The Translocation and Assembly Module (TAM) of Edwardsiella tarda Is Essential for Stress Resistance and Host Infection

Translocation and assembly module (TAM) is a protein channel known to mediate the secretion of virulence factors during pathogen infection. Edwardsiella tarda is a Gram-negative bacterium that is pathogenic to a wide range of farmed fish and other hosts including humans. In this study, we examined the function of the two components of the TAM, TamA and TamB, of E. tarda (named tamAEt and tamBEt, respectively). TamAEt was found to localize on the surface of E. tarda and be recognizable by TamAEt antibody. Compared to the wild type, the tamA and tamB knockouts, TX01ΔtamA and TX01ΔtamB, respectively, were significantly reduced in motility, flagella formation, invasion into host cells, intracellular replication, dissemination in host tissues, and inducing host mortality. The lost virulence capacities of TX01ΔtamA and TX01ΔtamB were restored by complementation with the tamAEt and tamBEt genes, respectively. Furthermore, TX01ΔtamA and TX01ΔtamB were significantly impaired in the ability to survive under low pH and oxidizing conditions, and were unable to maintain their internal pH balance and cellular structures in acidic environments, which led to increased susceptibility to lysozyme destruction. Taken together, these results indicate that TamAEt and TamBEt are essential for the virulence of E. tarda and required for E. tarda to survive under stress conditions.


INTRODUCTION
Edwardsiella tarda is a Gram-negative bacterium of the family Enterobacteriaceae. It is an important pathogen that causes systemic infection in a wide variety of marine and freshwater fish as well as other hosts, including birds, reptiles, and mammals (Mohanty and Sahoo, 2007;Leung et al., 2012Leung et al., , 2019. In aquaculture, E. tarda-induced edwardsiellosis in many fish species has led to heavy economic losses in Asia, United States, and Europe (Castro et al., 2006;Nelson et al., 2009;Park et al., 2012). In humans, E. tarda has been reported to cause gastroinhastestinal and extraintestinal diseases in immunocompromised people (Abayneh et al., 2013;Shao et al., 2015).
Bacterial virulence relies on membrane biogenesis pathways to assemble the outer membrane proteins (OMP) essential for the process of host-pathogen interactions, such as invasion and adhesion . In Gram-negative bacteria, the process of outer membrane assembly is dependent on the translocation and assembly module (TAM) and the β-barrel assembly machinery (BAM) complex (Selkrig et al., 2012;Gruss et al., 2013;Selkrig et al., 2014;Stubenrauch et al., 2016). TAM comprises two subunits: an integral OMP, TamA, and an inner membrane-anchored protein, TamB (Marani et al., 2006;Selkrig et al., 2012). Deletion of tamA or tamB reduces the virulence of Klebsiella pneumoniae, Proteus mirabilis, Citrobacter rodentium, Salmonella enterica, and Escherichia coli (Struve et al., 2003;Burall et al., 2004;Kelly et al., 2006;Selkrig et al., 2012). A recent study showed that the assembly of FimD, which is important for the deployment of fimbrial extensions from the surface of bacterial pathogens, and other usher proteins is mediated by the TAM complex (Stubenrauch et al., 2016).
In E. tarda, the tamA and tamB genes have been identified by genome sequencing; however, the role of TamA and TamB in E. tarda remains unknown. In the present work, we employed both in vitro and in vivo approaches to examine the function of E. tarda tamA and tamB (named tamA Et and tamB Et , respectively). Our results indicated that TamA Et and TamB Et were essential to the infectivity of E. tarda and to the survival of E. tarda under stress conditions, especially acidic conditions.

Ethics Statement
The experiments involving live animals in this study were approved by the Ethics Committee of Institute of Oceanology, Chinese Academy of Sciences. All methods were carried out in accordance with the relevant guidelines.

Fish
Clinically healthy tongue sole (Cynoglossus semilaevis) were purchased from a commercial fish farm in Shandong Province, China. Fish were maintained at 20 • C in aerated seawater with 144% ± 4% dissolved oxygen (detected with a fast response DO sensor RINKO I; ARO-USB, JFE Advantech Co., Ltd., Japan; Air saturation range: 0-200%). The fish were fed with  (Zhou and Sun, 2015). Fish were euthanized by immersion in seawater containing 10 mg/L of tricaine methanesulfonate (Sigma, St. Louis, MO, United States) before tissue collection.

Bacterial Culture
Bacterial strains used in this study are listed in

Plasmid Construction
To construct pETTamA, which expresses recombinant TamA Et (rTamA Et ), tamA Et was amplified by PCR with primers TamA-F/TamA-R ( Table 2). The PCR products were ligated with the T-A cloning vector T-Simple (TransGen Biotech., Beijing, China), and the recombinant plasmid was restriction digested with SmaI to retrieve the tamA Et -containing fragment, which was inserted into pET259 at the SwaI site, resulting in pETTamA. To construct pETTamB, which expresses the TamB domain of TamB Et (amino acid residues 780-1255) with a 6-histidine His-tag at the C-terminus, PCR was conducted with primers TamB-F/TamB-R ( Table 2), and the PCR products were inserted into pET259 as above. To construct the low copynumber plasmids pJTTamA and pJTTamB that express tamA Et and tamB Et , respectively, tamA Et and tamB Et were amplified by PCR with primers TamA-F3/TamA-R and TamB-F3/TamB-R ( Table 2), respectively; the PCR products were ligated with the TA cloning vector T-Simple, and the recombinant plasmids were digested with SmaI. The fragments containing tamA Et and tamB Et were retrieved and inserted into plasmid pBT3 at the EcoRV site, resulting in pBT3TamA and pBT3TamB, respectively. pBT3TamA and pBT3TamB were digested with SwaI, and the fragments carrying tamA Et and tamB Et were inserted into plasmid pJT at the SwaI site, resulting in pJTTamA and pJTTamB, respectively. All PCR products were verified by sequence analysis.

Construction of tamA Et and tamB Et Knockouts
To construct the mutant E. tarda with tamA Et knockout, i.e., TX01 tamA, in-frame deletion of a 1671 bp segment of tamA Et (residues 22-578) was performed by overlap extension PCR as follows: the first overlap PCR was performed with primers TamA-F1/TamA-R1, the second overlap PCR was performed with primers TamA-F2/TamA-R2, and the fusion PCR was performed with the primer pair TamA-F1/TamA-R2 ( Table 2). The PCR products were ligated into the suicide plasmid pDM4 (Milton et al., 1996) at the BglII site, resulting in pDMTamA. S17-1 λpir was transformed with pDMTamA, and the transformants were mated with E. tarda TX01 via conjugation as previously described (Li et al., 2015). Briefly, the donor strain (resistant to chloramphenicol, sensitive to polymyxin B) and the recipient strain (resistant to polymyxin B, sensitive to chloramphenicol) were cultured in LB medium to OD 600 of 0.8 and mixed at a ratio of 3:1. The mixture was spread onto a LB agar plate without antibiotics, and the plate was incubated at 28 • C for 24 h. After incubation, the bacteria on the plate were resuspended in 2 ml LB, from which 100 µl was taken and plated on a LB agar plate supplemented with polymixin B and chloramphenicol to select for transconjugants. Transconjugants were cured of pDMTamA by incubation on LB agar plates supplemented with 10% sucrose (which induces sacB-mediated plasmid curing), and chloramphenicol-sensitive strains were subsequently analyzed by DNA sequencing to confirm in-frame deletion of tamA Et . This strain was named TX01 tamA. To construct the mutant E. tarda with tamB Et knockout, i.e., TX01 tamB, in-frame deletion of a 987 bp segment of tamB (corresponding to amino acid residues 919-1247) was performed by overlap extension PCR as follows: the first overlap PCR was performed with primers TamB-F1/TamB-R1, the second overlap PCR was performed with primers TamB-F2/TamB-R2, and the fusion PCR was performed with the primer pair TamB-F1/TamB-R2 ( Table 2). S17-1 λpir was transformed with pDMTamB, and the transconjugants were selected as described above. One of the transconjugants was named TX01 tamB. To construct the tamA Et complement strain TX01 tamA/tamA, S17-1 λpir was transformed with pJTTamA, and the transformants were conjugated with TX01 tamA. The transconjugants were selected on LB agar plates supplemented with tetracycline (marker of pJT) and polymyxin B (marker of TX01 and its derivatives). One of the transformants was named TX01 tamA/tamA. To construct the tamB Et complement strain TX01 tamB/tamB, S17-1 λpir was transformed with pJTTamB, and the transformants were conjugated with TX01 tamB. The transconjugants were selected as described above. One of the transconjugants was named TX01 tamB/tamB.

Purification of Recombinant Proteins and Preparation of Antibodies
Escherichia coli BL21 (DE3) was transformed with pETTamA, pETTamB, or pET32a (which expresses the Trx tag). The transformants were cultured in LB medium at 37 • C to mid-log phase, and the expression of rTamA Et , rTamB Et , and rTrx was induced by adding isopropyl-β-D-thiogalactopyranoside to a final concentration of 1 mM. After growth at 16 • C for an additional 16 h, the cells were harvested by centrifugation, and recombinant proteins were purified using nickel-nitrilotriacetic acid columns (GE Healthcare, Piscataway, United States) as recommended by the manufacturer. The proteins were treated with Triton X-114 to remove endotoxin as reported previously (Zhang and Sun, 2015). The proteins were dialyzed for 24 h against phosphate buffered saline (PBS) and concentrated using PEG20000 (Solarbio, Beijing, China). The concentrations of the purified proteins were determined using the NanoPhotometer (Implen GmbH, Munich, Germany). Mouse antibodies against rTamA Et , rTamB Et , and rTrx were prepared as described previously (Li et al., 2016). The antibodies were purified using rProtein G Beads (Solarbio, Beijing, China). The specificity and titer of the antibodies were determined by Western blot and enzyme-linked immunosorbent assay (ELISA) as reported previously (Li et al., 2017).

Fluorescent Microscopy
Detecting of TamA Et on bacterial surface by fluorescence microscopy was performed as reported previously (Li et al., 2016). Briefly, E. tarda TX01 was cultured in LB medium to OD 600 of 0.8 and resuspended in PBS (pH 7) to 10 8 CFU/ml. The bacterial suspension was dropped on a glass slide and incubated for 12 h at 28 • C. The antibody against rTamA Et , rTamB Et , or rTrx was added to bacterial suspension. The cells were incubated at 37 • C for 1 h and then washed three times with PBS (pH 7). Fluorescein isothiocyanate (FITC)-labeled goat anti-mouse IgG (Abcam, Cambridge, United Kingdom) was added to the bacteria, followed by incubation at 37 • C for 1 h in the dark. After staining with 4, 6-diamino-2phenyl indole (DAPI) (Invitrogen, Carlsbad, CA, United States), bacteria were visualized using a confocal microscope (Carl Zeiss, Oberkochen, Germany). To determine bacterial damage under acidic conditions, E. tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were cultured as above and resuspended in PBS of pH 5 to 10 8 CFU/ml. The cells were incubated at 28 • C for 2 h. After incubation, bacterial cells were treated with propidium iodide (PI) (Majorbio Biotech, Shanghai, China) and DAPI for 15 min in the dark according to the manufacturer's instructions. The cells were then subjected to microscopy as above.

Western Blot
Edwardsiella tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were cultured in LB medium to an OD 600 of 0.8. Whole-cell proteins were prepared and subjected to Western blot as reported previously (Zhang M. et al., 2008) with mouse antibody against rTamB Et . RNA polymerase beta was used as an internal reference  and detected with anti-RNA polymerase beta antibody (Abcam, Cambridge, United Kingdom).

Electron Microscopy
To examine the flagella of E. tarda, E. tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were statically cultured in LB medium at 28 • C for 48 h and gently resuspended in PBS. Transmission electron microscope (TEM) examination was performed as previously reported (Givaudan and Lanois, 2000). Briefly, a drop of bacterial suspension was added to carbon-coated copper grids (200 mesh) and rinsed with ultrapure grade water, the bacteria were then negatively stained with 1% (wt/vol) phosphotungstic acid (5 s) and rinsed with ultrapure grade water. The grids were air dried and examined with a TEM (HT7700, Hitachi, Japan). To examine the structure of E. tarda under acidic conditions, E. tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were cultured with shaking in LB medium to OD 600 0.8, and resuspended in PBS buffer of pH 7 or pH 5 to 10 8 CFU/ml. The cells were incubated in PBS buffer of different pH at 28 • C for 2 h. After incubation, the cells were observed with a TEM as above.

Motility Assay
Edwardsiella tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were cultured with shaking in LB medium to OD 600 0.8, and 10 µl cell suspension were spotted onto the center of fresh LB plates containing 0.3 or 0.5% (w/v) agar. The plates were incubated at 28 • C for 2 days, and the motility of the bacteria was assessed by examining the diameter of the bacterial halo on the plate.

Bacterial Survival Under Acidic and Oxidizing Conditions
PBS buffer was adjusted to pH 7, pH 5, or pH 4.5 with hydrochloric acid. E. tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were cultured as above and resuspended in PBS buffers of different pH to 10 5 CFU/ml. The cells were incubated at 28 • C for 2 h. After incubation, the cells were diluted in PBS (pH 7) and plated on LB agar plates. The plates were incubated at 28 • C for 24 h, and the colonies emerged on the plates were counted. The survival rate was calculated as follows: (number of survived cells in different pH/number of survived cells in pH 7) × 100%.
To examine the effect of the acidic condition on the internal pH of E. tarda, above bacteria were suspended in PBS (pH 5) to 10 10 CFU/ml and incubated at 28 • C for 2 h. The cells were then pelleted by centrifugation and resuspended in 1 ml PBS (pH 7). The cells were boiled for 5 min at 100 • C and subjected to sonication in an ice-water bath.  (B) TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were incubated in PBS of pH 5 for 2 h; after incubation, the pH of the cell lysate was determined. (C) TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were incubated in PBS of different pH for 2 h; after incubation, the cells were treated with lysozyme, and bacterial survival was determined. (D) TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were incubated with different concentrations of H 2 O 2 for 1 h, and bacterial survival was determined. Data are the means of three independent assays and presented as means ± SEM. **P < 0.01.

Bacterial Infection of Peripheral Blood Leukocytes (PBL)
Blood was collected from the caudal vein of tongue sole. PBL were isolated from the blood with 61% Percoll and collected as described previously (Li et al., 2017). The cells were cultured in L-15 medium (Thermo Scientific HyClone, Beijing, China) in 96well culture plates (10 5 cells/well). E. tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were prepared as above and added to PBL (10 5 CFU/well). The cells were incubated at 28 • C for 0.5, 1, or 2 h. After incubation, the plates were washed with PBS (pH 7), and the cells were lysed with 100 µl PBS containing 1% Triton X-100. The cell lysate was diluted and plated in triplicate on LB agar plates. The plates were incubated at 28 • C for 48 h, and the colonies that emerged on the plates were counted. The genetic identities of the colonies were verified by PCR with specific primers and sequence analysis of the PCR products. The experiment was performed three times.

Intracellular
Edwardsiella tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were prepared as above and added to tongue sole PBL (10 5 CFU/well). The cells were incubated at 28 • C for 1 h and washed three times with PBS (pH 7). Fresh L-15 medium containing 100 µg/ml gentamicin (Solarbio, Beijing, China) was added to the cells, and the cells were incubated at 28 • C for 1 h to kill extracellular bacteria. The plates were washed three times with PBS (pH 7) and incubated at 28 • C for 0, 1, 2, 4, and 8 h. After incubation, the cells were lysed, and bacterial recovery was determined as above. The experiment was performed three times. tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were incubated at pH 5 for 2 h and stained with DAPI and PI. The cells were subjected to microscopy with red or blue fluorescence light. (B) TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were incubated at pH 5 or pH 7 for 2 h and examined with a transmission electron microscope.

In vivo Infection
In vivo infection was performed as reported previously (Li et al., 2015). Briefly, E. tarda TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, and TX01 tamB/tamB were cultured as above. The cells were washed with PBS (pH 7) and resuspended in PBS (pH 7) to 5 × 10 6 CFU/ml. Tongue sole (average 15.7 g) were randomly divided into five groups (15 fish/group) and infected via intramuscular injection with 100 µl TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA or TX01 tamB/tamB. At 12, 24, and 48 h post-infection, kidney, spleen, and blood were collected from the fish (five at each time point). The tissues were homogenized in PBS (pH 7). The homogenates was serially diluted and plated in triplicate on LB agar plates. The plates were incubated at 28 • C for 48 h, and the colonies that appeared on the plates were enumerated. The genetic identity of the colonies was verified as above. For mortality analysis, five groups (20 fish/group) of tongue sole were infected as above with TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, or TX01 tamB/tamB, and the fish were monitored daily for mortality for 15 days.

Statistical Analysis
All experiments were performed three times. Statistical analyses were carried out with SPSS 17.0 software (SPSS Inc., Chicago, IL, United States). Data were analyzed with analysis of variance (ANOVA), and statistical significance was defined as P < 0.05.

Characterization of the Sequences of TamA Et and TamB Et
A search of the genome of E. tarda revealed the presence of tamA and tamB homologs (named tamA Et and tamB Et , respectively). TamA Et is composed of 578 amino acid residues, with a predicted molecular mass of 64.27 kDa and a predicted pI of 9.1. TamB Et is composed of 1255 amino acid residues, with a predicted molecular mass of 134.99 kDa and a predicted pI of 9.3. TamA Et possesses an outer membrane translocation and assembly module A domain; TamB Et possesses an outer membrane translocation and assembly module B domain (Supplementary Figure S1). TamA Et has 67.1-97.8% overall sequence identities with other FIGURE 5 | Cellular infectivity of Edwardsiella tarda variants. (A) Tongue sole peripheral blood leukocytes (PBL) were infected with TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, or TX01 tamB/tamB for 0.5, 1, and 2 h, attached bacterial number was determined. (B) PBLs were infected as above for 1 h; the extracellular bacteria were killed with antibiotic, and the intracellular bacterial number was determined at different time points. Data are the means of three independent assays and presented as means ± SEM. **P < 0.01. bacterial OMP assembly factors, but only 12.8% sequence identity with E. tarda BamA. TamB Et has 59.8-96.4% overall sequence identities with other bacterial TamB (Supplementary Figure S2).

Localization of TamA Et Onto Bacterial Surface
Immunofluorescence microscopy showed that when E. tarda was treated with anti-rTamA Et antibody, the antibody was detected on the cells, whereas no cell-associated antibody was detected in E. tarda treated with anti-rTamB Et antibody or anti-rTrx antibody (Figure 1), suggesting that TamA Et is surface-exposed in E. tarda.

Growth and Motility
Two isogenetic mutants of E. tarda TX01, i.e., TX01 tamA and TX01 tamB, were constructed, which bear markerless deletions FIGURE 6 | The tissue dissemination capacities of Edwardsiella tarda variants. Tongue sole were infected with TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, or TX01 tamB/tamB for different hours, and bacterial recoveries from spleen (A), kidney (B), and blood (C) were determined. Data are the means of three independent assays and presented as means ± SEM. **P < 0.01.
of tamA Et and tamB Et , respectively. The deletion of tamB in TX01 tamB was verified by Western blot, which detected no TamB production in TX01 tamB (Supplementary Figure S3). Tongue sole were infected with TX01, TX01 tamA, TX01 tamB, TX01 tamA/tamA, or TX01 tamB/tamB. The mortality and survival of the fish were monitored daily for 15 days. The experiment was performed three times, and the mean survival rates are shown. For the significance test, TX01 was used as a control for its variants. **P < 0.01.
Growth analysis showed that when cultured in LB medium, TX01 tamA and TX01 tamB, displayed growth profiles similar to that of the parental strain TX01 (data not shown), which is consistent with previous observation in other bacterial species (Struve et al., 2003;Burall et al., 2004;Kelly et al., 2006;Selkrig et al., 2012). The swimming and swarming abilities of TX01 tamA and TX01 tamB were severely impaired in comparison to that of the wild type (Figure 2A and Table 3). In contrast, the swimming and swarming abilities of TX01 tamA/tamA and TX01 tamB/tamB, which have the tamA Et and tamB Et genes, respectively, introduced back to the mutant bacteria, were largely similar to that of the wild type (Figure 2A and Table 3). Electron microscopy showed that TX01, TX01 tamA/tamA, and TX01 tamB/tamB, but not TX01 tamA or TX01 tamB, possessed polar flagella ( Figure 2B).

Survival Under Acidic and Oxidizing Conditions
When incubated in PBS buffer of pH 7, the survival rates of TX01 tamA and TX01 tamB were comparable to that of the wild type TX01. However, at pH 5 and pH 4.5, the survival rates of TX01 tamA and TX01 tamB were significantly lower than that of TX01 ( Figure 3A). In contrast, the survival rates of TX01 tamA/tamA and TX01 tamB/tamB were largely similar to that of the wild type ( Figure 3A). Furthermore, following incubation in the pH 5 buffer, the cell lysates of TX01 tamA and TX01 tamB exhibited pH values of 6.25 and 6.27, respectively, which were significantly lower than that of the cell lysates of the wild type (pH 7.2), TX01 tamA/tamA (pH 7.1) and TX01 tamB/tamB (pH 7.05) ( Figure 3B). Pre-incubation in the pH 6 or pH 5 buffer, but not in the pH 7 buffer, significantly reduced the survival rates of TX01 tamA and TX01 tamB against lysozyme treatment ( Figure 3C). In contrast, preincubation in low pH buffers had no significant effects on the survival of TX01 tamA/tamA or TX01 tamB/tamB against lysozyme treatment (Figure 3C). Similarly, the survival rates of TX01 tamA and TX01 tamB, but not TX01 tamA/tamA and TX01 tamB/tamB, in 1 mM, 2 mM, and 3 mM H 2 O 2 were significantly reduced compared to that of the wild type ( Figure 3D).

Membrane Integrity of TX01 tamA and TX01 tamB Under Acidic Condition
Fluorescence microscopy showed that when incubated in PBS buffer of pH 5, TX01 tamA and TX01 tamB, but not TX01, TX01 tamA/tamA, or TX01 tamB/tamB, were markedly labeled by PI (Figure 4A), which can only penetrate into dead or damaged cells. TEM showed that compared to pH 7, pH 5 induced severe damages to the cellular structures of TX01 tamA and TX01 tamB, with the bacteria cells exhibiting swelling and formation of bubble-like protrusion structures ( Figure 4B). In contrast, no apparent change in the cellular structures of the wild type or the complemented strains was observed ( Figure 4B).
In vitro Infectivity of TX01 tamA and TX01 tamB In vitro study showed that when tongue sole PBLs were infected with E. tarda mutants or wild type for 0.5, 1, and 2 h, the numbers of TX01 tamA and TX01 tamB recovered from the cells were significantly lower than that of the wild type TX01, whereas the numbers of recovered TX01 tamA/tamA and TX01 tamB/tamB were comparable to that of the wild type ( Figure 5A). When the extracellular bacteria were removed by killing, the intracellular TX01, TX01 tamA/tamA, and TX01 tamB/tamB were found to continue to replicate and increase in number, whereas the intracellular TX01 tamA and TX01 tamB exhibited no detectable increase in number ( Figure 5B).
In vivo Infectivity and Lethality of TX01 tamA and TX01 tamB In vivo study showed that when inoculated into tongue sole, TX01 tamA and TX01 tamB exhibited dramatically reduced bacterial disseminations in kidney, spleen, and blood in comparison to the wild type TX01, whereas the tissue dissemination capacities of TX01 tamA/tamA and TX01 tamB/tamB were similar to that of the wild type ( Figure 6). Consistently, fish mortalities induced by TX01 tamA and TX01 tamB were significantly lower than that induced by TX01, TX01 tamA/tamA, or TX01 tamB/tamB (Figure 7).

DISCUSSION
In Gram-negative bacteria, the process of outer membrane assembly is dependent on the Omp85-family protein β-barrel assembly machinery (Hagan et al., 2011), and some membrane structures require a distinct subgroup of the Omp85 family protein, TamA (Selkrig et al., 2012;Gruss et al., 2013;Stubenrauch et al., 2016). TamA and BamA have similar domain structures, and both are Omp85-family proteins that function in parallel pathways for OMP assembly in bacteria . It has been postulated that in the evolution of some bacteria, duplicate copies of BamA form the origin of TamA . In our study, we found that TamA Et and TamB Et possess conserved domains of TamA and TamB, respectively. TamA Et has high levels of sequence identities with other bacterial OMP assembly factors but a low identity with the BamA of E. tarda, suggesting that the BAM and TAM complexes may have evolved to have different functions in E. tarda. Gram-negative bacteria are separated from the external environment by the inner and outer membranes, which are set apart by the periplasm and the peptidoglycan layer (Albenne and Ieva, 2017). In Proteobacteria, TAM is composed of two membrane proteins, TamA and TamB. TamA is integrated in the outer membrane, while TamB has a signal-anchor sequence embedded in the inner membrane, and the interaction of these two subunits depends on TamB penetrating through the peptidoglycan layer (Selkrig et al., 2012;Shen et al., 2014). Consistently, we found that TamA Et , but not TamB Et , was exposed on the surface of E. tarda, suggesting that, like the TAM complex of other bacterial species (Selkrig et al., 2012), TamA Et probably associates with the inner membrane protein TamB Et and functions as a complex with the latter.
In uropathogenic E. coli, TAM catalyzes the assembly of the OMP FimD, which is essential for the ordered assembly of type 1 fimbriae (Munera et al., 2007;Stubenrauch et al., 2016). In Proteus mirabilis, transposon mutants of TamA and TamB show no defect in swarming activity (Burall et al., 2004). In our study, TX01 tamA and TX01 tamB exhibited no apparent flagella, implying a requirement of TamA Et and TamB Et in the biosynthesis/transport of E. tarda flagella. Consistently, TX01 tamA and TX01 tamB displayed very little swimming and swarming capacities. These observations indicated that tamA Et and tamB Et mutations affected flagella formation, resulting in impaired motility.
pH is known to regulate the expression of membrane proteins, such as porins, that are required for bacterial survival under stress conditions associated with low pH, oxidation, osmotic pressure, and high temperature (Begic and Worobec, 2006;Klimentova et al., 2019). Low pH has been shown to induce conformational changes in the extracellular loop reign of OmpG and influence the transmembrane pore formation (Korkmaz-Ozkan et al., 2010). At pH 5, OmpF increased fluoroquinolone antibiotic permeability and accumulation, resulting in increased antibiotic sensitivity (Cama et al., 2015). In Mycobacterium bovis, the transcription of the ompA gene was increased at pH 5.5 (Singh and Verma, 2008). In many pathogenic Gramnegative bacteria, the type III secretion system, which consists of the inner and OMP s and a needle, is also regulated by low pH (Markham et al., 2008). In our study, the survivals of TX01 tamA and TX01 tamB were significantly decreased under acidic and oxidative conditions, and the internal pH of these mutants were significantly affected by environmental pH, implying that the mutants were unable to maintain pH homeostasis inside the bacteria. Furthermore, low pH rendered TX01 tamA and TX01 tamB susceptible to lysozyme damage. Consistently, low pH severely damaged the cellular structures of TX01 tamA and TX01 tamB, which likely accounted for the observed lysozyme sensitivity under acidic conditions. These results suggested that TamA Et and TamB Et likely play an important role in the regulation of the proteins that form the normal membrane structure and determine the membrane integrity and permeability of E. tarda. It is also likely that TamA Et and TamB Et, by forming a complex, are directly involved in the formation of membrane integrity by constituting a membrane channel with selective permeability. E. tarda is an intracellular pathogen and can survive in host macrophages (Okuda et al., 2006;Liu et al., 2017). As a strategy of intracellular survival, E. tarda is able to detoxify the reactive oxygen species (ROS) generated by host phagocytes through the production of catalase and superoxide dismutases (Srinivasa Rao et al., 2003;Cheng et al., 2010). In our study, the survival rates of the TX01 tamA and TX01 tamB mutants after treatment with H 2 O 2 , which produces ROS, were significantly lower than that of the wild type, suggesting that the defective cellular structure caused by TamA Et and TamB Et mutation facilitated the H 2 O 2 -derived ROS to penetrate into the bacterial cells and kill the cells more effectively. This observation further supports the above conclusion that TamA Et and TamB Et are vital to the structural integrity of E. tarda.
The outer membrane of Gram-negative bacteria protects cells from external aggressions and mediates the secretion of virulence factors, and intact OMPs can promote host adhesion (Henderson and Nataro, 2001;Ranava et al., 2018). Efficient and correct assembly of integral OMPs requires the TAM complex Ranava et al., 2018). In Citrobacter rodentium, Salmonella enterica and E. coli, TAM mutation eliminated the virulence of the bacteria (Selkrig et al., 2012(Selkrig et al., , 2014Stubenrauch et al., 2016). In our study, we found that TX01 tamA and TX01 tamB exhibited significantly decreased ability to invade into and replicate in fish cells, and were significantly attenuated in the ability of tissue dissemination and inducing mortality in the host. These results indicated that TamA Et and TamB Et were required for optimal bacterial virulence. It is possible that mutation of tamA Et and tamB Et affects not only the membrane structure and permeability of the bacteria, but also the secretion of some virulence factors, which together lead to decreased survival in the stressing environment of the host.

CONCLUSION
Our study demonstrated that TamA Et and TamB Et are required for motility, flagella development, stress survival, and infectivity of E. tarda. TamA Et and TamB Et exert a significant impact on membrane integrity, which affects both the physiology and the pathogenicity of the bacteria. These results add new insights into the function of bacterial TAM and the survival mechanism of E. tarda.

DATA AVAILABILITY STATEMENT
The datasets generated for this study can be found in the WP_012847252.1 of GenBank, WP_012847253.1 of GenBank.