Cell, Vol. 118, 69–82, July 9, 2004, Copyright 2004 by Cell Press
The Small RNA Chaperone Hfq and
Multiple Small RNAs Control Quorum Sensing
in Vibrio harveyi and Vibrio cholerae
those for bioluminescence (luciferase) (Bassler et al.,
1993, 1994a; Cao and Meighen, 1989; Chen et al.,
2002b), siderophore production (Lilley and Bassler,
2000), colony morphology, metalloprotease production
(Mok et al., 2003), and type III secretion (Henke and
In V. harveyi, AI-1 and AI-2 are produced by the syn-
Surette et al., 1999). LuxN detects AI-1, and LuxPQ de-
tects AI-2 (Figure 1A) (Bassler et al., 1993, 1994a; Chen
et al., 2002b; Freeman et al., 2000). LuxN and LuxQ are
proteins. LuxP, which binds AI-2 in the periplasm, is
required with LuxQ for the response to AI-2 (Bassler et
al., 1994a; Chen et al., 2002b). Sensory information from
both systems converges at the phosphorelay protein
LuxU, and LuxU transmits the signal to the response
regulator LuxO (Bassler et al., 1994b; Freeman and
Bassler, 1999, 2000). A transcriptional activator called
LuxR is also required for expression of lux and other
quorum sensing-controlled genes (Henke and Bassler,
2004; Martin et al., 1989; Miyamoto et al., 1994; Sho-
walter et al., 1990).
The human pathogen Vibrio cholerae possesses quo-
rum-sensing systems analogous to the two described
above for V. harveyi (Miller et al., 2002). The V. cholerae
autoinducers CAI-1 and AI-2 are synthesized by CqsA
and LuxS and detected by CqsS and LuxPQ, respec-
tively (Figure 1B). V. cholerae has an additional system
(System 3) that remains to be identified (Miller et al.,
2002). Sensory information from all three systems con-
verges at LuxO. The V. cholerae LuxR homolog is called
HapR (Joblingand Holmes,1997). Quorumsensing con-
2003; Zhu et al., 2002).
The V. harveyi and V. cholerae quorum-sensing cir-
cuits operate similarly (Miller et al., 2002). At low cell
density,i.e., intheabsenceof autoinducers,thesensors
are present, the sensors act as phosphatases. Phos-
phate flow through the circuit is reversed, resulting in
dephosphorylation and inactivation of LuxO (Freeman
and Bassler, 1999, 2000; Freeman et al., 2000). Under
this condition, the transcriptional regulators LuxR in
V. harveyi and HapR in V.cholerae bind the lux promoter
and activate transcription (Figure 1).
LuxO-P-mediated repression of lux is indirect (Lilley
and Bassler, 2000). LuxO is homologous to members of
the NtrC family of response regulators, which can act
either as transcriptional activators or repressors. Those
that are activators require the alternative sigma factor
?54for function, while those that are repressors do not
(Benson et al., 1994; North et al., 1996; Reitzer and Ma-
ton, 1997). LuxO is a member of the activator class of
NtrC homologs. We have suggested that, at low cell
Derrick H. Lenz,1,3Kenny C. Mok,1,3
Brendan N. Lilley,1,4Rahul V. Kulkarni,2
Ned S. Wingreen,1,2and Bonnie L. Bassler1,*
1Department of Molecular Biology
Princeton, New Jersey 08544
2NEC Laboratories America, Inc.
4 Independence Way
Princeton, New Jersey 08540
lular signal molecules called autoinducers. This pro-
cess allows community-wide synchronization of gene
expression. A screen for additional components of the
cuits revealed the protein Hfq. Hfq mediates interac-
tions between small, regulatory RNAs (sRNAs) and
specific messenger RNA (mRNA) targets. These inter-
actions typically alter the stability of the target tran-
of the mRNA encoding the quorum-sensing master
regulators LuxR (V. harveyi) and HapR (V. cholerae),
ics approach to identify putative sRNAs, we identified
deletion of all four sRNAs is required to stabilize hapR
mRNA. We propose that Hfq, together with these
sRNAs, creates an ultrasensitive regulatory switch
that controls the critical transition into the high cell
density, quorum-sensing mode.
Quorum sensing is a process of cell-to-cell communica-
tion that bacteria use to assess their population density
in order to coordinate the gene expression of the com-
munity (Miller and Bassler, 2001). Quorum sensing re-
quires the production, secretion, and detection of extra-
cellular signal molecules termed autoinducers. Diverse
behaviors are controlled by quorum sensing, but, typi-
cally, these behaviors are ones that would be ineffective
if only a small group of cells carried them out. Often,
bacteria produce and detect multiple autoinducers,
erle and Bassler, 2003; Fuqua et al., 2001; Xavier and
The marine bacterium Vibrio harveyi produces and
detects two autoinducers, AI-1 and AI-2, and these sig-
nals control the expression of multiple genes, including
3These authors contributed equally to this work.
4Present Address: Program in Biological and Biomedical Sciences,
Harvard Medical School, 200 Longwood Avenue, Boston, Massa-
Figure 1. Models of the V. harveyi and V. cholerae Quorum-Sensing Circuits
(A) Two quorum-sensing systems function in parallel to regulate gene expression in V. harveyi. Pentagons and triangles represent AI-1 and
(B) Three quorum-sensing systems function in parallel to regulate gene expression in V. cholerae. The functions making up the third circuit
(denoted System 3) remain to be identified. Diamonds and triangles represent CAI-1 and AI-2, respectively. In both circuits, phosphate flows
in the direction indicated by the arrows at low cell density and in the opposite direction at high cell density.
density, LuxO-P activates the expression of a repressor
that controls the downstream target genes. We show
that multiple, redundantsmall regulatory RNAs (sRNAs),
together with the sRNA binding protein Hfq, fulfill this
repressor role. Specifically, at low cell density, the Hfq-
sRNA repressor complexes destabilize the V. harveyi
luxR and V. cholerae hapR mRNAs (Figure 1).
tained overlapping regions of DNA, suggesting that a
single locus was responsible. One cosmid, pBNL2014,
siblefor luxrepression.The regionidentified wascloned
and sequenced and found to contain the gene hfq (Fig-
ure 2A). The V. harveyi hfq gene displays high homology
to hfq genes from other vibrio species, including Vibrio
parahaemolyticus, V. cholerae, and Vibrio vulnificus,
with 100%, 95%, and 94% identity, respectively (NCBI
A Genetic Screen for the V. harveyi
Quorum-Sensing Repressor Reveals Hfq
To identify the putative repressor that acts downstream
tion alters the site of phosphorylation, and “locks” the
LuxO D47E protein into a state mimicking LuxO-P.
LuxO-P activates transcription of the putative repressor
of lux at low cell density. Consistent with the model,
V. harveyi luxO D47E strains are dark, presumably due
genized a V. harveyi luxO D47E strain with the transpo-
son Mini-MulacZ and screened for colonies that had
acquired a bright phenotype, indicating that they had
Of the 40,000 transposon insertion mutants gener-
transposon insertions in either luxO or rpoN, the gene
encoding ?54. Three bright mutants did not harbor muta-
tions in either of these genes. A V. harveyi genomic
cosmidlibrary wasintroducedintoone ofthesemutants
(BNL211) and screened for restoration of the dark phe-
notype. All cosmids conferring a dark phenotype con-
Hfq Is Required for Quorum-Sensing Repression
To verify that Hfq has a role in quorum-sensing regula-
tion in V. harveyi, hfq null mutations were introduced
onto the chromosomes of the V. harveyi wild-type and
luxO D47E strains. The Lux phenotypes of the single hfq
and double luxO D47E, hfq mutants were examined and
V. harveyi strains (Figure 2B). Wild-type V. harveyi dis-
plays typical quorum-sensing behavior (squares): it is
very bright immediately following dilution into fresh me-
dium, but, early in the assay, luminescence decreases
precipitously (?1000-fold) due to dilution of the autoin-
ducers to a level below that required for activation of
tion. Light production commences and increases 1000-
fold, ultimately reaching the predilution level. The luxO
null strain (diamonds) is constitutively bright because,
in the absence of LuxO, no lux repressor is produced.
The hfq mutant (circles) has a phenotype indistinguish-
able from the luxO mutant, demonstrating that Hfq is
required for repression of lux expression at low cell
Hfq-sRNA Regulation of Quorum Sensing in Vibrios
Figure 2. Hfq Is Required for Quorum-Sensing Repression in V. harveyi and V. cholerae
(A) The hfq locus in V. harveyi. miaA and hflC were not fully sequenced (unsequenced regions are denoted by light-colored shading). (B)
Bioluminescence assays for V. harveyi strains are: BB120 (WT, squares), JAF78 (?luxO::cmr, diamonds), JAF548 (luxO D47E, open triangles),
BNL258 (hfq::Tn5lacZ, circles), and BNL211 (luxO D47E, hfq::Mini-MulacZ, closed triangles). Relative light units for V. harveyi are defined as
counts min?1ml?1? 103/cfu ml?1. (C) Bioluminescence assays for V. cholerae strains are: MM227 (WT, squares), MM349 (?luxO, diamonds),
BH48 (luxO D47E, open triangles), DL2078 (?hfq, circles), and DL2378 (luxO D47E, ?hfq, closed triangles). Relative light units for V. cholerae
are defined as counts min?1ml?1/OD600nm. In (B) and (C), the dotted lines represent the limit of detection for light. (D) V. cholerae strains
analyzed for TcpA production by Western blot are: C6706str2 (WT), MM307 (?luxO), BH38 (luxO D47E), MM194 (?hapR), DL2066 (?hfq), DL2146
(luxO D47E, ?hfq), and DL2607 (?hapR, ?hfq).
densities. The luxO D47E, hfq double mutant (closed
triangles) is also constitutively bright, showing that re-
pression of lux by Hfq occurs downstream of LuxO.
We confirmed that Hfq has the identical role in V.
cholerae that it has in V. harveyi quorum sensing by
measuring density-dependent light production from V.
cholerae strains carrying V. harveyi lux. The Lux pheno-
types of the wild-type (squares), luxO (diamonds), luxO
D47E (open triangles), hfq (circles), and luxO D47E, hfq
(closed triangles) single and double V. cholerae mutants
mimic the corresponding V. harveyi mutant phenotypes
1991). TcpA is present in the wild-type strain because,
leading to TcpA production, which enables its detection
at high cell density (Zhu et al., 2002). No TcpA is ob-
served in the luxO strain, because LuxO-P is required
at low cell density to initiate the production of the TcpA
observed in the wild-type. In contrast, high levels of
TcpA are observed in the luxO D47E strain. Similarly,
the hapR mutant that is also locked in low cell density
mode produces high levels of TcpA. Importantly, low
TcpA is detected in the hfq mutant, demonstrating that
Hfq is indeed required for virulence-factor expression.
hfq mutants demonstrates that Hfq acts downstream of
LuxO and upstream of HapR in the quorum-sensing
Hfq Is Required for Regulation of Virulence Gene
Expression in V. cholerae
expression at low cell density (Figure 1B). To show that
the quorum-sensing activity of Hfq is not restricted to
the nonnative lux target in V. cholerae, we measured
TcpA (the major subunit of the toxin-coregulated pilus)
production using Western blots (Figure 2D) (Taylor,
Predictions for Hfq Involvement
in Quorum Sensing
tion between the sRNAs and their target mRNAs (Masse
et al., 2003b; Valentin-Hansen et al., 2004). These Hfq-
sRNA complexesalter thestability/translation ofthe tar-
get mRNAs. Our finding that Hfq is required for quorum-
sensing repression in V. harveyi and V. cholerae led
us to two predictions: first, quorum-sensing repression
occurs posttranscriptionally, and, second, there must
and ?54do not control transcription of hfq (data not
shown) led us to predict that, at low cell density, the
LuxO-P-?54complex activates the transcription of the
gene(s) encoding the sRNA(s). The remainder of the ex-
periments presented here test these predictions.
is fused to the predicted site of transcription initiation
(?1 site) (Figure 4C). These results suggest that LuxO-P
regulation of hapR is posttranscriptional and show that
the region between the predicted ?1 site and the hapR
coding region is required for Hfq control.
Identification of sRNAs Involved in Quorum-
Sensing using Bioinformatics
Our findings point toward a LuxO-?54-regulated sRNA
in the quorum-sensing signal-transduction circuits of V.
harveyi and V. cholerae. Because these small genes are
very difficult to identify by traditional genetic ap-
proaches, we developed a method to scan the V. chol-
erae genome for candidate sRNA loci. We could not
perform this analysis in V. harveyi, because its genome
has not been sequenced. We used the following param-
eters in our analysis: (1) expression of the sRNA is acti-
vated by the LuxO-P-?54complex, thus the upstream
region of the locus must contain a ?54binding site; (2)
most sRNAs identified to date have Rho-independent
terminators, and we assumed this to be the case for the
putative sRNAs in our analysis (Argaman et al., 2001;
Chen et al., 2002a; Wassarman et al., 2001); (3) most
our search to regions between annotated genes (Arga-
man et al., 2001; Wassarman et al., 2001); and (4) the
sRNA must be conserved in V. cholerae, V. parahaemo-
lyticus, and V. vulnificus. The completed genome se-
quences of these vibrios show that they possess homo-
logs of luxR/hapR, luxO, luxU, and hfq, suggesting that
they have a conserved quorum-sensing regulatory
mechanism (Chen et al., 2003; Heidelberg et al., 2000;
Makino et al., 2003).
Using PATSER, we scanned the V. cholerae genome
for potential ?54binding sites with a weight matrix con-
structed from a compiled set of ?180 ?54binding sites
from multiple bacterial species (Barrios et al., 1999;
Dombrecht et al., 2002; Hertz and Stormo, 1999; van
Helden, 2003). We considered all hits above a cutoff
score chosen to include all binding sites upstream of
genes in V. cholerae that are known to be regulated by
?54. In a parallel procedure, the upstream regions of the
known V. cholerae ?54-regulated genes were extracted,
and we used the program CONSENSUS to search for a
16 bp motif in these sequences (Hertz and Stormo,
1999). The motif so obtained corresponded perfectly
to the known ?54binding sites in V. cholerae, with the
consensus sequence 5?-TGGCAC-N5-TTGCA/T-3?. The
aligned set of binding sites was used to construct a
weight matrix for ?54sites specific to V. cholerae. Be-
causethe weightmatricesobtained bythese twoproce-
dures were quite similar, the final result did not depend
on the weight matrix used to scan the genome.
?54binding sites in intergenic regions. We examined
these regions for conservation across the specified vib-
rio genomes and for the presence of Rho-independent
terminators. These constraints narrowed the search to
four intergenic regions. The sequences and alignment
of these four regions are shown in Figure 5A, along with
the corresponding regions from V. parahaemolyticus
and V. vulnificus. These four loci are highly homologous
Hfq Affects the Stability of luxR/hapR mRNA
LuxR and HapR appear to be the master regulators of
their respective quorum-sensing regulons (Henke and
Bassler, 2004; Zhu et al., 2002). Knowing this suggested
two possible methods by which Hfq could repress quo-
rum sensing-controlled gene expression. First, Hfq
could directly act on the mRNA encoding each of the
known target genes of the quorum-sensing regulons.
Second, Hfq could act on the mRNAs encoding LuxR
in V. harveyi and HapR in V. cholerae. We reasoned
this latter possibility was more likely, because, in this
scenario, Hfq need only act on a single mRNA in each
Northern blots were used to determine the effect of
hfq mutations on luxR and hapR mRNA stability in V.
harveyi and V. cholerae. Rifampicin was added to cul-
was performed in luxO D47E strains to assess the fate
of the luxR and hapR transcripts at low cell density.
Figure 3A shows that, under these conditions, both the
luxR and hapR transcripts disappear immediately fol-
lowing termination of transcription (panels labeled luxO
D47E). However, in the luxO D47E, hfq double mutants,
the transcripts show significantly increased longevity
(Figure 3A, panels labeled luxO D47E, hfq). The control
shows that mutation of hfq has no effect on the stability
of rpsL mRNA (Figure 3A, four lower panels). Western
blots show that the increased stability of the luxR and
hapR mRNAs in the hfq mutants leads to increased
levels of the LuxR and HapR proteins (Figure 3B). These
results demonstrate that, at low cell density, Hfq desta-
bilizes the luxR and hapR mRNA in V. harveyi and V.
cholerae, respectively, which leads to reduced LuxR
and HapR protein in the cells.
LuxO-P Regulation of hapR Is Posttranscriptional
and Requires Hfq
Previous analyseshave suggestedthat LuxO-Pcontrols
transcription of hapR in V. cholerae (Zhu et al., 2002).
Our inability to detect hapR mRNA in the Northern blots
at time zero does not allow us to distinguish between
transcriptional and posttranscriptional regulation of
hapR. We constructed chromosomal hapR-lacZ tran-
scriptional, translational, and promoter fusions and
measured their activities in the V. cholerae wild-type,
luxO D47E, hfq, and luxO D47E, hfq strains. The tran-
scriptional and translational fusionsare repressed in the
luxO D47E strain, and repression requires Hfq (Figures
4A and 4B, respectively). In contrast, LuxO D47E does
Hfq-sRNA Regulation of Quorum Sensing in Vibrios
Figure 3. Hfq Regulates the Expression of luxR and hapR Posttranscriptionally
Non-steady-state Northern blots were used to analyze luxR/hapR transcript stability in the following: (A) V. harveyi JAF548 (luxO D47E) and
BNL211 (luxO D47E, hfq::Mini-MulacZ); and (B) V. cholerae BH38 (luxO D47E) and DL2146 (luxO D47E, ?hfq). (C) Western blots on lysates of
V. harveyi BB120 (WT), JAF548 (luxO D47E), BNL258 (hfq::Tn5lacZ), BNL211 (luxO D47E, hfq::Mini-MulacZ), and (D) V. cholerae C6706str2
(WT), BH38 (luxO D47E), DL2066 (?hfq), DL2146 (luxO D47E, ?hfq) measured LuxR and HapR protein, respectively.
site and a terminator, suggesting that these elements
are independently transcribed loci. We name these loci
qrr1, qrr2, qrr3, and qrr4 (for quorum regulatory rna
1–4). Interestingly, one of the sRNA loci, qrr1, is located
immediately upstream of luxO. In V. parahaemolyticus
and V. vulnificus, a fifth putative sRNA locus was identi-
fied that fulfills all of our search criteria (denoted qrr5).
Using theRNAFOLD program, we foundthe predicted
secondary structures of the candidate sRNAs (Figure
5B) (Hofacker, 2003). The predicted structures of Qrr2
and Qrr3 are very similar, as is the structure of Qrr4 if
the three nucleotide pairs joining the two center loops
are melted (and note these three nucleotide pairs are
not conserved across species). Thus, only the predicted
structure of Qrr1 is obviously distinct. Qrr2 and Qrr3
show a site similar to the proposed Hfq binding site,
which is an 8–12 nucleotide AU-rich region adjacent to
stem loops (Moll et al., 2003). The composition of the
conserved, supporting the folding predictions. Many
small regulatory RNAs act by base pairing to comple-
mentary regions in the 5? untranslated region of the
mRNA. Using the program LALIGN, which finds the best
local alignment of the input sequences, we aligned the
complement of the hapR untranslated upstream region
with all four V. cholerae sRNAs and the luxR upstream
region with V. harveyi Qrr1 (Figures 5C and 5D, respec-
tively). Interestingly, the region we identify as being po-
tentially involved in the complementary base pairing is
absolutely conserved among all four sRNA candidates,
with a single base difference in V. harveyi qrr1 (Figure
5A). The highly conserved region overlaps the hapR and
luxR putative ribosome binding sites (AAGGAUAU for
hapR and AAGGAAAA for luxR). Finally, analysis of the
upstream regions of hapR and luxR and their orthologs
tive region of interaction with the sRNAs is strongly con-
We propose that the four putative sRNA loci in
V. cholerae, and possibly five in V. parahaemolyticus,
V. vulnificus, and V. harveyi (see Discussion), are regu-
lated by LuxO-P together with ?54. Analysis of the up-
stream regions of the candidate sRNA loci shows a
highly conserved region upstream of the ?54site. This
site has dyad symmetry, and the consensus sequence
is TTGCAW3TGCAA (where W corresponds to A/T). We
hypothesize that this region could be important for
LuxO-P-?54Controls the Expression
of the sRNA Loci
To ascertain whether any of the candidate sRNA loci is
blot analysis was used to quantify transcript levels. The
DNA encoding the putative sRNAs was amplified by
PCR and used to probe identical Northern blots con-
Figure 4. LuxO-P and Hfq Regulate hapR
?-galactosidase activity of (A) hapR-lacZ
transcription in V. cholerae DL2106 (WT,
?lacZ), DL2099 (luxO D47E, ?lacZ), DL2523
(?hfq, ?lacZ), and DL2441 (luxO D47E, ?hfq,
?lacZ). (B) hapR-lacZ translation in V. chol-
?lacZ), DL2531 (?hfq, ?lacZ), and DL2533
(luxO D47E, ?hfq, ?lacZ). (C) hapR-lacZ pro-
moter activity in V. cholerae DL2748 (WT,
?lacZ), DL2771 (luxO D47E, ?lacZ), DL2703
(?hfq, ?lacZ), and DL2698 (luxO D47E,
taining RNA isolated from low cell density cultures of
Figure 6A (hapR?panel) shows that, surprisingly, only
Qrr4 is obviously regulated by LuxO-P-?54. A very small
amount of this sRNA is detected in wild-type cells,
whereas high levels are present in the luxO D47E strain.
Importantly, Qrr4 is undetectable in both the luxO and
rpoN mutants, consistent with a requirement for both
encoding Qrr4 at low cell density. We did not detect the
other three V. cholerae sRNAs in this analysis.
Recently, it was shown in E. coli that, upon binding
its mRNA target, the sRNA RyhB is degraded along
with the target by RNaseE (Masse et al., 2003a). In the
absence of the mRNA targets, increased stability of
RyhB is observed (see Discussion). We wondered if the
Qrr sRNAs were being degraded along with the hapR
target mRNA. To test this, we deleted hapR in the wild-
type, luxO, luxO D47E, and rpoN V. cholerae strains,
prepared RNA, performed Northern blots, and probed
them for all four sRNAs (Figure 6A, hapR?panel). In the
absence of hapR mRNA, an increase in the level of Qrr4
is observed in the wild-type and luxO D47E strains. We
also detect minor amounts of Qrr2 and Qrr3 in the luxO
Hfq-sRNA Regulation of Quorum Sensing in Vibrios
Figure 5. Bioinformatic Analysis of the Qrr sRNAs in V. cholerae, V. parahaemolyticus, V. vulnificus, and V. harveyi
(A) Multiple sequence alignmentof the qrr genes encoding thesRNAs identified in V. cholerae, V. parahaemolyticus,V. vulnificus, and V. harveyi.
Annotations for the genes flanking each sRNA are given in the brackets. Numbering of sRNAs is based on orthology of flanking genes.
Nucleotides in black indicate perfect alignment. The putative ?54binding site is marked as ?12 and ?24, the predicted start of transcription
is labeled as ?1, and the terminator is noted by the line over the sequence. (B) Lowest-energy secondary-structural predictions for the Qrr
sRNAs identified in V. cholerae. Bold typeface indicates the regions conserved across all sRNAs in V. cholerae, V. parahaemolyticus, and
V. vulnificus. (C) Alignment of the complement of the hapR UTR with a portion of the Qrr sRNAs identified in V. cholerae. (D) Alignment of the
complement of the luxR UTR with a portion of sRNA Qrr1 identified in V. harveyi. In (C) and (D), the translational start site (Start), the ribosome
binding site (RBS), and the transcriptional start site (?1) are indicated. (A), (C), and (D) were produced using CLUSTALW (Thompson et al.,
1994) and ESPript (Gouet et al., 1999).
Figure 6. Regulation of Expression of the sRNAs by Quorum Sensing
RNA isolated from (A) V. cholerae C6706str2 (WT), MM307 (?luxO), BH38 (luxO D47E), BH76 (?rpoN) was probed for sRNAs Qrr1, Qrr2, Qrr3,
and Qrr4, and V. cholerae rpsL is shown as the loading control. (B) RNA isolated from V. harveyi BB120 (WT), BB721 (luxO::Tn5lacZ), JAF548
(luxO D47E), and BNL240 (rpoN::cmr) was probed for sRNA Qrr1 with a probe made against V. harveyi qrr1 and for sRNA Qrr4 with a probe
made against V. cholerae qrr4. V. harveyi rpsL is shown as the loading control. (C) Single time point RLU for V. cholerae strains DL3212 (luxO)
and DL3213 (luxO D47E) containing the qrr1-lux transcriptional fusion in trans.
D47E strain, showing that they are indeed regulated by
LuxO-P and that their levels increase in the absence of
hapR mRNA. However, we could not detect the sRNA
We know from our sequencing that qrr1 resides up-
stream of luxO in V. harveyi; however, we could not
detect this sRNA by Northern blot (Figure 6B). Because
the genome of V. harveyi has not been sequenced, we
do notknow whether genes correspondingto qrr2, qrr3,
are present. We successfully detected Qrr4 from V. har-
veyi by probing total RNA with the DNA probe made
from the V. cholerae qrr4 PCR product, showing that
such an sRNA exists in V. harveyi, and its expression is
induced by LuxO D47E (Figure 6B).
The inability to detect V. cholerae and V. harveyi Qrr1
could be a consequence of extremely low expression
of qrr1 or instability of the qrr1 transcript, coupled with
the insensitivity of the Northern blot procedure. Alterna-
tively, expression of qrr1 might not be controlled by
LuxO-P; however, this seemed unlikely, based on the
bioinformatics analysis. To eliminate this latter possibil-
Hfq-sRNA Regulation of Quorum Sensing in Vibrios
Figure 7. Simultaneous Deletion of the Four sRNAs Is Required to Affect Quorum-Sensing in V. cholerae
(A) Bioluminescence assays were performed on V. cholerae: MM227 (WT, open squares), MM349 (?luxO, open diamonds), DL2998 (?qrr2,
?qrr3, ?qrr4, closed squares), DL2996 (?qrr1, ?qrr3, ?qrr4, closed diamonds), DL2955 (?qrr1, ?qrr2, ?qrr4, closed triangles), DL2997 (?qrr1,
?qrr2, ?qrr3, closed circles), DL2956 (?qrr1, ?qrr2, ?qrr3, ?qrr4, open circles).
(B) Single time point RLU for V. cholerae strains MM227 (WT), MM349 (?luxO), BH48 (luxO D47E), DL2956 (?qrr1, ?qrr2, ?qrr3, ?qrr4), and
DL3024 (luxO D47E, ?qrr1, ?qrr2, ?qrr3, ?qrr4). Western blots probed for HapR and TcpA from V. cholerae strains C6706str2 (WT), MM307
(?luxO), BH38 (luxO D47E), DL2953 (?qrr1, ?qrr2, ?qrr3, ?qrr4), and DL3020 (luxO D47E, ?qrr1, ?qrr2, ?qrr3, ?qrr4).
ity, we constructed a qrr1 transcriptional reporter by
fusing the upstream region of V. cholerae qrr1 to the
luxCDABE (luciferase) operon and tested whether this
construct was regulated by LuxO-P. We measured the
expression of the qrr1-lux fusion in the luxO null and
luxO D47E V. cholerae strains. The presence of LuxO
D47E causes a 220-fold increase in expression from the
qrr1 promoter, verifying that, indeed, qrr1 is regulated
by LuxO-P (Figure 6C). We conclude that transcription
of all four qrr genes is regulated by LuxO-P.
sensing repression in V. cholerae, it follows that overex-
pression of the sRNAs should result in constitutive re-
pression. We tested this by overexpressing V. cholerae
qrr1 in various V. cholerae and V. harveyi strains and
examining the impact on light production. Compared
to the vector-alone control, when V. cholerae qrr1 is
overexpressed in V. cholerae, light production is re-
duced to 21% in the wild-type, 10% in the luxO null
strain, and 1% in the quadruple sRNA mutant (Table 1).
Overexpression of V. cholerae qrr1 in V. harveyi reduces
light production to 12% in the wild-type and to 3% in
the luxO null strain. Thus, the V. cholerae sRNA Qrr1
functions in both V. cholerae and V. harveyi to repress
be obtained with any of the other three sRNAs identified
in this work.
All of the data presented here suggest that multiple
sRNAs act downstream of LuxO-P to destabilize luxR/
Four sRNAs Are Involved in Quorum-Sensing
Repression in V. cholerae
in quorum sensing, single, double, triple, and the qua-
druple qrr sRNA deletions were constructed in V. chol-
erae. Subsequently, we measured density-dependent
light production and found that only the simultaneous
deletion of all four sRNAs affected bioluminescence ex-
pression (Figure 7A). The results are shown only for
double sRNA deletion mutants behaved similarly to the
triple mutants. Remarkably, the results show that if any
one of the four sRNAs is present, V. cholerae expresses
density-dependent bioluminescence similar to the wild-
type. However, deletion of all four sRNA genes together
results in a constitutive lux phenotype identical to the
luxO null mutant. Thus, all four sRNAs participate in
quorum-sensing repression, although any one alone is
Because deletion of the sRNAs eliminates quorum-
Table 1. Overexpression of V. cholerae qrr1 in V. cholerae
and V. harveyi
StrainPercent Light Remaining
Compared to Vector Alone
V. cholerae wt
V. cholerae ?luxO
V. cholerae 4 sRNA?
V. harveyi wt
V. harveyi ?luxO
These experiments are representative of several trials that did not
hapR mRNA and regulate quorum-sensing dependent
gene expression in V. harveyi and V. cholerae. As a final
verification of this model, we performed an epistasis
test in V. cholerae. We measured light production and
HapR and TcpA protein levels in the V. cholerae wild-
type strain, the luxO null mutant, the luxO D47E mutant,
the quadruple sRNA deletion mutant, and the luxO D47E
mutant containing the quadruple deletion of the qrr
genes (Figure 7B). Maximal light is produced and a cor-
responding high level of HapR protein is observed in
the high cell density wild-type and luxO strains. Both
light and HapR protein levels are severely reduced in
the luxO D47E strain. However, deletion of the four
sRNAs alone or in the luxO D47E background restores
maximal light production and maximal HapR protein
production, showing that the four sRNAs are required
for repression and act downstream of LuxO. Because
lux and tcpA are regulated in an opposite manner by
rocally with those of lux expression and HapR concen-
trationin thequorum-sensingmutants.Figure 7Bshows
this is the case and, importantly, that the four sRNAs
are epistatic to LuxO-P in regulation of tcpA.
gene expression at a variety of levels, including but not
limited to mRNA stability and translation (Masse et al.,
2003b). Regulation occurs via RNA-protein interactions
and RNA-RNA base pairing. Bacterial sRNAs are usually
on the order of 100 nucleotides in length, and they can
act either positively or negatively on their targets, de-
pending on the location within the message of the RNA-
RNA interaction (Masse et al., 2003b). For example, the
sRNA OxyS binds to the rpoS mRNA and prevents its
translation (Zhang et al., 1998, 2002). In contrast, the
sRNAs DsrA and RprA enhance translation of rpoS
mRNA by binding to an upstream leader region and
eliminating the formation of a particular secondary struc-
the succinate dehydrogenase (sdhCDAB) operon and fa-
cilitates degradation of the transcript (Masse et al.,
2003a; Masse and Gottesman, 2002). In many known
cases, the sRNA chaperone protein Hfq is required to
enhance the interaction between sRNAs and their target
mRNAs (Moller et al., 2002; Zhang et al., 2002). In addi-
tion, many newly identified sRNAs that have no known
regulatory function can bind to Hfq directly, suggesting
that Hfq is an RNA chaperone for a large number of
sRNAs (Wassarman et al., 2001; Zhang et al., 2003). Hfq
has long been known to be a global regulator of gene
expression, and it is now believed that this property of
Hfq stems from its interaction with numerous regula-
as potential partners to Hfq in quorum-sensing regula-
tion in V. cholerae, (Figures 5–7). While experimental
verification is required, bioinformatics suggests five
sRNAs are involved in the analogous V. parahaemolyti-
cus and V. vulnificus quorum-sensing circuits (Figure 5).
lyticus than to V. cholerae (Rowe-Magnus et al., 2001),
we suspect that five sRNAs are likely to partner with
Hfq in V. harveyi quorum-sensing regulation.
Interestingly, the gene encoding sRNA Qrr1 is located
immediately upstream of the luxOU operon in all the
analyzed vibrios (Figure 5). Conservation of gene order
across species is generally indicative of functions that
act in the same process (Dandekar et al., 1998). We
hypothesize that the qrr1-luxOU locus represents an
ancient evolutionary unit. This sRNA is noticeably differ-
ent from the other three in that it is predicted to fold
into a stem-loop structure with a loop that is much
smaller than those predicted for the other three sRNAs
(Figure 5B). This is partially accounted for by the fact
thatQrr1 ismissingseveral stretchesof nucleotidesthat
are conserved in Qrr2, Qrr3, and Qrr4 (see Figure 5A).
Preliminary phylogenetic analysis indicates that the four
sRNAs are paralogs, i.e., they derive from duplications
ably the sRNA linked to luxOU) followed by speciation
into the family of vibrios. This evolutionary scenario
leaves open the question of why multiple sRNAs are
used to control quorum sensing in these organisms.
Extensive redundancy exists among the four sRNAs:
the simultaneous inactivation of all four is necessary
to eliminate Hfq-mediated quorum-sensing repression
(Figure 7A), and, consistent with this, overexpression of
only one sRNA is sufficient for repression (Table 1). In
Both V. harveyi and V. cholerae use quorum sensing to
regulate gene expression in response to changes in cell
density. Many of the regulatory components making up
these vibrio communication systems have been identi-
fied and their roles characterized (Federle and Bassler,
2003; Xavier and Bassler, 2003). Genetic analyses have
shown that one such component, LuxO, is phosphory-
lated at low cell density, and, in this form, interacts with
the alternative sigma factor ?54to activate a putative
downstream repressor (Lilley and Bassler, 2000). Here
we show that, in V. cholerae, the repressor is the sRNA
chaperone Hfq and four sRNAs. LuxO-P, together with
?54, activates the expression of the loci encoding all four
sRNAs, and repression occurs via Hfq-sRNA-mediated
destabilization of the hapR mRNA transcript (Figure 1).
Surprisingly, while four sRNAs are involved, any one is
sufficient for complete quorum-sensing repression. In
V. harveyi, most probably five sRNAs work in conjunc-
tion with Hfq to destabilize the luxR mRNA (see below).
Small, untranslated RNA molecules have roles in con-
trolling geneexpression in bacteria(Masse etal., 2003b)
and eukaryotes (Carrington and Ambros, 2003). In eu-
by microRNAs (miRNAs). For example, in Caenorhab-
ditis elegans and Drosophila melanogaster, some un-
translated regions of mRNAs contain multiple sites to
which miRNAs can bind. In other instances, one miRNA
can control the expression of another miRNA. Finally,
in D. melanogaster, distinct miRNAs can affect a single
process (such as apoptosis), but each miRNA acts on
a different function in the pathway (Carrington and
Ambros, 2003). In bacteria, sRNAs regulate a variety of
cellular processes, including carbon storage and utiliza-
tion (Romeo, 1998), response to iron limitation (Masse
et al., 2003a; Masse and Gottesman, 2002), response
to oxidative stress (Zhang et al., 1998), and transition
Hfq-sRNA Regulation of Quorum Sensing in Vibrios
Figure 8. Mutual Destruction of sRNA and Target mRNA Produces an Ultrasensitive Response to the Rate of sRNA Synthesis.
If the rate of sRNA synthesis, kx, drops below the rate of target mRNA synthesis, ky, the steady-state pool of target message rises abruptly.
In the quorum-sensing circuit, this implies that an ultrasensitive increase in hapR/luxR mRNA levels occurs with decreasing levels of LuxO-P
as cell density increases. The curves are generated from Equation 7 of Elf et al. (2003), which describes the production of two chemical
species (the sRNA and the mRNA) that undergo mutual destruction with a second-order rate constant kmd(? vmax/KxKy) and also undergo
intrinsic, first-order degradation at a rate ?. In the curves shown, we have set the one adjustable parameter kmdky/?2to equal 5 ? 107, which
is in the regime where degradation of the minority species is primarily due to mutual-destruction processes. Concentrations are given in units
principle, a single sRNA activated by LuxO-P could ac-
complish the transition between low and high cell den-
sity states. One possibility is that the presence of multi-
ple sRNAs is important to fine tune this transition. If
so, the presence of four sRNAs may allow additional
regulatory inputs (e.g., metabolic) to influence when the
transition occurs. Indeed, we have identified conserved
motifs, consistent with regulatory-factor binding sites,
in the upstream regions of the qrr2 and qrr3 genes, in
addition to the putative LuxO-P binding sites located
upstream of all four sRNAs (R.V.K., unpublished data).
We cannot detect sRNAs Qrr1, Qrr2, or Qrr3 in the
luxO D47E strain by Northernanalysis (Figure 6A, hapR?
panel). Nonetheless, their inactivation is required to
eliminate Hfq-directed quorum-sensing repression (Fig-
ure 7), so they must be present, and they definitely act
under our experimental conditions. Recently, it was re-
the degradation of the mRNA encoding SodB in E. coli
(Masse et al., 2003a). The sRNA RyhB when paired to
its target mRNA is rapidly degraded by RNaseE. Other
sRNAs were also shown to be stabilized in the absence
latory mechanism is generalizable (Masse et al., 2003a).
We found that the levels of sRNA Qrr2, Qrr3, and Qrr4
mRNA (Figure 6). This finding suggests that, first, like
qrr4, qrr2 and qrr3 are controlled by LuxO-P and, sec-
ond, that as in E. coli, the presence of the target mRNA
reduces the stability of these three sRNAs. If messages
specifying genes other than hapR are also targets of
regulation by this set of sRNAs, it might be necessary
to eliminate these mRNAs to observe increased stability
and/or abundance of sRNA Qrr1 and further increased
stability and/or abundance of sRNAs Qrr2 and Qrr3.
A transcriptional reporter fusion to the qrr1 promoter
demonstrates that it too is controlled by LuxO-P (Figure
6C). We suspect that we cannot detect this sRNA by
Northern blot, because the technique does not have
the sensitivity required to detect RNAs with very low
abundance. In one additional attempt to study Qrr1, we
hypothesized that the absence of the Qrr2–4 sRNAs
might lead to enhanced expression of the remaining qrr
gene, qrr1. However, no change in transcription of qrr1-
lux occurred in the triple mutant, suggesting that the
presence or absence of qrr2–4 does not affect the ex-
pression of qrr1 (data not shown).
Why does LuxO regulate hapR/luxR via sRNAs rather
than directly? One reason may be that the sRNAs allow
a simple “inversion” of regulatory control, so that the
activator LuxO can repress hapR/luxR. However, this
inversion could also be accomplished in other ways,
e.g., by switching the regulation pattern of hapR/luxR
with respect to its target genes. A more fundamental
motivation for control via sRNAs may be to achieve
an ultrasensitive (switchlike) response to the level of
(HI), or autoinducer bioassay (AB) broth (Bassler et al., 1994b; Free-
man and Bassler, 1999). V. cholerae strains are derivatives of El Tor
strain C6706str2 (Thelin and Taylor, 1996) and were grown at 30?C
with aeration in Luria-Bertani (LB) or SOC broth (Sambrook et al.,
1989). For studies of toxin coregulated pilus (TCP), V. cholerae was
grown at 37?C in AKI medium under AKI conditions (Iwanaga et al.,
1986). E. coli S17-1?pir (de Lorenzo and Timmis, 1994) and JM109
(Yanisch-Perron et al., 1985) were used to propagate plasmids at
37?C in LB. The following antibiotics were used: ampicillin (amp),
100?g/ml; tetracycline (tet), 10?g/ml; kanamycin (kan), 100?g/ml;
chloramphenicol (cm), 10?g/ml; and gentamicin (gent), 100?g/ml.
Streptomycin (strep) was used at 1 mg/ml and polymyxcin B (pb)
at 50 units/ml.
LuxO-P. As base pairing of an sRNA with its target mes-
sage isknown to promotedegradation of boththe sRNA
and the message, this “mutual destruction” provides an
elegant mechanism for ultrasensitivity. Specifically, as
shown in Figure 8, if the rate of synthesis of a particular
sRNA exceeds the rate of synthesis of its target mes-
in the cell, and target message levels can be reduced
to very low levels. In contrast,if the rate of synthesis of a
particular target message exceeds that of its regulatory
sRNA, then the message can accumulate (Figure 8). The
ultrasensitive mechanism described here also applies
in the case of multiple sRNAs interacting with one or
more mRNA targets (Paulsson and Ehrenberg, 2001).
The use of sRNAs to accomplish an ultrasensitive re-
sponse may be particularly apt for processes such as
quorum sensing in which an all-or-nothing response is
indicated. Similarly, this all-or-nothing requirement
could explain why sRNAs control the entry into station-
ary phase. A different kind of sRNA switch was high-
lighted by Masse et al. (2003a) for the RyhB system. In
reversible switch in time in response to a large change
in input (e.g., addition of iron to the medium). In the
quorum-sensing circuit, by contrast, the switch occurs
in response to a small change in input (e.g., LuxO-P
two discrete states (e.g., low and high cell density). How
fast this switch occurs in time will depend on the rate
of change of LuxO-P levels as well as on the rate of
accumulation and/or degradation of luxR/hapR mRNA
and LuxR/HapR protein. Interestingly, an ultrasensitive
response to LuxO-P via the rate of sRNA production
rate of the sRNAs, consistent with our hypothesis that
for fine tuning the transition between low and high cell
The decision to transition from acting alone to partici-
pating in a group activity is a critical one for bacteria.
In vibrios, sophisticated regulatory devices are located
at different positions in the quorum-sensing signal-
transduction relay to ensure that this decision occurs
under the appropriate set of circumstances and with
high fidelity. Previously, we showed that a coincidence
detector regulates entry into high cell density mode:
the simultaneous presence of multiple autoinducers is
the system and thus to initiate the critical transition from
The coincidence-detection scheme likely protects the
quorum-sensing circuit from molecules in the environ-
ment that resemble the bona fide autoinducers. In the
present work, we show that an ultrasensitive switch
involving multiple sRNAs exists to make the commit-
ment step into quorum-sensing mode definitive. We ar-
gue that this transition is not graded but rather an on/
carried out alone and turns on behaviors that are pro-
ductive when carried out as a community.
All DNA manipulations were performed according to Sambrook et
al. (1989). PFU turbo polymerase (Stratgene) was used for PCR
reactions used in cloning, whereas Taq polymerase (Roche) was
used for all other PCR reactions. dNTPs, restriction endonucleases,
and T4 ligase were obtained from New England Biolabs. DNA purifi-
able upon request. V. harveyi deletions were constructed using the
method of Datsenko and Wanner (2000). Constructions were placed
on the V. harveyi chromosome by allele replacement (Bassler et
al., 1993). In-frame deletions were constructed by the method of
Skorupski and Taylor (1996). hapR-lacZ reporter fusions were con-
structed via the method of Kalogeraki and Winans (1997). qrr1 from
V. cholerae was overexpressed from plasmid pKK177-3RI (gift of
G.Storz). ForV.harveyi,a kanresistancecassettewas alsoincorpo-
rated into pKK177-3RI. The qrr1-lux transcriptional fusion plasmid
was constructed by ligating a PacI fragment from pCS26-Pac (Bjar-
nason et al., 2003) into an engineered PacI site in pBBR1MCS (Ko-
vach et al., 1994). The vector was digested with BamHI, which elimi-
nated a roughly 2 kb DNA fragment, and a PCR-amplified fragment
containing the promoter region of V. cholerae qrr1 was cloned into
the BamHI site.
V. harveyi cultures were grown in AB broth for 14 hr at 30?C with
aeration. The cultures were diluted 1:5000 prior to bioluminescence
assays, which were performed as described (Bassler et al., 1993).
Relative light units for V. harveyi are defined as counts min?1ml?1?
103/cfu ml?1. V. cholerae bioluminescence assays were performed
following 10 hr growth at 30?C in SOC containing tet to maintain
the plasmid pBB1 carrying V. harveyi luxCDABE. OD600nmfor each
culture was measured, and the cultures were diluted such that each
culture was at the same cell density (?1:1000 dilution). Light and
OD600were measured every 45 min as described (Miller et al., 2002).
Relative light units for V. cholerae are defined as counts min?1
Silhavy (1991). ?-galactosidase units are defined as [Vmax][dilution
Western Blot Analysis and Antibody Preparation
Western blot analysis was performed as described (Henke and
and chemiluminescence detection (Amersham) was used (Sun et
al., 1991). To analyze HapR and LuxR protein levels, HapR and LuxR
were purified (Chen et al. 2002b), and polyclonal antibodies were
generated (Henke and Bassler, 2004). Polyclonal antisera were ad-
sorbed to both E. coli pGEX-4T-1 lysates and either a V. cholerae
hapR mutant lysate or a V. harveyi luxR mutant lysate prior to use.
Northern Blot Analysis
Cultures used for RNA preparations were grown to OD600nmof 0.5.
Rifampicin was added at 100 ?g/ml, and each culture was further
incubated with aeration at 30?C. Aliquots were taken at the appro-
priate times, and RNA was extracted with TRIzol (Invitrogen) and
chloroform. RNA was precipitated with isopropanol, washed with
75% ethanol, and resuspended in DEPC water. Northern blots were
Bacterial Strains and Media
V. harveyi strains are derived from BB120 (Bassler et al., 1997) and
weregrownat 30?CwithaerationinLuria-marine (LM),heartinfusion
Hfq-sRNA Regulation of Quorum Sensing in Vibrios
performed as described (Martin et al. 1989). Steady-state Northern
of a bacterial quorum-sensing signal containing boron. Nature
Liu, Y.M., Chen, H.J., Shen, A.B., Li, J.C., et al. (2003). Comparative
genome analysis of Vibrio vulnificus, a marine pathogen. Genome
Res. 13, 2577–2587.
of gene order: a fingerprint of proteins that physically interact.
Trends Biochem. Sci. 23, 324–328.
Datsenko, K.A., and Wanner, B.L. (2000). One-step inactivation of
chromosomal genes in Escherichia coli K-12 using PCR products.
Proc. Natl. Acad. Sci. USA 97, 6640–6645.
de Lorenzo, V., and Timmis, K.N. (1994). Analysis and construction
of stable phenotypes in gram-negative bacteria with Tn5- and Tn10-
derived minitransposons. Methods Enzymol. 235, 386–405.
Dombrecht, B., Marchal, K., Vanderleyden, J., and Michiels, J.
(2002). Prediction and overview of the RpoN-regulon in closely re-
lated species of the Rhizobiales. Genome Biol 3. Published online
November 26, 2002. RESEARCH0076.1–0076.11.
Elf, J., Paulsson, J., Berg, O.G., and Ehrenberg, M. (2003). Near-
critical phenomena in intracellular metabolite pools. Biophys. J.
Federle, M.J., and Bassler, B.L. (2003). Interspecies communication
in bacteria. J. Clin. Invest. 112, 1291–1299.
Freeman, J.A., and Bassler, B.L. (1999). A genetic analysis of the
function of LuxO, a two-component response regulator involved in
quorum sensing in Vibrio harveyi. Mol. Microbiol. 31, 665–677.
Freeman, J.A., and Bassler, B.L. (2000). Sequence and function of
LuxU: a two-component phosphorelay protein that regulates quo-
rum sensing in Vibrio harveyi. J. Bacteriol. 181, 899–906.
Freeman, J.A., Lilley, B.N., and Bassler, B.L. (2000). A genetic analy-
sis of the functions of LuxN: a two-component hybrid sensor kinase
that regulates quorum sensing in Vibrio harveyi. Mol. Microbiol.
Fuqua, C., Parsek, M.R., and Greenberg, E.P. (2001). Regulation of
gene expression by cell-to-cell communication: acyl-homoserine
lactone quorum sensing. Annu. Rev. Genet. 35, 439–468.
Gouet, P., Courcelle, E., Stuart, D.I., and Metoz, F. (1999). ESPript:
ics 15, 305–308.
Hammer, B.K., and Bassler, B.L. (2003). Quorum sensing controls
biofilm formation in Vibrio cholerae. Mol. Microbiol. 50, 101–104.
Heidelberg, J.F., Eisen, J.A., Nelson, W.C., Clayton, R.A., Gwinn,
L., et al. (2000). DNA sequence of both chromosomes of the cholera
pathogen Vibrio cholerae. Nature 406, 477–483.
Henke, J.M., and Bassler, B.L. (2004). Quorum sensing regulates
type III secretion in Vibrio harveyi and Vibrio parahaemolyticus. J.
Bacteriol., in press.
Hertz, G.Z., and Stormo, G.D. (1999). Identifying DNA and protein
patterns with statistically significant alignments of multiple se-
quences. Bioinformatics 15, 563–577.
Hofacker, I.L. (2003). Vienna RNA secondary structure server. Nu-
cleic Acids Res. 31, 3429–3431.
Iwanaga, M., Yamamoto, K., Higa, N., Ichinose, Y., Nakasone, N.,
and Tanabe, M. (1986). Culture conditions for stimulating cholera
toxin production by Vibrio cholerae O1 El Tor. Microbiol. Immunol.
Jobling, M.G., and Holmes, R.K. (1997). Characterization of hapR,
a positive regulator of the Vibrio cholerae HA/protease gene hap,
and its identification as a functional homologue of the Vibrio harveyi
luxR gene. Mol. Microbiol. 26, 1023–1034.
Kalogeraki, V.S., and Winans, S.C. (1997). Suicide plasmids con-
taining promoterless reporter genes can simultaneously disrupt and
create fusions to target genes of diverse bacteria. Gene 188, 69–75.
Kovach, M.E., Phillips, R.W., Elzer, P.H., Roop, R.M., II, and Pe-
Biotechniques 16, 800–802.
Kovacikova, G., and Skorupski, K. (2002). Regulation of virulence
Genetic Screen to Identify hfq
V. harveyi strain JAF548 (luxO D47E kanr) was mutagenized with
Mini-MulacZ (cmr) as described (Martin et al., 1989). Bright colonies
were isolated, and insertions in luxO and rpoN were identified by
PCR and complementation. Tn5lacZ mutagenesis of hfq in cosmid
pBNL2014 was carried out as described previously (Showalter et
al., 1990). Transposon insertions were mapped by restriction analy-
sis and sequencing. Cosmid pBNL2031, containing a Tn5lacZ inser-
This work was supported by NSF grant MCB-0343821, NIH grant
5RO1 GM065859, ONR grant N00014-03-0183, and HHMI Predoc-
toral Fellowship (D.H.L). We are grateful to the members of the
Bassler, Gottesman, and Storz labs; and to Dr. Johan Paulsson for
for the TcpA antibody; Dr. Brian Hammer for the lux transcriptional
reporter construct; and Mr. Robert Kelly for help in cloning the qrr1-
lux promoter fusion.
Received: March 24, 2004
Revised: May 18, 2004
Accepted: May 18, 2004
Published: July 8, 2004
Argaman, L., Hershberg, R., Vogel, J., Bejerano, G., Wagner, E.G.,
in the intergenic regions of Escherichia coli. Curr. Biol. 11, 941–950.
Barrios, H., Valderrama, B., and Morett, E. (1999). Compilation and
analysis of sigma(54)-dependent promoter sequences. Nucleic
Acids Res. 27, 4305–4313.
Bassler, B.L., Wright, M., Showalter, R.E., and Silverman, M.R.
(1993). Intercellular signalling in Vibrio harveyi: sequence and func-
Bassler, B.L., Wright, M., and Silverman, M.R. (1994a). Multiple sig-
nalling systems controlling expression of luminescence in Vibrio
pathway. Mol. Microbiol. 13, 273–286.
Bassler, B.L., Wright, M., and Silverman, M.R. (1994b). Sequence
and function of LuxO, a negative regulator of luminescence in Vibrio
harveyi. Mol. Microbiol. 12, 403–412.
Bassler, B.L., Greenberg, E.P., and Stevens, A.M. (1997). Cross-
Vibrio harveyi. J. Bacteriol. 179, 4043–4045.
Benson, A.K., Ramakrishnan, G., Ohta, N., Feng, J., Ninfa, A.J., and
Newton, A. (1994). The Caulobacter crescentus FlbD protein acts
at ftr sequence elements both to activate and to repress transcrip-
tion of cell cycle-regulated flagellar genes. Proc. Natl. Acad. Sci.
USA 91, 4989–4993.
Bjarnason, J., Southward, C.M., and Surette, M.G. (2003). Genomic
profiling of iron-responsive genes in Salmonella enterica serovar
typhimurium by high-throughput screening of a random promoter
library. J. Bacteriol. 185, 4973–4982.
Cao, J.G., and Meighen, E.A.(1989). Purification and structural iden-
tification of an autoinducer for the luminescence system of Vibrio
harveyi. J. Biol. Chem. 264, 21670–21676.
Carrington, J.C., and Ambros, V. (2003). Role of microRNAs in plant
and animal development. Science 301, 336–338.
Chen, S., Lesnik, E.A., Hall, T.A., Sampath, R., Griffey, R.H., Ecker,
D.J., and Blyn, L.B. (2002a). A bioinformatics based approach to
discover small RNA genes in the Escherichia coli genome. Biosys-
tems 65, 157–177.
Chen, X., Schauder, S., Potier, N., Van Dorsselaer, A., Pelczer, I.,
Bassler, B.L., and Hughson, F.M. (2002b). Structural identification
Cell Download full-text
gene expression in Vibrio cholerae by quorum sensing: HapR func-
tions at the aphA promoter. Mol. Microbiol. 46, 1135–1147.
Lilley, B.N., and Bassler, B.L. (2000). Regulation of quorum sensing
in Vibrioharveyi by LuxOand sigma-54. Mol. Microbiol.36, 940–954.
Majdalani, N., Cunning, C., Sledjeski, D., Elliott, T., and Gottesman,
S. (1998). DsrA RNA regulates translation of RpoS message by an
anti-antisense mechanism, independent of its action as an antisi-
lencer of transcription. Proc. Natl. Acad. Sci. USA 95, 12462–12467.
Majdalani, N., Hernandez, D., and Gottesman, S. (2002). Regulation
and mode of action of the second small RNA activator of RpoS
translation, RprA. Mol. Microbiol. 46, 813–826.
Makino, K., Oshima, K., Kurokawa, K., Yokoyama, K., Uda, T., Tago-
mori, K., Iijima, Y., Najima, M., Nakano, M., Yamashita, A., et al.
(2003). Genome sequence of Vibrio parahaemolyticus: a pathogenic
mechanism distinct from that of V cholerae. Lancet 361, 743–749.
Martin, M., Showalter, R., and Silverman, M. (1989). Identification
of a locus controlling expression of luminescence genes in Vibrio
harveyi. J. Bacteriol. 171, 2406–2414.
Masse, E., and Gottesman, S. (2002). A small RNA regulates the
expression of genes involved in iron metabolism in Escherichia coli.
Proc. Natl. Acad. Sci. USA 99, 4620–4625.
Masse, E., Escorcia, F.E., and Gottesman, S. (2003a). Coupled deg-
radation of a small regulatory RNA and its mRNA targets in Esche-
richia coli. Genes Dev. 17, 2374–2383.
Masse, E., Majdalani, N., and Gottesman, S. (2003b). Regulatory
roles for small RNAs in bacteria. Curr. Opin. Microbiol. 6, 120–124.
Miller, M.B., and Bassler, B.L. (2001). Quorum sensing in bacteria.
Annu. Rev. Microbiol. 55, 165–199.
Miller, M.B., Skorupski, K., Lenz, D.H., Taylor, R.K., and Bassler,
B.L. (2002). Parallel quorum sensing systems converge to regulate
virulence in Vibrio cholerae. Cell 110, 303–314.
Miyamoto, C.M., Smith, E.E., Swartzman, E., Cao, J.G., Graham,
A.F., and Meighen, E.A. (1994). Proximal and distal sites bind LuxR
independently and activate expression of the Vibrio harveyi lux op-
eron. Mol. Microbiol. 14, 255–262.
Mok, K.C., Wingreen, N.S., and Bassler, B.L. (2003). Vibrio harveyi
quorum sensing: a coincidence detector for two autoinducers con-
trols gene expression. EMBO J. 22, 870–881.
Moll, I., Afonyushkin, T., Vytvytska, O., Kaberdin, V.R., and Blasi, U.
and small regulatory RNAs. RNA 9, 1308–1314.
Moller, T.,Franch, T., Hojrup,P., Keene, D.R., Bachinger,H.P., Bren-
nan, R.G., and Valentin-Hansen, P. (2002). Hfq: a bacterial Sm-like
protein that mediates RNA-RNA interaction. Mol. Cell 9, 23–30.
North, A.K., Weiss, D.S., Suzuki, H., Flashner, Y., and Kustu, S.
(1996). Repressor formsof the enhancer-binding proteinNrtC: some
fail in coupling ATP hydrolysis to open complex formation by sigma
54-holoenzyme. J. Mol. Biol. 260, 317–331.
network: plasmid copy number control. Q. Rev. Biophys. 34, 1–59.
and nucleotide sequence of luxR, a regulatory gene controlling bio-
luminescence in Vibrio harveyi. J. Bacteriol. 172, 2946–2954.
Skorupski, K., and Taylor, R.K. (1996). Positive selection vectors for
allelic exchange. Gene 169, 47–52.
Slauch, J.M., and Silhavy, T.J. (1991). cis-acting ompF mutations
thatresult inOmpR-dependentconstitutiveexpression. J.Bacteriol.
Localization of protective epitopes within the pilin subunit of the Vibrio
cholerae toxin-coregulated pilus. Infect. Immun. 59, 114–118.
Surette, M.G.,Miller, M.B.,and Bassler,B.L. (1999).Quorum sensing
in Escherichia coli, Salmonella typhimurium, and Vibrio harveyi: a
new family of genes responsible for autoinducer production. Proc.
Natl. Acad. Sci. USA 96, 1639–1644.
Taylor, R.K. (1991). Bacterial adhesion to mucosal surfaces. J. Che-
mother. Suppl. 1 3, 190–195.
Thelin, K.H., and Taylor, R.K. (1996). Toxin-coregulated pilus, but
not mannose-sensitive hemagglutinin, is required for colonization
Thompson, J.D., Higgins, D.G., and Gibson, T.J. (1994). CLUSTAL
W: improving the sensitivity of progressive multiple sequence align-
ment through sequence weighting, position-specific gap penalties
and weight matrix choice. Nucleic Acids Res. 22, 4673–4680.
Valentin-Hansen, P., Eriksen, M., and Udesen, C. (2004). The bacte-
rial Sm-like protein Hfq: a key player in RNA transactions. Mol.
Microbiol. 51, 1525–1533.
Vance, R.E., Zhu, J., and Mekalanos, J.J. (2003). A constitutively
active variant of the quorum-sensing regulator LuxO affects prote-
ase production and biofilm formation in Vibrio cholerae. Infect. Im-
mun. 71, 2571–2576.
van Helden, J. (2003). Regulatory sequence analysis tools. Nucleic
Acids Res. 31, 3593–3596.
Wassarman, K.M., Repoila, F., Rosenow, C., Storz, G., and Gottes-
man, S. (2001). Identification ofnovel small RNAs using comparative
genomics and microarrays. Genes Dev. 15, 1637–1651.
Wingrove, J.A., and Gober, J.W. (1994). A sigma 54 transcriptional
activator also functions as a pole-specific repressor in Caulobacter.
Genes Dev. 8, 1839–1852.
gene hierarchy; not just for motility. Mol. Microbiol. 24, 233–239.
Xavier, K.B., and Bassler, B.L. (2003). LuxS quorum sensing: more
than just a numbers game. Curr. Opin. Microbiol. 6, 191–197.
Yanisch-Perron, C., Vieira, J., and Messing, J. (1985). Improved M13
phage cloning vectors and host strains: nucleotide sequences of
the M13mp18 and pUC19 vectors. Gene 33, 103–119.
Zhang, A., Altuvia, S., Tiwari, A., Argaman, L., Hengge-Aronis, R.,
lation and binds the Hfq (HF-I) protein. EMBO J. 17, 6061–6068.
Zhang, A., Wassarman, K.M., Ortega, J., Steven, A.C., and Storz, G.
(2002).The Sm-likeHfq proteinincreases OxySRNA interactionwith
target mRNAs. Mol. Cell 9, 11–22.
Zhang, A., Wassarman, K.M., Rosenow, C., Tjaden, B.C., Storz, G.,
and Gottesman, S. (2003). Global analysis of small RNA and mRNA
targets of Hfq. Mol. Microbiol. 50, 1111–1124.
Zhu, J., and Mekalanos, J.J. (2003). Quorum sensing-dependent bio-
films enhance colonization in Vibrio cholerae. Dev. Cell 5, 647–656.
Zhu, J., Miller, M.B., Vance, R.E., Dziejman, M., Bassler, B.L., and
gene expression in Vibrio cholerae. Proc. Natl. Acad. Sci. USA 99,
Reitzer, L.J., and Magasanik, B. (1985). Expression of glnA in Esche-
richia coli is regulated at tandem promoters. Proc. Natl. Acad. Sci.
USA 82, 1979–1983.
Repoila, F., Majdalani, N., and Gottesman, S. (2003). Small non-
coding RNAs, co-ordinators of adaptation processes in Escherichia
coli: the RpoS paradigm. Mol. Microbiol. 48, 855–861.
Romeo, T. (1998). Global regulation by the small RNA-binding pro-
tein CsrA and the non-coding RNA molecule CsrB. Mol. Microbiol.
Rowe-Magnus, D.A., Guerout, A.M., Ploncard, P., Dychinco, B., Da-
vies, J., and Mazel, D. (2001). The evolutionary history of chromo-
somal super-integrons provides an ancestry for multiresistant inte-
grons. Proc. Natl. Acad. Sci. USA 98, 652–657.
in this paper is AY578785.
Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989). Molecular Clon-
ing: A Laboratory Manual (Cold Spring Harbor, NY: Cold Spring
Harbor Laboratory Press).
Showalter, R.E., Martin, M.O., and Silverman, M.R. (1990). Cloning