Molecular Microbiology (2003)

47

(6) 1613ndash1625

copy 2003 Blackwell Publishing Ltd

Blackwell Science LtdOxford UKMMIMolecular Microbiology0950-382XBlackwell Publishing Ltd 200347Original Article

Analysis of the PrfA regulon of L monocytogenesE Milohanic et al

Accepted 2 December 2002 For correspondence E-mailcbuchpasteurfr Tel (

+

33) 1 45 68 83 72 Fax (

+

33) 1 45 68 87 86

dagger

Present address Molecular Microbiology Section Department ofClinical Veterinary Science University of Bristol Langford BS405DU UK

Transcriptome analysis of

Listeria monocytogenes

identifies three groups of genes differently regulatedby PrfA

Eliane Milohanic

12

Philippe Glaser

2

Jean-Yves Coppeacutee

3

Lionel Frangeul

2

Yolanda Vega

4

Joseacute A Vaacutezquez-Boland

4dagger

Frank Kunst

2

Pascale Cossart

1

and Carmen Buchrieser

2

1

Uniteacute des Interactions Bacteacuteries-Cellules

2

Laboratoire de Geacutenomique des Microorganismes Pathogegravenes and

3

Plateau Technique Puces agrave ADN Institut Pasteur 28 Rue du Docteur Roux 75724 Paris Cedex 15 France

4

Grupo de Patogeacutenesis Molecular y Genoacutemica Bacteriana Facultad de Veterinaria Universidad de Leon 24071 Spain

Summary

PrfA is the major regulator of

Listeria

virulence geneexpression This protein is a member of the CrpFnrfamily of transcription regulators To gain a deeperunderstanding of the PrfA regulon we constructed awhole-genome array based on the complete genomesequence of

Listeria monocytogenes

strain EGDe andevaluated the expression profiles of the wild-typeEGDe and a

prfA

-deleted mutant (EGDe

DDDD

prfA

) Bothstrains were grown at 37

infininfininfininfin

C in brainndashheart infusionbroth (BHI) and BHI supplemented with either acti-vated charcoal a compound known to enhancevirulence gene expression or cellobiose a sugarreported to downregulate virulence gene expressionin spite of full expression of PrfA We identified threegroups of genes that are regulated differently GroupI comprises in addition to the 10 already knowngenes two new genes

lmo2219

and

lmo0788

bothpositively regulated and preceded by a putative PrfAbox Group II comprises eight negatively regulatedgenes

lmo0278

is preceded by a putative PrfA boxand the remaining seven genes (

lmo0178ndashlmo0184

)are organized in an operon Group III comprises 53genes of which only two (

lmo0596

and

lmo2067

) arepreceded by a putative PrfA box Charcoal additioninduced upregulation of group I genes but abolished

regulation by PrfA of most group III genes In thepresence of cellobiose all the group I genes weredownregulated whereas group III genes remainedfully activated Group II genes were repressed in allconditions tested A comparison of the expressionprofiles between a second

L monocytogenes

strain(P14) its spontaneous mutant expressing a constitu-tively active PrfA variant (P14

prfA

) and its corre-sponding

prfA

-deleted mutant (P14

DDDD

prfA

) and theEGDe strain revealed interesting strain-specific differ-ences Sequences strongly similar to a sigma B-dependent promoter were identified upstream of 22group III genes These results suggest that PrfApositively regulates a core set of 12 genes precededby a PrfA box and probably expressed from a sigmaA-dependent promoter In contrast a second setof PrfA-regulated genes lack a PrfA box and areexpressed from a sigma B-dependent promoter Thisstudy reveals that PrfA can act as an activator or arepressor and suggests that PrfA may directly or indi-rectly activate different sets of genes in associationwith different sigma factors

Introduction

Listeria monocytogenes

is a Gram-positive facultativeintracellular bacterial pathogen that causes severe food-borne infections in humans and animals The mainsymptoms include meningoencephalitis septicaemiaand abortions and as shown recently gastroenteritis(Vazquez-Boland

et al

2001) Several

L monocytogenes

virulence genes have been identified They include genesinvolved in adherence and uptake by the host cell escapefrom the phagocytic vacuoles and intracellular replicationor in intra- and intercellular movement (Cossart andLecuit 1998 Vazquez-Boland

et al

2001) The onlyregulatory factor identified to date necessary for the reg-ulation of the expression of most of these virulence genesis PrfA PrfA activates all genes of the so-called virulencegene cluster of

L monocytogenes

(

prfA

plcA

hly

mpl

actA

and

plcB

) as well as the expression of

inlA

and

inlB

which encode two invasion proteins (InlA and InlB)(Dramsi

et al

1996)

inlC

which encodes a smallsecreted internalin-like protein (InlC) (Engelbrecht

et al

1996 Lingnau

et al

1996) and

hpt

a gene encoding a

1614

E Milohanic

et al

copy 2003 Blackwell Publishing Ltd

Molecular Microbiology

47

1613ndash1625

UhpT-related sugar phosphate transporter that mediatesrapid intracellular proliferation (Chico-Calero

et al

2002)PrfA is structurally and functionally related to the Crp

Fnr family of transcription regulators (Lampidis

et al

1994) Most members of the CrpFnr family have beenstudied in Gram-negative bacteria except the Flp pro-tein of

Lactobacillus casei

(Irvine and Guest 1993) Fnrof

Bacillus subtilis

a regulator involved in anaerobicrespiration (Cruz Ramos

et al

1995) and PrfA one ofthe best-studied representatives of this family in Gram-positive bacteria CrpFnr-like regulators are site-specificDNA-binding proteins possessing a C-terminal helixndashturnndashhelix (HTH) DNA-binding motif and a

b

-rollstructure in their N-terminal domains PrfA binds to apalindromic PrfA recognition sequence (PrfA box)located at position

-

40 from the transcription start sitein PrfA-dependent promoters (Mengaud

et al

1989Sheehan

et al

1996) PrfA-dependent transcription ofvirulence genes is weak below 30

infin

C but becomesinduced at 37

infin

C because of a switch of the non-trans-lated leader RNA of PrfA between an inactive structureat low temperatures and an active structure at hightemperatures (Johansson

et al

2002) The mediumcomposition is critical for full expression of PrfA-controlled genes in

L monocytogenes

Starvation con-ditions (incubation in minimal essential medium Bohne

et al

1994) or incubation in brainndashheart infusion (BHI)medium supplemented with activated charcoal (Ripio

et al

1996) induce the PrfA regulon In contrastmetabolizable unphosphorylated sugars such as glu-cose maltose fructose mannose trehalose andcellobiose inhibit the expression of PrfA-dependentvirulence genes (Park and Kroll 1993 Milenbachs

et al

1997 Ripio

et al

1997a) However in thepresence of cellobiose PrfA is fully expressed suggest-ing that PrfA is post-transcriptionally modified (Renzoni

et al

1997) and can switch between a transcriptionallyactive and inactive form upon interaction with ahypothetical activating factor (Renzoni

et al

1997 Ripio

et al

1997b Vega

et al

1998)Using the extensive data set obtained during the

genome sequencing project of

L monocytogenes

EGDe(Glaser

et al

2001) to construct whole-genome macro-arrays we have been able to study the global regulatorycapacity of PrfA Gene expression patterns of two

Lmonocytogenes

wild-type strains (EGDe and P14) andtheir isogenic

prfA

-deleted mutants as well as a

prfA

mutant (P14

prfA

) constitutively overexpressing all PrfA-dependent virulence genes were studied and comparedin different conditions Three groups of genes differentlyregulated by PrfA were identified Real-time quantitativepolymerase chain reaction (PCR) was used to corrobo-rate the gene expression patterns revealed by themacroarrays

Results

Listeria

monocytogenes

DNA macroarray and experimental design

Whole-genome arrays were constructed by amplifying

ordf

500-bp-long PCR products specific for each gene usingopen reading frame (ORF)-specific primers Ninety-nineper cent (2816 ORFs) of the 2853 predicted ORFs of the

L monocytogenes

EGDe genome were amplified suc-cessfully The PCR products were spotted onto nylonmembranes and then hybridized with

33

P-labelled single-stranded cDNA derived from total cellular RNA Theexpression profiles of the different

L monocytogenes

wild-type and mutant strains were studied after growth at 37

infin

Cin BHI and BHI supplemented with activated charcoal(BHIC) or cellobiose (BHICel) Charcoal increases thetranscription levels of

prfA

(Ripio

et al

1996) and genesunder its control (Geoffroy

et al

1989) whereas cellobi-ose downregulates virulence gene expression (Park andKroll 1993 Brehm

et al

1996 Renzoni

et al

1997)Total RNA was purified from a culture grown to OD

600

=

06as it has been shown previously that the overall amountof

prfA

transcript drops in stationary phase (Mengaud

et al

1991) For each growth condition two differentmRNA preparations from independent cultures were usedfor cDNA synthesis and subsequent hybridization to twosets of arrays To identify statistically significant differ-ences in gene expression we used the Statistical Analysisfor Microarrays (

SAM

) program (Tusher

et al

2001) Allgenes with statistically significant changes in the level ofexpression and with at least a twofold change were con-sidered in this analysis All primary data from the tran-scriptome experiments and the statistical analysis areavailable as

Supplementary material

Expression profiles of wild-type

L monocytogenes

EGDe compared with its isogenic

prfA

mutant in BHI

Analysis of the expression profiles in BHI revealed 70genes organized in 47 predicted transcriptional units (13predicted operons and 34 single genes) differentially tran-scribed in the

D

prfA

mutant relative to the wild-type EGDestrain Sixty-two of these genes were upregulatedwhereas eight (organized in two transcriptional units) weredownregulated in the wild-type EGDe strain when com-pared with the

prfA

mutant strain (Table 1) These genesencode primarily known or predicted transport proteins(20) proteins involved in stress response (15) and pro-teins of unknown function (21) (Table 1) From the 10previously known PrfA-dependent virulence genes seven(

prfA

plcA

hly

mpl

actA

plcB

and

inlA

) were detectedas clearly upregulated by PrfA whereas three (

inlB

inlC

and

hpt

) were slightly or not PrfA activated in BHI Sixty-

Analysis of the PrfA regulon of

L monocytogenes 1615

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

Table 1 Genes differentially regulated in wild-type L monocytogenes EGDe and P14prfA relative to their isogenic prfA deleted mutants

1616 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

three PrfA-regulated genes are newly identified in thisstudy

As PrfA binds to a specific palindromic sequence thePrfA box (consensus TTAACAnnTGTTAA) we searchedfor putative PrfA boxes upstream of all the newly identifiedgenes Five of these genes (lmo2219 lmo0788 lmo0278lmo0596 lmo2067) are preceded by a putative PrfA box(Table 2) Gene lmo2219 encodes a protein similar toPrsA of Bacillus subtilis a post-translocation molecularchaperon (Kontinen et al 1991 Kontinen and Sarvas1993) genes lmo0788 and lmo0596 code for probablemembrane proteins of unknown function lmo0278 codesfor a sugar ABC transporter and lmo2067 is a genecoding for a bile salt hydrolase recently shown to be PrfAregulated and involved in the intestinal and hepaticphases of listeriosis (Dussurget et al 2002) In contrastfor the remaining 37 genes (single genes or first genesof a predicted operon) no sequences similar to PrfAboxes were identified Using MEME (Bailey and Elkan1994) a tool for discovering motifs in a group ofrelated DNA sequences and BIOPROSPECTOR (httpbioprospector standfordedu) allowing for the modellingof gapped motifs and motifs with palindromic patterns (Liuet al 2001) we searched for an alternative PrfA bindingsite upstream of these newly identified genes No otherputative PrfA binding sites were found Interestingly manyof these genes have homologues which belong to thestress-induced sigma B regulon in B subtilis (Petersohnet al 2001 Price et al 2001) Therefore a search forputative sigma B promoters using a matrix based on47 B subtilis sigma B promoter sequences extracted fromthe B subtilis regulatorypromoter database DBTB(httpelmoimsu-tokyoacjpdbtbs) was undertaken inthe L monocytogenes EGDe genome sequence (FChetouani and M S Gelfand unpublished data) We alsoconducted a motif search using the B subtilis sigma Bconsensus sequence AGGTTT-N17-GGGTAT with a max-imum of two mismatches and a maximum distance of 300nucleotides upstream from the start codon of the gene ascriteria These searches allowed us to identify a sequencesimilar to a sigma B-dependent promoter upstream of 22of the 63 newly identified genes Six of these are the firstgene of putative co-transcribed units accounting for 33genes in total (Table 3) The primary data and the statis-tical analysis are available as Supplementary material(Tables S1A and S1B)

Expression profiles of wild-type L monocytogenesEGDe compared with its isogenic prfA mutant incharcoal-supplemented BHI (BHIC)

Transcriptome analysis of the EGDe wild-type strain com-pared with its isogenic mutant grown in BHI indicated thatthree genes (inlB inlC and hpt) previously shown to be

regulated by PrfA were not activated in BHI (see above)It has been reported that activated charcoal in the culturemedium increases the synthesis of listeriolysin (LLO)encoded by the hly gene and also that of the lecithinaseencoded by the plcB gene (Geoffroy et al 1989 1991Ripio et al 1996) suggesting that charcoal somehowexerts an effect on the L monocytogenes virulence generegulation mechanisms Analysis of the expression levelsshowed that after growth in BHIC all previouslydescribed PrfA-regulated genes including inlB inlC andhpt were indeed regulated by PrfA in the EGDe wildtype The level of PrfA activation was increased betweentwo- and fourfold compared with growth without charcoal(Table 1) The upregulation of two of the newly identifiedgenes lmo2219 and lmo0788 was similar to that of theknown virulence genes The eight genes identified asbeing negatively regulated by PrfA (lmo0278 lmo0178ndashlmo0184) in BHI remained repressed at a similar level inBHIC (Table 1) Surprisingly 52 of the remaining 53 PrfA-regulated genes were no longer activated by PrfA inBHIC (Table 1) They showed expression levels similar tothat of EGDeDprfA grown in BHI (Supplementarymaterial) In summary the addition of charcoal to themedium allowed us to define three groups of genesGroup I comprises 12 genes including the previouslyknown virulence genes the activation of which by PrfA isincreased upon charcoal treatment Group II compriseseight negatively regulated genes the regulation of whichis not altered by the presence of charcoal and group IIIcomprises the 53 remaining genes the activation ofwhich by PrfA is abolished in growth in BHIC (Table 1)Furthermore 12 additional upregulated and four down-regulated genes were identified in BHIC compared withgrowth in BHI The primary data and the statistical analy-sis are available as Supplementary material (Tables S2Aand S2B)

Transcriptome of wild-type L monocytogenes P14its isogenic DprfA and P14prfA mutants in BHIand BHIC

Listeria monocytogenes strain EGDe has been chosenfor transcriptome analysis as the complete genomesequence for strain EGDe is available Moreover EGDeis an intensively studied strain in numerous laboratoriesHowever virulence heterogeneity among L monocytoge-nes strains is well documented (Brosch et al 1993)DNAndashDNA hybridization studies indicate considerablegenetic differences among different strains of L monocy-togenes (M Doumith et al unpublished data) and straindifferences in the levels of virulence gene expression havebeen reported (Ripio et al 1996) To date all known PrfA-regulated genes are implicated in virulence Analysis andcomparison of the PrfA regulon of different Listeria iso-

Analysis of the PrfA regulon of

L monocytogenes 1615

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

Table 1 Genes differentially regulated in wild-type L monocytogenes EGDe and P14prfA relative to their isogenic prfA deleted mutants

1616 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

three PrfA-regulated genes are newly identified in thisstudy

As PrfA binds to a specific palindromic sequence thePrfA box (consensus TTAACAnnTGTTAA) we searchedfor putative PrfA boxes upstream of all the newly identifiedgenes Five of these genes (lmo2219 lmo0788 lmo0278lmo0596 lmo2067) are preceded by a putative PrfA box(Table 2) Gene lmo2219 encodes a protein similar toPrsA of Bacillus subtilis a post-translocation molecularchaperon (Kontinen et al 1991 Kontinen and Sarvas1993) genes lmo0788 and lmo0596 code for probablemembrane proteins of unknown function lmo0278 codesfor a sugar ABC transporter and lmo2067 is a genecoding for a bile salt hydrolase recently shown to be PrfAregulated and involved in the intestinal and hepaticphases of listeriosis (Dussurget et al 2002) In contrastfor the remaining 37 genes (single genes or first genesof a predicted operon) no sequences similar to PrfAboxes were identified Using MEME (Bailey and Elkan1994) a tool for discovering motifs in a group ofrelated DNA sequences and BIOPROSPECTOR (httpbioprospector standfordedu) allowing for the modellingof gapped motifs and motifs with palindromic patterns (Liuet al 2001) we searched for an alternative PrfA bindingsite upstream of these newly identified genes No otherputative PrfA binding sites were found Interestingly manyof these genes have homologues which belong to thestress-induced sigma B regulon in B subtilis (Petersohnet al 2001 Price et al 2001) Therefore a search forputative sigma B promoters using a matrix based on47 B subtilis sigma B promoter sequences extracted fromthe B subtilis regulatorypromoter database DBTB(httpelmoimsu-tokyoacjpdbtbs) was undertaken inthe L monocytogenes EGDe genome sequence (FChetouani and M S Gelfand unpublished data) We alsoconducted a motif search using the B subtilis sigma Bconsensus sequence AGGTTT-N17-GGGTAT with a max-imum of two mismatches and a maximum distance of 300nucleotides upstream from the start codon of the gene ascriteria These searches allowed us to identify a sequencesimilar to a sigma B-dependent promoter upstream of 22of the 63 newly identified genes Six of these are the firstgene of putative co-transcribed units accounting for 33genes in total (Table 3) The primary data and the statis-tical analysis are available as Supplementary material(Tables S1A and S1B)

Expression profiles of wild-type L monocytogenesEGDe compared with its isogenic prfA mutant incharcoal-supplemented BHI (BHIC)

Transcriptome analysis of the EGDe wild-type strain com-pared with its isogenic mutant grown in BHI indicated thatthree genes (inlB inlC and hpt) previously shown to be

regulated by PrfA were not activated in BHI (see above)It has been reported that activated charcoal in the culturemedium increases the synthesis of listeriolysin (LLO)encoded by the hly gene and also that of the lecithinaseencoded by the plcB gene (Geoffroy et al 1989 1991Ripio et al 1996) suggesting that charcoal somehowexerts an effect on the L monocytogenes virulence generegulation mechanisms Analysis of the expression levelsshowed that after growth in BHIC all previouslydescribed PrfA-regulated genes including inlB inlC andhpt were indeed regulated by PrfA in the EGDe wildtype The level of PrfA activation was increased betweentwo- and fourfold compared with growth without charcoal(Table 1) The upregulation of two of the newly identifiedgenes lmo2219 and lmo0788 was similar to that of theknown virulence genes The eight genes identified asbeing negatively regulated by PrfA (lmo0278 lmo0178ndashlmo0184) in BHI remained repressed at a similar level inBHIC (Table 1) Surprisingly 52 of the remaining 53 PrfA-regulated genes were no longer activated by PrfA inBHIC (Table 1) They showed expression levels similar tothat of EGDeDprfA grown in BHI (Supplementarymaterial) In summary the addition of charcoal to themedium allowed us to define three groups of genesGroup I comprises 12 genes including the previouslyknown virulence genes the activation of which by PrfA isincreased upon charcoal treatment Group II compriseseight negatively regulated genes the regulation of whichis not altered by the presence of charcoal and group IIIcomprises the 53 remaining genes the activation ofwhich by PrfA is abolished in growth in BHIC (Table 1)Furthermore 12 additional upregulated and four down-regulated genes were identified in BHIC compared withgrowth in BHI The primary data and the statistical analy-sis are available as Supplementary material (Tables S2Aand S2B)

Transcriptome of wild-type L monocytogenes P14its isogenic DprfA and P14prfA mutants in BHIand BHIC

Listeria monocytogenes strain EGDe has been chosenfor transcriptome analysis as the complete genomesequence for strain EGDe is available Moreover EGDeis an intensively studied strain in numerous laboratoriesHowever virulence heterogeneity among L monocytoge-nes strains is well documented (Brosch et al 1993)DNAndashDNA hybridization studies indicate considerablegenetic differences among different strains of L monocy-togenes (M Doumith et al unpublished data) and straindifferences in the levels of virulence gene expression havebeen reported (Ripio et al 1996) To date all known PrfA-regulated genes are implicated in virulence Analysis andcomparison of the PrfA regulon of different Listeria iso-

1616 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

three PrfA-regulated genes are newly identified in thisstudy

As PrfA binds to a specific palindromic sequence thePrfA box (consensus TTAACAnnTGTTAA) we searchedfor putative PrfA boxes upstream of all the newly identifiedgenes Five of these genes (lmo2219 lmo0788 lmo0278lmo0596 lmo2067) are preceded by a putative PrfA box(Table 2) Gene lmo2219 encodes a protein similar toPrsA of Bacillus subtilis a post-translocation molecularchaperon (Kontinen et al 1991 Kontinen and Sarvas1993) genes lmo0788 and lmo0596 code for probablemembrane proteins of unknown function lmo0278 codesfor a sugar ABC transporter and lmo2067 is a genecoding for a bile salt hydrolase recently shown to be PrfAregulated and involved in the intestinal and hepaticphases of listeriosis (Dussurget et al 2002) In contrastfor the remaining 37 genes (single genes or first genesof a predicted operon) no sequences similar to PrfAboxes were identified Using MEME (Bailey and Elkan1994) a tool for discovering motifs in a group ofrelated DNA sequences and BIOPROSPECTOR (httpbioprospector standfordedu) allowing for the modellingof gapped motifs and motifs with palindromic patterns (Liuet al 2001) we searched for an alternative PrfA bindingsite upstream of these newly identified genes No otherputative PrfA binding sites were found Interestingly manyof these genes have homologues which belong to thestress-induced sigma B regulon in B subtilis (Petersohnet al 2001 Price et al 2001) Therefore a search forputative sigma B promoters using a matrix based on47 B subtilis sigma B promoter sequences extracted fromthe B subtilis regulatorypromoter database DBTB(httpelmoimsu-tokyoacjpdbtbs) was undertaken inthe L monocytogenes EGDe genome sequence (FChetouani and M S Gelfand unpublished data) We alsoconducted a motif search using the B subtilis sigma Bconsensus sequence AGGTTT-N17-GGGTAT with a max-imum of two mismatches and a maximum distance of 300nucleotides upstream from the start codon of the gene ascriteria These searches allowed us to identify a sequencesimilar to a sigma B-dependent promoter upstream of 22of the 63 newly identified genes Six of these are the firstgene of putative co-transcribed units accounting for 33genes in total (Table 3) The primary data and the statis-tical analysis are available as Supplementary material(Tables S1A and S1B)

Expression profiles of wild-type L monocytogenesEGDe compared with its isogenic prfA mutant incharcoal-supplemented BHI (BHIC)

Transcriptome analysis of the EGDe wild-type strain com-pared with its isogenic mutant grown in BHI indicated thatthree genes (inlB inlC and hpt) previously shown to be

regulated by PrfA were not activated in BHI (see above)It has been reported that activated charcoal in the culturemedium increases the synthesis of listeriolysin (LLO)encoded by the hly gene and also that of the lecithinaseencoded by the plcB gene (Geoffroy et al 1989 1991Ripio et al 1996) suggesting that charcoal somehowexerts an effect on the L monocytogenes virulence generegulation mechanisms Analysis of the expression levelsshowed that after growth in BHIC all previouslydescribed PrfA-regulated genes including inlB inlC andhpt were indeed regulated by PrfA in the EGDe wildtype The level of PrfA activation was increased betweentwo- and fourfold compared with growth without charcoal(Table 1) The upregulation of two of the newly identifiedgenes lmo2219 and lmo0788 was similar to that of theknown virulence genes The eight genes identified asbeing negatively regulated by PrfA (lmo0278 lmo0178ndashlmo0184) in BHI remained repressed at a similar level inBHIC (Table 1) Surprisingly 52 of the remaining 53 PrfA-regulated genes were no longer activated by PrfA inBHIC (Table 1) They showed expression levels similar tothat of EGDeDprfA grown in BHI (Supplementarymaterial) In summary the addition of charcoal to themedium allowed us to define three groups of genesGroup I comprises 12 genes including the previouslyknown virulence genes the activation of which by PrfA isincreased upon charcoal treatment Group II compriseseight negatively regulated genes the regulation of whichis not altered by the presence of charcoal and group IIIcomprises the 53 remaining genes the activation ofwhich by PrfA is abolished in growth in BHIC (Table 1)Furthermore 12 additional upregulated and four down-regulated genes were identified in BHIC compared withgrowth in BHI The primary data and the statistical analy-sis are available as Supplementary material (Tables S2Aand S2B)

Transcriptome of wild-type L monocytogenes P14its isogenic DprfA and P14prfA mutants in BHIand BHIC

Listeria monocytogenes strain EGDe has been chosenfor transcriptome analysis as the complete genomesequence for strain EGDe is available Moreover EGDeis an intensively studied strain in numerous laboratoriesHowever virulence heterogeneity among L monocytoge-nes strains is well documented (Brosch et al 1993)DNAndashDNA hybridization studies indicate considerablegenetic differences among different strains of L monocy-togenes (M Doumith et al unpublished data) and straindifferences in the levels of virulence gene expression havebeen reported (Ripio et al 1996) To date all known PrfA-regulated genes are implicated in virulence Analysis andcomparison of the PrfA regulon of different Listeria iso-

Analysis of the PrfA regulon of L monocytogenes 1617

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

lates might shed light on the molecular basis of virulencedifferences In order to address this question we investi-gated a second L monocytogenes strain L monocytoge-nes P14 a serovar 4b clinical isolate from a humanlisteriosis outbreak in Spain in 1989 (Vazquez-Bolandet al 1991) Like EGDe and all wild-type strains of Lmonocytogenes P14 shows weak haemolytic and lecithi-nase activities (Ripio et al 1996 1997b) The main rea-son for choosing this strain was the availability of a prfAmutant (P14prfA) expressing a constitutive PrfA formresulting from a single amino acid substitution in PrfA

Gly145Ser which lsquofreezesrsquo the regulatory protein in itsactive conformation (Ripio et al 1996) A prfA deletedmutant of P14 (P14DprfA) was constructed by doublecross-over as described previously (Chico-Calero et al2002)

Virulence gene expression was very low in wild-typeP14 after growth in BHI After growth in BHIC the viru-lence genes and one of the newly identified group I genes(lmo0778) were activated However the expression levelsof all other genes remained low (data not shown) Wetherefore took advantage of strain P14prfA and com-

Table 2 Sequence and position of putative PrfA boxes in the five new PrfA box-containing genes

Gene Function Present in L innocua Reg +-

PrfA box

Distance from start codon Sequence

lmo2219 Similar to PrsA from B subtilis Yes + -206 TTTACAcaTATTAAlmo0788 Unknown Yes + -79 TAAACAacTATTTAlmo0278 Similar to ATP-binding protein Yes ndash -30 TGAACAcaAGTTAAlmo0596 Unknown Yes + -106 TTAAAAggTTTTAAlmo 2067 Bile salt hydrolase No + -147 TTAAAAatTTTTAA

Small capital letters indicate mismatches compared with the consensus PrfA box (TTAACAnnTGTTAA) Gene names correspond to the Lmonocytogenes EGDe gene names on the ListiList server httpgenolistpasteurfrListiList Reg +- regulated positively or negatively respec-tively Numbers indicate the position from the start codon of the gene

Table 3 Putative sigma B promoter regions of group III genes

Gene Description Distance from start codon Putative sigma B promoter sequence

lmo0596 Unknown -103 GTTTTA-N13-GGCTATlmo 2067 Similar to conjugated bile salt hydrolase -65 GTTTTA-N13-GGGTACopuCA Similar to betainecarnitinecholineABC transporter -83 GTTTAA-N14-GGGAAA

opuCB Similar to betainecarnitinecholineABC transporteropuCC Similar to betainecarnitinecholineABC transporteropuCD Similar to betainecarnitinecholineABC transporter

lmo1602 Similar to general stress protein -53 GTTTTA-N14-GGGTATlmo1601 Similar to general stress protein

lmo2748 Similar to B subtilis stress protein YdaG -58 GTTTGA-N14-TGGAAAlmo2230 Similar to arsenate reductase -144 GTTTCT-N13-GGGTAG

lmo2231 Similar to cation efflux systemlmo0913 Similar to succinate semi-aldehyde dehydrogenase -82 GATTAA-N13-TGGAAAlmo0669 Similar to oxidoreductase -175 GTTTTA-N13-GGGAAG

lmo0670 Unknownlmo2695 Similar to dihydroxyacetone kinase -66 GTTTTG-N13-GGGAAA

lmo2696 Similar to hypothetical dihydroxyacetone kinaselmo2697 Unknown

lmo1694 Similar to CDP-abequose synthase -58 GTTTTA-N13-GGGAATlmo0539 Similar to tagatose 16-diphosphate aldolase -86 GTTTTA-N14-TGGTATlmo0784 Similar to mannose specific PTS component IIA -230 GTTTTC-N14-GGGTAA

lmo0783 Similar to mannose specific PTS component IIBlmo0782 Similar to mannose specific PTS component IIClmo0781 Similar to mannose specific PTS component IID

lmo0602 Weakly similar to transcriptional regulator (PaiA) -37 GTTTCA-N13-GTGAAAlmo2391 Similar to B subtilis YhfK protein -61 GTTTTA-N13-GGGAAAlmo0937 Unknown -69 GTTTAA-N13-GGGAATlmo0994 Unknown -63 GTTTAT-N15-GGGAATlmo0794 Unknown -80 GTTTCC-N14-GGGAATlmo2213 Unknown -79 GTTTCA-N13-TGGAAAsepA Unknown -72 GTTTTG-N13-AGGTATlmo1261 Unknown -67 GTTTAA-N14-GGGAATlmo0439 Unknown -65 GTTTCA-N14-GGGAAArsbV Anti-antisigma factor (antagonist of RsbW) -63 GTTTTA-N15-GGGTAA

1618 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

pared its expression profile with that of P14DprfA Expres-sion of prfA in strain P14prfA after growth in BHI isninefold higher than in the P14 wild-type strain thus rep-resenting a significant activation of virulence gene expres-sion (Table 1) Only one (lmo0788) of the two newlyidentified group I genes of EGDe was activated inP14prfA relative to P14DprfA and the negatively regu-lated group II genes of strain EGDe were not repressedin P14prfA relative to P14DprfA (Table 1) Forty-six of the53 positively regulated group III genes identified in strainEGDe after growth in BHI corresponded to the positivelyregulated genes in strain P14prfA One hundred and fiveadditional PrfA activated genes specific to strainP14prfA were also identified (Table S4B) Growth ofstrain P14prfA in BHIC did not result in increased activa-tion of group I genes except for inlA in agreement withprevious results obtained for the hly and plc genes (Ripioet al 1996 1997b) The negatively regulated group IIgenes of strain EGDe remained uninduced by PrfA Acti-vation of all but five (lmo0596 lmo2230 lmo2231lmo0669 lmo0670) of the group III genes was abolishedafter growth in BHIC as in strain EGDe The primary datafor the transcriptome experiments for L monocytogenesP14prfA and P14DprfA are available as Supplementarymaterial (Tables S3A S3B S4A and S4B)

Effect of cellobiose on PrfA-regulated genes

Several studies have shown that the presence and utiliza-tion of different carbohydrates have a remarkable impacton virulence gene expression in L monocytogenes (for areview see Kreft and Vazquez-Boland 2001) For exam-ple metabolizable unphosphorylated sugars have beenshown to inhibit the expression of PrfA-dependent viru-lence genes However when L monocytogenes is grownin the presence of cellobiose transcription of prfA is notinhibited and the amount of PrfA protein is only slightlydecreased (Renzoni et al 1997) We were interested inthe impact of cellobiose on the expression of all the genesregulated by PrfA identified in our study The four L mono-cytogenes strains EGDe EGDeDprfA P14prfA andP14DprfA were grown in BHI supplemented with 25 mMcellobiose Analysis of the expression profiles againrevealed the existence of three groups of genes differentlyregulated by PrfA in the EGDe strain and two groups inthe P14prfA strain As reported prfA itself was still tran-scribed in both strains in the presence of cellobiose In Lmonocytogenes EGDe all the previously known virulencegenes as well as the two newly identified (lmo2219 andlmo0788) group I genes were not activated in the pres-ence of cellobiose The group II genes belonging to thesugar transport operon (lmo0178ndash0184 lmo278) were stillrepressed Surprisingly all but nine of the group III genesremained activated by PrfA in BHICel (Table 1) Therefore

cellobiose in contrast to charcoal abolished the regula-tion of group I genes except for prfA but not that ofgroups II and III further suggesting two different mecha-nisms of regulation by PrfA Analysis of the response ofstrain P14prfA relative to its DprfA counterpart showedthat the presence of cellobiose did not alter the activationof group I genes but that all except 17 of the group IIIgenes remained activated as observed for strain EGDe(Table 1) The primary data and the statistical analysis areavailable as Supplementary material (Tables S5A S5BS6A and S6B)

Confirmation of the macroarray results using real-time quantitative PCR

Real-time quantitative PCR analysis of 18 genes fiverepresentative of the group I genes (hly actA inlAlmo2219 lmo0788) three representative of the group IIgenes (lmo0178 lmo0184 lmo0278) and 10 representa-tive of the group III genes (lmo0596 lmo2067 opuCDlmo1694 lmo2231 lmo0913 lmo0670 lmo2570lmo2697 lmo0781) was conducted to verify the macroar-ray transcription profiling data For group III one generepresentative of a predicted operon was chosen forquantitative PCR The average quantity of DNA moleculespresent in the EGDe wild-type strain grown at 37infinC in BHIrelative to the EGDeDprfA strain and in P14prfA relativeto its DprfA counterpart grown in BHICel was determinedFigure 1 shows that there was a very strong positive cor-relation (r = 082 for EGDe and r = 060 for P14) betweenthe data obtained by the two different techniques Primersused for the PCR are available as Supplementary material(Table S5)

Analysis of the differentially regulated genes

The majority of the newly identified genes code for trans-port proteins proteins involved in stress response andproteins of unknown function Twenty genes encode pro-teins constituting different transport systems Sixteen ofthese are organized in three operons two of which arededicated to carbohydrate transport Operon lmo0784ndashlmo0781 codes for proteins of a phosphoenolpyruvate-dependent phosphotransferase system (PTS) probablydevoted to mannose transport and operon lmo178ndashlmo0184 constitutes a sugar ABC transport systembelonging to the OSP family of ABC systems specific fordi- and oligosaccharides and polyols (Dassa and Bouige2001) lmo0278 coding for the only ABC ATPase presentin the genome of L monocytogenes belonging to the OSPfamily is co-ordinately regulated with lmo178ndashlmo0184The enzymes encoded by this operon are probablyinvolved in the transport of oligo alpha-16 or oligo alpha-14 saccharides (E Dassa personal communication) The

Analysis of the PrfA regulon of L monocytogenes 1619

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

third operon identified is the opuCAndashopuCD operonwhich has been shown to be an ATP-driven osmoregu-lated glycine transport system implicated in the ability ofL monocytogenes to tolerate environments of elevatedosmolarity and reduced temperature (Ko and Smith1999) Fifteen genes encode proteins probably involvedin stress response Among these are gene lmo1601 cod-ing a protein 59 similar to the general stress proteinYtxH from B subtilis and lmo2748 which is 67 similarto the B subtilis stress protein YdaG Gene lmo2230encodes a putative arsenate reductase which might beinvolved in detoxification lmo0913 and lmo0669 encodea potential succinate semi-aldehyde dehydrogenase anda hypothetical oxidoreductase respectively One wouldexpect these enzymes to be involved in the maintenance

of the redox balance of the cell Twenty-one genes encodeproteins of unknown function (Table 1)

Genes lmo2219 and lmo0178 were analysed furtherGene lmo2219 is preceded by a putative PrfA box(Table 2) and belongs to group I genes PrfA activated inBHI and BHIC but not in the presence of cellobiose Itsdeduced protein sequence shows high homology to thelipoprotein PrsA of B subtilis (63 protein similarity overthe entire length) and to the Lactococcus lactis PrtMlipoprotein (52 protein similarity over the entire lengthVos et al 1989) In B subtilis PrsA is located at theouter side of the membrane and is important for therefolding of several mature proteins after their trans-location through the membrane (Jacobs et al 1993Kontinen and Sarvas 1993) We attempted to constructan in frame deletion mutation of lmo2219 in the chromo-some of the EGDe wild-type strain However severalattempts to construct this mutant were unsuccessfulThis might result from the fact that as in B subtilis(Kontinen and Sarvas 1993) PrsA has an essential rolein L monocytogenes

Gene lmo0178 belongs to group II genes which arerepressed by PrfA in the wild-type EGDe It is the firstgene of a putative operon of seven genes which are allrepressed in EGDe Its deduced amino acid sequenceshows high sequence similarity to different xylose repres-sor proteins The highest homology is to XylR of Anaero-cellum thermophilum (54 amino acid similarity over theentire length of the protein) and to XylR of B subtilis (36amino acid similarity over the entire length of the protein)where it acts as a transcriptional repressor of xylose-usingenzymes (Kreuzer et al 1989) To investigate the effectof xylose on gene lmo0178 we performed reverse tran-scription (RT)-PCR assays Equal amounts of total RNAwere extracted from EGDe and the EGDeDprfA mutantstrain grown in BHI with and without 2 xylose Expres-sion of ami which is not regulated by PrfA but isexpressed at a very high level in BHI was used as apositive control Fluorescent analysis of the PCR productsrevealed a five- to 10-fold higher level of the xylR mRNAin EGDeDprfA compared with EGDe whatever the xyloseconcentration (data not shown) Thus xylR is indeedrepressed by PrfA and xylose is not the inducing sugarof this operon as is the case for B subtilis (Kreuzer et al1989)

Discussion

The primary goal of this study was to analyse the PrfAregulon at the level of the complete genome PrfA is themaster regulator of L monocytogenes virulence genes(prfA plcA hly mpl actA plcB inlA inlB inlC and hpt)which are all absent from L innocua the closest non-pathogenic Listeria species Thus we reasoned that the

Fig 1 Correlation of macroarray and quantitative real-time PCR results The fold changes in transcripts present of 18 genes in the EGDe wild-type strain grown at 37infinC in BHI relative to the EGDeDprfA strain (A) and in the P14prfA relative to its DprfA mutant grown in BHICel (B) were log transformed and values were plotted against each other to evaluate their correlation The points on the graph represent genes that were analysed by both methods

1620 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

study of gene expression of all L monocytogenes genesdependent on the presence of PrfA might identify novelvirulence genes or elucidate whether this regulon containsadditional genes This approach identified three groups ofgenes regulated differently by PrfA in strain L monocyto-genes EGDe Group I comprises the 10 known virulencegenes and two newly identified genes (lmo0788 andlmo2219) lmo0788 encodes a protein of unknown func-tion and lmo2219 codes for a protein similar to a post-translocation molecular chaperone In contrast to theknown virulence genes which are all absent from the Linnocua genome these genes are present in L innocuaThe group I genes are PrfA activated in BHI Activation isincreased in BHIC and abolished in BHICel Group IIcomprises eight genes seven (lmo0178ndashlmo0184) ofwhich are organized in an operon constituting a predictedABC transport system involved in sugar transport andmetabolism and are repressed in the EGDe wild-typestrain relative to its prfA deleted mutant independently ofthe conditions tested (BHI BHIC or BHICel) Group IIIcomprises 53 genes of which 44 are also PrfA activatedin BHICel but not in BHIC and only two are absent fromL innocua (lmo2067 and sepA) (Table 1) QuantitativePCR for a subset of genes representative of each of thethree groups (after growth in BHI) showed a strong posi-tive correlation with the macroarray data (Fig 1A)

Comparison of these results with the expression pro-files of a second L monocytogenes strain expressing aconstitutively active form of PrfA (P14prfA) identified allbut one (lmo2219) of the group I genes and all but sevenof the group III genes compared with EGDe (Table 1)DNAndashDNA hybridization of the EGDe macroarray withgenomic DNA from strain P14 indicated that the genesexpressed differently between these two strains areindeed present in both isolates Interestingly the putativesugar transport operon lmo0178ndashlmo0184 and the OSPABC ATPase lmo0278 which are repressed in EGDewere not repressed in the three conditions tested Thismight result from the replacement of the conserved ade-nine at position 13 of the consensus PrfA box sequenceby a guanine in the P14prfA strain which might preventPrfA from acting as a repressor Group I genes were PrfAregulated in the P14prfA strain constitutively overex-pressing PrfA independently of the conditions studiedthus confirming that PrfA in its active conformationbypasses the repressor mechanism triggered by ferment-able sugars (Ripio et al 1997a Brehm et al 1999) Thequantitative PCR results for the P14 strain also showed astrong positive correlation with the macroarray data(Fig 1B)The comparative analysis of the expression pro-files of two different L monocytogenes strains reveals thatdifferences in gene expression and gene regulation existamong different isolates of the same bacterial species

Some of these may have important consequences forglobal virulence

The major findings of our study are that PrfA can act asan activator and as a repressor and that PrfA regulatesdifferent groups of genes depending on the growth condi-tions PrfA is a DNA-binding protein that recognizes a so-called PrfA box to which the PrfAndashRNA polymerase(RNAP) complex binds under appropriate conditions toinitiate transcription (for a review see Kreft and Vazquez-Boland 2001) We identified putative PrfA boxesupstream of all group I genes and one group II gene(lmo0278) The putative PrfA box identified for the ABCATPase lmo0278 is situated at position -30 from the startcodon (Table 2) in agreement with a possible direct rolefor PrfA in repression In contrast a putative PrfA box ispresent upstream of only two group III genes lmo0596encoding a protein of unknown function and lmo2067coding for a bile salt hydrolase recently shown to beinvolved in the intestinal and hepatic phases of listeriosisand indeed PrfA regulated (Dussurget et al 2002)(Table 2) This finding is puzzling as up to now genesshown to be regulated by PrfA contain a PrfA box Theabsence of a PrfA box and other conserved motifs sug-gests indirect regulation of the group III genes by PrfA

Interestingly the group III genes mainly encode proteinsinvolved in different stress responses including a trans-port system involved in cold shock response and osmo-larity stress (opuCAndashopuCD) (Ko and Smith 1999) twoproteins (lmo1602 and lmo1601) similar to general stressresponse proteins or proteins that might be involved in themaintenance of the redox balance of the cell Some homo-logues of these proteins in B subtilis belong to the stress-induced sigma B regulon of B subtilis (Petersohn et al2001 Price et al 2001)

In bacteria alternative sigma factors of RNAP areknown to play a crucial role in regulating gene expressionupon major changes in the environment The alternativesigma factor sigma B was shown to play an importantrole in stress-induced gene regulation and its targets areparticularly well studied in B subtilis where sigma B isactivated when cells encounter growth-limiting energy andenvironmental stresses (Akbar et al 2001) Sigma Bbinds to a specific promoter sequence which has beenidentified for many genes of the B subtilis sigma B regu-lon In B subtilis sigma B itself is under a complex regu-latory network (Hecker et al 1996) In L monocytogenessigma B contributes to survival in conditions of oxidativestress starvation and reduced osmolarity (Ferreira et al2001) and enables L monocytogenes in stationary phaseto adapt and to resume growth at reduced temperature(Becker et al 2000) A search for a probable sigma B-dependent promoter among all the genes identified in thisstudy using a matrix constructed from 47 sigma B pro-

Analysis of the PrfA regulon of L monocytogenes 1621

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

moter sequences of B subtilis and a motif search usingthe sigma B consensus sequence of B subtilis identified22 promoter regions accounting for 33 genes in total assome are organized in operons (Table 3) Genes pre-ceded by a putative sigma B promoter belonged to groupIII and were also included In contrast none of the groupI or group II genes displayed a sigma B-like promoterexcept prfA itself PrfA can be transcribed from two differ-ent promoters The P1 and P2 promoters of prfA are nottranscribed in the same way Depending on the conditionsprfA has been shown to be transcribed from either P1prfAor P2prfA resulting in either a 09 or a 08 kb transcript(Freitag et al 1993 Renzoni et al 1997) Recent geneticevidence suggests that P2prfA is a sigma B-dependentpromoter (Nadon et al 2002) This situation is reminis-cent of that of the global virulence gene regulator SarA inStaphylococcus aureus the transcription of which is atleast partially controlled by sigma B (Deora et al 1997)It can be transcribed from three different promoters ofwhich P1 and P2 have been reported to be sigma Adependent and P3 sigma B dependent (Deora et al1997) Our results strongly suggest an interplay betweenthe PrfA regulon and the sigma B regulon It will be inter-esting to decipher at what stage of the infectious processthe PrfA-regulated genes are expressed Whether othergenes are regulated by PrfA in other growth conditionsalso deserves further investigation

Experimental procedures

Strains and growth conditions

Listeria monocytogenes EGDe and its derivative DprfA42(Bockmann et al 1996) L monocytogenes P14 P14prfA(also named P14-A in other studies) (Ripio et al 1996) andP14DprfA (this study) were used For all experiments a singlecolony of each Listeria strain was grown in BHI (Difco) BHIsupplemented with 02 activated charcoal (BHIC Merck) orBHI supplemented with 25 mM cellobiose (BHICel Sigma)overnight at 37infinC with shaking

Primer design amplification of PCR products and construction of whole-genome macroarrays

Pairs of specific oligonucleotide primers were designed foreach of the 2853 ORFs of the L monocytogenes EGDegenome For primer design a modified version of primer 3which allows testing of the specificity of the obtained PCRproducts against the complete genome sequence to avoidcross-hybridization was used (CAAT-Box L Frangeul et alunpublished) Primers (Metabion) were chosen in order toamplify a fragment of 300ndash600 bp specific for each ORF witha melting temperature of 55ndash65infinC Amplification reactionswere performed in 96-well plates (Perkin-Elmer) in a 100 mlreaction volume containing 20 ng of chromosomal DNA fromL monocytogenes EGDe 2 U of DyNazyme EXT DNA poly-

merase (Finnzymes) 06 mM each primer and 02 mMdNTPs (Perkin-Elmer) Reactions were cycled 35 times(94infinC for 15 s 55infinC for 30 s 72infinC for 60 s) with one finalcycle of 72infinC for 7 min in a thermocycler Successful ampli-fication of each PCR product was verified on agarose gelsand negative PCRs were repeated resulting in the finalamplification of 2816 ORFs of the 2853 ORFs (99) identi-fied in the L monocytogenes EGDe genome For macroarraypreparation nylon membranes (Qfilter Genetix) were soakedin 10 mM TE pH 76 Spot blots of ORF-specific PCR prod-ucts and controls were printed using a Qpix robot (Genetix)Immediately after spot deposition membranes were neutral-ized for 15 min in 05 M NaOH 15 M NaCl washed threetimes with distilled water and stored wet at -20infinC until use

Validation of the macroarrays

To ensure that DNA samples were successfully deposited onthe nylon membranes and to assess differential hybridizationto the target genes 33P-labelled genomic DNA from L mono-cytogenes EGDe was hybridized before transcriptome anal-ysis to the macroarray The total intensity of all pixels withineach spot was determined after scanning of exposed phos-phor screens Fifty-six spots scored as undetectable andwere therefore not taken into account in our study In orderto establish optimal experimental conditions RNA from theEGDe wild-type strain and its prfA-deleted mutant grown inBHI or BHIC was tested To determine the best labellingmethod random hexamers (Boehringer Mannheim) or a mix-ture of all 3cent oligonucleotides (Metabion) each at a concen-tration of 0033 mM specific for each amplified gene wereused Similar results were obtained using the two differentlabelling methods suggesting that the observed distributionreflected the true distribution of transcripts (data not shown)For all further experiments a mixture of specific oligonucle-otides was chosen The intensity distribution patterns showedgood reproducibility (r = 0959) between independent exper-iments In contrast the correlation between the intensity ofspots generated with cDNAs from wild-type cells and thosegenerated with cDNAs from prfA mutant cells showed thatthe mutation caused significant changes in the expressionprofile (r = 0905)

Sample collection cell lysis and total RNA isolation

BHI BHIC or BHICel was inoculated with 100 ml of the over-night culture and incubated at 37infinC with shaking Exponen-tially growing cells (OD600 06ndash08) were harvested for 2 minat 6000 g at 4infinC The pellets were flash frozen on dry icendashethanol and stored at -80infinC For extraction cells were resus-pended by vortexing in 400 ml of resuspension buffer(125 mM Tris 5 mM EDTA and 10 glucose) Then 500 mlof acid phenol (pH 46) and 04 g of glass beads (02ndash03 mmdiameter Sigma) were added The cells were shearedmechanically using a Fastprep apparatus (Bio101) After cen-trifugation at 13 000 g for 5 min the supernatant was trans-ferred to a fresh tube and 1 ml of Trizol reagent (Gibco BRL)was added The sample was incubated for 5 min at roomtemperature Total RNA was extracted twice with chloroformndash

1622 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

isoamyl alcohol (241 vv) and precipitated in 07 volumes ofisopropanol After a washing step with 70 ethanol the RNApellet was dissolved in sterile DNase- and RNase-free water(ICN Biomedicals) and quantified by absorbance at 260 and280 nm Purity and integrity of RNA were controlled on aga-rose gels and RNA was stored at -80infinC until use

Genomic DNA labelling and cDNA labelling

Chromosomal DNA (30 ng) isolated from L monocytogenesstrains EGDe or P14 was 33P-labelled using a random-primedDNA labelling kit (Boehringer Mannheim) Labelled genomicDNAs were purified before hybridization using QIAquick col-umns (Qiagen)

For cDNA synthesis random hexamers or a set of 3cent-specific oligonucleotide primers were used in reversetranscription reactions in the presence of [a-33P]-dCTP(2000ndash3000 Ci mmol-1 Amersham) Total RNA (1 mg) and50 U of AMV reverse transcriptase (Roche) with RNase Hactivity were used Labelled cDNA was purified to removeunincorporated nucleotides before hybridization using aQIAquick column (Qiagen)

Hybridization

Hybridization and washing steps were carried out usingSSPE buffer (018 M NaCl 10 mM NaH2PO4 1 mM EDTApH 77) Macroarrays were prewet in 2yen SSC and prehybrid-ized for at least 2 h in 10 ml of hybridization solution (5yenSSPE 2 SDS 1yen Denhardtrsquos reagent 100 mg of shearedsalmon sperm DNA ml-1) at 65infinC in roller bottles Hybridiza-tion was carried out for 20 h at 65infinC with 5 ml of hybridizationsolution and the entire spin-purified cDNA probe Afterhybridization membranes were washed twice at room tem-perature and twice at 65infinC in 05yen SSC 02 SDS Arrayswere then sealed in thin bags and exposed to a phosphorscreen (Molecular Dynamics) for 24ndash72 h For each straintwo independent RNA preparations were tested and twocDNAs from each of the RNA preparations were hybridizedto two sets of arrays and analysed

Data analysis

Membranes were scanned using a 445SI PhosphorImager(Molecular Dynamics) The ARRAYVISION software (ImagingResearch) was used for quantification of the hybridizationintensities and for normalization The intensity of each spotwas normalized according to the median value of the totalintensities of all spots on each array which allowed directcomparison of the two strains The global background wascalculated from the average intensity of 610 lsquono-DNArsquo spotshomogeneously distributed throughout the membrane Onlyspots for which the sum of the normalized intensity values ofthe wild type and the mutant strain was higher than the sumof the global background of the wild type and the mutantmembrane were taken into account

For identification of genes with statistically significantchanges in expression SAM a statistical technique for findingsignificant genes in a set of microarray experiments was used(Tusher et al 2001) (httpwww-statstanfordedu~tibs

SAM) We chose a delta value corresponding to a falsediscovery rate (FDR) lt10 The FDR was computed as theratio of the estimated number of lsquofalse significantrsquo genes tothe total number of lsquosignificantrsquo genes Genes with a twofoldexpression change that were significant according to thisanalysis were taken into account The complete data set ofthe statistical analysis is available online as Supplementarymaterial

Real-time quantitative PCR

Real-time quantitative PCR was conducted on the same totalcellular RNAs as those used for transcriptome experimentsPrimers (Table S7) were designed with PRIMEREXPRESS soft-ware (Applied Biosystems) and purchased from EurogentecRNA (20 mg) was treated with 50 U of RNase-free DNase I(Roche) for 30 min at 37infinC DNase was inactivated by phenolchloroform extraction and RNA was precipitated cDNA syn-thesis was performed for 60 min at 42infinC using 10 mg of RNA2 ml of AMV reverse transcriptase (25 U ml-1 Roche) and 2 mlof the primer mix To determine the primer concentration withthe best efficiency concentrations of 100 nM 200 nM and400 nM in a 25 ml reaction volume were tested for eachprimer The optimal concentration of 200 nM was used in thefinal experiments Real-time quantitative PCR was performedin a 25 ml reaction volume containing 200 ng of cDNA 125 mlof SYBR PCR master mix (Applied Biosystems) and 2 ml ofgene-specific primers (200 nM) Amplification and detectionof specific products were performed with the ABI Prism 7700sequence detection system (PE Applied Biosystems) with thefollowing cycle profile one cycle at 50infinC for 2 min one cycleat 95infinC for 10 min 40 cycles at 95infinC for 15 s and 60infinC for1 min Each product was deposited on agarose gels to verifythe presence of a single band The quantity of cDNA for eachexperimental gene was normalized to the quantity of rpoBcDNA in each sample For each gene triplicate assays weredone To check whether contaminating chromosomal DNAwas present each sample was tested in control reactions thatdid not contain reverse transcriptase

Acknowledgements

This work received financial support from the EuropeanCommission contract QLG2-CT-1999-00932 the SpanishMinisterio de Ciencia y Technologiacutea (BMC-2000-0553) theMinistere de lrsquoEducation Nationale de la Recherche et de laTechnologie (PFRMMIP) and the Pasteur Institut (PTR 6)EM holds a post-doctoral fellowship from SOREDAB SASF Chetouani we thank for access to unpublished results Wethank the following persons for help in preparing macroar-rays E Zalachas C Lacroix A Schiavo and E CouveacuteSpecial thanks to A Bourdet who took time to help us withthe quantitative PCR experiments Tim Stinear Claire Poyartand Chester Price for critical reading of the manuscript andA Maitournam and M Zouine for help with statistical analy-sis We also thank the reviewers for their constructive com-ments P Cossart is an international investigator from theHoward Hughes Medical Institute

Analysis of the PrfA regulon of L monocytogenes 1619

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

third operon identified is the opuCAndashopuCD operonwhich has been shown to be an ATP-driven osmoregu-lated glycine transport system implicated in the ability ofL monocytogenes to tolerate environments of elevatedosmolarity and reduced temperature (Ko and Smith1999) Fifteen genes encode proteins probably involvedin stress response Among these are gene lmo1601 cod-ing a protein 59 similar to the general stress proteinYtxH from B subtilis and lmo2748 which is 67 similarto the B subtilis stress protein YdaG Gene lmo2230encodes a putative arsenate reductase which might beinvolved in detoxification lmo0913 and lmo0669 encodea potential succinate semi-aldehyde dehydrogenase anda hypothetical oxidoreductase respectively One wouldexpect these enzymes to be involved in the maintenance

of the redox balance of the cell Twenty-one genes encodeproteins of unknown function (Table 1)

Genes lmo2219 and lmo0178 were analysed furtherGene lmo2219 is preceded by a putative PrfA box(Table 2) and belongs to group I genes PrfA activated inBHI and BHIC but not in the presence of cellobiose Itsdeduced protein sequence shows high homology to thelipoprotein PrsA of B subtilis (63 protein similarity overthe entire length) and to the Lactococcus lactis PrtMlipoprotein (52 protein similarity over the entire lengthVos et al 1989) In B subtilis PrsA is located at theouter side of the membrane and is important for therefolding of several mature proteins after their trans-location through the membrane (Jacobs et al 1993Kontinen and Sarvas 1993) We attempted to constructan in frame deletion mutation of lmo2219 in the chromo-some of the EGDe wild-type strain However severalattempts to construct this mutant were unsuccessfulThis might result from the fact that as in B subtilis(Kontinen and Sarvas 1993) PrsA has an essential rolein L monocytogenes

Gene lmo0178 belongs to group II genes which arerepressed by PrfA in the wild-type EGDe It is the firstgene of a putative operon of seven genes which are allrepressed in EGDe Its deduced amino acid sequenceshows high sequence similarity to different xylose repres-sor proteins The highest homology is to XylR of Anaero-cellum thermophilum (54 amino acid similarity over theentire length of the protein) and to XylR of B subtilis (36amino acid similarity over the entire length of the protein)where it acts as a transcriptional repressor of xylose-usingenzymes (Kreuzer et al 1989) To investigate the effectof xylose on gene lmo0178 we performed reverse tran-scription (RT)-PCR assays Equal amounts of total RNAwere extracted from EGDe and the EGDeDprfA mutantstrain grown in BHI with and without 2 xylose Expres-sion of ami which is not regulated by PrfA but isexpressed at a very high level in BHI was used as apositive control Fluorescent analysis of the PCR productsrevealed a five- to 10-fold higher level of the xylR mRNAin EGDeDprfA compared with EGDe whatever the xyloseconcentration (data not shown) Thus xylR is indeedrepressed by PrfA and xylose is not the inducing sugarof this operon as is the case for B subtilis (Kreuzer et al1989)

Discussion

The primary goal of this study was to analyse the PrfAregulon at the level of the complete genome PrfA is themaster regulator of L monocytogenes virulence genes(prfA plcA hly mpl actA plcB inlA inlB inlC and hpt)which are all absent from L innocua the closest non-pathogenic Listeria species Thus we reasoned that the

Fig 1 Correlation of macroarray and quantitative real-time PCR results The fold changes in transcripts present of 18 genes in the EGDe wild-type strain grown at 37infinC in BHI relative to the EGDeDprfA strain (A) and in the P14prfA relative to its DprfA mutant grown in BHICel (B) were log transformed and values were plotted against each other to evaluate their correlation The points on the graph represent genes that were analysed by both methods

1620 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

study of gene expression of all L monocytogenes genesdependent on the presence of PrfA might identify novelvirulence genes or elucidate whether this regulon containsadditional genes This approach identified three groups ofgenes regulated differently by PrfA in strain L monocyto-genes EGDe Group I comprises the 10 known virulencegenes and two newly identified genes (lmo0788 andlmo2219) lmo0788 encodes a protein of unknown func-tion and lmo2219 codes for a protein similar to a post-translocation molecular chaperone In contrast to theknown virulence genes which are all absent from the Linnocua genome these genes are present in L innocuaThe group I genes are PrfA activated in BHI Activation isincreased in BHIC and abolished in BHICel Group IIcomprises eight genes seven (lmo0178ndashlmo0184) ofwhich are organized in an operon constituting a predictedABC transport system involved in sugar transport andmetabolism and are repressed in the EGDe wild-typestrain relative to its prfA deleted mutant independently ofthe conditions tested (BHI BHIC or BHICel) Group IIIcomprises 53 genes of which 44 are also PrfA activatedin BHICel but not in BHIC and only two are absent fromL innocua (lmo2067 and sepA) (Table 1) QuantitativePCR for a subset of genes representative of each of thethree groups (after growth in BHI) showed a strong posi-tive correlation with the macroarray data (Fig 1A)

Comparison of these results with the expression pro-files of a second L monocytogenes strain expressing aconstitutively active form of PrfA (P14prfA) identified allbut one (lmo2219) of the group I genes and all but sevenof the group III genes compared with EGDe (Table 1)DNAndashDNA hybridization of the EGDe macroarray withgenomic DNA from strain P14 indicated that the genesexpressed differently between these two strains areindeed present in both isolates Interestingly the putativesugar transport operon lmo0178ndashlmo0184 and the OSPABC ATPase lmo0278 which are repressed in EGDewere not repressed in the three conditions tested Thismight result from the replacement of the conserved ade-nine at position 13 of the consensus PrfA box sequenceby a guanine in the P14prfA strain which might preventPrfA from acting as a repressor Group I genes were PrfAregulated in the P14prfA strain constitutively overex-pressing PrfA independently of the conditions studiedthus confirming that PrfA in its active conformationbypasses the repressor mechanism triggered by ferment-able sugars (Ripio et al 1997a Brehm et al 1999) Thequantitative PCR results for the P14 strain also showed astrong positive correlation with the macroarray data(Fig 1B)The comparative analysis of the expression pro-files of two different L monocytogenes strains reveals thatdifferences in gene expression and gene regulation existamong different isolates of the same bacterial species

Some of these may have important consequences forglobal virulence

The major findings of our study are that PrfA can act asan activator and as a repressor and that PrfA regulatesdifferent groups of genes depending on the growth condi-tions PrfA is a DNA-binding protein that recognizes a so-called PrfA box to which the PrfAndashRNA polymerase(RNAP) complex binds under appropriate conditions toinitiate transcription (for a review see Kreft and Vazquez-Boland 2001) We identified putative PrfA boxesupstream of all group I genes and one group II gene(lmo0278) The putative PrfA box identified for the ABCATPase lmo0278 is situated at position -30 from the startcodon (Table 2) in agreement with a possible direct rolefor PrfA in repression In contrast a putative PrfA box ispresent upstream of only two group III genes lmo0596encoding a protein of unknown function and lmo2067coding for a bile salt hydrolase recently shown to beinvolved in the intestinal and hepatic phases of listeriosisand indeed PrfA regulated (Dussurget et al 2002)(Table 2) This finding is puzzling as up to now genesshown to be regulated by PrfA contain a PrfA box Theabsence of a PrfA box and other conserved motifs sug-gests indirect regulation of the group III genes by PrfA

Interestingly the group III genes mainly encode proteinsinvolved in different stress responses including a trans-port system involved in cold shock response and osmo-larity stress (opuCAndashopuCD) (Ko and Smith 1999) twoproteins (lmo1602 and lmo1601) similar to general stressresponse proteins or proteins that might be involved in themaintenance of the redox balance of the cell Some homo-logues of these proteins in B subtilis belong to the stress-induced sigma B regulon of B subtilis (Petersohn et al2001 Price et al 2001)

In bacteria alternative sigma factors of RNAP areknown to play a crucial role in regulating gene expressionupon major changes in the environment The alternativesigma factor sigma B was shown to play an importantrole in stress-induced gene regulation and its targets areparticularly well studied in B subtilis where sigma B isactivated when cells encounter growth-limiting energy andenvironmental stresses (Akbar et al 2001) Sigma Bbinds to a specific promoter sequence which has beenidentified for many genes of the B subtilis sigma B regu-lon In B subtilis sigma B itself is under a complex regu-latory network (Hecker et al 1996) In L monocytogenessigma B contributes to survival in conditions of oxidativestress starvation and reduced osmolarity (Ferreira et al2001) and enables L monocytogenes in stationary phaseto adapt and to resume growth at reduced temperature(Becker et al 2000) A search for a probable sigma B-dependent promoter among all the genes identified in thisstudy using a matrix constructed from 47 sigma B pro-

Analysis of the PrfA regulon of L monocytogenes 1621

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

moter sequences of B subtilis and a motif search usingthe sigma B consensus sequence of B subtilis identified22 promoter regions accounting for 33 genes in total assome are organized in operons (Table 3) Genes pre-ceded by a putative sigma B promoter belonged to groupIII and were also included In contrast none of the groupI or group II genes displayed a sigma B-like promoterexcept prfA itself PrfA can be transcribed from two differ-ent promoters The P1 and P2 promoters of prfA are nottranscribed in the same way Depending on the conditionsprfA has been shown to be transcribed from either P1prfAor P2prfA resulting in either a 09 or a 08 kb transcript(Freitag et al 1993 Renzoni et al 1997) Recent geneticevidence suggests that P2prfA is a sigma B-dependentpromoter (Nadon et al 2002) This situation is reminis-cent of that of the global virulence gene regulator SarA inStaphylococcus aureus the transcription of which is atleast partially controlled by sigma B (Deora et al 1997)It can be transcribed from three different promoters ofwhich P1 and P2 have been reported to be sigma Adependent and P3 sigma B dependent (Deora et al1997) Our results strongly suggest an interplay betweenthe PrfA regulon and the sigma B regulon It will be inter-esting to decipher at what stage of the infectious processthe PrfA-regulated genes are expressed Whether othergenes are regulated by PrfA in other growth conditionsalso deserves further investigation

Experimental procedures

Strains and growth conditions

Listeria monocytogenes EGDe and its derivative DprfA42(Bockmann et al 1996) L monocytogenes P14 P14prfA(also named P14-A in other studies) (Ripio et al 1996) andP14DprfA (this study) were used For all experiments a singlecolony of each Listeria strain was grown in BHI (Difco) BHIsupplemented with 02 activated charcoal (BHIC Merck) orBHI supplemented with 25 mM cellobiose (BHICel Sigma)overnight at 37infinC with shaking

Primer design amplification of PCR products and construction of whole-genome macroarrays

Pairs of specific oligonucleotide primers were designed foreach of the 2853 ORFs of the L monocytogenes EGDegenome For primer design a modified version of primer 3which allows testing of the specificity of the obtained PCRproducts against the complete genome sequence to avoidcross-hybridization was used (CAAT-Box L Frangeul et alunpublished) Primers (Metabion) were chosen in order toamplify a fragment of 300ndash600 bp specific for each ORF witha melting temperature of 55ndash65infinC Amplification reactionswere performed in 96-well plates (Perkin-Elmer) in a 100 mlreaction volume containing 20 ng of chromosomal DNA fromL monocytogenes EGDe 2 U of DyNazyme EXT DNA poly-

merase (Finnzymes) 06 mM each primer and 02 mMdNTPs (Perkin-Elmer) Reactions were cycled 35 times(94infinC for 15 s 55infinC for 30 s 72infinC for 60 s) with one finalcycle of 72infinC for 7 min in a thermocycler Successful ampli-fication of each PCR product was verified on agarose gelsand negative PCRs were repeated resulting in the finalamplification of 2816 ORFs of the 2853 ORFs (99) identi-fied in the L monocytogenes EGDe genome For macroarraypreparation nylon membranes (Qfilter Genetix) were soakedin 10 mM TE pH 76 Spot blots of ORF-specific PCR prod-ucts and controls were printed using a Qpix robot (Genetix)Immediately after spot deposition membranes were neutral-ized for 15 min in 05 M NaOH 15 M NaCl washed threetimes with distilled water and stored wet at -20infinC until use

Validation of the macroarrays

To ensure that DNA samples were successfully deposited onthe nylon membranes and to assess differential hybridizationto the target genes 33P-labelled genomic DNA from L mono-cytogenes EGDe was hybridized before transcriptome anal-ysis to the macroarray The total intensity of all pixels withineach spot was determined after scanning of exposed phos-phor screens Fifty-six spots scored as undetectable andwere therefore not taken into account in our study In orderto establish optimal experimental conditions RNA from theEGDe wild-type strain and its prfA-deleted mutant grown inBHI or BHIC was tested To determine the best labellingmethod random hexamers (Boehringer Mannheim) or a mix-ture of all 3cent oligonucleotides (Metabion) each at a concen-tration of 0033 mM specific for each amplified gene wereused Similar results were obtained using the two differentlabelling methods suggesting that the observed distributionreflected the true distribution of transcripts (data not shown)For all further experiments a mixture of specific oligonucle-otides was chosen The intensity distribution patterns showedgood reproducibility (r = 0959) between independent exper-iments In contrast the correlation between the intensity ofspots generated with cDNAs from wild-type cells and thosegenerated with cDNAs from prfA mutant cells showed thatthe mutation caused significant changes in the expressionprofile (r = 0905)

Sample collection cell lysis and total RNA isolation

BHI BHIC or BHICel was inoculated with 100 ml of the over-night culture and incubated at 37infinC with shaking Exponen-tially growing cells (OD600 06ndash08) were harvested for 2 minat 6000 g at 4infinC The pellets were flash frozen on dry icendashethanol and stored at -80infinC For extraction cells were resus-pended by vortexing in 400 ml of resuspension buffer(125 mM Tris 5 mM EDTA and 10 glucose) Then 500 mlof acid phenol (pH 46) and 04 g of glass beads (02ndash03 mmdiameter Sigma) were added The cells were shearedmechanically using a Fastprep apparatus (Bio101) After cen-trifugation at 13 000 g for 5 min the supernatant was trans-ferred to a fresh tube and 1 ml of Trizol reagent (Gibco BRL)was added The sample was incubated for 5 min at roomtemperature Total RNA was extracted twice with chloroformndash

1622 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

isoamyl alcohol (241 vv) and precipitated in 07 volumes ofisopropanol After a washing step with 70 ethanol the RNApellet was dissolved in sterile DNase- and RNase-free water(ICN Biomedicals) and quantified by absorbance at 260 and280 nm Purity and integrity of RNA were controlled on aga-rose gels and RNA was stored at -80infinC until use

Genomic DNA labelling and cDNA labelling

Chromosomal DNA (30 ng) isolated from L monocytogenesstrains EGDe or P14 was 33P-labelled using a random-primedDNA labelling kit (Boehringer Mannheim) Labelled genomicDNAs were purified before hybridization using QIAquick col-umns (Qiagen)

For cDNA synthesis random hexamers or a set of 3cent-specific oligonucleotide primers were used in reversetranscription reactions in the presence of [a-33P]-dCTP(2000ndash3000 Ci mmol-1 Amersham) Total RNA (1 mg) and50 U of AMV reverse transcriptase (Roche) with RNase Hactivity were used Labelled cDNA was purified to removeunincorporated nucleotides before hybridization using aQIAquick column (Qiagen)

Hybridization

Hybridization and washing steps were carried out usingSSPE buffer (018 M NaCl 10 mM NaH2PO4 1 mM EDTApH 77) Macroarrays were prewet in 2yen SSC and prehybrid-ized for at least 2 h in 10 ml of hybridization solution (5yenSSPE 2 SDS 1yen Denhardtrsquos reagent 100 mg of shearedsalmon sperm DNA ml-1) at 65infinC in roller bottles Hybridiza-tion was carried out for 20 h at 65infinC with 5 ml of hybridizationsolution and the entire spin-purified cDNA probe Afterhybridization membranes were washed twice at room tem-perature and twice at 65infinC in 05yen SSC 02 SDS Arrayswere then sealed in thin bags and exposed to a phosphorscreen (Molecular Dynamics) for 24ndash72 h For each straintwo independent RNA preparations were tested and twocDNAs from each of the RNA preparations were hybridizedto two sets of arrays and analysed

Data analysis

Membranes were scanned using a 445SI PhosphorImager(Molecular Dynamics) The ARRAYVISION software (ImagingResearch) was used for quantification of the hybridizationintensities and for normalization The intensity of each spotwas normalized according to the median value of the totalintensities of all spots on each array which allowed directcomparison of the two strains The global background wascalculated from the average intensity of 610 lsquono-DNArsquo spotshomogeneously distributed throughout the membrane Onlyspots for which the sum of the normalized intensity values ofthe wild type and the mutant strain was higher than the sumof the global background of the wild type and the mutantmembrane were taken into account

For identification of genes with statistically significantchanges in expression SAM a statistical technique for findingsignificant genes in a set of microarray experiments was used(Tusher et al 2001) (httpwww-statstanfordedu~tibs

SAM) We chose a delta value corresponding to a falsediscovery rate (FDR) lt10 The FDR was computed as theratio of the estimated number of lsquofalse significantrsquo genes tothe total number of lsquosignificantrsquo genes Genes with a twofoldexpression change that were significant according to thisanalysis were taken into account The complete data set ofthe statistical analysis is available online as Supplementarymaterial

Real-time quantitative PCR

Real-time quantitative PCR was conducted on the same totalcellular RNAs as those used for transcriptome experimentsPrimers (Table S7) were designed with PRIMEREXPRESS soft-ware (Applied Biosystems) and purchased from EurogentecRNA (20 mg) was treated with 50 U of RNase-free DNase I(Roche) for 30 min at 37infinC DNase was inactivated by phenolchloroform extraction and RNA was precipitated cDNA syn-thesis was performed for 60 min at 42infinC using 10 mg of RNA2 ml of AMV reverse transcriptase (25 U ml-1 Roche) and 2 mlof the primer mix To determine the primer concentration withthe best efficiency concentrations of 100 nM 200 nM and400 nM in a 25 ml reaction volume were tested for eachprimer The optimal concentration of 200 nM was used in thefinal experiments Real-time quantitative PCR was performedin a 25 ml reaction volume containing 200 ng of cDNA 125 mlof SYBR PCR master mix (Applied Biosystems) and 2 ml ofgene-specific primers (200 nM) Amplification and detectionof specific products were performed with the ABI Prism 7700sequence detection system (PE Applied Biosystems) with thefollowing cycle profile one cycle at 50infinC for 2 min one cycleat 95infinC for 10 min 40 cycles at 95infinC for 15 s and 60infinC for1 min Each product was deposited on agarose gels to verifythe presence of a single band The quantity of cDNA for eachexperimental gene was normalized to the quantity of rpoBcDNA in each sample For each gene triplicate assays weredone To check whether contaminating chromosomal DNAwas present each sample was tested in control reactions thatdid not contain reverse transcriptase

Acknowledgements

This work received financial support from the EuropeanCommission contract QLG2-CT-1999-00932 the SpanishMinisterio de Ciencia y Technologiacutea (BMC-2000-0553) theMinistere de lrsquoEducation Nationale de la Recherche et de laTechnologie (PFRMMIP) and the Pasteur Institut (PTR 6)EM holds a post-doctoral fellowship from SOREDAB SASF Chetouani we thank for access to unpublished results Wethank the following persons for help in preparing macroar-rays E Zalachas C Lacroix A Schiavo and E CouveacuteSpecial thanks to A Bourdet who took time to help us withthe quantitative PCR experiments Tim Stinear Claire Poyartand Chester Price for critical reading of the manuscript andA Maitournam and M Zouine for help with statistical analy-sis We also thank the reviewers for their constructive com-ments P Cossart is an international investigator from theHoward Hughes Medical Institute

1620 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

study of gene expression of all L monocytogenes genesdependent on the presence of PrfA might identify novelvirulence genes or elucidate whether this regulon containsadditional genes This approach identified three groups ofgenes regulated differently by PrfA in strain L monocyto-genes EGDe Group I comprises the 10 known virulencegenes and two newly identified genes (lmo0788 andlmo2219) lmo0788 encodes a protein of unknown func-tion and lmo2219 codes for a protein similar to a post-translocation molecular chaperone In contrast to theknown virulence genes which are all absent from the Linnocua genome these genes are present in L innocuaThe group I genes are PrfA activated in BHI Activation isincreased in BHIC and abolished in BHICel Group IIcomprises eight genes seven (lmo0178ndashlmo0184) ofwhich are organized in an operon constituting a predictedABC transport system involved in sugar transport andmetabolism and are repressed in the EGDe wild-typestrain relative to its prfA deleted mutant independently ofthe conditions tested (BHI BHIC or BHICel) Group IIIcomprises 53 genes of which 44 are also PrfA activatedin BHICel but not in BHIC and only two are absent fromL innocua (lmo2067 and sepA) (Table 1) QuantitativePCR for a subset of genes representative of each of thethree groups (after growth in BHI) showed a strong posi-tive correlation with the macroarray data (Fig 1A)

Comparison of these results with the expression pro-files of a second L monocytogenes strain expressing aconstitutively active form of PrfA (P14prfA) identified allbut one (lmo2219) of the group I genes and all but sevenof the group III genes compared with EGDe (Table 1)DNAndashDNA hybridization of the EGDe macroarray withgenomic DNA from strain P14 indicated that the genesexpressed differently between these two strains areindeed present in both isolates Interestingly the putativesugar transport operon lmo0178ndashlmo0184 and the OSPABC ATPase lmo0278 which are repressed in EGDewere not repressed in the three conditions tested Thismight result from the replacement of the conserved ade-nine at position 13 of the consensus PrfA box sequenceby a guanine in the P14prfA strain which might preventPrfA from acting as a repressor Group I genes were PrfAregulated in the P14prfA strain constitutively overex-pressing PrfA independently of the conditions studiedthus confirming that PrfA in its active conformationbypasses the repressor mechanism triggered by ferment-able sugars (Ripio et al 1997a Brehm et al 1999) Thequantitative PCR results for the P14 strain also showed astrong positive correlation with the macroarray data(Fig 1B)The comparative analysis of the expression pro-files of two different L monocytogenes strains reveals thatdifferences in gene expression and gene regulation existamong different isolates of the same bacterial species

Some of these may have important consequences forglobal virulence

The major findings of our study are that PrfA can act asan activator and as a repressor and that PrfA regulatesdifferent groups of genes depending on the growth condi-tions PrfA is a DNA-binding protein that recognizes a so-called PrfA box to which the PrfAndashRNA polymerase(RNAP) complex binds under appropriate conditions toinitiate transcription (for a review see Kreft and Vazquez-Boland 2001) We identified putative PrfA boxesupstream of all group I genes and one group II gene(lmo0278) The putative PrfA box identified for the ABCATPase lmo0278 is situated at position -30 from the startcodon (Table 2) in agreement with a possible direct rolefor PrfA in repression In contrast a putative PrfA box ispresent upstream of only two group III genes lmo0596encoding a protein of unknown function and lmo2067coding for a bile salt hydrolase recently shown to beinvolved in the intestinal and hepatic phases of listeriosisand indeed PrfA regulated (Dussurget et al 2002)(Table 2) This finding is puzzling as up to now genesshown to be regulated by PrfA contain a PrfA box Theabsence of a PrfA box and other conserved motifs sug-gests indirect regulation of the group III genes by PrfA

Interestingly the group III genes mainly encode proteinsinvolved in different stress responses including a trans-port system involved in cold shock response and osmo-larity stress (opuCAndashopuCD) (Ko and Smith 1999) twoproteins (lmo1602 and lmo1601) similar to general stressresponse proteins or proteins that might be involved in themaintenance of the redox balance of the cell Some homo-logues of these proteins in B subtilis belong to the stress-induced sigma B regulon of B subtilis (Petersohn et al2001 Price et al 2001)

In bacteria alternative sigma factors of RNAP areknown to play a crucial role in regulating gene expressionupon major changes in the environment The alternativesigma factor sigma B was shown to play an importantrole in stress-induced gene regulation and its targets areparticularly well studied in B subtilis where sigma B isactivated when cells encounter growth-limiting energy andenvironmental stresses (Akbar et al 2001) Sigma Bbinds to a specific promoter sequence which has beenidentified for many genes of the B subtilis sigma B regu-lon In B subtilis sigma B itself is under a complex regu-latory network (Hecker et al 1996) In L monocytogenessigma B contributes to survival in conditions of oxidativestress starvation and reduced osmolarity (Ferreira et al2001) and enables L monocytogenes in stationary phaseto adapt and to resume growth at reduced temperature(Becker et al 2000) A search for a probable sigma B-dependent promoter among all the genes identified in thisstudy using a matrix constructed from 47 sigma B pro-

Analysis of the PrfA regulon of L monocytogenes 1621

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

moter sequences of B subtilis and a motif search usingthe sigma B consensus sequence of B subtilis identified22 promoter regions accounting for 33 genes in total assome are organized in operons (Table 3) Genes pre-ceded by a putative sigma B promoter belonged to groupIII and were also included In contrast none of the groupI or group II genes displayed a sigma B-like promoterexcept prfA itself PrfA can be transcribed from two differ-ent promoters The P1 and P2 promoters of prfA are nottranscribed in the same way Depending on the conditionsprfA has been shown to be transcribed from either P1prfAor P2prfA resulting in either a 09 or a 08 kb transcript(Freitag et al 1993 Renzoni et al 1997) Recent geneticevidence suggests that P2prfA is a sigma B-dependentpromoter (Nadon et al 2002) This situation is reminis-cent of that of the global virulence gene regulator SarA inStaphylococcus aureus the transcription of which is atleast partially controlled by sigma B (Deora et al 1997)It can be transcribed from three different promoters ofwhich P1 and P2 have been reported to be sigma Adependent and P3 sigma B dependent (Deora et al1997) Our results strongly suggest an interplay betweenthe PrfA regulon and the sigma B regulon It will be inter-esting to decipher at what stage of the infectious processthe PrfA-regulated genes are expressed Whether othergenes are regulated by PrfA in other growth conditionsalso deserves further investigation

Experimental procedures

Strains and growth conditions

Listeria monocytogenes EGDe and its derivative DprfA42(Bockmann et al 1996) L monocytogenes P14 P14prfA(also named P14-A in other studies) (Ripio et al 1996) andP14DprfA (this study) were used For all experiments a singlecolony of each Listeria strain was grown in BHI (Difco) BHIsupplemented with 02 activated charcoal (BHIC Merck) orBHI supplemented with 25 mM cellobiose (BHICel Sigma)overnight at 37infinC with shaking

Primer design amplification of PCR products and construction of whole-genome macroarrays

Pairs of specific oligonucleotide primers were designed foreach of the 2853 ORFs of the L monocytogenes EGDegenome For primer design a modified version of primer 3which allows testing of the specificity of the obtained PCRproducts against the complete genome sequence to avoidcross-hybridization was used (CAAT-Box L Frangeul et alunpublished) Primers (Metabion) were chosen in order toamplify a fragment of 300ndash600 bp specific for each ORF witha melting temperature of 55ndash65infinC Amplification reactionswere performed in 96-well plates (Perkin-Elmer) in a 100 mlreaction volume containing 20 ng of chromosomal DNA fromL monocytogenes EGDe 2 U of DyNazyme EXT DNA poly-

merase (Finnzymes) 06 mM each primer and 02 mMdNTPs (Perkin-Elmer) Reactions were cycled 35 times(94infinC for 15 s 55infinC for 30 s 72infinC for 60 s) with one finalcycle of 72infinC for 7 min in a thermocycler Successful ampli-fication of each PCR product was verified on agarose gelsand negative PCRs were repeated resulting in the finalamplification of 2816 ORFs of the 2853 ORFs (99) identi-fied in the L monocytogenes EGDe genome For macroarraypreparation nylon membranes (Qfilter Genetix) were soakedin 10 mM TE pH 76 Spot blots of ORF-specific PCR prod-ucts and controls were printed using a Qpix robot (Genetix)Immediately after spot deposition membranes were neutral-ized for 15 min in 05 M NaOH 15 M NaCl washed threetimes with distilled water and stored wet at -20infinC until use

Validation of the macroarrays

To ensure that DNA samples were successfully deposited onthe nylon membranes and to assess differential hybridizationto the target genes 33P-labelled genomic DNA from L mono-cytogenes EGDe was hybridized before transcriptome anal-ysis to the macroarray The total intensity of all pixels withineach spot was determined after scanning of exposed phos-phor screens Fifty-six spots scored as undetectable andwere therefore not taken into account in our study In orderto establish optimal experimental conditions RNA from theEGDe wild-type strain and its prfA-deleted mutant grown inBHI or BHIC was tested To determine the best labellingmethod random hexamers (Boehringer Mannheim) or a mix-ture of all 3cent oligonucleotides (Metabion) each at a concen-tration of 0033 mM specific for each amplified gene wereused Similar results were obtained using the two differentlabelling methods suggesting that the observed distributionreflected the true distribution of transcripts (data not shown)For all further experiments a mixture of specific oligonucle-otides was chosen The intensity distribution patterns showedgood reproducibility (r = 0959) between independent exper-iments In contrast the correlation between the intensity ofspots generated with cDNAs from wild-type cells and thosegenerated with cDNAs from prfA mutant cells showed thatthe mutation caused significant changes in the expressionprofile (r = 0905)

Sample collection cell lysis and total RNA isolation

BHI BHIC or BHICel was inoculated with 100 ml of the over-night culture and incubated at 37infinC with shaking Exponen-tially growing cells (OD600 06ndash08) were harvested for 2 minat 6000 g at 4infinC The pellets were flash frozen on dry icendashethanol and stored at -80infinC For extraction cells were resus-pended by vortexing in 400 ml of resuspension buffer(125 mM Tris 5 mM EDTA and 10 glucose) Then 500 mlof acid phenol (pH 46) and 04 g of glass beads (02ndash03 mmdiameter Sigma) were added The cells were shearedmechanically using a Fastprep apparatus (Bio101) After cen-trifugation at 13 000 g for 5 min the supernatant was trans-ferred to a fresh tube and 1 ml of Trizol reagent (Gibco BRL)was added The sample was incubated for 5 min at roomtemperature Total RNA was extracted twice with chloroformndash

1622 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

isoamyl alcohol (241 vv) and precipitated in 07 volumes ofisopropanol After a washing step with 70 ethanol the RNApellet was dissolved in sterile DNase- and RNase-free water(ICN Biomedicals) and quantified by absorbance at 260 and280 nm Purity and integrity of RNA were controlled on aga-rose gels and RNA was stored at -80infinC until use

Genomic DNA labelling and cDNA labelling

Chromosomal DNA (30 ng) isolated from L monocytogenesstrains EGDe or P14 was 33P-labelled using a random-primedDNA labelling kit (Boehringer Mannheim) Labelled genomicDNAs were purified before hybridization using QIAquick col-umns (Qiagen)

For cDNA synthesis random hexamers or a set of 3cent-specific oligonucleotide primers were used in reversetranscription reactions in the presence of [a-33P]-dCTP(2000ndash3000 Ci mmol-1 Amersham) Total RNA (1 mg) and50 U of AMV reverse transcriptase (Roche) with RNase Hactivity were used Labelled cDNA was purified to removeunincorporated nucleotides before hybridization using aQIAquick column (Qiagen)

Hybridization

Hybridization and washing steps were carried out usingSSPE buffer (018 M NaCl 10 mM NaH2PO4 1 mM EDTApH 77) Macroarrays were prewet in 2yen SSC and prehybrid-ized for at least 2 h in 10 ml of hybridization solution (5yenSSPE 2 SDS 1yen Denhardtrsquos reagent 100 mg of shearedsalmon sperm DNA ml-1) at 65infinC in roller bottles Hybridiza-tion was carried out for 20 h at 65infinC with 5 ml of hybridizationsolution and the entire spin-purified cDNA probe Afterhybridization membranes were washed twice at room tem-perature and twice at 65infinC in 05yen SSC 02 SDS Arrayswere then sealed in thin bags and exposed to a phosphorscreen (Molecular Dynamics) for 24ndash72 h For each straintwo independent RNA preparations were tested and twocDNAs from each of the RNA preparations were hybridizedto two sets of arrays and analysed

Data analysis

Membranes were scanned using a 445SI PhosphorImager(Molecular Dynamics) The ARRAYVISION software (ImagingResearch) was used for quantification of the hybridizationintensities and for normalization The intensity of each spotwas normalized according to the median value of the totalintensities of all spots on each array which allowed directcomparison of the two strains The global background wascalculated from the average intensity of 610 lsquono-DNArsquo spotshomogeneously distributed throughout the membrane Onlyspots for which the sum of the normalized intensity values ofthe wild type and the mutant strain was higher than the sumof the global background of the wild type and the mutantmembrane were taken into account

For identification of genes with statistically significantchanges in expression SAM a statistical technique for findingsignificant genes in a set of microarray experiments was used(Tusher et al 2001) (httpwww-statstanfordedu~tibs

SAM) We chose a delta value corresponding to a falsediscovery rate (FDR) lt10 The FDR was computed as theratio of the estimated number of lsquofalse significantrsquo genes tothe total number of lsquosignificantrsquo genes Genes with a twofoldexpression change that were significant according to thisanalysis were taken into account The complete data set ofthe statistical analysis is available online as Supplementarymaterial

Real-time quantitative PCR

Real-time quantitative PCR was conducted on the same totalcellular RNAs as those used for transcriptome experimentsPrimers (Table S7) were designed with PRIMEREXPRESS soft-ware (Applied Biosystems) and purchased from EurogentecRNA (20 mg) was treated with 50 U of RNase-free DNase I(Roche) for 30 min at 37infinC DNase was inactivated by phenolchloroform extraction and RNA was precipitated cDNA syn-thesis was performed for 60 min at 42infinC using 10 mg of RNA2 ml of AMV reverse transcriptase (25 U ml-1 Roche) and 2 mlof the primer mix To determine the primer concentration withthe best efficiency concentrations of 100 nM 200 nM and400 nM in a 25 ml reaction volume were tested for eachprimer The optimal concentration of 200 nM was used in thefinal experiments Real-time quantitative PCR was performedin a 25 ml reaction volume containing 200 ng of cDNA 125 mlof SYBR PCR master mix (Applied Biosystems) and 2 ml ofgene-specific primers (200 nM) Amplification and detectionof specific products were performed with the ABI Prism 7700sequence detection system (PE Applied Biosystems) with thefollowing cycle profile one cycle at 50infinC for 2 min one cycleat 95infinC for 10 min 40 cycles at 95infinC for 15 s and 60infinC for1 min Each product was deposited on agarose gels to verifythe presence of a single band The quantity of cDNA for eachexperimental gene was normalized to the quantity of rpoBcDNA in each sample For each gene triplicate assays weredone To check whether contaminating chromosomal DNAwas present each sample was tested in control reactions thatdid not contain reverse transcriptase

Acknowledgements

This work received financial support from the EuropeanCommission contract QLG2-CT-1999-00932 the SpanishMinisterio de Ciencia y Technologiacutea (BMC-2000-0553) theMinistere de lrsquoEducation Nationale de la Recherche et de laTechnologie (PFRMMIP) and the Pasteur Institut (PTR 6)EM holds a post-doctoral fellowship from SOREDAB SASF Chetouani we thank for access to unpublished results Wethank the following persons for help in preparing macroar-rays E Zalachas C Lacroix A Schiavo and E CouveacuteSpecial thanks to A Bourdet who took time to help us withthe quantitative PCR experiments Tim Stinear Claire Poyartand Chester Price for critical reading of the manuscript andA Maitournam and M Zouine for help with statistical analy-sis We also thank the reviewers for their constructive com-ments P Cossart is an international investigator from theHoward Hughes Medical Institute

Analysis of the PrfA regulon of L monocytogenes 1621

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

moter sequences of B subtilis and a motif search usingthe sigma B consensus sequence of B subtilis identified22 promoter regions accounting for 33 genes in total assome are organized in operons (Table 3) Genes pre-ceded by a putative sigma B promoter belonged to groupIII and were also included In contrast none of the groupI or group II genes displayed a sigma B-like promoterexcept prfA itself PrfA can be transcribed from two differ-ent promoters The P1 and P2 promoters of prfA are nottranscribed in the same way Depending on the conditionsprfA has been shown to be transcribed from either P1prfAor P2prfA resulting in either a 09 or a 08 kb transcript(Freitag et al 1993 Renzoni et al 1997) Recent geneticevidence suggests that P2prfA is a sigma B-dependentpromoter (Nadon et al 2002) This situation is reminis-cent of that of the global virulence gene regulator SarA inStaphylococcus aureus the transcription of which is atleast partially controlled by sigma B (Deora et al 1997)It can be transcribed from three different promoters ofwhich P1 and P2 have been reported to be sigma Adependent and P3 sigma B dependent (Deora et al1997) Our results strongly suggest an interplay betweenthe PrfA regulon and the sigma B regulon It will be inter-esting to decipher at what stage of the infectious processthe PrfA-regulated genes are expressed Whether othergenes are regulated by PrfA in other growth conditionsalso deserves further investigation

Experimental procedures

Strains and growth conditions

Listeria monocytogenes EGDe and its derivative DprfA42(Bockmann et al 1996) L monocytogenes P14 P14prfA(also named P14-A in other studies) (Ripio et al 1996) andP14DprfA (this study) were used For all experiments a singlecolony of each Listeria strain was grown in BHI (Difco) BHIsupplemented with 02 activated charcoal (BHIC Merck) orBHI supplemented with 25 mM cellobiose (BHICel Sigma)overnight at 37infinC with shaking

Primer design amplification of PCR products and construction of whole-genome macroarrays

Pairs of specific oligonucleotide primers were designed foreach of the 2853 ORFs of the L monocytogenes EGDegenome For primer design a modified version of primer 3which allows testing of the specificity of the obtained PCRproducts against the complete genome sequence to avoidcross-hybridization was used (CAAT-Box L Frangeul et alunpublished) Primers (Metabion) were chosen in order toamplify a fragment of 300ndash600 bp specific for each ORF witha melting temperature of 55ndash65infinC Amplification reactionswere performed in 96-well plates (Perkin-Elmer) in a 100 mlreaction volume containing 20 ng of chromosomal DNA fromL monocytogenes EGDe 2 U of DyNazyme EXT DNA poly-

merase (Finnzymes) 06 mM each primer and 02 mMdNTPs (Perkin-Elmer) Reactions were cycled 35 times(94infinC for 15 s 55infinC for 30 s 72infinC for 60 s) with one finalcycle of 72infinC for 7 min in a thermocycler Successful ampli-fication of each PCR product was verified on agarose gelsand negative PCRs were repeated resulting in the finalamplification of 2816 ORFs of the 2853 ORFs (99) identi-fied in the L monocytogenes EGDe genome For macroarraypreparation nylon membranes (Qfilter Genetix) were soakedin 10 mM TE pH 76 Spot blots of ORF-specific PCR prod-ucts and controls were printed using a Qpix robot (Genetix)Immediately after spot deposition membranes were neutral-ized for 15 min in 05 M NaOH 15 M NaCl washed threetimes with distilled water and stored wet at -20infinC until use

Validation of the macroarrays

To ensure that DNA samples were successfully deposited onthe nylon membranes and to assess differential hybridizationto the target genes 33P-labelled genomic DNA from L mono-cytogenes EGDe was hybridized before transcriptome anal-ysis to the macroarray The total intensity of all pixels withineach spot was determined after scanning of exposed phos-phor screens Fifty-six spots scored as undetectable andwere therefore not taken into account in our study In orderto establish optimal experimental conditions RNA from theEGDe wild-type strain and its prfA-deleted mutant grown inBHI or BHIC was tested To determine the best labellingmethod random hexamers (Boehringer Mannheim) or a mix-ture of all 3cent oligonucleotides (Metabion) each at a concen-tration of 0033 mM specific for each amplified gene wereused Similar results were obtained using the two differentlabelling methods suggesting that the observed distributionreflected the true distribution of transcripts (data not shown)For all further experiments a mixture of specific oligonucle-otides was chosen The intensity distribution patterns showedgood reproducibility (r = 0959) between independent exper-iments In contrast the correlation between the intensity ofspots generated with cDNAs from wild-type cells and thosegenerated with cDNAs from prfA mutant cells showed thatthe mutation caused significant changes in the expressionprofile (r = 0905)

Sample collection cell lysis and total RNA isolation

BHI BHIC or BHICel was inoculated with 100 ml of the over-night culture and incubated at 37infinC with shaking Exponen-tially growing cells (OD600 06ndash08) were harvested for 2 minat 6000 g at 4infinC The pellets were flash frozen on dry icendashethanol and stored at -80infinC For extraction cells were resus-pended by vortexing in 400 ml of resuspension buffer(125 mM Tris 5 mM EDTA and 10 glucose) Then 500 mlof acid phenol (pH 46) and 04 g of glass beads (02ndash03 mmdiameter Sigma) were added The cells were shearedmechanically using a Fastprep apparatus (Bio101) After cen-trifugation at 13 000 g for 5 min the supernatant was trans-ferred to a fresh tube and 1 ml of Trizol reagent (Gibco BRL)was added The sample was incubated for 5 min at roomtemperature Total RNA was extracted twice with chloroformndash

1622 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

isoamyl alcohol (241 vv) and precipitated in 07 volumes ofisopropanol After a washing step with 70 ethanol the RNApellet was dissolved in sterile DNase- and RNase-free water(ICN Biomedicals) and quantified by absorbance at 260 and280 nm Purity and integrity of RNA were controlled on aga-rose gels and RNA was stored at -80infinC until use

Genomic DNA labelling and cDNA labelling

Chromosomal DNA (30 ng) isolated from L monocytogenesstrains EGDe or P14 was 33P-labelled using a random-primedDNA labelling kit (Boehringer Mannheim) Labelled genomicDNAs were purified before hybridization using QIAquick col-umns (Qiagen)

For cDNA synthesis random hexamers or a set of 3cent-specific oligonucleotide primers were used in reversetranscription reactions in the presence of [a-33P]-dCTP(2000ndash3000 Ci mmol-1 Amersham) Total RNA (1 mg) and50 U of AMV reverse transcriptase (Roche) with RNase Hactivity were used Labelled cDNA was purified to removeunincorporated nucleotides before hybridization using aQIAquick column (Qiagen)

Hybridization

Hybridization and washing steps were carried out usingSSPE buffer (018 M NaCl 10 mM NaH2PO4 1 mM EDTApH 77) Macroarrays were prewet in 2yen SSC and prehybrid-ized for at least 2 h in 10 ml of hybridization solution (5yenSSPE 2 SDS 1yen Denhardtrsquos reagent 100 mg of shearedsalmon sperm DNA ml-1) at 65infinC in roller bottles Hybridiza-tion was carried out for 20 h at 65infinC with 5 ml of hybridizationsolution and the entire spin-purified cDNA probe Afterhybridization membranes were washed twice at room tem-perature and twice at 65infinC in 05yen SSC 02 SDS Arrayswere then sealed in thin bags and exposed to a phosphorscreen (Molecular Dynamics) for 24ndash72 h For each straintwo independent RNA preparations were tested and twocDNAs from each of the RNA preparations were hybridizedto two sets of arrays and analysed

Data analysis

Membranes were scanned using a 445SI PhosphorImager(Molecular Dynamics) The ARRAYVISION software (ImagingResearch) was used for quantification of the hybridizationintensities and for normalization The intensity of each spotwas normalized according to the median value of the totalintensities of all spots on each array which allowed directcomparison of the two strains The global background wascalculated from the average intensity of 610 lsquono-DNArsquo spotshomogeneously distributed throughout the membrane Onlyspots for which the sum of the normalized intensity values ofthe wild type and the mutant strain was higher than the sumof the global background of the wild type and the mutantmembrane were taken into account

For identification of genes with statistically significantchanges in expression SAM a statistical technique for findingsignificant genes in a set of microarray experiments was used(Tusher et al 2001) (httpwww-statstanfordedu~tibs

SAM) We chose a delta value corresponding to a falsediscovery rate (FDR) lt10 The FDR was computed as theratio of the estimated number of lsquofalse significantrsquo genes tothe total number of lsquosignificantrsquo genes Genes with a twofoldexpression change that were significant according to thisanalysis were taken into account The complete data set ofthe statistical analysis is available online as Supplementarymaterial

Real-time quantitative PCR

Real-time quantitative PCR was conducted on the same totalcellular RNAs as those used for transcriptome experimentsPrimers (Table S7) were designed with PRIMEREXPRESS soft-ware (Applied Biosystems) and purchased from EurogentecRNA (20 mg) was treated with 50 U of RNase-free DNase I(Roche) for 30 min at 37infinC DNase was inactivated by phenolchloroform extraction and RNA was precipitated cDNA syn-thesis was performed for 60 min at 42infinC using 10 mg of RNA2 ml of AMV reverse transcriptase (25 U ml-1 Roche) and 2 mlof the primer mix To determine the primer concentration withthe best efficiency concentrations of 100 nM 200 nM and400 nM in a 25 ml reaction volume were tested for eachprimer The optimal concentration of 200 nM was used in thefinal experiments Real-time quantitative PCR was performedin a 25 ml reaction volume containing 200 ng of cDNA 125 mlof SYBR PCR master mix (Applied Biosystems) and 2 ml ofgene-specific primers (200 nM) Amplification and detectionof specific products were performed with the ABI Prism 7700sequence detection system (PE Applied Biosystems) with thefollowing cycle profile one cycle at 50infinC for 2 min one cycleat 95infinC for 10 min 40 cycles at 95infinC for 15 s and 60infinC for1 min Each product was deposited on agarose gels to verifythe presence of a single band The quantity of cDNA for eachexperimental gene was normalized to the quantity of rpoBcDNA in each sample For each gene triplicate assays weredone To check whether contaminating chromosomal DNAwas present each sample was tested in control reactions thatdid not contain reverse transcriptase

Acknowledgements

This work received financial support from the EuropeanCommission contract QLG2-CT-1999-00932 the SpanishMinisterio de Ciencia y Technologiacutea (BMC-2000-0553) theMinistere de lrsquoEducation Nationale de la Recherche et de laTechnologie (PFRMMIP) and the Pasteur Institut (PTR 6)EM holds a post-doctoral fellowship from SOREDAB SASF Chetouani we thank for access to unpublished results Wethank the following persons for help in preparing macroar-rays E Zalachas C Lacroix A Schiavo and E CouveacuteSpecial thanks to A Bourdet who took time to help us withthe quantitative PCR experiments Tim Stinear Claire Poyartand Chester Price for critical reading of the manuscript andA Maitournam and M Zouine for help with statistical analy-sis We also thank the reviewers for their constructive com-ments P Cossart is an international investigator from theHoward Hughes Medical Institute

1622 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

isoamyl alcohol (241 vv) and precipitated in 07 volumes ofisopropanol After a washing step with 70 ethanol the RNApellet was dissolved in sterile DNase- and RNase-free water(ICN Biomedicals) and quantified by absorbance at 260 and280 nm Purity and integrity of RNA were controlled on aga-rose gels and RNA was stored at -80infinC until use

Genomic DNA labelling and cDNA labelling

Chromosomal DNA (30 ng) isolated from L monocytogenesstrains EGDe or P14 was 33P-labelled using a random-primedDNA labelling kit (Boehringer Mannheim) Labelled genomicDNAs were purified before hybridization using QIAquick col-umns (Qiagen)

For cDNA synthesis random hexamers or a set of 3cent-specific oligonucleotide primers were used in reversetranscription reactions in the presence of [a-33P]-dCTP(2000ndash3000 Ci mmol-1 Amersham) Total RNA (1 mg) and50 U of AMV reverse transcriptase (Roche) with RNase Hactivity were used Labelled cDNA was purified to removeunincorporated nucleotides before hybridization using aQIAquick column (Qiagen)

Hybridization

Hybridization and washing steps were carried out usingSSPE buffer (018 M NaCl 10 mM NaH2PO4 1 mM EDTApH 77) Macroarrays were prewet in 2yen SSC and prehybrid-ized for at least 2 h in 10 ml of hybridization solution (5yenSSPE 2 SDS 1yen Denhardtrsquos reagent 100 mg of shearedsalmon sperm DNA ml-1) at 65infinC in roller bottles Hybridiza-tion was carried out for 20 h at 65infinC with 5 ml of hybridizationsolution and the entire spin-purified cDNA probe Afterhybridization membranes were washed twice at room tem-perature and twice at 65infinC in 05yen SSC 02 SDS Arrayswere then sealed in thin bags and exposed to a phosphorscreen (Molecular Dynamics) for 24ndash72 h For each straintwo independent RNA preparations were tested and twocDNAs from each of the RNA preparations were hybridizedto two sets of arrays and analysed

Data analysis

Membranes were scanned using a 445SI PhosphorImager(Molecular Dynamics) The ARRAYVISION software (ImagingResearch) was used for quantification of the hybridizationintensities and for normalization The intensity of each spotwas normalized according to the median value of the totalintensities of all spots on each array which allowed directcomparison of the two strains The global background wascalculated from the average intensity of 610 lsquono-DNArsquo spotshomogeneously distributed throughout the membrane Onlyspots for which the sum of the normalized intensity values ofthe wild type and the mutant strain was higher than the sumof the global background of the wild type and the mutantmembrane were taken into account

For identification of genes with statistically significantchanges in expression SAM a statistical technique for findingsignificant genes in a set of microarray experiments was used(Tusher et al 2001) (httpwww-statstanfordedu~tibs

SAM) We chose a delta value corresponding to a falsediscovery rate (FDR) lt10 The FDR was computed as theratio of the estimated number of lsquofalse significantrsquo genes tothe total number of lsquosignificantrsquo genes Genes with a twofoldexpression change that were significant according to thisanalysis were taken into account The complete data set ofthe statistical analysis is available online as Supplementarymaterial

Real-time quantitative PCR

Real-time quantitative PCR was conducted on the same totalcellular RNAs as those used for transcriptome experimentsPrimers (Table S7) were designed with PRIMEREXPRESS soft-ware (Applied Biosystems) and purchased from EurogentecRNA (20 mg) was treated with 50 U of RNase-free DNase I(Roche) for 30 min at 37infinC DNase was inactivated by phenolchloroform extraction and RNA was precipitated cDNA syn-thesis was performed for 60 min at 42infinC using 10 mg of RNA2 ml of AMV reverse transcriptase (25 U ml-1 Roche) and 2 mlof the primer mix To determine the primer concentration withthe best efficiency concentrations of 100 nM 200 nM and400 nM in a 25 ml reaction volume were tested for eachprimer The optimal concentration of 200 nM was used in thefinal experiments Real-time quantitative PCR was performedin a 25 ml reaction volume containing 200 ng of cDNA 125 mlof SYBR PCR master mix (Applied Biosystems) and 2 ml ofgene-specific primers (200 nM) Amplification and detectionof specific products were performed with the ABI Prism 7700sequence detection system (PE Applied Biosystems) with thefollowing cycle profile one cycle at 50infinC for 2 min one cycleat 95infinC for 10 min 40 cycles at 95infinC for 15 s and 60infinC for1 min Each product was deposited on agarose gels to verifythe presence of a single band The quantity of cDNA for eachexperimental gene was normalized to the quantity of rpoBcDNA in each sample For each gene triplicate assays weredone To check whether contaminating chromosomal DNAwas present each sample was tested in control reactions thatdid not contain reverse transcriptase

Acknowledgements

This work received financial support from the EuropeanCommission contract QLG2-CT-1999-00932 the SpanishMinisterio de Ciencia y Technologiacutea (BMC-2000-0553) theMinistere de lrsquoEducation Nationale de la Recherche et de laTechnologie (PFRMMIP) and the Pasteur Institut (PTR 6)EM holds a post-doctoral fellowship from SOREDAB SASF Chetouani we thank for access to unpublished results Wethank the following persons for help in preparing macroar-rays E Zalachas C Lacroix A Schiavo and E CouveacuteSpecial thanks to A Bourdet who took time to help us withthe quantitative PCR experiments Tim Stinear Claire Poyartand Chester Price for critical reading of the manuscript andA Maitournam and M Zouine for help with statistical analy-sis We also thank the reviewers for their constructive com-ments P Cossart is an international investigator from theHoward Hughes Medical Institute

Analysis of the PrfA regulon of L monocytogenes 1623

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

Supplementary material

The following material is available from httpwwwblackwellpublishingcomproductsjournalssuppmatmolemole3413mmi3413smhtmTable S1A Primary expression data obtained for wild-type Lmonocytogenes EGDe and its isogenic prfA deleted mutantafter growth in BHITable S1B Genes whose expression was significantlydifferent between L monocytogenes EGDe and EGD-eDprfA after growth in BHITable S2A Primary expression data obtained for wild-type Lmonocytogenes EGDe and its isogenic prfA deleted mutantafter growth in BHICTable S2B Genes whose expression was significantlydifferent between L monocytogenes EGDe and EGD-eDprfA after growth in BHICTable S3A Primary expression data obtained for wild-type Lmonocytogenes P14prfA and its isogenic prfA deletedmutant after growth in BHITable S3B Genes whose expression was significantlydifferent between L monocytogenes P14prfA and P14DprfAafter growth in BHITable S4A Primary expression data obtained for wild-type Lmonocytogenes P14prfA and its isogenic prfA deletedmutant after growth in BHICTable S4B Genes whose expression was significantlydifferent between L monocytogenes P14prfA andP14DprfA after growth in BHICTable S5A Primary expression data obtained for wild-type Lmonocytogenes EGDe and its isogenic prfA deleted mutantafter growth in BHICelTable S5B Genes whose expression was significantly differ-ent between L monocytogenes EGDe and EGDeDprfAafter growth in BHICelTable S6A Primary expression data obtained for wild-type Lmonocytogenes P14prfA and its isogenic prfA deletedmutant after growth in BHICelTable S6B Genes whose expression was significantly differ-ent between L monocytogenes P14prfA and P14DprfA aftergrowth in BHICelTable S7 Genes and corresponding primers used forquantitative PCR

References

Akbar S Gaidenko TA Kang CM OrsquoReilly M DevineKM and Price CW (2001) New family of regulators inthe environmental signaling pathway which activates thegeneral transcription factor sigma (B) of Bacillus subtilis JBacteriol 183 1329ndash1338

Bailey TL and Elkan C (1994) Fitting a mixture model byexpectation maximization to discover motifs in biopoly-mers Proc Int Conf Intell Syst Mol Biol 2 28ndash36

Becker LA Evans SN Hutkins RW and Benson AK(2000) Role of sigma (B) in adaptation of Listeria monocy-togenes to growth at low temperature J Bacteriol 1827083ndash7087

Bockmann R Bickneite C Middendorf B Goebel Wand Sokolovic Z (1996) Specific binding of the Listeriamonocytogenes transcriptional regulator PrfA to target

sequences requires additional factor(s) and is influencedby iron Mol Microbiol 22 643ndash653

Bohne J Sokolovic Z and Goebel W (1994) Transcrip-tional regulation of prfA and PrfA-regulated virulencegenes in Listeria monocytogenes Mol Microbiol 11 1141ndash1150

Brehm K Kreft J Ripio MT and Vazquez-Boland JA(1996) Regulation of virulence gene expression in patho-genic Listeria Microbiologia 12 219ndash236

Brehm K Ripio MT Kreft J and Vazquez-Boland JA(1999) The bvr locus of Listeria monocytogenes mediatesvirulence gene repression by beta-glucosides J Bacteriol181 5024ndash5032

Brosch R Catimel B Milon G Buchrieser C Vindel Eand Rocourt J (1993) Virulence heterogeneity of Listeriamonocytogenes strains from various sources (foodhuman animal) in immunocompetent mice and its associ-ation with typing characteristics J Food Prot 56 296ndash301

Chico-Calero I Suarez M Gonzalez-Zorn B Scortti MSlaghuis J Goebel W and Vazquez-Boland JA (2002)Hpt a bacterial homolog of the microsomal glucose-6-phosphate translocase mediates rapid intracellular prolif-eration in Listeria Proc Natl Acad Sci USA 99 431ndash436

Cossart P and Lecuit M (1998) Interactions of Listeriamonocytogenes with mammalian cells during entry andactin-based movement bacterial factors cellular ligandsand signaling EMBO J 17 3797ndash3806

Cruz Ramos H Boursier L Moszer I Kunst F DanchinA and Glaser P (1995) Anaerobic transcription activationin Bacillus subtilis identification of distinct FNR-dependentand -independent regulatory mechanisms EMBO J 14(23) 5984ndash5994

Dassa E and Bouige P (2001) The ABC of ABCS aphylogenetic and functional classification of ABC systemsin living organisms Res Microbiol 152 211ndash229

Deora R Tseng T and Misra TK (1997) Alternativetranscription factor sigmaSB of Staphylococcus aureuscharacterization and role in transcription of the globalregulatory locus sar J Bacteriol 179 6355ndash6359

Dramsi S Lebrun M and Cossart P (1996) Molecular andgenetic determinants involved in invasion of mammaliancells by Listeria monocytogenes Curr Topics MicrobiolImmunol 209 61ndash77

Dussurget O Cabanes D Dehoux P Lecuit MConsortium ELG Buchrieser C et al (2002) Listeriamonocytogenes bile salt hydrolase is a virulence factorinvolved in the intestinal and hepatic phase of listeriosisMol Microbiol 45 1095ndash1106

Engelbrecht F Chun S-K Ochs C Hess J LottspeichF Goebel W and Sokolovic Z (1996) A new PrfA-regulated gene of Listeria monocytogenes encoding asmall secreted protein which belongs to the family of inter-nalins Mol Microbiol 21 823ndash837

Ferreira A OrsquoByrne CP and Boor KJ (2001) Role ofsigma (B) in heat ethanol acid and oxidative stress resis-tance and during carbon starvation in Listeria monocyto-genes Appl Environ Microbiol 67 4454ndash4457

Freitag NE Rong L and Portnoy DA (1993) Regulationof the prfA transcriptional activator of Listeria monocytoge-nes multiple promoter elements contribute to intracellular

1624 E Milohanic et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

growth and cell-to-cell spread Infect Immun 61 2537ndash2544

Geoffroy C Gaillard JL Alouf J and Berche P (1989)Production of thiol-dependent hemolysins by Listeriamonocytogenes and related species J Gen Microbiol 135481ndash487

Geoffroy C Raveneau J Beretti JL Lecroisey AVazquez-Boland JA Alouf JE and Berche P (1991)Purification and characterization of an extracellular 29-kilodalton phospholipase C from Listeria monocytogenesInfect Immun 59 2382ndash2388

Glaser P Frangeul L Buchrieser C Rusniok C AmendA Baquero F et al (2001) Comparative genomics ofListeria species Science 294 849ndash852

Hecker M Schumann W and Volker U (1996) Heat-shockand general stress response in Bacillus subtilis Mol Micro-biol 19 417ndash428

Irvine AS and Guest JR (1993) Lactobacillus casei con-tains a member of the CRP-FNR family Nucleic Acids Res21 753

Jacobs M Andersen JB Kontinen V and Sarvas M(1993) Bacillus subtilis PrsA is required in vivo as an extra-cytoplasmic chaperone for secretion of active enzymessynthesized either with or without pro-sequences MolMicrobiol 1993 957ndash966

Johansson J Mandin P Renzoni A Chiaruttini CSpringer M and Cossart P (2002) An RNA thermosen-sor controls expression of virulence genes in Listeriamonocytogenes Cell 110 551ndash561

Ko R and Smith LT (1999) Identification of an ATP-drivenosmoregulated glycine betaine transport system inListeria monocytogenes Appl Environ Microbiol 65 4040ndash4048

Kontinen VP and Sarvas M (1993) The PrsA lipoproteinis essential for protein secretion in Bacillus subtilis and setsa limit for high-level secretion Mol Microbiol 4 727ndash737

Kontinen VP Saris P and Sarvas M (1991) A gene(prsA) of Bacillus subtilis involved in a novel late stage ofprotein export Mol Microbiol 5 1273ndash1283

Kreft J and Vazquez-Boland JA (2001) Regulation of vir-ulence genes in Listeria Int J Med Microbiol 291 849ndash852

Kreuzer P Gartner D Allmansberger R and Hillen W(1989) Identification and sequence analysis of the Bacillussubtilis W23 xylR gene and xyl operator J Bacteriol 13840ndash3845

Lampidis R Gross R Sokolovic Z Goebel W and KreftJ (1994) The virulence regulator protein of Listeria ivanoviiis highly homologous to PrfA from Listeria monocytogenesand both belong to the Crp-Fnr family of transcription reg-ulators Mol Microbiol 13 141ndash151

Lingnau A Chakraborty T Niebuhr K Domann E andWehland J (1996) Identification and purification of novelinternalin-related proteins in Listeria monocytogenes andListeria ivanovii Infect Immun 64 1002ndash1006

Liu X Brutlag DL and Liu JS (2001) BioProspectordiscovering conserved DNA motifs in upstream regulatoryregions of co-expressed genes Pac Symp Biocomput127ndash138

Mengaud J Vicente MF and Cossart P (1989) Tran-scriptional mapping and nucleotide sequence of the Liste-ria monocytogenes hlyA region reveal structural features

that may be involved in regulation Infect Immun 57 3695ndash3701

Mengaud J Dramsi S Gouin E Vazquez-Boland JAMilon G and Cossart P (1991) Pleiotropic control ofListeria monocytogenes virulence factors by a gene whichis autoregulated Mol Microbiol 5 2273ndash2283

Milenbachs A Brown DP Moors M and Youngman P(1997) Carbon-source regulation of virulence gene expres-sion in Listeria monocytogenes Mol Microbiol 23 1075ndash1085

Nadon CA Bowen BM Wiedmann M and BoorKJ (2002) Sigma B Contributes to PrfA-mediated viru-lence in Listeria monocytogenes Infect Immun 70 3948ndash3952

Park SF and Kroll RG (1993) Expression of listeriolysinand phosphatidylinositol-specific phospholipase C isrepressed by the plant-derived molecule cellobiose in List-eria monocytogenes Mol Microbiol 8 653ndash661

Petersohn A Brigulla M Haas S Hoheisel JD VolkerU and Hecker M (2001) Global analysis of the generalstress response of Bacillus subtilis J Bacteriol 183 5617ndash5631

Price CW Fawcett P Ceacuteremonie H Su N MurphyCK and Youngman P (2001) Genome-wide analysis ofthe general stress response in Bacillus subtilis Mol Micro-biol 41 757ndash774

Renzoni A Klarsfeld A Dramsi S and Cossart P (1997)Evidence that PrfA the pleiotropic activator of virulencegenes in Listeria monocytogenes can be present but inac-tive Infect Immun 65 1515ndash1518

Ripio MT Dominguez-Bernal G Suarez M Brehm KBerche P and Vazquez-Boland J (1996) Transcriptionalactivation of virulence genes in wild type strains of Listeriamonocytogenes response to a change in extracellularmedium composition Res Microbiol 147 371ndash384

Ripio M Brehm K Lara M Suarez M and Vazquez-Boland JA (1997a) Glucose-1-phosphate utilization byListeria monocytogenes is PrfA dependent and coordi-nately expressed with virulence factors J Bacteriol 1797174ndash7180

Ripio MT Dominguez-Bernal G Lara M Suarez M andVazquez-Boland JA (1997b) A Gly145Ser substitution inthe transcriptional activator PrfA causes constitutive over-expression of virulence factors in Listeria monocytogenesJ Bacteriol 179 1533ndash1540

Sheehan B Klarsfeld A Ebright R and Cossart P(1996) A single substitution in the putative helix-turn-helixmotif of the pleiotropic activator PrfA attenuates Listeriamonocytogenes virulence Mol Microbiol 20 785ndash797

Tusher V Tibshirani R and Chu C (2001) Significanceanalysis of microarrays applied to ionizing radiationresponse Proc Natl Acad Sci USA 98 5116ndash5121

Vazquez-Boland JA Ferrer D and Rocourt J (1991)Heterogeneity of strains of Listeria monocytogenes iso-lated during an outbreak of listeriosis among adults inValencia in 1989 Enferm Infect Microbiol Clin 7 442ndash444

Vazquez-Boland J-A Kuhn M Berche P ChakrabortyT Dominguez-Bernal G Goebel W et al (2001) Liste-ria pathogenesis and molecular virulence determinantsClin Microbiol Rev 14 1ndash57

Analysis of the PrfA regulon of L monocytogenes 1625

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

Vega Y Dickneite C Ripio MT Bockmann RGonzalez-Zorn B Novella S et al (1998) Functionalsimilarities between the Listeria monocytogenes virulenceregulator PrfA and cyclic AMP receptor protein the PrfA(Gly145Ser) mutation increases binding affinity for targetDNA J Bacteriol 180 6655ndash6660

Vos P van Asseldonk M van Jeveren F Siezen RSimons G and de Vos WM (1989) A maturation pro-tein is essential for production of active forms of Lacto-coccus lactis SK11 serine proteinase located in orsecreted from the cell envelope J Bacteriol 171 2795ndash2802

Analysis of the PrfA regulon of L monocytogenes 1625

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 47 1613ndash1625

Vega Y Dickneite C Ripio MT Bockmann RGonzalez-Zorn B Novella S et al (1998) Functionalsimilarities between the Listeria monocytogenes virulenceregulator PrfA and cyclic AMP receptor protein the PrfA(Gly145Ser) mutation increases binding affinity for targetDNA J Bacteriol 180 6655ndash6660

Vos P van Asseldonk M van Jeveren F Siezen RSimons G and de Vos WM (1989) A maturation pro-tein is essential for production of active forms of Lacto-coccus lactis SK11 serine proteinase located in orsecreted from the cell envelope J Bacteriol 171 2795ndash2802