A quorum sensing-based in vivo expression system and its application in multivalent bacterial vaccine

Background Delivery of antigens by live bacterial carriers can elicit effective humoral and cellular responses and may be an attractive strategy for live bacterial vaccine production through introduction of a vector that expresses an exogenous protective antigen. To overcome the instability and metabolic burden associated with plasmid introduction, alternative strategies, such as the use of in vivo-inducible promoters, have been proposed. However, screening an ideal in vivo-activated promoter with high efficiency and low leak expression in a particular strain poses great challenges to many researchers. Results In this work, we constructed an in vivo antigen-expressing vector suitable for Edwardsiella tarda, an enteric Gram-negative invasive intracellular pathogen of both animals and humans. By combining quorum sensing genes from Vibrio fischeri with iron uptake regulons, a synthetic binary regulation system (ironQS) for E. tarda was designed. In vitro expression assay demonstrated that the ironQS system is only initiated in the absence of Fe2+ in the medium when the cell density reaches its threshold. The ironQS system was further confirmed in vivo to present an in vivo-triggered and cell density-dependent expression pattern in larvae and adult zebrafish. A recombinant E. tarda vector vaccine candidate WED(ironQS-G) was established by introducing gapA34, which encodes the protective antigen glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from the fish pathogen Aeromonas hydrophila LSA34 into ironQS system, and the immune protection afforded by this vaccine was assessed in turbot (Scophtalmus maximus). Most of the vaccinated fish survived under the challenge with A. hydrophila LSA34 (RPS = 67.0%) or E. tarda EIB202 (RPS = 72.3%). Conclusions Quorum sensing system has been extensively used in various gene structures in synthetic biology as a well-functioning and population-dependent gene circuit. In this work, the in vivo expression system, ironQS, maintained the high expression efficiency of the quorum sensing circuit and achieved excellent expression regulation of the Fur box. The ironQS system has great potential in applications requiring in vivo protein expression, such as vector vaccines. Considering its high compatibility, ironQS system could function as a universal expression platform for a variety of bacterial hosts.


Background
Vaccination constitutes the most cost-effective tool for prophylaxis of infectious diseases, and the use of bacterial carriers is probably one of the most successful strategies ever developed to deliver vaccine antigens [1]. The efficacy of a live bacterial vector vaccine rests in its ability to present sufficient antigens to the host immune system and initiate the desired protective immune response [2]. Selecting an appropriate expression strategy to optimize the production of the recombinant antigen is one of the most important issues relevant to the use of bacterial strains as vaccine carriers [3]. To date, certain strategies have been developed to achieve stable heterologous gene expression in vaccine vectors. Application of low copy number vectors and modification by introducing a balanced lethal system can improve the stability and safety of vaccines [4][5][6]. However, expression levels may be insufficient to stimulate the desired responses [7]. High copy number vectors achieve sufficient antigen expression but can cause over-attenuation of the carrier and lack of immunogenicity [8]. To circumvent these problems, optimal and controllable expression, such as with the use of in vivo-activated expression systems, is a crucial step in multivalent vaccine design.
In vivo environmental stimuli are represented by aggregates of unique signals, including anaerobic conditions [9], oxidizing agent availability [10], and low iron concentrations [11]. Several in vivo-inducible promoters that respond to these signals have been previously investigated [12]. For example, iron is often bound to metalchelating proteins in vivo, making it extremely limited in hosts [13]. Thus, lack of free iron is an important in vivo environmental signal [14]. To survive from the iron limited conditions of their hosts, bacteria have evolved several iron uptake, storage, and metabolism systems to obtain adequate iron in such an environment [15]. The promoter from iron-uptake regulon is strongly repressed in iron-rich conditions by Fur, a typical ferric uptake regulator protein, but fully de-repressed in the absence of iron [16]. A 19-bp inverted repeat consensus sequence, Fur box, which is known as the binding site of the Fur protein complexing with ferrous irons, is the control core of iron-responsive promoters in bacteria [17]. When iron is abundant, Fur protein forms a complex with ferrous iron and blocks the entry of RNA polymerase by binding to the Fur box in the relevant promoter area [18]. In our previous work, several ironregulated promoters and Fur boxes were applied to build in vivo-inducible regulation circuits for antigen expression, bacterial ghost preparation, and toxic protein synthesis [19][20][21].
Quorum sensing (QS) is a system that bacteria release and respond to membrane-penetrating molecules called autoinducers to regulate their gene expressions according to cell density [22]. This system allows bacteria to monitor a variety of processes, including competence, bioluminescence, virulence factor secretion, biofilm formation, and sporulation [23]. The first quorum sensing circuit was discovered in the bioluminescent marine bacterium Vibrio fischeri [24]. Two proteins, LuxI and LuxR, are essential for the quorum-sensing control of bioluminescence in V. fischeri. LuxI is the autoinducer synthase that produces an AHL, N-(3-oxohexanoyl)homoserine lactone [25]. LuxR is the cytoplasmic receptor of the autoinducer and the transcriptional activator of the luciferase luxICDABE operon [26,27]. At low cell density, the luxI gene is transcribed at a low basal level. As the culture grows and autoinducers accumulate to a specific threshold, cytoplasmic LuxR proteins combine with autoinducers and bind to the luxICDABE promoters [28]. Given that the expression of luxI is also activated by the autoinducer-bound LuxR, this autoinduction positive feedback loop is presumed to enforce synchrony as the cell population switches from low cell density mode to high cell density quorum sensing mode [29]. Quorum sensing systems have been widely used in various gene circuits in synthetic biology, serving as oscillator [30], amplifier [31] and so on [32][33][34], because of their high expression efficacy and cell density regulated characteristic. However, although a number of quorum sensing systems participate in various synthetic architectures; very few have been designed for bacterial vector vaccine.
In this work, several in vivo expression systems, including ironQS1-4, were developed for potential application in multivalent bacterial vaccine based on quorum sensing system of V. fischeri. To activate the quorum sensing system in vivo, iron starvation of the in vivo environment was utilized to adopt different strategies, including iron-regulated promoter substitution and iron-regulated regulator introduction. The screened ironQS system was demonstrated to be only initiated in iron-limited medium during in vitro expression assay when the cell density reached a threshold. The ironQS system was also well regulated by the iron signals and cell density in the subsequent in vivo expression assay. Furthermore, a protective antigen, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from the important fish pathogen Aeromonas hydrophila, was introduced to the ironQS system. The immune protection of the resultant recombinant E. tarda vector vaccine candidates was evaluated in turbot.

Results and discussion
Construction of quorum sensing-based in vivo expression systems The quorum sensing components including luxI, luxR and their respective promoters were cloned from V. fischeri. A reporter gene named katushka, which encoded a far-red fluorescent protein, was used to measure the expression efficiency of our systems. First, a basal quorum sensing regulated expression circuit was constructed through sequentially inserting a RBS component and a reporter gene katushka behind the intact QS regulon. A stable plasmid pUTat [35] containing this circuit was named pQS ( Figure 1A), and four quorum sensing-based in vivo expression systems ironQS1-4 were designed on this basis ( Figure 1B). The ironQS1 was a pQS derivative in which the original P luxR promoter was substituted for a strict low-iron-triggered promoter P viuA [35], so that ironQS1 would not express Katushka protein in ironrich culture medium. IronQS2 was constructed by inserting a standard Fur box sequence into the −10 region of luxI promoter, and ironQS3 and ironQS4 were the derivatives of ironQS2 with additional Fur boxes inserted striving for higher strictness. In culture medium, bacterial Fur proteins complexing with ferrous irons would bind to the Fur box sites to prevent transcriptional process of RNA polymerases, while in in vivo low-iron situation, ironQS2-4 systems would be de-repressed and started to produce interest proteins because of the dissociation of Fur proteins.
To evaluate the regulation and expression performances of four candidate systems, E. tarda recombinants harboring ironQS1-4 plasmids were cultured in both ironlimited and iron-rich media. Samples were adjusted to OD 600 (optical density at 600 nm) = 1 to measure the red fluorescence emitted by Katushka. As seen in Figure 1C, in iron-limited environment, all four candidates showed strong fluorescence intensities with similar RFVs (relative fluorescence values) to that of the control system pQS. This was evidence that insertion of Fur box exerted no effects on expression efficiency. However, in iron-rich environment, their performances were divergent. IronQS2 presented the lowest leak expression level in iron-rich environment, while ironQS1 system showed the reduced expression strictness by the substitution of promoter P luxR by P viuA . On the other hand, the additional Fur box in ironQS3 and ironQS4 did not play the desired effects to enhance their strictness. After comprehensive comparison of the four candidate expression systems, ironQS2, abbreviated to ironQS, was finally chosen as our desired expression system with the combined advantages of high induced expression level, high stringency and relatively simple genetic structure.

In vitro expression of ironQS system
To verify the iron-regulated and cell density-regulated performances of ironQS system, ironQS plasmid was transformed into E. tarda for the in vitro assay by adding FeSO 4 or iron-chelator DP (2,2-dipyridyl) into the medium during cultivation. As showed in Figure 2A, there was a low expression level of ironQS system in the presence of Fe 2+ . This indicated that Fur protein complexing with ferrous iron bond to the Fur box region and the QS system was remarkably repressed. On the contrary, ironQS was considerably activated in the absence of iron when Fe 2+ in the medium was plundered by DP. tarda strains containing ironQS1-4 after cultured in iron-rich and iron-limited media respectively. E. tarda loaded with pQS and blank strain were set as controls. * means a significant deviation.
All these data confirmed that Fur box sequence could negatively control QS system by signaling to Fe 2+ in the medium. Furthermore, to identify whether the cell density-regulated feature was still preserved in ironQS system, a fed-batch cultivation was conducted to control the bacterial cell density always under the threshold by continuously feeding fresh iron-free medium, and a control experiment was carried out by normally cultivating the bacteria to a high cell density. As shown in Figure 2B, RFV of ironQS strain in fed-batch model maintained at a low level, showing that the ironQS was always kept off even in iron-free medium when the cell density was below its threshold. Meanwhile, the RFV of control group rose continuously as the cell density exceeded threshold. These results suggested that ironQS system exhibited a precise iron and density dual-regulated expression feature by Fur box and quorum sensing circuit, respectively.

In vivo expression of ironQS system
In vivo condition is a typical iron-limited environment and a crucial site where bacterial vaccine plays its role. Although ironQS has been proved to be an efficient binary system responding to Fe 2+ and cell density in vitro, the in vivo behaviors are particularly concerned. Since E. tarda is known as a facultative intracellular bacterium and able to persist and replicate within phagosomes of macrophages, it is reasonable to apply macrophages as a cell model for in vivo performance evaluation of ironQS. Suspensions of recombinant E. tarda WED containing ironQS were incubated in the macrophagelike cell J774A.1, and the fluorescence intensity was determined at regular time intervals with a fluorescence microscope. As shown in Figure 3A, cells infected by the bacterium showed only very low fluorescence in the first 2 h post-infection. During the whole period, the bacterial number maintained at 4.83 ± 1.46 per cell. Although WED was not likely to replicate in the macrophages, fluorescence intensity still became stronger gradually with the accumulation of autoinducers, indicating that ironQS system could sense the iron-limited signal in macrophage and unlock quorum sensing to express target protein. Meanwhile, the positive feedback of quorum sensing increased the expression of Katushka protein and enhanced the fluorescence intensity [36].
On the other hand, zebrafish larvae and adults were further applied as animal models to evaluate in vivo performance of ironQS. The larvae were infected by immersion with recombinant E. tarda WED strain, and the fluorescence in zebrafish larvae was determined at regular time intervals. As shown in Figure 3B, no obvious Katushka fluorescence signal was detected until 2 h post immersion and the fluorescence signals were mainly found in the gastrointestinal tract of fish. These results indicated that the ironQS system could function well in zebrafish larvae. Responding to the ironfree signal in larvae, ironQS was de-repressed after Fur protein dissociating from Fur box, and simultaneously, it was activated by locally accumulated autoinducers from an increasing bacterial population in vivo. For zebrafish adults, ironQS-loaded E. tarda strains were administrated by intraperitoneal injection, and then the in vivo expression pattern of ironQS was evaluated by Western blot. As showed in Figure 3C, ironQS was activated significantly after 12 h post injection upon in vivo propagation of the bacteria. This result indicated that attenuated E. tarda WED could colonize and persist in fish adult, in which the low iron concentration would trigger the expression of heterologous protein by ironQS. Figure 2 Iron-regulated and cell density-regulated performances of ironQS system. (A) Expression curves of ironQS system regulated by Fe 2+ concentration. FeSO 4 or iron-chelator DP was added into the medium during cultivation. (B) Expression curves of ironQS system regulated by cell density. E. tarda harboring ironQS system was cultivated via fed batch and the OD 600 was maintained under 0.3 by feeding fresh iron-limited media continuously. Meanwhile a control group was carried on by cultivating the bacteria normally to a high cell density. * means a significant deviation.
All the data confirmed that ironQS system could be effectively activated in vivo and the bacterial densitydependent expression feature, illustrated in vitro, was finely preserved. IronQS system maintained its dualregulated characteristics and made a good performance in in vivo environments such as in macrophage and zebrafish.

Stress on bacterial growth and colonization
Constitutive expression systems are often believed to produce significant metabolic burden on the host bacterium and thus may have negative effects on immune protection of bacterial vector vaccine. Therefore in vivo inducible expression system is highly recommended for bacterial vector vaccine design [3,8]. In this work, an in vivo inducible and cell density-dependent expression system ironQS was established. Compared with the original QS expression system, ironQS system exerted few effects on microbial in vitro growth and in vivo colonization. As showed in Figure 4A, when bacterial hosts containing pQS, ironQS or blank plasmid pUTat were inoculated into fresh medium, WED(ironQS) showed a normal growth, which is similar to that of WED(pUTat), while the growth of WED(pQS) was obviously inhibited, especially during the early stage after inoculation. When three strains were injected into adult zebrafish, the colonization abilities were measured by counting viable bacterial numbers in zebrafish organs. As Figure 4B showed, a dramatic drop of viable bacterial numbers for WED(pQS) was found within zebrafish, while in vivo WED(ironQS) propagated significantly, similarly to the control strain WED(pUTat). These results suggested that the expression system ironQS, controlled by cell density and iron-limitation signal, brought about less growth stress on bacterial host and greatly contributed to the in vivo colonization and propagation of WED(ironQS), which would be a valuable attribute for a bacterial vector vaccine to evoke effective immune response in vivo.

Expression of ironQS in broad bacterial hosts
Adaptation to different bacterial hosts is an important aspect of a good heterologous expression system. IronQS system is composed of quorum sensing genes from V. fischeri and an artificial Fur box, and it is of paramount importance to work well in different bacterial hosts. Here, ironQS was transformed into gram-negative bacteria including genetic engineering strain Escherichia coli, fish-originated Vibrio anguillarum, and humanoriginated Salmonella typhimurium, as well as grampositive bacterium Staphylococcus aureus, respectively. In in vitro assays, ironQS in all the four strains presented no expression in normal iron-rich medium, and high expression in DP supplemented medium ( Figure 5). According to the specificity of the quorum sensing system, LuxR receptor could only sense the autoinducer molecules produced by its homologous LuxI. Thus, quorum sensing circuit from V. fischeri had no interference with these bacterial hosts. Fur proteins in bacterial hosts could bind to or dissociate from the standard Fur box depending on iron concentration. The fine adaptability of ironQS system suggested that it might be potential to be used in a wide range of bacterial vaccine hosts.

Application of ironQS in multivalent bacterial vaccine
GAPDH of A. hydrophila has been shown to be an effective protective antigen [37]. Edwardsiella tarda is an important facultative intracellular pathogen of both animals and humans, and its attenuated strain WED, constructed by in-frame deleting eseBCD and aroC, is an excellent bacterial vector for use in recombinant vaccine design [38]. The ironQS vector expressing heterologous antigen GAPDH was transformed into live attenuated bacterial vaccine E. tarda WED, and a multivalent bacterial vaccine candidate WED(ironQS-G) was constructed.
To investigate the potential application of WED (ironQS-G) as a multivalent vaccine, protection efficacy was evaluated in turbot, an important mariculture species in China. Vaccinated turbot were maintained for 30 days and then challenged by intraperitoneal injection with wild type A. hydrophila LSA34 or E. tarda EIB202. As shown in Table 1, WED(ironQS-G) showed significant multivalent protection against E. tarda EIB202 (RPS = 72.3%) and A. hydrophila LSA34 (RPS = 67.0%) over saline control. Meanwhile, the attenuated vector strain WED triggered significant protection against E. tarda EIB202 (RPS = 76.7%), but only slight protection against A. hydrophila LSA34 (RPS = 13.9%). This confirmed that in vivo expression of GAPDH controlled by ironQS system conferred effective protection against experimental challenge with A. hydrophila LSA34, and also showed no adverse effect on the protection of E. tarda wildtype.

Conclusions
Quorum sensing system has been widely used in various gene structures in synthetic biology as a well-functioned and population-dependent gene circuit. Although these delicate genetic architectures fully embody the density dependent features and high expression efficiency, few of such systems have been designed and implemented in practical applications [39]. In this work, an efficient and finely-regulated in vivo expression system, ironQS, was designed by integrating Fur box with quorum sensing circuit of V. fischeri. Binding of the Fur-iron complex functioned as a lock that prevents gene transcription of quorum sensing circuit and then shuts down the entire expression system. However, given the extraordinary lack of free iron in vivo, Fur protein dissociated from Fur box and activated quorum sensing circuit. Experiments on in vitro simulation and zebrafish model confirmed that the ironQS system exhibits an in vivo-triggered and cell density-dependent expression pattern.
For practical application, we established a bacterial vector vaccine using the ironQS system as an antigen presenter. Several characteristics dominate the efficiency of a live bacterial vector vaccine: stability, infectivity and  ability to present sufficient antigen [3]. How to improve antigen delivery efficiency without reducing vaccine stability and vitality has long challenged researchers. The use of plasmids in vector vaccines is often associated with instability of the recombinant phenotype and declining ability of colonization. In the ironQS system, several strategies were adopted to render it more suitable for use in a vector vaccine. First, as the main switch of the entire system, Fur box prevented unnecessary expression during in vitro cultivation, thereby alleviating the metabolic burden during cultivation and enhancing the ability of colonization of the bacterial host at early infection stages [40]. Second, thorough de-repression of the quorum sensing system completely restored the high expression efficiency of the system to guarantee sufficient antigen production in the host. Benefited from the population-dependent expression pattern of quorum sensing circuit, undesirable antigen protein expression was controlled to a minimum during the first stage of colonization because of the low cell density, until the bacterial population reached its threshold and then evoked multiple protective immune responses in fish host. Taken these results together, the ironQS system proposed in this work has great potential in designs of a stable and efficient bacterial vector vaccine.
The iron-dependent gene regulation mechanism has been well studied in different bacteria, and certain ironregulated promoters have been used to build in vivo inducible regulation circuits for controlled antigen expression [19,35]. Prior to designing a vector vaccine, a desirable iron-regulated promoter must be screened for the corresponding host because promoters borrowed from other host strains often function unsatisfactorily. This problem could be readily circumvented by introduction of the ironQS system. Although the various host strains used in this work possess their own QS systems and produce their own autoinducers, the LuxR receptors only identify their own autoinducers because these signal molecules vary in their acyl side chains [22,23]. Thus, no crosstalk or interference was observed between the native host QS and the heterologous ironQS because of the specificity of quorum sensing system. As well, the 19-bp consensus Fur box sequence is highly conserved [16]. Hence, the ironQS system displays high compatibility with a variety of bacterial hosts, and this plasmid forms a selfcontained and closed-loop expression circuit that is less dependent on its hosts. In fact, ironQS can function as a universal platform that is suitable for multiple hosts, rather than as a specific vector for E. tarda, in this work. This broad suitability suggests that, theoretically, any antigen protein can be loaded into the ironQS system and this system may be introduced into any bacterial host to prepare the desired vaccines simply and directly.
In summary, we established an in vivo programmed expression circuit that is de-repressed by iron-free signals in vivo and then programmed to produce heterologous proteins in a cell density-dependent manner in vivo. This artificial synthetic circuit might have potential use for in vivo applications including, but not limited to, bacterial vector vaccines.

Plasmid construction
General DNA operations were conducted following standard protocols. Automated DNA sequencing and primer The mortality was recorded for 4 weeks after challenge, and the observation of surviving fish was extended to 6 weeks. b RPS is the relative percent survival, which is defined relative to the saline group. RPS (%) = (1 -mortality of vaccinated fish / mortality of control fish) × 100.
c If RPS > 60%, it represents that the vaccine candidate showed protection against challenge, marked as "+". If RPS < 30%, it represents no protection against challenge strain, marked as "-". The "±" is standard deviation.
synthesis were completed by Life Technologies (Shanghai, China). Seamless cloning operations were carried out using ClonExpress II One Step Cloning Kit (Vazyme Co. Ltd., China) following standard procedures. The plasmids used or constructed in this work were listed in Table 1. First, a basal quorum sensing regulated expression plasmid pQS was constructed on the basis of pUTat expression vector by sequentially inserting a RBS sequence and a reporter gene katushka behind the intact QS regulon. The DNA fragments of pUTat, katushka and QS gene were respectively amplified with their primers that overlapping with each other (primers pUT-F/R for linear pUTat, katushka-F/R for gene katushka, QS-F/R for quorum sensing circuit). By seamless cloning operation, the fragments were spliced to form the intact pQS plasmid. On the basis of pQS, four quorum sensing-based in vivo expression plasmids ironQS1-4 were designed by replacing promoters or inserting Fur boxes into basal plasmid pQS. IronQS1 was constructed by replacing the original luxR promoter with an iron-regulated promoter P viuA . P viuA was amplified from the genome of Vibrio anguillarum MVM425 using primers P viuA -F/R and the rest was amplified from pQS using ironQS1-F/R. These two fractions were then jointed through seamless cloning operation. IronQS2 was constructed by inserting a standard Fur box behind luxI promoter to block its transcription in ironrich condition. Using primers ironQS2-F/R, pQS plasmid was amplified, linearized and then self-ligated via seamless cloning to form ironQS2. To enhance this modification, another Fur box sequence was subsequently added behind luxR promoter to form ironQS3 using primers ironQS3-F/R. Similarly, in ironQS4, two continuous Fur box sequences were inserted behind luxI promoter for the purpose of improving iron regulation. IronQS4 was constructed by amplifying and linearizing pQS using primers ironQS4-F/R and make it self-ligated via seamless cloning operation. For construction of vaccine vector ironQS-G, the gene gapA34 encoding glyceraldehyde-3-phosphate dehydrogenase from Aeromonas hydrophila LSA34 was amplified using primers gapA-F/R, and ironQS was linearized to remove the reporter gene using primers ironQSG-F/R. These two fragments were assembled together by seamless cloning operation. Primers used in this work were listed in Table 3.

Expression assay of ironQS strains in vitro
Overnight bacterial cultures were inoculated (1:1000, v/v) into fresh LB medium with antibiotics and FeSO 4 . For fed-batch cultivation, fresh LB medium with antibiotics (and DP or FeSO 4 if needed) was continuously added into culture vessel to keep OD 600 under 0.3 all the time, below the cell density threshold to activate the established pQS system in this work. At different time intervals, each culture sample was centrifuged at 8000 × g for 5 min, and the

Expression assay of ironQS strains in zebrafish model
Overnight bacterial cultures of E. tarda(ironQS) were inoculated (1:1000, v/v) into fresh LB medium with antibiotics and FeSO 4 . The bacteria were incubated for 9 h, harvested, and suspended in PBS. Zebrafish larvae at an age of the 6th to 8th day were immersed in cell suspension at a concentration of 10 8 viable bacteria per ml for 2 h and then were washed and transferred into freshwater. At target time point, larvae were washed with PBS at least three times, anesthetized with 0.01% (w/v) anesthetic MS-222 and observed with an inverted fluorescence microscope. For zebrafish adults, bacterial cultures were treated as above. Healthy adult zebrafish with an average weight of 0.2 g were intraperitoneally injected with E. tarda (ironQS) suspension at the dose of 10 5 CFU (colony forming unit) per tail. At 4 h, 12 h and 24 h post injection, the internal organs of 20 fish from each group were surgically extracted. After homogenized and concentrated with ultrafiltration, the tissue samples were further analyzed by Western blot using the antibody specific to Katushka.

Vaccination and challenge
The recombinant plasmid ironQS-G was transformed into attenuated E. tarda vaccine strain WED to get WED(ironQS-G). This potential multivalent vaccine candidate WED(ironQS-G) was assayed for the immune protection in turbot. Attenuated E. tarda strain WED was used as the control. All the vaccination and challenge experiments were done in triplicate. For vaccination, turbot (Scophtalmus maximus) weighed about 10 g from an aquaculture farm in Shandong Province, China, were fed and acclimated for 30 days before experiment. Three turbot groups were intraperitoneally injected with sterilized saline (0.1 ml per tail), WED suspension (10 7 CFU per tail) and WED(ironQS-G) suspension (10 7 CFU per tail), respectively. The immunized fish were reared in aquaria supplied with a continuous flow of recycling water at 16-18°C. For challenge, 4 weeks later, each immunized group was divided into 2 challenge subgroups which were intramuscularly injected with wild type A. hydrophila (8 × 10 7 CFU per tail) and