A distinctive dual-channel quorum-sensing system operates in Vibrio anguillarum: Vibrio anguillarum quorum sensing

Molecular Microbiology (2004)

52

(6), 1677–1689 doi:10.1111/j.1365-2958.2004.04083.x

© 2004 Blackwell Publishing Ltd

Blackwell Science, LtdOxford, UKMMIMolecular Microbiology0950-382XBlackwell Publishing Ltd, 2004

? 2004

52

616771689

Original Article

Vibrio anguillarum quorum sensingA. Croxatto

et al

.

Accepted 13 February, 2004. *For correspondence. [email protected]; Tel. (

+

46) 90 785 6755; Fax (

+

46) 90771 420.

A distinctive dual-channel quorum-sensing system operates in

Vibrio anguillarum

Antony Croxatto,

1

John Pride,

2

Andrea Hardman,

2

Paul Williams,

2,3

Miguel Cámara

2,3

and Debra L. Milton

1

*

1

Department of Molecular Biology, Umeå University, S-901 87 Umeå, Sweden.

2

School of Pharmacy and

3

Institute of Infection, Immunity and Inflammation, University of Nottingham, Nottingham NG7 2RD, UK.

Summary

Many bacterial cells communicate using diffusiblesignal molecules to monitor cell population densityvia a process termed quorum sensing. In marine

Vibrio

species, the

Vibrio harveyi

-type LuxR proteinis a key player in a quorum-sensing phosphorelaycascade, which controls the expression of virulence,symbiotic and survival genes. Previously, we charac-terized

Vibrio anguillarum

homologues of LuxR(VanT) and LuxMN (VanMN) and, in this study, we haveidentified homologues of LuxPQ (VanPQ) and LuxOU(VanOU). In contrast to other

Vibrio

species,

vanT

wasexpressed at low cell density and showed no signi-ficant induction as the cell number increased. Inaddition, although the loss of VanO increased

vanT

expression, the loss of VanU, unexpectedly,decreased it. Both VanN and VanQ were required forrepression of

vanT

even in a

vanU

mutant, suggestingan alternative route for VanNQ signal transductionother than via VanU. VanT negatively regulated its ownexpression by binding and repressing the

vanT

promoter and by binding and activating the

vanOU

promoter. The signal relay results in a cellularresponse as expression of the metalloprotease,

empA

, was altered similar to that of

vanT

in all themutants. Consequently, the

V. anguillarum

quorum-sensing phosphorelay systems work differently fromthose of

V. harveyi

and may be used to limit ratherthan induce

vanT

expression.

Introduction

Vibrio anguillarum

causes a terminal haemorrhagic

septicaemia in marine fish and has been associated withhigh mortality within aquaculture. Little is known abouthow this bacterium survives in the environment or how itcauses disease in fish (for reviews, see Actis

et al

., 1999;Austin and Austin, 1999).

V. anguillarum

constitutes partof the normal microflora of the aquatic environment(Oppenheimer, 1962; West

et al

., 1983; Muroga

et al

.,1986) and can survive for prolonged periods (

>

50 months)in a sea-water microcosm (Hoff, 1989).

Diverse bacteria use a type of cell–cell signalling,termed quorum sensing, to monitor their populationdensity. Small diffusible molecules, such as

N

-acylhomoserine lactones (AHLs) in Gram-negative bacte-ria, are secreted by one individual and sensed by a sec-ond individual of the same species resulting in a specificaction (for reviews, see Dunny and Winans, 1999;Lazazzera, 2000; Miller and Bassler, 2001; Withers

et al

.,2001). Thus, bacteria can co-ordinate activities as a pop-ulation instead of as single cells. A unified action is advan-tageous for survival of bacteria, as they often alter theirmorphology and physiology quickly to adapt to environ-mental changes that may be harsh, such as movingbetween fish tissue and sea water.

In

V. anguillarum

, the transcriptional regulator, VanT,was shown to regulate positively extracellular proteaseactivity, pigment production and biofilm formation(Croxatto

et al

., 2002). Each of these activities may playa role in the survival of

V. anguillarum

in sea water or thefish host. VanT is a homologue of the

Vibrio harveyi

LuxRtranscriptional activator for bioluminescence, which isregulated in a cell density-dependent manner through theproduction and sensing of

N-

(3-hydroxybutanoyl)-

L

-homoserine lactone (3-hydroxy-C4-HSL) (for a review, seeMiller and Bassler, 2001). Thus, quorum sensing probablyregulates VanT-regulated genes as it does biolumines-cence in

V. harveyi

.Quorum sensing in

V. harveyi

uses two cell-signallingsystems that function in parallel to regulate biolumines-cence in a cell density-dependent manner. The firstquorum-sensing system (system 1) relies on 3-hydroxy-C4-HSL (Bassler

et al

., 1993), the synthesis of which isdirected by

luxM

, whereas the second quorum-sensingsystem (system 2) uses a signal molecule (termed AI-2),which is probably a furanosyl borate diester. AI-2 is chemi-cally distinct from AHLs and is synthesized via LuxS(Surette

et al

., 1999; Schauder

et al

., 2001; Chen

et al

.,

1678

A. Croxatto

et al.

© 2004 Blackwell Publishing Ltd,

Molecular Microbiology

,

52

, 1677–1689

2002). The sensors for 3-hydroxy-C4-HSL and AI-2,named LuxN and LuxQ, respectively, resemble proteinsbelonging to two-component signalling systems (Bassler

et al

., 1993; 1994a), and each possesses a conservedhistidine kinase and response regulator domain but noDNA-binding domain. At low cell densities, in the absenceof signal molecules, LuxN and LuxQ are suggested towork in parallel by relaying the phosphates from theirresponse regulator domains to a shared phosphorelayprotein LuxU (Freeman and Bassler, 1999a). The phos-phate from LuxU is then transferred to the response reg-ulator domain of the

s

54

-dependent activator LuxO, which,when phosphorylated, represses bioluminescence byactivating the expression of an as yet unidentified repres-sor (Bassler

et al

., 1994b; Freeman and Bassler, 1999b;Lilly and Bassler, 2000). This repressor is predicted toinhibit the expression of

luxR

as LuxO has been shownto repress the expression of

luxR

mRNA (Miyamoto

et al

.,2003). In contrast, at high cell densities, the signalmolecules accumulate and are believed to bind to theirrespective sensors (Freeman and Bassler, 1999a,b). 3-Hydroxy-C4-HSL may bind directly to LuxN, whereas AI-2 is postulated to bind to LuxQ via interaction with aputative periplasmic protein LuxP. Binding of the signalsis suggested to switch the sensor kinase activities of LuxNand LuxQ into phosphatases, leading to the dephospho-rylation of LuxO. Expression of

luxR

is derepressed, andthus bioluminescence is induced.

Part of a

V. harveyi

-like quorum-sensing system hasbeen found in

V. anguillarum

(Milton

et al

., 2001; Croxatto

et al

., 2002). In addition to the transcriptional activatorVanT, homologues of the LuxM–LuxN signalling systemare known. VanM synthesizes both

N

-hexanoyl-

L

-homoserine lactone (C6-HSL) and

N-

(3-hydroxyhex-anoyl)-

L

-homoserine lactone (3-hydroxy-C6-HSL). Unlike

V. harveyi

,

V. anguillarum

strains are rarely luminous;however, genes encoding other activities regulated byVanT have been characterized (Croxatto

et al

., 2002). The

empA

gene, encoding a metalloprotease, requires VanTfor its expression and hence is a good candidate for func-tional analysis of this regulator. Unexpectedly, mutationsin the

vanM

and

vanN

genes did not affect proteolyticactivity as might be predicted from the

V. harveyi

model.There are several possible explanations for this observa-tion. First, a second parallel quorum-sensing system sim-ilar to LuxPQ may exist in

V. anguillarum

and, thus, aneffect on

empA

expression may be undetectable. Sec-ondly, LuxO and LuxU homologues may not exist. Thirdly,VanT may not be part of the quorum-sensing signallingcascade. Fourthly, this type of cascade may work differ-ently in

V. anguillarum

than in

V. harveyi

.To begin answering these questions, genes similar to

the

V. harveyi luxPQ

and

luxOU

were cloned from

V.anguillarum

. Null mutations in these genes were made,

and the effects of the mutations on the expression of

vanT

and

empA

were determined. This paper demonstratesthat (i) there is more than one quorum-sensing phospho-relay system in

V. anguillarum

; (ii) the systems work dif-ferently from their homologues in

V. harveyi

; and (iii) theymay function to limit the expression of

vanT

instead ofderepressing its expression as described in other

Vibrio

species.

Results

Identification of a second

V. harveyi

-like quorum-sensing system and its regulators

Vibrio anguillarum

contains gene homologues for the

V.harveyi

-like quorum-sensing system 1 and for the tran-scriptional regulator of bioluminescence (Milton

et al

.,2001; Croxatto

et al

., 2002). Thus, we speculated that

V.anguillarum

could also have the homologues for thosegenes encoding the

V. harveyi

-like quorum-sensing sys-tem 2 (

luxPQ

), the shared phosphorelay (

luxU

) and regu-latory (

luxO

) proteins. To search for these genes,degenerate primers based on the

V. harveyi luxO

and

luxQ

coding sequences were designed (Bassler

et al

.,1994a,b). Using these primers, polymerase chain reaction(PCR) fragments were generated and used to screen a

V.anguillarum

genomic library. Candidate cloned DNA frag-ments were sequenced and found to contain homologuesof

luxOU

and

luxPQ

. The predicted proteins had highpercentages of protein identity (VanO, 71–85%; VanU,45–58%; VanP, 61–64%; VanQ, 47–56%) to the corre-sponding homologues in five other

Vibrio

species,

V. har-veyi

,

Vibrio cholerae

,

Vibrio parahaemolyticus

,

Vibriofischeri

and

Vibrio vulnificus

. Because of the high proteinidentities, we have named the

V. anguillarum

proteinsVanP, VanQ, VanO and VanU. Interestingly, the proteinsencoded by the flanking open reading frames (ORFs) from

V. anguillarum

are also closely related to their equivalentsfrom

V. cholerae

(Heidelberg

et al

., 2000),

V. para-haemolyticus

(Makino

et al

., 2003),

V. fischeri

(Miyamotoet al., 2000) and V. vulnificus (GenBank, unpublished),indicating that the genetic loci are conserved in Vibriospecies.

For functional analyses, in frame null mutations weremade in vanO, vanU, vanQ and vanN, and the deletedcodons for each gene are given in Table 1.

VanT is expressed throughout growth

Repression of LuxR homologues in early log phase viathe V. harveyi-like quorum-sensing systems has beensuggested to be a common regulatory mechanism inVibrio species (Miyamoto et al., 2003). Transcripts of V.harveyi luxR are induced three- to fourfold from early

Vibrio anguillarum quorum sensing 1679

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

(OD600 0.1) to late log (OD600 1.0) phase growth (Martinet al., 1989; Miyamoto et al., 2003). A similar high induc-tion is also seen for V. fischeri LitR (Miyamoto et al., 2003)and for V. cholerae HapR (Zhu et al., 2002). In fact, luxR,litR and hapR mRNA is expressed minimally, if at all, atlow cell density. To investigate whether vanT expressionfollowed a similar pattern to the other luxR homologues,a lacZ transcriptional gene fusion was made with the vanTpromoter and inserted into the chromosome of the V.anguillarum wild type. Figure 1A shows that LacZ activitywas detected at both low and high cell density in the wildtype and showed no significant induction as the cell num-ber increased. Northern analysis was also done on bac-teria grown to an OD600 of 0.2, 0.7, 2.5 and 3.8. Figure 1Bshows that VanT transcripts were produced at low celldensity and that the amount of transcript did not differmuch during growth. These data suggest that VanT, unlikeother LuxR homologues, is expressed at early stages ofgrowth and is not significantly induced during growth.

VanT represses its own expression throughout growth

Autoregulation is a common trait for many prokaryoticregulatory proteins. To determine whether VanT affects itsown expression, the above vanT::lacZ transcriptionalgene fusion was inserted into the chromosome of a vanTnull mutant (AC10), and LacZ activity was measured.Expression was increased in the vanT mutant comparedwith that of the wild type (Fig. 1A), indicating that VanTrepresses its own expression. Moreover, electrophoreticmobility shift assays (EMSAs) showed that purified VanTacts directly to repress its expression by binding the vanTpromoter (Fig. 1C).

VanO and VanU have antagonistic roles in the expression of vanT

As LuxO homologues in V. harveyi, V. fischeri and V.cholerae repress the expression of their respective LuxR

Table 1. Bacterial strains and plasmids used in the study.

Strain or plasmid Genotype or relevant markers Reference or source

E. coli strainsBL21 (DE3) hsdS, lon, ompT, gal (lcIts857 ind1 Sam7 nin5 lacUV5-T7 gene 1) Studier and Moffatt (1986)SY327 D(lac pro), argE(am), rif, malA, recA56, lpir Miller and Mekalanos (1988)S17-1 thi, pro, hsdR, hsdM+, recA, RP4-2-Tc::Mu-Km::Tn7, lpir Simon et al. (1983)XL1-Blue recA1, endA1, gyrA96, thi-1, hsdR17, supE44, relA1, D(lac-pro), Stratagene

[F¢proAB lacIq lacZDM15 Tn10(Tetr)]V. anguillarum strainsNB10 Wild type, serotype 01, clinical isolate from the Gulf of Bothnia Norqvist et al. (1989)AC10 vanT in frame deletion; NB10 derivative; start codon fused to last codon Croxatto et al. (2002)AC11 vanO in frame deletion; NB10 derivative; codons 1–99 fused to codons 321–486 This studyDM63 vanQ in frame deletion; NB10 derivative; codons 1–417 fused to codons 841–848 This studyDM64 vanN in frame deletion; NB10 derivative; codons 1–367 fused to codons 851–859 This studyDM65 vanNQ in frame deletions; same mutations as DM63 and DM64 This studyDM71 vanU in frame deletion; NB10 derivative; codons 1–25 fused to the stop codon This studyDM80 vanUQ in frame deletion; same mutations as DM71 and DM63 This studyDM81 vanUN in frame deletion; same mutations as DM71 and DM64 This studyDM82 vanUQN in frame deletion; same mutations as DM71 and DM65 This study

PlasmidspBluescript Apr; ColE1 origin StratagenepBSVanO-285 Apr; pBluescript containing a 285 bp PCR fragment from vanO (bp 1389–1673) This studypBSVanQ-1030 Apr; pBluescript containing a 1035 bp PCR fragment from vanQ (bp 3317–4330) This studypBSVanO-16 Apr; pBluescript containing a cloned fragment with the vanOU genes This studypBSVanQ-11 Apr; pBluescript containing a cloned fragment with the vanQ gene and a partial vanP This studypBSVanQ-27 Apr; pBluescript containing a cloned fragment with the vanP gene and a partial vanQ This studypDM4 Cmr; suicide vector with an R6K origin (pir requiring) and sacBR of Bacillus subtilis Milton et al. (1996)pDMVanO2 Cmr; pDM4 derivative containing vanO bp 635–872 fused in frame to bp 1536–1774 This studypDMVanU2 Cmr; pDM4 derivative containing vanU bp 1955–2115 fused in frame to bp 2377–2554 This studypDMVanN3 Cmr; pDM4 derivative containing vanN bp 2328–2620 fused in frame to bp 4070–4307 This studypDMVanQ1 Cmr; pDM4 derivative containing vanQ bp 2865–3103 fused in frame to bp 4373–4604 This studypDM8 Cmr, Tcr; pSup202 derivative containing the promoterless lacZ gene cassette from Croxatto et al. (2002)

pTL61T used for transcriptional gene fusion studiespDM8-VanO Cmr, Tcr; pDM8 containing a vanO::lacZ transcriptional gene fusion (265 bp fusion) This studypDM8-Vps73 Cmr, Tcr; pDM8 containing a vps73::lacZ transcriptional gene fusion (295 bp fusion) Croxatto et al. (2002)pNQ705-1 Cmr; suicide vector that contains an R6K origin (pir requiring) McGee et al. (1996)pDM35 Cmr, pNQ705-1 derivative containing the promoterless lacZ gene cassette from This study

pTL61T used for transcriptional gene fusion studiespDM35-VanT Cmr, pDM35 containing an vanT::lacZ transcriptional gene fusion This studypDM35-EmpA Cmr, pDM35 containing an empA::lacZ transcriptional gene fusion This studypET30a Kanr, protein expression vector using pBR322 origin and the T7 promoter NovagenpET30aVanT Kanr, pET30a containing vanT fused to the T7 promoter This study

1680 A. Croxatto et al.

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

homologues (Zhu et al., 2002; Miyamoto et al., 2003), wewondered whether VanO represses VanT expression. Ifthis is true, then, according to the proposed model for V.harveyi, inactivation of both quorum-sensing systems 1and 2 (a vanNQ double mutation) and inactivation ofvanU, which encodes the phosphorelay protein, shouldhave a similar phenotype to that of a vanO mutant.Thus, the above vanT::lacZ transcriptional gene fusionwas inserted into the chromosome of the vanO (AC11),vanU (DM71), vanN (DM64), vanQ (DM63) and vanNQ(DM65) mutants. LacZ activity was assayed throughoutgrowth and compared with that of the wild type. For thevanO, vanN, vanQ and vanNQ mutants (Fig. 2A), a slightincrease in LacZ activity compared with the wild type was

Fig. 1. Autoregulation of vanT expression.A. A vanT::lacZ transcriptional gene fusion was expressed from the chromosome of the wild type (open squares) and the vanT mutant (filled stars). Samples at various time points were analysed for growth (OD600, dotted lines) and b-galactosidase expression (solid lines). Background expression for b-galactosidase activity is 15 units.B. Northern analysis. Total RNA was isolated from wild-type cultures with an OD600 of 0.2, 0.7, 2.5 and 3.8, and 5 mg from each was hybridized to a DNA fragment complementary to vanT. RNA tran-scripts were quantified, and the fold increase or decrease was com-pared with the first time point and given below the lanes.C. EMSA performed with DIG-labelled vanT promoter DNA and puri-fied VanT protein. All lanes contain ª2 ng of DNA with variable amounts of purified VanT protein as indicated. Experiments confirm-ing the specificity of DNA binding were also performed using both specific (unlabelled vanT promoter DNA) and non-specific (digested lambda DNA) competitor DNA (data not shown).

b-ga

lact

osid

ase

Uni

ts

OD

600

Time (hr)

A

BOD600 0.2 0.7 2.5 3.8

1 1.2

0.5

0.8

0.8 kb

VanTamount

boundpromoter

unboundpromoter

3 mg

0.3

mg

30 n

g

none

C

0 5 10 15 20 25 30

0,01

0,1

1

200

400

600

800

1000

1200T

WT

Fig. 2. Regulation of vanT expression by two quorum-sensing systems.A. A vanT::lacZ transcriptional gene fusion was expressed from the chromosome in the wild type (open squares), the vanU (filled circles), vanNQ (filled triangles), vanN (filled inverted triangles), vanQ (filled diamonds) and vanO (open circles) mutants. Samples at various time points were analysed for growth (OD600, dotted lines) and b-galactosidase expression (solid lines). Background expression for b-galactosidase activity is 15 units.B. Northern analysis. Total RNA from cultures of the wild type and from the vanO, vanQ, vanN, vanNQ and vanU mutants with an OD600 of 0.6 and 3.5 was isolated, and 5 mg from each was hybridized to a DNA fragment complementary to vanT. RNA transcripts were quan-tified, and the mutant/wild type ratio is given below the lanes.

WT O Q N NQ U WT O Q N NQ U

3.50.6

A

B

1 1.5

1.3

1.5

1.4

0.7

1 2.3

2.1

2.4

2.6

0.6

0.8kb

Time (hr)

OD600

b-ga

lact

osid

ase

Uni

ts

OD

600

0 5 10 15 20 25 30 35

0,01

0,1

1

U

NQOQN

WT

0

200

400

600

800

1000

1200

Vibrio anguillarum quorum sensing 1681

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

seen throughout growth. Surprisingly, LacZ activity wasdecreased in a vanU mutant. Although the differences inactivity were moderate, they were reproducible. To confirmthat the LacZ activities were a reflection of the amount ofvanT transcript present, Northern analysis was also donefor each mutant grown to an OD600 of 0.6 and 3.5. Theamounts of vanT transcripts were measured and are pre-sented in Fig. 2B as a ratio of the mutant to wild-typeamount. Similar to the LacZ data, vanT transcript levelswere increased at an OD600 of both 0.6 and 3.5 for allmutants compared with the wild type except for the vanUmutant, for which the transcript levels were decreased.These data show that, in V. anguillarum as in other Vibriospecies, VanO does repress expression of vanT althoughless significantly. Moreover, VanN and VanQ may functionco-operatively as loss of one or the other gave a pheno-type similar to a vanO and a vanNQ mutant. What isunique to V. anguillarum is that VanU appears to play arole in the activation of vanT expression, possibly givingVanU a more pivotal role in regulating the quorum-sensingregulon.

VanT positively regulates vanOU

As vanT expression does not vary greatly during cellgrowth and as VanU and VanO play antagonistic roles invanT expression, we speculated that, in addition to adirect autoregulation, VanT may regulate the possiblevanOU operon to modulate its own regulation further. Forthis analysis, a lacZ transcriptional gene fusion was madewith the vanOU promoter and carried on plasmid pDM8-VanO. This plasmid was mobilized into the wild type andthe vanT mutant (AC10), and LacZ activity was measured(Fig. 3A). During late log phase growth for the vanTmutant, LacZ activity decreased, indicating that VanT isrequired for greater expression of vanOU in stationaryphase. Confirming these data, Northern analysis after24 h of growth showed that the vanOU transcript isdecreased in the vanT mutant compared with the wild type(Fig. 3B). An estimated transcript size was shown to be1.9 kb (DNA sequence predicts 1.8 kb), indicating thatthese two genes are part of the same operon. Further-more, EMSA studies using purified VanT showed thatactivation of vanOU is via a direct interaction of VanT withthe vanOU promoter (Fig. 3C). Taken together, these datasuggest that regulation of the vanT gene is complex andthat this complexity may be necessary to limit the level ofVanT within the cell.

The VanT-regulated metalloprotease gene empA is similarly regulated by quorum-sensing systems 1 and 2

Vibrio anguillarum NB10 is not a luminescent bacteriumand does not have functional lux genes that can be used

to characterize a cellular response resulting from quorumsensing. Ideally, a native V. anguillarum gene that isdirectly regulated by VanT and can be assayed biologicallyis needed. Previously, VanT was shown to be required for

Fig. 3. VanT positively regulates vanOU expression.A. A vanO::lacZ transcriptional gene fusion was expressed from a plasmid (pDM8-VanO) in the wild type (open squares) and the vanT mutant (filled stars). Samples at various time points were analysed for growth (OD600, dotted lines) and b-galactosidase expression (solid lines). Background expression for b-galactosidase activity is 200 units.B. Northern analysis. Total RNA from 24 h cultures of the wild type (OD600 of 3.1) and vanT mutant (OD600 of 2.0) was isolated, and 5 mg from each was hybridized to a DNA fragment complementary to vanO. RNA transcripts were quantified, and the mutant/wild type ratio is given below the lane.C. EMSA performed with DIG-labelled vanO promoter DNA and puri-fied VanT protein. All lanes contain ª2 ng of DNA with variable amounts of VanT protein as indicated. Experiments confirming the specificity of DNA binding were also performed using both specific (unlabelled vanO promoter DNA) and non-specific (digested lambda DNA) competitor DNA (data not shown).

OD

600

Time (hr)

b-ga

lact

osid

ase

Uni

ts

Wt DvanT

A

B 1.9kb

C

3 mg

0.3

mg

30 n

g

noneVanT

amount

boundpromoter

unboundpromoter

1 0.6

0 5 10 15 20 25 30

0,01

0,1

1

T

WT

0

1500

3000

4500

6000

7500

9000

10500

1682 A. Croxatto et al.

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

activation of the metalloprotease gene, empA (Croxattoet al., 2002). However, it is not known whether the regu-lation is mediated directly via the interaction of VanT withthe empA promoter. EMSA studies revealed that purifiedVanT does indeed bind the empA promoter (Fig. 4). Con-sequently, empA expression was used to measure cellularresponses resulting from the V. harveyi-like quorum-sensing regulatory cascade in V. anguillarum.

Three methods were used to assay empA expressionin the vanO (AC11), vanU (DM71), vanN (DM64), vanQ(DM63) and vanT (AC10) mutants as well as a vanNQdouble mutant (DM65) and the wild type. First, a lacZtranscriptional gene fusion was made with the empA pro-moter and inserted into the chromosome of the V. anguil-larum strains. LacZ activity was measured throughoutgrowth (Fig. 5A). Secondly, Northern analysis was done,and the amount of empA transcript was measured in abacterial culture grown to an OD600 of 0.6 and 3.5 andpresented as ratio of the mutant to wild-type amount(Fig. 5B). Thirdly, a quantitative protease assay previouslyused to assay EmpA was done for each strain (Fig. 5C).These three methods result in comparable conclusions.However, this may not be obvious because of the scaleused in Fig. 5A at the OD600 of 0.6. The LacZ activity forthe vanU and vanT mutants and wild type was withinbackground levels at 15 units, and the vanO, vanQ, vanNand vanNQ were 43, 24, 43 and 45 units respectively. Asthe vanU and vanT units were background units, we couldnot detect in this assay anything lower than that. However,the protease activity (Fig. 5C) clearly showed a decreasein activity at this period of growth for both the vanU andthe vanT mutants (0.006 and 0.007 absorbance respec-tively) compared with the wild type (0.055 absorbance).As expected from previous studies (Croxatto et al., 2002),the empA promoter was inactive in the vanT mutant, indi-cating that VanT is essential for expression of empA under

Fig. 4. VanT binds the empA promoter. EMSA performed with DIG-labelled empA promoter DNA and purified VanT protein. All lanes contain ª2 ng of DNA with variable amounts of VanT protein as indicated. Experiments confirming the specificity of DNA binding were also performed using both specific (unlabelled empA promoter DNA) and non-specific (digested lambda DNA) competitor DNA (data not shown).

2.5

mg

25 n

g

noneVanT

amount

boundpromoter

unboundpromoter

0.25

mg

Fig. 5. Regulation of empA expression by two quorum-sensing systems.A. An empA::lacZ transcriptional gene fusion was expressed from the chromosome of the wild type (open squares), the vanU (filled circles), vanNQ (filled triangles), vanN (filled inverted triangles), vanQ (filled diamonds), vanO (open circles) and vanT (filled stars) mutants. Sam-ples at various time points were analysed for growth (OD600, dotted lines) and b-galactosidase expression (solid lines). Background expression for b-galactosidase activity is 15 units.B. Northern analysis. Total RNA from cultures of the wild type and from the vanO, vanQ, vanN, vanNQ, vanU and vanT mutants with an OD600 of 0.6 and 3.5 was isolated, and 5 mg from each was hybridized to a DNA fragment complementary to empA. RNA transcripts were quantified, and the mutant/wild type ratio is given below the lanes.C. Protease activity. Samples from the wild type (open squares), the vanU (filled circles), vanNQ (filled triangles), vanN (filled inverted triangles), vanQ (filled diamonds), vanO (open circles) and vanT (filled stars) mutants were taken at various times and analysed for growth (OD600, dotted lines) and for azocasein degradation (OD442, solid lines).

OD

600

Time (hr)

b-ga

lact

osid

ase

Uni

ts

WT O Q N NQ U T WT O Q N NQ U T

3.50.6

A

B

1 2.1

0.9

2.1

2.2

0.3

0.1

1 2.9

3.1

2.7

2.9

0.6

0.2

1.8 kb

OD

600

Time (hr)

Abs

orba

nce

442n

m

C

OD600

0 5 10 15 20 25 30

0,01

0,1

1

0

200

400

600

800

1000

1200

1400

1600

1800ONQ

N

Q

T

WT

U

0 5 10 15 20 25 30

0,01

0,1

1

0,0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

T

U

WT

O

NNQ

Q

Vibrio anguillarum quorum sensing 1683

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

the growth conditions used. A derepression of empAexpression is seen in the vanO, vanN, vanQ and vanNQmutants compared with the wild type, suggesting that thequorum-sensing systems 1 and 2 repress the expressionof empA. As for vanT expression, empA expression wasdown for the vanU mutant compared with the wild type.Thus, the quorum-sensing cascade regulates proteolyticactivity similarly to vanT expression.

VanU is not essential for VanN and VanQ repression of vanT and empA

The decreased expression of vanT and empA in a vanUmutant compared with the wild type was quite unex-pected. There are two possible explanations for theseresults. First, VanU aids in the decay of the phosphatesignal from VanO, thus decreasing the duration of vanTand empA repression. This would require that VanO isphosphorylated in the absence of VanU. Secondly, VanUtransfers a phosphoryl group to a second response regu-lator that activates expression of vanT and empA. Todetermine whether VanN and VanQ are required for acti-vation of VanO in the absence of VanU, double vanUN andvanUQ mutants and a triple vanUNQ mutant were made,and expression of vanT and empA was measured byNorthern analysis (Fig. 6). Surprisingly, the multiplemutants gave a phenotype similar to that for the singlevanN and vanQ mutants, suggesting that loss of VanN orVanQ in a vanU mutant relieves the repression on vanTand empA expression and that VanO is probably phos-phorylated by VanN and VanQ via an alternative route forsignal relay. However, this still does not rule out that VanUmay transfer a phosphoryl group to a second, as yetunidentified, response regulator.

Discussion

Genes encoding part or all of two parallel V. harveyi-likequorum-sensing systems have been identified in severalVibrio species, V. cholerae, V. vulnificus, V. parahaemolyt-icus, V. anguillarum and V. fischeri. Signal transduction inthe V. harveyi systems occurs via a phosphorylation relaythat uses hybrid two-component-type sensors thatrespond to two types of signals, 3-hydroxy-C4-HSL andAI-2 (for a review, see Miller and Bassler, 2001). Thesignals from the two sensors converge on to a singlephosphorelay protein, LuxU, which activates the responseregulator LuxO. LuxO then represses LuxR, the activatorof bioluminescence genes. This type of quorum-sensingsystem has been found only in Vibrio species and is aglobal regulatory system in these bacteria.

In this study, genes encoding additional proteins of theV. harveyi-like quorum-sensing system in V. anguillarumwere cloned, and null mutants were constructed. These

mutants were used in functional studies that were aimedat understanding how quorum sensing in V. anguillarumregulates gene expression. Similarities to, as well as strik-ing differences from, the other Vibrio systems were seen.A model is suggested for the V. anguillarum quorum-sensing system and presented in Fig. 7.

LuxO homologues from V. harveyi, V. cholerae and V.fischeri have been shown to regulate negatively theexpression of their respective transcriptional activatorsluxR, hapR and litR (Zhu et al., 2002; Miyamoto et al.,2003). LuxO homologues are believed to work indirectlyvia activation of an as yet unidentified repressor. Thismechanism is suggested to be a general regulatory mech-anism in vibrios (Miyamoto et al., 2003). In V. anguillarum,VanO also negatively regulated the expression of vanT.However, expression of vanT was not completelyrepressed at low cell density in V. anguillarum as are luxRhomologues in the other Vibrios. In fact, the amount ofvanT transcript at an OD600 of 0.2 does not differ muchfrom transcript amounts at an OD600 of 0.7, 2.5 and 3.8,indicating that VanO does not completely repress theexpression of vanT at low cell density.

The level of vanT expression was also limited by auto-

Fig. 6. VanU is not required for repression of empA and vanT.A. Northern analysis. Total RNA from cultures of the wild type and from the vanUN, vanUQ and vanUNQ mutants with an OD600 of 0.6 and 3.5 was isolated, and 5 mg from each was hybridized to a DNA fragment complementary to empA. RNA transcripts were quantified, and the mutant/wild type ratio is given below the lanes.B. Northern analysis. Total RNA from cultures of the wild type and from the vanUN, vanUQ and vanUNQ mutants with an OD600 of 0.6 and 3.5 was isolated, and 5 mg from each was hybridized to a DNA fragment complementary to vanT. RNA transcripts were quantified, and the mutant/wild type ratio is given below the lanes.

WT UN UQ UNQ WT UN UQ UNQ

3.50.6

1 1.6

2.0

1.9

1 1.7

2.0

1.5

0.8 kb

WT UN UQ UNQ WT UN UQ UNQ

3.50.6

1 2.1

1.4

2.2

1 2.1

1.5

1.7

1.8 kb

A

B vanT

empA

OD600

OD600

1684 A. Croxatto et al.

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

repression via two mechanisms. The first mechanismresulted from a direct effect of VanT on its own promoter.The vanT promoter had a slightly (1.3- to 1.5-fold) higheractivity in a vanT mutant than in the wild type, suggesting

a minor role in self-regulation. However, VanT did bind itsown promoter, indicating a direct effect on self-expression.An autorepressor mechanism was indicated previously forLuxR from V. harveyi in which LuxR binds its own pro-moter and interferes with binding of RNA polymerase tothe promoter (Chatterjee et al., 1996). The second mech-anism, not observed in other vibrios, results from an indi-rect effect, in which VanT binds to the vanOU promoterand increases expression at the onset of stationaryphase. As the expression of vanT does not vary muchthroughout growth and as VanT increases expression ofvanOU late in growth, the V. harveyi-like quorum-sensingsystems in V. anguillarum probably function to limit theexpression of VanT instead of inducing expression as inother vibrios. Why VanT expression needs to be limited isnot yet apparent.

The difference in vanT and vanO expression in the vanTmutant compared with the wild type appears to coincidewith entry into stationary phase. Although we cannot ruleout the possibility that the effect seen results from specificregulators that are activated by other signalling mecha-nisms during entry into stationary phase instead of lossof the global regulator VanT, we believe the regulatoryeffect of VanT is direct as the protein binds the vanT andthe vanO promoter DNA.

A prediction from the V. harveyi model is that activationof VanO would depend on phosphorylation via theupstream components VanN, VanQ and VanU. In V. har-veyi, LuxN and LuxQ are sensor kinases that provide twoindependent phosphorylation channels that converge onto the LuxU phosphorelay protein. To prevent phosphory-lation of LuxO and thus repression of bioluminescence asseen with a luxO mutation, a luxNQ double mutation or aluxU mutation is required in V. harveyi (Freeman andBassler, 1999a,b). However, in V. anguillarum, anincrease in vanT and empA expression similar to that ina vanO mutant was achieved in a single vanN or vanQmutant as well as in a double vanNQ mutant. These datasuggest that activation of VanO requires both VanN andVanQ. These two systems may not work independently ofeach other as in V. harveyi (Freeman and Bassler, 1999b).Unexpectedly, a vanU mutant exhibited decreasedexpression of vanT and empA compared with the wildtype, suggesting that VanU has additional roles in V.anguillarum other than that of transferring a phosphorylgroup to VanO.

VanU, similar to other LuxU homologues, is a small 113-amino-acid protein that contains a detached histidine-containing phosphostransferase (HPt) domain, an inter-mediate in multistep phosphorelay systems that transfersa phosphoryl group from one response regulatory domain(Asp-1) to another (Asp-2). HPt domains are often local-ized within a hybrid two-component sensor kinase thatcontains three of the amino acid residues required for

Fig. 7. A model for V. harveyi-like quorum sensing systems in V. anguillarum. Dashed lines indicate possible regulatory actions pre-dicted by this study that need further investigation. Solid lines indicate regulatory actions that we show in this study and that are predicted from previous models for V. harveyi, V. cholerae and V. fischeri. Double arrowheads indicate phosphorylation relay, and a single arrowhead indicates gene activation. VanN and VanQ, both sensor kinases, are suggested to work together to initiate the phosphoryla-tion relay. Transduction of the signal may occur via VanU; however, VanU is not essential for transfer of the phosphoryl group to VanO. VanU, a phosphotransferase, is probably multifunctional and plays a role in either the decay of phosphoryl group from VanO and/or the activation of a second response regulator that activates VanT expres-sion. VanO represses VanT expression and may function to limit the levels of VanT in the cell instead of inducing expression, as suggested for luxR expression in V. harveyi. The transduction of signals through this regulatory cascade is measured as a cellular response via VanT activation of empA, which encodes a metalloprotease.

VanN VanQ

VanU

VanO Activator

RR 2

RR 1

HkHk

RR 3

RR 4

Repressor

EmpA

VanT

H1 P

P D1

P H2

D2 P

H1 P

D2 P

Unknown Mechanism

?

?

D1 P

P P

P

Vibrio anguillarum quorum sensing 1685

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

phosphorelay, His-1, Asp-1 and His-2 (for a review, seePerraud et al., 1999). VanU contains the His-2 required forphosphorelay signalling and is detached from the hybridsensor kinases VanQ and VanN, which contain therequired residues His-1 and Asp-1. A few other detachedphosphorelay proteins have been identified outside thegenus Vibrio. SpoOB from Bacillus subtilis is involved insporulation (Burbulys et al., 1991), RdeA from Dictyoste-lium discoideum is involved in development (Chang et al.,1998), and Ypd1 from Saccharomyces cerevisiae isinvolved in osmoregulation and oxidative stress (Posaset al., 1996; Li et al., 1998).

The decrease in vanT expression exhibited in a vanUmutant might be explained in one of two ways based onmore specific functions of HPt domains and the geneticanalysis that we have to date. These possibilities aredepicted in Fig. 7 as dashed lines. First, VanU may phos-phorylate an alternative response regulator that interfereswith repression of vanT expression. How VanU wouldachieve a balance between increased and decreasedvanT expression by phosphorylating two downstreamresponse regulators is not clear. One speculation may bethat VanU interacts with the response regulatorsunequally. An interesting example for this has been shownfor the detached HPt domain YPD1 from S. cerevisiae,which phosphorylates two downstream response regula-tor domains (Asp-2), SSK1 involved in osmoregulationand SKN7 involved in oxidative stress (Posas et al., 1996;Li et al., 1998). Although YPD1 relays phosphoryl groupsto both proteins, it interacts with the two response regu-lators differently. A protein–protein complex is formed withSSK1 stabilizing the phosphorylation state but not withSKN7 (Janiak-Spens et al., 2000). Secondly, VanU maybe important for aiding signal decay or dephosphorylationof VanO, as has been shown for ArcB from Escherichiacoli (Georgellis et al., 1998). Such a role would affect thehalf-life of the phosphorylated response regulator, andthe length of phosphorylation would affect the duration ofthe cellular response. If VanU is absent, then phosphory-lated VanO is present in higher amounts, and this wouldprobably enhance repression of vanT expression. For thisto be true, phosphorylation of VanO cannot depend onVanU. This, in fact, is suggested by the vanT expressionstudies using double vanNU and vanQU and the triplevanNQU mutant. Each of these multiple mutants had alevel of derepression similar to that for the vanO mutant,suggesting that, in a single vanU mutant, VanO is stillphosphorylated, and that this phosphorylation is depen-dent on VanN and VanQ. Hence, phosphorylation viaVanNQ is not dependent on VanU. Either an additionalHPt domain is involved or possibly phosphorylation istransferred from a His-1 to an Asp-2 residue directly, ashort-cut phosphotransfer (Perraud et al., 1999).

This study clearly shows that the regulation of vanT and

VanT-regulated genes such as empA via the V. harveyi-like quorum-sensing systems is distinct and more complexthan would be predicted from the V. harveyi model. In thisinitial study, we propose that this quorum-sensing systemmay function in V. anguillarum differently from that of V.harveyi. First, VanT is not strongly induced like other LuxRhomologues during late log phase growth. Instead, vanTis expressed at low cell density and does not changemuch during growth. Secondly, VanU and VanO affect theexpression of vanT oppositely and, thirdly, VanT positivelyregulates the expression of vanOU during late growth. Wepropose that this quorum-sensing system is used to limitthe expression of VanT, a global regulator that induces theexpression of several genes during the late stationaryphase of growth (Croxatto et al., 2002), and that VanU, aphosphotransferase protein, may be the pivotal protein forreceiving both positive and negative signals, which couldbe external signals other than AHLs, and integrating theminto a cellular response via VanO or possibly otherresponse regulators yet to be identified.

Experimental procedures

Strains, plasmids and media

Bacterial strains and plasmids are described in Table 1.Escherichia coli SY327 (lpir) was used for transformationafter subcloning fragments into either the pNQ705-1 or thepDM4 suicide vector derivatives. All plasmids to be conju-gated into V. anguillarum were transformed into E. coli S17-1 (lpir), which was used as the donor strain. Plasmid trans-fers from E. coli to V. anguillarum were done as describedpreviously (Milton et al., 1996). E. coli XL1-Blue was used forbacteriophage lambda infections and for most transforma-tions. E. coli BL21 (DE3) was used for overexpressing andpurification of proteins.

Escherichia coli was grown routinely in Luria broth(10 g l-1 Bacto tryptone, 5 g l-1 Bacto yeast extract and10 g l-1 sodium chloride) with the following antibiotic concen-trations: ampicillin at 100 mg ml-1 and chloramphenicol at25 mg ml-1. For V. anguillarum, trypticase soy medium (TSB)from BBL was used routinely. For selection against E. coliafter conjugation, the Vibrio selective medium TCBS agar(Difco Laboratories) was used. Antibiotic concentrations forV. anguillarum were tetracycline at 5 mg ml-1 and chloram-phenicol at 5 mg ml-1.

DNA techniques, PCR conditions and sequencing

Unless otherwise stated, all conditions for the various DNAtechniques were as described by Sambrook et al. (1989).Reaction conditions for the DNA-modifying enzymes andDNA restriction enzymes were performed as suggested bythe manufacturers. Double-strand DNA sequencing wasperformed using automated sequencing on an ABI Prism377 DNA sequencer and by primer walking in two direc-tions from known regions of DNA sequence. PCR wasperformed as described previously (McGee et al., 1996;

1686 A. Croxatto et al.

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

Croxatto et al., 2002). PCR optimization kit from Stratagene(buffer 10) was used to generate the 285 bp vanO PCRfragment.

Cloning and sequencing of vanOU and vanPQ DNA loci

For the cloning of the vanOU locus, PCR was performedusing the degenerate primers LuxO-3 [5¢-GGCACTAGT(CT)TGICGIAC(GA)TTICCIGGCCA-3¢], which contains an SpeIsite at the 5¢ end and is complementary to V. harveyi luxOresidues 2121–2140 bp encoding amino acids WPGNVRQ,and LuxO-4 [5¢-GGTGAGCTCGGTTCTTCTAA(AG)ATGAA-3¢], which contains a SacI site at the 5¢ end and is com-plementary to V. harveyi luxO residues 1857–1872 bpencoding amino acids GSSKMK (Bassler et al., 1994b).These two primers generated a 285 bp fragment from thechromosome of V. anguillarum. For the cloning of thevanPQ locus, PCR was performed using the degenerateprimers LuxQ-1 [5¢-GGTGAGCTCAGICA(TC)GA(GA)ATICGIACICC-3¢], which contains a SacI site at the 5¢ end and iscomplementary to V. harveyi luxQ residues 2407–2426 bpencoding amino acids SHEIRTP, and LuxQ-5 [5¢-GGCACTAGTGACATAGTTTGCACCTGC(AGTC)GCCAT-3¢], whichcontains an SpeI site at the 5¢ end and is complementary toV. harveyi luxQ residues 3412–3435 bp encoding aminoacids MAAGANYV (Bassler et al., 1994a). These two prim-ers generated a 1030 bp fragment from the chromosome ofV. anguillarum.

Each PCR fragment was purified from a 1% agarose gelusing Ultrafree-DA spin columns (Millipore), digested over-night with SacI and SpeI and cloned into similarly digestedpBluescript (Stratagene), creating pBSVanO-285 andpBSVanQ-1030. Each fragment was sequenced to confirmthat the deduced protein sequence was significantly identicalto the respective V. harveyi gene and used as a probe toscreen a previously described (Milton et al., 1992) genomiclibrary from V. anguillarum in the Lambda Zap II bacterioph-age (Stratagene). The probes were labelled by random prim-ing using [a-32P]-dCTP as described by Sambrook et al.(1989). The pBluescript plasmids containing a chromosomalinsert were excised from positive plaques as described pre-viously (Milton et al., 1992). One plasmid that contained theentire DNA locus for vanOU, pBSVanO-16, was chosen forsequencing of these genes. For vanPQ, no clone containedboth genes. Thus, two plasmids, pBSVanQ-11 andpBSVanQ-27, which together contained sequence for bothgenes, were used to sequence vanPQ.

In frame deletion mutagenesis

Vibrio anguillarum strains were made that carried single,double or triple in frame deletions that created null mutations.The construction of in frame deletions by allelic exchange hasbeen described in detail previously (Milton et al., 1996).Mutant strains and the residues missing for each gene aswell as the pDM4 derivatives carrying the mutant alleles usedfor making the in frame deletions are listed in Table 1. All inframe deletions were confirmed by PCR amplification of therespective DNA loci and subsequent DNA sequencing ofeach PCR product.

Construction of transcriptional b-galactosidase fusions

Transcriptional gene fusions were constructed between thereporter gene lacZ from E. coli and the V. anguillarum pro-moters, vanO, vanT and empA. For these studies, two typesof plasmid vectors that contain the entire lacZ gene and itsribosomal binding site but lack the lacZ promoter wereused, pDM8 and pDM35. For lacZ gene fusions carried ona plasmid, pDM8, a medium-copy pSup202 derivative wasused that was described previously (Croxatto et al., 2002).To integrate the lacZ gene fusions into the chromosome,pDM35, a pir-requiring suicide vector that contains thesame promoterless lacZ gene as pDM8, was made. To cre-ate pDM35, pDM8-Vps73 (Croxatto et al., 2002) waspartially digested with BamHI. The 3400 bp fragment con-taining the promoterless lacZ gene from pDM8 plus thevps73 promoter was gel purified from a 1% agarose gel andligated to the suicide vector pNQ705-1 that was cut withBglII. The resulting plasmid was digested with SmaI toremove the vps73 promoter and then religated creatingpDM35. For both pDM8 and pDM35, the SmaI siteupstream of the promoterless lacZ gene is unique and wasused for the fusion of promoter regions. For all transcrip-tional fusions, DNA fragments containing promoters butlacking the possible ribosomal binding site were amplifiedby PCR using primers that contained SmaI sites at the 5¢ends. The PCR primer pairs are as follows: VanT-bgal-1(5¢-TCCCCCGGGATGATGGCGTCGCGTAACT-3¢) andVanT-bgal-3 (5¢-TCCCCCGGGGCCAATGACTGTTGAATT-3¢); VanO-bgal-1 (5¢-TCCCCCGGGTCAGCAGTTCATTAC-3¢) and VanO-bgal-3 (5¢-TCCCCCGGGTATTTTGCACTTTGCC-3¢); and EmpA-bgal-1 (5¢-TCCCCCGGGTTATATTGATAGTTATGT-3¢) and EmpA-bgal-3 (5¢-TCCCCCGGGGAGAGTTATTATTAGCAT-3¢). The DNA fragments were gel puri-fied, digested with SmaI and ligated into the unique SmaIsite on both vectors, resulting in pDM8-VanO, pDM35-VanTand pDM35-EmpA. All constructs were sequenced toensure that the gene fusions were made properly. Chromo-somal integration of pDM35-VanT and pDM35-EmpA wasvia the promoter sequence used to make the gene fusion.Thus, each gene fusion was integrated into the promoterregion of the respective gene. As the promoters were dupli-cated on the chromosome, these insertions did not disrupteither empA or vanT. The wild-type genes were still active.Both chromosomal integrations were checked by PCR toensure proper insertion in the chromosome.

b-Galactosidase assays

Vibrio anguillarum cultures were grown overnight at 24∞C inTSB containing chloramphenicol (3 mg ml-1). Cell cultureswere diluted to an OD600 of 0.01 (107 cells ml-1) forempA::lacZ and 0.001 for all other lacZ gene fusions (106

cells ml-1) in the same medium and incubated further at 24∞Cwith shaking. Samples were taken at the various time points,and b-galactosidase assays were performed in triplicateaccording to the method of Miller (1972). After the assay wasstopped, the bacterial debris was pelleted, and the A420 wasmeasured for the reaction supernatant. Specific activitywas determined as 1000 ¥ A420 ¥ min-1 ¥ ml-1 ¥ A600

-1.Growth was determined by measuring the A600 after diluting

Vibrio anguillarum quorum sensing 1687

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

the culture in fresh culture medium to give a reading withinthe linear range of the spectrophotometer.

Measurement of protease activity

Vibrio anguillarum cultures were grown overnight at 24∞C inTSB containing chloramphenicol (3 mg ml-1). Cell cultureswere diluted to an OD600 of 0.01 (107 cells ml-1) in the samemedium and incubated further at 24∞C with shaking. Culturesupernatants from various time points were assayed for pro-tease activity as described previously (Croxatto et al., 2002).Briefly, 100 ml of filtered-sterilized culture supernatant wasmixed with 100 ml of azocasein solution and incubated at30∞C for 2 h. The reaction was stopped by the addition ofTCA, the unreacted azocasein was removed by centrifuga-tion, and the absorbance at 442 nm was determined.

Purification of VanT

Overnight cultures of E. coli harbouring pET30VanT weresubcultured into fresh LB medium, incubated at 37∞C for 2 hat 200 r.p.m. and then induced with 1 mM IPTG. Incubationwas continued for an additional 2 h, after which the cells wereharvested and washed in PBS (pH 7.4) and then sonicated.The cell lysates were centrifuged for 5 min at 13 000 g (4∞C)to remove cellular debris. A 10 ml volume of nickel resin (NewEngland Biolabs) contained within a disposable column wasequilibrated with 100 ml of cold PBS. Cell lysates wereapplied to the column followed by a wash with 100 ml of PBS.Non-specifically bound proteins were eluted with 20 ml of0.1 M imidazole in PBS. The bound protein was eluted using1 M imidazole in PBS, and the eluant was collected as 1 mlfractions. The protein content of the fractions was assayedby absorbance at 280 nm (A280), and the purity of VanT wasanalysed by SDS-PAGE.

DNA gel shifts

Digoxigenin (DIG)-labelled target DNA promoter regionswere generated by PCR amplification using one standardoligonucleotide and one bearing a 5¢ DIG label. PCR primersfor each promoter region were as follows: vanTGS1DIG -101(5¢-CCATATAGCTAAAGCCGTTC-3¢); vanTGS2 +82 (5¢-CAAGAGCGATTTCCATCAG-3¢); vanOUGS1DIG -335 (5¢-GTACAAACACGCACAAAACC-3¢); vanOUGS2 +127 (5-CCTGTACCAACAATATTGAT-3¢); empAGS1DIG -212 (5¢-TGATAGTTATGTGCACTATT-3¢); and empAGS2 +168 (5¢-ATCATCAACCTGCACCATTT-3¢). PCR products were quantified bymeasuring absorbance at 260 nm using a Genequant spec-trophotometer (Pharmacia).

Polyacrylamide gels (5%) were prepared by mixing 6 ml ofBio-Rad Protogel (30:08), 8 ml of 5¥ TBE (0.5 M Tris, 0.5 Mboric acid, 0.01 M EDTA, pH 8.0), 250 ml of glycerol, 25.6 mlof sterile distilled water, 560 ml of 10% (w/v) ammonium per-sulphate and 16 ml of TEMED. DNA–protein binding reactionswere prepared using 0.5 ml of DIG-labelled DNA, 4 ml of 5¥binding buffer [100 mM Hepes, pH 7.6, 50 mM (NH4)2SO4,5 mM dithiothreitol (DTT), 1% (v/v) Tween 20, 150 mM NaCl],10 ml of purified protein and sterile distilled water to a finalreaction volume of 20 ml. The reactions were incubated at

room temperature for 10 min, after which 5 ml of 50% (v/v)glycerol was added, and the reactions were loaded on to thegel. Gels were run on ice at 100 V in 0.5¥ TBE for 1 h. DNAwas subsequently transferred to a nylon membrane overnightat 4∞C using electrophoretic transfer at 100 mA in transferbuffer [2.88% (w/v) glycine, 0.6% (w/v) Tris-HCl]. Blots werethen blocked for 1 h in 1% (w/v) blocking reagent (Roche)reconstituted in maleic acid buffer (0.1 M maleic acid, 0.15 MNaCl, pH 7.5). Chemiluminescent labelled antibodies hybrid-izing to the DIG DNA were detected using the DIG lumines-cent detection kit (Roche). Gel shifts were also conducted inthe presence of competitor DNA. Digested lambda DNA wasused as random competitor at a concentration 100 timesgreater than that of the target DNA. Non-labelled target DNAwas used as specific competitor. Blots were then exposed toautoradiograph film (Kodak) and developed according to themanufacturer’s instructions.

Northern analysis

Vibrio anguillarum cultures were grown overnight at 24∞C inTSB containing chloramphenicol (3 mg ml-1). Cell cultureswere diluted to an OD600 of 0.01 (107 cells ml-1) in the samemedium and incubated further at 24∞C with shaking. BeforeRNA isolation, bacterial cultures at various time points weretreated with RNAprotect bacteria reagent (Qiagen). TotalRNA was then isolated using the Rneasy minikit (Qiagen).The RNA (5 mg) was separated in a 1.2% formaldehyde aga-rose gel as described in the Qiagen RNeasy minikit hand-book and transferred to a ZetaProbe GT membrane (Bio-RadLaboratories), as suggested by the manufacturers. Hybridiza-tion was also performed as suggested by the manufacturer.DNA probes were DIG labelled and generated by PCRamplification using the PCR DIG probe synthesis kit fromRoche. PCR primers for each gene probe are as follows: forvanT, VanT-1 (5¢-CCACGCAGATATTGCTG-3¢) and VanT-13(5¢-ACGTTCAATGGCTTTGAT-3¢); for empA, 69-6 (5¢-AACAAAAGCAAGCGGTT-3¢) and 69-2R (5¢-TTGATAAACAAGCCCTG-3¢); for vanO, VanO-E (5¢-TTCCGTTGCGGCGCT-3¢) and VanO-4 (5¢-TGCGCCTTTCACGTGGCC-3¢).Chemiluminescent probes hybridized to RNA transcriptswere activated using the DIG luminescent detection kit(Roche), and bands were visualized using a Fluor-STM Multi-Imager (Bio-Rad). Luminescence was then quantified usingthe QUANTITY ONE version 4.2.3 software from Bio-Rad.

Computer analysis

Database searches were done using the Genetics ComputerGroup sequence analysis software (Devereux et al., 1984).

Nucleotide sequence accession number

Sequence data have been submitted to GenBank underaccession number AY525157 for vanOU and AY525158 forvanPQ.

Acknowledgements

This work was supported by a grant from the SwedishCouncil for Environment, Agricultural Sciences and Spatial

1688 A. Croxatto et al.

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

Planning (to D.M.), by a grant from the Carl Tryggers Foun-dation, Sweden (to D.M.) and by a grant and a studentshipfrom the Biotechnology and Biological Sciences ResearchCouncil, UK (to A.H., P.W. and M.C.), which are gratefullyacknowledged.

References

Actis, L.A., Tolmasky, M.E., and Crosa, J.H. (1999) Vibriosis.In Fish Diseases and Disorders, Vol. 3. Viral, Bacterial andFungal Infections. Woo, P.T.K., and Bruno, E.W. (eds).Wallingford: CABI, pp. 523–557.

Austin, B., and Austin, D.A. (1999) Bacterial Fish Pathogens:Diseases of Farmed and Wild Fish. Chichester: Springer-Praxis.

Bassler, B.L., Wright, M., Showalter, R.E., and Silverman,M.R. (1993) Intercellular signalling in Vibrio harveyi:sequence and function of genes regulating expression ofluminescence. Mol Microbiol 9: 773–786.

Bassler, B.L., Wright, M., and Silverman, M.R. (1994a) Mul-tiple signaling systems controlling expression of lumines-cence in Vibrio harveyi: sequence and function of genesencoding a second sensory pathway. Mol Microbiol 13:273–286.

Bassler, B.L., Wright, M., and Silverman, M.R. (1994b)Sequence and function of LuxO, a negative regulator ofluminescence in Vibrio harveyi. Mol Microbiol 12: 403–412.

Burbulys, D.K., Trach, A., and Hoch, J.A. (1991) The initiationof sporulation in Bacillus subtilis is controlled by a multi-component phosphorelay. Cell 64: 545–552.

Chang, W.-T., Thomason, P.A., Gross, J.D., and Newell, P.C.(1998) Evidence that the RdeA protein is a component ofa multistep phosphorelay modulating rate of developmentin Dictyostelium. EMBO J 17: 2809–2816.

Chatterjee, J., Miyamoto, C.M., and Meighen, E.A. (1996)Autoregulation of luxR: the Vibrio harveyi lux-operon acti-vator functions as a repressor. Mol Microbiol 20: 415–425.

Chen, X., Schauder, S., Potier, N., Van Dorsselaer, A.,Pelczer, I., Bassler, B.L., and Hughson, F.M. (2002) Struc-tural identification of a bacterial quorum-sensing signalcontaining boron. Nature 415: 545–549.

Croxatto, A., Chalker, V.J., Lauritz, J., Jass, J., Hardman, A.,Williams, P., et al. (2002) VanT, a homologue of Vibrioharveyi LuxR, regulates serine, metalloprotease, pigment,and biofilm production in Vibrio anguillarum. J Bacteriol184: 1617–1629.

Devereux, J., Haeberli, P., and Smithies, O. (1984) A com-prehensive set of sequence analysis programs for the VAX.Nucleic Acids Res 12: 387–395.

Dunny, G.M., and Winans, S.C. (1999) Cell–cell Signaling inBacteria. Washington, DC: American Society for Microbi-ology Press.

Freeman, J.A., and Bassler, B.L. (1999a) Sequence andfunction of LuxU: a two-component phosphorelay proteinthat regulates quorum sensing in Vibrio harveyi. J Bacteriol181: 899–906.

Freeman, J.A., and Bassler, B.L. (1999b) A genetic analysisof the function of LuxO, a two-component response regu-lator involved in quorum sensing in Vibrio harveyi. MolMicrobiol 31: 665–677.

Georgellis, D., Kwon, O., DeWulf, P., and Lin, W.C.C. (1998)Signal decay through a reverse phosphorelay in the Arctwo-component signal transduction system. J Biol Chem273: 32864–32869.

Heidelberg, J.F., Eisen, J.A., Nelson, W.C., Clayton, R.A.,Gwinn, M.L., Dodson, R.J., et al. (2000) DNA sequence ofboth chromosomes of the cholera pathogen Vibrio chol-erae. Nature 406: 477–484.

Hoff, K.A. (1989) Survival of Vibrio anguillarum and Vibriosalmonicida at different salinities. Appl Environ Microbiol55: 1775–1786.

Janiak-Spens, F., Sparling, D.P., and West, A.H. (2000)Novel role for an HPt domain in stabilizing the phosphory-lated state of a response regulator domain. J Bacteriol 182:6673–6678.

Lazazzera, B.A. (2000) Quorum sensing and starvation: sig-nals for entry into stationary phase. Curr Opin Microbiol 3:177–182.

Li, S., Ault, A., Malone, C.L., Raitt, D., Dean, S., Johnston,L.H., et al. (1998) The yeast histine protein kinase, Sln1p,mediates phosphotransfer to two response regulators,Ssk1p and Skn7p. EMBO J 17: 6952–6962.

Lilly, B.N., and Bassler, B.L. (2000) Regulation of quorumsensing in Vibrio harveyi by LuxO and Sigma-54. MolMicrobiol 36: 940–954.

McGee, K., Hörstedt, P., and Milton, D. (1996) Identificationand characterization of additional flagellin genes fromVibrio anguillarum. J Bacteriol 178: 5188–5198.

Makino, K., Oshima, K., Kurokawa, K., Yokoyama, K., Uda,T., Tagomori, K., et al. (2003) Genome sequence of Vibrioparahaemolyticus: a pathogenic mechanism distinct fromthat of V. cholerae. Lancet 361: 743–749.

Martin, M., Showalter, R., and Silverman, M. (1989)Identification of a locus controlling expression ofluminescence genes in Vibrio harveyi. J Bacteriol 171:2406–2414.

Miller, J.H. (1972) Experiments in Molecular Genetics. ColdSpring Harbor, NY: Cold Spring Harbor Laboratory Press.

Miller, M.B., and Bassler, B.L. (2001) Quorum sensing inbacteria. Annu Rev Microbiol 55: 165–199.

Miller, V.L., and Mekalanos, J.J. (1988) A novel suicide vectorand its use in construction of insertion mutations: osmo-regulation of outer membrane proteins and virulence deter-minants in Vibrio cholerae requires toxR. J Bacteriol 170:2575–2583.

Milton, D.L., Norqvist, A., and Wolf-Watz, H. (1992) Cloningof a metalloprotease gene involved in the virulencemechanism of Vibrio anguillarum. J Bacteriol 174: 7235–7244.

Milton, D.L., O'Toole, R., Hörstedt, P., and Wolf-Watz, H.(1996) Flagellin A is essential for the virulence of Vibrioanguillarum. J Bacteriol 178: 1310–1319.

Milton, D.L., Chalker, V.J., Kirke, D., Hardman, A., Cámara,M., and Williams, P. (2001) The LuxM homologue, VanM,from Vibrio anguillarum directs the synthesis of N-(3-hydroxyhexanoyl)-L-homoserine lactone and N-hexanoyl-L-homoserine lactone. J Bacteriol 183: 3537–3547.

Miyamoto, C.M., Dunlap, P.V., Ruby, E.G., and Meighen,E.A. (2003) LuxO controls luxR expression in Vibrio har-veyi: evidence for a common regulatory mechanism inVibrio. Mol Microbiol 48: 537–548.

Vibrio anguillarum quorum sensing 1689

© 2004 Blackwell Publishing Ltd, Molecular Microbiology, 52, 1677–1689

Miyamoto, C., Lin, Y.-H., and Meighen, E.A. (2000) Controlof bioluminescence in Vibrio fischeri by the LuxO signalresponse regulator. Mol Microbiol 36: 594–607.

Muroga, K., Iida, M., Matsumoto, H., and Nakai, T. (1986)Detection of Vibrio anguillarum from waters. Bull Jap SocSci Fish 52: 641–647.

Norqvist, A., Hagström, Å., and Wolf-Watz, H. (1989) Protec-tion of rainbow trout against vibriosis and furunculosis bythe use of attenuated strains of Vibrio anguillarum. ApplEnviron Microbiol 55: 1400–1405.

Oppenheimer, C.H. (1962) Marine fish diseases. In Fish asFood, Vol. 2. Borgström, G. (ed.). New York: AcademicPress, pp. 541–572.

Perraud, A.-L., Weiss, V., and Gross, R. (1999) Signallingpathways in two-component phosphorelay systems.Trends Microbiol 7: 115–120.

Posas, F., Wurgler-Murphy, S.M., Maeda, T., Witten, E.A.,Thai, T.C., and Saito, H. (1996) Yeast HOG1 MAP kinasecascade is regulated by a multistep phosphorelay mecha-nism in the SLN1-YPD1-SSK1 ‘two-component’ osmosen-sor. Cell 86: 865–875.

Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989) Molecu-lar Cloning: a Laboratory Manual, 2nd edn. Cold SpringHarbor, NY: Cold Spring Harbor Laboratory Press.

Schauder, S., Shokat, K., Surette, M.G., and Bassler, B.L.(2001) The LuxS family of bacterial antoinducers: biosyn-

thesis of a novel quorum-sensing signal molecule. MolMicrobiol 41: 463–476.

Simon, R., Priefer, U., and Pühler, A. (1983) A broad hostrange mobilization system for in vivo genetic engineering:transposon mutagenesis in gram negative bacteria. Bio/Technology 1: 787–796.

Studier, F.W., and Moffatt, B.A. (1986) Use of bacteriophageT7 RNA polymerase to direct selective high-level expres-sion of cloned genes. J Mol Biol 189: 113.

Surette, M.G., Miller, M.B., and Bassler, B.L. (1999) Quorumsensing in Escherichia coli, Salmonella typhimurium, andVibrio harveyi: a new family of genes responsible for auto-inducer production. Proc Natl Acad Sci USA 96: 1639–1644.

West, P.A., Lee, J.V., and Bryant, T.N. (1983) A numericaltaxonomic study of species of Vibrio isolated from theaquatic environment and birds in Kent, England. J ApplBacteriol 55: 263–282.

Withers, H., Swift, S., and Williams, P. (2001) Quorum sens-ing as an integral component of gene regulatory networksin Gram-negative bacteria. Curr Opin Microbiol 4: 186–193.

Zhu, J., Miller, M.B., Vance, R.E., Dziejman, M., Bassler,B.L., and Mekalanos, J.J. (2002) Quorum-sensing regula-tors control virulence gene expression in Vibrio cholerae.Proc Natl Acad Sci USA 99: 3129–3134.