Role of srnas in the regulatory network controlling virulence in vibrio spendidus

gigas - extracellular superoxide dismutase ChIP-seq Chromatin immunoprecipitation DNA sequencing Csr Carbon storage regulator CT/CTX Cholera toxin DAP Diaminopimelate HCT8 Human colon c

UNIVERSITÉ PARIS SUD UFR SCIENTIFIQUE D’ORSAY ÉCOLE DOCTORALE :

GÈNES, GÉNOMES, CELLULES

Président du Jury Pr Cécile FAIRHEAD, Université Paris Sud

Rapporteur Pr Sun Nyunt WAI, Université Umeå

Rapporteur Dr Francis REPOILA, INRA, Jouy-en-Josas

Examinateur Pr Didier MAZEL, Institut Pasteur, Paris

Directrice de thèse Dr Annick JACQ, CNRS, Orsay

Laboratoire d’accueil

Signalisation et Réseaux de Régulations Bactériens

Institut de Génétique et Microbiologie, Orsay, France

Rôle des petits ARN régulateurs dans le réseau

de régulation contrôlant la virulence chez

Vibrio splendidus

1

UNIVERSITÉ PARIS SUD UFR SCIENTIFIQUE D’ORSAY ÉCOLE DOCTORALE :

GÈNES, GÉNOMES, CELLULES

DISCIPLINE: MICROBIOLOGIE

Thèse Présentée pour obtenir le grade de DOCTEUR EN SCIENCES DE L’UNIVERSITÉ PARIS SUD

Le 16 décembre 2013

Ngoc-An NGUYEN

Composition du jury:

Président du Jury Pr Cécile FAIRHEAD, Université Paris Sud

Rapporteur Pr Sun Nyunt WAI, Université Umeå

Rapporteur Dr Francis REPOILA, INRA, Jouy-en-Josas

Examinateur Pr Didier MAZEL, Institut Pasteur, Paris

Directrice de thèse Dr Annick JACQ, CNRS, Orsay

Laboratoire d’accueil

Signalisation et Réseaux de Régulations Bactériens

Institut de Génétique et Microbiologie, Orsay, France

Role of sRNAs in the regulatory network

controlling virulence in Vibrio spendidus

2

3

ACKNOWLEDGEMENTS

I would like to express my deep gratitude to Annick Jacq for her dedicated guidance, sympathy, generosity and supports at all times and in every way It is my great honor to have the opportunity of working in this lab and learning from you and other members of our team

With great honor, I would like to thank Mrs Sun Nyunt Wai, Mrs Cécile Fairhead, Mr Didier Mazel and Mr Francis Repoila for accepting to be in the dissertation committee Mrs Cécile Fairhead and Mr Didier Mazel are also my

tutors from whom I received a lot of encouragements and advices

I especially want to thank Philippe Bouloc, Chantal Bohn, Frédérique Lartigue, Tatiana Rochat, Elena Disconzi, Audrey Vingadassalon, Le Lam Thao Nguyen, Rémy Bonnin and Wang Ji for their whole-hearted supports and kindness I

have received plenty of devoted help and encouragements not only in lab work and report writing but also in all aspects of my life in France

I would like to thank Erwin van Dick and Yan Jaszczyszyn for the deep sequencing; all the collaborators and especially Claire Toffano-Nioche for their great work and supports; Muriel Decraene, Catherine Drouet and my colleagues at the Institute of Genetics and Microbiology, Orsay, France for helping me in many occasions; Simone Séror and Barry Holland for helping me to improve my English

I would like to thank the Comité pour la Coopération Scientifique et Technique avec le Vietnam for financial support during the extension period of my

PhD

I would not be here today without My Parents and Dr Tran Thu Hoa I am

grateful to you all from the bottom of my heart

I would never forget All The Professors who taught me during the Master course

for their precious knowledge, great affections and supports to all of us

Finally, I thank “Équipe de TG” for supporting and sharing with me during my 4 years in France

4

CAI Cholerae autoinducer

cAMP Cyclic adenosine-monophosphate

c-di-GMP Cyclic diguanosine-monophosphate

CDS Coding sequence

Cg-BPI C gigas bactericidal permeability increasing protein

Cg-Def C gigas defensin

Cg-EcSOD C gigas - extracellular superoxide dismutase

ChIP-seq Chromatin immunoprecipitation DNA sequencing

Csr Carbon storage regulator

CT/CTX Cholera toxin

DAP Diaminopimelate

HCT8 Human colon carcinoma cell line

HGT Horizontal gene transfer

HTS High-throughput sequencing

IGR Intergenic region

LB Luria-Bertani

LBS Saline Luria-Bertani

Lux Luminescence expression

ncRNA Non-coding regulatory RNA

NIH 3T3 Mouse embryonic fibroblast cell line

ORF Open reading frame

Pck Phosphoenolpyruvate carboxykinase

PTS Phosphotransferase system

QS

r5’ UTR Quorum-sensing Regulatory 5’ UTRs, including riboswitches

RBS Ribosomal binding site

RLU Relative light units

SD

sRNA

Shine-Dalgarno

Trans-encoded regulatory RNA

TCP Toxin co-regulated pilus

Tdh Thermostable direct hemolysin

TIR Transcription initiation region

tmRNA tRNA-mRNA like small RNA

UTR Untranslated region

VPI Vibrio pathogenicity island

6

7

TABLE OF CONTENTS

GENERAL INTRODUCTION - 9

I Regulatory small RNAs in bacteria - 9

1 Regulatory small RNAs targeting mRNAs - 11

1.1 Trans-encoded RNAs (sRNAs) - 11

1.1.1 Modes of action - 11

1.1.2 Protein partners of sRNAs - 14

1.1.2.1 The RNA chaperone Hfq - 14

1.1.2.2 The endonuclease RNase E - 16

1.1.2.3 RNases and sRNAs in Gram positive bacteria - 18

1.1.3 Bi-functional coding sRNAs - 19

1.2 Cis-encoded antisense RNAs - 21

2 sRNAs targeting proteins - 22

3 Cis-regulatory elements - 24

II Marine pathogenic Vibrios - 26

1 Vibrio cholerae - 27

2 Vibrio splendidus - 28

III Regulatory non-coding RNAs in the Vibrio genus - 29

1 Identification of small RNAs by computational approaches - 30

2 sRNA discovery in Vibrios by High Throughput Sequencing - 31

3 Toward characterization of sRNAs in Vibrios - 31

3.1 The role of Hfq in Vibrios - 31

3.2 Some Vibrio sRNAs - 32

3.3 RyhB and two antisense sRNAs involved in iron homeostasis - 33

3.4 sRNAs involved in virulence and quorum-sensing in V cholerae and V harveyi - 35

3.5 Characterized antisense RNAs in Vibrios, the tip of the iceberg? - 38

3.6 sRNAs in other Vibrios - 38

AIMS AND OUTLINE OF THE THESIS - 40

CHAPTER 1 - 42

CHAPTER 2 - 45

Abstract - 69

Introduction - 70

Experimental Procedures - 73

Bacterial strains and media - 73

Mutant construction - 73

Bacterial growth kinetics - 74

Bioluminescence assays - 74

Colony morphology and biofilm formation assay - 74

RT-PCRs - 74

Dot blot analyses - 74

SDS-PAGE analysis and detection of protease activity - 75

Results - 76

The extra copy, CsrB4 of V splendidus LGP32 is functional - 76

csrBs are regulated by VarS/VarA except csrB1 - 77

The CsrB pathway does not affect hapR mRNA level in V splendidus - 78

Conditional growth defects are dependent both on VarS/VarS and CsrBs - 79

The VarA/VarS/CsrB pathway controls biofilm formation but not motility - 81

The VarA/VarS/CsrB system controls secreted protease production - 83

Discussion - 85

Acknowledgments - 88

References - 88

Supplementary data - 93

GENERAL CONCLUSION AND PERSPECTIVES - 95

ANNEX 1 - 95

ANNEX 2 - 122

BIBLIOGRAPHY - 123

8

9

GENERAL INTRODUCTION

I Regulatory small RNAs in bacteria

The discovery of regulatory non-coding RNAs (ncRNAs) in bacteria as modulators of gene expression in response to environmental changes brought new insights into gene regulation A vast number of ncRNAs have now been discovered in both Gram negative and positive bacterial species, including pathogen These ncRNAs control a wide range of processes in cell physiology from carbon metabolism, iron homeostasis, envelope homeostasis, toxin-antitoxin systems to quorum-sensing (QS) and virulence (Repoila and Darfeuille, 2009)

The identification of bacterial ncRNAs other than tRNAs and 5S rRNAs was pioneered with the fractionation of 32P-labeled total cellular RNA about 40 years ago (Hindley, 1967) The few thus discovered RNAs were assigned to important housekeeping functions, such as M1 RNA of RNase P that acts as an endoribonuclease to process the 5′ leader sequence of precursor tRNA (Guerriertakada et al., 1983); 4.5S RNA, a component of the signal recognition particle complex, that upon binding and release of the signal peptide, directs the traffic of proteins towards membrane translocation (Herskovits et al., 2000); tmRNA, a versatile tRNA-mRNA-like small RNA that plays an important role in translational quality control (Withey and Friedman, 2003)

The majority of ncRNAs are regulators that play important roles in adaptation to living conditions with the benefits of reduced metabolic cost, additional levels of regulation, faster regulation and unique regulatory properties (Storz et al., 2011) Regulatory RNAs in bacteria

can be subdivided into two categories: cis-acting and trans-acting RNAs The cis-acting

regulatory RNAs generally correspond to the 5' untranslated region (5’ UTR) of a transcript

(so called cis-regulatory elements), of which riboswitches are important representatives (Figure 1A) (Serganov and Nudler, 2013) Trans-acting RNAs, generally 50 – 300 nt in length

(Figure 1 B to G) can further be subdivided into two categories: antisense RNAs (asRNAs)

that are cis-encoded RNAs transcribed from the opposite strand of their target, with which they have perfect complementarity; and regulatory trans-encoded RNAs (sRNAs) transcribed

from a different locus than their targets sRNAs and asRNA can act in two ways: by base pairing with mRNA(s) to activate/repress translation or affect their stability (Figure 1 B,C,D,E), sRNAs can also act by binding to/titrating a protein, often a regulator, of which they modulate the activity (Gottesman and Storz, 2011) (Figure 1F,G)

10

Figure 1 Main modes of actions of bacterial regulatory RNAs

mRNAs and cis-regulatory elements are in dark blue, trans-encoded RNAs are in red, ribosomes and

RNases are in gray, regulatory proteins are in yellow, RNApol is in light blue, σ70 is in violet and DNAs

are in black More details are described in following parts Adapted from (Bouloc and Felden, 2011).

11

1 Regulatory small RNAs targeting mRNAs

1.1 Trans-encoded RNAs (sRNAs)

Up to date, hundreds of putative sRNAs have been found in many bacterial species and

more than 120 sRNAs have been experimentally validated in E coli (Raghavan et al., 2011)

where they have been associated with a diversity of physiological roles such as controlling expression of outer membrane proteins, regulating metabolism, modulating transcriptional/translational factors, responding to various stress (Gottesman and Storz,

2011)

1.1.1 Modes of action

In this sub-category, the trans-encoded RNA genes are at a locus different from this of

its target, in a non-coding region (often called Intergenic region or IGR) As a result, the sRNA and its target have only partial complementary

Interaction of the sRNA with its mRNA target can have several consequences: inhibiting

or activating translation and/or affecting mRNA stability

Typically, upon binding to a region overlapping or adjacent to the RBS (ranging from

nucleotide -70 to +15 respective to the translation start), trans-encoded RNAs repress

translational initiation of their target mRNAs (Bouvier et al., 2008; Sharma et al., 2007) This may subsequently lead to degradation of the target However, in the case of a few sRNAs

such as MicC in Salmonella and RyhB in E coli, base pairing occurs downstream of the RBS

within the coding region Thus, while MicC represses the translational initiation of the porin

encoding gene ompC in E coli (Chen et al., 2004), the interaction between MicC and ompD

mRNA (encoding another porin) at codons 23-26, instead of preventing ribosome loading, accelerates mRNA degradation by RNase E (Pfeiffer et al., 2009) Likewise, base pairing

between RyhB and the iscRSUA polycistronic mRNA (encoding a Fe-S cluster biosynthesis machinery) in the iscR-iscS intergenic region induces the degradation of the downstream

region (Desnoyers et al., 2009) A recent study showed that bacterial sRNAs may even use target-site multiplicity to enhance the efficiency and stringency of regulation such as in the

case of SgrS and the manXYZ operon (encoding a low specificity PTS family sugar transporter) in E coli SgrS, with the help of Hfq binds to the coding region of manX and inhibits manX but not manYZ translation In addition, base pairing between SgrS and manX

CDS promotes the degradation of the whole operon by RNase E (Rice and Vanderpool,

2011) Interestingly, SgrS was also found to bind a second site, the manX-Y IGR, and

repress translation of the two downstream genes (Rice et al., 2012) Although pairing at each site is sufficient for translational repression of each gene individually, interactions at both

12

sites are considered to be required for efficient SgrS-dependent mRNA degradation and maximal recovery from stress (Rice et al., 2012) (Figure 2)

Figure 2 Regulation of manY and manZ mRNA translation by SgrS

(Upper) The manXYZ operon is organized so that 62 nt separate the stop codon of manX and the start codon of manY, and 3 nt separate the stop codon of manY and the start codon of manZ Ribosomes

are represented by stacked blue ovals

(Lower) When exposed to the non-metabolizable glucose analogs αMG or 2DG, SgrS is produced,

base pairs with manX CDS and manX-manY IGR to inhibit translation of manX and manYZ, respectively Both interactions are required to promote efficient degradation of the manXYZ mRNA

(Rice et al., 2012)

Rather than inhibiting translation, some sRNAs can also activate translation by binding and disrupting a 5' UTR inhibitory secondary structure, which normally masks the RBS

(Kozak, 2005) One of the most extensively studied case is the regulation of the rpoS mRNA,

which encodes the stationary phase specific sigma factor important for certain stress responses such as high osmolarity, low temperature and/or, aerobic/anaerobic transitions In such conditions, the expression of 3 different sRNAs RprA, DsrA, and ArcZ are respectively induced (Hengge-Aronis, 2002; Repoila et al., 2003; Soper et al., 2010), all targeting the

rpoS mRNA Translation of this mRNA is self-repressed by a stem loop in its 5′ leader that blocks ribosome access With the help of the RNA chaperone Hfq, RprA, DsrA, and ArcZ

activate translation by base-pairing to the same region in the rpoS leader, opening the stem

loop thus unmasking the RBS and leading to the expression of RpoS (Majdalani et al., 2001; Mandin and Gottesman, 2010; Sledjeski et al., 1996; Soper et al., 2010) (Figure 3A)

Interestingly, some multi-target sRNAs can inhibit expression of some targets while activating expression of others DsrA, which can adopt two different conformations

depending on its target, activates rpoS translation, but represses the expression of hns

mRNA, encoding a histone-like nucleoid-structuring protein which plays a role in formation

13

of nucleoid and negatively controls expression of various genes including rpoS (Battesti et

al., 2012; Lease and Belfort, 2000; Zhou and Gottesman, 2006) (Figure 3A, 3B) On the other hand, some genes can be regulated by different sRNAs either positively or negatively

For example, the flhDC operon in E coli (encoding the transcriptional activators for flagellum

biosynthesis and chemotaxis) is negatively controlled by OxyS, ArcZ and OmrA/B regulated sRNAs A/B) under oxidative stress, anaerobic and osmotic stress conditions respectively while positively controlled by the 90 nt sRNA McaS (multicellular adhesive sRNA) in low glucose condition leading to an increase in motility (De Lay and Gottesman, 2012; Thomason et al., 2012) It should also be noted that, contrary to ArcZ, OxyS

(OmpR-downregulates rpoS upon induction by oxidative stress (Altuvia et al., 1997) (Figure 3A)

Figure 3: Gene regulation by multitarget sRNAs

(A) Complexity of gene regulation by multi-target sRNAs upon induction by stress condition

(B) Details of activation of RpoS and repression of H-NS expression by DsrA showing secondary structures and base-paring regions Adapted from (Lease and Belfort, 2000)

14

For base-pairing sRNAs, the essential minimum length of the base pairing region (seed

region) was reported to be about 7 – 14 nucleotides as in the case of SgrS (sugar

transport-related sRNA) (Kawamoto et al., 2006; Maki et al., 2010) and RybB (Balbontin et al., 2010;

Papenfort et al., 2010) SgrS down-regulates the ptsG gene, encoding the glucose

transporter, in response to a stress condition caused by the accumulation of sugar 6-phosphate leading to growth inhibition (Vanderpool and Gottesman, 2004) RybB

negatively controls the expression of many major and minor outer membrane proteins

(OMPs) during membrane stress conditions (Vogel et al., 2003; Wassarman et al., 2001)

sRNAs are generally highly structured Target interacting regions are in many cases

located within single-strand stretches of the secondary structure (Peer and Margalit, 2011) or

in the loop of stem-loop regions of some sRNAs such as RNAIII and RsaE in Staphylococcus

aureus (Bohn et al., 2010; Geissmann et al., 2009)

As said above, some sRNAs have more than one target, other examples being RyhB

and GcvB in E coli and Salmonella It is common that the seed region within the 5' end of the

sRNA is conserved GcvB is expressed in response to increasing levels of glycine in the cell,

and is at peak in fast-growing cells in rich medium (Sharma et al., 2007; Urbanowski et al.,

2000) GcvB has a conserved GU-rich domain that overlaps the SD and/or AUG of the target

mRNA (e.g., dppA and oppA, encoding components of ABC transport systems involved in

amino acid uptake) (Pulvermacher et al., 2009; Sharma et al., 2007) In addition, a survey of

the sRNA RyhB and its targets in 19 enterobacterial species revealed that the interaction site

in the sRNA within a species is well conserved and is enriched for uridines (Richter and

Backofen, 2012)

Because trans-encoded RNAs interact through limited and imperfect pairing with their

targets, the interaction generally requires the RNA chaperone Hfq, at least in Gram-negative

bacteria In addition, the endoribonuclease RNase E is often recruited to degrade targeted

mRNAs

1.1.2 Protein partners of sRNAs

1.1.2.1 The RNA chaperone Hfq

Hfq in E coli is a thermostable protein of 70–110 amino acids, forming a hexamer of

about 66.6 kDa The Hfq protein is a member of the conserved RNA-binding Lsm (Sm-like)

protein family Sm and Lsm sub-family found in eukaryotes and archaea have various cellular

functions such as in mRNA splicing, RNA decapping, 3' end histone mRNA processing and

RNA stabilization (Mura et al., 2013; Wilusz and Wilusz, 2013)

Hfq was first identified as a bacterial protein required for the replication of the RNA

plus-strand of bacteriophage Qβ (Franze de Fernandez et al., 1968) Subsequent

15

phylogenetic analyses revealed the presence of Hfq orthologs in about half of the sequenced

Gram-positive and negative bacteria while some bacteria such as Bacillus anthracis and

Burkholderia cenocepacia contain more than one hfq gene (Ramos et al., 2011; Sun et al.,

2002) At least one archaeon, Methanococcus jannaschii, contains a protein that is related to Hfq (Mja Hfq) (Nielsen et al., 2007) In addition to the genomic encoded Mja Hfq, “Hfq-like”

protein encoding genes were recently found in four Thermococcus plasmids and three unrelated Methanococcal plasmids (Krupovic et al., 2013)

In E coli, Hfq is highly abundant and beyond Qβ replication, other important functions have been uncovered through the characterization of an hfq null mutant In E coli, the

mutant showed pleiotropic defects, such as decreased growth rate, increased sensitivity to

UV light, mutagens and oxidants and increased cell length (Tsui et al., 1994) Deletion of Hfq

can also lead to some problems in translational fidelity or increase persister cell formation (Kim and Wood, 2010) In addition, later discoveries have also shown that Δhfq mutants

have pleiotropic phenotypes such as impaired stress responses, population behavior

changes, altered metabolic regulation in other species such as Rhodobacter sphaeroides

(Berghoff et al., 2011) and/or notably, loss of virulence in a number of bacterial pathogens

such as Yersinia pseudotuberculosis (Schiano et al., 2010), S typhimurium (Sittka et al., 2007), Vibrios (Ding et al., 2004; Liu et al., 2011) and Burkholderia cenocepacia (Ramos et

al., 2011; Sousa et al., 2010) Hfq expression is proposed to be self-regulated, by a mechanism whereby Hfq binds to its own 5 'UTR region, blocks 30S subunit access and subsequently destabilizes its own mRNA in an RNase E-dependent manner (Sobrero and Valverde, 2011; Tsui et al., 1997; Vecerek et al., 2005) CsrA (carbon storage regulator) has also been reported to regulate Hfq expression (Baker et al., 2007)

Hfq carries out its post transcriptional regulatory function, such as repressing/activating translation initiation and/or accelerating decay of target mRNAs, by binding and facilitating

the limited base-pairing between trans-encoded RNAs and their mRNA targets, thus enabling

the formation of stable sRNA – mRNA duplexes (Geissmann and Touati, 2004; Moller et al., 2002) Although the mechanism is not yet fully understood, it was proposed that Hfq increases the local concentration of RNAs and/or restructures their conformation to facilitate sRNA – mRNA interaction (Vogel and Luisi, 2011) It should be noted that, for a particular sRNA, Hfq mode of action may depend on the target, as in the case of Spot42 sRNA in

E coli, which responds to glucose availability (Moller et al., 2002) and plays an important role

in catabolic repression (Beisel and Storz, 2011) Hfq is an essential partner for binding of

Spot42 to galK mRNA at the SD region of the galETKM mRNA to dis-coordinate expression

of the gal operon (involved in galactose metabolism) (Moller et al., 2002) Spot42 on the other hand, binds far upstream of SD region of the sdhCDAB polycistronic mRNA (encoding

succinate dehydrogenase) without affecting the target stability or directly repressing

16

translational initiation but recruits Hfq to bind and compete with 30S ribosomal subunits (Desnoyers and Masse, 2012) Besides its function in facilitating sRNA–target interaction, Hfq can also play a role in protecting sRNAs from degradation by PNPase (polynucleotide phosphorylase) since the levels of MicA, GlmY, RyhB, and SgrS sRNAs have been shown to

be dramatically increased upon PNPase inactivation in Hfq(-) cells (Andrade et al., 2012) Hfq-like proteins possess an evolutionarily conserved N-terminal domain of 65 amino acids, whereas the C-terminus is quite variable in length and sequence The Hfq extended C-terminus is found in γ- and β-proteobacteria whereas in the case of Gram-positive bacteria

such as in S aureus, extensions are short (Sun et al., 2002) Unlike in Gram negative

bacteria, the role of Hfq is not well-understood in Gram positive bacteria because either no Hfq homolog has been identified (e.g in Streptomyces and Actinomyces) (Sun et al., 2002)

or its actual function could not be identified In S aureus, disruption of hfq in some strains did

not impact their physiology (Bohn et al., 2007), but resulted in decreased toxicity and virulence in other strains (Liu et al., 2010) These contradictory results suggest that there

may exist other proteins that could act as functional analogue of Hfq such as Sinorhizobium

meliloti YbeY, which has structural similarities with a domain of the Argonaute proteins in

eukaryote and impacts gene expression in a way reminiscent of Hfq (Pandey et al., 2011)

1.1.2.2 The endonuclease RNase E

Another protein partner important for the action of sRNAs together with Hfq is the endoribonuclease RNase E RNase E is an endonuclease that has been extensively studied

in E coli and Salmonella In these bacteria, the main mRNA degradation pathway depends

on the RNA degradosome, a protein complex composed of an endoribonuclease (RNase E),

a 3'-5' polynucleotide phosphorylase (PNPase), and an RNA helicase (RhlB) RNase E cuts RNAs in single-stranded regions and its activity is determined by the 5' monophosphate terminus of the substrate, even if it lies at a distance from the cutting site (Prevost et al., 2011) The crystal structure of the catalytic domain of RNase E identified a sensor pocket that can accommodate the 5' monophosphate Callaghan and coworkers proposed that binding of the RNA terminal 5' monophosphate causes structural rearrangements in RNase

E, which reorient the RNA substrate, allowing greatest catalytic activity (Callaghan et al., 2005)

Hfq has been shown to co-purify with the components of the RNase E degradosome, and the C-terminal region of RNase E is involved in binding to Hfq (Morita et al., 2005) Moreover the same study demonstrated the presence of sRNAs bound to Hfq in the complex formed by Hfq and the RNA degradosome, thereby forming a ribonucleo-protein complex

A recent study by Ikeda et al (2011) also indicated that Hfq binds to the RhlB-recognition region of RNase E and that this binding is crucial for the rapid degradation of target mRNAs

17

mediated by sRNAs (Ikeda et al., 2011) Although most bacterial mRNAs are synthesized with a 5' triphosphate, and in principle protected from RNase E action, some can still be degraded upon interaction with their sRNA regulator In this case, upon base-paring with its target, the sRNA guides RNase E to a defined site and its 5' monophosphate can then be used as an alternative to allosterically activates the enzyme for cleavage of the target mRNA

This model has been demonstrated by in vitro study of MicC sRNA which acts in conjunction with Hfq and RNase E to negatively regulate the ompD mRNA (Bandyra et al., 2012)

RNase E is also important for the processing of some RNAs Processing of the mature

E coli 6S RNA, generated through a multi-step pathway by RNase E, is an example of such

function (Kim and Lee, 2004) For sRNA-Hfq mediated RNA degradation, at least two mechanisms have been described: i) sRNA acts stoechiometrically and is co-degraded with its target mRNA In this case, this provides a way to shut-off rapidly the response when the sRNA is not made anymore ii) Catalytically: there is only degradation of the target and the sRNA is recycled

Massé and coworkers have demonstrated that the turn-over of RyhB is much more rapid when transcription is active as compared to when transcription is blocked by rifampicin This can be explained by the fact that RyhB get degraded stoechiometrically with its target,

the sodB mRNA When rifampicin blocks target transcription, there is no more substrate for

the sRNA to bind to, resulting in higher stability of RyhB (Masse et al., 2003) Conversely, the group of Valentin-Hansen described that the sRNA MicM, which silences the expression of

an outer membrane protein, YbfM under many growth conditions does not become destabilized by target mRNA overexpression, indicating that the small RNA regulator acts catalytically (Overgaard et al., 2009)

sRNAs can also protect form RNase E degradation In the case of SgrS, responding to

phosphosugar stress, one of its Hfq dependent target is the yigL mRNA coding for a phosphatase SgrS was shown to activate yigL expression ~14.5-fold but not the first gene

pldB (encoding Phospholipase 2) of the bicistronic mRNA pld-yigL Further characterization

revealed that by binding to the 948 – 955 nt of the pldB mRNA, a cleavage site for RNase E,

SgrS protects yigL from degradation and increases its half-life about six times (Papenfort et

al., 2013) Once made, YigL will dephosphorylate the accumulating sugar-phosphates These sugars are proposed to be pumped out of the cell by a yet to be identified mechanism Interestingly, mRNA processing by RNase E upstream of the SgrS binding site is a prerequisite for the action of SgrS In absence of this processing event, translation by ribosomes may block SgrS binding and/or displace the sRNA-mRNA duplex (Papenfort et al., 2013)

In a reverse case, it is the mRNA that controls the stability of a sRNA Hfq-dependent

ChiX (MicM) sRNA in Salmonella is responsible for the down-regulation of a chitoporin gene

18

by binding to the RBS of ybfM (chiP) mRNA, of the ybfMN operon, required for chitin

utilization and induced by low chitosugar concentration When the level of chitosugar is high,

the expression of the chbBCARFG operon (comprising genes for chito-oligosaccharide metabolism) is induced Upon induction, the chbBCARFG mRNA is processed by RNase E

to generate the chbBC mRNA (~300 nt) which acts as a “trap” that efficiently binds ChiX This interaction results in a slight change in the 3′ terminal stem-loop of ChiX, which makes it

sensitive to degradation, thus relieving its repression on ybfM expression, leading to

upregulation of chitoporin production and increase of sugar uptake (Figueroa-Bossi et al., 2010; Overgaard et al., 2009)

1.1.2.3 RNases and sRNAs in Gram positive bacteria

Remarkably, RNase E is absent from Gram positive bacteria species such as B subtilis and S aureus Instead, RNase III in S aureus, an endonuclease that specifically binds to

and cleaves double-stranded RNA was proven to play the same role as RNase E in mRNA stability mediated by sRNA For instance, Boisset and co-workers have found that RNase III

plays an important role for the stability control of the rot mRNA (repressor of toxins) by the

RNAIII sRNA (see below for more details about RNAIII) (Boisset et al., 2007) The interaction between RNAIII and the mRNA results in repression of translation initiation and triggers RNase III hydrolysis at two loop–loop interaction sites in the RNAIII–rot mRNA complex (Figure 4) (Boisset et al., 2007)

Figure 4 The regulatory mechanism of rot mRNA by RNAIII-RNase III

RNAIII binds to its target mRNAs via two loop–loop interactions The initial base pairings is propagated, leading to the formation of an extended irregular duplex Binding of RNAIII hinders

ribosome binding and promotes the access of RNase III, which will in turn cleave the rot mRNA

Adapted from (Boisset, Geissmann et al 2007)

19

Compared to other mode of actions, positive regulation of target mRNA stability by

sRNAs seems to be rare In Clostridium perfringens, the VR-RNA sRNA base pairs with the

colA mRNA (encoding a toxin - collagenase) and induces cleavage at position -79 and -78

from the ATG This leads to partial removal of an inhibitory 5' terminal stem loop that

normally blocks expression, unmasking of the SD and activation of colA expression Moreover, processed colA mRNA is apparently more stable than the primary transcripts

possibly because of ribosome binding preventing endoribonuclease access (Dreyfus, 2009; Obana et al., 2010)

In the group A Streptococcus (GAS), another sRNA FasX binding/hemolytic-activity/streptokinase-regulator-X) has also been shown to positively

(fibronectin/fibrinogen-regulate ska which encodes a virulence factor - the secreted protein streptokinase that

promotes the degradation of blood clots While binding of FasX to the transcriptional initiation

site of the cpa mRNA (encoding a minor pilin protein) leads to inhibition of translation and

destabilization of the transcript to favor the transition from colonization to dissemination during infection, base-pairing of FasX to the 5’ end of ska mRNA at position less than 30 nt

from the start codon enhances ska transcript stability, resulting in ~10-fold increase in SKA

activity (Liu et al., 2012; Ramirez-Pena et al., 2010) The authors hypothesize that one of or both RNase J1 and J2, both essential endoribonucleases in GAS, are responsible for

initiating ska mRNA degradation via a process inhibited by FasX:ska mRNA hybridization

(Ramirez-Pena et al., 2010)

1.1.3 Bi-functional coding sRNAs

Although the majority of regulatory sRNAs in bacteria are non-coding, some sRNAs have been found to encode small polypeptide acting in the same physiological pathway as its sRNA SgrS is again an example of such sRNA Being 227 nt in length, SgrS contains a small ORF in the 5’ region which is translated into a 43-aa polypeptide named SgrT (Wadler and Vanderpool, 2007) Upon induction by sugar-phosphate stress, SgrT interacts with pre-existing PtsG glucose transporters, acting as a “plug” to block further sugar entry This

would reinforce the effect of downregulation of ptsG expression by SgrS SgrS and SgrT are generally conserved in enteric bacteria but since a few SgrS homologs lack the sgrT ORF or

an appropriate start codon, riboregulation by SgrS is considered to be more conserved than

translation (Horler and Vanderpool, 2009) Indeed, further characterization of sgrS mutants in

Salmonella showed that while abolishing sgrT expression has little effect on riboregulation,

SgrS defect in base paring led to increased sgrT expression (Balasubramanian and Vanderpool, 2013) This suggests that riboregulation of ptsG expression is the first and

20

indispensable step to rescue the cells from the stress and that SgrT acts later in response to prolonged stress condition (Balasubramanian and Vanderpool, 2013)

RNAIII sRNA, an important regulator of virulence in the Gram positive S aureus is

another example RNAIII transcription is activated by the two component system AgrC/ArgA which senses the quorum-sensing signal AIP (autoinducing peptide) when the bacterial population reaches its optimal density RNAIII presents many specific characteristics: it is large (514 nt), regulates many targets either positively or negatively and codes for a secreted 26-amino-acid long δ-hemolysin (hld) (Toledo-Arana et al., 2007) While carrying the ORF

near its 5’ end, RNAIII uses its 3’ end to repress the rot and spa mRNAs (the latter encoding

Protein A which acts as an immunological disguise and helps to prevent phagocytosis)

(Geisinger et al., 2006; Huntzinger et al., 2005) Although the regulation of hld translation is

not yet fully understood, it is possible that its translation may be repressed during the

activation of hla (encoding an α-hemolysin) by RNAIII since the hld SD sequence is masked

by the base pairing between the sRNA and its target hla (Morfeldt et al., 1995)

Figure 5: A multi-output coherent feed forward loop controlled by RNAIII in

Staphylococcus aureus Adapted from (Felden et al., 2011)

Another interesting point about RNAIII regulation is that, in response to quorum-sensing,

it regulates rot, as mentioned above, and Rot itself also controls positively and negatively the expression of spa and hla respectively The resulting network thus forms a so-called multi-

output coherent feed forward loop (Figure 5) (Felden et al., 2011)

21

1.2 Cis-encoded antisense RNAs

Antisense sRNAs or cis-encoded antisense RNAs (asRNAs) are transcribed from the

complementary strand of the genes they regulate As a consequence, asRNAs and their targets form perfect, extended duplexes The overlap of asRNAs with their targets involves either the coding or the UTRs of the gene Although many computing programs have been developed to predict sRNAs in bacteria, no program exists to predict asRNAs until recently (Pichon et al., 2012) Experimental detection and functional studies by conventional methods such as Northern blot and RT-PCR remain difficult because of low level of expression, and/or very high turnover rate Mutations affecting expression of asRNAs would also likely affect the coding gene located on the opposite strand (Georg and Hess, 2011) making genetic approaches unavailable However, recent transcriptomic studies by high-throughput sequencing (HTS) technology have revealed an unexpected reservoir of asRNAs RNA-sequencing in various species of both Gram negative and positive bacteria including

bacterial pathogens such as H pylori, Listeria sp, B subtilis, S aureus, S enterica serovar

Typhimurium and P aeruginosa led to the discovery of hundreds of potential asRNAs that

overlap about 1-25% of protein coding genes or even 46% in H pylori (Bohn et al., 2010;

Kroger et al., 2012; Mraheil et al., 2011; Nicolas et al., 2012; Sharma et al., 2010) Although interactions between asRNAs and their targets are independent of Hfq because of the

perfect complementarity, asRNAs have similar mode of action as trans-encoded RNAs

asRNAs can induce degradation of the complementary transcript by RNase E In

Salmonella, the expression of the mgtCBR operon (composed of mgtC encoding a

membrane protein needed for survival in macrophages and growth in low Mg2+ condition,

mgtB encoding a Mg2+ transporter; and mgtR encoding a peptide required for MgtC

proteolysis) is activated by the two component system PhoP/PhoQ in low Mg2+ condition

(Shin et al., 2006) The long IGR between mgtC and mgtB of the polycistronic mgtCBR

mRNA carries a promoter in the opposite strand that controls the production of an unusual long asRNA name AmgR (1.2 kb) which upon interaction with the mRNA induces its degradation by RNase E independently of Hfq (Lee and Groisman, 2010)

Many asRNAs act as antitoxins in type I toxin-antitoxin systems, in which the asRNA represses the synthesis of a toxic protein as in the case of the 60 nt SprA1AS in S aureus

(Sayed et al., 2012) SprA1AS is expressed from the opposite strand of the sprA1 gene,

encoding a pore-forming peptide having both hemolytic activity against human erythrocytes and antimicrobial activity against Gram-negative and Gram-positive bacteria (Sayed et al., 2012) SprA1AS is constitutively expressed to repress expression of the peptide during growth Interestingly, this repression is not due to the region of the asRNA which is perfectly

complementary to the sprA1 message but rather to its 5’ end which upon folding, traps the

22

SD of sprA1 mRNA to inhibit translation (Sayed et al., 2012) This study therefore raises the possibility that other cis-encoded asRNAs may not act through an antisense mechanism More generally, genome-wide asRNA transcription in S aureus is proposed to be

involved in adjusting the sense mRNA levels in a RNase III-dependent manner as a collection of short RNAs generated genome-wide (75% of sense RNAs) by RNase III digestion of overlapping sense/antisense transcripts was detected by high through-put sequencing (Lasa et al., 2011) This broad role of antisense transcription was later supported

by the fact that a number of asRNAs corresponding to 44% of mRNAs were precipitated with an RNase III defective in ribonuclease activity (Lioliou et al., 2012)

co-immuno-2 sRNAs targeting proteins

The CsrB family is among a specific class of sRNAs (100 nt to more than 400 nt in length) that instead of interacting with mRNAs, modulate the activity of the protein CsrA by mimicking its binding site on the target gene (Babitzke and Romeo, 2007; Liu et al., 1997) (Figure 6)

Figure 6 Outline of the E coli Csr system

Binding of the CsrA dimer to an mRNA can block translation Non-coding RNAs (CsrB/C) containing multiple CsrA binding sites within the loops of predicted stem-loops compete with mRNAs for CsrA binding, thus antagonizing CsrA activity Free CsrA is regenerated by the turnover of CsrB/C RNAs,

which requires in E.coli RNase E, PNPase and CsrD Adapted from (Romeo et al., 2013)

23

Depending on the species, CsrA was found to be involved in the regulation of carbon metabolism, motility, biofilm formation, secondary metabolite production, and/or quorum-sensing, etc (Table 1) (Romeo et al., 2013)

CsrA was initially identified as a negative regulator of glycogen biosynthesis in E coli

and is in most cases a post-transcriptional regulator that inhibits translation by occluding the ribosome binding site and/or inducing the degradation of various mRNA targets The translational repression mechanism generally involves binding of CsrA to the SD sequence,

preventing ribosome access However, in the case of the psl mRNA which is responsible for the synthesis of the biofilm polysaccharide Psl in P aeruginosa, repression is achieved by

binding of RsmA (an orthologue of CsrA) to the loop of the inhibitory hair-pin structure normally masking the SD, thereby stabilizing this structure (Irie et al., 2010)

Despite its generally negative regulatory function, there is a reported case in which

CsrA activates expression of the flhDC operon by stabilizing the mRNA upon binding to the

untranslated leader sequence, possibly preventing RNase E interaction with the 5’ UTR of the transcript (Wei et al., 2001; Yakhnin et al., 2013) CsrA self-represses its expression by binding to four sites in its mRNA leader thus blocking ribosome entry In addition, two of the

five promoters of csrA are specific for the stationary phase sigma factor RpoS (Yakhnin et

al., 2011)

CsrB, which adapts a conserved secondary structure containing several CsrA binding motifs (AGGA/ARGGA - where R stands for T/C/G) in the loop of many stem-loops, binds and titrates multiple copies of CsrA away from the regulated targets (Liu et al., 1997) CsrB

expression in E coli is activated by the two-component system BarA/UvrY and its stability

was shown to be controlled by CsrD, and the degradosome (Suzuki et al., 2006; Suzuki et al., 2002) The CsrB/CsrA systems are widely distributed, especially in the γ-proteobacteria, with CsrB usually present in multicopies (Table 1) (Babitzke and Romeo, 2007)

The computing program CSRNA_FIND was later developed in order to systematically detect putative CsrB homologs in various bacterial species, taking into account both their genomic location (the intergenic regions) and their secondary structure containing several CsrA binding motifs presenting in the loop regions (Kulkarni et al., 2006) Interestingly, while

the number of CsrB varies from two to three copies per species, Photobacterium profundum

possesses four putative CsrBs (Kulkarni et al., 2006)

24

Table 1: CsrB homologs and their role in various bacterial species

(Babitzke and Romeo, 2007)

Additionally, the recent discovery of a sRNA named McaS, which seems restricted to

the E coli and Shigella genera, provided an example of a sRNA acting both by direct pairing with its mRNA targets and via sequestration of a regulator protein (Jorgensen et al., 2013) While activating translation of flhDC to increase motility, McaS represses in an Hfq dependent manner the translation of csgD, coding for a transcriptional activator of curli and

base-cellulose biosynthesis (Jorgensen et al., 2012; Thomason et al., 2012) Curli and base-cellulose production, which are important for adhesion and biofilm formation are inversely correlated with motility Interestingly, McaS also directly interacts with CsrA thanks to two CsrA binding

sites (Jorgensen et al., 2013) It is still not totally clear why McaS directly activates flhDC while repressing CsrA, itself an activator of flhDC There is an increase in motility in mutants

overexpressing McaS and the increase of CsrB and CsrC in late exponential phase possibly reduces the effect of McaS on CsrA to favor its riboregulation activity (Thomason et al., 2012)

3 Cis-regulatory elements

Another class of regulatory RNAs are built-in regulatory RNA elements located in the

5’ UTR region of prokaryotic mRNAs and hence act in cis, such as riboswitches These

elements sense a wide range of signals including uncharged tRNAs (T-Box), various metabolites (riboswitches), metal ions, pH and temperature (thermoswitches) to control expression of the downstream gene (Serganov and Nudler, 2013) A total of 17 riboswitch classes with at least some experimental validation have been so far classified and more remain to be discovered (Breaker, 2012) It has been estimated that as many as 2% of genes

25

in L monocytogenes may be regulated by such cis-elements (Toledo-Arana et al., 2009)

Riboswitches are generally longer than 50 nt and consist of two parts: the aptamer region, which binds the ligand, and the so-called expression platform, which regulates gene expression through alternative RNA structures Upon binding its ligand with high affinity and selectivity, the riboswitch undergoes a change in conformation These changes usually involve alternative hairpin structures that form or disrupt transcriptional terminators or anti-terminators or that occlude or expose ribosome-binding sites (Garst et al., 2011) (Waters and Storz, 2009) (Figure 7)

In one recently described case, a MOCO-sensing riboswitch (MOCO being the molybdenum cofactor which functions as a redox center for enzymes of anaerobic metabolism) located in the 5’ leader of the moaABCDE operon in E coli, which encodes the enzymes responsible for the biosynthesis of MOCO, was shown to act by two additive mechanisms Whereas binding of MOCO would result in transcription termination of the downstream operon, CsrA binds to two sites in the 5’ leader of the moaA mRNA, inducing a change of the highly structured conformation of the riboswitch, leading to the activation of

moaA translation (Edwards et al., 2011; Patterson-Fortin et al., 2013)

Interestingly, riboswitches can also act in trans like sRNAs For instance, in the case of the S-adenosylmethionine (SAM) riboswitch SreA in L monocytogenes (Loh et al., 2009), the

binding of SAM to SreA leads to a change in conformation of the riboswitch from antiterminator to transcriptional termination structure, resulting in no transcription of the downstream mRNA and production of a short transcript (229 nt) (Loh et al., 2009) Importantly, at high temperature (37oC) the short SreA is able to bind to the 5’ UTR of the

prfA mRNA (encoding a regulator of virulence), itself a thermosensor, inhibiting prfA

expression (Johansson et al., 2002; Loh et al., 2009) Thus, in this case, a thermosensor can

be regulated by a sRNA, itself derived from a riboswitch

26

Figure 7: Riboswitch mode of actions

Riboswitches are composed of an aptamer region (purple) and an expression platform (orange) in the 5’ UTR of an mRNA (blue) Ligand (purple filled circle) binding can result in transcriptional or translational control

(1) In the absence of ligand, the expression platform assumes a conformation permissive of transcription When the ligand binds to the aptamer region, a conformational change leads to the formation of a transcriptional terminator, inhibiting gene expression

(2) In the absence of ligand, the riboswitch initially forms a terminator Upon ligand binding, this terminator is disrupted, allowing transcription to continue

(3) In the absence of ligand, the RBS is accessible, but, upon ligand binding, is sequestered into an inhibitory stem loop, preventing translation

(4) In the absence of ligand, the expression platform forms a repressive secondary structure in which the RBS is occluded Upon ligand binding, the RBS is released and translation can be initiated

Adapted from (Waters and Storz, 2009)

II Marine pathogenic Vibrios

Vibrios, which belong to the -proteobacteria branch of Gram-negative bacteria, are highly motile curve-shaped bacteria They are facultative aerobes, mostly halophilic

(requiring up to 3% salt for growth) with the exception of a few species such as Vibrio

cholerae which can grow in water of low salinity if warm and containing sufficient organic

nutrients (Colwell, 1996) These bacteria form a complex group, inhabiting a variety of ecological niches, either from free-living forms to those attached to biotic and abiotic surfaces

or associated with various hosts, the associations ranging from symbiosis to full virulence

(Johnson, 2013; Reen et al., 2006a, b) Though some species such as V cholerae, Vibrio

parahemolyticus and Vibrio vulnificus are human pathogens, the majority of Vibrios are

pathogenic to fish or marine invertebrates of environmental (corals) or economic (shrimp, shellfish) importance (Austin, 2010) Their adaptability relies on the capacity to generate

27

genetic diversity at high rates, which at least involves horizontal gene transfer (HGT) by natural transformation, transduction and conjugation (Hazen et al., 2010)

Vibrio species possess two unique circular chromosomes and many of them also

contain plasmids (Okada et al., 2005) Chromosome I contains the vast majority of housekeeping functions In contrast, hyper-variable regions constitute the majority of chromosome II Besides containing most of the unknown function ORFs, this chromosome codes for many accessory functions required for adaptation to specific conditions (Dryselius

et al., 2007; Jha et al., 2012) Therefore, in the Vibrionaceae, the evolution of virulence is

intimately associated with the genomic structure both from a functional and an evolutionary point of view

1 Vibrio cholerae

In the Vibrionaceae, V cholereae is among the most extensively studied species

Indeed, this species has been the cause of numerous cholera outbreaks and pandemics resulting in hundreds of thousands of death worldwide, going back as early as 1817 in Bengal - Eastern India (Rosenberg, 1962) Although V cholerae can belong to more than

200 serogroups based on the O antigen of the lipopolysaccharide, only toxigenic strains of

O1 and O139 serogroups cause outbreaks of cholera (Morris, 2003) V cholerae O1 is composed of two biotypes, classical and El Tor (Sack et al., 2004) V cholerae O139 is identical to V cholerae O1 El Tor except the O1 antigen was substituted by O139 by HTG of

a genomic island (Bik et al., 1995) After the seventh cholera pandemic, in 2000, the

4,033,460 base pairs (bp) complete genomic sequence of Vibrio cholerae El Tor N16961

was reported The genome consists of two circular chromosomes I and II, which are 2,961,146 and 1,072,314 bp in length, respectively (Heidelberg et al., 2000) About 3,885 open reading frames in total were predicted, and while chromosome I comprises the majority

of housekeeping and pathogenicity genes, chromosome II carries a gene capture system (the integron island) and replication initiation and host 'addiction' genes that are typically found on plasmids suggesting that this chromosome may have originally been a

megaplasmid that was captured by an ancestral Vibrio species (Heidelberg et al., 2000) The complete genomic sequence V cholerae has also provided information relevant to

studies of multi-chromosomal prokaryotic organism for which it is now a well-established model (Val et al., 2012b; Xu et al., 2003)

Cholera infection results from consumption of contaminated food and/or water Upon infection, the two most critical and well-studied virulent factors responsible for the main symptom of the disease, a profuse watery diarrhea, are the Cholera toxin (CT/CTX) and the toxin co-regulated pilus (TCP) Cholera toxin is coded by the CTX lysogenic bacteriophage

When secreted by the Vibrio into epithelial cells of human intestine (enterocytes) CT causes

28

a dramatic efflux of Cl ions and water (Waldor and Mekalanos, 1996) TCP (itself a receptor

of the phage CTXΦ) is a type-IV pilus required for colonization of the intestine by V cholerae

(Waldor and Mekalanos, 1996) Both cholera toxin coding genes ctxAB and the toxin co-regulated pilus genes tcpA-F are directly activated by the transcriptional factor ToxT

(Brown and Taylor, 1995; Matson et al., 2007) In addition, CTP and ToxT are encoded

within a genomic island - VPI-1 (vibrio pathogenicity island – ~40 kb) that possibly derives

from the genome of a filamentous bacteriophage VPI (Karaolis et al., 1998; Karaolis et al.,

1999) There exists in the genome a second pathogenicity island VPI-2 (~57.3 kb) in

V cholerae, which together with VPI-1, is present in all seventh pandemic strains of

V cholerae O1 El Tor (Dziejman et al., 2002; Jermyn and Boyd, 2002)

2 Vibrio splendidus

The vast majority of marine vibrios are commensals or pathogens of fish and shellfish

and can cause important economic losses in aquaculture Vibrio splendidus, first described

by Reichelt and coworkers in 1976, is a facultative intracellular pathogen which has not only

been associated with major oyster mortality outbreaks over the past 15 years but can also

infect fish and shrimps (Gay et al., 2004; Jensen et al., 2003; Leano et al., 1998; Reichelt et

al., 1976) The genome of the type strain, V splendidus strain LGP32, which has been

responsible for the ‘summer mortality’ syndrome having caused high financial losses in

commercial oyster farming in France since 1991, was sequenced by Le Roux and coworkers

in 2009 (Le Roux et al., 2009) Both chromosomes of V splendidus are larger than those in

V cholerae with 3299 (chromosome I) and 1675 (chromosome II) Mb respectively (Le Roux

et al., 2009)

Two main virulence factors have been identified in V splendidus using as host model

the oyster Crassostrea gigas (Binesse et al., 2008b; Duperthuy et al., 2010; Duperthuy et al.,

2011a; Le Roux et al., 2007)

The first virulence factor is an extracellular zinc-containing metalloprotease (encoded

by the vsm gene) The absence of Vsm causes a 4.5-fold reduction in oyster mortality upon

injection of its extracellular products (ECPs) compared to the wild-type (Le Roux et al.,

2007) Additionally, both ECPs and purified Vsm were able to induce detachment and cell

death of NIH 3T3, a fibroblastic mouse cell line sensitive to extracellular matrix attachment

(Binesse et al., 2008) It is still not known whether the Vsm metalloprotease plays a

predominant role during oyster infection or whether it is coregulated with additional virulence

factors Le Roux and coworkers proposed that Vsm promotes host tissue damages to

facilitate the invasion by the bacteria as well as degrades host proteins to provide readily

available nutrients upon infection (Binesse et al., 2008) Interestingly, the level of the

VS_II1062 protein, another metalloprotease, was found to be increased in the ECPs of the

29

Δvsm mutant compared to the wild type While protease activity in the Δvsm strain was 20%

less than the wild-type, the Δvsm-II1062 double mutant had no detectable protease activity

suggesting that VS_II1062 contributes significantly to the proteolytic activity of the ECPs

However, the ECPs of the Δvsm-II1062 double mutant displayed the same cytotoxicity as

this of the Δvsm ECPs, indicating that VS_II1062 plays no role in cytotoxicity (Binesse et al.,

2008) In contrast, PrtV metalloprotease, a homolog of VS_II1062 in V cholerae, was found

to have high cytotoxic effect against HCT8, a human intestinal cell line (Vaitkevicius et al.,

2008) Since VS_II1062 shares 41% identity with immune inhibitor A precursor (InHA), a

virulence factor of B thuringiensis, Binesse and coworkers suggested that VS_II1062 may

provide V splendidus with a way to evade the oyster immune system (Binesse et al., 2008)

The precise function of Vsm and the role of VS_II1062 is yet to be determined

OmpU, an outer membrane protein of V splendidus, is also considered as a major

virulence factor V splendidus LGP32 mutant lacking OmpU shows a 5-fold reduction in

oyster mortality and this mutant was out-competed by the wild-type strain in co-infection

assays (Duperthuy et al., 2010) Furthermore, ompU has been proven to be involved in

antimicrobial peptides/proteins (AMPs) resistance since ompU mutation increases the

sensitivity to Cg-Def (C gigas defensin) and Cg-BPI (C gigas bactericidal permeability

increasing protein), the two AMPs produced by the C gigas immune system (Duperthuy et

al., 2010) In addition, OmpU plays a role in invasion of oyster hemocytes through

highjacking the host-cell actin cytoskeleton In this process, Cg-EcSOD (an extracellular

superoxide dismutase, a major protein of oyster plasma) acts as an opsonin that mediates

the interaction between LGP32 OmpU and a β-integrin localized at the surface of the

hemocytes LGP32 then induces actin and clathrin polymerization in host cell, resulting in

phagocytosis The internalized bacteria are then able to escape the host cellular defenses by

preventing the acidic vacuole formation and limiting reactive oxygen species production by a

mechanism that remains to be explored (Duperthuy et al., 2011b)

III Regulatory non-coding RNAs in the Vibrio genus

So far, the majority of studies regarding sRNAs in Vibrios have focused on V cholerae

Therefore, information is still scarce about regulatory sRNAs in the Vibrio genus compared to

enterobacteriaceae In V cholerae, a few sRNAs were initially identified based on sequence

homology with known E coli sRNAs (e.g RyhB, CsrB, 6S RNA, Spot 42 RNA, 4.5S RNA,

M1 RNA) (Hershberg et al., 2003) and only a limited number of studies have identified

functions for specific sRNAs in this species Recent global approaches such as bio-computing or high-throughput sequencing have uncovered hundreds of potential sRNAs,

most of them being still in need of validation and functional characterization

30

1 Identification of small RNAs by computational approaches

The very first computational approach in V cholerae to look for additional components involved in the quorum-sensing circuit led to the discovery of 4 trans-encoded RNAs in

V cholerae named Qrr1-4 (Quorum regulatory RNA and Qrr5 in V parahaemolyticus,

V vulnificus and V harveyi (~96-108 nt), whose functions will be described in more details

below (Lenz et al., 2004) LuxO (encoding a central response regulator of the QS circuit), σ54,

a nitrogen-limitation sigma factor and Hfq were found to mediate the destabilization of the mRNA encoding the QS master regulator LuxR/HapR, suggesting the existence of sRNAs controlled by σ54 in the network Lenz and coworkers thus looked for σ54-regulated sRNAs based on several parameters: each sRNA was supposed to be coded within an IGR, and flanked by both a σ54 binding site and a Rho-independent terminator and conserved in

different Vibrios such as V cholerae, V parahaemolyticus, V vulnificus (Lenz et al., 2004)

One of the limitations of such approach is that accurate prediction of promoters or transcription factor binding sites (TFBSs) requires reliable species-specific consensus sequences; few of these have been experimentally determined in bacterial species other

than E coli (Livny et al., 2005) To overcome this, the program sRNAPredict was developed

to identify putative sRNAs in V cholerae relying only on putative terminators and regions of sequence conservation in IGRs 9 out of 10 known or putative V cholerae sRNAs as well as

32 candidates for novel sRNAs were found, and transcripts for 6 out of 9 predicted novel sRNAs were detected by Northern blots, underscoring the effectiveness of the method (Livny

et al., 2005)

Nonetheless, prediction based not only on sequence homology and Rho-independent terminator but also genomic context such as conserved synteny and TFBSs can give more insight into the potential biological roles of sRNAs (Livny et al., 2008) The SIPHT (sRNA identification protocol using high-throughput technologies) program was used to enable high-throughput, kingdom wide prediction, functional annotation of bacterial sRNA as well as differentiation between putative sRNA-encoding genes and other types of intergenic loci such

as conserved UTRs, repeat sequences, and cis-regulatory elements (Livny et al., 2008) Using SIPHT, 42 trans-encoded RNAs were identified in V cholerae More than 50% of

these were new, did not exist in Rfam (a registry of all identified functional RNAs) and had not been previously annotated

However, there exist many limitations to bio-computing approaches, especially for the detection of asRNAs In addition, the majority of sRNAs are not well conserved between species, whereas programs rely mostly on conservation Hence, high-throughput sequencing technologies have lately become the method of choice to explore bacterial transcriptomes and discover new sRNA genes

31

2 sRNA discovery in Vibrios by High Throughput Sequencing

The first attempt of experimental genome-wide detection of sRNAs in V cholerae was

based on direct cloning of cDNAs and massive-parallel sequencing by 454-pyrosequencing (Liu et al., 2009; Margulies et al., 2005) Based on their general average size, sRNAs were isolated from 4 independent samples by gel size-selection and the fractions were enriched in

putative sRNAs by depletion of 5S and tRNAs by complementary oligonucleotide-capture

This step was made necessary by the relative low depth achieved by 454 sequencing After cDNA preparation and sequencing, analysis of the data confirmed the existence of 20 previously identified sRNAs (Livny et al., 2005; Song et al., 2008) and detected hundreds of

new candidates for trans-encoded RNAs and asRNAs ranging from 16 to 225 nt in length

6 out of 7 tested new trans-encoded RNAs and 4 out of 9 tested new asRNAs were

confirmed by northern blot, demonstrating the efficiency of the method, especially in the case

of asRNAs However, because of the size selection step, either larger precursors or potential

sRNA species larger than 225 nt would have been missed CsrBs, a family of trans-encoded

RNAs or RNAβ, an asRNA (involved in iron acquisition in Vibrio anguillarum) are examples of those being more than 350 nt in length (Lenz et al., 2005; Salinas et al., 1993)

sRNA encoding genes are often expressed only under specific conditions, in particular those involved in pathogenicity To identify potential sRNAs regulated by ToxT (the major

V cholerae transcriptional regulator involved in the regulation of virulence genes), Bradley et

al (2011) compared the transcriptome of a ToxT mutant with this of the wild-type strain by

Illumina/Solexa RNA-seq (Bradley et al., 2011b) The RNA-seq data were further compared with putative ToxT binding sites determined by ToxT chromatin immunoprecipitation (ChIP-seq) 18 sRNAs were thus identified, amongst which 17 were novel Another set of 77 new

putative trans-encoded RNAs were found in a study by Mandlik et al (2011) comparing gene expression in V cholerae grown in laboratory medium, from the cecal fluid of orally infected

infant rabbits, and from small intestine homogenates of orally infected infant mice, also using Illumina HT RNA-seq (Mandlik et al., 2011)

3 Toward characterization of sRNAs in Vibrios

3.1 The role of Hfq in Vibrios

As described above, Hfq plays an important role in cell physiology and gene regulation

by sRNAs in Gram-negative bacteria However, unlike in E coli, deletion of the hfq in

V cholerae does not alter σS (RpoS) level but leads to σE over expression (σE codes for the extracytoplasmic stress response sigma factor RpoE) (Ding et al., 2004) This can be in part explained by the fact that the three sRNAs: OxyS, DsrA and RprA, positively controlling

32

translation of σS have no homologs in V cholerae (Hershberg et al., 2003) In contrast, in

V alginolyticus, a fish pathogen, the decrease of the σS level resulting from an hfq deletion,

led to attenuation in a zebra fish infection model and sensitivity to various σS-inducing

stresses such as osmotic shock, ethanol and temperature shift (Liu et al., 2011)

RpoE in E coli is involved in maintaining envelope homeostasis in response to

extracytoplasmic stress (Ades, 2008; Rouviere et al., 1995) In Vibrios, σE deletion mutants

are sensitive to such stress, i.e ethanol and/or high temperature (Brown and Gulig, 2009;

Kovacikova and Skorupski, 2002a) Ding et al (2004) showed that a V cholerae Hfq deletion

mutant was deficient in colonization of the suckling mouse small intestine and this defect was

not related to either overproduction of RpoE or minor change in toxin-coregulated pilus

expression (Ding et al., 2004) In another close relative human pathogen

V parahaemolyticus, Hfq was found to negatively regulate transcriptionally various genes

involved in virulence For instance, loss of Hfq resulted in increased expression level of tdh

(encoding a thermostable direct hemolysin), VP1680, vopC, and vopT (encoding TTSS

effectors) as well as sod and cat (encoding a superoxide dismutase and a catalase

respectively) that play a role in oxidative stress response (Nakano et al., 2008; Su et al.,

2010)

Since Hfq usually acts in conjunction with trans-encoded RNAs, the multiple effects of

an hfq deletion imply the existence of Hfq dependent sRNAs, which play important roles in

vibrio physiology

3.2 Some Vibrio sRNAs

MicX sRNA, previously known as A10, was first identified in V cholerae by a

bio-computing approach and then confirmed by northern blot showing three distinct bands of

189 nt, 241 nt and 346 nt in size (Livny et al., 2005) The long transcript has a half-life 10-fold

shorter compared to the short one and the short form is almost non-detectable in a Δhfq

mutant as well as at non-permissive temperature in a temperature-sensitive rne mutant

These are good indications that the long primary transcript is processed by RNase E in a Hfq

dependent manner to generate a stable mature form of MicX (Davis and Waldor, 2007)

A bioinformatic approach using the TargetRNA program (an mRNA target prediction

software) in combination with comparing wild-type and micX Δ196-263 transcriptomes by

microarrays uncovered two putative targets whose expression was significantly up-regulated

in the mutant: vc0972 (encoding a putative outer membrane protein) and vc0620 (encoding a

putative periplasmic peptide-binding protein for a peptide ABC transporter) (Davis and

Waldor, 2007; Tjaden et al., 2006) Moreover, the over expression of both the long and short

forms of MicX in either the wild-type or Δhfq background led to a similar reduction in vc0972

33

and vc0620 transcript levels, suggesting that Hfq is not required for the regulation of these

two targets by MicX (Davis and Waldor, 2007) and that both forms of MicX are active

In another study, Yamamoto et al (2011) described a small RNA called TfoR that involved in the regulation of natural competence In Vibrios, the ability to take up exogenous

DNA, used both as nutrient and transformation, is induced by growth on chitin, a component

of the crustacean exoskeleton and an abundant carbon and nitrogen source in marine environment (Meibom et al., 2005) The ability to uptake free DNA from the environment provides the bacteria an advantage to acquire new genetic elements that could help them to

adapt to environmental changes TfoX VC in V cholerae is an important regulator of

chitin-induced natural competence, and positively controls expression of a series of genes responsible for extra cytoplasmic chitin degradation as well as those encoding a type IV pilus-like DNA uptake machinery (Meibom et al., 2005; Yamamoto et al., 2010) In the presence of chitin or its disaccharide derivative (GlcNAc)2, tfoX VC is activated both at the transcriptional and translational level Both the TfoR sRNA and Hfq are required for activating

translation of tfoX VC Mature TfoR (about 102 nt in length) positively regulates tfoX VC

translation initiation by binding to the 5’ UTR of the mRNA and thus preventing the formation

of an inhibitory stem-loop structure that masks the RBS TfoR expression decreases

remarkably in the absence of Hfq suggesting that Hfq may be involved in tfoR turnover

(Yamamoto et al., 2010)

3.3 RyhB and two antisense sRNAs involved in iron homeostasis

Among known Hfq associated sRNAs in Vibrios, RyhB, a sRNA involved in iron homeostasis, has been well-characterized Vibrios RyhBs are 233–214 nt long and have distinct yet overlapping regulons with E coli RyhB regulon (Davis et al., 2005b; Mey et al., 2005) Iron is

an important element needed for survival and growth of bacteria since it is a cofactor of crucial enzymes required for processes such as the Krebs cycle, electron transport, DNA metabolism, and response to oxidative stress However, an excess in intracellular free iron leads to reactions that generate toxic reactive oxygen species Therefore, there must be a tight regulation of iron level which is controlled by two important factors: the ferric uptake regulator Fur and the RyhB sRNA When the cells are in high iron condition, Fur complexes

iron and bind to the promoter of the ryhB gene to inhibit its expression In low iron condition, Fur is inactivated, ryhB is now expressed and the RyhB sRNA in turn inhibits the expression

of many target mRNAs by blocking RBS entry, facilitating their degradation by RNase E with the help of Hfq (Prevost et al., 2011) In agreement with this model, RyhB was more

abundant in V cholerae grown in M9 medium and infecting the mouse intestine, having low

expression between wild-type and ΔryhB strains (Davis et al., 2005a) or between ΔryhB

mutant and RyhB-overexpressing strains (Mey et al., 2005) Interestingly, these targets represent just a small proportion of highly up- or down-regulated genes not involved in iron homeostasis or encoding iron-containing proteins, among them, were some possible biofilm- and virulence-related genes In addition, ΔryhB mutant had reduced chemotactic motility in

low-iron medium as well as reduced biofilm formation in comparison with the wild-type Defect in biofilm formation could be rescued by providing an excess of iron or succinate (Mey

et al., 2005) In V parahaemolyticus, the vibrioferrin biosynthesis operon (pvs) (encoding a

siderophore involved in iron acquisition) was also found to be under the positive control of

RyhB and Hfq (Tanabe et al., 2013)

Besides RyhB, there are two antisense RNAs that have been reported to control genes involved in iron acquisition and are located on the virulence plasmid pJM1 of the fish

pathogen Vibrio anguillarum RNAα and RNAβ are long (650 nt and 427 nt) antisense RNAs

that are transcribed on the opposite strand of the iron capture and transport-biosynthesis

operon fatDCBA angRT In this operon, fatDCBA encode proteins involved in the transport of the ferric siderophore anguibactin across the membranes and angRT are required for anguibactin biosynthesis These genes are essential for the ability of V anguillarum to cause

septicemia (Di Lorenzo et al., 2003) In iron-rich conditions Fur activate RNAα expression

while iron plays a role in increasing RNAα stability RNAα, which is an antisense of fatB, then promotes fatB mRNA degradation thereby leading to a dramatic reduction of FatA and FatB

production in order to shut down iron uptake (Chen and Crosa, 1996) Interestingly, RNAβ is expressed in both high and low iron conditions thus fine-tuning the relative ratio between iron transport and biosynthesis genes since a small amount of biosynthetic enzymes could provide enough siderophore, while more of the transport proteins increases the chances to encounter ferric siderophore molecules in the environment RNAβ, which is located on the

opposite strand of the fatA angR IGR, interacts with the IGR stem-loop inducing transcription termination of the downstream genes angRT thus reducing their expression (Stork et al.,

2007)

35

3.4 sRNAs involved in virulence and quorum-sensing in V cholerae and V harveyi

Fitness and virulence in V cholerae involve many factors such as iron acquisition,

motility, biofilm formation and colonization potential, secretion of toxins, production of outer membrane vesicles that may help to protect against UV irradiation, etc Expression of many

of these factors and their associated sRNAs are controlled by quorum-sensing (QS) (Bardill and Hammer, 2012; Rutherford and Bassler, 2012) Quorum-sensing is a process of cell-to-cell communication via small signal molecules called autoinducers (AI) that bacteria use to assess their population density in order to coordinate gene expression in the community Four redundant copies of Qrr1-4 (“quorum regulatory RNA1-4”) in V cholerae (five additive

copies in V harveyi) together with three redundant CsrBs (CsrB, CsrC, CsrD, 351-416 nt)

have been shown to be the main players connecting QS and virulence as well as

luminescence in V harveyi (Lenz et al., 2005; Lenz et al., 2004; Tu and Bassler, 2007)

Figure 8 presents a simplified schematic representation of this complex regulatory cascade

Figure 8: Quorum sensing regulation involving sRNAs in V cholerae

See text for details Adapted from (Bardill and Hammer, 2012)

36

Signaling molecule autoinducers CAI-1 (cholerae autoinducer-1 – intraspecies signal)

and AI-2 (interspecies signal) are synthesized by CqsA and LuxS, respectively During

growth, these molecules accumulate in the medium, and are then sensed by the sensor

CqsS and the complex LuxPQ, respectively, LuxQ being a sensor histidine kinase

In parallel, the histidine kinase VarS senses another unknown signal (Rutherford and Bassler

2012)

At low cell density, in absence of autoinducers, LuxPQ acts as a kinase and transfers

phosphate through LuxU (phosphorelay protein) to LuxO (response regulator) The active

LuxO in concert with σ54 then activates transcription of the sRNAs qrr1-4 Qrr sRNAs in

partnership with Hfq regulate positively AphA and negatively HapR (Rutherford et al., 2011)

While Qrrs interact with Hfq to inhibit translation and destabilize the mRNA of hapR, they

bind to the 5' UTR of the aphA mRNA to prevent the formation of a translation inhibiting

stem-loop and activate its expression (Lenz et al., 2004; Shao and Bassler, 2012;

Svenningsen et al., 2008) Moreover, two regulatory feedback loops between qrr/luxO and

qrr/aphA have been described, contributing to fine-tuning the QS system (Rutherford et al.,

2011; Svenningsen et al., 2009) In addition, a 4th target positively controlled by Qrrs named

vca0939 encodes a putative GGDEF protein that synthesizes the intracellular small-molecule

second messenger cyclic di-GMP (Hammer and Bassler, 2007) and finally Shao et al., 2013

recently identified 16 additional targets for Qrrs in V harveyi. Interestingly, the absence of

HapR led to the down-regulation of hapA (encoding the hemagglutinin protease, and a

homologue of vsm) and the up-regulation of aphA, vpsT as well as of 14 genes that are

responsible for synthesizing and degrading c-di-GMP (Srivastava, Harris et al 2011) ToxT,

a transcriptional activator, is under the direct negative control of HapR and positive control of

AphA and regulates two operons vpsA-K and vpsL-Q responsible for exopolysaccharide

production and biofilm formation (Casper-Lindley and Yildiz, 2004; Yang et al., 2010) aphA

expression and biofilm formation are induced by cyclic di-GMP (Krasteva et al., 2010;

Srivastava et al., 2011; Waters et al., 2008; Yang et al., 2010) In a parallel pathway, the

phosphorylated response regulator VarA activates the transcription of the sRNAs CsrBCD,

which inhibit the protein CsrA, itself an indirect activator of LuxO At low cell density, VarS is

inactive, VarA is not phosphorylated and CsrA indirectly enhances the activity of LuxO-P

(Figure 8)

At high cell density, LuxQ switches from kinase to phosphatase, resulting in

dephosphorylation and inactivation of LuxO On the contrary, VarS is active as a kinase;

VarA-P activates expression of csrBCD, CsrA is thus inhibited, leading to a decrease of

active LuxO As a result, qrr1-4 are not transcribed and hapR is expressed The protein

HapR on the other hand can also be activated by PckA (phosphoenolpyruvate

carboxykinase) whose expression is under the control of CsrA by an unknown mechanism

37

(Romeo et al., 1993; Tsou et al., 2011) However, it should be noted that, qrr expression control in V cholerae and V harveyi may be different from that in other Vibrios Indeed, among the five qrrs in V campbellii BAA-1116, Qrr4 and Qrr2 were found to be expressed at

high cell density state (mid-exponential-phase) (Silveira et al., 2010) More recently, all four

qrrs in V anguillarum were also reported to be expressed at high cell density or at low cell

density upon the addition of autoinducer molecules (Weber et al., 2011)

In addition, production of two additional Hfq-dependent sRNAs, TarA (~91 nt) and TarB (~69 nt) were shown to be under the direct control of ToxT (Bradley et al., 2011b; Richard et

al., 2010) The only known target of TarA is ptsG which is a target of the SgrS sRNA in

E coli (Richard et al., 2010) E coli SgrS is much longer (227 nt) than TarA and has no

homolog in Vibrios It is unclear why ToxT, a virulence regulator, should also regulate

glucose uptake but it has been shown that the ΔtarA mutant had a small defect in infant

mouse colonization in competition with the wild-type (Richard et al., 2010) Moreover, together with CsrB and TarB, TarA also appeared to be overexpressed in rabbits compared

to LB and M9 media (Mandlik et al., 2011)

Bradley et al (2011) showed that TarB, which is directly activated by ToxT, represses the expression of tcpF, the gene encoding V cholerae toxin co-regulated pilus biosynthesis protein F A V cholerae mutant strain in which the interaction between TarB and tcpF 5' UTR

was abolished could out-compete the wild type strain (Bradley et al., 2011a) According to this study, the action of TarB was independent of Hfq, in contradiction with (Davies et al., 2012) When comparing the expression profiles between the wild-type strain and the ΔtarB

mutant, no differential expression of tcpF was detected but two other genes were overexpressed: vc2760 (encoding a putative inner membrane) and vc0177 (hypothetical protein) vc0177, located in the Vibrio 7th pandemic island-1 (VPS-1), was the most strongly

upregulated gene and proposed to be a direct target of TarB, according to a biocomputing prediction of interaction (Davies et al., 2012) Further characterization has shown that

VC0177 shares structural homology with metallo-regulator repressor proteins and was renamed VspR (cholerae 7th pandemic regulator) VspR in turn, negatively regulates a gene that encodes a dinucleotide cyclase (DncV) Most interestingly, DncV is the first described enzyme capable of synthesizing both cyclic-di-AMP and c-di-GMP and cyclic-AMP-GMP, a molecule which has not been previously described Though the exact functional pathway is still under investigation, increasing the expression of DncV causes a strong inhibition of chemotaxis, and thus, significantly enhances V cholerae intestinal colonization (Davies et al., 2012)

Finally, a 140 nt sRNA named VrrA (Vibrio regulatory RNA of ompA) may also be involved in virulence and adaptation of V cholerae since a ΔvrrA mutant showed a 5-fold

increased ability to colonize the intestines of infant mice compared with the wild type This

38

could be partially due to the increase in TCP production as tcpA is a predicted target of VrrA

In addition, VrrA inhibits the expression of ompA (encoding outer membrane porin) by

binding to the SD and coding region of the mRNA and is itself under the control of σE

suggesting that VrrA plays a role in envelope stress response (Song et al., 2008) Moreover,

upon repressing ompA, VrrA induces the release of outer membrane vesicles Vesicles have

been proposed to protect the bacteria from UV damage and/or to be used for escaping the host immune system as well as delivering toxins (Kesty et al., 2004; Saunders et al., 1999;

Song and Wai, 2009) More recently, ompT (encoding an abundant porin controlling the flow

of hydrophilic solutes) whose expression is repressed by ToxR, has also been shown to be regulated by VrrA Despite using the same region of the sRNA to base-pair with its targets,

VrrA requires Hfq for regulation of ompT but not ompA (Li et al., 2000; Song et al., 2010)

3.5 Characterized antisense RNAs in Vibrios, the tip of the iceberg?

Compared to trans-encoded RNAs, only a few asRNAs have been characterized in

Vibrios while many more may exist (Liu et al., 2009; Raabe et al., 2011) Besides the two

antisense RNAs involved in iron acquisition in V anguillarum (mentioned above), there are two well-characterized additional asRNAs The first one, discovered in V cholerae, MtlS (120 nt), is transcribed from the opposite strand of the 5’ UTR of the mtlA mRNA encoding a

mannitol-specific phospho-transferase system transporter MtlS thereby acts to inhibit translation of the target gene when cells grow on carbon sources other than mannitol (Mustachio et al., 2012a) The second, 68-nt long, asRNA, RNAI, is involved in controlling

the replication of the virulence-linked plasmid pB1067 of Vibrio nigripulchritudo (Le Roux et

al., 2011) RNAI and its target RNAII are located within a 500 bp pB1067 fragment that mediates plasmid autonomous replication where RNAII serves as a primer of DNA replication Interestingly, while RNAI is fully complementary with RNAII, experimental results have shown that the interaction sites between RNAI and RNAII are located in two loops of the stem-loop structures in each molecule, since even single nucleotide change in these regions could dramatically affect this interaction (Le Roux et al., 2011)

3.6 sRNAs in other Vibrios

As more and more genome sequences of other Vibrio species have been annotated over the past 10 years, in silico searches using Infernal and Rfam database for the identification of putative ncRNA-encoding genes in four environmental Vibrio species Vibrio

alginolyticus 40B, Vibrio communis 1DA3, Vibrio mimicus VM573 and Vibrio campbellii

BAA-1116 were carried out by Silveira and coworkers in order to explore sRNA diversity (Silveira

et al., 2010) 31–38 putative sRNAs genes per species were identified: trans-encoded

39

ncRNAs, ncRNAs that modulate protein activity (CsrB) and riboswitches Most of the sRNAs thus identified either carry house-keeping functions and/or are well conserved among different bacteria (e.g 6S, tmRNA, CsrB, RyhB, Spot 42, etc.) (Silveira et al., 2010)

In conclusion, compared to other bacterial species, information about sRNAs is still

scarce in Vibrionaceae, with the vast majority of the studies concerning V cholerae sRNAs

vary a lot between species, except for a few exception already mentioned Therefore, studies

in other Vibrios such as those pathogenic to marine hosts, should uncover a world of new

RNAs regulating virulence as well as new functions Transcriptome analyses by sequencing remains the method of choice to explore fully this universe In addition, metagenomics in combination with metatranscriptomics will be the future direction to study microbial population biology