T6SS and ExoA of flesh-eating Aeromonas hydrophila in peritonitis and necrotizing fasciitis during mono- and polymicrobial infections

Significance

Natural infections with flesh-eating bacteria from exposure to water can have devastating patient outcomes. Aeromonas hydrophila causes necrotizing fasciitis (NF). A patient was infected with 4 A. hydrophila strains, 3 clonal (NF2, NF3, and NF4) and 1 (NF1) phylogenetically distinct. By traditional clinical methods, this case was classified as a monomicrobial infection. Metagenomics analysis revealed polymicrobial infection. Clonal strains harbored exotoxin A (ExoA), and NF1 strain possessed a unique effector of the type 6 secretion system (T6SS). The role of T6SS and its effector TseC in direct killing of NF2 in vitro and in vivo and in bacterial phagocytosis/intracellular survival was demonstrated, and the evidence provided that T6SS and ExoA influence dynamics and outcome of mixed infection in NF disease.

Abstract

An earlier report described a human case of necrotizing fasciitis (NF) caused by mixed infection with 4 Aeromonas hydrophila strains (NF1–NF4). While the NF2, NF3, and NF4 strains were clonal and possessed exotoxin A (ExoA), the NF1 strain was determined to be phylogenetically distinct, harboring a unique type 6 secretion system (T6SS) effector (TseC). During NF1 and NF2 mixed infection, only NF1 disseminated, while NF2 was rapidly killed by a contact-dependent mechanism and macrophage phagocytosis, as was demonstrated by using in vitro models. To confirm these findings, we developed 2 NF1 mutants (NF1ΔtseC and NF1ΔvasK); vasK encodes an essential T6SS structural component. NF1 VasK and TseC were proven to be involved in contact-dependent killing of NF2 in vitro, as well as in its elimination at the intramuscular injection site in vivo during mixed infection, with overall reduced mouse mortality. ExoA was shown to have an important role in NF by both NF1-exoA (with cis exoA) and NF2 during monomicrobial infection. However, the contribution of ExoA was more important for NF2 than NF1 in the murine peritonitis model. The NF2∆exoA mutant did not significantly alter animal mortality or NF1 dissemination during mixed infection in the NF model, suggesting that the ExoA activity was significant at the injection site. Immunization of mice to ExoA protected animals from NF2 monomicrobial challenge, but not from polymicrobial infection because of NF2 clearance. This study clarified the roles of T6SS and ExoA in pathogenesis caused by A. hydrophila NF strains in both mouse peritonitis and NF models in monomicrobial and polymicrobial infections.

Necrotizing fasciitis (NF) is a rapidly progressing, life-threatening infection involving necrotic inflammation of the fascial planes and surrounding soft tissues (1–3). Clinical features of NF include hypotension, fever, necrosis, and gangrene (2). Even with rapid surgical intervention (4, 5), case fatalities can be as high as 50% and leave surviving patients with lifelong disabilities and disfigurement (4–7). Group A Streptococcus pyogenes is the most common causative agent of NF, among other pathogens that include species of Staphylococcus, Clostridium, Klebsiella, Vibrio, and Aeromonas (8, 9). NF is classified into 2 types, polymicrobial (type I), usually seen in the elderly or in those with underlying illnesses, while the monomicrobial (type II) NF occurs in immunocompetent individuals (10, 11). Aeromonas hydrophila is the major species among aeromonads involved in NF cases with infections occurring via contaminated water (2, 12–15).

Two previous studies described complex and dynamic interactions among multiple strains of A. hydrophila isolated from an immunocompetent individual who developed NF (1, 16). This particular NF case would traditionally have been considered monomicrobial as only a single strain of A. hydrophila was presumably involved. However, metagenomic analysis showed infection with 4 strains, designated as NF1–NF4, which represented 2 paraphyletic lineages of A. hydrophila, with strains NF2 to NF4 identified as clonal and distinct from NF1. A significant difference was that the strains NF2, NF3, and NF4 harbored a gene, exoA, which encodes a homolog of Pseudomonas aeruginosa exotoxin A (ExoA), whereas strain NF1 proved to be exoA negative, and hence does not produce ExoA toxin. ExoA has ADP ribosylating activity and targets eukaryotic elongation factor-2 (16, 17).

Our previous data demonstrated that the exoA-negative strain NF1 was more virulent than the exoA-positive strain NF2 in a mouse peritonitis model (16). Moreover, a higher mortality rate was observed in mice infected with exoA-positive strains NF2 or NF1-exoA (chromosomally integrated with a single copy of exoA from NF2) than in mice infected with exoA-negative strains NF1 or NF2ΔexoA in a mouse NF model during monomicrobial infection (1, 16). At some infection doses, exoA-negative NF strains (e.g., NF1) were generally localized at the infection site and eliminated by the host defense mechanism, while exoA-positive NF strains (e.g., NF2) disseminated to peripheral organs (1, 16). However, when mice were i.m. (intramuscular) injected with a mixed culture (1:1) of NF1 and NF2, only NF1 systemically disseminated, while NF2 localized at the injection site and was eventually eliminated. Considering the importance of ExoA in both NF1 and NF2 monomicrobial infections, dissemination of NF1 in mixed infection was considered enabled by ExoA destruction of the local tissue barrier (1).

Genetic analysis revealed that both NF1 and NF2 strains possessed functional type 3 and 6 secretion systems (T3SS and T6SS) (16). However, the NF1 genome harbored a unique T6SS effector TseC and its cognate immunity protein TsiC (18), both absent in NF2–NF4 (16). TseC was first characterized as a T6SS effector in A. hydrophila SSU (reclassified as Aeromonas dhakensis) with a putative colicin domain that exhibited bactericidal activity for Escherichia coli and Aeromonas strains lacking the immunity protein TsiC (18). We hypothesized that NF1 was responsible for contact-dependent killing of NF2 in vitro, presumably by deploying T6SS (1). T6SS initially was concluded to target eukaryotic cells, but now more likely its action is in the elimination of bacterial competitors (19). An antibacterial trait provides the pathogen with a competitive advantage against other bacteria in both the host or the environment in the context of mixed infections, enabling proliferation and successful infection (19).

In the present study, 2 NF1 T6SS mutants were generated and the roles of T6SS and ExoA in A. hydrophila pathogenesis elucidated using both mouse peritonitis and NF models. Direct evidence was obtained showing that T6SS and ExoA significantly influenced the dynamics and outcome of mixed infection with nontoxigenic NF1 and toxigenic NF2 in NF disease.

Results

T6SS, but Not ExoA, Affects Phagocytosis and Intracellular Survival of A. hydrophila NF Strains in Macrophages.

To evaluate the role of NF1 T6SS, tseC (encoding effector TseC), and vasK (coding for an essential component for functionality of T6SS) were in-frame deleted, generating NF1∆tseC and NF1∆vasK mutants. Previously generated ExoA-positive and -negative NF strain pairs (NF1/NF1-exoA and NF2/NF2ΔexoA) were used to study the role of ExoA (1) (SI Appendix, Table S1).

Murine macrophages (RAW 264.7) were infected with the NF1 and NF2 strains described above. A significantly higher number of NF1 T6SS mutants were engulfed by macrophages compared to the parental NF1 strain (Fig. 1A). Importantly, a difference in the extent of phagocytosis was also observed between NF1∆tseC and NF1∆vasK. Similarly, a significantly higher rate of phagocytosis was observed for NF2 compared to NF1. In contrast, ExoA did not show a significant impact on bacterial phagocytosis, since the same number of bacteria, ExoA positive and negative in corresponding pairs (NF1/NF1-exoA and NF2/NF2ΔexoA) was phagocytized by macrophages (Fig. 1B).

Fig. 1.

Role of T6SS and ExoA in bacterial phagocytosis by macrophages. Murine macrophages (RAW 264.7) were infected at a MOI of 5 with parental or various T6SS mutants of A. hydrophila NF1 strain (A) or with various ExoA-positive/negative NF1 and NF2 strains of A. hydrophila (B). The percentages of engulfed bacteria were determined by plate counting. Results were expressed as the arithmetic mean ± SD. Data were analyzed by using either the unpaired t test (NF1 vs. NF1ΔtseC) or by 1-way ANOVA with Tukey post hoc test. Asterisks indicate statistically significant differences as compared to parental NF1 strain in A or indicated by a line comparing designated groups (A and B). *P < 0.05, **P < 0.01, and ***P < 0.001.

Deletion of vasK significantly reduced NF1 survival in macrophages compared to the parental strain or ΔtseC mutant (Fig. 2A). A slightly lower survival rate was also observed for the ΔtseC mutant compared to parental NF1 (Fig. 2A). In contrast, both NF1 and NF2 had a similar survival rate in macrophages (Fig. 2B). ExoA did not contribute to bacterial survival inside macrophages, since both ExoA-positive (NF1-exoA and NF2) and -negative NF strains (NF1 and NF2∆exoA) exhibited similar survival rates in macrophages as parental NF1 and NF2 (Fig. 2B).

Fig. 2.

Role of T6SS and ExoA in bacterial intracellular survival in macrophages. Murine macrophages (RAW 264.7) were infected at a MOI of 5 with parental or various T6SS mutants of A. hydrophila NF1 strain (A) or with various ExoA-positive/negative NF1 and NF2 strains of A. hydrophila (B). The intracellular survival of bacteria was evaluated via the gentamicin protection assay after 4 h and 6 h of phagocytosis. The percent survival of intracellular bacteria was calculated and expressed as the arithmetic mean ± SD. Data were analyzed by using 2-way ANOVA with Sidak post hoc test. Asterisks indicate statistically significant differences as compared to designated groups indicated by a line. **P < 0.01 and ****P < 0.0001.

During mixed infection of macrophages with wild-type (WT) NF1 and NF2-lux (1:1), these phagocytic cells still engulfed significantly more NF2-lux than NF1. A similar phenomenon was observed when individual NF strains were used to infect macrophages (Fig. 1B). However, the intracellular survival of NF2-lux was reduced to only 1%, while the survival of NF1 was at 32% after 4 h of phagocytosis, indicting killing of NF2-lux by NF1.

NF1 T6SS Mediates NF2 Killing.

NF1 or its T6SS mutants mixed in a ratio of 5:1 or 1:1 with NF2-lux or NF2-lux expressing tsiC (NF2-lux-pBR322-tsiC) were spotted on Luria Bertani (LB) agar plates and incubated for 4 h at 37 °C. The tsiC encodes the immunity protein for T6SS effector TseC. Colony counts for NF1 and NF2 in the mixture were enumerated before and after incubation for 4 h. An average 9% of NF2-lux survived after incubation for 4 h with NF1, while survival of NF2-lux increased on average of 31% and 63% when mixed with NF1∆tseC and NF1∆vasK, respectively (Fig. 3A). In contrast, an average increase of 30% in bacterial number was observed across all NF1 strains (parental and T6SS mutants) after incubation with NF2-lux, indicating that the interruption of T6SS did not alter fitness of NF1 in the mixed culture.

Fig. 3.

NF1 T6SS-mediated NF2 killing and effect of ExoA on NF1 T6SS. (A) Parental NF1 or its T6SS mutants were mixed with the bioluminescent parental NF2 strain (NF2-lux) or the NF2 strain that expresses the gene encoding immunity protein (TsiC) of NF1 T6SS effector TseC via pBR322 vector (NF2-lux-pBR322-tsiC) in a ratio of 5:1. The mixed cultures were spotted on LB agar plates and incubated at 37 °C for 4 h. The NF2 strains were enumerated in the mixed culture before and after 4 h of the incubation period. The percent survival of NF2 strains after the incubation period was then calculated and the results were plotted as arithmetic means ± SD. Data were analyzed by using either the unpaired t test (NF1ΔtseC/NF2-lux vs. NF1/NF2-lux-pBR322-tsiC) or by 1-way ANOVA with Tukey post hoc test. Asterisks indicate statistically significant differences as compared to the NF1/NF2-lux group or between groups as indicated by a line. (B) Parental NF1 and bioluminescent NF2 exoA mutant (NF2ΔexoA-lux) were mixed in a ratio of 5:1. The mixed cultures were spotted on LB agar plates and incubated at 37 °C for 4 h. Both NF1 and NF2ΔexoA-lux strains were enumerated in the mixed culture at the time of mixing (0 h) and after a 4-h incubation on the agar plates. The percent survival of NF strains after the 4-h incubation period was then calculated, and the results were plotted as arithmetic means ± SD. Data were analyzed by using 2-way ANOVA with Sidak post hoc test. Asterisks indicate statistically significant differences as compared to designated groups indicated by a line. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. A similar result was observed when the cultures were mixed in a ratio of 1:1.

A significant difference in killing by NF1∆tseC and NF1∆vasK of NF2-lux suggested that other T6SS effectors also contribute to NF2 clearance. In comparison to NF2-lux, a higher survival rate (42%) for NF2-lux-pBR322-tsiC was observed when mixed with NF1, indicating TsiC provided protection for NF2 (Fig. 3A).

To evaluate whether NF2 ExoA affected NF1 T6SS-mediated contact killing, NF1 was mixed with NF2∆exoA-lux at a ratio of 5:1 or 1:1. As shown in Fig. 3B, a significant decrease of NF2∆exoA-lux (from 100% to 32%) was observed in mixed culture after a 4-h incubation, while the number of NF1 increased from initial 100% to 137%, indicating ExoA did not influence NF1-mediated killing of NF2.

NF1 T6SS Is Critical in NF and Peritonitis Mouse Models of Infection.

Mice were injected i.m. with either NF1 or its mutant in 2 doses to evaluate the role of T6SS during NF1 monomicrobial infection. At the higher challenge dose, while delayed death was observed for both T6SS mutants, neither T6SS mutant nor NF1-infected groups survived, but at the lower challenge, 88% of T6SS mutant-infected animals survived, whereas all of the NF1-infected mice died (Fig. 4A). A similar pattern was observed when animals were injected i.p. (intraperitoneally), with 100% and 12% survival of T6SS mutants and parental NF1, respectively (Fig. 4B).

Fig. 4.

Role of T6SS in pathogenesis of A. hydrophila NF1 monomicrobial infection. Mice (n = 8) were injected i.m. (A) or i.p. (B) with NF1 strain and its T6SS mutants (∆tseC and ∆vasK) at the indicated doses. Animals were monitored for 2 wk after which percent survival of mice was recorded. Data were analyzed by Log-rank (Mantel–Cox) test. Asterisks indicate statistical significance with P values compared to their corresponding control groups.

To assess the contribution of NF1 T6SS in mixed infection, NF1 or T6SS mutant were mixed with NF2-lux at 1:1 and mice were injected i.m. As shown in Fig. 5A, all mice succumbed to NF1/NF2-lux mixed infection. However, mice infected with either NF1∆tseC/NF2-lux or NF1∆vasK/NF2-lux had 64% and 75% survival rates, respectively.

Fig. 5.

Role of T6SS in the pathogenesis of polymicrobial infection caused by A. hydrophila NF strains in a mouse model of NF. Parental bioluminescent NF2-lux strain was 1:1 mixed with parental or various T6SS mutants of NF1. Mice (10 to 11 per group) were injected i.m. with various mixed cultures at a dose of 5 × 10⁷ CFU (2.5 × 10⁷ CFU per strain). Percent survival of mice was plotted. Data were analyzed by Log-rank (Mantel–Cox) test. Asterisks indicate statistical significance with P values when indicated groups were compared to the control group (NF1/NF2-lux) (A). In a parallel study, mice (7 per group) were injected with the mixed cultures at the same dose. After 24 to 48 h postinfection, animal spleens and muscle at the injection site were collected, and bacterial loads in muscle (B) and spleen (C) enumerated. Results were expressed as arithmetic mean ± SD. Data were analyzed by using the 2-way ANOVA with Sidak post hoc test. Asterisks indicate statistically significant differences among groups indicated by lines. **P < 0.01 and ****P < 0.0001.

In a parallel experiment, spleen and muscle tissue at the injection site were collected 24 to 48 h postinfection (p.i.) from mice injected with the mixed culture and bacterial loads enumerated. In muscle (Fig. 5B), a larger number of NF1 was detected compared to NF2-lux in all cases. The difference was much larger and significant for NF1/NF2-lux, average 1.3 × 10⁹ colony-forming units (CFU) for NF1, while this number decreased to 7.7 × 10⁷ CFU and 4.6 × 10⁷ CFU for the NF1∆tseC/NF2-lux and NF1∆vasK/NF2-lux injected mice, respectively.

Interestingly, NF2-lux was detected at only 124 CFU at the muscle injection site for the NF1/NF2-lux injected group, and 1.4 × 10⁷ CFU and 3.5 × 10⁶ CFU for NF1∆tseC/NF2-lux and NF1∆vasK/NF2-lux injected mice, respectively. These results indicate that NF2-lux survived better when mixed with NF1 T6SS mutants. In spleen (Fig. 5C), only NF1 strains were detected and the average bacterial counts were 1.5 × 10⁶ CFU, 71 CFU, and 29 CFU for NF1/NF2-lux, NF1∆tseC/NF2-lux, and NF1∆vasK/NF2-lux, respectively. Reduced bacterial loads in the spleen correlated with higher animal survival for the NF1 T6SS mutant (Fig. 5A).

ExoA Is Critical for NF2 Peritonitis Mouse Infection.

In earlier studies, ExoA-positive NF2 and NF1-exoA proved more virulent than corresponding ExoA-negative NF2∆exoA and NF1 in monomicrobial infection in the mouse NF model (1). In addition, NF1 was more virulent than NF2 in the mouse i.p. model (16). To elucidate the contribution of ExoA to virulence in mice, ExoA-positive and -negative NF strains were injected at different doses and comparisons focused on ExoA-positive and -negative paired strains of the same genetic background. As shown in SI Appendix, Fig. S1A, the same animal mortality was observed for NF1 and NF1-exoA strains. For the NF2/NF2∆exoA pair (SI Appendix, Fig. S1B), all of the mice in all groups survived the challenge dose of 2.5 × 10⁷ CFU. However, at 8.0 × 10⁷ CFU, 90% of mice injected with NF2∆exoA survived, while survival rate was only 10% when injected with NF2. These results indicate that ExoA plays a more important role in NF2 pathogenesis than NF1 in mouse peritonitis monomicrobial infection (SI Appendix, Fig. S1).

ExoA Is Critical to NF2 in NF.

Previous results showed that when mice were injected i.m. with a mixed culture (1:1) of NF1 and NF2, only NF1 systemically disseminated and NF2 localized at the injection site and eventually was eliminated (1). To elucidate the role of ExoA in NF, mice were injected i.m. to immunize against ExoA. As shown in Fig. 6A, significantly higher anti-ExoA antibody titers were detected in sera of immunized mice compared to animals receiving only alum. The immunized mice were challenged i.m. with NF2 in pure culture or a mixture of NF1 and NF2 (1:1). For the latter, only a marginal difference in mortality (100% and 80%) was observed for both control and immunized mice. However, all immunized animals survived, while all control animals died, in the NF2 single culture challenge (Fig. 6B). These data strongly indicate that ExoA plays an important role in NF2 monomicrobial infection. In NF1/NF2 polymicrobial infection, NF2 was eliminated rapidly at the injection site and only NF1 disseminated, the contribution of ExoA in protection was, thus, dramatically lower (Fig. 6B).

Fig. 6.

ExoA immunization of mice and bacterial challenge. Mice were immunized i.m. with 100 ng of ExoA mixed with alum (1:1). Boosters with the same dose of ExoA were given every 2 wk for a total of 6 immunizations. Mice receiving alum only served as a control. Blood was collected before immunization (preimmune) and after the third and final immunization to evaluate anti-ExoA antibody titers (geometric mean ± SD) (A). Mice (8 per group) were challenged i.m. with either A. hydrophila NF2 alone at a dose of 1 × 10⁸ CFU or with a mixed culture of NF1 and NF2 at a dose of 2.0 × 10⁸ CFU (1 × 10⁸ CFU per strain). Percent survival of mice was recorded (B). Data were analyzed by using the unpaired t test for antibody titers or Log-rank (Mantel–Cox) test for animal survival. Asterisks indicate statistical significance for groups compared to their respective controls. *P < 0.05 and ****P < 0.0001.

The role of ExoA in mixed infection was further studied by injecting mice i.m. with NF1/NF2 or NF1/NF2ΔexoA. Mice of both groups succumbed to infection within 48 h p.i. (Fig. 7A). Neither NF2 nor NF2ΔexoA was detected in the muscle, whereas NF1 counts were 6.6 × 10⁷ to 1.52 × 10⁸ CFU (Fig. 7B). Similarly, NF2 strains did not disseminate to the spleen and the NF1 count in the spleen ranged from 2.2 × 10⁴ to 2.3 × 10⁵ CFU (Fig. 7C).

Fig. 7.

Role of ExoA during mixed infection. In A–C, parental bioluminescent NF1-lux was mixed 1:1 with parental NF2 or its ΔexoA mutant. Mice (7 per group) were infected i.m. with the mixed cultures at a dose of 5 × 10⁷ CFU (2.5 × 10⁷ CFU per strain). Survival of mice was recorded (A). Tissues (spleen and muscle from the injection site) were collected from terminal animals during the course of infection, and bacterial loads in muscle (B) and spleen (C) were enumerated. In D and E, mice (3 per group) were injected i.m. with ExoA-positive/negative (1:1) pairs of A. hydrophila NF1 or NF2 at a dose of 2 × 10⁸ (1 × 10⁸ per strain). After 24 h postinfection, mouse tissues were collected and bacterial loads for each strain at the injection site in muscle (D) and spleen (E) were evaluated. Results were expressed as arithmetic mean ± SD. Data were analyzed by the Log-rank (Mantel–Cox) test for mouse survival, while 2-way ANOVA with Sidak post hoc test was used for analyzing bacterial loads. Asterisks indicate statistical significance for compared groups as indicated by the lines. *P < 0.05, **P < 0.01, and ***P < 0.001.

Mice injected i.m. with mixed cultures of ExoA-positive and -negative strains (NF1-lux/NF1-exoA and NF2/NF2ΔexoA-lux) 24 h p.i. had bacterial loads in both muscle and spleen (Fig. 7 D and E). The NF1-lux/NF1-exoA pair showed similar survival and dissemination, while NF2 (ExoA positive) showed significant advantage over NF2ΔexoA-lux by persisting in muscle and disseminating to the spleen.

Discussion

Polymicrobial infections, specifically those contracted directly from the natural environment, can lead to serious and often fatal outcomes. Interactions among microbes modulate virulence and can influence progression of an infection, posing challenges to treatment (20–22). Unfortunately, little is known about polymicrobial infections (23).

T6SS, a contact-dependent secretion system operates by injecting effectors into other bacteria and eukaryotic cells (24–26). In addition, T6SS is implicated in emergence of multidrug resistant (MDR) superbugs (27–29). Since microbial symbionts require T6SS to persist in the gut (30, 31), it is reasonable to hypothesize that T6SSs are key in modulating the ecology of the gut flora (31).

Genomes of both NF1 and NF2 possess an intact T6SS cluster (16), but only NF1 harbors the effector-immunity pair TseC-TsiC. Contact-dependent killing of NF2 by NF1 was suggested to be mediated by T6SS (1). Deletion of tseC or vasK from NF1 significantly increased survival of NF2 in mixed NF1 and NF2 cultures in vitro. Furthermore, expression of tsiC in trans provided protection to NF2 against killing by NF1 (Fig. 3A). Our observation of NF2 reduction in mixed culture with NF1∆vasK indicated that T6SS effectors still were able to be secreted out or translocated into NF2 from NF1∆vasK at a much reduced level. We previously showed that deletion of the vasK gene from A. dhakensis SSU only prevented secretion but not translocation of Hcp into the host cells (32).

T6SS of NF1 modulates phagocytosis and enhances survival in macrophages (Figs. 1A and 2A), a similar phenomenon observed in strain SSU, with hemolysin-coregulated protein (Hcp) secreted via T6SS mediating the antiphagocytic activity (32, 33). The hallmark of a functional T6SS is the presence of Hcp and VgrG in the culture supernatant (34). Surprisingly, Hcp was detected only in the NF1 pellet and not in the supernatant (16). Whether Hcp participates in the antiphagocytic activity of NF1 and/or interacts with TseC, it requires further study.

We have shown deletion of tseC increased phagocytosis of the ∆tseC mutant as compared to the WT bacterium (Fig. 1), indicating TseC might also possess antieukaryotic activity. However, such an antieukaryotic activity domain has not yet been characterized within TseC. Therefore, it is possible that deletion of tseC might somewhat alter secretion dynamics of other T6SS effectors with antieukaryotic activity from the NF1 strain. Indeed, significant differences between NF1∆tseC and NF1∆vasK were observed with respect to NF2 killing, bacterial phagocytosis, and intracellular survival, suggesting other T6SS effectors may exist and contribute to these biological activities (Figs. 1A, 2A, and 3A).

Attenuation of NF1 T6SS mutants in both mouse peritonitis and NF models in the monomicrobial infection (Fig. 4) was anticipated, since it had been shown that T6SS was associated with pathogenicity of SSU (32, 33).

Considering that NF2 was more virulent than NF1 in the mouse NF model (1, 16), it was reasonable to speculate that mixed infection of NF2 with NF1 T6SS mutants would exhibit a hypervirulent phenotype due to reduced killing of NF2 by NF1. On the contrary, increased animal survival was observed in groups with NF1 T6SS mutants (Fig. 5A). This attenuated phenotype could be explained as being related to the overall reduced virulence of NF1 T6SS mutants (Fig. 4). Alternatively, overall bacterial loads (both NF1 and NF2) at the local injection site (muscle) marginally shifted in infection groups associated with NF1 T6SS mutants at 24 to 48 h after challenged with a dose of 5.0 × 10⁷ CFU (Fig. 5B). In contrast, this number significantly increased to 10⁹ CFU and was dominated by NF1 in the infection group associated with parental NF1 (Fig. 5B).

Interestingly, significant differences between NF1∆tseC and NF1∆vasK were observed with respect to phagocytosis, intracellular survival, and NF2 killing (Figs. 1A, 2A, and 3A), while almost similar animal survival rates were observed for both NF1∆tseC and NF1∆vasK (Figs. 4 and 5A), suggesting that the role of TseC was more prominent in vivo than in vitro.

The role that toxins play in triggering NF remains elusive, although some studies have shown that cytotoxic necrotizing factors produced by E. coli could lead to cell necrosis shown here for NF strains by activation of Rho GTPases (35, 36). Similarly, in a hyperinvasive S. pyogenes strain, increased production of C3-like ADP ribosyltransferase (SpyA) occurs (37), which is functionally related to ExoA of NF2 or P. aeruginosa (38–40). However, while deletion of spyA attenuates S. pyogenes in a s.c. (subcutaneous) mouse soft tissue infection model, the mutant is more virulent when injected i.v. (37, 38). These results suggest that the role of SpyA on bacterial virulence is dependent on infection routes as we also noted for NF1 and NF2 in peritonitis and NF models (16). Furthermore, both SpyA and ExoA of P. aeruginosa trigger pyroptosis in macrophages promoting pathogen clearance and ∆spyA and ∆exoA (∆toxA) mutants exhibit higher phagocytosis and intracellular survival (37).

Neither the deletion of exoA from NF2 nor the addition of exoA to NF1 altered bacterial phagocytosis and intracellular survival in macrophages (Figs. 1B and 2B). Furthermore, in the invasive mouse peritonitis model, addition of exoA to NF1 (NF1-exoA) did not attenuate bacteria, while deletion of exoA from NF2 (NF2∆exoA) significantly reduced bacterial virulence (SI Appendix, Fig. S1).

These discrepancies noted could be attributed to the different cell lines employed. While we used murine macrophages RAW 264.7, bone marrow-derived macrophages (BMDMs), and murine macrophages J774.1 were used to evaluate ∆spyA and ∆toxA mutants. Moreover, ∆spyA showed lower intracellular survival in J774.1, but not in BMDMs (37). However, the in vivo data strongly suggest that ExoA is a key virulence factor for NF2 in both mouse NF and peritonitis models.

As expected, ExoA immunization provided protection only for animals in single NF2, but not in mixed NF1 and NF2 infection (Fig. 6B) due solely to dissemination of NF1 (ExoA negative) and rapid killing of NF2. In addition, when NF1 was mixed with either NF2 or NF2∆exoA, a similar virulence phenotype was observed in the NF mouse model (Fig. 7A). It could be anticipated that NF2 and NF2∆exoA were confined and eliminated at the injection site (muscle), because of the bactericidal activity of NF1 T6SS (Fig. 7B).

Systemic dissemination of NF1 was observed in this study, as well as animals succumbing to mixed infection with NF2∆exoA (Fig. 7 A and C). Most likely ExoA from NF2 destroys the local tissue barrier, facilitating dissemination of NF1 (1). To confirm the role of ExoA in a mixed infection, particularly bacterial dissemination, mice were injected i.m. with a mixture of exoA-positive and -negative pairs with the same genetic background (NF1/NF1-exoA and NF2/NF2∆exoA). While NF1 and NF1-exoA persisted in muscle, as well as distally disseminating, NF2∆exoA showed a significant disadvantage compared to NF2 (Fig. 7 D and E). Thus, the importance of ExoA in NF2 pathogenesis during monomicrobial infection is clarified. The role of ExoA in NF1 and NF2 mixed infection appears to be limited to destruction of the local tissue barrier due to rapid elimination of NF2 from the injection site. With reduced NF2 competition and absence of ExoA, which increases bacterial motility (16), NF1 disseminates readily to cause animal mortality.

In summary, both NF2 and NF2∆exoA assist NF1 in persistence and ability to proliferate at the injection site, eventually leading to systemic dissemination of NF1 and death in a mixed infection (Fig. 7). NF1 T6SS and NF2 ExoA play important roles in the pathogenesis of infection in both mouse NF and peritonitis. Conclusive evidence is provided showing that T6SS mediates interaction between NF1 and NF2, shown both in vitro and in vivo. Our findings are of significant interest showing how interaction among microbes of the same species modulates overall bacterial virulence, thus eventually influencing progression of infection and posing challenges in therapeutic interventions.

Materials and Methods

Bacterial Strains, Plasmids, and Reagents.

The source of A. hydrophila and E. coli strains and plasmids are listed in SI Appendix, Table S1. Bacteria were grown for 16 to 18 h in LB medium at 37 °C with continuous shaking (180 rpm). Suicide vector pDMS197 was used to prepare isogenic mutants (41). The antibiotics tetracycline (Tc), ampicillin (Ap), kanamycin (Km), and rifampin (Rif) were used at concentrations of 15, 250, 100, and 200 μg/mL, respectively.

Construction of Bioluminescent NF2∆exoA Strain and NF1 T6SS In-Frame Deletion Mutants.

Tn7 transposon was used to generate bioluminescent NF2∆exoA strain (NF2∆exoA-lux) (1). Briefly, a spontaneous Rif^R NF2ΔexoA clone was conjugated with 2 E. coli SM10 (λpir) strains carrying either pTNS2 or pUC18-mini-Tn7T::Km-lux plasmid as described elsewhere (1, 16, 42).

The flanking DNA fragments to NF1 vasK and tseC were PCR amplified, respectively, as described in SI Appendix, Fig. S2 by using indicated primer sets in SI Appendix, Table S2. PCR fragments were cloned into pDMS197 suicide vector, resulting in recombinant plasmids pDMSvasK and pDMStseC, respectively (SI Appendix, Table S1).

The recombinant plasmids were electroporated into the parental NF1 strain and double-crossover mutants (NF1ΔtseC and ΔvasK, Tc^S) were verified via PCR by using verification (VR) series of primer sets, followed by genomic sequencing with the confirmation (CON) series of primers (SI Appendix, Table S2).

Construction of NF2-lux Strain Expressing Immunoprotein Encoding Gene tsiC.

tsiC is downstream, overlaps with T6SS effector tseC in NF1, and is cotranscribed. Using the primer pair tsiCF/tsiCR (SI Appendix, Table S2), the tsiC coding region, with its promoter (200 bp upstream of tseC), was PCR amplified from the genome of NF1ΔtseC and cloned into pBR322, yielding recombinant plasmid pBR322-tsiC. The plasmid was transformed into the NF2-lux strain by electroporation to create the NF2-lux-pBR322-tsiC strain (SI Appendix, Table S1).

Phagocytic Assay and Intracellular Survival of Bacteria in Macrophages.

Murine macrophage cell line RAW 264.7 was maintained at 37 °C and 5% CO₂ in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS and 1 mM sodium pyruvate.

RAW 264.7 cells seeded in 24-well plates were infected with overnight grown A. hydrophila strains at multiplicity of infection (MOI) of 5 (8 wells per strain) individually or in mixed culture (i.e., NF1/NF2-lux) and incubated at 37 °C with 5% CO₂ for 1 h. The infected macrophages were washed with PBS and the number of bacteria associated with the cells determined from 4 wells of each strain by serial dilution and plating (1, 16). Infected macrophages in the remaining 4 wells were treated with gentamicin (50 µg/mL) for 1 h to kill extracellular bacteria. After washing, the number of the bacteria inside the macrophages was determined. Percent phagocytosis was calculated from number of bacteria after gentamicin treatment and before gentamicin treatment.

For the intracellular survival assay, macrophages were infected with A. hydrophila as described above, and bacterial loads first determined (4 wells per strain) after 1-h gentamicin treatment (50 µg/mL). Incubation of infected macrophages in the remaining 4 wells continued with fresh DMEM and maintenance dose of gentamicin (2 µg/mL) for an additional 4 to 6 h at 37 °C with 5% CO₂. After washing with PBS, the number of bacteria inside the macrophages was counted. Percent survival was calculated from number of bacteria after 4 to 6 h of incubation and after gentamicin treatment.

In Vitro Bacterial Direct Contact Killing Assay.

Overnight cultures of A. hydrophila strains were diluted (1:20) with fresh LB broth and allowed continued growth for 2 h at 37 °C and 180 rpm shaking. Selected pairs of bacterial strains were mixed in a ratio of 1:1 or 1:5 and spotted on nonselective LB agar plates and incubated for 4 h at 37 °C. The number of each strain before and after incubation for 4 h in mixtures was determined by serial dilution and plating on selective medium with appropriate antibiotic or on nonselective LB plates as previously described (1). Percent bacterial survival was calculated from the number of bacteria after 4 h of incubation and before incubation.

Animal Studies.

Four- to 6-wk-old Swiss-Webster mice (Taconic Biosciences, Rensselaer, NY) were used by following the approved Institutional Animal Care and Use Committee protocol of the University of Texas Medical Branch at Galveston.

For infection studies, 16- to 18-h bacterial cultures were centrifuged, the pellets washed 3× in sterile PBS, and suspended in 1/10 the original culture volume of PBS. Subsequently, each culture was titrated and bacterial inocula prepared to obtain 100-μL volume containing the intended infectious dose.

Mouse model of peritonitis.

Animals (8 to 10 per group) were infected with 100 μL of A. hydrophila NF strains (NF1, NF1 T6SS mutants, NF1-exoA, NF2, and NF2∆exoA-lux) at various doses via i.p. injection. After infection, animals were observed for disease progression over a period of 14 d.

Mouse model of NF.

Mice were anesthetized with isoflurane and 100 μL Aeromonas culture injected i.m. into one leg. Inocula contained either a pure single culture (monomicrobial infection) or mixed culture containing different strains at 1:1 ratio (polymicrobial infection). After infection, animals were observed for disease progression.

Assessment of bacterial loads in muscle and spleen of infected mice.

At 24 to 48 h p.i., spleen and muscle tissue (200 to 300 mg) at the injection site were excised from euthanized mice and homogenized in 1 mL PBS. Tissue homogenates were serially diluted and plated on LB agar amended with appropriate antibiotics. After a 24-h incubation at 37 °C, bacterial colonies were enumerated per organ (spleen) or per gram (muscle tissue).

ExoA immunization and antibody titers.

Mice were immunized i.m. with 100 ng P. aeruginosa ExoA (Sigma-Aldrich) mixed with alum (Thermo Scientific) in 1:1. Boosters, with the same dose of ExoA, were given every 2 wk for a total of 6 immunizations. Mice receiving alum only served as control. Blood was collected by retroorbital bleeding before immunization (preimmune) and after the third and final immunization. ELISA plates were coated with ExoA protein (100 ng/well) and incubated overnight at 4 °C. Total IgG against ExoA in mouse serum (1:5 serially diluted) was determined as previously described (43).

Statistical Analysis.

All experiments were performed in triplicate (minimum). Statistical analyses were performed by using either the unpaired t test or 1-way ANOVA with Tukey post hoc test or 2-way ANOVA with Sidak post hoc test. Animal survival data were subjected to Log-rank (Mantel–Cox) analysis. P value of <0.05 was considered statistically significant.

Materials and Data Availability.

All experimental procedures and data from this study are provided in the main text as well as in SI Appendix. All bacterial strains and plasmids used in the study are available on request by qualified researchers for their own use.