Expression of the AHPND Toxins PirAvp and PirBvp Is Regulated by Components of the Vibrio parahaemolyticus Quorum Sensing (QS) System

Acute hepatopancreatic necrosis disease (AHPND) in shrimp is caused by Vibrio strains that harbor a pVA1-like plasmid containing the pirA and pirB genes. It is also known that the production of the PirA and PirB proteins, which are the key factors that drive the observed symptoms of AHPND, can be influenced by environmental conditions and that this leads to changes in the virulence of the bacteria. However, to our knowledge, the mechanisms involved in regulating the expression of the pirA/pirB genes have not previously been investigated. In this study, we show that in the AHPND-causing Vibrio parahaemolyticus 3HP strain, the pirAvp and pirBvp genes are highly expressed in the early log phase of the growth curve. Subsequently, the expression of the PirAvp and PirBvp proteins continues throughout the log phase. When we compared mutant strains with a deletion or substitution in two of the quorum sensing (QS) master regulators, luxO and/or opaR (luxOD47E, ΔopaR, ΔluxO, and ΔopaRΔluxO), our results suggested that expression of the pirAvp and pirBvp genes was related to the QS system, with luxO acting as a negative regulator of pirAvp and pirBvp without any mediation by opaRvp. In the promoter region of the pirAvp/pirBvp operon, we also identified a putative consensus binding site for the QS transcriptional regulator AphB. Real-time PCR further showed that aphBvp was negatively controlled by LuxOvp, and that its expression paralleled the expression patterns of pirAvp and pirBvp. An electrophoretic mobility shift assay (EMSA) showed that AphBvp could bind to this predicted region, even though another QS transcriptional regulator, AphAvp, could not. Taken together, these findings suggest that the QS system may regulate pirAvp/pirBvp expression through AphBvp.


Introduction
Vibrio parahaemolyticus is an opportunistic marine pathogen often found in the ocean and estuary environment [1]. In general, V. parahaemolyticus is recognized as an etiologic agent that causes acute gastroenterocolitis and diarrhea after human consumption of contaminated seafood [2,3]. Consequently, most earlier studies were on the pathogenesis of V. parahaemolyticus in humans and focused on virulence factors such as its hemolysins, T3SSs, and T6SSs [4][5][6]. However, several strains of V. parahaemolyticus were also identified as causative agents of the newly emergent acute hepatopancreatic necrosis disease (AHPND) in shrimp [7]. Since the first outbreak in China in 2009, AHPND has rapidly spread across southeast Asia and reached as far as central and south America, leading to huge losses in the aquaculture industry [8][9][10]. AHPND induces early mortality, usually within 35 to 45 days after stocking post-larvae shrimp in cultivation ponds [11]. Characteristic AHPND symptoms in shrimp include a pale and atrophied hepatopancreas (HP) with an empty stomach and midgut [7,9]. Histological examination has further shown that AHPND causes sloughing of the HP tubule epithelial cells into the HP tubule lumens [7,9], and this has become the main diagnostic criterion to confirm an AHPND infection. AHPND-causing strains harbor a 70-kbp plasmid (pVA1) that produces the "Photorhabdus insect-related" (Pir)-like binary toxins, PirA vp and PirB vp [12]. These two AHPND-associated toxins form a complex that is structurally homologous to the insecticidal Cry toxin [12][13][14], and they have been confirmed as the key factors that drive AHPND pathogenesis [12]. However, while previous studies have increased our understanding of the structural and functional characteristics of PirA vp and PirB vp , the mechanism by which the pirA vp and pirB vp genes are regulated still remains unknown. Here, we investigate this mechanism and show that the quorum sensing (QS) system may play an important role.
QS is a cell density-dependent process that regulates the expression of a number of genes in both Gram-positive and Gram-negative bacteria [15]. The QS system achieves this regulation by a series of control factors, including LuxO, OpaR (a homolog of LuxR), and AphA [16,17]. QS-regulated genes are involved in many important physiological activities, such as biofilm formation, bioluminescence, virulence factor production, conjugation, plasmid transfer, antibiotic production, cell mobility, and sporulation [15]. The importance of the QS system in AHPND pathogenicity was also recently demonstrated: extract from V. alginolyticus BC25 contained the anti-QS compounds Cyclo-(L-Leu-L-Pro) and Cyclo-(L-Phe-L-Pro), both of which had anti-QS activity, and pre-treatment with V. alginolyticus BC25 reduced mortality after challenge with the AHPND-causing strain V. parahaemolyticus PSU5591 [18]. However, this study did not investigate the mechanism by which the QS system regulates the virulence of AHPND-causing V. parahaemolyticus.
Here we show how LuxO vp , which is an important regulator of QS in Vibrio spp., affects the gene expression of the key AHPND pathogenic factors pirA vp and pirB vp . First, by monitoring the gene/protein expression of PirA vp and PirB vp during the growth of V. parahaemolyticus, we found the appearance of these two toxins is cell density-dependent. We then confirmed that, at the low cell density stage (LCD, OD 600 ≈ 0.6), the deletion of LuxO vp significantly increased the gene/protein expression of PirA vp and PirB vp in V. parahaemolyticus. Next, in a PirA vp /PirB vp operon analysis, a possible DNA binding site for the Vibrio regulator AphB was identified. AphB is known as an activator, which plays a central role in virulence gene expression in both Vibrio cholerae and Vibrio alginolyticus [19,20]. It also plays a regulatory role as a QS control factor of LuxR [20]. Here, using EMSA (electrophoretic mobility shift assay), we confirmed that V. parahaemolyticus AphB (AphB vp ) bound to the proposed promoter region of the PirA vp /PirB vp operon. Using real-time polymerase chain reactions, we also found a correlation between the expression levels of LuxO vp and AphB vp . At low cell density, the expression of AphB vp was increased by 1.7 fold in LuxO vp -deleted V. parahaemolyticus, and this increase was positively correlated to the gene/protein expression of PirA vp and PirB vp under the same conditions. Taken together, our results show that the pirA vp /pirB vp genes are regulated by components of the QS system, particularly by AphB vp . This is the first report to investigate the influence of the bacterial physiological system on the pirA vp /pirB vp genes. Our findings will be helpful for the development of APHND prevention and/or control strategies in the future.

Results
2.1. The Expression Levels of pirA vp /pirB vp during Different Growth Phases of V. parahaemolyticus In Vibrio, since QS is involved in the regulation of many physiological processes, including several virulence-related systems, we first wanted to determine if the expression level of pirA vp /pirB vp was cell density-dependent. We, therefore, recorded the growth curve of the wild type V. parahaemolyticus, 3HP strain, and analyzed the gene expression patterns of pirA vp and pirB vp . As shown in Figure 1A, the lag phase lasted until about 3.5 h (OD 600 from 0.01 to~0.6), the log phase ran until 9 h (OD 600 from 0.6 to~7.7), and the stationary phase ran from 9~13 h (OD 600 from 7.7 to~8.8). The expression of pirA vp and pirB vp remained low until the curve entered the log phase ( Figure 1B). Expression levels reached their peaks at 5 h, then declined again at 6 h and continued to remain low ( Figure 1B). To determine the protein expression patterns, we used specific anti-PirA vp and anti-PirB vp antibodies to detect the PirA vp and PirB vp proteins at each time point. As shown in Figure 2, the PirA vp and PirB vp expression levels were low in the lag phase except for 1 h. The over-presence at this time point might be due to carry-over from 15 h culture prior to inoculation. The expression levels were high in the log phase through to the early part of the stationary phase (4~10 h), after which they returned to lower levels until the end of the recording. our results show that the pirA vp /pirB vp genes are regulated by components of the QS system, particularly by AphB vp . This is the first report to investigate the influence of the bacterial physiological system on the pirA vp /pirB vp genes. Our findings will be helpful for the development of APHND prevention and/or control strategies in the future.

The Expression Levels of pirA vp /pirB vp during Different Growth Phases of V. Parahaemolyticus
In Vibrio, since QS is involved in the regulation of many physiological processes, including several virulence-related systems, we first wanted to determine if the expression level of pirA vp /pirB vp was cell density-dependent. We, therefore, recorded the growth curve of the wild type V. parahaemolyticus, 3HP strain, and analyzed the gene expression patterns of pirA vp and pirB vp . As shown in Figure 1A, the lag phase lasted until about 3.5 h (OD600 from 0.01 to ~0.6), the log phase ran until 9 h (OD600 from 0.6 to ~7.7), and the stationary phase ran from 9~13 h (OD600 from 7.7 to ~8.8). The expression of pirA vp and pirB vp remained low until the curve entered the log phase ( Figure 1B). Expression levels reached their peaks at 5 h, then declined again at 6 h and continued to remain low ( Figure 1B). To determine the protein expression patterns, we used specific anti-PirA vp and anti-PirB vp antibodies to detect the PirA vp and PirB vp proteins at each time point. As shown in Figure 2, the PirA vp and PirB vp expression levels were low in the lag phase except for 1 h. The overpresence at this time point might be due to carry-over from 15 h culture prior to inoculation. The expression levels were high in the log phase through to the early part of the stationary phase (4~10 h), after which they returned to lower levels until the end of the recording.  collected at the indicated time points (OD600 0.01-8.82), and analyzed by real-time qPCR with specific primers for pirA vp and pirB vp , respectively. Expression levels are shown relative to those of the gyrB reference gene.

LuxO Is a Negative Regulator for the Expression of pirA vp and pirB vp
To better understand whether QS regulation is involved in the expression of pirA vp and pirB vp , isogenic mutants of opaR and luxO were derived from the AHPND-causing strain 3HP. These mutants had a deletion in opaR (ΔopaR) or luxO (ΔluxO), or in both opaR and luxO (ΔopaRΔluxO), and they were constructed by allelic gene exchange as described previously [21]. Successful deletion in the different mutants was confirmed by using opaRspecific and luxO-specific primer sets ( Figure 3). In addition, we also mutated the 47th amino acid of LuxO vp from aspartic acid (D) to glutamic acid (E) to mimic the permanently active form of LuxO vp (luxO D47E ). All mutants exhibited similar growth rates to that of the wild-type strain in LB+ medium ( Figure S1). As shown in Figure 4, the gene expression levels of pirA vp and pirB vp were down-regulated to about 60% when the active form of Lux-O vp was mimicked, while the expression levels were up-regulated by about 2 folds with the luxO vp deletion mutants (i.e., ΔluxO and ΔopaRΔluxO). By contrast, the opaR vp deletion mutant had only a small, statistically insignificant effect on the expression of pirA vp and

LuxO Is a Negative Regulator for the Expression of pirA vp and pirB vp
To better understand whether QS regulation is involved in the expression of pirA vp and pirB vp , isogenic mutants of opaR and luxO were derived from the AHPND-causing strain 3HP. These mutants had a deletion in opaR (∆opaR) or luxO (∆luxO), or in both opaR and luxO (∆opaR∆luxO), and they were constructed by allelic gene exchange as described previously [21]. Successful deletion in the different mutants was confirmed by using opaRspecific and luxO-specific primer sets ( Figure 3). In addition, we also mutated the 47th amino acid of LuxO vp from aspartic acid (D) to glutamic acid (E) to mimic the permanently active form of LuxO vp (luxO D47E ). All mutants exhibited similar growth rates to that of the wild-type strain in LB+ medium ( Figure S1). As shown in Figure 4, the gene expression levels of pirA vp and pirB vp were down-regulated to about 60% when the active form of LuxO vp was mimicked, while the expression levels were up-regulated by about 2 folds with the luxO vp deletion mutants (i.e., ∆luxO and ∆opaR∆luxO). By contrast, the opaR vp deletion mutant had only a small, statistically insignificant effect on the expression of pirA vp and pirB vp , suggesting that although opaR vp is one of the core regulators of the QS system, unlike luxO vp , it plays only a minor role in the expression of pirA vp and pirB vp . Similar effects were also seen in the protein expression levels of PirA vp and PirB vp , although we note that, despite the reduction in mRNA expression, protein levels were not reduced by the luxO D47E mimic ( Figure 4A,B, upper panels). Taken together, these results suggest that the expression of pirA vp and pirB vp is negatively regulated by luxO vp , but not by opaR vp . pirB vp , suggesting that although opaR vp is one of the core regulators of the QS system, unlike luxO vp , it plays only a minor role in the expression of pirA vp and pirB vp . Similar effects were also seen in the protein expression levels of PirA vp and PirB vp , although we note that, despite the reduction in mRNA expression, protein levels were not reduced by the luxO D47E mimic ( Figure 4A,B, upper panels). Taken together, these results suggest that the expression of pirA vp and pirB vp is negatively regulated by luxO vp , but not by opaR vp .   D47E , ∆opaR, ∆luxO, and ∆opaR∆luxO mutants. Soluble proteins were extracted from cells collected at OD600~0.6 and analyzed by immunoblotting with anti-PirA vp or anti-PirB vp antibodies (upper panels). Total RNA was extracted from the same batch of bacterial samples, and real-time RT-PCR was carried out using specific primer sets for pirA vp and pirB vp (lower panels), respectively. The housekeeping gene, gyrB served as an internal control. *: p < 0.05.

The QS Transcription Factor AphB vp Is a Possible Regulator for pirA vp and pirB vp
Since the key QS transcription factor, OpaR is evidently not involved in the regulation of pirA vp and pirB vp , we further analyzed the predicted promoter region of the pi-rA vp /pirB vp operon, and found that there was a sequence (5'-TGCATAATTTTGTGCAA-3'), which was similar to the consensus sequence of the AphB binding site, 5 Figure 5). Using real-time PCR, we found that for aphB vp , the elevated gene expression levels of the luxO vp deletion mutants ΔluxO and ΔopaRΔluxO were similar to those seen for pirA vp and pirB vp ( Figure 6B). Conversely, the expression levels of another important QS transcription factor, aphA vp , did not correspond so closely to the expression patterns of pirA vp and pirB vp ( Figure 6A). Taken pirB vp , suggesting that although opaR vp is one of the core regulators of the QS system, unlike luxO vp , it plays only a minor role in the expression of pirA vp and pirB vp . Similar effects were also seen in the protein expression levels of PirA vp and PirB vp , although we note that, despite the reduction in mRNA expression, protein levels were not reduced by the luxO D47E mimic ( Figure 4A,B, upper panels). Taken together, these results suggest that the expression of pirA vp and pirB vp is negatively regulated by luxO vp , but not by opaR vp .

The QS Transcription Factor AphB vp Is a Possible Regulator for pirA vp and pirB vp
Since the key QS transcription factor, OpaR is evidently not involved in the regulation of pirA vp and pirB vp , we further analyzed the predicted promoter region of the pi-rA vp /pirB vp operon, and found that there was a sequence (5'-TGCATAATTTTGTGCAA-3'), which was similar to the consensus sequence of the AphB binding site, 5 Figure 5). Using real-time PCR, we found that for aphB vp , the elevated gene expression levels of the luxO vp deletion mutants ΔluxO and ΔopaRΔluxO were similar to those seen for pirA vp and pirB vp ( Figure 6B). Conversely, the expression levels of another important QS transcription factor, aphA vp , did not correspond so closely to the expression patterns of pirA vp and pirB vp ( Figure 6A). Taken . The gene and protein expression levels of (A) pirA vp and (B) pirB vp in the wild type strain (3HP) compared to those in the luxO D47E , ∆opaR, ∆luxO, and ∆opaR∆luxO mutants. Soluble proteins were extracted from cells collected at OD 600~0 .6 and analyzed by immunoblotting with anti-PirA vp or anti-PirB vp antibodies (upper panels). Total RNA was extracted from the same batch of bacterial samples, and real-time RT-PCR was carried out using specific primer sets for pirA vp and pirB vp (lower panels), respectively. The housekeeping gene, gyrB served as an internal control. *: p < 0.05.

The QS Transcription Factor AphB vp Is a Possible Regulator for pirA vp and pirB vp
Since the key QS transcription factor, OpaR is evidently not involved in the regulation of pirA vp and pirB vp , we further analyzed the predicted promoter region of the pirA vp /pirB vp operon, and found that there was a sequence (5 -TGCATAATTTTGTGCAA-3 ), which was similar to the consensus sequence of the AphB binding site, Figure 5). Using real-time PCR, we found that for aphB vp , the elevated gene expression levels of the luxO vp deletion mutants ∆luxO and ∆opaR∆luxO were similar to those seen for pirA vp and pirB vp ( Figure 6B). Conversely, the expression levels of another important QS transcription factor, aphA vp , did not correspond so closely to the expression patterns of pirA vp and pirB vp ( Figure 6A). Taken together, these results suggest that AphB vp , but not AphA vp , may be important for the expression of pirA vp and pirB vp . together, these results suggest that AphB vp , but not AphA vp , may be important for the expression of pirA vp and pirB vp .   together, these results suggest that AphB vp , but not AphA vp , may be important for the expression of pirA vp and pirB vp .

His-AphB vp Binds with the Predicted Promoter Region of pirA vp /pirB vp
In order to verify whether AphB vp could bind directly to the predicted AphB vp binding site located upstream of pirA vp /pirB vp , we amplified the fragment 300 bp upstream of the pirA vp /pirB vp operon by PCR, and mixed it with the indicated concentrations (0, 0.5, 1, 2, 5, 10, 20, and 40 µM) of the recombinant His-tagged Aph proteins His-AphB vp and His-AphA vp . As shown in Figure 7B, the DNA fragments shifted upward at His-AphB vp concentrations of 10 µM and above, showing that His-AphB vp was able to bind to this DNA fragment. By contrast, no DNA band shift was seen for His-AphA vp (Figure 7A), suggesting that binding to the predicted promoter region of pirA vp /pirB vp is His-AphB vp -specific.

His-AphB vp Binds with the Predicted Promoter Region of pirA vp /pirB vp
In order to verify whether AphB vp could bind directly to the predicted AphB vp bin ing site located upstream of pirA vp /pirB vp , we amplified the fragment 300 bp upstream the pirA vp /pirB vp operon by PCR, and mixed it with the indicated concentrations (0, 0.5 2, 5, 10, 20, and 40 μM) of the recombinant His-tagged Aph proteins His-AphB vp and H AphA vp . As shown in Figure 7B, the DNA fragments shifted upward at His-AphB vp co centrations of 10 μM and above, showing that His-AphB vp was able to bind to this DN fragment. By contrast, no DNA band shift was seen for His-AphA vp (Figure 7A), sugge ing that binding to the predicted promoter region of pirA vp /pirB vp is His-AphB vp -specifi To further verify the importance of the putative AphB binding sequence in the p dicted promoter region of pirA vp /pirB vp , we truncated the DNA fragment into Fragmen (136 bp), which contained the complete AphB binding sequence, and Fragment 2 (116 b which contained only half of the AphB binding sequence ( Figure 8A). After incubati with the indicated concentrations of His-AphB vp , we observed shifting with Fragment 1 concentrations of 5~10 μM and above, whereas there were no shifts observed at any co centration with Fragment 2 ( Figure 8B). These results confirmed the importance of predicted sequence for AphB vp binding and thus for its potential role in the regulation pirA vp and pirB vp expression. To further verify the importance of the putative AphB binding sequence in the predicted promoter region of pirA vp /pirB vp , we truncated the DNA fragment into Fragment 1 (136 bp), which contained the complete AphB binding sequence, and Fragment 2 (116 bp), which contained only half of the AphB binding sequence ( Figure 8A). After incubation with the indicated concentrations of His-AphB vp , we observed shifting with Fragment 1 at concentrations of 5~10 µM and above, whereas there were no shifts observed at any concentration with Fragment 2 ( Figure 8B). These results confirmed the importance of the predicted sequence for AphB vp binding and thus for its potential role in the regulation of pirA vp and pirB vp expression.

Discussion
PirA vp and PirB vp have been confirmed as critical pathogenic factors of AHPND [12]. In a previous study, we further showed that PirA vp and PirB vp formed a heterotetramer in solution, and based on this tetramer's structural similarity to the Cry toxin, we proposed that the binary toxin may destroy host cells by means of a mechanism similar to that used by Cry [14]. In particular, we suggested that the role of the PirA vp component is to recognize the glycan of its receptor on the host cells, after which PirB vp penetrates the cell membrane to form an unregulated channel, ultimately leading to critical cell damage [14]. Other reports have further shown that both PirA vp and PirB vp can bind directly to the receptor LvAPN1, and that PirB vp was translocated to the cytoplasm and nucleus of hemocytes [22,23], suggesting that the PirA vp and PirB vp toxins might be involved in other, additional pathogenic mechanisms. Until the present study, however, the mechanisms that might regulate expression of the pirA and pirB genes themselves had remained unknown. Here, we found a relationship between cell density and high expression levels of pirA vp and pirB vp in the early log phase (Figure 1). In addition, starting at around the same time as the peak of the gene expression levels, we also observed sustained elevated levels of PirA vp and PirB vp protein throughout the entire log phase and on into the stationary phase (Figures 1 and 2).
It is already known that QS system is involved in the regulation of many virulence factors in Vibrio, including the genes of the type-III secretion systems (T3SS1 and T3SS2), type-VI secretion systems (T6SS1 and T6SS2), and the thermostable direct hemolysin genes tdh1 and tdh2 [24][25][26]. Here, by using strains with mutations in key molecules of the

Discussion
PirA vp and PirB vp have been confirmed as critical pathogenic factors of AHPND [12]. In a previous study, we further showed that PirA vp and PirB vp formed a heterotetramer in solution, and based on this tetramer's structural similarity to the Cry toxin, we proposed that the binary toxin may destroy host cells by means of a mechanism similar to that used by Cry [14]. In particular, we suggested that the role of the PirA vp component is to recognize the glycan of its receptor on the host cells, after which PirB vp penetrates the cell membrane to form an unregulated channel, ultimately leading to critical cell damage [14]. Other reports have further shown that both PirA vp and PirB vp can bind directly to the receptor LvAPN1, and that PirB vp was translocated to the cytoplasm and nucleus of hemocytes [22,23], suggesting that the PirA vp and PirB vp toxins might be involved in other, additional pathogenic mechanisms. Until the present study, however, the mechanisms that might regulate expression of the pirA and pirB genes themselves had remained unknown. Here, we found a relationship between cell density and high expression levels of pirA vp and pirB vp in the early log phase (Figure 1). In addition, starting at around the same time as the peak of the gene expression levels, we also observed sustained elevated levels of PirA vp and PirB vp protein throughout the entire log phase and on into the stationary phase (Figures 1 and 2).
It is already known that QS system is involved in the regulation of many virulence factors in Vibrio, including the genes of the type-III secretion systems (T3SS1 and T3SS2), type-VI secretion systems (T6SS1 and T6SS2), and the thermostable direct hemolysin genes tdh1 and tdh2 [24][25][26]. Here, by using strains with mutations in key molecules of the QS system, we found that the expression of pirA vp and pirB vp was also regulated by components of the QS system and that it was negatively regulated by LuxO vp (Figure 4).
Although the upstream region of the pirA vp /pirB vp operon did not include a consensus LuxO binding sequence (5 -TTGCAW 3 TGCAA-3 , where W stands for A or T; [27]), we found a possible AphB binding site as shown in Figure 5. In addition, we also found that the expression of aphB vp was affected by mutation of the QS component luxO vp (Figure 6). The upregulated expression pattern of aphB vp was similar to the pattern seen for pirA vp and pirB vp under the same conditions (Figure 4). The possible AphB binding sequence we observed (5 -TGCATAATTTTGTGCAA-3 ) diverges from other established binding motifs, such as those of tcpP (5 -TGCAAN 7 TTGCA-3 ), toxR (5 -TGCAAN 7 ATGGA-3 ), aphB (5 -TGCAAN 7 TGTCA-3 ), and a consensus sequence (5 -TGCAGN 7 TGTTG-3 ), as well as cadC I (5 -TTAAAN 7 ACTTA-3 ) and cadC II (5 -TACGTN 7 GGCTA-3 ) [20]. Nevertheless, despite this high divergence, our EMSA results showed that the binding between AphB vp and the predicted AphB binding site was both direct and specific (Figures 7 and 8). AphB vp (but not AphA vp ) thus appears to act as an enhancer of pirA vp and pirB vp . Given AphB vp 's role as a regulator of QS, we hypothesize that the QS system regulates the expression of pirA vp and pirB vp via AphB vp (Figure 9, upper left panel). We also found that pirA vp and pirB vp were down-regulated by the permanently active luxO vp mutant even though there was no significant change in aphB vp . This suggests that the regulation of the PirA vp /PirB vp genes may not be controlled only by AphB vp , but that other LuxO vp -related effectors may also be involved (Figure 9, upper center panel). The proposed regulatory mechanisms and outcomes for LuxO vp , AphB vp , and PirA vp /PirB vp in the wild type (3HP) and mutant strains (luxO D47E , ∆opaR, ∆luxO, and ∆opaR∆luxO) are shown in Figure 9. QS system, we found that the expression of pirA vp and pirB vp was also regulated by components of the QS system and that it was negatively regulated by LuxO vp (Figure 4). Although the upstream region of the pirA vp /pirB vp operon did not include a consensus LuxO binding sequence (5'-TTGCAW3TGCAA-3', where W stands for A or T; [27]), we found a possible AphB binding site as shown in Figure 5. In addition, we also found that the expression of aphB vp was affected by mutation of the QS component luxO vp (Figure 6). The upregulated expression pattern of aphB vp was similar to the pattern seen for pirA vp and pirB vp under the same conditions (Figure 4). The possible AphB binding sequence we observed (5'-TGCATAATTTTGTGCAA-3') diverges from other established binding motifs, such as those of tcpP (5'-TGCAAN7TTGCA-3'), toxR (5'-TGCAAN7ATGGA-3'), aphB (5'-TGCAAN7TGTCA-3'), and a consensus sequence (5'-TGCAGN7TGTTG-3'), as well as cadC I (5'-TTAAAN7ACTTA-3') and cadC II (5'-TACGTN7GGCTA-3') [20]. Nevertheless, despite this high divergence, our EMSA results showed that the binding between AphB vp and the predicted AphB binding site was both direct and specific (Figures 7 and 8). AphB vp (but not AphA vp ) thus appears to act as an enhancer of pirA vp and pirB vp . Given AphB vp 's role as a regulator of QS, we hypothesize that the QS system regulates the expression of pirA vp and pirB vp via AphB vp (Figure 9, upper left panel). We also found that pirA vp and pirB vp were down-regulated by the permanently active luxO vp mutant even though there was no significant change in aphB vp . This suggests that the regulation of the PirA vp /PirB vp genes may not be controlled only by AphB vp , but that other LuxO vp -related effectors may also be involved (Figure 9, upper center panel). The proposed regulatory mechanisms and outcomes for LuxO vp , AphB vp , and PirA vp /PirB vp in the wild type (3HP) and mutant strains (luxO D47E , ΔopaR, ΔluxO, and ΔopaRΔluxO) are shown in Figure 9.  6. In the first three strains (i.e., wild type strain 3HP, the activated mimic strain luxO D47E , and the OpaR deletion strain ΔopaR), the phosphorylated LuxO vp acted to limit the free expression of AphB vp and its downstream genes, pirA vp /pirB vp . Conversely, when LuxO was deleted (strains ΔluxO and ΔopaRΔluxO), AphB vp was significantly up-regulated, and this further promoted the expression of PirA vp /PirB vp . Figure 9. The relationships between LuxO vp , AphB vp and PirA vp /PirB vp at OD 600~0 .6. In the first three strains (i.e., wild type strain 3HP, the activated mimic strain luxO D47E , and the OpaR deletion strain ∆opaR), the phosphorylated LuxO vp acted to limit the free expression of AphB vp and its downstream genes, pirA vp /pirB vp . Conversely, when LuxO was deleted (strains ∆luxO and ∆opaR∆luxO), AphB vp was significantly up-regulated, and this further promoted the expression of While AphA is a well-studied QS regulator [28,29], there are relatively few studies on AphB. These studies include its regulation of virulence [19,20] and the role it plays in survival under particular conditions [30]. AphB is also a positive regulator of LuxR/OpaR activity, and it activates the expression of the exotoxin Asp [20]. These studies, together with the recent finding that anti-QS compounds may reduce AHPND pathogenicity [18], all suggest that the QS system might be critically important for regulating the virulence of AHPND-causing bacteria.
To this body of evidence, we now add the results of the present study, which suggests that the QS system might be modulating virulence by regulating expression of the pirA vp /pirB vp genes through AphB vp . This new insight into the pathogenic mechanisms of AHPND points toward the QS system as a possible target for therapeutics that might one day be able to control the virulence of AHPND-causing bacteria and prevent AHPND.

Growth Curve Measurement and Sample Collection
All V. parahaemolyticus strains (3HP, LuxO D47E , ∆opaR, ∆luxO, and ∆opaR ∆luxO) were activated by culturing on LB+ agar plates (that is, LB agar plates that contained 2% NaCl) at 30 • C for 16 h. From these plates, 3 single colonies were transferred into 5 mL LB+ medium (LB medium that contained 2% NaCl) and incubated at 30 • C for 3 h to OD 600~1 .0. The culture was then diluted 100-fold, transferred into 50 mL fresh LB+ medium and cultured at 30 • C for 15 h with shaking at 200 rpm. Dilutions (1:1000) of these overnight cultures were further sub-cultured into 500 mL LB+ medium in a 2L flask, incubated at 30 • C with shaking at 200 rpm, and the OD 600 was measured every hour for a total of 13 h. At the same time, 3HP cells were collected from part of the culture every hour, and temporarily stored at −80 • C prior to subsequent RNA and protein extractions. Other batches of sample cells of 3HP, LuxO D47E , ∆opaR, ∆luxO, and ∆opaR ∆luxO were collected at low cell density (LCD; OD 600~0 .6) and temporarily stored at −80 • C for later use.

RNA Extraction and Real-Time PCR
Total bacterial RNA was extracted using RareRNA reagent (Blossom Biotechnologies, Inc.; Taipei, ROC). DNaseI (Invitrogen) was used to digest any DNA contamination. From 1 µg of total RNA, cDNA was synthesized by using M-MLV reverse transcriptase (Promega; Madison, WI, USA) with random hexamers. Using specific primers (Table 1), real-time PCR was carried out to quantify the expression levels of target genes (pirA vp , pirB vp , aphA vp , and aphB vp ) and an internal control gene (gryB vp , a housekeeping gene). The reaction mixture contained 1 µL cDNA, 0.2 µM forward primer, 0.2 µM reverse primer, 1× ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech Co. Ltd.; Nanjing, China) in a total volume of 20 µL, and thermal cycling was performed using a LightCycler ® 96 System (Roche; Basel, Schweiz) as follows: 95 • C for 2 min followed by 40 cycles of 95 • C for 10 s and 60 • C for 30 s, and an extension cycle of 95 • C for 10 s, 65 • C for 60 s and 97 • C for 60 s.
To calculate ∆C t , the threshold cycle values of the internal control gene (C t gyrB ) were first subtracted from the threshold cycle values of the target genes. Next, either the initial ∆C t for the "1-h" timepoint was subtracted from the ∆C t s of the other timepoints (i.e., 2-13 h), or else the ∆C t for the wild-type 3HP strain was subtracted from the ∆C t of all the groups (3HP, luxO D47E , ∆opaR, ∆luxO, and ∆opaR ∆luxO) to obtain ∆∆Ct values. The fold change of the gene expression levels was then expressed as 2 −∆∆Ct [31], and the data presented as mean ± SD. Statistically significant differences were tested using an unpaired Student's t-test (p < 0.05).

Protein Extraction and Western Blots
Two batches of bacterial samples were removed from storage at −80 • C and lysed immediately using B-PER™ Bacterial Protein Extraction Reagent (ThermoFisher Scientific; Waltham, MA, USA) containing 1 mM PMSF, 1 mM EDTA, and 120 µg/mL DNaseI. The suspensions were inverted at room temperature for 10 min, and the cell debris was removed by centrifuging at 13,000× g for 10 min. Two micrograms of cell lysates were separated by 12.5% SDS-PAGE, transferred onto a PVDF membrane, and blocked with 5% skim milk at 4 • C overnight. The blots were then hybridized with chicken anti-PirA vp or chicken anti-PirB vp polyclonal antibodies (1:5000 diluted with 5% skim milk). After 1 hour of incubation at room temperature, the blots were washed 3 times with PBST solution (1× PBS contained 0.1% Tween-20). The blots were further incubated with donkey anti-chicken-HRP conjugated secondary antibody (Jackson; West Grove, PA; 1:10,000 diluted with 5% skim milk) at room temperature for 1 h. Following 3 more washes, the protein bands were visualized using a chemiluminescence reagent (GE Healthcare; Chicago, IL, USA) and detected with an Amersham Imager 600 (GE Healthcare; Chicago, IL, USA). For loading controls, 8 µg of cell lysates were separated with another 12.5% SDS-PAGE, and stained with Coomassie blue.

Construction of the ∆opaR, ∆luxO, ∆opaR∆luxO, and luxO D47E Mutants
The in-frame ∆opaR, ∆luxO, and ∆opaR ∆luxO mutants were constructed by in vivo allelic exchange as described previously [21]. Briefly, DNA fragments from the down-and up-stream regions of opaR and luxO were amplified, respectively, with the primer sets opaR-1/opaR-2 and opaR-3/opaR-4, and luxO-1/luxO-2 and luxO-3/luxO-4 (Table 1). These fragments were cloned into pGEM-T ® Easy vector (Promega; Madison, WI, USA) in the correct orientation to generate a recombinant fragment containing a 627 bp-and a 1323 bp-deletion in opaR and luxO, respectively. The DNA fragments were removed from the pGEMT ® -easy vector by enzyme digestion with SacI and SalI, respectively, and then cloned into the suicide vector pDS132. The suicide plasmids containing either the ∆opaR or ∆luxO fragment were transformed into Escherichia coli S17-1λpir [21], and then transferred into V. parahaemolyticus 3HP by conjugation to facilitate allelic exchange to produce the mutants. The double mutant was obtained by introducing the ∆luxO fragment into an already-constructed ∆opaR mutant by the method described above. For the luxO D47E mutant, a 2340-bp fragment amplified by the primers luxO-1 and luxO-16 was given a single nucleotide mutation (GAT to GAG) using a QuikChange ® Site-Directed Mutagenesis Kit (Stratagene; La Jolla, CA, USA). This luxO D47E -containing fragment was introduced into V. parahaemolyticus 3HP by allelic exchange to generate the mutant.

Confirmation of the ∆opaR, luxO, ∆opaR∆luxO and luxO D47E Mutants by PCR
Genomic DNA was extracted from 3HP and the mutants using a Genomic DNA Extraction Kit (Bioman; New Taipei City, ROC) according to the manufacturer's instructions. The opaR and luxO genes from all of the isolated mutants were then checked by PCR with the opaRand luxO-specific primers opaR-5/opaR-6 and luxO-5/luxO-6 ( Table 1), respectively. The DNA sequences were also determined to confirm that the deletion was in-frame, or, in the case of luxO D47E , that the single nucleotide mutation was correct.

Expression and Purification of Recombinant His-AphA and His-AphB
To express the recombinant AphA vp and AphB vp , the aphA vp -pET28a and aphB vp -pET28a plasmids were, respectively, transformed into E. coli strain BL21 (DE3) cells. For AphA vp , the transformed cells were inoculated into 50 mL of fresh LB medium and grown at 37 • C for 12-14 h. Three ml of this overnight culture was then added to 500 mL of fresh LB medium in a 2L flask, and grown at 37 • C until the OD 600 of the culture reached 0.4. IPTG was then added to a final concentration of 0.4 mM, and the culture was incubated at 16 • C for 20 h. The cells were then collected, resuspended in binding buffer (20 mM Tris-base, 500 mM NaCl, 20 mM imidazole, pH 8.0) containing 1 mM PMSF, 100 µg/mL lysozyme and 10 µg/mL DNase I, and homogenized by sonication on ice. After the cell debris was removed by centrifugation, the supernatant was filtrated using a 0.45 µm filter and loaded onto a 5 mL HisTrap HP column (GE Healthcare; Chicago, IL, USA). The column was washed with 100 mL of binding buffer and then eluted with a 20-500 mM imidazole gradient. The eluted recombinant protein was concentrated and loaded onto a Superdex 75 gel filtration column (GE Healthcare; Chicago, IL, USA) using 20 mM Tris-base, 500 mM NaCl, pH 8.0 as a running buffer. The protein concentration was measured by the Bradford method. For AphB vp , all of the culture and purification processes were the same as for AphA vp , except that for the subculture step, 10 mL of the overnight culture was inoculated into the 500 mL of fresh LB+ medium.