In Vivo Programmed Gene Expression Based on Artificial QuorumNetworks

Teng Chu,a Yajun Huang,a Mingyu Hou,a Qiyao Wang,a Jingfan Xiao,a Qin Liu,a,b Yuanxing Zhanga,b

State Key Laboratory of Bioreactor Engineering, East China University of Science and Technology, Shanghai, Chinaa; Shanghai Collaborative Innovation Center forBiomanufacturing, Shanghai, Chinab

The quorum sensing (QS) system, as a well-functioning population-dependent gene switch, has been widely applied in manygene circuits in synthetic biology. In our work, an efficient cell density-controlled expression system (QS) was established viaengineering of the Vibrio fischeri luxI-luxR quorum sensing system. In order to achieve in vivo programmed gene expression, asynthetic binary regulation circuit (araQS) was constructed by assembling multiple genetic components, including the quorumquenching protein AiiA and the arabinose promoter ParaBAD, into the QS system. In vitro expression assays verified that thearaQS system was initiated only in the absence of arabinose in the medium at a high cell density. In vivo expression assays con-firmed that the araQS system presented an in vivo-triggered and cell density-dependent expression pattern. Furthermore, thearaQS system was demonstrated to function well in different bacteria, indicating a wide range of bacterial hosts for use. To ex-plore its potential applications in vivo, the araQS system was used to control the production of a heterologous protective antigenin an attenuated Edwardsiella tarda strain, which successfully evoked efficient immune protection in a fish model. This worksuggested that the araQS system could program bacterial expression in vivo and might have potential uses, including, but notlimited to, bacterial vector vaccines.

Bacteria release, detect, and respond to the accumulation ofself-generated chemical signaling molecules called autoinduc-

ers (AIs), whose concentration correlates to the abundance of se-creting microorganisms in the vicinity (1, 2). This process, termedquorum sensing (QS), allows bacteria to monitor a variety of pro-cesses, including competence, bioluminescence, virulence factorsecretion, biofilm formation, and sporulation (3). A paradigm ofquorum sensing regulation is the luminescence (lux) operon inVibrio fischeri (4). The lux genes are organized in two divergentlytranscribed units (5). LuxI is the autoinducer synthase producingan acyl-homoserine lactone (AHL), N-(3-oxohexanoyl)-homo-serine lactone (6). At low cell density, it is transcribed at a low basallevel (7, 8). As the V. fischeri culture grows and the autoinducersaccumulate to a specific threshold, cytoplasmic LuxR proteinsform a complex with autoinducers and initiate a positive autoin-duction circuit by binding to the luxICDABE promoter (9, 10).Numerous bacteria secrete enzymes that can degrade the QS sig-nals of heterologous species. The approach of interfering with ordestroying the QS signal is referred to as quorum quenching (11).The AiiA protein produced by Bacillus sp. is a typical AHL lacto-nase, which can quench quorum sensing systems by hydrolyzinglactone rings of AHL signal molecules (12). Therefore, introduc-tion of the aiiA gene into quorum sensing circuits renders the QSsystem under fine negative control.

Bacterial QS signal synthases, receptors, and cognate promoterelements are important components of a wide variety of engi-neered biological devices (13). As a well-functioning and popula-tion-dependent gene switch, the QS system has been widely ap-plied to many gene circuits in synthetic biology. Several syntheticbiology studies have employed the entire lux regulatory module(14–16), which served as an oscillator (17), amplifier (18), and soon. For example, Basu et al. (15) constructed a synthetic multicel-lular system in which the engineered receiver cells were designedto form ringlike patterns by expressing reporter proteins of differ-ent colors according to AHL concentrations. You et al. (16) built

and characterized a population control circuit that autonomouslyregulated the density of an Escherichia coli population by couplingQS with the synthesis of toxic proteins. Bulter et al. (19) developedan artificial QS system to establish a gene-metabolic networkwhere bacterial cells exchanged acetate molecules that were pro-duced in proportion to cell growth.

Although synthetic QS genetic circuits carried by prokaryotesperformed fairly well in vitro with the addition of exogenous in-ducers to medium to switch the system on or off, how to achievetheir in vivo application in eukaryote hosts is a new topic. Becausethe addition of inducers in vivo is infeasible, it is essential to designa gene switch to initiate the QS system by sensing a specific in vivoenvironmental signal. Guan et al. (20) established an in vivo in-ducible suicide circuit by introducing an iron-regulated pro-moter, which could activate the expression of a lysis gene in re-sponse to the iron-limiting signal in vivo and lyse the bacteriaeffectively. Besides low iron concentrations (21), the lack of ara-binose is another typical in vivo signal. Kong et al. (22) introduceda plasmid carrying the arabinose-regulated genes murA and asdAinto a Salmonella enterica serovar Typhimurium �murA �asdAmutant, and the in vivo expression of murA and asdA was strictlyturned off after the bacteria invaded host tissues.

The main purpose of our work was to achieve the self-initiation

Received 3 April 2015 Accepted 11 May 2015

Accepted manuscript posted online 15 May 2015

Citation Chu T, Huang Y, Hou M, Wang Q, Xiao J, Liu Q, Zhang Y. 2015. In vivoprogrammed gene expression based on artificial quorum networks. Appl EnvironMicrobiol 81:4984 –4992. doi:10.1128/AEM.01113-15.

Editor: M. A. Elliot

Address correspondence to Qin Liu, qinliu@ecust.edu.cn.

Copyright © 2015, American Society for Microbiology. All Rights Reserved.

doi:10.1128/AEM.01113-15

4984 aem.asm.org August 2015 Volume 81 Number 15Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from

of the quorum sensing system only in an in vivo environment. Itwas essential to design a gene switch to initiate the QS system bysensing a specific signal in the in vivo environment. We con-structed an in vivo-activated expression system called araQS bycoupling an artificial QS expression module and an arabinose-regulated QS quenching control module. The expression moduleensured the efficient production of the target protein by the quo-rum sensing system at high cell densities. The control module wascomposed of the arabinose-activated promoter ParaBAD and thequorum quenching gene aiiA. The quorum quenching proteinAiiA effectively stopped the function of the quorum sensing sys-tem by degrading autoinducers in the presence of arabinose. Theexpression of AiiA was controlled by the ParaBAD promoter, whichcould sense arabinose in the environment. Thus, when the bacte-rial host harboring the araQS system was cultivated in an arabi-nose-rich in vitro environment, the expression of the AiiA proteininhibited the QS system and cut off the whole expression module.However, in an arabinose-free environment, such as in vivo, theabsence of the AiiA protein derepressed the QS system, and thetarget protein was expressed effectively. Subsequent expressionassays, both in vitro and in vivo, demonstrated that the araQSsystem was well regulated by the signals of arabinose and cell den-sity and has great potential for in vivo applications, such as servingin multivalent bacterial vector vaccines.

MATERIALS AND METHODSStrains and medium. Strains and plasmids used in this study are listed inTable 1. Escherichia coli strains were grown at 37°C in Luria-Bertani (LB)medium (1% tryptone, 0.5% yeast extract, 1% NaCl). Aeromonas hydro-phila, Vibrio anguillarum, Salmonella Typhimurium, Staphylococcus au-reus, and Edwardsiella tarda strains were grown at 30°C in LB medium.Strains were grown in 250-ml flasks with a 50-ml working volume in ashaker at 200 rpm. When required, ampicillin (100 �g/ml) or kanamycinsulfate (100 �g/ml) was added to the medium for plasmid selection, andarabinose (0.4 mg/ml) or the autoinducer (10 nM) was added for induc-tion (23).

Plasmid construction. General DNA operations were conducted ac-cording to standard protocols. Seamless cloning operations were carried

out by using a ClonExpress II one-step cloning kit (Vazyme Co. Ltd.,China) according to standard procedures. DNA sequencing and primersynthesis (Table 2) were carried out by Generay Co. Ltd. (Shanghai,China).

Three cell density-dependent expression systems, QS1 to QS3, weredesigned on the basis of plasmid pUTat. The fundamental plasmid pUTatwas constructed by replacing the lac promoter, the multiple-cloning site(MCS), and the replicon of pUC18 with a 506-bp sequence containing theMCS and the rrnBT1T2 terminator from plasmid pBV220 and a pAT153replicon, respectively (24). Plasmid QS1 was constructed by inserting aluxI promoter fused with a Katushka reporter gene into pUTat. PlasmidQS2 was constructed by adding an intact luxI-luxR quorum sensing cir-cuit into QS1. Plasmid QS3 was constructed by inserting a luxI-luxR quo-rum sensing circuit into pUTat, with a Katushka reporter gene fused inframe with luxI. Finally, a ParaBAD-aiiA control module was inserted intoQS3 with a transcription terminator between the control module andquorum sensing module to form a binary QS system (araQS). PlasmidDNA was introduced into E. coli by heat shock and was introduced intoother strains by electroporation.

Expression assay in vitro. Bacterial cultures grown overnight wereinoculated (1:1,000, vol/vol) into fresh LB medium with antibiotics,and inducers such as arabinose were preadded if needed. For fed-batchcultures, fresh LB medium with antibiotics (and arabinose or autoin-ducer if needed) was continuously added to the culture vessel to keepthe optical density at 600 nm (OD600) lower than 0.3, below the celldensity threshold for the activation of the established quorum sensingsystem. At different time intervals, each culture was sampled and cen-trifuged at 8,000 � g for 5 min, and the harvested cells were washedwith phosphate-buffered saline (PBS) three times and resuspended inPBS. To measure the expression level per bacterial unit, all the sampleswere diluted to the same OD600 value (OD600 � 1.0), and 100 �l ofeach cell suspension was added to a 96-well flat-bottom polystyreneplate (Costar, USA) for fluorescence detection. Katushka protein wasdetected at excitation and emission wavelengths of 588 nm and 633nm, respectively. The relative fluorescence value (RFV) per bacterialunit (RFV/OD600) was set as the evaluation parameter to compare theexpression efficiencies of different systems.

Expression assay in zebrafish larvae. Bacterial cultures of E. coli-(araQS) grown overnight were inoculated (1:1,000, vol/vol) into fresh LBmedium with antibiotics and arabinose. The bacteria were harvested and

TABLE 1 Strains and plasmids used in this study

Bacterial strain or plasmid Description Source

StrainsE. coli BL21 Expression host; F� ompT gal dcm lon hsdSB Life TechnologiesVibrio fischeri MJ11 Wild type; used for cloning of the luxR-luxI quorum sensing system MCCCa

Vibrio anguillarum MVM425 Wild type; fish pathogen, broad-range testing host Our laboratoryE. tarda EIB202 Wild type; fish pathogen, broad-range testing host Our laboratoryE. tarda WED Mutant disrupted in type III secretion system and chorismic acid synthesis; live attenuated vaccine Our laboratoryS. aureus Wild type; human pathogen, broad-range testing host Our laboratoryS. Typhimurium Wild type; human pathogen, broad-range testing host Our laboratoryB. thuringiensis BT4Q7 Gene source of the quorum quenching gene aiiA Our laboratory

PlasmidspET-28a Expression vector; Kanr NovagenpUTat Expression vector; Ampr Our laboratoryQS1 pUTat vector containing a Katushka reporter gene and an isolated PluxI promoter This workQS2 QS1 plasmid inserted with intact V. fischeri QS gene elements This workQS3 pUTat vector containing intact V. fischeri QS gene elements and a Katushka reporter gene

cotranscribed with luxIThis work

araQS QS3 plasmid with insertion of a ParaBAD-aiiA gene module This workaraQS-G araQS derivative in which the reporter gene was replaced by the gapA34 gene This work

a MCCC, Marine Culture Collection of China.

In Vivo Gene Expression Based on Quorum Networks

August 2015 Volume 81 Number 15 aem.asm.org 4985Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from

suspended in PBS. Zebrafish larvae at 6 to 8 days of age were immersed inthe cell suspension at a concentration of 108 viable bacteria per ml for 2 hand then washed and transferred into freshwater. At the target time point,larvae were washed with PBS at least three times and anesthetized with0.01% (wt/vol) MS-222 anesthetic. The fish larvae were observed by usingan inverted fluorescence microscope (DMI3000 B; Leica). Meanwhile, 5larvae at each time point were homogenized with lysis buffer and used tocoat LB agar plates to count the corresponding number of bacteria.

Expression assay in zebrafish adults. The bacterial suspension of E.coli(araQS) was prepared as described above. Healthy adult zebrafish withan average weight of 0.2 g were intraperitoneally injected with an E. coli-(araQS) suspension at a dose of 107 CFU per tail. At 4 h, 12 h, and 24 hpostinjection, the internal organs of 20 fish from each group were surgi-cally extracted. After being homogenized and concentrated by ultrafiltra-tion, the tissue samples were analyzed by Western blotting using an anti-body specific to Katushka. Meanwhile, zebrafish internal organs taken atdifferent time points were treated with tissue lysis buffer and used to coatLB agar plates to count the number of bacteria in zebrafish.

All in vitro and in vivo assays were repeated at least three times foraccuracy.

Vaccination and challenge. The gapA34 gene, encoding glyceralde-hyde-3-phosphate dehydrogenase (GAPDH) from Aeromonas hydro-phila, was inserted into araQS, and the resultant recombinant plasmid,araQS-G, was transformed into attenuated E. tarda vaccine strain WED tocreate WED(araQS-G). The potential multivalent vaccine candidateWED(araQS-G) was evaluated for immune protection in turbot (Scoph-thalmus maximus). Attenuated E. tarda strain WED was used as the con-trol. All the vaccination and challenge experiments were done in triplicate.

The turbot, from a mariculture farm in Shandong Province, China,and weighing �10 g, were fed and acclimated for 30 days before theexperiment. Three turbot groups were intraperitoneally injected withsterilized saline (0.1 ml per tail), a WED suspension (107 CFU per tail),and a WED(araQS-G) suspension (107 CFU per tail). The immunized fishwere reared in aquaria supplied with a continuous flow of recycling waterat �16°C to 18°C. Four weeks later, each immunized group was dividedinto 2 challenge subgroups, which were intramuscularly injected withwild-type A. hydrophila (8 � 107 CFU per tail) and wild-type E. tardaEIB202 (6 � 103 CFU per tail). The mortality was recorded for 4 weeksafter challenge, and the observation of surviving fish was extended to 6weeks. Significant differences and relative percent survival (RPS) valueswere calculated by using Fisher’s exact test and according to the formula

RPS � (1 � mortality rate of vaccinated fish/mortality rate of control fish) �100, respectively.

All the animal experiments were approved by the Institutional AnimalCare and Use Committee of East China University of Science and Tech-nology.

RESULTSConstruction of cell density-dependent expression systems. Thequorum sensing elements, including luxI, luxR, and their respec-tive promoters, were cloned from V. fischeri. Three plasmids, QS1to QS3, were designed based on different strategies (Fig. 1A) andtransformed into E. coli BL21 strains for function tests in vitro.QS1, containing only the luxI promoter, almost failed to expressKatushka. This finding suggested that the luxI-luxR quorum sens-ing circuit from the E. coli genome could not activate the luxIpromoter of V. fischeri (Fig. 1B), possibly because of species spec-ificity (25, 26). More importantly, the single luxI promoter lost itsvitality without the help of other quorum sensing components.Comparatively, both QS2 and QS3 were designed to contain intactV. fischeri QS gene elements, and the only difference between themis that the Katushka gene was transcribed from a separate PluxI

promoter in QS2, while it was cotranscribed with luxI in QS3. Theresults (Fig. 1B) showed that QS3 displayed a dramatically in-creased expression level, comparable that of to highly efficientpET expression system, while QS2 displayed only a low expressionlevel. The low efficiency of QS2 might be caused by the presence oftwo individual luxI promoters competitively binding with a lim-ited number of LuxR-AI complexes, which act as transcriptionactivators to initiate the QS system. For QS3, the expression levelcontinued to decrease in the first 3 h and then started to increaseafter 3 h. These data indicated that QS3 was not activated until thecell density reached the threshold (OD of 0.47 for QS3 system) at3 h postinoculation. To further confirm the cell density-depen-dent behavior of QS3, a fed-batch culture model was establishedto keep the cell density of the bacterial culture below the thresholdby feeding fresh medium continuously. Concurrently, a contrast-ing experiment was conducted by adding medium with autoin-ducer molecules into the fed-batch culture. As shown in Fig. 1C,

TABLE 2 Primers used in this study

Primer Sequence (5=–3=)PluxI-F CATAAGTACCTGTAGGATCGTACAGGTTTACGCAAGAAAATGGTTTGTTATAGTCGAATAAACGCAAGGGAGGTTGGTPluxI-R ACCAACCTCCCTTGCGTTTATTCGACTATAACAAACCATTTTCTTGCGTAAACCTGTACGATCCTACAGGTACTTATGPIka-F GAATAAACGCAAGGGAGGTTGGTATGGTGGGCGAGGATAGCGTGPIkaTT-R AAGCTTGGCTGCAGGTCGACTCAGCTGTGACCCAGCTTGCQSkaTT-F ATAAAGAATTCCTTACGTACTTAATTTTTAQSkaTTO-R CGCCCTTGCTCACCATATAATTTCCTTTAATTAATTTAAGACTGCQSkaTTO-F GCAGTCTTAAATTAATTAAAGGAAATTATATGGTGAGCAAGGGCGQSkaTT-R ATCCGGATCCCAAAAAACCCCTCAAGACCCGTTTAGAGGCCCCAAGGGGTTATGCTAGTTACTTGTACAGCTCGTCCATpUTQS-F TTCCCGGGGATCCGTCGACTTAACTTTTAAAGTATGGGCAATCAATTGCQSkatushka-OR CGCTATCCTCGCCCACCATATAATTTCCTTTAATTAATTTAAGACTGCQSkatushka-OF GCAGTCTTAAATTAATTAAAGGAAATTATATGGTGGGCGAGGATAGCGkatushkapUT-R AACAGCCAAGCTTGGCTGCAGTCAGCTGTGACCCAGCTTGCTCpUTQK-F CTGCAGCCAAGCTTGGCTGTTTTGGpUTQK-R GTCGACGGATCCCCGGGAATTCCTCCaraQS1 ATGGTGCACTCTCAGTACAATCAAAAAACCCCTCAAGACCCGTTTAGAGaraQS2 GGCTAACAGGAGGAATTAACCATGACAGTAAAGAAGCaraQS3 GCTTCTTTACTGTCATGGTTAATTCCTCCTGTTAGCCpUTPA-F ATTGTACTGAGAGTGCACCATATGGAGTTTGpUTPA-R CTGCTCTGATGCCGCATAGTTAAGCCAG

Chu et al.

4986 aem.asm.org August 2015 Volume 81 Number 15Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from

the RFV of the QS3 strain in the fed-batch culture dropped in-stantly and stayed low the entire time, showing that the quorumsensing circuit remained inactivated when the microbial popula-tion density was kept below the threshold level. This result con-firmed that a small bacterial population of QS3 could not synthe-size enough autoinducer molecules to activate the QS system.Simultaneously, the addition of exogenous autoinducers signifi-cantly activated QS3 in fed-batch cultivation, even with the celldensity being below the threshold level. All these data confirmedthat the artificial QS3 system is an efficient and cell density-regu-lated expression system.

Design of a binary QS system that responds to arabinose andcell density. Although QS3 performed well in regulating gene ex-pression in a cell density-dependent manner in vitro, it is ratherchallenging to control the switch of the QS system by a specific invivo signal for potential applications in eukaryotic hosts. Here, aParaBAD-aiiA gene module was inserted into QS3 to form a newbinary QS system, araQS. In this circuit (Fig. 2A), arabinose, as aninducer, activates the transcription of the aiiA gene from ParaBAD

(27, 28), and the expressed AiiA proteins repress the PluxI pro-moter by degrading autoinducers and quench the QS3 systemeven at high cell density. Since arabinose is often absent in eukary-otic hosts, the ParaBAD-aiiA gene module was designed to act as anarabinose-responsive molecular switch to negatively regulate theQS system in vivo.

The performance of araQS in the E. coli BL21 host was as-sayed by adding arabinose to the culture medium. The araQS sys-tem was severely inhibited in the presence of arabinose (Fig. 2B),indicating that arabinose-induced AiiA proteins remarkably re-pressed the QS system by degrading autoinducer molecules. On

the contrary, the araQS system was significantly derepressed in theabsence of arabinose because the quorum quenching protein AiiAwas not synthesized. All these data confirmed that the ParaBAD-aiiAgene module could negatively control the QS system by respond-ing to arabinose in the medium. The araQS system was furtheranalyzed in the fed-batch culture model (Fig. 2C). As long as thecell density was below the threshold value, the araQS system wasalways kept switched off, even in arabinose-free medium, unlessan exogenous autoinducer was added. The results confirmed thatthe araQS system, composed of the ParaBAD-aiiA regulation mod-ule and a quorum sensing circuit, exhibits a precise dually arabi-nose- and density-regulated expression feature.

In vivo performance of the araQS system. Although araQSwas demonstrated to be an efficient binary system that responds toarabinose and cell density in vitro, its performances in vivo stillremained unknown. In the following work, zebrafish larvae andadults were applied as animal models to evaluate the in vivo per-formances of araQS. The recombinant E. coli strain carryingaraQS was administered to zebrafish larvae by immersion, andboth fluorescence intensity and bacterial counts in larvae werethen monitored at different intervals. Until 6 h postimmersion, noobvious Katushka fluorescence signal was detected (Fig. 3A),which was in accordance with the in vivo propagation kinetics ofthe E. coli strains (Fig. 3B). The results indicated that the araQSsystem functioned well in zebrafish larvae. Responding to the ar-abinose-free signal in vivo, the araQS system was derepressed byswitching off aiiA transcription from ParaBAD, and simultaneously,araQS was activated by locally accumulated autoinducers from anincreasing microbial population in vivo. In addition, E. coli-(araQS) was administered to zebrafish adults by injection, and the

FIG 1 Construction and in vitro performance of QS. (A) Structures of the cell density-regulated expression systems QS1 to QS3. PluxI, promoter of the luxI gene;PluxR, promoter of the luxR gene; TT, transcription terminator; RBS, ribosome binding site. Katushka is a reporter gene encoding red fluorescent protein. (B)Protein expression levels of E. coli BL21 strains containing QS1 to QS3 were evaluated by determining the relative fluorescence values (RFV) of red fluorescentprotein. The BL21(DE3) strain containing a traditional T7 expression system was configured as a control. (C) Cell density-regulated expression of the QS3system. E. coli BL21 harboring the QS3 system was cultivated in a fed-batch model, and the OD600 was maintained below 0.3 by feeding fresh mediumcontinuously. Autoinducers of the QS system were exogenously added to the medium or not.

In Vivo Gene Expression Based on Quorum Networks

August 2015 Volume 81 Number 15 aem.asm.org 4987Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from

in vivo expression pattern of araQS was evaluated. As shown inFig. 3C and D, the araQS system was activated significantly after 4h postinjection upon in vivo propagation of the bacteria. All thedata confirmed that the araQS system was effectively activated in

the zebrafish model through responding to both arabinose-freeand cell density signals, displaying a dually regulated expressionfeature.

Expression of araQS in a wide range of bacterial hosts. Adap-

FIG 2 Construction and in vitro performance of araQS. (A) Linear sketch of araQS. aiiA is quorum quenching gene from B. thuringiensis; ParaBAD is the arabinosepromoter. (B) Protein expression of the araQS system regulated by an arabinose signal. Arabinose was added to the medium or not. Cell cultures were harvestedand adjusted to an OD600 of 1 for fluorescence detection at the indicated time points. (C) Cell density-regulated expression of araQS. E. coli BL21 harboring araQSwas cultivated via fed-batch culture, and the OD600 was maintained below 0.3 by feeding fresh medium continuously. Autoinducers of QS were exogenouslyadded to the medium or not.

FIG 3 In vivo performance of araQS. (A and B) Protein expression of E. coli(araQS) in zebrafish larvae. Fish larvae, immersed in a bacterial suspension for 2 h,were observed by using an inverted fluorescence microscope at different time intervals, and the corresponding number of bacteria in larvae was counted bycoating LB agar plates with homogenized larvae at each time point. (C and D) Protein expression of E. coli(araQS) in adult zebrafish. The internal organs of 20fish from each group were surgically extracted at different time intervals. The tissue samples were further analyzed by Western blotting using an antibody specificto Katushka, and the corresponding number of bacteria in adult zebrafish was counted by coating LB agar plates with the tissue samples.

Chu et al.

4988 aem.asm.org August 2015 Volume 81 Number 15Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from

tation to different bacterial host environments is an importantindicator of a trustworthy expression system. Since araQS wascomposed of the quorum-related genes luxI and luxR from V.fischeri, aiiA from Bacillus thuringiensis, and the arabinose pro-moter, it is of interest to test whether these heterologous geneelements could be coordinated in different bacterial hosts. Here,araQS was introduced into different Gram-negative bacteria, in-cluding the fish pathogens Edwardsiella tarda and Vibrio anguilla-rum and the human pathogen Salmonella Typhimurium, and theGram-positive bacterium Staphylococcus aureus. The expressionperformances of these recombinant strains were then assayed invitro (Fig. 4). All four strains bearing araQS presented high proteinexpression levels and a binary regulation pattern similar to that inE. coli. These data suggested that the heterologous gene elementsof araQS were compatible with different bacteria and might havethe potential to be applied in a wide range of bacterial hosts.

Application of araQS in a multivalent bacterial vaccine. Amultivalent bacterial vaccine is constructed via the expression of aheterologous antigen in a live bacterial vector, and this is an at-tractive vaccine strategy to induce an immune response to thecarried protective antigen. How to enhance antigen expressionand delivery without reducing vaccine stability and vitality haslong been challenging researchers. Constitutive expression of aheterologous antigen in attenuated bacteria often causes obviousmetabolic burdens and thus results in declining in vivo coloniza-tion for the live bacterial vaccine (29). To overcome the shortcom-ings of constitutive expression, in vivo-inducible promoters have

been used to regulate antigen expression (30). In this work, the invivo-triggered araQS system was used to produce antigen in vivoafter the live recombinant vaccine was administered to the host.

Edwardsiella tarda is an important facultative intracellularpathogen of both animals and humans (31). An attenuated E.tarda strain, called WED, was constructed by in-frame deletion ofeseBCD and aroC, and this attenuated strain was proven to be anefficient live bacterial vaccine for fish (32). In order to establish asuitable expression system for application in a multivalent E. tardavaccine, the in vivo-triggered araQS system and the in vitro-ex-pressed QS3 system were used to express heterologous protein inWED. Compared to WED(araQS), WED(QS3) displayed signifi-cantly inhibited growth in vitro (Fig. 5A) and also an impairedcolonization ability in vivo in fish (Fig. 5B), possibly due to themetabolic burden of heterologous protein expression. The resultssuggest that in vivo-triggered expression of a heterologous antigenfacilitates bacterial colonization in vivo and thus is more suited forvector vaccine design.

In order to design a multivalent E. tarda vaccine using thearaQS system, the protective antigen gene gapA34 from the patho-genic organism A. hydrophila, encoding GAPDH (33, 34), wasinserted into the araQS plasmid and transformed into attenuatedE. tarda strain WED, forming a multivalent vaccine candidate,WED(araQS-G) (Fig. 6A). As shown in Fig. 6B, GADPH was ef-fectively produced in WED(araQS-G) under arabinose-free andhigh-cell-density conditions. To further investigate the in vivo ap-plication of WED(araQS-G) as a multivalent vaccine, protection

FIG 4 In vitro performance of araQS in different bacterial hosts. The araQS plasmid was transformed into Edwardsiella tarda, Vibrio anguillarum, SalmonellaTyphimurium, and Staphylococcus aureus, and the resultant recombinant bacterial hosts were cultivated in LB medium with or without the addition of arabinose.Simultaneously, the recombinant bacteria were cultivated in a fed-batch model to maintain the OD600 below 0.3 by feeding fresh medium continuously. Each ofthe cultures was harvested and adjusted to an OD600 of 1 for fluorescence detection at the indicated time points.

In Vivo Gene Expression Based on Quorum Networks

August 2015 Volume 81 Number 15 aem.asm.org 4989Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from

efficacy was evaluated in turbot, an important cultured fish inChina. The multivalent vaccine WED(araQS-G) showed signifi-cant protection against challenge with E. tarda EIB202 (RPS �73.3%) and A. hydrophila LSA34 (RPS � 68.6%) over the salinecontrol (Table 3), while the monovalent vaccine WED triggeredsignificant protection against E. tarda EIB202 (RPS � 76.7%) butonly a slight response against A. hydrophila LSA34 (RPS � 13.9%).These data confirmed that araQS-mediated expression of GAPDHin WED effectively protected turbots from experimental challengewith A. hydrophila LSA34 while exerting no adverse effect on pro-tection against E. tarda infection.

DISCUSSION

The engineering of gene regulatory networks is a cornerstone ofsynthetic biology (35). Cell-cell communication represents a valu-able mechanism to engineer novel signaling constructs in bacteria(36). The genetic elements of the V. fischeri quorum sensing circuithave been successfully used in several artificial genetic systems (15,17, 37). However, all of these systems were intended to work invitro by adding inducers to initiate the pathway of circuits. Ahighly efficient and cell density-dependent QS system has beendesigned by rewiring quorum sensing genetic elements of V. fis-

cheri into E. coli (38). In order to engineer the QS system for use invivo, it is essential to integrate an additional regulation module tokeep the artificial network inactivated in vitro unless it senses aspecial in vivo signal. Generally, there are two ways to manipulatethe QS circuit, either positively by an in vivo-induced activator ornegatively by an in vitro-induced inhibitor. In this work, the araQSsystem with binary regulation was designed for in vivo applica-tions based on the latter strategy. Because of the absence of arabi-nose in almost all kinds of in vivo environments, the ParaBAD-aiiAmodule, consisting of the arabinose-activated promoter ParaBAD

and the quorum quenching gene aiiA, was constructed and cou-pled to the QS system. The resultant araQS system exhibited an invivo-triggered and cell density-dependent expression pattern in invivo environments such as zebrafish models.

The araQS system was established as a chimeric system, com-prising an optimized QS circuit from V. fischeri, the arabinosepromoter ParaBAD, and the aiiA gene from B. thuringiensis. ThearaQS system displayed a high level of compatibility with a varietyof bacterial hosts, indicating araQS forms a self-contained andclosed-loop expression circuit that is less dependent on hosts. Al-though some bacterial hosts used in this work might possess theirown QS systems and produce their own AHL autoinducers, nocross talk or interference between the native bacterial host QS andthe heterologous araQS systems was found. This could be attrib-uted mainly to the fact that the AHL autoinducers differ in theshape of the acyl side chain between species and specifically acti-vate the QS circuit of their own species (6, 26, 39). All these datasuggested that the araQS system has great potential as a generalexpression tool in a wide variety of bacterial hosts. However, thearaQS system was shown to be initiated at a very early stage, when

FIG 5 In vitro growth and in vivo colonization of QS3 and araQS. (A) Growthof E. tarda WED loaded with the QS3 or araQS plasmid in arabinose-richmedium. The blank plasmid pUTat was used as the control. (B) Counts ofviable bacteria in zebrafish injected with plasmid-loaded WED strains (5 � 104

CFU per tail) at 12, 24, and 48 h. Ten fish were set as a pool, and three parallelexperiments were done. At the indicated time points, 10 fish from each groupwere randomly picked and homogenized, and the corresponding number ofbacteria was counted by coating LB agar plates with homogenized fish.

FIG 6 Construction of the recombinant vector vaccine WED(araQS-G). (A)Plasmid construction of araQS-G. (B) Expression analysis of GAPDH inWED(araQS-G) by SDS-PAGE. Recombinant bacteria were cultivated underdifferent conditions for 9 h. HCD, high cell density; LCD, low cell density;ara �, medium with arabinose added; ara �, arabinose-free medium; M, pro-tein molecular marker.

Chu et al.

4990 aem.asm.org August 2015 Volume 81 Number 15Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from

the bacteria were still in a small population both in vitro and invivo. This might be explained by the fact that multiple copies of thearaQS plasmid make bacteria produce higher levels of autoinduc-ers, and thus, accumulated autoinducers could reach the thresh-old at a lower cell density. Thus, bacteria bearing araQS exhibit amuch lower threshold than V. fischeri with a single copy of the QSsystem on its chromosome. Therefore, if a delayed initiation ofaraQS in vivo is expected, a low-copy-number plasmid or single-copy integration into the chromosome is a reasonable choice.

Bacterial vector vaccines are an attractive vaccination strategyto induce multiple immune responses to both the bacterial vectorand the carried protective antigen (40). The superiorities of livebacterial vaccines are essentially based on their mimicry of naturalinfection in hosts (41). A typical bacterial infection involves around of sequential in vivo events, including invasion, coloniza-tion, propagation, and then dissemination. At the early stage ofinfection, invading bacteria are in a small population at the site ofinfection, and they are required to contend desperately with rig-orous host stresses for successful colonization. The additionalmetabolic burden of heterologous expression at this stage will un-wittingly attenuate the bacterium to some extent and then impaircolonization. Therefore, it is rather beneficial to turn off antigenexpression in bacteria until colonization is established in vivo (30,42). In this work, the araQS system, with an in vivo-triggered andpopulation-dependent expression pattern, was used to express aheterologous protective antigen in an attenuated bacterial strain.Compared with QS3, the araQS system kept protein expressionswitched off in vitro, which significantly facilitates bacterialgrowth in vitro and bacterial colonization in vivo. Therefore, thearaQS system has great potential for the design of a bacterial vectorvaccine.

In summary, an in vivo programmed expression circuit wasestablished in this work, which was derepressed by the arabinose-free signal in vivo and then programmed to produce heterologousproteins in a cell density-dependent manner. This artificial syn-thetic circuit might have the potential for in vivo applications,including, but not limited to, bacterial vector vaccines.

ACKNOWLEDGMENTS

This work was financially supported by grants from the National NaturalScience Funds of China (31272700), the National Key Technology Sup-port Program of China (2012BAD17B02), the National High Technology

Research and Development Program of China (2013AA093101), and theShanghai Rising Star Program (13QA1401000).

REFERENCES1. Garg N, Manchanda G, Kumar A. 2014. Bacterial quorum sensing:

circuits and applications. Antonie Van Leeuwenhoek 105:289 –305. http://dx.doi.org/10.1007/s10482-013-0082-3.

2. Miller MB, Bassler BL. 2001. Quorum sensing in bacteria. Annu Rev Micro-biol 55:165–199. http://dx.doi.org/10.1146/annurev.micro.55.1.165.

3. Bassler BL, Losick R. 2006. Bacterially speaking. Cell 125:237–246. http://dx.doi.org/10.1016/j.cell.2006.04.001.

4. Collins CH, Arnold FH, Leadbetter JR. 2005. Directed evolution ofVibrio fischeri LuxR for increased sensitivity to a broad spectrum of acyl-homoserine lactones. Mol Microbiol 55:712–723. http://dx.doi.org/10.1111/j.1365-2958.2004.04437.x.

5. Schaefer AL, Hanzelka BL, Eberhard A, Greenberg E. 1996. Quorumsensing in Vibrio fischeri: probing autoinducer-LuxR interactions withautoinducer analogs. J Bacteriol 178:2897–2901.

6. Eberhard A, Burlingame A, Eberhard C, Kenyon G, Nealson K, Op-penheimer N. 1981. Structural identification of autoinducer of Photobac-terium fischeri luciferase. Biochemistry 20:2444 –2449. http://dx.doi.org/10.1021/bi00512a013.

7. Egland KA, Greenberg E. 1999. Quorum sensing in Vibrio fischeri: ele-ments of the luxI promoter. Mol Microbiol 31:1197–1204. http://dx.doi.org/10.1046/j.1365-2958.1999.01261.x.

8. Kaplan HB, Greenberg E. 1985. Diffusion of autoinducer is involved inregulation of the Vibrio fischeri luminescence system. J Bacteriol 163:1210 –1214.

9. Hanzelka BL, Greenberg E. 1995. Evidence that the N-terminal region ofthe Vibrio fischeri LuxR protein constitutes an autoinducer-binding do-main. J Bacteriol 177:815– 817.

10. Urbanowski M, Lostroh C, Greenberg E. 2004. Reversible acyl-homoserine lactone binding to purified Vibrio fischeri LuxR protein. JBacteriol 186:631– 637. http://dx.doi.org/10.1128/JB.186.3.631-637.2004.

11. Raina S, Vizio DD, Odell M, Clements M, Vanhulle S, Keshavarz T.2009. Microbial quorum sensing: a tool or a target for antimicrobial ther-apy? Biotechnol Appl Biochem 54:65– 84. http://dx.doi.org/10.1042/BA20090072.

12. Dong Y-H, Xu J-L, Li X-Z, Zhang L-H. 2000. AiiA, an enzyme thatinactivates the acylhomoserine lactone quorum-sensing signal and atten-uates the virulence of Erwinia carotovora. Proc Natl Acad Sci U S A 97:3526 –3531. http://dx.doi.org/10.1073/pnas.97.7.3526.

13. Choudhary S, Schmidt-Dannert C. 2010. Applications of quorum sens-ing in biotechnology. Appl Microbiol Biotechnol 86:1267–1279. http://dx.doi.org/10.1007/s00253-010-2521-7.

14. Anderson JC, Clarke EJ, Arkin AP, Voigt CA. 2006. Environmentallycontrolled invasion of cancer cells by engineered bacteria. J Mol Biol 355:619 – 627. http://dx.doi.org/10.1016/j.jmb.2005.10.076.

15. Basu S, Gerchman Y, Collins CH, Arnold FH, Weiss R. 2005. A syntheticmulticellular system for programmed pattern formation. Nature 434:1130 –1134. http://dx.doi.org/10.1038/nature03461.

TABLE 3 Efficacy of the vaccine candidate WED(araQS-G) against E. tarda EIB202 and A. hydrophila LSA34 in turbot

ImmunogenVaccinationdose/tail

Challengestrain

Challenge dose(CFU/tail)

No. offishd

Mean mortality(%) SDa

Mean RPS SDb Effectc

Saline 0.1 ml EIB202 6 � 103 30 100 0LSA34 8 � 107 30 85.0 5.8

WED 107 CFU EIB202 6 � 103 30 23.3 4.7 76.7 4.7 �LSA34 8 � 107 30 73.3 5.8 13.9 6.8 �

WED(araQS-G) 107 CFU EIB202 6 � 103 30 26.7 4.0 73.3 4.0 �LSA34 8 � 107 30 26.7 6.7 68.6 7.8 �

a 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 and is calculated as follows: RPS � (1 � mortality of vaccinated fish/mortality of control fish) �100.c If the RPS was 60%, the vaccine candidate showed protection against challenge, indicated by a “�.” If the RPS was �30%, the vaccine candidate showed no protection againstthe challenge strain, indicated by a “�.”d The vaccination and challenge experiments were repeated three times.

In Vivo Gene Expression Based on Quorum Networks

August 2015 Volume 81 Number 15 aem.asm.org 4991Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from

16. You L, Cox RS, Weiss R, Arnold FH. 2004. Programmed populationcontrol by cell-cell communication and regulated killing. Nature 428:868 – 871. http://dx.doi.org/10.1038/nature02491.

17. Danino T, Mondragón-Palomino O, Tsimring L, Hasty J. 2010. Asynchronized quorum of genetic clocks. Nature 463:326 –330. http://dx.doi.org/10.1038/nature08753.

18. Nistala GJ, Wu K, Rao CV, Bhalerao KD. 2010. A modular positivefeedback-based gene amplifier. J Biol Eng 4:4. http://dx.doi.org/10.1186/1754-1611-4-4.

19. Bulter T, Lee SG, Wong WW, Fung E, Connor MR, Liao JC. 2004.Design of artificial cell-cell communication using gene and metabolic net-works. Proc Natl Acad Sci U S A 101:2299 –2304. http://dx.doi.org/10.1073/pnas.0306484101.

20. Guan L, Mu W, Champeimont J, Wang Q, Wu H, Xiao J, Lubitz W,Zhang Y, Liu Q. 2011. Iron-regulated lysis of recombinant Escherichia coliin host releases protective antigen and confers biological containment.Infect Immun 79:2608 –2618. http://dx.doi.org/10.1128/IAI.01219-10.

21. Touati D. 2000. Iron and oxidative stress in bacteria. Arch Biochem Bio-phys 373:1– 6. http://dx.doi.org/10.1006/abbi.1999.1518.

22. Kong W, Wanda SY, Zhang X, Bollen W, Tinge SA, Roland KL, CurtissR, III. 2008. Regulated programmed lysis of recombinant Salmonella inhost tissues to release protective antigens and confer biological contain-ment. Proc Natl Acad Sci U S A 105:9361–9366. http://dx.doi.org/10.1073/pnas.0803801105.

23. Hawkins AC, Arnold FH, Stuermer R, Hauer B, Leadbetter JR. 2007.Directed evolution of Vibrio fischeri LuxR for improved response to bu-tanoyl-homoserine lactone. Appl Environ Microbiol 73:5775–5781. http://dx.doi.org/10.1128/AEM.00060-07.

24. Xiao Y, Liu Q, Chen H, Zhang YX. 2011. A stable plasmid system forheterologous antigen expression in attenuated Vibrio anguillarum. Vac-cine 29:6986 – 6993. http://dx.doi.org/10.1016/j.vaccine.2011.07.036.

25. Ng WL, Bassler BL. 2009. Bacterial quorum-sensing network architec-tures. Annu Rev Genet 43:197–222. http://dx.doi.org/10.1146/annurev-genet-102108-134304.

26. Yao Y, Martinez-Yamout MA, Dickerson TJ, Brogan AP, Wright PE,Dyson HJ. 2006. Structure of the Escherichia coli quorum sensing proteinSdiA: activation of the folding switch by acyl homoserine lactones. J MolBiol 355:262–273. http://dx.doi.org/10.1016/j.jmb.2005.10.041.

27. Guzman LM, Belin D, Carson MJ, Beckwith J. 1995. Tight regulation,modulation, and high-level expression by vectors containing the arabi-nose PBAD promoter. J Bacteriol 177:4121– 4130.

28. Reeder T, Schleif R. 1991. Mapping, sequence, and apparent lack offunction of araJ, a gene of the Escherichia coli arabinose regulon. J Bacteriol173:7765–7771.

29. Loessner H, Endmann A, Leschner S, Bauer H, Zelmer A, zur Lage S,Westphal K, Weiss S. 2008. Improving live attenuated bacterial carriersfor vaccination and therapy. Int J Med Microbiol 298:21–26. http://dx.doi.org/10.1016/j.ijmm.2007.07.005.

30. Medina E, Guzmán CA. 2001. Use of live bacterial vaccine vectors forantigen delivery: potential and limitations. Vaccine 19:1573–1580. http://dx.doi.org/10.1016/S0264-410X(00)00354-6.

31. Leung KY, Siame BA, Tenkink BJ, Noort RJ, Mok Y-K. 2012.Edwardsiella tarda—virulence mechanisms of an emerging gastroenteritispathogen. Microbes Infect 14:26 –34. http://dx.doi.org/10.1016/j.micinf.2011.08.005.

32. Xiao J, Chen T, Liu B, Yang W, Wang Q, Qu J, Zhang Y. 2013.Edwardsiella tarda mutant disrupted in type III secretion system and cho-rismic acid synthesis and cured of a plasmid as a live attenuated vaccine inturbot. Fish Shellfish Immunol 35:632– 641. http://dx.doi.org/10.1016/j.fsi.2013.05.022.

33. Zhou L, Wang X, Liu Q, Wang Q, Zhao Y, Zhang Y. 2010. A novelmultivalent vaccine based on secretary antigen-delivery induces protec-tive immunity against Vibrio anguillarum and Aeromonas hydrophila. JBiotechnol 146:25–30. http://dx.doi.org/10.1016/j.jbiotec.2009.12.010.

34. Mu W, Guan L, Yan Y, Liu Q, Zhang Y. 2011. A novel in vivo inducibleexpression system in Edwardsiella tarda for potential application in bac-terial polyvalence vaccine. Fish Shellfish Immunol 31:1097–1105. http://dx.doi.org/10.1016/j.fsi.2011.09.015.

35. Benner SA. 2003. Synthetic biology: act natural. Nature 421:118. http://dx.doi.org/10.1038/421118a.

36. Purnick PE, Weiss R. 2009. The second wave of synthetic biology: frommodules to systems. Nat Rev Mol Cell Biol 10:410 – 422. http://dx.doi.org/10.1038/nrm2698.

37. Neddermann P, Gargioli C, Muraglia E, Sambucini S, Bonelli F, DeFrancesco R, Cortese R. 2003. A novel, inducible, eukaryotic gene expres-sion system based on the quorum-sensing transcription factor TraR.EMBO Rep 4:159 –165. http://dx.doi.org/10.1038/sj.embor.embor734.

38. Haseltine EL, Arnold FH. 2008. Implications of rewiring bacterial quo-rum sensing. Appl Environ Microbiol 74:437– 445. http://dx.doi.org/10.1128/AEM.01688-07.

39. Morohoshi T, Inaba T, Kato N, Kanai K, Ikeda T. 2004. Identificationof quorum-sensing signal molecules and the LuxRI homologs in fishpathogen Edwardsiella tarda. J Biosci Bioeng 98:274 –281. http://dx.doi.org/10.1016/S1389-1723(04)00281-6.

40. Shata MT, Stevceva L, Agwale S, Lewis GK, Hone DM. 2000. Recentadvances with recombinant bacterial vaccine vectors. Mol Med Today6:66 –71. http://dx.doi.org/10.1016/S1357-4310(99)01633-0.

41. Spreng S, Dietrich G, Weidinger G. 2006. Rational design of Salmonella-based vaccination strategies. Methods 38:133–143. http://dx.doi.org/10.1016/j.ymeth.2005.09.012.

42. Wang S, Li Y, Shi H, Sun W, Roland KL, Curtiss R, III. 2011. Comparisonof a regulated delayed antigen synthesis system with in vivo-inducible pro-moters for antigen delivery by live attenuated Salmonella vaccines. InfectImmun 79:937–949. http://dx.doi.org/10.1128/IAI.00445-10.

Chu et al.

4992 aem.asm.org August 2015 Volume 81 Number 15Applied and Environmental Microbiology

on February 11, 2020 by guest

http://aem.asm

.org/D

ownloaded from