Sigma E Regulators Control Hemolytic Activity and Virulence in a Shrimp Pathogenic Vibrio harveyi

Members of the genus Vibrio are important marine and aquaculture pathogens. Hemolytic activity has been identified as a virulence factor in many pathogenic vibrios including V. cholerae, V. parahaemolyticus, V. alginolyticus, V. harveyi and V. vulnificus. We have used transposon mutagenesis to identify genes involved in the hemolytic activity of shrimp-pathogenic V. harveyi strain PSU3316. Out of 1,764 mutants screened, five mutants showed reduced hemolytic activity on sheep blood agar and exhibited virulence attenuation in shrimp (Litopenaeus vannamei). Mutants were identified by comparing transposon junction sequences to a draft of assembly of the PSU3316 genome. Surprisingly none of the disrupted open reading frames or gene neighborhoods contained genes annotated as hemolysins. The gene encoding RseB, a negative regulator of the sigma factor (σE), was interrupted in 2 out of 5 transposon mutants, in addition, the transcription factor CytR, a threonine synthetase, and an efflux-associated cytoplasmic protein were also identified. Knockout mutations introduced into the rpoE operon at the rseB gene exhibited low hemolytic activity in sheep blood agar, and were 3-to 7-fold attenuated for colonization in shrimp. Comparison of whole cell extracted proteins in the rseB mutant (PSU4030) to the wild-type by 2-D gel electrophoresis revealed 6 differentially expressed proteins, including two down-regulated porins (OmpC-like and OmpN) and an upregulated protease (DegQ) which have been associated with σE in other organisms. Our study is the first report linking hemolytic activity to the σE regulators in pathogenic Vibrio species and suggests expression of this virulence-linked phenotype is governed by multiple regulatory pathways within the V. harveyi.


Introduction
Vibrio harveyi is one of several closely-related species of Vibrio that cause disease in marine organisms [1][2][3]. Outbreaks of highly virulent Vibrio strains have caused major losses to shrimp farmers in Thailand and elsewhere [4][5][6][7][8]. The mechanisms of pathogenesis in these vibrios are not clearly understood and are likely mediated in part by strain-specific virulence factors. Hemolytic activity has been linked to virulence in many species of Vibrio where several different classes of hemolysins have been described. The V. harveyi hemolysin (VHH) is a member of the broadly distributed thermolabile hemolysin (TLH) family [9][10][11][12] and appears to be sufficient for hemolytic activity in some, but not all, strains of V. harveyi [12,13]. We have previously determined that the vhh gene is present in all V. harveyi isolates from both healthy and diseased marine animals collected in Southern Thailand [14]. However, hemolytic activity on blood agar was variable in the vhh bearing isolates, suggesting that hemolytic activity is influenced by additional factors.
In order to better understand mediators of hemolytic activity and virulence in V. harveyi we have investigated genes that control hemolytic activity in the shrimp-virulent V. harveyi strain PSU3316.
Results of an initial transposon screen led us to hypothesize that regulators of the cell envelope stress sigma factor RpoE control the elaboration of V. harveyi hemolytic activity and that this phenotype correlates with virulence in shrimp. Recent work in Vibrio vulnificus has revealed that the regulatory network of RpoE controls virulence in a mouse model [15]. To investigate whether the RpoE-operon regulatory proteins mediate the virulence of V. harveyi we have carried out targeted mutagenesis, in vitro and in vivo activity assays, and proteomic analysis to identify genes differentially expressed in wild-type and non-hemolytic mutants. Surprisingly, none of the differentially regulated proteins or gene products of loci inactivated in hemolysis-attenuated transposon mutants were similar to known hemolysins. These results suggest that hemolysis activity per se may be a phenotype that is influenced by a variety of different factors including the RpoE-operon mediated cell envelope stress response of V. harveyi.

Bacterial strains and plasmids
All bacterial strains and plasmids used in this study are listed in Table 1. V. harveyi PSU3316 was isolated in 2004 from the hemolymph of a diseased shrimp (Penaeus monodon) in Southern Thailand, and an early passage of this strain was archived in glycerol at 280uC. PSU3316 was identified as being most similar to type strains of V. harveyi by biochemical testing, genome sequencing, and phylogenetic analysis of the gyrB gene [16]. Although the systematics of the V. harveyi-group is subject to debate based on emerging genomic data we have chosen to retain the original designation pending official taxonomic revision of type strains. PSU3545 is a spontaneous streptomycin-resistant (Sm R ) derivative of PSU3316. Confirmation of virulence of V. harveyi PSU3316 and PSU3545 was performed by determining the LD 50 of each strain in shrimp (P. monodon) and through a competition assay involving co-infection of these strains and measurement of hepatopancreas colonization levels. All strains were maintained in 20% glycerol at 280uC.

Sequencing and draft assembly of the PSU3316 genome
Libraries suitable for sequencing using the Illumina Genome Analyzer (Illumina, Inc.) were generated using a modified version of the standard Illumina GA protocol. 5 mg of genomic DNA from strain PSU3316 was sheared using Adaptive Focused Acoustic technology (Covaris, Inc.) to generate fragments 100-300 bp in length. Fragments were blunt-ended, A-tailed and ligated with T nucleotide overhang Illumina forked paired end-sequencing adapters (Illumina, Inc.) containing custom bar-codes for multiplex sequencing. Libraries were then PCR amplified for 16 cycles after identifying the optimum number of cycles using qPCR, sequenced to a depth of ,66 and assembled into contigs using CLC Genomics Workbench 4 (Aarhus, Denmark). Genome fragments were uploaded to the Rapid Annotations using Subsystems Technology (RAST) server for annotation [17]. Genome regions containing open reading frames for genes and proteins identified in this study have been deposited at NCBI with accessions XXX-YYY (to be provided with final submission).

Transposon mutagenesis of V. harveyi
The mariner-based transposon pSC189 (Km R ) [18] containing a kanamycin resistance gene was modified by insertion of a chloramphenicol marker into the PstI restriction site to obtain pJT064. This plasmid carries a transposon fragment which was designated as TnJT064. pJT064 was transformed into competent E. coli BW20767 [19]. Conjugation of E. coli donor strain and V. harveyi PSU3316 was performed by mixing each strain at ratio of 1:1 on LB plates supplemented with 1% NaCl and incubation at 30uC for 6 h followed by selection of V. harveyi mutants carrying TnJT064 insertions on Vibrio-selective Thiosulfate citrate bile salt-sucrose agar (TCBS) containing chloramphenicol (2 mg mL 21 ).

Screening transposon insertion mutants for hemolytic activity and shrimp virulence
TnJT064 transposon insertion mutants were screened for hemolytic activity by spotting cultures on sheep blood agar (PML Microbiologicals, USA) and then scoring for the appearance of a lytic zone after incubation at 30uC for 48 h. Wild type V. harveyi PSU3316 and its streptomycin resistant derivative PSU3345 are b-hemolytic in this assay and any mutant that caused incomplete hemolysis (defined here as a-hemolysis) were subjected to a second screen for virulence in shrimp. Cells grown overnight in TSB were harvested by centrifugation, washed, and suspended in a sterile saline solution. Shrimp were challenged by intramuscular injection of a 100 ml of saline suspensions of approximately 2.4610 6 CFU (corresponding to four times LD 50 for wild-type PSU3316), of either selected transposon mutants, PSU3316, PSU3545, or sterile saline (mock infection control). The challenge experiment was repeated twice, first with three shrimp per challenge followed by a second experiment with seven shrimp per challenge. Differences in mortality after 18 h in wild-type strains (PSU3316 and PSU3545) and mutant strains with attenuated hemolysis were evaluated using the Fisher's exact test. Relative risk (RR) of mortality compared with the wild type-strain was also determined for each mutant. Characterization of genes involved in hemolytic activity The identity of genes disrupted in hemolysis-and virulenceattenuated mutants was determined by PCR amplification and sequencing of the transposon insertion junctions using primers targeting outward facing priming sites in the TnJT064 transposon (Mar2018 and MarEC6) and arbitrary primers ARB6/7 and ARB2 which bind throughout the genome [20]. Nucleotide sequences from transposon insertion junctions were compared to the PSU3316 genome by BLAST and identified based on annotation of the corresponding open reading frames disrupted by the transposon insertion.

rseBC disruption
A spontaneous streptomycin resistant V. harveyi PSU3316 (designated as PSU3545) was used as the parental strain in the construction of two independent insertions in the rseB gene by homologous recombination-based gene disruption [21] which is expected to also inactivate the downstream gene rseC. PCR amplification of inserts for gene disruption was performed using specific primers designated in this study ( Table 2). The PCR products were cloned into pJT084 (a derivative of suicide plasmid pDTR801 which carries the R6K origin of replication , the RP4 mob region, the sacB gene for sucrose-based counter selection from pCVD422 [22] modified by removal of the bla gene encoding ampicillin resistance by partial PstI digestion (NEB) and addition of a chloramphenicol acetyltransferase (cat) gene for chloramphenicol resistance), transformed into E. coli BW20767/lpir and conjugated with PSU3545. V. harveyi mutants containing gene disruptions were selected on TCBS containing chloramphenicol (2 mg mL 21 ) and streptomycin (200 mg mL 21 ).

Quantification of hemolytic activity
Hemolytic activity expressed by rseBC mutant and wild-type strains on solid media was visualized using a standard sheep blood agar assay [23,24]. Cell-associated hemolytic activity was quantified as the percentage of red blood cells lysed in liquid suspension assay. In brief, bacterial strains were grown 18 h at 30uC in LB agar containing 1.5% NaCl.
Bacterial cells and sheep erythrocytes were washed twice with sterile artificial seawater or Artificial SeaWater (Mariscience) prior to the assay. Bacterial cell suspensions were adjusted to 1.5610 8 CFU mL 21 by optical density at 600 nm and 190 ml of each strain was mixed with 10 ml of packed sheep erythrocytes in 96-well microtiter plate and incubated at 30uC for 5 h. Supernatants were obtained by centrifugation (40006 g; 10 min; 4uC) and the amount of hemoglobin released from the lysed erythrocytes was determined by spectrophotometer at 540 nm. Isotonic artificial seawater (9 ppt) was used as a non-hemolytic negative control and complete hemolysis was induced through osmotic shock with deionized water. The percentage of hemolysis was calculated using the following formula: % Hemolysis = 1006(OD 540 of sample -OD 540 of negative control)/(OD 540 of positive control -OD 540 of negative control). Differences in hemolytic activity in wild-type and mutant strains were evaluated using the student's t-test SPSS version 14 software (SPSS Inc., Chicago, Illinois).

Growth and competition assays during shrimp infections
Growth rates of V. harveyi rseBC mutants (PSU4030 and PSU4031) were compared to the wild-type strain PSU3545. Strains were grown in LB broth containing 1.5% NaCl for 16 h at 30uC and growth was quantified by optical density at 600 nm every 60 min. Colony morphologies were observed on sheep blood agar. The competitive index (CI) of mutant strains was determined in vivo using juvenile L. vannamei shrimp (10 to 13 g with a length of 4 to 5 inches). Shrimp were obtained from a farm in Pattani province, Thailand, and were maintained in a 70 L glass tank containing artificial seawater (salinity 17 ppt) at a temperature of 2961uC for at least 7 days before testing. For competition assays, cell suspensions of wild-type PSU3545 (Cm S Sm R ) and a mutant strain (Cm R Sm R ) were mixed at a ratio of 1:1 in sterile saline and then approximately 1610 6 CFU were injected intramuscularly into each shrimp. Each competition was performed in duplicate using 7 shrimp per group. Approximately 18 h post infection, the hepatopancreas of each infected shrimp was removed, homogenized, and the proportion of V. harveyi parental and mutant strains determined by plating on TCBS supplemented with streptomycin. After colonies were recovered several hundred were picked on to TCBS supplemented with chloramphenicol and streptomycin to determine the ratio of PSU3545 (Cm S Sm R ) to either PSU4030 and PSU4031 (Cm R Sm R ).
The CI was calculated as follows: CI = (number of mutant CFU/number of wild type CFU isolated from hepatopancreas)/ (number of mutant CFU/number of wild type CFU in injected cell suspension). A CI of less than 1 indicates the ability of the mutant strain to infect shrimp is lower than the wild type. Statistical analysis of results obtained from competition assays were evaluated by the Mann-Whitney U test [25,26].

Characterization of whole cell proteins from rseBC mutant
Protein extracts of rseBC mutant (PSU4030) and wild-type V. harveyi (PSU3545) were investigated to identify differentially expressed proteins that may mediate hemolytic activity. Strains were grown on sheep blood agar at 30uC for 24 h. Whole-cell protein extracts were prepared by suspending bacterial cells in 1 mL phosphate-buffered saline (PBS). Cells were washed three times with PBS. The cell pellets were then resuspended in lysis buffer (7 M Urea, 2 M Thiourea, 4% CHAPS, 2% Pharmalyte 4-7, 40 mM DTT) and incubated at 4uC for 6 h. The supernatant was collected by centrifugation at 14,0006 g for 5 min at 4uC and was kept at 270uC until required.
2-D gel electrophoresis was performed according to the manufacturer's protocol (GE Healthcare). Briefly, tested samples were cleaned by 2D clean-Up Kits (GE Healthcare) and the concentration of protein was determined using PlusOne 2-D Quant Kit (GE Healthcare). The test sample was then applied to Immobiline Drystrip (IPG gel strip) and was transferred to

Results
Transposon mutagenesis of V. harveyi and screen for hemolysis and virulence Using the mariner-based transposon mutagenesis, 1,764 V. harveyi PSU3316 mutants were generated and screened for a loss of hemolytic activity on sheep blood agar. Five mutants displayed low hemolytic activity (Table 3) and were subsequently screened for virulence in shrimp at a dose of four times the LD 50 of the wildtype strain (2.4610 6 CFU shrimp 21 ). This dose of wild-type V. harveyi PSU3316 (or PSU3545, see below) consistently resulted in 100% shrimp mortality. Hemolysis attenuated mutants exhibited lower rates of mortality (Fisher's exact test p,0.05) where the relative risk of mortality of mutant strains was 2.5 to 5-fold less than wild-type. Other randomly chosen transposon mutants without disruption of the hemolysis phenotype were not attenuated for virulence (data not shown), indicating that the acquisition of the transposon per se is not associated with reduced shrimp virulence. Shrimp infected with V. harveyi exhibited characteristic disease symptoms including lethargy, and upon dissection, their hepatopancreas appeared discolored.
The DNA sequences corresponding to the junction of transposon insertions were determined for the attenuated V. harveyi mutants. Annotation of sequences adjacent to the transposon insertion sites revealed that 2 hemolysis-attenuated mutants (designated as FH3 and GG7) carried insertions that disrupted the gene encoding the s E factor negative regulatory protein (RseB) ( Table 3 & Figure 1A). The other mutants carried insertions that interrupted genes encoding putative proteins annotated as a transcriptional repressor cytR, a threonine synthase, and a hypothetical protein in the kefGB operon associated with glutathione-regulated potassium-efflux (Table 3 & Figure 1B, C, D). These genes were also present in the genomes of closely-related strains (i.e. V. harveyi BAA-1116, V. harveyi HY01, and V. parahaemolyticus RIMD 2210633). Analysis of neighboring genes on contigs assembled from the V. harveyi PSU3316 genome and in the closely related V. harveyi genomes (BAA-1116 and HY01) did not reveal any known hemolysins that could be inactivated by polar effects of the transposon insertions characterized above.

Involvement of rpoE regulators in the hemolytic activity of V. harveyi
Since inactivation of the rpoE regulatory protein RseB was linked to reduced hemolytic activity, we hypothesized that this gene or the product of the downstream gene rseC may mediate hemolysin gene expression or activity. PSU3316 harbors a canonical rpoE operon structure comprised of four genes (rpoE-rseA-rseB-rseC) which are co-transcribed as a poly-cistronic operon in closely related model systems [27] ( Figure 1A). Ideally, construction of in-frame mutations in all four of these genes would produce the appropriate a panel of mutations that might reveal the relationship between the rpoE-rseA-rseB-rseC operon and hemolytic activity in V. harveyi PSU3316. However, after repeated attempts to inactivate genes in this region and to create in-frame deletions using counter-selection, we were only able to construct two independent plasmid insertion mutations in rseB in V. harveyi PSU3545 using homologous recombination mediated by internal fragments of the rseB gene. Disruption of rseB is expected to also inactivate the downstream rseC gene. The two defined rseBC mutant strains (PSU4030 and PSU4031) displayed wild-type rates of growth in LB broth ( Figure 2) and a defective hemolytic phenotype on sheep blood agar (Figure 3). Hemolytic activity of cell suspensions derived from these mutants strains grown on LB agar containing 1.5% NaCl was measured using a quantitative assay for hemoglobin release from sheep erythrocytes (see Materials and Methods). Compared to the parental strain PSU3545, hemolytic activity in the V. harveyi rseBC mutants   (PSU4030 and PSU4031) was 20% lower but this reduction reached statistical significance (p,0.05) (Figure 4).

In vivo competition assays
The virulence of V. harveyi rseBC mutants (PSU4030 and PSU4031) was evaluated by quantification of shrimp colonization relative to wild-type V. harveyi (PSU3545). The competitive indexes (CI) for rseBC mutants PSU4030 and PSU4031 against the wild type were determined to be 0.15 and 0.31, respectively and both were significantly different from 1.0 (p,0.05) ( Figure 5) indicating three to seven-fold reduction in the ability of either rseBC mutant to colonize shrimp relative to the wildtype strain.
Comparative analysis and identification of proteins using 2D gel electrophoresis Two-dimensional protein gel electrophoresis revealed 6 proteins that were differentially expressed in the rseBC mutant (PSU4030) compared to the wild-type V. harveyi (PSU3545). These proteins were analyzed by nano LC-ESI-MS/MS and five proteins could be identified with significant identity (p,0.05) to proteins in the NCBInr database. In the non-hemolytic V. harveyi rseBC mutant, three proteins were significantly under-expressed ( Figure 6-A, spots 1, 2, 3) corresponding to two outer membrane porins (OmpC-like and OmpN) and an unidentified protein (Figure 7-A,  B; Table 4). Expression of three proteins was elevated in the rseBC mutant relative to the wild type ( Figure 6-B, spots 4, 5, 6) corresponding to a phosphosugar mutase, S-(hydroxymethyl) glutathione dehydrogenase, and the protease (DegQ) (Figure 7 C, D, E & Table 4).

Discussion
In the present study transposon and targeted insertion mutagenesis of V. harveyi PSU3316 revealed that disruption of regulatory elements in the rpoE operon modulate hemolytic activity and the ability of this pathogenic organism to colonize shrimp. The ability to lyse blood cells is an important virulence factor for V. harveyi [12] and other microbial pathogens [28] however modulation of hemolysis by the activity of RpoE operon regulatory proteins was previously unknown. In E. coli the s E operon contains the gene for the sigma(E) factor (rpoE) as well as the regulator elements rseA, rseB and rseC which are co-transcribed by a promoter upstream of the rpoE gene [29]. RseA and RseB are transmembrane and periplasmic negative regulatory proteins while RseC is a positive regulatory protein [30,31]. Genome sequencing and draft assembly revealed an identical rpoE-rseABC operon structure in V. harveyi PSU3316 ( Figure 1A). RpoE has been implicated in virulence as well as adaptive responses for survival in natural habitats [30,[32][33][34][35] and in Gram negative bacteria, stressors trigger accumulation of unfolded-or misfoldedproteins in the periplasm, and this in turn activates s E through RseA and RseB signal transduction [36,37]. In E. coli the transcription factor encoded by rpoE (s E ) becomes more active when rseB is disrupted leading to increased transcription of s E  dependent genes [27]. In V. vulnificus rseB mutants are attenuated for virulence in a mouse model through putative modulation of rpoE activity [15]. If regulation of the rpoE operon in V. harveyi is the same as in E. coli and as proposed for V. vulnificus, then the attenuated virulence phenotypes associated with the V. harveyi rseBC mutants may be due to changes in activity of RpoE due to relief of regulatory control by RseB or RseC. Studies in E. coli suggest deletion of the negative regulator rseB induces RpoE activity while deletion of the positive regulator rseC has a negligible effect [27,30]. In E. coli the double mutant (rseBC) either has a net inducing effect on RpoE activity similar to a rseB deletion mutant (in the case of a transposon insertion in rseB with expected polar effects on rseC) [27] or revealed less induction of RpoE activity than the single rseB deletion mutant (in the case of a double deletion mutant in both rseB and rseC) suggesting competing effects of negative and positive regulation by RseB and RseC respectively on RpoE activity [30] In V. harveyi, if the activity of RseB is dominant to that of RseC then elevated RpoE activity in the rseB gene-disruption mutants would be responsible for alterations in protein expression that are negatively correlated with hemolysis and virulence. Alternatively, our transposon and plasmid insertions in rseB may have had a polar effect on expression of the downstream RpoE positive regulatory gene rseC. In the latter case, inactivation of either rpoE or rseC would be predicted to produce the same phenotype as such postulated polar rseB insertion mutations. Unfortunately we were not successful in making mutations in the rpoE region other than in rseB and it is possible that rpoE is an essential gene in V. harveyi. In addition, we were unable to select for in-frame deletions of rseB using sucrose-or streptomycin-based counter selection via the sacB or rpsL genes, respectively, due to the limited genetic tractability of this shrimp pathogenic V. harveyi strain. Additional genetic analysis (including the construction of double mutants in rseB together with rseC or rpoE and complementation with each gene) will be needed to resolve these two possible models for the phenotype associated with rseB insertion mutations. To shed light on how disruption of the RpoE operon at the rseB gene controls virulence, we used proteome analysis to determine whether any protein level changes could be distinguished between non-hemolytic (rseBC mutant) and wild-type V. harveyi grown on sheep blood agar. We found that two outer membrane porins (OmpN-and OmpC-like) and an additional unidentified protein were decreased in the rseBC mutant relative to the wild-type (Table 4). This observation is consistent with down-regulation of OmpN during enhanced expression of s E in E. coli [38]. Several porins are established components of the RpoE regulon, controlling transport of solutes through the outer membrane and modulating the periplasmic environment [39]. Our data show that reduced porin expression and reduced hemolytic activity are both linked to disruption of the RpoE operon at the rseB gene in V. harveyi and characterization of these proteins and their potential involvement in RpoE-regulated virulence of V. harveyi will be a priority for future work in this system.
We also observed that three proteins were over-expressed in the rseBC mutant relative to the wild type. These proteins are a phosphosugar mutase, a S-(hydroxymethyl)glutathione dehydrogenase, and the DegQ protease which is homologous, and shares overlapping function with DegP [40,41], a positively regulated member of the rpoE regulon in E. coli [42]. RpoE -directed regulation of DegQ in fish pathogenic V. harveyi has been suggested recently [41] and our data supports this model. It should be noted that differentially expressed RpoE, RseA, RseB, and RseC were not detected during proteomic analysis because the theoretical molecular weight of RpoE, RseA and RseC proteins were less than 30 kDa (21.7, 23.5 kDa, and 16.6 kDa respectively) and the theoretical pI of RseB was more than 7 (pI = 8.6) which were out of molecular weight and pI ranges investigated in this study.
Transposon mutagenesis, confirmed by targeted gene disruption, has allowed us to establish the first link between regulation of the RpoE-operon, hemolytic activity, and virulence in V. harveyi. This would not be possible in the closely-related and more genetically tractable V. harveyi strain BAA-1116 because it is weakly hemolytic and avirulent in the shrimp model used in this study. Thus, although limited by the availability of optimized genetic methods including in-frame deletion construction and complementation, analysis of non-laboratory adapted pathogens can provide insights into the regulation of virulence that may not be accessible in more highly characterized and optimized laboratory model systems. We have sequenced the genome of PSU3316 in anticipation of future optimized genetics.
In conclusion, we have demonstrated that disruption of RpoE operon at the rseB gene in a shrimp-pathogenic V. harveyi strain leads to reduction of hemolytic activity, shrimp colonization ability, and virulence. These changes are concomitant with changes in protein expression including decreased expression of two porins (OmpC-like and N) and increased expression of the stress-responsive periplasmic protease DegQ. If potential polar effects on rseC are negligible, as has been suggested in E. coli [27] our results suggest a model where enhanced activity of rpoE upon loss of rseB activity triggers periplasmic stress-responses that lead to diminished virulence by altering periplasmic homeostasis that controls some aspect of functional hemolytic activity. This study represents the first report of the role of the rpoE operon in the hemolytic activity of V. harveyi and paves the way for further work to determine the molecular mechanisms for hemolytic activity and virulence in this important group of marine pathogens.