 h after infection (9,10). In addition, acute V. vulnificus infection induces severe rapidly progressive necrotizing fasciitis and tissue necrosis in host skeletal muscle. According to epidemiological studies, V. vulnificus exhibited the highest rate of infection and accounted for 25% of the Vibrio species detected in parenteral infections in the United States (1). Another study reported that V. vulnificus causes V. vulnificus necrotizing skin and soft tissue infection (VNSSTIs), and more than 95% of cases of VNSSTIs are associated with the subtropical western Pacific and Atlantic coastal regions of the Northern Hemisphere, such as Taiwan, South Korea, Japan, or the Gulf of Mexico (11).
Vibrio vulnificus hemolysin (VVH) and the multifunctional-autoprocessing repeatsin-toxin (MARTX) are regarded as significant toxins of V. vulnificus. Researchers have identified several functions for each toxin. VVH is a pore-forming toxin that promotes bacterial translocation from the intestine to the bloodstream by causing intestinal tissue damage with cell death (12). VVH exerts several functions in host cells, such as causing permeabilization of the intestinal epithelium in association with apoptosis and cell death, with reactive-oxygen-species (ROS) production (13) or autophagy activation (14). MARTX is commonly conserved in several Gram-negative bacteria, including another Vibrio species (15). MARTX is associated with numerous cytopathic and cytotoxic functions in eukaryotic cells, as it causes cytoskeletal dysfunction (16), inflammasome activation (17), inhibition of phagocytosis (18,19), and induction of apoptosis (20). Several studies have reported that V. vulnificus toxins injured host epithelial cells with the disruption of the host mitochondrial function, and that these damages played a role in the disruption of the central metabolic system in host cells in vitro (21,22). Those studies suggest that V. vulnificus induces the degradation of skeletal muscle and host metabolic changes via the action of its characteristic toxins. Necrotizing fasciitis is a major symptom of V. vulnificus infection that is found in severe cases of infection with dramatic symptoms in skeletal muscle. However, the relationship between necrotizing fasciitis and the metabolic changes that occur during V. vulnificus infection remains unknown.
Omics analyses (genomics, transcriptomics, proteomics, and metabolomics) have been used to investigate the mechanism of necrotizing fasciitis in bacterial infection. In particular, a metabolic analysis indicated differences in metabolites during the acute phase of infection in patients. Moreover, previous studies have indicated that bacterial growth and viability were closely associated and resulted in host metabolic changes. In fact, a variety of group A streptococci (GAS) respond to specific metabolites (23). According to a recent study, metabolic changes in infected tissues of the host may contribute to promoting bacterial growth and/or virulence, which exacerbates host symptoms and physiological conditions (24). However, it is unknown how V. vulnificus triggers metabolic changes in the host. To identify the mechanism of bacterial infection, it is necessary to investigate the metabolic changes that occur in V. vulnificus-infected tissues.
In this study, we attempted to estimate the metabolic changes in V. vulnificusinfected skeletal muscles in a mouse wound infection model using capillary electrophoresis-time-of-flight mass spectrometry (CE-TOFMS). We found that muscle catabolites changed upon infection by Vibrio spp., whereas V. vulnificus alone significantly affected the succinate-fumarate pathway of the tricarboxylic acid (TCA) cycle in skeletal muscle. The V. vulnificus toxin MARTX facilitated the proliferation of bacteria in skeletal muscle and was indirectly important for triggering metabolic changes in mouse skeletal muscle. Even during short infection, these metabolic changes were detected in skeletal muscle. This study is the first report showing metabolic changes in V. vulnificus-infected muscle tissue.

RESULTS
Catabolites of mouse skeletal muscle were detected in the mouse skeletal muscle and blood after infection with V. vulnificus. Prior to the investigation of the metabolic changes in the V. vulnificus wound infection model, we checked that V. vulnificus induces inflammation of skeletal muscle (25). After 9 h of infection, V. vulnificus-injected skeletal muscle tissues exhibited edema and neutrophil infiltration (see Fig. S1A in the supplemental material) and exhibited a high transcriptional level of MIP-2, TNF-a, and IL-6 (Fig. S1B). These data showed that V. vulnificus induced inflammatory reactions in the injected tissues of our mouse model.
In accordance with inflammation of the skeletal muscle, next we measured changes in the metabolite profile of the infection model mice. Creatine has been used as a creatine-phosphokinase-mediated muscle catabolism indicator in general clinical tests (26). Therefore, to estimate the muscle catabolism in V. vulnificus infection, we analyzed the metabolic conversion of ornithine to creatinine in host skeletal muscle after 9 h of V. vulnificus infection. Mouse skeletal muscles were treated by subcutaneous injection into the right femoral region of phosphate-buffered saline (PBS; control) or V. vulnificus (Vv; wild type), and each opposite leg was defined as OP (leg opposite the PBS leg) or OV (leg opposite the V. vulnificus leg). These skeletal muscles were collected after 9 h of infection, and the metabolites were analyzed by CE-TOFMS. We found that creatine and creatinine were decreased in the Vv group (Fig. 1A). Conversely, among the blood metabolites, creatine and creatinine were increased in the Vv group sample (Fig. S1C). In this mouse model, we found an increase of creatine and creatinine in blood similarly to the general clinical test, whereas we found a decrease of creatine and creatinine in V. vulnificus-infected skeletal muscle. These data indicated that V. vulnificus infection causes metabolic changes in skeletal muscle and imply the existence of other uncovered metabolic changes.
Metabolic changes in the skeletal muscles of the V. vulnificus wound infection mouse model. Next, we conducted a comprehensive analysis of skeletal muscle metabolites and searched for significantly altered metabolites. A total 110 metabolites were identified (54 and 56 metabolites were identified in the cationic and anionic mode, respectively), using CE-TOFMS (Table S1). These data were analyzed via a principal-component analysis (PCA) and a hierarchical clustering heat map analysis. The PCA plots indicated a clear metabolite difference between the metabolites of the PBS and Vv groups (Fig. 1B). The heat map data showed characteristic metabolite patterns in the PBS and Vv groups (Fig. 1C). These data showed that V. vulnificus infection induced apparent metabolic changes. Furthermore, a volcano plot showed that there were 13 increased metabolites and 36 decreased metabolites in V. vulnificus-infected skeletal muscle compared with those of PBS-treated muscle (Fig. 1D). The data show the relative values of each metabolite identified by CE-TOFMS in V. vulnificus-infected mice. The detected metabolites are classified into six metabolic pathways (tricarboxylic acid [TCA] cycle, carbohydrate metabolism, pentose phosphate pathway, amino acid metabolism, ornithine cycle, and nucleic acid metabolism). Those data indicated changes in metabolite associated with infection damage (Fig. 1E). Upon analysis of each metabolic pathway, the amino acid levels in infected tissues were elevated, which suggests the degradation of proteins in infected tissues. Analysis of catabolism in tissues infected with V. vulnificus showed that most metabolites, such as nucleic acids, pentose phosphate, and carbohydrates, were decreased. Conversely, the metabolites of the TCA cycle were specifically increased and decreased. In this experiment, we also analyzed the metabolites in blood samples from PBS-treated mice and the V. vulnificus-infected mouse wound infection model (Table S1). In contrast with the muscle samples, we could not identify a characteristic difference in blood samples using PCA (Fig. S2A), the heat map data (Fig. S2B), or the volcano plot (Fig. S2C). Furthermore, most metabolic pathways were not changed in the V. vulnificus-infected group relative to the PBS-treated group (Fig. S2D). These data indicate that V. vulnificus affected the host metabolism and that its metabolic changes were specifically induced at the site of infection in our mouse wound infection model.
Changes to the TCA cycle caused by V. vulnificus infection. The TCA cycle is a central pathway for oxidative phosphorylation that works in the generation of energy. These enzymes are localized in mitochondria, which are organelles that are often injured by x axis, 63.46%; y axis, 13.43%; z axis, 9.03% (3/group). (C) Heat map analysis of the metabolic changes to the fold change of each metabolite using the autoscaling method. Red and blue indicate high and low concentrations of metabolites, respectively (3/group). (D) Volcano plot showing relative changes in the metabolites in the Vv group, with red denoting significantly changed metabolites (by more than 2-fold), blue indicating metabolites that changed (by more than 2-fold) but not significantly, and orange indicating significant differences which were less than 2-fold (3/group). (E) Volcano plot showing relative changes of the metabolites in the Vv group, categorized into six metabolic pathways (TCA cycle, carbohydrate metabolism, pentose phosphate pathway, amino acid metabolism, ornithine cycle, and nucleic acid metabolism) (3/group). P values were determined by one-way ANOVA followed by Dunnett's test. *, FDR , 0.05 compared with the Vv group. bacterial infections. In fact, P. aeruginosa modulates the host TCA cycle in a variety of tissues (27). To capture the metabolic changes of the TCA cycle in V. vulnificus infection, we measured the organic acids in the TCA cycle (i.e., citric acid, cis-aconitic acid, isocitric acid, 2-oxoglutaric acid, succinic acid, fumaric acid and malic acid) ( Fig. 2A and B). In V. vulnificus-infected muscle, the citric acid and succinic acid was increased, whereas fumaric acid and malic acid were decreased. Interestingly, the pattern of metabolic change of succinic acid was completely opposite to that of fumaric acid. In contrast to V. vulnificus-infected muscle, metabolic change patterns were not found in blood. These data indicated that V. vulnificus caused metabolic changes in the TCA cycle in an infection site-specific manner.
In order to reveal the metabolic change in the early infection phase, we investigated the bacterial cell counts between the different infection time points in the host skeletal muscle and measured metabolites in the host skeletal muscle after 3 h of infection ( Fig. 2C to E). The bacterial cell counts in infected skeletal muscle were dramatically increased after 9 h Concentrations of the TCA cycle metabolites in whole blood were extracted from CE-TOFMS data at 9 h after infection (3 to 5/group). (C) The number of bacteria in skeletal muscle tissues was estimated using the CFU assay. The bacteria were collected in skeletal muscles after 1, 3, and 9 h of infection by V. vulnificus (4/group). (D) Metabolites of the arginine-to-creatinine pathway in skeletal muscles were extracted from CE-TOFMS data at 3 h after infection (4/group). (E) Metabolites of the TCA cycle in skeletal muscles were extracted from CE-TOFMS data at 3 h after infection (4/group). Each bar indicates the mean concentration. *, P , 0.05, and **, P , 0.01, versus the control group (by t test). One-way ANOVA with Tukey's test was used to determine the differences in each group. P values were determined by one-way ANOVA followed by Dunnett's test. † , FDR , 0.05, versus the Vv group.
Metabolic Changing in V. vulnificus-Infected Tissue mSystems of infection. Moreover, even at 3 h of infection, our data indicated the presence of changes in muscle catabolism and TCA cycle metabolites in V. vulnificus-infected skeletal muscle. These data imply that the muscle catabolism and TCA cycle metabolites are associated with bacterial growth. Association between V. vulnificus secreted toxins and metabolic changes in mouse skeletal muscle. Previous studies revealed that VVH and MARTX induced apoptosis and mitochondrial damage (20,21). To estimate the effect of the toxins on the metabolic changes of the TCA cycle, we used mutant strains with single and double deletions of the VVH and MARTX toxins (the DvvhA and DrtxA1 mutants and the double knockout [DKO] mutant) in the V. vulnificus wound infection model. First, we confirmed the inflammation of the skeletal muscle in the presence of V. vulnificus toxins ( Fig. 3A and B). The rtxA1 deletion mutant and the DKO strain exhibited a decrease in bacterial survival and proliferation in skeletal muscle and a decrease in inflammation in skeletal muscle ( Fig. S3A and B). Taken together, these data confirm that MARTX contributes to inflammation in the skeletal muscle of V. vulnificus infection. The rtxA1 mutant strain and double deletion strain (DrtxA1 and DKO) showed a decrease in metabolic changes in the creatine to creatinine pathway in skeletal muscle. As shown in Fig. 3B, the vvhA deletion mutant exhibited metabolic changes in the TCA cycle metabolites in host skeletal muscle that were similar to those observed under wild-type conditions. However, these metabolic changes were abolished in the rtxA1 Metabolic Changing in V. vulnificus-Infected Tissue mSystems deletion mutant. These data indicate that MARTX is associated with the modulation of the host TCA cycle and skeletal muscle catabolism. MARTX does not cause TCA cycle metabolite changes directly. According to the data presented in Fig. S3C, the wild-type and vvhA deletion mutant strains yielded a gradual increase in bacterial cell counts in mouse skeletal muscle, whereas the rtxA1 deletion mutant strain did not (Fig. S3C). Similarly, the mRNA levels of the bacterial 16S rRNA were decreased in rtxA1 deletion mutants in infected skeletal muscle (Fig. S3D). These data indicate that the rtxA1 deletion mutant strain was removed from the infected site by the host immune system. V. vulnificus may cause metabolic changes in the TCA cycle via two possible mechanisms: changes to the host metabolic system directly by MARTX or the contribution by MARTX to the metabolic changes via the promotion of infection by evading the host immune system. We evaluated these hypotheses using neutropenic mice treated with cyclophosphamide monohydrate (CPA). We examined the number of white blood cells (Fig. S3E) and living bacterial cells ( Fig. 4A; Fig. S3F and G) in each group. The number of white blood cells was decreased in the CPA-treated mice, and the bacterial cell count of the DrtxA1 mutant in CPA-treated mice was not decreased at 6 h of infection. Similarly, the mRNA levels of 16S rRNA were elevated in the DrtxA11CPA group in infected skeletal muscle. Moreover, living bacteria were detected in mouse blood in the Vv, Vv1CPA, and DrtxA11CPA groups. These data indicate that the DrtxA1 mutant was affected by treatment with CPA in skeletal muscle. Next, we assessed inflammation in CPA-treated mice using hematoxylinand-eosin (H&E) staining (Fig. S3H) and Western blotting (Fig. S3I). Histopathological changes and myeloperoxidase protein levels indicated the migration of neutrophils in normal mice, whereas CPA-treated mice did not exhibit migration of neutrophils. In addition, the reduction in the expression levels of the MIP-2 mRNA in infection with the DrtxA1 strain was recovered by CPA treatment in skeletal muscle (Fig. 4B). These results suggest that MARTX contributes to evasion of the clearance of the bacteria by the host immune system and that CPA treatment enables the infection of the host skeletal muscle, even for the rtxA1 mutant.
Under these conditions, we checked the metabolic changes in the creatine-to-creatinine pathway (Fig. S3J) and the TCA cycle (Fig. 4C). The levels of metabolites in skeletal muscle infected with wild-type (WT) and DrtxA1 strains were reduced significantly in the CPA-treated mice. Moreover, the V. vulnificus-mediated TCA metabolic changes were abolished in mice infected with the DrtxA1 strain but were recovered in CPA-treated mice. These data indicate that MARTX does not change the TCA metabolism directly. However, MARTX was necessary for the infiltration, survival, and replication of bacterial cells in skeletal muscle, and bacterial infection caused changes in metabolites in the TCA cycle and catabolism.
Metabolic changes in mouse skeletal muscle during Vibrio infection. Finally, we confirmed the specificity of the host metabolic changes during V. vulnificus infection. Mice were infected with other Vibrio spp. (i.e., Vibrio parahaemolyticus, Vibrio mimicus, and Vibrio cholerae) for 9 h, and the resulting metabolic changes were analyzed ( Fig. 5B and C;  Fig. S4C). With the analysis of the metabolites, we confirmed the number of living bacteria and the inflammation of skeletal muscles in those infected with Vibrio spp. based on histopathological changes and the mRNA expression levels of MIP-2, TNF-a, and IL-6 ( Fig. 5A;  Fig. S4A and B). Our data indicated that V. parahaemolyticus and V. mimicus exhibited higher bacterial loads and induced greater inflammatory responses. In contrast, V. cholerae did not show an increase in live bacteria and caused an inflammatory response in infected skeletal muscle. These data implied that these bacteria, known to cause wound infection as observed in V. vulnificus infection, induced metabolic changes in the TCA cycle with the inflammation in skeletal muscle.
Next, the mouse skeletal muscle was isolated, and its metabolites were analyzed by CE-TOFMS. The metabolites associated with muscle catabolites, the creatine-to-creatinine pathway, were changed in V. parahaemolyticus-and V. mimicus-infected muscle, similar to V. vulnificus infection. Despite the similar catabolic pattern, citric acid, isocitric acid, and succinic acid were not changed in V. parahaemolyticus and V. mimicus infection. These results indicated that the increase in TCA cycle metabolites was specific to V. vulnificus-infected skeletal muscle.

DISCUSSION
Metabolome analyses are instrumental to elucidate host responses to bacterial infection (23,28). In a previous study, a mouse skeletal muscle infection model was established and revealed a characteristic inflammation after V. vulnificus infection (25,29). It is known that skeletal muscle has an exacerbated catabolism under conditions of inflammation (30); however, the metabolic changes in the host skeletal muscle in V. vulnificus-infected mice have not been determined. Therefore, we attempted to clarify the metabolic changes in V. vulnificus-infected muscle tissues using CE-TOFMS. Our experimental model also indicated the migration of neutrophils and upregulation of inflammatory cytokines during V. vulnificus infection (Fig. S1A and B). Our metabolome analysis revealed characteristic changes in the TCA cycle and the arginine-to-creatine metabolic pathway after V. vulnificus infection (Fig. 1E). In addition, we found that a V. vulnificus toxin, multifunctional-autoprocessing repeat-in-toxin (MARTX), affected the changes in metabolism simultaneously with the increase in the bacterial counts in skeletal muscle (Fig. 3A and B; Fig. S3C). Importantly, despite the similar creatine and creatinine metabolite levels, the citric acid, isocitric acid, and succinic acid metabolites were not changed in other wound infection models, such as those of V. parahaemolyticus or V. mimicus infection ( Fig. 5B and C). These results indicated that metabolic changes in the energy metabolic pathway occur in V. vulnificus-infected skeletal muscle.
In general, creatine or creatine-phosphokinase is used as a marker of body muscle injury in clinical blood tests. A previous study reported that the serum creatine-phosphokinase level was increased during V. vulnificus infection (26). In agreement with that study, our data indicated an increase in creatine metabolites in the blood of V. vulnificus-infected mice (Fig. S1E). Conversely, creatine and creatinine metabolites were clearly decreased in V. vulnificus-infected skeletal muscle (Fig. 1A). In turn, skeletal muscles infected with V. parahaemolyticus or V. mimicus showed a reduction of creatine and creatinine metabolites, similar to that observed for V. vulnificus infection (Fig. 5B). Moreover, the bacterial counts of V. vulnificus, V. parahaemolyticus, and V. mimicus were increased in infected skeletal muscles (Fig. 5A). Taken together, our data suggest that wound infection with Vibrio spp. causes the degradation of muscle tissue simultaneously with the propagation of Vibrio cells in host tissues and that catabolic metabolites seem to leak into the blood from skeletal muscle. In addition, previous studies reported that the blood creatine phosphokinase level was increased in necrotizing fasciitis by group A beta-hemolytic streptococci (31,32). Therefore, the measurement of metabolites associated with muscle catabolism is expected to be useful for estimating the seriousness of wound infections, including those caused by Vibrio spp., in blood clinical biochemistry tests.
To elucidate the wound infection caused by V. vulnificus, it is necessary to identify the specific metabolic changes triggered by infection with this bacterium. Pathway analysis revealed characteristic changes in metabolites in the TCA cycle (Fig. 1E). In particular, V. vulnificus caused an increase in succinic acid content in muscle tissue, and the metabolic pattern of succinic acid was completely opposite to that of fumaric acid in the TCA cycle ( Fig. 2A). In addition, in contrast with that observed for catabolic metabolites, the levels of the citric acid, isocitric acid, and succinic acid metabolites were increased in V. vulnificus-infected skeletal muscle exclusively (Fig. 5C). It is possible that V. vulnificus affects the host energy metabolic pathway to utilize host succinate for survival in host muscle tissues. Recent studies reported that host succinate was used for increasing the virulence of a variety of bacteria. It is known that host succinate activates Salmonella enterica serovar Typhimurium virulence factors. Furthermore, gut microbiota-produced succinate is utilized by a variety of bacteria to enhance their pathogenesis (33,34). Periodontal ligament cells (PDLSCs) cause succinate accumulation and promote the inflammation triggered by Porphyromonas gingivalis infection (35). Thus, we thought that the catabolism of mouse skeletal muscle triggered by toxins could accelerate the supply of succinate to V. vulnificus, thus enhancing bacterial growth, survival, and virulence activation in host tissues. Moreover, we observed that the expression of the host succinate dehydrogenase mRNA in mitochondria was decreased by V. vulnificus infection (data not shown). These data supported the contention that V. vulnificus promotes succinate accumulation via the downregulation of host succinate dehydrogenase. Taken together, these findings suggest that the inhibition of the succinate-fumarate metabolic pathway provides succinate more effectively to V. vulnificus, with consequent development of dramatic inflammation in infected skeletal muscles. VVH and MARTX have been correlated with cytotoxicity during host infection (36)(37)(38)(39). Therefore, we predicted that VVH and MARTX contribute to the changes in creatine and creatinine metabolites in infected skeletal muscles. Our data indicate that the MARTX deletion mutant alone abolished the changes in the metabolites of muscle catabolism ( Fig. S1C; Fig. 3A). Interestingly, TCA cycle metabolites were also abolished in the MARTX deletion mutant simultaneously with the catabolic changes (Fig. 3B). Conversely, V. parahaemolyticus and V. mimicus also secrete pore-forming toxins (such as thermostable direct hemolysin [TDH]), which induce permeability, cytotoxicity, and apoptosis in host cells (40)(41)(42)(43). According to our data, V. parahaemolyticus and V. mimicus caused creatine and creatinine metabolic changes that were similar to those afforded by V. vulnificus (Fig. 5B); however, these pathogens did not affect the TCA cycle (Fig. 5C). These results suggest that the accumulation of energy metabolites in skeletal muscle is associated with MARTX of V. vulnificus, independent of catabolism. A previous study reported that MARTX induces mitochondrion-mediated apoptosis in vitro (44). MARTX may cause energy metabolic changes directly, similarly to catabolites. Remarkably, the bacterial counts of the rtxA1 mutant strain in skeletal muscle gradually decreased over time in our infection model ( Fig. S3C and D). A previous study reported that MARTX impaired the ability of host phagocytes to clear V. vulnificus infection (45). We considered that MARTX had a role in evasion of the host clearance system. Therefore, to investigate the exact contribution of bacterial proliferation in the metabolic change of infected skeletal muscle, we used a neutropenic mouse model of infection with a MARTX deletion mutant. During infection with the rtxA1 mutant strain in the neutropenic mouse model, the creatine and creatinine metabolites and the expression levels of inflammatory cytokines were similar to those observed for infection with the WT strain, and the bacterial cell counts of the rtxA1 strain were recovered in the neutropenic mice ( Fig. 4A and B; Fig. S3E to J). Based on these results, we concluded that MARTX contributes to host metabolic changes by suppressing the host immune system and promoting proliferation in infected skeletal muscles.
In summary, we identified characteristic metabolic changes in protein catabolism and the TCA cycle triggered by V. vulnificus infection. MARTX contributed to metabolic changes by suppressing the host immune system and promoting bacterial cell infiltration into the host skeletal muscle during wound infection. Based on our results, we propose that the metabolic changes in infected skeletal muscles were correlated with dramatic inflammatory response triggered by V. vulnificus infection. However, the virulence factors that directly affect host metabolism have not been clarified; therefore, additional research is needed to identify the virulence factors associated with these metabolic changes. Future studies focusing on metabolic changes can be used to assist in the design of new therapeutic strategies and elucidate the mechanism of V. vulnificus wound infection and the resulting characteristic inflammation. The bacterial strains used in this experiment are listed in Table 1. V. vulnificus strains were cultured in 2% NaCl-LB broth (1% tryptone, 0.5% yeast extract), V. parahaemolyticus was cultured in 3% NaCl-LB broth, and V. mimicus and V. cholerae were cultured in 1% NaCl-LB broth overnight at 37°C with shaking (170 rpm). Bacterial culture media were centrifuged at 12,000 rpm for 2 min and concentrated, and organisms were grown on thiosulfate-citrate-bile salts-sucrose (TCBS) agar overnight. A single colony was picked from the TCBS agar and cultured in LB broth for 8 h. The bacterial suspension was mixed with a glycerol solution (final concentration of glycerol, 15%) and stored at 280°C. For experiments, bacterial strains were picked from the stock and were cultured in 1%, 2%, or 3% LB broth overnight. Subsequently, the culture broth was recultured in new LB broth for 2 h and then concentrated to an optical density at 600 nm (OD 600 ) of 1.0 with PBS. Animals and infection. We purchased 5-week-old male Crl:CD1 mice (Charles River Laboratories Japan, Kanagawa, Japan). The mice were housed in a controlled environment with a 12-h/12-h light/ dark cycle and maintained at 23°C. One week later, the mice were utilized in the animal experiments.

MATERIALS AND METHODS
Previously, Yamazaki et al. established a V. vulnificus mouse wound infection model (25). Mice were subcutaneously inoculated with 100 mL of a 10Â-diluted bacterial solution adjusted to an OD 600 of 1.0 into the right caudal thighs and infected (10 6 CFU/head) at 23°C after anesthesia using isoflurane. The mice were sacrificed 9 h after the injections. The mouse muscles were dissected and placed in a 1.5-mL tube. Mouse blood samples were collected in 1.5-mL tubes and EDTA dipotassium (EDTA-2K) at 1 mg/ mL was added, to prevent blood coagulation.
Neutropenic mice, 6 weeks old, were prepared via a 2-dose intraperitoneal administration of cyclophosphamide monohydrate at 150 mg/kg of body weight at day 24 and 100 mg/kg at day 21 from infection. Neutropenic mice were infected with bacteria using the method used in another experiment. The mice were sacrificed 6 h after the injections. The mouse muscles were dissected and placed in 1.5-mL tubes. Mouse blood samples were collected in 1.5-mL tubes, and EDTA-2K at 1 mg/mL was added to prevent blood coagulation.
Metabolome analysis of skeletal muscles and blood by capillary electrophoresis time-of-flight mass spectrometry. The tissues were frozen rapidly in liquid nitrogen and stored at 280°C. The skeletal muscles were cut on ice, and 50-mg samples were collected in 1.5-mL tubes. The skeletal muscles were mixed with 500 mL of methanol containing 10 mM methionine sulfone and camphor-10-sulfonic acid, as an internal control, in each tube. The muscles were homogenized using a beads cell disrupter (Micro Smash, MS-100R; Tomy) and the lysates were removed into a new tube. The blood (50 mL) was placed in 2-mL tubes and mixed with 500 mL of methanol containing 10 mM methionine sulfone and camphor-10sulfonic acid, as an internal control, in each tube. These samples were mixed with 200 mL of ultrapure water and 500 mL of chloroform. After vortex mixing, the samples were centrifuged at 9,100 Â g for 5 min at 4°C, and the supernatant was filtered using ultracentrifugal 5-kDa-cutoff filters (Millipore) at 9,100 Â g for 4 h at 4°C. The filtrates were concentrated in a vacuum evaporator and dissolved with ultrapure water containing internal controls (3-aminopyrrolidine, N,N-diethyl-2-phenylacetamide, trimesic acid, and 3-hydroxynaphthalene-2,7-disulfonic acid).
The samples were analyzed using Agilent 7100 capillary electrophoresis (CE) system coupled with an Agilent 6230 TOF LC-MS (Agilent Technologies, Palo Alto, CA) by Human Metabolome Technologies, Inc. (HMT, Tsuruoka, Japan). CE-TOFMS analysis of cationic and anionic metabolites was performed as described previously (48)(49)(50)(51). Cationic metabolites were analyzed with a fused silica capillary (50 mm [internal diameter] by 80 cm [total length]) with cationic electrophoresis buffer (H3302-2031; HMT) as the electrolyte. The sample solution was injected at a pressure of 5,000 Pa for 10 s. The applied voltage was set at 30 kV. Anionic metabolites were analyzed with a fused silica capillary (50 mm by 80 cm) with anionic electrophoresis buffer (H3301-2031, HMT) as the electrolyte. The sample solution was injected at a pressure of 5,000 Pa for 10 s. The applied voltage was set at 230 kV. TOF LC-MS was conducted in the positive-ion mode (4,000 V) and the negative-ion mode (4,000 V) for cationic and anionic metabolites, respectively. Exact mass data were scanned at a rate of 1.5 cycles/s over a 50-to-1,000 m/z range.
Analyzed raw data were processed using Agilent MassHunter qualitative analysis (version B.07.00). Each metabolite was identified and quantified based on the reference peak information, including m/z, migration time, and peak area to compare with the standard (Human Metabolome Technologies, Inc., Tsuruoka, Japan; HMT-H3304-3031). Each peak area was normalized based on internal standard levels, and then the resultant relative area values were normalized by sample amounts to obtain a relative level for each metabolite. Methionine sulfone was used for the density correction of cationic metabolites, and camphor-10-sulfonic acid was used for the density correction of anionic metabolites.
Detection of living Vibrio cells in mouse skeletal muscles and blood. The homogenized Mouse skeletal muscles were weighed to 100 mg, added to 900 mL of PBS, homogenized using a homogenizer pestle, and diluted gradually (10 21 to 10 25 ). Mouse blood was collected in 100-mL portions in 1.5-mL Quantitative real-time PCR analysis. Total RNA was isolated from tissue homogenates using the TRIzol reagent (Thermo Fisher Scientific Inc, Massachusetts, USA). cDNA was synthesized using the Primescript RT reagent kit (TaKaRa, Kyoto, Japan). Gene expression levels were measured using quantitative real-time PCR with TB Green premix Ex Taq (TaKaRa, Kyoto, Japan). The mouse-specific primer pairs used in this experiment are listed in Table 2 (52,53).
H&E staining. Muscle tissues were fixed in 4% paraformaldehyde for 24 h, washed in PBS, and embedded in paraffin. Two-micrometer sections were stained with H&E according to standard procedures.
Measurement of white blood cell counts. Mouse whole blood was collected in 10-mL portions in 1.5-mL tubes, and the number of white blood cells was measured using an automatic blood cell counter for animals (Microsemi LC-662; Horiba Medical, Kyoto, Japan).
Statistical analysis. All studies were conducted in triplicate and on 3 separate days. The data are presented as means and standard errors of the means (SEM) for all experiments. P values were calculated using the data presented as the means and SEM for all experiments. Hierarchical clustering by heat map and PCA of metabolite data sets were computed using MassHunter Mass Profiler Professional (Agilent Technologies, Palo Alto, CA). P values were calculated using Student's t test with the threshold for significance set at a P value of ,0.05. For metabolic analysis, comparisons between groups were estimated by Dunnett's test with Benjamini-Hochberg normalization using IBM SPSS Statistics 25 (SPSS Inc., Chicago, IL, USA). A Benjamini-Hochberg false discovery rate (FDR; q value) of ,0.05 was considered statistically significant. Multiple data analysis was performed based on one-way analysis of variance (ANOVA) and Tukey's honestly significant difference test using IBM SPSS Statistics 25 (SPSS Inc., Chicago, IL, USA). A P value of ,0.05 was considered significant.
Data availability. The data from metabolic analysis using CE-MS are publicly available at EMBL-EBI's MetaboLights repository (https://www.ebi.ac.uk/metabolights) with the data set identifier MTBLS6837; these data are summarized in Table S1.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.